Mutations in the N-terminal Cooperativity Domain of Gene 32 Protein Alter Properties of the T4 DNA Replication and Recombination Systems*

The gene 32 protein (gp32) of bacteriophage T4 is the essential single-stranded DNA (ssDNA)-binding protein required for phage DNA replication and recombination. gp32 binds ssDNA with high affinity and cooperativity, forming contiguous clusters that optimally configure the ssDNA for recognition by DNA polymerase or recombination enzymes. The precise roles of gp32 affinity and cooperativity in promoting replication and recombination have yet to be defined, however. Previous work established that the N-terminal “B-domain” of gp32 is essential for cooperativity and that point mutations at Arg4 and Lys3 positions have varying and dramatic effects on gp32-ssDNA interactions. Therefore, we examined the effects of six different gp32 B-domain mutants on T4 in vitro systems for DNA synthesis and homologous pairing. We find that the B-domain is essential for gp32's stimulation of these reactions. The stimulatory efficacy of gp32 B-domain mutants generally correlates with the hierarchy of relative ssDNA binding affinities,i.e. wild-type gp32 ≈ R4K > K3A ≈ R4Q > R4T > R4G ≫ gp32-B. However, the functional defect of a particular mutant is often greater than can be explained simply by its ability to saturate the ssDNA at equilibrium, suggesting additional defects in the proper assembly and activity of DNA polymerase and recombinase complexes on ssDNA, which may derive from a decreased lifetime of gp32-ssDNA clusters.

The gene 32 protein (gp32) of bacteriophage T4 is the essential single-stranded DNA (ssDNA)-binding protein required for phage DNA replication and recombination. gp32 binds ssDNA with high affinity and cooperativity, forming contiguous clusters that optimally configure the ssDNA for recognition by DNA polymerase or recombination enzymes. The precise roles of gp32 affinity and cooperativity in promoting replication and recombination have yet to be defined, however. Previous work established that the N-terminal "B-domain" of gp32 is essential for cooperativity and that point mutations at Arg 4 and Lys 3 positions have varying and dramatic effects on gp32-ssDNA interactions. Therefore, we examined the effects of six different gp32 B-domain mutants on T4 in vitro systems for DNA synthesis and homologous pairing. We find that the B-domain is essential for gp32's stimulation of these reactions. The stimulatory efficacy of gp32 B-domain mutants generally correlates with the hierarchy of relative ssDNA binding affinities, i.e. wild-type gp32 Ϸ R4K > K3A Ϸ R4Q > R4T > R4G > > gp32-B. However, the functional defect of a particular mutant is often greater than can be explained simply by its ability to saturate the ssDNA at equilibrium, suggesting additional defects in the proper assembly and activity of DNA polymerase and recombinase complexes on ssDNA, which may derive from a decreased lifetime of gp32-ssDNA clusters.
The gene 32 protein (gp32) 1 of bacteriophage T4 serves as the paradigm of sequence-nonspecific single-stranded DNA-binding proteins or SSBs, which as a class play vital roles in DNA replication, recombination, and repair processes in virtually all organisms (1). The interactions of gp32 with polynucleotides are characterized by its pronounced preference for singlestranded versus double-stranded DNA and by the high cooperativity of gp32-ssDNA interactions. The high cooperativity of binding by monomeric gp32 results in the formation of protein clusters on regions of ssDNA that form transiently during DNA replication, recombination, and repair. In addition to protecting ssDNA from intracellular nucleases, cooperative binding by gp32 appears to be a prerequisite for removing ssDNA secondary structure and for inducing other DNA structural changes necessary for optimal recognition of the ssDNA by T4 enzymes, including DNA polymerase (gp43), and for stimulating homologous recombination by binding to the displaced strand during DNA strand exchange (1)(2)(3).
Native gp32 consists of a single polypeptide chain of 301 amino acid residues (34 kDa). Early studies revealed that the protein contains three distinct structural and functional domains, which map to specific regions of the primary structure of gp32 (4). The ssDNA binding core domain spans residues 22-253, and contains an intrinsic Zn(II) ion (5). The structure of the core domain co-crystallized with oligo(dT) 6 has been determined to 2.2-Å resolution (6). Appended to the core domain at its N and C termini are the so-called "B" or basic domain (residues 1-21), and the "A" or acidic domain (residues 254 -301), respectively, neither of which has been characterized structurally. The A-domain is known to be important for heterologous protein-protein interactions with several T4 DNA replication and recombination proteins, including gp43 (DNA polymerase), gp61 (primase), UvsX (general recombinase), the recombination mediator protein UvsY, the primosome assembly factor gp59, and the nonessential DNA helicase Dda (7)(8)(9). The B-domain contains elements essential for the cooperativity of gp32-ssDNA interactions. Characterization of gp32-B (gp32 residues 22-301), a truncated form of gp32 lacking the Nterminal B-domain, reveals that this protein binds noncooperatively to ssDNA and also fails to exhibit self-association in solution (10), consistent with a model in which major stabilization of the interactions between contiguously ssDNA-bound gp32 molecules occurs via the B-domain.
A more recent characterization of gp32 species that contain specific amino acid substitutions within the B-domain has revealed unexpected complexity in this simple model, while identifying some of the residues required for high affinity, cooperative binding to single-stranded polynucleotides (11)(12)(13). Of central importance is Arg 4 , the positive charge of which appears necessary both for the cooperativity of gp32-ssDNA binding and for the helix-destabilizing activity of gp32 (11). Four gp32 species containing different amino acid substitutions at this position (the R4K, R4Q, R4T, and R4G derivatives) have been shown to vary widely in their intrinsic affinity and cooperativity of binding to model single-stranded polynucleotides (11). In addition, K3A gp32 has also been characterized (12), as has the truncated species gp32-B (10). Experimentally determined binding parameters for each of these gp32 species to a poly(A) lattice are listed in Table I, which shows that the apparent binding constant (K app ) value decreases according to the following hierarchy: wild-type gp32 Ϸ R4K Ͼ K3A Ϸ R4Q Ͼ R4T Ͼ R4G Ͼ Ͼ gp32-B (10 -12). Surprisingly, both the cooperativity () and intrinsic affinity (K int ) constants appear to be adversely affected by nonconservative substitution of Lys 3 and Arg 4 , with primary perturbations in K int (Table I), suggesting that these mutations affect gp32-ssDNA stability globally (11)(12). Consistent with this, mechanistic studies reveal that virtually all of the perturbation in equilibrium affinity originates with an enhanced rate at which cooperatively bound mutant gp32 monomers dissociate from the ends of protein clusters (13). These studies reveal that the N-terminal B-domain plays a critical role in modulating the lifetime of cooperatively bound gp32-polynucleotide complexes.
A thorough understanding of the functional ramifications of B-domain mutations requires an assessment of their effects on the DNA replication and recombination processes in which gp32 participates. Under defined conditions in vitro, homologous pairing reactions catalyzed by the T4 UvsX recombinase require gp32 as a nearly essential cofactor and are highly sensitive to gp32 concentration (3). Likewise, there are several in vitro DNA synthesis reactions catalyzed by the T4 DNA polymerase holoenzyme (gp43 polymerase plus accessory proteins gp44/62 and gp45) that absolutely depend on the presence of gp32. One of these is strand displacement synthesis on a nicked duplex template, which, in the absence of the gp41 DNA helicase, requires gp32 in approximately stoichiometric amounts with respect to the DNA and in large excess over other components of the T4 replication fork (14). gp32 is also required for DNA synthesis past a stable hairpin in a single-stranded DNA template (15).
Due to the strong dependence of each of these three in vitro reactions on gp32 concentration, they offer an unparalleled opportunity to correlate well defined physicochemical defects in mutant gp32-ssDNA complexes with defects in the nonequilibrium processes of DNA replication and recombination. Accordingly, we have studied the extent to which each of the six gp32 B-domain mutants previously characterized stimulate homologous pairing, strand displacement DNA synthesis, and transhairpin DNA synthesis. Our results demonstrate that the Bdomain of gp32 is essential for the stimulatory effects of gp32 on these in vitro reactions. Specific reductions in recombination and replication functional efficacy exhibited by individual mutants correlate, to a first approximation, with their relative affinities for ssDNA at equilibrium. However, some mutants are more defective in replication and recombination assays than would be predicted simply on the basis of their ability or inability to saturate ssDNA. Instead, the data suggest that these mutations give rise to gp32-ssDNA complexes that are kinetically compromised in their functional stability and/or defective in establishing important interactions between gp32 and the DNA polymerase and recombinase enzymes. T4 DNA Replication and Recombination Proteins-Purification and storage conditions for T4 proteins including gp43 (DNA polymerase), gp44/62 and gp45 (DNA polymerase processivity factors), and gp32 (ssDNA-binding protein) were as described previously (16,17). gp32 mutant species R4K, R4Q, R4T, R4G, K3A, and gp32-B were also purified and stored according to published procedures (10 -13). The concentrations of gp32 wild-type and B-domain mutant species were determined by the absorbance at 280 nm using an extinction coefficient of ⑀ 280 ϭ 4.13 ϫ 10 4 M Ϫ1 cm Ϫ1 determined from the amino acid sequence (18). The concentrations of other T4 protein stock solutions were determined by a spectrophotometric Bradford assay using bovine serum albumin standards. All purified protein stocks were tested for contaminating endo-or exonuclease activities by incubating protein (at reaction concentrations) with circular plus linear ssDNA, supercoiled plus nicked circular dsDNA, or linear dsDNA samples as appropriate and then examining the DNA on agarose gels. All protein stocks used in these studies were nuclease-free according to these criteria.

Reagents and Enzymes-Radionuclides
Nucleic Acids-Supercoiled RFI DNA from bacteriophage M13mp19 was isolated as described (19). Linearized RFIII dsDNA was produced by treating RFI with the SmaI restriction enzyme and then 3Ј-endlabeled with [␣-32 P]dCTP via the T4 DNA polymerase reaction (20). Nicked RFII DNA was generated by treating M13mp19 RFI DNA with the bacteriophage fd gene 2 protein as described (21). Circular singlestranded DNA from bacteriophages M13mp19 and M13mp4 (15) was isolated by extraction from purified phage particles (19,22). Oligonucleotide primers used for trans-hairpin DNA replication experiments and controls on primed templates were purchased from Operon Technologies, Inc. Primer molecules were 5Ј-32 P-labeled using T4 polynucleotide kinase as described (20). All DNA concentrations were determined by the absorbance at 260 nm using conversion factors of 50 g/ml/A 260 for dsDNA and 36 g/ml/A 260 for ssDNA and are expressed as mol of nucleotide residues/liter except as noted.
Homologous Pairing Assays-UvsX-catalyzed homologous pairing reactions were carried out essentially as described previously (23,24). Recombination buffer contained final concentrations of 10 mM Tris acetate, pH 7.4, 90 mM potassium acetate, 10 mM magnesium acetate, 10 mM creatine phosphate, 10 g/ml creatine phosphokinase, and 1 mM 2-mercaptoethanol. Each 40-l reaction mixture contained 1 M (40 g/ml) UvsX protein, 9.2 M (3 g/ml) 3Ј-32 P-labeled M13mp19 RFIII DNA, 15.4 M (5 g/ml) M13mp19 ssDNA circles, and concentrations of gp32 wild-type or mutant species as indicated. Each reaction mixture was preincubated for 10 min at 37°C, and then 2 mM ATP (final concentration) was added to initiate the reaction. The reaction was quenched after 15 min by the addition of a quench buffer to bring the final solution concentrations of EDTA and SDS to 20 mM and 0.5%, respectively. The samples were then electrophoresed on a 0.5% agarose gel in TBE buffer (89 mM Tris base, 89 mM boric acid, pH 7.5, 2 mM EDTA). The recombination products were visualized by autoradiography of the dried gel.
Homologous pairing experiments performed to study the time dependence of the pairing reaction were carried out similarly at a constant protein concentration (3 M) of wild-type or R4Q gp32. For each reaction, all constituents except ATP were combined, followed by preincu- bation for 10 min at 37°C before an aliquot corresponding to time 0 was withdrawn. 2 mM ATP was then added to initiate the reaction, and aliquots were withdrawn at the times indicated and quenched with 20 mM EDTA and 0.5% SDS.
Agarose Gel Mobility Shift Assays-Protein-ssDNA complexes were formed by incubating 15.4 M (nucleotides) M13mp19 ssDNA with an indicated concentration of wild-type or mutant gp32 for 10 min at 37°C in 40 l (final volume) of recombination buffer. 5 l of gel loading buffer (50% glycerol, 0.04% bromphenol blue) was added to each sample; samples were then electrophoresed on a 0.5% agarose gel in low salt buffer (20 mM Tris, pH 7.8, 0.4 mM sodium acetate, and 0.2 mM Na 3 EDTA) at a constant voltage of 7 V/cm with continuous buffer recirculation (25). The protein-ssDNA complexes contained in the gel were dissociated by soaking the gel for 1-2 h in high salt buffer (10 mM Tris, pH 7.8, 0.1 mM Na 3 EDTA, 1 M NaCl) in the presence of Sybergreen II RNA gel stain, which was found to stain ssDNA more efficiently than ethidium bromide under these conditions. Normalized relative mobility values were calculated from the data by dividing the relative mobility of a protein-ssDNA complex band by the relative mobility of the saturated ssDNA-wild-type gp32 complex. The relative mobility was measured as the distance of migration of a protein-ssDNA complex from the middle of the free ssDNA band to the middle of the most intense region of the protein-ssDNA complex band.
Strand Displacement DNA Synthesis Reactions-Strand displacement DNA synthesis reactions were carried out at 37°C in reaction volumes of 25 l. Reaction mixtures contained the following components (final concentrations): 20 mM Tris acetate, pH 7.4, 80 mM potassium acetate, 10 mM magnesium acetate, 100 g/ml bovine serum albumin, 0.5 mM dithiothreitol, 10 g/ml creatine phosphokinase, 10 mM creatine phosphate, 2 mM each of ATP and GTP, 100 M each of dATP, dTTP, dGTP, and dCTP, 10 Ci of [␣-32 P]dTTP (specific activity ϭ 3000 Ci/mmol), and 10 M (nucleotides) M13mp19 RFII DNA, plus T4 proteins (4 g/ml gp43, 24 g/ml gp44/62, 8 g/ml gp45, and variable amounts of wild-type or mutant gp32 species as indicated in the figure legends). For each reaction, all reaction components except DNA, dNTPs, and [␣-32 P]dTTP were preincubated for 3 min at 37°C, and then the reaction was initiated by the addition of the missing components. Reactions were allowed to proceed for 10 min, at which time they were stopped by the addition of 1.5 l of 0.5 M EDTA to each tube (yielding a final EDTA concentration of 28 mM) and transferred to an ice-water bath. The DNA synthesis that occurred in each sample was evaluated quantitatively by trichloroacetic acid precipitation and/or qualitatively by electrophoresis on 0.8% alkaline agarose gels. For trichloroacetic acid precipitation, 15 l of each stopped reaction sample was spotted onto a Whatman GF/A glass filter; these were soaked for 15 min in 250 ml of cold 5% trichloroacetic acid, 10% saturated sodium pyrophosphate solution at 4°C and then washed successively with 4 ϫ 250 ml of cold 1 M HCl and 2 ϫ 250 ml of cold 95% ethanol, dried under a heat lamp, and transferred to a scintillation counter. For alkaline agarose gel electrophoresis, the remainder of each sample (11.5 l) was brought to final concentrations of 10% sucrose, 30 mM NaOH, and 0.04% bromcresol green, respectively, by adding 3.5 l of a 4ϫ concentrated solution of these gel loading buffer components. Samples were loaded onto a horizontal 0.8% alkaline agarose gel, which was prepared and run as described (26). Following electrophoresis, gels were neutralized by soaking for 30 min in TBE (89 mM Tris borate, 89 mM boric acid, 2 mM EDTA, pH 7.5) and then stained in TBE containing 1 g/ml ethidium bromide. Stained gels were photographed on a UV light box to record marker positions and then dried under vacuum onto Whatman DE81 DEAE-cellulose paper. Dried gels were autoradiographed at Ϫ70°C using Kodak XAR-5 x-ray film.
Trans-hairpin DNA Replication on Primed Templates-Primer 1 is a 5Ј-32 P-labeled 20-mer of sequence 5Ј-CGATTAAGTTGGGTAACGCC-3Ј, which anneals to the M13mp4 ssDNA template with its 3Ј-end 40 bases upstream of the stable 15-base pair hairpin described by Hacker and Alberts (15). Primer 2 is a 5Ј-32 P-labeled 20-mer of sequence 5Ј-TC-CGCTCACAATTCCACACA-3Ј, which anneals to M13mp4 ssDNA with its 3Ј-end 60 bases downstream of the hairpin. Annealing conditions were as follows: 0.2 pmol (molecules) of either primer 1 or 2, depending on the experiment, or control was mixed with 0.2 pmol (molecules) of M13mp4 ssDNA circles in 20 l of annealing buffer (20 mM Tris acetate, pH 7.4, 80 mM potassium acetate, 10 mM magnesium acetate). The mixture was heated to 75°C for 4 min and then slowly cooled to 25°C, forming a concentrated stock solution of 75 M (nucleotides) annealed primer-template. Replication reactions were carried out at 25°C, in reaction volumes of 15 l. Reaction mixtures contained (final concentrations) 20 mM Tris acetate, pH 7.4, 80 mM potassium acetate, 10 mM magnesium acetate, 0.5 mM dithiothreitol, 100 g/ml bovine serum albumin, 5 M (nucleotides) annealed primer-template, 2 mM ATP, 100 M each of dATP, dTTP, dGTP, and dCTP, 0.04 g/ml gp43, 24 g/ml gp44/62, 8 g/ml gp45, and variable amounts of wild-type or mutant gp32 species as indicated in the figure legends. Reaction mixtures containing all components except primer-template were preincubated for 2 min at 25°C, and then reactions were initiated by the addition of primer-template. Reactions at 25°C were stopped after 1 min by the addition of EDTA to 30 mM final concentration and transferred to an ice bath. Samples were phenol/chloroform-extracted once to remove protein, passed through a Sephadex spin cartridge to remove salt, dried in a speed-vac, dissolved in 5 l of sequencing gel loading buffer (98% deionized formamide, 10 mM EDTA, 0.025% xylene cyanol, 0.025% bromphenol blue), boiled for 4 min, and then rapidly transferred to an ice-water bath. Samples were loaded onto a 12% polyacrylamide/urea denaturing gel and electrophoresed at 500 V. The gel was dried and autoradiographed as described above.

RESULTS
The Efficiency of gp32 B-domain Mutants to Stimulate UvsXcatalyzed Homologous Pairing-The efficiency of each B-domain mutant to stimulate the homologous pairing reaction catalyzed by UvsX protein was monitored by observing the conversion of 32 P-labeled linear dsDNA (RFIII) into high molecular weight aggregates as evidenced by agarose gel electrophoresis (24). Since UvsX protein alone can catalyze this reaction at saturating concentrations, care was taken in choosing the conditions of the experiment so that pairing would be co-dependent on UvsX and gp32. These conditions, identified previously (23), require 1 M UvsX protein, along with 9.2 M linear duplex and 15.4 M ssDNA circles in each reaction with an indicated amount of gp32. When reactions are performed under these conditions in the presence of wild-type gp32, autoradiography of the deproteinized samples reveals a gp32 concentration-dependent disappearance of the labeled duplex with the simultaneous appearance of a diffuse, slower migrating band (Fig. 1A). This diffuse or smeared band is subsequently converted into a band consisting of high molecular weight aggregates that do not enter the gel (Fig. 1A). The proficiency with which each gp32 derivative assists in this reaction provides an indication of its relative ability to function as an accessory protein in recombination events.
For ease of comparison between wild-type and mutant gp32 species, homologous pairing at the fixed UvsX concentration of 1 M was carried out over similar concentration ranges of each gp32 derivative, except where a higher mutant gp32 concentration was required to effect homologous pairing (Fig. 1). A reaction time of 15 min was previously shown to be sufficient for complete conversion of linear duplex to aggregates with wild-type gp32 (23). However, to ensure that results found for some gp32 mutants were not kinetically determined, a systematic variation of the reaction time as compared with wild-type gp32 was carried out with modestly defective R4Q gp32 (K app is reduced 6 -8-fold depending on the salt concentration) at a constant protein concentration of 3 M (Fig. 2). From these experiments, it can be concluded that the distribution of high molecular weight aggregates formed as products of the homologous pairing reaction changes little if at all after 15 min for both wild-type and R4Q gp32s. Therefore, a reaction time of 15 min appears sufficient for the observable pairing reaction to reach completion and was used in all subsequent experiments. Fig. 1, A-G, shows the results of homologous pairing assays carried out with wild-type gp32 and with each B-domain mutant species (R4K, R4Q, K3A, R4T, and R4G gp32s and gp32-B). In order to quantitate the differences between gp32 species, we compare the lowest gp32 species concentration(s) in each panel at which all of the input linear dsDNA appears to be converted into high molecular weight aggregates that do not enter the gel. It appears from these data that 3 M wild-type gp32 (Fig. 1A) is sufficient for the conversion of all of the input gp32 Mutants in T4 DNA Replication and Recombination labeled duplex into high molecular weight aggregates in the UvsX-catalyzed reaction. R4K gp32 appears to be slightly more efficient than wild-type gp32, since complete conversion of duplex to aggregates is seen at 2 M R4K (Fig. 1B). R4Q and K3A gp32s (Fig. 1, C and D) show similar results when compared with each other, consistent with their roughly equivalent equilibrium binding affinities for polynucleotides (12). However, neither R4Q nor K3A gp32s are as efficient in stimulating homologous pairing as wild-type and R4K gp32s. Complete conversion of linear duplex to aggregates remaining in the well is observed at 2 and 3 M R4K and wild-type gp32s, respectively; in contrast, R4Q and K3A gp32s require at least 6 and 5 M concentrations, respectively, for complete conversion (Fig.  1, C and D). R4T gp32 (Fig. 1E) is clearly more deficient in stimulating homologous pairing than wild-type, R4K, R4Q, and K3A gp32s, requiring approximately 9 -10 M R4T for aggregate formation to reach completion. Results with R4G gp32 (Fig. 1F) are even more dramatic, since complete conversion of linear duplex to aggregates is not observed even at the highest R4G concentration tested (10 M). Finally, gp32-B fails to stimulate UvsX-catalyzed homologous pairing at any of the concentrations tested (Fig. 1G). 2 Differential Abilities of gp32 B-domain Mutants to Saturate ssDNA at Equilibrium-Results obtained in the homologous pairing assays (Figs. 1 and 2) show that overall, the functional efficiencies of gp32 B-domain mutants correlate well with the energetics of ssDNA binding described for these species previously (10 -12); i.e. wild-type gp32 Ϸ R4K Ͼ R4Q Ϸ K3A Ͼ R4T Ͼ R4G Ͼ Ͼ gp32-B. This raises the possibility that the ability or inability of a gp32 species to saturate ssDNA under the homologous pairing assay conditions may dictate its efficiency as an accessory protein for homologous pairing. To investigate this possibility, gel mobility shift assays were performed, in which wild-type and mutant gp32-M13mp9 ssDNA complexes were formed in recombination buffer and then electrophoresed on a nondenaturing agarose gel. To facilitate a direct comparison, the same gp32 concentration ranges tested in the homologous pairing reactions were also used for the gel mobility shift assays. These results are shown in Fig. 3. Fig. 3A shows that wild-type gp32 binds nearly stoichiometrically to the M13mp19 ssDNA under these conditions, since 2 Note that there is a high background of "aggregates" seen in all lanes of Fig. 1, F and G. These aggregates were a property of the 32 P-labeled linear duplex used in experiments with the R4G and gp32-B species, which was a different preparation from that used in Fig. 1, A-E. Aggregates in Fig. 1, F-G, were observed in the absence as well as the presence of R4G or gp32-B and were also independent of UvsX protein (J. L. Villemain, D. P. Giedroc, unpublished results). Nevertheless, failure of the R4G and gp32-B species to efficiently stimulate the UvsXcatalyzed homologous strand pairing reaction is evident from the persistence of the input linear duplex DNA at the highest concentrations of mutant gp32 proteins tested (up to 10 M). complete complex formation is observed at ϳ2.5 M protein, equivalent to 6.2 nucleotide residues/gp32 monomer. The previously published binding site size of gp32 on single-stranded nucleic acids is ϳ7 nucleotide residues (27). Interestingly, R4K gp32 appears to completely saturate the ssDNA at slightly lower protein concentrations (ϳ2 M); this may explain why R4K gp32 appears to be slightly more effective than wild-type gp32 in stimulating the homologous pairing reaction (see Fig.  1). The K3A and R4Q gp32 complexes (Fig. 3, C and D) appear fully saturated at approximately stoichiometric protein concentrations as well (2-2.5 M), signifying tight binding. Interestingly, these complexes appear to migrate slightly faster than the saturated wild-type gp32 complex (e.g. compare the last two lanes of Fig. 3, A, C, and D), which might be reporting on slightly different structures formed by mutant gp32-ssDNA complexes. The R4T (Fig. 3E) mutant exhibits markedly weaker ssDNA binding activity under these solution conditions than do wild-type, R4K, K3A, and R4Q gp32s. Saturation appears to require at least 5 M R4T, and the saturated complex migrates markedly faster than the equivalent complex formed with wild-type gp32 (compare last two lanes of Fig. 3E). The binding of mutants R4G and gp32-B (Fig. 3, F-G) is clearly far weaker than that of the other species under recombination solution conditions. The R4G mutant barely approaches saturation of the ssDNA at the highest protein concentration tested (11 M), whereas 11 M gp32-B yields very poor binding and is far from saturation.
Is the efficiency of stimulation of homologous pairing directly related to the ability of a gp32 species to form a saturated complex with the ssDNA substrate? When the data from Fig. 3 are plotted as relative mobility of the protein-ssDNA complex versus total protein concentration (Fig. 4), it becomes apparent that for some mutant gp32 species the ability or inability to saturate the ssDNA substrate does not account for the entire deficiency in stimulating homologous pairing. The true situation is more complex. For example, the R4Q and K3A gp32s in particular do not show significantly lower saturation of the ssDNA relative to wild-type and R4K gp32s under these solu-tion conditions; however, complete formation of aggregates in the homologous pairing experiment requires significantly (approximately 2-fold) higher concentrations of these mutants to drive the reaction to completion (see Fig. 1). Complete conversion of labeled duplex to aggregates is observed at 2-3 M R4K or wild-type gp32s (Fig. 1, A and B), which corresponds to an ssDNA saturation level of ϳ90% according to Fig. 4. With the R4Q and K3A mutants, however, a significant concentration of recombination intermediates remain in the pairing reactions for these proteins at 90% saturation at equilibrium (compare Fig. 4 with Fig. 1, C and D). The efficiency of homologous pairing in reactions involving R4T and R4G gp32s also appears lower than can be accounted for by a simple reduction in equilibrium binding affinity toward the ssDNA substrate. For example, R4G gp32 reaches ϳ90% saturation of the ssDNA substrate at a concentration of 9 -10 M (Figs. 3 and 4), yet homologous pairing activity is weak at these concentrations, and complete conversion of linear duplex to high molecular weight aggregates is not observed (Fig. 1F). gp32-B, as expected, does not approach saturation of the ssDNA even at a concentration of 11 M (Fig. 4) and does not detectably stimulate homologous pairing (Fig. 1G).
Abilities of gp32 Mutants to Facilitate Strand Displacement DNA Synthesis on a Nicked Template-In the absence of the T4-encoded gp41 DNA helicase, strand displacement DNA synthesis reactions catalyzed by the T4 DNA polymerase holoenzyme (gp43 plus gp44/62 and gp45) on nicked templates in vitro absolutely require gp32 (14). Extensive strand displacement synthesis requires high concentrations of gp32, roughly stoichiometric to the amounts of ssDNA produced and in large excess over holoenzyme components. Thus, it appears likely that the ability of gp32 to saturate and thus sequester the displaced strand is important for stimulation of strand displacement synthesis.
Strand displacement DNA synthesis reactions were carried out using a specifically nicked M13mp19 dsDNA template (RFII). DNA synthesis was measured by incorporating [␣-32 P]dTTP, allowing quantitative analysis of acid-insoluble material retained on glass filters and qualitative analysis of reaction products on denaturing alkaline agarose gels. Except for the presence of the nucleoside triphosphates, the solution conditions are otherwise very similar to those of the homologous pairing and gel mobility shift assays (see "Materials and Methods"); therefore, the extent of ssDNA binding by different gp32 species should be comparable in all three experiments. Although the concentration of ssDNA in this experiment is initially zero, 1.45 M gp32 was present in each reaction, which is sufficient to cover the ssDNA generated from, on average, two complete rounds of rolling circle DNA synthesis from an initial concentration of 10 M (nucleotides) RFII template (assuming n ϭ 7). The results from these experiments are shown in Table II.
The data in Table II show that strand displacement DNA synthesis strongly requires active gp32. Here, the low level of background label incorporation is due to a "chew-back/fill-in" synthesis reaction catalyzed by the T4 DNA polymerase holoenzyme at the nick in the template, not strand displacement synthesis. As expected, R4K gp32 supports wild-type levels of DNA synthesis. In striking contrast, the R4Q and K3A gp32s are only about 10 -20% (corrected for background) as efficient as wild-type and R4K gp32s in promoting DNA synthesis. Worse still are the R4T, R4G, and gp32-B mutant species, each of which shows incorporation levels only modestly above background (Table II), representing only 3-5% of the activity (corrected for background) seen with wild-type and R4K gp32. The product size distributions of replication products from identical reactions were analyzed on alkaline agarose gels (Fig. 5). At the 1.45 M concentration level, the R4K mutant clearly supports wild-type rates and processivities of replication fork movement, since long products indistinguishable from that of wild-type gp32 are observed (Fig. 5A). However, reactions containing 1.45 M of either R4Q or K3A gp32 produced dramatically shorter products, with R4Q gp32 less efficient than K3A gp32 (Fig. 5B). Although we cannot conclude from this assay whether the R4Q and K3A substitutions reduce the number of replication fork initiations, it is clear that each mutation dramatically lowers the rate and/or net processivity of replication fork movement, since those forks that do initiate clearly do not travel as far. Thus, the low incorporation levels observed with R4Q and K3A gp32s under these conditions (Table II) are not merely due to fewer initiation events. In the reactions containing a 1.45 M concentration of either R4T, R4G, or gp32-B mutant, no strand displacement DNA synthesis products are visible (Fig. 5B). Doubling the input concentrations of various gp32s to 2.90 M leads to qualitatively the same conclusions (Fig. 5B). For example, although doubling the concentration of R4Q and K3A gp32s leads to a significant increase in the average product length, the product size distribution is still clearly smaller than those obtained for wild-type and R4K gp32s at the lower concentration. Doubling the R4T gp32 concentration gives rise to only very short strand displacement synthesis products, which run slightly more slowly than the chew-back/fill-in product band. Doubling the concentrations of R4G gp32 and gp32-B has no effect on visible product distributions. These results suggest that R4G gp32 and gp32-B simply do not support strand displacement DNA synthesis under these conditions, although it remains possible that very high concentrations of these proteins might allow low levels of strand displacement synthesis to occur.

Abilities of gp32 Mutants to Facilitate DNA Synthesis Past a Stable Hairpin in a ssDNA
Template-Wild-type gp32 greatly enhances the replication of a primed ssDNA template by the T4 DNA polymerase holoenzyme, at least partly through the elimination of secondary structure within the ssDNA, which otherwise interferes with polymerase movement (1). Hacker and Alberts (15) reported conditions in which replication by T4 DNA polymerase holoenzyme is completely blocked by a stable hairpin structure within the ssDNA template. The addition of saturating gp32 efficiently removes the blockage, which seems to require the helix-destabilizing activity of wild-type gp32. It seems reasonable to predict that this reaction or perhaps other aspects of primer-initiated DNA synthesis might also be sensitive to mutations that lower the stability of gp32-ssDNA complexes. We therefore determined the extent to which various gp32 B-domain mutants stimulate DNA synthesis through a stable 15-base pair hairpin by the T4 holoenzyme on a primed M13mp4 ssDNA template (15).
Two replication primers were used for these experiments (see "Materials and Methods"). Primer 1 anneals upstream, while primer 2 anneals downstream of the 15-base pair hairpin in M13mp4. The results of initiation of DNA synthesis with primers 1 and 2 are shown in Fig. 6, A and B, respectively. The 15-base pair hairpin clearly acts as a severe block to DNA synthesis by the T4 DNA polymerase holoenzyme in the absence of gp32, as demonstrated by the strong, specific pause band on the gel, corresponding to arrested synthesis at the base of the hairpin (Fig. 6A, lane 2). Essentially none of the primers that initiated DNA synthesis were elongated past the hairpin in the absence of gp32. In contrast, DNA synthesis in the presence of 1.45 M wild-type gp32 (a 2-fold excess over ssDNA binding sites) proceeds past the hairpin efficiently, yielding no pause band and long products from virtually all primers initiated (Fig. 6A, lane 3). A similar result is obtained with R4K gp32 (Fig. 6A, lane 4), which shows only a very faint pause band, consistent with its nearly wild-type behavior in other assays (see above). Both K3A and R4Q mutants support moderate levels of trans-hairpin DNA synthesis (Fig. 6A, lanes 5  and 6), although both display visible pause bands corresponding to replication arrest at the base of the hairpin. The yield of longer products is also reduced with respect to wild-type and R4K gp32s. The remaining three mutants, R4T and R4G gp32s and gp32-B, show larger deficiencies in supporting trans-hairpin DNA synthesis, with R4T gp32 more similar to than different from R4Q gp32 under these conditions (Fig. 6A, lanes 7-9).
In addition to the defect of certain gp32 mutants in transhairpin synthesis, the frequency of primer utilization in these reactions is significantly reduced as well (Fig. 6A). This phenomenon was also observed when DNA synthesis was primed with primer 2 (Fig. 6B). Both primers were extended with poor efficiency by T4 DNA polymerase holoenzyme in the presence of R4T and R4G gp32s and gp32-B, the latter of which gives similar results as the reaction with no gp32 added. Primer utilization was much more efficient with R4Q and K3A gp32s, although substantially reduced relative to wild-type and R4K gp32s. Taken collectively, the data suggest that the nonconservative B-domain mutations of gp32 strongly diminish the ability of polymerase to initiate DNA synthesis on primed templates.
Further inspection of Fig. 6B also suggests that the gp32 substitutions have a significant impact on the overall processivity of DNA synthesis by the T4 holoenzyme. For example, reactions containing the weaker binding gp32 species (R4T and R4G gp32s and gp32-B) exhibit few long replication products and many shorter synthesis products not observed in reactions with more tightly binding gp32 species. As expected, intermediate results are obtained with K3A and R4Q gp32s, which also show significant accumulations of shorter synthesis products. The data suggest that an intact gp32 B-domain is necessary both for optimal initiation of DNA synthesis on a primed ssDNA template and to maintain the processivity of polymerase movement along a ssDNA template, even in the presence of polymerase processivity factors gp45 and gp44/62, and at saturating concentrations of gp32. DISCUSSION All organisms encode SSBs to facilitate or stimulate biological processes in which single-stranded DNA is an intermediate, including DNA replication, recombination, and repair. SSBs bind preferentially to ssDNA relative to duplex DNA and can therefore stabilize ssDNA against intermolecular duplex formation, denature adventitious intramolecular secondary structure that might form in the single-stranded DNA, and/or impart a particular conformation on the ssDNA that is efficiently utilized or specifically recognized, via protein-protein interactions, by the enzymatic machinery in each case (recombinases, helicases, polymerases, etc.). Thus, SSBs play a critical accessory role in stimulating these processes. Extensive studies of the specificity and thermodynamics of singlestranded nucleic acid binding of various SSBs (25,(27)(28)(29)(30)(31)(37)(38)(39) and, more recently, x-ray crystallographic studies of SSBs with and without bound nucleic acid (6,(33)(34)(35)(36) reveal that SSBs represent a structurally diverse family of proteins that are characterized by widely differing tertiary and quaternary structures, binding site sizes, paths that the ssDNA takes on the molecule or oligomeric assembly, and degrees of cooperativity of binding. T4 gene 32 protein is relatively unusual among SSBs in that it clearly functions as a monomer, has a rather small site size (n ϭ 7 nucleotides), and binds highly cooperatively at equilibrium to single-stranded nucleic acids (29,30). Cooperative binding at equilibrium enables gp32 to form clusters of monomers at gp32 concentrations less than that required to completely saturate all of the available ssDNA binding sites (32).
The mechanism and functional significance of highly cooperative binding by any SSB is unclear. For example, in bacteriophage T4-infected Escherichia coli, sufficient gp32 is always thought to be present to fully saturate all available ssDNA binding sites (1). To address the mechanism and functional importance of cooperative binding by gp32, mutant gp32s have been created that contain a single amino acid substitution in the N-terminal cooperativity domain of gp32, with the goal to identify amino acids critical for maintaining highly cooperative binding as well as to generate select mutant gp32s with modestly altered binding properties that could be used as tools to probe these processes (11)(12)(13). Previous studies have examined the binding of B-domain mutant gp32s to the model singlestranded homopolymer, poly(A), at equilibrium as well as the kinetics of bimolecular association and salt-induced dissociation of these gp32s from poly(A) under the same solution conditions (10 -13). These studies reveal that R4K, R4Q, R4T, and K3A gp32s bind with high cooperativity and show relatively small diminutions (ϳ2-50-fold) in binding affinity to poly(A) at equilibrium (Table I); in contrast, R4G gp32 and gp32-B were found to bind weakly and nearly noncooperatively to polynucleotides. The kinetics studies quantitatively show that the entire observed defect in poly(A) binding at equilibrium is largely manifested as an increase in the rate at which mutant gp32 monomers dissociate from the ends of protein clusters (k e ) relative to wild-type gp32 (13). The bimolecular association rate constant appears unaffected (13). A short extrapolation of the salt dependence of k e determined in the original study (0.25-0.45 M NaCl) (13) to solution conditions that closely mimic those used here (0.15-0.20 M NaCl) suggests that k e is at most increased 3-fold for R4K gp32, a small effect, but may be as much as ϳ20 -50-fold faster for K3A and R4Q gp32s, with K3A gp32 forming a slightly more stable complex relative to R4Q gp32 under all conditions (13). If these results with poly(A) can be extended to other single-stranded nucleic acids, which seems likely (28), the findings reveal that even fully saturated ssDNA-gp32 complexes formed with K3A and R4Q gp32s at equilibrium will exhibit significantly shorter lifetimes under the solution conditions used here, in contrast to wildtype and R4K gp32s.
In this study, we show that the ability of mutant gp32s to stimulate homologous pairing by the recombinase T4 UvsX or DNA synthesis by the T4 DNA polymerase holoenzyme complex appears to correlate more strongly with a reduced lifetime or altered structure of mutant gp32-ssDNA complexes rather than with simple affinity for ssDNA at equilibrium. The homologous pairing assay carried out here is biologically relevant, since it is thought to faithfully mimic the first step of homologous recombination in vivo. The dependence of UvsX-catalyzed homologous pairing on gp32 is well documented (3,23), with conditions identified in which the formation of products is absolutely dependent on the concentration of gp32 present. While the precise mechanism of gp32 action in this assay is not completely clear, there is evidence to suggest that the primary means of stimulation of UvsX-catalyzed homologous pairing by gp32 involves the binding of gp32 to the displaced duplex strand (3). At suboptimal UvsX concentrations, gp32 stimulates homologous pairing at approximately stoichiometric concentrations with respect to the ssDNA substrate, suggesting that saturation of the ssDNA by gp32 is an important component of the stimulation of this activity by gp32.
Wild-type and R4K gp32s are most efficient at stimulating UvsX-catalyzed homologous pairing, followed by K3A, R4Q, R4T, and R4G gp32s in that order. gp32-B does not stimulate homologous pairing even at the highest concentrations tested. Thus, the relative functional efficacy of these gp32 species parallels the hierarchy of relative binding affinities at equilibrium; nevertheless, our gel shift assays suggest that overall binding affinity only partially dictates the stimulatory activity of each gp32 mutant toward UvsX-catalyzed homologous pairing. As shown in Figs. 3 and 4, the concentrations of wild-type, R4K, R4Q, and K3A gp32s required to saturate the ssDNA substrate under homologous pairing solution conditions are similar. However, higher concentrations of K3A and R4Q gp32s are clearly required to stimulate the homologous pairing reaction to the same degree as wild-type gp32. Although both K3A and R4Q gp32s are capable of highly cooperative binding to ssDNA (i.e. the magnitude of is similar to that of wild-type gp32) and can therefore form long clusters on ssDNA, this property does not fully rescue the ability of either of these mutants to stimulate pairing. Only the addition of excess mutant gp32 beyond that which is required to form a saturated ssDNA complex is able to stimulate homologous pairing by UvsX. Excess gp32 increases the effective rate at which gp32 binds to the ssDNA relative to the rate at which gp32s dissociate from the lattice, thereby decreasing the lability of the mutant gp32-ssDNA complexes. Although the molecular basis of the reduced lifetime exhibited by mutant gp32-ssDNA complexes is unknown, other additional factors may well be important. For example, different gp32 species may undergo altered protein-protein interactions with UvsX or have different effects on the presynapsis phase of recombination, e.g. the formation of UvsX-ssDNA filaments. Similar conclusions can be drawn from our DNA replication experiments (Figs. 5 and 6, Table II). Here, however, the functional defects of several gp32 mutants appear far more severe than those observed in homologous pairing reactions. For example, the functional deficiencies of the K3A and R4Q gp32s in strand displacement DNA synthesis reactions are far greater than anticipated on the basis of their ssDNA binding properties relative to wild-type and R4K gp32s under the solution conditions and gp32 concentrations employed (see Figs. 3 and 4). R4Q and K3A gp32s are strongly defective in stimulation of the initiation of ssDNA synthesis from a nicked template (Fig. 5); furthermore, successful initiation events do not result in fully extended replication products in the presence of R4Q and K3A gp32s. This is suggestive of a decreased processivity of DNA synthesis by the polymerase holoenzyme, despite the presence of a significant excess of gp32 in each case. A particularly dramatic example of this phenomenon occurs with R4T gp32, which, under the conditions employed here, should fully saturate the displaced ssDNA strand generated by the polymerase yet largely fails to stimulate DNA synthesis. As expected, R4G gp32 and gp32-B are inactive in this assay, which can be traced to low ssDNA affinity.
The relative abilities of mutant gp32s to stimulate DNA synthesis from primed ssDNA templates are largely consistent with the results from the strand displacement experiments (Fig. 6). All gp32 mutants, with the exception of gp32-B, stimulate some formation of long replication products. However, K3A, R4Q, and R4T gp32s all give rise to an appreciable build-up of short products of intermediate length; these complexes also support significantly fewer initiation events by the DNA polymerase holoenzyme. In addition, all mutant gp32s, with the exception of R4K gp32, have difficulty replicating through the stable hairpin; this may be explained by previously determined defects in the helix-destabilizing activity of these mutant gp32s (11). These results taken collectively suggest that the lifetimes of the mutant gp32-ssDNA complexes are not optimized to effectively stimulate the initiation and processivity of DNA synthesis by the polymerase holoenzyme complex, since excess gp32 rescues the reaction to some extent. Alternatively or additionally, mutant gp32s may also be compromised in properly positioning the DNA polymerase on a moving primer terminus; defective gp32-gp43 interactions could also be important here.
In conclusion, we have demonstrated that moderately defective gp32s with well defined mechanistic deficiencies in cooperative binding to ssDNA are strongly affected in their abilities to stimulate homologous pairing by UvsX and DNA synthesis by the T4 DNA polymerase holoenzyme complex to a degree more severe than anticipated on the basis of their equilibrium affinity for ssDNA. It is tempting to attribute these defects to an enhanced lability or reduced lifetime of the cooperatively bound saturated mutant gp32-ssDNA complexes, due principally to the fact that increasing the concentration of gp32 to an extent that exceeds that required to saturate the ssDNA binding sites at least partially restores functional efficacy in vitro. It is anticipated that these mutant gp32s would be strongly defective in vivo, although these experiments have not been carried out. How mutant gp32s assemble on a ssDNA lattice relative to the wild-type gp32 complex is not known. Crystallographic studies of a functional gp32-ssDNA complex that includes the N-terminal B-domain as well the C-terminal A-domain, the latter of which mediates intermolecular interactions with the replication and recombination enzymes, remain unreported. We have speculated that the N-terminal domain may function as a gate that closes over the bound ssDNA and effectively links monomers together on the lattice (13). Mutations in the B-domain may, aside from functionally stabilizing the gp32-ssDNA complex, also help orient the C-terminal A-domain for productive interactions with other proteins in these multiprotein machines.