PCR fidelity of
Pfu
DNA polymerase and other thermostable DNA polymerases
PCR fidelity of Pfu DNA polymerase and other thermostable DNA polymerases
Janice
Cline
,
Jeffery C.
Braman
and
Holly H.
Hogrefe*
Stratagene Cloning Systems, 11011 North Torrey Pines Road,
La Jolla
, CA 92037,
USA
Received June 12, 1996;
Revised and Accepted July 29, 1996
ABSTRACT
The replication fidelities of
Pfu, Taq, Vent, Deep Vent
and
UlTma
DNA polymerases were compared using a PCR-based forward mutation assay. Average error rates (mutation
frequency/bp/duplication) increased as follows:
Pfu
(1.3*10-6) <
Deep Vent
(2.7*10-6)
<
Vent
(2.8*10-6) <
Taq
(8.0*10-6) << exo-
Pfu
and
UlTma
(~5*10-5). Buffer optimization experiments indicated that
Pfu
fidelity was highest in the presence of 2-3 mM MgSO
4
and 100-300
[mu]
M each dNTP and at pH 8.5-9.1. Under these conditions, the error rate of exo
-
Pfu
was
~
40-fold higher (5*10-5
) than the error rate of
Pfu
. As the reaction pH was raised from pH 8 to 9, the error rate of
Pfu
decreased
~
2-fold, while the error rate of exo
-
Pfu
increased
~
9-fold. An increase in error rate with pH has also been noted for the
exonuclease-deficient DNA polymerases
Taq
and
exo
-
Klenow, suggesting that the parameters which influence replication error rates
may be similar in pol I- and
[alpha]
-like polymerases. Finally, the fidelity of `long PCR' DNA polymerase
mixtures was examined. The error rates of a
Taq
/
Pfu
DNA polymerase mixture and a Klen
taq
/
Pfu
DNA polymerase mixture were found to be less than the error rate of
Taq
DNA polymerase, but
~
3-4-fold higher than the error rate of
Pfu
DNA polymerase.
INTRODUCTION
The use of high fidelity DNA polymerases in the polymerase chain reaction (PCR)
is essential for reducing the introduction of amplification errors in PCR
products that will be cloned, sequenced and expressed. Several thermostable DNA
polymerases with 3' -> 5' exonuclease-dependent proofreading activity
(Pfu
,
Vent
,
Deep Vent
and
UlTma
) have been introduced for high fidelity PCR amplification (
1
-
3
). Flaman
et al
. have reported that the error rate of
Pfu
was 5- and 30-fold lower than the error rates of the proofreading enzymes
Deep Vent
and
UlTma
, respectively (
4
). Using several different fidelity assays, the error rate of
Pfu
has been found to be ~10-fold lower than that of the non-proofreading enzyme
Taq
(
1
,
4
,
5
).
The parameters which contribute to the replication fidelity of DNA polymerases
need to be investigated, as very little is known about the molecular features
of these enzymes which give rise to variations in replication fidelity and
mutational spectra. A number of factors are thought to contribute to the
overall fidelity of a DNA polymerase (reviewed in
6
-
8
). These parameters include the tendency of a polymerase to incorporate
incorrect nucleotides and the presence of an integral 3' -> 5' exonuclease activity which can remove mispaired bases
(proofreading activity).
The importance of proofreading activity to replication fidelity has been
demonstrated for both the Klenow fragment (
9
) and for
Vent
polymerase (
10
), which exhibit 10- and 5-fold increases in error rates, respectively, when the associated 3' -> 5' exonuclease activity is inactivated. The
contribution of proofreading activity to DNA polymerase fidelity is also
evident when the error rates of proofreading and non-proofreading enzymes are compared. Kunkel has noted that the average base
substitution error rates exhibited by non-proofreading DNA polymerases range from 10
-2
to >= 10
-6
, while the error rates of proofreading enzymes range from 10
-6
to 10
-7
(
7
). The parameters which contribute to error rate variations among proofreading
enzymes may reflect inherent differences in 3' -> 5' exonuclease activity, the tendency to discriminate
mispaired versus correctly paired bases and/or the efficiency of shuttling
between polymerization and proofreading modes.
Recently, mixtures of non-proofreading and proofreading DNA polymerases have been reported to
synthesize higher yields of PCR product and to allow amplification of longer
templates than is possible with single enzyme formulations (`long PCR') (
5
). The addition of a low level of a proofreading enzyme (e.g.
Pfu
DNA polymerase) to PCR reaction mixtures has been proposed to improve the
performance of non-proofreading polymerases (e.g.
Taq
DNA polymerase) by correcting mismatches introduced during PCR which prevent
the efficient synthesis of full-length products (
5
). The PCR fidelity of DNA polymerase mixtures has not yet been determined, but
error rates are likely to reflect the fidelity of the component polymerases and
the ratio of non-proofreading to proofreading enzyme activities.
Pfu
DNA polymerase has been found to be useful in high fidelity amplifications (
1
,
4
) of DNA targets up to 25 kb (K. Nielson, personal communication). In this report we use the previously described
lacI
PCR mutation assay (
1
) to compare the error rate of
Pfu
with an expanded number of PCR polymerases, including exo
-
Pfu
,
Deep Vent
,
Vent
,
UlTma
and
Taq
, as well as `long PCR' DNA polymerase mixtures. Polymerase error rates have
been found to vary with buffer composition, including pH, Mg
2+
concentration and nucleotide concentration (
11
-
13
). PCR reaction conditions have been optimized with respect to fidelity for both
Vent
and
Taq
DNA polymerases (
11
). Buffer optimization studies with
Pfu
DNA polymerase were performed here to assess whether the fidelity of
Pfu
DNA polymerase could be further enhanced. Error rate comparisons between
Pfu
and exo
-
Pfu
are expected to illuminate the contribution of proofreading activity to the
fidelity of
Pfu
DNA polymerase. Finally, PCR fidelity comparisons between
Pfu
DNA polymerase and
Pfu
-containing DNA polymerase mixtures will allow evaluation of the
contribution of the predominant non-proofreading enzyme to the error rate of `long PCR' mixtures.
MATERIALS AND METHODS
DNA polymerases
Cloned
Pfu
, exo
-
Pfu
and
Taq
DNA polymerases were prepared at Stratagene.
Deep Vent
and
Vent
polymerases were purchased from New England BioLabs,
UlTma
was obtained from Perkin-Elmer and KlentaqLA (KTLA) was provided by Wayne Barnes (Washington
University School of Medicine, St Louis, MO). Except where indicated, PCR
amplifications were performed in the presence of buffers supplied by the
manufacturers. The KTLA PCR buffer used was buffer PC2 of Barnes (
5
).
PCR fidelity assay
The fidelity of DNA replication during PCR was measured using a previously
described assay (
1
,
14
). Briefly, a 1.9 kb sequence encoding
lacIOZ
[alpha] was PCR amplified as described below with oligonucleotide primers
containing 5'
Eco
RI restriction sites (
1
). The amplified fragments were digested with
Eco
RI, purified by gel electrophoresis and ligated into [lambda]gt10 arms. The ligation reactions were packaged and the [lambda] phage used to infect an [alpha]-complementing
Escherichia coli
host strain. Aliquots of infected cells were plated on LB plates with top agar
containing either X-gal (1 mg/ml) or X-gal plus IPTG (1.5 mM). Error rates were calculated as described in
the legend to Table
1
.
PCR amplifications
Except where indicated, PCR amplifications were performed in 100 [mu]l reaction volumes in the presence of the appropriate Tris-based buffer, using 5 U polymerase, 200 [mu]M each dNTP, 250 ng each primer and 24 ng
lacIOZ
[alpha] target (50 ng
lacIOZ
[alpha] plasmid template). The PCR mixtures were denatured by heating at 95oC for 30 s. Thirty cycles of amplification were performed using the
following conditions: 5 s at 95oC; 1 min at 55oC; 2.5 min at 72oC.
RESULTS
PCR fidelity of thermostable DNA polymerases
Replication fidelities of thermostable DNA polymerases were compared using a
previously described assay (
1
) which measures the frequency of mutations introduced into the
lacI
target gene during PCR amplification. PCR amplification was performed in the
presence of each enzyme's optimal PCR buffer. All other PCR parameters remained
constant, including the dNTP, primer and template concentrations, the PCR
cycling parameters and the number of PCR cycles performed.
Pfu
DNA polymerase exhibited the greatest PCR fidelity, with an average error rate of 1.3 * 10
-6
mutation frequency/bp/ duplication (Table
1
). The
lacI
target size used in these calculations was estimated to be 349 bp, based upon
the most recent analysis of
lacI
-
mutant DNA sequences (
15
). Previous error rate calculations assumed a
lacI
target size of 182 bp (
1
). After recalculating error rates based on a
lacI
target size of 349 bp, the mean error rate of
Pfu
DNA polymerase obtained in this study (1.3 * 10
-6
mutation frequency/bp/duplication) was found to be similar to previous
estimates obtained using an identical assay (0.8 * 10
-6
;
1
) or an alternative PCR-based assay employing a p53 target gene ( <= 1.0 * 10
-6
;
4
).
.
Average error rates of thermostable DNA polymerases during PCR
a
DNA polymerase
No. of PCRs
Target (ng)
Template doublings
b
lacI
-
plaques
c
(% +- SD)
Error rate
d
(*10
-6
+- SD)
Pfu
10
24
9.7
0.42 +- 0.08
1.3 +- 0.2
2
2
12.7
0.30 +- 0.06
e,f
0.7 +- 0.1
e
2
0.2
16.0
0.47 +- 0.03
e,f
0.8 +- 0.02
e
2
0.02
19.4
0.66 +- 0.03
e,f
1.0 +- 0.04
e
Deep Vent
4
24
9.7-10
0.9 +- 0.1
2.7 +- 0.2
Vent
6
24
8.7-10
0.9 +- 0.3
2.8 +- 0.9
Taq
11
24
8.7-11
2.7 +- 1.1
8.0 +- 3.9
UlTma
2
24
9.7
18.8 +- 0.8
e
55 +- 2
e
a
PCRs were performed in each manufacturers' recommended buffer (all pH 8.8) in
the presence of 200 [mu]M each dNTP and 2 mM MgSO
4
(
Pfu
,
Deep Vent
and
Vent
), 2 mM MgCl
2
(
UlTma
) or 1.5 mM MgCl
2
(
Taq
).
b
Template doublings (
d
) were determined using the equation 2
d
= (amount of PCR product)/( amount of starting target). 24 ng
lacI
target corresponds to 50 ng
lacIOZ
[alpha] plasmid template. The range of
d
obtained is indicated.
c
Mutant frequencies (
mf
) were determined by dividing the total number of blue plaques (
lacI
-
mutants) on X-gal plates by the total number of plaques containing a functional
lacZ
[alpha] sequence (blue plaques on X-gal plus IPTG plates).
d
Error rates were calculated using the equation
ER
=
mf
/(
bp
*
d
), where
mf
is the mutation frequency
,
bp
is the number of detectable sites in
lacI
(=349; 15) and
d
is the number of template doublings.
e
+- indicates range of duplicate measurements.
f
Mutation frequencies for
Pfu
amplification of 0.02-2 ng target were normalized such that the mean mutation frequency for
Pfu
amplification of 24 ng target (assay internal control) was 0.42%.
Average error rates of thermostable DNA polymerases were found to increase in
the following order:
Pfu
(1.3 * 10
-6
)
<
Deep Vent
(2.7 * 10
-6
) <
Vent
(2.8 * 10
-6
) <
Taq
(8.0 * 10
-6
) <<
UlTma
(5.5 * 10
-5
). These results are in excellent agreement with the relative error rates
measured by Flaman
et al
. (
4
), who reported that
Pfu
exhibits the greatest PCR fidelity, followed by
Deep Vent
,
Taq
and
UlTma
DNA polymerases. The relative error rates obtained here are also consistent
with DGGE analyses showing that
Pfu
exhibits a lower error rate than
Vent
and
Taq
DNA polymerases (
16
). We found that relative error rates observed using the
lacI
screening assay were consistent from PCR reaction to PCR reaction.
The influence of template doublings (
d
) on error rate estimates of
Pfu
DNA polymerase was also examined (Table
1
). Amplification reactions described above and resulting in the
Pfu
error rate of 1.3 * 10
-6
employed 24 ng
lacI
target (10
10
copies). Approximately 10 doublings were observed in 30 PCR cycles. When the
input
lacI
target DNA was decreased from 10
10
copies (24 ng) to 10
7
copies (0.02 ng), the number of template doublings increased from 9.7 (~900-fold amplification) to 19.4 (~700 000- fold amplification) after 30 cycles of PCR. The error
rate of
Pfu
DNA polymerase varied from 0.7 to 1.3 * 10
-6
over the 1000-fold range of DNA target concentrations tested. Flaman
et al
. have also reported that polymerase error rates do not appear to be
significantly influenced by the number of template doublings (
4
).
Optimization of the PCR fidelity of
Pfu
We attempted to further improve the fidelity of
Pfu
by optimizing PCR reaction conditions. PCR error rates were measured at varying
concentrations of MgSO
4
(Fig.
1
) and dNTPs (Fig.
2
) and at varying pHs (Fig.
3
). The indicated pH values are those measured at room temperature. Where noted,
the pH of Tris buffers at elevated temperatures was estimated using the
formula: pH
T
= pH
25oC
+ [(
T
oC - 25oC) * (-0.03 pH U/oC)] (where
T
is the reaction temperature;
17
). The lowest error rates for
Pfu
were observed when PCR amplifications were performed in the presence of 2-3 mM MgSO
4
, 100-300 [mu]M each dNTP and in a pH range between 8.5 and 9.1 (pH ~7.1-7.7 at 72oC). These conditions have been found to give optimal
yield of PCR product as well (
18
).
pH dependency of the fidelity of
Pfu
and exo
-
Pfu
The error rates of
Pfu
and exo
-
Pfu
were compared to assess the contribution of 3' -> 5' exonuclease activity to fidelity. In the presence of
Pfu
PCR buffer (2 mM MgSO
4
, 200 [mu]M each dNTP, pH 8.8), exo
-
Pfu
exhibited an error rate of 4.7 * 10
-5
mutation frequency/bp/duplication, which is ~40-fold higher than that determined for exonuclease-proficient
Pfu
.
Figure
3
shows the error rate variation of exo
-
Pfu
and
Pfu
as a function of pH. Exo
-
Pfu
shows a dramatic increase in error rate (~9-fold) as the reaction pH is raised from pH 8 to 9.1 (or from 6.6 to
7.7 at 72oC). In contrast to exo
-
Pfu
, the error rate of
Pfu
decreased ~2-fold in this pH range. Presumably, the fidelity of
Pfu
is maintained at high pH (pH 9) by enhanced proofreading activity, which
accompanies the dramatic increase in nucleotide misincorporation occurring
between pH 8 and 9.1 (identified using exo
-
Pfu
)
.
These results and those reported by others for
Taq
and exo
-
Klenow (
11
-
13
) indicate that the error rates of exonuclease-deficient enzymes,
Taq
, exo
-
Klenow and exo
-
Pfu
, are similarly increased by pH. The significance of the apparent biphasic
relationship between error rate and pH is currently under investigation.
PCR fidelity of `long PCR' DNA polymerase mixtures
The fidelities of
Pfu
and
Taq
DNA polymerases were compared with the fidelities of two
Pfu
-containing DNA polymerase mixtures (Table
2
). A
Taq
/
Pfu
(16 U:1 U) mixture was prepared and shown to amplify DNA targets >30 kb (data
not shown). The
Taq
/
Pfu
mixture exhibited an average error rate of 5.6 * 10
-6
mutation frequency/bp/duplication when amplifications were performed in
Taq
PCR buffer. The mean error rate of the
Taq
/
Pfu
mixture was 30% lower than the mean error rate of
Taq
DNA polymerase when amplifications were conducted in
Taq
PCR buffer. When compared with the error rate of
Pfu
DNA polymerase in the same buffer system, the error rate of the
Taq
/
Pfu
mixture was found to be 6-fold higher.
Similar observations were made for a second `long PCR' mixture, KTLA, which
consists of Klentaq (N-terminally truncated
Taq
) and
Pfu
DNA polymerases (
5
). When PCR amplifications were conducted as described in this report, KTLA
exhibited a mean error rate of 3.9 * 10
-6
mutation frequency/bp/duplication, which was 3-fold higher than the error rate of
Pfu
DNA polymerase (Table
2
). When PCR conditions from Barnes (
5
) were used (Table
2
, condition 2), KTLA exhibited a mean error rate (9.4 * 10
-6
) which was 4-fold higher than the error rate of
Pfu
DNA polymerase assayed under identical conditions.
.
Error rate comparisons of DNA polymerases and `long PCR' DNA polymerase
mixtures
PCR condition
a
DNA polymerase
No. of PCRs
Template doublings
b
Error rate (* 10
-6
+- SD)
b
1
Pfu
10
8.7-9.7
1.3 +- 0.2
Taq
11
8.7-11
8.0 +- 3.9
Taq
/
Pfu
(16:1)
c
Taq
buffer
8
9.7-10
5.6 +- 1.6
Pfu
buffer
11
9.7-11
7.6 +- 1.2
KTLA
2
9.7
3.9 +- 0.1
d
2
Pfu
2
8.1
2.3 +- 0.2
d
KTLA
2
8.1
9.4 +- 0.9
d
a
PCR condition 1 is described in Materials and Methods, PCR amplification. PCR
condition 2 is from Barnes (5) and differs in the following respects. PCR
amplifications were performed on a Robocycler 40 (Stratagene) using thin-walled PCR tubes and 7.2 ng target (15 ng
lacIOZ
[alpha] plasmid). Sixteen cycles of amplification were performed using the
following conditions: 30 s at 99oC, 30 s at 67oC, 3 min at 68oC.
b
Defined in the legend to Table 1.
c
The
Taq/Pfu
mixture consists of
Taq
(5 U/[mu]l) and
Pfu
(0.31 U/[mu]l) DNA polymerases.
d
+- indicates range of duplicate measurements.
DISCUSSION
The intrinsic properties of thermostable DNA polymerases which contribute to
variation in PCR fidelity are not fully understood. In general, enzymes which
possess an associated 3' -> 5' exonuclease-dependent proofreading activity are thought to exhibit
higher replication fidelity than non-proofreading DNA polymerases (
7
). Variation in fidelity among proofreading enzymes, such as
Pfu
,
Vent
and
Deep Vent
, may reflect differences in the rate of mispair excision, the level of
discrimination between mispaired and correctly paired bases, the rate of
mispair extension and/or the efficiency of shuttling the 3' primer terminus between the polymerase and exonuclease active sites.
The contribution of 3' -> 5' exonuclease activity to the PCR fidelity of
Pfu
was demonstrated directly by comparing the error rates of
Pfu
and exo
-
Pfu.
The error rate of exo
-
Pfu
was found to be 7-fold higher than the error rate of exo
+
Pfu
at pH 8.0 and 40-fold higher at pH 8.8 (
Pfu
PCR buffer).
Despite the importance of proofreading activity to the fidelity of
Pfu
and
Vent
(
10
), the presence of 3' -> 5' exonuclease activity does not necessarily guarantee high
fidelity DNA synthesis, as illustrated by
UlTma
DNA polymerase. The poor fidelity of
UlTma
DNA polymerase may be related to the relatively low level of 3' -> 5' exonuclease activity exhibited by this enzyme. In a
preliminary analysis of exonuclease activity,
UlTma
was found to exhibit significantly lower levels of 3' -> 5' exonuclease activity than
Pfu
,
Deep Vent
and
Vent
DNA polymerases (A. Lovejoy, personal communication). However, other parameters
are likely to contribute to low fidelity, since
UlTma
, an N-terminally deleted version
of
Thermatoga maritima
DNA polymerase (
20
), exhibits an ~7-fold higher error rate than
Taq
, which is completely devoid of proofreading activity.
In the absence of proofreading activity, a DNA polymerase like
Taq
is thought to accomplish high fidelity DNA synthesis by inefficient
incorporation of non-complementary dNTPs and a reduced tendency to extend from mismatched 3' primer termini. Huang
et al.
(
21
) have shown that, with the exception of C-T mispairs,
Taq
polymerase exhibits ~100-1000-fold greater discrimination against mispair extension, as compared
with avian myeloblastosis and HIV-1 reverse transcriptases, which extend most mispairs permissively. The
rate at which DNA polymerases extend from mispaired 3' primer termini, however, does not contribute to the actual fidelity of
non-proofreading enzymes. The mismatch extension rate only contributes to
fidelity in the sense that if the mismatch is extended inefficiently, the DNA
will not be replicated to completion and the mutation will not be scored.
Therefore, the mispair extension rate influences the number of detected
mutants, rather than reflecting the inherent fidelity of a non-proofreading DNA polymerase.
The observed 6-fold difference in error rate between
Taq
(8 * 10
-6
) and exo
-
Pfu
(4.7 * 10
-5
)
suggests that the misincorporation and/or misextension rates of
Pfu
(as measured with exo
-
Pfu
)
are significantly higher than those of
Taq
. Apparently, a lower degree of discrimination against misinsertion or mispair
extension errors can be tolerated when an associated proofreading activity is
present, as is the case with exonuclease-proficient
Pfu
.
Further fidelity measurements with exo
-
Pfu
revealed that the fidelity of dNTP incorporation was significantly influenced
by the pH of the PCR buffer. The error rate increased by ~9-fold as the pH was raised from pH 6.6 to 7.7 (pH at 72oC). The error rates of both
Taq
(
12
) and exo
-
Klenow (
13
) increase similarly at higher pH. Eckert and Kunkel have attributed the lower
fidelity of exo
-
Klenow at high pH to an increase in both nucleotide misinsertion and mispair
extension (
13
).
It is tempting to speculate that the lower fidelity of exo
-
Pfu
at high pH may also reflect increased misinsertion and mispair extension,
analogous to the observations made for exo
-
Klenow (
13
). If so, it would suggest that the parameters which contribute to fidelity are
similar, despite the structural differences which are thought to exist between
the [alpha]-like (exo
-
Pfu
;
22
) and pol I-like (exo
-
Klenow and
Taq
) DNA polymerases. For example, the observed variation in error rates with pH
suggests that an active site histidine residue may play a role in fidelity,
possibly in the discrimination of mismatched 3' primer termini. Alternatively, protonation of the primer, template or
substrate dNTP may enhance error discrimination (
13
). Finally, pol I- and [alpha]-like polymerases may undergo a similar conformational change
at low pH which may alter template binding properties, thereby improving error
discrimination. Such a mechanism was proposed for exo
-
Klenow by Eckert and Kunkel (
13
) and was supported by additional data showing that lower error rates at low pH
were accompanied by an increase in polymerase processivity.
The relative error rates for
Pfu
,
Vent
and
Taq
were found to parallel the terminal transferase activities of DNA polymerases.
Hu (
23
) has compared the tendency of DNA polymerases to catalyze the addition of non-template-directed bases to the 3'-end of a DNA fragment (terminal transferase activity).
Terminal transferase activity is high in
Taq
but low (Klenow and
Vent
) or absent (
Pfu
,
T4
and
T7
) in proofreading enzymes, which presumably edit the misextended base. The
absence of terminal transferase activity appears to correlate with high
fidelity. Fidelity measurements compiled by Cha and Thilly show that the error
rates of
Pfu
,
T4
and
T7
DNA polymerases are lower than the error rates of
Vent
and Klenow (
16
). Thus, the parameters which give rise to terminal transferase activity may be
similar to those which contribute to lower fidelity. The lower error rate and
lack of terminal transferase activity for
Pfu
(as compared with
Vent
) may be the result of a reduced tendency of
Pfu
to incorporate a mismatch or a base opposite an abasic site. Alternatively,
Pfu
may excise misincorporated bases more readily or shuttle between the
exonuclease and polymerase active sites more efficiently.
Finally, fidelity comparisons with
Pfu
-containing `long PCR' DNA polymerase mixtures have shown that the error
rate of mixtures appears to be intermediate between the error rate of
Pfu
and the non-proofreading DNA polymerase. The lower error rate of a
Taq
/
Pfu
mixture, as compared with
Taq
alone, suggests that
Pfu
is editing a certain percentage of mismatches that have been introduced by
Taq
during the PCR process. Editing may occur at the 3'-terminus after
Taq
has introduced a mismatch and dissociated from the incomplete PCR product (
5
). In the absence of
Pfu
,
Taq
presumably extends some of these putative stalling mismatches during the course
of the PCR process; otherwise the mutations would not be scored in the
lacI
-
screening assay and there would be no apparent difference in error rate between
Taq
and the
Taq
/
Pfu
mixture.
Pfu
may also reduce the overall error rate of
Taq
DNA polymerase by degrading
Taq
-generated duplex DNA containing mismatches and resynthesizing the correct sequence.
Although the error rate of the
Taq
/
Pfu
mixture is somewhat lower than the error rate of
Taq
alone, it is still 4-6-fold higher than the error rate of
Pfu
alone (Table
2
). These results
indicate that the majority of PCR products are synthesized by
Taq
. This result is not surprising, since
Taq
is present in this particular mixture at a 16-fold higher polymerase unit concentration than
Pfu
DNA polymerase. Hence, the misincorporation rate of
Taq
DNA polymerase contributes significantly to the error rate of
Taq
/
Pfu
DNA polymerase mixtures.
KTLA, a `long PCR' mixture of Klentaq and
Pfu
DNA polymerases, was also found to exhibit an error rate significantly higher
than the error rate of
Pfu
. Our results are inconsistent with the results of Barnes (
5
), who has compared the error rates of
Pfu
, Klentaq and KTLA-64 (~640 U Klentaq:1 U
Pfu
) using a similar PCR forward mutation assay based on the mutational target gene
lacZ
. Barnes reported that the error rate of the KTLA mixture was 2-fold lower than the error rate of
Pfu
DNA polymerase (
5
). There are several differences between the Barnes assay and the assay
performed here, including PCR amplification conditions (see Table
2
legend), number of clones screened [500-4200 clones/1 PCR in Barnes (
5
) versus 10 000-50 000 clones/PCR/4 PCRs in this study] and the mutational target gene
used (
lacZ
versus
lacI
), as well as possible unknown variations in the KTLA mixtures. The results in
Table
2
demonstrate that differences in the PCR amplification conditions employed are
not likely to contribute to the differences in relative error rates observed in
the two studies. Fidelity analyses of additional DNA polymerase mixtures are currently under way to help elucidate
the role of component enzymes and buffer composition in the fidelity of `long PCR' amplifications.
