Nanopore sequencing elucidates in vivo development of meropenem resistance by insertion of a mobile genetic element in the porin gene ompC in E. coli

Abstract

An ESBL-producing E. coli isolate recovered from a patient undergoing long-term treatment developed resistance to meropenem without acquiring carbapenem-hydrolysing enzymes. We performed Nanopore and Illumina sequencing and subsequent full hybrid genome assembly of this isolate and the meropenem-susceptible isolate recovered almost 8 weeks prior. Whole genome MLST patterns did not differ between isolates. However, we found the insertion of an IS5-like element in the sequence of the ompC gene and an increase in the number of copies of the CTX-M-15 gene in the resistant isolate. These results show that E. coli can develop meropenem resistance under antibiotic pressure by mutations in ompC genes and increasing the copy number of ESBL genes, and the value of next generation sequencing to reveal resistance mechanisms not detected by conventional PCR.

Case presentation

A patient with a history of unresectable perihilar cholangiocarcinoma was transferred to our tertiary hospital with cholangitis and liver abscesses from a secondary hospital. An ESBL-producing Escherichia coli isolate recovered from blood cultures almost 8 weeks prior to the transfer (isolate A) showed no meropenem resistance (MIC 0.047 mg l⁻¹) (Table 1). Under long-term treatment with meropenem and vancomycin, this patient developed fever and shaking chills. Blood cultures performed at our hospital showed growth of an extended-spectrum β-lactamase (ESBL)-producing E. coli (isolate B) that was meropenem resistant with a minimum inhibitory concentration (MIC) of 16 mg l⁻¹ as determined by gradient test (Liofilchem), whereas the strain retained imipenem-susceptibility (MIC 1.5 mg l⁻¹). No carbapenemase activity was detected by carbapenemase inactivation method [1] and no IMP, KPC, NDM, OXA-48 or VIM carbapenemase genes by qPCR. We stopped meropenem and started treatment with cotrimoxazole.

Table 1.

Susceptibility profiles of the E. coli isolates

	Isolate A		Isolate B
Date of isolation	24-4-2022		11-6-2022
Antibiotic	Classification	MIC (mg l−1)	Classification	MIC (mg l−1)
Amoxicillin	R	>16	R	>16
Amoxicillin/clavulanic acid	R	>16	R	>16
Piperacillin/tazobactam	S	8	R	>64
Cefuroxime	R	>32	R	>32
Cefotaxime	R	>32	R	>32
Ceftazidime	R	32	R	>32
Ceftazidime/avibactam		nd	S	1*
Cefoxitin	S	≤4	R	32
Ertapenem		nd	R	>32*
Imipenem	S	≤0.25	S	1.5*
Meropenem	S	0.047*	R	16*
Cotrimoxazole	S	≤20	S	≤20
Trimethoprim	S	≤0.5	S	≤0.5
Ciprofloxacin	R	>2	R	>2
Colistin	S	≤0.5	S	≤0.5
Gentamicin	R	>8	R	>8
Tobramycin	R	>8	R	>8
Fosfomycin	S	≤16	S	≤16
Nitrofurantoin	S	≤16	S	≤16

Antibiotic MICs were measured with the automated susceptibility testing system Vitek 2 (bioMérieux) except those with *, by gradient strip (Liofilchem). EUCAST: Clinical breakpoints and dosing of antibiotics v 13.1 was used Interpretation of MICs for classification of susceptibility. Grey background indicates where a difference in susceptibility between both isolates was observed.

MIC, minimum inhibitory concentration; nd, not detemined R, resistant; S, susceptible.

Both isolates were subjected to Illumina whole genome sequencing, multi locus sequence typing (MLST) and whole genome MLST (wgMLST) using Bionumerics 8 (Applied Maths, Belgium) and AMR genes detection using ResFinder 4.0 [2]. Both isolates belonged to ST131 and had the same wgMLST profile (<10 alleles difference). Bioinformatics analyses of the whole genomes did not reveal in isolate B any additional resistance genes or other (point) mutations known to confer carbapenem resistance such as polymorphisms in genes gyrA, parC or others.

Mutations, deletions and insertions in the gene sequences of OmpC and OmpF porine proteins can increase carbapenem MIC’s by decreasing the drug influx in the bacterial cell [3]. To analyse this mechanism in our isolates, we performed genome assembly using SKESA [4] after removing low quality reads with Trimmomatic v0.38 [5]. A full sequence of the ompC gene could be retrieved from the genome assembly of isolate A, but the gene sequence of isolate B was located on two different contigs. Due to the nature of short reads genome assembly, this difference could be due to a bioinformatics artefact or a biological reason such as a recombination or insertion. We found no other difference in the sequence of the ompC gene.

Additionally, we sequenced both isolates with Oxford Nanopore technology using R9.4 Flow Cell on a MinIon following manufacturer’s instructions. We performed trimming of Nanopore reads for quality using NanoFilt [6] and hybrid assembly using both Illumina and Nanopore reads using UniCycler [7].

Complete assembly of a circular chromosome and several plasmids was achieved for both isolates. Isolate A has a 5 111 702 bp chromosome and four plasmids (112 kbp; 53 kbp; 4 kbp and 3 kbp). Isolate B has a 5 129 626 bp chromosome and four similar plasmids with some insertions (112 kbp; 51 kbp; 8 kbp; 3 kbp). ResFinder revealed that isolate B possessed three copies of the CTX-M-15 gene compared to one for isolate A. One additional copy was located on the chromosome and the second one on the 8 kbp plasmid. When aligning the ompC genes from isolates A and B with that from another E. coli as reference (accession number NC_002695), the alignments revealed a 1200 bp insertion starting at nucleotide position 1026 (relative to the start of sequence in the sequence NC_002695) of the ompC gene of isolate B (Fig. 1).

Fig. 1.

Alignment of the ompC genes from isolates A and B shows a 1200 bp insertion. Sequence NC 002695 is used as a reference for the position of the alignment. Grey sections of the alignment indicate 100 % similarity between the sequences; black sections indicate discrepancies.

We extracted the insertion sequence (IS), translated it into amino acids and compared to the NCBI non-redundant protein sequence database using Blast-P. It revealed a high identity between our sequence and an IS5-like element of the ISEc68 transposase family. IS elements including ISEc68 have previously been described as disrupting the sequence of ompK36 gene, the K. pneumoniae homolog of ompC, and thereby elevating MICs of β-lactam antibiotics [8, 9]. In vitro studies showed porin-deficient subpopulations emerged in ESBL-producing E. coli during exposure to ertapenem at concentrations simulating human pharmacokinetics [10]. Additionally, IS26-mediated mechanisms underlying bla _OXA-1 and bla _CTX-M-1 group β-lactamase gene amplification with concurrent outer membrane porin disruption was found to lead to carbapenem resistance [11]. To our knowledge however, this is the first publication describing the in vivo evolution of this mechanism in E. coli isolates. Although we have not verified the level of expression, it has been shown previously that disruptions in the sequence of the ompC gene such as by insertion of IS elements leads to a significantly reduced expression of the OmpC protein [12]. These sequence disruptions also lead to a loss of fitness compared to the wild-type strains, but both the loss of fitness and the antimicrobial resistance traits can be reversed by complementing the mutants strains with ompC-carrying plasmids [13]. Additional experiments are needed to determine if the clinical isolates from the present study behave in a similar way.

Antimicrobial resistance in potentially pathogenic bacteria such as E. coli ST131 is a serious threat to hospitalized patients. In the present case, the carbapenemase resistance mechanism is probably porine deficiency combined with amplification of CTX-M-15 ESBL gene and arose in less than 8 weeks while under antibiotic treatment. No resistance determinant was detected by our qPCR method. We show that Nanopore sequencing is a valuable addition to the clinical microbiology laboratory to investigate the mechanism of unusual antibiotic resistance patterns in E. coli.