A Novel Point Mutation Promotes Growth Phase-Dependent Daptomycin Tolerance in Staphylococcus aureus

Abstract

Recalcitrance of genetically susceptible bacteria to antibiotic killing is a hallmark of bacterial drug tolerance. This phenomenon is prevalent in biofilms, persisters, and also planktonic cells and is associated with chronic or relapsing infections with pathogens such as Staphylococcus aureus. Here we report the in vitro evolution of an S. aureus strain that exhibits a high degree of nonsusceptibility to daptomycin as a result of cyclic challenges with bactericidal concentrations of the drug. This phenotype was attributed to stationary growth phase-dependent drug tolerance and was clearly distinguished from resistance. The underlying genetic basis was revealed to be an adaptive point mutation in the putative inorganic phosphate (P_i) transporter gene pitA. Drug tolerance caused by this allele, termed pitA6, was abrogated when the upstream gene pitR was inactivated. Enhanced tolerance toward daptomycin, as well as the acyldepsipeptide antibiotic ADEP4 and various combinations of other drugs, was accompanied by elevated intracellular concentrations of P_i and polyphosphate, which may reversibly interfere with critical cellular functions. The evolved strain displayed increased rates of survival within human endothelial cells, demonstrating the correlation of intracellular persistence and drug tolerance. These findings will be useful for further investigations of S. aureus drug tolerance, toward the development of additional antipersister compounds and strategies.

INTRODUCTION

Staphylococcus aureus is a Gram-positive coccoid bacterium that colonizes the skin and mucosa of approximately one-third of the world's human population (1). Besides being an asymptomatic commensal organism, S. aureus is an etiological agent of various pyogenic or toxin-mediated illnesses, particularly among immunocompromised patients, who experience superficial infections to life-threatening inflammation and sepsis (2). Numerous vaccination approaches against S. aureus have failed to date (3), which underscores the need for a suitable armamentarium of drugs to combat staphylococcal infections. An efficient agent in the treatment of staphylococcal infections is the lipopeptide antibiotic daptomycin (DAP), which interferes with bacterial membrane polarization and the integrity of Gram-positive bacteria (4) and is generally suited to combat methicillin-resistant S. aureus (MRSA) (5). S. aureus strains that are resistant to β-lactams and other drugs frequently used in antibiotic therapy have been isolated (6), and underlying genetic factors are continuously being elucidated (7).

Besides resistance, bacterial drug tolerance represents another reason for the failure of antibiotic therapy. In contrast to resistant mutants, tolerant cells are unable to grow under exposure to an antibiotic and do not show changes in MIC values. However, such cells are capable of enduring high concentrations of antibiotics for at least a limited period (8). Antibiotic tolerance can be divided into genetic tolerance and phenotypic tolerance, which is not transmitted vertically (9). Phenotypic tolerance is represented by the persister state, in which only a small subpopulation is reversibly recalcitrant to killing by antibiotics (10, 11), or drug indifference, in which antibiotic nonsusceptibility extends over an entire culture (12); both are independent of variances in the genomic sequence. In contrast, genetic tolerance requires a mutation that renders a whole population more tolerant against antibiotic treatment (9).

A number of studies have investigated staphylococcal phenotypic tolerance (13–21). However, except for small-colony-variant (SCV) strains that exhibit intrinsically poor susceptibility to selected drugs (22), mutations that affect antibiotic tolerance in strains with ordinary colony morphotypes have not been described previously for this genus (23). In this study, we exploited an experimental approach with cycles of high-level antibiotic treatment of a wild-type S. aureus strain to select for mutations that increase drug tolerance. DAP was chosen as a selective agent because it efficiently kills S. aureus cells in the stationary growth phase (15). We obtained a mutant strain with drastically elevated DAP tolerance, beginning at the onset of the stationary growth phase. This phenotype correlates with increased cytoplasmic phosphate levels and requires a currently uncharacterized gene that we termed pitR. Furthermore, the phenotype was causally linked to a single nonsynonymous mutation within the downstream gene pitA, which encodes a putative phosphate transporter.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

The bacterial strains used in this study are listed in Table S2 in the supplemental material. Molecular cloning was performed in Escherichia coli DH5α using the plasmid pMAD (24), which was also exploited for markerless allelic exchange and gene inactivation in S. aureus. E. coli cells bearing pMAD or its derivatives were grown at 37°C in basic medium (BM) (1% [wt/vol] soy peptone, 0.5% [wt/vol] yeast extract, 0.5% [wt/vol] NaCl, 0.1% [wt/vol] K₂HPO₄, 0.1% [wt/vol] glucose) supplemented with 100 μg/ml ampicillin (Carl Roth), and S. aureus cells with episomal pMAD plasmids were grown at 30°C in BM supplemented with 2.5 μg/ml erythromycin (Carl Roth). To facilitate identification of genetically modified S. aureus strains that had undergone homologous recombination with pMAD derivatives, cells were plated on BM agar plates supplemented with X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) (Peqlab) and 2.5 μg/ml erythromycin, if necessary. For all other experiments except genetic manipulation, S. aureus was cultured in baffle flasks containing tryptic soy broth (TSB) (1.7% [wt/vol] peptone from casein, pancreatic digest, 3% [wt/vol] soy peptone, 5% [wt/vol] NaCl, 2.5% [wt/vol] K₂HPO₄, 2.5% [wt/vol] glucose [pH 7.3 ± 0.2]), at a 1:6.25 culture/flask ratio, at 37°C with shaking (150 rpm) or plated on tryptic soy agar (TSA) (TSB with 1.5% [wt/vol] agar). The following antibiotics were purchased and used: daptomycin (Cubicin) (powder, Novartis; Etest stripes, bioMérieux), ampicillin (Carl Roth), erythromycin, rifampin, vancomycin, gentamicin (Sigma-Aldrich), and ciprofloxacin (Fluka). ADEP4 was custom synthesized by EMC Microcollections. Notably, we observed different in vitro killing efficiencies with two different batches of DAP. In the first set of experiments, we determined 4 μg/ml as the MIC with the contents of one flask, in contrast to 1 μg/ml quantified with another vessel. The MIC and minimal bactericidal concentration (MBC) values of antibiotics tested against strains used in this study are listed in Table S1 in the supplemental material.

Molecular cloning and isolation of DNA.

Genomic DNA from S. aureus was isolated as described previously (25). Plasmids were isolated from bacterial cells using Qiagen kits. Plasmid constructs were verified by colony PCR and sequencing (GATC Biotech). E. coli cells were made chemically competent with RbCl (26). Electrocompetent S. aureus was prepared by collecting cells, by centrifugation for 10 min at 4,000 × g, from a 50-ml culture grown in TSB to an optical density at 578 nm (OD₅₇₈) of ∼0.5. Pelleted cells were washed successively in 100 ml, 50 ml, and 25 ml of ice-cold 10% (vol/vol) glycerin. Cells were finally suspended in 400 μl of ice-cold 10% (vol/vol) glycerin, and 75-μl aliquots were frozen at −80°C or immediately used for transformation. Plasmids that had been propagated in E. coli DH5α were shuttled through S. aureus RN4220 (27) before they were used for transformation into other S. aureus strains for homologous recombination.

In vitro evolution protocol with cyclic exposure to DAP.

S. aureus HG003 was incubated in TSB for 16 h at 37°C, with shaking, to obtain an overnight culture. Subsequently, DAP (with a final concentration of 50 μg/ml Ca²⁺ cations from CaCl₂) was added to a final concentration of 100 times the MIC. Cultures were incubated for 7.5 h at 37°C with shaking, and then the antibiotic was removed by washing pelleted cells twice in 1% saline solution. Collected cells were resuspended in fresh TSB and incubated for 16 h at 37°C in TSB without antibiotics to start a new cycle. Bacteria from culture D6, which had been obtained after six cycles of the in vitro evolution protocol, were repeatedly streaked four times to obtain single colonies, to ensure a clonal strain for subsequent experiments. One randomly chosen colony gave rise to strain D6.

Antibiotic susceptibility and survival assays.

The MIC was determined as described previously (15), and the MBC was determined according to the guidelines of the Clinical and Laboratory Standards Institute (28). To determine the numbers of drug-tolerant cells in cultures, strains were grown in TSB for 16 h at 37°C with shaking. Killing curves were recorded with cells in the stationary growth phase and 100 times the MIC of DAP/Ca²⁺, unless stated otherwise. Aliquots of 2.24 ml were transferred to 14-ml Greiner tubes, and antibiotics were added to the final concentrations indicated in the text and figure legends. Samples were incubated at 37°C for 8 h or up to 100 h, with shaking. To track levels of DAP-tolerant cells in a growth phase-dependent manner, overnight cultures of tested stains were inoculated at an OD₅₇₈ of 0.07 in 40 ml of fresh TSB in 250-ml baffle flasks and incubated for 10 h at 37°C with shaking. On an hourly basis, 2.24 ml of the cultures was transferred to 14-ml Greiner tubes and incubated with 100 times the MIC of DAP for 4.5 h at 37°C with shaking. For determinations of CFU per milliliter, 100-μl aliquots were taken before and during antibiotic challenge, at the indicated time points, washed and diluted with 1% saline solution, and subsequently plated on nonselective TSA. Colonies were counted after 24 h or up to 48 h of incubation. The lower limit of quantification was 100 CFU/ml. All time-kill experiments were conducted using at least three biological replicates.

Lag-phase determination.

Cells grown in TSB overnight at 37°C with agitation were treated with 100 times the MIC of DAP for 8 h. Cells were then washed in 1% saline solution and diluted to 1 CFU/10 μl; 10 μl of this suspension was mixed with 190 μl TSB for each well of a 96-well plate and incubated at 37°C for up to 48 h. The OD₅₇₈ was measured every 20 min. The lag phase was defined as the period without an increase in OD₅₇₈. For comparison, the lag phase of overnight cultures without DAP treatment was also determined. CFU concentrations were assayed by plating 10 μl on TSA. Only data from experiments in which were ≤1 CFU/10 μl grew on TSA after 48 h at 37°C were used. Experiments were repeated at least three times, on different days. Sample sizes are indicated in the figures.

Allelic exchange constructs and homologous recombination.

The putative pitRA6 operon (gene no. SAOUHSC_00669 and mutant gene no. SAOUHSC_00670) was amplified from D6 by PCR using primers SAO_00669-f1 and SAO_00671-r (for oligonucleotides, see Table S3 in the supplemental material) and was cloned into the plasmid pMAD via BglII, to yield pMAD-pitA6. Gibson Assembly (29) was used for cloning of the constructs for chromosomal deletion of pitA or pitR and single-copy complementation of the resulting strain HG003ΔpitA. The primers SAO_670down_f and SAO_670down_r, as well as SAO_670up_f and SAO_670up_r, were used for amplification of the genomic fragments bracketing pitA. The upstream fragment for deletion of pitR was amplified with primers SAO_670up_f and 669_GA#2_r, and the downstream fragment was generated using 670_GA#2_f and SAO_670down_r. Genomic DNA from wild-type HG003 was used for cloning of the plasmid pMAD-ΔpitRpitA, and DNA from HG003pitA6 served as the template for pMAD-ΔpitRpitA6. Complementation of HG003ΔpitA with the wild-type pitA allele was achieved with the plasmid pMAD-pitA, for which primers 670_GA_f and 670_GA_r were used for PCR, with genomic DNA from HG003 as the template. The target vector pMAD was digested with the enzymes BglII and SalI, and PCR products were incubated in 2.5-fold excess over the plasmid backbone for 60 min at 50°C, in a final volume of reaction mixture of 20 μl. The reaction mixture contained 0.08 U T5 exonuclease, 0.5 U Phusion DNA polymerase, 80 U Taq DNA ligase (all from New England Biolabs), and 4 μl of 5× isothermal reaction buffer (25% [wt/vol] polyethylene glycol [PEG] 8000, 500 mM Tris-HCl [pH 7.5], 50 mM MgCl₂, 50 mM dithiothreitol [DTT], 1 mM levels of each of the four deoxynucleoside triphosphates [dNTPs], and 5 mM NAD). S. aureus strains containing derivatives of pMAD episomally for homologous recombination were grown overnight at 30°C in TSB supplemented with erythromycin. All incubation steps in batch cultures were performed with shaking at 150 rpm. The overnight cultures were diluted 1:1,000 in fresh TSB without antibiotics and incubated at 30°C for 2 h before transfer to 42°C for 6 h. Bacterial cells were serially diluted to 1:10,000, plated on TSA supplemented with erythromycin and X-Gal, and incubated at 42°C until blue colonies grew. Several light blue colonies were picked for inoculation of fresh TSB without antibiotics. These cultures were incubated for 8 h at 30°C before they were again diluted 1:1,000 in fresh TSB and grown overnight at 42°C. The cultures were again diluted 1:1,000 in fresh TSB, and cells were grown for 4 h at 30°C and 4 h at 42°C. All such cultures were serially diluted and plated to obtain single colonies on TSA supplemented with X-Gal. Colonies that appeared white on TSA with X-Gal were tested for sensitivity to erythromycin and successful genetic replacement, by PCR and sequencing.

Inorganic phosphate determination.

A commercially available kit system (no. ab65622; Abcam) was used to quantify intracellular inorganic phosphate (P_i) levels. S. aureus strains were inoculated in 16 ml fresh TSB in 100-ml baffle flasks and incubated at 37°C with shaking. After 4, 6, and 16 h, the cultures were chilled on ice for 15 min before cells were harvested by centrifugation at 4,000 × g for 10 min at 4°C. The pellet was washed twice in double-distilled water and adjusted to an OD₅₇₈ of 20 in double-distilled water; 1 ml of that suspension was used for cell disruption (FastPrep-24; MPBio) at 6.5 m/s for 30 s, in three rounds, with 0.5 ml of 0.1-mm glass zirconium/silica beads. Homogenized samples were centrifuged at 15,000 × g for 15 min at 4°C. The supernatant was diluted 1:200 in double-distilled water, and P_i levels were determined according to the manufacturer's instructions.

Polyphosphate determination.

Intracellular polyphosphate (polyP) levels were determined using DAPI (4′,6-diamidino-2-phenylindole) (Sigma), as described previously (30, 31). Bacterial cultures were grown in TSB for 16 h at 37°C at 150 rpm. Cells were washed twice and resuspended in Tris-HCl buffer (100 mM Tris [pH 7.5]). Cell suspensions were adjusted to an OD₅₇₈ of 0.5, and DAPI was added to a final concentration of 20 μM. After 15 min of agitation at 37°C, DAPI fluorescence spectra were determined in a 96-well plate using a microplate reader (Tecan Infinite 200M; excitation, 415 nm; emission, 450 to 650 nm). An emission wavelength of 550 nm was used for the determination of fluorescence intensity, which is given as relative fluorescence units (RFU) of the DAPI-polyP complex, as emission signals of free DAPI and DAPI bound to DNA are minimal at this wavelength (31). Confocal images of strains stained with DAPI (final concentration, 10 μM) were obtained with a Zeiss LSM-710 NLO microscope, using a 63×/1.40 oil-immersion objective and Zen software. DAPI-polyP complexes were excited with a 415-nm laser and emission at 550 nm was recorded, whereas DAPI bound to DNA was visualized with excitation at 358 nm and emission at 461 nm.

RNA isolation and RT-PCR.

RNA from S. aureus was isolated after 4, 6, and 16 h of growth in 16 ml TSB at 37°C with shaking. Cells were adjusted to an OD₅₇₈ of 70 and lysed in 1 ml TRIzol (Life Technologies). RNA was isolated and digested twice with DNase I, as described previously (32). cDNA was prepared using a first-strand cDNA synthesis kit (Thermo Scientific), as described in the manufacturer's instructions. cDNA was diluted 1:2 in double-distilled water, and 2 μl was used as the template for PCRs.

Sequencing and genomic analysis.

An Illumina HighSeq 2500 system was used to sequence the wild-type HG003 strain and the D6 strain at CeGaT (Tübingen). The S. aureus NCTC8325-2 genome (NCBI accession no. NC_007795.1) was used as a reference for bioinformatics analysis. Paired-end sequencing reads were mapped to the reference using the Burrows-Wheeler aligner (33) using the BWA-MEM algorithm in the software package. Postprocessing of read mappings was performed with SAMtools (34). Single-nucleotide polymorphisms (SNPs) were determined using the Genome Analysis Toolkit (35). We filtered variant calls for a minimal coverage of five reads, a minimal quality of 30, and a minimal allele frequency of 90%. Effects on protein-coding genes were predicted for all detected variants using SnpEff (36). Sanger sequencing of PCR products was performed at GATC Biotech, as recommended by the manufacturer. The pan-genome of S. aureus was investigated with respect to pitA using PanGee (unpublished software), an alignment-based method that uses the SuperGenome approach (37).

Invasion and intracellular survival assays.

The human endothelial cell line Ea.Hy926 was a kind gift from Sucharit Bhakdi (Mainz, Germany), who had obtained the cell line from C. J. S. Edgell (38). EA.hy926, which is also available from the ATCC (strain CRL-2922), was cultivated in Dulbecco's modified Eagle's medium (DMEM)/F-12 medium supplemented with 10% (vol/vol) heat-inactivated fetal calf serum, 100 μg/ml streptomycin, and 100 U/ml penicillin, in tissue culture flasks, in a humidified atmosphere of 5% CO₂ at 37°C. Depending on the time required to achieve confluence, cells were passaged every 2 to 3 d. One day prior to infection with S. aureus, 1 × 10⁵ cells were seeded in 12-well plates. The medium was replaced with antibiotic-free DMEM/F-12 medium 1 h before infection. S. aureus HG003 and D6 cultures were grown for 4, 6, and 16 h in TSB and were used to inoculate host cells at a multiplicity of infection of 20. For that purpose, 1 ml of bacterial culture was pelleted and washed twice with phosphate-buffered saline (PBS). The pellet was resuspended in 1 ml of cell culture medium, cells were counted, and the required amount of culture was transferred to the 12-well plate. The plates were centrifuged for 10 min at 1,500 × g to synchronize infection, and the plates were then incubated for 1 h at 37°C in 5% CO₂. Extracellular bacteria were removed by treatment with 20 μg/ml lysostaphin (Ambi) for 30 min, after which the host cells were rinsed with PBS, tissue culture medium was added, and the cells were cocultured for an additional 1 h (to assay invasion efficiency) or 24 h (to assay intracellular survival). Intracellular bacteria were recovered by lysis of the host cells with distilled sterile water, and dilutions of the bacteria were plated for the enumeration of CFU. Results were obtained from at least three biological duplicates, with two technical replicates each. Significance was assessed using Student's t test.

Wax moth (Galleria mellonella) virulence assay.

Final-stage instar larvae of G. mellonella were ordered from R. J. Mous Live Bait. Larvae were grouped to a number of 10 for each experiment and were infected with S. aureus HG003 or D6 for time-to-death measurements, as described previously (39). For infection of G. mellonella, S. aureus cells were grown in TSB overnight at 37°C with agitation, washed and resuspended in PBS (140 mM NaCl, 10 mM Na₂HPO₄, 2.7 mM KCl, 1.8 mM KH₂PO₄, pH 7.3), and adjusted to levels of approximately 1 × 10⁶, 2.5 × 10⁶, or 1 × 10⁷ CFU of HG003 or D6 suspended in 10 μl of PBS; 10 μl with the indicated amount of cells was injected into the last left proleg of each caterpillar, using a 500-μl high-performance liquid chromatography (HPLC) syringe and a semiautomated PB600 repeating dispenser. A control group was injected with PBS only. Dead and living larvae were counted every 24 h for 5 days. Each experiment was repeated three times, on different days.

RESULTS

Cyclic treatment of S. aureus in the stationary growth phase with bactericidal concentrations of DAP in vitro yields a highly drug-tolerant strain.

We asked whether consecutive cycles of DAP treatment could evolve an S. aureus strain that produced elevated levels of drug-tolerant cells within a culture. To that end, cultures of the laboratory strain HG003 were first incubated for 16 h under nonselective conditions (to reach the stationary growth phase), followed by 7.5 h of challenge with 100 times the MIC of DAP (see Materials and Methods for details). This experiment was conducted for up to 10 days. Inspection of the CFU of stationary-growth-phase cultures treated with DAP revealed dramatically elevated levels of survivors after 6 days, compared to the parental HG003 strain. The latter cultures showed a typical biphasic killing pattern, with a >100-fold decrease in viable counts within the first 2 h of antibiotic treatment (from ∼2.6 × 10⁹ CFU/ml to ∼1.3 × 10⁷ CFU/ml), followed by slower eradication to ∼2.3 × 10⁴ CFU/ml after 24 h. In contrast, cultures of strain D6 (designated thus due to isolation on day 6) decreased merely from ∼1.4 × 10⁹ CFU/ml to ∼1.2 × 10⁹ CFU/ml within 2 h and exhibited more than ∼3.8 × 10⁸ CFU/ml after 24 h of incubation in DAP-containing medium (Fig. 1A). In a long-term experiment involving DAP treatment, a HG003 culture was sterilized after 48 h, whereas the viable counts of D6 were reduced by only 2 to 3 log units after 4 days (Fig. 1B). We calculated the minimal duration required to kill 99% of the cells (MDK₉₉), a measure of drug susceptibility (8). The MDK₉₉ for strain HG003 treated with DAP was 2 h, in stark contrast to ≥96 h (98.1% killed in 96 h) determined for strain D6.

FIG 1.

Time-dependent killing of S. aureus by daptomycin. Stationary-growth-phase wild-type (wt) HG003 and D6 were challenged with 100 times the MIC of DAP, and CFU were recorded. (A) Percentages of CFU after 2, 8, and 24 h were compared to the initial inoculum. (B) Killing curves from a long-term experiment. Controls were cultures without antibiotic. Values are presented as the arithmetic means of three independent experiments, and error bars represent standard deviations.

S. aureus D6 exhibits growth phase-dependent drug tolerance and not resistance.

As indicated by Etest strips, with a predefined gradient of antibiotic, and quantified by broth dilution, the DAP MICs for D6 and HG003 were identical (1 μg/ml) (see Fig. S1 and Table S1 in the supplemental material). The same value was determined as the minimal bactericidal concentration (MBC) for both strains (see Table S1 in the supplemental material). Growth of D6 in nonselective liquid TSB ceased at a lower final cell density (see Fig. S2 in the supplemental material). Next, the DAP tolerance of D6 and HG003 cells was investigated in relation to the growth phase of the cultures. As shown in Fig. 2, the levels of survivors in both strains exceeded the detection threshold (1 × 10² CFU/ml) 6 h after inoculation but, whereas the concentration of DAP-tolerant cells in the HG003 culture increased rather moderately, from approximately 1.4 × 10³ CFU/ml to ∼6.7 × 10⁴ CFU/ml at 10 h, a drastic increase in the drug-tolerant cell concentration was observed with the D6 culture, reflected by values increasing from ∼5.4 × 10⁴ CFU/ml to ∼1.0 × 10⁹ CFU/ml during the same period. Of note, exponential-growth-phase cultures of D6 were as susceptible to DAP killing as HG003 cultures, which, together with unaffected MIC and MBC values, rules out a canonical drug resistance phenomenon. We next monitored resuscitation times of survivors in liquid medium at a single-cell level. When drug-tolerant cells were shifted to fresh antibiotic-free medium after drug challenge, the increases in viable cell numbers were delayed to similar degrees in the two strains, with lag times ranging from 10 to ≥20 h (means of 17.8 ± 3.7 h [HG003] and 15.6 ± 2.7 h [D6] and medians of 18.0 h [HG003] and 16.3 h [D6]). Notably, lag times were shorter (∼8 to 10 h) and much less widely distributed when overnight cultures were shifted to fresh medium without DAP pretreatment (means of 8.7 ± 0.6 h [HG003] and 9.2 ± 1.1 h [D6] and medians of 8.7 h [HG003] and 9.0 h [D6]) (see Fig. S3 in the supplemental material).

FIG 2.

Growth phase-dependent daptomycin tolerance of S. aureus. DAP tolerance of wild-type (wt) HG003 and D6 was monitored by means of CFU recording at the indicated time points. All values are averages of three independent experiments. Error bars indicate standard deviations. The limit of detection was 1 × 10² CFU/ml.

Drug tolerance of S. aureus D6 extends beyond DAP.

In order to investigate whether strain D6 exhibits multidrug tolerance, we subjected stationary-growth-phase cultures to treatment with a number of other antibiotics. As observed previously, only a few drugs are capable of rapidly decreasing the viable counts of S. aureus in the stationary growth phase (13, 15, 18), consistent with our observations regarding ampicillin, ciprofloxacin, gentamicin, and rifampin. Of note, this also pertained to vancomycin, tolerance to which is clinically relevant for multidrug-resistant S. aureus strains (see Fig. S4 in the supplemental material). One exception was the acyldepsipeptide ADEP4, which targets the Clp machinery (40) (Fig. 3A). In a long-term experiment with drug challenge, viable counts of D6 exceeded those of HG003 until day 2, which indicates that the tolerance phenotype of D6 extends beyond DAP. The increases in CFU values for both strains after 3 days of ADEP4 challenge are attributed to clp mutations that manifest themselves in the bacterial population (18, 40). Combined treatment with ADEP4 and rifampin, which has proven useful in eradicating stationary-growth-phase S. aureus cells (18), was also effective against HG003 and D6; again, more survivors were detected in D6 cultures from day 2 (Fig. 3B). Killing curves for cultures treated with ADEP4 and DAP showed a trajectory comparable to that for the first combination (Fig. 3C). Notably, when DAP-challenged cultures were supplemented with ampicillin or gentamicin 4 h later, increased killing was observed with HG003 but not with D6. Similar combinations of DAP plus rifampin, ciprofloxacin, or vancomycin did not affect the viability of either culture type (see Fig. S5 in the supplemental material).

FIG 3.

Long-term killing experiment with S. aureus and ADEP4. Stationary-growth-phase wild-type (wt) HG003 and D6 were challenged with ADEP4 (A), ADEP4 in combination with rifampin (RIF) (B), or ADEP4 in combination with DAP (C). Controls were cultures without antibiotic. All values are averages of three independent experiments. Error bars indicate standard deviations. The limit of detection was 1 × 10² CFU/ml.

Drug tolerance of S. aureus D6 is caused by a single point mutation in a putative inorganic phosphate transporter.

S. aureus strains D6 and HG003 were subjected to whole-genome sequencing to decipher the genetic basis for the observed drug tolerance. Rigid filtering yielded two nonsynonymous single-nucleotide polymorphisms (SNPs) in D6, compared to ancestral HG003. These mapped to open reading frames SAOUHSC_02622 (codon 92, GGC [Gly] to AGC [Ser], termed SNP1), putatively encoding a sodium/glutamate transporter, and SAOUHSC_00670 (codon 57, GCT [Ala] to GGT [Val], termed SNP2) (Fig. 4A), which exhibits signatures of an inorganic phosphate (P_i) transporter that is designated pitA in S. aureus strain JKD6159 (41). PitA is predicted to contain seven transmembrane helices, with the A57V mutation being located in the second helix. Subsequent Sanger sequencing of PCR products confirmed the presence of both mutations in strain D6.

FIG 4.

Responsibility of pitA6 for increased drug tolerance. (A) Genetic organization of the pitRA locus (top row), pitRA6 (middle row), and pitRΔpitA (bottom row). Putative transcriptional and translational signals are depicted only in the top row, as a bent arrow (promoter; sequence in bold), a lollipop sign (terminator), and sequences in italics (Shine-Dalgarno [SD] sequences). Numbers in brackets indicate distances in base pairs. (B) Stationary-growth-phase wild-type (wt) HG003, D6, pitA6, and ΔpitA strains and corresponding ΔpitA→pitA and ΔpitA→pitA6 complementation strains challenged with 100 times the MIC of DAP. All values are averages of three independent experiments. Error bars indicate standard deviations. The limit of detection was 1 × 10² CFU/ml.

To identify the mutation responsible for the DAP tolerance phenotype, the wild-type SAOUHSC_00670 and SAOUHSC_02622 genes of HG003 were replaced with the respective mutant alleles from D6. Whereas the eradication kinetics of HG003 SNP1 with DAP were comparable to those of the ancestral HG003 strain, the strain bearing the mutant pitA (designated pitA6) displayed a DAP-dependent killing curve virtually indistinguishable from that for D6. Intriguingly, a markerless pitA deletion mutant of HG003 was killed in a fashion comparable to that for the ancestral HG003 (Fig. 4B), indicating that pitA6 is not a loss-of-function allele. Single-copy complementation of HG003ΔpitA by pitA or pitA6 restored the previously observed phenotypes, ruling out critical second-site mutations (e.g., in a global regulator). An exchange of pitA for pitA6 in the frequently used S. aureus laboratory strains SA113 and Newman resulted in growth and DAP tolerance alterations comparable to those of HG003pitA6 or D6 (see Fig. S6 in the supplemental material).

Strains with pitA6 display elevated intracellular levels of inorganic phosphate and polyphosphate.

According to the assumed function of PitA, we quantified the intracellular P_i concentrations of the HG003, D6, pitA6, and ΔpitA strains and complementation strains after 4, 6, and 16 h of growth (Fig. 5A). P_i levels increased steadily in strains harboring pitA6 and were elevated at least 3-fold, in comparison with the HG003 and ΔpitA strains, at 6 h and 16 h. The same effects were observed when the ΔpitA strain was complemented with pitA or with pitA6 (Fig. 5B).

FIG 5.

Increases in intracellular P_i levels in pitA6 strains. Intracellular P_i concentrations were measured for wild-type (wt) HG003, D6, pitA6, and ΔpitA strains (A) and for complementation strains with restored wild-type (ΔpitA to pitA) or mutant (ΔpitA to pitA6) alleles (B). All values are averages of at least three independent experiments. Error bars indicate standard deviations, and significance versus the wild-type values was determined using Student's t test. n.s., not significant.

Bacteria can store P_i as polyphosphate (polyP), which can form chains made up of tens to hundreds of P_i moieties (42). Fluorescence microscopy of DAPI-stained cells revealed increased intracellular concentrations of polyP in all inspected cells of D6 versus HG003 after 16 h of growth (Fig. 6A). Microplate reader measurements confirmed 2- to 3-fold increased polyP levels in D6 and pitA6 cells, compared to HG003 cells (Fig. 6B).

FIG 6.

Increases in polyphosphate levels in pitA6 strains. (A) Microscopic images of stationary-growth-phase wild-type (wt) HG003 and D6. BF, bright field; DNA, stained with DAPI with emission at 460 nm; polyP, stained with DAPI with emission at 550 nm. (B) Quantification of fluorescence at 550 nm using a microplate reader.

In order to assess a possible role for the stringent response alarmone (p)ppGpp, which has been associated with drug tolerance (43) and the persister state (together with polyP) in other bacteria (44), we changed pitA to pitA6 in strain HG001(p)ppGpp⁰, which is defective in the production of this messenger (45). An involvement of (p)ppGpp in DAP tolerance was ruled out, as the pitA6 mutation evoked the same phenotype as in HG003 (see Fig. S7 in the supplemental material).

pitR is required for the high-tolerance phenotype of HG003 pitA6.

pitA is located downstream of a gene putatively encoding a regulator or accessory protein for phosphate transport, which we coined pitR. An obvious σ^A-dependent promoter is positioned ∼40 bp upstream of pitR. Two well-conserved ribosome binding sites precede pitR and pitA, and a putative Rho-independent transcriptional terminator was found downstream of pitA, strongly indicating a bicistronic operon (Fig. 4A). The correlation of the DAP tolerance of D6 with the onset of the stationary growth phase led us to examine the transcription profiles of pitA and pitR during culture growth (see Fig. S8 in the supplemental material). Reverse transcription (RT)-PCR indicated constitutive expression of both pitR and pitA/pitA6 at either time point checked, which suggests that the activity of pitA is controlled posttranscriptionally or at the protein level, if at all. We found that pitR plays a critical role in the drug tolerance phenomenon. A generated HG003pitA6ΔpitR strain exhibited DAP killing behavior, pitA expression profiles, and intracellular P_i levels (at 16 h) similar to those of ancestral HG003 (Fig. 7), which shows that the pitA6 mutation requires an intact pitR gene to be effective.

FIG 7.

Deletion of pitR abolishing the pitA6 effect. (A) Stationary-growth-phase cultures of HG003ΔpitR and HG003pitA6ΔpitR and corresponding HG003ΔpitR→pitR and HG003pitA6ΔpitR→pitR complementation strains challenged with 100 times the MIC of DAP. (B) Intracellular P_i concentrations. (C) Transcriptional profiles of the genes pitR, pitA/pitA6, and pykA (control), monitored by RT-PCR after 4, 6, or 16 h of growth. All values are averages of at least three independent experiments. Error bars indicate standard deviations, and significance was determined using Student's t test. n.s., not significant. *, P < 0.05.

Intracellular survival of S. aureus D6 in human epithelial cells is increased.

Besides colonizing (a)biotic surfaces, S. aureus is capable of dwelling in tissues and inside eukaryotic host cells (46, 47). We tested HG003 and D6 grown in liquid cultures for 4, 6, and 16 h for their capacities to invade and to survive within the human endothelium-like cell line Ea.Hy926 (38). Invasion of both HG003 and D6 decreased with the growth phase of the bacterial culture, and this effect was more pronounced with D6 (Fig. 8A). Interestingly, regarding the number of intracellular bacteria recovered 1 h postinfection, D6 values exceeded those of HG003 at 4 h and 6 h, whereas the tendency was inverted at 16 h. In contrast, the numbers of D6 cells recovered after 24 h were elevated for the mutant irrespective of the growth phase used for infection. Cells from D6 cultures harvested at 4 h and 16 h were recovered at significantly elevated levels, compared to HG003 (Fig. 8B), suggesting that D6 is better suited for intracellular survival. Since the proficiency of intracellular survival could alter virulence, we tested HG003 and D6 in a Galleria mellonella wax moth larval model, which is well established for infections with bacterial pathogens, including S. aureus (39, 43, 48). Increasing the CFU led to faster killing of caterpillars for both strains, but significant differences in virulence between the two strains were not observed (see Fig. S9 in the supplemental material).

FIG 8.

Invasion and intracellular survival of S. aureus wild-type (wt) HG003 and D6 in the eukaryotic cell line Ea.Hy926. HG003 and D6 cells grown for 4, 6, and 16 h were used to inoculate the host cells at a multiplicity of infection of 20. Intracellular bacteria were counted after 1 h to determine invasion capacity (A) and after 24 h to assay intracellular survival (B). All values are averages of at least three independent experiments. Error bars indicate standard deviations, and significance versus wild-type values was determined using Student's t test. n.s., not significant.

DISCUSSION

Antibiotic-tolerant bacterial pathogens pose a formidable challenge in treating infectious diseases. Although the pathogens are genetically susceptible, antibiotic-dependent eradication of pathogenic bacteria in vivo cannot be completely achieved, which translates into chronic or relapsing infections (23, 49). Even worse, bacterial pathogens can acquire mutations that result in increasing levels of drug-tolerant persisters during the course of infection (50). Comparable studies focusing on staphylococcal persister development in vivo are currently lacking, which prompted us to obtain an S. aureus strain producing elevated levels of drug-tolerant cells in the laboratory.

We pursued an in vitro evolution strategy that employed repeated cycles of drug exposure followed by cultivation under nonselective conditions. Similar approaches have yielded E. coli strains with enhanced drug tolerance that is not based on canonical resistance (51, 52). Our strategy closely resembles the one used by Fridman et al. (8), in which a wild-type strain acquired decreased drug susceptibility traits under antibiotic challenge through adaptive evolution. We chose the clinically important anti-MRSA drug DAP as a selective agent because it rapidly eradicates the bulk of genetically susceptible S. aureus cells in liquid stationary-growth-phase cultures, leaving very small proportions of survivors (Fig. 1 and reference 15).

To our knowledge, S. aureus strain D6 represents the first example of evolved drug tolerance in S. aureus accompanied by unchanged DAP MIC values. It was noted previously that DAP tolerance is increased in adherent staphylococci, but a suggested link to the physiological status had not been proven and underlying genetic mechanisms had not been elucidated (53). We deciphered the genetic basis of the drug-tolerant phenotype in our evolved strain as a single nonsynonymous point mutation in the putative inorganic phosphate (P_i) transporter PitA (Fig. 4), which, according to the literature, is largely uncharacterized and is not associated with DAP resistance in S. aureus (54–58). A pan-genomic analysis conducted with 49 S. aureus strains available at the NCBI and sequencing of pitA in the chromosomes of 16 clinical S. aureus strains isolated from DAP-treated patients did not identify the pitA6 mutation in our study (data not shown). However, pitA of S. aureus was found to be mutated at elevated frequencies when strain RN4220 was subjected to long-term single-, mixed-, or alternating-drug treatments with three different antibiotics (59); pitA was particularly affected when trimethoprim, which is unrelated to DAP in structure and function, was involved. Most of the 20 mutations in pitA resulted in frameshifts, in contrast to the functional pitA6 allele we obtained. The accumulation of pitA mutations during challenges with different antibiotics indicates the relevance of this gene to drug nonsusceptibility.

In our study, increased intracellular P_i and polyP levels were linked to the pitA6 allele and the phenotype of high drug tolerance in the stationary growth phase (Fig. 5 and 6). As indicated by pitA/pitA6 mRNA abundances during all stages of growth tested, the activity of PitA/PitA6 seems not to be controlled at the transcriptional level. It is conceivable that the affinity of PitA6 for the substrate P_i is increased, compared to PitA, but this notion, as well as the role of PitR in the observed phenotype, needs to be addressed further.

A connection between P_i and the tolerance of S. aureus to membrane-acting cationic antimicrobial peptides was suggested previously, when treatment with ranalexin was found to result in upregulation of the global central metabolic regulator PhoU (60). Studies in E. coli showed that PhoU negatively controls the high-affinity phosphate transporter system pst (61), which is also present in S. aureus. Disruption or deletion of phoU in E. coli resulted in increased P_i uptake (62) and accumulation of polyP (63), similar to findings we observed for S. aureus pitA6 (Fig. 5 and 6). Gao et al. (43) previously identified a single point mutation in relA that rendered a S. aureus clinical isolate less susceptible to the antibiotic linezolid. RelA synthesizes the stringent response alarmone (p)ppGpp, which is associated with persister formation among Gram-negative bacteria (44, 64, 65). In E. coli, the intracellular concentrations of (p)ppGpp were shown to affect the levels of polyP, which stimulates the Lon protease (66). Among its substrates are components of toxin-antitoxin systems (67), which are prominent drivers of antibiotic tolerance (68). S. aureus encodes a polyP kinase (SAOUHSC_00943 in strain NCTC8325/HG003), which may use elevated P_i concentrations to synthesize more polyP. However, no homologs of Lon have been identified, and direct experimental evidence for the involvement of staphylococcal TA systems in the persister state and drug tolerance is lacking (69–72). Furthermore, in S. aureus, a link between the stringent response and persister formation has not been established to date (23), which is in line with our results obtained with a (p)ppGpp-defective strain (see Fig. S7 in the supplemental material). In general, polyP plays pivotal roles in various stress-related networks in bacteria (73), and it is conceivable that interactions between polyP and RNA polymerase (74) and/or ribosomes (75) are causative for the observed phenotypes of S. aureus pitA6 strains. Accordingly, a fatty acid signaling molecule that was shown to render bacterial persisters more drug sensitive favors enzymatic degradation of polyP (76).

The monophasic eradication trajectories of the stationary-growth-phase pitA6 strains treated with DAP indicate a population that uniformly displays similar degrees of drug tolerance, as was the case with similarly evolved E. coli strains (8). In our previous study, S. aureus strain SA113 challenged with 100 times the MIC of DAP in the stationary growth phase required at least 10 h to resuscitate and to resume growth after a shift to nonselective medium (16). This “postantibiotic effect” (77) was also observed with strains HG003 and D6, which displayed prolonged lag phases, averaging around 16 to 18 h, after DAP challenge in our study (see Fig. S3 in the supplemental material). The similarity in the resuscitation delays for the two strains contrasts with the phenotype of “tolerance by lag” that was ascribed to E. coli strains evolved for high drug tolerance (8). Also, the earlier entrance of pitA6 strains into the stationary phase, compared to HG003, cannot alone account for the observed drug tolerance, since DAP is generally efficient against S. aureus in later growth stages (Fig. 1 and reference 15).

Despite enhanced tolerance toward single drugs, the killing of strain D6 was accelerated by antibiotic combinations, although the effects were not as pronounced as in HG003. This finding pertains to combinations of ADEP4 with DAP and ADEP4 with rifampin, which were reported to eliminate S. aureus persisters (18). This emphasizes that targeting more than one bacterial structure or process is a promising strategy to combat drug-tolerant cells (78).

S. aureus is capable of dwelling within host cells after phagocytosis (46, 47). Therefore, we tested HG003 and D6 for their abilities to invade and to survive within endothelial cells. The invasion efficiency decreased with the growth phase of the culture, according to the agr quorum-sensing-system-dependent downregulation of fibronectin-binding proteins (79–84). Interestingly, this effect was more pronounced in D6 than in HG003, leading to higher invasion rates with 4- and 6-h cultures, although the 16-h cultures showed the reverse trend. Nevertheless, D6 generally exhibited higher survival rates, suggesting that the drug-tolerant phenotype can influence virulence. The in vivo relevance of drug-tolerant S. aureus persisters has recently been underlined (18), and further countermeasures against these specialized cells and other representatives of drug tolerance are critical (85). The strains generated in this study will facilitate elucidation of drug tolerance and the persister state of S. aureus and determination of the role of phosphate metabolism on a molecular level.