PlEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Proj(!ct Report Sheet

Surname or Family name: Grehan

First name: Martin Othe£ name/s: James Abbreviation for degree as given in the University calendar: PhD

School: Biote<;hnology & Faculty:· Science Title: Biomolecular Sciences Molecular studies of the human and murine intestinal micro biota.

Studies of the diversity of bacterial species in the human colon have recently been based on 16S ribosomal RNA sequences using molecular methods such as PCR and denaturing gradient gel electrophoresis (DGGE). However due to the limited number of published primer sets, an incomplete representation of faecal bacteria is obtained with this approach. This thesis developed molecular methods for detecting Helicobacter species in faecal samples and examined whether Helicobacter spp. were associated with inflammatory bowel disease. A PCR-DGGE assay for the detection and speciation of lower bowel helicobacters was developed and validated in mice. In a human study of 43 controls and 27 inflammatory bowel disease cases with nested PCR, there was no significant difference in the prevalence of Helicobacter spp. DNA found in IBD cases and controls. In all positive subjects the amplified DNA was identical to Helicobacter pylori, suggesting that wash-down of gastric organisms was responsible for the positive results. In order to increase the number of bacterial species that were represented by PCR- DGGE, primer sets were also developed for the three major groups of anaerobes present in human faeces; the Bacteroides-prevotella group, Clostridium coccoides group and Clostridium leptum subgroup. Using these assays it was shown that the bacterial species present in murine faecal and caecal samples were nearly identical. The primers were also applied to show that preparation of mice and humans with antibiotics and cathartics reduced colonisation resistance such that the bacterial populations of the large intestine could be altered. In particular, the administration of a suspension of donated faeces to a human or murine subject that had been prepared with antibiotics and cathartics resulted in a hybrid microbiota in the colon of that individual. In humans, the hybrid microbiota was predominantly derived from the infused faecal suspension (rather than the subjects own baseline microbiota) and was largely stable over the following 24 weeks. The new PCR-DGGE primer sets in combination produced a more complete representation of bacterial species diversity than existing universal primers. Using these assays it was shown for the first time that the composition of the adult colonic microbiota can be altered. Declaration relating to disposition of project report/thesis

1 am. fully aware of the policy of the University relating to the retention and use of higher degree project· reports and thes_e7, namely that th~ UniversitY retains the copies submitted for examination and is free to allow them to be consulted or borrowed. Subject to the prov1s1ons of the Copynght Act 1968, the University may issue a proj.ect report or thesis in whole or in part, in photostat or microfilm or other copying medium.

1 also author-ise the publication by University Microfilms of 1- 350 word abstract in Dissertation Abstracts International (applicable to doctorates only).

Signature ...... I U I ...... ?.12 .. t~ . Y: ...... U\ Witness Date

The University recognises that there may be exceptional circumstances requiring restrictions on_copying o~ c_onditions on use_. Requ.ests for r~striction for a period of up to 2 years must be made in writing to the Registrar. Requests for a longer penod of restnct10n may be c.ons1de~ed m except10n?l circumstances if accompanied by a letter of support from the Supervisor or Head of School. Such requests must be subm1tted wtth the thests/project re art. FOR OFFICE USE ONLY Date of completion of requirements for Award: CJ 4 /I j{f) Re!!:istrar ann DPnutv Princioal

THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS

N: \ FlORENCE\ABSTRACT

Molecular studies of the human and murine intestinal microbiota.

Martin James Grehan

A thesis submitted for the degree of Doctor of Philosophy

School of Biotechnology and Biomolecular Sciences The University of New South Wales Sydney, Australia

March2004 UNSW 11 FEB 2005 LIBRARY Acknowledgments

I would like to sincerely thank each of the following people for their support and encouragement during this project.

Most importantly, Associate Professor Hazel Mitchell for her support during the past four years and for being a particularly good listener. Your capacity for hard work and ability to balance this with family life is inspirational.

Professor Adrian Lee for his enthusiasm and for generously offering me the opportunity to join his research team.

Dr Stephen Danon for his patient teaching, friendship and good humour.

Dr Jani O'Rourke for generously sharing her great wealth of experience and common-sense.

To all of my co-workers in the lab in particular Andrew Harris, Gauri Tamotia, Martin Wiseman, Daniel Sieveking, Li Zhang, Thosapom Coldham and Mai Dung Ha- thank you for your friendship.

John Wilson and Kathleen Kimpton for their friendship and animal handling expertise.

And finally my wife Penny for her gentle persuasion and encouragement during these studies, and my father Peter for his anecdotes and perspective. Certificate of Originality

I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, nor material which to a substantial extent has been accepted for publication for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgment is made in the thesis. Any contribution made to research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis.

I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.

(Signed) ABSTRACT

Studies of the diversity of bacterial species in the human colon have recently been based on 16S ribosomal RNA sequences using molecular methods such as PCR and denaturing gradient gel electrophoresis (DOGE). However due to the limited number of published primer sets, an incomplete representation of faecal bacteria is obtained with this approach.

This thesis developed molecular methods for detecting Helicobacter species in faecal samples and examined whether Helicobacter spp. were associated with inflammatory bowel disease. A PCR-DGGE assay for the detection and speciation of lower bowel helicobacters was developed and validated in mice. In a human study of 43 controls and 27 inflammatory bowel disease cases with nested PCR, there was no significant difference in the prevalence of Helicobacter spp. DNA found in IBD cases and controls. In all positive subjects the amplified DNA was identical to Helicobacter pylori, suggesting that wash-down of gastric organisms was responsible for the positive results.

In order to increase the number of bacterial species that were represented by PCR­ DGGE, primer sets were also developed for the three major groups of anaerobes present in human faeces; the Bacteroides-prevotella group, Clostridium coccoides group and Clostridium leptum subgroup. Using these assays it was shown that the bacterial species present in murine faecal and caecal samples were nearly identical. The primers were also applied to show that preparation of mice and humans with antibiotics and cathartics reduced colonisation resistance such that the bacterial populations of the large intestine could be altered. In particular, the administration of a suspension of donated faeces to a human or murine subject that had been prepared with antibiotics and cathartics resulted in a hybrid microbiota in the colon of that individual. In humans, the hybrid microbiota was predominantly derived from the infused faecal suspension (rather than the subjects own baseline micro biota) and was largely stable over the following 24 weeks.

The new PCR-DGGE primer sets in combination produced a more complete representation of bacterial species diversity than existing universal primers. Using these assays it was shown for the first time that the composition of the adult colonic microbiota can be altered. TABLE OF CONTENTS

CHAPTER 1: Introduction

1.1 The anatomy and function of the human colon 1 1.2 Bacterial metabolism in the colon 2 1.3 Culture based assessment of human gut bacterial ecology 6 1.4 Molecular studies of colonic bacterial species diversity 11 1.5 The acquisition of the colonic microbiota 15 1.6 Overview of human colonic bacterial ecology 17 1.7 Colonic bacteria and disease 21 1.8 Molecular methods for assessing the colonic microbiota 29 1.8.1 Qualitative molecular methods 29 1.8.1.1 PCR-denaturing gradient gel electrophoresis 29 1.8.1.2 PCR-cloning 31 1.8.2 Quantitative molecular methods 33 1.9 Helicobacter species and human IBD 35 1.10 Colonisation Resistance 39 1.11 Probiotics and colonic disease 42 1.12 Overview, Hypotheses and Aims 45

CHAPTER 2: Materials and Methods

2.1 Methods 49 2.1.1 Bacterial Culture 49 2.1.1.1 Culture of reference and laboratory strains of Helicobacter and Campylobacter species 49 2.1.1.2 Culture of reference and laboratory strains of other bacterial species 50 2.1.1.3 Culture of Desulfovibrio desulfuricans 51 2.1.1.4 Culture of B. adolescentis 51 2.1.1.5 Quantitation of bacteria harvested from plate cultures 52 2.1.1.6 Culture of Helicobacter species from murine caecum and faeces, and human colonic biopsies 52 2.1.2 DNA extraction 53 2.1.2.1 DNA extraction from plate cultures 53 2.1.2.2 DNA extraction from murine and human samples 53 2.1.2.3 DNA quantitation 54 2.1.3 PCR amplification 54 2.1.3.1 Development of novel primer sets 54 2.1.3.2 PCR conditions 55 2.1.4 Denaturing gradient gel electrophoresis 56 2.1.5 Analysis of DGGE banding profiles 57 2.1.5.1 Calculation of the Bray-Curtis similarity for lane comparisons 57 2.1.5.2 Statistical analysis of pooled Bray-Curtis similarities 57 2.1.6 Sequencing 57 2.1.7 Extraction of PCR product from polyacrylamide gels 58 2.1.8 PCR-cloning 58 2.1.8.1 Generation of Competent E. coli DH5a with the Rubidium Chloride method 58 2.1.8.2 Clone library construction 59 2.1.8.3 Analysis of PCR-clone libraries 60 2.1.9 Quantitative comparison of clone libraries 61 2.1.1 0 Animals 62 2.1.11 Ethics 62 2.2 MATERIALS 63 CHAPTER 3. Development of primers targeting the major phylogenetic groups of bacteria present in the lower bowel and their application to the comparison of murine faecal and caecal microbiota.

3.1 INTRODUCTION 75 3.2 METHODS 79 3.2.1 Primer development 79 3.2.2 Comparison of the primers with 16S rRNA sequences from the murine lower bowel microbiota 81 3.2.3 Optimisation of PCR and DGGE 81 3.2.4 Animals and sample collection for comparison of caecal and faecal micro biota 82 3.2.5 Sample processing 82 3.2.6 Statistics 82 3.3 RESULTS 84 3.3.1 Primer development 84 3.3.2 Comparison of the primers with 16S rRNA sequences from the murine lower bowel microbiota 85 3.3.3 Application of PCR-DGGE to murine faecal and caecal samples 86 3.4 DISCUSSION 103

CHAPTER 4: Development and validation of Helicobacter genus specific PCR­ DGGE

4.1 INTRODUCTION 111 4.2 METHODS 118 4.2.1 Development and validation of the Helicobacter genus- specific primer set and PCR-DGGE 118 4.2.2 Limit of detection ofHelicobacter genus-specific PCR-DGGE 119 4.2.3 Comparison of the sensitivity ofHelicobacter genus- specific PCR-DGGE with Species-specific PCR and Culture 119 4.2.3.1 Sample collection and culture 119 4.2.3.2 Species-specific PCR 120 4.2.3.3 Nested PCR 120 4.2.4 Sequencing 121 4.2.5 Application ofHelicobacter genus-specific PCR-DGGE to mice from the BABS Animal Facility 121 4.3 RESULTS 122 4.3.1 Development of the Helicobacter genus-specific PCR-DGGE 122 4.3.2 Limit of detection for H. hepaticus using Helicobacter genus-specific PCR-DGGE 123 4.3.3 Comparison of the Helicobacter genus-specific PCR-DGGE with species-specific PCR and culture 123 4.3.3.1 Culture of caecal mucus scrapings and faecal samples 123 4.3.3.2 Species-specific PCR using faecal template DNA 124 4.3.3.3 Helicobacter genus-specific PCR-DGGE 124 4.3.3.4 Comparison of the sensitivity of Helicobacter genus-specific PCR, species-specific PCR and culture 124 4.3.4 Application ofPCR-DGGE to mice from the BABS Animal Facility 125 4.4 DISCUSSION 138

CHAPTER 5: Absence of mucosa-associated colonic helicobacters in an Australian urban population

5.1 INTRODUCTION 143 5.2 METHODS 150 5 .2.1 Australian study 150 5.2.1.1 Subjects 150 5.2.1.2 Sample collection 150 5.2.1.3 Controls for PCR 151 5.2.1.4 Nested PCR 151 5 .2.1.5 Statistical analysis 152 5.2.1.6 Culture 152 5.2.2 Malaysian study 152 5.3 RESULTS 153 5.3.1 Australian study 153 5.3.1.1 Nested PCR 153 5.3.1.2 Sequencing 153 5.3 .1.3 Statistical analysis 154 5.3.2 Malaysian study 154 5.4 DISCUSSION 157

CHAPTER 6: Colonisation resistance in a murine model

6.1 INTRODUCTION 161 6.2 METHODS 168 6.2.1 Experiment 1: Effectiveness of different cathartic regimens for reducing luminal content in the murine caecum and colon 168

6.2.1.1 ~imals 168 6.2.1.2 Experimental regimen 168 6.2.1.3 Outcome measures 169 6.2.2 Experiment 2: Colonisation resistance 169 6.2.2.1 Animals 169 6.2.2.2 Faecal suspension 169 6.2.2.3 Experimental regimen 170 6.2.2.4 Histology 170 6.2.2.5 PCR-DGGE 171 6.2.2.6 Gel analysis 171 6.2.2. 7 Statistics 172 6.2.3 Experiment 3: Colonisation resistance- improved protocol 173 6.2.3.1 Animals 173 6.2.3.2 Faecal suspension 173 6.2.3.3 Experimental regimen 173 6.2.3.4 Histology 174 6.2.3.5 PCR-DGGE 174 6.2.3.6 Gel analysis 174 6.2.3.7 Statistics . 175 6.3 RESULTS 176 6.3.1 Experiment 1: Effectiveness of different cathartic regimens for reducing luminal content in the murine caecum and colon 176 6.3.2 Experiment 2: Colonisation resistance 177 6.3.3 Experiment 3: Colonisation resistance- improved protocol 178 6.4 DISCUSSION 201

CHAPTER 7: Colonisation resistance in humans

7.1 INTRODUCTION 208 7.2 METHODS 212 7.2.1 Subjects 7.2.2 Treatment protocol 212 7 .2.2.1 Test subject preparation 212 7 .2.2.2 Preparation of the faecal suspensions from the source subjects 213 7.2.3 Sample collection 213 7.2.4 DNA extraction and PCR-DGGE 214 7.2.5 Assessment of the RNA polymerase f3 subunit PCR-DGGE 214 7.2.6 Gel analysis 214 7.2.7 Statistical analysis 215 7.3 RESULTS 216 7.3.1 Assessment of the RNA polymerase f3 subunit PCR-DGGE 216 7.3.2 Assessment of the test subjects' samples with RNA polymerase f3 subunit PCR-DGGE 216 7.3.3 Assessment of the test subjects' samples with PCR-DGGE for the domain Bacteria 217 7.3.4 Assessment of the test subjects' samples with PCR-DGGE for Bifidobacterium 217 7.3.5 Assessment of the test subjects' samples with PCR-DGGE for the Bacteroides-prevotella group 218 7.3.6 Assessment of the test subjects' samples with PCR-DGGE for the Clostridium coccoides group 219 7.3.7 Assessment of the test subjects' samples with PCR-DGGE for the Clostridium leptum subgroup 219 7.3.8 Assessment of the stability of the banding patterns post-procedure 220 7.4 DISCUSSION 230

CHAPTER 8: Comparison of clone libraries derived from studies of colonisation resistance in humans

8.1 INTRODUCTION 237 8.2 METHODS 241 8.2.1 Subjects 241 8.2.2 DNA extraction and PCR-cloning 241 8.2.3 Sequencing, identification of homologues and tree construction 241 8.2.4 Quantitative comparison of clone libraries 242 8.2.5 Assessment of the effect of library size on UBSHUFF comparisons 242 8.3 RESULTS 244 8.4 DISCUSSION 271

CHAPTER 9: Discussion

9.1 Principal findings 278 9.2 The further application of PCR-DGGE in studies of colonic bacterial ecology 279 9.3 The issue of sampling - are the species that are present in human faeces representative of the caecal microbiota? 283 9.4 Do humans have mucus-colonising bacterial populations in the colon? 284 9.5 Can changing the human colonic microbiota have therapeutic application? 285 9.6 Conclusion 287

REFERENCES 289

APPENDIX: LIBSHUFFpvalues 329 CHAPTER 1: Introduction

1.1 The anatomy and function of the human colon

The human large intestine is an open system around 1.2 m long that receives approximately 1.5 kg of ileal chyme and generates 100 to 200 grams of faecal material per day (63). This dramatic reduction in volume is primarily due to the absorption of sodium and water from ileal chyme. In the colonic lumen, the anaerobic fermentation of non-absorbed dietary carbohydrates and proteins results in the production of short chain fatty acids (SCFA) that function as the primary energy substrate for human colonocytes (279, 280) and promote the absorption of sodium and water from luminal content.

A degree of regional specialisation within the human colon is suggested by studies of colonic motility. The caecum and right colon act as a reservoir for mixing and fermentation while the rectosigmoid colon functions as a storage area to delay defecation to a socially appropriate time. Radioisotope marker studies have shown that the caecum/ascending colon and the rectosigmoid colon are the two regions within the colon where the luminal contents spend the largest proportion of total colonic transit time (221). This observation has been directly confirmed in studies of humans post-mortem (14). The presence of a functional colonic reservoir is also suggested by the exponential faecal excretion of ingested radio-opaque markers (375). The turnover time of the colonic pool is reported to be around one third of the whole gut transit time, or approximately 20 hours in an individual with an average transit time of 60 hours (375).

Manometric and scintigraphic studies provide a dynamic view of the activity of the colon. Measurements with nasally or rectally passed colonic luminal catheters show that increases in intracolonic pressure can be divided into non-propagating and propagating sequences based on recordings of pressure waves at spatially separated recording points (13). Using this approach the motor activity of the human colon has

- 1- been shown to be relatively quiescent at night compared with morning and evening periods (13, 58, 270). In a study of the unprepared colon of healthy subjects by Bampton et al (2001) propagating sequences were more frequent in the morning and evening (21 ± 3 and 20 ± 4 per 8 hour period respectively) than at night (14 ± 3 per 8 hours)(13). In particular, morning awakening was associated with an increased frequency of propagating sequences. High amplitude propagating sequences are defined as propagating sequences where at least one component of the pressure wave has an amplitude greater than around 100 mmHg (58, 270). In the study of Bampton et al these sequences occurred 9.9 ± 1.4 times per subject per day, were more frequent after meals and were associated with 62% of episodes of urge to defecate. However, there is only a partial correlation between manometric propagating sequences and movement of luminal content. Combined manometric and scintigraphic studies reveal that those propagating sequences that are associated with the movement of luminal content (i.e. propulsive sequences) have a higher amplitude and slower velocity than non-propulsive sequences. A study of the colon of volunteers who had been prepared with colonic lavage solution showed that 64% of propagating sequences caused no detectable propulsion of content, 25% transported the radioisotope the full length of the propagating sequence and 11% showed incomplete transport (58). Manometric studies also clearly demonstrate both antegrade and retrograde propagating sequences in the colon (13, 58), with some studies suggesting that retrograde propagating sequences are more frequent in the proximal colon (58). Scintigraphic studies also show clear evidence for retrograde movement of luminal content in the human colon (179, 258). Overall the scintigraphic and manometric data confirm that the colon does not act as a simple reservoir for fermentation but has a range of regional functions and diurnal variation in activity. In particular, the caecum acts as a reservoir for mixing and fermentation.

1.2 Bacterial metabolism in the colon

Bacteria comprise between 45 and 60 % of the dry weight of human faeces (334 ). Approximately 65 to 85% of these are viable at the time of defaecation as assessed

-2- by cell well integrity or cultivated isolates as a fraction of microscopic clump counts (7, 226, 229). Although bacteria are viable in faeces it is clear that their replication in vivo is not as rapid as under ideal conditions in vitro. For example in a study conducted in mice, comparison of the daily bacterial faecal output and total gastrointestinal bacterial numbers suggested that overall only 1 to 5 bacterial generations occur per day (118). Similar findings were reported in mice that were only colonised with Escherichia coli, where 1.2 divisions per day occurred (118). These results indicate that in the gut there are significant limitations to bacterial growth compared with growth in the log phase in vitro.

The luminal 0 2 tension in the unprepared human caecum has not been directly measured in vivo, however a number of lines of evidence suggest that it is primarily an anaerobic environment with focal areas containing higher oxygen tensions. Studies would suggest that in mixed bacterial environments oxygen concentrations can vary significantly over distances ofless than 1 mm (301). Thus the coexistence of closely opposed aerobic, microaerophilic and anaerobic environments is entirely plausible. Evidence to support the view that conditions in the caecum are predominantly anaerobic comes from direct observations of the oxygen tension in the caecum and colonic lumen of animals. For example, the partial pressures of 0 2 in the colon of germfree and conventional rats (measured by inserting a canula 8 em into the colon and analysing luminal gas by mass spectrometry) were 15.1 ± 6.3 and

7.3 ± 2. 7 rnmHg respectively (29). This compares with a partial pressure of 0 2 of 150 rnmHg in air. Culture based and molecular studies of the bacterial populations that reside in the human ileum and colon also suggest that a transition from a predominantly aerobic to an anaerobic environment occurs near the ileocaecal junction with anaerobes the dominant group in the caecum. In an early study of the bacterial populations of the intestine, Gorbach et al (1967) obtained simultaneous aspirates via a sterilised double lumen polyvinyl tube from the terminal ileum and colon of adults 2 to 4 hours post meal (136). This study showed that the counts of all of the bacterial groups assessed were higher in the caecal than the ileal samples (total aerobes, coliforms, staphylococci, streptococci, lactobacilli and anaerobes).

-3- 9 1 Total anaerobes were 106.4 to 10 · per millilitre (ml) of caecal content compared with

3 1 5 8 10 · to 10 · per ml in ileal samples. Counts of coliforms (facultative bacteria) were

5 6 7 4 4 9 10 · to 10 · per ml in the caecal samples as compared with 0 to 10 · per ml in ileal samples. The quantity of anaerobes is likely to have been underestimated because samples were not transported under anaerobic conditions and an inferior form of anaerobic culture was performed (i.e. 10 fold dilution of samples in aerobic conditions) (229). A recent molecular study using dot-blot hybridisation has confirmed the predominance of anaerobes in the caecum (213). This study suggested that facultative anaerobes comprise 25% of the caecal microbiota and 1% of the . faecal microbiota with the remainder belonging to strictly anaerobic groups (213). The presence of a predominantly anaerobic flora however does not necessarily imply strictly anaerobic conditions. Some anaerobes can grow in the presence of oxygen if the oxidation-reduction (redox or Eh) potential is held at a sufficiently low level (15). A low redox potential is likely to be present in the human colon although this has not been directly measured in vivo (82). Anaerobic bacteria vary in their tolerance of oxygen and Eh conditions, a fact that relates to the susceptibility of critical enzyme systems to 0 2 and oxidised substances (152, 161).

The bacterial metabolic processes that occur within the human large intestine are well understood and are defined by the available substrates and predominantly anaerobic conditions of the colon (63, 64). The principal substrates in Western diets that are available for bacterial utilisation are resistant starch (8 - 40 g), non-starch polysaccharides (principally cellulose, 8- 18 g), oligosaccharides (2- 8 g), protein (3- 9 g), pancreatic enzymes (4- 6 g) and mucins ( 2-3 g) (63). As oxygen is generally unavailable as a terminal electron acceptor in the colon, either the products of substrate metabolism themselves (fermentation generating adenosine triphosphate - ATP- through substrate-level phosphorylation) or inorganic molecules such as

S04 or N03 act as electron acceptors (anaerobic respiration) (63). Fermentation of substrates such as resistant starch, nonstarch polysaccharides and protein leads to the production of SCFA and H2• The energy yield from fermentation is low, so that the removal of H2 by nitrite reduction, methanogenesis, acetogenesis, or sulfate

-4- reduction is essential. Further bacterial utilisation of the SCFA acetate, propionate and butyrate is limited by the marginal energy benefit of doing so under anaerobic conditions, leaving these end products to accumulate for their absorption by the host and subsequent utilisation in muscle, liver and the colonic epithelium respectively (64). The main products of bacterial fermentation in the colon are SCFA, other organic acids, lactate, succinate, phenols, indoles, amines and ammonia (63).

The metabolic activity of bacteria varies at different sites within the colon and between different individuals. Direct examination of the colonic contents from two sudden-death victims by Macfarlane et al (1992) revealed differences in the fermentation reactions occurring in the caecum and colon (201). This is despite the relatively homogeneous composition of the dominant bacterial species in the ascending colon, transverse colon, descending colon and rectum assessed at post­ mortem by Moore et al (1978) (the caecum was not assessed however) (227). In the study of Macfarlane et al, SCFA concentrations were found to be higher and the luminal pH lower in the caecum and right colon as compared with the left colon (201). This was subsequently shown in vitro to be the result of the greater availability of fermentable carbohydrate in the right colon. The products of protein breakdown such as phenol, p-cresol and phenolic acids were higher in the left colon.

Bacterial metabolic activity may differ significantly between individuals. In only one of the two subjects studied by Macfarlane et al were methanogens present, and in this subject, methane production and methanogen counts were much higher in the left colon than the right (201). Methanogenesis and sulfate reduction are competing processes for the utilisation of H2 with either one or the other dominating in the colon of a given individual (121). In a comparative in vitro study of faecal samples from 20 Britons and 20 rural black South Africans, Gibson et al demonstrated that 70% of Britons and 15% of South Africans harboured sulfate-reducing bacteria while the remainder produced methane (120, 121). In a study of faeces from French volunteers, Pochart et al found that bacterial counts of methanogens were much higher in the 10 methane producers than non-producers (109 versus 104 per gram)

- 5- 6 5 and that the methane producers had lower numbers of sulfate-reducing bacteria (10 ·

7 3 versus 10 · ) (260). Overall these studies suggest that while the core processes of fermentation are common to all, other metabolic reactions such as sulfate reduction and methanogenesis do vary between individuals.

By using media containing a restricted range of substrates for bacterial growth it has been possible to identify the phylogenetic groups of bacteria that participate in the catabolism of specific substrates. For example, bacteria that degrade complex carbohydrates, proteins and amino acids or grow on intermediate products of fermentation such as lactate, succinate and ethanol have been described in this manner (summarised by Cummings and Macfarlane (1991)) (63). In the colon, a range of bacterial genera contribute to the majority of known metabolic activities, with a few genera being linked to specific activities. Examples of the former include the broad range of genera capable of degrading complex carbohydrates, including Bacteroides, eubacteria, bifidobacteria, ruminococci and clostridia (63). Mucins have been reported to be degraded by bifidobacteria, ruminococci and Bacteroides species (63). Examples of organisms with specific metabolic activities include the sulfate-reducing bacteria (mainly Desulfovibrio species) and methanogenic archaea (mainly Methanobrevibacter species) of the human colon that compete for hydrogen (119, 121, 201). In groups such as these, there appears to be a good correlation between phylogenetic classification and metabolic activity. It is important to note however, that most colonic bacteria are uncultivable and studies have shown that bacterial phylogeny may be a poor indicator of metabolic activity (264).

1.3 Culture based assessment of human gut bacterial ecology

Early attempts to define the microbiota of the human colon used in vitro culture and anaerobic methods such as the anaerobic jar, anaerobic chamber and roll tubes (160, 226, 326). In the early literature there was considerable controversy regarding the appropriate methods for sampling the intestinal contents (136, 346) and for

-6- transporting intestinal samples, as well as the appropriate conditions for their incubation.

Sampling methods included direct sampling at surgery or post-mortem (201), needle aspiration (346), sampling via a nasally passed polyvinyl tube (136), capsule studies (77, 318) and direct analysis of faecal samples (93, 133, 226, 391). Studies of the small bowel of dogs suggested that direct sampling provided the highest bacterial counts followed by sampling through a polyvinyl tube and then needle aspiration (136). By repeated sampling of one human subject, Gorbach et al (1967) demonstrated that sampling via a polyvinyl tube was reproducible (136). Using this method, streptococci, staphylococci, lactobacilli and fungi were the most populous isolates from most human small bowel samples. Subsequent culture-based studies of the colon have most commonly studied faecal samples as a method for describing the microbiota of the colonic lumen (93, 155, 226, 230). In support of this approach, a culture-based study of the luminal flora of cadavers within 4 hours of death suggested that the flora of the ascending, transverse, descending colon and rectum were similar (227). This suggests that faecal samples are a reasonable proxy for bacterial populations in the colon (227).

The methods used to transport and store samples include storage at room temperature under C02, storage at 4°C, freezing at -20°C or lower with reduced glycerol containing medium, and direct inoculation into roll tubes for transport under

C02 (62, 226,229, 331). To assess these methods, Crowther (1971) cultured faecal samples on a range of selective and non-selective media after storage under a variety of conditions. Significant quantitative losses in cultivable bacteria were observed with freezing, although this was ameliorated by making faeces into a suspension in 10% glycerol broth, prior to storage at -80°C (62). At room temperature and above, enterobacteria and lactobacilli proliferated dramatically and distorted the cultivable populations (62). Like Crowther, Moore and Holdeman also suggested that 2 hours of air exposure may have altered the results of faecal culture however this was not examined directly (155, 227). They stated that in their experience, freezing in a

-7- reduced glycerol containing medium or inoculation into roll tubes under C02 resulted in the best recovery of viable bacteria from human faecal samples (226, 229).

Although anaerobic jars, anaerobic chambers and roll tubes will all produce anaerobic conditions for incubation, an often overlooked factor in culture studies has been the method used for the dilution of samples prior to incubation. In many cases this was performed aerobically in non-reduced media (133, 136, 391). As a result, these studies overemphasise the importance of aerobes and facultative bacteria in the human intestine as compared with more advanced methods in which faecal samples are diluted in pre-reduced media in an anaerobic chamber (93, 226, 229).

The method by which bacterial numbers were quantitated and reported also differs significantly between studies. Most commonly 10 fold dilutions of faecal suspensions were directly plated onto a whole range of selective and non-selective media (62, 77, 93, 116, 133, 136) and the number of bacterial colonies on a given plate type reported as a count to the base 10 (62, 77, 93, 116, 133, 136). Others such as Finegold et al (1974) combined this approach with the identification of individual colonies at the species level, based on a range of phenotypic characteristics such as gram stain, biochemical testing and chromatography of cellular fatty acids (93). Subsequent studies of the bacterial species cultured from murine and human faeces using selective media and examining the gram stain, biochemical testing or 16S ribosomal DNA sequences of individual colonies have demonstrated that the selective media used in early studies are indeed selective but not specific (7, 149, 238). For example in a study of murine faeces only 24% of colonies on Rogosa agar were lactobacilli (238). Similarly a study of human infant faeces has shown that many different colonic bacterial species commonly grow on selective media for Bacteroides spp. (reinforced clostridial medium), Clostridium spp. (sulphite-polymixin-milk agar), Bifidobacterium spp. (raffinose-bifidobacterium agar, Beerens medium and Bifidobacterium spp. medium) and coliforms (MacConkey agar) (149). In addition, combined molecular and culture based

-8- approaches to the quantitation of bifidobacteria in human faeces have shown that many selective media for bifidobacteria grossly underestimate the numbers of B. adolescentis - the principal species in adults (7).

In contrast the cultural method of Moore and Holdeman (1974) involved initial incubation on non-selective media followed by the identification of large numbers of individual colonies rather than plate counts (155, 226, 229). The results of these studies are reported as a ranking of the frequency of species isolated and not counts to the base 10. In their studies, stools were immediately transported to the laboratory in a plastic bag where they were flushed with C02 and kneaded. All subsequent work was performed ~nder C02• One gram of stool was then diluted in 9 ml of a pre­ reduced salt solution and 10 fold dilutions performed. The 108,109 and 10 10 dilutions were then inoculated into rumen-fluid-glucose-cellobiose agar roll tubes and incubated at 37°C for 5 days. Fifty-five colonies were then picked in succession from the top of the tube downwards and inoculated into Sweet "E" broth. The resulting isolates were then characterised using around 40 different media, biochemical tests, gas production and subsequently gas chromatography. Between 44 and 69 isolates were obtained from each subject. The 10 most frequently isolated species were: Bacteroides vulgatus, Faecalibacterium prausnitzii, Bijidobacterium adolescentis, Eubacterium aerofaciens, Propionibacterium productus-ll, Bacteroides thetaiotaomicron, Eubacterium eligens, Propionibacterium productus-l, Eubacterium biforme and Eubacterium aerofaciens-lll (226). Facultative bacteria were not specifically examined in these studies by microaerophilic culture and are therefore not represented in their results.

It is difficult to make any definite statement concerning the reproducibility of culture of faecal samples because of a real paucity of direct evidence, however reproducibility is likely to be the main drawback of this approach. An ideal assessment of this would require repeated culture of the same sample, however this has rarely been performed. Duplicate culture of the same intestinal sample using the early methods of plate dilution on selective media revealed variations in culture

-9- counts of at least 1log (62, 133). Studies using the more involved culture methods that identified each individual colony have not directly examined reproducibility, despite the large amounts of published literature using this approach (93, 226, 230). This makes the interpretation of any differences observed in such studies difficult. Comparisons of the microbiota of four regions of the colon sampled at post-mortem suggested that culture is reproducible to the extent that the most frequently isolated species were usually represented in each of the four samples from a given individual. However, species that were represented by less than 7 isolates in any of the 4 samples were not always detected in each sample (227). The same authors also reported that "different 1 g sub-samples of the stool often differed in bacterial composition", but did not provide data to illustrate this (226). Repeated sampling of faeces from three individuals over a 3-month period did suggest that there were differences in the frequency of different isolates (155). Whether these changes represented a real change in micro flora or the effects of limited sampling from a large faecal microbiota is unknown. This questionable reproducibility has hampered studies that attempted to determine whether the microflora of an individual varied over time and whether there were quantitative differences between individuals (93, 133, 155, 230, 391). The time consuming and laborious nature of phenotypic identification of each colony is a second drawback to this approach (93, 155, 226, 229).

In summary, carefully conducted and exhaustive cultural studies of human faeces (93, 155, 226, 230) have been invaluable in improving our knowledge of the range of species present in the human colon and their metabolic activities. Moore and Holdeman estimated that they had cultured between 94 and 98% of the cultivable bacteria present in the human colon by pooling the isolates in their studies (155, 226). However, the questionable reproducibility of the technique means that prior to the development of molecular approaches to bacterial ecology, there was no clear consensus as to whether the microbiota of an individual varied in different colonic sites, was stable over time or whether it was influenced by various external factors. Although it is likely that this work has defined most of the cultivable species in the

- 10- human colon, molecular studies have subsequently demonstrated that a majority of the approximately 135 species present in the human colon have not been cultivated previously (157). Several recent studies estimate that culture only reveals 20 to 40% of the bacterial species present in human faeces (186, 335, 376). To make further progress in understanding the bacterial populations of the colon, the application of molecular techniques was required.

1.4 Molecular studies of colonic bacterial species diversity

Molecular studies of bacterial ecology are based primarily on studies of the sequence of 16S ribosomal RNA. Each bacterial cell has one or more copies of 16S rDNA that encode a large number of cytoplasmic 16S rRNA molecules. In molecular ecology studies, either the RNA or DNA can be amplified or probed (5). The 16S rRNA sequence is particularly useful in studies of phylogeny because this essential gene is universal and has a mosaic structure depicting both ancient and recent evolutionary relationships (378). Probes or primers can often be designed to target a particular genus or species based on common regions of 16S rDNA sequence (5). In addition, molecular studies are supported by the large public databases that contain 16S rRNA sequences of cultured and uncultured bacteria (GenBank (19), the European Molecular Biology Laboratory database (EI\1BL) and the Ribosomal Database Project (55)).

In molecular studies of ecology, ribosomal DNA from Archaea, fungi and protozoa are not amplified by the universal primer sets used to target the domain Bacteria. The domain Archaea includes organisms that are morphologically similar to bacteria however their cell walls, ribosomal RNA and antibiotic sensitivity are quite distinct. Methanogens, belonging to the genus Methanobrevibacter, have been cultivated from the human colon of 72 to 100% of subjects, in counts of 103 to 109 per gram (wet weight) of stool (222, 223, 260). Whether other archeal genera are present in the human colon has not been directly addressed but it is known that a range of genera are present in other anaerobic environments such as the rumen (342). The

- 11- literature that follows specifically relates to the domain Bacteria in the human colon and excludes the assessment of the other domains - the Archaea and Eucarya.

Studies of clone libraries have revealed the true diversity of the bacterial ecology of the human colon (28, 154, 335, 376). PCR-cloning facilitates the sequencing of individual 16S ribosomalJ?NA molecules from a mixed population of PCR products. In this procedure DNA is extracted from the test sample (for example faeces) and the 16S rDNA is amplified by PCR. This PCR product is then ligated into a vector (usually a plasmid) which is inserted into a bacterium. The bacterium is then cultured so as to produce individual colonies, each of which contains a unique vector. The vector inserts from multiple colonies can then be individually sequenced so that a list of 16S rRNA sequences present in the source DNA is generated. These sequences can be presumptively identified by comparison with public databases including GenBank and RDP.

The most comprehensive molecular study of the microbiota of the human faecal microbiota to date, involved the sequencing of284 16S ribosomal DNA (16S rDNA) clones from the faecal sample of a 40 year old male (335). For the purposes of their study Suau et al (1999) defined a "species" or Operational Taxonomic Unit (OTU) as those sequences with 98% or greater sequence homology. Table 1.1 shows the number of clones and OTU' s detected in the study of Suau et al placed within each of the relevant phylogenetic groups or subgroups of the Ribosomal Database Project (55). Clones placed within the Bacteroides -prevotella group, Clostridium coccoides group (cluster XIVa) and Clostridium leptum subgroup (cluster IV) were most frequent. Sixty-three of the 83 OTUs from the clone library of Suau et al were shown to have greater than 2% divergence from the sequences of named bacteria in the RDP and GenBank and these therefore represent potentially novel (previously uncultivated) species. Other PCR-cloning studies of human faecal samples that have used a larger number of cycles of PCR amplification or different primer sets have confirmed the results of Suau et al (28, 376). Similarly, clone libraries based on DNA obtained from washed colonic tissue samples thought to represent mucus-

- 12- associated flora have shown the same frequency and phylogenetic placement of sequences as studies of faecal samples (154). Interestingly, sequences representing Enterobacteriaceae are very infrequent in 16S rRNA clone libraries from human faeces or colonic biopsies (28, 154, 335, 376) despite their prominence in early cultural studies.

Clone libraries representing the caecal or colonic microbiota from animals have generally shown that their bacterial species originate within the same phylogenetic groups as the faecal microbiota of humans. Analysis of 16S rRNA clone libraries from the caecum or faeces of broiler chickens (387), mice (288), and pigs (192) have shown that a majority of sequences also belong within the three major phylogenetic groups identified in human faeces (that is, the Bacteroides-prevotella group, Clostridium coccoides group and the Clostridium leptum subgroup). However, the individual sequences within these three phylogenetic groups from the lower bowel of different animals and birds are distinct, suggesting that each has a unique microbiota with common ancestors (69, 387). In addition to these common types, several bacterial groups that were not found in significant numbers in human faeces are frequently represented in clone libraries derived from the caecal or colonic samples of animals and birds. For example, the Bacillus-Lactobacillus­ Streptococcus subdivision (RDP 2.30.7), and Sporomusa and relatives group (RDP 2.30.3) that are infrequent in human clone libraries were commonly found in 16S rDNA clone libraries from mice, pigs and chickens, while Proteobacteria (RDP 2.28) were also represented in significant numbers in libraries from pigs and chickens (192, 288, 387).

The qualitative representation of species diversity in human faeces from clone libraries is complemented by quantitative estimates made with 16S rDNA targeted fluorescent or radiolabelled probes. In a study by Harmsen et al (2002) the mean proportions of various target species present in faecal samples from 11 adult volunteers was determined using fluorescent in situ hybridisation (FISH) (148). The results of this study are summarised in Table 1.2.

-13- The probes used by Harmsen et al produced a nearly complete representation of the bacterial cells fluorescing with the bacterial probe Bact338. Of the total number of cells fluorescing with this probe, 95% were represented by the combined set of probes of Harmsen et al. However, only 60% of bacterial cells staining with the DNA stain 4' ,6-diamidino-2-phenylindole (DAPI) were detected with Bact338. This may be due to the fact that these cells were impermeable to the probe, possessed a 16S rRNA secondary structure that prevented probe hybridisation, were Archaea, or perhaps most likely, were nonviable (148). Significantly, the relative quantities of the phylogenetic groups detected by Harmsen et al are in agreement with the frequency of clone placements from the clone libraries of Suau et al (335) and Hold et al (154). The high degree of similarity in the findings of these studies, that used very different methodologies, suggests that Tables 1.1 and 1.2 are a reasonable representation of the major phylogenetic groups of bacteria present in human faeces.

The metabolic activities of many of these bacteria are unknown. For the majority of species of the human colonic microbiota that have not been cultivated, presumptive metabolic activities can only be ascribed tentatively, based on the known activities of their cultivated relatives. Although large numbers of cultivated human faecal bacteria have been phenotypically characterised (93, 226) many are as yet unnamed and unpublished (230). For example, culture of the faecal flora of 88 humans revealed 371 taxa of which 261 were still unnamed in 1995 (230). Studies of the phylogeny of cultivable butyrate-producing members of the C. coccoides and C. leptum clusters show that they are interspersed with non-butyrate producing species (264). This highlights the pitfalls of assessing the likely metabolic activities of colonic bacteria based on phylogeny. Therefore, although the overall metabolic outcomes in the colon are well understood, the contribution of each bacterial species requires further research.

- 14- 1.5 The acquisition of the colonic microbiota

The gut of the healthy unborn human infant is sterile and the colonic microbiota of human infants is acquired orally. Strong evidence for this proposition is the observation that the bacterial flora of the intestine, just above congenital jejunal atresia resembles the micro biota of the normal large intestine, whereas distal to the area of atresia the intestine remains sterile (23). Studies suggest that there is little resistance to colonisation with bacterial species in the germ-free gut. For example, germ-free mice can be monoassociated with a broad range of bacterial species including bacteroides, staphylococci, lactobacilli, enterococci and even species that are not usually part of the normal flora of the gastrointestinal tract such as Pseudomonas species (298).

Animal models and studies of human infants have revealed that bacterial succession occurs within the colon. That is, bacterial species will colonise an available niche in the large intestine or displace other species in a competitive process. For example, colonisation with facultative anaerobes often occurs prior to colonisation with strict anaerobes, presumably because the former contribute to the generation of reduced anaerobic conditions required by the latter (299). Studies of mice and humans suggest that although this sequence is common, anaerobes can also colonise the colon without the apriori presence of facultative anaerobes (89, 298).

Studies of germ-free mice have provided an insight into the process of bacterial succession in the colon. Germfree mice associated with only a few murine strains of lactobacilli, anaerobic streptococci or coliforms are rapidly colonised throughout the gastrointestinal tract, by very large numbers of these bacteria - 107 to 109 per gram of intestinal content (298). Mice monoassociated with bacteroides also have very high levels of colonic colonisation -1010 -but colonisation of only 103 occurs in the stomach and small intestine. In general, these high levels of colonisation persist if no other bacterial exposure occurs. However, exposure of mice associated with a paucity of bacterial strains to the faecal microbiota of conventional mice, results in a

- 15- "precipitous fall" in cultured counts of these strains and the development of a more diverse microbiota (298).

In human infants the gut is initially colonised by bacteria from the environment. In the first week of life this may include the maternal oral, skin, breast, vaginal, perineal or faecal flora and other environmental bacteria. Coliforms and enterococci are often the initial colonisers of the gut, followed by anaerobic bacteria such as bacteroides and bifidobacteria after 4 to 7 days (202, 331). The type of delivery (vaginal or caesarean section) and feeding (breast versus bottle) have significant effects on the initial bacteriology of the infant colon. Bottle fed infants have significantly higher numbers of enterococci and coliforms than breast-fed infants (149, 331). Bifidobacterium and Bacteroides species are quantitatively the most important faecal bacteria of breast-fed and bottle-fed infants respectively before the introduction of solid foods (149, 331). The introduction of solids is accompanied by the development of a more diverse micro biota similar to that of adults, confirming the importance of substrate for determining the composition of the colonic bacterial populations (214, 331).

Recent molecular studies have confirmed the process of bacterial succession and the range of organisms observed in previous culture studies. A dot-blot hybridisation study of faecal samples from breast-fed infants suggested that soon after birth the faecal microbiota was dominated by bifidobacteria (21.7 ± 6.8% of total bacteria), lactobacilli (12.3 ± 3.2%) and "enterics" (34.6 ± 0.15%). After weaning, bacteroides (30.5 ± 9.4%), Clostridium leptum subgroup (3.5 ± 2.7%) and Clostridium coccoides group (10.6 ± 5.7%) appeared in the dominant microbiota (214). In a combined culture and molecular study, Harmsen et al (2000) cultured faecal samples from 6 breast and 6 bottle-fed infants during the first 20 days of life on 8 selective media (149). Colonies were typed by random amplification of polymorphic DNA (RAPD), grouped and the 16S rRNA of representatives from each RAPD group were sequenced. Bifidobacteria and E. coli were isolated from all of the babies. In contrast, Lactobacillus species and streptococci were only isolated from breast-fed

- 16- infants while clostridia and some species of staphylococci were only isolated from bottle-fed infants. In addition, FISH with eight different 16S rRNA probes was used to quantitate relative bacterial numbers within defined groups. From the fourth day onwards, bifidobacteria were by far the most numerous species in the faeces of breast-fed infants (around 50% by number) whereas Bacteroides spp. and E. coli were represented to a much greater extent in the bottle-fed infants. Interestingly, despite the prominence of Bacteroides species in the bottle-fed infants detected by FISH, these organisms were rarely cultured. A similar progression of bacterial succession was also clearly demonstrated in a PCR-denaturing gradient gel electrophoresis (PCR-DGGE) study' of faecal samples from 2 infants from birth until 12 months of age by Favier et al (2002) (this method is discussed in detail in section 1.8.1.1). The time-course of colonisation corresponded to the pattern observed in earlier studies (89). Thus there is clear evidence, from culture and molecular-based ecological studies, for a defined pattern of bacterial succession in the infant gut.

The main lesson to be learned from studies of infants, is that the microbiota develops via the acquisition of new species and the loss of others over time, in a process that is strongly influenced by the available substrate. This evolution implies a competitive process in the colon. In contrast a number of studies of the ecology of the adult colon suggest that the species composition of the microbiota is stable, presumably representing the final evolutionary result of this process (155, 291, 388). Further research is needed to define the specific competative advantages possessed by each bacterial species in the colon and how interactions occur in vivo to produce bacterial succession.

1.6 Overview of human colonic bacterial ecology

The results of these more recent molecular studies plus those of previous culture based studies has led to an improved knowledge of the bacterial ecology of the human colon. Taken as a whole, the existing scientific literature relating to human colonic bacteria suggests the following points. Firstly, the range of detectable

- 17- species present in the human colon consists of a core group of indigenous bacteria and a smaller group of transients derived from oral intake (110, 148, 151, 154, 226, 313, 368). Molecular studies of adults suggest that this core group of bacteria has a stable species composition over months to years, despite quantitative variation in bacterial numbers (110, 148, 151, 291, 313, 332, 388). The stability of this core group of species is supported by the results of PCR-DGGE or PCR-temperature gradient gel electrophoresis with universal bacterial primers and bifidobacterial primers. These studies have shown stable banding profiles using faecal DNA obtained from individuals over time periods of up to 6 months (291, 332, 388). This suggests that in the human colon, the bacterial species composition within the groups targeted by these primer sets is likely to be stable over time. Presumably this reflects the presence of stable populations of caecal bacteria that proliferate and are excreted (375). In contrast, PCR-DGGE with primers targeting Lactobacillus species reveals a stable background population of lactobacilli, plus a changing population that is probably of dietary origin and represents part of the transient population of bacteria (151, 368). This transient population reflects the bacterial populations present in food (particularly fermented food) and water, including lactobacilli, bifidobacteria and coliforms.

In contrast to the results ofPCR-DGGE, culture based studies of the stability of colonic bacterial populations within individuals over time, have produced conflicting results depending upon the method used for culture and the rigour of the analysis (133). For example, the study ofZubrzycki and Spaulding (1961) suggested that the faecal microbiota of 4 individuals was unchanged over months (391). However, in this study, stool samples were not diluted in reduced media, and quantitation was only reported using plate counts on seven types of media. Counts of bacteria classified as "bacteroides", "enterococcus", "diphtheroids", coliforms and "lactobacilli" appeared stable, however statistical analysis was not performed. In a comprehensive culture study, the faecal micro biota of 33 Japanese-Americans on Western and Japanese diets was assessed using 27 types of media (93). Initial dilutions were conducted in an anaerobic chamber. A second sample was obtained in 7 of the 33 individuals after an interval of between 2 and 19 weeks. In this study Eubacterium len tum was detected in 26 of the 40 samples at a mean concentration of 1010 per gram of stool. However analysis of the 7 paired samples revealed that E. lentum was detected in only one of the samples in 5 of the cases. Examination of the other bacterial groups showed that differences in counts of up to 4 log were detected. Whether these differences reflect changes in the microbiota of these individuals over time, or are merely a function of sampling and the cultural method per se, is unknown. In contrast, a second study of faecal samples obtained over a 3 month period from adult males in the Skylab programme, studied with the method of Moore and Holdeman, suggested that the species composition of each individual was stable over time (155). That is, the bacterial species that constituted a significant fraction of the identified species from a given subject were likely to be identified in subsequent samples from that subject (155).

Faecal populations of the facultative bacterium E. coli have also been assessed over time using culture methods (307, 308, 379). Typing of E. coli strains is generally performed by serotyping, antibiogram or electromorph typing of bacterial enzymes. Using this approach several studies suggest that there are resident populations of E. coli strains in the bowel that are commonly detected within the one subject over time, and transient strains that are only detected once (307, 308, 379). Furthermore, studies performed over several years suggest that the resident strains of E. coli may change over long periods of time but are stable over periods of weeks to months (307, 308). However, none of these studies aimed to comprehensively assess E. coli strain diversity at a given time point but instead characterised 1 to 3 dominant isolates (307, 308, 379). Therefore the disappearance of dominant "resident" strains for periods of up to 12 months followed by their reappearance suggests that the results of these studies may reflect sampling issues rather than real changes in the colonic microbiota (307). However, at face value these studies suggest that there are resident and transient strains of E. coli in the human faecal microbiota.

- 19- The second major point revealed by molecular studies of faecal composition is that despite a persistent core group of bacterial species that are consistently represented in faecal samples from a given adult individual, there is quantitative variation in bacterial numbers over time. Franks et al (1998) performed FISH with probes for bacteroides (Bfra/Bdis), C. coccoides group (Erec482), C. leptum subgroup (Lowgc2P) and bifidobacteria (Bif164) on faecal samples collected from 9 volunteers over a period of 8 months using an automated analysis (110). Hybridisation of 10 separate aliquots of a single fixed faecal sample with the bacterial probe Bact338 established that the coefficient of variation of the assay was 0.18. An earlier study had shown that this coefficient was of a similar magnitude for different probes (164). That is, the variation in the results attributable to the assay itself were known. The recorded coefficients of variation for within subject comparisons over time were 0.27, 0.45, 0.39, 0.44 and 0.59 for Bacteria, Bacteroides group, C. coccoides group, C. leptum subgroup and bifidobacteria respectively. As the coefficients of variation are larger than those due to the assay per se , these results confirm that substantial quantitative variation occurs within these bacterial groups over time (110).

The third significant observation is that despite the commonality of most metabolic activities and core phylogenetic groups, the species composition of the colon of different individuals is unique. For example, PCR-DGGE with primer sets for the domain Bacteria, bifidobacteria, and lactobacilli generate different banding patterns (i.e."profiles") for different individuals (151, 291, 368, 388). The more comprehensive culture based studies confrrm this finding, with larger differences being observed in the species composition between, as compared with within individuals (93, 155, 226). In the study of Franks et al the coefficients of variation for bacterial numbers were also higher for between, as compared to within-subject comparisons (110). Thus studies would suggest that quantitative as well as compositional differences are present when the microbiota of different individuals is compared. There is also some suggestion that the similarity of the faecal microbiotas

-20- of individuals from different ethnic groups may be less than that from within a given ethnic group (230).

Fourth, once an adult microbiota is acquired, the existing ecology prevents colonisation by other introduced bacterial species. This is known as colonisation resistance and has been convincingly demonstrated in mice with members of the Enterobacteriaceae(357, 358). Colonisation resistance can be significantly reduced in mice by the administration of antibiotics (360) or caecectomy (367) suggesting that the anaerobic caecal bacteria may be responsible for this phenomenon. This concept will be discussed in detail later.

In summary it is clear that a core group of predominantly anaerobic bacterial species derived from the Bacteroides-prevotella group, C. coccoides group and C. leptum subgroups persist in the colon of adults over time. Each individual has their own combination of species drawn from within these groups and their quantity in faecal samples varies over time. This information provides a framework from which methods can be developed to study the relationship between bacterial populations and disease. It is clear that if quantitative changes in bacterial numbers are associated with disease then these will only be detected with current methods if those changes are substantial. However, changes in species composition within an individual over time are amenable to detection with targeted PCR-DGGE.

1. 7 Colonic bacteria and disease

Research into a link between the colonic microbiota and the aetiology of colorectal cancer was stimulated by epidemiologic data that showed that migrants who moved from an area of low cancer incidence to areas with a high cancer incidence developed a higher cancer risk (145). Despite the obvious impact of diet in this assessment, questions were raised as to whether the observed differences in cancer incidence were due to changes in the bacterial populations of the colon (228). A mechanistic basis for this link is founded on the knowledge that certain colonic

-21- bacteria produce enzymes such as ~-glucosidase, 7-a.-dehydroxylase nitroreductase

and ~-glucuronidase that can interact with dietary substances to form mutagenic substances (321). The overall hypothesis is supported by studies of animal models in which the presence of an intestinal microbiota and its composition has been shown to influence the effect of dietary mutagens on the colonic epithelium (169, 321).

In 1971 the Japan-Hawaii Cancer Study was established to compare the faecal microbiota of populations with different incidences of colorectal cancer. The results of this study were published in 1995 (230). Culture based assessment of the faecal microbiota of Japanese-Hawaiians with (N=18) and without (N=15) polyps, American Caucasians (N=17), rural Japanese (N=22) and rural Africans (N=16) was undertaken (230). The subjects with an intermediate and high risk (American Caucasians, and the two groups of Japanese-Hawaiians) were compared with the lower risk subjects (rural Africans and Japanese). A distance matrix was constructed showing the relatedness of each subject's microbiota (expressed as minimum percent identities) and each group of subjects was compared with lambda analysis (128). This analysis examined the mean percent relatedness within and between populations and compared the observed value with 1000 randomly shuffled analyses for significance. Using this analysis the microbiotas of low risk groups (rural Africans and Japanese) were significantly different to the high risk groups (Japanese-Hawaiians with or without polyps) and the American Caucasians of intermediate risk were not significantly different from either group. Fifteen bacterial species were found to be in a greater percentage of the total isolates in the high risk group and 5 species to be a greater percentage in the low-risk groups (p value of< 0.05). Interestingly the species that were different between the high and low risk groups are commonly found in the faecal microbiota. The observed differences might therefore be related to slight differences in the initial culture of samples on site in Africa or Japan or might relate to the effect of transport from distant locations. The number ofT-tests performed is unclear but 5% of these would have been significant by chance alone (potentially resulting in 10 positive results in 200 comparisons). In summary, whether the microbiota is different in populations with

-22- different rates of colorectal cancer independent of an effect of diet is uncertain, however this study suggests that the microbiota of these populations may be different (230). Certainly this is worthy of further investigation. Although bacterial populations play an undefined role in colorectal cancer risk there is a clear contribution from genetic factors.

The inflammatory bowel diseases (IBD), ulcerative colitis and Crohn's disease are characterised by chronic or relapsing inflammation of the large and/or small intestine. Although genetic factors are clearly important, there is also indisputable evidence for the role of bacteria in IBD pathogenesis (94).

A number of murine models of IBD have been described (153). In well characterised murine models of colitis such as the interleukin 10 (IL-l 0) knockout mouse, interleukin 2 (IL-2) deficient mouse, T -cell receptor mutant mouse, and CD4 reconstituted T and B cell deficient (SCID) mice, the presence of luminal bacteria is essential for disease development (52, 73, 74, 181, 183,224,287, 311). A bacterial flora is also required for the development of colitis in HLA B27/Human ~2 microglobulin transgenic rats (271). In all of these studies exposure to environments associated with a broader diversity of bacterial species accelerated colitis development. Further support for the role of bacteria in murine colitis comes from studies in which antibiotics and selected probiotics prevented and, to a lesser extent ameliorated, colitis in mouse and rat models (203, 204, 272, 304, 305).

In addition to these animal studies, there is also a large body of evidence supporting a role for luminal bacteria in human IBD. Crohn's disease of the terminal ileum and colon is highly dependent on the presence of the faecal stream and is temporarily ameliorated by surgical diversion of the small bowel (65, 150,247, 284). Antibiotics, particularly metronidazole, have also been shown to have modest efficacy in some forms ofiBD, such as perianal and fistulising Crohn's disease, in the prevention of post-operative recurrence, and in the treatment of pouchitis following surgery for ulcerative colitis (21, 285, 337, 353).

-23- Despite this dependence on the presence of luminal bacteria, previous studies of the faecal microbiota of human IBD cases and controls have not revealed consistent and reproducible differences in the bacterial populations of each (116, 252, 355). In an attempt to define any differences in the faecal microbiota Giaffer et al (1991) used a simplified semi-quantitative plate streaking method to enumerate total aerobes, enterobacteria, lactobacilli, bifidobacteria, total anaerobes and Bacteroides in faecal samples from 42 Crohn' s cases (22 of which had active disease), 37 ulcerative colitis cases (18 had active disease) and 21 controls. The results showed that counts of total aerobes and coliforms were increased in active Crohn's disease while counts of bifidobacteria were decreased. Otherwise there were no significant differences in bacterial numbers (116). In contrast, culture of frozen intestinal tissue obtained at surgery from patients with active Crohn's disease and controls failed to reveal any significant differences (252). A study ofthe faecal microbiota of 15 Crohn's disease cases, 45 first degree relatives and 12 controls found increased counts of anaerobic gram-positive coccoid rods and gram-negative rods in the Crohn's cases. The disease activity of the subjects was however not assessed (355). Fabia et al cultured colonic biopsies obtained from the lavaged colon of 30 ulcerative colitis cases (12 had active disease) and 30 controls (86). Decreased counts of total anaerobes, gram negative anaerobes and lactobacilli were observed in the group of ulcerative colitis cases. Other authors have reported significant increases in total aerobes and coliforms amongst cases of severe ulcerative colitis, however not with less severe or extensive disease (134). More recent studies using quantitative dot-blot hybridisation of faecal RNA have also revealed a trend towards increased number of Enterobacteriaceae and decreased anaerobes such as Bacteroides and C. coccoides groups in active Crohn's disease as compared with controls (86). This observation, plus the earlier culture based studies, suggest that facultative bacteria and aerobes are increased in faecal samples from cases with active ulcerative colitis and Crohn's disease, while anaerobes are reduced.

-24- However, this may represent a non-specific effect of rapid-transit per se which generally results in an increase in the ratio of facultative bacteria versus anaerobes, perhaps reflecting a more aerobic colonic environment or inadequate fermentation (3, 135, 156). Supporting this view is a study of faecal samples from 5 healthy elderly subjects and 4 elderly subjects with Clostridium difficile diarrhoea. This study showed decreased numbers of bacteroides and increased numbers of enterobacteria, clostridia and lactobacilli were present in plate culture from the diarrhoea cases (156). Dot-blot hybridisation in the subjects with diarrhoea confirmed that enterobacteria increased, while bacteroides decreased as a proportion of total faecal bacteria. Similarly, per-oral saline perfusion of normal volunteers has been found to increase the counts of aerobes in faeces (135). Likewise, children with infectious diarrhoea of any cause display increased faecal aerobe counts - particularly Enterobacteriaceae- and decreased counts of anaerobes particularly Bacteroides, bifidobacteria and Veillonellae compared with controls (3). Thus differences in the microbiota of IBD cases and controls, unrelated to an effect of rapid transit per se, have not yet been discernible using previously applied methods.

Two recent studies have focused on bacterial populations that are closely associated with the mucosal surface of the colon in IBD. Swidsinski et al (2002) performed culture and FISH on washed colonoscopic biopsies from IBD cases, self-limiting colitis cases and controls (340). The results of culture showed that the concentrations of all bacterial groupings (total anaerobes, bacteroides, total aerobes and enterobacteriaceae) were higher in the IBD cases than in controls. However within the IBD group, the concentrations of bacteria from biopsies of histologically normal mucosa were higher than from inflamed mucosa in the IBD cases. FISH revealed no differences in the quantity or appearance of surface bacteria in the biopsies from different groups (340). In contrast, another study using FISH reported increased numbers of surface bacteria in rectal biopsies from IBD cases as compared with controls (303). However in this study no bacteria at all were observed in 71% of controls and 32% of IBD cases. Due to the unknown and potentially variable effect of different bowel preparations, sampling and sample processing these studies are

-25- potentially flawed. There is not enough data to draw firm conclusions on the microbiota that is adjacent to the mucosa in human IBD.

In addition to studies of the microbiota generally, a number of authors have investigated the potential role of E. coli populations in human ulcerative colitis and Crohn's disease (31, 32, 59, 67, 215, 296). Current evidence suggests that genetically related E. coli strains, with the phenotypic property of diffuse adherence to the human colonic carcinoma-derived cell line Caco-2, are commonly found in ileal samples from cases of Crohn' s disease and that many of these isolates are genetically distinct from those of controls (31, 67, 215,296, 302). In these studies, E. coli was cultured from the ileal samples of two-thirds of cases and controls, while diffuse adherence to the cell line Caco-2 was observed in 60 to 85% of the E. coli isolates from the Crohn's disease cases and only 33% of the isolates from controls. The mechanism by which cellular adherence occurred could not be determined for a majority of the isolates (31, 215, 302). Of the 136 strains isolated, only 7 Crohn's disease and 1 control strain had genes encoding a Pap adhesin commonly found in uropathogenic strains. In addition, genes encoding the other known virulence factors of common pathovars of E. coli such as enterotoxigenic, enteroinvasive, enterohaemorrhagic, enteropathogenic or enteroaggregative types were not detected in strains from the Crohn's cases. Subsequent ribotyping of the isolates revealed a large cluster of genetically related E. coli strains that was more frequently isolated from chronic or recurrent Crohn's disease than controls (215). Strains within this cluster from Crohn' s disease patients were also more likely to be adherent than strains from the same cluster that were isolated from controls.

Whether these findings are significant and whether E. coli strains with the diffusely adhering phenotype are pathogenic is controversial (296). Earlier studies suggested that cellular adherence is more frequent in E. coli isolated from both ulcerative colitis and Crohn's disease cases than controls (39, 117, 315). However other studies have also demonstrated the frequent isolation of E. coli strains with cellular adherence from controls (302, 379). In addition, diarrhoea per se can result in a

-26- change in the detected strains of E. coli that then revert to baseline in samples taken after resolution of the diarrhoea (307). All of these observations would be consistent with the presence of this E. coli phenotype in most individuals and increased numbers (and hence detectability) in association with any diarrhoeal illness. This situation is analogous to that of necrotoxin and haemolysin producing strains of E. coli which are more frequently isolated in ulcerative colitis cases than controls because of increased ease of cultivation rather than being a primary event in disease initiation and persistence (59). The pathogenicity of these strains must also be questioned because this pattern of adherence was frequently observed in E. coli isolates from children of the developing world with persistent diarrhoea, but these failed to produce diarrhoea when cultured and administered to healthy adult volunteers (296). Further longitudinal studies of Crohn's cases are required.

While luminal bacteria are important, the inflammatory bowel diseases are clearly based on a host genetic predisposition. In recent years there have been significant advances in this area. Evidence from twin and sibling-pair studies strongly suggests a heritable predisposition to both diseases. Concordance rates for Crohn's disease in monozygotic and dizygotic twins are estimated to be 37% and 7%, while for ulcerative colitis the estimates are 10 and 3% respectively (292). Furthermore, specific disease phenotypes such as ileal and fistulising Crohn's disease and extraintestinal manifestations show concordance. Genome-wide scanning and candidate gene studies have subsequently identified a large number of markers of IBD susceptibility, some of which are associated with specific disease phenotypes (292). The most important of these is the IBD 1locus on Chromosome 16. Single nucleotide polymorphisms (SNPs) in this gene have been shown to be associated with an approximately 2 and 42 times relative risk of Crohn's disease for heterozygotes and homozygotes respectively (146). In addition SNPs in this gene are associated with a predisposition to ileal rather than colonic Crohn' s disease. Interestingly this gene encodes the NOD2 protein that is present in monocytes and paneth cells (185) and recognises bacterial peptidoglycan (125, 248). This would tie in with an abnormality in innate immunity in Crohn's disease. Phenotypic

-27- abnormalities in the first-degree relatives of IBD cases also support the importance of heredity. For example intestinal permeability is increased in 10 to 25% of first­ degree relatives of IBD patients (253). Calprotectin, a neutrophil cytoplasmic protein that is increased in the faeces of patients with active inflammatory bowel disease is also increased in the stool of 49% of first degree relatives (347). In summary, host and bacterial factors interact in IBD pathogenesis although which bacterial factors are of importance is unclear.

A role for bacteria in irritable bowel syndrome (IBS) is more speculative. IBS is a chronic benign disorder of uncertain pathogenesis characterised by abdominal pain and/or altered bowel habit, in the absence of identifiable colonic pathology. Twelve percent of Australian adults meet the Rome Criteria for diagnosis (78, 344). These criteria are based on more than 3 months of altered bowel habit and abdominal pain. The key objective findings are altered gut motility and visceral perception that are thought to have a neurophysiologic basis (276). A role for gut bacteria in the initiation of the disease has been suggested by the fact that approximately 20% of IBS patients describe an acute onset of symptoms consistent with an initiating episode of gastroenteritis (45). One-quarter of individuals with documented bacterial enterocolitis have a disturbed bowel habit 6-months post infection and between 7 and 31% meet the diagnostic criteria forms 12-months after infection (143, 218, 236). Bacterial enterocolitis is a common illness and may therefore account for a significant proportion of the burden ofiBS (139). Interestingly it has been reported that subtle increases in the enteroendocrine cells and lymphocytes of the colonic mucosa may persist in post-dysenteric IBS long after clearance of the inciting pathogen (144, 327). Whether the post-infective symptoms might also be related to persistent changes in the bacterial populations of the colon after an infective episode is unknown. How bacterial populations might relate to the initiation of IBS and the persistence of symptoms is unclear. It is clear that there is still much to be learned about intestinal bacteria and their interactions with the host to produce colonic disease.

-28- 1.8 Molecular methods for assessing the colonic microbiota

The problems associated with culture-based assessment of the human faecal micro biota are well known. These include the fastidious nature of many colonic bacteria resulting in only 20-40% of species having been cultivated previously (186, 335, 376). In addition culture appears to have limited reproducibility. If the comprehensive method of Moore and Holdeman is applied, then the number of individual colonies that need to be identified from each sample to reliably assess significant differences between microbiotas within a given genus or for a given species, is large. It is apparent that only differences in the most frequently isolated species will be evident in such comprehensive culture-based studies.

Given these limitations, if the field is to advance further the application of other methods to studies of colonic microbial ecology is required. Molecular approaches based on 16S rDNA gene sequences have the advantage that cultivability is not required and sample storage and transport are less critical.

1.8.1 Qualitative molecular methods

Qualitative methods that produce a detailed species list or representation thereof using 16S ribosomal RNA or its encoding DNA, include PCR-cloning and PeR­ denaturing gradient gel electrophoresis (DGGE). These methods have been applied widely in studies of bacterial environments where bacteria are not cultivable by current methods (234).

1.8.1.1 PCR-denaturing gradient gel electrophoresis

To study bacterial populations by PCR-DGGE, DNA is extracted from an environmental sample and amplified with a primer set specific for the target group of bacteria. In this technique, one of the primers has a 30 to 40 base sequence of guanine and cytosine nucleotides (known as a GC-clamp) at its 5' end. Following

-29- amplification, the PCR product is electrophoresed in a polyacrylamide gel containing an increasing gradient of denaturants in the direction of electrophoresis. As the double stranded product passes through the gel, denaturation occurs resulting in greatly diminished electrophoretic mobility (95). The point at which denaturation occurs is dependent on the sequence of the PCR product in the melting domain that has the lowest denaturation temperature; which in the case of 16S rDNA may reflect strain, species or genus differences (234). The GC-clamp prevents complete denaturation of the double stranded product (235). In this way a banding profile is generated which represents the genospecies in each sample (234). The advantage of PCR-DGGE with universal primers, i.e. primers for all bacteria, is the stable banding patterns produced when the method is applied to faecal samples collected from one adult human subject over time (291, 388). Given this, events that alter the faecal microbiota from baseline may be easily detected by this method. For example, following antibiotic therapy dramatic changes in the PCR-DOGE patterns generated from faecal DNA with universal primers can be observed (75). Although relatively quick and highly reproducible, PCR and gradient gel electrophoresis does have a number of limitations (388). DOGE band positions are often altered by single base differences in the first melting domain, whereas strains within a species might vary by 12 or more base differences over a 600 base PCR product (235). In addition a single DOGE band position may represent multiple bacterial species (234, 309). In a study of DNA from river water amplified with universal bacterial primers for the V3 region of 16S rRNA, a single band position represented 7 different genospecies (309). A single bacterial species can also produce multiple DOGE bands due to the presence of multiple 16S rDNA copies or errors introduced by non-proof-reading Taq DNA polymerase (56, 171, 244). Despite these difficulties, PCR-DGGE is a powerful method for the detection of changes in bacterial composition of samples over time. The addition of a new species to a mixed population such that the new species comprises only 1% of the total population may still be detectable by PCR­ DGGE (234). The banding pattern produced by PCR-DGGE is a description of the species and strains of bacteria in a given sample, however whether band intensity is a reliable measure of relative bacterial numbers is a controversial issue (37, 88).

-30- 1.8.1.2 PCR-cloning

PCR-cloning and sequencing generates a library of 16S ribosomal RNA sequences from environmental samples (28, 154, 335). Comparison of these sequences with public databases allows the putative identification of the bacterial species present in the sample. This is extremely valuable for the study of bacteria that cannot be cultivated. However the method is laborious and expensive. In addition there are many potential biases in PCR-cloning relating to differences in the efficiency of cell lysis and PCR amplification from different bacterial species (377). This approach generates a long list of the species present in a sample, like the method of Moore and Holdeman in their culture-based studies (155, 226, 230). Making meaningful statistical comparisons of bacterial populations based on clone libraries is challenging and requires the sequencing of a very large number of clones (28, 34, 113, 324). PCR-cloning is discussed in more detail in Chapter 8.

The qualitative methods of cloning and DGGE are based on PCR amplification that introduces a host of potential biases into the assessment of bacterial ecology. Amplification is necessary to increase the number of 16S rRNA copies to a level that is detectable by gel electrophoresis or can be efficiently ligated into a vector for cloning. It is known that PCR results in significant quantitative biases due to preferential amplification when mixed 16S rRNA template is used, particularly as the cycle number increases (339, 376, 388). For example, in a study by Wilson and Blitchington (1996) which used PCR and cloning to study the human faecal microbiota, a single low G+C 16S rDNA sequence made up 25% of clones after a nine cycle PCR and 42% of clones after 35 cycles (376). An additional example of preferential amplification is the fact that bacterial sequences representing a large proportion of the microbiota may not be represented after PCR despite being perfectly matched to the primer set. For example Bacteroides spp. were not represented on gels in the study of human faeces by Zoetendal et al (388). PCR amplification of mixed templates with significant homology may also be associated

- 31- with the formation of chimeric molecules, where PCR product is composed of parts of 2 different template sequences ( 17 5). Chimeras are thought to occur in the later cycles of PCR when partially elongated sequences compete with primer for binding sites on template molecules (369). In spite of these difficulties some authors have suggested that PCR-DGGE represents the "dominant" bacterial flora in a sample, but given the biases above whether this is accurate is unclear (234, 388).

Several additional important caveats specifically relate to the interpretation of 16S rRNA data. While bacteria belonging to the same genus generally have more than 94% 16S rDNA sequence homology, and bacteria from the same species greater than 97% homology (363), this is by no means universally true. The misidentification of bacterial species using complete 16S rDNA sequences is well described in the literature (362). This is because bacteria with less than 97.5% 16S rDNA sequence similarity are highly unlikely to represent the same species but bacteria with greater than 99% 16S rDNA sequence identity are not necessarily members of the same species (328). That is, 16S rDNA sequence comparisons can more reliably be used to confirm that 2 bacteria are not related than the converse. Nevertheless an arbitrary cut-off of around 98% sequence homology is often used to define an Operational Taxonomic Unit (OTU or molecular equivalent of a species) in molecular studies (28, 335).

An additional difficulty for molecular studies of colonic bacteria arises because a majority of the 16S rDNA sequences of colonic bacteria fall within the heterogeneous taxon Clostridium (130). In phylogenetic trees, clostridial sequences are interspersed with the sequences of bowel bacteria that belong to other named genera such as Fusobacterium, Ruminococcus, and Eubacterium (335). Thus 16S rDNA sequences have limited power to discriminate between these bacterial genera and the genus Clostridium, increasing the probability that a sequence will be misclassified based on 16S rDNA data alone (57, 335). There is little correlation between phenotypic characteristics and phylogenetic classification within this genus and a major taxonomic revision has been suggested (57).

-32- 1.8.2 Quantitative molecular methods

Bacterial group quantitation based on 16S rRNA can be achieved by in situ hybridisation of whole cells with fluorescent probes (110, 148, 149, 164) or using extracted RNA and radiolabelled probes (213, 310, 313). The key consideration in these studies is developing a probe that is specific for the target group of organisms compared with other bacteria in the sample, under the conditions used in the assay.

Dot-blot hybridisation with radiolabelled probes allows a direct quantitative assessment of bacterial rRNA within a probe-targeted group, as compared with total rRNA assessed with a probe for the domain Bacteria (213, 310, 313). This method has a number of advantages. Firstly, each sample is generally quantitated in triplicate and reported as a mean ± standard deviation which gives an indication of the reproducibility of the method (213, 310, 313). Secondly, the quantitation of relative bacterial numbers is substantially less labour intensive than manual counting for FISH. However one potential bias with this technique is that some bacteria posses cell walls that are resistant to lysis (e.g. some gram-positive bacteria) and may be under-represented in the extracted RNA. However this is thought to represent only a minor bias in studies of gut bacteria (76).

Bacterial quantitation using FISH involves hybridising fixed bacterial cells with fluorescently-labelled oligonucleotides and counting them under epifluorescence or confocal microscopy. A number of authors have published descriptions of automated counting techniques for use with FISH and human faecal samples (148, 164). These were shown to be a reliable method for the quantitation of the major groups of bacteria present in each sample (148, 164). However, several factors can affect the results of FISH, particularly the resulting signal or brightness. Probe design is complicated by the varying accessibility of different regions of the ribosome to hybridisation with a fluorescent oligonucleotide (17, 111). This means that probes designed to target certain regions are often poorly functional in practice (17, 111).

-33- Cell wall permeability can also affect the results of FISH. This applies particularly to gram-positive cells such as streptococci, enterococci and lactobacilli that need to be treated with lysozyme to permeablise them prior to hybridisation (110, 231). Brightness is also dependent on the rate of protein synthesis (i.e. growth) in the target cells. Actively growing cells are brighter than older cultures because of their higher 16S rRNA content (262). For the best results, samples must be fixed immediately after collection and stored at -20°C or lower until processed, because the ribosomal RNA degrades over time. Measurements of the main groups of bacteria in faeces using FISH are reproducible (148, 164). The coefficient of variation of automated assays is around 0.2 (range 0.07 to 0.28 depending on the probe) (148, 164). At present at least 15 probes for major subgroups of the human faecal flora have been published and successfully applied (110, 148). However, bacterial groups that are present in small numbers only in a sample may not be reliably quantitated (148, 164).

Despite these problems molecular approaches based on 16S rDNA have greatly expanded our knowledge of human colonic bacterial ecology, and will be the basis for studies of disease association in the future. PCR-cloning is a useful tool for describing bacterial diversity while PCR-DGGE is invaluable for detecting changes over time. When the quantitative methods of dot-blot hybridisation and FISH are used appropriately accurate quantitation of major subgroups of the microbiota is possible.

Three additional specific areas of the literature on colonic bacterial ecology are relevant to the work that follows. These are the role of Helicobacter species in human IBD, colonisation resistance and probiotic therapies. Each of these is discussed below.

-34- 1.9 Helicobacter species and human ffiD

A question remains as to whether a specific bacterial population colonises the mucus layer lining the human colon that is distinct from the luminal bacterial populations. This ecological niche has been clearly identified to have its own bacterial species in animals (190, 257, 294, 343) and is often assumed to be present in discussions of the human colon by analogy (202, 390). However, comprehensive culture of the bacteria associated with the gut wall and the lumen of the colon from human cadavers have not shown any differences between these sites (227, 228). The results of this study have not been reported in detail in the literature and have only been alluded to in passing (227, 228). Molecular studies using 16S rRNA PCR-cloning have also failed to show significant differences in clone libraries obtained from the faeces and washed colonic biopsies of the same subject (154). In contrast, examination of the banding patterns obtained with PCR-DGGE using universal16S rRNA primers have shown different banding patterns to be present for faecal samples and colonic biopsies obtained from the prepared colon of individual subjects (390). Further evidence questioning the presence of a mucus-adapted microbiota in humans ~.omes from 2 FISH studies. A report of FISH performed on cryostat sections of endoscopic biopsies taken from lavaged and unlavaged patients failed to demonstrate bacteria populating the inner layers of the human colonic mucus or crypts (361). Swidinski et al reported similar fmdings in washed and unwashed biopsies from the human small bowel and colon (340). In summary, to date based on the very limited data that is available, there is minimal evidence to support the presence of a mucus-adapted colonic microbiota in humans.

Several authors have speculated on a potential role for mucus-adapted Helicobacter species in the aetiology of human IBD (100, 196, 249, 349). The mucus layer that is adjacent to the epithelium of the large bowel of animals is often colonised by bacterial species that are phenotypically distinct from luminal populations. Subgroups of these bacteria are adapted to specific regions of the bowel, with different phenotypes found in the terminal ileum, caecum and colon of the rat (257).

-35- Studies of this mucus-associated microbiota have shown that many of these bacteria belong to the Helicobacter genus, a subgroup of the epsilon subclass of Proteobacteria. Helicobacter species are microaerophilic or anaerobic, fusiform, curved or spiral bacteria that are specifically adapted for motility at the interfaces of mucus layers (382). When Helicobacter species are present in the colon of animals, they are often observed in large numbers within the surface and crypt mucus of the large bowel (103, 190) and are generally very effectively transmitted by the faecal­ oral route (197, 374). In their host species lower bowel helicobacters are usually commensals, however they can be associated with chronic bowel inflammation in immunocompromised host strains (104, 183, 194) and cause transient bacterial colitis if acquired by another species (115, 267, 350).

There are three lines of evidence that suggest that Helicobacter species colonise the human colon. Firstly, bacteria with a spiral morphology have been observed in the mucus layer adjacent to the colonic mucosa of2 out of 4 individuals whose samples were collected within 6 hours of death, and examined by scanning electron microscopy (61). However, these bacteria had a long spiral morphology that was more suggestive of spirochetes than Helicobacter species.

Secondly, Helicobacter species have been isolated from the stools of humans with acute diarrhoeal illnesses. These include Helicobacter cinaedi, Helicobacter fennelliae, Helicobacter canis, Helicobacter pullorum, and "Helicobacter rappini" (246). H. cinaedi and H. fennelliae were first isolated from a group of homosexual males with anorectal and intestinal symptoms attending a sexually transmitted disease clinic in Seattle (267, 350). The sigmoidoscopic and histologic findings in these men were similar to those found with Campylobacter jejuni infection (266) and subsequently animal studies have confirmed the pathogenic nature of the organisms (96). It is likely that large bowel helicobacter infection was first demonstrated in this population because homosexual males have an increased risk of acquiring bacterial intestinal pathogens as a result of their sexual practices. Subsequently, H. cinaedi has been recognised to be part of the commensal

-36- microbiota of hamsters, making it likely that the outbreak in Seattle originated as a zoonosis (115). Other zoonotic cases of human diarrhoeal illness associated with H. canis and H. pullorum (whose known hosts are dogs and chickens respectively) have also appeared in the literature (329, 330). Although current evidence suggests that animal Helicobacter species can transiently infect the bowel of humans whether long-term colonisation can occur is unknown.

Thirdly, helicobacter DNA has been detected in human colonic tissue samples (27, 196, 249, 349). In 2000, Linskens et al reported the detection of 16S rDNA from the Helicobacter genus by PCR in colonic biopsies from 9 of 18 healthy persons, 6 of 9 ulcerative colitis cases, 6 of 10 Crohn's cases and 4 of7 patients with diverticulosis (196). These authors could not detect helicobacter DNA in stool samples. PCR with Helicobacter bilis specific primers was positive in all of the biopsy samples in which helicobacter DNA had been detected using the genus-specific PCR raising the possibility that a Helicobacter species was colonising the colonic mucus of a large proportion of humans. Subsequently, these authors conceded that their Helicobacter PCR was not specific and would amplify 16S rDNA from other bacterial species (J. Kusters, personal communication). Using a different primer set, Bohr et al detected helicobacter 16S rDNA in colonic biopsies from 4 of 10 subjects- 3 of whom had documented concurrent gastric Helicobacter pylori infection. This result suggests that wash-through of helicobacter DNA of gastric origin occurred during bowel preparation (27). In a third study, helicobacter DNA was detected by southern blot hybridisation in 16S rDNA PCR product generated from endoscopic or surgical biopsies of the ileum in 8 of 11 Crohn' s disease cases and 4 of 11 controls (349). Finally Helicobacter DNA was recently identified in colonic biopsies from 8 of 42 ulcerative colitis cases and 7 of 74 controls using nested PCR by Oliveira et al (2004) (249). However in this study, the 16S rDNA sequences that were obtained showed greater than 99% sequence homology to H. pylori, and this organism was cultured from the colonic samples of 3 of the ulcer.ative colitis cases and 1 control. This finding would again suggest that wash-down of gastric organisms was responsible for the positive results.

-37- The principal argument for the possible role of a commensal Helicobacter species in human IBD is the evidence from murine models. The lower bowel of laboratory mice from commercial and research colonies is often colonised with Helicobacter species (107, 314). Three of the 8 officially named murine lower bowel species have been associated with inflammation of the caecum and colon with or without concomitant hepatitis - Helicobacter hepaticus, H. bilis and Helicobacter typhlonius (102, 104, 106, 107, 109, 320, 372). In a situation analogous to current theories of IBD pathogenesis, Helicobacter species are generally associated with colitis in murine strains that are genetically predisposed to the development of inflammation. Such strains include interleukin-10 deficient mice (183), BandT lymphocyte deficient (SCID) mice (106, 194), T lymphocyte deficient strains (371) and NF­ kappaB deficient mice (85). It is important to realise that these murine strains will often develop colitis unless maintained in a germ-free state, however inflammation is more severe in mice with a microbiota that includes Helicobacter species than in those with background microbiota alone (183). The potentially pathogenic nature of lower bowel Helicobacter species is supported by the presence of virulence factors encoded within their genomes. For example, sequencing of the genome of H. hepaticus, the species most frequently associated with colitis, has revealed multiple virulence factors (336). These include a candidate adhesin (HH1481) with homology to PEB1 of Campylobacter jejuni, and a genomic island, HHG1, containing proteins with homology to structural components of Type IV secretion systems (336). However, to date Koch's postulates have only been satisfied for the link between H. hepaticus and hepatitis in A/JCr mice (372) and not all studies support a role for Helicobacter species in the induction of colitis (74, 203).

In addition to the evidence from studies of mice, Helicobacter species have been cultured from faecal samples obtained from Cotton-Top Tamarins held in captivity that spontaneously develop colitis (293). In a study by Saunders et al (1999) bacteria of the "Helicobacter rappini" morphotype (i.e. fusiform with periplasmic fibres and bipolar flagella) were cultured from faecal samples of 8 of 34 Tamarins and detected

-38- in colonic biopsies from 18. This Helicobacter species was closely related to H. fennelliae but it has been suggested that this species probably requires its own taxon within the genus Helicobacter (70). While it is possible that this organism is related to the initiation or perpetuation of colitis in these animals, disease association and Koch's postulates were not tested. To summarise at present it remains unclear whether Helicobacter species colonise the human colon or are involved in the initiation of IBD.

1.10 Colonisation Resistance

Colonisation refers to the persistence of viable organisms in the gut. For practical purposes, the presence of detectable quantities of organisms 2 weeks or more after oral administration has ceased, is a useful definition. Other authors have defined colonisation as a rate of proliferation of an organism in the gut that is at least equal to the rate of loss (202). Colonisation resistance refers to the fact that the oral administration of viable bacteria to adults usually does not result in the persistence of those bacteria within the large intestine of the host for periods exceeding 2 weeks (26, 358). This concept was originally coined by van der Waaij et al (1971), who initially defined colonisation resistance as the logarithm of the oral dose of bacteria required to produce a "persistent take" in 50% of mice (358).

Subsequently, van der Waaij and Berghuis also proposed that the colonisation resistance of individual mice could be expressed as, the base 10 logarithm of the concentration of test bacteria in faeces, at any time during the 2 weeks after oral administration (357). This conclusion was based on the linear relationship that van der Waaij and Berghuis observed between the percentage of mice within a group excreting streptomycin-resistant (s.r.) E. coli after oral dosing, and the base 10 logarithm of the mean s.r.-E. coli concentration cultured from the faecal samples of that group. In their study the results for many groups of mice were pooled regardless of the timing of sampling after the dose or the dose administered. As a result the authors argued that this linear relationship was independent of the dose of s.r.-E. coli

-39- administered and the timing of the sample post dose, because this relationship was observed in pooled data. However this data was based on mean concentrations for groups of mice and not individuals. In addition, the study's results contradict this proposition as the mean concentration of test bacteria in the faeces of individual mice is clearly related to the dose administered and the time at which the faecal sample was taken after dosing (357, 358). That is the reported colonisation resistance of an individual mouse would be reported as low after dosing (with lower resistance reported for higher doses) and increase over time. Therefore van der Waaij's generalisation seems unjustified.

An extension of this flawed concept arose from data obtained from 8 of the mice - four of which had been lethally irradiated. This showed that the number of biotypes of gram negative bacilli that were isolated from faeces, and the faecal concentration of orally administered s.r.-E. coli both increased in tandem. This led to the proposition that an individual's colonisation resistance could be measured by quantitative biotyping of Enterobacteriaceae in faeces. That is, the presence of a higher number of biotypes indicated a reduced colonisation resistance. In this ~tudy biotyping of the gram-negative bacilli was undertaken with biochemical testing. This concept has led to a very confusing literature on colonisation resistance and its inappropriate measurement by quantitation of Enterobacteriaceae biotypes in stool (366).

Colonisation resistance can be dramatically reduced in a number of ways including antibiotic treatment, cathartic administration, irradiation and caecectomy (26, 358, 367). Several studies have examined the effect of antibiotic treatment (26, 358). For example, 102 E. coli, P. aeruginosa, Salmonella enteriditis and Klebsiella pneumoniae colonised all of the mice that had been treated with streptomycin and/or neomycin, whereas in mice with an intact conventional micro biota doses of 109 P. aeruginosa and K. pneumoniae failed to colonise and doses of 105 of E. coli and S. enteriditis were required to produce colonisation (26, 358). In the studies of van der Waaij et al, colonisation resistance began to recover within 7 days after ceasing

-40- antibiotics and this correlated with the clearance of antibiotics from the gut (299, 358-360). A regeneration of the microbiota, similar to that seen during the initial colonisation of the newborn mouse, was observed (299, 358-360). Radiation treatment also reduces colonisation resistance, perhaps through a direct antibacterial effect (357, 358) and the administration of the cathartic MgC03 has been shown to increase susceptibility to S. enteriditis (26). Surgical caecectomy is associated with an increased mortality during S. enteriditis infection and may also reduce colonisation resistance (367). Although these methods have not been directly compared, antibiotics appear to be the most effective method for reducing colonisation resistance.

Several published experimental results suggest that the concept of colonisation resistance also applies to humans. For example, a neomycin and gentamicin resistant

11 strain of E. coli that was orally administered at a dose of 10 , was detectable in faeces from volunteers prepared with colonic lavage fluid containing neomycin and metronidazole for more than 10 weeks, but for less than 1 week in subjects treated with lavage alone and unprepared subjects (356). In another study, specific serotypes of E. coli taken orally by volunteers at a dose of 106 could not be detected when they were evaluated 4 weeks later (308). In contrast, another human study suggested that orally administered neomycin resistant E. coli were often detectable for greater than 6 months in unprepared subjects (8). In this study, 4 of 10 volunteers who ingested 109 neomycin resistant E. coli had detectable antibiotic resistant E. coli in their faeces 300 days later. Like mice, bacterial species that are not part of the normal microbiota of humans e.g. K. pneumoniae and Enterobacter cloacae do not colonise the bowel after oral exposure (366). However, colonisation with these strains for up to 3 weeks may occur if subjects are prepared with antibiotics (366). Overall these studies suggest that the concept of colonisation resistance does apply to humans and that antibiotics, with or without colonic lavage, reduce colonisation resistance.

-41- Many of these investigations of colonisation resistance have used Enterobacteriaceae as the test species because many gastrointestinal bacterial pathogens are from this group. Whether the concept of colonisation resistance applies to other bacterial species including strict anaerobes is unknown (26, 357, 358, 367).

It is often asserted that colonisation resistance is mediated by caecal anaerobes or their SCFA products, based on the observation that the introduction of caecal anaerobes to the microbiota results in decreases in coliform numbers where these have occupied the caecum previously (26, 360, 367). However there is some evidence to suggest that a complete microbiota is necessary to mediate colonisation resistance. For example, in a study of murine susceptibility to Listeria monocytogenes , the presence of a complete specific pathogen free (SPF) flora was protective against infection while colonisation with one murine Clostridium species and Bacteroides species was not (385). Further research is needed to identify which bacterial species mediate colonisation resistance to individual introduced strains of bacteria in humans.

1.11 Probiotics and colonic disease

The potential for luminal bacteria to have beneficial effects on the host is currently being extensively investigated. Probiotics have been defined as live microbial food supplements that alter the composition or metabolic activities of the microbiota, or modulate the immune system reactivity in a way that benefits health beyond simple nutritional effects (140, 199). Various strains of Lactobacillus spp., Bifidobacterium spp., Streptococcus spp. and E. coli are purported to have these therapeutic benefits (199). There is evidence for the efficacy of a large variety ofprobiotic therapies in a broad-range of apparently unrelated clinical disorders (such as eczema, traveller's diarrhoea, prevention of neonatal rotavirus outbreaks, antibiotic-associated diarrhoea, ulcerative colitis, and bacterial vaginosis) (132, 199, 259, 286).

-42- The mechanisms by which the therapeutic benefits of probiotics are derived in vivo have not been elucidated, however in vitro studies have suggested several possibilities. Lactobacillus species have been reported to inhibit bacterial adherence to colonic epithelial cells in vitro (191). In addition, different Lactobacillus species can induce different patterns of cytokine expression by murine bone marrow dendritic cells, depending on the bacterial species tested (50). As cytokine expression by dendritic cells may direct individual T -cell responses towards Th1, Th2 or Th3 types this indicates a putative immunomodulatory mechanism for probiotic therapies. For example, irradiated Lactobacillus reuteri inhibited the induction of interleukin 12 (ll...-12), interleukin 6 (ll...-6) and Tumour Necrosis Factor alpha (TNF-a) secretion by cultured dendritic cells in response to Lactobacillus casei and other Lactobacillus species (50). To put this in context however, lysed bacteria and bacterial DNA per se are also known to have immunomodulatory effects. For example, the administration of these bacterial products has been shown to ameliorate colitis in chemically-induced and cytokine-knockout murine models of colitis (173, 268). At this time there is only limited direct evidence for in vivo modulation of the microbiota or immune responses, by viable probiotic bacteria, in order to provide a mechanistic explanation for their observed therapeutic effects (345, 365).

A number of studies have confirmed that the probiotics studied to date do not colonise the colon, as assessed by their absence once oral administration is discontinued (132, 345, 365, 368) and their lack of proliferation in vivo relative to an inert marker (250). Despite the absence of colonisation with current probiotics, a number of studies have demonstrated that probiotic bacteria will survive transit through the gut, and can be cultured in the faeces of a majority of subjects during the period of daily intake (122, 123, 345, 368). In addition, the synthesis of protein in vivo has been demonstrated by Oozer et al using a luminescent strain of L. casei in human flora-associated mice (250). Thus it would appear that many probiotics are

-43- viable in the colon, however whether they proliferate or reach the intestinal mucosa to interact directly with the host is unknown (250).

The best evidence for the efficacy of probiotics in human colonic disease is found in studies of pouchitis (122, 123). Pouchitis is the chronic inflammation of the ileal reservoir (or pouch) that is constructed after proctocolectomy for patients with ulcerative colitis. In a randomised double-blind placebo-controlled study of 40 individuals with a history of pouchitis, daily ingestion of a mixture known as VSL#3 containing 4 strains of lactobacilli, 3 strains of bifidobacteria and 1 of Streptococcus salivarius subspthermophilus reduced the relapse rate of pouchitis, after antibiotic induced remission, from 100% to 15% during nine months of follow up (123). All of the patients in remission at the end of the study relapsed within 4 months after the cessation of the probiotic supplement (124). A subsequent preventative study, in which VSL#3 was commenced at the time of ileostomy closure in ulcerative colitis cases with a pouch, revealed that pouchitis occurred in 40% of the placebo group and only 10% of the treated patients during 12 months of follow up (122). There is also some evidence for the efficacy of other probiotics in human IBD (180, 273) particularly in the maintenance of remission, rather than remission induction. In contrast to studies suggesting that probiotics may be efficacious in inflammatory bowel disease, probiotic therapies have not been convincingly shown to be beneficial in the irritable bowel syndrome (172, 241, 312).

In summary the role of probiotic therapies in the treatment of human colonic diseases and their mechanisms of action in vivo remain to be fully elucidated. There is preliminary evidence for significant efficacy in some subgroups of IBD patients. Interestingly, the studies that report the most convincing benefit for probiotics have administered the bacteria after the use of antibiotics (123).

-44- 1.12 Overview, Hypotheses and Aims

Clearly there are still substantial gaps in our knowledge of the bacterial species that are present in the human colon, their metabolic activity, interaction with the host and impact on colonic disease. The application of molecular methods to the study of this field has led to significant advances but a great deal of work remains to be done. The overall goal of this work was to expand the molecular approaches available to study the human and murine colonic microbiota and to apply these methods to studies of colonisation resistance in mice and humans.

Hypotheses

1. Specific colonic bacterial species are important in the pathogenesis of inflammatory bowel disease in mouse models and humans. In particular, commensal Helicobacter species induce IBD in genetically susceptible individuals. 2. Faeces and caecal contents from the one individual contain the same bacterial species. 3. Durable changes in host caecal microbiota can be induced by the administration of bacterial strains adapted to the colon during antibiotic and cathartic induced disruption of colonisation resistance. Changing the caecal microbiota might be of benefit in IBD.

Aims

1. To develop PCR-DGGE methods for assessment of the species compositibn of the major anaerobic groups of the micro biota of the large intestine.

2. To develop PCR-DGGE methods targeting lower bowel Helicobacter species.

-45- 3. To determine whether lower bowel Helicobacter species play a role in the pathogenesis of human ulcerative colitis or Crohn's disease.

4. To assess whether faecal samples represent the species composition present in the caecal micro biota of mice.

5. To determine whether colonisation resistance for anaerobic bacterial species can be reduced with antibiotics and cathartics in mice.

6. To determine whether colonisation resistance for anaerobic bacterial species can be reduced with antibiotics and cathartics in humans.

-46- TABLE 1.1: The phylogenetic placement of clones from the study of Suau et al (335) within the RDP database.

Group RDP registration No. of No. of No. of Number1 clones OTUs novel2 OTUs Bacteroides-prevotella group 2.15.1.2 88 20 13 Clostridium coccoides group 2.30.9.1 125 31 24 Clostridium leptum subgroup 2.30.10.1.2 57 20 17 Mycoplasma spp. and relatives 2.30.8 4 3 3 Phascolarctobacterium faecium 2.30.3.1.4 2 2 1 subgroup (Sporomusa group) Clostridium therrnocellum 2.30.9.1.1 1 1 1 subgroup Clostridium propionicum subgroup 2.30.4.2.1 1 1 1 Atopobium group 2.30.1.5 1 1 1 Eubacterium group 2.30.4.4 1 1 1 Prosthecobacter group 2.10 1 1 1 Streptococcus salivarius subgroup 2.30.7.21.6 1 1 Streptococcus pneumoniae 2.30.7.4.9 1 1 subgroup

1updated RDP registrations from RDP IT. 2novel OTUs are those with greater than 2% sequence divergence from the sequences of named bacteria in GenBank.

-47- TABLE 1.2: Percent of the faecal microbiota by target group for 11 adults (148).

Target bacterial Group Probe Cells/gram of faeces %of dry weight(SD) microbiota1 11 Total bacteria Bact338 1.3(0.6) X 10 100

10 Bacteroides/Prevotella Bac303 3.6(2.3) X 10 27.7

10 Eubacterium rectale/ C. Erec482 2.9(1.9) X 10 22.7 coccoides

2 10 Eubacterium low G+C#2 Group Elgc01 1.4(1.2) X 10 10.8

2 10 Ruminococcus group Rbro729/ 1.4(1.6) X 10 10.3 Rfla730 10 Atopobium group Ato291 1.4(0.9) X 10 11.9

9 Bifidobacterium Bif164 6.0(4.0) X 10 4.8 9 Eubacterium cylindroides Ecyl387 1.5(2.8) X 10 1.4

8 Phascolarctobacterium Phasco741 9.0(15) X 10 0.6 8 Enterobacteriaceae Ecoli1531 3.2(9.1) X 10 0.2 8 Veillonella Vefil223 1.0(2.3) X 10 0.08

7 Lactobacillus/Enterococcus Lab158 1.2(2.6) X 10 0.01

1The denominator used in the calculation of the percentage of the microbiota is the total sum of bacterial cells fluorescing with the eubacterial probe Bact338. 2The species that are targeted by the probes for Eubacterium low G+C#2 and Ruminococcus group are found within the C. leptum subgroup of the RDP.

-48- CHAPTER 2: Materials and Methods

2.1 METHODS

2.1.1 Bacterial Culture

2.1.1.1 Culture of reference and laboratory strains of Helicobacter and Campylobacter species

The following reference strains were used:

Helicobacter bilis * ATCC51630

Helicobacter hepaticus # ATCC 51448 Helicobacter muridarum * ATCC49282

Helicobacter rodentium # ATCC 700285 Helicobacter trogontum * ATCC 700114

The following laboratory strains were used:

Campylobacter coli* Campylobacter fetus *

Two strains of Helicobacter ganmanzo~~ were used. These were Helicobacter ganmani directly isolated from the caecum of interleukin 10 deficient (IL-l o-'-) C57BL/6 mice and wild-type C57BL/6 mice respectively, in the School of Biotechnology and Biomolecular Science animal facility.

Helicobacter and Campylobacter species were cultured at 37°C for 3 to 7 days on Horse Blood Agar (HBA) containing 3.7% Blood agar base No. 2 (Oxoid, Basingstoke, Hampshire, England), 5% sterile defibrinated horse blood (Oxoid) and 2.5 mg!L amphotericin B (Bristol-Myers Squibb, Princeton, New Jersey) under anaerobic# or microaerophilic* conditions in anaerobic jars (Oxoid). Anaerobic conditions were attained using an Anaerobic system gas generating kit (Oxoid) and

-49- catalyst. Microaerophilic conditions were attained with a Campylobacter gas generating kit (Oxoid).

2.1.1.2 Culture of reference and laboratory strains of other bacterial species

Bifidobacterium longum ATCC 15707 Clostridium leptum DSM753 Fusobacterium mortiferum ATCC25557 Lactobacillus salivarius ATCC 11741

The following strains were obtained from the University of New South Wales School of Microbiology Culture Collection:

Bacillus cereus 052300 Bacteroides fragilis 035100 Bacteroides vulgatus 035202 Bifidobacterium adolescentis 509400 Clostridium histolyticum 057600 Clostridium nexile 096900 Desulfovibrio desulfuricans 060900 Enterococcus faecalis 054400 Escherichia coli Kl2 002900 Eubacterium limosum 041600 Lactobacillus acidophilus 060800 P eptostreptococcus anaerobilus 041900 Propionibacterium acnes 042200

E. coli, E. faecalis and B. cereus were incubated aerobically at room temperature on HBA for 48 hours. With the exception of B. adolescentis, D. desulfuricans, C. nexile and C. leptum the remainder were incubated under anaerobic conditions on HBA for 48 hours at 37°C. Anaerobic conditions were attained using Anaerobic system gas

-50- generating kit (Oxoid), catalyst and anaerobic jars (Oxoid). C. leptum and C. nexile were cultivated in cooked meat medium for 72 hours under anaerobic conditions at 37°C.

2.1.1.3 Culture of Desulfovibrio desulfuricans

Stock cultures of D. desulfuricans were grown in Postgate's Medium E in sterile McCartney bottles in an anaerobic chamber at 37°C for 5 days (200).

Postgate's medium E was made by mixing all of the components listed below, adjusting the pH to 7.4 and autoclaving for 15 minutes at 121°C.

K2HP04 (Asia Pacific Speciality Chemicals (APS), Seven Hills, NSW) 0.5 g NH4Cl (APS) 1.0 g NazS04 (APS) 1.0 g

CaC12.2H20 (Merck, Darmstadt, Germany) 0.1 g

MgS04.7H20 (APS) 2.0 g 70% Sodium Lactate (BDH Laboratory Supplies, Poole, UK) 7.2 ml

FeS04.7H20 (APS) 0.5 g Mercaptoacetic acid, sodium salt (Sigma, St. Louis, Missouri, USA) 0.1 g L-Ascorbic acid (BDH Laboratory Supplies) 0.1 g Yeast Extract (Oxoid) 1.0 g

Made up to 1L with reverse osmosis water.

2.1.1.4 Culture of B. adolescentis

Reinforced Clostridial Agar (RCA, Oxoid) was made according to the manufacturer's instructions. B. adolescentis was cultured on RCA under anaerobic conditions for 72 hours at 37°C.

-51- 2.1.1.5 Quantitation of bacteria harvested from plate cultures

A 1 in 100 dilution of the harvested bacteria in phosphate buffered saline (PBS) was placed on an Improved Neubauer Haemocytometer. The mean number of bacteria in each of 10 squares was multiplied by 4 x 106 per m1 to obtain an estimate of concentration in the 1 in i 00 dilution of suspension.

2.1.1.6 Culture of Helicobacter species from murine caecum and faeces, and human colonic biopsies

Samples were cultured on Campylobacter Selective Agar (CSA) and HBA under microaerophilic and anaerobic conditions. CSA plates were prepared as described for HBA with the addition of Skirrow's supplement (2 milL) at the same time as the amphotericin. Skirrow' s supplement consists of filter sterilised distilled water containing 5 mg/ml vancomycin (Eli Lilly, West Ryde, NSW), 1250 U/ml polymyxin B (Sigma) and 2.5 mg/ml trimethoprim (Sigma). Filter sterilisation was performed using a 0.2 J.Lm filter (Nalge Nunc International, Rochester, NY).

In the case of murine caecal tissue, the caecum was opened and scraped with a No. 11 scalpel blade. Murine faecal samples were made into a slurry in sterile PBS by macerating each with a 20 J.Ll pipette tip followed by vortexing for 30 seconds. For culture on HBA, the murine caecal scrapings, murine faecal slurry and human colonic biopsies were loaded onto 0.45 J.Lm filters (Millipore Corporation, Bedford, MA) that had been placed on the surface of the culture plates. After 2 hours incubation at 37°C under microaerophilic conditions, the filters were removed and the plates were then re-incubated for 3 to 7 days under microaerophilic and anaerobic conditions. For culture on CSA, human colonic biopsies were placed directly on the plates and incubated for 3 to 7 days under microaerophilic and anaerobic conditions. After 3 days, a loopful of the resulting growth was examined by phase contrast microscopy and subcultured if bacteria with a helicobacter-like morphology and motility were visible.

-52- 2.1.2 DNA extraction

2.1.2.1 DNA extraction from plate cultures

Bacteria were harvested from plates in sterile PBS. After centrifugation for 5 minutes at 7200 x g, DNA was isolated from the resulting pellet according to the gram-negative bacteria protocol of the Puregene DNA purification kit (Gentra systems, Minneapolis, Minnesota).

2.1.2.2 DNA extraction from murine and human samples

Human faecal samples were transported to the laboratory at -20°C for analysis. The samples were thawed on ice and a 1 g aliquot was added to 10 ml of sterile PBS, vigorously shaken and then vortexed until a slurry had formed. Particulate matter was allowed to settle by leaving the sample on ice for 10 seconds. One and a half ml aliquots of this slurry were stored at -80°C for batch processing. At batch processing, 1 ml of thawed slurry was transferred into a 1.5 ml tube and centrifuged at 7200 x g for 5 minutes. The supernatant was discarded, and DNA was isolated from the pellet resuspended in cell lysis solution (Gentra systems) using the animal tissue protocol of the Pure gene DNA purification kit per the manufacturer's instructions (Gentra systems).

Murine faecal pellets and caecal tips were vortexed into a slurry in 600 J1l of cell lysis solution containing 0.01% Proteinase K (Sigma) at 55°C. Three 2 minute episodes of vortexing over 30 minutes were required to completely break down the pellets. Once a slurry had formed, the samples were then incubated for 30 minutes at 55°C. DNA was then extracted with the Puregene DNA purification kit according to the animal tissue protocol and the manufacturer's instructions.

-53- DNA was isolated from human colonic biopsies and murine caecal and colonic tissue samples after overnight incubation in cell lysis solution containing 0.01% Proteinase K at 55°C using the animal tissue protocol of the Puregene DNA purification Kit per the manufacturer's instructions.

2.1.2.3 DNA quantitation

DNA was quantitated with UV spectrophotometry by the comparison of absorbances at 260 and 280 nm.

2.1.3 PCR amplification

2.1.3.1 Development of novel primer sets

Primer sets for use in PCR-denaturing gradient gel electrophoresis (PCR-DOGE) were developed to target specific phylogenetic groups based on the Ribosomal Database Project (55). The principle criteria used to develop primer sets were:

1. The product of the PCR was less than 600 base pairs so as to be suitable for DOGE. 2. Sites were selected that were common to the targeted bacterial groups and different to non-target groups frequently found in the same environmental samples, particularly at the 3' end of the primer. 3. Where possible, one of the primers was adjacent to an area of sequence variability amongst the targeted groups to facilitate separation by sequence dependent melting during DOGE (234). This primer was not GC clamped.

The 16S rRNA sequences of target and non-target bacteria were downloaded manually from GenBank (19) or using the HIERARCHY_BROWSER of the Ribosomal Database Project (RDP-II) (55). These sequences were aligned with CLUSTALW (348) and potential primer sites manually selected. Candidate primer

-54- sequences were checked for their theoretical specificity using PROBE_MATCH of the RDP-II to generate a list of matches allowing for 1 or 2 mismatches at the 3' end of the primer. The PCR conditions for each primer pair were then empirically determined using the Sprint PCR Express (Hybaid, Ashford, Middlesex, UK) with a gradient of annealing temperatures. The specificity of each reaction was assessed using template DNA from pure cultures.

2.1.3.2 PCR conditions

Details of the primer sets, MgC12 concentrations and thermal cycling conditions used in these studies are detailed in Tables 2.1, 2.2 and 2.3. Primers were manufactured by Sigma-Genosys, Castle Hill, NSW.

Unless otherwise indicated, hot start reactions were performed on a PCR Sprint thermal cycler (Hybaid). Taq DNA polymerase was added to each tube after it had reached 94°C.

PCR reaction volumes of 25 1.11 or 50 1.11 were used. Each 50 j..tl reaction contained 67 mM Tris HCl (pH 8.8), 16.6 mM [NH4h S04, 0.45% Triton X-100, 0.01 mg gelatin, 200 nM of each nucleotide triphosphate, 20 picomoles of each primer, 1.1 units of Taq DNA polymerase (Biotech International, Belmont, Western Australia,

Australia), template DNA and a variable concentration of MgC12 (Table 2.3). The 25 j..tl reactions were the same except that 10 picomoles of each primer and 0.55 units of Taq DNA polymerase were added.

The product of each PCR was examined by electrophoresis at 70V for 45 minutes in a 1.5% agarose gel in 1x TAB buffer followed by ethidium bromide staining for 5 minutes and UV transillumination (Gel Doc 2000, Bio-Rad, Hercules, California).

-55- 2.1.4 Denaturing gradient gel electrophoresis

Sixteen centimetre, 6% polyacrylamide gels (acrylamide/bis 37.5:1) containing a gradient of the denaturants urea and formamide were poured using a Model 4 7 5 gradient delivery system (Bio-Rad). Solutions with the desired concentration of denaturants in a 15 ml volume were made by mixing the required proportions of 0% and 100% stock solutions (detailed below in Table 2.3). A 100% denaturant solution contained 7M urea (sigma) and 40% deionised formamide (Sigma). Seventeen microlitres of the initiators TEMED (Sigma) and 47% ammonium persulfate (Sigma) were added to each 15 ml solution and the gel was poured. A 16 or 20 tooth comb preheated to 37°C in a waterbath was inserted and the gel covered with aluminium foil to exclude air for at least 1 hour for polymerisation to occur.

The amount of PCR product loaded into each lane was initially estimated based on the relative band intensity seen after agarose gel electrophoresis. However, a single relatively bright band will effectively mask the presence of less intense bands because the bright band limits the exposure time during transillumination. For example, the same quantity of PCR amplified DNA may result in 14 DOGE bands or one extremely bright band. Therefore, when multiple PCR products were loaded onto a single DOGE gel, a repeat gel with adjusted loading volumes was often required to obtain a satisfactory representation of all of the bands at similar intensities.

Electrophoresis was performed for 16 hours at 75 V and 60°C in 1x TAB buffer using the Dcode system (Bio-Rad). The electrophoresis bath was placed on a magnetic stirrer to improve the resolution of the rear gel when two gels were run concurrently. Gels were stained in ethidium bromide solution for 5 minutes and destained in water for 15 minutes. Banding profiles were visualised with UV transillumination (Gel Doc 2000).

-56- 2.1.5 Analysis of DGGE banding profiles

2.1.5.1 Calculation of the Bray-Curtis similarity for lane comparisons

The banding profiles were manually converted into binary arrays in Microsoft Excel. For each band position, the presence of a band in a given lane was recorded as 1 and its absence as 0. Each lane in the gel became a column in Excel and each band position a row. The arrays were analysed with the Plymouth Routines In Multivariate Ecological Research software package (PRIMER 5 Version 5.2.2, PRIMER-E Ltd, Plymouth, UK) using Bray-Curtis similarity for binary data (35). Cluster analysis and dendrogram construction were performed using the arithmetic average algorithm of PRIMER 5.

2.1.5.2 Statistical analysis of pooled Bray-Curtis similarities

All statistical analyses were performed with Graphpad InStat version 3.00 (GraphPad Software, San Diego, California, www.graphpad.com). The Bray-Curtis similarity measures for groups of like comparisons were pooled for statistical analysis. Pooled Bray-Curtis similarities were analysed by group using the Friedman test for paired samples and non-parametric data. Where the results were significant, post-hoc testing was performed with the Wilcoxon Matched-Pairs Signed-Ranks test.

2.1.6 Sequencing

Sequencing was performed using the ABI PRISM BigDye Terminator Sequencing Kit™ (Applied Biosystems, Foster City, CA). In brief, the PCR product to be sequenced was cleaned up by precipitation with 2 volumes of 80% ethanol and 1110 volume of 3 M sodium acetate (APS) for 10 minutes at room temperature. DNA was then pelleted by centrifugation at 7200 x g for 15 minutes and resuspended in 12.5 J..Ll of sterile deionised water. One micro litre of this DNA solution was added to 4 J..Ll

-57- of BigDye and 10 picomoles of primer and made up to a total volume of 10 J.tl. Thermal cycling using this mixture was performed on a Hybaid Sprint thermal cycler. Cycling consisted of 96°C for 1 minute; 30 cycles of 96°C for 10 s, 50°C for 5 s, 60°C for 4 minutes; hold at 15°C. The sequencing product was then precipitated with 4 volumes of 75% ethanol and 112 volume of 3 M sodium acetate for 15 minutes at room temperature. After centrifugation at 7200 x g for 20 minutes the resulting pellet was washed with 250 J..Ll of 70% ethanol and dried in a vacuum centrifuge. The product was separated on a model 377 DNA sequencer (Applied Biosystems) in the Automated DNA Sequencing Facility at the UNSW. Overlapping sequences were assembled using the INHERIT software package (Applied Biosystems).

2.1.7 Extraction of PCR product from polyacrylamide gels

The method used was a variation on the "crush and soak" method of Sambrook et al 1989 (289). The band of interest was cut out of the gel using a sterile scalpel and placed in a sterile 1.5 m1 tube. Then, 100 J..Ll of sterile deionised water was added and the tube was stored in the dark at room temperature for 2 days. Two microlitres of this solution was then used as template in subsequent PCR.

2.1.8 PCR-cloning

2.1.8.1 Generation of Competent E. coli DHSa with the Rubidium Chloride method

A variation of the method of Maniatis et al (211) was used to generate the clone libraries. One millilitre from an overnight culture of E. coli DH5a was inoculated into 50 ml of Luria Bertani (LB) broth in a sidearm flask. The broth was then incubated at 37°C on a shaking table until the absorbance at 550 nm (A550) was 0.48 (a sidearm flask containing LB broth alone was required as a control to determine

-58- the A550). The flask was then cooled on ice for 15 minutes and the contents transferred to a 50 ml Falcon tube for centrifugation at 3000 x g for 5 minutes at 4°C (Mistral3000i Centrifuge, Sanyo). The supernatant was discarded and 20 ml of Transformation Buffer 1 (see below) was added. The cell pellet was then gently resuspended and cooled on ice for 15 minutes. Finally, the cells were pelleted by centrifugation, the supernatant discarded and 2 ml of Transformation Buffer 2 added (see below). The cells were gently resuspended and the tube was then placed on ice for 15 minutes. One hundred microlitre aliquots were snap frozen for storage at minus 70°C.

2.1.8.2 Clone library construction

Clone libraries containing 16S rRNA sequences were generated from faecal samples using a modification of the method of Suau et al (335).

Seventy-five nanograms of faecal template DNA was added to a 50 1-Ll reaction containing 20 picomoles of primers F27 and R1492 (Table 2.1). Thermal cycling consisted of 94°C for 10 minutes; 10 cycles of 92°C for 1 minute, 48°C for 1 minute and 72°C for 1.5 minutes; and finally 72°C for 15 minutes (Table 2.3). Six PCR reactions derived from each faecal sample were pooled and purified using the Wizard PCR preps DNA purification system (Promega, Madison, WI) according to the manufacturer's instructions. The purified PCR product was quantitated by UV spectrophotometry.

PCR products were then ligated into the pGEM-T Easy Vector (Promega) according to the manufacturer's instructions. Briefly, 40 ng of PCR product was added to 50 ng of vector, 5 1-Ll of Rapid Ligation Buffer and 1 I-Ll of T4 DNA Ligase and the volume made up to 10 1-Ll with sterile deionised water. After incubation for 2 hours at room temperature, competent E. coli DH5a. were transformed with this ligation reaction and then plated onto LB plates containing ampicillin (Sigma), isopropyl-P -D­ thiogalactopyranoside (IPTG, Progen Industries, Darra, QLD) and 5-bromo-4-

-59- chloro-3- iodyl-B-D-galactoside (X-Gal, Progen). To transform the E. coli, 2~-tl of the ligation reaction was added to a 50 1-tl aliquot of competent E. coli DH5a and incubated on ice for 1 hour. Cells were then heat shocked in a 37°C waterbath for 2 minutes after which they were replaced on ice for 5 minutes. The cells were then grown out in 1 ml of LB broth supplemented with 20 rnM glucose for 1 hour on a shaking table at 37°C. A 100 ~-tl aliquot was then spread evenly over an LB/ampicillin/IPTG/X-Gal plate. The remaining broth was centrifuged at 7500 x g for 15 seconds. The supernatant was removed and the cells resuspended in 100 ~-tl of LB broth for a second plating out.

After overnight incubation at 37°C, the white colonies were picked with a 200 1-tl pipette tip and subcultured onto gridded LB/ampicillin/IPTG/X-Gal plates. The inserts of colonies on the subcultured plates were then amplified by colony PCR targeting the plasmid insert. Bacterial cells were picked from the plates with a pipette tip and dipped briefly into a 200 1-tl PCR tube containing 5 1-tl of deionized water. Twenty microlitres of PCR mix was then added so that the resulting solution contained 2.5 rnM MgC12, 10 picomoles of the primers FpUC/M13 and RpUC/M13 and 0.55 units of Taq DNA polymerase (Table 2.2). A cold start PCR reaction was then performed according to the thermal cycling conditions outlined in Table 2.3. The size of the PCR product was checked by agarose gel electrophoresis and ethidium bromide staining. Inserts of the correct size (approximately 1.6 Kb) were sequenced with primers F27 and 356F as previously described in Section 2.1.6.

2.1.8.3 Analysis of PCR-clone libraries

Sequences were viewed and assembled using software contained in the INHERIT™ package (Applied Biosystems). Chimeric sequences detected with the CHECK_CHIMERA function of the Ribosomal Database Project (55) were discarded. Clones producing less than 350 bases of readable sequence were also discarded. Each remaining sequence was compared with the GenBank database (19)

-60- and the sequence with the greatest homology and the most homologous sequence with a genus and species designation were recorded.

The number of operational taxonomic units (OTUs) present in each library was calculated by defining an OTU as a cluster of 16S ribosomal DNA sequences that differed by 2% or less (335). The coverage of the OTUs amplified by PCR from each sample by each library was estimated by applying Good's formula (129). The total number of OTUs present in each library was estimated with the Chao1

2 estimator (43) Schaol =Sobs+ n1 12n2 where Sobs is the number of OTUs observed, n1

is the number of OTUs observed once only and n2 is the number of OTUs observed only twice (43). The equation for the variance of Chao1 (42) as discussed in Hughes et al (157) was applied to estimate an upper limit for total OTU numbers, where the upper limit was the Chao 1 estimate plus twice the square root of the variance.

The sequences were uploaded and analysed in BioManager (http://www.angis.org.au). The complete 16S ribosomal DNA sequences of named homologues were downloaded from GenBank for use as reference sequences in tree diagrams. All sequences for each sample were aligned with CLUSTALW (348) and a distance matrix calculated by DNADIST (91) using the Jukes-Cantor algorithm(l67). Phylogenetic trees displaying the degree of sequence homology were produced with NEIGHBOR (91) and edited in Microsoft Powerpoint (Microsoft Corporation, www .microsoft.com).

2.1.9 Quantitative comparison of clone libraries

Paired clone libraries were compared using the method of Singleton et al (324). Multiple sequence alignments of paired clone libraries were downloaded into Microsoft WORD (Microsoft) and the 5' and 3' ends removed such that a 600 to 700 base pair segment of complete sequence common to all of the sequences remained. The alignment was uploaded into BioManager and a distance matrix constructed with DNADIST (91) using the Jukes-Cantor method (167). A sample file

- 61- (http://www.arches.uga.edu/~whitmanllibshuff.html) was constructed per the authors instructions (http://www .arches. uga.edu/-w hitman/description.html) and the program LIB SHUFF was run.

2.1.10 Animals

All mice were housed in conventional conditions in the School of Biotechnology and Biomolecular Science (BABS) Animal Facility and had access to autoclaved water and rodent chow (Gordon's Speciality Stock Feed Pty Ltd, Yanderra, NSW) unless otherwise stated.

Wild-type C57BL/6 mice and BALB/C mice were obtained from the Animal Resources Centre, Canning Vale, Western Australia at 6 weeks of age. Interleukin- 10 deficient C57BL/6 mice originating from the Australian National University (ANU, Canberra, ACT) and bred in the BABS animal facility were housed under similar conditions. Where required mice were ear notched under ketamine/xylazine anaesthesia. C57BL/6 WEJ crosses were bred within the BABS animal facility.

Mice of various ages housed within the School of Biotechnology and Biomolecular Science Animal Facility, but originating from two other suppliers, were assessed for Helicobacter colonisation. These suppliers were the Biological Resources Centre, Sydney, NSW and the Walter and Elisa Hall Institute (WEill), Kew,VIC.

2.1.11 Ethics

All animal studies were approved by the UNSW Animal Care and Ethics Committee with approval numbers 99/84 and 02/97. Human studies were approved by the UNSW Human Research Ethics Committee with approval numbers: 00110 and 01256.

-62- 2.2 MATERIALS

Phosphate Buffered Saline pH 7.2

NaH2P04.2H20 (BDH) 8.74 g N~HP04 .12H2 0 (BDH) 51.62 g NaCI (APS) 17.0 g

Add 1.8 L of deionised water and adjust pH to 7.2 with NaOH pellets (APS). Made up to 2 Land autoclaved at 121 °C for 15 minutes.

SOx TAE buffer

Tris base 242 g (Sigma) Acetic acid, glacial57.1 ml (AlliedSignal, Seelze, Germany) 0.5 M EDTA pH 8.0 100 ml (Sigma)

Made up to 1L with deionised water and autoclaved. lx TAE buffer

40 ml of 50x TAB made up to 2litres with deionised water.

0.5 M EDTA (pH 8.0)

EDTA 186.1 g (Sigma) NaOH pellets 20 g (APS)

Made up to 1 litre with deionized water. pH adjusted to 8.0 and autoclaved.

-63- 3M Sodium Acetate

12.3 g of Sodium acetate (APS)

Made up to 50 ml in deionized water and autoclaved.

2x Gel loading dye for DGGE

bromophenol blue (Sigma) 10mg xylene cyanol (Sigma) 10mg glycerol (APS) 14ml Water 6ml

0% Denaturing solution for DGGE (for low denaturant concentration syringe)

30 ml of 40% acrylamide/bis 37.5:1 (Sigma) 4 ml of SOx T AE buffer 166 ml of deionised water

Store at 4 °C and minimise light exposure.

0% Denaturing solution for DGGE (for high denaturant concentration syringe)

40 ml of glycerol (APS) 30 ml of 40% acrylamide/bis 37.5:1(Sigma) 4 ml of SOx T AE buffer 126 ml of deionised water

Store at 4°C and minimise light exposure.

-64- 100% Denaturing solution for DGGE

30 ml of 40% acrylamide/bis 37.5:1 (Sigma) 4 ml of 50x T AE buffer 80 ml of deionised formamide (Sigma) 84 g urea (Sigma) 20 ml of deionized water

Dissolve urea crystals by heating the solution in a 37°C water bath. Store at 4°C and minimise light exposure.

Ethidium bromide stain

Add ethidium bromide stock solution (10 mg/ml) to 1x TAE to obtain a 1 !lg/ml solution.

Polymerisation activators

17 Ill ofTEMED (Sigma)

17111 of 47% ammonium persulfate

0.235 g ammonium persulfate (Sigma) made up to 0.5 ml with deionized water. A freshly made solution is required for each gel.

1.5% agarose for gels

3 g of agarose (Progen) made up to 200 ml with 1x TAB buffer. Heated to 70°C in a microwave with the bottle cap loosened.

-65- Neutral buffered formalin

100 ml of 40% formaldehyde

4 g NaH2P04.2H20 (BDH) 6.5 g N~HP04 or 16.4 g N~HP04 .12H2 0 (BDH) 900 ml of deionized water

Luria Bertani broth (500 ml)

5 g Bacto-tryptone (Oxoid) 2.5 g Bacto-yeast extract (Oxoid) 5 g NaCl (APS)

Madeupto 1L Adjust pH to 7.0 with NaOH (APS) Autoclave.

LB plates with ampicillin (100 !lg/ml)

500 ml LB broth 7.5 g Bacteriological Agar (Oxoid) Ampicillin (Sigma) 50 mg in 1 ml of deionized water. Filter sterilised with a 0.2 !liD filter (Nalge Nunc International, Rochester, NY).

Add agar to LB medium. Autoclave. Allow to cool to 50°C before adding ampicillin and then pouring the plates.

IPTG stock solution

47.6 mg isopropyl-~ -D-thiogalactopyranoside (IPTG, Progen) 2 ml of deionised water

-66- Filter sterilise and store at 4°C.

X-Gal (10 ml)

400 mg 5-bromo-4-chloro-3- iodyl-B-D-galactoside (Progen) dissolve in 10 ml ofN,N'-dimethyl-formamide (Sigma) Cover with aluminium foil and store at -20°C.

LB plates with ampicillin/IPTG/X-Gal

500 ml LB broth 7.5 g Bacteriological Agar (Oxoid) 1 ml of filter sterilised 50 mg/ml Ampicillin (Sigma) 0.25 ml IPTG solution 1 ml of X -Gal solution

Add agar to LB broth. Autoclave. Allow to cool to 50°C before adding ampicillin, IPTG and X-gal then pour the plates.

Transformation buffer 1 (per 200 ml)

Potassium acetate 0.588 g (APS)

RbC1 2 2.42 g (Sigma)

CaC12.2H20 0.294 g (Merck)

MnC12.4H20 2.0 g (APS) Glycerol 30 ml (APS)

Make up to 200 ml with deionised water and adjust pH to 5.8 with dilute acetic acid. Filter sterilise before use.

-67- Transformation buffer 2 (per 100 ml)

MOPS 0.21 g (Sigma)

CaC12 1.1g (Merck)

RbC12 0.121 g (APS) Glycerol 15 ml (APS)

Make up to 100 ml with deionised water and adjust pH to 6.5 with dilute NaOH. Filter sterilise.

-68- TABLE 2.1: 16S rRNA Primers

Primer Sequence (5'-3') Locusa Tmb Reference GC658F CGC CCC CCG CGC CCC GCG CCC GGC 658-676 86 Riley et al (21St CCG CCG CCC CCG CCC TGG GAG AGG TAGGTGGAA T 1067R GCC GTG CAG CAC CTG TTT TCA 1047-1067 61 Grehan et al (138) D86 GTC CTT AGT TGC TAA CTA TT 1117-1136 54 Shen et al (316) D87 AGA TTT GCT CCA TTT CAC AA 1264-1283 54 Shen et al (316) H276f CTA TGA CGG GTA TCC GGC 276-293 58 Riley et al (275) H676r ATT CCA CCT ACC TCT CCC A 658-676 58 Riley et al (275) Hbr TCT CCC ATA CTC TAG AAA AGT 644-664 58 Riley et al (275) F27 AGA GTT TGA TCC TGG CTC AG 8-27 60 Weisburg et al (373) R1492 ACG GCT ACC TTG TTA CGA CTT 1492-1512 61 Neilan et al (237) R1494 TAC GGC TAC CTT GTT ACG AC 1494-1513 59 Weisburg et al (373) B38 GCA TTT GAA ACT GTT ACT CTG 633-653 58 Shames et al (314) B39 CTG TTT TCA AGC TCC CC 1031-1047 62 Shames et al (314) Hmur ACA GAA GTG GCA CTC CCA 1019-1032 56 Grehan et al (138) TABLE 2.1: 16S rRNA Primers (continued).

Primer Sequence (5' -3') Locusa Tmb Reference U968GC CGC CCG GGG CGC GCC CCG GGC GGG 968-984 85 Nubel et al (243) GCG GGG GCA CGG GGG GAA CGC GAA GAACCTTAC L1401 GCG TGT GTA CAA GAC CC 1385-1401 54 Zoetendal et al (388) 356F CTC CTA CGG GAG GCA GCA G 338-356 67 Neilan et al (237) Bif164f GGG TGG T AA TGC CGG ATG 164-182 66 Satokari et al (291) Bif662r CCA CCG TTA CAC CGG GAA 645-662 67 Satokari et al (291) Bif662GCr CGC CCG CCG CGC GCG GCG GGC GGG 645-662 87 Satokari et al (291) GCG GGG GCA CGG GGG GCC ACC GTT ACACCGGGAA Bac948F ATG TGG TTT AAT TCG ATG ATA 948-968 56 This study Bac1307R ATG CGC GAT TAC TAG CGA A 1289-1307 63 This study Bac1307RGC CGC CCC CCG CCG CCC CGC CGC CCG 1289-1307 84 This study GCC CGC CGC CCC CGC CAT GCG CGA TTACTAGCGAA Ccoc447F TGA CGG TAC CTG ACT AAG 447-464 53 This study

Ccoc986R TTG AGT TIC ATT CTT GCG AA 967-986 61 This study TABLE 2.1: 16S rRNA Primers (continued).

Primer Sequence (5' -3') Locusa Tmb Reference Ccoc986RGC CGC CCC CCG CGC CCC GCG CCC GGC 967-986 83 This study CCG CCG CCC CCG CCC TTG AGT TTC ATT CTT GCG AA Clept751F GTG CCG CAG TTA ACA CAA 751-768 61 This study Cleptl246R GCG ATT ACT AGC AAT TCC GA 1227-1246 62 This study Clept1246RGC CGC CCC CCG CGC CCC GCG CCC GGC 1227-1246 85 This study CCG CCG CCC CCG CCC GCG ATT ACT AGCAATTCCGA

aApproximate position of target 16S rRNA based on the numbering of E. coli 16S rRNA. bTheoretical temperature for primer-template dissociation in degrees celsius. cGC658F is a GC clamped and reversed version ofH676r. TABLE2.2: Other primer sets

Primer Sequence (5'-3') Locusa Tmb Reference GCF20 CGC CCC CCG CGC CCC GCG CCC GGC CCG Beta subunit of RNA 85 R. Case (personal communication) CCG CCC CCG CCC CCC AAT TGA AAC ACC polymerase TGAAGG R19 ACT GCC TGA CGT TGC ATG Beta subunit of RNA 64 R. Case (personal communication) polymerase FpUC/M13 GTT TTC CCA GTC ACG AC pGEM-T easy vector 56 insertion site RpUC/Ml3 CAG GAA ACA GCT ATG AC pGEM-T easy vector 51 insertion site

~argetDNA. bTheoretical temperature for primer-template dissociation in degrees celsius. TABLE 2.3: PCR and Thermal Cycling conditions

Target Forward Reverse MgCI2 Thermal cycling conditions Gradient for primer primer concentration DGGE (mM)

Helicobacter species H276f H676r 2.5 94°C 4 min; 35 cycles of: 94°C 5 s, 57°C 5 s, 72°C 30 s; 72°C 2 min Helicobacter species GC658F 1067R 2.5 94°C 5 min; 30 cycles of: 94°C lOs, 41-48% 62°C lOs, 72°C 30 s; 72°C 2 min H. rodentium and H. D86 D87 2.5 94°C 4 min; 35 cycles of: 94°C lOs, ganmani 58°C 10 s, 72°C 30 s; 72°C 2 min H. bilis H276f Hbr 1.5 94°C 5 min; 35 cycles of: 94°C 20s, 53°C 20 s, 72°C 30 s; 72°C 2 min H. hepaticus B38 B39 3 94°C 4 min; 35 cycles of: 94°C lOs, 61°C 10 s, 72°C 30 s; 72°C 2 min H. muridarum H276f Hmur 2.5 94°C 4 min; 35 cycles of: 94°C lOs, 58°C 10 s, 72°C 30 s; 72°C 2 min domain Bacteria F27 R1494 2.5 94°C 5 min; 30 cycles of: 94°C 15s, 50°C 20 s, 72°C 2 min; 72°C 7min domain Bacteria F27 R1492 2.5 94°C 10 min; 10 cycles of: 92°C 1 (for PCR-cloning) min, 48°C 1 min, 72°C 1.5 min; 72°C 15 min domain Bacteria U968GC L1401 2.5 94°C 5 min; 30 cycles of: 94°C 30s, 43-57% 56°C 30 s, 72°C 60 s; 72°C 4 min Bacteroides-prevotella Bac948F Bac1307RGC 2.5 94°C 4 min; 30 cycles of: 94°C 20s, 40-70% group (RDP 2.15.1.2) 64°C 30 s, 72°C 60 s; 72°C 4 min Clostridium coccoides Ccoc447F Ccoc986RGC 2.5 94°C 4 min; 30 cycles of: 94°C 20s, 43-60% group (RDP 2.30.4.1) 63°C 30 s, 72°C 60 s; 72°C 4 min TABLE 2.3: PCR and Thermal Cycling conditions (continued).

Target Forward Reverse MgCI2 Thermal cycling conditions Gradient for primer primer concentration DGGE (mM) Clostridium leptum Clept751F Clept1246RGC 2.5 94°C 4 min; 30 cycles of: 94°C 38-65% subgroup (RDP 20 s, 63°C 30 s, 72°C 90 s; 72°C 7 2.30.9.1) mm Bifidobacterium Bif164f Bif662GCr 3 94°C 5 min; 30 cycles of: 94°C 43-65% 30 s, 62°C 20 s, 72°C 60 s; 72°C 7 min Insert in pGEM-T easy FpUC/M13 RpUC/Ml3 2.5 94°C 10 min; 30 cycles of: 94°C vector 10 s, 54°C 20 s, 72°C 2 min; 72°C 7 min Bsubunit of RNA GCF20 R19 3 94°C 5 min; 30 cycles of: 94°C 20-65% polymerase 30 s, 45°C 30 s, 72°C 60 s; 72°C 2 min CHAPTER 3. Development of primers targeting the major phylogenetic groups of bacteria present in the lower bowel and their application to the comparison of murine faecal and caecal microbiota.

3.1 INTRODUCTION

In spite of substantial efforts in the past to characterise the bacterial populations that are present in the human colon in health and disease, no clear picture has emerged of an association between particular bacterial groups and many prevalent colonic diseases. However the problems associated with current methods are such that significant differences could be present in the microbiota of individuals with disease, compared to unaffected subjects, without these having been detected.

Studies of human faecal samples using in vitro culture have assessed whether there is an association between the colonic microbiota and diseases such as ulcerative colitis, Crohn's disease and colonic carcinoma (86, 116, 134,230,252, 355). These studies however did not reveal consistent differences in the colonic microbiota of cases compared with controls. There are several reasons why culture may have failed to detect a real difference. Molecular studies suggest that many colonic bacterial species are not represented in culture studies (157, 335). In addition, it is likely that the processes of sampling, dilution and incubation lead to significant variance in the results of culture (62, 133). This variance limits the usefulness of the technique to the detection of large quantitative differences only (62, 133). Thus although culture has a proven role in the detection of known colonic pathogens such as Campylobacter jejuni and Salmonella enterica, the method appears to be of limited usefulness in the description of the faecal microbiota as a whole.

Molecular methods such as PCR-DGGE, PCR-cloning, 16S rRNA blotting and FISH have significantly advanced our understanding of the bacterial populations of the lower bowel and avoided many of the problems associated with in vitro culture (148, 213, 335, 376, 390). To date the application of these methods to studies of

-75- disease-association have been hampered by the complexity of the lower bowel flora (335) and the fact that the methods that produce detailed descriptions of the diversity . of species in a sample, such as PCR-cloning and PCR-DGGE, are not quantitative. Conversely, the methods of 16S rRNA blotting and FISH that can quantitate groups or species of bacteria do not produce a description of the range of species in each sample.

In order to examine whether there is an association between a particular bacterial species and disease, the generation of a list of the species present in faecal samples or a representation of this list by a qualitative method is desirable. Of the qualitative methods, PCR-cloning of 16S rDNA produces a series of cloned sequences that can be compared with the GenBank database so that the bacterial species may be putatively identified. However the method of DNA extraction, the amount of template DNA added to the PCR, the types of 16S rDNA template, the PCR primers, the number of cycles of amplification, the cloning kit and non-target DNA can all affect the final library composition (28, 147, 269, 341, 377). Furthermore the method is expensive and labour intensive due to the large number of sequences required to compare the bacterial populations in different samples.

The alternative qualitative method, PCR-DGGE, has the advantage of producing quick and reproducible estimates of species diversity but does not identify each species. This method is also PCR-based and so in reality is also biased by similar factors to those for clone libraries. In addition, when applied to faecal samples, the number of bands detected with primers for the domain Bacteria are much less than estimates of species diversity based on the sequencing of clone libraries (157, 335, 388). For example, a commonly used universal primer set predominantly produces bands representing only one of the 3 major phylogenetic groups in faeces- the Clostridium coccoides group (243, 387, 388). One approach to improving the representation of faecal bacteria is to divide up the spectrum of bacterial species present in the microbiota and analyse samples with primers targeting the major phylogenetic groups represented in clone libraries derived from faecal samples

-76- (216). However, few such single-step non-degenerate primer sets suitable for DGGE have been published so far. In fact, the primers for Bifidobacterium species are the only common colonic bacterial species with a suitable primer set (291). Degenerate and semi-nested primer sets have also been published for Lactobacillus species and related bacteria (151, 368).

One of the main strengths of PCR-DGGE applied to the study of the adult human intestine is that the banding patterns produced with the Bifidobacterium and domain Bacteria primer sets appear to be unchanging over periods of months (291, 388). Therefore it should be possible to examine changes in the species present in the faeces of individuals in association with disease over time if other suitable primer sets were available. That is, the loss or gain of species in association with disease onset, flare or remission in an individual could be evaluated. Unfortunately however, the technique does not appear to be useful for comparing the microbiota of different individuals, as preliminary studies comparing the faecal microbiota of individuals with and without disease using primers for the domain Bacteria, have not yielded useful findings (332, 388). This is likely to reflect the limited depiction of the microbiota obtained with these primers, that only partly reflects the species and strain diversity present.

Apart from the lack of suitable primer sets an additional issue hampering disease­ association studies in humans, is the accumulating evidence that the bacterial populations of the caecal and faecal microbiota within a given person may differ in species composition and quantity (213, 390). These observed differences have important implications for sampling in studies of colonic ecology. Ideally studies of the microbiota of the human colon would use samples from the caecum because the caecal microbiota constitutes the mixing portion of the right colon, a region that is commonly involved in diseases such as Crohn's disease. However faeces are the only. sample that can be repeatedly and non-invasively collected from individuals that represents this microbiota from a practical point of view. Samples can be directly obtained from the right colon at colonoscopy, however this requires gut

-77- lavage that eliminates much of the luminal content and, in addition, is associated with some risk to the subject (114). Alternatively, a sample may be aspirated from a nasocaecal tube but this requires 6 to 10 hours to insert (213).

The aim of this work was to develop primer sets for PCR-DGGE of the major groups of lower bowel microbiota and to apply these to a comparison of caecal and faecal bacterial populations in a model not constrained by sampling difficulties. PCR primers were developed to target 3 of the major phylogenetic groups from the mammalian anaerobic flora present in clone libraries derived from the lower bowel (192, 288, 335). These are the Bacteroides-prevotella group, Clostridium coccoides group and Clostridium leptum subgroup. Subsequently the primer sets were used to compare the caecal and faecal microbiota of a murine model.

-78- 3.2 METHODS

3.2.1 Primer development

Primer sets were developed to target the Bacteroides-Prevotella group (RDP 2.15.1.2), Clostridium coccoides group (RDP 2.30.4.1) and Clostridium leptum subgroups (RDP 2.30.9.1) of the Ribosomal Database Project (RDP-II) (55) as these constitute the numerically dominant anaerobic flora of the colon. Initial attempts were made to design non-degenerate primers of 15 bases in length with homology to 90% of RDP sequences within the target groups and less than 50 non-target matches using the published freeware program PRlMROSE (9). However this did not yield suitable sequences for DGGE, necessitating manual primer design.

For manual primer design the named sequences within target and non-target groups were downloaded using the HIERARCHY_BROWSER (55) and aligned with CLUSTALW (348). Primer sites for DGGE were chosen using 3 criteria. Firstly the product was less than 600 base pairs in length. Secondly, primer sites were selected from 16S rRNA sites that were common to the targeted bacterial group (particularly at the 3' end of the proposed primer) and different to other gut bacterial sequences within the Ribosomal Database Project. Thirdly, to facilitate the separation ofPCR products, where possible one of the primer sites (for the undamped primer) was adjacent to an area of sequence variability among the targeted bacterial sequences (234).

The target sequences for the Bacteroides-prevotella group, C. coccoides group and the C. leptum subgroup that were downloaded from the RDP are listed in Tables 3.1, 3.2 and 3.3 respectively. Representative sequences from the RDP within all subgroups of the target group were included; where possible the sequences of type strains were selected. The non-target sequences used in all of the alignments are shown in Table 3.4. To generate a list of non-target sequences for each primer set, the sequences shown in Table 3.4 were combined with a selection of sequences from

-79- the 2 other RDP groups in the lower bowel microbiota. For example, the non-target sequences for the Bacteroides-Prevotella primer set included the sequences in Table 3.4 plus 10 sequences from each of Tables 3.2 and 3.3. Using this approach 6 potential primer sites were identified for the Bacteroides-prevotella group, 5 for the C. coccoides group and 4 for the C. leptum subgroup.

The sequences of potential primers were compared with the RDP database using PROBE_MATCH (55) to generate a list of matches and near-matches with 1 or 2 mismatched bases at the 3' end of the primer. For each of the primer sets it was not possible to encompass all of the sequences within the target groups or exclude all non-target groups with a single non-degenerate primer. Therefore potential primer sites that resulted in the greatest number of target and least non-target matches suitable for PCR-DGGE were selected. The selected primers are shown in Table 2.1, Chapter 2.

The non-target groups within the RDP containing one or more sequences that were complete matches with each selected primer are listed in Table 3.5. In general, where a complete match was present with one primer this was not present with the other primer. The exceptions to this were 4 sequences in the Clostridium lentocellum group that perfectly matched the forward and reverse primers for the C. coccoides group. An additional problem in designing comprehensive primers was the presence of a guanine(G) or cytosine( C) base at the target sequence position corresponding to the 7th base of the forward primer for the C. leptum subgroup (base position 757 of 16S rRNA with E. coli numbering). A "C" was empirically chosen in this position for the primer and a 1 base mismatch to some of the target sequences was accepted.

- 80- 3.2.2 Comparison of the primers with 168 rRNA sequences from the murine lower bowel microbiota

The proposed primers were then compared with the unique murine lower bowel16S rRNA PCR-clone library sequences published by Salzman et al (2002) (288) and the sequences of the Altered Schaedler Flora published by Dewhirst et al (1999) (69).

3.2.3 Optimisation of PCR and DGGE

For each of the new GC-clamped primer sets the annealing temperature for PCR was determined using the gradient PCR Express (Hybaid) and DNA template from positive control strains. These strains were B. fragilis and B. vulgatus for the Bacteroides-prevotella group primer set, C. nexile for the C. coccoides group primer set and C. leptum for the C. leptum subgroup primer set. PCR amplification and agarose gel electrophoresis was performed as described in Section 2.1.3, Chapter 2.

Following the determination of the optimal thermal cycling conditions the PCR was applied to positive and negative control strains to assess specificity. The following bacterial strains served as positive and negative controls for PCR: B. longum (ATCC 15707), F. mortiferum (ATCC 25557), L. salivarius (ATCC 11741), C. leptum (DSM 753) and University of New South Wales School of Microbiology Culture Collection strains of: B. vulgatus, B. fragilis, C. histolyticum, C. nexile, B. adolescentis, E. limosum, P. anaerobilus, E. coli, E.faecalis, L. acidophilus, P. acnes, D. desulfuricans and B. cereus.

PCR product derived from the positive control bacterial strains and murine and human faeces were then used to optimise the gradients for DOGE. DOGE was performed as described in 2.1.4, Chapter 2. All electrophoresis was initially conducted with a 20-70% gradient of denaturants and this was gradually narrowed to maximise band separation.

- 81- 3.2.4 Animals and sample collection for comparison ofcaecal and faecal microbiota

Faecal samples were collected from 16 four or six week old C57BL/6 interleukin-10 deficient mice housed in 4 cages, under conventional conditions, in the BABS Animal Facility. The four cages consisted of three 4 week old males, five 4 week old females, four 6 week old males and four 6 week old females. The mice were sacrificed after stool collection and the caecal tip (approximately 20 mg) was collected. Animal experimentation was approved by the Animal Care and Ethics Committee of the UNSW (approval number: 02/97).

3.2.5 Sample processing

Immediately following collection, faecal and caecal samples were vortexed into a slurry in cell lysis solution and DNA was extracted using the Puregene DNA purification kit (Gentra) according to the manufacturer's instructions, Section 2.1.2.2, Chapter 2. Fifty nanograms of this DNA was used as template in PCR­ DGGE for the Bacteroides-prevotella group, C. coccoides group, C. leptum subgroup and the domain Bacteria (243). The resulting banding profiles were compared and the Bray-Curtis similarity for each pair of lanes was determined as described in Section 2.1.5, Chapter 2.

3.2.6 Statistics

Each Bray-Curtis similarity for presence-absence (binary data) represents the relatedness of two banding patterns on the gel, that is, two lanes. The Bray-Curtis similarity measures for groups of like comparisons were pooled. Sets of Bray-Curtis similarities were than compared with the Wilcoxon Matched-Pairs Signed-Ranks test. For example, pooled similarity measures for faecal versus caecal samples from the same mouse (within mouse Bray-Curtis similarities) were compared with the similarity measures for faecal samples from a given mouse versus

-82- the caecal sample of next mouse depicted on the gel (between mice Bray-Curtis similarities). The banding profiles of mice grouped by age (4 or 6 weeks), sex (male or female) and cage (divided into cagemates and non-cagemates) were also contrasted by pooling the similarity measures for samples of a given category versus other samples of the same category (within group Bray-Curtis similarities) and comparing this with the similarity measures for samples of a given category compared with samples from the alternative category (between group Bray-Curtis similarities). For example the similarity of the caecal banding patterns for 4 week old mice versus other 4 week old mice was compared with the caecal banding patterns of 4 week old mice versus 6 week old mice to determine whether there were significant differences in the banding patterns representing caecal samples from mice grouped by age.

-83- 3.3 RESULTS

3.3.1 Primer development

Amplification of B. vulgatus template DNA with the Bacteroides-prevotella group primer set was evident at annealing temperatures up to 68°C with the most intense bands visible at annealing temperatures up to 66°C (Figure 3.1). When the PCR was applied to negative control strains with an annealing temperature of 64°C, no amplification was detected (Figure 3.2). While initial DOGE analysis revealed a single band with the PCR product derived from B. fragilis template DNA, up to 5 bands were visible with DOGE using the PCR-amplified B. vulgatus DNA. Further studies with gradient PCR using a range of annealing temperatures between 63°C and 69°C revealed that the number of distinct bands from B. vulgatus template DNA was reduced to 3 with annealing temperatures greater than or equal to 64°C (Lane 2, Figure 3.3). This annealing temperature was used in subsequent studies.

Template DNA from C. nexile was amplified with the C. coccoides group primer set at annealing temperatures up to 65°C (Figure 3.4). Intense bands were visible with annealing temperatures up to 63°C, therefore this was used for further PCR with this primer set. At an annealing temperature of 63°C there was no amplification of negative control strains of bacteria (Figure 3.5). A single DOGE band was produced at this temperature with PCR product from C. nexile (Lane 3, Figure 3.3).

A PCR product was produced with template DNA from C. leptum using the C. leptum subgroup primer set at annealing temperatures up to 67°C with intense bands evident at up to 65°C (Figure 3.6). A slightly lower annealing temperature was selected for use in this PCR to allow for the fact that the forward primer had a 1 base mismatch to a significant number of sequences in the C. leptum subgroup of the RDP. Using an annealing temperature of 63°C there was no amplification of negative control strains with the exception of E. limosum which produced a faint

- 84- band under these conditions (Figure 3.7, Lane 8). A single DGGE band was produced with PCR product from C. leptum (Lane 4, Figure 3.3).

The primer sequences and optimised thermal cycling conditions for each PCR are listed in Tables 2.1 and 2.3, Chapter 2.

The gradient of denaturants for DGGE was determined by initial electrophoresis of PCR product in a 20-70% gradient of denaturants and gradual narrowing of the gradient to increase band separation in the region where the PCR products denatured. The optimised gradients for DGGE were 40-70% for the Bacteroides­ prevotella group PCR, 43-60% for the C. coccoides group primer set and 38-65% for the C. leptum subgroup primer set (Table 2.3, Chapter 2). Gels depicting the optimised gradients with control strains of bacteria are shown in Figure 3.3.

3.3.2 Comparison of the primers with 16S rRNA sequences from the murine lower bowel microbiota " In order to assess the species diversity that might be amplified by the new primer sets from the murine lower bowel, the primer sequences were compared with the 40 unique 16S rRNA sequences of Salzman et al (2002) derived from murine lower bowel and faecal samples (288). Of the 40 unique sequences in this set, 14 belonged to the Bacteroides group and 11 of these (79%) were complete matches to the Bacteroides-prevotella group primers (Table 3.6). Six of the 8 sequences within the C. coccoides group had sequences of sufficient length for evaluation. All of these sequences were complete matches to the C. coccoides group primer set (Table 3.6). One of the two sequences from the C. leptum subgroup matched the C. leptum primer set. Overall, 18 of the 38 unique sequences of Salzman et al that were of evaluable length (47%) were appropriate matches to the 3 primer sets. Furthermore, three of the 38 sequences matched primers targeting a different bacterial group; two sequences from the Eubacterium rectale-Clostridium coccoides group and one from

-85- the "miscellaneous" group matched the Bacteroides-prevotella group primers (Table 3.6).

The primers were also compared with the sequences of the Altered Schaedler flora of Dewhirst et al (1999) (69). Two of the eight sequences from the Altered Schaedler Flora were completely homologous to their respective primer sets - ASF 519 to the Bacteroides-prevotella group primers and ASF 502 to the C. coccoides group primers. In contrast ASF 492 and ASF 500 each had 2 bases mismatched with one primer from the C. coccoides group and C. leptum subgroup primer sets respectively. The remaining 4 sequences were not within the target groups.

3.3.3 Application of PCR-DGGE to murine faecal and caecal samples

In order to compare the caecal and faecal samples of the mice, a DGGE banding pattern was produced for each sample with all three of the new primer sets. Examples of the DGGE gels obtained with primers for the domain Bacteria, the Bacteroides-prevotella group, C. coccoides group and C. leptum subgroup from the faecal and caecal samples of two six week old male cagemates (denoted Ml and M2) and a six week old female (denoted M3) are shown in Figures 3.8 to 3.11. The banding profiles of caecal and faecal samples from the same mouse were highly similar but different to the banding profiles of other mice (Figures 3.8 to 3.11). The same result was evident with all16 of the mice that were studied. The mean Bray­ Curtis similarities for within and between mouse comparisons were 87.5 ± 10.1 and 52.7 ± 18.9 for the domain Bacteria primers, 96.7 ± 4.4 and 71.5 ± 24.6 for the Bacteroides-prevotella group primer set, 90.8 ± 11.2 and 41.8 ± 22.3 for the C. coccoides group primer set and 96.7 ± 8.7 and 54.9 ± 25.6 for the C. leptum subgroup primer set, respectively. The results for each primer set were statistically significant with the Wilcoxon test at p < 0.01 (Table 3.7). The similarities are illustrated in Figure 3.12. That is, for all of the primer sets the similarity measures for faecal and caecal samples from the same mouse were significantly higher than similarity measures for faecal and caecal samples from different mice (Table 3.7).

-86- The Bray-Curtis similarities for within cage comparisons were divided into comparisons of faecal and caecal samples within groups of cagemates and between groups of cagemates and non-cagemates. For comparisons of faecal samples, the within-group and between-group Bray-Curtis similarities were 48.6 ± 20.7 and 39.8 ± 20 with the domain Bacteria primers, 77.6 ± 35.6 and 65.2 ± 46.9 with the Bacteroides-prevotella group primer set, 38.9 ± 44 and 37.7 ± 42.5 with the C. coccoides group primer set and 54.3 ± 57.2 and 54.3 ± 54.4 with the C. leptum subgroup primer set. The similarity measures for within cage comparisons were significantly higher than between cage comparisons with the Bacteroides-prevotella group primer set only (p = 0.027, Table 3.7). For comparisons of the caecal samples the within-group and between-group Bray-Curtis similarities were 54 ± 20.9 and 40.1 ± 21.5 with the domain Bacteria primers, 78.6 ± 30 and 63.3 ± 39.1 with the Bacteroides-prevotella group primer set, 43.9 ± 48.2 and 37.2 ± 44.8 with the C. coccoides group primer set, and 54.9 ± 55.3 and 57.7 ± 45.7 with the C. leptum subgroup primer set. The results for domain Bacteria primers and Bacteroides­ prevotella primer sets only were significant p = 0.036 and p =0.01 respectively. A box plot demonstrating the range of similarity measures obtained for these comparisons by primer set is shown in Figure 3.13. The results with C. coccoides group and C. leptum subgroup primer sets by cage were not significant (Table 3.7). Further comparisons made between mice grouped by age and sex were found to be significant with the Bacteroides-prevotella group primer set, but this was not supported by the results of any other primer set (Table 3.7).

The mean number of bands produced with each primer set using DNA template from caecal samples from each of the 16 mice was 10.7 ± 3.9, 13.1 ± 4.3, 5.4 ± 1.7 and 3.2 ± 1.6 for the domain Bacteria, Bacteroides-prevotella group, C. coccoides group and C. leptum subgroups respectively.

-87- TABLE 3.1: Target sequences from RDP 2.15.1.2 for the Bacteroides-prevotella group pnmers. Target sequence GenBank accession "Anaeroflexus maritimus" str. PL12FS DSM 2831. * Bacteroides ASF519 str. ASF 519. AF157056 Bacteroides caccae ATCC 43185 (T). X83951 Bacteroides distasonis ATCC 8503 (T). M86695 Bacteroides eggerthii ATCC 27754 (T). L16485 Bacteroides forsythus FDC 331. X73962 Bacteroides forsythus FDC 338 (T). L16495 Bacteroides fragilis ATCC 25285 (T). M61006 Bacteroides ovatus A TCC 8483 (T). X83952 Bacteroides putredinis ATCC 29800 (T). L16497 Bacteroides splanchnicus NCTC 10825 (T). L16496 Bacteroides stercoris ATCC 43183 (T). X83953 Bacteroides thetaiotaomicron ATCC 29148 (T). L16489 Bacteroides un{formis ATCC 8492 (T). L16486 Bacteroides vulgatus ATCC 8482 (T). M58762 Cytophagafermentans ATCC 19072 (T). M58766 Cytophagafermentans NCIMB 2218 (T). D12661 Cytophaga xylanolytica str. XM3. M80585 Porphyromonas asaccharolytica A TCC 25260 (T). L16490 Porphyromonas cangingivalis VPB 4874 (T). X76259 Porphyromonas gingiva/is ATCC 33277 (T). L16492 Porphyromonas gingiva/is DSM 20709. X73964 Porphyromonas macacae ATCC 33141 (T). L16494 Prevotella bivia ATCC 29303 (T). L16475 Prevotella brevis ATCC 19188 (T). AJ011682 Prevotella bryantii DSM 11371 (T). AJ006457 Prevotella buccae ATCC 33690. L16478 Prevotella buccalis ATCC 35310 (T). L16476 Prevotella dentalis DSM 3688 (T). X81876 Prevotella denticola ATCC 33185. L16466 Prevotella loescheii ATCC 15930 (T). L16481 Prevotella melaninogenica ATCC 43982. L16470 Prevotella nigrescens ATCC 33563 (T). L16471 Prevotella oralis ATCC 33269 (T). L16480 Prevotella oris ATCC 33573 (T). L16474 Prevotella tannerae ATCC 51259 (T). AJ005634 Rikenella microfusus ATCC 29728 (T). L16498 *No entry number in GenBank (T) - indicates type strain ATCC- American Type Culture Collection (www.atcc.org) DSM -Deutsche Sammlung von Mikroorganismen und Zellkulturen Gmbh (www. dsmz. de) NCFB -National Collection of Food Bacteria (NCIMB Ltd, Aberdeen, Scotland, UK) NCDO- National Collection of Dairy Organisms (NCIMB Ltd, Aberdeen, Scotland, UK) FERM- Patent and Bioresource Center, Tsukuba, Japan.

-88- TABLE 3.2: Target sequences from RDP 2.30.4.1 for the Clostridium coccoides group primers.

Target sequence GenBank accession Butyrivibrio crossotus NCDO 2416 (T). X89981 Butyrivibrio fibrisolvens NCDO 2221 (T). X89970 Catonella morbi ATCC 51271 (T). X87151 Clostridium aerotolerans DSM 5434 (T). X76163 Clostridium aminophilum ATCC 49906 L04165 Clostridium aminovalericum ATCC 13725 (T). M23929 Clostridium celerecrescens DSM 5628 (T) .. X71848 Clostridium clostridiifonne ATCC 25537 (T). M59089 Clostridium coccoides ATCC 29236 (T). M59090 Clostridium herbivorans ATCC 49925 (T). L34418 Clostridium nexile DSM 1787 (T). X73443 Clostridium oroticum ATCC 13619 (T). M59109 Clostridium polysaccharolyticum ATCC 33142 (T). X71858 Clostridium populeti ATCC 35295 (T). X71853 Clostridium sphenoides ATCC 19403 (T). X73449 Clostridium xylanolyticum ATCC 49623 (T). X71855 Coprococcus eutactus ATCC 27759 (T). D14148 Desulfotomaculum guttoideum DSM 4024 (T). Y11568 Eubacterium cellulosolvens ATCC 43171 (T). L34613 Eubacterium contortum ATCC 25540 (T). L34615 Eubacterium eligens ATCC 27750 (T). L34420 Eubacteriumformicigenerans ATCC 27755 (T). L34619 Eubacterium hadrum ATCC 29173 (T). * Eubacterium hallii ATCC 27751 (T). L34621 Eubacterium plexicaudatum ATCC 27514. AF157058 Eubacterium ramulus ATCC 29099 (T). L34623 Eubacterium rectale ATCC 33656 (T). L34627 Eubacterium ruminantium ATCC 17233 (T). AB008552 Eubacterium saburreum DSM 3986 (T). * Eubacterium uniforme ATCC 35992 (T). L34626 Eubacterium ventriosum ATCC 27560 (T). L34421 Eubacterium xylanophilum A TCC 35991 (T). L34628 Johnsonella ignava ATCC 51276 (T). X87152 Lachnospira multipara ATCC 19207 (T). L14719 Roseburia cecicola ATCC 33874 (T). L14676 Ruminococcus gnavus ATCC 29149 (T). D14136 Ruminococcus hansenii ATCC 27752 (T). M59114 Ruminococcus hydrogenotrophicus DSM 10507 (T). X95624 Ruminococcus lactaris ATCC 29176 (T). L76602 Ruminococcus obeum ATCC 29174 (T). L76601 Ruminococcus productus ATCC 27340 (T). X94966 Ruminococcus schinkii DSM 10518 (T). X94965 Ruminococcus torques ATCC 27756 (T). D14137 *No entry number in GenBank

-89- TABLE 3.3: Target sequences from RDP 2.30.9.1 for the Clostridium leptum subgroup primers

Target sequence GenBank accession Acetivibrio cellulolyticus ATCC 33288 (T). L35516 Acetivibrio cellulolyticus ATCC 35928 (T). L35515 Anaero.filum aRile DSM 4272 (T). X98011 Anaerofilum pentosovorans DSM 7168 (T). X97852 Bacteroides cellulosolvens ATCC 35603 (T). L35517 Clostridium aldrichii DSM 6159 (T). X71846 Clostridium cellobioparum DSM 1351 (T). X71856 Clostridium cellulolyticum ATCC 35319 (T). X71847 Clostridium cellulosi (T). L09177 Clostridium hungatei ATCC 700212. AF020429 Clostridium josui FERM P-9684 (T). AB011057 Clostridium leptum ATCC 29065 (T). M59095 Clostridium papyrosolvens ATCC 35413 (T). X71852 Clostridium sporosphaeroides ATCC 25781 (T). M59116 Clostridium stercorarium NCIMB 11754 (T). L09174 Clostridium termitidis DSM 5396 (T). X71854 Clostridium thermocellum DSM 1237 (T). L09173 Clostridium thermolacticum DSM 2911. L09176 Clostridium viride DSM 6836 (T). X81125 Eubacterium desmolans ATCC 43058 (T). L34618 Eubacterium plautii DSM 4000 (T). * Eubacterium siraeum ATCC 29066 (T). L34625 Faecalibacterium prausnitzii ATCC 27766. X85022 Ruminococcus albus ATCC 27210 (T). L76598 Ruminococcus albus ATCC 27211. AF030451 Ruminococcus bromii ATCC 27255 (T). L76600 Ruminococcus callidus ATCC 27760 (T). L76596 Ruminococcus .flavefaciens ATCC 19208 (T). L76603 Ruminococcusflavefaciens NCFB 2213 (T). X83430 Sporobacter termitidis DSM 10068 (T). Z49863 *No entry number m GenBank (T) - indicates type strain ATCC- American Type Culture Collection (www.atcc.org) DSM - Deutsche Sammlung von Mikroorganismen und Zellkulturen Gmbh (www. dsmz. de) NCFB -National Collection of Food Bacteria (NCIMB Ltd, Aberdeen, Scotland, UK) NCDO -National Collection of Dairy Organisms (NCIMB Ltd, Aberdeen, Scotland, UK) FERM- Patent and Bioresource Center, Tsukuba, Japan.

-90- TABLE 3.4: Non-target sequences used in alignments for primer design

Non-target sequence GenBank accession Bifidobacterium adolescentis AF275881 Bifidobacterium longum ATCC 15707 M58739 Desulfovibrio piger ATCC 29098 AF192152 Escherichia coli ATCC 25922 X80724 Enterococcus faecalis AB036835 Fusobacterium necrophorum AF044948 Faecalibactrerium prausnitzii ATCC 27766 X85022 Eubacterium limosum AF064242 Lactobacillus salivarius subsp. salivarius ATCC AF089108 11741 Propionibacterium acnes AB041617

-91- TABLE 3.5: Non-target groups within the RDP containing sequences that were a complete match to each primer of the 3 developed primer sets.

Bacteroides-prevotella group Clostridium coccoides group Clostridium leptum subgroup

Bac948F Bac1307R Ccoc447F Ccoc986R Clept751F* Clept1246R

Cytophaga groups land 2 Clostridium Clostridium lentocellum Clostridium botulinum lentocellum group group group Sphingobacterium group Chloroplasts and cyanelles Clostridium purinolyticum group Lewinwella group Borellia gro Chlamydophila subgroup

Persicobacter group Clostridium purinolyticum Paenibacillus group group Cytophaga aurantiaca group Mycoplasma pneumoniae group Brachyspira group Bacteroides group

Treponema group Cytophaga groups 1 and 2

Borrelia group

Leptospira group

*G or C was required in position 7 of this primer to target all C. leptum species. As the aim was to utilise this primer set under PCR conditions in which sequences with either a G or a C at this position would be amplified the results in this table report matches for either a C or G at this position. TABLE 3.6: Comparison of primer sequences with murine lower bowel and faecal 16S rRNA sequences of Salzman et al (288).

Primer set Complete matches to both Non-matches within Matches with non- primer sequences target groups target groups

1 Bacteroides- Bacteroides spp.(3/3) AJ400266 (MIB ) AJ400247 (Clostridium

1 prevotella "mouse intestinal bacteria" AJ400264 (MIB ) clostridiifo rmes)

1 (8/11) AJ400241 (MIB ) AJ400265 (Clostridium celerecrescens) AJ400237 (Clostridium methylpentosum) Clostridium E. rectale-C. coccoides Nil coccoides 2 group (6/6)

Clostridium AJ400270 AJ418058 3 Nil leptum

1 MIB refers to the "Mouse intestinal bacteria" operational taxonomic unit described by Salzman et al (288). 2 AJ400247 and AJ400250 did not have sequence in the region of the forward primer and could not be evaluated. 3 1 mismatch with the forward primer for the C. leptum subgroup.

-93- TABLE 3.7: p-values for the statistical analysis of pooled Bray-Curtis similarities Pooled similarity measures p-values for comparisons of Groups 1 and 2 using each primer set domain Bacteroides C. coccoides C. leptum Group 1 Group2 Bacteria - prevotella Comparison of caecal and faecal samples within the same mouse and between mice Faeces vs caecum Faeces vs caecum 0.0007* 0.0012* 0.0004* 0.0012* of same mouse of next mouse on el Comparisons by cage Caecum vs Caecum vs 0.036* 0.01* 0.65 0.2 caecum of caecum of non- cagemates cagemates Faeces vs faeces Faeces vs faeces 0.13 0.027* 0.34 0.1 of cagemates of non-cagemates Comparisons by age Caecum of 4 week Caecum of 4 week ND 0.42 0.57 0.11 mice vs other 4 mice vs 6 week week old mice mice Faeces of 4 week Faeces of 4 week ND 0.45 0.59 0.12 mice vs other 4 mice vs 6 week week old mice mice Caecum of 6 week Caecum of 6 week ND 0.046* 0.61 0.35 mice vs other 6 mice vs 4 week week old mice mice Faeces of 6 week Faeces of 6 week ND 0.06 0.73 0.25 mice vs other 6 mice vs 4 week week old mice mice Comparisons by sex Male caecum vs Male caecum vs 0.59 0.51 0.67 0.18 other male caecum female caecum Male faeces vs Male faeces vs 0.37 0.89 1 0.21 other male faeces female faeces Female caecum vs Female caecum vs 0.27 0.01* 0.98 0.98 other female male caecum caecum Female faeces vs Female faeces vs 0.61 0.02* 0.96 0.34 other female male faeces faeces ND - not done as 4 and 6 week samples were displayed on separate gels. *significant at p < 0.05 level.

-94- Figure 3.1: Gradient PCR with B. vulgatus template DNA. Lane 1: Annealing temperature 63°C, Lane 2: 63.3°C, Lane 3: 63.60C, Lane 4: 64.30C, Lane 5: 650C, Lane 6: 65.90<::, Lane 7: 66.SOC, Lane 8: 67.SOC, Lane 9: 68.SOC, Lane 10: 69.5°C, Lane 11: 69.90C, Lane 12: 70.10C, Lane 13: Marker FN-1 (Biotech International). 1 2 3 4 5 6 7 8 9 1011 12 13

692 404

Figure 3.2: Specificity of Bacteroides-prevotella primer set. Lane 1: Negative control, Lane 2: B. vulgatus, Lane 3: B.fragilis, Lane 4: B. cereus, Lane 5: B. longum, Lane 6: E. limo um, Lane 7: L. salivarius, Lane 8: L. acidophilus, Lane 9: E.faecalis, Lane 10: F. mortiferum, Lane 11: E. coli, Lane 12: P. anaerobilus, Lane 13: C. histolyticum, Lane 14: P. acnes, Lane 15: D. desulfuricans. Results for C. nexile and C. leptum were also negative (not shown). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Figure 3.3: PCR-DGGE with 1 2 3 4 optimised gradients and positive control bacterial species -(separate gels for each DGGE). Lanes 1 and 2: Bacteroides­ prevotella group PCR-DGGE with B. fragilis and B. vulgatus template DNA respectively using a 40-70% gradient. Lane 3: C. coccoides group PCR-DGGE with C. nexile DNA and a 43 - 60% gradient. Lane 4: C. leptum subgroup PCR-DGGE with C. Ieptum DNA and a 38- 65% gradient. Figure 3.4: Gradient PCR with C. coccoides group primer set and C. nexile template DNA. Lane 1: Marker FN-1~ Lane 2: Annealing temperature 55.40C, Lane 3: 56.40C~ Lane 4: 57.70C~ Lane 5: 59.40C, Lane 6: 61.40C, Lane 7: 63.30C~ Lane 8: 65.30C, Lane 9: 67.60C, Lane 10: 6g>C, Lane 11: 70.20C, Lane 12: Marker FN-1. 1 2 3 4 5 6 7 8 9 10 11 12

6'n

404

Figure 3.5: Specificity of the C. coccoides group primer set. Lane 1: Marker FN-1, Lane 2: Negative control, Lane 3: C. nexile, Lane 4: B. vulgatus, Lane 5: B.fragilis, Lane 6: B. cereus, Lane 7: B. longum, Lane 8: E. limosum, Lane 9: L. salivarius, Lane 10: L. acidophilus, Lane 11: E.jaecalis, Lane 12: F. mortiferum, Lane 13: E. coli, Lane 14: P. anaerobilus, Lane 15: C. histolyticum. Results for P. acnes, D. desulfuricans and C. leptum were also negative (not shown).

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

692 404 Figure 3.6: Gradient PCR with the C. leptum subgroup primer set and C. leptum template DNA. Lane 1: Marker FN-1, Lane 2: Annealing temperature 55.40C, Lane 3: 56.40C, Lane 4: 57.7le, Lane 5: 59.40C, Lane 6: 61.40C, Lane 7: 63.30C, Lane 8: 65.30C, Lane 9: 67.60C, Lane 10: 690C, Lane 11: 70.20C, Lane 12: Marker FN-1. 1 2 3 4 5 6 7 8 9 10 11 12

692 404

•

Figure 3.7: Specificity of the C. leptum subgroup primer set. Lane 1: Marker FN-1, Lane 2: Negative control, Lane 3: C. leptum, Lane 4: B. vulgatus, Lane 5: B.jragilis, Lane 6: B. cereus, Lane 7: B. longum, Lane 8: E. limo sum, Lane 9: L. salivarius, Lane 10: L. acidophilus, Lane 11: E.jaecalis, Lane 12: F. mortiferum, Lane 13: E. coli, Lane 14: P. anaerobilus, Lane 15: C. histolyticum. Results for P. acnes, D. desuljuricans and C. nexile were negative (not shown). The arrow points out the faint band amplified from E. limosum template DNA. 1 2 3 4 5 6 7 8 9 10 1112 13 1415

692 489 Figure 3.8: PCR-DGGE with domain Bacteria primers. Lane 1: M1 caecum, Lane 2: Ml faeces, Lane 3: M2 caecum, Lane 4: M2 faeces, Lane 5: M3 caecum, Lane 6: M3 faeces.

1 2 3 4 5 6

Figure 3.9: PCR-DGGE with Bacteroides-prevotella group primers. Lane 1: B.fragilis, Lane 2: B. vulgaJus, Lane 3: M1 caecum, Lane 4: M1 faeces, Lane 5: M2 caecum, Lane 6: M2 faeces, Lane 7: M3 caecum, Lane 8: M3 faeces. 12345678 Figure 3.10: PCR-OOGE with C. coccoides group primers. Lane 1: C. nexile, Lane 2: M1 caecum, Lane 3: M1 faeces, Lane 4: M2 caecum, Lane 5: M2 faeces, Lane 6: M3 caecum, Lane 7: M3 faeces.

12 3 4 56 7

Figure 3.11: PCR-DGGE with C. leptum subgroup primers. Lane 1: C. leptum, Lane 2: M1 caecum, Lane 3: M1 faeces, Lane 4: M2 caecum, Lane 5: M2 faeces, Lane 6: M3 caecum, Lane 7: M3 faeces

12 3 4 56 7 Figure 3.12: Box and whiskers plot of Bray-Curtis similarities for faecal vs caecal samples by primer set. For each paired plot the fust bar represents the ''within mouse" similarities and the second the "between mouse" similarities. *p < 0.05.

100 * * * * * .c ·;::(1J 80 .E 'iii ·E..._ 60 * :::l u :>.. ~ 40

20

0 Domain Bacteroides- C. coccoides C. leptum Bacteria prevotella group group subgroup

Faeces vs Caecum within mouse • Faeces vs Caecum between mice Figure 3.13: Box and whiskers plot of Bray-Curtis similarities for cagemates and non-cagemates by primer set. For each set of four bars the first bar represents the ''within cagemates" similarities for faeces, the second the "between cagemates and non-cagemates" similarities for faeces. The third and fourth bars represent the same comparisons for caecal samples. *p < 0.05

100 * ....>. * ro ao .E

U) U) ..;J 60 L. ::3 u >. ~ 40 CD

20

0

Domain Bacteroides- C. coccoides C. leptum Bacteria prevotella group subgroup group

Comparisons

within cagemates for faeces

between cagemates and • non-cagemates for faeces within cagemates for caecum • between cagemates and non-cagemates for caecum 3.4 DISCUSSION

The key advantage ofPCR-DGGE methods for studies of colonic bacterial ecology is the rapid detection of species changes within a given subject from baseline over time. The major obstacle to the application of this method to studies of disease­ association is the limited representation of the lower bowel flora with universal bacterial primer sets (157, 335, 388). This is illustrated by the limited number of bands generated from murine faecal samples seen in Figure 3.8. Therefore, the development of novel primer sets targeting the major phylogenetic groups within the mammalian bowel microbiota that will increase the information available from PCR­ DGGE, is clearly desirable. The primer sets described in this study divide the lower bowel microbiota into its three major constituent phylogenetic groups so that each can be represented on separate DGGE gels. Using this approach a representation of the species composition within each group can be obtained rapidly and compared with previous samples from the same individual. However many of the less numerically significant groups of colonic bacteria are not represented by these primer sets including the Atopobium group, Phascolarctobacterium group, Enterobacteriaceae, Veillonellae and enterococci.

Furthermore, certain caveats must be applied to the interpretation of the resulting gels. One of these is the fact that the C. leptum subgroup primer set also amplified E. limosum under the conditions used for PCR. This may be explained by the fmding that the forward primer Clept751 is identical to the sequence of E. limosum (ATCC 8486)- this species falls within the C. purinolyticum group of the RDP (Table 3.5). However E. limosum has multiple mismatches to the non-GC clamped reverse primer sequence of Clept1246R including the last two 3' bases (E. limosum sequence 5' -GTGGAA-3' compared with the reverse complement of the primer sequence 5'-TCGGAA-3'). It is likely therefore that 16S rDNAfrom some non­ target bacterial species are amplified by this primer set. However the new GC­ clamped primer sets for the Bacteroides-prevotella group and the C. coccoides group were specific for their positive controls.

-103- An additional factor to be considered in the interpretation of the gels is that three bands were produced with the Bacteroides-prevotella primer set following the amplification of DNA template from B. vulgatus (Figure 3.3). The presence of multiple DGGE bands for DNA from a pure culture may reflect multiple 16S ribosomal DNA copy number. Multiple 16S rDNA copies are present in a broad range of bacterial species and have been reported to result in multiple DGGE bands for PCR product from pure bacterial cultures in other studies (56, 88, 291). In fact B. vulgatus is known to contain 7 16S rDNA copies (158). Evidence supporting this proposition as an explanation for the multiple bands, was the finding that increasing the annealing temperature above 64°C did not change the banding pattern until temperatures exceeded 68°C and no detectable amplification occurred.

Comparison of the sequences of the three new primer sets with the murine lower bowel16S rRNA libraries of Salzman et al (2002) and Dewhirst et al (1999) revealed that the utility of a given primer is strongly influenced by the quality and breadth of the database on which the primers design is based (69, 288). It is important to note that many of the sequences in these published libraries are not represented in the database that was used to design the primer sets (i.e. the Rib0somal Database Project). Three of the unique sequences from the library of Salzman et al (288) were inappropriate matches to the Bacteroides-prevotella group primer set (Table 3.6). Using the SEQUENCE MATCH program of the RDP (55) it was found that there were no close relatives of each of the three sequences in the RDP database. The closest match to AJ400247 was C. clostridiiforrne of the C. coccoides group with 88.2% homology. Similarly, R. lactaris of the C. coccoides group was the closest match to AJ400265 with 87% homology while an environmental clone, U81697, was the closest match to AJ400237 with 80% homology. This comparison ~mphasises that primers optimally designed for a given database of sequences may not prove to be perfectly specific or comprehensive in the light of new sequence data. Nevertheless the new primer sets did produce significant coverage of the unique sequences of Salzman et al. The coverage of the

-104- sequences of sufficient length for evaluation was 20 of 38 or 53%. This calculation includes the 18 appropriate primer matches, 3 "inappropriate" matches and the fact that AJ400265 matches both the Bacteroides-prevotella group and C. coccoides group primer sets. If sequences from groups that were not targeted by the three primer sets such as Helicobacter species, Lactobacillus species andVerrucomicrobium species are excluded then the coverage increases to 20 out of 27 or74%.

The number of bands produced with the Bacteroides-prevotella group, C. coccoides group and C. leptum subgroup primer sets using DNA template from caecal samples of each mouse was 13.1 ± 4.3, 5.4 ± 1.7 and 3.2 ± 1.6 respectively. The number of bands amplified with each of the new primer sets (13:5:3 respectively) approximates the relative diversity of unique clones within the library of Salzman et al within these groups (14:8:2) (288). This would suggest that the gel banding patterns may constitute a good representation of species diversity within the three target groups. However direct examination of this question would require generation of clone libraries using each primer set that were large enough to estimate coverage by each library of the amplifiable sequences within a given sample. This would then be followed by the identification of the corresponding DGGE band position of each clone to determine how well the amplified genospecies diversity was represented on the DGGE gels.

The aim of developing primer sets to represent the lower bowel microbiota was principally to detect any changes that may occur within a given subset of bacteria over time. Thus, the fact that these primers are not absolutely specific or completely inclusive does not greatly impair their use for this application. However the caveat that the primers are not perfectly specific or completely inclusive must be considered in the interpretation of any results using this method. For example the finding that the banding profile obtained from serial samples of an individual with colitis does not change over time using the C. coccoides primer set could be interpreted as "No change in the bacterial populations of the C. coccoides group within the limits

-105- imposed by the primer set, PCR amplification and DGGE". It is possible that bacterial species within the C. coccoides group were not amplified by the primer set or were not present in sufficient numbers for their detection.

The banding patterns produced with each primer set were compared with Bray­ Curtis similarity for binary data. Each band position was encoded as a 0 or 1 for the absence or presence of a band at that position respectively. Other studies in bacterial ecology have compared banding patterns with other similarity measures such as Sorensen's index, however this requires an assessment of band intensity (217, 322). Whether it is reasonable to use band intensity in community analyses is controversial because of the known biases involved in PCR amplification discussed in Section 1.8.1, Chapter 1. Therefore an algorithm that did not rely on band intensity was applied.

The application of primer sets for theBacteroides-prevotella group, C. coccoides group and C. leptum subgroups to a murine model suggests that the species composition of caecal and faecal bacteria from an individual mouse within the targeted groups is nearly identical (Table 3.7, Figure 3.12). Therefore faecal samples can be used to study murine caecal microbiota within these, and possibly other groups, over time. Furthermore, the banding profiles obtained with all 4 primer sets examined in this study suggest that each mouse appears to have its own relatively distinct lower-bowel microbiota as has been previously reported in humans (388) (Figures 3.8 to 3.11).

Comparisons based on which cage the mice originated from showed that banding patterns derived from faecal or caecal samples of cagemates were more alike than comparisons between mice from different cages (Table 3.7, Figure 3.13). This was supported by the results of PCR-DGGE with primers for the domain Bacteria and the Bacteroides-prevotella group but not with the C. coccoides group or C. leptum subgroup primer sets (Table 3.7). Similarly while several of the comparisons by age and sex were significant with the Bacteroides-prevotella group primer set (Table

-106- 3.7) this was not so with the other primer sets. Potentially the different findings with different primers sets may reflect real differences in the acquisition of gram-negative and gram-positive bacteria. Alternatively the results might reflect a lack of diversity in the bacterial species represented by the C. coccoides group and C. leptum subgroup primer sets in different mice. Overall the results suggest that the bacterial species composition of the faecal microbiota of mice raised in the same cage has a greater similarity than mice raised in different cages. This is not surprising as the cagemates were generally from one or two litters and presumably acquired their bowel bacteria from the same dam(s) and cage environment (202, 354).

An additional explanation for the significant results obtained with comparisons by age and sex using the Bacteroides-prevotella group primer set is that this result may reflect the highly similar microbiota of cagemates. This possibility arises because each comparison based on age or sex compared groups of 2 cages of mice (e.g. the 2 cages of females with the 2 cages of males or the 2 cages of 4 week mice with the 2 cages of 6 week mice). Therefore each group in the age or sex analysis consisted of two groups of cagemates and the significant results for these comparisons may be due to an effect of mice being cagemates per se rather than an effect of age or sex.

An important question that arises from the observation that the murine caecal and faecal microbiota have a similar species composition is whether the same is true in human subjects? Two recent studies have examined the relationship between caecal and faecal microbiota in humans. In the first study, Marteau et al (2001) compared the bacterial populations of faecal samples and caecal content from eight subjects (213). The caecal sample was collected 2 to 3 hours after a standardised meal using a nasocaecal tube. In their study, the cultured counts of anaerobes, bifidobacteria and Bacteroides spp. were significantly lower in caecal than in faecal samples, while the counts of facultative anaerobes were similar in both sample types. A molecular approach to quantitation with 16S rRNA probes targeting bacteria from Bacteroides group, Clostridium leptum group and bifidobacteria confirmed the culture findings. For all three probes, hybridisation was significantly higher in the faecal as compared

- 107- with the caecal samples. Overall their results suggested that facultative anaerobes were much more numerous and important in the human caecum than would have been apparent from the examination of faecal samples.

In the second study by Zoetendal et al (2002) of the "mucosa-associated" flora, the bacterial populations of faecal samples and colonic biopsies from the prepared ascending colon of the same subject were compared using universal16S rRNA PCR-DGGE for the domain Bacteria (390). Their results supported the observations of Marteau et al in that the banding profiles representing the microbiota of colonic biopsies taken from the right colon and faeces were different. However there are significant sampling issues associated with this type of study. The colon had been washed-out with bowel preparation prior to colonoscopy. In addition, the biopsy forceps pass through the suction channel of the colonoscope and are therefore usually contaminated with the luminal fluid that is frequently evident at colonoscopy (51). This contamination may then coat the biopsy itself. So it is not clear whether the ascending colon samples were representative of the mucosa of the unprepared right colon or merely represent the luminal fluid that was present at the colonoscopy.

The studies of Marteau et al and Zoetendal et al suggest that the microbiota found in faeces may be different to that present in the caecum (213, 390). This finding is in contrast to the results reported in this study in mice. However there are substantial differences between the human and murine lower bowel that might explain this. For example dietary intake, anatomy (a proportionally much larger caecum in the mouse), and transit time are all different. The human colonic transit time of around 35 hours (221) is substantially longer than the rodent colonic transit time (less than 24 hours) (250, 351) and this difference is likely to be ecologically significant in view of the doubling times for many colonic bacteria. These differences could explain why the PCR-DGGE banding patterns obtained from the caecum and faeces of an individual mouse are nearly identical while those obtained from the human right colon and faeces by Zoetendal et al (2002) were not (390). However sampling issues and the effect of bowel preparation make it difficult to be sure that the

-108- observed differences in banding patterns for faecal samples and ascending colon biopsies from Zoetendal et al' s study reflect real differences in species composition in the unprepared state. In addition, the clear quantitative differences elucidated by Marteau et al (2001) are not necessarily inconsistent with similar DGGE banding profiles for caecal and faecal samples as the majority of the groups tested by probe hybridisation were detected in both samples (213). That is the range of species present in each sample may have been very similar and resulted in similar DGGE profiles despite the fact that quantitative differences in bacterial numbers were present. At this time therefore it remains an open question as to whether the human faecal PCR-DGGE banding pattern is representative of the caecal pattern.

The known minor limitations of the primer sets developed in this thesis do not detract greatly from their use in studies of the microbiota within individuals over time. The foundation for this approach is the fact that a number of studies in human adults have confirmed that the species composition of the bacteria in faeces is stable with PCR-DGGE for the domain Bacteria (290, 310, 332, 388, 390). Therefore any changes that are observed in the banding patterns of an individual in association with clinical events may be significant. If a change is seen at PCR-DGGE then this can be further investigated with species-specific PCR for bacteria from within that bacterial group or by cutting out and sequencing the bands of interest (this is not always possible because multiple genospecies may be present in the one band (309)). Alternatively cloning and sequencing of the PCR product, with comparison of the band position of each clone to the banding profile, can also provide evidence concerning which bacterial species may have changed (388). The knowledge that such a change occurs can also direct further studies in other individuals using FISH or culture targeting bacteria from within that group. In this way it should be possible to identify changes in bacterial species composition occurring in association with disease development in individuals.

In summary, this work highlights the strengths of primer sets for PCR-DGGE targeting major groups within the of lower bowel flora of mammals. The application

-109- of these primer sets to faecal samples collected over time should reveal whether changes in species composition have occurred, and facilitate studies of colonic disease. In a murine model free of the sampling issues affecting human studies, the species present in faecal samples was strongly representative of the caecal microbiota.

-110- CHAPTER 4: Development and validation of Helicobacter genus specific PCR­ DGGE

4.1 INTRODUCTION

Bacteria with the capacity to colonise the mucus layer of the large intestine are known to be potentially pathogenic. For example, in both humans and animals Campylobacter species can cause a range of disease states, including acute colitis and less frequently bacteraemia. Following the discovery of the pathogenic role of Helicobacter pylori in human gastroduodenal disease in 1984, the identification of several murine lower bowel Helicobacter species raised questions about the potential role of these mucus-colonising bacteria in diseases of the large intestine (212).

There are eight murine lower bowel Helicobacter species; H. muridarum (190), H. hepaticus (372), H. bilis (106), H. trogontum (220), H. rodentium (316), H. ganmani (278), Helicobacter typhlonius (107) and Helicobacter muricola (380). To date only 4 isolates of H. muricola have been reported by Won et al (2002), thus technically this species does not meet the criteria for official naming (71, 380). All eight species are motile gram-negative microaerophilic or anaerobic bacteria that have adapted to live in the intestinal mucus layer and crypts. These bacteria are primarily mouse colonisers with the exception of H. trogontum, a rat colonic organism that will colonise the lower bowel of gavaged germ-free mice (232).

Colonisation of the lower bowel of mice by Helicobacter spp. is prevalent in animal research laboratories (102, 137, 374). Franklin et al (2001) performed PCR with Helicobacter species and genus specific primers on samples from 1271 mice obtained from a range of research centres. In their study 16.4% of mice were shown to be positive for H. hepaticus, 15.1% for H. rodentium, 4.9% for H: typhlonius, 4.3% for H. bilis and 10.5% for other Helicobacter species (107). Similarly 62 of 167 mice (37%) from a large number of different murine strains housed at The Jackson Laboratory (Bar Harbor,

-111- Maine) were positive for Helicobacter species by PCR or culture in a study by Mahler et al (1998) (207). Of the 167 mice, 16.7% were found to be colonised by H. hepaticus, 13.2% by H. muridarum and 6% by H. bilis. The high prevalence of this genus in the colon of mice relates to two factors. Firstly, murine lower bowel Helicobacter species persistently colonise the colon despite the fact that a specific immune response develops in the infected mice (particularly those infected with H. hepaticus) (74, 104, 105, 197, 374). The second reason for the high prevalence of infection is that Helicobacter species are known to be easily transmitted by the faecal-oral route. In a study by Livingston et al (1998) 66% of uninfected mice that were exposed to bedding from H. hepaticus infected mice were colonised within 2 weeks. By 4 weeks all of the mice were infected (197). An interesting study by Whary et al (2000) has suggested that the efficiency of faecal­ oral transmission may vary depending upon the Helicobacter species (374). For example, these authors detected H. hepaticus by PCR in faecal samples from 60% of sentinel mice at 1 month and 100% by 3 months when their cages were contaminated with bedding from H. hepaticus infected mice. In comparison, the rate of H. rodentium

acquisition was 43% at one month, 81% at 3 months and 88% at~ months when bedding from H. rodentium infected mice was used. Whether the lower apparent transmission rate for H. rodentium reflects a real difference in acquisition, or other factors such as relatively lower burdens of colonic infection or a lower PCR sensitivity, is not known. Regardless, the transmission of both organisms appears to be relatively efficient.

Murine lower bowel helicobacters have been shown to cause inflammation and neoplasia in the colon and liver of susceptible murine strains (11, 48, 104, 107, 108, 183, 194, 320, 372). The prevalence and severity of the pathological changes are a function of both the infecting Helicobacter species and host immunodeficiency. Occasional reports of mild typhlocolitis in outbred mice, or inbred strains that are thought to be immunocompetent have also been published (105). Pathological changes are most frequently reported in severely immunocompromised mice including severe combined immunodeficient (SCID), nuclear factor kappa B (NF-ld3) deficient and interleukin 10 (IL-10) knockout mice (85, 102, 108, 109, 194, 319, 320). For example, typhlocolitis has

- 112- been reported in SCID and multiple drug resistant- deficient mice infected with H. bilis (205, 320); SCID, T cell deficient, IL-10 knockout and NF-KB-deficient mice infected with H. hepaticus (48, 85, 183, 194, 371) and SCID and IL-10 knockout mice infected with H. typhlonius (102, 107, 109). However not all studies have supported the contention that Helicobacter species and in particular H. hepaticus play an important role in inducing colitis in immunodeficient mice (74, 203).

The severity of typhlocolitis in immunodeficient mice in response to lower bowel helicobacter colonisation also appears to relate to the strain of mouse. For example, IL- 10 deficient mice on a C57BL/6 background show less severe colitis than IL-10 deficient 129/SvEv or BALB/C mice when infected with H. hepaticus (20, 183). Interestingly, susceptibility to colitis in chemically-induced (such as the dextran sulfate­ sodium model) and other cytokine-knockout models is also dependent on the strain of mouse (208, 224, 287). A study of back-crosses from two interleukin-10 deficient murine strains with differing susceptibility to colitis by genome-wide scanning revealed that several chromosomal loci influenced disease development (209). Indeed in this study by Mahler et al (2002) six colitis susceptibility loci on chromosomes 1, 2, 3, 8, 17 and 18 were reported to modify disease severity (209). A similar study in the dextran sulfate-sodium model, identified colitis susceptibility loci on chromosomes 1, 2, 5 and 18 (208). Host genetics is therefore a key factor in the severity of many forms of colitis and, in particular, that which develops in association with lower bowel Helicobacter species.

The pathogenic potential of each lower bowel Helicobacter species appears to relate to its virulence factors. The species that are most commonly associated with colonic disease in mice include H. hepaticus, H. bilis and H. typhlonius (48, 104, 105, 108, 109, 183, 194, 319, 320). All three of these species possess genes encoding orthologs of the cytolethal distending toxin of pathogenic strains of E. coli and Campylobacter species (47, 107, 176, 384). Both H. hepaticus and H. bilis have been shown to produce a functional protein with cytopathic effects however it is not clear whether this is also true

-113- for H. typhlonius (47, 176, 384). The sequencing of the genome of H. hepaticus has revealed several other putative virulence factors (336). For example, most strains of H. hepaticus possess a large genomic island (HHGil) that may encode a Type IV secretion system (336). H. hepaticus also has genes that encode an ortholog of the C. jejuni adhesin PEB1 (336). In summary, murine Helicobacter species vary in their pathogenicity (with H. hepaticus appearing to be of the greatest virulence) and can induce colitis in immunodeficient mice, the severity of which is determined by a range of susceptibility genes.

While H. hepaticus and H. bilis have been shown to translocate from the bowel to the liver, other species do not readily do so (101, 102, 104-107, 336). Subsequent hepatitis development appears to be determined by host genetic factors, with immunodeficient and certain inbred strains being susceptible, while outbred C57BL/6 mice are relatively resistant (104, 372). SCID mice have been reported to develop hepatitis following H. hepaticus and H. bilis infection (194, 320). Inbred AJJCr mice also develop hepatitis and hepatocellular tumors when infected with H. hepaticus (101, 104, 105). Intraperitoneal injections of liver homogenates from infected AJJCr mice have also been shown to reproduce the hepatitis in previously uninfected mice, although it is important to note that significant rates of liver pathology were seen in uninfected AJJCr mice in a later study (11, 372). Like colonic disease, the development of hepatic disease in response to Helicobacter species is dependent upon the infecting species and genetically determined host susceptibility.

In view of the evidence suggesting that Helicobacter species are present in the human colon and that a number of lower bowel helicobacters are pathogenic in certain susceptible strains of mice, the hypothesis arose that lower bowel Helicobacter species might be of aetiologic significance in humans with inflammatory bowel diseases. As discussed in Section 1.9 of Chapter 1 and later in Chapter 5 a number of research findings have suggested that Helicobacter species may be present in the human colon. DNA from Helicobacter species has been detected in biopsies and tissue samples from

-114- the human ileum, colon and liver by PCR with or without probe hybridisation (12, 27, 46, 99, 112, 196, 239, 240, 249, 349). In addition, bacteria with a spiral morphology have been observed using electron microscopy (EM) in the mucus layer that lines the colon of a small number of individuals examined immediately after death (61). Studies in animals suggest that mucus-associated bacteria with a spiral morphology are commonly found to be Helicobacter species when isolated in pure culture (70, 72, 101, 103, 190). However direct evidence for long-term colonisation of the human colon with commensal Helicobacter species or any association with chronic disease is lacking. In view of this important question in humans, and the potentially confounding effect of helicobacter colonisation in mice, appropriate assays for the detection and speciation of helicobacters are needed.

A number of culture-based and molecular methods have been used to determine whether the lower bowel of a given mouse is colonised by Helicobacter species and if so which species. The most commonly applied methods include genus-specific PCR, species­ specific PCR and culture. In contrast to PCR-DGGE, these methods usually involve much greater time and effort, and culture in particular may lack sensitivity. For example, it can take 5-7 days for initial growth of some Helicobacter species to be apparent in plate culture (386). In addition, culture for Helicobacter species has a highly variable reported sensitivity that appears to depend upon the growth characteristics of the individual Helicobacter species (11, 207, 314, 320). To culture Helicobacter species, scrapings of the intestinal mucosa or faecal samples can be plated directly onto Campylobacter Selective Agar (CSA) containing vancomycin, polymyxin and trimethoprim or cultured on non-selective media such as Horse Blood Agar using a filter technique (333). One species, H. ganmani, will not however grow on CSA (278). The reported sensitivity of culture using murine faecal or caecal samples ranges from 16 to 90% for H. hepaticus (11, 207, 314), 64% for H. muridarum (207) and 10 to 100% for H. bilis (207, 320). The isolation of helicobacters in pure culture is particularly difficult in mice colonised with two or more Helicobacter species as growth occurs as a thin spreading film on agar plates, and not as discrete colonies (101, 106, 190,220, 278). The

-115- subsequent identification of cultured bacteria at a species level by biochemical testing or species-specific PCR is also time consuming (386).

The principal alternatives to culture are the PCR-based assays. While genus-specific PCR appears to be sensitive (207, 275), assays for each individual species are then required for speciation (16, 207, 314). The widely applied genus-specific PCRprimer set of Riley et al (1996) has been shown to be 100% sensitive for H. hepaticus and H. bilis in intestinal tissue from mice proven to be infected with each organism by culture (275). In a second study of murine caecal samples that compared PCR for the Helicobacter genus with culture, Mahler et al (1998) found that of 167 mice tested 105 were negative by both methods, while 60 were positive by PCR and only 37 were positive by culture (207). In contrast, a study of H. hepaticus -specific PCR and culture in 56 mice showed almost 100% concordance of results with 30 animals neg~tive by both methods, 26 positive by PCR and 24 positive by culture (314). In general, PCR assays are reported to be more sensitive than culture for Helicobacter species. However in the publications that describe species-specific PCRs for H. typhlonius, H. ganmani, H. trogontum and H. muridarum, direct assessments of PCR sensitivity in comparison with culture have not been performed (107, 138, 220, 278). An additional limitation of the PCR approach is that H. ganmani and H. rodentium cannot be distinguished by current PCR assays despite only 98.2% 16S rRNA sequence homology (278). As at least 8 Helicobacter species have been reported to be present in the murine lower bowel, detection and speciation with individual species-specific PCRs is also very laborious.

The principal aim of this study was to develop a PCR-DGGE assay that would allow the speciation of bacteria from the Helicobacter genus, and in addition, detect Helicobacter spp. that are not easily cultivable or not detectable with published species-specific primer sets (151, 291, 368). PCR-DGGE assays targeting bacteria from a particular genus found within the colon have been developed for Bifidobacterium species (291), Lactobacillus and related species (151, 368) and, following the publication of the work described in this chapter, by other authors for Helicobacter species (1, 138). The main

-116- advantage of PCR-DGGE is the ability to represent the species within a targeted genus in a single PCR and gel electrophoresis.

The second aim of this work was to validate the PCR-DGGE assay for the Helicobacter genus using faecal template DNA. This method was developed in mice because of the known diversity of murine lower bowel Helicobacter species and the fact that concurrent colonisation with multiple species can occur (137, 319, 374). The utility of the method to detect and differentiate concurrent colonisation with multiple species could therefore be directly assessed. Furthermore, lower bowel Helicobacter species have been shown to be pathogenic in susceptible murine strains, thus knowledge of the helicobacter colonisation status of mice used in research studies of colonic and hepatic disease is desirable in its own right. The assay was developed such that it could also be applied in future studies of human lower bowel samples from IBD cases and controls.

-117- 4.2 METHODS

4.2.1 Development and validation of the Helicobacter genus-specific primer set and PCR-DGGE

Seventy-three 16S ribosomal DNA (16S rDNA) sequences from Helicobacter species including all named murine lower bowel helicobacters and all helicobacters isolated from humans were downloaded from the GenBank database (Table 4.1a) (19). These sequences were compared with the 16S rDNA sequences of 30 other colonic bacteria including Campylobacter species (Table 4.1b). An alignment of all sequences was performed with CLUSTALW (348). Regions of 16S rDNA with complete homology among the Helicobacter species were identified. The regions that differed from other colonic bacteria were identified as potential primer sites. Primer 1067R was chosen to take advantage of a region of variability among the Helicobacter species between bases 1012 and 1047 (E. coli 16s rDNA numbering). This variability was targeted to facilitate the separation of PCR products derived from different species at DGGE. A reversed and GC clamped version of primer H676 described by Riley et al (275) was paired with 1067R to create an expected product of 421 base pairs that was suitable for DGGE.

The primer sequences and conditions for PCR amplification are shown in Tables 2.1 and 2.3 of Chapter 2 respectively. For pure cultures and murine faecal samples, 10 ng and 30 ng of DNA respectively were used as PCR template. The optimal annealing temperature was determined by Gradient PCR (Hybaid) using annealing temperatures ranging between 55°C and 70°C. The optimised thermal cycling conditions consisted of 94°C for 5 minutes, 30 cycles of 94°C for 10 s, 62°C for 10 s, 72°C for 30 s and 72°C for 2 minutes. To assess the specificity of amplification, DNA from reference and laboratory strains of murine lower bowel helicobacters (H. rodentium, H. muridarum, H. trogontum, H. bilis, H. hepaticus and H. ganmani) and 15 other colonic bacteria (B. longum, F. mortiferum, L. salivarius, B. vulgatus, B. fragilis, C. histolyticum, E.

-118- limosum, P. anaerobilus, E. coli, E. faecalis, L. acidophilus, P. acnes, B. cereus, C. coli and C. fetus) were used as template.

The PCR product amplified from the DNA of the Helicobacter species above was used to optimise the gradient for DOGE so that maximal band separation was achieved. The initial gradient used was 20 to 70% (100% denaturant solutions contained 40% deionised formamide and 7M urea (Sigma)). Based on preliminary results this gradient was gradually narrowed to maximise band separation such that a gradient of 41-48% was used for all further experiments. Electrophoresis was performed for 16 hours at 75V and 60°C (Dcode system, Bio-rad) and the gels were stained with ethidium bromide.

4.2.2 Limit of detection ofHelicobacter genus-specific PCR-DGGE

The limit of detection of the PCR-DGGE assay for Helicobacter species in murine faecal samples was determined by dividing a fresh faecal pellet from a C57BL/6 IL-10_,_ mouse known to be colonised with H. ganmani (and not H. hepaticus by species-specific PCR) into 5 mg portions in separate 1.5 mL tubes. A stock solution of H. hepaticus was made by culturing H. hepaticus on HBA under anaerobic conditions for 3 days. The resulting growth was harvested in sterile PBS and bacterial numbers were quantitated in triplicate using an Improved Neubauer Haemocytometer. Each portion of stool was then spiked with 10 f..Ll of serial 10 fold dilutions of the stock solution of H. hepaticus ranging from approximately 108 to 103 per mL. A control sample of stool remained unspiked. After seeding of the faecal samples DNA isolation was performed (Section 2.1.2.2, Chapter 2), followed by Helicobacter genus-specific PCR-DGGE.

4.2.3 Comparison of the sensitivity of Helicobacter genus-specific PCR-DGGE with Species-specific PCR and Culture

4.2.3.1 Sample collection and culture

-119- To compare the sensitivity of the Helicobacter genus-specific PCR-DGGE with that of species-specific PCR and culture for assessing helicobacter colonisation status, thirteen 12 week-old wild-type C57BL/6 mice from the same supplier housed in 2 cages were sacrificed by C02 asphyxiation, followed by cervical dislocation. The ileum and large bowel were removed and a stool pellet from the distal colon or rectum was divided into 2 equal parts. DNA was extracted from the first part of the stool for use in the Helicobacter genus-specific PCR-DGGE and species-specific PCRs for H. rodentium and H. ganmani, H. hepaticus and H. muridarum (see below). Caecal mucus scrapings and the second part of the faecal pellet were homogenised in saline and cultured under anaerobic and microaerophilic conditions on HBA using the filter technique (Section 2.1.1.6, Chapter 2). HBA alone was used as H. ganmani will not grow on CSA (278). Plates were examined macroscopically for a thin spreading film characteristic of Helicobacter species after 3 days of incubation. Phase contrast microscopy and subculture was then performed every 3 to 5 days for microaerophilic plates, and every 5 to 7 days for anaerobic plates, for up to 4 weeks. The identity of isolates obtained in pure culture was confirmed by DNA ~xtraction and Helicobacter genus-specific PCR-DGGE.

4.2.3.2 Species-specific PCR

Thirty nanograms of DNA isolated from the faecal samples was used as template in the species-specific PCRs for H. rodentium and H. ganmani, H. hepaticus, H. muridarum, and H. bilis. The primers and conditions used for these PCR reactions are shown in Tables 2.1 and 2.3, Chapter 2.

4.2.3.3 Nested PCR

Samples that were negative in the first round PCR with primers for H. rodentium and H. ganmani were subjected to nested PCR. Nested PCR for H. rodentium and H. ganmani used the product of a reaction for 16S rRNA with universal bacterial primers F27 and R1494 diluted 1:25 in water as template (Tables 2.1 and 2.3, Chapter 2). Mice were

-120- considered to be colonised by a given Helicobacter species if they were positive by any method (species-specific PCR, nested species-specific PCR, culture or PCR-DGGE).

4.2.4 Sequencing

PCR products obtained using DNA template from all of the reference and laboratory strains of Helicobacter species using the primers GC658F and 1067R were directly sequenced using the techniques outlined in Chapter 2, Section 2.1.6. In addition, representative DGGE bands obtained using PCR product amplified from the thirteen C57BL/6 mice were excised from ethidium bromide stained gels using the "crush and soak" method outlined in Section 2.1.7. The resulting solution was re-amplified by PCR with primers GC658F and 1067R using the conditions discussed above and the resulting product was directly sequenced.

4.2.5 Application of Helicobacter genus-specific PCR-DGGE to mice from the BABS Animal Facility

To determine the helicobacter colonisation status of mice housed in the School of Biotechnology and Biomolecular Science Animal Facility, fresh faecal samples were obtained from eight mice (6 BALB/C, and 2 C57BL/6) housed in 8 different cages (designated A to H) in four rooms of the animal facility. The mice ranged in age from 3 to 13 months and had previously originated from 3 Australian suppliers (Animal Resources Centre, Canningvale, W A; Biological Resources Centre, Sydney, NSW; Walter and Elisa Hall Institute, Kew, VIC). DNA isolation, Helicobacter genus-specific PCR and DGGE were performed as described previously. After extraction from the gel using the "crush and soak" method all of the DGGE bands obtained from the 8 mice were sequenced.

- 121- 4.3 RESULTS

4.3.1 Development of the Helicobacter genus-specific PCR-DGGE

Gradient PCR with H. hepaticus template DNA revealed amplification at annealing temperatures of up to 67.6°C, with strong amplification at 61.4°C or below (Figure 4.1). Gradient PCR with C. coli template revealed amplification of 16S rDNA from Campylobacter species at annealing temperatures below 61 °C (Figure 4.2), thus the use of an annealing temperature greater than 61 °C was critical for specificity. By combining these results, an annealing temperature of 62°C was chosen. At this annealing temperature Helicobacter genus-specific PCR amplified DNA from all of the Helicobacter species tested including H. hepaticus, H. rodentium, H. muridarum, H. bilis, H. trogontum, and H. ganmani (Figure 4.3). In contrast, DNA from the other colonic bacteria -·F. mortiferum, L. salivarius, P. anaerobilus, B. cereus, E. faecalis, C. histolyticum, E. limosum, B. longum, B. vulgatus, B. fragilis, C. fetus, C. coli, and E. coli - was not amplified (Figure 4.4). The optimal gradient for DGGE was gradually narrowed from 20 to 70% (Figure 4.5) to 41 to 48% (Figure 4.6). The bands for most species were well separated at this optimised gradient of denaturants. The DGGE bands obtained with reference and laboratory strains of Helicobacter species are shown in Figure 4.6. Band positions were specific for reference strains of H. trogontum, H. bilis, H. hepatic us, H. muridarum and a laboratory strain of H. ganmani from ll..-1 o-t­ C57BL/6 mice, and each could easily be differentiated. However, H. ganmani isolated from wild-type C57BL/6 mice in our animal facility had an identical band position to the reference strain of H. rodentium. (Lanes 5 and 7 in Figure 4.6).

-122- 4.3.2 Limit of detection for H. hepaticus using Helicobacter genus-specific PCR­ DGGE

Bands representing H. hepaticus were detected by PCR-DGGE from samples containing 106 and 105H. hepaticus per gram of faeces but not in those containing 104 per gram or less, that is, the limit of detection was 105 H. hepaticus per gram of faeces (Figure 4.7).

4.3.3 Comparison of the Helicobacter genus-specific PCR-DGGE with species­ specific PCR and culture

Using a combination of species-specific PCR, nested species-specific PCR, culture and genus specific PCR-DGGE all13 mice were found to be colonised with H. bilis and H. ganmani (Table 4.2).

4.3.3.1 Culture of caecal mucus scrapings and faecal samples

H. bilis grew rapidly on HBA plates incubated under microaerophilic conditions and required subculture every 3 days. H. ganmani grew slowly on HBA under anaerobic conditions only, and required subculture every 7 days. Using phase contrast microscopy, bacteria with a morphology consistent with H. bilis were observed in microaerophilic cultures of faecal and/or caecal samples from 10 of the 13 mice. Similarly, bacteria morphologically identical to H. ganmani were observed in anaerobic cultures from 7 of the 13 mice. H. bilis is a long tapering motile rod while H. ganmani is a relatively short S-shaped bacterium (106, 278). Unfortunately overgrowth of other bacteria prevented the isolation of these helicobacters in pure culture in the majority of cases, so that H. bilis and H. ganmani were each isolated in pure culture from faecal and/or caecal samples of only 3 mice (Table 4.2). The identity of the isolates obtained in pure culture was confirmed by PCR-DGGE using DNA extracted from culture plates.

- 123- 4.3.3.2 Species-specific PCR using faecal template DNA

The species-specific PCR for H. bilis was positive in 12 of 13 cases (Table 4.2). The sample that was negative in the H. bilis PCR was positive by Helicobacter genus­ specific PCR-DGGE. H. rodentium and H. ganmani PCR was also positive in 12 out of 13 mice (Table 4.2). The sample that was negative in the H. rodentium and H. ganmani PCR was however positive by nested PCR using 16S rDNA PCR product as template. Species-specific PCRs for H. hepaticus and H. muridarum were negative for all mice.

4.3.3.3 Helicobacter genus-specific PCR-DGGE

PCR-DGGE using faecal samples from the 13 mice resulted in bands representing two Helicobacter species. Results for 8 of the mice are shown in Figure 4.8. The upper and lower bands had the same gel positions as the reference strains of H. rodentium and H. bilis respectively. Sequencing of PCR products obtained from the DGGE gels showed that the sequences for the upper and lower bands were identical to H. ganmani and H. bilis respectively. In all13 mice bands representing H. bilis were observed, and in 12 of 13 mice bands representing H. ganmani were observed. The sample that was negative for H. ganmani by PCR-DGGE represented in lane 5 of Figure 4.8 was the same sample that was negative by H. rodentium and H. ganmani PCR, but positive by nested PCR.

4.3.3.4 Comparison of the sensitivity of Helicobacter genus-specific PCR, species­ specific PCR and culture

Combining the results of species-specific PCR, nested H. rodentium and H. ganmani PCR and PCR-DGGE all 13 mice were found to be colonised by H. bilis and H. ganmani. Compared with this result, isolation in pure culture was only 23% sensitive for either bacterium. However, the sensitivity of culture was greatly increased if the definition of a positive result was changed from isolation in pure culture to visualisation of bacteria with a consistent morphology on plates using phase-contrast microscopy after 3 and 7 days of microaerophilic and anaerobic culture, respectively. Using this less

-124- stringent approach, culture detected H. bilis in 77% of the mice and H. ganmani in 54% of the mice. In comparison, species-specific PCR for H. bilis and H. rodentium and H. ganmani were both 92% sensitive, while PCR-DGGE was 92% sensitive for H. ganmani and 100% sensitive for H. bilis.

4.3.4 Application of PCR-DGGE to mice from the BABS Animal Facility

The results of PCR-DGGE using template DNA extracted from the faeces of mice from different cages (A to E) is shown in Figure 4.9. Where multiple bands were present from one mouse these are numbered from the top of the gel down. The sequences obtained by PCR amplification of DNA from each band are shown in Figure 4.10. The sequencing results for reference strains of H. rodentium, H. hepaticus, H. bilis, H. muridarum, H. trogontum, and laboratory strains of H. ganmani isolated from wild-type and IL-10·'­ C57BL/6 mice are also shown for comparison in Figure 4.10. The overall results of PCR-DGGE and sequencing are summarised in Table 4.3.

The PCR product from 6 of the 8 mice examined produced bands that matched the gel position of the reference and laboratory strains (Figure 4.9). Bands A2, Dl and E1 had identical gel positions to the reference strain of H. rodentium. However, the sequencing results for these bands were identical to H. ganmani (Figure 4.10). B1 had an identical band position and sequence to the laboratory strain of H. ganmani isolated from other IL-10·'- mice (Figure 4.9). This strain differed from the other strain of H. ganmani found in mice from cages A, D, and E by a single base only- a thymidine replacing guanidine at position 971 (E. coli 16S rDNA numbering), Figure 4.10. The PCR product from mice from Cage C, Cage F and Cage H produced a single band matching the gel position and sequence of H. hepaticus (Figure 4.9). The mice from cages D and E produced 2 bands. The first represented H. ganmani and the second matched the gel position and sequence of H. bilis (Figure 4.9). The amplification of DNA from the Cage G mouse did not produce any PCR product or DOGE band. Thus the most frequent types of infection were co-colonisation with H. bilis and H. ganmani or colonisation with H. hepatic us or

-125- H. ganmani alone. Interestingly, electrophoresis of PCR product from the mouse from cage A showed 3 bands (Figure 4.9). When compared with the reference strains two of these bands (A1 and A3 in Lane 3, Figure 4.9) did not match any of those represented in the marker. The sequence of band A1 was identical to the 16S rRNA of a helicobacter closely related to H. rodentium (HSU96299) (316) while band A3 was 98.3% homologous to a helicobacter isolated from dog stomach (HSU51874) (81).

-126- TABLE 4.1a. Target sequences used in the multiple sequence alignment for primer design.

GenBank Target sequence accession M88148 Helicobacter acinonyx (strain Eaton 90-119-3) AF297868 Helicobacter aurati U51873 Helicobacter bilis U18766 Helicobacter bilis Hb 1 AF054570 Helicobacter bilis strain HRI3caefr AF047844 Helicobacter bilis strain MIT 93-3055 AF047847 Helicobacter bilis strain MIT 97-6456 Y09404 Helicobacter bizzozeronii AF103883 Helicobacter bizzozeronii AF302107 Helicobacter bizzozeronii AF127027 Helicobacter bovis AF262037 Helicobacter canadensis U65102 Helicobacter canis AF177475 Helicobacter canis U04344 Helicobacter canis NCTC 12220 U46129 Helicobacter cholecystus M88150 Helicobacter cinaedi (CCVG 18818) AF207738 Helicobacter cinaedi strain CCUG 33887 M57398 Helicobacter.felis AF103879 Helicobacter .felis strain 937-12 M88154 Helicobacter.fennelliae (CCUG 18820) AF000223 Helicobacter ganmani strain ABHU H11 AF000221 Helicobacter ganmani strain CMRI H02 AF000222 Helicobacter ganmani strain CMRI H03 AF000224 Helicobacter ganmani strain UNSW H16 AF058768 Helicobacter heilmannii isolate 1 L39122 Helicobacter hepaticus AJ007931 Helicobacter hepaticus AF302103 Helicobacter hepaticus AF072471 H elicobacter mesocricetorum AF010140 Helicobacter muridarum AF013464 Helicobacter muridarum M35048 Helicobacter mustelae AF348617 Helicobacter nemestrinae X67854 Helicobacter nemestrinae (ATCC 49396T) AF302105 Helicobacter pametensis AF047850 Helicobacter pullorum U00679 Helicobacter pylori TABLE 4.1a. Target sequences used in the multiple sequence alignment for primer design (continued).

GenBank Target sequence accession U01331 Helicobacter pylori isolate MC937 U96296 Helicobacter rodentium U96297 Helicobacter rodentium Y09405 Helicobacter salomonis M88147 Helicobacter sp. (strain B9A Seymour) AF054574 Helicobacter sp. ABHU3cae U96299 Helicobacter sp. 'Eaton 94-536' AF225546 Helicobacter sp. 'Flexispira taxon 2' AF225547 Helicobacter sp. 'Flexispira taxon 3' AF225548 Helicobacter sp. 'Flexispira taxon 4' M88138 Helicobacter sp. 'Flexispira taxon 8' ATCC 43879 AF072333 Helicobacter sp. 'hamster B' AF142583 Helicobacter sp. 'liver 1' AF142584 Helicobacter sp. 'liver 2' AF142585 Helicobacter sp. 'liver 3' AF225550 Helicobacter sp. 'MIT 94-022' U96298 Helicobacter sp. 'MIT 95-2011' AF107494 Helicobacter sp. 'MIT 97-6194-5' AF320621 Helicobacter sp. MIT 98-5357 AF333341 Helicobacter sp. :MIT 98-6070 AF333339 Helicobacter sp. MIT 99-5504 AF292377 Helicobacter sp. 'MIT 99-5657' AF292376 Helicobacter sp. 'MIT 99-5660-6' AF292379 Helicobacter sp. ':MIT 99-5665' AF142147 Helicobacter sp. 'pig B 1' AF142151 Helicobacter sp. 'pig F8' AF292381 Helicobacter sp. 'Solnick 9Al-T71' AJ011431 Helicobacter sp. strain MZ 640285 AF054576 Helicobacter sp. UNSW3SBsp AF237612 Helicobacter sp. Wee Tee AF127028 Helicobacter suis AB006148 Helicobacter suncus strain:Kaz-2 U65103 Helicobacter trogontum AF061104 Helicobacter typhlonius TABLE 4.lb. Non-target sequences used in the multiple sequence alignment for primer design.

GenBank accession Non-target sequence X83935 Bacteroides fragilis (ATCC 25285T). X83944 Bacteroides fragilis L16489 Bacteroides thetaiotaomicron (ATCC 29148) M58762 Bacteroides vulgatus L37787 Bacteroides gracilis M58739 Bifidobacterium longum ATCC 15707 L04313 Campylobacter curvus AF219233 Campylobacter fetus MGH 97-2126 U03022 Campylobacter helveticus AF062491 Campylobacter hominis HS-B AF097682 Campylobacter hyointestinalis subsp. hyointestinalis strain 0943 AF097685 Campylobacter hyointestinalis subsp. lawsonii AF550627 Campylobacter jejuni AF043423 Campylobacter lanienae strain UB 993 L04317 Campylobacter rectus L06974 Campylobacter showae AJ002592 Clostridium butyricum strain MW8 AF072474 Clostridium dif.ficile strain 79685 M59094 Clostridium histolyticum AF262239 Clostridium leptum M59096 Clostridium limosum AB041865 Clostridium novyi (ATCC27323) X80724 Escherichia coli (ATCC 25922) X80731 Escherichia coli (pk3) AB036835 Enterococcus faecalis AF064242 Eubacterium limosum X85022 F aecalibacterium prausnitzii AF044948 Fusobacterium necro_p_horum AF089108 Lactobacillus salivarius sub~ salivarius AB041617 Propionibacterium acnes TABLE 4.2: The results of genus-specific PCR, species-specific PCR and culture* Helicobacter genus specific Species-specific Culture PCR-DGGE PCR H. bilis N=13 13 (100%) 12 (92%) 3 (23%) H. ganmani N=13 12 (92%) 12 (92%) 3 (23%) *One mouse was negative for H .ganmani in all three assays. However, using amplified 16S rDNA as the template in the H. rodentium and H. ganmani PCR revealed that this mouse was colonised by H. ganmani. TABLE 4.3. The combined results of Helicobacter genus-specific PCR-DGGE and sequencing of the 8 mice from the BABS animal facility: Cage Strain Age in Helicobacter species months A BALB/C 9 HSU96299 (316) H. ganmania HSU51874 (81) B C57BL/6 ILIO·'· 8 H. ganmanib c BALB/C 12 H. hepaticus D C57BL/6 2 H. ganmania H. bilis E BALB/C 5 H. ganmania H. bilis F BALB/C 12 H. hepaticus G BALB/C 11 Negative H BALB/C 12 H. hepaticus

a laboratory strain of H. ganmani with guanidine at position 971 (E. coli numbering). blaboratory strain of H. ganmani with thymidine at position 971 (E. coli numbering). Figure 4.1: The gradient PCR with H. hepaticus DNA template. Lane 1: annealing temperature of 55.40C, Lane 2: 56.40C, Lane 3: 57.70C, Lane 4: 59.40C, Lane 5: 61.40C, Lane 6: 63.30C, Lane 7: 65.30C, Lane 8: 67.60C, Lane 9: 690<::, Lane 10: 7f!JC, Lane 11: marker FN-1.

1 2 3 4 5 6 7 8 9 10 11

692 404

Figure 4.2: The gradient PCR with C. coli template DNA. Lane 1: 55.80C, Lane 2: 56.90C, Lane 3: 58.2°C, Lane 4: 59.50C, Lane 5: 60.90<::, Lane 6: 62.40C. 12 3 4 56 7

692 404 Figure 4.3: The results of Helicobacter genus-specific PCR for murine colonic Helicobacter species: Lane 1: H. hepaticus, Lane 2: H. rodentium, Lane 3: H. muridarum, Lane 4: H. hili , Lane 5: H. trogontum, Lane 6: H. ganmani, Lane 7: Negative control, Lane 8: Marker FN -1.

Figure 4.4: The results of Helicobacter genus-specific PCR with non-helicobacter species. Lane 1: Negative control, Lane 2: H. hepaticus, Lane 3: B. longum, Lane 4: F. mortiferum, Lane 5: L. salivarius, Lane 6: B. vulgatus, Lane 7: B.fragilis, Lane 8: C. histolyticum, Lane 9: E. limosum, Lane 10: P. anaerobilus, Lane 11: C. coli, Lane 12: C. fetus, Lane 13: Marker FN-1.

1 2 3 4 5 6 7 8 9 10 11 12 13

692 404 1 2 3 4 5 Figure 4.5: The initial PCR­ DGGE gel with a non­ optimised denaturant gradient of 20 to 70%. Lane 1: H. hepaticus, Lane 2: H. ganmani, Lane 3: H. rodentium, Lane 4: H. muridarum, Lane 5: H. bilis.

Figure 4.6: The results of Helicobacter genus-specific PCR and DGGE for reference and laboratory strains of Helicobacter species: Lane 1: H. trogontum, Lane 2: H. bilis, Lane 3: H. hepaticus, Lane 4: H. muridarum, Lane 5: H. rodentium, Lane 6: H. ganmani isolated from IL-10-1-C57BU6 mice, Lane 7: H. ganmani isolated from wild type C57BU6 mice. Figure 4.7: The results of the Helicobacter genus-specific PCR­ DGGE limit of detection experiment. Lane 1: Marker lane containing PCR product from H. muridarum, H. hepaticus and H. bilis, Lane 2: unspiked stool sample revealing H. ganmani, Lanes 3 to 7: stool aliquots spiked with 10 fold dilutions of H. hepaticus representing 106, lOS, 1()4, 1(}1 and 1()2 H. hepaticus per gram of faeces respectively. 1 2 3 4 5 6 7

H. muridarum H. ganmani IL -1 o-t- H. hepaticus

H. bilis Figure 4.8: The results of the Helicobacter genus-specific PCR-DGGE for 8 of the 13 C57BU6 mice. Lanes 1 to 8: PCR product from 8 of the cagemale , Lane 9: Marker: H. muridarum and H. trogontum, Lane 10: Marker: H. rodentium, H. hepaticus and H. bilis.

H. rodentium H. muridarum

H. hepaticus

H. bilis H. trogontum

Figure 4.9: The Helicobacter genus-specific PCR-DGGE result for BABS Animal Facility mice: Lane 1 and 9: Marker containing H. rodentium, H. hepaticus and H. bilis, Lane 2 and 8: Marker containing H. ganmani of IL-lQ-1- mice, H. muridarum and H. trogontum, Lane 3: Cage A mouse, Lane 4: Cage B mouse, Lane 5: Cage C mouse: Lane 6: CageD mouse, Lane 7: Cage E mouse.

H. ganmari IL - 1o- ,_ H. rodentium H. muridarum H. hepaticus Figure 4.10: The results of sequencing for PCR product from reference and laboratory strains of Helicobacter species and DGGE gel bands Laboratory or Band Sequence Reference strain 743 763 781 809 849 883 969 1009 1021 1031 1041 aGCTGGAACATT-GATGCGCGA-G-AAT-CC TGCTT--GTCAGGGCAG--T---AAGATACGCG-AATCCGCTAGAGATAGTGGAGTGCTAGCTTGCTAGAGC H. ganmanib A2,D1,E1 . A. .A TGGA . .TCT. •.••••• A ...... •.•.•...•..•••...... •. A. H. ganmanic B1 . A. .A TGGA. .TCT. . .T •.•• A ••••.•••..••.•.••••.••..••.••.•••.••.• A • A1d . .A. . .. A .. . T TGGA. .TCT ...... TA ...... TG ...... CC-T .. TAGG ... . H. rodentium . . . A .GAG . .CTT . ...•••• A ...••..••••••.••..••.•....•.••.••••••• A. H. muridarum A. c. . A. .T. .A ...... TA ...... TG...... C. CT .CTG.G .. . A3e . . A. . G . .T . . . .C ..•.• A ....•...... •.•...... •..• A. H. hepaticus C1,F1,H1 . . . . A. . G . .T. . . c ..•. TA ...... C-CT.C.G-G ... . H. bilis D2, E2 . . c . .G ••.••.•.•..••.••••.•..•..••...•.. G .••.•. C •.••• H. trogontum . c. . ... T ...... C ...... C-CT.C.G-G ... .

aconsensus sequence bH. ganmani isolated from wild type C57BL/6 mice. cH. ganmani isolated from IL-10·'· C57BL/6 mice. dsequence identical to HSU96299 (316) csequence with 98% homology to HSU51874 (81) 4.4 DISCUSSION

The PCR-DGGE assay developed for detecting and determining the species of helicobacters that are present in murine faecal samples has been shown to be extremely effective. Band positions for reference strains of H. trogontum, H. bilis, H. hepaticus, H. muridarum and a laboratory strain of H. ganmani from IL-10_,_ C57BL/6 mice were shown to be specific. However, H. ganmani isolated from wild-type C57BL/6 mice resident in our animal facility had an identical band position to the reference strain of H. rodentium. Sequencing of the PCR products obtained from H. rodentium and the strain of H. ganmani from wild-type C57BL/6 mice revealed that they cannot be distinguished by PCR-DGGE, due to the very high degree of sequence homology at the non-GC clamped end of the resulting PCR product (Figure 4.1 0) (278). This lack of species resolution is a problem inherent to PCR-DGGE, and has also been evident with both genus and group-specific assays for bifidobacteria and lactobacilli (151, 291, 368). Therefore if this technique is to be used to define the Helicobacter spp. status of mice from a new supplier then a selection of bands should be sequenced.

A further finding of this study is that the two laboratory strains of H. ganmani, the first isolated from IL-10·'- C57BL/6 mice and the second from wild-type C57BL/6 mice, have different band positions. This observation relates to the fact that the method is sensitive to small differences in sequence within the domain with the lowest melting temperature, this being the domain that determines the bands position on the gel (95). Sequencing of the PCR products amplified from H. ganmani strains of wild-type and IL-10 _,_ mice revealed that only a one base difference existed between the two strains, however despite this, a large difference in gel band position was observed (Figure 4.6). The fact that different species within the Helicobacter genus give identical band positions while different strains of the same species produce different band positions highlights the caveats that must be used in interpreting PCR-DGGE data. Certainly, differences in migration distance (or band separation) cannot be used to infer the relatedness of organisms by genus-specific PCR-DGGE.

- 138- The limit of detection of 105 H. hepaticus per gram of faeces is comparable to the published sensitivities of PCR and culture for the detection of bacterial colonic pathogens from faecal samples (80, 187). For example, in a study of human faeces seeded with known quantities of Campylobacter species, spiking with 105 colony forming units (cfu) per gram of faeces was required for their detection by culture or PCR (187). In addition the assay compares favourably with the other methods for speciating murine lower bowel helicobacters. The sensitivity of PCR-DGGE for Helicobacter species in murine faeces was similar to species-specific PCRs for H. ganmani and H. bilis, with both organisms being detected in greater than 90% of mice. Given that these assays are affected by similar variables, such as the ratio of target to non-target DNA, and the presence of PCR inhibitors in faeces, this is perhaps not surprising. The finding that the limit of detection for H. hepaticus was 105 bacteria per gram of murine faeces would suggest that a reasonably high density of organisms must be present in colon for their detection by such PCR based methods.

Nevertheless the molecular approach is far easier and more sensitive than culture. Contamination of plates with fast growing non-target species is a recognised difficulty of the filter method on non-selective media for slow growing bacteria (10, 84, 106, 107, 190, 278, 314). In particular, the comparatively slow growth of Helicobacter species (84, 106, 278) hampers the isolation of these bacteria when there is any contamination on the plate (131, 333). In this study the sensitivity of isolation in pure culture for H. bilis and H. ganmani was 23%. Using phase contrast microscopy for the detection of bacteria with a morphology consistent with Helicobacter species in samples taken from plate culture substantially increased the rate of detection of helicobacters. However, there are several obvious problems with this approach. Firstly, several Helicobacter species have similar morphologies making their preliminary identification by phase­ contrast microscopy problematic. For example, H. bilis and H. trogontum are motile tapered rods of similar dimensions while H. rodentium and H. ganmani are both short motile S-shaped bacteria (106, 220, 278, 316). Secondly, concurrent colonisation with

-139- multiple lower bowel Helicobacter species is commonly reported in the literature and was frequently observed in this study (Figure 4.9) (319). Although considerably less sensitive than PCR, culture has the distinct advantage of facilitating the phenotypic characterisation of Helicobacter species.

These observations concerning PCR and culture for Helicobacter species parallel the literature regarding the detection of Campylobacter species in faecal samples (162, 182, 206). Comparisons of culture and PCR based methods for the detection of Campylobacter species in faecal samples suggest that PCR based approaches are more sensitive than selective culture (162, 182, 206). For example in the study of Maher et al (2003) campylobacters were detected in human diarrhoeal stools from only 6% of cases by culture while a PCR-probe assay detected Campylobacter spp. DNA in 38% of cases (206). The interpretation of this PCR data is however complicated by the frequent presence of Campylobacter species in the oral microbiota (84) and the very sensitive nature of methods based on probe hybridisation to PCR product. Direct single-step PCR has been shown to be 5-10% more sensitive than culture based techniques for the detection of Campylobacter species (162, 182). A further relevant fmding from these studies is that filter-based culture methods for Campylobacter species have been reported to be less sensitive than culture on selective media (84, 182, 187). However the filter method allows the detection of Campylobacter species that are sensitive to cefoperazone, an antibiotic supplement that is present in many selective media (84). In this regard culture of Campylobacter species is analogous to the situation with H. ganmani that will not grow on selective media.

A distinct advantage of PCR-DGGE is that Helicobacter species that were nof previously known to be present in a given murine facility, and species that are difficult to cultivate, may be detected. The presence of two bands for the mouse from Cage A in gel positions that are distinct from the species used in the marker, suggests that novel Helicobacter species are present in this mouse. When multiple bands are seen in genus­ specific PCR-DGGE, recombination of PCR products from 2 or more species should

-140- always be considered (291). Heteroduplexes representing annealed single stranded PCR product from two different species can often be present in the upper part of PCR-DGGE gels. This occurs due to their tendency to denature under low denaturant conditions (291). However the two new band positions observed for the mouse from Cage A were within the region of the gel where PCR product from known Helicobacter species denatured. Furthermore, comparison of the sequences generated from these bands with other Helicobacter species present in the mouse colony suggested that these were not chimeras (Figure 4.10). The finding of a novel sequence (band A3) suggesting that an unnamed species is present in the colony is not surprising as helicobacter colonisation is common in laboratory mice (107), and to date many members of this genus have not been named (70). Supporting this observation is a survey of more than 1000 laboratory mice in the USA that revealed that 10.5% of mice that were positive for Helicobacter species by PCR were negative with primer sets for H. typhlonius, H. hepaticus, H. bilis and H. rodentium (107).

In summary, a novel PCR-DGGE assay for directly detecting and speciating l

pathology (11, 48, 104, 107, 108, 18~, 194, 320, 372). This assay provides a sensitive approach for research and commercial murine colonies to examine the helicobacter colonisation status of individual mice or sentinels representing cages of mice. The fmding that murine faecal samples can be stored at room temperature for up to a week without affecting the outcome ofPCR for Helicobacter species (16) suggests that sample transport is unlikely to affect the results of PCR assays for centralised testing. Furthermore, the ease with which helicobacters are transmitted to sentinels (197, 374)

-141- suggests that screening one or two mice from each cage would provide a useful description of the helicobacter populations within a given facility. The broad nature of the helicobacter sequences used to design the primer sets also suggests that the assay may be of value in defining the lower bowel helicobacters of other animal species.

-142- CHAPTER 5: Absence of mucosa-associated colonic helicobacters in an Australian urban population

5.1 INTRODUCTION

The scientific evidence implicating lumenal bacteria in the pathogenesis of inflammatory bowel diseases is compelling (94). Whether specific bacterial populations are important in initiating or perpetuating inflammation, or the mere presence of a luminal flora is sufficient, is however unknown. If specific bacterial populations contribute to the pathogenesis of Crohn's disease or ulcerative colitis then this may relate to bacterial metabolic, antigenic or virulence characteristics. Specific virulence factors that might be involved would include factors such as motility and epithelial contact, adherence or invasion.

In view of these possibilities the research effort directed at the role of the microbiota in ulcerative colitis and Crohn's disease has had a broad focus ranging from identifying a single pathogenic organism to defining the characteristics of a disease promoting bacterial microflora (31, 32, 59, 65, 67, 86, 97, 134, 150, 193,215,252, 284, 310, 338, 349, 355). In particular, the bacterial populations adjacent to the colonic mucus layer have attracted attention in the study of IBD. This relates to the fact that their proximity to the epithelium may allow greater interactions between these organisms and the host (303, 340). However, as alluded to in Section 1.9 of Chapter 1, at this stage there is little direct evidence for a specific bacterial species that durably colonises the mucus layer of the human colon (227, 303, 340).

Among the potential candidate species that might colonise the colonic mucus of humans helicobacters feature prominently. Helicobacter species colonise the colonic crypts and the adjacent mucus layer of a broad range of animals and birds (72, 106, 115, 220, 329, 330). The number of named species within this genus has expanded rapidly over the past 2 decades, such that there are currently 23 validly published Helicobacter species (71). There are also many other new species represented by a small number of isolates that do

-143- not yet meet the recommended criteria for official naming (70, 71). Increased detection of these Helicobacter species in lower bowel samples of a range of animals has been related to the development of improved cultural techniques that were originally developed for the isolation of Campylobacter species. These include the use of a filter technique and an appreciation of the unusual macroscopic appearance of Helicobacter species on culture plates, their slower growth rate and fastidious nature (84, 333).

As discussed in Chapter 4, in immunodeficient strains of mice (such as IL-10·'· and SCID mice) it appears likely that helicobacter colonisation can accentuate the development of colitis (102, 109, 183, 320). Helicobacter species have a number of characteristics associated with pathogenicity that could mediate colitis induction. The presence of flagella and periplasmic fibrils results in motility at the interface with mucus layers, representing an adaptation to this environment (92). As a result of this motility Helicobacter species are often seen in large numbers in the colonic crypts and surface mucus of rodents, and are sometimes found within the colonic epithelial cells (190, 257, 300). H. hepaticus and H. bilis have been shown to produce Cytolethal Distending Toxin (CDT), and several other rodent helicobacters contain the genes encoding it's production (176, 384). In addition, the sequencing ofthe complete genome of H. hepaticus has revealed a candidate adhesin and a genomic island with homology to structural components of Type IV secretion systems (336).

Helicobacter species have also been isolated from 2 primate models of IBD - Saguinus oedipus (the Cotton-Top Tamarin or CTT) andMacaca mulatta (the rhesus monkey). The New World primate Saguinus oedipus is prone to the development of chronic pancolitis and carcinoma when kept in captivity (22, 166, 293, 381). In a recent study looking for lower bowel Helicobacter species in CTTs, Saunders et al (1999) cultured CTT faeces using a filtration method and helicobacter-selective media (293). Eight helicobacter isolates were cultured from 34 Cotton-Top Tamarins, some of whom had colitis (293). Complete 16S rRNA sequences were obtained from 2 of the isolates and these were shown to be most closely related to Helicobacter fennelliae. Using PCR, DNA from the Helicobacter genus was detected in colonic biopsies from 18 of the 34 animals studied.

-144- The authors did not report whether there was an association between the detection of helicobacter DNA and the presence or severity of disease (293).

Colitis development in the Cotton-Top Tamarin appears to be strongly influenced by environmental factors (166, 381). Interestingly, captive CTTs have a much higher prevalence and severity of colitis than their wild counterparts (381). For example, in a histologic study of colonic biopsies from 88 wild CTTs in Columbia and 67 captive CTTs in the USA, Wood et al found severe colitis in 64% and 0% of the CTTs and moderate colitis in 19% and 13% of the CTTs respectively. A broad range of environmental factors that may promote disease development in captive animals have been suggested, however researchers have found it difficult to conduct studies where only one environmental variable is examined. To date, the factors thought to induce disease include stress related to captivity, ambient temperatures less than 32°C, lack of exposure to sunlight and gastrointestinal pathogens such as Campylobacter species (166, 381). This latter factor was highlighted in a study by Johnson et al (1996) of 212 CTT twins, where one of each pair of newborn twins was separated from their parents within 12 hours of birth and hand-reared in a separate environmental unit, while the other sibling remained with their parents in the main colony (166). This study found that the cumulative incidence of acute colitis was lower in unit-raised than in colony-raised monkeys. Of significance, unit­ raised monkeys were less likely to be exposed to Campylobacter jejuni and Campylobacter coli than colony raised animals and they were also more likely to have been treated for diarrhoea with the anti-inflammatory agent sulfasalazine. In addition, the unit-raised CTTs with campylobacter infection were twice as likely to develop colitis as uninfected CTTs. In this context, the reported colonisation of these primates with Helicobacter species may also be of aetiologic significance, but so far this research question has not been addressed (293). The fact that the prevalence of colitis is different in different animal research centres also potentially supports the role of a transmissible agent in disease aetiology (166).

Helicobacter species related to H. fennelliae have also been isolated from rhesus monkeys (Macaca mulatta) that are prone to colitis development in captivity (103). In a

-145- study of 8 monkeys with colitis by Fox et al (2001), Helicobacter species closely related to H.fennelliae were isolated from each monkey (103). Silver staining of colonic samples taken from these monkeys revealed dense colonisation of the surface and crypt mucus by Helicobacter species. However studies to directly examine whether there was an association between colonisation and disease development or severity, in these rhesus monkeys were not performed. In an earlier study by Flores et al (1990) the virulence of closely related Helicobacter species in infant monkeys was examined (96). Experimental oral infection of infant macaque monkeys with H. cinaedi or H. fennelliae caused an acute diarrhoeal illness and bacteraemia (96). Each species was detectable in the faeces of infected monkeys for at least 4 weeks after experimental infection. Whether long-term colonisation occurred was not assessed however.

These observations in animal models have led to the hypothesis that human lower bowel commensal Helicobacter species might precipitate colitis in genetically predisposed individuals. In support of this hypothesis, four investigators have reported the detection of helicobacter DNA by polymerase chain reaction (PCR) amplification in humans with normal colonic mucosa, diverticulosis and inflammatory bowel disease (27, 196,249, 349). However two of these studies have significant methodological flaws.

In the first of these studies, Tiveljung et al (1999) extracted DNA from surgical and endoscopic biopsies of the ileum of 11 Crohn' s disease patients and 11 controls (349). This DNA was amplified with universal16S rDNA primers and the resulting PCR product hybridised with radiolabelled probes. The helicobacter probe hybridised to 8 of 11 Crohn' s biopsies, however hybridisation was also reported in 4 of 11 controls. In addition, using the same method other potential pathogens previously suggested to be important in Crohn' s disease such as Mycobacterium paratuberculosis and Listeria monocytogenes were detected in a significant number of Crohn' s cases (5 and 8 respectively) and some controls (1 and 4 respectively). These observations raise questions regarding the stringency of the method (which probed PCR product and not DNA or RNA directly obtained from the sample) and complicate the interpretation of the positive result for Helicobacter species. In fact, the probe used in their study appears to be

-146- specific for H. pylori and should not detect other Helicobacter species under stringent conditions. That is, comparison of their helicobacter probe sequence with helicobacter sequences in the RDP (RDP 2.28.5.1) reveals that the probe has 2 mismatches with H. pullorum, H. bilis and H. cinaedi and 3 bases mismatched with H. heilmanni, H. ganmani and H. fennelliae. Therefore it is likely that either H. pylori DNA was detected in these ileal samples or that non-stringent hybridisation occurred.

A second study by Linskens et al (2000) reported the detection of helicobacters by PCR in sigmoid biopsies from 9 of 18 controls (50%), 6 of9 ulcerative colitis cases (66%), 6 of 10 Crohn's cases (60%) and4 of7 diverticulosis cases (57%) (196). Subsequently all of these samples were also positive with a Helicobacter bilis-specific PCR. To date this work has only been published in abstract form. It would appear however that the Helicobacter genus PCR ofLinskens et al was not specific for Helicobacter species (J.G. Kusters, personal communication). Therefore the results of this study are most likely to represent non-specific amplification of 16S rDNA from other bacterial species. This non­ specific amplification may relate to the use of a universal primer (for all bacteria) in combination with a genus-specific primer rather than two genus-specific primers.

In the third study, Bohr et al (2002) performed a Helicobacter genus-specific PCR on DNA template from pooled endoscopic ileal and colonic biopsies of 10 patients with known gastric H. pylori infection status (27). Three of these patients had gastric H. pylori infection and helicobacters were detected by PCR in all3. However a fourth individual with Crohn's disease had a positive PCR result despite being free of gastric H. pylori infection. Sequencing of 764 bases of this PCR product showed a high level of similarity to the 16S rDNA of H. pullorum. Stool cultures for bacterial pathogens were negative in this individual, however specialised culture for Helicobacter species was not undertaken. Biopsies from the remaining 6 individuals without gastric H. pylori infection were negative by PCR.

In the fourth study, Oliveira et al (2004) performed nested PCR and culture for Helicobacter species on colonic biopsies from 42 ulcerative colitis cases and 74 controls

-147- (249). Helicobacter DNA was detected in 8 of the ulcerative colitis cases and 7 of the controls and a bacterium with the morphology and 16S rRNA sequence of H. pylori was cultured from 3 ulcerative colitis cases and 1 control. There was no significant difference in the frequency of detection of H. pylori in the cases and controls. All of the subjects with positive culture or PCR fmdings had serological evidence of gastric H. pylori infection suggesting that the positive results were due to wash down of gastric organisms during bowel preparation.

To put these PCR findings in context, several animal intestinal helicobacters can cause a transient zoonotic campylobacter-like illness in humans (for example H. cinaedi and H. fennelliae), but this is not common (266, 267, 350). A number of other animal Helicobacter species are rare causes of acute diarrhoeal illness and bacteraemia in normal and immunocompromised humans (246). These include Helicobacter pullorum, "Helicobacter rappini" and Helicobacter canis. Whether long-term colonisation of the colon occurs after acute infection has not been systematically examined but clinical recovery is the most common outcome of infection.

A number of studies also suggest that helicobacter DNA is present in the human liver or biliary tree in association with disease (11, 12, 99, 101, 104, 194, 240}. The hypothesis that Helicobacter species colonise the human liver or biliary tree and promote diseases such as Primary Biliary Cirrhosis (PBC) and Primary Sclerosing Cholangitis (PSC) is based on observations of mice. In specific murine strains, enteric helicobacters can translocate to the liver and contribute to the development of hepatitis, adenomas and carcinomas (11, 101, 104, 194). The suggestion that Helicobacter species are present in the human hepatobiliary system is founded solely on PCR-based studies. There have not been any published reports of the culture or ultrastructural demonstration of helicobacters other than H. pylori in human hepatobiliary samples (246). In addition, the PCR-based studies that report the detection of helicobacter DNA in human hepatobiliary samples have conflicting results. Helicobacter DNA has been detected in liver or biliary samples from patients with PBC, PSC, cirrhosis, cholelithiasis, liver tumors and controls (12, 46, 99, 112, 239, 240). However, other research directly contradicts these findings reporting

-148- that Helicobacter species were not detected by PCR in hepatobiliary samples from other populations with similar conditions (60, 87,282, 283). When PCR products were sequenced in the positive studies the amplicons were most often H. pylori -like (12, 46, 239, 282). There are several possible explanations for these conflicting results. Firstly translocation from the stomach to the liver, of H. pylori or phagocytic cells containing H. pylori DNA, could explain the results (24). Secondly the positive results could be due to PCR contamination.

In view of this literature, a carefully controlled nested PCR study was undertaken to determine if Helicobacter species were colonising the colon of Australian subjects undergoing colonoscopy. As helicobacters are likely to contribute to colitis development in mice, Crohn' s disease and ulcerative colitis cases were enrolled as well as a group of control subjects, to determine if lower bowel colonisation was associated with disease. The opportunity also arose to examine a small number of colonic biopsies from Malaysian subjects, thus allowing for the inclusion of subjects from a developing country in the study.

-149- 5.2 METHODS

5.2.1 Australian study

5.2.1.1 Subjects

Seventy patients undergoing colonoscopy (N=64) or sigmoidoscopy (N=6) at the Inner West Endoscopy Centre in Sydney participated in the study. The age, sex, ethnicity, indication for procedure and endoscopic findings were recorded for each subject. Fifteen subjects met standard diagnostic criteria for Crohn's disease (7 Males, 8 Females; mean age 40), 12 for ulcerative colitis (6 Males, 6 Females; mean age 49) and 43 subjects who did not have inflammatory bowel disease formed the control group (23 Males, 20 Females; mean age 61) (44, 210). Prior to the procedures, subjects were administered commercially available bowel preparation. The presence of gastric Helicobacter pylori infection was determined by review of the clinical records. If testing had not been performed patients were classified as having "unknown H. pylori status". Subjects who had been tested and shown to be positive (by urease test, phase-contrast microscopy, culture for H. pylori or histology) were considered positive, otherwise they were recorded as negative.

5.2.1.2 Sample collection

At colonoscopy biopsies were collected from the ileum, caecum (in duplicate), transverse colon, descending colon and rectum into separate sterile 1.5 rnl tubes. At sigmoidoscopy biopsies were collected from the sigmoid (in duplicate) and rectum. If luminal fluid was present, 5 to 10 rnl was aspirated through the endoscope into a disposable polyp trap.

- 150- 5.2.1.3 Controls for PCR

As human faecal samples are known to contain inhibitors of PCR, two positive controls were set up for each subject. To prepare the positive controls for each subject, one of the duplicate biopsies and a 100 jll aliquot of lumenal fluid were spiked with approximately 102 and 103 Helicobacter muridarum (ATCC 492820) respectively prior to DNA extraction. The stock solution of H. muridarum in normal saline was prepared by harvesting freshly cultured bacteria from Horse Blood Agar plates (190) and the mean number of organisms per ml was estimated, in triplicate, using a haemocytometer. In addition, two negative controls were included in each PCR reaction, the first with no template DNA and the second containing solution from the negative control for DNA extraction. The negative controls from the first round PCR were also diluted and amplified in the second round of PCR.

5.2.1.4 Nested PCR

To determine if 16S ribosomal DNA derived from the genus Helicobacter was present in any of the samples a nested PCR was performed. DNA extraction, PCR amplification and sequencing were performed according to the protocols detailed in Section 2.1.2 and 2.1.3, Chapter 2. In brief, DNA was extracted with the Puregene kit (Gentra systems) and initially amplified with primers F27 (378) and R1494 (237) (Tables 2.1 and 2.3, Chapter 2). A 1:25 dilution of the first round PCR product was amplified in the second round with the Helicobacter-genus specific primers H276f and H676r (275) (Tables 2.1 and 2.3, Chapter 2). The PCR product from unspiked positive samples was sequenced and compared with GenBank (19). Their homology to the corresponding 16S rDNA sequences of H. pylori (U00679), H. cinaedi (AF348745), H. fennelliae (AF348746), Helicobacter heilmanni (AF058768), H. bilis (AF047843) and H. hepaticus (AF302103) was calculated.

- 151- 5.2.1.5 Statistical analysis

For the purposes of analysis the ulcerative colitis and Crohn' s disease groups were pooled as an IBD group. In addition, subjects with helicobacter DNA in any unspiked colonic or luminal fluid sample were considered positive. The proportion of positive subjects in the IBD and control groups was compared using the Chi-squared test.

5.2.1.6 Culture

Culture using a number of techniques was not successful and not pursued further.

5.2.2 Malaysian study

A colonic biopsy was taken from a randomly selected colonic site in 20 adult Malaysian subjects (9 Males, 11 Females; mean age 48) between March 1999 and February 2000. The colonoscopic findings were normal (15), ulcerative colitis (1), rectal cancer (2), and diverticulosis (2). The ethnicity of the subjects was Indian (12), Malaysian (7) and Chinese (1). DNA was extracted in Malaysia using the Phenol-Chloroform method (49) and transported by airfreight to Australia on dry ice. These DNA samples were diluted 1:10 and nested PCR amplification and sequencing performed as above. No spiked controls were used in this study.

-152- 5.3 RESULTS

5.3.1 Australian study

5.3.1.1 Nested PCR

A total of 360 biopsies (mean 5.1 per subject, range 0 to 6) and 56 luminal fluid aspirates were collected from the 70 subjects. The number of subjects belonging to each ethnic group was as follows: Australian 25, Southern European 30, South East Asian 4, and "Other" 11. The common indications for colonoscopy included previous colonic polyps (27%), change in bowel habit (24%), rectal bleeding (19%), and a family history of colonic carcinoma (11% ). Four subjects were known to be H. pylori infected at the time of examination, 41 were negative and 25 were of unknown infection status.

Helicobacter spp. DNA was detected in 9 samples obtained from 6 of the 70 subjects - 2 controls, 2 ulcerative colitis and 2 Crohn's cases (Table 5.1). An example of a gel showing the positive result for one patient is shown in Figure 5.1 (a faint band is present in Lane 5). For 3 of the positive subjects two samples were positive (2 biopsies for 2 cases and a biopsy and luminal fluid aspirate in the other) and in the remaining 3 only one sample was positive (a biopsy in 2 cases and a luminal fluid aspirate in the other). Three hundred and thirty-eight samples were negative. In 68 of 70 subjects the biopsy and lumenal fluid aliquot spiked with H. muridarum were positive. In the other 2 subjects the spiked biopsy sample was negative but in both of these cases the spiked lumenal fluid sample was positive. All of the negative control samples were negative.

5.3.1.2 Sequencing

In all 9 positive samples sequencing of the PCR product revealed an identical 356 base pair DNA sequence that completely matched the 16S rDNA of H. pylori. Using a BLAST search (4) the sequence obtained from the PCR products differed from other helicobacter sequences by at least 2%, with differences to the sequences of H. cinaedi, H. fennelliae,

-153- H. heilmannii, H. bilis and H. hepaticus of 2.5%, 2.5%, 3.6%,3.9% and 4.2% respectively.

Three of the subjects positive for H. pylori DNA had known gastric H. pylori infection, 2 were of unknown status and 1 was negative. The negative individual had previously underwent Bilroth II partial gastrectomy and had eradication therapy in 1995 for documented H. pylori infection. This patient had negative histology, phase-contrast microscopy and culture at gastroscopy on the day of colonoscopy. Compared with the "gold standard" of the combined methods for detecting gastric H. pylori infection, nested Helicobacter genus PCR had a sensitivity of75% (3 of 4) and a specificity of 97% (40 of 41) for gastric infection.

5.3.1.3 Statistical analysis

There was no significant difference in the proportion of IBD cases who had detectable helicobacter DNA in their colonic samples compared with controls (.XZ p = 0.32).

5.3.2 Malaysian study

Positive results were obtained with biopsies from 4 subjects (including the subject with ulcerative colitis). Sequencing was successful for 3 of these which had 95%, 98% and 99% homology to the 16S rRNA of H. pylori. The sequence lengths were however only 291, 296 and 319 bases respectively. Due to the small number of samples statistical analysis was not performed.

-154- TABLE 5.1: Results of nested PCR for Helicobacter species DNA and sequencing for the Australian subjects.

Group Number Mean Subjects with H. pylori positive samples Subjects with all samples negative of age subjects (years) Number Gastric H. p_)!_lori infection status Number Gastric H. pylori infection status Positive Unknown Negative Positive Unlmown Negative Controls 43 61 2 1 0 1 41 1 10 30

Ulcerative 12 49 2 1 1 0 10 0 6 4 colitis Crohn's 15 40 2 1 1 0 13 0 7 6 disease Figure 5.1: PCR for Helicobacter species 16S rDNA. Lane 116S PCR: H. muridarum, Lane 2 16S PCR: colonic biopsy, Lane 3 negative control 16S PCR, Lane 4 nested Helicobacter genus PCR: H. muridarum spiked colonic biopsy, Lane 5 nested Helicobacter genus PCR: positive colonic biopsy (not spiked), Lane 6 Helicobacter genus PCR with H. muridarum, Lane 7 nested negative control. 5.4 DISCUSSION

Only 8.6% of the Australian subjects and 20% of the small study of Malaysian subjects reported here had detectable helicobacter DNA in their colonic samples. This compares with previous studies applying PCR methods to the detection helicobacter DNA in human ileal and colonic biopsies that have reported positive results in 19 to 73% of subjects (27, 196, 249, 349). Although several of these studies have methodological flaws, the results still raise the question as to why the prevalence of detectable helicobacter DNA in this study is comparatively low.

It is well known that samples of faecal material may contain substances that inhibit the PCR reaction (225, 389) and there is some evidence suggesting that the degree of inhibition varies in samples obtained from different individuals (219). The use of a spiked sample from each Australian patient in this study confirmed that 102 helicobacters in a biopsy and 104 per m1 of luminal fluid (that is 103 in each 100 !J.l aliquot) would have been detected had they been present in 68 of the 70 individuals. This spiked control accounted for the effect of inhibitors in the samples of each individual subject, and facilitates the conclusion that Helicobacter species were not present in the colon of these subjects in significant numbers and thus are unlikely to play a significant role in the perpetuation of colitis in IBD cases.

It could be argued that the low proportion of positive subjects in the current study may relate to the removal of the mucus layer by bowel preparation. However previous studies have shown that a layer of mucus remains on the surface of the colon of patients who have taken bowel preparation similar to that used in these subjects, so it is unlikely that this layer was completely removed in any of the subjects or can explain the largely negative results (265, 340). In addition, the subjects in other studies reporting the detection of Helicobacter species in ileal or colonic tissue samples were also prepared with bowel preparation (27, 196, 249, 349).

-157- The helicobacter DNA that was detected in samples from 6 Australian subjects is most likely to have originated from gastric H. pylori infection. This is in agreement with other studies that have also reported the detection of H. pylori DNA of gastric origin in faecal or colonic samples using a PCR based methodology (27, 168, 249). In fact, in a similar study by Oliveira et al (2004) H. pylori was cultured from colonic biopsies taken from 4 of 60 subjects (7%) with gastric H. pylori infection (249). All of the Australian sequences were identical to H. pylori and at least 2% different to those of other Helicobacter spp. isolated from humans or rodents. The 350 base pairs of sequence obtained is inadequate to state that H. pylori was the source of the DNA with complete certainty (362), however this remains highly probable.

The interpretation of the results for the 4 positive Malaysian subjects is limited by the small number of subjects and the short DNA sequences obtained. However the sequences were only 95 to 99% homologous to the 16S rRNA of H. pylori suggesting that other lower bowel Helicobacter species might be present in individuals from this developing nation. Further research with a larger number of subjects and appropriate positive and negative controls is required to clarify this issue.

On balance, the most likely explanation for the relatively small proportion of subjects in the Australian study with detectable helicobacter DNA in colonic samples compared with other studies is the fact that many of the study population had previously been treated with eradication therapy for gastric H. pylori infection. Forty-one of the 70 subjects were known to be negative while only 4 were known to be positive. Similarly the higher proportion of subjects with detectable H. pylori DNA in the Malaysian population may reflect the higher prevalence of gastric H. pylori infection in these subjects although as noted above the sequencing results are not strongly supportive of this interpretation (127).

Of note, one Australian who was not infected by H. pylori at the time of colonoscopy returned a positive PCR for H. pylori DNA. There are two possible explanations for this finding. Firstly, gastric H. pylori infection may not have been detected. This individual had previously had a Bilroth IT gastrectomy and it is not clear how sensitive histology,

-158- microscopy and culture are for H. pylori infection following this procedure (245, 317), however all3 screening tests were negative in this case. Alternatively, this positive result may represent the persistence of bacterial DNA within cleaned colonoscopes or on biopsy forceps as has previously been noted elsewhere (170, 281).

How can the overall results be integrated with previous studies reporting the detection of helicobacter DNA in human ileal or colonic samples with PCR? By utilising appropriate positive and negative controls, the presence of significant numbers of Helicobacter species within the colonic mucus of the Australian population studied, as defined by the primer set used, has been effectively excluded. Given the finding that in all 9 positive samples the DNA detected was likely to be derived from H. pylori, the previous reports of helicobacter DNA in colonic samples may also reflect the presence of DNA derived from gastric H. pylori infection (27, 249). Alternatively this might be due to differences in the subject populations studied, the primer sets and PCR conditions used or the presence of false positive results (196, 349).

False positive results due to laboratory contamination should be considered whenever a nested PCR methodology is used (254). This most commonly occurs because of contamination of DNA extraction solutions or PCR solutions by trace amounts of PCR product. In a study in which five laboratories performed PCR analyses on the same 48 specimens for human herpesvirus 8 (HHV-8) Pellett et al found that 3 of the laboratories displayed evidence of PCR contamination (254). In all cases this was associated with the use of a nested PCR assay. Two of the three laboratories with contamination problems had previously reported high prevalence rates of HHV-8 in semen, and thus their results may have contributed to the wide variance in the reported prevalence of this virus in the literature at the time (254).

A second problem with PCR studies targeting bacterial16S rDNAis that unknown to the investigators, primers may not be entirely specific for the target species or genus. For example, the primer set used in this study was previously thought to be specific for the helicobacter genus but has subsequently been shown to amplify certain species of

- 159- Brevundimonas (38). However in the current study this would have been identified had it occurred because the product sequence differs significantly from that of Helicobacter species.

PCR studies of potential pathogens in inflammatory bowel disease have often led to confusing and contradictory results, some of which may be related to the problems outlined above. To avoid misleading PCR based reports of pathogen detection in gastrointestinal disease, confirmation of positive results by another method such as culture or by visualisation with fluorescent in situ hybridisation is desirable. The results for Australian subjects support the view that Helicobacter species are not commensals of the colon or involved in the perpetuation of IBD in this population. A role for transient infection with lower bowel helicobacters in disease initiation however cannot be excluded definitively. In view of the absence of Helicobacter spp. DNA other than that from H .pylori, the previously developed Helicobacter genus PCR-DGGE was not applied in this study.

-160- CHAPTER 6: Colonisation resistance in a murine model

6.1 INTRODUCTION

The resistance of the colon to colonisation by newly introduced bacterial species is known as colonisation resistance, a concept frrst coined by Van der Waaij in 1971(358). This resistance to colonisation is a key barrier to attempts to modify the species composition of the human colon, with the aim of preventing or modifying the course of large intestinal diseases. Although at this time there is little direct evidence that changing the species composition of the human colonic contents will be of therapeutic benefit, interest in this area has increased with the publication of several trials that suggest that certain strains of bacteria, known as probiotics, when taken orally may have beneficial effects on health (122, 123, 172). Despite these reported beneficial effects most studies suggest that probiotic strains do not persist in the colon for periods lasting longer than 2 weeks after administration is stopped (reviewed below), that is, colonisation does not occur.

Previous studies of colonisation resistance have largely been performed in mice (26, 357, 358). These studies were conducted with Campylobacter jejuni and members of the Enterobacteriaceae because organisms such as Salmonella enterica and C. jejuni are the common intestinal bacterial pathogens (26, 357, 358). That is, the resistance to colonisation by these organisms was of direct practical significance in describing the susceptibility of individual hosts to gastrointestinal infections. However, recent interest in probiotic therapies has raised questions regarding the resistance of the gut to colonisation with other bacterial groups, as many probiotic therapies use strains of lactobacilli and bifidobacteria some of which are anaerobic (122, 124, 172, 199, 203, 250, 259, 286, 304, 305, 312, 345, 365).

One of the aims of probiotic therapies has been to produce durable colonisation of the human colon, however to date the literature suggests that such colonisation usually does not occur in adults. For example, Gionchetti et al (2000) studied

- 161- patients with inflammation in the small bowel pouch that is constructed following proctocolectomy for ulcerative colitis who had been treated with probiotics (123). In this study, a probiotic preparation named VSL#3, consisting of 4 strains of lactobacilli, 3 strains ofbifidobacteria and Streptococcus salivarius subsp. thermophilus was administered twice daily to patients with chronic pouchitis after remission had been induced with antibiotics. At 9 months, 17 of the 20 patients on the probiotic preparation remained in remission while all of the placebo group had relapsed. In this study the cultured faecal concentrations of total lactobacilli and bifidobacteria were shown to increase during treatment and to return to pre­ treatment levels one month after dosing was complete (123). Subsequently PCR assays for the Bifidobacterium and S. thermophilus species of VSL#3 were developed (36). Applying these assays to the faeces of subjects who had consumed the probiotic preparation showed that the Bifidobacterium and S. thermophilus species of VSL#3 did not persist for more than a week after dosing was ceased (36). Similar observations have been made in studies of orally administered L. acidophilus, L. casei, Lactobacillus delbrueckii and Lactobacillus rhamnosus (163, 195, 297, 345). All of these failed to colonise. For example, the five strains of lactobacilli studied by Jacobsen et al (1999) were not detectable in the faeces of any of 12 subjects a week after twice daily dosing with 1010 bacteria was ceased (163). Similarly a study by Tannock et al (2000) found that L. rhamnosus DR20 could not be detected in the faeces of.9 of 10 subjects after dosing ceased, despite a daily intake of 109 bacteria for 6 months (345). In the remaining subject, L. rhamnosus was detected at 2 months but not 3 months after dosing was completed. Whether orally administered bacteria may persist at specific sites in the gut and not be detectable with studies of faecal samples remains an open question. For example, Alander et al (1999) showed that L. rhamnosus GG was not detectable in faecal samples 2 weeks after completion of oral dosing, however low numbers of L. rhamnosus GG were detectable in colonic biopsies from 2 of the 7 subjects at that time (2). A number of studies do suggest that probiotic bacteria may persist for longer periods after their oral administration however in these studies the species present in the colon prior to the study were not clearly defmed (68, 165). For

-162- example, in a study by Johansson et al (1993) the lactobacillus populations of 13 subjects were assessed 1 and 11 days after dosing with a mixture of 19 Lactobacillus species in fermented oatmeal soup for 11 days. In this study, lactobacilli were cultured from faeces on Rogosa agar and 10 colonies from each sample were typed by biochemical testing, plasmid profiles and restriction endonuclease patterns (165). One day after dosing a broad variety of test strains were isolated from the faeces of 11 of the 13 subjects, while at eleven days post-dosing only L. plantarum was isolated from 11 of the 13 subjects. It therefore appears that L. plantarum originating in the oatmeal soup may have colonised the gut of the test subjects, however data on the Lactobacillus strains present in the gut of the subjects prior to the administration of the soup are not reported in detail (165). Similarly, De Champs et al (2003) reported that L. casei subsp. rhamnosus Lcr35 colonised the human gut after oral administration, however all of the subjects in this study had bacterial populations that hybridised with the test probe for L. casei prior to the commencement of dosing (68). Thus it would appear that the problem of producing long-term colonisation of the human colon with probiotic bacteria has not yet been solved.

If colonisation of the adult colon with new bacterial species is desirable, then a working hypothesis would be that this is most likely to occur when a bacterial species that has a significant competitive advantage in a given ecological environment, reaches its niche in a viable state and in significant numbers. An example of this process is the bacterial succession that is observed in the infant gut (detailed in Section 1.5, Chapter 1). It is likely that a significant competitive advantage for gut survival is possessed by bacterial species that have been isolated from gut samples. Indeed it has been demonstrated that the resistance of the colon to colonisation is substantially less for bacterial strains that are adapted to the gut than exogenous strains (358, 366). There is also evidence to suggest that gut-derived bacteria from a given animal species are highly adapted, and have a significant competitive advantage that is relatively specific for that particular animal species. For example, it has been shown that each animal species has its own strains of

-163- common gut-adapted bacterial species (69, 387). Furthermore, in the study of Johansson et al (1993) oral administration of Lactobacillus reuteri of human origin was shown to be more likely to persist in the human jejunum than a strain of L. reuteri of rat origin that had previously been shown to preferentially colonise the rat intestine (165). A natural extension of this hypothesis is that bacterial strains of human origin may be more likely to colonise the gut of humans than strains of animal or food origin.

Although many of the bacteria studied in trials of probiotic therapies have not been of human origin (263, 345), several ofthe strains tested were (163, 165). Thus the failure of probiotic strains to colonise the human colon in previous studies cannot only be attributed to the use of non-human isolates. To increase the probability of producing durable colonisation there are a number of potential approaches including: 1) the use of larger doses ofprobiotic bacteria for longer time periods, 2) direct administration of cultured bacteria into the caecum, 3) use of highly host adapted strains, or 4) application of methods to reduce colonisation resistance.

The studies that are described below evaluate a method for reducing the colonisation resistance of a murine model prior to probiotic administration. In addition to the non-specific barriers to colonisation by probiotic bacteria such as gastric acidity and peristalsis, it may be that the resistance of the colon to colonisation by a given bacterial strain is specifically related to the presence of bacterial species that are already occupying the environmental niche favoured by that strain. That is, the probability of colonisation by a given strain of probiotic lactobacilli may be improved by specifically eliminating the existing lactobacillus populations. Experimental evidence supporting this concept principally applies to aerobic gram­ negative bacteria. For example. in the study of Koopman et al (1978) the resistance of the gut to colonisation by Enterobacteriaceae was shown to be significantly less when Enterobacteriaceae were absent from the microbiota compared with animals that had these organisms (174). In this study, the oral dose of streptomycin-resistant E. coli that that was required to produce cultivable streptomycin -resistant E. coli

-164- populations 5 days after dosing was determined in 3 groups of mice. Two of the groups were free of E. coli at the commencement of the study and in these mice 102 to 103 E. coli were required to produce cultivable sr-E. coli. In contrast, 104 to 109 E. coli were required to produce cultivable sr-E. coli in the mice with a pre-existing E. coli containing microbiota. Similarly, as discussed in Section 1.10 of Chapter 1, mice treated with streptomycin and neomycin to eradicate facultative bacteria were shown to have a dramatically reduced colonisation resistance to mouse-derived strains of E. coli (358). However, other studies indirectly suggest that the anaerobic microbiota may have an inhibitory effect on Enterobacteriaceae, and this may partly mediate colonisation resistance to coliforms. In germfree mice monoassociated with E. coli a very high level of persistent colonisation of the colon occurs (109 per gram) (298). However, once a mixed microflora from other mice is introduced the E. coli populations fall dramatically (298). Although these studies suggest a role for anaerobes in colonisation resistance, it is clear that existing populations of Enterobacteriaceae mediate at least part of the colonisation resistance to coliforms.

Although it is theoretically desirable to reduce existing lactobacillus populations prior to administering probiotic strains, at present this is not technically feasible. How then can colonisation resistance in general be abrogated? Experimental evidence from humans and mice suggests that colonisation resistance for coliforms can be reduced with antibiotic treatment. A small and uncontrolled study of 15 human subjects by van den Bogaard et al(1986) found that an aminoglycoside resistant strain of orally administered E. coli persisted in faecal samples of patients treated with lavage alone for less than 4 weeks, but remained detectable in subjects treated with lavage plus 1 day of oral neomycin and metronidazole for 6 to 12 weeks (356). In mice, antibiotics have also been clearly demonstrated to reduce colonisation resistance to Enterobacteriaceae (357, 358). At least two mechanisms may mediate the efficacy of antibiotics in this regard. Firstly, there are the direct bacteriocidal and bacteriostatic effects that reduce competition for substrate and antibacterial factors of bacterial origin, such as bacteriocins. Secondly, treatment of mice with antibiotics increases the fluidity of the caecal contents and caecal volume

-165- at 24 hours (295), suggesting that bulk movement of caecal content may be part of their action in reducing colonisation resistance. In previous studies this increased fluidity of caecal contents was found to be more marked with penicillin than kanamycin treatment (295). In the same study penicillin was shown to have a dramatic effect on the anaerobic microbiota of the caecum, whereas kanamycin did not. This suggests that the inhibition of the anaerobic flora may be responsible for the increased fluidity of the caecal contents that was observed (295). That is, the antibacterial effect of antibiotic treatment may indirectly account for the increased fluidity of the caecal contents by the following mechanism. The anaerobic metabolism of carbohydrate and protein by caecal anaerobes is known to result in the production of short chain fatty acids, these have been shown to stimulate sodium (and therefore water) uptake in the colon by upregulating the expression of the apical membrane sodium hydrogen exchanger 3 (233). As sodium hydrogen exchangers are the dominant pathway of colonic sodium absorption, the increase in luminal fluidity observed after antibiotic treatment may be due to reduced butyrate formation, secondary to reduced anaerobe activity.

This proposed mechanism for part of the efficacy of antibiotics in reducing colonisation resistance leads on to a second potential method for reducing colonisation resistance. This method is quantitative reduction of the endogenous caecal flora with gut lavage. Indeed there is some evidence from a human study discussed above that this approach may be effective. In this study by van den Bogaard et al (1986), a combination of antibiotics and lavage prolonged the carriage of a marked species of E. coli compared with subjects treated with lavage alone. Unfortunately there was no group in this study in which antibiotics alone were given (356).

In mice, the effect of gut lavage on colonisation resistance has not been systematically evaluated. Despite some suggestion of efficacy, there is little published data on the use of cathartics in mice that specifically examines the degree of large bowel luminal cleansing that is achieved and the safety of their use (189). In

-166- one study of this issue by Lee et al (1986) the addition of magnesium sulphate gavage to antibiotic treatment was shown to reduce the number of caecal crypts containing spiral organisms (189). However in this study the effect on the luminal microbiota was not quantified. Potential cathartic regimes for use in mice include isotonic solutions such as a polyethylene glycol balanced salt solution and hypertonic solutions such as sodium phosphate solution, saturated magnesium sulphate and 20% sorbitol. Magnesium sulfate catharsis has previously been successfully used in rodents however some mortality was observed when this approach was applied in murine experiments (189). In contrast, the safety of short courses of treatment with sorbitol in drinking water can be inferred from previous dietary studies in mice however their efficacy is unknown (261).

Finally, surgical removal of the murine caecum is also likely to significantly reduce colonisation resistance (367). Clearly this is a drastic method of inducing susceptibility to probiotic bacterial colonisation, with very limited acceptability for human studies and thus this will not be discussed further.

The aim of this series of experiments was to develop a safe method of reducing the resistance of the mouse gastrointestinal tract to colonisation by anaerobes. To increase the probability of successfully colonising the gut, the anaerobic flora of the lower bowel of another mouse was used as the test bacterial species. Changes in the microbiota were assessed with PCR-DGGE for the domain Bacteria and the major bacterial groups described in Chapter 3.

-167- 6.2 METHODS

6.2.1 Experiment 1: Effectiveness of different cathartic regimens for reducing luminal content in the murine caecum and colon

6.2.1.1 Animals

Twenty-one 8 to 12 week old female BALB/C mice from the Animal Resources Centre (ARC, Canningvale, W A) housed in the BABS animal facility were used in this study. All mice were weighed prior to the experiment. In view of mortality experienced with magnesium sulfate catharsis in the past (A. Lee, personal communication) the five mice with the lowest weights were allocated to the control group. The remaining mice were randomly divided into four groups of 4 mice each.

6.2.1.2 Experimental regimen

Each group received the following treatments: Group 1: Polyethylene glycol solution (Glycoprep-C, Pharmatel, Hornsby, NSW) made up according to the manufacturer's instructions in the water bottle. Group 2: 20% sorbitol (Sorbilax, Pharmacia & Upjohn, Perth W A) in the water bottle. Group 3: Sodium phosphate solution (Fleet, B.B.Fleet, Braeside, VIC) given by orogastric gavage in four 0.1 ml doses. Access to water ad libitum. Group 4: Saturated magnesium sulfate (APS) given by orogastric gavage in four doses of 0.15 ml. Access to water ad libitum. Saturated magnesium sulfate was

made by placing 2 g ofMgS04 in 10 ml of deionised water and shaking for 10 seconds. Group 5: Controls. Access to water ad libitum.

The protocol for treatment administration was as follows: Day 1: At 17:00 remove feed. Mice remain fasting until sacrifice.

-168- Day 2: At 08:00 add cathartic containing water bottles made up to 100 ml exactly to the cages for Groups 1 and 2. Add water bottles containing 100 ml of water to the other groups (3,4 and 5). Cathartics were administered by gavage at: 09:00, 11:00, 14:00 and 17:00 hours for groups 3 and 4.

Day 3: At 08:00 all of the mice were sacrificed by C02 asphyxiation and cervical dislocation.

6.2.1.3 Outcome measures

All mice were weighed at sacrifice and the caecum and colon removed. The number of faecal pellets in the colon and the appearance of the caecum was also recorded. In addition the caecum and colon combined as well as the caecum alone were weighed. The volume of water, polyethylene glycol and sorbitol solutions remaining in each water bottle was recorded.

6.2.2 Experiment 2: Colonisation resistance

6.2.2.1 Animals

Twenty-four 8 week-old female BALB/C mice obtained from ARC at 6 weeks of age and housed in the BABS animal facility were randomly divided into filter-top cages of 4 mice and ear notched for individual identification. Twelve mice (3 cages) were assigned to each of the treatment and control groups.

6.2.2.2 Faecal suspension

The faecal suspension used to test colonisation resistance was obtained from the faecal samples of three individually housed 16 week-old C57BL/6 mice bred within the BABS animal facility. Faecal suspensions were produced by vortexing a single freshly collected faecal pellet for 2 minutes in sterile PBS immediately prior' to use.

-169- Each cage of treated BALB/C mice received faecal suspensions from the same C57BL/6 mouse for the duration of the study.

6.2.2.3 Experimental regimen

Prior to the commencement of the study all mice were ear notched and a baseline faecal sample was collected. This sample was collected by placing individual mice in a clean plastic cage until defaecation occurred. On Days 1 to 5, mice in both groups (control and treated) were gavaged twice daily with antibiotics. The dose of antibiotics used was vancomycin 0.5 mg (Eli Lilly, Indianapolis, USA), amoxycillin 0.57 mg (Sigma) and metronidazole 0.45 mg (Sigma) per 20 g mouse per day in a volume of 0.1 ml of sterile PBS. From 18:00 on Day 5 the mice were fasted (with access to water ad libitum) until the completion of magnesium sulfate cartharsis. This consisted of 4 doses of saturated magnesium sulfate (150 J..Ll each) at 18:00 on Day 5 and 11:00, 14:00 and 17:00 on Day 6. On Day 6 at 18:00 and twice daily for Days 7 to 11 the treated group of mice were gavaged with 150 J..Ll of faecal suspension. Control mice were gavaged with sterile PBS only. Access to food recommenced after the first treatment at 18:00 on Day 6. Faecal pellets were collected from all mice 4 weeks after the completion of the treatment protpcol. At 8

weeks all of the mice were sacrificed by C02 asphyxiation followed by cervical dislocation. The caecum and colon were harvested and fixed in 10% neutral buffered formalin. The protocol for treatment is summarised in Table 6.1 and illustrated in a flow-diagram in Figure 6.1.

6.2.2.4 Histology

5 J..Lm sections of the fixed tissues were cut and stained with haematoxylin and eosin. Blinded histologic assessment of the caecum and colon was performed using the scoring system of Berg et al (20).

-170- 6.2.2.5 PCR-DGGE

DNA was extracted from the faecal pellets as described in Section 2.1.2.2, Chapter 2. Each sample was subjected to PCR-DGGE with the domain Bacteria primer set using the conditions outlined in Tables 2.1 and 2.3, Chapter 2. To facilitate the analysis samples were grouped on the gels such that each gel represented the baseline and 4 week samples of a cage of treated and control mice for comparison, plus the faecal suspension used to gavage the treated group.

6.2.2.6 Gel analysis ,;

As the banding profiles present on separate gels are not easily comparable, all profiles to be compared had to be represented on a single gel. The Bray-Curtis similarity for each lane comparison was calculated using the PRIMER 5 software package (PRIMER-E, Plymouth, UK).

Two primary outcomes were examined. Firstly, the effect of treatment with faecal suspension on the microbiota of the treated mice was evaluated by comparing the baseline and 4 week samples of all treated and control mice with the faecal suspension depicted on the gel. If the faecal suspension altered the microbiota of the treated mice, then the similarity of the 4 week samples from the treated group to the respective faecal suspension would be greater than the other similarity comparisons.

Secondly, the effect of antibiotics and lavage per se on the faecal microbiota was assessed by comparing the similarity of the banding patterns from baseline and 4 week samples of all treated and control mice to the 4 week sample of a control · · mouse. For treated mice the 4 week sample of the first control mouse represented on the gel was used for this comparison while for control mice the 4 week sample of the next control mouse depicted on the gel was used. The basis for this comparison was that if treatment with antibiotics and lavage had a significant effect on the microbiota at 4 weeks, then the similarity of the banding profiles of mice from the

- 171- control group at 4 weeks to the 4 week banding profiles of other control mice would be greater than the other similarity for other comparisons.

6.2.2. 7 Statistics

The similarity measures for comparisons to the 4 week control sample and the faecal suspension were pooled by group and assessed for significance using the Friedman test. For example, in the comparison to the faecal suspension the groups assessed with the Friedman test were the Bray-Curtis similarities of: i) Control mice baseline samples to the faecal suspension ii) Control mice 4 week samples to the faecal suspension iii) Treated group baseline sample to the faecal suspension iv) Treated group 4 week sample to the faecal suspension

Similarly, in the comparison to the 4 week sample of control mice the groups assessed with the Friedman test were the Bray-Curtis similarities of: i) Control mice baseline samples to the 4 week sample of the next control mouse depicted on the gel ii) Control mice 4 week samples to the 4 week sample of the next control mouse depicted on the gel iii) Treated group baseline sample to the first 4 week control mouse sample depicted on the gel iv) Treated group 4 week sample to the first 4 week control mouse sample depicted on the gel

Post-hoc testing for the comparison of individual groups pre- and post-treatment was made using the Wilcoxon Matcl!ed Pairs Signed Ranks Test with a p value for significance of 0.05.

-172- 6.2.3 Experiment 3: Colonisation resistance - improved protocol

6.2.3.1 Animals

Eighty-four 12 to 16 week-old female BALB/C mice obtained from the ARC at 6 weeks of age and housed in the BABS animal facility were randomly divided into cages of 4 mice and ear notched for individual identification. Twelve mice (3 cages containing 4 mice) were assigned to each of the 7 treatment groups.

6.2.3.2 Faecal suspension

The faecal suspension to test colonisation resistance was obtained from nine 20 to 28 week-old C57BL/6 WEJ crosses bred within the BABS animal facility. These mice could be individually identified by their black and white "patchwork coat markings". Each cage of BALB/C mice received faecal suspensions from the same C57BL/6 WEJ mouse for the duration of treatment.

6.2.3.3 Experimental regimen

The 7 groups of mice were allocated to a range of interventions described in Table 6.2. After ear notching a baseline faecal sample was collected. For Days 1 to 4, mice in groups 3, 4, 6 and 7 were gavaged twice daily with antibiotics while the other groups of mice were gavaged with autoclaved PBS alone (Table 6.2). The antibiotic doses were the same as for Experiment 2. Mice were fasted (with access to water ad libitum) from 0:800 on Day 5 until the completion of magnesium sulfate cartharsis. The mice in groups 2 to 7 were gavaged with 4 doses of saturated magnesium sulfate (150 J.Ll each) at 9:00, 12:00, 15:00 and 18:00 on Day 5. The fmal2 doses were reduced to 100 J.Ll each if there was faecal staining of the fur around the tail indicating significant diarrhoea. Group 1 was gavaged with PBS only on Day 5. On Day 6 mice in groups 1 to 6 were gavaged at 08:00 and 16:00 with 150 J.Ll of faecal suspension, B. adolescentis suspension or PBS depending on treatment assignment.

-173- All suspensions given by gavage on Day 6 were preceded by 50 J..Ll of 5% NaHC03• Group 7 received a single 200 J..Ll enema of faecal suspension at 08:00 via a 20 gauge canula (Johnson &Johnson Medical, Arlington, USA) under ketamine/xylazine anaesthesia. Access to food recommenced after the first treatment at 08:00. Faecal pellets were collected from all mice 4 weeks after the completion of the treatment protocol. At 8 weeks all BALB/C mice were sacrificed by C02 asphyxiation followed by cervical dislocation. The caecum and colon were harvested and fixed in 10% neutral buffered formalin.

6.2.3.4 Histology

5 Jlm sections of the fixed tissues were cut and stained with haematoxylin and eosin. Blinded histologic assessment of the caecum and colon was performed using the scoring system of Berg et al (20).

6.2.3.5 PCR-DGGE

DNA extracted from the faecal pellets was subjected to PCR-DGGE with the domain Bacteria primer set as outlined in Section 2.1.3 and 2.1.4 of Chapter 2. PCR­ DGGE for Bifidobacterium was performed with samples from Groups 1 and 4 per the method of Satokari et al (291) outlined in Tables 2.1 and 2.3, Chapter 2. The samples from groups with significant results in domain Bacteria PCR-DGGE were also analysed with the Bacteroides-prevotella group and C. coccoides group primer sets as per the method described in summarised in Tables 2.1 and 2.3, Chapter 2. When this work was performed the PCR-DGGE assay for the C. leptum subgroup had not been developed.

6.2.3.6 Gel analysis

Each gel depicted samples from two cages of mice (with baseline and 4 week samples), a faecal suspension and a week 4 sample from a Group 3 mouse that

-174- represented the post-antibiotic and lavage pattern observed in Experiment 2 (discussed below). The faecal suspension sample used for PCR-DGGE comparison with Groups 5, 6 and 7 was a faecal sample from the respective source of faecal suspension for each cage collected at the start of the experiment. A randomly selected faecal suspension sample was used to generate a banding profile for comparison with other the remaining groups. The post-antibiotic and lavage profile used on each gel was generated from the 4 week sample from one randomly selected mouse from Group 3. For post-antibiotic and lavage comparisons within Group 3, each mouse's profile was compared with a cage mate's 4 week sample i.e. mouse 1 with mouse 2' s 4 weeks sample, mouse 2 with mouse 3' s week sample, 3 with 4 and 4 with 1. The Bray-Curtis similarity was calculated for each mouse's baseline and 4 week samples compared to a faecal suspension, and a post-antibiotic and lavage sample using the PRIMER 5 software package.

Two primary outcomes were assessed in this study. Firstly, the effect of treatment with faecal suspension on the micro biota of the treated mice was evaluated by comparing the baseline and 4 week samples of all groups of mice with the faecal suspension depicted on the gel. Secondly, the effect of antibiotics and lavage per se on the faecal microbiota was assessed by comparing the similarity of the banding patterns from baseline and 4 week samples of all treated and control mice to the 4 week sample of a Group 3 mouse.

6.2.3.7 Statistics

The similarity measures for comparisons to the faecal suspension and 4 week Group 3 sample were pooled by group and assessed for significance using the Friedman test. Post-hoc testing for comparison of individual groups pre- and post-treatment was made using the Wilcoxon Matched Pairs Signed Ranks Test. As a large number of groups (7) were included in this experiment an arbitrary significance level of p < 0.01 was set for the Wilcoxon Matched Pairs Signed Ranks Test. Results between 0.01 and 0.05 were considered of borderline significance.

-175- 6.3 RESULTS

6.3.1 Experiment 1: Effectiveness of different cathartic regimens for reducing luminal content in the murine caecum and colon

Two of the four mice in the group gavaged with sodium phosphate solution died overnight on Day 2 suggesting that this agent is not suitable for use in mice. As only two mice remained in this group the results for these mice were not analysed further.

There were no other deaths. Mice gavaged with MgS04 were observed to have wet fur around their tails suggesting the presence of significant watery diarrhoea.

The volume of water consumed per mouse during the study was 4 ml for the controls as compared with 9.25 ml for the MgS04 group. The polyethylene glycol treated group consumed 6.75 ml of solution while the sorbitol treated group consumed 10.75 mi. Caecal weights were observed to be higher in treated mice than the controls even allowing for the lower body weight of control mice (Table 6.3). The mean number of colonic pellets was lower in the MgS04 and polyethylene glycol solution treated groups than in the controls (0.5 ± 0.6, 2.25 ± 2 versus 5.2 ± 1.8 respectively). The appearance of the caecal fluid was clear in the MgS04 group and light or dark in the remaining cathartic treated groups. Overall the number of mice was too small for statistical comparisons of any parameter. However the number of colonic pellets and the appearance of the caecum strongly favoured the efficacy of saturated magnesium sulfate solution for producing a cathartic effect in the mice. Using the same criteria, polyethylene glycol was the next most effective agent, with sorbitol of limited usefulness.

-176- 6.3.2 Experiment 2: Colonisation resistance

This study combined the optimal method of catharsis with antibiotics, with the aim of reducing colonisation resistance. Whether resistance was reduced was assessed by gavage with faecal suspension.

Two mice in each of the treatment and control groups died during the study, possibly as a result of the stress of frequent handling and gavage combined with abnormally low ambient temperatures in the animal facility related to an air-conditioning malfunction at the time. There was no significant difference in the weights of the treated versus the control mice throughout the study (Figure 6.2). However the body weight of both groups of mice declined substantially during the period in which twice daily handling and gavage occurred (Figure 6.2).

The banding patterns obtained with baseline and 4 week faecal samples in three treated and control mice using the domain Bacteria primers are shown in Figure 6.3. As can be observed, the control mice have homologous bariding profiles at 4 weeks (lanes 9, 11 and 13) suggesting that a common microbiota has developed after antibiotic treatment and lavage. This change is not evident in the treated mice. The similarity measures for samples from all mice to the post antibiotic and lavage banding profile, by group, was not significant (Friedman test X2 7.2 p = 0.066 ); shown in Table 6.4. However as this was close to a significant level, assessment with the Wilcoxon test was performed. Direct comparisons of the similarity measures for treated mice to the post antibiotic and lavage profile were not significant (baseline similarity 40.4 ± 15.3 versus 4-week similarity 47.2 ± 12, Wilcoxon test p = 0.77) while similarities for control mice were significant (baseline similarity 48.4 ± 11.6 versus 4-week similarity 75.9 ± 26.3, Wilcoxon test p = 0.03).

The similarities of baseline and 4 week samples to the faecal suspension by group were significant (Friedman test X2 14.9 p = 0.002) as shown in Table 6.4. Post hoc testing showed that the similarity of baseline and 4 week samples of control mice to

-177- the faecal suspension was not significant (baseline similarity 47.8 ± 12.4 versus 4- week similarity 44.5 ± 9, Wilcoxon test p = 0.43). In contrast, the comparisons for treated mice were highly significant (baseline similarity 47.5 ± 16.5 versus 4-week similarity 74.2 ± 7.9, Wilcoxon test p = 0.004).

Caecal and colonic histology was normal in all of the mice.

6.3.3 Experiment 3: Colonisation resistance- improved protocol

In view of the encouraging results from experiment 2 and the two deaths that occurred, an abbreviated protocol was applied to older groups of mice in a follow-up study. Additional primer sets were also employed to confirm the results obtained with the domain Bacteria primers. The key elements of the protocol to reduce colonisation resistance were assessed, and preliminary studies were undertaken into the optimal method of administering the faecal suspensions.

This protocol differed from that of experiment 2 in that only 4 rather than 5 days of antibiotics were given and that the faecal suspensions were only administered for one as compared with 6 days. There were no deaths in this older cohort of mice treated with a shorter protocol.

In this study the primer set for Bifidobacterium species did not produce a PCR product from the baseline or 4 week samples of control mice (Group 1) or mice treated with antibiotics, lavage and B. adolescentis suspension (Group 4 mice). In contrast, the positive control was amplified.

The similarity measures for baseline and 4 week samples to the post-antibiotic and lavage sample and the faecal suspension samples generated with domain Bacteria primers were each significantly different when analysed by group (Friedman's test X2 56.5 p <0.0001 and X2 28.7 p =0.003 respectively) as sho~n in Table 6.5. Post­ hoc analysis with the Wilcoxon test revealed that the groups treated with lavage

-178- alone or antibiotics and lavage only (Groups 2 and 3 respectively), had a significant change in their 4 week banding patterns towards that of the post-antibiotic and lavage pattern. The gel depicting the results for mice from Group 3 is shown in Figure 6.4. Group 2 had a similarity of33.6 ± 9 at baseline to the post-antibiotic and lavage pattern that increased to 44 ± 11.9 at 4 weeks (p =0.006). The corresponding similarities for Group 3 were 38.5 ± 10.7 and 68.3 ± 18.1 (p =0.001). The mice treated with antibiotics and lavage followed by gavage with faecal suspension (Group 6) were the only group that had a significant increase in similarity to the faecal suspension in post-hoc analysis (p = 0.009 Wilcoxon test). A gel illustrating the results for Group 6 mice is shown in Figure 6.5. The similarity at baseline was 54.5 ± 15.9 and at 4 weeks 66.6 ± 6.1. The mice treated with lavage (not antibiotics) and then gavaged with faecal suspension (Group 5) and the mice given faecal suspension by enema (Group 7) did not show significant increases in similarity to the faecal suspension banding profile. A dendrogram showing the relatedness of the banding profiles of 4 mice from each of Groups 3 (antibiotics and lavage alone) and 6 (antibiotics, lavage and gavage with faecal suspension) to the post-antibiotic and lavage banding pattern and faecal suspension is shown in Figure 6.6. In this dendrogram the week 4 banding patterns cluster together. Banding patterns for Group 6 are clustered with the faecal suspension pattern while the Group 3 banding patterns cluster together suggesting a common post-antibiotic and lavage pattern.

Examination of the banding patterns obtained with domain Bacteria primers from the baseline and 4 week faecal samples of the control mice (Group 1) revealed a number of differences in their banding profiles (Figure 6.7). The mean Bray-Curtis similarity for this comparison was 67 ± 15.1. A dendrogram showing the relatedness of baseline and 4 week samples from this group is shown in Figure 6.8. There was no significant change for control mice towards the post-antibiotic and lavage pattern or the faecal suspension pattern during the 5 week study period (p = 0.08 and p = 0.42 respectively, Wilcoxon test).

-179- Two of the groups with significant domain Bacteria PCR-DGGE similarity comparisons in post-hoc testing, Groups 3 (antibiotics and lavage only) and 6 (as for Group 3 plus gavage with faecal suspension) were further evaluated with the primer sets for the Bacteroides-prevotella group and the C. coccoides group. Gels depicting results for some of the mice from Groups 3 and 6 with each of these primer sets are shown in Figures 6.9, 6.10, 6.12 and 6.13. Dendrograms representing the Bray­ Curtis similarities obtained with the two primer sets from some of the mice are shown in Figures 6.11 and 6.14. The similarities of Group 3 and 6 samples to the post-antibiotic and lavage pattern obtained with the Bacteroides-prevotella group and C. coccoides group primer sets were significant (Friedman test X2 23.8 p < 0.0001 and X2 9.9 p = 0.02 respectively) as shown in Tables 6.6 and 6.7. Similarities to the faecal suspension pattern were also found to be significant (Friedman test X2 13.2 p = 0.004 and X2 12.7 p = 0.005 respectively), Tables 6.6 and 6.7. The increase in the similarity of mice treated with antibiotics and lavage only (Group 3) to the post-antibiotic and lavage profile was significant for the Bacteroides-prevotella group (39.9 ± 8.9 to 82.4 ± 9.2, p = 0.0005 Wilcoxon test) and of borderline

significance for ~e C. coccoides group (37.7 ± 10.8 to 55.7 ± 18.7, p = 0.015.. Wilcoxon test). The same comparisons for Group 6 mice were not significant (p = 0.28 and p = 0.08 respectively, Wilcoxon test). Conversely, the similarity of Group 6 samples at baseline and 4 weeks to the faecal suspension increased significantly for C. coccoides group primers (40 ± 14 to 59.6 ± 10.8, p =0.005 Wilcoxon test) and was of borderline significance for the Bacteroides-prevotella group (50.2 ± 14 to 66.1 ± 15.2, p = 0.04 Wilcoxon test). The same comparisons for Group 3 were not significant (p = 0.27 and p = 0.21 respectively, Wilcoxon test). In summary, the results confmned the observations made with the domain Bacteria PCR-DGGE in Experiments 2 and 3.

There was no significant difference in the weights of the mice from different groups during the study (Friedman test X2 22.1 p = 0.10). The histology of the caecum and colon of all mice was normal.

-180- TABLE 6.1: Treatment regimen for experiment 2. Colonisation resistance.

Group Treatment Day 1 to5 Day5to6 Day 6 to 11

1 Controls Antibiotics MgS04 PBS

2 Treatment Antibiotics MgS04 Faecal suspension

TABLE 6.2: Treatment regimen for experiment 3. Colonisation resistance­ improved protocol.

Group Day 1 to 4 DayS Day6

1 Controls PBS PBS PBS+HC03

2 Lavage only PBS MgS04 PBS+HC03

3 Antibiotics + Lavage Antibiotics MgS04 PBS+HC03

4 B. adolescentis Antibiotics MgS04 B. adolescentis + HC03

5 Lavage + faecal suspension PBS MgS04 faecal suspension + HC03

6 Full protocol Antibiotics MgS04 faecal suspension + HC03

7 Full protocol with faecal Antibiotics MgS04 faecal suspension by suspension by enema enema TABLE 6.3: Results of Experiment 1. Effectiveness of different cathartic regimens.

Treatment Baseline Sacrifice Number Caecal caecal Caecum and allocation weight g weight g of colonic content weightg colon weight pellets g Polyethylene 18.8 ± 2.1 19.8 ±2.9 2.25 ±2 Light or 0.46 ± 0.18 0.78±0.22 glycol dark solution liquid (N=4) Sorbitol 18.4 ± 0.5 18.6 ± 0.8 6± 1.4 Light or 0.34±0.09 0.73±0.13 (N=4) dark liquid MgS04 19.8 ± 1.5 18.5 ± 3.1 0.5 ±0.6 Clear 0.36 ±0.22 0.61 ± 0.22 (N=4) liquid Control 16.4± 1.2 16.4 ± 1.7 5.2 ± 1.8 Dark 0.20±0.03 0.48 ±0.05 (N=S) semi- solid TABLE 6.4: Bray-Curtis similarity by group with the domain Bacteria primer set in Experiment 2.

Group Similarity to post- Friedman Wilcoxon Similarity to faecal Friedman Wilcoxon antibiotic + lavage profile test test suspension test test (mean±SD) (mean±SD) Baseline 4 weeks Baseline 4weeks 1 Controls 48.4 ± 11.6 75.9 ± 26.3 0.066 0.03* 47.8 ± 12.4 44.5±9 0.002* 0.43 2Treatment 40.4± 15.3 47.2± 12 0.77 47.5 ± 16.5 74.2±7.9 0.004* * significant at p < 0.05 level in the Friedman or Wilcoxon tests. TABLE 6.5: Bray-Curtis similarity by group with the domain Bacteria primer set from Experiment 3.

Group Similarity to post- Friedman Wilcoxon Similarity to faecal Friedman Wilcoxon antibiotic + lavage profile test test suspension test test (mean±SD) (mean±SD) Baseline 4 weeks Baseline 4weeks

1 Controls 46.1 ± 13.6 54.2± 10.2 <0.0001* 0.08 52.3 ± 11.2 49 ± 9.4 0.003* 0.42

2 Lavage only 33.6 ± 9 44± 11.9 0.006** 49.6 ± 14.4 45.1 ± 10.5 0.51

3 Antibiotics 38.5 ± 10.7 68.3 ± 18.1 0.001 ** 52.8 ± 6.1 60.7 ± 8.8 0.054 and lavage

5 Lavage and 34.5 ± 8.6 41.6 ± 7.5 0.09 48.7 ± 13.8 52.2±7.5 0.38 faecal suspension 6 Full 52.4± 6.8 42±3.8 ND1 54.5 ± 15.9 66.6 ± 6.1 0.009** protocol

7 Faecal 40.5 ±7.4 36.2±6.2 0.1 55.1 ± 13 57.5 ±8.2 0.32 suspension by enema

* significant at p < 0.05 level in the Friedman test. **significant at P < 0.01level in the Wilcoxon test. 1 inadequate data for assessment i.e. the post-antibiotic and lavage pattern was not represented on the gel for 2 of the 3 cages in this group. TABLE 6.6: Bray-Curtis similarity by group with the Bacteroides-prevotella group primer set in Experiment 3.

Group Similarity to post- Friedman Wilcoxon Similarity to faecal Friedman Wilcoxon antibiotic + lavage profile test test suspension test test (mean±SD) (mean±SD) Baseline 4 weeks Baseline 4weeks 3 Antibiotics 39.9 ± 8.9 82.4 ± 9.2 <0.0001* 0.0005** 51.9 ± 16.8 46.4± 8.9 0.004* 0.27 and lavage 6 Full 40.1 ± 13 44.3 ±9.4 0.28 50.2 ± 14 66.1 ± 15.2 0.04 protocol

* significant at p < 0.05 level in the Friedman test. **significant at p < 0.01level in the Wilcoxon test. TABLE 6.7: Bray-Curtis similarity by group with the Clostridium coccoides group primer set in Experiment 3.

Group Similarity to post- Friedman Wilcoxon Similarity to faecal Friedman Wilcoxon antibiotic + lavage proide test test suspension test test (mean±SD) (mean±SD) Baseline 4weeks Baseline 4 weeks 3 Antibiotics 37.7 ± 10.8 55.7 ± 18.7 0.02* 0.015 42± 13 47.1 ± 15.4 0.005* 0.21 and lavage 6 Full 48.7 ±9.7 41.2± 15.2 0.08 40± 14 59.6± 10.8 0.005** protocol

* significant at p < 0.05level in the Friedman test. **significant at p < 0.01level in the Wilcoxon test. Figure 6.1: Experimental protocol for BALB/C mice in Experiment 2.

·-control mice Treated mice

Baseline faecal sample collected

Twice daily gavage with antibiotics• on Days 1 to 5

Magnesium sulfate catharsis

Day 5 18:00: Fasting. First dose MgS04 Day 6 three doses of MgSO4

PBS Faecal suspension Day 6: 18:00 Recommence diet . Day 6: 18:00 Recommence diet . Day 7 to 11:gavage twice daily Day 7 to 11:gavage twice daily with PBS with faecal suspension

Faecal sample collected 4 weeks post-treatment

Sacrifice and collect large bowel 8 weeks post-treatment Figure 6.2: The weight of mice in grams during Experiment 2. Bars show the mean and standard deviation. During the period of gavage with antibiotics, magnesium sulfate and faecal suspension weight gain did not occur.

25

···············-····-·-··················;r···-··i··············-··-·············-l----··········· f i U I 'I • Control group ·-······-·-····-········f·····-l·-·········-··-············--··-·····················-···········- ..... • Treatment group "§, 10 ------...... ·a; 3: 5 ·p:~tibi;;i~~-Iava9e"ancr~iava9in·9·······················································

0 0 2 4 6 8 10 12 Weeks Figure 6.3: The results of domain Bacteria PCR-DGGE for 3 treated and 3 control mice from Experiment 2. Lanes 1, 3 and 5: Baseline samples from treated mice, Lanes 2,4 and 6: Week 4 samples from treated mice; Lane 7: Faecal suspension used to gavage the treated mice. Lanes 8,10 and 12: Baseline samples from control mice, Lanes 9,11 and 13: Week 4 samples from control mice.

Treated mice Control mice

Faecal I I I I I I I I Base 4wks Base 4wks Base 4wks ~uspens - Base 4 wks Base 4 wks Base 4 wks 100 1 2 3 4 5 6 7 8 9 10 11 12 13 Figure 6.4: The results of domain Bacteria PCR-DGGE with mice from Group 3, Experiment 3. Lane !:Baseline mouse 1, Lane 2: Week 4 mouse 1, Lane 3: Baseline mouse 2, Lane 4: Week 4 mouse 2, Lane 5: Baseline mouse 3, Lane 6: Week 4 mouse 3, Lane 7: Baseline mouse 4, Lane 8: Week 4 mouse 4. A similar banding pattern is evident in all of the 4 week samples (Lanes 2, 4, 6 and 8) -arrows.

Base 4 wks Base 4 wks Base 4 wks Base 4 wks 1 2 3 4 5 6 7 8 Figure 6.5: The results of domain Bacteria PCR-DGGE with mice from Group 6, Experiment 3. Lane !:Baseline mouse 1, Lane 2: Week 4 mouse 1, Lane 3: Baseline mouse 2, Lane 4: Week 4 mouse 2, Lane 5: Baseline mouse 3, Lane 6: Week 4 mouse 3, Lane 7: Faecal suspension used to gavage these mice.

Faecal Base 4 wks Base 4 wks Base4wks suspension 1 2 3 4 5 6 7 Figure 6.6: Dendrogram based on Bray-Curtis similarity showing the relationship between Baseline and Week 4 samples from Group 3 and Group 6 mice with PCR-OOGE using domain Bacteria primers. Baseline samples are in red boxes and 4 week samples in blue boxes. Faecal suspension used to gavage the group 6 mice in the black box. Boxes for the group 6 mice have dashed borders. The week 4 samples of Group 3 cluster together. In addition the week 4 samples of Group 6 cluster with the faecal suspenston.

50

60

90

100 Figure 6.7: The results of domain Bacteria PC.:R-DGGE with samples from mice of Group 1, Experiment 3. Lane 1: Faecal suspension, Lane 2: Baseline mouse 1, Lane 3: Week 4 mouse 1, Lane 4: Baseline mouse 2, Lane 5: Week 4 mouse 2, Lane 6: Baseline mouse 3, Lane 7: Week 4 mouse 3.

Faecal suspension Base 4wks Base 4wks Base 4wks 12 3 4 56 7 Figure 6.8: Dendrogram based on Bray-Curtis similarity showing the relationship between Baseline and Week 4 samples from Group 1 mice with PCR-DGGE using domain Bacteria primers. Baseline samples are in red boxes and 4 week samples in blue boxes. Faecal suspension in the black box. Post antibiotic and lavage sample in green. Note that the 4 week samples (blue) do not cluster together.

40

>. 80 ·~ -...... 8 80 CZl

100 Figure 6.9: The results of Bacteroides-prevotella group PCR-DGGE with mice from Group 3, Experiment 3. Lane 1: Faecal suspension, Lane 2: Baseline mouse 1, Lane 3: Week 4 mouse 1, Lane 4: Baseline mouse 2, Lane 5: Week 4 mouse 2, Lane 6: Baseline mouse 3, Lane 7: Week 4 mouse 3.

Faecal suspension Base 4 wks Base 4 wks Base 4 wks 1 2 3 4 56 7 Figure 6.10: The results of Bacteroides-Prevotella group PCR­ DGGE with mice from Group 6, Experiment 3. Lane !:Baseline mouse 1, Lane 2: Week 4 mouse 1, Lane 3: Baseline mouse 2, Lane 4: Week 4 mouse 2, Lane 5: Baseline mouse 3, Lane 6: Week 4 mouse 3, Lane 7: Faecal suspension used to gavage these mice.

Faecal Base 4wks Base 4wksBase 4wks suspension Figure 6.11: Dendrogram based on Bray-Curtis similarity showing the relationship between Baseline and Week 4 samples from Group 3 and Group 6 mice with PCR-DGGE using Bacteroides-prevotella group primers. Baseline samples are in red boxes and 4 week samples in blue boxes. Faecal suspension used to gavage the group 6 mice in the black box. Boxes for the Group 6 mice have dashed borders. The week 4 samples of Group 3 cluster together. In addition the week 4 samples of Group 6 cluster with the faecal suspension.

20

40

100 • ' - . 1 I u 1 fJ I ~ I • I ~ ~ rJ l.. ~ I l.. =- l.. <;! I L ~ ~ o Figure 6.12: The results of C. coccoides PCR-DGGE with mice from Group 3, Experiment 3. Lane 1: Faecal suspension, Lane 2: Baseline mouse 1, Lane 3: Week 4 mouse 1, Lane 4: Baseline mouse 2, Lane 5: Week 4 mouse 2, Lane 6: Baseline mouse 3, Lane 7: Week 4 mouse 3. A similar banding pattern is evident in the 4 week samples (Lanes 3, 5, and 7) that differs from baseline.

Faecal suspensiOn Base 4wks Base 4wks Base 4wks 1 2 3 4 5 6 7 Figure 6.13: The results of C. coccoides PCR-OOGE with mice from Group 6, Experiment 3. Lane 1: Baseline mouse 1, Lane 2: Week 4 mouse 1, Lane 3: Baseline mouse 2, Lane 4: Week 4 mouse 2, Lane 5: Baseline mouse 3, Lane 6: Week 4 mouse 3, Lane 7: Faecal suspension used to gavage these mice.

,---., Base 4wks Base 4wks Base 4wks Faecal . suspensiOn 1 2 3 4 5 6 7 Figure 6.14: Dendrogram based on Bray-Curtis similarity showing the relationship between Baseline and Week 4 samples from Group 3 and Group 6 mice with PCR-DGGE using C. coccoides group primers. Baseline samples are in red boxes and 4 week samples in blue boxes. Faecal suspension used to gavage the group 6 mice in the black box. Boxes for the Group 6 mice have dashed borders. The week 4 samples of Group 3 cluster together. In addition the week 4 samples of Group 6 cluster with the faecal suspension.

20

80

100 6.4 DISCUSSION

The comparative effectiveness of different cathartic regimens for reducing luminal content in the murine caecum and colon was initially examined to determine the best protocol for future studies. The degree of reduction in luminal content achieved appeared to be better in the group gavaged with MgS04 than in any of the other groups as assessed by macroscopic caecal appearance and the number of colonic pellets. The intake of fluid from the water bottles was higher in the MgS04 and sorbitol groups as compared with the polyethylene glycol solution and control groups. This larger consumption of fluid by these groups of mice was probably related to cathartic-induced dehydration in the case ofMgS04 treated mice and the sweet taste of sorbitol in that group. Higher caecal weights were also observed in the treated mice than in the controls, even allowing for the difference in body weight at the commencement of the study. Given the short duration of the study, the observed increase in caecal and colonic weights for all treated groups is likely to be due to increased luminal or tissue fluid volume, the former being more likely and most probably due to the osmotic effect of the cathartic agents. Consistent with this was the observed reduction in the body weight of the MgS04 treated group (6.6% of pre­ study weight) that is likely to reflect volume depletion due to diarrhoeal losses. This reduction in body weight is concerning and could be addressed by reducing the

MgS04 dosage or subcutaneous injection of saline during the protocol. Certainly

longer durations or larger doses of MgS04 treatment could increase the risk of mortality if further reductions in body weight were to occur.

The addition of polyethylene glycol solution to the water bottles was the next most effective method. An average of 2 pellets remained in the colon of these mice at the end of this protocol suggesting that a longer duration of treatment may have been as

effective as MgS04• However the duration of fasting required would raise ethical issues and increase stress on the animals. Sorbitol appeared largely ineffective over this duration of treatment.

-201- The results of this preliminary experiment also suggested that the use of sodium phosphate solution is unacceptable as 2 of the 4 mice in this group died. Therefore magnesium sulfate was adopted as the cathartic of choice for the experiments that followed.

Having determined the best method for reducing the large intestinal luminal contents, this approach was combined with broad-spectrum antibiotics in the second experiment in an effort to substantially reduce colonisation resistance. Whether colonisation resistance was affected was assessed by treating the mice with a faecal slurry from another mouse and assessing the effect on the microbiota with domain Bacteria PCR-DGGE.

The antibiotics vancomycin, amoxycillin and metronidazole were chosen because of their antibacterial spectrum, absorption characteristics or proven safety in previous murine experiments by the Helicobacter Research Group at the University of New South Wales. Vancomycin is active against gram-positive organisms including clostridia and enterococci. It is poorly absorbed from the gastrointestinal tract resulting in high luminal concentrations (277). Oral penicillins are well tolerated in mice and have been shown to have dose-related effects against a broad range of bacterial species including lactobacilli, enterococci and other anaerobes (98, 295). Metronidazole was used for its bacteriocidal effect against most obligate anaerobic bacteria and its reported safe use in mice (98, 242).

Observation of the banding patterns obtained from the mice treated with antibiotics and lavage only (controls) in this experiment suggested that a common microbiota may have developed after treatment with antibiotics and lavage. In these mice a

common microbiota w~s evident at 4 weeks post treatment that was not present in the mice gavaged with the faecal suspension. The p value for the comparisons of samples from each group to the 4 week samples of controls was close to significance (p = 0.066) and post hoc testing suggested that the results for the control group may have been responsible for this (Table 6.4). In Chapter 3 the banding patterns of

-202- faecal samples from cagemates were of higher similarity than comparisons of cagemates and non-cagemates with domain Bacteria primers. In the current experiment only 3 cages of mice were studied in the control group, however an effect of these mice being cagemates is not likely to explain the results. Subsequently studies of untreated mice (Group 1) from experiment 3 did not reveal an increasing similarity to other cagemates over time. Therefore the relatively uniform nature of the banding patterns of mice treated with antibiotics and lavage at 4 weeks suggests that a common post-treatment microbiota may develop despite some differences in baseline microbiota. Previous culture-based studies have shown that antibiotic treatment can produce durable changes in the murine caecal flora 30 and 90 days after treatment with antibiotics (295). It may be that this microbiota predominantly represents the antibiotic resistant components of the baseline flora.

The 4 week banding patterns of mice gavaged with faecal suspension in experiment 2 were not similar to those of the control group in spite of treatment with antibiotics and lavage. That is, the increased similarity of the 4 week as compared with baseline samples to the post-antibiotic and lavage pattern of the control mice was not seen. In addition there was a significant increase in the similarity of the banding profile at 4 weeks to the faecal suspension, as compared with the similarity of the baseline sample to the faecal suspension. The combination of these results suggests that the treatment regimen reduced colonisation resistance to a degree sufficient to allow the development of a hybrid microbiota, consisting of bacteria from the faecal suspension derived from another mouse, plus part of their own baseline microbiota. This implies survival of faecal bacterial populations through the process required to produce a suspension as well as the defences of the murine upper gastrointestinal tract. As the primer set of Zoetendal et al appears to predominantly represent · anaerobic bacteria, it is likely that the changes in the microbiota affected the anaerobic populations (388).

Unfortunately there was significant mortality associated with the regimen applied in experiment 2. Two mice in each of the treatment and control groups died during the

-203- study. These deaths were most likely related to the stress caused by frequent handling and gavage in mice of this age and the low ambient temperatures in the animal facility at that time due to an air conditioning malfunction. Given these findings, further work was required to produce a regimen that was not associated with such a high mortality. A shorter protocol was introduced and older mice were studied.

In the third experiment further work was performed that confirmed the observations of experiment 2 in a larger group of older mice using additional primer sets. A common post-antibiotic and lavage banding pattern was again observed with domain Bacteria primers. Both mice treated with lavage (Group 2) or antibiotics and lavage (Group 3) had a significant change in their 4 week banding profiles towards that of the post-antibiotic and lavage profile. The magnitude of this change was much higher in the Group 3 mice as compared with Group 2 (a 30% increase in similarity versus 10%) suggesting that antibiotics have the most substantial effect in inducing a common microbiota. Analysis of the results for samples from Group 3 with primers for the Bacteroides-prevotella group and C. coccoides group confirmed that increases in similarity to the post-antibiotic and lavage banding profile had occurred. These were significant with the former primer set and of borderline significance with the latter. It is likely that this convergence towards a common microbiota after treatment reflects the dominance of antibiotic resistant strains of bacteria present in the pre-treatment microbiota. Other authors have also shown a convergence of banding profiles after antibiotic treatment in mice using a different universal 16S rRNA primer set (217).

The third experiment also confirmed that antibiotics and lavage reduced colonisation resistance such that dosing with a mouse-adapted microbiota induced a hybrid microbiota that was detected by the analysis of faecal samples. The mice prepared with antibiotics and lavage and then gavaged with faecal suspension (Group 6) were the only group of mice that had a significant increase in similarity to the faecal suspension in post-hoc analysis with the domain Bacteria primer set. These results

-204- were confrrmed with the Bacteroides-prevotella group and C. coccoides group primer sets although the statistical significance was borderline with the former (p = 0.04). Therefore, it is likely that the bacterial populations represented by both of these primer sets were significantly altered by the protocol. The lack of significant results for c0mparisons of samples from the Group 3 mice to the faecal suspension shows that convergence to a common microbiota after treatment with antibiotics and lavage does not account for the results.

Antibiotics appear to be a key component of the regimen to reduce colonisation resistance as lavage alone did not allow the development of a hybrid flora. The group treated with lavage and faecal suspension but no antibiotics· (Group 5) did not have a significant increase in similarity to the faecal suspension.

In addition, the failure of the administration of faecal suspension by enema to induce a hybrid microbiota suggests that introducing new bacteria directly into the caecum may be critical to generating a hybrid microbiota. That is, despite treatment with a regimen that proved to be effective in reducing colonisation resistance, the administration of faecal suspension by enema did not result in an increase in the post-treatment banding patterns similarity to the faecal suspension (Group 7). This failure to develop a hybrid microbiota is most likely due to the inability of the enema volume to reach the caecum in these mice. This may be due to distension of the caecum and colon with fluid, as was observed at sacrifice in mice treated with

MgS04 catharsis in experiment 1 (Table 6.3). Thus the effectiveness of this method of administration could possibly be improved by lengthening the period of fasting after catharsis so that there is less fluid in the colon. A second alternative would be to increase the enema volume, however this may increase the risk of bowel perforation during administration. The result of the current study would suggest that to introduce new bacterial strains into the lower bowel microbiota of mice, the strains must reach the caecum during the period of reduced colonisation resistance.

-205- Interestingly, despite using a regimen shown to reduce colonisation resistance, a human strain of B. adolescentis did not colonise the murine lower bowel in mice prepared with antibiotics and lavage (Group 4). A human strain of B. adolescentis was chosen as the anaerobic species for testing as this strain is usually represented by a single band when analysed with the PCR-DGGE method of Satokari et al (291) and because B. adolescentis has not been reported to be present in the faeces of mice. A previous study of bifidobacteria with species-specific PCR showed B. longum but not B. adolescentis to be present in the faeces of mice (370). Therefore, had this species colonised the mice it should have been identifiable as a single separate band at PCR-DGGE. The absence of PCR product when the faeces of untreated (Group 1) mice were tested suggests that Bifidobacterium species were not a significant part of the normal faecal flora of the BALB/C mice studied. The absence of PCR product from the 4 week samples of Group 4 mice, demonstrates that treatment with antibiotics and catharsis followed by dosing with freshly harvested bifidobacteria, did not result in durable colonisation with a non-murine bacterial strain. Therefore this preliminary result suggests that it is likely that bacterial species that are not adapted to specific animal hosts are unlikely to colonise the gut, even when introduced during a period of reduced colonisation resistance.

Finally the changing banding patterns for control mice observed with domain Bacteria primers suggests that at 12 to 16 weeks of age the murine lower bowel microbiota has not yet reached its final adult composition or "climax community". This result obtained for the untreated mice (Group 1) contrasts with the known stability of the faecal PCR-DGGE profile obtained with this primer set in adult humans (388). These changes could reflect the ongoing development of the murine caecal microbiota in these young mice or exposure to novel anaerobic caecal microflora in a new animal facility. Certainly over the same time period in infant humans it is clear that the colonic microbiota is still developing.

The histology of the caecum and colon of all mice was within normal limits i.e. all scores were zero using the scoring of Berg et al. This suggests that there was no

-206- immediate pathological consequence to changing the host anaerobic flora. These results call into question the biological significance of the observations of Duchmann et al who found evidence for host tolerance to "self' bacterial flora and a T cell response to non-self bacterial flora in human controls and a loss of "self'­ tolerance in IBD cases (79). If the mice with a hybrid bacterial flora (Group 6) had a T -cell response to the luminal bacteria this was not sufficient to induce colitis in the subsequent 4 weeks. That is aT-cell response to non-"self' flora either did not occur, was delayed or was not sufficient to induce colitis.

To summarise, treatment with broad-spectrum antibiotics and lavage reduced colonisation resistance while lavage alone did not. A mixed population of mouse faecal bacteria colonised the murine lower bowel when introduced during a period of reduced colonisation resistance. However a bacterial strain of human origin failed to colonise mice effectively. Mouse-adapted bacterial strains administered into the murine colon by enema (as opposed to the caecum) did not colonise under the conditions used in the study, suggesting that the caecum may be the critical site for introducing new species into the microbiota of the lower bowel. Ideally theref?re to introduce probiotic bacterial strains into the lower bowel and produce colonisation, an effective regimen to reduce colonisation resistance will be required plus the use of host-adapted species. Furthermore a method of administration that results in viable bacteria reaching the target environmental niche is essential.

-207- CHAPTER 7: Colonisation resistance in humans

7.1 INTRODUCTION

The opportunity arose during the course of this research to study an alternative health practice that specifically tested the concept of colonisation resistance in humans. In this treatment, an attempt is made to alter the bacterial flora of the colon by the administration of a suspension of faeces from one subject into the colon of a second subject with gastrointestinal or non-gastrointestinal disease. This procedure is analogous to the studies conducted in mice reported in Chapter 6.

The administration of a suspension of faeces from a normal donor into the colon of a recipient with colonic disease has been reported by a number of researchers to be beneficial in a several uncontrolled studies of a range of colonic diseases. For example, clinically significant responses have been described in antibiotic­ associated diarrhoea (83), Clostridium difficile associated diarrhoea (33, 142, 251, 256, 306) and ulcerative colitis (18, 30). In addition, enemas containing one or more cultured colonic bacterial species have also been reported to be efficacious in the treatment of chronic C. difficile associated diarrhoea (352) and constipation (6, 274).

Clostridium difficile associated diarrhoea usually arises in the context of antibiotic use and is thought to reflect a disturbance of the normal anaerobic colonic bacteria, that results in overgrowth of the organism and subsequent toxin production (53, 90). Up to 30% of patients with C. difficile diarrhoea will relapse after standard antibiotic therapy and a small proportion of these will have refractory disease (53, 90). As the disease is relapsing and remitting, in studies of prospective therapies it is essep.tial to include controls if meaningful conclusions are to be reached (90). C. difficile associated disease that is refractory to, or relapses following standard therapy with oral vancomycin or metronidazole has been reported to be successfully treated with colonic lavage followed by donor faecal enemas (256) or by donor faecal enemas alone (33, 83, 142, 251, 306). While these authors reported significant improvement

-208- in 80% or more of cases, the findings of these studies are questionable as the series were uncontrolled and as stated earlier C. difficile diarrhoea may resolve without specific treatment.

Mixtures of colonic bacteria grown in pure culture and given by rectal instillation have been reported to be beneficial in C. difficile diarrhoea and constipation. For example, Tvede and Rask-Madsen (1989) gave rectal instillations of a mixture of 10 colonic bacteria into the unprepared colon of 5 patients with refractory C. difficile diarrhoea resulting in immediate clinical responses (352). Andrews and Borody (1993) have also reported significant clinical improvement in 25 of 33 patients with chronic constipation given colonic lavage, antibiotics and a caecal infusion of a mixture of 18 bowel bacteria at colonoscopy (6).

The mechanism underlying the observed response to these faecal or bacterial enemas has not been systematically investigated previously. It is generally held that colonic lavage washes out C. difficile spores and that donor faecal enemas restore a normal microbial flora in C. difficile associated diarrhoea (141, 142, 352). In the other diseases, such as constipation and ulcerative colitis, therapeutic benefit is thought to result from a change in the microbiota of the colon. However, apart from the small study of Tvede and Rask-Madsen (1989), the composition of the patients faecal microbiota has not been examined pre- and post- enema (352). In this study of six patients with diarrhoea related to C. difficile that was resistant to multiple courses of antibiotics, 1 patient was treated with enemas of donor faeces, 4 patients with enemas containing 10 cultured bacterial species and 1 with both. In all six subjects the treatments resulted in a resolution of their bowel symptoms within 2.4 hours. Prior to treatment, culture of the patients' faeces revealed predominantly aerobic or facultative aerobic species as is common in diarrhoeal states (3, 135, 156). However following treatment a more diverse microbiota including anaerobes was established. Whether the additional anaerobic species present in faecal samples obtained after treatment originated from the introduced bacterial species or represented a regeneration of the patients' pre-treatment microbiota is uncertain. Tvede and Rask-

-209- Madsen (1989) reported that several of the bacterial strains given in the enemas could be cultured from the faeces of the treated patients for 6 months after therapy, however these only represented a small proportion of the flora that was present.

Studies by Gustafsson et al in 1998 and 1999 documented that short-chain fatty acid levels and several faecal enzyme activities normalised after treatment with faecal enemas (141, 142). In this study, nine patients with antibiotic-associated and C. difficile-associated diarrhoea treated with donor faecal enemas were examined (141, 142). In this case the enema consisted of a suspension of faeces in milk that had been frozen at -20°C and thawed just prior to use. In two of the nine patients two treatments were required as they failed to respond to the first treatment. Following treatment, Gustafsson et al reported a return of relatively normal short-chain fatty acid levels and enzyme activities in the faecal samples of the treated cases. The interpretation of this data suggests that it is likely that the results are documenting a return to relatively normal conditions in the colon. Unfortunately these investigators did not include controls with antibiotic-associated diarrhoea that responded to antibiotics. Had they done so, any specific effect of the faecal enema may have been evident.

Although the best (but limited) evidence for the efficacy of this practice is in the treatment of C. difficile related diarrhoea, it has also been applied to other gastrointestinal diseases. Bennet and Brinkman (1989) reported a case of continuously active severe ulcerative colitis that had failed to respond to standard therapy with sulfasalazine and prednisone, but rapidly improved following a week of donor faecal enemas (18). Similarly Borody et al (2003) reported sustained remission of ulcerative colitis in 6 patients treated with faecal enemas (30).

In studies of faecal infusions the underlying mechanism that has been proposed to

mediate the effect of therapy is t~at the microbiota of the colon is altered. However, despite anecdotal evidence of the benefit of treatment in specific conditions there is little supportive data for this proposal. The aim of the current study was to determine

-210- whether the microbiota was altered by faecal infusions, using PCR-DGGE targeting major phylogenetic groups within the colon.

In addition to the 16S rRNA gene primers ofNubel et al (1996), a novel primer set for a second universal bacterial gene, the gene for the RNA polymerase B subunit (RPOB), was used (243). This was applied in order to corroborate the results . obtained with the primer set for the domain Bacteria. The principal advantage of the RPOB gene is that each bacterium has only one DNA copy as opposed to 16S rRNA where multiple copies and intraspecies heterogeneity can lead to multiple DGGE bands for a single species (66). In a preliminary study by Dahllof et al (2000) an initial set of primers were designed based on the sequences of only 4 bacterial species: E. coli, Bacillus subtilis, Staphylococcus aureus and H. pylori. Even so mismatches in the primer sequences (to target) had to be accepted (66). Subsequently the primer set was altered and updated by Dr Rebecca Case based on the 111 complete bacterial genome sequences available from the NCBI in July 2003 (www.ncbi.nlm.nih.gov/genomes/MICROBES/Complete.html). At the time of primer development there were only two complete genomes of bacteria from the three major phylogenetic groups of anaerobes in this database. These were Bacteroides thetaiotaomicron and Clostridium acetobutylicum. To date this primer set has not yet been published.

PCR-DGGE is particularly suited to the assessment of changes in the microbiota of adults because the banding patterns produced with primers for bifidobacteria and the domain Bacteria from faecal DNA template from a given individual do not change over time (291, 388). Thus in this study, the banding patterns generated from post­

procedure samples were compared with the pre-treatment sample of that subj~ct and a sample of infused stool. Using this approach, the theoretical basis for this alternative practice was critically examined for the first time with molecular methods.

-211- 7.2 METHODS

7.2.1 Subjects

Ten test subjects (6 males, 4 females, aged 22 to 57 years) who were to receive infusions of donated faeces were recruited into the study between October 2001 and Apri12003. The test subjects were adults who had not taken antibiotics for at least 2 months prior to the study. Demographic and clinical information was collected for all test subjects (Table 7.1). This included their age, sex, diagnosis and medications during the study period. Their clinical diagnoses were irritable bowel syndrome (N = 5), constipation (N = 4), Crohn's disease (N=1). Four males (aged 40 to 49) that acted as the source of the donated faecal suspensions - "source subjects" were also enrolled.

All of the subjects were enrolled and treated at the Probiotic Therapy Research Centre (PTRC) in Five Dock, Sydney, NSW. The collection of faecal samples was approved by the UNSW Human Research Ethics Committee with approval number HREC01256.

7 .2.2 Treatment protocol

7.2.2.1 Test subject preparation

The protocol for the treatment of each test subject was empirically determined by the alternative health provider and was not controlled by the investigators. Details of the regimens used to prepare each test subject are summarised in Table 7.1. In general test subjects were prepared with various combinations of oral vancomycin 500 mg bd, rifampicin 150 mg bd, doxycycline 50 mg bd or metronidazole 400 mg bd for 5 to 10 days. This was immediately followed by gut lavage using polyethylene glycol solution (Glycoprep™, Pharmatel, Hornsby, NSW) according to the manufacturer's instructions.

-212- 7.2.2.2 Preparation of the faecal suspensions from the source subjects

Faecal suspensions were prepared from fresh stool samples that were collected from the source subjects each morning and blended for 10 seconds with 250 ml of sterile normal saline. Mter completion of preparation of the test subjects with antibiotics and lavage, two to four hundred millilitres of this faecal suspension was infused daily into the test subjects colon for 5 to 15 days. The first infusion was administered into the caecum via a colonoscope, and subsequent infusions were given over 60 minutes via a nasojejunal tube, enema or combination thereof, at the discretion of the alternative health provider (details in Table 7.1). There was one exception to this general protocol- subject TS3 -for whom the faecal suspension was only administered by enema, that is, the subject did not have a colonoscopic infusion. The allocation of the source subject to a given test subject was at the discretion of the alternative health provider.

7.2.3 Sample collection

In order to assess the bacterial ecology of the colon pre- and post-procedure, faecal samples were collected from the test subjects at enrolment prior to the procedure, from the first stool after commencing polyethylene glycol solution and at 4, 8 and 24 weeks after the infusions were completed. The samples from each of these timepoints were subsequently described as T1, T2, T3, T4 and T5 respectively. Two samples were also collected from the source subjects. These were samples from the

faeces used in the first faecal infusion for each test subject (denoted S1) and at 4

weeks after the completion of the infusions (denoted S4weeks) for each test subject.

Faecal samples were collected in standard sterile containers and frozen at -20°C in the subjects' home freezers. The samples were transferred to the PTRC and again stored at -20°C prior to transport to the University of New South Wales. The transport of samples between locations was performed in less than one hour using an esky containing an ice brick.

-213- 7.2.4 DNA extraction and PCR-DGGE

Sample storage and batch processing for DNA extraction were performed as described in section 2.1.2.2, Chapter 2. PCR-DGGE was performed with domain Bacteria, Bacteroides-prevotella group, C. coccoides group, C. leptum subgroup, bifidobacterial and RNA polymerase p subunit primers. The conditions for PCR amplification and DGGE for each group are summarised in Tables 2.1, 2.2 and 2.3 of Chapter 2.

7.2.5 Assessment of the RNA polymerase f3 subunit PCR-DGGE

As the RNA polymerase p subunit primers ofDahllof et al (2000) (66) modified by Dr R. Case had not previously been applied to faecal samples the amplification of DNA from B. longum, B. cereus, E. limosum, L. salivarius, E. faecalis, F. mortiferum, E. coli , C. histolyticum, B. vulgatus, B. fragilis, L. acidophilus , C. leptum, C. nexile, D. desulfuricans and P. acnes was assessed at annealing temperatures between 40 and 50°C. PCR product from the samples that amplified were initially applied to DGGE with a 20 to 70% gradient of denaturants and this was empirically narrowed to optimise band separation.

7.2.6 Gel analysis

All of the samples from a given test subject (T1 to T5) and representative source

subject sample (S1) that were amplified with a given primer set were electrophoresed

on the same gel to facilitate analysis. In addition the S4 weeks sample was included on gels with primers from the domain Bacteria to confirm the findings of Zoetendal et al that the banding patterns were unchanging over time(388). The DGGE gels were stained with ethidium bromide and banding profiles converted into a binary array as described in Section 2.1.5, Chapter 2. The arrays were analysed using the PRIMER 5 Software package (PRIMER-E) and Bray-Curtis similarity for binary data.

-214- Dendrograms showing the relatedness of each sample were constructed using an arithmetic average algorithm of PRIMER 5.

7.2. 7 Statistical analysis

To determine whether the banding patterns of post-procedure samples were more like Tl or S1 the Bray-Curtis similarity measures for T3, T4 and T5 were compared with Tl and S1, pooled for each primer set and assessed for statistical significance using the Friedman test. The p value for statistical significance was 0.05. Post-hoc analysis of the significant results was performed with the Wilcoxon Matched Pairs Signed Ranks test to make direct comparisons of the results for each time point versus Tl and S1• For example, the similarities for T3 and Tl were compared with

T3 and S1 for each primer set. The significance level for the post-hoc comparisons was arbitrarily set at 0.01. Results of post-hoc tests between 0.01 and 0.05 were described as being of borderline significance.

To determine if the microbiota of the test subjects changed in species composition in follow up betweentimepoints T3, T4 and T5, the Bray-Curtis similarities for comparisons of each pair of time points (T3 vs T4, T4 vs TS and T3 vs T5) were pooled by primer set and assessed for significance with the Friedman test. Post hoc testing of significant results was performed with the Wilcoxon Matched Pairs Signed Ranks test.

Finally an estimate of the Bray-Curtis similarity for faecal samples from unrelated individuals was obtained for qualitative comparison with the above results. The similarity of unrelated samples was calculated by comparing the banding patterns of samples from different test subjects, test subjects to other source subjects, and source subjects to each other when these were represented on the same gel. This result was not used in any statistical analysis.

-215- 7.3 RESULTS

7 .3.1 Assessment of the RNA polymerase f3 subunit PCR-DGGE

Using the RPO~ primers at an annealing temperature of 50°C the amplification of genomic DNA from pure cultures was observed for B. longum, B. cereus, E. limosum, L. salivarius, E. faecalis, F. mortiferum, E. coli and C. histolyticum but not B. vulgatus, B. fragilis, C. leptum, L. acidophilus or P. acnes. At an annealing temperature of 40°C PCR products were generated for all of these bacteria except L. acidophilus and P. acnes. However at this annealing temperature a very large number of non-specific products were also generated. In order to amplify a broad range of templates and limit non-specific amplification products an annealing temperature of 45°C was used for subsequent experiments. The results with DNA template from the pure cultures at this annealing temperature are shown in Figure 7 .1. A PCR product of the correct size was produced for B. longum, B. cereus, E. limosum, L. salivarius, E.faecalis, F. mortiferum, E. coli, C. histolyticum, B. vulgatus, C. nexile and D. desulfuricans but not C. leptum, B. fragilis, L. acidophilus, or P. acnes at this temperature. Non-specific amplification products are evident with several of the templates (Figure 7.1). Subsequently, the gradient of denaturants required for efficient band separation was narrowed revealing an optimised denaturant gradient of20 to 65%. An example of the gel banding pattern

obtained with the RPO~ primer set from test subject TS 10 and source subject S2 is shown in Figure 7 .2. In general the gels obtained with this primer set revealed only a small number of poorly-defined bands in contrast to the clearly defined bands obtained with the other primer sets (see below).

7.3.2 Assessment ofthe test subjects' samples with RNA polymerase f3 subunit PCR-DGGE

Pooled similarity statistics for comparisons of the 4, 8 and 24 week test subject samples (T3, T4 and T5) compared with baseline (Tl) and the source subject first

-216- infusion sample (S1) are represented in Figure 7 .3a. The Bray-Curtis similarities for comparisons of T3, T4 and T5 to Tl were 33.1 ± 22.2, 29.1 ± 19.0 and 26.4 ± 21.3 respectively (Table 7.2). In contrast the similarities ofT3, T4 and T5 to S1 were 43.6 ± 28.5, 38.3 ± 23.5 and 44.3 ± 30.1. The similarity of unrelated samples was 23.2± 20.1. A trend is seen in Figure 7 .3a suggesting that the similarity measures for comparisons of the 4, 8 and 24 week samples with the source subject were greater than the test subject's baseline sample. However there was no significant difference 2 in the similarity measures for the comparisons to Tl and S1 by group (X 7.4, p = 0.19).

7.3.3 Assessment of the test subjects' samples with PCR-DGGEfor the domain Bacteria

An example of a gel obtained with the second universal primer set for Bacteria from test subject TS1 and source subject S1 is shown in Figure 7.4a. Pooled similarity

statistics for comparisons ofT3, T4 and T5 with T1 and S1are represented in Figure 7.3b. The Bray-Curtis similarities for comparisons ofT3, T4 and T5 to T1 were 56.5 ± 11.3, 55.8 ± 8.5 and 53.9 ± 6.6 respectively (Table 7.2). The similarities ofT3, T4

and T5 to S1 were 70.5 ± 10.8, 69.5 ± 11.3 and 72.1 ± 11. For unrelated samples the similarity was 51± 11.7. There was a significant difference in the similarity

2 measures for the comparisons to T1 and S1 by group (X = 14.7, p = 0.012). The results of post-hoc testing were of borderline significance for each timepoint (p values of 0.027, 0.027, and 0.02 at timepoints T3, T4 and T5 respectively, Table 7 .2). However Figure 7 .3b reveals a clear trend towards higher levels of similarity to

S1 than T1 at T3, T4 and T5.

7.3.4 Assessment ofthe test subjects' samples with PCR-DGGEfor Bifidobacterium

Using the primer set for bifidobacteria in PCR-DGGE resulted in a limited number of distinct banding profiles with most test and source subjects having bands in one or

-217- more of six distinct band positions. Figure 7 .4b shows the banding profiles for samples from TS 10 and S2 with the bifidobacteria primer set. Five of the six common band positions are illustrated in Lane 6 of Figure 7 .4b. Figure 7 .3c presents the pooled Bray-Curtis similarity measures for T3, T4 and T5 compared with T1 and

S1• In this Figure the pooled similarity measures for comparisons to the source subject infusion sample (Sr) are higher than the comparisons to the test subject baseline sample (T1) at each timepoint. The Bray-Curtis similarities for comparisons ofT3, T4 and T5 to T1 were 45.4 ± 28.3, 45.2 ± 27.3 and 43.9 ± 26.2 respectively

(Table 7.2). The similarities ofT3, T4 and T5 to S1 were 85.9 ± 15.3, 81.8 ± 11.9 and 76.8 ± 13.8. For unrelated samples the similarity was 45 ± 24.4. There was a

significant difference in the similarity measures for the comparisons to T1 and S1 by group (X2 = 21.1 p = 0.0008 Friedman test). Comparisons of each post-procedure sample with the test subject baseline and source subject infusion samples using the Wilcoxon matched-pairs signed-ranks test resulted in p-values ranging between 0.01 and 0.03 that were considered of borderline significance with reference to the arbitrary p-value of 0.01 (Table 7.2).

7.3.5 Assessment of the test subjects' samples with PCR-DGGEfor the Bacteroides-prevotella group

An example of a gel displaying the banding patterns obtained with the Bacteroides­ prevotella primer set is shown in Figure 7 .4c. The pooled similarity statistics for these comparisons are represented in Figure 7.3d. Using the Bacteroides-prevotella primers the Bray-Curtis similarities for comparisons ofT3, T4 and T5 to T1 were 44.8 ± 24.9, 41.1 ± 22.1 and 39.3 ± 20.9 respectively (Table 7.2). The similarities of

T3, T4 and T5 to S1 were 70.5 ± 24.1, 77.1 ± 15.5 and 81.4 ± 13.9. For unrelated samples the similarity was 37.1 ± 20. The similarity of the test subject samples at T3, T4 and T5 to the source subject infusion sample was higher than their similarity

to baseline. The Friedmans Test statistic and p-value for comparisons to T1 and S1 was 38.1, p < 0.0001. Post-hoc testing was significant for comparisons ofT4 and T5

to T1 and S1 (Table 7.2).

-218- 7.3.6 Assessment of the test subjects' samples with PCR-DGGEfor the C. coccoides group

An example of a gel obtained with the C. coccoides group primer set is shown in

Figure 7.4d and the overall results of comparisons to T1 and S1 in Figure 7.3e. The Bray-Curtis similarities for comparisons ofT3, T4 and T5 to T1 were 59.8 ± 13.1, 53.4± 12.3 and 54± 16.8 respectively (Table 7.2). The similarities ofT3, T4 and T5 to S1 were 74.1 ± 11.4, 77.3 ± 17.2 and 77.8 ± 15.5. The similarity of unrelated samples using the C. coccoides group primers was 53.6 ± 18.3. The Friedman test statistic was 17 .6, p =0.004 and post-hoc testing was significant for comparisons of

T4 to T1 and S1 (Table 7.2).

7.3.7 Assessment of the test subjects' samples with PCR-DGGEfor the C. leptum subgroup

Figure 7 .4e displays a gel for subject TS 10 using this primer set. The pooled similarity statistics for all subjects are represented in Figure 7 .3f. The Bray-Curtis similarities for comparisons of T3, T4 and T5 to T1 were 49.4 ± 24.2, 49.8 ± 24.1 and 44.2 ± 24.4 respectively (Table 7.2). The similarities ofT3, T4 and T5 to S1 were 69.3 ± 20.9, 67.9 ± 18.3 and 71.5 ± 18. The similarity of unrelated samples was

2 39.1 ± 19.4. Comparisons of T3,T4 and T5 to T1 and S1 were significant (X 21, p = 0.0008). Post-hoc testing did not reveal which comparisons were responsible for the significant result (Table 7.2).

Overall, the results obtained with all six primer sets suggest that after the procedure the banding patterns obtained with each of these defined primer sets was more like that of the source subject infusion sample, than the test subject's baseline.

-219- 7 .3.8 Assessment of the stability of the banding patterns post-procedure

The stability of the microbiota within the groups defined by the primer sets after the procedure was assessed by comparing the similarity of the banding patterns at T3, T4 and T5. Figure 7.5 displays the Bray-Curtis similarities for comparisons ofT3 vs T4, T3 vs T5 and T4 vs T5 by primer set. There was no significant difference in the similarity of T3 vs T4, T3 vs T5 and T4 vs T5 for all but one of the primer sets. The result of the Friedman test was X2 3.0, p = 0.22 for the Bacteroides-prevotella group primers; X2 3.0, p = 0.22 for c. coccoides group primers; X2 0.21, p = 0.90 for c. leptum subgroup primers; X2 1.7, p = 0.42 for Bifidobacterium primers and X2 1.6, p

= 0.44 for RPO~ primers. However with the domain Bacteria primer set a significant difference was observed for these comparisons (X2 9.9, p = 0.007). The similarity measures for comparisons of T3 vs T4, T4 vs T5 and T3 vs T5 were 92.8 ± 5.4, 88.4 ± 7.9 and 84.4 ± 7.9 respectively. Post hoc testing revealed a significant difference between T3 vs T4 compared with T3 vs T5 (p < 0.01) while the other comparisons were not significant. This suggests that using the domain Bacteria primer set, the similarity of T3 to T5 was lower than the similarity of T3 to T4 and T4 to T5. While not significant, a similar trend was seen in the results for most of the other primer sets with the exception of C. leptum subgroup (Figure 7 .5).

Dendrograms showing the similarity of samples from timepoints T1, T2, T3, T4, T5 for three test subjects and their respective source subject infusion sample (Sr ) are shown in Figure 7.6a,b and c. In general the post-procedure samples (T3, T4 and T5) were found to cluster with Sr rather than Tl.

-220- TABLE 7.1: Details of the test subjects enrolled in the study. Test Age& Diagnosis Medication Antibiotic preparation Method of administration of Source Subject Sex faecal sus~ension subject TS1 47F Diarrhoea 5 days vancomycin and 1st at colonoscopy S1 predominant IBS metronidazole then 4 enemas TS2 38M Diarrhoea 5 days vancomycin and 1st at colonoscopy S2 predominant IBS metronidazole then 4 nasojejunal infusions then 5 enemas TS3 34F Constipation Colchicine 5 days vancomycin and 15 enemas S3 Lactulose metronidazole Bisacodyl TS4 22F Constipation Colchicine 5 days vancomycin and 1st at colonoscopy S4 metronidazole then 9 enemas TS5 57M Crohn's colitis Mesalazine 10 days vancomycin 1st at colonoscopy S2 Azathioprine and metronidazole then 9 enemas TS6 38M IBS 5 days vancomycin and 1st at colonoscopy then S4 metronidazole 5 combined nasojejunal infusions + enemas TS7 51M IBS Pantoprazole, 5 days doxycycline and 1st at colonoscopy S2 Bismuth metronidazole then 9 enemas TS8 50M Constipation Esomeprazole 5 days doxycycline and 1st at colonoscopy S2 Aspirin rifampicin then 5 combined nasojejunal Atorvastatin infusions + enemas then Colchicine 5 enemas TS9 56M Diarrhoea Bismuth 5 days vancomycin and 1st at colonoscopy S4 predominant IBS Lisinopril metronidazole then 4 enemas TSlO 41 F Constipation Betainterferon 5 days vancomycin and 1st at colonoscopy S2 metronidazole then 9 enemas TABLE 7.2: Bray-Curtis similarities by primer set.

Primer set Pooled Bray-Curtis similarities for sample comparisons Friedman p-values of the Wilcoxon matched- (mean±SD) Test pairs signed-ranl':s test for comparisons with Tl and S r of Tl Sr Unrelated T3 T4 T5 T3 T4 TS T3 T4 TS samples RPOB 33.1 ± 29.1 ± 26.4± 43.6± 38.3 ± 44.3± 23.2 ±20.1 x2 =7.4 22.2 19 21.3 28.5 23.5 30.1 p = 0.19 domain Bacteria 56.5± 55.8± 53.9± 70.5± 69.5± 72.1 ± 51± 11.7 x2= 14.7 0.027 0.027 0.02 11.3 8.5 6.6 10.8 11.3 11 p = 0.012* Bacteroides- 44.8± 41.1± 39.3 ± 70.5± 77.1 ± 81.4± 37.1 ± 20 x2 = 38.1 0.01 0.004** 0.004** prevotella 24.9 22.1 20.9 24.1 15.5 13.9 p < 0.0001 * C. coccoides 59.8± 53.4± 54± 74.1 ± 77.3± 77.8 53.6 ± 18.3 x2 = 17.6 0.04 0.008** 0.03 13.1 12.3 16.8 11.4 17.2 15.5 p = 0.004* C.leptum 49.4± 49.8± 44.2± 69.3± 67.9± 71.5 39.1 ± 19.4 x2 =21 0.08 0.048 0.07 24.2 24.1 24.4 20.9 18.3 18 p = 0.0008* Bifidobacterium 45.4± 45.2± 43.9± 85.9± 81.8 ± 76.8± 45 ± 24.4 x2 = 21.1 0.03 0.01 0.03 28.3 27.3 26.2 15.3 11.9 13.8 p= 0.0008*

*significant at p < 0.05 level in the Friedman test. **significant at p < O.Ollevel in the Wilcoxon test. 1group comparisons with Wilcoxon only performed if Friedmans Test significant. Figure 7.1: PCR for pure bacterial cultures with the primers for RNA polymerase B subunit at an annealing temperature of 450C. Lane1: marker FN-1, Lane 2: negative control, Lane 3: C. leptum, Lane 4: C. nexile, Lane 5: C. histolyticum, Lane 6: B.jragilis, Lane 7: B. vulgatus, Lane 8: B. longum, Lane 9: B. cereus, Lane 10: L. acidophilus, Lane 11: L. alivarius, Lane 12: P. acnes, Lane 13: E. limosum, Lane 14: E. coli, Lane 15: F. mortiferum. PCR product of the correct size and non-specific products were also found with E. faecalis and D. desulfuricans (not shown). 1 2 3 4 5 6 7 8 9 10 11 12 13 1415

non-specific products

1 2 3 4 5 6

Figure 7.2: OOGE gel with the RNA polymerase B subunit primer set obtained from the samples for test subject TS10 and source subject S2. Lane 1: TS 10 baseline T 1, Lane 2: TS10 day of lavage T2, Lane 3: TS10 4 weeks T3, Lane 4: TS10 8 weeks T4, Lane 5: TS10 24 weeks T5, Lane 6:

Source subject S2, sample S1• Only a few bands were derived from each sample with this primer set and diffuse bands or "smearing" was common. Figure 7.3: Box and whiskers plots of the pooled Bray-Curtis similarities for comparisons of test subject samples at T3, T4 and

T5 to Tl and~ by primer set. In general the similarity to S1 is greater than to T 1. The comparisons for each primer set were statistically significant with the exception of RPO~ (Friedman test).

Figure 7 .Ja: RNA polymerase Jl Figure 7 .3d: Bacteroilk•-pre•otella group 100 T1 T1 s s 0 ] ·s iZi ! 13 T4 s....,~e

Figure 7 .3b: domain Bacteria Figure 7 .3e: Clostridium coccoides group

Tl T1 .. s 0 0 .. s ] ] ·s ~~ 8~ D~ ·a ~~ iZi iZi ~~ 1~

13 1lt TS 1J 1lt TS Sample Sample

Figure 7 .3c: Bifidobacterium Figure 7.3f: Clostridium kptum subgroup

T1 T1 s s

1J 1lt 13 T4 TS Sample Sample Figure 7.4: Examples of gels obtained with each primer set . Lane 1: baseline T1, Lane 2: day of lavage T2, Lane 3: 4 weeks T3, Lane 4: 8 weeks T4, Lane 5: 24 weeks T5, Lane 6: Source subject

S1, sample S1,and where shown Lane 7: sample S4weeks Figure 7.4a: DGGE gel with the domain Bacteria primer set obtained from the samples for test subject TS 1 and source subject S 1. 12 34 56 7 Figure 7.4d: DGGE gel with the C. coccoides group primer set obtained from the samples for test subject TS10 and source subject S2. 123 4 5 67

.... ~ _. • ~ Figure 7.4b: DGGE gel with the I l •" I. ·n ~ ' Bifidobacterium primer set obtained from f the samples for test subject TS 10 and ti,. ..- -" -· t!Jil source subject S2. -• - .. 123456 ' Figure 7.4e: DGGE gel with the C. leptum primer set obtained from the samples for test subject TS10 and source subject S2. 1 2 3 4 5 6

Figure 7.4c: DGGE gel with the Bacteroides­ prevotella primer set obtained from the samples for test subject TS 10 and source subject S2. 1 2 3 4 5 6 Figure 7.5: Box and whiskers plots of the Bray-Curtis similarities for comparisons of samples at post-procedure time points by primer set. In general the banding profiles were highly similar. There is a trend for T3 vs T5 (middle bar) to be lower than the other comparisons.

Bacteroides-prevotella group C. coccoides group ]I c a 0 ] ] ·e ~ ·e r Ci) r Ci)

C. leptum subgroup Bifidobacterium ~ r ~ ~ ~ ~

RPO~ domain Bacteria ~ c c ~ n ] ] ·e '6 Ci) ! ~ Ci) 'l'JI1I 'D'S 1m 13R 1315 1lrl5 Satpes cuopud &npes cmpn:d Figure 7.6a: Dendrograms for each primer set using the samples from TSl and the respective source subject Sl. Tl is marked with a red box; S1 with a black stripped box; and the post procedure samples T3, T4 and T5 with blue boxes

domain Bacteria Bacteroides-prevotella group

a 0 ·a 20 - 40 "§ ·- 60 1':1)&>

T2 T1 S S4 T5 T3 T4 • IS • • •

RPOfJ C. coccoides group

0 ;;... -~ 20 . :;::::: 40 , .§ 60 1':1) 11 80 , S T4 T3 T5 T2 S S4 Tl T4 T3 T5 IS ••• IS • •••

bifidobacteria C. leptum sub_group

0 a a 0 I ·a 20 ·a 20 :-;:::;40 :;::::: 40 .§ 60 1':1) .§ 60 1':1) 80 80 100 I I T2 T1 S4 T4 T5 T3 S T2 Tl s T5 S4 T3 T4 • • •• IS • IS • • • Fignre 7.6b: Dendrograms for each primer set using the samples from TS9 and the respective source subject S4. Tl is marked with a red box; S1 with a black stripped box; and the post procedure samples T3, T4 and T5 with blue boxes

domain Bacteria Bacteroides-prevotella group a o a o - ~ii2o ==40E Ill _ _ I fi)60 1:100 1 ,----.,;:=>---,. Ill 1 I T2 Tl S S4 T5 T3 T4 100 I I ===-- • lSJ ••• T2 Tl T3 s T4 T5 • · ~ • •

RPOf3 C. coccoides group ]- ~ 40~: 11 ~ rn : . I I 1 T2 Tl S T5 T3 T4 S T4 T5 T3 Tl T2 lSJ ••• • . ~ ...

bifidobacteria C. leptum subgroup

·; 2() , 0 0 I ffi)60..] o · I ~ 40 . fi) 60 ' ll) , ~ ~ 100 r=----'-1 T2 T5 T3 T4 Tl S T2 T1 S T5 T3 T4 ~ .... . ~ ... Figure 7 .6c: Dendrograms for each primer set using the samples from TSlO and the respective source subject S2. Tl is marked with a red box; S1 with a black stripped box; and the post procedure samples T3, T4 and T5 with blue boxes

domain Bacteria Bacteroides-prevotella group c o. ·a 20 . ==40e 0060 80 100, T2 T1 S S4 T5 T3 T4 100 • ISl ••• T2 T1 T4 T3 T5 S • • • • ISl

C. coccoides group

0 ,

T2 Tl S T5 T3 T4 • ISl •••

bifidobacteria C. leptum subgroup

o' o' c c 20' ·a 20 ·a :-::40e ·-a 40 ' iZi 60· iZi 60 ' 80 . 80 . J()() , T2 Tl T5 T4 T3 S T2 T1 S T5 T3 T4 • ••• ISl • ISl • • • 7.4 DISCUSSION

The is the first study to show conclusively that a novel microbiota, consisting of predominantly source subject bacterial species, developed after a procedure in which a faecal slurry from a source subject was infused into the caecum (and colon) of a test subject that had been prepared with antibiotics and cathartics. The change involved all of the bacterial groups for which assays were performed including the domain Bacteria, Bacteroides-prevotella group, C. coccoides group, C. leptum subgroup and bifidobacteria. A second important fmding is that this novel microbiota was reasonably stable in composition over the following 6 months.

The findings suggest that a majority of the species in the colon of the test subjects post-procedure, originated from the source subject infusions. This conclusion is supported by two sets of results. Firstly, for each primer set and each post-procedure time point (T3, T4 and T5), the Bray-Curtis similarity for comparisons to the source infusion sample were greater than for comparisons to the test subject baseline sample. These results are best illustrated in Figure 7 .3b to 7 .3f. These fmdings were statistically significant for 5 of the 6 primer sets (Table 7.2), the exception being the problematic primer set for RPO~ (discussed below). Post-hoc testing of individual time points was significant at the p < O.Ollevel for T4 and T5 with the Bacteroides­ prevotella group primers, and T4 with the C. coccoides group primers. Many more of the post-hoc comparisons were of borderline significance including all of the results with the domain Bacteria and bifidobacteria primers, T3 with the Bacteroides-prevotella group primers and T3 and T5 with the C. coccoides group primers. The effect of the procedure on groups of colonic bacteria that were not specifically targeted by the primer sets used such as lactobacilli and Enterobacteriaceae was not assessed.

The second observation that supports the view that a majority of the species in the colon of the test subjects post-procedure originated from the source infusion, is the finding that the Bray-Curtis similarity for comparisons of T3, T4 and T5 to Tl were

-230- only modestly greater than the comparisons for unrelated samples with each primer set. The similarity measure for unrelated samples, reflects the commonality of the species amplified by a given primer in different individuals. Previous studies using culture (226), PCR-DGGE (388) and PCR-cloning (154, 335) of human faecal samples have confirmed a moderate overlap in the bacterial species composition of faeces from different humans. For example, with the Bacteroides-prevotella primer set the similarity of unrelated samples was 37.1 ± 20, which was very similar to the results for comparisons ofTl to T3, T4 and T5 these being 44.8 ± 24.9, 41.1 ± 22.1 and 39.3 ± 20.9 respectively. With the primer set for the domain Bacteria the similarity of unrelated samples was 51± 11.7 compared with 56.5 ± 11.3, 55.8 ± 8.5 and 53.9 ± 6.6 for comparisons to Tl ofT3, T4 and T5 respectively. Thus post­ procedure a hybrid microbiota developed that consisted predominantly of source subject microbiota.

This hybrid microbiota appeared to be reasonably stable over the 24 weeks of follow up. The stability of the post-procedure bacterial populations over time was assessed by comparing the similarity of the 4, 8 and 24 week test subject s~ples to each other (Figure 7.5) i.e. T3 compared with T4 and T5, and T4 compared with T5. With the exception of the RPO~ primer set all of these comparisons revealed high degrees of similarity (>85% in a majority of cases) as shown in Figure 7.5. A trend towards lower similarities for comparisons ofT3 vs T5 compared with T3 vs T4 and T4 vs T5 is also evident in Figure 7.5. Thus there were larger differences between the banding patterns at T3 vs T5 (a period of20 weeks) as compared with the shorter time frames of 4 and 16 weeks for T3 vs T4 and T4 vs T5 respectively. This pattern of similarities would be consistent with a gradual change in the species composition of the colon post-procedure making it less similar to the S1 sample. This was most marked with the bifidobacteria primer set and to a lesser extent perhaps with the primer set for the domain Bacteria (Figure 7.3b and 7.3c). Statistical analysis confirmed that there were no significant differences between the post-procedure sample patterns with all of the primer sets except the domain Bacteria primers. With this primer set the result of the Friedman test was X2 9.9 with a p value of 0.007.

-231- Post-hoc testing of the results with the domain Bacteria primer set revealed a significant difference between T3 vs T4 and T3 vs T5, a result consistent with the hypothesis above. That is, the results suggest that the microbiota post procedure may have been gradually changing in contrast to the usual known stability of the human faecal microbiota of adults over time (291, 388). Such a change could reflect an adjustment of the novel microbiota to variation in the colonic environment over time. An example of this variation would be changes in the available substrate due to dietary intake.

In this study an effort was made to confirm the results of the universal domain Bacteria primer set with universal primers targeting a different gene. The gene for RNA polymerase was chosen as there is only a single copy of this gene in each bacterial cell. Like the other primer sets, after the procedure at T3, T4 and T5 a trend was noted towards a higher similarity to the source infusion sample than the test subject's baseline (Figure 7 .3a). However the results were not statistically significant. This may be due to the very broad range of similarities that were present between samples (Figure 7 .3a) as a result of the small number of discernible bands that were present (Figure 7 .2). When applied to human faecal DNA, the banding profiles generated with the RPOB primers generally resulted in far fewer bands per sample than the 16S rDNA primers (Figure 7.2). This is likely to reflect a lack of homology between the primer sequences and the sequences of many gut bacteria, due to the limited number of gut bacterial sequences used in the database for their design. In addition the similarity of banding patterns produced with unrelated samples using the RPOB primer set was low (23.2 ± 20.1) compared with the other primer sets (Table 7.2). One speculative explanation for this phenomenon is that bacteria that are transients within the colon (i.e. ingested or of upper gastrointestinal origin) are preferentially amplified by this primer set. This outcome could reflect the development of this primer set using RNA polymerase Bsequences from bacteria that are predominantly from aerobic or facultative environments.

-232- It is also interesting that a restricted band diversity was observed with the bifidobacterial primer set. That is, the bands obtained with DNA template from faecal samples were only seen in six specific gel positions. Five of the six common band positions observed with this primer set are illustrated in Lane 6, Figure 7 .4b. This finding of a small number of bands is consistent with the results of Satokari et al (2001) who produced 8 or 9 band positions from human faeces (291). This finding is likely to reflect the small number of different Bifidobacterium species that colonise the bowel of an individual host and the fact that PCR products derived from different species often denature in the same gel position with this and other primer sets (7, 291). Despite the limited band diversity, the results for this primer set did support the overall findings of the study, in that there was a statistically significant change in the bifidobacterial populations of the colon of the test subject to resemble that ofthe source infusion, after the procedure (Table 7.2).

What are the implications of these fmdings? It is clear that treatment with broad­ spectrum antibiotics followed by colonic lavage dramatically reduces the colonisation resistance of the human colon. This is not surprising in view of the known effects of antibiotics on colonisation resistance in humans and mice (174, 356-358), and the findings relating to a similar regimen in mice that were reported in Chapter 6. This effect is likely to relate directly to the antibacterial effect of antibiotics and the dramatic increase in gut transit associated with lavage.

Whether this method of reducing the colonisation resistance of the colon can or should be applied to increase the probability of colonisation by current strains of probiotic bacteria such as selected Lactobacillus and Bifidobacterium species is unclear. There are several differences between infusions of faecal suspension and the administration of monobacterial or oligobacterial cultures, that may impact on the probability of colonisation by introduced bacterial strains. For example, it may be that a microbiota containing organisms representing all of the common metabolic activities of the caecal microflora is much more likely to colonise than a single bacterial species. In addition whether non-human strains of gut bacteria or non-gut

-233- bacteria will colonise as effectively as human-adapted colonic bacteria is questionable. Large doses of orally administered probiotic bacteria often result in significant contemporaneous cultivable populations in human faeces suggesting that viable populations must pass through the entire gut (2, 68, 122, 123, 163, 165, 345). Therefore oral administration after preparation with antibiotics and lavage may result in a comparable rate of colonisation to their infusion at colonoscopy and avoid much of the risk and expense associated with the protocol at the PTRC. The benefits of introducing probiotic strains in this way would need to be shown to outweigh the risks and inconvenience of bowel preparation and treatment with broad-spectrum antibiotics to justify such an approach.

It is clear that colonisation resistance was severely impaired by the preparation that the test subjects underwent. If a less rigorous regimen is effective, then this would be beneficial if this type of procedure is clinically applied in the future. Such a regimen might be a much shorter course of a single antibiotic followed by catharsis. Furthermore, it may be that a single caecal infusion at a time when colonisation resistance is substantially lowered may be all that is required to change the colonic microbiota. This possibility is supported by the finding that a single day of treatment with faecal suspension produced a change in the microbiota of mice (Chapter 6). That is in humans, further infusions of faecal suspension by enema over one to two weeks may not be necessary to change the test subject's colonic microbiota. Certainly this would be a relatively easy question to answer with the methods applied in this chapter.

While it would have been interesting to examine the therapeutic efficacy of the alternative health treatment, this was not evaluated as part of the study for two reasons. Firstly only 10 subjects were enrolled in the study, and secondly a control group, placebo infusion and blinded assessment were not feasible. The inclusion of a control group is important because placebo responses may be very significant in irritable bowel syndrome sufferers (364).

-234- The most important question is whether the procedure that was studied in this chapter has any role in treating gut diseases at the present time. Clearly this is a matter of risk and benefit for individual patients with little clinical evidence available to guide decision making from the published uncontrolled case series (6, 18, 30, 33, 83, 141, 142, 251, 256, 306, 352). It would seem essential therefore, that any use of this technique be restricted to clinical trials with appropriate safety monitoring, so that meaningful data on outcomes and adverse effects is obtained. The key issues at present are safety and efficacy.

Most importantly the safety of changing the microbiota of the colon needs to be established over long periods of follow up. Such a monitoring process would ideally include assessment for the major adverse outcomes that could conceivably arise from this process - these might include infectious diseases, colitis, colonic neoplasia and other non-colonic diseases in which gut antigens or chronic inflammation may play a role, such as arthritis or biliary disease. Although significant adverse effects were not identified in previous studies, these have not included structured safety monitoring or large enough numbers of patients to draw meaningful conclusions.

For some researchers, the results of this study will be enough to stimulate trials of treatment for conditions such as, ulcerative colitis and Crohn's disease. Others would certainly be happier to know what type of microbiota was of benefit in these diseases, before embarking on such trials. At this time there is strong evidence that the microbiota of the bowel is involved in the pathogenesis of inflammatory bowel disease, but no clear documentation of whether any particular bacterial species or group of species play an aetiologic role. If the groups of bacteria that had an aetiologic role were known, then this procedure would have obvious merit in changing the bacterial populations to a less pathogenic microbiota. However, in the absence of this knowledge it is difficult to know whether changing the microbiota will be beneficial or harmful.

-235- In summary, it is clear that the administration of antibiotics and gut lavage dramatically reduces the colonisation resistance of the human colon. In addition, this is the first study to clearly demonstrate that the administration of a faecal suspension from another human subject can establish a novel hybrid microbiota if this material is administered into the caecum when colonisation resistance is critically impaired. The bulk of this new hybrid microbiota appears to originate from the source of the infusion, and is reasonably stable over a 24 week period. This approach may form the basis for novel treatments of diseases such as inflammatory bowel disease in future.

-236- CHAPTER 8: Comparison of clone libraries derived from studies of colonisation resistance in humans

8.1 INTRODUCTION

PCR-cloning can generate a species list that complements the results of other methods of assessing the bacterial populations present in faecal samples such as the PCR-DGGE that was applied in Chapter 7. PCR cloning and putative sequence identification by comparison with GenBank is the only method that generates a detailed species list for samples containing uncultivable bacterial isolates. As a result this PCR-cloning approach has been widely used in studies of veterinary and human gut ecology. Fpr example, clone libraries representing PCR amplified bacterial ribosomal DNA have been generated from human faeces (284 clones (335)) and lower bowel samples from a range of animals including pigs (4270 clones (192)) and mice (152 clones (288)). In addition, a European co-operative study is currently underway that aims to significantly increase the database of 16S rDNA sequences derived from the human colonic microbiota (The European Union Human Gut Flora Project (25)).

Despite the utility of the PCR-cloning approach the method is hampered by a number of significant biases that limit its reproducibility (377). The preferential amplification of a given genospecies during PCR appears to be the major problem that interferes with the production of clone libraries that qualitatively and quantitatively represent the bacterial sequences in a given sample. It has been hypothesised that preferential amplification of template may occur in the first 10 thermal cycles and is exaggerated during the exponential phase that follows (147, 341). Such bias can have a massive impact on clone library composition. For example, a study of duplicate PCR-clone libraries generated from a single DNA sample extracted from pig intestine found that genospecies from the Bacteroides group constituted 22.1% of one library, but were not found in the other (192).

-237- Despite this problem, PCR-cloning remains the most useful method for generating species lists that describe complex microbial communities.

Until recently, clone libraries were frequently compared qualitatively by observations of taxon frequency and there was no direct method of quantitatively comparing the sequences (28, 387). Statistical approaches to comparisons of taxon frequency in order to assess overall library relatedness were complicated, required very large numbers of sequenced clones to yield meaningful results, and did not give a single statistic for overall library comparisons (28). For example in Bonnet et al's recent comparison of 69 and 284 16S rDNA clones from the same human faecal sample obtained with 25 and 10 cycles ofPCR respectively, the number of clones within a given Operational Taxonomic Unit (OTU) required for statistical significance was calculated. This study showed that if an OTU was represented by one 25-cycle clone, then to achieve significance at least 20 clones were required in the 10-cycle library. Two clones within the one OTU in the 25-cycle library without a match were significant, while for the same comparisons in the 10-cycle library, at least 14 clones without a match were required. In view of the labour intensive nature of cloning and sequencing this requirement for very large clone libraries is prohibitive.

Singleton et al (2001) have published a method to quantitatively compare clone libraries using a computer-based analysis (324). Using their method the coverage of a clone library by a sample of sequences is calculated using Good's formula (129). That is, the coverage of the library equals the number of sequences the sample and library have in common, divided by the number of sequences in the library. Coverage is dependent on the size of the difference in sequence homology tolerated when defining a sequence as "common". Singleton et al defmed the percentage difference in sequence homology tolerated as "evolutionary distance" or D. As the degree of sequence difference or D allowed is increased, the level of coverage generally increases. This is graphically depicted as a coverage curve displaying coverage (on theY axis) versus evolutionary distance-D- (on the X axis).

-238- Using their methodology two libraries can be compared by calculating the difference between homologous and heterologous coverage curves at each D. The homologous coverage curve Cx for library X represents the coverage of the library by sequences within library X at each level of evolutionary distance between 0 and 0.5 in 0.01 increments. Similarly, the heterologous coverage curve CXY represents the coverage of library X by the sequences in library Y at each D. The square of the difference between the two curves at a given level of evolutionary distance is 11CXY i.e. 11Cxy =

2 (Cx-CXY) • A Cramer-von Mises test statistic 2:,11Cxy is generated by adding the 11CXY values at each D(178).

In their work Singleton et al applied a Monte Carlo test (177) procedure to determine if the observed test statistic was significant using a program called LIB SHUFF. In this program sequences are randomly shuffled between libraries X andY, and for the new libraries 11CXY at each D and :L.11Cxy are calculated. This process is repeated 1000 times and the values of 11Cxy at each D and L,/1CXY are ranked from lowest to highest. Libraries are reported to be significantly different if the observed L,/1CXY is greater than or equal to the 950th ranked :L.11Cxy generated during the Monte Carlo procedure. In addition, comparison of the observed 11Cxy to the 950th ranked 11CXY at each D provides information on the level at which the differences exist. For example, significant differences at D less than 0.1 (i.e. less than 10%} suggest that similar groups of bacteria are present in both libraries with variation at the species and genus level.

To demonstrate the usefulness of their method Singleton et al (324) compared libraries from grassland soils and bioreactors and showed that these were significantly different when compared using LIBSHUFF. Comparison of clone libraries taken from two different bioreactors i.e. phosphate-removing and non­ phosphate-removing bioreactors were however not found to be significantly different. Similarly, libraries from different grassland soils were not significantly different. The probability that two libraries from different sources would be correctly

-239- identified as different increased with increasing numbers of clones in each library. Soil and bioreactor clone libraries were consistently detected as different when 40 or 0 more sequences were present in each library. This analysis has subsequently been applied to clone library studies of the depth distribution of bacterial populations in a lake (159) and Antarctic ice (34) as well as comparisons of the bacterial populations of soil and earthworm casts (113). It should be borne in mind that the statistic reported in the LIBSHUFF analysis reflects the probability that the two libraries being compared are significantly different and does not account for variation in the libraries related to sampling or the biases of PCR-cloning.

In this chapter, PCR-cloning of 16S ribosomal DNA was performed to compare the species composition of faecal samples obtained from the first test subject and the corresponding source subject. An evaluation was also made of the LIB SHUFF program for the comparison of clone libraries.

-240- 8.2 METHODS

8.2.1 Subjects

The samples from the first patient TS 1 and the source of the faecal suspension S 1 were studied. Specifically, 16S rDNA from samples T1, T3 and T5 from TS1 and the faecal suspension sample (S1) from S 1 was amplified, cloned and sequenced.

8.2.2 DNA extraction and PCR-cloning

DNA extraction and Hot-start PCR with primers F27 and R1492 using 10 cycles of amplification was performed as outlined in Section 2.1.8, Chapter 2. For each sample six PCR reactions were pooled and purified. This product was ligated into the pGEM-T Easy Vector (Promega) and used to transform competent E. coli DH5a. White colonies growing on LB/ampicillin/IPTG/X-Gal plates were picked with a sterile 200 f..Ll pipette tip and subcultured. Subsequently the inserts were amplified by colony PCR with primers FpUC/M13 and RpUC/M13 as described in Section 2.1.8, Chapter 2. Inserts of the correct size (1.6 Kb) were sequenced with primers F27 and 356F.

8.2.3 Sequencing, identification of homologues and tree construction

As discussed in Section 2.1.8 of Chapter 2, the sequences were assembled and chimeric sequences and sequences less than 350 base pairs were discarded. Each remaining sequence was compared with the GenBank database (19). The number of operational taxonomic units present in the sequences from each sample was calculated using the defmition of an OTU as a cluster of 16S ribosomal DNA sequences that differed by 2% or less (335). An estimate of the coverage ofthe total number of amplifiable OTUs by the OTUs amplified from each sample was estimated by applying Good's formula (129). The total number of OTU's amplifiable from each sample was estimated with the Chao1 estimator (43). Tree

-241- diagrams showing the relatedness of the sequences within each library were generated with the BioManager interface (http://www.angis.org.au). The complete 16S ribosomal DNA sequences of named homologues of the sequences were downloaded from GenBank for use as references in phylogenetic trees. The sequences amplified from each sample were aligned with CLUSTAL W (348) and a distance matrix was calculated by DNADIST(91). Trees displaying the degree of homology of the sequences were produced with NEIGHBOR (91) and edited in Microsoft Powerpoint.

8.2.4 Quantitative comparison of clone libraries

Groups of sequences were compared using the method of Singleton et al (324). A multiple sequence alignment of the two sets of sequences to be compared was downloaded into Microsoft WORD(Microsoft) and the 5' and 3' ends removed such that a 600 to 700 base pair segment of complete sequence common to all of the sequences remained. Sequences that did not have complete sequence in this common region were not included in the analysis. The alignment was uploaded into BioManager and a distance matrix constructed with DNADIST (91) using the Jukes -Cantor method (167). A sample file was constructed per the author's instructions (http://www.arches.uga.edu/-whitman/description.html) and the program LffiSHUFF (http://www.arches.uga.edu/-whitmanllibshuff.html) was run. To assess the utility of LffiSHUFF with gut microbiota libraries, the odd and even numbered sequences from each sample library were compared with LffiSHUFF. The LffiSHUFF program was also applied to comparisons of each of the 4 complete sets of sequences.

8.2.5 Assessment of the effect of library size on LIBSHUFF comparisons

To evaluate the effect of library size (the number of clones) on significance level the program LffiSHUFF was altered so that random subsets of the clone libraries of a particular size could be compared and a significance level determined for each

-242- comparison (see Appendix A for program text in MacPerl- LIBSHUFFpvalues ). For each sub-library size, 20 p values were generated with LIBSHUFF and the mean p value was calculated. The clone libraries representing each sample were compared using this program.

-243- 8.3 RESULTS

Clone libraries were generated from the samples obtained from subjects TS 1 (samples T1, T3 and TS) and S1 (sample Sr). For each sample, the pooled product from six 10 cycle PCRs was used in ligation for each sample. In order to obtain a satisfactory number of clones two transformations (with the same ligation product) were required for the S1 and T1 samples. Table 8.1 summarises the number of sequences in each sample's library, the number of Operational Taxonomic Units, an estimate of coverage and the estimate of the total number of OTU' s amplifiable from the sample. Tables 8.2a, b, c and d detail the sequences contained within the four libraries. This table lists the sequence designation, sequence length, the closest matching sequence in GenBank and percent homology, and lastly the closest matching sequence in GenBank with a genus and species name and the percent homology.

The matching sequences with genus and species names from all libraries were pooled and aligned to produce a reference for the comparison of libraries representing each sample. These reference sequences were aligned with each sample library individually, and phylogenetic trees were produced. The four trees are shown in Figure 8.1a, b, c and d. On visual inspection the phylogenetic trees obtained with

T3 and S1 appear most similar, while the other trees appear different.

To assess the utility of LIBSHUFF for the comparison of gastrointestinal16S rDNA clone libraries, each sample library was divided into odd numbered and even

numbered sequences and compared. Table 8.3 displays the delta-C (L,~CXY) and p­ values for each comparison. None of the comparisons were significant.

Pairwise comparisons of the complete libraries representing each sample were also made with LIB SHUFF (324). Figures 8.2a to 8.2f show the homologous and heterologous coverage curves for library comparisons. The squared difference in the

coverage values (~CXY) and the 950th ranked ~Cxy are also shown. Table 8.4lists the

-244- delta-C (L,.L1CXY) and p-values for each comparison. In contrast to the results for the odd and even half libraries, all of these comparisons were significant.

The reliability of LIB SHUFF analysis for the examination of these clone libraries was then compared with the program LffiSHUFFpvalues. The results for comparisons of complete sample libraries are shown in Figure 8.3. Using a cut-off of all twenty p-values being less than 0.05, all of the libraries were consistently differentiated at random sub-library sizes containing 37 or more clones except for comparisons of libraries representing T3 and Sr where 59 or more clones were required.

-245- TABLES.l: Overview of the clone libraries. Sample Number Genbank Number Estimate Estimate Upper of accessions for ofOTU's of of total limit of Clones the sequences observed1 coverage2 OTU's3 total OTU'sin sample4 SI 76 AY343242-317 23 86% 36 57 Tl 74 A Y338285-358 23 84% 59 126 T3 79 A Y343094-172 29 78% 65 117 T5 69 AY343173-241 33 76% 54 83

1 An Operational Taxonomic Unit refers to sequences with less than or equal to 2% sequence difference (335). 2 Coverage was estimated with Good's formula (129). 3 This estimate was made with the Chaol estimator (43). 4 The upper limit was defined as twice the square root of the estimated variance in Chao1 plus the estimate of total OTU's (42). Table 8.2a: Clones present in the S1 library.

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology s'" 1 624 Uncultured bacterium adhufec13 AF132237 99 Faecalibacterium prausnitiii AJ413954 92 s.-2 712 Uncultured bacterium adhufec296 AF132258 99 Eubacterium desmolans L34618 96

S1-3 787 F. prausnitzii X85022 98 81"4 772 Clostridium lituseburense M59107 97 s.-5 853 Catenibacterium mitsuokai AB030221 97

S1-6 764 C. mitsuokai AB030221 97

S1-7 761 Uncultured bacterium adhufec296 AF132258 99 E. desmolans L34618 97

81-8 745 Eubacterium ramulus AJ011522 99 s.- 9 784 C. mitsuokai AB030221 98

S1-1o 780 C. mitsuokai AB030221 98

81-11 793 Uncultured bacterium ckncm312-B6-5 AF376216 98 F. prausnitzii AJ413954 92

8 1-12 746 C. mitsuokai AB030221 97

81-14 773 Eggerthella sp. MLG043 AF304434 89 s'" 15 783 Eubacterium biforme M59230 97 s.-16 758 C. mitsuokai AB030221 97 s'" 18 790 Uncultured bacterium (human infant) L37A AF253389 98 Ruminococcus lactaris L76602 97

81-19 766 E. biforme M59230 97 s.-2o 719 C. lituseburense M59107 97 s.- 21 810 Eubacterium formicigenerans L34619 98

81-22 820 Unidentified butyrate-producing bacterium A2-231 AJ270484 99

81-23 826 C. mitsuokai AB030221 98

81-24 703 E. formicigenerans L34619· 97

81-25 735 Unidentified butyrate-producing bacterium A2-231 AJ270484 99 Clostridium nexile X73443 95

81-26 741 Unidentified butyrate-producing bacterium A2-231 AJ270484 99 C. nexile X73443 95

81-27 676 Unidentified butyrate-producing bacterium A2-207 AJ270471 98 E. desmolans L34618 '97

81-29 804 Uncultured bacterium CB25 AB050851 99 F. prausnitzii X85022 92

S1-30 608 Ruminococcus obeum X85101 99 Table 8.2a: Clones present in the S1 library (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession .. homology match accession homology s,-31 763 Bacterium mpn-isolate group 1 9 AF357567 99 R. obeum X85101 99 s,-37 868 Uncultured bacterium CB25 AB050851 99 F. prausnitzii X85022 93 s,-4o 714 E. ramulus AJ011522 99 s,-41 848 Uncultured bacterium adhufec420 AF132271 99 R. gnavus X94967 96 s,-42 784 Butyrate-producing bacterium A2-165 AJ270469 98 F. prausnitzii X85022 97 s,-43 699 Unidentified butyrate-producing bacterium A2-231 AJ270484 99 C. nexi/e X73443 95 s,-44 595 E. ramulus AJ011522 99 Sr51 750 Uncultured bacterium A21 AF052418 100 Clostridium saccharolyticum Y18185 94 s,-52 769 E. biforme M59230 97 Sr53 737 Unidentified butyrate-producing bacterium A 1-86 AJ270475 99 Eubacterium rectale L34627 98 Sr54 777 E. biforme M59230 97 Sr55 695 E. ramulus AJ011522 99 s,-56 558 C. mitsuokai AB030221 97 s,-57 678 C. mitsuokai AB030221 97 s,-58 591 Uncultured bacterium (human infant) L37A AF253389 97 Ruminococcus torques L76604 97 Sr60 496 Uncultured bacterium adhufec296 AF132258 99 E. desmolans L34618 95 s,- 61 658 Unidentified butyrate-producing bacterium A 1-86 AJ270475 99 E. rectale L34627 98 s,-62 502 C. mitsuokai AB030222 97 s,-63 725 C. mitsuokai AB030221 97 s,-64 384 B. adolescentis clone nru-1 AF275881 99 Sr65 707 E. formicigenerans L34619 97 Sr66 677 Unidentified butyrate-producing bacterium A2-231 AJ270484 96 E. formicigenerans L34619 95 s,-68 663 C. mitsuokai AB030221 97 s,-69 707 Unidentified butyrate-producing bacterium A2-231 AJ270484 99 R. torques L76604 95 Sr70 710 Unidentified butyrate-producing bacterium A 1-86 AJ270475 99 E. rectale L34627 98 s,-71 739 Uncultured bacterium CB25 AB050851 99 F. prausnitzii X85022 92 Table 8.2a: Clones present in the S1 library (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology 81-12 437 Uncultured bacterium adhufec13 AF132237 100 F. prausnitzii X85022 93 81-73 747 Uncultured bacterium CB25 AB050851 99 F. prausnitzii X85022 92 8~-74 712 B. ado/escentis clone nru-5 AF275882 100 81-75 736 Uncultured bacterium CB25 AB050851 99 F. prausnitzii X85022 92 81-76 728 Uncultured bacterium CB25 AB050851 99 F. prausnitzii AJ413954 92 8~-77 717 Uncultured bacterium adhufec335 AF132262 94 Clostridium populeti X71853 93 8~-79 390 Unidentified butyrate-producing bacterium L2-21 AJ270477 99 E. rectale L34627 97 81·80 675 Unidentified butyrate-producing bacterium A 1-86 AJ270475 99 E. rectale L34627 98 81-81 728 Bacterium mpn-isolate group 1 9 AF357567 96 R. obeum X85101 96 81-83 756 C. mitsuokai AB030221 97 8~-84 623 C. mitsuokai AB030222 97 81·85 642 C. mitsuokai AB030221 97 81-86 722 Unidentified butyrate-producing bacterium A 1-8 6 AJ270475 99 E. rectale L34627 97 8r87 728 C. mitsuokai AB030221 96 8r88 739 C. mitsuokai AB030221 98 81·89 727 Uncultured bacterium CB25 AB050851 98 F. prausnitzii AJ413954 93 81-90 690 Clostridium methy/pentosum Y18181 91 81-91 715 Uncultured bacterium adhufec13 AF132237 99 F. prausnitzii AJ413954 92 81·92 675 Uncultured bacterium CB25 AB050851 98 F. prausnitzii AJ413954 92 81-93 703 C. mitsuokai AB030221 97 81·94 710 Unidentified butyrate-producing bacterium A 1-86 AJ270475 99 E. rectale L34627 98 8r95 676 C. mitsuokai AB030221 97 8~-96 400 Unidentified butyrate-producing bacterium L2-21 AJ270477 99 E. recta/e L34627 98 81·97 708 Clostridium co/inurn X76748 96 8~-98 685 Uncultured bacterium (human infant) L37 A AF253389 98 R. /actaris L76602 97 81·99 720 E. formicigenerans L34619 97 81-1 oo 681 Coriobacterium sp. strain CCUG 33918 AJ131150 100 Table 8.2a: Clones present in the S1 library (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology s,-1 01 583 C. mitsuokai AB030222 97 Sr 102 688 C. mitsuokai AB030221 97 s,-1 o3 738 Uncultured bacterium adhufec335 AF132262 94 Eubacterium eligens L34420 93 Table 8.2b: Clones present in the Tllibrary.

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession ' homology match accession homology T1-3 622 Uncultured bacterium (human infant) L37 A AF253389 97 R. torques L76604 97 T1-4 390 Clostridium ramosum M23731 99 T1-5 780 Unidentified butyrate-producing bacterium A2-194 AJ270474 99 Eubacterium oxidoreducens AF202259 95 T1-6 738 Bifidobacterium dentium 086183 99 T1-7 721 Unidentified butyrate-producing bacterium L 1-83 AJ270474 99 E. oxidoreducens AF202259 95 T1-8 430 Uncultured bacterium (human infant) L37A AF253389 99 R. lactaris L76602 96 T1-9 751 Uncultured bacterium adhufec35.25 AF153853 96 Ruminococcus schinkii RSPBIE16 95 T1-10 755 Uncultured bacterium A21 AF052418 99 C. saccharolyticum Y18185 94 T1-11 731 Clostridium propionicum X77841 92 T1-12 729 Uncultured bacterium (human infant) L37A AF253389 97 R. lactaris L76602 97 T1-13 729 Ruminococcus productus X94966 96 T1-14 743 Unidentified butyrate-producing bacterium L 1-83 AJ270474 99 E. oxidoreducens AF202259 95 T1-15 658 Uncultured bacterium (human infant) L37A AF253389 96 R. lactaris L76602 95 T1-16 400 Holdemania filiformis Y11466 92 T1-18 750 Uncultured bacterium A21 AF052418 99 C. saccharolyticum Y18185 94 T1-19 453 Bacterium mpn-isolate group 1 9 AF357567 96 R. obeum X85101 95 T1-20 752 Uncultured bacterium (human infant) L37A AF253389 97 R. lactaris L76602 97 T1-21 748 Clostridium clostridiiformes M59089 99 T1-22 754 Bacterium mpn-isolate group 1 8 AF357566 98 R. obeum X85101 96 T1-23 721 Uncultured bacterium adhufec35.25 AF153853 96 R. schinkii X94964 95 T1-24 772 Unidentified butyrate-producing bacterium A2-194 AJ270473 99 E. oxidoreducens AF202259 95 T1-28 506 Uncultured bacterium A21 AF052418 99 C. saccharolyticum Y18185 94 T1-29 705 Unidentified butyrate-producing bacterium L 1-83 AJ270474 99 E. oxidoreducens AF202259 95 T1-30 768 Uncultured bacterium A21 AF052418 99 C. saccharolyticum Y18185 94 T1-31 679 Unidentified butyrate-producing bacterium L 1-83 AJ270474 99 E. oxidoreducens AF202259 94 T1-32 415 Uncultured bacterium A54 AF052421 98 C. saccharolyticum Y18185 95 T1-33 370 Uncultured rumen bacterium 4C2Bd-12 AB034126 94 R. flavefaciens AF104840 91 Table 8.2b: Clones present in the Tllibrary (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology T1-34 410 Uncultured rumen bacterium 4C28d-12 AB034126 94 R. f/avefaciens AF104840 91 T1-35 390 R. bromii X85099 99 T1-36 736 B. dentium 086183 98 T1-37 761 Uncultured bacterium adhufec85 AF132285 99 Eubacterium e/igens L34420 98 T1-38 730 Bacteroides uniformis AB05011 0 99 T1-39 748 Eubacterium contortum L34615 97 T1-40 780 Uncultured bacterium A 71 AF052423 96 Clostridium indo/is Y18184 95 T1-42 695 Unidentified butyrate-producing bacterium L 1-83 AJ270474 99 E. oxidoreducens AF202259 94 T1-43 720 Unidentified butyrate-producing bacterium L 1-83 AJ270474 99 E. oxidoreducens AF202259 95 T1-44 728 Butyrate-producing bacterium A2-165 AJ270469 99 F. prausnitzii X85022 98 T1-45 464 Uncultured bacterium (human infant) L37A AF253389 98 Ruminococcus torques L76604 97 T1-46 722 R. bromii X85099 99 T1-47 713 Uncultured bacterium (human infant) L37A AF253389 97 R. lactaris L76602 96 T1-48 773 Bifidobacterium dentium 086183 99 T1-49 788 Uncultured bacterium adhufec80.25 AF153858 97 R. /actaris L76602 96 T1-50 710 Uncultured rumen bacterium 4C28d-12 AB034126 94 C. cel/u/olyticum X71847 90 T1-51 737 Uncultured bacterium A21 AF052418 99 C. saccharolyticum Y18185 95 T1-52 407 Unidentified butyrate-producing bacterium L 1-81 AJ270480 94 Syntrophococcus sucromutans Y18191 93 T1-53 707 R. bromii X85099 99 T1-54 756 C. co/inurn X76748 96 T1-55 736 R. gnavus X94967 99 T1-56 757 Lachnospira pectinoschiza L14675 92 T1-57 370 Uncultured rumen bacterium 4C28d-12 AB034126 94 R. f/avefaciens AF104840 91 T1-58 730 C. spiroforme X75908 99 T1-59 747 R. bromii X85099 99 T1-60 729 R. productus X94966 96 Table 8.2b: Clones present in the Tllibrary (continued).

Clone length Closest GenBank sequence match GenBank· % Closest named GenBank GenBank % accession homology match accession homology T1-61 725 Uncultured bacterium (human infant) L37A AF253389 97 R. /actaris L76602 97 T1-62 723 Uncultured bacterium (human infant) L37A AF253389 98 R. /actaris L76602 97 T1-63 410 Bacterium mpn-isolate group 1 8 AF357566 99 R. schinkii X94964 97 T1-65 735 Unidentified butyrate-producing bacterium L 1-83 AJ270474 99 E. oxidoreducens AF202259 95 T1-66 708 R. gnavus L76597 99 T1-67 720 R. productus X94966 96 T1-68 771 Unidentified butyrate-producing bacterium A2-194 AJ270473 99 E. oxidoreducens AF202259 95 T1-69 729 Uncultured bacterium adhufec80.25 AF153858 97 R. /actaris L76602 97 T1-70 395 Uncultured bacterium adhufec217 AF132245 99 C. clostridiiformes M59089 95 T1-71 738 R. productus X94966 96 T1-73 745 R. productus X94966 96 T1-74 742 C. c/ostridiiformes M59089 99 T1-75 534 Uncultured bacterium (human infant) L37A AF253389 97 R. torques L76604 97 T1-76 737 Unidentified butyrate-producing bacterium L 1-83 AJ270474 99 E. oxidoreducens AF202259 95 T1-77 400 C. spiroforme X75908 99 T1-79 723 R. bromii X85099 99 T1-80 828 Uncultured bacterium adhufec80.25 AF153858 97 R. /actaris L76602 96 T1-81 805 E. formicigenerans L34619 97 T1-82 817 Uncultured bacterium A 11 AF052412 98 E. oxidoreducens AF202258 94 T1-83 726 Uncultured bacterium (human infant) L37A AF253389 98 R. lactaris L7.6602 97 T1-84 731 B. dentium 086183 99 T1-85 720 Uncultured bacterium (human infant) L37A AF253389 97 R. /actaris L76602 97 T1-86 771 R. torques L76604 96 T1-87 743 Uncultured bacterium adhufec25 AF132254 99 Clostridium indo/is AF028351 94 T1-88 750 R. bromii X85099 99 T1-89 783 Uncultured bacterium A 11 AF052412 98 E. oxidoreducens AF202258 94 Table 8.2b: Clones present in the Tllibrary (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology T1-90 732 Uncultured bacterium A 11 AF052412 97 E. oxidoreducens AF202258 95 T1-91 796 Unidentified butyrate-producing bacterium A2-194 AJ270473 99 E. oxidoreducens AF202259 95 T1-92 320 Clostridium sp. Y10584 95 T1-93 708 R. productus X94966 96 T1-94 762 R. bromii X85099 99 T1-95 746 Uncultured bacterium (human infant) L37A AF253389 97 R. /actaris L76602 97 T1-96 705 Uncultured bacterium (human infant) P36G AF253346 99 R. obeum X85101 96 T1-97 784 Unidentified butyrate-producing bacterium A2-194 AJ270473 99 E. oxidoreducens AF202259 95 T1-98 770 R. bromii X85099 99 T1-99 757 C. clostridiiformes M59089 99 Table 8.2c: Clones present in the T3library.

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology T3- 1 709 Unidentified butyrate-producing bacterium A2-231 AJ270484 99 C. nexile X73443 95 T3- 2 651 Butyrate-producing bacterium A2-165 AJ270469 99 F. prausnitzii AJ413954 97 T3- 3 546 Uncultured bacterium adhufec168 AF132242 99 Clostridium sporosphaeroides M59116 91 T3- 4 465 Uncultured bacterium adhufec13 AF132237 99 F. prausnitzii AJ413954 94 T3- 5 697 Uncultured bacterium clone p-2679-65A5 AF371811 97 Clostridium methylpentosum Y18181 86 T3- 6 675 Uncultured bacterium adhufec13 AF132237 99 F. prausnitzii X85022 92 T3- 7 669 Uncultured bacterium adhufec13 AF132237 99 F. prausnitzii X85022 92 T3- 8 543 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 T3- 9 697 Uncultured Streptococcus sp. clone KL-27-1-5 AF408263 100 Streptococcus thermophilus X68418 99 T3- 10 699 Uncultured bacterium adhufec150 AF132238 99 Roseburia intestinalis AJ312385 98 T3- 11 633 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 T3- 12 666 Uncultured bacterium clone p-2559-9F5 AF371715 99 F. prausnitzii X85022 94 T3- 13 694 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 T3- 14 664 Uncultured bacterium adhufec113 AF132236 99 F. prausnitzii AJ413954 98 T3- 15 721 Uncultured bacterium adhufec25 AF132254 99 C. indo/is AF028351 93 T3- 16 690 Butyrate-producing bacterium A2-165 AJ270469 99 F. prausnitzii AJ413954 98 T3- 17 694 Uncultured bacterium A20 AF052417 99 R.obeum X85101 96 T3- 18 729 Ruminococcus cal/idus L76596 99 T3- 19 682 Butyrate-producing bacterium A2-165 168 AJ270469 98 F. prausnitzii AJ413954 97 T3- 20 711 Uncultured bacterium clone p-2559-9F5 AF371715 100 F. prausnitzii X85022 92 T3- 21 723 Uncultured bacterium clone p-2053-s959-5 AF371800 98 Oscil/ospira guillermondii AB040497 94 T3- 22 710 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 T3- 23 675 Uncultured bacterium clone p-2197 -s959-3 AF371794 97 Termitobacter aceticus Z49863 94 T3- 24 677 Uncultured bacterium clone p-595-a5 AF371830 97 Acidaminobacter AF016691 92 hydrogenoformans T3- 26 719 Uncultured bacterium clone p-3567-9F3 AF371775 99 C. sporosphaeroides M59116 90 T3- 28 647 Uncultured bacterium adhufec250 AF132253 97 C. clostridiiformes M59089 94 Table 8.2c: Clones present in the T3 library (continued).

Clone length Closest GEm Bank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology T3- 31 704 Uncultured Bifidobacterium sp. 150 AF275886 99 Bifidobacterium ruminantium 086197 97 T3- 32 625 Uncultured bacterium clone p-392-o3 AF371512 99 E. biforme M59230 97 T3- 33 705 Uncultured bacterium clone p-2559-9F5 AF371715 99 F. prausnitzii X85022 92 T3- 34 610 Uncultured Eubacterium sp. clone LabF83 AF335908 99 E. biforme M59230 97 T3- 35 260 Uncultured bacterium clone p-4205-6Wa5 AF371520 . 98 C. mitsuokai AB030222 96 T3- 36 732 Uncultured bacterium clone p-4205-6Wa5 AF371520 99 C. mitsuokai AB030221 97 T3- 39 407 Uncultured bacterium clone p-3024-SwA5 AF371827 96 R. f/avefaciens AF104836 90 T3- 40 691 Uncultured bacterium clone HuCA 10 AJ409008 99 F. prausnitzii AJ413954 98 T3- 42 724 Uncultured bacterium clone p-392-o3 AF371512 99 E. biforme M59230 97 T3- 43 724 Uncultured bacterium (human infant) P36G AF253346 99 R. obeum X851 01 95 T3- 44 410 Swine fecal bacterium FPC63 AF445206 98 C. sporosphaeroides M59116 88 T3- 45 778 Uncultured bacterium clone p-1735-b3 AF371530 94 Paenibacil/us sp. AJ297715 84 T3- 46 793 Uncultured bacterium clone p-2168-959-3 AF371571 98 Clostridium fimetarium AF126687 92 T3- 47 735 Uncultured bacterium clone p-2168-959-3 AF371571 93 E. eligens L34420 92 T3- 48 725 Uncultured bacterium clone p-392-o3 AF371512 99 E. biforme M59230 97 T3- 49 707 Uncultured bacterium clone p-596-a5 AF371595 99 E. formicigenerans L34619 97 T3- 50 684 Eubacterium cylindroides L34617 98 T3- 51 733 Uncultured Streptococcus sp. clone KL-48-1-4 AF408260 99 S. thermophilus X68418 99 T3- 52 724 Uncultured bacterium clone p-392-o3 AF371512 99 E. biforme M59230 97 T3- 53 575 Uncultured bacterium clone p-2571-9F5 AF371719 97 F. prausnitzii AJ413954 <90 T3- 55 661 Uncultured bacterium adhufec61.25 AF153867 98 Prevotella ou/ora L16472 93 T3- 56 729 Uncultured Streptococcus sp. clone KL-48-1-4 AF408260 99 S. thermophilus X68418 99 T3- 57 730 Uncultured Streptococcus sp. clone KL-27-1-5 AF408263 99 S. thermophilus X68418 99 T3- 58 741 Uncultured Streptococcus sp. clone KL-48-1-4 AF408260 99 S. thermophilus X68418 99 T3- 59 708 Uncultured Streptococcus sp. clone KL-27-1-5 AF408263 99 S. thermophilus X68418 99 T3- 60 716 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 Table 8.2c: Clones present in the T3 library (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology T3- 61 723 Uncultured bacterium clone cadhufec030h7 AF530318 99 F. prausnitzii AJ413954 93 T3- 62 654 Uncultured bacterium clone p-2406-55G5 AF371824 93 C. ce/lulo/yticum X71847 88 T3- 63 748 Uncultured bacterium clone p-392-o3 AF371512 99 E. biforme M59230 97 T3- 64 695 Uncultured bacterium clone p-2617-9F5 AF371717 99 F. prausnitzii X85022 <90 T3- 65 724 Uncultured bacterium clone p-4205-6Wa5 AF371520 99 C. mitsuokai AB030221 98 T3- 66 678 Uncultured bacterium clone p-4205-6Wa5 AF371520 99 C. mitsuokai AB030221 98 T3- 67 719 Uncultured bacterium clone p-1877 -s962-3 AF371805 99 Papillibacter cinnaminovorans AF167711 90 T3- 68 689 Uncultured Streptococcus sp. clone KL-27-1-5 AF408263 100 S. thermophi/us X68418 99 T3- 69 711 Uncultured bacterium CB25 AB050851 98 F. prausnitzii X85022 <90 T3- 70 694 Uncultured bacterium clone p-4205-6Wa5 AF371520 99 C. mitsuokai AB030221 98 T3- 71 714 Uncultured bacterium clone p-1877 -s962-3 AF371805 99 P. cinnaminovorans AF167711 90 T3- 72 700 Uncultured Streptococcus sp. clone KL-27-1-5 AF408263 100 S. thermophi/us X68418 99 T3- 73 620 Uncultured bacterium CB25 AB050851 97 F. prausnitzii X85022 <90 T3- 74 746 Uncultured bacterium clone p-5228-4Wb5 AF371943 97 F. prausnitzii FPR413954 92 T3- 75 674 Uncultured bacterium adhufec405 AF132268 98 C. xy/ano/yticum X71855 94 T3- 76 723 Bacteroides merdae X83954 99 T3- 77 603 R. callidus L76596 99 T3- 78 565 Uncultured Eubacterium sp. clone LabF83 AF335908 99 E. biforme M59230 97 T3- 79 626 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 T3- 80 727 Uncultured bacterium clone p-392-o3 AF371512 99 E. biforme M59230 97 T3- 81 764 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 T3- 82 709 Uncultured bacterium clone p-987 -s962-5 AF37191 0 90 B. distasonis M86695 87 T3- 83 771 Uncultured bacterium clone p-596-a5 AF371595 99 E. formicigenerans L34619 97 T3- 84 722 Uncultured bacterium clone p-596-a5 AF371595 99 E. formicigenerans L34619 97 T3- 85 608 Uncultured bacterium clone p-2617-9F5 AF371717 99 F. prausnitzii X85022 <90 T3- 86 723 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 T3- 87 726 Uncultured bacterium clone p-4205-6Wa5 AF371520 98 C. mitsuokai AB030221 97 Table 8.2c: Clones present in the T3 library (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology T3- 88 617 Uncultured bacterium clone HuCB29 UEU409002 99 F. prausnitzii FPR413954 98 T3- 89 724 Uncultured bacterium A20 AF052417 99 R. obeum L76601 96 T3- 90 729 Uncultured Streptococcus sp. clone KL-48-1-4 AF408260 99 S. thermophilus X68418 99 T3- 91 722 Uncultured Streptococcus sp. clone KL-48-1-4 AF408260 99 S. thermophilus X68418 99 Table 8.2d: Clones present in the T5 library.

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology T5- 1 350 Uncultured bacterium adhufec406 AF132269 97 Clostridium aminophilum L04165 94 T5- 2 791 Uncultured rumen bacterium 5COd-12 AB034045 94 Clostridium methy/pentosum Y18181 88 T5- 3 360 Uncultured bacterium ckncm143-F1 M.1 E3 AF376376 97 B. distasonis M86695 97 T5- 4 310 Uncultured bacterium clone L 1 0-6 AJ400275 98 Verrucomicrobium spinosum X90515 92 T5- 5 420 F. prausnitzii AJ413954 99 T5- 6 371 Uncultured rumen bacterium 4COd-9 AB034022 96 P. cinnaminovorans AF167711 93 T5- 7 721 Uncultured bacterium adhufec55 AF132275 99 B. merdae X83954 99 T5- 8 752 Bacteroides caccae X83951 88 T5- 9 405 uncultured eubacterium A 16-K1 AJ405026 99 S. thermophi/us X68418 98 T5- 10 792 Butyrate-producing bacterium A2-165 AJ270469 99 F. prausnitzii X85022 98 T5- 11 758 Uncultured bacterium A21 AF052418 99 C. saccharolyticum Y18185 94 T5- 12 765 Eggerthe/la /enta AF292375 90 T5- 13 723 Uncultured bacterium adhufec29 AF132258 99 E. desmolans L34618 96 T5- 14 450 Uncultured bacterium adhufec27 AF132256 99 Bacteroides vulgatus AB050111 99 T5- 15 740 B. caccae X83951 88 T5- 16 784 Uncultured bacterium adhufec61.25 AF153867 99 P. ou/ora L16472 92 T5- 17 580 Bacteroides thetaiotaomicron AB050109 96 T5- 18 778 Uncultured bacterium adhufec406 AF132269 99 Clostridium xy/anolyticum X71855 93 T5- 19 691 Uncultured bacterium adhufec81 AF132280 98 Eubacterium siraeum 168 L34625 97 T5- 20 502 Bilophila wadsworthia U82813 99 T5- 21 773 Unidentified eubacterium from anoxic bulk soil AJ229190 92 Tindallia magadii Y15626 91 T5- 22 721 Uncultured rumen bacterium 5COd-12 AB034045 95 R. flavefaciens AF104846 87 T5- 23 593 Prevotella oralis L16480 90 T5- 25 695 Uncultured bacterium partial AJ400275 99 Candidatus AF217461 88 Xiphinematobacter rivesi T5- 26 410 Uncultured bacterium adhufec61.25 AF153867 99 P. oulora L16472 94 T5- 27 710 Uncultured bacterium adhufec406 AF132269 99 C. xy/anolyticum X71855 95 Table 8.2d: Clones present in the TSlibrary (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession. homology match accession homology T5- 28 400 Uncultured bacterium adhufec21 AF132247 91 R. bromii X85099 90 T5- 29 782 Uncultured bacterium adhufec367 AF132266 99 Bacteroides ovatus X83952 93 T5- 30 760 B. caccae X83951 89 T5- 31 698 Unidentified butyrate-producing bacterium L2-21 AJ270477 99 E. recta/e L34627 98 T5- 32 744 Uncultured Streptococcus sp. clone KL-48-1-4 AF408260 99 S. thermophilus X68418 99 T5- 33 702 Uncultured bacterium adhufec81 AF132280 98 E. siraeum L34625 98 T5- 34 532 Uncultured bacterium clone L 10-6. AJ400275 99 Candidatus X. rivesi AF217461 88 T5- 35 715 Uncultured Streptococcus sp. clone KL -48-1-4 AF408260 99 S. thermophi/us X68418 99 T5- 36 389 Uncultured bacterium clone L 1 0-6 AJ400275 98 V. spinosum X90515 92 T5- 37 744 Uncultured bacterium A21 168 AF052418 99 C. saccharo/yticum Y18185 94 T5- 38 735 E. ramulus AJ011522 99 T5- 39 328 Prevotella sp. oral strain 831 FD AY005061 91 T5- 40 572 Prevote/la sp. oral strain 831 FD AY005061 90 T5- 41 719 Eggerthella sp. MLG043 AF304434 90 T5- 42 759 Uncultured bacterium A21 AF052418 98 C. saccharo/yticum Y18185 94 T5- 44 778 Bacteroides sp. AR20 AF139524 99 B. uniformis L16486 95 T5- 45 760 Uncultured bacterium adhufec27 AF132256 99 B. vulgatus A8050111 99 T5- 46 739 Uncultured rumen bacterium 5COd-12 A8034045 95 R. f/avefaciens AF030450 87 T5- 47 741 Uncultured rumen bacterium 3COd-3 A8034003 93 Ruminococcus a/bus AF104833 91 T5- 49 737 Uncultured bacterium adhufec395 AF132234 99 Phasco/arctobacterium X72866 93 faecium T5- 50 430 Uncultured bacterium C825 A8050851 99 F. prausnitzii AJ413954 92 T5- 51 440 Uncultured bacterium adhufec77.25 AF153865 99 Porphyromonas sp. oral clone AY008310 89 T5- 52 440 Uncultured bacterium (human infant) L13dG AF253379 99 B. vulgatus A8050111 99 T5- 54 413 Unidentified eubacterium from anoxic bulk soil AJ229190 93 Clostridium fervidus L09187 95 T5- 55 415 Uncultured bacterium adhufec296 AF132258 99 E. desmolans 16S L34618 95 T5- 56 718 Unidentified eubacterium from anoxic bulk soil AJ229190 91 T. magadii Y15626 91 Table 8.2d: Clones present in the T51ibrary (continued).

Clone length Closest GenBank sequence match GenBank % Closest named GenBank GenBank % accession homology match accession homology T5- 57 764 Uncultured bacterium adhufec55 AF132275 99 B. merdae X83954 99 T5- 58 705 Uncultured bacterium adhufec365 AF132265 99 F. prausnitzii AJ413954 98 T5- 59 746 Bacteroides sp. AR20 AF139524 99 T5- 60 734 Eubacterium ventriosum L34421 99 T5- 61 686 Uncultured bacterium clone L 1 0-6 AJ400275 99 V. spinosum X90515 92 T5- 62 719 Uncultured bacterium CB25 AB050851 99 F. prausnitzii X85022 91 T5- 63 743 Uncultured bacterium adhufec13 AF132237 99 F.prausnitzii X85022 92 T5- 64 420 Bacteroides sp. AR29 AF139525 99 B. thetaiotaomicron L16489 98 T5- 65 730 Uncultured bacterium adhufec55 AF132275 99 B. merdae X83954 99 T5- 66 730 Uncultured bacterium adhufec55 AF132275 99 B. merdae X83954 99 T5- 67 735 Uncultured bacterium L 10-6 AJ400275 99 V. spinosum X90515 92 T5- 68 710 Uncultured bacterium A20 AF052417 99 R. obeum X85101 96 T5- 70 752 B. uniformis AB05011 0 99 T5- 71 788 Uncultured rumen bacterium 4C28d-4 AB034125 95 0. guillermondii AB040497 94 T5- 72 400 Uncultured bacterium adhufec168 AF132242 99 Clostridium leptum AJ305238 92 T5- 73 732 Uncultured bacterium adhufec363 AF132264 96 Butyrivibrio crossotus X89981 95 T5- 74 703 Uncultured bacterium clone L 1 0-6 AJ400275 99 V. spinosum X90515 91 T5- 75 724 Uncultured bacterium adhufec406 AF132269 99 C. xylanolyticum X71855 95 T5- 76 767 S/ackia exigua AF101240 92 T5- 77 740 Uncultured bacterium A21 AF052418 100 C. saccharolyticum Y18185 94 T5- 78 742 Uncultured bacterium adhufec367 AF132266 99 B. ovatus X83952 94 T5- 79 744 Uncultured bacterium adhufec296 AF132258 99 E. desmolans L34618 96 T5- 80 712 Uncultured bacterium adhufec363 AF132264 96 B. crossotus X89981 95 T5- 81 716 Eubacterium ha/ii L34621 99 T5- 82 618 E. formicigenerans L34619 97 T5- 83 717 Uncultured bacterium CB25 AB050851 100 F. prausnitzii AJ413954 92 T5- 84 696 Uncultured bacterium adhufec363 AF132264 96 B. crossotus X89981 95 Table 8.2d: Clones present in the T5library (continued).

Clone length Closest GenBank sequence match GenBan~ % Closest named GenBank GenBank % accession homology match accession homology T5- 85 406 Uncultured rumen bacterium 4C28d-4 AB034125 97 0. guil/ermondii AB040499 92 T5- 86 623 R. a/bus strain Ra8 AF104833 92 T5- 87 677 Unidentified rumen bacterium RFN80 AB009228 90 Desulfotomaculum ha/ophilum U88891 89 T5- 88 616 Prevotella sp. oral strain 831 FD AY005061 89 Prevotella /oescheii L16481 88 T5- 89 400 0. guil/ermondii AB040499 94 T5- 90 693 B. uniformis AB05011 0 100 T5- 92 671 Uncultured bacterium adhufec367 AF132266 99 B. ovatus X83952 93 T5- 93 350 Unidentified eubacterium U43698 90 Prevotella oris L16474 89 T5- 95 563 C. propionicum X77841 90 T5- 96 673 Unidentified eubacterium from anoxic bulk soil AJ229190 91 Moorel/a glycerini U82327 86 T5- 97 655 Uncultured bacterium CB25 AB050851 99 F. prausnitzii AJ413954 92 TABLE8.3: Results of LIB SHUFF comparisons of Odd and Even numbered sequences from each sample library. Library delta-C for XY p-value for XY delta-C for YX p-value for YX comparison comparison comparison comparison (2:.6.CXY )1 (2:.6.Cvx) SI 0.076 0.815 0.089 0.76 Tl 0.016 0.995 0.044 0.922 T3 0.334 0.221 0.153 0.519 T5 0.402 0.314 0.266 0.489

1 X refers to the odd-numbered sequences and Y to the even-numbered sequences. TABLE8.4: Results of LIB SHUFF comparison of complete sample libraries. Library X LibraryY delta-Cfor p-valuefor delta-C for p-value for XY XY YX YX comparison comparison comparison comparison (:LilCxv) (LilCYX) Tl T3 2.679 0.001 5.243 0.001 Tl TS 3.474 0.001 5.164 0.001 Sr T3 0.549 0.001 0.992 0.001 Sr TS 3.84 0.001 7.033 0.001 Sr Tl 3.464 0.001 0.438 0.001 T3 TS 3.448 0.001 2.4323 0.001

TABLE8.5: Bray-Curtis similarities by primer set for subjects TS 1 and S 1. Primer set Bray-Curtis similarity for sample comparisons Tl Sr T3 T4 TS T3 T4 TS domain Bacteria 62.9 62.9 58.1 82.8 82.8 80

RPO~ 57.1 20 46.2 50 33.3 66.7 Bacteroides- 20 20 20 45.5 45.5 54.5 prevotella group C. coccoides group 82.4 70.6 87.5 70.6 82.4 75 C. leptum subgroup 23.5 25 11.8 88.9 84.6 88.9 Bifidobacterium 44.4 40 36.4 90.9 83.3 92.3 Figure S.la: Phylogenetic tree

representing the clones in the S1 library.

'----Ciosllldlum~DSM 2544 AT~propionlcmiDSM 1682

ss-~g

·------~~~~'!arbfonneATCC 27806

~f~~Sl-15

__ ,·------unltonnls~~ ATCC 8492

L---i:=:=:=:=:=:=:=:=:=:=~;:;;::;:~~~v~fllloph/lllwlldsworlhlll U82813 spblisiMn DSM 4136

'------Closflld/IJm lenllt:UI ATCC 43204

0.1 Figure S.lb: Phylogenetic tree representing the clones in the Tl library.

T~lorquesATccmse ~===:~aossorus NC002416 II ~~ATCC275e0

...... "------c;;;= fl'_~unltonnisATCC 8492

'------Cbslliram lll.lsleblnnse ATCC 257511 '---~~~n~- 1~1 CtJslriram f1IOI1Iot*:lln DSM 1662

0.1 Figure S.lc: Phylogenetic tree representing the clones in the T3 lr-- Closlridlum cb!t*•omi8S ATCC 25537 library. Clost1lflm~OSM 2544 r;==.ii~~~=Uf88mgy

------T3-82T3-76

------Bacteroides oolloml ATCC 8492 ~------T3-~

Closlridlum 1eptum OSM 753 .L~c==-~~~:uslfiJwJillclensAF104848

0.1 '----f fg:J~ ~------~- --~~~~1

r------<====~~=- Vetri.ICCmofctOO sptJoslm OSM 4136

------Closllfdrum lenldJs ATCC 43Z04 ~------~

0.1 Figure 8.2: Graphs generated by LIBSHUFF for comparisons of the sample libraries. X is the coverage of each library by sequences from library X; XY is the coverage of library X by -t--X sequences from library Y; ACxv is the squared difference between --- XY X and XY at each evolutionary distance; 95% ACxv is the 95Qth ranked ACxv in 1000 randomly shuffled libraries from X andY. ACxv --*- 95%ACxv

....:---8.2a Figure8.2b .O.J16oan. T1 - T3 S-I T3 1.0 .-..r- c __.r_.,..-J I i I 1 Q) C) I C'O r I Q; 0.5 > 0 u J Lt": 0 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5 Evolutionary distance Evolutionary distance

Figure 8.2c Figure s.2d S - T 5 T1- TS ,- ~~ - 1.0 ..,.. D'" _...--r T _/_ .....!:_ --1 Q) r J C) ~ C'O 1 Q; 0.5 .,..; > 0 ! u j

l!.-' 0 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5 Evolutionary distance Evolutionary distance ....:---8.2e ... JI6U&~ T3- TS Figure 8.2r S - T 1 r-- l 1.0 ~ .I L ~ ,..... _,...... ,., ( Q) I C) 1 C'O Q; 0.5 ...... > I _1 0 I 1 u ...- I lL 0 ...... 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5 Evolutionary distance Evolutionary distance Figure 8.3: The mean p value obtained • Tl- ~ after the comparison of 20 randomly • T3- 51 selected sub-libraries vs the number of TS- 51 clones in each library obtained with X T1-T3 e Tl-TS LffiSHUFFpvalues. Libraries T3 and S1 are most alike. )K T3-TS

Mean p value vs number of clones

0.8 ~------~

Q) ~ 0. 6 +------~------t-P\:---A------~ «S > ~o.sr------~----~~~~------~ c: Q)«S 0.4 r------~~------+~------~ ~ 0.3

0.2r------~~~._------~~~------~

10 20 30 40 so 60 Number of clones 8.4 DISCUSSION

In this study PCR cloning was performed to further investigate the findings obtained with PCR-DGGE in chapter 7. In that chapter PCR-DGGE analysis showed that a hybrid microbiota developed after infusion of donated faeces into the colon during periods of reduced colonisation resistance. That is, the bacterial populations in faecal

samples at T3 and T5 were more like S1 than Tl and furthermore, there appeared to be a gradual change between timepoints T3, T4 and T5. To assess this in more detail16S rDNA clone libraries were constructed for one of the test subjects (TSl) and the source of the donated faeces (Sl).

Both the results of visual inspection of the phylogenetic trees representing each of

the libraries, and the analysis with LIBSHUFF, suggest that libraries T3 and S1 are most similar. For example, multiple clones closely related to E. forrnicigenerans, C.

mitsuokai, E. biforrne and F. prausnitzii are present in the S1 and T3 trees (48 out of 76 and 46 of 79 clones respectively) and are very infrequent in the trees depicting the clones in Tl and T5 (3 out of74 and 6 of 69 respectively). The Tl and T5 trees did not appear similar to each other or any of the other trees. LIBSHUFF analysis suggested that all of the complete libraries were significantly different and that the odd and even halves of each sample's library were not. The smallest differences in the complete library comparisons were between S1 and T3, with delta-C values for XY and YX comparisons of 0.55 and 0.99 respectively. Subsequent analysis with

LIBSHUFFpvalues confrrmed that S1 and T3 were clearly the most similar libraries with a minimum of 59 randomly selected clones being required to distinguish the two libraries.

In order to facilitate further discussion of these results for test subject TS 1 and source subject Sl, the results of banding pattern comparisons for these subjects with each of the primer sets used in Chapter seven are listed in Table 8.5 and a dendrogram illustrating these results is shown in Figure 7.6a, Chapter 7. The findings of PCR-cloning and LIB SHUFF would tend to partly corroborate the results

-271- ofPCR-DGGE for individuals TS1 and S1, shown in Table 8.5, particularly for the T3 sample compared with T1 and Sr. However the trees representing libraries T5 and Sr appear different and significant differences were observed in the LIB SHUFF analysis, with only 34 randomly selected clones required to distinguish the libraries using LIBSHUFFpvalues. This result is clearly not in keeping with the results in Table 8.5 that demonstrate nearly equivalent Bray-Curtis similarities for comparisons of T3 with Sr and T5 with Sr.

There are several observations that suggest that the clone libraries are unlikely to be good representations of the bacterial species present in the samples and this may explain the incongruous result for T5 and S1• This suggestion is supported by the fact that bacterial species from major phylogenetic groups within the colon were underrepresented in the libraries, while other less common groups were over­ represented. For example, genospecies related to the Bacteroides group were very infrequent in libraries from 3 of the 4 samples and were absent in one. As PCR amplification and cloning are stoichiometric processes, it is common for relatively important bacterial groups not to be represented on DGGE gels of PCR amplified faecal DNA (388) and in clone libraries, as they are significantly outnumbered by the most numerically dominant groups of bacteria. However, the numbers of Bacteroides spp. have been shown to be only slightly less than those of the dominant C. coccoides group, that is, by less than one order of magnitude (110). In addition, Bacteroides spp. were cultured from all of the 20 faecal samples studied by Moore and Holdeman (226) and were represented in counts of at least 1010 per cells per gram (dry weight) of faeces in the FISH study of 9 volunteers by Franks et al (110). In other words, it is highly probable that significant numbers of bacteria from this phylogenetic group were present in each faecal sample and their absence in the clone libraries is unlikely to be due to stoichiometric effects alone.

Other sequences are likely to have been significantly over-represented. This effect can be clearly seen by looking at the large groups of clones that are evident in each library. In the Sr library sequences related to C. mitsuokai and F. prausnitzii make up

-272- 21 and 13 ofthe 76 clones (45%) respectively. In T1, sequences related to Roseburia intestinalis and R. obeum make up 13 and 9 of the 74 clones (30% ), while in T3 F. prausnitzii and C. mitsuokai make up 17 and 13 of the 79 clones (38%). This gross over-representation of some sequences at the expense of others is most likely to reflect preferential amplification during PCR.

Preferential amplification is known to increase as the number of PCR cycles increases (28, 376, 377). However the clone libraries in this study were constructed after only 10 cycles of amplification. This fmding would suggest that preferential amplification can be a problem even with a modest number of cycles of amplification. An explanation for this occurrence may be found in the observations ofTajima et al (2001) with real-time PCR (341). Using real-time PCR with a universal primer set similar to the one applied in this study, and DNA template from pure bacterial cultures, the number of cycles of amplification that were required before the fluorescence signal reached threshold was measured (341). The results of this study showed that the number of cycles required to reach threshold was a function of which bacterial species template was studied. For example, after only 10 cycles of amplification there was a very significant signal from Streptococcus bovis while Fibrobacter succinogenes had not yet reached threshold. This finding provided a possible explanation for the fact that although F. succinogenes was the third most numerous of the bacterial species in the rumen of cows on a hay diet, it was only represented by a single sequence in the three clone libraries published at the time (341). These authors hypothesised that DNA-associated molecules may have been responsible for this effect by inhibiting amplification. Hansen et al (1998) have suggested that the secondary structure of the DNA template itself might inhibit the amplification of some 16S rDNA molecules (147). In their study the biased amplification of mixed DNA template from pure cultures of 4 species could be eliminated when PCR amplified 16S rDNA was used as template instead of native DNA. These authors suggested that DNA flanking the template region interfered with the initial cycles of amplification due to the formation of a secondary structure, and that this inhibition was specific to a given species. In support of their

-273- theory, the study also demonstrated that the use of different sets of 16S rDNA primers resulted in the preferential amplification of 16S rDNA from different organisms. One possibility is that this amplification bias is reproducible for a given species with a given primer set, and therefore might be accounted for in the analysis of PCR results, however this hypothesis is as yet unproven.

Thus, the large numbers of very similar sequences that are over-represented in the libraries of this study and the under-representation of other sequences is most likely to be a function of the effects proposed by Tajima et al and Hansen et al resulting in preferential amplification (147, 341).

It could be argued that the clone libraries are an adequate representation of the species present in each sample as each was estimated to result in a reasonable coverage of the OTUs present in each library- ranging from 76 to 86%. However, preferential amplification of template results in an overestimation of coverage because increases in multiply represented clones reduce the number of unique clones as a proportion of the total. For example, in two independent studies of the same human faecal sample using 10 and 25 cycles of PCR followed by cloning and sequencing, 284 and 69 clones with estimated coverage values of 85 and 75% resulted, with each library containing 82 and 35 OTUs respectively (28, 335). It is clear that preferential template amplification occurred during the second study and resulted in a reduction of observed OTU diversity and an overestimation of coverage (28). Similarly, in a study of a single human faecal sample by Wilson and Blitchington (1996), 50 clones obtained after 9 cycles ofPCR amplification revealed 27 OTUs and resulted in an estimate of coverage of 59% while after 35 cycles of amplification the 39 clones that were obtained revealed only 13 OTUs and estimated coverage at 74% (376).

The relatively high coverage values obtained with each library in this study are a measure of the completeness with which each clone library represents the 16S rDNA genospecies in the PCR product, rather than in the sample itself. Measurement of the

-274- completeness of the coverage of the underlying bacterial species present in the actual sample is not possible with a clone library alone. Similarly, the coverage of the faecal microbiota by cultured isolates reflects the coverage of the cultivable bacteria in the sample. This point is well illustrated by Moore and Holdeman's estimate of coverage based on comprehensive culture of faecal samples from 20 individuals (226). Forty-four to 69 bacterial species were isolated from each sample and the coverage estimated using Good's formula ranged from 71 to 95%. To put these results in context Suau et al's PCR-cloning study subsequently demonstrated that 76% of 16S rDNA sequences do not correspond to known organisms (represented by sequences available in public databases) (335). That is, a majority of human faecal bacteria have not been cultivated and it is unlikely that Moore and Holdeman achieved a 95% coverage of the bacteria present in each faecal sample. Their figures represent the coverage by their cultured isolates of the bacterial species in each sample that were cultivable by the method they used.

Hughes et alestimated that the actual number of OTUs in the PCR product of Suau et al's 10-cycle faecal PCR-clone library was 135 with 95% confidence limits of 110 to 170 (Chaol estimator) (157). Applying the same method to the four clone libraries in the present study resulted in lower estimates of OTU number, these being 33 to 61 with upper 95% confidence limits of 51 to 105. How did such a difference arise? Although the method ofPCR-cloning and sequencing used in the present study was very similar to Suau et al and Bonnet et al there are identifiable differences (28, 335). In particular, the method of DNA extraction was different. Suau et al used bead-beating to extract DNA from their sample whereas a chemical form of lysis was used in this study (126). Evidence to support the view that the method of DNA extraction is important arises from the observation that different extraction methods are known to influence the sensitivity of PCR for specific bacterial templates in faecal samples (219, 389). Bacteria are also known to vary in their susceptibility to chemical lysis. While this makes comparisons between studies difficult it is unlikely to impact upon the comparison of clone libraries obtained by the same method.

-275- The LIB SHUFF analysis determined that all of the libraries were significantly different despite the high levels of Bray-Curtis similarity that were observed at PCR­ DGGE with multiple primer sets for the comparison of several of the samples. The LIB SHUFF analysis is an algorithm for determining if two libraries are significantly different, based on a comparison between the measured distances between their constituents and the distances between random recombinations of their component sequences. This is a mathematical algorithm that states whether the two libraries can be distinguished and the statement of probability refers to whether the libraries per se are likely to be different, rather than being a measure of certainty about the populations of bacteria within the sample itself. This analysis does not take account of sampling issues or the uncertainties involved in generating a PCR clone library, such as preferential amplification. To achieve an estimate of these effects for a given microbiota repeated sampling and multiple clone-libraries would need to be performed - a substantial task.

The comparison of libraries by the LIB SHUFF method is known to be strongly affected by the number of clones in each library. For example, Singleton et al found that a minimum of 25 clones were required in each library from bioreactors and grassland soils for them to be distinguished. Clone libraries representing samples of the same type such as paired samples from grassland soils or bioreactors were not differentiated by LIBSHUFF even when 90 or more clones were compared. The exception to this was the two clone libraries obtained from arid soils that were significantly different with only 53 and 59 clones (324). Like the result for arid soils, the clone libraries in this study of human faecal samples were significantly different despite originating from a similar environmental site. The results of . LIBSHUFFpvalues revealed that random sub-libraries containing between 34 and 39 clones from the complete clone libraries were reliably differentiated except for

comparisons of S1 and T3. The odd and even halves of each library were of similar size, and were not found to be significantly different by the LIB SHUFF method, suggesting a degree of homogeneity within each library and supporting the validity

-276- of the results for complete library comparisons. After the completion of the present study Singleton et al stated that the published LIBSHUFF program was incorrect and evaluated clone libraries using ACXY = (Cx-CTii rather than the correct formula

2 ACxy = (Cx-CXY) • The authors stated that this error could lead to erroneously high p­ values (DR Singleton, personal communication). In the present study this error is unlikely to have impacted on the results for complete clone library comparisons as the p values in this studies were already highly significant. A corrected version of the script (LIBSHUFF vl.2) has been posted on the web (http://www. arches. uga.edu/-whitman/libshuff.html).

In summary, although the results ofPCR-cloning are broadly supportive of the findings of Chapter 7 it is clear that this method is seriously limited in its application to the comparison of different samples. Ideally such comparisons would require repeated samplings and multiple clone libraries so that the effects of sampling and PCR-amplification could be accounted for statistically. However, the cost and labour involved in such an approach would be prohibitive.

-277- CHAPTER 9: Discussion

9.1 Principal imdings

The principal findings of this work are summarised below.

1. A distinct banding pattern representing the 3 numerically dominant phylogenetic groups of anaerobes -the Bacteroides-prevotella group, Clostridium coccoides group and Clostridium leptum subgroup- can be produced with PCR-DGGE. This method can be applied to detect changes in the species composition of the human or murine faecal microbiota.

2. The Helicobacter species of the murine large intestine can be detected and speciated by PCR-DGGE. In practice this method has a similar sensitivity to species-specific PCR but has the advantage that novel species, or Helicobacter species that were not suspected to be present in a given murine facility, may be detected. The assay will also detect more than one Helicobacter species in a given sample with a single PCR reaction.

3. Helicobacter species, other than H. pylori of gastric origin, were not detected in colonic biopsies from Australian controls undergoing colonoscopy or cases of ulcerative colitis or Crohn's disease at levels greater than 102 per biopsy.

4. Treatment with antibiotics and cathartics dramatically reduces colonisation resistance in the human and murine lower bowel. The data obtained from a murine study suggests that antibiotics are the most important component of this process as catharsis alone did not reduce colonisation resistance significantly.

5. The administration of a suspension of donated faeces, when colonisation resistance is impaired in mice and humans, can change the bacterial populations

-278- of the colon such that a hybrid microbiota consisting of parts of the donated faecal microbiota and the subject's baseline microbiota develops.

6. In humans this hybrid microbiota appears to consist predominantly of the donated bacterial populations and only gradual and modest changes occur during the subsequent 24 weeks.

7. Murine studies suggest that the caecum may be the anatomical site where viable bacteria from the donated microbiota must reach, in order to establish a hybrid microbiota. Furthermore, a single day of faecal infusions when colonisation resistance is impaired may be enough to induce a hybrid microbiota.

8. PCR-DGGE is ideally suited to examining changes in the microbiota of individuals over time, while PCR-cloning is not useful for this purpose.

9.2 The further application of PCR-DGGE in studies of colonic bacterial ecology

A key observation from this work and the work of others (291, 388) is that PCR­ DGGE methods are most valuable in determining whether a change in the species composition of the microbiota of an individual has occurred over time. The observation that a change has occurred can be further investigated to identify which species have been lost or gained by the sequencing of excised DGGE bands. Alternatively cloning of the PCR product, followed by PCR-DGGE can be performed to determine which cloned products denature at the gel position of interest. Methods such as FISH or even culture may be directed by the results of PCR-DGGE. An example of an important research question that could be approached by this method is the determination of whether a change in the microbiota of the colon occurs prior to the onset of disease, development of flares, or onset of remission in ulcerative colitis or Crohn' s disease. For example, faecal samples could be collected at regular intervals from cohorts of individuals at risk for

-279- the development of Crohn's disease, such as the siblings of affected individuals. An existing database of such individuals is already established in Australia for genetic studies (40). The analysis of serial faecal samples from individuals in this cohort that subsequently developed disease could determine if a change in species composition preceded disease expression. The identification of which species changed in association with disease development or exacerbation could form the basis for new preventative or therapeutic regimens for these diseases. Conversely, even a study showing that there did not appear to be a change in the species composition of the human colon in various phases of IBD would be an extremely valuable addition to our knowledge of this disease.

Despite the strength of PCR-DGGE methods applied to the study of individuals over time there are significant barriers to the application of this method to the comparison of microbiotas from different individuals (332, 390). An example of such a comparison would be the study of cases of IBD and controls. The principal problem with PCR-DGGE is the limited representation of the bacterial populations present using the domain Bacteria primer set, as well as the fact that samples electrophoresed on different gels are not easily comparable. These problems have been partly addressed in this thesis through the development of primer sets in Chapter 3 that significantly increase the representation of faecal bacterial populations compared with the universal primer set. Just how comprehensive this representation is requires further study. A logical approach to describing the diversity of species amplified with each primer set would be cloning of the PCR product obtained with each primer set. This would identify which bacterial species are commonly represented at DGGE during their use. Furthermore, the application of PCR-cloning with these new primer sets to the samples from the subjects studied in Chapter 7, might produce more meaningful results in a PCR-cloning study than the universal primer set that was applied in Chapter 8, by reducing the diversity of species to be amplified. However, given the difficulties ofPCR-cloning a useful result is not guaranteed.

-280- The PCR-DGGE approach to the study of the faecal microbiota could be enhanced by the addition of primer sets for other groups that are represented in faecal clone libraries, such as the Atopobium group, Phascolarctobacterium group, Veillonellae and Enterobacteriaceae. In this way a more comprehensive assessment of the faecal microbiota could be made. If a method were developed such that gels could be readily compared then this approach might be a useful tool for the study of the microbiota as a whole in different individuals. Examples of how this might be achieved would include the use of specific markers and a highly reproducible denaturant or temperature gradient perhaps in combination with an automated system with well-measured variances for band positions. Automated gel electrophoresis and analysis is commonplace in sequencing and ribotyping, and with appropriate funding this could probably be applied to temperature gradient gel electrophoresis (54, 388). Using such a method a single gel might produce a visual representation of the species present in the faecal microbiota from an individual that could be compared with gels from other individuals.

The clone libraries that were produced in this study show clear evidence of preferential template amplification and the resulting bias in library composition. While the method is useful it could be significantly improved if this bias was addressed. Studies by Tajima et al and Hansen et al suggest that either DNA itself or DNA associated molecules interfere with the amplification of some 16S rDNA templates (147, 341). If this effect could be eliminated without disrupting primer binding, then significantly better clone libraries might be produced. Potential approaches to this problem would include cutting the mixed bacterial DNA template into smaller segments, with the elimination of segments that did not contain 16S rDNA, prior to PCR. DNA containing 16S rDNA sequence is readily identified with a combination of oligonucleotide and antibody probes during ribotyping, and this method could be adapted to produce DNA separation (54). Alternatively, mixed template DNA could be denatured in a matrix that interfered partially with the annealing of long complimentary strands of DNA, to facilitate primer binding in subsequent PCR. Real-time PCR provides a simple method for evaluating the

-281- efficacy of any new methods that could be applied to cloning, by producing a cycle­ by-cycle measurement of the amplification of template DNA from different bacterial species during PCR with a universal primer set (341). The amplification of different DNA templates is therefore readily compared and any improvement in preferential amplification should be apparent. Of course the additional processing of the template DNA in either of these approaches is associated with potential bias in the resulting clone libraries. For example, 16S rDNA templates that were cut by the restriction enzyme would be under-represented in the resulting clone libraries.

A relatively new technology that is being applied in the field of bacterial ecology with a modest degree of early success is microarray analysis (41, 198, 255, 325). This approach has the advantage of directly detecting bacterial16S rRNA without PCR amplification. For example, Small et al could identify Geobacter chapellei and D. desulfuricans 16S rRNA directly in soil extracts using a microarray (325). However the secondary structure of rRNA is a barrier to specific hybridisation on microarrays (41, 255, 325). For example, it has been shown that the removal of RNA secondary structure through fragmentation by heating at 95°C for 30 minutes, significantly improved hybridisation specificity and signal intensity (325). Nevertheless obtaining specific hybridisation or accounting for non-specific hybridisation in mixed rRNA templates of unknown composition remain major challenges. For example, a "PhyloChip" described by Loy et al (2002) that was based on 132 oligonucleotides specific for all known sulfate-reducing prokaryotes correctly identified Desulfomicrobium species in PCR amplified 16S rDNA from the periodontal tooth pockets of 2 of 5 patients (198). This microarray result was confirmed with species-specific 16S rRNA PCR and a PCR for dissimilatory sulfite reductase. However, the results from a hypersaline microbial mat containing a more diverse population of sulfate-reducing bacteria using the same microarray were reported to be difficult to interpret (198). In this study, positive and negative results were obtained from the same sample with different oligonucleotides specific for the genus Desulfofaba (198). The authors could not confirm the presence of this genus in the samples by other methods. It is difficult to interpret this type of conflicting

-282- result. The microarray approach for 16S rRNA that has shown the greatest promise in reducing non-specific hybridisation is the use of a two probe system; a species­ specific capture probe attached to the glass slide and a chaperone detector probe that binds to target 16S rRNA approximately 10 nucleotides away from the capture probe (41). In the study of Chandler et al (2003), capture probes 10 nucleotides away from the detector probe performed vastly better than capture probes immediately adjacent to the detector probe (41). Capture probes immediately adjacent to the detector produced significant cross-hybridisation to non-target detector probes (41). Whether this method can eventually be accurately applied to assess species diversity and even be used in a semi-quantitative manner only the future will tell.

9.3 The issue of sampling - are the species that are present in human faeces representative of the caecal microbiota?

The study performed in Chapter 3 demonstrates that the microbiota of the caecum and faeces from the same mouse are highly similar within the Bacteroides-prevotella group, C. coccoides group and C. leptum subgroups. The data of Marteau et al clearly shows that there are quantitative differences in the common bacterial groups present in post-meal caecal samples and faecal samples from the same subject (213). In that study, bacteria from the Bacteroides group, Clostridium leptum group and bifidobacteria made up a much larger proportion of total bacteria in faecal compared with caecal samples. Whether the samples that were aspirated through a nasocaecal tube represent the real composition of the caecum is uncertain because theoretically such sampling is biased by containing only the liquid portion of caecal and ileal content. However, Marteau et al's results are not inconsistent with faecal and caecal samples having the same species composition despite quantitative differences. To be able to make unqualified conclusions about the species composition of the human caecal and faecal microbiota, samples must be directly obtained from the unprepared human caecum. The development of a method of push-enteroscopy that can achieve insertion of a steerable orally inserted enteroscope with a biopsy channel into the

-283- unprepared caecum opens up a real possibility to obtain the samples that are necessary to directly answer this question in humans (383).

9.4 Do humans have mucus-colonising bacterial populations in the colon?

A key issue raised by this research is why, contrary to our initial hypothesis, humans do not appear to have commensal lower bowel Helicobacter species when animals frequently do so. It is necessary however to qualify this statement by noting that only a small number of subjects and ethnic groups were sampled in the study reported in Chapter 5. Therefore further studies in other populations would be needed to defmitively conclude that humans do not have commensal lower bowel Helicobacter species, however all of the currently available evidence (that is not complicated by methodological flaws) would support these findings (27, 249). One possible explanation for the absence of commensal Helicobacter species in the human colon is that the faecal-oral transmission of bacteria occurs less frequently in human populations compared with animals as a result of comparatively higher 'standards of hygiene. This results in less opportunity for members of this genus of animal origin present in the environment to evolve and become "humanised". Alternatively, the crypt defences of the human colon may be fundamentally different and more effective than those of animal species.

The putative absence of Helicobacter species in human colonic mucus and crypts suggests that humans may not have a commensal mucus-colonising microbiota as this genus are frequent mucus-layer commensals in animals (72, 220, 293, 329, 330, 386). However definitive studies of the mucus layer of the human colon have not been achieved because of sampling issues related to bowel preparation for colonoscopy or surgery. The limited evidence to date suggests that such a microbiota is not present. If a mucus-associated microbiota is present in some human populations then subjects from developing countries with higher rates of disease transmitted via the faecal-oral route would be an ideal study cohort, as a higher prevalence of colonisation would be anticipated. Intestinal tissue obtained

-284- immediately after death and evaluated with FISH and a domain Bacteria probe using cryostat sections to directly visualise bacteria in the mucus layer would be an ideal method (subject to ethics approval).

9.5 Can changing the human colonic micro biota have therapeutic application?

A key question arising form the results of this study is whether the alternative health practice studied in Chapter 7 should be applied at this time in patients with inflammatory bowel disease. Certainly, were it to be applied, then it would provide a major opportunity to determine whether bacterial factors are involved in perpetuating disease, and for directly assessing which bacterial species are important. In an ideal experimental system a small number of species would be targeted and removed from or added to the microbiota of diseased subjects and markers of clinical response assessed. However, the procedure described in Chapter 7 produces a dramatic change in the colonic microbiota.

In the absence of direct control of the individual species comprising the donated microbiota, it is likely that the species changes that occurred in groups of responders and non-responders in such studies would be pooled and then assessed statistically to identify species that were associated with response and non-response. For example a species list describing the composition of the microbiota pre- and post procedure could be directly compared in those individuals with a clinical response and without. If large resources of time and money were available then an overall species list could be generated with the cultural method of Moore and Holdeman or with PCR-cloning for each sample. Alternatively, specific phylogenetic groups or subgroups could be compared using culture, PCR-DGGE or PCR-cloning with novel primers targeting components of the microbiota thought to be important in IBD, for example the Enterobacteriaceae. Such a study would supply basic direct information concerning the role of specific bacterial species in IBD pathogenesis.

-285- The use of mixtures of cultured bacteria would reduce the risk of transmission of infection during the procedure and has been reported to be of benefit in the past (6, 274, 352). The bacterial suspensions used in these therapies ranged from Enterococcusfaecium alone (274) to a mixture of 10 bacteria in the studies ofTvede and Rask-Madsen (352) and 18 species in the study of Andrews and Borody (6). The ten species used by Tvede and Rask-Madsen were Enterococcusfaecalis, Clostridium innocuum, Clostridium ramosum, Clostridium bifermentans, Bacteroides ovatus, B. vulgatus, Bacteroides thetaiotaomicron, 2 strains of E. coli and Ruminococcus productus (352). In all cases the bacterial species were cultivated separately and mixed just prior to their administration (6, 352). However whether a novel stable bacterial community can be promoted in the recipients of such therapy must be seriously questioned given the known diversity of species in the human colon (157, 335).

Ideally, clear evidence that certain species of the microbiota are implicated in the pathogenesis of IBD would be obtained prior to the use of this type of treatment. That is, a specific rationale should be developed first and the use of the procedure restricted to clinical trials with appropriate ethics oversight and safety monitoring. In the absence of a specific rationale, the whole question of whether the procedure should be applied to study the microbiota in IBD or should be withheld until a rationale is discovered, hinges on it's safety. That is, if years of experience informed us that the procedure had no significant adverse effects then it's use in clinical trials in IBD might be justified. Unfortunately there is little published data on safety to guide us in this assessment. The use of this procedure where there does not appear to be a clear rationale, for example in IBS, is difficult to sustain.

There is no reason why, prior to any human studies, the procedure could not be studied in an animal model ofiBD. Murine models may not be ideal for a number of reasons, including the fact that murine models of IBD are not particularly like human IBD (94). Several of these have major rather than subtle immunological deficits, or require a chemical treatment to induce colitis (153). In addition, the

-286- bacterial species that are important in disease induction and persistence in rodents are unknown (271, 272, 311). An important consideration is the fact that the change in the microbiota induced in mice during this work (presented in Chapter 6) appeared less complete than that achieved in humans (presented in Chapter 7) suggesting that mice may be a suboptimal model. For example, the Bray-Curtis similarity of the post-procedure time-points to the infused faecal suspension were much higher in humans than in mice. This lower apparent effectiveness in mice could relate to multiple factors, the most obvious difference being that the faecal suspension was directly introduced into the caecum in the humans and not in the mice. Other differences are readily apparent in the different types of preparation and treatment regimen used in the two studies.

The use of a larger animal model, for example rats, gerbils, hamsters, guinea pigs or rabbits, where administration of faecal suspension into the caecum might be more feasible than in mice, could prove a better model. However at present there are no well-defmed spontaneously occurring or knock-out models ofiBD in these larger animals.

9.6 Conclusion

In reality very little is known about the interactions of the bacterial species in the human colon with each other and the host and their impact. This lack of knowledge is directly attributable to the difficulties associated with accurately measuring the microbiota in terms of species composition and relative numbers. At present there is no convenient method that will accurately describe and quantitate the bacterial species present in a human faecal sample. Were such a method to be developed then determining if specific bacterial populations play a role in the pathogenesis of bowel diseases would be relatively simple. In this work, PCR-DGGE was the ideal approach to examine the experimental question of whether the species composition of the colon had changed. The resulting gels unequivocally demonstrate clear changes in the microbiota. In contrast, PCR-cloning failed to confirm the results of

-287- PCR-DGGE due to problems inherent to the method. In the absence of an ideal method, future studies of the human colonic microbiota must be carefully designed to account for the limitations of each of the current methods.

-288- REFERENCES

1. Al-Soud W A, M Bennedsen, SL On, IS Ouis, P Vandamme, HO Nilsson, A Ljungh, and T Wadstrom. 2003. Assessment ofPCR-DGGE for the identification of diverse Helicobacter species, and application to faecal samples from zoo animals to determine Helicobacter prevalence. J Med Microbial. 52(9):765-71. 2. Alander M, R Satokari, R Korpela, M Saxelin, T Vilpponen-Salmela, T Mattila-Sandholm, and A von Wright. 1999. Persistence of colonization of human colonic mucosa by a probiotic strain, Lactobacillus rhamnosus GG, after oral consumption. Appl Environ Microbial. 65(1):351-4. 3. Albert MJ, P Bhat, D Rajan, PP Maiya, SM Pereira, and SJ Baker. 1978. Faecal flora of South Indian infants and young children in health and with acute gastroenteritis. J MedMicrobiol.11:137-43. 4. Altschul SF, TL Madden, AA Schaffer, J Zhang, Z Zhang, W Miller, and DJ Lipman. 1997. Gapped BLAST and PSI-BLAST:a new generation of protein database programs. Nucleic Acids Res. 25:3389-402. 5. Amann RI, W Ludwig, and K-H Schleifer. 1995. Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbial Rev. 59(1):143-69. 6. Andrews PJ, and TJ Borody. 1993. Putting back the bugs: Bacterial treatment relieves chronic constipation and symptoms of irritable bowel syndrome. Med J Aust. 159:633-4. 7. Apajalahti JHA, A Kettunen, PH Nurminen, H Jatila, and WE Holben. 2003. Selective plating underestimates abundance and shows differential recovery of bifidobacterial species from human feces. Appl Environ Microbial. 69(9):5731-5. 8. Apperloo-Renkema HZ, and D Vander Waaij. 1991. Study of colonization resistance for Enterobacteriaceae in man by experimental contamination and biotyping as well as the possible role of antibodies in the clearance of these bacteria from the intestines. Epidemiol Infect. 107:619-26.

-289- 9. Ashelford KE, AJ Weightman, and JC Fry. 2002. PRIMROSE: a computer program for generating and estimating the phylogenetic range of 16S rRNA oligonucleotide probes and primers in conjunction with the RDP-ll database. Nucleic Acids Res. 30(15):3481-9. 10. Aspinall ST, DR Wareing, PG Hayward, and DN Hutchinson. 1993. Selective medium for thermophilic campylobacters including Campylobacter upsaliensis. 1 Clin Pathol. 46:829-31. 11. A venaud P, B Le Bail, K Mayo, A Marais, R Fawaz, P Bioulac-Sage, and F Megraud. 2003. Natural history of Helicobacter hepaticus infection in conventional A/J mice, with special reference to liver involvement. Infect Immun. 71(6):3667-72. 12. Avenaud P, A Marais, L Monteiro, B Le Bail, P Bioulac-Sage, C Balabaud, and F Megraud. 2000. Detection of Helicobacter species in the liver of patients with and without primary liver carcinoma. Cancer. 89(7):1431-9. 13. Bampton PA, PG Dinning, BS Kennedy, DZ Lubowski, and U Cook. 2001. Prolonged multi-point recording of colonic manometry in the unprepared human colon: providing insight into potentially relevant pressure wave parameters. Am 1 Gastroenterol. 96(6):1838-48. 14. Banwell JG, W Branch, and JH Cummings. 1981. The microbial mass in the human large intestine. Gastroenterology. 80(5):A1104. 15. Barnes EM, and M Ingram. 1956. The effect of redox potential on the growth of Clostridium welchii strains isolated from horse muscle. 1 Appl Bact. 19:117. 16. Beckwith CS, CL Franklin, RR Hook, CL Beschwilliford, and LK Riley. 1997. Fecal PCR assay for diagnosis of helicobacter infection in laboratory rodents. 1 Clin Microbial. 35:1620-3. 17. Behrens S, C Ruhland, J Inacio, H Huber, A Fonseca, I Spencer- Martins, BM Fuchs, and R Amann. 2003. In situ accessibility of small-subunit rRNA of members of the domains Bacteria, Archaea, and Eucarya to Cy3-labeled oligonucleotide probes. Appl Environ Microbial. 69(3):1748-58.

-290- 18. Bennet JD, and M Brinkman. 1989. Treatment of ulcerative colitis by implantation of normal colonic flora. Lancet. i(164). 19. Benson DA, I Karsch-Mizrachi, DJ Lipman, J Ostell, BA Rapp, and DL Wheeler. 2000. GenBank. Nucl Acids Res. 28:15-8. 20. Berg DJ, N Davidson, R Kuhn, W Muller, S Menon, G Holland, L Thompson-Snipes, MW Leach, and D Rennick. 1996. Enterocolitis and colon cancer in interleukin-10 deficient mice are associated with aberrant cytokine production and CD4(+) TH-1like responses. J Clin Invest. 98(4):1010-20. 21. Bernstein LH, MS Frank, LJ Brandt, and SJ Boley. 1980. Healing of perianal Crohn's disease with metronidazole. Gastroenterology. 79(2):357-65. 22. Bertone ER, EL Giovannucci, NW King, AJ Petto, and LD Johnson. 1998. Family history as a risk factor for ulcerative colitis-associated colon cancer in Cotton-Top Tamarin. Gastroenterology. 114(4):669-74. 23. Bishop RF, and CM Anderson. 1960. The bacterial flora of the stomach and small intestine in children with intestinal obstruction. Arch Dis Child. 35:487- 91. 24. Blaser MJ. 1998. Helicobacters and biliary tract disease. Gastroenterology. 114(4):840-2. 25. Blaut M, MD Collins, GW Welling, J Dore, J van Loo, and W de Vos. 2002. Molecular biological methods for studying the gut microbiota: the EU human gut flora project. Br 1 Nutr. 87(Suppl. 2):S203-11. 26. BohnhoffM, CP Miller, and WR Martin. 1964. Resistance of the mouse's intestinal tract to experimental salmonella infection. I. Factors which interfere with the initiation of infection by oral innoculation. 1 Exp Med. 120:805-16. 27. Bohr URM, A Primus, A Zagoura, B Glasbrenner, T Wex, and P Malfertheiner. 2002. A group-specific PCR assay for the detection of Helicobacteraceae in human gut. Helicobacter. 7(6):378-83. 28. Bonnet R, A Suau, J Dore, GR Gibson, and MD Collins. 2002. Differences in rDNA libraries of faecal bacteria derived from 10- and 25-cycle PCRs. Int 1 Syst Evol Microbial. 52:757-63.

-291- 29. Bornside GH, GW Cherry, and B Myers. 1973. Intracolonic oxygen tension and in vivo bactericidal effect of hyperbaric oxygen on rat colonic flora. Aerospace Med. 44(11):1282-6. 30. Borody TJ, EF Warren, S Leis, R Surace, and 0 Ashman. 2003. Treatment of ulcerative colitis using fecal bacteriotherapy. J Clin Gastroenterol. 37(1):42-7. 31. Boudeau J, N Barnich, and A Darfeuille-Michaud. 2001. Type 1 pili- mediated adherence of Escherichia coli strain LF82 isolated from Crohn's disease is involved in bacterial invasion of intestinal epithelial cells. Mol Microbial. 39(5): 1272-84. 32. Boudeau J, A-L Glasser, E Masseret, B Joly, and A Darfeuille-Michaud. 1999. Invasive ability of an Escherichia coli strain isolated from the ileal mucosa of a patient with Crohn's disease. Infect Immun. 67(9):4499-509. 33. Bowden TA, AR Mansberger, and LE Lykins. 1981. Pseudomembranous enterocolitis: mechanism of restoring floral homeostasis. Am Surg. 47 (4): 178-83. 34. Bowman JP, and RD McCuaig. 2003. Biodiversity, community structural shifts, and biogeography of prokaryotes within Antarctic continental shelf sediment. Appl Environ Microbial. 69(5):2463-83. 35. Bray JR, and JT Curtis. 1957. An ordination of the upland forest communities of Southern Wisconsin. Ecol Monogr. 27:325-49. 36. Brigidi P, E Swennen, B Vitali, M Rossi, and D Matteuzzi. 2003. PCR detection of Bifidobacterium strains and Streptococcus thermophilus in feces of human subjects after oral bacteriotherapy and yoghurt consumption. Int J Food Microbial. 81:203-9. 37. Bruggemann J, JR Stephen, Y-J Chang, SJ Macnaughton, GA Kokwalchuk, E Kline, and DC White. 2000. Competitive PCR-DGGE analysis of bacterial mixtures an internal standard and an appraisal of template enumeration accuracy. J MicrobiolMethods. 40:111-23. 38. Buczolits S, R Rosengarten, R Hirt, and H Busse. 2001. Classification of a Brevundimonas strain detectable after PCR with a Helicobacter-specifi.c primer pair. Syst Appl Microbial. 24:368-76.

-292- 39. Burke DA, and ATR Axon. 1988. Adhesive Escherichia coli in inflammatory bowel disease and infective diarrhoea. BMJ. 297:102-4. 40. Cavanaugh JA, DF Callen, SR Wilson, PM Stanford, ME Sraml, M Gorska, J Crawford, SA Whitmore, C Shlegel, S Foote, M Kohonen-Corish, and P Pavli. 1998. Analysis of Australian Crohn's disease pedigrees refines the localization for susceptibility to inflammatory bowel disease on chromosome 16. Ann Hum Genet. 62(4):291-8. 41. Chandler DP, GJ Newton, JA Small, and DS Daly. 2003. Sequence versus structure for the direct detection of 16S rRNA on planar oligonucleotide microarrays. Appl Environ Microbial. 69(5):2950-8. 42. Chao A. 1987. Estimating the population size for capture-recapture data with unequal catchability. Biometrics. 43:783-91. 43. Chao A. 1984. Non-parametric estimation of the number of classes in a population. Scand J Stat. 11:265-70. 44. Chapman RW, WS Selby, and DP Jewell. 1986. Controlled trial of intravenous metronidazole as an adjunct to corticosteroids in severe ulcerative colitis. Gut. 27:1210-2. 45. Chaudhury NA, and SC Truelove. 1962. The irritable colon. Q J Med. 31:307-22. 46. Chen W, D Li, RJ Cannan, and RS Stubbs. 2003. Common presence of Helicobacter DNA in the gallbladder of patients with gallstone diseases and controls. Dig Liver Dis. 35(4):237-43. 47. Chien C-C, NS Taylor, Z Ge, DB Schauer, B Young, andJG Fox. 2000. Identification of cdtB homologues and cytolethal distending toxin activity in enterohepatic Helicobacter spp. J Med Microbial. 49:525-34. 48. Chin EY, CA Dangler, JG Fox, and DB Schauer. 2000. Helicobacter hepaticus infection triggers inflammatory bowel disease in T cell receptor alphabeta mutant mice. Comparative Med. 50(6):586-94. 49. Chomczynski P, and N Sacchi. 1987. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem. 162:156-9.

-293- 50. Christensen HR, H Frokiaer, and JJ Pestka. 2002. Lactobacilli differentially modulate expression of cytokines and maturation surface markers in murine dendritic cells. J Immunol. 168:171-8. 51. Chu NS, D McAlister, and PA Antonoplos. 1998. Natural bioburden levels detected on flexible gastrointestinal endoscopes after clinical use and manual cleaning. Gastrointest Endosc. 48(2):137-42. 52. Claesson MH, A Rudolphi, K Kofoed, SS Poulsen, and J Reimann. 1996. CD4+T lymphocytes injected into severe combined immunodeficient (SCID) mice lead to an inflammatory and lethal bowel disease. Clin Exp Immunol. 104:491-500. 53. Cleary R. 1998. Clostridium difficile-associated diarrhea and colitis: clinical manifestations, diagnosis, and treatment. Dis Colon Rectum. 41(11):1435-49. 54. Clermont 0, C Cordevant, S Bonacorsi, A Marecat, M Lange, and E Bingen. 2001. Automated ribotyping provides rapid phylogenetic subgroup affiliation of clinical extraintestinal pathogenic Escherichia coli strains. J Clin Microbial. 39(12):4549-53. 55. Cole JR, B Chai, TL Marsh, RJ Farris, Q Wang, SA Kulam, S Chandra, DM McGarrell, TM Schmidt, GM Garrity, and JM Tiedje. 2003. The Ribosomal Database Project (RDP-ll): previewing a new autoaligner that allows regular updates and the new prokaryotic taxonomy. Nucleic Acids Res. 31(1):442-3. 56. Cole ST, and IS Girons. 1994. Bacterial genomics. FEMS Microbial Rev. 14:139-60. 57. Collins MD, PA Lawson, A Willems, JJ Cordoba, J Fernandez- Garayzabal, P Garcia, J Cai, H Hippe, and JAE Farrow. 1994. The phylogeny of the genus Clostridium: proposal of five new genera and eleven new species combinations. Int J Syst Bacterial. 44(4):812-26. 58. Cook IJ, Y Furukawa, V Panagopoulos, PJ Collins, and J Dent. 2000. Relationships between spatial patterns of colonic pressure and individual movements of content. Am J Physiol Gastrointest Liver Physiol. 278:G329-41. 59. Cooke EM, SP Ewins, J Hywel-Jones, and JE Lennard-Jones. 1974. Properties of strains of Escherichia coli carried in different phases of ulcerative colitis. Gut. 15:143-6.

-294- 60. Coppola N, G De Stefano, C Marrocco, F Scarano, C Scolastico, L Tarantino, F Piccinino, E Sagnelli, A Giorgio, and P Filippini. 2002. Absence of Helicobacter spp in the liver of patients with primary or metastatic liver cancer. Hepatology. 36(5):1300-1. 61. Croucher SC, AP Houston, CE Bayliss, and RJ Turner. 1983. Bacterial populations associated with different regions of the human colon wall. Appl Environ Microbial. 45(3):1025-33. 62. Crowther JS. 1971. Transport and storage of faeces for bacteriological examination. J Appl Bacterial. 34:477-83. 63. Cummings JH, and GT Macfarlane. 1991. The control and consequences of bacterial fermentation in the human colon. J Appl Bacterial. 70:443-59. 64. Cummings JH, and GT MacFarlane. 1997. Role of intestinal bacteria in nutrient metabolism. J Parenteral Enteral Nutr. 21(6):357-64. 65. D'Haens GR, K Geboes, M Peeters, F Baert, F Penninickx, and P Rutgeerts. 1998. Early lesions of recurrent Crohn's disease caused by infusion of intestinal contents in excluded ileum. Gastroenterology. 114(2):262-7. 66. Dahllof I, H Baillie, and S Kjelleberg. 2000. rpoB-bas·ed microbial community analysis avoids limitations inherent in 16S rRNA gene intraspecies heterogeneity. Appl Environ Microbial. 66(8):3376-80. 67. Darfeuille-Michaud A, C Neut, N Barnich, E Lederman, P DiMartino, P Desreumaux, L Gambiez, B Joly, A Cortot, and J-F Colombe!. 1998. Presence of adherent Escherichia coli strains in ileal mucosa of patients with Crohn's disease. Gastroenterology. 115(6): 1405-13. 68. De Champs C, N Maroncle, D Balestrino, C Rich, and C Forestier. 2003. Persistence of colonization of intestinal mucosa by a probiotic strain, Lactobacillus casei subsp. rhamnosus Lcr35, after oral consumption. J Clin Microbial. 41(3): 1270-3. 69. Dewhirst FE, C-C Chien, BJ Paster, RL Ericson, RP Orcutt, DB Schauer, and JG Fox. 1999. Phylogeny ofthe defined murine microbiota: Altered Schaedler Flora. Appl Environ Microbial. 65(8):3287-92.

-295- 70. Dewhirst FE, JG Fox, EN Mendes, BJ Paster, CE Gates, CA Kirkbride, and KA Eaton. 2000. 'Flexispira rappini' strains represent at least 10 Helicobacter taxa. Int J Syst Evol Microbial. 50: 1781-7. 71. Dewhirst FE, JG Fox, and SLW On. 2000. Recommended minimal standards for describing new species of the genus Helicobacter. Int J Syst Evol Microbial. 50:2231-7. 72. Dewhirst FE, C Seymour, GJ Fraser, BJ Paster, and JG Fox. 1994. Phylogeny of Helicobacter isolates from bird and swine faeces and description of Helicobacter pametensis sp. nov. Int J Syst Bacterial. 44:553-60. 73. Dianda L, AM Hanby, NA Wright, A Sebesteny, AC Hayday, and MJ

Owen. 1997. T cell receptor-a~-deficient mice fail to develop colitis in the absence of a microbial environment. Am J Pathol. 150(1):91-7. 74. Dieleman LA, A Arends, SL Tonkonogy, MS Goerres, DW Craft, W Grenther, RK Sellon, E Balish, and RB Sartor. 2000. Helicobacter hepaticus does not induce or potentiate colitis in interleukin-1 0-deficient mice. Infect Immun. 68(9):5107-13. 75. Donskey CJ, AM Hujer, SM Das, NJ Pultz, RA Bonomo, and LB Rice. 2003. Use of denaturing gradient gel electrophoresis for analysis of the stool microbiota of hospitalized patients. J Microbial Methods. 54:249-56. 76. Dore J, A Sghir, G Hannequart-Gramet, G Corthier, and P Pochart. 1998. Design and evaluation of a 16S rRNA-targeted oligonucleotide probe for specific detection and quantitation of human faecal Bacteroides populations. System Appl Microbial. 21:65-71. 77. Drasar BS, and M Shiner. 1969. Studies on the intestinal flora. Part II Bacterial flora of the small intestine in patients with gastrointestinal disorders. Gut. 10:812-9. 78. Drossman DA, P Funch-Jensen, J Janssens, NJ Talley, WG Thompson, and WE Whitehead. 1990. Identification of subgroups of functional bowel disorders. Gastroenterol Int. 3: 169-72.

-296- 79. Duchmann R, I Kaiser, E Hermann, W Mayet, K Ewe, and K-H Meyer Zum Buschenfelde. 1995. Tolerance exists towards resident intestinal flora but is broken in active inflammatory bowel disease(IBD). Clin Exp Immunol. 102:448-55. 80. Dutta S, A Chaterjee, P Dutta, K Rajendran, S Roy, KC Pramanik, and SK Buttacharya. 2001. Sensitivity and performance characteristics of a direct PCR with stool samples in comparison to conventional techniques for diagnosis of shigella and enteroinvasive Escherichia coli infection in children with acute diarrhoea in Calcutta. India J Med Microbial. 50:667-74. 81. Eaton KA, FE Dewhirst, BJ Paster, N Tzellas, BE Coleman, J Paola, and R Sherding. 1996. Prevalence and varieties ofHelicobacter species in dogs from random sources and pet dogs - animal and public health implications. J Clin Microbial. 34(12):3165-70. 82. Edwards CA, BI Duerden, and NW Read. 1985. Metabolism of mixed human colonic bacteria in a continuous culture mimicking the human cecal contents. Gastroenterology. 88(6):1903-9. 83. Eiseman B, W Silen, GS Bascom, and AJ Kauvar. 1958. Fecal enema as an adjunct in the treatment of pseudomembranous enterocolitis. Surgery. 44(5):854- 9. 84. Engberg J, SLW On, CS Harrington, and P Gerner-Smidt. 2000. Prevalence of Campylobacter, Arcobacter, Helicobacter, and Sutterella spp. in human fecal samples as estimated by a reevaluation of isolation methods for campylobacters. J Clin Microbial. 38(1):286-91. 85. Erdman S, JG Fox, CA Dangler, D Feldman, and BH Horwitz. 2001. Typhlocolitis in NF-KB-deficient mice. J Immunol. 166(3):1443-7. 86. Fabia R, A Ar'Rajab, M-L Johansson, R Andersson, R Willen, B Jeppsson, G Molin, and S Bengmark. 1993. Impairment of bacterial flora in human ulcerative colitis and experimental colitis in the rat. Digestion. 54:248-55. 87. Fallone CA, S Tran, M Semret, F Discepola, M Behr, and AN Barkun. 2003. Helicobacter DNA in bile: correlation with hepato-biliary diseases. Aliment Pharmacal Ther. 17:453-8.

-297- 88. Farrelly V, FA Rainey, and E Stackebrandt. 1995. Effect of genome size and rrn gene copy number on PCR amplification of 16S rRNA genes from a mixture of bacterial species. Appl Environ Microbial. 61(7):2798-801. 89. Favier CF, EE Vaughan, WM DeVos, and ADL Akkermans. 2002. Molecular monitoring of succession of bacterial communities in human neonates. Appl Environ Microbial. 68(1):219-26. 90. Fekety R. 1997. Guidelines for the diagnosis and management of Clostridium dif.ficile-associated diarrhea and colitis. Am J Gastroenterol. 92(5):739- 50. 91. Felsenstein J. 1989. PHYLIP --Phylogeny Inference Package (Version 3.2). Cladistics. 5:164-6. 92. Ferrero RL, and A Lee. 1988. Motility of Campylobacter jejuni in a viscous environment: comparison with conventional rod-shaped bacteria. J Gen Microbial. 134(Pt 1):53-9. 93. Finegold SM, HR Attebery, and VL Sutter. 1974. Effect of diet on human fecal flora: comparison of Japanese and American diets. Am J Clin Nutr. 27:1456- 69. 94. Fiocchi C. 1998. Inflammatory bowel disease: etiology and pathogenesis. Gastroenterology. 115:182-205. 95. Fischer SG, and LS Lerman. 1983. DNA fragments differing by single base-pair substitutions are separated in denaturing gradient gels: correspondence with melting theory. Proc NatlAcad Sci USA. 80:1579-83. 96. Flores BM, CL Fennell, L Kuller, MA Bronsdon, WR Morton, and WE Stamm. 1990. Experimental infection of pig-tailed macaques (Macaca nemestrina) with Campylobacter cinaedi and Campylobacter fennelliae. Infect Immun. 58(12):3947-53. 97. Florin THJ, GR Gibson, G Neale, and JH Cummings. 1990. A role for sulfate reducing bacteria in ulcerative colitis. Gastroenterology. 98:A170. 98. Foltz CJ, JG Fox, L Yan, and B Shames. 1995. Evaluation of antibiotic therapies for eradication of Helicobacter hepaticus. Antimicrob Agents Chemother. 39(6): 1292-4.

-298- 99. Fox JG, FE Dewhirst, Z Shen, Y Fen, NS Taylor, BJ Paster, RL Ericson, CN Lau, P Correa, JC Araya, and I Roa. 1998. Hepatic Helicobacter species identified in bile and gallbladder tissue from Chileans with chronic cholecystitis. Gastroenterology. 114(4):755-63. 100. Fox JG. 2002. The non-H. pylori helicobacters: their expanding role in gastrointestinal and systemic disease. Gut. 50:273-83. 101. Fox JG, FE Dewhirst, JG Tully, BJ Paster, L Yan, NS Taylor, MJ Collins, PL Gorelick, and JM Ward. 1994. Helicobacter hepaticus sp. nov., a microaerophilic bacterium isolated from livers and intestinal mucosal scrapings from mice. J Clin Microbial. 32(5): 1238-45. 102. Fox JG, PL Gorelick, MC Kullberg, ZM Ge, FE Dewhirst, and JM Ward. 1999. A novel urease-negative Helicobacter species associated with colitis and typhlitis in IL-10-deficient mice. Infect Immun. 67:1757-62. 103. Fox JG, L Handt, S Xu, Z Shen, FE Dewhirst, BJ Paster, CA Dangler, K Lodge, S Motzel, and H Klein. 2001. Novel Helicobacter species isolated from rhesus monkeys with chronic idiopathic colitis. J Med Microbial. 50:421-9. 104. Fox JG, X Li, L Yan, RJ Cahill, R Hurley, R Lewis, and JC Murphy. 1996. Chronic proliferative hepatitis in NJCr mice associated with persistent Helicobacter hepaticus infection: a model of Helicobacter-induced carcinogenesis. Infect Immun. 64(5):1548-58. 105. Fox JG, L Yan, B Shames, J Campbell, JC Murphy, and X Li. 1996. Persistent hepatitis and enterocolitis in germfree mice infected with Helicobacter hepaticus. Infect Immun. 64(9):3673-81. 106. Fox JG, LL Yan, FE Dewhirst, BJ Paster, B Shames, JC Murphy, A Hayward, JC Belcher, and EN Mendes. 1995. Helicobacter bilis sp. nov., a novel Helicobacter species isolated from bile, livers, and intestines of aged, inbred mice. J Clin Microbial. 33(2):445-54. 107. Franklin CL, PL Gorelick, LK Riley, FE Dewhirst, RS Livingston, JM Ward, CS Beckwith, and JG Fox. 2001. Helicobacter typhlonius sp. nov., a novel murine urease-negative Helicobacter species. J Clin Microbial. 39(11):3920-6.

-299- 108. Franklin CL, LK Riley, RS Livingston, CS Beckwith, CL Beschwilliford, and RR Hook. 1998. Enterohepatic lesions in SCID mice infected with Helicobacter bilis. Lab Anim Sci. 48:334-9. 109. Franklin CL, LK Riley, RS Livingston, CS Beckwith, RR Hook, CL Besch-Williford, R Hunziker, and PL Gorelick. 1999. Enteric lesions in SCID mice infected with "Helicobacter typhlonicus", a novel urease-negative Helicobacter species. Lab Anim Sci. 49:496-505. 110. Franks AH, HJM Harmsen, GC Raangs, GJ Jansen, F Schut, and GW Welling. 1998. Variations of bacterial populations in human feces measured by fluorescent in situ hybridisation with group-specific 16S rRNA-targeted oligonucleotide probes. Appl Environ Microbial. 64(9):3336-45. 111. Fuchs BM, G Wallner, W Beisker, I Schwippl, W Ludwig, and R Amann. 1998. Flow cytometric analysis of the in situ accessibility of Escherichia coli 16S rRNA for fluorescently labeled oligonucleotide probes. Appl Environ Microbial. 64(12):4973-82. 112. Fukuda K, T Kuroki, Y Tajima, N Tsuneoka, T Kitajima, S Matsuzaki, J Furui, and T Kanematsu. 2002. Comparative analysis of Helicobacter DNAs and biliary pathology in patients with and without hepatobiliary cancer. Carcinogenesis. 23(11):1927-31. 113. Furlong MA, DR Singleton, DC Coleman, and WB Whitman. 2002. Molecular and culture-based analyses of prokaryotic communities from an agricultural soil and the burrows and casts of the earthworm Lumbricus rubellus. Appl Environ Microbial. 68(3):1265-79. 114. Gatto NM, H Frucht, V Sundararajan, JS Jacobson, VR Grann, and AI Neugut. 2003. Risk of perforation after colonoscopy and sigmoidoscopy: a population-based study. J Natl Cancer Inst. 95(3):230-6. 115. Gebhart CJ, CL Fennell, MP Murtaugh, and WE Stamm. 1989. Campylobacter cinaedi is normal intestinal flora in hamsters. J Clin Microbial. 27(7):1692-4.

-300- 116. Giaffer MH, CD Holdsworth, and BI Duerden. 1991. The assessment of faecal flora in patients with inflammatory bowel disease by a simplified bacteriological technique. J Med Microbial. 35:238-43. 117. Giaffer MH, CD Holdsworth, and BI Duerden. 1992. Virulence properties of Escherichia coli strains isolated from patients with inflammatory bowel disease. Gut. 33:646-50. 118. Gibbons RJ, and B Kapsimalis. 1967. Estimates of the overall rate of growth of the intestinal microflora in hamsters, ginea pigs, and mice. J Bacterial. 93(1):510-2. 119. Gibson GR, ER Beatty, X Wang, and JH Cummings. 1995. Selective stimulation of bifidobacteria in the human colon by oligofructose and inulin. Gastroenterology. 108:975-82. 120. Gibson GR, JH Cummings, GT Macfarlane, CAllison, I Segal, HH Vorster, and ARP Walker. 1990. Alternative pathways for hydrogen disposal during fermentation in the human colon. Gut. 31:679-83. 121. Gibson GR, GT Macfarlane, and JH Cummings. 1988. Occurrence of sulphate-reducing bacteria in human faeces and the relationship of dissimilatory sulphate reduction to methanogenesis in the large gut. J Appl Bacterial. 65:103-111. 122. Gionchetti P, F Rizzello, U Helwig, A Venturi, KM Lammers, P Brigidi, B Vitali, G Poggioli, M Miglioli, and M Campieri. 2003. Prophylaxis ofpouchitis onset with probiotic therapy: a double-blind, placebo-controlled trial. Gastroenterology. 124(5):1202-9. 123. Gionchetti P, F Rizzello, A Venturi, P Brigidi, D Matteuzzi, G Bazzocchi, G Poggioli, M Miglioli, and M Campieri. 2000. Oral bacteriotherapy as maintenance treatment in patients with chronic pouchitis: a double-blind, placebo­ controlled trial. Gastroenterology. 119(2):305-9. 124. Gionchetti P, F Rizzello, A Venturi, and M Campieri. 2000. Probiotics in infective diarrhoea and inflammatory bowel diseases. J Gastroenterol Hepatol. 15(5):489-93. 125. Girardin SE, IG Boneca, LAM Carneiro, A Antignac, M Jehanno, J Viala, K Tedin, M-K Taha, A Labigne, U Zahringer, AJ Coyle, PS DiStefano, J

-301- Bertin, PJ Sansonetti, and DJ Philpott. 2003. Nod1 detects a unique muropeptide from gram-negative bacterial peptidoglycan. Science. 300(625):1584-7. 126. Godon J, E Zumstein, P Dabert, F Habouzit, and R Moletta. 1997. Molecular microbial diversity of an anaerobic digester as determined by small­ subunit rDNA sequence analysis. Appl Environ Microbial. 63(7):2802-13. 127. Goh KL, and N Parasakthi. 2001. The racial cohort phenomenon: seroepidemiology of Helicobacter pylori infection in a multiracial South-East Asian country. Eur J Gastroenterol Hepatol. 13(2):177-83. 128. Good IJ. 1982. An index of separateness of clusters and a permutation test for its significance. J Statist Comput Simul. 15:81-4. 129. Good IJ. 1953. The population frequencies of species and the estimation of population parameters. Biometrica. 40:237-64. 130. Goodfellow M, GP Manfio, and J Chun. 1997. Towards a practical species concept for cultivable bacteria, p. 25-59. In Claridge MF, Dawah HA, and Wilson MR. (ed.), Species: the units of biodiversity. Chapman & Hall, London. 131. Goossens H, M De Boeck, H Coignau, L Vlaes, C van den Borre, and J -P Butzler. 1986. Modified selective medium for isolation of Campylobacter spp. from feces: comparison with Preston medium, a blood-free medium, and a filtration system. J Clin Microbial. 24(5):840-3. 132. Gorbach SL. 2000. Probiotics and gastrointestinal health. Am J Gastroenterol. 95(1):S2-S4. 133. Gorbach SL, L Nahas, PI Lerner, and L Weinstein. 1967. Studies of intestinal microflora. I. Effects of diet, age, and periodic sampling on numbers of fecal microorganisms in man. Gastroenterology. 53(6):845-55. 134. Gorbach SL, L Nahas, AG Plaut, L Weinstein, JF Patterson, and R Levitan. 1968. Studies of intestinal microflora. V. Fecal microbial ecology in ulcerative colitis and regional enteritis: relationship to severity of disease and chemotherapy. Gastroenterology. 54(4):575-87. 135. Gorbach SL, G Neale, R Levitan, and GW Hepner. 1970. Alterations in human intestinal microflora during experimental diarrhoea. Gut. 11(1):1-6.

-302- 136. Gorbach SL, AG Plaut, L Nahas, L Weinstein, G Spanknebel, and R Levitan. 1967. Studies of intestinal microflora. IT. Microorganims of the small intestine and their relations to oral and fecal flora. Gastroenterology. 53(6):856-67. 137. Goto K, H Ohashi, A Takakura, and T Itoh. 2000. Current status of Helicobacter contamination of laboratory mice, rats, gerbils, and house musk shrews in Japan. Current Microbial. 41:161-6. 138. Grehan M, G Tamotia, B Robertson, and H Mitchell. 2002. Detection of Helicobacter colonization of the murine lower bowel by genus-specific PeR­ denaturing gradient gel electrophoresis. Appl Environ Microbial. 68(10):5164-6. 139. OzFoodNet Working Group. 2003. Foodbome disease in Australia: incidence, notifications and outbreaks. Annual report of the OzFoodNet network, 2002. Commun Dis Intell. 27(2):209-43. 140. Guarner F, and GJ Schaafsma. 1998. Probiotics. Int J Food Microbial. 39:237-8. 141. Gustafsson A, A Berstad, S Lund-Tonnesen, T Midtvedt, and E Norin. 1999. The effect of faecal enema on five microflora-associated characteristics in patients with antibiotic-associated diarrhoea. Scand J Gastroenterol. 6:580-5. 142. Gustafsson A, S Lund-Tonnesen, A Berstad, T Midtvedt, and E Norin. 1998. Faecal short-chain fatty acids in patients with antibiotic-associated diarrhoea, before and after faecal enema treatment. Scand J Gastroenterol. 33(7):721-7. 143. Gwee KA, JC Graham, MW McKendrick, SM Collins, JS Marshall, SJ Walters, and NW Read. 1996. Psychometric scores and persistence of irritable bowel after infectious diarrhoea. Lancet. 347:150-3. 144. Gwee KA, YL Leong, C Graham, MW McKendrick, SM Collins, SJ Walters, JE Underwood, and NW Read. 1999. The role of psychological and biological factors in postinfective gut dysfunction. Gut. 44:400-6. 145. Haenszel R, JW Berg, M Segi, M Kurihara, and FB Locke. 1973. Large­ bowel cancer in Hawaiian Japanese. J Natl Cancer Inst. 51:17 65-79. 146. Hampe J, A Cuthbert, PJ Croucher, MM Mirza, S Mascheretti, D Fischer, H Frenzel, K King, A Hasselmeyer, AJ MacPherson, S Bridger, S van Deventer, A Forbes, S Nikolaus, JE Lennard-Jones, UR Foelsch, M Krawkzak,

-303- CLewis, S Schreiber, and CG Mathew. 2001. Association between insertion mutation in NOD2 gene and Crohn's disease in German and British populations. Lancet. 357(9272):1925-8. 147. Hansen MC, T Tolker-Nielsen, M Givskov, and SMolin. 1998. Biased 16S rDNA PCR amplification caused by interference from DNA flanking the template region. FEMS Microbial Ecol. 26:141-9. 148. Harmsen HJM, GC Raangs, THe, JE Degener, and GW Welling. 2002. Extensive set of 16S rRNA-based probes for detection of bacteria in human feces. Appl Environ Microbial. 68(6):2982-90. 149. Harmsen HJM, ACM Wildeboer-Veloo, GC Raangs, AA Wagendorp, N Klijn, JG Bindels, and GW Welling. 2000. Analysis of intestinal flora development in breast-fed and formula-fed infants by using molecular identification and detection methods. J Ped Gastroenterol Nutr. 30:61-7. 150. Harper PH, ECG Lee, MGW Kettlewell, MK Bennett, and DP Jewell. 1985. Role of the faecal stream in the maintenance ofCrohn s colitis. Gut. 26:279- 84. 151. Heilig HGHJ, EG Zoetendal, EE Vaughan, P Marteau, ADL Akkermans, and WM de Vos. 2002. Molecular diversity of Lactobacillus spp. and other lactic acid bacteria in the human intestine as determined by specific amplification of 16S ribosomal DNA. Appl Environ Microbial. 68(1):114-23. 152. Hentges DJ, and BR Maier. 1972. Theoretical basis for anaerobic methodology. Am J Clin Nutr. 25:1299-305. 153. Hoffmann JC, NN Pawlowski, AA Kuhl, W Hohne, and M Zeitz. 2002. Animal models of inflammatory bowel disease: an overview. Pathobiology. 70:121- 30. 154. Hold LG, SE Pryde, VJ Russell, E Furrie, and HJ Flint. 2002. Assessment of microbial diversity in human colonic samples by 16S rDNA sequence analysis. FEMS Microbial Ecol. 39:33-9. 155. Holdeman LV, U Good, and WE Moore. 1976. Human fecal flora: variation in bacterial composition within individuals and a possible effect of emotional stress. Appl Environ Microbial. 31(3):359-75.

-304- 156. Hopkins MJ, R Sharp, and GT Macfarlane. 2001. Age and disease related changes in intestinal bacterial populations assessed by cell culture, 16S rRNA abundance, and community cellular fatty acid profiles. Gut. 48:198-205. 157. Hughes JB, JJ Hellmann, TH Ricketts, and BJM Bohannan. 2001. Counting the uncountable: statistical approaches to estimating microbial diversity. Appl Environ Microbial. 67(10):4399-406. 158. Huijsdens XW, RK Linskens, M Mak, SGM Meuwissen, CMJE Vandenbroucke-Grauls, and PHM Savelkoul. 2002. Quantification of bacteria adherent to gastrointestinal mucosa by real-time PCR. J Clin Microbial. 40(12):4423-7. 159. Humayoun SB, N Bano, and JT Hollibaugh. 2003. Depth distribution of microbial diversity in Mono Lake, a meromictic soda lake in California. Appl Environ Microbial. 69(2): 1030-42. 160. Hungate RE. 1950. The anaerobic mesophilic cellulolytic bacteria. Bacterial Rev. 14:1-50. 161. Imlay JA. 2002. How oxygen damages microbes: oxygen tolerance and obligate anaerobiosis. Advances Microb Physiol. 46:111-53. 162. Inglis GD, and LD Kalischuk. 2003. Use ofPCR for direct detection of Campylobacter species in bovine feces. Appl Environ Microbial. 69(6):3435-47. 163. Jacobsen CN, V Rosenfeldt Nielsen, AE Hayford, PL Moller, KF Michaelsen, A Paerregaard, B Sandstrom, M Tvede, and M Jakobsen. 1999. Screening of probiotic activities of forty-seven strains of Lactobacillus spp. by in vitro techniques and evaluation of the colonization ability of five selected strains in humans. Appl Environ Microbial. 65(11):4949-56. 164. Jansen GJ, ACM Wildeboer-Veloo, RHJ Tonk, AH Franks, and GW Welling. 1999. Development and validation of an automated, microscopy-based method for enumeration of groups of intestinal bacteria. J Microbial Methods. 37:215-21. 165. Johansson M-L, G Molin, B Jeppsson, S Nobaek, S Ahrne, and S Bengmark. 1993. Administration of different Lactobacillus strains in fermented

-305- oatmeal soup: in vivo colonization of human intestinal mucosa and effect on the indigenous flora. Appl Environ Microbial. 59(1):15-20. 166. Johnson LD, LM Ausman, PK Sehgal, and NW King. 1996. A prospective study of the epidemiology of colitis and colon cancer in Cotton-top Tamarins (Saguinus oedipus oedipus). Gastroenterology. 110(1):102-15. 167. Jukes TH, and CR Cantor. 1969. Evolution of protein molecules, p. 21- 132. In Munro HN (ed.), Mammalian Protein Metabolism. Academic Press, New York. 168. Kabir S. 2001. Detection of Helicobacter pylori in faeces by culture, PCR and enzyme immunoassay. J Med Microbial. 50(12):1021-9. 169. Kassie F, S Rabot, M Kundi, M Chabicovski, H-M Qin, and S Knasmuller. 2001. Intestinal microflora plays a crucial role in the genotxicity of the cooked food mutagen 2-amino-3-methylimidazo[4,5-f]quinoline (IQ). Carcinogenesis. 22(10):1721-5. 170. Kaul K, S Luke, C McGurn, N Snowden, C Monti, and W Fry. 1996. Amplification of residual DNA sequences in sterile bronchoscopes leading to false­ positive PCR results. J Clin Microbial. 34(9):1949-51. 171. Keohavong P, and WG Thilly. 1989. Fidelity of DNA polymerases in DNA amplification. Proc NatlAcad Sci USA. 86:9253-7. 172. Kim HJ, M Camilleri, S McKinzie, MB Lempke, DD Burton, GM Thomforde, and AR Zinsmeister. 2003. A randomized controlled trial of a probiotic, VSL#3, on gut transit and symptoms in diarrhoea-predominant irritable bowel syndrome. Aliment Pharmacal Ther.17:895-904. 173. Konrad A, M Mahler, B Flogerzi, MB Kalousek, J Lange, L Varga, and F Seibold. 2003. Amelioration of murine colitis by feeding a solution of lysed Escherichia coli. Scand J Gastroenterol. 38(2):172-9. 174. Koopman JP, HM Kennis, and JAM van Druten. 1978. Colonization resistance of the digestive tract and gastrointestinal transit time in SPF mice. Laboratory Animals. 12:223-6.

-306- 175. Kopczynski ED, MM Bateson, and DM Ward. 1994. Recognition of chimeric small-subunit ribosomal DNAs composed of genes from uncultivated microorganisms. Appl Environ Microbial. 60(2):746-8. 176. Kostia S, P Veijalainen, U Hirvi, and M-L Hanninen. 2000. Cytolethal distending toxin B gene (cdtB) homologues in taxa 2,3 and 8 in six canine isolates of Helicobacter sp. flexispira. J Med Microbial. 52:103-8. 177. Kotz S, and NL Johnson (ed.). 1985. Encyclopedia of statistical sciences, vol. 5. John Wiley & Sons, New York. 178. Kotz S, and NL Johnson (ed.). 1985. Encyclopedia of statistical sciences, vol. 2. John Wiley & Sons, New York. 179. Krevsky B, L Malmud, F D'Ercole, A Maurer, and R Fisher. 1986. Colonic transit scintigraphy. A physiologic approach to the quantitative measurement of colonic transit in humans. Gastroenterology. 91(5):1102-12. 180. Kruis W, E Schuts, P Fric, B Fixa, G Judmaier, and M Stolte. 1997. Double-blind comparison of an oral Escherichia coli preparation and mesalazine in maintaining remission of ulcerative colitis. Aliment Pharmacal Ther. 11:853-8. 181. Kuhn R, J Lohler, D Rennick, K Rajewsky, and W Muller. 1993. Interleukin-10-deficient mice develop chronic enterocolitis. Cell. 75:263-74. 182. Kulkarni SP, S Lever, JMJ Logan, AJ Lawson, J Stanley, and MS Shafi. 2002. Detection of Campylobacter species: a comparison of culture and polymerase chain reaction based methods. J Clin Pathol. 55:749-53. 183. Kullberg MC, JM Ward, PL Gorelick, P Caspar, S Hieny, A Cheever, D Jankovic, and ASher. 1998. Helicobacter hepaticus triggers colitis in specific­ pathogen-free interleukin-10 (ll.--10)-deficient mice through an ll.--12 and gamma interferon-dependent mechanism. Infect Immun. 66:5157-66. 185. Lala S, Y Ogura, C Osborne, SY Hor, A Bromfield, S Davies, 0 Ogunbyi, G Nunes, and S Kesazv. 2003. Crohn's disease and the NOD2 gene: a role for paneth cells. Gastroenterology. 125(1):47-57. 186. Langendijk PS, F Schut, GJ Jansen, GC Raangs, GR Kamphuis, MHF Wilkinson, and GW Welling. 1995. Quantitative fluorescence in situ hybridization

-307- of Bifidobacterium spp. with genus-specific 16S rRNA-targeted probes and its application to fecal samples. Appl Environ Microbial. 61(8):3069-75. 187. Lawson AJ, D Linton, J Stanley, and RJ Owen. 1997. Polymerase chain reaction detection and speciation of Campylobacter upsaliensis and C. helveticus in human faeces and comparison with culture techniques. J Appl Microbial. 83:375-80. 189. Lee A, JL O'Rourke, PJ Barrington, and TJ Trust. 1986. Mucus colonization as a determinant of pathogenicity in intestinal infection by Campylobacter jejuni: A mouse cecal model. Infect Immun. 51:536-46. 190. Lee A, MW Phillips, JL O'Rourke, BJ Paster, FE Dewhirst, GJ Fraser, JG Fox, LI Sly, PJ Romaniuk, TJ Trust, and S Kouprach. 1992. Helicobacter muridarum sp. nov., a microaerophilic helical bacterium with a novel ultrastructure isolated from the intestinal mucosa of rodents. Int J Syst Bacterial. 42(1):27-36. 191. Lehto E, and S Salminen. 1997. Inhibition of Salmonella typhimurium adhesion to Caco-2 cell cultures by Lactobacillus strain GG spent culture supernate: Only a pH effect? FEMS ImmunolMedMicrobiol.18:125-32. 192. Leser TD, JZ Amenuvor, TK Jensen, RK Lindecrona, M Boye, and K Moller. 2002. Culture-independent analysis of gut bacteria: the pig gastrointestinal tract microbiota revisited. Appl Environ Microbial. 68(2):673-90. 193. Levine J, CJ Ellis, JK Furne, JR Springfield, and MD Levitt. 1996. Sulfate reducing bacteria and ulcerative colitis. Gastroenterology. 110:A959. 194. Li X, JG Fox, MT Whary, LL Yan, B Shames, and Z Zhao. 1998. SCID/NCr mice naturally infected with Helicobacter hepaticus develop progressive hepatitis, proliferative typhlitis and colitis. Infect Immun. 66( 11) :5477-84. 195. Lidbeck A, J-A Gustafsson, and CE Nord. 1987. Impact of Lactobacillus acidophilus supplements on the human oropharyngeal and intestinal microflora. Scand J Infect Dis. 19:531-7. 196. Linskens RK, JG Kusters, XW Huijsdens, LA Dieleman, J Stoof, MM Gerrits, CM Vandenbroucke-Grauls, SG Meuwissen, and EJ Kuipers. 2000. Identification by PCR of Helicobacter species in intestinal mucosal samples of healthy humans and patients with IBD. Gastroenterology. 118(4,Suppl2,Part 1):A325.

-308- 197. Livingston PS, LK Riley, CL Beschwilliford, RR Hook, and CL Franklin. 1998. Transmission of Helicobacter hepaticus infection to sentinel mice by contaminated bedding. Lab Anim Sci. 48:291-3. 198. Loy A, A Lehner, N Lee, J Adamczyk, H Meier, J Ernst, K-H Schleifer, and M Wagner. 2002. Oligonucleotide microarray for 16S rRNA gene-based detection of all recognized lineages of sulfate-reducing prokaryotes in the environment. Appl Environ Microbial. 68(10):5064-81. 199. Macfarlane GT, and JH Cummings. 2002. Probiotics, infection and immunity. Curr Opin Infect Dis. 15:501-6. 200. Macfarlane GT, and GR Gibson. 1991. Sulphate-reducing bacteria, p. 201- 222. In Levett PN (ed.), Anaerobic microbiology: a practical approach. Oxford University Press, Oxford. 201. Macfarlane GT, GR Gibson, and JH Cummings. 1992. Comparison of fermentation reactions in different regions of the human colon. J Appl Bacterial. 72:57-64. 202. Mackie RI, A Sghir, and HR Gaskins. 1999. Developmental microbial ecology of the neonatal gastrointestinal tract. Am J Clin Nutr. 69(suppl): 1035S-45S. 203. Madsen KL, JS Doyle, LD Jewell, MM Tavernini, and RN Fedorak. 1999. Lactobacillus species prevents colitis in interleukin 10 gene-deficient mice. Gastroenterology.116(5):1101-14. 204. Madsen KL, JS Doyle, MM Tavernini, LD Jewell, RP Rennie, and RN Fedorak. 2000. Antibiotic therapy attenuates colitis in interleukin 10 gene-deficient mice. Gastroenterology. 118(6):1094-105. 205. Maggio-Price L, D Shows, K Waggie, A Burich, W Zeng, S Escobar, P Morrissey, and JL Viney. 2003. Helicobacter bilis infection accelerates and H. hepaticus infection delays the development of colitis in multiple drug resistance­ deficient (mdr1a-/-) mice. Am J Pathol. 160(2):739-51. 206. Maher M, C Finnegan, E Collins, B Ward, C Carroll, and M Cormican. 2003. Evaluation of culture methods and a DNA probe-based PCR assay for detection of Campylobacter species in clinical specimens of feces. J Clin Microbial. 41(7):2980-6.

-309- 207. Mahler M, HG Bedigian, BL Burgett, RJ Bates, ME Hogan, and JP Sundberg. 1998. Comparison of four diagnostic methods for detection of Helicobacter species in laboratory mice. Lab Anim Sci. 48:85-91. 208. Mahler M, IJ Bristol, JP Sundberg, GA Churchill, EH Birkenmeier, CO Elson, and EH Leiter. 1999. Genetic analysis of susceptibility of dextran sulfate sodium-induced colitis in mice. Genomics. 55:147-56. 209. Mahler M, C Most, S Schmidtke, JP Sundberg, R Li, HJ Hedrich, and GA Churchill. 2002. Genetics of colitis susceptibility in IL-l 0 deficient mice: backcross versus F2 results contrasted by principal component analysis. Genomics. 80(3):274-82. 210. Malchow H, K Ewe, JW Brandes, HM Goebell, H Ehms, H Sommer, and H Jesdinsky. 1984. European Cooperative Crohn's Disease Study (ECCDS): results of drug treatment. Gastroenterology. 86(2):249-66. 211. Maniatis T, EF Fritsch, and J Sambrook. 1982. Introduction of plasmid and Bacteriophage 1 DNA into Escherichia coli, p. 249-53, Molecular Cloning. (A laboratory manual). Cold Spring Harbor Laboratory, New York. 212. Marshall BJ, and JR Warren. 1984. Unidentified curved bacilli in the stomach of patients with gastritis and peptic ulceration. Lancet. i(8390): 1311-5. 213. Marteau P, P Pochart, J Dore, C Bera-Maillet, A BernaUer, and G Corthier. 2001. Comparative study of bacterial groups within the human cecal and fecal microbiota. Appl Environ Microbial. 67(10):4939-42. 214. Martin F, SAH Savage, AM Parrett, G Gramet, J Dore, and C Edwards. 2000. Dynamics of colonisation of the colon in breast-fed infants. Reprod Nutr Dev. 40:180. 215. Masseret E, J Boudeau, JF Colombel, C Neut, P Desreumaux, B Joly, A Cortot, and A Darfeuille-Michaud. 2001. Genetically related Escherichia coli strains associated with Crohn's disease. Gut. 48(3):320-5. 216. Matsuki T, K Watanabe, J Fujimoto, Y Miyamoto, T Takada, K Matsumoto, H Oyaizu, and R Tanaka. 2002. Development of 16S rRNA-gene­ targeted group-specific primers for the detection and identification of predominant bacteria in human feces. Appl Environ Microbial. 68(11):5445-51.

-310- 217. McCracken VJ, JM Simpson, RI Mackie, and HR Gaskins. 2001. Molecular ecological analysis of dietary and antibiotic-induced alterations of the mouse intestinal micro biota. J Nutr. 131(6): 1862-70. 218. McKendrick MW, and NW Read. 1994. Irritable bowel syndrome-post salmonella infection. J Infect. 29:1-3. 219. McOrist AL, M Jackson, and AR Bird. 2002. A comparison of five methods for extraction of bacterial DNA from human faecal samples. J Microbial Methods. 50:131-9. 220. Mendes EN, DMM Queiroz, FE Dewhirst, BJ Paster, SB Moura, and JG Fox. 1996. Helicobacter trogontum sp. nov., isolated from the rat intestine. Int J Syst Bacterial. 46:916-21. 221. Metcalf AM, SF Phillips, AR Zinsmeister, RL McCarty, RW Beart, and BG Wolff. 1987. Simplified assessment of segmental colonic transit. Gastroenterology. 92(1):40-7. 222. Miller TL, and MJ Wolin. 1982. Enumeration of Methanobrevibacter smithii in human feces. Arch Microbial. 131:14-18. 223. Miller TL, and MJ Wolin. 1983. Stability of Methanobrevibacter smithii populations in the microbial flora excreted from the human large bowel. Appl Environ Microbial. 45(1):317-8. 224. Mombaerts P, E Mizoguchi, MJ Grusby, LH Glimcher, AK Bhan, and S Tonegawa. 1993. Spontaneous development of inflammatory bowel disease in T cell receptor mutant mice. Cell. 75:275-82. 225. Montiero L, D Bonnemaison, A Vekris, KG Petry, J Bonnet, R Vidal, J Cabrita, and F Megraud. 1997. Complex polysaccharides as PCR inhibitors in feces: Helicobacter pylori model. J Clin Microbial. 35:995-8. 226. Moore WE, and LV Holdeman. 1974. Human fecal flora: the normal flora of20 Japanese-Hawaiians. ApplMicrobiol. 27:961-79. 227. Moore WEC, EP Cato, and LV Holdeman. 1978. Some current concepts in intestinal bacteriology. Am J Clin Nutr. 31:S33-S42.

-311- 228. Moore WEC, and LV Holdeman. 1975. Discussion of current bacteriological investigations of the relationships between intestinal flora, diet, and colon cancer. Cancer Res. 35:3418-20. 229. Moore WEC, and LV Holdeman. 1974. Special problems associated with the isolation and identification of intestinal bacteria in fecal flora studies. Am J Clin Nutr. 27:1450-5. 230. Moore WEC, and LH Moore. 1995. Intestinal floras of populations that have a high risk of colon cancer. Appl Environ Microbial. 61(9):3202-7. 231. Moter A, and UB Gobel. 2000. Fluorescence in situ hybridization (FISH) for direct visualization of microorganisms. J Microbial Methods. 41:85-112. 232. Moura SB, EN Mendes, DMM Queiroz, ERS Camargos, EF Fonseca, GA Rocha, and JR Nicoli. 1998. Ultrastructure of Helicobacter trogontum in culture and in the gastrointestinal tract of gnotobiotic mice. J Med Microbial. 47:513-20. 233. Musch MW, C Bookstein, Y Xie, JH Sellin, and EB Chang. 2001. SCFA increase intestinal Na+ absorption by induction ofNHE3 in rat colon and human intestinal C2/bbe cells: Am J Physiol Gastrointest Liver Physiol. 280:0687-0693. 234. Muyzer G, EC de Waal, and AG Uitterlinden. 1993. Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Appl Environ Microbial. 59(3):695-700. 235. Myers RM, SG Fischer, LS Lerman, and T Maniatis. 1985. Nearly all single base substitutions in DNA fragments joined to a GC-clamp can be detected by denaturing gradient gel electrophoresis. Nucleic Acids Res. 13:3131-45. 236. Neal KR, J Hebden, and R Spiller. 1997. Prevalence of gastrointestinal symptoms six months after bacterial gastroenteritis and risk factors for development of the irritable bowel syndrome: postal survey of patients. Br Med J. 314(7083):779- 82. 237. Neilan BA, D Jacobs, T Del Dot, LL Blackall, PR Hawkins, PT Cox, and A Goodman. 1997. rRNA sequences and evolutionary relationships among toxic

-312- and nontoxic Cyanobacteria of the genus Microcystis. Int J Syst Bacterial. 47(3):693-7. 238. Nelson GM, and SE George. 1995. Comparison of media for selection and enumeration of mouse fecal flora populations. J Microbial Methods. 22:293-300. 239. Nilsson HO, R Mulchandani, KG Tranberg, U Stenram, and T Wadstrom. 2001. Helicobacter species identified in liver from patients with cholangiocarcinoma and hepatocellular carcinoma. Gastroenterology. 120(1):323-4. 240. Nilsson HO, J Taneera, M Castedal, E Glatz, R Olsson, and T Wadstrom. 2000. Identification of Helicobacter pylori and other Helicobacter species by PCR, hybridization, and partial DNA sequencing in human liver samples from patients with primary sclerosing cholangitis or primary biliary cirrhosis. J Clin Microbial. 38(3):1072-6. 241. Nobaek S, M-L Johansson, G Molin, S Ahrne, and B Jeppsson. 2000. Alteration of intestinal microflora is associated with reduction of abdominal bloating and pain in patients with Irritable Bowel Syndrome. Am J Gastroenterol. 95(5):1231-8. 242. Noble JI, and FP Tally. 1984. Metronidazole, p. 255-62. In Ristuccia AM and Cunha BC (ed.), Antimicrobial Therapy. Raven Press, New York. 243. Nubel U, B Engelen, A Felske, J Snaidr, A Wieshuber, RI Amann, W Ludwig, and H Backhaus. 1996. Sequence heterogeneities of genes encoding 16S rRNAs in Paenibacillus polymyxa detected by temperature gradient gel electrophoresis. J Bacterial. 178(19):5636-43. 244. Nubel U, F Garcia-Piche!, M Kuhl, and G Muyzer. 1999. Quantifying microbial diversity: morphotypes, 16S rRNA genes, and carotenoids of oxygenic phototrophs in microbial mats. Appl Environ Microbial. 65(2):422-30. 245. O'Connor HJ, KM Newbold, WJ Alexander, H Thompson, J Drumm, and lA Donovan. 1989. Effect of Roux-en-Y biliary diversion on Campylobacter pylori. Gastroenterology. 97:958-64. 246. O'Rourke JL, M Grehan, and A Lee. 2001. Non-pylori helicobacter species in humans. Gut. 49:601-6.

-313- 247. Oberhelman HA, S Kohatsu, KB Taylor, and RM Kivel. 1968. Diverting ileostomy in the surgical management of Crohn's disease of the colon. Am J Surg. 115:231-9. 248. Ogura Y, DK Bonen, N Inohara, DL Nicolae, FF Chen, R Ramos, H Britton, T Moran, R Karaliuskas, RH Duerr, J Achkar, SR Brant, TM Bayless, BS Kirschner, SB Hanauer, G Nunez, and JH Cho. 2001. A frameshift mutation in NOD2 associated with susceptibility to Crohn's disease. Nature. 411:603-606. 249. Oliveira AG, MGP Sanna, GA Rocha, AMC Rocha, A Santos, R Dani, FP Marinho, LPF Castro, and DMM Queiroz. 2004. Helicobacter species in the intestinal mucosa of patients with ulcerative colitis. J Clin Microbial. 42(1):384-6. 250. Oozer R, N Goupil-Feuillerat, CA Alpert, M van de Guchte, J Anba, J Mengaud, and G Corthier. 2002. Lactobacillus casei is able to survive and initiate protein synthesis during its transit in the digestive tract of human flora-associated mice. Appl Environ Microbial. 68(7):3570-4. 251. Paterson DL, J Iredell, and M Whitby. 1994. Putting back the bugs: bacterial treatment relieves chronic diarrhoea. Med J Aust. 160(21):232-3. 252. Peach S, MR Lock, D Katz, IP Todd, and S Tabaqchali. 1978. Mucosal­ associated bacterial flora of the intestine in patients with Crohn's disease and in a control group. Gut. 19:1034-42. 253. Peeters M, B Geypens, D Claus, H Nevens, Y Ghoos, G Verbeke, F Baert, S Vermeire, R Vlietinck, and P Rutgeerts. 1997. Clustering of increased small intestinal permeability in families with Crohn's disease. Gastroenterology. 113(3):802-7. 254. Pellett PE, TJ Spira, 0 Bagasra, C Boshoff, L Corey, L de Lellis, M Huang, J Lin, S Matthews, P Monini, P Rimessi, C Sosa, C Wood, and JA Stewart. 1999. Multicenter comparison ofPCR assays for detection of Human Herpesvirus 8 DNA in semen. J Clin Microbial. 37(5):1298-301. 255. Peplies J, FO Glockner, and R Amann. 2003. Optimization strategies for DNA microarray-based detection of bacteria with 16S rRNA-targeting oligonucleotide probes. Appl Environ Microbial. 69(3):1397-407.

-314- 256. Persky SE, and LJ Brandt. 2000. Treatment of recurrent Clostridium difficile-associated diarrhea by administration of donated stool directly through a colonoscope. Am J Gastroenterol. 95(11):3283-5. 257. Phillips M, A Lee, and WD Leach. 1978. The mucosa-associated microflora of the rat intestine: A study of normal distribution and magnesium sulphate induced diarrhoea. Aust J Exp Biol Med Sci. 56(6):649-62. 258. Picon L, M Lemann, B Flourie, J-C Rambaud, J-D Rain, and R Jain. 1992. Right and left colonic transit after eating assessed by a dual isotopic technique in healthy humans. Gastroenterology. 103(1):80-5. 259. Pochapin M. 2000. The effect of probiotics on Clostridium difficile diarrhea. Am J Gastroenterol. 95(1):S11-13. 260. Pochart P, J Dore, F Lemann, I Goderel, and J-C Rambaud. 1992. Interrelations between populations of methanogenic archaea and sulfate-reducing bacteria in the human colon. FEMS Microbial Lett. 98:225-8. 261. Potezny N, J McClure, AM Rofe, and RA Conyers. 1986. The long-term effect of dietary administration of refined sugars and sugar alcohols on plasma biochemistry, urine biochemistry and tissue histology in mice given a limited degree of dietary self-selection. Food Chem Toxicol. 24(5):389-96. 262. Poulsen LK, G Ballard, and DA Stahl. 1993. Use ofrRNA fluorescence in situ hybridization for measuring the activity of single cells in young and established biofilms. Appl Environ Microbial. 59(5):1354-60. 263. Prasad J, H Gill, J Smart, and PK Gopal. 1998. Selection and characterisation of Lactobacillus and Bifidobacterium strains for use as probiotics. Int Dairy J. 8:993-1002. 264. Pryde SE, SH Duncan, GL Hold, CS Stewart, and HJ Flint. 2002. The microbiology of butyrate formation in the human colon. FEMS Microbial Lett. 217:133-9. 265. Pullan RD, GAO Thomas, M Rhodes, RG Newcombe, GT Williams, A Allen, and J Rhodes. 1994. Thickness of adherent mucus gel on colonic mucosa in humans and its relevance to colitis. Gut. 35(3):353-9.

-315- 266. Quinn TC, SE Goodell, C Fennell, SP Wang, MD Schuffler, KK Holmes, and WE Stamm. 1984. Infections with Campylobacter jejuni and Campylobacter­ like organisms in homosexual men. Ann Intern Med. 101(2):187-92. 267. Quinn TC, WE Stamm, SE Goodell, E l\1krtichian, J Benedetti, L Corey, MD Schuffler, and KK Holmes. 1983. The polymicrobial origin of intestinal infections in homosexual men. N Engl J Med. 309(10):576-82. 268. Rachmilewitz D, F Karmeli, K Takabayashi, T Hayashi, L Leider-Trejo, J Lee, LM Leoni, and E Raz. 2002. Immunostimulatory DNA ameliorates experimental and spontaneous murine colitis. Gastroenterology. 122(5):1428-41. 269. Rainey FA, N Ward, LI Sly, and E Stackebrandt. 1994. Dependence of the taxon composition of clone libraries for PCR amplified, naturally occurring 16s rDNA, on the primer pair and the cloning system used. Experientia. 50:796-7. 270. Rao SS, P Sadeghi, J Beaty, R Kavlock, and K Ackerson. 2001. Ambulatory 24-h colonic manometry in healthy humans. Am J Physiol Gastrointest Liver Physiol. 280:G629-G639. 271. Rath HC, HH Herfarth, JS Ikeda, WB Grenther, TE Hamm, E Balish, JD Taurog, RE Hammer, KH Wilson, and RB Sartor. 1996. Normal luminal bacteria, especially Bacteroides species, mediate chronic colitis, gastritis, and arthritis in HLA-B27/Human B microglobulin transgenic rats. J Clin Invest. 2 98:945-53. 272. Rath HC, M Schultz, R Freitag, LA Dieleman, F Li, H-J Linde, J Scholmerich, and B Sartor. 2001. Different subsets of enteric bacteria induce and perpetuate experimental colitis in rats and mice. Infect Immun. 69(4):2277-85. 273. Rembacken BJ, AM Snelling, PM Hawkey, DM Chalmers, and ATR Axon. 1999. Non-pathogenic E. coli vs mesalazine for the treatment of ulcerative colitis: a randomised trial. Lancet. 354:635-9. 274. Ricci N, and M Caselli. 1983. Rectal infusions of bacterial preparations for intestinal disorders. Lancet. ii: 1494. 275. Riley LK, CL Franklin, RR Hook, and C Beschwilliford. 1996. Identification of murine helicobacters by PCR and restriction enzyme analyses. J Clin Microbial. 34:942-6.

-316- 276. Ringel Y, AD Sperber, and DA Drossman. 2001. Irritable bowel syndrome. Annu Rev Med. 52:319-38. 277. Ristuccia AM, and BC Cunha. 1984. Vancomycin, p. 265-70. In Ristuccia AM and Cunha BC (ed.), Antimicrobial Therapy. Raven Press, New York. 278. Robertson BR, JL O'Rourke, P Vandamme, SOn, and A Lee. 2001. Helicobacter ganmani sp. nov., a urease-negative anaerobe isolated from the intestines of laboratory mice. Int J Syst Evol Microbial. 51: 1881-9. 279. Roediger WEW. 1980. The role of anaerobic bacteria in the metabolic welfare of the colonic mucosa of man. Gut. 21:793-8. 280. Roediger WEW. 1982. Utilization of nutrients by isolated epithelial cells of the rat colon. Gastroenterology. 83(2):424-9. 281. Roosendaal R, EJ Kuipers, AJ van den Brule, AS Pena, SG Meuwissen, JM Walboomers, and J de Graaff. 1993. Detection of Helicobacter pylori DNA by PCR in gastrointestinal equipment. Lancet. 341(8849):900. 282. Roosendaal R, EJ Kuipers, CM Vandenbroucke-Grauls, and JG Kusters. 2002. Helicobacter species are not detectable by 16S rDNA PCR in bile from Dutch patients with common bile duct stones. Digestion. 66(2):89-91. 283. Rudi J, A Rudi, M Maiwald, and W Stremmel. 1999. Helicobacter sp. are not detectable in bile from German patients with biliary disease. Gastroenterology. 116(4):1016-7. 284. Rutgeerts P, K Goboes, M Peeters, M Hiele, F Penninckx, R Aerts, R Kerremans, and G Vantrappen. 1991. Effect of faecal stream diversion on recurrence ofCrohn s disease in the neoterminal ileum. Lancet. 338:771-4. 285. Rutgeerts P, M Hiele, K Gerboes, M Peeters, P F., R Aerts, and R Kerremans. 1995. Controlled trial of metronidazole for prevention of Crohn's recurrence after ileal resection. Gastroenterology. 108(6): 1617-21. 286. Saavedra JM, NA Bauman, I Oung, JA Perman, and RH Yolken. 1994. Feeding of Bijidobacterium bifidum, and Streptococcus thermophilus to infants in hospital for prevention of diarrhea, and shedding of rotavirus. Lancet. 344:1046-9.

-317- 287. Sadlack B, H Merz, H Schorle, A Schimpl, AC Feller, and I Horak. 1993. Ulcerative colitis-like disease in mice with a disrupted interleukin-2 gene. Cell. 75:253-61. 288. Salzman NH, H de Jong, Y Paterson, HJM Harmsen, GW Welling, and NA Bos. 2002. Analysis of 16S libraries of mouse gastrointestinal microflora reveals a large new group of mouse intestinal bacteria. Microbiology. 148:3651-60. 289. Sambrook J, EF Fritsch, and T Maniatis. 1989. Isolation of DNA fragments from polyacrylamide gels, p. 6.46-6.48. In Nolan C (ed.), Molecular Cloning. A Laboratory Manual, 2 ed, vol. 1. Cold Spring Harbor Laboratory Press, Plainview. 290. Sasaki K, Y Tajiri, M Sata, Y Fujii, F Matsubara, M Zhao, S Shimizu, A Toyonaga, and K Tanikawa. 1999. Helicobacter pylori in the natural environment. Scand J Infect Dis. 31(3):275-9. 291. Satokari RM, EE Vaughan, AD Akkermans, M Saarela, and WM de Vos. 2001. Bifidobacterial diversity in human feces detected by genus-specific PCR and denaturing gradient gel electrophoresis. Appl Environ Microbial. 67(2):504-13. 292. Satsangi J, J Morecroft, NB Shah, and E Nimmo. 2003. Genetics of inflammatory bowel disease: scientific and clinical implications. Baillieres Best Pract Res Clin Gastroenterol. 17(1):3-18. 293. Saunders KE, Z Shen, FE Dewhirst, BJ Paster, CA Dangler, and JG Fox. 1999. Novel intestinal Helicobacter species isolated from Cotton-Top Tamarins (Saguinus oedipus) with chronic colitis. J Clin Microbial. 37(1):146-51. 294. Savage DC. 1970. Associations of indigenous microorganisms with gastrointestinal mucosal epithelia. Am J Clin Nutr. 24:1495-1501. 295. Savage DC, and R Dubos. 1968. Alterations in the mouse caecum and its flora produced by antibacterial drugs. J Exp Med. 128:97-110. 296. Savarino SJ. 1993. Enteroadherent Escherichia coli: a heterogeneous group of E. coli implicated as diarrhoeal pathogens. Trans R Soc Trap Med Hyg. 87(suppl 3):49-53.

-318- 297. Saxelin M, S Elo, S Salminen, and H Vapaatalo. 1991. Dose response colonisation of faeces after oral administration of Lactobacillus casei strain GG. Microb Ecol Health Dis. 4:209-14. 298. Schaedler RW, R Dubos, and R Costello. 1965. Association of germfree mice with bacteria isolated from normal mice. J Exp Med. 122:77-83. 299. Schaedler RW, R Dubos, and R Costello. 1965. The development of the bacterial flora in the gastrointestinal tract of mice. J Exp Med. 122:59-66. 300. Schauer DB, N Ghori, and S Falkow. 1993. Isolation and characterization of "Flexispira rappini" from laboratory mice. J Clin Microbial. 31(10):2709-14. 301. Schramm A, CM Santegoeds, HK Nielsen, H Ploug, M Wagner, M Pribyl, J Wanner, R Amann, and D de Beer. 1999. On the occurrence of anoxic microniches, denitrification, and sulfate reduction in aerated activated sludge. Appl Environ Microbial. 65(9):4189-96. 302. Schultsz C, M Moussa, R van Ketel, GNJ Tytgat, and J Dankert. 1997. Frequency of pathogenic and enteroadherent Escherichia coli in patients with inflammatory bowel disease and controls. J Clin Pathol. 50(7):573-9. 303. Schultsz C, FM Van Den Berg, FW Ten Kate, GNJ Tytgat, and J Dankert. 1999. The intestinal mucus layer from patients with inflammatory bowel disease harbors high numbers of bacteria compared with controls. Gastroenterology. 117(5):1089-97. 304. Schultz M, C Velptkampe, LA Dieleman, WB Grenther, PB Wyrick, SL Tonkonogy, and B Sartor. 2002. Lactobacillus plantarum 299V in the treament and prevention of spontaneous colitis in interleukin-1 0-defecient mice. Inflamm Bowel Dis. 8:71-80. 305. Schultz M, C Veltkampe, LA Dieleman, PB Wyrick, SL Tonkonogy, and RB Sartor. 1998. Continuous feeding of Lactobacillus plantarum attenuates· established colitis in interleukin-10 deficient mice. Gastroenterology. 114:Al081. 306. Schwan A, S Sjolin, U Trottestam, and B Aronsson. 1984. Relapsing Clostridium difficile enterocolitis cured by rectal infusion of normal faeces. Scand J Infect Dis. 16:211-5.

-319- 307. Sears HJ, and I Brownlee. 1952. Further observations on the persistence of individual strains of Escherichia coli in the intestinal tract of man. J Bacterial. 63:47-57. 308. Sears HJ, I Brownlee, and JK Uchiyama. 1950. Persistence of individual strains of Escherichia coli in the intestinal tract of man. J Bacterial. 59:293-301. 309. Sekiguchi H, N Tomioka, T Nakahara, and H Uchiyama. 2001. A single band does not always represent single bacterial strains in denaturing gradient gel electrophoresis analysis. Biotechnol Lett. 23:1205-8. 310. Seksik P, L Rigottier-Gois, G Gramet, M Sutren, P Pochart, P Marteau, R Jian, and J Dore. 2003. Alterations of the dominant faecal bacterial groups in patients with Crohn's disease of the colon. Gut. 52:237-42. 311. Sellon RK, S Tonkonogy, M Schultz, LA Dieleman, W Grenther, E Balish, DM Rennick, and RB Sartor. 1998. Resident enteric bacteria are necessary for the development of spontaneous colitis and immune system activation in interleukin-10-deficientmice. Infect Immun. 66(11):5224-31. 312. Sen S, MM Mullan, TJ Parker, JT Woolner, SA Tarry, and JO Hunter. 2002. Effect of Lactobacillus plantarum 299v on colonic fermentation and symptoms of irritable bowel syndrome. Dig Dis Sci. 47(11):2615-20. 313. Sghir A, G Gramet, A Suau, V Rochet, P Pochart, and J Dore. 2000. Quantification of bacterial groups within human fecal flora by oligonucleotide probe hybridization. Appl Environ Microbial. 66(5):2263-6. 314. Shames B, JG Fox, F Dewhirst, LL Yan, ZL Shen, and NS Taylor. 1995. Identification of widespread Helicobacter hepaticus infection in feces in commercial mouse colonies by culture and PCR assay. J Clin Microbial. 33:2968-72. 315. Shen W, H Steinruck, and A Ljungh. 1995. Expression of binding of plasminogen, thrombospondin, vitronectin, and fibrinogen, and adhesive properties by Escherichia coli strains isolated from patients with colonic diseases. Gut. 36:401- 6. 316. Shen Z, JG Fox, FE Dewhirst, BJ Paster, CJ Foltz, LL Yan, B Shames, and L Perry. 1997. Helicobacter rodentium sp. nov., a urease-negative Helicobacter species isolated from laboratory mice. Int J Syst Bacterial. 47:627-34.

-320- 317. Sheu BS, SC Lee, PW Lin, ST Wang, YC Chang, HB Yang, CH Chuang, and XZ Lin. 2000. Carbon urea breath test is not as accurate as endoscopy to detect Helicobacter pylori after gastrectomy. Gastrointest Endosc. 51(6):670-5. 318. Shiner M. 1963. A capsule for obtaining sterile samples of gastrointestinal fluids. Lancet. 1:532-3. 319. Shomer NH, CA Dangler, RP Marini, and JG Fox. 1998. Helicobacter bilis/Helicobacter rodentium co-infection associated with diarrhoea in a colony of scid mice. Lab Anim Sci. 48(5):455-9. 320.Shomer NH, CA Dangler, MD Schrenzel, and JG Fox. 1997. Helicobacter bilis-induced inflammatory bowel disease in SCID mice with defined flora. Infect Immun. 65:4858-64. 321. Simon GL, and SL Gorbach. 1984. Intestinal flora in health and disease. Gastroenterology. 86:174-93. 322. Simpson JM, VJ McCracken, HR Gaskins, and RI Mackie. 2000. Denaturing gradient gel electrophoresis analysis of 16S ribosomal DNA amplicons to monitor changes in fecal bacterial populations of weanling pigs after introduction of Lactobacillus reuteri strain MM53. Appl Environ Microbial. 66(11):4705-14. 324. Singleton DR, MA Furlong, SL Rathbun, and WB Whitman. 2001. Quantitative comparisons of 16S rRNA gene sequence libraries from environmental samples. Appl Environ Microbial. 67(9):4374-6. 325. Small J, DR Call, FJ Brockman, TM Straub, and DP Chandler. 2001. Direct detection of 16S rRNA in soil extracts by using oligonucleotide microarrays. Appl Environ Microbial. 67(10):4708-16. 326. Sonnenwirth AC. 1972. Evolution of anaerobic methodology. Am J Clin Nutr. 25:1295-8. 327. Spiller RC, D Jenkins, JP Thornley, JM Hebden, T Wright, M Skinner, and KR Neal. 2000. Increased rectal mucosal enteroendocrine cells, T lymphocytes, and increased gut permeability following acute Campylobacter enteritis and in post­ dysenteric irritable bowel syndrome. Gut. 47(6):804-11.

-321- 328. Stackebrandt E, and BM Goebel. 1994. Taxonomic note: a place for DNA­ DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology. Int J Syst Bacterial. 44(4):846-9. 329. Stanley J, D Linton, AP Burnens, FE Dewhirst, SLW On, A Porter, RJ Owen, and M Costas. 1994. Helicobacter pullorum sp. nov.- genotype and phenotype of a new species isolated from poultry and from human patients with gastroenteritis. Microbiology. 140:3441-9. 330. Stanley J, D Linton, AP Burnens, FE Dewhirst, RJ Owen, A Porter, SLW On, and M Costas. 1993. Helicobacter canis sp. nov., a new species from dogs: an integrated study of phenotype and genotype. J Gen Microbial. 139:2495- 504. 331. Stark PL, and A Lee. 1982. The microbial ecology of the large bowel of breast-fed and formula-fed infants during the first year of life. J Med Microbial. 15:189-203. 332. Stebbings S, K Munro, MA Simon, G Tannock, J Highton, H Harmsen, G Welling, P Seksik, J Dore, G Grame, and A Tilsala-Timisjarvi. 2002. Comparison of the faecal microflora of patients with ankylosing spondylitis and controls using molecular methods of analysis. Rheumatology. 41(41):1395-401. 333. Steele TW, and SN McDermott. 1984. The use of membrane filters applied directly to the surface of agar plates for the isolation of Campylobacter jejuni from feces. Pathology. 16:263-5. 334. Stephen AM, and JH Cummings. 1980. The microbial contribution to human faecal mass. J Med Microbial. 13:45-56. 335. Suau A, R Bonnet, M Sutren, J Godon, GR Gibson, MD Collins, and J Dore. 1999. Direct analysis of genes encoding 16S rRNA from complex communities reveals many novel molecular species within the human gut. Appl Environ Microbial. 65(11):4799-807. 336. Suerbaum S, C Josenhans, T Sterzenbach, B Drescher, P Brandt, M Bell, M Droge, B Fartmann, H-P Fischer, G Zhongming, A Horster, R Holland, K Klein, J Konig, L Macko, GL Mendz, G Nyakatura, DB Schauer, Z Shen, J Weber, M Frosch, and JG Fox. 2003. The complete genome sequence of the

-322- carcinogenic bacterium Helicobacter hepaticus. Proc Natl Acad Sci USA. 100(13):7901-6. 337. Sutherland L, J Singleton, J Sessions, S Hanauer, E Krawitt, G Rankin, R Summers, H Mekhjian, N Greenberger, M Kelly, J Levine, A Thomson, E Alpert, and E Prokipchuk. 1991. Double blind, placebo controlled trial of metronidazole in Crohn's disease. Gut. 32:1071-5. 338. Sutton CL, J Kim, A Yamane, H Dalwadi, B Wei, C Landers, SR Targan, and J Braun. 2000. Identification of a novel bacterial sequence associated with Crohn's disease. Gastroenterology. 119(1):23-31. 339. Suzuki MT, and SJ Giovannoni. 1996. Bias caused by template annealing in the amplification of mixtures of 16S rRNA genes by PCR. Appl Environ Microbial. 62(2):625-30. 340. Swidsinski A, A Ladhoff, A Pernthaler, S Swinsinski, V Loening­ Baucke, M Ortner, J Weber, U Hoffmann, S Schreiber, M Dietel, and H Lochs. 2002. Mucosal flora in Inflammatory Bowel Disease. Gastroenterology. 122(1):44- 54. 341. Tajima K, RI Aminov, T Nagamine, H Matsui, M Nakamura, andY · Benno. 2001. Diet-dependent shifts in the bacterial population ofthe rumen revealed with real-time PCR. Appl Environ Microbial. 67(6):2766-74. 342. Tajima K, T Nagamine, H Matsui, M Nakamura, and RI Aminov. 2001. Phylogenetic analysis of archaeal16S rRNA libraries from the rumen suggests the existence of a novel group of archaea not associated with known methanogens. FEMS Microbial Lett. 200:67-72. 343. Takeuchi A, HA Jervis, H Nakazawa, and DM Robinson. 1974. Spiral­ shaped organisms on the surface colonic epithelium of the monkey and man. Am J ClinNutr. 27:1287-96. 344. Talley NJ, P Boyce, and M Jones. 1998. Identification of distinct upper and lower gastrointestinal symptom groupings in an urban population. Gut. 42(5):690-5. 345. Tannock GW, K Munro, HJM Harmsen, GW Welling, J Smart, and PK Go pal. 2000. Analysis of the fecal microflora of human subjects consuming a

-323- probiotic product containing Lactobacillus rhamnosus DR20. Appl Environ Microbial. 66(6):2578-88. 346. Thadepalli H, MA Lou, VT Bach, TK Matsui, and AK Mandai. 1979. Microflora of the human small intestine. Am J Surg. 138:845-50. 347. Thjodleifsson B, G Sigthorsson, N Cariglia, I Reynisdottir, DF Gudbjartsson, K Kristjansson, JB Meddings, V Gudnason, JH Wandall, LP Andersen, R Sherwood, M Kjeld, E Oddsson, H Gudjonsson, and I Bjarnason. 2003. Subclinical intestinal inflammation: an inherited abnormality in Crohn's Disease relatives? Gastroenterology. 124:1728-37. 348. Thompson JD, DG Higgins, and TJ Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucl Acids Res. 22:4673-80. 349. Tiveljung A, JD Soderholm, G Olaison, J Jonasson, and HJ Monstein. 1999. Presence of eubacteria in biopsies from Crohn's disease inflammatory lesions as determined by 16S rRNA gene-based PCR. J Med Microbial. 48(3):263-8. 350. Totten PA, CL Fennell, FC Tenover, JM Wezenberg, PL Perine, WE Stamm, and KK Holmes. 1985. Campylobacter cinaedi (sp. nov.) and Campylobacter fennelliae (sp. nov.): Two new Campylobacter species associated with enteric disease in homosexual men. J Infect Dis. 151(1):131-9. 351. Tuleu C, C Andrieux, P Boy, and JC Chaumeil. 1999. Gastrointestinal transit of pellets in rats: effect of size and density. Int J Pharm. 180:123-131. 352. Tvede M, and J Rask-Madsen. 1989. Bacteriotherapy for chronic relapsing Clostridium difficile diarrhoea in six patients. Lancet: 1156-60. 353. Ursing B, T Aim, F Barany, I Bergelin, K Garnot-Norlin, J Hoevels, B Huitfeld, G Jarnerot, U Krause, A Krook, B Lindstrom, 0 Nordle, and A Rosen. 1982. A comparative study of metronidazole and sulfasalazine for active Crohn's disease: the cooperative Crohn's disease study in Sweden. IT Results. Gastroenterology. 83:550-62.

-324- 354. Vaahtovuo J, P Toivanen, and E Eerola. 2001. Study of murine faecal microflora by cellular fatty acid analysis; effect of age and mouse strain. Antonie van Leeuwenhoek. 80:35-42. 355. van de Merwe JP, AM Schroder, F Wensinck, and MP Hazenberg. 1988. The obligate anaerobic faecal flora of patients with Crohn's disease and their first­ degree relatives. Scand J Gastroenterol. 23:1125-31. 356. van den Bogaard AE, WF Weidema, CP van Boven, and D van der Waay. 1986. Recolonization and colonization resistance of the large bowel after three methods of preoperative preparation of the gastrointestinal tract for elective colorectal surgery. J Hyg. 97(1):49-59. 357. van der Waaij D, and JM Berghuis. 1973. Determination of the colonization resistance of the digestive tract of individual mice. J Hyg. 72:379-87. 358. van der Waaij D, JM Berghuis-deVries, and JEC Lekkerkerk-van der Wees. 1971. Colonization resistance of the digestive tract in conventional and antibiotic treated mice. J Hyg. 69:405-11. 359. van der Waaij D, JEC Lekkerkerk-van der Wees, and PJ Heidt. 1974. Clearance of antibiotics from the intestines after termination of antibiotic decontamination. J Hyg. 73:409-14. 360. van der Waaij D, JM Vossen, C Korthals Altes, and C Hartgrink. 1977. Reconventionalization following antibiotic decontamination in man and animals. Am J Clin Nutr. 30:1887-95. 361. van der Waaij LA, HJM Harmsen, M Madjipour, FGM Kroese, HM Van Dullemen, NK De Boer, GW Welling, and PLM Jansen. 2001. 16S rRNA fluorescent in situ hybridization of human colon and terminal ileum biopsies: a mucus-adherent bacterial flora does not exist. Gastroenterology. 120(5):A3810. 362. Vandamme P, CS Harrington, KJalava, and SLW On. 2000. Misidentifying helicobacters: the Helicobacter cinaedi example. J Clin Microbial. 38(6):2261-6. 363. Vandamme P, B Pot, M Gillis, P de Vos, K Kersters, and J Swings. 1996. Polyphasic taxonomy, a consensus approach to bacterial systematics. Microbial Rev. 60:407-38.

-325- 364. Vase L, ME Robinson, GN Verne, and DD Price. 2003. The contributions of suggestion, desire, and expectation to placebo effects in irritable bowel syndrome patients. An empirical investigation. Pain.105:17-25. 365. Venturi A, P Gionchetti, F Rizzello, R Johansson, E Zucconi, P Brigidi, D Matteuzzi, and M Camperri. 1999. Impact on the composition of the faecal flora by a new probiotic preparation: preliminary data on maintenance treatment of patients with ulcerative colitis. Aliment Pharmacal Ther. 13(8):1103-8. 366. Vollaard EJ, and HAL Clasener. 1994. Colonization resistance. Antimicrob Agents Chemother. 38(3):409-14. 367. Voravuthikunchai SP, and A Lee. 1987. Cecectomy causes long-term reduction of colonization resistance in the mouse gastrointestinal tract. Inject Immun. 55(4):995-9. 368. Walter J, C Hertel, GW Tannock, CM Lis, K Munro, and WP Hammes. 2001. Detection of Lactobacillus, Pediococcus, Leuconostoc, and Weissella species in human faeces by using group-specific PCR primers and denaturing gradient gel electrophoresis. Appl Environ Microbial. 67(6):2578-85. 369. Wang GC, and Y Wang . 1997. The frequency of formation of chimeric molecules as a consequence of PCR coamplification of 16S rRNA genes from mixed bacterial genomes. Appl Environ Microbial. 63(12):4645-50. 370. Wang R-F, W-W Cao, and CE Cerniglia. 1996. PCR detection and quantitation of predominant anaerobic bacteria in human and animal fecal samples. Appl Environ Microbial. 61(4):1242-7. 371. Ward JM, MR Anver, DC Haines, JM Melhorn, PL Gorelick, LL Yan, and JG Fox. 1996. Inflammatory large bowel disease in immunodeficient mice naturally infected with Helicobacter hepaticus. Lab Anim Sci. 46(1):15-20. 372. Ward JM, JG Fox, MR Anver, DC Haines, CV George, MJ Collins, PL Gorelick, K Nagashima, MA Gonda, RV Gilden, JG Tully, RJ Russell, RE Benveniste, BJ Paster, FE Dewhirst, JC Donovan, LM Anderson, and JM Rice. 1994. Chronic active hepatitis and associated liver tumors in mice caused by a persistent bacterial infection with a novel Helicobacter species. J Natl Cancer Inst. 86:1222-7.

-326- 373. Weisburg WG, SM Barns, DA Pelletier, and DJ Lane. 1991. 16S ribosomal DNA amplification for phylogenetic study. J Bacterial. 173:697-703. 374. Whary MT, JH Cline, AE King, KM Hewes, D Chojnacky, A Salvarrey, and JG Fox. 2000. Monitoring sentinel mice for Helicobacter hepaticus, H. rodentium and H. bilis infection by use of polymerase chain reaction analysis and serologic testing. Comparative Med. 50:436-43. 375. Wiggins HS, and JH Cummings. 1976. Evidence for the mixing of residue in the human gut. Gut. 17: 1007-11. 376. Wilson KH, and RB Blitchington. 1996. Human colonic biota studied by ribosomal DNA sequence analysis. Appl Environ Microbial. 62(7):2273-8. 377. Wintzingerode FV, UB Gobel, and E Stackebrandt. 1997. Determination of microbial diversity in environmental samples: pitfalls of PCR-based rRNA analysis. FEMS Microbial Rev. 21:213-29. 378. Woese CR. 1987. Bacterial evolution. Microbial Rev. 51:221-71. 379. Wold AE, DA Caugant, G Lidin-Janson, P de Man, and C Svanborg. 1992. Resident colonic Escherichia coli strains frequently display uropathogenic characteristics. J Infect Dis. 165:46-52. 380. Won YS, JH Yoon, CH Lee, BH Kim, BH Huyn, and YK Choi. 2002. Helicobacter muricola sp. nov., a novel Helicobacter species isolated from the ceca and feces of Korean wild mouse (Mus musculus molossinus). FEMS Microbial Lett. 209(1):45-51. 381. Wood JD, OC Peck, KS Tefend, MA Rodriguez, JV Rodriguez, H J.I., MJ Stonebrook, and HM Sharma. 1998. Colitis and colon cancer in cotton-top tamarins (Saguinus oedipus oedipus) living wild in their natural habitat. Dig Dis Sci. 43:1443-53. 382. Worku ML, RL Sidebotham, JH Baron, JJ Misiewicz, RP Logan, T Keshavarz, and ON Karim. 1999. Motility of Helicobacter pylori in a viscous environment. Eur J Gastroenterol Hepatol. 11(10):1143-50. 383. Yamamoto H, Y Sekine, Y Sato, T Higashizawa, T Miyata, S Iino, K ldo, and Sugano, K. 2001. Total enteroscopy with a nonsurgical steerable double­ balloon method. Gastrointest Endosc. 53:216-20.

-327- 384. Young VB, KA Knox, and DB Schauer. 2000. Cytolethal distending toxin sequence and activity in the enterohepatic pathogen Helicobacter hepaticus. Infect Immun. 68: 184-91. 385. Zachar Z, and D Savage. 1979. Microbial interference and colonization of the murine gastrointestinal tract by Listeria monocytogenes. Infect Immun. 23(1):168-74. 386. Zenner L. 1999. Pathology, diagnosis and epidemiology of the rodent Helicobacter infection. Camp Immun Microbial Infect Dis. 22:41-61. 387. Zhu XY, T Zhong, Y Pandya, and RD Joerger. 2002. 16S rRNA-based analysis of microbiota from the cecum of broiler chickens. Appl Environ Microbial. 68(1):124-37. 388. Zoetendal EG, ADL Akkermans, and WM DeVos. 1998. Temperature gradient gel electrophoresis analysis of 16S rRNA from human fecal samples reveals stable and host-specific communities of active bacteria. Appl Environ Microbial. 64(10):3854-9. 389. Zoetendal EG, K Ben-amor, ADL Akkermans, T Abee, and WM de Vos. 2001. DNA isolation protocols affect the detection limit ofPCR approaches of bacteria in samples from the human gastrointestinal tract. Syst Appl Microbial. 24:405-10. 390. Zoetendal EG, A von Wright, T Vilpponen-Salmela, K Ben-Amor, ADL Akkermans, and WM de Vos. 2002. Mucosa-associated bacteria in the human gastrointestinal tract are uniformly distributed along the colon and differ from the community recovered from feces. Appl Environ Microbial. 68(7):3401-7. 391. Zubrzycki L, and EH Spaulding. 1961. Studies on the stability of the normal human fecal flora. J Bacterial. 83:968-74.

-328- APPENDIX: LIBSHUFFpvalues

# Adapted from script by David Singleton using MacPerl # 3/6/2003 # first load the source matrix from the file "sample" # and put it into the variable @sourcematrix

open (MATRIX, "sample") II open (MATRIX, "sample.txt") II die "Could not find sample file.\n";

while () { @tmp=split; push @sourcematrix, [ @tmp ]; }

open (PVALUES, ">pvalues"); print PVALUES ("\n Matrixsize Repetition PvalueXY PvalueYX \n\n");

# ask total seqs/how many columns in sourcematrix

print ("\n\nWelcome to LIBSHUFF pvalues vl.O\n\n");

print ("How many sequences are in X? ") ; $numXtotal = ; print ("How many sequences are in Y? ") ; $numYtotal = ; print ("How many p values for each matrix size?"); $psize = ; print ("Lower limit of matrix size x?"); $lower = ; print ("Upper limit of matrix size x?"); $upper = ; print ("Step size between upper and lower limits?"); $step = ;

# q is matrix size counter # r is repetition counter for the number of p values

$q = $lower; $totalXYseq $numXtotal + $numYtotal;

while ($q < $upper+l){

$numX = $q; $numY = $q; $totalseqs $q + $q; $r = 0;

while ($r < $psize){

# Create a random matrix derived from the source matrix for analysis with # the appropriate size # Choose random X and Y rows q times and save in arrays # selectXarray and selectYarray # First set up a dummy arrays filled with zeros for X and Y

$i = 0; for ($i .. $totalXYseq) @duromyX [ $i J = 0; @dummyY[$i] =" 0; $i++; }

# Pick random numbers and change zeros in the dummy array for X into ones # once one is chosen

$i = 0; while ($i != $q) { XRANDOM: $randomnumber = int (rand($numXtotal)); if ($dummyX[$randomnumber] == 1) { goto XRANDOM; } @selectXarray[$i] = $randomnumber; @duromyX[$randomnumber] = 1; $i++; }

# Pick random numbers and change zeros in the dummy array for Y # into ones once one is chosen

$i = 0; while ($i != $q) { YRANDOM: $randomnumber = int (rand($numYtotal)); if ($duromyY[$randomnumber] == 1) { goto YRANDOM; } @selectYarray[$i] = $randomnumber; @dummyY[$randomnumber] = 1; $i++; }

#create input matrix for libshuff from source matrix

$i = 0; for (1. .$q) $j = 0; for (1.. $q) { $matrix[$i] [$j] = $sourcematrix[$selectXarray[$i]] [$selectXarray[$j]]; $matrix[$i] [$j+$q] = $sourcematrix[$selectXarray[$i]] [$numXtotal + $selectYarray[$j]]; $matrix[$i+$q] [$j] = $sourcematrix[$numXtotal + $selectYarray[$i]] [$selectXarray[$j]]; $matrix[$i+$q] [$j+$q] = $sourcematrix[$numXtotal + $selectYarray[$i]] [$numXtotal + $selectYarray[$j]]; $j++; } $i++; } # Now to run Libshuff for this matrix

$rounds = 1000;

# Now to get the LD for each sequence in X minus self-identities. # Store all LDs in ld_array

$i = 0; for ($i .. $numX-1) $ld = 100; $j = 0; for ($j .. $numX-1) { if ($matrix [$j] [$i] < $ld && ($i != $j)) { $ld = $matrix[$j] [$i]; @X_array[$i] = $ld; } $j++; } $i++; }

# Calculate the homologous LD of Y

$i = $numX; $k = 0; for ($i .. $totalseqs-1) $ld = 100; $j = $numX; for ($j .. $totalseqs-1) { if ($matrix [$j] [$i] < $ld && ($i ! = $j) ) { $ld = $matrix [$j] [$i]; @Y_array[$k] = $ld; } $j++; } $i++; $k++; }

# Calculate the heterologous LD XY (X in columns, Y in rows)

$i = 0; for ($i .. $numX-1) $ld = 100; $j = $numX; for($j .. $totalseqs-1) { if ($matrix [$j] [$i] < $ld) { $ld = $matrix[$j] [$i]; @XY_array[$i]= $ld; $j++; } $i++; }

# Calculate the heterologous LD YX (Y in columns, X in rows)

$i = $nurnX; $k = 0; for ($i .. $totalseqs-1) $ld = 100; $j = 0; for ($j .. $nurnX-1) { if($matrix[$j] [$i] < $ld) $ld = $matrix[$j] [$i]; @YX_array[$k]= $ld; } $j++; } $i++; $k++; }

# Convert all arrays into No values

$d = 0; $j = 0; while ($d < 0.51) $i = 0; $sum = 0; for ($i .. $nurnX-1) { if ($X_array[$i] < $d I I $X_array[$i] $d) { $sum++; } $i++; } @X_array_No[$j] $sum; $d = $d + 0.01; $j++; $sum = 0; }

$d = 0; $j = 0; while ($d < 0.51) { $i = 0; $sum = 0; for ($i .. $numY-1) { if ($Y_array[$i] < $d I I $Y_array[$i] $d) { $sum++; } $i++; } @Y_array_No[$j] $sum; $d = $d + 0.01; $j++; $sum = 0; }

$d = 0; $j = 0; while ($d < 0.51) $i = 0; $sum = 0; for ($i .. $numX-1) if ($XY_array[$i] < $d I I $XY_array[$i] $d) { $sum++; } $i++; } @XY_array_No[$j] $sum; $d = $d + 0.01; $j++; $sum = 0; }

$d = 0; $j = 0; while ($d < 0.51) $i = 0; $sum = 0; for ($i .. $numY-1) if ($YX_array[$i] < $d I I $YX_array[$i] $d) { $sum++; } $i++; } @YX_array_No[$j] $sum; $d = $d + 0.01; $j++; $sum = 0; }

* Now convert all No values into C values but don't print them $i = 0; $d = 0; for ($i. .50) { $X_array_No[$i] = (1-(($numX-$X_array_No[$i])/$numX)); $Y_array_No[$i] = (1-(($numY-$Y_array_No[$i])/$numY)); $XY array No[$i] (1-(($numX-$XY_array_No[$i])/$numX)); $YX=array=No[$i] = (1-(($numY-$YX_array_No[$i])/$numY)); $i++; $d = $d + 0.01; } # Now to calculate p-values for X/XY and Y/YX

$XY_delta 0; $YX_delta 0; $i = 0;

for ($i.. 50) { $XY_delta $XY_delta + (($X_array_No[$i] - $XY_array_No[$i])**2); $YX_delta $YX_delta + (($Y_array_No[$i] - $YX_array_No[$i])**2); $i++; }

# Begin the shuffling of the matrix # since this needs to be done twice, we need an additional variable to # change the size of the matrix from X x Y to Y x X

$trip = 0;

while ($trip != 2)

if($trip == 1) { $temporary = $numX; $numX $numY; $numY = $temporary; }

$sigh = 0; while ($sigh != $rounds) {

# devise an array of random numbers equal to numcolumns within totalseqs # however, cannot duplicate random numbers within the list

# First, set up a dummy array filled with zeros.

$i = 0; for ($i .. $totalseqs) @dummy [$i] = 0; $i++; }

# Pick random numbers and change the zeros in the dummy array into # ones once one is chosen.

$i = 0; while ($i != $numY) { RANDOM: $randomnumber = int(rand($totalseqs)); #ranges between zero and $totalseqs if ($dummy[$randomnumber] == 1) { goto RANDOM; } @columnarray[$i] = $randomnumber; @dummy[$randomnumber] = 1; $i++ } # Ok, let's put every other number into the rowarray

$i 0; $j 0; while ($i != $totalseqs) if ($durnmy[$i] == 0) { @rowarray[$j] = $i; $j++; } $i++; }

# Now to get the LD of each column using ONLY the row numbers. # Store all LOs in ld_array

$i = 0; for ($i .. $numY-l) $ld = 100; $j = 0; for ($j .. $numX-1) { if ($matrix[$rowarray[$j]] [$columnarray[$i]] < $ld) $ld = $matrix[$rowarray[$j]] [$columnarray[$i]]; @ld_array[$i] = $ld; } $j++; } $i++; }

# Determine the LD for each sample in the rowarray WITHOUT # self identities. Place these LOs in the array called 'homol ld'

$i = 0; for ($i .. $numX-1) $ld = 100; $j = 0; for ($j .. $numY-l) { if ($matrix[$rowarray[$j]] [$rowarray[$i]] < $ld && ($i != $j)) { $ld = $matrix[$rowarray[$j]] [$rowarray[$i]]; @homol_ld_array[$i] = $ld; } $j++; } $i++; }

# Now to determine the No for each value of D in both LD arrays

$d = 0; $j = 0; #First, ld_array containing the heterologous comparisons ... while ($d < 0.51) { $i = 0; $sum = 0; for ($i .. $numY-1) if ($ld_array[$i] < $d I I $ld_array[$i] $d) { $sum++; } $i++; } @sum_array[$j] $sum; $d = $d + 0.01; $j++; $sum = 0; }

#Now, the homol_ld_array containing the homologous comparisons ...

$d 0; $j 0; while ($d < 0.51) { $i = 0; $sum = 0; for ($i .. $numX-l) { if ($homol_ld_array[$i] < $d I I $homol_ld_array[$i] $d) { $sum++; } $i++; } @homol_sum_array[$j] $sum; $d = $d + 0.01; $j++; $sum = 0; }

# Now to convert the values in sum_array and homol_sum_array # into coverage values

$i 0; $d 0; for ($i. .50) { $sum_array[$i] = (1-(($numY-$sum array[$i])/$numY)); $homol_sum_array[$i] (1-(($numX-$homol_sum_array[$i])/$numX)); $i++; $d = $d + 0.01; }

#Compare the coverage values of the homologous to the heterologous ...

$i = 0; $delta = 0; for ($i.. 50) { $deltad = (($sum_array[$i]-$homol_sum_array[$i])**2); $delta = $delta + $deltad; if($trip == 0) { $XYmatrix[$i] [$sigh] = ( $deltad ); } if($trip == 1) { $YXmatrix [$i] [$sigh] ( $deltad ) ; } $i++; }

@dumrny=O; @ld_array=O; @column_array=O; @rowarray=O; @sum_array=O; if ($trip == 0) { @p_values1[$sigh] $delta; } if ($trip == 1) { @p_values2[$sigh] $delta; } $sigh++; } $trip++; }

@results = sort @p_values1; @results2 = sort @p_values2;

$XYp 1; $YXp 1;

$i = 0; for ($i .. 998) { if($XY_delta > $results[$i] I I $XY_delta == $results[$i]) { $XYp = $XYp - 0.001; } if($YX_delta > $results2[$i] I I $YX_delta == $results2[$i]) { $YXp = $YXp - 0.001; } $i++; } print PVALUES (" "); printf PVALUES ("%.3f",$q); print PVALUES (" "); printf PVALUES ("%.3f",$r); print PVALUES (" ") ; printf PVALUES ("%.3f",$XYp); print PVALUES (" ") ; printf PVALUES ("%.3f",$YXp); print PVALUES ("\n"); print ( rr rr) ; printf ("%.3f",$q); print ( rr rr) ; printf ("%.3f",$r); print ( rr rr) ; printf ("%.3f",$XYp); print (" ") ; printf ("%.3f",$YXp); print ( "\n");

$r++; } $q = $q + $step; } close PVALUES; print("Program finished.\n"); ALLBOOK BINDERY 91 RYEDALE ROAD WEST RYDE 2114 PHONE: 9807 6026