Purification and Properties of Wild-type and Exonuclease-deficient DNA Polymerase II from Escherichia coli(*)
- From the (1) Department of Biological Science, Hedco Molecular Biology Laboratories, University of Southern California, Los Angeles, California 90089-1340, the
- (2)Department of Biological Chemistry and the Molecular Biology Institute, UCLA School of Medicine, Los Angeles, California 90024, and the
- (3)Laboratory of Molecular Genetics, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina 27709
Abstract
Wild-type DNA polymerase II (pol II) and an exonuclease-deficient pol II mutant (D155A/E157A) have been overexpressed and
purified in high yield from Escherichia coli. Wild-type pol II exhibits a high proofreading 3′-exonuclease to polymerase ratio, similar in magnitude to that observed
for bacteriophage T4 DNA polymerase. While copying a 250-nucleotide region of the lacZα gene, the fidelity of wild-type pol II is high, with error rates for single-base substitution and frameshift errors being
≤10
. In contrast, the pol II exonuclease-deficient mutant generated a variety of base substitution and single base frameshift
errors, as well as deletions between both perfect and imperfect directly repeated sequences separated by a few to hundreds
of nucleotides. Error rates for the pol II exonuclease-deficient mutant were from ≥13- to ≥240-fold higher than for wild-type
pol II, depending on the type of error considered. These data suggest that from 90 to >99% of base substitutions, frameshifts,
and large deletions are efficiently proofread by the enzyme. The results of these experiments together with recent in vivo studies suggest an important role for pol II in the fidelity of DNA synthesis in cells.
INTRODUCTION
Three DNA polymerases have been identified in Escherichia coli(1). Pol I1(1)is involved in repair of damaged DNA and processing RNA-primed Okazaki fragments(1), while pol III is required for replication of the bacterial genome(2). Although pol II was discovered more than 24 years ago(3), its role in either replication or repair has been difficult to define. It has been shown that pol II expression is regulated by the lexA repressor, as part of the SOS regulon, where the locus of the structural gene for pol II is coincident with the DNA damage-inducible dinA gene(4, 5). In vitro studies have established that DNA pol II activity (6, 7) and processivity (8) are strongly stimulated by DNA pol III accessory proteins suggesting that the two polymerases might share overlapping roles in the cell. An E. coli strain containing a deletion in the pol II structural gene exhibited an increase in adaptive mutation rate(9), suggesting a role for pol II in localized DNA synthesis or repair in nondividing cells(10). Recently, we have shown that replacement of wild-type polB with an exonuclease-deficient polB allele increases the rate adaptive mutation in E. coli.2(2)
As a consequence of these newly-established pol II phenotypes(10, 12), 2 biochemical investigations of pol II(7, 8, 13, 14) are especially timely and important to pursue. Pol II is classified as a type II, human α-like polymerase (4) having considerable sequence similarity to bacteriophage T4 pol. The enzyme contains both polymerase and 3′-exonuclease activities within a single 89.9-kDa polypeptide chain(4, 15, 16). In this paper, we describe the purification and biochemical properties of wild-type pol II and an exonuclease-deficient pol II derivative (pol II ex1).
EXPERIMENTAL PROCEDURES
Chemicals
All nonradioactive dNTP substrates were purchased from Pharmacia LKB Biotechnology Inc. Radioactive [
-
P]ATP (4000 Ci/mmol), [α-
P]dNTP (4000 Ci/mmol), and [methyl-3H]TTP (20 Ci/mmol) were purchased from ICN Radiochemicals, Inc. (Irvine, CA). Diethylaminoethyl-cellulose (DE52) and phosphocellulose
(P-11) resins were obtained from Whatman BioSystems Inc.
Enzymes
Phage T4 DNA polymerase (3.5 
106 units/mg(17), 1 unit is that amount of enzyme required to incorporate 1 pmol of labeled dTMP into nucleic acid product in 1 min at 37°C)
was kindly provided by Dr. L. Reha-Krantz (University of Alberta, Canada). A measurement of T4 pol specific activity was also
carried out at 30°C for direct comparison with published data; the value obtained, 2 
106 units/mg is consistent with the data in Ref.(17). Klenow fragment (7000 units/mg) was purchased from Pharmacia LKB Biotechnology Inc. Pol II wild-type and a 3′-exonuclease-deficient
pol II derivative were purified as described below.
Primer/Template
Single-stranded M13 DNA template was prepared as described(18). Primers were synthesized using an Applied Biosciences DNA synthesizer by Lynn Williams (Kenneth Norris Jr. Comprehensive Cancer Center, University of Southern California, Los Angeles) and used after purification by polyacrylamide gel electrophoresis. Primer sequence p5, 5′-ATTAAATCCTTTGCCCG-3′, is complementary to the numbered positions on wild-type M13 strand: 4640-4657. Primer p15C (5′-ATGATTAAGACTCCTTA-3′) can be annealed to template TGG-1 (3′-TACTAATTCTGAGGAATGGTGAGTCATACAATCGT-5′). The sequence of primer Prsp2 is 5′-GATCAGTCCTGTACGGTACTGACTGACC-3′.
Bacteria Strain and Plasmid
E. coli JM109 (recA1 Δ(lac-proAB
) F′(traD36 proAB
lacI
lacZΔM15)) was kindly provided by Dr. Miriam Susskind, University of Southern California. Plasmid pPROK-1 was purchased from CLONTECH
Laboratories, Inc. (Palo Alto, CA).
Construction of Pol II Overproducing Plasmid (pHY400) and a Pol II 3′-Exonuclease Mutant (D155A/E157A) Overproducing Plasmid (pHC700)
A 2.4-kb DNA fragment containing the dinA/polB3(3)open reading frame was obtained from plasmid pSH100 by polymerase chain reaction amplification of the polB coding region(4). The original “inefficient” GTG translation initiation codon was changed to ATG using a polymerase chain reaction primer containing the appropriate base change. This 2.4-kb polymerase chain reaction fragment was subcloned into the pPROK-1 vector (a 4.6-kb plasmid vector containing a Ptac promotor) to give a 7.0-kb plasmid construct, pHY400. The sequence of polB was confirmed by DNA sequence analysis. The expression of polB is under the control of the Ptac promotor that is regulated by Lac Iq. The polB gene containing substitutions D155A/E157A was engineered using standard oligonucleotide-directed mutagenesis procedures(19). Mutations in the plasmid were screened initially by restriction endonuclease mapping (the mutant oligonucleotide encoding the alanine substitution introduced a new restriction site for AflII endonuclease) and later by DNA sequencing of the polB gene. A 2.4-kb fragment corresponding to the polB open reading frame containing the desired mutations was inserted into the pPROK-1 vector resulting in a 7.0-kb plasmid, pHC700.
Cell Growth and Cell Lysis
E. coli JM109 cells carrying the polB gene on the overproducing plasmid pHY400 (wild-type pol II) or pHC700 (3′-exonuclease-deficient pol II, i.e. pol II ex1) were grown in LB with 50 μg/ml ampicillin in a 180-liter fermenter at 37°C. The overproduction of pol II protein
was induced by adding isopropyl β-D-thiogalactoside to the cells at midlog phase (A
about 0.8) to a final concentration of 0.4 mM. Two hours after isopropyl β-D-thiogalactoside induction, cells were harvested
and resuspended in a volume of storage buffer (sterile 50 mM Tris-HCl, pH 7.5, 10% sucrose) equal to the wet weight of the
cells in grams. Cells were rapidly frozen by resuspension in liquid nitrogen and were stored at −70°C. Lysis buffer (50 mM
Tris-HCl, pH 7.5, 10% sucrose, 0.1 M NaCl, 15 mM spermidine) was added to frozen cells to achieve a final concentration of
0.2 g of cells/ml. Cells were thawed at 4°C and the pH was adjusted to 7.7 with 2 M Tris base. Lysozyme was added to a final
concentration of 0.2 mg/ml. The cell slurry was incubated for 1 h at 4°C, 4 min at 37°C, and centrifuged at 11,800 rpm for
1 h in a Sorvall GSA rotor. The supernatant (Fraction I) was decanted and kept on ice.
Purification of Pol II
Ammonium sulfate was added slowly to Fraction I, to a final concentration of 30% (w/v), and the suspension was kept at 4°C
overnight, without stirring. The suspension was centrifuged in a Sorvall GSA rotor (11,800 rpm, 40 min) and the supernatant
was discarded. A volume of PC/25 buffer (50 mM Tris-HCl, pH 7.5, 20% glycerol, 1 mM EDTA, 5 mM dithiothreitol, 25 mM NaCl)
equal to one-fifth to one-tenth of the original volume of Fraction I was added to the ammonium sulfate pellet and the protein
was gently resuspended (Fraction II). Fraction II was dialyzed against PC/25 buffer until the conductivity reached a value
equivalent to that of 40 mM NaCl. After dialysis, Fraction II was diluted with PC/25 buffer to a protein concentration of
approximately 10 mg/ml and then loaded onto a P-11 phosphocellulose column, 5 
70 cm (inner diameter 
length). The column was washed with 1 column volume of PC/25 buffer and an additional column volume of PC/200 buffer (the
same components as buffer PC/25 except that the NaCl concentration was 200 mM) to elute DNA polymerase III. Pol II protein
was eluted with a gradient of 200-500 mM NaCl in buffer PC (8 column volumes). The pol II fractions (eluting between 225 and
250 mM NaCl) were pooled to give Fraction III. Fraction III was dialyzed against PK20 buffer (20 mM potassium phosphate, pH
6.8, 15% glycerol, 1 mM EDTA, 5 mM dithiothreitol) until the conductivity reached that of PK30 buffer (30 mM potassium phosphate,
pH 6.8) and loaded on a DEAE column, 5 
70 cm (inner diameter 
length). The DEAE column was washed with 2-column volumes of PK20 followed by elution with an 8-column volume gradient of
20-350 mM potassium phosphate (PK20 to PK350). The pol II fractions (typically eluting at 100-140 mM potassium phosphate)
were pooled as Fraction IV and stored at −70°C. The procedure used to purify the exonuclease deficient pol II mutant was the
same used for wild-type pol II. The specific activity and recovery at each step of purification for wild-type pol II is contained
in.
Deoxyribonucleotide Incorporation Assay
The deoxynucleotide incorporation activity of pol II was measured using [methyl-3H]TTP and activated salmon sperm DNA as described(20, 21). Reactions contained 2.5 mM dithiothreitol, 20 mM Tris-HCl, pH 7.5, 7.3 mM MgCl2, 6 mM spermidine/HCl, 1 mg/ml bovine serum albumin, 1.1 mM activated DNA, 60 μM dATP, dCTP, dTTP, dGTP, and [3H]dTTP. One unit of enzyme catalyzed the incorporation of 1 pmol of [3H]TMP into acid insoluble material in 1 min at 37°C. T4 pol deoxynucleotide incorporation activity was measured by the same method except the reaction buffer contained 67 mM Tris-HCl, pH 8.8, 16.7 mM (NH4)2SO4, 0.5 mM dithiothreitol, 6.7 mM MgCl2, and 167 μg/ml bovine serum albumin.
Protein Assay
Protein concentration was measured by the Bradford assay using materials and protocol supplied by Bio-Rad. Silver staining of protein gels were carried out using protocols and materials supplied in the ICN Rapid Ag-Staining Kit.
Active Site Titration of Purified Pol II Protein Using DNA-Heparin Trap
5′ End-labeling of primer p15C was carried out in a labeling reaction (20 μl) containing 50 mM Tris-HCl, pH 7.5, 2 mM MgCl2, and 0.5 mM dithiothreitol, 3.0 μM primer, 0.3 μM [
-
P]ATP, and 10 units of T4 polynucleotide kinase. The labeling reaction was performed at 37°C for 1 h and terminated by heating
it to 90°C for 5 min. 2 μl of 100 mM EDTA (dissolved in 5 mM Tris-HCl, pH 8.0) was added to the labeling reaction to chelate
Mg
ions. Template TGG-1 was added to the above solution and conditions were adjusted to attain 17 mM Tris-HCl, pH 7.5, 0.7 mM
MgCl2, 0.17 mM dithiothreitol, 1.0 μM primer, 1.25 μM template TGG-1, and 3 mM EDTA. Annealing was carried out by heating the annealing
reaction mixture to 90°C for 3 min and then cooling to room temperature over a 1-h time period. Under these conditions, more
than 96% of annealed primer could be utilized by polymerase as judged by the primer extension experiment described below (data
not shown). A combination of 0.1 mg/ml heparin sulfate and 0.66 mg/ml calf thymus DNA, prepared as described (22) was completely effective in trapping free (i.e. dissociated) pol II. For a typical reaction (20 μl), 4 μl of enzyme was preincubated with 8 μl of primer/template/dNTP mixture
for 1 min at 37°C to form enzyme-primer-template complexes, and primer extension was initiated with 8 μl of MgCl2/trap mixture. The final concentrations were 3-50 nM polymerase, 50 mM Tris-HCl, pH 7.5, 50 nM primer/template DNA, 50 μg/ml
bovine serum albumin, 0.5 mM dithiothreitol, 60 μM dNTPs, 7.3 mM MgCl2, 0.1 mg/ml heparin sulfate, and 0.66 mg/ml calf thymus DNA. Following incubation, the reaction was quenched by addition of
20 mM EDTA (40 μl) in 95% formamide. The efficiency of the trap was determined by preincubating pol II with the trapping mixture
before addition of labeled primer/template. Separation of 
P-labeled primers of different lengths was carried out by gel electrophoresis, and the fraction of primer molecules extended
by addition of 1 or more nucleotides was determined using a PhosphorImager (Molecular Dynamics). The amount of primer extended
represents the minimum fraction of active enzyme-primer-template complex and therefore determines a lower limit for the fraction of active pol II
(see “Results”). The minimum fraction of active pol II is given by the slope of the curve obtained by plotting the amount
of primer extended (nM) versus the amount of enzyme (nM) in the reaction.
Measurement of Exonuclease Activity by Primer Degradation
5′ End-labeling and gel electrophoresis were carried out as described(23). To measure 3′-exonuclease activity, 20 μl total reaction volume, 50 nM primer p5 (as single-stranded DNA substrate) was incubated with pol II, T4 pol, or KF. The concentrations of the three enzymes were adjusted to an equivalent of 0.04 pol II units. Incubations were carried out for different times at 37°C and reactions were terminated as described above. Reaction buffers for pol II and KF contained 50 mM Tris-HCl, pH 7.5, 50 μg/ml bovine serum albumin, 0.5 mM dithiothreitol, and 7.3 mM MgCl2. The reaction buffer for T4 pol was given above. The rate of primer degradation was determined from the slope of the linear region of a plot of percent primer degraded versus time. A comparison of the exonuclease activities of wild-type pol II and pol II ex1 (D155A/E157A) was carried out by the same method, except that primer Prsp2 was used as the substrate (28-mer, at a concentration of 180 nM) and 0.4 units of wild-type pol II (or 4 units of pol II ex1) was present in the reaction mixture (40 μl total volume).
Turnover Assay
Primer/template p5-M13 was used for measuring dNTP 
dNMP turnover during DNA synthesis. The reaction conditions were the same as those used to measure 3′-exonuclease activity
by primer degradation except the primer p5 was unlabeled and [α-
P]dCTP or [α-
P]dTTP (4000 Ci/mmol, 3.3 μM) and 95 μM dATP were used to allow primers to be extended by addition of 3 nucleotides. 0.4 pol
II units were used to measure turnover of the correctly inserted nucleotide (dCTP), and 1.2 pol II units were used to measure
turnover of the incorrectly inserted nucleotide (dTTP). The concentrations of T4 pol and KF were adjusted to have the same
number of polymerase units that was used for pol II, i.e. the nucleotide insertion rates were adjusted to be the same for all three enzymes. Product dNMP was separated from unreacted
dNTP and contaminating dNDP by electrophoresis on an 8% polyacrylamide gel for 1 h at 2000 V. The rate of turnover was determined
from the slope of the linear region of a plot of the amount of dNMP formed versus time. Contaminating levels of radioactive dNDP in dNTP substrates were typically about 0.5-1% and were easily resolved from
the product dNMP (see, e.g.Fig. 4); levels of contaminating dNMP in freshly prepared substates were negligible compared to the amount of product formed during
the reaction.
Comparison of pol II, T4 pol, and KF dNTP 
dNMP turnover activities. A, incorporation and excision of a correct deoxynucleotide [α-
P]dCTP opposite template G on the unlabeled primer/template DNA shown at the top of the figure. Incorporation of labeled dCTP
is indicated by the location of the extended primer band corresponding, primarily, to the addition of three nucleotides. Turnover
of the dCTP substrate is indicated by the dCMP radioactive spot. The heavy radioactive spot migrating at the bottom of the
gel corresponds to unreacted dCTP, and the spot migrating between the unreacted dCTP and dCMP product is dCDP contaminant
present in the dCTP substrate. B, excision of an incorrectly inserted deoxynucleotide [α-
P]dTTP opposite template G on the unlabeled primer/template shown at the top of the figure. No stable incorporation of dTTP is observed as shown by the absence of a detectable radioactive band at the
location denoted by “extended primer.” Turnover of the dTTP substrate is indicated by the dTMP radioactive spot. The radioactive
spot migrating at the bottom of the gel is unreacted dTTP, and the spot located between dTTP and dTMP product is dTDP contaminant present in dTTP. Reactions
using pol II, T4 pol, and KF were carried out for the indicated times. The position of the unextended primer (5′-
P-labeled 17-mer) is shown in the left-hand lane of each gel. The concentrations of the three enzymes used for turnover of the correct nucleotide (dCMP) were adjusted to
have the same number of polymerase units, equivalent to 0.4 pol II units. Three-fold higher concentrations of each polymerase
(1.2 pol II units) were used to measure turnover of the incorrect nucleotide (dTMP).
Forward Mutagenesis Assay
The assay, described in detail in Ref. 24, measures errors in the wild-type lacZα gene of M13 mp2. Correct polymerization during synthesis to fill a 390-base single-stranded gap produces DNA that yields dark blue M13 plaques upon transfection of an appropriate E. coli host strain. A wide variety of errors at many different sites are scored as lighter blue or colorless plaques. After sequence analysis of a collection of independent mutants, errors rates per detectable nucleotide incorporated were calculated as described(24). The DNA polymerase reaction conditions and description of product analysis are in the legend to. E. coli strains, bacteriophage M13 mp2, and the sources of materials used for the M13 mp2 fidelity assay have been described(24).
RESULTS
E. coli JM109 cells carrying the polB gene under Ptac control on a high copy number plasmid were used to overproduce wild-type pol II and the 3′-exonuclease-deficient
pol II derivative (pol II ex1). A typical preparative scale purification starting with 300 g of dry cells yielded about 300
mg of purified pol II (). Approximately 75-80% of the enzyme activity was recovered with a corresponding 35-40-fold purification
(). The degree of purification and recovery of polymerase activity for the exonuclease-deficient enzyme was similar to wild-type
pol II (data not shown). Fractions from each step of the purification were analyzed by SDS-polyacrylamide gel electrophoresis
and visualized by silver staining (Fig. 1). Each enzyme preparation appeared to be at least 95% chromatographically pure. Crystals suitable for high resolution x-ray
diffraction have been obtained from the fraction IV material for the wild-type protein(25). The specific activity of wild-type pol II was routinely 3-fold higher than for the pol II ex1 enzyme suggesting that the
Asp
to Ala and Glu
to Ala mutations in the exonuclease domain, that reduce levels of exonuclease to 1/1000th the level of wild-type, appeared
to cause a small reduction in the polymerization rate of the enzyme.
Silver-stained polyacrylamide gel showing protein bands during purification of E. coli wild-type and exonuclease-deficient (exo
) pol II. Lane 1, prestained molecular weight markers; lane 2, F1 crude lysate (17 μg total protein); lane 3, F2 ammonium sulfate fraction (20 μg total protein); lane 4, phosphocellulose column fraction (5 μg total protein); lane 5, DEAE-cellulose column fraction (5 μg total protein). The purification procedure is described under “Experimental Procedures.”
The specific activity and recovery at each stage of purification for the wild-type pol II is given in Table I.
A titration using increasing levels of pol II at a constant primer/template DNA concentration was carried out to determine
the fraction of pol II (fraction IV) that was active (Fig. 2A). Pol II was allowed to bind to primer/template DNA and extend a 
P-labeled primer molecule, at most once, by carrying out the reaction in the presence of excess unlabeled DNA and heparin
to trap free polymerase molecules (“Experimental Procedures”). The primer extension reaction in the absence of trap is shown
in lane 1, where pol II was able to engage and elongate primer/template DNA multiple times. The effectiveness of the trap is seen in
lane 8, where the addition of trap prior to initiation of the polymerization reaction, by addition of Mg
, was successful in eliminating detectable elongation of 
P-labeled primer strands. Increased levels of pol II resulted in increased primer utilization (lanes 3-7), and the fraction of primer/template DNA extended is plotted as a function of pol II concentration in Fig. 2B. From the slope of the line we conclude that at least 50% of the pol II molecules were active. The spots corresponding to unelongated primers (P) are overexposed (Fig. 2A), but we have verified by integrating all of the bands on the gel that the disappearance of intensity in the P band is converted
quantitatively into primer extension bands. Since not all polymerase binding events are productive, i.e. a bound pol II-DNA complex might dissociate prior to elongation followed by trapping of the dissociated enzyme, the observed
molar ratio of extended DNA/pol II (
0.5) represents a lower limit for the fraction of active pol II (Fig. 2B). The same criteria was used to verify that at least half of the purified exonuclease-deficient pol II ex1 (D155A/E157A)
molecules were active (data not shown).
Active site titration of pol II. A, lane 1, primer extension was carried out using 2.5 nM pol II in the absence of DNA-heparin trap (see “Experimental Procedures”);
lane 2, 46 nM pol II incubated with labeled primer/template DNA and all four dNTP substrates; lane 3, primer extension was carried out using 3 nM pol II in the presence of a DNA-heparin trap; lanes 4-7, same as lane 3 using pol II at 6, 12, 24, 46 nM, respectively; lane 8, 46 nM pol II was preincubated with DNA-heparin trap and all four dNTP substrates and the reaction was then initiated by
addition of Mg
. The primer/template DNA was present at 50 nM. Lane 2 serves as a control showing that trace amounts of Mg
present in the primer/template DNA solution are insufficient to cause measurable primer elongation. Lane 8 is a control demonstrating the effectiveness of the DNA-heparin trap; preincubation with trap eliminates elongation of 5′-
P-labeled primer molecules. B, plot showing the amount of active pol II-DNA complex (given by the amount of primer extended) versus the amount of pol II present in the reaction. The minimum fraction of active purified pol II, given by the slope of the line, is equal to 50%.
A Comparison of Exonuclease Activities of Pol II, T4 Pol, and Pol I(KF)
It was of interest to compare the polymerase and 3′-exonuclease proofreading activities of pol II and bacteriophage T4 pol
because both enzymes are class B polymerases (human α-like polymerases) sharing five highly conserved sequence motifs(4). Wild-type T4 pol has an exceptionally high 3′-exonuclease activity and is able to excise 10-20% of correctly paired nucleotides(26, 27, 28), depending on the identity of the the base pair and sequence context (29). Equal polymerase units of pol II, T4 pol, and
KF were incubated in the presence of a 5′-
P-labeled single-stranded 17-mer without dNTPs (Fig. 3). Degradation of the 17-mer occurred more rapidly for pol II than T4 pol, with full-length input primer disappearing within
1 min of incubation at 37°C. In the case of KF, the primer remained essentially intact for incubations of up to 10 min. Degradation
rates were determined to be 6.9 
10
mol/min, 4.7 
10
mol/min, and 1.5 
10
mol/min for pol II, T4 pol, and KF, respectively.
Comparison of pol II, T4 pol, and KF 3′-exonuclease activities on single-stranded DNA. A 5′-
P-labeled single-stranded DNA oligonucleotide (17-mer) was incubated with pol II, T4 pol, and KF for the reaction times shown.
The concentrations of the three enzymes were adjusted to have the same number of polymerase units, equivalent to 0.04 pol
II units. The first lane on the left-hand side indicates the position of undegraded 5′-
P-labeled single-stranded 17-mer.
A measurement of 3′-exonuclease activity under synthesizing conditions (dNTP 
dNMP turnover), was carried out for insertion of a correct (dCTP, Fig. 4A) and an incorrect (dTTP, Fig. 4B) deoxynucleotide opposite G by pol II, T4 pol, and KF. An increase in dNTP 
dNMP turnover as a function of incubation time was observed for each enzyme. Using equal polymerase units, turnover of both
correct and incorrect inserted dNMP's were slightly larger for pol II than T4 pol, with both polymerases exhibiting far greater
proofreading activity than KF. The radioactive spot migrating in between the dense spot of unreacted α-
P-labeled dNTP substrate and product dNMP is dNDP, which likely resulted from the spontaneous hydrolysis of dNTP. The level
of dNDP is about 0.5% of the input dNTP and remained constant during the course of the reaction. A small amount of stable
correct incorporation of dCMP was observed as an extended primer band (Fig. 4A), but there was no detectable band corresponding to the stable misincorporation of dTMP (Fig. 4B). Turnover rates of the correct nucleotide (dCTP) were about 6.7 
10
mol/min, 5.6 
10
mol/min, and 1.1 
10
mol/min for pol II, T4 pol, and KF, respectively (Fig. 4A). For the incorrect nucleotide (dTTP), turnover rates were 5.6 
10
mol/min, 2.3 
10
, and 1.5 
10
for pol II, T4 pol, and KF, respectively (Fig. 4B). The rates of turnover of incorrect nucleotide per polymerase unit were less than correct nucleotide because dNTP 
dNMP turnover measures insertion followed by excision, and insertion of incorrect nucleotides occur much less frequently
than correct nucleotides. For these experiments, a 3-fold higher concentration of each of the polymerases was used for measurements
of incorrect compared to correct turnover.
Exonucleolytic Degradation of Single-stranded DNA
We compared the ability of wild-type pol II and the pol II ex1 mutant (D155A/E157A) to degrade single-stranded DNA (Fig. 5). During incubations with equivalent polymerase units, degradation of a 28-mer was detected within a 5-s incubation with the wild-type enzyme, but was just barely detectable after 80 s with the exonuclease-deficient mutant (Fig. 5). The residual level of 3′-exonuclease activity in the mutant enzyme was calculated to be approximately 0.1% of the wild-type activity, based on the fraction of 28-mer remaining intact.
Comparison of 3′-exonuclease activities of wild-type pol II and an exonuclease-deficient (exo
) pol II mutant (D155A/E157A). A, wild-type pol II (0.4 polymerase units) and B, pol II exo
mutant (4 polymerase units) were used to hydrolyze a single-stranded 5′-
P-labeled single-stranded DNA oligomer as described under “Experimental Procedures.” The DNA oligomer (28-mer) was present
at a concentration of 180 nM.
Fidelity of Wild-type and Exonuclease-deficient Forms of Pol II
The fidelity of both pol II forms was determined during synthesis to fill a 390-nucleotide single-stranded gap containing the lacZα complementation reporter gene sequence. Gap-filling synthesis by the wild-type polymerase in a reaction containing 50 μM dNTPs generated products that had a lacZ mutant frequency that was elevated about 2-fold compared to an uncopied control DNA substrate (). When a wild-type pol II reaction was performed using 1 mM dNTPs, the mutant frequency increased to a value more than 10-fold higher than the uncopied control value. This increase is consistent with stimulation of polymerization at the expense of exonucleolytic proofreading(28, 30). The 2-fold higher mutant frequency observed at 1 mM dNTPs is consistent with a dNTP-dependent increase in extension of mispaired and misaligned intermediates(31). In comparison, pol II ex1 generated substantially higher lacZ mutant frequencies than did the wild-type enzyme at both dNTP concentrations ().
In order to determine the nature of the errors generated by pol II ex1 and to determine quantitative error rates and the contribution
of the exonuclease to fidelity, collections of independent lacZ mutants from reactions containing 50 μM dNTPs were analyzed
by DNA sequence analysis. The data are summarized in I, and the distribution of mutants containing single base substitutions
and frameshifts are shown in Fig. 6. Of 54 mutants from reactions catalyzed by wild-type pol II, 33 contained a single C
T transition mutation. These mutants, which account for most of the increase in mutant frequency above the control value ()
are likely the result of cytosine deamination, with wild-type pol II “correctly” inserting dAMP opposite template U residues.
Most of the other classes of mutants from the wild-type pol II reaction occurred at background frequencies, suggesting that
they were not the result of synthesis errors in vitro. The frequency data were used to calculate (see legend to I) “less than or equal to” error rates per detectable nucleotide
polymerized, for the single base errors and the two nucleotide deletions.
Spectra of single-base errors by E. coli DNA polymerase II. Three lines of primary DNA sequence for the lacZ α-complementation gene in M13 mp2 are shown. The sequence
is that of the viral (+) template strand. Position +1 is the first transcribed base. The spectrum above the lines of sequence is of mutants generated by wild-type enzyme, while that below the lines of sequence is of mutants generated by the exonuclease-deficient polymerase (pol II ex1). In both cases the reactions
contained 50 μM of each dNTP. Substitutions are indicated by a single letter and are the new base found in the viral DNA.
For frameshifts, the loss of a base is indicated by a Δ, while the addition of a base is indicated by an 
. When frameshifts occur at iterated nucleotides, it is not possible to distinguish which base was lost or added.
The only mutants observed in the wild-type pol II collection having a frequency above that of the uncopied control values
were the eight deletions of more than two nucleotides (Fig. 7, top). None of these mutants has been observed previously in sequence analysis of over 200 mutants from uncopied controls, suggesting
that they are errors generated by wild-type pol II. These mutants had lost between 49 and 268 nucleotides and in each instance
the deletion end points were flanked by short repeated sequences. One particular mutation, a 182-nucleotide deletion between
a perfect 7-base pair direct repeat, was recovered four times in this small collection. Only one of the eight mutants was
flanked by perfect repeat shorter than 5 base pairs, and even here the repeat homology is higher if formation of G
T pairs in the repeat is permitted (see legend to Fig. 7 and “Discussion”).
Deletion mutations recovered from reactions with wild-type and exonuclease-deficient pol II. Shown are lacZ mutants recovered
from pol II reactions wherein 22 or more nucleotides have been deleted. Nucleotide +1 is the first transcribed nucleotide
of the gene. The deletion end points are indicated by slashes, with the number of nucleotides deleted as listed and the homology at the deletion end points indicated in the last column.
For example, the first deletion listed for wild-type pol II had lost 268 nucleotides, from nucleotide −126 through nucleotide
142, and a perfect 8-base pair direct repeat sequence was present at the deletion end points. Complementarity at the end points
is indicated by capital letters, while non-complementary nucleotides are indicated by lower case letters. Underlined lower case letters indicate positions where a G
T or T
G mispair would be located in a putative misaligned intermediate formed during synthesis proceeding from right to left using
the template strand shown. Numbers in parentheses under “Homology” indicate homologies obtained by including the involvement of these two types of mispairs. Numbers in parentheses under “Bases Lost” indicate the number of times the particular mutation was recovered.
The collection of mutants from the pol II ex1 reaction was much more diverse, containing a variety of single base substitutions and frameshifts and deletions of two or more nucleotides (I, Figs. 6 and 7). Rates calculated for the single-base substitution and frameshift errors and two-base deletions are higher by values ranging from ≥13-fold to ≥240-fold for exo-pol II than for wild-type pol II (I), consistent with a role for the exonuclease in proofreading each of these classes of errors. Among deletions involving the loss of more than two bases, 10 had lost more than 20 nucleotides, and each of these had substantial but often imperfect homology at the deletion end points (Fig. 7, bottom). The others (not shown) had lost fewer than 10 nucleotides and most had limited homology at the end points.
DISCUSSION
Although DNA polymerase II was discovered in 1970 (3) its precise role in DNA replication or repair has been difficult to establish. Pol II contains both 5′-polymerase and 3′- exonuclease activities (1) on a single 89.9-kDa polypeptide chain(4, 5, 32). It has been shown that pol II can bypass abasic lesions in vitro(20) and is regulated by the LexA repressor (4, 5, 20) as part of the SOS regulon of E. coli(33). Recent experiments by Tessman and Kennedy (12) suggest that pol II may be required for bypassing abasic lesions in vivo under conditions where heat shock proteins are not induced. We have presented evidence for the involvement of pol II in the repair of or tolerance to oxidative damage and in influencing the rate of adaptive mutation(10).2 The purification and properties of wild-type pol II and an exonuclease-deficient pol II mutant (pol II ex1) provide additional insights into the role of this polymerase in cells.
A purification scheme has been presented, which makes use of an overexpression system for wild-type pol II and exonuclease-deficient enzyme, that can be carried out rapidly, yielding large quantities of highly purified polymerase that has been used to prepare single crystals suitable for x-ray diffraction analysis(25). Based on an active site titration of our preparations (Fig. 2), we have determined that at least 50% of the purified wild-type and exonuclease-deficient pol II molecules were active in extending primer/template DNA in the presence of a DNA-heparin trap. This 50% value is a minimum estimate of the fraction of active enzymes molecules since the assay does not detect those enzymes that dissociated from the primer/template DNA before incorporating a deoxynucleotide.
The specific activity of our most highly purified pol II fraction (Fraction IV) was calculated to be 1.8 
104 units/mg. Recently, another laboratory has reported a specific activity for purified DNA polymerase II which is nearly 11,000
times larger(13). This reported value is undoubtedly in error since their steady state rate exceeds the fastest measured presteady state
synthesis rate by 4-5-fold(34). Our specific activity measurements are consistent with values for other DNA polymerases. As we have previously reported,
DNA polymerase II can be made more processive by the addition of DNA polymerase III accessory factors which increase the specific
activity of pol II by preventing dissociation from the DNA template(8).
Bacteriophage T4 polymerase contains a highly active 3′-exonuclease proofreading activity(26). Pol II and T4 pol have been classified as group B “α-like” polymerases because of their similarity in sequence to five conserved regions of eucaryotic pol α(4, 35). It is of considerable interest to compare the relative levels of nuclease/polymerase activities for pol II and T4 pol, not only because of their sequence similarities, but also because of their effects on mutagenesis. Mutations in T4 pol that decreased nuclease-to-polymerase ratios in vitro were shown to confer mutator phenotypes for base substitution mutations in vivo and conversely mutants that increased nuclease/polymerase ratios conferred antimutator phenotypes(17, 26, 36, 37).
Pol II appeared to degrade single-stranded DNA about 1.5-fold more rapidly than T4 pol and 400-500-fold more rapidly than
KF, when equivalent polymerase units were used for each enzyme (Fig. 3). The relative 3′-exonuclease activities for the three polymerases were also determined under DNA synthesizing conditions
by measuring the turnover of [α-
P]dNTP to [α-
P]dNMP (Fig. 4). The rates of turnover of correctly inserted dCMP opposite G were roughly similar for pol II and T4 pol, and about 60-fold
greater than observed for KF (Fig. 4A). Turnover rates of misinserted dTMP opposite G were similar for pol II and T4 and approximately 20-fold greater than for
KF (Fig. 4B). Thus pol II and T4 pol have essentially the same ratio of nuclease/polymerase activities even though the specific activity
of T4 pol is approximately 100-fold greater than pol II, which undoubtedly reflects its role as the replicative polymerase.
A pol II exonuclease-deficient (pol II ex1) derivative was constructed by introducing two amino acid changes (D155A/E157A) analogous to those made for a proofreading-deficient mutant of bacteriophage T4 (D112A/E114A). The exonuclease-deficient mutant pol II exhibited about a 1000-fold reduction in exonuclease activity compared to wild-type pol II (Fig. 5), and a 3-4-fold reduction in polymerase activity was also observed. The residual exonuclease activity that represents 0.1% of the normal level may be due to contaminant wild-type DNA polymerase II encoded by the chromosome.
Wild-type pol II is highly accurate, having base substitution and frameshift error rates of ≤10
(I). In contrast, the exonuclease-deficient polymerase generated a variety of different errors during DNA synthesis in vitro. Assuming that the two amino acid substitutions in the conserved exonuclease motif that are responsible for diminished exonuclease
activity do not affect the selectivity of the polymerase, the differences in error rates between the two polymerases under
identical reaction conditions represent the contribution of proofreading to fidelity for each type of error. The data suggest
that single base substitutions, single base additions, and a variety of deletion errors are proofread by the highly active
intrinsic 3′- exonuclease activity of pol II. The substitutions likely result from direct misinsertion followed by mispair
extension by the polymerase, while models involving misaligned template/primers initiated by strand slippage (38) or base misinsertion (39) have been presented that can account for the one- and two-base frameshift errors observed at repetitive and non-reiterated
sequences (I).
An unexpected and interesting result from the present study was the frequent generation of errors involving loss of a large
number of nucleotides flanked by directly repeated sequences (Fig. 7). The simplest model to account for these involves pol II synthesis through the first repeat, disruption of repetitive terminal
base pairs, formation of base pairs involving the newly-made DNA and the downstream direct repeat, and continued synthesis.
Deletions consistent with this model have been found in previous studies of the fidelity of several different exonuclease-deficient
DNA polymerases (reviewed in Refs. 40 and 41). However, here they are even observed above background mutant frequencies in
the lacZ mutant collection from the exonuclease-proficient pol II reaction (Fig. 7, top). In 7 of 8 such mutants, perfect direct repeat homology involving 5, 7, or 8 base pairs was present at the deletion end
points, with the eighth example being homologous for 9 of 10 base pairs (if one G
T pair is allowed). This suggests that, if the degree of homology is sufficient, deletions between direct repeats can be generated
by wild-type pol II despite the presence of a highly active exonuclease that successfully proofreads point mutations.
The frequency of deletions between direct repeats is substantially higher in the exonuclease-deficient pol II mutant collection (I) and, among those observed, several may involve imperfect direct repeats (Fig. 7, bottom). Intermediates involving imperfect repeats would contain mismatches in the double-stranded template/primer region that could slow polymerization and lead to proofreading by the wild-type enzyme. The frequency and specificity data thus suggest that some direct repeat-dependent polymerase errors are in fact proofread. It is even possible that the absence of exonuclease activity may promote such deletion errors, by a mechanism wherein unedited nucleotide misinsertion generates a terminus that frays and realigns with a downstream sequence that is more homologous by virtue of the original misinsertion. This idea is supported by numerous observations suggesting that one-base deletions are initiated by misinsertion followed by realignment with a downstream sequence (for review, see Ref. 40).
Our results demonstrate that the 3′-proofreading exonuclease of DNA polymerase II plays an important role in rectifying replication errors that can lead to frameshifts and deletions in vitro. Recently we have obtained evidence that strains deficient in the exonuclease function of pol II show a mutator phenotype in nondividing cells.2 These biochemical and genetic results suggest a fundamental role for DNA polymerase II in maintaining the sequence integrity of DNA.
Purification of E. coli DNA polymerase II from E. coli JM109
Cells were induced with isopropyl-1-thio-β-D-galactopyranoside to overproduce pol II.
Fidelity of wild-type and exonuclease-deficient forms of E. coli DNA polymerase II
Reactions (25 μl) contained 20 mM Tris-HCl buffer, pH 7.5, 2 mM dithiothreitol, 10 mM MgCl2, dATP, dTTP, dGTP, and dCTP as indicated, 30 fmol of (150 ng) gapped M13mp2 DNA and 5-14 units of DNA polymerase II. Reactions
were incubated for 60 min at 37°C and terminated by adding EDTA to 15 mM. Analysis of 15 μl of the reaction by agarose gel
electrophoresis showed that all reactions filled the gap to the extent that the DNA migrated coincident with the fully double-stranded
RFII standard (e.g. see Ref. 24). The mutant frequency of uncopied DNA ranges from 5.1 to 7.0 
10
.
Average error rates by class
Footnotes
-
↵* This research was supported by National Institutes of Health Grants GM21422, GM42554, and GM29558. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore by hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-
↵1 The abbreviations used are:
- pol I
- DNA polymerase I from E. coli
- pol II
- E. coli DNA polymerase II
- pol III
- E. coli DNA polymerase III
- KF
- Klenow fragment of pol I
- pol II ex1
- an exonuclease-deficient pol II mutant (D155A/E157A)
- T4 pol
- bacteriophage T4 DNA polymerase
- dNTP
- deoxyribonucleoside triphosphate
- kb
- kilobase pair.
-
↵2Foster, P. L., Gudmundsson, G., Trimarchi, J. M., Cai, H., and Goodman, M. F. (1995) Proc. Natl. Acad. Sci. U. S. A.92, in press.
-
↵3The polB gene originally described in Refs. 42 and 43 corresponds to the DNA polymerase II structural gene (Z-H. Qiu, K. McEntee, and M. F. Goodman, manuscript in preparation).
- © 1995 by The American Society for Biochemistry and Molecular Biology, Inc.
