The χψ Subunits of DNA Polymerase III Holoenzyme Bind to Single-stranded DNA-binding Protein (SSB) and Facilitate Replication of an SSB-coated Template*

A complex of the χ and ψ proteins is required to confer resistance to high levels of glutamate on the DNA polymerase III holoenzyme-catalyzed reaction (Olson, M., Dallmann, H. G., and McHenry, C. (1995) J. Biol. Chem. 270, 29570–29577). We demonstrate that this salt resistance also requires templates to be coated with the Escherichia coli single-stranded DNA-binding protein (SSB). We show that this is the result of a direct χψ–SSB interaction that is strengthened approximately 1000-fold when SSB is bound to DNA. On model oligonucleotide templates, DNA polymerase III core is inhibited by SSB. We show that the minimal polymerase assembly that will synthesize DNA on SSB-coated templates is polymerase III–τ–ψχ. γ, the alternative product of thednaX gene, will not replace τ in this reaction, indicating that τ’s unique ability to bind to DNA polymerase III holding χψ in the same complex is essential. All of our findings are consistent with χψ strengthening DNA polymerase III holoenzyme interactions with the SSB-coated lagging strand at the replication fork, facilitating complex assembly and elongation.

The 10-subunit DNA polymerase III holoenzyme is the major replicative polymerase of Escherichia coli, responsible for synthesizing the entire bacterial chromosome. Like other replicases from eukaryotes and prokaryotes, the holoenzyme 1 can be resolved into three primary functional units: a polymerase core (␣⅐⑀⅐), a sliding clamp processivity factor (␤ 2 ), and a clamp assembly apparatus (DnaX complex, 2 ␥ 2 ␦ 1 ␦Ј 1 1 1 ). and ␥ subunits are both products of the dnaX gene (1, 2); they comprise the ATPase that drives ␤ loading on a primed template and replication complex assembly (3)(4)(5). The ␥-subunit is a truncated version of arising from a Ϫ1 ribosomal frameshift (6 -9). The carboxyl-terminal extension of , absent from ␥, is responsible for dimerization of pol III (10 -12) and binding to the DnaB helicase, effectively coupling all of the replicative activities of the fork into one complex (13,14). The DnaX complex auxiliary subunits ␦ and ␦Ј function in clamp assembly (11)(12)(13)(14)(15)(16)(17). The ␦-subunit directly contacts ␤ (16). and perform an ancillary, nonessential role in simple single-stranded assays (15)(16)(17). The presence of and makes the holoenzyme resistant to glutamate concentrations up to 800 mM and dramatically increases the affinity of DnaX for ␦ and ␦Ј, dropping the K D to a point where they saturate DnaX at physiological concentrations (17).
In addition to the holoenzyme, the polymerase can be resolved into three distinct subassemblies of decreasing complexity: pol III*, pol IIIЈ, and pol III. The activity of pol III core is limited in processivity and is inhibited by both SSB and physiological levels of spermidine (18). Association of forms the dimeric pol IIIЈ that is slightly more processive and resistant to spermidine inhibition but still sensitive to inhibition by SSB (18 -21). Pol III* (pol IIIЈ ϩ ␥ 2 ␦ 1 ␦Ј 1 1 1 ) becomes resistant to SSB inhibition and exhibits an increased processivity.
DNA polymerases that are stimulated by their cognate SSBs also bind to them. T4 gp32 and T7 gene 2.5 protein SSBs stimulate and bind to T4 and T7 DNA polymerase, respectively (22)(23)(24)(25)(26). Human RP-A protein has been shown to interact directly with DNA polymerase ␣ (27). In E. coli, SSB stimulates and binds to DNA polymerase II, but not DNA polymerase III (28). Since pol III* is stimulated by SSB and pol IIIЈ is inhibited, ␥, ␦, ␦', , or would be expected to interact with SSB. Consistent with this hypothesis, Fradkin and Kornberg (29) have observed ␥ complex binding to DNA that is SSB-dependent.
In this report, we demonstrate that binds to SSB in the presence or absence of other subunits of the DnaX complex and that the interaction of SSB with is responsible for the previously observed salt resistance conferred upon the holoenzyme by . We also demonstrate an additional function of in forming a minimal polymerase (pol IIIЈ ϩ ) capable of replicating SSB-coated DNA.
Nucleic Acids-Nonlabeled nucleotides and tritiated dTTP were purchased from Amersham Pharmacia Biotech, and the 32 P-nucleotides were purchased from ICN. Oligonucleotides were synthesized by Synthetic Genetics Inc. The sequence of the 52-mer was 5Ј-TTGGACAGA-TGAACGGTGTACAGACCAGGCGCATAGGCTGGCTGACCTTCAT-3Ј. The 102-mer's sequence was 5Ј-TTACGTTGATTTGGGTAATGAATAT-CCGGTTCTTGTCAAGATTACTCTTGATGAAGGTCAGCCAGCCTAT-GCGCCTGGTCTGTACACCGTTCATCTGTCCAA-3Ј. The F52-mer has * This work was supported by National Institutes of Health Grant GM035695. 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.
the same sequence as the 52-mer above but has a fluorescein attached to the 5Ј-OH. 3 The 102-mer has a biotin molecule attached to the 5Ј-end.
The 3Ј-ends of the F52-mer and 102-mer were modified to impart ⑀-exonuclease resistance to the oligonucleotides. In each, the third base from the 3Ј-end is a ribonucleotide connected to the penultimate base through a phosphorothioate linkage. These modifications increase the half-life of the oligonucleotide from 2 s to 18 min (33). M13Gori singlestranded DNA was prepared according to Johanson et al. (31).
FIG. 1. Salt resistance is conferred to a -reconstituted DNA polymerase III holoenzyme only in the presence of both and SSB. DNA synthesis on a 52-mer/M13Gori primer-template was measured as described under "Experimental Procedures" using a holoenzyme reconstituted with in the presence of the indicated potassium glutamate concentration. A DNA 52-mer was used to prime M13 as a template for the reaction. A, each assay contained 500 fmol of 4 , 600 fmol of pol III core (␣⑀), 600 fmol of ␦, 600 fmol of ␦Ј, 500 fmol of , 500 fmol of ␤ 2 , and 480 pmol of 52-mer/M13 (as nucleotide). B, same as in A, only the assay was done in the absence of SSB. Data represent DNA synthesis by -reconstituted holoenzyme in the presence (q) or absence (E) of . Each data point represents the average of a duplicate determination.
Protein Determinations-Protein concentration was determined using the Pierce Coomassie Plus assay according to the manufacturer's specifications. Bovine serum albumin (fat-free, Sigma) was used as an assay standard.

RESULTS
A Re-examination of the Salt Resistance Conferred to the -Reconstituted DNA Polymerase III Holoenzyme by -Previously, we demonstrated that confers salt resistance to a -reconstituted DNA polymerase III holoenzyme but not a ␥-reconstituted holoenzyme (17). The experiments in that report were all performed in the presence of SSB. We sought to ascertain whether SSB played a role in the observed salt effects. We bypassed the need for a DnaG primase and the accompanying SSB requirement by annealing a DNA 52-mer primer to an M13 template. Using this 52-mer/M13 primer-template, we reproduced the -dependent salt resistance of a -reconstituted holoenzyme. In the presence of , the reconstituted holoenzyme retains Ͼ75% of its activity in the presence of 800 mM potassium glutamate. DNA replication of the reconstituted holoenzyme in the absence of is inhibited by high salt (Fig.  1A).
In order to determine whether the presence of SSB influenced salt resistance, we performed the same experiment in the absence of SSB (Fig. 1B). In the absence of SSB and added salt, the reconstituted holoenzyme was still able to replicate 41-43% of the template whether was present or not. As the salt concentration was increased, the activities of both the -proficient and -less -reconstituted enzymes were inhibited equally (Fig. 1B). Thus, the salt resistance previously observed (17) required both and SSB. This result prompted us to examine whether SSB physically interacts with .
Gel Filtration Demonstrates That and SSB Interact Directly-Incubation of equimolar amounts of and SSB 4 resulted in a complex of the components as revealed by Superose 12 gel filtration. The SSB-complex (Fig. 2C) elutes earlier than either of the individual proteins alone (Fig. 2, A and B), demonstrating that and SSB interact directly. The concentration of the and SSB 4 applied to the gel filtration column was 8.3 M with the eluant being diluted 6-fold, suggesting that the -SSB K D is probably in the micromolar range.
SSB Binds to the ␥-Complex via a -SSB Interaction-SSB interacts with the ␥-complex only when is present. Complexes containing only ␥, ␦, and ␦Ј do not bind SSB (Fig. 2D). Minimally, ␥ and are required for ␥ to enter into a complex with SSB (Fig. 2E). The addition of ␦ and ␦Ј to ␥ increases the affinity of ␥ for (17) resulting in more in the high molecular weight complex and an accompanying larger amount of SSB (Fig. 2F). SSB also binds to the -complex in a -dependent manner (data not shown).
Determination of the -SSB K D -Having demonstrated that and SSB interact directly by gel filtration, we determined the K D for their interaction using the BIAcore™. Our initial investigation into the interaction between and SSB on the BIAcore found that the on and off rates for the interaction were too fast to be amenable to a kinetic analysis. Therefore, we utilized an equilibrium approach on the BIAcore to determine the dissociation constant. Various concentrations of were passed over SSB covalently attached to a CM5 sensor chip. Identical injections were made over a bovine serum albuminderivatized chip, and the resulting sensograms were subtracted from the data as background. Fig. 3A shows an overlay plot of sensograms representing injections of varying concentrations. The R max from each sensogram was plotted as a function of the concentration (Fig. 3B). A hyperbolic relationship is observed, indicating that saturation was nearly achieved at concentrations equaling or exceeding 10 M . The data were fit to a rectangular hyperbola (RU bound ϭ [] (RU max )/([] ϩ K D )) generating a K D of 2.7 M for the interaction.
The -SSB Interaction Increases the Affinity of the DnaX Complex for the DNA Primer-Template-Three oligonucleotides were synthesized for use in experiments in solution and on the BIAcore™. The F52-mer was annealed to the 5Ј-biotinylated 102-mer with the 5Ј-fluoresceinated end of the F52-mer being flush with the 3Ј-end of the 102-mer. This construct was immobilized on a streptavidin-derivatized BIAcore™ sensor chip for binding studies (Fig. 4, A and B). The primer-template construct showed no dissociation or loss in RUs over the time course of the experiments. SSB was then passed over the DNA, binding with a stoichiometry of 1:1.1 DNA to SSB 4 (Fig. 4C). Under these ionic conditions, one SSB tetramer would be expected to bind to the exposed 50 nucleotides of single-stranded DNA (35).
We next wanted to ascertain the role of the -SSB interaction during the association of the DnaX complex with SSBcoated primer-template. Either 400 nM -complex (␦␦Ј) or ␦␦Ј complex was passed over primer-template in the presence or absence of SSB prebound to the DNA (Fig. 5, A-D). The -complex does not significantly interact with the primer-template in the absence of SSB (stoichiometry Ͻ 1:0.04) (Fig. 4A).
In the presence of SSB and absence of , the clamp loader complex does not interact stably with primer-template (stoichiometry Ͻ 1:0.04) (Fig. 5D). It is only when both SSB and are present that the -complex interacts with the F52/102-mer (stoichiometry 1:1.9 DNA to -complex) (Fig. 5B). In a parallel experiment using a range of protein concentrations, we determined the apparent K D of theand ␥-complex for SSB-coated DNA to be 3 and 9 nM, respectively. This range is approximately 1000-fold lower than the value determined for and SSB alone, suggesting that the presence of the DNA and auxiliary subunits significantly strengthens the interaction. The binding and subsequent stable complex formed when the -complex interacts with the SSB-coated primer-template prompted us to test whether the ␥-complex functions in a similar manner. Four experiments performed using the ␥-complex (Fig. 6) in place of the -complex demonstrated that significant binding only occurred in the presence of both and SSB (stoichiometry 1:1 DNA to ␥-complex).
The -SSB Interaction Facilitates Primer Elongation by Pol III-In order to assign a functional significance to the -SSB interaction, we investigated its role in DNA replication on short defined primer-templates. Fig. 7 demonstrates that various holoenzyme subassemblies can elongate the F52/102-mer primer-template. Bands represent incorporation of [␣-32 P]TTP arising from DNA synthesis. In the absence of auxiliary subunits, pol III can elongate a small percentage of the primer to full length (Fig. 7, lane 1). This elongation is prevented when SSB coats the primer-template (lane 2). The SSB-coated F52/102- FIG. 5. mediates a strong interaction of the -complex with an SSB-coated primer-template. 400 nM ␦␦Ј or ␦␦Ј was injected over the SSB-coated or uncoated F52/102-mer 3 immobilized on the sensor chip via a biotin-streptavidin linkage (Fig. 3). All proteins were diluted in HKGM buffer plus 100 M ATP. In order to dissociate all bound protein on the F52/102-mer, a 35-l pulse of 1.5 M guanidine, 50 mM HEPES, 0.005% P20 detergent (Amersham Pharmacia Biotech) guanidine regeneration was done between each analysis on the primer-template. Regeneration and removal of all bound protein was effective, since the increased experimental signal was returned to the initial RU value for the DNA primer-template. A, 400 nM ␦␦Ј was injected over primer-template. B, the same conditions as in A except that the primer-template was coated (1:1 stoichiometry SSB 4 to F52/102-mer) with SSB. C, 400 nM ␦␦Ј over the primer-template. D, the same conditions as C except that the primer-template was coated with SSB. mer can be elongated significantly by and pol III only when is present (compare lanes 3 and 4). The addition of ␤ in the presence of the -clamp assembly complex facilitates elongation in the presence and absence (compare lanes 5 and 6) of SSB on the DNA. In the presence of ␤ and absence of the -complex, no elongation is observed in the presence of SSB (lane 7).
The experiments presented in Fig. 7 prompted us to determine the minimum requirements necessary in order to overcome the SSB inhibition of pol III. We end-labeled the 52-mer with 32 P and incorporated unlabeled dNTPs with various combinations of holoenzyme subunits (Fig. 8). The trend seen in the experiment reported in Fig. 7 was also observed here; pol III is inhibited by SSB on the primer-template (Fig. 8, compare lanes  1 and 2). and by themselves (lanes 3 and 4) produce replication products identical to those generated by the inhibited pol III core and do not support full-length product formation. The addition of ␦␦Ј to produces replication products identical to alone (lane 6). The concerted action of both and allows full-length primer extension (lane 5). Additionally, we sought to determine if 's ability to interact with the core polymerase played a role in the elongation process. In comparing the ability of theand ␥-complex to facilitate the primer elongation we found that the ␥-subunit is unable to facilitate primer extension in the presence of (compare lanes 7 and 8).
The presence of rescues ␥ in the presence of and permits pol III to replicate through SSB to full-length product (lane 9). Quantitation of the full-length product arising from each subassembly reaction was done by integrating each lane and examining the resultant percentages (Table I). In the presence of pol III and SSB, 100% full-length product formation occurs when, minimally, both and are present. DISCUSSION The interaction of a single-stranded DNA-binding protein with its cognate polymerase has been well documented (22,24,27,28,36,37). The benefits arising from this interaction are FIG. 6. mediates a strong interaction of the ␥-complex with an SSB-coated primer-template. The experiments shown in Fig. 4 were repeated identically for ␥␦␦Ј and ␥␦␦Ј on the F52/102-mer 3 primer-template in the presence and absence of SSB. A, 400 nM ␥␦␦Ј over a primer-template with no SSB. B, the same conditions as in A but with an SSB-coated primer-template. C, 400 nM ␥␦␦Ј over the primer-template in the absence of SSB. D, the same conditions as in C but with an SSB-coated primer-template. Proteins were diluted in HKGM plus 100 M ATP. Injection volumes were 35 l at a flow rate of 5 l/min. obvious, since any interaction promoting the co-localization of an enzyme and its substrate is probably a favorable one. The observations that pol III* and not pol IIIЈ is stimulated by SSB (18) and that ␥-complex binds to single-stranded DNA coated with SSB, but not to uncoated single-stranded DNA (29), suggest that a component of the ␥-complex (␥, ␦, ␦Ј, ␦, , ) binds to E. coli SSB.
We directly demonstrate, by gel filtration and surface plasmon resonance, that and no other holoenzyme component forms a tight complex with SSB. Thus, the SSB stimulation of higher forms of DNA polymerase III probably arises from this interaction. The DnaX gene products form a complex with (17,38). SSB binds to and gel filters with whether it exists alone or in a protein complex. Thus, can bind to SSB in the presence of its partners in the DnaX complex and holoenzyme. This shows that binding of SSB and DnaX are not mutually exclusive, consistent with an SSB-interaction during holoenzyme initiation complex assembly and elongation.
The interaction between the DnaX complex subunits and SSB prompted us to investigate whether this interaction could strengthen the interaction between the complex and the DNA. The -SSB interaction brings together entities that must interact for both clamp loading and replication to occur. Consid-ering the relatively high equilibrium constant for the proteins in the absence of DNA, and SSB probably do not interact significantly free in the cell. It is only when SSB is bound to the DNA primer-template and the clamp loader associates with the primer terminus that the -SSB interaction occurs. The apparent K D for the -SSB interaction is significantly (approximately 1000-fold) lower under conditions where a DNA template and auxiliary subunits are present. The relatively low affinity of free SSB for probably prevents SSB from acting as a competititve inhibitor of -SSB-DNA binding, enabling holoenzyme to be targeted to the replication fork.
We have demonstrated that the DnaX complex binds to SSB via on an SSB-coated primer-template. The -complex binds to the SSB/DNA with nearly twice the stoichiometry of the ␥-complex, suggesting perhaps that the C-terminal portion of provides some added stability over ␥, enabling interactions away from the primer terminus. Further experiments will be necessary in order to resolve the reason for the higher stoichiometry.
The ability of the -SSB interaction to link a polymerase subassembly and a DNA template led us to investigate its functional role in replication. The -SSB interaction enables pol III* and not pol III or pol IIIЈ to replicate through an SSB-coated DNA template. The minimal assembly that can replicate on an SSB template is pol III--. ␥, which can interact with , but not pol III, is unable to replace in the assembly, consistent with needing to be held in the proximity of the polymerase to stimulate it during the replicative reaction. This provides further evidence for our observations that the DnaX complex does not cycle during replication and is a required component for elongation.
In a previous study, we found that relieved the salt inhibition of a -reconstituted holoenzyme and not a ␥-reconstituted one that lacked (17). Here, we reproduced this effect and added the finding that salt resistance occurs only when and SSB are both present. The most direct explanation that is consistent with all of our observations is that interacts with SSB during replication, stabilizing the complex, and that at elevated salt, when electrostatic interactions are diminished, this interaction becomes dominant in stabilizing polymerasetemplate-primer interactions. It would follow that a significant portion of the -SSB binding energy derives from hydrophobic interactions. In a review article, O'Donnell and colleagues (38) refer to an unpublished observation that and SSB interact. This finding would extend our own to identifying the component of the pair as the interacting subunit.
Our results show that is important for stabilizing and interaction with a primer-template in the presence of SSB, especially at elevated salt, but show that the interaction is not required for efficient replication, at least on single-stranded templates. If ␤ is loaded by a DnaX complex lacking , efficient elongation can still be achieved on an SSB-coated template. This would suggest that a signal that triggers SSB to release during replication resides in a component other than , perhaps the polymerase itself.
Since there is only one copy of /dimeric DNA polymerase III holoenzyme, probably functions in the lagging strand half, stabilizing interactions with the SSB-coated template during elongation and in establishing new initiation complexes on nascent primers.
Note Added in Proof-After this manuscript was prepared for publication, a report appeared (Kelman, Z., Yuzhakov, A., Andjelkovic, J., and O'Donnell, M. (1998) EMBO J. 17, 2436 -2449) that overlapped with a subset of our findings. Based on the behavior of reactions reconstituted with a mutant SSB, it was inferred that an SSB-interaction was responsible for the observed salt resistance of elongation com-FIG. 8. Minimal holoenzyme subunits required for overcoming SSB inhibition. Amounts of subunits used in the various experiments were 3.2 pmol of , ␦, ␦Ј, , and ␥; 12 pmol of core; and 2 pmol of SSB 4 . One pmol of the 5Ј-32 P-labeled 52/102-mer was elongated in an assay containing 18 M TTP and 48 M dATP, dCTP, and dGTP in enzyme dilution buffer (50 mM HEPES-KOH, pH 7.5, 10% (v/v) glycerol, 0.1 M potassium glutamate, 10 mM dithiothreitol, 10 mM magnesium acetate, 200 g/ml bovine serum albumin, 0.02% (v/v) Tween 20) and incubated at 30°C for 5 min. Here, P refers to 5Ј-32 P-labeled 52-mer, while M refers to molecular weight standards. Replication products were separated on a 12% denaturing polyacrylamide gel and visualized on a Molecular Dynamics PhosphorImager.  Fig. 8. b Percentage of elongation was determined by dividing the percentage of fully elongated product from each subassembly by the percentage of fully elongated product from pol III alone.