Characterization of Cefoxitin-Resistant Escherichia coli Isolates from Recreational Beaches and Private Drinking Water in Canada between 2004 and 2006

Abstract

A total of 142 cefoxitin-resistant Escherichia coli isolates from water sources were collected across Canada. Multidrug resistance was observed in 65/142 (45.8%) isolates. The bla_CMY-2 gene was identified in 110/142 (77.5%) isolates. Sequencing of the chromosomal ampC promoter region showed mutations from the wild type, previously shown to hyperproduce AmpC. CMY-2-producing plasmids predominantly belonged to replicon groups I1-Iγ, A/C, and K/B. The majority of the E. coli isolates belonged to the nonvirulent phylogenetic groups A and B1.

The predominant mechanism of resistance to β-lactams in gram-negative bacteria is the synthesis of β-lactamases (2, 28, 31). AmpC β-lactamases belong to Ambler class C; they can be either chromosomal or plasmid-mediated and have activity against most penicillins, cephalosporins, and cephamycins (31, 37). Wild-type Escherichia coli strains carry a chromosomal ampC gene which constitutively produces AmpC at low levels (20). The E. coli ampC gene is not induced by β-lactams, as in other bacterial species in the family Enterobacteriaceae, due to the absence of the ampR gene (17). It has been well documented that the acquisition of AmpC genes is associated with the acquisition of additional resistance genes to non-β-lactam antimicrobials (1, 5, 28, 30, 41). The presence of additional resistance determinants increases the spectrum of resistance displayed by these organisms and is causing concern over the appropriate treatment for AmpC producers in humans and animals. In this study, we characterized a sample of cefoxitin-resistant (FOX-R) E. coli isolates from water collected across Canada. Using molecular analysis, we were able to determine the mechanism of FOX-R and the relatedness of plasmids bearing acquired AmpC-type β-lactamase genes.

FOX-R E. coli isolates were collected through the Antimicrobial Resistant Organisms (ARO) Water Study, which utilized existing microbial water-testing programs in Alberta, Ontario, and Quebec, Canada, over 2.5 years beginning 31 May 2004. Sampling was spread proportionately over the study period and across the participating laboratories for each province to reflect previous testing volumes throughout the year. A total of 15,119 samples were tested; 3,195 came from Alberta (70.5% of beach origin), 10,051 from Ontario (78.1% of drinking water origin), and 1,873 from Quebec (100% of beach origin). The detection of E. coli isolates in private drinking water (n = 8,793) and recreational water (n = 6,326) samples was performed using standard methodologies (21) and confirmed by API 20E (bioMérieux, Canada). A total of 142 E. coli strains (76/6,326 from recreational water sources and 66/8,793 drinking water sources) with MICs of ≥16 mg/liter for cefoxitin (FOX-R) were identified and characterized over the 2.5-year study. Antimicrobial susceptibilities were determined on 96-well susceptibility panels (catalogue number CMVIAGNF; Trek Diagnostics, OH) using an automated microdilution broth method (Sensititre; Trek Diagnostics, OH). The Clinical and Laboratory Standards Institute (CLSI) provided breakpoints used to determine resistance for the antimicrobials tested (11) with the exception of ceftiofur (≥8 mg/liter) and streptomycin (≥64 mg/liter) (16), for which CLSI does not provide breakpoints. All isolates were screened for ampC genes as previously described (25). The amplification of the ampC promoter region was performed as previously described (8). Amplicons were sequenced and compared to the corresponding E. coli K-12 sequence (GenBank database accession number U00096). Incompatibility grouping (6) and phylogenetic analysis (10) were conducted using PCR-based methods. E. coli was characterized by pulsed-field gel electrophoresis by the standardized protocol for E. coli (O157:H7) (9, 19). Plasmidic DNA was isolated using Qiagen plasmid kits (Qiagen, Inc., Mississauga, Canada), transformed to electrocompetent E. coli ElectroMax DH10B cells (Invitrogen, CA), and digested using BglII (Roche Diagnostics, Laval, Quebec, Canada) as previously described (26). Fisher's two-tailed test was used to evaluate statistical significance using GraphPad QuickCalcs (http://www.graphpad.com/quickcalcs/index.cfm).

Of the total 15,119 samples collected, FOX-R E. coli strains made up 1.02% of the isolates from Ontario (n = 103), 0.7% from Alberta (n = 21), and 1.0% from Quebec (n = 18).

The susceptibility data obtained from the E. coli FOX-R environmental isolates are summarized in Table 1. Multidrug resistance (MDR), which is defined as resistance to three or more different antibiotic classes, was observed in 65 of 142 (45.8%) isolates.

TABLE 1.

Antimicrobial resistance patterns of environmental FOX-R study isolates

Antimicrobial agent	% (no.) of resistant strains
Amikacin	0 (0)
Amoxicillin-clavulanic acid	97.2 (138)
Ampicillin	97.2 (138)
Cefoxitin	100 (142)
Ceftiofur	68.3 (97)
Ceftriaxone	2.1 (3)
Chloramphenicol	14.1 (20)
Ciprofloxacin	0.7 (1)
Gentamicin	4.2 (6)
Kanamycin	16.2 (23)
Nalidixic acid	4.2 (6)
Streptomycin	39.4 (56)
Sulfamethoxazole	35.2 (50)
Tetracycline	57 (81)
Trimethoprim-sulfamethoxazole	14.1 (20)

Studies involving Canadian human clinical isolates of E. coli have previously shown that the majority of the FOX-R isolates were due to ampC promoter mutations (15, 25). Contrary to these previous reports, we have found that FOX-R in E. coli was due to a higher proportion of plasmid-mediated AmpC enzymes (77.5%), all of which were 100% identical to bla_CMY-2 compared to the promoter mutations (21.1%). The remaining two isolates did not contain an acquired AmpC or promoter mutation, and it is possible that the FOX-R phenotype may be due to the loss of outer-membrane porins (23). The predominance of CMY-2 in this study is not surprising, as bla_CMY-2 has been described as encoding the most prevalent of the plasmid-mediated AmpC enzymes in Enterobacteriaceae (33) and the most widely distributed worldwide (28). Indeed, AmpCs, particularly those encoded by bla_CMY-2, have become more prevalent than ESBLs in Canadian clinical cases (3, 29).

The ampC promoter region was sequenced for all isolates that did not contain a mobile ampC gene (32/142 [22.5%]). A total of 30 of 32 (93.8%) isolates fell into one of five mutation types. The most common mutations identified in the promoter region of the ampC gene in this study occurred in the previously reported sites which created new −35 and −10 box sequences and increased ampC transcription (15, 20, 22, 25, 35, 36) (Table 2).

TABLE 2.

Mutations occurring in the ampC promoter region^a

Mutation typeb	No. of isolatesc	Mutation at positiond:
		−42	−18	−15	−1	+22	+26	+27	+31	+32	+58	+70
WT	2	C	G	G	C	C	T	A	C	G	C	C
1	23	T	A		T						T
2	4	T	A		T				A		T
3	1	T	A	A	T						T
4	1		A		T						T
5	1					T	G	T		A		T

^aNumbering as described by Jaurin et al. (20).

^bWT, wild-type promoter from E. coli K-12. The mutation type numbers were assigned arbitrarily.

^cNumber of isolates in each type representing a total of 32 isolates.

^dPositions −42, −18, and −15 are located in the alternate promoter; +22, +26, +27, +31, and + 32 are located in the attenuator region; and +70 is located in the coding region.

It has been previously reported that plasmids expressing AmpC are associated with resistance to non-β-lactams (1, 28, 31, 38). In this study, 22/110 (20%) of the CMY-2 plasmids conferred an MDR phenotype (Table 3).

TABLE 3.

CMY-2 plasmids (n = 110) isolated from FOX-R E. coli strains

CMY-2 plasmid replicon typea	No. of isolates	Antimicrobial resistance phenotypes on CMY-2 plasmidsc	Avg plasmid size (range) (kb)
A/C	11	AMC, AMP, FOX, TIO, CHL, SMX, TETd	102 (74-131)
	3	AMC, AMP, FOX, TIO, CHL, SMX, TET, STX
	2	AMC, AMP, FOX, TIO, SMX, TET
	1	AMC, AMP, FOX, TIO, KAN, SMX, TET
	1	AMC, AMP, FOX, TIO
	2	AMC, AMP, FOX, TIO, CHL, KAN, SMX, TET
F	3	AMC, AMP, FOX, TIO	90 (34-122)
	1	AMC, AMP, FOX, TIO, CHL, SMX, TET
I1-Iγ	51	AMC, AMP, FOX, TIO	98 (62-134)
	1	AMC, AMP, FOX, TIO, CHL, SMX, TET
K/B	13	AMC, AMP, FOX, TIO	83 (66-104)
	1	AMC, AMP, FOX, TIO, TET
UIb	14	AMC, AMP, FOX, TIO	83 (63-139)
FIB, I1-Iγ	3	AMC, AMP, FOX, TIO	97 (71-133)
	1	AMC, AMP, FOX, TIO, CRO, SMX, TET
K/B, I1-Iγ	2	AMC, AMP, FOX, TIO	91.5 (90-93)

^aAll replicon types screened included HI1, H12, I1-Iγ, X, L/M, N, FIA, FIB, W, Y, P, FIC, A/C, T, FIIA, F, and K/B as described by Carattoli et al. (6).

^bUI refers to unidentified replicon type.

^cThe antimicrobial drugs used in panels are as follows: amikacin (AMK), amoxicillin-clavulanic acid (AMC), ampicillin (AMP), cefoxitin (FOX), ceftiofur (TIO), ceftriaxone (CRO), chloramphenicol (CHL), ciprofloxacin (CIP), gentamicin (GEN), kanamycin (KAN), nalidixic acid (NAL), streptomycin (STR), sulfamethoxazole (SMX), tetracycline (TET), and trimethoprim-sulfamethoxazole (STX).

^dResistance phenotype previously reported (4, 18).

The E. coli water isolates contained at least 1 of 10 different replicon types (I1-Iγ, N, FIA, FIB, Y, P, FIC, A/C, F, and K/B), whereas the CMY-2 plasmids contained at least 1 of 5 replicon types (I1-Iγ, FIB, A/C, F, and K/B). The predominant replicon types observed in the E. coli environmental isolates were FIB (80.9%), I1-Iγ (67.3%), F (65.5%), A/C (24.5%), and K/B (15.5%), of which 95/110 (86.4%) contained multiple replicon types. The CMY-2 plasmids contained predominantly I1-Iγ (52.7%, n = 58), followed by A/C (18.2%, n = 20) and K/B (14.5%, n = 16) plasmids (see the supplemental material). Although previous reports have indicated the wide distribution of replicon type N plasmids in β-lactamase-producing gram-negative organisms (4, 7, 14, 18, 27, 32, 34), we have only reported two isolates containing this plasmid type, neither of which contained bla_CMY-2. The I1-Iγ and A/C plasmids have previously been associated with CMY-2 plasmids from both animals and humans (7, 18, 24). Of the CMY-2 plasmids encoding MDR phenotypes, 86.4% (19/22) were associated with the replicon type A/C. Common MDR resistance phenotypes in A/C plasmids harboring bla_CMY-2 have previously been reported for both human and animal isolates (Table 3) (4, 18, 24), suggesting a possible relationship between replicon type and MDR.

Previous reports have described indistinguishable restriction patterns from E. coli CMY-2 plasmids isolated from different hospitals (40) as well as Salmonella isolates from cattle (13). In this study, seven unique pRFLP clusters were observed, each of which corresponding to specific replicon types (Fig. 1). To our knowledge, this is the first time that indistinguishable K/B and F plasmids harboring bla_CMY-2 have been identified.

FIG. 1.

E. coli plasmid restriction fragment length polymorphisms were analyzed by restricting CMY-2 plasmids with BglII. Seven clusters were observed corresponding to prevalent replicon types. Group I corresponds with the largest I1-Iγ cluster, groups II and III with two smaller I1-Iγ clusters, group IV with A/C, group V with K/B, group VI with an unidentified replicon type, and group VII with replicon type F. DW, drinking water.

E. coli isolates could not be grouped based on DNA fingerprinting techniques by using pulsed-field gel electrophoresis due to their inherent diversity, supporting previous data (13, 39) and indicating that these isolates are not clonal (data not shown).

Using phylogenetic analysis, we observed that 113/142 (79.6%) of the environmental isolates and 82/110 (74.5%) of the isolates harboring CMY-2 plasmids belonged to avirulent groups A and B1, which were significantly more likely to be MDR than isolates belonging to virulent groups B2 and D (P = 0.0314). Isolates with ampC promoter mutations predominantly belonged to group A (53.3%) or B1 (43.3%), which is in agreement with previous data (12).

In this report, we have identified bla_CMY-2 as the only acquired AmpC-type resistance gene in FOX-R E. coli strains from Canadian water sources. We have also demonstrated the similarities of CMY-2-containing plasmids identified in E. coli isolates of water origin as well as human and animal E. coli clinical isolates. Additional studies focused on the dissemination of plasmids identified in bacterial isolates from domestic animals, water, and humans are warranted to determine the significance and route of zoonotic transmission with respect to the emerging MDR resistance problem.