Screening for enterotoxigenic Bacteroides fragilis in stool samples

Accepted Manuscript Screening for enterotoxigenic Bacteroides fragilis in stool samples Jacqueline I. Keenan, Alan Aitchison, Rachel V. Purcell, Rosie Greenlees, John F. Pearson, Frank A. Frizelle PII:

S1075-9964(16)30057-9

DOI:

10.1016/j.anaerobe.2016.05.004

Reference:

YANAE 1576

To appear in:

Anaerobe

Received Date: 3 March 2016 Revised Date:

21 April 2016

Accepted Date: 6 May 2016

Please cite this article as: Keenan JI, Aitchison A, Purcell RV, Greenlees R, Pearson JF, Frizelle FA, Screening for enterotoxigenic Bacteroides fragilis in stool samples, Anaerobe (2016), doi: 10.1016/ j.anaerobe.2016.05.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

1

Screening for enterotoxigenic Bacteroides fragilis in stool samples

2

Jacqueline I Keenana,#, Alan Aitchisona, Rachel V Purcella, Rosie Greenleesb, John F

4

Pearsonc, & Frank A Frizellea.

RI PT

3

5 6 7

a

8

b

9

c

SC

Microbiology Department , Canterbury Health Laboratories, Christchurch, New Zealand

Biostatistics and Computational Biology Unit, University of Otago Christchurch,

M AN U

10

Department of Surgery, University of Otago Christchurch, Christchurch, New Zealand

Christchurch, New Zealand

11 12 #

Dr Jacqui Keenan, [email protected], Tel +64 3 3640 570; Fax +64 3 3641 427

14

TE D

13

Key words: enterotoxigenic Bacteroides fragilis, colorectal cancer, bft gene, PCR, stool

16

samples

18 19 20 21

AC C

17

EP

15

22

1

ACCEPTED MANUSCRIPT

Abstract

24

Bacteroides fragilis is a commensal bacterium found in the gut of most humans, however

25

enterotoxigenic B.fragilis strains (ETBF) have been associated with diarrhoea and colorectal

26

cancer (CRC). The purpose of this study was to establish a method of screening for the

27

Bacteroides fragilis toxin (bft) gene in stool samples, as a means of determining if carriage of

28

ETBF is detected more often in CRC patients than in age-matched healthy controls. Stool

29

samples from 71 patients recently diagnosed with CRC, and 71 age-matched controls, were

30

screened by standard and quantitative PCR using primers specific for the detection of the bft

31

gene. Bacterial template DNA from stool samples was prepared by two methods: a sweep,

32

where all colonies growing on Bacteroides Bile Esculin agar following stool culture for 48 h

33

at 37°C in an anaerobic environment were swept into sterile water and heat treated; and a

34

direct DNA extraction from each stool sample. The bft gene was detected more frequently

35

from DNA isolated from bacterial sweeps than from matched direct DNA extractions. qPCR

36

was found to be more sensitive than standard PCR in detecting bft. The cumulative total of

37

positive qPCR assays from both sample types revealed that 19 of the CRC patients had

38

evidence of the toxin gene in their stool sample (27%), compared to seven of the age-matched

39

controls (10%). This difference was significant (P = 0.016). Overall, ETBF carriage was

40

detected more often in CRC patient stool samples compared to controls, but disparate

41

findings from the different DNA preparations and testing methods suggests that poor

42

sensitivity may limit molecular detection of ETBF in stool samples.

AC C

EP

TE D

M AN U

SC

RI PT

23

2

ACCEPTED MANUSCRIPT

Introduction

44

Bacteroides fragilis are anaerobic gram negative bacteria commonly found as part of the

45

human faecal microbiota. Infection most likely occurs in early childhood, and the inherent

46

ability of B. fragilis to evade the host immune response suggests that colonisation may persist

47

for life [1]. Whereas carriage of non-enterotoxigenic strains of B. fragilis may promote

48

mucosal health [2], some strains produce a toxin that is identified as causative agent of human

49

diarrhoea [3]. Colonic carriage of toxin-producing strains is also reportedly more prevalent in

50

people presenting with colorectal cancer (CRC) compared to healthy controls [4-6], fuelling

51

speculation that persistent carriage of these bacteria may “drive” colon carcinogenesis [7].

52

CRC rates of 37.3 per 100,000 New Zealanders are amongst the highest in the world [8]. Our

53

aim was to establish a method for screening stool samples for evidence of the B. fragilis toxin

54

(bft) gene, as a first step in determining if this potential marker of colon carcinogenesis is

55

detected more frequently in stool of CRC patients than in age-matched healthy controls. This

56

approach involved the processing of stool samples by two different methods: culture-based

57

amplification [9, 10] and direction extraction [11, 12], and testing each of these samples by

58

standard and quantitative PCR (qPCR).

61 62 63

SC

M AN U

TE D

EP

60

AC C

59

RI PT

43

64 65 66 3

ACCEPTED MANUSCRIPT

Materials & Methods

68

Study Participants. Seventy-one individuals, diagnosed with CRC between 2012 and 2014,

69

using standard endoscopic, histological or radiological criteria, provided a stool sample prior

70

to chemo-radiation and/or surgery. Patients found to have had pre-operative chemo-radiation

71

therapy or those with adenomas were not included in the study. Seventy-one stool samples

72

from a cohort of 125 collected from healthy volunteers who self-reported no evidence of

73

bowel problems at the time of sampling were age-matched to the 71 patients (Table 1). The

74

study was approved by the Upper South A Regional Ethics Committee of New Zealand, and

75

all participants provided written informed consent.

M AN U

SC

RI PT

67

76

Sample preparation. Each stool was processed by two different methods. The first was

78

culture-based amplification, where a loopful of sample was plated onto Bacteroides Bile

79

Esculin (BBE) agar (Fort Richard Laboratories, Auckland, NZ), a selective medium for

80

Bacteroides [13]. After 48 h anaerobic incubation at 37°C, all colonies were swept off the

81

plate into 500 µl of sterile water and heated for 10 min at 99°C. These “sweeps” were kept

82

frozen at -20°C until PCR was performed. The second method involved extraction of DNA

83

directly from 100 mg of each stool sample using a commercially available kit (Dynabeads

84

DNA DIRECTTM Universal extraction kit, Life Technologie AS, Oslo, Norway). This

85

“extract” was diluted 100-fold to reduce any inhibitory factors [9]. Extracts were also kept

86

frozen at -20°C.

EP

AC C

87

TE D

77

88

Standard PCR analysis of bacterial DNA. Each sample was screened for evidence of the bft

89

gene using a previously published primer set [14]. Briefly, each 10 µl PCR reaction mixture

90

contained 5 pmol of each primer, 200 nM dNTPs, 2 mM MgCl2, 1x enzyme buffer, 0.5 U 4

ACCEPTED MANUSCRIPT

HotFire Polymerase (Solis Biodyne, Tartu, Estonia), and 0.5 µl DNA template. The PCR

92

amplification program consisted of a 15 minute incubation at 95°C, followed by 35 cycles of

93

30 sec denaturation at 95°C, 30 sec annealing at 66°C, and 30 sec extension at 72°C, followed

94

by a final extension step of 2 min at 72°C. DNA from ETBF strain VPI 13784 was used as a

95

positive bft control. We have previously shown that the primer set used will detect all three bft

96

subtypes [15]. Products were visualised on a 1.5% agarose gel containing SYBRSafe

97

(Invitrogen, Carlsbad, CA, USA).

SC

RI PT

91

98

Quantitative PCR. Quantitative real-time polymerase chain reaction (qPCR) was carried out

100

to quantify levels of ETBF in stool samples using the LightCycler®480 thermocycler (Roche

101

Diagnostics, Indianapolis, IN, USA) using SYBR-green chemistry. Each 10 µl reaction

102

consisted of 25–35 ng of genomic DNA, 0.5 µM of each primer pair (Odamaki primers, Table

103

1), 5 µl of SYBR Green Master Mix (Roche Diagnostics), and 1.5 µl of water. Thermal

104

cycling conditions were as follows: 1 cycle of 95°C for 5 mins, followed by 60 cycles of 95°C

105

for 10 secs, 66°C for 10 secs and 72°C for 20 secs. Melting curves were obtained by heating

106

samples from 65°C to 97°C in increments of 0.11 °C/s with 5 acquisitions per °C. It was

107

assumed that each ETBF contained a single copy of the bft gene. The concentration of ETBF

108

was expressed as genome copy numbers by calculating the weight of one ETBF genome, as

109

described by Dolezel et al [16]. ETBF has an estimated genome size of 5.2 Mb, and a single

110

ETBF genome weighs approximately 5.32 fg (5.2 Mb/978Mb [one picogram of double

111

stranded DNA] = 0.00532 pg). Therefore, 1 ng of ETBF DNA contains approximately 187,970

112

copies of the genome (1000 pg/0.00532 pg). A standard curve was constructed by serially

113

diluting DNA of known concentration, extracted from purified ETBF (strain VPI 13784).

114

Concentrations of ETBF in stool samples were calculated using the standard curve following

AC C

EP

TE D

M AN U

99

5

ACCEPTED MANUSCRIPT

115

amplification of bft.

116

Reference strains. Three ETBF reference strains containing the three Bft subtypes (VPI

118

13784 [Bft-1] [17], 86-5443-2-2 [Bft-2] [18], and Korea 570 [Bft-3] [19] were generously

119

supplied by Professor Cynthia Sears, Baltimore, USA. DNA was extracted from colonies

120

following anaerobic culture of each strain on sheep blood agar (Fort Richard Laboratories).

RI PT

117

SC

121

Preparation of spiked faecal samples for PCR analysis. Serial ten-fold dilutions of a

123

suspension of toxigenic B. fragilis strain 86-5443-2-2, containing approximately 1 x 109

124

CFU/ml, were added to pre-weighed stool samples from two healthy volunteers negative for

125

carriage of ETBF. Standard and quantitative PCR was performed on DNA (sweeps and

126

extracts) from each spiked sample to assess the sensitivity of the assays.

M AN U

122

TE D

127

Statistical analysis. Sample counts were compared with Fisher’s exact test with odds-ratio

129

and 95% confidence intervals estimated by unconditional maximum likelihood. Logistic

130

regression including covariates for age (continuous in years) and gender (dichotomous) was

131

used to produce age and gender adjusted P-values for sample counts. All tests were two-sided

132

with a type II error rate of 0.05. Analysis was performed in R version 3.2.1 (Vienna, Austria).

AC C

EP

128

6

ACCEPTED MANUSCRIPT

Results

134

Incidence of bft gene in stool samples analysed by standard PCR. Stool samples from 71

135

patients recently diagnosed with CRC, and 71 age-matched controls were screened using

136

primers specific for the detection of the B. fragilis toxin (bft) gene in both bacterial sweeps and

137

direct DNA extractions. Standard PCR revealed that sweeps yielded more positive PCR results

138

(16/142 (11%)) than DNA extracted from stool samples (11/142 (8%)), indicating that the

139

sweep method is more sensitive (Table 2). This was confirmed by assaying serial dilutions of

140

stool samples spiked with ETBF. The threshold of detection was approximately 1 x 105 CFUs

141

per gram of stool when DNA was directly extracted from the stool. In contrast, the threshold

142

for DNA extracted from cultures grown from spiked stool was 1 x 104 CFUs per gram of stool.

143

More positive PCR results (11/71) were detected from bacterial sweeps of CRC patient

144

samples than from sweeps of age-matched controls (5/71) (16% vs 7%; P > 0.05). Ten CRC

145

patients were also found to be positive for the bft gene when the stool DNA extract was tested.

146

However, notably fewer positive samples were detected in the control group using this method

147

(1/71), which resulted in a significant difference between the number of positive patient and

148

control samples using this method of sample preparation (P = 0.046) (Table 2).

149

Only seven patient samples tested positive by both methods of sample preparation. Bacterial

150

sweeps from four patients were found to be positive while the corresponding extract was

151

negative, while three stool samples had evidence of the bft gene in DNA extract only, as

152

illustrated by Figure 1. The cumulative total of positive PCR assays from both sample types

153

was used to determine the presence of the bft gene in the two patient cohorts. We found that 14

154

of the individuals in the CRC group had evidence of the toxin gene in their stool sample

155

(20%), compared to five of the age-matched controls (7%). This difference was significant (P

156

= 0.046). Logistic regression analysis showed that age and gender did not materially affect the

AC C

EP

TE D

M AN U

SC

RI PT

133

7

ACCEPTED MANUSCRIPT

157

results.

158

Incidence of bft gene in stool samples analysed by qPCR. Both DNA samples from each

160

individual’s stool (bacterial sweep and Dynabead extraction) were then screened for evidence

161

of bft using qPCR. The bacterial sweeps yielded a total of 22 positives compared to the 16

162

positive samples detected by standard PCR, whereas 19 and 11 of the 142 extracts were

163

positive by quantitative and standard PCR, respectively. The increased sensitivity of qPCR

164

was confirmed by analysis of DNA extracted directly from serial dilutions of spiked stool

165

samples. The threshold for detecting bft from DNA extracted from cultures grown from spiked

166

stool was approximately 1 x 101 CFUs per gram of stool; a higher detection threshold was

167

observed when DNA was extracted directly from the same samples using Dynabead

168

technology (approximately 1 x 102 CFUs per gram of stool). However, despite the increased

169

sensitivity of qPCR, this assay did not detect the bft gene significantly more often in individual

170

samples when compared to standard PCR (P > 0.05).

171

All of the patients that tested positive for bft in either sample by standard PCR also tested

172

positive using qPCR. A further seven samples, negative by standard PCR, were additionally

173

detected using qPCR. Five of the seven samples were from CRC patients; the other two were

174

from age-matched controls (Table 2). Of note, all of these seven samples were detected

175

between 35 cycles and the 45 cycle cut-off, which was set to minimise the detection of non-

176

specific products; this is consistent with low abundance of bft in these samples. Interestingly,

177

six of the seven samples that tested positive by quantitative but not standard PCR were

178

bacterial sweeps generated following stool culture on selective media. These additional seven

179

samples were confirmed as true positives by sequencing of the qPCR product.

180

However, while qPCR detected bft in all the patients/controls positive for bft by standard PCR,

AC C

EP

TE D

M AN U

SC

RI PT

159

8

ACCEPTED MANUSCRIPT

a comparison of standard and qPCR findings on individual samples rather than per patient

182

revealed again the results were not consistent. Of the 11 CRC patient samples where bft was

183

detected in the sweeps by standard PCR, only ten were positive by qPCR. Likewise, qPCR

184

failed to detect bft in an extracted patient sample that was positive by standard PCR. These

185

results suggest sampling error due to low abundance of the target DNA in the samples.

186

Accordingly, the cumulative total of positive qPCR assays from both sample types was used to

187

determine the presence of the bft gene in patient and age-matched control cohorts. Nineteen of

188

the individuals in the CRC group had evidence of the toxin gene in their stool sample (27%),

189

compared to seven of the age-matched controls (10%). This difference was significant (P =

190

0.016). Logistic regression analysis showed that age and gender did not materially affect the

191

results (Table 2).

M AN U

SC

RI PT

181

192

ETBF genome copy number in stool samples analysed by qPCR. The generation of a

194

standard curve using DNA from a reference strain of ETBF allowed us to determine the copy

195

number of ETBF genomes in DNA samples using qPCR. The copy number ranged from

196

0.0005 to 32000 among bft-positive samples. There was no difference in relative abundance

197

between CRC samples (1955 ± 977) and controls (1706 ± 1193), or between sweep samples

198

and DNA extracts (2121 ± 1681 and 1305 ± 384, respectively).

AC C

EP

TE D

193

9

ACCEPTED MANUSCRIPT

Discussion

200

There is a growing awareness that certain species of gut bacteria may drive toxin- and/or

201

inflammation-mediated colorectal carcinogenesis [20]. We have been collecting evidence of a

202

significant association between long-term colonic carriage of ETBF and risk of low-grade

203

colonic dysplasia in our community. This led us to evaluate two methods for use as faecal

204

diagnostic tests for ETBF in individuals considered at risk of developing CRC.

205

Culture-based amplification was shown to provide greater sensitivity for ETBF detection than

206

DNA extracted directly from stool by using spiked stool samples, as reported elsewhere [9,

207

10]. Accordingly, selective enrichment of stool samples resulted in a slight increase in the

208

number of bft-positive samples compared to screening of DNA extracted directly from the

209

stool, suggesting an increase in the relative abundance of ETBF, if present. In this study solid

210

media was used to enrich for Bacteroides spp in the stool samples whereas a recent study

211

described the use of a defined broth culture medium [10]; either approach would likely serve

212

the same purpose.

213

Screening of stool DNA for bft using standard PCR methods showed a significant difference

214

between patient and control groups (P = 0.046). However, the detection rates of bft found in

215

the New Zealand patients were notably lower than those in the only other study to describe

216

increased incidence of ETBF in CRC patient stool samples [4]. To investigate whether the use

217

of standard PCR was responsible for under-reporting of bft-positivity in our study, the samples

218

were re-screened using a SYBR green qPCR assay with the same primers. This method was

219

shown to detect a signal at higher dilutions of spiked samples when compared to standard

220

PCR, similar to the TaqMan assay for ETBF recently described by Chen et al. [10]. Culture-

221

based amplification of ETBF served to further increase qPCR sensitivity: the number of CRC

222

samples with evidence of ETBF carriage increased from 14 (20%) to 19 (27%), and two

AC C

EP

TE D

M AN U

SC

RI PT

199

10

ACCEPTED MANUSCRIPT

further age-matched controls were found to be positive (from five (7%) to seven patients

224

(10%). The difference in detection rates between the groups was significantly different (P =

225

0.016) but still notably lower than that reported in Turkish patients [4].

226

Toprak and colleagues also used a culture-based system but instead used PCR to analyse

227

single colonies that were identified as B. fragilis by conventional methods [4]. Thus, the

228

differences between the two studies may reflect methodological as well as geographical

229

variation, particularly if the relative abundance of ETBF in samples is low. This is highlighted

230

by the finding that six of the seven stool samples in our study, positive for bft only by qPCR,

231

were sweep samples. This suggests a very low abundance of ETBF that was undetectable in

232

DNA extracted directly from stool samples. The viable cell number of B. fragilis in faecal

233

samples is reportedly 10- to 100-fold lower than other intestinal Bacteroides [21] and these

234

bacteria account for a very small proportion of the normal colonic microflora (0.1%-0.5%)

235

[22]; culturing of samples on selective media may have allowed ETBF to grow to numbers

236

detectable in subsequent qPCR.

237

In conclusion, this study aimed to determine if molecular screening of stool samples using bft

238

as a marker would detect ETBF more often in stool of CRC patients than in age-matched

239

healthy controls. Our results support the concept that ETBF may act to promote colon

240

carcinogenesis [20]. However, we stress that care needs to be taken with assay design and

241

interpretation following the observation of disparate findings in some stool samples where

242

DNA was prepared by different processes and/or tested by different methods. We conclude

243

that using qPCR with a combination of culture-based amplification and direct extraction is

244

likely to increase sensitivity.

AC C

EP

TE D

M AN U

SC

RI PT

223

11

ACCEPTED MANUSCRIPT

245

Conflict of interest declaration

246

The authors declare that they have no conflict of interest.

247

Acknowledgements

249

This study was funded by grants from Genesis Oncology Trust NZ and the Rotary Club of

250

Christchurch Sunrise.

RI PT

248

SC

251 252

M AN U

253 254 255 256

260 261 262 263 264

EP

259

AC C

258

TE D

257

265 266 267

12

ACCEPTED MANUSCRIPT

268

References

269

1

2

3

Microbiol Rev. 2009; 22: 349-369.

274 275

Sears CL. Enterotoxigenic bacteroides fragilis: A rogue among symbiotes. Clin

4

SC

273

Mazmanian SK, Round JL, Kasper DL. A microbial symbiosis factor prevents intestinal inflammatory disease. Nature. 2008; 453: 620-625.

272

RI PT

colon carcinogenesis. Infect Agent Cancer. 2014; 9: 31.

270

271

Keenan JI, Frizelle FA. Bacteria flying under the radar linking a bacterial infection to

Toprak NU, Yagci A, Gulluoglu BM, Akin ML, Demirkalem P, Celenk T, Soyletir G. A possible role of bacteroides fragilis enterotoxin in the aetiology of colorectal cancer.

277

Clin Microbiol Infect. 2006; 12: 782-786.

278

5

M AN U

276

Viljoen KS, Dakshinamurthy A, Goldberg P, Blackburn JM. Quantitative profiling of colorectal cancer-associated bacteria reveals associations between fusobacterium spp.,

280

enterotoxigenic bacteroides fragilis (etbf) and clinicopathological features of colorectal

281

cancer. PloS one. 2015; 10: e0119462.

282

6

TE D

279

Boleij A, Hechenbleikner EM, Goodwin AC, Badani R, Stein EM, Lazarev MG, Ellis B, Carroll KC, Albesiano E, Wick EC, Platz EA, Pardoll DM, Sears CL The

284

bacteroides fragilis toxin gene is prevalent in the colon mucosa of colorectal cancer

285

patients. Clin Infect Dis. 2015; 60: 208-215. 7

Tjalsma H, Boleij A, Marchesi JR, Dutilh BE. A bacterial driver-passenger model for colorectal cancer: Beyond the usual suspects. Nat Rev Microbiol. 2012; 10: 575-582.

287 288

AC C

286

EP

283

8

Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM,

289

Forman D, Bray F. Cancer incidence and mortality worldwide: Sources, methods and

290

major patterns in globocan 2012. Int J Cancer. 2015; 136: E359-386.

13

ACCEPTED MANUSCRIPT

291

9

enterotoxigenic bacteroides fragilis by pcr. J Clin Microbiol. 1997; 35: 2482-2486.

292 293

Pantosti A, Malpeli M, Wilks M, Menozzi MG, D'Ambrosio F. Detection of

10

Chen LA, Van Meerbeke S, Albesiano E, Goodwin A, Wu S, Yu H, Carroll K, Sears C. Fecal detection of enterotoxigenic bacteroides fragilis. Eur J Clin Microbiol Infect

295

Dis. 2015; 34: 1871-1877.

296

11

RI PT

294

Shetab R, Cohen SH, Prindiville T, Tang YJ, Cantrell M, Rahmani D, Silva J, Jr.

Detection of bacteroides fragilis enterotoxin gene by pcr. J Clin Microbiol. 1998; 36:

298

1729-1732. 12

bacteroides fragilis. J Clin Microbiol. 1998; 36: 3545-3548.

300 301

13

Livingston SJ, Kominos SD, Yee RB. New medium for selection and presumptive identification of the bacteroides fragilis group. J Clin Microbiol. 1978; 7: 448-453.

302

14

Odamaki T, Sugahara H, Yonezawa S, Yaeshima T, Iwatsuki K, Tanabe S, Tominaga

TE D

303

Zhang G, Weintraub A. Rapid and sensitive assay for detection of enterotoxigenic

M AN U

299

SC

297

T, Togashi H, Benno Y, Xiao JZ. Effect of the oral intake of yogurt containing

305

bifidobacterium longum bb536 on the cell numbers of enterotoxigenic bacteroides

306

fragilis in microbiota. Anaerobe. 2012; 18: 14-18. 15

gene relies on robust primer design. J Clin Microbiol. 2015; 54:239-240.

308 309

16

Dolezel J, Bartos J, Voglmayr H, Greilhuber J. Nuclear DNA content and genome size of trout and human. Cytometry Part A 2003; 51: 127-128; author reply 129.

310 311

Aitchison A, Frizelle FA, Keenan JI. PCR detection of bacteroides fragilis enterotoxin

AC C

307

EP

304

17

Moncrief JS, Duncan AJ, Wright RL, Barroso LA, Wilkins TD. Molecular

312

characterization of the fragilysin pathogenicity islet of enterotoxigenic bacteroides

313

fragilis. Infect Immun. 1998; 66: 1735-1739.

14

ACCEPTED MANUSCRIPT

314

18

Franco AA, Cheng RK, Goodman A, Sears CL. Modulation of bft expression by the

315

bacteroides fragilis pathogenicity island and its flanking region. Mol Microbiol. 2002;

316

45: 1067-1077. 19

Rhee KJ, Wu S, Wu X, Huso DL, Karim B, Franco AA, Rabizadeh S, Golub JE,

RI PT

317

Mathews LE, Shin J, Sartor RB, Golenbock D, Hamad AR, Gan CM, Housseau F,

319

Sears CL. Induction of persistent colitis by a human commensal, enterotoxigenic

320

bacteroides fragilis, in wild-type c57bl/6 mice. Infect Immun. 2009; 77: 1708-1718. 20

21

328

Moore WE, Holdeman LV. Human fecal flora: The normal flora of 20 japanesehawaiians. Appl Microbiol. 1974; 27: 961-979.

TE D

327

22

EP

326

Salyers AA. Bacteroides of the human lower intestinal tract. Ann Rev Microbiol. 1984; 38: 293-313.

324 325

M AN U

2014; 15: 317-328.

322 323

Sears CL, Garrett WS. Microbes, microbiota, and colon cancer. Cell Host Microbe.

AC C

321

SC

318

15

ACCEPTED MANUSCRIPT

329

Figure legend

330

Figure 1. B. fragilis toxin PCR amplification from patient stool samples using both DNA

332

extracted directly from stool (top row) and sweeps of bacteria cultured on BBE agar (bottom

333

row). Discrepancies between the two DNA preparations can be seen for samples 203 and 277.

334

DNA from ETBF strain VPI 13784 and water were used as positive and negative PCR controls

335

respectively.

AC C

EP

TE D

M AN U

SC

RI PT

331

16

ACCEPTED MANUSCRIPT

336

Tables

337

Table 1. Characteristics of CRC patients and controls

340

Age median (10%, 90%)

CRC

Control

71

71

72 (53,81)

64 (53, 80)

M AN U

Gender n (%)

SC

339

RI PT

338

Female

29 (41%)

39 (55%)

Male

42 (59%)

32 (45%)

343 344 345 346 347 348

EP

342

AC C

341

TE D

CRC, colorectal cancer; n, number of patients

349 350 351 17

ACCEPTED MANUSCRIPT

352

Table 2. Patient and control bft status by standard and qPCR

353 Extract (%)

11 (16)

10 (14)

Controls

5 (7)

1 (1)

P-value

0.183

0.005*

0.046*

P-value (adj)

0.112

0.023*

0.030*

bft qPCR CRC patients

15 (21)

Controls

7 (10)

P-value

0.103

P-value (adj)

0.098

M AN U

CRC patients

SC

bft standard PCR

Total (%)

RI PT

Sweep (%)

14 (20) 5 (7)

14 (20)

19 (27)

5 (7)

7 (10)

0.046*

0.016*

0.024*

0.015*

CRC, colorectal cancer; BFT, Bacteroides fragilis toxin; PCR, polymerase chain reaction; qPCR,

355

quantitative PCR; P-value (adj) adjusted for age and gender; * P-value < 0.05.

358 359 360 361

EP

357

AC C

356

TE D

354

362 363 364

18

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Highlights Molecular identification of enterotoxigenic B. fragilis (ETBF) in stool samples.

•

Comparison of different methods of DNA preparation and ETBF amplification.

•

Selective enrichment of Bacteroides increases identification of ETBF.

•

Quantitative PCR increases sensitivity over conventional PCR.

•

Use of multiple methods offers greatest detection rates.

AC C

EP

TE D

M AN U

SC

RI PT

•