Error-Prone Processing of Apurinic/Apyrimidinic (AP) Sites by PolX Underlies a Novel Mechanism That Promotes Adaptive Mutagenesis in Bacillus subtilis

Abstract

In growing cells, apurinic/apyrimidinic (AP) sites generated spontaneously or resulting from the enzymatic elimination of oxidized bases must be processed by AP endonucleases before they compromise cell integrity. Here, we investigated how AP sites and the processing of these noncoding lesions by the AP endonucleases Nfo, ExoA, and Nth contribute to the production of mutations (hisC952, metB5, and leuC427) in starved cells of the Bacillus subtilis YB955 strain. Interestingly, cells from this strain that were deficient for Nfo, ExoA, and Nth accumulated a greater amount of AP sites in the stationary phase than during exponential growth. Moreover, under growth-limiting conditions, the triple nfo exoA nth knockout strain significantly increased the amounts of adaptive his, met, and leu revertants produced by the B. subtilis YB955 parental strain. Of note, the number of stationary-phase-associated reversions in the his, met, and leu alleles produced by the nfo exoA nth strain was significantly decreased following disruption of polX. In contrast, during growth, the reversion rates in the three alleles tested were significantly increased in cells of the nfo exoA nth knockout strain deficient for polymerase X (PolX). Therefore, we postulate that adaptive mutations in B. subtilis can be generated through a novel mechanism mediated by error-prone processing of AP sites accumulated in the stationary phase by the PolX DNA polymerase.

INTRODUCTION

The genetic alterations that allow organisms to escape from growth-limiting conditions in response to natural or artificial selection during prolonged nonlethal selective pressure are referred to as adaptive or stationary-phase mutagenesis (1). This biological process, originally discovered in Escherichia coli (2, 3), was later found to occur in other prokaryotes (1, 4) as well as in some eukaryotes (5). The existence of adaptive mutagenesis was demonstrated in Bacillus subtilis by employing strain YB955, which allows measuring the reversion frequencies to chromosomal auxotrophies of hisC952 (TAG nonsense mutation), metB5 (TAA nonsense mutation), and leuC427 (missense mutation) (1).

It has been proposed that during periods of environmental stress, such as those occurring in the stationary phase of growth, a group of cells in a B. subtilis culture can be differentiated into a subpopulation with suppressed DNA repair systems in which adaptive mutations may be generated (1, 6). In agreement with this idea, it has been shown that the genetic inactivation of the mismatch (MMR) and guanine-oxidized (GO) systems potentiates the mutagenic events that occur in nongrowing B. subtilis cells (6, 7). Thus, it appears that the accumulation of mismatched and oxidized DNA bases in nongrowing B. subtilis cells is a key factor that promotes mutations under conditions of nutritional or metabolic stress (6, 7).

Reactive oxygen species (ROS) generated in cells either as byproducts of normal cellular metabolism or by exogenous agents have the potential to react with lipids, proteins, and DNA (8, 9). Accordingly, it has been shown that attack of DNA by ROS results in the formation of a myriad of oxidized bases, including uracil glycol and thymine glycol, 5-hydroxy-uracyl and 5-hydroxy-cytocine, and 8-oxo-adenine and 8-oxo-guanine (8-oxo-G), among others (10, 11). However, in addition to inducing the formation of oxidized bases, ROS may generate other types of genetic injuries, including formation of apurinic/apyrimidinic (AP) sites, damage to the deoxyribose sugar, and fragmentation of the DNA backbone, producing single-strand and/or double-strand DNA breaks (12). AP sites are among of the most frequently formed lesions in DNA, and they may arise spontaneously or following the catalytic action of specific DNA glycosylases that hydrolyze damaged bases from DNA; if these enzymes possess a lyase activity, a second catalytic event causes the rupture of the deoxyribose sugar, generating a single-strand break (13–15). AP sites and strand breaks are potentially mutagenic and toxic for cells; therefore, if left unrepaired, they affect replication, transcription, and cell survival (16, 17). The first step in the processing of AP sites is carried out by AP endonucleases, a group of enzymes which excise the DNA backbone at the 5′ end of the AP site. This cut generates 5′-phosphate deoxyribose and 3′-hydroxyl deoxyribose ends that are recognized by a DNA polymerase that is responsible for incorporating the appropriate nucleotide(s); finally, the DNA ligase seals the DNA patch (8). B. subtilis possesses Nfo and ExoA, two AP endonucleases that are frequently found in organisms of the three domains of life (17). In addition to processing AP sites, these proteins may process 3′-OH blocking lesions, including those that result from elimination of modified bases by glycosylases with associated lyase activity such as Nth. Interestingly, in comparison with a wild-type parental strain, B. subtilis cells lacking Nfo and ExoA are modestly more mutagenic and susceptible to oxidizing agents (18). These results suggest the existence of proteins that process AP sites and single-strand breaks in addition to Nfo and ExoA in B. subtilis. In fact, results from recent reports have demonstrated that the DNA glycosylase-AP lyase Nth and DNA polymerase X (PolX) have the ability to repair these types of DNA lesions (19, 20).

In this work, we investigated how the absence of Nfo, ExoA, Nth, and PolX, involved in processing AP sites and 3′ blocking lesions, affects the production of mutations in growing and starved B. subtilis cells.

Results revealed that (i) AP sites accumulate significantly during stationary phase, (ii) the combined deficiencies of nfo, exoA, and nth significantly increased the amount of reversions in the his, met, and leu alleles of strain B. subtilis YB955, and (iii) PolX is required to produce His⁺, Met⁺, and Leu⁺ reversions in the stationary phase but not during growth. Therefore, we propose that adaptive mutations can be generated by a novel mechanism that requires PolX-dependent error-prone processing of AP sites in starved B. subtilis cells.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

B. subtilis strains and plasmids used in this study are listed in Table 1. Liquid cultures of B. subtilis strains were routinely growth in Penassay broth (PAB) (antibiotic medium 3; Difco Laboratories, Sparks, MD). Growth was monitored with a Pharmacia Ultrospec 2000 spectrophotometer set at 600 nm. When required, erythromycin (Ery, 1 μg/ml), tetracycline (Tet, 10 μg/ml), neomycin (Neo, 12.5 μg/ml), spectinomycin (Sp, 100 μg/ml), chloramphenicol (Cm, 5 μg/ml), or isopropyl-β-d-thiogalactopyranoside (IPTG; 0.5 mM) was added to media. E. coli cultures were maintained on Luria-Bertani (LB) medium containing ampicillin (Amp, 100 μg/ml), chloramphenicol (Cm, 5 μg/ml), or kanamycin (Kan, 50 μg/ml) when required. Liquid cultures were incubated at 37°C with vigorous aeration.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Genotype and descriptiona	Reference, source, or transformationb
B. subtilis strains
YB955	hisC952 metB5 leuC427 xin-1 SpβSENS	1
SL7360	ΔrecA::neo; Neor	Patrick Piggot
PERM971	Δnth::ery; Eryr	pPERM768 → YB955
PERM1035	ΔexoA::tet nth::ery; Eryr Tetr	pPERM374 → PERM971
PERM1036	Δnfo::neo exoA::tet nth::ery; Neor Tetr Eryr	pPERM337 → PERM1035
PERM1067	Δnfo::neo; Neor	pPERM337 → YB955
PERM1068	Δnfo::neo exoA::tet; Neor Tetr	pPERM374 → PERM1067
PERM1137	Δnfo::neo exoA::tet nth::ery amyE::Phs-nth; Neor Tetr Eryr Spr	pPERM1096 → PERM1036
PERM1138	Δnfo::neo exoA::tet nth::ery amyE::Phs-empty; Neor Tetr Eryr Spr	pDR111 → PERM1036
PERM1165	ΔpolX::cm; Cmr	pPERM1152 → YB955
PERM1168	Δnfo::neo exoA::tet polX::cm; Neor Tetr Cmr	pPERM1152 → PERM1068
PERM1169	Δnfo::neo exoA::tet nth::ery polX::cm; Neor Tetr Eryr Cmr	pPERM1152 → PERM1036
PERM1238	amyE::PrecA-gfpmut3a; Spr	Laboratory stock
PERM1308	Δnfo::neo exoA::tet nth::ery amyE::PrecA-gfpmut3a; Neor Tetr Eryr Spr	pPERM1237 → PERM1036
Plasmids
pDR111	amyE::Phyper-spank promoter (Phs); Ampr Spr	David Rudner
pMUTIN4	Integrational lacZ fusion vector; Ampr Eryr	BGSCc
pPERM337	pUC18 containing the nfo gene interrupted by a Neor cassette; Ampr Neor	18
pPERM374	pDG1515 with 469-bp XbaI-BamHI and 537-bp EcoRI-HindIII PCR products containing the 5′ and 3′ regions of exoA, respectively; Ampr Tetr	18
pPERM768	pMUTIN4 containing an internal 306-bp HindIII-SalI fragment of the nth ORF; Ampr Eyrr	This study
pPERM1096	pDR111 containing a Phs-nth construct; Ampr Spr	This study
pPERM1149	pMUTIN4-cat; Cmr	This study
pPERM1152	pPERM1149 containing an internal 466-bp HindIII-BamHI fragment of the polX ORF; Cmr	This study
pPERM1237	pDR111 containing a PrecA-gfpmut3a construct; Ampr Spr	Laboratory stock

^aAmp, ampicillin; Ery, erythromycin; Neo, neomycin; Tet, tetracycline; Cm, chloramphenicol; Sp, spectinomycin; ORF, open reading frame.

^b“X” → “Y” indicates that “strain Y” was transformed with plasmid DNA from “source X.”

^cBGSC, Bacillus Genetic Stock Center.

Genetic and molecular biology techniques.

Preparation of competent E. coli or B. subtilis cells and their transformation with plasmid DNA were performed as described previously (21, 22). Chromosomal DNA from B. subtilis was purified as described by Cutting and Vander Horn (23). Large-scale preparation and purification of plasmid DNA were accomplished using commercial ion-exchange columns, following the instructions of the manufacturer (Qiagen, Inc., Valencia, CA). Small-scale preparation of plasmid DNA from E. coli cells, enzymatic manipulations, and agarose gel electrophoresis were performed by standard techniques (22). PCR products were obtained with homologous oligonucleotide primers and Vent DNA polymerase (New England BioLabs, Ipswich, MA).

Construction of mutant strains.

To obtain the Δnth mutant strain, a 306-bp HindIII-SalI internal fragment (from bp 199 to bp 505 downstream of the nth translational start codon) was amplified by PCR using chromosomal DNA from strain B. subtilis YB955 and the oligonucleotide primers 5′-GCAAGCTTGCTGTTCCGCTGGAAGAGC-3′ (forward) and 5′-GCGTCGACCCTTGCGCATCAGCGTCTT-3′ (reverse) (restriction sites are underlined). The PCR fragment was cloned between HindIII and SalI of pMUTIN4. The resulting construct, designated pPERM768, was used to transform B. subtilis strain YB955, generating strain PERM971 (Δnth). To obtain the triple Δnfo exoA nth mutant, plasmid pPERM374 was used to transform B. subtilis PERM971, generating strain PERM1035 (ΔexoA nth); this strain was then transformed with plasmid pPERM337, producing strain B. subtilis PERM1036 (Δnfo exoA nth). To obtain a double Δnfo exoA mutant, B. subtilis YB955 was transformed with plasmid pPERM337, generating strain B. subtilis PERM1067 (Δnfo); subsequently, competent cells of this strain were transformed with plasmid PERM374 to obtain strain B. subtilis PERM1068 (Δnfo exoA) (Table 1).

A construct to disrupt polX was created using the pMUTIN4-cat integrative vector. This vector was constructed by removing the complete erythromycin resistance cassette and part of the ampicillin resistance cassette from pMUTIN4 (24) by restriction with PstI. The cat (chloramphenicol acetyltransferase) gene was amplified by PCR using plasmid pBGSC6 as the template and oligonucleotide primers 5′-GCCTGCAGCGCTCATGAGACAATAACC-3′ (forward) and 5′-GCCTGCAGGTACAGTCGGCATTATCTC-3′ (reverse) that inserted PstI restriction sites (underlined) in both sides of the amplified fragment. The PCR fragment was cloned in the PstI site of pMUTIN4. The correct insertion of the cat gene was confirmed by restriction assay and PCR (data not shown). The resulting construct, designated pPERM1149 (pMUTIN4-cat), was propagated in E. coli DH5α cells. To interrupt polX, a 466-bp HindIII-BamHI internal fragment (from bp 465 to bp 930 downstream of the polX translational start codon) was PCR amplified using chromosomal DNA from B. subtilis YB955 as the template and the oligonucleotide primers 5′-GCAAGCTTCGGCTACGCCCTCCGGATC-3′ (forward) and 5′-GCGGATCCCGGCGGAATCAGCGGCAGG-3′ (reverse) (restriction sites are underlined). The PCR fragment was cloned between the HindIII and BamHI sites of pMUTIN4-cat. The resulting construct, designated pPERM1152, was used to transform B. subtilis strains YB955, PERM1068, and PERM1036, generating strains PERM1165 (ΔpolX), PERM1168 (Δnfo exoA polX), and PERM1169 (Δnfo exoA nth polX), respectively.

Construction of a plasmid to overexpress nth.

For complementation of the nth mutation, a copy of nth was placed ectopically at the amyE locus, under the control of the IPTG-inducible hyperspank promoter (P_hs). To this end, the nth allele and its Shine-Dalgarno sequence were amplified by PCR using genomic DNA of B. subtilis YB955 as the template and oligonucleotide primers 5′GCAAGCTTGCAGAATGAACCGCAA-3′ (forward) and 5′-GCGCTAGCTCATCGTTTCACCAGT-3′ (reverse) containing HindIII and NheI restriction sites, respectively (underlined). The PCR product was cloned between the HindIII and NheI sites of the pDR111 integrative vector (a generous gift of David Rudner), generating plasmid pPERM1096 (P_hs-nth). This plasmid was linearized and used to transform strain B. subtilis PERM1036, generating B. subtilis strain PERM1137. As an experimental control, the linearized empty pDR111 vector (P_hs-empty) was inserted into the amyE locus by transforming B. subtilis PERM1036 to generate strain PERM1138 (Table 1).

For all strains generated, the single- and double-crossover events leading to inactivation of the appropriate genes and in trans complementation were corroborated by PCR using the oligonucleotide primers described in Table 2.

TABLE 2.

Oligonucleotide primers used to corroborate the mutant strains of B. subtilis constructed in this study

Oligonucleotide	Sequence (5′ to 3′)	Description
F-nth	GCAAGCTTGTGTTAAATCTAAAAC	This forward primer extends from nt 4 to nt 21 downstream of the translational start codon of nth
F-polX	GCACGCTCGAATCGCTTA	This forward primer extends from nt 221 to nt 238 downstream of the translational start codon of polX
R-lacZ	GCAGCAACGAGACGTCAG	This reverse primer extends from nt 618 to nt 633 downstream of the translational start codon of lacZ
F-exoA	GCGGATCCATGAAGTTGATTTCATGG	This set of primers amplifies the complete ORF of exoA
R-exoA	GCCTCGAGTCATATATTGATGATAAG
F-nfo	GCGGATCCTTGCTGAGAATAGGCTC	This set of primers amplifies the complete ORF of nfo
R-nfo	GCCTCGAGTTATTGCTGTAAAATCT
F-amyE	GATCAAAAGCGGAACCATTCTTC	This set of primers amplifies a region of amyE encompassing nt 138 to nt 1980 downstream of the amyE translational start codon
R-amyE	TTGAGCTCAATGGGGAAGAGAACCGC

Determination of sensitivity to hydrogen peroxide.

B. subtilis strains were propagated at 37°C in PAB medium to an optical density at 600 nm (OD₆₀₀) of 1.0. Cell samples (2 ml) collected at this point were treated with H₂O₂ to reach a final concentration of 5 mM. H₂O₂-treated cultures, together with untreated cultures of each strain tested, were incubated at 37°C with shaking for an additional period of 1 h. Finally, the cultures (10 μl) were spot plated in 10-fold serial dilutions on solid LB medium; the plates were incubated overnight at 37°C and then photographed in a Gel Doc EZ Imaging system (Bio-Rad Laboratories, Hercules, CA). Furthermore, the 50% lethal doses (LD₅₀) for H₂O₂ were also determined for these strains. In brief, cell samples (2 ml) collected at an OD₆₀₀ of 1.0 were exposed to different concentrations of this oxidant chemical and the cultures were incubated for additional 1 h. Cell survival was measured by plating aliquots of serial dilutions on LB agar plates. CFU were counted after 16 h of incubation at 37°C.

Stationary-phase mutagenesis assays.

Cultures were grown in flasks containing PAB medium with vigorous aeration at 37°C until 90 min after the cessation of exponential growth. The stationary-phase mutagenesis assays were performed as previously described (1, 6) by plating cell aliquots (0.1 ml) on six plates of solid Spizizen minimal medium (SMM; 1× Spizizen salts supplemented with 0.5% glucose and either 50 μg/ml or 200 ng/ml of the required amino acid and 50 μg/ml each of isoleucine and glutamic acid). The concentration of the amino acid used depended on the reversion that was being selected. For instance, to select for His⁺ revertants, 50 μg/ml of methionine and leucine and 200 ng/ml of histidine were added to the medium. Isoleucine and glutamic acid were added as described previously (25) in order to protect the viability of the cells. When required, the selection medium was supplemented with IPTG (0.5 mM final concentration). The number of revertants from the six plates was scored daily. The initial number of bacteria for each experiment was determined by serial dilution of the bacterial cultures and then by plating the cells on a minimal medium containing all three essential amino acids. The experiments were performed at least three times.

The survival rates of the bacteria plated onto the minimal selective medium were determined as follows. Three agar plugs were removed from each selection plate every 2 days. The plugs were removed with sterile Pasteur pipettes and taken from areas of the plates where no growth was observed. The plugs were suspended in 1 ml of 1× Spizizen salts, diluted, and plated onto LB plates. Again, the number of colonies was determined following 16 h of growth at 37°C.

Analysis of mutation rates.

The growth-dependent reversion rates for the His⁺, Met⁺, and Leu⁺ revertants were measured by fluctuation tests with the Lea-Coulson formula as follows: r/m − ln(m) = 1.24 (26). Three parallel cultures were used to determine the total number of CFU plated onto each plate (N_t) by titration. The mutation rates were calculated as previously described using the following formula: mutation rate = m/2N_t (1, 6).

Detection of AP sites in chromosomal DNA.

Strains were propagated at 37°C in PAB, and samples (6 ml) were collected during exponential growth (OD₆₀₀ = 1.0) and 90 min after the cessation of exponential growth. Cell samples were collected by centrifugation at 18,200 × g for 1 min and washed 1× with 0.05 M EDTA–0.1 M NaCl (pH 7.5), and the cell pellets were stored at −20°C. Chromosomal DNA was purified from the cells according to the protocol of Cutting and Vander Horn (23). To detect AP sites, DNA was digested with 18 U of endonuclease IV (EndoIV) (New England, BioLabs). To detect single-strand DNA breaks generated by EndoIV cleavage at AP sites, enzymatically digested DNA samples were denatured with 0.3 N NaOH and electrophoresed through a 0.8% alkaline agarose gel (22). The gels were stained with ethidium bromide and photographed using a Gene-Genius BioImaging system (Syngene, Frederick, MD); digital photographic images of the gels were scanned, and DNA was quantified by densitometry using ImageJ 1.47n software (http://imagej.nih.gov/ij/). The intensity of the chromosomal DNA band remaining in the gel well after EndoIV treatment was determined and compared to the intensity of the chromosomal band of the untreated control reaction.

Statistical analysis.

For determination of hydrogen peroxide sensitivity, differences between strains were calculated by performing one-way analysis of variance (ANOVA) followed by a Tukey's post hoc analysis. Significance was set at P < 0.05. In the case of the determination of AP sites in the chromosomal DNA and in the quantification of fluorescence by spectrometry, the statistical significance was determined with unpaired t tests. Analyses were performed with GraphPad Prism software, version 6.3.

RESULTS

Nth confers protection from hydrogen peroxide to B. subtilis cells deficient for Nfo and ExoA.

Reactive oxygen species, generated as byproducts of the aerobic metabolism, are able to attack DNA, producing different types of lesions, including 8-oxo-G and AP sites (9). AP sites are potentially mutagenic and highly genotoxic for cells; therefore, these lesions must be removed through the base excision repair (BER) pathway that requires strict participation of a group of specialized enzymes termed AP endonucleases (16, 17). B. subtilis possesses the conserved AP endonucleases Nfo and ExoA (18, 27–29); however, in reference to the wild-type parental strain, the absence of these proteins slightly increased both the susceptibility of B. subtilis to hydrogen peroxide and the mutation frequency in this microorganism (18, 27). It has recently been shown that Nth, a DNA repair protein of the MutY/Nth family, has the ability to process AP sites in B. subtilis (20). This protein confers protection against the toxic effects of hydrogen peroxide to B. subtilis cells and is involved in mutagenesis in E. coli (20, 30, 31). Therefore, we determined whether Nth is responsible for conferring protection against oxidative stress to B. subtilis cells deficient for Nfo and ExoA. We found that the sole absence of Nth increased the susceptibility of B. subtilis to hydrogen peroxide to levels similar to those observed in the Δnfo exoA mutant (Fig. 1) (20). Interestingly, disruption of nth in the Nfo ExoA-deficient strain induced a significant increase in the susceptibility of B. subtilis cells to this oxidizing agent (Fig. 1). Together, these results support the hypothesis that in the absence of Nfo and ExoA, Nth plays an important role in counteracting the genotoxic effects of H₂O₂ in growing B. subtilis cells.

FIG 1.

Effects of nth, polX, nfo, and exoA mutations on the resistance of growing cells to H₂O₂. Exponentially growing cells (OD₆₀₀ = 1.0) of different strains were treated with H₂O₂ (5 mM, final concentration). Cultures that were treated (+) or not treated (−) were incubated for an additional hour before 10-fold serial dilutions were spotted onto LB plates. Two independent experiments were performed and yielded similar results. (Bottom) The LD₅₀ for H₂O₂ was determined for each strain as described in Materials and Methods, and values were expressed as averages ± standard deviations (SD) of the results from three independent experiments. Superscripts a, b, and c indicate statistically significant differences between strains as determined by one-way ANOVA followed by a Tukey's post hoc test; P < 0.05.

Stationary-phase mutagenesis in B. subtilis cells deficient for the AP endonucleases Nfo, ExoA, and Nth.

We next investigated the role played by Nth, Nfo, and ExoA in stationary-phase-associated mutagenesis of B. subtilis. To test this, the nth, nfo, and exoA alleles were disrupted in strain B. subtilis YB955, a model system widely employed to understand how mutations are generated in amino-acid-starved cells (7, 32). This strain is auxotrophic for three amino acids due to the chromosomal mutations hisC952 (amber), metB5 (ochre), and leuC427 (missense) (25). Analysis of frequencies of reversion to his, met, and leu in cell cultures that were starved for each of these amino acids revealed that Nfo and ExoA made a minor contribution to adaptive mutagenesis compared to that of Nth. As shown in Fig. 2, using as a reference the parental YB955 strain, the single absence of Nth increased the number of adaptive His⁺, Met⁺, and Leu⁺ revertants ∼2-, 5-, and 3.5-fold, respectively. In contrast, the rates of reversion to his, met, and leu in the Δnfo exoA strain were 1.5, 3, and 1.7 times higher than those calculated for the YB955 parental strain (Fig. 2). Interestingly, the strain that was deficient for Nfo, ExoA, and Nth showed a strong propensity to produce large amounts of revertants for the three alleles tested compared to the YB955 parental strain and its Δnth- and Δnfo exoA-derived mutant strains (Fig. 2). Together, these results indicate that deficiencies in DNA repair proteins that process AP sites and single-strand breaks promote adaptive mutagenesis in nongrowing B. subtilis cells. In agreement with this hypothesis, the production of His⁺, Met⁺, and Leu⁺ colonies by the strain deficient for Nfo, ExoA, and Nth was restored to the level seen with the Δnfo exoA strain following expression of nth from the IPTG-inducible P_hs promoter (Table 1 and Fig. 3). Of note, this effect was not observed in a control strain harboring the same construct but lacking the nth gene downstream of the IPTG-inducible P_hs promoter (Table 1 and Fig. 3). Furthermore, the survival rates, during the time course of the adaptive mutagenesis experiments, were determined for the four strains tested. As shown in Fig. S1 in the supplemental material, the Δnth, Δnfo exoA, and Δnfo exoA nth mutants had survival rates similar to those of the parental YB955 strain in cultures that were starved for each amino acid. Therefore, the increase in the numbers of revertant His⁺, Met⁺, and Leu⁺ colonies observed in the three mutant strains cannot be attributed to differential rates of growth or survival with respect to the parental strain.

FIG 2.

Frequencies of stationary-phase reversions for his (A), met (B), and leu (C) mutant alleles of B. subtilis strains YB955 (parental strain [○]), PERM971 (Δnth [●]), PERM1068 (Δnfo exoA [◇]), and PERM1036 (Δnfo exoA nth [◆]) as described in Materials and Methods. Data represent counts from six plates averaged from three separate tests normalized to initial cell titers ± SD.

FIG 3.

Frequencies of stationary-phase reversions for his (A), met (B), and leu (C) mutant alleles of B. subtilis strains YB955 (parental strain [○]), PERM1036 (Δnfo exoA nth [●]), PERM1068 (Δnfo exoA [△]), PERM1137 (Δnfo exoA nth amyE::P_hs-nth [◆]), and PERM1138 (Δnfo exoA nth amyE::P_hs-empty [◇]) as described in Materials and Methods. Data represent counts from six plates averaged from three separate tests normalized to initial cell titers ± SD.

As previously described, both the his and met auxotrophies in strain YB955 are due to nonsense mutations, and it was reported that more than 90% of Met⁺ revertants arise due to the generation of a nonsense suppressor mutation, while ∼20% of the His⁺ revertants are the result of this type of mutation (1); therefore, we decided to investigate what types of mutations (nonsense suppressors versus true revertants) gave rise to the His⁺, Met⁺, and Leu⁺ revertants in the strains lacking Nth, Nfo, and ExoA. The results of this analysis revealed that around 60% of the colonies with a His⁺ phenotype were also revertants in the met allele; in contrast, around 99% of the initially isolated Met⁺ colonies were also His⁺ revertants (Table 3). These results strongly suggest that most of the His⁺ and Met⁺ revertants were generated by suppressor mutations presumably occurring in tRNAs as result of oxidative-stress-induced DNA damage accumulated in the genome of nutritionally stressed cells (7). In contrast, the totality of the Leu⁺ colonies, in starved cells, did not possess a His⁺ or Met⁺ phenotype; therefore, the reversions in the leu allele produced by the strain deficient for Nfo, ExoA, and Nth were most probably generated by a mutagenic event in this allele (Table 3).

TABLE 3.

Growth of stationary-phase revertants on alternative selective media^a

Revertant	No. of revertants that grew/no. tested (% revertants that grew)
	Leu− medium	Met− medium	His− medium
PERM1036 (Δnfo exoA nth)
His+	0/100 (0)	63/100 (63)	100/100 (100)
Met+	0/100 (0)	100/100 (100)	99/100 (99)
Leu+	40/40 (100)	0/40 (0)	0/40 (0)

^aMet⁺, His⁺, and Leu⁺ revertant colonies from day 7 were tested on 1× SMM missing one required amino acid (His, Met, or Leu) to screen for suppressor mutations. Plates were scored after 48 h of incubation.

Growth-dependent his, met, and leu reversions in AP endonuclease-deficient strains.

The growth-dependent mutation rates of the Δnth, Δnfo exoA, and Δnfo exoA nth strains for the generation of His⁺, Met⁺, and Leu⁺ colonies were also determined and compared with those of the YB955 parental strain. The results shown in Fig. 4 revealed that disruption of nth slightly increased the generation of His⁺ (∼1.5 times), Met⁺ (2 times), and Leu⁺ (∼2.5 times) revertants compared to the YB955 parental strain results. The absence of Nfo and ExoA did not significantly promote mutagenesis in growing B. subtilis cells; thus, no significant differences between the Δnfo exoA strain and the YB955 parental strain in the levels of reversion of his, met, and leu alleles were found (Fig. 4). However, in comparison with parental strain YB955, significant increases in the numbers of His⁺ (2.5-fold), Met⁺ (8-fold), and Leu⁺ (10-fold) revertants were observed in the strain deficient for Nfo, ExoA, and Nth (Fig. 4). Therefore, Nth not only confers protection from the noxious effects of hydrogen peroxide to growing cells but also suppresses the mutagenic effects of intracellular factors that promote the generation of AP sites.

FIG 4.

Analysis of exponential mutation rates. B. subtilis strains YB955 (parental strain), PERM971 (Δnth), PERM1068 (Δnfo exoA), PERM1036 (Δnfo exoA nth), PERM1137 (Δnfo exoA nth amyE::P_hs-nth), and PERM1169 (Δnfo exoA nth polX) were tested for their ability to produce His⁺ (white bars), Met⁺ (gray bars), and Leu⁺ (black bars) revertants during exponential growth as described in Materials and Methods. The mutation rates were calculated with the formula m/2Nt as previously described (1, 6, 26). Results show the average mutation rates from two individual fluctuation tests; error bars represent the SD from two independent experiments.

AP sites promote stationary-phase mutagenesis in B. subtilis.

The results presented in Fig. 2 and 3 suggest that in cells facing nutritional stress, the lack of Nfo and ExoA and especially of Nth increases the production of mutations detected by the reversions to his, met, and leu alleles. Therefore, we investigated whether an exacerbated accumulation of AP sites present in DNA of the strain lacking Nfo, ExoA, and Nth correlated with the mutagenic phenotypes observed in growing and starved B. subtilis cells. To this end, chromosomal DNA samples isolated from strains YB955 and Δnfo exoA nth during the exponential and stationary phases of growth were treated with AP endonuclease EndoIV and separated by alkaline gel electrophoresis (AGE) to detect AP sites (22). The results of this analysis revealed that both the YB955 and Δnfo exoA nth strains accumulated a significant amount of AP sites during the stationary phase compared to the amount generated during the exponential phase of growth (compare Fig. 5A and C). However, as shown in Fig. 5A and B, a major accumulation of AP sites was detected in the genomic DNA of the Δnfo exoA nth mutant compared to the level in the chromosomal DNA of the YB955 parental strain during the stationary phase. Furthermore, the AGE analysis performed with DNA samples obtained from growing cells treated with EndoIV did not show significant differences between the parental strain and the Nfo ExoA Nth-deficient strains (Fig. 5C and D). Taken together, these results strongly suggest that AP sites are a major source of lesions involved in generating mutations in starved B. subtilis cells.

FIG 5.

Alkaline agarose gel electrophoresis analysis of DNA isolated from stationary-phase cells (A and B) or exponential-growth-phase cells (C and D) of strain YB955 and strain YB955 nfo exoA nth with or without treatment with EndoIV. Data represent chromosomal DNA from stationary-phase (A) or exponential-growth-phase (B) strain YB955 (lanes 1 and 2) or strain YB955 Δnfo exoA nth (lanes 3 and 4) cells that were left untreated (lanes 1 and 3) or were treated with 18 units of EndoIV (lanes 2 and 4), all as described in Materials and Methods. Reaction samples were electrophoresed on a 0.8% alkaline agarose gel that was then stained with ethidium bromide as described in Materials and Methods. The results shown in panels A and C are representative of the results of two independent experiments. The results shown in panels B and D correspond to the quantification of chromosomal DNA degradation determined by densitometry using ImageJ 1.47n software; values are averages ± SD of the results of duplicate determinations in two separate experiments (with different batches of cultures). Statistical significance (P value) was determined with an unpaired t test.

Stationary-phase mutations produced by the Δnfo exoA nth strain are dependent on PolX.

It has recently been reported that DNA polymerase X (PolX) could act as a support mechanism for the BER pathway as it possesses three activities usually associated with different proteins that belong to this repair system, including AP endonuclease, 3′→5′ exonuclease, and DNA polymerase (19). Indeed, our results revealed that a strain lacking PolX was more susceptible to hydrogen peroxide than the parental YB955 strain; disruption of this gene significantly increased the susceptibility of the Δnfo exoA strain to this oxidizing agent (Fig. 1). Importantly, during growth, the absence of PolX induced a significant increase in the levels of his, met, and leu reversion of the Δnfo exoA nth strain (Fig. 4).

After confirming that polX counteracts mutagenesis in growing B. subtilis cells, we next investigated whether the absence of this gene enhances the production of the his, met, and leu reversions generated by the Δnfo exoA nth strain in amino-acid-starved cultures. Strikingly, our results revealed that the genetic inactivation of PolX in the Δnfo exoA nth genetic background significantly reduced the production of His⁺, Met⁺, and Leu⁺ revertants (Fig. 6). Of note, the absence of PolX almost completely suppressed the production of Leu⁺ revertants generated by the Δnfo exoA nth strain, as it showed levels of reversion to the leu allele similar to those exhibited by the YB955 parental strain (Fig. 6). Furthermore, the number of His⁺ and Met⁺ revertants produced by the Δnfo exoA nth strain was diminished by around 50% in the strain deficient for Nfo, ExoA, Nth, and PolX (Fig. 6). We confirmed that during the course of these experiments, there were no significant differences in the survival rates of the Δnfo exoA nth polX quadruple-knockout strain (see Fig. S1 in the supplemental material). These results strongly suggest that PolX is capable of processing the AP sites accumulated in amino-acid-starved cells deficient for Nfo, ExoA, and Nth in an error-prone manner, thus promoting reversions in the three alleles of this strain.

FIG 6.

Frequencies of stationary-phase reversions for his (A), met (B), and leu (C) mutant alleles of B. subtilis strains YB955 (parental strain [◇]), PERM1036 (Δnfo exoA nth [○]), and PERM1169 (Δnfo exoA nth polX [●]) as described in Materials and Methods. Data represent counts from six plates averaged from three separate tests and normalized to initial cell titers ± SD.

DISCUSSION

B. subtilis possesses the exoA and nfo genes encoding proteins with AP endonuclease activity (33; GenoList/SubtiList from Institut Pasteur Genopole). These proteins conserve a high level of homology with ExoIII (xth) and EndoIV (nfo) of E. coli, respectively (27–29). In contrast to E. coli results (34), the loss of ExoA and Nfo did not dramatically decrease the resistance of B. subtilis cells to oxidizing agents (H₂O₂ and t-BHP [tert-butyl hydroperoxide]) or significantly increase spontaneous mutagenesis (18). This result suggested that an alternative protein(s) with AP endonuclease activity compensates for the absence of Nfo and ExoA in growing B. subtilis cells. To commence to investigate this point, we initially considered the nth gene, which encodes a protein with 46.2% identity to the endonuclease III of E. coli (EndoIII_Ec) (12, 35). This enzyme, which processes a wide variety of oxidized bases, including 8-oxo-G (36–39), was also found capable of recognizing and processing AP sites generated in vitro (40). In this work, instead of using halos of growth inhibition (18), we measured the susceptibility of strains lacking Nfo, ExoA and/or Nth to H₂O₂ by determining the LD₅₀ for H₂O₂ and by spotting serial dilutions of cultures treated with this agent or left untreated. Following this approach, we found that, in comparison with the YB955 parental strain, cells of the Δnfo exoA mutant exhibited a significant increase in susceptibility to H₂O₂ treatment. Importantly, we also found that the genetic disruption of Nth was enough to sensitize B. subtilis cells to hydrogen peroxide. Notably, in comparison with the Δnfo exoA and Δnth mutants, a strain lacking these three functions (namely, those associated with Nfo, ExoA, and Nth) significantly increased its susceptibility to this oxidizing agent. On the basis of these results as well as those of a previous study showing that Nth possesses AP endonuclease activity (20), we propose that in growing B. subtilis cells, Nth plays a more major role than Nfo and ExoA in counteracting the DNA-damaging effects of oxidative stress.

It has been proposed that oxidative stress is a crucial factor that promotes stationary-phase-associated mutagenesis in strain B. subtilis YB955 (his, met, leu). Accordingly, starved cells of this strain, lacking a functional guanine-oxidized (GO) DNA system, had a strong propensity to generate His⁺ and Met⁺ revertants (7). However, as noted above, in addition to promoting the synthesis of 8-oxo-G, oxidative stress is a potential source of AP sites and these lesions, if left unrepaired, may also promote mutagenesis (8, 9, 12). Therefore, we initially investigated how a lack of proteins that repair AP sites in nongrowing B. subtilis cells affects adaptive mutagenesis. It was found that, although the absence of the typical AP endonucleases Nfo and ExoA significantly increased the number of reversion events occurring in the his, met, and leu alleles of strain YB955, the single defect in Nth had a stronger effect on this parameter; it is noteworthy that the levels of reversion of the three alleles tested were even higher when Nth was also absent from the Δnfo exoA mutant strain. These results strongly suggest that accumulation of AP sites is an important factor that promotes mutagenesis in metabolically stressed B. subtilis cells. It was evident in these experiments that the higher levels of reversion observed in the Δnfo exoA nth triple mutant were mainly attributable to the absence of Nth and to a lesser degree to the absence of the two AP endonucleases Nfo and ExoA. In support of this concept, the overexpression of nth from an IPTG-inducible promoter induced a significant reduction in the number of revertants produced by the Δnfo exoA nth strain (Fig. 3). We speculate that the multifunctional properties of Nth are responsible for this effect, as this enzyme has the ability to hydrolyze the glycosidic linkage of several oxidized bases, including 8-oxo-G, as well as to promote the repair of the generated AP sites in situ (20, 36–39, 41, 42). Indeed, the ability of enzymes with AP endonuclease activity such as Nth to keep tightly bound to AP sites is advantageous with respect to protecting the sugar in its open aldehyde form from reacting with proteins and/or lipids that may result in blocks for DNA replication (12). It has also been shown that bound AP endonucleases can stimulate the recruitment of the DNA polymerases that fill the generated gap, making the repair of these lesions more efficient (12, 20).

Given that ExoA and Nfo are typical AP endonucleases and that Nth also displays this activity (20), we suspected that the mutagenic phenotype exhibited by the Δnfo exoA nth triple-mutant strain was most probably caused by an exacerbated accumulation of AP sites. AGE analysis of chromosomal DNAs treated with E. coli EndoIV (Nfo) confirmed our hypothesis, as cells of the triple-knockout strain accumulated a significantly larger number of AP sites during the stationary phase than during the exponential phase of growth. It was also found that, with respect to the parental strain, actively growing cells of the Δnfo exoA nth mutant generated a major amount of His⁺, Met⁺, and Leu⁺ revertants despite the two strains containing similar numbers of AP sites. Thus, processing of AP sites to permit growth may at least in part proceed in an error-prone manner; in support of this contention, experimental evidence has revealed that a functional coupling between the A- and Y-family polymerases promotes translesion synthesis (TLS) during B. subtilis growth (43). Therefore, although AP sites are a source for mutations during the exponential and stationary phases of growth, our results suggest that these lesions make a major contribution to mutagenesis in nutritionally stressed cells.

The construction of an E. coli strain deficient for Xth, Nfo, and Nth has previously been reported (34). Although this mutant did not display an increased susceptibility to oxidizing agents, such as H₂O₂, compared to a Δnfo xth double mutant (34), this strain seems to possess a stronger mutagenic phenotype than its counterpart in B. subtilis, yielding levels of reversion to the argE3 allele ∼40 times superior to those exhibited by the wild-type parental strain (44). Interestingly, the Δnfo exoA nth B. subtilis mutant presented significantly lower levels of reversion of the his, met, and leu alleles that were ∼2, 7, and 9 times higher than those calculated for the parental YB955 strain, respectively. Notably, the Δxth nfo nth E. coli mutant also activated the SOS response putatively due to accumulation of unrepaired AP sites (44). We found no evidence of similar behavior in B. subtilis since the Δnfo exoA nth B. subtilis mutant was unable to spontaneously induce the expression of a P_recA-gfpmut3a fusion or generate filamented cells during growth (Table 1; see also Fig. S2 and S3 in the supplemental material). These observations strongly suggest that AP sites could still be processed during growth in the strain lacking Nfo, ExoA, and Nth. Additional proteins that could back up the functions of Nfo, ExoA, and Nth in B. subtilis are PolX and YwqL, as both enzymes have been revealed to possess AP endonuclease activity in this microorganism (19, 45; V. M. Ayala-García and M. Pedraza-Reyes, unpublished results). Regarding PolX, in vitro studies with the purified protein showed that this DNA polymerase has all the enzymatic activities necessary to repair AP sites, including those corresponding to AP endonuclease, 3′→5′ exonuclease, and DNA synthesis (19). Following a genetic approach, our results revealed that PolX did provide growing B. subtilis cells with an alternative pathway to process AP sites; thus, disruption of polX not only increased the levels of reversion of the his, met, and leu alleles but also made the Δnfo exoA nth strain more sensitive to hydrogen peroxide (Fig. 1 and 4).

Considering the antimutagenic role shown by PolX during growth, we anticipated a similar effect of the absence of this polymerase in the mutagenesis experiments performed with nutritionally stressed B. subtilis cells. Surprisingly, it was found that the number of adaptive His⁺, Met⁺, and Leu⁺ revertants produced by the Δnfo exoA nth polX quadruple-knockout strain showed a significant decline with respect to the levels of reversion calculated for the Δnfo exoA nth mutant, which contained an intact polX gene. Remarkably, as mentioned above, an opposite result was found in growing cells; namely, disruption of polX did significantly increase the reversion frequencies of the three alleles tested in the Δnfo exoA nth strain. Therefore, it is tempting to speculate that the accumulation of AP sites in nongrowing cells that overwhelms the capacity of Nfo, ExoA and/or Nth promotes adaptive mutations following a mechanism that involves PolX processing of AP sites in an error-prone manner. The ability of PolX to process AP sites may depend on the metallic cofactor bound to it and may in turn cause mutations only in starved B. subtilis cells (19). Indeed, results of in vitro assays have shown that the presence of Mg²⁺ promoted the addition of a single nucleotide to the 3′ OH generated by this enzyme; in contrast, Mn²⁺ stimulated 3′→5′ exonuclease activity encoded in PolX, followed by the synthesis of a longer DNA patch (19). Although it can be expected that the synthesis of longer patches can correlate with a higher probability of base misinsertions, it remains to be investigated whether the gap-filling properties of the Mn²⁺-PolX form are connected with its ability to process AP sites in an error-prone manner exclusively in nongrowing B. subtilis cells. However, it is a fact that PolX proteins of different origins are less accurate during filling of DNA gaps than typical replicative and repair polymerases (46, 47). For instance, a recent report showed that a human lymphoblastoid cell line that overexpressed the PolX-family Polβ protein, compared to normal control cells, showed an increased frequency of 1-nucleotide (nt) deletions (48).

Although our results showed that the majority of the Leu⁺ revertants produced by the Δnfo exoA nth mutant were fully dependent on the presence of PolX, not all of the His⁺ and Met⁺ revertants generated by the Δnfo exoA nth strain were dependent on this polymerase (Fig. 6). Therefore, it remains to be investigated whether alternative DNA replicases, including error-prone polymerases YqjH and/or YqjW, are involved in the generation of adaptive His⁺ and Met⁺ revertants promoted by the AP sites.

Finally, to our knowledge, this is the first report demonstrating how the error-prone repair of AP sites present in the genome of nonreplicating B. subtilis cells promotes adaptive mutagenesis. However, in nonreplicating cells, AP sites are also the source of transcriptional mutagenesis (TM). In this respect, Clauson et al. (49) reported that nonreplicating cells of a Δnfo exoA E. coli strain increased the levels of TM in response to the presence of AP sites, single-strand breaks, and oxidized bases such as 8-oxo-G. In B. subtilis, a variant of this process has been shown to be involved in adaptive mutagenesis and has been termed transcription-associated mutagenesis (50, 51). In this work, we found that the totality of Leu⁺ revertants as well as a significant fraction (∼40%) of the His⁺ revertants produced by the Δnfo exoA nth strain were not generated by suppressor mutations (Table 3); therefore, is not unreasonable to speculate that transcriptional errors followed by retromutagenesis could be also involved in generating adaptive Leu⁺ and Met⁺ revertants.