ABSTRACT
The biological activity of TA*, the major photoproduct of thymidylyl-(3
'
,5
'
)-deoxyadenosine, has remained speculative since it was identified a decade
ago. To determine the mutagenicity of TA* in
Escherichia coli
, we constructed the replicative form of an M13mp18-derived phage containing TA* in the (-)-strand by polymerase-catalyzed elongation of a TA*-containing 49mer opposite a uracil-containing (+)-strand of the phage. The
in vitro
synthesis mixture was transfected into an
ung
+
, phr
-
E.coli
host and the progeny were screened with a hybridization probe unique for the (-)-strand. TA* was found to block DNA replication substantially in the
absence of SOS, but under SOS, TA* was bypassed more efficiently and was highly
mutagenic. Among 56 analyzed (-)-strand progeny from two transfections, 46 (82%) were mutants,
including six (11%) tandem mutants. The most abundant mutation was a 3
'
A
-> T substitution (31/46, 56%). The possible biological consequences of TA*
formation in the highly conserved TATA box consensus sequence on gene
expression are discussed in light of the mutagenicity of TA*.
Each gene carries not only protein sequence coding sequences, but also
regulatory elements for its expression, such as promoters, enhancers and
silencers. Mutations in the coding region could lead to a mutant protein with
altered or defective properties, whereas mutations in the promoter region could
affect proper gene regulation. The TATA box sequence of promoters is highly
conserved (
1
,
2
) and critical for gene expression in many cases, though the effect of a
particular mutation is gene-dependent (
3
-
7
). In 1983, Davies and co-workers discovered that 254 nm irradiation of TpdA produced a product
d(TpA)* (
8
) that was proposed to be the [2+2] cycloaddition adduct between the 5,6 double
bonds of the T and the A (
9
,
10
) (Fig.
1
, structure 1a). We have recently carried out further spectroscopic studies
which indicate that TA* photoproduct is actually a valence isomer of the [2+2]
adduct, (Fig.
1
, structure 2a) (
11
) which explains why TA* is not photoreversed by 254 nm irradiation (
1
). This photoproduct is also produced by 254 nm irradiation of DNA (hereafter
referred to as TA*, structure 2b) with a quantum yield of 10-100 less than dipyrimidine products (
12
). Despite its low yield, its discoverers noted that the TATA sequence of
promoters has four potential sites for TA* photoproduct formation that could
lead to mutations that permanently interfere with expression of the downstream
gene (
8
,
12
). The mutagenic properties of the TA* photoproduct have remained unknown,
however, primarily because of a lack of methods for preparing pure, well
characterized TA*-containing substrates for
in vitro
and
in vivo
replication studies.
Recently, we isolated two TA* photoproduct-containing octamers produced by 254 nm irradiation of d(GTATTATG), and
were able to ligate one of these two, d(GTATTA*TG), into a 49mer (
13
). We found that the TA* product could be bypassed by both exo
-
KF and an exo
-
T7 DNA polymerase (Sequenase Version 2.0), but we were unable to determine if
the bypass reactions were mutagenic. Herein, we report that bypass of TA* in
the (-)-strand of the replicative form of an M13mp18-derived vector in a repair-deficient
E.coli
host under SOS conditions leads predominantly to TA* -> TT mutations. The implications of the mutagenicity of TA* for
transcription involving TATA sequences is also discussed.
The TA*-containing 49mer was prepared as previously described (
13
) by ligation of a TA*-containing octamer to two other oligonucleotides in the presence of an
overlapping 35mer scaffold. The M13 clone XZ1 was prepared by Kunkel's method
for site-directed mutagenesis (
14
) of (+)-CS1 (
15
) with the 35mer shown in Figure
2
. Uracil substituted (+)-XZ1 was isolated from
dut
-
ung
-
E.coli
CJ236. The details of the procedures used to obtain the TA* mutation spectrum
were the same as previously described for the cis-syn dimer of TT and TU (
16
) with some modifications as noted. Briefly, 4 pmol 5'-phosphorylated 49-TA*-mer was used to prime (-)-strand synthesis by T4 polymerase and dNTPs
opposite 1 pmol uracil-containing (+)-XZ1 in the presence of T4 ligase and ATP for 10 min at 0oC, 10 min at 10oC and 2 h at 37oC. Subsequent treatment with uracil glycosylase was
omitted, and the reaction was allowed to proceed overnight incubation at 17oC. One third of the reaction mixture was used to transfect competent SOS
+
and SOS
-
CSRO6F'
E.coli
according to standard procedures (
17
). SOS-induced cells were prepared by 254 nm irradiation at 4oC followed by incubation at 37oC in LB for 30 min prior to being made competent in the same way
as the non-SOS induced cells. The transfected cells were mixed with top agar
containing log-phase DH5[alpha]F' cells and plated on LB plates. Plaques were then
transferred to a master plate and to replica plates that were covered with
nitrocellulose filters. The nitrocellulose filters were probed by hybridization
overnight with either 19-CC-mer or 19-TA-mer at 37oC and selected plaques were sequenced by the
dideoxy method. The chi-square ([chi]
2
) test was used to evaluate the significance of differences in mutation spectra
by sorting mutations into three categories, TA* -> TT, TA and other, for which the expected values were >= 5 which avoids significant errors in the analysis.
The mutation spectra of TA* were obtained by a general method previously
described by us for obtaining the mutation spectra of the cis-syn dimer of TU and TT (
16
). Our method is based on Kunkel's method for site-directed mutagenesis (
14
,
18
) and involves constructing a heteroduplex containing a small fraction of the
thymines replaced by uracils in the (+)-strand and a site specific photoproduct in the (-)-strand (Fig.
2
). The idea is that, when transfected into a uracil glycosylase active (
ung
+
), photoproduct repair deficient (
phr
-
,
uvr
-
)
E.coli
host, the (+)-strand will be rapidly and selectively degraded, thereby favoring
replication of the photoproduct-containing (-)-strand. To distinguish progeny of the (-)-strand from progeny of the (+)-strand, a double mismatch is introduced
into the RF phage, either adjacent to, or opposite the photoproduct.
Originally, we had hoped that a single double mismatch could be introduced at
some distance from the photoproduct site so as not to alter the sequence and
hence structure of the photodamaged DNA. In experiments with other
photoproducts, however, we have found that a small amount of double mismatch
correction occurs in the (+)-strand which diminishes the apparent mutagenicity of the photoproduct (
15
). It appears that the repair of the double mismatch results from nick
translation synthesis initiated at sites of Us, which would explain why the
correction is confined to the (+)-strand when mismatch correction would have directed repair to the (-)-strand (
19
). As a result, we now know that the double mismatch must be incorporated
opposite the photoproduct to obtain an accurate estimation of the mutagenicity
of a photoproduct, as we had originally done in obtaining the mutation spectra
of the cis-syn dimers of TT and TU (
16
).
To produce a heteroduplex M13 phage containing a GG double mismatch opposite
TA*, bacteriophage XZ1 was prepared by oligonucleotide-directed mutagenesis (
14
,
18
) of CS1 (
15
), which is complementary to TA*-49-mer, except for a double mismatch 12 nucleotides to the 3'-side of TA* (Fig.
2
). Uracil-containing (+)-XZ1 was then isolated from transfected
dut
-
ung
-
E.coli
CJ236 and used as a (+)-strand template for primer extension of TA*-49-mer by T4 DNA polymerase to form the TA*-containing (-)-strand of the bacteriophage DNA following
ligation with T4 DNA ligase and ATP.
Table 1
Progeny of the (+)-strand resulting from transfection into photoproduct repair deficient (
phr
-1,
uvr
A6)
E.coli
strain CSRO6F' (
20
) were detected by hybridization with 19-CC-mer, whereas progeny resulting from non-mutagenic bypass of the TA* product were detected by
hybridization with 19-TA-mer. In the absence of SOS induction, the proportion of progeny of
the TA*-containing (-)-strand was only 3/250 or 1% of the randomly selected
bacteriophage plaques, indicating that the photoproduct was a substantial block
to replication (Table
1
). The three (-)-strand progeny were sequenced and found to have the wild-type sequence. In the SOS-induced cells, the proportion of the (-)-strand progeny increased to an average of
8.5% for two independent transfections (55/650). When non-uracil-containing (+)-XZ1 was used to construct the TA*-containing phage, no progeny of the photoproduct-containing (-)-strand were detected out of 300
plaques that were probed. While the 19-CC-mer could distinguish (+)- and (-)-strand progeny successfully, the 19-TA-mer could not distinguish between
wild type and mutant (-)-strand progeny and all (-)-strand progeny were sequenced. From a total of 55
sequenced (-)-strand progeny from two independent transfection experiments, 46
(82%) were mutants and of these 42 (91%) were substitution mutants targeted
opposite the TA* site. Of the targeted mutations, 35/46 (76%) were single base
substitutions opposite the A of TA*, and only 1/46 (2%) was a single base
substitution opposite the T of TA*. The frequency of tandem mutations was also
rather high and constituted 6/46 (13%) of the targeted mutations. The most
frequent targeted mutation (31/46 or 67%) was the substitution of A for T
opposite the A of TA*. Non-targeted substitution and deletion mutations were also observed nearby the
photoproduct site with a low frequency (4/55, 9%).
The mutation spectra for the two independent transfections were not
significantly different (
P
< 0.25), but because the two transfections were carried out a month apart we
considered the possibility that the apparent difference was due to partial
decomposition of TA*. In our original preparation of the 49-TA*-mer, 14% of the sample contained degradation products corresponding
to strand cleavage in the vicinity of the TA* (
16
). In a second preparation, in which care was taken not to unnecessarily heat
the sample, only 3% of the degradation products were observed. To further
investigate the stability of TA*, we carried out an HPLC analysis of TA*-containing octamer, 8-TA*-mer, under a variety of conditions. The TA*-containing octamer was found not to decompose to any
products that were separable by C-18 HPLC when kept frozen in water at -20oC for 2 months. After one week at room temperature ~12% of a product with a longer retention time was
observed, which could be produced in ~40% yield after heating for only 20 min at 100oC. With regard to possible decomposition of the samples used to obtain
the mutation spectra, we note that the first transfection was carried out with
TA*-containing bacteriophage that had been synthesized the day before with the
second preparation of the 49-TA*-mer. This 49mer had been constructed ~3 months previously from the 8-TA*-mer and had been stored at -20oC. The second transfection was carried
out with the same 49-TA*-mer 1 month later with a new batch of competent cells and TA*-containing bacteriophage prepared two days earlier. The only
times that the 8-TA*-mer used in the preparation of the TA*-49-mer was allowed to warm above -20oC was when it was incorporated into the
49mer by ligation for 10 h from 0-16oC followed by purification at room temperature for 2 h by gel
electrophoresis and elution from the gel at 4oC overnight. The 49-TA*-mer was again warmed for 1 h at 37oC for RF bacteriophage synthesis, and 1 h at 0oC for transfection followed by heat shock at 42oC for 90 s before plating. Though everything
possible was done to minimize decomposition of TA* for the
in vivo
mutation studies, it is possible that some small fraction (<5%) of TA* decomposed to at least one other product. Because most biological
systems of interest are normally at 37oC, we would also expect that TA* product produced
in vivo
would also decompose to some small extent before DNA synthesis bypass.
In non-SOS cells, TA* appeared to block replication as judged by the very low
proportion (1%) of progeny from the photoproduct-containing (-)-strand relative to those from the uracil-containing (+)-strand. The ability of TA* to block normal
replication in
E.coli
is similar to what has been observed for the cis-syn, (6-4) and Dewar photoproducts of TT (
21
,
22
), which are also dinucleotide photoproducts. One notable exception is the trans-syn dimer of TT which has been reported to be bypassed to a significant
extent under non-SOS conditions in one sequence context (
23
) but not in another (
15
). Photodimerization of adjacent bases fixes their relative orientation and
restricts the conformational flexibility of the entire dinucleotide subunit,
which would explain why bypass of dinucleotide photoproducts is inhibited. In
SOS cells, however, the proportion of (-)-strand progeny increased to 8.5%, suggesting that TA* is more
easily bypassed than in the non-SOS cells, again consistent with what has been observed for the other DNA
photoproducts, and what is known about the SOS response (
24
,
25
).
In contrast to the observation that none of the three progeny of the TA*-containing-strand in non-SOS cells were mutants, 82% of the TA* progeny in SOS cells
were mutants (Table
2
). The majority of these mutants (78%) were single base substitutions and the
rest were tandem mutations (13%), one near targeted substitution (2%) and three
29 bp deletions coupled to a substitution (7%). One possible mechanism for the
origin of the targeted single and tandem substitution mutations is one in which
nucleotides are sequentially incorporated opposite the TA* product (Fig.
3
). To account for the observed mutations, nucleotides would have to be first
incorporated opposite the 3'-A of TA* with an overall selectivity of A >> T > G > C. In the
second step, nucleotides would have to be incorporated opposite the 5'-T of TA* with an overall selectivity of A >> G > T > C. We use the
term overall selectivity to refer to the net result of a competition between
nucleotide incorporation, 3' -> 5' exonucleolytic cleavage (proofreading), and subsequent
nucleotide incorporation and proofreading steps opposite and completely past
the photoproduct site. The origin of the near targeted substitution and 29 bp
deletion coupled with a substitution is not understood at this time.
TA* is a minor photoproduct of DNA, having a quantum yield for induction by 254
nm that is ~0-100 times less than for the cis-syn and (6-4) products respectively (
12
,
32
). When the frequency of TA sites were taken into account, it was estimated that
10 TA* products would be produced in the
E.coli
K12 genome following exposure to one mean lethal fluence of 254 nm (F
37
= 50 Jm
-2
) (
12
). Even a minor photoproduct, however, can have a significant biological effect
if it is highly mutagenic, not repaired rapidly, and there is an abundant
target site which is sensitive to the photoproduct-induced mutations. Davies and co-workers recognized that one abundant and possibly important
biological target for TA* formation is the highly conserved TATA sequence of
promoters which has four potential photoproduct sites (
8
,
12
). We now know that TA* in the sequence context studied is highly mutagenic and
causes primarily TA* -> TT mutations. The mutagenicity of TA* is comparable to that of the (6-4) product of TT (57-85% TT -> TC) (
15
,
21
) and the cis-syn dimer of TU, the deamination product of the cis-syn dimer of TC (96% TC -> TT) (
16
). If TA* -> TT is also the major mutation in a TATA sequence, 254 nm radiation would
induce TATA -> TTTA, TATT, TAAA and AATA mutations by way of the four potential TA*
sites. In a study of the P22 promoter activity for the antirepressor gene, all
the possible T[middot]A -> A[middot]T transversions in TATA resulted in a reduction of
promoter activity for
E.coli
RNA polymerase
in vitro
, and was most severe for transversion mutations at the first two sites (
33
). It was also found that all of these transversion were induced by UV
irradiation of P22 followed by transfection into
S.typhimurium
strain DB7168 which is repair defective and carries the pKM101 plasmid that
enables error-prone bypass of UV damage. In a forward mutation assay of 254 nm
irradiated single-strand M13mp2, however, no TA -> TT mutation was observed in the -10 TATG sequence of the
lac
promoter (
34
). Only TA -> CA, 2 TG, TC and AC mutations were observed, the first three of which were
only minor mutations in our system, and the last of which was not observed at
all. Why the TA -> TT mutation was not detected in this system is unknown at the moment, and
will have to await further studies of TA* mutagenesis.
This work was supported by NIH Grant R37 CA40463. We also thank Jeremy Davies
for helpful discussions.
Cell
(+)-strand
(-)-strand progeny/
(-) mutants/
(+) mutants/
no. probed (%)
no. sequenced
no. sequenced
1
SOS+
uracil-containing
24/250 (10%)
22/24
0/9
2
SOS+
uracil-containing
31/400 (8%)
23/31
1/27
3
SOS-
uracil-containing
3/250 (1%)
0/3
0/15
4
SOS+
non-uracil containing
0/300 (0%)
-
-
REFERENCES
Return