Mutations in ArgS Arginine-tRNA Synthetase Confer Additional Antibiotic Tolerance Protection to Extended-Spectrum-β-Lactamase-Producing Burkholderia thailandensis

Highly conserved PenI-type class A β-lactamase in pathogenic members of Burkholderia species can evolve to extended-spectrum β-lactamase (ESBL), which exhibits hydrolytic activity toward third-generation cephalosporins, while losing its activity toward the original penicillin substrates. We describe three single-amino-acid-substitution mutations in the ArgS arginine-tRNA synthetase that confer extra antibiotic tolerance protection to ESBL-producing Burkholderia thailandensis.

ABSTRACT

Highly conserved PenI-type class A β-lactamase in pathogenic members of Burkholderia species can evolve to extended-spectrum β-lactamase (ESBL), which exhibits hydrolytic activity toward third-generation cephalosporins, while losing its activity toward the original penicillin substrates. We describe three single-amino-acid-substitution mutations in the ArgS arginine-tRNA synthetase that confer extra antibiotic tolerance protection to ESBL-producing Burkholderia thailandensis. This pathway can be exploited to evade antibiotic tolerance induction in developing therapeutic measures against Burkholderia species, targeting their essential aminoacyl-tRNA synthetases.

INTRODUCTION

PenI-type class A β-lactamase acts as a major defense armory against penicillin derivatives in Burkholderia pathogens, including the highly pathogenic Burkholderia pseudomallei and Burkholderia mallei strains (1–3). This enzyme can evolve via a simple nucleotide substitution, deletion, or duplication mutation to an extended-spectrum β-lactamase (ESBL), which can hydrolyze third-generation cephalosporins, including ceftazidime (4–6). However, these ESBLs tend to lose hydrolyzing activity for their original substrates (penicillin derivatives) (4–6). PenL (previously called PenA) from the nonhuman pathogenic species Burkholderia thailandensis has been extensively studied as a model for the PenI-type class A β-lactamase, particularly with regard to its transition into an ESBL (4–6).

We used a selection experiment to investigate cellular survival responses in ESBL-producing Burkholderia species facing a lethal dose of penicillins. Specifically, a single colony of B. thailandensis strain E260 harboring an ESBL PenL (Glu168Del) (4) was grown overnight in 2 ml of LB broth at 37°C and shaken at 250 rpm. The overnight culture was pelleted by centrifugation at 2,600 × g for 5 min at 4°C, and the cells were resuspended in fresh LB broth to ∼10⁹ CFU/ml. Cell suspension (100 μl) was spread on LB agar plates containing 16 μg/ml (2× MIC) of ampicillin (AMP) and was incubated at 37°C. The frequency of survivors was estimated to be 10⁻⁸ to 10⁻⁶. Genomic DNA was purified from an isolate using a Wizard genomic DNA purification kit (Promega, Madison, WI), and whole-genome sequencing was performed using HiSeq 2000 (Illumina, San Diego, CA). The sequence data were analyzed using CLC Genomics Workbench software (CLC Bio, Redwood City, CA). We found that a single-nucleotide-substitution mutation occurred in the argS gene (BTH_I0355) encoding arginine-tRNA synthetase (ArgS). We further screened survivor isolates targeting the argS gene via PCR (primers argS-F [5′-GTAGACCGCGAAATCCTTCA-3′] and argS-R [5′-GGTTTTGCCATTGCTTGAGT-3′]) before sequencing using a 3730XL DNA analyzer (Applied Biosystems, Foster City, CA) in both directions using primers argS-F, argS-R, argS-MF (5′-AAAACAGCGTGAAGCAGGTC-3′), and argS-MR (5′-GCGCTTCGAGATCTTCACTT-3′). We ultimately obtained six argS mutants. These mutants had one of three different single-nucleotide-substitution mutations that led to Ala93Thr, Asp314Asn, or Pro332Gln in the enzyme (Fig. 1A). The MIC values for AMP were measured with the mutants and the wild-type (WT) strain using the agar dilution method as previously described (6). The argS mutants exhibited 4-fold-higher AMP MICs than the parental strain (Fig. 1A).

FIG 1.

Mutations against AMP occurring in ESBL-producing B. thailandensis. (A) Survivors of AMP selection and their mutations. Point mutations in the argS gene found in survivors are shown with their associated MICs. ND, not determined. (B) Structural model of B. thailandensis ArgS complexed with tRNA^Arg. Three mutation sites are denoted in red: pink circle, arginine-binding cavity; yellow circles, tRNA^Arg-binding sites.

To investigate the ArgS function affected by the mutations, we first computationally predicted the B. thailandensis ArgS structure by using the homologous structure of Neisseria gonorrhoeae (PDB accession no. 6AO8) using Modeller (7). We then simulated the ArgS complexed with tRNA^Arg using the structure from Pyrococcus horikoshii (PDB accession no. 2ZUE) (Fig. 1B). The results showed that all three mutations were located at the predicted arginine- or tRNA^Arg-binding sites (Fig. 1B). Pro332Gln was mapped to the arginine-binding cavity, and Ala93Thr and Asp314Asn were mapped to the tRNA-binding sites separated into two regions (Fig. 1B). The specific localization of the three amino acid changes suggested that they might affect arginine (Pro332Gln) or tRNA^Arg (Ala93Thr and Asp314Asn) binding, increasing the uncharged tRNA^Arg pool in the cell. In Escherichia coli isolates, uncharged tRNA interacts with RelA and binds to the ribosome, which leads to the activation of RelA to synthesize (p)ppGpp, the alarmone that induces the stringent response (SR) (8). In bacteria, (p)ppGpp alarmones accumulate in the cell in response to diverse disfavored conditions. SR promotes rapid reallocation of cellular resources and increases functions that facilitate bacterial survival (9, 10).

Because these argS mutations may affect tRNA metabolism and growth, we compared the mutants to the WT strain using growth curves after culture of the isolates in LB broth shaken at 250 rpm at 37°C. The argS mutants exhibited significantly reduced growth rates (the slope values from the linear region of the plots of the natural log of optical density versus time) compared with the WT strain (Fig. 2A). The growth rate of the WT strain was 0.88/h, but those of the argS mutants were 66% to 79% of this value (Asp314Asn, 0.58/h; Pro332Gln, 0.64/h; Ala93Thr, 0.69/h). Growth retardation is a main symptom of SR leading to antibiotic tolerance in bacteria (11, 12). Consistent with SR induction, the mutants exhibited increased MICs for various antibiotics with different modes of action, including ceftazidime (a third-generation cephalosporin), ciprofloxacin (a quinolone), and kanamycin (an aminoglycoside), in addition to AMP (Fig. 2B). Note that the mutations augmented ceftazidime insusceptibility, against which ESBL is already highly active. To confirm the involvement of SR, we disrupted relA, the key gene for SR induction, in these argS mutants as previously described (13). The relA-disrupted argS mutants exhibited significantly decreased antibiotic MICs (Fig. 2B). Killing curves were generated in the presence of a lethal dose of AMP (400 μg/ml) or ceftazidime (512 μg/ml); the minimal duration for killing 99% (MDK₉₉ values) of each strain (a measure of antibiotic tolerance [14, 15]) was determined as previously described (16). The MDK₉₉ values for argS mutants were markedly higher than those for the WT strain, the WT strain with disrupted relA, or the same mutants with disrupted relA, demonstrating the integral role of SR in antibiotic tolerance in argS mutants (Fig. 2C). Note that the WT strain and the WT strain with disrupted relA exhibited highly similar killing curves with the same MDK₉₉ values (Fig. 2C). This result indicates that SR was not induced by any of the antibiotics in the WT strain without an argS mutation. Mutations in argS conferring reduced susceptibility to a β-lactam antibiotic, amdinocillin, in clinical isolates of E. coli have been reported (17). Although these mutations were not analyzed for the underlying mechanism, it is likely that similar SR-associated tolerance shaped the phenotype. The occurrence of argS mutations in different genera challenged by β-lactam antibiotics suggests the significance of the argS-mutation-mediated tolerance in the survival of Gram-negative pathogens.

FIG 2.

Stringent response phenotypes. (A) Growth curves of three argS mutants compared with the WT strain. Experiments were performed in triplicate. (B) MIC table of argS mutants for various antibiotics. MEM, meropenem; CAZ, ceftazidime; CTX, cefotaxime; KAN, kanamycin; CIP, ciprofloxacin; TET, tetracycline; ND, not determined. (C) Killing curves of argS and ΔrelA mutants in the presence of ampicillin or ceftazidime. Pink horizontal line, point of 99% killing in each mutant population. Vertical dashed arrows, x axis values corresponding to the minimal duration for killing 99% (MDK₉₉ in h) of each mutant population. Experiments were performed in triplicate.

Besides the original substrates, ESBLs cannot protect Burkholderia spp. against meropenem, which constitutes a common antibiotic regimen for Burkholderia infections (1, 3, 4). We previously reported that mutations arose in metG or trmD genes encoding methionyl-tRNA synthetase or tRNA (guanine-N1)-methyltransferase, respectively, in ESBL-producing B. thailandensis in response to a lethal dose of meropenem; the mutations induced SR-associated antibiotic tolerance (13). Note that different antibiotic challenges, meropenem (13) and AMP in this study, led to the rise of mutations that affected tRNA metabolism, inducing tolerance to both antibiotics, but the affected genes differed. The essential tRNA synthetases in pathogens are attractive therapeutic targets from a practical perspective (18, 19). The SR induction pathway elucidated here will be an asset for developing means to target ArgS and other aminoacyl-tRNA synthetases without unnecessarily causing antibiotic tolerance.