 |
INTRODUCTION |
Accurate replication by the mitochondrial DNA polymerase
(pol)1 
is achieved
through discrimination of the correct nucleotides during polymerization
and during error correction by the 3'-5' exonuclease, so that the
overall fidelity is a product of the contributions from each reaction.
Previous biochemical studies have indicated that pol exonuclease
exhibits a 4-15-fold preference for 3'-mismatched substrates (1-3).
However, the actual fidelity contribution of DNA polymerases is based
upon the kinetic partitioning between forward polymerization and
exonuclease error correction (4). Chick embryo pol was estimated to
have at least a 100-fold increase in fidelity due to exonuclease
activity based on the opal codon reversion assay (5) and was able to
correct 98% of the T/C2 and
99% of the A/C mispairs in a gap-filling polymerization assay (6).
Additionally, mutation of one residue required for yeast pol exonuclease activity resulted in 100-200-fold higher frequency of
spontaneous erythromycin-resistant mutants (1) and a 1500-fold increase
in chloramphenicol-, erythromycin-, and oligomycin-resistant mutants
due to single substitution errors in mitochondrial DNA (7).
Although, the contributions of pol exonuclease activity to faithful
mitochondrial DNA replication are apparent, little is known about the
mechanisms of selectivity and proofreading by pol exonuclease and
the role of the accessory subunit in error correction.
Our understanding of the mechanistic basis for exonuclease error
correction by polymerases is based in part on studies done on the T7
DNA polymerase (4). Selectivity of the exonuclease (the decision to
excise or not) occurs at the polymerase site. For correctly base-paired
DNA, forward polymerization is fast (300 s
1) relative to
the rate of transfer of correctly base-paired DNA from the polymerase
to the exonuclease active site (0.2 s1), and the
"cost" of proofreading is correspondingly small, 0.2/(300 + 0.2) = 0.06%. In contrast, after incorporation of a mismatch, the
rate of polymerization on top of a mismatch is reduced to 0.01 s1, while the rate of transfer of the DNA to the
exonuclease site is increased to 2.3 s1, so that the net
probability of excision is large, 2.3/(0.01 + 2.3) = 99.6%. Thus,
it is this 30,000-fold reduction in the rate of correct base pair
formation after a mismatch that leads to active site selectivity
favoring removal of a mismatch by allowing time for the transfer of the
mismatched DNA to the exonuclease active site. The probability of
excising a mismatch relative to the rate of polymerization over the
mismatch defines the overall contribution of the exonuclease to
polymerase fidelity. Since the rate of incorporation is determined by
the rate of the protein conformational change and a mismatched
primer-template inhibits this step and not ground state
nucleotide binding, it is clear that the conformational change provides
the checkpoint for polymerase editing (8).
In this report, we analyze the contributions of exonuclease
proofreading to fidelity by pol , following upon our analysis of the
rates of incorporation of correct and incorrect base pairs in the
previous article (23). Here we measured the rates of excision of
matched and mismatched DNA and determined the selectivity of the pol
3'-5' exonuclease for DNA containing incorrect base pairs. For
clarity, the kinetic parameters governing misincorporation determined
in the previous paper (23) and those describing proofreading determined
from the data to be presented here are summarized in Scheme
1. Single turnover conditions were used
to directly examine the selectivity and efficiency of the exonuclease
for removing various mismatches as well as the effectiveness of
proofreading in correcting those mismatches. Excision of DNA by pol is suggested to occur by an intramolecular strand transfer mechanism
(3, 9), and we provide direct evidence for this intramolecular transfer
without dissociation of the DNA from the enzyme.
View larger version (13K):
[in this window]
[in a new window]
|
Scheme 1.
Kinetic parameters of polymerization by DNA
pol . The starting DNA used in this
scheme was 25/45 DNA, and the species of DNA at each step is designated
by the resulting primer length (i.e. DNA25
signifies 25/45 DNA).
|
|
| |
EXPERIMENTAL PROCEDURES |
Experimental Set-up--
Preparation of enzyme and DNA, reaction
conditions, and product analysis are as described in the preceding
article (23). Since exonuclease studies required the use of wild-type
pol , only this form of the enzyme was used in the experiments
presented here. Pol holoenzyme was reconstituted by mixing a 1:5
ratio of the catalytic/accessory subunits. Protein concentrations are given based on the concentration of active holoenzyme determined by
active site titration. The synthetic oligonucleotides used in this
study are shown in Table I, which also
defines the nomenclature used to refer to each DNA.
A 27-mer containing a 3'-[
-32P]dCMP was created by
combining 1 µM pol , 10 µM 26/45-mer
(with a 45-mer to allow correct base pairing), 300 nM
[-32P]dCTP, and 50 µM dCTP in reaction
buffer for 30 min at 37 °C. Product 27-mer was purified by
electrophoresis through a 15% denaturing polyacrylamide gel and used
as described for 5'-32P-labeled 27-mer. A similar reaction
was used to make a 27-mer primer containing [-32P]dTMP
at the T:T mismatch position, followed by an unlabeled dCMP at the 3'
correct base pair position.
Reactions were performed using a RQF-3 rapid quench apparatus to
explore fast reaction times (KinTek Corp., Austin, TX). Product and
substrate DNAs were separated on 15% denaturing polyacrylamide gels
and analyzed using a PhosphorImager (Storm 860; Molecular Dynamics,
Inc., Sunnyvale, CA). The loss of full-length primer was monitored over
time. Alternatively, DNA products were separated from excised
3'-[-32P]dCMP or [-32P]dTMP by
polyethyleneimine-cellulose thin layer chromatography for a direct
measure of exonuclease activity.
Excision Reactions--
Enzyme (100 nM) was
preincubated with 75 nM DNA. Mg2+ was added to
initiate the reaction. The loss of full-length substrate primer due to
exonuclease hydrolysis was plotted against time and fit to a single or
double exponential, as appropriate. Alternatively, reactions were
initiated by the addition of Mg2+ and DNA to the enzyme as noted.
Proofreading and Extension of Mismatched DNA--
Enzyme (100 nM) was preincubated with 75 nM DNA.
Mg2+ and 25 µM dNTP were added to initiate
the reaction. Accumulation of DNA products was plotted against time and
fit to a single or double exponential, as appropriate. For experiments
analyzing incorporation of [-32P]dATP to
correct a mismatch, 16 nM radioactive nucleotide was added
to 25 µM unlabeled dATP to provide a signal.
| |
RESULTS |
Exonuclease Cleavage of Single-stranded DNA--
The rate of
hydrolysis of the 3'-base from a single-stranded DNA was measured by
mixing enzyme-25-mer with Mg2+ and monitoring loss of the
full-length substrate (formation of 24-mer and shorter). Excision of
single-stranded DNA terminated in dGMP, dTMP, or dCMP resulted in
single exponential kinetics with rates of excision of 0.17, 0.11, and
0.07 s1, respectively. Fig.
1 shows a representative trace for
removal of dGMP. Excision of primer terminated with dAMP fit a double exponential consisting of a brief lag followed by excision at a rate of
0.07 s1. The reason for the lag, occurring at a rate of
~0.17 s1, is not known.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 1.
Excision of single-stranded DNA.
Polymerase (100 nM) was preincubated with 75 nM
single-stranded 25-mer DNA. Mg2+ was added to initiate the
cleavage reaction. The percentage of 25-mer remaining was plotted
versus time and fit to a single exponential to yield an
excision rate of 0.16 ± 0.004 s1 for primer
terminated with a 3' dGMP ( ). Cleavage of 25-mer terminated
in dCMP or dTMP appeared similar, with rates of 0.073 ± 0.002 and
0.11 ± 0.004 s1, respectively. Cleavage of 25-mer
terminated in a 3' dAMP resulted in a double exponential fit to yield
rates of 0.17 ± 0.02 and 0.07 ± 0.004 s1 for
the fast and slow phases ( ). The fast phase defines the
kinetics of the reaction governing the lag phase (occurring with a
half-life of 4 s), while the slower phase defined the rate of
excision of 97% of the DNA occurring over the remainder of the time
course.
|
|
Exonuclease Cleavage of Duplex DNA Containing 0-7
Mismatches--
To characterize exonuclease activity on
double-stranded DNA, we examined excision of the 3'-terminal base from
the primer strand of correctly base-paired DNA or DNA containing one,
four, or seven terminal mismatches. Cleavage was examined after mixing Mg2+ with a preformed enzyme-DNA complex (Fig.
2A) or by mixing the enzyme
with both DNA and Mg2+ to initiate the reaction (Fig.
2B). Excision of correctly base-paired DNA occurred with
single exponential kinetics at a rate of 0.05 s1.
Cleavage of DNA containing 1, 4, or 7 mismatches was biphasic. The
rates and amplitudes of the fast phase increased from 1.1 ± 0.1 s1 (for 31% of the DNA) to 8.8 ± 0.8 s1 (42% of the DNA) and 9.2 ± 1.6 s1
(70% of the DNA) for DNA containing 1, 4, or 7 mismatches,
respectively, as summarized in Table
II. The excision rate appeared to
reach a maximum of ~9 s1 with 4-7 mismatches.
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 2.
Excision of DNA containing an increasing
number of mismatches. The effect of zero (), one (),
four ( ), and seven ( ) mismatches on the excision rate and
amplitude was examined. A, polymerase was preincubated with
75 nM 25/45 DNA containing an increasing number of
mismatches, and Mg2+ was added to initiate the cleavage
reaction. The percentage of 25-mer remaining was plotted against time
and fit to a single (no mismatch) or double (mismatches) exponential.
The rates and amplitudes derived from the data fits are listed in Table
II. The percentage of DNA excised at the fast rate increases as the
number of mismatches increases. B, a solution containing
Mg2+ and duplex DNA with one, four, or seven mismatches was
combined with polymerase to initiate the cleavage reaction. Data for
the single and double exponential fits are listed in Table II. Slower
rates of excision as well as lower amplitudes of the fast phase were
observed when the DNA and enzyme were not preincubated.
|
|
View this table:
[in this window]
[in a new window]
|
Table II
Excision rates and amplitudes for DNA containing an increasing number
of mismatches
Primer-template substrates used are defined in Table I. Rates and
amplitudes were obtained by fits to a double exponential defining the
kinetics of excision of the first base from the 3'-end of the primer.
NA, not applicable.
|
|
When the reaction was initiated by mixing enzyme with DNA, the rates
and amplitudes of the fast excision phase were lower (Fig.
2B, Table II), suggesting that DNA binding may have been rate-limiting. The fraction of DNA undergoing fast cleavage was only
23-30% of the total DNA, compared with 30-70% seen after preincubation of the enzyme with the DNA. Since the experiments were
performed with enzyme in excess over the DNA, the two reaction phases
may represent the binding of the enzyme to the DNA in two modes. For
example, the relative amplitudes of the fast and slow phases may
reflect the binding of DNA at the exonuclease and polymerase sites,
respectively (see "Discussion").
Role of the Accessory Subunit in Exonuclease Cleavage--
The
catalytic subunit contains both polymerase and exonuclease active
sites. Excision of DNA containing four mismatches (25 × 4/45 DNA)
by the catalytic subunit alone or by holoenzyme were compared to
determine what role the accessory subunit might play in error
correction. The rates of the fast phase of hydrolysis were 9.3 ± 0.7 and 15.9 ± 2.9 s1 for the holoenzyme and
catalytic subunit, respectively (Fig. 3A). The rates of cleavage
during the slower phase were 0.25 ± 0.03 and 0.10 ± 0.03 s1 for the catalytic subunit and holoenzyme,
respectively. Thus, in both the fast and slow phases, the rate of
excision was approximately 2-fold slower in the presence of the
accessory protein. Following excision to remove the 4-base mismatch
region, both enzyme forms proceeded to degrade properly base-paired DNA
(Fig. 3, C and D). However, the continued
degradation was much more prominent in the absence of the accessory
protein, suggesting a possible role for the accessory subunit in
recognition of properly base-paired DNA. The catalytic subunit
converted 31% of the DNA to products smaller than 21/45 DNA, while the
increased ability of the holoenzyme to recognize correctly base-paired
DNA led to only 14% of the DNA continuing to smaller products.
Additionally, the rates of formation of these smaller products were
0.26 ± 0.06 and 0.15 ± 0.06 s1 for the
catalytic subunit and holoenzyme, respectively. Thus, the accessory
subunit helps to discriminate against hydrolysis of properly
base-paired DNA.
View larger version (91K):
[in this window]
[in a new window]
|
Fig. 3.
Role of the accessory subunit in exonuclease
activity. Holoenzyme () or catalytic subunit
() (100 nM) were preincubated with 75 nM 25x4/45 DNA, and Mg2+ was added to initiate
the cleavage reaction. A, the percentage of 25-mer remaining
was plotted against time. Data were fit to a double exponential to
yield rates of 15.9 ± 3.0 s1 for cleavage of 25%
of the DNA by the catalytic subunit and 9.3 ± 0.7 s1 for cleavage of 42% of the DNA by the holoenzyme. The
slow phases proceeded at rates of 0.22 ± 0.02 s1
(51%) and 0.12 ± 0.02 s1 (31%) for the catalytic
subunit and holoenzyme, respectively. B, excision products
smaller than 21-mer, resulting from removal of the four mismatches
followed by continued excision of correct base pairs, were plotted
against time. Data were fit to a double exponential to yield maximal
amplitudes of 14 and 31% of the DNA for the holoenzyme and catalytic
subunit, respectively. Gels show the products of the cleavage reaction
of holoenzyme (C) and catalytic subunit (D). Note
the accumulation of 21-mer during excision by the holoenzyme in
contrast to the continued excision to create smaller products during
excision by the catalytic subunit alone.
|
|
Exonuclease Cleavage of 26xN/45 DNA Containing Different
3'-Terminal Nucleotides--
Exonuclease hydrolysis of DNA substrates
containing a T:T, C:T, or G:T mismatch was examined to determine the
effect of each base pair on the excision rate in a single turnover with
enzyme in excess (Fig. 4). The decreases
in concentrations of 26/45 DNA were fit to single exponential curves.
Excision of 26xN/45 DNA containing the T:T, C:T, and G:T mismatches
occurred at rates of 0.40 ± 0.04, 0.31 ± 0.01, and
0.57 ± 0.03 s1, respectively.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 4.
Base specificity of exonuclease
activity. Polymerase (100 nM) was preincubated with 75 nM 26/45 DNA containing a primer 3' dTMP (), dGMP
(), or dCMP () opposite a template dTMP. Mg2+ was
added to initiate the cleavage reaction. The percentage of 26-mer
primer remaining was plotted against time, and the data were fit to a
single exponential. Excision rates were 0.40 ± 0.04, 0.31 ± 0.01, and 0.57 ± 0.03 s1 for primer terminated with
dTMP, dCMP, and dGMP, respectively. For comparison, the
dashed line illustrates excision of correctly
base-paired DNA containing a 3' A:T base pair, as shown in Fig.
2A.
|
|
Intramolecular Correction of a T:T Mismatch by pol --
Models
for proofreading often suggest that the DNA can traverse from the
polymerase site to the exonuclease site and back again to remove and
correct an error without requiring the DNA to dissociate from the
enzyme. However, for pol the DNA dissociation rate of 0.02 s1 is significant relative to the rates of the
exonuclease reaction. To determine whether DNA strand transfer between
the polymerase and exonuclease active sites occurs without release of
the DNA, we examined the excision/extension reactions in the presence
and absence of a DNA trap consisting of excess unlabeled DNA. We
examined the reaction of 26xT/45(TGA) DNA to first remove the T:T
mismatch and then extend the primer to form a 27-mer in the presence of dATP and dCTP. The 45(TGA) template allowed the T:T mismatch to be
replaced by an A:T base pair and then to extend that DNA by the
incorporation of dCMP to distinguish the product from the starting
material. Formation of 27/45 product DNA occurred at rates of 0.41 ± 0.05 and 0.68 ± 0.06 s1 for polymerase in the
absence and presence of a DNA trap, respectively (Fig.
5). Additionally, the maximum amplitude
of the reaction decreased from 39 to 31 nM upon the
addition of the DNA trap. In the absence of the trap, 53% of the DNA
underwent both the excision and extension reactions, while this number
decreased to 41% in the presence of a trap. Thus, nearly 80% of the
reactive E-DNA complexes proceeded to completion without dissociation. Similar results were observed for 26xT/45 DNA, where the template allowed two dCMP incorporations to create 28-mer product following excision of the mismatch.
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 5.
Error correction in the presence of a DNA
trap. Polymerase (100 nM) was preincubated with 75 nM 26/45(TGA) DNA (see Table I). Mg2+, 25 µM dATP, and 25 µM dCTP were added to
initiate the reaction. Excision of the mismatch followed by extension
of the DNA to create correctly paired 27/45-mer was examined. Fit of
the data to a single exponential yielded rates of 0.41 ± 0.05 and 0.68 ± 0.06 s1 in the absence () and
presence () of a DNA trap, respectively. The amplitude
decreased from 39 nM (52%) to 31 nM (41%) by
the addition of the DNA trap.
|
|
Error Correction by pol --
As shown in the previous paper
(23), polymerization of a correct base pair over a T:T mismatch by the
exo mutant occurred at a rate of 0.5 s1 but
with a Kd of 404 µM for ground state
nucleotide binding. Accordingly, one would expect a rate of correct
incorporation over the T:T mismatch of ~0.03 s1 at 25 µM nucleotide. To verify that error correction
was occurring with wild-type enzyme as opposed to polymerization over
the mismatch in the experiment described above (Fig. 5), an excision
reaction was performed using 26xT/45 DNA in the presence of
[-32P]dATP to monitor formation of a correct A:T base
pair following mismatch hydrolysis. Excision of 26xT/45 DNA (unlabeled)
in the presence of [-32P]dATP resulted in the
formation of radioactively labeled 26-mer at a rate of 0.77 ± 0.27 s1, with an amplitude representing 40% of the DNA
(data not shown), which is equal to the amplitude of the reaction in
the presence of a trap (Fig. 5). Since the rate of formation of an A:G
mismatch over a T:T mismatch is negligible, this experiment showed that the observed reaction did indeed involve removal of the T:T mismatch followed by formation of a proper A:T base pair.
Error Correction of a Buried Mismatch--
Even with error
correction, polymerization will proceed to bury the mismatch by a
correct base pair a small fraction of the time. However, if forward
polymerization is still inhibited by the now buried mismatch, the
exonuclease is still able to correct the mistake. We examined the
ability of the wild-type polymerase to remove and subsequently correct
a buried mismatch by providing the correct nucleotide to extend the
27xTC/45 DNA substrate following excision of both the correct C:G base
pair and the buried T:T mismatch (see Table I). Three experiments were
done to compare 1) excision, 2) excision and correction of the
mismatch, and 3) excision followed by forward polymerization. Two forms
of the DNA were used, where the label was present either on the 5'- or 3'-end of the 27-mer primer.
First, the rate of excision of DNA containing a buried mismatch was
determined. We examined excision of 5'-32P-labeled 27xTC/45
DNA containing a T:T mismatch followed by a 3'-terminal C:G base pair
(Fig. 6, A-C). Loss of
full-length substrate occurred at a rate of 3.0 ± 0.3 s1 for 55% of the DNA, followed by a slower phase at a
rate of 0.08 ± 0.01 s1 for 42% of the DNA. The
major products of the reaction were 24- and 25-mer DNA, as illustrated
by the accumulation of these bands in Fig. 6B. An identical
reaction utilizing 27xTC/45 DNA containing a
3'-[-32P]dCMP directly showed release of the
3'-terminal nucleotide at a rate of 2.7 ± 0.2 s1
(data not shown). A slower phase was not observed. Release of [-32P]dTMP (the buried mismatch base) from the
27xTC/45 DNA occurred at a rate of 0.4 ± 0.1 s1, as
was observed in Fig. 4 for excision of a single mismatch (data not
shown). These results show that excision of the correct base pair after
a buried mismatch occurs at a rate of 3 s1. This rate is
faster than the rate of 0.4 s1 observed for removal of
the original mismatch (Fig. 4).
View larger version (55K):
[in this window]
[in a new window]
|
Fig. 6.
Proofreading and polymerization of a buried
mismatch. Polymerase (100 nM) was preincubated with 75 nM 27xTC/45 DNA (see Table I). Mg2+ was added
to initiate the exonuclease reaction (A-C), and
Mg2+ and 10 µM dATP were added to initiate
the proofreading and polymerization reaction (D-F).
A, product DNA was calculated as the ratio of 27-mer
relative to the sum of all of the DNA. Data were fit to a double
exponential to yield excision rates of 3.0 ± 0.3 s1
for 55% of the DNA and 0.079 ± 0.010 s1 for 43%
of the DNA. B, the gel shown illustrates the reaction
quantified in A. Lanes correspond to time points
in the graph. C, diagram of reaction shows excision of
27-mer primer to remove the buried mismatch. D, nucleotide
was added to a parallel reaction to examine excision followed by
polymerization, resulting in formation of 26-mer. The concentration of
DNA that was extended to 26-mer following exonuclease proofreading
() was plotted against time. Data were fit to a double
exponential to yield a rate of 3.4 ± 0.4 s1 for
40% of the DNA. The slow phase proceeded at a rate of 0.04 ± 0.01 s1 for 30% of the DNA. Products smaller than 26-mer
are created by continued excision of correctly base-paired DNA. In this
example (), pol continues excision of 17% of the
correctly paired DNA, despite the presence of the correct nucleotide
required for polymerization. E, the gel illustrates the
reaction quantified in D. Lanes correspond to
time points in the graph. F, diagram of reaction shows
excision followed by correct A:T base pair formation to correct the
error at the 26th primer position.
|
|
Second, the same cleavage reaction was performed but in the presence of
dATP added with the Mg2+ solution to initiate the reaction.
Following removal of the 3'-terminal C and the buried T mismatch, error
correction can occur by insertion of a correct A at the 26th primer
position. Following removal of the two 3'-terminal primer bases, the
correct dATP was inserted to form 26-mer product as shown in Fig. 6,
E and F. If excision results in formation of a
24-mer, this DNA can also be extended by two dATP incorporation events
to form a 26-mer. The conversion of 27/45 DNA to 26/45 DNA was
monitored over time to yield a rate of 3.2 ± 0.4 s1
for 50% of the DNA, followed by a slower phase at a rate of 0.04 ± 0.01 s1 for 30% of the DNA (Fig. 6D).
Correct incorporation of dATP was previously shown to occur at a rate
of 45 s1 (10), indicating that the error correction
reaction was limited by the excision rate.
The third experiment examined polymerization beyond the length of the
original 27/45 DNA substrate, as would occur upon error detection
during processive polymerization. Wild-type polymerase was preincubated
with 5'-32P-labeled 27xTC/45 DNA, and a solution of
Mg2+, dATP, and dCTP was added to initiate the reaction.
Formation of 28/45 DNA occurred at a rate of 39.4 ± 4.1 s1 rather than the rate of ~3 s1 observed
in the two previous reactions (Fig. 7).
The amplitude of the reaction represents ~50% of the original DNA
concentration, and a similar result was obtained when a DNA trap was
added to the reaction. This surprising result suggests that the
presence of the second correct nucleotide somehow stimulates the rate
of proofreading.
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 7.
Effect of next correct nucleotide on
exonuclease rate. Polymerase (100 nM) was preincubated
with 75 nM 27xTC/45 DNA. Mg2+, 25 µM dATP, and 25 µM dCTP were added to
initiate the exonuclease/polymerase reaction. Excision of the 3' dCMP
and dTMP allow the polymerase to form a correct A:T base pair followed
by two C:G base pairs. Accumulation of 28/45 DNA () was
monitored over time. Fit of the data to a single exponential yields a
rate of 39.4 ± 4.1 s1 and a maximal amplitude of 40 nM (53%). The mismatched primer base,
[-32P]dTMP (), was released at a rate of
0.8 ± 0.1 s1. Parallel reactions performed with the
addition of 1) Mg2+, dATP, and dCMP or 2) Mg2+,
dATP, and dTTP resulted in no observed cleavage (data not shown).
|
|
We performed two experiments to attempt to distinguish whether the fast
rate of 39 s1 was due to an accelerated excision of the
buried mismatch or due to faster than expected polymerization over a
buried mismatch. First, an identical experiment was performed using the
exonuclease-deficient mutant, and we observed extension over the buried
mismatch at a rate of 2 s1 in the absence of proofreading
(see accompanying paper (23)). Second, we used the same 27xTC/45 DNA
with an [-32P]dTMP label at the position of the buried
mismatch, to monitor directly the excision of the buried T:T mismatch.
The rate of removal of labeled T in the presence of dCTP and dATP was
0.8 ± 0.1 s1, similar to the rate observed for
excision of a single mismatch. We are thus faced with a dilemma. Either
the exonuclease-deficient mutant catalyzes much slower rates of
polymerization over a buried mismatch (2 versus 39 s1), or the observed excision of radiolabeled T in the
buried mismatch reflects a slower or more complex process involving
multiple reactions.
It was unclear from the experiment shown in Fig. 7 how the presence of
a nucleotide could enhance the rate of proofreading. It appeared that
the excision reaction in the absence of nucleotides proceeded at a rate of ~3 s1, as observed for both
cleavage of 5'- and 3'-labeled 27-mer primer. Since proofreading in the
presence of dATP was limited by the rate of excision, the role of dCTP
in these reactions was explored further.
Several variations to the experiment shown in Fig. 7 were performed, to
determine the importance of the identity of nucleotides present in a proofreading reaction to enhance the 3'-base excision rate. First, identical reactions performed with dATP/dTTP or dATP/dCMP combinations yielded little or no cleavage or extension products, indicating that the identity of the added nucleotide (dCTP in this
case) is a requirement for the fast formation of 28/45 DNA (data not shown).
Second, alternative 27xTT/45 DNA sequences were created to examine the
effects of changing the 3'-base pair or the next template base on the
proofreading reaction. The DNAs used are listed in Table I as
27xTT/45(TAG) and 27xTT/45(TAA). In each case, the 27th primer base
pair was changed from C:G to T:A, and inclusion of dTTP in the
proofreading reaction was required for rate enhancement in the "error
correction" reaction. Note that in these experiments, an increase in
rate was observed when the reaction was performed in the presence of
dTTP (same as the 3-' base of 27xTTmer), but the template base
opposite the 28th primer position was changed from a G to an A. This
demonstrated that the nucleotide at the position immediately adjacent
to the buried mismatch is not essential to achieve the increase in the
observed polymerization-proofreading rate. Using the alternate DNA
template sequences, excision of the buried mismatch occurred at rates
of ~3 s1, and the rate of 28-mer formation in the
presence of the nucleotides required for proper base pairing was
enhanced to 10-20 s1 (data not shown).
We also examined the error correction proofreading reactions
using wild-type polymerase and utilizing
5'-[-32P]dATP- or
3'-[-32P]dCMP-labeled 27-mer primers in parallel
experiments in the presence of Mg2+ and dCTP. Formation of
28/45 DNA with 5'-labeled DNA would allow the observation of 28/45 DNA
product that is created through extension over the 27/45 DNA as well as
any product formed through various excision and polymerization
reactions that may occur, but by following the label in each position
we could ascertain the extent to which the dCMP was removed prior to
polymerization. This resulted in the formation of 28/45 DNA at a rate
of 18.6 ± 1.7 s1, to an amplitude of 39% of the
DNA (Fig. 8, left
gel); and over the same time course, the majority of the
labeled dCMP was removed. Little formation of 28/45-mer was observed
when monitoring excision of 27/45-mer DNA containing a
3'-[-32P]dCMP in the presence of just dCTP. (Fig. 8,
right gel). Loss of the 3'-dCMP from the 27/45
DNA occurred at a rate of 21.5 ± 5.7 s1 for 61% of
the DNA. This experiment provided direct evidence that the presence of
dCTP afforded an increased rate of excision of the 3'-terminal base.
Furthermore, maximally 8% of the DNA transiently proceeded from
27/45 DNA to 28/45 DNA followed by subsequent loss of the
radioactive label during proofreading.
View larger version (29K):
[in this window]
[in a new window]
|
Fig. 8.
Role of dCTP in error correction.
Polymerase (100 nM) was incubated with either
5'-32P-labeled () or
3'-[-32P]dCMP-labeled () 27xTC/45 DNA. The
reaction was initiated by the addition of Mg2+ and 50 µM dCTP. Formation of 5'-32P-labeled 28/45
DNA occurred at a rate of 18.6 ± 1.7 s1 to a
maximal amplitude of 29 nM, or 39% of the DNA. Excision of
3'-[-32P]dCMP-labeled 27/45 DNA occurred at a rate of
21.5 ± 5.7 s1 for 62% of the DNA. Extension of
3'-[-32P]dCMP-labeled 27/45 DNA increased to a maximum
of 5.6 nM DNA (8%) and then decreased over time.
|
|
In a separate experiment, we examined the extension to full-length
product in the presence of all four nucleotides, starting with 27xTC/45
DNA labeled with either [-32P]dTMP or
[-32P]dCMP at the two 3'-terminal bases in the primer.
Only 9 or 13% of the labeled dCMP or dTMP, respectively, were
recovered in the full-length product (data not shown). The small
fraction of 28/45 DNA or 45/45 DNA accumulating with label in the
buried mismatch demonstrates that pol recognizes a buried mismatch
and will correct rather than polymerize over the buried mismatch
90% of the time.
| |
DISCUSSION |
We have previously examined the contributions of the polymerase
reaction to the overall pol fidelity by defining the parameters governing formation of an incorrect base pair and the consequences of a
mismatch during subsequent polymerization steps. Here we have used
single-turnover conditions to characterize the exonuclease activity of
pol in an effort to identify the contributions of the exonuclease
to overall polymerase fidelity. By definition, an efficient exonuclease
would be able to quickly and specifically remove errors without
unnecessarily cleaving correctly base-paired DNA. Exonuclease cleavage
of correctly base-paired and mismatched DNA was analyzed to define the
role of mismatch selectivity during proofreading by pol .
Excision of Single-stranded DNA--
The rate of hydrolysis of
single-stranded DNA by proofreading DNA polymerases is generally
fast. For example, excision rates greater than 700, 280, and 100 s1 were observed for T7 DNA polymerase, Escherichia
coli polymerase III, and T4 DNA polymerase, respectively (4, 11,
12). For these reactions, hydrolysis of single-stranded DNA defines the minimum excision rate for single-stranded DNA bound to the exonuclease site and, thereby, the maximum rate one would expect with duplex DNA.
Accordingly, the slower excision rate seen with duplex DNA is thought
to be due to rate-limiting strand melting and/or transfer to the
exonuclease site (4, 11, 12) and preferential binding of the DNA to the
polymerase active site (13). In contrast, single-stranded DNA appears
not to be a preferred substrate for excision by human pol , since
hydrolysis is remarkably slow. As an alternative, we have attempted to
estimate the minimum excision rate using duplex DNA containing multiple
3'-terminal mismatches and have observed a maximum rate of 9 s1.
The pol family of DNA polymerases includes the pol I, bacteriophage
T7, and T4 DNA polymerases. Members of this family have a catalytic
lysine residing in the exonuclease active site that forms a salt bridge
with a catalytic aspartate (14). This interaction increases the
exonuclease activity of the T4 DNA polymerase 3'-5' exonuclease; and a
lack of a similar salt bridge in Klenow fragment correlates with the
1000-fold lower exonuclease activity for that polymerase. Protein
sequence alignment with these polymerases shows that pol may also
lack this residue, perhaps contributing to the low potency of this exonuclease.
The report of a 7.5-fold preference of porcine pol for
single-stranded homopolymeric DNA over double-stranded DNA (15) probably reflects the difference in excision of single-stranded (e.g. 0.16 s1 for dGMP-terminated 25-mer) and
correctly base-paired double-stranded DNA (0.02 s1)
observed here. In another report, human pol catalytic subunit was
observed to have a similar preference for single- and double-stranded DNA (9), but the differences in rates we observed may not have been
apparent in the visual gel analysis used to support those conclusions.
Human pol catalyzes excision of single-stranded DNA at a rate that
is only slightly faster than excision of correctly base-paired duplex
DNA. Structural studies on other DNA polymerases have shown that
single-stranded DNA binds into the exonuclease active site, and in fact
double-stranded DNA is excluded due to spatial constraints (16). It is
possible that single-stranded DNA binds more weakly or at a different
site on pol , limiting the fraction that binds to the exonuclease
site. Alternatively, pol may require duplex DNA to bind to the
polymerase and to be frayed to extend the 3'-end of the primer to the
exonuclease site, thereby occupying both double-stranded DNA and
single-stranded DNA sites simultaneously to achieve specificity and
affinity. The slow rate of excision of single-stranded DNA may also
result from a lack of structural rigidity of the polymerase, since it
is possible that efficient proofreading by pol requires
double-stranded DNA for positioning of the primer 3'-end into the
exonuclease active site. This hypothesis may be supported by
observation of enhancement of the excision rate by the presence of a
nucleotide that can interrupt the terminal base pair of a duplex and
favor partitioning of the 3'-end into the exonuclease site (see below).
In the absence of structural data for pol , it is difficult to
determine the cause of the slow rate of excision of single-stranded DNA.
Cost of Proofreading--
A measure of the efficiency of an
exonuclease is the ability to discriminate between correctly and
incorrectly base paired DNA. The "cost" of having a proofreading
exonuclease is defined by the frequency of loss of correct base pairs.
Pol removes the 3'-base of correctly base paired 25/45 DNA at a
rate of 0.05 s1. The polymerase incorporates correct
nucleotides at an average rate of 37 s1. Thus, the
average "cost of proofreading" as defined by the probability of
forward polymerization versus removal of a correctly
base-paired nucleotide is 0.14%. Excision of correctly base-paired DNA
does not significantly affect processive polymerization. A
similar cost of proofreading was observed for T7 DNA
polymerase (4).
Selectivity of Exonuclease for Mismatched DNA
Substrates--
Brutlag and Kornberg (17) first demonstrated the
preference of the 3'-5' exonuclease for terminally mismatched DNA
substrates. The selectivity of pol for mismatched DNA was examined
by comparing excision of correctly and incorrectly base-paired DNA.
Indeed, the pol exonuclease is selective for mismatched DNA
substrates, as evident from the increased amplitudes and rates of
excision observed with increasing numbers of mismatches in the DNA
(Fig. 2). The rate of excision increases 20-fold when one mismatch is introduced onto a primer 3'-end and increases greater than 170-fold with four or seven mismatches present. It is reasonable that the ability of the DNA to transfer to or bind into the exonuclease active
site increases with the length of incorrectly base-paired DNA, due to
the expanded duplex melting. However, this increase in excision rate is
not sufficient to favor exonuclease removal of mismatches by pol or
any other known polymerase. Rather, stalling of the polymerase after a
mismatch alters the kinetic partitioning from favoring forward
polymerization to favoring exonuclease removal. The markedly slower
polymerization rate combined with a slightly faster excision rate,
provides for selective excision of mismatches.
Effect of Sequence on the Excision Rate--
Excision rates vary
in the presence of alternate DNA sequences. Excision of 26xN/45 DNA
containing a single 3'-terminal mismatch occurred at an average rate of
0.4 s1 (Fig. 4). Only slight differences in rates of
excision were observed for 26/45 DNA containing dTMP, dGMP, or dCMP as
the 3'-base of the primer, indicating that the pol exonuclease has
little selectivity for the identity of a mismatch at this position. The
rate of excision by pol does not correlate with stability of the
adjacent base pair, as might be expected (18). For example, in studies
analyzing DNA sequence effects with T4 DNA polymerase, the presence of
a C:G base pair prior to the 3'-terminal correct base pair contributed to 4-fold slower kinetics of excision (19). In our DNA sequences, a C:G
and an A:T base pair precede the T:T mismatches in the 25x1/45 and
26/45 DNA, respectively. The rate of the cleavage reaction was slightly
faster when the T:T mismatch resided at the 25th primer position
(25×1/45) rather than the 26th. But the possible influence of adjacent
base pair stability may be observed in the larger amplitude of excision
of 26xN/45 DNA (80%) compared with 25/45 DNA (31%), where ease of
melting may contribute to an increased fraction of DNA transfer to the
exonuclease active site. Finally, single exponential kinetics were
observed for excision of correctly base-paired DNA as well as 26xN/45
DNA. These reactions provided the slowest excision rates that we
observed and may have occluded observation of two distinct phases. It
is likely that the reactions were dominated by the slow rate of
transfer of the DNA from the polymerase to the exonuclease active site,
since identical kinetics were observed for excision of 26xN/45
regardless of whether the DNA was preincubated with the enzyme.
Is Strand Transfer Intra- or Intermolecular?--
Several previous
studies examining pol exonuclease activity in the presence of a DNA
trap have suggested that DNA strand transfer from the polymerase to the
exonuclease site occurs by release and rebinding. In previously
published studies, human and Drosophila pol were
observed to dissociate prior to excision of DNA, since no hydrolysis of
a mismatched DNA (Drosophila) and correctly paired DNA
(human) were observed in the presence of a DNA trap (3, 9). We observed
that the steady state rate of excision of correctly base-paired DNA
(0.01 s1; data not shown) was slightly slower than the
dissociation rate (0.02 s1; previous article (23)), and
even in a single turnover experiment, the rate of excision of this DNA
(0.05 s1; Fig. 2) was only slightly faster than the
dissociation rate. Therefore, it is unlikely that excision of correctly
base-paired DNA would be observed in the presence of a trap in a steady
state experiment. The observation that Drosophila pol may dissociate from mismatched DNA prior to any excision was surprising.
To examine the extent of dissociation during proofreading by human pol
, several of the cleavage reactions were repeated, but with
the addition of 1 µM unlabeled 25/45 DNA as a trap. The rates and amplitudes of the reactions were equal to those previously observed for the fast phase of excision. For example, ~40% of the
25x4/45 DNA underwent the fast excision by the holoenzyme as shown in
Figs. 2A and 3A. When a DNA trap was added to the reaction, the initial amplitude and rate were unchanged, but in the
presence of a trap, a second, slower phase was not discernible (data
not shown). Similar amplitudes were observed for excision by T4 DNA
polymerase, due to the ratio of the rate of dissociation to the rate of
excision (12). The crystal structure of RB69 DNA polymerase, which is
closely related to T4 DNA polymerase, shows that the polymerase is in
an "open" conformation when the DNA is bound to the exonuclease
active site (16), perhaps leading to the ~50% dissociation of the
DNA during excision reactions (12).
Error correction requires two strand transfers involving movement of
the DNA from the polymerase site to the exonuclease site and then back
to the polymerase site. One might expect a higher rate of dissociation
of the DNA from the enzyme while it is loosely bound in transit. In
direct measurement in a single turnover experiment, we observed a
10-20% decrease in the amplitude of the reaction when analyzing
excision followed by incorporation in the presence of a trap. Thus, it
is important to note that human pol can excise and correct an error
in one DNA binding event without dissociation. In these reactions,
generally 80% of the productive E-DNA complexes (defined by the amplitude in the absence of the trap) proceeded through
the complete intramolecular reaction sequence involving transfer at
least from the exonuclease to the polymerase without dissociation.
Role of the Accessory Subunit in Excision--
The accessory
subunit allows the polymerase to more tightly control excision through
increased selectivity. Increased recognition of correctly base-paired
substrate by the holoenzyme was visually evident in Fig. 3C
compared with Fig. 3D (and quantified in Fig. 3B), where excision of DNA containing four mismatches
resulted in increased 21/45 DNA accumulation by the holoenzyme. This
accumulation was not observed during cleavage by the catalytic subunit
alone. Additionally, cleavage of mismatched DNA by the catalytic
subunit occurred at a faster rate than was observed for the holoenzyme. The slower excision rate by the holoenzyme may reflect a more stable
polymerase conformation that is better able to recognize correct DNA
structure or bind duplex DNA at the polymerase site more tightly. This
may contrast with a previous suggestion that the accessory subunit
enhances pol exonuclease activity (20), but the experiment
performed to provide this conclusion was not described.
Kinetic Partitioning during Correction of a Single
Mismatch--
Formation of DNA containing one mismatch, such as
26xN/45 DNA, is likely to occur. The exonuclease contribution to
fidelity can be calculated by determining the probability of excision
of a single mismatch from the ratio of the rate of excision (0.4 s1) relative to the rate of polymerization over the
mismatch (4). The rate of polymerization over the mismatch is
nucleotide concentration-dependent, with
kcat/Km = 0.5 s1/400 µM for pol . At an intracellular
nucleotide concentration of 100 µM, the expected rate
will be 0.12 s1. Accordingly, the probability of
exonuclease correction of this mismatch is as follows.
|
(Eq. 1)
|
This calculation indicates that 80% of single mismatches are
likely to be excised without dissociation of the enzyme from the DNA.
Since the closed system of a test tube or a mitochondria allows for the
rebinding of DNA that dissociates from the polymerase, we have not
expressly included dissociation in the probability of excision.
Error Correction--
The fate of the 80% of mismatched DNA that
can proceed through polymerization to become a buried mismatch (27/45
DNA) was examined more fully in experiments reported here, as
summarized in Table III. The most
surprising result from this study was the faster rate of 3'-base
excision during proofreading of a buried mismatch occurred in the
presence of the nucleotides required to extend a primer past the site
of the original mismatch. Upon the addition of one nucleotide to the
reaction to allow correction of the mismatch (Fig. 6D), no
change in the excision rate was observed, indicating that the
exonuclease reaction was rate-limiting. Then, upon examination of
polymerization beyond correction of the mismatch, the rate of the
reaction increased 13-fold, from 3 to 39 s1 (Fig. 7).
The observed 39 s1 proofreading rate may simply represent
the rate of nucleotide incorporation to form 28/45 DNA containing the
buried mismatch, or it may represent the excision and polymerization to
correct the error. Using substrate labeled with
-32P-labeled dNMP, we have shown by lack of accumulation
of 28/45 DNA that direct incorporation over the buried mismatch
at this rate did not occur. Therefore, the presence of nucleotide must be increasing the rate of hydrolysis of the 27/45 DNA. Indeed, an
increase in the rate of release of the 3'-[-32P]dCMP
label from 3 to 21 s1 was observed when the reaction was
performed with the addition of Mg2+ and dCTP. It appears
that this mechanism does not affect excision of the mismatch once the
correctly paired 3'-dCMP is removed. While no change in the rate of
hydrolysis of the mismatched base was observed in the presence of
nucleotides, accelerated removal of the 3'-base may allow the mismatch
to be exposed for excision, thus allowing correction of the mismatch.
Exonuclease activity is not processive; and removal of subsequent
nucleotides requires a trip to the exonuclease active site for each
hydrolysis reaction. Faster release of the 3'-nucleotide (dCMP) may act
to increase the chance of excision of the buried mismatch (dTMP), due
to exposure of that base in the polymerase active site.
Based on the evidence presented here, it seems reasonable to speculate
that the presence of dCTP increased the rate of DNA transfer for pol
, although the mechanism through which this enhancement is
accomplished is unclear. We propose an exonuclease mechanism where
excision of 27/45 DNA occurs by two pathways (Scheme 2). The rate and amplitude of excision
accomplished through the first pathway are governed by the fraction of
27/45 DNA that is able to extend toward the exonuclease active site.
This fraction undergoes cleavage of the base at a rate of 3 s1, as shown in Fig. 6. The rate of the reaction is
limited by the rate of transfer of the DNA to the exonuclease active
site, where the 3'-end of the DNA is not able to efficiently extend
into the exonuclease active site. The second pathway explains how this transfer can be enhanced. In the second pathway, the addition of
nucleotides (dCTP or both dATP and dCTP together) causes an increase in
the rate of excision to between 21 and 39 s1. We propose
that this may occur through nucleotide interruption of the 3'-base
pair, which may widen the gap between the duplex and push the 3'-end of
the DNA closer to the exonuclease active site. Since the rate of 39 s1 for formation of the 28/45 DNA was only observed in
the presence of both dATP and dCTP, possibly both of the nucleotides
that allow error correction and extension are required for full
reaction enhancement. The presence of both dNTPs would allow a larger
wedge to aid in duplex melting and fast excision of the 3'-terminal nucleotide.
View larger version (14K):
[in this window]
[in a new window]
|
Scheme 2.
Proposed pathways of correction of a buried
mismatch. The bend in the primer 3'-end indicates transfer of the
DNA into the exonuclease active site. Excision of DNA through pathway 1 occurs at a rate of ~3 s1. Excision of DNA through
pathway 2 occurs at a rate of 39 s1 due to the presence
of dATP and dCTP. These nucleotides can base-pair with the template and
facilitate an increased rate of excision.
|
|
Several lines of evidence support the proposed mechanism. First, only
the presence of dCTP was required for an enhanced rate of 3' mismatch
excision, and this is the nucleotide that would be able to interrupt
the 3'-base pair. Second, rate enhancement was observed in the presence
of dATP and dTTP, with modified sequences to place a correct A:T base
pair following the T:T mismatch. Third, if dTTP (wrong nucleotide) or
dCMP were included in the reaction, little or no cleavage was observed.
Both dTTP and dCMP lack the proper nucleotide structure (base hydrogen
bonds or triphosphate contacts) required for efficient binding of an
incoming nucleotide opposite the template dGMP. Fourth, if the
partitioning of the rates of excision and forward polymerization for
26xT/45 and 27xTC/45 DNA are taken into account, 4% of the DNA could
proceed from 26xT/45 DNA to 28/45 DNA, without correction of the T:T
mismatch ((0.1/1.1) × (2/5) = 0.04). Experimentally,
incubation of 27/45 DNA containing the 3'-labeled dCMP, with all four
dNTPs, results in extension of only 9% of the DNA to full-length 45/45
DNA (data not shown). Combined, it is evident that error correction
does occur and that a nucleotide identical to the primer 3'-base
specifically enhances the efficiency of this reaction.
Contribution of Exonuclease to pol Fidelity--
Processive
DNA synthesis catalyzed by pol proceeds at an average rate of 37 s1, with an occasional dissociation (0.02 s1) or transfer to the exonuclease active site (0.05 s1). Overall, pol selects the correct nucleotide and
catalyzes efficient incorporation compared with dissociation and
excision of correctly base-paired DNA. We have calculated a cost of
proofreading of 0.1%, meaning that polymerization continues for 99.9%
of correct base pairs.
The exonuclease contribution to fidelity in correcting a single
mismatch is the probability of excision versus
extension = 4. This value is low compared with the estimated
100-fold enhancement of fidelity by chick embryo pol (5) and the
200-fold contribution of the exonuclease defined for T7 DNA polymerase
(4). However, this calculation includes the rate of polymerization over
a mismatch (kpol,over), which allows the
formation of a buried mismatch. We have shown that excision of the
3'-base of a buried mismatch occurred at a fast rate (21-39
s1) in the presence of nucleotides. The increased
exposure of the mismatch to excision may act to increase the
exonuclease contribution to fidelity to ~200, based on the 21 s1 rate of excision of 27/45 DNA in the presence of dCTP.
Future examination of the effects of nucleotides upon pol proofreading may show that the exonuclease contribution is in fact higher.
Overall Fidelity of Mitochondrial DNA Replication--
Combining
the results of the previous paper (23) with the data presented here,
the net average fidelity of pol is ~1 error in 1.1 million base
pairs assuming a contribution of only 4× in the exonuclease step. This
number could be 18-fold higher if the presence of the correct
nucleotides improves fidelity as indicated by some experiments. Thus,
the overall fidelity is in the range of 1 error in 1-20 million. This
number is considerably lower than the error frequencies of 1 in
109 to 1010 bases expected for the replication
of nuclear DNA and accounts for the observed faster rates of mutation
in mitochondrial DNA. However, this simple analysis neglects the
effects of mismatch repair and of multiple copy numbers of the
mitochondrial genome. Mammalian cells have 1000-5000 copies of the
mitochondrial genome (21, 22). On the one hand, multiple copies
increase the probability of introducing an error with each generation,
but the redundancy allows fatal errors to be masked. The process of
selection from the pool of mitochondria within a cell is not well
understood and is probably an important factor affecting the rate of
accumulation of persistent mitochondrial DNA mutations. The current
data only provide the rate of formation of mutations during replication under the conditions we have chosen to mimic the interior of the mitochondria. Our results must be interpreted in the context of events
that follow mitochondrial replication within eukaryotic cells.
TOP
ABSTRACT
INTRODUCTION
To whom correspondence should be addressed: Institute for Cellular
and Molecular Biology, MBB 3.122, A4800, University of Texas, Austin,
TX 78712. Tel.: 512-471-0434; Fax: 512-471-0435; E-mail:
kajohnson@mail.utexas.edu.
