1

In situ ESBL conjugation from avian to human E. coli during cefotaxime 1

administration 2

Annemieke Smet,1,2*

Geertrui Rasschaert,2 An Martel,

1 Davy Persoons,

2,3 Jeroen Dewulf,

3 Patrick 3

Butaye,1,4

Boudewijn Catry,5 Freddy Haesebrouck,

1 Lieve Herman,

2 and Marc Heyndrickx

2 4

5

Department of Pathology, Bacteriology and Avian Diseases, Faculty of Veterinary Medicine, 6

Ghent University, Salisburylaan 133, 9820 Merelbeke, Belgium1;

7

Institute for Agricultural and Fisheries Research, Unit Technology and Food, Brusselsesteenweg 8

370, B-9090 Melle, Belgium2; 9

Department of Reproduction, Obstetrics and Herd Health, Veterinary Epidemiology Unit, 10

Faculty of Veterinary Medicine, Ghent University, Salisburylaan 133, B-9820 Merelbeke, 11

Belgium3; 12

Department of Bacteriology and Immunology, CODA-CERVA-VAR, Groeselenberg 99, 1180 13

Brussels, Belgium4 14

Public Health & Surveillance, Scientific Institute of Public Health, J. Wytsmanstraat 14 1050 15

Brussels, Belgium5

16

17

* Corresponding author. Mailing address: Department of Pathology, Bacteriology and Avian 18

Diseases, Faculty of Veterinary Medicine, Ghent University, Salisburylaan 133, 9820 Merelbeke, 19

Belgium. Phone: 003292647435. Fax: 003292647494. 20

E-mail: Annemieke.Smet@Ugent.be 21

Keywords: Keywords: conjugation, extended-spectrum β-lactams, transfer frequency, continuous 22

flow model 23

2

Abstract 24

Aims: The behaviour of an E. coli isolate of broiler origin harbouring a blaTEM-52-carrying 25

plasmid (lactose negative mutant of B1-54, IncII group) was studied in an in situ continuous flow 26

culture system, simulating the human caecum and the ascending colon during cefotaxime 27

administration. 28

Methods and Results: Fresh faeces from a healthy volunteer, negative for cephalosporin-29

resistant E. coli, were selected to prepare inocula. The microbiota was monitored by plating on 30

diverse selective media, and a shift in the populations of bacteria was examined by 16S rDNA 31

PCR denaturing gradient gel electrophoresis. E. coli transconjugants were verified by plasmid 32

and pulsed field gel electrophoresis profiles (PFGE). The avian extended-spectrum β-lactamase 33

(ESBL) positive E. coli was able to proliferate without selective pressure of cefotaxime and E. 34

coli transconjugants of human origin were detected 24 h after inoculation of the donor strain. 35

Upon administration of cefotaxime to the fresh medium, an increase of the population size of E. 36

coli B1-54 and the transconjugants was observed. PFGE and plasmid analysis revealed a limited 37

number of human E. coli clones receptive for the blaTEM-52-carrying plasmid. 38

Conclusions: These observations provide evidence of the maintenance of an E. coli strain of 39

poultry origin and the horizontal gene transfer in the human commensal bowel microbiota even 40

without antimicrobial treatment. 41

Significance and Impact of Study: The fact that an E. coli strain of poultry origin might 42

establish itself and transfer its bla gene to commensal human E. coli raises public health 43

concerns. 44

45

46

3

Introduction 47

Escherichia coli, the best studied member of the bacterial family of Enterobacteriaceae, is a 48

Gram-negative rod-shaped bacterium that mostly occurs as a normal commensal in the intestinal 49

tract of animals and humans. Some strains are, however, important intestinal and extraintestinal 50

pathogens. Human and animal pathogenic E. coli are able to cause a spectrum of illnesses ranging 51

from self-limiting gastrointestinal infections to life threatening bacteraemia. 52

In human and veterinary medicine, β-lactams are extensively used. The predominant cause of 53

resistance to β-lactam antibiotics in Gram-negative bacteria is the production of β-lactamases. 54

These enzymes can be located on the chromosome or on plasmids (Sheldon 2005). 55

Extended-spectrum β-lactamases (ESBLs) (e.g. TEM-52), capable of efficiently hydrolyzing 56

extended-spectrum cephalosporins and associated with several treatment failures, disseminate 57

through plasmid transmission among the family of the Enterobacteriaceae. These bacteria are 58

part of the total microbiota of the digestive tract (Shah et al. 2004; Sheldon 2005; Smet et al. 59

2009). 60

The presence of cephalosporin-resistant E. coli in the intestinal tract of food-producing animals is 61

extensively described in many reports (Cloeckaert et al. 2007; Hasman et al. 2005; Kojima et al. 62

2005; Liu et al. 2007; Riaño et al. 2006). There is recently a high concern among clinical 63

microbiologists about ESBL-producing E. coli on poultry meat as a result of monitoring reports 64

in the EU, e.g. in the Netherlands and Belgium, showing high prevalence levels (Kojima et al. 65

2005; Liu et al. 2007; Riaño et al. 2006; Smet et al. 2009). The diversity of ESBLs (e.g TEM-52, 66

CTX-M-1, CTX-M-2) among these bacteria is high and they may act as a reservoir of ESBL 67

genes for pathogens causing disease in humans. This raises a potential public health concern 68

(Smet et al. 2008; Vollaard and Clasener 1994). The human digestive tract, colonized by a 69

4

complex microbiota, has been proposed as a suitable environment for horizontal transfer of 70

plasmids carrying antimicrobial resistance genes and contributes to the maintaining and the 71

dissemination of resistance (Netherwood et al. 1997). In the case of ESBL genes, it is not exactly 72

known how easy humans become carriers as the result of the consumption of poultry meat. The 73

determination of the transfer frequency between E. coli populations of human and poultry origin 74

is a first step in this study. 75

In vitro transfer of β-lactamase encoding (bla) genes is relatively easily to demonstrate between 76

individual strains and has been extensively studied (Smet et al. 2009). In vivo transfer of a bla 77

gene in the intestine of mice has been shown to occur (Moubareck et al. 2003; Schjorring et al. 78

2008) as also indications of transfer of bla genes between commensal members of 79

Enterobacteriaceae of human origin in clinical cases due to the treatment with a β-lactam 80

antimicrobial agent (Bidet et al. 2005; Karami et al. 2007). 81

More information is however needed on the transfer of specific antibiotic resistance genes from 82

animal or food derived strains to human strains in the gut. A recent controlled experiment where 83

healthy humans ingested a sulphonamide-susceptible recipient strain of human origin and a 84

sulphonamide-resistant donor strain of animal origin, showed the transfer of this sul gene to E. 85

coli of human origin in the gut. It has to be noted that the volunteer where this transfer took place, 86

was already colonized with commensal sulphonamide-resistant E. coli before the start of the 87

experiment. The reason that these volunteers also ingested an E. coli strain of animal origin was 88

to shed light on the risk associated with antimicrobial drug resistant E. coli in food. However, 89

evidence of transfer of resistance genes from an animal-derived E. coli to a recipient of human 90

origin was not provided in the human intestinal tract (Trobos et al. 2009). Another study reported 91

the transfer of vancomycin resistance from an E. faecium isolate of animal origin to an E. faecium 92

isolate from human origin in the intestines of humans (Lester et al. 2006). 93

5

These human controlled experiments requiring volunteers are possibly not without risk and 94

usually do not investigate the behaviour of the antibiotic resistant strains (donors and 95

transconjugants) with and without antibiotic treatment. Ethical and health constraints however 96

hamper experimental studies in humans. Therefore, the objective of this study was to investigate 97

whether an E. coli isolate of broiler origin could transfer its blaTEM-52-carrying plasmid to E. coli 98

of human origin in an in situ continuous flow culture system, simulating the human intestinal 99

tract in different situations (disturbed or stabilized gut microbiota) and this to highlight the risk 100

for public health associated with ESBL-producing E. coli in food. The in situ continuous flow 101

culture system underwent different culture scenarios that simultaneously allowed assessing the 102

impact of a 4 day course of a third-generation cephalosporine (cefotaxime). This study also 103

allowed us to measure to what extent the cephalosporin-resistant E. coli strains, both the donor 104

from broiler origin and the human transconjugants, could colonize in this fermentation system in 105

different situations (antibiotic treatment or not). Such data about transfer frequency and 106

colonization rate are necessary to perform risk assessment of human intake of food containing 107

antimicrobial drug resistant bacteria. 108

109

Materials and Methods 110

Strain preparation 111

E. coli strain B1-54, isolated from a faecal sample of broilers during an earlier survey (Smet et al. 112

2009) was used as donor, after selection of a lactose negative mutant to allow accurate 113

enumeration of transconjugants. The B1-54 isolate carried a blaTEM-52-carrying plasmid that 114

belonged to the incompatibility group IncI1 and had a size of > 100 kb. B1-54 also showed 115

phenotypic resistance to nalidixic acid and tetracycline but these resistance genes are not located 116

6

on the blaTEM-52-carrying plasmid (Smet et al. 2009). The methodology was slightly modified 117

from the one described by Varela and colleagues (1997). Briefly, mutants were isolated on 118

minimal succinate (0.2%) plates containing 3mM β-thio-o-nitrophenyl-β-D-galactopyranoside 119

(TONPG) (Sigma, Bornem, Belgium) plus 0.5mM isopropyl-β-D-thio-galactopyranoside (IPTG) 120

(Sigma, Bornem, Belgium) and incubated at 37°C. Mutants which grew on these plates were 121

streaked on 1% (w/v) melibiose MacConkey agar plates (plus 0.5mM IPTG). Red colonies were 122

picked and inoculated on 1% (w/v) lactose MacConkey agar plates. One white colony was 123

selected for this study, and this lactose negative mutant of E. coli B1-54 was grown overnight on 124

a MacConkey agar plate supplemented with 2.5 mg/l cefotaxime. 125

The presence of red colonies (lactose positive) on the MacConkey agar plates (Oxoid LTD, 126

Basingstoke, Hampshire, England) (with 2.5 mg/ ml cefotaxime) refers to transconjugants (of 127

human origin), while white colonies (lactose negative) refers to the lactose mutant of E. coli B1-128

54. 129

130

Fermentation system 131

Growth conditions for the caecum and ascending colon were mimicked in an in situ system, 132

which operated as a continuous system. The growth medium for the human caecum and the 133

ascending colon was prepared as described by MacFarlane et al. (2005). One 1.3 l glass 134

fermentation vessel, operated under the control of a BioFlo 110 unit (New Brunswick Scientific, 135

Edison, New Jersey, USA), was used in this study to simulate the microbial growth conditions of 136

the caecum and ascending colon. Via a peristaltic pump, fresh sterilized medium was added at a 137

constant rate of about 1.8 ml/min and spent culture liquid was wasted at the same rate to maintain 138

a constant working volume of 500 ml. Culture pH was maintained at 6.2, the temperature at 37°C 139

and moderate agitation at 200 rpm. Cultures were sparged by flushing the headspace of the vessel 140

7

with a nitrogen carbon dioxide gas mixture (80 and 20%, respectively) at a flow rate of 20 ml/min 141

to create anaerobic conditions. Fresh faeces from a healthy volunteer (adult), negative for 142

cephalosporin-resistant E. coli, were used to prepare inocula. The stools were homogenized with 143

a kitchen blender, divided in aliquots of 30 ml, supplemented with 15% glycerol and frozen at -144

80°C. After inoculation with 5 ml pooled stool content (corresponding to 1g feces), the fermentor 145

was operated in batch mode for 24 h. Continuous culture was started by switching on the 146

peristaltic pumps (Messens et al. 2009). The lactose mutant of E. coli B1-54 was grown 147

overnight on a MacConkey agar plate (supplemented with 2.5 mg/l cefotaxime). Two loopfull of 148

colonies were suspended in 10 ml PBS up to a final concentration of approximately 2 x 108 149

CFU/ml and added to the fermentation system at the indicated times (Fig. 1-3). Immediately after 150

inoculation of the lactose mutant of E. coli B1-54 (the donor strain), a sample was taken in order 151

to calculate the start number of the donor strain in the fermentation system. 152

153

Study design 154

Three experiments were carried out and in all three cases cefoxatime was supplied during 4 155

consecutive days. In a first negative control experiment (I) no ESBL+ donor strain (lactose 156

mutant of E. coli B1-54) was added in order to evaluate the stability of the model. 157

A second experiment (II) consisted of adding the ESBL+ donor strain simultaneously with the 158

intenstinal content at the iniation of the batch mode, 24 h prior to the peristaltic simulation. 159

During a final third experiment (III) the ESBL+donor strain was added 24 hours after the 160

peristaltic simulation had started. 161

162

163

164

8

Bacterial population dynamics 165

The cultivatable microbiota was monitored and counted by dilution plating on diverse selective 166

media (Oxoid LTD, Basingstoke, Hampshire, England): Reinforced Clostrial Medium (RCM) for 167

total anaerobes (37°C, 24h, anaerobic conditions); de Man, Rogosa and Sharpe medium (MRS) 168

for lactic acid bacteria (37°C, 24h, anaerobic conditions); Tryptone Soy Agar (TSA) for total 169

aerobes (37°C, 24h, aerobic conditions); and MacConkey Agar (with and without 2.5 mg/l 170

cefotaxime) for lactose negative donor strains of E. coli (of poultry origin, white colonies), 171

population of potential recipient E. coli (of human origin and cefotaxime susceptible, pink 172

colonies) and E. coli transconjugants (of human origin and cefotaxime resistant, pink colonies) 173

(37°C, 24h, aerobic conditions). 174

A shift in bacterial populations, due to a continuous administration of cefotaxime (2.5mg/l) 175

during 4 consecutive days, was examined by 16S (V3 region) rDNA PCR denaturing gradient gel 176

electrophoresis (PCR-DGGE), as described previously (Boon et al. 2000). 177

178

Transconjugant identification and horizontal gene transfer confirmation 179

A representative number of pink colonies were phenotypically identified as E. coli (Smet et al. 180

2008) after overnight aerobic incubation at 37°C on MacConkey agar with 2.5 mg/l cefotaxime. 181

The selected E. coli transconjugants were genotyped using XbaI-pulsed field gel electrophoresis 182

(PFGE) (Bertrand et al. 2006). In order to demonstrate whether these transconjugants obtained 183

the blaTEM-52-carrying plasmid, plasmid DNA was obtained as described by Takahashi and 184

Nagano (1984), the incompatibility (Inc) group was defined by the PCR-based replicon typing 185

(PBRT) method (Carattoli et al. 2005) and PCR and sequencing was performed for the 186

identification of the blaTEM-52 gene (Smet et al. 2009). These transconjugants were tested for 187

susceptibility to: ampicillin, amoxicillin-clavulanic acid, cefoxitin, ceftazidim, cefepime, 188

9

nalidixic acid, tetracycline, sulfonamides, trimethoprim, streptomycin, gentamicin and 189

chloramphenicol as earlier described (Smet et al. 2008).

190

At a given sample point, the average transfer frequency was estimated by dividing the number of 191

transconjugants per ml by the number of recipients (human E. coli) per ml. 192

193

Results 194

Experiment I: Caecal bacterial dynamics without ESBL+ donor strain 195

In a blank experiment (without administration of the lactose mutant of E. coli B1-54), simulating 196

the human caecum and the ascending colon, E. coli of human origin (and cefotaxime susceptible) 197

were able to stabilize at a population size of 6±0.2 log10 cfu/ml for 12 days in continuous culture 198

(Fig. 1). No spontaneous lactose negative E. coli mutants were detected during the experiment. 199

From day 12 onwards, the normal incubation medium was replaced by a stock containing 2.5 200

mg/l cefotaxime and maintained until day 15. Antimicrobial treatment gave an immediate 201

decrease of the population size (from 6 log10 to 3 log10 cfu/ml) for E. coli of human origin. At day 202

16, the medium with 2.5 mg/l cefotaxime was replaced again by the normal incubation medium. 203

After day 16, the population size of the cefotaxime susceptible E. coli of human origin increased 204

again and restored itself after discontinuation of the antimicrobial administration (Fig. 1). 205

206

Experiment II: Caecal bacterial dynamics with ESBL+ donor strain at the initiation of 207

experiment 208

The aim of the second experiment was to investigate whether the lactose mutant of E. coli B1-54 209

of broiler origin was able to maintain itself and transfer its blaTEM-52-carrying plasmid to E. coli 210

10

strains of human origin as well as the behaviour of the human transconjugants, in a fermentation 211

condition simulating a disturbance of the human microbial community. 212

The lactose mutant of E. coli B1-54 was added as donor strain together with the human stool 213

sample in the fermentation system, which was operated in batch mode for 24 h. Continuous 214

culture was then started by switching on the peristaltic pumps. Fluctuations in population size of 215

several bacterial groups were seen during the first five days (Fig. 2a). The lactose mutant of E. 216

coli B1-54 and the E. coli of human origin (and cefotaxime susceptible) could establish 217

themselves after 5 days at a population size of 7 ± 0.1 log10 cfu/ml and 7 ± 0.2 log10 cfu/ml, 218

respectively (Fig. 2a). Other average population sizes were 8 ± 1 log10, 9 ± 2 log10 and 9 ± 2 log10 219

cfu/ml for total aerobic bacteria, lactic acid bacteria and total anaerobes, respectively (Fig. 2b). 220

Twenty-four hours after administration of the lactose mutant of E. coli B1-54, transconjugants of 221

human origin, resistant to cefotaxime, were detected at a population size of 5 log10 cfu/ml (Fig. 222

2a). The average transfer frequency of ESBL-resistance was 2.5 x 10-3

at day 2 of the 223

fermentation experiment. From day 11 onwards, the normal incubation medium was replaced by 224

a stock containing 2.5 mg/l cefotaxime. The antimicrobial treatment was maintained until day 14. 225

During antimicrobial treatment (day 11-14), a decrease of the population size of the normal 226

cefotaxime susceptible human microbial community (E. coli (from 6 to 3 ± 0.04 log10) and total 227

aerobic bacteria (from 6 to 3 ± 0.1 log10)) was seen. On the contrary, an increase of the 228

population size of the lactose mutant of E. coli B1-54 (from 6 to 9 ± 0.3 log10) and of the human 229

transconjugants (from 5 to 8 ± 0.2 log10), both resistant to cefotaxime, was noted (Fig. 2a and 2b). 230

After stopping the antimicrobial treatment, the population size of the human E. coli community, 231

susceptible to cefotaxime, restored itself to the same level as before cefotaxime administration, 232

while the population size of the cefotaxime resistant lactose mutant of E. coli B1-54 and the 233

11

human transconjugants was maintained at the same high level as during antimicrobial treatment 234

(Fig. 2a). 235

The microbiota was disturbed as indicated by a different denaturing gradient gel electrophoresis 236

(DGGE) pattern compared to the fermentation condition without antibiotic (data not shown). At 237

the start of antimicrobial treatment, one band disappeared, but reappeared during antimicrobial 238

treatment. Also a few bands appeared more strongly during antimicrobial treatment. After 239

stopping the antimicrobial treatment, the DGGE pattern of the microbiota returned to the initial 240

condition, but the few more prominent bands which appeared during antimicrobial treatment, 241

remained in the pattern. Pure cultures of the lactose mutant of E. coli B1-54, of a human 242

transconjugant and of a human cefotaxime susceptible E. coli showed similar DGGE patterns 243

(data not shown). 244

245

Experiment III: Caecal bacterial dynamics with ESBL+ donor strain 24 hours after the 246

initiation of experiment 247

The aim of this third experiment was to verify whether the lactose mutant of E. coli B1-54 of 248

broiler origin was able to maintain itself and to transfer its blaTEM-52-carrying plasmid to E. coli 249

strains of human origin as well as the behaviour of the human transconjugants in a fermentation 250

condition in which the human microbial community was already dominantly present. 251

The human stool sample was added in the fermentation system and operated in batch mode for 24 252

h so that the human microbial community was predominantly present in the fermentation system. 253

The lactose mutant of E. coli B1-54 was supplemented after the switch to continuous culture 254

(after 1 day). The lactose mutant of E. coli B1-54 and the cefotaxime susceptible E. coli of 255

human origin could establish themselves at a population size of 3 ± 1 log10 and 7 ± 1 log10 cfu/ml, 256

respectively (Fig 3a). Other average population sizes were 7 ± 0.1 log10 cfu/ml for total aerobic 257

12

bacteria, lactic acid bacteria and total anaerobes (Fig. 3b). As seen in experiment II, 24h after 258

administration of the lactose mutant of E. coli B1-54, transconjugants of human origin were 259

detected. The population size (1 log10 cfu/ml) of the transconjugants was remarkably lower in 260

comparison with the second experiment. The average transfer frequency was 2.5 x 10-5

at day 3.

261

After 11 days of normal operation, the switch was made to medium stock with 2.5 mg/l 262

cefotaxime. Gradual replacement of the regular medium in the fermentor by medium 263

supplemented with 2.5 mg/l cefotaxime, resulted in a decrease of the E. coli of human origin (and 264

cefotaxime susceptible) (from 7 to 5 ± 0.04 log10) and an increase of the cefotaxime resistant 265

lactose mutant of E. coli B1-54 (from 3 to 5 ± 0.1 log10 cfu/ml) and the transconjugants of human 266

origin (from 2 to 5 ± 0.1 log10 cfu/ml) (Fig. 3a and 3b). At day 14, the feed was switched back to 267

the medium without 2.5 mg/l cefotaxime. After day 14, the population size of the cefotaxime 268

resistant lactose mutant of E. coli B1-54 and the transconjugants was maintained at the same high 269

level as during antimicrobial treatment, and the population size of the cefotaxime susceptible 270

human E. coli community gradually restored (Fig. 3a and 3b). 271

DGGE observations and characterisation showed highly comparable results with experiment II. 272

273

Verification of the E. coli transconjugants 274

A total of 60 E. coli transconjugants were retrieved from the MacConkey agar plates 275

(supplemented with 2.5 mg/l cefotaxime) during fermentation experiments II and III and were 276

verified by plasmid analysis (data not shown) and PFGE profiles. All isolated transconjugants 277

contained a large plasmid of >100 Kb from incompatibility group IncI1 and PCR and sequencing 278

revealed the presence of a blaTEM-52 gene on this plasmid (data not shown). Antimicrobial 279

susceptibility revealed that the E. coli transconjugants, besides showing an ESBL phenotype (i.e. 280

a marked synergistic effect between resistance to extended-spectrum cephalosporins and 281

13

amoxicillin with clavulanic acid), were susceptible to all other tested non-β-lactam antimicrobial 282

agents. PFGE revealed only two different fingerprint patterns among the isolated E. coli 283

transconjugants of human origin (Fig. 4). The most predominant fingerprint pattern was obtained 284

for 55 of the 60 isolates (91.7%) (lane 3 and 5 in Fig. 4), with a distinct fingerprint pattern 285

obtained for the other isolates. PFGE was also performed for a total of 20 random cefotaxime 286

susceptible E. coli strains isolated from the stool sample. Surprisingly, only one PFGE pattern 287

was seen and it was interesting to notice that this fingerprint pattern was identical to the 288

predominant fingerprint pattern found among the transconjugants. 289

290

Discussion 291

In this study, it was experimentally shown that a lactose mutant of ESBL-producing E. coli B1-292

54, isolated from a faecal sample of broilers, was able to proliferate in the fermentation system 293

simulating a part of the human gastrointestinal tract. The results suggest that the population size 294

of the donor strain depends on ongoing microbial competitions during the experiments that might 295

suffer from standardisation. Previous reports suggested that in order for antibiotic resistance 296

transfer to occur, high numbers (> 106 cfu/g) of both donor and recipient (compatible) strains 297

should coincide at the same time in the gut (Lester et al. 2006). In our study, E. coli 298

transconjugants of human origin were detected 24 h after inoculation of the donor strain B1-54 in 299

both experiments, even with the low concentration of the donor strain (3 log cfu/ml) in 300

experiment 3. Nevertheless, a substantial difference in transfer frequency (10-5

versus 10-3

) was 301

calculated in the two fermentation conditions. Thus, attention should be given to assess the 302

reproducibility of the fermentation conditions to allow accurate comparison of different 303

scenarios. Other limitations of the system are the fact that only the latter part of the digestive tract 304

14

is simulated while dynamics in resistance throughout the digestive tract (Catry et al. 2007) can 305

have an influence on transfer frequencies. Also factors like age, immunity, disease conditions 306

cannot be evaluated in the presented model (Graffunder et al. 2007; Levy 2002). The normal 307

human bacterial microbiota, dominantly present and acting as barrier for other bacteria, gives this 308

in situ model nevertheless advantages in mimicking the human gastrointestinal tract in 309

comparison to animal models. Also the possibility to measure the transfer frequency of 310

antimicrobial resistance determinants and to monitor the behaviour of the donor, recipient and 311

transconjugants are important advantages of this in situ model in comparison to animal models 312

(Schjorring et al. 2008). Despite the continuous input and output of the growth medium, 313

mimicking gastrointestinal transit, it has to be noted that conjugation of the blaTEM-52-carrying 314

plasmid still appeared even without selective pressure of an antimicrobial agent. This easy 315

transfer of the blaTEM-52-carrying plasmid to another recipient was not so surprising since the 316

transfer of this plasmid has also been investigated by in vitro liquid mating experiments between 317

individual strains in a previous report, for which the transfer frequency was 1.27 x 10-3

(Smet et 318

al. 2009). A comparison between these transfer frequencies reveals that the transfer of blaTEM-52 319

encoded resistance between E. coli strains can be influenced by the human gut microbiota under 320

stabilized conditions. Probably the lower transfer frequency of 10-5

is a more realistic estimation 321

of the normal situation. 322

Due to antimicrobial cefotaxime administration, both the lactose mutant of E. coli B1-54 of 323

broiler origin and the human E. coli transconjugants were able to proliferate quickly. In both 324

simulated conditions (disturbed and stabilized gut microbiota), an increase with 2-3 log of both 325

resistant E. coli populations was observed. After antimicrobial administration, the population size 326

of the normal microbial community, susceptible to cephalosporins, increased again to previously 327

obtained levels, while the cephalosporin resistant lactose mutant of E. coli B1-54 and the E. coli 328

15

transconjugants were able to maintain at the increased levels in the fermentation system. 329

Nevertheless, our experiments clearly support other reports (Cloeckaert et al. 2007; Persoons et 330

al. 2010) in which third-generation cephalosporins substantially enhance the proliferation of 331

ESBL-producing bacteria in the digestive tract. From the molecular cultivation independent 332

DGGE approach, it was also evident that after antimicrobial administration some bacterial 333

groups, other than the E. coli population and probably cefotaxime (inherently or acquired) 334

resistant, had increased in population size. These results indicate that antimicrobial agents may 335

not only select for resistant bacteria, but may also increase the potential of these resistant bacteria 336

as donors of resistance genes by increasing their population size. 337

The human E. coli transconjugants were genotyped using PFGE. Since E. coli has a polyclonal 338

population structure, it would be expected to find several PFGE patterns. However, only two 339

PFGE patterns were found, indicating that only two human E. coli clones received the blaTEM-52-340

carrying plasmid. Besides a dominant transconjugant clone which was genetically identical to the 341

dominant clone in the faecal E. coli population of the human donor, also a minor transconjugant 342

clone was selected with a genotype which was not picked up in the random selection of the faecal 343

E. coli isolates. The reason behind this phenomenon is unknown and needs further investigation. 344

To our knowledge, this is the first time that an in situ system simulating the human caecum and 345

the ascending colon was used to study the behaviour of an E. coli isolate of broiler origin as well 346

as of the human E. coli transconjugants after the transfer of an ESBL gene from a poultry-derived 347

E. coli to (a) recipient(s) of human origin. These observations provide evidence of the 348

maintenance of an E. coli strain of poultry origin and the horizontal gene transfer in the human 349

commensal bowel microbiota even without antimicrobial treatment. 350

In conclusion, the ability of an ESBL-producing E. coli strain of animal origin to be maintained 351

under normal or disturbed simulated physiological conditions in the human gut and to transfer the 352

16

ESBL-resistance gene residing on a large plasmid to commensal bacteria of human origin without 353

antimicrobial treatment raises public health concerns. In addition, cephalosporin treatment 354

increases potential health risk by increasing the population size of the cephalosporin resistant 355

microbial population. Further research is necessary to determine the risk of carriage in the human 356

gut of ESBL-producing microbiota from consumption of food products such as poultry with a 357

high prevalence of ESBL-producing E. coli and the potential risk of long-term carriage of a high 358

population level of this resistant microbiota in the human gut after antimicrobial cephalosporin 359

treatment. 360

361

Acknowledgements 362

This work was supported by a grant of Federal Public Service of Health, Food Chain Safety and 363

Environment (Grant number RT 06/3 ABRISK). 364

365

366

367

368

369

370

371

372

373

374

17

References 375

Bertrand, S., Weill, F.X., Cloeckaert, A., Vrints, M. Mairiaux, E., Praud, K., Dierick, K., 376

Wildemauve, C., Godard, C., Butaye, P., Imbrechts, H., Grimont, P.A. and Collard, J.M. 377

(2006) Clonal emergence of extended-spectrum beta-lactamase (CTX-M-2)-producing 378

Salmonella enterica serovar Virchow isolates with reduced susceptibilities to 379

ciprofloxacin among poultry and humans in Belgium and France (2000 to 2003). J Clin 380

Microbiol 44, 2897-2903. 381

Bidet, P., Burghoffer, B., Gautier, V., Brahimi, N., Mariani-Kurkdjian, P., El-Ghonemi, A., 382

Bingen, E. and Arlet, A. (2005) In vivo transfer of plasmid-encoded ACC-1 AmpC from 383

Klebsiella pneumoniae to Escherichia coli in an infant and selection of impermiability to 384

imipenem in K. pneumoniae. Antimicrob Agents Chemother 49, 3562-3565. 385

Boon, N., Goris, J., De Vos, P., Verstraete, W. and Top, E.M. (2000) Bioaugmentation of 386

activated sludge by an indigenious 3-chloroaniline-degradin Comamonas 387

testosteronistrain, 12gfp. Appl Environ Microbiol 66, 2906-2913. 388

Carattoli, A., Bertini, A., Villa, L., Falbo, V., Hopkins, K.L. and Threlfall, E.J. (2005) 389

Identification of plasmids by PCR-based replicon typing. J Microbiol Methods 63, 219-390

228. 391

Catry, B., Dewulf, J., Goffin, T., Decostere, A., Haesebrouck, F. and De Kruif, A. (2007) 392

Antimicrobial resistance patterns of Escherichia coli through the digestive tract of veal 393

calves. Microb Drug Resist 13, 147-150. 394

Cloeckaert, A., Praud, K., Doublet, A., Bertini, A. Carattoli, A., Butaye, P., Imbrechts, H., 395

Bertrand, S., Collard, J.M., Arlet, G. and Weill, F.X. (2007) Dissemination of an 396

extended-spectrum-β-lactamase blaTEM-52 gene-carrying IncI1 plasmid in various 397

18

Salmonella enterica Serovars isolated from poultry and humans in Belgium and France. 398

Antimicrob Agents Chemother 51, 1872-1875. 399

Graffunder, E.M., Preston, K.E., Evans, A.M. and Venezia, R.A. (2005) Risk factors associated 400

with extended-spectrum beta-lactamase producing organisms at a tertiary hospital. J 401

Antimicrob Chemother 56, 139-145. 402

Hasman, H., Mevius, D., Veldman, K., Olesen, I. and Aarestrup F.M. (2005) β-lactamases among 403

extended-spectrum β-lactamase (ESBL)-resistant Salmonella from poultry, poultry 404

products and human patients in the Netherlands. J Antimicrob Chemother 56, 115-121. 405

Karami, N. Martner, A., Enne, V.I., Swerkersson, S., Adlerberth, A. and Wold, A.E. (2007) 406

Transfer of an ampicillin resistance gene between two Escherichia coli strains in the 407

bowel microbiota of an infant treated with antibiotics. J Antimicrob Chemother 60,1142-408

1145. 409

Kojima, A., Ishii, Y. and Ishihara, K. (2005) Extended-spectrum-beta-lactamase-producing 410

Escherichia coli strains isolated from farm animals from 1999 to 2002: report from the 411

Japanese Veterinary Antimicrobial Resistance Monitoring Program. Antimicrob Agents 412

Chemother 49, 3533-3537. 413

Lester, C.H., Frimodt-Moller, N., Sorensen, T.L., Monnet, D.L. and Hammerum, A.M. (2006) In 414

vivo transfer of the vanA resistance gene from an Enterococcus faecium isolate of animal 415

origin to an E. faecium isolate of human origin in the intestine of human volunteers. 416

Antimicrob Agents Chemother 50: 596-599. 417

Levy, S.B. (2002) The 2000 Garrod lecture: factors impacting on the problem of antibiotic 418

resistance. J Antimicrob Chemother 49, 25-30. 419

Liu, J.H., Wei, S.Y., Ma, J.Y., Zeng, Z.L., Lü, D.H., Yang, G.X. and Chen, Z.L. (2007) Detection 420

and Characterization of CTX-M and CMY-2 β-lactamases among Escherichia coli 421

19

isolates from farm animals in Guangdong province of China. Int J Antimicrob Agents 29, 422

576-581. 423

MacFarlane, S., Woodmansey, E.J. and Macfarlane, G.T. (2005) Colonization of mucin by 424

human intestinal bacteria and establishment of biofilm communities in a two stage 425

continuous culture system. Appl Environ Microbiol 71, 7483-7492. 426

Messens, W., Goris, J., Dierick, N., Herman, L. and Heyndrick, M. (2009) Inhibition of 427

Salmonella Typhimurium by medium-chain fatty acids in an in vitro simulation of the 428

porcine cecum. Vet Microbiol doi:10.1016/j.vetmic.2009.08.002. 429

Moubareck, C., Bourgeois, N., Courvalin, P. and Doucet-Populaire, F. (2003) Multiple antibiotic 430

resistance gene transfer from animal to human enterococci in the digestive tract of 431

gnotobiotic mice. Antimicrob Agents Chemother 47: 2993-2996. 432

Netherwood, T., Bowden, R., Harrison, P., O’Donnell, A.G., Parker, D.S. and Gilbert, H.J. 433

(1999) Gene transfer in the gastrointestinal tract. Appl Environ Microbiol 65, 5139-5141. 434

Persoons, D., Dewulf, J., Smet, A., Herman, L., Heyndrickx, M., Martel, A., Catry, B., Butaye, P. 435

and Haesebrouck, F. (2010) Prevalence and persistence of antimicrobial resistance in E. 436

coli from broiler chickens. Microb Drug Res 16, 67-74. 437

Riaño, I., Moreno, M.A., Teshager, T., Saenz, Y., Dominguez, L. and Torres, C. (2006) Detection 438

and characterization of extended- spectrum beta-lactamases in Salmonella enterica strains 439

of healthy food animals in Spain. J Antimicrob Chemother 58, 844-847 440

Schjorring S., Struve, C. and Krogfelt, K.A. (2008) Transfer of antimicrobial resistance plasmids 441

from Klebsiella pneumoniae to Escherichia coli in the mouse intestine. J Antimicrob 442

Chemother 62, 1086-1093. 443

20

Shah, A.A., Hasan, F., Ahmed, S. and Hameed, A. (2004) Characteristics, epidemiology and 444

clinical importance of emerging strains of gram-negative bacilli producing extended-445

spectrum β-lactamases. Res Microbiol 155, 409-421. 446

Sheldon, A.T. (2005) Antibiotic resistance: a survival strategy. Clin Lab Sci 18, 170-180. 447

Smet A., Martel, A., Persoons, D., Dewulf, J., Heyndrickx, M., Cloeckaert, A., Praud, K., 448

Claeys, G., Catry, B., Herman, L., Haesebrouck, F. and Butaye, P. (2009) Comparative 449

analysis of Extended-spectrum-β-lactamase (ESBL)-carrying plasmids from different 450

members of Enterobacteriaceae isolated from poultry, pigs and humans: evidence for a 451

shared β-lactam resistance gene pool? J Antimicrob Chemother 63, 1286-1288. 452

Smet, A., Martel, A., Persoons, D., Dewulf, J., Heyndrickx, M., Catry, B., Herman, L., 453

Haesebrouck, F. and Butaye, P. (2008) Diversity of extended-spectrum β-lactamases and 454

class C β-lactamases among cloacal Escherichia coli in Belgian broiler farms. Antimicrob 455

Agents Chemother 52, 1238-1243. 456

Takahashi, S. and Nagano, Y. (1984) Rapid procedure for isolation of plasmid DNA and 457

application to epidemiological analysis. J Clin Microbiol 20, 608-613. 458

Trobos, M., Lester, C.H., Olsen, J.E., Frimodt-Moller, N. and Hammerum, A.M. (2009) Natural 459

transfer of sulphonamide and ampicillin resistance between Escherichia coli residing in 460

the human intestine. J Antimicrob Chemother 63, 80-86. 461

Varela, M.F., Brooker, R.J. and Wilsons, T.H. (1997) Lactose carrier mutants of Escherichia coli 462

with changes in sugar recognition (lactose versus melibiose). J Bacteriol 179, 5570-5573. 463

Vollaard, E.J. and Clasener, H.A. (1994) Colonization resistance. Antimicrob Agents Chemother 464

38, 409-414. 465

466

467

21

Figure Legends 468

Figure 1 Influence of antimicrobial treatment on the microbial community of the human caecum 469

and the ascending colon (blank experiment). Start continuous culture at day 1. The start and end 470

of the antimicrobial treatment (from day 12 to day 15) is indicated by the box. 471

472

Figure 2 Administration of E. coli B1-54 before the microbial community of the human ceacum 473

and the ascending colon reached a steady state. Start continuous culture at day 1. The start and 474

end of the antimicrobial treatment (from day 11 to day 14) is indicated by the box. 475

476

Figure 3 Administration of E. coli B1-54 after the microbial community of the human ceacum 477

and the ascending colon reached a steady state. Start continuous culture at day 1. The start and 478

end of the antimicrobial treatment (from day 11 to day 14) is indicated by the box. 479

480

481

Figure 4 PFGE profiles of the donor, recipient and the E. coli transconjugants (experiment 2). M: 482

marker; 1 and 2: E. coli of human origin (susceptible to cefotaxime; recipient); 3: E. coli 483

transconjugant (dominant PFGE fingerprint pattern, experiment 2 before antimicrobial 484

treatment); 4: E. coli transconjugant; 5: E. coli transconjugant (dominant PFGE fingerprint 485

pattern, experiment 3, before antimicrobial treatment); 6: E. coli B1-54 (donor). 486

22

Figure 1 487

1,00E+00

1,00E+01

1,00E+02

1,00E+03

1,00E+04

1,00E+05

1,00E+06

1,00E+07

1,00E+08

1,00E+09

1,00E+10

0 1 2 3 4 5 6 8 12 13 15 16 17 18 19 20 21 24 27

time after inoculation (days)

CF

U/m

l

E. coli (human)

total aerobic bacteria

488

489

490

491

492

493

494

495

496

497

498

499

500

501

502

503

504

23

Figure 2 505

(a) 506

1,00E+00

1,00E+01

1,00E+02

1,00E+03

1,00E+04

1,00E+05

1,00E+06

1,00E+07

1,00E+08

1,00E+09

1,00E+10

1,00E+11

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

time after inoculation (days)

CF

U/m

l E. coli (human)

B1-54

transconjugants

507

508

509

510

511

512

513

514

515

516

517

518

519

520

521

522

523

24

(b) 524

1,00E+00

1,00E+01

1,00E+02

1,00E+03

1,00E+04

1,00E+05

1,00E+06

1,00E+07

1,00E+08

1,00E+09

1,00E+10

1,00E+11

1,00E+12

0 1 2 3 4 6 7 8 9 10 11 12 13 14 15 16 17

time after inoculation (days)

CF

U/m

l

lactic acid bacteria

anaerobes

total aerobic bacteria

525

526

527

528

529

530

531

532

533

534

535

536

537

538

539

540

541

25

Figure 3 542

(a) 543

544

1,00E+00

1,00E+01

1,00E+02

1,00E+03

1,00E+04

1,00E+05

1,00E+06

1,00E+07

1,00E+08

0 1 2 3 4 6 7 8 9 10 11 12 13 14 16 17

time after inoculation (days)

CF

U/m

l E. coli (human)

B1-54

transconjugants

545

546

547

548

549

550

551

552

553

554

555

556

557

558

559

26

(b) 560

1,00E+00

1,00E+01

1,00E+02

1,00E+03

1,00E+04

1,00E+05

1,00E+06

1,00E+07

1,00E+08

0 1 2 3 4 6 7 8 9 10 11 12 13 14 16 17

time after inoculation (days)

CF

U/m

l lactic acid bacteria

anaerobes

total aerobic bacteria

561

562

563

564

565

566

567

568

569

570

571

572

573

574

575

576

577

27

Figure 4 578

M 1 2 579

580

581

582

583

584

585

M 3 4 5 6