Effect of aminoacyl-tRNA synthetase mutations on susceptibility to ciprofloxacin in Escherichia coli

(1)

Effect of aminoacyl-tRNA synthetase mutations on susceptibility to ciprofloxacin in Escherichia coli

Linne´a Garoff, Douglas L. Huseby, Lisa Praski Alzrigat and Diarmaid Hughes *

Uppsala University, Department of Medical Biochemistry and Microbiology, Biomedical Center, Uppsala, Sweden

*Corresponding author. Tel: !46 18 471 4507; Fax: !46 18 471 4673; E-mail: diarmaid.hughes@imbim.uu.se orcid.org/0000-0002-7456-9182

Received 8 May 2018; returned 6 June 2018; revised 3 August 2018; accepted 9 August 2018

Background: Chromosomal mutations that reduce ciprofloxacin susceptibility in Escherichia coli characteristical- ly map to drug target genes (gyrAB and parCE), and genes encoding regulators of the AcrAB-TolC efflux pump.

Mutations in RNA polymerase can also reduce susceptibility, by up-regulating the MdtK efflux pump.

Objectives: We asked whether mutations in additional chromosomal gene classes could reduce susceptibility to ciprofloxacin.

Methods: Experimental evolution, complemented by WGS analysis, was used to select and identify mutations that reduce susceptibility to ciprofloxacin. Transcriptome analysis, genetic reconstructions, susceptibility measurements and competition assays were used to identify significant genes and explore the mechanism of resistance.

Results: Mutations in three different aminoacyl-tRNA synthetase genes (leuS, aspS and thrS) were shown to reduce susceptibility to ciprofloxacin. For two of the genes (leuS and aspS) the mechanism was partially dependent on RelA activity. Two independently selected mutations in leuS (Asp162Asn and Ser496Pro) were studied in most detail, re- vealing that they induce transcriptome changes similar to a stringent response, including up-regulation of three efflux-associated loci (mdtK, acrZ and ydhIJK). Genetic analysis showed that reduced susceptibility depended on the activity of these loci. Broader antimicrobial susceptibility testing showed that the leuS mutations also reduce susceptibility to additional classes of antibiotics (chloramphenicol, rifampicin, mecillinam, ampicillin and trimethoprim).

Conclusions: The identification of mutations in multiple tRNA synthetase genes that reduce susceptibility to ciprofloxacin and other antibiotics reveals the existence of a large mutational target that could contribute to resistance development by up-regulation of an array of efflux pumps.

Introduction

The antibacterial drug ciprofloxacin targets DNA gyrase and topoisomerase IV, inhibiting DNA synthesis^1,2and causing lethal double-stranded breaks in the chromosome.³ Resistant isolates typically carry chromosomal mutations in some or all of the target genes, gyrA, gyrB, parC and parE,^4–6often accompanied by mutations in marR, soxR and/or acrR^3,7–11that increase expression of the AcrAB-TolC efflux pump, down-regulate the OmpF porin (an entry route for ciprofloxacin) and up-regulate the more restrictive OmpC porin.^6,12,13In addition, several plasmid-borne genes can contribute to resistance in Escherichia coli. These include genes encoding different Qnr family proteins that protect the drug targets,^14–16 a gene encoding an amino-acetyl-transferase enzyme (aac-6⁰-Ib-cr) that inactivates the drug,¹⁷and genes encoding efflux pumps such as QepA^18,19or OqxAB.¹⁶Single mutations or acquired resistance genes increase MIC 2-fold (e.g. marR) or

up to 24-fold (gyrA and qnr).^6,20The absence of individual genetic alterations that cause larger fold increases in MIC explains why the evolution of resistance to ciprofloxacin in E. coli requires the accumulation of multiple genetic alterations.

This multi-step evolution of resistance can result in very complex genotypes. In the absence of detailed knowledge of the evolutionary history and experimental tests, it is problematic to determine the complete set of genes and alleles that contribute to reduced susceptibility in an evolved resistant isolate. There are indications in the literature that alterations in additional, non- canonical genes might be involved in resistance evolution.^21–24 Indeed, it was recently shown that mutations in rpoB, encoding a subunit of RNA polymerase, reduce susceptibility to ciprofloxacin by increasing expression of the MdtK inner membrane efflux pump.²⁵Here we asked whether mutations in additional chromosomal genes in E. coli could contribute significantly to the evolution of resistance to ciprofloxacin. Our approach was to experimentally

V^C The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy.

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://

creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the

Downloaded from https://academic.oup.com/jac/article-abstract/73/12/3285/5098535 by Beurlingbiblioteket user on 10 January 2019

(2)

evolve E. coli to resistance, then use genome sequencing and genetic analysis to identify mutations in non-canonical genes that contributed to resistance. We found that mutations in aminoacyl- tRNA synthetase (aaRS) genes are selected by ciprofloxacin, and we report the mechanism by which mutations in leuS contribute to the resistance phenotype.

Materials and methods Media and growth conditions

Strains (TableS1, available asSupplementary dataat JAC Online) were rou- tinely grown in LB (10 g tryptone, 5 g yeast extract, 10 g NaCl per litre of water) or on LA (LB with the addition of 1.5% agar), with antibiotics or sucrose added as described in the text. Mueller–Hinton II medium (MHII) (Becton, Dickinson, France) was used for in vitro evolution, MIC measurements and antibiotic gradient plates.

In vitro evolution

From 10 independent cultures of CH1378 (E. coli MG1655 DmutS, ciprofloxacin MIC 0.016 mg/L) aliquots of 2%10⁸cfu were spread on MH agar with ciprofloxacin at 2, 4, 6, 8 and 10-fold MIC. After 48 h at 37C, a single colony from each culture was picked from the highest drug concentration where growth occurred (defined as colonies of diameter 1 mm) and streak-purified three times on MHII agar at the selective concentration.

Second-step mutants were selected in the same manner, by spreading onto plates with ciprofloxacin concentrations at 2, 4, 6, 8 and 10 times the mutant-selective concentration. Selections were continued for three or four cycles, until the selective concentration reached or exceeded 1 mg/L.

DNA sequencing

DNA for WGS prepared using a Genomic DNA Buffer Set (Qiagen, Sweden) was sent to Beijing Genomic Institute (BGI; Tai Po, Hong Kong) for library preparation and sequencing and analysed using CLC Genomic Workbench 6 (CLC Bio, Denmark). Local DNA sequencing was performed at the Macrogen Europe Laboratory (Amsterdam, Netherlands) and analysed using CLC Main Workbench 7.7.1.

RNA sequencing

Bacteria were grown in liquid culture to mid-log phase (OD6000.25–0.4), and 1.5 mL of each culture was mixed with 3 mL of RNAprotect Bacteria Reagent (Qiagen). RNA was isolated using an RNeasy Mini Kit (Qiagen) and DNase Turbo Free (Ambion, Life Technologies) was used to remove DNA.

Purified RNA (4500 lg) from each sample (two biological replicates for each strain) was sent to BGI (Tai Po, Hong Kong) for transcriptome analysis.

qPCR

RNA was isolated in three biological replicates per strain, at OD6000.25–0.3, using the RNeasy Mini Kit (Qiagen). DNA was removed with the DNase Turbo Free (Ambion, Life Technologies) kit, and the High Capacity Reverse Transcription kit (Applied Biosystems) was used to generate cDNA.

Quantifications of nucleotide concentrations were made using the Nanodrop NO-1000 spectrophotometer. The Eco Real-Time PCR System (Illumina) was used for qPCR, with 20 lL reactions containing 1 lL of diluted cDNA (dilutions 1:10, 1:100, 1:1000), 12.5 lL of PerfeCTa SYBR Green FastMix (Quanta Biosciences), 1.25 lL of 10 lM forward and reverse primers (TableS2) and 4 lL of ddH2O. Standard housekeeping genes were idnT, cysG and hcaT.²⁶

Strain construction

Lambda red recombineering²⁷was used to place selectable markers onto the chromosomes of bacterial strain CH1940 or CH3302 (TableS1). P1 bacteriophage was used to transduce selectable cassettes between strains.

The kan-sacB cassette linked to leuS Asp162Asn and Ser496Pro was removed by P1 transduction with a lysate grown on a WT E. coli, selecting for sucrose-resistant recombinants on salt-free LA containing 5% sucrose.

Drug cassettes bounded by Flp recombination target sequences (FRT) were removed by transformation of pCP20, expressing Flp enzyme,²⁸with selection on ampicillin (100 mg/L), leaving an FRT scar at the site of insertion.

Local sequencing confirmed all genetic constructs. Oligonucleotides for PCR, sequencing and recombineering are listed in TableS2.

Growth rate measurements

Four biological replicates of liquid overnight cultures were diluted 1:1000 in LB media. Aliquots (300 lL/well) were pipetted into a Honeycomb 2 100- well plate (Oy Growth Curves AB Ltd, Finland), and growth was measured using a Bioscreen C (Oy Growth Curves AB Ltd, Finland) with readings at 600 nm every 5 min for 18 h at 37C with continuous shaking. Negative con- trols were included to correct OD600values by subtraction. The natural loga- rithm of each OD600 value was plotted against time, and the linear regression slope for 10 subsequent OD600 readings was calculated.

Doubling time (minutes) was calculated by dividing ln2by the value of the maximal slope.

MIC

MIC was determined using broth microdilution in MHII medium. Bacterial suspension (McFarland 0.5) in 0.9% NaCl, determined using a Sensititre^TM Nephelometer (Oxoid AB), was diluted 1:100 in MHII broth. Aliquots of 50 lL were mixed with 50 lL of MHII broth containing serially diluted antibiotic in 96-well round-bottomed plates. Mixtures were incubated for 18 h at 37C before visual reading. Each strain was measured in at least three biological replicates. For mupirocin, bacterial suspensions were prepared in MHII broth supplemented with 30 mg/L mupirocin.

Growth competition assays

For growth competition assays, strains carried a yellow fluorescent marker (SYFP, CH2037) or a dTomato fluorescent marker (dTomato, CH6016).

Overnight cultures were mixed 1:1, with dye-swaps included, diluted 1:142 in 180 lL of LB, with or without ciprofloxacin, in a round-bottomed 96-well plate and grown overnight at 37C at 900 rpm in a PHMP-4 Thermoshaker (Grant-Bio, Cambridge). For measurements, cultures were diluted 1:128 in 200 lL of PBS in a round-bottomed 96-well plate. Proportions of strains in each culture were determined using a magnetically activated cell sorter (MACSQuant^V^R VYB, Miltenyi Biotech, Germany) counting 10⁴cells. Ratios were measured at timepoints 0 and 24 h as the fraction dTomato-labelled cells divided by SYFP-labelled cells. Values plotted against time generated an exponential trend line for each competition. From the slope a selection coefficient for one of the competing strains was determined, representing the change relative to the competing strain expressed as percentage per generation.²⁹

Antibiotic gradient plates

Antibiotic gradient plates were cast in tilted 12%12 cm dishes (Greiner Bio- One, GmbH) with 50 mL of MHII agar supplemented with antibiotic at a set concentration. After solidifying (1.5 h at room temperature), 50 mL of MHII agar without antibiotic was poured on top (plates now lying flat) and left to solidify for at least 1.5 h. Colonies dispersed in 5 mL of 0.9% NaCl were diluted to 10⁴cfu/mL and 5 lL aliquots were spotted across the gradient.

Garoff et al.

(3)

Plates were incubated at 37C for 48 or 72 h. At least three independent measurements were made for each strain.

Results

Selection of novel resistant mutants

Resistance to ciprofloxacin was evolved in 10 independent lineages of E. coli K-12 mutS::Tn10. One endpoint strain from each lin- eage was subjected to WGS (TableS3). Genotypes were examined for mutations occurring independently in different lineages in genes not typically associated with resistance. Mutations in rpoB and rpoC were noted, in agreement with the recent finding that RNA polymerase mutations contribute to ciprofloxacin resistance.²⁵Mutations were also noted in genes encoding aaRS enzymes in 5/10 lineages (Table S3), with two independently selected in leuS (Asp162Asn and Ser496Pro), suggesting these might be associated with resistance to ciprofloxacin.

leuS mutations reduce susceptibility to ciprofloxacin

We asked whether the leuS mutations were involved in reducing susceptibility to ciprofloxacin. Repairing each leuS mutation in the complex evolved strains caused a 2-fold reduction in ciprofloxacin MIC (TableS4). We next introduced each leuS mutation into several simple genetic backgrounds (Table1) to test whether each was sufficient to reduce susceptibility. Ciprofloxacin MIC increased 2- to 4-fold in the presence of each leuS mutation in strains carrying gyrA Ser83Leu and/or DmarR DacrR resistance mutations (Table1) and significantly increased competitive fitness as a function of increasing ciprofloxacin concentration (Figure 1). No increase in MIC was observed in the fully susceptible WT. However, MIC is a relatively crude assay and might not detect subtle changes in susceptibility. To evaluate this we performed two tests. Growth competition (WT versus each isogenic leuS mutant) as a function of ciprofloxacin concentration showed an advantage for each leuS mutant (FigureS1A, B). Growth on antibiotic gradient plates also showed that each leuS mutation reduced susceptibility to ciprofloxacin in an otherwise WT strain (FigureS1C). We conclude that the leuS mutations reduce susceptibility to ciprofloxacin in multiple genetic backgrounds.

We also performed a complementation assay for the leuS mutant with the stronger phenotype, Ser496Pro, to determine whether expression of a WT leuS gene would reduce the resistance phenotype. We constructed strains with a WT leuS gene expressed under the control of the arabinose promoter (pBAD) on the chromosome, and also carrying either WT leuS or leuS Ser496Pro at its native location. Expressing the leuS^!gene with arabinose induction reduced ciprofloxacin resistance in the leuS Ser496Pro strain, measured both in MIC assays and as growth on antibiotic gradient plates (FigureS2). In contrast, overexpression of leuS^!in the WT background did not affect ciprofloxacin susceptibility.

These experiments provide strong additional evidence that the leuS mutation is involved in reducing ciprofloxacin susceptibility, and furthermore indicate that the phenotype of the mutation is due to a reduced functionality of the synthetase.

leuS mutations alter global transcription patterns

We hypothesized that aaRS mutations might influence ciprofloxacin susceptibility indirectly by triggering a stringent response, resulting in global alterations in gene expression.³⁰RNA sequencing showed that both leuS alleles had similar, large impacts on global mRNA levels (TablesS5andS6), with leuS Ser496Pro affecting

200 more genes than leuS Asp162Asn, and causing a larger fold change in transcript level (2.1-fold greater, P,0.01) (Figure2a).

The overall pattern of changes was strikingly similar to that associated with the onset of a stringent response, with 90% of genes pre- viously found to be up-regulated in a microarray-based analysis of E. coli MG1655³¹also being up-regulated by the presence of the leuS mutations. To test this similarity to a stringent response we deleted relA in strains carrying the leuS mutations and found that this reversed the MIC increase associated with the leuS mutations (TableS1), showing that the reduced susceptibility is RelA dependent. We also treated LM709 (leuS WT) with 30 mg/L mupirocin, an antibiotic that inhibits the isoleucine tRNA synthetase (ileRS) and thus induces ppGpp production in E. coli,³²and found this increased the ciprofloxacin MIC 2-fold. Together, these data show that inhibition of aaRS enzymes, either by chemical inhibition (mupirocin) or by mutation (leuS mutations), reduces susceptibility to ciprofloxacin by inducing a stringent-like transcriptional response.

Table 1. MICs of ciprofloxacin (mg/L) and relative growth rates for isogenic strains of E. coli with or without leuS mutations

Strain Genotype MIC Growth rate (SD)

LM179 Control strain MG1655 0.016 1.00 (0.05)

CH4985 leuS Asp162Asn 0.016 0.83 (0.03)

CH4986 leuS Ser496Pro 0.016 0.53 (0.11)

HS25 DmarR DacrR 0.047 0.96 (0.04)

CH4987 DmarR DacrR, leuS Asp162Asn 0.094 0.72 (0.01)

CH4988 DmarR DacrR, leuS Ser496Pro 0.094 0.58 (0.06)

CH4980 gyrA Ser83Leu 0.19 1.00 (0.01)

CH4989 gyrA Ser83Leu, leuS Asp162Asn 0.38 0.71 (0.03)

CH4990 gyrA Ser83Leu, leuS Ser496Pro 0.38 0.52 (0.11)

LM709 gyrA Ser83Leu, DmarR DacrR 1 0.90 (0.06)

CH7381 gyrA Ser83Leu, DmarR DacrR, leuS Asp162Asn 3 0.69 (0.08)

CH7382 gyrA Ser83Leu, DmarR DacrR, leuS Ser496Pro 4 0.58 (0.05)

(4)

Among the transcripts up-regulated.2-fold (TablesS5andS6) were several that could plausibly be associated with affecting susceptibility to ciprofloxacin: (i) mdtK, encoding an inner membrane efflux pump for which increased transcription reduces susceptibility to ciprofloxacin;²⁵ (ii) acrZ, a component of the AcrAB-TolC efflux pump possibly involved in determining substrate specificity;^33,34(iii) ydhIJK, a putative efflux pump operon;³⁵and (iv) ompF, encoding an outer membrane protein that favours ciprofloxacin influx.³⁶We hypothesized that increases in the levels of these and possibly other transcripts play a role in reducing susceptibility to ciprofloxacin. We tested the significance of the RNA sequencing results by performing qPCR to measure transcript levels of the three genes mdtK, acrZ and ydhIJK. These measurements were made on strains carrying the two different leuS mutants, on a leuS WT control strain (LM709), and on the corresponding isogenic DrelA mutants. LM709 was used as a relative reference point for the transcript levels of mdtK, acrZ and ydhIJK.

The results showed that a DrelA mutation in an leuS WT background has no effect on the expression levels of mdtK, acrZ or ydhIJK. In contrast, the levels of mdtK and ydhIJK were found to increase in strains carrying either of the leuS mutations, and this increase was dependent on having an active relA gene in the leuS mutant strains (FigureS3). Changes in level were not significant for acrZ. In summary, we were able to show for two of the three genes implicated in leuS-mediated resistance, mdtK and ydhIJK, that transcription levels were increased in leuS mutant strains, and that the increase was dependent on an active relA gene.

To test whether expression of mdtK, acrZ and ydhIJK influences ciprofloxacin susceptibility we constructed isogenic strains with these loci deleted, individually and in combination (TableS1) and measured MIC and growth competition as a function of ciprofloxacin concentration. Single- and double-deletion mutants had relatively small or insignificant effects on MIC (TableS1) but the triple deletion DmdtK DacrZ DydhIJK caused a large reduction specifical- ly associated with leuS mutations (Asp162Asn, 8-fold reduction;

Ser496Pro, 20-fold reduction) with no significant change when leuS was WT (TableS1). The surprisingly large reduction in MIC in the triple-pump deletion leuS strains is probably due to increased ciprofloxacin influx mediated by increased expression of ompF.

Indeed, when we deleted ompF from the DmdtK DacrZ DydhIJK strains the MIC increased in the leuS mutants, from 0.375 to 1.5 mg/L (leuS Asp162Asn), and from 0.19 to 3 mg/L (leuS Ser496Pro) (Table S1). The growth competition assays were in broad agreement with MIC data, with single- and double-pump loci deletions having minor or allele-specific effects on competitive ability (FigureS4), whereas the triple deletion abolished the competitive ability of each leuS mutant strain but had no significant impact on the leuS WT strain (Figure2b). These data show that reduced susceptibility of the leuS mutant strains is strongly dependent on the activities of the mdtK acrZ and ydhIJK efflux pump loci.

Mutations in leuS reduce susceptibility to other classes of antibiotics

We asked whether leuS mutations decreased susceptibility to antibiotic classes other than ciprofloxacin. Using antibiotic gradient plates and MIC measurements, we assayed the influence of leuS mutations on susceptibility to gentamicin, tetracycline, rifampicin, chloramphenicol, mecillinam, ampicillin and trimethoprim. The assays demonstrated reduced susceptibility to rifampicin, chloramphenicol, mecillinam, ampicillin and trimethoprim, showing that both leuS mutations reduce susceptibility to diverse classes of antibiotics (Figure3, FigureS5and TableS7). For several of these antibiotics the reduction in susceptibility associated with the leuS mutations was partly or completely reversed in DrelA strains (rifampicin, mecillinam, ampicillin and trimethoprim) or DmdtK DacrZ DydhIJK strains (mecillinam, ampicillin, trimethoprim and chloramphenicol). These data show that the mechanism by which leuS mutations reduce susceptibility to ciprofloxacin also acts to reduce susceptibility to antibiotics of different classes (FigureS6and TableS7panels B, C).

Mutations in additional tRNA synthetases reduce susceptibility to ciprofloxacin

We asked whether the phenotype of reduced susceptibility to ciprofloxacin was specific to these two leuS mutations. Mutations in additional aaRS genes, tyrS, aspS and pheS, were observed in the 30

(a) (b)

leuS Asp162Asn 30 leuS Ser496Pro

20 10 0 –10 –20 –30 –40

0 0.5 1 1.5

Ciprofloxacin (mg/L)

2 2.5 3

20 10 0 –10

Selection coefficient (%) Selection coefficient (%)

–20 –30 –40

0 0.5 1 1.5

2 2.5 3

Figure 1. Competitive advantage conferred by leuS mutations. Selection coefficients as a function of increasing concentrations of ciprofloxacin are shown for leuS mutants competing against LM709, an isogenic strain with a WT leuS gene. Positive values indicate an increased competitive fitness advantage. (a) leuS Asp162Asn. (b) leuS Ser496Pro.

Garoff et al.

(5)

original in vitro evolution experiment (TableS3). In separate evolution experiments carried out in mutS^!E. coli strains, we selected additional aaRS mutations (leuS Leu41His, aspS Asp207Ala, thrS His244Pro and thrS Ile582Ser). We constructed isogenic strains carrying leuS Leu41His, aspS Asp207Ala, thrS His244Pro, thrS Ile582Ser and pheS Thr110Ala. Resistance to bacteriophage P1 in some evolved strains prevented a similar reconstruction of the remaining two aaRS mutation in tyrS and aspS. Strains carrying seven different aaRS mutations (including leuS Asp162Asn and leuS Ser496Pro) were then tested for their ability to grow on antibiotic gradient agar plates. With the exception of pheS Thr110Ala, each of the other six aaRS mutations (leuS Asp162Asn, leuS Ser496Pro, leuS Leu41His, aspS Asp207Ala, thrS His244Pro and thrS

Ile582Ser) significantly reduced susceptibility to ciprofloxacin (Figure4and FigureS7), and also caused an increase in ciprofloxacin MIC (TableS1). Interestingly, whereas the reduced susceptibility caused by the leuS and aspS mutations is at least partly dependent on RelA activity, the thrS mutations did not show a de- pendence on relA (Figure4, FigureS7and TableS1), suggesting that they may operate through a different mechanism. We conclude that mutations in several different aaRS genes can reduce susceptibility to ciprofloxacin.

Discussion

Here we have shown that mutations in different aaRS genes (leuS, aspS and thrS) can reduce susceptibility to ciprofloxacin (Figures1 and4). Focusing on two different mutations in leuS (Asp162Asn and Ser496Pro), we showed that this phenotype was RelA dependent and involved changes in global gene expression, similar to those induced by the stringent response, including up-regulation of several efflux pump genes whose activity we showed to be required for reduced susceptibility (Figures2and4). Each of the leuS mutations also reduced susceptibility to additional classes of antibiotics: rifampicin, chloramphenicol, mecillinam, ampicillin and trimethoprim (Figure3). Taken together, these results show that aaRS genes could represent a large genetic target capable of gen- erating mutations that reduce susceptibility to fluoroquinolones y = 2.1304x – 3.2843

R² = 0.8289

0.001 0.01 0.1 1 10 100 1000 10 000

mRNA fold changes with leuS Asp162Asn

mRNA fold changes with leuS Ser496Pro

–100 –80 –60 –40 –20 0

Relative selection coefficient (%)

0 0.5 1 1.5 2 2.5 3

leuS WT

leuS Ser496Pro

leuS Asp162Asn (a)

(b)

Figure 2. Mutations in leuS influence mRNA transcript levels.

(a) Changes in global mRNA levels relative to a leuS^!strain (LM709), with the fold changes of leuS Ser496Pro (CH7383) plotted against fold changes in leuS Asp162Asn (CH7381). A linear regression line is depicted, plus the associated equation. (b) Deletion of mdtK, acrZ and ydhIJK (efflux components up-regulated by the leuS mutants) reduced competitive fitness in the presence of ciprofloxacin. Strains were competed against an isogenic leuS strain without deletions of mdtK, acrZ or ydhIJK.

Competitions involving single and double gene deletions are shown in FigureS2.

GEN 3

Antibiotic gradient

16 30

Bacterial growth across gradient

leuS Ser496Pro leuS Asp162Asn

leuS⁺

60 3 80 1

TET RIF CHL MEC AMP TMP

Figure 3. Mutations in leuS reduce susceptibility to several antibiotic classes. Bacterial suspensions of 10⁴cfu/mL were spotted in 5 lL aliquots for LM709 (leuS^!), CH7381 (leuS Asp162Asn) and CH7383 (leuS Ser496Pro), seven times across gradient plates (increasing gradient indicated on the y-axis) carrying different antibiotics (maximum antibiotic concentrations indicated in mg/L), and photographed after 48 h of in- cubation at 37C. The extent of positive growth (indicated by the height of the bars) was observed by visual examination with a cutoff value of .5 colonies. The heights of leuS^!bars are not normalized but are a function of the concentration of the individual antibiotics in the gradient.

Green background indicates decreased susceptibility with the leuS mutants. The photographs may be viewed in FigureS5, and MIC values can be read in TableS7. GEN, gentamicin; TET, tetracycline; RIF, rifampicin; CHL, chloramphenicol; MEC, mecillinam; AMP, ampicillin; TMP, trimethoprim.

(6)

and other antibiotics. The current data argue that, at least for the leuS mutations, the reduced susceptibility to ciprofloxacin is mediated through up-regulation of drug efflux genes (Figure 2 and FigureS4).

The aaRS mutations occur in both classes of aaRS enzymes:

LeuS belonging to class I, and AspS and ThrS belonging to class II.

In a genetic complementation assay we found that expression of the WT leuS gene suppressed the resistance phenotype associated with the leuS Ser496Pro mutation (Figure S2). This is consistent with the mutation causing antibiotic resistance by reducing aaRS function. With one exception, all of the selected aaRS mutations are located in the catalytic regions of their respective enzymes (leuS Leu41His, leuS Ser496Pro, aspS Asp207Ala, thrS His244Pro and thrS Ile582Ser).^37–40The only exception, leuS Asp162Asn, is situated in the zinc-binding region of the protein, important for positioning the tRNA acceptor arm.⁴⁰The locations of the mutations are consistent with reduced enzymatic functionality being the cause of the resistance phenotype. This is also consistent with the increased resistance to ciprofloxacin we observed in the presence of mupirocin, which chemically inhibits ileRS functionality (see the Results section). Together, the data suggest that aaRS genes represent a large target where the occurrence of mutations reducing functionality has the potential to affect drug susceptibility.

There are previous reports linking mutations in aaRS genes, or induction of a stringent response, with reduced susceptibility to

antimicrobial drugs. Mutations conferring resistance to mecillinam have been selected in alaS, argS, aspS and thrS,^41,42and for novobiocin in ileS, alaS, argS and leuS,⁴³where the mutations were also shown to confer cross-resistance to mecillinam.⁴³In the case of alaS and argS mutations, the mecillinam/novobiocin resistance phenotype was shown to be RelA dependent.^42,43Although the mechanism of mutational resistance to mecillinam is not known in detail, it was shown by deletion analysis that it makes PBP2, the target of mecillinam, dispensable for cell growth.⁴⁴Induction of the stringent response has also been shown to reduce susceptibility to penicillins⁴⁵and to trimethoprim in E. coli.⁴⁶Interestingly, increased expression of mdtK has been reported to increase MICs of trimethoprim and chloramphenicol in E. coli.⁴⁷Accordingly, although the earlier literature showed that mutations in aaRS genes, via induction of a stringent response, reduce susceptibility to mecillinam/penicillins and novobiocin, it did not provide any specific mechanism by which the resistance phenotype was mediated.

Here, we have shown that mutations in aaRS genes reduce susceptibility to the clinically important fluoroquinolones, that the ciprofloxacin resistance phenotype of mutations selected in leuS is RelA dependent, and that the phenotype depends on the activity of the efflux-related genes mdtK, acrZ and ydhIJK. The current work does not rule out that other gene products affected by the aaRS mutations might also play a role in the observed net reduction in susceptibility.

Whereas the reduction in ciprofloxacin susceptibility associated with leuS and aspS mutations is RelA dependent (Figure 4 and Table S1), mutations selected for the same phenotype in thrS are unaffected by deletion of relA (Figure 4and Table S1).

Interestingly, in a study of LpxC inhibitors (LpxC is an essential enzyme involved in lipid A synthesis), a resistance mutation selected in thrS (Ser517Ala) was found to increase the MIC of ciprofloxacin 2-fold.⁴⁸This thrS mutation reduced protein synthesis and growth rates, which the authors suggested might be the cause of reduced susceptibility to ciprofloxacin. The possibility that the rate of protein synthesis might influence susceptibility to fluoroquinolones is supported by evidence that pre-treatment of E. coli with chloramphenicol, a protein synthesis inhibitor, reduces killing by fluoroquinolones.⁴⁹This raises the possibility that the thrS mutations operate through a mechanism involving a direct effect on protein synthesis, but this remains to be elucidated in future studies.

Whereas the effect of the leuS mutations is significant in the fully susceptible WT genetic background (FigureS1), it is magnified in the gyrA Ser83Leu DmarR DacrR genetic background (Table1).

The mechanism of resistance we have identified (Figure2b) sug- gests a possible explanation for this observation. Increased expression of inner membrane pumps would remove some of the ciprofloxacin from the cytoplasm to the periplasm, from where it could either be pumped out of the cell (for example by AcrAB-TolC) or diffuse back into the cytoplasm. It is plausible that the combined effects of reduced affinity for the drug target and multiple efflux pumps acting on different cellular compartments provide a significant boost to the net removal of ciprofloxacin from the cell. The influence of overall genotype on phenotype was also seen in the large differences in relative fitness costs associated with leuS mutations in different genetic backgrounds (Asp162Asn, 17%–

29%; Ser496Pro, 18%–48%; Table1and TableS4), suggesting that there could also be a complex interplay between genotype and phenotype in resistant clinical isolates. One could argue that aaRS WT

Asp162AsnSer496ProLeu41HisHis244ProIle582SerAsp207AlaThr110Ala Bacterial growth across gradient

leuS

relA⁺ ΔrelA

CIP gradient (<2.5 mg/L)

thrS aspS pheS

Figure 4. Ciprofloxacin (CIP) susceptibility as a function of different tRNA synthetase mutations, with or without RelA activity. Gradient plates with maximum concentration of 2.5 mg/L ciprofloxacin were used, and 5 lL of bacterial suspension of 10⁴cfu/mL was spotted seven times across the gradient (increasing gradient indicated on the y-axis). Plates were incubated at 37C for 48 h, or 72 h in the case of thrS mutants owing to slow growth rates. The extent of positive growth (indicated by the height of the bars) was observed by visual examination with a cut-off value of .5 colonies. The dashed red line indicates the extent of positive growth for the aaRS WT strain LM709. Green background indicates aaRS mutants with RelA-dependent decrease in ciprofloxacin susceptibility and red background indicates aaRS mutants with RelA-independent decrease in ciprofloxacin susceptibility. Photographs of the gradient plates may be viewed in FigureS7, and MIC values can be read in TableS1.

Garoff et al.

(7)

fitness costs of this magnitude would preclude any clinical rele- vance for aaRS mutations. However, mutations that in the laboratory have fitness costs up to 20%,²⁰such as those that up-regulate drug efflux, are found in clinical isolates.^7,8Accordingly, the influence of genotype on phenotype (concerning both susceptibility and fitness) makes it difficult to predict whether aaRS mutations contribute to the evolution of resistance in clinical isolates. We have searched WGS data on clinical isolates^8,50–53for genetic var- iations in aaRS genes. We observed many SNPs predicted to cause amino acid substitutions, including several in close proximity to the aaRS mutations described in this study, but none that was identi- cal. This shows a tolerance for variation in aaRS genes in clinical isolates, but determining whether particular aaRS variants have been selected to reduce susceptibility is non-trivial because infor- mation on the ancestry and evolutionary trajectories of clinical isolates is incomplete. However, given the association between the mutations described in this report and the induction of a stringent response, one could hypothesize that any mutation or condition that induced a stringent response would make bacteria in a clinical setting less susceptible to killing by fluoroquinolones. Interestingly, ciprofloxacin-resistant mutants with up-regulated expression of mdtK, one of the genes influenced by the leuS mutations, are found in clinical isolates, although the genetic basis for the up- regulation is currently not known.¹¹

The results of this study open up the possibility that a novel, and relatively large, genetic target could contribute to the evolution of resistance to ciprofloxacin, and other antibiotic classes.

Acknowledgements

An abstract of this project was published in the proceedings of the 27th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID) 2017, with the title ‘Functional significance of leuS mutations in Escherichia coli resistance to ciprofloxacin’, abstract number 3451.

Funding

This research was supported by grants from Vetenskapsra˚det (the Swedish Research Council) to Diarmaid Hughes (grants 2013–02904, 2016–04449, 2017–03593) and from the Scandinavian Society for Antimicrobial Chemotherapy (grant SLS-693211). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Transparency declarations

None to declare.

Supplementary data

FiguresS1toS7and TablesS1toS7appear asSupplementary dataat JAC Online.

References

1 Hooper DC, Wolfson JS, Ng EY et al. Mechanisms of action of and resistance to ciprofloxacin. Am J Med 1987; 82: 12–20.

2 Khodursky AB, Zechiedrich EL, Cozzarelli NR. Topoisomerase IV is a target of quinolones in Escherichia coli. Proc Natl Acad Sci USA 1995; 92: 11801–5.

3 Hooper DC. Mechanisms of action of antimicrobials: focus on fluoroquinolones. Clin Infect Dis 2001; 32 Suppl 1: S9–15.

4 Barnard FM, Maxwell A. Interaction between DNA gyrase and quinolones:

effects of alanine mutations at GyrA subunit residues Ser(83) and Asp(87).

Antimicrob Agents Chemother 2001; 45: 1994–2000.

5 Hooper DC. Mechanisms of fluoroquinolone resistance. Drug Resist Updat 1999; 2: 38–55.

6 Jacoby GA. Mechanisms of resistance to quinolones. Clin Infect Dis 2005;

41 Suppl 2: S120–6.

7 Mazzariol A, Tokue Y, Kanegawa TM et al. High-level fluoroquinolone- resistant clinical isolates of Escherichia coli overproduce multidrug efflux protein AcrA. Antimicrob Agents Chemother 2000; 44: 3441–3.

8 Komp Lindgren P, Karlsson A˚, Hughes D. Mutation rate and evolution of fluoroquinolone resistance in Escherichia coli isolates from patients with urinary tract infections. Antimicrob Agents Chemother 2003; 47: 3222–32.

9 Wang H, Dzink-Fox JL, Chen M et al. Genetic characterization of highly fluoroquinolone-resistant clinical Escherichia coli strains from China: role of acrR mutations. Antimicrob Agents Chemother 2001; 45: 1515–21.

10 Koutsolioutsou A, Pena-Llopis S, Demple B. Constitutive soxR mutations contribute to multiple-antibiotic resistance in clinical Escherichia coli isolates.

11 Vinue L, Hooper DC, Jacoby GA. Chromosomal mutations that accom- pany qnr in clinical isolates of Escherichia coli. Int J Antimicrob Agents 2018; 51: 479–83.

12 Cohen SP, McMurry LM, Hooper DC et al. Cross-resistance to fluoroquinolones in multiple-antibiotic-resistant (Mar) Escherichia coli selected by tetracycline or chloramphenicol: decreased drug accumulation associated with membrane changes in addition to OmpF reduction. Antimicrob Agents Chemother 1989; 33: 1318–25.

13 Nikaido H. Multidrug efflux pumps of gram-negative bacteria. J Bacteriol 1996; 178: 5853–9.

14 Martinez-Martinez L, Pascual A, Jacoby GA. Quinolone resistance from a transferable plasmid. Lancet 1998; 351: 797–9.

15 Mammeri H, Van De Loo M, Poirel L et al. Emergence of plasmid- mediated quinolone resistance in Escherichia coli in Europe. Antimicrob Agents Chemother 2005; 49: 71–6.

16 Jacoby GA, Strahilevitz J, Hooper DC. Plasmid-mediated quinolone resistance. Microbiol Spectr 2014; 2: doi:10.1128/microbiolspec.

17 Robicsek A, Strahilevitz J, Jacoby GA et al. Fluoroquinolone-modifying enzyme: a new adaptation of a common aminoglycoside acetyltransferase.

Nat Med 2006; 12: 83–8.

18 Yamane K, Wachino J, Suzuki S et al. New plasmid-mediated fluoroquinolone efflux pump, QepA, found in an Escherichia coli clinical isolate.

19 Perichon B, Courvalin P, Galimand M. Transferable resistance to amino- glycosides by methylation of G1405 in 16S rRNA and to hydrophilic fluoroquinolones by QepA-mediated efflux in Escherichia coli. Antimicrob Agents Chemother 2007; 51: 2464–9.

20 Marcusson LL, Frimodt-Moller N, Hughes D. Interplay in the selection of fluoroquinolone resistance and bacterial fitness. PLoS Pathog 2009; 5:

e1000541.

21 Komp Lindgren P, Marcusson LL, Sandvang D et al. Biological cost of single and multiple norfloxacin resistance mutations in Escherichia coli implicated in urinary tract infections. Antimicrob Agents Chemother 2005; 49:

2343–51.

22 Sato T, Yokota S, Uchida I et al. A fluoroquinolone-resistant Escherichia coli clinical isolate without quinolone resistance-determining region mutations found in Japan. Antimicrob Agents Chemother 2011; 55: 3964–5.

(8)

23 Sato T, Yokota SI, Uchida I et al. Fluoroquinolone resistance mechanisms in an Escherichia coli isolate, HUE1, without quinolone resistance-determining region mutations. Front Microbiol 2013; 4: 125.

24 Vinue L, Corcoran MA, Hooper DC et al. Mutations that enhance the ciprofloxacin resistance of Escherichia coli with qnrA1. Antimicrob Agents Chemother 2015; 60: 1537–45.

25 Pietsch F, Bergman JM, Brandis G et al. Ciprofloxacin selects for RNA polymerase mutations with pleiotropic antibiotic resistance effects. J Antimicrob Chemother 2017; 72: 75–84.

26 Zhou K, Zhou L, Lim Q et al. Novel reference genes for quantifying transcriptional responses of Escherichia coli to protein overexpression by quantita- tive PCR. BMC Mol Biol 2011; 12: 18.

27 Sharan SK, Thomason LC, Kuznetsov SG et al. Recombineering: a homologous recombination-based method of genetic engineering. Nat Protoc 2009; 4: 206–23.

28 Cherepanov PP, Wackernagel W. Gene disruption in Escherichia coli: tcR and KmR cassettes with the option of Flp-catalyzed excision of the antibiotic- resistance determinant. Gene 1995; 158: 9–14.

29 Dykhuizen DE. Experimental studies of natural selection in bacteria. Annu Rev Ecol Syst 1990; 21: 373–98.

30 Cashel M, Gentry DR, Hernandez VJ et al. The stringent response. In:

Neidhardt FC, Curtiss R III, Ingraham JL et al., eds. Escherichia coli and Salmonella: Cellular and Molecular Biology. Washington, DC, USA: ASM Press, 1996; 1458–96.

31 Durfee T, Hansen AM, Zhi H et al. Transcription profiling of the stringent response in Escherichia coli. J Bacteriol 2008; 190: 1084–96.

32 Kudrin P, Varik V, Oliveira SR et al. Subinhibitory concentrations of bac- teriostatic antibiotics induce relA-dependent and relA-independent tolerance to b-lactams. Antimicrob Agents Chemother 2017; 61: pii: e02173–16.

33 Hobbs EC, Yin X, Paul BJ et al. Conserved small protein associates with the multidrug efflux pump AcrB and differentially affects antibiotic resistance.

Proc Natl Acad Sci USA 2012; 109: 16696–701.

34 Du D, Voss J, Wang Z et al. The pseudo-atomic structure of an RND-type tripartite multidrug efflux pump. Biol Chem 2015; 396: 1073–82.

35 Saier MH, Reddy VS, Tsu BV et al. The transporter classification database (TCDB): recent advances. Nucleic Acids Res 2016; 44: D372–9.

36 Hirai K, Aoyama H, Irikura T et al. Differences in susceptibility to quinolones of outer-membrane mutants of Salmonella typhimurium and Escherichia coli. Antimicrob Agents Chemother 1986; 29: 535–8.

37 Eiler S, Dock-Bregeon A, Moulinier L et al. Synthesis of aspartyl-tRNA(Asp) in Escherichia coli—a snapshot of the second step. EMBO J 1999; 18: 6532–41.

38 Mermershtain I, Finarov I, Klipcan L et al. Idiosyncrasy and identity in the prokaryotic Phe-system: crystal structure of E. coli phenylalanyl-tRNA synthetase complexed with phenylalanine and AMP. Protein Sci 2011; 20:

160–7.

39 Sankaranarayanan R, Dock-Bregeon AC, Romby P et al. The structure of threonyl-tRNA synthetase-tRNA(Thr) complex enlightens its repressor activity and reveals an essential zinc ion in the active site. Cell 1999; 97: 371–81.

40 Palencia A, Crepin T, Vu MT et al. Structural dynamics of the aminoacyla- tion and proofreading functional cycle of bacterial leucyl-tRNA synthetase.

Nat Struct Mol Biol 2012; 19: 677–84.

41 Thulin E, Sundqvist M, Andersson DI. Amdinocillin (mecillinam) resistance mutations in clinical isolates and laboratory-selected mutants of Escherichia coli. Antimicrob Agents Chemother 2015; 59: 1723–32.

42 Vinella D, D’Ari R, Jaffe A et al. Penicillin binding protein 2 is dispensable in Escherichia coli when ppGpp synthesis is induced. EMBO J 1992; 11:

1493–501.

43 Jovanovic M, Lilic M, Janjusevic R et al. tRNA synthetase mutants of Escherichia coli K-12 are resistant to the gyrase inhibitor novobiocin. J Bacteriol 1999; 181: 2979–83.

44 Ogura T, Bouloc P, Niki H et al. Penicillin-binding protein-2 is essential in wild-type Escherichia coli but not in lov or cya mutants. J Bacteriol 1989; 171:

3025–30.

45 Rodionov DG, Ishiguro EE. Direct correlation between overproduction of guanosine 3⁰, 5⁰-bispyrophosphate (ppGpp) and penicillin tolerance in Escherichia coli. J Bacteriol 1995; 177: 4224–9.

46 Kwon YK, Higgins MB, Rabinowitz JD. Antifolate-induced depletion of intracellular glycine and purines inhibits thymineless death in E. coli. ACS Chem Biol 2010; 5: 787–95.

47 Nishino K, Yamaguchi A. Analysis of a complete library of putative drug transporter genes in Escherichia coli. J Bacteriol 2001; 183: 5803–12.

48 Zeng D, Zhao J, Chung HS et al. Mutants resistant to LpxC inhibitors by rebalancing cellular homeostasis. J Biol Chem 2013; 288: 5475–86.

49 Chen CR, Malik M, Snyder M et al. DNA gyrase and topoisomerase IV on the bacterial chromosome: quinolone-induced DNA cleavage. J Mol Biol 1996; 258: 627–37.

50 Clark G, Paszkiewicz K, Hale J et al. Genomic analysis uncovers a phenotypically diverse but genetically homogeneous Escherichia coli ST131 clone circulating in unrelated urinary tract infections. J Antimicrob Chemother 2012; 67: 868–77.

51 Price LB, Johnson JR, Aziz M et al. The epidemic of extended-spectrum-b- lactamase-producing Escherichia coli ST131 is driven by a single highly patho- genic subclone, H30-Rx. MBio 2013; 4: e00377–13.

52 Petty NK, Ben Zakour NL, Stanton-Cook M et al. Global dissemination of a multidrug resistant Escherichia coli clone. Proc Natl Acad Sci USA 2014; 111:

5694–9.

53 Xia E, Khong WX, Marimuthu K et al. Draft genome sequence of a multidrug-resistant New Delhi metallo-b-lactamase-1 (NDM-1)-producing Escherichia coli isolate obtained in Singapore. Genome Announc 2013; 1:

e01020–13.