Development of an In Vitro Fermentation Model to Culture the

Human Distal Gut Microbiota

by

Julie A.K. McDonald

A thesis

presented to

The University of Guelph

In partial fulfilment of requirements

for the degree of

Doctor of Philosophy

in Molecular and Cellular Biology

Guelph, Ontario, Canada

© Julie A.K. McDonald, May, 2013

ABSTRACT

DEVELOPMENT OF AN IN VITRO FERMENTATION MODEL TO CULTURE THE

HUMAN DISTAL GUT MICROBIOTA

Julie A.K. McDonald Advisor:

University of Guelph, 2013 Dr. Emma Allen-Vercoe

In vitro gut models provide several advantages over in vivo models for the study of the

human gut microbiota. However, because communities developed in these models are simplified

simulations of the in vivo environment it is necessary to characterize the reproducibility,

repeatability and stability of cultured communities. We also need to broadly define the

differences between in vitro consortia and the communities from which they are derived. In this

study we characterized and validated a twin-vessel (independent, identical) single-stage

chemostat model of the human distal gut. Samples were analyzed using a molecular

fingerprinting technique (Denaturing Gradient Gel Electrophoresis) to compare and monitor

changes in the overall structure of the communities while a phylogenetic microarray (Human

Intestinal Tract Chip) was used to obtain phylogenetic information. We found that twin-vessels

inoculated with feces developed and maintained diverse communities that reached stable

compositions by at most 36 days post-inoculation. Communities were enriched in Bacteroidetes

but not Clostridium cluster XIVa, Bacilli or other Firmicutes relative to the fecal inocula.

Vessels were very reproducible when inoculated with identical fecal inocula, less similar when

inoculated with consecutive fecal donations from the same donor, and maintained donor-specific

identities when inoculated with feces from different donors. Norepinephrine exposure (undefined

perturbation) did not appear to have a substantial effect on the structure of chemostat

communities, while clindamycin treatment (defined perturbation) caused large changes in the

structure of chemostat communities. Packed-column biofilm reactors incorporated a simulated

mucosal environment into our chemostat system, allowing us to simultaneously culture

biologically relevant planktonic and biofilm communities that were complex, reproducible, and

distinct. Defined communities were comparable to fecal communities at the phylum/class-level

but established stable compositions more rapidly. While it was difficult to assess the persistence

of synthetic stool in a healthy fecal chemostat community (+/- antibiotic perturbation), mixing

communities from two donors resulted in a mixed community that more closely resembled one

donor over the other. Although future experimentation is required, the results presented here

show our twin-vessel single-stage chemostat model represents a valid simulation of the human

distal gut environment and can support complex, representative microbial communities ideal for

experimental manipulation.

iv

ACKNOWLEDGEMENTS

First and foremost I would like to thank my supervisor, Dr. Emma Allen-Vercoe, for her

expertise, support, dedication and friendship. It's difficult to express how grateful I am for her

help. She has been a wonderful role model and has been instrumental in developing my

confidence and academic research skills. I would also like to thank members of my advisory

committee, Dr. Cezar Khursigara, Dr. Wendy Keenleyside, and Dr. Marc Habash for their

guidance and wisdom.

My gratitude also goes out to past and present members of the Allen-Vercoe, Lo, and

Khursigara labs who have made my graduate research experience so enjoyable. I would like to

give special thanks to Kathleen Schroeter for her help, advice, and friendship. Her support has

been invaluable and I really appreciate all the hard work she has contributed to my project. I'd

also like to sincerely thank Chris Ambrose, Michelle Daigneault, Kyla Cochrane, Mike Toh, and

Jaclyn Strauss for all their help and for making the Allen-Vercoe lab such an amazing place to

work and study. I am also indebted to Dr. Paula Russell for her academic advice and kind words

of encouragement.

I would like to express my appreciation to Dr. Willem de Vos and his research group

whose collaboration allowed us to analyze many of my samples using his phylogenetic

microarray (HITChip). I would especially like to thank Dr. Susana Fuentes for working hard to

send me HITChip data and answer all of my questions.

I would also like to acknowledge the Canadian Institutes of Health Research (CIHR) who

provided financial support as master's and doctoral scholarships as well as the Government of

Ontario for an Ontario Graduate scholarship. Importantly, I'd also like to thank the CIHR and

Ontario Ministry of Agriculture, Food & Rural Affairs for their financial contributions towards

v

the Allen-Vercoe lab for my research project.

I appreciate all the love and support from my family and friends who have always been

there for me, no matter how busy I was with experiments, data analysis and writing. Most of all,

I am grateful to Alex for his love, patience, and encouragement throughout my graduate degree.

vi

Author's Declaration of Work Completed

I declare that all of the work carried out in this thesis is my own, with the following

exceptions as outlined below.

Kathleen Schroeter, Christian Ambrose, and Michelle Daigneault assisted me with setting

up and operating several chemostat runs (e.g. sampling vessels, changing bottles or media,

changing waste bottles).

HITChip analysis was carried out in collaboration with Dr. Willem de Vos' research group

at Wageningen University. DNA samples were sent to Ineke Heikamp-deJong who performed

the HITChip hybridizations and Dr. Susana Fuentes who performed the bioinformatic analysis of

the HITChip data. On the advice of Dr. Fuentes I interpreted the HITChip data and prepared

additional figures and tables as required.

Michelle Daigneault and Eric Brown isolated and identified the strains used to comprise

the defined microbial communities (referred to as DEC and MET-1 in this thesis). Kathleen

Schroeter and I prepared the DEC inocula (for Chapter 6), while Michelle Daigneault prepared

the MET-1 inocula (for Chapters 6 and 7). Christian Carlucci and Kathleen Schroeter set up,

operated, and collected samples from VM1 (Chapter 6), however analysis of VM1 chemostat

samples was conducted by me (DNA extraction, DGGE analysis, and HITChip data

interpretation), Ineke Heikamp-deJong (HITChip microarray procedure) and Susana Fuentes

(HITChip data analysis).

Michelle Daigneault also tested Subdoligranulum variabile and Faecalibacterium

prausnitzii isolates for clindamycin susceptibility.

vii

TABLE OF CONTENTS

Acknowledgements ........................................................................................................... iv

Author’s Declaration of Work Completed ..................................................................... vi

List of Figures .................................................................................................................... xii

List of Tables ..................................................................................................................... xv

List of Equations ............................................................................................................... xviii

Glossary of Abbreviations ................................................................................................ xix

Chapter 1. INTRODUCTION ......................................................................................... 1

1.1 Human gut microbiota .............................................................................................. 1

1.2 Stability and perturbation ......................................................................................... 9

1.3 Gut microbiota and disease ...................................................................................... 11

1.4 Modeling the gut microbiota in vivo ........................................................................ 18

1.5 In vitro human gut fermentation modeling .............................................................. 20

1.6 Research rationale .................................................................................................... 26

1.7 Research objectives .................................................................................................. 28

Chapter 2. MATERIALS AND METHODS .................................................................. 30

2.1 Preparation of single-stage chemostats .................................................................... 30

2.2 Collection and preparation of fecal inocula ............................................................. 34

2.3 Preparation of defined communities ........................................................................ 36

2.4 Inoculation, operation and sampling ........................................................................ 40

2.5 Combining steady state chemostat communities ..................................................... 40

2.6 Norepinephrine-induced perturbation ...................................................................... 41

2.7 Clindamycin-induced perturbation ........................................................................... 41

viii

2.8 Packed-column biofilm reactor preparation ............................................................. 42

2.9 Biofilm harvest from packed-columns ..................................................................... 42

2.10 Adding MET-1 to clindamycin-perturbed chemostat communities ....................... 46

2.11 DNA extraction ...................................................................................................... 46

2.12 DGGE analysis ....................................................................................................... 47

2.13 Community dynamics ............................................................................................ 49

2.14 Compensation for DGGE gel-to-gel variation ....................................................... 50

2.15 Phylogenetic microarray (HITChip) analysis ........................................................ 50

2.16 HITChip statistical data analysis ............................................................................ 53

2.17 Shannon diversity index ......................................................................................... 54

2.18 Shannon equitability index ..................................................................................... 54

2.19 Richness ................................................................................................................. 55

Chapter 3. EVALUATION OF MICROBIAL COMMUNITY REPRODUCIBILITY,

STABILITY, AND COMPOSITION IN A HUMAN DISTAL GUT

CHEMOSTAT MODEL................................................................................ 56

3.1 Introduction .............................................................................................................. 56

3.2 Results ...................................................................................................................... 57

3.2.1 Establishment of steady state microbial community compositions in a twin-vessel

single stage distal gut model ............................................................................ 58

3.2.2 Reproducibility of fecal inocula and chemostat communities (technical

replicates) ......................................................................................................... 58

3.2.3 Repeatability of steady state communities seeded with consecutive fecal donations

from the same donor (biological replicates) .................................................... 73

3.2.4 Development of microbial communities seeded with feces from different

donors ............................................................................................................... 74

3.3 Discussion ................................................................................................................ 75

ix

Chapter 4. CHARACTERIZING PERTURBATION IN A SIMULATED DISTAL

GUT ECOSYSTEM: EFFECT OF NOREPINEPHRINE AND

CLINDAMYCIN TREATMENT ................................................................. 84

4.1 Introduction .............................................................................................................. 84

4.2 Results ...................................................................................................................... 86

4.2.1 Perturbation of chemostat communities using a single dose of

norepinephrine ................................................................................................. 86

4.2.2 Effects of sustained norepinephrine perturbation on gut chemostat

communities ..................................................................................................... 90

4.2.3 Effects of sustained clindamycin perturbation on gut chemostat

communities ..................................................................................................... 102

4.3 Discussion ................................................................................................................ 104

Chapter 5. MODELING SIMULATED MUCOSAL AND LUMINAL COMMUNITIES

USING PACKED-COLUMN BIOFILM REACTORS AND AN IN VITRO

DISTAL GUT CHEMOSTAT MODEL ...................................................... 109

5.1 Introduction .............................................................................................................. 109

5.2 Results ...................................................................................................................... 111

5.2.1 Planktonic communities developed in mucosal twin-vessel single-stage chemostats

are reproducible and stable .............................................................................. 111

5.2.2 Twin-vessel biofilm growth is reproducible both between twin vessels and within

the same vessel ................................................................................................. 125

5.2.3 Planktonic and biofilm communities have distinct community

compositions .................................................................................................... 128

5.2.4 Clindamycin treatment causes a shift in planktonic community

composition ...................................................................................................... 128

5.2.5 Clindamycin treatment alters the structure of simulated mucosal

biofilms ............................................................................................................ 134

5.3 Discussion ................................................................................................................ 135

x

Chapter 6. USING DERIVED, DEFINED EXPERIMENTAL COMMUNITIES TO

MODEL THE HUMAN DISTAL GUT IN VITRO .................................... 144

6.1 Introduction .............................................................................................................. 144

6.2 Results ...................................................................................................................... 146

6.2.1 Twin-vessel reproducibility .............................................................................. 147

6.2.2 VF1, VD1 and VM1 correlation coefficients ................................................... 152

6.2.3 VF1, VD1, VM1 stability and diversity ........................................................... 157

6.2.4 Phylogenetic analysis of VF1, VD1, and VM1 ................................................ 161

6.3 Discussion ................................................................................................................ 167

Chapter 7. ASSESSING INTERACTIONS BETWEEN GUT COMMUNITIES FROM

DIFFERENT DONORS IN AN IN VITRO MODEL OF THE HUMAN DISTAL

COLON ........................................................................................................... 175

7.1 Introduction .............................................................................................................. 175

7.2 Results ...................................................................................................................... 177

7.2.1 Effect of clindamycin treatment and MET-1 supplementation on healthy fecal

chemostat communities .................................................................................... 177

7.2.2 Mixing donor A and B chemostat communities results in a community that

more closely resembled the donor A community than the donor B

community ....................................................................................................... 187

7.3 Discussion ................................................................................................................ 199

Chapter 8. SUMMARY, CONCLUSIONS AND FUTURE WORK ........................... 209

REFERENCES .................................................................................................................. 224

APPENDICES ................................................................................................................... 253

Appendix 1. SUPPLEMENTARY MATERIAL FROM CHAPTER 2 ....................... 253

Appendix 1.1: Supplementary figures and tables from Chapter 2 ................................. 253

Appendix 1.2: Preparation of spent LS174T growth medium ....................................... 259

xi

1.2.1 Setting up a flask of LS174T tissue culture cells ............................................. 259

1.2.2 Changing spent LS174T growth medium......................................................... 259

1.2.3 Splitting LS174T cells ...................................................................................... 260

1.2.4 Preparation of heat inactivated FBS ................................................................. 261

Appendix 1.3: Freezing media preparation and storage of bacterial isolates ................ 262

Appendix 1.4: Running DGGE gels ............................................................................... 263

1.4.1 Agarose gel electrophoresis to check DGGE PCR products ............................ 263

1.4.2 Preparation of 50x TAE (Tris-acetate-EDTA) running buffer ......................... 264

1.4.3 Preparation of PCR samples to load onto DGGE gels ..................................... 264

1.4.4 Pouring DGGE gels .......................................................................................... 265

1.4.5 Running DGGE gels ......................................................................................... 266

Appendix 2. SUPPLEMENTARY FIGURES AND TABLES FROM

CHAPTER 3 ................................................................................................. 268

Appendix 3. SUPPLEMENTARY FIGURES AND TABLES FROM

CHAPTER 5 ................................................................................................. 275

Appendix 4. SUPPLEMENTARY TABLE FROM CHAPTER 6................................ 278

xii

List of Figures

Figure 1.1 ............................................................................................................................ 2

Figure 1.2 ............................................................................................................................ 24

Figure 2.1 ............................................................................................................................ 32

Figure 2.2 ............................................................................................................................ 44

Figure 2.3 ............................................................................................................................ 51

Figure A2.1 ......................................................................................................................... 253

Figure A2.2 ......................................................................................................................... 254

Figure A2.3 ......................................................................................................................... 255

Figure 3.1 ............................................................................................................................ 60

Figure 3.2 ............................................................................................................................ 63

Figure 3.3 ............................................................................................................................ 64

Figure 3.4 ............................................................................................................................ 67

Figure 3.5 ............................................................................................................................ 69

Figure 3.6 ............................................................................................................................ 72

Figure A3.1 ......................................................................................................................... 269

Figure A3.2 ......................................................................................................................... 271

Figure 4.1 ............................................................................................................................ 87

Figure 4.2 ............................................................................................................................ 89

Figure 4.3 ............................................................................................................................ 93

Figure 4.4 ............................................................................................................................ 96

Figure 4.5 ............................................................................................................................ 98

Figure 4.6 ............................................................................................................................ 101

xiii

Figure 5.1 ............................................................................................................................ 112

Figure 5.2 ............................................................................................................................ 113

Figure 5.3 ............................................................................................................................ 117

Figure 5.4 ............................................................................................................................ 120

Figure 5.5 ............................................................................................................................ 121

Figure 5.6 ............................................................................................................................ 123

Figure 5.7 ............................................................................................................................ 126

Figure A5.1 ......................................................................................................................... 275

Figure A5.2 ......................................................................................................................... 277

Figure 6.1 ............................................................................................................................ 149

Figure 6.2 ............................................................................................................................ 151

Figure 6.3 ............................................................................................................................ 155

Figure 6.4 ............................................................................................................................ 156

Figure 6.5 ............................................................................................................................ 159

Figure 6.6 ............................................................................................................................ 160

Figure 6.7 ............................................................................................................................ 163

Figure 7.1 ............................................................................................................................ 180

Figure 7.2 ............................................................................................................................ 182

Figure 7.3 ............................................................................................................................ 184

Figure 7.4 ............................................................................................................................ 189

Figure 7.5 ............................................................................................................................ 190

Figure 7.6 ............................................................................................................................ 194

Figure 7.7 ............................................................................................................................ 196

xiv

Figure 7.8 ............................................................................................................................ 197

xv

List of Tables

Table 1.1 ............................................................................................................................. 5

Table 1.2 ............................................................................................................................. 7

Table 1.3 ............................................................................................................................. 13

Table 1.4 ............................................................................................................................. 14

Table 1.5 ............................................................................................................................. 27

Table 2.1 ............................................................................................................................. 33

Table 2.2 ............................................................................................................................. 35

Table 2.3 ............................................................................................................................. 37

Table 2.4 ............................................................................................................................. 45

Table 2.5 ............................................................................................................................. 52

Table A2.1........................................................................................................................... 256

Table A2.2........................................................................................................................... 257

Table A2.3........................................................................................................................... 258

Table A2.4........................................................................................................................... 261

Table A2.5........................................................................................................................... 261

Table A2.6........................................................................................................................... 266

Table 3.1 ............................................................................................................................. 61

Table 3.2 ............................................................................................................................. 71

Table 3.3 ............................................................................................................................. 76

Table A3.1........................................................................................................................... 270

Table A3.2........................................................................................................................... 272

Table A3.3........................................................................................................................... 273

xvi

Table A3.4........................................................................................................................... 274

Table 4.1 ............................................................................................................................. 91

Table 4.2 ............................................................................................................................. 94

Table 4.3 ............................................................................................................................. 99

Table 4.4 ............................................................................................................................. 103

Table 5.1 ............................................................................................................................. 114

Table 5.2 ............................................................................................................................. 116

Table 5.3 ............................................................................................................................. 118

Table 5.4 ............................................................................................................................. 124

Table 5.5 ............................................................................................................................. 127

Table 5.6 ............................................................................................................................. 130

Table 5.7 ............................................................................................................................. 131

Table 5.8 ............................................................................................................................. 133

Table 5.9 ............................................................................................................................. 136

Table A5.1........................................................................................................................... 276

Table 6.1 ............................................................................................................................. 150

Table 6.2 ............................................................................................................................. 153

Table 6.3 ............................................................................................................................. 158

Table 6.4 ............................................................................................................................. 162

Table 6.5 ............................................................................................................................. 164

Table 6.6 ............................................................................................................................. 165

Table 6.7 ............................................................................................................................. 166

Table A6.1........................................................................................................................... 279

xvii

Table 7.1 ............................................................................................................................. 181

Table 7.2 ............................................................................................................................. 186

Table 7.3 ............................................................................................................................. 191

Table 7.4 ............................................................................................................................. 192

Table 7.5 ............................................................................................................................. 198

Table 7.6 ............................................................................................................................. 200

xviii

List of Equations

Equation 2.1: Shannon diversity index ............................................................................... 54

Equation 2.2: Shannon equitability index ........................................................................... 54

xix

Glossary of Abbreviations

APS, Ammonium persulfate

CD, Crohn's disease

CDI, Clostridium difficile infection

DEC, Defined experimental community

DGGE, Denaturing gradient gel electrophoresis

DMEM, Dulbecco's modified eagle medium

DMSO, Dimethyl sulfoxide

EDTA, Ethylenediaminetetraacetic acid

EH', Shannon equitability index

EHEC, Enterohaemorrhagic Escherichia coli

ETEC, Enterotoxigenic Escherichia coli

FAA, Fastidious anaerobe agar

FBS, Fetal bovine serum

GABA, Gamma-aminobutyric acid

GI, Gastrointestinal

H', Shannon diversity index

HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

HITChip, Human Intestinal Tract Chip

HMP, Human Microbiome Project

HPLC, High-performance liquid chromatography

IBD, Inflammatory bowel disease

IBS, Irritable bowel syndrome

xx

IgA, Immunoglobulin A

L-SHIME, Luminal simulator of the human intestinal microbial ecosystem

M-SHIME, Mucosal simulator of the human intestinal microbial ecosystem

MET, Microbial ecosystem therapeutic

NE, Norepinephrine

NEC, Necrotizing enterocolitis

NMR, Nuclear magnetic resonance

PBS, Phosphate buffered saline

PCA, Principal component analysis

RDA, Redundancy analysis

S, Richness

SCFA, Short chain fatty acid

SD, Standard deviation

sFAA, Fastidious anaerobe agar supplemented with 5% defibrinated sheep blood

SHIME, Simulator of the human intestinal microbial ecosystem

SI, Similarity index

SIHUMI, Simplified human intestinal microbiota

sp., Species

spp., Species (plural)

TAE, Tris-acetate-EDTA

TE, Tris EDTA

TEMED, Tetramethylethylenediamine

TIM, TNO intestinal model

xxi

TNBS, Trinitrobenzene sulfonic acid

TNO, Nederlandse Organisatie voor Toegepast Natuurwetenschappelijk Onderzoek

(Dutch Organization for Applied Scientific Research)

UC, Ulcerative colitis

UPGMA, Unweighted Pair Group Method with Arithmetic Mean

UV, Ultraviolet

1

Chapter 1

Introduction

1.1 Human gut microbiota

The human gastrointestinal (GI) tract is a specialized digestive organ with several distinct

anatomical regions, including the mouth, esophagus, stomach, small intestine and large intestine.

The large intestine consists of the proximal bowel (cecum and ascending colon), transverse

colon, descending colon and sigmoid/rectum. The arrangement of the large intestine can be

thought of as an ‘open’ system, where food residues from the small intestine enter at the cecum

and are periodically voided from the rectum (1).

Due to the low pH, microbial counts are low in the stomach (<103 cells/g) and while they

increase in the small intestine (104-10

7 cells/g), growth is still limited by rapid transit time, bile

secretion, and the presence of pancreatic juice (Figure 1.1; 2). Bacterial counts increase in the

large intestine, with 1010

-1011

cells/g in the proximal colon, 1011

-1012

cells/g in the transverse

colon and ≥1012

cells/g in the descending colon (2). Counts are higher in the large intestine

because this compartment contains the most favourable conditions for microbial growth: slower

transit time, high nutrient availability (dietary and host origin), and favourable pH (2-4).

The gut microbiota is one of the most dense and diverse microbial ecosystems known; the

average healthy human gut is estimated to contain ~1000 different species both in intestinal

material and on mucosal surfaces (5-9). The gut microbiota is largely composed of indigenous or

commensal bacterial species that are permanent (autochthonous) residents of the gut, as well as a

smaller, variable collection of microbial species that are transient (allochthonous) occupants

2

Figure 1.1 Diagram illustrating different segments of the human gastrointestinal tract. Listed are

the approximate number of microbial cells per gram of intestinal contents, nutrient availability,

and types of fermentation typical of each segment of a healthy gut. Text adapted from Payne et

al. (2). Image from http://www.loveyourgut.com/wp-content/uploads/digestive-system.jpg

Proximal colon

1010-1011 cells/g- High concentration of nutrients

- Active microbial growth

- Fermentation of carbohydrates

- High SCFA production

- pH 5.4-5.9

Transverse colon

1011-1012 cells/g- Depletion of nutrients

- Decrease in bacterial activity

- Reduction in SCFA production

- pH 6.1-6.4

Distal colon

≥1012 cells/g- Low concentration of nutrients

- Slow microbial growth

- Fermentation of protein

- Low SCFA production

-pH 6.1-6.9

Small intestine

104-107 cells/g

Stomach

<103 cells/g

3

(10, 11). Autochthonous gut microbial communities provide a significant contribution to human

digestive physiology and play a large role in gut homeostasis (4). In fact, the gut microbiota has

been referred to as a personalized "human organ" with a collective metabolic activity rivaling

that of the liver (12, 13). Allochthonous microorganisms can include neutral microbes that don't

affect host health, infectious agents that cause disease, or "probiotic" strains that are being

exploited for their health benefits (see section 1.3).

Although the human gut microbiota was once regarded as a homogeneous microbial

community, it is now clear that substrate availability, bacterial populations and microbial

metabolism vary along the length of the gut (Figure 1.1; 14, 15). Within the large intestine

microbial populations colonize food residues in the cecum, maintain populations within the

proximal colon, and serve as inocula for residues entering the distal colon (16). Further

examination of the gut microbiota has shown that microorganisms occupy a large number of

microhabitats and metabolic niches in addition to the gut lumen, including the intestinal mucosa

and the surfaces of incompletely degraded dietary residues (16). Microenvironments within the

gut contribute to the diversity of the microbiota and are in a dynamic state of change according

to the consumption and production of resources (16). Microhabitats may play an important role

in host health. For example, within the appendix microorganisms mostly grow adherent to the

mucosal surface as biofilms (17). It has been proposed that microorganisms growing in the

appendix may allow for re-inoculation of the colon following a perturbation event (for example,

infection with an enteric virus) to help re-establish the gut ecosystem (17).

For many years it has been known that bacteria colonize the gut in abundance, but until

fairly recently the true significance of this colonization to human health was greatly

underestimated (18). It is now clear that the healthy gut microbiota exerts various protective,

4

structural, and metabolic functions on the host gut epithelium (Table 1.1; 12). The relationship

between gut microbiota and host is mutualistic; in return resident microbes can acquire nutrients,

a favourable habitat, and an efficient means of dispersal (19).

Despite its importance in host health and disease, the gut microbiota is incompletely

characterized and has a poorly defined microbial diversity (6, 20). However, we do know that the

low oxygen tension environment of the colon favours obligately anaerobic microbes, with only

~0.1 % of the total species diversity made up of facultatively anaerobic species (21-23). Gut

microbiota are mainly composed of bacteria, but methanogenic archaea (e.g.

Methanobrevibacter smithii), eukaryotes (mostly yeasts) and viruses (largely bacteriophage) are

also present (24).The majority of the bacterial species of the gut microbiota (99%) consists of

just four phyla, including Bacteroidetes and Firmicutes (which together dominate the gut

microbiota, accounting for > 90% of the species present) as well as Proteobacteria and

Actinobacteria (which are present at lower levels; 5, 6, 25, 26). Gut microbial diversity is

therefore in contrast to that of Earth’s total biosphere, which is comprised of up to 70 bacterial

phyla (27). However, despite a relatively small number of phyla, gut microbial diversity is

considered to be high at the level of genera and species within these phyla (27). The dominant

anaerobic genera found in the human gut include Bifidobacterium, Clostridium, Bacteroides, and

Eubacterium, while the dominant facultatively anaerobic genera include Escherichia,

Enterococcus, Streptococcus, and Klebsiella (12).

Attempts to define the composition of "healthy" gut microbiota have been difficult as gut

microbiota show significant inter-individual variation (6, 28). It is important to understand both

the factors that give rise to host specificity of a given gut microbiota, as well as the contributions

of microbial community members to microbiota functionality, as differences in these factors may

5

Table 1.1 Functions of the gut microbiota that benefit human health. Modified from O’Hara and

Shanahan (12) and Lutgendorff et al. (10).

Metabolic functions Protective functions Structural functions

Ferment non-digestible

dietary residue and

endogenous epithelial-

derived mucus

Pathogen displacement Barrier fortification

Salvage of energy Nutrient competition Apical tightening of

tight junctions

Synthesis of micronutrients

including vitamins, short

chain fatty acids, and

polyamines

Receptor competition

Immune system

development (innate

and adaptive)

Control intestinal epithelial

cell differentiation and

proliferation

Production of anti-

microbial factors

(bacteriocins, lactic acids)

Induction of

immunoglobulin A

(IgA)

Processing and detoxification

of xenobiotics

Ion absorption

6

have important implications in health outcomes (29). Factors that drive normal compositional

variation of the gut microbiota include diet, genetics, age and environment (Table 1.2; 24).

Diet is one of the most important factors that drives normal compositional variation of the

gut microbiota since substrate preference, use, and composition change the relative abundance of

species within a given microbial community (30). For example, it has been shown that in the

fecal microbiotas from healthy individuals who consumed long-term diets rich in animal protein,

the abundances of several amino acids and saturated fats in these diets were associated with

Bacteroides populations. In contrast, long-term diets rich in carbohydrates and simple sugars

were associated with Prevotella populations (31). De Filippo et al. (2010) showed that the fecal

microbiotas from rural African children consuming a high fiber diet were enriched in

Bacteroidetes and depleted in Firmicutes compared to the fecal microbiotas from European

children consuming a Western diet (high in saturated and unsaturated fats; 32).

Studies have shown that the gut microbiotas sampled from family members are more

similar to each other than to those sampled from unrelated individuals (33-36). However, while

these similarities may be linked to the genetic relatedness of the hosts, they could also be

correlated with a common living environment (30). Human twin studies comparing the similarity

of gut microbiotas from monozygotic (identical) and dizygotic (fraternal) twins has yielded

contrasting results (33, 37-39). Although the fecal microbiotas of twins are generally more

similar to each other than to unrelated individuals (37, 38), some studies found that the gut

microbiotas from monozygotic twins were more similar to each other than the gut microbiotas

from dizygotic twins (37), while others found a comparable level of similarity (33). Diet and

lifestyle differences between twin subjects as well as differences in the methodologies employed

in experimental techniques may account for the varying results of these studies (30). Future work

7

Table 1.2 Factors that influence the composition of the human gut microbiota (4).

Host factors Microbiological factors Environmental factors

Diet Competition for nutrients Substrate availability

Genetics

Digestive secretions Competition for adhesion sites Local pH

Digestive physiology

Innate and adaptive immunity Metabolic cooperation Redox potential

Disease

Antibiotics Bacterial antagonism Geography

Drugs

8

utilizing a larger number of healthy twin subjects is required to confirm these observations (30).

Several studies have examined the influence of age on the composition and function of the

gut microbiota, describing differences between healthy infants, adults, and elderly subjects (6,

40-42). Compared to adult fecal microbiotas, infant microbiotas are relatively unstable and have

lower phylogenetic and species richness (enriched in Bifidobacterium spp.; 43). The

compositions of infant gut microbiotas is affected by the method of delivery (vaginal or

caesarean-section), diet (breastfed or formula fed), and antibiotic use (44-46). As infants age and

solid foods are introduced, fecal community stability and diversity increase and the relative

proportions of bacterial phyla begin to resemble those of an adult (47, 48). The gut microbiotas

of the elderly (adults over the age of 65) differ from the gut microbiotas of younger adults and

are characterized by lower diversity and stability as well as higher inter-individual variation (49-

51). Physiological changes associated with aging may account for differences in the composition

and function of elderly gut microbiotas (e.g. decreased intestinal motility, reduced nutritional

intake, immunosenescence, and use of laxatives and other medications; 51-54).

The evident stability and homeostasis of the adult human gut microbiota has led several

groups to propose species that comprise a "core gut microbiota," that is, species that are shared

components between many individuals (9, 33, 39, 40, 50, 55-57). However, currently there is no

consensus on the microbes that might comprise this core (24, 27, 58). Instead, the concept of a

core functional microbiome (or core microbiome-encoded gene set) has become an increasingly

popular proposal (27). Several studies have shown that individuals with highly divergent gut

microbial compositions have very similar functional gene profiles (9, 33, 59). Healthy gut

microbiomes contain genes involved in central pathways for carbohydrate and amino acid

metabolism (33). However, finer differences in function between individuals can be concealed

9

by grouping genes into broad functional categories (24). While specific metabolic functions

comprising a "healthy" core functional microbiome have yet to be defined, it is possible that

changes in these core sets of genes may be important for maintaining host health, or alternatively

for promoting the development of disease (27, 60).

1.2 Stability and perturbation

Aside from compositional and functional descriptions, "healthy" gut microbiotas can also

be described in terms of stability (61). The stability of a given microbial community is not

maintained due to the unresponsiveness of the microbial populations, but rather by its restorative

abilities that allow it to act as a dynamic system (62). Ecosystems are considered to be stable if

they possess or display five key characteristics: (1) return to and maintain baseline values

following displacement (recover); (2) re-establish baseline values rapidly following displacement

(resilience); (3) are maintained at a set value (persistence); (4) are resistant to change

(resistance); and (5) have limited variability over time (63). The stability of healthy gut microbial

ecosystems has been observed over several time windows, ranging from days to weeks to months

to years (56, 64, 65). These studies, among others, showed that the gut microbiotas of healthy

adults are dynamic but relatively stable in the absence of major perturbation (59, 66-68).

Longitudinal studies tracking changes in the gut microbiota allow us to test hypotheses, as

it is then possible to relate changes in microbial community composition or function to changes

in the health status of the host (61). For example, longitudinal analyses of gut mucosal

microbiotas in ulcerative colitis (UC) patients showed that compositional variations occurred

over time and that the structures of mucosal communities were associated with disease severity

(69). High clinical activity scores and sigmoidoscopy scores were positively associated with

10

Enterobacteriaceae, Desulfovibrio spp., type E Clostridium perfringens, and Enterococcus

faecalis and negatively associated with Clostridium butyricum, Ruminococcus albus, and

Eubacterium rectale. Low clinical activity scores were associated with Lactobacillus spp. and

Bifidobacterium spp.

The stability of the gut microbiota can be disrupted by several different kinds of

disturbances (events, biological processes, or physical processes) that separately or together can

cause discrete structural changes in the composition or function of the community (62). Such

changes can disrupt the diversity (including richness and evenness; 70, 71) and damage the

functionality of the ecosystem (72, 73). Changes may include die-off, displacement, or damage

to community members, each of which can allow other organisms to establish themselves within

the ecosystem (74). The environmental factors that can effect such changes to the gut microbiota

include diet, medications (in particular, antibiotics), stress, smoking, chemotherapeutic drugs,

probiotic ingestion and microbial infections (Table 1.2; 18, 27).

The intensity and frequency of disturbance also influences the response patterns of

microbial communities (75). Even brief disturbances to the ecosystem are known to cause lasting

changes to the structure of the gut microbiota. Dethlefsen et al. (2011) found that the fecal

microbiotas from healthy volunteers receiving a 5-day course of ciprofloxacin experienced a loss

of diversity and a compositional shift within 3 to 4 days of antibiotic initiation (76). While fecal

communities began to return to their initial state 1 week after the end of the ciprofloxacin course,

recovery was incomplete. Complete taxonomic recovery of a microbial community following

perturbation may not be as important as the recovery of functions essential for the service output

of the ecosystem (76). In the study by Dethlefsen et al. (2011), the gut microbiota stabilized but

was different from its initial state by the end of the experiment (76). Researchers have proposed

11

that multiple stable states may exist for the gut microbiota, and that perturbation (e.g. brief

antibiotic exposure) may shift the gut microbial community from its current stable state into an

alternative stable state (29, 76-78). However, because what constitutes a healthy stable state is

not currently known or defined, the consequences to health of shifting to alternative stable states

are poorly understood (24, 76).

The ability of the gut microbiota to remain stable in the presence of exogenous stimuli is

expected to be important for host health (61). Due to the intimate relationship between the gut

microbiota and the host, changes in intestinal microbial communities could disrupt physiological

functions, potentially causing disease (61, 79). Low diversity (including richness and evenness)

within a microbial community can reduce its ability to resist perturbation (80). While the host

may not show any signs of disease, their reduced gut microbiota diversity may promote

susceptibility to development of disease in the future (61). Therefore, measures of diversity may

be useful indicators of the stability or "health" of a gut ecosystem and as such, clinical studies

may benefit from incorporating these measures into their experimental design (61).

Characterization of perturbed gut microbiota may aid in designing novel methodologies to

restore dysbiotic gut ecosystems following disturbance, to increase the resilience of disrupted

communities, or to take pre-emptive measures to improve the resistance of the gut microbiota in

susceptible individuals (62).

1.3 Gut microbiota and disease

Dysbiosis is defined as a shift in the balance of the intestinal microbiota to an unbalanced,

non-physiological state, resulting in increased growth of harmful bacteria and deceased growth

of protective bacteria (81, 82). Changes in the microbiota can be quantitative or qualitative and

12

may involve changes in metabolic activity or distribution of commensal organisms (10, 83).

Dysbioses of the gut microbiota may have important implications for host health. Increases in the

incidence of dysbiosis have stimulated interest in manipulating the composition and function of

gut communities using novel therapeutic tools in order to benefit host health (60).

While traditional medical paradigms for infectious disease focus on a single, specific

microbial pathogen as the causative agent, current research has shifted to ecological approaches

proposing the concept of a "microbial community as a pathogen" (62). Comparison of different

compositional backgrounds from diseased patients can allow researchers to identify a disease

phenotype that correlates with specific microbial members or microbial patterns (24). Several

studies have found that the diversity, composition and function of the gut microbiota can vary

according the physiological state of the host (24). For example, the gut microbiotas of obese

individuals have lower Bacteroidetes:Firmicutes ratios than lean people (84-86). In fact, obese

and lean people can be classified solely based on the composition of their gut microbiota with

90% accuracy (87, 88).

A number of human diseases and other forms of pathology have been associated with

alterations in the structure of the gut microbiota (62). However, while special cases have

described the association of aberrant gut microbiota with a specific disease state, it is not always

clear whether dysbiotic gut microbiota are a cause or consequence of disease (13). To date,

associations have been made with aberrant gut microbiota and over 20 diseases, syndromes, or

functional disorders (Tables 1.3, 1.4; 13). These associations have varying degrees of support,

ranging from anecdotal indicators to stronger evidence gleaned from the study of large patient

cohorts (13). Table 1.3 lists disorders with the strongest associations (supported by multiple

studies in humans) along with potential causation (where possible). Table 1.4 lists disorders with

13

Table 1.3 Diseases associated with dysbiotic gut communities based on multiple studies.

Modified from de Vos and de Vos (13).

Disorder Most relevant observations and potential correlation References

Crohn’s disease Decreased gut community diversity, reduced

Faecalibacterium prausnitzii 89-93

Ulcerative colitis Decreased gut community diversity, reduced

Akkermansia muciniphila 89, 92, 94-96

Irritable bowel

syndrome

Differences in global signatures, increased Dorea spp.

and Ruminococcus spp. 97-100

Clostridium difficile

infection

Dramatic decreased gut community diversity, detection

of Clostridium difficile 101-104

Colorectal cancer Differences in Bacteroides spp., increased

Fusobacterium spp. 105-108

Allergy/atopy Decreased gut community diversity, specific signatures 109-112

Celiac disease Differences in the composition of gut communities,

especially in the small intestine 113-115

Type 1 diabetes Signature differences 116-118

Type 2 diabetes Signature differences 119-122

Obesity Differences in the Bacteroidetes/Firmicutes ratio 33, 84, 123,

124

14

Table 1.4 Diseases with preliminary data suggesting an association with dysbiotic gut

communities. Modified from de Vos and de Vos (13).

Disorder Most relevant observations and potential correlation Recent

reference

Alzheimer’s disease Differences in the composition of gut communities (murine

model) 125

Atherosclerosis Correlation between plasma cholesterol levels and several

gut bacterial taxa (human study) 126

Autistic spectrum

disorders

Differences in the composition of gut communities,

reduced Bacteroidetes, increased Betaproteobacteria

(human study)

127

Chronic fatigue

syndrome

Increased abundance of Gram positive facultative anaerobic

microbial species (human study) 128

Colicky babies

Decreased gut community diversity, increased

Proteobacteria, reduced Bifidobacterium spp. and

Lactobacillus spp. (human study)

129

Cardiovascular

disease

Germ-free and antibiotic-treated mice were less susceptible

to disease 130

Depression and

anxiety

Emotional behavior and central gamma-aminobutyric acid

(GABA) receptor expression regulated by the ingestion of a

Lactobacillus strain (murine model)

131

Graft-vs.-host

disease Germ-free mice were less susceptible to disease 132

Multiple sclerosis Germ-free mice were less susceptible to disease 133

Non-alcoholic fatty

liver disease

Correlation between choline levels and the composition of

gut communities (human study) 134

Retrovirus infection Germ-free and antibiotic-treated mice were less susceptible

to disease 135

Poliovirus infection Antibiotic-treated mice were less susceptible to disease 136

15

indications of associations (for example, first report studies), where the data are less strong than

for those listed in Table 1.3.

While the mechanisms underlying these disorders are not always clear, several studies

(including those using germ-free animals) have shown that gut microbes can play an important

role in the initiation and maintenance of disease (137). For example, several studies have

illustrated the importance of gut bacteria in Inflammatory Bowel Disease (IBD). Transgenic and

gene knockout rodents, which normally develop chronic intestinal inflammation when raised

conventionally, fail to develop colitis when raised under germ-free conditions (138). Moreover,

while trinitrobenzene sulfonic acid (TNBS) induces colitis in normal rats, eradication of the gut

microbiota with antibiotics followed by TNBS administration fails to induce colitis (139). As

another example, germ-free mice colonized with the gut microbiota from obese mice experience

a significantly larger increase in their total body fat than when colonized with the gut microbiota

from lean mice (86).

Alterations of the gut microbiota may in some cases be more consistent in individuals with

the same disease (24). Willing et al. (2010) found reproducible alterations in the gut microbiota

of ileal Crohn's disease (CD) patients relative to controls (93). In addition, more subtle

alterations were also noted in patients with colonic CD (93). Metabolomics studies of these

samples also showed consistent differences in the metabolic profiles of Crohn's patients (140). In

contrast, some diseases show inconsistent alterations of the gut microbiota in individuals with the

same disease (24). For example, patients with recurrent Clostridium difficile infection (CDI)

have a reduction in gut microbiota phylum-level diversity that was different from controls but

not similar to other CDI patients (141).

16

Recently, reduced microbial diversity has been linked to disease states associated with

dysbiotic intestinal microbiota (Table 1.3). For example, perturbation of the normal gut

microbiota using broad-spectrum antibiotics creates the largest risk for development of CDI

(142). Patients with recurrent CDI have significant reductions in the diversity of their gut

microbiota compared to both healthy controls as well as patients suffering from an initial episode

of disease (141). Moreover, in the absence of antibiotic treatment, other diseases have been

linked to reduced gut microbiota diversity. Previous studies have shown that both CD and UC

patients had a statistically significant reduction in the diversity of their gut microbiota during the

active disease state (89) as well as during remission (90, 94). Consistent with ecological theory,

the gut microbiota from CD and UC patients showed temporal instability, while healthy

individuals had a stable composition over time (94, 143). Lower gut microbial diversity has also

been associated with obesity (33) and necrotizing enterocolitis (NEC; 144).

Several different therapeutic strategies have been designed to manipulate the gut

microbiota in order to benefit host health. These strategies include use of antibiotics, as well as

ingestion of probiotics, prebiotics, and synbiotics (61, 104). Other potential, but yet unexplored

therapies include immune modulators and bacteriophage therapy (61).

While antibiotics have been shown to perturb gut microbiota, recent studies have suggested

they can also be used as "microbiota-shifting agents" to correct gut microbiota dysbiosis (61).

Antibiotic therapy may be able to induce remission in active CD and UC (145) and has been

shown to relieve Irritable Bowel Syndrome (IBS) symptoms (146). However, it is not clear

whether antibiotic use alleviated disease through the correction of dysbiosis, through the removal

of a specific bacterial pathogen, or by the modulation of host physiology in some other indirect

manner (61).

17

Probiotics are cultures of live, non-pathogenic bacteria (not members of the host

microbiota) that are beneficial to the host when administered in adequate amounts (147). Clinical

trials have indicated that antibiotic-associated diarrhea, NEC, pouchitis, UC, and IBS patients

have benefited from probiotic interventions (148). Probiotics have also been used in association

with prebiotics. Prebiotics are non-digestible food residues (typically carbohydrates) that

selectively stimulate the growth and/or activity of gut microorganisms to benefit host health (3).

Prebiotics have been shown to induce beneficial effects in patients with bowel cancer, IBD,

infections, coronary heart disease, NEC, poor mineral absorption, and obesity (149). Syngeristic

combinations of probiotics and prebiotics (synbiotics) may further benefit host health (61).

Multi-species probiotics, including fecal bacteriotherapy (fecal transplants) and defined

microbial ecosystem therapeutics (MET), have also been developed to treat GI disorders (103,

104). The goal of these therapies is to restore the diversity of a diseased, dysbiotic microbiota

through displacement or augmentation with a healthy microbial community (150). While much

of the data to date has consisted of case studies, a recent clinical trial has shown that fecal

transplants may be an effective treatment for CDI (103). Case reports have also shown the

potential to treat UC (151), IBS (152) and obesity (153) using fecal bacteriotherapy. While

investigations using defined communities are still in their infancy, preliminary data has shown

that ‘synthetic stool’ (MET) was able to resolve disease in two patients with recurrent CDI (104).

Before the gut microbiota can be manipulated to benefit host health, a more complete

characterization of these complex microbial ecosystems is required. An understanding of key

factors that regulate community dynamics would allow researchers to predict the behaviour of

gut ecosystems and manipulate key parameters to control them (154).

18

1.4 Modeling the gut microbiota in vivo

Because of the complexity of the microbial communities in the GI tract, pure culture study

of component species only provides limited information about their role in the ecosystem. The

function and behaviour of these ecosystems are often best studied as a whole.

Human gut microbiota can be studied in a wide variety of individuals, including healthy

volunteers, hospital patients, patients with ileostomies, and sudden death victims (1). Direct

study of the gut microbiota of human subjects has the advantage of biological significance, but

also has several drawbacks. There are a wide range of influences that may affect the gut

microbiota and these can be difficult to control in human studies (27). Inter-individual variation

of the gut microbiota, combined with individualized responses to the same treatment, can also

make it difficult to compare results between treatment and control groups and to correlate

changes in the gut microbiota to the treatment of interest (76). For example, gut microbial

communities from different individuals were shown to have varying short- and long-term

responses to ciprofloxacin perturbation (28, 76). Within the same subject, differences in the

initial compositions of gut microbiota meant that repeat ciprofloxacin exposure also resulted in

varying responses (76). In a separate study, Walker et al. (2011) studied the fecal microbiotas of

14 overweight men consuming precisely controlled diets for 10 weeks (155). While specific

bacterial populations rapidly "bloomed" following changes between the diets, inter-individual

variation was marked and samples clustered more closely by individual than by diet.

There are several other drawbacks to human studies of the gut microbiota. Ethical

considerations often restrict human studies to the analysis of fecal samples, because otherwise,

invasive medical procedures would be required to gain access to intestinal material and tissues

(1, 2). Additionally, experiments involving humans must pass necessarily stringent research

19

ethics approval that would restrict the administration of unsafe or untested compounds and/or

may restrict experimental design (1). Furthermore, human studies are generally expensive, time

consuming, and may require the use of specialist facilities, all of which can limit subject

numbers and therefore study power (1). This issue can be compounded by low volunteer

compliance and high dropout rates (1).

The technical and ethical limitations of studying the gut microbiota in human subjects have

driven the development of several animal and in vitro model systems. Animal studies allow

researchers to conduct more controlled, reproducible experiments, keeping variability due to

environmental (e.g. diet) and genetic influences at a minimum (156). There are a variety of

different types of animals that have been used as models of the gut microbiota, including mice

(157), rats (158), pigs (159), and zebrafish (160), among others. These model animals can be

germ-free, gnotobiotic (germ-free animals inoculated with specific microbes), humanized

(gnotobiotic animals inoculated with human feces), or conventionally raised (native gut

microbiota; 161). Some laboratory animals, such as mice and rats, are available as inbred strains,

outbred stocks, or as a variety of knockouts (1, 162). The clear benefit of using in vivo animal

models to study the gut microbiota is that researchers have direct access to intestinal contents,

tissues, and organs at necropsy, and have more freedom to investigate the effects of various

drugs and toxins (1).

Despite these advantages, differences between the digestive physiologies of humans and

animals raises concerns regarding the relevance of the data obtained from animal studies (163,

164). For example, rodents are coprophages (feed on fecal matter), and their intestinal anatomies,

and physico-chemical and biological environments, are very different from those of humans (1,

13). While the porcine digestive tract is commonly considered to be more similar to that of

20

humans than rodents, the pig’s proximal gut is more densely colonized and the lower bowel is

proportionally larger than in humans. Moreover, animal studies are generally more expensive

than in vitro models and may require the use of specialist facilities (1).

In addition to the other drawbacks of human and animal studies, in vivo studies often limit

the dynamic monitoring of the gut microbiota by deriving their data from end-point

measurements (1, 165). Furthermore, mechanistic studies are often confounded due to the

inability to focus on gut communities without host interference (165). The use of in vitro models

to study the microbial communities of the gut can potentially overcome many of the limitations

of in vivo models (1).

1.5 In vitro human gut fermentation modeling

Like other complex microbial ecosystems, most members of the human gut microbiota

have not been cultured (6), likely because the individual species of the gut microbiota are

difficult to culture axenically in vitro (166). In fact, of the hundreds of different bacterial species

that colonize the human GI tract, up to 80% have not been cultured using conventional

techniques (19). These recalcitrant microorganisms may be hard to grow because their cells are

dead or stressed, they may have obligate requirements for interactions with other bacteria or the

host, or they may require essential nutrients or particular culture conditions (166). It is widely

recognized that novel culture techniques are required to grow these “unculturable” microbes.

In vitro gut fermentation models are technological platforms that can simulate the spatial,

temporal and environmental features that microbes experience within the gut environment (1).

The key objective of in vitro gut fermentation models is to culture stable, complex intestinal

microbial communities in a highly controlled environment (2). Depending on the model, single

21

or multiple vessels are inoculated with fecal samples and operated under anaerobic conditions,

using a temperature, pH, and growth medium that mimic the intestinal segment of interest (2).

Being host-free systems, in vitro gut fermentation models make ideal systems in which to

study microbial perturbations resulting from the addition of exogenous stimuli, as microbial

changes can be measured in isolation of any concurrent effects on the host (1, 2). Whilst this can

be seen as a disadvantage of an in vitro model, a reductionist approach to understanding the

complex interactions of the gut microbiota in isolation of the host also has great value in terms of

mechanistic studies (1). The experimental parameters can be tightly controlled by the researcher

(enabling high stability and reproducibility) and can be sampled as frequently as necessary

without ethical constraints (1). The relative lack of ethical consideration also means that in vitro

modeling is the method of choice for studying the effects of potentially lethal substances on the

microbiota, including toxic, radioactive, and genotoxic compounds (1). Compared to in vivo

models, in vitro gut fermentation models are less expensive to operate and generally easy to set

up (1).

Adaptation, survival, and growth of fecal microorganisms within gut fermentation models

depends on the user-defined experimental parameters of the system, including the composition of

the growth medium, pH, retention time, temperature and the maintenance of an anaerobic

environment (1, 2). Strict control of environmental parameters allows for the maintenance of

steady state conditions within each chemostat vessel where microbial composition and activity

remain consistent over time (2). This results in the development of stable, reproducible

chemostat communities that are ideal for experimental manipulation (2).

In vitro experiments with gut fermentation models are virtually unlimited (2). Vessels can

be seeded with a single species, a defined microbial community, or feces (from either healthy or

22

diseased donors; 1). They have been used to study many different microbial processes, including

carbohydrate and protein fermentation, metabolism of steroids, bile acids, lignans and

phytooestrogens, production and removal of hydrogen and mutagens, and the transformations of

xenobiotic compounds (1). Studies have monitored changes in the composition and metabolic

activity of specific bacterial populations using single or multiple dietary compounds, changes in

gut microbiota functionality due to the administration of potential probiotic strains, and

pathogen-commensal microbiota interactions (167-170).

There are several different versions of in vitro gut fermentation models; these vary in

design and complexity, each simulating different spatial, temporal, nutritional and

physicochemical properties of different colonic regions (Figure 1.2; 2). Because each model has

its own advantages and drawbacks, selection of the model should correspond with the objectives

of the study (1, 2).

Batch fermentation models are closed systems (sealed bottles or reactors) that facilitate the

growth of microorganisms without adding or removing material post-inoculation (Figure 1.2a;

2). Batch cultures are especially useful for determining short chain fatty acid (SCFA) profiles

generated by the metabolism of dietary compounds by fecal communities (2). Since batch

cultures are inexpensive and easy to set up, they facilitate the rapid testing of a wide variety of

substrates or fecal samples (1). However, batch cultures are limited to short-term studies as

changes in substrate availability, pH, and redox potential can result in the selection of non-

representative microbial populations and distorted fermentation profiles (2). This means batch

fermentation models cannot establish or maintain steady state conditions (2).

As stated at the beginning of this chapter, the digestive tract is an open system, where food

enters through the mouth and exits through the rectum. With this in mind, uninterrupted flow of

23

Figure 1.2 Different in vitro gut fermentation models used to culture gut microbial communities.

a) Batch fermentation model used to culture microorganisms without adding additional nutrients.

b) Single-stage chemostat used to simulate a single segment of the GI tract (171). Fresh growth

medium is added and spent culture medium is removed at the same, consistent rate. c) Multi-

stage chemostat used to simulate multiple segments of the GI tract (e.g. proximal, transverse, and

distal colon; 172). Vessels are connected in series and liquid is continuously transferred from the

growth medium bottle to the culture vessels and the effluent bottle at the same rate. d) TIM-1

model simulating the upper intestine (stomach, duodenum, jejunum, and ileum; 173). The TIM-1

model can also mimic host bile secretion, motility, pH and absorption abilities of this region of

the GI tract. e) TIM-2 model simulating the proximal colon (174). The TIM-2 model mimics

host peristaltic mixing, water absorption and metabolite absorption. f) SHIME model simulating

the numerous segments of the GI tract (including the stomach, small intestine, ascending colon,

transverse colon, and descending colon; 165). This semi-continuous model adds growth medium

to the stomach compartment three times per day (to simulate three meals) and then transfers

liquid to other compartments (when available).

24

Media in Culture outP

Media in Culture outP P P P

Proximal

colon

Transverse

colon

Distal

colon

Culture outSHIME feed in

Stomach Small

intestine

Ascending

colon

Transverse

colon

Descending

colon

P

Pancreatic juice inP

P P P P P

Gastric acid

and pepsin

Pancreatin

and bile

Dialysates out

Dialysates out

Culture out

Duodenum

Jejunum

Ileum

P P

Electrolytes in

Electrolytes in

P

P

P

P

Media in

Electrolytes in Dialysates out

Media in

P P

PStomach

a b

c

d e

f

25

liquid through continuous culture fermentation models is thought to better mimic the in vivo

situation (1, 2). Long-term studies can be carried out using continuous culture fermentation

models (or 'chemostats') because these systems persistently add fresh growth medium

(replenishing nutrients) and remove spent culture medium (including toxic products; 2). Similar

to the large intestine, chemostats can effectively culture microbial communities with high cell

densities (1). Chemostat designs can be simple or complex; single-stage chemostats consist of

one vessel to mimic one segment of the GI tract (Figure 1.2b; 2), whereas multi-stage chemostats

consist of several vessels connected in series to replicate multiple sections of the GI tract, usually

the ascending, transverse, and distal colon regions (Figure 1.2c; 172).

In an effort to improve the biological significance of multi-stage gut fermentation models,

artificial digestive systems have been developed that mimic additional host physiologic functions

(2). Two TNO (Dutch Organization for Applied Scientific Research) Intestinal Models (TIMs)

have been developed to simulate different sections of the GI tract. The TIM-1 small intestine

model simulates host bile secretion, motility, pH and absorption abilities of the upper intestine

(Figure 1.2d; 173). The TIM-2 proximal colon model mimics additional host functions including

peristaltic mixing, water and metabolite absorption (Figure 1.2e; 174). TIM-1 and TIM-2 models

have also been combined to create an artificial digestive system to investigate drug delivery

methods in pharmaceutical studies and for more advanced nutritional studies (175-177). In

addition to the TIM models, another artificial digestive system includes the simulator of the

human intestinal microbial ecosystem (SHIME) model (Figure 1.2f). The SHIME is a multi-

stage sequential batch model (semi-continuous culture; 178). It is composed of five vessels

connected in series, with parameters set to mimic the stomach, small intestine and large colon

(ascending, transverse and descending; 178).

26

1.6 Research rationale

While in vitro gut fermentation models have many advantages, like in vivo models they

also have limitations (Table 1.5). Depending on the research question, certain models may be

more appropriate to use than others (2). Model selection often requires a compromise between

technical complexity, biological significance, and cost (2).

The inaccessibility of intestinal material from different segments of the human GI tract

means that gut fermentation models are normally inoculated with feces (1). However, feces are

not representative of the content from of each segment of the GI tract, because substrate

availability, bacterial populations, and microbial metabolism vary along the length of the gut (14,

15, 82). Fecal communities are more representative of communities from the distal gut than

communities from other intestinal segments, thus the most relevant in vitro model that is seeded

with feces is a model of the distal colon. In vitro characterization and modeling of distal gut

diseases with proposed microbial community involvement (including UC, colorectal cancer, and

CDI) might aid in determining the etiology of these diseases (18).

In vitro gut models have been used for many years, however most models have yet to fully

characterize the reproducibility (technical replicates), repeatability (biological replicates), and

stability (time to and maintenance of steady state) of cultured communities (2). A notable

exception is the SHIME system, as studies have confirmed that it can develop reproducible,

stable communities that generally mimic those of the human gut (165).

Aside from dietary degradation experiments, gut models have the potential to answer a

wide variety of ecological questions; for example, studies investigating biofilm formation,

defined microbial communities, novel sources of ecosystem perturbation, and development of

experimental therapies can be facilitated using a relevant in vitro model system. Responses to a

27

Table 1.5 Advantages and limitations of different in vitro gut fermentation models. Modified from Payne et al. (2).

In vitro gut

fermentation models Advantages Limitations References

Batch culture

- Easy to set up

- Useful for substrate digestion and

fermentation studies

- Short-term studies (hours)

- Lack of environmental control 179, 180

Continuous culture

- Continuous flow simulates in vivo

environment

- Highly controlled environmental parameters

- Lack of host functionality 181, 182

Multistage

continuous culture

- Continuous flow into several compartments

simulates different segments of the digestive

tract

- Lack of host functionality

- Inoculated with feces (not

representative of each intestinal

segment)

165, 181

Artificial digestive

system

- Continuous flow simulates in vivo

environment

- Simulates the exchange of metabolites and

water found in vivo

- Short-term studies (days)

- Inoculated with feces (not

representative of each intestinal

segment)

175, 183

28

variety of stimuli with previously characterized outcomes (observed in in vivo models and

validated in vitro models) can aid in confirming the validity of a novel in vitro gut fermentation

model. In addition, modeling distal gut communities allows for the study of extremes of

microbial diversity in this microbial ecosystem, because the distal gut is one of the most diverse

microbial ecosystems known (5, 8).

1.7 Research objectives

The overall objective of my doctoral research was to characterize and validate a twin-

vessel (independent, identical) single-stage chemostat model of the human distal gut. Hypotheses

and objectives relating to different investigations are outlined for each chapter. In chapter 3 we

assessed the time to (and maintenance of) steady state compositions, within-run reproducibility

(technical replicates), between-run repeatability (biological replicates using the same donor), and

inter-individual differences (alternative biological replicates). In chapter 4 we characterized the

effects of defined (clindamycin) and undefined (norepinephrine, NE) perturbation on the

structure of chemostat communities. In chapter 5 we characterized planktonic and biofilm

communities developed in a chemostat model with packed-column biofilm reactors (simulating

mucosal environments) in the presence and absence of antibiotic perturbation. In chapter 6 we

compared simplified, defined communities to complex, undefined fecal communities developed

in our chemostat model. Finally, in chapter 7 we investigated long-term persistence of

allochthonous (non-resident) microorganisms (from defined or fecal communities) within

complex microbial communities simulating the human gut following perturbation (with

antibiotics or by microbial pathogen invasion). Samples were analyzed using a molecular

fingerprinting technique (Denaturing Gradient Gel Electrophoresis, DGGE) to compare and

29

monitor changes in the overall structure of the communities; in addition a phylogenetic

microarray (Human Intestinal Tract Chip, HITChip; 56) was used to obtain phylogenetic

information. Throughout these investigations the comparability between DGGE and HITChip-

calculated parameters were also assessed.

30

Chapter 2

Materials and Methods

2.1 Preparation of single-stage chemostats

An Infors Multifors bioreactor system was used for this work (Infors, Switzerland; Figure

2.1). Conversion from a fermentation system into a chemostat was accomplished by blocking off

the condenser vent and bubbling nitrogen gas through the culture to allow positive pressure to

maintain a constant volume of culture (400 mL), as well as to displace oxygen (Figure A2.1;

184, 185). The vessels were maintained at a temperature of 37ᵒC and a pH of 6.9-7.0 (184).

While both acid (5% (v/v) HCl) and base (5% (w/v) NaOH) were required immediately

following inoculation, only base was required to main culture pH throughout the chemostat runs.

Growth medium (see below) was continuously fed at a rate of 400 mL/day to give a

retention time of 24 hours (165, 186). Each vessel was aseptically sampled to check for absence

of contaminant growth on fastidious anaerobe agar (Acumedia; Lansing, Michigan)

supplemented with 5% defibrinated sheep blood (Hemostat Laboratories; Dixon, California)

(sFAA) prior to inoculation.

Growth medium was based on previous studies attempting to mimic the human gut (172,

187) with several modifications. Media was prepared from 4 separate stock preparations as

outlined in Table 2.1. For 1 L of growth medium, reagents were dissolved in either 800 mL H2O

(preparation 1), 50 mL H2O (preparations 2 and 3) or 100 mL H2O (preparation 4). Preparations

1 and 4 were autoclaved while preparations 2 and 3 were filtered through 0.22µm filters. 2.5 mL

of antifoam B silicone emulsion (J.T. Baker; Center Valley, Pennsylvania) were added to each

31

Figure 2.1 Infors Multifors fermentation vessels adapted into a single-stage chemostat. a) Three

computer-controlled units are shown (each supporting 2 vessels). b) Close-up of twin-vessels

from a single unit (tubes and bottles moved aside to better view vessels). c) Schematic depicting

features of a single vessel run as a chemostat.

32

33

Table 2.1 Composition of the growth medium used to culture microbial communities within the

twin-vessel single-stage chemostat model (per litre of media). Numbers in superscripts denote

chemical suppliers as follows: (1) Sigma-Aldrich (Oakville, Ontario); (2) Thermo Fisher

Scientific (Ottawa, Ontario); (3) BD (Franklin Lakes, New Jersey); (4) Alfa Aesar (Ward Hill,

Massachusetts); (5) BDH (Radnor, Pennsylvania).

Preparation Reagent Weight (g)

1

peptone water2

2

yeast extract3

2

NaHCO32

2

CaCl21

0.01

pectin (from citrus)1

2

xylan (from beechwood)1 2

arabinogalactan1

2

starch (from wheat, unmodified)1 5

casein4

3

inulin (from Dahlia tubers)4 1

NaCl1 0.1

2

K2HPO41

0.04

KH2PO42

0.04

MgSO45

0.01

Hemin5

0.005

Menadione1 0.001

3 bile salts

1 0.5

L-cysteine HCl2 0.5

4 porcine gastric mucin (type II)1 4

34

litre of prepared media (Figure A2.2). Media was stored at 4ᵒC for up to 2 weeks and checked for

sterility by plating on sFAA as above.

2.2 Collection and preparation of fecal inocula

The Research Ethics Board of the University of Guelph approved this study

(REB#09AP011). Four healthy donors provided fresh fecal samples: donor A (male, 44 years-

old), donor B (female, 42 years-old), donor C (male, 26 years-old) and donor D (female, 25

years-old). None of these donors had a recent history of antibiotic treatment prior to this study

(within 6 years for donor A, 8 years for donor B, 9 months for donor C and 11 months for donor

D). Chemostat vessels were inoculated with fresh feces from the donors, as described in Table

2.2.

Fecal samples were provided in a washroom located within the building, in close proximity

to the laboratory in which the anaerobic chamber used to process the sample was located. A

sterile, lidded container was provided for each sample, and this was immediately sealed and

placed in an anaerobic chamber (atmosphere of 90% N2, 5% CO2 and 5% H2) within 5-10

minutes of voiding. A 10% (w/v) fecal slurry was prepared by homogenizing 5g of feces in 50

mL of degassed growth medium for 1 minute using a stomacher (Tekmar Stomacher Lab

Blender, Seward; Worthing, West Sussex, UK). The resulting slurry was centrifuged for 10

minutes at 175 x g to remove large food residues (188). The resulting post-spin supernatant was

used as the inoculum for this study.

There is precedent in the literature for gently centrifuging stool prior to inoculating

chemostat vessels to remove large food particles from the inoculum (188-190). However, we

also carried out a DGGE-based comparison of the fecal slurry to the post-spin supernatant

35

Table 2.2 Composition of the inoculum used to seed the chemostat vessels discussed in this thesis.

Thesis

Chapter Experiments Vessels (chemostat run) Inoculum

3

Time to and maintenance of steady state compositions

Within-run reproducibility (technical replicates)

Between-run repeatability (biological replicates using the

same donor)

Inter-individual differences (alternative biological

replicates)

VA1-1, VA1-2 (run A1)

VA2-1, VA2-2 (run A2)

VB3-1, VB3-2 (run B3)

VB4-1, VB4-2 (run B4)

VC5-1, VC5-2 (run C5)

Donor A feces (first donation)

Donor A feces (second donation)

Donor B feces (first donation)

Donor B feces (second donation)

Donor C feces

4

Perturbation using a single dose of norepinephrine (NE)

Perturbation using a 5 day course of NE

Perturbation using a 5 day course of clindamycin

VB-N1, VB-N2

VC-N1, VB-N2

VA-N1, VA-N2

VA-C1, VA-C2

Donor B feces

Donor C feces

Donor A feces (first donation)

Donor A feces (second donation)

5

Planktonic and biofilm communities developed in a

chemostat model with packed-column biofilm reactors

(simulating mucosal environments) in the presence and

absence of antibiotic perturbation

V1, V2 Donor D feces

6 Simplified, defined communities (DEC, MET-1) vs.

complex, undefined communities (feces)

VF1, VF2

VD1, VD2

VM1

Donor B feces

DEC

MET-1

7

Persistence of members from MET-1 in donor A fecal

chemostat communities in the presence and absence of

clindamycin perturbation

Persistence of microorganisms from donor A and donor B

chemostat communities following mixing (microbial

pathogen invasion)

V(+/+),V(+/- ), V(-/+),V(-/-)

VA

VB

VAB

Donor A feces (first donation)

Donor A feces (second donation)

Donor B feces

VA+VB steady state chemostat communities

36

(inoculum) to ensure that low-speed spinning did not introduce bias. DGGE analysis showed that

the post-spin supernatant is very similar to the fecal sample (87.6%, Figure A2.3).

2.3 Preparation of defined communities (for Chapter 6)

Defined communities were prepared by axenically culturing the microbial community from

a Donor B's stool using a variety of media types (Tables A2.1, A2.2, A2.3). Media included

fastidious anaerobe agar (supplemented with 5% defibrinated sheep blood, 3% spent LS174T

medium + 5% defibrinated sheep blood, or 3% sterile spent chemostat medium + 5%

defibrinated sheep blood), Brain Heart Infusion agar, Wilkins–Chalgren agar, Clostridial

Differential agar, deMan, Rogosa & Sharpe agar, 'Faecali' agar (191), and 'Mucin' agar (192).

Plates were incubated at 37ºC under strictly anaerobic conditions (Ruskinn Bug Box anaerobic

chamber, in an atmosphere of 90% N2, 5% CO2 and 5% H2). Purified isolates were identified by

16S rRNA gene sequencing as previously described (193). Isolates were stored in degassed

freezing media at -80ºC until needed (see Appendix 1.3 for details).

A defined experimental community (DEC) was composed of 50 strains representing the

cultivable community from the fecal donor (Table 2.3). A microbial ecosystem therapeutic

formulation (MET-1) was composed of 33 strains representing a collection of commensal species

isolated from the same fecal donor (Table 2.3; 104). DEC and MET-1 formulations were

prepared by culturing each strain in pure culture on the growth medium from which it was

originally isolated. Culture time was staggered to allow for the growth of slow-growing strains.

The DEC inocula were prepared by scraping bacterial isolates from plates (using the ratios

outlined in Table 2.3) and mixing together in 400 mL of warm (37ºC) degassed saline. The

MET-1 inoculum was prepared by scraping bacterial isolates from plates (using the ratios

37

Table 2.3 List of cultured isolates from the healthy donor B comprising DEC and MET-1 synthetic communities. For the relative

biomass values, zero ("0") indicates the strain was not included in the formulation.

Higher taxonomic group

Closest species match,

inferred by alignment of

16S rRNA sequence

to GreenGenes database

% identity to

closest match

Relative abundance

(by biomass)

in DEC

formulation

(Chapter 6)

in MET-1

formulation

(Chapter 6)

in MET-1

formulation

(Chapter 7)

Actinobacteria

Bifidobacterium adolescentis

(two different strains)

99.79

99.79

1

1

1.5

1.5

1.5

1.5

Bifidobacterium longum

(two different strains)

99.86

99.16

1

1

0.5

2

2.5

2

Bifidobacterium pseudocatenulatum 98.83 1 0 0

Collinsella aerofaciens 98.73 1 1 1

Atopobium minutum 100 0.75 0 0

Bacteroidetes

Bacteroides fragilis 100 1 0 0

Bacteroides ovatus 99.52 1 1.5 1.5

Bacteroides vulgatus 99.2 1 0 0

Parabacteroides distasonis 99.45 1 1.5 1.5

Firmicutes

Bacilli

Lactobacillus casei/paracasei 99.47 1 1 1

Lactobacillus casei 99.74 1 1 1

Lactobacillus ruminis 100 1 0 0

Streptococcus mitis 99.79 0 0.75 1

Clostridium cluster IV

Eubacterium desmolans 94.90 1 1 1.5

Faecalibacterium prausnitzii 99.17 0.33 2 1

Flavonifractor plautii 99.37 1 0 0

Clostridium cluster VIII Clostridium cocleatum 91.92 1 1 1

Clostridium cluster IX Acidaminococcus intestini 100 0 1 1

Clostridium cluster XIVa

Anaerostipes hadrus

(two different strains)

99.59

99.56

0.67

1

0

0

0

0

Blautia sp. 99.55 1 0.75 0.75

38

Table 2.3 Continued

Higher taxonomic group

Closest species match,

inferred by alignment of

16S rRNA sequence

to GreenGenes database

% identity to

closest match

Relative abundance

(by biomass)

in DEC

formulation

(Chapter 6)

in MET-1

formulation

(Chapter 6)

in MET-1

formulation

(Chapter 7)

Firmicutes Clostridium cluster XIVa

(continued)

Clostridium bolteae

(Clostridium sp.)

93.36

(93.78) 0.5 0 0

Clostridium citroniae 99.78 1 0 0

Clostridium hathewayi 99.36 1 0 0

Coprococcus comes

(Clostridium sp.)

98.64

(99.78) 1 0 0

Dorea formicigenerans 100 1 0 0

Dorea longicatena

(two different strains)

99.62

99.60

1

1

1

2

1

2

Eubacterium sp.

(three different strains)

99.33

99.34

99.77

1

1

1

0

0

0

0

0

0

Eubacterium eligens 98.15 0.5 2 2

Eubacterium fissicatena 99.78 1 0 0

Eubacterium rectale

(four different strains)

99.59

99.60

99.19

99.53

1

0.33

1

1

2

1.5

1

1

2

2

1

1

Eubacterium ventriosum 100 0 1 1.5

Lachnospira pectinoshiza 95.22 0.5 0.5 0.5

Roseburia faecalis 99.65 1 1.5 1.5

Roseburia intestinalis 100 1 1 1

Ruminococcus torques

(two different strains)

99.15

99.29

1

1

1

2

2

2

Ruminococcus gnavus 96.48 1 0 0

Ruminococcus sp.

(two different strains)

97.35

98.40

1

1

1.5

0.5

1.5

2

39

Table 2.3 Continued

Higher taxonomic group

Closest species match,

inferred by alignment of

16S rRNA sequence

to GreenGenes database

% identity to

closest match

Relative abundance

(by biomass)

in DEC

formulation

(Chapter 6)

in MET-1

formulation

(Chapter 6)

in MET-1

formulation

(Chapter 7)

Firmicutes

Clostridium cluster XIVb Clostridium lactatifermentans 97.97 1 0 0

Clostridium cluster XV Eubacterium limosum 97.05 1 1.5 1.5

Clostridium cluster XVIII Clostridium spiroforme 90.00 0.5 0 0

Proteobacteria

Escherichia coli

(two different strains)

99.80

99.60

0

1

0.5

0

0.5

0

Raoultella sp. 99.40 1 1 1

40

outlined in Table 2.3) and mixing together in 150 mL of warm (37ºC) degassed saline.

2.4 Inoculation, operation and sampling

'Twin' chemostat vessels consisted of two identical chemostat vessels inoculated with

identical inocula. Chemostat vessels seeded with feces were inoculated by adding 100 mL of

10% fecal inoculum (post-spin supernatant) to 300 mL of sterile growth medium. VD1 and VD2

were inoculated by adding 50 mL of the DEC formulation to 350 mL of sterile growth medium.

VM1 was inoculated by adding 100 mL of MET-1 formulation to 300 mL of sterile growth

medium.

Immediately following inoculation, cultures were set to continual gentle agitation

(maintained throughout the run) and vessel pH was adjusted to 6.9-7.0. Feeding was started 24

hours post-inoculation. Each chemostat vessel was sampled daily by removing 4 mL of culture

through the sampling port and aliquots of each sample were archived at -80ᵒC.

2.5 Combining steady state chemostat communities (Chapter 7)

Stable steady state chemostat communities seeded with feces from donors A (VA) and B

(VB) were combined in a third vessel (VAB). VAB was filled with ~350 mL sterile media and

operated overnight (heating, degassing, and stirring). Media from VAB was sampled and plated

on sFAA to confirm sterility prior to inoculation, following which 217 mL of liquid was drained

from the vessel. Next, using sterile syringes and needles, 133 mL of culture were removed from

VA and transferred to VAB, and 133 mL of culture were removed from VB and transferred to

VAB using a syringe and needle. Thus, following inoculation with VA and VB chemostat

culture, the volume of VAB was 400 mL. 133 mL of sterile growth medium were then pumped

41

into VA and VB to regain a volume of 400 mL in each vessel. All three vessels were then

operated as previously described.

2.6 Norepinephrine (NE)-induced perturbation (Chapter 4)

VB-N1 and VB-N2 and VC-N1 and VC-N2 were operated without experimental

manipulation until 36 days post-inoculation. Immediately following sampling on day 36, a single

physiologically relevant dose of norepinephrine (L-(-)-norepinephrine (+)-bitartrate salt

monohydrate, Sigma-Aldrich) was added to VB-N1 and VC-N1 to a final concentration of 100

µM (194, 195). An equal volume of sterile water was added to VB-N2 and VC-N2 (control

vessels).

VA-N1 and VA-N2 were operated without experimental manipulation until 48 days post-

inoculation. NE was added to VA-N1 to a final concentration of 100 µM at the following time

points (days of chemostat run): 48, 48.5, 49, 49.5, 50, 50.5, 51, 51.5, 52, 52.5. Each time a dose

of NE was added to VA-N1, an equal volume of sterile water was added to VA-N2 (control

vessel).

2.7 Clindamycin-induced perturbation (Chapters 5 and 7)

Vessels were operated without experimental manipulation until 36 days post-inoculation to

allow communities to reach steady state compositions. Clindamycin (clindamycin 2-phosphate,

Sigma-Aldrich) was added to the vessel according to a physiologically relevant regime

simulating a typical clinical course of this antibiotic and its pharmacological accumulation in

stool. Each day dosages were given twice, 12 hours apart: day 36 (of chemostat run) = 12.6

µg/mL, day 37 = 16.6 µg/mL, day 38 = 22.4 µg/mL, day 39 = 124.7 µg/mL, day 40 = 203.8

42

µg/mL (196). Equal dosage volumes of sterile water were added to vessels not receiving

antibiotic treatment.

In Chapter 5 clindamycin was added to V1 while water was added to V2. In Chapter 7

clindamycin was added to V(+/+) and V(+/-) while water was added to V(-/+) and V(-/-).

2.8 Packed-column biofilm reactor preparation (Chapter 5)

A simulated mucosal environment was incorporated into our distal gut chemostat model

using packed-column biofilm reactors (Figure 2.2; 197). A peristaltic pump was used to circulate

culture from the vessel, through the packed column, independently of the pumps used to

maintain continuous culture conditions in the vessel. Flow rate through the column was the same

as that in the parent vessel. Each glass jacketed distilling condenser column (10 mm inner

diameter, 280 mm length, 34 mL total volume) was packed with pieces of silicone tubing

(approximately 2.5 mm width, 5 mm length) to provide a large surface for microbial attachment.

Silicone was chosen as a substrate for biofilm growth because of its mucin-binding capabilities

and autoclavability (198). Columns were maintained at 37ᵒC using a circulating water bath.

Rubber stoppers fitted with 2 mm diameter steel tubing were used to seal each end of the

column, and attached to each vessel with silicone tubing (VWR; Mississauga, Ontario). Two

packed-column reactors were attached to each vessel throughout the course of the experiment, as

outlined in Table 2.4.

2.9 Biofilm harvest from packed-columns (Chapter 5)

To harvest biofilms from the packed-column reactors, the stopper was removed from the

top of the column and 25 mL of sterile phosphate buffered saline (PBS, Gibco, Life

43

Figure 2.2 Packed-column biofilm reactor set-up illustrating the attachment of silicone tubing to

the columns and vessel. P represents the location of the peristaltic pumps and the dashed arrows

illustrate the direction of culture circulation. A-G represents different lengths of silicone tubing

while YC1-3 represent y-connectors used to attach these lengths of tubing. Tubing E was clipped

off while culture was circulating through the columns.

44

P P

G

AB

C

YC1

YC2

YC3

F

E

D

45

Table 2.4 Biofilm initiation, harvest, and development times for biofilms cultured in packed-

column reactors.

Packed-column

reactor

Biofilm growth

initiation day

Biofilm

harvest day

Days of biofilm

development

B1 0 36 36

B2 0 41 41

B3 36 48 12

B4 41 48 7

46

Technologies Corporation; Grand Island, New York) was gently dripped through (to wash away

excess non-adherent microorganisms). Column packing material was transferred to a sterile

conical tube and diluted in 15 mL of Tris EDTA (TE, pH 8.0). To aid in removing adherent

biofilms from the packing material, these tubes were sonicated for 3 minutes in a sonicating

water bath (Branson; Shelton, Connecticut; 199). Aliquots of sonicated column material were

stored at -80ᵒC.

2.10 Adding MET-1 to clindamycin-perturbed chemostat communities (Chapter 7)

Following clindamycin-induced perturbation of chemostat communities (as outlined in

section 2.7), MET-1 was added to clindamycin-treated and untreated vessels (on day 42). The

MET-1 formulation was prepared as described in section 2.3, except that the 33 bacterial isolates

were mixed in 30 mL of warm, degassed saline instead of 150 mL of saline. This was done to

reduce the volume MET-1 added to the vessel to avoid removal of chemostat culture following

inoculation. 10 mL of the MET-1 formulation were added to V(+/+) and V(-/+) while 10 mL of

saline were added to V(+/-) and V(-/-).

2.11 DNA extraction

DNA was extracted from archived samples using bead beating and modified protocols

from commercially available kits. Briefly, 200 µL of each sample were added to 300 µL of SLX

buffer (from the Omega Bio-Tek E.Z.N.A.® Stool DNA kit; Norcross, Georgia), 10 µL of 20

mg/mL proteinase K (in 0.1 mM CaCl2) and 200 mg of glass beads and placed into a bead beater

(Biospec Products; Bartlesville, Oklahoma) for 3 minutes. The samples were then incubated at

70ᵒC for 10 minutes, 95

ᵒC for 5 minutes and on ice for 2 minutes. Next, 100 µL of Buffer P2

47

(from the E.Z.N.A. kit) were added to each tube and vortexed for 30 seconds. This was followed

by incubation on ice for 5 minutes and centrifugation at 20450 x g for 5 minutes. The supernatant

was added to 200 µL of HTR reagent (from the E.Z.N.A. kit), vortexed for 10 seconds and

incubated at room temperature for 2 minutes. Samples were then centrifuged at 20450 x g for 2

minutes and supernatants were transferred into Maxwell®16 DNA Purification Kit cartridges

(Promega; Madison, Wisconsin). The remainder of the extraction protocol was carried out

according to the Maxwell kit instructions.

2.12 DGGE analysis

For DGGE analysis, DNA was extracted from each vessel every 2 days, starting

immediately following inoculation (day 0) until the end of the run. The V3 region (339–539 bp,

Escherichia coli numbering) of the 16S rRNA gene was amplified using primers HDA1 and

HDA2-GC (200). DNA from each sample was amplified in three identical PCR reactions. The

PCR master mix consisted of Tsg DNA polymerase (Bio Basic; Markham, Ontario) and 1x

Thermopol buffer with 2 mM MgSO4 (NEB; Whitby, Ontario), using extracted DNA as a

template. The cycling conditions were: 92ᵒC for 2 min, (92

ᵒC for 1 min, 55

ᵒC for 30 sec, 72

ᵒC for

1 min) x 35; 72ᵒC for 10 min. To confirm the PCR reactions were successful, PCR products were

analyzed by agarose gel electrophoresis (Appendix 1.4.1).

A DGGE standard ladder was prepared using strains available in house from our extensive

collection of human bacterial gut isolates. Strains used were as follows: Coprobacillus sp.

(1/2/53), Enterococcaceae sp. (30/1 aka HMP#323), Veillonella sp. (5/2/43 FAA), Clostridium

sp. (1/1/41 A1 FAA CT2, aka HMP#174), and Propionibacterium sp. (7/6/55B FAA). Strains

with an associated HMP number have been sequenced through the Human Microbiome Project

48

(HMP, http://www.hmpdacc.org/reference_genomes/reference_genomes.php). DNA from these

strains was extracted using the method described by Strauss et al. (193). To prepare the ladder,

the V3 region of the 16S gene was amplified for each of these strains and the reactions were

pooled in equal ratios. Identical ladder samples were run on the first and middle lanes of each

DGGE gel.

Samples were concentrated with an EZ-10 Spin Column PCR Products Purification Kit

(Bio Basic), eluted in high-performance liquid chromatography (HPLC) grade water and mixed

with loading dye (0.05% (v/v) bromophenol blue; 0.05% (v/v) xylene cyanol; 70% (v/v) glycerol

in HPLC grade water; Bio-Rad DCode Manual), and loaded on a prepared gel (Appendix 1.4.3).

DGGE was performed using the DCode System (Bio-Rad Laboratories, Hercules, California)

using the method of Muyzer et al. (201) with a 6% (v/v) polyacrylamide gel (Appendix 1.4.4 and

1.4.5). Amplicons were separated using a denaturing gradient of 30-55%, where 100%

denaturing acrylamide was defined as 7 M urea and 40% formamide. Electrophoresis was

performed for 5 h at 60ᵒC and 120 V in 1x Tris-acetate-EDTA (TAE) buffer. Gels were stained

for 10 minutes with ethidium bromide (Sigma-Aldrich), destained for 10 minutes in ddH2O, and

images captured using a SynGene G-Box gel documentation system and GeneSnap software

(version 6.08.04, Synoptics Ltd; Cambridge, UK). The gels were normalized for saturation while

the images were captured.

DGGE gel images were analyzed using the Syngene GeneTools software (version 4.01.03,

Synoptics Ltd). Similarities between DGGE banding patterns were analyzed by calculating

Pearson correlation coefficient values (similarity index values). Similarity index (SI) values

ranged from 0 to 1, with values of 0 indicating two profiles had no bands in common, while

values of 1 indicate the two profiles had identical banding patterns. Correlation coefficients (%

49

similarity index values, %SI) were calculated by multiplying the similarity index value by 100. A

dendrogram was generated using Pearson correlation coefficient values and the unweighted pair

group with mathematical averages (UPGMA) method.

Gel-specific vessel comparison “cut-off thresholds” were calculated by comparing the

similarity of identical standard ladder profiles within the same DGGE gel. Identical ladder

profiles should be identical (100% similar), however, due to variations in the gradient across the

width of the DGGE gel or experimental error generally associated with DGGE, these values are

always less than 100%. Sample profiles with correlation coefficient values higher than the cut-

off threshold were considered identical. Sample profiles were considered similar if the

correlation coefficient values were within 5% of the cut-off value.

2.13 Community dynamics

The changes within a community over a fixed time frame were measured using moving

window correlation analysis according to the method of Marzorati et al. (202, 203). Each point in

the plot represents the similarity between DGGE profiles from the same vessel on days (x-2) vs.

(x) (203). Plots were used to assess the stability of these communities and the time required to

reach steady state.

Steady state compositions were maintained once rate of change values remained below or

within 5% of the gel-defined cut-off values. The rate of change cut-off thresholds were

calculated by subtracting the vessel comparison cut-off threshold values from 100%.

50

2.14 Compensation for DGGE gel-to-gel variation

DGGE Shannon diversity index (H'), Shannon equitability index (EH'), and richness (S)

values (described in the following sections) were adjusted to create plots which illustrated

general patterns of ecological change over the course of a chemostat run. Within a series of gels

the inoculum from each vessel was loaded onto each gel. In addition, the last time point from one

gel was run as the first time point on the next gel to create overlapping time points. H', EH', and S

values for the inocula were compared between all gels within a chemostat run. The gel which

contained H’, EH’, and S values closest to the average values was set as the 'reference gel' to

which values from other gels were aligned (see Figure 2.3). A 'correction value' was determined

for each parameter and gel by calculating the average difference between identical samples

(inoculum and overlapping time points for both vessels). Sample values for aligned gels were

adjusted by adding (or subtracting) the correction value to all samples within the gel. This shifted

all values up (or down) to align overlapping values with the flanking gel. Parameter values for

overlapping time points were averaged. Since the same correction value was added to (or

subtracted from) all values within a gel, relative within-gel differences were maintained. Gels

that flanked the reference gel were aligned to the reference gel, while gels that did not

immediately flank the reference gel were aligned to gels that had been aligned to the reference

gel.

2.15 Phylogenetic microarray (HITChip) analysis

Microbial community composition was evaluated using the Human Intestinal Tract Chip

(HITChip), a phylogenetic microarray that contains over 5000 probes based on 16S rRNA gene

sequences of over 1,100 intestinal bacterial phylotypes (56). Table 2.5 lists the 46 samples

51

Figure 2.3 Example of an adjusted band richness plot using data from run A1 (Chapter 3 data).

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36 40 44 48

S

Day

VA1-1 Unadjusted

VA1-2 Unadjusted

VA1-1 Adjusted

VA1-2 Adjusted

Day of Overlap

52

Table 2.5 DNA samples analyzed by HITChip analysis.

Chapter 3 samples Chapter 5 samples Chapter 6 samples Chapter 7 samples

VA1-1/2 inoculum V1/2 fecal inoculum VF1 inoculum VA inoculum

VA1-1 day 36 V1-P day 36 VF1 day 36 VA day 36

VA1-2 day 36 V2-P day 36 VD1 inoculum VA day 44

VA1-1 day 48 V1-P day 41 VD1 day 28 VB inoculum

VA1-2 day 48 V2-P day 41 VM1 inoculum VB day 36

VA2-1/2 inoculum V1-P day 48 VM1 day 28 VB day 44

VA2-1 day 36 V2-P day 48 VAB day 36

VA2-2 day 36 V1- B1 days 0-36 VAB day 44

VB3-1/2 inoculum V2- B1 days 0-36

VB3-1 day 36 V1-B2 days 0-41

VB3-2 day 36 V2-B2 days 0-41

VB4-1/2 inoculum V1-B3 days 36-48

VB4-1 day 36 V2-B3 days 36-48

VB4-2 day 36 V1-B4 days 41-48

VC5-1/2 inoculum V2-B4 days 41-48

VC5-1 day 36

VC5-2 day 36

53

analyzed using the HITChip. Two independent hybridisations were performed for each sample

and duplicates with a Pearson correlation over 0.98 were considered for further analysis. Slides’

data were extracted from the microarray images using Agilent Feature Extraction software,

version 10.7.3.1 (http://www.agilent.com). Array normalization was performed with a set of R-

based scripts (http://r-project.org) in combination with a designed relational database that runs

under the MySQL database management system (http://www.mysql.com). Ward’s minimum

variance method was used for the construction of hierarchical clusters of the total microbiota

probe profiles, while the distance matrix between the samples was based on Pearson correlations.

2.16 HITChip statistical data analysis

The stability of the total microbial composition was assessed by Pearson correlation

between the log transformed hybridisation signals for 3699 unique HITChip probes. To evaluate

the significant differences between the microbial composition of the fecal inocula and steady

state (day 36) chemostat communities (Chapter 3) or between planktonic and biofilm samples

(Chapter 5), Wilcoxon signed-rank test corrected for multiple comparisons (using the Benjamini

& Hochberg method) was applied at the genus-like level.

Principal component analysis (PCA) and redundancy analysis (RDA) were performed to

assess bacterial associations under different conditions as implemented in the CANOCO 4.5

software package (Biometrics, Wageningen, the Netherlands). The log transformed hybridization

signals of all probes included in the HITChip or the 131 genus-like level phylogenetic groups

targeted by the HITChip were used as response variables for PCA and RDA respectively.

54

2.17 Shannon diversity index

The Shannon diversity index (H’) was calculated as follows (204, 205):

(Eq. 2.1) ))(ln('1

S

i

ii ppH

For DGGE profiles, S represents the total number of DGGE bands and pi represents the

proportion of the ith band (band peak height). For HITChip data, S represents the total number of

hybridized probe sequences and pi represents the proportion of the ith hybridized probe sequence

(hybridization signal intensity). The minimum value of this index is zero, equal to the value of

H’ for a community with a single species. An increase in H’ may be the result of an increase in

richness, an increase in evenness, or an increase in both.

2.18 Shannon equitability index

The Shannon equitability index (EH’) was calculated as follows (206, 207):

(Eq. 2.2) EH’ = H’/lnS

where H’ represents the Shannon diversity index and S represents the total number of bands (for

DGGE profiles) or total number of hybridized probe sequences (for HITChip data). An EH’ value

of 0 represents complete community unevenness and a value of 1 represents complete

community evenness.

55

2.19 Richness

The number of species present in the ecosystem (richness, S) was approximated by

enumerating the total number of bands (band richness) present in a given DGGE profile, or by

enumerating the total number of hybridized probe sequences (meeting or exceeding a specified

cut-off) for HITChip data (205, 208).

56

Chapter 3

Evaluation of microbial community

reproducibility, stability, and composition

in a human distal gut chemostat model

The data presented in Chapter 3 has been prepared as a manuscript and submitted to the

Journal of Microbiological Methods.

3.1 Introduction

Microbial diversity and community composition within the gut are influenced by several

physical, biochemical, and physiological factors (12), and thus it is important to control and

mimic these factors as closely as possible when designing in vitro simulators. Computer-

operated chemostat models allow for experimental parameters such as pH, temperature, feed

rate, and oxygen levels to be continuously monitored and automatically adjusted to maintain

consistent environmental conditions. However, while in vitro models provide several advantages

over in vivo models, they are inevitably simplified simulations of the in vivo environment (2).

The reproducibility and stability of microbial communities developed in gut fermentation

models are often poorly defined and frequently criticized (2). To represent a valid model of the

human gut, the communities that develop within these vessels must be reproducible, stable, and

must retain the high level of diversity (including species richness and evenness) seen in their

native counterparts (165). They must also share some similarity to the fecal inocula from which

they were derived (165).

In this study we used a molecular fingerprinting technique (DGGE) and a phylogenetic

57

microarray (HITChip; 56) as complementary tools to characterize communities developed in a

'twin-vessel' (independent, identical) single-stage chemostat model of the healthy human distal

gut. The aims of this study were (i) to determine the length of time required for microbial

communities to reach stable or 'steady state' compositions and the length of time these

compositions could be maintained; (ii) to evaluate the within-run reproducibility of communities

developed in twin-vessel single-stage chemostats (technical replicates) as well as the between-

run repeatability of communities developed in twin-vessel single-stage chemostats using

consecutive fecal donations (biological replicates); (iii) to assess how communities seeded with

feces from different donors develop within our system (alternative biological replicates); (iv) to

obtain phylogenetic information on the communities developed and maintained within this

system, whilst also assessing the comparability between DGGE and HITChip-calculated

parameters.

3.2 Results

Five separate chemostat runs were analyzed in this study: twin-vessels seeded with donor

A feces, first donation (VA1-1 and VA1-2); twin-vessels seeded with donor A feces, second

donation (VA2-1 and VA2-2); twin-vessels seeded with donor B feces, first donation (VB3-1

and VB3-2); twin-vessels seeded with donor B feces, second donation (VB4-1 and VB4-2); twin-

vessels seeded with donor C feces, single donation (VC5-1 and VC5-2). VA2-1, VA2-2, VB3-1,

VB3-2, VB4-1, VB4-2, VC5-1 and VC5-2 were analyzed until day 36, while VA1-1 and VA1-2

were analyzed until day 48.

58

3.2.1 Establishment of steady state microbial community compositions in a twin-vessel single

stage distal gut model

Twin-vessels from five different chemostat runs were analyzed by DGGE to determine the

time required for communities to reach steady state compositions. The length of time required for

twin-vessel correlation coefficients to rise above the gel-defined cut-off thresholds varied

depending on the chemostat run being analyzed. However, all five chemostats were at steady

state by 36 days post-inoculation (Figure 3.1a, Table 3.1). Moving window correlation analysis

showed that rate of change (Δt) values were the highest between days 0-18, but decreased and

stabilized as steady state compositions were established (Figure 3.1b). As with the vessel

correlation coefficients, rate of change values for twin-vessels took varying lengths of time to

drop below the gel-specific cut-off thresholds, but occurred by 34-days post-inoculation in all

five runs (Table 3.1). Band richness values fluctuated in the beginning of the experiment, but

stabilized by 34 days post-inoculation in all five runs (Figure 3.1d, Table 3.1). Shannon diversity

and equitability index values remained relatively stable (H' ± 0.12, EH' ± 0.04 on average)

compared to the other parameters measured, and were therefore less informative regarding the

time required for chemostat communities to reach steady state (Figure 3.1c,e).

3.2.2 Reproducibility of fecal inocula and chemostat communities (technical replicates)

To assess between-vessel reproducibility, two identical twin-vessels (VA1-1 and VA1-2)

were compared until 48 days post-inoculation. VA1-1 and VA1-2 DGGE profiles were 98.1%

similar immediately following inoculation (day 0) and 93.3% similar during steady state (36 days

post-inoculation, Figure 3.2a). VA1-1 and VA1-2 HITChip profiles were 96.3% similar on day

36 and clustered together on PCA and RDA plots (Figures 3.2b, 3.3, and A3.1). Twin-vessel

59

Figure 3.1 DGGE community analysis of chemostat runs seeded with feces from three different

healthy donors (donors A, B, and C). With the exception of the plot shown in panel a, each line

represents the average values of samples taken from two identical chemostat vessels (where

similar values were seen between both vessels). Samples were analyzed every two days until

steady state conditions were met (days 0-36). Gel-specific cut-off values are not shown for

simplification purposes. a) Correlation coefficients (expressed as percentages) comparing the

profiles of each vessel at identical time points. b) Community dynamics calculated using moving

window correlation analysis. c) Adjusted Shannon Diversity Index (H’). d) Adjusted band

richness (S). e) Adjusted Shannon equitability index (EH’).

60

0

10

20

30

40

50

60

70

80

90

100

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36

%S

I

Day

a

0

10

20

30

40

50

60

70

80

90

100

2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36

%S

I

Day (x-2) vs. (x)

b

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36

H’

Day

c

0

10

20

30

40

50

60

70

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36

S

Day

d

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36

EH

’

Day

Run A1

Run A2

Run B3

Run B4

Run C5

e

61

Table 3.1 Time (in days) to steady state compositions based on measured parameters from

DGGE gels (bracketed numbers refer to values that met the 5% cut-off threshold by day 36).

Parameter Run A1 Run A2 Run B3 Run B4 Run C5

Vessel Comparison (%SI) 20 34 30 (36) 0

Rate of change (Δt) 18 (34) 28 (34) 30

Band richness (S) 34 34 28 30 30

Time to steady state (days) 34 34 30 36 30

62

Figure 3.2 Dendrogram comparing communities present in twin-vessels for three different

donors (A, B, and C). Dendrograms are based on (a) Pearson and UPGMA correlation (DGGE

profiles) or (b) Pearson correlation and Ward's minimum variance method (HITChip profiles).

HITChip profiles also list the highest phylogenetic level of specificity of probes (level 1) on the

right of panel b. Runs A1, A2, B3, and B4 were used to assess biological repeatability and runs

A1, B3 and C5 were used to compare between-donor differences.

63

VA

1-1

/2 D

ay 0

VA

2-1

/2 D

ay 0

VB

3-1

/2 D

ay 0

VB

4-1

/2 D

ay 0

VB

4-2

Day

36

VB

4-1

Day

36

VB

3-1

Day

36

VB

3-2

Day

36

VA

2-1

Day

36

VA

2-2

Day

36

VA

1-1

Day

36

VA

1-2

Day

36

VC

5-1

Day

36

VC

5-2

Day

36

VC

5-1/

2 D

ay 0

VA

1-1

Day

0

VA

1-2

Day

0

VA

2-1

Day

0

VA

2-2

Day

0

VB

3-1

Day

0

VB

3-2

Day

0V

B4

-1 D

ay 0

VB

4-2

Day

0V

B4

-2 D

ay 3

6

VB

4-1

Day

36

VB

3-1

Day

36

VB

3-2

Day

36

VA

2-1

Day

36

VA

2-2

Day

36

VA

1-1

Day

36

VA

1-2

Day

36

VC

5-1

Day

36

VC

5-2

Day

36

VC

5-1

Day

0V

C5

-2 D

ay 0

Actinobacteria

Asteroleplasma

Bacilli

Bacteroidetes

Clostridium cluster I

Clostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XI

Clostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XVClostridium cluster XVIClostridium cluster XVIIClostridium cluster XVIIICyanobacteriaFusobacteria

Proteobacteria

Spirochaetes

Uncultured Clostridiales

Uncultured MollicutesVerrucomicrobia

b a

64

Figure 3.3 Principal component analysis (PCA) at the oligonucleotide-level based on HITChip

data of the fecal inocula and steady state chemostat communities.

65

DGGE correlation coefficients remained above their gel-defined cut-off thresholds between days

20-42 and within 5% of the cut-off thresholds between days 44-46 (Figure 3.4a). On day 48, the

correlation coefficient dropped to 81.7% (by DGGE) or 93.9% (by HITChip).

DGGE cluster tree analysis showed that VA1-1 and VA1-2 fecal inocula (day 0) grouped

with each other but separately from day 34-48 samples (by DGGE, Figure 3.5). VA1-1 and VA1-

2 grouped together between days 34-44, however samples began to cluster separately on day 46.

Fecal inocula and steady state community profiles from each vessel were compared to

determine how the community structure changed following the in vivo to in vitro transition.

DGGE and HITChip analysis showed reproducible changes in microbial populations during

steady state conditions relative to the fecal inoculum (Tables 3.2 and A3.1, Figures 3.6 and

A3.2). These differences are also illustrated in PCA and RDA plots (Figures 3.3 and A3.1).

Moving window correlation analysis showed that twin-vessel communities developed in a

similar manner for the duration of the run (Figure 3.4b). Δt values were larger and more variable

at the beginning of the run (during the stabilization period). Δt values decreased and stabilized as

steady state conditions were established, dropping and remaining below the gel-defined cut-off in

both vessels by day 18. On day 36 (during steady state conditions) Δt values were 9.3% in VA1-1

and 8.3% in VA1-2.

Shannon diversity and equitability index values were reproducible and relatively stable

throughout the course of this chemostat run (Figure 3.4c,e). By DGGE, day 0 H’ values were

3.33 and 3.34 for VA1-1 and VA1-2, respectively, while EH’ values were 0.82 for both vessels.

On day 36, H’ values were 3.30 and 3.34 and EH’ values were 0.80 and 0.81 for VA1-1 and VA1-

2, respectively. By HITChip, the fecal inoculum had an H’ value of 5.93 and EH’ value of 0.82.

On day 36, H’ values were 5.45 and 5.53 and EH’ values were 0.81 and 0.82 for VA1-1 and

66

Figure 3.4 DGGE community analysis of two identical chemostat vessels modeling the human

distal gut. The vertical dashed line represents the beginning of steady state conditions. a)

Correlation coefficients (expressed as percentages) comparing the profiles of each vessel at

identical time points. The horizontal dotted line represents the cut-off threshold while the

horizontal dashed line represents the cut-off threshold -5%. b) Community dynamics as shown

using moving window correlation analysis. Variability of the community within each vessel was

calculated by comparing the profile of day (x) and day (x – 2). The horizontal dotted line

represents 100-(cut-off threshold) and the horizontal dashed line represents 100-(cut-off

threshold -5%). c) Shannon Diversity Index (H’) representing the adjusted diversity of each

vessel. d) Band richness (S) representing the adjusted richness of each vessel. e) Shannon

equitability index (EH’) representing the adjusted evenness of each vessel.

67

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36 40 44 48

%S

I

Day

VA1-1 vs. VA1-2

Steady state

Cut-off

Cut-off -5%

a

0

10

20

30

40

50

60

70

80

90

100

2 6 10 14 18 22 26 30 34 38 42 46

%S

I

Day (x-2) vs. (x)

VA1-1 VA1-2 Steady state Cut-off Cut-off+5%

b

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36 40 44 48

H'

Day

VA1-1 VA1-2 Steady state

c

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36 40 44 48 S

Day

VA1-1 VA1-2 Steady state

d

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36 40 44 48

EH

'

Day

VA1-1 VA1-2 Steady state

e

68

Figure 3.5 Dendrogram based on Pearson and UPGMA correlation of the DGGE profiles

comparing communities cultured in two identical chemostat vessels inoculated with feces from a

healthy donor (Donor A).

69

VA

1-2

Day

0V

A1

-1 D

ay 4

6V

A1

-1 D

ay 3

6

VA

1-2

Day

34

VA

1-1

Day

34

VA

1-1

Day

44

VA

1-1

Day

38

VA

1-1

Day

40

VA

1-1

Day

42

VA

1-2

Day

44

VA

1-2

Day

42

VA

1-2

Day

40

VA

1-2

Day

36

VA

1-1

Day

48

VA

1-2

Day

38

VA

1-2

Day

46

VA

1-2

Day

48

VA

1-1

Day

0

47

100

95

89

84

79

74

68

63

58

52

70

Table 3.2 Percent abundance of higher taxonomic groups (~ phylum/class level) based on

HITChip analysis for the fecal inocula and twin-vessel steady state (day 36) chemostat

communities.

71

Higher taxonomic groups

Donor A Donor B Donor C

Feces Day 36 Feces Day 36 Feces Day 36

Run

A1

Run

A2

VA1-

1

VA1-

2

VA2-

1

VA2-

2

Run

B3

Run

B4

VB3-

1

VB3-

2

VB4-

1

VB4-

2

Run

C5

VC5-

1

VC5-

2

Actinobacteria 1.71 1.31 0.09 0.08 0.10 0.13 9.38 4.83 0.17 0.21 0.20 0.23 5.27 0.35 0.15

Bacteroidetes 19.07 9.65 57.83 53.27 59.70 48.37 17.11 34.87 56.39 51.62 57.40 49.71 47.57 63.42 71.32

Firmicutes

Bacilli 1.32 0.80 0.18 0.21 0.21 0.28 3.61 5.55 0.29 0.38 0.22 0.30 0.37 0.32 0.21

Clostridium cluster I 0.20 0.55 0.05 0.05 0.06 0.07 0.05 0.05 0.06 0.09 0.07 0.09 0.06 0.10 0.06

Clostridium cluster III 0.27 0.73 0.19 0.20 0.08 0.16 0.31 0.34 0.02 0.04 0.14 0.04 0.13 0.05 0.04

Clostridium cluster IV 25.29 25.41 23.00 24.53 26.61 31.52 14.56 15.44 13.05 8.82 3.19 8.55 9.05 15.68 12.17

Clostridium cluster IX 0.11 0.10 0.05 0.05 0.06 0.07 0.12 0.11 0.40 0.33 0.44 0.43 0.09 0.13 0.10

Clostridium cluster XI 0.38 0.63 0.13 0.14 0.15 0.17 0.10 0.08 0.13 0.13 0.09 0.13 0.13 0.12 0.07

Clostridium cluster XIII 0.02 0.02 0.01 0.01 0.01 0.02 0.01 0.01 0.02 0.02 0.02 0.02 0.01 0.03 0.02

Clostridium cluster XIVa 50.10 57.24 14.47 17.95 9.84 16.02 54.08 37.88 21.76 31.27 29.88 32.05 36.51 17.22 14.01

Clostridium cluster XV 0.02 0.02 0.12 0.02 0.16 0.20 0.02 0.02 0.61 0.82 0.64 0.81 0.02 0.03 0.02

Clostridium cluster XVI 0.12 0.11 0.04 0.03 0.03 0.04 0.03 0.04 0.04 0.06 0.04 0.05 0.27 0.06 0.05

Clostridium cluster XVII 0.01 0.01 0.00 0.00 0.00 0.01 0.01 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.01

Clostridium cluster XVIII 0.02 0.03 0.02 0.02 0.02 0.03 0.12 0.21 0.02 0.04 0.03 0.04 0.07 0.04 0.03

Uncultured Clostridiales 0.81 2.66 1.12 1.83 1.31 0.97 0.11 0.12 1.02 1.14 1.47 1.43 0.20 2.04 1.47

Total Firmicutes 78.67 88.31 39.38 45.04 38.56 49.55 73.13 59.86 37.42 43.14 36.23 43.94 46.91 35.81 28.25

Cyanobacteria 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00

Fusobacteria 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.01 0.02 0.01 0.02 0.01

Proteobacteria 0.40 0.55 1.34 0.95 1.09 1.17 0.27 0.34 3.23 2.32 2.78 3.36 0.19 0.31 0.21

Spirochaetes 0.01 0.01 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

Tenericutes Asteroleplasma 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.01 0.00 0.01 0.00

Uncultured Mollicutes 0.04 0.05 0.03 0.03 0.05 0.06 0.06 0.05 0.12 0.08 0.04 0.06 0.03 0.06 0.04

Verrucomicrobia 0.08 0.11 1.30 0.60 0.48 0.69 0.02 0.03 2.66 2.59 3.33 2.66 0.01 0.01 0.01

72

Figure 3.6 Relative abundance of higher taxonomic groups (~ phylum/class level) based on HITChip analysis for the fecal

inocula and steady state (day 36) chemostat communities.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Run A1

Feces

Run A2

Feces

Run A1

Day 36

Run A2

Day 36

Run B3

Feces

Run B4

Feces

Run B3

Day 36

Run B4

Day 36

Run C5

Feces

Run C5

Day 36

Rel

ati

ve

Con

trib

uti

on

Actinobacteria

Bacteroidetes

Bacilli

Clostridium cluster I

Clostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XI

Clostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XV

Clostridium cluster XVI

Clostridium cluster XVII

Clostridium cluster XVIII

Uncultured Clostridiales

Cyanobacteria

Fusobacteria

Proteobacteria

Spirochaetes

Asteroleplasma

Uncultured Mollicutes

Verrucomicrobia

73

VA1-2, respectively. DGGE band richness was very reproducible throughout the duration of the

experiment (Figure 3.4d). Reproducible fluctuations in DGGE band numbers were seen between

days 0-34, however values were reproducible and stable between days 34-48. On day 0, VA1-1

and VA1-2 DGGE profiles had 60 bands, while day 36 profiles had 64 bands. HITChip data

showed that the fecal inoculum had hits against 1041 probes on day 0 while VA1-1 and VA1-2

had hits against 692 and 698 probes on day 36, respectively.

3.2.3 Repeatability of steady state communities seeded with consecutive fecal donations from

the same donor (biological replicates)

To characterize biological repeatability, donors A and B provided fecal samples on two

separate occasions to seed 2 consecutive chemostat runs (2 runs per donor, 4 runs total). Fecal

inocula (day 0) and steady state (day 36) communities were compared for donor A (runs A1 and

A2) or B (runs B3 and B4) on a single DGGE gel to allow for direct numerical comparisons.

Note that the use of different DGGE gels for calculation of these values within the technical

replicates section above resulted in minor gel-to-gel variations and thus calculated values here

show minor differences to those above.

Consecutive chemostat runs seeded with different fecal donations from the same donor had

higher within-run (between-vessel) reproducibility than between-run repeatability. Table A3.2

compares Donor A profiles between runs A1 and A2 and Donor B between runs B3 and B4. H’,

EH’, and S values confirmed that within a given run, twin-vessel communities were more

reproducible than the corresponding communities from a consecutive chemostat run (Table

A3.3). Clustering tree analysis showed that between-run day 0 DGGE profiles clustered together,

and between-run day 36 profiles clustered together (for each donor, Figure 3.2). As for the

74

technical replicates described above, comparison of fecal inocula and steady state chemostat

communities showed consistent twin-vessel differences between consecutive runs (Table A3.1).

Consecutive chemostat runs had more repeatable fecal inocula than steady state chemostat

communities. Donor A DGGE profiles from runs A1 and A2 were 88.1% similar on day 0 and

72.7% on day 36 (on average). Donor A HITChip profiles from runs A1 and A2 were 94.3%

similar on day 0 and 90.6% on day 36 (on average). Donor B DGGE profiles from runs B3 and

B4 were 86.6% similar on day 0 and 82.3% similar on day 36 (on average). Donor B HITChip

profiles from runs B3 and B4 were 97.4% similar on day 0 and 94.4% similar on day 36 (on

average). Again, H’, EH’, and S values confirmed this observation (Table A3.3).

HITChip data from runs A1 vs. A2 and runs B3 vs. B4 on days 0 and 36 provided

phylogenetic information and illustrated between-run repeatability of fecal and steady state

communities using consecutive fecal donations (Figure 3.6). Twin-vessel day 36 samples had

similar compositions, as shown in Table 3.2 and Figure A3.2. PCA and RDA plots illustrate

between-vessel and between-run grouping of fecal and steady state chemostat samples (Figures

3.3 and A3.1).

3.2.4 Development of microbial communities seeded with feces from different donors

Three chemostat runs (A1, B3, and C5) seeded with feces from different donors (A, B, and

C) were compared to characterize inter-individual variation of communities that developed

within our model. Fecal inocula (day 0) and steady state (day 36) communities from runs A1, B3,

and C5 were compared on a single DGGE gel to allow for direct numerical comparisons.

The steady state profiles from donors A, B, and C retained donor specific signatures

throughout each run. Consecutive runs seeded with feces from the same donor were more similar

75

to each other on days 0 or 36 than runs seeded with feces from different donors (Tables A3.2 and

A3.4). For example, on day 0 run A1 HITChip profiles were 94.3% similar to run A2 samples,

but only 70.4% similar to run B3 and 64.9% similar to run C5 samples. On day 36, run A1

HITChip profiles were 90.6% similar to run A2 samples, but only 63.2% similar to run B3 and

70.3% similar to run C5 samples.

Clustering tree analysis showed that day 0 samples clustered more closely with other day 0

samples than with day 36 samples from the same run (and vice versa; Figure 3.2). In general,

comparison of day 0 and 36 samples within each run showed that our chemostat model was not

biased towards more accurate modelling for one donor sample over another (Table A3.1).

HITChip data from runs A1, B3, and C5 on days 0 and 36 provided phylogenetic

information and illustrated between-donor variability of fecal and steady state communities

(Figure 3.6, Table 3.2). PCA and RDA plots illustrate between-run and between-donor grouping

of fecal and steady state chemostat samples (Figures 3.3 and A3.1). HITChip analysis of all five

chemostat runs showed that steady state cultures were enriched in Bacteroidetes but not

Clostridium cluster XIVa, Bacilli or other Firmicutes relative to the corresponding fecal

inoculum (Table 3.3, Figure A3.1).

H’, EH’, and S values indicated that vessels seeded with feces from different donors

resulted in reproducible steady state communities with stable, biologically-relevant levels of

diversity, evenness, and richness (Table A3.3).

3.3 Discussion

Single-stage in vitro models can develop and maintain complex communities derived from

human fecal samples. We compared the colonization process in a twin-vessel single-stage

76

Table 3.3 Comparison of bacterial groups (belonging to higher taxonomic groups) of average

fecal inocula and average steady state (day 36) chemostat communities from all 5 chemostat

runs. P values were calculated using the Wilcoxon signed-rank test corrected for multiple

comparisons (applied at the genus-like level). Shaded p values remain significantly different

after correction. SD = standard deviation.

Level 1

(~ phylum/class level)

Level 2

(genus-like level) p values

Fecal inoculum

abundance (%) SD

Day 36

abundance (%) SD

Bacteroidetes

Allistipes et rel. 0.153 4.00 1.11 5.71 1.20

Bacteroides fragilis et rel. 0.006 1.03 0.73 6.59 2.70

Bacteroides ovatus et rel. 0.006 1.49 1.15 8.52 1.72

Bacteroides splachnicus et rel. 0.153 2.21 1.53 4.46 1.00

Bacteroides stercoris et rel. 0.063 1.29 0.83 2.61 0.54

Parabacteroides distasonis et rel. 0.010 3.12 1.63 7.58 2.34

Prevotella ruminicola et rel. 0.006 0.55 0.36 2.10 0.41

Tannerella et rel. 0.043 1.85 0.60 2.91 0.50

Bacilli

Aneurinibacillus 0.090 0.01 0.00 0.01 0.00

Bacillus 0.121 0.01 0.00 0.02 0.01

Streptococcus bovis et rel. 0.006 1.36 1.26 0.05 0.01

Streptococcus intermedius et rel. 0.010 0.11 0.07 0.02 0.01

Streptococcus mitis et rel. 0.010 0.65 0.59 0.04 0.02

Clostridium cluster III Clostridium stercorarium et rel. 0.043 0.35 0.20 0.09 0.07

Clostridium cluster IV Papillibacter cinnamivorans et rel. 0.006 0.65 0.25 0.07 0.04

Clostridium cluster IX Peptococcus niger et rel. 0.121 0.03 0.01 0.01 0.00

Clostridium cluster XIVa

Bryantella formatexigens et rel. 0.063 1.87 0.74 0.68 0.89

Clostridium nexile et rel. 0.019 2.48 0.57 0.71 0.64

Coprococcus eutactus et rel. 0.006 1.56 1.12 0.13 0.05

Dorea formicigenerans et rel. 0.090 3.93 1.03 1.68 1.33

Eubacterium hallii et rel. 0.029 0.93 0.42 0.21 0.20

Eubacterium ventriosum et rel. 0.006 1.29 0.24 0.30 0.19

Lachnospira pectinoschiza et rel. 0.029 4.70 2.68 0.96 0.69

Outgrouping Clostridium cluster XIVa 0.006 2.78 1.23 0.61 0.23

Roseburia intestinalis et rel. 0.006 4.28 0.85 0.61 0.54

Ruminococcus gnavus et rel. 0.006 1.02 0.17 0.32 0.14

Ruminococcus lactaris et rel. 0.006 0.83 0.19 0.12 0.06

Ruminococcus obeum et rel. 0.006 5.43 1.55 0.66 0.22

Uncultured Clostridiales Uncultured Clostridiales II 0.006 0.18 0.11 1.24 0.29

Proteobacteria Burkholderia 0.043 0.00 0.00 0.03 0.03

Escherichia coli et rel. 0.029 0.02 0.00 0.03 0.01

Verrucomicrobia Akkermansia 0.153 0.05 0.04 1.43 1.19

77

chemostat mimicking the distal colon. Five different runs were analyzed for three different

donors for at least 36-days post inoculation. DGGE was used as a molecular fingerprinting

technique to provide an overall view of the microbial community structure, diversity, and

dynamics. After selecting key samples based on DGGE analysis, HITChip analysis was

performed to gather phylogenetic information of the fecal inocula and steady state chemostat

communities.

Inoculation of in vitro chemostat models results in the development of a “newly balanced”

gut microbial ecosystem that is influenced by the initial fecal diversity as well as environmental

conditions imposed by the system (2). Since chemostat models are simplified systems, these

applied environmental conditions can only attempt to simulate those of the host and are unlikely

to be exact replications of the in vivo environment (2). For this reason we have not attempted to

produce an exact copy of the human distal gut in vitro, but rather to culture stable, diverse

communities representative of gut communities that are suitable for the study of experimental

perturbations (2). Invariably, communities developed in the twin-vessel single-stage chemostat

did not contain the initial quantitative composition of the fecal inocula as differences in the ratios

of bacterial populations were noted (as expected; 165, 209). Overall, HITChip analysis showed

that steady state cultures were enriched in Bacteroidetes but not Clostridium cluster XIVa,

Bacilli or other Firmicutes relative to the corresponding fecal inoculum. HITChip data from a

SHIME (multistage semi-continuous gut model) study (165) and a TIM-2 (proximal colon gut

model) study (209) were consistent with our results, as in vitro communities were enriched in

Bacteroidetes but not some Firmicutes (including Clostridium cluster XIVa) in addition to other

differences. It is also important to remember that feces may contain transient species ingested by

the host (e.g. from food) that are not maintained as part of the gut system (210). Overall, the

78

compositions of steady state chemostat communities developed within our model were in general

agreement with other in vitro models and representative of the corresponding human fecal

communities.

While growing the gut microbiota outside of the host is the goal of in vitro systems such as

chemostats, these vessels cannot completely mimic the gut environment, including activities

such as the specific absorption of metabolites, secretion of host molecules, or host defensive

mechanisms (including innate and adaptive immunity; 1). In addition, microbes within chemostat

communities could have been lacking essential (not yet identified) nutrients, particular culture

conditions (166), or essential interactions with washed out bacteria (1, 166). Finally, transition

from an in vivo to an in vitro environment could kill sensitive populations of bacteria (165) or

convert stressed populations into a viable but non-culturable state resulting in their wash out

from the system (166).

The ability of a chemostat model to develop robust microbial communities depends on the

certainty that perturbation of composition and function are due to the applied experimental

treatment and not the dynamics associated with adaption to a simulated distal colon environment

(2). Compositional steady state was achieved in our system by 36 days post-inoculation in all

five chemostat runs, however the time to steady state varied according to chemostat run and was

achieved earlier in some runs. Analysis of run A1 over 48 days showed that identical

communities were maintained until day 42, while the communities remained very similar to each

other until day 48. Steady state communities were most easily evaluated by examining the

correlation coefficients comparing duplicate vessels, the moving-window correlation analysis of

both vessels, and band richness plots (Figure 3.1a,b,d). Communities were considered to be at

steady state once the twin-vessel DGGE correlation coefficients rose and stayed above the cut-

79

off threshold, once the rate of change values dropped and stayed below the cut-off threshold, and

once band richness stabilized. Diversity and evenness were generally stable regardless of the

time point.

Similar to other gut models, we found that microbial richness was lost following

inoculation of the vessels (165). It may be tempting to initiate experimentation earlier in the run

to examine the effects of perturbation on biologically important taxa lost over time. However, the

development of stable, highly reproducible communities is required for placebo-controlled or

multi-treatment experiments to be conducted on diverse communities simulating the human gut

microbiota (165). To assess the effect of a treatment on community composition and structure it

is important to operate a “test” vessel in parallel with a “control” vessel that contains identical

steady state communities at the time of treatment. Otherwise it cannot be guaranteed that

changes in complex chemostat communities are due to the applied treatment or to adaptation of

the fecal communities to the chemostat system. We propose that the requirement for stability

justifies the loss of diversity in order to ensure reproducibility of developed communities.

Other published studies using in vitro gut models initiated experimental perturbation after

allowing the communities to develop for 1-48 days (165, 172, 184, 187, 203, 211). While some

studies justified the time required to reach steady state, the motivation behind the time period

used in other studies was not always clear. We were unable to find any studies that defined the

steady state window during which stable community compositions could be maintained. While it

is not unexpected that the time to steady state would vary using different chemostat models with

different media compositions, operational parameters and fecal donors, it is important to

establish the time it takes for a community to reach steady state and the length of time steady

state can be maintained before experimentation can begin. It is also important to have a method

80

to determine whether steady state compositions have been perturbed following treatment.

In agreement with previously published work (165), our model was able to culture

transient and steady state communities that developed in a similar manner over the course of

each experiment. An initial peak of change was noted at the beginning of each run as the

communities transitioned from an in vivo to an in vitro environment. Changes in community

composition during this transition period resulted in reproducible differences between the fecal

inocula and the corresponding steady state communities. As the communities stabilized within

both vessels the community dynamics decreased and steady state conditions were maintained.

H’, EH’ and S were reproducible upon the establishment of steady state conditions.

While some studies have made an effort to ensure the reproducibility of the communities

developed within their in vitro system (165), other studies never fully confirmed this requirement

(212, 213). Our single-stage chemostat system was able to support the development of two

identical steady state communities which were as reproducible as communities developed in a

multi-stage chemostat system. Van den Abbeele et al. (2010) evaluated the reproducibility of

communities developed in a Simulator of the Human Intestinal Microbial Ecosystem (SHIME)

by simultaneously inoculating two SHIME units with identical fecal material (165). By day 26

they found that the descending colon compartments showed a 76% correlation (by DGGE).

These authors also concluded that steady state communities were developed 21 days post-

inoculation. In our single-stage chemostat model of the distal gut we found that the communities

developed in two identical vessels were 93.7% similar on day 26 (by DGGE). On day 36, once

the vessels reached steady state, the vessels showed a 93.3% correlation (by DGGE).

There is a general lack of information in the literature regarding the ability of individual

systems to develop reproducible steady state communities seeded from consecutive fecal

81

samples. Because biological replicates require consecutive testing of the same donor to account

for sources of error (2), we compared consecutive fecal donations from healthy donors (Donors

A and B) over a 5-6 month period. Within-donor comparison showed that steady state chemostat

communities had high within-run reproducibility, but lower between-run repeatability when

using consecutive fecal donations. Temporal shifts of the fecal microbiota between donations

may be responsible for the inconsistencies in the reproducibility of steady state chemostat

communities (2). In addition, differences between consecutive fecal inocula, combined with the

in vivo to in vitro transition seen upon inoculation of a chemostat, may result in the development

of alternative equilibrium or steady states, as described by the concept of stability landscapes

(62). While this means that direct comparison of results between runs is challenging, high

within-run, between-vessel reproducibility indicates that multiple vessels can be run in parallel

and maintain similar communities that are ideal for experimentation. An alternative approach

would be to inoculate consecutive chemostat runs with a frozen fecal inoculum (214). While this

may increase the repeatability between runs, it could potentially decrease the diversity of the

resulting chemostat cultures. Longer storages times of frozen feces between different runs have

been shown to alter the profiles of communities developed within an in vitro model (214).

Another way to evaluate biological replicates in a chemostat system is to culture fecal

communities from multiple donors (2). Communities developed in our model showed similar

patterns of community development regardless of the fecal donor. Chemostat runs seeded with

feces from each donor resulted in reproducible, stable, and diverse steady state communities.

However, these communities all showed varying degrees of compositional changes as they

transitioned from an in vivo to an in vitro environment, with some steady state communities more

similar to the fecal inoculum than others. Donor-specific variations in the diet, transit time, pH,

82

and other physiochemical conditions mean that in vitro systems may more closely mimic the gut

of one individual over another.

Another aim of the between-donor comparisons was to determine whether our system was

able to culture steady state communities that retained unique community profiles (and did not

develop a 'generic' steady state composition). Our data indicates that steady state communities

generally became more similar to each other on day 36 relative to the fecal inoculum, however

donor-specific identities were maintained.

The final alternative to biological replicates is to examine chemostat runs inoculated by

pooling feces from multiple donors. However, this practice may lead to microbe-microbe

interactions not normally seen within the gut microbiota of a single healthy individual, resulting

in communities that may not be representative of a normal human gut ecosystem (2). Future

work will determine how community development is affected by the pooling of fecal

communities in our in vitro gut model.

In conclusion, our study shows that twin-vessel, single stage chemostats can be inoculated

with identical fresh fecal samples from a healthy donor to obtain and maintain stable, diverse,

and reproducible communities that reach steady state by 36-days post-inoculation in all five

chemostat runs analyzed. We note that steady state was achieved earlier in some runs compared

to others, and that, as expected, it was possible to discern a higher level of community similarity

using consecutive donations from the same donor than from different donors. However, we have

determined that care must be taken when using consecutive fecal samples from the same donor

to inoculate successive runs as between-run repeatability may be compromised. In general,

HITChip analysis is a more sensitive technique than DGGE that can provide compositional data

on fecal and chemostat samples. Moreover, we found that vessel correlation coefficients (%SI

83

values) were higher when samples were compared by HITChip analyses than when compared by

DGGE (discussed in more detail in chapter 8). Nevertheless, DGGE is a useful, rapid, low-cost

alternative to the HITChip that can be used to target key samples for further analysis.

A well-characterized distal gut model allows the exploration of the etiology of disease-

associated distal gut microbiota in the absence of host involvement. There is further potential to

facilitate the development of simulated mucosal gut communities in this model using

biologically significant scaffolds. Given the emerging importance of the role of the gut

microbiota in the maintenance of health, as well as increasing evidence that gut microbial

dysbiosis is associated with several disease states (18), modeling the gut ecosystem in vitro

represents a valuable experimental technique. In this work we have attempted to define the

limitations of the technique, as well as to recommend windows of appropriate experimentation.

84

Chapter 4

Characterizing perturbation in a

simulated distal gut ecosystem: Effect of

norepinephrine and clindamycin

treatment

4.1 Introduction

Ecosystem stability is thought to be reliant on its complexity, with higher stability found in

more complex ecosystems (215). High diversity results in redundancy, allowing for

compensatory growth following perturbation where the growth of one species increases in

response to the loss of another species in the same functional group (216). Because chemostat

communities are less diverse compared to their fecal inocula (as outlined in Chapter 3), we

wanted to investigate the effect of perturbation on these communities in an effort to further

validate our model.

Microbial communities can be disrupted by several factors, including antibiotics, dietary

changes, microbial infections, surgery or drug treatments (217-219). Several studies in humans

and mice have shown that broad-spectrum antibiotic therapy causes rapid and pronounced

alterations in the composition and function of the gut microbiota (220). Following the cessation

of antibiotic therapy gut communities experience partial taxonomic recovery, arriving at a new,

stable equilibrium (156).

Antibiotic perturbation studies are relatively straightforward as the biological activity of

antibiotics is limited to bacterial members of the gut microbiota. However, gut microbiota

85

perturbation studies are often confounded by host involvement, as it is not always clear whether

the biological activity of the stimulus is directed at the gut microbiota or at host physiological

processes. For example, when an individual is subjected to psychological or physical stress, the

sympathetic and enteric nervous systems respond by releasing catecholamines (such as

norepinephrine, epinephrine, and dopamine), which spill over into the gut lumen (221, 222).

Many people insist that stress can worsen gastrointestinal symptoms, and IBD patients believe

stress has an influence on the onset and relapse of illness (223, 224). However, in vivo studies are

limited as they cannot directly study perturbations following stress responses or catecholamine

addition in isolation of their effects on host physiology. Pure culture studies demonstrate that

some species of pathogens can increase their growth, chemotaxis, motility, adherence to

epithelial cells, and virulence gene expression in response to catecholamines, with

norepinephrine (NE) consistently causing the greatest increase in growth (225-230). However,

pure culture studies fail to replicate the interactions between different members of the gut

microbiota, which is more relevant to the in vivo situation. To date there have not been any

studies investigating the direct effects of catecholamines on gut communities maintained outside

of the host.

Therefore, we used DGGE as a molecular fingerprinting tool to characterize the effect of

defined (treatment with clindamycin) and undefined (treatment with NE) perturbation on the

structure of simulated gut communities developed in our chemostat model. The aims of this

study were to evaluate the effects of (i) a single dose of NE; (ii) sustained NE dosing; and (iii)

sustained clindamycin dosing on steady state chemostat communities.

86

4.2 Results

Four separate chemostat runs were analyzed in this study: twin-vessels seeded with donor

B feces and perturbed with a single dose of NE (VB-N1 and VB-N2), twin-vessels seeded with

donor C feces and perturbed with a single dose of NE (VC-N1 and VC-N2), twin-vessels seeded

with donor A feces and perturbed with a sustained dose of NE (VA-N1 and VA-N2), and twin-

vessels seeded with donor A feces and perturbed with a sustained dose of clindamycin (VA-C1

and VA-C2). VB-N1, VB-N2, VC-N1, VC-N2, VA-C1 and VA-C2 perturbation was initiated on

day 36 and analyzed until day 48. VA-N1 and VA-N2 perturbation was initiated on day 48 and

analyzed until day 60.

4.2.1 Perturbation of chemostat communities using a single dose of norepinephrine

Twin-vessels from two different chemostat runs were analyzed by DGGE to determine

whether a single dose of NE exposure affected the diversity and stability of simulated distal gut

communities. Paired time-points from VB-N1 (+NE) and VB-N2 (+H2O), which were seeded

with donor B feces, were run on the first series of DGGE gels until 48-days post-inoculation

(Figure 4.1a). Paired time-points from VC-N1 (+NE) and VC-N2 (+H2O), seeded with donor C

feces, were run on the second series of DGGE gels until 48 days post-inoculation (Figure 4.1b).

VB-N1 and VB-N2 DGGE correlation coefficients remained above their gel-defined cut-

off thresholds between days 30-36 (Figure 4.2a) while VC-N1 and VC-N2 correlation

coefficients remained above their gel-defined cut-off thresholds between days 0-36 (Figure

4.2b), suggesting that steady state was achieved prior to NE addition (36-days post-inoculation).

Following NE addition, VB-N1 and VB-N2 correlation coefficients were above the cut-off

between days 38-40, within the 5% cut-off on day 42, and rose above the cut-off between days

87

Figure 4.1 DGGE profiles illustrating within-vessel comparisons of chemostat communities

seeded with feces from donor B (VB-N1 and VB-N2) or donor C (VC-N1 and VC-N2)

immediately after inoculation (day 0), during steady state conditions (day 36) and following a

single dose of NE (days 38, 40, 42).

VC

-N2

Day

36

VC

-N2

Day

38

VC

-N2

Day

40

VC

-N2

Day

42

VC

-N2

Day

0

VC

-N1

Day

0V

C-N

1 D

ay 3

6V

C-N

1 D

ay 3

8V

C-N

1 D

ay 4

0

VC

-N1

Day

42

VB

-N1

Day

0V

B-N

1 D

ay

36

VB

-N1

Day

38

VB

-N1

Day

40

VB

-N1

Day

42

VB

-N2

Day

36

VB

-N2

Day

38

VB

-N2

Day

40

VB

-N2

Day

42

VB

-N2

Day

0

a b

88

Figure 4.2 DGGE analysis assessing the effect of a single dose of NE on chemostat communities

seeded with feces from donor B (VB-N1 and VB-N2, panels on the left) or donor C (VC-N1 and

VC-N2, panels on the right). Panels a and b) Correlation coefficients (expressed as percentages)

comparing the profiles of each vessel at identical time points. Panels c and d) Community

dynamics calculated using moving window correlation analysis. Panels e and f) Adjusted

Shannon Diversity Index (H’) plots. Panels g and h) Adjusted Shannon equitability index (EH’)

plots. Panels i and j) Adjusted band richness (S) plots.

89

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36 40 44 48

%S

I

Day

VB-N1 vs. VB-N2 Added NE Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36 40 44 48

%S

I

Day

VC-N1 vs. VC-N2 Added NE Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

2 6 10 14 18 22 26 30 34 38 42 46

%S

I

Day (x-2) vs. (x)

VB-N1 VB-N2 Added NE Cut-off Cut-off +5%

0

10

20

30

40

50

60

70

80

90

100

2 6 10 14 18 22 26 30 34 38 42 46

%S

I Day (x-2) vs. (x)

VC-N1 VC-N2 Added NE Cut-off Cut-off +5%

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36 40 44 48

H'

Day

VB-N1 VB-N2 Added NE

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36 40 44 48

H'

Day

VC-N1 VC-N2 Added NE

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36 40 44 48

EH

'

Day

VB-N1 VB-N2 Added NE

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36 40 44 48

EH

'

Day

VC-N1 VC-N2 Added NE

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36 40 44 48

S

Day

VB-N1 VB-N2 Added NE

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36 40 44 48

S

Day

VC-N1 VC-N2 Added NE

c

a b

d

e f

g h

i j

90

44-48 (Figure 4.2a). VC-N1 and VC-N2 correlation coefficients remained above their gel-

defined cut-off thresholds following NE addition (between days 36-48; Figure 4.2b). Table 4.1

lists the similarities between twin-vessel DGGE profiles on days 0, 36, 38, 40 and 42.

Clustering tree analysis showed that VB-N1 and VB-N2 profiles clustered with each other

immediately following inoculation (day 0) and during steady state conditions (days 34 and 36;

Figure 4.3a). Following NE addition, VB-N1 and VB-N2 samples clustered together with the

pre-NE samples (days 38-48) with the exception of VB-N1 day 42, which grouped farther from

the rest of the samples. VC-N1 and VC-N2 profiles clustered with each other immediately

following inoculation (day 0) and during steady state conditions (days 34 and 36; Figure 4.3b).

Following NE addition, VC-N1 and VC-N2 samples (days 38-48) clustered with each other and

with pre-NE samples.

VB-N1 and VB-N2 had reproducible rate-of-change values below the gel-defined cut-offs

between days 30-48 with the exception of VB-N1 between days 40-42, which was within 5% of

the cut-off between (VB-N1 Δt = 11.6%, VB-N2 Δt = 5.9; Figure 4.2c). VC-N1 and VC-N2 had

reproducible rate-of-change values that remained below the gel-defined cut-off between days 32-

48 (Figure 4.2d).

Regardless of the donor, H’, EH’, and S values were reproducible and stable in NE-treated

and control vessels prior to and following NE addition (Figure 4.2e-j, Table 4.2).

4.2.2 Effects of sustained norepinephrine perturbation on gut chemostat communities

Because a single dose of NE did not appear to cause substantial perturbation to chemostat

communities, we decided to assess the reproducibility, stability, and diversity of chemostat

communities exposed to a 5-day course of NE. Paired time-points from VA-N1 (+NE) and

91

Table 4.1 Correlation coefficients (%SI) comparing chemostat communities seeded with feces

from donor B (VB-N1 and VB-N2) or donor C (VC-N1 and VC-N2) immediately after

inoculation (day 0), during steady state conditions (day 36) and following a single dose of NE

(days 38, 40, 42).

Day VB-N1 vs. VB-N2 VC-N1 vs. VC-N2

0 95.7 88.4

36 91.6 97.3

38 93.9 97.4

40 92.9 93.5

42 85.5 96.5

92

Figure 4.3 Dendrograms comparing the effect of a single dose of NE on chemostat communities

seeded with feces from a) donor B (VB-N1 and VB-N2) or b) donor C (VC-N1 and VC-N2).

Dendrograms are based on Pearson and UPGMA correlation of DGGE profiles.

93

94

Table 4.2 Ecological parameters (diversity, evenness, and richness) for chemostat communities seeded with feces from donor B

(VB-N1 and VB-N2) or donor C (VC-N1 and VC-N2) immediately after inoculation (day 0), during steady state conditions (day

36) and following a single dose of NE (days 38, 40, 42).

Day

Shannon Diversity Index

(H')

Shannon Equitability Index

(EH')

Band Richness

(S)

VB-N1

(+NE)

VB-N2

(-NE)

VC-N1

(+NE)

VC-N2

(-NE)

VB-N1

(+NE)

VB-N2

(-NE)

VC-N1

(+NE)

VC-N2

(-NE)

VB-N1

(+NE)

VB-N2

(-NE)

VC-N1

(+NE)

VC-N2

(-NE)

0 3.32 3.34 3.24 3.27 0.80 0.81 0.81 0.82 62 62 56 55

36 3.39 3.36 3.09 3.00 0.79 0.78 0.79 0.77 72 72 51 51

38 3.37 3.36 3.05 3.04 0.79 0.79 0.78 0.77 72 72 51 51

40 3.40 3.41 3.00 3.01 0.80 0.80 0.77 0.77 72 72 51 51

42 3.33 3.42 2.96 2.98 0.79 0.81 0.76 0.76 72 72 50 51

95

VA-N2 (+H2O), seeded with donor A feces, were run on the third series of DGGE gels until 60

days post-inoculation (Figure 4.4a).

Prior to NE addition, VA-N1 and VA-N2 DGGE correlation coefficients were not within

5% of the gel-defined cut-off thresholds (7.6% from cut-off on day 48; Figure 4.5a). VA-N1 and

VA-N2 were 97.9% similar on day 0 and 83.7% similar on day 48 (Table 4.3). However,

HITChip data obtained for VA-N1 and VA-N2 day 48 samples indicate that they are 93.9%

similar (these samples are referred to as VA1-1 and VA1-2 day 48 samples in chapter 3). VA-N1

and VA-N2 DGGE correlation coefficients increased to within 5% of the gel-defined cut-off

threshold during the NE addition period (days 48-53). Following the NE addition period the VA-

N1 and VA-N2 correlation coefficients hovered within 0-6.5% of the gel-defined cut-off

threshold. VA-N1 and VA-N2 were 92.7% similar on day 54 and 91.2% similar on day 60

(Table 4.3).

Clustering tree analysis showed that VA-N1 and VA-N2 profiles clustered with each other

immediately following inoculation (day 0; Figure 4.6a). VA-N2 day 48 and VA-N2 day 54

clustered with each other, while VA-N1 day 48 clusters farther out from VA-N2 day 48 and VA-

N2 day 54. VA-N1 day 54 clusters farther out from VA-N2 day 48 and VA-N2 day 54, but

closer than VA-N1 day 48. While VA-N1 and VA-N2 samples clustered together on day 60, they

clustered farther out from day 48 and 54 samples.

VA-N1 and VA-N2 had reproducible rate-of-change values between days 0-60 (Figure

4.5c). VA-N1 rate-of-change values showed a peak of change between days 52-54, however the

Δt value was within 5% of the gel-defined cut-off (VA-N1 Δt = 8.4%, VA-N2 Δt = 5.0%). With

the exception of VA-N1 days 52-54, VA-N1 and VA-N2 rate-of-change values remained below

the gel-defined cut-off until days 58-60 (average VA-N1 Δt = 6.4% between days 48-58, average

96

Figure 4.4 DGGE profiles illustrating within-vessel comparisons of chemostat communities

seeded with feces from donor A a) following a course of NE (VA-N1 and VA-N2) or b)

following a course of clindamycin (VA-C1 and VA-C2).

VA

-N1 D

ay 0

VA

-N1 D

ay 4

8V

A-N

1 D

ay 5

4V

A-N

1 D

ay 6

0

VA

-N2

Day

0V

A-N

2 D

ay 4

8V

A-N

2 D

ay 5

4V

A-N

2 D

ay 6

0

VA

-C2 D

ay 0

VA

-C1

Day

0

VA

-C2 D

ay 3

6

VA

-C1 D

ay 3

6

VA

-C2

Day

42

VA

-C1 D

ay 4

2

VA

-C2

Day

48

VA

-C1 D

ay 4

8

a b

97

Figure 4.5 DGGE analysis assessing the effect of a course of NE (VA-N1 and VA-N2, panels on

the left) or a course of clindamycin (VA-C1 and VA-C2, panels on the right) on chemostat

communities seeded with feces from donor A. Panels a and b) Correlation coefficients

(expressed as percentages) comparing the profiles of each vessel at identical time points. Panels

c and d) Community dynamics calculated using moving window correlation analysis. Panels e

and f) Adjusted Shannon Diversity Index (H’) plots. Panels g and h) Adjusted Shannon

equitability index (EH’) plots. Panels i and j) Adjusted band richness (S) plots. C = clindamycin.

98

0

10

20

30

40

50

60

70

80

90

100

0 6 12 18 24 30 36 42 48 54 60

%S

I

Day

VA-N1 vs. VA-N2 Began adding NE Finished Adding NE Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36 40 44 48

%S

I

Day

VA-C1 vs. VA-C2 Began adding C Finished adding C Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

0 6 12 18 24 30 36 42 48 54 60

%S

I

Day (x-2) vs. (x)

VA-N1 VA-N2 Began adding NE Finished adding NE Cut-off Cut-off +5%

0

10

20

30

40

50

60

70

80

90

100

2 6 10 14 18 22 26 30 34 38 42 46

%S

I Day (x-2) vs. (x)

VA-C1 VA-C2 Began adding C Finished adding C Cut-off Cut-off + 5%

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 6 12 18 24 30 36 42 48 54 60

H'

Day

VA-N1 VA-N2 Began adding NE Finished adding NE

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36 40 44 48

H'

Day

VA-C1 VA-C2 Began adding C Finished adding C

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 6 12 18 24 30 36 42 48 54 60

EH

'

Day

VA-N1 VA-N2 Began adding NE Finished adding NE

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36 40 44 48

EH

'

Day

VA-C1 VA-C2 Began adding C Finished adding C

0

10

20

30

40

50

60

70

0 6 12 18 24 30 36 42 48 54 60

S

Day

VA-N1 VA-N2 Began adding NE Finished adding NE

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36 40 44 48

S

Day

VA-C1 VA-C2 Began adding C Finished adding C

a b

c d

e f

g h

i j

99

Table 4.3 Correlation coefficients (%SI) comparing chemostat communities seeded with feces

from donor A following exposure to a course of NE (VA-N1 and VA-N2) or clindamycin (VA-

C1 and VA-C2). Samples are compared immediately following inoculation (day 0), during

steady state conditions (VA-N1/2 = day 48, VA-C1/2 =day 36), 1 day following perturbation

(VA-N1/2= day 54, VA-C1/2= day 42) , and 7 days following perturbation (VA-N1/2= day 60,

VA-C1/2= day 48). C = clindamycin.

Time point VA-N1 (+NE) vs. VA-N2 (-NE) VA-C1 (+C) vs. VA-C2 (-C)

Day 0 97.9 90.0

Immediately before perturbation 83.7 91.3

1 day after end of perturbation 92.7 28.2

7 days after end of perturbation 91.2 80.5

100

Figure 4.6 Dendrograms comparing the effects of a) a course of NE (VA-N1 and VA-N2) or b) a

course of clindamycin (VA-C1 and VA-C2) on chemostat communities seeded with feces from

donor A. Dendrograms are based on Pearson and UPGMA correlation of DGGE profiles.

101

102

VA-N2 Δt = 4.5% between days 48-58; Figure 4.5c).

VA-N1 and VA-N2 H’, EH’, and S values were also reproducible and stable between days

34-48 (Figure 4.5e,g,i, Table 4.4).

4.2.3 Effects of sustained clindamycin perturbation on gut chemostat communities

Because a single dose of NE and a 5 day course of NE did not appear to cause substantial

perturbation to chemostat communities, we also decided to assess the reproducibility, stability,

and diversity of chemostat communities exposed to a 5 day course of a compound with known

effects (clindamycin). Clindamycin is an antibiotic that is active against most anaerobic bacteria

and has been associated with the greatest risk of C. difficile infection (231, 232). Paired time-

points from VA-C1 (+ clindamycin) and VA-C2 (+H2O), seeded with donor A feces, were run

on the third series of DGGE gels until 48 days post-inoculation (Figure 4.4b).

Prior to clindamycin addition, VA-C1 and VA-C2 DGGE correlation coefficients remained

within 5% of their gel-defined cut-off thresholds between days 32-34 and was above the gel-

defined cut-off on day 36 (Figure 4.5b). VA-C1 and VA-C2 were 90.0% similar on day 0 and

91.3% similar on day 36 (Table 4.3). Clustering tree analysis showed that VA-C1 and VA-C2

profiles clustered with each other immediately following inoculation (day 0) and during steady

state conditions (day 36; Figure 4.6b). Clustering tree analysis also showed that VA-C2 samples

from days 36, 42, and 48 clustered together.

VA-C1 and VA-C2 correlation coefficients decreased during the clindamycin treatment

period and increased once clindamycin addition was stopped (Figure 4.5b). On day 42 (one day

after the clindamycin treatment period), VA-C1 and VA-C2 were 28.2% similar (Table 4.3).

VA-C1 and VA-C2 correlation coefficients did not recover to pre-clindamycin levels by the end

103

Table 4.4 Ecological parameters (diversity, evenness, and richness) for chemostat communities seeded with feces from donor A

following exposure to a course of NE (VA-N1 and VA-N2) or clindamycin (VA-C1 and VA-C2). Samples are compared

immediately following inoculation (day 0), during steady state conditions (VA-N1/2 = day 48, VA-C1/2 =d ay 36), 1 day

following perturbation (VA-N1/2= day 54, VA-C1/2= day 42) , and 7 days following perturbation (VA-N1/2= day 60, VA-C1/2=

day 48).

Time point

Shannon Diversity Index

(H')

Shannon Equitability Index

(EH')

Band Richness

(S)

VA-N1

(+NE)

VA-N2

(-NE)

VA-C1

(+C)

VA-C2

(-C)

VA-N1

(+NE)

VA-N2

(-NE)

VA-C1

(+C)

VA-C2

(-C)

VA-N1

(+NE)

VA-N2

(-NE)

VA-C1

(+C)

VA-C2

(-C)

0 3.42 3.41 3.49 3.47 0.83 0.83 0.85 0.84 61 61 61 61

Immediately before

perturbation 3.26 3.30 3.42 3.49 0.78 0.79 0.82 0.84 64 65 65 65

1 day after end

of perturbation 3.28 3.29 2.58 3.44 0.78 0.79 0.74 0.82 65 65 33 65

7 days after end

of perturbation 3.29 3.25 3.21 3.36 0.79 0.78 0.84 0.80 66 66 46 65

104

of the experiment (Figure 4.5b). On day 48, VA-C1 and VA-C2 were 80.5% similar (Table 4.3).

VA-C1 day 42 clustered separately from other untreated samples (VA-C1 day 36 and VA-

C2 days 36, 41, and 48; Figure 4.6b). VA-C1 day 48 clustered closer to the untreated samples,

but still farther out relative to other untreated samples.

VA-C1 and VA-C2 had reproducible rate-of-change values between days 0-36 (Figure

4.5d). Higher rate-of-change values were noted at the beginning of the experiment during the

transition period (days 0-14). However, the rate-of-change values decreased as the community

developed (days 16-32) and stabilized as steady state was reached (days 34-36). VA-C1 rate-of-

change values showed a peak of change between days 40-42 (VA-C1 Δt= 59.3%, VA-C2 Δt =

17.6%). After day 36, VA-C2 (control vessel) communities maintained low rate-of-change

values (average Δt = 9.2% between days 34-46).

VA-C1 and VA-C2 H’, EH’, and S values were also reproducible between days 16-36 and

stable between days 34-36 (Figure 4.5f,h,j, Table 4.4). VA-C1 H’, EH’, and S values decreased

during the clindamycin treatment period and began to increase once clindamycin addition was

stopped. While VA-C1 H’ and EH’ values recovered to VA-C1 day 36 (pre-clindamycin) and the

corresponding VA-C2 (control) values by the end of the experiment, S values did not fully

recover. VA-C2 H’, EH’ and S remained stable between days 34-48.

4.3 Discussion

We compared the effects of short term (1 dose) and long term (5 day course) NE exposure

on steady state chemostat communities simulating the human distal gut. In both experiments, NE

did not appear to have a substantial effect on the structure of chemostat communities.

Clindamycin treatment allowed us to characterize the effect of perturbation within our system

105

using a stimulus that is known to alter the structure of gut communities both in vivo and in vitro

(and will be discussed in detail in Chapter 5). Below, we suggest and discuss several reasons

why no major differences were seen in community structure following NE exposure.

In the first instance, the literature does not clearly define NE concentrations found in the

gut of individuals experiencing heightened stress responses. However, Cogan et al. (2007) used

100 µM NE to test the effects of NE on Campylobacter jejuni virulence (194). Hedge et al.

(2009) used 50 µM NE (baseline NE levels) and 500 µM NE ("supraphysiological" NE levels) to

test the effect of NE on Pseudomonas aeruginosa virulence and gene expression in the gut (233).

Naresh and Hampson (2011) used 50, 100, 500 and 1000 µM NE to test the effects of NE on the

activity of the intestinal intestinal spirochaete Brachyspira pilosicoli (234). Peterson et al.(2011)

used 0, 5, 50, 100, or 2000 µM NE to investigate conjugative gene transfer between enteric

bacteria (195). While the NE concentration used in our study (100 µM) was in line with the

physiological concentration used in previous studies, future studies could use

"supraphysiological" NE concentrations (i.e. 500 µM) to ensure low NE concentrations were not

responsible for the perceived lack of NE-driven perturbation within our system.

A second issue was that chemostat communities analyzed in this study were separately

seeded with feces from healthy donors (donors A, B, and C). Since these donors were healthy,

their gut microbiotas may have been particularly resilient and thus less susceptible to NE-

induced gut dysbiosis relative to unhealthy gut communities (215). Microbial imbalances,

reduced diversity, and general instability of microbial communities associated with IBD patients

may instead result in increased susceptibility to stress-induced perturbation (235, 236).

Therefore, future studies will concentrate on investigating the effects of NE exposure on

chemostat communities seeded with feces from IBD patients.

106

A third explanation for the perceived lack of NE activity could be due to a lack of host

physiological conditions in our in vitro system. Catecholamines are involved in host

physiological processes, including gastrointestinal motility, transepithelial ion transport, and

immune suppression (237, 238). An alteration in these physiological processes following

exposure to NE in the host has the potential to indirectly alter the composition and function of

the gut microbiota. Our in vitro model is not capable of mimicking these physiological

processes; therefore subtle changes to the gut microbiota driven by these effects will be missed

in a chemostat.

Tsavkelova et al. (2000) reported that some microorganisms can produce NE in an in vitro

system (239). Some species of bacteria have been shown to produce transcripts that share some

similarity to the tryosine hydroxylase enzyme found in mammals (a rate-limiting enzyme in the

NE synthesis pathway; 240, 241). We did not measure NE concentrations or putative tryosine

hydroxylase transcript levels potentially present in our chemostat vessels prior to, during, or

following the NE treatment periods, and thus a fourth reason for our findings is that such

enzymatic activity may have compensated for the increase in NE in our system.

Iron availability is a further factor that may have influenced the results of this work. Pure

culture studies have shown that NE can indirectly increase the growth of members of the gut

microbiota by facilitating bacterial access to host sequestered iron (242). In many pathogens iron

is essential, and iron restriction makes bacterial growth difficult (242). In the intestinal mucosa,

iron is bound by high affinity iron binding proteins (for example, lactoferrin and transferrin;

243). To scavenge iron from the host environment, pathogenic and commensal bacteria secrete

siderophores – low molecular weight catecholate or hydroxamate molecules that capture iron and

present it to bacterial cell surface receptor proteins (194, 242, 244). However, siderophores are

107

not always effective at removing iron from host high affinity iron binding proteins (242). It has

been shown that NE, epinephrine, and dopamine help to mediate removal of sequestered iron

from host iron binding proteins, and this has been confirmed in several studies by Freestone et al.

(243, 245-248). Increases in NE during stress could cause the iron bound to lactoferrin or

transferrin in the gut to become more available to the gut microbiota, causing an increase in the

growth of certain bacteria (245, 249, 250). We did not investigate the effect on iron acquisition

by the microbiota within our model, as we first wanted to establish whether there were any direct

effects of NE on microbiota members. To further determine the role of host–derived iron-binding

proteins on the microbiota would require the addition of large amounts of iron-saturated

lactoferrin or transferrin to chemostat cultures, the cost of which would be prohibitive (251).

NE exposure has been proposed to upregulate the expression of virulence factors in

addition to its known effects on growth (252). In enterohaemorrhagic E. coli (EHEC), NE has

been shown to increase chemotaxis, motility, adherence to epithelial cells and the expression of

virulence genes (253). It has also been shown to increase the internalization of EHEC and

Salmonella enterica serovar Choleraesuis in epithelial cells of the jejunal mucosa (254, 255).

Enterotoxigenic E. coli (ETEC) and EHEC strains showed increases in the expression of the K99

pilus adhesin, in response to NE exposure; this adhesin is responsible for adhesion to and

penetration of host tissues (83, 256, 257). In addition, commensal E. coli demonstrated increased

expression of type 1 pili when exposed to NE (258). As well as adhesion molecule elaboration in

response to NE, EHEC also showed an increase in Shiga-like toxin-I and –II (256). When taken

together this evidence suggests that there might have been changes in the expression of virulence

factors by certain microbiota members, but such changes in expression may not have had a

measurable impact on community structure and were therefore undetectable by DGGE. Future

108

studies should monitor virulence gene expression in this model, focusing on virulence factors,

such as those mentioned above, that have been shown to be upregulated following exposure to

NE in pure culture studies.

In conclusion, while sustained clindamycin dosing caused large changes in the structure of

chemostat communities (as expected), a single dose of NE or sustained NE dosing did not appear

to have a substantial effects on the structure of chemostat communities. As mentioned

previously, in vitro gut fermentation models provide several advantages over in vivo models and

other in vitro models for studying the effects of perturbation on the gut microbiota. While the

lack of host physiological processes can be viewed as a drawback of in vitro models, a key

benefit is the ability to use such models to determine if a stimulus directly effects gut ecosystem

composition or functionality. However, before we can make conclusions regarding the potential

ineffectiveness of a compound on communities developed within our model, we must first utilize

analysis techniques that do not solely rely on changes in community structure (DNA).

Improvements in technology are currently reaching a point whereby the levels of transcription of

genes within complex communities can be determined, and the application of such

transcriptomic approaches will allow a more detailed study of system perturbation where effects

on community structure may not be a measurable outcome. In addition to identifying the

functions, microbial interactions, and core species important to gut microbiota, defining the

response of gut communities to perturbation may allow us to predict, prevent, or treat future

imbalances or restore diseased microbiota to a healthy state (62).

109

Chapter 5

Modeling simulated mucosal and luminal

communities using packed-column

biofilm reactors and an in vitro distal gut

chemostat model

5.1 Introduction

The composition of the gut microbiota varies depending on the segment of the GI tract

being sampled, with the most diverse community found in the distal gut (5, 8, 161). In addition

to heterogeneity along the length of the GI tract, gut communities also vary within an intestinal

segment (161). Previous literature has characterized distinct communities harboured in the gut

lumen (intestinal material or feces) and mucosa (biopsy; 259, 260).

The intestinal epithelium is separated from the gut lumen by a thick mucus layer (161).

Community composition within this layer depends on a variety of factors, including the ability of

microorganisms to bind to, degrade and gain nutrients from mucins, as well as their ability to

resist the actions of host antimicrobial peptides and to tolerate the relatively increased oxygen

levels in this niche (261).

Microbial associations with host mucins may allow microorganisms to persist within the

gut environment through the avoidance of washout as well as protection from the greater

disturbances encountered by their luminal counterparts (262). For example, Van den Abbeele et

al. (2012) found that biofilm communities cultured in a SHIME system with a simulated mucosal

environment (M-SHIME) were enriched in Firmicutes over Bacteroidetes and Proteobacteria

110

(263). This is in contrast to planktonic communities cultured from conventional gut fermentation

models, which are enriched in Bacteroidetes and Proteobacteria (165, 172, 184, 209, 264) over

Firmicutes (165, 209). In another study, Van den Abbeele et al. (2012) found that planktonic

populations of Lactobacillus mucosae recovered from antibiotic treatment more quickly when

growing in a M-SHIME unit than when growing in a conventional SHIME model lacking a

mucosal environment (L-SHIME; 262).

Mucosal communities are thought to be particularly important to host health due to their

proximity to the host epithelium (265). While human mucosal communities have been

characterized in vivo by analyzing microorganisms adherent to colonic biopsy specimens (259,

260), specimens collected via endoscopy are often limited to end-point analyses and studies may

be confounded by the use of prior bowel cleansing procedures (260, 262). Thus, dynamic

monitoring of gut microbial communities is often conducted using fecal samples, which are not

representative of mucosal community composition (262).

In vitro models represent a useful alternative to in vivo biopsies for the study of mucosal

communities (262, 263, 266-269). However, such models would benefit from a more detailed

characterization of planktonic and biofilm reproducibility, diversity, and response to

perturbation. In addition, applying strategies to encourage physiologically relevant biofilm

growth could improve in vitro models (263).

In this study, we used packed-column biofilm reactors to incorporate a simulated mucosal

environment into our twin-vessel single-stage chemostat model of the human distal gut. Our aims

were (i) to determine whether planktonic and biofilm communities were reproducible and stable,

(ii) to determine whether planktonic and biofilm communities had distinct microbial

111

compositions, and (iii) to determine whether planktonic and biofilm communities responded to

perturbation in a biologically relevant manner.

5.2 Results

In this study, planktonic samples from V1 and V2 (vessels seeded with donor D feces) are

referred to as V1-P (added clindamycin between days 36-41) and V2-P (added water between

days 36-41). Biofilm samples cultured from V1 or V2 between days 0-36 are referred to as V1-

B1 or V2-B1 (not exposed to clindamycin). Biofilm samples cultured from V1 or V2 between

days 0-41 are referred to as V1-B2 (exposed to clindamycin between days 36-41) or V2-B2

(exposed to water between days 36-41). Biofilm samples cultured from V1 or V2 between days

36-48 are referred to as V1-B3 (exposed to clindamycin between days 36-41) or V2-B3 (exposed

to water between days 36-41). Biofilm samples cultured from V1 or V2 between days 41-48 are

referred to as V1-B4 (growth initiated following the clindamycin treatment period) or V2-B4

(growth initiated following the water addition period).

5.2.1 Planktonic communities developed in mucosal twin-vessel single-stage chemostats were

reproducible and stable

V1 and V2 planktonic community reproducibility, stability, and diversity were compared

to determine whether the incorporation of a simulated mucosal environment interfered with

planktonic community development. V1-P and V2-P samples had reproducible DGGE profiles

between days 0-36 (Figures 5.1, 5.2a). On day 0, V1-P and V2-P were 96.3% similar by DGGE

(Table 5.1). On day 36, V1-P and V2-P were 96.7% similar by DGGE and 98.7% similar by

HITChip (Table 5.1). Vessel correlation coefficient values were above the gel-defined cut-off

112

Figure 5.1 DGGE analysis of the planktonic communities from two chemostat vessels modeling

the human distal gut prior to, during, and following a clindamycin treatment period. a)

Correlation coefficients (expressed as percentages) comparing the profiles of each vessel at

identical time points. b) Community dynamics calculated using moving window correlation

analysis. c) Adjusted Shannon Diversity Index (H’). d) Adjusted band richness (S). e) Adjusted

Shannon equitability index (EH’).

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36 40 44 48

%S

I

Day

V1-P vs. V2-P Began adding C Finished adding C Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36 40 44 48

%S

I

Day (x-2) vs (x)

V1-P V2-P Began adding C Finished adding C Cut-off Cut-off+5%

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36 40 44 48

H'

Day

V1-P V2-P Began adding C Finished adding C

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36 40 44 48 S

Day

V1-P V2-P Began adding C Finished adding C

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36 40 44 48

EH

'

Day

V1-P V2-P Began adding C Finished adding C

a b

c d

e

113

Figure 5.2 Molecular fingerprints representing planktonic (P) and biofilm (B) communities

cultured within a twin-vessel single-stage chemostat. (a) DGGE profiles (b) HITChip profiles.

HITChip profiles also list the highest phylogenetic level of specificity of probes (level 1) on the

right of panel b.

V1

-P D

ay 0

V2

-P D

ay 0

V1

-P D

ay 3

6

V2

-P D

ay 3

6

V1

-B1

Day

s 0

-36

V2

-B1

Day

s 0

-36

V2

-B2

Day

s 0

-41

V1

-B2

Day

s 0

-41

V2

-P D

ay 4

1

V1

-P D

ay 4

1

V1

-P D

ay 4

8

V2

-P D

ay 4

8

V1

-B3

Day

s 3

6-4

8

V2

-B3

Day

s 3

6-4

8

V1

-B4

Day

s 4

1-4

8

V2

-B4

Day

s 4

1-4

8

V1

/2-P

Day

0

V1

-P D

ay 3

6

V2

-P D

ay 3

6

V1

-B1

Day

s 0

-36

V2

-B1

Day

s 0

-36

V2

-B2

Day

s 0

-41

V1

-B2

Day

s 0

-41

V2

-P D

ay 4

1

V1

-P D

ay 4

1

V1

-P D

ay 4

8

V2

-P D

ay 4

8

V1

-B3

Day

s 3

6-4

8

V2

-B3

Day

s 3

6-4

8

V1

-B4

Day

s 4

1-4

8

V2

-B4

Day

s 4

1-4

8

a b

Actinobacteria

Asteroleplasma

Bacilli

Bacteroidetes

Clostridium cluster IClostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XIClostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XVClostridium cluster XVIClostridium cluster XVIIClostridium cluster XVIIICyanobacteriaFusobacteria

Proteobacteria

Spirochaetes

Uncultured Clostridiales

Uncultured MollicutesVerrucomicrobia

114

Table 5.1 Correlation coefficients (%SI) comparing twin-vessel planktonic communities and

twin-vessel biofilm communities by DGGE and HITChip. P=planktonic sample, B=biofilm

sample, NA=not available.

Sample %SI, by DGGE %SI, by HITChip

P Day 0 96.3 NA

P Day 36 96.7 98.7

P Day 41 48.2 81.4

P Day 48 75.8 87.1

B1 Days 0-36 98.7 99.0

B2 Days 0-41 87.2 97.1

B3 Days 36-48 80.6 90.7

B4 Days 41-48 75.4 83.9

115

thresholds immediately following inoculation until day 36 (Figure 5.1a). Similarities between

V2-P samples from days 36, 41, and 48 are shown in Table 5.2. HITChip analysis showed that

V1-P and V2-P had similar compositions on day 36 and V2-P had similar compositions on days

36, 41, and 48 (Figures 5.2b, 5.3, Table 5.3).

Clustering tree analysis showed that V1-P and V2-P profiles clustered with each other

immediately following inoculation (day 0) and during steady state conditions (day 36; Figure

5.4). Clustering tree analysis showed that V2 planktonic samples from days 36, 41, and 48

clustered together; however day 36 and 41 profiles clustered more closely with each other than

with the day 48 profile (Figure 5.4). Planktonic steady state profiles clustered together on PCA

and RDA plots (Figures 5.5, A5.1).

V1-P and V2-P had reproducible rate of change values between days 0-36 (Figure 5.1b).

Higher rate of change values were noted at the beginning of the experiment during the transition

period (days 0-16). However, the rate of change values decreased as the community developed

(days 18-32) and stabilized as steady state was reached (days 34-36; Figure 5.6). After day 36,

V2-P (control vessel) communities maintained low rate of change values (average Δt = 6.6%

between days 34-46) until days 46-48 (Δt = 22.3%; Figures 5.1b, 5.6).

H’, EH’, and S values were also reproducible in V1-P and V2-P samples between days 0-36

(Figure 5.1c,d,e, Table 5.4). Following the inoculation of both vessels, reproducible decreases in

diversity and evenness and a reproducible increase and decrease in band richness was observed.

Upon the establishment of steady state conditions, H’, EH’ and S values stabilized.

116

Table 5.2 Correlation coefficients (%SI) comparing within-vessel planktonic samples by DGGE

and HITChip.

Within-vessel

planktonic comparison

%SI, by DGGE %SI, by HITChip

V1 V2 V1 V2

Day 36 vs. Day 41 43.7 96.7 81.5 99.2

Day 36 vs. Day 48 54.8 82.0 89.2 97.3

Day 41 vs. Day 48 79.7 87.4 88.4 96.8

117

Figure 5.3 Relative abundance of higher taxonomic groups (~ phylum/class level) based on HITChip analysis for the fecal

inoculum, planktonic communities and biofilm communities.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

V1/2-P

inoculum

V1-P

Day 36

V2-P

Day 36

V1-P

Day 41

V2-P

Day 41

V1-P

Day 48

V2-P

Day 48

V1-B1 V2-B1 V1-B2 V2-B2 V1-B3 V2-B3 V1-B4 V2-B4

Rel

ati

ve

Co

ntr

ibu

tio

nActinobacteria

Bacteroidetes

Bacilli

Clostridium cluster I

Clostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XI

Clostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XV

Clostridium cluster XVI

Clostridium cluster XVII

Clostridium cluster XVIII

Uncultured Clostridiales

Cyanobacteria

Fusobacteria

Proteobacteria

Spirochaetes

Asteroleplasma

Uncultured Mollicutes

Verrucomicrobia

118

Table 5.3 Percent abundance of higher taxonomic groups (~ phylum/class level) based on HITChip analysis for the fecal

inoculum, planktonic communities and biofilm communities.

Higher taxonomic groups

Feces Planktonic Biofilm

V1/2-P

Day 0

V1-P

Day 36

V2-P

Day 36

V1-P

Day 41

V2-P

Day 41

V1-P

Day 48

V2-P

Day 48

V1-

B1

V2-

B1

V1-

B2

V2-

B2

V1-

B3

V2-

B3

V1-

B4

V2-

B4

Actinobacteria 0.45 0.36 0.52 0.73 0.44 0.50 0.54 1.50 1.08 2.52 1.86 0.56 0.50 0.54 0.43

Bacteroidetes 21.69 50.01 45.49 63.61 49.15 36.43 47.63 18.72 19.55 27.03 20.70 38.86 22.58 48.68 19.15

Firmicutes

Bacilli 0.79 0.65 0.94 1.23 0.77 0.87 0.95 0.79 0.69 1.01 0.87 0.91 0.79 0.93 0.74

Clostridium cluster I 0.18 0.20 0.29 0.39 0.24 0.26 0.30 0.29 0.32 0.40 0.35 0.28 0.26 0.29 0.24

Clostridium cluster III 0.15 0.09 0.14 0.13 0.14 0.11 0.13 0.14 0.25 0.22 0.26 0.19 0.43 0.12 0.37

Clostridium cluster IV 31.01 14.90 15.20 18.45 18.84 29.25 12.24 30.46 35.26 28.21 30.16 31.95 26.06 21.62 31.11

Clostridium cluster IX 1.26 0.31 0.46 0.53 0.38 0.39 0.46 0.46 0.39 0.54 0.52 0.43 0.46 0.41 0.46

Clostridium cluster XI 1.31 0.27 0.37 0.44 0.31 0.31 0.36 0.45 0.38 0.71 0.65 0.32 0.27 0.32 0.25

Clostridium cluster XIII 0.03 0.04 0.05 0.08 0.04 0.05 0.06 0.04 0.04 0.06 0.05 0.05 0.04 0.05 0.04

Clostridium cluster XIVa 39.26 28.86 30.83 9.53 25.06 27.66 22.93 36.01 32.32 29.63 32.62 21.05 30.84 13.92 29.97

Clostridium cluster XV 0.03 0.03 0.05 0.08 0.04 0.05 0.05 0.04 0.04 0.05 0.04 0.05 0.04 0.05 0.04

Clostridium cluster XVI 0.11 0.11 0.16 0.23 0.13 0.16 0.16 0.13 0.12 0.17 0.15 0.16 0.14 0.16 0.13

Clostridium cluster XVII 0.02 0.03 0.04 0.06 0.04 0.04 0.05 0.04 0.03 0.05 0.04 0.05 0.04 0.05 0.03

Clostridium cluster XVIII 0.07 0.08 0.11 0.47 0.09 0.10 0.11 0.10 0.07 0.14 0.10 0.14 0.09 0.15 0.08

Uncultured Clostridiales 1.67 0.79 1.06 1.01 0.71 0.75 10.19 1.45 1.17 1.68 1.41 1.22 1.53 0.73 1.79

Total Firmicutes 75.90 46.36 49.69 32.63 46.78 60.00 47.97 70.40 71.09 62.85 67.23 56.79 61.00 38.81 65.26

Cyanobacteria 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

Fusobacteria 0.05 0.07 0.10 0.14 0.08 0.09 0.10 0.08 0.07 0.10 0.09 0.10 0.08 0.10 0.07

Proteobacteria 0.69 1.41 1.93 2.49 1.68 1.79 1.86 5.37 5.34 3.17 4.35 2.94 11.30 11.35 8.45

Spirochaetes 0.01 0.01 0.02 0.03 0.02 0.02 0.02 0.02 0.01 0.02 0.02 0.02 0.02 0.02 0.02

Tenericutes Asteroleplasma 0.01 0.01 0.01 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

Uncultured Mollicutes 0.08 0.11 0.16 0.21 0.13 0.15 0.16 0.12 0.11 0.16 0.14 0.15 0.13 0.16 0.11

Verrucomicrobia 1.11 1.66 2.07 0.13 1.72 0.99 1.70 3.77 2.73 4.11 5.61 0.57 4.39 0.31 6.50

119

Figure 5.4 Dendrogram comparing planktonic (P) and biofilm (B) communities cultured in two

identical chemostat vessels. Dendrograms are based on (a) Pearson and UPGMA correlation

(DGGE profiles) or (b) Pearson correlation and Ward's minimum variance method (HITChip

profiles).

120

121

Figure 5.5 Principal component analysis (PCA) at the genus-like level based on HITChip data of the fecal inoculum, planktonic

communities, and biofilm communities. Red arrows indicate shifts in V1-P samples following clindamycin treatment, while green

arrows indicate shifts in V2-P samples (control) following water addition. Light grey circles indicate clustering of planktonic or

biofilm samples not exposed to clindamycin.

-1.0 1.5

-0.6

0.8

Bacteroidetes

V1/2 inoculum

V1-P Day 36

V2-P Day 36

V1-P Day 41

V2-P Day 41

V1-P Day 48

V2-P Day 48

V1-B1

V2-B1

V1-B2V2-B2 V1-B3

V2-B3

V1-B4

V2-B4

PC1: 47.2%

PC

2:

22.3

%

Clostridium cluster IV

122

Figure 5.6 DGGE profiles illustrating within-vessel comparisons of planktonic communities

from V1 and V2 immediately prior to the clindamycin treatment period (days 34 and 36), during

the clindamycin treatment period (days 38 and 40), and following the clindamycin treatment

period (recovery period, days 42, 44, 46, and 48).

123

V1

-P D

ay

0

V1

-P D

ay

34

V1

-P D

ay

36

V1

-P D

ay

38

V1

-P D

ay

40

V1

-P D

ay

42

V1

-P D

ay

44

V1

-P D

ay

46

V1

-P D

ay

48

V2

-P D

ay

0

V2

-P D

ay

34

V2

-P D

ay

36

V2

-P D

ay

38

V2

-P D

ay

40

V2

-P D

ay

42

V2

-P D

ay

44

V2

-P D

ay

46

V2

-P D

ay

48

Added

clindamycinAdded

H2O

124

Table 5.4 Ecological parameters (diversity, evenness, and richness) for the fecal inocula,

planktonic communities, and biofilm communities as measured using DGGE and the HITChip.

P=planktonic sample, B=biofilm sample, NA=not available.

Analysis Method Sample

Shannon Diversity

Index (H’)

Shannon Equitability

Index (EH’)

Richness

(S)

V1 V2 V1 V2 V1 V2

DGGE

P Day 0 3.45 3.47 0.84 0.84 61 61

P Day 36 3.11 3.19 0.75 0.77 64 64

P Day 41 2.89 3.24 0.74 0.78 49 64

P Day 48 2.82 3.17 0.73 0.76 49 64

B1 Days 0-36 3.08 3.12 0.75 0.76 61 61

B2 Days 0-41 3.07 3.09 0.74 0.75 62 61

B3 Days 36-48 2.96 2.96 0.74 0.72 54 61

B4 Days 41-48 2.74 2.97 0.69 0.72 52 61

HITChip

P Day 0 5.71 0.80 1105

P Day 36 5.13 5.18 0.76 0.78 644 552

P Day 41 4.68 5.08 0.72 0.76 415 580

P Day 48 5.08 5.13 0.78 0.77 480 533

B1 Days 0-36 5.60 5.38 0.80 0.78 736 743

B2 Days 0-41 5.40 5.42 0.78 0.78 689 687

B3 Days 36-48 5.09 5.33 0.76 0.78 582 636

B4 Days 41-48 4.84 5.40 0.75 0.79 469 665

125

5.2.2 Twin-vessel biofilm growth was reproducible both between twin vessels and within the

same vessel

V1-B1 and V2-B1 samples were harvested during steady state conditions (day 36) to

characterize between-vessel reproducibility of biofilm communities. Photographs of the biofilms

present in V1-B1 and V1-B2 packed-column reactors are shown in Figure 5.7. V1-B1 and V2-B1

were 98.7% similar by DGGE and 99.0% similar by HITChip (Table 5.1). HITChip analysis

showed that V1-B1 and V2-B1 had similar compositions on day 36 (Figures 5.2b, 5.3, Table

5.3). Clustering tree analysis and the PCA and RDA plots showed that V1-B1 and V2-B1

clustered together (Figures 5.2, 5.4, 5.5, A5.1). V1-B1 and V2-B1 also had reproducible H’, EH’,

and S values, as shown in Table 5.4.

V2 biofilm samples were compared to determine whether biofilm initiation, harvest, and

development times impacted the reproducibility of biofilms cultured from a single vessel.

Photographs of the biofilms present in the V2 packed-column reactors are shown in Figure 5.7.

Within-vessel biofilm formation was the most reproducible when biofilm initiation, harvest, and

development times were more similar, as V2-B1 clustered more closely with V2-B2 while V2-

B3 clustered more closely with V2-B4 (Figure 5.4, Table 5.5). Clustering tree analysis and the

PCA and RDA plots showed that V2-B1, V2-B2, V2-B3, and V2-B4 clustered together (Figures

5.2, 5.4, 5.5, A5.1, Table 5.5). HITChip analysis showed that V2 biofilms had similar

compositions (Figures 5.2b, 5.3, Table 5.3). In addition, H’ EH’, and S values were also the most

reproducible in these profiles (Table 5.4).

126

Figure 5.7 Photographs of columns attached to V1 and V2 on (a) day 36, (b) day 41, and (c) day

48.

127

Table 5.5 Correlation coefficients (%SI) comparing within-vessel biofilm samples (B) by DGGE

and HITChip. Values in italics indicate biofilm samples with more comparable initiation,

harvest, and development times, while values not in italics indicate biofilm samples with less

comparable initiation, harvest, and development times.

Within-vessel

biofilm comparison

%SI, by DGGE %SI, by HITChip

V1 V2 V1 V2

B1 vs. B2 80.2 96.6 97.3 99.1

B1 vs. B3 79.7 91.7 88.1 96.9

B1 vs. B4 69.4 87.7 81.0 96.6

B2 vs. B3 77.2 86.9 92.0 95.7

B2 vs. B4 57.4 85.3 85.3 95.5

B3 vs. B4 82.0 95.9 96.3 99.5

128

5.2.3 Planktonic and biofilm communities had distinct community compositions

Biofilm samples were compared to the planktonic communities present at the time of

harvest to determine whether planktonic and biofilm communities had distinct compositions.

HITChip analysis showed that, in the absence of clindamycin, biofilm samples were positively

associated with Firmicutes (especially Clostridium cluster IV), and negatively associated with

Bacteroidetes compared to the steady state planktonic communities (Figure 5.3, Tables 5.3, 5.6).

Table 5.7 lists the correlation coefficient values comparing biofilm samples to samples of the

planktonic community present at the time of biofilm harvest. Clustering tree analysis and the

PCA and RDA plots showed that planktonic and biofilm samples grouped separately from each

other in the absence of clindamycin treatment (Figures 5.4, 5.5, A5.1).

Reproducible differences were seen between V1-B1 and V2-B1 communities and their

corresponding day 36 planktonic samples (Figure 5.2, Table 5.7). These differences were

consistent over time, as comparable differences were seen betweenV2 biofilm samples and the

corresponding V2 planktonic samples (Figure 5.2, Table 5.7). While DGGE showed that H’, EH’,

and S values were slightly lower in biofilm samples compared to planktonic communities present

at time of harvest, HITChip showed that H’, EH’, and S values were slightly higher in biofilm

samples compared to the planktonic communities (Table 5.4).

5.2.4 Clindamycin treatment caused a shift in planktonic community composition

V1 and V2 planktonic communities were compared to characterize the impact of

biologically relevant doses of clindamycin on simulated luminal communities of the distal colon.

V1-P (+ clindamycin) and V2-P (+ H2O) similarity decreased during the clindamycin treatment

129

Table 5.6 Genus-like level bacterial groups significantly different (p= 0.02, q= 0.05) between

average planktonic communities and average biofilm communities for samples not exposed to

clindamycin treatment. P values were calculated using the Wilcoxon signed-rank test corrected

for multiple comparisons. The fold changes (biofilm/planktonic samples) shown in green are

>2.5 (higher in biofilm samples) and shown in red are <0.5 (lower in biofilm samples). SD =

standard deviation.

130

Level 1

(~ phylum/class level)

Level 2

(genus-like level)

Planktonic

abundance (%) SD

Biofilm

abundance (%) SD

Fold change

(biofilm/planktonic)

Bacteroidetes

Alistipes et rel. 1.14 0.20 0.63 0.06 0.55

Bacteroides fragilis et rel. 5.25 0.94 1.61 0.27 0.31

Bacteroides ovatus et rel. 7.52 1.52 1.93 0.41 0.26

Bacteroides plebeius et rel. 0.93 0.14 0.58 0.09 0.62

Bacteroides splachnicus et rel. 0.59 0.07 0.40 0.03 0.68

Bacteroides stercoris et rel. 0.92 0.17 0.52 0.07 0.57

Bacteroides vulgatus et rel. 19.71 2.62 7.85 1.27 0.40

Prevotella melaninogenica et rel. 0.66 0.26 0.27 0.04 0.41

Prevotella oralis et rel. 0.56 0.16 0.24 0.03 0.43

Prevotella ruminicola et rel. 0.06 0.01 0.05 0.00 0.73

Prevotella tannerae et rel. 0.50 0.04 0.26 0.02 0.52

Tannerella et rel. 0.54 0.06 0.34 0.03 0.63

Bacilli Streptococcus mitis et rel. 0.08 0.01 0.11 0.01 1.37

Clostridium cluster III Clostridium stercorarium et rel. 0.12 0.02 0.28 0.11 2.43

Clostridium cluster IV

Anaerotruncus colihominis et rel. 0.27 0.03 0.69 0.12 2.50

Clostridium cellulosi et rel. 0.84 0.18 2.49 0.30 2.96

Clostridium leptum et rel. 1.81 0.48 3.13 0.29 1.73

Clostridium orbiscindens et rel. 2.06 0.33 7.88 1.65 3.84

Oscillospira guillermondii et rel. 4.96 1.47 12.49 1.28 2.52

Ruminococcus bromii et rel. 2.79 0.61 0.51 0.06 0.18

Subdoligranulum variable at rel. 0.79 0.04 1.08 0.29 1.37

Clostridium cluster IX Dialister 0.12 0.02 0.19 0.02 1.62

Clostridium cluster XIVa

Anaerostipes caccae et rel. 0.29 0.03 0.42 0.07 1.43

Clostridium sphenoides et rel. 2.45 0.22 1.79 0.24 0.73

Clostridium symbiosum et rel. 19.28 3.11 25.67 1.87 1.33

Lachnobacillus bovis et rel. 0.28 0.08 0.16 0.02 0.56

Lachnospira pectinoschiza et rel. 0.22 0.03 0.33 0.03 1.46

Ruminococcus obeum et rel. 1.07 0.07 0.89 0.06 0.83

Proteobacteria

Burkholderia 0.05 0.01 0.02 0.00 0.49

Enterobacter aerogenes et rel. 0.21 0.03 0.13 0.01 0.61

Escherichia coli et rel. 0.27 0.07 5.78 2.84 21.42

Klebisiella pneumoniae et rel. 0.09 0.01 0.12 0.02 1.32

Oceanospirillum. 0.05 0.01 0.04 0.00 0.73

Oxalobacter formigenes et rel. 0.20 0.04 0.11 0.01 0.56

Serratia 0.01 0.00 0.09 0.03 6.66

Sutterella wadsworthia et rel. 0.22 0.02 0.14 0.02 0.63

Xanthomonadaceae 0.08 0.01 0.05 0.01 0.55

Verrucomicrobia Akkermansia 1.79 0.19 4.60 1.49 2.57

131

Table 5.7 Correlation coefficients (%SI) comparing biofilm samples to the planktonic samples

from the time of harvest (by DGGE and HITChip).

Comparison %SI, by DGGE %SI, by HITChip

V1 V2 V1 V2

B1 vs. P Day 36 45.3 54.7 89.8 90.6

B2 vs. P Day 41 60.8 52.9 79.0 89.7

B3 vs. P Day 48 78.9 49.1 95.7 91.1

B4 vs. P Day 48 54.4 49.4 94.2 90.9

132

period and increased once clindamycin addition was stopped (Figures 5.1a, 5.2, 5.6). On day 41,

V1-P and V2-P were 48.2% similar by DGGE and 81.4% similar by HITChip (Table 5.1).

However, V1-P and V2-P similarity did not recover to pre-clindamycin levels by the end of the

experiment (Figure 5.1a). On day 48, V1-P and V2-P were 75.8% similar by DGGE and 87.1%

similar by HITChip (Table 5.1). Clustering tree analysis and the PCA and RDA plots showed

that V1-P samples taken after the clindamycin treatment period clustered separately from other

untreated planktonic samples (Figures 5.4, 5.5, A5.1).

V1 rate of change values showed a peak of change between days 36-38 (V1 Δt= 46.6%, V2

Δt = 6.3%). Δt values were stable in V1 and V2 between days 40-46, however V1 Δt values were

higher than V2 Δt values (average V1 Δt = 13.1%, V2 Δt= 6.1).

HITChip analysis showed that V1-P day 41 was positively associated with Bacteroides

fragilis et rel., Bacteroides ovatus et rel., Subdoligranulum variabile at rel., and Escherichia coli

et rel. compared to planktonic communities not exposed to clindamycin (Figure 5.3, Table 5.8).

V1-P day 41 was negatively associated with some members of the Bacteroidetes and

Firmicutes (especially groups belonging to Clostridium clusters IV and XIVa). V1-P day 48 was

positively associated with Bacteroides fragilis et rel., Clostridium orbiscindens et rel.,

Subdoligranulum variabile at rel., and Dorea formicigenerans et rel. Resistant Bacteroides spp.

appeared to be recovering to normal levels, while other Bacteroides spp. that were reduced

following clindamycin exposure remained present in lower abundances. Clostridium

orbiscindens et rel. showed a 'rebound' effect, where its abundance was decreased immediately

following antibiotic treatment but grew beyond baseline levels during the recovery period.

Subdoligranulum variabile at rel. were still present in higher abundances than baseline levels but

greatly reduced in comparison to their abundance immediately following antibiotic treatment.

133

Table 5.8 Percent abundance of major genus-like level bacterial groups (present at an abundance >0.5% in at least 1 sample) and

fold differences (V1-P clindamycin-exposed sample/average of planktonic samples not exposed to clindamycin). Genus-like

levels are only shown where the fold differences were >2.5 (higher in clindamycin-exposed samples, shown in green) or <0.5

(lower in clindamycin-exposed samples, shown in red).

Higher taxonomic groups Level 2

(genus-like level)

Average planktonic

abundance

(%) without

clindamycin

V1-P Day 41

abundance (%)

V1-P Day 48

abundance (%)

Fold (V1-P Day 41/

Average planktonic)

Fold (V1-P Day 48/

Average planktonic)

Bacteroidetes

Bacteroides fragilis et rel. 5.25 25.39 13.52 4.84 2.58 Bacteroides ovatus et rel. 7.52 26.46 14.85 3.52 1.97

Bacteroides plebeius et rel. 0.93 0.38 0.22 0.41 0.24

Bacteroides stercoris et rel. 0.92 0.51 0.34 0.56 0.37 Bacteroides uniformis et rel. 7.73 0.14 0.10 0.02 0.01

Bacteroides vulgatus et rel. 19.71 2.60 2.02 0.13 0.10

Prevotella melaninogenica et rel. 0.66 0.30 0.20 0.45 0.30 Prevotella oralis et rel. 0.56 0.37 0.27 0.66 0.47

Prevotella tannerae et rel. 0.50 0.31 0.24 0.61 0.49

Firmicutes

Clostridium cluster IV

Clostridium cellulosi et rel. 0.84 0.38 0.30 0.45 0.36

Clostridium leptum et rel. 1.81 0.29 0.30 0.16 0.16

Clostridium orbiscindens et rel. 2.06 0.70 16.90 0.34 8.22

Ruminococcus bromii et rel. 2.79 0.09 0.10 0.03 0.04

Subdoligranulum variabile et rel. 0.79 12.11 5.35 15.40 6.80

Clostridium cluster XIVa

Butyrivibrio crossotus et rel. 0.81 0.35 0.54 0.44 0.67

Clostridium sphenoides et rel. 2.45 0.94 2.77 0.39 1.13

Clostridium symbiosum et rel. 19.28 2.38 17.73 0.12 0.92

Dorea formicigenerans et rel. 0.90 1.04 3.08 1.16 3.44

Uncultured Clostridiales Uncultured Clostridiales I 2.81 0.59 0.43 0.21 0.15

Proteobacteria Escherichia coli et rel. 0.27 0.73 0.35 2.69 1.30

Verrucomicrobia Akkermansia 1.79 0.13 0.99 0.07 0.56

134

Dorea formicigenerans et rel. was not greatly affected by clindamycin treatment but grew

beyond baseline levels after antibiotic treatment was stopped. V1-P day 48 was negatively

associated with some members of the Bacteroidetes and Firmicutes, however members of

Clostridium cluster XIVa affected by antibiotic treatment generally recovered to baseline

abundances.

V1-P H’, EH’, and S values decreased during the clindamycin treatment period and began

to increase once clindamycin addition was stopped (Figure 5.1c,d,e; Table 5.4). However, V1-P

H’, EH’, and S values did not recover to V1-P day 36 (pre-clindamycin) or the corresponding V2-

P (control) values by the end of the experiment. V2-P H’, EH’, and S values remained stable

between days 36-48 (Figure 5.1c,d,e; Table 5.4).

5.2.5 Clindamycin treatment altered the structure of simulated mucosal biofilms

V1-B2, V1-B3, and V1-B4 were compared to between-vessel (V2-B2, V2-B3, and V2-B4)

and within-vessel (V1-B1) controls to characterize the effect of community disturbance on

simulated mucosal communities. Photographs of the biofilms present in B2, B3, and B4 packed-

column reactors on days 41 and 48 are shown in Figure 5.7.

Following clindamycin treatment there was a loss of distinction between biofilm and

planktonic communities. B2, B3, and B4 samples from V1 and V2 became less similar to each

other than B1 samples (Table 5.1). In addition, V1-B2, V1-B3, and V1-B4 were less similar to

V1-B1 compared to the corresponding V2 comparisons (Figure 5.2, Table 5.5). Clustering tree

analysis showed that V1-B2 clustered farther from V1-B1 and V2-B2 (Figures 5.2, 5.4). V1-B3

and V1-B4 clustered farther from other untreated biofilm samples than V1-B2 and clustered

closer to V1-P days 41 and 48. PCA and RDA plots show that V1-B3 and V1-B4 cluster closer

135

to V1-P day 48 than other biofilm samples (Figures 5.5, A5.1). PCA and RDA plots also show

that the composition of V1-B2 grouped between the cluster of biofilm samples not exposed to

clindamycin and the cluster of V1-P day 48, V1-B3, and V1-B4.

HITChip analysis indicated that V1-B2 had similar compositions compared to biofilms not

exposed to clindamycin (Figures 5.2, 5.3, Tables 5.1, 5.9). V1-B2 was positively associated with

Bifidobacterium spp., Bacteroides fragilis et rel., and Bacteroides ovatus et rel. compared to

biofilm communities not exposed to clindamycin. V1-B2 was negatively associated with

Bacteroides plebeius et rel., Bacteroides uniformis et rel., and Escherichia coli et rel., however

in general Firmicutes members appeared less affected than those in planktonic samples. V1-B3

and V1-B4 gave comparable results, however V1-B4 was less similar to biofilms not exposed to

clindamycin than V1-B3. V1-B3 and V1-B4 were positively associated with Bacteroides fragilis

et rel., Bacteroides ovatus et rel., Parabacteroides distasonis et rel., Subdoligranulum variable

et rel. and Dorea formicigenerans et rel. and negatively associated with some members of the

Bacteroidetes and Firmicutes.

Diversity, evenness, and richness values from V1 and V2 biofilm samples became less

comparable following exposure to clindamycin (Table 5.4). These differences increased in the

following order: V1-B2, V1-B3, V1-B4.

5.3 Discussion

In this study we characterized simulated luminal and mucosal distal gut communities in our

twin-vessel single-stage chemostat model. We showed that planktonic and biofilm communities

had between- and within-vessel reproducibility upon the establishment of steady state conditions.

Planktonic and biofilm communities had distinct compositions, which were perturbed to different

136

Table 5.9 Percent abundance of major genus-like level bacterial groups (present at an abundance >0.5% in at least 1 sample) and

fold differences (V1-B2, V1-B3 or V1-B4/average of biofilm samples not exposed to clindamycin). Genus-like levels are only

shown where the fold differences were >2.5 (higher in clindamycin-exposed biofilm samples, shown in green) or <0.5 (lower in

clindamycin-exposed biofilm samples, shown in red).

Higher taxonomic groups Level 2

(genus-like level)

Average biofilm

abundance

(%) without

clindamycin

V1-B2

abundance

(%)

V1-B3

abundance

(%)

V1-B4

abundance

(%)

Fold (V1-B2/

Average

biofilm)

Fold (V1-B3/

Average

biofilm)

Fold (V1-B4/

Average

biofilm)

Actinobacteria Bifidobacterium 0.85 2.23 0.29 0.26 2.61 0.34 0.30

Bacteroidetes

Bacteroides fragilis et rel. 1.61 5.97 13.39 18.37 3.71 8.33 11.42 Bacteroides ovatus et rel. 1.93 7.94 13.95 18.25 4.11 7.22 9.45

Bacteroides plebeius et rel. 0.58 0.27 0.27 0.31 0.46 0.47 0.53

Bacteroides uniformis et rel. 3.76 0.51 0.27 0.14 0.14 0.07 0.04 Bacteroides vulgatus et rel. 7.85 6.39 2.33 3.12 0.81 0.30 0.40

Parabacteroides distasonis et rel. 1.50 2.88 4.60 3.79 1.93 3.07 2.53

Firmicutes

Clostridium cluster IV

Anaerotruncus colihominis et rel. 0.69 0.67 0.41 0.28 0.98 0.60 0.41

Clostridium cellulosi et rel. 2.49 2.21 0.87 0.43 0.89 0.35 0.17

Clostridium leptum et rel. 3.13 3.36 1.01 0.41 1.07 0.32 0.13

Oscillospira guillermondii et rel. 12.49 8.85 9.05 3.82 0.71 0.72 0.31

Ruminococcus bromii et rel. 0.51 0.32 0.16 0.07 0.63 0.32 0.14

Sporobacter termitidis et rel. 1.72 2.80 1.04 0.58 1.63 0.61 0.33

Subdoligranulum variable et rel. 1.08 2.03 13.48 11.29 1.89 12.51 10.48

Clostridium cluster IX Dialister 0.19 0.18 0.12 0.09 0.92 0.60 0.49

Clostridium cluster XIVa

Butyrivibrio crossotus et rel. 0.73 0.77 0.39 0.36 1.04 0.53 0.48

Clostridium symbiosum et rel. 25.67 21.74 12.28 6.31 0.85 0.48 0.25

Dorea formicigenerans et rel. 0.76 1.12 1.99 1.99 1.49 2.63 2.63

Uncultured Clostridiales Uncultured Clostridiales I 1.05 1.09 0.87 0.41 1.04 0.83 0.39

Proteobacteria Escherichia coli et rel. 5.78 1.75 1.72 9.85 0.30 0.30 1.70

Verrucomicrobia Akkermansia 4.60 4.11 0.57 0.31 0.89 0.12 0.07

137

degrees following antibiotic treatment. Therefore, we conclude that the addition of a packed

column biofilm reactor to our distal gut model provides a convenient method to simultaneously

simulate both mucosal and luminal communities of the human distal gut.

Incorporation of simulated mucosal environments into in vitro gut models creates a more

biologically significant, heterogeneous environment for the development of microbial

ecosystems. However, planktonic and biofilm communities developed in these systems must be

characterized and validated before placebo-controlled or multi-treatment experiments can be

conducted. While previous studies have shown that planktonic communities developed in

luminal gut models are reproducible, planktonic community reproducibility in mucosal gut

models still requires validation (165). In our system, planktonic communities from twin vessels

were stable and had similar compositions by 34-36 days post-inoculation.

In addition to planktonic community diversity, biofilm community reproducibility must

also be validated. To our knowledge, this is the first study characterizing the reproducibility of

simulated mucosal biofilms developed in a gut chemostat model. In our study, simulated

mucosal biofilms developed in packed-column reactors showed high between- and within-vessel

reproducibility. Biofilms harvested from the same vessel were more reproducible when biofilm

initiation, harvest, and development times were similar. Future experiments could characterize

biofilms developed for shorter periods and harvested at regularly spaced, staggered time points.

This would allow for a more dynamic view of biofilm formation that would mimic the regular

renewal of the mucus layer and sloughing off of host epithelial cells in vivo (270).

Simulated luminal and mucosal communities developed within our system had distinct

compositions. Although, as noted in chapter 3, correlation coefficients were higher when

samples were compared by HITChip than by DGGE (~50% similar by DGGE vs. ~90% similar

138

by HITChip, see discussion in chapter 8). Studies using an M-SHIME have found that planktonic

and biofilm communities are 14-19% similar by DGGE (on average; 263) or less than 60%

similar by DGGE (262). An in vivo study by Zoetendal et al. (2002) found that matched fecal

samples and descending colon biopsy specimens from healthy individuals were 13.6-91.3%

similar by DGGE (259). In their study, the jaws of the colonoscope were carefully washed with

tap water following biopsy sampling to minimize contamination by non-adherent

microorganisms from the intestinal content. Comparison of our DGGE correlation coefficient

values to the in vivo values obtained by Zoetendal et al. (2002) confirms that both gut models

can develop planktonic and biofilm communities with biologically significant differences in

community composition.

In vivo studies have found an enrichment of Firmicutes (especially Clostridium cluster

XIVa) over Bacteroidetes in human biopsy specimens (compared to fecal samples; 6, 93, 271-

274). In our study, HITChip analysis showed biofilm samples were positively associated with

Firmicutes (especially Clostridium cluster IV), and negatively associated with Bacteroidetes

compared to steady state planktonic communities. Van den Abbeele et al. (2012) used an M-

SHIME (mucosal simulator of the human intestinal microbial ecosystem) to culture simulated

luminal and mucosal gut communities (263). They found that Bacteroidetes and Proteobacteria

were enriched in the planktonic phase while Firmicutes (especially Clostridium cluster XIVa)

were enriched in the biofilm phase (263). In another study, Macfarlane et al. (2005) used a two-

stage mucosal chemostat to model the proximal and distal colon (267). They found that

Bacteroides, Clostridium, and Bifidobacterium spp. were enriched in the planktonic phase while

members of the Bacteroides fragilis group, Enterobacterium spp., and Clostridium spp. were

enriched in the biofilm phase (of the distal colon vessel; 267). Variations in the composition of

139

biofilm communities developed in our model relative to previous studies may be due to

differences in the composition of the donor's fecal inoculum, in the type of gut model employed

(e.g. single-stage vs. multistage), in the device used to culture biofilms (e.g. packed-column

reactor vs. microcosms in a netted bag vs. gels in glass tubes), and the analysis techniques (e.g.

HITChip vs. culturing).

In the absence of antibiotic treatment, biofilm sample H’, EH’, and S values were lower

than planktonic sample values by DGGE, but higher than planktonic sample values by HITChip

(Table 5.4). Increased sensitivity of the HITChip relative to DGGE may account for these

differences, as band co-migration can limit proper DGGE band separation (275, 276). Durbán et

al. (2011) found few differences between the average H’ values for fecal and biopsy specimens

from healthy human subjects (260). However, data from each individual showed that biopsy H’

values were higher or lower than the corresponding fecal H’ values depending on the subject of

interest. In addition, it has been proposed that feces consist of luminal and shed/poorly adherent

mucosal microorganisms (260). This means the diversity of planktonic communities may be

higher than that of biofilm communities.

Clindamycin-induced perturbation of chemostat communities caused biologically

significant shifts in the composition of planktonic communities. HITChip results indicated that

clindamycin treatment resulted in a broad reduction of the Bacteroidetes and Firmicutes

(especially groups belonging to Clostridium clusters IV and XIVa) and an increase in the

abundance of clindamycin-resistant Bacteriodetes (Bacteroides fragilis et rel., Bacteroides

ovatus et rel.), Subdoligranulum variabile at rel., and Escherichia coli et rel. While planktonic

communities recovered following clindamycin treatment, recovery was incomplete. Previous

pure culture studies have shown that Bacteroides fragilis (277), Bacteroides ovatus (278), and

140

Escherichia coli (279) are resistant to clindamycin. While we were unable to find published

studies describing clindamycin sensitivity in Subdoligranulum variabile isolates, characterization

of isolates from our laboratory suggest that clindamycin sensitivity varies by strain (Table A5.1).

There are several published human studies that have investigated the effects of

clindamycin on the structure of fecal communities. Kager et al. (1981) found that colorectal

surgery patients receiving clindamycin prophylaxis had decreased fecal populations of anaerobic

(Bacteroides spp., Clostridium spp., Bifidobacterium spp., Lactobacillus spp., Fusobacterium

spp., Eubacterium spp., Veillonella spp., Megasphaera spp.) and aerobic (Enterococcus spp. and

Streptococcus spp.) bacteria that recovered following clindamycin treatment (196). Jernburg et

al. (2007) found that clindamycin treatment caused significant disturbances in human fecal

communities (specifically, a sharp decline in Bacteroides diversity), which did not return to their

starting composition following clindamycin treatment, as well as the persistence of clindamycin-

resistant Bacteroides spp. (consistent with our results, as shown in Figure A5.2; 231).

The effect of clindamycin on the structure of fecal communities has also been studied in

animal models and in vitro gut fermentation models. Fecal samples from mice and hamsters

exposed to clindamycin showed a decrease in gut microbiota diversity, with a reduction in the

abundance of Bacteroidetes and Firmicutes species and an increase in the abundance of

clindamycin-resistant microbes, including some members of the Proteobacteria (280-282).

Edwards et al. (1986) found that the addition of clindamycin to a luminal chemostat model of the

proximal colon resulted in a decrease in planktonic anaerobe diversity, resulting in partial

community recovery for specific clindamycin-resistant Bacteroides species (283). Comparison of

our data to data from other studies is difficult, as previous work has shown that responses to

antibiotics are individualized and influenced by past exposure to the same antibiotic (62).

141

However, overall we found that planktonic communities developed within our model responded

to clindamycin treatment in a biologically relevant manner.

Clindamycin treatment also altered the composition of simulated mucosal biofilms. The

impact of clindamycin exposure increased in the following order: mature biofilms (V1-B2),

biofilms forming during clindamycin exposure (V1-B3), biofilms formed following clindamycin

exposure (V1-B4). Previously, pure culture chemostat studies with Pseudomonas aeruginosa

have found that antibiotic susceptibility increased in the following order: old biofilms, young

biofilms, planktonic cells (217, 218). V1-B2 was the oldest biofilm exposed to clindamycin

(cultured for 36 days prior to antibiotic treatment) and the most resistant to clindamycin

treatment. V1-B3 could have initiated biofilm growth early during antibiotic treatment (bacteria

can bind to mucin within 15 minutes; 284, 285), or may have seeded biofilm growth from

sloughed off biofilm cells from V1-B1 or V1-B2 (which could have been more resistant to

antibiotic exposure than planktonic communities). V1-B4 biofilm growth was initiated following

clindamycin exposure, after susceptible planktonic cells may have been removed from the

system. However, sloughed off biofilm cells from V1-B2 or V1-B3 may have been circulated

into the planktonic culture in our system, and thus would have been available to seed V1-B4

biofilm growth.

To date, few studies have investigated the impact of antibiotic exposure on healthy gut

mucosal communities. However, Peterfreund et al. (2012) found that hamsters sensitized to

Clostridium difficile infection using clindamycin had a loss of distinction between cecal luminal

contents and tissue-associated communities (280). In our study we found that biofilms that

formed either during or following clindamycin exposure were more similar to recovering

planktonic communities than to biofilms not exposed to antibiotics. These findings suggest that

142

biofilms developed in our model responded to clindamycin in a biologically relevant manner.

Other studies have suggested that antibiotics can influence the composition of microbial

communities. Nomura et al. (2005) found that UC patients receiving three antibiotics

(amoxicillin, tetracycline, and metranidazole) had a significant reduction in Fusobacterium

varium numbers in biopsy specimens and improved clinical and histological scores (286).

Swidsinski et al. (2008) found that UC and indeterminate colitis patients receiving two

antibiotics (metronidazole and ciprofloxacin) had reduced total numbers of bacteria associated

with biopsy specimens, which increased one week following the cessation of antibiotic therapy

(287). As mentioned previously, the proximity of mucosal communities to the host epithelium

means that changes in mucosal community composition may be important to host health (265).

In conclusion, packed-column biofilm reactors were able to successfully incorporate a

simulated mucosal environment in our twin-vessel single-stage chemostat model of the human

distal gut. Planktonic and biofilm communities developed in this model were complex,

reproducible, and biologically significant with distinct compositions. Moreover, these planktonic

and biofilm communities were able to respond to perturbations in a biologically relevant manner.

As a consequence, this model can be effectively used for placebo-controlled or multi-treatment

experiments to simultaneously investigate the effect of perturbations on the gut microbial

ecosystem independent of the effects of stimuli on the host. Examples of studies that could be

conducted in this system include modeling the effects of novel probiotics (single or multi-

species), prebiotics, antibiotics, drugs, or genetically modified microorganisms on planktonic and

biofilm communities (262, 263). In vitro infection experiments or fecal transplant experiments

could also be conducted within this system. If diseased mucosal and luminal communities can

also be successfully modeled, novel treatments could be tested in vitro prior to in vivo validation

143

(266).

144

Chapter 6

Using derived, defined experimental

communities to model the human distal

gut in vitro

6.1 Introduction

Chemostat models of the human gut are generally seeded with fresh feces and cultured

under conditions set to simulate the gut environment. Although fecal donations provide a

convenient source of inoculating material, the resulting complex chemostat communities require

extended periods of time to establish steady state compositions (up to 36 days post-inoculation;

see Chapter 3). This limits the number of experiments that can be conducted within a fixed time

frame and increases the consumption of laboratory materials required to maintain the microbial

communities. In addition, alterations in the composition of a donor's fecal communities between

consecutive donations results in reduced between-run repeatability (relative to within-run

reproducibility; see Chapter 3). Studies have attempted to compensate for between-run variation

by freezing the fecal inoculum in order to preserve it, however previous work has suggested that

culturing microorganisms from frozen fecal samples results in reduced recovery of specific fecal

microorganisms (288). Feria-Gervasio et al. (2011) also showed that increased frozen feces

storage times between different runs resulted in less similar community profiles in an in vitro gut

model (214).

Simplified, defined communities may represent a useful alternative to feces for the study of

human distal gut communities. Defined community composition can be controlled to ensure

145

reproducibility between consecutive chemostat runs or to suit the experiment of interest. In

addition, as improvements are made in culturing techniques the composition of defined

communities can be increased to more accurately reflect the native fecal diversity of the host,

both in terms of membership and function. Conversely, defined communities can also be

manipulated to easily model what happens when a given taxonomic group (e.g. genus, family, or

class) is removed from a simulated gut community.

By using defined communities consisting of members with sequenced genomes, more

detailed mechanistic studies can be conducted which would not be possible using undefined fecal

communities. Improvements to genomic sequencing, with the advent of next-generation

sequencing technologies (low-cost, high throughput methods of genomic sequencing), means

microbial genomes are now easy to obtain (289). For example, metatranscriptomic analysis of a

microbial community can provide a snapshot of the gene expression levels (including actively

transcribed ribosomal and messenger RNA) present in the community at the time of sampling

(290). These gene expression profiles allow researchers to directly study how microbial

communities respond to changes in their environment (290). While researchers can obtain these

profiles from undefined communities, changes in gene expression profiles are difficult to

correlate with specific microbes. Access to a defined community consisting of members with

sequenced genomes would allow us to link the expression of transcripts back to members of the

defined community.

To date few attempts have been made to mimic the human gut microbiota using simplified,

defined communities because the fastidious nutritional and environmental conditions required for

growth of most fecal microorganisms often challenges their construction. Of the previous studies

that have formulated defined communities, most were composed of 10 or less species and

146

usually consisted of strains isolated from the feces of different donors (25, 154, 291-294). To

more accurately model the interactions that occur in vivo, defined communities should be

composed of microorganisms isolated from the feces of the same donor. Pooling microorganisms

from the feces of different hosts may result in inter-microbe interactions that favour the growth

of specific microbial populations, resulting in steady state communities that are less

representative of a human distal gut ecosystem (2).

Before defined communities can be used as a simulated distal gut community they must be

compared to the fecal communities from which they were derived. Defined communities must

maintain the general characteristics of the distal gut communities if they are to be used for

ecological studies. In this study we compared simplified, defined communities to their respective

originating, complex, undefined fecal communities in a single-stage chemostat model of the

human distal gut. The aims of this study were (i) to evaluate the within-run reproducibility of

fecal and defined communities developed in twin-vessel single-stage chemostats (technical

replicates); (ii) to determine the length of time required for fecal and defined communities to

reach steady state compositions; (iii) to compare fecal and defined steady state chemostat

community compositions to each other and to the fecal inoculum; and (iv) to characterize

community diversity, evenness, and richness between vessels and within the same vessel over

time.

6.2 Results

Three separate chemostat runs were analyzed in this study: twin-vessels seeded with donor

B feces (VF1 and VF2), twin-vessels seeded with DEC (50 cultivable strains from donor B

feces) (VD1 and VD2) and a single vessel seeded with MET-1 (33 cultivable commensal strains

147

from donor B feces) (VM1). Twin-vessels seeded with fecal communities (VF1 and VF2) were

analyzed until 36 days post-inoculation. Of the two vessels seeded with DEC, VD1 was analyzed

until 28 days post-inoculation and VD2 was operated until 14 days post-inoculation. The vessel

seeded with MET-1 (VM1) was analyzed until 28 days post-inoculation.

6.2.1 Twin-vessel reproducibility

Community development in VD1 and VD2 twin-vessels were compared to VF1 and VF2

twin-vessels to characterize ecosystem reproducibility and time to steady state compositions (by

DGGE). Paired time-points from VF1 and VF2 were run on the first series of DGGE gels until

36-days post-inoculation to allow for the direct comparison of the reproducibility and stability of

chemostat communities seeded with feces. Paired time-points from VD1 and VD2 were run on

the second series of DGGE gels until 14-days post-inoculation to allow for the direct comparison

of the reproducibility and stability chemostat communities seeded with DEC.

The length of time required for twin-vessel correlation coefficients to rise above the gel-

defined cut-off thresholds varied depending on the inoculum (Figure 6.1a,b, Table 6.1). Twin-

vessel correlation coefficients in VF1 and VF2 rose above the cut-off by 30 days post-

inoculation (Figure 6.1a). VF1 and VF2 DGGE profiles were 95.1% similar immediately

following inoculation (day 0) and 97.4% similar upon the establishment of steady state (30 days

post-inoculation; Figure 6.2a). Twin vessels seeded with defined communities (VD1 and VD2)

did not drop below their gel-defined cut-off (Figure 6.1b). VD1 and VD2 DGGE profiles were

95.5% similar immediately following inoculation (day 0) and 97.4% similar upon the

establishment of steady state (12 days post-inoculation; Figure 6.2b).

148

Figure 6.1 DGGE analyses of twin-vessel chemostat communities seeded with feces (panels on

left) or DEC (panels on right). The vertical dashed line represents the beginning of steady state

conditions. Panels a and b) Correlation coefficients (expressed as percentages) comparing the

profiles of each vessel at identical time points. Panels c and d) Community dynamics calculated

using moving window correlation analysis. Panels e and f) Adjusted Shannon Diversity Index

(H’). Panels g and h) Adjusted Shannon equitability index (EH’). Panels i and j) Adjusted band

richness (S).

149

0

10

20

30

40

50

60

70

80

90

100

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36

%S

I

Day

VF1 vs. VF2 Steady state Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

0 2 4 6 8 10 12 14

%S

I

Day

VD1 vs. VD2 Steady state Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36

%S

I

Day (x-2) vs. (x)

VF1 VF2 Steady state Cut-off Cut-off+5%

0

10

20

30

40

50

60

70

80

90

100

2 4 6 8 10 12 14

%S

I

Day (x-2) vs. (x)

VD1 VD2 Steady state Cut-off Cut-off+5%

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36

H'

Day

VF1 VF2 Steady state

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 2 4 6 8 10 12 14

H'

Day

VD1 VD2 Steady state

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36

EH

'

Day

VF1 VF2 Steady state

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 2 4 6 8 10 12 14

EH

'

Day

VD1 VD2 Steady state

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36

S

Day

VF1 VF2 Steady state

0

10

20

30

40

50

60

70

0 2 4 6 8 10 12 14

S

Day

VD1 VD2 Steady state

a b

c d

e f

g h

i j

150

Table 6.1 Time (in days) to steady state compositions based on measured parameters from

DGGE gels. NA = not available.

Parameter VF1 VD1 VM1

Vessel comparison (%SI) 30 0 NA

Rate of change (Δt) 30 12 12

Band richness (S) 28 12 18

Time to steady state (days) 30 12 18

151

Figure 6.2 DGGE profiles representing inocula and steady state communities from twin-vessels.

a) Twin-vessels seeded with feces (VF1 and VF2). b) Twin-vessels seeded with DEC (VD1 and

VD2).

VF

-1 D

ay

30

VF

-2 D

ay

30

VF

-1 D

ay

0

VF

-2 D

ay

0

VD

-1 D

ay

12

VD

-2 D

ay

12

VD

-1 D

ay

0

VD

-2 D

ay

0

ba

152

Moving window correlation analysis showed that twin-vessel communities developed in a

similar manner for the duration of the run (Figure 6.1c,d). Δt values were larger and more

variable at the beginning of the run (during the stabilization period). VF1 and VF2 Δt values

stabilized as steady state conditions were established, dropping and remaining below the gel-

defined cut-off in both vessels by 28-30 days post-inoculation, where Δt values were 1.0% in

VF1 and 5.2% in VF2. VD1 and VD2 Δt values also stabilized as steady state conditions were

established, dropping and remaining below the gel-defined cut-off in both vessels by 10-12 days

post-inoculation, where Δt values were 6.9% in VD1 and 4.1% in VD2.

Twin-vessels seeded with feces or DEC had reproducible Shannon diversity index (H'),

Shannon equitability index (EH'), and band richness (S) values (Figure 6.1e-j). Twin-vessel H',

EH', and S values are listed in Table 6.2.

6.2.2 VF1, VD1 and VM1 correlation coefficients

VD1 and VM1 were compared to VF1 to determine whether defined communities are

capable of mimicking fecal communities by maintaining overall fecal community structure.

Corresponding time-points from VF1, VD1, and VM1were run on the third series of DGGE gels

until 28-days post-inoculation (for VD1 and VM1) or 36-days post-inoculation (for VF1) to

allow for the direct comparison of chemostat communities seeded with feces to simplified,

defined communities with varying diversity (Figure 6.3). Several samples were analyzed using

the HITChip, including: VF1 inoculum (feces), VF1 day 36, VD1 inoculum (DEC), VD1 day 28,

VM1 inoculum (MET-1), VM1 day 28.

Vessel correlation coefficients showed that communities from VF1, VD1, and VM1 had

distinct compositions (Figure 6.4a). With the exception of VD1 vs. VM1 on day 2, VF1, VD1

153

Table 6.2 Ecological parameters (diversity, evenness, and richness) for VF1, VD1, and VM1 inocula (day 0) and steady state

communities (VF1/VF2 = day 30, VD1/VD2 = day 12) as measured using DGGE.

Ecological Parameter VF-1

Day 0

VF-2

Day 0

VF-1

Day 30

VF-2

Day 30

VD-1

Day 0

VD-2

Day 0

VD-1

Day 12

VD-2

Day 12

Shannon Diversity Index (H’) 3.08 3.07 3.10 3.05 3.44 3.41 3.14 3.17

Shannon Equitability Index (EH’) 0.75 0.75 0.73 0.72 0.83 0.82 0.76 0.76

Richness (S) 60 60 71 71 63 63 63 63

154

Figure 6.3 Molecular fingerprints representing the fecal inoculum and steady state communities

from vessels seeded with feces (VF1), DEC (VD1) or MET-1 (VM1). (a) DGGE profiles (b)

HITChip profiles. HITChip profiles also list the highest phylogenetic level of specificity of

probes (level 1) on the right of panel b.

155

VF

1 D

ay 0

VF

1 D

ay 3

6

VD

1 D

ay 2

8

VM

1 D

ay 2

8

b

VF

-1 D

ay 3

6

VM

-1 D

ay 2

8

VD

-1 D

ay 2

8

VF

-1 D

ay 0

a

Clostridium cluster XVIII

Actinobacteria

Asteroleplasma

Bacilli

Bacteroidetes

Clostridium cluster I

Clostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XI

Clostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XVClostridium cluster XVIClostridium cluster XVII

CyanobacteriaFusobacteria

Proteobacteria

Spirochaetes

Uncultured Clostridiales

Uncultured MollicutesVerrucomicrobia

156

Figure 6.4 DGGE analysis of chemostat communities seeded with feces (VF1), DEC (VF1) or

MET-1 (VM1). a) Correlation coefficients (expressed as percentages) comparing the profiles of

each vessel at identical time points. b) Correlation coefficients comparing the profiles of each

vessel over the course of the chemostat run to the fecal inoculum (VF1 day 0). c) Community

dynamics calculated using moving window correlation analysis. d) Adjusted Shannon Diversity

Index (H’). e) Adjusted band richness (S). f) Adjusted Shannon equitability index (EH’).

0

10

20

30

40

50

60

70

80

90

100

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28

%S

I

Day

VF1 vs. VD1 VF1 vs. VM1 VD1 vs. VM1 Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36

%S

I

Day

VF1 VD1 VM1 Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

2 6 10 14 18 22 26 30 34

%S

I

Day (x-2) vs. (x)

VF1 VD1 VM1 Cut-off Cut-off +5%

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36

H'

Day

VF1 VD1 VM1

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36

S

Day

VF1 VD1 VM1

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36

EH

'

Day

VF1 VD1 VM1

a b

c d

e f

157

and VM1 correlation coefficients remained below the gel-defined cut-off thresholds. Similarities

between VF1, VD1 and VM1 inocula and steady state communities are shown in Table 6.3.

Again, the %SI for these samples were higher when they were calculated using the HITChip than

when they were calculated using DGGE (see Chapter 8). Clustering tree analysis (calculated at

the oligonucleotide-level) showed that the VD1 inoculum clustered more closely with the VM1

inoculum than the VF1 inoculum, but the VD1 steady state (day 28) community clustered more

closely with the VF1 steady state (day 36) community than with the VM1 steady state (day 28)

community (Figure 6.5). The PCA plot (calculated at the genus-like level) showed that the VF1

and VD1 inoculum clustered more closely with each other than with the VM1 inoculum, and that

the VF1 and VD1 steady state chemostat communities clustered more closely with each other

than with the VM1 steady state chemostat community (Figure 6.6).

Similarities between the fecal inoculum and VF1, VD1, or VM1 steady state communities

are shown in Figure 6.4b. By DGGE, the fecal inoculum was 39.4% similar to VF1 day 36, 43%

similar to VD1 day 28 and 34% similar to VM1 day 28. By HITChip, the fecal inoculum was

68.0% similar to VF1 day 36, 60.5% similar to VD1 on day 28 and 67.5% similar to VM1 on day

28.

6.2.3 VF1, VD1, VM1 stability and diversity

Moving window correlation analysis showed that the time required for communities to

maintain low rate of change (Δt) values varied depending on the composition of the inoculum

(Figure 6.4c). Low Δt values were maintained by 28-30 days post-inoculation for VF1,

(Δt=5.9%), 10-12 days post-inoculation for VD1 (Δt=4.2%), and 10-12 days post-inoculation for

VM1 (Δt=4.1%; Figure 6.4c).

158

Table 6.3 Correlation coefficients (%SI) comparing VF1, VD1, and VM1 inocula (day 0) and

steady state communities (VF1 = day 36, VD1 and VM1 = day 28).

Analysis Method Vessel Comparison %SI, inocula %SI, steady state

DGGE

VF1 vs. VD1 37.0 66.5

VF1 vs. VM1 48.2 47.3

VD1 vs. VM1 60.0 35.5

HITChip

VF1 vs. VD1 71.3 79.2

VF1 vs. VM1 75.4 66.0

VD1 vs. VM1 83.0 75.1

159

Figure 6.5 Dendrogram comparing chemostat communities seeded with feces (VF1), DEC

(VD1), or MET-1 (VM1). Dendrograms are based on (a) Pearson and UPGMA correlation

(DGGE profiles) or (b) Pearson correlation and Ward's minimum variance method (HITChip

profiles).

160

Figure 6.6 Principal component analysis (PCA) at the genus-like level based on HITChip data of the VF1, VD1, and VM1

inocula (day 0) and steady state chemostat communities (VF1 = day 36, VD1 and VM1 = day 28). Arrows shown in the plot

correspond to the top 30 genus-like groups that explain most of the variability in the dataset.

-1.0 1.5

-0.6

1.0

VF1 inoc

VF1 Day 36

VD1 inoc

VD1 Day 28

VM1 inoc

VM1 Day 28

PC1: 48.3%

PC

2:

20.7

%

Bacteroidetes

Clostridium cluster IX,

Proteobacteria and Bacilli

Bifidobacteria and

Clostridium cluster XIVa

161

H’, EH’, and S values for VF1, VD1, and VM1 inocula and steady state communities are

listed in Table 6.4. DGGE band richness (S) values fluctuated in all three vessels following

inoculation, but stabilized by 28 days post-inoculation in VF1, 12 days post-inoculation in VD1,

and 18 days post-inoculation in VM1 (Figure 6.4e). VF1 Shannon diversity index (H') and

Shannon equitability index (EH') values were stable over the course of the experiment (Figure

6.4d,f). VD1 H' and EH' values were stable from day 2 onwards, after experiencing an initial drop

between days 0-2 (Figure 6.4d,f). However, while VM1 H’ and EH’ values generally stabilized

after day 18, values gradually declined until the end of the experiment (Figure 6.4d,f).

6.2.4 Phylogenetic analysis of VF1, VD1, and VM1

HITChip data provided phylogenetic information for the inocula and steady state

communities (Figure 6.7, Table 6.5). Bacterial groups positively or negatively associated with

steady state defined communities (day 28) relative to the steady state fecal chemostat community

(day 36) are shown for higher-level (~phylum/class) level (Table 6.6) and genus-like level (Table

6.7) comparisons. Table A6.1 lists the species that the HITChip microarray has been validated to

detect for each genus-like group shown in Table 6.7. Even though differences are observed at the

genus-like level, overall comparisons at the phylum/class level show more similar patterns of

community structure.

As expected (5, 6, 25, 26), the fecal inoculum was dominated by four phyla accounting for

99.7% of the microbial diversity: Firmicutes (62.8%), Bacteroidetes (32.0%), Actinobacteria

(4.4%) and Proteobacteria (0.6%) (Table 6.5). Following the establishment of steady state

compositions VF1, VD1, and VM1 chemostat communities were still dominated by these four

phyla. For VF1 day 36, Firmicutes (47.1%) and Bacteroidetes (47.9%) were the most abundant,

162

Table 6.4 Ecological parameters (diversity, evenness, and richness) for VF1, VD1, and VM1 inocula (day 0) and steady state

communities (VF1 = day 36, VD1 and VM1 = day 28).

Ecological

Parameter

By DGGE By HITChip

Inocula Steady State Communities Inocula Steady State Communities

VF1

Day 0

VD1

Day 0

VM1

Day 0

VF1

Day 36

VD1

Day 28

VM1

Day 28

VF1

Day 0

VD1

Day 0

VM1

Day 0

VF1

Day 36

VD1

Day 28

VM1

Day 28

Shannon

Diversity

Index (H’) 3.22 3.43 3.36 3.35 3.17 2.98 5.71 5.35 5.30 5.47 5.18 4.92

Shannon

Equitability

Index (EH’) 0.78 0.83 0.81 0.80 0.76 0.72 0.82 0.79 0.80 0.80 0.78 0.77

Richness

(S) 60 62 62 66 64 61 868 658 581 688 537 465

163

Figure 6.7 Relative abundance of higher taxonomic groups (~ phylum/class level) based on HITChip analysis VF1, VD1, and

VM1 inocula (day 0) and steady state communities (VF1 = day 36, VD1 and VM1 = day 28).

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

VF1 Day 0 VD1 Day 0 VM1 Day 0 VF1 Day 36 VD1 Day 28 VM1 Day 28

Rel

ati

ve

Con

trib

uti

on

Actinobacteria

Bacteroidetes

Bacilli

Clostridium cluster I

Clostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XI

Clostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XV

Clostridium cluster XVI

Clostridium cluster XVII

Clostridium cluster XVIII

Uncultured Clostridiales

Cyanobacteria

Fusobacteria

Proteobacteria

Spirochaetes

Asteroleplasma

Uncultured Mollicutes

Verrucomicrobia

164

Table 6.5 Percent abundance of higher taxonomic groups (~ phylum/class level) based on

HITChip analysis for VF1, VD1, and VM1 inocula (day 0) and steady state chemostat

communities (VF1 = day 36, VD1 and VM1 = day 28). Percent abundances in lighter grey-

shaded cells represent values for higher taxonomic groups present in the undefined fecal

communities. Percent abundances in darker grey-shaded cells represent values for higher

taxonomic groups used to compose defined communities.

Higher taxonomic groups

Inocula Steady State Communities

VF1

Day 0

VD1

Day 0

VM1

Day 0

VF1

Day 36

VD1

Day 28

VM1

Day 28

Actinobacteria 4.36 4.38 4.32 0.39 0.54 1.05

Bacteroidetes 31.98 28.65 14.13 47.93 67.71 54.63

Firmicutes

Bacilli 3.86 1.95 3.06 0.59 0.65 1.39

Clostridium cluster I 0.12 0.25 0.30 0.18 0.19 0.36

Clostridium cluster III 0.14 0.06 0.07 0.05 0.05 0.09

Clostridium cluster IV 15.49 7.95 13.45 10.92 15.00 16.05

Clostridium cluster IX 0.25 0.35 1.88 1.69 3.04 1.82

Clostridium cluster XI 0.17 0.28 0.36 0.21 0.23 0.44

Clostridium cluster XIII 0.02 0.05 0.06 0.03 0.04 0.07

Clostridium cluster XIVa 42.14 34.78 53.93 32.45 10.03 19.77

Clostridium cluster XV 0.03 7.21 4.08 0.31 0.41 0.75

Clostridium cluster XVI 0.16 0.14 0.16 0.14 0.11 0.19

Clostridium cluster XVII 0.02 0.06 0.06 0.03 0.03 0.06

Clostridium cluster XVIII 0.12 0.39 0.54 0.07 0.10 0.21

Uncultured Clostridiales 0.27 0.57 0.65 0.41 0.46 0.78

Total Firmicutes 62.79 54.04 78.59 47.07 30.34 41.97

Cyanobacteria 0.00 0.01 0.01 0.01 0.01 0.01

Fusobacteria 0.04 0.09 0.10 0.06 0.07 0.12

Proteobacteria 0.60 12.16 2.23 1.15 1.13 1.85

Spirochaetes 0.01 0.02 0.02 0.01 0.02 0.03

Tenericutes Asteroleplasma 0.00 0.01 0.01 0.01 0.01 0.01

Uncultured Mollicutes 0.07 0.14 0.18 0.10 0.12 0.22

Verrucomicrobia 0.14 0.49 0.40 3.27 0.06 0.10

165

Table 6.6 Fold differences of higher taxonomic groups (~ phylum/class level) comparing

defined steady state communities (VD1 day 28 or VM1 day 28) to a steady state chemostat

community seeded with feces (VF1 day 36). The fold differences (defined steady state

community/VF1 day 36) marked in green are >2.5 (higher in defined steady state samples) and

marked in red are <0.5 (lower in defined steady state samples). Fold differences in light grey-

shaded cells represent values for higher taxonomic groups used to compose defined

communities.

Higher taxonomic groups VD1 Day 28

/VF1 Day 36

VM1 Day 28

/VF1 Day 36

Actinobacteria 1.38 2.69

Bacteroidetes 1.41 1.14

Firmicutes

Bacilli 1.10 2.37

Clostridium cluster I 1.11 2.05

Clostridium cluster III 1.05 1.92

Clostridium cluster IV 1.37 1.47

Clostridium cluster IX 1.80 1.07

Clostridium cluster XI 1.08 2.06

Clostridium cluster XIII 1.13 2.02

Clostridium cluster XIVa 0.31 0.61

Clostridium cluster XV 1.33 2.41

Clostridium cluster XVI 0.77 1.36

Clostridium cluster XVII 1.16 2.21

Clostridium cluster XVIII 1.50 3.04

Uncultured Clostridiales 1.12 1.90

Total Firmicutes 0.64 0.89

Cyanobacteria 1.14 2.01

Fusobacteria 1.10 1.97

Proteobacteria 0.99 1.61

Spirochaetes 1.11 2.02

Tenericutes Asteroleplasma 1.17 2.04

Uncultured Mollicutes 1.16 2.18

Verrucomicrobia 0.02 0.03

166

Table 6.7 Percent abundance of major genus-like level bacterial groups (present at an abundance >0.5% in at least 1 sample) and

fold differences (defined steady state community/VF1 day 36). Genus-like levels are only shown where the fold differences were

>2.5 (higher in defined steady state samples, marked in green) or <0.5 (lower in defined steady state samples, marked in red).

Higher taxonomic groups Level 2

(genus-like level)

Percent abundance Fold difference

VF1

Day 0

VD1

Day 0

VM1

Day 0

VF1

Day 36

VD1

Day 28

VM1

Day 28

VD1 Day 28

/VF1 Day 36

VM1 Day 28

/VF1 Day 36

Actinobacteria Bifidobacterium 4.19 3.79 3.85 0.20 0.23 0.60 1.15 3.01

Bacteroidetes

Alistipes et rel. 0.32 0.56 0.73 0.70 0.94 1.80 1.35 2.57

Bacteroides ovatus et rel. 1.11 1.03 1.75 5.31 3.10 19.97 0.58 3.76

Bacteroides splachnicus et rel. 0.22 0.39 0.46 0.38 0.47 1.13 1.23 2.95

Bacteroides stercoris et rel. 0.11 0.15 0.16 0.18 0.24 0.54 1.31 3.00

Bacteroides uniformis et rel. 0.06 1.13 0.04 2.08 2.83 0.12 1.36 0.06

Bacteroides vulgatus et rel. 28.28 12.74 0.63 25.31 41.00 3.35 1.62 0.13

Parabacteroides distasonis et rel. 0.46 6.28 7.92 1.96 2.17 14.34 1.11 7.30

Firmicutes

Bacilli Streptococcus bovis et rel. 2.59 0.19 0.38 0.10 0.11 0.38 1.09 3.67

Clostridium

cluster IV

Faecalibacterium prausnitzii et rel. 10.92 0.38 10.36 4.12 0.26 12.35 0.06 3.00

Oscillospira guillermondii et rel. 0.14 0.31 0.32 1.61 0.25 0.36 0.15 0.23

Papillibacter cinnamivorans et rel. 1.75 1.99 0.16 0.09 0.25 0.17 2.84 1.92

Ruminococcus callidus et rel. 0.62 0.14 0.99 0.22 0.12 1.33 0.57 6.09

Subdoligranulum variable at rel. 0.99 1.76 0.21 3.34 11.50 0.25 3.44 0.08

Clostridium

cluster IX

Dialister 0.03 0.06 0.07 1.27 2.10 0.09 1.65 0.07

Phascolarctobacterium faecium et rel. 0.07 0.08 1.52 0.22 0.75 1.39 3.46 6.40

Clostridium

cluster XIVa

Anaerostipes caccae et rel. 4.77 0.66 0.35 0.53 0.23 0.33 0.43 0.62

Butyrivibrio crossotus et rel. 1.16 0.77 0.64 5.00 0.24 0.37 0.05 0.07

Clostridium symbiosum et rel. 5.47 3.31 5.42 9.05 3.81 7.87 0.42 0.87

Eubacterium hallii et rel. 1.46 0.36 1.10 0.08 0.06 0.24 0.68 2.85

Eubacterium rectale et rel. 2.07 0.67 1.58 0.87 0.25 1.77 0.28 2.04

Lachnobacillus bovis et rel. 0.43 0.41 0.58 12.16 0.18 0.37 0.02 0.03

Lachnospira pectinoschiza et rel. 6.59 0.89 3.77 0.65 0.48 2.07 0.74 3.18

Outgrouping Clostridium cluster XIVa 5.68 7.60 13.17 0.64 0.31 0.82 0.48 1.29

Roseburia intestinalis et rel. 1.01 0.81 2.26 0.13 0.72 0.86 5.54 6.60

Proteobacteria Enterobacter aerogenes et rel. 0.09 2.92 0.54 0.19 0.09 0.28 0.49 1.45

Verrucomicrobia Akkermansia 0.14 0.49 0.40 3.27 0.06 0.10 0.02 0.03

167

and Actinobacteria (0.4%) and Proteobacteria (1.1%) were present at lower levels. For VD1 day

28, Firmicutes (30.3%) and Bacteroidetes (67.7%) were the most abundant, and Actinobacteria

(0.5%) and Proteobacteria (1.1%) were present at lower levels. For VM1 day 28, Firmicutes

(42.0%) and Bacteroidetes (54.6%) were the most abundant, and Actinobacteria (1.1%) and

Proteobacteria (1.8%) were present at lower levels.

Phylum/class level comparisons showed that VD1 day 28 was comparable to VF1 day 36,

but was negatively associated with Clostridium cluster XIVa and Verrucomicrobia (Table 6.6).

Although VM1 day 28 showed more differences than VD1 day 28, it was still comparable to

VF1 day 36. VM1 day 28 was positively associated with Actinobacteria and Clostridium cluster

XVIII and negatively associated with Verrucomicrobia relative to VF1 day 36 (Table 6.6).

6.3 Discussion

Simplified communities with known compositions provide a more easily studied

alternative to complex, undefined fecal communities for chemostat studies characterizing distal

gut microbial ecology. However, these simplified communities require characterization before

they can be used for experimental purposes to ensure that broad ecological measures are not

compromised by the reduced diversity. In our study, we compared simplified (defined)

chemostat communities to complex, undefined fecal communities derived from the same donor.

We were able to show that successful technical replicates could be obtained using both fecal and

defined communities, as demonstrated by the reproducible steady state communities developed

by twin-vessels inoculated with DEC or feces. Vessels seeded with feces, DEC and MET-1 were

all able to establish steady state compositions, however defined communities (DEC and MET-1)

reached steady state more quickly than fecal chemostat communities. Upon the establishment of

168

steady state compositions, vessels seeded with feces or DEC could maintain stable diversity

(including richness and evenness), while the MET-1 community diversity and evenness slightly

decreased between days 18-28 (although richness remained stable). Although DEC and MET-1

communities differed at the genus-like level, overall the steady state communities from all three

vessels showed biologically relevant phylum-level compositions that were comparable to the

fecal inoculum and steady state chemostat communities. We therefore conclude that simplified,

defined communities provide a promising alternative to fecal communities for studies of the

human distal gut ecosystem.

The grouping of VF1, VD1, and VM1 inocula in the clustering trees and PCA plot

depended upon the taxonomic level of analysis (Figures 6.5, 6.6). The HITChip clustering tree

(calculated at the oligonucleotide-level) showed that the VD1 and VM1 inocula clustered more

closely with each other than with the VF1 inoculum (Figure 6.5b). The PCA plot (calculated at

the genus-like level) showed that VF1 and VD1 inocula clustered more closely with each other

than with the VM1 inoculum. Calculations at the oligonucleotide-level (using all the probes in

the chip), analyze samples at the probe level, meaning groupings are more precise. However,

oligonucleotide-level PCA plots are less informative regarding the bacterial associations with

samples as they only illustrate probe associations (e.g. probe HIT1018). Grouping of the data at

the genus-like level allows us to more clearly define bacterial associations (e.g. Roseburia

intestinalis et rel.), but because higher-level groupings are used to calculate these plots, the

distribution of the samples in the PCA plot might differ. It should be noted that the VF1 and

VD1 steady state communities clustered more closely with each other than with the VM1 steady

state community in both the clustering trees and the PCA plot.

The ratios of each bacterial isolate used to construct the defined communities did not

169

appear to affect the structure of the resulting steady state chemostat communities. The DEC

inoculum consisted of approximately equal ratios of each microorganism (with the exception of a

few slow growing isolates where less biomass was introduced compared to faster growing

strains). DEC and fecal steady state communities became more similar to each other following

inoculation (at the oligonucleotide-level). In contrast, the MET-1 inoculum consisted of

members added at ratios approximating those noted in human stool samples (104). MET-1 and

fecal steady state communities did not become more similar to each other following inoculation

(at the oligonucleotide-level). Previous experimental results from our laboratory suggested that

the relative population abundances within the defined community inoculum did not affect the

structure of steady state communities (Schroeter, unpublished data). Inoculation of a chemostat

vessel with MET-1 and a reciprocal MET-1 community (where microbial isolates added to

MET-1 at higher abundances were added to reciprocal MET-1 at lower abundances and vice

versa), resulted in similar steady state compositions. These findings are consistent with in vivo

studies, as Becker et al. (2011) showed that inoculation of germ-free rats with equal ratios of 7 or

8 bacterial species isolated from humans resulted in comparable single strain population numbers

in the feces of rats compared to those found in human feces (25).

Defined communities must be able to maintain reproducible, steady state compositions in

order to ensure that modification of the composition or function of a chemostat community is

due to experimentation rather than adaption to the chemostat environment (2). However, because

previous reports in micro- and macroecology have shown that stability is increased with

increases in community diversity, it is important to ensure that the lower diversity of DEC and

MET-1 chemostat communities does not compromise their stability (295-299). In our study we

found that both defined and fecal communities were able to develop reproducible, stable

170

communities. Compared to chemostat vessels seeded with feces, defined communities reached

steady state compositions in a shorter time period. Few studies have evaluated these parameters

using defined communities, although Becker et al. (2011) found that germ-free rats inoculated

with a simplified human intestinal microbiota (SIHUMI, consisting of 7 or 8 bacterial species)

developed gut microbial communities that were reproducible and stable over time (25). The high

reproducibility and rapid stabilization of defined communities as well as the lack of ethical

approval required to obtain an inoculum makes them a more attractive alternative to feces for use

as a simulated human gut microbial ecosystem.

Chemostat communities seeded with feces or DEC maintained stable diversity, richness,

and evenness values upon the establishment of steady state compositions. However, while

chemostat communities seeded with MET-1 maintained stable richness values, evenness (and

therefore diversity) values gradually decreased over time. In spite of this, decreases in evenness

and diversity values did not compromise the stability of MET-1 chemostat communities,

although these observations do suggest that defined communities should contain as much

cultivable diversity as possible to maintain balanced compositions. This is especially important if

the defined communities are exposed to experimental perturbations, as functional redundancy is

higher in more complex communities (300).

Because the inocula used for VF1, VD1, and VM1 were composed of different

microorganisms (both in richness and abundance), it cannot be expected that steady state

communities would share identical compositions at the genus-like level. Indeed, at the

oligonucleotide-level and genus-like level communities seeded with feces, DEC or MET-1 had

distinct compositions (Table 6.3 and 6.7). Instead, the aim of this study was to develop defined

communities that were able to maintain the overall characteristics of a fecal chemostat

171

community. Broader comparisons indicated that VF1, VD1 and VM1 chemostat communities

had comparable phylum/class-level abundances (Figure 6.7, Table 6.6). During steady state

conditions, VF1 community structure more closely resembled the VD1 community than theVM1

community, stressing the importance of including as many strains as possible when designing

defined communities. Like VF1 day 0 and VF1 day 36, VD1 day 28 and VM1 day 28

communities were dominated by Bacteroidetes and Firmicutes, with less Proteobacteria and

Actinobacteria. These results are consistent with a previous study conducted by Trosvik et al.

(2010). In this work, a chemostat system was inoculated with a defined microbial community

consisting of four bacterial species, chosen to represent the four main phyla that colonize the

human gastrointestinal tract (154). These species included Bacteroides thetaiotaomicron

(representing Bacteriodetes), Clostridium perfringens (representing Firmicutes), Escherichia coli

(representing Proteobacteria) and Bifidobacterium longum (representing Actinobacteria; 154).

As the community developed, the abundances of these four species gradually mimicked the

phylum-level abundances noted in vivo, with Bacteriodetes and Firmicutes dominating the

community and Proteobacteria and Actinobacteria present at lower levels (154). Inter-microbe

interactions as well as the chemostat environment (e.g. composition of the growth medium, set to

mimic the human gut) may be responsible for the maintenance of these phylum-level ratios

(154).

In this study it was sometimes difficult to correlate genus-like groups with specific

members of the defined communities using HITChip data. HITChip analysis of defined

communities detected positive hits (hybridized probe sequences) for genus-like groups (sequence

similarity ≥ 90%; 56) that did not precisely correspond to the identities of microorganisms that

were used to formulate the defined community inocula. Cross-hybridization with non-target

172

sequences and differences in probe binding affinities might result in the classification of genus-

like groups as closely related taxa, especially when the specific taxa are not represented on the

microarray (64). However, genus-like groups detected in the defined communities generally

corresponded to the higher taxonomic groups that were included in the defined community

formulations.

HITChip analysis of defined communities also detected low levels of positive hits for

higher taxonomic groups (phylum/class levels) that did not correspond to microorganisms that

were used to formulate the defined community inocula. Cross-hybridization with non-target

sequences and differences in probe binding affinities might also explain the low levels of

"background noise" that correspond to false positives for higher taxonomic groups (64).

Additional HITChip hybridizations for specific cases with lower diversity than fecal samples

(e.g. defined communities) may need to use stricter stringencies to improve the specificity of the

array.

Higher taxonomic level comparisons showed that VD1 day 28 was comparable to VF1 day

36, but was negatively associated with Clostridium cluster XIVa and Verrucomicrobia. VM1 day

28 showed more differences than VD1 day 28, but it was still comparable to VF1 day 36. VM1

day 28 was positively associated with Actinobacteria and Clostridium cluster XVIII and

negatively associated with Verrucomicrobia. It should be noted that the defined communities did

not contain members of the Verrucomicrobia phylum, and thus detected levels of this phylum

may correspond to background noise (Verrucomicrobia abundance = 0.06% in VD1 day 28 and

0.10% in VM1 day 28). Subsequent defined community designs should incorporate

representatives of Verrucomicrobia into their formulations to more accurately mimic the fecal

chemostat community composition. Although fold difference values indicated that members

173

from Clostridium cluster XVIII were higher in VM1 day 28 compared to VF1 day 36, the

percent abundance of this group was low in VM1 and may also correspond to background noise

(Clostridium cluster XVIII abundance = 0.21% in VM1 day 28). Likewise, despite the fact that

the abundance of Actinobacteria is higher in VM1 day 28 compared to VF1 day 36, the levels of

this phylum are relatively low in both communities (Actinobacteria abundance = 1.05% in VM1

day 28 and 0.39% in VF1 day 36).

A drawback of the HITChip is that it can only detect taxa which are covered by the

reference sequences (301). In our study Clostridium cocleatum (Clostridium cluster VIII) and

Clostridium lactatifermentans (Clostridium cluster XIVb) were included in defined community

formulations, however they were not specifically targeted by HITChip probe sequences. Future

studies will determine whether these microorganisms could be detected by cross-hybridizations

with probe sequences already present on the HITChip microarray. However, subsequent

microarray designs should incorporate representatives of Clostridium clusters VIII and XVIb

into their set of targeted sequences.

Analysis of fecal communities comparing HITChip and pyrosequencing data showed that

HITChip hybridizations and community profiles correspond well with pyrosequencing data

(301). However, to date there has been no work to compare HITChip data with next-generation

sequencing data from defined communities. Thus, future studies should include an analysis of

archived DNA samples from VF1, VD1, and VM1 using a suitable next-generation sequencing

platform. 16S rDNA profiling data from these samples will allow us to analyze our samples in

more detail (species-level comparisons) and will further validate the HITChip microarray.

In spite of their advantages over fecal communities, the construction of defined

communities provides several challenges. Firstly, obligatory symbiotic relationships between gut

174

microbes can make it difficult to culture some microorganisms axenically on agar plates, as what

would appear to be a single colony consisting of one microbial population could actually be a

mixture of two or more microorganisms. Also, passaging of fecal microorganisms on laboratory

growth media may cause isolates to become lab-adapted, resulting in microbial behaviour that is

not characteristic of the behaviour exhibited in vivo (e.g. laboratory reference strains, such as

Escherichia coli strain K12, Pseudomonas aeruginosa strain PAO1 and Staphylococcus aureus

strain COL; 302). However, these challenges are approachable (for example, novel approaches to

culture of the human gut microbiota species are allowing for growth of previously unculturable

organisms) and may be compensated for by the many benefits of using defined communities for

distal gut ecological research.

In conclusion, chemostat vessels seeded with simplified, defined communities can develop

reproducible, stable communities that are generally representative of the distal gut microbiota

community from which they are derived. As a consequence, this opens up many possibilities to

more easily examine microbial interactions of a synthetic fecal community at the molecular level

in a highly controllable manner. As novel culturing techniques facilitate the growth of fastidious

microorganisms, the number of microbes included in preparations of defined communities can be

increased. In addition, formulations of defined communities derived from the stools of diseased

donors (for example, IBD patients) may offer a new approach to the study of gut microbial

dysbiosis. The functionality, resistance and resilience responses of defined communities and the

fecal communities from which they are derived should be carefully compared before any defined

community can be endorsed as a biologically relevant model. The standardization of human

microbial ecology research using innovations such as defined communities may allow

researchers to normalize gut microbiota studies in laboratories worldwide.

175

Chapter 7

Assessing interactions between gut

communities from different donors in an

in vitro model of the human distal colon

7.1 Introduction

A major advantage for in vitro gut models is their relative lack of ethical restraints

regarding the experimental manipulation of simulated gut communities (1). In addition to

observational studies, experimental manipulation of gut communities can guide the design of

new therapeutic strategies based on their response to relevant stimuli (137).

Depending on the stimulus applied, disruption of the gut microbiota can be transient or

prolonged. Probiotics and prebiotics only appear to cause transient effects on host physiology

and don't appear to result in long-term, persistent alterations of the gut microbiota (137). As

previously mentioned, antibiotic exposure can cause major short-term disruptions of the gut

microbiota (137). Most gut microbiota experience at least partial taxonomic recovery following

antibiotic treatment, however exposure can shift the gut microbiota to a new, altered steady state

(24). Persistent alterations of the gut microbiota following perturbation suggest experimental

manipulation to benefit host health is a viable option (137).

Some diseases can cause major changes in the structure of gut communities that persist

over time. For example, recurrent CDI has been associated with the collapse of gut microbial

ecosystems where patients cannot recover (or partially recover) their indigenous gut microbiota

following exposure to antibiotics (102, 141, 303). Traditional treatment options for CDI are

176

limited, and as such new approaches are currently under development (103, 104, 304).

Fecal bacteriotherapy ('stool transplants') involves administering stool from a healthy

donor to a CDI patient to restore the microbial diversity of the gut (103). This approach has had a

high rate of success at treating CDI, as indicated in both case studies (92% success rate; 305) and

a recent clinical trial (94% success rate; 103). Despite its high effectiveness there are several

limitations to stool transplants (150). Although fecal donors undergo an extensive health screen

to ensure potential pathogens are not transferred to the patient, there is no guarantee the donor

will not become infected with a pathogen in the window between the health screen and donation

(150). In addition, the donor may not show any symptoms of disease but they may carry

opportunistic pathogens that can cause disease in the recipient (especially if the patient is

immunocompromised; 150). Other drawbacks to stool transplants include minimal

reproducibility, low controllability following infusion (not easily monitored or removed), limited

stability (must be processed and administered within a few hours of the donation), and poor

patient acceptability (unpleasant procedure; 150).

An alternative to stool transplants is microbial ecosystem therapeutics (MET, 'synthetic

stool'). MET consists of defined microbial communities comprised of bacterial species isolated

from the feces of individual healthy donors (104). Because these strains have already interacted

and co-evolved with each other in vivo, it is expected that they can co-exist without deleterious

inter-microbe interactions (150). However, it is also possible that they could provide the host

with a synergistic beneficial effect to promote quicker and more complete recovery from

dysbiosis (150). Despite the fact that MET is still in its early developmental stages compared to

stool transplants, preliminary proof-of-principle studies have shown that the prototype MET

formulation, MET-1, was able to resolve disease in two patients with severe recurrent CDI (104).

177

Even though MET-1 was only administered to each patient on one occasion, many members of

this introduced community persisted in the patients' gut microbiota for at least 6 months

following treatment (104). Other advantages of MET include moderate reproducibility

(potentially improved when grown in a chemostat, see chapter 6), very good safety (defined

microbial composition), excellent controllability (can be monitored and if necessary removed

using specific antibiotics), excellent stability (can be preserved in the same manner as other

probiotics, e.g. through lyophilisation techniques), and very good patient acceptability (150).

We hypothesized that we could manipulate gut communities established in a chemostat

model to increase the retention of allochthonous (non-resident) microorganisms by first

disrupting them with antibiotics. The aims of this study were thus to i) to determine whether

antibiotic perturbation would aid in the establishment of donor B-derived synthetic stool (MET-

1) in chemostat communities seeded with donor A feces, and ii) to determine whether the mixture

of two different gut communities results in a donor A-like community, donor B-like community

or a hybrid community.

7.2 Results

7.2.1 Effect of clindamycin treatment and MET-1 supplementation on healthy fecal chemostat

communities

In the first chemostat run, 4 identical vessels were inoculated with donor A feces and

allowed to operate without experimental perturbation until 36 days post-inoculation. Between

days 36-41 increasing doses of clindamycin were added to V( /+) and V( /-) and an equal

volume of water was added to V( /+) and V( /-) (as a control). On day 42 (after one full

changeover of media within each vessel), a single dose of MET-1 was added to V(+/ ) and

178

V(-/ ) and an equal volume of saline was added to V(+/ ) and V(-/ ) (as a control). Communities

were maintained in these four vessels until 48 days post-inoculation. These four vessels were

analyzed on the first series of DGGE gels (Figure 7.1).

All four vessels had very similar starting communities, with day 0 DGGE profiles sharing

93.4% similarity (Table 7.1). Steady state was achieved between days 34-36, where V(+/-), V(-

/+) and V(-/-) shared similar community profiles (Figure 7.2a, Table 7.1). Between days 32-36,

V(+/+) was less similar to the other three vessels with correlation coefficient values below the

gel-defined cut-offs.

V(+/+) and V(+/-) profiles became more similar to each other following clindamycin

treatment, where correlation coefficient values rose above the cut-off threshold on days 40 and

42 (95.6% similar on day 42, Figure 7.2a, Table 7.1). V(-/+) and V(-/-) profiles remained similar

while receiving water dosage between days 36-41 (93.8% similar on day 42). On average,

clindamycin-treated vessels (V(+/+) and V(+/-)) were 33.6% similar to vessels not receiving

clindamycin treatment (V(-/+) and V(-/-)) on day 42 (before MET-1 addition).

Immediately after sampling on day 42 (day 42 (pre) samples), donor B-derived MET-1 was

added to V(+/+) and V(-/+) and saline was added to V(+/-) and V(-/-) (day 42 (post) samples).

Following MET-1/saline addition, day 42 (pre) and day 42 (post) samples were 75.5% similar for

V(+/+), 94.6% similar for V(-/+), 91.1% similar for V(+/-) and 96.3% similar for V(-/-).

By DGGE, the addition of MET-1 to the donor A clindamycin- and water-treated vessels

did not appear to have a lasting impact on the community composition. Following the

clindamycin treatment period and MET-1 addition, V(+/+) and V(+/-) profiles rapidly increased

in similarity to each other, increasing above the gel-defined cut-off between days 46 and 48

(90.9% similar on day 48, Figure 7.2a). Although both V(+/+) and V(+/-) rapidly increased in

179

Figure 7.1 DGGE profiles comparing chemostat communities seeded with feces from donor A.

a) immediately following inoculation on day 0, b) during steady state conditions on day 36, c) 1

day following the end of the clindamycin treatment/water addition period and just prior to MET-

1/saline addition on day 42, d) MET-1 inoculum, e) immediately following MET-1/saline

addition on day 42, and f) on the last day of the experiment on day 48.

180

V(+

/-)

Da

y 4

2 (

po

st)

V(+

/+)

Da

y 4

2 (

po

st)

V(-

/-)

Da

y 0

V(-

/+)

Da

y 0

V(+

/-)

Da

y 0

V(+

/+)

Da

y 0

V(-

/-)

Da

y 3

6

V(-

/+)

Da

y 3

6

V(+

/-)

Da

y 3

6

V(+

/+)

Da

y 3

6

V(-

/-)

Da

y 4

2 (

pre

)

V(-

/+)

Da

y 4

2 (

pre

)

V(+

/-)

Da

y 4

2 (

pre

)

V(+

/+)

Da

y 4

2 (

pre

)

V(-

/-)

Da

y 4

2 (

po

st)

V(-

/+)

Da

y 4

2 (

po

st)

V(-

/-)

Da

y 4

8

V(-

/+)

Da

y 4

8

V(+

/-)

Da

y 4

8

V(+

/+)

Da

y 4

8

ME

T-1

a b c d e f

181

Table 7.1 Correlation coefficients (%SI) comparing chemostat communities seeded with donor

A feces immediately following inoculation (day 0), during steady state conditions (day 36), 1 day

following clindamycin treatment prior to MET-1 addition (day 42 (pre)), 1 day following

clindamycin treatment and MET-1 addition (day 42 (post)),and at the end of the experiment (day

48).

Day V(+/+) vs. V(+/-) V(+/+ ) vs. V(-/-) V(+/-) vs. V(-/-) V(-/+) vs. V(-/-)

0 96.8 89.8 90.0 93.4

36 76.5 72.6 91.3 96.9

42 (pre-MET-1) 95.6 37.6 28.2 93.8

42 (post-MET-1) 78.6 47.5 41.3 93.7

48 90.9 67.6 80.5 94.9

182

Figure 7.2 DGGE analysis comparing chemostat communities seeded with donor A feces during

a clindamycin or water addition period followed by a single dose of MET-1or saline. a)

Correlation coefficients (expressed as percentages) comparing the profiles of each vessel at

identical time points. b) Community dynamics calculated using moving window correlation

analysis. c) Adjusted Shannon Diversity Index (H’). d) Adjusted band richness (S). e) Adjusted

Shannon equitability index (EH’).

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36 40 44 48

%S

I

Day

V(+/+) vs. V(+/-) V(+/+) vs. V(-/-) V(+/-) vs. V(-/-) V(+/+) vs. V(-/+) V(-/+) vs. V(-/-) Began adding C Finished adding C Added MET-1

0

10

20

30

40

50

60

70

80

90

100

2 6 10 14 18 22 26 30 34 38 42 46

%S

I

Day (x-2) vs. (x)

V(+/+) V(+/-) V(-/+) V(-/-) Began adding C Finished adding C Added MET-1

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36 40 44 48

H'

Day

V(+/+) V(+/-) V(-/+) V(-/-) Began adding C Finished adding C Added MET-1

0

10

20

30

40

50

60

70

0 4 8 12 16 20 24 28 32 36 40 44 48

S

Day

V(+/+) V(+/-) V(-/+) V(-/-) Began adding C Finished adding C Added MET-1

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36 40 44 48

EH

'

Day

V(+/+) V(+/-) V(-/+) V(-/-) Began adding C Finished adding C Added MET-1

a b

c d

e

183

similarity to V(-/-), V(+/-) recovered to within 5% of the gel-defined cut-off threshold by day 48

while V(+/+) was just below the 5% of the gel-defined cut-off threshold. Following the addition

of MET-1 to V(-/+), V(-/+) and V(-/-) correlation coefficients remained above the gel-defined

cut-off thresholds (94.9% similar on day 48).

Clustering tree analysis showed that day 0 samples clustered together and day 36 samples

clustered together, with V(+/+) clustering farther out from the other day 36 samples (Figure 7.3).

Prior to MET-1 addition on day 42, samples from clindamycin treated vessels clustered

separately from non-clindamycin treated vessels, with V(+/+) and V(+/-) profiles clustering

together and V(-/+) and V(-/-) profiles clustering with day 36 samples (not exposed to

clindamycin). Following MET-1 addition, V(+/+) day 42 (post) clustered with other clindamycin

treated vessels, however it did not cluster as closely to V(+/+) day 42 (pre), V(+/-) day 42 (pre),

and V(+/-) day 42 (post). V(-/+) day 42 (post) clustered closely with non-clindamycin treated

vessels. At the end of the experiment, V(+/+) and V(+/-) day 48 samples clustered more closely

with samples from non-clindamycin treated vessels. V(-/+) and V(-/-) samples from day 48 also

clustered with samples from non-clindamycin treated vessels.

All four vessels had comparable rate of change profiles between days 0-36, with higher

values at the beginning of the experiment (days 0-14) which decreased as the communities

developed (days 14-32) until steady state was achieved (days 32-36, Figure 7.2b). Large changes

in community dynamics were seen in the vessels that received clindamycin treatment. V(+/+)

and V(+/-) rate of change values increased between days 36-42 (V(+/+) Δt = 62.0% between

days 36-38, V(+/-) Δt = 59.3% between days 40-42). Between days 36-42 V(-/+) and V(-/-) rate

of change values fluctuated slightly, but not to the degree noted in V(+/+) and V(+/-) (average

V(-/+) Δt = 16.3%, V(-/-) Δt = 10.6%). Following MET-1 addition, V(+/+) showed another peak

184

Figure 7.3 Dendrograms comparing chemostat communities seeded with donor A feces during a

clindamycin or water addition period followed by a single dose of MET-1or saline. Dendrograms

are based on Pearson and UPGMA correlation of DGGE profiles.

185

in the rate of change values between days 42-44 (Δt =51.5%) corresponding with MET-1

addition. V(+/+) and V(+/-) showed similar rate of change values between days 44-48. Between

days 46-48, V(+/+) Δt values dropped below the cut-off threshold while V(+/-) Δt values

dropped to within 5% of the cut-off threshold. Following MET-1 addition V(-/+) showed

increased Δt values between days 42-48 (average Δt= 21.4%) relative to V(-/-) (average Δt=

11.1%) and V(-/+) prior to MET-1 addition (average Δt= 16.3% between days 36-42).

All four vessels had comparable H', EH' and S values between days 0-36 (Figure 7.2c,d,e,

Table 7.2). These values fluctuated at the beginning of the experiment, but stabilized as steady

state was reached (days 34-36). Following clindamycin treatment in V(+/+) and V(+/-) there was

a reproducible reduction in community diversity, evenness, and richness. Between days 36-44

V(-/+) and V(-/-) maintained high H', EH', and S values while receiving water dosage (control).

Immediately following the addition of MET-1 to V(+/+), H, EH', and S values increased.

V(-/+) S values increased following the inoculation of MET-1, however, H' and EH' values did

not have a dramatic increase. Even though the addition of saline to V(+/-) and V(-/-) did not

impact S values, V(+/-) H' and EH' values increased slightly while V(-/-) H' and EH' values

decreased slightly.

Following the clindamycin treatment period and MET-1 addition, V(+/+) maintained stable

H' values between days 42-48. V(+/+) EH' values decreased between days 42-44 but were

maintained between days 44-48, while V(+/+) S values increased between days 42-44 and were

maintained between days 44-48. V(+/-) H', EH', and S values increased between days 42-44 and

maintained these values between days 44-48. Between days 44-48, V(+/+) H' and EH' values

were lower than V(+/-) values, however V(+/+) S values were higher than V(-/+) values.

Following MET-1 addition, V(-/+) H' values decreased slightly between days 42-44 and

186

Table 7.2 Ecological parameters (diversity, evenness, and richness) for chemostat communities seeded with donor A feces.

Samples taken immediately following inoculation (day 0), during steady state conditions (day 36), 1 day following clindamycin

treatment prior to MET-1 addition (day 42 (pre)), 1 day following clindamycin treatment and MET-1 addition (day 42 (post)),and

at the end of the experiment (day 48).

Day Shannon Diversity Index (H') Shannon Equitability Index (EH') Band Richness (S)

V(+/+) V(+/- ) V(-/+) V(-/-) V(+/+) V(+/- ) V(-/+) V(-/-) V(+/+) V(+/- ) V(-/+) V(-/-)

0 3.50 3.49 3.48 3.47 0.85 0.85 0.85 0.84 61 61 61 61

36 3.49 3.42 3.43 3.49 0.84 0.82 0.82 0.84 65 65 65 65

42 (pre-MET-1) 2.70 2.58 3.43 3.44 0.77 0.74 0.82 0.82 33 33 65 65

42 (post-MET-1) 3.28 2.56 3.51 3.39 0.81 0.73 0.83 0.81 57 33 69 65

48 3.20 3.21 3.39 3.36 0.84 0.84 0.81 0.80 46 46 67 65

187

were maintained between days 44-48. Although V(-/+) EH' values were maintained between days

42-48, S values decreased between days 42-44 and were maintained between days 44-48. V(-/+)

H' and S values were higher than V(-/-) values between days 42-44, but were comparable

between days 44-48. V(-/+) EH' values comparable to V(-/-) values between days 42-48.

7.2.2 Mixing donor A and B chemostat communities results in a community that more closely

resembled the donor A community than the donor B community

Another aim of this study was to determine whether the fecal community from one healthy

donor would be more successful at persisting in the presence of a fecal community from another

donor. If this is the case, a "dominant" donor gut microbiota may be able to better persist in the

presence of a diseased recipient's intestine, therefore making an ideal candidate for stool

transplants or novel MET formulations.

Therefore, in the second chemostat run one vessel was seeded with donor A feces (VA)

and one vessel was seeded with donor B feces (VB). At 36 days post-inoculation, a third vessel

(VAB) was inoculated with an equal volume of VA and VB chemostat cultures. All three vessels

were maintained until 44 days post inoculation (relative to the VA and VB inoculation time

point). These three vessels were analyzed on the second series of DGGE gels and day 0, 36, and

44 samples were also analyzed by HITChip (Figure 7.4).

VA and VB fecal inocula had distinct compositions at the beginning of the experiment

(Figure 7.4, 7.5, Table 7.3). Table 7.4 lists the vessel correlation coefficients comparing VA, VB,

and VAB on days 0, 36, and 44 using DGGE and HITChip. Again, %SI values were higher for

HITChip data compared to DGGE data (see chapter 8). VA and VB were able to maintain

distinct compositions over the course of the chemostat run, with vessel correlation coefficients

188

Figure 7.4 Molecular fingerprints representing communities seeded with donor A feces (VA),

donor B feces (VB) or a mixture of donor A and B steady state communities (VAB) cultured

within a twin-vessel single-stage chemostat. (a) DGGE profiles (b) HITChip profiles. HITChip

profiles also list the highest phylogenetic level of specificity of probes (level 1) on the right of

panel b.

189

a

VA

Day

0

VB

Day

0

VA

Day

36

VB

Day

36

VA

B D

ay 3

6

VA

Day

44

VB

Day

44

VA

B D

ay 4

4

VA

Day

0

VB

Day

0

VA

Day

36

VB

Day

36

VA

B D

ay 3

6

VA

Day

44

VB

Day

44

VA

B D

ay 4

4

b

Clostridium cluster XVIII

Actinobacteria

Asteroleplasma

Bacilli

Bacteroidetes

Clostridium cluster IClostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XIClostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XVClostridium cluster XVIClostridium cluster XVII

CyanobacteriaFusobacteria

Proteobacteria

Spirochaetes

Uncultured Clostridiales

Uncultured MollicutesVerrucomicrobia

190

Figure 7.5 Relative abundance of higher taxonomic groups (~ phylum/class level) based on HITChip analysis for the VA and VB

fecal inoculum (day 0) and steady state communities (VA and VB days 36 and 44) as well as VAB cultures immediately

following mixture (VAB day 36) and at the end of the experiment (VAB day 44).

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

VA Day 0 VB Day 0 VA Day 36 VAB Day 36 VB Day 36 VA Day 44 VAB Day 44 VB Day 44

Rel

ati

ve

Con

trib

uti

on

Actinobacteria

Bacteroidetes

Bacilli

Clostridium cluster I

Clostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XI

Clostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XV

Clostridium cluster XVI

Clostridium cluster XVII

Clostridium cluster XVIII

Uncultured Clostridiales

Cyanobacteria

Fusobacteria

Proteobacteria

Spirochaetes

Asteroleplasma

Uncultured Mollicutes

Verrucomicrobia

191

Table 7.3 Percent abundance of higher taxonomic groups (~ phylum/class level) based on HITChip analysis for VA, VB, and

VAB communities by DGGE and HITChip on days 0, 36, and 44.

Higher taxonomic groups VA Day 0 VB Day 0 VA Day 36 VAB Day 36 VB Day 36 VA Day 44 VAB Day 44 VB Day 44

Actinobacteria 0.30 2.04 0.27 0.31 0.34 0.31 0.28 0.28

Bacteroidetes 7.72 43.53 25.01 41.42 48.70 22.21 18.11 59.12

Firmicutes

Bacilli 0.64 6.12 0.49 0.47 0.46 0.56 0.55 0.39

Clostridium cluster I 0.20 0.13 0.15 0.15 0.14 0.16 0.16 0.12

Clostridium cluster III 0.13 0.14 0.06 0.07 0.07 0.06 0.06 0.05

Clostridium cluster IV 54.15 17.60 56.46 28.84 7.15 60.66 65.01 4.53

Clostridium cluster IX 0.24 0.25 0.22 0.89 2.89 0.25 0.26 1.33

Clostridium cluster XI 0.64 0.15 0.23 0.21 0.18 0.25 0.25 0.15

Clostridium cluster XIII 0.03 0.02 0.03 0.03 0.03 0.03 0.03 0.02

Clostridium cluster XIVa 33.41 28.49 8.67 21.21 33.54 9.51 11.57 28.69

Clostridium cluster XV 0.03 0.03 0.03 0.18 0.28 0.04 0.26 0.21

Clostridium cluster XVI 0.28 0.12 0.10 0.11 0.11 0.11 0.13 0.09

Clostridium cluster XVII 0.06 0.02 0.02 0.02 0.02 0.03 0.03 0.02

Clostridium cluster XVIII 0.07 0.15 0.06 0.06 0.06 0.06 0.06 0.05

Uncultured Clostridiales 0.77 0.28 4.14 1.83 0.33 1.06 0.61 0.30

Total Firmicutes 90.66 53.51 70.66 54.06 45.26 72.78 78.97 35.97

Cyanobacteria 0.01 0.00 0.01 0.01 0.01 0.01 0.01 0.00

Fusobacteria 0.05 0.04 0.05 0.05 0.05 0.06 0.05 0.04

Proteobacteria 0.72 0.61 1.29 1.18 1.10 1.54 1.46 0.94

Spirochaetes 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

Tenericutes Asteroleplasma 0.01 0.00 0.01 0.01 0.01 0.01 0.01 0.00

Uncultured Mollicutes 0.08 0.07 0.08 0.08 0.08 0.09 0.09 0.07

Verrucomicrobia 0.44 0.18 2.60 2.87 4.45 2.98 1.01 3.55

192

Table 7.4 Correlation coefficients (%SI) comparing VA, VB, and VAB communities by DGGE

and HITChip on days 0, 36, and 44. NA= Not applicable.

Analysis Method Day VA vs. VB VA vs. VAB VB vs. VAB

DGGE

0 29.3 NA NA

36 45.2 84.3 75.5

44 48.3 88.5 40.0

HITChip

0 67.2 NA NA

36 66.7 90.9 86.8

44 70.2 94.8 73.0

193

maintained below the gel-defined cut-off thresholds (Figure 7.6a). On day 36, an equal volume

of chemostat culture from VA and VB were mixed in a new vessel (VAB). Following mixing,

VA and VAB correlation coefficients remained above the cut-off threshold for the rest of the

experiment (days 36-44, Figure 7.6a). VB and VAB correlation coefficients were initially close

to but below the cut-off thresholds and decreased rapidly between days 36-38. Between days 40-

44 the VB vs. VAB correlation coefficient values were comparable to values observed for VA

vs. VB comparisons.

Clustering tree and PCA analysis showed that VA and VB profiles didn't cluster closely

with each other immediately following inoculation (day 0) or during steady state conditions

(days 36 and 44; Figures 7.7, 7.8). VA steady state samples from days 36 and 44 clustered

together and VB steady state samples from days 36 and 44 clustered together. Although the

clustering trees showed that the VAB sample clustered more closely with the VA sample than

the VB sample on day 36, the PCA plot showed that the VAB sample sat midway between the

VA and VB samples. On day 44 both the clustering trees and the PCA plot showed that the VAB

day 44 sample clustered more closely with the VA sample than with the VB sample.

The HITChip also provided phylogenetic information for VA, VB, and VAB on days 0, 36,

and 44 (Figure 7.5, Table 7.3). Analysis of samples at both the phylum/class level and the genus-

like level showed that the VAB community did not remain a mixture of the two communities

following mixing, but rather more closely resembled VA compositions compared to VB

compositions (Figure 7.5, Table 7.3, 7.5). Immediately following mixing on day 36 the VAB

HITChip profile contained a mixture of microorganisms that were also identified in VA and VB

profiles (Figure 7.5, Table 7.3, 7.5). On day 44 the abundances of phyla/class level groups

(Figure 7.5, Table 7.3) and genus-like groups (Table 7.5) detected in VAB were more

194

Figure 7.6 DGGE analysis of VA, VB, and VAB prior to and following the mixture of steady

state VA and VB chemostat communities. a) Correlation coefficients (expressed as percentages)

comparing the profiles of each vessel at identical time points. b) Community dynamics

calculated using moving window correlation analysis. c) Adjusted Shannon Diversity Index (H’).

d) Adjusted band richness (S). e) Adjusted Shannon equitability index (EH’).

0

10

20

30

40

50

60

70

80

90

100

0 4 8 12 16 20 24 28 32 36 40 44

%S

I

Day

VA vs. VB VA vs. VAB VB vs. VAB Mixed VA and VB Cut-off Cut-off -5%

0

10

20

30

40

50

60

70

80

90

100

2 6 10 14 18 22 26 30 34 38 42

%S

I

Day

VA VB VAB Mixed VA and VB Cut-off Cut-off +5%

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 4 8 12 16 20 24 28 32 36 40 44

H'

Day

VA VB VAB Mixed VA and VB

0

10

20

30

40

50

60

0 4 8 12 16 20 24 28 32 36 40 44

S

Day

VA VB VAB Mixed VA and VB

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 4 8 12 16 20 24 28 32 36 40 44

EH

'

Day

VA VB VAB Mixed VA and VB

a b

c d

e

195

Figure 7.7 Dendrogram comparing VA, VB, and VAB prior to and following the mixture of

steady state VA and VB chemostat communities. Dendrograms are based on (a) Pearson and

UPGMA correlation (DGGE profiles) or (b) Pearson correlation and Ward's minimum variance

method (HITChip profiles).

196

197

Figure 7.8 Principal component analysis (PCA) at the genus-like level based on HITChip data of the VA and VB fecal inocula

(VA and VB day 0) and steady state communities (VA and VB days 36 and 44), as well as VAB cultures immediately following

mixture (VAB day 36) and at the end of the experiment (VAB day 44). Arrows shown in the plot correspond to the top 30 genus-

like groups that explain most of the variability in the dataset.

-1.0 1.5

-0.6

1.0

VA inoc

VA Day 44VB Day 44

VAB Day 44

VB inoc

VA Day 36 VB Day 36

VAB Day 36

PC1: 51.5%

PC

2:

30.7

%Clostridium clusters IV and XIVa

Clostridium cluster IV and Bacteroidetes

198

Table 7.5 Percent abundance of major genus-like level bacterial groups (present at an abundance >0.5% in at least 1 sample) and

fold differences comparing VA, VB, and VAB communities on days 36 and 44. Genus-like levels are only shown where the fold

differences were >2.5 (higher in VAB Day (x), marked in green) or <0.5 (lower in VAB Day (x), marked in red).

Higher taxonomic groups Level 2

(genus-like level)

Percent abundance Fold difference

VA

Day

36

VAB

Day

36

VB

Day

36

VA

Day

44

VAB

Day

44

VB

Day

44

VAB

Day 36

VA

Day 36

VAB

Day 36

VB

Day 36

VAB

Day 44

VA

Day 44

VAB

Day 44

VB

Day 44

VAB

Day 44

VAB

Day 36

Bacteroidetes

Alistipes et rel. 1.65 1.05 0.85 2.01 3.27 0.64 0.64 1.24 1.63 5.11 3.11 Bacteroides fragilis et rel. 3.56 8.12 6.97 3.28 2.58 7.49 2.28 1.17 0.79 0.34 0.32

Bacteroides intestinalis et rel. 1.13 0.56 0.07 1.30 1.32 0.05 0.49 7.59 1.02 28.58 2.36

Bacteroides ovatus et rel. 2.44 4.95 7.35 2.31 2.18 5.99 2.03 0.67 0.94 0.36 0.44 Bacteroides plebeius et rel. 0.75 0.42 0.20 0.91 0.75 0.12 0.56 2.09 0.83 6.47 1.80

Bacteroides uniformis et rel. 5.25 2.59 1.24 4.78 2.96 1.66 0.49 2.10 0.62 1.78 1.14

Bacteroides vulgatus et rel. 7.46 20.09 26.32 4.89 2.47 38.70 2.69 0.76 0.51 0.06 0.12 Parabacteroides distasonis et rel. 0.64 1.94 4.05 0.64 0.79 2.75 3.02 0.48 1.23 0.29 0.41

Prevotella tannerae et rel. 0.28 0.40 0.42 0.24 0.18 0.63 1.42 0.96 0.75 0.28 0.45

Firmicutes

Clostridium

cluster IV

Anaerotruncus colihominis et rel. 0.32 0.25 0.11 0.51 0.47 0.09 0.78 2.24 0.92 4.96 1.84

Clostridium cellulosi et rel. 0.47 0.32 0.13 0.62 0.67 0.11 0.69 2.50 1.08 5.93 2.09

Clostridium leptum et rel. 0.86 0.56 0.11 1.38 1.55 0.10 0.64 5.28 1.12 16.21 2.79

Faecalibacterium prausnitzii et rel. 3.37 1.30 0.40 3.65 3.35 0.41 0.39 3.27 0.92 8.14 2.56

Oscillospira guillermondii et rel. 10.54 6.82 2.34 13.04 7.58 1.64 0.65 2.92 0.58 4.62 1.11

Ruminococcus bromii et rel. 0.79 0.57 0.02 2.32 2.08 0.03 0.72 24.48 0.90 64.09 3.65

Ruminococcus callidus et rel. 0.28 0.19 0.10 0.29 0.36 0.09 0.69 1.99 1.27 4.12 1.87

Sporobacter termitidis et rel. 2.22 1.34 0.37 2.81 1.74 0.31 0.60 3.58 0.62 5.61 1.30

Subdoligranulum variable at rel. 36.45 16.32 2.60 34.51 45.93 0.85 0.45 6.27 1.33 54.12 2.81

Clostridium cluster IX

Dialister 0.04 0.66 2.56 0.04 0.04 1.09 18.64 0.26 0.98 0.04 0.06

Clostridium

cluster XIVa

Butyrivibrio crossotus et rel. 0.84 2.94 4.93 0.94 0.68 5.07 3.50 0.60 0.72 0.13 0.23

Eubacterium rectale et rel. 0.17 0.47 0.61 0.33 0.46 0.59 2.76 0.78 1.38 0.79 0.99

Lachnobacillus bovis et rel. 0.13 6.11 12.18 0.16 0.21 10.56 46.93 0.50 1.31 0.02 0.04

Ruminococcus gnavus et rel. 0.11 0.13 0.15 0.13 0.67 0.14 1.19 0.89 5.26 4.86 4.99

Uncultured

Clostridiales Uncultured Clostridiales I 3.87 1.61 0.17 0.73 0.36 0.15 0.42 9.72 0.49 2.38 0.22

Proteobacteria Sutterella wadsworthensis et rel. 0.60 0.37 0.11 0.60 0.73 0.11 0.62 3.23 1.21 6.37 1.96

Verrucomicrobia Akkermansia 2.60 2.87 4.45 2.98 1.01 3.55 1.10 0.65 0.34 0.28 0.35

199

comparable to the abundances found in VA compared to VB.

Higher rate of change values were noted in VA and VB at the beginning of the experiment

as communities transitioned from an in vivo to an in vitro environment (days 0-12 for VA and

days 0-4 for VB; Figure 7.6b). These values decreased as communities approached steady state

(days 12-26 for VA and days 4-22 for VB), where low rate of change values were maintained

(days 26-44 for VA and days 22-44 for VB). On average, VA Δt was 6.2% between days 26-36

and 8.2% between days 36-44. Average VB Δt values were 3.8% between days 26-36 and 5.1%

between days 36-44. There was a relatively large change in VAB community dynamics as the

community shifted from the equal VA and VB mixture to a more VA-like community. VAB rate

of change values showed a peak of change between days 36-38 (VAB Δt= 20.3%). VAB rate of

change values decreased over the rest of the chemostat run (Δt= 11.5% between days 42-44).

H’, EH’, and S values were stable in VA between days 28-44 (Figure 7.6c,d,e, Table 7.6).

Even though S values were stable in VB between days 26-44 (Figure 7.6d), H’ and EH’ values

were generally stable between days 22-32 and gradually decreased after day 32 (Figure 7.6c,e).

VAB H' and EH' values remained relatively stable between days 36-42 and decreased between

days 42-44 (Figure 7.6c,e), while S values fluctuated between days 36-44 (Figure 7.6d).

7.3 Discussion

In this study we investigated the long-term persistence of allochthonous microorganisms

within complex microbial communities simulating the human gut following perturbation. We

used DGGE to determine whether antibiotic treatment would aid in the establishment of donor

B-derived synthetic stool (MET-1) in chemostat communities seeded with donor A feces. We

found that donor A chemostat communities appeared to resist perturbation with antibiotics and/or

200

Table 7.6 Ecological parameters (diversity, evenness, and richness) for VA, VB, and VAB

communities on days 0, 36, and 44, as measured using DGGE and the HITChip. NA=not

applicable.

Analysis

Method Day

Shannon Diversity

Index (H')

Shannon Equitability

Index (EH')

Richness

(S)

VA VB VAB VA VB VAB VA VB VAB

DGGE

0 3.37 3.06 NA 0.86 0.79 NA 51 49 NA

36 3.21 3.06 3.13 0.80 0.77 0.81 55 54 49

44 3.23 2.94 3.03 0.81 0.74 0.78 55 55 50

HITChip

0 5.81 5.67 NA 0.81 0.83 NA 964 755 NA

36 5.21 5.32 5.57 0.78 0.82 0.82 573 485 624

44 5.25 5.16 5.32 0.78 0.79 0.79 588 486 600

201

MET-1. We also used DGGE and HITChip to determine whether chemostat communities seeded

with feces from two different donors resulted in the more successful persistence of

microorganisms from one donor compared to the other donor (microbial invasion perturbation).

Following mixing we found that the structure of the mixed community tended to more closely

resemble the structure of the donor A chemostat community than the donor B chemostat

community.

We first wanted to determine whether antibiotic treatment resulted in reproducible

alterations of steady state chemostat communities. Aside from being important in terms of our

experimental design, stereotypic responses by the gut microbiota to the same stimulus can help

us understand the factors that influence the structure of gut communities (156). This can help us

design therapeutic strategies to intentionally manipulate the gut microbiota to maintain or restore

host health following perturbation (156). Although V(+/+) and V(+/-) steady state communities

were less similar to each other prior to clindamycin exposure (relative to steady state

communities from V(-/+) and V(-/+)), communities from both vessels responded to clindamycin

treatment in a reproducible manner and increased in similarity during the clindamycin treatment

period (Figure 7.2). In vivo studies investigating the reproducibility of antibiotic disturbances

found that inbred mice had reproducible baseline gut microbiota that underwent significant,

reproducible compositional shifts following exposure to antibiotics (156). However, despite

these findings there were still significant differences between in the composition of the gut

microbiota of control and antibiotic-treated groups. Although inbred murine models undoubtedly

have their value, chemostats can culture steady state communities under even more tightly

controlled environmental conditions to produce complex, stable communities that have higher

within-run, between vessel reproducibility (see chapter 3).

202

The persistence of fecal microorganisms following stool transplants has been investigated

in conventionally raised Lewis rats in the presence and absence of antibiotic treatment (137).

Similar to our experimental design, Lewis rats were divided into four groups: transplantation,

antibiotics, transplantation following antibiotics, and controls. The stool transplant consisted of

pooled cecal content from Sprague-Dawley and Wistar strain rats to ensure the donor microbial

communities were highly diverse and different from Lewis rat feces. Following the

administration of the donor feces (in the presence and absence of antibiotic perturbation) the

recipient gut microbiota increased in diversity and became more similar to the donor stool.

Phylotypes from donor rats persisted in recipient rats at 3 months post-transplantation. Although

it was expected that antibiotic perturbation would facilitate the establishment of donor gut

microbes within the recipient gut microbiota, rats that received a stool transplant following

antibiotic perturbation combined the reshaping effects of the two treatments (including a

decrease in gut microbiota diversity), reducing the establishment of donor gut microbes.

Although the stool transplant aided in recovering the primary loss of diversity, Manichanh et al.

(2010) proposed that the altered state of the gut microbiota following antibiotic perturbation may

be more important for the persistence of donor microbes (137).

In our experiment MET-1 supplementation in the presence or absence of clindamycin-

induced perturbation did not appear to have a significant impact on the structure of chemostat

communities (as determined using DGGE). However, although comparisons can be made

between V(+/+) and V(+/-) using DGGE, at the end of the experiment we cannot determine

whether specific DGGE bands from V(+/+) correspond to bacterial species recovering from the

chemostat community, bacterial species introduced through MET-1, or a combination of the two.

Similarly, the bacterial diversity present in V(-/+) prior to MET-1 supplementation made it

203

difficult to determine whether strains from MET-1 were able to persist within an unperturbed

community. While this is not unexpected, we had hypothesized that MET-1 supplementation

would allow antibiotic-treated communities to recover more fully and rapidly. Following

antibiotic treatment and MET-1 addition, there was an initial increase in V(+/+) diversity

(including richness and evenness). However, V(+/-) recovery was also very rapid and appeared

to follow the same pattern as seen in V(+/+) between days 44-48.

V(+/+), V(+/-), V(-/+) and V(-/-) were seeded with feces from a healthy donor (donor A).

Many studies have shown that the gut microbiota from healthy donors is stable over time and

resilient to perturbation (24, 68). This might explain why V(+/+) and V(+/-) were able to rapidly

recover from clindamycin exposure (Figure 7.2). In addition, the healthy gut microbiota

generally shows colonization resistance where the indigenous gut microbiota prevents

allochthonous microorganisms from persisting within the gut environment (24). This might also

explain why MET-1 did not appear to persist in V(-/+) (by DGGE). Even though we were

interested in investigating the response of antibiotic perturbation and MET-1 supplementation in

a healthy donor, the major advantage of choosing a healthy donor for this preliminary

experiment was the relative ease of access to the fecal donation. Future experiments will

investigate the effects of antibiotic exposure and MET-1 supplementation using chemostat

communities seeded with feces from an unhealthy donor whose gut microbiota is suspected to be

in a state of dysbiosis (e.g. an ulcerative colitis patient). If chemostat communities seeded with

feces derived from unhealthy donors are less stable following perturbation, it would be expected

that these communities would display a prolonged recovery time, which in turn would make it

easier to determine if the addition of MET-1 supplementation aids in more rapid and full

recovery of microbial diversity under these conditions.

204

When designing MET strategies it is important to isolate strains from a healthy fecal donor

with a diverse, stable gut microbiota that can resist perturbation. However, to date it is unclear

whether the gut microbiotas from different healthy donors would be equally effective at

displacing dysbiotic communities, or whether the microbiota from one donor may be superior to

that of another for this purpose. In a controlled, randomized clinical trial Van Nood et al. (2013)

found that a single infusion of stool from a healthy donor cured disease in 13 of 16 recurrent CDI

patients (103). The remaining 3 patients were subsequently treated with feces derived from new

donors, resulting in disease resolution (103). While it is unclear whether some patients simply

require multiple stool transplants regardless of the fecal donor, it is possible that some fecal

donors may be superior to others under certain conditions.

In this study we found that mixing steady state chemostat communities seeded with feces

from donors A and B resulted in a community that more closely resembled the donor A

community structure than the donor B community structure. For obvious ethical reasons there is

no data regarding the persistence of exogenous fecal microbial communities in the gut

microbiota of healthy human recipients. However, our observations are consistent with the rat

experiments conducted by Manichanh et al. (2010), who found that rat feces taken 1 and 3

months following a fecal transplant clustered closer to the donor fecal sample than with pre-

transplantation samples (137).

The diversity of the chemostat communities present in VA and VB may explain the shifts

in the VAB community structure following mixing. In general, the donor A fecal inoculum and

steady state chemostat communities had higher richness and diversity values than donor B. As

mentioned in chapter 4, ecosystems with higher diversity exhibit higher stability compared to

less complex ecosystems (215). Although this is consistent with ecological theory, future studies

205

using similar experiments to investigate additional donors are required to confirm this

observation. Should this observation prove to be consistent, screening the microbial diversity of

a stool sample may provide a rapid way of determining the suitability of a stool transplant donor.

It is also possible that the chemostat environment altered the structure of donor A

chemostat communities following inoculation (in vivo to in vitro transition), causing the

communities to shift to a more chemostat adapted-state. To confirm that mixing donor A and B

chemostat communities was not biased by changes in the structure of the communities following

inoculation, further experiments should explore the influence of altering environmental

conditions within the model to see if we can repeat the same results. For example, we could alter

the composition of the growth medium to mimic a diet with higher protein content and lower

fibre content. Finally, to add biological significance, experiments should be verified in vivo, for

example using a humanized animal model colonized with donor A and/or donor B fecal

communities.

Despite the fact that communities from VA, VB, and VAB were analyzed by HITChip, in

addition to DGGE, it was still difficult to determine whether the microbial populations

comprising the VAB day 44 sample originated from microorganisms from VA, VB, or both.

Because the structure of the VAB day 44 community changed following mixing on day 36 to

more closely resemble that of the VA day 44 community, it is tempting to conclude that the

bacterial populations from VB were not able to persist. However, microbial interactions

following mixing might have not only influenced the growth and persistence of VA bacterial

populations, but also the microbial populations of closely related taxa originating from VB. A

limitation of the HITChip is that it is unable to distinguish between different strains of species-

like groups (64). Moreover, potential cross-hybridization of closely-related taxa and differences

206

in probe binding affinities make it difficult to completely correlate the origin of bacterial

populations comprising the VAB day 44 community (see chapter 6; 64).

The use of defined microbial communities can help to overcome some of these challenges.

For example, one approach would be to construct defined communities from the feces of donors

A and B. DEC-A would be composed of strains cultured from donor A feces, while DEC-B

would be composed of strains cultured from donor B feces. After sequencing the genomes of all

the members of these two defined communities, we could repeat our experiment and analyze the

communities using metagenomics. Alternatively, we could also design DEC-A and DEC-B such

that each defined community contained microorganisms from different species (if possible with

no overlapping genera). Because each defined community would contain different species this

would lend more confidence to a finding that a given strain originated from a specific defined

community. A final advantage of using defined communities over undefined fecal communities

is that we would not need to rely on the 16S rRNA gene to analyze our chemostat communities;

instead it would be possible to use other genes characteristic of specific members (e.g. butyryl-

CoA:acetate CoA-transferase gene in the butyrate production pathway; 306) to monitor changes

in the community using highly accurate techniques such as quantitative PCR.

Following mixing of VA and VB in VAB, the liquid culture would likely contain a

metabolite signature that could affect the structure of the VAB community following mixing. For

example, metabolites produced by one microorganism can serve as a food source for the growth

of another microorganism (cross-feeding; 307). Thus, if a particular, required metabolite source

was lacking in VB, the growth of specific bacterial populations within this community could

have been hampered. Following mixing of the two communities, the required growth-promoting

metabolite might be donated by microbial components of VA. Thus, it is important that future

207

studies examine the functional output of each community in a mixed community experiment, and

that the effects of these metabolites on both communities separately, as well as when mixed

together, be monitored.

In conclusion, we found that the long-term persistence of allochthonous microorganisms

within complex microbial communities depends on a variety of factors, including the

composition of the donor and recipient communities. We found that chemostat communities

seeded from donor A feces resisted perturbation with antibiotics and/or MET-1. We also found

that, following mixing under chemostat conditions, the structure of the combined community

from donors A and B more closely resembled the organization of the donor A than the donor B

consortium.

Work from our lab has resulted in the design of a new MET formulation (MET-2)

constructed from commensal bacterial species isolated from donor A feces. Similar to donor B-

derived MET-1, MET-2 is composed of strains with favourable antibiotic resistance profiles that

will facilitate its complete eradication using minimal antibiotic treatment. Future experiments

will determine whether novel MET formulations (e.g. donor A-derived MET-2) can persist in

antibiotic perturbed and unperturbed chemostat cultures seeded with feces from different donors.

We will also mix steady state chemostat communities seeded with MET-1 and MET-2

communities to determine whether the simplified, defined communities behave in the same

manner as their fecal counterparts. Moreover, we will study the stability of MET-1 and MET-2

in the chemostat following perturbation with various antibiotics. Experiments such as these will

aid the design of future MET formulations.

Although further experimentation is required to confirm our findings, preliminary results

suggest that a hierarchy exists between different fecal donors, where the fecal community from

208

one donor tends to predominate over the fecal community from another. Future experiments are

required to determine whether this observation is consistent, and whether we can find an

emerging pattern that will help to explain this phenomenon. It is important to determine whether

dominant fecal donors share a similar community structure (composition and abundance of

microorganisms), because communities from these donors may have superior attributes

amenable to the further development of the MET concept.

209

Chapter 8

Summary, conclusions and future work

Although the gut microbiota is incompletely characterized, technological advancements in

the field of microbial ecology have allowed researchers to describe the diversity of the human

microbiome in more detail than ever before (59, 308). However, because the compositional and

functional characteristics of "healthy" gut microbiotas are not fully described, attempts to

correlate differences or changes of the gut microbiota in diseased states remains a challenge (61).

Future studies must focus on strengthening established associations between the structure and

functionality of gut microbiota in healthy and diseased states to determine causality (13).

To date in vivo models have not been fully successful at demonstrating the mechanisms of

gut diseases (2). Even though gut dysbiosis has been associated with several diseased states, the

inability to focus on the gut microbiota in the absence of host involvement makes it difficult to

determine whether imbalances in the gut microbiota are a cause or consequence of disease (13,

61). Novel models are required to study key factors that control the dynamics of gut

communities. Understanding these factors would allow researchers to predict and control the

behaviour of gut microbiotas (154).

Researchers propose that gut microbiota studies should use a multidisciplinary "human gut

systems biology approach" where the causative relationships between disease states and

dysbiotic gut microbiota are supported by in vitro models (gut fermentation models and human

intestinal cell lines) as well as in vivo studies (in animals and humans; 2). In our research we

characterized and validated a twin-vessel single-stage in vitro gut fermentation model. This

210

model was particularly useful as it allowed us to investigate the distal colon - a region of the

gastrointestinal tract that is involved in several disease states (18).

Top-down and bottom-up systems biology approaches (metagenomics,

metatranscriptomics, metaproteomics, metabolomics) allow researchers to identify key

interactions and metabolic pathways important for health and disease (2). Although traditional

systems biology approaches have been used in in vivo and in situ studies (309-312), they have

not been routinely applied to in vitro gut fermentation models (2). However, a systems biology

tactic could allow for the simultaneous exploration of the gut microbiota at several molecular

levels, including the identities of the members of the gut microbiota (16S rRNA sequencing

data), their metabolic potential (metagenomics), transcribed rRNA (metatranscriptomics), protein

expression (metaproteomics), and metabolic products (metabolomics; 61). Future studies

analyzing chemostat communities encompassing all of these techniques will allow a more

complete characterization of communities developed within our model.

In our study microbial community samples were analyzed using a molecular fingerprinting

technique (DGGE) and a phylogenetic microarray (HITChip). Throughout our investigations we

evaluated the comparability between DGGE and HITChip-calculated parameters. We found that

DGGE was a convenient, high throughput molecular fingerprinting technique that provided

information on differences in community structure (between samples and over time). However,

compared to the HITChip we found DGGE lacked sensitivity (as expected based on previous

work; 276). Although correlation coefficient values calculated by DGGE and the HITChip

usually showed the same relative between-sample differences, DGGE correlation coefficient

values were usually lower than the corresponding HITChip values. Moreover, microbial

community diversity was also underrepresented in DGGE profiles relative to HITChip profiles.

211

Differences in correlation coefficient values and ecological parameters can be explained by

differences in the number of data points generated in each profile, as DGGE generates fewer data

points than the HITChip (56). Therefore, direct comparisons of correlation coefficient values

should only be made if they were calculated using the same techniques (e.g. DGGE vs. DGGE,

HITChip vs. HITChip).

Another disadvantage of DGGE is gel-to-gel variation, which can hamper comparisons of

community development over the course of a chemostat run. H’, EH’, and S values are

susceptible to gel-to-gel variation as they depend on band number and peak height. Temporal H’,

EH’, and S analysis is possible looking at uncorrected plots, however the overall patterns of

change are much easier to interpret when consecutive gels are aligned. Adjustment using

correction values creates plots that still maintain relative between-vessel, within-gel differences.

It also allows for an overall pattern of stability or instability to emerge more clearly. Although it

is preferable to extract sample-specific numerical comparisons for samples run on a single gel,

the gel-to-gel adjustment method outlined in this study still allows for multi-gel analysis of

overall patterns of change over time. However, there is a limit to the amount of correction that

can be done to overcome stochastic variability between and within gels. For example, variation

in the denaturing gradient along the width of a DGGE gel can cause "smiling" and less consistent

band separation in different DGGE lanes. This means the gel analysis software might find DGGE

profiles to be less similar to each other than they would be otherwise. If these differences affect

the relative differences for H', EH', and S values within a gel this can make between-gel

adjustments more difficult.

DGGE can provide useful information regarding changes in community structure for a

large number of samples, however it does not provide phylogenetic information regarding the

212

composition of the community. Phylogenetic information is not only required to determine the

extent to which the in vitro communities resemble the in vivo communities, but it also allows us

to determine which members of the community might respond to an experimental treatment. The

HITChip is a phylogenetic fingerprinting tool with a higher resolution, reproducibility and

therefore reliability relative to DGGE (56, 203). Furthermore, the HITChip provided us with

taxonomic information for key samples that allowed us to assess the diversity of communities

developed within our model, compare compositional differences between samples, and track

changes in specific populations of bacteria over time and following experimental treatment.

However, like all techniques, the HITChip has its limitations. Firstly, it can only detect

taxa which are covered by the reference sequences (301). Unless there are distinct differences

between the variable regions, species/strain level resolution is challenging (301). Cross-

hybridization with non-target sequences and differences in probe binding affinities might result

in the classification of genus-like groups as closely related taxa, especially when the specific taxa

are not represented on the microarray (64). However, cross-hybridization with related but

unidentified taxa can still provide information on the composition of the community (56, 301,

313). Also, cross-hybridization and differences in probe binding affinities may also result in low

levels of "background noise" resulting in false positives (64). Finally, compared to next-

generation sequencing there are fewer options for downstream analysis (301).

We chose to analyze our samples using DGGE and the HITChip because, for our purposes,

the benefits of these techniques outweighed their drawbacks. Although quantitative comparison

between specific samples is best achieved using more sensitive techniques (e.g. HITChip, 16S

rDNA sequencing, quantitative PCR), DGGE is a more rapid and inexpensive method of

assessing the overall pattern of community development and stability over time (314). Vessel

213

correlation coefficients and moving window analysis (rate of change) calculated from DGGE

profiles have previously been used to monitor community development in SHIME units (165,

203) and sequential batch reactors (315, 316). In addition, the Shannon diversity index and band

richness has previously been used to compare DGGE profiles between samples (205, 208). There

are benefits to high-level views of complex microbial community structure as the reduction in

detail can allow broader patterns to emerge more clearly. Having said that, the HITChip is a

robust technique in which hybridizations and community profiles correlate well with

pyrosequencing-based comparisons (301). Moreover, HITChip analysis tends to be faster,

cheaper, and more straightforward than high-coverage amplicon sequencing (301). Taking this

all into consideration, we decided to use DGGE to analyze large numbers of samples, enabling

us to follow long-term changes in community dynamics. Key samples of interest were targeted

by DGGE for HITChip analysis in collaboration with the laboratory of Prof. Dr. Willem de Vos,

a leading research group with expertise in this method.

The stability of gut microbial ecosystems can be assessed by analyzing two different

components: compositional stability and functional stability. Monitoring metabolic activity is

also an important marker of microbial community stability within a chemostat model. However,

given the extensive experimentation to characterize our model using DNA-based techniques, the

evaluation of metabolic activity was beyond the scope of the work described in this thesis. In

collaboration with an analytical chemistry research laboratory (Aucoin Lab, University of

Waterloo) we are currently further-developing our model to establish the functional capacity and

metabolic stability of our chemostat-grown communities over time using 1H-NMR. Although it

is too early to comment on this 1H-NMR data, the preliminary results support the reproducibility

and stability of chemostat communities, the defined time frame to steady state, differences in

214

chemostat communities seeded with consecutive fecal donations from the same donor, inter-

individual variation, and response to perturbation (with NE, clindamycin, and/or MET-1). Even

though I do not explicitly discuss metabolic activity in our system in my thesis work, it has been

previously shown that extremely dynamic bacterial communities (unstable composition) can still

maintain a functionally stable ecosystem (317). Even without our 1H-NMR data, we have shown

that communities developed within our model have high compositional stability, which would

suggest they have functional stability as well.

In chapter 3 we showed that twin-vessel, single-stage chemostats inoculated with identical

fresh fecal samples from a healthy donor can develop and maintain stable, diverse, and

reproducible communities. By 36-days post-inoculation chemostat communities were at steady

state compositions, however steady state was achieved earlier in some runs compared to others.

The high reproducibility and stability of communities developed within our model allows us to

conduct placebo-controlled or multi-treatment experiments, as we can compare communities

between vessels or within the same vessel over time (165). As noted in other in vitro studies,

steady state communities were enriched in Bacteroidetes but not Clostridium cluster XIVa,

Bacilli or other Firmicutes relative to the fecal inocula. Communities developed within this

model had higher within-run reproducibility than between-run repeatability when using

consecutive fecal donations. Even though between-donor steady state community similarity

increased following inoculation of the vessels, donor-specific identities were maintained.

Although we used our model to develop and maintain chemostat communities mimicking

healthy adult gut microbiota, our model also has the potential to simulate a diverse range of

different gut microbiota, such as gut microbiota from disease states with known distal gut

involvement (e.g. IBD, CDI, colorectal cancer; 18). In vitro gut fermentation modeling could be

215

particularly useful to advance the understanding of diseases such as CD and UC, which normally

focus on immune system functionality (2). Additionally, it would also be interesting to model gut

microbiota that experience normal compositional variation in the absence of host involvement,

for example by genetics (related vs. unrelated individuals; 33, 39), age (infants vs. adults vs.

elderly; 318), or diet (Western diet vs. rural African diet; 32).

In chapter 4 we characterized the effects of defined (clindamycin) and undefined (NE;

norepinephrine) perturbation on the structure of chemostat communities. Both short term (1

dose) and long term (5 day course) NE exposure did not appear to have a substantial effect on the

structure of chemostat communities. On the other hand, a 5-day course of clindamycin treatment

caused large changes in the structure of chemostat communities that did not fully recover by the

end of the experiment. Characterizing the effects of clindamycin treatment allowed us to

characterize the effect of perturbation within our system using a stimulus that is known to alter

the structure of gut communities both in vivo and in vitro.

Although we did not notice changes in chemostat communities following NE

administration, further experimentation is required before we can conclude that NE does not

have an effect on the structure or functionality of the human gut microbiota. Characterizing NE-

treated chemostat communities using systems biology approaches may enable us to identify

community responses not identified using DGGE alone (2). Future studies should also

investigate the effect of NE on chemostat communities seeded with feces from IBD patients (in

remission), who believe stress has an influence on the onset and relapse of illness (223). If we

are still unable to identify any responses to NE treatment after analyzing chemostat communities

using these techniques, this will give us more confidence that NE causes indirect changes in the

gut microbiota of stressed individuals through the alteration of host physiological processes, and

216

not through microbiota-driven processes per se.

Characterizing the responses of the gut microbiota to specific perturbation allows

researchers to identify key species, microbial interactions, and functions which can be important

for maintenance of the stability of gut communities (62). A greater understanding of these factors

will allow manipulation of the gut microbiota in order to prevent gut dysbiosis in healthy

individuals or restore balance in diseased individuals (62). Because in vitro gut fermentation

models are not limited by ethical considerations (compared to in vivo models) perturbation

studies can be conducted using a wide variety of endogenous (e.g. cortisol metabolites) or

exogenous (e.g. novel drugs) stimuli (2).

In chapter 5 we developed a chemostat model using packed-column biofilm reactors to

simulate mucosal environments. In combination with our chemostat model, packed-column

reactors allowed us to simultaneously culture biologically relevant planktonic and biofilm

communities that were complex, reproducible, and distinct. Moreover, planktonic and biofilm

communities responded to antibiotic perturbation in a biologically relevant manner.

The results outlined in chapter 5 represent preliminary data gained through the

development of an appropriate model for the study of biofilms in the gut, and as such there are

still several questions that should be investigated. As with our planktonic communities, the

metabolic activity of microorganisms growing as biofilms should be assessed, as previous

studies have reported that microorganisms growing in planktonic and biofilm states are

metabolically distinct (16). Furthermore, we analyzed biofilms grown over longer time periods

(36 or 41 days) and shorter time periods (7 or 12 days) to ensure biofilm growth reached a stable,

reproducible composition. Future experiments should characterize biofilms developed for shorter

time periods and harvested at regularly spaced, staggered time points to more closely mimic the

217

regular renewal of the mucus layer and sloughing off of host epithelial cells in vivo (270). This

would also provide a more dynamic view of biofilm formation within our model. Finally,

alternative packing material could also be tested for its ability to bind mucin, and a variety of

different mucin sources could also be tested (e.g. spent LS174T tissue culture cell medium, a

source of human mucins).

As mentioned in chapter 5, mucosal communities may be particularly important to host

health due to their proximity to the host epithelium (265). As such, in vitro gut fermentation

models that incorporate a mucosal environment could be especially useful for the study of

diseased mucosal biofilms. Future studies should investigate biofilm formation in chemostat

communities seeded with feces from donors displaying gut microbial dysbiosis as well as

different healthy donors. Common taxa or functions consistently observed in healthy biofilm

communities that are lacking or altered in diseased biofilm communities may help steer future

therapeutic interventions.

The SHIME system has been used to model mucosal biofilms from UC patients using

mucin-coated microcosms in netted bags (266). However, these experiments were limited to 2

days, as longer fermentation times altered the ratios of bacterial populations within biofilm

communities to ratios not routinely observed in vivo (Tom Van de Wiele, personal

communication). Because these cultures cannot be maintained for longer time periods, steady

state communities cannot be established and therefore placebo-controlled or multi-treatment

studies cannot be applied. Since microcosms in netted bags were less successful at maintaining

cultures for longer time periods, future studies should investigate whether packed-column

reactors are more successful at culturing UC biofilm communities.

Packed-column reactors could also be used as a screening method to determine whether

218

novel probiotic strains can persist within the gut as part of a mucosal biofilm community. Both in

vivo and in vitro studies have shown that Lactobacillus rhamnosus GG (one of the best

characterized probiotic strains; 319) strongly adheres to the intestinal epithelium (320). Packed-

column reactors can be used to characterize the biofilm-forming ability of mono-species

probiotics, synthetic stool (e.g. MET-1), or stool replacement (transplant) within a gut

fermentation model.

Packed-column reactors are flexible and convenient in that they can be filled with many

different indigestible dietary residues or prebiotics, such as inulin-type fructans (321-323) or

arabinoxylan oligosaccharides (324). In addition, packed-column reactors can be filled with

packing material coated in other scaffolds for biofilm formation, such as sIgA (325, 326).

Investigations utilizing these alternative packing materials would allow researchers to screen

potential prebiotics for their ability to stimulate biofilm and planktonic growth in a tightly

controlled environment. It would also allow researchers to investigate the importance of biofilm

growth in additional niches within the gut environment.

In chapter 6 we cultured simplified, defined communities (DEC and MET-1) and a

complex, undefined fecal community in our chemostat model. Despite the reduced diversity in

DEC and MET-1, both simplified communities were able to establish steady state compositions

more rapidly than fecal chemostat communities. Fecal and DEC chemostat communities

maintained stable diversity, richness, and evenness values upon the establishment of steady state

compositions. Although MET-1 chemostat community maintained stable richness values,

diversity and evenness values slightly decreased over time. As expected, DEC and MET-1

chemostat communities differed from fecal chemostat communities at the genus-like level,

however overall the composition of DEC and MET chemostat communities were comparable at

219

the phylum/class-level.

Using well-characterized, standardized synthetic gut communities we can explore

molecular-level microbial interactions using systems biology techniques (e.g. metagenomics).

However, when using defined communities to represent a complex native community it is

important to ensure that important characteristics of the native community are not lost. In

addition to the HITChip and DGGE data, future studies will compare the functionality of defined

communities to fecal communities to ensure important metabolic pathways are adequately

represented. As novel culturing techniques are developed we can improve the yield of

microorganisms cultured from feces and increase the number of strains included in defined

community formulations. Characterizing the resistance and resilience responses following

perturbation using a well-defined stimulus (e.g. specific antibiotics) will also allow us to

determine whether additional strains must be incorporated into defined communities to maintain

ecosystem functionality.

New formulations of defined communities can be developed to suit the research question

of interest. Future formulations could include strains cultured from the feces of different healthy

donors as well as diseased donors. Moreover, different compositional backgrounds could be

investigated, such as hosts with different diets (e.g. Western vs. rural diets) or ages (infants vs.

adults vs. elderly).

Packed-column biofilm reactors could be attached to chemostat vessels culturing defined

communities to investigate the biofilm-forming ability of DEC or MET formulations.

Alternatively, strains could be cultured from biofilms seeded with fecal microorganisms.

Depending on the column packing material, isolated microorganisms could be used to construct

"defined biofilm communities" mimicking gut mucosal communities or communities adhering to

220

dietary residues or prebiotics. Defined mucosal biofilm communities could provide strong

microbial ecosystem candidates, while defined prebiotic-adherent communities could be used in

conjunction with the prebiotic to develop novel synbiotics.

In chapter 7 we investigated long-term persistence of allochthonous microorganisms within

fecal chemostat communities in the presence and absence of perturbation (with antibiotics or

microorganisms). We found that chemostat communities seeded with feces from a healthy donor

rapidly recovered from antibiotic perturbation in the presence and absence of MET-1

supplementation (although recovery was incomplete). The complexity of fecal chemostat

communities in the absence of antibiotic perturbation as well as rapid recovery of fecal

chemostat communities in the absence of MET-1 supplementation meant that it was difficult to

assess the persistence of MET-1 strains using DGGE. However, this led us to question whether a

MET formulation derived from one donor would be more successful at persisting in a fecal

chemostat community than a MET formulation derived from another donor. This also led us to

question whether the fecal community from one healthy donor would be more successful at

persisting in the presence of a fecal community from another donor. A "dominant" donor gut

microbiota may be able to better persist in the presence of a diseased recipient's intestine,

therefore making an ideal candidate for stool transplants or novel MET formulations. Following

mixing of chemostat communities seeded with feces from two different healthy donors, we found

the structure of the mixed community more closely resembled the structure of one donor

chemostat community over the other. Further experiments are required to determine whether this

observation is consistent, and whether we can find an emerging pattern in the structure or

functionality of dominant gut microbiota to explain this phenomenon.

Future experiments should explore the persistence of a donor A-derived MET in a donor B

221

fecal chemostat community (+/- antibiotic perturbation). However, if we fail to notice changes in

the structure of the donor B fecal chemostat community there are several other options to

explore. First, we could test the persistence of different MET formulations in diseased gut

microbiota (again, +/- antibiotic perturbation). The lower diversity and stability of gut

microbiotas found in certain diseased states (e.g. UC and CD) suggests that they may be more

disturbed by perturbation events, possibly making it easier to evaluate the effects of MET

supplementation (89, 90, 94, 143).

Future studies should also begin to explore the factors that drive structural shifts following

mixing of chemostat communities seeded with feces from different donors. To ensure that the in

vivo to in vitro transition didn't bias the structure of donor A and B chemostat communities prior

to mixing, future experiments will alter the vessel environment to ensure we can repeat the same

experimental results. One method of accomplishing this is to alter the composition of the growth

medium. If the mixed community still resembles the donor A chemostat community using a

variety of growth media, we could then use a germ-free animal model to further support these

observations. We could also investigate shifts in the structure of mixed defined communities

(donor A-derived DEC and donor B-derived DEC). If all the members of the donor A and B

defined communities are sequenced we can follow changes in the mixed community structure

using metagenomic approaches. We could also alter the composition of defined communities to

identify microorganisms that drive structural changes.

It would also be interesting to determine whether we could manipulate the donor B

chemostat community to more closely resemble the donor A chemostat community without

mixing the two communities. In addition to microorganisms, mixing the donor A and B

chemostat communities also combined metabolic products from both chemostat cultures. Future

222

experiments should explore whether altering donor B metabolic profiles to a more donor A-like

profile can shift the donor B community structure to a more donor A-like structure.

Despite the numerous studies that can be carried out using in vitro gut fermentation models

they should be considered to be complementary to in vivo studies (2). Experimental designs

should include both in vitro and in vivo models to strengthen the validity of each model and

distinguish between human physiological processes and microbial ecosystem processes (2).

However, gut fermentation models can be modified to more accurately simulate host responses.

Previous studies have applied effluent from gut fermentation models to monolayers of human

intestinal cell models (e.g. Caco-2 or HT29-MTX cells; 327-329). Kim et al. (2012) developed a

human "gut-on-a-chip" microdevice which was composed of two microfluidic channels

separated by a porous membrane, coated with extracellular matrix and lined with Caco-2 human

intestinal epithelial cells (330). Not only can the gut-on-a-chip simulate the structure,

physiology, and peristaltic motions of a human intestine, but Lactobacillus rhamnosus GG (a

normal intestinal microorganism) could be co-cultured on the luminal surface of the Caco-2 cells

for more than one week without affecting Caco-2 viability (330). Attachment of the human gut-

on-a-chip to gut fermentation models could greatly enhance the data extracted from in vitro gut

microbiota studies. Marzorati et al. (2010) have developed an in vitro adhesion module

consisting of two compartments (luminal and basal) separated by a semi-permeable membrane

(331). The luminal chamber has an artificial mucus layer and anaerobic conditions to culture

SHIME communities while the basal chamber has aerobic conditions to culture human intestinal

epithelial cells. This device allows SHIME microbial communities to adhere to and colonize

host-related surfaces, simulates the transport of chemical compounds across epithelial surfaces,

and mimics host-microbe interactions in a tightly controlled environment. Combination of

223

human intestinal epithelial cell models and gut fermentation models allows researchers to assess

the effects of gut communities on host epithelial responses in a highly controlled manner (2).

In summary, the microorganisms colonizing the human gut are essential to host health,

however the compositional and functional characteristics of these communities, in both healthy

and diseased states, requires further characterization (61). Although dysbiotic gut microbiota

have been associated with several gut diseases, it is unclear whether dysbiosis is the cause or

consequence of disease (61). Establishing causal relationships will greatly aid the design of novel

therapeutic approaches (13). To this end, our twin-vessel single-stage distal gut chemostat model

provides a powerful tool to elucidate the role of gut microbial ecosystems in health and a wide

range of diseased states. The relative lack of ethical considerations means the experimental

capacity of this model is essentially limitless (2). Because this system lacks host physiological

processes, changes in the structure and functionality of the chemostat microbial communities can

be directly correlated to the applied treatment. Additionally, the flexibility in the design of this

model means that it can be modified or upgraded to more accurately mimic the host environment

as required.

224

References

1. Macfarlane GT, and Macfarlane S. 2007. Models for intestinal fermentation: association

between food components, delivery systems, bioavailability and functional interactions in

the gut. Curr. Opin. Biotechnol. 18:156-162.

2. Payne AN, Zihler A, Chassard C, Lacroix C. 2012. Advances and perspectives in in

vitro human gut fermentation modeling. Trends Biotechnol. 30:17-25.

3. Gibson GR, and Roberfroid MB. 1995. Dietary modulation of the human colonic

microbiota: introducing the concept of prebiotics. J. Nutr. 125:1401-1412.

4. Macfarlane GT, and Macfarlane S. 2012. Bacteria, colonic fermentation, and

gastrointestinal health. J. AOAC Int. 95:50-60.

5. Backhed F, Ley RE, Sonnenburg JL, Peterson DA, Gordon JI. 2005. Host-bacterial

mutualism in the human intestine. Science. 307:1915-1920.

6. Eckburg PB, Bik EM, Bernstein CN, Purdom E, Dethlefsen L, Sargent M, Gill SR,

Nelson KE, Relman DA. 2005. Diversity of the human intestinal microbial flora. Science.

308:1635-1638.

7. Rajilic-Stojanovic M, Smidt H, de Vos WM. 2007. Diversity of the human

gastrointestinal tract microbiota revisited. Environ. Microbiol. 9:2125-2136.

8. Gaskins HR, Croix JA, Nakamura N, Nava GM. 2008. Impact of the intestinal

microbiota on the development of mucosal defense. Clin. Infect. Dis. 46 Suppl 2:S80-6;

discussion S144-51.

9. Qin J, Li R, Raes J, Arumugam M, Burgdorf KS, Manichanh C, Nielsen T, Pons N,

Levenez F, Yamada T, Mende DR, Li J, Xu J, Li S, Li D, Cao J, Wang B, Liang H,

Zheng H, Xie Y, Tap J, Lepage P, Bertalan M, Batto JM, Hansen T, Le Paslier D,

Linneberg A, Nielsen HB, Pelletier E, Renault P, Sicheritz-Ponten T, Turner K, Zhu

H, Yu C, Li S, Jian M, Zhou Y, Li Y, Zhang X, Li S, Qin N, Yang H, Wang J, Brunak

S, Dore J, Guarner F, Kristiansen K, Pedersen O, Parkhill J, Weissenbach J,

MetaHIT Consortium, Bork P, Ehrlich SD, Wang J. 2010. A human gut microbial gene

catalogue established by metagenomic sequencing. Nature. 464:59-65.

10. Lutgendorff F, Akkermans LM, Soderholm JD. 2008. The role of microbiota and

probiotics in stress-induced gastro-intestinal damage. Curr. Mol. Med. 8:282-298.

11. Neish AS. 2009. Microbes in gastrointestinal health and disease. Gastroenterology.

136:65-80.

225

12. O'Hara AM, and Shanahan F. 2006. The gut flora as a forgotten organ. EMBO Rep.

7:688-693.

13. de Vos WM, and de Vos EA. 2012. Role of the intestinal microbiome in health and

disease: from correlation to causation. Nutr. Rev. 70 Suppl 1:S45-56.

14. Macfarlane GT, Gibson GR, Cummings JH. 1992. Comparison of fermentation

reactions in different regions of the human colon. J. Appl. Bacteriol. 72:57-64.

15. Macfarlane GT, Macfarlane S, Gibson GR. 1998. Use of a three-stage compound

continuous culture system to investigate bacterial growth and metabolism in the human

colonic microbiota. Microbiol. Ecol. 35:180-187.

16. Macfarlane S, and Macfarlane GT. 2006. Composition and metabolic activities of

bacterial biofilms colonizing food residues in the human gut. Appl. Environ. Microbiol.

72:6204-6211.

17. Randal Bollinger R, Barbas AS, Bush EL, Lin SS, Parker W. 2007. Biofilms in the

large bowel suggest an apparent function of the human vermiform appendix. J. Theor. Biol.

249:826-831.

18. DuPont AW, and DuPont HL. 2011. The intestinal microbiota and chronic disorders of

the gut. Nat. Rev. Gastroenterol. Hepatol. 8:523-531.

19. Dethlefsen L, McFall-Ngai M, Relman DA. 2007. An ecological and evolutionary

perspective on human-microbe mutualism and disease. Nature. 449:811-818.

20. Hooper LV, and Gordon JI. 2001. Commensal host-bacterial relationships in the gut.

Science. 292:1115-1118.

21. Hayashi H, Sakamoto M, Benno Y. 2002. Phylogenetic analysis of the human gut

microbiota using 16S rDNA clone libraries and strictly anaerobic culture-based methods.

Microbiol. Immunol. 46:535-548.

22. Suau A, Bonnet R, Sutren M, Godon JJ, Gibson GR, Collins MD, Dore J. 1999. Direct

analysis of genes encoding 16S rRNA from complex communities reveals many novel

molecular species within the human gut. Appl. Environ. Microbiol. 65:4799-4807.

23. Wang X, Heazlewood SP, Krause DO, Florin TH. 2003. Molecular characterization of

the microbial species that colonize human ileal and colonic mucosa by using 16S rDNA

sequence analysis. J. Appl. Microbiol. 95:508-520.

24. Lozupone CA, Stombaugh JI, Gordon JI, Jansson JK, Knight R. 2012. Diversity,

stability and resilience of the human gut microbiota. Nature. 489:220-230.

226

25. Becker N, Kunath J, Loh G, Blaut M. 2011. Human intestinal microbiota:

characterization of a simplified and stable gnotobiotic rat model. Gut Microbes. 2:25-33.

26. Zoetendal EG, Vaughan EE, de Vos WM. 2006. A microbial world within us. Mol.

Microbiol. 59:1639-1650.

27. Dave M, Higgins PD, Middha S, Rioux KP. 2012. The human gut microbiome: current

knowledge, challenges, and future directions. Transl. Res. 160:246-257.

28. Dethlefsen L, Huse S, Sogin ML, Relman DA. 2008. The pervasive effects of an

antibiotic on the human gut microbiota, as revealed by deep 16S rRNA sequencing. PLoS

Biol. 6:e280.

29. Pepper JW, and Rosenfeld S. 2012. The emerging medical ecology of the human gut

microbiome. Trends Ecol. Evol. 27:381-384.

30. Spor A, Koren O, Ley R. 2011. Unravelling the effects of the environment and host

genotype on the gut microbiome. Nat. Rev. Microbiol. 9:279-290.

31. Wu GD, Chen J, Hoffmann C, Bittinger K, Chen YY, Keilbaugh SA, Bewtra M,

Knights D, Walters WA, Knight R, Sinha R, Gilroy E, Gupta K, Baldassano R, Nessel

L, Li H, Bushman FD, Lewis JD. 2011. Linking long-term dietary patterns with gut

microbial enterotypes. Science. 334:105-108.

32. De Filippo C, Cavalieri D, Di Paola M, Ramazzotti M, Poullet JB, Massart S, Collini

S, Pieraccini G, Lionetti P. 2010. Impact of diet in shaping gut microbiota revealed by a

comparative study in children from Europe and rural Africa. Proc. Natl. Acad. Sci. U. S. A.

107:14691-14696.

33. Turnbaugh PJ, Hamady M, Yatsunenko T, Cantarel BL, Duncan A, Ley RE, Sogin

ML, Jones WJ, Roe BA, Affourtit JP, Egholm M, Henrissat B, Heath AC, Knight R,

Gordon JI. 2009. A core gut microbiome in obese and lean twins. Nature. 457:480-484.

34. Moodley Y, Linz B, Yamaoka Y, Windsor HM, Breurec S, Wu JY, Maady A,

Bernhoft S, Thiberge JM, Phuanukoonnon S, Jobb G, Siba P, Graham DY, Marshall

BJ, Achtman M. 2009. The peopling of the Pacific from a bacterial perspective. Science.

323:527-530.

35. Schwarz S, Morelli G, Kusecek B, Manica A, Balloux F, Owen RJ, Graham DY, van

der Merwe S, Achtman M, Suerbaum S. 2008. Horizontal versus familial transmission

of Helicobacter pylori. PLoS Pathog. 4:e1000180.

36. Vaishampayan PA, Kuehl JV, Froula JL, Morgan JL, Ochman H, Francino MP.

2010. Comparative metagenomics and population dynamics of the gut microbiota in

mother and infant. Genome Biol. Evol. 2:53-66.

227

37. Stewart JA, Chadwick VS, Murray A. 2005. Investigations into the influence of host

genetics on the predominant eubacteria in the faecal microflora of children. J. Med.

Microbiol. 54:1239-1242.

38. Zoetendal EG, Akkermans AD, Akkermans-van Vliet WM, de Visser JA, de Vos

WM. 2001. The host genotype affects the bacterial community in the human

gastronintestinal tract. Microb. Ecol. Health Dis. 13:129-134.

39. Turnbaugh PJ, Quince C, Faith JJ, McHardy AC, Yatsunenko T, Niazi F, Affourtit

J, Egholm M, Henrissat B, Knight R, Gordon JI. 2010. Organismal, genetic, and

transcriptional variation in the deeply sequenced gut microbiomes of identical twins. Proc.

Natl. Acad. Sci. U. S. A. 107:7503-7508.

40. Gillevet P, Sikaroodi M, Keshavarzian A, Mutlu EA. 2010. Quantitative assessment of

the human gut microbiome using multitag pyrosequencing. Chem. Biodivers. 7:1065-1075.

41. Marteau P, Pochart P, Dore J, Bera-Maillet C, Bernalier A, Corthier G. 2001.

Comparative study of bacterial groups within the human cecal and fecal microbiota. Appl.

Environ. Microbiol. 67:4939-4942.

42. Dave M, Johnson LA, Walk ST, Young VB, Stidham RW, Chaudhary MN, Funnell

J, Higgins PD. 2011. A randomised trial of sheathed versus standard forceps for obtaining

uncontaminated biopsy specimens of microbiota from the terminal ileum. Gut. 60:1043-

1049.

43. Yatsunenko T, Rey FE, Manary MJ, Trehan I, Dominguez-Bello MG, Contreras M,

Magris M, Hidalgo G, Baldassano RN, Anokhin AP, Heath AC, Warner B, Reeder J,

Kuczynski J, Caporaso JG, Lozupone CA, Lauber C, Clemente JC, Knights D,

Knight R, Gordon JI. 2012. Human gut microbiome viewed across age and geography.

Nature. 486:222-227.

44. Dominguez-Bello MG, Costello EK, Contreras M, Magris M, Hidalgo G, Fierer N,

Knight R. 2010. Delivery mode shapes the acquisition and structure of the initial

microbiota across multiple body habitats in newborns. Proc. Natl. Acad. Sci. U. S. A.

107:11971-11975.

45. Kozyrskyj AL, Bahreinian S, Azad MB. 2011. Early life exposures: impact on asthma

and allergic disease. Curr. Opin. Allergy Clin. Immunol. 11:400-406.

46. Azad MB, Konya T, Maughan H, Guttman DS, Field CJ, Chari RS, Sears MR,

Becker AB, Scott JA, Kozyrskyj AL. 2013. Gut microbiota of healthy Canadian infants:

profiles by mode of delivery and infant diet at 4 months. CMAJ. .

47. Koenig JE, Spor A, Scalfone N, Fricker AD, Stombaugh J, Knight R, Angenent LT,

Ley RE. 2011. Succession of microbial consortia in the developing infant gut microbiome.

Proc. Natl. Acad. Sci. U. S. A. 108 Suppl 1:4578-4585.

228

48. Palmer C, Bik EM, DiGiulio DB, Relman DA, Brown PO. 2007. Development of the

human infant intestinal microbiota. PLoS Biol. 5:e177.

49. Biagi E, Nylund L, Candela M, Ostan R, Bucci L, Pini E, Nikkila J, Monti D,

Satokari R, Franceschi C, Brigidi P, De Vos W. 2010. Through ageing, and beyond: gut

microbiota and inflammatory status in seniors and centenarians. PLoS One. 5:e10667.

50. Claesson MJ, Cusack S, O'Sullivan O, Greene-Diniz R, de Weerd H, Flannery E,

Marchesi JR, Falush D, Dinan T, Fitzgerald G, Stanton C, van Sinderen D, O'Connor

M, Harnedy N, O'Connor K, Henry C, O'Mahony D, Fitzgerald AP, Shanahan F,

Twomey C, Hill C, Ross RP, O'Toole PW. 2011. Composition, variability, and temporal

stability of the intestinal microbiota of the elderly. Proc. Natl. Acad. Sci. U. S. A. 108

Suppl 1:4586-4591.

51. Tiihonen K, Ouwehand AC, Rautonen N. 2010. Human intestinal microbiota and

healthy ageing. Ageing Res. Rev. 9:107-116.

52. Woodmansey EJ. 2007. Intestinal bacteria and ageing. J. Appl. Microbiol. 102:1178-

1186.

53. Schiffrin EJ, Morley JE, Donnet-Hughes A, Guigoz Y. 2010. The inflammatory status

of the elderly: the intestinal contribution. Mutat. Res. 690:50-56.

54. Gerritsen J, Smidt H, Rijkers GT, de Vos WM. 2011. Intestinal microbiota in human

health and disease: the impact of probiotics. Genes Nutr. 6:209-240.

55. Tap J, Mondot S, Levenez F, Pelletier E, Caron C, Furet JP, Ugarte E, Munoz-

Tamayo R, Paslier DL, Nalin R, Dore J, Leclerc M. 2009. Towards the human intestinal

microbiota phylogenetic core. Environ. Microbiol. 11:2574-2584.

56. Rajilic-Stojanovic M, Heilig HG, Molenaar D, Kajander K, Surakka A, Smidt H, de

Vos WM. 2009. Development and application of the human intestinal tract chip, a

phylogenetic microarray: analysis of universally conserved phylotypes in the abundant

microbiota of young and elderly adults. Environ. Microbiol. 11:1736-1751.

57. Andersson AF, Lindberg M, Jakobsson H, Backhed F, Nyren P, Engstrand L. 2008.

Comparative analysis of human gut microbiota by barcoded pyrosequencing. PLoS One.

3:e2836.

58. Marchesi JR. 2011. Shifting from a gene-centric to metabolite-centric strategy to

determine the core gut microbiome. Bioeng. Bugs. 2:309-314.

59. Human Microbiome Project Consortium. 2012. Structure, function and diversity of the

healthy human microbiome. Nature. 486:207-214.

229

60. Nicholson JK, Holmes E, Kinross J, Burcelin R, Gibson G, Jia W, Pettersson S. 2012.

Host-gut microbiota metabolic interactions. Science. 336:1262-1267.

61. Backhed F, Fraser CM, Ringel Y, Sanders ME, Sartor RB, Sherman PM, Versalovic

J, Young V, Finlay BB. 2012. Defining a healthy human gut microbiome: current

concepts, future directions, and clinical applications. Cell. Host Microbe. 12:611-622.

62. Relman DA. 2012. The human microbiome: ecosystem resilience and health. Nutr. Rev.

70 Suppl 1:S2-9.

63. Pimm SL. 1991. The Balance of Nature? Ecological Issues in the Conservation of Species

and Communities. University of Chicago Press, Chicago.

64. Jalanka-Tuovinen J, Salonen A, Nikkila J, Immonen O, Kekkonen R, Lahti L, Palva

A, de Vos WM. 2011. Intestinal microbiota in healthy adults: temporal analysis reveals

individual and common core and relation to intestinal symptoms. PLoS One. 6:e23035.

65. Rajilic-Stojanovic M. 2007. Diversity of the Human Gastrointestinal Microbiota: Novel

Perspectives from High Throughput Analysis. PhD. Wageningen University, Wageningen,

The Netherlands.

66. Costello EK, Lauber CL, Hamady M, Fierer N, Gordon JI, Knight R. 2009. Bacterial

community variation in human body habitats across space and time. Science. 326:1694-

1697.

67. Franks AH, Harmsen HJ, Raangs GC, Jansen GJ, Schut F, Welling GW. 1998.

Variations of bacterial populations in human feces measured by fluorescent in situ

hybridization with group-specific 16S rRNA-targeted oligonucleotide probes. Appl.

Environ. Microbiol. 64:3336-3345.

68. Zoetendal EG, Akkermans AD, De Vos WM. 1998. Temperature gradient gel

electrophoresis analysis of 16S rRNA from human fecal samples reveals stable and host-

specific communities of active bacteria. Appl. Environ. Microbiol. 64:3854-3859.

69. Fite A, Macfarlane S, Furrie E, Bahrami B, Cummings JH, Steinke DT, Macfarlane

GT. 2013. Longitudinal analyses of gut mucosal microbiotas in ulcerative colitis in

relation to patient age and disease severity and duration. J. Clin. Microbiol. 51:849-856.

70. Mackey RL, and Currie DJ. 2001. The diversity-disturbance relationship: Is it generally

strong and peaked? Ecology. 82:3479-3492.

71. Svensson JR, Lindegarth M, Pavia H. 2009. Equal rates of disturbance cause different

patterns of diversity. Ecology. 90:496-505.

72. Hooper DU, Chapin FS, Ewel JJ, Hector A, Inchausti P, Lavorel S, Lawton JH,

Lodge DM, Loreau M, Naeem S, Schmid B, Setälä H, Symstad AJ, Vandermeer J,

230

Wardle DA. 2005. Effects of biodiversity on ecosystem functioning: A consensus of

current knowledge. Ecol. Monogr. 75:3-35.

73. Naeem S, Thompson LJ, Lawler SP, Lawton JH, Woodfin RM. 1994. Declining

biodiversity can alter the performance of ecosystems. Nature. 368:734-737.

74. Sousa WP. 1984. The role of disturbance in natural communities. Annu Rev Ecol Syst.

15:353-391.

75. Berga M, Szekely AJ, Langenheder S. 2012. Effects of disturbance intensity and

frequency on bacterial community composition and function. PLoS One. 7:e36959.

76. Dethlefsen L, and Relman DA. 2011. Incomplete recovery and individualized responses

of the human distal gut microbiota to repeated antibiotic perturbation. Proc. Natl. Acad.

Sci. U. S. A. 108 Suppl 1:4554-4561.

77. Beisner BE, Haydon DT, Cuddington K. 2003. Alternative stable states in ecology.

Front. Ecol. Environ. 1:376-382.

78. Schröder A, Persson L, De Roos AM. 2005. Direct experimental evidence for alternative

stable states: a review. Oikos. 110:3-19.

79. Thompson-Chagoyan OC, Maldonado J, Gil A. 2005. Aetiology of inflammatory bowel

disease (IBD): role of intestinal microbiota and gut-associated lymphoid tissue immune

response. Clin. Nutr. 24:339-352.

80. Virgin HW, and Todd JA. 2011. Metagenomics and personalized medicine. Cell.

147:44-56.

81. Tamboli CP, Neut C, Desreumaux P, Colombel JF. 2004. Dysbiosis in inflammatory

bowel disease. Gut. 53:1-4.

82. Salzman NH, and Bevins CL. 2008. Negative interactions with the microbiota: IBD.

Adv. Exp. Med. Biol. 635:67-78.

83. Hawrelak JA, and Myers SP. 2004. The causes of intestinal dysbiosis: a review. Altern.

Med. Rev. 9:180-197.

84. Ley RE, Turnbaugh PJ, Klein S, Gordon JI. 2006. Microbial ecology: human gut

microbes associated with obesity. Nature. 444:1022-1023.

85. Turnbaugh PJ, Backhed F, Fulton L, Gordon JI. 2008. Diet-induced obesity is linked

to marked but reversible alterations in the mouse distal gut microbiome. Cell. Host

Microbe. 3:213-223.

231

86. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, Gordon JI. 2006. An

obesity-associated gut microbiome with increased capacity for energy harvest. Nature.

444:1027-1031.

87. Sun Y, Cai Y, Mai V, Farmerie W, Yu F, Li J, Goodison S. 2010. Advanced

computational algorithms for microbial community analysis using massive 16S rRNA

sequence data. Nucleic Acids Res. 38:e205.

88. Knights D, Parfrey LW, Zaneveld J, Lozupone C, Knight R. 2011. Human-associated

microbial signatures: examining their predictive value. Cell. Host Microbe. 10:292-296.

89. Ott SJ, Musfeldt M, Wenderoth DF, Hampe J, Brant O, Folsch UR, Timmis KN,

Schreiber S. 2004. Reduction in diversity of the colonic mucosa associated bacterial

microflora in patients with active inflammatory bowel disease. Gut. 53:685-693.

90. Manichanh C, Rigottier-Gois L, Bonnaud E, Gloux K, Pelletier E, Frangeul L, Nalin

R, Jarrin C, Chardon P, Marteau P, Roca J, Dore J. 2006. Reduced diversity of faecal

microbiota in Crohn's disease revealed by a metagenomic approach. Gut. 55:205-211.

91. Sokol H, Pigneur B, Watterlot L, Lakhdari O, Bermudez-Humaran LG, Gratadoux

JJ, Blugeon S, Bridonneau C, Furet JP, Corthier G, Grangette C, Vasquez N, Pochart

P, Trugnan G, Thomas G, Blottiere HM, Dore J, Marteau P, Seksik P, Langella P.

2008. Faecalibacterium prausnitzii is an anti-inflammatory commensal bacterium

identified by gut microbiota analysis of Crohn disease patients. Proc. Natl. Acad. Sci. U. S.

A. 105:16731-16736.

92. Kaser A, Zeissig S, Blumberg RS. 2010. Inflammatory bowel disease. Annu. Rev.

Immunol. 28:573-621.

93. Willing BP, Dicksved J, Halfvarson J, Andersson AF, Lucio M, Zheng Z, Jarnerot G,

Tysk C, Jansson JK, Engstrand L. 2010. A pyrosequencing study in twins shows that

gastrointestinal microbial profiles vary with inflammatory bowel disease phenotypes.

Gastroenterology. 139:1844-1854.e1.

94. Martinez C, Antolin M, Santos J, Torrejon A, Casellas F, Borruel N, Guarner F,

Malagelada JR. 2008. Unstable composition of the fecal microbiota in ulcerative colitis

during clinical remission. Am. J. Gastroenterol. 103:643-648.

95. Png CW, Linden SK, Gilshenan KS, Zoetendal EG, McSweeney CS, Sly LI,

McGuckin MA, Florin TH. 2010. Mucolytic bacteria with increased prevalence in IBD

mucosa augment in vitro utilization of mucin by other bacteria. Am. J. Gastroenterol.

105:2420-2428.

96. Lepage P, Hasler R, Spehlmann ME, Rehman A, Zvirbliene A, Begun A, Ott S,

Kupcinskas L, Dore J, Raedler A, Schreiber S. 2011. Twin study indicates loss of

232

interaction between microbiota and mucosa of patients with ulcerative colitis.

Gastroenterology. 141:227-236.

97. Salonen A, de Vos WM, Palva A. 2010. Gastrointestinal microbiota in irritable bowel

syndrome: present state and perspectives. Microbiology. 156:3205-3215.

98. Saulnier DM, Riehle K, Mistretta TA, Diaz MA, Mandal D, Raza S, Weidler EM, Qin

X, Coarfa C, Milosavljevic A, Petrosino JF, Highlander S, Gibbs R, Lynch SV,

Shulman RJ, Versalovic J. 2011. Gastrointestinal microbiome signatures of pediatric

patients with irritable bowel syndrome. Gastroenterology. 141:1782-1791.

99. Rajilic-Stojanovic M, Biagi E, Heilig HG, Kajander K, Kekkonen RA, Tims S, de Vos

WM. 2011. Global and deep molecular analysis of microbiota signatures in fecal samples

from patients with irritable bowel syndrome. Gastroenterology. 141:1792-1801.

100. Rigsbee L, Agans R, Shankar V, Kenche H, Khamis HJ, Michail S, Paliy O. 2012.

Quantitative profiling of gut microbiota of children with diarrhea-predominant irritable

bowel syndrome. Am. J. Gastroenterol. 107:1740-1751.

101. Grehan MJ, Borody TJ, Leis SM, Campbell J, Mitchell H, Wettstein A. 2010. Durable

alteration of the colonic microbiota by the administration of donor fecal flora. J. Clin.

Gastroenterol. 44:551-561.

102. Khoruts A, Dicksved J, Jansson JK, Sadowsky MJ. 2010. Changes in the composition

of the human fecal microbiome after bacteriotherapy for recurrent Clostridium difficile-

associated diarrhea. J. Clin. Gastroenterol. 44:354-360.

103. van Nood E, Vrieze A, Nieuwdorp M, Fuentes S, Zoetendal EG, de Vos WM, Visser

CE, Kuijper EJ, Bartelsman JF, Tijssen JG, Speelman P, Dijkgraaf MG, Keller JJ.

2013. Duodenal infusion of donor feces for recurrent Clostridium difficile. N. Engl. J. Med.

368:407-415.

104. Petrof EO, Gloor GB, Vanner SJ, Weese JS, Carter D, Daigneault MC, Brown EM,

Schroeter K, Allen-Vercoe E. 2013. Stool substitute transplant therapy for the eradication

of C. difficile infection: RePOOPulating the gut. Microbiome. 1:1-12.

105. Marchesi JR, Dutilh BE, Hall N, Peters WH, Roelofs R, Boleij A, Tjalsma H. 2011.

Towards the human colorectal cancer microbiome. PLoS One. 6:e20447.

106. Sobhani I, Tap J, Roudot-Thoraval F, Roperch JP, Letulle S, Langella P, Corthier G,

Tran Van Nhieu J, Furet JP. 2011. Microbial dysbiosis in colorectal cancer (CRC)

patients. PLoS One. 6:e16393.

107. Wang T, Cai G, Qiu Y, Fei N, Zhang M, Pang X, Jia W, Cai S, Zhao L. 2012.

Structural segregation of gut microbiota between colorectal cancer patients and healthy

volunteers. ISME J. 6:320-329.

233

108. McCoy AN, Araujo-Perez F, Azcarate-Peril A, Yeh JJ, Sandler RS, Keku TO. 2013.

Fusobacterium is associated with colorectal adenomas. PLoS One. 8:e53653.

109. Stsepetova J, Sepp E, Julge K, Vaughan E, Mikelsaar M, de Vos WM. 2007.

Molecularly assessed shifts of Bifidobacterium ssp. and less diverse microbial communities

are characteristic of 5-year-old allergic children. FEMS Immunol. Med. Microbiol. 51:260-

269.

110. Bisgaard H, Li N, Bonnelykke K, Chawes BL, Skov T, Paludan-Muller G, Stokholm

J, Smith B, Krogfelt KA. 2011. Reduced diversity of the intestinal microbiota during

infancy is associated with increased risk of allergic disease at school age. J. Allergy Clin.

Immunol. 128:646-52.e1-5.

111. Storro O, Oien T, Langsrud O, Rudi K, Dotterud C, Johnsen R. 2011. Temporal

variations in early gut microbial colonization are associated with allergen-specific

immunoglobulin E but not atopic eczema at 2 years of age. Clin. Exp. Allergy. 41:1545-

1554.

112. Noval Rivas M, Burton OT, Wise P, Zhang YQ, Hobson SA, Garcia Lloret M,

Chehoud C, Kuczynski J, DeSantis T, Warrington J, Hyde ER, Petrosino JF, Gerber

GK, Bry L, Oettgen HC, Mazmanian SK, Chatila TA. 2013. A microbiota signature

associated with experimental food allergy promotes allergic sensitization and anaphylaxis.

J. Allergy Clin. Immunol. 131:201-212.

113. Nistal E, Caminero A, Herran AR, Arias L, Vivas S, de Morales JM, Calleja S, de

Miera LE, Arroyo P, Casqueiro J. 2012. Differences of small intestinal bacteria

populations in adults and children with/without celiac disease: effect of age, gluten diet,

and disease. Inflamm. Bowel Dis. 18:649-656.

114. Di Cagno R, De Angelis M, De Pasquale I, Ndagijimana M, Vernocchi P, Ricciuti P,

Gagliardi F, Laghi L, Crecchio C, Guerzoni ME, Gobbetti M, Francavilla R. 2011.

Duodenal and faecal microbiota of celiac children: molecular, phenotype and metabolome

characterization. BMC Microbiol. 11:219-2180-11-219.

115. Kalliomaki M, Satokari R, Lahteenoja H, Vahamiko S, Gronlund J, Routi T,

Salminen S. 2012. Expression of microbiota, Toll-like receptors, and their regulators in the

small intestinal mucosa in celiac disease. J. Pediatr. Gastroenterol. Nutr. 54:727-732.

116. Vaarala O. 2011. The gut as a regulator of early inflammation in type 1 diabetes. Curr.

Opin. Endocrinol. Diabetes Obes. 18:241-247.

117. Giongo A, Gano KA, Crabb DB, Mukherjee N, Novelo LL, Casella G, Drew JC,

Ilonen J, Knip M, Hyoty H, Veijola R, Simell T, Simell O, Neu J, Wasserfall CH,

Schatz D, Atkinson MA, Triplett EW. 2011. Toward defining the autoimmune

microbiome for type 1 diabetes. ISME J. 5:82-91.

234

118. Brown CT, Davis-Richardson AG, Giongo A, Gano KA, Crabb DB, Mukherjee N,

Casella G, Drew JC, Ilonen J, Knip M, Hyoty H, Veijola R, Simell T, Simell O, Neu J,

Wasserfall CH, Schatz D, Atkinson MA, Triplett EW. 2011. Gut microbiome

metagenomics analysis suggests a functional model for the development of autoimmunity

for type 1 diabetes. PLoS One. 6:e25792.

119. Larsen N, Vogensen FK, van den Berg FW, Nielsen DS, Andreasen AS, Pedersen BK,

Al-Soud WA, Sorensen SJ, Hansen LH, Jakobsen M. 2010. Gut microbiota in human

adults with type 2 diabetes differs from non-diabetic adults. PLoS One. 5:e9085.

120. Wu X, Ma C, Han L, Nawaz M, Gao F, Zhang X, Yu P, Zhao C, Li L, Zhou A, Wang

J, Moore JE, Millar BC, Xu J. 2010. Molecular characterisation of the faecal microbiota

in patients with type II diabetes. Curr. Microbiol. 61:69-78.

121. Kootte RS, Vrieze A, Holleman F, Dallinga-Thie GM, Zoetendal EG, de Vos WM,

Groen AK, Hoekstra JB, Stroes ES, Nieuwdorp M. 2012. The therapeutic potential of

manipulating gut microbiota in obesity and type 2 diabetes mellitus. Diabetes Obes. Metab.

14:112-120.

122. Qin J, Li Y, Cai Z, Li S, Zhu J, Zhang F, Liang S, Zhang W, Guan Y, Shen D, Peng

Y, Zhang D, Jie Z, Wu W, Qin Y, Xue W, Li J, Han L, Lu D, Wu P, Dai Y, Sun X, Li

Z, Tang A, Zhong S, Li X, Chen W, Xu R, Wang M, Feng Q, Gong M, Yu J, Zhang Y,

Zhang M, Hansen T, Sanchez G, Raes J, Falony G, Okuda S, Almeida M, LeChatelier

E, Renault P, Pons N, Batto JM, Zhang Z, Chen H, Yang R, Zheng W, Li S, Yang H,

Wang J, Ehrlich SD, Nielsen R, Pedersen O, Kristiansen K, Wang J. 2012. A

metagenome-wide association study of gut microbiota in type 2 diabetes. Nature. 490:55-

60.

123. Musso G, Gambino R, Cassader M. 2011. Interactions between gut microbiota and host

metabolism predisposing to obesity and diabetes. Annu. Rev. Med. 62:361-380.

124. Xu P, Li M, Zhang J, Zhang T. 2012. Correlation of intestinal microbiota with

overweight and obesity in Kazakh school children. BMC Microbiol. 12:283-2180-12-283.

125. Karri S, Acosta-Martinez V, Coimbatore G. 2010. Effect of dihydrotestosterone on

gastrointestinal tract of male Alzheimer's disease transgenic mice. Indian. J. Exp. Biol.

48:453-465.

126. Koren O, Spor A, Felin J, Fak F, Stombaugh J, Tremaroli V, Behre CJ, Knight R,

Fagerberg B, Ley RE, Backhed F. 2011. Human oral, gut, and plaque microbiota in

patients with atherosclerosis. Proc. Natl. Acad. Sci. U. S. A. 108 Suppl 1:4592-4598.

127. Williams BL, Hornig M, Buie T, Bauman ML, Cho Paik M, Wick I, Bennett A,

Jabado O, Hirschberg DL, Lipkin WI. 2011. Impaired carbohydrate digestion and

transport and mucosal dysbiosis in the intestines of children with autism and

gastrointestinal disturbances. PLoS One. 6:e24585.

235

128. Sheedy JR, Wettenhall RE, Scanlon D, Gooley PR, Lewis DP, McGregor N, Stapleton

DI, Butt HL, DE Meirleir KL. 2009. Increased d-lactic Acid intestinal bacteria in patients

with chronic fatigue syndrome. In Vivo. 23:621-628.

129. de Weerth C, Fuentes S, Puylaert P, de Vos WM. 2013. Intestinal microbiota of infants

with colic: development and specific signatures. Pediatrics. 131:e550-8.

130. Wang Z, Klipfell E, Bennett BJ, Koeth R, Levison BS, Dugar B, Feldstein AE, Britt

EB, Fu X, Chung YM, Wu Y, Schauer P, Smith JD, Allayee H, Tang WH, DiDonato

JA, Lusis AJ, Hazen SL. 2011. Gut flora metabolism of phosphatidylcholine promotes

cardiovascular disease. Nature. 472:57-63.

131. Bravo JA, Forsythe P, Chew MV, Escaravage E, Savignac HM, Dinan TG,

Bienenstock J, Cryan JF. 2011. Ingestion of Lactobacillus strain regulates emotional

behavior and central GABA receptor expression in a mouse via the vagus nerve. Proc.

Natl. Acad. Sci. U. S. A. 108:16050-16055.

132. Murphy S, and Nguyen VH. 2011. Role of gut microbiota in graft-versus-host disease.

Leuk. Lymphoma. 52:1844-1856.

133. Berer K, Mues M, Koutrolos M, Rasbi ZA, Boziki M, Johner C, Wekerle H,

Krishnamoorthy G. 2011. Commensal microbiota and myelin autoantigen cooperate to

trigger autoimmune demyelination. Nature. 479:538-541.

134. Spencer MD, Hamp TJ, Reid RW, Fischer LM, Zeisel SH, Fodor AA. 2011.

Association between composition of the human gastrointestinal microbiome and

development of fatty liver with choline deficiency. Gastroenterology. 140:976-986.

135. Kane M, Case LK, Kopaskie K, Kozlova A, MacDearmid C, Chervonsky AV,

Golovkina TV. 2011. Successful transmission of a retrovirus depends on the commensal

microbiota. Science. 334:245-249.

136. Kuss SK, Best GT, Etheredge CA, Pruijssers AJ, Frierson JM, Hooper LV, Dermody

TS, Pfeiffer JK. 2011. Intestinal microbiota promote enteric virus replication and systemic

pathogenesis. Science. 334:249-252.

137. Manichanh C, Reeder J, Gibert P, Varela E, Llopis M, Antolin M, Guigo R, Knight

R, Guarner F. 2010. Reshaping the gut microbiome with bacterial transplantation and

antibiotic intake. Genome Res. 20:1411-1419.

138. Sartor RB. 2004. Therapeutic manipulation of the enteric microflora in inflammatory

bowel diseases: antibiotics, probiotics, and prebiotics. Gastroenterology. 126:1620-1633.

139. Llopis M, Antolin M, Guarner F, Salas A, Malagelada JR. 2005. Mucosal colonisation

with Lactobacillus casei mitigates barrier injury induced by exposure to trinitronbenzene

sulphonic acid. Gut. 54:955-959.

236

140. Jansson J, Willing B, Lucio M, Fekete A, Dicksved J, Halfvarson J, Tysk C, Schmitt-

Kopplin P. 2009. Metabolomics reveals metabolic biomarkers of Crohn's disease. PLoS

One. 4:e6386.

141. Chang JY, Antonopoulos DA, Kalra A, Tonelli A, Khalife WT, Schmidt TM, Young

VB. 2008. Decreased diversity of the fecal Microbiome in recurrent Clostridium difficile-

associated diarrhea. J. Infect. Dis. 197:435-438.

142. Carroll KC, and Bartlett JG. 2011. Biology of Clostridium difficile: implications for

epidemiology and diagnosis. Annu. Rev. Microbiol. 65:501-521.

143. Scanlan PD, Shanahan F, O'Mahony C, Marchesi JR. 2006. Culture-independent

analyses of temporal variation of the dominant fecal microbiota and targeted bacterial

subgroups in Crohn's disease. J. Clin. Microbiol. 44:3980-3988.

144. Wang Y, Hoenig JD, Malin KJ, Qamar S, Petrof EO, Sun J, Antonopoulos DA,

Chang EB, Claud EC. 2009. 16S rRNA gene-based analysis of fecal microbiota from

preterm infants with and without necrotizing enterocolitis. ISME J. 3:944-954.

145. Khan KJ, Ullman TA, Ford AC, Abreu MT, Abadir A, Marshall JK, Talley NJ,

Moayyedi P. 2011. Antibiotic therapy in inflammatory bowel disease: a systematic review

and meta-analysis. Am. J. Gastroenterol. 106:661-673.

146. Pimentel M, Lembo A, Chey WD, Zakko S, Ringel Y, Yu J, Mareya SM, Shaw AL,

Bortey E, Forbes WP, TARGET Study Group. 2011. Rifaximin therapy for patients

with irritable bowel syndrome without constipation. N. Engl. J. Med. 364:22-32.

147. Holzapfel WH, Haberer P, Snel J, Schillinger U, Huis in't Veld JH. 1998. Overview of

gut flora and probiotics. Int. J. Food Microbiol. 41:85-101.

148. Ringel Y, Quigley E, Lin H. 2012. Using Probiotics in Gastrointestinal Disorders. Am. J.

Gastroenterol. Suppl 1:34-40.

149. Gibson GR, Probert HM, Loo JV, Rastall RA, Roberfroid MB. 2004. Dietary

modulation of the human colonic microbiota: updating the concept of prebiotics. Nutr. Res.

Rev. 17:259-275.

150. Petrof EO, Claud EC, Gloor GB, Allen-Vercoe E. 2013. Microbial Ecosystem

therapeutics: a new paradigm in medicine? Benef. Microbes. 4:53-65.

151. Borody TJ, Warren EF, Leis S, Surace R, Ashman O. 2003. Treatment of ulcerative

colitis using fecal bacteriotherapy. J. Clin. Gastroenterol. 37:42-47.

152. Andrews P. 1992. Chronic constipation reversed by restoration of bowel flora. A case and

a hypothesis. Eur. J. Gastroenterol. Hepatol. 4:245-247.

237

153. Vrieze A, Van Nood E, Holleman F, Salojarvi J, Kootte RS, Bartelsman JF, Dallinga-

Thie GM, Ackermans MT, Serlie MJ, Oozeer R, Derrien M, Druesne A, Van

Hylckama Vlieg JE, Bloks VW, Groen AK, Heilig HG, Zoetendal EG, Stroes ES, de

Vos WM, Hoekstra JB, Nieuwdorp M. 2012. Transfer of intestinal microbiota from lean

donors increases insulin sensitivity in individuals with metabolic syndrome.

Gastroenterology. 143:913-6.e7.

154. Trosvik P, Rudi K, Straetkvern KO, Jakobsen KS, Naes T, Stenseth NC. 2010. Web

of ecological interactions in an experimental gut microbiota. Environ. Microbiol. 12:2677-

2687.

155. Walker AW, Ince J, Duncan SH, Webster LM, Holtrop G, Ze X, Brown D, Stares

MD, Scott P, Bergerat A, Louis P, McIntosh F, Johnstone AM, Lobley GE, Parkhill J,

Flint HJ. 2011. Dominant and diet-responsive groups of bacteria within the human colonic

microbiota. ISME J. 5:220-230.

156. Antonopoulos DA, Huse SM, Morrison HG, Schmidt TM, Sogin ML, Young VB.

2009. Reproducible community dynamics of the gastrointestinal microbiota following

antibiotic perturbation. Infect. Immun. 77:2367-2375.

157. Bry L, Falk PG, Midtvedt T, Gordon JI. 1996. A model of host-microbial interactions

in an open mammalian ecosystem. Science. 273:1380-1383.

158. Gerard P, Beguet F, Lepercq P, Rigottier-Gois L, Rochet V, Andrieux C, Juste C.

2004. Gnotobiotic rats harboring human intestinal microbiota as a model for studying

cholesterol-to-coprostanol conversion. FEMS Microbiol. Ecol. 47:337-343.

159. Meurens F, Berri M, Siggers RH, Willing BP, Salmon H, Van Kessel AG, Gerdts V.

2007. Commensal bacteria and expression of two major intestinal chemokines,

TECK/CCL25 and MEC/CCL28, and their receptors. PLoS One. 2:e677.

160. Rawls JF, Samuel BS, Gordon JI. 2004. Gnotobiotic zebrafish reveal evolutionarily

conserved responses to the gut microbiota. Proc. Natl. Acad. Sci. U. S. A. 101:4596-4601.

161. Sekirov I, Russell SL, Antunes LC, Finlay BB. 2010. Gut microbiota in health and

disease. Physiol. Rev. 90:859-904.

162. Hufeldt MR, Nielsen DS, Vogensen FK, Midtvedt T, Hansen AK. 2010. Variation in

the gut microbiota of laboratory mice is related to both genetic and environmental factors.

Comp. Med. 60:336-347.

163. Hegsted DM. 1975. Relevance of animal studies to human disease. Cancer Res. 35:3537-

3539.

164. Cohen SM. 1995. Human relevance of animal carcinogenicity studies. Regul. Toxicol.

Pharmacol. 21:75-80; discussion 81-6.

238

165. Van den Abbeele P, Grootaert C, Marzorati M, Possemiers S, Verstraete W, Gerard

P, Rabot S, Bruneau A, El Aidy S, Derrien M, Zoetendal E, Kleerebezem M, Smidt H,

Van de Wiele T. 2010. Microbial community development in a dynamic gut model is

reproducible, colon region specific, and selective for Bacteroidetes and Clostridium cluster

IX. Appl. Environ. Microbiol. 76:5237-5246.

166. Hart AL, Stagg AJ, Graffener H, Glise H, Falk P, Kamm MA (eds.), 2002. Gut

Ecology. Martin Dunitz Ltd, United Kingdom.

167. Le Blay G, Rytka J, Zihler A, Lacroix C. 2009. New in vitro colonic fermentation model

for Salmonella infection in the child gut. FEMS Microbiol. Ecol. 67:198-207.

168. Zihler A, Gagnon M, Chassard C, Hegland A, Stevens MJ, Braegger CP, Lacroix C.

2010. Unexpected consequences of administering bacteriocinogenic probiotic strains for

Salmonella populations, revealed by an in vitro colonic model of the child gut.

Microbiology. 156:3342-3353.

169. Kheadr E, Zihler A, Dabour N, Lacroix C, Le Blay G, Fliss I. 2010. Study of the

physicochemical and biological stability of pediocin PA-1 in the upper gastrointestinal

tract conditions using a dynamic in vitro model. J. Appl. Microbiol. 109:54-64.

170. Gagnon M, Zihler A, Chassard C, Lacroix C. 2011. Ecology of probiotics and enteric

protection, p. 65-85. In Malago JJ, Konikx JFJG, Marinsek-Loger R (eds.), Probiotic and

Enteric Infections: Cryoprotection by Probiotic Bacteria, 1st ed. Springer

Science+Business Media BV.

171. Freter R, Stauffer E, Cleven D, Holdeman LV, Moore WE. 1983. Continuous-flow

cultures as in vitro models of the ecology of large intestinal flora. Infect. Immun. 39:666-

675.

172. Macfarlane GT, Macfarlane S, Gibson GR. 1998. Validation of a Three-Stage

Compound Continuous Culture System for Investigating the Effect of Retention Time on

the Ecology and Metabolism of Bacteria in the Human Colon. Microb. Ecol. 35:180-187.

173. Minekus M, Marteau P, Havenaar R, Huis in 't Veld, J.H.H. 1995. A multi-

compartmental dynamic computer-controlled model simulating the stomach and small

intestine. Altern. Lab. Anim. 23:197-209.

174. Minekus M, Smeets-Peeters M, Bernalier A, Marol-Bonnin S, Havenaar R, Marteau

P, Alric M, Fonty G, Huis in't Veld JH. 1999. A computer-controlled system to simulate

conditions of the large intestine with peristaltic mixing, water absorption and absorption of

fermentation products. Appl. Microbiol. Biotechnol. 53:108-114.

175. Blanquet-Diot S, Soufi M, Rambeau M, Rock E, Alric M. 2009. Digestive stability of

xanthophylls exceeds that of carotenes as studied in a dynamic in vitro gastrointestinal

system. J. Nutr. 139:876-883.

239

176. Souliman S, Beyssac E, Cardot JM, Denis S, Alric M. 2007. Investigation of the

biopharmaceutical behavior of theophylline hydrophilic matrix tablets using USP methods

and an artificial digestive system. Drug Dev. Ind. Pharm. 33:475-483.

177. Souliman S, Blanquet S, Beyssac E, Cardot JM. 2006. A level A in vitro/in vivo

correlation in fasted and fed states using different methods: applied to solid immediate

release oral dosage form. Eur. J. Pharm. Sci. 27:72-79.

178. Molly K, Vande Woestyne M, Verstraete W. 1993. Development of a 5-step multi-

chamber reactor as a simulation of the human intestinal microbial ecosystem. Appl.

Microbiol. Biotechnol. 39:254-258.

179. Pompei A, Cordisco L, Raimondi S, Amaretti A, Pagnoni UM, Matteuzzi D, Rossi M.

2008. In vitro comparison of the prebiotic effects of two inulin-type fructans. Anaerobe.

14:280-286.

180. Gumienna M, Lasik M, Czarnecki Z. 2011. Bioconversion of grape and chokeberry

wine polyphenols during simulated gastrointestinal in vitro digestion. Int. J. Food Sci.

Nutr. 62:226-233.

181. Maccaferri S, Vitali B, Klinder A, Kolida S, Ndagijimana M, Laghi L, Calanni F,

Brigidi P, Gibson GR, Costabile A. 2010. Rifaximin modulates the colonic microbiota of

patients with Crohn's disease: an in vitro approach using a continuous culture colonic

model system. J. Antimicrob. Chemother. 65:2556-2565.

182. Duncan SH, Louis P, Thomson JM, Flint HJ. 2009. The role of pH in determining the

species composition of the human colonic microbiota. Environ. Microbiol. 11:2112-2122.

183. Kovatcheva-Datchary P, Egert M, Maathuis A, Rajilic-Stojanovic M, de Graaf AA,

Smidt H, de Vos WM, Venema K. 2009. Linking phylogenetic identities of bacteria to

starch fermentation in an in vitro model of the large intestine by RNA-based stable isotope

probing. Environ. Microbiol. 11:914-926.

184. Allison C, McFarlan C, MacFarlane GT. 1989. Studies on mixed populations of human

intestinal bacteria grown in single-stage and multistage continuous culture systems. Appl.

Environ. Microbiol. 55:672-678.

185. Fooks LJ, and Gibson GR. 2003. Mixed culture fermentation studies on the effects of

synbiotics on the human intestinal pathogens Campylobacter jejuni and Escherichia coli.

Anaerobe. 9:231-242.

186. Duncan SH, Scott KP, Ramsay AG, Harmsen HJ, Welling GW, Stewart CS, Flint HJ.

2003. Effects of alternative dietary substrates on competition between human colonic

bacteria in an anaerobic fermentor system. Appl. Environ. Microbiol. 69:1136-1142.

240

187. Lesmes U, Beards EJ, Gibson GR, Tuohy KM, Shimoni E. 2008. Effects of resistant

starch type III polymorphs on human colon microbiota and short chain fatty acids in

human gut models. J. Agric. Food Chem. 56:5415-5421.

188. De Boever P, Deplancke B, Verstraete W. 2000. Fermentation by gut microbiota

cultured in a simulator of the human intestinal microbial ecosystem is improved by

supplementing a soygerm powder. J. Nutr. 130:2599-2606.

189. Cinquin C, Le Blay G, Fliss I, Lacroix C. 2004. Immobilization of infant fecal

microbiota and utilization in an in vitro colonic fermentation model. Microb. Ecol. 48:128-

138.

190. Haug MC, Tanner SA, Lacroix C, Stevens MJ, Meile L. 2011. Monitoring horizontal

antibiotic resistance gene transfer in a colonic fermentation model. FEMS Microbiol. Ecol.

78:210-219.

191. Barcenilla A, Pryde SE, Martin JC, Duncan SH, Stewart CS, Henderson C, Flint HJ.

2000. Phylogenetic relationships of butyrate-producing bacteria from the human gut. Appl.

Environ. Microbiol. 66:1654-1661.

192. Derrien M, Vaughan EE, Plugge CM, de Vos WM. 2004. Akkermansia muciniphila

gen. nov., sp. nov., a human intestinal mucin-degrading bacterium. Int. J. Syst. Evol.

Microbiol. 54:1469-1476.

193. Strauss J, White A, Ambrose C, McDonald J, Allen-Vercoe E. 2008. Phenotypic and

genotypic analyses of clinical Fusobacterium nucleatum and Fusobacterium

periodonticum isolates from the human gut. Anaerobe. 14:301-309.

194. Cogan TA, Thomas AO, Rees LE, Taylor AH, Jepson MA, Williams PH, Ketley J,

Humphrey TJ. 2007. Norepinephrine increases the pathogenic potential of Campylobacter

jejuni. Gut. 56:1060-1065.

195. Peterson G, Kumar A, Gart E, Narayanan S. 2011. Catecholamines increase

conjugative gene transfer between enteric bacteria. Microb. Pathog. 51:1-8.

196. Kager L, Liljeqvist L, Malmborg AS, Nord CE. 1981. Effect of clindamycin

prophylaxis on the colonic microflora in patients undergoing colorectal surgery.

Antimicrob. Agents Chemother. 20:736-740.

197. Ferrera I, Sanchez O, Mas J. 2004. A new non-aerated illuminated packed-column

reactor for the development of sulfide-oxidizing biofilms. Appl. Microbiol. Biotechnol.

64:659-664.

198. Shi L, Ardehali R, Caldwell K, Valint P. 2000. Mucin coating on polymeric material

surfaces to suppress bacterial adhesion. Colloid. Surface. B. 17:229-239.

241

199. Sanchez O, Ferrera I, Mas J. 2008. Laboratory Model Systems for the Study of

Microbial Communities, p. 87-114. In Van Dijk T (ed.), Microbial Ecology Research

Trends. Nova Biomedical Books, New York.

200. Tannock GW, Munro K, Harmsen HJ, Welling GW, Smart J, Gopal PK. 2000.

Analysis of the fecal microflora of human subjects consuming a probiotic product

containing Lactobacillus rhamnosus DR20. Appl. Environ. Microbiol. 66:2578-2588.

201. Muyzer G, de Waal EC, Uitterlinden AG. 1993. Profiling of complex microbial

populations by denaturing gradient gel electrophoresis analysis of polymerase chain

reaction-amplified genes coding for 16S rRNA. Appl. Environ. Microbiol. 59:695-700.

202. Marzorati M, Wittebolle L, Boon N, Daffonchio D, Verstraete W. 2008. How to get

more out of molecular fingerprints: practical tools for microbial ecology. Environ.

Microbiol. 10:1571-1581.

203. Possemiers S, Verthe K, Uyttendaele S, Verstraete W. 2004. PCR-DGGE-based

quantification of stability of the microbial community in a simulator of the human

intestinal microbial ecosystem. FEMS Microbiol. Ecol. 49:495-507.

204. Magurran AE. 2004. Measuring Biological Diversity. Blackwell Science Ltd., Malden,

MA, USA.

205. Gafan GP, Lucas VS, Roberts GJ, Petrie A, Wilson M, Spratt DA. 2005. Statistical

analyses of complex denaturing gradient gel electrophoresis profiles. J. Clin. Microbiol.

43:3971-3978.

206. Smith B, and Wilson JB. 1996. A consumer's guide to evenness indices. Oikos. 76:70-82.

207. Pielou EC. 1975. Ecological diversity. Wiley, New York.

208. Fromin N, Hamelin J, Tarnawski S, Roesti D, Jourdain-Miserez K, Forestier N,

Teyssier-Cuvelle S, Gillet F, Aragno M, Rossi P. 2002. Statistical analysis of denaturing

gel electrophoresis (DGE) fingerprinting patterns. Environ. Microbiol. 4:634-643.

209. Rajilic-Stojanovic M, Maathuis A, Heilig HG, Venema K, de Vos WM, Smidt H.

2010. Evaluating the microbial diversity of an in vitro model of the human large intestine

by phylogenetic microarray analysis. Microbiology. 156:3270-3281.

210. Sonnenburg JL, Angenent LT, Gordon JI. 2004. Getting a grip on things: how do

communities of bacterial symbionts become established in our intestine? Nat. Immunol.

5:569-573.

211. Gibson GR, Cummings JH, Macfarlane GT. 1988. Use of a three-stage continuous

culture system to study the effect of mucin on dissimilatory sulfate reduction and

242

methanogenesis by mixed populations of human gut bacteria. Appl. Environ. Microbiol.

54:2750-2755.

212. Grootaert C, Van den Abbeele P, Marzorati M, Broekaert WF, Courtin CM, Delcour

JA, Verstraete W, Van de Wiele T. 2009. Comparison of prebiotic effects of

arabinoxylan oligosaccharides and inulin in a simulator of the human intestinal microbial

ecosystem. FEMS Microbiol. Ecol. 69:231-242.

213. Possemiers S, Rabot S, Espin JC, Bruneau A, Philippe C, Gonzalez-Sarrias A,

Heyerick A, Tomas-Barberan FA, De Keukeleire D, Verstraete W. 2008. Eubacterium

limosum activates isoxanthohumol from hops (Humulus lupulus L.) into the potent

phytoestrogen 8-prenylnaringenin in vitro and in rat intestine. J. Nutr. 138:1310-1316.

214. Feria-Gervasio D, Denis S, Alric M, Brugere JF. 2011. In vitro maintenance of a human

proximal colon microbiota using the continuous fermentation system P-ECSIM. Appl.

Microbiol. Biotechnol. 91:1425-1433.

215. MacArthur R. 1955. Fluctuations of Animal Populations and a Measure of Community

Stability. Ecology. 36:533-536.

216. Naeem S, and Li SB. 1997. Biodiversity enhances ecosystem reliability. Nature. 390:507-

509.

217. Anwar H, Strap JL, Chen K, Costerton JW. 1992. Dynamic interactions of biofilms of

mucoid Pseudomonas aeruginosa with tobramycin and piperacillin. Antimicrob. Agents

Chemother. 36:1208-1214.

218. Anwar H, van Biesen T, Dasgupta M, Lam K, Costerton JW. 1989. Interaction of

biofilm bacteria with antibiotics in a novel in vitro chemostat system. Antimicrob. Agents

Chemother. 33:1824-1826.

219. Clemente JC, Ursell LK, Parfrey LW, Knight R. 2012. The impact of the gut

microbiota on human health: an integrative view. Cell. 148:1258-1270.

220. Hill DA, Hoffmann C, Abt MC, Du Y, Kobuley D, Kirn TJ, Bushman FD, Artis D.

2010. Metagenomic analyses reveal antibiotic-induced temporal and spatial changes in

intestinal microbiota with associated alterations in immune cell homeostasis. Mucosal

Immunol. 3:148-158.

221. Charmandari E, Tsigos C, Chrousos G. 2005. Endocrinology of the stress response.

Annu. Rev. Physiol. 67:259-284.

222. Goyal RK, and Hirano I. 1996. The enteric nervous system. N. Engl. J. Med. 334:1106-

1115.

243

223. Moser G, Maeir-Dobersberger T, Vogelsang H, Lochs H. 1993. Inflammatory bowel

disease: Patient’s beliefs about the etiology of their disease-a controlled study. Psychosom.

Med. 55:131.

224. Harris A, and Martin BJ. 1994. Increased abdominal pain during final examinations.

Dig. Dis. Sci. 39:104-108.

225. Lyte M. 2004. Microbial endocrinology and infectious disease in the 21st century. Trends

Microbiol. 12:14-20.

226. Lyte M, and Ernst S. 1992. Catecholamine induced growth of gram negative bacteria.

Life Sci. 50:203-212.

227. Lyte M, Frank CD, Green BT. 1996. Production of an autoinducer of growth by

norepinephrine cultured Escherichia coli O157:H7. FEMS Microbiol. Lett. 139:155-159.

228. Clarke MB, Hughes DT, Zhu C, Boedeker EC, Sperandio V. 2006. The QseC sensor

kinase: a bacterial adrenergic receptor. Proc. Natl. Acad. Sci. U. S. A. 103:10420-10425.

229. Hughes DT, and Sperandio V. 2008. Inter-kingdom signalling: communication between

bacteria and their hosts. Nat. Rev. Microbiol. 6:111-120.

230. Bansal T, Englert D, Lee J, Hegde M, Wood TK, Jayaraman A. 2007. Differential

effects of epinephrine, norepinephrine, and indole on Escherichia coli O157:H7

chemotaxis, colonization, and gene expression. Infect. Immun. 75:4597-4607.

231. Jernberg C, Lofmark S, Edlund C, Jansson JK. 2007. Long-term ecological impacts of

antibiotic administration on the human intestinal microbiota. ISME J. 1:56-66.

232. Freeman J, Baines SD, Jabes D, Wilcox MH. 2005. Comparison of the efficacy of

ramoplanin and vancomycin in both in vitro and in vivo models of clindamycin-induced

Clostridium difficile infection. J. Antimicrob. Chemother. 56:717-725.

233. Hegde M, Wood TK, Jayaraman A. 2009. The neuroendocrine hormone norepinephrine

increases Pseudomonas aeruginosa PA14 virulence through the las quorum-sensing

pathway. Appl. Microbiol. Biotechnol. 84:763-776.

234. Naresh R, and Hampson DJ. 2011. Exposure to norepinephrine enhances Brachyspira

pilosicoli growth, attraction to mucin and attachment to Caco-2 cells. Microbiology.

157:543-547.

235. Lydiard RB, Fossey MD, Marsh W, Ballenger JC. 1993. Prevalence of psychiatric

disorders in patients with irritable bowel syndrome. Psychosomatics. 34:229-234.

244

236. Nemoto H, Kataoka K, Ishikawa H, Ikata K, Arimochi H, Iwasaki T, Ohnishi Y,

Kuwahara T, Yasutomo K. 2012. Reduced diversity and imbalance of fecal microbiota in

patients with ulcerative colitis. Dig. Dis. Sci. 57:2955-2964.

237. Mawdsley JE, and Rampton DS. 2005. Psychological stress in IBD: new insights into

pathogenic and therapeutic implications. Gut. 54:1481-1491.

238. Lyte M, Vulchanova L, Brown DR. 2011. Stress at the intestinal surface: catecholamines

and mucosa-bacteria interactions. Cell Tissue Res. 343:23-32.

239. Tsavkelova EA, Botvinko IV, Kudrin VS, Oleskin AV. 2000. Detection of

neurotransmitter amines in microorganisms with the use of high-performance liquid

chromatography. Dokl. Biochem. 372:115-117.

240. Hernandez-Romero D, Sanchez-Amat A, Solano F. 2006. A tyrosinase with an

abnormally high tyrosine hydroxylase/dopa oxidase ratio. FEBS J. 273:257-270.

241. Solano F, Garcia E, Perez D, Sanchez-Amat A. 1997. Isolation and Characterization of

Strain MMB-1 (CECT 4803), a Novel Melanogenic Marine Bacterium. Appl. Environ.

Microbiol. 63:3499-3506.

242. Freestone PP, and Lyte M. 2008. Microbial endocrinology: experimental design issues in

the study of interkingdom signalling in infectious disease. Adv. Appl. Microbiol. 64:75-

105.

243. Freestone PP, Lyte M, Neal CP, Maggs AF, Haigh RD, Williams PH. 2000. The

mammalian neuroendocrine hormone norepinephrine supplies iron for bacterial growth in

the presence of transferrin or lactoferrin. J. Bacteriol. 182:6091-6098.

244. Andrews SC, Robinson AK, Rodriguez-Quinones F. 2003. Bacterial iron homeostasis.

FEMS Microbiol. Rev. 27:215-237.

245. Freestone PP, Williams PH, Haigh RD, Maggs AF, Neal CP, Lyte M. 2002. Growth

stimulation of intestinal commensal Escherichia coli by catecholamines: a possible

contributory factor in trauma-induced sepsis. Shock. 18:465-470.

246. Freestone PP, Haigh RD, Williams PH, Lyte M. 2003. Involvement of enterobactin in

norepinephrine-mediated iron supply from transferrin to enterohaemorrhagic Escherichia

coli. FEMS Microbiol. Lett. 222:39-43.

247. Freestone PP, Haigh RD, Lyte M. 2007. Specificity of catecholamine-induced growth in

Escherichia coli O157:H7, Salmonella enterica and Yersinia enterocolitica. FEMS

Microbiol. Lett. 269:221-228.

245

248. Freestone PP, Haigh RD, Lyte M. 2007. Blockade of catecholamine-induced growth by

adrenergic and dopaminergic receptor antagonists in Escherichia coli O157:H7,

Salmonella enterica and Yersinia enterocolitica. BMC Microbiol. 7:8.

249. Lyte M, and Bailey MT. 1997. Neuroendocrine-bacterial interactions in a neurotoxin-

induced model of trauma. J. Surg. Res. 70:195-201.

250. Bailey MT, Engler H, Sheridan JF. 2006. Stress induces the translocation of cutaneous

and gastrointestinal microflora to secondary lymphoid organs of C57BL/6 mice. J.

Neuroimmunol. 171:29-37.

251. Griffiths EA, Duffy LC, Schanbacher FL, Dryja D, Leavens A, Neiswander RL, Qiao

H, DiRienzo D, Ogra P. 2003. In vitro growth responses of bifidobacteria and

enteropathogens to bovine and human lactoferrin. Dig. Dis. Sci. 48:1324-1332.

252. Lyte M, and Freestone P. 2009. Microbial Endocrinology Comes of Age. Microbe.

4:169-175.

253. Bansal T, Englert D, Lee J, Hegde M, Wood TK, Jayaraman A. 2007. Differential

effects of epinephrine, norepinephrine, and indole on Escherichia coli O157:H7

chemotaxis, colonization, and gene expression. Infect. Immun. 75:4597-4607.

254. Green BT, Lyte M, Kulkarni-Narla A, Brown DR. 2003. Neuromodulation of

enteropathogen internalization in Peyer's patches from porcine jejunum. J. Neuroimmunol.

141:74-82.

255. Brown DR, and Price LD. 2008. Catecholamines and sympathomimetic drugs decrease

early Salmonella Typhimurium uptake into porcine Peyer's patches. FEMS Immunol. Med.

Microbiol. 52:29-35.

256. Lyte M, Arulanandam B, Nguyen K, Frank C, Erickson A, Francis D. 1997.

Norepinephrine induced growth and expression of virulence associated factors in

enterotoxigenic and enterohemorrhagic strains of Escherichia coli. Adv. Exp. Med. Biol.

412:331-339.

257. Lyte M, Erickson AK, Arulanandam BP, Frank CD, Crawford MA, Francis DH.

1997. Norepinephrine-induced expression of the K99 pilus adhesin of enterotoxigenic

Escherichia coli. Biochem. Biophys. Res. Commun. 232:682-686.

258. Hendrickson BA, Guo J, Laughlin R, Chen Y, Alverdy JC. 1999. Increased type 1

fimbrial expression among commensal Escherichia coli isolates in the murine cecum

following catabolic stress. Infect. Immun. 67:745-753.

259. Zoetendal EG, von Wright A, Vilpponen-Salmela T, Ben-Amor K, Akkermans AD,

de Vos WM. 2002. Mucosa-associated bacteria in the human gastrointestinal tract are

246

uniformly distributed along the colon and differ from the community recovered from feces.

Appl. Environ. Microbiol. 68:3401-3407.

260. Durban A, Abellan JJ, Jimenez-Hernandez N, Ponce M, Ponce J, Sala T, D'Auria G,

Latorre A, Moya A. 2011. Assessing gut microbial diversity from feces and rectal

mucosa. Microb. Ecol. 61:123-133.

261. Van den Abbeele P, Van de Wiele T, Verstraete W, Possemiers S. 2011. The host

selects mucosal and luminal associations of coevolved gut microorganisms: a novel

concept. FEMS Microbiol. Rev. 35:681-704.

262. Van den Abbeele P, Roos S, Eeckhaut V, MacKenzie DA, Derde M, Verstraete W,

Marzorati M, Possemiers S, Vanhoecke B, Van Immerseel F, Van de Wiele T. 2012.

Incorporating a mucosal environment in a dynamic gut model results in a more

representative colonization by lactobacilli. Microb. Biotechnol. 5:106-115.

263. Van den Abbeele P, Belzer C, Goossens M, Kleerebezem M, De Vos WM, Thas O, De

Weirdt R, Kerckhof FM, Van de Wiele T. 2012. Butyrate-producing Clostridium cluster

XIVa species specifically colonize mucins in an in vitro gut model. ISME J. Epub ahead of

print.

264. Makelainen H, Hasselwander O, Rautonen N, Ouwehand AC. 2009. Panose, a new

prebiotic candidate. Lett. Appl. Microbiol. 49:666-672.

265. Sun J, Zhou TT, Le GW, Shi YH. 2010. Association of Lactobacillus acidophilus with

mice Peyer's patches. Nutrition. 26:1008-1013.

266. Vermeiren J, Van den Abbeele P, Laukens D, Vigsnaes LK, De Vos M, Boon N, Van

de Wiele T. 2012. Decreased colonization of fecal Clostridium coccoides/Eubacterium

rectale species from ulcerative colitis patients in an in vitro dynamic gut model with mucin

environment. FEMS Microbiol. Ecol. 79:685-696.

267. Macfarlane S, Woodmansey EJ, Macfarlane GT. 2005. Colonization of mucin by

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

continuous culture system. Appl. Environ. Microbiol. 71:7483-7492.

268. Probert HM, and Gibson GR. 2004. Development of a fermentation system to model

sessile bacterial populations in the human colon. Biofilms. 1:13-19.

269. Van den Abbeele P, Grootaert C, Possemiers S, Verstraete W, Verbeken K, Van de

Wiele T. 2009. In vitro model to study the modulation of the mucin-adhered bacterial

community. Appl. Microbiol. Biotechnol. 83:349-359.

270. Ouwehand A, and Salminen S. 2003. In vitro Adhesion Assays for Probiotics and their in

vivo Relevance: A Review. Microb. Ecol. Health D. 15:175-184.

247

271. Frank DN, St Amand AL, Feldman RA, Boedeker EC, Harpaz N, Pace NR. 2007.

Molecular-phylogenetic characterization of microbial community imbalances in human

inflammatory bowel diseases. Proc. Natl. Acad. Sci. U. S. A. 104:13780-13785.

272. Hong PY, Croix JA, Greenberg E, Gaskins HR, Mackie RI. 2011. Pyrosequencing-

based analysis of the mucosal microbiota in healthy individuals reveals ubiquitous bacterial

groups and micro-heterogeneity. PLoS One. 6:e25042.

273. Shen XJ, Rawls JF, Randall T, Burcal L, Mpande CN, Jenkins N, Jovov B, Abdo Z,

Sandler RS, Keku TO. 2010. Molecular characterization of mucosal adherent bacteria and

associations with colorectal adenomas. Gut Microbes. 1:138-147.

274. Wang Y, Antonopoulos DA, Zhu X, Harrell L, Hanan I, Alverdy JC, Meyer F, Musch

MW, Young VB, Chang EB. 2010. Laser capture microdissection and metagenomic

analysis of intact mucosa-associated microbial communities of human colon. Appl.

Microbiol. Biotechnol. 88:1333-1342.

275. Jackson C, Roden E, Churchill P. 2000. Denaturing Gradient Gel Electrophoresis Can

Fail to Separate 16S rDNA Fragments with Multiple Base Differences. Mol. Biol. Today.

1:49-51.

276. Gafan GP, and Spratt DA. 2005. Denaturing gradient gel electrophoresis gel expansion

(DGGEGE)--an attempt to resolve the limitations of co-migration in the DGGE of complex

polymicrobial communities. FEMS Microbiol. Lett. 253:303-307.

277. Oteo J, Aracil B, Alos JI, Gomez-Garces JL. 2000. High prevalence of resistance to

clindamycin in Bacteroides fragilis group isolates. J. Antimicrob. Chemother. 45:691-693.

278. Smith CJ. 1985. Characterization of Bacteroides ovatus plasmid pBI136 and structure of

its clindamycin resistance region. J. Bacteriol. 161:1069-1073.

279. Nyberg SD, Osterblad M, Hakanen AJ, Lofmark S, Edlund C, Huovinen P, Jalava J.

2007. Long-term antimicrobial resistance in Escherichia coli from human intestinal

microbiota after administration of clindamycin. Scand. J. Infect. Dis. 39:514-520.

280. Peterfreund GL, Vandivier LE, Sinha R, Marozsan AJ, Olson WC, Zhu J, Bushman

FD. 2012. Succession in the gut microbiome following antibiotic and antibody therapies

for Clostridium difficile. PLoS One. 7:e46966.

281. Reeves AE, Theriot CM, Bergin IL, Huffnagle GB, Schloss PD, Young VB. 2011. The

interplay between microbiome dynamics and pathogen dynamics in a murine model of

Clostridium difficile Infection. Gut Microbes. 2:145-158.

282. Buffie CG, Jarchum I, Equinda M, Lipuma L, Gobourne A, Viale A, Ubeda C,

Xavier J, Pamer EG. 2012. Profound alterations of intestinal microbiota following a

248

single dose of clindamycin results in sustained susceptibility to Clostridium difficile-

induced colitis. Infect. Immun. 80:62-73.

283. Edwards CA, Duerden BI, Read NW. 1986. Effect of clindamycin on the ability of a

continuous culture of colonic bacteria to ferment carbohydrate. Gut. 27:411-417.

284. Freter R, O'Brien PC, Macsai MS. 1981. Role of chemotaxis in the association of motile

bacteria with intestinal mucosa: in vivo studies. Infect. Immun. 34:234-240.

285. Freter R, Allweiss B, O'Brien PC, Halstead SA, Macsai MS. 1981. Role of chemotaxis

in the association of motile bacteria with intestinal mucosa: in vitro studies. Infect. Immun.

34:241-249.

286. Nomura T, Ohkusa T, Okayasu I, Yoshida T, Sakamoto M, Hayashi H, Benno Y,

Hirai S, Hojo M, Kobayashi O, Terai T, Miwa H, Takei Y, Ogihara T, Sato N. 2005.

Mucosa-associated bacteria in ulcerative colitis before and after antibiotic combination

therapy. Aliment. Pharmacol. Ther. 21:1017-1027.

287. Swidsinski A, Loening-Baucke V, Bengmark S, Scholze J, Doerffel Y. 2008. Bacterial

biofilm suppression with antibiotics for ulcerative and indeterminate colitis: consequences

of aggressive treatment. Arch. Med. Res. 39:198-204.

288. Merga JY, Leatherbarrow AJ, Winstanley C, Bennett M, Hart CA, Miller WG,

Williams NJ. 2011. Comparison of Arcobacter isolation methods, and diversity of

Arcobacter spp. in Cheshire, United Kingdom. Appl. Environ. Microbiol. 77:1646-1650.

289. Forde BM, and O'Toole PW. 2013. Next-generation sequencing technologies and their

impact on microbial genomics. Brief Funct. Genomics. Epub ahead of print.

290. Gilbert JA, and Hughes M. 2011. Gene expression profiling: metatranscriptomics.

Methods Mol. Biol. 733:195-205.

291. Martin FP, Dumas ME, Wang Y, Legido-Quigley C, Yap IK, Tang H, Zirah S,

Murphy GM, Cloarec O, Lindon JC, Sprenger N, Fay LB, Kochhar S, van Bladeren

P, Holmes E, Nicholson JK. 2007. A top-down systems biology view of microbiome-

mammalian metabolic interactions in a mouse model. Mol. Syst. Biol. 3:112.

292. Mahowald MA, Rey FE, Seedorf H, Turnbaugh PJ, Fulton RS, Wollam A, Shah N,

Wang C, Magrini V, Wilson RK, Cantarel BL, Coutinho PM, Henrissat B, Crock

LW, Russell A, Verberkmoes NC, Hettich RL, Gordon JI. 2009. Characterizing a

model human gut microbiota composed of members of its two dominant bacterial phyla.

Proc. Natl. Acad. Sci. U. S. A. 106:5859-5864.

293. Rezzonico E, Mestdagh R, Delley M, Combremont S, Dumas ME, Holmes E,

Nicholson J, Bibiloni R. 2011. Bacterial adaptation to the gut environment favors

249

successful colonization: microbial and metabonomic characterization of a simplified

microbiota mouse model. Gut Microbes. 2:307-318.

294. Lawley TD, Clare S, Walker AW, Stares MD, Connor TR, Raisen C, Goulding D,

Rad R, Schreiber F, Brandt C, Deakin LJ, Pickard DJ, Duncan SH, Flint HJ, Clark

TG, Parkhill J, Dougan G. 2012. Targeted Restoration of the Intestinal Microbiota with a

Simple, Defined Bacteriotherapy Resolves Relapsing Clostridium difficile Disease in Mice.

PLoS Pathog. 8:e1002995.

295. Robinson CJ, Bohannan BJ, Young VB. 2010. From structure to function: the ecology

of host-associated microbial communities. Microbiol. Mol. Biol. Rev. 74:453-476.

296. Jiang L, and Pu Z. 2009. Different effects of species diversity on temporal stability in

single-trophic and multitrophic communities. Am. Nat. 174:651-659.

297. Lehman CL, and Tilman D. 2000. Biodiversity, stability, and productivity in competitive

communities. Amer. Nat. 156:534-552.

298. Girvan MS, Campbell CD, Killham K, Prosser JI, Glover LA. 2005. Bacterial diversity

promotes community stability and functional resilience after perturbation. Environ.

Microbiol. 7:301-313.

299. Reed HE, and Martiny JB. 2007. Testing the functional significance of microbial

composition in natural communities. FEMS Microbiol. Ecol. 62:161-170.

300. Yachi S, and Loreau M. 1999. Biodiversity and ecosystem productivity in a fluctuating

environment: the insurance hypothesis. Proc. Natl. Acad. Sci. U. S. A. 96:1463-1468.

301. Claesson MJ, O'Sullivan O, Wang Q, Nikkila J, Marchesi JR, Smidt H, de Vos WM,

Ross RP, O'Toole PW. 2009. Comparative analysis of pyrosequencing and a phylogenetic

microarray for exploring microbial community structures in the human distal intestine.

PLoS One. 4:e6669.

302. Fux CA, Shirtliff M, Stoodley P, Costerton JW. 2005. Can laboratory reference strains

mirror "real-world" pathogenesis? Trends Microbiol. 13:58-63.

303. Tvede M, and Rask-Madsen J. 1989. Bacteriotherapy for chronic relapsing Clostridium

difficile diarrhoea in six patients. Lancet. 1:1156-1160.

304. Huebner ES, and Surawicz CM. 2006. Treatment of recurrent Clostridium difficile

diarrhea. J. Gastroen. Hepatol. 2:203-208.

305. Gough E, Shaikh H, Manges AR. 2011. Systematic review of intestinal microbiota

transplantation (fecal bacteriotherapy) for recurrent Clostridium difficile infection. Clin.

Infect. Dis. 53:994-1002.

250

306. Louis P, Young P, Holtrop G, Flint HJ. 2010. Diversity of human colonic butyrate-

producing bacteria revealed by analysis of the butyryl-CoA:acetate CoA-transferase gene.

Environ. Microbiol. 12:304-314.

307. Belenguer A, Duncan SH, Calder AG, Holtrop G, Louis P, Lobley GE, Flint HJ.

2006. Two routes of metabolic cross-feeding between Bifidobacterium adolescentis and

butyrate-producing anaerobes from the human gut. Appl. Environ. Microbiol. 72:3593-

3599.

308. Human Microbiome Project Consortium. 2012. A framework for human microbiome

research. Nature. 486:215-221.

309. Waldram A, Holmes E, Wang Y, Rantalainen M, Wilson ID, Tuohy KM, McCartney

AL, Gibson GR, Nicholson JK. 2009. Top-down systems biology modeling of host

metabotype-microbiome associations in obese rodents. J. Proteome Res. 8:2361-2375.

310. Urich T, Lanzen A, Qi J, Huson DH, Schleper C, Schuster SC. 2008. Simultaneous

assessment of soil microbial community structure and function through analysis of the

meta-transcriptome. PLoS One. 3:e2527.

311. Frias-Lopez J, Shi Y, Tyson GW, Coleman ML, Schuster SC, Chisholm SW, Delong

EF. 2008. Microbial community gene expression in ocean surface waters. Proc. Natl.

Acad. Sci. U. S. A. 105:3805-3810.

312. van Duynhoven J, Vaughan EE, Jacobs DM, Kemperman RA, van Velzen EJ, Gross

G, Roger LC, Possemiers S, Smilde AK, Dore J, Westerhuis JA, Van de Wiele T.

2011. Metabolic fate of polyphenols in the human superorganism. Proc. Natl. Acad. Sci. U.

S. A. 108 Suppl 1:4531-4538.

313. Gentry TJ, Wickham GS, Schadt CW, He Z, Zhou J. 2006. Microarray applications in

microbial ecology research. Microb. Ecol. 52:159-175.

314. Muyzer G. 1999. DGGE/TGGE a method for identifying genes from natural ecosystems.

Curr. Opin. Microbiol. 2:317-322.

315. Wittebolle L, Van Vooren N, Verstraete W, Boon N. 2009. High reproducibility of

ammonia-oxidizing bacterial communities in parallel sequential batch reactors. J. Appl.

Microbiol. 107:385-394.

316. Wittebolle L, Verstraete W, Boon N. 2009. The inoculum effect on the ammonia-

oxidizing bacterial communities in parallel sequential batch reactors. Water Res. 43:4149-

4158.

317. Fernandez A, Huang S, Seston S, Xing J, Hickey R, Criddle C, Tiedje J. 1999. How

stable is stable? Function versus community composition. Appl. Environ. Microbiol.

65:3697-3704.

251

318. Mariat D, Firmesse O, Levenez F, Guimaraes V, Sokol H, Dore J, Corthier G, Furet

JP. 2009. The Firmicutes/Bacteroidetes ratio of the human microbiota changes with age.

BMC Microbiol. 9:123.

319. Saxelin M, and Kajander K. 2009. Commercially Available Human Probiotic

Microorganisms: Lactobacillus rhamnosus GG, LGG, p. 469-470. In Kun Lee Y and

Salminen S (eds.), Handbook of Probiotics and Prebiotics, 2nd ed. John Wiley & Sons,

Inc., Canada.

320. Alander M, Satokari R, Korpela R, Saxelin M, Vilpponen-Salmela T, Mattila-

Sandholm T, von Wright A. 1999. Persistence of colonization of human colonic mucosa

by a probiotic strain, Lactobacillus rhamnosus GG, after oral consumption. Appl. Environ.

Microbiol. 65:351-354.

321. Dewulf EM, Cani PD, Neyrinck AM, Possemiers S, Van Holle A, Muccioli GG,

Deldicque L, Bindels LB, Pachikian BD, Sohet FM, Mignolet E, Francaux M,

Larondelle Y, Delzenne NM. 2011. Inulin-type fructans with prebiotic properties

counteract GPR43 overexpression and PPARgamma-related adipogenesis in the white

adipose tissue of high-fat diet-fed mice. J. Nutr. Biochem. 22:712-722.

322. Cani PD, Neyrinck AM, Fava F, Knauf C, Burcelin RG, Tuohy KM, Gibson GR,

Delzenne NM. 2007. Selective increases of bifidobacteria in gut microflora improve high-

fat-diet-induced diabetes in mice through a mechanism associated with endotoxaemia.

Diabetologia. 50:2374-2383.

323. Cani PD, Possemiers S, Van de Wiele T, Guiot Y, Everard A, Rottier O, Geurts L,

Naslain D, Neyrinck A, Lambert DM, Muccioli GG, Delzenne NM. 2009. Changes in

gut microbiota control inflammation in obese mice through a mechanism involving GLP-2-

driven improvement of gut permeability. Gut. 58:1091-1103.

324. Neyrinck AM, Van Hee VF, Piront N, De Backer F, Toussaint O, Cani PD, Delzenne

NM. 2012. Wheat-derived arabinoxylan oligosaccharides with prebiotic effect increase

satietogenic gut peptides and reduce metabolic endotoxemia in diet-induced obese mice.

Nutr. Diabetes. 2:e28.

325. Bollinger RR, Everett ML, Palestrant D, Love SD, Lin SS, Parker W. 2003. Human

secretory immunoglobulin A may contribute to biofilm formation in the gut. Immunology.

109:580-587.

326. Bollinger RR, Everett ML, Wahl SD, Lee YH, Orndorff PE, Parker W. 2006.

Secretory IgA and mucin-mediated biofilm formation by environmental strains of

Escherichia coli: role of type 1 pili. Mol. Immunol. 43:378-387.

327. Deat E, Blanquet-Diot S, Jarrige JF, Denis S, Beyssac E, Alric M. 2009. Combining

the dynamic TNO-gastrointestinal tract system with a Caco-2 cell culture model:

252

application to the assessment of lycopene and alpha-tocopherol bioavailability from a

whole food. J. Agric. Food Chem. 57:11314-11320.

328. Bahrami B, Child MW, Macfarlane S, Macfarlane GT. 2011. Adherence and cytokine

induction in Caco-2 cells by bacterial populations from a three-stage continuous-culture

model of the large intestine. Appl. Environ. Microbiol. 77:2934-2942.

329. Zihler A. 2010. In vitro assessment of bacteriocinogenic probiotics for prevention and

treatment of Salmonella in children using novel in vitro continuous colonic fermentation

and cellular models. PhD thesis. ETH, Zürich.

330. Kim HJ, Huh D, Hamilton G, Ingber DE. 2012. Human gut-on-a-chip inhabited by

microbial flora that experiences intestinal peristalsis-like motions and flow. Lab. Chip.

12:2165-2174.

331. Marzorati M, Possemiers S, Van den Abbeele P, Van de Wiele T, Vanhoecke B,

Verstraete W. 2010. Technology and Method to Study Microbial Growth and Adhesion to

Host-Related Surface and the Host-Microbiota Interactions. WO/2010/118857.

332. Holmstrom K, Collins MD, Moller T, Falsen E, Lawson PA. 2004. Subdoligranulum

variabile gen. nov., sp. nov. from human feces. Anaerobe. 10:197-203.

253

Appendices

Appendix 1:

Supplementary Material from Chapter 2

APPENDIX 1.1: SUPPLEMENTARY FIGURES AND TABLES FROM CHAPTER 2

Figure A2.1 Maintenance of continuous flow of chemostat communities using gas overpressure.

a) Schematic diagram of a single chemostat vessel at 400 mL working volume. b) Culture

volume rises above the bottom of the waste pipe due to the addition of culture medium

(exaggerated for illustrative purposes). c) Gas overpressure forces liquid culture out of the vessel

via the waste pipe, reducing the culture volume back to 400 mL.

400 mL 400 mL

Media in Culture out

400 mL

a b c

Waste pipe

254

Figure A2.2 DGGE profiles comparing the effect of antifoam addition on a steady state

chemostat culture not previously exposed to antifoam. Unexposed steady state cultures were

95.7% similar to communities 1 day after beginning antifoam addition, and 97.7% similar 5 days

after beginning antifoam addition.

255

Figure A2.3 Example of DGGE profiles comparing the fecal slurry to the inoculum (post-spin

supernatant). Preparation of the fecal slurry and inoculum are described in the Materials and

Methods.

256

Table A2.1 Agar used to isolate bacterial strains from donor B feces for DEC and MET-1

formulations. For each recipe reagents were dissolved in ddH2O (to 1L), transferred to 2 x 500

mL bottles, sterilized by autoclaving at 121ºC for 30 min, and poured into petri dishes.

Agar growth medium Reagent Manufacturer Quantity

Fastidious anaerobe agar (FAA) +

defibrinated sheep blood1

FAA Acumedia 45.7 g

Defibrinated sheep blood Hemostat

Laboratories 50 mL

FAA + spent LS174T medium +

defibrinated sheep blood1,2

FAA Acumedia 45.7 g

Spent LS174T medium

(see Appendix 1.2) See Appendix 1.2 30 mL

Defibrinated sheep blood Hemostat

Laboratories 50 mL

FAA + sterile spent chemostat

medium + defibrinated sheep

blood1,3

FAA Acumedia 45.7 g

Spent chemostat medium3

(see Table 2.1) See Table 2.1 30 mL

Defibrinated sheep blood Hemostat

Laboratories 50 mL

Brain Heart Infusion agar

Brain Heart Infusion broth BD 37 g

Agar Thermo Fisher

Scientific 15 g

Wilkins–Chalgren agar

Wilkins–Chalgren broth Acumedia 33 g

Agar Thermo Fisher

Scientific 15 g

Clostridial Differential agar

Clostridial Differential

broth Sigma-Aldrich 27.5 g

Agar Thermo Fisher

Scientific 15 g

deMan, Rogosa & Sharpe agar

deMan, Rogosa & Sharpe

broth BD 55 g

Agar Thermo Fisher

Scientific 15 g

'Faecali' agar See Table A2.2

'Mucin' agar See Table A2.3 1FAA and water were autoclaved and cooled to 50 ºC before adding the supplement (blood, spent

LS174T medium, or spent chemostat culture). 2Spent LS174T medium was collected from LS174T tissue culture cells as described in

Appendix 1.2. 3Spent chemostat culture medium was collected from a chemostat vessel seeded with donor B

feces. Spent culture medium was filter sterilized through a 0.22µm filter and used as an

additional supplement for FAA plates to encourage the growth of fastidious microorganisms.

257

Table A2.2 Recipe for 'Fecali' growth medium (191). To prepare the growth medium, the

reagents listed in the table were combined and ddH2O was added to 1L. The mixture was

transferred to 2 x 500 mL bottles, and sterilized it in a liquid autoclave cycle for 45 minutes.

Once the media cooled to 50°C, 30mL filter-sterilized rumen fluid were added (University of

Guelph Animal and Poultry Science group) and it was poured into petri dishes.

Reagent Manufacturer Weight (g)

Casitone Difco (Detroit, Michigan) 10

Yeast Extract BD 2.5

NaHCO3 Fisher 4

Dextrose EMD (USA) 2

Cellobiose Sigma 2

Starch (from wheat, unmodified) Sigma-Aldrich 2

Cysteine Sigma 1

K2HPO4 Sigma 0.45

KH2PO4 Fisher 0.45

(NH4)2SO4 EM Science (Darmstadt, Germany) 0.9

NaCl Fisher 0.9

MgSO4 BDH 0.09

CaCl2 BDH 0.09

Agar Fisher 15

258

Table A2.3 Recipe for 'Mucin' growth medium (192). To prepare the growth medium, the

reagents listed in the table were combined and ddH2O was added to 1L. The mixture was then

transferred to 2 x 500 mL bottles, and sterilized in a liquid autoclave cycle for 45 minutes. Once

the media cooled to 50°C, 7 mL filter-sterilized rumen fluid and 2.5 mL mucin were added and

the mixture was poured into petri dishes.

Reagent Manufacturer Weight (g)

KH2PO4 Fisher 0.4

Na2HPO4 Sigma 0.53

NH4Cl BDH 0.3

NaCl Fisher 0.3

MgCl2.6H2O Sigma 0.1

CaCl2 BDH 0.11

NaHCO3 Fisher 4

Na2S•7-9H2O Acros (New Jersey, USA) 0.25

Porcine gastric mucin (Type II) Sigma-Aldrich 2.5

Agar Fisher 15

259

APPENDIX 1.2: PREPARATION OF SPENT LS174T GROWTH MEDIUM

1.2.1 Setting up a flask of LS174T tissue culture cells

1) LS174T media (Table A2.4) was warmed to 37°C in a heated water bath

2) The tissue culture hood was sterilized using a UV light for 15 minutes

3) In the tissue culture hood, 20 mL of fresh LS174T media and 40 µL of plasmosin

(InvivoGen; San Diego, California) were added to a sterile small (75 cm2) flask

4) A stock of frozen LS174T cells were removed from the liquid nitrogen dewar and thawed

slightly until it became a slurry (liquid with ice crystals)

5) LS174T cells were transferred to the flask

6) The flask was transferred to and incubated in a tissue culture cell incubator (37°C with

5% CO2)

1.2.2 Changing spent LS174T growth medium (media changed every 3-4 days)

1) LS174T media and 1X Versene (Table A2.5) were warmed to 37°C in a heated water

bath

2) The tissue culture hood was sterilized using a UV light for 15 minutes

3) Flask was transferred from the tissue culture cell incubator to the tissue culture hood

4) Spent LS174T media was pipetted out of the flask and stored in a sterile conical tube at

4°C until it was ready to use (see Table A2.1)

5) 30 mL of 1X Versene were added to the flask

6) 1X Versene bathed LS174T cells for ~1 minute

7) 1X Versene was pipetted out of the flask and discarded

260

8) 30 mL of fresh LS174T media and 60 µL of plasmosin were added to a new, sterile large

(150 cm2) flask

9) Flask was returned to and incubated in the tissue culture cell incubator

1.2.3 Splitting LS174T cells (Once cells reached confluence, after approximately one week of

growth)

1) LS147T media, 1X Versene and trypsin (Gibco, Life Technologies Corporation) were

warmed to 37°C in a heated water bath

2) The tissue culture hood was sterilized using a UV light for 15 minutes

3) The flask was transferred from the tissue culture cell incubator to the tissue culture hood

4) Spent LS174T media was pipetted out of the flask and stored in a sterile conical tube at

4°C until it was ready to use (see Table A2.1)

5) 30 mL of 1X Versene were added to the flask

6) 1X Versene bathed LS174T cells for ~1 minute

7) 1X Versene was pipetted out of the flask and discarded

8) 10 mL of trypsin were added to the flask

9) The flask was incubated in the tissue culture cell incubator for 5 minutes or until the

cells sloughed off from bottom of flask

10) 30 mL of fresh LS174T media and 60 µL of plasmosin were added to a new large flask

11) Once trypsinized, 4-5 mL of cells were added to the new flask, and pipetted up and

down several times to mix

12) The new flask was transferred to and incubated in the tissue culture cell incubator

261

Table A2.4 Recipe for LS174T tissue culture cell growth medium. To prepare the growth

medium, the reagents listed in the table were combined, ddH2O was added to 1L, and the mixture

was filter sterilized using a 0.22 µm filter. The prepared growth medium was stored at 4°C and

used within 3 months.

Reagent Manufacturer Volume or

Weight

DMEM/High Glucose Thermo Scientific 500mL

Sodium bicarbonate Thermo Fisher Scientific 2.2 g

HEPES BDH 4.766g

100 mM sodium pyruvate Cellgro, Mediatech Inc

(Manassas, Virginia) 10 mL

Heat inactivated fetal bovine serum (FBS,

see Appendix 1.2.4)

Gibco, Life Technologies

Corporation 100mL

Table A2.5 Recipe for 10x Versene. To prepare 10x versene, the reagents listed in the table were

combined, ddH2O was added to 1L, and the mixture was filter sterilized using a 0.22 µm filter.

Reagent Manufacturer Weight (g)

EDTA VWR 2

NaCl Fisher Scientific 8

Na2HPO4 Sigma Aldrich 11.5

KCl Fisher 2

KH2PO4 Fisher Scientific 2

dextrose Fisher Scientific 2

1.2.4 Preparation of heat inactivated FBS

1) 2 X 50 mL conical tubes of aliquoted FBS were removed from the freezer

2) Tubes were warmed to 37°C in a heating water bath (tubes were inverted every 10

minutes)

3) Heated water bath temperature was turned up to 56°C

4) After the temperature of the water bath reached 56°C, the tubes were incubated for an

additional 30 minutes (tubes were inverted every 10 minutes)

5) Tubes were removed and allowed to cool to room temperature

262

APPENDIX 1.3: FREEZING MEDIA PREPARATION AND STORAGE OF

BACTERIAL ISOLATES

1) 60 g fat free instant skim milk powder (Carnation; Markham, Ontario), 5 mL dimethyl

sulfoxide (DMSO; Fisher) and 5 mL glycerol (Fisher) were mixed.

2) ddH2O was added to 500 mL

3) Mixture was autoclaved at 121ºC for 15 minutes

4) Once cooled, freezing media was aliquoted into 1.5 mL cryovials and transferred into an

anaerobic chamber to degas

5) Bacterial colonies were scrapped from a petri dish and resuspended in the pre-reduced

freezing media

6) Cryovials were sealed and stored at -80ºC

263

APPENDIX 1.4: RUNNING DGGE GELS

1.4.1 Agarose gel electrophoresis to check DGGE PCR products

1) Agarose powder was dissolved in Tris/EDTA (TE) buffer (10 mM Tris-HCl pH 8.0 + 1

mM EDTA pH 8.0) at a final concentration of 2%

2) Agarose-TE mixture was heated until the agarose powder melted

3) Molten agarose was poured into a medium (10 cm x 10 cm) gel casting tray

4) 5 µL of ethidium bromide (0.5 µg/mL) were mixed into the molten agarose

5) Gel was allowed to solidify (approximately 20 minutes)

6) Gel was transferred to an electrophoresis tank containing 1x TAE buffer (see Appendix

1.4.2)

7) DGGE loading dye was added to the PCR products:

a. For chemostat samples: 2 µL of each PCR product were mixed with 3µL DGGE

loading dye

b. For DGGE standard ladder strains: 12.5 µL DGGE loading dye were added

directly to each of the 5 ladder sample PCR tubes

c. For PCR blank: add 5 µL DGGE loading dye were added directly to each blank

PCR tube

8) PCR product-dye mixture was loaded into the lanes of the agarose gel

a. For chemostat samples: 5µL of the mixture were loaded onto each lane

b. For DGGE standard ladder strains: 3 µL of the mixture were loaded onto each

lane

c. For PCR blank: 5 µL and 10 µL of the mixture were loaded onto each lane

264

9) 1 µL 100 bp DNA molecular size ladder (Invitrogen; Carlsbad, California) was added to

a lane of the agarose gel

10) Gel electrophoresis was carried out at 80V for 45 minutes

11) After electrophoresis was complete, the agarose gel was imaged using the SynGene G-

box gel documentation system using the Genesnap software

1.4.2 Preparation of 50x TAE (Tris-acetate-EDTA) running buffer

1) 242g Tris base were dissolved in approximately 750 mL ddH2O

2) 750 mL glacial acetic acid were added to the Tris solution

3) 100 mL 0.5 M EDTA (pH 8.0) were added to the mixture

4) ddH2O was added to 1L

1.4.3 Preparation of PCR samples to load onto DGGE gels

Three DGGE PCR reactions were concentrated for each sample using the EZ-10 Spin

Column PCR Products Purification Kit (Biobasic; Markham, Ontario).

1) A tube of sterile ddH2O was warmed to 50-60°C in a hot box

2) For each chemostat sample, three DGGE PCR reactions were pooled in a 1.5 mL tube

3) 450 µL Binding Buffer 1 were added to the tube

4) The mixture was transferred to the column (supplied as part of the kit), and incubated at

room temperature for 2 minutes

5) The tubes were centrifuged at 9725 x g for 2 minutes and the flow through was discarded

6) 500 µL of Wash Buffer were added to each column, the tubes were centrifuged at 9725 x

g for 2 minutes, and the flow through was discarded

265

7) Step 6 was repeated

8) The tubes were centrifuged at 14000 x g for 3 minutes

9) The columns were transferred into a new 1.5 mL tube and 45 µL warm ddH2O (heated in

the hot box) were added to each column

10) The columns were incubated at room temperature for 2 minutes

11) The tubes were centrifuged at 9725 x g for 2 minutes and the columns were discarded

12) 10 µL DGGE loading dye were added to each tube and the tubes were gently vortexed to

mix

13) Samples were stored at 4ºC until ready to load onto the DGGE gel

1.4.4 Pouring DGGE gels

1) Glass plates were cleaned using 70% EtOH

2) Plates were arranged with 1 mm spacers between big and small plates

3) Casters were slid on and tightened (using the guide card)

4) Glass plates were placed into the pouring stand

5) Two "low" and two "high" conicals of DGGE gel mixtures were prepared (for list of

reagents see Table A2.6)

6) Urea, formamide, bis-acrylamide, TAE, and water were added to a 50 mL conical and

mixed gently (until the urea dissolved)

7) The following reagents were added to one high and one low mixture and gently mixed:

- 35 µL of 2% bromophenol blue (added to the high mixture only)

- 95 µL 10% APS (added to both mixtures)

- 55 µL TEMED (added to both mixtures)

266

8) 16 mL of the low mixture were loaded into a 30 mL syringe and the syringe was attached

to the gradient pourer (on the low side)

9) 16 mL of the high mixture were loaded into another 30 mL syringe and the syringe was

attached to the gradient pourer (on the high side)

10) A y-connector was attached to the tubes attached to the high and low syringes

11) Gels were poured slowly using the gradient pourer and the comb was inserted

12) Steps 7-11 were repeated to pour the second gel

13) Gels were allowed to polymerize for at least 2 hours, then each gel was covered with a

moist paper towel and saran wrap and stored at 4ºC overnight

14) 140 mL 50x TAE was added to the DGGE running tank and dH2O was added to the

“FILL” line

15) The DGGE tank was set so it would automatically heat to 60ºC the following morning

Table A2.6 Composition of DGGE gel mixtures.

Reagent Low Mixture High Mixture

Urea 3.15 g 5.78 g

Formamide 3.0 mL 5.5 mL

Bis-acrylamide (37.5:1) 3.75 mL 3.75 mL

50x TAE 500 µL 500 µL

dH2O to 25 mL to 25 mL

1.4.5 Running DGGE gels

1) Slowly removed combs from the gels and cut away excess gel

2) Cleaned wells using 1x TAE, attached gels to gel core, placed the core into the DGGE

running tank, and heated to 60ºC

3) Loaded DGGE gel lanes

267

- Loaded 40 µL DGGE loading dye into the outside lanes

- Loaded 40 µL of the DGGE ladder into the first, middle, and last lanes (ladder

consisted of an equal mixture of the 5 ladder sample PCR reactions)

- Loaded 40 µL of each sample into the rest of the lanes

4) DGGE running tank buffer was reheated to 60ºC

5) DGGE gel was run at 120V for 5 hours (buffer circulating pump turned on 1 minute after

turning on the voltage)

6) Once the gel electrophoresis was complete the DGGE gels were placed into an ethidium

bromide bath (1L 1x TAE + 100 µL 0.5 µg/mL ethidium bromide) and stained for 10

minutes

7) Gels were transferred to a ddH2O water bath and destained for 10 minutes

8) Gels were transferred to the SynGene G-box gel documentation system and imaged using

the Genesnap software (saturated images were captured using the "autosat" feature)

268

Appendix 2:

Supplementary Figures and Tables from

Chapter 3

269

Figure A3.1 Redundancy analysis (RDA) at the genus-like group-level based on HITChip data of the fecal inocula and steady

state chemostat communities. The variation in the abundance of 44 genus-like level bacterial groups (represented by the arrows) is

explained to at least 30% by sample characteristics. Fecal inocula are significantly different from steady state communities as

calculated by Monte Carlo permutation test (p=0.002).

270

Table A3.1 Correlation coefficients (%SI) comparing the fecal inoculum (day 0) to the

corresponding steady state community (day 36).

Vessel %SI, Day 0 vs. Day 36

By DGGE By HITChip

VA1-1 45.3 68.8

VA1-2 49.1 67.8

VA2-1 41.6 56.4

VA2-2 39.4 61.7

VB3-1 55.1 68.5

VB3-2 54.9 68.0

VB4-1 48.5 66.1

VB4-2 46.5 67.4

VC5-1 37.5 76.7

VC5-2 35.7 71.9

271

Figure A3.2 Relative abundance of higher taxonomic groups (~ phylum/class level) based on HITChip analysis for twin-vessel

steady state (day 36) chemostat communities.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

VA1-1 VA1-2 VA2-1 VA2-2 VB3-1 VB3-2 VB4-1 VB4-2 VC5-1 VC5-2

Rel

ati

ve

Con

trib

uti

on

Actinobacteria

Bacteroidetes

Bacilli

Clostridium cluster I

Clostridium cluster III

Clostridium cluster IV

Clostridium cluster IX

Clostridium cluster XI

Clostridium cluster XIII

Clostridium cluster XIVa

Clostridium cluster XV

Clostridium cluster XVI

Clostridium cluster XVII

Clostridium cluster XVIII

Uncultured Clostridiales

Cyanobacteria

Fusobacteria

Proteobacteria

Spirochaetes

Asteroleplasma

Uncultured Mollicutes

Verrucomicrobia

272

Table A3.2 Correlation coefficients (%SI) comparing within-vessel reproducibility to between-

vessel repeatability (biological replicates) for donors A and B on days 0 and 36. NA = not

available.

Comparison %SI, by DGGE %SI, by HITChip

Day 0 Day 36 Day 0 Day 36

VA1-1 vs. VA1-2 99.8 94.2 NA 96.3

VA2-1 vs. VA2-2 97.4 95.8 NA 96.5

Run A1 vs. Run A2 88.1 72.7 94.3 90.6

VB3-1 vs. VB3-2 99.1 91.6 NA 97.5

VB4-1 vs. VB4-2 95.0 85.1 NA 95.4

Run B3 vs. Run B4 86.6 82.3 97.4 94.4

273

Table A3.3 Ecological parameters (diversity, evenness, and richness) for each vessel immediately following inoculation (day 0)

and during steady state (day 36) as measured using DGGE and the HITChip. Runs A1, A2, B3, and B4 were used to assess

biological repeatability and runs A1, B3 and C5 were used to compare between-donor differences.

Analysis Method Vessel Shannon Diversity Index (H’) Shannon Equitability Index (EH’) Richness (S)

Day 0 Day 36 Day 0 Day 36 Day 0 Day 36

DGGE

VA1-1 3.46 3.34 0.81 0.78 71 71

VA1-2 3.48 3.38 0.82 0.79 71 71

VA2-1 3.44 3.29 0.81 0.78 71 68

VA2-2 3.45 3.23 0.81 0.77 71 68

VB3-1 3.18 3.31 0.76 0.79 66 65

VB3-2 3.17 3.24 0.76 0.78 66 65

VB4-1 3.22 3.13 0.77 0.75 66 64

VB4-2 3.22 3.18 0.77 0.77 66 64

VC5-1 3.32 3.20 0.80 0.79 63 57

VC5-2 3.32 3.07 0.80 0.76 63 57

HITChip

VA1-1 5.93

5.45 0.82

0.81 1041

692

VA1-2 5.53 0.82 698

VA2-1 5.89

5.31 0.82

0.79 980

646

VA2-2 5.51 0.81 692

VB3-1 5.71

5.46 0.84

0.82 715

583

VB3-2 5.47 0.83 524

VB4-1 5.67

5.32 0.83

0.82 734

480

VB4-2 5.40 0.82 498

VC5-1 5.49

5.11 0.82

0.80 647

391

VC5-2 4.95 0.79 396

274

Table A3.4 Correlation coefficients (%SI) comparing runs seeded with feces from different

donors on days 0 and 36.

Comparison %SI by DGGE %SI by HITChip

Day 0 vs. Day 0 Day 36 vs. Day 36 Day 0 vs. Day 0 Day 36 vs. Day 36

Run A1 vs. Run B3 63.2 63.2 70.4 63.2

Run A1 vs. Run C5 40.2 53.1 64.9 70.3

Run B3 vs. Run C5 40.2 46.2 63.7 74.1

275

Appendix 3:

Supplementary Figures and Tables from

Chapter 5

Figure A5.1 Redundancy analysis (RDA) at the genus-like level based on HITChip

compositional data of planktonic and biofilm samples. The variation in the abundance of 27

genus-like bacterial groups shown (represented by the black arrows) is explained to at least 50%

by sample characteristics. Biofilm communities are significantly different from steady state

planktonic communities as calculated by the Monte Carlo permutation test (p=0.03). Red arrows

indicate shifts in V1-P samples following clindamycin treatment, while green arrows indicate

shifts in V2-P samples (control) following water addition. Light grey circles indicate clustering

of planktonic or biofilm samples not exposed to clindamycin.

-1.5 1.0

-1.0

1.0 V1/2 inoculum

V1-P Day 36

V2-P Day 36

V1-P Day 41

V2-P Day 41

V1-P Day 48

V2-P Day 48

V1-B1

V2-B1 V2-B2

V1-B3

V2-B3

V1-B4

V2-B4

Clostridium cluster XIVaClostridium cluster IV

Bacteroidetes

Proteobacteria

30%

14.2

%

V1-B2

276

Table A5.1 Clindamycin susceptibility profiles for 1Subdoligranulum variabile isolates and

2Faecalibacterium prausnitzii isolates (closest relative of S. variabile; 332). S = clindamycin

sensitive, R = clindamycin resistant. NA = not available.

Strain 0.2 µg clindamycin 2 µg clindamycin 4 µg clindamycin

22-5-S 19 D5 FAA1 R R R

6/1/47 FAA1 NA NA S

22-5-I 24 FAA1 S S S

16-6-I 40 FAA2 R R R

22-5-I 5 FAA NB2 R R R

277

Figure A5.2 Percent abundance of genus-like members (level 2) from the Bacteroidetes-like phylum (level 1) based on HITChip

analysis for the fecal inoculum, planktonic communities and biofilm communities. As seen here, clindamycin treatment caused a

sharp decline in Bacteroides diversity, which did not return to their starting composition following clindamycin treatment. These

results are comparable to data obtained from human subjects in a study by Jernberg et al. (231).

0

10

20

30

40

50

60

70

80

90

100

V1/2-P

inoculum

V1-P

Day 36

V2-P

Day 36

V1-P

Day 41

V2-P

Day 41

V1-P

Day 48

V2-P

Day 48

V1-B1 V2-B1 V1-B2 V2-B2 V1-B3 V2-B3 V1-B4 V2-B4

Per

cen

t a

bu

nd

an

ce (

%)

Allistipes et rel.

Bacteroides fragilis et rel.

Bacteroides intestinalis et rel.

Bacteroides ovatus et rel.

Bacteroides plebeius et rel.

Bacteroides splachnicus et rel.

Bacteroides stercoris et rel.

Bacteroides uniformis et rel.

Bacteroides vulgatus et rel.

Parabacteroides distasonis et rel.

Prevotella melaninogenica et rel.

Prevotella oralis et rel.

Prevotella ruminicola et rel.

Prevotella tannerae et rel.

Tannerella et rel.

Uncultured Bacteroidetes

278

Appendix 4:

Supplementary Table from Chapter 6

279

Table A6.1 Species-like groups that can be detected for each of the genus-like levels listed in Table 6.7 using the HITChip. N/A

= not applicable.

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Actinobacteria Bifidobacterium

Bifidobacterium adolescentis

Bifidobacterium adolescentis

Bifidobacterium longum

Bifidobacterium pseudocatenulatum

Bifidobacterium angulatum

Bifidobacterium animalis

Bifidobacterium bifidum

Bifidobacterium breve

Bifidobacterium catenulatum

Bifidobacterium dentium

Bifidobacterium gallicum

Bifidobacterium infantis

Bifidobacterium longum

Bifidobacterium pseudocatenulatum

Bifidobacterium pseudolongum

Bifidobacterium thermophilum

uncultured bacterium Adhufec069rbh

Uncultured bacterium clone Eldhufec085

Uncultured bacterium clone Eldhufec088

uncultured Bifidobacterium sp. 13D

uncultured Bifidobacterium sp. 15D

uncultured Bifidobacterium sp. 16B

uncultured Bifidobacterium sp. 16C

uncultured Bifidobacterium sp. 9A

uncultured Bifidobacterium sp. 9C

Bacteroidetes Alistipes et rel.

Alistipes finegoldii

Bacteroides spp.

Alistipes putredinis

Bacteroides sp. ANH 2438

Bacteroides sp. CJ44

Bacteroides sp. DSM 12148

uncultured bacterium adhufec52.25

uncultured bacterium C706

uncultured bacterium D080

uncultured bacterium HuCA7

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

280

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Bacteroidetes

cont.

Alistipes et rel.

cont.

uncultured bacterium LJ01

Bacteroides spp.

uncultured bacterium M162

uncultured bacterium M516

uncultured bacterium MG06

uncultured bacterium NH37

uncultured bacterium NN46

Bacteroides ovatus et rel.

Bacteroides caccae

Bacteroides ovatus

Bacteroides ovatus

Bacteroides sp. CS3

Uncultured bacterium clone Eldhufec030

uncultured bacterium HuCA34

uncultured bacterium HuCC30

uncultured bacterium LCLC24

uncultured bacterium NC94

uncultured bacterium NP35

Bacteroides splachnicus et rel.

uncultured bacterium NN84

Bacteroides spp. uncultured bacterium NP53

uncultured bacterium NX93

Bacteroides stercoris et rel.

bacterium adhufec303

Bacteroides spp.

Bacteroides eggerthii

Bacteroides stercoris

Uncultured bacterium clone Eldhufec025

Uncultured bacterium clone Eldhufec057

uncultured bacterium NO48

Bacteroides uniformis et rel. Bacteroides uniformis

Bacteroides spp. uncultured Bacteroides sp. NS2A11

Bacteroides vulgatus et rel.

bacterium adhufec27

Bacteroides vulgatus

Bacteroides vulgatus

uncultured bacterium ABLC36

uncultured bacterium ABLC8

uncultured bacterium ABLCf3

uncultured bacterium HuCA9

uncultured bacterium HuCB3

uncultured bacterium HuCB6

uncultured bacterium LCLC45

uncultured bacterium LCLC5

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

281

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Bacteroidetes

cont.

Bacteroides vulgatus et rel. cont.

uncultured bacterium LCLC6

Bacteroides vulgatus uncultured bacterium LCRC22

uncultured bacterium OLDA-A2

Parabacteroides distasonis et rel.

Parabacteroides distasonis

Parabacteroides distasonis

Parabacteroides goldsteinii

Parabacteroides merdae

uncultured bacterium ABLCf15

Uncultured bacterium clone Eldhufec042

uncultured bacterium LCLC20

uncultured bacterium LV88

uncultured bacterium M270

uncultured bacterium MH76

uncultured bacterium Y112

Firmicutes

Bacilli Streptococcus bovis et rel.

Streptococcus agalactiae

Streptococcus mitis

Streptococcus bovis

Streptococcus equi subsp. zooepidemicus

Streptococcus equinus

Streptococcus equisimilis

Streptococcus infantarius subsp. coli

Streptococcus pyogenes

Streptococcus salivarius

Streptococcus sp. CB1

Streptococcus thermophilus

Streptococcus uberis

uncultured bacterium OLDA-B7

Clostridium cluster IV Faecalibacterium prausnitzii et rel.

bacterium adhufec113

Faecalibacterium prausnitzii

bacterium adhufec218

bacterium adhufec365

butyrate-producing bacterium A2-165

Faecalibacterium prausnitzii

uncultured bacterium A10

uncultured bacterium ABLCf5

uncultured bacterium adhufec08.25

Uncultured bacterium clone Eldhufec227

Uncultured bacterium clone Eldhufec228

Uncultured bacterium clone Eldhufec252

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

282

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Firmicutes cont.

Clostridium cluster IV cont.

Faecalibacterium prausnitzii et rel.

cont.

Uncultured bacterium clone Eldhufec255

Faecalibacterium prausnitzii

Uncultured bacterium clone Eldhufec256

Uncultured bacterium clone Eldhufec259

Uncultured bacterium clone Eldhufec261

Uncultured bacterium clone Eldhufec276

Uncultured bacterium clone Eldhufec281

Uncultured bacterium clone Eldhufec282

Uncultured bacterium clone Eldhufec288

Uncultured bacterium clone Eldhufec289

uncultured bacterium HuCA11

uncultured bacterium KP66

uncultured bacterium L308

uncultured bacterium L420

uncultured bacterium M484

uncultured bacterium MG85

uncultured bacterium MJ65

uncultured bacterium MN45

uncultured bacterium NC10

Oscillospira guillermondii et rel.

bacterium adhufec269

Flavonifractor plautii

Eubacterium desmolans

uncultured bacterium A051

uncultured bacterium B811

uncultured bacterium C574

Uncultured bacterium clone Eldhufec223

Uncultured bacterium clone Eldhufec257

Uncultured bacterium clone Eldhufec283

Uncultured bacterium clone Eldhufec285

Uncultured bacterium clone Eldhufec286

Uncultured bacterium clone Eldhufec301

uncultured bacterium D134

uncultured bacterium D288

uncultured bacterium D440

uncultured bacterium HuCB7

uncultured bacterium HuRC81

uncultured bacterium LD59

uncultured bacterium LE02

uncultured bacterium MA30

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

283

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Firmicutes cont.

Clostridium cluster IV cont.

Oscillospira guillermondii et rel.

cont.

uncultured bacterium MM71

Flavonifractor plautii

Eubacterium desmolans

uncultured bacterium N139

uncultured bacterium OLDA-D11

uncultured bacterium OLDA-H2

uncultured bacterium OLDB-B7

uncultured bacterium V239

uncultured human gut bacterium JW1C11

uncultured human gut bacterium JW2D11

Papillibacter cinnamivorans et rel.

bacterium adhufec296

Flavonifractor plautii

Eubacterium desmolans

butyrate-producing bacterium A2-207

uncultured bacterium cadhufec32c10

Uncultured bacterium clone Eldhufec258

uncultured bacterium ZO15

uncultured Gram-positive bacterium NB2C3

uncultured Gram-positive bacterium NB5F9

uncultured Gram-positive bacterium NO2-21

uncultured Gram-positive bacterium NS2D7

uncultured Gram-positive bacterium NS2H9

Ruminococcus callidus et rel.

bacterium adhufec58

Faecalibacterium prausnitzii

Clostridium methylpentosum

Ruminococcus albus

Ruminococcus callidus

Ruminococcus flavefaciens

Uncultured bacterium clone Eldhufec284

uncultured bacterium D005

uncultured bacterium D739

uncultured bacterium D789

uncultured bacterium MF20

uncultured bacterium MH26

uncultured Gram-positive bacterium NS4G9

uncultured Ruminococcus sp. NB2F11

uncultured Ruminococcus sp. NO11

Subdoligranulum variable at rel.

bacterium A03

Faecalibacterium prausnitzii bacterium adhufec13

human intestinal firmicute CJ7

Subdoligranulum variabile

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

284

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Firmicutes cont.

Clostridium cluster IV

cont.

Subdoligranulum variable at rel.

cont.

uncultured bacterium ABLCf22

Faecalibacterium prausnitzii

Uncultured bacterium clone Eldhufec222

Uncultured bacterium clone Eldhufec243

Uncultured bacterium clone Eldhufec268

Uncultured bacterium clone Eldhufec271

Uncultured bacterium clone Eldhufec300

Uncultured bacterium clone Eldhufec302

uncultured bacterium HuCB5

uncultured bacterium LC79

uncultured bacterium M479

uncultured bacterium OLDA-E4

uncultured Gram-positive bacterium NO2-41

uncultured Gram-positive bacterium NB4C12

uncultured Gram-positive bacterium NB5C6

uncultured human gut bacterium JW1D4

Clostridium cluster IX

Dialister

Dialister invisus

Acidaminococcus intestini

Dialister pneumosintes

Uncultured bacterium clone Eldhufec089

Uncultured bacterium clone Eldhufec093

uncultured bacterium MG10

uncultured Gram-positive bacterium NS2B11

Phascolarctobacterium faecium et rel.

Acidaminococcus fermentans

Acidaminococcus intestini

bacterium adhufec395

Uncultured bacterium clone Eldhufec094

Uncultured bacterium clone Eldhufec097

uncultured bacterium D115

uncultured bacterium OLDB-B2

uncultured bacterium OLDB-D6

uncultured Gram-positive bacterium NB4G9

Clostridium cluster XIVa Anaerostipes caccae et rel.

Anaerostipes caccae

Anaerostipes hadrus

bacterium adhufec25

Clostridium indolis

uncultured bacterium HuCA20

uncultured bacterium L902

uncultured bacterium LCRC9

uncultured Gram-positive bacterium NB2G8

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

285

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Firmicutes cont.

Clostridium cluster XIVa cont.

Anaerostipes caccae et rel. cont.

uncultured Gram-positive bacterium NO2-51

Anaerostipes hadrus uncultured human gut bacterium JW1A1

uncultured human gut bacterium JW2C7

Butyrivibrio crossotus et rel.

Butyrivibrio crossotus

Eubacterium eligens

Eubacterium ramulus

uncultured bacterium ABLCf20

uncultured bacterium adhufec43.25

Uncultured bacterium clone Eldhufec138

Uncultured bacterium clone Eldhufec147

Uncultured bacterium clone Eldhufec155

Uncultured bacterium clone Eldhufec167

Uncultured bacterium clone Eldhufec244

uncultured bacterium D373

uncultured bacterium D680

uncultured bacterium D692

uncultured bacterium D726

uncultured bacterium D738

uncultured bacterium HuCB40

uncultured bacterium MG71

uncultured bacterium OLDA-D12

uncultured Gram-positive bacterium NB5D7

uncultured Gram-positive bacterium NS4A6

Clostridium symbiosum et rel.

butyrate-producing bacterium GM2/1

Clostridium bolteae

butyrate-producing bacterium M62/1

Clostridium bolteae

Clostridium clostridiiformes

Clostridium sp. N6

Clostridium symbiosum

uncultured bacterium Adhufec071rbh

uncultured bacterium B147

uncultured bacterium B395

uncultured bacterium B840

Uncultured bacterium clone Eldhufec100

uncultured bacterium E024

uncultured bacterium E108

uncultured bacterium HuCC34

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

286

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Firmicutes cont.

Clostridium cluster XIVa cont.

Clostridium symbiosum et rel. cont.

uncultured bacterium inhufecA-32

Clostridium bolteae

uncultured bacterium K094

uncultured bacterium K375

uncultured bacterium L812

uncultured bacterium LCTI22

uncultured bacterium M431

uncultured bacterium M985

uncultured bacterium MB66

uncultured bacterium MD61

uncultured bacterium MI29

uncultured bacterium N322

uncultured bacterium V213

uncultured Gram-positive bacterium NB2A1

uncultured Gram-positive bacterium NO59

uncultured Gram-positive bacterium NS2G4

uncultured human gut bacterium JW1C1

uncultured human gut bacterium JW1D8

Eubacterium hallii et rel.

Eubacterium hallii

Eubacterium spp. uncultured bacterium HuCA15

uncultured bacterium HuCB26

uncultured bacterium HuCC15

Eubacterium rectale et rel.

butyrate-producing bacterium L1-83

Eubacterium rectale

Butyrivibrio fibrisolvens

Eubacterium rectale

Lachnobacterium sp. wal 14165

uncultured bacterium A22

uncultured bacterium D522

uncultured bacterium HuCA8

uncultured bacterium HuCB37

uncultured bacterium M372

Lachnobacillus bovis et rel.

bacterium A11

Eubacterium rectale

Roseburia faecalis

Roseburia intestinalis

bacterium adhufec68

uncultured bacterium B558

Uncultured bacterium clone Eldhufec128

Uncultured bacterium clone Eldhufec137

Uncultured bacterium clone Eldhufec139

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

287

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Firmicutes cont.

Clostridium cluster XIVa cont.

Lachnobacillus bovis et rel. cont.

Uncultured bacterium clone Eldhufec153

Eubacterium rectale

Roseburia faecalis

Roseburia intestinalis

uncultured bacterium D695

uncultured bacterium HuRC64

uncultured bacterium ME11

Uncultured bacterium UC7-39

Lachnospira pectinoschiza et rel.

butyrate-producing bacterium A2-175

Lachnospira pectinoschiza

butyrate-producing bacterium L2-10

Coprococcus catus

Eubacterium eligens

Lachnospira pectinoschiza

uncultured bacterium ABLCf6

uncultured bacterium Adhufec019rbh

Uncultured bacterium clone Eldhufec108

Uncultured bacterium clone Eldhufec140

uncultured bacterium KO89

uncultured bacterium LZ58

uncultured bacterium NW71

Uncultured bacterium UC7-131

Uncultured bacterium UC7-62

uncultured human gut bacterium JW1G3

Outgrouping Clostridium cluster XIVa

bacterium adhufec236

Eubacterium spp.

bacterium adhufec405

bacterium adhufec52

Clostridium aminovalericum

human intestinal firmicute CO2

uncultured bacterium A12

Uncultured bacterium clone Eldhufec129

Uncultured bacterium clone Eldhufec143

Uncultured bacterium clone Eldhufec145

uncultured bacterium HuCB56

uncultured bacterium LL95

uncultured bacterium MK42

uncultured Gram-positive bacterium NB5G12

uncultured Gram-positive bacterium NO19

uncultured Gram-positive bacterium NS4B5

Roseburia intestinalis et rel. bacterium adhufec225 Roseburia intestinalis

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/

288

Table A6.1 Continued

Higher taxonomic groups Level 2

(genus-like level)

Species/strains that have been

validated with the HITChip

Members of the defined communities

that are expected to be classified

within the genus-like group*

Firmicutes cont.

Clostridium cluster XIVa cont.

Roseburia intestinalis et rel. cont.

butyrate-producing bacterium L1-81

Roseburia intestinalis Roseburia intestinalis

uncultured bacterium HuCA28

Proteobacteria Enterobacter aerogenes et rel.

Citrobacter braakii

Raoultella sp.

Citrobacter freundii

Citrobacter murliniae

Citrobacter werkmanii

Enterobacter aerogenes

Enterobacter cancerogenus

Uncultured bacterium clone Eldhufec065

Uncultured bacterium clone Eldhufec068

uncultured bacterium OLDA-E9

Verrucomicrobia Akkermansia

Akkermansia muciniphila N/A

Uncultured bacterium clone Eldhufec002

uncultured bacterium HuRC51

*Based on the All-Species Living Tree found at http://www.arb-silva.de/projects/living-tree/