Introduction

Transcription has been proposed to be the link between physiology and mutation occurrence [7, 14, 19]. Two mechanisms have been proposed to explain why transcription increases mutation rates. First, the nontranscribed strand is unprotected in the transcription bubble and is thus significantly more vulnerable to mutations than the transcribed strand. Second, transcription drives localized supercoiling which facilitates secondary structures formation, thus favoring lesions on exposed bases [23, 25]. In nonreplicating E. coli cells, genome instability propitiated by DNA/RNA hybrids and nicks, and triggers Mfd-dependent mutations and amplifications [24]. A mutagenic process dependent on Mfd, excision repair, and translation synthesis that prevents conflicts between transcription and replication have been recently described in growing B. subtilis cells [13]. The occurrence of stationary-phase-associated mutations (SPM) in the Gram-positive bacteria, B. subtilis has been previously demonstrated [18, 20]. In addition to suppressed or reduced DNA repair, error-prone repair of lesions has been proposed to contribute to the generation of mutations in nondividing B. subtilis cells [3, 5, 8, 21]. Experimental evidences in mouse embryonic fibroblasts (MEF) and Saccharomyces cerevisiae have revealed that BER deficiencies promote transcription-associated mutations (TAM) [7, 14]. However, the role of spontaneous lesions, including deaminated and oxidized bases in modulating this process in B. subtilis, is currently unknown. Results have demonstrated that in a bacterial culture, there is cell heterogeneity [6, 16], and that the batch methods employed for physiologic characterization are limited, since the contribution of small populations could be overlooked [2, 9]. Since mutation rates are low, even in the stressful starving conditions of stationary phase, a method of single-cell analysis such as flow cytometry (FCM) may be better suited for measuring mutations with the appropriate reporter system. Here, we describe the use of gfp as a reporter gene to evaluate at the single-cell level transcription-associated mutation events in nondividing stationary-phase B. subtilis cells. Nonfunctional point mutant alleles of the gfp gene were combined with FCM analysis of cells displaying gain-of-function phenotypes to show that a positive correlation between transcriptional derepression and the mutation frequency was enhanced in nongrowing B. subtilis cells deficient in base excision DNA repair.

Materials and Methods

Bacterial strains and growth conditions. Strains used are listed in Table S1 and were grown in antibiotic medium no.3 (A3) (Difco Laboratories, Sparks, MD) supplemented with 1× HoLe trace elements [20].

Reporter System Construction for Analysis of gfp Transcription Levels

The open reading frame (ORF) of the gfpmut3a allele from pAD123 [4] amplified by PCR was cloned between the NheI and SphI sites of plasmid pMPR025 [11]. This construct contained the s-box-gfp, where gfp expression is under positive transcriptional control with an IPTG-inducible hyperspank promoter (Phs) and under negative transcriptional control, by methionine attenuation through an s-box riboswitch (s-box) [12]. The s-box-gfp (pPERM1082) construct and an empty s-box (pMPR025) were independently recombined into the amyE locus to obtain the reporter amyE::sbox-gfp (PERM1087) and control amyE::s-box (PERM1114) strains (Fig. 1). Two-point-mutated gfp alleles encoding nonfunctional Gfp proteins were designed as follows. In the first one (gfp ns ), the third base of the 57th TGG codon was replaced for adenine, resulting in a TGA stop codon. For the second one, the 66th tyrosine codon, essential for Gfp fluorescence was mutated to a histidine codon, generating a full-sized nonfluorescent product, named gfp Tyr–His . The gfpmut3a gene was used as a template for introducing the gfp ns and gfp Tyr–His mutations using a set of mutagenic oligonucleotide primers and the QuickChange II system (Agilent Technologies; Santa Clara, CA). Once corroborated by DNA sequencing, the nonfunctional phenotype of the point-mutated alleles was determined by fluorescence microscopy with cells expressing the nonsense or missense gfp alleles from the IPTG-inducible Hyperspank promoter (Results not shown).

Fig. 1
figure 1

Diagram of the reporter (a) and control (b) constructs integrated in the amyE locus. Arrows above the boxes indicate the orientation of the genes

Full size image

Evaluation of Transcription-associated Mutations

Strains were propagated in A3 medium at 37 °C with vigorous aeration; cell samples collected at an OD600nm of 1 or 90 min after cessation of exponential growth were washed and resuspended in Spizizen Minimal Medium (SMM) [15] supplemented or not with 0.1 mM IPTG or methionine (50 µg mL−1). After 2 h of incubation at 37 °C, the cells were collected and resuspended in 0.2-µm-filtered phosphate buffered saline (PBS) for further analysis.

Microscopy and Flow Cytometry Analysis

Suspensions of cells of interest were processed for epifluorescence microscopy analysis as it was previously described [17]. Alternatively, bacterial suspensions were analyzed in a Flow cytometer MoFlo XDP (Beckman Coulter, Brea, CA, USA), equipped with a 488 nm laser for excitation. Bacterial cells stained with hexidium iodide (10 µg mL−1), washed, and resuspended in PBS were used for adjusting appropriate settings to discriminate debris and electronic noise. Control strains bearing an empty s-box construct were used for adjusting appropriate voltage settings in the FL1 filter (515/510 nm BP filter) in logarithmic scale, to establish the negative population, and for determination of background fluorescence. To discriminate singlet events, the events located in a lineal region in a dot-plot Forward Scatter (FS) logA vs FS logH were considered. At least, 3 million of singlet’s were acquired for gain-of Gfp function analysis.

Statistical Analysis

Where indicated, values presented in plots are averages from three triplicate experiments, with variations between experiments of ≤10 %. A nonparametric Mann–Whitney test was carried out without adjusting for ties. Minitab Express (Minitab Inc., State College, PA, USA) for Mac and Statistica 7 for Windows (StatSoft, Inc., Tulsa, OK) softwares were employed. Differences with P values ≤0.05 were considered statistically significant.

Results

Reporter System to Analyze TAM at a Single-cell Level

The reporter amyE::sbox-gfp strains (Fig. 1) cultured with 0.1 mM IPTG but lacking methionine and analyzed by microscopy, exhibited a high level of fluorescence (Fig. 2c). However, the presence of methionine in the medium induced a dramatic decrease in the Gfp-associated fluorescence (Fig. 2b), close to the level of the auto-fluorescence that is naturally exhibited by B. subtilis cells, carrying only the empty construction (Fig. 2a, e). Spectrofluorometric quantitation showed that under induction conditions, in three triplicate independent experiments, the fluorescent intensity of the cell bearing the s-box-gfp construct increased ~10 times with respect to the attenuated condition (i.e., 52 ± 4.5 vs. 5.1 ± 0.8, respectively). Moreover, as shown in Fig. 2c, f, the trace amount of tryptophan added to limit growth did not affect the synthesis of Gfp.

Fig. 2
figure 2

Control of gfp expression by IPTG induction and methionine attenuation. The control (amyE::s-box) (a, d) and reporter (amyE::s-box-gfp) (b, c, e, and f) strains were grown to exponential growth collected and incubated under conditions of attenuation (b, e) or induction (c, f) and analyzed by epifluorescence microscopy (a–c) and by flow cytometry (d–f). In a a bright-field image of cells of the control strain is also shown. (d–e): histogram of control strain in induction conditions (d), and overlay of reporter strains with the control strain in attenuation (e; green histogram) and induction (f; cyan histogram) conditions, control strain histograms shown in red. A total amount of 2500,000 singlet events were analyzed. In d–f, Y-axis = number of events; X-axis = fluorescent intensity detected in FL1 as log height (510–520 nm)

Full size image

The results shown in Fig. 2 analyzed by FCM to quantitate the cell population exhibiting a Gfp-associated fluorescence phenotype allowed us to establish two levels of gfp transcription, (i) attenuation, when methionine was supplemented in the medium (Fig. 2e), and (ii) induction, when cultures were supplemented with IPTG and lacked methionine (Fig. 2f). Therefore, these levels of control for expression of the gfp reporter allowed us to inquire whether levels of transcription influences mutagenesis in nongrowing bacterial populations (SPM), and whether base excision repair pathways play a role in this cellular process.

FCM Analysis of TAM by Gain of Gfp Function

To determine whether transcriptional levels have an impact in the population of cells that acquired a fluorescent phenotype, stationary-phase cells carrying the gfp ns and gfp Tyr-His point-mutated constructs were analyzed by FCM. A positive correlation between transcriptional induction and the number of fluorescent cells was found in cells of the parental strain carrying both mutant alleles (Fig. 3). However, it was evident that transcriptional derepression generated a higher amount of fluorescent cells in the s-box-gfp ns strain than in the strain bearing the s-box-gfp Tyr–His construction (Fig. 3 and Table 1).

Fig. 3
figure 3

Analysis of gain-of-function of mutant gfp alleles by FCM. Cells carrying gfp Tyr–His missense (a) and gfp ns nonsense (b) and the respective control cells were recovered from stationary phase cultures and incubated under conditions of attenuation (A) or induction (I). 3 × 106 cells were analyzed by FCM as described above. Box and Whisker plot, representing the number of fluorescent cells of 3 × 106 of singlet events analyzed. The box represents the 25–75 % interquartile range, the whiskers represent the nonoutlier range, the squares represent the median and circles represent the outliers’ values

Full size image
Table 1 Stationary-phase mutation frequencies determined by gain-of gfp function
Full size table

Role of Base Excision Repair in Promoting TAM

FCM analysis was conducted on stationary-phase cultures of strains deficient for deaminated (Δung ywqL) [10] or guanine oxidized (ΔGO) [22] systems in genetic backgrounds that contained the s-box-gfp ns or the s-box-gfp Tyr-His constructs, respectively. As shown in Fig. 3a, in cells that overexpressed the gfp Tyr-His allele, the absence of base deamination repair proteins (Ung, YwqL) significantly increased the fluorescent population in comparison with cells proficient for these repair proteins. Remarkably, the FCM analysis conducted under derepression of the nonsense gfp mutant allele identified a significantly higher proportion of cells with a fluorescent phenotype in the strains lacking the deaminated and guanine oxidized repair systems (Fig. 3b).

Stationary-phase Subpopulations with a Gain-of-gfp-function Phenotype also Exhibited a Hyper-mutagenic Rifr Phenotype

It has been proposed that adaptive mutations may arise in a hyper-mutagenic subpopulation of nutritionally stressed nongrowing, bacterial cells [8, 18, 20]. To provide support for this hypothesis, a stationary-phase IPTG-induced culture of the strain B. subtilis Phs-sbox-gfp ns was analyzed by FCM to identify the subpopulation of cells that gained a fluorescent phenotype. The mutation frequency to Rifr of cells sorted from this subpopulation and of those separated from a nonfluorescent region was determined as detailed above. The results shown in Table 2 revealed that the mutation frequencies levels to Rifr, in both subpopulations, were similar to those obtained with the gain-of function nonsense gfp allele (Table 1). However, the fluorescent cell subpopulation was ~3 times more mutagenic than the subpopulation of nonfluorescent cells (Table 2).

Table 2 Mutation frequencies (Rifr) in sorted B. subtilis stationary-phase subpopulations exhibiting or not a gain-of Gfp function phenotype
Full size table

Discussion

Here, we described a FCM-based approach to study TAM that offers the possibility of directly detecting in bacterial cultures, at single-cell level, the population of cells that gained or lost a fluorescent phenotype owing to mutational events occurring in a gfp reporter gene. After corroborating by fluorescence microscopy, the nonfunctional phenotype of the point mutated gfp alleles employed we focus our analysis on those mutations occurring in nutritionally stressed nongrowing bacteria using tryptophan auxotrophic isogenic strains carrying a defective trpC2 allele [1]. Accordingly, viable counts of the strains collected from stationary-phase cultures and starved for tryptophan did not significantly change during a period of 2 h (Results not shown). Overall with the gain-of-function system implemented, it was found that although transcriptional derepression promoted mutations that increased the number of cells with a functional Gfp phenotype, such events were more frequent in stationary-phase cultures of the WT strain bearing the gfp ns allele. As shown in Table 1, under conditions of derepression, the mutation frequencies in this allele were approximately 31-fold higher than those produced by the missense alleles. Thus, due to the nonessential character of the Trp57 residue, it is not unreasonable to speculate that a broad range of mutagenic events can revert the gfp mutation in this stop codon.

Previous reports have suggested that in nongrowing B. subtilis cells, the accumulation of spontaneous genetic lesions saturate the repair capacity of DNA repair systems promoting transcriptional-associated mutagenic events [19, 20]. Our results supported this suggestion as the genetic inactivation of the base deamination [10] and guanine oxidized excision repair pathways [22] enhanced the population of fluorescent cells in stationary-phase cultures that overexpressed the missense and nonsense gfp alleles. However, compared with the reversion frequency of the missense allele, under conditions of derepression, the mutation frequencies in the nonsense allele were approximately 60-fold for the ung ywqL mutant and 107-fold for the GO deficient strain (Table 1). Taken collectively, these results strongly suggest that deamination and oxidation events taking place in the TGA codon 57 promote true reversions, as well as compensatory amino acid changes in this gfp ns allele. Importantly, under growth-limiting conditions processing of damaged DNA bases can proceed in an error-prone manner promoting adaptive mutagenesis in B. subtilis [3, 5, 10, 22]. Furthermore, components of the BER pathway, including the 8-oxoguanine glycosylase (Ogg1) from MEF, the GO system and Mfd have been involved in modulating TAM in eukaryotes and bacteria [5, 11, 14, 15, 24].

We employed the rpoB gene as a different target to determine mutagenesis in sorted stationary-phase cell subpopulations from derepressed cultures of B. subtilis carrying the nonsense gfp allele. The mutation frequencies to Rifr obtained (Table 2) not only paralleled those obtained with the gain-of-function nonsense gfp allele (Table 1) but also were in agreement with SPM reversion levels of the hisC952, metB5, and leuC427 alleles reported in strain B. subtilis YB955 [5, 20, 22]. Interestingly in reference to a sorted cell subpopulation that did not acquired a Gfp functional phenotype, the Rifr mutagenesis levels were three times higher in the Gfp+ cell subpopulation (Table 2). These results, together with the increased mutagenesis levels exhibited by B. subtilis cell populations deficient for BER (Table 1), support the concepts that (i) adaptive mutations arise in nutritionally stressed growth-limited cell subpopulations with suppressed or saturated DNA repair systems and (ii) that such type of mutations are enhanced by transcriptional derepression. Finally, it must be pointed that although TAM is not limited to cells with a limited growth [13, 19], results from a FCM analysis measuring the loss of function of a wild-type gfp gene revealed that a minor proportion of growing cells lost the fluorescent phenotype in comparison with cells collected from nongrowing stationary-phase cultures suggesting that TAM are most frequent in nutritionally stressed cells (Data not shown).