Role of Ribonucleotide Reductase in Bacillus subtilis Stress-Associated Mutagenesis

ABSTRACT

The Gram-positive microorganism Bacillus subtilis relies on a single class Ib ribonucleotide reductase (RNR) to generate 2′-deoxyribonucleotides (dNDPs) for DNA replication and repair. In this work, we investigated the influence of RNR levels on B. subtilis stationary-phase-associated mutagenesis (SPM). Since RNR is essential in this bacterium, we engineered a conditional mutant of strain B. subtilis YB955 (hisC952 metB5 leu427) in which expression of the nrdEF operon was modulated by isopropyl-β-d-thiogalactopyranoside (IPTG). Moreover, genetic inactivation of ytcG, predicted to encode a repressor (NrdR) of nrdEF in this strain, dramatically increased the expression levels of a transcriptional nrdE-lacZ fusion. The frequencies of mutations conferring amino acid prototrophy in three genes were measured in cultures under conditions that repressed or induced RNR-encoding genes. The results revealed that RNR was necessary for SPM and overexpression of nrdEF promoted growth-dependent mutagenesis and SPM. We also found that nrdEF expression was induced by H₂O₂ and such induction was dependent on the master regulator PerR. These observations strongly suggest that the metabolic conditions operating in starved B. subtilis cells increase the levels of RNR, which have a direct impact on SPM.

IMPORTANCE Results presented in this study support the concept that the adverse metabolic conditions prevailing in nutritionally stressed bacteria activate an oxidative stress response that disturbs ribonucleotide reductase (RNR) levels. Such an alteration of RNR levels promotes mutagenic events that allow Bacillus subtilis to escape from growth-limited conditions.

INTRODUCTION

Ribonucleotide reductases (RNRs) are essential enzymes present in all life forms and are responsible for the conversion of ribonucleotides to 2′-deoxyribonucleotides, the precursors for the replication and repair of DNA. Three different RNR classes (classes I, II, and III) differing in the metal cofactor required for their catalytic activity, allosteric regulation, and quaternary structure have been described (1, 2). Class I RNRs are composed of two homodimeric proteins, α₂ and β₂; the larger subunit, α, contains the catalytic and the allosteric site that controls the specificity of reduction, while the smaller subunit, β, contains a stable tyrosyl radical and a dinuclear metal center (3, 4). In Bacillus subtilis, the nrdE and nrdF genes encode the α and β subunits of the class Ib RNR (3, 4). Although the activity of the RNRs is controlled mainly by allosteric regulation in Escherichia coli and other bacteria, the intracellular levels of this enzyme can also be regulated by different transcriptional factors, including DnaA, Fur, and NrdR. Such transcription factors respond to changes in metabolic or environmental conditions and mediate the repression or induction of RNR-encoding genes (2, 3, 5–7). The transcriptional repressor NrdR controls the expression of genes encoding the three RNR classes in virtually all bacteria studied, although it exerts a stronger effect over genes encoding class Ib proteins (7, 8).

The genetic alterations that occur in response to a persistent nonlethal selective pressure that allows organisms to escape from growth-limiting conditions are known as “adaptive mutagenesis” or “stationary-phase-associated mutagenesis” (SPM) (9–11). In Bacillus subtilis, strain YB955, bearing the chromosomal auxotrophies encoded by hisC952, metB5, and leuC427, has been successfully employed as a well-validated system to explore the mechanisms that promote adaptive mutations in soil bacteria (11). It has been suggested that under growth-limiting conditions, a subpopulation of cells in a culture may experience higher levels of DNA damage, activating error-prone repair events that promote adaptive mutagenesis (12–16). Several reports of studies with B. subtilis support this hypothesis; thus, B. subtilis cells lacking functional guanine-oxidized (GO) or mismatch repair (MMR) systems are highly prone to the accumulation of stationary-phase-associated mutations (17, 18). In nongrowing B. subtilis cells, saturation of the MMR system promotes mutagenesis through a mechanism that involves the glycosylase MutY and processing of A-C or A-G mispairs (15). Moreover, repair of deaminated cytosine in B. subtilis by the uracil DNA glycosylase Ung and the endonuclease V YwqL occurs in an error-prone manner, promoting adaptive mutagenesis (19). A recent report revealed that the numbers of spontaneously or enzymatically generated apurinic/apyrimidinic (AP) sites increase during the stationary phase in B. subtilis and that such lesions can be processed in an error-prone manner by the low-fidelity (LF) DNA polymerase PolX (20). Other reports have documented the participation of LF DNA polymerases, including YqjH (a member of the Y DNA polymerase family), in generating mutations in nutritionally stressed B. subtilis cells (21).

Compelling evidence has revealed that cellular levels of deoxynucleoside triphosphates (dNTPs) can affect the fidelity of replication. In growing E. coli cells, overexpression of RNR genes and the subsequent increase in dNTP pools not only stimulated the activity of translesion synthesis (TLS) DNA polymerases but also transiently modified the proofreading functions of replicative polymerases (22). A similar effect was observed in actively growing cells of Saccharomyces cerevisiae, where increased dNTP levels promoted mutagenic events by conferring on DNA polymerases the capacity to extend over DNA mismatches (23, 24). Therefore, it is clear that changes in RNR expression and subsequent dNTP levels affect mutagenesis in actively growing cells. However, little is currently known about how these processes impact the genetic changes occurring in nutritionally stressed nongrowing bacteria.

In the present work, the impact of RNR levels on growth-associated mutagenesis and SPM in B. subtilis YB955 (hisC952 metB5 leuC427) was investigated. Because the class Ib RNR encoded by the nrdEF genes is essential for growth (3, 4, 25), we used a conditional mutant to modulate nrdEF expression in growth-limited B. subtilis cells. Genetic disruption of ytcG, predicted to encode the NrdR repressor in B. subtilis, resulted in a strain that overexpressed the nrdEF operon. Analysis of hisC952, metB5, and leuC427 reversion rates revealed that RNR-encoding genes are necessary for the production of adaptive His⁺, Met⁺, and Leu⁺ revertants. Notably, overexpression of nrdEF had a positive impact on the production of prototrophic His⁺, Met⁺, and Leu⁺ colonies during growth as well as on nutritionally stressed, nongrowing B. subtilis cells. Finally, we found that, in addition to the negative regulation exerted by NrdR, the expression of the nrdEF operon was affected by the oxidative stress response regulator PerR. In summary, our results support the concept that RNR levels are disturbed by the adverse metabolic conditions prevailing in nutritionally stressed cells and have a direct impact on B. subtilis SPM.

RESULTS

Construction and characterization of a conditional nrdEF mutant of B. subtilis YB955.

B. subtilis YB955 (hisC952 metB5 leuC427) possesses a single essential class Ib RNR whose subunits are encoded by the nrdEF operon (4); therefore, one would expect NrdEF to be the only protein responsible for generating 2′-deoxyribonucleotides for DNA replication and repair in this strain (3, 4). To investigate whether fluctuations in NrdEF levels impact mutagenesis in B. subtilis cells, we constructed a YB955 mutant in which the expression of nrdEF was conditioned by the presence of isopropyl-β-d-thiogalactopyranoside (IPTG) in the culture medium (Fig. 1). Of note, since the product of nrdI, the cistron preceding nrdE-nrdF, has been shown to be necessary for full RNR activity in other bacteria (26–28), expression of nrdI in the conditional nrdE-nrdF mutant was left under the control of the native regulatory regions of the nrdI-nrdE-nrdF operon (Fig. 1). Therefore, under inducing conditions, the conditional RNR mutant strain expresses nrdE-nrdF as well as nrdI for full RNR activity. As shown in Fig. 2A, as assessed by determination of the optical density at 600 nm (OD₆₀₀) and viable counts, the absence of IPTG in the culture medium abrogated the growth of this conditional mutant (PERM1017), whereas supplementation with ≥50 μM IPTG resulted in growth comparable to that of the parental strain in broth (Fig. 2B) or agar medium (Fig. 2C).

FIG 1.

Diagram showing the recombination event that generated a conditional nrdEF mutant, B. subtilis PERM1017. Expression of nrdE-nrdF in this strain is under the control of the IPTG-inducible Pspac promoter. In this strain, a 5′-end fragment of nrdE is fused to lacZ, and the resulting nrdE-lacZ fusion and nrdI are under the transcriptional control of the native regulatory regions of the nrdIEF operon. The 3′ end of the construct provides the cell with a functional RNR when IPTG is present; the 5′ end of the construct indicates native promoter activity.

FIG 2.

(A) Transcriptional induction of nrdEF and growth phenotype of strain B. subtilis PERM1017 in the absence or presence of increasing amounts of IPTG. (B) Viable counts of cultures for which the results are shown in panel A during the exponential phase (OD₆₀₀ = 0.7) and stationary phase (OD₆₀₀ = 1.6) of growth. (C) Ability of B. subtilis strain PERM1017 to grow in minimal medium supplemented or not with increasing amounts of IPTG. B. subtilis PERM1017 was propagated in liquid LB supplemented with 1 mM IPTG until saturation. Cells collected by centrifugation were washed and resuspended in 1× SS. Serial dilutions of the cell suspension were plated in SMM supplemented or not with different concentrations of IPTG. Colony formation was monitored after 48 h of incubation at 37°C. 0, no IPTG addition; PS, parental strain YB955.

YtcG represses nrdEF expression in B. subtilis.

It has been proposed that the nrdEF genes from B. subtilis are transcribed together with nrdI and ymaB as part of the nrdIEF-ymaB operon (3, 4). Further, results from a transcriptomic study revealed that nrdE and nrdF seem to be regulated by a DnaA-mediated transcriptional response (29, 30). Aside from this, little is known regarding additional factors involved in modulating nrdEF expression in B. subtilis. In this bacterium, the gene ytcG encodes a 152-amino-acid protein with 49% identity and 67% similarity to the transcriptional repressor NrdR from E. coli (7). Depending on the bacterial species, the length of the NrdR repressor varies from 150 to 200 amino acids. The N-terminal portion displays a zinc finger-like DNA-binding domain, whereas the C-terminal region encodes a putative class Ia ATP/dATP allosteric site (31–33). On the basis of these properties, it is proposed that increasing cellular levels of nucleotides activate the repressing properties of NrdR, which recognizes a palindromic element (NrdR box) located in the promoter region of the nrdEF operon (34, 35). In fact, results from an in silico analysis revealed the presence of two potential NrdR boxes in the regulatory region that control the expression of the nrdIEF operon (34). We then investigated whether YtcG is involved in regulating the expression of nrdEF; to this end, we first analyzed the expression patterns of nrdEF and ytcG during the life cycle of B. subtilis. B. subtilis strains bearing transcriptional in-frame nrdEF-lacZ and ytcG-lacZ fusions were propagated in antibiotic medium 3 (A3 medium [also called Penassay broth]; Difco Laboratories, Sparks, MD), and cell samples collected at different growth stages were assayed for β-galactosidase activity. As shown in Fig. 3A, expression of the nrdEF-lacZ fusion was maximal (∼255 ± 10 Miller units) during exponential growth and decreased ∼3-fold during the transition and stationary phases of growth. On the other hand, the levels of β-galactosidase produced by the B. subtilis ytcG-lacZ strain were about 30-fold lower than those produced by the nrdEF-lacZ strain and remained barely constant during all the growth stages of B. subtilis (Fig. 3B). These results suggest that YtcG is present during the entire life cycle of B. subtilis putatively to modulate the levels of nrdEF. To determine whether ytcG encodes a functional repressor for nrdEF transcription, the level of β-galactosidase activity of the nrdEF-lacZ fusion strain was measured in a background with a defective ytcG gene. As shown in Fig. 3C, in reference to the YtcG-proficient strain, the level of nrdE expression was increased ∼6-fold during exponential, early, and late stages of growth in the YtcG-deficient strain. These results, together with the presence of two putative NrdR boxes upstream of nrdIEF overlapping −10 and −35 promoter elements (34), strongly suggest that (i) YtcG is the functional repressor for nrdEF transcription, (ii) its regulatory functions are exerted throughout the life cycle of B. subtilis, and (iii) the temporal pattern of nrdEF expression is independent of YtcG.

FIG 3.

Analysis of expression of nrdE-lacZ (A), ytcG-lacZ (B), and ΔytcG nrdE-lacZ (C) transcriptional fusions during exponential and stationary phases of B. subtilis growth. Strains were propagated in PAB medium. Cells collected at the indicated times were disrupted with lysozyme, and the cell extracts were assayed for β-galactosidase with ONPG as described in Materials and Methods. The data shown are average values from triplicate independent experiments ± standard deviations (SDs) for β-galactosidase specific activity (□) and for OD₆₀₀ (●).

nrdEF expression in nutritionally stressed B. subtilis cells.

The strain B. subtilis YB955, bearing the chromosomal auxotrophic markers hisC952, metB5, and leuC427, has been successfully employed as a model system to understand how mutations are generated under conditions of limited growth (11, 15, 17–20, 36–39). It is possible that under those conditions dNTPs are required for DNA repair; however, the expression levels of nrdEF in B. subtilis cells under conditions of growth arrest are currently unknown. We investigated nondividing cells of strains YB955, PERM1017, and PERM1202 for the presence of nrdE transcript. These strains were grown and subsequently exposed to nutritional stress in agar medium lacking histidine, methionine, and leucine and incubated for a period of 8 days at 37°C. Reverse transcription-PCR (RT-PCR) analysis was performed with total RNA extracted from the three strains at the beginning of incubation and at 3 and 8 days of incubation. As shown in Fig. 4, in parental strain YB955, the nrdE transcript levels decreased by about three times by day 8 of starvation. In contrast, in the conditional nrdEF mutant, in the absence of IPTG, the nrdE transcript levels decreased by about three times by day 4 and completely disappeared by day 8 of starvation (Fig. 4). On the other hand, the ΔytcG strain expressed about 1.8 times higher levels of the nrdE transcript than strain YB955 beginning on the first starvation day; moreover, such levels increased 3 and 6 times at days 3 and 8 poststarvation, respectively (Fig. 4).

FIG 4.

nrdE expression in starved B. subtilis cells. (A) RT-PCR analysis of nrdE expression in strains YB955, PERM1017 (nrdEF), and PERM1202 (ytcG) on day 1, 3, or 8 after incubation in 1× SMM lacking His, Met, Leu, and IPTG. (B) Densitometry analysis of the results shown in panel A using the veg gene as an endogenous control for constitutive expression (31). The results shown in panel B are average values from triplicate independent experiments ± SDs.

Inactivation of nrdEF inhibits B. subtilis stationary-phase-associated mutagenesis.

The conditional RNR mutant constructed in the YB955 genetic background (PERM1017) was used to establish the role of NrdEF in producing His⁺, Met⁺, and Leu⁺ adaptive revertants. To this end, after the mutant strain was propagated under permissive conditions, cells were collected at 90 min after the cessation of growth and plated on Spizizen minimal medium (SMM) lacking IPTG and two of the required amino acids (histidine and leucine dropout medium or methionine and leucine dropout medium), and then after different periods of incubation, a set of plates was taken and supplemented with 0.1 mM IPTG (to induce nrdEF expression) and one of the initially absent amino acids. In this way, the initial absence of IPTG allowed us to temporarily inactivate nrdEF expression to evaluate the contribution of RNR to mutagenesis, whereas the subsequent addition of this agent guaranteed supplementation of the dNTPs required for DNA replication in the revertant cells. Moreover, the initial absence of two essential amino acids prevented the growth of revertants with variant or less efficient enzymes conferring growth (33, 36). Our results revealed that in comparison with the levels of reversion in parental strain YB955, inactivation of nrdEF expression in the conditional mutant PERM1017 induced a significant decrease in the levels of production of the His⁺, Met⁺, and Leu⁺ revertants (Fig. 5). Importantly, during the course of the experiments, it was shown that the absence of amino acids or IPTG did not affect the survival rates of the studied strains (see Fig. S1 in the supplemental material). Therefore, the levels of SPM-dependent His⁺, Met⁺, or Leu⁺ reversion obtained cannot be attributed to problems with the viability of the strains tested. Taken together, these results support the notion that NrdEF is required to provide the precursors necessary for events of DNA synthesis that promote mutations in nongrowing B. subtilis cells.

FIG 5.

The frequencies of stationary-phase reversions to His⁺ (A), Met⁺ (B), and Leu⁺ (C) of B. subtilis strains YB955 (○), PERM1202 (ΔytcG) (▲), and PERM1017 (Pspac-nrdEF) (■) were determined as described in Materials and Methods. Data were normalized to the initial cell titers ± SDs and represent the counts averaged from three separate tests.

Overexpression of nrdEF promotes B. subtilis SPM.

To establish the impact of RNR derepression in B. subtilis SPM, we determined the reversion frequencies to His⁺, Met⁺, and Leu⁺ in cultures of starved B. subtilis ΔytcG cells. As shown in Fig. 5, the YtcG-deficient strain showed a significant increase in the frequencies of His⁺, Met⁺, and Leu⁺ reversions in reference to those generated by the strain proficient for this repressor. Importantly, as shown in Fig. S1, the levels of SPM-dependent amino acid reversions could not be attributed to differences in the viabilities of the strains. Together, these results and those from the experiments in which nrdEF was repressed strongly support the participation of RNR in B. subtilis SPM.

Analysis of suppressor mutations in the nrdEF-overexpressing strain.

Histidine and methionine auxotrophies were generated by nonsense mutations in the hisC952 and metB5 genes, and then it was suggested that His⁺ Met⁺ revertants could be generated by a suppressor sup-3 allele that simultaneously confers prototrophy to both His⁺ and Met⁺ revertants, in addition to other less well characterized suppressors (11, 40). Our analysis for suppressor mutations showed that in strain YB955 a high percentage (47%) of the Met⁺ revertants presented a His⁺ phenotype; moreover, 10% of the His⁺ colonies were also Met⁺ revertants (Table 1), suggesting that the His⁺ Met⁺ prototrophs were also generated by suppressor mutations. As also shown in Table 1, repression of nrdEF in the conditional mutant did not promote a significant change in the proportion of His⁺ Met⁺ colonies in comparison with that for YB955. Of note, the absence of YtcG (NrdR) generated a lower but more diverse group of suppressors bearing the His⁺ Met⁺ (4%), Leu⁺ Met⁺ (7%), His⁺ Leu⁺ (2%), and Leu⁺ His⁺ (2%) phenotypes.

TABLE 1.

Growth of stationary-phase revertants on alternative selection media^d

Strain and selection medium	No. of revertants that showed the following phenotype/no. of revertants analyzed (%)
	His+ Met+ Leu+	His+ Met+	Met+ Leu+	His+ Leu+	Single revertant (His+, Met+, or Leu+)
ΔytcG strain
His−	27/100 (27)			2/100 (2)	71/100 (71)a
Met−	7/100 (7)	4/100 (4)			89/100 (89)b
Leu−	22/100 (22)		7/100 (7)	2/100 (2)	69/100 (69)c
YB955
His−	3/100 (3)	10/100 (10)			87/100 (77)a
Met−	2/100 (2)	47/100 (47)			51/100 (51)b
Leu−					100/100 (100)c
Pspac-nrdEF strain (with nrdEF inactivated)
His−		3/54 (5)			51/54 (95)a
Met−		5/19 (26)			14/19 (74)b
Leu−					50/50

^aHis⁺ revertants.

^bMet⁺ revertants.

^cLeu⁺ revertants.

^dHis⁺, Met⁺, and Leu⁺ revertants from day 4, 6, or 8 were screened on 1× SMM missing one required amino acid (His, Met, or Leu [His⁻, Met⁻, or Leu⁻, respectively) to screen for suppressor mutations. Plates were scored after 48 h of incubation.

Derepression of nrdEF promotes the formation of intragenic suppressors.

The high propensity to accumulate adaptive colonies with reversion in the three genes tested exhibited by the YtcG-deficient strain prompted us to investigate the type of mutation that gave rise to this phenotype. The hisC952, metB5, and leuC427 genes of 30 independent adaptive His⁺ Met⁺ Leu⁺ revertants generated at day 3, 4, or 5 by the YtcG-deficient strain were sequenced. Surprisingly, about two-thirds of the colonies analyzed contained changes that restored the wild-type sequence of the three genes analyzed (Table 2). However, the genetic changes associated with one-third of the triple His⁺ Met⁺ Leu⁺ revertants were not identified in the amplicons sequenced, suggesting that missense suppressors generated these types of revertants (Table 2). Sequencing analysis of single Met⁺, His⁺, or Leu⁺ revertants revealed the presence of mutations that changed the stop codons to distinct amino acid codons in the first two types of revertants. The mutations that reestablished the Leu⁺ phenotype were not located in the mutant codon of the leuC427 gene.

TABLE 2.

Base changes in revertants of mutant alleles^e

Phenotype and revertant allele	Position(s) of mutation (bp)	No. of revertants with mutation/no. of revertants sequenced	Type of mutation	DNA change	Result of mutation
His+ Met+ Leu+a
hisC	952	22/30	Transition	T → C	Stop → Gln
hisC	NF	8/30	NF	NF	NF
metB	346	19/30	Transversion	T → G	Stop → Glu
metB	346 and 348	2/30	Transversion	T → G and A → T	Stop → Asp
metB	NF	9/30	NF	NF	NF
leuC	427	19/30	Transition	A → G	Arg → Gly
leuC	NF	11/30	NF	NF	NF
Met+b
metB	346	6/10	Transition	T → C	Stop → Gln
metB	346	2/10	Transition	T → A	Stop → Lys
metB	NF	2/10	NF	NF	NF
His+c
hisC	952	4/10	Transition	T → C	Stop → Gln
hisC	952	2/10	Transversion	T → G	Stop → Glu
hisC	953	4/10	Transition	A → G	Stop → Trp
Leu+,d leuC	NF	10/10	NF	NF	NF

^aRevertants were selected in histidine dropout medium (7/30 revertants) or methionine dropout medium (23/30 revertants).

^bRevertants were selected in methionine dropout medium (10 revertants).

^cRevertants were selected in histidine dropout medium (10 revertants).

^dRevertants were selected in leucine dropout medium (10 revertants).

^eRevertants were tested for their ability to grow on 1× SMM lacking histidine, methionine, and leucine and classified for their phenotype. Fragments of the hisC, metB, and leuC genes were sequenced depending on the observed reversion. NF, no changes were found in the sequenced fragment.

Mutation rates in growing B. subtilis cells with derepressed nrdEF.

The reversion levels of the hisC952, metB5, and leuC427 alleles in exponentially growing cells of the strain deficient for the repressor YtcG were determined and compared with those generated in parental strain YB955. As shown in Fig. S2, the levels of production of His⁺, Met⁺, or Leu⁺ revertants by the ΔytcG strain increased ∼2.5, 2, and 1.5 times, respectively, compared to those by the YB955 parental strain. Therefore, higher levels of RNR also promoted mutagenesis in growing B. subtilis cells.

Transcription of nrdEF-lacZ is increased by oxidative stress.

In E. coli, Staphylococcus aureus, Bacillus anthracis, and Corynebacterium ammoniagenes, genes coding for class Ib RNR are induced in response to oxidative stress (41–45); therefore, we inquired whether B. subtilis nrdEF could be also induced by the reactive oxygen species-promoting agent hydrogen peroxide (41, 44, 46–48). To investigate this point, strain PERM1017 carrying the nrdE-lacZ fusion was cultured in A3 medium to an OD₆₀₀ of 0.5 and treated with hydrogen peroxide (1 mM). The results revealed that exposure to H₂O₂ increased the level of nrdEF expression 1.5-fold compared to that in a nontreated culture (Fig. 6A). In B. subtilis, the PerR repressor controls the expression of an oxidative stress regulon that is triggered by the presence of H₂O₂ in the culture medium (46, 48). As shown in Fig. 6B, disruption of perR increased about five times the expression levels of the nrdEF-lacZ fusion in comparison with the level for the B. subtilis strain that was proficient for PerR. A similar result was found for group A Streptococcus, where expression of the ribonucleotide reductase nrdF2-nrdI2-nrdE2 operon was directly regulated by the PerR repressor (47). Altogether, our results suggest that in B. subtilis, PerR directly or indirectly regulates the expression of RNR-encoding genes in response to oxidative stress.

FIG 6.

H₂O₂ induction of an nrdE-lacZ transcriptional fusion and its expression in B. subtilis strains deficient or not for PerR. (A) Induction of nrdE-lacZ fusion in B. subtilis after exposure to H₂O₂. B. subtilis PERM1017 was propagated in A3 medium to an OD₆₀₀ of 0.5. At this point the culture was equally divided into two flasks. One of the subcultures was left untreated (circles), and the other one was challenged with 1 mM H₂O₂ (triangles). Cells harvested at the indicated times were assayed for β-galactosidase activity as described in Materials and Methods. Values represent averages from triplicate independent experiments ± SDs. (B) Expression of an nrdE-lacZ fusion in a PerR genetic background. The levels of β-galactosidase produced by B. subtilis strains PERM1017 (nrdE-lacZ) (circles) and PERM1159 (perR nrdE-lacZ) (triangles) are shown. Strains were grown in A3 medium. Cell samples were collected at the indicated times and treated with lysozyme, and the extracts were assayed for β-galactosidase as described in Materials and Methods. The data shown are average values from triplicate independent experiments ± SDs.

DISCUSSION

Ribonucleotide reductases, the rate-limiting factor for the synthesis of 2′-deoxyribonucleotides, are subjected to extensive allosteric regulation in most organisms (1, 2). However, the intracellular amount of these enzymes is also regulated at the transcriptional level; such control is of particular relevance in bacterial species possessing different types of RNRs, including E. coli, Salmonella, and Pseudomonas aeruginosa (7, 8, 35). B. subtilis possesses one class Ib RNR, encoded by the nrdEF operon (3, 4). Disruption of either the nrdE or the nrdF open reading frame generates a growth arrest in this microorganism (4). Studies aimed at understanding the physiological role of NrdEF in B. subtilis have been performed with temperature-sensitive conditional mutants. These studies have employed 45°C as the nonpermissive condition (25). However, under these conditions the heat shock regulon is activated (49, 50) and may confound the effects of RNR activity. The conditional mutant B. subtilis YB955 Pspac-nrdEF, constructed in this work, permitted analysis of the mutagenic events leading to the production of colonies with His⁺, Met⁺, and Leu⁺ phenotypes under growth-limiting conditions and at 37°C. This avoided the undesired pleiotropic or confounding effects triggered at 45°C. Importantly, the amount of IPTG (namely, 0.1 mM) chosen to propagate the conditional mutant promoted growth levels indistinguishable from those of the parental strain (Fig. 2). Furthermore, experiments employing a different marker for mutagenesis (i.e., the rpoB gene) demonstrated that increased levels of nrdEF expression correlated with increased frequencies of mutation to rifampin resistance (Rif^r) in the conditional mutant strain PERM1017 (see Fig. S3 in the supplemental material).

In B. subtilis, the DNA region preceding the nrdIEF operon possesses two putative conserved DNA sequences predicted to be bound by NrdR (31). This protein represses the expression of genes encoding the class 1b RNR in E. coli and other bacteria (7, 8, 35). B. subtilis contains a putative ortholog of this protein encoded by ytcG. In this work, we showed that disruption of this gene resulted in the increased expression of an nrdE-lacZ transcriptional fusion, strongly suggesting that YtcG corresponds to the NrdR ortholog in this microorganism.

Oxidative stress is an important factor that promotes adaptive mutagenesis in B. subtilis (18); thus, deficiency of the GO system exacerbated the production of His⁺ and Met⁺ revertants in nondividing cells of this microorganism (18). Interestingly, as previously described for group A Streptococcus (47), our results showed that the oxidative stress promoted by H₂O₂ activated the expression of the nrdEF operon. Further, nrdEF expression was significantly increased in a B. subtilis genetic background deficient for the repressor PerR. Of note, conditions that interfere with replication or cell elongation in this microorganism induce PerR as well as the RNR operon in a way dependent on DnaA (30). On the basis of these results, it is feasible to propose that the adverse metabolic conditions operating in nutritionally stressed cells may affect RNR levels. Our results also showed that repression of nrdEF transcription in strain PERM1017 induced a significant decrease in the levels of production of adaptive His⁺, Met⁺, and Leu⁺ revertants compared to those for the parental strain. In contrast, the RNR conditional mutant deficient in YtcG overexpressed the nrdEF operon and displayed increases, by severalfold, in reversion in the three genes tested (Fig. 5). These results revealed that reversions in the tested genes increased in parallel with nrdEF induction. Interestingly, the expression levels of nrdE decreased in the nonrevertant populations of YB955 and PERM1017 under conditions of prolonged starvation; however, a fraction of starved YB955 cells produced revertants (Fig. 5), suggesting that changes in RNR transcription take place in a subpopulation of cells to promote mutagenic events. Previous reports revealed that adaptive mutations arise in a subpopulation of nutritionally stressed nongrowing, bacterial cells (11, 17). Taken together, our results support the concept that induction of RNR is necessary to generate adaptive mutations in B. subtilis and suggest that the metabolic conditions prevailing in nongrowing cells increase the expression levels of RNR, which have a direct impact on stationary-phase-associated mutagenesis (SPM).

Our suppressor analysis revealed that an important proportion of the His⁺ and Met⁺ colonies generated by the conditional nrdEF mutant were also prototrophic for methionine or leucine, respectively, whereas none of the Leu⁺ colonies tested exhibited additional amino acid prototrophies (Table 1). Of note, these analyses also revealed that a very high proportion (∼20%) of the His⁺, Met⁺, or Leu⁺ revertants generated by nondividing cells of the strain deficient for NrdR acquired mutations that conferred prototrophy in two other genes (Table 1). Remarkably, DNA sequencing analysis showed that about two-thirds of the His⁺ Met⁺ Leu⁺ colonies carried mutations that changed not only the missense leuC427 but also the nonsense hisC952 and metB5 genes to amino acid codons that reestablished the wild-type sequences of their open reading frames. Moreover, in one-third of the triple His⁺ Met⁺ Leu⁺ revertants, the mutations were not detected in the defective genes; however, these colonies exhibited robust growth in each of the three selective media (histidine dropout medium, methionine dropout medium, and leucine dropout medium). Further sequence analyses showed differences between mutant colonies bearing triple and single reversions; for instance, Met⁺ reversions in the triple His⁺ Met⁺ Leu⁺ colonies mainly resulted from a T → G transversion event. In contrast, T → C and T → A suppressor mutations were involved in generating colonies with single Met⁺ reversions. Taken together, these results point to the existence of a common mechanism that increases the occurrence of mutations in genes under selection in starved cells of B. subtilis deprived of NrdR.

In starved B. subtilis YB955 cells, true reversions to His⁺, Met⁺, and Leu⁺ have been associated with transcription repair coupling events mediated by Mfd (38); notably, the frequency of Mfd-dependent true reversions increased proportionally to the transcription levels of the gene under selection (36, 37). Perhaps the high proportion of true reversions in starved cells overexpressing the RNR-encoding genes is linked to mutagenic events occurring at genes undergoing transcriptional derepression. Since our assays were conducted in the absence of histidine, methionine, or leucine, one would expect that genes like hisC955, metB5, and leuC427 would be highly transcribed and would be subjected to transcription-mediated mutagenesis.

The stringent response, activated during starvation and mediated by GTP levels, induces the expression of genes involved in amino acid biosynthesis (51). Such a response has been shown to increase bacterial survival during amino acid starvation (52). In contrast, high concentrations of GTP activate CodY, which represses the expression of genes required for the synthesis and utilization of branched-chain amino acids (BCAAs), in addition to methionine, histidine, and arginine, among others (52, 53) Furthermore, it has been reported that changes in RNR levels affect the fidelity of DNA synthesis by increasing and/or biasing the dNTP pools (22–24, 54, 55). In agreement with these observations, in nongrowing B. subtilis cells, the low-fidelity DNA polymerases PolX, YqjH, and/or PolA have been shown to promote error-prone repair events of DNA synthesis (20, 21, 56–59).

Based on these observations, we propose that in growth-arrested B. subtilis cells, changes in expression of RNR genes alters dNTP/NTP pools. Changes in dNTP/NTP pools precipitate the activation of the stringent response; change transcription, particularly in amino acid biosynthetic genes; and alter the fidelity of DNA polymerases. All these conditions lead to a mutagenic program that increases the likelihood that stressed cells will escape from growth-limiting conditions.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The B. subtilis strains and plasmids used in this study are listed in Table 3. The procedures used for the transformation and isolation of chromosomal and plasmid DNA were described previously (60, 61). Liquid cultures of B. subtilis strains were routinely grown in Penassay broth (PAB; Difco Laboratories, Sparks, MD). Growth was monitored with a Pharmacia Ultrospec 2000 spectrophotometer set at 600 nm. When required, erythromycin (Ery; 1 μg/ml), neomycin (Neo; 10 μg/ml), kanamycin (Kan; 10 μg/ml), or isopropyl-β-d-thiogalactopyranoside (IPTG; 0.1 mM) was added to the medium. PCR products were obtained with homologous oligonucleotide primers (Table 4) and Vent DNA polymerase (New England BioLabs, Ipswich, MA).

TABLE 3.

Strains and plasmids used in this study

Strain or plasmid	Genotype or descriptiona	Reference, source, or transformationb
Strains
B. subtilis
YB955	hisC952 metB5 leuC427 xin-1 Spβs	11
HB2078	CU1065 ΔperR::kan Kanr	46
PERM1017	ΔnrdE::lacZ Pspac-nrdEF::ery Eryr	pPERM1011 → YB955
PERM1202	ΔytcG::neo Neor	pPERM1179 → YB955
PERM1207	ΔytcG::neo ΔnrdE::lacZ Pspac-nrdEF::ery Neor Eryr	pPERM1179 → PERM1017
PERM1159	ΔperR::kan ΔnrdE::lacZ Pspac-nrdEF::ery Kanr Eryr	pPERM1011 → HB2078
PERM1448	ytcG::lacZ	pPERM1330 → YB955
E. coli XL10-Gold	endA1 glnV44 recA1 thi-1 gyrA96 relA1 lac Hte Δ(mcrA)183 Δ(mcrCB-hsdSMR-mrr)Tn10 Tetr Cmr	Laboratory stock
Plasmids
pMUTIN4	Integrational lacZ fusion vector; Ampr Eryr	62
pBEST501	Integrational neomycin replacement vector; Ampr Neor	63
pPERM1011	pMUTIN4 containing a 5′ fragment (from bp −47 to 290) of the nrdE ORF cloned into the HindIII-SacII sites; Ampr Eryr	This study
pPERM1179	pBEST501 containing a 5′ fragment (from bp −367 to −22) and a 3′ fragment (from bp +23 to +365) adjacent to the ytcG ORF cloned into the HindIII-SalI and BamHI-SacI sites, respectively; Ampr Eryr	This study
pPERM1330	pMUTIN4 containing a 3′ fragment (from bp −179 to 456) of the ytcG ORF cloned into BamHI-EcoRI sites; Ampr Eryr	This study

^aAmp, ampicillin; Ery, erythromycin; Neo, neomycin; Kan, kanamycin; ORF, open reading frame.

^bTransformations are represented by X → Y, where strain Y was transformed with plasmid DNA from source X.

TABLE 4.

Oligonucleotides used for PCRs

Primer no.	Oligonucleotidea	Sequenceb (5′ to 3′)	Amplified region
1	F-nrdE	GAAGCTTCGGCACATCTAAAGA	nt −47 to 290 relative to translational start codon of nrdE
2	R-nrdE	CGCCGCGGCACTCATGAAGGAAG
3	F-ytcG-US	CGCCGCGGCACTCATGAAGGAAG	nt −367 to −22 relative to translational start codon of ytcG
4	R-ytcG-US	CGGTCGACCCCAACTATATCGTTCCCG
5	F-ytcG-DS	GCGGATCCGCTAAGAGGATTCTTCTAGC	nt +23 to +365 relative to translational stop codon of ytcG
6	R-ytcG-DS	GCGAGCTCCCTGTTTTCCTGATGGATCGT
7	F-ytcG	GGATCCAAATGTCCTTCATGCCA	nt 179 to 456 of ytcG open reading frame
8	R-ytcG	GAATTCCTCTTTCTTTATTAAGTC
9	F-hisC	GCAGGCCTTCAGCAGTATTATGAT	nt 840 to +71 relative to translational stop codon of hisC
10	R-hisC	GACCGGCGAGCAATATTGTATCTTTCA
11	F-metB	ATCCCAAGCACACACACGCCG	nt 210 to 516 of metB open reading frame
12	R-metB	AACGGTATGTTCGAACACACCGAAGAT
13	F-leuC	CGCTTCCAGGGAAAACGATTG	nt 353 to 655 of leuC open reading frame
14	R-leuC	CGATGGATGAACGAATGACTG
15	F-nrdE	GCTGCAGGCATCTCAAGTTT	nt 1155 to 1344 of nrdE open reading frame
16	R-nrdE	CGCATTACGGATGTCTGTTG
17	F-veg	TGGCGAAGACGTTGTCCGATATTA	nt 2 to 83 of veg open reading frame
18	R-veg	CGGCCACCGTTTGCTTTTAAC

^aF, forward; R, reverse.

^bUnderlining indicates the following restriction sites: primers 1 and 3, HindII; primer 2, SacII; primer 4, SalI; primers 5 and 7, BamHI; primer 6, SacI; and primer 8, EcoRI.

Construction of mutant strains.

To obtain a conditional nrdEF mutant in the B. subtilis YB955 genetic background, a 333-bp HindIII-SacII 5′ fragment of the nrdE gene (from bp −47 to bp 290 of the nrdE translational start codon) was amplified by PCR using chromosomal DNA from B. subtilis 168 and oligonucleotide primers 1 and 2 (Table 4). The PCR fragment was inserted between the HindIII and SacII sites of the integrative plasmid pMUTIN4 (62). The resulting construction, designated pPERM1011 (nrdE::lacZ), was used to transform B. subtilis YB955, generating strain PERM1017 (Pspac-nrdEF) (Fig. 1).

To generate a ytcG (nrdR) mutant, a 345-bp fragment from the 5′ ytcG region (nucleotides [nt] −367 to −22 relative to the ytcG start codon) and a 342-bp fragment from the 3′ ytcG region (nt +23 to +365 relative to the ytcG stop codon) were PCR amplified using chromosomal DNA from B. subtilis 168. Oligonucleotide primers 3 and 4 and oligonucleotide primers 5 and 6 (Table 4) were used for amplification of the 5′ ytcG and 3′ ytcG fragments, respectively. The amplified 5′ ytcG and 3′ ytcG fragments were cloned between the HindIII/SalI and BamHI/SacI sites of pBEST501 (63), respectively. The resulting construction, designated pPERM1179, was propagated in E. coli XL10-Gold cells and used to transform competent cells of B. subtilis YB955 to generate strain PERM1202 (ΔytcG). The double-crossover event replacing ytcG with a neomycin resistance cassette was confirmed by PCR (results not shown).

Construction of ytcG-lacZ and nrdE-lacZ fusions.

To generate a strain carrying a ytcG-lacZ fusion, a fragment from positions 179 to 456 of the ytcG open reading frame was amplified by PCR using chromosomal DNA from B. subtilis 168 and oligonucleotide primers 7 and 8 (Table 4). The PCR product was cloned between the BamHI and EcoRI sites of the integrative plasmid pMUTIN4 immediately upstream of the lacZ gene. The resulting construct, designated pPERM1330, was used to transform B. subtilis YB955, generating strain PERM1448 carrying a transcriptional in-frame ytcG-lacZ fusion.

To recombine the transcriptional in-frame nrdE-lacZ fusion in a PerR-deficient genetic background, competent cells of strain B. subtilis HB2078 (perR) (46) were transformed with plasmid pPERM1011, thus generating strain B. subtilis PERM1159 (Table 3). The correct integration of the construction into the nrdE locus was verified by PCR using specific oligonucleotide primers (data not shown).

Stationary-phase mutagenesis soft-agar overlay assays.

The stationary-phase mutagenesis assays were performed as previously described (36) with modifications. Strains of B. subtilis were grown in 10 ml of PAB medium to 90 min after time zero, which was the time point in the culture when the slopes of the logarithmic and stationary phases of growth intercepted. After centrifugation, the cell pellets were washed 3 times and resuspended in 10 ml of 1× Spizizen minimal salts (1× SS) (64). Aliquots of 100 μl were plated onto Spizizen minimal medium (SMM; 1.5% agar, 1× SS, 0.5% glucose, 50 μg/ml isoleucine, 50 μg/ml glutamic acid; the content of histidine, methionine, and leucine depended on the reversion assay). Two types of minimal medium were used: histidine and leucine dropout medium (SMM with 50 μg/ml of methionine, 200 ng/ml of leucine, and 200 ng/ml of histidine) and methionine and leucine dropout medium (SMM with 50 μg/ml of histidine, 200 ng/ml of leucine, and 200 ng/ml of methionine). Starting from 48 h and every 2 days thereafter, a set of plates was overlaid with soft agar (0.7% agar prewarmed at 42°C) amended with one of the amino acids initially absent to select His⁺, Met⁺, or Leu⁺ revertants. For strain PERM1017, the soft agar was supplemented with IPTG to a final concentration of 1 mM to induce expression of the nrdEF genes. The plates were incubated for 2 days, and the number of revertant colonies was scored. The initial number of bacteria plated for each experiment was determined by serial dilution of the bacterial cultures that were plated on SMM containing all three essential amino acids. The experiments were repeated at least three times.

Cell survival was determined by removing three agar plugs daily, using sterile Pasteur pipettes, from areas of the plates where no growth of revertants was observed (20). The plugs were suspended in 400 μl of 1× SS, mixed, diluted, and plated on SMM containing all the essential amino acids (50 μg/ml). The number of colonies was determined following 48 h of incubation at 37°C.

Fluctuation test.

The growth-dependent reversion rates for His, Met, and Leu were measured by fluctuation tests with the Lea-Coulson formula (65, 66), as previously described (11, 15, 18, 20).

Analysis of frequencies of mutation to Rif^r.

The frequencies of spontaneous mutation to rifampin resistance (Rif^r) were determined as previously described (18). Essentially, the appropriate strains were grown for 12 h at 37°C in antibiotic A3 medium with proper antibiotics. Independent cultures of strain PERM1017 were amended with 0.025, 0.1, or 1 mM IPTG, whereas the ΔnrdR strain was cultured in the absence of the inducer. Mutation frequencies were determined by plating aliquots on six LB plates containing 10 μg ml⁻¹ rifampin, and the Rif^r colonies were counted after 1 day of incubation at 37°C. The number of cells used to calculate the frequency of mutation to Rif^r was determined by plating aliquots of appropriate dilutions on LB plates without rifampin and incubating the plates for 24 to 48 h at 37°C. These experiments were repeated at least three times.

β-Galactosidase assays.

B. subtilis strains PERM1017 and PERM1207, both of which contain the transcriptional nrdE-lacZ fusion, were propagated in liquid A3 medium supplemented with 0.1 mM IPTG. Aliquots of 0.5 ml were collected from the cultures during the exponential and stationary phases of growth at the times indicated in Fig. 3 and 6 and processed for determination of β-galactosidase activity using o-nitrophenyl-β-d-galactopyranoside (ONPG) as a substrate (67).

DNA sequencing.

Revertant colonies with the ΔytcG background were collected from plates from the stationary-phase mutagenesis assays on days 3 to 5 of incubation, independently propagated in liquid A3 medium, and subjected to DNA isolation. Internal fragments of the genes hisC952, metB5, and leuC427 were amplified by PCR using high-fidelity DNA polymerase and specific oligonucleotide primers (Table 4). Sequencing services were carried out by Functional Biosciences, Inc. (Madison, WI).

RT-PCR.

Strains YB955, PERM1017, and PERM1202 were grown under the conditions of the stationary-phase mutagenesis assays described above. Post-exponential-phase cultures (10 ml) were centrifuged, and the cell pellets were resuspended in 1 ml of 1× SS. Aliquots (100 μl) of this suspension were plated onto SMM containing growth-limiting amounts of histidine, methionine, and leucine and incubated at 37°C. At different intervals of incubation, the nonrevertant background was scraped off the agar medium with 2 ml of 1× SS and recovered by centrifugation. Total RNA was isolated by using a Tri Reagent kit (Molecular Research Center, Inc. Cincinnati, OH). Reverse transcription-PCRs (RT-PCRs) were performed with RNA samples and the Verso reverse transcriptase enzyme from a 1-step quantitative RT-PCR kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. Primers 15 and 16 (Table 4) were used to amplify an 81-bp RT-PCR product extending from nt 2 to 83 of the veg transcript, which was used as an internal control for constitutive expression (36), whereas primers 17 and 18 (Table 4) generated a 189-bp RT-PCR product extending from positions 1155 to 1344 of the nrdE transcript. In each experiment, the absence of chromosomal DNA in the RNA samples was assessed by PCRs carried out with Vent DNA polymerase (New England BioLabs) and the set of primers described above. The size of the RT-PCR products was determined by utilizing a 1-kb-Plus DNA ladder (Life Technologies, Rockville, MD) during agarose gel electrophoresis. Gels were stained with ethidium bromide and photographed using a Gel Doc EZ imager (Bio-Rad Laboratories, Hercules CA); digital photographic images of the gels were scanned, and the RT-PCR products were quantified by densitometry using Image Lab (version 4.1) software and the veg RT-PCR product corresponding to each condition as a reference.