Replication bypass and mutagenic effect of
[alpha]-deoxyadenosine site-specifically incorporated into single-stranded vectors
Replication bypass and mutagenic effect of [alpha]-deoxyadenosine site-specifically incorporated into single-stranded vectors
Hironori
Shimizu
,
Ryohei
Yagi
,
Yoshiharu
Kimura
,
Keisuke
Makino
,
Hiroaki
Terato
1
,
Yoshihiko
Ohyama
1
and
Hiroshi
Ide
1,
*
Department of Polymer Science and Engineering, Kyoto Institute of Technology,
Matsugasaki,
Sakyo-ku
, Kyoto 606,
Japan
and
1
Graduate Department of Gene Science, Faculty of Science, Hiroshima University,
Kagamiyama
, Higashi-Hiroshima 739,
Japan
Received October 1, 1996
;
Revised and Accepted December 2, 1996
ABSTRACT
[alpha]
-2
'
-Deoxyadenosine (
[alpha]
) is a major adenine lesion produced by
[gamma]
-ray irradiation of DNA under anoxic conditions. In this study, single-stranded recombinant M13 vectors containing
[alpha]
were constructed and transfected into
Escherichia coli
to assess lethal and mutagenic effects of this lesion. The data for
[alpha]
were further compared with those obtained with M13 vectors containing normal A
or a model abasic site (F) at the same site. The transfection assay revealed
that
[alpha]
constituted a moderate block to DNA replication. The
in vivo
replication capacity to pass through
[alpha]
was
~
20% relative to normal A, but 20-fold higher than that of F constituting an almost absolute replication
block. Similar data were obtained by
in vitro
replication of oligonucleotide templates containing
[alpha]
or F by
E.coli
DNA polymerase I. The mutagenic consequence of replicating M13 DNA containing
[alpha]
was analyzed by direct DNA sequencing of progeny phage. Mutagenesis was totally
targeted at the site of
[alpha]
introduced into the vector. Mutation was exclusively a single nucleotide
deletion and no base substitutions were detected. The deletion frequency
associated
[alpha]
was dependent on the 3
'
-nearest neighbor base: with the 3
'
-nearest neighbor base T mutation (deletion) frequency was 26%, whereas 1%
with the 3
'
-nearest neighbor base G. A possible mechanism of the single nucleotide
deletion associated with
[alpha] is discussed on the basis of the misinsertion-strand slippage model.
INTRODUCTION
Cellular DNA is continuously exposed to endogenous and exogenous genotoxic
agents that generate a wide variety of structural defects in DNA. These
structural defects are restored in cells by multiple pathways such as base or
nucleotide excision repair pathways (
1
). However, recent evidence shows that lesions present in DNA are not repaired
at an equal rate, but those in transcribed strands in expressed genes are
preferentially repaired (for review, see ref.
2
). This raises the possibility that the DNA replication fork encounters DNA
lesions before they are restored. In this case, DNA replication could be either
aborted due to the lesions or proceed through the sites with a risk of
mutation. Accumulated data indicate that the response of DNA polymerases to the
encountered lesion is not unique and how they cope with it depends on the
structure of the lesion, sequence contexts flanking to the lesion, enzymatic
properties of DNA polymerases and accessory proteins (
3
,
4
).
We have been focusing our attention on the structural factors of DNA lesions
that perturb hydrogen bonding and base stacking interactions, showing that
alteration of these interactions exerts differential effects on DNA replication
(
5
-
7
). In a series of these studies, we have recently shown by
in vitro
experiments that [alpha]-2'-deoxyadenosine ([alpha]) site-specifically introduced into
oligonucleotide templates transiently inhibits DNA synthesis (
8
). The data also suggest that [alpha] is potentially mutagenic. [alpha] was originally shown to be produced in [gamma]-irradiation of aqueous deoxyadenosine under anoxic
conditions (
9
) and later this product was found in DNA, poly(dA[middot]dT) and poly(dA) irradiated under similar conditions (
10
). [alpha] has the following unique structural feature. The N-glycosidic bond linking adenine and deoxyribose moieties is flipped
due to the abstraction of the H1' atom by OH radicals, but the base moiety is totally intact. Although
basal and induced levels of [alpha] in cellular DNA are not known, several lines of evidence imply that this
lesion, in particular, or lesions with the [alpha] configuration with respect to the N-glycosidic bond might, in general, have some biological relevance.
The cell nucleus is a very poorly oxygenated intracellular compartment (
11
,
12
), and cell hypoxia is ubiquitously observed in tumor cells and anaerobe. In
addition,
Escherichia coli
and yeast cells have repair enzymes such as endonuclease IV (
13
) and Apn (Ide
et al
., unpublished data) that recognize this lesion. The nucleoside of 5-formyluracil, resulting from oxidative damage of thymine, readily undergoes base-catalyzed anomerization with respect to the N-glycosidic bond (
14
).
Molecular mechanics and thermodynamic studies on the duplex DNA containing this
lesion have revealed that [alpha] paired with pyrimidines generates little distortions in DNA, whereas
purines introduce distinct kinks and budges (
15
). These studies also suggest that the space created in the major groove by the
flipped base is key to the specific recognition by endonuclease IV, a repair
enzyme from
E.coli
(
13
).
In this study, we extended our investigations on [alpha] to an
in vivo
system. Single-stranded M13 vectors containing [alpha] at the defined position were constructed and transfected into
E.coli
. Biological consequences such as lethal and mutagenic effects resulting from
this lesion were assessed based on the plaque forming efficiency of
transfecting DNA and direct DNA sequencing of progeny phage. The data were
further compared with those obtained in parallel experiments for normal A and a
model abasic site (F) introduced into the same site of M13 vectors.
MATERIALS AND METHODS
Materials
Four 2'-deoxyribonucleoside 5'-triphosphates (dNTPs, purity >99.5%) and ATP (purity
>95%) were purchased from Takara and Boehringer Mannheim, respectively. [[alpha]-
32
P]dATP (110 TBq/mmol) and [[gamma]-
32
P]ATP (220 TBq/mmol) were from Amersham.
E.coli
DNA polymerase I Klenow fragment (Pol I), T4 DNA ligase and a restriction
enzyme
Bam
HI were obtained from New England Biolabs. T4 polynucleotide kinase was from
Toyobo. Oligonucleotides used in this study are listed in Figure
1
. Oligonucleotides except 32temps A, G[alpha], T[alpha] and F were synthesized by the standard phosphoramidite chemistry
and extensively purified by reversed phase HPLC and polyacrylamide gel
electrophoresis if necessary, and desalted as described (
16
,
17
). Characterization of DNA lesions ([alpha] or F) introduced into oligonucleotides has been published previously (
13
). 32mer templates (32tempA, G[alpha], T[alpha] and F) were constructed by ligation of an appropriate 17mer (17A,
17G[alpha], 17T[alpha] or 17F) and a 5'-phosphorylated 15mer (15linker) using T4 DNA ligase
and 57scaffold(+) (
18
). The constructed 32mer templates were purified by reversed phase HPLC and
desalted (
17
). Concentrations of oligonucleotides were determined based on UV absorption at
260 nm and calculated molecular absorption coefficients (
19
).
Construction of single-stranded M13 vectors containing
[alpha]
or F
Single-stranded M13 vectors containing a unique lesion were constructed basically
following the reported methods (
18
,
20
,
21
). Single-stranded closed circular M13mp18 DNA was first linearized. M13mp18 DNA (40
[mu]g) and 16
Bam
HI(+) oligonucleotide (108 pmol, ~6-fold excess over M13 DNA), which is complementary to the franking
regions of a unique
Bam
HI site present in the multiple cloning site, were heated in 100 [mu]l 150 mM NaCl, 10 mM Tris-HCl (pH 7.9), 10 mM MgCl
2
, 1 mM DTT (total volume 100 [mu]l) at 70oC for 5 min and slowly cooled to room temperature. The mixture was
incubated with 100 U
Bam
HI at 37oC for 1 h, then 100 U
Bam
HI was further added to the mixture and incubation was continued for another
hour. The reaction was terminated by adding EDTA (final concentration 50 mM).
Completion of the restriction digestion was confirmed by 1% agarose gel
electrophoresis. To remove 16
Bam
HI(+) and digested oligonucleotides, the reaction mixture was subjected to
alkaline gel permeation chromatography on a Sephadex G-100 (
22
). Fractions containing M13 DNA were determined by ethidium bromide staining,
neutralized and DNA precipitated by ethanol. Linearized M13 DNA was stored in
10 mM Tris-HCl (pH 7.5).
Recombinant M13 DNA containing a 17mer insert was constructed as follows.
Linearized M13 DNA (10 [mu]g), a 5'-phosphorylated 17mer (17A, 17G[alpha], 17T[alpha], 17F, 13.6 pmol, 3-fold excess over M13 DNA), 57scaffold(+)
(13.6 pmol) and 16
Bam
HI(-) (22.7 pmol) in 500 [mu]l 50 mM Tris-HCl (pH 7.8), 10 mM MgCl
2
, 1 mM DTT, 1 mM ATP and 50 [mu]g/ml BSA were heated at 75oC for 15 min and slowly cooled to room temperature. 16
Bam
HI(-) complementary to 16
Bam
HI(+) was included in the reaction mixture to avoid annealing of trace amounts
of remaining 16
Bam
HI(+) to linearized M13 DNA, which would generate original circular M13 DNA
without any insert by subsequent ligation. Ligation reaction mixture was
incubated with T4 DNA ligase (12000 U) at 14oC for 12 h. DNA was precipitated by ethanol, resuspended in 10 mM Tris-HCl (pH 7.5) with a final concentration 200 ng/[mu]l, and used for transfection. An aliquot of the sample was
analyzed by 1% agarose gel electrophoresis. Gel was stained by ethidium
bromide, photographed and ligation efficiency and concentration of DNA were
determined by densitometric scanning of the picture (see below). Recombinant
M13 vectors containing 17mer inserts (17A, 17G[alpha], 17T[alpha] or 17F) are designated as M1317A, M1317G[alpha], M1317T[alpha] and M1317F, respectively.
Transfection assay
Recombinant M13 vector (1 [mu]g) and 57scaffold(-) (6.8 pmol) in 6 [mu]l Tris-HCl (pH 7.5) were heated at 90oC for 3 min and quickly cooled on ice. DNA sequencing
analysis of progeny phage revealed that, without the oligonucleotide
complementary to the 57scaffold(+), a significant portion of progeny phage
obtained by transfection contained a 17mer insert derived from 57scaffold(+).
This could be readily identified by a G:T mismatch specifically introduced at
the 5' side of the lesion as a genetic marker [see the sequences of the 17mer
inserts and 57scaffold(+) in Fig.
1
]. Presumably the 17mer gap region in the bridging scaffold was filled
in vivo
and DNA containing this insert was replicated. The progeny phage containing the
17mer insert from 57scaffold(+) was completely eliminated by the addition of
complementary 57scaffold(-) in the denaturing step before transfection. Electrocompetent JM101 ([Delta]
lacpro, supE, thi, F
'
traD36, proAB, laqI
q
Z
[Delta]
M15
) prepared by following the instruction of BioRad (0.4 ml) and a recombinant M13
vector (200 ng) were gently mixed and transferred into an electroporation
cuvette with a 0.1 cm electrode gap. Electroporation was performed with a time
constant ~4.5 [mu]s and at 1.8 kV using a BioRad
E.coli
pulser. After appropriate dilution, the electroporation mixture was mixed with an 1 ml SOC media, 10 [mu]l 100 mM IPTG, 50 [mu]l 2% X-gal, 0.2 ml freshly cultured host JM101 and 3 ml 2* YT top agar, and plated on an L-Broth plate. The plate was incubated overnight at 37oC. Colorless plaques were counted to measure
in vivo
DNA replication capacity to pass through the introduced lesions.
Mutation analysis
For each lesion, progeny phage were randomly isolated from total populations of
colorless plaques obtained from four independent transfections. Nucleotide
sequences were determined by the dideoxy method (
23
) using the 17primer, complementary to positions 25-41 downstream of the
Hin
dIII site, and a Sequenase version 2.0 DNA sequencing kit (USB).
In vitro
DNA polymerase reaction
5'-
32
P labeled 15primer (0.2 pmol) and an appropriate 32temp (1.5 pmol, template excess to ensure complete annealing of the primer) were
heated in 10 [mu]l 20 mM Tris-HCl (pH 7.5) and 50 mM NaCl, and slowly cooled to room temperature.
Pol I reaction buffer (final composition 10 mM Tris-HCl pH 7.5, 5 mM MgCl
2
, 7.5 mM DTT) and four dNTPs (final concentrations 50 [mu]M) were added to the reaction mixture (final volume 19 [mu]l) and preincubated 5 min at 37oC. The replication reaction was initiated by adding 0.1 U Pol I (1 [mu]l) and incubation was continued at 37oC for 5-30 min. The reaction was terminated by the addition
of loading buffer consisting of 98% formamide, 0.05% bromophenol blue, 0.05%
xylene cyanol and 20 mM EDTA. Reaction products were analyzed by 16%
polyacrylamide gel electrophoresis under denaturing conditions (8 M urea).
After electrophoresis, the gel was dried and subjected to autoradiography using
a Fuji RX film with varying exposure times to ensure linear correlation between
radioactivity and band intensity in the autoradiogram. Reaction products
obtained with 32tempG[alpha] and T[alpha] were isolated from gels as described (
8
), and full length (32mer) and one nucleotide shorter (31mer) products were
separately subjected to Maxam-Gilbert sequencing reactions (
24
).
Densitometry
Images of autoradiograms and agarose gel pictures were recorded on an Epson
GT6000 scanner using the Color Magician III software. Intensity of bands in the
recorded images was analyzed by the NIH Image (ver 1.55) using a built-in gel scan macro.
RESULTS
In vitro
replication bypass of
[alpha]
In the previous study, we have shown that [alpha] constitutes a transient block to DNA replication catalyzed by Pol I
in vitro
(
8
). However,
in vitro
replication efficiency of Pol I to pass through this lesion has not been
assessed on the quantitative basis. Thus, time course of the replication of
oligonucleotide templates containing normal A, [alpha] and a model abasic site (F) incorporated at the same position were
followed and the relative replication efficiencies were compared. For this
purpose, 32temps were primed by 15primer (Fig.
1
) and replicated by Pol I as described in Materials and Methods. Reaction
products were analyzed by 16% denaturing polyacrylamide gel electrophoresis and
the relative band intensities of the reaction products were determined by
densitometry of autoradiograms. The position of the band was determined by
comparison with standard dideoxy sequencing ladders prepared using an M1317A
DNA template containing the same sequence as 32tempA.
Figure
2
shows a typical autoradiogram obtained by 30 min incubation. Replication of the
template was completely arrested at F (strong band) or one nucleotide prior to
the lesion (weak band) so that bypassed products were not detected within the
present detection limit (Fig.
2
, lane 6). In contrast to F, significant fractions of the template containing [alpha] were replicated to full length (32mer) or one nucleotide shorter
products (31mer) at the same incubation time (Fig.
2
, lanes 7 and 8). Both full length and one nucleotide shorter products were also
produced in the replication of 32tempA containing no lesions (Fig.
2
, lane 5). Interestingly, the band intensity of a stalled product appearing one
nucleotide prior to [alpha] was different for 32tempG[alpha] and T[alpha], indicating a nearest neighbor effect on the replication
capacity of Pol I to pass through [alpha]. In Maxam-Gilbert sequencing analysis of the reaction products obtained with
32tempG[alpha] and T[alpha], bands implying incorporation of dTMP, dCMP and dAMP opposite [alpha] were present for both full length and one nucleotide
shorter products (data not shown). Similar results were obtained in the
previous study (
8
). Thus, one nucleotide shorter products commonly observed for 32tempA, G[alpha] and T[alpha] are likely generated by premature dissociation of Pol I at the
end of DNA templates. Figure
3
shows the time course of the replication of 32temps determined by the
densitometry. Whether or not the template contained a lesion, the original
primer was converted to extended products within 5-10 min, suggesting that initial extension of the primer is independent of
the lesions and occurs at similar rates. Replication of control template
32tempA completed in 10 min (Fig.
3
A). According to the time course plots of the stalled intermediates, extension
of the intermediates for 32tempG[alpha] was more efficient than 32tempT[alpha] (Fig.
3
C and D). For the former, only 4% of the stalled intermediate remained at 20
min, while for the latter, 77% was left unextended at the same incubation time.
The stalled products at F were extremely resistant to subsequent elongation by
Pol I (Fig.
3
B), which is consistent with published data (
25
).
In vivo
replication bypass of
[alpha]
To evaluate the
in vivo
replication efficiency to pass through lesions, recombinant M13 vectors
containing [alpha], F or normal A at the same position were constructed and transfected
into JM101 cells. The recombinant M13 vectors contain 17mer inserts in the
Bam
HI site of the multicloning site so that progeny phages resulting from the
vectors form colorless plaques due to the (3
n
+ 2) frameshift in the lacZ([alpha]) gene. Therefore, progeny phages derived from the recombinant and
parental M13 vectors are readily distinguished by the color of plaques, the
latter of which forms blue plaques in the presence of X-gal and IPTG. In addition, the potential contamination of colorless
plaques derived from replication of the 17mer gap region of the scaffold was
ruled out by the single base mismatch serving as a genetic marker (see
Materials and Methods).
.
Replication efficiency of recombinant M13 vectors containing a single lesion [alpha] or F
a
Experiment No.
Plaque forming efficiency (%) relative to control 17A
17A
b
17G[alpha]
b
17T[alpha]
b
17F
b
1
100
10.0
10.0
1.4
2
100
22.6
18.9
2.2
3
100
36.7
18.1
0.8
4
100
20.8
16.7
0.9
Average (+-SD)
c
100
22.5 (+-11.0)
15.9 (+-4.0)
1.3 (+-0.6)
a
Average numbers of colorless plaques per 1 [mu]g of closed circular recombinant M13 DNA were 8900 (17A), 1600 (17G[alpha]), 1400 (T[alpha]) and 130 (F). Blue plaques were excluded from data since they
did not contain the 17mer insert. Blue plaques consisted of <2% of total plaques obtained with control M1317A.
b
17mer inserts introduced into the
Bam
HI site in M13 vector.
c
Average of four experiments with a standard deviation in the parenthesis.
Table
0
shows the replication efficiencies of recombinant M13 vectors containing [alpha], F or normal A. The data are based on the number of colorless plaques.
Typically, blue plaques excluded from the data consisted of <2% of total plaques obtained by the transfection of M1317A. M1317G[alpha] and M1317T[alpha] containing [alpha] were replicated with 22.5 and 15.9% efficiencies,
respectively, relative to control M1317A. These transfection data indicate that
[alpha] allows appreciable translesion bypass when DNA templates containing this
lesion are replicated in
E.coli
. In contrast, M1317F containing a model abasic site was replicated with much
lower efficiency (1.3%) than M1317G[alpha] and M1317T[alpha]. The replication efficiency obtained for M1317F is in good
agreement with those (0.3-0.6%) for single-stranded recombinant M13 DNA containing a single natural abasic
site (
21
).
Mutagenic effects of
[alpha]
The mutagenic consequences of replicating M13 DNA containing [alpha] were examined by sequencing total 201 progeny phage DNA. In control
experiments, M13 phage DNA containing a model abasic site (F) or normal A was
also transfected and progeny phage DNA was sequenced. The mutagenic spectra and
frequency associated with abasic sites have been extensively studied using
transfecting DNA (for example,
21
,
26
-
28
), thereby providing an appropriate standard in comparing mutagenic effects of [alpha] and other damages. Nucleotide sequences of all isolated phages were
analyzed for 30 nucleotides (nt) on the 3' side of the 17mer insert and >= 50 nt on the 5' side. Sequence analysis showed that except for the lesion
site, the nucleotide sequence of all isolated phage was normal. In addition,
none of DNA randomly isolated from blue plaques contained the 17mer insert and
all had the same sequence of original M13mp18 (data not shown). Potential -2 or +1 frameshift mutation that restores the reading frame and gives
rise to blue plaques and extensive deletions that could occur during the
construction of vectors and give rise to colorless plaques were practically
eliminated by these data.
Table
0
summarizes the mutation spectra and frequency determined for recombinant M13
vectors. Out of 100 progeny phages derived from M1317G[alpha], where the lesion [alpha] is flanked by G on the 3' side, 99% contained A at the site of [alpha]. The rest (1%) was a single nucleotide deletion at
the lesion. However, with M1317T[alpha], where [alpha] is flanked by 3' T, the deletion frequency markedly increased up to 26% in
total 101 phage sequenced. The rest (74%) contained A at [alpha]. In both vectors containing [alpha], the deletion was totally targeted at the lesion and no base
substitutions such as [alpha] -> G transitions or [alpha] -> T, [alpha] -> C transversions were detected. These
results indicate that template [alpha] primarily directs incorporation of T
in vivo
, which does not result in mutation since [alpha] is derived from A by the anomerization of the N-glycosidic bond. However, a single nucleotide deletion becomes a
major consequence depending on the 3' nearest neighbor base.
.
Numbers and spectrum of mutations derived from replication of DNA lesions (X = [alpha], F or A) site-specifically introduced into M13 vectors
a
Lesion (X)
Mutation
A
G[alpha]
b
T[alpha]
b
F
b
X -> A
20 (100%)
99 (99%)
75 (74%)
11 (39%)
c
X -> T
0
0
0
13 (46%)
X -> C
0
0
0
1 (4%)
X -> G
0
0
0
0
1 deletion
0
1 (1%)
26 (26%)
3 (11%)
total
d
20
100
101
28
a
Numbers in the parentheses indicate percentage of the total number of sequences
analyzed.
b
G[alpha]:[alpha] with the 3' nearest neighbor base G; T[alpha]:[alpha] with the 3' nearest neighbor base T; F:abasic
site.
c
Two of them were tandem mutation (3'-GF -> 3'-AA).
d
Total number of plaque sequenced.
In vivo
replication of M1317F containing a model abasic site resulted in the nucleotide
insertion (89%) or a single nucleotide deletion (11%). The nucleotide insertion
or deletion was targeted at the lesion except for two cases, which were tandem
changes at the sites of the lesion and the immediately flanking 3' nucleotide. The progeny phage DNA contained the following nucleotides at
the lesion: T (46%) > A (39%) > C (4%) > G (0%), indicating that nucleotides
were incorporated opposite template F in the following order of the frequency:
A > T > G > C. The preferential incorporation of A at non-instructive lesions such as abasic sites, called A-rule, has been well established
in vitro
(
25
,
29
,
30
), and
E.coli
(
21
,
26
,
27
) although there are exceptions in eukaryotic cells (
28
,
31
,
32
). Furthermore, the present results on the relative order of nucleotide
insertion frequencies as well as the nucleotide deletion frequency agree fairly
well with the those obtained by the transfection of M13 DNA containing natural
abasic sites (
21
,
27
).
DISCUSSION
In this study, single-stranded recombinant M13 vectors containing an [alpha] lesion at the defined position were constructed and transfected
into
E.coli
to assess
in vivo
replication capacity to pass through this lesion and its mutagenic potential.
The data obtained for [alpha] were further compared with those obtained with M13 vectors containing
normal A or a model abasic site (F) at the same site. Biological consequences
associated with DNA lesions are generally combined outcome of DNA replication
and repair (
1
). However, the latter contribution is virtually eliminated (or at least
minimized) in this study since [alpha] and F lesions incorporated in single-stranded DNA are poor substrates of
E.coli
endonuclease IV and exonuclease III that are responsible for the repair of
these lesions (
13
,
33
,
34
). Thus, lethal and mutagenic effects of the introduced lesions in this study
are primarily correlated with the events occurring in DNA replication.
The transfection data show that [alpha] constitutes a moderate block to DNA replication
in vivo
(Table
0
). The
in vivo
replication capacity of [alpha] was ~20% relative to normal A, but 20-fold higher than that of the abasic site that was an almost
absolute replication block. These data are consistent with the time course
study obtained by
in vitro
replication of templates containing [alpha] or F (Fig.
3
). The
in vivo
replication efficiency for [alpha] is comparable to those obtained for etheno-deoxycytidine (
35
,
36
), etheno-deoxyadenosine (
35
) and a
cis
-diamminedichloroplatinum adduct of guanine (
37
) using uninduced
E.coli
hosts. These lesions including [alpha] maintain base ring structures, hence retaining potential base stacking
interactions. Interestingly, the efficiency of replication bypass of [alpha] was higher when this lesion is flanked by the nearest neighbor base G
than T at the 3' position both
in vitro
and
in vivo
, although
in vivo
data are not totally conclusive because of the small difference in the
efficiencies and relatively large standard deviations.
The mutagenic consequences of replicating M13 DNA containing [alpha] were analyzed by direct DNA sequencing of progeny phage (Table
0
). The mutagenesis was totally targeted at the site of [alpha] introduced into the vector. Mutation was exclusively a single nucleotide
deletion and no base substitutions were detected. This result is rather
interesting and suggests some differences in the responses of
E.coli
DNA polymerases I (Pol I) and III holoenzyme (Pol III). Pol I incorporated C
and A in addition to T opposite [alpha] although the frequencies for C and A were lower than that of T (
8
). However, base substitution mutations resulting from incorporation of C or A
opposite [alpha] were not observed in the transfection of M13 vectors containing [alpha], where the replicating enzyme was Pol III. Differential
processing of template lesions by Pol I and Pol III has been implicated in
transfection studies (
36
,
38
), but the molecular mechanism of the differential processing remains to be
clarified. The frequency of the deletion was dependent on the 3' nearest neighbor base of [alpha]. With the 3' nearest neighbor base T, mutation frequency was 26%,
whereas 1% with the 3' nearest neighbor base G. These data suggest that although [alpha] primarily directs incorporation of the correct nucleotide T
during DNA replication
in vivo
, a single nucleotide deletion will be a significant event depending on the
flanking sequences. It is also noted that the mutation spectrum for [alpha] is unique. Several DNA lesions site-specifically introduced into vectors are known to generate a single
nucleotide deletion
in vivo
(
21
,
35
,
39
-
41
). However, unlike [alpha] in this study, deletions account for a minor part of the mutation
spectra (for an exception, see ref.
42
).
The mechanism of the single nucleotide deletion by [alpha] is likely to be explained by the `misinsertion-strand slippage model' (Fig.
4
) (
43
-
47
) proposed based on the original model for undamaged DNA (
48
). In a previous primer extension assay using Pol I, we have shown that
nucleotide insertion frequency opposite [alpha] was T > C >= A >> G (
8
). We have also shown by the molecular modeling that pyrimidine nucleotides (T
and C) placed opposite [alpha] fit into B form DNA without introducing any significant distortion, but
purine nucleotides (A and G) produce distinct distortions due to the steric
hindrance between [alpha] and purine bases (
15
). In addition, an [alpha]:T pair forms two stable hydrogen bondings, one of which is not canonical
and is locating in the minor groove.
T
m
measurements have also revealed that thermodynamic stability of 9mer duplexes
containing a central [alpha]:N base pair (N = A, G, C, T) is [alpha]:T > [alpha]:C [approx] [alpha]:A > [alpha]:G (
15
). The duplex containing an [alpha]:T pair is as stable as that contains a normal A:T pair at the same site.
In light of these structural and thermodynamic data, it is likely that when
encounters a template [alpha] lesion, DNA polymerase III holoenzyme (Pol III) incorporates T most
frequently opposite template [alpha], followed by C. When T is incorporated opposite [alpha], the resulting primer terminus mimics canonical one in geometry
and stability as described above. Thus, this will allow Pol III for subsequent
polymerization, but no mutations result in this case (Fig.
4
, path A). On the contrary, incorporation of C opposite [alpha] will produce a breathing primer terminus since C fits in B geometry but
can not form proper hydrogen bondings with [alpha] (
11
). Accordingly, while Pol III is stalling at this site, transient misalignment
between C at the primer terminus and template G located at 5' to [alpha] occurs, and extension of the transiently misaligned primer
terminus fixes a single nucleotide deletion (Fig.
4
, path B). Such a mechanism involving misinsertion followed by misalignment of
template-primer has been proposed to explain frame shifts observed in iterated
nucleotide sequences (
44
,
48
) and a single nucleotide deletion at DNA lesions such as abasic sites (
46
,
49
) or other DNA adducts (
45
-
47
).
ACKNOWLEDGEMENT
This work was supported by grants from the Ministry of Education, Science and
Culture, Japan (H.I.).
REFERENCES
1 Friedberg,E.C. (1995) DNA Repair and Mutagenesis, ASM Press, Washington, DC.
