Tau protects beta in the leading-strand polymerase complex at the replication fork.

Replication forks formed in the absence of the tau subunit of the DNA polymerase III holoenzyme produce shorter leading and lagging strands than when tau is present. We show that one reason for this is that in the absence of tau, but in the presence of the gamma-complex, leading-strand synthesis is no longer highly processive. In the absence of tau, the size of the leading strand becomes proportional to the concentration of beta and inversely proportional to the concentration of the gamma-complex. In addition, the beta in the leading-strand complex is no longer resistant to challenge by either anti-beta antibodies or poly(dA):oligo(dT). Thus, tau is required to cement a processive leading-strand complex, presumably by preventing removal of beta catalyzed by the gamma-complex.

The replication fork of Escherichia coli is extraordinarily processive. Two replication forks form at oriC and synthesize roughly 2.2 megabase pairs of DNA before they meet again in the terminus region. The enzymatic machinery at the replication fork accomplishes this task while supporting two distinct modes of DNA synthesis. Whereas the leading strand is synthesized in a continuous fashion that reflects the overall processivity of the replication fork, the lagging strand is synthesized discontinuously in short Okazaki fragments 2 kb 1 in length (1). This issue is compounded by the fact that the mechanism responsible for maintaining the processivity of the replicative polymerase on either strand is the same, a complex between the ␤ subunit and the polymerase core of the DNA polymerase III holoenzyme (Pol III HE) (2,3).
The core (4), the catalytic polymerase/exonuclease subassembly of the Pol III HE, is essentially a distributive enzyme, synthesizing only a few nucleotides per primer binding event (5). For conversion to a processive enzyme, the core must be clamped onto DNA by associating with the ␤ subunit (2, 3), a dimer that encircles double-stranded DNA (6,7). ␤ can be loaded onto the primer terminus via the action of five other HE subunits that themselves associate, forming the ␥-complex (␥, ␦, ␦Ј, , and ) (8).
Using rolling circle DNA replication supported by a tailed form II DNA template (TFII) and the X-type primosomal proteins (PriA, PriB, PriC, DnaT, DnaB, DnaC, and DnaG), the single-stranded DNA-binding protein (SSB), and either bona fide Pol III HE or HE reconstituted from purified subunits, we have reconstituted the coordinated leading-and lagging-strand synthesis of the E. coli replication fork (9 -13). These replication forks were shown to have processivities of at least 0.5 megabase, yet they also made Okazaki fragments whose average length was about 1.8 kb (9). This suggested that the required protein elements for establishing processivity and triggering lagging-strand polymerase cycling were present.
The mechanism for inducing polymerase release on the lagging strand is yet to be established firmly. We have proposed that protein-protein interactions between a primase synthesizing the primer for the next Okazaki fragment and the laggingstrand polymerase initiates termination of synthesis and keys recycling of the polymerase to the new primer (11,13). In the bacteriophage T4 system, Hacker and Alberts (14) have argued that the lagging-strand polymerase will dissociate spontaneously from the gene 45 protein clamp when it hits the 5Ј-end of the previous Okazaki fragment. Stukenberg et al. (15) have made similar arguments for the E. coli fork.
Challenge experiments have shown that both the polymerase and helicase on the leading-strand side of the fork are processive (16). Our recent studies have demonstrated that a proteinprotein interaction between the subunit of the Pol III HE and DnaB is required to mediate rapid replication fork movement. 2 We show here that contributes in yet another way to processivity on the leading strand by protecting ␤ in the leadingstrand polymerase complex from being removed by the action of the ␥-complex.
[␣-32 P]dATP was from Amersham, Bio-Gel A-150m was from Bio-Rad. Oligo(dT) 20 was synthesized using an Applied Biosystems 380A DNA Synthesizer and was used after gel purification. Alkaline phosphatase was from Boehringer Mannheim. DNAs from bacteriophages flAY-7M and flR229-A/33 were prepared as described previously (17).

Rolling Circle DNA Replication
TFII DNA template was prepared as described by Mok and Marians (16) using [ 3 H]TTP. Rolling circle reaction mixtures (12 l final volume) * These studies were supported by National Institutes of Health Grants GM36255 and GM34557 (to C. S. M. and K. J. M., respectively). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 The abbreviations used are: kb, kilobase(s); Pol III HE, DNA polymerase III holoenzyme; TFII, tailed form II; SSB, E. coli single-stranded DNA-binding protein.  Fig. 1), 28 nM DnaT, 2.5 nM PriA, 2.5 nM PriB, 2.5 nM PriC, and HE subunits and subassemblies (formed by mixing purified subunits at equimolar ratios) at 28 nM unless indicated otherwise, were incubated at 30°C for 2 min. ATP was then added to 1 mM along with GTP, CTP, and UTP each to 200 M and the dNTPs to 40 M, and the incubation continued for 1.5 min. [␣-32 P]dATP (to 2000 -4000 cpm/pmol) was then added and the incubation continued for 10 min. The reaction was terminated by the addition of EDTA to 40 mM. Total DNA synthesis was determined by acid-precipitating an aliquot of the reaction mixture, and the DNA products were analyzed by alkaline agarose gel electrophoresis as described previously (9).

Processivity Challenge Experiments
Poly(dA):Oligo(dT)-Poly(dA):oligo(dT) was prepared by annealing 7 M poly(dA) with 8.6 M oligo(dT) 20 in a reaction mixture (30 l) containing 50 mM Tris-OAc (pH 7.5), 200 mM NaCl, and 20 mM MgCl 2 at 75°C for 2 min followed by slow cooling (2 h) to room temperature. A mixture (2 l) of the preprimosomal proteins (the primosomal proteins in the absence of primase), SSB, core, and ␥-complex, either in the presence or absence of , was added to 20 l of the peak fraction of the ␤-TFII DNA complex to give their standard reaction concentrations. The reaction was initiated by the addition of ATP to 1 mM, GTP, CTP, and UTP each to 200 M, and dATP and dTTP each to 40 M, and incubated at 30°C for 1.5 min. [␣-32 P]dATP and poly(dA):oligo(dT) 20 (to a final concentration of 120 nM 3Ј-ends), as indicated, were then added, and the reaction continued for 10 min at 30°C. Reactions were terminated by the addition of EDTA to 40 mM and processed and analyzed as described above.
Anti-␤ Antibody-Standard rolling circle reaction mixtures were assembled either in the presence or absence of . The effect of the anti-␤ antibody on initiation was assessed by including it in the reaction mixture from the start, before the ATP concentration was raised and before the dNTPs and other NTPs were added. The effect of the antibody on elongation was determined by adding the antibody along with the [␣-32 P]dATP, 1.5 min after the reactions had been initiated. Reactions were processed and analyzed as described above.

Replication Forks Formed in the Absence of Synthesize
Shorter Leading and Lagging Strands-In the course of our studies on the roles of the subunits of the Pol III HE at the replication fork, we noted that replication forks formed in the absence of synthesized shorter leading and lagging strands than those reconstituted with (Fig. 1). We found that played a central role in ensuring the generation of a processive replication fork. A protein-protein interaction between and the replication fork helicase DnaB was required to mediate rapid fork movement 2 and, as described here, in the absence of , the leading-strand side of the fork becomes nonprocessive.
The processivity of the E. coli replication fork in vitro is reflected in its high rate of speed (9,16), the inaccessibility of ␤ on the leading strand to challenge (16,23), and the insensitivity of the length of the leading strand to the concentration of ␤ (9). Forks that lacked and free ␥-complex were still processive. 2 We examined whether this held true in the presence of added ␥-complex as well.
The length of the leading strand synthesized by -less replication forks was dependent on the concentration of the ␤ subunit (Fig. 2). This is strikingly different from the situation in the presence of where, although overall DNA synthesis is dependent on ␤, the size of the leading strand is independent of the ␤ concentration (9). This is consistent with the leadingstrand ␤-core complex needing to form only once for synthesis of a long continuous DNA product. This observation suggested that multiple ␤s were required in the absence of in order to synthesize the leading strand. That is, ␤ was being cycled in and out of the leading-strand complex.
The only group of proteins available in these reactions that could affect ␤ loading onto 3Ј-ends was the ␥-complex. It followed that it was the ␥-complex that was removing ␤ from the leading-strand complex as well. If this were true, then the size of the leading strand synthesized in the absence of should also be affected by the concentration of the ␥-complex. This proved to be the case.
In the presence of , the concentration of the ␥-complex had no effect on leading-strand synthesis (Fig. 3A), whereas in the absence of , the size of the leading strand synthesized decreased progressively as the concentration of ␥-complex in- creased (Fig. 3B). At the highest concentrations of ␥-complex tested, DNA synthesis was significantly inhibited. These observations suggested that, in the absence of , ␤ was being cycled on and off the leading strand by the action of the ␥-complex.
Replication Forks Formed in the Absence of Are Nonprocessive-We used two types of challenges in order to test the processivity of replication forks formed in the absence of : anti-␤ antibody and poly(dA):oligo(dT) 20 . The former is, of course, specific for ␤, whereas the latter serves to capture ␤, core, and probably DnaB, all of which are normally processive on the leading strand (9,16).
Replication forks capable of synthesizing leading strands were formed using a ␤-TFII DNA complex, the preprimosomal proteins (the primosomal proteins minus DnaG), SSB, core, and the ␥-complex either in the presence or absence of . After 1.5 min, the poly(dA):oligo(dT) 20 challenge was added. Forks formed in the presence of were completely resistant to the challenge (Fig. 4). On the other hand, those formed in the absence of were sensitive, as evidenced by the sharply reduced size of the leading-strand product (Fig. 4). This demonstrated that, in the absence of , at least one of the normally processive enzymatic components on the leading strand was now acting distributively. The data described above suggested that this component was ␤. To assess this, we repeated the challenge experiment using anti-␤ antibody.
␤ becomes inaccessible to anti-␤ antibody once it forms an initiation complex with core (24). We used this observation previously to show that the ␤ on the leading strand was processive, i.e. it was insensitive to the presence of the antibody (16). We repeated those experiments here using standard rolling circle replication reactions reconstituted in either the absence or presence of (Fig. 5). In each case, replication was inhibited if the anti-␤ antibody was added before initiation. However, if the anti-␤ antibody was added 1.5 min after initiation, only replication forks formed in the presence of were resistant, those formed in its absence were inhibited almost completely.
These data show that -less replication forks are nonprocessive because the ␤ in the leading-strand complex has now become vulnerable to disassembly catalyzed by the ␥-complex.

DISCUSSION
During DNA replication, the processivity of the replication fork arises from two contributing sources: the DNA helicase and the leading-strand polymerase. We have recently shown that rapid fork movement requires a physical connection between the polymerase and the helicase that is mediated by a -DnaB protein-protein interaction. 2 In the absence of this interaction, the polymerase follows behind the helicase at a rate equal to the slow (ca. 40 nt/s) unwinding rate of the helicase alone, whereas upon establishing a -DnaB contact, DnaB becomes a more effective helicase, increasing its translocation rate by more than ten-fold. We show here that also contributes directly to the processivity of the leading-strand polymerase by preventing the ␥-complex-catalyzed removal of ␤ from the leading strand.
In the absence of , the length of the leading strand was dependent on the concentration of ␤ and inversely proportional to the concentration of the ␥-complex. In the presence of , the concentrations of these subunits had no effect on leadingstrand synthesis (9). This suggested that at a -less replication fork, ␤, and probably core, were being reused to synthesize the leading strand in short bursts. The distributive nature of the -less replication fork was demonstrated directly in challenge experiments. Thus, not only is essential for maintaining a high rate of fork movement, 2 it is also essential for maintaining the leading-strand ␤-core complex in a form that is resistant to the action of the ␥-complex.
This suggests that, at least on the leading strand, may contact ␤, perhaps covering a face of the protein that is essential for interaction with the ␥-complex. This is consistent with the genetic finding that mutations in dnaX (encoding both and ␥) can be suppressed by mutations in dnaN (encoding ␤) (24). Our recent studies show that the C-terminal domain of ␣ binds both and ␤, thus placing them in proximity and presumably permitting direct protection. 3 Alternatively, contacting core (probably via ␣) causes a rearrangement of the ␤-core complex.
It is not clear when, at a -less replication fork, ␤ becomes available for attack by the ␥-complex. In the absence of both and the ␥-complex, the leading-strand ␤-core assembly is processive for at least 20 kb, following along on the DNA behind the slow moving helicase. 2 It has also been shown that the combination of ␤ and core will synthesize full-length product in a single binding event on poly(dA):oligo(dT) (5). Perhaps the -less leading-strand polymerase moves fitfully behind the helicase, occasionally stalling, and occasionally jumping ahead, and it is while the polymerase is stalled that ␤ can be targeted by the ␥-complex for recycling. On the other hand, in the absence of , ␤ may be bound less tightly to the pol III core, occasionally dissociating and diffusing away unidirectionally. Reassociation with the core would be rapid because of the constraints imposed by diffusion in two-dimensional space along the DNA fiber. It may be that ␤ is removed by ␥-complex while diffusing on the DNA.
In solution, dimerizes the core to give Pol IIIЈ (25). If this structure is similar at the fork, then the core-␤ complex on the lagging strand should be affected by in a similar fashion as the core-␤ complex on the leading strand. Thus, the laggingstrand polymerase should be refractory to recycling. Yet, this is clearly not the case. The lagging-strand core has been shown to cycle from the just-completed Okazaki fragment to the next primer terminus (12,13). It seems likely, then, that mechanisms exist on the lagging strand that override the protective effect of .
These mechanisms could take one of many forms. Conformational rearrangements could occur upon termination of Okazaki fragment synthesis, by whatever mechanism, that expose the lagging-strand core-␤ to disassembly by the ␥-complex. On the other hand, it may be that the protective effect of is specific to the leading-strand polymerase.
Stukenberg et al. (15) demonstrated that ␤ would dissociate from core when the polymerase collided with the 5Ј-end of a DNA strand bound to a template, a situation similar to the lagging-strand polymerase colliding with the 5Ј-triphosphate end of the primer on the previous Okazaki fragment. These experiments were conducted using Pol III* and ␤ on gapped duplex bacteriophage RF DNA. was present, yet disassembly of the core-␤ complex occurred. Thus, does not prevent recycling. This suggests that, in solution, and when a dimeric HE subassembly binds a primer end, the polymerase bound to the primer cannot sense, a priori, whether it should be a leading-or a lagging-strand polymerase. Functional asymmetry that defines the leading-and lagging-strand polymerase may arise from a protein-protein interaction that is established between a properly oriented and DnaB. Establishment of this contact literally cements the replication fork together, activating the helicase and ensuring rapid fork movement, as well as causing a rearrangement that makes ␤ refractory to the action of the ␥-complex.
FIG. 5. ␤ on the leading strand of -less replication forks is sensitive to anti-␤ antibody. Standard rolling circle replication reactions were assembled as indicated in the absence of primase and in either the absence or presence of . In the lanes labeled with an I, anti-␤ antibody was added before ␤ to the reaction mixtures. In the lanes labeled with an E, anti-␤ antibody was added 1.5 min after initiation of DNA synthesis, as indicated under "Materials and Methods." The distinct bands that appear at about 7 and 14 kb are inactive monomer and dimer templates that become labeled by the addition of a few nucleotide residues.