In vivo assembly of overproduced DNA polymerase III. Overproduction, purification, and characterization of the alpha, alpha-epsilon, and alpha-epsilon-theta subunits.

The genes for the polymerase core (αϵθ) of the DNA polymerase III holoenzyme map to widely separated loci on the Escherichia coli chromosome. To enable efficient overproduction and in vivo assembly of DNA polymerase III core, artificial operons containing the three structural genes, dnaE, dnaQ, and holE, were placed in an expression plasmid. The proteins α, αϵ and αϵθ were overexpressed and assembled in E. coli and purified to homogeneity. The three purified polymerases had a similar specific activity of about 6.0 × 106 units/mg in a gap-filling assay. Kinetics studies showed that neither ϵ nor θ influenced the Km of α for deoxynucleotide triphosphate and only slightly decreased the Km of α for DNA, although ϵ was absolutely required for maximal DNA synthesis. The rate of DNA synthesis by α-reconstituted holoenzyme using τ complex was about 5-fold less than that of αϵ or αϵθ-reconstituted holoenzyme as determined by a gel analysis. The processivity of α-reconstituted holoenzyme was very similar to that of αϵθ-reconstituted holoenzyme when τ complex was used as a clamp loader.

The genes for the polymerase core (␣⑀) of the DNA polymerase III holoenzyme map to widely separated loci on the Escherichia coli chromosome. To enable efficient overproduction and in vivo assembly of DNA polymerase III core, artificial operons containing the three structural genes, dnaE, dnaQ, and holE, were placed in an expression plasmid. The proteins ␣, ␣⑀ and ␣⑀ were overexpressed and assembled in E. coli and purified to homogeneity. The three purified polymerases had a similar specific activity of about 6.0 ؋ 10 6 units/mg in a gap-filling assay. Kinetics studies showed that neither ⑀ nor influenced the K m of ␣ for deoxynucleotide triphosphate and only slightly decreased the K m of ␣ for DNA, although ⑀ was absolutely required for maximal DNA synthesis. The rate of DNA synthesis by ␣-reconstituted holoenzyme using complex was about 5-fold less than that of ␣⑀ or ␣⑀-reconstituted holoenzyme as determined by a gel analysis. The processivity of ␣-reconstituted holoenzyme was very similar to that of ␣⑀-reconstituted holoenzyme when complex was used as a clamp loader.
Individual subunits of pol III have been overexpressed, purified, and characterized. The ␣ subunit contains catalytic polymerase activity and synthesizes DNA at a rate of approximately 10 nucleotides/s (Maki and Kornberg, 1987;Maki and Kornberg, 1985). The ⑀ subunit (the dnaQ product) contains 3Ј 3 5Ј exonuclease activity for the proofreading function of DNA replication (Scheuermann and Echols, 1984); thus, dnaQ (mutD) has a strong mutator phenotype (DiFrancesco et al., 1984). The function of the subunit in DNA replication is unclear.
Pol III is dimerized via the interaction between the and ␣ subunits, resulting in the formation of a dimeric polymerase that enables the coordinated synthesis of the leading and lagging strands (McHenry, 1982;Studwell-Vaughan and O'Donnell, 1991). The C-terminal region of ␣ binds to a dimer with a high affinity (K D ϭ 70 pM) (Kim and McHenry, 1996). Pol III itself is distributive but becomes a processive and rapid polymerase on a primed template with other accessory subunits (Fay et al., 1981). The DnaX complex ( 4 ␦␦Ј or ␥ 4 ␦␦Ј) loads the ␤ sliding clamp onto the primed template by coupling ATP hydrolysis Onrust et al., 1995). The ␤ sliding clamp provides pol III with high processivity by tethering it to the template (LaDuca et al., 1986;Stukenberg et al., 1991).
Each subunit of pol III works cooperatively and stimulates the activity of other subunits. For instance, the ␣ subunit can stimulate the exonuclease activity of the ⑀ subunit 10 -80-fold by increasing the affinity of ⑀ for the 3Ј-hydroxyl terminus (Maki and Kornberg, 1987). The ⑀ exonuclease activity is also slightly stimulated by the subunit (Studwell-Vaughan and O'Donnell, 1993). Additionally, ⑀ induces a 3-fold increase in the polymerase activity of the ␣ subunit (Maki and Kornberg, 1987). Thus, three pol III subunits are functionally cooperative. In fact, most DNA polymerases contain separate domains for the polymerase and exonuclease activities in a single polypeptide, suggesting that the two activities are interactive (Blanco et al., 1991).
In this present study, we constructed artificial operons that overexpress either ␣, ␣⑀, or ␣⑀ complexes assembled in vivo and purified them to homogeneity without the denaturationrenaturation step required for the purification of ⑀ due to its insolubility. The three purified polymerases were characterized, and their function and kinetics in DNA replication were compared.
Oligonucleotides-Two primers used to amplify the dnaQ gene were synthesized on a Biosearch 8600 DNA synthesizer. The 3Ј-primer is 5Ј-GGG GGA GAT CTA GGA GGT TTA AAA TAA TGA GCA CTG CAA TTA CAC GCC-3Ј (48-mer), and the 5Ј-primer is 5Ј-CCC CCC CAA GCT TCA CCC AGT GGC GGC CGC TGC AGT TAT GCT CGC CAG A-3Ј (50-mer). These two oligonucleotides were purified by DE52 column chromatography as described (Hagerman, 1985). Another four oligonucleotides (5Ј-AGG CGC ATA GGC TGG CTG ACC TT(N)-3Ј) complementary to the M13Gori DNA (nucleotides 993-1016) were purified by gel electrophoresis (Hagerman, 1985). They have the same DNA sequence except for the nucleotide at the 3Ј end (either C, G, A, or T), yielding a matched base pair (C) and three mismatched base pairs (G, A, T) to the template (G).
Construction of the Artificial Operon of Pol III Core-The dnaQ gene of pNS121 (Scheuermann et al., 1983) was amplified using two primers, the 5Ј-primer contains a 22-nucleotide sequence complementary to dnaQ, a Shine-Dalgarno site (AGGAGG), and a BglII restriction enzyme site; the 3Ј-primer has a 16-nucleotide sequence complementary to dnaQ and three cloning sites (PstI, DraIII, and HindIII). The polymerase chain reaction was conducted as described (Saiki et al., 1988). The polymerase chain reaction products of dnaQ were digested with BglII and HindIII and ligated to the large fragment of pJC1 (You and McHenry, 1993) digested at the same restriction sites (Fig. 1). The resulting plasmid was named pHN1, an ⑀ overexpressing plasmid. The plasmid pHN3, an ␣⑀ overexpressing plasmid, was generated by ligation of the PstI-DraIII fragment of pHN1 and the same restriction enzyme-digested fragment of pOPPA50 -4a2, an ␣ overexpressing plasmid (Tomasiewicz, 1991). Finally, the holE gene, encoding the subunit, from pHN100 (Carter et al., 1993) was inserted into pHN3 at the PstI site between dnaQ and dnaE to generate plasmid pHN4, which overexpressed pol III core (␣⑀) complex. Each gene of pol III in this plasmid contains its own Shine-Dalgarno sequence in front of a start codon (ATG) with an AT-rich, 9-nucleotide spacer.
Determination of Molar Extinction Coefficients-The extinction coefficients of the ␣ subunit, ␣⑀ complex, and pol III core (␣⑀) complex at ⑀ 280 were 99,920, 112,370, and 123,000 liters mol Ϫ1 cm Ϫ1 , respectively, as determined by the method of Edelhoch (1967). Proteins were dialyzed against buffer E overnight, and their extinction coefficients were determined in buffer E, in the presence or absence of 6 M guanidine hydrochloride. Spectra of the three polymerases were measured on a Hewlett-Packard 8450Z diode array spectrophotometer between 240 and 340 nm. Extinction coefficients of denatured proteins were calculated from the number of tryptophan and tyrosine residues in each protein (Edelhoch, 1967) and corrected by the ratio of the absorbances of the native proteins to the absorbances of the proteins in 6 M guanidine hydrochloride.
Gap-filling Polymerase Assay-This assay was performed by a modification of the method of McHenry and Crow (1979) and used to detect enzyme activity during protein purification. The reaction was initiated by the addition of enzyme to a 25-l solution containing four dNTPs (100 cpm of 3 H/pmol dNTPs), 10 mM MgCl 2 , and 5 g of activated calf thymus DNA in 50 mM HEPES (pH 7.5), 10 mM DTT, 200 mg/ml BSA, 0.02% Nonidet P-40, and 20% glycerol and incubated at 30°C for 5 min. One unit is defined as the amount of enzyme catalyzing the incorporation of 1 pmol of dNTPs per min at 30°C.
FIG. 1. Construction of plasmids to overexpress ␣⑀ and pol III. Plasmids were constructed as described under "Experimental Procedures." The backbone of the vectors was derived from pJC1, an HIV nucleocapsid (NC)-overexpressing plasmid (You and McHenry, 1993), which has a tac promoter, a replication origin (Ori), lacI q gene, two transcriptional terminators (T1 and T2), (Brosius et al., 1981), and the structural gene for ␤-lactamase (amp R ). S/D indicates a Shine-Dalgarno site.
FIG. 2. Overexpression of ␣, ⑀, and from pHN4. Total cell proteins before and after induction of E. coli strain HB101 (pHN4) were prepared as described (Kim and McHenry, 1996), and 20 l of each sample was loaded on a 10 -20% gradient SDS-polyacrylamide gel. Proteins were separated at a constant 65 V overnight. The gel was stained with Coomassie Brilliant Blue overnight and destained in a solution of 10% methanol and 10% acetic acid. Lane 1, protein markers; lane 2, uninduced total cell proteins; lane 3, induced total cell proteins.
Preparation of Activated Calf Thymus DNA-Calf thymus DNA (100 mg) was dissolved in 50 ml of 20 mM KCl with stirring overnight at 4°C. The dissolved DNA was treated with 0.4 g of DNase I (1 mg/ml in 1 mM CaCl 2 and 50% glycerol) per 10 mg DNA in a reaction mix containing 50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , and 50 g/ml BSA at 37°C for 20 min. DNase I was inactivated at 65°C for 10 min, followed by phenol: chloroform extraction, and an additional chloroform extraction. The DNA was precipitated by addition of 2 volumes of ethanol and 1/10 volume of 3 M sodium acetate (pH 5.2) (Ϫ20°C overnight). DNA samples were centrifuged at 10,000 ϫ g for 35 min and rinsed with 70% ethanol twice. The DNA pellet was initially dissolved in 30 ml of buffer TE and dialyzed against 2 liters of buffer TE for 4 h with one buffer change. The DNA concentration was determined spectrophotometrically at 260 nm (17 A 260 ϭ 1 mg/ml) and adjusted to 5 mg/ml final concentration by dilution with buffer TE.
Determination of Steady-state Kinetic Parameters-Kinetic parameters for dNTPs and activated DNA of the three polymerases (␣, ␣⑀, ␣⑀) were determined in the gap-filling polymerase assay described above in the presence of 67 fmol of each polymerase at 30°C for various times. dNTPs were titrated in the presence of 605 M activated calf thymus DNA (as nucleotide), or activated DNA was titrated in the presence of 60 M dNTPs to determine an initial velocity at each substrate concentration. The K m and V max were calculated from Lineweaver-Burk plots or by nonlinear least squares curve fitting (Kaleidagraph 3.0.1 soft- where v 0 is initial velocity at a given substrate concentration, and S is substrate concentration. The kinetic parameters (K m and V max ) determined by these two different methods agree to within ϳ5%.
Preparation of Primed M13Gori DNA-RNA-primed and SSB-coated M13Gori DNA was prepared (Fay et al., 1981). A reaction mix (1 ml) containing M13Gori DNA (1 mol as nucleotide), SSB (1.6 mg), DnaG primase (87,500 units), four NTPs (each 0.5 mM), and magnesium acetate (10 mM) in buffer B was incubated at 30°C for 10 min and applied to a Bio-Gel A-5m column (1 ϫ 25 cm) equilibrated with buffer B at 4°C. Primed M13Gori DNA was eluted with 40 ml of buffer B (flow rate 125 l/min) and detected by assaying with holoenzyme subunits and dNTPs . The peak fractions (total 1.8 ml) that incorporate more than 40 pmol of nucleotides in a 25-l assay using 1 l of each fraction were combined (83% yield based on replication assay).
Analysis of DNA Elongation Rate-The DNA synthesis rate of the holoenzyme-like activity reconstituted with three polymerases (␣, ␣⑀, ␣⑀) and complex and ␤ was determined by a modification of the method of Fay et al. (1981). A reaction mix (50 l) containing RNAprimed, SSB-coated M13Gori DNA template (480 fmol as a circle),  Bradford (1976). The actual yield of pol III from S-300 was 18 mg as determined using the extinction coefficient. Thus, the true specific activity of pure pol III is 5.9 ϫ 10 6 units/mg. complex (1.64 pmol as 4 ␦␦Ј), ␤ subunit (2.26 pmol as dimer), and 200 M ATP in enzyme dilution buffer was incubated at 30°C for 5 min with 1 pmol of ␣, ␣⑀, or ␣⑀ to allow formation of an initiation complex. The reaction was then placed in a 22°C water bath for 5 min to permit thermal equilibration and started by the addition of 6 l of dNTPs (each 0.8 mM) at 22°C. All reaction mixtures were initially made in a batch, and a 50-l sample was removed at the indicated time and quenched in 200 l of ethanol and 5 l of 4 M NaCl in a dry ice/ethanol bath. DNA samples were precipitated at Ϫ80°C overnight, spun at 15,000 ϫ g for 45 min at 4°C, and resuspended in 26 l of H 2 O. The DNA was digested with BbvI in a 30-l volume at 37°C for 1 h, loaded on an 8% native polyacrylamide gel (1.5 ϫ 25 ϫ 15 cm), and run at 100 V overnight. The gel was stained with ethidium bromide (25 g/ml) solution for 30 min and destained in a 1 mM MgSO 4 solution for 10 min to visualize DNA fragments with a UV illuminator.
Other Methods-Protein concentration during protein purification was determined by the method of Bradford (1976). Protein concentration of all purified proteins was determined using the extinction coefficient. SDS-polyacrylamide gel electrophoresis was performed by a modification of the method of Laemmli (1970). RESULTS We constructed artificial operons to overexpress pol III subunit complexes assembled in vivo. The dnaE, dnaQ, and holE genes were inserted into a vector with a tac promoter to produce the ␣⑀ complex, whereas dnaE and dnaQ were inserted to obtain a plasmid expressing the ␣⑀ complex ( Fig. 1). Three polymerases (␣, ␣⑀ or ␣⑀) were overexpressed and purified to 99% homogeneity from overexpressing E. coli strains. All steps of preparation were performed at 0 -4°C unless noted otherwise.

Purification of Pol III (␣⑀)
E. coli HB101 containing pHN4 was used to express pol III. The ␣, ⑀, and subunits were expressed at ϳ1, 13, and 6% of total proteins, respectively, as determined by densitometric  a The yield determined using the extinction coefficient was 9.5 mg. b Thus, the true specific activity of ␣ is 6.6 ϫ 10 6 units/mg.  (Fig. 2, lane 3). The presence of free ␣ or ␣⑀ would not be expected since ␣ was the limiting subunit. The extra ⑀ and subunits formed a soluble, separable complex (data not shown) through a direct interaction; excess ⑀ itself was insoluble.
Cell Lysis and Ammonium Sulfate Precipitation-Frozen cells (180 g) were thawed and lysed (2 mg of lysozyme per g of cells) to prepare cell lysates (Fr I) as described (Cull and McHenry, 1995). Initially, 0.226 g of ammonium sulfate (40% saturation at 0°C) for each ml of cell lysate was added, followed by two sequential backwashes with 0.200 and 0.170 g of ammonium sulfate added to each ml as described (Cull and McHenry, 1995). The final ammonium sulfate precipitate was resuspended in buffer I ϩ 25 mM NaCl to yield Fr II (44 ml, Table I).
Bio-Rex 70 Cation Exchange Chromatography-Fr II was dialyzed overnight against buffer I ϩ 25 mM NaCl, diluted to a conductivity equivalent to buffer I ϩ 25 mM NaCl by the addition of buffer I, and applied to a 190-ml pre-equilibrated Bio-Rex 70 column (5.75 ϫ 7.32 cm). The column was washed with 3-column volumes of buffer I ϩ 25 mM NaCl, and proteins were eluted with a 10-column volume gradient of buffer I ϩ 25 mM NaCl to buffer I ϩ 300 mM NaCl (flow rate ϭ 1-column volume per h). Pol III started eluting at a conductivity of buffer I ϩ 125 mM NaCl (Fig. 3A). The pool of peak fractions (72-98) (Fr III, 490 ml, Table I) was precipitated by the addition of an equal volume of saturated ammonium sulfate solution. The purity of pol III after this column was more than 80% based on densitometric scan of an SDS gel (Fig. 3B). All three subunits (␣, ⑀, and ) eluted in a constant ratio across the peak (Fig. 3B).
Sephacryl S-300 HR Gel Filtration Chromatography-The protein pellet obtained from ammonium sulfate precipitation of Fr III was dissolved in buffer A ϩ 100 mM KCl. The resulting protein solution (2 ml) was loaded onto a 110-ml, equilibrated Sephacryl S-300 column (1.5 ϫ 62.3 cm), and 80 fractions were collected at a flow rate of 0.1-column volume/h. Pol III eluted at fractions 44 -54 (Fig. 3C), resulting in Fr IV (15.1 ml, Table I). Individual fractions (50 l) were analyzed by SDS-polyacrylamide gel electrophoresis (Fig. 3D). The gel was overloaded (about 175 g at fractions 46 -52), so that trace contaminants could be detected. However, when 10 -20 g of this pool was loaded and resolved on a gel, no contamination was detected (Fig. 5). The overall yield of pol III in the purification was 34% ( Table I).

Purification of ␣⑀ Complex
Cell Lysis and Ammonium Sulfate Precipitation-E. coli HB101 containing pHN3 produced the ␣⑀ complex at a level of 2% ␣ and 20% ⑀ of total proteins (data not shown). Cell lysis (286 g of cells) and ammonium sulfate precipitation were performed under the same conditions as described for pol III purification except that the 0.170 backwash was skipped because it resulted in solubilization of a significant amount of ␣⑀. ␣ and ⑀ formed a tight complex at a 1:1 ratio, and the excess insoluble ⑀ subunits sedimented with cell debris (data not shown).
Bio-Rex 70 and Sephacryl S-300 Chromatography-A 280-ml Bio-Rex 70 column (2.7 ϫ 49.5 cm) was prepared, and proteins were loaded onto the equilibrated column as described for pol III. The column was washed with buffer I ϩ 75 mM NaCl, and proteins were eluted with a 10-column volume buffer I ϩ 75 to buffer I ϩ 300 mM NaCl gradient. The ␣⑀ complex started to elute at a conductivity of buffer I ϩ 100 mM NaCl, but some proportion of ␣⑀ complex eluted in early fractions (data not shown). Like pol III, the ␣⑀ complex was purified to nearhomogeneity from the cation exchange chromatographic step. The pool (Fr III, 965 ml, Table II) of fractions 37 to 78 was concentrated by ammonium sulfate precipitation. A 160-ml Sephacryl S-300 column (1.5 ϫ 90 cm) was used for the final purification. The protein pellet obtained from ammonium sulfate precipitation of Fr III was dissolved in buffer A (2.0 ml) and loaded onto the column. The activity eluted at fractions 42-49, resulting in Fr IV (16 ml). The overall yield of ␣⑀ complex was 20% (Table II), and the overall purification of the ␣⑀ complex was very similar to pol III in terms of column profile and purity. FIG. 5. Homogeneity of three purified polymerase forms. Purified ␣, ␣⑀, and ␣⑀ (10 g each) were loaded on a 12.5% SDS-polyacrylamide gel. After separation of proteins at a 65 V overnight, the gel was stained with Coomassie Brilliant Blue overnight, destained in a solution of 10% methanol and 10% acetic acid, and subjected to a densitometric scan (Molecular Dynamics). Lane 1, purified pol III core; lane 2, purified ␣⑀ complex; lane 3, purified ␣ subunit.

Purification of the ␣ Subunit
Cell Lysis and Ammonium Sulfate Precipitation-E. coli MC1061 containing pOPPA50 -4a2 (Tomasiewicz, 1991) was used to express the ␣ subunit to ϳ5% of total cell proteins (data not shown). Cell lysis (239 g of cells) was performed as described for pol III purification. The ␣ subunit alone was much more soluble in ammonium sulfate than the other two polymerase forms; it did not precipitate significantly under conditions described for pol III or ␣⑀ complex purification. Initially, 0.164 g of ammonium sulfate (30% saturation at 0°C) was added to each ml of Fr I lysate. Insoluble protein was removed by centrifugation (23,300 ϫ g at 0°C for 1 h). The supernatant, containing ␣, was adjusted to a final concentration 0.291 g/ml ammonium sulfate (50% saturation at 0°C). Precipitates (Fr II) were collected by centrifugation as described above.
Bio-Rex 70 Chromatography-A 450-ml Bio-Rex 70 column (5.75 ϫ 17.3 cm) was used in an identical manner as described for pol III. The ␣ subunit eluted at the same conductivity as the ␣⑀ and ␣⑀ complexes (data not shown). Fractions 78 -93 (Fr III, 445 ml, Table III) were pooled and precipitated by the addition of an equal volume of saturated ammonium sulfate solution.
Toyopearl Phenyl-650M Hydrophobic Chromatography-Although ␣ was purified to near-homogeneity from the Bio-Rex 70 column, it was less pure than pol III or ␣⑀. Thus, additional chromatographic steps were required. Ammonium sulfate-precipitated Fr III was dissolved in 30% A.S. buffer and loaded onto a pre-equilibrated hydrophobic Toyopearl phenyl-650M column (60 ml, 1.5 ϫ 36 cm). The column was washed with 1-column volume of 30% A.S. buffer followed by 2-column volumes of 15% A.S. buffer. Proteins were eluted with a 10-column volume 15-0% ammonium sulfate gradient at a flow rate of 0.84-column volume per h. The ␣ subunit eluted at 8 -6% saturating ammonium sulfate (Fig. 4, A and B). Fractions 42-60 were combined, resulting in Fr IV (150 ml, Table III). The pool of fractions was precipitated by the addition of an equal volume of saturated ammonium sulfate.
Sephacryl S-300 Chromatography-Gel filtration chromatography to remove aggregated proteins as well as to purify ␣ further was carried out as for ␣⑀ and ␣⑀. Ammonium sulfateprecipitated Fr IV was dissolved in 2 ml of buffer A ϩ 100 mM KCl and loaded onto a Sephacryl S-300 gel filtration column (160 ml, 1.5 ϫ 90 cm). ␣ eluted at fractions 46 -52. The pool (Fr V, 13.5 ml) of this column was nearly pure, but some contamination was detected when more than 75 g of protein was loaded on a gel (data not shown).
Heparin-Sepharose Chromatography-We conducted another chromatographic step to remove all trace contaminants. Fr V was dialyzed against buffer I ϩ 10 mM NaCl and loaded onto a 19-ml heparin-Sepharose column (0.75 ϫ 44 cm) equilibrated with buffer I ϩ 10 mM NaCl. The column was washed with 2-column volumes of buffer I ϩ10 mM NaCl and eluted with 10-column volumes of gradient (buffer I ϩ 10 3 300 mM NaCl) at a flow rate of 1-column volume per h. Fractions 58 -62 (Fr VI, 15 ml, Table III) were combined. The ␣ subunit eluted at a conductivity of buffer I ϩ 150 mM NaCl (Fig. 4, C and D).
SDS-Polyacrylamide Gel Electrophoresis of Purified Polymerase Forms-The three purified pol III polymerases (each 10 g) were analyzed by SDS-gel electrophoresis (Fig. 5). Based on densitometric scan, all pol III polymerases were purified to 99% homogeneity. Stoichiometries of ␣⑀ and ␣⑀ were determined by laser densitometry of complexes subjected to SDS-polyacrylamide gel electrophoresis, stained with Coomassie Blue, and corrected for molecular weight. The stoichiometry of the three pol III subunits was 1.0:1.1:0.9 (␣:⑀:). In the ␣⑀ preparation, the ratio was 1.0:1.1 (␣:⑀).
Steady-state Kinetics of Polymerases in Gap-filling Assay-To examine and compare the kinetic properties of the three polymerase forms, the K m and k cat values for dNTPs or activated DNA substrates (Table IV) were calculated by an iterative fit to the equation v 0 ϭ V max [S]/(K m ϩ[S]) as described under "Experimental Procedures" (Fig. 6) and from Lineweaver-Burk plots. The k cat was calculated from the equation where E t is total enzyme concentration. The K m of all three polymerase forms for dNTPs was in the range of 21-24 M. The k cat value (ϳ12 s Ϫ1 ) of ␣ was 2-fold less than that of the other two polymerase forms (Table IV). The K m of all three polymerase forms for activated calf thymus DNA was very similar, although K m for pol III was slightly lower. Because activated calf thymus DNA is such a heterogeneous template, K m values were defined in terms of total nucleotide concentration, permitting a relative comparison. The k cat /K m of pol III for both dNTP and DNA was about 2-fold higher than that of ␣ alone, indicating that ⑀ and made a modest contribution to the gap-filling polymerase activity of ␣. The 3Ј-OH concentration of activated DNA was estimated from the average size of DNA fragments determined by denaturing gel electrophoresis and dNTPs incorporated between gaps, ϳ150 M as nucleotide was equivalent to 1 M of 3Ј-OH.
Primer Extension from Mismatched 3Ј Ends by Holoenzymes Reconstituted with ␣, ␣⑀ and ␣⑀-The ␣ subunit itself does not have proofreading activity to remove misincorporated bases during replication. We asked whether ␣-reconstituted holoenzyme extends nucleotides further from the mispaired 3Ј end or, instead, pauses in DNA elongation when an incorrect nucleotide is incorporated. Using four oligonucleotides annealed to M13Gori DNA (Fig. 7A) and saturating levels of polymerases, we carried out assays for various times to examine utilization of mispaired primer termini and found that ␣ elongated DNA from the 3Ј ends of a paired C-G and mispaired T-G but could not overcome G-G and A-G mismatches (Fig. 7, B-E). Perhaps T was elongated by forming a transient base pair with G. Both ␣⑀ and pol III utilized each 3Ј end at almost the same rate because of their 3Ј 3 5Ј proofreading activity which removes mispaired nucleotides (Fig. 7, B-E).
DNA Elongation Rates of Holoenzymes Reconstituted with the Three Polymerase Forms-Earlier studies concluded that ⑀ is required for ␣ to achieve rapid and highly processive DNA synthesis (Studwell-Vaughan and O'Donnell, 1990). The avail- FIG. 6. Determination of steady-state kinetic constants for ␣, ␣⑀, and ␣⑀. The initial velocity (v 0 ) of ␣ (E), ␣⑀ (Ⅺ), and ␣⑀ (Ç) determined by gap-filling assay was plotted versus the concentration (as nucleotide) of activated DNA (S). This plot was generated by a nonlinear curve fitting program as described under "Experimental Procedures." The estimated K m and V max values from this plot are 87 and 2 M/min for ␣, 108 and 3.3 M/min for ␣⑀, and 76 and 3.5 M/min for ␣⑀, respectively. ability of three highly purified polymerase forms allowed us to study the effect of ⑀ or on DNA synthesis rates by reconstituted holoenzyme. Initiation complexes were formed using saturating levels of either ␣, ␣⑀, or ␣⑀ as described under "Experimental Procedures." Reactions were started by the addition of dNTPs at 22°C (Fig. 8A). The DNA fragments generated from BbvI digestion are a (900 bp), b (175 bp), c (690 bp), d (1739 bp), e (611 bp), f (1154 bp), g (410 bp), h (2000 bp), i (688 bp), and j (256 bp) in the order produced. Fragment d appeared at 2 min for ␣-reconstituted holoenzyme and at 30 s for both ␣⑀-and ␣⑀-reconstituted holoenzymes, and fragment h appeared at 5 min for ␣-reconstituted holoenzyme and at 1 min for ␣⑀and ␣⑀-reconstituted holoenzymes (Fig. 8B). Based on the production of these fragments, the elongation rates of reconstituted holoenzymes at 22°C were calculated as 28 Ϯ 2, 126 Ϯ 4, and 126 Ϯ 4 nucleotides/s for ␣-, ␣⑀,and ␣⑀-reconstituted holoenzymes, respectively. The elongation rate of ␣-reconstituted holoenzyme was approximately 5-fold slower than that of ␣⑀and ␣⑀-reconstituted holoenzymes.
plex-In previous studies, the ␣-reconstituted holoenzyme showed a processivity of 1-3 kb when ␥ complex was used as a clamp loader, whereas both ␣⑀and ␣⑀-reconstituted holoenzymes were fully processive for a full cycle of replication on an M13 DNA template (Studwell-Vaughan and O'Donnell, 1990). We determined the processivity of the ␣-reconstituted holoenzyme using anti-␤ IgG to prevent reinitiation onto the DNA template once an initiation complex was formed (Fay et al., 1981). Holoenzymes reconstituted using either the ␣ subunit or pol III in the absence of anti-␤ IgG synthesized a full circle of M13Gori DNA regardless of which DnaX complex ( or ␥ complex) was used (Fig. 9, lanes 2, 3, 6, and 7). When anti-␤ IgG was added after initiation complex formation, the ␣and ␣⑀reconstituted holoenzyme using complex was fully processive (compare lanes 4 and 5 of Fig. 9), while the ␣-reconstituted holoenzyme using ␥ complex showed a processivity of 3.6 kb based on the last fragment (d) synthesized from this reaction (Fig. 9, lane 8). Anti-␤ IgG inhibited 99% of holoenzyme activity if it was added to a reaction mix before pol III (or ␣), DnaX complex, and ␤ followed by incubations for sufficient time for replication of a full circle of M13Gori DNA (15 s for pol III and 1 min for ␣). The intensity of each fragment in the same lane was determined by a densitometric scan. Comparison of ratios of early fragments (a, b, or d) and late fragments (f or g) in the presence or absence of anti-␤ IgG showed that once holoenzyme is reconstituted with either ␣ or ␣⑀ using complex to initiate DNA synthesis, it processively replicates M13Gori DNA (until fragment g). However, the ratio of early fragments and one particular late fragment (h) of ␣-reconstituted holoenzyme using complex was 3-4-fold higher in the absence of anti-␤ IgG (lane 2) than in its presence (lane 4), indicating that only 25-30% of initiation complexes completed the synthesis of fragment h in the presence of anti-␤ IgG during the time course of the reaction. Perhaps the movement of holoenzyme is impeded by a strong hairpin in the bacteriophage M13 replication origin (van Wezenbeek et al., 1980) located in fragment h.

DISCUSSION
Three subunits of pol III are tightly associated and can be isolated from wild-type cell lysates (McHenry and Crow, 1979). Since a cell contains only 10 -20 molecules of pol III (Wu et al., 1984), purification of pol III from wild-type E. coli required a 30,000-fold purification and enormous quantities of cells (McHenry and Crow, 1979). We have constructed an artificial operon containing combinations of the three pol III genes (dnaE, dnaQ, and holE) which permitted overexpression of three pol III subunits from a single promoter and the assembly of complexes in vivo. From 180 g of pol III overexpressing cells, FIG. 8. Determination of the elongation rate of ␣, ␣⑀, and ␣⑀ in reconstituted holoenzyme. A, scheme of elongation assay. The initiation complex on RNA-primed, SSB-coated M13Gori DNA was formed at 30°C for 5 min. The reaction was started at 22°C by the addition of dNTPs. The numbers on replicated double-stranded DNA indicate the distance of BbvI cleavage sites from the 3Ј-OH primer terminus. Fragments generated from the restriction enzyme digestion are given as a  12 and 19), and 5 min (lanes 13 and 20) for ␣⑀ and ␣⑀-reconstituted holoenzymes. The DNA fragments (a-j) were separated by 8% native polyacrylamide gel electrophoresis and stained with ethidium bromide.
FIG. 9. Processivity of ␣-reconstituted holoenzyme. Initiation complexes were formed by addition of either (lanes 1-5) or ␥ (lanes 6-10) complex (1.2 pmol as 4 ␦␦Ј or ␥ 4 ␦␦Ј), ␤ (565 fmol as dimer), and either 1 pmol of ␣ (lanes 2, 4, 6, and 8) or ␣⑀ (lanes 3, 5, 7, and 9) to primed M13Gori DNA (500 fmol as circle) as prepared under "Experimental Procedures" in 50 l of enzyme dilution buffer containing 10 mM magnesium acetate and 200 M ATP by an incubation at 30°C for 5 min. Elongation reaction was initiated by the addition of each 48 M of dATP, dCTP, and dGTP and 18 M of [ 3 H]dTTP (100 cpm/pmol) at 30°C for 5 min. To block reinitiation during polymerase cycling so that processivity could be measured, 20 g of anti-␤ IgG was added to tubes (lanes 4, 5, 8, and 9) prior to dNTP addition. Anti-␤ IgG was omitted from lanes 2, 3, 6, and 7. Subsequent steps were performed as described in Fig. 8. Lanes 1 and 10 are controls in the absence of polymerases. As a control to demonstrate that anti-␤ IgG under these reaction conditions can rapidly inhibit cycling polymerase, tubes containing all components except polymerase (either ␣ or ␣⑀), DnaX complex, and ␤ either in the presence or absence of anti-␤ IgG (20 g) were incubated at 30°C for 5 min, immediately added to a mix of polymerase, DnaX complex, and ␤ and incubated for an additional 15 s and 1 min for ␣⑀and ␣-reconstituted holoenzymes, respectively. Radioactivity of [ 3 H]dTMP-incorporated DNA was measured as described (McHenry and Crow, 1979), and anti-␤ IgG inhibited 99% of holoenzyme activity. Letters on the right side of the gel indicate the DNA fragments (a-j) generated by restriction enzyme digestion as described in Fig. 8. 18 mg of pol III was purified to 99% homogeneity in two chromatographic steps. It is interesting that ⑀, normally insoluble when overproduced alone (Sheuermann and Echols, 1984), forms a defined 1:1 complex with ␣ when coexpressed in vivo, resulting in a soluble complex. Excess ⑀ is found in inclusion bodies.
Gap-filling polymerase assays were used to determine the contribution of ⑀ or to the kinetic properties of ␣. All three polymerase forms, ␣, ␣⑀, and ␣⑀, had identical affinities for dNTP, suggesting that ⑀ and are not involved in dNTP binding by ␣. The K m of ␣ for DNA was about 2-fold higher than ␣⑀. The k cat of ␣ was 2-fold lower than that of ␣⑀ and ␣⑀, indicating a modest contribution from ⑀, perhaps by stabilizing a more active conformation of ␣; made no detectable contributions, consistent with other biochemical (Studwell-Vaughan and O'Donnell, 1990) and genetic (Slater et al., 1994) studies.
To complete the synthesis of 4.4-megabase pairs of the E. coli genome within 40 min, holoenzyme must synthesize DNA at a rate of about 1 kb per s. In vitro replication assays have shown that naturally purified as well as reconstituted holoenzyme synthesize DNA at a rate of about 500 nucleotides/s at 30°C . Our gel analysis of restriction fragments indicated a DNA synthesis rate of holoenzyme reconstituted with pol III, ␤, and complex of about 130 nucleotides per s at 22°C. The ␣⑀-reconstituted holoenzyme synthesized DNA at the same rate as ␣⑀-reconstituted holoenzyme, but the ␣-reconstituted holoenzyme elongated DNA at a rate ϳ5-fold slower. Therefore, rapid DNA synthesis by holoenzyme is dependent on ⑀. By contrast, appears to have no influence on DNA elongation at 22°C. Although the ␣-reconstituted holoenzyme using complex showed a 5-fold slower elongation rate than holoenzyme reconstituted using ␣⑀ or ␣⑀, its processivity was very similar to that of the ␣⑀-reconstituted holoenzyme. When ␥ complex replaced complex, the processivity of ␣-reconstituted holoenzyme decreased (Fig. 9). protects ␤ from removal by ␥ complex (Kim et al. 1996a). Presumably, in the absence of , the slower moving ␣ provides more time for ␤ removal, resulting in a lower apparent processivity.
The function of in DNA replication is not clear. Analysis of a null mutation in holE has shown that is dispensable for E. coli growth (Slater et al., 1994). Our kinetic and functional studies of three polymerases did not show any significant effect of in DNA replication, although it slightly increased ␣'s binding to DNA substrates. The subunit has also been shown to slightly stimulate the 3Ј 3 5Ј proofreading exonuclease activity of the ⑀ subunit at a mismatched base pair (Studwell-Vaughan and O'Donnell, 1993). Another possible role of is communication with other replication proteins such as primase or helicase at the replication fork. One of two holoenzymes at the replication fork continuously recycles onto newly synthesized primer after completion of the synthesis of a preceding Okazaki fragment, which requires a signal between a new primer and subunits of holoenzyme (Wu et al., 1992). Clearly a -DnaB inter-action is required for replicase-primosome coupling (Kim et al., 1996b) but that does not preclude additional primosome-holoenzyme interactions. Finally, might somehow mediate the conformational change in pol III that is probably necessary in switching between the polymerase and exonuclease activity of pol III during replication. Firm conclusions about the function of await further genetic, structural, and functional studies.