INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Escherichia coli has evolved several strategies to cope with DNA damage, including a primitive form of cell cycle checkpoint control, nucleotide and base excision repair, UmuD'₂C/RecA-assisted translesion synthesis, and damage avoidance-tolerance mechanisms such as recombination repair (6, 24). DNA damage induces SOS functions and halts progression through the cell cycle, providing time for DNA repair. However, some damage may escape repair and new damage may be introduced during DNA replication. These unrepaired lesions are believed to be the major sources of induced mutations (6).

Many DNA lesions block DNA synthesis, and translesion DNA synthesis (TLS) is often associated with replication errors. In E. coli, TLS, catalyzed by inducible SOS DNA polymerases (7, 10, 42), and DNA damage tolerance mechanisms (6) operate to overcome this block. RecA is activated upon DNA damage and facilitates autodigestion of LexA, a repressor of more than 20 SOS genes (6). RecA, UmuC (DNA polymerase V [pol V]), UmuD, DinB (DNA pol IV), and DNA pol II are encoded by prominent SOS genes. UmuD is converted to UmuD' by proteolytic cleavage facilitated by RecA. An SOS DNA polymerase, DNA pol V, associates with two molecules of UmuD' and catalyzes DNA synthesis across lesions (37). This synthesis is distributive, and pol V is thought to be replaced by DNA pol III after synthesis of a short stretch of DNA, during which pol V may incorporate an incorrect nucleotide opposite a lesion. Therefore, this pathway is error prone, inducing targeted point mutations.

E. coli has other mechanisms by which to overcome a DNA synthesis block. Restart of DNA synthesis downstream from the site of the block generates a gap in newly synthesized DNA. This gap subsequently may be filled with the parental strand of the sister molecule by recombination. The recA, recF, recO, recR, ruvA, ruvB, ruvC, and recG genes are believed to be involved in this process, termed daughter strand gap repair, in which the first four genes are involved in the presynapsis and synapsis stages and the remainder are involved in the postsynapsis stage (6, 12, 13). A hypothetical mechanism named template strand switching may also operate, in which the nascent strand switches its template to its sister molecule upon encountering a blocking DNA lesion and then switches back to the original template after bypassing the lesion (6). Both mechanisms are mechanistically error free.

The aim of this study was to determine the contributions of the various pathways described above to the overall response to DNA damage. Our strategy was to incorporate a single DNA adduct into double-stranded (ds) heteroduplex (HD) DNA bearing several mismatched regions. These mismatched regions serve as strand-specific markers. Following replication in a mismatch repair-deficient host cell, progeny plasmids are analyzed for each marker. Each pathway generates a unique linkage of markers, and the number of progeny plasmid derived from a given pathway represents its relative contribution to damage response in the host cell. The results are used to analyze mechanisms of responses to a DNA adduct. In this study, we conducted experiments using a single 1,N⁶-ethenodeoxyadenosine (dA). dA is one of four etheno DNA adducts, including 3,N⁴-etheno dC, N²,3-etheno dG, and 1,N²-etheno dG, that are produced by both endogenous and exogenous agents (23, 36). These adducts are mutagenic in vivo (2, 4, 14, 22, 25, 26). Although dA is barely mutagenic in E. coli, it is strongly mutagenic in mammalian cells (26).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

E. coli strains and oligodeoxyribonucleotides. The E. coli strains used in this study are shown in Table 1. All MO strains were constructed by P1 transduction (20). The alkA1 tag-1 genotype was confirmed by testing sensitivity to methyl methanesulfonate; the uvrA recF recO recR ruvABC and recA genotype was tested by determining sensitivity to UV irradiation; and mutS was tested by an increased spontaneous mutation frequency as determined by resistance to rifampin. The synthesis, purification, and characterization of the oligodeoxyribonucleotide containing dA have been described previously (26).

Plasmid vectors. The construction of plasmid pSBK (8.4 kbp) has already been described (15). It is a shuttle vector containing the simian virus 40, BK, ColE1, and f1 origins of replication, the BK T-antigen gene, and the neomycin and ampicillin resistance genes. Strand-specific marker sequences were introduced by site-directed mutagenesis. Unique SpeI and NheI sites were constructed in one pSBK molecule, and unique AatII and BamHI sites were constructed in another. The SpeI and AatII sites were constructed at identical positions. BamHI and NheI sites are located 220 nucleotides upstream and 150 nucleotides downstream, respectively, from the site of the DNA adduct (see Fig. 3). The plasmid bearing SpeI and NheI sites and that bearing AatII and BamHI sites are named pS and pA, respectively (Fig. 1). Hybrid duplex DNA constructed from pS and pA forms mismatches at three regions (SpeI/AatII, NheI, and BamHI sites), which serve as strand-specific tags (see Fig. 3).

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Structure of pA and pS. pA contains A, C, and D strand-specific markers, and pS contains a, c, and d marker sequences. B and b markers are generated during HD DNA construction; otherwise, pA and pS are identical. For the marker sequences, see Fig. 3. SV40, simian virus 40.

Construction of HD DNA containing a single dA adduct. The scheme for the procedure used to construct HD DNA containing a single dA adduct is shown in Fig. 2. Detailed procedures have been described elsewhere (15); in brief, ds pA was digested with EcoRV (step I) and the linearized plasmid DNA was ligated to a blunt-ended duplex 13-mer, 5' d(AGGTACGTAGGAG)/3' d(TCCATGCATCCTC) containing a SnaBI site (5'TACGTA) (step II). Two constructs, each containing a single insert, with opposite orientations were isolated; one of these, pA106, was used in this study. Single-stranded (ss) pA106 was mixed with EcoRV-digested ds pS (steps III and IV). This mixture was treated with NaOH to denature ds pS and then neutralized to form ds DNA. Circular ss pA106 and its complementary strand from ds pS anneal to form HD DNA with a 13-nucleotide gap. Unmodified and modified 13-mers [3' d(TCCATAXCTCCTC), where X is dA or dA] were phosphorylated at the 5' termini using T4 polynucleotide kinase and ATP, annealed to the gap, and ligated by T4 DNA ligase. The 13-mers are not perfectly complementary to the gap sequence, forming three base mismatches at and adjacent (5' and 3') to the site of dA (Fig. 3). These mismatches are located opposite the SnaBI site and also serve as strand-specific markers. The ligation mixture was treated with SpeI and EcoRV to remove residual ds pS. Closed circular ds DNA was purified by ultracentrifugation in a CsCl-ethidium bromide solution. DNA was concentrated by Centricon 30 (Amicon, Beverly, Mass.), and the concentration was determined spectrophotometrically.

View larger version (19K): [in this window] [in a new window]   FIG. 2.   Construction of HD DNA containing a single dA (X) residue. (I) EcoRV digestion of pA and pS. (II) Ligation of a duplex 13-mer containing a SnaBI site (5' TACGTA). (III) Preparation of ss DNA. (IV) Preparation of gapped HD DNA. (V) Ligation of a modified 13-mer. Note that there are three contiguous base mismatches (highlighted) in the SnaBI site. There are also mismatches at the SpeI (d)/AatII (D), BamHI (a/A), and NheI (c/C) sites (not shown here for simplicity; see Fig. 3 for details). These mismatched regions serve as strand-specific markers.

View larger version (18K): [in this window] [in a new window]   FIG. 3.   DNA sequence of regions containing strand-specific marker sequences and probes used for oligonucleotide hybridization. Marker sequences in the sequence-specific probes, A to D (overscored) and a to d (underlined), are highlighted. L and R probes (underlined) were used to identify progeny containing the 13-mer insert. Probes b (SA), ST, SG, SC, and SD were used to determine which base replaced dA. -, direct connection of bases (e.g., G-G is GG); ~, sequence interruption. Approximate distances between markers are shown in parentheses.

Transformation of E. coli and analysis of progeny plasmid DNA. Control or modified DNA (6 ng) was introduced into mismatch repair-deficient (mutS) MO strains by electroporation. The mutS mutation prevents mismatch repair. 2× YT medium (950 µl) was added to the electroporation mixture (50 µl) and then incubated for 20 min at 37°C (2 × YT medium contains [per liter] 16 g of tryptone, 10 g of yeast extract, and 5 g of NaCl [pH7]). A portion (10 to 50 µl of a 100× dilution) was plated onto a plate of 1× YT medium containing ampicillin (100 µg/ml) to determine the number of transformants in the mixture. The remaining mixture was incubated for 40 min and then added to 10 ml of 2× YT medium containing ampicillin. The mixture was cultured overnight, and progeny plasmid DNA was prepared by alkaline lysis. Purified plasmid DNA was used to transform E. coli DH5. This second transformation segregates progeny plasmid DNA derived from each strand of HD DNA. Transformants were inoculated into 96-well plates and cultured for several hours. Bacteria were spotted onto filter paper placed on a 1× YT medium-ampicillin plate and cultured overnight. Filters were treated with 0.5 M NaOH, neutralized in 1 M Tris-HCl buffer (pH 7.4), washed in 1× SSC (0.15 M NaCl plus 0.015 M sodium citrate) and then in ethanol, and baked at 80°C for 2 h. To detect each strand-specific marker sequence, differential oligonucleotide hybridization (26) was conducted using the ³²P-labeled probes shown in Fig. 3. An example of hybridization is shown in Fig. 4.

View larger version (116K): [in this window] [in a new window]   FIG. 4.   Oligonucleotide hybridization of E. coli transformants. MO199 was transformed with dA-containing HD DNA. Progeny plasmids were used to transform E. coli DH5. DH5 transformants were hybridized with all of the probes shown in Fig. 3.

Interpretation of results: progeny plasmid DNA derived from possible pathways. When the HD construct bearing a single dA residue is introduced into a mismatch repair-deficient host, several events may occur (Fig. 5). If the adduct is removed by excision repair before encountering the DNA replication apparatus, the 5' and 3' flanking mismatches are also removed and gap-filling DNA synthesis converts 5' CXA (b) to 5' ACG (B). As a result, progeny DNAs derived from the repaired strand contain the a-B-c-d linkage (progeny III) (step 1 of Fig. 5). If the construct is replicated without repair, progeny DNAs produced from the unmodified strand contain the A-B-C-D linkage (progeny I, step 2) while those from the modified strand show the a-b-c-d linkage (progeny II) when TLS occurs (step 3). When DNA synthesis is blocked by the adduct, it may be overcome by one of three mechanisms. One pathway involves UmuD'₂C/RecA-assisted TLS. Other possibilities are template strand switching and daughter strand gap repair (6). In the former mechanism, upon encountering a blocking lesion, the 3' end of the nascent strand switches its template to the newly synthesized strand of its sister molecule (step 9). When this mechanism operates, our construct will generate progeny with the a-B-C-d linkage (progeny IV, steps 98). In the daughter strand gap repair model, a ss gap, resulting from a synthesis block, is filled by strand transfer of the unmodified parental strand (step 4). The 3' end of the blocked nascent strand is used to replicate the transferred region (step 5). These processes generate a Holliday junction. When the RuvA-RuvB complex migrates the Holliday junction and the Holliday junction-specific endonuclease, RuvC, resolves the Holliday junction (steps 6 and 7), four types of progeny arise, i.e., progeny with the linkages A-B-C-D (progeny I), a-B-C-d (progeny IV), a-B-C-D (progeny V), and A-B-C-d (progeny VI). Progeny I and progeny IV may also be produced when the Holliday junction is resolved by reverse branch migration catalyzed by RecG (step 8).

View larger version (44K): [in this window] [in a new window]   FIG. 5.   E. coli pathways involved in the processing of DNA damage. Modified input DNA (X represents dA) is boxed. The ColE1 origin of replication (short vertical bars) and the direction of replication (short arrow) are shown. A to D and a to d correspond to those shown in Fig. 3. Progeny I to VI correspond to those listed in Tables 2 to 4. Excision repair of dA (step 1) converts the sequence of the adducted region to B; replication of a repaired molecule produces progeny I and III. When modified DNA replicates (step 2), the complementary strand produces progeny I and the modified strand yields progeny II following TLS (step 3). When the adduct inhibits DNA synthesis, DNA damage avoidance mechanisms, daughter strand gap repair (steps 4 to 8) and/or template strand switching (steps 9 and 8), operate to overcome the inhibition. The daughter strand gap repair mechanism involves strand transfer (step 5), formation of a Holliday junction, gap-filling synthesis (step 5), and branch migration (step 6). Resolution of a Holliday junction by RuvC resolvase produces progeny I, IV, V, and VI (step 7). When a Holliday junction is resolved by reverse branch migration (step 8), progeny I and IV are produced. In the template strand switch mechanism, upon encountering a blocking lesion, the nascent strand switches its template to the sister molecule and synthesis continues (step 9). After bypassing the adducted region, the strand reassociates with the original template strand and continues synthesis (step 8). This mechanism produces progeny I and IV.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Inhibition of DNA synthesis and translesional DNA synthesis. The HD construct was introduced into excision repair-deficient strain MO934 (alkA1 tag1 uvrA6) and strain MO937, a recA derivative of MO934 (Table 2). When the unmodified construct was used, more than 97% of the progeny consisted of progeny I and II, derived from the replication of each strand. In both strains, the number of progeny II was approximately 1.8-fold more than that of progeny I. A similar phenomenon has been reported by others (9). When the modified construct was used, the number of progeny II decreased markedly, indicating that dA strongly blocks DNA synthesis. In excision repair- and recombination-deficient strain MO937, the approximate efficiency of TLS was 11% (7 of 63). The fraction of progeny II ranged from 7 to 14% in the strains used (Tables 2 to 4). To determine the base inserted opposite dA, differential oligonucleotide hybridization was performed. This method allows the detection of a single base mismatch using oligonucleotide probes (38) and is frequently used in site-specific mutagenesis experiments (9, 21). Purified progeny plasmids were digested with SnaBI and used for transformation. This digestion results in enrichment in progeny II derived from the TLS pathway (Fig. 5). Hybridization using the b (SA), ST, SG, SC, and SD probes revealed that all of the 312 progeny II analyzed had dA at the position of dA, indicating that the correct nucleotide, dTMP, had been inserted opposite this adduct. Thus, TLS is highly accurate and probe b will detect TLS events.

Excision repair of dA. The fractions of progeny III having the linkage for excision repair events were >2% in excision repair-deficient strains (Table 2). The fractions increased marginally in repair-proficient strains, ranging from 3 to 8% (Tables 3 to 5), indicating slow repair of this adduct located in the mismatched region. This result is in contrast to that obtained with 3H-8-hydroxy-3-(-D-2'-deoxyribofuranosyl)-5,6,7,8-tetrahydropyrido[3,2-a]purine-9-one [-(OH)-PdG], an endogenous DNA adduct that is repaired by nucleotide excision repair, for which progeny III accounts for >30% when a modified construct is introduced into excision repair-proficient cells (unpublished result). Our results suggest that dA is not a good substrate for either base or nucleotide excision repair in E. coli. An in vitro study also has shown that dA is repaired very inefficiently by base excision repair catalyzed by the E. coli alkA gene product, 3-methyladenine DNA glycosylase (30, 43).

Recombination repair. Recombinants (progeny IV, V, VI, and others) accounted for <3% of the progeny when the unmodified control construct was introduced into MO934 (recA⁺) and MO937 (recA) (Table 2). The relative number of recombinants increased 10-fold (24%) when the dA-modified construct was introduced into MO934. Progeny IV, V, and VI account for the majority (91%) of recombinants. When the recA mutant (MO937) was used, there was no increase in the number of recombinants. These results clearly indicate that progeny IV, V, and VI are generated by a RecA-dependent pathway(s) in response to dA.

Genes involved in the damage tolerance-avoidance mechanism(s). Recombinants may be derived from the daughter strand gap repair and/or strand switching pathways depicted in Fig. 5. First, we examined the requirement for genes hypothesized to be involved in the presynapsis and synapsis stages of daughter strand gap repair (11, 12). When the recF, recO, or recR gene was inactivated, the number of recombinants decreased markedly, revealing a requirement for these genes (Table 3). Inactivation of the recN gene, which is thought to be involved in ds break repair (12), showed no effect.

View larger version (18K): [in this window] [in a new window]   FIG. 6.   Frequency of recombinant progeny in various E. coli strains transformed with dA-containing plasmid DNA. The y axis represents the percentage of recombinants (progeny IV, V, and VI) among all of the progeny. recA+, recA, and "wild" denote MO934, MO937, and MO199, respectively. Refer to Tables 3 to 5 for the other strains. mc indicates treatment with mitomycin C; refer to the legend to Table 5. An asterisk indicates P < 0.001, as determined by the 2 test, compared with both controls (recA+ and "wild").

Effects of SOS induction and inactivation of proofreading activity of pol III. The induction of SOS functions by mitomycin C treatment increased the fraction of progeny II twofold, suggesting that inducible SOS DNA polymerases such as pol II, IV, and V may catalyze TLS accurately. Enhancement of the accurate TLS is also observed by the inactivation, due to mutD5, of the proofreading function of DNA pol III. The SOS induction or the mutD5 mutation did not influence the fraction of recombinants (Table 4).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Several pathways operate in E. coli in response to DNA damage (Fig. 5). To examine the individual contributions of these pathways, we have devised a novel plasmid-based approach using a single DNA adduct. Our strategy is to sort progeny plasmids according to the linkage of marker sequences, since each pathway produces progeny with a unique linkage. Results of this study indicate that dA strongly inhibits DNA synthesis in E. coli; the efficiency of TLS was approximately 11%, and the induction of SOS functions mitigates the blocking effects. However, a significant fraction of DNA synthesis was still inhibited and a portion of the blocked DNA synthesis was rescued by daughter strand gap repair. Inactivation of the proofreading activity of DNA pol III also increased TLS, possibly by forcing pol III to perform TLS. TLS was highly accurate in the presence and in the absence of induced SOS functions and also in the presence of the mutD5 mutation. These findings suggest strongly that the correct nucleotide is almost exclusively incorporated opposite dA in E. coli.

Our results clearly show, at the DNA sequence level, that the E. coli damage avoidance-tolerance mechanism operates to rescue nascent strands when progression of DNA synthesis is blocked by dA (Table 2). Our results are consistent with daughter strand gap repair. We observed an increase in the number of progeny IV, V, and VI, which are predicted to emerge from steps 5 to 10 in Fig. 5. Production of these recombinants depends on the functions of the recA, recF, recO, and recR genes. Biochemical studies have revealed that the RecO-RecR complex facilitates binding of RecA protein to an ss DNA-binding protein-coated region of ss DNA and prevents the end-dependent disassembly of the RecA filament (8, 32). The RecF-RecR complex binds to ds DNA and is thought to inhibit excessive extension of a RecA filament into ds DNA (39). RecA is known to search for homology and to catalyze strand transfer. These biochemical activities explain the lack of recombinants in these mutants in our assay. RecF is also thought to be required for the resumption of DNA synthesis (5).

The results obtained in this study argue against the idea that the template strand switching mechanism plays a major role in damage avoidance of dA. Our experimental system predicts the appearance of progeny IV if this mechanism were to operate (Fig. 5). Inactivation of the recA, recF, recO, and recR genes reduced markedly the numbers of all types of recombinants (Table 3). Therefore, it must be assumed that these genes are also required for any postulated template strand switching mechanism, if it exists.

Compared with the inactivation of genes involved in the presynapsis and synapsis stages, the effects of inactivation of the genes involved in the postsynapsis stage of recombination repair are complex. While inactivation of the recG gene does not affect the number of recombinants, inactivation of the ruvABC genes increased the incidence of recombinants (Table 4). RecG has ATPase and 3'5' helicase activity, binds a Holliday junction, and promotes branch migration (12). RuvA also binds a Holliday junction and targets RuvB (5'3' helicase) to a Holliday junction (12), which then promotes branch migration. RecG and RuvAB are thought to migrate a Holliday junction in opposite directions (41); RuvC resolves a Holliday junction by endonucleolytic cleavages (12), while RecG may resolve it by reverse branch migration (41). In considering these activities, it is understandable that the recG mutation had no effect on the incidence of progeny IV, V, and VI. On the other hand, the enhancing effects of mutations in the ruv genes (MO204, MO206, and MO210) are unexpected, suggesting that the RecG-catalyzed Holliday junction resolution pathway is more active than the RuvABC-catalyzed pathway and that the RuvABC pathway may suppress this RecG pathway. Enhanced daughter strand gap repair in the ruv mutants is an apparent paradox to the UV sensitivity of the ruv mutants. It is likely that a complicated combination of multiple Holliday junctions generated by daughter strand gap repair and double-strand break repair in the E. coli chromosome requires the endonucleolytic resolution mediated by RuvABC.

The second puzzle is how progeny V and VI are produced in the absence of RuvC resolvase. This is not due to the activity of cryptic RusA, since the introduction of rusA into the ruvA-C mutant showed no effect. If the Holliday junction is resolved by reverse branch migration, progeny IV and I are expected. If the Holliday junction is not resolved, a plasmid dimer containing the linkage A-B-C-d-a-B-C-D is expected. Our analysis showed that recombinant progeny hybridized only one of the two probes at the A/a and D/d sites. Additionally, the 11 recombinants of progeny IV, V, and VI were confirmed to be monomers (8.4 kbp) by agarose gel electrophoresis.

The E. coli chromosome and ColE1 plasmid contain dif and cer sites, respectively, at which a dimer is monomerized by XerCD site-specific recombinase (33). Our pUC-based plasmid does not carry the cer or dif site. Therefore, a plasmid dimer can not be monomerized. This information suggests that progeny plasmids are monomerized at the completion of recombination events. Figure 7 shows our model for the generation of progeny V and VI by the RecG-catalyzed pathway. When a daughter strand gap is filled by strand transfer, the parental strand invades the adjacent ds regions to displace a marker: d is displaced by D in the left column, and a is displaced by A in the right column. A subsequent filling reaction and reverse branch migration will generate progeny V (left column) and VI (right column). In the latter case, the marker a must be removed by a nuclease before initiation of the gap-filling reaction. Since the number of progeny VI did not increase in the ruv mutants (Table 4), the latter pathway in Fig. 6 may be a rare event. If the high incidence of progeny V in the ruv mutants is ascribed to this pathway (left column), it is predicted that DNA synthesis is not resumed between marker c and the origin of replication, which is 3,100 nucleotides away. Hence, inactivation of the polB gene, which has been recently suggested to function in the resumption of DNA synthesis (28), will have no effect on the incidence of progeny V in the ruv strains. The introduction of another marker between c/C and d/D may help to answer this question.

View larger version (33K): [in this window] [in a new window]   FIG. 7.   Possible mechanisms generating progeny V and VI in ruv mutants. Refer to the legend to Fig. 5 and the text for an explanation.

When a recG mutation was introduced into ruvA-C strain MO208, the incidence of recombinants decreased markedly, clearly showing that RecG plays an essential role in resolving the Holliday junction in the absence of RuvABC. However, a significant number of recombinants derived from the daughter strand gap repair mechanism were observed in MO208 (ruvA-C recG). This suggests that RuvAB- and RecG-independent branch migration still occurs, although not efficiently, in E. coli.