Escherichia coli O80 hybrid pathotype strains producing Shiga toxin and ESBL: molecular characterization and potential therapeutic options

Abstract

Objectives

Enterohaemorrhagic Escherichia coli (EHEC) infections may be complicated by haemolytic uraemic syndrome (HUS). The emerging worldwide EHEC serogroup O80 has acquired a mosaic plasmid combining extraintestinal virulence and antibiotic resistance. This hybrid pathotype is associated with invasive infections that require antibiotic therapy, classically not recommended in EHEC infections, increasing the risk of HUS. We characterized two ESBL-producing O80 EHEC strains, which is an unusual resistance mechanism among EHECs, and determined the safest therapy to be used for invasive infections.

Methods

WGS of two strains isolated from the stools of an asymptomatic carrier and a patient with HUS was performed using Illumina and Nanopore technologies. Generated reads were combined to assemble genomes. We determined the safest therapy by comparing Shiga toxin (Stx) production by the two strains in the presence of several antibiotics.

Results

The strains were genetically close to the O80 EHEC clone, belonging to ST301 and harbouring stx2d, eae-ξ, ehxA and genes characteristic of the extraintestinal virulence plasmid pS88. Long-read sequencing identified the acquisition of an additional plasmid harbouring CTX-M-type genes (bla_CTX-M-14 and bla_CTX-M-1). Azithromycin decreased Stx production at subinhibitory concentrations, ciprofloxacin increased it and imipenem had no major effect. The combination of azithromycin and imipenem overall reduced Stx production.

Conclusions

Acquisition of an additional plasmid harbouring ESBL genes is a step towards increasing the risk of O80 EHEC dissemination and represents a serious public health concern. The combination of azithromycin and imipenem reduced Stx production and suggests that this combination could be tested in clinical trials.

Introduction

Enterohaemorrhagic Escherichia coli (EHEC) is a major foodborne pathogen that can cause haemolytic uraemic syndrome (HUS), the most common aetiology of acute renal failure in young children, due to the action of the phage-encoded Shiga toxin (Stx).¹ The adhesin intimin (eae) and the enterohaemolysin (ehxA) genes are both hallmarks of EHEC strains.

The historical serogroup O157 is the predominant EHEC serogroup worldwide but non-O157 serogroups are becoming increasingly associated with post-diarrhoeal HUS and the unusual serogroup O80 has recently emerged in Europe.^2–8

O80 is unique among EHEC serogroups as it is associated with multiple determinants of antibiotic resistance and can cause invasive infections,^7,8 whereas EHEC is usually a strictly intestinal pathogen. Indeed, O80 EHEC appears to be a hybrid pathotype that combines intestinal virulence factors (VFs) (stx, eae and ehxA genes) and genes characteristic of the plasmid pS88, a key determinant of extraintestinal E. coli virulence.² In cases of invasive infection, clinicians face a dilemma as rapid and effective antibiotic therapy is required but antibiotics may increase the risk of HUS by increasing Stx production.⁹

We previously showed that genes characteristic of pS88 and MDR genes are colocalized on the same plasmid (pR444_A) constituting a ‘mosaic’ plasmid shared by all members of this serogroup. In addition to this plasmid, the O80 clone harbours a plasmid carrying ehxA (pR444_C) and a cryptic plasmid of unknown function (pR444_B).² Plasmids pR444_A, pR444_B and pR444_C from the representative O80 strain RDEx444 have been fully sequenced and previously described.²

This study characterizes two EHEC isolates harbouring hybrid virulence and unusual resistance traits, providing a competitive advantage to inactivate critically important antibiotics and to establish and propagate clonally, increasing the chances of untreatable invasive infections.

Materials and methods

Bacterial strains

The two isolates were recovered in France in 2017 from stools of an HUS patient (PEY) and of an asymptomatic carrier (L113) collected as part of a paediatric research network approved by the Saint-Germain-en-Laye Hospital Ethics Committee (10/10/2010-CPP06063). Strain EDL933 (NC_002655.2),¹⁰ characteristic of serogroup O157, was used as a positive control for the comparison of Stx production.

Sequence analysis

Illumina (San Diego, CA, USA) sequencing was performed using the Nextera kit and the MiSeq reagent kit v3 (600 cycles). Nanopore (Oxford, UK) sequencing was performed using the Rapid Sequencing kit SQK-RAD and a MinION device with a R9.4 flowcell. Hybrid assembly of the Illumina and Nanopore sequences was performed using Unicycler software (v0.4.0).¹¹ Sequences have been deposited at DDBJ/ENA/GenBank (PRJNA530533). Scaffolds representing the different plasmids were aligned using MAUVE software and annotated by RAST software.

Strain typing, resistome investigation, VF detection and plasmid characterization were performed as previously described.²

Comparison of Stx production

Stx was quantified in filtered supernatants of bacterial cultures using a chemiluminescent immunoassay for EHEC toxins (Liaison, DiaSorin, Italy) as previously described.⁸ Stx production was assessed in the presence or absence of ciprofloxacin, azithromycin and imipenem, tested alone or in combination for the last two, at subinhibitory concentrations (½ and ¼ of the MIC).

Three independent assays were performed for each antimicrobial agent alone and two for the combination. Each measurement was performed twice. The results are also expressed as the percentage of production relative to the control without antibiotic for each series (relative secretion) to minimize the interseries variability.

Results

Sequence analysis

Hybrid assembly produced 127 contigs with an N50 of 280 177 bp for the PEY strain and 72 contigs with an N50 of 959 987 bp for the L113 strain.

These two strains appear to be very close to the O80 clone. Indeed, as for the representative strain RDEx444, they belong to ST301, harbour the fimH54 gene and the classical O80 EHEC resistance and virulence (stx2d, eae-ξ and ehxA) genes. Moreover, they carry the three plasmids of interest (p_A, p_B and p_C) sharing strong homology (99% nucleotide identity, 84%–98% coverage) with those of RDEx444 previously described² [Figure 1 and Table S1 (available as Supplementary data at JAC Online)].

Figure 1.

Plasmid comparative genomic analysis of the two ESBL-producing O80 EHEC strains isolated from an asymptomatic carrier (L113) or involved in HUS (PEY). The main genes are schematized by triangles coloured according to their function. Homology areas between two plasmid scaffold regions are symbolized by a coloured area according to the function of the main genes present in this area. Comparison of plasmid representative scaffolds of each strain, L113 and PEY, with different plasmids of RDEx444, the representative strain of the O80:H2 hybrid pathotype EHEC clone: the mosaic plasmid carrying extraintestinal VFs and antibiotic/heavy metal resistance genes (p_A) (a); the cryptic plasmid (p_B) (b); and the plasmid carrying the enterohaemolysin gene (ehxA) (p_C) (c). Comparative genomic analysis of L113’s ESBL-resistance plasmid (pL113_R) and the two closest relative plasmids publicly available (pC59-112, accession number KJ484637.1; DH5α p644-7, accession number MG692702.1) (d). Comparative genomic analysis of PEY’s ESBL-resistance plasmid (pPEY_R) and the two closest relative plasmids publicly available (pCT of strain C159/11, accession number FN868832.1; DH5α p310-1, accession number MG692685.1) (e). This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Furthermore, they have acquired an additional resistance plasmid (p_R) conferring greater resistance to β-lactams, due to ESBL production (bla_CTX-M-14 for PEY and bla_CTX-M-1 for L113), and closely related to conjugative resistance plasmids previously reported (Figure 1 and Table S1). pL113_R displayed 95% query coverage and 99% identity to IncI1 ST3 E. coli conjugative plasmids previously isolated from chickens: pC59-112 [112 330 bp long (KJ484637.1)] and DH5α p644-7 [107 977 bp long (MG692702.1)].^12,13 A maximum of 39 SNPs were observed within the 96 217 bp common to these three plasmids. All carried bla_CTX-M-1, a copy of the colicin gene (cia) and other resistance genes. pPEY_R shared strong homology (99% identity) with IncB/O/K/Z conjugative plasmids carrying bla_CTX-M-14 identified in E. coli strains isolated from a calf [pCT strain C159-11, 93 629 bp long (FN868832.1), 97% query coverage]¹⁴ and a chicken [DH5α p310-1, 83 938 bp long (MG692685.1), 90% query coverage].¹³ A maximum of one SNP was observed within the 83 874 bp common to these three plasmids. As for the PEY strain, these plasmids contained no other resistance or virulence genes.

Comparison of Stx production

Basal Stx production was assessed in the absence of antibiotic (Figure 2a). The positive control (EDL933), produced large amounts of toxin, with an average quantity of 4500 ng/mL Stx. Strain L113, isolated from an asymptomatic carrier, produced an average of only 18 ng/mL Stx and strain PEY, involved in HUS, produced an average of 670 ng/mL Stx.

Figure 2.

Expression of Stx by two ESBL-producing O80 EHEC strains (L113 and PEY) and the reference strain O157 EHEC (EDL933) in the absence or presence of different antibiotics. Production of Stx is either expressed in ng/mL, in the absence of antibiotics, as the basal production for each strain (a) or as relative to basal production (corresponding to 100%) in the presence of ciprofloxacin, imipenem, azithromycin or imipenem + azithromycin at subinhibitory concentrations for each strain [PEY (b), L113 (c) or EDL933 (d)]. Error bars indicate SDs. Significant differences (P < 0.05), using the Mann–Whitney test, between two conditions are represented by brackets. ATB, antibiotic; AZM, azithromycin; CIP, ciprofloxacin; IPM, imipenem.

We then measured Stx production in the presence of antibiotics (Figure 2b, c and d). We expressed toxin production as the relative quantity of Stx (%) compared with the concentration obtained in the absence of antibiotic (basal concentration equivalent to 100%) for each series. Ciprofloxacin, which could potentially be used for invasive infections, significantly induced Stx production, regardless of the strain and subinhibitory concentration. The magnitude of induction was particularly high for strain L113 (average overproduction of 65-fold), which produced a low quantity of Stx at the basal state. Azithromycin, first-line treatment for paediatric digestive infections, decreased Stx production for all conditions and strains studied. Imipenem, the drug of choice for severe infections caused by ESBL-producing bacteria, did not significantly modify Stx production for the two tested O80 strains. The combination of azithromycin and imipenem reduced Stx production overall, significantly for the PEY strain, regardless of the conditions tested, and at ¼ of the MIC for strain L113, but not for strain EDL933. The benefit of azithromycin addition on Stx production seemed to vary according to the condition, but azithromycin never increased the toxin production in the experiments carried out.

Discussion

The hybrid pathotype O80 EHEC has emerged in France over the last few years and has diffused within Europe.^2–8 Here, we characterized two O80 EHEC strains carrying ESBL genes (bla_CTX-M-1 and bla_CTX-M-14). Apart from clone O104:H4, responsible for the 2011 German outbreak and carrying the bla_CTX-M-15 gene,¹⁵ EHECs are generally antibiotic susceptible. A Belgian study recently described an O80 EHEC isolate carrying bla_CTX-M-1 but, in contrast to ours, this strain did not possess genes characteristic of the extraintestinal virulence plasmid pS88.⁴

In spite of their diverse origins (animal and human), a high level of similarity and a common set of backbone modules between ESBL-carrier plasmids (p_R) were observed for each strain. In all cases, ESBL genes were flanked by ISEcp1, as previously described,^12–14 and inserted in the conjugal transfer tra locus. This low diversity suggests vertical transmission and evokes high fitness and adaptability of these resistance plasmids.

It is worth noting that IncI1 group plasmids (pL113_R closest relative plasmids) also carry genes conferring resistance to tetracyclines and trimethoprim/sulphonamides, antimicrobials massively used in veterinary use, thus conferring a major selective advantage.¹⁶ On the contrary, bla_CTX-M-14 is the only known gene conferring selective advantage identified for IncB/O/K/Z group plasmids (pPEY_R closest relative plasmids). A hypothesis for their persistence and spread, in the absence of β-lactam pressure, can be the presence of the pil locus encoding a type IV pilus that may be involved in bacterial adhesion to surfaces and eukaryotic epithelial cells in vitro and in biofilm formation.¹⁴ pL113_R also harbours this additional conjugal transfer system that jointly with the presence of a bacteriocin gene (cia) may promote its maintenance in the digestive tract and explain isolation of this strain in an asymptomatic carrier.

These multiresistant strains, which carry the same plasmid virulome as the O80 clone, are perfectly equipped to induce invasive infections that may require antibiotic therapy. The basal Stx production of the strain responsible for HUS (PEY) was nearly 40 times higher than that of the carrier strain (L113). Producing a low amount of toxin at the basal state, L113 had a high capacity for toxin induction upon exposure to ciprofloxacin (average overproduction of 65-fold). These results suggest that the degree of virulence of EHECs may be associated, at least in part, with their level of Stx basal production. Fine-resolution genetic analysis of the Stx prophage and its chromosomic environment on more O80 EHEC isolates may help to explain this difference in basal production.

The effect of antibiotics on Stx production has mainly been studied for serogroup O157. Fluoroquinolones and trimethoprim/sulfamethoxazole significantly increase Stx2 production, contraindicating their use in EHEC infections.^9,17 Conversely, protein synthesis-inhibiting antibiotics, in particular azithromycin, decrease Stx production.^9,17 We obtained identical results for strains of serotype O80:H2. However, the use of macrolides to treat invasive infections caused by ESBL-producing bacteria is inappropriate and carbapenems may be required. Our results show that imipenem did not induce Stx production with O80 strains as previously described with EHEC strains of serogroups O157,^9,18 O26 and O111.¹⁸ Although imipenem did not induce Stx production, carbapenems markedly induce the production of outer membrane vesicles (OMVs) in EHEC strains,¹⁹ which may transport Stx.²⁰ Overproduction of OMVs, even without that of Stx, could negatively affect the clinical prognosis by inducing the production of IL-8 involved in the pathophysiology of HUS.¹⁹ In contrast, azithromycin decreases OMV production.¹⁹ Moreover, azithromycin is highly effective for the eradication of EHEC faecal carriage and thus may prevent relapses of HUS, which may occur with O80 EHEC infections.⁷ The imipenem/azithromycin combination is yet to be evaluated in this context. Pending knowledge of which combination therapy both limits Stx production and decreases OMV production, it seems prudent to add a protein synthesis inhibitor, such as azithromycin, in the treatment of invasive infections requiring systemic antibiotherapy.

The current spread of the emerging clone of O80 EHEC in various countries represents a serious public health concern. Acquisition of an additional plasmid harbouring ESBL genes is a step towards increasing the risk of its dissemination and may complicate the treatment of infections. The combination of azithromycin and imipenem, which remains to be evaluated in clinical trials, may represent a promising option for invasive infections with such hybrid pathotype strains by limiting Stx production.

Acknowledgements

We thank Mrs Marty (DiaSorin, Italy) for her technical assistance. We also thank Dr Thollot (Essey les Nancy, France) and Mrs Sobral (ACTIV) for the collection and identification of the L113 strain.

Funding

This study was supported by internal funding.

Transparency declarations

None to declare.