To investigate plasmid-mediated fosfomycin resistance related to fosA3 in Escherichia coli isolates collected from different animals in Hong Kong, China, 2008–2010.
To investigate plasmid-mediated fosfomycin resistance related to fosA3 in Escherichia coli isolates collected from different animals in Hong Kong, China, 2008–2010.
In total, 2106 faecal specimens from 210 cattle, 214 pigs, 460 chickens, 398 stray cats, 368 stray dogs and 456 wild rodents were cultured. The faecal colonization rates of fosfomycin-resistant E. coli were as follows: 11·2% in pigs, 8·6% in cattle, 7·3% in chickens, 2·4% in dogs, 0·8% in cats and 1·5% in rodents. The cultures yielded 1693 isolates of which 831 were extended-spectrum β-lactamases (ESBL) producers. Fosfomycin-resistant isolates were more likely than fosfomycin-susceptible isolates to be producers of ESBL and to have resistance to chloramphenicol, ciprofloxacin, cotrimoxazole, gentamicin and tetracycline. Of the 101 fosfomycin-resistant isolates, 97 (96·0%) isolates were fosA3 positive and 94 (93·1%) were blaCTX-M positive. PCR mapping showed that the fosA3-containing regions were flanked by IS26, both upstream and downstream in 81 (83·5%) isolates, and by an upstream blaCTX-M-14-containing transposon-like structure (ΔISEcp1-blaCTX-M-14-ΔIS903 or ISEcp1-IS10 -blaCTX-M-14-ΔIS903) and a downstream IS26 in 14 (14·4%) isolates. For the remaining two isolates, fosA3 was flanked by a downstream IS26 but the upstream part cannot be defined. In a random subset of 18 isolates, fosA3 was carried on transferable plasmids with sizes of 50–200 kb and the following replicons: F2:A-B- (n = 3), F16:A1:B- (n = 2), F24:A-B- (n = 1), N (n = 1), B/O (n = 1) and untypeable (n = 3).
This study demonstrates the emergence of fosA3-mediated fosfomycin resistance among multidrug-resistant E. coli isolates from various animals. IS26 transposon-like structures might be the main vehicles for dissemination of fosA3.
The spread of extended-spectrum β-lactamases (ESBLs) and carbapenemases in Enterobacteriaceae has renewed interest to revisit the clinical use of old antibiotics such as fosfomycin for treating infections (Falagas et al. 2010; Gutierrez et al. 2010; Soraci et al. 2011). In Escherichia coli, resistance to fosfomycin is uncommon and is mainly caused by chromosomal factors (Takahata et al. 2010). However, two novel plasmid-mediated fosfomycin enzymes, FosA3 and FosC2 have been found among CTX-M-producing E. coli in Japan (Wachino et al. 2010). Furthermore, a recent study involving strains from pets in China demonstrated that 29 of 33 fosfomycin-resistant E. coli were positive for the plasmid-mediated fosA3 gene and all of them were CTX-M producers (Hou et al. 2012). It is often held that exposure to antibiotics is a major factor in maintaining antimicrobial resistance. Nonetheless, high prevalence of certain antibiotic-resistant genes has been detected in bacteria originating from wild animals (Gilliver et al. 1999), indicating that resistance may not decline in absence of antibiotic treatment. Here, we investigate the prevalence of faecal carriage of fosfomycin-resistant E. coli in domestic and wild animals. The mechanism of fosfomycin resistance in the isolates was characterized.
During September 2008–August 2010, 2106 faecal specimens were collected from 210 beef cattle, 214 pigs, 460 broiler chickens, 398 stray cats, 368 stray dogs and 456 wild rodents. Of the 456 rodents, 182 were Rattus norvegicus (brown rat), 18 were R. rattus (black rat), 123 were R. andamanensis (Indochinese forest rat), and 92 were Niviventer fulvescens (chestnut spiny rat). The species for the remaining 41 rodents were not identified. All the specimens were collected from animals in Hong Kong in a central slaughterhouse (cattle and pigs), wet markets (chicken) and in urban areas (stray dogs, stray cats, wild rodents) by veterinary staff of the Hong Kong government (Ho et al. 2011a, 2012). All the cattle were imported from mainland China. The pigs and chicken included animals produced at local farms and those imported from mainland China. Samples from the stray dogs, stray cats and wild rodents were collected at Governmental Animal Management Centres. The stray animals and wild rodents were captured from urban areas from all over Hong Kong. The animals were selected for sampling at random in batches: chicken (20 animals per batch), cattle (10 animals per batch), pigs (2–7 animals per batch), stray cats (1–10 animals per batch), stray dogs (1–10 animals per batch) and urban rodents (2–23 animals per batch) (Ho et al. 2012).
Samples (approximately 0·1 g of faeces) were seeded with Dacron swabs onto three MacConkey agars (one plain and two supplemented with 2 μg ml−1 cefotaxime or 2 μg ml−1 ceftazidime, BD Diagnostics, Hong Kong, China). For each agar plate, at least one and up to five colonies were investigated. Bacteria were identified as E. coli by the Vitek GNI system (Biome′rieux Vitek Inc., Hazelwood, MO, USA), and their antibiotic susceptibilities were determined by the disc diffusion method according to the CLSI (Clinical and Laboratory Standards Institute 2011). The following antibiotic discs were used: amoxicillin–clavulanate (20/10 μg), cefotaxime (30 μg), ceftazidime (30 μg), amikacin (30 μ), chloramphenicol (30 μg), ciprofloxacin (5 μg), cotrimoxazole (1·25/23·75 μg), fosfomycin (200 μg), gentamicin (10 μg), nalidixic acid (30 μg), nitrofurantoin (300 μg) and tetracycline (30 μg) (BD Diagnostics). The double-disc synergy test (with cefotaxime, amoxicillin–clavulanate, ceftazidime at centre to centre, interdisc distances of 25 mm) was used for detection of extended-spectrum β-lactamases (ESBLs) (Ho et al. 1998). Isolates from the same animal were considered to be unique if the resistance profiles for amikacin, tetracycline, cotrimoxazole, ciprofloxacin, gentamicin, chloramphenicol and nitrofurantoin differed by at least one drug. One isolate from each MacConkey agar plate was included and up to three isolates per animal were included in the final analysis.
Susceptibility of the isolates to fosfomycin was determined by the disc diffusion method and interpreted according to the CLSI (inhibition zone diameters: ≥16 mm, sensitive; intermediate, 13–15 mm; and ≤12, resistant) (Clinical and Laboratory Standards Institute 2011). The fosfomycin MIC for the transconjugants was determined by Etest (AB Biodisk, Solna, Sweden) on Mueller–Hinton agar. The faecal carriage rate of fosfomycin-resistant E. coli was calculated as the number of animals with fosfomycin-resistant E. coli divided by the total number of animals tested.
The blaCTX-M genes were identified by PCR and sequencing (Lo et al. 2010; Ho et al. 2011a). Plasmid-mediated fosA3 was investigated by PCR, and the genetic environment of fosA3 (i.e. sequences flanking the open reading frame of fosA3) was determined by PCR mapping (Table 1). The primers for PCR mapping were selected to cover insertion sequences (IS26, ISEcp1, IS10) previously described to occur in the regions flanking fosA3 and blaCTX-M-14 (Wachino et al. 2010; Ho et al. 2012). J53 transconjugants carrying all the targets in two completely sequenced plasmids (pHK01 and pHK23a) and the J53 recipient strain were used as positive and negative controls, respectively (Ho et al. 2011b, 2012).
|Primer name||Direction||Nucleotide sequence (5′ to 3′)||Position||Accession No.||Amplicon size (bp)||Source|
|tnpIS26-R||F||AAC TCT GCT TAC CAG GCG||65-82||AB522970||1365||Ho et al. (2012)|
|fosA3-R||B||CGG TTA TCT TTC CAT ACC TCA G||c1429-1408||This study|
|fosA3-F||F||CCT GGC ATT TTA TCA GCA GT||1196-1215||AB522970||This study|
|fosA3-2R||B||GAG AAC ATG GAC AAA GAG AAC G||c1944-1923||AB522970||749||This study|
|tnpIS26-F||B||TGA CAT CAT TCT GTG GGC||c4042-4025||AB522970||2847||This study|
|IS10-F||F||GAC TGG TCT GAT ATT CGT GA||1904-1923||JQ318854||1377||This study|
|ISEcp1U1||F||AAA AAT GAT TGA AAG GTG GT||2915-2934||JQ318854||366||Saladin et al. (2002)|
|CTX-M-9B||B||ATT GGA AAG CGT TCA TCA CC||c3280-3261||Woodford et al. (2006)|
|MA1||F||SCV ATG TGC AGY ACC AGT AA||1167-1186||JQ343849||1567||Saladin et al. (2002)|
|fosA3-R||B||CGG TTA TCT TTC CAT ACC TCA G||C2733-2712||This study|
A subset of 18 fosA3 positive isolates originating from different animals (two cats, three dogs, two rodents, three cattle, four chickens and four pigs), collection year (two from 2008, seven from 2009 and nine from 2010) and carrying different CTX-M group of genes (seven positive for CTX-M-1 group, nine positive for CTX-M-9 group and two CTX-M negative) were randomly chosen for characterization of the fosA3 encoding plasmids. Transferability of the fosfomycin resistance determinant was investigated by filter mating using E. coli J53 (azide resistant) as the recipient (at donor recipient ratio 1 : 2). Transconjugants were selected on MacConkey agar containing 150 μg ml−1 sodium azide, 4 μg ml−1 fosfomycin (Sigma) and 25 μg ml−1 glucose-6-phosphate (Sigma, Hong Kong, China). The sizes of the plasmids were determined by S1-PFGE (Ho et al. 2011b). Plasmid typing was achieved by PCR-based replicon typing (PBRT) which recognizes FIA, FIB, FIC, HI1, HI2, I1, L/M, N, P, W, T, A/C, K, B/O, X, Y, F and FIIA replicons (Carattoli et al. 2005). The IncF plasmids were further typed by the FAB (FII, FIA and FIB) scheme which involves sequencing the copA (FII and FIC replicons), iterons-repE (FIA replicon) and the repB (FIB replicon) regions as previously described (Villa et al. 2010). In all the isolates, plasmid location of the blaCTX-M, fosA3 and replicons was confirmed by hybridization using specific PCR products as probes (Ho et al. 2012).
In total, 2106 faecal specimens from 210 cattle, 214 pigs, 460 chickens, 398 stray cats, 368 stray dogs and 456 rodents were cultured. Overall, 4·5% (95/2106) of the animals were found to be colonized by fosfomycin-resistant E. coli. The colonization rates by animal species were as follows: 11·2% (24/214) in pigs, 8·6% (18/210) in cattle, 7·3% (34/460) in chickens, 2·4% (9/368) in dogs, 0·8% (3/398) in cats and 1·5% (7/456) in rodents. Of the seven fosfomycin-resistant E. coli from rodents, five were recovered from R. norvegicus and one each was recovered from R. andamanensis and N. fulvescens. The fosfomycin-resistant E. coli carriage rate for food producing animals (chickens, pigs and cattle; 8·6%, 76/884; group A) was significantly higher than that for stray animals (cats and dogs; group B; 1·6%, 12/766; P < 0·001) and wild rodents (group C; 1·5%, 7/456, P < 0·001, chi-square test).
The animal samples were collected in 304 batches. The proportions of animal batches with at least one fosfomycin-resistant E. coli were as follows: cats, 5·1% (4/78 batches); dogs, 13·7% (10/73 batches); rodents, 15·4% (4/26 batches); chicken, 54·2% (13/24 batches); pigs, 37·3% (22/59 batches) and cattle, 40·9% (9/22 batches). For all the animals, fosfomycin-resistant isolates were found in multiple sampling batches scattered over the 2-year period.
The cultures yielded 1693 E. coli isolates of which 831 were ESBL producers. Overall, 93·6% (1584/1693) of the isolates were fosfomycin susceptible, 0·5% (8/1693) were fosfomycin intermediate, and 6·0% (101/1693) were fosfomycin resistant. Most (98/101) of the fosfomycin-resistant isolates grew to the edges of the fosfomycin discs. The antimicrobial resistance rates stratified by fosfomycin susceptibility categories are summarized in Table 2. Fosfomycin-resistant isolates were significantly more likely to be ESBL positive than were fosfomycin-susceptible isolates (P < 0·001). In addition, the resistance rates for eight non-β-lactam drugs (amikacin, chloramphenicol, ciprofloxacin, gentamicin, nalidixic acid, netilmicin and tetracycline) were significantly higher in fosfomycin-resistant isolates than fosfomycin-susceptible isolates (P < 0·05 for all comparisons). Thus, fosfomycin-resistant isolates were more likely to be multidrug resistant than fosfomycin-susceptible isolates (64·4% vs 38·4%, P < 0·001).
|% resistanta||P c|
|Fos-S (n = 1584)||Fos-I (n = 8)||Fos-R (n = 101)|
PCR showed that 96·0% (97/101) of the fosfomycin-resistant isolates were positive for the resistance determinant, fosA3. Five of the seven fosfomycin-resistant isolates from rodents were fosA3 positive and all five were recovered from R. norvegicus. Eight of the nine fosfomycin-resistant isolates and 26 of 27 fosfomycin-resistant isolates from dogs and pigs were fosA3 positive, respectively. All fosfomycin-resistant isolates from cats, cattle and chickens were fosA3 positive. Among fosfomycin-resistant isolates, 93·1% (94/101) carried CTX-M-type ESBL genes; 46·5% (47/101) had CTX-M-1 group genes; 43·6% (44/101) had CTX-M-9 group genes and 3·0% (3/101) had both CTX-M-1 and -9 groups of genes.
The regions surrounding fosA3 were explored by PCR mapping and sequencing of representative products and were arbitrarily designated as types A to G (Table 3 and Fig. 1). The insertion sequence IS26 was identified in the downstream region of fosA3 in all the isolates. The intergenic regions between fosA3 and the downstream IS26 included sequences of variable lengths (247, 536, 1075, 1222 and 1758 bp) having homology to regulatory genes (orf1, orf2 and orf3) in the chromosome of Klebsiella pneumoniae 342 (GenBank accession no. CP000964). Three different genetic organizations were identified in the region upstream of fosA3. In 81 (83·5%) isolates, IS26 was the only mobile element identified in the 5′ end (types A, B and C). The length of the upstream IS26/fosA3 intergenic space was 316 bp. In 14 (14·4%) isolates, the intergenic region between upstream IS26/fosA3 was interrupted by transposon-like structures, including ΔISEcp1-blaCTX-M-14-ΔIS903 (types D, n = 10) and ΔISEcp1-IS10-blaCTX-M-14-ΔIS903 (types E, n = 4). In two isolates (types F and G), the genetic elements upstream of fosA3 could not be defined.
|Typea||Genetic environment of fosA3||No|
A subset of 18 fosA3 positive isolates originating from different animals (two cats, three dogs, two rodents, three cattle, four chickens and four pigs), collection year (two from 2008, seven from 2009 and nine from 2010) and carrying different CTX-M group of genes (seven positive for CTX-M-1 group, nine positive for CTX-M-9 group and two CTX-M negative) were randomly chosen for characterization of the fosA3 encoding plasmids.
In conjugation experiments, the fosfomycin resistance trait could be successfully transferred to recipient E. coli strain in 11 isolates at frequencies of 10−5–10−1 per donor cells (Table 4). Fosfomycin MIC for all transconjugants was >1024 μg ml−1. In addition to fosfomycin resistance, there was cotransfer of the CTX-M-type ESBL trait in ten transconjugants. Four transconjugants had additional resistance involving chloramphenicol (one isolate), cotrimoxazole (four isolates), gentamicin (three isolates) and/or tetracycline (four isolates) in different combinations. In all but one isolate, a single plasmid with sizes of 50–200 kb was transferred into the recipient.
|Plasmid||Date||Origin||Size, kb||Replicona||Other resistance cotransferredb||CTX-M gene contenta||fosA3 genetic environmentc|
|1||pHK23a||Dec 08||Pig||80||F2:A-B-||–||CTX-M-3||Type A|
|2||pP121T||Dec 09||Pig||120||F2:A-B-||–||CTX-M-3||Type A|
|3||pN385T||Jan 09||Dog||80||F2:A-B-||–||CTX-M-3||Type A|
|4||pC353T||Feb 10||Chicken||150||F16:A1:B-||Gen, Sxt, Tet||CTX-M 65||Type A|
|5||pHK22a||Feb 10||Chicken||150||F16:A1:B-||Gen, Sxt, Tet||CTX-M-65||Type A|
|6||pC121T||Feb 09||Chicken||180||F24:A-B-||Gen, Sxt, Tet||CTX-M-14||Type E|
|7||pN863T||Feb 10||Dog||50, 200||N||–||CTX-M-14||Type D|
|8||pX6SA||Oct 08||Cattle||120||B/O||–||–||Type A|
|9||pX100T||Jul 09||Cattle||200||Untypeable||Chl, Sxt, Tet||CTX-M-3||Type A|
|10||pR424T||Jun 10||Rodent||80||Untypeable||–||CTX-M-55||Type A|
|11||pX56T||Feb 09||Cattle||80||Untypeable||–||CTX-M-55||Type A|
Three different replicons (FII, N and B/O) were found among the 11 fosA3-carrying plasmids (Table 4). Sequencing showed that the FII replicon fell into three different subtypes: F2:A-B-, F16:A1:B- and F24:A-B-. The replicon type for three plasmids could not be determined by the PBRT method. In ten of the transconjugants, hybridization experiments confirmed that the fosA3 and blaCTX-M genes were harboured on the same plasmids. The remaining isolate (pX6SA) was not a ESBL producer.
This study revealed significant rates of fosfomycin resistance among ESBL-positive E. coli isolates obtained from food animals, stray dogs/cats and wild rodents. To our knowledge, this is the first report of the occurrence of fosA3 among isolates in food animals and wild rodents. Importantly, the occurrence of fosfomycin-resistant strains was several folds higher among food animals than in dogs/cats. This may possibly be explained by a lower level of antibiotic exposure in dogs/cats. In food animals, fosfomycin has been studied for the treatment of various infections in chickens, cattle and postweaning piglets (Gutierrez et al. 2010; Soraci et al. 2011). However, no information was available on whether the food animals in the present study had been treated with antibiotics at the source farms. In China, fosfomycin resistance was also reported to have an association with CTX-M type ESBL among E. coli isolates from ten pet hospitals in the Guangdong province which has a land border with Hong Kong (Hou et al. 2012). It is possible that such organisms could have been carried across the border, in either direction through pets trade, thus providing an explanation on the prevalence of fosA3 among dogs/cats in Hong Kong and Guangdong.
Our data demonstrated that fosfomycin-resistant E. coli carrying fosA3 were detected from wild rodents (R. norvegicus) in urban areas. The finding corroborates previous observations that increasing prevalence of antimicrobial-resistant E. coli in human and veterinary medicine could be accompanied by coemergence of similar antimicrobial-resistant E. coli in wildlife species (Gilliver et al. 1999; Osterblad et al. 2001; Guenther et al. 2010; Ho et al. 2011a). The emergence of antimicrobial-resistant bacteria in wildlife animals could possibly be related to environmental exposure to antimicrobial residues, resistant bacteria or resistance genes (Gilliver et al. 1999; Osterblad et al. 2001; Allen et al. 2011). A study of enterobacteria from wild rodents living in close proximity to humans in North West England showed high antimicrobial resistance (Gilliver et al. 1999). In contrast, there was an absence of antimicrobial resistance among E. coli from wild rodents in areas in Finland not populated by humans (Osterblad et al. 2001), suggesting that carriage of antimicrobial-resistant bacteria in wildlife could be a result of human use of antibiotics. In addition, living in close proximity to farm animals is another factor reported to be associated with carriage of antimicrobial-resistant bacteria in wild rodents (Allen et al. 2011).
In the present study, the fosA3 and blaCTX-M genes were coharboured by conjugative plasmids of multiple incompatibility groups. In China, there is a huge burden of CTX-M-producing E. coli among animals (Ho et al. 2011a; Zheng et al. 2012). Therefore, the accumulation of fosA3 in multidrug resistant and CTX-M-producing strains is worrisome. Other studies have further demonstrated that these CTX-M-encoding plasmids may sometimes carry the rmtB gene (encoding 16S rRNA methylase) and various plasmid-mediated quinolone resistance determinants (Liu et al. 2012; Sun et al. 2012). In China, a high prevalence of a FII plasmid carrying four resistance genes (fosA3, blaCTX-M-24, blaTEM-1, and rmtB) has been reported among E. coli isolates from chickens and ducks in different geographical locations (Sun et al. 2012). As demonstrated by our conjugation experiments, these FII plasmids could carry resistance determinants encoding resistance to other antibiotics such as chloramphenicol and cotrimoxazole. Therefore, fosA3 carrying plasmids are subjected to coselection by multiple other antimicrobials. Our findings showed that fosA3 was flanked by IS26 under multiple genetic organizations. The IS26 elements might be involved in the horizontal spread of fosA3 across different plasmids. Beside IS26, the other mobile elements identified in the regions surrounding fosA3 include IS10, ISEcp1 and IS903. These elements are believed to play major roles in mobilization and dissemination of blaCTX-M. The presence of multiple insertion sequences and the complex transposon-like structures may facilitate the codissemination of fosA3 and blaCTX-M. In isolates with genetic environment other than types D and E, it was not clear whether the blaCTX-M was present upstream or downstream of fosA3. In future, it would be useful to completely sequence representative plasmids for delineating the relationship between the full spectrum of resistance determinants and how they relate to insertion sequences (Hou et al. 2012; Sun et al. 2012).
In conclusion, this study demonstrated the dissemination of transferable fosA3-mediated fosfomycin resistance among CTX-M-positive, multidrug-resistant strains in different types of animals. Future studies should investigate the prevalence of fosA3-mediated fosfomycin resistance determinants in human clinical isolates. The clonal structure of isolates and the fosA3 carrying plasmids from different sources should also be compared (Ho et al. 2009, 2012; Giufre et al. 2012).
This work was supported by grants from the Research Fund for the Control of Infectious Diseases (RFCID) of the Health, Welfare and Food Bureau of the Government of the HKSAR and Consultancy Service for Enhancing Laboratory Surveillance of Emerging Infectious Disease for the HKSAR Department of Health. We thank the staff at the Agriculture, Fisheries and Conservation Department (AFCD) and Food and Environmental Hygiene Department (FEHD) of the Hong Kong Special Administrative Region for assistance with specimen collection.
Authors have no conflicts of interest to declare.