Research in Veterinary Science 93 (2012) 1142–1150

Contents lists available at SciVerse ScienceDirect

Research in Veterinary Science

journal homepage: www.elsevier .com/locate / rvsc

Molecular and phenotypic characterisation of Extended Spectrum b-lactamaseCTX-M Escherichia coli from farm animals in Great Britain

Monique Toszeghy, Neil Phillips, Hannah Reeves, Guanghui Wu, Chris Teale, Nick Coldham,Luke Randall ⇑Animal Health and Veterinary Laboratories Agency (Weybridge), New Haw, Addlestone, Surrey KT15 3NB, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history:Received 7 December 2011Accepted 7 May 2012

Keywords:Antibiotic resistancePoultryCattleVirulenceReplicon

0034-5288/$ - see front matter Crown Copyright � 2http://dx.doi.org/10.1016/j.rvsc.2012.05.001

⇑ Corresponding author. Tel.: +44 (0) 1932 357906E-mail address: luke.randall@ahvla.gsi.gov.uk (L. R

The aim of this study was to characterise CTX-M Escherichia coli isolates from cattle, chickens and turkeysin Great Britain with respect to CTX-M sequence type, replicon type, ability to transfer plasmids, and forthe presence of antibiotic resistance, fitness and virulence genes as determined by micro-arrays. Themain CTX-M enzymes identified in E. coli from cattle, chicken and turkeys were 14 and 15, 1 and 15,and 1 and 14 respectively. Most isolates from different animal species transferred their plasmids withsimilar frequencies. The plasmid replicon type I1-k was most common and seen in 23%, 95% and 50%of the isolates tested from cattle, chickens and turkeys respectively, whilst types F, FIA, FIB and K werecommon to isolates from cattle and turkeys only. Thirty-eight different antibiotic resistance genes weredetected by micro-array including aad genes, blaCTX-M, blaTEM, cat genes dfrA, floR, strA, strB, sul, sul2 tetAand tetB. Thirty-nine different fitness and virulence genes were also detected by-micro-array, includingespP, ireA, lpfA, mchF, prfB and tsh. Fisher exact test and hierarchical clustering of the antibiotic resistanceand virulence gene results showed some genes were more commonly associated with isolates from chick-ens or cattle. This study provides a baseline of the characteristics of CTX-M E. coli isolates from animals inGreat Britain and suggests that chicken and cattle CTX-M E. coli represent different populations.

Crown Copyright � 2012 Published by Elsevier Ltd. All rights reserved.

1. Introduction

At the start of 2012 there were approximately 120 differentCTX-M b-lactamases listed on the Lahey clinic web-site (Lahey-clinic, 2012) and for some years Escherichia coli carrying CTX-M Ex-tended Spectrum b-Lactamase (ESBL) plasmids have been consid-ered to be a threat to the effective use of third generationcephalosporin antibiotics for these and similar organisms (Liver-more et al., 2007). Of particular concern in human medicine isthe human pandemic O25:H4-ST131 E. coli clone which is oftenamikacin resistant and carries CTX-M-15 on an Inc FII plasmid(Lau et al., 2008). In the UK, the dominant CTX-M sequence typesseen in E. coli from humans are CTX-M-15/-3 and, much less fre-quently, CTX-M-14/-9 (Tarrant et al., 2007).

The isolation of the first CTX-M-producing E. coli from animalsin 2000 and 2001 was reported in Spain from healthy chickens(Brinas et al., 2003) and in Japan from cattle (Shiraki et al., 2004).Since these first isolations of CTX-M-producing E. coli they havebeen isolated from chickens in many countries such as Belgium(Smet et al., 2008), Japan (Kojima et al., 2005) and Portugal(Machado et al., 2008) as well from chickens in other countries

012 Published by Elsevier Ltd. All r

; fax: +44 (0) 1932 347046.andall).

and from other animal species and as outlined previously (Randallet al., 2011a).

Recently in the UK CTX-M-carrying E. coli were isolated from3.6% of individual chicken broiler caecal samples collected frombroilers at slaughter, and from 5.2% of turkey meat productionfarms and 6.9% of turkey breeder farms (Randall et al., 2011a).

Although the survey of CTX-M isolates from chickens and tur-keys in GB (Randall et al., 2011a) suggested that the isolates werein general different from the main ones causing disease in humansin GB, some studies suggest that a common epidemiological pool ofsimilar plasmids encoding CTX-M resistance may circulate in ani-mals and humans or that animals may comprise a source of ESBLsin humans (Cottell et al., 2008). In Belgium, plasmids with identicalRFLP patterns carrying a blaTEM-52 ESBL were found in Enterobacte-riaceae isolated from poultry and a human suggesting a sharedsource (Smet et al., 2009). In another example, a multi-drug resis-tant E. coli indistinguishable from a type of ESBL found in poultrywas recently associated with a death of a human patient in theNetherlands (Sheldon, 2010).

Bearing in mind the potential for organisms resistant to antimi-crobials to enter the food chain from animals, there is need to beable to characterise bacteria from animals with respect to antibi-otic, virulence and fitness genes, ability to pass on plasmids andplasmid types. This information can help to inform whether certain

ights reserved.

M. Toszeghy et al. / Research in Veterinary Science 93 (2012) 1142–1150 1143

bacteria are likely to present problems if they enter the food chain,but can also serve to ‘‘fingerprint’’ bacteria from animals and assuch help to determine if they are related to those causing diseasein humans.

Recent studies in cattle, chickens and turkeys have provided apanel of CTX-M E. coli for further examination (Randall et al.,2011a). The aim of this work therefore was to report on and per-form further work on isolates from previous published studiesfor chicken and turkey isolates (Randall et al., 2011a), and fromunpublished studies for cattle isolates with respect to CTX-M type,replicon type and relative conjugation rate, plasmid size and anti-biotic resistance, fitness and virulence genes as determined by mi-cro-array (Anjum et al., 2011; Wu et al., 2010a) and plasmid size.

2. Materials and methods

2.1. Control isolates

Recipient strains for conjugations were a rifampicin resistantmutant of Salmonella enterica serovar Typhimurium 26R andE. coli K12. All PCRs were carried out using suitable positive andnegative controls.

2.2. Isolates from animals

E. coli isolates from chickens and turkeys (Table 1) were mainlycollected from healthy birds as part of randomised survey, re-ported previously (Randall et al., 2011a) but also included threeisolates recovered from chicken diagnostic samples (Randallet al., 2011b). Isolates from cattle (Table 1) were derived fromthose received at the Animal Health and Veterinary LaboratoriesAgency (AHVLA) since 2004 and originated both from diagnosticsamples submitted from cases of (usually unrelated) clinical dis-ease in cattle to AHVLA laboratories and from surveillance forCTX-M in healthy cattle on farms (Horton et al., 2012).

2.3. Confirmation of isolates as E. coli

All CTX-M-positive chicken and turkey isolates were confirmedas E. coli using Matrix-assisted laser desorption ionisation-time offlight mass spectrometry (MALDI-TOF). Standard Bruker protocolsas described previously (Eigner et al., 2009) were used for MAL-DI-TOF analysis using a Bruker Autoflex II machine. Resulting un-known spectra were compared to reference spectra (MALDIBiotyper reference library version 3, Bruker Daltronics) using MAL-DI Biotyper software version 2.0.10.0 (� Bruker Daltronics, 2008).All CTX-M positive cattle isolates were confirmed as E. coli eitherby MALDI-TOF as part of a previous study (Horton et al., 2012) orphenotypically by ability to grow on CHROMagar ECC and by bio-chemical tests.

2.4. PCR and sequencing

The presence of the blaCTX-M gene in isolates was confirmedusing a real-time PCR assay (Batchelor et al., 2005) which can alsoassign the isolates as CTX-M groups 1, 2 or 9. All positive isolateswere sequenced to determine CTX-M sequence type (Carattoliet al., 2008; Sabate et al., 2002).

2.5. Conjugations

The ability of isolates to transfer CTX-M genes to rifampicinresistant Salmonella Typhimurium (isolate 26R) or E. coli K12 byconjugation was determined using a filter mating technique as pre-viously described (Randall et al., 2011a). In order to ascertain the

optimum time point for conjugation studies, two chicken isolatesof different CTX-M sequence types were conjugated after 0.5, 1,2, 3 and 5 hr time intervals.

2.6. Replicon typing

Characterisation of plasmids, based on incompatibility (Inc)grouping was performed on transconjugants of isolates from differ-ent animal species based on methods described previously (Carat-toli et al., 2005). Prior to preparing cell lysates, transconjugantswere grown on media containing 2 mg/L cefotaxime to maintainselective pressure for the plasmid.

2.7. Sizing of plasmids

To determine the number of plasmids in transconjugants andtheir approximate sizes, plasmid sizing was carried out on cattle,chicken and turkey isolates, using an adapted PFGE method of (Bar-ton et al., 1995) as reported previously (Randall et al., 2011a). Plas-mids were extracted from transconjugants, separated by PFGE,blotted to a nitrocellulose membrane and probed with a CTX-Mprobe to confirm the presence of the CTX-M gene in the plasmid.

2.8. Genes on the arrays

All 67 antibiotic resistance and 74 fitness and virulence geneson the arrays used are listed on the Identibac web site for the Iden-tibac AMR �ve antibiotic resistance gene array (Identibac, 2011-A)and for the pathotyping array (Identibac, 2011-B). These geneshave also been described in previous publications for antibioticresistance genes (Anjum et al., 2011) and fitness and virulencegenes (Wu et al., 2010a).

2.9. Determination of genes by array

Mainly transconjugants, but some wild type strains also, wereanalysed for antimicrobial resistance and for virulence genes, usingthe Identibac Gram-negative gene array (Identibac, cataloguenumber RAK 6251) and the Identibac E. coli pathotyping array(Identibac, catalogue number RAK 6255). These two arrays are sim-ilar to those described previously (Anjum et al., 2011; Wu et al.,2010a), but have a few additional probes. Strains were grown onLuria–Bertani (LB) agar with 2 mg/L cefotaxime, arrays were per-formed as per manufacturer’s instructions.

2.10. Statistical comparison

To determine if some of the virulence and fitness genes in wildtype isolates were more strongly associated with isolates from cer-tain animal species, the Fisher Exact test was used (StatXact 8, cy-tel, USA).

2.11. Previous work

This paper reports on some work that the authors have pub-lished previously (Randall et al., 2011a,b) but then builds on it. InTable 1 the characteristics of different E. coli CTX-M strains are re-ported. For the cattle isolates these results have not been publishedpreviously, but these results have been previously published forchicken and turkey isolates (Randall et al., 2011a) which are in-cluded in the table for completeness and to give a background tothe array data. In Table 2, again the information about the cattleisolates has not been previously published nor has the compari-sons of all animal CTX-M types to replicon types and plasmid sizesbeen published for combined isolates from cattle, chickens and tur-keys, although for chicken and turkey isolates, the replicon types

Table 1Characteristics of cattle, chickena and turkey blaCTX-M Escherichia coli isolates with respect to CTX-M sequence type and plasmids.

Animalspecies

Characteristics of CTX-M strains from different animal species

CTX-M sequence (n) Conjugation rate range (n) Replicon types (n) Range of plasmid sizesb

Types [% each type]c [mean/SD] [% I1-k and K] (n)

Cattle 1 [7.4%], 3 [0.5%], 14 [36%], 14B [11.3%], 15b [37.5%], 20 [0.5%],27 [0.5%], 28 [2.4%], 55 [1.5%], NT [1.5%] (202)

9.09 � 10�8 to 1.22 � 10�4

[5.44 � 10�5/1.18 � 10�4] (31)A/C, F, FIA, FIB, I1-k[23%], K [10%], P, Y (31)

70–90 kb (9), 90–148 kb(16) >148 kb (3)

Chickens 1 [73%], 3 [6.6%], 15 [17%], NT [3.3%] (27) 6.22 � 10�6 to 1.01 � 10�4

[1.00 � 10�4/1.08 � 10�4] (20)A/C, I1-k [95%], P (19) 90–148 kb (18), >148 kb (1)

Turkeys 1 [36.5%], 14 [45.1%], 15 [12.9%], 55 [1.0%] NT [5.3%] (93) 9.11 � 10�7 to 1.05 � 10�4

[6.77 � 10�5/1.61 � 10�4] (32)B/O, F, FIA, FIB, I1-k[50%], K [42%] (30)

70–90 kb (7), 90–148 kb(23) 90 kb and 130 kb (2)

n, number for a given animal species for a given test or characteristic. NT – Not tested.a For chicken and turkey isolates these characteristics have been reported previously (Randall et al., 2011a) but are included here to give background to the array work.b <70 kb plasmid sizes not included.c 23.2% of the CTX-M 15 isolates were sequenced as CTX-M 15/28, but as CTX-M type 28 is not seen in the UK and further sequencing has shown such isolates to be CTX-M

15, CTX-M 15/28 isolates have been listed as CTX-M 15.

1144 M. Toszeghy et al. / Research in Veterinary Science 93 (2012) 1142–1150

and plasmid sizes have been previously published as for Table 1(Randall et al., 2011a). The array data is the main new work of thispaper, and none of this has been published previously, except forvirulence array data the chicken isolates (Randall et al., 2011b).Isolates were selected for arrays from the panel of cattle, chickenand turkey CTX-M isolates held at AHVLA to represent a range ofCTX-M types and isolates from different locations.

3. Results

3.1. Conjugation rates

A one hour conjugation time was chosen for all experiments(Fig. 1), as this ensured time for plasmid transfer, but did not givetime for the recipients to become donors.

Conjugations were performed on a subpanel of 31, 20 and 32cattle, chicken and turkey isolates respectively (Table 1) and re-ported previously for the poultry isolates (Randall et al., 2011a).Of the cattle, chicken and turkey isolates, approximately 60%,75% and 64% transferred their plasmids to recipient strains. How-ever a number of isolates with CTX-M 1 (chicken, turkey), 3 (chick-en), 14 (cattle, turkey) and 15 (cattle, chicken, turkey) sequencetypes were unable to conjugate.

The relative conjugation frequencies where conjugation oc-curred, ranged from �10�4 to �10�8 for cattle isolates, from�10�4 to 10�6 for chicken isolates and from �10�4 to 10�7 for tur-key isolates. Two turkey isolates which did not conjugate againstSalmonella Typhimurium (isolate 26R), did so with E.coli K12.

3.2. Identification of isolates

All poultry isolates were confirmed E. coli by MALDI-TOF analy-sis as reported previously (Randall et al., 2011a) whilst cattle iso-lates were confirmed E. coli by biochemical tests or by MALDI-TOF as reported previously (Horton et al., 2012).

Table 2Replicon types and plasmid sizes for specific CTX-M sequence types for cattle, chicken an

CTX-M Sequence Type (n) Number of isolates with replicon types

A/C B/O F FIA FIB I1-k

1 (32) 3 5 273 (2) 2 114 (26) 1 9 10 1 515 (10) 10 3 520 (1) 155 (1) 1

Type 1 associated with I1-k, 14 with IncK and FIAor F and 15 with FIA.

3.3. CTX genes, replicon types and plasmid sizes in isolates fromdifferent species

CTX-M sequence types in the 202 E. coli from cattle tested weremainly those in CTX-M group 1 (�45%) and group 9 (�45%), ofwhich the main sequence types were CTX-M 15 (37.5%) and CTX-M 14 (36%) respectively (Tables 1 and 2). Less common CTX-M se-quence types in cattle isolates were 14B, 20, 27 and 55.

CTX-M sequence types in the 27 E. coli from chickens were all inCTX-M group 1 and were predominantly CTX-M-1 (73%, Table 1) asreported previously (Randall et al., 2011b).

CTX-M sequence types in the 93 E. coli from turkeys were CTX-M 14 (45.1%) and CTX-M 1 (36.5%). Sequence types 15 and 55 werepresent at lower numbers (Table 1) as reported previously (Randallet al., 2011a).

Replicon type IncI1- k was found in 23%, 95% and 50% of thetested cattle (n = 31), chicken (n = 19) and turkey (n = 30) isolatesrespectively and 72% of these isolates were CTX-M 1 sequence type(Tables 1 and 2). Inc type A/C was not present in any of the turkeyisolates (Randall et al., 2011a); however it was present in two iso-lates from cattle and chicken (Randall et al., 2011b) where it wasassociated with CTX-M 3 sequence type. In the majority of cases,Inc type K was found in isolates with CTX-M 14 sequence typeand was only present in cattle and turkey (Randall et al., 2011b)isolates (Table 1 and 2). None of the turkey isolates tested were po-sitive for Inc type P (Randall et al., 2011b), but this Inc type waspresent in a few isolates from cattle and chickens (Randall et al.,2011b) with CTX-M 1 and CTX-M 3 sequence types (Table 2).One turkey (Randall et al., 2011b) and one cattle isolate, both withCTX-M 14 sequence type had one Inc B/O and one Inc Y plasmidrespectively. Inc type FIA was present in much higher numbersin isolates from cattle than from isolates in chickens and turkeys(Randall et al., 2011b). Cattle and turkey (Randall et al., 2011b) iso-lates were also positive for Inc types F, FIB in CTX-M 1, 14, 15 and55 sequence types (Tables 1 and 2).

d turkey Escherichia coli.

Number of isolates with range of plasmid sizes

K P Y 70–90 kb 90–148 kb >148 kb >one plasmid

2 1 1 312 2

16 1 12 14 23 7

11

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1 2 3 4 5

Time in hours

Rel

ativ

e co

njug

atio

n fr

eque

ncy

CTX-M 1strain

CTX-M 3strain

Fig. 1. Relative conjugation frequency over time for two strains carrying blaCTX-M genes.

M. Toszeghy et al. / Research in Veterinary Science 93 (2012) 1142–1150 1145

The majority of plasmids detected in E. coli from all three animalspecies were in the size range of �90–148 kb (Tables 1 and 2), andthese were present in isolates with all the CTX-M sequence typesdetected, except CTX-M 3 sequence type (Table 2). Cattle and tur-key isolates (Randall et al., 2011b) that had plasmids of �70–90 kb were of sequence type 14 (Table 2). Chicken isolates (Randallet al., 2011b) did not have plasmids in the range �70–90 kb, butotherwise had similar sized plasmids to plasmids from cattle andturkey isolates. Two isolates (one cattle, one chicken) had plasmids>148 kb associated with sequence type 3 (Table 2).

Multiple replicon types were often present on one plasmid,however two turkey isolates (Randall et al., 2011b) that both wereCTX-M 14 sequence type had two plasmids of �90 and �130 kb.

3.4. Resistance genes in isolates from different species

A total of 101/18, 17/17, 14/12 wild type/transconjugants fromcattle, chickens and turkeys respectively were tested for the pres-ence of antibiotic resistance and integron genes (Table 3). Trans-conjugants from cattle, chicken and turkey isolates were found tocarry in addition to blaCTX-M genes, a total of 10, 8 and 7 other anti-microbial resistance genes respectively, and these mainly encodedresistances to older antibiotics (Table 3). The fact that these geneswere found on transconjugants shows that they must be co-locatedon the same plasmid with the blaCTX-M gene, though of course, notall isolates carried all of these genes.

Transconjugants carried genes conferring resistance to amino-glycosides, b-lactams, trimethoprim, sulphonamides, and tetracy-clines (Table 3). Fifty-six percent of transconjugants from cattleisolates, 29% from chickens and 50% of turkey transconjugants car-ried genes encoding trimethoprim resistance, whilst 11%, 65% and8% of transconjugants from cattle, chicken, turkey isolates carriedthe gene tetA encoding tetracycline resistance respectively. Thegene dfrA, was the most prevalent antibiotic resistance gene intransconjugants from cattle and turkey isolates (all isolates ofcourse possessed blaCTX-M genes, as this was the preliminary screenfor inclusion). A sulphonamide resistance gene was detected in 53%of chicken transconjugants, in comparison with 22% cattle and 8%turkey transconjugants. Many antibiotic resistance genes wereonly detected in wild type (i.e. non-transconjugant) isolates partic-ularly from cattle isolates, but as there were many more wild typecattle isolates (n = 101) tested than the transconjugants (n = 18)from different animal species it is not possible to draw conclusions

from the data (Table 3). Of specific interest, about 10% and 30% ofthe cattle isolates, but none of the other isolates, were positive forthe catB and floR genes respectively encoding chloramphenicolresistance.

3.5. Fitness and virulence genes in isolates from different species

A total of 101/9, 26/15, 19/7 wild type/transconjugants fromcattle, chickens (Randall et al., 2011b) and turkeys respectivelywere tested for the presence of fitness and virulence genes (Ta-ble 4). The number of fitness and virulence genes detected in trans-conjugants from cattle, chicken (Randall et al., 2011b) and turkeyisolates were 2, 2 and 0 respectively. The number of fitness and vir-ulence genes detected in wild type strains from cattle, chicken(Randall et al., 2011b) and turkey isolates were 36, 14 and 11respectively (Table 4). Although it was not possible to make directcomparisons between wild types and transconjugants, only a fewof the fitness and virulence genes seemed to be carried on plasmidsas evidenced by presence of genes in transconjugants (Table 4).However, the conjugation method we used primarily selected forCTX-M plasmids, so it could be that the fitness and virulence geneswere carried on other plasmids that were not selected for. Exam-ples of fitness and virulence genes that seemed to be carried onthe plasmids selected were mcmA and rpeA.

The genes espP, ireA, iroN, iss, lpfA, mcmA, mchF, prfB and tshwere among the commonest seen, with the iss gene encoding forincreased serum survival being the most common in isolates fromall three animal species. The iroN gene was particularly common inisolates from chickens and turkeys, whilst the tsh gene was partic-ularly common in isolates from chickens. The espP, ireA, mcmA andprfB genes were particularly common in isolates from cattle, butnot in isolates from chickens and turkeys in the population of iso-lates examined (Table 4).

3.6. Gene maps

The antibiotic resistance genes located on plasmids in the cattle,chicken and turkey isolates are shown by a heat map diagram inFig. 2, whilst Fig. 3 is a heat map showing the presence of fitnessand virulence genes in the chicken and turkey isolates and someof the cattle isolates.

It was interesting that both these hierarchial clustering dia-grams for totally different sets of genes tended to cluster most

1146 M. Toszeghy et al. / Research in Veterinary Science 93 (2012) 1142–1150

cattle, chicken and turkey isolates as groups to some extent,although there were overlaps. However, there was more overlapbetween chicken and turkey isolates, whilst cattle isolates tendedto be slightly distinct.

Whilst all of the isolates were positive for the blaCTX-M gene byPCR, some of the array results for this gene were ambiguous (0.3–0.399 result) and as such some results for this gene appear as blue.This could be due to the probes have reduced sensitivity comparedto PCR for some CTX-M types and warrants further work.

In Fig. 3 the lpfA, mchF and tsh gene was mainly associated withchicken isolates, whilst the gene iroN was mainly associated withchicken and turkey isolates, and several genes were only associatedwith cattle isolates. In general, the isolates from chickens and tur-keys had a higher number of fitness and virulence genes per isolatethan the cattle isolates.

3.7. Association of virulence and fitness genes with isolates fromcertain animal species

The results of the Fisher Exact test showed that the cba, cma,iroN, iss, lpfA, mchF and tsh genes were all significantly more asso-ciated with chicken and turkeys isolates than cattle isolates at

Table 3Presence of antibiotic resistance genes in CTX-M positive isolates from different animal sp

Gene name Gene group and or general function Number (N

Cattle isola

WT (101)

aac(30)Ia Aminoglycoside resistance 1aac(30)IV Aminoglycoside resistanceaac(60)-Ib Aminoglycoside, possible quinolone (crb) 12aac2 Aminoglycoside resistance 1aadA1 Aminoglycoside resistance 40aadA2 Aminoglycoside resistance 6aadA4 Aminoglycoside (STR, SPT) resistance 16ant(20)Ia Aminoglycoside resistance 1blaCMY Beta-lactam - AmpC 1blaCTX-M group-1 CTX-M group 1 – ESBL 58c

blaCTX-M group-2 CTX-M group 2 – ESBL 2blaCTX-M group-9 CTX-M group 9 – ESBL 34blaDHA-1 Beta-lactam - AmpC 1blaFOX Beta-lactam - AmpC 1blaLEN-1 Beta-lactam 1blaOXA

d Beta-lactam 1blaSHV Beta-lactam 1blaTEM Beta-lactam 68catA Chloramphenicol resistance 24catB Chloramphenicol resistance 9cat3 Chloramphenicol resistance 1cmlA Efflux pump 3dfr12 Trimethoprim resistance 4dfrAe Trimethoprim resistance 54dfrV Trimethoprim resistance 2floR Chloramphenicol resistance 32Intl1 Class 1 integrase 40Intl2 Class 2 integrase 5strA Streptomycin resistance 8strB Streptomycin resistance 53sul1 Sulphonamide resistance 43sul2 Sulphonamide resistance 36sul3 Sulphonamide resistance 5tetA Tetracycline resistance 34tetB Tetracycline resistance 42tetC Tetracycline resistancetetD Tetracycline resistance 1tetG Tetracycline resistance

WT, wild type. TC, transconjugant. SPT, spectinomycin. STR, streptomycin.a Positive is taken as array results >0.399.b The aac-(6’)-Ib-cr variant of the aac-(6’)-Ib gene is involved with reduced susceptibic All isolates were positive for the CTX-M gene by PCR, but some array results for the C

in this table.d The blaOXA probe may not detect all OXA types.e dhfA probe variants 1, 7, 14, 15, 17, 19 were seen.

p < 0.001. Conversely, the espP and mcmA genes were morestrongly associated with cattle isolates than chicken or turkeysisolates.

The lpFA and tsh genes were also present in isolates from chick-ens at significantly higher levels compared to isolates from turkeys(p = 0.036 and p < 0.001 respectively).

4. Discussion

In previous publications the prevalence and characteristics ofCTX-M sequence types in E. coli from chickens and turkeys in GB(Randall et al., 2011b) and the presence of virulence genes in chick-en CTX-M producing E. coli isolates has been reported (Randallet al., 2011b). In this study we complement and add to previousfindings, to include characteristics of CTX-M E. coli isolates fromcattle, to include new array results for antibiotic resistance genesfor E. coli from all animal species and to include new fitness andvirulence array data for E. coli from cattle and turkeys.

The highest number of isolates that were tested for CTX-M dif-ferent sequence type were the cattle isolates, and these alsoshowed the most variability with respect to CTX-M sequence types

ecies (positive genes only).

o. tested) of isolates positivea for probes

tes Chicken isolates Turkey isolates

TC (18) WT (17) TC (17) WT (14) TC (12)

1

10 1 5 11 1 1

3 2 2 2 2

111 14c 17 7 5

7 7 7

1

6 10 1 12

1

10 2 5 8 6

2 2 3 5 23

1 13 2 1 31 5 8 3 1

62 7 11 8 11 8 9

4

1 1

lity to fluoroquinolones.TX-M gene were intermediate (value 0.3–0.399) and these are not shown as positive

Table 4Presence of virulencea genes in CTX-M positive isolates from different animal species (positive genes only).

Gene and or probe name Gene group and or general function Number (No. tested) of isolates positive for probes

Cattle isolates Chicken isolates Turkey isolates

WT (101) TC (9) WT (26) TC (15) WT (19) TC (7)

astA Heat stable enterotoxin 1 7 1cba Colicin B pore forming 1 3 6celB Endonuclease colicin E2 2cif Type III secreted effector 1cma Colicin M 4 5 8eae Intimin 6efa EHEC factor for adherence 1epeA SPATE 1espF Type III secretion system 2espI SPATE 2espJ Prophage encoded type III ss effector 1espP Putative exoprotein-precursor 22F17-A Subunit A of F17 fimbrial protein 7F17-G Adhesion subunit of F17 fimbrial protein 9hylA Haemolysin A 2Iha Adherence protein 11 3ireA Sidephore receptor 18 2 2iroN Enterobactin sidephore receptor protein 4 19 14Iss Increased serum survival 47 26 17katP Catalase peroxidise, plasmid encoded 2lpfA Long polar fimbriae 17 6mchB Microcin H47 part of colicin H 3 3mchC MchC protein 3 3mchF ABC transporter protein MchF 6 16 6mcmA Microcin M part of colicin H 25 1 1nleA Non LEE encoded effector A 4nleC Non LEE encoded effector C 2pic SPATE 10 1prfB P-related fimbriae regulatory gene 25 3 1rpeA SPATE 1saa Auto agglutinating adhesion 1senB Plasmid encoded enterotoxin 3sepA SPATE 1tccP Tir-cytoskeleton coupling protein 2tir Translocated intimin receptor protein 2 1toxB Toxin B 1 1 1tsh Temperature sensitive haemaglutin 3 20 3vat SPATE 8

WT, wild type; TC, transconjugant.a For chicken isolates these characteristics have been reported previously (Randall et al., 2011b) but are included here for comparative purposes.

M. Toszeghy et al. / Research in Veterinary Science 93 (2012) 1142–1150 1147

(Tables 1 and 2). In all 9, 3 and 4 CTX-M sequence types were seenfor isolates from cattle, chickens and turkeys respectively, andthese were CTX-M types 1, 3, 14, 14B, 15, 20, 27, 28 and 55 for cat-tle isolates, 1, 3 and 15 for chicken isolates (Randall et al., 2011b)and 1, 14, 15 and 55 for turkey isolates (Randall et al., 2011b).All CTX-M types occurring in chickens and turkeys were also de-tected in cattle. CTX-M sequence types 14B, 20, 27 and 28 wereonly detected in cattle isolates and CTX-M sequence types 14and 55 were only found in both cattle and turkey isolates.

The association of different replicon types with the chicken andturkey isolates has been discussed previously (Randall et al.,2011b). Certain replicon types appear to be associated with certainCTX-M sequence types, for example CTX-M-14 E. coli has recentlybeen reported as the most widespread CTX-M sequence type in hu-mans in Spain (Valverde et al., 2009). In the work of Valverde et al.,the human CTX-M-14 E. coli isolates were mainly associated withIncK plasmids; a similar association between IncK plasmids andCTX-M-14 was detected in this study in the cattle and turkey iso-lates as shown in Tables 1 and 2 and as previously published (Ran-dall et al., 2011b). Some replicon types, such as F, FIB and I1-k werefound in isolates from cattle, chickens and turkeys, whilst IncKplasmids were only found in isolates from cattle and turkeys, Inctypes A/C and P were only found in cattle and chicken isolates,Inc B/O was only found in turkey isolates and IncY was only foundin a single cattle isolate.

In a recent study 1015 E. coli isolates from healthy and diseasedchickens, from retail meat and from faeces of healthy human vol-unteers and humans with urinary tract infections were tested forplasmid types (Johnson et al., 2007). All of the replicon types seenin this study were also seen in the 1015 E. coli from chickens andhumans in the previous study (Johnson et al., 2007). In particular,replicon types FIB and I1-k, which were found in cattle, chickensand turkeys in this study, were found in isolates from humans(Johnson et al., 2007). Conversely, replicon type K found in cattleand turkey isolates in this study was found in poultry isolatesbut not human isolates in the previous study (Johnson et al.,2007). This suggests that similar plasmids exist in E. coli from farmanimals and people to some extent, although comparison are oflimited benefit as the study of Johnson et al. did not focus onCTX-M plasmids.

There was limited evidence of certain plasmid sizes being asso-ciated with certain CTX-M types, for example CTX-M sequencetype 3 isolates had plasmids that were >148 kb, CTX-M sequencetype 1 had plasmids that were mainly 90 to 148 kb and CTX-M se-quence type 14 had an almost equal split of plasmids in the range70 to 90 kb and 90 to 148 kb. There were also two isolates withmore than one plasmid. In previous studies where IncK FIA andFIB plasmids from E. coli with CTX-M 15 sequence type isolatedfrom humans have been sized, they have been in the range45.5–194 kb (Mshana et al., 2009) a broadly similar size range to

Fig. 2. Antibiotic resistance gene profiles for all cattle, chicken and turkey transconjugant isolates.

Fig. 3. Virulence gene profiles for all wild types chicken and cattle isolates and for some cattle (n = 26) of the wild type cattle isolates.

1148 M. Toszeghy et al. / Research in Veterinary Science 93 (2012) 1142–1150

plasmids in this study, except that some smaller plasmids werealso seen.

The antibiotic resistance genes present were in general thosecommonly detected in E. coli from farm animals (Anjum et al.,

M. Toszeghy et al. / Research in Veterinary Science 93 (2012) 1142–1150 1149

2011). Some of the wild type strains were positive for genes thatwould be likely to be plasmid encoded, but these genes were notdetected on the transconjugants. This is likely to be due to the factthat the conjugation protocol specifically selected plasmids bear-ing the blaCTX-M gene and not all plasmids.

Of note, most or all of the isolates were not positive for genes onthe arrays such as blaAMP-C genes, genes conferring transferablequinolone resistance for example qnrA, qnrB and qnrS, or for otherESBL or metallo-Beta-Lactamase resistance genes such as blaIMP,blaOXA, blaPER, blaPSE, blaVEB, blaVIM. In a recent study in Portugal,the pandemic O25-ST131 human virulent E. coli CTX-M-15-pro-ducing clone harbouring the qnrB2 and aac(60)-Ib-cr genes was iso-lated from dogs (Pomba et al., 2009). Although 12 of the cattleisolates were positive for the aac(60)-Ib gene, further analysis wasnot performed to determine if these isolates harboured theaac(60)-Ib-cr variant of this gene involved in reduced susceptibilityto quinolones. It was reassuring in this study that no qnrA, B or S ormetallo-ESBL genes were detected in the isolates screened.

There are a number of studies published that describe the pres-ence of virulence genes in E. coli from cattle, chickens and turkeysor their products (Badri et al., 2009; Cortes et al., 2010; Ghanbar-pour et al., 2010; Janben et al., 2001; Rodriguez-Siek et al.,2005a,b; Wu et al., 2010a,b), but each study has a different set ofvirulence genes, making comparisons between the different stud-ies and between our study and these studies difficult.

In a recent study we showed that the virulence genes in isolatesfrom chickens were in the main different from the virulence genesin isolates from humans for a small sample set, but the virulencegenes were similar for a limited number of isolates from diseasedand healthy birds (Randall et al., 2011b). Other workers have com-mented that seven virulence associated genes are significantly dis-tributed among E. coli strains isolates from poultry withcolibacillosis, namely iucD, tsh, fimC, fyuA, irp2, papC and astA (Jan-ben et al., 2001). The majority (79.3%) of Avian Pathogenic E. coli(APEC) strains however possess the three genes fimC, iucD andtsh, and tsh is considered a marker for APEC strains (Janben et al.,2001). The array used in this study did not include probes for iucD,fimC, fyuA, irp2 and papC genes, but did include many genes notused in previous studies. The astA and tsh genes were found in 1/26 and 20/26 chicken isolates, but not in any of the 15 transconju-gants from chicken strains, suggesting that these genes are not lo-cated on CTX-M plasmids, which are likely to have been the onlymobilised plasmids in the conjugation protocol used. Several genesthough, such as espJ, rpeA and toxB appeared to be located on theCTX-M plasmids, or on plasmids that were co-transferred withthe CTX-M plasmids.

The hierarchical clustering did show some clustering for iso-lates from different animal species, for example the cba, cma, iroN,iss, lpfA, mchF and tsh genes were all significantly more associatedwith chicken and turkeys isolates than cattle isolates at whilst theespP and mcmA genes were more strongly associated with cattleisolates than chicken or turkeys isolates and the lpFA and tsh geneswere present in isolates from chickens at significantly higher levelscompared to isolates from turkeys.

In previous studies, PFGE analysis of CTX-M positive E. coli fromchicken and turkey isolates showed them to be less than 80% sim-ilar, although within chicken or turkey isolates, some clustered atgreater than 80% similarity (Randall et al., 2011b). This suggestsdifferent populations of E. coli are present in chickens and turkeys.It was interesting to note in this study, the clustering of virulencegenes in chicken and turkey CTX-M E. coli isolates clustered someof the isolates as distinct from each other.

Previous workers have found that among healthy cattle isolates,170, 163, 161, 155, 112 and 84 isolates were positive for lpfA (99%),ehaA (95%), iha (94%), saa (91%), subAB (65%), and cdt-V (49%),respectively, while two were positive for eae (1.2%) and efa1

(1.2%) each (Wu et al., 2010b). Most of these genes were also foundin at least some the cattle isolates tested (Table 4) but not withsuch high percentages.

In conclusion, this study has provided a benchmark of charac-teristics of CTX-M isolates from cattle, chickens and turkeys inGreat Britain. The results will be of use when investigating andmaking temporal comparisons regarding the characteristics ofCTX-M isolates recovered from people and animals in Great Britainand in other countries. These findings are also of potential impor-tance in relation to the use of different antimicrobials such as thirdand four generation cephalosporins. The hierarchical clusteringclearly showed some distinct separation of many cattle isolatesfrom poultry isolates for both antibiotic resistance genes and viru-lence genes, and to a lesser extent the same was seen betweensome of the chicken and turkey isolates, suggesting to some extentdistinct populations of CTX-M positive E. coli isolates from cattle,chickens and turkeys.

Transparency declarations

None.

Acknowledgements

Many thanks to the Department for Environment, Food andRural Affairs (Defra) for funding this work. For chicken and turkeyisolates from healthy birds, samples were collected during surveyfunded by Defra and the Food Standards Agency and the authorsthank the survey team for provision of material for testing. Fund-ing: This work was funded by the Department for Environment,Food and Rural Affairs (project grants OD2023, OZ0328 andVM2207).

References

Anjum, M.F., Choudhary, S., Morrison, V., Snow, L.C., Mafura, M., Slickers, P., Ehricht,R., Woodward, M.J., 2011. Identifying antimicrobial resistance genes of humanclinical relevance within Salmonella isolated from food animals in Great Britain.Journal of Antimicrobial Chemotherapy 66, 550–559.

Badri, S., Fassouane, A., Filliol, I., Hassar, M., Cohen, N., 2009, Molecular typing ofEscherichia coli strains isolated from food in Casablanca (Morocco). Cellular andmolecular biology (Noisy-le-Grand, France) 55 Suppl, OL1132-1137.

Barton, B.M., Harding, G.P., Zuccarelli, A.J., 1995. A general method for detecting andsizing large plasmids. Analytical Biochemistry 226, 235–240.

Batchelor, M., Hopkins, K., Threlfall, E.J., Clifton-Hadley, F.A., Stallwood, A.D., Davies,R.H., Liebana, E., 2005. Bla (CTX-M) genes in clinical Salmonella isolatesrecovered from humans in England and Wales from 1992 to 2003.Antimicrobial Agents and Chemotherapy 49, 1319–1322.

Brinas, L., Moreno, M.A., Zarazaga, M., Porrero, C., Saenz, Y., Garcia, M., Dominguez,L., Torres, C., 2003. Detection of CMY-2, CTX-M-14, and SHV-12 beta-lactamasesin Escherichia coli fecal-sample isolates from healthy chickens. AntimicrobialAgents and Chemotherapy 47, 2056–2058.

Carattoli, A., Bertini, A., Villa, L., Falbo, V., Hopkins, K.L., Threlfall, E.J., 2005.Identification of plasmids by PCR-based replicon typing. Journal ofMicrobiological Methods 63, 219–228.

Carattoli, A., Garcia-Fernandez, A., Varesi, P., Fortini, D., Gerardi, S., Penni, A.,Mancini, C., Giordano, A., 2008. Molecular epidemiology of Escherichia coliproducing extended-spectrum beta-lactamases isolated in Rome, Italy. Journalof Clinical Microbiology 46, 103–108.

Cortes, P., Blanc, V., Mora, A., Dahbi, G., Blanco, J.E., Blanco, M., Lopez, C., Andreu, A.,Navarro, F., Alonso, M.P., Bou, G., Blanco, J., Llagostera, M., 2010. Isolation andcharacterization of potentially pathogenic antimicrobial-resistant Escherichiacoli strains from chicken and pig farms in Spain. Applied and EnvironmentalMicrobiology 76, 2799–2805.

Cottell, J.L., Webber, M.A., Coldham, N.G., Taylor, D.L., Cerdeno-Tarraga, A.M.,Hauser, H., Thomson, N.R., Woodward, M.J., Piddock, L.J., 2008. Completesequence and molecular epidemiology of IncK epidemic plasmid encodingblaCTX-M-14. Emerging Infectious Diseases 17, 645–652.

Eigner, U., Holfelder, M., Oberdorfer, K., Betz-Wild, U., Bertsch, D., Fahr, A.M., 2009.Performance of a matrix-assisted laser desorption ionization-time-of-flightmass spectrometry system for the identification of bacterial isolates in theclinical routine laboratory. Clinics in Laboratory Medicine 55, 289–296.

Ghanbarpour, R., Sami, M., Salehi, M., Ouromiei, M., 2010. Phylogenetic backgroundand virulence genes of Escherichia coli isolates from colisepticemic and healthybroiler chickens in Iran. Tropical Animal Health and Production 43, 153–157.

1150 M. Toszeghy et al. / Research in Veterinary Science 93 (2012) 1142–1150

Horton, R.A., Randall, L.P., Snary, E.L., Cockrem, H., Lotz, S., Wearing, H., Duncan, D.,Rabie, A., McLaren, I., Watson, E., La Ragione, R.M., Coldham, N.G., 2012. Fecalcarriage and shedding density of CTX-M extended-spectrum {beta}-lactamase-producing Escherichia coli in cattle, chickens, and pigs: implications forenvironmental contamination and food production. Applied andEnvironmental Microbiology 77, 3715–3719.

Identibac 2011-A. Pdf file giving full list of antibiotic resistance genes in the AMR-vemicro-array.(on-line, last accessed 30-4-12). <http://identibac.com/fileadmin/Media/Downloads/Genelist__E.coli_03_m.pdf>.

Identibac, 2011-B, Pdf file giving full list of virulence genes in the E. coligenotypingmicro-array (on-line, last accessed 30-4-12). <http://identibac.com/fileadmin/Media/Downloads/Genelist__E.coli_03_m.pdf>.

Janben, T., Schwarz, C., Preikschat, P., Voss, M., Philipp, H.C., Wieler, L.H., 2001.Virulence-associated genes in avian pathogenic Escherichia coli (APEC) isolatedfrom internal organs of poultry having died from colibacillosis. InternationalJournal of Medical Microbiology 291, 371–378.

Johnson, T.J., Wannemuehler, Y.M., Johnson, S.J., Logue, C.M., White, D.G., Doetkott,C., Nolan, L.K., 2007. Plasmid replicon typing of commensal and pathogenicEscherichia coli isolates. Applied and Environmental Microbiology 73, 1976–1983.

Kojima, A., Ishii, Y., Ishihara, K., Esaki, H., Asai, T., Oda, C., Tamura, Y., Takahashi, T.,Yamaguchi, K., 2005. Extended-spectrum-beta-lactamase-producing Escherichiacoli strains isolated from farm animals from 1999 to 2002: report from theJapanese Veterinary Antimicrobial Resistance Monitoring Program.Antimicrobial Agents and Chemotherapy 49, 3533–3537.

Lahey-clinic, 2012, ß-Lactamase Classification and Amino Acid Sequences for TEM,SHV and OXA Extended-Spectrum and Inhibitor Resistant Enzymes. (on-line,last accessed 30-4-12). <http://www.lahey.org/studies/webt.asp>.

Lau, S.H., Kaufmann, M.E., Livermore, D.M., Woodford, N., Willshaw, G.A., Cheasty,T., Stamper, K., Reddy, S., Cheesbrough, J., Bolton, F.J., Fox, A.J., Upton, M., 2008.UK epidemic Escherichia coli strains A-E, with CTX-M-15 beta-lactamase, allbelong to the international O25:H4-ST131 clone. Journal of AntimicrobialChemotherapy 62, 1241–1244.

Livermore, D.M., Canton, R., Gniadkowski, M., Nordmann, P., Rossolini, G.M., Arlet,G., Ayala, J., Coque, T.M., Kern-Zdanowicz, I., Luzzaro, F., Poirel, L., Woodford, N.,2007. CTX-M: Changing the face of ESBLs in Europe. Journal of AntimicrobialChemotherapy 59, 165–174.

Machado, E., Coque, T.M., Canton, R., Sousa, J.C., Peixe, L., 2008. Antibiotic resistanceintegrons and extended-spectrum {beta}-lactamases among Enterobacteriaceaeisolates recovered from chickens and swine in Portugal. Journal of AntimicrobialChemotherapy 62, 296–302.

Mshana, S.E., Imirzalioglu, C., Hossain, H., Hain, T., Domann, E., Chakraborty, T.,2009. Conjugative IncFI plasmids carrying CTX-M-15 among Escherichia coliESBL producing isolates at a University hospital in Germany. Biomedical CentralInfectious Diseases 9, 97.

Pomba, C., da Fonseca, J.D., Baptista, B.C., Correia, J.D., Martinez-Martinez, L., 2009.Detection of the pandemic O25-ST131 human virulent Escherichia coli CTX-M-15-producing clone harboring the qnrB2 and aac(6’)-Ib-cr genes in a dog.Antimicrobial Agents and Chemotherapy 53, 327–328.

Randall, L.P., Clouting, C., Horton, R.A., Coldham, N.G., Wu, G., Clifton-Hadley, F.A.,Davies, R.H., Teale, C.J., 2011a. Prevalence of Escherichia coli carrying extended-spectrum beta-lactamases (CTX-M and TEM-52) from broiler chickens andturkeys in Great Britain between 2006 and 2009. Journal of AntimicrobialChemotherapy 66, 86–95.

Randall, L.P., Wu, G., Phillips, N., Coldham, N., Mevius, D., Teale, C., 2011b, Virulencegenes in blaCTX-M Escherichia coli isolates from chickens and humans. Researchin Veterinary Science Jul 12. (Epub ahead of print).

Rodriguez-Siek, K.E., Giddings, C.W., Doetkott, C., Johnson, T.J., Fakhr, M.K., Nolan,L.K., 2005a. Comparison of Escherichia coli isolates implicated in human urinarytract infection and avian colibacillosis. Microbiology 151, 2097–2110.

Rodriguez-Siek, K.E., Giddings, C.W., Doetkott, C., Johnson, T.J., Nolan, L.K., 2005b.Characterizing the APEC pathotype. Veterinary Research 36, 241–256.

Sabate, M., Navarro, F., Miro, E., Campoy, S., Mirelis, B., Barbe, J., Prats, G., 2002.Novel complex sul1-type integron in Escherichia coli carrying bla(CTX-M-9).Antimicrobial Agents and Chemotherapy 46, 2656–2661.

Sheldon, T., 2010. Dutch doctors warn of dangers of overuse of antibiotics infarming. British Medical Journal 341, 5677.

Shiraki, Y., Shibata, N., Doi, Y., Arakawa, Y., 2004. Escherichia coli producing CTX-M-2beta-lactamase in cattle, Japan. Emerging infectious diseases 10, 69–75.

Smet, A., Martel, A., Persoons, D., Dewulf, J., Heyndrickx, M., Catry, B., Herman, L.,Haesebrouck, F., Butaye, P., 2008. Diversity of extended-spectrum beta-lactamases and class C beta-lactamases among cloacal Escherichia coli Isolatesin Belgian broiler farms. Antimicrobial Agents and Chemotherapy 52, 1238–1243.

Smet, A., Martel, A., Persoons, D., Dewulf, J., Heyndrickx, M., Cloeckaert, A., Praud, K.,Claeys, G., Catry, B., Herman, L., Haesebrouck, F., Butaye, P., 2009. Comparativeanalysis of extended-spectrum-{beta}-lactamase-carrying plasmids fromdifferent members of Enterobacteriaceae isolated from poultry, pigs andhumans: Evidence for a shared {beta}-lactam resistance gene pool? Journal ofAntimicrobial Chemotherapy 63, 1286–1288.

Tarrant, F., MacGowan, A.P., Walsh, T.R., 2007. Occurrence and current frequency ofCTX-M-type beta-lactamases from a regional hospital in the South West ofEngland. Journal of Antimicrobial Chemotherapy 59, 815–816.

Valverde, A., Canton, R., Garcillan-Barcia, M.P., Novais, A., Galan, J.C., Alvarado, A.,de la Cruz, F., Baquero, F., Coque, T.M., 2009. Spread of bla(CTX-M-14) isdriven mainly by IncK plasmids disseminated among Escherichia coliphylogroups A, B1, and D in Spain. Antimicrobial Agents and Chemotherapy53, 5204–5212.

Wu, G., Mafura, M., Carter, B., Lynch, K., Anjum, M.F., Woodward, M.J., Pritchard,G.C., 2010a. Genes associated with Escherichia coli isolates from calves withdiarrhoea and/or septicaemia. The Veterinary Record 166, 691–692.

Wu, Y., Hinenoya, A., Taguchi, T., Nagita, A., Shima, K., Tsukamoto, T., Sugimoto, N.,Asakura, M., Yamasaki, S., 2010b. Distribution of virulence genes related toadhesins and toxins in shiga toxin-producing Escherichia coli strains isolatedfrom healthy cattle and diarrheal patients in Japan. Journal of VeterinaryMedical science 72, 589–597.