Resistance to piperacillin/tazobactam in Escherichia coli resulting from extensive IS26-associated gene amplification of bla_TEM-1

Abstract

Background

bla_TEM-1 encodes a narrow-spectrum β-lactamase that is inhibited by β-lactamase inhibitors and commonly present in Escherichia coli. Hyperproduction of bla_TEM-1 may cause resistance to penicillin/β-lactamase inhibitor (P/BLI) combinations.

Objectives

To characterize EC78, an E. coli bloodstream isolate, resistant to P/BLI combinations, which contains extensive amplification of bla_TEM-1 within the chromosome.

Methods

EC78 was sequenced using Illumina and Oxford Nanopore Technology (ONT) methodology. Configuration of bla_TEM-1 amplification was probed using PCR. Expression of bla_TEM-1 mRNA was determined using quantitative PCR and β-lactamase activity was determined spectrophotometrically in a nitrocefin conversion assay. Growth rate was assessed to determine fitness and stability of the gene amplification was assessed by passage in the absence of antibiotics.

Results

Illumina sequencing of EC78 identified bla_TEM-1B as the only acquired β-lactamase preceded by the WT P3 promoter and present at a copy number of 182.6 with bla_TEM-1B bracketed by IS26 elements. The chromosomal location of the IS26-bla_TEM-1B amplification was confirmed by ONT sequencing. Hyperproduction of bla_TEM-1 was confirmed by increased transcription of bla_TEM-1 and β-lactamase activity and associated with a significant fitness cost; however, the array was maintained at a relatively high copy number for 150 generations. PCR screening for bla_TEM amplification of isolates resistant to P/BLI combinations identified an additional strain containing an IS26-associated amplification of a bla_TEM gene.

Conclusions

IS26-associated amplification of bla_TEM can cause resistance to P/BLI combinations. This adaptive mechanism of resistance may be overlooked if simple methods of genotypic prediction (e.g. gene presence/absence) are used to predict antimicrobial susceptibility from sequencing data.

Introduction

In Denmark, penicillin/β-lactamase inhibitor (P/BLI) combinations have increased from 5.2% of total antibiotic consumption in 2008 to 15.6% in 2017 in hospitals.¹ In the USA and the UK, usage of P/BLI combinations has also increased since the beginning of the millennium and is estimated to account for 13.4% and 22% of in-hospital use of antibiotics in 2012 in the USA and the UK, respectively.^2,3 In the UK, prescription of piperacillin/tazobactam increased by 94.8% between 2008 and 2013.³ Despite increased use, resistance rates have so far remained at a modest level in Denmark with 4.7% of Escherichia coli blood culture isolates being reported as resistant to piperacillin/tazobactam in 2017.¹ Likewise, a recent study from the USA found that between 2011 and 2015 4.2% of E. coli bloodstream infections were piperacillin/tazobactam resistant, but ceftriaxone susceptible.⁴

Tazobactam inhibits most class A β-lactamases in the TEM, SHV and CTX-M series. However, some class A enzymes derived from bla_TEM-1 or bla_TEM-2 are refractory to the action of inhibitors and are classified as inhibitor resistant,⁵ although clinical isolates containing such enzymes often remain susceptible in vitro to piperacillin/tazobactam.⁶ Acquired class C (e.g. bla_CMY-2) and class D (e.g. bla_OXA-1 and bla_OXA-48-like) enzymes are not inhibited by tazobactam and are associated with resistance to piperacillin/tazobactam.^7–9 Acquired resistance genes may be selected if prevalent in a population subject to increasing use of P/BLI combinations.

Hyperproduction of β-lactamases (e.g. bla_TEM-1) may also cause resistance to P/BLI combinations. This can arise from promoter point mutations increasing promoter strength relative to the WT promoter P3, which directs relatively low levels of β-lactamase expression compared with Pa/Pb, P4 and P5 promoters.¹⁰ Also, gene duplication can cause hyperproduction.^11,12 Transient gene amplification has recently been causally linked to antibiotic heteroresistance¹³ and may be regarded as an adaptive change to selective environmental forces. Such genetic changes are likely to arise de novo because of increased use of an antibiotic in a hospital environment. Here we characterize a bloodstream isolate of E. coli resistant to P/BLI combinations because of the hyperproduction of TEM-1 β-lactamase resulting from bla_TEM-1 gene amplification within the chromosome.

Materials and methods

Study isolates

Clinical E. coli isolates resistant to P/BLI combinations (EC49, EC66, EC69, EC78, EC114 and EC120) are described in Table 1. EC101 was included as a control (resistant to ampicillin, but susceptible to P/BLI combinations, cephalosporins and cefoxitin). Study isolates were cultured and maintained without antibiotic selection.

Table 1.

Overview of isolates included in the study

Isolate	MLST	Origin	Zone diameter (mm)			TZP MIC (mg/L)b	blaTEM-1B copy number	blaTEM promotere	Porinsf	Doubling time (min)±SDg	Fold blaTEM-1 mRNA expression±SDh	β-Lactamase activity (μU/μg)±SDi
			MEC	AMC	TZP
EC49	ST131	blood	6/R	12/R	12/R	>32/R	8.7	Pa/Pb	full length	24.2 ± 1.3	29.6 ± 16.4	80.5 ± 12.3
EC66	ST131	blood	13/R	14/R	19/Ia	>32/Ra	1.5	IS3/P3	truncated ompC	25.9 ± 2.1	15.2 ± 4.9	41.3 ± 13.8
EC69	ST95	blood	9/R	13/R	13/R	>32/R	2.7c	P5	full length	24.8 ± 1.9	37.2 ± 10.9	95.6 ± 35.3
EC120	ST362	urine	20/S	10/R	14/R	>32/R	1/5d	P3/P4	full length	24.0	—	—
EC114	ST69	urine	11/R	12/R	19/Ia	>32/Ra	2.6	P4	full length	24.3 ± 0.7	13.4 ± 8.0	10.0 ± 5.9
EC78	ST567	blood	6/R	12/R	12/R	>32/R	182.6	P3	full length	32.9 ± 1.1	13.2 ± 5.7	31.6 ± 6.1
EC121	ST567	EC78 derivative	9/R	13/R	14/R	>32/R	70	P3	full length	27.0	—	—
EC122	ST567	EC78 derivative	12/R	15/R	20/S	≤1/S	36	P3	full length	28.1	—	—
EC101	ST131	urine	21/S	21/S	24/S	2/S	1.2	P3	full length	24.3 ± 0.5	—	—
MG1655	—	E. coli K-12	—	—	—	—	NA	NA	full length	27.7 ± 0.1	—	—
ATCC 25922	—	—	27/S	24/S	26/S	4/S	NA	NA	—	—	—	—

MEC, mecillinam; AMC, amoxicillin/clavulanate; TZP, piperacillin/tazobactam; NA, not applicable. S, susceptible; I, intermediate; R, resistant (as defined in the EUCAST clinical breakpoint tables version 8.1).

a

Discrepant interpretation of susceptibility to piperacillin/tazobactam when tested by disc diffusion and broth microdilution.

b

Piperacillin/tazobactam MIC determination was done using broth microdilution using EURGNCOL microtitre plates from Sensititre^TM (Thermo Fisher Scientific); piperacillin/tazobactam MIC breakpoints: susceptible ≤8 mg/L and resistant >16 mg/L.

c

bla_TEM-1C.

d

bla_TEM-1A (single copy) and bla_TEM-35 (five copies).

e

P3 WT promoter; Pa/Pb, P4 and P5 promoters are associated with increased expression compared with P3; and IS3/P3 is a hybrid promoter with the −35 box located within an IS3 element.

f

Porin genes investigated: ompC, ompF and phoE.

g

Measured in the absence of antibiotics. When SD is shown, data are the average from five independent experiments; if SD is not indicated, data are the average from a single experiment containing eight replicates.

h

bla_TEM-1 mRNA expression relative to EC101; data are the average from five independent experiments.

i

β-Lactamase activity relative to EC101; data are the average from five independent experiments.

Antimicrobial susceptibility testing

Susceptibility to amoxicillin/clavulanate, cefoxitin, cefpodoxime, mecillinam and piperacillin/tazobactam was determined using disc diffusion and EUCAST methodology. Further, piperacillin/tazobactam MIC determination was done using broth microdilution using EURGNCOL microtitre plates from Sensititre^TM (Thermo Fisher Scientific, East Grinstead, UK). Results were interpreted using the EUCAST clinical breakpoint tables version 8.1.

WGS

DNA was extracted from overnight cultures using the QIAGEN DNeasy Tissue and Blood Kit (QIAGEN, Hilden, Germany). Libraries were made using the Nextera XT Library Preparation Kit (Illumina, San Diego, CA, USA) and 2×150 bp paired-end reads were produced on an Illumina MiSeq instrument. Reads were assembled using SPAdes. Additionally, high molecular weight DNA was extracted from overnight cultures of EC78 and EC120 with the Nanobind CBB Big DNA Kit (Circulomics, Baltimore, MD, USA) and sequenced on an Oxford Nanopore Technology (ONT) MinION with the Rapid Barcoding Kit and an R9.4.1 flow cell (Oxford Nanopore Technology, Oxford, UK). Hybrid assembly using Illumina and ONT data was done using Unicycler.¹⁴ The copy number of bla_TEM-1 was estimated by comparing the read coverage of bla_TEM-1 with the average read coverage of the seven MLST genes. Sequence data from this study have been deposited at NCBI (BioProject accession PRJNA550338).

Determination of bla_TEM-1 mRNA

Total RNA was extracted from cultures in exponential growth phase using the Quick RNA™ Miniprep Kit (Zymo Research, Irvine, CA, USA), treated with DNase I (Thermo Fisher Scientific) and subjected to quantitative RT–PCRs targeting bla_TEM-1 and two housekeeping genes, mdh and gapA. Primers and probes are listed in Table S1 (available as Supplementary data at JAC Online). Expression of bla_TEM-1 was quantified using the ΔΔCt method, comparing the Ct value obtained for the bla_TEM-1 quantitative PCR with the average Ct value obtained in the mdh and gapA quantitative PCRs. For each isolate the ΔCt was normalized by subtracting the ΔCt of EC101.

Assay for β-lactamase activity

Cells from overnight cultures were resuspended in PBS, lysed by sonication and kept on ice for 5 min. Cellular debris was removed by centrifugation and the supernatant was kept for β-lactamase activity determination in duplicate using the Amplite^TM Colorimetric β-Lactamase Activity Assay Kit (AAT Bioquest, Sunnyvale, CA, USA) according to instructions of the manufacturer. β-Lactamase activity was normalized to total protein concentration measured using the Pierce Coomassie Plus Assay Kit (Bio-Rad, Hercules, CA, USA).

Determination of growth rate

For determination of growth rate an overnight culture was diluted 200× and cultured until late exponential phase. Cultures were then diluted 200-fold and growth measurements were done in eight replicates in a microtitre plate incubated at 37°C with agitation in a BioTek Synergy^TM HT Multi-Detection Microplate Reader (BioTek Instruments, Winooski, VT, USA) with OD₆₀₀ readings every 5 min. Doubling times were calculated from the exponential phase of the growth curve and averaged across replicates.

Passage of EC78

Two replicate cultures were passaged without antibiotic selection three times weekly by transferring 50 μL into 5 mL of fresh LB broth approximating 6.6 generations per passage. Cultures were maintained for 150 generations then isolates from each of the cultures were sequenced to establish bla_TEM-1 copy number.

Results

In an effort to characterize mechanisms underlying resistance to piperacillin/tazobactam, E. coli isolates EC49, EC66, EC69, EC78 and EC114, resistant to P/BLI combinations, but susceptible to cefoxitin and cephalosporins, were subjected to WGS (Table 1). In all isolates bla_TEM-1 was the only acquired β-lactamase identified. E. coli porin genes ompC, ompF and phoE were full length in all isolates except EC66, in which a truncated ompC gene was identified. In EC49, EC66, EC69 and EC114, bla_TEM-1 was preceded by strong Pa/Pb, IS3/P3 hybrid, P5 and P4 promoters, respectively (Table 1). In contrast, in EC78 bla_TEM-1B was associated with a P3 promoter. In this strain the copy number of bla_TEM-1B was 182.6 relative to the MLST gene copy number indicating that the observed resistance phenotype could be explained by bla_TEM-1B gene dosage. From Illumina sequencing data it was inferred that bla_TEM-1B was located on a 1905 bp long contig bracketed by IS26 elements. IS26 elements were only observed at one end each of two 130515 and 174779 bp large contigs having 80% and 88% sequence coverage in the chromosome of E. coli strain K-12 MG1655 (NC_000913). ONT sequencing confirmed bla_TEM-1B to be present within the chromosome of EC78. Individual ONT reads that contained up to 114 IS26-bla_TEM-1B repeats in head-to-tail configuration were identified. Hybrid assembly using Illumina and ONT data showed bla_TEM-1B to be present at a high copy number in a partially resolved region of the chromosome. The remaining resolved chromosome consisted of 4938917 bp. Orientation-specific PCRs directed against head-to-head, tail-to-tail and head-to-tail configurations of duplications only yielded amplification with the head-to-tail-specific primer set (Figure 1 and Figure S1).

Figure 1.

Genomic organization of bla_TEM amplifications in EC78 and EC120. In EC78 (top panel), the amplified 1933 bp region contained an IS26 element and a bla_TEM-1B controlled by a relatively weak P3 promotor and was situated in a prophage within the chromosome. The amplification was bracketed by IS26 elements with target site duplications (TSDs). EC120 (bottom panel) contained a 2563 bp region consisting of a 5′ 1743 bp fragment homologous to a part of Tn3 (nt 873 to 2615 of HM769901) and a 3′ 820 bp IS element. The 5′ 1743 bp fragment contained an apparent mobile element protein and a bla_TEM-35 encoding an inhibitor-resistant narrow-spectrum β-lactamase controlled by a relatively strong P4 promoter. The 3′ 2220 bp of the region was amplified and present in four additional copies. In the amplified copies the truncated ORF of the mobile element protein contained an in-frame initiation codon and encoded a putative transposase. The truncated reading frame had nine non-coding SNPs compared with the upstream longer reading frame.

To investigate whether bla_TEM-1B gene amplification resulted in hyperproduction of bla_TEM-1, we determined bla_TEM-1 mRNA expression and β-lactamase activity relative to the control strain EC101, which contains a single copy of bla_TEM-1 with a P3 promoter (Table 1). In EC78, transcription of bla_TEM-1 and β-lactamase activity was approximately 13-fold higher and 32-fold higher, respectively, than in EC101 (Table 1). The increased expression of bla_TEM-1 in EC78 was comparable to the clinical E. coli isolates EC49, EC66, EC69 and EC114, resistant to P/BLI combinations, containing bla_TEM-1 at lower copy numbers, but under the control of promoter variants associated with strong gene expression (Table 1).

The growth rate was measured for EC78 in the absence of antibiotics and compared with the growth rates of the clinical E. coli isolates EC49, EC66, EC69 and EC114 as well as EC101. Compared with EC101, EC78 was the only clinical isolate for which the doubling time was increased (difference = 8.6 min; 95% CI = 6.2–11.0; P = 0.0001). To assess the stability of the gene amplification without selection, EC78 was passaged in two replicate cultures for 150 generations generating EC121 and EC122. The MIC of piperacillin/tazobactam was unchanged at >32 mg/L for EC121, but was reduced to ≤1 mg/L for EC122. Both isolates were Illumina sequenced and analysis of coverage indicated a reduction in copy number of bla_TEM-1B to 70 and 36 for EC121 and EC122, respectively. The decreased bla_TEM-1B copy number was accompanied by a decrease in doubling time (Table 1).

To investigate the prevalence of bla_TEM amplifications in piperacillin/tazobactam resistance, we screened 36 E. coli isolates resistant to P/BLI combinations and susceptible to cephalosporins obtained from urine cultures using PCR specific for head-to-tail duplication of bla_TEM (Figure S1). The screen identified a single positive isolate, EC120. ONT and Illumina sequencing of this isolate disclosed that EC120 contained a single copy of bla_TEM-1B with a P3 promoter at one chromosomal site and at another chromosomal site a gene array of five head-to-tail replicates of a 2220 bp element containing bla_TEM-35 (an inhibitor-resistant β-lactamase)¹⁵ controlled by a P4 promoter and an IS26 element providing an explanation of the observed resistance pattern (Figure 1).

Discussion

bla_TEM-1 encodes a narrow-spectrum β-lactamase usually not associated with resistance to piperacillin/tazobactam. Here we demonstrate that both promoter variants and increased gene dosage, which will go undetected by simple algorithms for predicting susceptibility patterns from the presence or absence of resistance genes in sequence data, cause resistance to P/BLI combinations. However, as seen in this manuscript (Table 1; EC66 and EC114) and as observed by others,¹⁶ this type of isolate may yield discrepant results when tested by broth microdilution and disc diffusion. In version 9.0 of their breakpoint tables, EUCAST has adopted an area of technical uncertainty when testing P/BLI combinations reflecting a suboptimal correlation between inhibition zone obtained by disc diffusion and MIC. MIC is considered a key pharmacodynamic parameter for estimating the likelihood of treatment success; however, animal infection models studying piperacillin/tazobactam-resistant, but cephalosporin-susceptible, E. coli have shown relative success of piperacillin/tazobactam treatment.¹⁷ It may thus be questioned which susceptibility testing methodology best reflects the likelihood of piperacillin/tazobactam treatment success.

Gene amplification may arise as a consequence of recombination between separate homologous sites in sister chromosomes and may occur at rates approximating 10⁻⁵ per generation.¹⁸ Schechter et al.¹⁹ reported an isolate of E. coli that adapted to subinhibitory concentrations of piperacillin/tazobactam by amplification of a bla_TEM-1-containing 10 kb resistance module located on a plasmid. Repeated IS26 elements of composite transposons provide long homologous regions facilitating recombination, but IS26 elements may also translocate to adjacent sequences, preferentially to homologous IS elements, creating a gene array independent of RecA.²⁰ Adaptive amplifications of resistance genes can cause heteroresistance and most amplifications are unstable.¹³ Here we report an isolate with extensive IS26-associated amplification of bla_TEM-1 causing in vitro resistance to P/BLI combinations. The gene amplification was associated with a significant fitness loss compared with isolates hyperproducing bla_TEM-1 because of increased promoter strength. Despite this fitness cost, the array was maintained at a relatively high copy number for 150 generations. We screened a limited number of clinical isolates with a similar susceptibility pattern and identified an additional clinical isolate containing an IS26-associated amplification of a bla_TEM gene. Amplifications encompassing large regions may not have been detected by the PCR strategy used. This shows that gene amplifications, although here associated with a fitness cost, are not uncommon.

Acknowledgements

We thank Chih Man German Ma for his assistance with graphical artwork.

Funding

This study received funds from Fonden til Lægevidenskabens Fremme (grant 17-L-0495) and from the Scandinavian Society of Antimicrobial Chemotherapy Foundation (grant SLS-788501).

Transparency declarations

None to declare.

Author notes

Katrine Hartung Hansen and Minna Rud Andreasen authors contributed equally to this work.