Plasmid-Mediated Ciprofloxacin Resistance Imparts a Selective Advantage on Escherichia coli ST131

ABSTRACT

Escherichia coli ST131 is a recently emerged antibiotic resistant clone responsible for high rates of urinary tract and bloodstream infections. Despite its global dominance, the precise mechanisms that have driven the rapid dissemination of ST131 remain unknown. Here, we show that the plasmid-associated resistance gene encoding the AAC(6’)-Ib-cr enzyme that inactivates the fluoroquinolone (FQ) antibiotic ciprofloxacin is present in >70% of strains from the most rapidly expanding subgroup of multidrug resistant ST131. Using a series of genome-edited and plasmid-cured isogenic strains, we demonstrate that the aac(6’)-Ib-cr gene confers a selective advantage on ST131 in the presence of ciprofloxacin, even in strains containing chromosomal GyrA and ParC FQ-resistance mutations. Further, we identify a pattern of emerging carbapenem resistance in other common E. coli clones carrying both aac(6’)-Ib-cr and chromosomal FQ-resistance mutations, suggesting this dual resistance combination may also impart a selective advantage on these non-ST131 antibiotic resistant lineages.

INTRODUCTION

The Escherichia coli ST131 clone is of major clinical importance due to its multidrug resistance phenotype and high prevalence worldwide. Alarmingly, just over a decade since its first identification, ST131 has become the overwhelmingly predominant E. coli clone responsible for urinary tract and bloodstream infections. Genomic analyses have classified ST131 into three major clades (A, B, and C), with all strains from the globally dominant clade C (H30) sublineage containing target-modifying mutations in the DNA gyrase (GyrA; S83L, D87N) and topoisomerase IV (ParC; S80I) chromosome replication enzymes that together confer high-level resistance to fluoroquinolones (FQs) (1–4). Clade C strains can be further subdivided into clade C1 (H30 non-Rx) and C2 (H30Rx), with clade C2 strains largely positive for the CTX-M-15 extended-spectrum β-lactamase (ESBL) (1, 2), the most common type of ESBL worldwide (5).

FQs are one of the most commonly prescribed antibiotics globally, with resistance to FQs considered to be a strong driver of ST131 dissemination. While chromosomal GyrA and ParC modifications form a ubiquitous mechanism of FQ resistance in clade C ST131, other mechanisms of FQ resistance have also been characterized, including enhanced drug efflux and target modification by the plasmid-encoded AAC(6’)-Ib-cr enzyme (6). AAC(6’)-Ib-cr is a variant of the AAC(6’)-Ib aminoglycoside acetyltransferase that has been modified by the acquisition of two mutations (W102R and D179Y) necessary for its acetyltransferase function (7). Among the FQ antibiotics, ciprofloxacin and norfloxacin are subjected to N-acetylation (deactivation) by AAC(6’)-Ib-cr due to the availability of an unsubstituted piperazinyl group (8). Levofloxacin and moxifloxacin lack this unsubstituted piperazinyl group and therefore are not subjected to acetylation. AAC(6’)-Ib-cr activity confers low-level reduced susceptibility to ciprofloxacin and norfloxacin, however it can act additively with other mechanisms to achieve clinically significant resistance (7, 9).

The precise molecular mechanisms that have led to the dominance of ST131 over other FQ-resistant E. coli clones remain elusive. Several studies have reported an association between the aac(6’)-Ib-cr gene and clade C2 FQ-resistant ST131 (4, 6, 10, 11), supporting the assertion that expression of the plasmid-encoded AAC(6’)-Ib-cr enzyme amplifies the level of FQ resistance mediated by chromosomal GyrA and ParC modifications (6). Here, we curated a large ST131 genome data set to explore the prevalence of the aac(6’)-Ib-cr gene in different ST131 sublineages. We also generated a series of genome-edited and plasmid modified isogenic strains to experimentally define the precise contribution of the aac(6’)-Ib-cr gene to FQ resistance in ST131 backgrounds comprising different GyrA and ParC modifications. We show that the aac(6’)-Ib-cr gene provides a selective advantage in the presence of ciprofloxacin, and extend this observation to reveal an emerging link between aac(6’)-Ib-cr carriage, chromosomal FQ-resistance mutations, and resistance to last-line carbapenems. Overall, our data identify the acquisition of plasmids carrying the aac(6’)-Ib-cr gene as one element of a complex multifactorial genetic landscape linked to the global dominance of clade C2 ST131, and reveal that acquisition of the aac(6’)-Ib-cr gene can be linked to emerging carbapenem resistance in other FQ-resistant E. coli clones.

RESULTS

The plasmid-encoded aac(6’)-Ib-cr gene is highly prevalent in clade C2 ST131.

To examine the prevalence of the aac(6’)-Ib-cr gene in the context of the ST131 phylogeny, we downloaded a large data set of 3,391 ST131 genomes from EnteroBase (12) and built a core SNP maximum likelihood tree that reproduced its established clade A-B-C1/C2 structure. We then used a blast search to detect the presence of aac(6’)-Ib-cr in each genome. Among a total of 2,737 clade C strains, the aac(6’)-Ib-cr gene was present in significantly more clade C2 strains (1,378/1,902; 72%) compared to clade C1 strains (31/835; 3.7%) (Fig. 1A). Only 1% of clade A strains and 0.6% of clade B strains contained the aac(6’)-Ib-cr gene. We also used the same data set to probe the prevalence of the bla_CTX-M-15 gene, and revealed a very similar distribution to that observed for aac(6’)-Ib-cr, with 1,582/1,902 (83%) clade C2 strains harboring the bla_CTX-M-15 gene compared to 45/835 (5.4%) clade C1 strains (Fig. 1B). Taken together, this demonstrates the aac(6’)-Ib-cr gene is largely contained within clade C2 ST131, and for the majority of these strains the aac(6’)-Ib-cr gene is likely to be co-localized on a plasmid with the bla_CTX-M-15 gene.

FIG 1.

Prevalence of the aac(6’)-Ib-cr and bla_CTX-M-15 genes in ST131. (A) (i) Barplot showing the prevalence of aac(6’)-Ib-cr in each ST131 based on clade designation, and (ii) core SNP ML tree of 3,391 ST131 genomes with the presence of the aac(6’)-Ib-cr gene indicated by pink dots. (B) (i) Barplot showing the prevalence of bla_CTX-M-15 in each ST131 clade and (ii) core SNP ML tree of 3,391 ST131 genomes with the presence of the bla_CTX-M-15 gene indicated by blue dots. Statistical analyses based on Pearson's chi-squared test; P < 0.001 for pairs of C2-A, C2-B, C2-C1, and C2-Other.

The aac(6’)-Ib-cr gene imparts a growth advantage on ST131 in the presence of sub-MIC ciprofloxacin.

We hypothesized that the aac(6’)-Ib-cr gene confers a growth advantage in the presence of ciprofloxacin, and tested this by examining the growth kinetics of two clade C2 strains, EC958 and S101EC, as well as their plasmid-cured and aac(6’)-Ib-cr deletion-derivative mutants. Both EC958 and S101EC displayed an identical growth profile in the presence of sub-MIC ciprofloxacin (128 μg/ml; Fig. 2), consistent with their resistance phenotype (MIC ≥256 μg/ml). However, when the plasmid harboring the aac(6’)-Ib-cr gene was cured from both strains, they exhibited reduced growth in the presence of 128 μg/ml ciprofloxacin (Fig. 2). To confirm this growth defect was specifically due to loss of aac(6’)-Ib-cr, we deleted the gene in both strains and re-examined their growth profile. Both EC958 and S101EC aac(6’)-Ib-cr deletion-derivative mutants displayed the same growth-impaired phenotype as the plasmid-cured strains in the presence of 128 μg/ml ciprofloxacin (Fig. 2).

FIG 2.

Growth kinetics of wild-type ST131 strains and their aac(6’)-Ib-cr deletion derivatives in sub-MIC of ciprofloxacin. (A) Growth curves comparing growth of EC958 (WT), EC958cured (cured), and EC958Δaac(6’)-Ib-cr (Δaac). (B) Growth curves comparing growth of S101EC (WT), S101ECcured (cured), and S101ECΔaac(6’)-Ib-cr (Δaac). All experiments were performed in triplicate in MH supplemented with 128 μg/ml ciprofloxacin at 37°C. The gray shades indicate SE range.

To better understand the impact of aac(6’)-Ib-cr, we performed mixed competition assays using the EC958 aac(6’)-Ib-cr deletion strain and wild-type EC958. The competition was performed in both MH broth and human urine, with and without the presence of 64 μg/ml ciprofloxacin (Fig. 3). The aac(6’)-Ib-cr deletion strain was out-competed by wild-type EC958 in the presence of ciprofloxacin in both MH broth and human urine. However, no selective advantage was observed in the absence of ciprofloxacin. Together, these data demonstrate a selective advantage conferred by the aac(6’)-Ib-cr gene in the presence of ciprofloxacin in a strain already harboring the gyrA and parC FQ-resistance alleles.

FIG 3.

Competition assays showing the selective advantage of EC958 compared to EC958Δaac(6’)-Ib-cr in human urine and MH broth in the absence and presence of sub-MIC of ciprofloxacin (64 μg/ml). The data represents biological replicates with individual data points shown; the group median is indicated by a horizontal line. Statistical analyses were performed using the Wilcoxon signed rank test comparing each group against theoretical median of 0. *, P < 0.05.

Consistent with these results, both the plasmid-cured and aac(6’)-Ib-cr deletion mutants also exhibited a 4-fold reduction in MIC against ciprofloxacin (Table 1) despite the presence of FQ resistance-defining chromosomal GyrA and ParC modifications. AAC(6’)-Ib-cr is a dual function enzyme with aminoglycoside acetyltransferase activity, and its loss also caused a reduction in the MIC to amikacin (4-fold) and kanamycin (32-fold) (Table 1).

TABLE 1.

MIC of ST131 strains with and without the aac(6’)-Ib-cr gene^a

Strain	Ciprofloxacin	Amikacin	Kanamycin
EC958	>256	16	96
EC958cured	64	4	3
EC958Δaac(6’)-Ib-cr	64	3	3
S101EC	256	16	128
S101ECcured	64	4	4
S101ECΔaac(6’)-Ib-cr	64	3	3

^aMIC for ciprofloxacin was performed using broth microdilution method; amikacin and kanamycin were determined using the Etest.

Stepwise acquisition of aac(6’)-Ib-cr and chromosomal GyrA/ParC modifications shape resistance to ciprofloxacin.

All clade C ST131 strains possess chromosomal GyrA (S83L, D87N) and ParC (S80I) modifications, henceforth referred to as gyrA^RparC^R, that confer FQ resistance. To better understand the combinatory effect of the aac(6’)-Ib-cr gene and gyrA^RparC^R chromosomal modifications on ciprofloxacin susceptibility, we generated a unique set of genome-edited isogenic EC958 strains where the gyrA^R and parC^R alleles were reverted to the ‘sensitive’ allele type. Three isogenic mutants were constructed, namely, parC^S (EC958 ParC I80S), gyrA^S (EC958 GyrA L83S, N87D), and gyrA^SparC^S (EC958 GyrA L83S, N87D, and ParC I80S), and confirmed by genome sequencing (Table S2). These three mutants were also cured of the plasmid pEC958 (which harbors the aac(6’)-Ib-cr gene), yielding a set of eight isogenic strains containing different combinations of chromosomal and plasmid-encoded FQ susceptibility profiles. Analysis of the MIC against ciprofloxacin for all of these strains revealed the precise contributions of modifications to GyrA and ParC to FQ susceptibility (GyrA^RParC^R >256 μg/ml; GyrA^RParC^S = 8 μg/ml; GyrA^SParC^R = 0.125 μg/ml; GyrA^SParC^S <0.125 μg/ml) (Fig. 4). Notably, loss of plasmid pEC958, and hence the aac(6’)-Ib-cr gene, resulted in an ∼4-fold reduction in MIC to ciprofloxacin in all of the different FQ resistant/susceptible chromosome backgrounds.

FIG 4.

Comparative effects of chromosomal mutations and the aac(6’)-Ib-cr gene on EC958 susceptibility to ciprofloxacin. The four chromosomal backgrounds were WT (EC958), parC^S (EC958 ParC I80S), gyrA^S (EC958 GyrA L83S, N87D), gyrA^SparC^S (EC958 GyrA L83S, N87D, and ParC I80S) as shown in the lower panel. For each background, a strain with and a strain without plasmid pEC958 was tested. MIC determinations were performed using the broth microdilution method in triplicate. The data are presented as the mid-point and range of the three MIC values. The EUCAST breakpoint of 0.5 μg/ml is shown as a red dashed line.

Bacterial regrowth after ciprofloxacin treatment is mediated by aac(6’)-Ib-cr gene.

To further characterize the role of the aac(6’)-Ib-cr gene in response to ciprofloxacin treatment, we performed time-kill kinetic assays using the EC958 gyrA^RparC^R and EC958 gyrA^SparC^S strains in the presence and absence of aac(6’)-Ib-cr. Survival of the EC958 gyrA^RparC^R strain containing aac(6’)-Ib-cr was only minimally impacted in the presence of ciprofloxacin, and growth occurred even at a concentration of 1024 μg/ml (4x MIC). In contrast, the absence of aac(6’)-Ib-cr in EC958 gyrA^RparC^R led to a loss in viability after 24 h in the presence of 256 μg/ml (4x MIC) ciprofloxacin (Fig. 5A and B). Time-kill assays for the EC958 gyrA^SparC^S strains ± aac(6’)-Ib-cr were performed at lower concentrations of ciprofloxacin, in line with their increased FQ susceptibility. However, an identical pattern was observed, whereby compared to EC958 gyrA^SparC^S harboring aac(6’)-Ib-cr, loss of the aac(6’)-Ib-cr gene in EC958 gyrA^SparC^S resulted in complete killing after 24 h in the presence of 0.25 μg/ml (4× MIC) ciprofloxacin (Fig. 5C and D). Taken together, these data demonstrate that the AAC(6’)-Ib-cr enzyme enhances survival and growth against ciprofloxacin administered at concentrations above its MIC.

FIG 5.

Time-kill kinetic assays showing regrowth during ciprofloxacin treatment in the presence and absence of the aac(6’)-Ib-cr gene. The assays were performed on (A) EC958 wild-type, (B) EC958cured, (C) EC958 gyrA^SparC^S, and (D) EC958cured gyrA^SparC^S. The line type and symbols denote CFU counts following exposure to different ciprofloxacin concentrations, with concentrations relative to MIC shown in brackets as fold difference (i.e., 1×, 2×, and 4×). All experiments were performed in triplicate and the data are shown as mean ± SE.

Deactivation of ciprofloxacin by the AAC(6’)-Ib-cr enzyme enhances the survival of FQ resistant and FQ susceptible strains.

The AAC(6’)-Ib-cr enzyme converts ciprofloxacin to an inactive N-acetyl-ciprofloxacin form, and thus we hypothesized it would reduce the local concentration of ciprofloxacin during bacterial growth, leading to increased survival of both resistant and sensitive cells. To test this, we grew EC958 and the EC958 aac(6’)-Ib-cr deletion mutant in MH broth containing 64 μg/ml ciprofloxacin (Fig. 6A). At periodic intervals, samples were collected from both cultures, filter-sterilized, and the concentration of ciprofloxacin and N-acetyl-ciprofloxacin was measured directly by ultra-high-performance liquid chromatographic-photodiode array (UHPLC-PDA). These analyses demonstrated rapid inactivation of ciprofloxacin during growth, with the concentrations of ciprofloxacin and inactive N-acetyl-ciprofloxacin displaying an inverse relationship over time (Fig. 6B). To measure the impact of ciprofloxacin deactivation, we employed a disc diffusion assay using the ciprofloxacin sensitive E. coli strain ATCC 25922 in combination with filter sterilized broth collected during the growth assay. Strain ATCC 25922 was seeded onto a MH agar plate, a filter paper disc loaded with filtered broth was placed on the agar, and the plates were incubated overnight, with the size of the zone of inhibition used as an indicator of the amount of active ciprofloxacin in the filtered broth. We observed that the zone of inhibition from EC958 broth decreased from 35 mm at 0h (corresponding to 64 μg/ml ciprofloxacin) to 12 mm at 8h. In contrast, the zone of inhibition from the EC958 aac(6’)-Ib-cr deletion mutant was the same at both time points. Thus, the AAC(6’)-Ib-cr enzyme rapidly deactivates ciprofloxacin, thereby generating a local environment that enhances the survival of susceptible cells.

FIG 6.

Kinetics of ciprofloxacin conversion to N-acetyl-ciprofloxacin by AAC(6’)-Ib-cr during growth. (A) Growth kinetics of EC958 and EC958aac (EC958Δaac(6’)-Ib-cr) in MH broth supplemented with 64 μg/ml ciprofloxacin. (B) Measurements of ciprofloxacin or N-acetyl-ciprofloxacin concentrations performed by ultra-high-performance liquid chromatographic-photodiode array, demonstrating a decrease of ciprofloxacin and increase of N-acetyl-ciprofloxacin in EC958 broth cultures compared to EC958acc broth cultures, where the concentration of ciprofloxacin was unchanged. (C) Zones of inhibition from disc diffusion assays using 5 μl of filter-sterilized broth from each time point in (A) loaded onto filter paper and placed on MH agar seeded with the ciprofloxacin susceptible E. coli ATCC 25922 strain. In each case, the zone of inhibition is proportional to the amount of active ciprofloxacin in the respective broth culture.

Emergence of carbapenem resistant E. coli carrying both aac(6’)-Ib-cr and chromosomal FQ-resistance modifications.

Given the selective advantage of carrying both chromosomal FQ-resistance modifications and the plasmid-associated aac(6’)-Ib-cr gene, we hypothesized that E. coli isolates carrying this dual FQ-resistance combination would be more common than isolates that carry aac(6’)-Ib-cr alone. To investigate this, we randomly selected 100 isolates from the top 100 most common E. coli sequence types (STs) from EnteroBase and calculated the percentage of isolates carrying both FQ-resistance mechanisms compared to aac(6’)-Ib-cr alone. Isolates carrying both FQ-resistance mechanisms were identified in 22 STs, ranging from 1% to 58% (Fig. 7A). In contrast, isolates from 10 STs carried the aac(6’)-Ib-cr gene alone, ranging from 1% to 11%. These combinations were largely mutually exclusive, with carriage of the dual FQ-resistance combination excluding carriage of the plasmid-encoded aac(6’)-Ib-cr gene alone within an individual ST.

FIG 7.

ST-based phylogenetic clustering of strains carrying gyrA^RparC^R, aac(6’)-Ib-cr gene and carbapenem resistance gene combinations. (A) Barplot demonstrating the strong association between the aac(6’)-Ib-cr gene and chromosomal gyrA^RparC^R modifications compared to carriage of aac(6’)-Ib-cr gene alone in different E. coli STs. (B) to (F) Phylogenetic trees of each ST built from 100 randomly selected genomes using parsnps. The tips with red dots indicate genomes that possess the gyrA^RparC^R FQ resistance modifications, the aac(6’)-Ib-cr gene, and any of the indicated carbapenem resistance genes. The clustering of genomes that carry this resistance combination is indicated by pink shading. The side columns depict the absence (gray) or presence (blue) of gyrA^RparC^R modifications (GyrA S83L D87N and ParC S80I), the aac(6’)-Ib-cr gene, bla_CTX-M genes (CTX-M), bla_KPC genes (KPC), bla_VIM genes (VIM), bla_IMP genes (IMP), bla_NDM genes (NDM), and bla_OXA genes (OXA, including bla_OXA-48 and bla_OXA-181).

Several STs contained isolates that carried the FQ-resistance combination at levels approximately equal to ST131 (ST448 = 31% vs ST131 = 33%) or greater than ST131 (ST405 = 41%; ST648 = 42%; ST167 = 43%; ST410 = 46%; ST617 = 58%). Given the FQ-resistance combination in ST131 is strongly associated with the bla_CTX-M-15 gene, hence promoting additional resistance to 3rd-generation cephalosporins, we also examined the carriage of bla_CTX-M ESBL genes and genes encoding resistance to last-line carbapenems in five STs. Carriage of a bla_CTX-M ESBL gene was identified in >50% of isolates from all STs except ST448, with ST617, ST410, and ST167 displaying high co-carriage rates of the aac(6’)-Ib-cr and bla_CTX-M genes as we observed in clade C2 ST131 (Fig. 7B to F). In the case of carbapenem resistance genes, the bla_NDM gene was most common. We identified at least one cluster in each ST that was enriched for the combined carriage of the FQ-resistance combination as well as at least one type of carbapenem resistance gene (Fig. 7B to F). Interestingly, the possession of carbapenem resistance genes was less frequent in a similarly generated set of 100 randomly selected ST131 genomes (Fig. S1). Taken together, this highlights the existence of multiple global E. coli clones armed with critical resistance combinations that threaten to cause increased rates of infection associated with fewer treatment options.

DISCUSSION

FQ resistance in E. coli generally develops following a stepwise acquisition of resistance mutations in which alterations in the primary target, the DNA gyrase (gyrA), precede those in the secondary target, the topoisomerase IV (parC); mutations in both DNA gyrase and topoisomerase IV confer a higher level of resistance (13). Plasmid-mediated FQ resistance mechanisms only confer a small decrease in susceptibility and may prime the development of the first mutation in the DNA gyrase (14). All isolates from clade C of the E. coli ST131 clone harbor two mutations in DNA gyrase (S83L and D87N) and one in topoisomerase IV (S80I), and therefore the presence of the plasmid-encoded acc(6’)-Ib-cr gene in these isolates attracts little attention because its phenotype is dominated by the higher level of resistance provided by the chromosomal gyrA and parC mutations. Here, we show that the acc(6’)-Ib-cr gene is found in the overwhelming majority of multidrug resistant ST131 strains from clade C2, leading to a selective advantage in the presence of ciprofloxacin at a much higher concentrations than previously associated with the acc(6’)-Ib-cr gene.

AAC(6’)-Ib-cr is a dual function enzyme with activity against aminoglycosides and the FQ antibiotics ciprofloxacin and norfloxacin (7). Our data showed that loss of the aac(6’)-Ib-cr gene in ST131 clade C2 strains (EC958 and S101EC) caused a growth defect in the presence of sub-MIC ciprofloxacin compared to the corresponding WT strain, which translated into a 4-fold reduction in MIC. In contrast to a previous study that started with the sensitive ATCC 25922 strain and introduced isogenic FQ resistance mutations in gyrA and parC (15), we assessed the dynamics of FQ susceptibility by converting the gyrA and parC mutations in EC958, a clade C2 ST131 strain with clinically significant FQ resistance, into the sensitive allele type. These allelic reversion mutants were generated using the pORTMAGE vector exchange system, which uses transient suppression of the mismatch repair MutL enzyme to introduce oligonucleotide-directed mutations on the chromosome (16). The EC958 isogenic strains converted to the sensitive allele of parC^S (EC958 ParC I80S), gyrA^S (EC958 GyrA L83S, N87D), and gyrA^SparC^S (EC958 GyrA L83S, N87D, and ParC I80S) all displayed an ∼4-fold reduction in MIC against ciprofloxacin in the absence of aac(6’)-Ib-cr (compared to the corresponding parent strain containing aac(6’)-Ib-cr). Thus, we were able to show that the increased resistance conferred by AAC(6’)-Ib-cr is consistent across a very wide range of MICs (0.064 to >256 μg/ml), including a clinically significant range that could influence therapeutic outcomes.

The presence of aac(6’)-Ib-cr also led to a time-dependent reduction in ciprofloxacin-mediated killing. Strains carrying the gyrA/parC sensitive allele type were susceptible to the bactericidal effect of ciprofloxacin in the first 5 h, with subsequent bacterial regrowth driven by the presence of the aac(6’)-Ib-cr gene. This phenomenon has also been observed previously, but at lower ciprofloxacin concentrations and employing strains engineered to possess ciprofloxacin resistance (15, 17). It is worth noting that growth of the wild-type EC958 strain was only inhibited in the first 3h at a ciprofloxacin concentration of 1056 μg/ml. This concentration is the maximum human urine concentration achieved from a single treatment dose of 500 mg ciprofloxacin (ranging from 23–1000 μg/ml, with the median ranging from 237–407 μg/ml in the first 6h (18–20)). The ability of EC958 to grow in high concentrations of ciprofloxacin was also linked to the acetylation of ciprofloxacin by AAC(6’)-Ib-cr. During growth of EC958 in broth containing ciprofloxacin, AAC(6’)-Ib-cr activity caused the rapid conversion of active ciprofloxacin to inactive N-acetyl-ciprofloxacin, consistent with previous studies examining environmental E. coli isolates and other clinical Enterobacteriaceae isolates harboring the aac(6’)-Ib-cr gene (21, 22). In addition to priming the development of clinically relevant resistance, AAC(6’)-Ib-cr mediated reduction of local ciprofloxacin concentrations would lead to the enhanced survival of sensitive strains, thereby driving population-wide resistance. Indeed, this may contribute to the diversification of plasmid composition and resistance genes carried by ST131 that persist in the gut of patients that suffer recurrent UTI despite extensive antibiotic treatment (23).

The strong association between gyrA^RparC^R chromosomal mutations and the plasmid-encoded aac(6’)-Ib-cr gene in ST131 suggests this dual resistance combination may be a driver of its clonal expansion. Thus, we also explored the prevalence of this resistance combination in other common E. coli STs linked to increasing infection rates. Among the top 100 E. coli STs in Enterobase, 22 STs comprised strains harboring both FQ resistance mechanisms, some at an incidence as high as 58%. Closer examination of five STs with a high prevalence of the gyrA^RparC^R chromosomal mutations and the aac(6’)-Ib-cr gene revealed a strong association with carbapenem resistance genes. Indeed, three STs strongly linked to carriage of the bla_NDM gene, ST167, ST617 and ST410 (24), were among the top 5 STs in our analysis. Despite increasing antibiotic stewardship, ciprofloxacin remains one of the most widely used antibiotics worldwide, with its consumption (including both human and animal) reported by the WHO as high as 9.8% of total defined daily doses in the Regions of the Americas, 7% in the European Region, and 22.9% in the Western Pacific Region (WHO Report on Surveillance of Antibiotic Consumption (25)), emphasizing the importance of monitoring epidemiological co-resistance patterns.

In conclusion, we have shown that in E. coli ST131, the combination of aac(6’)-Ib-cr and chromosomal mutations in gyrA and parC represents an important dual resistance mechanism that has contributed to its global dissemination. Although the precise timeline of aac(6’)-Ib-cr acquisition remains to be determined, we hypothesize that the aac(6’)-Ib-cr gene was acquired after the gain of FQ resistance via gyrA and parC mutations (i.e., this occurred after the emergence of clade C), and that the aac(6’)-Ib-cr gene acts in concert with the chromosomal mutations to increase the selective advantage of clade C2 strains in the presence of ciprofloxacin. Furthermore, this combination has emerged in multiple other STs, including those that also harbor carbapenem resistance genes. As our dependence on last-line carbapenems grows, this presents the alarming scenario that multiple clones are poised to survive and spread, highlighting the need for ongoing surveillance of aac(6’)-Ib-cr carriage and improved control of ciprofloxacin usage.

MATERIALS AND METHODS

Ethics statement.

Approval for the collection of human urine was obtained from the University of Queensland Institutional Human Research Ethics Committee (2015000347). Participation was voluntary and all individuals provided informed consent prior to participation in the study.

Bacterial strains and growth conditions.

The E. coli strain EC958 was isolated from the urine of a patient with community-acquired UTI in the UK (26). E. coli strain S101EC was from a global collection of E. coli ST131 (1). The EC958Δlac mutant has been described previously (27). Bacterial strains were routinely cultured on or in lysogeny broth (LB) medium supplemented with appropriate antibiotics. Alternatively, bacterial strains were cultured in Mueller-Hinton (MH) broth or pooled human urine, or on MacConkey lactose agar, supplemented with appropriate antibiotics as required.

Plasmid curing.

The IncF plasmids harboring the aac(6’)-Ib-cr gene were cured from EC958 and S101EC using the curing vector pMDP5-cureEC958 (GenBank accession number MZ723317.1) designed based on the pCURE2 plasmid (28). The pMDP5-cureEC958 vector contained the FIA and FIIA replicons plus the ccdA, sok, pemI, and vagC anti-toxin genes from pEC958 (synthesized by Epoch Life Science). The curing procedure has been described previously (28).

Molecular methods.

Chromosomal DNA purification, PCR, and DNA Sanger sequencing of PCR products were performed as previously described (29). The full list of primers used are outlined in Table S1. Mutation of the aac(6’)-Ib-cr gene was performed using the λ-Red recombinase method with some modifications as previously described (27, 30). The chromosomal mutations on gyrA (GyrA L(TTG)-83-S(TCG), N(AAC)-87-D(GAC)) and parC (ParC I(ATT)-80-S(TCC)) were generated using the pORTMAGE method (16). Genome sequencing of these mutants was performed using Illumina methodology (by Australian Centre for Ecogenomics) to confirm the mutations and to examine for off-target mutations (Table S2).

Bacterial growth kinetics.

Bacterial growth kinetics were measured as changes in optical density (OD₆₀₀) over time using the FLUOstar OPTIMA Microplate Reader (BMG Labtech). Briefly, overnight cultures were standardized to OD₆₀₀ = 0.05 in cation adjusted MH broth (BD, USA) or MH supplemented with ciprofloxacin at specified concentrations, and loaded in triplicate onto a flat bottom 96-well plate at 200 μl per well. The plate was then sealed using the Breathe-Easy sealing membrane (Diversified Biotech) and incubated at 37°C with shaking inside the microplate reader. Optical measurements were recorded every 15 min for 12 h; all experiments were performed in triplicate.

Mixed competition assays.

Competition assays were performed using similar conditions as described for bacterial growth kinetics. A 50:50 mixture of wild-type (EC958Δlac) and mutant strains was used. Viable counts were determined at t = 0h and t = 18h by plating on MacConkey lactose agar, which allowed the differentiation of EC958Δlac (non-lactose fermenter) and the mutant strains. Human urine, when used as culture media for competition assays, was pooled from at least three healthy female donors with no recent history of antibiotic use.

Antimicrobial susceptibility testing.

MICs were determined by broth microdilution method (31). The interpretation of MIC was performed using EUCAST breakpoints (http://www.eucast.org/clinicalbreakpoints/). All MIC assays were performed in triplicate.

Time-kill kinetics assays.

The four strains EC958 wild-type, EC958cured, EC958 gyrA^SparC^S, and EC958cured gyrA^SparC^S were used to investigate the killing effects of ciprofloxacin at various concentrations. Overnight cultures were standardized to ∼10⁸ CFU/ml in MH or MH supplemented with ciprofloxacin at specified concentrations and loaded in triplicate onto a flat bottom 96-well plate at 200 μl per well, sealed with the Breathe-Easy membrane (Diversified Biotech) and incubated at 37°C with shaking. At each time point, a 10 μl sample was taken for serial dilution and enumeration of viable counts. All experiments were performed in triplicate.

Ultra-high-performance liquid chromatographic-photodiode array.

Total concentrations of ciprofloxacin and N-acetylciprofloxacin in cation adjusted Mueller-Hinton broth were measured by an ultra-high performance liquid chromatography-photo diode array (UHPLC-PDA) method on a Nexera2 liquid chromatograph connected to an SPD-M30 photo diode array detector (Shimadzu, Kyoto, Japan). The filtered sample (1 μl) was injected onto the UHPLC-PDA. The stationary phase was Pinnacle DB IBD, 50 × 2.1 mm (1.9 μm) column (Restek, Bellefonte, USA) preceded by a C18 UHPLC analytical guard column (Phenomenex, Torrence, USA). The mobile Phase A was 30 mM sodium phosphate at pH 3, and the Mobile Phase B was a mixture of 50% 30 mM sodium phosphate at pH 3 and 50% acetonitrile (vol/vol). Separation of ciprofloxacin and N-acetylciprofloxacin was effected by a gradient from 65% to 95% mobile phase B at a flow 0.3 ml/min, producing a backpressure of approximately 2600 lb/in². Peaks for the analytes were detected at 315 nm.

Bioinformatic analysis.

A total of 3,391 E. coli ST131 genomes were downloaded from EnteroBase (05/25/2017). The genome assemblies of 100 randomly selected isolates from each of the 100 most common STs in Enterobase were downloaded on 12/18/2020. The phylogenetic tree was built from a core genome alignment generated by parsnp with recombination filtering (32), using FastTree2 (33). The clade structure of ST131 was inferred from the tree. The presence of aac(6’)-Ib-cr and other resistance genes was detected by blastn (template sequences downloaded from the Comprehensive Antibiotic Resistance Database (34), August 2020 release) using a stringent threshold of 100% identity and 100% coverage. The resistance mutations in gyrA and parC were detected by pointfinder (35). The association of the aac(6’)-Ib-cr gene and clade C2 was tested using Pearson's Chi-squared test followed by post hoc pairwise Chi-squared tests (R v4.0).