Antimicrobial Resistance Testing of Verocytotoxin-Producing Escherichia coli and First Description of TEM-52 Extended-Spectrum β-Lactamase in Serogroup O26

Abstract

We have investigated the antimicrobial resistance of verocytotoxin-producing Escherichia coli (VTEC) strains isolated from humans, animals, food, and the environment in Belgium. Resistance was more frequent in non-O157 strains from humans than in O157 strains from humans or other sources, and among non-O157 VTEC strains, intimin-positive strains were more resistant than intimin-negative strains. We also report the first VTEC strain producing an IncI1 extended-spectrum β-lactamase encoded by plasmid-borne bla_TEM-52; this β-lactamase was previously associated with Salmonella enterica and E. coli isolates from different origins.

Verocytotoxin-producing Escherichia coli (VTEC) is associated with gastrointestinal illness, and especially strains belonging to serogroups O157, O26, O103, O111, and O145 may cause hemolytic-uremic syndrome (HUS) (4). The cardinal virulence trait involved in HUS development is the production of one or more verocytotoxin types encoded by genes located on temperate lambdoid bacteriophages (14). Although most postdiarrheal HUS cases are linked to VTEC infection, antibiotic therapy has not demonstrated beneficial effects after HUS development (11, 14, 17). The underlying mechanism is unknown, but bacterial lysis could increase the amount of verocytotoxin released into systemic circulation and/or induce verocytotoxin-containing bacteriophages.

Antimicrobials are widely used for disease prevention and growth promotion in cattle and other farm animals identified as important VTEC reservoirs [8; European Medicines Agency, Joint opinion on antimicrobial resistance (AMR) focused on zoonotic infections, 2009 (http://www.ema.europa.eu/pdfs/vet/sagam/44725909en.pdf)]. Consequently, resistance may be promoted in VTEC commensally present in the intestinal tracts of these animals. Recent reports indicate that antimicrobial resistance of VTEC is rising (15). Mora et al. reported that bovine VTEC O157:H7 strains were significantly more resistant to streptomycin, tetracycline, and sulfisoxazole than those from humans, whereas non-O157 VTEC strains isolated from humans and beef were more resistant than bovine non-O157 strains (8). Most non-O157 strains showing multidrug resistance belonged to HUS-associated serotypes. Although the overall frequency of β-lactamases in E. coli isolated from humans and farm animals is increasing (13, 16; European Antimicrobial Resistance Surveillance System, Susceptibility results for E. coli in Belgium, 2010 [http://www.rivm.nl/earss/database/]), only a few extended-spectrum β-lactamase (ESBL)-producing VTEC strains have been reported (3, 6, 10).

We evaluated the antimicrobial resistance of VTEC strains isolated in Belgium from different origins in relation to established virulence factors and report the first TEM-52-producing VTEC isolate.

(Part of this research has been presented at the Pathogenic Escherichia coli Network Conference Control and Management of Pathogenic Escherichia coli, Dublin, Ireland, 17 and 18 September 2009.)

A total of 302 unduplicated, consecutive VTEC strains isolated in Belgium from humans, as well as from animals (n = 48), food (n = 21), and the environment (n = 1), referred to our laboratory between 2004 and 2009 by several Belgian laboratories were investigated. Among strains from humans, 153 belonged to serogroup O157 and 149 to non-O157 serogroups, whereas all strains of nonhuman origin (n = 70) belonged to serogroup O157. These strains were isolated from cattle (n = 47), a dog (n = 1), ground beef (n = 20), cheese (n = 1), and dust (n = 1). Four established virulence genes, the verocytotoxin 1 and 2 genes (vtx1 and vtx2), the intimin gene (eaeA), and the enterohemolysin gene (ehxA) were searched for by PCR (9). The flagellar type (fliC) was determined by PCR-restriction fragment length polymorphism (RFLP) (7). In vitro susceptibility tests were performed by the disk diffusion method for the antimicrobials listed in Table 1 using the EUCAST and CLSI potency Neo-Sensitabs tablets (Rosco, Taastrup, Denmark), with interpretation of zones according to CLSI, as described by the manufacturer (Rosco Diagnostica A/S; Neo-Sensitabs user's guide, document 3.1.0, 2010 [http://www.rosco.dk/]). In addition, cefotaxime plus clavulanate and ceftazidime plus clavulanate were systematically tested, and an ESBL was considered to be present if the inhibition zone increased by ≥5 mm in comparison with that for the antibiotic alone.

TABLE 1.

Antimicrobial resistance and correlation with the presence or absence of the intimin gene (eaeA) in VTEC isolated from humans and other sources in Belgium

Antimicrobial(s)a	No. (%) of resistant isolates from:
	Humans					Animals and food (all O157; n = 70)
	All (n = 302)	O157 (n = 153)	Non-O157
			All (n = 149)	eaeA positiveb (n = 88)	eaeA negativeb (n = 61)
Ampicillin*&	43 (14.2)	8 (5.2)	35 (23.5)	23 (26.1)	12 (19.7)	5 (7.1)
Piperacillin-tazobactam*	1 (0.3)	0 (0.0)	1 (0.7)	1 (1.1)	0 (0.0)	0 (0.0)
Amoxicillin + clavulanic acid*&	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Cefazolin*&	1 (0.3)	0 (0.0)	1 (0.7)	1 (1.1)	0 (0.0)	0 (0.0)
Cefuroxime*	1 (0.3)	0 (0.0)	1 (0.7)	1 (1.1)	0 (0.0)	0 (0.0)
Cefotaxime*	1 (0.3)	0 (0.0)	1 (0.7)	1 (1.1)	0 (0.0)	0 (0.0)
Ceftriaxone*	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Ceftazidime*	1 (0.3)	0 (0.0)	1 (0.7)	1 (1.1)	0 (0.0)	0 (0.0)
Cefepime*	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Aztreonam*	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Meropenem*	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Nalidixic acid*	16 (5.3)	0 (0.0)	16 (10.7)	12 (13.6)	4 (6.6)	0 (0.0)
Ciprofloxacin*	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Streptomycin&	84 (27.8)	26 (17.0)	58 (38.9)	40 (45.5)	18 (29.5)	12 (17.1)
Gentamicin*&	2 (0.7)	0 (0.0)	2 (1.3)	2 (2.3)	0 (0.0)	0 (0.0)
Kanamycin&	25 (8.3)	5 (3.3)	20 (13.4)	17 (19.3)	3 (4.9)	2 (2.9)
Amikacin*	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Tetracycline*&	59 (19.5)	15 (9.8)	44 (29.5)	31 (35.2)	13 (21.3)	4 (5.7)
Chloramphenicol&	11 (3.6)	2 (1.3)	9 (6.0)	5 (5.7)	4 (6.6)	0 (0.0)
Sulfonamide*&	81 (26.8)	22 (14.4)	59 (39.6)	40 (45.5)	19 (31.1)	11 (15.7)
Trimethoprim*&	28 (9.3)	4 (2.6)	24 (16.1)	16 (18.2)	8 (13.1)	3 (4.3)

^a*, antimicrobials used in human medicine; &, antimicrobials used in veterinary medicine.

^bBy PCR.

Results are shown in Table 1. One hundred two (102/302; 33.8%) human isolates showed resistance to at least one antibiotic. Combined resistance to streptomycin, sulfonamide, and tetracycline occurred in 15 (9.8%) of 153 O157 VTEC isolates; 22 of 26 (84.6%) streptomycin-resistant strains were also resistant to sulfonamide. In non-O157 strains, ampicillin-streptomycin-sulfonamide-tetracycline was the most frequently observed multidrug resistance profile (29/149; 19.5%). Most (58/59) sulfonamide-resistant non-O157 strains were also streptomycin resistant, and 34 of 35 ampicillin-resistant strains were streptomycin and sulfonamide resistant. One O26:H- isolate produced an ESBL.

Compared to human O157 VTEC isolates, non-O157 isolates were significantly more resistant to ampicillin (23.5% versus 5.2%; χ² = 20.6; P < 0.0001), nalidixic acid (10.7% versus 0.0%; χ² = 17.3; P < 0.0001), streptomycin (38.9% versus 17.0%; χ² = 18.1; P < 0.0001), kanamycin (13.4% versus 3.3%; χ² = 10.3; P = 0.001), tetracycline (29.5% versus 9.8%; χ² = 18.7; P < 0.0001), chloramphenicol (6.0% versus 1.3%; χ² = 4.8; P = 0.03), sulfonamide (39.6% versus 14.4%; χ² = 24.5; P < 0.0001), and trimethoprim (16.1% versus 2.6%; χ² = 16.3; P < 0.0001) (Table 1).

Non-O157 intimin-positive VTEC strains were more resistant than intimin-negative strains to streptomycin (45.5% versus 29.5%; χ² = 3.85; P = 0.05), kanamycin (19.3% versus 4.9%; χ² = 6.43; P = 0.01), and tetracycline (35.2% versus 21.3%; χ² = 5.38; P = 0.02) (Table 1). No significant differences in resistance pattern were observed when verocytotoxin 1 and 2 or enterohemolysin was taken into account (data not shown).

No significant difference in levels of resistance was observed among O157 VTEC strains from humans and other sources, or when only isolates from cattle were considered.

ESBL production was detected in one isolate from an afebrile 70-year-old man with nonbloody diarrhea and abdominal cramps. Fecal cultures were positive for VTEC O26:H- (fliC type H11) and Campylobacter jejuni. The O26:H- strain showed resistance to ampicillin, susceptibility to β-lactam combinations, and intermediate susceptibility to all tested cephalosporins except cefepime. It was confirmed as an ESBL producer using Etest ESBL cefotaxime/cefotaxime plus clavulanic acid (CT/CTL) and ceftazidime/ceftazidime plus clavulanic acid (TZ/TZL) strips (AB Biodisk, Solna, Sweden). PCR sequencing of the whole 850-bp coding sequence using primers TEMFL-F (5′-ATG AGT ATT CAA CAT TTY CGT G-3′) and TEMFL-R (5′-TTA CCA ATG CTT AAT CAG TGA GG-3′) revealed a sequence identical to that of bla_TEM-52 (2). This gene was borne on an IncI1 replicon type-containing plasmid as determined by PCR-based replicon typing (1). The presence of TEM-52 and its association with plasmids belonging to the IncI1 incompatibility group have been previously demonstrated for several Salmonella enterica serovars isolated from poultry and humans and recently in E. coli strains from healthy humans and broilers (2, 12, 13, 16).

Antimicrobial resistance among food-borne bacteria has been rising worldwide since the early 1990s, albeit to a lesser extent in VTEC (15). Our data show that both O157 and non-O157 strains are frequently resistant to ampicillin, streptomycin, sulfonamide, and tetracycline. Compared to O157 isolates, non-O157 VTEC strains were significantly more resistant to 8 of the 21 antimicrobials tested. This is in contradiction to earlier findings by Mora et al., who showed similar resistance levels among O157 and non-O157 VTEC strains (8). Moreover, we provide further evidence for an enhanced resistance to streptomycin, kanamycin, and tetracycline among non-O157 strains carrying intimin, an adhesin associated with more-severe disease (5, 8). Antibiotics do not beneficially influence clinical outcome and may even increase HUS risk. Resistance could still worsen the outcome by selecting VTEC in the guts of treated patients. To our knowledge, only three ESBL-producing VTEC isolates have been described in the literature, two belonging to serogroup O26 (CTX-M-3 and CTX-M-18) and one to O157 (CTX-M-2) (3, 6, 10). With the isolation of a TEM-52 VTEC O26:H- strain, three of the four reported ESBL-positive VTEC strains belong to O26, suggesting a higher propensity of this O serogroup to acquire ESBL genes.