SOS and UVM Pathways Have Lesion-Specific Additive and Competing Effects on Mutation Fixation at Replication-Blocking DNA Lesions

Abstract

Escherichia coli cells have multiple mutagenic pathways that are induced in response to environmental and physiological stimuli. Unlike the well-investigated classical SOS response, little is known about newly recognized pathways such as the UVM (UV modulation of mutagenesis) response. In this study, we compared the contributions of the SOS and UVM pathways on mutation fixation at two representative noninstructive DNA lesions: 3,N⁴-ethenocytosine (ɛC) and abasic (AP) sites. Because both SOS and UVM responses are induced by DNA damage, and defined UVM-defective E. coli strains are not yet available, we first constructed strains in which expression of the SOS mutagenesis proteins UmuD′ and UmuC (and also RecA in some cases) is uncoupled from DNA damage by being placed under the control of a heterologous lac-derived promoter. M13 single-stranded viral DNA bearing site-specific lesions was transfected into cells induced for the SOS or UVM pathway. Survival effects were determined from transfection efficiency, and mutation fixation at the lesion was analyzed by a quantitative multiplex sequence analysis procedure. Our results suggest that induction of the SOS pathway can independently elevate mutagenesis at both lesions, whereas the UVM pathway significantly elevates mutagenesis at ɛC in an SOS-independent fashion and at AP sites in an SOS-dependent fashion. Although mutagenesis at ɛC appears to be elevated by the induction of either the SOS or the UVM pathway, the mutational specificity profiles for ɛC under SOS and UVM pathways are distinct. Interestingly, when both pathways are active, the UVM effect appears to predominate over the SOS effect on mutagenesis at ɛC, but the total mutation frequency is significantly increased over that observed when each pathway is individually induced. These observations suggest that the UVM response affects mutagenesis not only at class 2 noninstructive lesions (ɛC) but also at classical SOS-dependent (class 1) lesions such as AP sites. Our results add new layers of complexity to inducible mutagenic phenomena: DNA damage activates multiple pathways that have lesion-specific additive as well as suppressive effects on mutation fixation, and some of these pathways are not directly regulated by the SOS genetic network.

Autonomous organisms maintain a high level of replication fidelity through a variety of constitutive and inducible error avoidance mechanisms. However, in response to environmental and physiological stimuli, cells seem to have the ability to transiently reduce replication fidelity. The operation of these transient mutator pathways has broad implications for human health because of their potential impact on areas as diverse as cancer, aging, and emergence of resistant pathogenic organisms. The best-known example of a DNA damage-inducible mutagenic pathway is the Escherichia coli SOS response (for a review, see reference 4). The SOS response is mediated by the induced expression of some 20 member genes of a regulon that is normally transcriptionally repressed by the binding of the LexA protein to promoter sequences. DNA damage leads to replication arrest and generates a signal that activates the RecA protein (itself coded by an SOS gene) into the SOS-active RecA* (coprotease) form. The RecA* protein derepresses the SOS regulon by cleaving the LexA repressor protein and also assists in proteolytic activation of the UmuD protein to the SOS-active UmuD′ form. It is believed that RecA, UmuD′, and UmuC proteins alter the replication fidelity of the E. coli DNA polymerase III holoenzyme and accounts for the error-prone replication observed in cells subjected to DNA damage. Although the precise biochemical roles of the three SOS proteins have not been fully described, two recent reports indicate that the RecA and UmuD′C proteins may have biochemical functions as replication cofactors during translesion DNA synthesis (20, 27). Interestingly, overexpression of the SOS gene dinB was shown to confer a mutator phenotype in which mutagenesis was significantly elevated at apparently undamaged sites (8). Thus, dinB is implicated in at least some forms of untargeted mutagenesis accompanying SOS mutagenesis.

A second damage-inducible response, termed UVM (for UV modulation of mutagenesis), has been recently described. UVM is detected as increased mutation fixation at a site-specific 3,N⁴-ethenocytosine (ɛC) lesion borne on transfected M13 viral single-stranded DNA (ssDNA) in cells pretreated with a variety of DNA-damaging agents (6, 14). This response is distinct from the SOS response in its genetic requirements because UVM does not require functional recA, umuD, and umuC genes and occurs in cells under conditions where the SOS functions are not induced (17). UVM appears to significantly affect mutagenesis at class 2 noninstructive lesions such as ɛC and possibly 1,N⁶-ethenoadenine but not at mispairing lesions such as O⁶-methylguanine (19). The mechanisms underlying the UVM response are not known, but the preponderance of evidence points to the transient induction of an error-prone replication activity (12–14, 17).

In this study, we have compared the individual effects of the SOS and UVM pathways on mutation fixation at two noninstructive (ɛC and abasic [AP] site) lesions so that we may begin an assessment of the impact of these pathways on induced mutagenesis. ɛC is an exocyclic lesion induced by carcinogens such as vinyl chloride and ethyl carbamate but it is now recognized to be also induced by endogenous mutagens and may thus be part of the spontaneous DNA damage burden of the cell. Because the exocyclic amino nitrogen in ɛC is linked to the ring nitrogen 3 by a two-carbon bridge, two of the three Watson-Crick pairing positions become unavailable. ɛC is highly mutagenic and induces mostly base substitutions, with C→A and C→T mutations predominating (7). Although ɛC has the in vitro template characteristics of a noninstructive DNA lesion (23), it is highly mutagenic in ΔrecA cells (16). In contrast, AP-site, the other model lesion included here, is a classic example of a recA-dependent (SOS-dependent) noninstructive lesion. A large number of spontaneous as well as damage-initiated mechanisms generate AP sites. M13 ssDNA bearing AP sites is very poorly replicated in host cells unless the host has been induced for SOS functions, indicating that these sites act as strong replication blocks (9). The mutational specificity of AP sites suggests an AP:N base insertion-bypass mechanism in which N = A > T/G >> C.

The data presented here indicate that both SOS and UVM pathways significantly, and independently, elevate mutagenesis at ɛC. However, mutational specificity analyses show that the UVM effect predominates over the SOS effect when both pathways are induced. Mutagenesis at AP sites is strongly stimulated by the SOS pathway, with the UVM pathway causing a significant further enhanced mutagenesis for which SOS induction is a prerequisite.

MATERIALS AND METHODS

Strains.

Bacterial and plasmid strains are listed in Table 1. Plasmid pSR1000 (Fig. 1; 6,383 bp), expressing umuD′C genes under the control of a single inducible lac-derived promoter, P_trc, was constructed by ligating three fragments: (i) a 4,443-bp EcoRI-NcoI fragment from the P_trc vector plasmid pSE380 (2), (ii) a 1,897-bp EcoRI-ClaI fragment from plasmid pEC42 (3) containing the entire umuC gene as well as a 3′ part of the umuD gene, and (iii) a synthetic oligonucleotide duplex created by annealing a 45-mer (5′-CATGGGCTTTCCTTCACCGGCAGCAGATTACGTTGAACAGCGCAT) and a 43-mer (5′-CGATGCGCTGTTCAACGTAATCTGCTGCCGGTGAAGGAAAGCC). The synthetic duplex had ClaI- and NcoI-compatible ends and completed the 5′ sequence of the umuD′ gene that encodes the equivalent of the UmuD′ protein. To place the recA gene downstream of umuD′C genes as a part of the same operon, plasmid pSR1000 was modified to remove the transcription termination sequence at the end of the umuC gene as follows. A 1,649-bp fragment that included the umuD′C region of pSR1000 plasmid was amplified by PCR by using the primers P1 (same sequence as the 45-mer shown above) and P2 (5′-GCGGGAGCGCTTTTcTCgaGCCGCTAT). The P2 primer included three mismatches (lowercase letters) to create a site for XhoI (underlined) 9 bp downstream of the umuC translational termination codon. The PCR product was cut with BglII and XhoI to obtain a 1,102-bp fragment (containing the 3′ part of the umuC gene) that was ligated to a 4,695-bp BglII-XhoI fragment also from pSR1000 to generate plasmid pSR1007 (Fig. 1; 5,797 bp). The recA gene of E. coli KH₂ was amplified by PCR using a forward primer (5′-GCTTCAACAGAACtcgagGACTATCCGG) with five mismatches (lowercase letters) that inserted an XhoI site (underlined) 26 bp upstream of the Shine-Dalgarno sequence of the recA gene. The reverse primer (5′-CGACGGGATGTTGAactaGTCATGGCAT) with four mismatches (lowercase letters) was used to create an SpeI site (underlined) 71 bp downstream of the UAA stop codon of recA gene. The 1,207-bp PCR product was digested with XhoI and SpeI to obtain a 1,175-bp fragment containing a promoterless recA gene which was ligated with the 5,787-bp XhoI-SpeI fragment from plasmid pSR1007 to obtain pSR1718 (6,962 bp). Functional characterization of pSR1007 and pSR1718 is described elsewhere in the text.

TABLE 1.

Plasmids and bacterial strains used

Plasmid or strain	Relevant genotype	Source (reference)
Plasmids
pSE380	Vector (Ampr)	R. Maurer (2)
pSR1000	pSE380 derivative expressing umuD′ and umuC genes under the control of a Ptrc promoter	This work
pSR1007	pSR1000 derivative expressing umuD′ and umuC genes under the control of a Ptrc promoter	This work
pSR1718	pSR1007 derivative expressing umuD′, umuC, and recA genes under the control of a Ptrc promoter	This work
pEC42	pBR322-derived plasmid bearing umuD′ and umuC genes under the control of a lambda pL promoter	R. Woodgate (3)
E. coli
KH2	Sup0 Δlac-pro trpE9777 (F′ lacIqZΔM15 pro+)	This lab (21)
KH2R	Sup0 Δlac-pro trpE9777 Δ(srlR-recA)306::Tn10(Tetr) (F′ lacIqZΔM15 pro+)	This lab (16)
GY8347	F− Δ(umuD/C)595::cat	S. Sommer (25)
SR100	Δ(umuD/C)595::cat in KH2	This work
SR400	Δ(srlR-recA)306::Tn10(Tetr) in SR100	This work
SR230	SR100 bearing plasmid pSE380	This work
SR420	SR100 bearing plasmid pSR1007	This work
SR440	SR400 bearing plasmid pSR1718	This work

FIG. 1.

Maps of plasmids used in this study, depicting major features and relevant restriction sites. For construction details, see Materials and Methods.

Construction of ssDNA bearing site-specific lesions.

Briefly, phosphoramidite derivatives of ɛC (14, 17) and AP (29) residues were used to introduce these lesions into 17-mer oligonucleotides by chemical synthesis. Appropriate procedures at the deprotection stage to conserve the chemical integrity of the ɛC lesion were used as described previously (14, 17). The synthetic AP residues are known to be stable under conditions of oligonucleotide synthesis and were synthesized by Midland Certified Reagent Company (Midland, Tex.). All oligonucleotides were purified by a final step of electrophoresis on high-resolution (polyacrylamide-urea) gels and did not appear to have impurities at a detectable level. M13 ssDNA vectors bearing site-specific ɛC (14, 17) or synthetic AP (29) lesions were prepared as previously described in detail and summarized in Fig. 2A. The constructed ssDNAs were denatured in the presence of an antiscaffold oligonucleotide immediately before transfection (17, 18, 29).

FIG. 2.

Summary of procedures used for construction of M13 ssDNAs bearing lesions, for SOS or UVM induction, for transfection, and for multiplex sequence analyses. (A) Schematic representation of procedures for construction of M13 ssDNA molecules bearing site-specific lesions. Detailed methods have been described previously (17, 29). (B) Procedures used for SOS and/or UVM induction, transfection, and measurement of survival and mutagenic effects (for details of the methodology, see references 17 and 28). Briefly, cells grown in LB medium to mid-log phase (optical density at 600 nm of 0.35) in either the presence (for SOS induction) or absence of 1 mM IPTG were exposed to MNNG (for UVM induction; 10 μg/ml, final concentration) for 10 min at 37°C, followed by cell pelleting by centrifugation, washing by resuspension in LB medium, and then a second pelleting step. The washed cells were resuspended in 1/10 original volume of cold TSS solution (LB medium containing 10% [wt/vol] polyethylene glycol 3350, 5% [vol/vol] dimethyl sulfoxide, and 20 mM MgCl₂) and were processed to render them transfection competent (16). ssDNA (about 50 ng) was transfected into 1 ml of competent cells, and two 0.1-ml aliquots were plated for infectious centers (ic) to determine effects of the lesion and the strain on the transfected DNA. The remainder of the cells were used to prepare pooled progeny phage ssDNA as described elsewhere (14). (C) Principles of multiplex sequence analysis as applied to analysis of mutagenesis at ɛC and AP lesions. A prelabeled 19-mer primer is annealed to the pooled progeny phage ssDNA and allowed to elongate in the presence of dGTP, dCTP, and ddTTP. Depending on the base at position N, limit-elongation products of characteristic length are produced; these are fractionated on high-resolution gels and quantitated as described elsewhere (14, 15). In the case of ɛC, C→A transversions will yield a 22-mer and C→T transitions will yield a 21-mer. ɛC is also known to induce −1 nt deletions that will give rise to a 23-mer. Wild-type sequence will give rise to 24-mers (note that any C→G transversions can also give rise to a 24-mer, but ɛC does not induce C→G mutations at appreciable levels [7, 16]). In the case of AP sites, AP→T, AP→A, −1 nt, and AP→G/C mutations will yield a 21-mer, a 22-mer, a 23-mer, and a 24-mer, respectively.

Transfection, survival effects, and mutational effects.

Procedures for transfection of the DNA constructs, for measurement of survival effects, and for multiplex sequence analyses have been described previously (14, 15, 17) and are summarized in Fig. 2B and C and in the legend to Fig. 2.

RESULTS

Experimental system.

The experimental system used consists of transfecting M13 ssDNA bearing a site-specific ɛC or AP site lesion into cells in which the SOS pathway, the UVM response, or both can be induced through appropriate genetic manipulation and pretreatment with a UVM-inducing agent. Survival effects of the lesion are determined as transfection efficiency, and mutation fixation at the lesion site is monitored by a multiplex sequence analysis technology. M13 ssDNA replication is known to proceed through two stages. In stage I, complementary (minus)-strand initiation occurs at a unique site (Ori⁻) by the synthesis of a 20-nucleotide (nt)-long RNA primer by the host RNA polymerase. The primer is elongated around the genome by DNA polymerase III. Primer removal and gap filling are thought to be carried out by DNA polymerase I. Subsequent nick sealing by DNA ligase and negative supercoiling by DNA gyrase give rise to replicative form I (RF-I) DNA. Thus, mutation fixation is thought to occur during ssDNA→RF DNA (stage I) replication that depends exclusively on host replication proteins. In stage II, a rolling circle DNA replication mode that requires a phage-specified protein mediates both RF→RF replication and RF→ssDNA replication.

Construction of E. coli strains in which expression of SOS mutagenesis proteins is uncoupled from DNA damage.

While the SOS response is well characterized, and mutants defective for the SOS response are available, UVM-defective mutants are not yet available. Because DNA-damaging treatments can induce both pathways, we needed a means to allow for the expression of the SOS pathway without concomitant expression of the UVM response. To achieve this end, we opted for the strategy of placing expression of the SOS proteins under the control of a heterologous promoter rather than under the control of DNA damage-inducible SOS promoters. Plasmid pSR1718 (Fig. 1) was constructed by cloning the coding sequences for UmuD′, UmuC, and RecA proteins under the control of a P_trc promoter in the vector pSE380 (2). A ΔrecA ΔumuDC E. coli host cell bearing plasmid pSR1718 should express the three SOS proteins in response to the lac inducer isopropyl-β-d-thiogalactoside (IPTG) rather than to DNA damage.

Table 2 summarizes data on the functional characterization of the strains that we have created. As seen from Table 2, experiment A, exposure of wild-type (KH2) cells to a 50-J/m² dose of UV reduces survival to about 20% but increases mutagenesis measured as forward mutation to rifampin resistance very significantly (11.4 mutants/10⁶ survivors). Experiment B shows the effect of loss of the umuDC genes: a small though appreciable decrease in survival but drastic reduction in UV mutagenesis. Experiment C shows that, as expected, introduction of the cloning vector plasmid does not significantly alter the survival and mutagenesis patterns in the ΔumuDC strain in the presence of IPTG (the same pattern is observed in the absence of IPTG [data not shown]). In contrast, the data for experiment E show that the P_trc-umuD′C expression is able to completely restore both UV resistance and UV mutagenesis levels of the ΔumuDC strain. Comparison with the data for experiment D suggest that full restoration requires IPTG. Essentially similar results are seen with the ΔrecA ΔumuDC strain (experiment F) in combination with P_trc-umuD′C recA expression (experiment G), except for the extreme effects of UV (note the lower UV dose) on both survival and mutagenesis in the triply defective strain. In experiment H, there was a complete restoration of UV resistance as well as UV mutagenesis by P_trc-umuD′C recA expression in the presence of IPTG. Interestingly, the mere presence of P_trc-umuD′C recA genes, even in the absence of IPTG, confers significant UV resistance and UV mutagenesis in comparison to the triple mutant not bearing the plasmid. This observation indicates significant basal-level expression of the three SOS mutagenesis genes.

TABLE 2.

Functional characterization of plasmids expressing SOS functions

Expt	E. coli strains (relevant genotype) [plasmid-expressed SOS genes]	UV (J/m2)	IPTG (1 mM)	Survival (%)	No. of rifampin-resistant mutants/ 106 survivors
A	KH2	0	−	100	0.06
	KH2	50	−	21.4	11.4
B	SR100 (ΔumuDC)	0	−	100	0.14
	SR100 (ΔumuDC)	50	−	7.4	0.58
C	SR230 (ΔumuDC) [pSE380 vector]	0	+	100	0.04
	SR230 (ΔumuDC) [pSE380 vector]	50	+	9.5	0.97
D	SR420 (ΔumuDC) [Ptrc-umuD′C]	0	−	100	0.03
	SR420 (ΔumuDC) [Ptrc-umuD′C]	50	−	9.6	1
E	SR420 (ΔumuDC) [Ptrc-umuD′C]	0	+	100	0.16
	SR420 (ΔumuDC) [Ptrc-umuD′C]	50	+	18.1	15
F	SR400 (ΔumuDC ΔrecA)	0	−	100	0.09
	SR400 (ΔumuDC ΔrecA)	5	−	0.006	0.005
G	SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	0	−	100	0.14
	SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	50	−	0.04	2.7
H	SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	0	+	100	0.48
	SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	50	+	15	14.3

Effect of SOS and UVM pathways on survival of M13 ssDNA bearing site-specific ɛC or AP lesions.

To determine the individual effects of the SOS and UVM pathways on survival and mutagenesis at representative replication-blocking DNA lesions, we used the strategy summarized in Fig. 2A to construct M13 ssDNA genomes bearing single site-specific lesions. To induce the SOS pathway, cells were grown in the presence of IPTG; to induce the UVM pathway, cells were exposed for 10 min to N-methyl-N′-nitro-N-nitrosoguanidine (MNNG) at a concentration of 10 μg/ml. As summarized in Fig. 2B, induced cells were rendered transfection competent and were transfected with the appropriate ssDNA. Survival effects were measured as transfection efficiency (infectious centers [IC] per milliliter), and pooled progeny phage from each transfection were collected for analysis of mutation fixation. The data in Table 3 compare the effects of the induction of the SOS mutagenesis proteins (UmuD′ and UmuC), the UVM pathway, or both on the survival of ɛC-bearing ssDNA. In the absence of the induction of either pathway, survival is about 655 IC/ml. Survival increases approximately threefold by the induction of either the SOS mutagenesis pathway (umuD′C genes) or the UVM pathway. Interestingly, there is an additional twofold effect when both are induced, consistent with the simultaneous operation of two independent mechanisms. A similar pattern is observed in the ΔrecA ΔumuDC strain expressing plasmid-borne umuD′, umuC, and recA genes (Table 3). UVM or SOS induction elevates survival about twofold (Table 3; compare row 5 to rows 6 and 7), and induction of both elevates survival almost fourfold (compare rows 5 and 8). The UVM effect on survival does not require that the umuD, umuC, and recA genes be expressed at a basal level: it is observed in strain SR400, which is devoid of these genes (rows 9 and 10).

TABLE 3.

Effects of SOS and UVM pathways on mutagenesis at ɛC

E. coli strain (relevant genotype) [plasmid-expressed SOS genes]	Inductiona		Avg survivalb (IC/ml) ± SD	Avg % mutation frequency ± SDc
	SOS	UVM		C→A	C→T	−1 nt
SR420 (ΔumuDC) [Ptrc-umuD′C]	−	−	655 ± 332	24 ± 0.6	22.5 ± 3.6	4.7 ± 0.6
SR420 (ΔumuDC) [Ptrc-umuD′C]	−	+	2,360 ± 1,301	70 ± 2.4	15.8 ± 2.2	1.5 ± 0.3
SR420 (ΔumuDC) [Ptrc-umuD′C]	+	−	2,355 ± 658	25.9 ± 0.9	33.9 ± 1.6	0.4 ± 0.2
SR420 (ΔumuDC) [Ptrc-umuD′C]	+	+	4,650 ± 2,107	58 ± 1.5	24.7 ± 1.9	0.3 ± 0.2
SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	−	−	1,440 ± 919	30.5 ± 3.3	24.6 ± 4.9	1.1 ± 0.9
SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	−	+	2,820 ± 1,782	52.9 ± 2.4	10.2 ± 1.5	0.92 ± 0.4
SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	+	−	3,560 ± 1,696	31.6 ± 4.2	50.5 ± 8.8	0.47 ± 0.3
SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	+	+	5,213 ± 2,314	67.6 ± 4.3	15.3 ± 2.8	0.77 ± 0.5
SR400 (ΔumuDC ΔrecA)	−	−	210 ± 127	1 ± 0.1	0.4 ± 0.2	1.38 ± 0.5
SR400 (ΔumuDC ΔrecA)	−	+	875 ± 417	43.4 ± 1.5	7.8 ± 0.6	1.32 ± 0.3

^aExpression of the SOS genes is fully turned on (+) in strains SR440 and SR420 by including IPTG (1 mM) in the culture medium. UVM is induced (+) by exposing mid-log-phase cells to MNNG (10 μg/ml) at 37°C for 10 min followed by cell pelleting and washing. 

^bMeasured as transfection efficiency. Data shown are averages of two to three independent transfections. 

^cDetermined by quantitative computing densitometry as described in the text. Data shown are averages from at least two independent transfections (the same as the ones used for determining survival values) and six or more elongation reactions. 

Survival of AP-site-containing ssDNA is not significantly elevated independently by UVM induction in strain SR420 (Table 4; row 1 versus row 2) but is stimulated fourfold by SOS induction (Table 4, row 1 versus row 3). However, when the SOS mutagenesis proteins are fully induced, UVM induction has an almost twofold additive effect on survival (Table 4, row 3 versus row 4). An essentially similar pattern is observed in strain SR440 (Table 4, rows 5 to 8). That SOS expression is a prerequisite for the additive effect of UVM on survival of ssDNA bearing AP sites is confirmed by the lack of a significant UVM effect on AP sites in the SR400 triple mutant (Table 4, row 9 versus row 10).

TABLE 4.

Effects of SOS and UVM pathways on base insertion opposite an AP site

E. coli strain (relevant genotype) [plasmid-expressed SOS genes]	Inductiona		Avg survivalb (IC/ml) ± SD	Base insertion opposite AP site (avg % ± SD)c
	SOS	UVM		AP:A (AP→T)	AP:T (AP→A)	AP:G/C (AP→C/G)	AP:0 (AP→−1)
SR420 (ΔumuDC) [Ptrc-umuD′C]	−	−	405 ± 205	16.6 ± 3.9	1.2 ± 0.2	8.4 ± 5.3	73.8 ± 2.3
SR420 (ΔumuDC) [Ptrc-umuD′C]	−	+	470 ± 212	39.5 ± 9.1	2.9 ± 1.3	6.9 ± 6.9	50.7 ± 2.4
SR420 (ΔumuDC) [Ptrc-umuD′C]	+	−	1,805 ± 757	62.9 ± 6.3	3.7 ± 2.7	12.4 ± 2.5	21.1 ± 5.1
SR420 (ΔumuDC) [Ptrc-umuD′C]	+	+	2,885 ± 1,973	64.3 ± 7.5	6.1 ± 1.4	22.4 ± 3.3	7.2 ± 2.9
SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	−	−	520 ± 325	49.6 ± 3.3	1.3 ± 0.4	7.7 ± 2.8	41.3 ± 0.9
SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	−	+	700 ± 453	43.4 ± 2.4	6.8 ± 0.4	7.3 ± 2.6	42.5 ± 0.8
SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	+	−	1,643 ± 742	76.6 ± 5.9	3.3 ± 1.2	14.2 ± 6.3	5.9 ± 0.8
SR440 (ΔumuDC ΔrecA) [Ptrc-umuD′C recA]	+	+	2,910 ± 1,439	72.6 ± 8.2	4.4 ± 0.8	16.2 ± 5.3	6.9 ± 4.1
SR400 (ΔumuDC ΔrecA)	−	−	115 ± 35	2.4 ± 2.4	1.5 ± 0.6	3.1 ± 4	93 ± 6.4
SR400 (ΔumuDC ΔrecA)	−	+	100 ± 71	8.2 ± 2.7	1.8 ± 0.7	7 ± 2.6	83 ± 2.1

^aExpression of the SOS genes is fully turned on (+) in strains SR440 and SR420 by including IPTG (1 mM) in the culture medium. UVM is induced (+) by exposing mid-log-phase cells to MNNG (10 μg/ml) at 37°C for 10 min followed by cell pelleting and washing. 

^bMeasured as transfection efficiency. Data shown are averages of two to three independent transfections. 

^cData shown are averages from at least two independent transfections (the same as the ones used for determining survival values) and six or more elongation reactions. 

Expression of the UmuD′, UmuC, and RecA proteins did not significantly increase the survival of a control DNA construct in which cytosine replaced ɛC. For example, transfection of control DNA into E. coli SR400 (ΔumuDC ΔrecA) and SR440 (SR400 bearing the plasmid-expressed SOS genes) in the presence of 1 mM IPTG gave transfection efficiencies of 14,680 and 16,370 IC/ml, respectively. Thus, the increased survival of lesion-bearing DNAs in cells expressing the SOS proteins is most likely due to the site-specific lesion rather than to cryptic DNA damage in the ssDNA vectors.

Effect of SOS and UVM pathways on mutation fixation at ɛC.

The defining attribute of UVM is that its manifestation does not require functional recA, umuD, and umuC genes that are known to be required for SOS mutagenesis. To test whether the induction of the SOS pathway by itself can also affect mutation fixation at ɛC, we carried out the experiments in which SOS mutagenesis gene expression was placed under heterologous promoter control. Figure 3A shows the basic elements of the UVM response in wild-type (KH2; lanes 1 and 2), ΔrecA (KH2R; lanes 3 and 4), and ΔrecA ΔumuDC (SR400; lanes 5 and 6) strains. In uninduced cells, there is a low level of mutagenesis (little signal in 22-mer and 21-mer bands in comparison to the 24-mer in lanes 1, 3, and 5), whereas in UVM-induced cells there is a very high level of mutagenesis (lanes 2, 4, and 6).

FIG. 3.

Multiplex sequence analyses of mutagenesis at ɛC (A and B) and AP (C) lesions under conditions where SOS, UVM, or both are induced under conditions summarized in the legend to Fig. 2. WT, wild type.

Figure 3B shows the effects of placing SOS genes under the control of a heterologous promoter in strains in which the chromosomal umuDC (strain SR420) or umuDC as well as recA (strain SR440; triple mutant) genes are deleted. Figure 3B (lanes 1 and 5) and Table 3 (rows 1 and 5) show that mutation fixation at ɛC is high in plasmid-bearing cells even in the absence of IPTG, indicating that SOS expression at basal levels was sufficient to manifest this effect. Thus, although the UV resistance and UV mutagenesis data previously described (Table 2) also show constitutive expression of SOS mutagenesis activity, the site-specific mutagenesis assay appears to respond more sensitively. Full expression of the SOS genes further increases mutation fixation, as seen in Fig. 3B (lanes 3 and 7; compare the distribution of signal between the 24-mer wild-type band with the mutant 22- and 21-nt bands in lanes 1 versus 3 and 5 versus 7) and in Table 3 (rows 3 versus 1 and 7 versus 5). UVM induction of these strains has two effects: (i) there is a substantial further increase in overall mutagenesis the ΔumuDC strain expressing SOS proteins at basal levels (Table 3, row 2 versus row 1) and a more modest further increase in the triple mutant (row 6 versus row 5); and (ii) most strikingly, there is a difference in the specificity of mutation, as if the UVM effect supersedes the SOS effect. Thus, UVM induction results in an elevation in C→A mutations (22-mer band; lanes 2 and 6) and an apparent suppression of C→T mutations (21-mer band; lanes 2 and 6), such that the resulting pattern of mutagenesis closely resembles the UVM response, as shown in Fig. 3A, lanes 2, 4, and 6. This UVM dominance effect is equally pronounced in cells that are fully induced for expression of the SOS genes (lanes 4 versus 3 and 8 versus 7). The quantitative summary of these observations in Table 3 shows that under SOS, C→A and C→T mutations are produced in approximately equal proportions, whereas in UVM-induced cells, C→A mutations predominate two- to fivefold over C→T mutations.

Effect of SOS and UVM pathways on base insertion opposite an AP site.

The known genetic requirements for mutagenesis at AP sites indicate that SOS induction is necessary for mutation fixation opposite these classic noninstructive lesions. Because DNA damage can induce both SOS and UVM responses, it is interesting to assess the relative contributions of the SOS and UVM responses to mutation fixation at AP sites. Figure 3C, lane 1, shows that in the absence of UVM induction, a majority of mutational events in strain SR420 are −1 nt deletions (23-mer band), but base substitution events, such as AP→T (21 nt; product of an insertion of A opposite AP) and AP→C/G (AP:G and AP:C insertion events) are also visible. In quantitative terms (Table 4, row 1), deletions accounted for 74% of the bypass events, whereas AP→T (17%) and AP→C/G (8%) accounted for most of the remainder. Figure 3C (lane 2) and Table 4 (row 2) show that UVM induction causes a detectable redistribution of bypass events: deletions are apparently decreased to 51%, and there is an increase in AP→T mutations to 40% of the bypass events. Full induction of umuD′ and umuC proteins (Fig. 3C, lane 3) dramatically increases base substitution mutations apparently at the expense of deletions: Table 4 (row 3) shows that AP→T events now account for 63%, with more modest increases in AP→C/G (12%) and AP→A (4%) events, but a drop in deletions to about 21%. UVM induction (Table 4, row 4) appears to accentuate this trend, with increases in AP→C/G mutations to 22% and AP→A mutations to 6% and a further decrease in deletions to 7%.

In strain SR440 (triple mutant complemented by umuD′C and recA genes) expressing SOS mutagenesis genes at uninduced (basal) levels, 41% of the events are deletions (Fig. 3C, lane 5; Table 4, row 5). Full induction of the SOS mutagenesis proteins dramatically increases AP→T mutations to 77% and decreases the deletion fraction to 6% (Table 4, row 7). In triple-mutant cells not bearing plasmids, deletions account for a significant majority (Table 4, row 9; 93%) of the mutational events, and a small UVM effect is observed such that deletion events are reduced to about 83%, with a corresponding increase in base substitutions (Table 4, row 10).

The data in Table 4 indicate that UVM induction apparently does not cause a significant shift in the relative proportions of mutations at AP sites. However, in fully SOS-induced cells, UVM induction causes a significant further elevation in survival (Table 4, rows 3 versus 4 and 7 versus 8) of AP site DNA. The data in Fig. 4A take this survival effect into account and offer insights into the effects of UVM and SOS induction on mutagenic processing of AP lesions. Here, the total number of each type of bypass event in each transfection experiment, normalized for survival as shown in the legend for Fig. 4, is plotted. This analysis makes two interesting points regarding mutagenesis at AP sites. First, UVM appears to have a consistent additive effect on base substitution mutagenesis at AP sites, and this additive effect is dependent on the full induction of SOS functions. Second, the total number of deletion events (AP:0) is low and essentially constant (Fig. 4A and B); thus, cells appear to have a low but finite capacity to skip across AP sites, and this intrinsic ability appears to be unaffected by the UVM and SOS pathways. A similar analysis of the ɛC data (Fig. 4C and D) shows that when both SOS and UVM pathways are activated, there is a strong additive (or even synergistic) effect on C→A (ɛC:T insertion) mutations but a suppressive effect on C→T (ɛC:A insertion) mutations, suggesting complex interactions between the UVM and SOS pathways that ultimately result in higher mutagenesis than that induced by each pathway separately.

FIG. 4.

Number of insertion events at a site-specific AP site (A) or ɛC (C) in M13 ssDNA transfected into E. coli SR440. The numbers of insertion events (from Tables 3 or 4) were normalized to ssDNA survival under each experimental condition by multiplying the fraction of represented by each type of insertion with the number of IC per milliliter (i.e., per transfection). For example, the AP:A fractions in strain SR440 under UVM− SOS−, UVM+ SOS−, UVM− SOS+, and UVM+ SOS+ conditions are, respectively, 0.496, 0.434, 0.766, and 0.726 (Table 4, fifth column, rows 5 to 8). Multiplying each fractional value with the corresponding survival value yields the following normalized numbers for AP:A insertion events: 258, 304, 1,259, and 2113. (B) Analysis of insertion events at AP sites similar to that in panel A except that data from SR400 (Table 4, rows 9 and 10) were used to calculate SOS− UVM− and SOS− UVM+ conditions and data from SR440 were used for the UVM− SOS+ and UVM+ SOS+ conditions. Thus, the AP:A fractions used for normalizing the number of events under UVM− SOS−, UVM+ SOS−, UVM− SOS+, and UVM+ SOS+ conditions were 0.024, 0.082, 0.766, and 0.726, respectively; the corresponding normalized numbers were 3, 8, 1,259, and 2,113. (D) Analysis of insertion events at ɛC similar to that shown in panel C except that data from strain SR400 were used to represent the UVM− SOS− and UVM+ SOS− conditions as described above.

DISCUSSION

In this report, we present the results of an analysis of the relative contributions of two genetically distinct DNA damage-inducible pathways to mutagenesis at two representative noninstructive DNA lesions. A prerequisite for this study has been the development of strains in which SOS, UVM, or both can be induced. Because of the lack of defined UVM-defective mutations, we chose the strategy of uncoupling SOS induction from DNA damage by expressing SOS mutagenesis genes under the control of a heterologous promoter. A similar strategy was previously used by Boudsocq et al., who placed the umuD′ and umuC genes under the control of the pBAD promoter (1). Because of the subsequent report that expression of genes under pBAD control can be heterogeneous (22), we opted for the strategy of placing the SOS genes under the control of a lac-derived promoter. This strategy allows one to induce the SOS mutagenesis proteins by the simple means of adding IPTG to the growth medium. The data summarized in Table 2 confirm that SOS mutagenesis is IPTG inducible, as judged by the restoration of UV resistance and UV mutability to ΔumuDC cells and to ΔumuDC and ΔrecA cells by plasmids expressing the missing proteins under the control of the P_trc promoter. These data provide further confirmation to previous conclusions that expression of UmuD′ and UmuC proteins is necessary and sufficient for UV mutagenesis, and they also confirm that constitutive levels of RecA protein are sufficient for SOS mutagenesis and that the third role of RecA in SOS mutagenesis can be satisfied with uninduced levels of normal (as opposed to RecA*) RecA protein (24, 26). In these strains, UVM is induced by a 10-min exposure of growing cells to MNNG at a final concentration of 10 μg/ml. This exposure condition is known to strongly induce the UVM but not the SOS response (17, 28). Exposure of IPTG-induced cells to MNNG should induce both SOS and UVM pathways.

The UVM response is defined by elevated mutagenesis at an ɛC lesion (borne on a transfected M13 ssDNA genome) in cells subjected to DNA damage. This response has been shown to be independent of recA umuD and umuC genes and to be elicited by all major classes of DNA-damaging agents. The results presented in this report show that mutation fixation at ɛC is not only elevated by the UVM response but also elevated independently by the SOS response. However, there is a difference in the specificity of mutations induced by the two pathways at ɛC. The SOS pathway induces C→T and C→A mutations in about equal proportions, with perhaps a slight excess of the C→T mutations. In contrast, UVM induction produces more C→A mutations than C→T mutations. When both pathways are induced, the total number of mutations is increased further, but a strong bias toward C→A mutations that is a characteristic of UVM is maintained. Further analysis reveals that the apparent UVM dominance consists of both additive and competing effects (Fig. 4C and D): C→A (ɛC:T) mutations are elevated threefold by the combined actions of SOS and UVM pathways, whereas SOS-induced C→T (ɛC:A) mutations induced are reduced twofold by UVM.

A new insight to emerge from the AP site investigation is the possibility that the UVM response may contribute to mutagenesis at some SOS-dependent lesions. The data plotted in Fig. 4 show that in SOS-induced cells there is a consistent further increase in mutagenesis at AP sites upon UVM induction. The manifestation of this additive effect appears to require the full expression of SOS functions. Thus, our results show that the UVM response can affect mutagenesis at both class 2 (ɛC) and at class 1 (AP site) lesions. Another interesting finding here is that cells appear to have a low but finite capacity for skip-bypass replication across AP sites such that targeted 1-bp deletions arise at a constant rate unaffected by the induction status of UVM or SOS responses. (A less likely possibility is that AP site oligonucleotide used in the construction of the ssDNA vector had a low-level contaminating sequence that was the equivalent of a −1 deletion). Increased base substitution mutagenesis at AP sites does not appear to occur at the expense of deletions in induced cells, as the data in Table 4 might suggest. Thus, most of the transfected AP-containing ssDNA apparently suffers replication arrest at the lesion site. A small but appreciable proportion of the arrested molecules are released into replication by a loop-out of the AP site so that the polymerase in effect skips across the lesion site, and this capability is constitutively expressed. The remainder of the arrested molecules simply perish in the absence of inducible factors.

It is interesting to compare our results with previous analyses of translesion DNA synthesis and mutagenesis at single AP sites borne on M13 ssDNA vectors (9, 10). Lawrence et al. (9) showed that very low levels of survival occurred in unirradiated cells, with survival increasing 10-fold in UV-irradiated SOS-proficient cells, results that are similar to ours (Table 4). Also, their observation that survival is not increased in SOS-defective cells even when subjected to DNA damage is consistent with the present work (Table 4, rows 9 and 10). However, by uncoupling SOS induction from DNA damage, we show here that DNA damage can increase translesion synthesis through an SOS-independent (UVM) pathway in SOS-induced cells (Table 4, rows 7 and 8). The mutation spectrum observed in two previous studies using a similar (but not identical) experimental system is also largely in agreement with the current data, in that predominantly A is inserted opposite AP sites, followed by other bases such as T and G (9, 10) in SOS-induced cells. However, our data show that a significant fraction of translesion synthesis in cells defective for SOS functions (SR400; Table 4, row 9) proceeds through a −1 nt deletion, whereas the corresponding data of Lawrence et al. (10) show that −1 nt deletions constitute a minor fraction. Two major differences in the two experimental systems may account for this difference. First, we used a chemically stable synthetic AP site, whereas Lawrence et al. (9, 10) used an AP site created by treating an oligonucleotide bearing a DNA uracil with uracil glycosylase; it is possible that the synthetic AP site renders the transfected ssDNA less susceptible to spontaneous cleavage or endonucleolytic inactivation and thus may allow more translesion DNA synthesis. A second difference is that Lawrence et al. (10) used a uvrA- and ΔumuDC-deficient E. coli strain as a host, whereas we used a uvrA+ ΔumuDC ΔrecA strain. Finally, it should be noted that in terms of absolute numbers, −1 nt deletions are a minor event in our experiments as well, as indicated in Fig. 4A and B.

According to current understanding, the SOS mutagenesis proteins UmuD′ UmuC and RecA act at the site of replication arrest to enable translesion DNA synthesis by DNA polymerase III holoenzyme. Recently, inactivation of the epsilon subunit (3′→5′ editing activity) of DNA polymerase III was shown to obviate the need for SOS induction for a subset of frameshift mutations induced by the carcinogen N-2-acetylaminofluorene (5). This finding has been interpreted to mean that induced SOS proteins are required only under specific circumstances to overcome the polymerase cycling forced by normal editing at lesion sites. We have noted previously that the preponderance of evidence indicates that the UVM response is mediated through an alteration of DNA replication (13). Mutation fixation at ɛC residues was reported to be increased in mutD5 cells to the same extent as in UV-irradiated cells (11). Even though the authors of the study did not so hypothesize (11), their finding is consistent with the possibility that the UVM effect is mediated by a factor that blocks polymerase editing. The genetic requirements of the UVM response (17) suggest that the factor must be recA independent and therefore distinct from the recA-inducible hypothetical Npf factor proposed by Fuchs and Napolitano (5). The results presented here further strengthen the notion that mutagenesis observed in SOS-induced cells (i.e., cells exposed to DNA-damaging treatments) results from the operation of multiple mutagenic pathways. The knowledge that some of these pathways may not be a part of the recA- and lexA-regulated classical SOS network should facilitate a better understanding of the mechanisms underlying transient mutator responses in E. coli.