ARTICLE IN PRESS Clinical Nutrition (2007) 26, 208–215
Available at www.sciencedirect.com
journal homepage: www.elsevierhealth.com/journals/clnu
ORIGINAL ARTICLE
Commensal bacteria do translocate across the intestinal barrier in surgical patients$ Bala S. Reddya, John MacFiea,, Marcel Gatta, Louissa Macfarlane-Smithb, Kalliopi Bitzopouloub, Anna M. Snellingb a
Combined Gastroenterology Research Unit, Scarborough Hospital, Woodlands Drive, Scarborough, YO12 6QL, UK Department of Biomedical Sciences, University of Bradford, Bradford, BD7 1DP, UK
b
Received 15 June 2006; accepted 30 October 2006
KEYWORDS Gut barrier function; Bacterial translocation; Commensal microflora; E. coli; DNA fingerprinting
Summary Background: The ‘‘gut origin of sepsis’’ hypothesis proposes that enteric bacteria may cause sepsis at distant extra-intestinal sites. Whilst there is much circumstantial evidence to support this hypothesis, there is no conclusive proof in humans. The nature of translocating bacteria remains unclear. The aim of this study was to establish the origin of Escherichia coli (E. coli) cultured from mesenteric lymph nodes (MLN) and determine if they belonged to any recognized pathotypes known to cause infections in humans. Methods: MLN and faecal samples were obtained from 98 patients undergoing colonic resection. E. coli were isolated from 9/98 MLN samples. DNA fingerprints of MLN isolates were compared with faecal isolates from the same patient. MLN isolates were tested for adherence and invasion using HEp-2 epithelial cells, and screened for DNA markers indicative of different pathotypes of E. coli. MLN isolates were also tested for internalisation into Caco-2 cells. Results: All the nine E. coli cultured from MLNs were found to have identical DNA fingerprints to at least one and often several E. coli isolates cultured from faecal samples of the same patient. 8/9 (89%) MLN isolates were weakly adherent and 2/9 (22.2%) were invasive. 8/9 (89%) tested negative for DNA markers. All the nine MLN strains were internalised by Caco-2 cells. Conclusion: This study confirms the gut origin of translocating bacteria. Most translocating E. coli do not belong to any recognised pathotype and are therefore normal commensal
$
Awarded the Moynihan prize, for the best scientific paper presented at the Association of Surgeons of Great Britain & Ireland, Annual scientific meeting, Glasgow, 2005 (Abstract: Br J Surg. 2005; 92(S1): 3) Corresponding author. Tel.: +44 1723 342654; fax: +44 1723 354031. E-mail address:
[email protected] (J. MacFie). 0261-5614/$ - see front matter & 2006 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved. doi:10.1016/j.clnu.2006.10.006
ARTICLE IN PRESS Commensal bacteria do translocate across the intestinal barrier in surgical patients
209
microflora. Our results suggest that bacterial translocation is more dependent upon the gut epithelium rather than the virulence properties of resident enteric bacteria. & 2006 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Introduction Bacterial translocation is defined as the passage of viable bacteria or endotoxins across the intestinal epithelium to mesenteric lymph nodes (MLN) and beyond, possibly contributing to sepsis at distant extra-intestinal sites.1 Most of the evidence supporting the gut as the origin of these bacteria is derived from animal and in vitro models.2 In humans, the identification of enteric bacteria as a frequent cause of sepsis is regarded as indirect evidence to propose the gastrointestinal tract as an undrained abscess of multiorgan failure3 and support of the gut origin of sepsis hypothesis.4 The nature of these translocating bacteria remains unclear. Some authors have suggested that these organisms must have specific properties that predispose to translocation or belong to an invasive pathotype.5,6 Others however, have demonstrated that non-pathogenic commensal bacteria were able to translocate across an intact gut epithelium.7,8 The aims of this study therefore were twofold; firstly to confirm that the Escherichia coli (E. coli) cultured from MLN had originated from the lumen of the gastrointestinal tract, which by definition would confirm translocation across the gut barrier and secondly, to determine whether or not these bacteria have any specific attributes that predispose to translocation, such as adherence, invasion and DNA markers indicative of pathogenicity.
Patients and methods This was a prospective study conducted with the approval of the Scarborough Hospital’s Local Research Ethics Committee. A consecutive series of 98 elective surgical patients undergoing colectomy were recruited. Patients were excluded if they had evidence of intra-peritoneal sepsis or contamination and if sampling of lymph nodes was impractical or clinically inappropriate. All patients received intravenous cefuroxime and metronidazole as prophylactic antibiotics on induction of anaesthesia. To confirm bacterial translocation, bacterial phenotypes obtained from MLN were compared to isolates obtained simultaneously from the lumen of the gastrointestinal tract. Virulence properties of bacteria were assessed by evaluating adherence properties, invasive abilities and by identifying DNA markers indicative of pathogenicity. In order to confirm that these bacteria could passively internalise into gastrointestinal epithelial cells, we used a 3 h co-incubation assay with a Caco-2 cell line.
Lymph node and stool sampling Our objective was to identify bacterial phenotypes from MLN isolates and compare them with isolates obtained
simultaneously from the lumen of the gastrointestinal tract. The methodology of MLN harvesting has been published previously.4,6,9,10 In all patients, a sample of faeces was obtained from within the lumen of the resected colectomy specimen immediately after it was removed from the operative field. These faecal samples were homogenised in tryptone soya broth containing 15% glycerol (v/v) as cryoprotectant and stored at70 1C until culture.
Isolation of bacteria from lymph node and stool samples Lymph node samples were thoroughly rinsed in saline before analysis to remove any surface contamination, and then homogenized in 0.5–1 ml of peptone water. The homogenate was inoculated onto Cystine-Lactose-Electrolyte Deficient (CLED) agar and Columbia blood agar (Oxoid, Basingstoke UK) for incubation with and without air. All plates were incubated at 37 1C for 48 h. Faecal samples were serially diluted in phosphate buffered saline (PBS), and then plated onto MacConkey agar for incubation at 37 1C for 24 h. Four E. coli colonies were selected at random from amongst those cultured from each faecal sample. The identity of MLN and faecal isolates morphologically resembling E. coli was confirmed using API 20E identification strips (BioMerieux, Lyon, France), and checking for b-glucoronidase activity on TBX agar (Oxoid, Basingstoke, UK) at 44 1C.
DNA fingerprinting To compare E. coli isolates, DNA fingerprints were generated by amplifying regions of DNA with consensus primers for Enterobacterial Repetitive Intergenic Consensus (ERIC) and Repetitive Extragenic Palindromic (REP) sequences as described by Versalovic et al.11 Amplification was performed in 25 ml reaction volumes comprising 1.5u Taq polymerase, 1x Thermo Pol II buffer (New England Biolabs, Hertfordshire, UK), 1.5 mM Mg2+, 1.25 mM dNTPs (Invitrogen, Paisley, UK), 10% dimethyl sulfoxide, and 50 pm of primers (REPIR-1, or ERICIR plus ERIC2).11 Amplification reactions involved an initial denaturation (95 1C, 7 min) followed by 30 cycles of denaturation (90 1C, 30 s), annealing (42 1C for REP, 52 1C for ERIC, 1 min) and extension (65 1C, 8 min), then a single final extension of 65 1C for 16 min. Amplification products were resolved by running 10 ml aliquots on 1.5% agarose gels containing 1 TAE buffer, and visualised after staining in ethidium bromide. When comparing patterns, strains were not considered to match each other if the profiles differed by more than two bands.
ARTICLE IN PRESS 210
Assay for bacterial adherence using HEp-2 epithelial cells E. coli strains cultured from the MLNs were tested for their ability to adhere to HEp-2 human epithelial cells using the method described by Cravioto et al. with minor modifications.12 HEp-2 epithelial cells were grown at 37 1C in 5% CO2 in RPMI 1640 liquid media (Gibco/Invitrogen Corp, Paisley, UK.) supplemented with 10% foetal calf serum and antibiotics (penicillin+streptomycin). For the assay, cells were grown overnight on 16 mm diameter glass coverslips placed at the bottom of the wells of 12-well tissue culture plates. After 3 washes in antibiotic-free media, 2 ml of fresh RPMI was added to each well, plus 50 ml of an overnight culture of E. coli (Luria–Bertani media). Incubation was for 3 h at 37 1C in 5% CO2. Unbound bacteria were removed by further washing with RPMI media, and then the monolayers were fixed with ethanol and stained using 10% Giemsa stain. Coverslips were mounted on glass slides then viewed under oil immersion and 1000 magnification. All adherence assays were done in duplicate and appropriate adherent and non-adherent E. coli control strains were included in each run.
Assay for bacterial invasion using HEp-2 cells E. coli isolates displaying adherence to HEp-2 cells were tested for their ability to invade the epithelial cells using the gentamicin protection assay as described by Donnenberg et al.13 Briefly, the adherence assay was performed as described above, but after the 3 h incubation, the media was replaced with 2 ml RPMI containing 50mg of gentamicin to kill any bacteria still external to the epithelial cells. After 1 h of further incubation, the cells were washed 4 times with sterile phosphate-buffered saline and lysed by the addition of 2 ml of 0.5% (wt/vol) sodium deoxycholate. Serial dilutions of the lysate were plated onto nutrient agar to enumerate the internalised bacteria. Strain LF82, an adherent invasive E. coli (AIEC) was used as a positive control14 and strain DH1 was used as a negative control in this assay.
Assay for internalisation of viable bacteria into Caco-2 cells E. coli isolates were tested for their ability to internalise into Caco-2, human colonic adenocarcinoma cells using the gentamicin protection assay described by Donnenberg et al.13 The cells were cultured in 12-well plates in Minimum Essential Medium (Sigma) supplemented with 10% foetal calf serum (Lablech), 1% non-essential amino acid solution (Sigma), 1% L-glutamine (Gibco/Invitrogen Corp, Paisley, UK) and antibiotics (penicillin and streptomycin), at 37 1C under 5% CO2 atmosphere. The media was changed on alternate days, and confluent monolayers were ready for use 13 days post-seeding. E. coli strains were grown overnight in nutrient broth. Cultures were centrifuged and the supernatant removed. Bacterial pellets were washed with PBS and resuspended in tissue culture medium and the optical density adjusted by dilution, to give 2 108 cells/ml. Culture medium was
B.S. Reddy et al. aspirated from the Caco-2 monolayers and these were washed twice with PBS. Two ml of culture medium, without antibiotics, was added to each well plus 150 ml of E. coli suspension. After 3 h incubation (37 1C, 5% CO2), monolayers were washed 3 times with PBS and 50 ml of 1 mg/ml gentamicin (Sigma) was added to each well to kill bacteria adhering to cell surfaces, but not those within the cells. Incubation was continued for a further 1 h. Wells were washed a further 3 times with PBS and Caco-2 cells lysed by adding 2 ml 5% Nadeoxycholate (Sigma). E. coli is resistant to this lysis treatment. Ten-fold dilutions of well contents were prepared and plated out on to nutrient agar. After overnight aerobic incubation at 37 1C, cfu/ml lysate was calculated to enumerate the internalised bacteria. E. coli LF82, an adherent invasive E. coli (AIEC) and D64 an enteroinvasive E. coli (EIEC) were used as positive controls.15 E. coli DH1 was used as a non-invasive, negative control. Assays were done in duplicate and average results recorded.
Colony hybridisation DNA probes indicative of key pathotypes of E. coli were used to screen the MLN isolates by colony hybridisation.16 These probes included CVD432 (Enteroaggregative E. coli, EAEC), ipaH (Enteroinvasive E. coli, EIEC), bfpA (Enteropathogenic E. coli, EPEC), and daaC (Diffusely Adherent E. coli, DAEC).17–19 Probes were generated from control strains by PCR and labelled using a non-radioactive DIG-DNA labeling kit (Boehringer Mannheim, Germany). Hybridisation of membranes with DNA probes was performed overnight at 42 1C in the presence of deionised formamide. Detection of positive hybridisation was done using a non-radioactive DIG detection kit (Boheringer Mannheim, Germany).
Results The median (interquartile range; IQR) age of the 98 patients recruited to the study was 66 (54–74) years. Sixty seven of these patients had colorectal malignancies and 31 had benign disease including inflammatory bowel disease (n ¼ 22), tubulovillous adenomas (n ¼ 4) and diverticular disease (n ¼ 5). E. coli were isolated from 9/98 MLN samples. Details of the nine patients who had E. coli translocation are summarized in Table 1. Further studies were conducted on the E. coli isolates obtained from their homogenised nodes.
DNA fingerprinting DNA fingerprints of the E. coli cultured from the MLNs and corresponding four faecal isolates from each of these nine patients obtained by using ERIC and REP primers are shown as adjacent tracts in Fig. 1. ERIC- PCR results showed that the E. coli isolates from MLNs were genetically similar to at least one and often all the 4 E. coli isolated from the corresponding patient’s faeces (Fig. 1(A)). REP- PCR results again re-confirmed these findings (REP-PCR fingerprint of MLN isolate and one matched faecal isolate from each patient are shown in the Fig. 1(B)).
ARTICLE IN PRESS Commensal bacteria do translocate across the intestinal barrier in surgical patients
Table 1
211
Details of patients who were positive for E.coli on MLN culture.
n
Age
Sex
Diagnosis
Procedure
Growth in MLN
T1 T2 T3 T4
54 65 76 51
F F F F
Crohn’s disease Crohn’s disease Ca. Rectum Crohn’s disease
Right hemicolectomy Right hemicolectomy Anterior resection Right hemicolectomy
T5
43
F
Crohn’s disease
Right hemicolectomy
T6
70
M
Tubulovillous adenoma
Sigmoid colectomy
E. coli E. coli E. coli E. coli, Enterococcus faecalis Bacteroides fragilis E. coli, Bacteroides fragilis E. coli,
T7
58
M
Ca. Sigmoid colon
Left hemicolectomy
T8
82
M
Ca. Sigmoid colon
Left hemicolectomy
T9
78
M
Ca. Ceacum
Right hemicolectomy
GroupD Streptococci E. coli, Pseudomonas aeruginosa E. coli, Klebseilla oxytoca, Enterococcus faecalis, Pseudomonas aeruginosa E. coli
M: male, F: female; Ca, Carcinoma; MLN, mesenteric lymph node.
Adherence to HEp-2 cells One E. coli isolate (from patient T7, Figs. 2 and 3) showed a classic diffuse adherence pattern and also tested positive with the daaC probe, indicating it was a Diffusely Adherent E. coli (DAEC).20 Seven isolates showed weak adherence with no described pattern (Fig. 2) and one isolate was non-adherent.
Invasion of bacteria into HEp-2 Cells Each of the MLN isolates was confirmed to be sensitive to the antibiotic gentamicin prior to the assay. Of the nine E. coli isolates obtained from MLNs, seven were found to be noninvasive. Only two isolates were found to be invasive in the gentamicin protection assay (isolates from patients T3 and T7, cfu counts shown in Table 2).
Internalisation of bacteria into Caco-2 cells All the nine MLN isolates internalized after 3 h co-incubation with Caco-2 cells as did the positive control strains. Noninvasive, negative control strain DH1 was not recovered from the Caco-2 cell lysates (Table 3).
Colony hybridisation Of the nine E. coli isolates obtained from MLNs, eight tested negative with all the four probes. As stated above, one isolate from patient T7 was positive for daaC indicating it was a Diffusely Adherent E. coli (DAEC).20
Discussion The results of DNA fingerprinting confirm that the E. coli isolates from the MLNs were identical to E. coli isolates
cultured from faecal samples obtained simultaneously from the lumen of the bowel confirming the gut origin of translocating bacteria in surgical patients. Adherence patterns, invasive abilities and screening for virulence markers indicate that eight out of these nine E. coli isolates were not recognised pathotypes known to cause infections in humans and therefore represent normal commensal microflora. All these isolates showed internalisation into Caco-2 cells suggesting that intestinal epithelial cell function is fundamental to the process of bacterial translocation. Our results suggest, therefore, that it is intestinal barrier function of the host rather than virulence properties of enteric bacteria that is important in the process of bacterial translocation. Whilst the number of patients with translocation studied in this study is small, the accuracy and precision of experimental techniques employed are such that it is improbable that our conclusions are incorrect. Dispersed repetitive DNA sequences (repetitive extragenic palindromic [REP] and enterobacterial repetitive intergenic consensus [ERIC]) in the genomes of the E. coli were amplified using primers corresponding to REP and ERIC sequences. This method is recognised as a valuable tool for the analysis of genomic diversity among E. coli strains and is known to have a high discriminatory power.21 Furthermore, all possible precautions were taken to avoid contamination and a strict study protocol was adhered to for the collection of specimens. In order to maximise the probability of including the dominant faecal clone in the fingerprinting analysis, we randomly sampled four E. coli colonies from each faecal sample. This is based on the observation that the probability of including at least one isolate of the dominant clone in a small random sample of colonies was calculated to be 99% for four randomly selected colonies from faecal coliform flora of humans.22 Previous reports suggest a higher incidence of translocation in patients with Crohn’s disease and is often attributed
ARTICLE IN PRESS 212
B.S. Reddy et al.
A
B M
T1
T2
L F L F
T3
L
T4
F
L F
T5
L F
T6
L
T7
F
L F
M
T8
M
L F
T9
L F
Figure 1 (A) ERIC fingerprints of MLN and matched faecal isolates. T1–T9: Patients with E. coli translocation. L: DNA fingerprint of E. coli from MLN. A, B, C and D: DNA fingerprints of four E. coli isolates from faecal sample from the same patient. M: Marker. * Faecal Isolates whose finger prints were not identical to the MLN isolate are marked in colour. (B) REP fingerprints of MLN and matched faecal isolates. T1–T9: Patients with E. coli translocation. M: Marker. L: DNA fingerprint of E. coli from MLN. F: DNA fingerprint of E. coli isolate from faecal sample of the same patient that was found to be identical.
to discontinuity in the gut epithelium seen during the acute exacerbation of this disease.23,24 It may be argued that four of the nine patients (44%) who had E. coli transocation to MLNs in this study had active Crohn’s disease necessitating colectomy and therefore may not represent true translocation. However, we consider this unlikely as we have previously demonstrated in a large series of surgical patients using multivariate logistic regression analysis that inflammatory bowel diseases do not independently predispose to bacterial translocation in humans.9 Furthermore the results of this study, which show that only nine of the 98 patients had translocation, support the fact that mucosal abnormalities as seen in colorectal cancers and inflammatory bowel
diseases are unlikely to disrupt the intestinal barrier function in humans. It has previously been shown that indigenous enteric bacteria vary in the rate and efficacy in which they translocate across the gut epithelium.5 Gram-negative facultative aerobic bacteria such as E. coli, Klebsiella pneumoniae, and Proteus mirabilis translocate more qfrequently in humans than obligate anaerobes and Gram-positive bacteria.6 This has lead to the suggestion that these bacteria might have characteristics that predispose to translocation. This possibility was investigated in this study by measuring properties of adherence and invasion together with the use of specific DNA probes to
ARTICLE IN PRESS Commensal bacteria do translocate across the intestinal barrier in surgical patients
213
Table 3 Internalisation assay after 3 h co-incubation with Caco-2 cells.
Figure 2 Diffuse adherence pattern to HEp-2 cells seen with isolate from patient T7.
Strain
CFU/ml lysate
DH1 (negative control) Lf82 (positive control) D64 (positive control) T1 T2 T3 T4 T5 T6 T7 T8 T9
0 1.0 106 1.6 104 1.0 105 3.0 104 1.16 104 2.5 103 1.0 104 1.0 103 5.0 104 2.0 104 9.16 103
CFU: colony forming units.
Figure 3 No obvious pattern of adherence to HEp-2 cells seen with isolate from patient T3.
Table 2 cells.
Results of bacterial invasion assay using HEp-2
E.coli strains
CFU/ ml lysate
DH1 (Negative control) LF82 (Positive control) T3 T7
0 9.2 104 1.14 104 6.2 104
CFU: colony forming units. *Rest of the E. coli isolates did not show invasion into HEp-2 cells.
identify common pathotypes known to cause infections in humans. The adherence patterns of E. coli to HEp-2 epithelial cells are a recognised technique used to classify strains causing infections in humans.20 Animal studies have demonstrated that conditions such as bowel manipulation, haemorrhagic shock and trauma result in quantitative and qualitative alterations in the luminal E. coli, leading to an increased adherence onto mucosal epithelium which may contribute
towards bacterial translocation.25 In this study, most E. coli isolates did have the ability to adhere, albeit weakly, however they did not demonstrate adherence patterns previously recognised to be indicative of pathogenic E. coli.20 All the nine strains of E. coli internalised into Caco-2 cells although 7/9 strains did not have any invasive properties on HEp-2 cells. Clarke et al. have recently shown that poorly invasive commensal bacteria, in situations of inflammatory stress exploit the lipid-raft mediated transcytoctic pathway to cross the intact epithelium using Caco-2 cell lines.8 Zareie et al. demonstrated a time-dependant decrease in the trans-epithelial resistance (TER)26 and Nazli et al. have demonstrated the significance of the enterocyte cytoskeletal changes in the process of translocation of commensal bacteria across metabolically stressed intestinal epithelial cells.27 These observations may possibly explain how noninvasive commensal bacteria could translocate across the gut barrier in humans. Using the colony hybridization technique, E. coli isolates obtained from MLNs were tested for DNA markers indicative of common pathotypes known to cause infections in humans. Only one isolate (from patient T7) tested positive for probe daaC. This particular isolate (from patient T7) also exhibited a diffuse adherence pattern on HEp-2 cells. Taken together, these findings confirm that this isolate belongs to the diffusely adherent group of E. coli (DAEC).20 The association of DAEC strains with infection in humans is not as strong as with other categories of pathogenic E. coli. While some are thought to cause diarrhoea, byfar most DAEC survive as harmless commensals in the human gut.28 Numerous reports from other centers7,8 and results from this study have confirmed that normal gut commensals can translocate across the intestinal barrier. This observation has two implications; firstly from the teleological point of view it begs the question as to why commensals translocate beyond the sub-mucosal lymphoid tissue, and secondly it emphasizes the putative importance of intestinal barrier function in the prevention of translocation, as distinctive from the role of pathogenic bacteria. It is now well recognised that intestinal
ARTICLE IN PRESS 214 dendritic cells sample live commensal bacteria from the lumen of the gut, in order to stimulate IgA.29 This may protect against mucosal penetration by these organisms and can be considered part of the normal immune response, a process known as oral tolerance.30 Translocation of commensals can therefore, be considered a normal physiological phenomenon. It is important to emphasize that not all bacteria or endotoxins that pass through the intestinal barrier cause sepsis. The majority of bacteria that manage to cross the intestinal mucosa are likely to be destroyed by the gut associated lymphoid tissue (GALT) before they reach the systemic circulation.31 Animal experiments suggest that bacteria-loaded dendritic cells can migrate to the MLN, but do not travel further to reach systemic lymphoid organs.32 This geographic compartmentalisation of commensal loaded dendritic cells explains how induction of IgA to commensals is confined to the mucosal immune system, preserving systemic ignorance and avoiding extra-intestinal inflammatory response.33 It is only when the host is immunocompromised or critically ill that gut becomes a mortar for systemic inflammation. Intestinal barrier function describes the well-recognized ability of the gastrointestinal tract to separate the potentially harmful luminal contents such as bacteria and endotoxins from the closely regulated internal milieu of the human body.34 Several factors are known to influence the barrier function in humans including surgical stress, haemorrhagic shock, hypoxia, burns, acute pancreatitis, total parenteral nutrition and alterations in the resident microflora.34 The results of this study would suggest that it is the breakdown of normal barrier function that permits translocation of non-pathogenic commensal bacteria and therefore indirectly emphasises the importance of preserving intestinal barrier function in surgical patients. Measures to preserve the integrity of the gut barrier function, such as maintaining adequate splanchnic perfusion, early enteral nutrition, probiotics and the use of gut-specific nutrients such as glutamine could reduce bacterial translocation and may prove beneficial in attenuating post-operative inflammatory stress response and decreasing septic morbidity in surgical patients.
Acknowledgements The authors wish to thank Prof. Darfeuille-Michaud, Faculte de Pharmacie, Clermont-Ferrond, France for the gift of E. coli LF82 used in this study, Dr. Zohreh Khodaii, Dept of Biomedical Sciences, University of Bradford for assistance with the Caco-2 assays and Mr Paul Sudworth, Head of biomedical sciences, Scarborough hospital for the technical support in conducting this study. Authorship 1. Bala S. Reddy designed the study, recruited the patients, performed data analysis, prepared manuscript and critically reviewed the manuscript. 2. John MacFie designed the study, preparation of manuscript and critical review of the manuscript. 3. Marcel Gatt prepared the manuscript and critically reviewed the manuscript.
B.S. Reddy et al. 4. Louissa Macfarlane-Smith and Kalliopi Bitzopoulou performed the microbiological tests and DNA fingerprinting for the study. 5. Anna M. Snelling supervised the microbiological tests and the DNA fingerprinting and helped in the preparation and critical review of the manuscript.
References 1. Berg RD, Garlington AW. Translocation of certain indigenous bacteria from the gastrointestinal tract to the mesenteric lymph nodes and other organs in the gnotobiotic mouse model. Infect Immun 1979;23:403–11. 2. Berg RD. Bacterial translocation from the gastrointestinal tract. Adv Exp Med Biol 1999;473:11–30. 3. Marshall JC, Christou NV, Meakins JL. The gastrointestinal tract. The ‘‘undrained abscess’’ of multiple organ failure. Ann Surg 1993;218:111–9. 4. MacFie J, O’Boyle CJ, Mitchell CJ. Gut origin of sepsis: a prospective study investigating the association between bacterial translocation, gastric microflora and septic morbidity. GUT 1999;45:223–8. 5. Cruz N, Lu Q, Alvarez X, Ditch EA. Bacterial translocation is species dependent: results using the human Caco-2 intestinal cell line. J Trauma 1994;36:612–6. 6. O’Boyle CJ, MacFie J, Mitchell CJ, Johnstone D, Sagar PM, Sedman PC. Microbiology of bacterial translocation in humans. GUT 1998;42:29–35. 7. Nazli A, Yang PC, Jury J, et al. Epithelia under metabolic stress perceive commensal bacteria as a threat. American Journal of Pathology 2004;164:947–57. 8. Clark E, Hoare C, Tanianis-Hughes J, Carlson GL, Warhurst G. Interferon gamma induces translocation of commensal E. coli across gut epithelial cells via a lipid raft-mediated process. Gastroenterology 2005;128:1258–67. 9. MacFie J, Reddy BS, Gatt M, Jain PK, Sowdi R, Mitchell CJ. Bacterial translocation in 927 patients over 13 years. Br J Surg 2006;93:87–93. 10. Sedman PC, MacFie J, Sagar PM, Mitchell CJ, May J, ManceyJones B, et al. The prevalence of gut translocation in humans. Gastroenterology 1994;107:643–9. 11. Versalovic J, Koeuth T, Lupski JR. Distribution of repetitive DNA sequences in eubacteria and application to fingerprinting of bacterial genomes. Nucleic Acids Res. 1991;19:6823–31. 12. Cravioto A, Gross R J, Scotland S M, Rowe B. An adhesive factor found in strains of E. coli belonging to the traditional enteropathogenic serotypes. Curr Microbiol 1979;3:95–9. 13. Donnenberg MS, Donohue-Rolfe A, Keusch GT. Epithelial cell invasion: an overlooked property of enteropathogenic E. coli (EPEC) associated with the EPEC adherence factor. J Infect Dis 1989;160:452–9. 14. Darfeuille-Michaud A, Boudeau J, Bulois P, Neut C, Glasser AL, Barnich N. High prevalence of Adherent-E. coli associated with ileal mucosa in Crohn’s disease. Gastroenterology 2004;127: 412–21. 15. Boudeau J, Glasser AL, Masseret E, Joly B, Darfeuille-Michaud A. Invasive ability of an E. coli strain isolated from the ileal mucosa of a patient with Crohn’s disease. Infect Immun 1999;67:4499–509. 16. Grustein M, Hogness DS. Colony hybridization: a method for the isolation of cloned DNAs that contain a specific gene.1975. Biotechnology 1992;24:117–21. 17. Baudry B, Savarino SJ, Vial P, Kaper JB, Levine MM. A sensitive and specific DNA probe to identify enteroaggregative E. coli, a recently discovered diarrheal pathogen. J Infect Dis 1990;161: 1249–51.
ARTICLE IN PRESS Commensal bacteria do translocate across the intestinal barrier in surgical patients 18. Okeke IN, Borneman JA, Shin S, Mellies JL, Quinn LE, Kaper JB. Comparative sequence analysis of the plasmid-encoded regulator of enteropathogenic E. coli strains. Infect Immun 2001;69:5553–64. 19. Martin HM, Campbell BJ, Hart CA, et al. Enhanced E. coli adherence and invasion in Crohn’s disease and colon Cancer. Gastroenterology 2004;127:80–93. 20. Torres AG, Zhou X, Kaper JB. Adherence of diarrheagenic E. coli strains to epithelial cells. Infect Immun 2005;73:18–29. 21. Baldy-Chudzik K, Neidbach J, Stosik M. rep-PCR fingerptinting as a tool for the analysis of genomic diversity in E. coli strains isolated from an aqueous/freshwater environment. Cell Mol Biol Lett 2003;8:793–8. 22. Lidin-Janson G, Kaijser B, Lincoln K, Olling S, Wedel H. The homogeneity of the faecal coliform flora of normal school-girls, characterized by serological and biochemical properties. Med Microbiol Immunol 1978;164:247–53. 23. Takesue Y, Ohge H, Uemura K, et al. Bacterial translocation in patients with Crohn’s disease undergoing surgery. Dis Colon Rectum 2002;45:1665–71. 24. Ambrose NS, Johnson M, Burdon DW, Keighley MR. Incidence of pathogenic bacteria from mesenteric lymph nodes and ileal serosa during Crohn’s disease surgery. Br J Surg 1984;71:623–5. 25. Thomas S, Kang G, Balasubramanian KA. Surgical manipulation of the intestine results in quantitative and qualitative alterations in luminal E. coli. Ann Surg 2004;240:248–54.
215
26. Zareie M, Riff J, Donato K, et al. Novel effects of the prototype translocating E. coli, strain C25 on intestinal epithelial structure and barrier function. Cell Microbiol 2005;7: 1782–97. 27. Nazli A, Wang A, Steen O, et al. Enterocyte cytoskeleton changes are crucial for enhanced translocation of nonpathogenic E. coli across metabolically stressed gut epithelia. Infect Immun 2006;74:192–201. 28. Tacket CO, Moseley SL, Kay B, Losonsky G, Levin MM. Challenge studies in volunteers using E. coli strains with diffuse adherence to HEp-2 cells. J Infect Dis 1990;162:550–2. 29. Macpherson AJ, Uhr T. Induction of protective IgA by intestinal dendritic cells carrying commensal bacteria. Science 2004;303: 1624–5. 30. Garside P, Mowat AM. Oral Tolerance. Semin Immunol 2001;13: 177–85. 31. Deitch EA. Bacterial translocation or lymphatic drainage of toxic products from the gut: What is important in human beings? Surgery 2002;13:241–4. 32. Sansonetti PJ. Phagocytosis of bacterial pathogens: implications in host response. Semin Immunol 2001;13:381–90. 33. Macpherson AJ. Uhr. Compartmentalisation of the mucosal immune response to commensal intestinal bacteria. Ann N Y Acad Sci 2004;1029:36–43. 34. Baumgart DC, Digness AU. Intestinal barrier function. Curr Opin Clin Nutr Metab Care 2002;5:685–94.
