THE JOURNAL OF BIOLOGICAL CHEMISTRY
© 2002 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 277, No. 37, Issue of September 13, pp. 34198 –34207, 2002
Printed in U.S.A.
Fidelity of Escherichia coli DNA Polymerase IV
PREFERENTIAL GENERATION OF SMALL DELETION MUTATIONS BY dNTP-STABILIZED MISALIGNMENT*
Received for publication, May 16, 2002, and in revised form, July 1, 2002
Published, JBC Papers in Press, July 3, 2002, DOI 10.1074/jbc.M204826200
Sawami Kobayashi‡, Michael R. Valentine‡, Phuong Pham‡, Mike O’Donnell§,
and Myron F. Goodman‡¶
From the ‡Department of Biological Sciences and Chemistry, Hedco Molecular Biology Laboratories,
University of Southern California, University Park, Los Angeles, California 90089-1340 and the §Rockefeller
University and Howard Hughes Medical Institute, New York, New York 10021
More than 40 SOS genes, including three DNA polymerases,
are induced in Escherichia coli in response to DNA damage (1).
pol IV1 and pol V are “error-prone” Y family polymerases (2, 3),
and pol II is a high fidelity B family member (4, 5). Included
within the Y family is a subgroup sharing sequence homology
with E. coli dinB, the gene encoding pol IV. The dinB-like pols
have been identified in essentially all prokaryotic and eukaryotic organisms (2).
pol V is responsible for generating SOS-induced chromosomal mutations targeted at DNA damage sites (6, 7). pol II,
* This work was supported by National Institutes of Health Grants
GM42554 and GM21422. The costs of publication of this article were
defrayed in part by the payment of page charges. This article must
therefore be hereby marked “advertisement” in accordance with 18
U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed: Dept. of Biological
Sciences, University of Southern California, SHS Rm. 172, University
Park, Los Angeles, CA 90089-1340. Tel.: 213-740-5190; Fax: 213-7408631; E-mail: [email protected].
1
The abbreviations used are: pol, DNA polymerase; MBP, maltosebinding protein; 2AP, 2-aminopurine; nt, nucleotide; DTT, dithiothreitol; SSB, single-stranded DNA-binding protein; p/t DNA, primer-template DNA.
the only SOS pol having a 3⬘-exonuclease proofreading activity,
plays an instrumental role during “error-free” replication-restart (8, 9). pol II is induced almost immediately (⬃30 s) after
SOS induction (8), whereas the delayed appearance of pol V,
⬃30 – 45 min post-SOS induction (10), allows mutation-free
repair processes to occur prior to error-prone translesion
synthesis (8, 9).
In contrast to pol V, less is known about the cellular functions of pol IV (3). Although pol IV is not required for SOS
lesion-targeted UV mutagenesis, it does cause untargeted mutations on bacteriophage ␭, and its overexpression results in an
increase of untargeted frameshift and transversion mutations
on F⬘ and chromosomal DNA (11, 12). pol IV copies DNA lesions
of non-UV origin, for example, benzo(a)pyrene-guanine adducts
(13, 14) and 4-nitroquinoline N-oxide (15). pols II and V are
present in the cell at ⬃50 (16) and ⱕ15 (17) molecules/cell,
respectively, in the absence of SOS induction. In contrast,
constitutive levels of pol IV are ⬃250 molecules/cell (15).
The elevated level of expression in the absence of DNA damage implies a role for pol IV during normal DNA synthesis. For
example, pol IV might help in rescuing stalled replication forks.
Replication forks are thought to collapse in the absence of DNA
damage perhaps as often as once per round of replication (18)
and might result from mismatched or misaligned primer ends
refractory to proofreading by pol III (19). By interacting with
the ␤-processivity clamp (7, 20), pol IV might displace pol III
core to extend aberrant primer ends (21). Another role for pol
IV was revealed by genetic assays showing that it is primarily
responsible for increased mutability in starving non-dividing
cells (22, 23), a phenomenon referred to as “adaptive mutation.”
It has recently been shown that stationary phase E. coli lacking
in one or more of the SOS polymerases are less fit when grown
in the presence of wild type cells (24).
Given the paucity of information regarding the roles of pol IV
in vivo, it is important to investigate the biochemical properties
of this atypical polymerase. In this paper, we have investigated
the following: (i) the mutational spectrum for pol IV in the
absence or presence of ␤/␥ processivity complex; (ii) nucleotide
incorporation fidelity and mismatch extension efficiency of pol
IV on undamaged DNA; (iii) the biochemical mechanism responsible for the favored generation of ⫺1 frameshift mutations. To examine the mechanism governing pol IV-generated
⫺1 frameshifts, we have measured the change in fluorescence
intensity of a template 2-aminopurine while undergoing replication. Based on a crystal structure of Sulfolubus solfataricus
Dpo4, a Y family homolog of E. coli pol IV, it has been suggested
that ⫺1 frameshifts occur when an incoming dNTP bound at
the polymerase-active site is incorporated opposite a complementary downstream template base on a misaligned p/t DNA
34198
This paper is available on line at http://www.jbc.org
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
Escherichia coli DNA polymerase IV (pol IV), a member of the error-prone Y family, predominantly generates ⴚ1 frameshifts when copying DNA in vitro. T 3 G
transversions and T 3 C transitions are the most frequent base substitutions observed. The in vitro data
agree with mutational spectra obtained when pol IV is
overexpressed in vivo. Single base deletion and base
substitution rates measured in the lacZ␣ gene in vitro
are, on average, 2 ⴛ 10ⴚ4 and 5 ⴛ 10ⴚ5, respectively. The
range of misincorporation and mismatch extension efficiencies determined kinetically are 10ⴚ3 to 10ⴚ5. The
presence of ␤ sliding clamp and ␥-complex clamp loading proteins strongly enhance pol IV processivity but
have no discernible influence on fidelity. By analyzing
changes in fluorescence of a 2-aminopurine template
base undergoing replication in real time, we show that a
“dNTP-stabilized” misalignment mechanism is responsible for making ⴚ1 frameshift mutations on undamaged
DNA. In this mechanism, a dNTP substrate is paired
“correctly” opposite a downstream template base, on a
“looped out” template strand instead of mispairing opposite a next available template base. By using the same
mechanism, pol IV “skips” past an abasic template lesion
to generate a ⴚ1 frameshift. A crystal structure depicting dNTP-stabilized misalignment was reported recently for Sulfolubus solfataricus Dpo4, a Y family homolog of Escherichia coli pol IV.
E. coli pol IV Frameshift Mutation Mechanism
(25, 26), by a model which we have referred to as “dNTPstabilized misalignment” (19, 27).
EXPERIMENTAL PROCEDURES
(Vmax/Km)w/(Vmax/Km)r, where w and r refer to mispairs (wrong) and
correct pairs (right) for either incorporation or extension.
Processivity Measurement—A 32P-labeled 30-mer primer was annealed to ssM13mp2 DNA and extended by each of three versions of pol
IV, MBP-tagged, His-tagged, and native. Each reaction was performed
in the presence or absence of the processivity proteins ␤- and ␥⫺complex. Reactions were run at 37 °C in 1⫻ reaction buffer containing 10
nM p/t DNA, 200 ␮M each of dNTPs, and varying concentrations of pol
IV (0.62, 1.25, 2.5, 5, and 10 nM). SSB (1 ␮M), ␤- (40 nM), and ␥⫺complex
(10 nM), when present, were preincubated with p/t DNA and 500 ␮M
ATP for 3 min before adding dNTP and polymerase. Reactions were
quenched after 10 min at 37 °C with stop solution and resolved by 12%
PAGE. Average processivity is defined as the number of nucleotides
added divided by the number of primers extended.
Fluorescence Assay Measuring Extrahelical Template 2-Aminopurine—Fluorescence measurements were carried out on a QuantaMaster
(QM-1) Fluorometer (Photon Technology International) in which monochromators were used to select the wavelengths for excitation (310 nm)
and emission (370 nm) with a bandpass of 6 nm. Two 80-nucleotide
templates with 2-aminopurine (2AP) were designed as follows: 5⬘-TGAGCGTTTTTCCTGTTGCAATGX(2AP)AAAAAGTAATATTGTTCTGGATATTACCAGCAAGGCCGATAGTTTGAGTTCTTCT-3⬘ where X is
C or G. A 30-nt primer was annealed in the center of each template to
form p/t DNA. A cuvette containing reaction buffer (20 mM Tris-Cl, pH
7.5, 50 mM NaCl, 4 mM MgCl2) was placed in a fluorometer and scanned
as a function of time. An aliquot of p/t DNA (2 ␮l) was added to the
cuvette followed by pol IV (4 ␮l). A dNTP (10 ␮l) was then added as
either dTTP, dGTP, or dCTP. Emission was monitored for ⬃1 min prior
to the addition of each component. Final concentrations were 200 nM p/t
DNA, 1 ␮M pol IV, and 1 mM dNTP in a final volume of 200 ␮l.
Translesion Synthesis—120-nt templates were synthesized either
with or without an abasic site (tetrahydrofuran) located at a position 63
or 51 from the 5⬘-end of the template. p/t DNA substrates for lesion
bypass were constructed by annealing a 32P-labeled 30-mer primer
(complementary to positions 62–91) to the template. Translesion synthesis reactions (10 ␮l) were carried out at 37 °C in 1⫻ buffer containing
10 nM p/t DNA, 100 nM native pol IV, and 200 ␮M each of the four dNTP
substrates. SSB (250 nM), ␤- (200 nM), and ␥-complex (50 nM) were
added as indicated. When ␤- and ␥-complex were present, the reactions
were supplemented with 500 ␮M ATP. The reactions were quenched
after 5 min, and products were separated on 12% PAGE and visualized
by PhosphorImaging.
RESULTS
pol IV Mutational Spectrum—The types of errors catalyzed
by pol IV were measured using an in vitro LacZ␣ complementation assay that scores base substitution, addition, and deletion errors within a 407-nucleotide gapped region of M13mp2
DNA (29). Because it has become common practice to purify
proteins containing N- or C-terminal affinity tags, it is important to determine the extent to which specific protein interactions and functions might be impaired. A side-by-side comparison of MBP- and His-tagged pol IV with native pol IV ⫾ ␤/␥
show that His-tagged and native pol IV have similar average
processivities, ⬃50 nt in the presence of the accessory proteins.
In contrast, MBP-tagged pol IV has an average processivity of
only about 6 – 8 nt (Fig. 1). The various forms of pol IV catalyze
distributive synthesis in the absence of ␤/␥ (Fig. 1) (7, 20, 21).
The average processivity of native pol IV with ␤/␥ is about
6-fold lower than reported previously (20), possibly because of
differences in template sequences. Despite its modest average
processivity, native pol IV adds as many as 500 nt in a single
binding event (Fig. 1). Although the interaction between pol IV
and ␤/␥ appears to be unaffected by the presence of an Nterminal His tag, the processivity is severely impacted when
the enzyme is encumbered with a 42-kDa N-terminal MBP tag
(Fig. 1).
Analyses of the mutational spectra were carried out using
native pol IV ⫾ ␤/␥ and His-tagged pol IV (Fig. 2). The lacZ
average mutant frequencies and mutation distributions were
essentially indistinguishable for each form of the enzyme (Fig.
2 and Table I). The average mutation frequencies are 3.0 ⫻
10⫺2 and 3.6 ⫻ 10⫺2 for native pol IV in the presence and
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
Materials—Oligonucleotides were synthesized on an Applied Biosystems 392 DNA/RNA synthesizer and purified on a denaturing 16%
polyacrylamide gel. Tetrahydrofuran phosphoramidite was purchased
from Glenn Research. Primers were labeled with 32P using T4 polynucleotide kinase (U. S. Biochemical Corp.). Ultrapure deoxynucleoside
triphosphates and ATP were purchased from Amersham Biosciences,
and [␥-32P]ATP was purchased from ICN. Single-stranded M13 DNA
was purified as described by Sambrook et al. (28).
Buffers—Reaction buffer (1⫻) consists of 20 mM Tris, pH 7.5, 8 mM
MgCl2, 5 mM DTT, 0.1 mM EDTA, 25 mM sodium glutamate, 40 ␮g/ml
bovine serum albumin, and 4% glycerol. Stop solution is 95% formamide, 30 mM EDTA, and 0.25% bromphenol blue.
Enzymes—Maltose-binding protein tagged pol IV (MBP-pol IV), SSB,
␤- and ␥-complex were purified as described previously (7). The pET16b
(Novagen) was used to express histidine-tagged (His tag) pol IV and
native pol IV. The entire dinB-coding sequence was amplified by PCR
from E. coli genomic DNA by Pfu DNA polymerase. The resulting PCR
products were cloned into NdeI and BamHI or NcoI and BamHI sites of
pET16b to generate expression constructs for His tag and native pol IV,
respectively. His tag pol IV was overexpressed in BL21(DE3)/pLysS at
30 °C and purified by nickel affinity column followed by gel filtration on
Superdex-75 column similar to a protocol described by Wagner et al.
(21).
Native pol IV protein was overproduced in E. coli strain Turner/
pLysS (Novagen) and grown in LB medium in the presence of 20 ␮g/ml
chloramphenicol and 100 ␮g/ml ampicillin. Expression of pol IV was
induced by adding isopropyl-␤-D-1-thiogalactopyranoside to 1 mM and
growing for 3– 4 h at 30 °C. Cells (⬃20 g) were collected by centrifugation and were resuspended in 50 ml of lysis buffer (50 mM Tris-HCl, pH
7.5, 1 M NaCl, 10% sucrose, 2 mM DTT, 1 mM EDTA, and protease
inhibitor mixtures). Cells were lysed using freshly prepared lysozyme (2
mg/ml), and the clarified extract was collected. pol IV was then precipitated by ammonium sulfate added to 30% saturation (w/v) and stirring
for 10 min. The precipitate was subjected to gel filtration in GF buffer
(20 mM Tris-HCl, pH 7.5, 1 M NaCl, 0.1 mM EDTA, 1 mM DTT) using an
Amersham Biosciences Superdex-75 XK-26/60 gel filtration column. pol
IV fractions were pooled, dialyzed overnight in PC buffer (20 mM TrisHCl, pH 7.5, 0.1 mM EDTA 1 mM DTT, 10% glycerol), containing 200 mM
NaCl, and then subjected to phosphocellulose chromatography (P-11,
Whatman). After washing extensively with PC buffer ⫹ 200 mM NaCl,
pol IV was eluted with a linear gradient of 200 –500 mM NaCl. Fractions
containing native pol IV (⬎99% pure) were pooled and stored at ⫺70 °C.
Gap-filling Fidelity Assays—Fidelity of DNA synthesis by pol IV was
measured using an M13mp2 DNA substrate with a single-stranded gap
containing a portion of the lacZ ␣-complementation gene target constructed as described previously (29). Gap-filling synthesis reactions
(25-␮l volume) were carried out in 1⫻ reaction buffer containing 50 fmol
of gapped DNA, 5 pmol of His tag pol IV or native pol IV, and 1 mM each
of the four dNTP substrates. The reactions were supplemented with 500
␮M ATP when ␤- (200 fmol) and ␥-complex (50 fmol) were present.
Reaction mixtures were incubated at 37 °C for 30 min, quenched by
adding 2 ␮l of 200 mM EDTA, and analyzed as described (29).
Gel Fidelity Assays—Mispair formation and mispair extension fidelity were determined using a gel fidelity assay (30). Four templates and
five primers were used. The four 80-mer templates had the following
sequence: 5⬘-TGAGCGTTTTTCCTGTTGCAATGGCNGGCGGTAATATTGTTCTGGATATTACCAGCAAGGCCGATAGTTTGAGTTCTTCT3⬘, where N is A, C, G, or T for each template, respectively. The primer
used for the mispair formation assay is 29-mer complementary to the
template such that on annealing its 3⬘-end is adjacent to base N of the
template. The primers used for the mispair extension assay are 30-mers
identical in sequence to the mispair formation primer with the addition
of a different base to the 3⬘-end of each primer. Annealing a primer to
each of the four templates results in one complementary primer
template complex and three with a 3⬘-terminal mismatch.
Primer extension reactions were in reaction buffer using 4 nM 32Plabeled primer-template, 600 nM SSB, 80 nM ␤-complex, 20 nM ␥⫺complex, 400 ␮M ATP, 2 nM of either His-tagged or native pol IV, and dNTP
of varying concentrations in a total volume of 10 ␮l. Reactions were
quenched by the addition of 20 ␮l of stop solution. Apparent Vmax and
Km values were determined as described (31). The nucleotide misincor0
poration efficiency (finc) and intrinsic mismatch extension (f ext
) are
expressed as ratios of apparent second-order rate constants (31–34)
34199
34200
E. coli pol IV Frameshift Mutation Mechanism
absence of ␤/␥, respectively, and 3.7 ⫻ 10⫺2 for His-tagged pol
IV without ␤/␥ (Table I). Both the native and His-tagged proteins were able to fill the 407-nt gap completely with or without
␤/␥ (data not shown).
Mutational spectra were obtained by purifying and sequencing DNA obtained from randomly selected independent lacZa
mutants (light blue or colorless plaques), see under “Experimental Procedures.” A comparison of the mutation spectra of
His tag pol IV and native pol IV ⫾ ␤/␥, reveals a similar pattern
and distribution of mutations (Table I and Fig. 2). We have
therefore combined the data from all of the experiments to
arrive at a single pol IV mutational spectrum (Table II and Fig.
2, A and B).
The average frameshift and base substitution error rates are
2.1 ⫻ 10⫺4 and 5.1 ⫻ 10⫺5, respectively (Table I). Frameshifts
dominate the spectrum (⬃80%) including a sizable number of
⫺2 deletions (Table II and Fig. 2B). A small but not insignificant number of ⫺3, ⫺4, and ⫺5 deletions are also present (Fig.
2B). Base substitutions compose ⬃17% of the spectrum, and all
are represented, except for A 3 T transversions. T 3 G transversions (1.1 ⫻ 10⫺4) occur most frequently with the other
transitions and transversions having frequencies between
about 7 ⫻ 10⫺5 and 2.5 ⫻ 10⫺6 (Table II). A majority of base
substitutions (⬃69%) occur at template T sites. The overall
mutational specificities are consistent with mutational data
when pol IV is overexpressed in vivo (12).
An analysis of the frameshift pattern reveals that although
single base deletions are most prevalent (81%), there is an
unusually large fraction of 2-base deletions (15%) (Table II and
Fig. 2B). Two single base insertions were also detected. The
distribution of ⫺1 frameshifts favors some homopolymeric runs
depending on the number of identical bases in the run (Fig. 2A).
The rate of ⫺1 frameshifts in a non-homopolymeric run is 6.2 ⫻
10⫺5. Homopolymeric runs of four or five bases are the strongest hotspots for ⫺1 deletions by more than an order of magnitude. The number of 4- and 5-base runs in the lacZ␣ target
sequence are 3 and 1, respectively. The rates in runs of 2, 3, and
4 or more bases are 2.9 ⫻ 10⫺4, 7.5 ⫻ 10⫺5,and 1.1 ⫻ 10⫺3,
respectively. Nine of the 29 2-base runs have elevated ⫺1
frameshift rates, whereas none of the nine 3-base runs are
hotspots for ⫺1 frameshifts (Fig. 2A). If classical Streisinger
slippage, in which primer and template strands slip relative to
each in homopolymeric runs (35), were the principal mechanism for generating ⫺1 deletions, then one would expect to
observe more and not less mutations in homopolymeric runs of
3 compared with 2.
One common characteristic of sequences in which ⫺1 frameshifts are observed is the presence of an adjacent template 5⬘-G
whether or not there is a homopolymeric run. Although G is not
over-represented in the lacZ target sequence, 60% of all ⫺1
frameshifts have an adjacent 5⬘-G. Two of the three 4-base runs
have adjacent 5⬘-Gs, and these are the strongest hotspots.
There were a significant number of ⫺2 frameshift mutations,
along with 3-, 4-, and 5-consecutive base deletions. As with the
single base deletions, pyrimidines were preferentially deleted.
Of a total of 46 observed ⫺2 frameshifts, 34 (74%) are deletions
of two consecutive pyrimidine bases. In contrast, only three ⫺2
frameshifts involve deletions of two consecutive purines.
pol IV-generated ⫺1 Frameshifts by dNTP-stabilized Misalignment—In addition to the Streisinger slippage mechanism
(Fig. 3A, bottom panel), there are two other mechanisms that
can lead to ⫺1 deletions, dNTP-stabilized misalignment (19,
27) (Fig. 3A, top panel) and “misinsertion misalignment” (36,
37) (Fig. 3A, middle panel). A misinsertion misalignment occurs when a nucleotide misincorporation is followed by p/t DNA
slippage, i.e. repositioning of the misincorporated nucleotide
opposite a complementary downstream template base (37) (Fig.
3A, middle panel). However, a situation leading to the same
structural and mutational end point takes place when an incoming dNTP bound at the polymerase active site is incorporated opposite a complementary downstream template base on
a misaligned p/t DNA (19) (Fig. 3A, top panel).
We have obtained direct spectroscopic evidence supporting
the dNTP-stabilized misalignment model. Incorporation of
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
FIG. 1. A comparison of the processivity of native pol IV with MBP- and
His-tagged polymerases in the presence and absence of ␤/␥ processivity
subunits. The processivity of native pol
IV and pol IV containing N-terminal MBP
or His tag was measured in either the
presence or absence of ␤ sliding clamp
and ␥ clamp loading complex, using a 32Plabeled 30-mer primer annealed to a circular ssM13mp2 DNA. Reactions contained 10 nM p/t DNA substrate, 200 ␮M of
each of the four dNTP and, where indicated, 80 nM ␤-, 20 nM ␥⫺complex and 500
␮M ATP. The concentrations of pol IV
used in individual reactions were 0.63,
1.25, 2.5, 5, and 10 nM.
E. coli pol IV Frameshift Mutation Mechanism
34201
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
FIG. 2. pol IV base substitution and frameshift mutation spectra. A, spectrum of base substitution and single base deletion and addition
mutations. B, spectrum of 2-, 3-, 4- and 5-base deletion mutations. A gapped DNA template containing a lacZ␣ mutational target gene was copied
with either native or His-tagged pol IV. A, the sequence of the mutational target is shown with base substitutions indicated as the appropriate base
above the target sequence, single base deletions indicated as ⌬ below the sequence, and additions as ⫹ followed by the added base above the
sequence. The mutations are color-coded where black refers to mutations generated by His tag pol IV, blue refers to native pol IV, and red to native
pol IV ⫹ ␤/␥ processivity protein subunits. 375 independently isolated lacZ␣ mutants were sequenced.
dTMP opposite the fluorescent base analog 2AP is accompanied
by a quenching of the fluorescence signal (Fig. 3B, lower trace).
A reduction in 2AP fluorescence intensity occurs when 2AP
stacks within the helical plane while forming a W-C base pair
with T (38, 39). In contrast, an increase in fluorescence occurs
when 2AP is extrahelical (40, 41), i.e. when 2AP is “flipped” out
of the helical plane. The incorporation of dGMP opposite template C, located immediately downstream from 2AP, results in
34202
E. coli pol IV Frameshift Mutation Mechanism
TABLE I
Mutation frequency of pol IV when copying the LacZ␣
target sequence in vitro
Gapped DNA (control)
His tag pol IV
Native pol IV
Native pol IV ⫹ ␤/␥
Average
Average error rate (⫻10⫺4)
Mutant
frequencya
(⫻10⫺4)
Base
substitution
Frameshift
3.1
365
355
300
340
NDb
0.43
0.60
0.49
0.51
ND
2.5
2.0
1.7
2.1
a
The mutant frequency is the average of four determinations. In each
case 5,000 – 45,000 total plaques were scored.
b
ND, not determined.
TABLE II
Base substitutions and frameshifts generated by pol IV when copying
the LacZ␣ target sequence
Mutations
Occurrences
Error ratea
66 (17.6%)
25 (6.7%)
5
1
3
16
1.8 ⫻ 10⫺5
7.5 ⫻ 10⫺6
2.0 ⫻ 10⫺5
6.7 ⫻ 10⫺5
Transversions
A 3 C (A䡠dGMP)
A 3 T (A䡠dAMP)
G 3 C (G䡠dGMP)
G 3 T (G䡠dAMP)
T 3 A (T䡠dTMP)
T 3 G (T䡠dCMP)
C 3 A (C䡠dTMP)
C 3 G (C䡠dCMP)
41 (10.9%)
1
0
1
1
11
19
3
5
2.5 ⫻ 10⫺6b
⬍7.5 ⫻ 10⫺6
9.0 ⫻ 10⫺6
2.8 ⫻ 10⫺6b
5.8 ⫻ 10⫺5
1.1 ⫻ 10⫺4
6.2 ⫻ 10⫺6b
1.5 ⫻ 10⫺5b
Frameshifts
1-Base deletion,
addition
⌬A
⌬G
⌬C
⌬T
⫹A
⫹G
301 (80.3%)
41
17
94
91
1
1
1.6 ⫻ 10⫺4
6.8 ⫻ 10⫺5
2.6 ⫻ 10⫺4
3.2 ⫻ 10⫺4
2-Base deletion
⌬TT
⌬TC or ⌬CT
⌬CC
⌬AT or ⌬TA
⌬AC or ⌬CA
⌬GT or ⌬TG
⌬GA or ⌬AG
⌬GG
⌬AA
20
9
5
4
3
2
1
1
1
1.4 ⫻ 10⫺5
6.3 ⫻ 10⫺6
3.5 ⫻ 10⫺6
2.8 ⫻ 10⫺6
2.1 ⫻ 10⫺6
1.4 ⫻ 10⫺6
0.7 ⫻ 10⫺6
0.7 ⫻ 10⫺6
0.7 ⫻ 10⫺6
3-Base deletion
4
4-Base deletion
4
5-Base deletion
2
c
Others
Total
8 (2.1%)
375 (100%)
a
Except as noted, values determined according to Ref. 29.
Values based on number of mutations scored/((number of plaques
screened) ⫻ (number of occurrences of the template base in the sequence) ⫻ 0.6).
c
Others include 6 large deletion mutations and 2 tandem base substitution mutations.
b
an increase in fluorescence that is caused by a reduction in base
stacking when 2AP is extrahelical (Fig. 3B, upper trace).
There is no change in fluorescence intensity when the downstream template base is changed from C to G, because the
incoming dGTP cannot pair opposite G and can therefore no
longer stabilize the misaligned structure (Fig. 3B, inset panel).
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
Base substitutions
Transitions
A 3 G (A䡠dCMP)
G 3 A (G䡠dTMP)
C 3 T (C䡠dAMP)
T 3 C (T䡠dGMP)
The inability to incorporate C opposite 2AP also results in no
change in the fluorescence intensity (Fig. 3B, middle trace).
The incorporation of G opposite C cannot be attributed to a
slippage of a template A upstream of 2AP (i.e. Streisinger
slippage, Fig. 3A, bottom panel), because realignment of 2AP
opposite T at the 3⬘-primer end would result in a quench rather
than increase in fluorescence. The same experiment performed
with E. coli pol I Klenow fragment lacking 3⬘-exonuclease activity shows no evidence for 2AP flipping (data not shown).
A hallmark of the dNTP-stabilized misalignment mechanism
is that apparent Km,dNTP values are closer in magnitude to
Km,dNTP for correct nucleotide incorporation on properly
aligned p/t DNA, in contrast to much larger Km,dNTP values
associated with nucleotide misincorporations (19, 27). The apparent Km for incorporation of dTMP opposite 2AP is 264 ⫾ 183
␮M (corresponding to Fig. 3B, lower trace) whereas the Km for
incorporation of dGMP opposite C on misaligned p/t DNA, with
2AP out of the helical plane, is 740 ⫾ 474 ␮M (corresponding to
Fig. 3B, upper trace). In contrast, the Km values for incorporation of dCMP or dGMP directly opposite 2AP (corresponding to
Fig. 3B, middle trace and 3B, inset, respectively) are ⬎3 mM. In
the latter experiments, nucleotide misincorporation rates
measured as a function of dCTP or dGTP concentration remained linear at the highest substrate concentrations (⬃3 mM)
that did not inhibit pol IV synthesis.
pol IV-catalyzed Misincorporation Efficiencies—The gap-filling analysis (Fig. 2) shows that small frameshifts account for
roughly 80% of pol IV errors but that the remaining ⬃20% are
divided among all possible base substitutions, except for A to T
transversions that were not observed (Table II). T 3 C transitions and either T 3 G or T 3 A transversions are the most
prevalent base substitution (Table II). By using a steady-state
gel fidelity assay (see “Experimental Procedures”), we measured nucleotide misincorporation efficiencies, finc, for each
transition and transversion mutagenic pathway catalyzed by
either native or His-tagged pol IV in the presence of ␤/␥-complex and SSB (Table III).
For native pol IV, nucleotide misincorporation efficiencies
are typically larger at pyrimidine template sites (Table III and
Fig. 4). In contrast, the misincorporation efficiencies at purine
template sites are ⬃5–10-fold lower (Table III and Fig. 4). A
clustering of finc values are observed for pyrimidine and for
purine template bases, with the only outlier being the T䡠dCMP
mispair (Fig. 4). The only mutation absent from the gap-filling
spectrum (Table II) is an A 3 T transversion which has A䡠A as
a base mispaired intermediate. In agreement, the incorporation of dAMP opposite A, finc (A䡠A) ⫽ 2.7 ⫻ 10⫺5, is the least
efficient mismatch catalyzed by pol IV (Fig. 4 and Table III).
These results are consistent with the mutational spectrum
derived from the gap-filling mutational assay (Table II).
Misincorporations taking place via dNTP-stabilized misalignments occur most readily in AT-rich DNA (Fig. 3) but tend
to be inhibited in more stable surroundings (data not shown).
The template target site N used to measure misincorporation
kinetics is embedded in a GC-rich region, 5⬘ . . . GGCNGGCGG . . . 3⬘ (see “Experimental Procedures”). Consequently,
there is no increase in dGMP䡠N misincorporations relative to
other base substitution errors (Table III). However, the 3-fold
larger dGMP䡠G misincorporation efficiency (finc ⫽ 3.2 ⫻ 10⫺4)
compared with dTMP䡠G and dAMP䡠T (finc ⫽ 1.1 ⫻ 10⫺4) (Table
III) is an indication that Streisinger slippage may be occurring.
The presence of an N-terminal His tag does not appear to
exert an undue influence on pol IV-catalyzed incorporation
efficiencies for 9 of the 12 mispairs (Table III). The exceptions
are dAMP䡠A, dTMP䡠C, and dCMP䡠T where His-tagged pol IV
E. coli pol IV Frameshift Mutation Mechanism
34203
TABLE III
pol IV nucleotide misincorporation efficiency, finc
Kinetics of nucleotide incorporation by pol IV in the presence of ␤/␥ and SSB were measured using a standing start primer extension assay. The
velocity of nucleotide addition opposite a template base is measured at varying concentrations of dNTP substrates (see “Experimental Procedures”).
Reactions are run for each of the 16 canonical base pairs, and the nucleotide incorporation rate is plotted as a function of dNTP concentration. A
nonlinear least squares fit of the plots to a rectangular hyperbola is used to determine the kinetic parameters, apparent Km and Vmax. The apparent
Vmax/Km is a measure of reaction efficiency, and dividing the efficiency of formation of a mispair by that of a correct pair yields the misincorporation
efficiency (finc). The standard errors for finc are approximately ⫾30%.
Template 䡠 dNMP
% min
Native pol IV
His-tagged pol IV
finc
Vmax/Km
% min⫺1 ␮M⫺1
Vmax/Km
finc
% min⫺1 ␮M⫺1
T:A
T䡠G
T䡠C
T䡠T
1.8 ⫻ 10⫺1
9.2 ⫻ 10⫺5
9.8 ⫻ 10⫺5
1.1 ⫻ 10⫺4
1
5.1 ⫻ 10⫺4
5.5 ⫻ 10⫺5
6.6 ⫻ 10⫺4
4.5 ⫻ 10⫺2
3.5 ⫻ 10⫺5
4.4 ⫻ 10⫺5
2.4 ⫻ 10⫺5
1
7.8 ⫻ 10⫺4
9.7 ⫻ 10⫺4
5.3 ⫻ 10⫺4
C:G
C䡠A
C䡠T
C䡠C
9.1 ⫻ 10⫺2
1.5 ⫻ 10⫺4
4.6 ⫻ 10⫺5
5.2 ⫻ 10⫺5
1
1.7 ⫻ 10⫺3
5.1 ⫻ 10⫺4
5.7 ⫻ 10⫺4
6.7 ⫻ 10⫺3
5.5 ⫻ 10⫺6
8.7 ⫻ 10⫺5
3.9 ⫻ 10⫺6
1
8.1 ⫻ 10⫺4
1.3 ⫻ 10⫺2
5.8 ⫻ 10⫺4
G:C
G䡠T
G䡠A
G䡠G
8.6 ⫻ 10⫺1
9.5 ⫻ 10⫺5
9.7 ⫻ 10⫺6
2.7 ⫻ 10⫺4
1
1.1 ⫻ 10⫺4
1.1 ⫻ 10⫺4
3.2 ⫻ 10⫺4
3.0 ⫻ 10⫺1
1.1 ⫻ 10⫺4
6.8 ⫻ 10⫺6
1.8 ⫻ 10⫺4
1
3.7 ⫻ 10⫺4
2.2 ⫻ 10⫺5
5.8 ⫻ 10⫺4
A:T
A䡠C
A䡠G
A䡠A
6.8 ⫻ 10⫺1
1.9 ⫻ 10⫺4
7.3 ⫻ 10⫺5
1.8 ⫻ 10⫺5
1
2.9 ⫻ 10⫺4
1.1 ⫻ 10⫺4
2.7 ⫻ 10⫺5
3.8 ⫻ 10⫺2
2.5 ⫻ 10⫺5
1.8 ⫻ 10⫺5
3.5 ⫻ 10⫺5
1
6.6 ⫻ 10⫺4
4.7 ⫻ 10⫺4
9.1 ⫻ 10⫺4
exhibits 34-, 25-, and 20-fold increased error frequencies,
respectively.
pol IV-catalyzed Mismatch Extension Efficiencies—It has
been speculated that pol IV might play a role in relieving
stalled replication forks on undamaged DNA (18) by extending
mismatches that are refractory to proofreading by replicative
pol III core (19). In support of this possibility, of the 12 possible
natural base mismatches, 9 are extended by native pol IV with
efficiencies exceeding ⬃10⫺4 (Table IV and Fig. 4). The most
easily formed mismatch, finc (C䡠dAMP) ⫽ 1.7 ⫻ 10⫺3 (Table III),
0
is also extended with high efficiency, fext
(C䡠A-3⬘-primer end) ⫽
8.6 ⫻ 10⫺4 (Table IV). Two other easily formed mispairs are
also extended with efficiencies approaching 10⫺3, A䡠C-3⬘primer end and T䡠T (Fig. 4). However, the T䡠C mismatch, which
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
FIG. 3. The change in fluorescence intensity of 2-aminopurine undergoing replication by pol IV. A, three models for generation of ⫺1
frameshift mutations. B, traces showing changes in 2AP fluorescence intensity accompanying the incorporation of dGMP (upper trace), inability
to incorporate dCMP (middle trace), and the incorporation of dTMP (lower trace). Inset panel, the absence of a change in 2AP fluorescence intensity
showing that dGMP can no longer be incorporated when G replaces C downstream of 2AP.
34204
E. coli pol IV Frameshift Mutation Mechanism
TABLE IV
0
pol IV mismatch extension efficiency, fext
0
Intrinsic mismatch extension efficiencies (f ext) were calculated by dividing the efficiency of extension of a primer with a mispaired 3⬘-end,
0
apparent (Vmax/Km) value, by that of a correctly paired primer (see “Experimental Procedures”). The standard errors for f ext
are approximately
⫾30%.
Base pair,
template/primer
Native pol IV
His-tagged pol IV
0
fext
Vmax/Km
% min
⫺1
␮M
⫺1
Vmax/Km
% min
⫺1
␮M
f
o
ext
⫺1
T:A
T䡠G
T䡠C
T䡠T
3.2 ⫻ 10⫺1
1.1 ⫻ 10⫺4
2.4 ⫻ 10⫺4
2.8 ⫻ 10⫺4
1
3.6 ⫻ 10⫺4
7.3 ⫻ 10⫺4
8.7 ⫻ 10⫺4
1.5 ⫻ 10⫺2
3.6 ⫻ 10⫺5
7.8 ⫻ 10⫺6
1.5 ⫻ 10⫺5
1
2.3 ⫻ 10⫺3
5.1 ⫻ 10⫺4
1.0 ⫻ 10⫺3
C:G
C䡠A
C䡠T
C䡠C
7.3 ⫻ 10⫺1
6.2 ⫻ 10⫺4
1.1 ⫻ 10⫺4
8.8 ⫻ 10⫺5
1
8.6 ⫻ 10⫺4
1.5 ⫻ 10⫺4
1.2 ⫻ 10⫺4
4.7 ⫻ 10⫺1
1.2 ⫻ 10⫺4
3.3 ⫻ 10⫺5
2.0 ⫻ 10⫺5
1
2.6 ⫻ 10⫺4
7.0 ⫻ 10⫺5
4.2 ⫻ 10⫺5
G:C
G䡠T
G䡠A
G䡠G
1.6 ⫻ 100
1.7 ⫻ 10⫺4
5.8 ⫻ 10⫺5
1.7 ⫻ 10⫺5
1
1.1 ⫻ 10⫺4
3.7 ⫻ 10⫺5
1.1 ⫻ 10⫺5
3.1 ⫻ 10⫺1
1.1 ⫻ 10⫺4
2.7 ⫻ 10⫺5
1.7 ⫻ 10⫺5
1
3.6 ⫻ 10⫺4
8.6 ⫻ 10⫺5
5.4 ⫻ 10⫺5
A:T
A䡠C
A䡠G
A䡠A
5.7 ⫻ 10⫺1
5.2 ⫻ 10⫺4
2.9 ⫻ 10⫺5
7.0 ⫻ 10⫺5
1
9.1 ⫻ 10⫺4
5.1 ⫻ 10⫺5
1.2 ⫻ 10⫺4
3.2 ⫻ 10⫺2
2.3 ⫻ 10⫺4
1.6 ⫻ 10⫺5
4.2 ⫻ 10⫺5
1
7.2 ⫻ 10⫺3
5.0 ⫻ 10⫺4
1.3 ⫻ 10⫺3
is also extended with an efficiency ⬃10⫺3, is one of the most
difficult mismatches to make, finc (T䡠dCMP) ⫽ 5.5 ⫻ 10⫺5
(Table III and Fig. 4).
Three of the four Puo䡠Puo transversion mispaired intermediates, A䡠G, G䡠A, and G䡠G, are extended with lowest efficiencies,
0
fext
⬍4 ⫻ 10⫺5. The A䡠A mismatch, although extended some0
what more efficiently (fext
⬍1.2 ⫻ 10⫺4), is the least probable to
be formed (finc ⫽ 1.8 ⫻ 10⫺5) (Fig. 4). The latter observation
reflects the absence of A 3 T transversion in the lacZ␣ mutational spectrum (Table II). The other three improbable Puo䡠Puo
transversions appear just once each in the spectrum out of 66
base substitutions scored (Table II). In contrast, the two most
prevalent transversions in the spectrum, T 3 G (19/66) and
T 3 A (11/16), involve the Pyd䡠Pyd mispaired intermediates,
T䡠dCMP and T䡠dTMP, respectively. The T䡠dTMP intermediate
is both formed and extended with high efficiency ⬃10⫺3,
whereas the T䡠dCMP intermediate is formed with an efficiency
of ⬃10⫺4, but once formed is readily extended ⬃10⫺3 (Fig. 4).
One transition, T 3 C, appears multiple times in the lacZ␣
mutational spectrum (16/66). The T䡠dGMP mispaired intermediate is both formed and extended efficiently, ⬃3 to 5 ⫻ 10⫺4
(Tables II and III and Fig. 4). The presence of an N-terminal
His tag appears to have a small effect on mismatch extension
efficiencies, somewhat less than its effect on misincorporation
efficiencies. The most notable perturbations are an ⬃10-fold
increase in mismatch extension efficiencies for bases mispaired
opposite template A, a 2–5-fold increase for mispairs opposite
G, and an ⬃2–3-fold reduction in mispair extensions opposite
template C.
The general conclusion is that pol IV appears to extend a
significant fraction of mispaired intermediates with relatively
high efficiency (⬎10⫺4). The kinetic data are in overall agreement with the mutational spectrum. Thus, mismatches that
are both formed and extended easily appear multiple times as
independently scored mutations in the lacZ␣ mutational spectrum. Mismatches that are both poorly formed and extended
are either not observed (A 3 T transversions) or are only
sparsely represented in the spectrum.
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
FIG. 4. A plot of pol IV misincorporation efficiency, finc, versus intrinsic
0
mismatch extension efficiency, f ext
.
Nucleotide misincorporation and mismatch extension data were obtained for
pol IV ⫹ ␤/␥ using a gel kinetic fidelity
assay, see “Experimental Procedures.”
The misincorporation data are from Table
III, and the mismatch extension data are
from Table IV. The data points shown
above the line indicate lower values for
the efficiency misincorporation compared
with mismatch extension; data falling below the line indicate higher values for
misincorporation compared with mismatch extension. The misincorporation
efficiency finc is the reciprocal of the polymerase fidelity.
E. coli pol IV Frameshift Mutation Mechanism
34205
DISCUSSION
pol IV-catalyzed Abasic Site Translesion Synthesis—An examination of the mutational spectrum shows that pol IV is
heavily prone to make short deletion mutations on undamaged
DNA templates (Fig. 2). The “flipping” of a template 2AP out of
the helical plane (Fig. 3) provides a likely mechanism to explain
the ease with which ⫺1 frameshifts occur. However, since pol
IV also appears to be involved in copying damaged (13, 14) as
well as undamaged DNA (11, 22–24, 42), we have investigated
the products formed when copying a simple form of template
damage, a non-instructive abasic moiety (Fig. 5).
For the lesion-free control template, pol IV in the absence of
␤/␥ extends 61 bases, stopping two bases short of the end of the
template strand (Fig. 5A, 1st lane). The addition of SSB permits
the addition of one more base (Fig. 5A, 2nd lane). A full-length
product 63 bases long is synthesized in the presence of ␤/␥ and
does not require the presence of SSB (Fig. 5A, 3rd and 4th
lanes). For the templates with an abasic site, the products of
the extension reaction are 1 base shorter compared with those
of the normal template (Fig. 5, B and C), 60 added bases for pol
IV alone and 62 added bases for pol IV ⫹ ␤/␥ (Fig. 5, B and C).
pol IV makes shorter products in both sequence contexts 5⬘GGX-3⬘ (Fig. 5B) and 5⬘-TTX-3⬘ (Fig. 5C). Therefore, when
copying an abasic site, pol IV does not appear to insert a base
opposite the lesion but instead skips past the abasic site to
generate a single base deletion via dNTP-stabilized misalignment, as has been observed previously with eukaryotic pol ␤
but not with pol ␣ (27).
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
FIG. 5. pol IV generates deletions during abasic site translesion synthesis. A, pol IV-catalyzed synthesis on undamaged DNA in
the presence or absence of ␤/␥ and SSB. B, pol IV synthesis as in A on
a template containing an abasic moiety 1 nt from the 3⬘-primer end (63
nt from the 5⬘-end of the template strand). C, pol IV synthesis as in A
on a template containing an abasic moiety 13 nt from the 3⬘-primer end
(51 nt from the 5⬘-end of the template strand). Reactions were carried
out using a 32P-labeled 30-mer primer annealed to 120-mer synthetic
template (10 nM p/t DNA). Each reaction contained native pol IV (100
nM), the four dNTP substrates (200 ␮M each). SSB (250 nM), ␤- (200 nM)
and ␥-complex (50 nM) were added as indicated. X indicates the location
of the abasic moiety.
Three DNA polymerases are induced as part of the SOS
regulon in response to DNA damage in E. coli. pol V appears to
play a predominant role in SOS mutagenesis by generating
mutations targeted directly to DNA template lesions (7, 43). pol
IV, on the other hand, is believed responsible for generating
untargeted mutations (11, 22–24, 42). pol II, a much higher
fidelity polymerase (5), has been found to play a pivotal role
during error-free replication restart (8, 9). This paper is focused
on determining the fidelity properties of pol IV when copying
undamaged DNA and comparing the biochemical properties
determined in vitro with mutation spectra. We have investigated the mechanism of pol IV-generated ⫺1 frameshifts with
relation to unusual structural features of a pol IV homolog
(Dpo4 from the archeon S. solfataricus), which also makes ⫺1
frameshifts avidly (25, 26).
A dNTP-stabilized Misalignment Model to Account for ⫺1
Frameshifts—A characteristic property of pol IV is its ability to
generate small deletion mutations on presumably undamaged
DNA in vivo (42). In non-dividing cells, pol IV is responsible for
causing about 80% of the adaptive frameshift mutations concentrated within homopolymeric runs (22, 23). There are numerous different types of slipped p/t DNA structures that can
in principal lead to ⫺1 frameshift mutations (36).
We have previously provided kinetic evidence supporting a
particular type of misaligned DNA structure to explain single
base deletions on damaged (27) and undamaged (19) DNA. This
aberrant structure is characterized by having an incoming
dNTP bound at the polymerase active site but which is not
paired opposite the next available template position. Instead,
the bound dNTP is misaligned opposite a template base immediately downstream from next template base (Fig. 3A, upper
panel). We have coined the term dNTP-stabilized misalignment
(27) to describe this particular molecular model, which can
generate ⫺1 frameshifts in either the presence or absence of
homopolymeric template runs (19, 36).
Direct evidence supporting the dNTP-stabilized misalignment model as a molecular mechanism for pol IV-catalyzed ⫺1
frameshifts is provided by the observation of a “flipped out”
template 2AP occurring when dGMP is incorporated opposite a
downstream template C (Fig. 3B). There is a substantial increase in 2AP fluorescence intensity when the analog is no
longer stacked within a helix (Fig. 3B, upper trace) (40, 41). In
contrast to the increase in fluorescence when C is located to the
5⬘-side of 2AP, there is no change in fluorescence when the G
replaces C downstream from 2AP (Fig. 3B, inset). As expected,
the fluorescence intensity decreases when dTMP is incorporated directly opposite 2AP on a normally aligned p/t DNA (Fig.
3B, lower trace). The fluorescence intensity remains unchanged
when dCMP fails to be incorporated (Fig. 3B, middle trace).
We have shown previously that the apparent Km,dNTP values
that characterize “correct” nucleotide incorporation on misaligned p/t DNA ends (e.g. incorporation of dGMP opposite C,
Fig. 3A, upper panel) tend to be much closer to those corresponding to formation of standard W-C base pairs than to the
typically much higher values for direct misincorporations (e.g.
misincorporation of dGMP opposite 2AP, followed by slippage
to align dGMP opposite C, Fig. 3A, middle panel). In agreement
with our earlier observations, an apparent Km ⬃750 ␮M for
incorporation of dGMP opposite C on a misaligned p/t DNA is
closer to the value for the correct incorporation of dTMP opposite 2AP on a properly aligned p/t DNA (Km ⬃250 ␮M), compared with the misincorporation of dGMP opposite 2AP on a
properly aligned DNA molecule, which has a Km value exceeding 3 mM.
The structure of the S. solfataricus Dpo4 (a homolog of pol
34206
E. coli pol IV Frameshift Mutation Mechanism
in the minority, base substitutions nevertheless compose
roughly 18% of the pol IV-generated mutations in vitro (Table
II). T 3 G transversions and T 3 C transitions are the most
frequent base substitutions found in vitro (Table II and Fig.
2A), in agreement with in vivo mutational data when pol IV is
overexpressed (12). A 3 G and C 3 G mutations are also
favored in vitro, as mirrored by in vivo data, suggesting that pol
IV may tend to favor changes toward G:C pairs.
The obvious limitation when trying to compare in vitro fidelity data with in vivo mutational spectra stems from uncertainty surrounding the individual contributions made by the
five polymerases in vivo. For example, an unknown number of
untargeted mutations in vivo could result from pol IV replacing
pol III core at stalled replication forks (18) and perhaps extending mismatched ends that are refractory to proofreading. The
mutator effect emanating from overproduction of pol IV is
dependent on its interaction with the ␤ processivity clamp (46)
which suggests that switching between the two pols can take
place at the replication fork (3). Thus, pol III core-generated
errors could be extended by pol IV. Our data suggest that pol IV
0
is reasonably proficient at mismatch extension (fext
⬃10⫺3–
⫺5
10 , see Table IV and Fig. 4). However, pol IV appears to
catalyze mismatch extension far less efficiently than its human
counterpart, pol ␬ (47), by factors of 10 –1000-fold.
pol IV is regulated as part of the SOS response to DNA
damage. However, it is nevertheless expressed at high constitutive levels because the LexA repressor binds poorly to the
dinB/dinP operator consensus sequence (48). The uninduced
level of expression of pol IV of about 250 molecules per cell (15)
exceeds pol II, ⬃40 – 60 per cell (16), and pol V ⬍15 per cell (17).
Following SOS induction the levels of pols IV, II, and V are
increased to about 2500, 350, and 200 molecules per cell, respectively (15–17).
Although geared to function when DNA is damaged, there is
evidence supporting SOS polymerase function in the cell in the
absence of overt SOS induction (22–24). It has been suggested
that replication forks are likely to stall at least once per round
of replication when cells are growing normally, in the absence
of exogenous DNA damage (18). Perhaps pol IV is present at a
high constitutive level for the express purpose of rescuing
stalled replication forks on undamaged DNA, and in so doing
may even be responsible for a generating a significant fraction
of the spontaneous chromosomal mutations “blamed on” pol III.
pol II, also present at an elevated constitutive level, may be
engaged in rescuing replication forks that encounter an occasional spontaneously generated lesion, a so-called “cryptic” lesion (8, 9). The cell could, as a last resort, turn to pol V, whose
constitutive level is barely detectable, to copy heavily damaged
DNA requiring the induction of SOS.
Therefore, a degree of redundancy appears to be present
permitting each SOS pol to play either a primary or back-up
role when copying damaged or undamaged DNA in place of pol
III (3). The elimination of one or any combination of the SOS
pols results an inability of mutant cells to compete with wild
type cells when growing in stationary cultures in the absence of
exogenous DNA damage (24).
REFERENCES
1. Courcelle, J., Khodursky, A., Peter, B., Brown, P. O., and Hanawalt, P. C.
(2001) Genetics 158, 41– 64
2. Ohmori, H., Friedberg, E. C., Fuchs, R. P., Goodman, M. F., Hanaoka, F.,
Hinkle, D., Kunkel, T. A., Lawrence, C. W., Livneh, Z., Nohmi, T., Prakash,
L., Prakash, S., Todo, T., Walker, G. C., Wang, Z., and Woodgate, R. (2001)
Mol. Cell 8, 7– 8
3. Goodman, M. F. (2002) Annu. Rev. Biochem. 70, 17–50
4. Bonner, C. A., Hays, S., McEntee, K., and Goodman, M. F. (1990) Proc. Natl.
Acad. Sci. U. S. A. 87, 7663–7667
5. Cai, H., Yu, H., McEntee, K., Kunkel, T. A., and Goodman, M. F. (1995) J. Biol.
Chem. 270, 15327–15335
6. Goodman, M. F., and Tippin, B. (2000) Curr. Opin. Genet. & Dev. 10, 162–168
7. Tang, M., Pham, P., Shen, X., Taylor, J. S., O’Donnell, M., Woodgate, R., and
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
IV) ternary complex reveals limited and nonspecific contacts
with the replicating bases in the active site (25). Water might
not be effectively excluded from the active site of these polymerases, thus excluding an important source of discrimination
against mismatched base pairs (44, 45). The type II crystals
(25) of the Dpo4 ternary complex show that two template bases
occupy the active site simultaneously with the incoming nucleotide (ddGTP) bypassing an extrahelical template base (G),
aligning instead directly opposite the second template base (C).
The structural data are supported mechanistically by our spectroscopic data showing an extrahelical 2AP when pol IV incorporates dGMP opposite template C located immediately downstream of 2AP (Fig. 3B, upper trace; Fig. 3A, upper panel).
Relating pol IV Mutation Spectra to Fidelity—Before it became known that dinB encoded a DNA polymerase (21), there
was considerable evidence that the generation of small deletion
mutations coincided with the overexpression of DinB protein
(11). The biochemical data using a gap-filling lacZ␣ forward
mutational assay clearly illustrate the point that small deletions dominate the mutational spectrum by about a 4:1 margin
over base substitutions (Fig. 2B and Table II). Single base
deletions account for about 80% of the frameshifts observed,
accompanied by a substantial number of ⫺2 deletions (⬃15%).
According to the well documented Streisinger model (35),
slippage of primer and template strands relative to each other
in homopolymeric runs can result in frameshifts (Fig. 3A, bottom panel). A defining feature of the Streisinger slippage model
is that the mutation frequency at homopolymeric runs increases with the length of the run, because larger runs produce
increasingly more stable misaligned intermediates. The rates
of pol IV-generated ⫺1 frameshifts are as follows: 0.62 ⫻ 10⫺4
in a non-run, 2.9 ⫻ 10⫺4 in a 2-base run, 0.75 ⫻ 10⫺4 in a
3-base run, and 11 ⫻ 10⫺4 in runs of 4 or more bases (Table I
and Fig. 2A). The error rate for the deletion of a base in a run
of 4 or more identical bases is highest. However, the error rate
for a ⫺1 frameshift in a run of 3 bases is roughly equal to a
non-run and is about 4 times lower than observed for a run of
2 bases. Thus, whereas the Streisinger template slippage
mechanism might account for a subset of ⫺1 frameshifts, it is
unlikely to account for all of them. Instead, a sizable number of
⫺1 deletions could be occurring by dNTP-stabilized misalignment or misinsertion misalignment mechanisms (Fig. 3A, top
and middle panels). The dNTP-stabilized misalignment mechanism appears to explain a ⫺1 deletion occurring when pol IV
copies an abasic template lesion in vitro (Fig. 5). Similar properties have been observed with eukaryotic pol ␤, but not with
pol ␣ (27).
A potentially significant point is that whereas several deletions of two or more bases occur in short 2–3-base homopolymeric runs, others decidedly do not (Fig. 2B). Given the surprisingly large fraction of ⫺2-base deletions (15% of all
frameshifts) that appear in the spectrum, we speculate that the
active site of pol IV might also accommodate “looped out” template structures exceeding a single nucleotide, e.g. a slightly
more elaborate version of the dNTP-stabilized misalignment
mechanism.
The lacZ␣ in vitro mutational spectrum (Fig. 2) is generally
consistent with measurements performed in vivo for mutations
in the cII gene of ␭-phage when pol IV is overexpressed (12).
Single base deletions constitute a large majority of the mutations observed in vivo (11, 12) and in vitro (21); (Fig. 2A and
Table II); however, 2-base deletions (Fig. 2B and Table II) have
not been observed in vivo. Perhaps the large distortion introduced by 2-base deletions favor their excision by one of the
three proofreading-proficient polymerases in E. coli prior to
extension by proofreading-deficient pol IV. Although decidedly
E. coli pol IV Frameshift Mutation Mechanism
Press, Cold Spring Harbor, NY
29. Bebenek, K., and Kunkel, T. A. (1995) Methods Enzymol. 262, 217–232
30. Creighton, S., Bloom, L. B., and Goodman, M. F. (1995) Methods Enzymol. 262,
232–256
31. Boosalis, M. S., Petruska, J., and Goodman, M. F. (1987) J. Biol. Chem. 262,
14689 –14696
32. Mendelman, L. V., Petruska, J., and Goodman, M. F. (1990) J. Biol. Chem. 265,
2338 –2346
33. Goodman, M. F., Creighton, S., Bloom, L. B., and Petruska, J. (1993) Crit. Rev.
Biochem. Mol. Biol. 28, 83–126
34. Creighton, S., and Goodman, M. F. (1995) J. Biol. Chem. 270, 4759 – 4774
35. Streisinger, G., Okada, Y., Emrich, J., Newton, J., Tsugita, A., Terzaghi, E.,
and Inouye, M. (1966) Cold Spring Harbor Symp. Quant. Biol. 31, 77– 84
36. Kunkel, T., and Soni, A. (1988) J. Biol. Chem. 263, 14784 –14789
37. Bebenek, K., and Kunkel, T. A. (1990) Proc. Natl. Acad. Sci. U. S. A. 87,
4946 – 4950
38. Bloom, L. B., Otto, M. R., Eritja, R., Reha-Krantz, L. J., Goodman, M. F., and
Beechem, J. M. (1994) Biochemistry 33, 7576 –7586
39. Frey, M. W., Sowers, L. C., Millar, D. P., and Benkovic, S. J. (1995) Biochemistry 34, 9185–9192
40. Law, S. M., Eritja, R., Goodman, M. F., and Breslauer, K. J. (1996) Biochemistry 35, 12329 –12337
41. Beechem, J. M., Otto, M. R., Bloom, L. B., Eritja, R., Reha-Krantz, L. J., and
Goodman, M. F. (1998) Biochemistry 37, 10144 –10155
42. Brotcorne-Lannoye, A., and Maenhaut-Michel, G. (1986) Proc. Natl. Acad. Sci.
U. S. A. 83, 3904 –3908
43. Friedberg, E. C., Walker, G. C., and Siede, W. (1995) DNA Repair and Mutagenesis, pp. 467– 470, American Society for Microbiology, Washington,
D. C.
44. Petruska, J., Sowers, L. C., and Goodman, M. F. (1986) Proc. Natl. Acad. Sci.
U. S. A. 83, 1559 –1562
45. Ellenberger, T., and Silvian, L. F. (2001) Nat. Struct. Biol. 8, 827– 828
46. Lenne-Samuel, N., Wagner, J., Etienne, H., and Fuchs, R. P. (2002) EMBO
Rep. 3, 45– 49
47. Washington, M. T., Johnson, R. E., Prakash, L., and Prakash, S. (2002) Proc.
Natl. Acad. Sci. U. S. A. 99, 1910 –1914
48. Fernandez De Henestrosa, A. R., Ogi, T., Aoyagi, S., Chafin, D., Hayes, J. J.,
Ohmori, H., and Woodgate, R. (2000) Mol. Microbiol. 35, 1560 –1572
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
Goodman, M. F. (2000) Nature 404, 1014 –1018
8. Rangarajan, S., Woodgate, R., and Goodman, M. F. (1999) Proc. Natl. Acad.
Sci. U. S. A. 96, 9224 –9229
9. Rangarajan, S., Woodgate, R., and Goodman, M. F. (2002) Mol. Microbiol. 43,
617– 628
10. Sommer, S., Bailone, A., and Devoret, R. (1993) Mol. Microbiol. 10, 963–971
11. Kim, S. R., Maenhaut-Michel, G., Yamada, M., Yamamoto, Y., Matsui, K.,
Sofuni, T., Nohmi, T., and Ohmori, H. (1997) Proc. Natl. Acad. Sci. U. S. A.
94, 13792–13797
12. Wagner, J., and Nohmi, T. (2000) J. Bacteriol. 182, 4587– 4595
13. Napolitano, R., Janel-Bintz, R., Wagner, J., and Fuchs, R. P. (2000) EMBO J.
19, 6259 – 6265
14. Shen, X., Sayer, J. M., Kroth, H., Ponten, I., O’Donnell, M., Woodgate, R.,
Jerina, D. M., and Goodman, M. F. (2002) J. Biol. Chem. 277, 5265–5274
15. Kim, S. R., Matsui, K., Yamada, M., Gruz, P., and Nohmi, T. (2001) Mol. Genet.
Genomics 266, 207–215
16. Qiu, Z., and Goodman, M. F. (1997) J. Biol. Chem. 272, 8611– 8617
17. Woodgate, R., and Ennis, D. G. (1991) Mol. Gen. Genet. 229, 10 –16
18. Cox, M. M., Goodman, M. F., Kreuzer, K. N., Sherattt, D. J., Sandler, S. J., and
Marians, K. J. (2000) Nature 404, 37– 41
19. Bloom, L. B., Chen, X., Kuchnir Fygenson, D., Turner, J., O’Donnell, M., and
Goodman, M. F. (1997) J. Biol. Chem. 272, 27919 –27930
20. Wagner, J., Fujii, S., Gruz, P., Nohmi, T., and Fuchs, R. P. (2000) EMBO Rep.
1, 484 – 488
21. Wagner, J., Gruz, P., Kim, S. R., Yamada, M., Matsui, K., Fuchs, R. P., and
Nohmi, T. (1999) Mol. Cell 4, 281–286
22. Foster, P. L. (2000) Cold Spring Harbor Symp. Quant. Biol. 65, 21–29
23. McKenzie, G. J., Lee, P. L., Lombardo, M. J., Hastings, P. J., and Rosenberg,
S. M. (2001) Mol. Cell 7, 571–579
24. Yeiser, B., Pepper, E. D., Goodman, M. F., and Finkel, S. E. (2002) Proc. Natl.
Acad. Sci. U. S. A. 99, 8737– 8741
25. Ling, H., Boudsocq, F., Woodgate, R., and Yang, W. (2001) Cell 107, 91–102
26. Kokoska, R. J., Bebenek, K., Boudsocq, F., Woodgate, R., and Kunkel, T. A.
(2002) J. Biol. Chem. 277, 19633–19638
27. Efrati, E., Tocco, G., Eritja, R., Wilson, S. H., and Goodman, M. F. (1997)
J. Biol. Chem. 272, 2559 –2569
28. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A
Laboratory Manual, 2 Ed., pp. 4.21– 4.32, Cold Spring Harbor Laboratory
34207
DNA: REPLICATION REPAIR AND
RECOMBINATION:
Fidelity of Escherichia coli DNA
Polymerase IV: PREFERENTIAL
GENERATION OF SMALL DELETION
MUTATIONS BY dNTP-STABILIZED
MISALIGNMENT
J. Biol. Chem. 2002, 277:34198-34207.
doi: 10.1074/jbc.M204826200 originally published online July 3, 2002
Access the most updated version of this article at doi: 10.1074/jbc.M204826200
Find articles, minireviews, Reflections and Classics on similar topics on the JBC Affinity Sites.
Alerts:
• When this article is cited
• When a correction for this article is posted
Click here to choose from all of JBC's e-mail alerts
This article cites 46 references, 24 of which can be accessed free at
http://www.jbc.org/content/277/37/34198.full.html#ref-list-1
Downloaded from http://www.jbc.org/ at Rockefeller University Library on August 2, 2015
Sawami Kobayashi, Michael R. Valentine,
Phuong Pham, Mike O'Donnell and Myron F.
Goodman