Implementation of a loss-of-function system to determine growth and stress-associated mutagenesis in Bacillus subtilis

A forward mutagenesis system based on the acquisition of mutations that inactivate the thymidylate synthase gene (TMS) and confer a trimethoprim resistant (Tmpr) phenotype was developed and utilized to study transcription-mediated mutagenesis (TMM). In addition to thyA, Bacillus subtilis possesses thyB, whose expression occurs under conditions of cell stress; therefore, we generated a thyB- thyA+ mutant strain. Tmpr colonies of this strain were produced with a spontaneous mutation frequency of ~1.4 × 10−9. Genetic disruption of the canonical mismatch (MMR) and guanine oxidized (GO) repair pathways increased the Tmpr frequency of mutation by ~2–3 orders of magnitude. A wide spectrum of base substitutions as well as insertion and deletions in the ORF of thyA were found to confer a Tmpr phenotype. Stationary-phase-associated mutagenesis (SPM) assays revealed that colonies with a Tmpr phenotype, accumulated over a period of ten days with a frequency of ~ 60 ×10−7. The Tmpr system was further modified to study TMM by constructing a ΔthyA ΔthyB strain carrying an IPTG-inducible Pspac-thyA cassette. In conditions of transcriptional induction of thyA, the generation of Tmpr colonies increased ~3-fold compared to conditions of transcriptional repression. Further, the Mfd and GreA factors were necessary for the generation of Tmpr colonies in the presence of IPTG in B. subtilis. Because GreA and Mfd facilitate transcription-coupled repair, our results suggest that TMM is a mechanim to produce genetic diversity in highly transcribed regions in growth-limited B. subtilis cells.


Strains construction
Knockouts in the genes of interest were constructed by marker exchange between the chromosome-and plasmid-borne alleles. A plasmid to disrupt thyA was constructed as follows, a 493-bp fragment from the 5'-thyA region (-300 to +164 relative to thyA start codon) and a 473-bp fragment from the 3'-thyA region (+600 to +1073 relative to thyA start codon) were PCR amplified using chromosomal DNA from B. subtilis 168. The set of oligonucleotide primers 3, 4 and 5, 6 ( Table 2) were used for amplification of the 5'-thyA and 3'-thyA fragments, respectively. The amplified 5'-thyA and 3'-thyA fragments were cloned between the SphI/SalI and BamHI/SalI sites of pBEST502 [29]. The resulting construction pPERM1014 was replicated in E. coli XL10-GOLD Kan r (Stratagene, Cedar Creek, TX) and employed to transform the strain B. subtilis YB955 to neomycin resistance, thus generating strain B. subtilis PERM 1000 (Table 1). A ΔthyB strain was generated by amplifying a fragment from position 38 to 675 from the thyB open reading frame was amplified by PCR using chromosomal DNA from B. subtilis YB955 and the oligonucleotide primers 5 and 6 ( Table 2). The PCR product was cloned between the HindIII and BamHI sites of the integrative plasmid pMUTIN4 [30]. The resulting construct was designated pPERM1013 and used to transform B. subtilis YB955 and PERM 1000, thus generating strains PERM1037 (ΔthyB; Em r ) and PERM1074 (ΔthyA ΔthyB; Neo r Em r ), respectively (Table 1).
The strain containing the thyA gene under the control of an IPTG-inducible promoter was constructed as follows. The open reading frame (ORF) of thyA was PCR amplified from B. subtilis 168 chromosomal DNA and the oligonucleotide primers 7 and 8 ( Table 2). The PCR product (1,194 bp) was purified from a low-melting-point agarose gel and cloned between the SalI and SphI sites of the amyE integrative vector pdr111 (a gift from David Rudner), immediately downstream of the IPTG-inducible Phyperspank promoter (Phs). The resulting plasmid Stress-associated B. subtilis Tmp r mutagenesis (pPERM1099) was replicated in E. coli XL10-GOLD (Stratagene). This construct was transformed and integrated into the amyE locus of B. subtilis PERM1074 (ΔthyA thyB) to generate the strain B. subtilis PERM1100 (ΔthyA ΔthyB; amyE::Phs-thyA; Neo r Em r Spc r ) ( Table 1). As an experimental control, the empty pdr111 vector was also recombined in the amyE locus thus generating the strain B. subtilis PERM1075 (ΔthyA ΔthyB; amyE::Phs; Neo r Em r Spc r ) ( Table 1).
An integrative plasmid designed to inactivate greA was constructed as follows. A fragment of greA was first amplified by PCR using chromosomal DNA from B. subtilis 168 and the set of oligonucleotide primers 9 and 10 ( Table 2). The 222-bp PCR product, extending from position 115-bp to 318-bp downstream of the greA start codon was cloned between the EcoRI-BamHI sites of the integrative plasmid pMUTIN4-Cat [11]. The resulting construct designated pPERM 1191was used to transform B. subtilis PERM1100, generating strain PERM1192 (ΔthyA ΔthyB ΔgreA; amyE::Phs-thyA; Neo r Em r Cm r Spc r ) ( Table 1). The single-or double-crossover events leading to inactivation of the appropriate genes and integration of transcriptional cassettes were corroborated by PCR with specific oligonucleotide primers.

Stationary-phase mutagenesis soft-agar overlay assays
Cultures were grown in flasks containing antibiotic A3 medium with aeration (250 rpm) at 37˚C to 90 min after the cessation of exponential growth (designated T 90 ). Growth was monitored with a spectrophotometer measuring the optical density at 600nm (OD 600 ). The cultures were centrifuged at 10, 000 × g for 10 min and resuspended in 10 ml of 1X Spizizen Minimal Salts (SMS) [31]. Aliquots of 0.1 ml were then spread plated on Spizizen minimal medium (SMM) containing 1X SMS, 0.5% (w/v) dextrose, 50 μg isoleucine ml -1 , 50 μg glutamate ml -1 , 1.5% (w/v) agar (BD, Bioxom), 50 μg histidine ml -1 of and a trace amount (200 ng ml -1 ) of leucine and methionine (amino acids from Sigma-Aldrich, St. Louis, MO), with or without 0.07 mM IPTG. The initial titer was determined from the 10-ml culture. Starting from 48 h of incubation, a set of plates was overlaid with soft agar (0.7% (w/v) agar and prewarmed at 42˚C lacking histidine and containing 0.07 mM IPTG, 50 μg thymine ml -1 , 10 μg Tmp ml -1 , 50 μg methionine and leucine ml -1 . Of note, adjustment of the final concentrations for IPTG, methionine and leucine considered the volume and IPTG concentration in the medium dispensed previous to performing the overlay. The plates were incubated for two days, and the number of Tmp r colonies was scored. The initial titers were used to normalize the cumulative number of resistant colonies per day to the number of CFU plated. Assays were replicated three times, and the experiment was repeated at least twice. The viability of the non-revertant cell background was assessed every other day as follows. Using a sterile Pasteur pipette, a plug of agar was removed from the non-revertant background of each of five plates corresponding to one type of selection (no leucine and methionine without IPTG or no leucine and methionine with IPTG). The plugs were combined in 0.4 ml of 1X SMS, serially diluted, and plated in triplicate on SMM containing 50 μg ml -1 of the required amino acids. Colonies were counted after 48 h of incubation.

Phenotypic analysis of B. subtilis thy mutants
Trimethoprim resistance of the strains of interest was analyzed on solid SMM containing 1X SMS, 0.5% (w/v) dextrose, 50 μg ml -1 isoleucine, glutamate, histidine, methionine, leucine and thymine. All plates contained thymine at 50 μg ml -1 and the concentration (μg/ml) of trimethoprim indicated in Table 3. When necessary, the media was supplemented with IPTG to a final concentration of 0.1 mM. Growth was scored after 24 hr at 37˚.

Analysis of mutation frequencies to Tmp r and Rif r
Essentially, the appropriate strains were propagated for 12 h at 37˚C in A3 medium with proper antibiotics. For Tmp r , mutation frequencies were determined by plating aliquots on six LB plates containing 10 μg ml -1 trimethoprim and 50 μg ml -1 thymidine, and the trimethoprim-resistant (Tmp r ) colonies were counted after 2 days of incubation at 37˚C. The same procedure was employed to determine the Rif r phenotype, except that mutant colonies were selected in LB plates containing 10 μg ml -1 rifampin. The number of cells used to calculate the frequency of mutation to Tmp r or Rif r was determined by plating aliquots of appropriate dilutions on LB plates without antibiotics and incubating the plates for 24 to 48 h at 37˚C. These experiments were repeated at least three times.

Total RNA extraction
Cultures were grown to saturation in 1X SMS containing 0.5% (w/v) dextrose; 50 μg ml -1 of Ile, Glu, methionine, histidine, and leucine; Ho-Le trace elements; 5 mM MgSO 4 ; and 1 mM IPTG. Samples were removed at mid-exponential (approximately OD 600nm = 0.5) and stationary (150 min after onset of the stationary phase; T 150 ) growth phases, centrifuged at 5, 000 × g and 4˚C for 10 min and frozen at -20˚C. RNA was extracted from the pellets using a Tri reagent (Molecular Research Center, Inc. Cincinnati, OH). After DNAse I treatment, the samples were analyzed by PCR to confirm the absence of genomic DNA with Vent DNA polymerase and the set of primers 11 and 12 described in Table 2. The RNA content in the samples was quantitated using a Nanodrop spectrophotometer (Thermo Fisher Scientific, Dubuque, IA).

Reverse transcription and quantitative real-time PCR
One microgram of RNA was reverse transcribed using an ImPromII reverse transcriptase kit (Promega), as directed, with random hexamers (0.5-μg final concentration). No-reverse transcriptase (NRT) controls were included for all examples. Master mixes for real-time PCR contained Absolute QPCR SYBR green Mix (Thermo Fisher Scientific) and a 70 mM final concentration of the oligonucleotide primers thyART FW and thyART RV (191-bp amplicon) or of veg FW and veg RV (82-bp amplicon) described in Table 2. Three replicates from each culture condition containing 4 μl of cDNA were assayed and normalized to the expression of the internal control gene veg [32,33]. Two replicates were assessed for NRT and no-template controls. Quantitative real-time PCR was run on a Bio-Rad iCycler iQ Real-Time PCR Detection System (Bio-Rad, Hercules, CA), using the manufacturer´s suggested protocol and an thyA -thyB -; amyE:: Phs-thyA _a Strains were grown on solid Spizizen Minimal Medium (SMM) as described in Materials and Methods. ++, full growth; +, weak growth; -, no growth. b For this strain, solid SMM was supplemented with 0.1mM IPTG. https://doi.org/10.1371/journal.pone.0179625.t003 Stress-associated B. subtilis Tmp r mutagenesis annealing temperature of 57˚C, followed by a melting profile and assessment of amplicon size on an agarose gel. Results were calculated by the 2 -ΔΔC(T) (where C T is threshold cycle) method for relative fold expression [34].

DNA sequencing
Colonies with a Tmp r phenotype were independently propagated in liquid A3 medium supplemented with 10 μg/ml trimethoprim and 50 μg ml -1 thymidine and subjected to DNA isolation [27]. The thyA open reading frame of each colony was amplified by PCR using high fidelity and specific oligonucleotide primers (Table 2). Sequencing services were carry out by Functional Biosciences, Inc. (Madison, WI).

Results and discussion
Spontaneous mutation frequencies to trimethoprim resistance in growing B. subtilis cells To implement a loss-of function system to study mutagenesis in B. subtilis, the thyA gene that codes for TMS was considered as a possible target [23,24]. In B. subtilis the presence of ThyA prevents the incorporation of exogenous thymine into DNA synthesis [20]; therefore, strains deficient for this activity incorporate this metabolite more efficiently than ThyA-proficient strains [20]. Of note, thyA mutants of B. subtilis are able to grow in medium supplemented with DHFR inhibitors such as trimethoprim; if thymine is present in the culture medium [20], spontaneous colonies with a Tmp r phenotype can be selected [20,24,35]. This phenomenon has been described in several bacteria, including Escherichia coli [35][36][37]. B. subtilis possesses two TMS encoding genes; i.e., thyA and thyB respectively [20]. It has been reported that thyB expression takes place when B. subtilis is grown under temperatures below 37˚C contributing with only 5-8% of the total TMS activity present in this bacterium [23]. Since thyB is the first cistron of the thyB-dfrA-ypkP operon, we employed a gene construct (pPERM1013; Table 1) to only disrupt thyB and left dfrA-ypkP under control of the IPTG-inducible Pspac promoter. Under our experimental conditions, no significant differences were observed in the growth rates of this mutant with respect to those of the parental strain YB955. Thus, in A3 medium, in the absence or presence of 0.1 mM IPTG the thyB strain grew with similar doubling time values as those of its parental strain (i.e., 31 ± 2, 30 ± 2.5 and 31 ± 1.8 min, respectively); furthermore, this strain was incapable of growing in minimal medium supplemented with trimethoprim and thymidine ( Table 3). The ΔthyB strain was first used to analyze the spontaneous appearance of colonies resistant to trimethoprim in B. subtilis cells. As shown in Fig 1, colonies with a Tmp r phenotype appeared, in solid minimal medium supplemented with thymidine, with a spontaneous mutation frequency of~1.4 ± 0.11 × 10 −9 . These results are in good agreement with the mutation frequencies reported for the strain B. subtilis YB955 using the rpoB gene as a marker of mutagenesis [10,14]. Previous reports, employing the Rif r phenotype demonstrated that inactivation of the canonical mismatch (MMR) and guanine-oxidized (GO) repair systems increased, two and three orders of magnitude, respectively, the spontaneous mutagenesis in growing B. subtilis cells [8,9]. As shown in Fig 1, the genetic inactivation of the GO system, which in B. subtilis is composed of the MutY, MutM and YtkD proteins [9,38] increased the mutation frequency to Tmp r in the ThyB-deficient B. subtilis strain around 1000 times. Furthermore, our results revealed that disruption of the mutSL operon, encoding the MMR system, increased the frequency of mutation to trimethoprim resistance in the ΔthyB strain around 30 times. As shown in Fig 1, the mutation frequencies as determined with the rpoB marker in the mutSL and GO-deficient strains were similar to those determined with the thyA gene. Therefore, the loss-of-function mutagenesis system described in this report is robust and can be successfully employed to study the spectrum of mutagenic events that occur in thyA in a population of B. subtilis cells.

Mutational spectra of thyA in colonies exhibiting a Tmp r phenotype
We further investigated the mutagenic events associated with the loss of TMS function in the thyB-deficient strain. It was recently shown, that the heterologous expression of thyP3, a homologue of thyA, from the ITPG-inducible Pspac promoter, resulted in the production of mutations in the -5 region of Pspac conferring trimethoprim resistance in actively growing B. subtilis cells [39]. In this study, we were interested in identifying mutations affecting the function of ThyA under conditions of native thyA expression; therefore, the thyA gene of 40 colonies exhibiting a Tmp r phenotype was PCR amplified with high-fidelity DNA polymerase. All the samples produced amplification DNA bands of the expected size for the thyA ORF (namely, 837 bp), which were subsequently subjected to DNA sequencing to identify the type of mutations conferring the Tmp r phenotype. A wide spectrum of frameshift mutations and base substitutions were detected in the sequenced thyA mutants, predominating the insertions/deletions over the base substitutions in a proportion of 60% to 40% (Table 4 and Figs 2  and 3). Among the substitutions, 63% corresponded to transversions and 37% to transitions events, predominating the A!T (~29%) and A!G (~17%) changes (Table 4). Of note, three of the base changes generated non-sense mutations producing truncated ThyA proteins (Table 4 and Fig 2). It was found that most of the missense mutations in thyA resulted in nonconserved amino acid changes, moreover, two of the substitutions changed an alanine residue for the secondary structure-disrupting amino acid proline and one of them switched an isoleucine for the bulky aromatic residue phenylalanine (Table 4 and Fig 2). Importantly, 14 of the 17 missense mutations took place in codons encoding amino acids conserved in both thymidylate synthases (Table 4 and Fig 2). However, additional mutations in residues non-conserved in E. coli ThyA, were identified in the enzyme from B. subtilis, including, Ala 68 !Asp, Trp 31 ! Arg and Ala 68 !Pro (Fig 2). Thus, in addition to Arg 181 previously reported as essential for ThyA activity in E. coli [40], our results identified additional residues necessary for the proper function of ThyA in B. subtilis.
As noted above, insertion/deletions were also detected in the thyA sequence of colonies exhibiting a Tmp r phenotype. As shown in Fig 3, the insertion events took place between the positions 553-562 of the thyA ORF and consisted of single base additions, including, 4 adenines 2 thymines or 1 guanine. On the other hand, the frameshift deletions that took place between the positions 455-458 of the thyA ORF, involved the single loss of 7 thymines or the dinucleotide guanine-adenine (Fig 3). Notably, the insertion and deletions occurred in the base repeats GGGAAA and GAATT, respectively, and gave rise to truncated non-functional ThyA proteins. Thus, replication errors in these sequences that escaped the action of the mismatch repair machinery may be involved in generating these types of mutations. In support of this notion, the genetic inactivation of the MMR system (MutSL) in the ΔthyB strain increased by about two orders of magnitude the mutation frequency to trimethoprim resistance (Fig 1). Altogether, our results revealed that a wide spectrum of mutagenic events may lead to loss of ThyA function corroborating that the Tmp r system informs on the genetic events underlying stress-associated mutagenesis in B. subtilis.
Analysis of stationary-phase associated mutagenesis to trimethoprim resistance in nutritionally stressed B. subtilis cells We employed the ΔthyB strain to investigate whether loss-of thymidylate synthase (ThyA) function leading to Tmp r resistance can be employed to measure adaptive mutagenesis in B. subtilis.
To this end, B. subtilis cells of the ThyB-deficient strain collected from the stationary-phase of growth were extensively washed to eliminate residual nutrients and plated in selective Spizizen Stress-associated B. subtilis Tmp r mutagenesis Minimal Medium (SMM) as described in Materials and Methods. The number of colonies that acquired a Tmp r phenotype under growth-limited conditions was scored every two days for a ten days period. The results revealed that Tmp r colonies from the ΔthyB thyA + accumulated over a period of ten days with a frequency of~65 ± 11 × 10 −7 (Fig 4A), demonstrating the feasibility  Stress-associated B. subtilis Tmp r mutagenesis of employing this loss-of-function system to study mutagenesis in non-growing B. subtilis cells.
Notably, the increase in the number of Tmp r colonies took place without significant changes in the viable cell count number providing thus additional support for this assumption (Fig 4B).

Transcription-associated mutagenesis analysis in growth-limited B. subtilis cells acquiring a Tmp r phenotype
Previous studies have reported on the role of transcription in stationary-phase mutagenesis in B. subtilis [12]. It has been shown that a combination of transcriptional derepression and error-prone repair events in stress conditions can modulate the generation of mutations in highly transcribed DNA regions [13,41].
To investigate whether acquisition of Tmp r resistance can be used to study transcriptionassociated mutagenesis (TAM) in nutritionally stressed bacteria we engineered a double thyA thyB knock out strain that overexpressed a wild type copy of thyA from the IPTG-inducible Phyperspank (Phs) promoter; this strain was termed B. subtilis PERM1100 ( Table 1). As specified in Table 3, the thyA thyB mutant was able to grow in SMM supplemented with Tmp and thymidine. However, IPTG-induction of thyA expression abrogated the growth of the strain PERM1100 in this medium. Further, the lack of two amino acids (methionine and leucine) in the incubation agar media prevented the growth of cells. Therefore, the strain's properties and the double amino acid starvation allowed us to inquire whether derepression of thyA in growth-limited cells promote mutagenic events influencing the production of Tmp r colonies. As shown in Fig 5A in reference to the non-induced condition, derepression of the Phs-thyA construct resulted in a 3-fold increase in the production of colonies with a Tmp r phenotype.
Our loss-of function mutagenesis system indicate that increased transcription levels of thyA correlated with an increased production of colonies with a Tmp r phenotype in the ΔthyAthyB amyE::Phs-thyA strain. Of note, these results are in good agreement with former studies showing a positive correlation between derepression of a leuC allele and production of Leu + prototrophs in B. subtilis [13]. Furthermore, viability of the strain PERM1100 did not significantly change in the absence and/or presence of the inducer IPTG ( Fig 6A); therefore, the increase in the number of Tmp r can be uncoupled from the growth of the sensitive strain in the plate. To confirm that accumulation of Tmp r mutants were due to increases in transcription levels of Stress-associated B. subtilis Tmp r mutagenesis thyA, the mRNAs of this gene were quantified by qRT-PCR in stationary-phase cells of strain PERM1100 under induced and non-induced conditions. The results of this analysis showed that the mRNA levels of thyA increased over 3-fold (i.e. from 0.5 ± 0.06 to 4.5 ± 0.4) when IPTG was added to the medium. Therefore, the loss-of-function system implemented here can be successfully employed to investigate the mutagenic processes occurring in transcriptionally active DNA regions that allow cells to escape from non-proliferating conditions. Of note, the discovery of missense and nonsense mutations affecting B. subtilis ThyA function (Fig 2), will allow to adapt this system to determine the production of thyA + revertants in nutritionally stressed thyA -thyBcells overexpressing point-mutated alleles of thyA.
A recent study revealed that Mfd, a protein that couples DNA repair with transcription, promotes mutagenic events that increase the production of His + , Met + and Leu + revertants in starved cells of strain B. subtilis YB955 [12,13]. Here, we inquired whether the Mfd requirement for transcription-associated SPM can be detected with the loss-of-function Tmp r mutagenesis system. To this end, the gene mfd was genetically inactivated in the ΔthyA thyB strain bearing the Phs-thyA construct. The number of Tmp r colonies produced by this strain under starving conditions was determined over a period of ten days under conditions where the thyA gene was repressed or derepressed for transcription. As shown in Fig 5B, in reference to the strain harboring an intact copy of Mfd, the number of adaptive Tmp r colonies produced by the Mfd-deficient strain did not significantly differ under conditions of repression or derepression for the thyA gene. Therefore, employing a system that overexpresses a different gene; namely, thyA, our results provide further support for the concept that Mfd is a factor required for TMM in growth-limited B. subilis cells. As noted above, NusA, the elongation factor of the RNA polymerase has been found to be necessary for stress-induced mutagenesis in Escherichia coli [16]. Remarkably, disruption of greA in the thyA thyB strain abrogated the production of Tmp r colonies under conditions that induced thyA expression. Thus, as shown in Fig 5C, the levels of Tmp r mutagenesis in this strain were almost similar under conditions that induced or repressed thyA. Importantly, these results took place in the absence of growth since viability of the GreA-deficient strain did not significantly change in the absence and/or presence of the Stress-associated B. subtilis Tmp r mutagenesis inducer IPTG (Fig 6C). Based on these results is feasible to speculate that in nutritionally stressed B. subtilis cells, processing of paused or backtracked RNAP-DNA complexes promote mutagenic events that allow cells to escape from growth-limited conditions. Interestingly, in E. coli, both Mfd and GreA are speculated to prevent UvrD-dependent RNAP backtracking and repair during transcription [42]. However, the consequences of such transcriptional events on growth and stress-associated mutagenesis remain to be elucidated.
In summary, the mutagenesis method implemented here and the contribution of GreA to TMM in B. subtilis cells add further elements to our understanding on how bacteria develop beneficial mutations, including antibiotic resistance, under stressful conditions.