Azithromycin Resistance in Shiga Toxin-Producing Escherichia coli in France between 2004 and 2020 and Detection of mef(C)-mph(G) Genes

ABSTRACT

We described and characterized Shiga-toxin-producing Escherichia coli (STEC) strains with high levels of resistance to azithromycin isolated in France between 2004 and 2020. Nine of 1,715 (0.52%) STEC strains were resistant to azithromycin, with an increase since 2017. One isolate carried a plasmid-borne mef(C)-mph(G) gene combination, described here for the first time for E. coli. Azithromycin resistance, although rare, needs consideration, as this treatment may be useful in cases of STEC infection.

INTRODUCTION

Shiga-toxin-producing Escherichia coli (STEC) strains are major foodborne pathogens responsible for gastrointestinal diseases, ranging from diarrhea to hemolytic uremic syndrome (HUS) (1). Typical STEC isolates carry the intimin (eae gene) and the prophage-encoded Shiga-toxin (Stx). Although no clear risk factors for HUS development, except for young age, have been identified, higher levels of Stx production could be an important trigger of HUS occurrence (2). Unlike quinolones or beta-lactams, macrolides and especially azithromycin do not modify or increase Stx production in vitro (3, 4) and decrease bacterial shedding after HUS (5).

Azithromycin is one of the most prescribed antibiotics worldwide, among adults (6) and children (7), and could participate in the selection of resistant strains (8). Macrolide resistance is mainly observed in E. coli strains with phosphorylases [mph(A) to mph(O) genes (9, 10)], especially mph(A) (11). However, in STEC strains, resistance to azithromycin is poorly studied. Two studies address this subject; one, carried out in France, reported a prevalence of 0.3% (2/508 isolates in 2004 to 2014) (12), and another one, in England in 2017, reported a prevalence of 0.2% (1/430) (13).

Here, we aimed to describe and characterize STEC strains with high-level resistance to azithromycin, isolated in France between 2004 and 2020, and to assess alternative drugs in that case. To this end, we included all STEC strains collected in France between January 2004 and June 2020 by the French National Reference Center for E. coli (NRC). For azithromycin-resistant isolates, antimicrobial susceptibility testing was performed either by disk diffusion (rifaximin [40 μg; MAST] and spiramycin [100 μg; Bio-Rad]) or by using Etest (bioMérieux, France) (erythromycin, clindamycin, clarithromycin, fosfomycin, and rifampicin) (Table 1). We used the reference strain E. coli ATCC 25922 as a control. Fosfomycin MICs were interpreted using the European Committee on Antimicrobial Susceptibility Testing (EUCAST; https://mic.eucast.org/) breakpoints (14). The breakpoint of >16 mg/L has been proposed to define resistance to azithromycin in Salmonella and Shigella, and therefore, it is also often used for intestinal pathogenic E. coli (11). No resistance breakpoints are established for the other molecules (rifaximin, erythromycin, clindamycin, clarithromycin, spiramycin, and rifampicin) in E. coli.

TABLE 1.

Antibiotic susceptibility of azithromycin-resistant Shiga-toxin producing E. coli strains isolated between 2004 and 2020 in France

Isolate	MIC (mg/L)						Inhibition zone diam (mm) in disk diffusion
	Azithromycin	Erythromycin	Clindamycin	Clarithromycin	Fosfomycin	Rifampicin	Rifaximin	Spiramycin
34396	>256	>256	>256	>256	0.75	16	12	6
36493	>256	>256	>256	>256	0.75	16	16	8
42514	256	>256	>256	>256	0.38	16	13	9
43037	96	>256	>256	>256	1	16	16	9
45195	256	>256	>256	>256	0.5	16	13	6
45381	48	>256	>256	>256	1	16	16	6
45466	64	>256	>256	>256	12	16	17	8
47439	48	>256	>256	>256	0.75	12	15	10
47750	48	>256	>256	>256	0.5	12	17	6
ATCC 25922	4	32	>256	48	1,5	12	12	10

Azithromycin-resistant isolates were subjected to whole-genome sequencing (WGS) using Illumina, and sequences were analyzed as previously described (3, 15).

The genetic support of mef(C)-mph(G) genes was studied in E. coli isolate 45466 using long-read sequencing with Nanopore (Oxford, UK). It was performed with the rapid sequencing kit SQK-RAD and a MinION device with a R9.4 flow cell. We used Unicycler software (v0.4.0) to carry out hybrid assembly of the Illumina and Nanopore sequences (16). Sequences were analyzed with PlasmidFinder (17) and annotated with RAST software (18). Finally, we searched both for homologous plasmids from Enterobacterales and for mef(C)-mph(G) genes with 100% coverage and homology in all E. coli/Shigella sequences in the NCBI database.

Between 2004 and 2020, 1,715 STEC strains were isolated by the NRC. Azithromycin MIC₅₀ and MIC₉₀ were 4 mg/L and 6 mg/L, respectively (range, 0.5 to >256 mg/L; median, 4 mg/L). Nine isolates were resistant to azithromycin (MIC range, 48 to >256 mg/L; median, 64 mg/L), resulting in 0.52% resistance. Of note, only 2/807 (0.25%) azithromycin-resistant isolates were found before 2017 (in 2012 and in 2013 [12]), and the other 7 resistant isolates (7/908; 0.77%) were collected between January 2017 and June 2020 (Tables 1 and 2).

TABLE 2.

Whole-genome sequencing results of azithromycin-resistant Shiga toxin-producing E. coli strains isolated between 2004 and 2020 in France

Isolate	Collection yr	Patient age (yr)	Clinical diseasea	Quality sequencing data			Serotype	Sequence type	Virulence factor(s)	Macrolide resistance gene(s)
				N50 (bp)	Contig no.	Mean genome size (bp)
34396	2012	62	HUS	278,669	279	5,300,106	O17/O44:H18	69	Stx2d	mph(A), erm(B)
36493	2013	<1	HUS	94,290	408	5,562,733	O26:H11	29	Stx2d, Eaeβ1	mph(A)
42514	2017	28	BD	29,905	469	5,171,176	O117:H7	6880	Stx1a	mph(A)
43037	2017	5	HUS	48,855	301	5,125,621	O146:H8	8356	Stx1c	mph(A)
45195	2018	24	BD	39,047	332	5,095,835	ND:H7	5292	Stx1a	mph(A), erm(B)
45381	2019	29	BD	52,202	318	5,272,679	O146:H8	8356	Stx1c	mph(A), mph(B)
45466	2019	1	BD	64,220	472	5,942,917	O26:H11	21	Stx1a, Eaeβ1, EhxA	mph(G), mef(C)
47439	2020	1	HUS	79,390	382	5,411,518	O111:H8	16	Stx1a, Stx2a, Eaeγ2, EhxA	mph(A)
47750	2020	44	BD	70,538	333	5,642,051	O157:H7	11	Stx1a, Stx2c, Eaeγ1, EhxA	mph(B), erm(B)

^aHUS, hemolytic uremic syndrome; BD, bloody diarrhea.

For azithromycin-resistant isolates, MICs of rifampicin and fosfomycin ranged from 12 to 16 mg/L (median, 16 mg/L) and from 0.5 to 12 mg/L (median, 1.25 mg/L), respectively. MICs of other macrolides (erythromycin, clindamycin, and clarithromycin) were >256 mg/L, and spiramycin zone diameter inhibition was 6 to 10 mm (median, 8 mm). Zone diameter inhibition of rifaximin was between 12 and 17 mm (median, 16 mm). Of note, 45466 had the highest fosfomycin MIC (12 mg/L; however, this MIC remains susceptible according to EUCAST criteria) (Table 1).

Resistance to azithromycin was mostly mediated by mph(A) (n = 7), either alone (n = 4) or associated with mph(B) (n = 1) or erm(B) (n = 2). One isolate carried the association mph(B)/erm(B), and one carried the rare couple mef(C)/mph(G) (strain 45466) (Table 2).

For the latter, long-read sequencing allowed us to identify a 202,201-bp plasmid (p45466-R) harboring the mef(C)-mph(G) genes with IncHI1A, IncHI1B(R27), and IncFIA(HI1) incompatibility groups (17). These genes are localized in a cassette carrying several other resistance genes conferring resistance to penicillins (bla_TEM), sulfamethoxazole/trimethoprim (dhfr1 and sul2), or aminoglycosides [aph(3)-Ib and aph(6)-I]. The mef(C)-mph(G) genes are organized in tandem-pair arrangement only 5 nucleotides apart (Fig. 1 and Table 2) and are surrounded by IS6-like element IS26 family transposases and an IS91 family transposase.

FIG 1.

Annotation of the 202-kb plasmid p45466 from E. coli strain 45466, harboring the mef(C)-mph(G) gene pair involved in macrolide resistance.

After subjecting this plasmid to a BLAST search in the NCBI database, we found that it was close to several plasmids described in Enterobacterales, with the closest being p14ODMR, described in E. coli 14OD0056 with a 99.98% identity and a coverage of 94% (accession number MG904992.1). Interestingly, none of these plasmids carry the mef(C)-mph(G) genes.

Finally, we identified six isolates carrying the combination mef(C)-mph(G) with 100% identity and coverage that were E. coli/Shigella in the NCBI database. Of note, 2 of these also carried stx genes (Table 3).

TABLE 3.

Whole-genome sequence data (from public databases) of E.coli/Shigella isolates carrying mef(C)-mph(G) genes

Strain	Biosample ID	Country	Source	Origin	Serotype	Sequence type	Virulence factor(s)
IHIT32077	SAMN14279035	Spain		Environmental	O128:H26	2197
978891	SAMN15933666	United Kingdom	Human	Clinical	O166:H28	1819	Stx1c, EhxA
As Lw Down3-2	SAMN11124839	Germany	River sediment	Environmental	O93:H28	4038
B P Zu-1	SAMN11125139	Germany	Raw sewage	Environmental	O8:H19	201	Stx2e
S18-17	SAMN11125178	Germany	River sediment	Environmental	O149:H1	5748
Win2012_WWKa_NEU_19	SAMN06641869	Germany	Wastewater inflow	Environmental	O143:H4	117

While exploring antibiotic resistance among STEC strains in France, we observed low but increasing rates of high-level resistance to azithromycin. Isolates identified in our study were from various serotypes and sequence types, showing that the diffusion is not due to the emergence of a particular clone.

Interestingly, we identified an azithromycin-resistant E. coli strain (strain 45466) carrying the gene combination mef(C)-mph(G), encoding the Mph(G) phosphotransferase associated with the Mef(C) efflux pump. To our knowledge, this is the first description of an azithromycin-resistant E. coli strain carrying this combination. The mef(C)-mph(G) gene association was first described in Photobacterium damselae, an indigenous marine bacterium known as a zoonotic pathogen (19, 20). The combination of both genes is synergic and is associated with a high azithromycin MIC (19).

The mef(C)-mph(G) genes are located on a plasmid of 202 kb whose main structure was found in several other Enterobacterales. The presence of the mef(C)-mph(G) genes in the whole-genome sequences of six E. coli/Shigella in the NCBI database supports the idea that the diffusion of these genes is not an isolated event. However, as there are only genomic data, it is not possible to determine the azithromycin MICs for these isolates and thus be sure of the expression of these genes.

The triggering role of antibiotics in HUS is still debated (21). Some bactericidal antibiotics (co-trimoxazole and fluoroquinolones, for example) increase Stx production in vitro and therefore could increase the risk of progression to HUS (22, 23). Inversely, azithromycin is associated with a decrease in the STEC inoculum and in Stx release and therefore probably decreases the risk of developing HUS (3, 22). The use of azithromycin is currently being assessed in a clinical trial concerning patients with HUS (NCT02336516). Therapeutic alternatives to azithromycin were proposed in some countries, such as rifamycins (especially rifaximin) in the United States and salts of fosfomycin in Japan (24). Both are associated with a favorable clinical outcome when administered early after the first symptoms of bloody diarrhea (24–26). Although no critical inhibition zone diameters exist for these two antibiotics, all isolates had inhibition zone diameters or MICs similar to those of wild-type E. coli ATCC 25922, suggesting susceptibility, except for the 45466 isolates, which had a higher fosfomycin MIC than other isolates.

Together, these results show that resistance to azithromycin in STEC strains remains rare but tends to increase in France; hence the need to test sensitivity before therapeutic use. Here, we describe a new combination of genes, mef(C)-mph(G), in E. coli that was initially described in a marine bacterium. In vitro data suggest that rifaximin or fosfomycin could be interesting therapeutic alternatives in STEC infections with azithromycin-resistant E. coli.

Data availability.

Raw reads have been deposited in GenBank under BioProject ID PRJNA735027.