|Home | About | Journals | Submit | Contact Us | Français|
Overexpression of the Fst toxin in Enterococcus faecalis strain OG1X leads to defects in chromosome segregation, cell division and, eventually, membrane integrity. The M7 mutant derivative of OG1X is resistant to most of these effects but shows a slight growth defect in the absence of Fst. Full-genome sequencing revealed two differences between M7 and its OG1X parent. First, OG1X contains a frameshift mutation that inactivates the etaR response regulator gene, while M7 is a wild-type revertant for etaR. Second, the M7 mutant contains a missense mutation in the rpoC gene, which encodes the β′ subunit of RNA polymerase. Mutagenesis experiments revealed that the rpoC mutation was primarily responsible for the resistance phenotype. Microarray analysis revealed that a number of transporters were induced in OG1X when Fst was overexpressed. These transporters were not induced in M7 in response to Fst, and further experiments indicated that this had a direct protective effect on the mutant cells. Therefore, exposure of cells to Fst appears to have a cascading effect, first causing membrane stress and then potentiation of these effects by overexpression of certain transporters.
The only type I PSK system described for Gram-positive organisms is the par determinant of the pheromone-responsive, conjugative pAD1 plasmid of Enterococcus faecalis (16, 17). par encodes two RNA molecules, RNAI and RNAII, which serve as the toxin and the antitoxin of the system, respectively. When translated, RNAI produces a 33-amino-acid peptide toxin called Fst (18). Fst is the founding member of the Fst/Ldr peptides, a large family of RNA-regulated toxins that are widespread on plasmids and chromosomes of low-G+C Gram-positive bacteria as well as the Gram-negative enterobacteria (11, 12, 19). These peptides are characterized by the presence of a hydrophobic putative transmembrane domain and a highly charged C-terminal tail. Mutagenic analysis of Fst indicated that the hydrophobic domain is essential for toxicity while the C-terminal domain is dispensable (11). Recent nuclear magnetic resonance structural analysis of Fst in a membrane mimetic confirmed that amino acids 3 to 26 form a transmembrane domain with the relatively unstructured C-terminal tail protruding, likely into the cytoplasm (20).
Neither the target nor the mechanism of action of the Fst/Ldr peptide toxins is known. Unlike another RNA-regulated peptide toxin family, the Hok/Gef toxins, overexpression of Fst/Ldr toxins does not result in the formation of “ghost cells,” in which the cell center becomes translucent and the cell material appears to condense at the poles (21, 22). Rather, the primary effect of overexpression of Ldr in Escherichia coli (23) and Fst in E. faecalis, Bacillus subtilis, Staphylococcus aureus, and E. coli (11, 24) is the condensation of the nucleoid. In E. faecalis, nucleoid condensation is accompanied by aberrant chromosome segregation, irregular division septum placement, and loss of membrane integrity (24). The primary cause of these cell morphology changes is unclear.
In order to further illuminate the mechanism of action of the Fst toxin, genotypic and phenotypic analyses of the spontaneously isolated Fst-resistant M7 mutant were performed. DNA sequencing revealed two changes relative to the parental OG1X strain, one in the gene for the DNA response regulator etaR, and the other in the gene for the β′ subunit of RNA polymerase (RNAP), rpoC. Allelic exchange experiments revealed that resistance was primarily due to the latter mutation. Microarray analysis revealed that a number of transporters were induced late in OG1X after 60 min of exposure to Fst (45 min after the initial responses to Fst were observed). These transporters were not induced in M7 in response to Fst, and further experiments using reserpine (a broad-spectrum translocase inhibitor) and berberine (an antimicrobial indicator of efflux pump activity) established that the parent strain consequently possesses higher pump activities than M7 and that their inhibition is beneficial for growth of the parent strain. In conclusion, exposure of cells to Fst induces a nontranscriptional first event followed by later effects that include the upregulated expression of certain membrane transporter proteins.
Strains and plasmids used or constructed in this study are shown if Table 1. The E. faecalis strain used as the wild type in this study was OG1X (25), a streptomycin (Sm)-resistant, gelatinase-negative strain. E. faecalis strain M7 is a spontaneously Fst-resistant mutant derived from OG1X, as previously described (26). M7 shows nearly complete resistance to Fst induction and a growth defect in the absence of Fst. E. coli strain DH5α (Invitrogen, Grand Island, NY) was used in cloning procedures. E. coli and E. faecalis were routinely cultured in Luria-Bertani (LB) broth (27) and Todd-Hewitt Broth (THB; Sigma Aldrich, St. Louis, MO), respectively, at 37°C. Fst was produced in both OG1X and M7 cells and derivatives from the plasmid pAM2005K, an erythromycin-resistant, pAD1 miniplasmid in which the fst gene is fused to a pheromone cAD1-inducible promoter (28). The plasmid also contains a lacZ gene located between the promoter and fst, by which induction levels can be assessed. Strains containing pAM2005K were always cultured overnight in the presence of Erythromycin (Em), and cultures prepared for determination of growth curves for this work were Em free unless specified below, in order to observe the effects of toxin expression in the absence of possible complicating effects of the antibiotic.
When needed, the following antibiotics (all from Sigma Aldrich) were used: Em (for E. faecalis, 10 μg ml−1, for E. coli, 100 μg ml−1), Sm (1,000 μg ml−1), rifamycin (Rif; 100 μg ml−1), fusidic acid (Fus; 25 μg ml−1), and ampicillin (Amp; 100 μg ml−1). Isopropyl-β-d-thiogalactopyranoside (IPTG; 0.033 mM; Sigma Aldrich) and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal; 40 μg ml−1; Gold Biotechnology, St. Louis, MO) were used for the screening of pGEMTeasy (Promega, Madison, WI) clones and allelic exchange experiments. Berberine and reserpine were obtained from Sigma-Aldrich, and stock solutions were prepared using water and dimethyl sulfoxide, respectively. Solid medium was produced by adding 1.5% agar (Fisher Scientific, Pittsburgh, PA). All sequencing of plasmid constructs and PCR products was performed at the DNA sequencing facility at Iowa State University (Ames, IA). In all growth curves, cultures were started with a 2% dilution from overnight cultures grown with appropriate antibiotics. Fst was induced from pAM2005K with 20 ng ml−1 synthetic cAD1 (Mimotopes, Victoria, Australia) added after 1 h of growth.
Isolation of pAM2005K from E. faecalis was achieved using a modified Qiagen protocol (29) and the QIAprep Spin miniprep kit (Qiagen, Valencia, CA). Plasmid DNA was isolated from E. coli DH5α cells by using either a miniprep kit from Bio-Rad or a midiprep kit from Qiagen. Genomic DNA was prepared using the following protocol. Ten-milliliter cultures were grown overnight and centrifuged (Damon/IEC Division HN-SII) at 17,000 × g for 10 min at room temperature. Cells were then washed in 10 ml of Tris-EDTA (TE) buffer and centrifuged again. The cells were resuspended in 200 μl of IHB-1 solution (1 M glucose, 0.5 M Tris [pH 8.0], 0.5 M EDTA [pH 8.0]). To the resuspensions, 50 μl of lysozyme (50 mg ml−1) and 10 μl of mutanolysin (2500U ml−1) (Sigma Aldrich) were added, and the mixtures were incubated at 37°C for 2 h. After the 2-h incubation, 100 μl of 20% Sarkosyl (Sigma Aldrich) and 15 μl of RNase (10 mg ml−1; Sigma Aldrich) were added to the cultures, and cultures were incubated for 30 min at 37°C. Fifteen microliters of pronase (10 mg ml−1; Sigma Aldrich) was then added, and the samples were incubated for an additional 30 min at 37°C. Following the incubation, the final volume of each sample was brought up to 600 μl using TE buffer. Phenol-chloroform extraction was performed using a 1:1 ratio (final volume, 600 μl) five times, followed by a single chloroform extraction. DNA was precipitated at −20°C for at least 30 min with 500 μl of isopropanol and 50 μl of 3 M sodium acetate and centrifuged at top speed for 10 min at 4°C. DNA was washed with 700 μl of 70% ethanol and centrifuged for an additional 5 min, and samples were allowed to dry.
Routine PCR was performed using 45 μl of PCR HiFi Supermix (Invitrogen) and combining it with 2 μl of each primer (5 μM) and 1 μl of template DNA. Thermal cycling conditions were as follows: 2 min at 94°C, followed by 35 cycles of 45 s at 94°C, 45 s at 42°C, and 1 min at 72°C, with a final extension for 10 min at 72°C, on a Mastercycler Personal PCR machine (Eppendorf, Hauppauge, NY). For PCRs that required more stringent conditions due to product size or secondary structure, a Platinum PCR kit from Invitrogen was used. The reaction mixture was assembled per the manufacturer's instructions and performed under the following conditions: 1.5 min at 94°C, followed by 35 cycles of 30 s at 94°C, 30 s at 55°C, and 1 min/kb at 68°C. Colony PCR was used to screen colonies from allelic exchange experiments. Each reaction mixture contained the following: 2.5 μl of 10× Thermopol buffer (New England BioLabs, Ipswitch, MA), 0.2 μl of 25 mM solutions of deoxynucleoside triphosphates, 0.5 μl of 50 μM each primer, 0.5 μl Taq polymerase, and 21.3 μl distilled H2O. The colony PCRs were performed under the following conditions: 10 min at 95°C, followed by 30 cycles of 40 s at 95°C, 45 s at 50°C, and 1 min/kb at 72°C. Mismatch amplificiation mutation analysis-PCR (MAMA-PCR) (30) was used to screen colonies for single base changes (e.g., rpoC allelic exchanges). With this method, a single base pair change can be detected by using primers with either 1 or 2 mismatches on their 3′ ends. Primers with 2 mismatches will not form a PCR product, while those with only 1 mismatch will. The conditions for this PCR were as follows: 2 min at 95°C, followed by 30 cycles of 25 s at 95°C, 25 s at 61°C, and 1 min/kb at 72°C. All primers used are shown in Table 2 and were purchased from Sigma Aldrich.
Transformation of plasmid DNA into chemically competent E. coli DH5α cells was performed following the manufacturer's instructions (Invitrogen). E. faecalis electrocompetent cells were prepared as described previously (31). Electroporation was carried out under the following conditions: voltage, 2,000 V; capacitance, 25 μF; resistance, 200 Ω, using the GenePulser Xcell machine (Bio-Rad, Hercules, CA). Expression was for 90 min in THB at 37°C prior to plating on selective media. Electrocompetent EC1000 cells were prepared using the instructions for E. coli cells found in the GenePulser Xcell instruction manual. Conditions for electroporation of EC1000 cells were the same as with E. faecalis. Following electroporation, cells were resuspended in 1 ml of LB, and regrowth was allowed for 60 min at 37°C. Transformants were then selected on agar plates containing the appropriate antibiotic.
Allelic exchange in E. faecalis was performed essentially according to the protocol developed by Kristich et al. (32). Briefly, experiments proceeded in five steps. First, the allele to be introduced was constructed and cloned onto pCJK47, which cannot replicate in the recipient strain. Details of allele construction are described below. Second, the resulting construct was introduced into the recipient strain by conjugation from E. faecalis strain CK111(pCF10-101). Conjugation was modified from the Kristich protocol by performance for 2 h at 30°C. Third, single-crossover recombinants were isolated by selection for the pCJK47-carried Em resistance. Fourth, excisants that had deleted the plasmid via a second crossover were isolated by outgrowth in Em-free medium and counterselection on p-chloro-phenylalanine-containing plates. Excision was further confirmed by screening for loss of pCJK47-carried lacZ production on X-Gal-containing plates. Fifth, excisants containing either the donor or recipient allele were identified by PCR and confirmed by sequencing. Isolates with both exchanged and nonexchanged alleles were retained, the latter for use as control. For assessing sensitivity to Fst, pAM2005K was introduced into all strains by electroporation.
Construction of etaR deletion alleles was conducted by PCR amplification of approximately 800 bp of both the OG1X 5′ and 3′ ends of etaR, creating a frameshift and a stop codon at the 3′ end of the upstream fragment and deleting the central 685 nucleotides (nt) of the gene (primers 5′ 1050 Frag1 NotI and 3′ 1050 Frag1 SpeI for the 5′ fragment and 5′ 1050 Frag2 SpeI and 3′ 1050 Frag2 XmaI for the 3′ fragment). By using the restriction enzyme recognition sites introduced by the primers, fragments were directly cloned simultaneously into pCJK52. Restriction enzymes and T4 DNA ligase were used according to the manufacturer's instructions (New England BioLabs). The fused fragments were then subcloned from pCJK52 into pCJK47 using 5′NotI and 3′XmaI. The construct was electroporated into EC1000, isolated, and sequenced using the 5′ 1050 Frag1 NotI primer to ensure that no extraneous mutations had been introduced and then electroporated into E. faecalis CK111(pCF10-101) competent cells. Conjugation and excision were performed as described above, and the presence of full-length or knockout genes was determined using colony PCR (primers 5′ 1050 Frag1 NotI and 3′ 1050 Frag2 XmaI). Full-size etaR is ~2.5 kb, while the knockout is ~1.8 kb. Smaller PCR products were cleaned (Promega) and submitted for sequencing with the 5′ 1050 Frag1 NotI primer for confirmation.
Allelic exchange of rpoC was attempted in both the OG1X and M7 backgrounds. An ~840-bp PCR product containing the 3′ end of the rpoC gene was created for both wild-type and mutant alleles by using 5′ RpoC NotI and 3′ RpoC SphI (OG1X) or 3′ RpoC XmaI (M7) as primers with OG1X or M7 genomic DNA, respectively, as the template. The PCR was run using the Platinum PCR kit. The OG1X fragment was cloned into pGEM-T Easy according to the manufacturer's directions (Promega), transferred to pCJK47 using flanking EcoRI sites, and electroporated into EC1000 cells. The M7 fragment was directly cloned into pCJK47 using primer-introduced NotI and XmaI sites and then electroporated into EC1000 cells. Clones were screened by colony PCR using the corresponding NotI and XmaI primers mentioned above to detect the presence of the insert. When a positive clone was identified, plasmid DNA was purified, sequenced using the 5′ primer, and electroporated into CK111(pCF10-101) cells. Conjugation and excision were performed as described above. Clones were screened for the base pair change by MAMA-PCR with primers to detect OG1X rpoC (WT RpoC MAMA and 5′ RpoC MAMA Check) or M7 rpoC (M7 RpoC MAMA and 5′ RpoC MAMA Check). Once specific clones were identified by colony PCR, a separate fragment was made using 3′RpoC SphI and 5′RpoC NotI primers and sent for sequencing with the 3′ RpoC SphI primer to confirm the presence of the base pair change and that no unintended mutations had been introduced.
RNA for microarray analysis was prepared by growing a 2% dilution from overnight cultures of selected strains for 1 h, removing a 10-ml sample (zero time), and inducing the remaining culture with 20 ng ml−1 synthetic cAD1. Initially, OG1X(pAM2005K) 10-ml samples were removed 10, 30, and 60 min following induction of Fst. For M7(pAM2005K), only time zero and 60-min samples were analyzed, since few changes were observed in earlier samples in OG1X(pAM2005K). After removal of each time point, samples were mixed with an equal volume of 1:1 acetone-isopropanol and stored at −80°C. For RNA isolation, samples were thawed on ice and centrifuged at 17,000 × g at 4°C for 10 min and resuspended in 500 μl TE buffer (10 mM Tris-HCl and 1 mM EDTA). Resuspensions were transferred to lysing matrix B tubes (MP Biomedicals, Santa Ana, CA) and lysed using two cycles in an FP120 shaker (Thermo Scientific, Waltham, MA) at settings of 5.0 and 4.5 m s−1 for 20 s. Cell debris was removed by centrifugation at 17,000 × g at 4°C for 15 min in a microcentrifuge, and the supernatant was used for RNA isolation with the RNeasy kit following the manufacturer's protocol for prokaryotic RNA purification (Qiagen). RNA was quantitated by spectrophotometry at 260 nm on a SmartSpec 3000 apparatus (Bio-Rad). To prepare the RNA for microarray analysis, the RNA was reverse transcribed, fragmented, 3′ biotinylated, and hybridized to an E. faecalis GeneChip according to the manufacturer's protocol for antisense prokaryotic arrays (Affymetrix, Santa Clara, CA). The GeneChips used in these studies were custom-made Affymetrix E. faecalis GeneChips based on strain V583 (courtesy of M. Gilmore) (33). The signal intensity value for each RNA species interrogated on the GeneChip was normalized to the total signal across the entire microarray; biological replicates (at least duplicates) were averaged and compared using GeneSpring software. RNA transcripts exhibiting significant changes of ≥2-fold (Student's t test, P ≤ 0.05) in the titer between OG1X and M7 cells are reported.
Oligonucleotides and dually labeled probes (Table 3) were designed using Primer Express 2.0 software (ABI Prism; PE Biosystems, Framingham, MA) and ordered from Sigma Aldrich with 5′–6-carboxyfluorescein and 3′–6-carboxytetramethylrhodamine as the dye and quencher, respectively. RNA was collected from 3 separate experimental samples for each strain and prepared as described above using the RNeasy kit from Qiagen, quantitated using an RNA chip from the RNA 6000 nanokit (Agilent, Santa Clara, CA), and run on the Agilent 2100 BioAnalyzer according to the manufacturer's protocols. Amplification and detection were done using the Applied Biosystems StepOne Plus reverse transcription-PCR (RT-PCR) system and TaqMan One-Step RT-PCR master mix (Applied Biosystems, Carlsbad, CA) as recommended by the manufacturer.
Sequencing was performed at the University of Washington Center for Array Technologies, in the Bumgarner laboratory. The sequences were determined using 454 sequencing (34) and compared to the reference sequence, that of E. faecalis OG1RF (35). Assembly was performed using the Newbler Assembler, version 2.0.00.19 (454 Life Sciences, Roche Diagnostics Corporation, Branford, CT), which generated 79 contigs with lengths greater than 200 bp, and 99.72% of the bases had quality scores of 40 and above.
The OG1X sequence was assigned GenBank accession number AFHH00000000, and M7 was assigned GenBank accession number AGVN00000000.
Spontaneous Fst-resistant mutants of E. faecalis strain OG1X were previously isolated and shown to be resistant to growth inhibition in response to Fst induction from the pheromone (cAD1)-inducible promoter of pAM2005K (26). To determine the genetic changes responsible for Fst resistance, complete genomic sequencing of OG1X(pAM2005K) and M7(pAM2005K) was performed with comparisons to E. faecalis OG1RF (36) as a reference sequence (35) and to each other. To identify the high-confidence differences between M7 and OG1RF, a previously developed computational protocol was used (37). In brief, reads from each genome were mapped to the assembled contigs of the other genome to identify regions in which multiple reads from one genome differed in sequence from the consensus sequence of the other genome. Since the consensus sequence will have a small number of errors due to both assembly errors and low redundancy in some regions, hcSNPs (high-confidence single-nucleotide polymorphisms) were sought, and were found at the same positions in the reciprocal mapping. This protocol essentially allows screening for differences between two genomes that are supported by multiple raw sequencing reads in both genomes. The results were filtered to retain only high-confidence differences that were supported by at least 10 reads from each genome, where at least 80% of the reads spanning that region showed the same difference.
For the OG1X versus M7 genome comparison, only 5 hcSNPs were identified between the two strains. Two were G deletions adjacent to 6-nt poly(A) tracts in OG1RF_11774 and OG1RF_10534 (locus designations according to the OG1RF sequence). Sequencing of PCR products produced using primers 5′ and 3′ OG1RF_11774 Check and 5′ and 3′ OG1RF_10534 Check showed the same sequence as OG1RF for both OG1X and M7, suggesting that these were pyrosequencing errors. The mutation in OG1RF_11764 (C4824A) was not studied further because there was no change in the predicted amino acid sequence (proline; CCC → CCA).
The two remaining mutations were confirmed through sequencing of PCR products using primers with the gene designation followed by “Check” (Table 2). One mutation was in etaR (OG1RF_10783), the DNA response regulator of a two-component signal transduction system (EtaRS) (38). The mutation, a single nucleotide deletion at C96 of the coding sequence, occurs in OG1X, changing the reading frame and resulting in termination shortly downstream (Y32STOP). This mutation results in a truncation of 196 amino acids, deleting the receiver domain of the response regulator. M7 contains a reversion back to the OG1RF sequence and therefore is predicted to carry a fully functional etaR. The second mutation was a missense mutation in rpoC (OG1RF_12492) of M7 cells, which encodes the RNAP β′ subunit. The mutation (C2258A) results in an A753D substitution in the amino acid sequence. The mutation occurs in domain 4 of β′, which represents the funnel domain and serves as a binding site for various elongation factors necessary for proper transcription (39, 40). Genomic differences between OG1X and M7 are summarized in Table 4.
In addition to differences between OG1X and M7, mutations likely responsible for Sm resistance and the gelatinase defect conserved in both OG1X and M7 were identified. Mutations responsible for Rif and Fus resistance in OG1RF were also identified, increasing our confidence that all sequence changes of interest were detected. These mutations are shown in Table S1 of the supplemental material.
To determine the relative contributions of the etaR and rpoC mutations to Fst resistance, allelic replacements were performed using the method of Kristich et al. (32). Three possibilities were considered. First, the M7 phenotype could require both a wild-type etaR as well as the rpoC mutation for resistance, with neither providing significant resistance alone. Alternatively, both mutations may be independent of one another, contributing additively to the resistance phenotype. Finally, one or the other mutation could play the predominant role.
To determine the effects of the wild-type etaR and how it relates to Fst resistance, an etaR deletion was constructed and introduced into M7, creating M7ΔetaR. This alternative etaR allele also contained a frameshift mutation, as in OG1X, maintaining any possible polar effects that might be observed on the downstream etaS gene. Induction of Fst production in M7ΔetaR(pAM2005K) showed no detectable decrease in resistance in comparison to M7(pAM2005K) (Fig. 1), indicating that etaR plays little or no role in resistance to Fst in this strain.
To perform allelic exchanges of rpoC, a strategy was required that would maintain a functional rpoC gene at both the first crossover and excision step, since disruption or deletion would likely be lethal. To accomplish this, a fragment of approximately 840 nucleotides, including nucleotide 2258 and the 3′ end of rpoC, was made. The inclusion of the 3′ end allowed reconstitution of a functional rpoC gene at the single-crossover stage. Multiple attempts to replace the OG1X rpoC allele, designated rpoC(OG1X), with the M7 allele, rpoC(M7), on the OG1X chromosome were unsuccessful. All excisants examined retained the rpoC(OG1X) allele. If rpoC(M7) is responsible for the known growth defect of M7 (26), the failure to successfully replace the rpoC(OG1X) allele could be due to the competitive advantage of wild-type excisants during the outgrowth stage of the allelic exchange protocol (see Materials and Methods). If this were the case, it would be easier to replace rpoC(M7) on the M7 chromosome with rpoC(OG1X). Indeed, such allelic replacements were readily isolated at a frequency of approximately 50% of excisants. The rpoC genotype was confirmed by sequencing, and this strain was designated M7:rpoC(OG1X). Unfortunately, it proved impossible to introduce pAM2005K into this strain. While electroporations of pAM2005K into M7 and OG1X produced 10 to 20 transformants per electroporation, showing that the pAM2005K plasmid preparation was of sufficient quality for successful electroporation, no M7:rpoC(OG1X) transformants were ever isolated despite multiple attempts. Electroporation of M7:rpoC(OG1X) with the vector pAM401 occurred with transformation efficiencies comparable to electroporation into M7 cells (data not shown), indicating that the M7:rpoC(OG1X) cells were electrocompetent. It is possible that the combination of wild-type etaR with rpoC(OG1X) resulted in enhanced sensitivity to even uninduced levels of Fst produced from pAM2005K.
Since it was not possible to test M7:rpoC(OG1X) for sensitivity to Fst, sensitivity to nisin was used as a surrogate for examining the effect of the rpoC mutation. Previously published data showed that M7 is cross-resistant to nisin, and so it was reasoned that the mutation in rpoC might be responsible for both nisin and Fst resistance (26). As shown in Fig. 2, M7:rpoC(OG1X) showed sensitivity to nisin at 10 μg ml−1, similar to OG1X(pAM2005K), and was much more sensitive than the parental M7 strain, supporting a role for rpoC(M7) in nisin resistance. The growth defect of M7 was also eliminated by rpoC replacement with the wild-type allele, as can be seen by comparing growth of the unexposed strains (Fig. 2).
In order to better mimic OG1X, the rpoC(M7) allele was replaced with rpoC(OG1X) on the M7ΔetaR chromosome, creating M7ΔetaR:rpoC(OG1X), nicknamed M7Δ:OG1X. Following allelic exchange and confirmation of both the etaR and rpoC genotypes via sequencing, pAM2005K was successfully electroporated into both control [M7ΔetaR:rpoC(M7), nicknamed M7Δ:M7] and experimental (M7Δ:OG1X) strains, with efficiencies comparable to those obtained with M7 and OG1X. Transformants were selected that were white in the absence of cAD1 but changed to blue in the presence of cAD1, indicating proper induction.
When induced with cAD1, growth of M7Δ:OG1X(pAM2005K) was significantly inhibited, similarly to OG1X(pAM2005K) (Fig. 3). M7Δ:M7(pAM2005K) did not show any changes in resistance to Fst compared with M7(pAM2005K) (data not shown). As was observed with M7:rpoC(OG1X), M7Δ:OG1X(pAM2005K) grew as well as OG1X(pAM2005K) (Fig. 3) and showed similar sensitivity to nisin as OG1X (data not shown). Similar results were obtained in the presence and absence of Em-mediated selection for plasmid maintenance, indicating that resistance was not due to accelerated plasmid loss in the rpoC mutant. These results indicate that rpoC(M7) is responsible for the M7 growth defect as well as resistance to both Fst and nisin.
The involvement of the β′ subunit of RNAP in Fst resistance suggested that resistance might be due to transcriptional changes in multiple genes. To examine this possibility, microarray analysis was performed on OG1X(pAM2005K) and M7(pAM2005K) cells before and after induction with cAD1. After 10 and 30 min of induction, in OG1X(pAM2005K) only minor changes in gene transcription were noted, other than those plasmid genes expected to be induced by pheromone (see Table S2 in the supplemental material). After 60 min of Fst induction in OG1X(pAM2005K), 25 gene transcripts were increased above 4-fold. The majority of those transcripts encoded membrane-associated proteins, including numerous transporters of various types (Table 5). These included a number of cation and metal transporters, ABC transporters of the glucan exporter and glycine-betaine families, and a major facilitator superfamily (MFS) transporter. The genes for Nramp and MFS transporters are linked to genes for a universal stress protein (USP) and a phosphotidylethanolamine binding protein (PEBP), respectively, which were also induced. Other induced proteins included a Zn ribbon DNA binding protein, a SNARE domain protein, and a phage shock protein. Glycine betaine is an environmentally available osmoprotectant (41), and the phage shock protein is induced by proton motive force dissipation (42); the other induced transporters could also be induced as an attempt by cells to maintain osmotic and ionic balances within the cell in response to some form of membrane stress. This would be consistent with the membrane localization of the Fst toxin (20) and previous results indicating that Fst-exposed cells become permeable to membrane-impermeant dyes after 45 to 60 min (24). It was previously observed that nisin and other membrane-active peptides induce a specific response by ABC transporters linked to two-component regulators (43). The ABC regulators induced by Fst showed no specific homology to these transporters and were not linked to two-component regulatory systems, and therefore represented a novel response.
In M7(pAM2005K), 16 non-plasmid-associated genes were induced over 4-fold, and only 4 were induced over 6-fold (see Table S3 in the supplemental material). Most of the induced genes were hypothetical proteins, and only one was a transporter. None of the transporters highly induced in OG1X(pAM2005K) was induced in M7(pAM2005K) in response to Fst. Induction of the pheromone-responsive positive regulator TraE1, which stimulates the pheromone response in pAM2005K, was actually higher in M7(pAM2005K) than in OG1X(pAM2005K) (compare Tables S2 and S3 in the supplemental material). This was consistent with previous results that showed induction of the pAM2005K lacZ gene to be approximately 2-fold higher in M7 than in OG1X (26) and confirmed that the mutation does not cause a defect in the pheromone response.
As shown above, introducing the rpoC(OG1X) allele in M7 decreased resistance to Fst to levels similar to OG1X cells. To determine if the rpoC substitution also restored the ability to induce transporters in the presence of Fst, qRT-PCR was performed on cAD1-induced and uninduced OG1X, M7Δ:M7, and M7Δ:OG1X cells containing pAM2005K. Representative active transporters were selected based on the previous microarray experiments. The transporters were induced in response to cAD1 in OG1X but not in M7 cells (Table 5). These transporters were EF1304 (OG1RF_11074), a magnesium-translocating P-type ATPase, and EF1673 (OG1RF_11385) and EF2074 (OG1RF_11677), both of which are ABC transporters (V583 locus designators are given to correlate to microarray data, but primers and probes were generated with the OG1RF sequence). Transcript levels of the three genes were normalized to gyrB, which was presumed to be constitutively expressed under these conditions and showed no change in microarray experiments.
Results showed that transcript levels of all three genes were induced in response to Fst in OG1X(pAM2005K), as indicated in microarray experiments (Fig. 4A, ,B,B, and andC).C). Furthermore, induction of these transcripts occurred similarly in OG1X(pAM2005K) and M7Δ:OG1X, but not in M7Δ:M7. Thus, replacing rpoC(M7) with rpoC(OG1X) restored the cells' ability to induce active transporters in response to cAD1, indicating that the mutation in M7 rpoC is responsible for the inability of M7 cells to induce energy-dependent transporters in response to Fst.
The microarray results demonstrated that M7 failed to induce a number of transporters in response to Fst. It is possible that the rpoC mutation prevents transporter induction indirectly by preventing membrane damage in an undetermined manner that was not detected by microarray analysis, thereby making transporter induction unnecessary. Alternatively, the rpoC mutation may affect transcription of the transporters directly, making M7 cells intrinsically unable to induce them in response to membrane stress. To distinguish these possibilities, the relative sensitivities of OG1X and M7 cells to berberine were determined. Berberine is a hydrophobic cationic antimicrobial and the primary component of the medicinal plants golden seal (Hydrastis canadensis) and echinacea (Echinacea spp.) (44). Resistance to this substance in bacteria is based on multidrug-resistant (MDR) pumps, which can readily expel it from the cell (45). It was reasoned that, if M7 cells were intrinsically unable to respond to stress by inducing protective transporters, they would be more sensitive to berberine than OG1X cells. In contrast, if the rpoC mutation made cells more generally resistant to membrane stress, the M7 cells would be more resistant to berberine. As shown in Fig. 5, while growth was inhibited in both strains by berberine, inhibition was significantly greater in M7 cells, indicating that the M7 rpoC mutation has a direct effect on transporter induction.
The above results suggested that failure to induce one or more transporters has a protective effect on Fst-exposed M7 cells. If so, chemical inhibition of transporters should have a protective effect on Fst-exposed OG1X cells. Reserpine is a broad-spectrum translocase inhibitor, able to inhibit MDR pumps ranging from NorA in Staphylococcus aureus (an MFS translocase) to human P-glycoprotein (an ABC transporter) (45). It was reasoned that if hyperproduction of transporters in response to Fst leads to inhibition of cell growth, it might be possible to protect cells from the effects of Fst by exposure to reserpine. Improved growth of both streaks and spotted dilutions of OG1X(pAM2005K) was seen on cAD1-containing plates in the presence of reserpine (Fig. 6A and andB).B). There was no difference in growth of M7(pAM2005K) on plates containing reserpine (Fig. 6C). β-Galactosidase production, resulting in blue pigment on the X-Gal-containing plates, confirmed that reserpine did not interfere with cAD1 induction (data not shown). These results indicate that sensitivity to Fst is due at least in part to the excessive activity of the transporters induced in response to cAD1.
The E. faecalis M7 strain is a spontaneously derived Fst-resistant mutant of OG1X (26). The mutant shows nearly complete resistance to Fst, cross-resistance to nisin, and a growth defect in the absence of pheromone. Genomic sequencing and allelic exchange experiments demonstrated that all three of these phenotypes are due to a point mutation in the funnel domain of the β′ subunit of RNAP. A second base change restoring functionality to the gene for the EtaR response regulator was also present in M7, but mutations disrupting etaR in this strain had little or no effect on Fst sensitivity. Interestingly, a strain with a functional etaR and a wild-type rpoC could not act as a recipient for pAM2005K, suggesting that an etaR mutation might be a prerequisite for acquisition of pAM2005K in this strain. This possibility is under investigation.
Microarray analysis revealed that the primary transcriptional response of OG1X cells to Fst exposure was induction of numerous genes encoding various transporters. In contrast, these transporters were not induced in M7, and induction failure was linked to the mutant rpoC allele by qRT-PCR analysis of wild-type allelic replacements. Two possibilities could account for this observation. First, undetermined effects of the rpoC mutation could protect the cell from Fst-mediated (and nisin-mediated) membrane stress, making induction of the transporters unnecessary. Second, upregulated expression of the transporters could be detrimental to the cell, and the rpoC mutation protects M7 by directly suppressing their induction. The observation that M7 cells were more sensitive to berberine supports the conclusion that the inability to induce the transporters is a direct effect of the rpoC mutation and not due to some form of upstream protection against membrane-active peptides. The conclusion that transporter upregulation is detrimental to cells was also supported by the observation that transporter inhibition by reserpine was able to partially rescue OG1X cells induced for Fst expression.
Among the transporters induced by exposure to Fst were several ABC transporters. Interestingly, expression of ABC transporters has been previously associated with resistance to a number of peptide antibiotics that affect lipid II and membrane function, including nisin (43). So, it was somewhat surprising to find them associated with sensitivity to Fst here. However, in most cases, exposure to these peptide toxins induces a limited number of ABC transporters under the control of specific two-component regulatory systems. The observed response to Fst seems more diverse and nonspecific, and it is possible that the extensive overproduction of these transporters is detrimental in some way, perhaps by depleting cellular ATP stores or disrupting membrane integrity. Alternatively, rather than being a cumulative effect of all of the induced transporters, Fst sensitivity could be due to the upregulated expression of one key transporter.
Given the timing of induction of transporters (60 min) compared to the appearance of cellular effects of Fst overexpression (15 min) (24), it would appear that the induction of the transporters is not the primary target of Fst but rather a response to its initial effects that may potentiate and/or prolong growth inhibition. Considering the types of transcripts induced in the microarray experiments, this primary effect could be related to membrane stress. Since resistance in M7 is not complete, it could be that the residual sensitivity of this strain to Fst is due solely to these primary effects. If so, the primary effects of Fst are rather subtle, and cell killing may require prolonged toxin exposure. It is worth noting that in this work, as in most other work with TA toxins, examination of toxin action depends on overproduction of the toxin and may not entirely reflect the effects of transient toxin expression from natural plasmid-carried loci.
The molecular mechanism by which the rpoC A753D mutation specifically prevents induction of these transcripts is unclear. It is possible that it interferes with the interaction of the RNAP with a specific set of promoters or with a transcriptional regulator, possibilities that will require further study. Regardless of its specific mechanism of action, it appears that the mutant RpoC in some way alters transcription to prevent the induction of multiple transporters induced in response to Fst. It is interesting in this light that rpoC mutations have previously been implicated in surfactant resistance in E. coli (46) and daptomycin resistance in S. aureus (47), in addition to affecting the adaptive evolution of E. coli to growth in minimal media (48). Thus, global effects on cellular transcription due to an RNAP mutation may be an underappreciated mechanism for cells to respond to toxic compounds in their environment, although it also remains a formal and presumably more remote possibility that Fst directly targets RNAP itself. In conclusion, it has been determined that Fst resistance of the M7 mutant strain is due to a mutation in the β′ subunit of RNAP itself and that this mutation affects a function that ultimately prevents the upregulated expression of multiple membrane transport systems, resulting in a protective effect on Fst-exposed cells. The exact molecular mechanisms of the mutation's effect on RNAP transcription of the transporters, Fst-mediated induction of transporters, and transporter-dependent inhibition of cell growth will require further study.
We acknowledge technical assistance of and helpful discussions with Chris Johnson, Dawn Manias, and Gary Dunny in performing the allelic exchange experiments and Chris Kristich for helpful suggestions and for providing the strains required. We also acknowledge the generous contribution of multiple microarray chips from Michael Gilmore and assistance from E. Peter Greenberg in foreseeing the usefulness and possibility of sequencing the OG1X and M7 genomes and helping to arrange the collaboration with R.B.
K.E.W. was supported in part by Public Health Service grant GM55544 and the Division of Basic Biomedical Sciences, Sanford School of Medicine. R.B. was partially supported by the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through grants UL1 RR025014 and R01DE012212. P.M.D. was supported, in part, by NIH/NIAID award AI073780.
Published ahead of print 26 October 2012
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.01597-12.