METABOLISM OF A DEUTEROPORPHYRIN-NITROIMIDAZOLE ADDUCT BY GASTROINTESTINAL

Simon Alexander Dingsdag

A thesis submitted in fulfillment of the requirements for the degree of

Doctor of Philosophy

Institute of Dental Research

Faculty of Dentistry, The University of Sydney

December, 2016

DECLARATION

Work herein was carried out by Simon Alexander Dingsdag from April 2011 to December 2016 for the degree of Doctor of Philosophy in the Institute of Dental Research, The University of Sydney. None of this work has been submitted previously for the purpose of obtaining another degree and the work is the author’s own.

Signed: ______

Simon Alexander Dingsdag

December, 2016

ii TABLE OF CONTENTS

TABLE OF CONTENTS ...... iii DEDICATION ...... vii AKNOWLEDGEMENTS ...... viii A PUBLICATION, MANUSCRIPT AND PRESENTATIONS ARISING FROM THIS WORK ...... ix THESIS SUMMARY ...... x

CHAPTER I INTRODUCTION ...... 1 1.1 The mammalian gastrointestinal tract ...... 2

1.1.1 and microbial diversity ...... 2

1.1.2 Some nutritional and immunological interactions of microbial inhabitants with their mammalian hosts ...... 7

1.1.3 Dysbiosis of the gut ...... 9

1.1.4 Factors impeding the study of the gut, future directions and therapeutic strategies . 10

1.2 An overview of the Bacteroidetes phylum ...... 12

1.3 The biosynthesis, uptake and use of tetrapyrroles ...... 14

1.3.1 Tetrapyrrole biosynthetic pathways ...... 14

1.3.2 Bacterial tetrapyrrole acquisition systems ...... 19

1.3.3 Respiration and the tricarboxylic acid cycle...... 21

1.4 Microbial and nitroimidazole reduction (activation) ...... 25

1.4.1 Biosynthesis of azomycin the biotic precursor of metronidazole...... 25

1.4.2 The structure and use of nitroimidazoles ...... 26

1.4.3 Primary effects on microbial metabolism of the uptake and reductive activation of metronidazole ...... 28

1.4.4 Distribution and metabolism of metronidazole in human and murine models ...... 33

1.4.5 Metronidazole resistance ...... 39

iii 1.5 The rationale underpinning deuteroporphyrin-nitroimidazole adducts and selection of candidates for characterisation ...... 43

1.6 Chapter I references ...... 47

CHAPTER II AN EVALUATION OF DEUTEROPORPHYRIN-NITROIMIDAZOLE ADDUCTS IN MICROBIAL COMMUNITIES FROM HUMANS AND RATS ...... 71 2.0 PREFACE TO CHAPTER II ...... 72

2.1 INTRODUCTION ...... 73

2.2 METHODS ...... 74

2.2.1 Single Bacterial Strain Growth Curves and Kill Curves ...... 74

2.2.2 Plaque-derived bacterial growth curves ...... 75

2.2.3 DNA Extraction ...... 75

2.2.4 Real time polymerase chain reaction...... 76

2.2.5 Periodontitis model in F344 rats ...... 76

2.3 RESULTS...... 77

2.3.1 Single Species Growth Curves ...... 77

2.3.2 Human subgingival bacterial plaque growth assays ...... 80

2.3.3 Rat Plaque Model ...... 83

2.4 DISCUSSION ...... 83

2.5 Chapter II references...... 86

CHAPTER III THE METABOLISM OF A DEUTEROPORPHYRIN-NITROIMIDAZOLE ADDUCT BY NANAEROBIC BACTEROIDETES ...... 89 3.0 PREFACE TO CHAPTER III ...... 90

3.1 INTRODUCTION ...... 91

3.2 METHODS ...... 94

3.2.1 Medium Preparation and Growth Curves ...... 94

3.2.2 Human ex vivo distal colon culture model growth curves, DNA extraction and Illumina Sequencing ...... 96

iv 3.2.3 Micro-broth Dilution for Antibiotic Susceptibility, DNA Extraction and nim-Gene Amplification ...... 97

3.2.4 Fumarate Reductase and Succinate Dehydrogenase Enzyme Assays ...... 98

3.2.5 Chemical and Biological Reductive Activation of P5, Purification of Products and Liquid Chromatography and High Performance Mass Spectrometry of ...... 99

3.2.6 In Silico Analysis of Fumarate Reductase ...... 100

3.3 RESULTS...... 101

3.3.1 The response of single species and more complex bacterial populations of bacteria to P5 ...... 101

3.3.2 The metabolism of P5 by Bacteroides ...... 109

3.3.3 The Structural Metabolism of P5 by Bacteroides thetaiotaomicron ...... 113

3.3.3 Antibiotic Susceptibility and nim Gene Presence ...... 122

3.4 DISCUSSION ...... 124

3.5 Chapter III References ...... 128

CHAPTER IV THESIS DISCUSSION AND FUTURE DIRECTIONS ...... 134 4.1 THESIS DISCUSSION ...... 135

4.2 FUTURE DIRECTIONS ...... 142

4.2.1 Assessing P5 in a murine model of human C. difficile IBD ...... 142

4.2.2 The use of P5 to glean insight into gut development and ...... 143

4.2.3 Derivatisation of P5 to improve potency for Clostridia or improve growth of Bacteroides ...... 143

4.3 Thesis Discussion and Future Directions References ...... 144

CHAPTER V APPENDIX ...... 146 5.1 RESULTS PERTAINING TO CHAPTER II ...... 147

5.2 RESULTS PERTAINING TO CHAPTER III ...... 149

v APPENDIX CHAPTER VI INITIAL INVESTIGATIONS INTO THE RESPONSE OF RAT GASTROINSTESTINAL BACTERIA TO THE DEUTEROPORPHYRIN-NITROIMIDAZOLE ADDUCT P5 ...... 156 6.0 PREFACE TO APPENDIX CHAPTER VI ...... 157

6.1 INTRODUCTION ...... 158

6.2 METHODS ...... 160

6.2.1 Administration of treatments into rats ...... 160

6.2.2 Sampling of tissue or faeces and digesta for DNA Extraction and ...... 160

6.2.3 Real time PCR estimations of bacterial populations ...... 161

6.2.4 Pyrosequencing ...... 161

6.2.5 Viable bacterial counts ...... 162

6.3 RESULTS...... 162

6.3.1 The effect of 2mM P5 on rat faecal microbes...... 162

6.3.2 Optimising P5 and metronidazole treatment concentrations ...... 169

6.4 DISCUSSION ...... 172

6.5 Appendix Chapter VI References ...... 175

vi DEDICATION

This thesis is dedicated to my parents, Don Dingsdag and Jill Dunwoodie, without whom this endeavour would not be possible.

vii AKNOWLEDGEMENTS

I am fortunate to have been provided with resources and a wide berth to explore, to consider thermodynamics, the inefficiencies of antibiotics, periodicity, duplication and to be humbled by biological complexity.

Many people have helped in the completion of this thesis. I appreciate encouragement and necessary criticism from my mentor, Neil Hunter; provocative discussions with my colleague, Ramin Farahani and guidance synthesising deuteroporphyrin adducts from Ben Yap and Max Crossley. Also, support from my girlfriend, Carolina Firacative; the care, coffee and/or beer as appropriate from family and other friends, have also been helpful. I thank these people and many others I have not mentioned.

I am grateful for Michael Malamy’s generous gift of mutant Bacteroides fragilis. The Australian Dental Research Foundation and Colgate-Palmolive have provided funding, for which I am also grateful.

viii A PUBLICATION, MANUSCRIPT AND PRESENTATIONS ARISING FROM THIS WORK

Parts of the work presented herein are published in peer reviewed journals

S. A. Dingsdag*, B. C. M. Yap*, N. Hunter and M. J. Crossley (2014)

Amino acid-linked porphyrin-nitroimidazole antibiotics targeting

Organic and Biomolecular Chemistry, 13 (98–109). (DOI: 10.1039/C4OB01841A)

(*equal first authors)

Parts of the work presented herein have been prepared for publication

S. A. Dingsdag, N. Hunter. A deuteroporphyrin-nitroimidazole antibiotic promotes growth of Bacteroides and kills Clostridia

Parts of the work were presented at scientific meetings

2011 International Association for Dental Research, (Melbourne, Australia)

2013 Australian Dental Council meeting (Sydney, Australia)

2016 Tetrapyrroles, Chemistry & of, Gordon Research Conference (Salve Regina University, Rhode Island, USA)

ix THESIS SUMMARY

The human gastrointestinal tract, extending from mouth to rectum, hosts complex microbial communities. Dysbiosis is a term used to indicate a shift in the balance of these residents towards a pathogenic state, typically associated with a breakdown of the relationship with the and tissue degradation (disease onset). In the mouth, chronic inflammation may lead to periodontitis and destruction of the periodontium, a complex process in which the Gram-negative nanaerobe Porphyromonas gingivalis, has been implicated (Hajishengallis et al., 2011). Dysbiosis of the distal gut largely manifests as mucosal inflammation and loss of epithelial integrity, including a suite of inflammatory bowel diseases that may be associated with members of the genera. Broad-spectrum antibiotics, frequently chosen as a treatment, are a problematic solution to mediate restoration, with extensive ‘off target’ action. Also, colitis associated with overgrowth of endogenous Clostridium difficile is frequently linked to previous broad-spectrum antibiotic therapy. The unmet challenge in treating diseases associated with dysbiosis is the guided reconstitution of a so-called ‘normal microbiota’.

Previously, in an attempt to selectively control P. gingivalis, a porphyrin auxotroph sensitive to nitroimidazole antibiotics, deuteroporphyrin-linked nitroimidazole adducts were synthesised (Fig. 1). These adducts were designed as a ‘Trojan-horse’ delivery system for nitroimidazoles into P. gingivalis, through recognition of the porphyrin macrocycle. In this thesis, the activity in vitro of amide- and lysine-linked deuteroporphyrin- nitroimidazole adducts (termed ‘P5’ and ‘P8’

Figure 1: Deuteroporphyrin IX (DPIX, A), and amide-linked adduct, termed ‘P5’ (B). Carbons 13 and 17 of the DPIX macrocycle are labelled.

x respectively) was explored. Growth curve and viable count assays were used to assess the response of nanaerobic bacteria sensitive to metronidazole (a nitroimidazole drug), in enriched brain heart infusion (enriched BHI) , chosen to support growth of the fastidious nanaerobe . Like metronidazole, both adducts were bactericidal for P. gingivalis. Uptake and reductive activation of P5/P8 was rapid, with the majority (99% of 5 × 105) of inoculated bacteria being killed within 30 minutes. Since none of the other nanaerobic Bacteroidetes tested (Bacteroides fragilis, Prevotella melaninogenica, Tannerella forsythia) or Fusobacterium nucleatum were killed in enriched BHI medium, preliminary evidence indicated P5 and P8 were specific for P. gingivalis, killing this bacterium only.

In order to examine the specificity of adducts for P. gingivalis within complex ecosystems two models were explored. In the first human dental plaque was inoculated into growth media and were cultured under anaerobic conditions for short periods of time. Overgrowth by (presumably) facultative bacteria obscured the capacity to discriminate the fate of slower growing bacteria. Next a rat model for alveolar bone loss, due to introduced P. gingivalis ATCC 33277, a known infective strain, was examined. Kanamycin, an aminoglycoside which does not affect P. gingivalis, but does target dominant clades of bacteria inhabiting the rat oral cavity, including Gram- positive facultatives, Rothia () and (), was added to drinking water for two weeks. Despite daily gavages of ≈2 × 1010 viable cfu of P. gingivalis 33277, colonisation rate, assessed by real time PCR primers specific for P. gingivalis, was poor. Since none of the rats exhibited gross anatomical changes, introduction of ‘foreign’ bacteria such as P. gingivalis into the rat oral cavity is notoriously difficult (Manrique et al., 2013) and as the anatomy and bacterial composition differs from human counterparts, pursuit of this model was discontinued.

It was decided instead, that the gastrointestinal tract, dominated by nitroimidazole-sensitive nanaerobes, including Clostridia (of the Firmicutes phylum) and Bacteroidetes (including porphyromonads), was a more suitable model to test the specificity of porphyrin-nitroimidazole adducts. As a reference point, growth curves in BHI were prepared for strains representing major bacterial clades of human origin. Surprisingly, the growth of all representative laboratory strains of Bacteroides spp. (five strains), originating from human gut was enhanced by the amide-linked adduct ‘P5’. Further, when titrated below minimal inhibitory concentrations, growth of Prevotella spp. (two strains) was stimulated. In contrast, P5 retained nitroimidazole activity for nanaerobes within the Firmicutes clade, killing all examined Clostridiaceae (five strains) with a three-fold average increase in potency (1.6 µM) over metronidazole (5.6 µM). None of the examined facultative bacteria, including representatives from the Proteobacteria (two strains) or Actinobacteria (three strains) were killed by P5, nor was the metronidazole-sensitive F. nucleatum (a member of the Fusobacteria

xi phylum). These results were confirmed in an ex-vivo human faecal model, in which viable Bacteroides populations grown in P5 were approximately double the population of DMSO control at 10 hours. These results indicate not only growth support of Bacteroides, but taken with Illumina sequencing, confirm killing capacity for nanaerobic Clostridia in a complex bacterial human faecal population and secondly, that Bacteroides spp. were advantaged by depletion of Clostridia.

Based on these observations, and a re-examination of the literature, it is apparent that many Bacteroidetes do not encode the full porphyrin biosynthetic pathway. These Bacteroidetes incorporate porphyrin into a b-type cytochrome, encoded by the three-component fumarate reductase operon (frdC, frdA, frdB), reducing fumarate to succinate, thereby forming an essential component of the reductive arm of the tricarboxylic acid (rTCA) cycle. In Bacteroidetes that encode both oxidative and reductive arms of the TCA cycle (Bacteroides), porphyrin supplementation is not essential, but it permits optimal growth, whereas in genera that encode only a portion of the rTCA (Prevotella and Porphyromonads), a porphyrin source is essential to synthesise succinate and additional ATP, for redox maintenance (NAD+/NADH ratio) and for synthesis of amino acids including glutamic acid, arginine, proline and aspartic acid. Growth of two mutant Bacteroides fragilis defective in components of the fumarate reductase operon (frdC or frdB) was suppressed following challenge with P5 indicating the nitroimidazole moiety was activated, albeit with a lower potency than for metronidazole. Importantly, no stimulation of growth was detected for either mutant, indicating P5 or metabolites of P5 were not acting as a source of porphyrin in the absence of a functional fumarate reductase system. In -free biochemical assays utilising B. thetaiotaomicron, fumarate reduction was stimulated by a metabolite/s of P5 albeit around half the rate of biotic (haem) and abiotic (DPIX) controls. The data were interpreted as indicating that metabolised product/s of P5 were incorporated into a functional b-cytochrome. The B. thetaiotaomicron- metabolised and sodium-dithionite (reduced) products of P5 were studied by high resolution mass spectrometry. Structures consistent with the several ions shared between these treatments and candidates for incorporation into the b-cytochrome are proposed.

When P5 was reduced in sodium dithionite to inactivate the biologically-active nitro group of the adduct, the rate growth and final biomass of B. thetaiotaomicron was higher than for the control P5 treatment. Reduced P5 lost killing capacity for C. difficile, confirming that adduct bactericidal action was tethered to nitroimidazole activity (P5 retains prodrug-like properties). Data also indicated that P5 was activated in Bacteroides although this was insufficient to cause death. From these findings, it is apparent that nanaerobic Bacteroidetes have efficient porphyrin uptake machinery, including previously unappreciated flexibility for the types of porphyrin that can be taken up and incorporated into b-cytochromes. Both processes appear to be independent of the metal

xii ligand in examined Bacteroidetes. More broadly, growth curve and a faecal growth curve model indicate other clades, including representatives within the Actinobacteria and Firmicutes also have a previously unappreciated flexibility for the types of porphyrin that can be taken up.

Having characterised some of the elements of P5 metabolism in nanaerobic Bacteroidetes, the response of complex bacterial communities in the rat gut was explored. Despite concentrated five-day administration of metronidazole (25.7 mg kg-1 per day), little change was observed in the faecal bacteria of rats, based on real time PCR estimates of total bacterial load independently and with respect to the Bacteroidetes population. Next generation sequencing confirmed that at finer taxonomic resolution, little change occurred in faecal bacteria populations. The data are compatible with a report indicating reductions in viable Bacteroides in rats exposed to metronidazole, was restricted to mucosa-associated microbiota of the ileum and colon and not in the digesta (Mikelsaar and Siigur, 1992). Dysbiosis of the distal gut microbiota is not well defined, however prolonged mucosal inflammation and loss of barrier function, or inflammatory bowel disease (IBD), is at least partly mediated by bacteria, such as C. difficile from the Clostridia clade. Since growth of B. thetaiotaomicron was stimulated by therapeutic concentrations (20 nM - 20 µM) of P5, adducts may have a practical use in treating clostridial-mediated IBD, where it is envisaged that the guided restoration of Bacteroides, to the detriment of Clostridia, could be a new treatment modality.

Thesis aims

I) Characterise the metabolism of deuteroporphyrin-nitroimidazole adducts

II) Define the consequences of administering deuteroporphyrin-nitroimidazole adducts into complex bacterial communities in vitro and in vivo

xiii Thesis Summary References

HAJISHENGALLIS, G., LIANG, S., PAYNE, M. A., HASHIM, A., JOTWANI, R., ESKAN, M. A., MCINTOSH, M. L., ALSAM, A., KIRKWOOD, K. L., LAMBRIS, J. D., DARVEAU, R. P. & CURTIS, M. A. 2011. Low-Abundance Biofilm Species Orchestrates Inflammatory Periodontal Disease through the Commensal Microbiota and Complement. Cell Host & Microbe, 10, 497-506. MANRIQUE, P., FREIRE, M. O., CHEN, C., ZADEH, H. H., YOUNG, M. & SUCI, P. 2013. Perturbation of the indigenous rat oral microbiome by ciprofloxacin dosing. Molecular Oral , 28, 404-414. MIKELSAAR, M. & SIIGUR, U. 1992. Metronidazole and the intestinal microecology of rats. Microbial in health and disease, 5, 139-146.

xiv

xv Chapter I Introduction

CHAPTER I

INTRODUCTION

1 Chapter I Introduction

1.1 The mammalian gastrointestinal tract

1.1.1 Anatomy and microbial diversity

The human gastrointestinal tract, extending from the mouth to the anus, hosts an enormous diversity of microbial inhabitants, including Archaea (largely methanogens), Fungi, Bacteria and their viruses. In the mouth, the phyla Firmicutes, Proteobacteria, Bacteroidetes, Actinobacteria and Fusobacteria tend to dominate, with common genera including Streptococcus, Fusobacterium, Actinomyces, Capnocytophaga, Haemophilus, , Corynebacterium, Prevotella, Veillonella, and Rothia (Aas et al., 2005, Bik et al., 2010). Fungi are also present, as are Archaea. In the oral cavity, the imbalance of the microbial residents towards a pathogenic state, dysbiosis, may be accompanied by inflammation and degradation of the tissues supporting the teeth, termed periodontitis. The Gram-negative nanaerobe Porphyromonas gingivalis (see Fig. 1.1), is implicated in periodontitis, in concert with other oral bacteria, a process suggested to involve activation of complement components C3a and C5a (Hajishengallis et al., 2011). Other virulence characteristics of P. gingivalis are discussed in Section 1.2.

The mature gut epithelium consists of a series of undulating protrusions, termed villi, at the base of which are crypts of Lieberkühn with populations of epithelial stem cells that populate the midpoint of crypts with differentiated cell lineages, including columnar absorptive colonocytes,

Figure 1.1: Transmission electron micrograph of P. gingivalis W50 grown under ‘haem limited’ (A, 0.46 µM) and ‘haem excess’ (B, 7.7 µM) growth medium from McKee et al., 1986. Note that extracellular vesicles present in 0.46 µM haem, were effectively absent with 7.7 µM haem and were replaced with extensive fimbriae. Scale bar = 0.5 µM.

2 Chapter I Introduction enteroendocrine, Paneth, and mucin-producing goblet cells. Mature goblet cells secrete mucins, forming a barrier. Mucin is sparse or non-existent in the small intestine and increases in thickness with descent into the colon (Johansson et al., 2011). The underlying ≈50-100 µm mucus layer is thick, being primarily composed of mucin2 (Muc2) polymers (of around 100 MDa, consisting largely of sulfated glycans) (see Fig. 1.2). In healthy individuals limited strains of bacteria inhabit this thick primarly tightly linked layer (Davis et al., 1972, Swidsinski et al., 2005). In contrast, many bacteria inhabit a second more loosely arranged outer mucus layer which is up to 800 μm thick in the colon of rats (Johansson et al., 2008, Atuma et al., 2001). In the gut, the number and types of bacteria varies in a given site within individuals and between individuals. Constant turnover and shedding of epithelial cells, overlying mucus and peristaltic movements, all contribute to variation in microbial populations. Variation is also driven by diet (including carbon, nitrogen and vitamin flux), gastric, pancreatic and biliary from the bile duct (at the proximal end of the small intestine). Lastly, the appendix, an addendum to the caecum (Fig. 1.3) which contains mucus and also supports bacteria, has been proposed to re-seed the caecum and colon after catastrophic events, including diarrhoea (Bollinger et al., 2007). How these factors govern microbial stability is not well

Figure 1.2: A Section of the mouse distal colon showing mucosa and overlying mucus layer from Johansson et al., 2008. Goblet cell mucus was stained with anti-MUC2C3 antiserum (green). The inner mucus layer (marked a and s) and outer mucus layer (marked o) are linked by muc2 extensions from the epithelium (arrowed). DAPI was used to stain nuclei (blue) of the ‘invaginating’ crypts of Lieberkühn. The base of the crypts of Lieberkühn are not shown. The scale bar is 50 µm.

3 Chapter I Introduction characterised, partly because sampling is often performed from faeces, which is not ‘spatially defined’, so is a crude metric. Broad trends, with emphasis on development and spatial distinctions in microbial assemblage are summarised below.

In neonates, bacteria colonising the (sterile) gastrointestinal tract are thought to originate predominantly from the mother’s vagina. Organisms include Bifidobacterium and Lactobacilli, with lower colonisation of Proteobacteria (including Escherichia coli), Clostridia, and Bacteroidetes (Rotimi and Duerden, 1981, Koenig et al., 2011). In developing neonates, the acidity of the stomach is presumably low enough to allow passage of these microbes. Upon of H+ and Cl- ions (hydrochloric acid) by parietal cells, the pH in the gastric compartment is maintained at 1-2. The low pH, along with secretion of the proenzyme pepsinogen (which is converted to pepsin A) largely inhibits the transfer of viable microbes through the stomach. Facultative bacteria of the Firmicutes phylum, predominantly the Bacilli class (Lactobacilli and Streptococci) which are resistant to bile salts, maintain residence in the small intestine. Bacteria tend to increase in number with decent into the large intestine. These bacteria are widely considered anaerobic, although they are capable of respiring in nanomolar concentrations of oxygen, so are more accurately called ‘nanaerobes’, (Baughn and Malamy, 2004, Morris and Schmidt, 2013). These nanaerobes include Clostridia (a class of mostly spore forming bacteria, within the Firmicutes phylum) and the non-spore forming Bacteroidetes phylum, consisting largely of Bacteroides, Prevotella, Porphyromonas and Rikenella

(Eckburg et al., 2005) (Fig. 1.3). Depending on the region sampled, nanaerobes of the Bacteroidetes and Firmicutes phyla outnumber by up to 2 orders of magnitude facultative and aerobic bacteria including Actinobacteria, Proteobacteria, Fusobacteria (a nanaerobic phylum) and divisions of prokaryotes that are uncultured, although have sufficiently different 16S rDNA to be considered novel (candidate phyla), such as TM7 (Eckburg et al., 2005). Up to 1012 microbial cells per gram may be found in the colon, the vast majority of which reside in the digesta of the lumen and not on the mucosa. 16S rDNA surveys in mice, indicate Lachnospiraceae and Ruminococcaceae (Firmicutes) (Nava and Stappenbeck, 2011, Nava et al., 2011) dominate the colonic mucosa, while Bacteroidaceae, Enterococcaceae and Lactobacillaceae are enriched in the digesta (Pédron et al., 2012). Despite differences in diet, anatomy and , at a coarse phylum resolution, Bacteroidetes and Firmicutes phyla dominate the faeces of mammalia (Ley et al., 2008).

Several mechanisms have been proposed to explain how the host spatially confines microbial inhabitants and how populations of bacteria colonise the gut and maintain long-term residence. Some regulatory influences involve the host and others are intra-bacterial. As noted above, few bacterial species are closely associated with the intestinal epithelium. These include

4 Chapter I Introduction

Figure 1.3: Bacterial and proposed chemical gradients in the human gut from Donaldson et al.,2015. The duodenum, jejunum and ileum, making up the small intestine, are not labelled. Note that the appendix may serve as a reservoir, seeding the caecum, for example after diarrhoea (Bollinger et al., 2007).

Acinetobacter spp. and Proteobacteria associated with intestinal crypts of mice (Pédron et al., 2012), while in chickens (Yamauchi and Snel, 2000) and a variety of other non-mammalian vertebrates(Klaasen et al., 1993), Segmented Filamentous Bacteria (SFB, a spore forming clade, Candidatus Savagella of the Clostridia) are found. The secretion of an antibacterial lectin, RegIIIγ, has been proposed to limit microbial colonisation within 50 µm of the epithelium in the small intestine (Vaishnava et al., 2011). The secretion of other antimicrobials by Paneth cells, which are mostly confined to the small intestine, has also been proposed to limit colonisation of this region [see (Bevins and Salzman, 2011) for a detailed review]. The expression of polysaccharide A on (the

5 Chapter I Introduction surface of) B. fragilis, results in ‘immune tolerance’ allowing colonisation on the mucosa by down regulating TLR2-mediated recognition by T helper cells (CD4+ T cells) (Round et al., 2011). Modifications to the bacterial surface, such as the addition of L-fucose to surface glycoproteins and capsular polysaccharides by Bacteroides spp., mimicking mammalian intestinal epithelial cell glycans, which also contain L-fucose, may allow colonisation of the mucosal surface (Coyne et al., 2005).

Human faecal bacteria are generally amenable to culturing (Goodman et al., 2011) and remarkably, ex vivo culture models maintain the numerical ratio of bacterial populations found in the murine caecal lumen (Freter et al., 1983b, Freter et al., 1983a). The maintenance of bacterial populations in the absence of cues from the host, suggests these populations are partly internally regulated, for example by bacteria occupying specific nutritional niches or by the secretion of inhibitory molecules. For example, in a single-species gnotobiotic mouse model, carbohydrate utilisation systems uniquely encoded by B. fragilis and B. vulgatus enabled the catabolism of specific carbohydrates allowing colonisation and dominance of these clades in the gut (Lee et al., 2013).

Also, hydrogen sulfide (H2S) has been shown to supresses the growth of E. coli in an ex vivo culture model (Freter et al., 1983a). The excretion of H2S is probably the result of dissimilatory sulfur metabolism (anaerobic respiration). A possible mechanism is that dominant clades, rather than E. coli, take up sulfur containing proteins/peptides and liberate oxidised sulfur containing

2- 2- compounds (SO3 , SO4 ) and reduce them. Members of the Clostridium genus generally lack the capacity for dissimilatory sulfate reduction (Collins et al., 1994) so it is unlikely to be members of this clade that liberate H2S. These results could indicate that more dominant members, including bacteria within the Bacteroides and Clostridia have more efficient than E. coli. Members of these clades could also occupy specific nutritional niches not available to E. coli or generate inhibitors other than H2S, suppressing growth of E. coli. Consistent with these observations a decrease in members of the Clostridia is associated with an increase in E. coli on the Illeal mucosa of human subjects during Crohn’s disease (Baumgart et al., 2007). Overgrowth of Proteobacteria during IBD indicates that limitations of total numbers of these bacteria are not based on the lack of nutrients. Whilst not the focus of these investigations, it is intriguing that E. coli (and other Proteobacteria) or other low abundance clades maintain residence in the human gut, implying these clades perform a function/s such as the catabolism of ‘waste’ metabolites from more abundant clades or provide essential nutrients to bacteria or the host.

6 Chapter I Introduction

1.1.2 Some nutritional and immunological interactions of microbial inhabitants with their mammalian hosts

Given that mammals have co-evolved for millennia with microbes in their gastrointestinal tracts, it is not surprising there is an extensive network of inter- and intra-domain interactions, some of which are based on metabolism such as vitamin synthesis (folate, B12) and others based on immune stimulation and repression. While only beginning to be characterised, the gut microbiota collectively encode several functions absent in human hosts (Hooper and Gordon, 2001). Mass spectrometry of conventional and germ free mice detected up to a five-fold increase in the number of metabolites extracted from blood of conventional mice, including the absence of various flavones and phenyl derivatives in germ free mice (Wikoff et al., 2009). Also, in germ free rats the caecum is dramatically enlarged, in part due to undegraded mucus (Gustafsson et al., 1969) which is reversible by introduction of single bacterial strains including micros (Carlstedt-Duke et al., 1986).

Of the 97 glycoside hydrolases in the human genome, 8-17 are estimated to be involved in digestion (El Kaoutari et al., 2013). Human digestive glycan cleaving enzymes are limited to starch, sucrose, lactose, trehalose and possibly some chitins (Cantarel et al., 2012). Glycans not digested by humans are typically plant derived. These glycans are cleaved by intestinal microbiota and fermented into short chain fatty acids (SCFA), largely acetate, propionate and butyrate (in the range of 20-60 mmol kg-1, Fig 1.4) before excretion and absorption, particularly of butyrate in the large intestine (Macfarlane and Macfarlane, 2003). In germfree mice, reduced expression of enzymes supporting the TCA cycle in colonocytes leads to reductions in total NADH/NAD+ and ATP, resulting in autophagy which is ameliorated by butyrate (Donohoe et al., 2011). Butyrate production is considered to be limited to clades within the Clostridia class, including Eubacterium spp., Roseburia spp., Clostridium spp. and Faecalibacterium prausnitzii (Louis and Flint, 2009). However, Porphyromonas spp. and Prevotella spp. also produce butyrate [(Sawyer et al., 1962, McKee et al., 1986), see Section 1.2]. Also, emerging evidence indicates lactate is preferentially oxidised by neurons, even in the presence of glucose (Wyss et al., 2011). It is possible lactate is produced in the small intestine by bacterial clades, including the Lactobacilli, before absorption, transport in the bloodstream and oxidation by neurons.

The microbial inhabitants (Lathrop et al., 2011), B. fragilis (Round and Mazmanian, 2010) and Clostridium spp. (Atarashi et al., 2011) stimulate the development of T-regulatory cells (Tregs). More recently, bacterial fermentation products have been shown to stimulate the development of

Tregs (Arpaia et al., 2013, Smith et al., 2013). In Tregs isolated from mice, butyrate enhances surface

7 Chapter I Introduction

Figure 1.4: Simplified schema of microbial carbohydrate catabolism, showing fermentation (glycolytic/pentose phosphate pathways), portions of the TCA cycle (boxed) and excreted end products (from Macfarlane and Macfarlane, 2003). PEP=phosphoenolpyruvate. Note the redox cycle of NAD+/NADH, largely driven by glycolysis and TCA cycle, which is relevant to the activation of metronidazole.

expression of Foxp3 protein through the intronic enhancer conserved non-coding sequence 1 (CNS1) (Zheng et al., 2010, Arpaia et al., 2013) which in turn is proposed to dampen the immune response.

Propionate dampens the capacity of dendritic cells to drive extrathymic Treg development, probably through inhibition of histone deacetylase (EC 3.5.1.98) (Arpaia et al., 2013). Since Tregs limit the immune response in the intestine (Josefowicz et al., 2012), microbes in the gut indirectly modulate the immune response through butyrate and propionate production. SFB, noted above to inhabit

+ ‘privileged’ sites adjacent to the mucosa, also induce CD4 T-helper (Th) cells in a mono-associated mouse model (Ivanov et al., 2009). Induction of these Th cells results in cell lineages in the lamina

8 Chapter I Introduction propria producing interleukins 11 and 17, increasing expression of genes associated with inflammation (Ivanov et al., 2009). Lastly, in mice, B. fragilis produces sphingolipids that inhibit invariant natural killer T cells (iNKTC), limiting proliferation of these cells during infant development and reducing adult colonic iNKTC populations. This could limit autoimmune disorders mediated by overactive iNKTC responses (An et al., 2014).

1.1.3 Dysbiosis of the gut

As noted above, the healthy epithelium of the large intestine is lined by colonocytes, over which lies a sheath of tightly packed mucus generally harbouring low numbers of microbes. Degradation of the underlying mucus and mucosa (in the large intestine) of the gastrointestinal tract as a result of inflammation and microbially produced effectors, is a common feature of ulcerative colitis (primarily of the large intestine) and Crohn’s disease (occurs throughout the gastrointestinal tract). Notably, (typically E coli) are typically enriched during IBD (Darfeuille- Michaud et al., 1998, Frank et al., 2007, Baumgart et al., 2007), often at the expense of Bacteroidetes (Chang et al., 2008). Proponents of the so-called ‘like with like’ hypothesis suggest that a ‘bacterial bloom’, or overgrowth of a particular clade, is most likely to occur when genera are related (Stecher et al., 2010). For example, mice harbouring large populations of Proteobacteria are more susceptible to with enterica and Campylobacter jejuni (Stecher et al., 2010, Haag et al., 2012). However, Proteobacteria such as E. coli have a rapid growth capacity similar to genera within the Clostridia (see Chapter III). Therefore, it is likely that the Proteobacteria phylum is normally suppressed in the gut and the capacity for a new member of the Proteobacteria to colonise or expand is due to the failure of this mechanism/s. Alternatively, degradation of the

2- intestinal epithelium may provide molecules (tetrathionate, S4O6 ) permitting anaerobic respiration unique to members of the Proteobacteria (Winter et al., 2010).

Transplant from a ‘healthy’ human donor reduces active inflammation in recipients, suggesting at least in some instances, inflammation is mediated by resident microbes (Eiseman et al., 1958, Bennet and Brinkman, 1989). It is unclear to what extent the replacement of a dysbiotic with a healthy transplant remediates an inappropriate immune response by the host or the function of the dysbiotic community. Also, the inoculating-microbe/s required to ameliorate inflammation remains poorly defined. However, a diverse mixture of six bacteria ( warneri, hirae, reuteri, with three novel isolates: Anaerostipes sp., Bacteroides sp. and Enterorhabdus sp.) cultured from mice, displaced an epidemic strain of C. difficile (027/BI) in a murine model of persistent disease and also increased bacterial diversity

9 Chapter I Introduction

(Lawley et al., 2012). A recent model has also implicated a murine Lachnospiraceae isolate in reduced C. difficile colonisation and associated colonic degradation (Reeves et al., 2012).

Some explanations of how the mucus layer and underlying cells are degraded involve modulation of host immune response by changing cell surface ligands and the excretion of microbial effectors. For example, metalloproteinase 60 (M60-like) peptidases which degrade mucins, are present in many bacteria, including representatives within the Firmicutes, Bacteroidetes and Proteobacteria (Nakjang et al., 2012). The metalloprotease YghJ is secreted by enterotoxigenic Escherichia coli, facilitating delivery of heat labile and heat-stable toxin to receptors of intestinal enterocytes, disrupting fluid (Luo et al., 2014). Another example is the overgrowth of C. difficile and release of the AB toxin associated with pseudomembranous colitis and toxic megacolon. While C. difficile and M60-like proteinase generating bacteria are associated with inflammatory bowel diseases, other bacteria also contribute to inflammation, by production of effectors that cross mucus barriers, contacting host immune cells. In a mouse model, double mutants of T cell serine/threonine kinase TGFβRII receptor and interleukin-10 receptor 2 (IL-10R2) developed microbially driven spontaneous colitis, indicating a partial role for these receptors in reducing intestinal inflammation (Kang et al., 2008). Re-inoculation of Bacteroides spp. (Bacteroides vulgatus, Bacteroides TP5 and B. thetaiotaomicron), enriched from mice by selective antibiotics into this model also drove inflammation, similar to the full microbial assemblage (Bloom et al., 2011). More recently, mutant B. thetaiotaomicron, deficient in sulfatase maturation enzyme (anSME) and therefore sulfatase production, colonised although did not induce inflammation in the double knockout mouse model (Hickey et al., 2015). The investigators proposed sulfate-containing outer membrane vesicles (OMV) mediate degradation of host mucins and also activate underlying immune cells, including macrophages (Hickey et al., 2015). These results are a reminder that the function of B. thetaiotaomicron, normally considered a symbiotic organism (Goodman et al., 2009), is dependent on context.

1.1.4 Factors impeding the study of the gut, future directions and therapeutic strategies

Explaining why extreme microbial diversity (and function) is supported, characterising the interplay between tolerance, nutrient exchange and immune function, periodicity of microbial populations and how a healthy intestinal microbiota shifts towards dysbiosis continues. Several questions remain, for example, does the spatial segregation of certain microbes indicate a

10 Chapter I Introduction mutualism, parasitism or ‘tolerance’ with the host? Also, why do low abundance clades maintain long-term residence? Some of the technical hurdles in further understanding of the human gastrointestinal tract include the complexity of the harboured microbes, with inter-individual differences not only in numbers at phylum resolution, but also at strain and sub-strain resolution. Further, molecular profiling techniques based on ribosomal signatures such as those in 16S rDNA, provide a taxonomic framework to assess microbial populations, which may not concur with function per se. For example, the Clostridia are a phylogenetically (Collins et al., 1994) and metabolically (Seedorf et al., 2008) diverse clade, yet they are grouped together as a class. Indeed, taxonomic frameworks ignore functional redundancies. The abundance of a particular clade does not necessarily have a relationship to magnitude of interaction with the host or other microbes. Also, the contribution of other domains and their viruses is another appreciable layer of complexity. Lastly, distinguishing between causality and epiphenomena is difficult amongst such complexity. Dealing with such complexity requires a careful choice of model systems.

Regardless of these impedances, progress has been made. Germ-free animal models in which the absence of detectable harboured life results in poorly developed organs and a poor immune response to stimuli, suggest that gut microbes are so ingratiated in the biology of their host that they are now essential for proper development. Introduction of particular microbes into germ- free animals, termed gnotobionts (with defined inhabitants), may be used to infer interactions between host and introduced microbes, such as the augmentation of nutrient uptake or other molecules that are traded. Such models have been used to define trade between species of bacteria (Mahowald et al., 2009) or have been used to understand the capacity of a particular bacterium to modulate an immune response, as discussed above. However, making inferences from such models may be difficult, since both introduced microbes and host may differ from ‘normal’ animals in which the full microbiota is present (Freter et al., 1983a).

These vagaries make it difficult to modulate the microbiota in a rational way. For example, modulation of resident human gut microbes or their function with probiotic (live) strains of bacteria may be hampered by poor uptake of introduced strains. Faecal transplants, in which stools from healthy donors are used to replace the microbiota in unhealthy recipients have had some success, however these remedies remain crude, without an understanding of the specific microbes that remediate inflammation, nor how or where introduced microbes colonise. Ironically, without an understanding of the specific bacteria involved in dysbiosis, phage therapy, which relies of the recognition of specific receptors to target prokaryotes, may be too specific. However, the discovery that a murine RNA norovirus supports intestinal homeostasis in the absence of bacterial inhabitants, provides an intriguing therapeutic strategy (Kernbauer et al., 2014). Lastly, antibiotics are also

11 Chapter I Introduction problematic, often targeting broad groups of bacteria, impairing or killing non-target microbes, which may result in overgrowth of undesirable bacteria.

1.2 An overview of the Bacteroidetes phylum

The Bacteroidetes are a clade (phylum) of Gram-negative nanaerobes, including Bacteroides, Prevotella, Porphyromonas and Rikenella which are genera that inhabit the human gut and the mouth (only Prevotella and Porphyromonas). Aerobic genera of the Bacteroidetes include Cytophaga, Flavobacteria and Salinibacter, which dwell in soil, freshwater and oceans. Several changes to classification of these bacteria have been made over time. Most recently the human associated strains were split based on capacity to ferment saccharides. Prevotella may be highly fermentative (such as P. melaninogenica), intermediate (such as P. intermedia) or unlike the Bacteroides have little or no fermentative capacity, including the porphyromonad P. gingivalis (Shah and Williams, 1987).

Early growth curve studies established that Bacteroides tend to have relatively simple nutritional requirements including ammonia (NH3), which rather than amino acids supplies nitrogen (Varel and Bryant, 1974), a fermentable carbohydrate (such as glucose), sodium (~1mM) (Caldwell and Arcand, 1974), cysteine and sulfide (S2-) which serve as sources of sulfur (Varel and Bryant, 1974,

2- 2- Thauer, 1988). Sulfite (SO3 ) and sulfate (SO4 ) are not metabolised (Varel and Bryant, 1974). In the absence of methionine, a source of B12 is required for some, but not all Bacteroides (Varel and

Bryant, 1974). Vitamin B12 acts as a cofactor for methyl-malonyl-CoA mutase, catalysing the isomerisation of methylmalonyl-CoA to succinyl-CoA. Also methionine synthase catalyses methyl group addition to homocysteine, forming methionine. Bacteroides spp. have the most diverse and largest number of genes encoding polysaccharide cleavage enzymes, largely homologous to the starch utilisation system of B. thetaiotaomicron (Salyers et al., 1977a, Salyers et al., 1977b, Martens et al., 2008). The monomers of a wide variety of plant polysaccharides that humans are unable to cleave (Martens et al., 2008) are fermented and funnelled into the TCA cycle, the products of which are fumarate and malate, or succinate (in the presence of porphyrin) and propionate (if B12 is present) (Macy et al., 1975, Macy et al., 1978). Bacteroides also produce acetate and D-lactate (Macy et al., 1978) and tend to be weakly proteolytic (Moore and Holdeman, 1972). Interestingly, the Bacteroides encode oxidative and reductive portions of the TCA cycle, although use of the reductive portion requires a source of porphyrin (see Section 1.3.2 for an overview). Aconitase, an enzyme of the oxidative portion of the TCA cycle of B. fragilis (and other Bacteroides) has high homology to

12 Chapter I Introduction human mitochondrial aconitase, suggesting that enzymes of the TCA cycle may have a common ancestor (Baughn and Malamy, 2002). Also, interestingly, the ribosomal binding site and transcription machinery differ from that used by the canonical Shine-Dalgarno sequence (5’- AGAAAGGAG) (Wegmann et al., 2013).

As mentioned, Prevotella and Porphyromonas are considered to be intermediate- or non- fermentative. In intermediate-fermentative strains, such as P. melaninogenica, radiolabelled aspartate is oxidatively deaminated, forming oxaloacetate, which is then reduced via fumarate reductase, forming succinate (Wong et al., 1977). Aspartate may also be deaminated by P. melaninogenica, forming fumarate (Wong et al., 1977), a reaction suggested to be shared by P. gingivalis (Dashper et al., 2009). Notably, in contrast to current views that butyrate production is limited to Firmicutes (Louis and Flint, 2009), assacharolytic members of Prevotella and Porphyromonas genera produce butyrate (Sawyer et al., 1962, McKee et al., 1986), probably via the crotonyl-CoA pathway, a metabolism requiring exogenous porphyrin (Dashper et al., 2009). Fermentative strains produce acetate and lactate, whilst non-fermentative strains catabolise peptides, largely forming butyrate and propionate (Sawyer et al., 1962). The finding that Prevotella and Porphyromonas spp. invariably produce H2S, suggests they might have a functional dissimilatory sulfur-type metabolism (Sawyer et al., 1962). Oral Prevotella spp. tend to be slightly more proteolytic, degrading collagen (Sawyer et al., 1962). Growth curves and radiolabelled tracing studies indicate Bacteroides, like Prevotella and Porphyromonads tend not to encode the capacity to synthesise porphyrin (Macy et al., 1975, Macy et al., 1978, Caldwell and Arcand, 1974, Sperry et al., 1977, Lev et al., 1971, Rizza et al., 1968, Gibbons and Macdonald, 1960, Paramaesvaran et al., 2003), yet fumarate reductase, which requires a source of porphyrin for incorporation into a b-cytochrome linked FRD is conserved in these genera (Shah and Gharbia, 1993). These Bacteroidetes therefore generally take up exogenous porphyrin.

Being porphyrin auxotrophs, it is unsurprising that many of the virulence mechanisms of the the assacharolytic porphyromonads are haem-dependent. For example, a P. gingivalis mutant deficient in haem uptake upregulated extracellular membrane vesicles and was more virulent in a mouse periodontal model than the parental strain (Genco et al., 1995), again indicating haem (and other tetrapyrrole) limitation may lead to tissue damage. Other investigators have demonstrated that extracellular vesicles are highly proteolytic (Grenier and Mayrand, 1987), degrading tissue proteins, including collagen (Smalley and Birss, 1987, Smalley et al., 1988), immunoglobulin A and immunoglobulin G (Grenier et al., 1989). Being 50-200 nm in size, released vesicles could diffuse to regions inaccessible to P. gingivalis and contribute to degradation of the epithelial barrier. Also, under high haem P. gingivalis displays multiple penta- and tetra- acetylated lipid A structures (Al-

13 Chapter I Introduction

Qutub et al., 2006). In contrast, under low haem, one major penta-acetylated lipid A structure is displayed. The low haem penta-acetylated lipid A functions as a Toll-like receptor 4 agonist (Al- Qutub et al., 2006), promoting E-selectin expression on human endothelial cells (Reife et al., 2006). In contrast, high haem associated tetra acetylated lipid A is a TLR4 antagonist (Coats et al., 2005, Reife et al., 2006), expected to dampen this innate immune response.

1.3 The biosynthesis, uptake and use of tetrapyrroles

1.3.1 Tetrapyrrole biosynthetic pathways

Tetrapyrroles form the backbone of the ordered extraction of chemical energy (life); from maintenance of the proton motive force (cytochromes), assimilatory sulfur and nitrogen metabolism (haems), extraction of electrons from water (chlorophylls) and oxygen carriage (haemoglobin/myoglobin). Indeed, the co-option of chlorophyll by cyanobacteria, initiated the great oxygenation event (circa. 3.2 billion years ago), converting electromagnetic radiation into chemical energy (ATP), shaping biological systems henceforth. Tetrapyrrole structural and functional diversity in various life forms is not surprising given the relative ease of formation with gamma radiation (Szutka et al., 1959) visible/ultra violet light (Szutka, 1964) and heat (Hodgson and Baker, 1967). How the genes for tetrapyrrole biosynthesis arose remain mysterious, however, the formation of ionic water-soluble precursors, steadily modified into hydrophobic molecules that aggregate, agrees with the Oparin-Haldane hypothesis for genesis of life (Mauzerall, 1992).

Tetrapyrroles, as the name indicates, consist of four pyrroles, tethered by four methene bridges in a linear or cyclic fashion. Each pyrrole is 5-membered, consisting of 4 carbon atoms and a nitrogen with a lone pair of electrons. The lone pair of electrons on the inner nitrogen atoms allow chelation of metals with varying valencies, which in biological systems is typically magnesium, cobalt, nickel and iron. For example, haem (iron-protoporphyrin IX), the prosthetic group of several proteins, cycles between Fe2+ (ferrous) and Fe3+ (ferrous) oxidation states, allowing the carriage and release of an electron, depending on the local pH and other physicochemical parameters. Being aromatic (with 26 delocalised elections) porphyrins exhibit 5 absorbance peaks, one broad peak in the Soret region (around 400nm), followed by four additional shoulder peaks at higher wavelengths. The addition of light to the soret peak or q-band peaks, results in an emission of photons at a higher wavelength (typically within 20 nm) and depending on the porphyrin, also between 607-625 nm. Whether metallated or not, porphyrins absorb in the Soret region. When metallated, porphyrins display reduced emission during excitation (Das, 1972).

14 Chapter I Introduction

The canonical eight-enzyme tetrapyrrole biosynthesis pathway, in which protoporphyrin IX is metallated with iron to form haem b, mostly studied in Escherichia coli (a ) and subtilis (a Firmicute), has been commonly adopted by metazoans. In the first of eight steps, 5-aminolevulinc acid (ALA) is synthesised via two routes. The first (route I, Fig. 1.5), using one enzyme, is present in humans and some (Panek and O'Brian, 2002). The majority of plants and metazoans synthesise ALA via route II (Fig. 1.5) (Dailey et al., 2015). Following ALA synthesis, two molecules of ALA are condensed into porphobilinogen, four of which are combined, forming the linear hydroxymethylbilane. The four pyrroles of hydroxymethylbilane are cyclised into uroporphyrinogen III. Uroporphyrinogen III is the first major branching point in the tetrapyrrole biosynthetic pathway. It was recently demonstrated that haem b may be synthesised via both deviations (see below). Next, in the canonical tetrapyrrole pathway, a series of de- carboxylations and condensations of side chains occur, making the porphyrinogens less water soluble. Isomers and spontaneous oxidation products of uro- and copro-porphyrinogens are generated. Some of the enzymes in the canonical pathway are dissimilar, used during anaerobic (underlined), aerobic or both (bold) conditions (HemF/HemN and HemG/HemY/HemJ). Finally, oxidation of protoporphyrinogen occurs before ferrochelation to form haem b (Fig. 1.6). Interestingly, in metazoans, following the synthesis of ALA (route II of Fig. 1.5), condensation of propionate groups on coproporphyrinogen to vinyl groups and oxidation of protoporphyrinogen occurs in the mitochondria.

An ‘alternative haem biosynthesis’ pathway (AhbA,B,C,D), originally described in the Deltaproteobacterium Desulfovibrio vulgaris (Ishida et al., 1998, Akutsu et al., 1993), has recently been confirmed in other denitrifying/sulfate-reducing bacteria as well as Archaea (Bali et al., 2011) in which haem b and haem d1 (the cofactor in dissimilatory nitrate reductase cd1 or NirS) are synthesised via precorrin-2 (Nir proteins). Also, in a recent addendum to the canonical porphyrin biosynthetic pathway, Firmicutes and Actinobacteria (Gram-positive bacteria unable to synthesise protoporphyrin) oxidise coproporphyinogen III forming coproporphyrin III (HemY), before iron insertion (HemH) and then carboxylation (HemQ) to generate haem b (Dailey et al., 2015).

Figure 1.5: The two routes, ‘ALA synthase’ (I) and C5 (II) for ALA synthesis (ALA is shown in Fig. 1.6).

15 Chapter I Introduction

In the most comprehensive survey conducted (1606 genomes), haem b synthesis by the canonical pathway was present in approximately 50% of the Bacteroidetes and the alternative pathway (AhbA,B,C,D) was absent [(Sousa et al., 2013) and (Fig 1.7)]. The authors did not include specific details of the search strategy, nor were details given of subclades, including lifestyle (host associated or free living). A particularly scathing commentary was given regarding accuracy of the analysis (Ducluzeau et al., 2014), in part because the ‘HemQ’ pathway was discovered after the analysis was performed. Nevertheless, the study is currently the most inclusive overview of quinone and tetrapyrrole biosynthesis. In an analysis of the presence and absence of haem b synthesis (using refSEQ), none of the Bacteroidetes examined in this thesis synthesised haem b (data not shown). The ‘HemQ’ and alternative pathways, are limited to Firmicutes/Actinobacteria and most denitrifying/sulfate-reducing bacteria as well as Archaea. Therefore, it is likely that the true number of host-associated Bacteroidetes that synthesis haem b is lower than 50%, although a thorough analysis of the canonical haem b biosynthetic pathway in Bacteroidetes is required. Note also that the capacity to synthesis porphyrin by a particular bacterium does not mean that exogenous porphyrin is not taken up (Cavallaro et al., 2008).

16 Chapter I Introduction

Figure 1.6: Tetrapyrrole biosynthetic pathways. Dashed arrows represent summarised pathways. SAM=S-adenosyl-L-methionine). Only bacterial enzyme nomenclature is shown.

17 Chapter I Introduction

Figure 1.7: A summary of haem b ‘canonical’, ahbA-D (‘alternative’) and quinone biosynthetic pathways in Archaea (117 genomes) and Eubacteria (1489 genomes) from Sousa, 2013.

18 Chapter I Introduction

1.3.2 Bacterial tetrapyrrole acquisition systems

Several strategies and receptors have been described for porphyrin uptake, most of which are described from Proteobacteria and focus on haem uptake, based on a persistent dogma that iron, rather than the porphyrin macrocycle per se, is required for growth. Gram-positive bacteria use several strategies for recognition and import of haem. For example, recognition and transfer of haem across the cell envelope of is mediated by IsdA/B/C/H; after which the ATP binding cassette transporters (ABC-transporters) IsdDEF (note that IsdE=HemT and Isdf=HemU) ferries haem into the cytoplasm, largely for degradation by the oxygenases IsdG/I (Grigg et al., 2007, Andrade et al., 2002, Pluym et al., 2008) and iron acquisition. Corynebacterium diphtheriae also uses an ABC-type transport system, encoded by hmuT/U/V (Drazek et al., 2000). Notably, haem has been demonstrated to passively diffuse across lipid bilayers in vitro. However, the outer membrane permeability in Gram-negative bacteria is less than 600 daltons, so this is unlikely to be a mechanism for tetrapyrrole assimilation (Genco and Dixon, 2001).

A common strategy used by Gram-negative bacteria after recognition of haem, is TonB- dependent import through the periplasm, coupled to the ExbB/D import system (Higgs et al., 2002). Gram-negative bacteria also excrete haemophores, such as the HasA system of Serratia marscens, which then import haem from the periplasm into the cytoplasm with ABC transporters, much like those in Gram-positive bacteria (Izadi et al., 1997). In B. fragilis, an outer membrane protein and ‘haem binding complex’ are implicated in haem uptake (Otto et al., 1991, Otto et al., 1994). In response to low haem, Prevotella intermedia expresses a high affinity receptor, depending on the

-9 -11 strain, approximating 10 and 10 M Kd with around 4,000 binding sites per cell (Tompkins et al., 1997). More recently, an outer membrane lipoprotein with high homology to P. gingivalis HmuY (PIN0009) and a homologue of the tonB-dependent outer membrane receptor used by various Gram-negative bacteria, including the HmuR receptor of P. gingivalis (PINA0611) were detected in P. intermedia grown with PPIX and the ferrous iron chelator dipyridyl (Yu et al., 2007). PINA0611 product is predicted to function in a seven gene operon, which has recently been characterised in Capnocytophaga canimorsus, a Flavobacteriaeae within the Bacteroidetes phylum. The seven-gene polysaccharide utilization locus (‘PUL3’) encoded by C. canimorsus, has been shown to acquire iron from serotransferrin, although the authors suggested this operon does not facilitate haem uptake (Manfredi et al., 2015). Interestingly, the gene in the centre of this locus in P. gingivalis (hmuS) and P. intermedia (PIN0012) is annotated as a magnesium chelatase, while in C. canimorsus it is annotated as a methyl transferase (03660). In P. gingivalis and P. intermedia, it is likely these genes are misannotated and that this is in fact a magnesium chelatase subunit I. It is also possible the gene encodes a magnesium-protoporphyrin IX methyltransferase in C. canimorsus.

19 Chapter I Introduction

Several receptors for and types of tetrapyrroles recognised and imported are characterised in P. gingivalis, the growth of which is supported by a wide array of porphyrins, including protoporphyrin IX, plus the derivative deuteroporphyrin IX (PPIX without the vinyl groups) and mesoporphyrin IX, although not by protoporphyrin IV or chlorin E4 (Hunter et al., 2002, Paramaesvaran et al., 2003). Various modifications to the vinyl groups normally present on haem b supported growth, and although growth was not as well stimulated, modification to both propionic acid side chains, including deuteroporphyrin IX di-taurine amide stimulated improved growth relative to DMSO (Fig. 1.8) (Paramaesvaran et al., 2003). Indeed, on the basis that various derivatives of the propionic acid side chains are recognised and imported by P. gingivalis, several amide and lysine linked deuteroporphyrin adducts were synthesised (see also Section 1.5), the majority of which were also taken up (Yap et al., 2009, Dingsdag et al., 2015). However, several of these adducts did not support growth, suggesting tetrapyrrole uptake receptors in P. gingivalis have a broader recognition of porphyrin derivatives that can be incorporated into b-cytochrome to enable fumarate reductase activity.

The highest affinity receptors originally described in P. gingivalis under haem-depleted

-12 conditions approximated 0.4- 1 × 10 M Kd for haem, with around 1,500 binding sites per cell, depending on the strain (Tompkins et al., 1997). The same study described a low affinity receptor,

-8 5 with a lower binding affinity for haem (0.2-7 × 10 M Kd), and around 1.5 × 10 binding cites per cell. The outer membrane haem-binding protein HmuY and associated receptor HmuR, appear to form an operon in P. gingivalis that is also conserved in B. thetaiotaomicron and B. fragilis (Olczak et al., 2008). Also in P. gingivalis, Omp26 is a low affinity outer membrane protein, which binds haem before the protein-haem complex is imported (Bramanti and Holt, 1993). P. gingivalis also expresses on the outer membrane and secretes HusA, which has high affinity for haem (Gao et al., 2010).

Figure 1.8: Haem b (A), deuteroporphyrin IX di-taurine amide (B) and P5 (C).

20 Chapter I Introduction

HusA may be the receptor originally described by Tompkins et al., 1997. As mentioned, uptake of haem and other tetrapyrroles is better characterised in Proteobacteria than Bacteroidetes, including

-10 the Bacteroides. While high affinity (approximating 10 Kd M or higher) tetrapyrrole receptors have yet to be described in Bacteroides, it is known that B. thetaiotaomicron encodes three B12 transport systems, BtuB1-B3, which have some specialisation and also duplication in the types of corrinoids recognised and taken up (Degnan et al., 2014). Given redundancy is a general feature of biological systems, it is possible that Bacteroides also express more than one high affinity porphyrin-uptake receptor, with similar redundancy and/or specialisation for porphyrin recognition. It is also possible the BtuB1-B3 recognise and mediate uptake of porphyrins.

1.3.3 Respiration and the tricarboxylic acid cycle

Over 200 years ago, Charles MacMunn described respiratory pigments, ‘myohematin and histohematin’ in animal tissue (Munn, 1886). Then in the 1920s, working with horse tissue, David Keilin renamed these respiratory pigments as cytochromes, classifying them as a, b or c type, depending on the type of haem (a, b, c and later d) molecule that was coordinated with proteins and which region of the visible light spectrum absorbed the most photons (Keilin, 1925). The cytochromes involved in this electron transport chain function in the inner mitochondrial membrane using oxygen as a terminal electron acceptor and with mediation by dehydrogenases, ubiquinone

(coenzyme Q10) and menaquinone (K2) (Vos et al., 2012, Beyer, 1958). Soon after, Krebs described a series of cyclic reactions involving the complete oxidation of an unknown ‘triose’ (phosphoenolpyruvate) produced from the fermentation of sugars (glycolysis) or tricarboxylic acid cycle (TCA) (Krebs and Johnson, 1937). The TCA cycle, common in aerobic metabolism, oxidises acetate generated from carbohydrates, fatty acids and amino acids, giving rise to precursors for amino acids such as alanine, aspartate and glutamate. Operating in the mitochondria of metazoans, the TCA cycle also liberates NADH which is required for oxidative phosphorylation. The aerobic oxidation of NADH, which generates ATP, relies on electron transport phosphorylation (‘ETP’), using the proton translocating F1F0 ATPase (Thauer, 1988).

As early as 1886, Hoppe-Seyler described the complete oxidation of cellulose in anaerobic sediments supplemented with CaSO4 (Hoppe-Seyler, 1886). The quantitative recovery of H2S, suggested the existence of an anaerobic electron transport chain. Then, work in the 1950s established the requirement for haem as a growth factor for Prevotella spp. (Evans, 1951). It was known that Prevotella spp. stored ‘haematin’ (ferric-PPIX) on the surface (Schwabacher et al., 1947) and it was postulated that haem was involved in an electron transport system, although the function

21 Chapter I Introduction of this system was unknown (Gibbons and Macdonald, 1960). It was also known that photosynthetic (Vernon and Kamen, 1954), sulfate reducing (Postgate, 1954, Adams and Postgate, 1959) and other anaerobes used cytochromes (Hayward and Stadtman, 1959). However, it wasn’t until the 1960s that incorporation of haem b into a b-cytochrome, enabling fumarate reductase (FRD) activity, was described for Prevotella ruminicola (formerly Bacteroides ruminicola) (White et al., 1962). The use of FRD implied P. ruminicola was generating ATP by substrate level phosphorylation and also by anaerobic respiration (see Thauer et al., 1977 for an exhaustive review). Secondly, this implied succinate-forming members of the Bacteroidetes might also use b-cytochromes with a similar metabolism (Macy and Probst, 1979).

Soon the first reductive (or reverse) tricarboxylic acid cycle (rTCA) was described in the anoxygenic photosynthetic bacterium, Chlorobium thiosulfatophium, that used reduced ferredoxin to drive CO2 assimilation, generate α-ketoglutarate, a precursor of the amino acids (glutamate, aspartate and alanine) and lipid precursors (Evans et al., 1966). The rTCA cycle used by C. thiosulfatophium is endergonic (requiring energy in the form of ATP), so must be linked with energy yielding (exergonic) reactions, supplied by photochemical production of ATP and reduced ferredoxin. The rTCA cycle is considered the precursor to the Krebs cycle (Buchanan and Arnon, 1990). Portions of the rTCA cycle, particularly the reductive carboxylation of succinate to form α- ketoglutarate, would later be found to be shared by the Bacteroidetes.

The use of a b-cytochrome, forming a FRD, was later described in a diverse range of prokaryotes including: Bacteroidetes [Bacteroides spp. (Macy et al., 1975, Macy et al., 1978, Reddy and Bryant, 1967), Prevotella spp. (Rizza et al., 1968, Gibbons and Macdonald, 1960, Caldwell et al., 1965) and Porphyromonas spp. (Shah and Williams, 1987, Dashper et al., 2009)], Firmicutes [Clostridium spp., (Gottwald et al., 1975), Selenomonas rumantium and Veillonella alcalescens (de Vries et al., 1974)] and Proteobacteria [Wollinella succinogenes (Reddy and Peck, 1978)]. The b- cytochrome system in these bacteria employs two molecules of haem b as a prosthetic group, in concert with NADH, formate of H2 as an electron donor (Asanuma and Hino, 2000). Some Bacteroidetes maintain a portion of the oxidative (oTCA) and rTCA (as in Bacteroides) (Macy et al., 1975, Macy et al., 1978, Baughn and Malamy, 2003), while other nanaerobes only encode a reductive portion (as in the majority of the Prevotella and Porphyromonas) (Shah and Gharbia, 1993) (Fig. 1.9). Parts of the rTCA performed by Bacteroidetes are similar to the full rTCA cycle employed by C. thiosulfatophium, including the incorporation (fixation) of CO2 into larger molecules.

22 Chapter I Introduction

Figure 1.9: rTCA (‘haem-dependent’) and oTCA (haem-independent) portions of the TCA cycle proposed in B. fragilis (from Baughn and Malamy, 2002). The use of the rTCA is broadly conserved in Bacteroidetes, whereas the oTCA is limited to the Bacteroides.

The FRD of Bacteroides, Prevotella and Porphyromonas has high homology (see Chapter III and Appendix Table A4), consisting of three genes (frdC, frdA and frdB). In B. fragilis frdC encodes the b-cytochrome with five inner-membrane spanning helices that coordinate two molecules of haem. Two of these helices anchor an iron sulfur cluster encoded by frdB and a third unit, frdA, encodes a flavoprotein (Baughn and Malamy, 2003). In Bacteroides, exogenous haem is used for b- cytochrome formation enabling FRD activity (Macy et al., 1975, Macy et al., 1978, Allison et al., 1979,

Baughn and Malamy, 2003). FRD couples the two electron reduction of fumarate with NADH or H2, using menadione (vitamin K2) and membrane bound dehydrogenase/s, yielding succinate (Macy et al., 1975, Reddy and Harris, 1976, Macy et al., 1978). The reduction of fumarate to succinate is likely to occur in the cytoplasm, coupled with the cyclic re-oxidation of menaquinol (MQH2) to

+ menaquinone (MQ) and oxidation of NADH2 to NAD in the periplasm (Kotarski and Salyers, 1984, Kröger et al., 1992). This system will be coupled to proton or Na+ export across the inner membrane into the periplasm, creating chemiosmotic potential for phosphorylation of ADP (ATP formation) (Kröger et al., 1992). An F-type ATPase in B. fragilis may mediate this system (Baughn and Malamy, 2003), and is likely to be employed by other Bacteroides. More broadly, FRD plays a role in the

23 Chapter I Introduction reductive portion of the tricarboxylic acid cycle (rTCA), permitting redox maintenance (NAD+ regeneration) and through reductive carboxylation of succinate, forms α–ketoglutarate, the precursor of glutamate, arginine, proline and aspartate (Allison and Robinson, 1970, Macy et al., 1975, Macy et al., 1978). In Bacteroides exogenous supplementation of porphyrin enabling FRD, permits the extraction of more than double the amount of ATP per molecule of glucose than in the absence of glucose (Macy et al., 1975, Macy et al., 1978, Sperry et al., 1977). The presence of an oxidative portion of the TCA cycle, as demonstrated in B. fragilis (Baughn and Malamy, 2002), might obviate the need for external porphyrin supplementation. The absence of the oTCA pathway might explain porphyrin auxotrophy of various Prevotella and Porphyromonas spp. (Allison and Robinson, 1970). In the initial report of a b-cytochrome in Prevotella ruminicola, an ‘o type’ (oxidase) was also described (White et al., 1962). Soon after, an o-type cytochrome was described in B. fragilis (Reddy and Bryant, 1967). It was known that B. fragilis and , as well as other ‘anaerobes’ tolerated oxygen (Rolfe et al., 1977, Rolfe et al., 1978). It was also known FRD functions as a succinate oxidase (Kotarski and Salyers, 1984, Baughn and Malamy, 2003). More recently, under low (nanomolar) concentrations of oxygen, it was demonstrated that one of the FRD components

(FrdC) of B. fragilis appears to couple succinate oxidation to reduction of O2 via a cytochrome bd oxidase (CydAB) (Baughn and Malamy, 2004). The oxidation of succinate to fumarate by FRD represents a thermodynamic barrier [fumarate/succinate (E’0 = +30mV)], requiring a terminal electron acceptor with a more positive (E°’) redox potential than that of fumarate/succinate couple,

- - such as electron acceptors used by aerobes (including O2/H2O: E°’=+818mV and NO3 /NO2 : E°’=+433mV) (Thauer, 1988). Sulfur, sulfate and thiosulfate would also suffice in some anaerobes as a terminal electron acceptors (Thauer, 1988). In P. gingivalis, CydAB may also link the oxidation of succinate to fumarate via low concentrations of oxygen (Meuric et al., 2010). However, this model is based on studies of Desulfovibrio gigas, and has not been experimentally verified. During haem limitation, a slight reduction in FrdB and FrdA were observed for P. gingivalis, although not FrdC (Dashper et al., 2009), which is in keeping that FrdC is involved in an electron transport chain involving limited oxygen (CydA and CydB) (Meuric et al., 2010, Baughn and Malamy, 2004). Interestingly, in P. gingivalis, the F-type ATPase is absent, (Meuric et al., 2010), instead an operon of the V-type ATPase is present (Nelson et al., 2003). Since some Prevotella and Porphyromonas spp. produce H2S (Sawyer et al., 1962), it is also possible these strains use sulfur, sulfate and thiosulfate as electron acceptors in this pathway.

24 Chapter I Introduction

1.4 Microbial metabolism and nitroimidazole reduction (activation)

This thesis reviews the history, use and cytotoxicity of metronidazole with reference to other nitroimidazoles where relevant. Despite more than 60 years research, the metabolism of metronidazole and associated cytotoxicity is not definitively characterised. Nitroimidazoles are prodrugs that are reduced (activated) under anaerobic conditions. There is divided opinion regarding the enzymes responsible for reduction of nitroimidazole prodrugs. It also remains unclear if nitroimidazole reduction (activation) is a primary cytotoxic process, or whether subsequent fragmentation of the imidazole ring and formed metabolites alone mediate the cytotoxicity (Church et al., 1988, Tally et al., 1978, Church and Laishley, 1995). Lastly, the presence of oxygen reduces the effectiveness of nitroimidazole activation, confounding study of these prodrugs.

For purposes of clarity, in this review the metabolism of metronidazole is divided into the synthesis of azomycin, the natural precursor of metronidazole (Section 1.4.1), the structure and use of nitroimidazoles (Section 1.4.2), nitroimidazole activation (reduction) and associated primary effects on microbial metabolism (Section 1.4.3). The structural basis of metronidazole fragmentation and cytotoxicity of reduced fragments of metronidazole are also discussed (Section 1.4.4). Lastly, resistance to metronidazole is reviewed (Section 1.4.5). The current understanding of metronidazole metabolism is described in detail since the reduction of the nitro group appended to deuteroporphyrin adduct characterised in this thesis is predicted to be similar to the reduction of metronidazole and other nitroimidazoles.

1.4.1 Biosynthesis of azomycin the biotic precursor of metronidazole

Metronidazole is a bactericidal synthetic derivative of azomycin, originally detected during the 1950s from cultures of Streptomyces spp. (Maeda et al., 1953). Azomycin is largely produced by Actinobacteria including Streptomyces (Lancini et al., 1966, Lancini et al., 1968) such as Streptomyce eurocidicus (Osato et al., 1955) and Nocardia mesenterica (Nocardiaceae) (Okami et al., 1954, Maeda et al., 1953). Cultures of Proteobacteria (Pseudomonas fluorescens) also produce azomycin (Shoji et al., 1989). In S. eurocidicus, 2-aminoimidazole, the precursor of azomycin, is considered to be a metabolic product of L-arginine. The proposed metabolism involves three steps, forming 2-aminoimidazole (Fig. 1.10, compound I) before a final oxidation of 2-aminoimidazole to 2- nitroimidazole (Fig. 1.10, compound II) (Nakane et al., 1977). The reaction is localised to the

25 Chapter I Introduction

Figure 1.10: The proposed final step of ‘azomycin’ (2-nitroimidazole, compound II) synthesis in S. eurocidicus. The ‘oxygenase’ may be AurF or a homologue. Note that unlike the natural precursor azomycin, metronidazole (compound III, shown for reference purposes) contains a hydroxyethyl group at position 1, a methyl group at position 2 and a nitro group at position 5 (and not position 2) of the imidazole.

mycelium of S. eurocidicus (Seki et al., 1970). In Streptomyces thioluteus, the oxidation of various six- membered aromatic rings containing nitro groups, for instance p-aminobenzoate to p-nitrobenzoate of the antibiotic aureothin, involves the oxygenase AurF in a reaction requiring oxygen (Kawai et al., 1965, Krebs et al., 2007).

In Bacillus subtilis (an aerobe) and Escherichia coli (a facultative anaerobe) of the Firmicutes and Proteobacteria phyla respectively, it has been suggested that azomycin inhibits ribonucleotide reductase (‘RNR’, EC 1.17.4.1) (Saeki et al., 1974). Ribonucleotide reductases are conserved in all domains of life, catalysing the reduction of the four ribonucleotide diphosphates (NTPs) to corresponding deoxyribonucleotide triphosphates (dNTPs) the precursors of deoxyribonucleotides (DNA) that allow DNA synthesis and repair (Torrents, 2014). Inhibition of RNR should prevent DNA repair and prevent synthesis of DNA during replication. The possible inhibition of RNR by nitroimidazoles is discussed in more detail in Section 1.4.3. It is interesting that azomycin is produced by soil dwelling microbes such as Streptomyces, since azomycin presumably inhibits DNA replication in bacteria that produce this metabolite. It is possible that during nutrient and/or water depletion, azomycin may encourage spore formation in Streptomyces.

1.4.2 The structure and use of nitroimidazoles

As indicated by the nomenclature, 5-nitro antibiotics share a nitro functional group (-NO2) at the fifth position of a planar 5-membered ring, with a nitrogen (nitroimidazole), oxygen (furan) or sulfur (thiazole) at position 1. A variety of side chains may be appended to the imidazole, such as the

26 Chapter I Introduction methoxy-propanol (misonidazole), hydroxy-ethyl (metronidazole), and ethyl-amine (‘metronidazole- amine’) (Fig. 1.11). The heterocycles are aromatic, sharing a conjugated system of single and double bonds, in which a lone pair of electrons from the nitrogen, oxygen or sulfur is delocalised in the ring. The nitro group of these prodrugs is required for cytotoxicity [imidazole, 2-imidazole and the amino- derivative of metronidazole are inactive (see compound V of Fig 1.12 for structural details) (O'Brien and Morris, 1971, Ehlhardt et al., 1987).

Metronidazole was originally used against the anaerobic protozoan Trichomonas vaginalis. It has since been used against other protozoa, including Entamoeba histolytica (amoebic dysentry) and Giardia lamblia (giardiasis). However, it wasn’t until 1962 during treatment for bacterial vaginitis, that it was realised that metronidazole also treated bacterial gingivitis in the same patient (Shinn, 1962), making in-roads for the eventual use against other nanaerobic bacteria, such as in the treatment of a suite of acute and chronic inflammatory bowel diseases. Despite the diversity in physiology and life cycles of the nitroimidazole-sensitive organisms listed above, nitroimidazoles cross these taxonomic boundaries because these organisms occupy niches with low oxygen tension that allow reduction of the nitro group. Under conditions favouring aerobic respiration nitroimidazole prodrugs are ineffective; these prodrugs are either not reduced (activated) or oxygen prevents full reduction of the nitro group and generation of cytotoxic metabolites.

Notably, nitroimidazoles have also been demonstrated both in vitro and in vivo to act as radiosensitizers for cancerous cells (Asquith et al., 1974, Begg et al., 1974, Mohindra and Rauth, 1976). In an in vitro model, high concentrations of metronidazole (10 mM) have been shown to inhibit colony formation in Chinese hamster ovary cells (Ehlhardt et al., 1988). Introduction of E. coli nitroreductase (nfsB) as an expressed transgene in zebrafish, leads to death of targeted cells (Curado et al., 2008). NfsB is an oxygen-insensitive nitroreductase, predicted to catalyse the

Figure 1.11: Misonidazole, metronidazole and ‘metronidazole-amine’: a variety of aromatic nitroimidazoles sharing the common nitro functional group (-NO2) at position two or five of the imidazole. Note that ‘metronidazole-amine’ linked to deuteroporphyrin forms P5.

27 Chapter I Introduction reduction of the nitro group of metronidazole to the amino derivative that whilst non-toxic for bacteria, may be toxic for mammals by further non-enzymatic activation with thioesters (Knox et al., 1993). Mammalian selenoprotein thioredoxin reductase (TrxR) catalyses the 1- and 2-electron reduction of various poly-and mono- nitroaromatic compounds (Cenas et al., 2006). Mammalian NAD(P)H:quinone reductase (DT-diaphorase) also catalyses the 1-electron reduction of mono- nitroaromatic compounds (Knox et al., 1993). The reductive activation of nitroaromatic compounds such as metronidazole is inhibited by the presence of oxygen which is why these compounds are only reduced to cytotoxic intermediates under hypoxic conditions (see in Section 1.4.3). Interestingly, nitroimidazoles have a structural similarity to uncouplers of oxidative phosphorylation [reviewed in (Terada, 1990)], containing an electron withdrawing (nitro) group, a weak acid (hydroxethyl) and a hydrophobic ring (imidazole). It is possible that before reductive activation and cleavage, nitroimidazoles oscillate between the inner- and outer-mitochondrial membrane. The serial protonation and de-protonation of the hydroxethyl functional group could then dissipate the proton gradient.

1.4.3 Primary effects on microbial metabolism of the uptake and reductive activation of metronidazole

A receptor for metronidazole uptake has not been described and uptake is considered a passive process. However, nitro group reduction and associated cytotoxicity is correlated with the rate of metronidazole uptake (Narikawa, 1986) indicating uptake is not passive. Indeed, the oxidative phosphorylation inhibitor, carbonyl cyanide m-chlorophenyl hydrazone (CCCP) slows metronidazole uptake in Helicobacter pylori and Clostridium pasteuranium suggesting uptake is dependent on a proton gradient (Moore et al., 1995, Church and Laishley, 1995). This gradient is probably dependent on electron transport since uptake is also inhibited by potassium cyanide and oxygen (KCN) (Moore et al., 1995, Church and Laishley, 1995). The concentration of sucrose is also correlated with metronidazole uptake (Church and Laishley, 1995). Inhibitors of glycolysis (sodium fluoride, arsenate and iodoacetic acid) also slow uptake of metronidazole (Nseka and Müller, 1978, Church and Laishley, 1995). Uptake is rapid in Clostridium spp., Trichomonads and Entamoeba (Obrien and Morris, 1972, Church et al., 1988, O'Brien and Morris, 1971); intermediate in Fusobacteria spp. and slowest in Bacteroides spp. (Narikawa, 1986). Other than E. coli and Enteroccocus faecalis, few facultative anaerobes take up metronidazole (Tally et al., 1978, Müller, 1983, Narikawa, 1986).

28 Chapter I Introduction

Metronidazole is inert until taken up. Reduction of the nitro group of metronidazole occurs by two routes, with important ramifications for cell survival. Reduction of the nitro group may lead to activation of metronidazole resulting in imidazole fragmentation and cytotoxicity or inactivation of the nitro group to the non-toxic amino derivative (see overview in Fig. 1.12). The inactivation of metronidazole to the non-toxic amino derivative is associated with metronidazole resistance and is also discussed in Section 1.4.5. The reductive activation of metronidazole consumes 4-electrons and is proposed to occur in a series of 1- or 2-electron steps, leading to ring-fission and formation of transient cytotoxic derivatives (Church et al., 1988, Peterson et al., 1979, Lockerby et al., 1984). It has been proposed that the reductive activation of metronidazole exerts a primary cytotoxic effect, since metronidazole acts as an alternative electron acceptor, inhibiting the proton motive force and diminishing ATP production (Tally et al., 1978, Church et al., 1988). The first step of the reductive activation of metronidazole is proposed to form the nitro free radical (Fig 1.12). Subsequent steps in the reductive activation are proposed to form the nitroso, nitroso free radical and hydroxylamine derivatives. Due to their unstable , the intermediates have not been isolated from a biological context. Given that intermediates have been poorly studied, it is possible other enzyme/s are involved in heterocycle fission. Oxygen has a higher affinity for electrons than metronidazole. In the presence of oxygen, the nitroso radical has been suggested to be reduced to form a superoxide anion; followed by removal of an electron from the nitroso radical by oxygen, regenerating metronidazole in a process called ‘futile cycling’ (Church and Laishley, 1995). The reductive inactivation of the nitro group to the stable amino-derivative by nitroreductases occurs in 2-electron steps, consuming 6-electrons rendering metronidazole inactive and is oxygen insensitive (Roldán et al., 2008). Reductive inactivation of metronidazole is discussed further under metronidazole resistance (Section 1.4.5).

A thermodynamic logic was originally used to describe how metronidazole interferes with bacterial physiology, leading to the reduction of the nitro group and formation of cytotoxic products.

The thermodynamic logic employs the oxidation-reduction potential (E0), or the tendency of a chemical to accept electrons to predict the flow of electrons between low redox potential molecules and their enzyme carriers. In order to understand the reductive activation of metronidazole and the thermodynamic logic of activation, it is useful to outline pertinent aspects of microbial metabolism. In nanaerobic bacteria, in the cytosol of giardia and amoeba and in the hydrogenosome of trichomonads (Müller, 1986, Edwards and Mathison, 1970), pyruvate is formed during glycolysis, with a net production of two molecules of ATP and NADH per saccharide molecule (for example glucose). Next fermentation occurs, during which pyruvate is oxidatively decarboxylated. The

29 Chapter I Introduction

A) Reductive activation

B) Reductive inactivation

Figure 1.12: The reductive activation of metronidazole leading to heterocycle fission, transient compound formation and cytotoxicity (A) or reductive inactivation of metronidazole forming the 5- amino derivative (B). The reductive activation of metronidazole leading to cytotoxic derivative formation is hypothesised to occur via 1- or 2-electron steps forming transient intermediate compounds II – IV and fission of the heterocycle (marked with dashed lines in compound IV). Note compounds II – IV have not been isolated during biological reductive activation of metronidazole (A). The presence of oxygen is proposed to regenerate metronidazole in the so-called ‘futile cycle’ (A). Reductive activation of the nitro group leading to heterocycle fission (A) may be accomplished by redox transfer proteins, such as ferredoxin, acting either alone or in concert with hydrogenase, flavodoxin, and effectors of the dissimilatory sulphite pathway. The reductive inactivation of the nitro group to the 5-amino derivative (compound V) is proposed to occur in 2-electron steps by type I oxygen-insensitive nitroreductases, rendering metronidazole non-toxic (B).

30 Chapter I Introduction oxidation of pyruvate is driven by pyruvate:ferredoxin oxidoreductase (‘PFOR’, EC 1.2.7.1) forming acetyl phosphate, liberating carbon dioxide, and in turn reducing ferredoxin. Electrons from reduced ferredoxin are coupled to the reduction of low redox potential molecules by hydrogenase 1 (EC 1.12.1.1), an enzyme which couples hydrogen ions to these electrons, reversibly liberating molecular hydrogen (Müller, 1986) (Fig. 1.13). Normally, electrons flow from pyruvate/acetate and

CO2 couple [E0’ = -700 mV, (Thauer et al., 1977)] through PFOR [containing three iron-sulfur (FeS)

Glucose Metronidazole

Lactate

+ 2 PEP LDH NAD NADH + H+ 2NAD+ 2NADH +2H+ +2ATP 2e- + 2H+ Pyruvate Ferredoxin Metronidazole (reduced) PFOR HYD CoA-SH Ferredoxin ‘Reactive Cytotoxicity (oxidised) intermediates’

+ Acetyl-CoA +CO2 +2H H2

Acetyl-Phosphate N-(2-Hydroxyethyl)-oxamicacid

Acetate + Acetamide

Figure 1.13: Generalised schema of metronidazole reductive activation. One proposed metabolism involving hydrogenase is based on Church et al., 1988 (see other references in text). Some pathways are summarised by dashed arrows. PEP= phosphoenolpyruvate. During glycolysis, pyruvate is formed from the catabolism of saccharides such as glucose. During fermentation, electrons from pyruvate flow to ferredoxin, mediated by pyruvate:ferredoxin oxidoreductase (‘PFOR’). Reduced ferredoxin normally donates electrons to protons, by enzymatic coupling for example with hydrogenase 1 (‘HYD’). In susceptible species, metronidazole enters cells rapidly (dotted line), competing with hydrogen ions (protons) for electrons, so that reducing equivalents are consumed and the nitro functional group is reduced, resulting in heterocycle fission. Note that pyruvate may also be fermented by lactate dehydrogenase (‘LDH’, EC 1.1.1.27) to lactate. Fermentations culminating in butyrate, formate, ethanol and butanol, reflecting metabolism in some Clostridia or facultative bacteria, are not shown. Some products of reduced metronidazole are not shown.

31 Chapter I Introduction

clusters, the most negative of which is E0’ = -540 mV (Chabrière et al., 1999) to reduced ferredoxin

+ (E0’ = -420 mV) and to the proton [2 H /H2 = -414 mV, (Kaster et al., 2011)]]. Originally, it was proposed that PFOR/ferredoxin were the only couple in nanaerobes with a sufficiently low redox potential to reduce metronidazole [E0’ = - 470-510 mV (Lindmark and Müller, 1976, Church et al., 1988, O'Brien and Morris, 1971, Obrien and Morris, 1972, Edwards and Mathison, 1970)]. However, there are discrepancies as to the exact nature of metronidazole reduction. Originally, it was proposed metronidazole accepts electrons directly from ferredoxin (Lindmark and Müller, 1976, Obrien and Morris, 1972). Chen and colleagues next reported that hydrogenase reduces metronidazole at a low rate and that purified ferredoxin or flavodoxin from C. pasteuranium reduced metronidazole directly (Chen and Blanchard, 1979). Other reports have confirmed expression of flavodoxin from C. pasteuranium renders resistant E. coli sensitive to metronidazole (Santangelo et al., 1991). Later, metronidazole was demonstrated to be reduced by hydrogenase 1 from C. pasteuranium, effectively competing with protons (H+) as an acceptor of electrons (Church et al.,

2- 1988). Metronidazole also competes with sulfite (SO3 ) for reducing equivalents originating from ferredoxin in the dissimilatory sulfite pathway, as noted in studies of C. pasteuranium (Laishley and Krouse, 1978, Lockerby et al., 1984). PFOR inactive mutants of C. perfringens spp. and Trichomonad spp. are up to 100-fold more resistant to metronidazole although are still killed (Sindar et al., 1982, Čerkasovová et al., 1984, Hrdý et al., 2005). Similarly, a B. fragilis double mutant of PFOR and flavodoxin is around 10-fold more resistant to metronidazole, although is still killed (Diniz et al., 2004) confirming activation of metronidazole can occur independently of PFOR. Other candidates that may function in the reductive activation of metronidazole are the poorly characterised oxygen- sensitive (type II) nitroreductases (Roldán et al., 2008). Lastly, metronidazole is also reductively activated by sodium dithioninte [E0’ = - 660 mV, (Mayhew, 1978)] in a reaction consuming 4- electrons, which is irreversible because it induces ring fission (Chen and Blanchard, 1979).

Oxidation-reduction potential is typically measured using a hydrogen electrode, describing the thermodynamic feasibility of a particular redox reaction. Notably, enzymes such as hydrogenase that catalyse the reductive activation of metronidazole described above, exhibit variable reduction potentials depending on bound substrate/s and local environment. It is paradoxical that

+ metronidazole and other nitroimidazoles have a lower reduction potential than the H /H2 pair (E0’ = -414 mV (Thauer et al., 1977)), yet preferentially accept electrons. Also of note, oxygenated nitrogenous intermediates usually accept electrons during anaerobic respiration, although the redox

- potential of these electron acceptors, such as NO2 /NO pair [E0’ = +350 mV, (Thauer et al., 1977)] are more positive than metronidazole. The presence of a nitro (-NO2) functional group, similar to nitrite

- (NO2 ), apparently confers special properties to metronidazole.

32 Chapter I Introduction

There is conflicting evidence as to if and how metronidazole is metabolised by facultative anaerobic bacteria. It has been proposed that few facultative anaerobes, other than E. coli, E. faecalis and Klebsiella pneumoniae take up metronidazole (Tally et al., 1978, Müller, 1983, Narikawa, 1986). Furthermore, it has been reported that even if metronidazole is taken up by facultative bacteria such as E. coli, that this molecule is not activated due to the use of NAD+ coupled pyruvate-formate lyase (PDH) instead of PFOR during anaerobic cleavage of pyruvate to acetyl-CoA (Samuelson, 1999, Charon et al., 1999). However, metronidazole-mediated killing of E. coli is documented (Jackson et al., 1984, Tamae et al., 2008). Additionally, the rate and total amount of metronidazole taken up by E. coli is higher than for B. fragilis (Chrystal et al., 1980a). However, the rate of kill and liberation of acetamide, a major product formed after reduction of metronidazole and imidazole fission, is higher in B. fragilis than E. coli (Chrystal et al., 1980a), suggesting heterocycle fission and cytotoxic products are less efficiently formed in E. coli. Indeed, the cytotoxic concentration of metronidazole for facultative anaerobes is 1-3 orders of magnitude higher than corresponding concentrations for nanaerobic bacteria, even under strict anaerobic conditions (Tally et al., 1978, Ingham et al., 1980, Jackson et al., 1984, Narikawa, 1986, Yeung et al., 1984, Tamae et al., 2008, Lubbe et al., 1999). As mentioned, the reduction of the nitro group to the stable amino- derivative by nitroreductases renders metronidazole inactive. The NfsA, NfsB and YdjA nitroreductases of E. coli are type I oxygen-insensitive (see Fig. 1.12), reducing metronidazole to the amino derivative (Roldán et al., 2008). These type I oxygen-insensitive nitroreductases may account for higher rates of uptake accompanied by reduced cytotoxic potency in facultative bacteria such as E. coli than nanaerobic bacteria. It is likely that type II oxygen-sensitive nitroreductases of facultative anaerobes play a role in reductive activation of metronidazole. Pyruvate:flavodoxin oxidoreductase (YdbK) from E. coli and related enzymes in facultative bacteria may also reductively activate metronidazole, leading to ring fission and cytotoxic derivative formation (Nakayama et al., 2013).

1.4.4 Distribution and metabolism of metronidazole in human and murine models

Metronidazole is rapidly absorbed in humans (Ralph et al., 1974), mostly entering the gastrointestinal tract directly through the mucosa, rather than via enterohepatic circulation (through the bile duct in bile) possibly due to the molecules’s low molecular weight (Ings et al., 1975). Metronidazole is found in all tissues in mice and rats, after oral or intravenous injection. The highest concentrations are found in liver, bladder, kidneys, vagina and gastrointestinal tract (Placidi et al., 1970, Ings et al., 1975, Manthei et al., 1969). The majority of metronidazole in human or murine

33 Chapter I Introduction models is excreted in urine unchanged or as oxidised derivatives (compounds VIII - XII, Fig. 1.14) (Ings et al., 1966, Ings et al., 1975, Ralph et al., 1974). High concentrations of unchanged metronidazole are detected throughout the gastrointestinal tract of rats, except for the caecum. It is likely that this structure, containing the highest load of nanaerobic microorganisms (see Section 1.1.1 of this thesis for further details) is the major site of nitroimidazole reduction (Ings et al., 1975). Reduced products of metronidazole, including acetamide and hydroxyethyl-oxamic acid, are detected in urine of conventional rats and not germ-free rats, indicating that the intestinal microbes are required to reduce metronidazole (Koch et al., 1979).

From in vitro studies of the oxidised products of metronidazole, the ‘acid’ and ‘alcohol’ (IX and X respectively in Fig. 1.14) have been estimated to account for none-5% and 30-65% respectively of the toxicity exerted by metronidazole for Clostridium spp. and Bacteroides spp. (Ralph and Kirby, 1975, Haller, 1982, O'Keefe et al., 1982). Determining the reduced intermediates and how they might exert cytotoxicity after metronidazole activation (reduction) is challenging, since intermediates are short-lived (Church and Laishley, 1995). Nevertheless, following enzymatic or chemical nitro group reduction (activation of metronidazole), the imidazole ring is irreversibly fragmented, transiently forming reactive intermediates and finally, stable end products. In the urine of rats treated with metronidazole, N-(2-hydroxyethyl)-oxamic acid (Koch and Goldman, 1979, Ings et al., 1975) (compound VI, Fig. 1.14) and acetamide (Koch et al., 1979) (compound VII, Fig. 1.14) were detected. These metabolites are consistent with cleavage of the imidazole between positions 1 and 2 plus positions 3 and 4; however, they do not account for all of the carbon and nitrogen in metronidazole. Together these metabolites account for only 20% of total metronidazole administered to rats recovered from urine (Koch et al., 1979, Koch and Goldman, 1979). Importantly, while it is not possible to recover all of the metabolised metronidazole, around half of the reduced derivatives of total administered metronidazole are unaccounted for in the excreta of humans and rodents (Stambaugh et al., 1968, Nilsson-Ehle et al., 1981, Koch et al., 1979, Koch and Goldman, 1979) (see Fig. 1.14).

Two in vitro models have been used to study biological reduction of metronidazole. The first, an enzymatic model consisting of xanthine oxidase (EC 1.2.3.2) and metronidazole produced similar concentrations of N-(2-hydroxyethyl)oxamic acid (compound VI, Fig. 1.14) and acetamide (compound VII, Fig. 1.14) to those detected in rat studies. Additional products included N- acetylethanolamine, n-glycoylethanolamine, ethanolamine, acetate, and glycine (Chrystal et al., 1980b). In the second model, metronidazole was reduced electrolytically forming at least 12 products, including ethanolamine, N-acetylglycine, glycine and acetate, in agreement with a fragmentation between carbons 2 and 4 of the imidazole ring (Knox et al., 1983). The models of

34 Chapter I Introduction

A) Metronidazole or oxidised derivatives (~55-65%)

B) Reduced metabolites (~20%)

Figure 1.14: proposed metabolites of metronidazole and approximate proportions excreted in urine and faeces of mice (Stambaugh et al., 1968), humans (Stambaugh et al., 1968) and rats (Koch et al., 1979, Koch and Goldman, 1979, Ings et al., 1975) or from C. perfringens (Koch et al., 1979) (see references for full compound nomenclature). A portion of metabolised metronidazole was not detected in the excreta from murine models or humans, so the metabolites do not add to 100%. Note that intermediates in brackets (compounds II-IV) are presumptive and have not been isolated. Compounds XI and XII were found conjugated to glucuronide via an ether linkage. The carbons linking the hydroxyl functional group appended to metronidazole (I) and oxamic acid (VI) are marked with an asterisk for clarity.

metronidazole reduction agree, other than the absence of N-acetylethanolamine, nitrite and a possible azo or azoxy dimer, formed by condensation with the nitroso and hydroxylamine derivatives in the electrolytic model. None of the reduced electrolytic products of metronidazole have been shown to be toxic for Clostridium bifermentans, suggesting reduced products are not taken up or

35 Chapter I Introduction that intermediate and not final reduced products mediate toxicity (Knox et al., 1983). As mentioned above the amino derivative (V of Fig 1.14) is not toxic for E. coli or B. fragilis (Ehlhardt et al., 1987).

Conflicting reports exist as to how the ‘intermediate’ reductive products of metronidazole mediate cytotoxicity. Specifically, the capacity of hydroxylamine or nitroso ‘intermediate’ products (III and IV of Fig. 1.14) to exert cytotoxicity by direct DNA damage is contested (Sigeti et al., 1983, Church and Laishley, 1995). Sodium dithionite reduced metronidazole (but not metronidazole) was found to form covalent bonds with guanine and cytosine bases of mammalian, bacterial and phage DNA (Larusso et al., 1977) however, these covalent links cause no detectable damage to the DNA (LaRusso et al., 1978). Other investigators demonstrated the release of radiolabelled thymidine from synthetically radiolabelled DNA, in a model in which several nitroimidazoles were electrolytically reduced, suggesting that reductive intermediates do damage DNA (Knox et al., 1981). Yet another report found only 0.02% of radiolabelled reduced metronidazole associated with DNA from E. coli (Malliaros and Goldman, 1991). In exponentially growing cultures of Clostridium bifermentans, metronidazole inhibited DNA synthesis and degraded existing DNA while not inhibiting RNA synthesis (Plant and Edwards, 1976). In agreement, prior inhibition of protein synthesis by chloramphenicol does not alter the bactericidal rate of Bacteroides spp. subjected to metronidazole (Tally et al., 1978). Indeed, RNA and protein synthesis by B. fragilis, estimated by incorporation of radiolabelled uracil and phenylalanine, continued at linear rates even after 80 min exposure to metronidazole (Sigeti et al., 1983). Using radiolabelled thymidine, DNA synthesis was shown to cease within 20 minutes after addition of metronidazole to B. fragilis; however, no degradation of chromosomal DNA or single/double stranded ‘nicking’ of an endogenous plasmid were detected (Sigeti et al., 1983). In contrast, reduced metronidazole and dimetronidazole were found to cause GC to CG transversions within plasmids in B. fragilis as well as DNA degradation over a similar period of time (Trinh and Reysset, 1998).

Several reports concur that bacteria containing various DNA-repair defects are more susceptible to metronidazole. For instance, E. coli mutants, deficient in recA and lexA, that control the ‘SOS’ DNA repair system, were 10-fold more sensitive to metronidazole than parental strains (Jackson et al., 1984). E. coli uvrB mediates DNA repair by unwinding DNA in unison with ABC exinuclease. E. coli uvrB mutants are also more sensitive to metronidazole (Yeung et al., 1984). Double mutants affecting glutamate-cysteine ligase (gshA) and uvrD/recG/recC are also up to 40-fold more susceptible to metronidazole (Tamae et al., 2008). A AraC/XylS transcriptional regulator involved in DNA repair has been reported to aid survival after metronidazole challenge in B. fragilis (Casanueva et al., 2008). Also, B. fragilis recA DNA repair mutants (Steffens et al., 2010) and RecQ helicase (Paul et al., 2011) are more sensitive to metronidazole. However, the increased sensitivity of

36 Chapter I Introduction

DNA repair mutants does not prove that metronidazole or derivatives directly damage DNA in vivo. An alternative hypothesis is that failure to replicate DNA may be an indirect response following primary drug toxicity.

As discussed, metabolites detected in animal models and from those recovered from C. perfringens, are complementary. However, these metabolites do not account for all of the carbon and nitrogen in metronidazole. It is therefore possible that some of the fragments are liberated, causing cytotoxicity. It is known that hydroxylamine cleaves thioesters (Fishbein and Carbone, 1963, Knox et al., 1993) as shown in Reaction 2 below and hydroxylamine converts acetyl phosphate into acetyl hydroxamate (Chen and Blanchard, 1979). In mammals, hydroxylamine is likely to cleave acetyl-coenzyme A, forming acetohydroxamic acid (Fishbein and Carbone, 1963). Also, hydroxyurea inhibits RNR, the enzyme catalysing the conversion of RNA to DNA, and is relatively stable, although it is susceptible to hydrolysis producing hydroxylamine (Fishbein and Carbone, 1963) (see Reaction 1 below). As mentioned above, azomycin has been shown to inhibit ribonucleotide reductase and associated conversion of ribonucleotide diphosphates (ATP/CTP/GTP/TTP) to corresponding dNTPs (deoxyribonucleotide triphosphates), the monomers required for DNA synthesis and repair (Saeki et al., 1974). In agreement with observations by Sigeti (Sigeti et al., 1983), an explanation for the cytotoxicity of nitroimidazoles is that they fragment, one of the products being hydroxylamine.

Reaction 1: Hydrolysis of hydroxyurea.

Reaction 2: Cleavage of acetyl-Coenzyme A by hydroxylamine.

37 Chapter I Introduction

Released hydroxylamine could then cleave acetyl-CoA, reducing the total ATP pool and/or directly inhibit RNR, limiting DNA synthesis and thereby contributing to cell death. The short-lived nature of reactive intermediates explains some of the discrepancies in the literature and the difficulties in studying these mechanisms.

The ultimate fate of metronidazole depends on whether it is reductively activated or inactivated (summarised in Fig 1.15). If reductively activated, metronidazole competes for reducing equivalents generated primarily by ferredoxin-linked processes, preventing maintenance of redox state (NAD+/NADH ratios), halting cellular functions such as glycolysis. Following reductive activation, transient intermediate compounds are formed and the imidazole ring is fragmented. DNA synthesis and repair of existing DNA ceases (Church et al., 1988, Sigeti et al., 1983). The formation of hydroxylamine or other transient compounds may mediate these processes, for instance by direct inhibition of DNA synthesis or cleavage of thioesters.

A)

B)

Fig 1.15: Summary of metronidazole activation and associated cytotoxicity (A) or reductive inactivation and loss of cytotoxicity (B). Activation of metronidazole by reduction of the nitro group is proposed to fragment the imidazole ring producing transient intermediates such as hydroxylamine that inhibit DNA synthesis and repair. Hydroxylamine is proposed to inhibit DNA synthesis and repair by preventing the RNR (ribonucleotide reductase) catalysed conversion of NTPs (the four ribonucleotide diphosphates ATP/CTP/GTP/TTP) to corresponding dNTPs (deoxyribonucleotide triphosphates).

38 Chapter I Introduction

1.4.5 Metronidazole resistance

Resistance to metronidazole is complex. Resistance can manifest due to a reduced rate of uptake or by efflux. Mutations in effectors activating (reducing) metronidazole could de-sensitise organisms. Reducing the rate of metronidazole activation, for example by re-routing glucose fermentation, is another mechanism of metronidazole resistance. Presence of inactivating resistance determinants and increased DNA repair efficiency provide additional mechanisms (summarised in Fig. 1.16). Notably, several studies have detected considerable decreases in metronidazole sensitivity, although complete loss of sensitivity is rare in nanaerobes, consistent with metronidazole activation being complex (Sindar et al., 1982). Metronidazole resistance was first detected in clinical strains of B. fragilis during the late 1970s (Ingham,1978). Rates of clinical resistance in Bacteroides spp. (Löfmark et al., 2005, Gal and Brazier, 2004, Fernández-Canigia et al., 2012, Sadarangani et al., 2015) and Clostridium spp. (Pelaez et al., 2002, Freeman et al., 2015, Kullin et al., 2016) tend to range from absent to 11.6%, depending on the criteria of methods used to assess resistance and also the geographical region the isolates originate from.

Metronidazole uptake is one parameter involved in resistance to this drug. Generalised defects in metronidazole transport, accompanied by changes in cell wall and by reduced capsular polysaccharide have been described in B. fragilis (Britz and Wilkinson, 1979). E. coli that survived high-concentration metronidazole challenge exhibited an elongated filamentous morphology and colonies of these cells were anaerogenic (did not produce gas from glucose) (Jackson et al., 1984). These colonies were less susceptible to metronidazole and it is possible that the uptake rate of metronidazole was lower than wild type as found by others (Tally et al., 1978). Mutants lacking the ability to reduce nitrate and chlorate, presumably reducing the rate of metronidazole uptake, are also resistant (Yeung et al., 1984). A reduced growth rate is also associated with metronidazole resistance in C. perfringens (Sindar et al., 1982), C. difficile (Lynch et al., 2013) and B. fragilis (Gal and Brazier, 2004). Bacteroides multidrug efflux (bme) pumps also reduce sensitivity to metronidazole (Pumbwe et al., 2006). Deletion of the gene encoding the ferrous transport fusion protein (feoAB) also decreases sensitivity to metronidazole (Veeranagouda et al., 2014).

Changes in type of metabolism or down-regulation of metronidazole activators are also associated with resistance. For example, changes in fermentation account for some of the documented reports of resistance to metronidazole. The reduction of pyruvate ferredoxin oxidoreductase activity and up-regulation of lactate dehydrogenase activity is a strategy common to C. perfringens and Bacteroides spp. (Sindar et al., 1982, Narikawa et al., 1991, Diniz et al., 2004, Patel et al., 2009, Stanko et al., 2016). Also, down-regulation of flavodoxin in B. fragilis decreases

39 Chapter I Introduction sensitivity to metronidazole (Diniz et al., 2004). In B thetaiotaomicron, the up-regulation of rhamnose catabolism regulatory protein (RhaR) results in a similar re-routing and increased resistance to metronidazole (Patel et al., 2009). In genomic screens of metronidazole-sensitive and resistant C. difficile strains, mutations in a putative nitroreductase (nimB), hemN (porphyrin synthesis), thiH (thiamine biosynthesis), glycerol-3- oxidoreductase (glyC) and pyruvate-flavodoxin oxidoreductase (nifJ) (Lynch et al., 2013) were associated with resistance, consistent with a reduced

Glucose Metronidazole Metronidazole

Lactate

+ 2 PEP LDH NAD NADH + H+

2H+ Pyruvate Ferredoxin (reduced) PFOR HYD Ferredoxin (oxidised)

+ Acetyl-CoA +CO2 +2H H2

Acetyl-Phosphate

Acetate

Figure 1.16: Summary of resistance mechanisms to metronidazole and other nitroimidazoles. Red arrows indicate a change in gene expression that confers resistance. Underlined enzymes indicate that loss of function mutants affect sensitivity to metronidazole. Crosses indicate reduced activity or uptake. Enzymes are shown in green include pyruvate:ferredoxin oxidoreductase (‘PFOR’), lactate dehydrogenase (‘LDH’) and hydrogenase (‘HYD’). Metronidazole efflux is facilitated by Bacteroides multidrug efflux pump system (‘bme’). The nim genes, represented by ‘NIM’, are proposed to reductively inactivate the nitro group appended to nitroimidazoles to an amino derivative. The ferrous transport fusion protein is not shown. Fermentations culminating in butyrate, formate, ethanol and butanol, or the dissimilatory sulfate pathway, characteristic of metabolism in some Clostridia or other facultative bacteria, are not shown. Only some of the metabolites of reductively activated metronidazole are shown.

40 Chapter I Introduction rate of electron transport and alterations in intracellular redox. In Helicobacter pylori, thioredoxin reductase, alkyl hydroperoxide reductase and superoxide dismutase are involved in metronidazole resistance (Kaakoush et al., 2009).

Various enzymes inactivate metronidazole, primarily by reductive inactivation of the nitro group to the amino derivative. Nitroreductases consist of oxygen-insensitive NAD(P)H pairing (type I) and oxygen-sensitive (type II) types (Roldán et al., 2008). Type I nitroreductases are the primary enzymes described in metronidazole reductive inactivation, while type II nitroreductases may activate metronidazole, leading to imidazole fragmentation (Fig. 1.17). The biological role of nitroreductases is unclear, although they may function in redox maintenance, by dissipating excess reducing power (Roldán et al., 2008). Early studies of S. faecalis (Nagy and Földes, 1991) and E. coli (Ehlhardt et al., 1987) indicated they were able to inactivate metronidazole. For example, E. coli reduce around 20% of administered metronidazole to the amino derivative (V of Fig. 1.14) (Ehlhardt et al., 1987). The oxygen-insensitive major and minor (nsfA and nfsB respectively) flavoprotein nitroreductases of E. coli, have been shown to reduce nitrofurans and later metronidazole (McCalla et al., 1978, Curado et al., 2008). These nitroreductases, like the nim genes (described below) are dimers, with characteristic α+β folds. Mutants of both nfsA/nfsB in E. coli show reduced sensitivity to metronidazole, suggesting the third poorly characterised nim gene YdjA is involved in metronidazole reductive activation or that another uncharacterised enzyme reduces metronidazole to cytotoxic derivatives.

Figure 1.17: Bacterial reduction of nitroaromatic compounds by oxygen-insensitive (type I) and oxygen-sensitive (type II) nitroreductases (Roldán et al., 2008). In the presence of oxygen, type II nitroreductases are proposed to generate a superoxide anion, regenerating the nitro group.

41 Chapter I Introduction

Interestingly, metronidazole has also been demonstrated to be a substrate for thiaminase (EC 2.5.1.2), an enzyme excreted by Bacillus thiaminolyticus (Alston and Abeles, 1987). Thiaminase from B. thiaminolyticus may function to inactivate metronidazole. Additionally, metabolic products of thiaminase inhibit pyrophosphokinase in the mollusk Venus mercenariathiamine (Alston and Abeles, 1987). More recently, 2-nitroimidazole nitrohydrolase (NhhA, EC 3.5.99.9), which forms imidazole-2- one and nitrite from 2-nitroimidazole (azomycin) was shown to render E. coli resistant to 2- nitroimidazoles (Qu and Spain, 2011). In the limited number of examined nitroimidazoles, metronidazole was not inactivated, although other similar enzyme systems may function to render bacteria resistant to metronidazole.

Some of the nitroimidazole resistance can be ascribed to the ‘nim’ genes, conferring low level resistance to metronidazole (Trinh and Reysset, 1998). The nim genes are best characterised in Bacteroides and other genera of the Bacteroidetes phylum, although they are also encoded by genera of the Clostridia class, Proteobacteria and Archaea (Leiros et al., 2004, Theron et al., 2004, Marchandin et al., 2004). Currently nine nim genes are described in the Bacteroides, including nimA- nimH and nimJ which are chromosomally or plasmid borne (Gal and Brazier, 2004, Husain et al., 2013). A ‘silent’ nim gene, nimI, is encoded by Prevotella spp. (Alauzet et al., 2010). Of 1,502 isolates of the ‘B. fragilis’ group, around 2% were found to carry nim genes (Löfmark et al., 2005). In a study of resistant clinical isolates of Bacteroides, nimA, followed by nimB and nimE were most commonly identified (Gal and Brazier, 2004). Whether chromosomal or on a plasmid, resistance has been shown to be transferable in some Bacteroides spp. (Reysset, 1996) and is transferable by conjugation between Bacteroides spp. and Prevotella spp. (Trinh et al., 1996). In B. fragilis, nimA encodes a 5- nitroimidazole reductase reducing the metronidazole analogue dimetronidazole (1,2-dimethyl-5- nitroimidazole) to the amino derivative, preventing ring fission and associated toxicity (Carlier et al., 1997). Presumably, other nim genes encode 5-nitroimidazole reductases, although the exact mechanism of nim resistance is not known (Husain et al., 2013) and it has not been demonstrated that the ‘nim’ system produced the amino derivative in metronidazole. Structural analysis of NimA from Deinococcus radiodurans (from the Bacteroidetes phylum) indicates the protein is a homodimer, with two binding sites with weak affinity for nitroimidazoles including metronidazole (Leiros et al., 2010). Structural analyses of two nitroreductases from C. difficile indicate they are also homo-dimeric, both of which bind flavin mononucleotide (Wang et al., 2016). Interestingly, the nim genes share structural homology with the gene encoding CREG, a glycoprotein secreted by human embryonal carcinoma cells (Sacher et al., 2005).

The role of nim genes in metronidazole resistance is controversial. It has been established that over-expression of a NimA homologue from B. fragilis induces a three-fold increase in

42 Chapter I Introduction metronidazole resistance in E. coli (Meggersee and Abratt, 2015). Also, overexpression of nimE and nimJ on pMCL140 in B. fragilis induces a four-six fold increase in metronidazole resistance (Husain et al., 2013). However, several non-nim containing Bacteroides spp., Prevotella spp. and C. difficile and other genera within the Clostridia class are resistant to metronidazole or nim-containing isolates within these clades are sensitive to metronidazole (Theron et al., 2004, Gal and Brazier, 2004, Schaumann et al., 2005, Meggersee and Abratt, 2015, Kullin et al., 2016, Galvão et al., 2011, Alauzet et al., 2010, Stanko et al., 2016). Also, expression of the NimA-D proteins does not correlate with metronidazole resistance in B. fragilis (Leitsch et al., 2014). The possibility that insertion sequences upstream of nim genes may control expression is also contested (Löfmark et al., 2005). Lastly, up to a 256-fold increase in resistance to metronidazole can be induced in the absence of the nim genes (Schaumann et al., 2005, Alauzet et al., 2010), suggesting other factors also feature in metronidazole resistance (Schaumann et al., 2005, Alauzet et al., 2010, Husain et al., 2013). An explanation for these conflicting results is that very high levels of the Nim proteins are required to confer metronidazole resistance or that other factors determine metronidazole resistance.

1.5 The rationale underpinning deuteroporphyrin-nitroimidazole adducts and selection of candidates for characterisation

Current broad spectrum antibiotics may eliminate , although they also tend to have considerable off-target action, which may destabilise microbial communities. Secondly, uptake by non-target organisms may reduce availability of the antibiotic for target /s. Lastly, antibiotic use promotes drug resistance in bacteria (Jernberg et al., 2010, Jakobsson et al., 2010). These reasons provided impetus for the design of porphyrin-nitroimidazole adducts, designed to be narrow spectrum antibiotics for P. gingivalis.

The design of nitroimidazole-porphyrin adducts was based on several criteria. P. gingivalis was targeted because it is implicated in periodontitis (elaborated in Section 1.1.1). At the time porphyrin-nitroimidazole adducts were designed, P. gingivalis was considered one of a few bacteria in the human oral cavity that required porphyrin for growth yet was unable to synthesise this molecule. Typically in bacterial growth media hemin is used to satisfy this porphyrin growth requirement. Hemin is the chloride salt of the iron-oxidised form of haem (consisting of protoporphyrin IX with a ferric iron conjugated to a chloride ligand, see Fig. 1.18). P. gingivalis does encode three genes of the porphyrin biosynthetic pathway, a coporphyrinogen oxidase (hemN),

43 Chapter I Introduction protoporphyrinogen oxidase (hemG) and ferrochelatase (hemH) (Nelson et al., 2003, Naito et al., 2008)], see Section 1.3.1 for details of the porphyrin biosynthetic pathway). Ferrochelatase inserts Fe2+ into freebase (metal-free) porphyrins, such as protoporphyrin IX, forming haem b.

A series of growth curves and binding studies demonstrated recognition of porphyrin by P. gingivalis was independent of the central metal ligand and of a protein carrier (DeCarlo et al., 1999, Hunter et al., 2002, Paramaesvaran et al., 2003). These studies also identified some of the structural variations of the porphyrin macrocycle that P. gingivalis recognises, imports and that also stimulate growth (Fig. 1.18). The vinyl groups of protoporphyrin IX (PPIX, at positions 3 and 8 of the macrocycle) are heat labile, may undergo reversible hydration and be photo-oxidised, forming photoprotoporphyrin and isophotoprotoporphyrin (Smith, 1976). Since growth stimulation of P. gingivalis with deuteroporphyrin IX (DPIX) was identical to PPIX (DeCarlo et al., 1999, Hunter et al., 2002, Paramaesvaran et al., 2003), DPIX was chosen due to these instabilities of the vinyl groups of PPIX. Unlike haem b, DPIX does not feature in the porphyrin biosynthetic pathway. As a freebase porphyrin without the vinyl groups normally present on ‘biotic porphyrins’, it was envisaged that other bacteria in the human oral cavity would be less likely to recognise and import DPIX. Metronidazole or nitro- containing derivatives were chosen to link to DPIX, because nanaerobic microorganisms such as the porphyromonads are sensitive to these molecules. Together, it was anticipated that the recognition of DPIX portion of the DPIX-nitroimidazole adduct would facilitate the Trojan-horse style delivery of the toxic nitroimidazole molecule, killing P. gingivalis and few other bacteria during periodontitis in the human oral cavity (Fig. 1.19).

Substituent C2 C3 C7 C8 -H - Y - Y

-CH3 Y Y Y Y

-CHCH2 N Y N Y

-SO3H - Y - Y

-CHOHCH2OH - Y - Y

Figure 1.18: Porphyrin macrocycle numbering, in accord with the international Union of Pure and Applied Chemistry (IUPAC) scheme and tolerated substituents at the ‘northern aspect’ (carbons 2, 3, 7 and 8) of the porphyrin macrocycle. Porphyrins that stimulate or do not stimulate the growth of P. gingivalis are marked ‘Y’ and ‘N’ respectively. Data summarises work from DeCarlo, 1999, Hunter et al., 2003 and Paramaesvaran et al., 2003.

44 Chapter I Introduction

A

P. gingivalis

Metronidazole

B

P. gingivalis

P5

1) Abiotic porphyrins (ie. DPIX) Nitroimidazole sensitive Porphyrin auxotroph recognised and imported 2) Nitroimidazole sensitive

Figure 1.19: The rationale of synthesising deuteroporphyrin-nitroimidazole adducts and expected outcome after administering metronidazole (A) or porphyrin-nitroimidazole adduct ‘P5’ (B) in the human oral cavity during periodontitis. Unlike metronidazole, P5 is expected to kill P. gingivalis in the human oral cavity with minimal ‘by-stander’ damage for other bacteria.

A library of porphyrin adducts, their organic synthesis and preliminary characterisation were documented (Yap et al., 2009, Dingsdag et al., 2015). These porphyrin-nitroimidazole adducts consisted mostly of amide or lysine linkages, with a variety of nitroimidazoles. Porphyrin- nitroimidazole adducts are regioisomers. Separated Isomers of P5 have no difference in toxicity for P. gingivalis (Yap et al., 2009). Two candidates, amide linked (‘P5’) and lysine-linked (‘P8’) are characterised in this thesis (see fig. 1.20 for the structure of these adducts). P5 and P8 were studied because early biological characterisation indicated they had a similar potency to metronidazole and in preliminary assays, P5 was selective for P. gingivalis.

45 Chapter I Introduction

Figure 1.20: The structure of porphyrins, nitroimidazoles and deuterpporphyrin-nitroimidazole adducts used in this thesis. The lysine linkage between metronidazole-amine and deuteroporphyrin of P8 is highlighted in grey and the tert-butyloxycarbonyl group (BOC) is highlighted in green.

46 Chapter I Introduction

1.6 Chapter I references

AAS, J. A., PASTER, B. J., STOKES, L. N., OLSEN, I. & DEWHIRST, F. E. 2005. Defining the normal bacterial flora of the oral cavity. Journal of Clinical Microbiology, 43, 5721-5732. ADAMS, M. E. & POSTGATE, J. 1959. A new sulphate-reducing vibrio. Journal of general microbiology, 20, 252-257. AKUTSU, H., PARK, J. S. & SANO, S. 1993. L-Methionine methyl is specifically incorporated into the C- 2 and C-7 positions of the porphyrin of cytochrome c3 in a strictly anaerobic bacterium, Desulfovibrio vulgaris. Journal of the American Chemical Society, 115, 12185-12186. AL-QUTUB, M. N., BRAHAM, P. H., KARIMI-NASER, L. M., LIU, X. Y., GENCO, C. A. & DARVEAU, R. P. 2006. Hemin-dependent modulation of the lipid a structure of Porphyromonas gingivalis lipopolysaccharide. Infection and Immunity, 74, 4474-4485. ALAUZET, C., MORY, F., TEYSSIER, C., HALLAGE, H., CARLIER, J., GROLLIER, G. & LOZNIEWSKI, A. 2010. Metronidazole resistance in Prevotella spp. and description of a new nim gene in Prevotella baroniae. Antimicrobial agents and chemotherapy, 54, 60-64. ALSTON, T. A. & ABELES, R. H. 1987. Enzymatic conversion of the antibiotic metronidazole to an analog of thiamine. Archives of and , 257, 357-362. AN, D., OH, S. F., OLSZAK, T., NEVES, J. F., AVCI, F. Y., ERTURK-HASDEMIR, D., LU, X., ZEISSIG, S., BLUMBERG, R. S. & KASPER, D. L. 2014. Sphingolipids from a symbiotic microbe regulate homeostasis of host intestinal natural killer T cells. Cell, 156, 123-133. ANDRADE, M. A., CICCARELLI, F. D., PEREZ-IRATXETA, C. & BORK, P. 2002. NEAT: a domain duplicated in genes near the components of a putative Fe3+ siderophore transporter from Gram- positive . Genome Biol, 3, 0047.1-0047.5. ARPAIA, N., CAMPBELL, C., FAN, X., DIKIY, S., VAN DER VEEKEN, J., LIU, H., CROSS, J. R., PFEFFER, K., COFFER, P. J. & RUDENSKY, A. Y. 2013. Metabolites produced by commensal bacteria promote peripheral regulatory T-cell generation. Nature, 504, 451-455. ASANUMA, N. & HINO, T. 2000. Activity and properties of fumarate reductase in ruminal bacteria. The Journal of general and applied microbiology, 46, 119-125. ASQUITH, J., FOSTER, J., WILLSON, R., INGS, R. & MCFADZEAN, J. 1974. Metronidazole (“Flagyl”). A radiosensitizer of hypoxic cells. The British journal of radiology, 47, 474-481. ATARASHI, K., TANOUE, T., SHIMA, T., IMAOKA, A., KUWAHARA, T., MOMOSE, Y., CHENG, G., YAMASAKI, S., SAITO, T. & OHBA, Y. 2011. Induction of colonic regulatory T cells by indigenous Clostridium species. Science, 331, 337-341.

47 Chapter I Introduction

ATUMA, C., STRUGALA, V., ALLEN, A. & HOLM, L. 2001. The adherent gastrointestinal mucus gel layer: thickness and physical state in vivo. American Journal of Physiology-Gastrointestinal and Liver Physiology, 280, G922-G929. BALI, S., LAWRENCE, A. D., LOBO, S. A., SARAIVA, L. M., GOLDING, B. T., PALMER, D. J., HOWARD, M. J., FERGUSON, S. J. & WARREN, M. J. 2011. Molecular hijacking of siroheme for the synthesis of heme and d1 heme. Proceedings of the National Academy of Sciences, 108, 18260-18265. BAUGHN, A. D. & MALAMY, M. H. 2002. A mitochondrial-like aconitase in the bacterium Bacteroides fragilis: implications for the evolution of the mitochondrial Krebs cycle. Proceedings of the National Academy of Sciences, 99, 4662-4667. BAUGHN, A. D. & MALAMY, M. H. 2003. The essential role of fumarate reductase in haem- dependent growth stimulation of Bacteroides fragilis. Microbiology, 149, 1551-1558. BAUGHN, A. D. & MALAMY, M. H. 2004. The strict anaerobe Bacteroides fragilis grows in and benefits from nanomolar concentrations of oxygen. Nature, 427, 441-444. BAUMGART, M., DOGAN, B., RISHNIW, M., WEITZMAN, G., BOSWORTH, B., YANTISS, R., ORSI, R. H., WIEDMANN, M., MCDONOUGH, P. & KIM, S. G. 2007. Culture independent analysis of ileal mucosa reveals a selective increase in invasive Escherichia coli of novel phylogeny relative to depletion of Clostridiales in Crohn's disease involving the ileum. The ISME journal, 1, 403- 418. BEGG, A., SHELDON, P. & FOSTER, J. 1974. Demonstration of radiosensitization of hypoxic cells in solid tumours by metronidazole. The British journal of radiology, 47, 399-404. BENNET, J. & BRINKMAN, M. 1989. Treatment of ulcerative colitis by implantation of normal colonic flora. The Lancet, 333, 164. BEVINS, C. L. & SALZMAN, N. H. 2011. Paneth cells, antimicrobial peptides and maintenance of intestinal homeostasis. Nature Reviews Microbiology, 9, 356-368. BEYER, R. E. 1958. Vitamin K 1, a component of the mitochondrial oxidative phosphorylation system. Biochimica et biophysica acta, 28, 663-664. BIK, E. M., LONG, C. D., ARMITAGE, G. C., LOOMER, P., EMERSON, J., MONGODIN, E. F., NELSON, K. E., GILL, S. R., FRASER-LIGGETT, C. M. & RELMAN, D. A. 2010. Bacterial diversity in the oral cavity of 10 healthy individuals. Isme Journal, 4, 962-974. BLOOM, SETH M., BIJANKI, VINIETH N., NAVA, GERARDO M., SUN, L., MALVIN, NICOLE P., DONERMEYER, DAVID L., DUNNE JR, W. M., ALLEN, PAUL M. & STAPPENBECK, THADDEUS S. 2011. Commensal Bacteroides Species Induce Colitis in Host-Genotype-Specific Fashion in a Mouse Model of Inflammatory Bowel Disease. Cell Host & Microbe, 9, 390-403.

48 Chapter I Introduction

BOLLINGER, R. R., BARBAS, A. S., BUSH, E. L., LIN, S. S. & PARKER, W. 2007. Biofilms in the large bowel suggest an apparent function of the human vermiform appendix. Journal of theoretical biology, 249, 826-831. BRAMANTI, T. E. & HOLT, S. C. 1993. Hemin uptake in Porphyromonas gingivalis: Omp26 is a hemin- binding surface protein. Journal of bacteriology, 175, 7413-7420. BRITZ, M. & WILKINSON, R. 1979. Isolation and properties of metronidazole-resistant mutants of Bacteroides fragilis. Antimicrobial agents and chemotherapy, 16, 19-27. BUCHANAN, B. B. & ARNON, D. I. 1990. A reverse KREBS cycle in photosynthesis: consensus at last. Photosynthesis Research, 24, 47-53. CALDWELL, D., WHITE, D., BRYANT, M. & DOETSCH, R. 1965. Specificity of the heme requirement for growth of Bacteroides ruminicola. Journal of bacteriology, 90, 1645-1654. CALDWELL, D. R. & ARCAND, C. 1974. Inorganic and metal-organic growth requirements of the genus Bacteroides. Journal of bacteriology, 120, 322-333. CANTAREL, B. L., LOMBARD, V. & HENRISSAT, B. 2012. Complex carbohydrate utilization by the healthy human microbiome. PloS one, 7, e28742. CARLIER, J.-P., SELLIER, N., RAGER, M.-N. & REYSSET, G. 1997. Metabolism of a 5-nitroimidazole in susceptible and resistant isogenic strains of Bacteroides fragilis. Antimicrobial agents and chemotherapy, 41, 1495-1499. CARLSTEDT-DUKE, B., MIDTVEDT, T., NORD, C. E. & GUSTAFSSON, B. E. 1986. Isolation and characterization of a mucin-degrading strain of Peptostreptococcus from rat intestinal tract. Acta Pathol Microbiol Immunol Scand B, 94, 293-300. CASANUEVA, A. I., PAUL, L., PATRICK, S. & ABRATT, V. R. 2008. An AraC/XylS family transcriptional regulator homologue from Bacteroides fragilis is associated with cell survival following DNA damage. FEMS microbiology letters, 278, 249-256. CAVALLARO, G., DECARIA, L. & ROSATO, A. 2008. Genome-based analysis of heme biosynthesis and uptake in prokaryotic systems. Journal of proteome research, 7, 4946-4954. CENAS, N., PRAST, S., NIVINSKAS, H., SARLAUSKAS, J. & ARNÉR, E. S. 2006. Interactions of nitroaromatic compounds with the mammalian selenoprotein thioredoxin reductase and the relation to induction of apoptosis in human cancer cells. Journal of biological chemistry, 281, 5593-5603. ČERKASOVOVÁ, A., ČERKASOV, J. & KULDA, J. 1984. Metabolic differences between metronidazole resistant and susceptible strains of Tritrichomonas foetus. Molecular and biochemical , 11, 105-118.

49 Chapter I Introduction

CHABRIÈRE, E., CHARON, M. H., VOLBEDA, A., PIEULLE, L., HATCHIKIAN, E. C. & FONTECILLA–CAMPS, J. C. 1999. Crystal structures of the key anaerobic enzyme pyruvate: ferredoxin oxidoreductase, free and in complex with pyruvate. Nature structural & , 6, 182-190. CHANG, J. Y., ANTONOPOULOS, D. A., KALRA, A., TONELLI, A., KHALIFE, W. T., SCHMIDT, T. M. & YOUNG, V. B. 2008. Decreased diversity of the fecal microbiome in recurrent clostridium difficile—associated diarrhea. Journal of Infectious Diseases, 197, 435-438. CHARON, M.-H., VOLBEDA, A., CHABRIERE, E., PIEULLE, L. & FONTECILLA-CAMPS, J. C. 1999. Structure and electron transfer mechanism of pyruvate: ferredoxin oxidoreductase. Current opinion in , 9, 663-669. CHEN, J.-S. & BLANCHARD, D. K. 1979. A simple hydrogenase-linked assay for ferredoxin and flavodoxin. Analytical biochemistry, 93, 216-222. CHRYSTAL, E., KOCH, R. L., MCLAFFERTY, M. A. & GOLDMAN, P. 1980a. Relationship between metronidazole metabolism and bactericidal activity. Antimicrobial agents and chemotherapy, 18, 566-573. CHRYSTAL, E. J., KOCH, R. L. & GOLDMAN, P. 1980b. Metabolites from the reduction of metronidazole by xanthine oxidase. Molecular , 18, 105-111. CHURCH, D. L. & LAISHLEY, E. J. 1995. Reduction of metronidazole by hydrogenase from Clostridia. Anaerobe, 1, 81-92. CHURCH, D. L., RABIN, H. R. & LALSHLEY, E. J. 1988. Role of hydrogenase 1 of Clostridium pasteurianum in the reduction of metronidazole. Biochemical pharmacology, 37, 1525-1534. COATS, S. R., PHAM, T.-T. T., BAINBRIDGE, B. W., REIFE, R. A. & DARVEAU, R. P. 2005. MD-2 mediates the ability of tetra-acylated and penta-acylated lipopolysaccharides to antagonize Escherichia coli lipopolysaccharide at the TLR4 signaling complex. The Journal of , 175, 4490-4498. COLLINS, M. D., LAWSON, P. A., WILLEMS, A., CORDOBA, J. J., FERNANDEZ-GARAYZABAL, J., GARCIA, P., CAI, J., HIPPE, H. & FARROW, J. A. E. 1994. The Phylogeny of the Genus Clostridium: Proposal of Five New Genera and Eleven New Species Combinations. International Journal of Systematic Bacteriology, 44, 812-826. COYNE, M. J., REINAP, B., LEE, M. M. & COMSTOCK, L. E. 2005. Human symbionts use a host-like pathway for surface fucosylation. Science, 307, 1778-1781. CURADO, S., STAINIER, D. Y. & ANDERSON, R. M. 2008. Nitroreductase-mediated cell/tissue ablation in zebrafish: a spatially and temporally controlled ablation method with applications in developmental and regeneration studies. Nature protocols, 3, 948-954.

50 Chapter I Introduction

DAILEY, H. A., GERDES, S., DAILEY, T. A., BURCH, J. S. & PHILLIPS, J. D. 2015. Noncanonical coproporphyrin-dependent bacterial heme biosynthesis pathway that does not use protoporphyrin. Proceedings of the National Academy of Sciences, 112, 2210-2215. DARFEUILLE-MICHAUD, A., NEUT, C., BARNICH, N., LEDERMAN, E., DI MARTINO, P., DESREUMAUX, P., GAMBIEZ, L., JOLY, B., CORTOT, A. & COLOMBEL, J.-F. 1998. Presence of adherent Escherichia coli strains in ileal mucosa of patients with Crohn's disease. Gastroenterology, 115, 1405-1413. DAS, R. R. 1972. Kinetics of formation and displacement of metalloporphyrin in aqueous solutions. Journal of Inorganic and Nuclear Chemistry, 34, 1263-1269. DASHPER, S. G., ANG, C. S., VEITH, P. D., MITCHELL, H. L., LO, A. W. H., SEERS, C. A., WALSH, K. A., SLAKESKI, N., CHEN, D., LISSEL, J. P., BUTLER, C. A., O'BRIEN-SIMPSON, N. M., BARR, I. G. & REYNOLDS, E. C. 2009. Response of Porphyromonas gingivalis to Heme Limitation in Continuous Culture. Journal of Bacteriology, 191, 1044-1055. DAVIS, C. P., MULCAHY, D., TAKEUCHI, A. & SAVAGE, D. 1972. Location and description of spiral- shaped microorganisms in the normal rat cecum. Infection and immunity, 6, 184-192. DE VRIES, W., VAN WIJCK-KAPTEYN, W. M. & OOSTERHUIS, S. 1974. The presence and function of cytochromes in Selenomonas ruminantium, Anaerovibrio lipolytica and Veillonella alcalescens. Journal of general microbiology, 81, 69-78. DECARLO, A. A., PARAMAESVARAN, M., YUN, P. L. W., COLLYER, C. & HUNTER, N. 1999. Porphyrin- Mediated Binding to Hemoglobin by the HA2 Domain of Cysteine Proteinases (Gingipains) and Hemagglutinins from the Periodontal Pathogen Porphyromonas gingivalis. Journal of Bacteriology, 181, 3784-3791. DEGNAN, P. H., BARRY, N. A., MOK, K. C., TAGA, M. E. & GOODMAN, A. L. 2014. Human Gut Microbes Use Multiple Transporters to Distinguish Vitamin B< sub> 12 Analogs and Compete in the Gut. Cell host & microbe, 15, 47-57. DINGSDAG, S. A., YAP, B. C., HUNTER, N. & CROSSLEY, M. J. 2015. Amino acid-linked porphyrin- nitroimidazole antibiotics targeting Porphyromonas gingivalis. Organic & biomolecular chemistry, 13, 98-109. DINIZ, C. G., FARIAS, L. M., CARVALHO, M. A. R., ROCHA, E. R. & SMITH, C. J. 2004. Differential gene expression in a Bacteroides fragilis metronidazole-resistant mutant. Journal of Antimicrobial Chemotherapy, 54, 100-108. DONOHOE, D. R., GARGE, N., ZHANG, X., SUN, W., O'CONNELL, T. M., BUNGER, M. K. & BULTMAN, S. J. 2011. The microbiome and butyrate regulate energy metabolism and autophagy in the mammalian colon. Cell metabolism, 13, 517-526.

51 Chapter I Introduction

DRAZEK, E. S., HAMMACK, C. A. & SCHMITT, M. P. 2000. Corynebacterium diphtheriae genes required for acquisition of iron from haemin and haemoglobin are homologous to ABC haemin transporters. Molecular microbiology, 36, 68-84. DUCLUZEAU, A.-L., SCHOEPP-COTHENET, B., BAYMANN, F., RUSSELL, M. J. & NITSCHKE, W. 2014. Free energy conversion in the LUCA: quo vadis? Biochimica et Biophysica Acta (BBA)- Bioenergetics, 1837, 982-988. ECKBURG, P. B., BIK, E. M., BERNSTEIN, C. N., PURDOM, E., DETHLEFSEN, L., SARGENT, M., GILL, S. R., NELSON, K. E. & RELMAN, D. A. 2005. Diversity of the Human Intestinal Microbial Flora. Science, 308, 1635-1638. EDWARDS, D. I. & MATHISON, G. E. 1970. MODE OF ACTION OF METRONIDAZOLE AGAINST TRICHOMONAS-VAGINALIS. Journal of General Microbiology, 63, 297-&. EHLHARDT, W. J., BEAULIEU, B. B. & GOLDMAN, P. 1987. Formation of an amino reduction product of metronidazole in bacterial cultures: Lack of bactericidal activity. Biochemical Pharmacology, 36, 259-264. EHLHARDT, W. J., BEAULIEU, B. B. & GOLDMAN, P. 1988. Mammalian cell toxicity and bacterial mutagenicity of nitrosoimidazoles. Biochemical pharmacology, 37, 2603-2606. EISEMAN, B., SILEN, W., BASCOM, G. & KAUVAR, A. 1958. Fecal enema as an adjunct in the treatment of pseudomembranous enterocolitis. Surgery, 44, 854-859. EL KAOUTARI, A., ARMOUGOM, F., GORDON, J. I., RAOULT, D. & HENRISSAT, B. 2013. The abundance and variety of carbohydrate-active enzymes in the human gut microbiota. Nature Reviews Microbiology, 11, 497-504. EVANS, M., BUCHANAN, B. B. & ARNON, D. I. 1966. A new ferredoxin-dependent carbon reduction cycle in a photosynthetic bacterium. Proceedings of the National Academy of Sciences of the United States of America, 55, 928. EVANS, R. HAEMATIN AS A GROWTH FACTOR FOR A STRICT ANAEROBE, FUSIFORMIS- MELANINOGENICUS. JOURNAL OF GENERAL MICROBIOLOGY, 1951. SOC GENERAL MICROBIOLOGY MARLBOROUGH HOUSE, BASINGSTOKE RD, SPENCERS WOODS, READING RG7 1AE, BERKS, ENGLAND, R19-R19. FERNÁNDEZ-CANIGIA, L., LITTERIO, M., LEGARIA, M. C., CASTELLO, L., PREDARI, S. C., DI MARTINO, A., ROSSETTI, A., ROLLET, R., CARLONI, G. & BIANCHINI, H. 2012. First national survey of antibiotic susceptibility of the Bacteroides fragilis group: emerging resistance to carbapenems in Argentina. Antimicrobial agents and chemotherapy, AAC. 05622-11. FISHBEIN, W. N. & CARBONE, P. P. 1963. Hydroxyurea: Mechanism of Action. Science, 142, 1069- 1070.

52 Chapter I Introduction

FRANK, D. N., ST. AMAND, A. L., FELDMAN, R. A., BOEDEKER, E. C., HARPAZ, N. & PACE, N. R. 2007. Molecular-phylogenetic characterization of microbial community imbalances in human inflammatory bowel diseases. Proceedings of the National Academy of Sciences, 104, 13780- 13785. FREEMAN, J., VERNON, J., MORRIS, K., NICHOLSON, S., TODHUNTER, S., LONGSHAW, C. & WILCOX, M. H. 2015. Pan-European longitudinal surveillance of antibiotic resistance among prevalent Clostridium difficile ribotypes. Clinical Microbiology and Infection, 21, 248. e9-248. e16. FRETER, R., BRICKNER, H., BOTNEY, M., CLEVEN, D. & ARANKI, A. 1983a. Mechanisms that control bacterial populations in continuous-flow culture models of mouse large intestinal flora. Infection and immunity, 39, 676-685. FRETER, R., STAUFFER, E., CLEVEN, D., HOLDEMAN, L. & MOORE, W. 1983b. Continuous-flow cultures as in vitro models of the ecology of large intestinal flora. Infection and immunity, 39, 666- 675. GAL, M. & BRAZIER, J. 2004. Metronidazole resistance in Bacteroides spp. carrying nim genes and the selection of slow-growing metronidazole-resistant mutants. Journal of Antimicrobial Chemotherapy, 54, 109-116. GALVÃO, B., MEGGERSEE, R. & ABRATT, V. 2011. Antibiotic resistance and adhesion potential of Bacteroides fragilis clinical isolates from Cape Town, South Africa. Anaerobe, 17, 142-146. GAO, J.-L., NGUYEN, K.-A. & HUNTER, N. 2010. Characterization of a hemophore-like protein from Porphyromonas gingivalis. Journal of Biological Chemistry, 285, 40028-40038. GENCO, C., SIMPSON, W., FORNG, R., EGAL, M. & ODUSANYA, B. 1995. Characterization of a Tn4351- generated hemin uptake mutant of Porphyromonas gingivalis: evidence for the coordinate regulation of virulence factors by hemin. Infection and immunity, 63, 2459-2466. GENCO, C. A. & DIXON, D. W. 2001. Emerging strategies in microbial haem capture. Molecular microbiology, 39, 1-11. GIBBONS, R. J. & MACDONALD, J. B. 1960. HEMIN AND VITAMIN-K COMPOUNDS AS REQUIRED FACTORS FOR THE CULTIVATION OF CERTAIN STRAINS OF BACTEROIDES- MELANINOGENICUS. Journal of Bacteriology, 80, 164-170. GOODMAN, A. L., KALLSTROM, G., FAITH, J. J., REYES, A., MOORE, A., DANTAS, G. & GORDON, J. I. 2011. Extensive personal human gut microbiota culture collections characterized and manipulated in gnotobiotic mice. Proceedings of the National Academy of Sciences, 108, 6252-6257.

53 Chapter I Introduction

GOODMAN, A. L., MCNULTY, N. P., ZHAO, Y., LEIP, D., MITRA, R. D., LOZUPONE, C. A., KNIGHT, R. & GORDON, J. I. 2009. Identifying genetic determinants needed to establish a human gut symbiont in its . Cell Host Microbe, 6, 279-89. GOTTWALD, M., ANDREESEN, J. R., LEGALL, J. & LJUNGDAHL, L. G. 1975. Presence of cytochrome and menaquinone in Clostridium formicoaceticum and Clostridium thermoaceticum. Journal of bacteriology, 122, 325. GRENIER, D., CHAO, G. & MCBRIDE, B. 1989. Characterization of sodium dodecyl sulfate-stable Bacteroides gingivalis proteases by polyacrylamide gel electrophoresis. Infection and immunity, 57, 95-99. GRENIER, D. & MAYRAND, D. 1987. Functional characterization of extracellular vesicles produced by Bacteroides gingivalis. Infection and immunity, 55, 111-117. GRIGG, J. C., VERMEIREN, C. L., HEINRICHS, D. E. & MURPHY, M. E. 2007. Haem recognition by a Staphylococcus aureus NEAT domain. Molecular microbiology, 63, 139-149. GUSTAFSSON, B., MIDTVEDT, T. & STRANDBERG, K. 1969. Effects of microbial contamination on the cecum enlargement of germfree rats. Scandinavian journal of gastroenterology, 5, 309-314. HAAG, L.-M., FISCHER, A., OTTO, B., PLICKERT, R., KUHL, A. A., GOBEL, U. B., BERESWILL, S. & HEIMESAAT, M. M. 2012. Intestinal microbiota shifts towards elevated commensal Escherichia coli loads abrogate colonization resistance against Campylobacter jejuni in mice. PloS one, 7, e35988. HAJISHENGALLIS, G., LIANG, S., PAYNE, M. A., HASHIM, A., JOTWANI, R., ESKAN, M. A., MCINTOSH, M. L., ALSAM, A., KIRKWOOD, K. L., LAMBRIS, J. D., DARVEAU, R. P. & CURTIS, M. A. 2011. Low-Abundance Biofilm Species Orchestrates Inflammatory Periodontal Disease through the Commensal Microbiota and Complement. Cell Host & Microbe, 10, 497-506. HALLER, I. 1982. In vitro activity of the two principal oxidative metabolites of metronidazole against Bacteroides fragilis and related species. Antimicrobial agents and chemotherapy, 22, 165- 166. HAYWARD, H. & STADTMAN, T. 1959. ANAEROBIC DEGRADATION OF CHOLINE I., Vibrio cholinicusn. sp: Fermentation of Choline by an Anaerobic, Cytochrome-Producing Bacterium1. J Bacteriol, 78, 557-561. HICKEY, C. A., KUHN, K. A., DONERMEYER, D. L., PORTER, N. T., JIN, C., CAMERON, E. A., JUNG, H., KAIKO, G. E., WEGORZEWSKA, M. & MALVIN, N. P. 2015. Colitogenic Bacteroides thetaiotaomicron Antigens Access Host Immune Cells in a Sulfatase-Dependent Manner via Outer Membrane Vesicles. Cell host & microbe, 17, 672-680.

54 Chapter I Introduction

HIGGS, P. I., LARSEN, R. A. & POSTLE, K. 2002. Quantification of known components of the Escherichia coli TonB energy transduction system: TonB, ExbB, ExbD and FepA. Molecular microbiology, 44, 271-281. HODGSON, G. & BAKER, B. 1967. Porphyrin from pyrrole and formaldehyde under simulated geochemical conditions. Nature, 216, 29-32. HOOPER, L. V. & GORDON, J. I. 2001. Commensal host-bacterial relationships in the gut. Science, 292, 1115-1118. HOPPE-SEYLER, F. 1886. Ueber Gährung der Cellulose mit Bildung von Methan und Kohlensäure. Zeitschrift für physiologische Chemie, 10, 201-217. HRDÝ, I., CAMMACK, R., STOPKA, P., KULDA, J. & TACHEZY, J. 2005. Alternative Pathway of Metronidazole Activation in Trichomonas vaginalis Hydrogenosomes. Antimicrobial Agents and Chemotherapy, 49, 5033-5036. HUNTER, N., NGUYEN, K. A., MCDONALD, J. A., QUINN, M. J., LANGLEY, D. B., CROSSLEY, M. J. & COLLYER, C. A. 2002. Structural requirements for recognition of essential porphyrin by Porphyromonas gingivalis. Journal of Porphyrins and Phthalocyanines, 6, 774-782. HUSAIN, F., VEERANAGOUDA, Y., HSI, J., MEGGERSEE, R., ABRATT, V. & WEXLER, H. M. 2013. Two multidrug-resistant clinical isolates of Bacteroides fragilis carry a novel metronidazole resistance nim gene (nimJ). Antimicrobial agents and chemotherapy, 57, 3767-3774. INGHAM, H., HALL, C., SISSON, P. R., THARAGONNET, D. & SELKON, J. 1980. The activity of metronidazole against facultatively anaerobic bacteria. Journal of Antimicrobial Chemotherapy, 6, 343-347. INGS, R., MCFADZEAN, J. & ORMEROD, W. 1975. The fate of metronidazole and its implications in chemotherapy. Xenobiotica, 5, 223-235. INGS, R. M. J., LAW, G. L. & PARNELL, E. W. 1966. METABOLISM OF METRONIDAZOLE (1-2'- HYDROXYETHYL-2-METHYL-5-NITROIMIDAZOLE). Biochemical Pharmacology, 15, 515-&. ISHIDA, T., YU, L., AKUTSU, H., OZAWA, K., KAWANISHI, S., SETO, A., INUBUSHI, T. & SANO, S. 1998. A primitive pathway of porphyrin biosynthesis and enzymology in Desulfovibrio vulgaris. Proceedings of the National Academy of Sciences, 95, 4853-4858. IVANOV, I. I., ATARASHI, K., MANEL, N., BRODIE, E. L., SHIMA, T., KARAOZ, U., WEI, D., GOLDFARB, K. C., SANTEE, C. A. & LYNCH, S. V. 2009. Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell, 139, 485-498. IZADI, N., HENRY, Y., HALADJIAN, J., GOLDBERG, M. E., WANDERSMAN, C., DELEPIERRE, M. & LECROISEY, A. 1997. Purification and characterization of an extracellular heme-binding protein, HasA, involved in heme iron acquisition. Biochemistry, 36, 7050-7057.

55 Chapter I Introduction

JACKSON, D., SALEM, A. & COOMBS, G. 1984. The in-vitro activity of metronidazole against strains of Escherichia coli with impaired DNA repair systems. Journal of Antimicrobial Chemotherapy, 13, 227-236. JAKOBSSON, H. E., JERNBERG, C., ANDERSSON, A. F., SJOLUND-KARLSSON, M., JANSSON, J. K. & ENGSTRAND, L. 2010. Short-Term Antibiotic Treatment Has Differing Long-Term Impacts on the Human Throat and Gut Microbiome. Plos One, 5. JERNBERG, C., LÖFMARK, S., EDLUND, C. & JANSSON, J. K. 2010. Long-term impacts of antibiotic exposure on the human intestinal microbiota. Microbiology, 156, 3216-3223. JOHANSSON, M. E., AMBORT, D., PELASEYED, T., SCHÜTTE, A., GUSTAFSSON, J. K., ERMUND, A., SUBRAMANI, D. B., HOLMÉN-LARSSON, J. M., THOMSSON, K. A. & BERGSTRÖM, J. H. 2011. Composition and functional role of the mucus layers in the intestine. Cellular and molecular life sciences, 68, 3635-3641. JOHANSSON, M. E., PHILLIPSON, M., PETERSSON, J., VELCICH, A., HOLM, L. & HANSSON, G. C. 2008. The inner of the two Muc2 mucin-dependent mucus layers in colon is devoid of bacteria. Proceedings of the national academy of sciences, 105, 15064-15069. JOSEFOWICZ, S. Z., NIEC, R. E., KIM, H. Y., TREUTING, P., CHINEN, T., ZHENG, Y., UMETSU, D. T. & RUDENSKY, A. Y. 2012. Extrathymically generated regulatory T cells control mucosal TH2 inflammation. Nature, 482, 395-399. KAAKOUSH, N. O., ASENCIO, C., MEGRAUD, F. & MENDZ, G. L. 2009. A Redox Basis for Metronidazole Resistance in Helicobacter pylori. Antimicrobial Agents and Chemotherapy, 53, 1884-1891. KANG, S. S., BLOOM, S. M., NORIAN, L. A., GESKE, M. J., FLAVELL, R. A., STAPPENBECK, T. S. & ALLEN, P. M. 2008. An antibiotic-responsive mouse model of fulminant ulcerative colitis. PLoS Med, 5, e41. KASTER, A.-K., MOLL, J., PAREY, K. & THAUER, R. K. 2011. Coupling of ferredoxin and heterodisulfide reduction via electron bifurcation in hydrogenotrophic methanogenic archaea. Proceedings of the National Academy of Sciences, 108, 2981-2986. KAWAI, S., KOBAYASH.K, OSHIMA, T. & EGAMI, F. 1965. STUDIES ON OXIDATION OF P- AMINOBENZOATE TO P-NITROBENZOATE BY STREPTOMYCES THIOLUTEUS. Archives of Biochemistry and Biophysics, 112, 537-&. KEILIN, D. 1925. On cytochrome, a respiratory pigment, common to animals, yeast, and higher plants. Proceedings of the Royal Society of London. Series B, Containing Papers of a Biological Character, 312-339. KERNBAUER, E., DING, Y. & CADWELL, K. 2014. An enteric virus can replace the beneficial function of commensal bacteria. Nature, 516, 94-98.

56 Chapter I Introduction

KLAASEN, H., KOOPMAN, J., VAN DEN BRINK, M., BAKKER, M., POELMA, F. & BEYNEN, A. 1993. Intestinal, segmented, filamentous bacteria in a wide range of vertebrate species. Laboratory animals, 27, 141-150. KNOX, R. J., FRIEDLOS, F. & BOLAND, M. P. 1993. The bioactivation of CB 1954 and its use as a prodrug in antibody-directed enzyme prodrug therapy (ADEPT). Cancer and Metastasis Reviews, 12, 195-212. KNOX, R. J., KNIGHT, R. C. & EDWARDS, D. I. 1981. INTERACTION OF NITROIMIDAZOLE DRUGS WITH DNA INVITRO - STRUCTURE-ACTIVITY-RELATIONSHIPS. British Journal of Cancer, 44, 741-745. KNOX, R. J., KNIGHT, R. C. & EDWARDS, D. I. 1983. Studies on the action of nitroimidazole drugs. Biochemical Pharmacology, 32, 2149-2156. KOCH, R. L., CHRYSTAL, E. J., BEAULIEU, B. B. & GOLDMAN, P. 1979. Acetamide—a metabolite of metronidazole formed by the intestinal flora. Biochemical pharmacology, 28, 3611-3615. KOCH, R. L. & GOLDMAN, P. 1979. The anaerobic metabolism of metronidazole forms N-(2- hydroxyethyl)-oxamic acid. Journal of Pharmacology and Experimental Therapeutics, 208, 406-410. KOENIG, J. E., SPOR, A., SCALFONE, N., FRICKER, A. D., STOMBAUGH, J., KNIGHT, R., ANGENENT, L. T. & LEY, R. E. 2011. Succession of microbial consortia in the developing infant gut microbiome. Proceedings of the National Academy of Sciences, 108, 4578-4585. KOTARSKI, S. F. & SALYERS, A. A. 1984. Isolation and characterization of outer membranes of Bacteroides thetaiotaomicron grown on different carbohydrates. Journal of bacteriology, 158, 102-109. KREBS, C., MATTHEWS, M. L., JIANG, W. & BOLLINGER, J. M. 2007. AurF from Streptomyces thioluteus and a possible new family of manganese/iron oxygenases. Biochemistry, 46, 10413-10418. KREBS, H. & JOHNSON, W. 1937. The role of citric acid in intermediate metabolism in animal tissues. Enzymologia, 4, 148-156. KRÖGER, A., GEISLER, V., LEMMA, E., THEIS, F. & LENGER, R. 1992. Bacterial fumarate respiration. Archives of microbiology, 158, 311-314. KULLIN, B., BROCK, T., RAJABALLY, N., ANWAR, F., VEDANTAM, G., REID, S. & ABRATT, V. 2016. Characterisation of Clostridium difficile strains isolated from Groote Schuur Hospital, Cape Town, South Africa. European Journal of Clinical Microbiology & Infectious Diseases, 35, 1709-1718. LAISHLEY, E. & KROUSE, H. 1978. Stable isotope fractionation by Clostridium pasteurianum. 2. Regulation of sulfite reductases by sulfur amino acids and their influence on sulfur isotope

57 Chapter I Introduction

fractionation during SO32-and SO42-reduction. Canadian journal of microbiology, 24, 716- 724. LANCINI, C., LAZZARI, E. & SARTORI, C. 1968. Microbial oxidation of aminoimidazoles to nitroimidazoles. The Journal of antibiotics, 21, 387-392. LANCINI, G., KLUEPFEL, D., LAZZARI, E. & SARTORI, G. 1966. Origin of the nitro group of azomycin. Biochimica et Biophysica Acta (BBA)-General Subjects, 130, 37-44. LARUSSO, N. F., TOMASZ, M., KAPLAN, D. & MÜLLER, M. 1978. Absence of Strand Breaks in Deoxyribonucleic Acid Treated with Metronidazole. Antimicrobial Agents and Chemotherapy, 13, 19-24. LARUSSO, N. F., TOMASZ, M., MULLER, M. & LIPMAN, R. 1977. INTERACTION OF METRONIDAZOLE WITH NUCLEIC-ACIDS INVITRO. Molecular Pharmacology, 13, 872-882. LATHROP, S. K., BLOOM, S. M., RAO, S. M., NUTSCH, K., LIO, C.-W., SANTACRUZ, N., PETERSON, D. A., STAPPENBECK, T. S. & HSIEH, C.-S. 2011. Peripheral education of the immune system by colonic commensal microbiota. Nature, 478, 250-254. LAWLEY, T. D., CLARE, S., WALKER, A. W., STARES, M. D., CONNOR, T. R., RAISEN, C., GOULDING, D., RAD, R., SCHREIBER, F., BRANDT, C., DEAKIN, L. J., PICKARD, D. J., DUNCAN, S. H., FLINT, H. J., CLARK, T. G., 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. LEE, S. M., DONALDSON, G. P., MIKULSKI, Z., BOYAJIAN, S., LEY, K. & MAZMANIAN, S. K. 2013. Bacterial colonization factors control specificity and stability of the gut microbiota. Nature, 501, 426-429. LEIROS, H.-K. S., BRANDSDAL, B. O. & MCSWEENEY, S. M. 2010. Biophysical characterization and mutational analysis of the antibiotic resistance protein NimA from Deinococcus radiodurans. Biochimica et Biophysica Acta (BBA)-Proteins and Proteomics, 1804, 967-976. LEIROS, H.-K. S., KOZIELSKI-STUHRMANN, S., KAPP, U., TERRADOT, L., LEONARD, G. A. & MCSWEENEY, S. M. 2004. Structural Basis of 5-Nitroimidazole Antibiotic Resistance THE CRYSTAL STRUCTURE OF NimA FROM DEINOCOCCUS RADIODURANS. Journal of Biological Chemistry, 279, 55840-55849. LEITSCH, D., SÓKI, J., KOLARICH, D., URBÁN, E. & NAGY, E. 2014. A study on Nim expression in Bacteroides fragilis. Microbiology, 160, 616-622. LEV, M., KEUDELL, K. & MILFORD, A. 1971. Succinate as a growth factor for Bacteroides melaninogenicus. Journal of bacteriology, 108, 175-178.

58 Chapter I Introduction

LEY, R. E., HAMADY, M., LOZUPONE, C., TURNBAUGH, P. J., RAMEY, R. R., BIRCHER, J. S., SCHLEGEL, M. L., TUCKER, T. A., SCHRENZEL, M. D., KNIGHT, R. & GORDON, J. I. 2008. Evolution of Mammals and Their Gut Microbes. Science, 320, 1647-1651. LINDMARK, D. G. & MÜLLER, M. 1976. Antitrichomonad action, mutagenicity, and reduction of metronidazole and other nitroimidazoles. Antimicrobial agents and chemotherapy, 10, 476- 482. LOCKERBY, D. L., RABIN, H. R., BRYAN, L. & LAISHLEY, E. 1984. Ferredoxin-linked reduction of metronidazole in Clostridium pasteurianum. Antimicrobial agents and chemotherapy, 26, 665-669. LÖFMARK, S., FANG, H., HEDBERG, M. & EDLUND, C. 2005. Inducible metronidazole resistance and nim genes in clinical Bacteroides fragilis group isolates. Antimicrobial agents and chemotherapy, 49, 1253-1256. LOUIS, P. & FLINT, H. J. 2009. Diversity, metabolism and microbial ecology of butyrate-producing bacteria from the human large intestine. FEMS microbiology letters, 294, 1-8. LUBBE, M. M., STANLEY, K. & CHALKLEY, L. J. 1999. Prevalence of nim genes in anaerobic/facultative anaerobic bacteria isolated in South Africa. FEMS microbiology letters, 172, 79-83. LUO, Q., KUMAR, P., VICKERS, T. J., SHEIKH, A., LEWIS, W. G., RASKO, D. A., SISTRUNK, J. & FLECKENSTEIN, J. M. 2014. Enterotoxigenic Escherichia coli secretes a highly conserved mucin-degrading metalloprotease to effectively engage intestinal epithelial cells. Infection and immunity, 82, 509-521. LYNCH, T., CHONG, P., ZHANG, J., HIZON, R., DU, T., GRAHAM, M. R., BENIAC, D. R., BOOTH, T. F., KIBSEY, P. & MILLER, M. 2013. Characterization of a stable, metronidazole-resistant Clostridium difficile clinical isolate. PLoS One, 8, e53757. MACFARLANE, S. & MACFARLANE, G. T. 2003. Regulation of short-chain fatty acid production. Proceedings of the Society, 62, 67-72. MACY, J., PROBST, I. & GOTTSCHALK, G. 1975. Evidence for cytochrome involvement in fumarate reduction and adenosine 5'-triphosphate synthesis by Bacteroides fragilis grown in the presence of hemin. Journal of bacteriology, 123, 436-442. MACY, J. M., LJUNGDAHL, L. G. & GOTTSCHALK, G. 1978. Pathway of succinate and propionate formation in Bacteroides fragilis. Journal of Bacteriology, 134, 84-91. MACY, J. M. & PROBST, I. 1979. The biology of gastrointestinal bacteroides. Annual Reviews in Microbiology, 33, 561-594. MAEDA, K., OSATO, T. & UMEZAWA, H. 1953. A new antibiotic, azomycin. The Journal of antibiotics, 6, 182.

59 Chapter I Introduction

MAHOWALD, M. A., REY, F. E., SEEDORF, H., TURNBAUGH, P. J., FULTON, R. S., WOLLAM, A., SHAH, N., WANG, C., MAGRINI, V., WILSON, R. K., CANTAREL, B. L., COUTINHO, P. M., HENRISSAT, B., CROCK, L. W., RUSSELL, A., VERBERKMOES, N. C., HETTICH, R. L. & GORDON, J. I. 2009. Characterizing a model human gut microbiota composed of members of its two dominant bacterial phyla. Proceedings of the National Academy of Sciences, 106, 5859-5864. MALLIAROS, D. P. & GOLDMAN, P. 1991. Interaction of metronidazole with Escherichia coli deoxyribonucleic acid. Biochemical pharmacology, 42, 1739-1744. MANFREDI, P., LAUBER, F., RENZI, F., HACK, K., HESS, E. & CORNELIS, G. R. 2015. New iron acquisition system in Bacteroidetes. Infection and immunity, 83, 300-310. MANTHEI, R., FEO, L. & STAMBAUGH, J. 1969. Identification of the metabolites of metronidazole in the human vagina. Wiadomości parazytologiczne, 15, 403. MARCHANDIN, H., JEAN-PIERRE, H., CAMPOS, J., DUBREUIL, L., TEYSSIER, C. & JUMAS-BILAK, E. 2004. nimE gene in a metronidazole-susceptible Veillonella sp. strain. Antimicrobial agents and chemotherapy, 48, 3207-3208. MARTENS, E. C., CHIANG, H. C. & GORDON, J. I. 2008. Mucosal glycan foraging enhances fitness and transmission of a saccharolytic human gut bacterial symbiont. Cell host & microbe, 4, 447- 457. MAUZERALL, D. 1992. Light, iron, Sam Granick and the origin of life. Photosynthesis research, 33, 163-170. MAYHEW, S. G. 1978. The redox potential of dithionite and SO− 2 from equilibrium reactions with flavodoxins, methyl viologen and hydrogen plus hydrogenase. European Journal of Biochemistry, 85, 535-547. MCCALLA, D., KAISER, C. & GREEN, M. 1978. of nitrofurazone resistance in Escherichia coli. Journal of bacteriology, 133, 10-16. MCKEE, A., MCDERMID, A., BASKERVILLE, A., DOWSETT, A., ELLWOOD, D. & MARSH, P. 1986. Effect of hemin on the physiology and virulence of Bacteroides gingivalis W50. Infection and Immunity, 52, 349-355. MEGGERSEE, R. & ABRATT, V. 2015. The occurrence of antibiotic resistance genes in drug resistant Bacteroides fragilis isolates from Groote Schuur Hospital, South Africa. Anaerobe, 32, 1-6. MEURIC, V., ROUILLON, A., CHANDAD, F. & BONNAURE-MALLET, M. 2010. Putative respiratory chain of Porphyromonas gingivalis. Future microbiology, 5, 717-734. MOHINDRA, J. K. & RAUTH, A. M. 1976. Increased cell killing by metronidazole and nitrofurazone of hypoxic compared to aerobic mammalian cells. Cancer research, 36, 930-936.

60 Chapter I Introduction

MOORE, R. A., BECKTHOLD, B. & BRYAN, L. E. 1995. METRONIDAZOLE UPTAKE IN HELICOBACTER- PYLORI. Canadian Journal of Microbiology, 41, 746-749. MOORE, W. & HOLDEMAN, L. V. 1972. Identification of anaerobic bacteria. The American journal of clinical nutrition, 25, 1306-1313. MORRIS, R. L. & SCHMIDT, T. M. 2013. Shallow breathing: bacterial life at low O2. Nature Reviews Microbiology, 11, 205-212. MÜLLER, M. 1983. Mode of action of metronidazole on anaerobic bacteria and protozoa. Surgery, 93, 165-171. MÜLLER, M. 1986. Reductive activation of nitroimidazoles in anaerobic microorganisms. Biochemical pharmacology, 35, 37-41. MUNN, C. A. M. 1886. Researches on Myohaematin and the Histohaematins. Philosophical Transactions of the Royal Society of London, 177, 267-298. NAGY, E. & FÖLDES, J. 1991. Inactivatin of metronidazole by . Journal of antimicrobial chemotherapy, 27, 63-70. NAITO, M., HIRAKAWA, H., YAMASHITA, A., OHARA, N., SHOJI, M., YUKITAKE, H., NAKAYAMA, K., TOH, H., YOSHIMURA, F., KUHARA, S., HATTORI, M. & HAYASHI, T. 2008. Determination of the Genome Sequence of Porphyromonas gingivalis Strain ATCC 33277 and Genomic Comparison with Strain W83 Revealed Extensive Genome Rearrangements in P-gingivalis. DNA Research, 15, 215-225. NAKANE, A., NAKAMURA, T. & EGUCHI, Y. 1977. A novel metabolic fate of arginine in Streptomyces eurocidicus. Partial resolution of the pathway and identification of an intermediate. Journal of Biological Chemistry, 252, 5267-5273. NAKAYAMA, T., YONEKURA, S.-I., YONEI, S. & ZHANG-AKIYAMA, Q.-M. 2013. Escherichia coli pyruvate: flavodoxin oxidoreductase, YdbK-regulation of expression and biological roles in protection against oxidative stress. Genes & genetic systems, 88, 175-188. NAKJANG, S., NDEH, D. A., WIPAT, A., BOLAM, D. N. & HIRT, R. P. 2012. A novel extracellular metallopeptidase domain shared by animal host-associated mutualistic and pathogenic microbes. PLoS One, 7, e30287. NARIKAWA, S. 1986. DISTRIBUTION OF METRONIDAZOLE SUSCEPTIBILITY FACTORS IN OBLIGATE ANAEROBES. Journal of Antimicrobial Chemotherapy, 18, 565-574. NARIKAWA, S., SUZUKI, T., YAMAMOTO, M. & NAKAMURA, M. 1991. Lactate dehydrogenase activity as a cause of metronidazole resistance in Bacteroides fragilis NCTC 11295. Journal of Antimicrobial Chemotherapy, 28, 47-53.

61 Chapter I Introduction

NAVA, G. M., FRIEDRICHSEN, H. J. & STAPPENBECK, T. S. 2011. Spatial organization of intestinal microbiota in the mouse ascending colon. The ISME journal, 5, 627-638. NAVA, G. M. & STAPPENBECK, T. S. 2011. Diversity of the autochthonous colonic microbiota. Gut microbes, 2, 99-104. NELSON, K. E., FLEISCHMANN, R. D., DEBOY, R. T., PAULSEN, I. T., FOUTS, D. E., EISEN, J. A., DAUGHERTY, S. C., DODSON, R. J., DURKIN, A. S., GWINN, M., HAFT, D. H., KOLONAY, J. F., NELSON, W. C., MASON, T., TALLON, L., GRAY, J., GRANGER, D., TETTELIN, H., DONG, H., GALVIN, J. L., DUNCAN, M. J., DEWHIRST, F. E. & FRASER, C. M. 2003. Complete genome sequence of the oral pathogenic bacterium Porphyromonas gingivalis strain W83. Journal of Bacteriology, 185, 5591-5601. NILSSON-EHLE, I., URSING, B. & NILSSON-EHLE, P. 1981. Liquid chromatographic assay for metronidazole and tinidazole: pharmacokinetic and metabolic studies in human subjects. Antimicrobial agents and chemotherapy, 19, 754-760. NSEKA, K. & MÜLLER, M. 1978. [Effect of glycolysis inhibitors on the uptake of metronidazole by the protozoa Tritrichomonas foetus and Entamoeba invadens]. Comptes rendus des seances de la Societe de biologie et de ses filiales, 172, 1094-1098. O'BRIEN, R. & MORRIS, J. 1971. The ferredoxin-dependent reduction of chloramphenicol by Clostridium acetobutylicum. Journal of general microbiology, 67, 265-271. O'KEEFE, J. P., TROC, K. A. & THOMPSON, K. D. 1982. Activity of metronidazole and its hydroxy and acid metabolites against clinical isolates of anaerobic bacteria. Antimicrobial agents and chemotherapy, 22, 426-430. OBRIEN, R. W. & MORRIS, J. G. 1972. EFFECT OF METRONIDAZOLE ON HYDROGEN PRODUCTION BY CLOSTRIDIUM-ACETOBUTYLICUM. Archiv Fur Mikrobiologie, 84, 225-&. OKAMI, Y., MAEDA, K. & UMEZAWA, H. 1954. Studies on antibiotic actinomycetes. VII. Azomycin- producing strain resembling to Nocardia mesenterica. The Journal of antibiotics, 7, 53. OLCZAK, T., SROKA, A., POTEMPA, J. & OLCZAK, M. 2008. Porphyromonas gingivalis HmuY and HmuR: further characterization of a novel mechanism of heme utilization. Archives of microbiology, 189, 197-210. OSATO, T., UEDA, M., FUKUYAMA, S., YAGISHITA, K., OKAMI, Y. & UMEZAWA, H. 1955. Production of tertiomycin (a new antibiotic substance), azomycin and eurocidin by S. eurocidicus. The Journal of antibiotics, 8, 105. OTTO, B., SPARRIUS, M., WORS, D., DE GRAAF, F. & MACLAREN, D. 1994. Utilization of haem from the haptoglobin-haemoglobin complex by Bacteroides fragilis. Microbial pathogenesis, 17, 137-147.

62 Chapter I Introduction

OTTO, B., VERWEIJ, W., SPARRIUS, M., VERWEIJ-VAN VUGHT, A., NORD, C. & MACLAREN, D. 1991. Human immune response to an iron-repressible outer membrane protein of Bacteroides fragilis. Infection and immunity, 59, 2999-3003. PANEK, H. & O'BRIAN, M. R. 2002. A whole genome view of prokaryotic haem biosynthesis. Microbiology-Sgm, 148, 2273-2282. PARAMAESVARAN, M., NGUYEN, K. A., CALDON, E., MCDONALD, J. A., NAJDI, S., GONZAGA, G., LANGLEY, D. B., DECARLO, A., CROSSLEY, M. J., HUNTER, N. & COLLYER, C. A. 2003. Porphyrin-mediated cell surface heme capture from hemoglobin by Porphyromonas gingivalis. Journal of Bacteriology, 185, 2528-2537. PATEL, E. H., PAUL, L. V., CASANUEVA, A. I., PATRICK, S. & ABRATT, V. R. 2009. Overexpression of the rhamnose catabolism regulatory protein, RhaR: a novel mechanism for metronidazole resistance in Bacteroides thetaiotaomicron. Journal of antimicrobial chemotherapy, dkp203. PAUL, L., PATRICK, S., NORD, C. E. & ABRATT, V. 2011. The role of Bacteroides fragilis RecQ DNA helicases in cell survival after metronidazole exposure. FEMS microbiology letters, 319, 125- 132. PÉDRON, T., MULET, C., DAUGA, C., FRANGEUL, L., CHERVAUX, C., GROMPONE, G. & SANSONETTI, P. J. 2012. A crypt-specific core microbiota resides in the mouse colon. MBio, 3, e00116-12. PELAEZ, T., ALCALA, L., ALONSO, R., RODRIGUEZ-CREIXEMS, M., GARCIA-LECHUZ, J. & BOUZA, E. 2002. Reassessment of Clostridium difficile susceptibility to metronidazole and vancomycin. Antimicrobial Agents and Chemotherapy, 46, 1647-1650. PETERSON, F. J., MASON, R. P., HOVSEPIAN, J. & HOLTZMAN, J. L. 1979. Oxygen-sensitive and- insensitive nitroreduction by Escherichia coli and rat hepatic microsomes. Journal of Biological Chemistry, 254, 4009-4014. PLACIDI, G., MASUOKA, D., ALCARAZ, A., TAYLOR, J. & EARLE, R. 1970. Distribution and metabolism of 14C-metronidazole in mice. Archives Internationales de Pharmacodynamie et de Therapie, 188, 168-179. PLANT, C. & EDWARDS, D. 1976. The effect of tinidazole, metronidazole and nitrofurazone on nucleic acid synthesis in Clostridium bifermentans. Journal of Antimicrobial Chemotherapy, 2, 203- 209. PLUYM, M., MURYOI, N., HEINRICHS, D. E. & STILLMAN, M. J. 2008. Heme binding in the NEAT domains of IsdA and IsdC of Staphylococcus aureus. Journal of inorganic biochemistry, 102, 480-488. POSTGATE, J. 1954. Dependence of sulphate reduction and oxygen utilization on a cytochrome in Desulphovibrio. The Biochemical journal, 58, ix.

63 Chapter I Introduction

PUMBWE, L., GLASS, D. & WEXLER, H. M. 2006. Efflux pump overexpression in multiple-antibiotic- resistant mutants of Bacteroides fragilis. Antimicrobial Agents and Chemotherapy, 50, 3150- 3153. QU, Y. & SPAIN, J. C. 2011. Catabolic pathway for 2‐nitroimidazole involves a novel nitrohydrolase that also confers drug resistance. Environmental microbiology, 13, 1010-1017. RALPH, E. D., CLARKE, J. T., LIBKE, R. D., LUTHY, R. P. & KIRBY, W. M. 1974. Pharmacokinetics of metronidazole as determined by bioassay. Antimicrobial agents and chemotherapy, 6, 691- 696. RALPH, E. D. & KIRBY, W. M. 1975. Bioassay of metronidazole with either anaerobic or aerobic incubation. Journal of Infectious Diseases, 132, 587-591. REDDY, C. A. & BRYANT, M. P. 1967. DNA base composition and cytochromes in certain species of the genus Bacteroides. Bacteriological Proceedings, 40. REDDY, C. A. & PECK, H. 1978. Electron transport phosphorylation coupled to fumarate reduction by H2-and Mg2+-dependent adenosine triphosphatase activity in extracts of the rumen anaerobe Vibrio succinogenes. Journal of bacteriology, 134, 982-991. REEVES, A. E., KOENIGSKNECHT, M. J., BERGIN, I. L. & YOUNG, V. B. 2012. Suppression of Clostridium difficile in the Gastrointestinal Tracts of Germfree Mice Inoculated with a Murine Isolate from the Family Lachnospiraceae. Infection and Immunity, 80, 3786-3794. REIFE, R. A., COATS, S. R., AL‐QUTUB, M., DIXON, D. M., BRAHAM, P. A., BILLHARZ, R. J., HOWALD, W. N. & DARVEAU, R. P. 2006. Porphyromonas gingivalis lipopolysaccharide lipid A heterogeneity: differential activities of tetra‐and penta‐acylated lipid A structures on E‐selectin expression and TLR4 recognition. Cellular microbiology, 8, 857-868. REYSSET, G. 1996. Genetics of 5-nitroimidazole resistance in Bacteroides species. Anaerobe, 2, 59-69. RIZZA, V., SINCLAIR, P. R., WHITE, D. C. & CUORANT, P. R. 1968. Electron transport system of the protoheme-requiring anaerobe Bacteroides melaninogenicus. Journal of bacteriology, 96, 665-671. ROLDÁN, M. D., PÉREZ-REINADO, E., CASTILLO, F. & MORENO-VIVIÁN, C. 2008. Reduction of polynitroaromatic compounds: the bacterial nitroreductases. FEMS microbiology reviews, 32, 474-500. ROLFE, R. D., HENTGES, D. J., BARRETT, J. T. & CAMPBELL, B. J. 1977. Oxygen tolerance of human intestinal anaerobes1’2. The American journal of clinical nutrition, 1762-1769. ROLFE, R. D., HENTGES, D. J., CAMPBELL, B. J. & BARRETT, J. T. 1978. Factors related to the oxygen tolerance of anaerobic bacteria. Applied and environmental microbiology, 36, 306-313.

64 Chapter I Introduction

ROTIMI, V. & DUERDEN, B. 1981. The development of the bacterial flora in normal neonates. Journal of medical microbiology, 14, 51-62. ROUND, J. L., LEE, S. M., LI, J., TRAN, G., JABRI, B., CHATILA, T. A. & MAZMANIAN, S. K. 2011. The Toll- like receptor 2 pathway establishes colonization by a commensal of the human microbiota. Science, 332, 974-977. ROUND, J. L. & MAZMANIAN, S. K. 2010. Inducible Foxp3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proceedings of the National Academy of Sciences, 107, 12204-12209. SACHER, M., DI BACCO, A., LUNIN, V. V., YE, Z., WAGNER, J., GILL, G. & CYGLER, M. 2005. The crystal structure of CREG, a secreted glycoprotein involved in cellular growth and differentiation. Proceedings of the National Academy of Sciences of the United States of America, 102, 18326-18331. SADARANGANI, S. P., CUNNINGHAM, S. A., JERALDO, P. R., WILSON, J. W., KHARE, R. & PATEL, R. 2015. Metronidazole-and carbapenem-resistant Bacteroides thetaiotaomicron isolated in Rochester, Minnesota, in 2014. Antimicrobial agents and chemotherapy, 59, 4157-4161. SAEKI, T., UMEZAWA, H., TOKIEDA-FUJISHIGE, T. & HORI, M. 1974. BIOLOGICAL EFFECTS OF AZOMYCIN (2-NITRO-IMIDAZOLE): INHIBITION OF RIBONUCLEOTIDE REDUCTASE. The Journal of antibiotics, 27, 225-227. SALYERS, A., VERCELLOTTI, J., WEST, S. & WILKINS, T. 1977a. Fermentation of mucin and plant polysaccharides by strains of Bacteroides from the human colon. Applied and environmental microbiology, 33, 319-322. SALYERS, A., WEST, S., VERCELLOTTI, J. & WILKINS, T. 1977b. Fermentation of mucins and plant polysaccharides by anaerobic bacteria from the human colon. Applied and Environmental Microbiology, 34, 529-533. SAMUELSON, J. 1999. Why metronidazole is active against both bacteria and parasites. Antimicrobial agents and chemotherapy, 43, 1533-1541. SANTANGELO, J., JONES, D. & WOODS, D. 1991. Metronidazole activation and isolation of Clostridium acetobutylicum electron transport genes. Journal of bacteriology, 173, 1088- 1095. SAWYER, S. J., MACDONALD, J. & GIBBONS, R. 1962. Biochemical characteristics of Bacteroides melaninogenicus: A study of thirty-one strains. Archives of oral biology, 7, 685-691. SCHAUMANN, R., PETZOLD, S., FILLE, M. & RODLOFF, A. C. 2005. Inducible metronidazole resistance in nim-positive and nim-negative Bacteroides fragilis group strains after several passages on metronidazole containing Columbia agar plates. Infection, 33, 368-372.

65 Chapter I Introduction

SCHWABACHER, H., LUCAS, D. & RIMINGTON, C. 1947. Bacterium melaninogenicum—a misnomer. Journal of general microbiology, 1, 109-120. SEEDORF, H., FRICKE, W. F., VEITH, B., BRÜGGEMANN, H., LIESEGANG, H., STRITTMATTER, A., MIETHKE, M., BUCKEL, W., HINDERBERGER, J., LI, F., HAGEMEIER, C., THAUER, R. K. & GOTTSCHALK, G. 2008. The genome of Clostridium kluyveri, a strict anaerobe with unique metabolic features. Proceedings of the National Academy of Sciences, 105, 2128-2133. SEKI, Y., NAKAMURA, T. & OKAMI, Y. 1970. Accumulation of 2-Aminoimidazole by Streptomyces eurocidicus. Journal of biochemistry, 67, 389-396. SHAH, H. N. & GHARBIA, S. E. 1993. Ecophysiology and of Bacteroides and related taxa. Clinical infectious diseases, 16, S160-S167. SHAH, H. N. & WILLIAMS, R. A. D. 1987. UTILIZATION OF GLUCOSE AND AMINO-ACIDS BY BACTEROIDES-INTERMEDIUS AND BACTEROIDES-GINGIVALIS. Current Microbiology, 15, 241- 246. SHINN, D. L. S. 1962. METRONIDAZOLE IN ACUTE ULCERATIVE GINGIVITIS. Lancet, 1, 1191-&. SHOJI, J., HINOO, H., TERUI, Y., KIKUCHI, J., HATTORI, T., ISHII, K., MATSUMOTO, K. & YOSHIDA, T. 1989. Isolation of azomycin from Pseudomonas fluorescens. Journal of antibiotics, 42, 1513- 1514. SIGETI, J. S., GUINEY, D. G. & DAVIS, C. E. 1983. Mechanism of action of metronidazole on Bacteroides fragilis. Journal of Infectious Diseases, 148, 1083-1089. SINDAR, P., BRITZ, M. L. & WILKINSON, R. 1982. Isolation and properties of metronidazole-resistant mutants of Clostridium perfringens. Journal of medical microbiology, 15, 503-509. SMALLEY, J. & BIRSS, A. 1987. Trypsin-like enzyme activity of the extracellular membrane vesicles of Bacteroides gingivalis W50. Journal of general microbiology, 133, 2883-2894. SMALLEY, J., BIRSS, A. & SHUTTEWORTH, C. 1988. The degradation of type I collagen and human plasma fibronectin by the trypsin-like enzyme and extracellular membrane vesicles of Bacteroides gingivalis W50. Archives of oral biology, 33, 323-329. SMITH, K. M. 1976. Porphyrins and Metalloporphyrins. Elsevier Scientific Publishing Company: Amsterdam. SMITH, P. M., HOWITT, M. R., PANIKOV, N., MICHAUD, M., GALLINI, C. A., BOHLOOLY-Y, M., GLICKMAN, J. N. & GARRETT, W. S. 2013. The microbial metabolites, short-chain fatty acids, regulate colonic Treg cell homeostasis. Science, 341, 569-573. SOUSA, F. L., THIERGART, T., LANDAN, G., NELSON-SATHI, S., PEREIRA, I. A., ALLEN, J. F., LANE, N. & MARTIN, W. F. 2013. Early bioenergetic evolution. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 368, 20130088.

66 Chapter I Introduction

SPERRY, J., APPLEMAN, M. & WILKINS, T. D. 1977. Requirement of heme for growth of Bacteroides fragilis. Applied and environmental microbiology, 34, 386-390. STAMBAUGH, J., FEO, L. & MANTHEI, R. 1968. The isolation and identification of the urinary oxidative metabolites of metronidazole in man. Journal of Pharmacology and Experimental Therapeutics, 161, 373-381. STANKO, A. P., SÓKI, J., BRKIĆ, D. V. & PLEČKO, V. 2016. Lactate dehydrogenase activity in Bacteroides fragilis group strains with induced resistance to metronidazole. Journal of Global Antimicrobial Resistance, 5, 11-14. STECHER, B., CHAFFRON, S., KAPPELI, R., HAPFELMEIER, S., FREEDRICH, S., WEBER, T. C., KIRUNDI, J., SUAR, M., MCCOY, K. D. & VON MERING, C. 2010. Like will to like: abundances of closely related species can predict susceptibility to intestinal colonization by pathogenic and commensal bacteria. PLoS Pathog, 6, e1000711-e1000711. STEFFENS, L. S., NICHOLSON, S., PAUL, L. V., NORD, C. E., PATRICK, S. & ABRATT, V. R. 2010. Bacteroides fragilis RecA protein overexpression causes resistance to metronidazole. Research in Microbiology, 161, 346-354. SWIDSINSKI, A., WEBER, J., LOENING-BAUCKE, V., HALE, L. P. & LOCHS, H. 2005. Spatial organization and composition of the mucosal flora in patients with inflammatory bowel disease. Journal of clinical microbiology, 43, 3380-3389. SZUTKA, A. 1964. Porphine-like Substances: Probable Synthesis during Chemical Evolution. SZUTKA, A., HAZEL, J. & MCNABB, W. M. 1959. Synthesis of Porphine-Like Substances Induced by Gamma-Irradiation. Radiation research, 10, 597-603. TALLY, F. P., GOLDIN, B. R., SULLIVAN, N., JOHNSTON, J. & GORBACH, S. 1978. Antimicrobial activity of metronidazole in anaerobic bacteria. Antimicrobial agents and chemotherapy, 13, 460- 465. TAMAE, C., LIU, A., KIM, K., SITZ, D., HONG, J., BECKET, E., BUI, A., SOLAIMANI, P., TRAN, K. P. & YANG, H. 2008. Determination of antibiotic hypersensitivity among 4,000 single-gene- knockout mutants of Escherichia coli. Journal of bacteriology, 190, 5981-5988. TERADA, H. 1990. Uncouplers of oxidative phosphorylation. Environmental Health Perspectives, 87, 213. THAUER, R. K. 1988. Citric‐acid cycle, 50 years on. European Journal of Biochemistry, 176, 497-508. THAUER, R. K., JUNGERMANN, K. & DECKER, K. 1977. Energy conservation in chemotrophic anaerobic bacteria. Bacteriological reviews, 41, 100.

67 Chapter I Introduction

THERON, M. M., VAN RENSBURG, M. N. J. & CHALKLEY, L. J. 2004. Nitroimidazole resistance genes (nimB) in anaerobic Gram-positive cocci (previously Peptostreptococcus spp.). Journal of Antimicrobial Chemotherapy, 54, 240-242. TOMPKINS, G. R., WOOD, D. P. & BIRCHMEIER, K. R. 1997. Detection and comparison of specific hemin binding by Porphyromonas gingivalis and Prevotella intermedia. Journal of bacteriology, 179, 620-626. TORRENTS, E. 2014. Ribonucleotide reductases: essential enzymes for bacterial life. Frontiers in cellular and infection microbiology, 4, 52. TRINH, S., HAGGOUD, A. & REYSSET, G. 1996. Conjugal transfer of the 5-nitroimidazole resistance plasmid pIP417 from Bacteroides vulgatus BV-17: characterization and nucleotide sequence analysis of the mobilization region. Journal of Bacteriology, 178, 6671-6676. TRINH, S. & REYSSET, G. 1998. Mutagenic action of 5-nitroimidazoles: in vivo induction of GC→ CG transversion in two Bacteroides fragilis reporter genes. Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis, 398, 55-65. VAISHNAVA, S., YAMAMOTO, M., SEVERSON, K. M., RUHN, K. A., YU, X., KOREN, O., LEY, R., WAKELAND, E. K. & HOOPER, L. V. 2011. The Antibacterial Lectin RegIIIγ Promotes the Spatial Segregation of Microbiota and Host in the Intestine. Science, 334, 255-258. VAREL, V. H. & BRYANT, M. P. 1974. Nutritional features of Bacteroides fragilis subsp. fragilis. Applied microbiology, 28, 251-257. VEERANAGOUDA, Y., HUSAIN, F., BOENTE, R., MOORE, J., SMITH, C. J., ROCHA, E. R., PATRICK, S. & WEXLER, H. M. 2014. Deficiency of the ferrous iron transporter FeoAB is linked with metronidazole resistance in Bacteroides fragilis. Journal of Antimicrobial Chemotherapy, 69, 2634-2643. VERNON, L. P. & KAMEN, M. D. 1954. Hematin compounds in photosynthetic bacteria. Journal of Biological Chemistry, 211, 643-662. VOS, M., ESPOSITO, G., EDIRISINGHE, J. N., VILAIN, S., HADDAD, D. M., SLABBAERT, J. R., VAN MEENSEL, S., SCHAAP, O., DE STROOPER, B. & MEGANATHAN, R. 2012. Vitamin K2 is a mitochondrial electron carrier that rescues pink1 deficiency. Science, 336, 1306-1310. WANG, B., POWELL, S. M., HESSAMI, N., NAJAR, F. Z., THOMAS, L. M., KARR, E. A., WEST, A. H. & RICHTER-ADDO, G. B. 2016. Crystal structures of two nitroreductases from hypervirulent Clostridium difficile and functionally related interactions with the antibiotic metronidazole. Nitric Oxide. WEGMANN, U., HORN, N. & CARDING, S. R. 2013. Defining the Bacteroides ribosomal binding site. Applied and environmental microbiology, 79, 1980-1989.

68 Chapter I Introduction

WHITE, D. C., BRYANT, M. P. & CALDWELL, D. R. 1962. CYTOCHROME-LINKED FERMENTATION IN BACTEROIDES RUMINICOLA. Journal of Bacteriology, 84, 822-828. WIKOFF, W. R., ANFORA, A. T., LIU, J., SCHULTZ, P. G., LESLEY, S. A., PETERS, E. C. & SIUZDAK, G. 2009. Metabolomics analysis reveals large effects of gut microflora on mammalian blood metabolites. Proceedings of the National Academy of Sciences, 106, 3698-3703. WINTER, S. E., THIENNIMITR, P., WINTER, M. G., BUTLER, B. P., HUSEBY, D. L., CRAWFORD, R. W., RUSSELL, J. M., BEVINS, C. L., ADAMS, L. G. & TSOLIS, R. M. 2010. Gut inflammation provides a respiratory electron acceptor for Salmonella. Nature, 467, 426-429. WONG, J. C., DYER, J. K. & TRIBBLE, J. L. 1977. Fermentation of L-aspartate by a saccharolytic strain of Bacteroides melaninogenicus. Applied and environmental microbiology, 33, 69-73. WYSS, M. T., JOLIVET, R., BUCK, A., MAGISTRETTI, P. J. & WEBER, B. 2011. In vivo evidence for lactate as a neuronal energy source. The Journal of , 31, 7477-7485. YAMAUCHI, K.-E. & SNEL, J. 2000. Transmission electron microscopic demonstration of phagocytosis and intracellular processing of segmented filamentous bacteria by intestinal epithelial cells of the chick ileum. Infection and immunity, 68, 6496-6504. YAP, B. C. M., SIMPKINS, G. L., COLLYER, C. A., HUNTER, N. & CROSSLEY, M. J. 2009. Porphyrin-linked nitroimidazole antibiotics targeting Porphyromonas gingivalis. Organic & Biomolecular Chemistry, 7, 2855-2863. YEUNG, T., BEAULIEU, B., MCLAFFERTY, M. A. & GOLDMAN, P. 1984. Interaction of metronidazole with DNA repair mutants of Escherichia coli. Antimicrobial agents and chemotherapy, 25, 65- 70. YU, F., ANAYA, C. & LEWIS, J. P. 2007. Outer membrane proteome of Prevotella intermedia 17: Identification of thioredoxin and iron-repressible hemin uptake loci. Proteomics, 7, 403-412. ZHENG, Y., JOSEFOWICZ, S., CHAUDHRY, A., PENG, X. P., FORBUSH, K. & RUDENSKY, A. Y. 2010. Role of conserved non-coding DNA elements in the Foxp3 gene in regulatory T-cell fate. Nature, 463, 808-812.

69 Chapter I Introduction

70 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

CHAPTER II

AN EVALUATION OF DEUTEROPORPHYRIN-NITROIMIDAZOLE ADDUCTS IN MICROBIAL COMMUNITIES FROM HUMANS AND RATS

71 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

2.0 PREFACE TO CHAPTER II

Chapter II is comprised of three parts: in the first section the response of anaerobic, human- isolated bacteria to amide- (‘P5’) and lysine-deuteroporphyrin (‘P8’) nitroimidazole adducts is explored with growth curve and viable count assays (aim I). Based on this preliminary in vitro evidence, in which none of the four examined bacterial strains were killed by P5 or P8, a plaque growth model and a rat alveolar bone loss (periodontitis) model are also explored, to examine the specificity of adducts for P. gingivalis in complex bacterial communities (aim II). Parts of the first section of this Chapter have been published (Dingsdag et al., 2015).

72 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

2.1 INTRODUCTION

Single species growth curves give insight into the kinetics of uptake and processing of deuteroporphyrin-nitroimidazole adducts and metronidazole. While informative, such experiments are an inefficient approach to define specificity of deuteroporphyrin-nitroimidazole adducts. A dental plaque growth curve provides the opportunity to challenge complex bacterial populations under defined conditions and the capacity to evaluate specificity of adducts. It is expected that if adducts are specific for P. gingivalis, the total populations of bacteria (or yield of bacterial DNA) in adduct treatments would be similar to non-treated controls, unlike the metronidazole treatment, which should have a broader spectrum (Fig. 2.1). Deuteroporphyrin IX (DPIX) is included as a control with metronidazole in these experiments and throughout this thesis. DPIX is a strong growth stimulant for P. gingivalis. In contrast, metronidazole rapidly kills this bacterium. Including DPIX and metronidazole together indicates whether the DPIX growth stimulation overcomes the bactericidal effect of metronidazole reductive activation.

A B

) )

rDNA rDNA

Log(16S Log(16S Log(16S

Figure 2.1: Expected yield of bacterial DNA after incubation with treatments, if adducts were specific for P. gingivalis (A) and expected yield of DNA if adducts were not specific for P. gingivalis (B). ‘M’ = metronidazole.

Growth experiments are subject to experimental conditions, including inoculum concentration and phase of bacterial growth, the type and concentration of nutrients and pH. On this basis, a periodontal bone loss model is explored, using P. gingivalis ATCC 33277, a pathogenic strain, capable of inducing bone loss in murine models (Evans et al., 1992, Rajapakse et al., 2002, Polak et al., 2009). The rat oral cavity is dominated by Gram-positives, such as Rothia

73 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

(Actinobacteria) and Streptococcus (Firmicutes) (Manrique et al., 2013). Proteobacteria, Cyanobacteria, Bacteroidetes, Acidobacteria and Planctomycetes associated sequences are detected at low abundance (Manrique et al., 2013). In order to aid colonisation of P. gingivalis, kanamycin was co-administered. Kanamycin is a bactericidal aminoglycoside produced by Streptomyces kanamyceticus (Umezawa et al., 1957). Kanamycin prevents protein synthesis by inhibiting translocation of tRNA when N-acetyl-diPhe-tRNA is on the acceptor site and deacylated tRNA at the donor site of the bacterial 30S ribosome (Misumi and Tanaka, 1980) which is not present in . P. gingivalis spp. encode a kanamycin acetylase, inactivating kanamycin by acetylation (Kato et al., 1995). It is also possible that as with aminoglycoside uptake by Bacteroides, that kanamycin transport occurs poorly under anaerobic conditions because uptake is dependent on nitrate and oxygen (Bryan et al., 1979). Kanamycin predominantly targets facultative anaerobes (Davies et al., 1964, Waitz et al., 1972) and is expected to broadly kill members of the Proteobacteria and Actinobacteria. In both metronidazole and adduct treatments, a reduction in P. gingivalis and associated bone loss was expected with respect to untreated control. Populations of anaerobic bacteria in the rat oral cavity were expected to remain constant after P5 and P8 treatment, unlike the metronidazole treatment, in which it was anticipated anaerobic bacteria would be killed.

2.2 METHODS

2.2.1 Single Bacterial Strain Growth Curves and Kill Curves

Bacteria (Porphyromonas gingivalis ATCC 33277, Fusobacterium nucleatum ATCC 25586, Prevotella melaninogenica ATCC 25845, Bacteroides fragilis ATCC 25285 and Tannerella forsythia ATCC 43037) were prepared in enriched brain heart infusion broth (eBHI) in an anaerobic chamber

[Don Whitley Scientific, CO2 (5%), H2 (10%) and N2 (85%)] at 37 °C. Strict anaerobic conditions were maintained with a platinum-palladium-based catalyst (Don Whitley Scientific) which was regularly changed to remove O2 introduced during entry into the chamber. Annotox reagent (Don Whitley Scientific) was also frequently changed to absorb volatile compounds, such as hydrogen sulphide. eBHI medium contained brain heart infusion (37 g L-1, Oxoid, pH 7.7), yeast extract (5g L-1, Oxoid) and L-cysteine (0.5 mg L-1, Fluka), which was autoclaved before the addition of filter sterilised menadione (2 mg L-1, Sigma), fetal bovine serum (FBS, 5% v/v) and N-acetylmuramic acid (NAM, 10 mg L-1, Sigma). Porphyrins, adducts and metronidazole derivatives were prepared as 5 mM stock solutions in DMSO (Sigma) for dilution into growth medium. To control the possible inhibition of bacterial

74 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats growth by DMSO, all cultures contained levels of this solvent equivalent to the highest concentration used in the experiment (1% v/v).

-1 Bacterial strains were first grown in eBHI with haemin (5 mg L , Fluka) to an OD600 ≈ 1.0.

Bacteria were diluted 1/2000 into fresh eBHI medium. At OD600 ≈ 1.0 bacterial cultures were sampled for viable count on eBHI agar and were diluted 1/2000 into fresh eBHI medium with treatments at a final concentration of ≈5 × 105 cfu ml-1. At various times after addition of bacterial inolcula, viable bacteria were enumerated by plating in duplicate onto eBHI agar. Plates were incubated under completely anaerobic conditions at 37°C for 96 hours before counting. Data show mean values and standard error of the mean from two or three experiments for growth curves and kill curve assays respectively.

2.2.2 Plaque-derived bacterial growth curves

Subgingival plaque scrapings were taken from the upper molars (teeth 1-3), emulsified into pre-reduced phosphate buffered saline (PBS, pH 7.0, Sigma) and immediately transfered into an anaerobic chamber. Bacterial inoculum was adjusted to a final concentration of ≈ 1×108 cfu ml-1 in PBS, determined by optical density (600nm, OD600 DiluPhotometer, Implen GmbH), and were verified in triplicate by viable count on solid blood agar containing (4%) enriched BHI medium. Bacterial inoculum was introduced at a final concentration of ≈ 1×106 cfu ml-1 (1/100 dilution) into enriched BHI broth without haemin. Treatments were prepared from concentrated stock in DMSO, and were added to growth medium at a final concentration of 50 µM. Controls included 1% (v/v) DMSO (equivalent to the final concentration in other treatments) and 50 µM: deuteroporphyrin IX (DPIX), metronidazole plus DPIX and metronidazole. In each experiment, treatments were carried out in duplicate, and experiments were repeated twice from each donor.

2.2.3 DNA Extraction

500 µl aliquots were sampled from duplicate treatments at times indicated in text following established protocols (Nadkarni et al., 2009). Bacterial were pelleted at 13 000 rpm for 5 minutes at 4°C, before removal of supernatant, and suspensed in mutanolysin (200 units, Sigma-Aldrich), lysozyme (400 µg, Sigma-Aldrich), proteinase K (400 µg, Qiagen) and diethylpyrocarbonate (DEPC, 20mM, Sigma-Aldrich), made up to 200 µl with sodium phosphate buffer (10 mM, pH 6.7). Samples were vortexed during a 40 minute incubation at 56°C, before the addition of sodium dodecyl

75 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats sulphate (1% v/v, Sigma-Aldrich) and RNAse (400 µg, Sigma-Aldrich). Samples were then incubated at 37°C for 10 minutes. ‘Buffer AL’ (200 µl, Qiagen) was added to the samples, and they were incubated at 70°C for 10 minutes. Extracted DNA was eluted into 60 µL AE buffer. DNA was purified with the QIAamp DNA Mini kit (Qiagen). Purified DNA was electrophoresed in 1% agarose (Promega) in the presence of gelred (0.5×, Biotium).

2.2.4 Real time polymerase chain reaction

Real time polymerase chain reaction (qPCR) was performed as described previously (Nadkarni et al., 2002). For estimations of total bacteria, each reaction contained 300nM of 16S rRNA gene bacterial domain-specific forward (5’-TCCTACGGGAGGCAGCAGT) and reverse (5’- GGACTACCAGGGTATCTAATCCTGTT) primers, 175nM of the fluorogenic probe (5’-6-FAM- CGTATTACCGCGGCTGCTGGCAC-TAMRA, IDT) and 30nM ROX (SOLIS BioDyne) with Firepol (SOLIS

BioDyne) buffer made of to 20 µl with 18ΩM dH20. Reaction conditions included an initial 95 °C for 20 minutes followed by 40 cycles of 95 °C for 20 seconds and 60 °C for 1 minute using an Mx3005p (Agilent). Conditions for the estimation of P. gingivalis abundance were identical to the total bacterial estimations, except forward (5’-TCGGTAAGTCAGCGGTGAAAC), reverse (5’- GCAAGCTGCCTTCGCAAT) primers and probe (5’-6-FAM-CTCAACGTTCAGCCTGCCGTTGAAA-TAMRA) amplified a P. gingivalis specific 16S rRNA amplicon (Martin et al., 2002). The fluorescent emission of FAM, was corrected by ROX. 10-fold dilutions of P. gingivalis ATCC 33277 genomic DNA of known concentration was used to construct a standard curve for the conversion of threshold cycle (Ct) to cfu ml-1 following an established methodology (Nadkarni et al., 2002). Amplicons were assessed by agarose electrophoresis (1% agarose, Promega) in the presence of gelred (0.5×, Biotium) or by melt curve analysis. Linear acrylamide, DMSO, bovine serum albumin and betaine were provided by Sigma-Aldrich and were used at a final concentration of 10 µg 20 µL-1, 300 mM, 2 ng 20 µL-1 and 100 mM respectively.

2.2.5 Periodontitis model in F344 rats

10 Female F344 rats (5 per cage) were allowed unlimited access to drinking water with kanamycin sulfate [1.7 mM (1mg/ml), Sigma-Aldrich] for two weeks. Porphyromonas gingivalis ATCC 33277, grown to an optical density of 1.2 – 1.5 in BHI medium supplemented with haemin (5 mg L-1), was pelleted and emulsified into anaerobically equilibrated carboxymethyl cellulose (2% w/v) buffered with PBS, for daily introduction (≈5×1010 cfu made up to a final 100 µl volume) into the oral

76 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats cavity of rats over the two-week period. Autoclaved chow was available ad libitum and bedding was replaced bi-weekly. Bacterial purity and concentration was assessed daily in duplicate on solid medium. Bacteria were sampled by a subgingival scrape of front incisors with a wooden tooth pick. Approval of animal procedures was granted by Westmead Hospital Animal Ethics Committee (protocol 4185.12.11).

2.3 RESULTS

2.3.1 Single Species Growth Curves

In order to confirm a previous result in which P. gingivalis was challenged with P5 (Yap, 2009) and to establish rate of kill kinetics for plaque growth curves, anaerobic bacteria were profiled in enriched BHI. In enriched BHI, P5 (3 µM) and P8 (4 µM) killed P. gingivalis with a similar concentration to that of metronidazole (3 µM) (Fig. 2.2A). F. nucleatum was rapidly killed by 20 µM metronidazole (˂30 minutes), whilst equimolar P5 and P8 did not affect F. nucleatum (Fig. 2.2B and Fig. 2.3B, Table 2.1). F. nucleatum was unable to grow in 2 µM metronidazole (Fig. 2.2B). Like F. nucleatum, B. fragilis was unaffected by adducts in enriched BHI medium, but no growth was detected in the presence of 7 µM metronidazole (Fig. 2.2D and Table 2.1). P. melaninogenica was less sensitive to metronidazole than F. nucleatum, with 8 µM metronidazole preventing growth and 20 µM metronidazole taking 2.5 hours to kill 5 × 105 cfu mL-1 of this bacterium (Fig. 2.3C). P8 significantly increased the doubling time of P. melaninogenica, with recovery between 24-48 hours similar to the optical density of untreated controls (Fig. 2.2C, Fig. 2.3C and Table 2.1). In relation to the DMSO control, P5 (20µM) slowed the doubling time of P. melaninogenica slightly (Fig. 2.2C and Table 2.1).

77 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

Table 2.1: Doubling times of bacteria grown in enriched BHI medium. The reported doubling times are for P. gingivalis grown in 4 µM DPIX. Other bacteria were assayed in DMSO alone (1% v/v) or 20 µM DPIX, P5 and P8 with DMSO (1% v/v).

-1 TD (min ) Bacterium DMSO DPIX P5 P8 P. gingivalis 165.9 164.2 N/A N/A F. nucleatum 78.8 69.9 81.3 85.2 P. melaninogenica 75.7 59.2 87.6 229.4 B. fragilis 68.1 68.8 63.3 64.3 T. forsythia 446.9 501 2007.5 1047.2

The doubling time of T. forsythia was delayed significantly by P5 (20 µM) and to a lesser extent by P8 (20 µM) whereas 2.5 µM metronidazole prevented growth of T. forsythia (Fig. 2.2E and Table 2.1). Notably, in enriched BHI growth medium the doubling time of P. gingivalis and B. fragilis in the presence of DPIX was identical to the DMSO control (Fig. 2.2A, Fig. 2.2D and Table 2.1), suggesting that there are sources of porphyrin in enriched BHI or that the medium contains other nutrients (see discussion).

78 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

A 10 B DMSO DMSO 1 DPIX 4 µM DPIX 20 µM DPIX + M 1 µM DPIX + M 2 µM 600 0.1 M 3 µM

OD M 2 µM 0.01 P5 3 µM P5 20 µM P8 4 µM 0.001 P8 20 µM 0 20 40 60 80 100 0 20 40 60 80 100 Time (h) Time (h)

C 10 D DMSO DMSO 1 DPIX 2 µM DPIX 20 µM DPIX + M 8 µM DPIX + M 7 µM 600 0.1

OD M 8 µM M 7 µM 0.01 P5 20 µM P5 20 µM

0.001 P8 20 µM P8 20 µM 0 20 40 60 80 100 0 20 40 60 80 100 Time (h) Time (h)

E 10 DMSO

1 DPIX 20 µM DPIX + M 2.5 µM 600 0.1

OD M 2.5 µM 0.01 P5 20 µM 0.001 P8 20 µM 0 5 10 15 20 Time (d)

Figure 2.2: Bacterial growth curves in enriched BHI medium. Growth of P. gingivalis (A), F. nucleatum (B), P. melaninogenica (C), B. fragilis (D) and T. forsythia (E). Concentrations of treatments in growth curves are indicated micromoles. DPIX=deuteroporphyrin IX, M=metronidazole.

79 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

A B 8 8

DMSO

)] 1 - 6 DPIX 6 DPIX + M 4 4

(cfu (cfu ml M 10 2

Viable Count ViableCount P5 2 [Log P8 0 0 0 1 2 3 4 5 0 1 2 3 4 5 Time (h) Time (h)

C 8

DMSO )]

1 6

- DPIX DPIX + M 4

(cfu (cfu ml M 10

Viable Count ViableCount 2 P5 [Log P8 0 0 1 2 3 4 5 Time (h)

Figure 2.3: Bacterial kill curves in enriched BHI medium. Concentrations of treatments in kill curves are indicated in brackets. Growth of P. gingivalis (5 µM, A); F. nucleatum (20 µM, B); and P. melaninogenica (20 µM, C). DPIX=deuteroporphyrin IX, M=metronidazole.

2.3.2 Human subgingival bacterial plaque growth assays

Table 2.2: Details of patients who donated plaque for growth curves and real time PCR assays1.

Patient Age Sex Antibiotic use (previous six months) A 66 M N B 80 M N C 24 M N D 34 F N

1The sample from patient D consisted of subgingival plaque stored cryopreserved in glycerol (20% v/v) from a patient with chronic periodontitis.

80 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

Doubling times of bacteria from subgingival plaque of patients A-D varied from 24.3-52.7 minutes (Table 2.3). In all patients, the doubling time of bacteria in the metronidazole treatment was similar or decreased slightly than the DMSO controls (average TD of 37.6 and 30.8 minutes respectively, Table 2.3). The lag phase was longest and growth in metronidazole was the most delayed from plaque of patient D, with periodontitis (Table 2.3 and Fig. 2.4). In all patients, P5 increased the doubling time of subgingival bacteria in respect to the DMSO control by around 20% (Table 2.3), an affect more pronounced by the P8 adduct (50.7%, Table 2.3) which delayed initiation of exponential growth by 2.5-5 hours (Fig. 2.4). The average doubling time of nanaerobes originating from the oral cavity (P melaninogenica, F. nucleatum and P. gingivalis) in enriched BHI (with 1% DMSO) was 74.2 minutes (Table 2.1). Doubling times of subgingival plaque under the same growth conditions were around 1.4-3 fold lower than these three nanaerobes, more consistent with doubling times of facultative anaerobes. Doubling times imply facultatively anaerobic bacteria dominate the subgingival plaque samples.

A 10 B DMSO 1 DPIX DPIX + M 0.1

M OD OD (600nm) 0.01 P5

0.001 P8 0 5 10 15 20 25 0 5 10 15 20 25 Time (h) Time (h)

C 10 D DMSO 1 DPIX DPIX + M 0.1

M OD OD (600nm) 0.01 P5

0.001 P8 0 5 10 15 20 25 0 5 10 15 20 25 Time (h) Time (h)

Figure 2.4: Growth of subgingival plaque in eBHI medium from patients A-D (A – D) in treatments (50µM) or DMSO control (1% v/v). M=metronidazole, DPIX=deuteroporphyrin IX.

81 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

Table 2.3: Doubling times of subgingival plaque bacteria from patients A-D grown in enriched BHI medium. Subgingival plaque bacteria were grown in the presence of treatments (50µM) or DMSO control (1% v/v).

-1 TD (min ) Patient DMSO DPIX DPM M P5 P8 A 43.4 60 40.5 40.8 42.5 27.9 B 30.1 22.5 28.4 27 42.7 134.5 C 24.3 26 33.2 20.1 48.8 123.3 D 52.7 54.0 52.7 35.2 52.8 20.0 Average 37.6 (±6.4) 40.6 (±9.6) 38.7 (±5.3) 30.8 (±4.5) 46.7 (±2.5) 76.4 (±30.4)

The fate of subgingival plaque bacteria deposited into enriched growth media will vary from lysis, loss of viability or replication. The bacteria with the lowest doubling times are those most likely to be detected by an increase in the optical density after the lag phase (around 4 hours). It is expected that changes in the number of individual species of subgingival bacteria are complex. Nevertheless, by analysing bacterial number (estimated by 16 ribosomal copy number), it should be possible to detect reductions in nanaerobic bacteria, known to be most susceptible to metronidazole, during challenge with this compound for comparison with the number of bacteria after challenge with P5. For example, if nanaerobic bacteria consist of 20% of the subgingival plaque bacteria, real-time PCR has the sensitivity to detect the reduction of these bacteria after challenge with a lethal concentration of metronidazole. In viable count assays (Fig. 2.3), it was established that in enriched BHI medium, 20 µM metronidazole was sufficient to kill the majority of susceptible nanaerobes within 30 minutes. As such, 30 minutes and 2 hours were selected as reference points for analysis of the number of subgingival plaque bacteria by real time PCR. At these times, nanaerobes were expected to be killed by metronidazole. Real time PCR indicated total bacteria in the P5 treatment were typically similar or slightly higher than metronidazole treatment at 2 hours (Fig. A1), which was a similar pattern to that observed at 0.5 hours (data not shown). When enumerating bacteria on blood agar, dark brown/black colonies, consistent with the colony morphology of Prevotella and Porphyromonads, were observed, although at relatively low frequency (approximately 1-5 × 104 cfu mL-1). Consistent with colony count observations, based on the rapid growth of DMSO treatment samples and the relatively high number of bacteria remaining at 2 hour time point in the metronidazole treatment, it was concluded that insufficient nanaerobes were present to determine the specificity for adducts in the subgingival plaque of the four examined patients. This problem was exacerbated by rapid growth of presumably facultative bacteria, which

82 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats by examination of the total bacteria present, made it difficult to detect the fate of anaerobic bacteria. It was decided that an in vivo approach to determine the specificity of adducts may be more informative, so a rat plaque model was pursued.

2.3.3 Rat Plaque Model

During administration of P. gingivalis, inflammation was observed in the gingiva of rats, most prominent between the incisors of the mandible. Over the two-week period, no signs antisocial behaviour, changes in weight or in grooming were observed. The daily consumption of kanamycin by each rat was approximately 13-14 mg (65-70 mg kg-1 per day). Three weeks after conclusion of P. gingivalis and kanamycin administration, total bacteria and P. gingivalis population were compared by qPCR (Fig. A1 and Table A1). Each sample contained around 106 bacteria (0.6-1.4 ng genomic DNA) (Table A1). None of the samples contained detectable levels of P. gingivalis, a result verified by agarose gel electrophoresis (data not shown), despite the sensitivity of this assay being around 10 bacterial cells (20 fg genomic DNA). While the standard curve indicated the assay had excellent sensitivity, the samples may have contained substances that inhibited PCR or DNA may have been lost during extraction. Since the total bacterial population was low, acrylamide, a known carrier of DNA, was used to prevent loss of DNA during extraction. No significant increase in total DNA was detected, with serial dilutions of P. gingivalis genomic DNA mixed with acrylamide, in respect with non-treated controls (data not shown). PCR facilitators, DMSO, betaine and BSA did not significantly increase the concentration of amplified DNA (data not shown).

Since none of the rats exhibited gross anatomical changes, and as introduction of ‘foreign’ Gram-negative nanaerobes into the rat oral cavity is notoriously difficult (Manrique et al., 2013) and the anatomy and bacterial composition differs from human counterparts, pursuit of this model was discontinued.

2.4 DISCUSSION

In single species growth curves, metronidazole killed examined anaerobic bacteria (five strains). P5 and P8 had a similar potency for P. gingivalis to that of metronidazole. In contrast, all of the other anaerobic bacteria tested grew in the presence of P5 and P8 in concentrations approximately ten fold higher than in concentrations of metronidazole that prevented growth. To facilitate comparisons between fastidious T. fosythia and other nanaerobes, enriched BHI medium

83 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats was chosen. In enriched BHI, F. nucleatum was unaffected by P5, whereas in CDC medium it was previously demonstrated that growth of this organism was delayed (Yap et al., 2009). Compared with CDC medium used previously (Yap et al., 2009), enriched BHI facilitated an increased growth rate and higher final biomass of P. gingivalis and F. nucleatum. It is possible that these results are due to increased fitness of these strains in enriched BHI medium, however, the growth of P. gingivalis and B. fragilis in DPIX was similar to DMSO carrier control, indicating enriched BHI contains sources of porphyrin (for example from fetal bovine serum) or other nutrients (such as succinate). If there are sources of porphyrin in enriched BHI, adducts may be competing with these porphyrins for uptake via porphyrin uptake pathways. The results also suggest the effectiveness of porphyrin adducts will be impacted by local concentration of nutrients (including porphyrin).

The assessment of bacterial growth by optical density and real time PCR gave an overview of the response of complex subgingival bacterial communities to adducts and metronidazole. The fate of bacteria deposited in enriched BHI medium ranges from lysis, loss of viability or replication. Subgingival bacteria that are able to replicate in eBHI growth medium will have different nutrient requirements and different doubling times. Based on the number of black/brown pigmented colonies, consistent with the colony morphology of Porphyromonas and Prevotella, it was estimated that these clades together consisted of 1-5% of total inoculated bacteria.

The average doubling time of nanaerobes of the oral cavity (P melaninogenica, F. nucleatum and P. gingivalis) in enriched BHI medium was 74.2 minutes. Doubling times of subgingival plaque under the same growth conditions were around 1.4-3 fold lower than for these nanaerobes, consistent that subgingival plaque was dominated by facultative anaerobes. Even though Prevotella and porphyromonads were presumptively identified on blood agar, it was likely these genera were being outcompeted by other bacteria with more lower doubling times. It is possible the DNA extraction method, based on chemical-lysis of cells, may not have isolated DNA from all microbes in growth curve samples. However, the chosen DNA extraction method had previously been verified against a range of facultative nanaerobes and anaerobic bacteria. Also, in all growth curves at 2 hours, the concentration of DNA was above the 106 cfu mL-1 inoculum, indicating the extraction of bacterial DNA was effective. In order to examine the fate of presumably outcompeted nanaerobic bacteria, it would have been possible to slow the growth of facultative bacteria with antibiotics. Alternatively, the number of nanaerobic genera such as Porphyromonas could have been estimated by real time PCR using oligonucleotide primers specific for this clade. However, single species growth curves indicated sources of porphyrin or other nutrients might be obscuring the response of

84 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats nanaerobic bacteria to adducts. It was decided instead that a periodontal bone loss model may be more informative.

The total bacteria detected three weeks after kanamycin and P. gingivalis administration, assessed by 16S rDNA real time PCR, equated to 106 cfu per sample. Based on the sampling procedure, a subgingival scrape of the front mandible incisors, the detected population of bacteria (DNA) was in the expected range. As mentioned above, the DNA extraction method had already been optimised for the extraction of DNA from oral bacteria. Nevertheless in an attempt to increase sensitivity of extraction and amplification, a DNA carrier and three PCR facilitators were used, none of which increased the detection of total bacterial genomic DNA. It is presumed that the disruption of oral microbes by kanamycin caused the observed inflammation, which may have inhibited the colonisation of P. gingivalis. It is more likely however, that the disruption caused by kanamycin was insufficient and that an antibiotic cocktail and a physical insult (tooth ligature) was required. Alternatively, a two-week induction period may have been insufficient or co-administration of other bacteria was required with P. gingivalis to allow colonisation and bone resorption (Kesavalu et al., 2007). The rat oral cavity is dominated by facultative bacteria. If P. gingivalis colonisation was successful, there was no guarantee that the populations of native nanaerobes in the rat oral cavity would also increase. Rather than introduce P. gingivalis, a foreign anaerobic bacterium to an animal model, it was decided that it may be more informative to select a model replete with anaerobic bacteria, sensitive to nitroimidazoles, allowing interrogation of the specificity of adducts.

85 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

2.5 Chapter II references

BRYAN, L., KOWAND, S. & VAN DEN ELZEN, H. 1979. Mechanism of aminoglycoside antibiotic resistance in anaerobic bacteria: Clostridium perfringens and Bacteroides fragilis. Antimicrobial agents and chemotherapy, 15, 7-13. DAVIES, J., GILBERT, W. & GORINI, L. 1964. STREPTOMYCIN, SUPPRESSION, AND THE CODE. Proceedings of the National Academy of Sciences, 51, 883-890. DINGSDAG, S. A., YAP, B. C., HUNTER, N. & CROSSLEY, M. J. 2015. Amino acid-linked porphyrin- nitroimidazole antibiotics targeting Porphyromonas gingivalis. Organic & biomolecular chemistry, 13, 98-109. EVANS, R., KLAUSEN, B., SOJAR, H., BEDI, G., SFINTESCU, C., RAMAMURTHY, N., GOLUB, L. & GENCO, R. 1992. Immunization with Porphyromonas (Bacteroides) gingivalis fimbriae protects against periodontal destruction. Infection and immunity, 60, 2926-2935. KATO, T., HIRAI, K. & OKUDA, K. 1995. Identification of a chromosomally encoded kanamycin acetylase in Porphyromonas gingivalis. FEMS microbiology letters, 131, 301-306. KESAVALU, L., SATHISHKUMAR, S., BAKTHAVATCHALU, V., MATTHEWS, C., DAWSON, D., STEFFEN, M. & EBERSOLE, J. L. 2007. Rat model of polymicrobial infection, immunity, and alveolar bone resorption in periodontal disease. Infection and Immunity, 75, 1704-1712. MANRIQUE, P., FREIRE, M. O., CHEN, C., ZADEH, H. H., YOUNG, M. & SUCI, P. 2013. Perturbation of the indigenous rat oral microbiome by ciprofloxacin dosing. Molecular Oral Microbiology, 28, 404-414. MARTIN, F. E., NADKARNI, M. A., JACQUES, N. A. & HUNTER, N. 2002. Quantitative microbiological study of human carious dentine by culture and real-time PCR: Association of nanaerobes with histopathological changes in chronic pulpitis. Journal of Clinical Microbiology, 40, 1698- 1704. NADKARNI, M. A., MARTIN, F. E., HUNTER, N. & JACQUES, N. A. 2009. Methods for optimizing DNA extraction before quantifying oral bacterial numbers by real-time PCR. Fems Microbiology Letters, 296, 45-51. NADKARNI, M. A., MARTIN, F. E., JACQUES, N. A. & HUNTER, N. 2002. Determination of bacterial load by real-time PCR using a broad-range (universal) probe and primers set. Microbiology- Sgm, 148, 257-266. POLAK, D., WILENSKY, A., SHAPIRA, L., HALABI, A., GOLDSTEIN, D., WEISS, E. I. & HOURI‐HADDAD, Y. 2009. Mouse model of experimental periodontitis induced by Porphyromonas

86 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

gingivalis/Fusobacterium nucleatum infection: bone loss and host response. Journal of clinical periodontology, 36, 406-410. RAJAPAKSE, P. S., O'BRIEN-SIMPSON, N. M., SLAKESKI, N., HOFFMANN, B. & REYNOLDS, E. C. 2002. Immunization with the RgpA-Kgp proteinase-adhesin complexes of Porphyromonas gingivalis protects against periodontal bone loss in the rat periodontitis model. Infection and immunity, 70, 2480-2486. WAITZ, J. A., MOSS, E. L., DRUBE, C. G. & WEINSTEIN, M. J. 1972. Comparative Activity of Sisomicin, Gentamicin, Kanamycin, and Tobramycin. Antimicrobial Agents and Chemotherapy, 2, 431- 437. YAP, B. C. M., SIMPKINS, G. L., COLLYER, C. A., HUNTER, N. & CROSSLEY, M. J. 2009. Porphyrin-linked nitroimidazole antibiotics targeting Porphyromonas gingivalis. Organic & Biomolecular Chemistry, 7, 2855-2863.

87 Chapter II An evaluation of deuteroporphyrin-nitroimidazole adducts in microbial communities of humans and rats

88 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

CHAPTER III

THE METABOLISM OF A DEUTEROPORPHYRIN-NITROIMIDAZOLE ADDUCT BY NANAEROBIC BACTEROIDETES

89 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

3.0 PREFACE TO CHAPTER III

Studies using single species growth rates and by viable count determinations, as reported in Chapter II, revealed some pharmacokinetic properties of P5 and P8. Neither an ex vivo plaque model nor an in vivo rat P. gingivalis-induced periodontal bone loss model successfully interrogated the specificity of deuteroporphyrin-nitroimidazole adducts for P. gingivalis among complex microbial communities.

For reasons outlined in Chapter II, a human ex vivo distal colon culture model was considered a more suitable than the rat oral cavity to determine the specificity of deuteroporphyrin- nitroimidazole adducts, since it is dominated by nanaerobic nitroimidazole-sensitive Bacteroidetes (including Porphyromonas) and Clostridia (of the Firmicutes phylum). As a reference point, growth curves of 23 strains from dominant bacterial phyla originating from the human gastrointestinal tract are grown in brain heart infusion medium. Interesting findings relating to Bacteroides are further characterised. Using B. fragilis mutants, light-based biochemical, in silico analyses coupled with high performance liquid chromatography mass spectrometry, a schema is proposed for the metabolism of P5. The specificity of P5 is also compared with metronidazole by Illumina sequencing of the 16S ribosomal DNA of human bacteria challenged with these compounds in an ex vivo distal colon culture model.

90 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

3.1 INTRODUCTION

There are several ways to analyse complex bacterial populations, all of which have deficiencies. To overcome deficiencies inherent in each method, a combination of molecular and more traditional culture-based approaches are used in Chapter III. For example, bacteria may be analysed by using primers targeting conserved regions that flank variable portions of the universally conserved 16S rDNA. Bacteria may then be identified by sequencing the amplified variable region/s and comparing with databases of known 16S rDNA sequences. Misrepresentation of the types of bacteria in a given community in such analyses may be introduced from the DNA extraction, PCR primers and amplification conditions, sequencing platform and also post sequencing analyses. Specifically, a DNA extraction method may under represent different groups of bacteria. ‘Universal’ 16S rDNA primers are designed to target conserved regions of the bacterial ribosomal DNA on the basis of a ‘best fit’ consensus. Thus, the 16SrDNA template is less likely to be amplified as it diverges from the primers, introducing bias in the types of template (or bacterial clades) that are amplified (Lee, et al., 2012). Empirical testing, coupled with an in-silico examination of primer-template specificity helps determine the degree to which divergence from the primers (or consensus conserved region) will allow a given template 16S rDNA molecule (or bacterial group) to amplify (Klindworth et al., 2012, Schirmer et al., 2015). As a taxonomic group of bacteria diverges from the ‘universal primers’, it is less likely to be amplified, thus no 16S rDNA primers are truly universal (Lee, et al., 2012, Tremblay et al., 2015). The nine variable regions in the 16S rDNA give variable resolution, so the region/s chosen will inform resolution, which again may be interrogated by in silico analyses. Lastly, new bacterial phyla, are continually being detected/characterised and added to 16S rDNA databases, such as the non-photosynthetic anaerobic clade Melainabacteria, relatives of the Cyanobacteria recently found in the mammalian gut (Di Rienzi et al., 2013).

Several ‘next generation’ sequencing technologies, including 454 pyrosequencing and Illumina platforms (see Fig. 3.1) are commonly used to survey complex bacterial consortia targeting the 16S rRNA gene or total genomic DNA. For 16S rRNA gene surveys, 454 pyrosequencing was the dominant platform since it was released first (in 2005) and also because of the originally longer sequencing capacity over competing technologies (Mardis, 2008, Degnan and Ochman, 2012). The nature of error introduced by sequencing platforms may result from insertions, deletions and base substitutions. The most prominent source of error inherent to 454 and Illumina platforms depends on the sequencing strategy, although in both technologies error results primary due to substitutions (Tremblay et al., 15). The Illumina ‘MiSeq’ platform has the capacity to sequence ≈250 bp amplicons from both ends of the same sequence generally known as ‘paired end’ sequencing. Paired-end

91 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Figure 3.1: Solid phase amplification used in Illumina sequencing [from (Metzker, 2010)]. First, a ‘library’ is prepared, during which the 5’ and 3’ ends of single stranded DNA molecules are appended to linkers (represented by red or blue rectangles) during PCR amplification. Linkers that are complementary to a flow cell (hexagonal base), adhere and form a bridge. Next, DNA molecules are amplified and the template plus newly synthesised reverse strand are linearised. This process proceeds iteratively until thousands of clonal DNA molecules are synthesised. Finally, nucleotides modified with a 3’ blocker and a fluorescent tag are added, along with primers. The 3’ block allows only one nucleotide to be incorporated per DNA sequence (adenine, thymine, cytosine or guanine). After the addition of each nucleotide, an image is taken, to determine which was incorporated. To allow incorporation of new modified nucleotides (ie. thymine), DNA sequences are chemically ‘de- blocked’ until the DNA molecule is fully sequenced.

sequencing increases the fidelity of 16S rDNA sequencing if the amplicon is small enough, since each base is sequenced twice (Tremblay et al., 15). As noted, 454 pyrosequencing sequences longer amplicons, however, the accuracy of base calling falls after approximately 350 bp, such that some sequences do not reliably reflect the original DNA template (Gilles et al., 11). The quality of base calling using Illumina Miseq sequencing also falls as the DNA molecule is sequenced from the 5’ to 3’ end (Nelson, 2014). The highest fidelity base calling of paired end Illumina sequencing is achieved by near-full or full-length overlap of the amplicon being sequenced, currently limiting this technology to the variable region 4 (V4) of the 16S rDNA which is approximately 254 bp (Kozich et al., 2013).

92 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Both Illumina and 454 platforms provide a metric for the confidence of each base call against the parental sequence (similar to a Phred score associated with Sanger sequencing). For applications such as genome sequencing, a ten-fold coverage of the majority of target DNA is achieved, so that the fidelity (determined by the quality score, or ‘q score’) isn’t as important. However, for sequencing portions of a single gene such as the 16S rDNA, sequencing fidelity is important, as DNA molecules are sequenced once or twice. A q score of 20 to 30 corresponds to a predicted 99-99.9 % accuracy in base calling respectively (Ewing and Green, 1998). Genus level associations are currently accepted at 97% cut-off. A q-score less than or equal to 20, starts to impinge on the quality of binning sequenced DNA into taxonomic groups (Kozich et al., 2013). Notably, Illumina technologies are known to inflate the OTU number of sampled microbes from human faeces compared with longer reads (454 pyrosequencing Titanium technology), which is in part due to decreases in accuracy as a given molecule is sequenced (Claesson et al., 2010). After consideration of these technologies, Illumina Miseq paired-end sequencing of the V4 region was chosen to analyse complex bacterial populations, since this methodology provides the highest theoretical fidelity for one region of the 16S rDNA and gives the closest approximation to shotgun sequencing of environmental samples (Tremblay et al., 15).

In order to establish the specificity of deuteroporphyrin-nitroimidazole adducts, a human ex vivo distal colon culture model is used in Chapter III. The main advantage of an ex vivo distal colon culture model is that bacterial inhabitants of human distal gut are largely nitroimidazole-sensitive Bacteroidetes (including Porphyromonas) and Clostridia (a Class of nanaerobes of the Firmicutes phylum, see also Table 3.1 for a phylogeny of these bacteria). Since human faecal bacteria are generally amenable to culturing (Goodman et al., 2011, Rea et al., 2011, Maurice et al., 2013), human faeces provides a largely nitroimidazole-sensnitve bacterial population for challenge with P5 or metronidazole. Selective growth medium permit quantification of viable bacteria, instead of using DNA molecules as a proxy for the number or type of bacteria in a given sample. If bacteria are culturable, selective growth media may overcome some of the issues of 16S rDNA copy number, PCR amplification-bias, amplification of DNA from non-viable bacteria and/or errors associated with next generation sequencing mentioned above. Gentamicin is an aminoglycoside, produced primarily by Micromonospora spp. (such as M. purpurea, an Actinobacteria). Bacteria residing in the human gut, such as Prevotella and Fusobacterium tend to grow in high concentrations of gentamicin, while Bacteroides tend to be even more resistant to gentamicin (Finegold and Sutter, 1971). The vast majority of Bacteroides also hydrolyse esculin into esculetin and dextrose (Livingston et al., 1978). Taken together, gentamicin kills Proteobacteria and Firmicutes, whilst members of the Fusobacteria and non Bacteroides members of the phylum Bacteroidetes (including Prevotella and

93 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Porphyromonas) tend to be inhibited by bile. Of the remaining bacteria that grow on solid medium with bile and gentamicin, esculin hydrolysis may be detected by the black precipitant formed from esculetin and ferric ammonium citrate, providing presumptive identification of Bacteroides. Indeed, all 160 Bacteroides strains tested in vitro blackened ‘Bacteroides Bile Esculin’ (BBE) agar, other than two of seven Bacteroides vulgatus strains (Livingston et al., 1978). The recipe used for BBE agar is similar to recipe optimised for brain heart infusion gentamicin bile ‘BHIGB’ agar in Chapter III, although with 75% less bile (see Table A2 for optimisation of bile and gentamicin concentrations).

3.2 METHODS

3.2.1 Medium Preparation and Growth Curves

All bacterial growth curves were carried out using brain heart infusion medium (BHI), unless otherwise noted, in an anaerobic chamber at 37 °C in an atmosphere containing CO2 (5%), H2 (10%) and N2 (85%). Strict anaerobic conditions were maintained with a platinum-palladium-based catalyst

(Don Whitley Scientific) which was regularly changed to remove O2 introduced during entry into the chamber. Annotox reagent (Don Whitley Scientific) was also frequently changed to trap volatile gasses. BHI contained brain heart infusion (37 g L-1, Oxoid, pH 7.7), yeast extract (5g L-1, Oxoid), L- cysteine (0.5 mg L-1, Fluka), and when supplemented, haemin (5 mg L-1, Fluka) was included before autoclaving. Menadione (2 mg L-1, 0.22 µM filtered, Sigma) was added after autoclaving. Bacteria were first grown in BHI with haemin to OD600 ≈ 1.0 (or Faecalibacterium prausnitzii ≈ 0.8, and Parvimonas micra; ≈ 0.6). Bacteria were diluted 1/100 into fresh growth medium without haemin. At OD600 ≈ 1.0, bacteria were again diluted 1/2000 into fresh BHI medium without haemin at a final concentration of ≈5×105 cfu ml-1 and growth was monitored (by optical density at 600nm). This procedure limited porphyrin carryover, while allowing growth of porphyrin-requiring strains (Prevotella and Porphyromonas) as described previously (Paramaesvaran et al., 2003). For each growth curve, bacterial purity and concentration was confirmed in duplicate on BHI containing 4% sheep blood. Bacteroides thetaiotaomicron VPI 5482 was prepared in the same medium for viable counts and was enumerated in duplicate for each treatment on solid BHI supplemented with 4% sheep blood. Adducts or metronidazole were defined as bactericidal when no growth (turbidity) occurred at a given concentration after 96 hours. Bacterial strains (23 for growth curves and four additional strains used in validating BGBHI agar) were available from our culture collection; purchased from American Type Culture Collection (Bacteroides thetaiotaomicron ATCC 29148 (VPI 5482), Bacteroides vulgatus ATCC 8484 and Clostridium difficile ATCC 43255); or from Deutsche

94 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Sammlung von Mikroorganismen und Zellkulturen (F. prausnitzii DSM 17677 and Eubacterium rectale DSM 17629).

Bacteroides fragilis ADB77 (ΔthyA1) contains a deletion of the thymidylate synthase gene

(thyA1), requiring thymine supplementation for growth (Baughn and Malamy, 2002). B. fragilis ADB247 and ADB260 are derived from B. fragilis ADB77, with small deletions from frdC (of transmembrane helices 2-5, ΔfrdC) or frdB (iron-sulfur coordination sites, ΔfrdB) (Baughn and Malamy, 2003). Since mutants have portions deleted from their genomes, antibiotic or other selective agent is not required, although B. fragilis ADB77 and derivative mutants require thymine for growth. Mutant B. fragilis ADB77 and mutant derivatives ADB247 and ADB260 were grown as described (Baughn and Malamy, 2002), by passaging from BHI(S) with supplemented haem into a minimal medium (‘glucose minimal medium’, ‘AMM gluc’). At an OD600 of 0.5, strains were diluted 1/2000 into fresh AMM gluc medium. Bacterial purity and concentration was confirmed in duplicate on BHI(S) supplemented with haemin. Strains were not complemented, since under identical conditions, it was previously shown that mutants were complemented by vector plasmids carrying the frd operon (Baughn and Malamy, 2003).

BGBHI agar was based on the original composition developed by Livingston et al, (Livingston et al., 1978) in which brain heart infusion (37g L-1, pH 7.7), esculin (0.5g L-1, Sigma), ferric ammonium citrate (0.25g L-1, Sigma), L-cysteine (0.5mg L-1), oxgall (‘bile’, 5g L-1, Difco) and gentamicin (100 mg L- 1, Sigma) were autoclaved, before the addition of menadione (2 mg L-1).

Concentrated stock solutions of porphyrin/nitroimidazoles (<10mM) were dissolved in molecular grade DMSO (Sigma) before sterilisation with 0.22 µm non-pyrogenic cellulose acetate filters (Corning). In each assay, concentrated stock solutions were diluted 1/100 into growth medium so that each treatment contained 1% DMSO. Treatments were titrated in 1-2 µM intervals between 0.5-20 µM. Pyrex growth tubes and caps were soaked in detergent and bleach between experiments, before autoclaving and extensive washing to remove residual porphyrin.

-1 Deuteroporphyrin IX-Cl2 (511.2 g mol ) was provided by Frontier Scientific, metronidazole (171.2 g mol-1) by Sigma and ‘metronidazole-amine’ (1-(2-Aminoethyl)-2-methyl-5-nitroimidazole dihydrochloride monohydrate, 169.1 g mol-1) by Santa Cruz . P5 (663.5 g L-1) was synthesised as described previously (Yap et al., 2009). Unless stated otherwise, error incorporates standard error of the mean from two experiments containing duplicate treatments. The MIC is defined as the absence of growth after 96 hours incubation in growth medium.

95 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

3.2.2 Human ex vivo distal colon culture model growth curves, DNA extraction and Illumina Sequencing

One portion of fresh human faeces (100 mg) was immediately emulsified into 4900 µL anaerobically equilibrated warm PBS, before transfer into an anaerobic chamber. The second portion of faeces (200 mg) was immediately stored at -80°C for DNA extraction. Viable counts were made on BHI agar with 4% sheep blood or on BGBHI, before introduction into BHI broth at 1 × 106 cfu ml-1 with 20 µM treatments (see below) or DMSO carrier control (1%). The viable count of Bacteroides was estimated on BGBHI agar at 5 and 10 hours for comparison with the count obtained on BGBHI immediately before inoculation. At 10 hours, a 1 mL sample was stored -80 °C for DNA extraction and sequencing of 16S rDNA. Four ex vivo distal colon experiments were performed from one donor. Statistical analyses of viable counts were performed using unpaired, non-parametric t- test (GraphPad Prism, v.6).

For DNA extraction, 1 mL samples were defrosted on ice, and bacteria were isolated (14,000 g, 10 minutes, 4°C). DNA was extracted from bacterial pellets or faeces (200 mg) using QIAamp Fast DNA Stool MiniKit (QIAGEN), following the manufacturer’s protocol with the following amendments. 1) After a 10 minute incubation at 95C, lysozyme (800 µg, Sigma-Aldrich) and mutanolysin (400 U, Sigma-Aldrich) were added and samples were incubated for 1 hour at 56 °C. 2) All remaining supernatant lysate (~800 µL) was added to buffer AL (800 µL) with proteinase K (15 µL, QIAGEN) for a one hour incubation at 70°C. The standard protocol was followed until 3) samples were eluted into 100 µL ‘TAlowE’ buffer (10mM Tris, 0.5mM EDTA, pH 7.5). One extraction blank was included for each growth curve experiment. An average of 282 (±50) ng µL-1 genomic DNA, assessed by absorbance (Nanophotometer, Implen) was isolated from the four faecal inoculum samples.

Genomic DNA was diluted to ≈10 ng µL-1 and sent to Ramaciotti Centre for Genomic Analysis (UNSW, Australia) for PCR amplification and sequencing of the V4 region [with 515f (5’- GTGCCAGCMGCCGCGGTAA) and 806r (5’-GGACTACHVGGGTWTCTAAT) corresponding to positions 533-786 of Escherichia coli 83972] of the 16s rRNA gene (300-350bp) using a mate pair dual sequencing strategy as described elsewhere (Caporaso et al., 2012). An illumina Miseq instrument using V2 chemistry had an average PhiX 6.21% and a cluster density of 1072±44. Sequences were demultiplexed and paired end reads were joined into contigs using Mothur (Schloss and Westcott, 2011) leaving 9,391,212 sequences. Sequences were processed with Mothur (v. 1.36) following the ‘Illumina SOP’ (http://www.mothur.org/wiki/MiSeq_SOP). Sequences were removed if they exceeded 275 bp, contained any ambiguous base calls or more than 8 homopolymeric runs or did

96 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes not map to regions 13862-23444 of the 50,000 column SILVA reference alignment (r 119, (Pruesse et al., 2007). Sequences flagged as chimeric using the Mothur implementation of UCHIME (Edgar et al., 2011) were removed and sequences were amalgamated (‘preclustered’, diffs=2) into one sequence if they differed by less than 4 bases (Huse et al., 2010). Sequences were classified using the naïve Bayesian algorithm against the Mothur-formatted Greengenes reference 16S ribosomal database (‘gg_13_8_99’, with 202,421 bacterial and archaeal sequences) using an 80% pseudobootstrap value. Sequences were clustered into OTUs (3% similarity threshold). The number of quality filtered reads within a sample were converted to a percentage of the total reads within that sample. The average and standard error of the mean were calculated for remaining taxa within each sample type. Taxa that were less than 0.5% average abundance in all samples are not shown. Metastats, which uses a Student’s t-test (White et al., 2009) was used within Mothur to assess statistical significance of OTUs binned at 0.05 dissimilatory threshold (approximating genus level).

3.2.3 Micro-broth Dilution for Antibiotic Susceptibility, DNA Extraction and nim-Gene Amplification

Clostridum difficile were isolated from the stools of 100 symptomatic patients presenting to Westmead Hospital, Australia during July - October 2015. Strains were PCR-ribotyped and evaluated by Matrix-Assisted Laser Desorption Ionization-Time-of-Flight Mass Spectrometry (MALDI-TOF MS, Bruker MALDI Biotyper using MALDI Biotyper Automation Control, software version 2.0.43.8) following established protocols in a clinical laboratory (Stubbs et al., 1999, Schmitt et al., 2012). 100 strains were collected from 42 males and 58 females, which were characterised into 25 ribotypes. In an anaerobic chamber, strains were plated on pre-reduced BHI agar (1.5% w/v)]. Under strictly anaerobic conditions, strains were rapidly transferred to the Institute of Dental Research. After a 24 hour incubation, strains were inoculated into BHI growth medium in an anaerobic chamber. At an

OD600 of ≈1.0, strains were stored at -80°C in a final concentration of 10% (v/v) glycerol. 1 mL of the C. difficile cultures was retained at -80°C for DNA extraction following the protocol outlined in the ex vivo human faecal model.

In an anaerobic chamber, metronidazole (0.07 – 17.12 µg mL-1) and P5 (0.52 – 66.35 µg mL-1) were introduced in two-fold serial dilutions ranging from 0.4-100 µM into 96 well micro titre plates with BHI growth medium. At the same time, C. difficile strains were streaked onto BHI agar. After 24 hour incubation on BHI agar, C. difficile were introduced at a final concentration of 5×105 cfu mL-1 into micro titre plates containing metronidazole and P5. After a 48 hour incubation of C. difficile

97 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes strains with treatments, the optical density was recorded (570 nm, Benchmate microplate reader,

Bio rad). The concentration of metronidazole or P5 that prevented growth (OD570<0.01) after a 48 hour incubation was defined as the MIC. For each 96-well plate, growth of C. difficile, sterility of media and metronidazole/P5 were verified in triplicate. Antimicrobial breakpoints were established based on guidelines from the European Committee on Antimicrobial Susceptibility Testing (EUCAST, 2016). All media and 96-well plates were equilibrated for 24 hours in the anaerobic chamber before use. Strictly anaerobic conditions were maintained by frequent exchange of a platinum-palladium catalyst (Don Whiteley Scientific) and anotox (Don Whiteley Scientific).

PCR screening for the nimA-H genes (458 bp) was carried out using extensively validated nim primers (nim5: 5’-ATGTTCAGAGAAATGCGGCGTAAGCG and nim3: 5’-GCTTCCTTGCCTGTCATGTGCTC) (Trinh and Reysset, 1996, Husain, 2013, Meggersee and Abratt, 2015). PCRs were performed with 0.2 µM primers, approximately 1 ng genomic DNA with HotStarTaq Master Mix (1x, QIAGEN) made up to a final volume of in 25 µL. The cycling protocol for nimA-H consisted of 95°C for 15 min, followed by 35 cycles of 95°C for 30 s, 60°C for 30 s, 72°C for 1 min and a final 10 minute extension at 72°C. PCR screening for the nimJ was carried out using previously validated primers (615-616nimJ-qRT-F: 5’- TGACAAGGCTTCGTTCTGTG and 615-616nimJ-qRT-R: 5’-GTCGAAACGAATCATCAGCA) (Husain et al., 2013, Meggersee and Abratt, 2015). The cycling protocol for nimJ was identical, other than the use of a 55°C annealing temperature for 31 cycles (Husain et al., 2013, Meggersee and Abratt, 2015).

3.2.4 Fumarate Reductase and Succinate Dehydrogenase Enzyme Assays

B. thetaiotaomicron VPI-5482 was grown anaerobically and passaged in BHI medium as described for growth curves, before introduction into fresh growth medium with 20 µM haem, DPIX, P5 or DMSO (1% v/v) carrier in 50 mL Falcon tubes. During exponential phase of growth, Falcon tubes were tightly capped and bacteria were harvested at 10,000 g for 15 minutes at 4 °C. Cell pellets were washed with chilled buffer A and the washed pellet was made up to 5 ml with chilled pre-reduced buffer A (Macy et al., 1975). Buffer A consisted of potassium phosphate (50 mM), NaCl (150 mM), sucrose (100 mM) and sodium thioglycolate (10 mM), pH 7.0) as previously described (Macy et al., 1975). Cells were lysed via a French press (80 psi, Avestin) and aliquots were snap frozen by liquid nitrogen for storage at -80°C. All steps were carried out in an anaerobic chamber, except centrifugation and lysis.

Enzyme activity assays were carried out in quartz cuvettes with a 1 mL reaction volume. Reagents were prepared in an anaerobic chamber, before cuvettes were tightly sealed and removed

98 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes for recording assays. Enzyme activities were assayed in 50 mM potassium phosphate, pH 7.6, with

5 mM MgCl2 (Macy, 1975). Fumarate reductase activity was measured by recording the fumarate- dependent increase in absorbance of the low potential electron donor safranin O (200 µM, E°- 290 mV), indicated by increase in absorbance at 521 nm. Immediately prior to each fumarate reductase assay, fresh sodium dithionite (5 mM, E°-660 mV) was added to reduce safranin O stock

(5 mM) to a starting absorbance of A521≈0.01. Fumarate reductase activity is expressed in µmol safranin oxidised min-1 mg-1 protein in the presence of 10 mM sodium fumarate (using Ɛ = 27.8 cm-1 mM-1). Succinate dehydrogenase activity was determined by measuring the succinate-dependent reduction of methylene blue (44 µM, E°10 mV) as indicated by a decrease in absorbance at 667 nm (Baughn, 2003). Succinate dehydrogenase activity is expressed in µmol methylene blue reduced min- 1 mg-1 protein in the presence of 3 mM sodium succinate (using Ɛ = 25.0 cm-1 mM-1). Assays were initiated by the addition of cell lysate and were monitored for 10 minutes at 23 °C (Beckman DU800). Error incorporates mean ±SEM from two experiments prepared from one cell extract.

3.2.5 Chemical and Biological Reductive Activation of P5, Purification of Products and Liquid Chromatography and High Performance Mass Spectrometry of

Sodium dithionite is used to model the biological reductive activation of nitroimidazoles (LaRusso et al., 1978). In an anaerobic chamber, porphyrin or nitroimidazole stock (10-40mM) in DMSO were made up to final volume with 10 equivalents (100-400mM) sodium dithionite (1M stock in anaerobically equilibrated PBS, 0.22 µM filtered). After a 5 minute incubation, an aliquot was used for B. thetaiotaomicron or C. difficile dithionite growth curves or at indicated concentrations. Treatments for growth curves were treated or non-treated with sodium dithionite, including DMSO.

Extraction of the reduced products of P5 was based on an established porphyrin extraction method (McConville and Charles, 1979, Granick et al., 1972), in which porphyrins are partitioned into ethyl acetate with acetic acid (at pH 3.5, ‘fraction 1’) and a subsequently free-base porphyrins are purified in hydrochloride acid (at pH 1-2, ‘fraction 2’). B. thetaiotaomicron VPI-5482 was grown in defined ‘AMM Gluc’ medium (Baughn and Malamy, 2002) to OD600≈0.5, before dilution 1/2000 into AMM Gluc with P5 (20 µM). After 48 hours, pH was adjusted to 3.5 with acetic acid (≈70 mL per 500 mL) and the culture autoclaved (‘dry cycle’, 15 minutes, 121 °C), to assist release of protein bound porphyrin during extraction.

99 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

P5, P5-reductively activated by dithionite and DPIX or P5 metabolised by B. thetaiotaomicron were diluted 1/10 in methanol for liquid chromatography coupled mass spectrometry (LC-MS). Low- resolution measurements were obtained in positive mode from 150-1500 m/z on an LCQ Deca XP Plus mass spectrometer coupled in-line with PDA detection (Surveyor PDA Plus). Chromatography was performed on a C18 column (2.1 mm x 100 mm, Phenomenex Luna) using a Surveyor HPLC pump and auto sampler. Sample (1 µL) was injected and separated using a binary gradient of 0.1% formic acid in water against acetonitrile. The column eluent was directed through the flow cell of the PDA and into the electrospray source of the mass spectrometer. High resolution measurements were obtained using a Q-Exactive Plus mass spectrometer coupled to a U3000 UHPLC pump and auto sampler. A C18 BEH UHPLC column (2.1 mm x 100 mm, Waters) was employed on this system due to the differences in pumps. Chromatographic conditions and mass spectrometric detection were otherwise identical. Slight differences in retention times of ions between systems are due to slight changes in the employed systems.

LC-MS data was analysed with Xcalibur (Thermo, v. 3.0.63) and ChemBioDraw Ultra (Perkin Elmer, v. 13.0.2.3021). The absorbance at 400 nm in chromatograms, corresponding to the Soret absorption region of porphyrins is shown because it is the most intense region. Q-band absorbance signatures were used to verify retention time of possible porphyrin compounds (data not shown). Parental ions and fragments were examined manually 0.5 minutes before and after retention times of tetrapyrrole absorbance peaks (at 400 nm). The base peak, or ion that had the highest intensity was assigned an abundance of 100%. LC-MS is not necessarily quantitative, though the ionisation efficiency relative to the base peak is reported graphically. Ions were examined from the base peak abundance to 6 orders of magnitude in the lower range. At predicted retention times, peaks were also inspected using Xcalibur ion detection system. For example, to confirm that P5 was metabolised in relevant samples, searches were made manually and via Xcalibur for [m/z + H+] = 663.3043. Specific search strategies are further elaborated in text. Where possible, structures and formulae consistent with ions were manually solved and depicted using ChemBioDraw Ultra. Specific

3.2.6 In Silico Analysis of Fumarate Reductase

For fumarate reductase, BLASTP searches were conducted using B. thetaiotaomicron ATCC 29148 frdC (230 residues), frdA (647 residues) and frdB (251 residues). Curated hits were binned at the genus resolution or the next highest available taxonomic classification. Hits with an average homology of ≥60% of the total frdCAB operon are shown.

100 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

3.3 RESULTS

3.3.1 The response of single species and more complex bacterial populations of bacteria to P5

In Chapter II, the maximal growth of P. gingivalis ATCC 33277 was observed in enriched BHI medium with and without porphyrin supplementation. These results suggested that porphyrin or other nutrient sources in enriched BHI growth medium were obscuring the porphyrin-based maximal growth of P. gingivalis. To determine whether sources of porphyrin in enriched BHI growth medium were obscuring the growth stimulation of other bacteria, growth curves were repeated in non- enriched BHI media. The doubling times of P. gingivalis ATCC 33277 and W83 in the presence of 20 µM DPIX doubled and the final optical density was close to double the DMSO control treatment by 96 hours (Fig 3.1 and Table 3.2). Growth of Bacteroides spp. (n=5) and Prevotella spp. (n=2) also doubled when 20 µM DPIX was supplemented into media (average 1.9-fold) (Fig. 3.1 and Table 3.2). Surprisingly, relative to the DMSO control, the doubling time decreased 1.5-fold in Bacteroides spp. grown with DPIX amide linked-nitroimidazole adduct P5 at 20 µM (Fig. 3.1, Table 3.1 and Table 3.2). At 20 µM, P5 is more than 3-fold the concentration of metronidazole that suppresses growth of these Bacteroides spp. (Fig. 3.1, Table 3.1 and Table 3.2).

101 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Bt 10 Bf DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (8µM) DPIX+M (5µM)

0.01 M (6µM) M (5µM) OD(600nm) P5 (20µM) P5 (20µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100 Time (h) Time (h)

Bv 10 B DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (5µM) DPIX+M (5µM) M (5µM) M (4µM) 0.01 OD(600nm) P5 (20µM) P5 (20µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100

Bu 10 Pw DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (7µM) DPIX+M (0.5µM) M (6µM) M (0.5µM)

0.01 OD(600nm) P5 (20µM) P5 (0.5µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100

Pg 10 Pm DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (0.5µM) DPIX+M (5µM) M (0.5µM) M (2µM) 0.01 OD(600nm) P5 (0.5µM) P5 (1µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100

Pi 10 DMSO 1 DPIX (20µM) 0.1 DPIX+M (3µM) M (2µM) 0.01 OD(600nm) P5 (2µM) 0.001 0 20 40 60 80 100

Figure 3.1: Growth of Bacteroidetes in BHI medium. B. thetaiotaomicron VPI 8482 (‘Bt’), B. fragilis (‘Bf’), B. vulgatus (‘Bv’), B. thetaiotaomicron ATCC 29741 (‘B’), B. uniformis (‘Bu’), Porphyromonas gingivalis W83 (‘Pw’), P. gingivalis ATCC 33277 (‘Pg’), Prevotella melaninogenica (‘Pm’) and P. intermedia (‘Pi’).

102 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Further, P5 decreased the doubling time of Prevotella spp. two-fold when titrated below inhibitory concentrations. Since metronidazole in the presence of DPIX killed Bacteroides spp., the responses of other major bacterial phyla of the human gastrointestinal tract were surveyed. P5 and DPIX were assayed at 20 µM, which approaches the solubility limits in BHI growth medium. In contrast to the Bacteroidetes, DPIX or P5 did not support the growth of other major clades of the human gut, including representatives from the Firmicutes (n=8), Fusobacteria (n=1), Actinobacteria (n=3) and Proteobacteria (n=2) (Fig. 3.2, Fig. 3.3 and summary in Table 3.1). Growth of facultative bacteria in these clades in the presence of metronidazole was similar to DMSO carrier control (Table 3.2). Nanaerobes within the Clostridia class (n=5), were consistently more sensitive to P5 (MIC: 1-2 µM, mean 1.6 µM) than metronidazole (MIC: 3-11, µM mean 5.6 µM; see Fig. 3.2 and Table 3.1), with the exception of Veillonella parvula, which was not killed by P5 at 20 µM. Fusobacterium nucleatum grew in the presence of 20µM adduct, 10-fold the concentration of metronidazole required to kill this bacterium.

103 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Vp 10 Pm DMSO DMSO 1 DPIX (20µM) DPIX (12µM) 0.1 DPIX+M (9µM) DPIX+M (8µM) M (9µM) M (12µM)

OD(600nm) 0.01 P5 (20µM) P5 (2µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100 Time (h) Time (h)

Cp 10 Cd DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (6µM) DPIX+M (3µM) M (6µM) M (3µM) OD(600nm) 0.01 P5 (2µM) P5 (2µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100

Fp 10 Er DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (5µM) DPIX+M (3µM)

OD(600nm) M (5µM) M (3µM) 0.01 P5 (2µM) P5 (1µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100

Fn 10 DMSO 1 DPIX (20µM) 0.1 DPIX+M (2µM)

OD(600nm) 0.01 M (2µM) P5 (20µM) 0.001 0 20 40 60 80 100

Figure 3.2: Growth of Firmicutes and Fusobacteria in BHI medium. Veillonella parvula (‘Vp’), Parvimonas micra (‘Pm’), Clostridium perfringens (‘Cp’), C. difficile (‘Cd’), Faecalibacterium prausnitzii (‘Fp’), Eubacterium rectale (‘Er’) and Fusobacterium nucleatum (‘Fn’).

104 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Lr Sm 10 DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (20µM) DPIX+M (20µM) M (20µM)

OD(600nm) M (20µM) 0.01 P5 (20µM) P5 (20µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100 Time (h) Time (h) Ao Bd 10 DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (20µM) DPIX+M (20µM) M (20µM) M (20µM) OD(600nm) 0.01 P5 (20µM) P5 (20µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100

S An 10 DMSO DMSO 1 DPIX (20µM) DPIX (20µM) 0.1 DPIX+M (20µM) DPIX+M (20µM) M (20µM)

OD(600nm) M (20µM) 0.01 P5 (20µM) P5 (20µM) 0.001 0 20 40 60 80 100 0 20 40 60 80 100

Ec 10 DMSO

1 DPIX (20µM)

0.1 DPIX+M (20µM) M (20µM)

OD(600nm) 0.01 P5 (20µM) 0.001 0 20 40 60 80 100

Figure 3.3: Growth of Proteobacteria and Actinobacteria in BHI medium. (‘Sm’), Lactobacillus rhamnosus (‘Lr’), Bifidobacterium dentium (‘Bd’), Actinomyces odontolyticus (‘Ao’), Actinomyces naeslundii (‘An’), Serratia marcescens (‘S’) and Escherichia coli (‘Ec’).

105 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Table 3.1: Summary of minimum inhibitory concentrations (in µM) or growth stimulation (+) of bacterial strains in BHI growth medium with metronidazole (‘M’), DPIX and P5. The absence of growth stimulation is indicated by (-). Below inhibitory concentrations, P5 stimulated growth of bacteria marked with an asterisk (*).

Physiology Phylum Family Strain M DPIX P5 Bacteroides fragilis ATCC 25285 <5 + + Bacteroides thetaiotaomicron VPI-5482 (ATCC 29148) <7 + + Bacteroidaceae Bacteroides thetaiotaomicron ATCC 29741 <7 + + Bacteroides vulgatus ATCC 8482 <5 + + Bacteroidetes Bacteroides uniformis ATCC 8492 <7 + + Porphyromonas gingivalis W83 <0.5 + <0.5 Porphyromonadaceae Porphyromonas gingivalis ATCC 33277 <0.5 + <0.5 <2 + <3* Prevotellaceae Prevotella melaninogenica ATCC 25845 Nanaerobic Prevotella Intermedia ATCC 25611 <1 + <3* Fusobacteria Fusobacteriaceae Fusobacterium nucleatum ATCC 25586 <2 - >20 Veillonellacaeae Veilonella parvula ATCC 10790 <9 - >20 Parvimonas micra ATCC 33270 <11 - <1 Clostridium perfringens ATCC 13124 <6 - <2 Clostridiaceae Clostridium difficile VPI-10463 (ATCC 43255) <3 - <2 Firmicutes Faecalibacterium prausnitzii A2-165 (DSM 17677) <5 - <2 Eubacterium rectale A1-86 (DSM 17629) <3 - <1 Streptococcus mutans ATCC 25175 >20 - >20 Lactobacillaceae Lactobacillus rhamnosus ATCC 7469 >20 - >20 Bifidobacteriaceae Bifidobacterium dentium ATCC 27534 >20 - >20 Facultative Actinobacteria Actinomyces odontolyticus ATCC 17982 >20 - >20 Actinomycetaceae Actinomyces naeslundii ATCC 12104 >20 - >20 Serratia marcescens ATCC 274 >20 - >20 Proteobacteria Enterobacteriaceae Escherichia coli ATCC 25922 >20 - >20

Next an ex vivo faecal growth curve model was used to determine if the results obtained in single species growth curves held in complex bacterial populations. In this model, bacterial inocula were diluted to be within concentrations of metronidazole and P5 known to be lethal, as determined in Chapter II in enriched BHI growth media and confirmed in BHI growth media (Fig 3.4). Changes in viable Bacteroides were assessed with BHI medium containing gentamicin and bile (‘BHIBG’), a selective medium for Bacteroides, that are resistant to these molecules (Livingston et al., 1978). On BHIBG, Bacteroides hydrolyse esculin forming a dark precipitate with iron-citrate surrounding colonies (Livingston et al., 1978). By 10 hours, metronidazole and metronidazole with DPIX eliminated 97.9% and 97.5% respectively of inoculated Bacteroides, confirming escape of nitroimidazole-mediated death occurred only when DPIX was tethered to a nitroimidazole, as in P5 (P<0.05, Fig. 3.5A). Also after a 10 hour incubation with P5, viable Bacteroides were around double (1.9 fold) DMSO control and were higher than DPIX (17.6%), indicating not only did P5 enhance growth, but that depletion of other clades (such as Clostridia) advantaged the Bacteroides (Fig. 3.5A). Illumina sequencing of the 16S rDNA gene (V4 region) confirmed that by 10 hours, the Bacteroides population were enhanced by P5 (2.4 fold increase over DMSO, OTU1 P=0.0009 and

106 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Table 3.2: Summary of bacterial doubling times (TD) of strains in BHI growth media with metronidazole (‘M’), DPIX and P5. Doubling times are calculated in 20 µM DPIX, metronidazole and P5 except for Parvimonas micra and Prevotella spp. which were grown in the presence of 12 µM and 2 µM DPIX respectively.

-1 TD (min ) Physiology Phylum Family Strain DMSO DPIX M P5 Bacteroides fragilis ATCC 25285 128 64 NA 65 Bacteroides thetaiotaomicron VPI-5482 155 78 NA 77 Bacteroidaceae Bacteroides thetaiotaomicron ATCC 29741 111 72 NA 74 Bacteroides vulgatus ATCC 8482 138 74 NA 120 Bacteroidetes Bacteroides uniformis ATCC 8492 145 68 NA 139 Porphyromonas gingivalis W83 295 162 NA NA Porphyromonadaceae Porphyromonas gingivalis ATCC 33277 398 198 NA NA Prevotella melaninogenica ATCC 25845 345 139 NA 155 Nanaerobic Prevotellaceae Prevotella Intermedia ATCC 25611 332 206 NA 201 Fusobacteria Fusobacteriaceae Fusobacterium nucleatum ATCC 25586 110 103 NA 88 Veillonellacaeae Veilonella parvula ATCC 10790 205 218 NA 226 Parvimonas micra ATCC 33270 308 264 NA NA Clostridium perfringens ATCC 13124 29 35 NA NA Clostridiaceae Clostridium difficile VPI-10463 50 51 NA NA Firmicutes Faecalibacterium prausnitzii A2-165 97 95 NA NA Eubacterium rectale A1-86 92 95 NA NA Streptococcus mutans ATCC 25175 70 73 73 108 Lactobacillaceae Lactobacillus rhamnosus ATCC 7469 112 92 86 102 Bifidobacteriaceae Bifidobacterium dentium ATCC 27534 114 155 155 155 Actinobacteria Actinomyces odontolyticus ATCC 17982 128 180 128 272 Facultative Actinomycetaceae Actinomyces naeslundii ATCC 12104 177 227 156 156 Serratia marcescens ATCC 274 46 47 46 47 Proteobacteria Enterobacteriaceae Escherichia coli ATCC 25922 41 41 41 41

])

1 8 - DMSO 6

[cfu ml DPIX 10 4 DPIX + M M 2 P5

0 Viable Count (Log ViableCount 0 1 2 3 4 5 Time (h)

Figure 3.4: Kill curve of B. thetaiotaomicron VPI 5482 with 20 µM treatments or DMSO (1% v/v). B. thetaiotaomicron (≈5×105 cfu ml-1) was introduced into each treatment at time O. The doubling time for B. thetaiotaomicron in DMSO, DPIX and P5 were 136.4, 85.2 and 77.4 min-1 respectively. The lower limit of detection was 20 cfu mL-1.

107 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

OTU4 P=0.0002) and that metronidazole and metronidazole plus DPIX eliminated the majority (97.7% and 98.2% respectively) of inoculated Bacteroides (Fig. 3.5B and Table 3.3). In contrast, Clostridia were eliminated by metronidazole and by P5, with the exception of Lachnospiraceae (predominantly OTU5, P=0.007, Fig. 3B and Table 3.3). Interestingly, overgrowth of bacteria insensitive to metronidazole was reduced in the P5 treatments. For example, Streptococci (OTU2) dominated the metronidazole treatment (68.4%, P=0.0002), a 5-fold increase over DMSO control (12.9%), whereas P5 resulted in a 2-fold increase (25.5%, ns, see also Table 3.3).

A B *

100000 * Bacteroidaceae M 10000 DMSO 1000 Enterobacteriaceae P5 100 Lachnospiraceae

10 Streptococcaceae

Bacteroides (%) Bacteroides 1 Enterococcaceae

Bacteroides/Innoculated Bacteroides/Innoculated 0 DMSO DPIX DPM M P5 0 20 40 60 80 100 Abundance (%) Fig. 3.5: Ex vivo distal colon culture model with (A) Viable Bacteroides recovered at five hours (blue) and 10 hours (grey) as a percent of inoculated Bacteroides (‘100%’ at 0 hours), (mean±SEM, n=4). (B) Major changes in bacterial families based on 16S rDNA sequencing at 10 hours (mean±SEM, n=4). Bacterial families with >±5% variance between metronidazole and P5 treatments are shown (see Table 3.3 for full 16S rDNA sequencing results).

108 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Table 3.3: Ex vivo faecal culture model Illumina sequencing of the 16S rDNA (V4 region) of 10,864,916 quality filtered sequences. The absolute total was calculated from all quality filtered (10,864,916) sequences. The average (‘Av.’) and SEM were also calculated within each of the seven treatment groups from four experiments. Phylum summing less than 0.5% of the total sequences are summarised as ‘Other’ including: Unclassified (8661 sequences), Fusobacteria (107 sequences), Planctomycetes (6 sequences), Acidobacteria (3 sequences), OD1 (3 sequences), Chloroflexi (1 sequence) and OP11 (1 sequence).

Inocula DMSO DPIX DPM M P5 Taxa Av. SEM Av. SEM Av. SEM Av. SEM Av. SEM Av. SEM Bacteroidetes 27.0 0.7 13.6 1.9 16.1 2.1 0.6 0.3 0.8 0.5 32.7 3.5 Bacteroidaceae 23.6 0.6 13.1 1.8 15.3 1.8 0.5 0.3 0.7 0.5 31.7 3.5 Porphyromonadaceae 0.9 0.2 0.4 0.1 0.7 0.3 0.0 0.0 0.0 0.0 0.9 0.2 Rikenellaceae 1.8 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Firmicutes 63.8 1.0 84.0 2.5 81.3 3.2 76.3 11.2 71.5 10.2 55.7 8.0 Bacilli 0.6 0.0 13.5 3.4 10.3 2.4 73.9 10.9 70.0 9.8 33.0 4.4 Enterococcaceae 0.0 0.0 0.0 0.0 0.0 0.0 1.1 0.9 0.3 0.2 7.4 7.2 Streptococcaceae 0.4 0.0 12.9 3.5 9.6 2.5 72.5 11.0 69.7 9.8 25.5 7.9 Turicibacteraceae 0.2 0.1 0.5 0.3 0.6 0.3 0.0 0.0 0.0 0.0 0.0 0.0 Clostridia 59.0 1.5 63.4 4.5 64.1 4.2 2.0 1.0 1.2 0.4 22.3 6.9 Unclassified family 1.0 0.2 8.0 5.7 7.2 6.0 0.4 0.3 0.1 0.1 0.0 0.0 Clostridiaceae 1.1 0.4 2.2 1.0 1.2 0.8 0.1 0.0 0.0 0.0 0.0 0.0 Lachnospiraceae 30.5 1.9 39.8 2.0 43.5 5.1 1.2 0.5 0.8 0.3 20.1 7.0 Ruminococcaceae 25.4 3.6 12.7 1.1 11.5 1.9 0.4 0.1 0.2 0.1 0.3 0.1 Veillonellaceae 1.0 0.1 0.5 0.1 0.5 0.1 0.0 0.0 0.0 0.0 1.9 0.3 Erysipelotrichi 1.5 0.3 7.0 2.6 6.8 2.1 0.1 0.0 0.0 0.0 0.4 0.2 Erysipelotrichaceae 1.5 0.3 7.0 2.6 6.8 2.1 0.1 0.0 0.0 0.0 0.4 0.2 Unclassified class 2.6 0.9 0.0 0.0 0.0 0.0 0.3 0.2 0.2 0.1 0.1 0.0

Proteobacteria 7.1 0.4 0.7 0.4 1.3 0.6 21.2 11.6 26.3 10.2 10.9 5.4 1.0 0.0 0.1 0.0 0.2 0.1 5.0 1.4 4.9 1.4 1.1 0.3 Alcaligenaceae 1.0 0.0 0.1 0.0 0.2 0.1 5.0 1.4 4.9 1.4 1.1 0.3 Deltaproteobacteria 1.1 0.3 0.1 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.2 0.1 Desulfovibrionaceae 1.1 0.3 0.1 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.2 0.1 Gammaproteobacteria 5.0 0.5 0.5 0.3 1.0 0.6 16.2 12.6 21.4 11.3 9.6 5.4 Enterobacteriaceae 0.0 0.0 0.4 0.4 0.8 0.6 14.4 13.2 19.5 12.2 9.3 5.5 Pasteurellaceae 5.0 0.5 0.1 0.0 0.1 0.0 1.8 0.6 2.0 0.9 0.3 0.1

Actinobacteria 0.5 0.2 1.6 0.4 1.3 0.5 1.9 0.8 1.4 0.5 0.7 0.4 Bifidobacteriaceae 0.2 0.1 0.8 0.4 0.5 0.2 1.9 0.8 1.4 0.5 0.7 0.4 Coriobacteriaceae 0.3 0.2 0.8 0.6 0.7 0.6 0.0 0.0 0.0 0.0 0.0 0.0

Other 1.6 1.5 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0

3.3.2 The metabolism of P5 by Bacteroides

After establishing the growth enhancing effect of P5 for Bacteroides and Prevotella, it was hypothesised that these genera might incorporate a metabolite or metabolites of P5 into a b- cytochrome to enabling fumarate reductase activity. The homology of the fumarate reductase

109 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes operon within the Bacteroides and Prevotella genera was 87.5% (78 strains) and 75.6% (107 strains) respectively (Fig. 3.8). In B. fragilis FRD forms a three-component operon, frdC, frdA and frdB encoding the b-cytochrome, iron sulphur cluster protein and flavoprotein respectively. As previously described, doubling times for parental B. fragilis ADB77 were around 8.3 h-1 in the absence of a porphyrin source, compared to 2.3-2.4 h-1 when the growth medium was supplemented with haem, DPIX, or P5 (Fig. 3.6 and Table 3.4) (Macy et al., 1975, Baughn and Malamy, 2003). Whether grown with porphyrin supplementation or not, mutants derivatives of ADB77, which were deficient in b- cytochrome formation, B. fragilis ADB247 (ΔfrdC) and ADB260 (ΔfrdA), had similar doubling times (averaging 8.1-8.2 h-1 Fig. 3.6), similar to those reported (Macy et al., 1975, Baughn and Malamy, 2003). These results confirmed that in a defined growth medium, these genes were essential for incorporation of exogenous porphyrin into a b-cytochrome. Under identical conditions, it was previously shown that mutants complemented by vector plasmids carrying the frd operon, restored fumarate reductase activity (Baughn and Malamy, 2003).

A 1.0E+08 B C

1.0E+07 DMSO

1 - DPIX 1.0E+06

cfu cfu ml Haem 1.0E+05 P5

1.0E+04 0 10 20 30 0 10 20 30 0 10 20 30 Time (h) Time (h) Time (h) Figure 3.6: Growth of parental B. fragilis ADB77 (A) or derivative mutant strains B. fragilis ADB247 (frdC, B) and B. fragilis ADB260 (frdA, C) in defined medium. Data is from one representative experiment repeated twice. Haem, DPIX and P5 were assayed at 20 µM.

Next, the activity of FRD and SDH, using a lysate of B. thetaiotaomicron grown with P5 were compared with porphyrins that are known to form a b-cytochrome (DPIX and haem) or with no added porphyrin control (DMSO). No FRD or SDH activity were detected in a lysate of B. thetaiotaomicron grown without added porphyrin (DMSO control), indicating B. thetaiotaomicron was unable to form a detectable b-cytochrome in BHI medium (Table 3.5). In contrast, FRD activity similar in B. thetaiotaomicron lysate grown in the presence of haem or DPIX, which was approximately 0.3-fold higher than the activity of B. thetaiotaomicron grown with P5 (Table 3.5). SDH activity was similar in B. thetaiotaomicron lysate grown in the presence of haem or DPIX, which was 15-fold higher than the activity of B. thetaiotaomicron grown with P5 (Table 3.5). Next, using serial 10-fold serial dilutions of porphyrins, it was demonstrated that haem,

110 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Table 3.4: Doubling times of parental B. fragilis ADB77 or derivative mutant strains B. fragilis ADB247 and B. fragilis ADB260 in the presence and absence of porphyrins.

Td (h-1) Treatment ADB77 ADB247 ADB260 DMSO 8.3 8.1 8.1 DPIX 2.3 8.2 8.3 Haem 2.3 8.0 7.9 P5 2.4 8.7 8.3

DPIX and PPIX, all increased doubling time of B. thetaiotaomicron above 20 pM (>2.4 × 104 molecules cell-1) in BHI medium (Fig. 3.7 and Table 3.6). Above 2 nM (>2.4 × 106 molecules cell-1) P5, enhanced growth rate than DMSO control (Fig. 3.7). Growth stimulation by P5 at nanomolar concentrations is compatible with P5 forming a b-cytochrome. Finally 10-fold serial dilutions of B. thetaiotaomicron were prepared in BHI growth medium ranging from 5 × 105 cfu to 5 cfu containing 20 µM metronidazole or P5. None of the cultures of B. thetaiotaomicron grew in 20 µM metronidazole, whereas all grew in P5 (not shown). These results demonstrate that regardless of the concentration, B. thetaiotaomicron is not killed by P5. Possible interpretations of this data are considered in the discussion.

Table 3.5: Fumarate reductase and succinate dehydrogenase activity of cell free extracts of B. thetaiotaomicron grown with (DPIX, haem, P5) or without a source of porphyrin (DMSO). Error incorporates mean ±SEM from two independent experiments prepared from one cell extract. ND=no activity detected

Enzyme Activity [µmol min-1 (mg protein)-1] DMSO Haem DPIX P5 FRD ND 6.2 ±0.1 7.6 ±0.8 4.6 ±0.4 SDH ND 4.9 ±0.1 4.0 ±0.2 0.3 ±0.1

Table 3.6: Doubling time of B. thetaiotaomicron VPI 5482 titrated with 0.02 nM - 20 µM treatments (equating to 2.4×103-10 molecules per inoculated bacterium). The concentration of treatments is indicated in nmoles. The doubling time of B. thetaiotaomicron in DMSO was 116.1 min-1. Treatment Concentration (nM) Treatment 20000 2000 200 20 2 0.2 0.02 DPIX 69.4 66.3 68.3 66.3 65.4 78.8 116.1 PPIX 67.3 64.6 65.4 67.3 70.6 83.7 116.1 Haem 64.6 63.8 68.3 65.4 68.3 86.7 116.1 P5 75.0 63.8 70.6 69.4 86.7 100.0 116.1

111 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

10 20000 1 2000 200 0.1 20 2 0.01 OD OD (600nm) 0.2 DPIX 0.02 PPIX 0.001 0 50 100 0 50 100

10 20000 1 2000 200 0.1 20 2 0.01 OD OD (600nm) 0.2 Haem 0.02 P5 0.001 0 50 100 0 50 100

10 M 20000 1 2000 0.1 200

0.01 20 OD OD (600nm) DMSO 0.001 0 50 100 0 50 100 Time (h) Time (h)

Figure 3.8: Growth of B. thetaiotaomicron VPI 5482 titrated with 0.02 nM - 20 µM treatments (equating to 2.4×103-10 molecules per inoculated bacterium) or with DMSO. The concentration of treatments is indicated in nmoles. Concentrations below 2 pM for P5 or below 20 pM for metronidazole (‘M’) are not shown because growth in these treatments were identical to DMSO treatment. Note that from 20 pM – 20 µM growth of B. thetaiotaomicron in DPIX, PPIX and haem (haem) was saturated. Note that 20 µM treatment was corrected by blanks and that this treatment was otherwise identical to the 2 µM treatment.

112 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Figure 3.8: Homology searches (BLASTP) using B. thetaiotaomicron VPI 5482 FrdC (D), FrdA (C) and FrdB (B) fumarate reductase complex proteins. The average homology from gastrointestinal bacteria for the entire fumarate reductase complex is also summarised (A, hits ≥60% are shown). When genus identification was not available, next highest available taxonomic rank was presented.

3.3.3 The Structural Metabolism of P5 by Bacteroides thetaiotaomicron

Sodium dithionite is commonly used to chemically inactivate nitroimidazoles, causing nitro group reduction and heterocycle fragmentation. To validate this method, the growth of C. difficile (Fig. 3.9) and B. thetaiotaomicron (Fig. 3.10) was compared with metronidazole and P5 alone or with these compounds after chemical reduction by pre-incubation with sodium dithionite. C. difficile was unable to grow in the presence of lethal concentrations of metronidazole or P5 (10 µM, Fig 3.9).

113 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

When metronidazole and P5 (10 µM) were pre-incubated with sodium dithionite, growth of C. difficile was identical to the DMSO treatment (Fig 3.9) validating this method. Similar results were obtained for B. thetaiotaomicron with a lethal concentration of metronidazole (20 µM). Pre- treatment of metronidazole with sodium dithionite permitted identical growth to the DMSO control, whereas in the absence of sodium dithionite, growth was totally inhibited (Fig. 3.10). Maximal growth of B. thetaiotaomicron occurred with P5 and with dithionite treated P5 (Fig. 3.10). There was a slight increase in the growth of B. thetaiotaomicron (Fig. 3.10). Interpretations for this result are discussed below. Altogether, these results provide strong evidence toxicity associated with P5 is related to the nitro group reduction.

A 10 B C

1 DMSO DMSO- M P5 0.1 DPIX M- P5-

0.01 DPIX- OD(600nm) 0.001 0 20 40 0 20 40 0 20 40 Time (h) Time (h) Time (h)

Figure 3.9: Growth of C. difficile in the presence of DMSO, DPIX (10 µM), metronidazole (‘M’, 10 µM) or P5 (10 µM). Treatments with a minus sign (-) were chemically reduced with sodium dithionite as described in methods before introduction into growth medium. Doubling times of DMSO or reductively activated metronidazole and P5 were 53, 54 and 54 min-1 respectively.

A 10 B C

1 DMSO M DMSO- P5 0.1 DPIX M- P5- DPIX-

OD OD (600nm) 0.01

0.001 0 20 40 0 20 40 0 20 40 Time (h) Time (h) Time (h)

Figure 3.10: Growth of B. thetaiotaomicron in the presence of DMSO, DPIX (20 µM), metronidazole (‘M’, 20 µM) or P5 (20 µM). Treatments with a minus sign (-) were incubated with sodium dithionite (chemically reduced as described in text) before introduction into growth media. Doubling times of DMSO, P5 or reductively activated metronidazole and P5 were 130, 74, 127 and 69 min-1 respectively.

114 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

As described in the thesis introduction, the nitro group appended to a nitroimidazole may be reductively activated, leading to ring fragmentation and cytotoxicity. A possible alternative fate is that the nim system reductively inactivates the nitro group, forming an amino derivative. Two controls were used to help decipher how B. thetaiotaomicron was metabolising P5. P5 chemically reduced with dithionite was used as the first control, to model the biological reductive activation of P5, leading to ring fragmentation. The metabolism of DPIX by B. thetaiotaomicron was also examined. This control was included to help determine if P5 or derivatives were metallated and also whether metabolite/s of P5 were further processed by B. thetaiotaomicron. Since tetrapyrroles exhibit characteristic absorbance around 400 nm and also in the q-band region, ions at retention times within peaks that had these unique signatures were interrogated. In a previous study, low resolution LC-MS technology demonstrated metronidazole fragments into several species [at least 26 products (Knox et al., 1983)]. Data obtained with a high resoluton LC-MS was more complex (>100 ions). In order to stratify ions, structures consistent with the possible fragmentation of P5 were solved (Figs. 3.12A and 3.12B), also see summary in Fig. 3.15). Importantly, an ion with [m/z + H+] = 510.2502 (which was similar to the calculated ion of 510.2505,

+ for C30H31N5O3 + H ), was consistent with a similar structure to DPIX (Fig. 3.12C). This ion was detected with a relative abundance of approximately 0.04% (Fig. 3.12C and also Fig. 3.17). In an examination of DPIX metabolised B. thetaiotaomicron, hundreds of ions were also observed. Three main ions, consistent with DPIX or the addition of ethyl and methyl groups to DPIX, were detected (Fig. 3.13). The genome of B. thetaiotaomicron encodes three candidate enzymes that may add methyl or ethyl groups including tetrapyrrole methylase family-like protein (BT_2273), tetrapyrrole methylase family protein/MazG (BT_4350) and magnesium protoporphyrin O- methyltransferase (BT_4355). In DPIX metabolised by B. thetaiotoamicron, no ions consistent with iron metallated DPIX [m/z + H+] = 565.1538) or magnesium metallated DPIX ([m/z + H+] = 533.2039) were detected. Possible explanations for the absence of a metallated derivative of DPIX are discussed below. Notably, these data further support the notion that B. thetaiotaomicron is able to metabolise abiotic porphyrins. Next, using dithionite treated P5 and B. thetaiotaomicron metabolised-DPIX controls to inform analyses, the metabolism of P5 by B. thetaiotaomicron following incubation in defined growth medium was examined. Many of the ions detected at these retention times in B. thetaiotaomicron metabolised P5 were shared in dithionite-incubated P5 treatment (see examples of these ions and possible structures in (Figs. 3.14A - 3.14B). No ions were detected consistent with P5 ([m/z + H+] = 663.3043), indicating all of the P5 was taken up and metabolised by B. thetaiotaomicron. Secondly, no ion consistent with the amino derivative of P5 ([m/z + H+ =

115 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

A

B

C

D

E

F

G

Figure 3.11: HPLC traces with absorbance at 400 nm on the y-axis and retention time on the x-axis. A representative blank (A) is shown for samples (B), (C) and (D). A representative blank (E) pertains to remaining samples. P5 (B) is shown for reference, and (C) is P5 incubated (chemically reduced) with sodium dithionite. (F) and (G) are ethyl acetate extract and final HCl extract from B. thetaiotaomicron–metabolised P5 after 48 hour incubation respectively. The final HCl layer from B. thetaiotaomicron–metabolised DPIX after 48 hour incubation is shown as a reference (D).

+ 633.3301, for C36H40N8O3 + H ) was detected, indicating the absence of a nim system. An ion with [m/z + H+] = 510.2502, which was similar to ion detected in the dithionite-P5 treatment of abundance ~0.01%, was consistent with a derivative with a similar structure to DPIX (see summary Fig. 3.15 and Fig. 3.17). This ion is a possible candidate for b-cytochrome formation. Also, a

+ + prominent ion of around 10% ‘abundance’ ([m/z + H = 553.2919, calculated for C32H36N6O3 + H , Fig. 3.14A) was also shared in P5-dithionite control. This ion also contained a possible methyl

+ + derivative ([m/z + H = 567.3080, calculated for C33H38N6O3 + H , Fig. 3.14A) of abundance ~1%.

116 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

6 #3742-4247 RT: 11.72-13.03 AV: 73 NL: 4.09E7 T: FTMS + p ESI Full lock ms [150.00-1500.00] 319.6580 100 ^

90 510.2502

80 553.2926

594.3190 70 ^ 277.1498 622.3138 60 638.3087 50 ^

40

RelativeAbundance 638.3088 30

20 594.3190 10 365.1444 660.2905 390.8607 729.2813 526.2452 786.3033 0 200 300 400 500 600 700 800 900 m/z

Fig. 3.12A: LC-MS from sodium dithionite incubated P5 (chemically reduced) from 200-900 m/z. Where possible, formula and structures consistent with major detected ions (boxed) are shown. ^Possible doubly charged ion, often formed from the parental ion of tetrapyrroles.

6 #3743-4242 RT: 11.72-13.03 AV: 73 NL: 5.06E6 T: FTMS + p ESI Full lock ms [150.00-1500.00] 594.3190 100

90 553.2926 594.3190 80 553.2925 622.3138

70 638.3087 622.3138 60

50 580.3036 40 597.3282

RelativeAbundance 585.7875 30

20

10 580.3025 512.7373 526.2452 537.1665 565.2923 615.2056 0 500 520 540 560 580 600 620 m/z

Fig. 3.12B: LC-MS sub-set (of Fig 3.12A) from sodium dithionite incubated P5 from 500-650 m/z. Formula and structures consistent with major detected ions (boxed) are shown. Note the new scale of y-axis.

117 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

6 #3743-4242 RT: 11.72-13.03 AV: 73 NL: 1.63E4 T: FTMS + p ESI Full lock ms [150.00-1500.00] 512.7373 16000 15000 14000 13000

12000 519.1391 11000 513.2388 10000 9000 510.2502 8000 7000

6000 RelativeAbundance 5000 4000 519.4898 520.7348 3000 510.2501 512.2028 2000 513.7400 509.3093 510.7059 1000 515.2968 518.3321 0 510 512 514 516 518 520 m/z

Fig. 3.12C: LC-MS sub-set (of Fig 3.12B) from sodium dithionite incubated P5 from 500-520 m/z. Formula and structures consistent with major detected ions (boxed) are shown. Note the new scale of y-axis.

1 #1873-2456 RT: 5.99-7.57 AV: 84 NL: 7.66E6 T: FTMS + p ESI Full lock ms [150.00-1500.00] 539.2654 100

511.2341 567.2966 90

80 279.1591 511.2345 70 ^ ^ 539.2658

60 567.2971

50

40 301.1410 RelativeAbundance 30 ^ 20 354.2850 464.3730 408.2299 10 790.5594 615.2967 720.5539 844.6061 0 200 300 400 500 600 700 800 900 m/z

Fig. 3.13: LC-MS of an extract of DPIX metabolised by B. thetaiotaomicron from 200-900 m/z. Formula and structures consistent with major detected ions (boxed) are shown ^Possible double- charged ions are indicated.

118 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

12 #4715-4995 RT: 13.01-13.69 AV: 41 NL: 1.38E8 T: FTMS + p ESI Full lock ms [150.00-1500.00] 500.3793 100

90

80 510.2502

553.2926 70 ^ 277.1498 594.3190 60 638.3087 50

40 Relative Abundance Relative 30 ^ 20 297.6630 377.2356 483.3530 553.2923 10 225.1961 342.6739 677.1917 416.2333 594.3187 747.5077 794.2606 0 200 300 400 500 600 700 800 900 m/z

Fig. 3.14A: LC-MS from an ethyl acetate extract of P5 metabolised by B. thetaiotaomicron from 200- 900 m/z. Where possible, formula and structures consistent with major detected ions (boxed) are shown. ^Possible doubly charged ion, often formed from the parental ion of tetrapyrroles.

12 #4713-4995 RT: 13.01-13.69 AV: 41 NL: 2.88E7 T: FTMS + p ESI Full lock ms [150.00-1500.00] 100

553.2926 90 553.2926

567.3080 594.3190 80 638.3087 70

60

50 553.2923

40 677.1917 Relative Abundance Relative 30

20 528.4106 563.1486 594.3187 10 521.3085 684.3409 535.2818 567.3078 604.2115 639.3292 661.2331 699.2125 0 520 540 560 580 600 620 640 660 680 700 m/z

Fig. 3.14B: LC-MS sub-set (500-700 m/z) from an ethyl acetate extract of P5 metabolised by B. thetaiotaomicron (Fig. 3.14A). Formula and structures consistent with major detected ions (boxed) are shown. Note the new scale of axes.

119 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Various searches for m/z consistent with iron or magnesium metalation of these ions failed to identify candidates. Again, enzymes encoded by B. thetaiotaomicron (BT_2273, BT_4350 and BT_4355) may enable this metabolism, supporting the notion that this bacterium may metabolise abiotic porphyrins. Mass spectrometry is further discussed below and the major ions with possible structures are summarised in Fig. 3.15. Since P5 was around 100-fold less stimulatory than PPIX, haem and DPIX in growth curve titrations, the possibility that only selected metabolites of P5 are used to form a b-cytochrome is discussed below. Together with suppression of growth of mutants, these experiments disclosed that P5 was reduced by Bacteroides spp. and confirm P5 is a prodrug, exerting cytotoxic activity for C. difficile by nitro group reduction.

120 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

Figure 3.15: Potential structures of dithionite reduced P5, B. thetaiotaomicron metabolised P5 and DPIX derived from high resolution mass spectrometry. Mass spectrometry is not a structural technique, so some of the functional groups may be oriented slightly differently than proposed, such as methyl and ethyl additions to carboxylic acids of DPIX. Dotted lines indicate structures consistent with distinct ions, rather than fragments of parent ions.

121 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

3.3.3 Antibiotic Susceptibility and nim Gene Presence

In growth curves, susceptible Clostridia were approximately 3-fold more susceptible to P5 than metronidazole. A possible explanation for increased sensitivity to P5 is that these Clostridia encode a nim derivative with different specificities for P5 and metronidazole. The expression of the nim genes may then enable more efficient reductive inactivation of metronidazole than the nitro group appended to P5. Conversely, expression of nim genes may also enable the Bacteroides to overcome nitro group reduction of P5 and not of metronidazole. In a PCR screen of 23 bacterial strains used in Chapter III for nimA-I, no product was detected (data not shown). A second PCR screen for nimJ also failed to detect amplicons (data not shown).

To determine if differences in susceptibility more broadly reflect clinical isolates of C. difficile, 100 strains collected in a clinical setting were evaluated. The MIC of 100 strains of C. difficile ranged from 0.8-50 µM for metronidazole and P5. The average MIC for metronidazole (8.7±0.9 µM) was slightly higher than P5 (7.1±0.7 µM, Fig. 3.16). Based on EUCAST breakpoints, 10% of the strains (n=10) were considered resistant to metronidazole (>2 µg mL-1) growing in ≥23.4 µM metronidazole (EUCAST, 2016). P5 does not have an established EUCAST breakpoint, although at equimolar concentrations, 8% (n=8) of the strains were resistant to P5, growing in ≥23.4 µM P5. Strains within ribotypes 49, 6, 43, 103 and 17, classified resistant to metronidazole (MIC ≥23.4 µM), were 7.9, 5.0, 3.6, 2.0 and 1.8-fold more susceptible to P5 respectively (Fig. 3.16 and Table 3.7). One strain identified as ribotype 31 was two-fold more resistant to P5 than metronidazole (Table 3.7).

DNA was extracted from the strains of C. difficile, yielding between 10-70 ng µL-1 genomic DNA per extract. PCR amplification with oligonucleotide primers nimA-H and nimJ yielded no PCR products (data not shown). Having found that the nim genes were not present, it was concluded that increased sensitivity to P5 was caused by other factors than the nim genes. This is further elaborated in the discussion.

122 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

A B 50 40 30 20 M

Frequency 10 P5 0 0.8 1.6 3.1 6.3 12.5 25.0 50.0 Concentration (µM)

Figure 3.16: MIC of 100 C. difficile strains grown in BHI medium.

Table 3.7: the MIC of 100 C. difficile strains within each PCR-ribotype group.

Average MIC (µM) MIC Fold Change PCR-Ribotype Strains (n) Metronidazole P5 (Metronidazole/P5) 1 2 9.4 4.7 2.0 2 21 5.7 5.0 1.1 6 2 31.3 6.3 5.0 10 1 12.5 6.3 2.0 12 8 5.9 6.1 1.0 14 18 7.3 7.4 1.0 17 2 28.1 15.6 1.8 18 6 9.9 8.6 1.2 20 6 9.9 12.5 0.8 27 4 3.9 4.3 0.9 31 1 25.0 50.0 0.5 43 2 28.1 7.8 3.6 49 1 25.0 3.1 8.0 51 2 4.7 4.7 1.0 56 1 3.1 1.6 2.0 70 6 9.4 6.0 1.6 76 1 12.5 12.5 1.0 78 4 5.1 4.9 1.0 103 1 25.0 12.5 2.0 106 1 12.5 12.5 1.0 126 1 1.6 3.1 0.5 220 4 5.1 3.5 1.4 251 3 4.2 7.3 0.6 404 1 6.3 6.3 1.0 412 1 6.3 3.1 2.0

123 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

3.4 DISCUSSION

P5 was demonstrated to stimulate growth of Bacteroides spp. and below minimal inhibitory concentrations, Prevotella spp. In contrast to these genera of the Bacteroidetes, DPIX or P5 did not support the growth of other major clades of the human gut, including representatives from the Firmicutes, Fusobacteria, Actinobacteria and Proteobacteria. Nanaerobes within the Clostridia class, were consistently more sensitive to P5 than metronidazole. The possibility that the nim genes were inactivating the nitro-group to the amino derivative is unlikely, since nim genes were not detected in any of the 23 strains bacteria used in Chapter III or in 100 clinical strains of C. difficile. Metronidazole resistance in clinical isolates of C. difficile is not well understood and problematically strains may revert from resistant to susceptible after freeze-thawing (Lynch et al., 2013). Several investigators have failed to identify nim genes from clinical strains of C. difficile (Meggersee and Abratt, 2014, Meggersee and Abratt, 2015), Bacteroides spp. (Eitel et al., 2013, Kullin et al., 2016) or from complex bacterial populations of the oral cavity (Koukos et al., 2016). Many investigators have concluded the presence of nim genes does not necessarily convey resistance to metronidazole (Schaumann et al., 2005, Katsandri et al., 2006, Leitsch et al., 2014, Meggersee and Abratt, 2015, Kullin et al., 2016). There are several alternative explanations as to the increased sensitivity to P5. In C. difficile, mutations in coproporphyrinogen III oxidase (hemN) are associated with metronidazole resistance (Lynch, 2013). Clostridia do not use protoporphyrin, rather these bacteria oxidise coproporphyinogen III, forming coproporphyrin III, before iron is inserted and carboxylation to generate haem b [(Dailey, 2015), also see thesis section 1.3.1 for further details). It is possible P5 or metabolites interfere with iron-metalation or another aspect of porphyrin biosynthesis in C. difficile, accounting for increased sensitivity to P5 in respect to metronidazole. It is also possible P5 or metabolites modulate the regulation of hemN, inducing a change in the physiological response of C. difficile to P5, for example by increased rate of uptake. It is also possible the nitro group of P5 is more efficiently reductively activated by C. difficile. In defined medium, doubling times of B. fragilis were 3-4 fold lower when supplemented with a porphyrin source, including P5. Doubling times of derivative mutants, defective in b- cytochrome formation, were similar regardless of porphyrin supplementation. The FRD activity from a lysate of B. thetaiotaomicron grown with P5 was around 66% the levels detected from a lysate of haem and DPIX. Also, growth of Prevotella spp. was enhanced when P5 was titrated below minimal inhibitory concentrations. The homology of the fumarate reductase (FrdCAB) was similar between the Bacteroides and Prevotella clades. Together, these results indicate P5 is metabolised by Bacteroides spp. and Prevotella spp. with at least one metabolite of P5 forming a functional b- cytochrome.

124 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

The surprising survival of Bacteroides challenged with P5 could be ascribed to enhanced regeneration of ATP and NAD+ through FRD activity. However, B. fragilis mutants, unable to derive ATP or NAD+ from P5, but susceptible to reductive activation of P5, were not killed by P5. The increased doubling times of Bacteroides spp. (B. vulgatus and B. uniformis) grown with P5 supplementation rather than DPIX, together with a slight reduction in doubling time in the presence of chemically reduced P5 (inactive) versus P5 (active nitroimidazole) all confirm that reductive activation of the nitro group on P5 is a metabolic challenge for Bacteroides. Nevertheless, P5 supplementation consistently decreased the doubling time Bacteroides by 2-3 fold, indicating metabolic derivatives of P5 may be suitable for incorporation into the b-cytochrome. Since increasing the relative concentration of metronidazole to B. thetaiotaomicron killed this bacterium, though P5 did not, it is possible that Bacteroides limit uptake of porphyrins, taking up sufficient P5 to form a b-cytochrome, whilst at the same time effectively limiting uptake of the toxic nitro-group. It is also possible that the rate of reductive activation of P5 may be lower than metronidazole or the formation of FRD and associated regeneration of ATP and NAD+ mitigate the toxic effects of nitroimidazole reduction. The activity of the FRD formed in B. thetaiotaomicron exposed to P5 was around half that of DPIX and haem controls. Yet in growth curves with serial dilutions of DPIX/haem/PPIX, maximum growth of B. thetaiotaomicron was reached at 20 nM, whereas 100-fold more P5 (2 µM) was required to reach maximal growth. One possibility is that only a limited range of derivatives formed during reduction of P5 form a functional b-cytochrome (Fig. 3.17). The three subunits of the magnesium chelatase were detected in almost all examined Bacteroides (results not shown). Ions consistent with magnesium metalation were not found, possibly related to removal of magnesium during the extraction procedure (Granick et al., 1972).

125 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

P5 A B Glucose 2NAD+ NAD+ 2NADH +2ATP NADH 2 PEP

NAD+ D-Lactate NADH LDH 2H+ Pyruvate Fumarate Ferredoxin P5 NADH (reduced) MQH2 PFOR HYD FRD Ferredoxin Reduced NAD+ MQ (oxidised) P5 (4e-) +2ATP + Acetyl-CoA +CO2 +2H H2 Succinate

Acetyl-Phosphate

Acetate Propionate CO2

Fig: 3.17: proposed metabolism of P5 by Bacteroides, involving a series of electron transfers, represented by arrows. A) P5 is proposed to be first reduced, by hydrogenase (HYD) or pyruvate:ferredoxin oxidoreductase (PFOR). A metabolite/s of P5 is then proposed to function as source of porphyrin for b-cytochrome formation, permitting fumarate reductase activity (FRD). Magnesium chelation of the possible metabolite of P5 is not shown. Some pathways are summarised by dashed arrows. Enzymes (in green) include LDH=Lactate dehydrogenase, PFOR, HYD. PEP=phosphoenolpyruvate and MQ=menaquinone. B) A metabolite of P5 that may enable FRD activity. See main text for references and details.

An ex-vivo distal colon culture model was chosen for this study to explore the effect of P5 in complex human-derived bacterial populations of the distal colon, since such a model ameliorates confounders such as uptake by the host, allowing a direct evaluation of the impact of P5 activity. Broadly, since growth of members within the Bacteroidetes, Firmicutes, Proteobacteria and Actinobacteria was perturbed in the ex vivo distal colon culture model or strains were killed in single species growth curves, these results indicate cell surface receptors, that recognised P5, a metal free porphyrin with a ‘bulky’ nitroimidazole appendage, have a previously unappreciated flexibility, apparently independent of the metal ligand and vinyl groups of the porphyrin macrocycle. Also in this model, P5 reproducibly killed Clostridia while promoting growth of Bacteroides. Based on the high homology to the Bacteroides FRD, it is likely that growth of other members of the Bacteroidetes, such as Paraprevotella, and even saccharolytic Porphyromonas with metabolic fermentations similar to Bacteroides rather than proteolytic Porphyromonas (used herein for growth

126 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes curves), will also be stimulated. As with Prevotella, it is predicted that the response of these bacteria will depend on the concentration of P5, whereby below the minimal bactericidal nitroimidazole concentration, growth may be stimulated, whereas at higher concentrations, growth may be inhibited. Scant reports indicate bacteria are able to subsist on antibiotics as a sole carbon source (Dantas et al., 2008), however, to the best of our knowledge P5 is the first molecule reported to both stimulate growth of one clade (the Bacteroides) via formation of a b-cytochrome and suppress growth of another major clade (the Clostridia), of the human gut. On a molar basis, P5 is more potent for susceptible Clostridia than metronidazole and P5 supports the formation of a functional b-cytochrome stimulating growth of B. thetaiotaomicron at nanomolar concentrations (≥2.4 × 106 molecules per cell). Use of metronidazole alone or with vancomycin, during a suite of diseases including colitis or toxic megacolon in which C. difficile is involved, is problematic, since off-target bacteria may be killed, including the Bacteroides. Destabilisation of the gastrointestinal microbiota may also encourage overgrowth of facultative bacteria. P5 presents an alternative strategy, where it is envisaged that the killing of Clostridia with concomitant growth stimulation of Bacteroides, will aid restoration of microbial populations to pre-diseased states, helping to prevent relapse of C. difficile infection and overgrowth of facultative bacteria.

127 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

3.5 Chapter III References

BAUGHN, A. D. & MALAMY, M. H. 2002. A mitochondrial-like aconitase in the bacterium Bacteroides fragilis: implications for the evolution of the mitochondrial Krebs cycle. Proceedings of the National Academy of Sciences, 99, 4662-4667.

BAUGHN, A. D. & MALAMY, M. H. 2003. The essential role of fumarate reductase in haem- dependent growth stimulation of Bacteroides fragilis. Microbiology, 149, 1551-1558.

CAPORASO, J. G., LAUBER, C. L., WALTERS, W. A., BERG-LYONS, D., HUNTLEY, J., FIERER, N., OWENS, S. M., BETLEY, J., FRASER, L. & BAUER, M. 2012. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. The ISME journal, 6, 1621-1624.

CLAESSON, M. J., WANG, Q., O'SULLIVAN, O., GREENE-DINIZ, R., COLE, J. R., ROSS, R. P. & O'TOOLE, P. W. 2010. Comparison of two next-generation sequencing technologies for resolving highly complex microbiota composition using tandem variable 16S rRNA gene regions. Nucleic acids research, gkq873.

DANTAS, G., SOMMER, M. O., OLUWASEGUN, R. D. & CHURCH, G. M. 2008. Bacteria subsisting on antibiotics. Science, 320, 100-103.

DEGNAN, P. H. & OCHMAN, H. 2012. Illumina-based analysis of microbial community diversity. The ISME journal, 6, 183-194.

DI RIENZI, S. C., SHARON, I., WRIGHTON, K. C., KOREN, O., HUG, L. A., THOMAS, B. C., GOODRICH, J. K., BELL, J. T., SPECTOR, T. D. & BANFIELD, J. F. 2013. The human gut and groundwater harbor non-photosynthetic bacteria belonging to a new candidate phylum sibling to Cyanobacteria. Elife, 2, e01102.

EDGAR, R. C., HAAS, B. J., CLEMENTE, J. C., QUINCE, C. & KNIGHT, R. 2011. UCHIME improves sensitivity and speed of chimera detection. , 27, 2194-2200.

EITEL, Z., SÓKI, J., URBÁN, E. & NAGY, E. 2013. The prevalence of antibiotic resistance genes in Bacteroides fragilis group strains isolated in different European countries. Anaerobe, 21, 43- 49.

EWING, B. & GREEN, P. 1998. Base-calling of automated sequencer traces using phred. II. Error probabilities. Genome Research, 8, 186-194.

FINEGOLD, S. M. & SUTTER, V. L. 1971. Susceptibility of gram-negative anaerobic bacilli to gentamicin and other aminoglycosides. Journal of Infectious Diseases, 124, S56-S58.

128 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

GAL, M. & BRAZIER, J. 2004. Metronidazole resistance in Bacteroides spp. carrying nim genes and the selection of slow-growing metronidazole-resistant mutants. Journal of Antimicrobial Chemotherapy, 54, 109-116.

GILLES, A., MEGLÉCZ, E., PECH, N., FERREIRA, S., MALAUSA, T. & MARTIN, J.-F. 2011. Accuracy and quality assessment of 454 GS-FLX Titanium pyrosequencing. BMC , 12, 245.

GOODMAN, A. L., KALLSTROM, G., FAITH, J. J., REYES, A., MOORE, A., DANTAS, G. & GORDON, J. I. 2011. Extensive personal human gut microbiota culture collections characterized and manipulated in gnotobiotic mice. Proceedings of the National Academy of Sciences, 108, 6252-6257.

GRANICK, S., SASSA, S., GRANICK, J., LEVERE, R. & KAPPAS, A. 1972. Assays for porphyrins, δ- aminolevulinic-acid dehydratase, and porphyrinogen synthetase in microliter samples of whole blood: Applications to metabolic defects involving the heme pathway. Proceedings of the National Academy of Sciences, 69, 2381-2385.

HUSAIN, F., VEERANAGOUDA, Y., HSI, J., MEGGERSEE, R., ABRATT, V. & WEXLER, H. M. 2013. Two multidrug-resistant clinical isolates of Bacteroides fragilis carry a novel metronidazole resistance nim gene (nimJ). Antimicrobial agents and chemotherapy, 57, 3767-3774.

HUSE, S. M., WELCH, D. M., MORRISON, H. G. & SOGIN, M. L. 2010. Ironing out the wrinkles in the rare biosphere through improved OTU clustering. Environ Microbiol, 12, 1889-1898.

KATSANDRI, A., AVLAMIS, A., PANTAZATOU, A., HOUHOULA, D. P. & PAPAPARASKEVAS, J. 2006. Dissemination of nim-class genes, encoding nitroimidazole resistance, among different species of Gram-negative anaerobic bacteria isolated in Athens, Greece. Journal of antimicrobial chemotherapy, 58, 705-706.

KLINDWORTH, A., PRUESSE, E., SCHWEER, T., PEPLIES, J., QUAST, C., HORN, M. & GLÖCKNER, F. O. 2012. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next- generation sequencing-based diversity studies. Nucleic acids research, gks808.X, R. J., KNIGHT, R. C. & EDWARDS, D. I. 1983. Studies on the action of nitroimidazole drugs. Biochemical Pharmacology, 32, 2149-2156.

KOZICH, J. J., WESTCOTT, S. L., BAXTER, N. T., HIGHLANDER, S. K. & SCHLOSS, P. D. 2013. Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Applied and environmental microbiology, 79, 5112-5120.

129 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

KOUKOS, G., KONSTANTINIDIS, A., TSALIKIS, L., ARSENAKIS, M., SLINI, T. & SAKELLARI, D. 2016. Prevalence of β-lactam (blaTEM) and Metronidazole (nim) Resistance Genes in the Oral Cavity of Greek Subjects. The open dentistry journal, 10, 89.

KULLIN, B., BROCK, T., RAJABALLY, N., ANWAR, F., VEDANTAM, G., REID, S. & ABRATT, V. 2016. Characterisation of Clostridium difficile strains isolated from Groote Schuur Hospital, Cape Town, South Africa. European Journal of Clinical Microbiology & Infectious Diseases, 35, 1709-1718.

LARUSSO, N. F., TOMASZ, M., KAPLAN, D. & MÜLLER, M. 1978. Absence of Strand Breaks in Deoxyribonucleic Acid Treated with Metronidazole. Antimicrobial Agents and Chemotherapy, 13, 19-24.

LEITSCH, D., SÓKI, J., KOLARICH, D., URBÁN, E. & NAGY, E. 2014. A study on Nim expression in Bacteroides fragilis. Microbiology, 160, 616-622.

LIVINGSTON, S. J., KOMINOS, S. D. & YEE, R. B. 1978. New medium for selection and presumptive identification of the Bacteroides fragilis group. Journal of Clinical Microbiology, 7, 448-453.

MACY, J., PROBST, I. & GOTTSCHALK, G. 1975. Evidence for cytochrome involvement in fumarate reduction and adenosine 5'-triphosphate synthesis by Bacteroides fragilis grown in the presence of hemin. Journal of bacteriology, 123, 436-442.

MARDIS, E. R. 2008. Next-generation DNA sequencing methods. Annu Rev Genomics Hum Genet, 9, 387-402.

MAURICE, C. F., HAISER, H. J. & TURNBAUGH, P. J. 2013. Xenobiotics shape the physiology and gene expression of the active human gut microbiome. Cell, 152, 39-50.

MCCONVILLE, M. L. & CHARLES, H. 1979. Isolation of haemin-requiring mutants of Escherichia coli K12. Journal of general microbiology, 113, 155-164.

MEGGERSEE, R. L. 2014. Metronidazole resistance in clinical Bacteroides fragilis isolates from Groote Schuur Hospital, Cape Town, South Africa. MA thesis., accessed https://open.uct.ac.za/bitstream/item/9482/thesis_sci_2014_meggersee_rl.pdf?sequence= 1. MEGGERSEE, R. & ABRATT, V. 2015. The occurrence of antibiotic resistance genes in drug resistant Bacteroides fragilis isolates from Groote Schuur Hospital, South Africa. Anaerobe, 32, 1-6. NELSON, M. C., MORRISON, H. G., BENJAMINO, J., GRIM, S. L. & GRAF, J. 2014. Analysis, optimization and verification of Illumina-generated 16S rRNA gene amplicon surveys. PloS one, 9, e94249.

130 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

PAN, Y. P., TENG, D., BURKE, A. C., HAASE, E. M. & SCANNAPIECO, F. A. 2009. Oral bacteria modulate invasion and induction of apoptosis in HEp-2 cells by . Microbial Pathogenesis, 46, 73-79.

PARAMAESVARAN, M., NGUYEN, K. A., CALDON, E., MCDONALD, J. A., NAJDI, S., GONZAGA, G., LANGLEY, D. B., DECARLO, A., CROSSLEY, M. J., HUNTER, N. & COLLYER, C. A. 2003. Porphyrin-mediated cell surface heme capture from hemoglobin by Porphyromonas gingivalis. Journal of Bacteriology, 185, 2528-2537.

PRUESSE, E., QUAST, C., KNITTEL, K., FUCHS, B. M., LUDWIG, W., PEPLIES, J. & GLÖCKNER, F. O. 2007. SILVA: a comprehensive online for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic acids research, 35, 7188-7196.

REA, M. C., DOBSON, A., O'SULLIVAN, O., CRISPIE, F., FOUHY, F., COTTER, P. D., SHANAHAN, F., KIELY, B., HILL, C. & ROSS, R. P. 2011. Effect of broad- and narrow-spectrum antimicrobials on Clostridium difficile and microbial diversity in a model of the distal colon. Proceedings of the National Academy of Sciences, 108, 4639-4644.

SCHAUMANN, R., PETZOLD, S., FILLE, M. & RODLOFF, A. C. 2005. Inducible metronidazole resistance in nim-positive and nim-negative Bacteroides fragilis group strains after several passages on metronidazole containing Columbia agar plates. Infection, 33, 368-372.

SCHLOSS, P. D. & WESTCOTT, S. L. 2011. Assessing and Improving Methods Used in Operational Taxonomic Unit-Based Approaches for 16S rRNA Gene Sequence Analysis. Applied and Environmental Microbiology, 77, 3219-3226.

SCHIRMER, M., IJAZ, U. Z., D'AMORE, R., HALL, N., SLOAN, W. T. & QUINCE, C. 2015. Insight into biases and sequencing errors for amplicon sequencing with the Illumina MiSeq platform. Nucleic acids research, gku1341.

SCHMITT, B. H., CUNNINGHAM, S. A., DAILEY, A. L., GUSTAFSON, D. R. & PATEL, R. 2012. Identification of anaerobic bacteria by Bruker Biotyper matrix assisted laser desorption ionization-time of flight mass spectrometry with on-plate formic acid preparation. Journal of clinical microbiology, 02420-12.

SOKI, J., GAL, M., BRAZIER, J. S., ROTIMI, V. O., URBAN, E., NAGY, E. & DUERDEN, B. I. 2006. Molecular investigation of genetic elements contributing to metronidazole resistance in Bacteroides strains. Journal of Antimicrobial Chemotherapy, 57, 212-220.

131 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

TREMBLAY, J., SINGH, K., FERN, A., KIRTON, E. S., HE, S., WOYKE, T., LEE, J., CHEN, F., DANGL, J. L. & TRINGE, S. G. 2015. Primer and platform effects on 16S rRNA tag sequencing. Frontiers in Microbiology, 6, 771.

TRINH, S. & REYSSET, G. 1996. Detection by PCR of the nim genes encoding 5-nitroimidazole resistance in Bacteroides spp. Journal of Clinical Microbiology, 34, 2078-2084.

WHITE, J. R., NAGARAJAN, N. & POP, M. 2009. Statistical methods for detecting differentially abundant features in clinical metagenomic samples. PLoS Comput Biol, 5, e1000352.

132 Chapter III The metabolism of a deuteroporphyrin-nitroimidazole adduct by nanaerobic Bacteroidetes

133 Chapter IV Discussion and future directions

CHAPTER IV

THESIS DISCUSSION AND FUTURE DIRECTIONS

134 Chapter IV Discussion and future directions

4.1 THESIS DISCUSSION

A deuteroporphyrin-nitromidazole adduct (‘P5’), designed to deliver nitroimidazoles into P. gingivalis unexpectedly enhanced the growth of the related and also nitroimidazole sensitive Bacteroides and Prevotella spp. Whilst not designed for this purpose, P5 consists of a deuteroporphyrin base, tethered to a nitroimidazole through an amide linkage. By comparison with DPIX and untreated controls in growth curves and rate of kill assays, P5 was used as a proxy to infer the rate of porphyrin uptake and other aspects of porphyrin metabolism. For example, in single species growth curves and an ex vivo distal human faecal model (Chapter III), P5 killed the majority of Clostridia and the growth of Bacteroides and Prevotella spp. was stimulated and the growth of some Proteobacteria plus Actinobacteria was slowed. These results highlight a previously unappreciated flexibility in the types of porphyrin that are recognised by membrane receptors and taken up by several clades of human-associated bacteria. Indeed, recognition and uptake appear to be independent of the vinyl groups (highlighted in blue, Fig. 4.1) of the porphyrin macrocycle. Secondly, assuming metals are not inserted enzymatically or non-enzymatically before uptake, recognition of P5 and perhaps other tetrapyrroles, also appears to be metal independent. In agreement with these results, a literature review highlighted that porphyrin auxotrophy was broader than previously appreciated in members of the Bacteroidetes. These results broadly agree that many Firmicutes (Bacilli and Clostridia), Fusobacteria and Actinobacteria, take up porphyrin analogues, such as P5, regardless of whether they synthesise haem b or not (summarised in Fig. 4.2). The growth stimulation of the Bacteroides, which for B. thetaiotaomicron was linear from 2 nM - 2µM (or 2.4 × 106-9 molecules of P5 per cell), that the growth of two mutant B. fragilis, incapable of forming a b-cytochrome, were not stimulated by P5 (or DPIX control) and that the FRD activity of cell free lysate was similar for P5 and porphyrin controls, provide evidence that Bacteroides incorporate a metabolite/s of P5 into a cytochrome. Since the growth of examined

Figure 4.1: Haem b (A) and P5 (B).

135 Chapter IV Discussion and future directions

Prevotella spp. was also stimulated, and the fumarate reductase of this clade has a high homology to that of Bacteroides, it is likely Prevotella also form a b-cytochrome from a metabolite/s of P5. It is likely that the uptake of P5 by Bacteroides was delayed, possibly due to an active mechanism to limit porphyrin uptake, although it is also possible that recognition and uptake of P5 is impaired, relative to unconjugated porphyrins. The metabolism of P5 by B. thetaiotaomicron was probed with liquid chromatography mass spectrometry. The products of P5 reduced by sodium-dithionite, were used to study nitroimidazoles and their reduction (Church et al., 1988, Lockerby et al., 1984). Several of the ions (and possible corresponding formulas) of sodium dithionite-reduced P5 were consistent with the fragments formed from the ‘electrolytic’ and enzymatic reduction of metronidazole, producing structures consistent with the fragmentation at carbons 2 and 4 of the imidazole ring (Chrystal et al., 1980, Knox et al., 1983). Also, many of the ions observed in sodium dithionite-reduced P5 were observed in a porphyrin-based extraction of B. thetaiotaomicron incubated with P5. A primary ion, consistent with the formation of 2-amino-N-methyl-2-oxoacetamido-ethyl-amino-3-oxopropyl derivative from DPIX-metronidazole-amine at position 13 of the porphyrin macrocycle was observed, which is similar to the fragmentation of metronidazole to n-(2-hydroxyethyl) oxamic acid. Several sequential fragments of 2-amino-N-methyl-2-oxoacetamido-ethyl-amino-3-oxopropyl side chain were also detected as were ions consistent with fragmentations of the DPIX-nitroimidazole linker, forming 3- amino-3-oxopropyl and 2-aminoethyl-amino-3-oxopropyl side chains. A metronidazole-amine control was not included because mass spectrometry below a mass to charge ratio of approximately 200 has an insufficient accuracy to elucidate formulae consistent with ions. Interestingly, in the B. thetaiotaomicron metabolised P5, several ions were detected that were consistent with the sequential addition of two methyl (–CH3) functional groups to the porphyrin macrocycle. Ions consistent with the addition of two methyl functional groups were also detected in a control culture of B. thetaiotaomicron incubated with DPIX, and these ions were not observed in dithionite-reduced P5. These results strongly indicate that B. thetaiotaomicron has a broader flexibility than previously recognised for the types of abiotic tetrapyrroles incorporated into its metabolism other than a b-cytochrome. Also, modification of P5 is consistent with loss of toxicity when nitroimidazoles are reduced (Knox et al., 1983). Mass spectrometry is not a structural technique, although by the fragmentation signature of the parent ion (how a larger, or parent ion fragments into constituent ions by mass spectrometry of the parental fragment or ms/ms), and from the biology of other microbes, it is possible that these -CH3 groups are appended to one or both of the carboxylic acids (positions 13 and 17 of the porphyrin macrocycle). For example, pheophytins and chlorophyll a contain a 16-carbon chain appended to the carboxyl group at position 17 of the

136 Chapter IV Discussion and future directions macrocycle (Fig. 4.3). One possibility, shown in Fig. 4.2, is that B. thetaiotaomicron appends a long chain hydrocarbon to the carboxylic acid at position 17 of the macrocycle. The addition of a long chain hydrocarbon would decrease the polarity of the molecule, which might be why it was not detected in the final fraction of the B. thetaiotaomicron incubated DPIX extract. Alternatively, it is also possible B. thetaiotaomicron appends only four methyl groups at this position. The smaller ion, which was a fragment of a parent ion, but a distinct ion, could be the result of addition of –CH3 to the methyl group at position 8 of the macrocycle and to the carboxylic acid at position 13 or 17 of the macrocycle, for example by Mg-protoporphyrin IX methyl transferase. It is also possible this fragment was generated by the extraction procedure, for example from autoclaving the B. thetaiotaomicron culture.

137 Chapter IV Discussion and future directions

Figure 4.2: Potential structures of dithionite reduced P5, B. thetaiotaomicron metabolised P5 and DPIX derived from high resolution mass spectrometry. Mass spectrometry is not a structural technique, so some of the functional groups may be oriented slightly differently than proposed, such as methyl and ethyl additions to carboxylic acids of DPIX. Dotted lines indicate structures consistent with distinct ions, rather than fragments of parent ions.

138 Chapter IV Discussion and future directions

Figure 4.3: chlorophyll a, synthesised for example by Cyanobacteria (A). DPIX (B) with carbons 8, 13 and 17 labelled, is shown for reference purposes.

Interestingly, in single species growth curves, the majority of examined Clostridia (five strains) were around three-fold more sensitive to P5 than metronidazole. It is possible that like

2- metronidazole, P5 competes with sulfite (SO3 ) for reducing equivalents originating from ferredoxin in the dissimilatory sulfite pathway, as described in Clostridium spp. (Lockerby et al., 1984, Laishley and Krouse, 1978). It is also possible P5 perturbs the growth of members within the Clostridia by disrupting porphyrin biosynthesis or function of another pathway. In the same vein, in Chapter II, P8 was observed to delay the growth of several bacteria insensitive to nitroimidazoles. It is possible that P8 perturbs the formation of a b-cytochrome in these organisms, including Actinomyces spp. and Faecalibacterium spp., both of which have relatively high amino acid homology (47% and 42% respectively) to the B. thetaiotaomicron fumarate reductase operon (see Appendix Table A4). In Chapter III, findings from single species growth curves were confirmed by selective culturing of Bacteroides and by Illumina sequencing of the 16S rDNA in a human ex vivo faecal culture model. In this model, faecal bacteria were diluted more than 5 orders of magnitude (from ≈5 × 1011 to ≈1 × 1066 cells per mL-1) to be within lethal concentrations of P5 and metronidazole. Strikingly, across the four replicates, there was a remarkable preservation of bacterial genera comparing the inoculum to the DMSO and DPIX treatments, even though bacteria underwent several divisions over the course of the 10 hour period (from the initial ≈1 × 106 cells mL-1 to the endpoint total of ≈5 × 109 cells per mL-1). These results agree with similar models (Freter et al., 1983b). Intriguingly in this model, growth stimulation by DPIX was not as beneficial as P5, yet DPIX has a greater stimulatory capacity for Bacteroides in single species growth curves. One interpretation is that Bacteroides are outcompeted by members of the Clostridia, in a proteinaceous broth such as BHI medium. The faecal culture model does not preserve spatial distribution of gut bacteria, but it could be inferred that in the gut, members of the Clostridia more efficiently metabolise proteinaceous nutrients than Bacteroides. As mentioned in the thesis introduction (Section 1.1.1),

139 Chapter IV Discussion and future directions

E. coli and examined Clostridium spp. (see Chapter III) have similar rapid growth rates in BHI medium. Broad removal of the Clostridia class by P5 leads to overgrowth of Bacilli (mostly streptococci, consisting of 26% of the total population) and a 19-fold increase of Proteobacteria (mostly Gammaproteobacteria including E. coli), relative to DMSO control (see Table 3.3, Chapter III). Interestingly, removal of remaining Clostridia (Lachnospiraceae) and Bacteroides by metronidazole, facilitated growth of Streptococci, which dominated this treatment (70% of the total bacterial population) and also E. coli (21% of this treatment). These results agree that there is a reciprocal relationship between E. coli and more dominant clades. In the presence of bile and P5, a control not shown in Chapter III (see Appendix Table A3 for this control), the removal of the remaining Lachnospiraceae, although not Bacteroides, returned E. coli to levels observed in metronidazole control (22%), suggesting Lachnospiraceae may exert control over E. coli. As suggested by others, this relationship could be mediated by H2S (Freter et al., 1983a). However, it is paradoxical that E. coli populations maintain residence in the gut, or grow in this model, unless they also occupy a unique nutrient niche or provide a nutrient for other bacteria. If such a relationship exists, it is most likely with members within the Clostridial class. On the basis of these results, it is anticipated that the depletion of Clostridia with P5 rather than metronidazole, will reduce the available nutritional and spatial niche for overgrowth of non-target organisms, such as Streptococcus spp. or E. coli.

Consistent with structural studies of metronidazole metabolism and microbiological analyses in murine models (Placidi et al., 1970, Mikelsaar and Siigur, 1992), a remarkable concentration of metronidazole is required to displace nanaerobes residing in the gut. Based on the bactericidal concentration of metronidazole in BHI medium established in Chapters II and III, ranging from 1-11 µM for 5 × 105 viable bacterial cells, extrapolated for 3 × 1012 total bacterial cells in the gut of rats, the minimum concentration of metronidazole is around 65 mg. However, these calculations oversimply the metabolism of metronidazole (see Appendix for calculations of required metronidazole). For example, more than half of the administered metronidazole is oxidised and excreted (see further details in section 1.4.1). Also, as established in Chapter III, when displaced from the human distal gut and diluted into BHI growth medium, the numbers of bacteria within each genera were remarkably stable. The stability of bacterial genera in the ex vivo distal colon culture model indicate that the number and type of bacteria are also likely to be relatively stable over short periods of time and it is possible that the quantity of metronidazole required during health is probably larger than that required during a dysbiotic state. On the basis of ten-fold serial dilutions of metronidazole and porphyrins presented in Chapter III, the minimum concentration of metronidazole required to elicit an effect would be 65 mg per day and around ten-fold lower for P5

140 Chapter IV Discussion and future directions

Table 4.1: Estimated bioactivity of compounds based on extrapolations of the growth of B. thetaiotaomicron in BHI medium. The calculations are based on a rat weighing 200g. For P5, this equates to a 2.3 gram dose in a 70 kg human. See the thesis appendix for full details of these calculations.

Minimally active Molecules bacterial Compound Mass rat-1 (mg) concentration (nM) cell-1 DPIX 0.02 2.409 × 104 0.065 Metronidazole 20 2.409 × 107 65.19 P5 2 2.409 × 106 6.519

to stimulate the growth of Bacteroides (Table 4.1). Interestingly, the calculated concentration of metronidazole is close to that established by Mikelsaar et al. However, as demonstrated in Chapter III, the removal of the Clostridia is probably required to open a spatial and nutritional niche to facilitate the growth of the Bacteroides, requiring a much higher concentration of P5.

In an initial exploration of the response of rat gastrointestinal bacteria to P5, pyrosequencing established there was a surprisingly high variation (2-3 fold) in the populations of bacterial clades of rats housed in one cage (see Table 6.8 of Appendix Chapter VI). Appendix Chapter VI established this variability, indicating that at least a 10 to 100-fold change in anaerobic populations would be required to assess the effect of metronidazole and P5. The pyrosequencing technology used in Chapter VI, with longer reads than Illumina technology, will not be available for future studies. One advantage of the Illumina technology is ‘sequencing depth’, which depending on the number of samples, increases the number of sequences per sample, allowing more accurate enumeration of lower abundance population clades.

In order to determine to the extent that in vitro activity of P5 determined in Chapters II and III reflects biological changes in vivo, rats were also dosed with high concentrations of P5, DPIX and metronidazole (Appendix Chapter VI). To achieve relatively large quantities of required P5, the previous synthesis described in (Yap et al., 2009) was ‘up-scaled’ to produce gram quantities. Regardless of treatment regime, the weight of rats was similar. Strikingly, the caecal volume of the metronidazole-treated rats doubled over a period of 5 days (Appendix Chapter VI). No other differences in gross anatomy of rats were observed. Anatomical changes at a finer resolution were next examined. In relation to the carrier control, the large intestine of the metronidazole, DPIX and P5 treatments all exhibited an increase in lymphoid follicles, particularly adjacent to the crypts. The metronidazole treatment also exhibited goblet cell hyperplasia and an increase in infiltrating

141 Chapter IV Discussion and future directions leukocytes. These promising results call for further exploration into the microbiology and also the response of the host.

4.2 FUTURE DIRECTIONS

4.2.1 Assessing P5 in a murine model of human C. difficile IBD

Unfortunately, time constraints prevented fully testing the hypothesis that results garnered in vitro reflect those in vivo. Planned future studies depend on the demonstration that P5 functions in a rat model to at least limit growth of members within the Clostridia and either have no effect on Bacteroides or facilitate the growth of this clade directly and/or by removal of Clostridia spp. P5 functioning in this way is dependent on many factors, including how P5 is taken up (Ings et al., 1975). Pending success, translational and basic lines of investigation would be warranted. A C. difficile model of human colitis would be explored in a murine model and the derivatisation of P5 to increase potency for Clostridia or to improve stimulation of Bacteroides, would be considered. The possibility of using P5 to examine basic aspects of bacterial populations and their function are also considered below.

As gram quantities of P5 are required to assess the response of nitroimidazole-sensitive microbes of the gastrointestinal tract, the logistical challenges of upscaling synthesis must first be overcome. The planned study would extend the study described in Appendix Chapter VI, to determine if (and where) changes in total bacteria, Clostridia and or Bacteroides occur after a five day concentrated dose of metronidazole, P5, carrier and DPIX controls. It would be possible to determine to what extent changes in bacterial communities reflect function of these populations.

Finally, a planned future study will focus on the host response during Clostridial-insult by examining epithelial integrity (crypt morphology, tight junction integrity), overlying mucus layer and alteration of cells lining the mucosa, including goblet cells, colonocytes and (in the small intestine) Paneth cells. The presence of inflammatory markers (including RegIIIγ, interleukins 11, 17 and 25) or transcript of these markers would be used as a metric to determine to what extent the removal of Clostridia dampened active inflammation, as would metrics such as animal weight and faecal consistency. Microbial screens would be carried out with qPCR and next generation sequencing methodologies to assess changes associated with bacterial genera in illeal and colonic tissue.

142 Chapter IV Discussion and future directions

4.2.2 The use of P5 to glean insight into gut development and function

The removal of microbes from animal hosts has illuminated various principles, such as the stimulation of host development. An extension of this approach involves the addition of one or two bacterial species back into the gut, to demonstrate trade and stimulation. Another approach to answer basic questions involves the removal of specific bacteria or a clade of bacteria and examining the outcome for the host and resident microbes. For example, if it was possible to largely sterilise the gut of Clostridia with P5, would this affect host development, for example by reducing total available butyrate for colonocytes? Would it be possible to confirm that H2S controls populations of E. coli in vivo and that Lachnospiraceae mediate this interaction? Would a reduction in Clostridia impinge upon or aid the growth of Bacteroides and other clades, such as E. coli?

4.2.3 Derivatisation of P5 to improve potency for Clostridia or improve growth of Bacteroides

It was noted in Chapter III that P5 stimulated the growth of B. thetaiotaomicron less efficiently (2 nM) than other porphyrins (200 pM: DPIX, PPIX and Fe-PPIX). In Bacteroides, magnesium chelatase inserts magnesium into the PPIX-methyl ester. To increase efficacy, P5 could be derivatised to P5-methyl ester and/or magnesium could be inserted. Whilst this may help Bacteroides to metabolise P5, it may also modify solubility of the drug in the stomach, since the inner pyrrolic hydrogens would not be available for protonation. It would also be possible to increase the solubility of P5 by addition of urea linkers between nitroimidazole and DPIX.

143 Chapter IV Discussion and future directions

4.3 Thesis Discussion and Future Directions References

CHRYSTAL, E. J., KOCH, R. L. & GOLDMAN, P. 1980. Metabolites from the reduction of metronidazole by xanthine oxidase. Molecular pharmacology, 18, 105-111. CHURCH, D. L., RABIN, H. R. & LALSHLEY, E. J. 1988. Role of hydrogenase 1 of Clostridium pasteurianum in the reduction of metronidazole. Biochemical pharmacology, 37, 1525-1534. FRETER, R., BRICKNER, H., BOTNEY, M., CLEVEN, D. & ARANKI, A. 1983a. Mechanisms that control bacterial populations in continuous-flow culture models of mouse large intestinal flora. Infection and immunity, 39, 676-685. FRETER, R., STAUFFER, E., CLEVEN, D., HOLDEMAN, L. & MOORE, W. 1983b. Continuous-flow cultures as in vitro models of the ecology of large intestinal flora. Infection and immunity, 39, 666- 675. INGS, R., MCFADZEAN, J. & ORMEROD, W. 1975. The fate of metronidazole and its implications in chemotherapy. Xenobiotica, 5, 223-235. KNOX, R. J., KNIGHT, R. C. & EDWARDS, D. I. 1983. Studies on the action of nitroimidazole drugs. Biochemical Pharmacology, 32, 2149-2156. LAISHLEY, E. & KROUSE, H. 1978. Stable isotope fractionation by Clostridium pasteurianum. 2. Regulation of sulfite reductases by sulfur amino acids and their influence on sulfur isotope fractionation during SO32-and SO42-reduction. Canadian journal of microbiology, 24, 716- 724. LOCKERBY, D. L., RABIN, H. R., BRYAN, L. & LAISHLEY, E. 1984. Ferredoxin-linked reduction of metronidazole in Clostridium pasteurianum. Antimicrobial agents and chemotherapy, 26, 665-669. MIKELSAAR, M. & SIIGUR, U. 1992. Metronidazole and the intestinal microecology of rats. Microbial ecology in health and disease, 5, 139-146. PLACIDI, G., MASUOKA, D., ALCARAZ, A., TAYLOR, J. & EARLE, R. 1970. Distribution and metabolism of 14C-metronidazole in mice. Archives Internationales de Pharmacodynamie et de Therapie, 188, 168-179.

144 Chapter IV Discussion and future directions

145 Appendix

CHAPTER V

APPENDIX

146 Appendix

5.1 RESULTS PERTAINING TO CHAPTER II

A 40 y = -1.709ln(x) + 28.211 30 R² = 0.9947

Ct 20

10 0.1 1 10 100 1000 10000 DNA (pg µl-1)

B 8 C 8

)

1 -

7 7 (cfu ml (cfu

10 6 6 LOG

5 5

D 7

E 7

)

1 -

6 6

(cfu ml (cfu 10

5 5 LOG

4 4

Figure A1: A representative standard curve used to convert fluorescent intensity of real time PCR amplified 16S rDNA to the corresponding bacterial viable count [(A), from growth curve of sample C]. Total bacteria detected from plaque curves of patient A (B), patient B (C), patient C (D) and patient D (E) from DNA extracted at 2 hours.

147 Appendix

A 40 B 35 y = -1.62ln(x) + 24.262 y = -1.742ln(x) + 26.743 30 R² = 0.9965 R² = 0.9996

CT 25 20 15 10 0 0.010 1 100 10000 0 0.10 1 10 100 1000 10000 DNA (picograms µl-1) DNA (picograms µl-1)

Figure A2: qPCR standard curves generated with Bacteria domain-specific primers and probe for 16S rDNA gene (A) and P. gingivalis 16S rRNA gene specific primers and probe (B).

Table A1: qPCR fluorescent output and corresponding concentration of total genomic DNA in controls and from two rat groups using 16S rDNA specific primers and taqman probe.

-1 CT DNA (pg µl ) DNA elution buffer 37.25 (±0.49) 0.003 (± 0.0010) qPCR no template control 38.27 (±0.48) 0.002 (± 0.0005) DNA reagent negative 38.20 (±0.74) 0.002 (± 0.0008) Group 1 22.64 (±0.39) 23.35 (±4.40) Group 2 24.25 (±0.52) 10.00 (±1.93)

148 Appendix

5.2 RESULTS PERTAINING TO CHAPTER III

Table A2: Growth of bacterial cultures streaked on BHI agar media with gentamicin (100 mg L-1) or with gentamicin and bile (titrated at 5, 10, 15, 20 g L-1). For bacteria that grew after 4 days on gentamicin BHI Esculin Agar, growth was assessed as excellent (++), impaired (+), scant (~) or non- existent (-) at the four bile treatment concentrations. Esculin hydrolysis (indicated by a dark halo surrounding colonies) was not assessable for strains that failed to grow on gentamicin BHI agar (denoted N/A).

Gentamicin BHI Gentamicin BHI Esculin Agar Bile Esculin Agar Esculin Bile Family Strain Growth Hydrolysis (5/10/15/20 g L-1) Porphyromonas gingivalis W83 + - - Porphyromonadaceae Porphyromonas gingivalis ATCC 33277 + - - Prevotella melaninogenica ATCC 25845 - N/A - Prevotellaceae Prevotella Intermedia ATCC 25611 - N/A - Bacteroides fragilis ATCC 25285 ++ + ++ Bacteroides thetaiotaomicron ATCC 29148 ++ + +/+/+/~ Bacteroides thetaiotaomicron ATCC 29741 ++ + +/+/+/~ Bacteroidaceae Bacteroides vulgatus ATCC 8482 ++ - ~/-/-/- Bacteroides uniformis ATCC 8492 ++ + ++/++/++/+ Bacteroides eggerthii ATCC 27754 ++ + ++/++/++/+ Fusobacterium nucleatum ATCC 25586 + - - Fusobacteriaceae 'Leptotrichia buccalis' - N/A - Veillonellacaeae Veilonella parvula ATCC 10790 ~ - - Parvimonas micra ATCC 33270 - N/A - Clostridiaceae Clostridium perfringens ATCC 13124 + - - Clostridium difficile ATCC 43255 ~ - - Enterococcaceae Enterococcus faecalis ATCC 29212 ~ + - Serratia marcescens ATCC 274 - N/A - Enterobacteriaceae Escherichia coli ATCC 25922 - N/A - Bifidobacteriaceae Bifidobacterium dentium ATCC 27534 - N/A - Coriobacteriaceae Eggerthella lenta ATTC 43055 - N/A -

149 Appendix

Table A3: Illumina sequencing of the ex vivo faecal culture model (V4 region 16S rDNA) at 10 hours. The absolute total was calculated from 10,864,916 sequences that passed quality filtering. The average (‘Av.’) and SEM were calculated within each of the seven treatment groups from four experiments. Phylum summing less than 0.5% of the total sequences are summarised as ‘Other’ including: unclassified (8661 sequences), Fusobacteria (107 sequences), Planctomycetes (6 sequences), Acidobacteria (3 sequences), OD1 (3 sequences), Chloroflexi (1 sequence) and OP11 (1 sequence). P5(B) treatment group consisted of P5 (20 µM) plus bile (‘oxgal’, 5gL-1).

Inocula DMSO DPIX DPM M P5 P5(B) Taxa Av. SEM Av. SEM Av. SEM Av. SEM Av. SEM Av. SEM Av. SEM Bacteroidetes 27.0 0.7 13.6 1.9 16.1 2.1 0.6 0.3 0.8 0.5 32.7 3.5 47.1 5.9 Bacteroidaceae 23.6 0.6 13.1 1.8 15.3 1.8 0.5 0.3 0.7 0.5 31.7 3.5 46.5 5.9 Porphyromonadaceae 0.9 0.2 0.4 0.1 0.7 0.3 0.0 0.0 0.0 0.0 0.9 0.2 0.3 0.1 Rikenellaceae 1.8 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.1

Firmicutes 63.8 1.0 84.0 2.5 81.3 3.2 76.3 11.2 71.5 10.2 55.7 8.0 28.5 6.8 Bacilli 0.6 0.0 13.5 3.4 10.3 2.4 73.9 10.9 70.0 9.8 33.0 4.4 3.6 2.8 Enterococcaceae 0.0 0.0 0.0 0.0 0.0 0.0 1.1 0.9 0.3 0.2 7.4 7.2 0.1 0.1 Streptococcaceae 0.4 0.0 12.9 3.5 9.6 2.5 72.5 11.0 69.7 9.8 25.5 7.9 0.1 0.0 Turicibacteraceae 0.2 0.1 0.5 0.3 0.6 0.3 0.0 0.0 0.0 0.0 0.0 0.0 3.4 2.9 Clostridia 59.0 1.5 63.4 4.5 64.1 4.2 2.0 1.0 1.2 0.4 22.3 6.9 20.0 4.9 Unclassified family 1.0 0.2 8.0 5.7 7.2 6.0 0.4 0.3 0.1 0.1 0.0 0.0 0.0 0.0 Clostridiaceae 1.1 0.4 2.2 1.0 1.2 0.8 0.1 0.0 0.0 0.0 0.0 0.0 2.6 2.5 Lachnospiraceae 30.5 1.9 39.8 2.0 43.5 5.1 1.2 0.5 0.8 0.3 20.1 7.0 6.3 1.7 Ruminococcaceae 25.4 3.6 12.7 1.1 11.5 1.9 0.4 0.1 0.2 0.1 0.3 0.1 0.1 0.0 Veillonellaceae 1.0 0.1 0.5 0.1 0.5 0.1 0.0 0.0 0.0 0.0 1.9 0.3 11.1 1.8 Erysipelotrichi 1.5 0.3 7.0 2.6 6.8 2.1 0.1 0.0 0.0 0.0 0.4 0.2 4.8 1.4 Erysipelotrichaceae 1.5 0.3 7.0 2.6 6.8 2.1 0.1 0.0 0.0 0.0 0.4 0.2 4.8 1.4 Unclassified class 2.6 0.9 0.0 0.0 0.0 0.0 0.3 0.2 0.2 0.1 0.1 0.0 0.0 0.0

Proteobacteria 7.1 0.4 0.7 0.4 1.3 0.6 21.2 11.6 26.3 10.2 10.9 5.4 22.0 13.2 Betaproteobacteria 1.0 0.0 0.1 0.0 0.2 0.1 5.0 1.4 4.9 1.4 1.1 0.3 0.1 0.1 Alcaligenaceae 1.0 0.0 0.1 0.0 0.2 0.1 5.0 1.4 4.9 1.4 1.1 0.3 0.1 0.1 Deltaproteobacteria 1.1 0.3 0.1 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.2 0.1 0.2 0.0 Desulfovibrionaceae 1.1 0.3 0.1 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.2 0.1 0.2 0.0 Gammaproteobacteria 5.0 0.5 0.5 0.3 1.0 0.6 16.2 12.6 21.4 11.3 9.6 5.4 21.7 13.2 Enterobacteriaceae 0.0 0.0 0.4 0.4 0.8 0.6 14.4 13.2 19.5 12.2 9.3 5.5 21.6 13.2 Pasteurellaceae 5.0 0.5 0.1 0.0 0.1 0.0 1.8 0.6 2.0 0.9 0.3 0.1 0.0 0.0

Actinobacteria 0.5 0.2 1.6 0.4 1.3 0.5 1.9 0.8 1.4 0.5 0.7 0.4 2.5 1.1 Bifidobacteriaceae 0.2 0.1 0.8 0.4 0.5 0.2 1.9 0.8 1.4 0.5 0.7 0.4 2.2 1.2 Coriobacteriaceae 0.3 0.2 0.8 0.6 0.7 0.6 0.0 0.0 0.0 0.0 0.0 0.0 0.3 0.1

Other 1.6 1.5 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0

150 Appendix

13

12 BA

BGBHI

(Viable Count) (Viable 11

10 Log 10 0 1 2 3 4 5 6 7 8 Day

Figure A3: Estimations of viable Bacteroides and total bacteria per gram human faeces (from the ex vivo distal colon culture model donor) over seven days on bile and gentamicin containing BHI agar (‘BGBHI’) and BHI blood agar (‘BA’) plates. Over seven days, viable Bacteroides averaged 19% of total bacteria recovered on BHI blood agar (‘BA’) plates. Each bar incorporates the mean from triplicate plate counts.

10 DMSO 1 DPIX (20µM)

0.1 DPIX+M (20µM) M (20µM)

OD(600nm) 0.01 P5 (20µM) 0.001 0 10 20 Time (h)

Figure A4: growth curves of ex vivo distal colon culture model. Bacterial growth was estimated by optical density at 600 nm, with the mean and SEM from four experiments. The average total bacterial inoculum was 1.95±0.72 × 106 cfu ml-1, 15% of which were Bacteroides 2.97±0.51 × 105 cfu ml-1, assessed on culture medium as described in Chapter III. Doubling times for DMSO, DPIX, DPIX and metronidazole, M and P5 were 42.8, 37.5, 54.0, 54.0 and 50.0 min-1 respectively. The optical density of treatments containing gentamicin and bile was not recorded.

151 Appendix

Table A4: Summary of BLASTP homology searches to the Bacteroides thetaiotaomicron ATCC 29148 fumarate reductase. ‘FRD units’ indicates how many of the three genes were present in a given taxon and ‘average hits’ refers to the average number of taxa (bacterial strains) within each clade. Bacterial clades with representatives that were examined with in this thesis with growth curves are in bold font. The highest resolution is given when genus descriptions were not available.

Taxon FRD units Average Homology (%) Average Hits Bacteroides 3 87.2 78 Paraprevotella 3 80.5 3 Phocaeicola 3 77.0 1 Prevotella 3 75.6 107 Barnesiella 3 73.3 2 Coprobacter 3 73.2 2 Alloprevotella 3 72.9 3 Parabacteroides 3 70.2 8 Tannerella 3 70.0 10 Porphyromonas 3 69.9 32 Paludibacter 3 69.7 1 Dysgonomonas 3 69.7 5 Proteiniphilum 3 68.7 1 Porphyromonadaceae 3 68.0 2 Bacteroidetes 3 67.6 2 Alistipes 3 65.5 18 Mucinivorans 3 65.3 1 Niastella 2 64.0 1 Sporocytophaga 2 63.8 3 Fibrobacter 2 63.8 2 Thermophagus 3 63.7 1 Anaerophaga 3 63.5 2 Geothrix 2 63.0 1 Haloferula 2 62.5 1 Marinilabilia 3 62.3 1 Rikenella 3 62.0 1 Actinobaculum 2 62.0 1 Rubritalea 2 62.0 1 Alkaliflexus 3 61.7 1 Saccharicrinis 3 61.7 2 Geofilum 3 61.0 1 Enhygromyxa 2 61.0 1 Halothece 2 60.5 1 Serinicoccus 2 60.0 2 Sanguibacteroides 3 59.7 1 Butyricimonas 3 59.7 2 Zavarzinella 2 59.5 1 Sunxiuqinia 3 59.2 1 Draconibacterium 3 59.0 2 Methylacidiphilum 2 59.0 4 Odoribacter 3 58.7 4 Aphanocapsa 2 58.5 2 Sandaracinus 2 58.5 2 Persicobacter 3 58.0 1 Prolixibacter 3 58.0 1 Ornithinimicrobium 2 58.0 1 Granulicoccus 2 58.0 1 Fulvivirga 3 56.8 1 Methylophaga 2 56.3 7 Flavobacteriales 3 55.8 3

152 Appendix

Calculations extrapolated from in vitro potency of metronidazole and porphyrins

The minimum inhibitory concentration of metronidazole required to kill 5 × 105 viable bacteria in BHI medium ranged from 1-11 µM. If it is taken that 10 µM (1.172 ng mL-1) metronidazole is required to kill 5 × 105 cells mL-1:

5 × 105 bacterial cells / 1.172 ng metronidazole

1 bacterial cell / 2.344 pg metronidazole

If it is taken that the rat caecal volume is 3 mL, with 1 × 1012 bacterial cells per ml, we have 3 × 1012 bacterial cells:

1 bacterial cell / 2.344 pg metronidazole

3 × 1012 bacterial cells / [(2.344 pg metronidazole) × (3 × 1012)]

3 × 1012 bacterial cells / 7.032 g metronidazole

Culture conditions (in vitro) are expected to provide conditions more favourable for carbon and nitrogen acquisition, as well as the maintenance of membrane polarity and integrity. Cultured bacteria may therefore be more resistant to metronidazole than bacteria residing in the rat gut. Also, metronidazole may be inactivated by non-target bacteria or metabolised by the host. Ignoring factors that may impede the efficacy of metronidazole in rats, based in vitro efficacy of metronidazole, 7 grams of metronidazole are required to kill the majority of anaerobic bacterial inhabitants in rats.

The molecular weight of metronidazole is 171.15 g mol-1. Thus, to deliver 7.032 g metronidazole in 200 µL, would require a stock solution of 35.16 g mL-1. In water, metronidazole solubility approximates 10 mg ml-1, whilst in DMSO the concentration approaches 30 mg ml-1.

153 Appendix

Extrapolations of the minimum concentration of DPIX required for in vivo growth stimulation of Bacteroides

For deuteroporphyrin IX, the in vitro concentration required to promote excellent growth of Bacteroides approximates 2 nM, (around 3-log lower than the concentration of metronidazole required to kill susceptible bacteria). If it is taken that 2 nM (6.022 ×1011 molecules ml-1) DPIX is required to stimulate growth of 5 × 105 cells ml-1:

5 × 105 bacterial cells / 6.022 × 1011 molecules of DPIX

1 bacterial cell / 1.204 × 106 molecules of DPIX

If it is taken that the rat caecal volume is 3 mL, with 1 × 1012 bacterial cells per ml, we have 3 × 1012 bacterial cells:

3 × 1012 bacterial cells / [(1.204 × 106 molecules of DPIX) × (3 × 1012)]

3 × 1012 bacterial cells / 3.613 × 1018 molecules of DPIX required

In a rat weighing 200g, a 3.613 × 1018 molecule dose of DPIX equates to:

3.502 mg of DPIX (17.51 mg kg-1)

154 Appendix

155 Appendix

APPENDIX CHAPTER VI

INITIAL INVESTIGATIONS INTO THE RESPONSE OF RAT GASTROINSTESTINAL BACTERIA TO THE DEUTEROPORPHYRIN- NITROIMIDAZOLE ADDUCT P5

156 Appendix

6.0 PREFACE TO APPENDIX CHAPTER VI

In Chapter III, P5 was found to behave like biotic porphyrins, stimulating the growth of Bacteroides spp. (five strains) and below growth inhibitory concentration, Prevotella spp. (two strains) in brain heart infusion medium. None the facultative bacteria (seven strains) or the nanaerobe F. nucleatum were killed by P5. In contrast, P5 retained nitroimidazole qualities for nanaerobic Clostridiales (five strains tested), which were killed by P5. Findings were confirmed by Illumina sequencing and viable counts of an ex-vivo human faecal culture model, in which the growth of Bacteroides was stimulated and capacity to kill Clostridia was retained by P5. Notably, in faecal growth curves, DPIX only slightly stimulated growth of Bacteroides, suggesting by reference to the P5 treatment, that removal of Clostridia was required to open a nutritional niche for the Bacteroides. The possible formation of a b-cytochrome using a metabolite/s of P5 in Bacteroides, which would link products of glycolysis to the terminal respiration of fumarate was described. Together with in silico data and a re-examination of the literature, the previously unappreciated aspect of porphyrin auxotrophy of Prevotella and Porphyromonas was noted, and it was suggested Prevotella also incorporate metabolites of P5 into a b-cytochrome, enabling fumarate reductase activity.

In appendix Chapter VI, a rat gastrointestinal model is used to test the predicted metabolism of P5 based in vitro results outlined in Chapter III. It is found that high concentrations of P5 (and metronidazole) are required to kill susceptible nanaerobes. The synthesis of P5 is up-scaled and a preliminary investigation of the biology of rats is examined after administering high doses of metronidazole and P5.

157 Appendix

6.1 INTRODUCTION

In Chapter III, the effects of P5 observed in single species growth curves reflected those detected by Illumina sequencing in a human faecal growth curve model. While the bacterial populations were remarkably stable, such a model ignores spatial distributions of bacteria in the gastrointestinal tract, pH and concentration gradients of nutrient/waste products. Specifically, how mucosal bacterial populations will respond to P5, or what extent the availability of molecules such as menadione impinge upon fumarate reductase activity in the Bacteroides. The choice of the rat gastrointestinal tract to examine the ecological effects of P5 is based on several criteria. Clone libraries with few sequences suggest the types of bacteria in the faeces of laboratory rats at the genera level is similar to humans, sharing Bacteroidetes (Porphyromonas, Prevotella and Bacteroides), Firmicutes (Lactobacilli, Clostridium, Ruminococcus, Lachnospira and Enterococcus), Actinobacteria (Corynebacterium, Actinomyces and Eubacterium) and Proteobacteria (Sutterella and Escherichia) (Inoue and Ushida, 2003, Brooks et al., 2003). Secondly, as in humans, Bacteroidetes and Firmicutes dominate the rat gut, which was established by Ley et al., 2008 (see Introduction). Anaerobic genera of these clades are sensitive to nitroimidazoles and the number of viable bacteria per gram of stool (>5 × 1011) is similar to human counterparts. The gross anatomy in rats is similar to humans, other than an enlarged caecum. Lastly, inbred laboratory rats, such as specific ‘pathogen free’ Fischer 344 rats have identical housing, diet and genetics, all of which should reduce biological confounders in assessing the ecological effects of P5 or metronidazole.

Some of the advantages and disadvantages of analysing complex bacterial populations are discussed in Chapter III (see introduction to Chapter III). In appendix Chapter VI, the primary reason for selecting 454 pyrosequencing for analysing complex bacterial populations from rat gastrointestinal tracts is that this technology has the capacity to sequence longer reads than Illumina paired-end Miseq platform (see Fig. 6.1 for an overview of this technology). Secondly, it is known that shorter but more numerous paired end reads (Illumina technology) technology inflate the OTU number of sampled microbes from human faeces compared with longer reads (454 pyrosequencing Titanium technology) (Claesson et al., 2010), which is in part due to decreases in accuracy as a given molecule is sequenced. Regions V3-V4 of the 16S rDNA molecule were chosen, using the same primer set for total estimations of bacterial populations, allowing direct comparisons between qPCR and 454 pyrosequencing data sets. Sequencing technologies are prone to error (as detailed above and in the introduction of Chapter III). Mothur was chosen for analysis, following a conservative methodology for reducing error associated with next generation sequencing. For example, to reduce inflated estimates of bacterial diversity, reads that were similar to each other (within 4 nucleotides

158 Appendix

Fig 6.1: An overview of 454 pyrosequencing [from (Metzker, 2010)]. Template DNA is PCR amplified with universal primers appended to unique DNA adapters. DNA molecules are denatured, and captured onto beads under conditions that favour one sequence per bead. Single DNA molecules are then PCR amplified and typically deposited into wells for sequencing. As a specific nucleotide is incorporated (A, T, C or G) by polymerase, pyrophosphate is released. Pyrophosphate is converted to ATP by ATP sulfurylase. ATP in turn acts as a substrate for luciferase mediated conversion of luciferin to oxyluciferin and release of light. The emitted light is correlated to which nucleotide (A, T, C or G) was added, in an iterative process. The unique DNA adapter, which identifies the sample the sequence originates from, may then be coupled to the sequence of the DNA molecule.

or ~1% of the entire read length) were merged into one read (Huse S.M. et al., 2010). Secondly, sequences were screened for putative chimeras, again reducing arbitrarily inflated estimates of bacterial diversity.

As mentioned in Chapter II, selective growth medium overcomes some of the issues associated with molecular approaches, including 16S rDNA copy number, PCR amplification-bias, amplification of DNA from non-viable bacteria and error associated with next generation sequencing. Kanamycin is a bactericidal aminoglycoside produced by Streptomyces kanamyceticus (Umezawa et al., 1957). Kanamycin prevents protein synthesis by binding the bacterial 30S ribosome, inhibiting translocation of tRNA when N-acetyl-diPhe-tRNA is on the acceptor site and deacylated tRNA at the donor site (Misumi and Tanaka, 1980). Kanamycin kills facultative anaerobes and not Eukaryotes, since the 30S ribosome is not present (Waitz et al., 1972). Uptake of aminoglycosides results from an energy-dependent transport system using oxygen or nitrate as a terminal electron acceptor (Bryan et al., 1979). As with other aminoglycosides, uptake of kanamycin is probably low under anaerobic conditions, facilitating survival of the Bacteroidetes. Proteobacteria and Actinobacteria are broadly expected to be killed by kanamycin. Vancomycin inhibits cell wall synthesis in Gram-positive bacteria (Jordan, 1961, Rogosa et al., 1958, Halebian et al., 1981). Low concentrations of Vancomycin (˂5 µg/ml) are bactericidal for Actinomyces, Clostridium and Lactobacilli in vitro (Rogosa et al., 1958, Halebian et al., 1981). Together, kanamycin and vancomycin

159 Appendix are expected to allow survival of nanaerobes from the Bacteroidetes phylum, such as Prevotella, Porphyromonas and Bacteroides, making it an appropriate choice to determine if metronidazole kills these genera and/or if P5 stimulates growth.

6.2 METHODS

6.2.1 Administration of treatments into rats

Inbred ‘specific pathogen free’ female Fischer 344 rats (Animal Resource Centre, Australia) were housed in groups of five at the Department of Animal Care, Westmead Hospital. Rats were afforded ad libitum access to autoclaved water and chow (Specialty Feeds, Australia). The age of the rats is indicated for each experiment. Bedding was replaced bi-weekly (allowing coprophagy). For initial administration (2mM final concentration), 10 mM stock treatments were solubilised into 40 µL dimethyl sulfoxide (DMSO) before dilution into 160 µL carboxy methyl cellulose (CMC, final concentration 1% (w/v)). 2mM treatments included DMSO (40 µL made up to 200 µL with CMC), P5 (663.5, g L-1), deuteroporphyrin IX [(DPIX) Frontier Scientific, 511.2 g L-1] and metronidazole (171.2, g L-1). Carboxymethyl cellulose (‘carrier’, 160 µL) was also gavaged daily into rats. Treatments (200 µl) were delivered daily by oral gavage during isoflurane anesthetisation (c. 3.5% per minute) for 5 days. In additional experiments following ‘high-dosing’ protocol (Mikelsaar and Siigur, 1992), treatments were emulsified into 150 µl CMC [1% (w/v)] only. Approval of animal procedures was granted by Westmead Hospital Animal Ethics Committee (protocol 4185.12.11). Reagents were supplied by Sigma-Aldrich unless stated otherwise.

6.2.2 Sampling of tissue or faeces and digesta for DNA Extraction and Microscopy

One day after a 5 day administration of treatments, rats were culled with carbon dioxide. In the initial experiment, faeces was sampled as indicated in text. In the second experiment, approximately 30 mm from the anus, a 10 mm section of colon was sampled. The colon segment was prepared in 5 ml paraformaldehyde (2%) with sucrose (5%) and PBS (2x) at 4°C for 24 hours, before washing with PBS (2x) and soaking in sucrose (30% v/v, filtered) with PBS (1x) for 24 hours at 4°C. Approximately 30 mm from the anus, the colonic contents were extracted from a 10 mm piece of colonic tissue.

160 Appendix

Paraformaldehyde fixed sections were next embedded in Tissue Freezing Medium (23°C, Tissue Plus, Scigen) for 1 hour and frozen in liquid nitrogen. Embedded samples were stored at -80°C until microtomy. 5-6µm thick sections were cut with a microtome (Microm, HM 505E). By serial sectioning, a library of 10 sections per sample was made. 2 sections (totalling 40) from each sample were warmed to room temperature and stained with toluidine blue before a glass cover slip (22×50 mm, Matsunami) was mounted (Permount). The 40 sections were examined by a pathologist with more than 30 years experience in tissue . Representative images were obtained with a Bx60 light microscope (Olympus) with a 40x water immersion lense.

6.2.3 Real time PCR estimations of bacterial populations

Real time PCR was performed as described in Chapter II, except 25 µL SYBR GreenER (Invitrogen) was used instead of Taqman probes and template consisted of 1 µL of a 1/50 dilution of extracted DNA (typically reactions contained low ng total DNA). The fluorescent emission of SYBR Green, was corrected by ROX. The viable count of B. thetaiotaomicron VPI-5482 grown for 20 hours in BHI broth, was taken in triplicate on BHI agar containing 4% sheep blood. An aliquot of this culture was extracted using methods described in Chapter II. Genomic DNA from B. thetaiotaomicron was quantified with a nanophotometer (Implen), used to construct standard curves, consisting of duplicate 10-fold dilutions, over 5-6 orders, for the conversion of threshold cycle (CT) to DNA concentration, which was converted to the viable count ascertained above. Bacteroidetes populations were assessed with forward (‘BactF341’, 5’-AATGGGCGAGAGCCTGAAC) and reverse (‘BactR700’ 5’-AAGCTGCCTTCGCAATCG) primers targeting a region of 16S rDNA specific for Bacteroidetes (Nadkarni et al., 2012). Amplicons were assessed by agarose electrophoresis (1% agarose, Promega) in the presence of gelred (0.5×, Biotium) or by melt curve analysis.

6.2.4 Pyrosequencing

Bacterial domain specific forward (5’-TCCTACGGGAGGCAGCAGT) and reverse (5’- GGACTACCAGGGTATCTAATCCTGTT) primers, targeting regions V3-V4 of the 16S rDNA gene (generating a 466 base pair amplicon, corresponding to positions 331 – 797 of Escherichia coli 16S rRNA) (Nadkarni et al., 2002)) were appended with DNA barcodes as previously described (Kianoush et al., 2014). Amplicons were amplified, such that each sample had a unique DNA barcode. PCR products were normalised to 20 ng µL-1 and sent to the Australian Centre for Genomic Analysis (AGRF) for sequencing from the forward primer using a 454 FLX instrument with Titanium chemistry.

161 Appendix

Sequences were demultiplexed by AGRF and were analysed with Mothur (Schloss and Westcott, 2011) essentially as described (‘454 SOP’, http://www.mothur.org/wiki/454_SOP). Amplicons were removed if they had: a length less than 250 bp, a moving average quality score less than 30, more than eight homopolymers, or more than two differences to the forward primer. Sequences were aligned against the Mothur formatted SILVA reference alignment. Sequences were merged together (‘clustered’) if they differed by more than 4 bases (diffs=2) (Huse S.M. et al., 2010). After chimera removal, filtered reads were classified against the Greengenes 16S rDNA reference taxonomy (‘gg_13_5_99’) using a 70% pseudobootstrap value. Sequences were clustered into OTUs (3% similarity threshold). The number of quality filtered sequences within a sample were converted to a percentage of the total sequences within the same sample. Taxa that were less than 0.5% abundance within a given sample are not shown. The average and standard error of the mean were calculated for remaining taxa within each sample type. Metastats, which uses a Student’s T-test (White et al., 2009) was used within Mothur to assess statistical significance of OTUs binned at 0.05 dissimilatory threshold (approximating genus).

6.2.5 Viable bacterial counts

For testing ‘high-dose’ protocol by (Mikelsaar and Siigur, 1992), fresh faecal content was collected from rats (n=4, housed in one cage), before dilution (50 mg) into 4950 µL anaerobically equilibrated warmed PBS. For each rat, diluted faeces was plated onto brain heart infusion agar with kanamycin (206.4 µM, 100 µg ml-1) and vancomycin (10.3 µM, 15 µg ml-1) (‘KVBHI’) over three dilutions (at 10-7, 10-8 and 10-9). KVBHI plates were incubated anaerobically at 37 °C for 4 days before counts were made. Diluted faeces was also plated onto BHI agar supplemented with 4% sheep blood, and incubated aerobically at 37 °C for 4 days before counts were made.

6.3 RESULTS

6.3.1 The effect of 2mM P5 on rat faecal microbes

An initial five day administration of ‘2 mM’ treatments (0.2-0.4 mg per day into 200 gram rats, Table 6.1) was carried out to assess the effect of P5 on rat gastrointestinal bacteria and also the toxicity of P5 for rats. The total viable faecal count (1.6-3.3 ×1010 bacteria g-1, Table 6.2) was slightly lower than expected from the literature and Bacteroidetes populations tended to be slightly higher than expected, particularly in the DPIX treatment. When the total bacterial load increased or

162 Appendix

Table 6.1: Daily treatments (200 µL of 2 mM stock) given for five days rats weighing ≈180 g. Rats were 6-8 weeks old at time of treatment administration. This dose of P5 equates to 103 mg per day in a 70 kg human.

Treatment mg mg kg-1

Deuteroporpyrhin IX (Cl2) 0.2 1.2 Metronidazole 0.1 0.3 P5 0.3 1.3

decreased, so did the total Bacteroidetes, averaging 54-57%, other than the DPIX treated rats, in which the Bacteroidetes populations consisted of 73% of the total detected Bacteria (Table 6.3). After a five day administration of ‘2 mM’ treatments into rats, DNA was extracted following ‘DNA from stool for pathogen detection’ protocol. Non-treated and DMSO treatments contained similar concentrations of bacteria, approximating 1.9 × 1010 bacteria per gram wet weight faeces, which was slightly higher than P5 treatment and around half metronidazole and DPIX treatments 2.5-3.3 × 1010 (Table 6.2).

Table 6.2: Total bacteria and Bacteroidetes detected in rat faeces immediately after ‘2mM treatment’ by qPCR. No template control (NTC) contained qPCR reagents without DNA [the fluorescent threshold of the NTC was 36.9 (±0.4)]. BD=below detection.

Bacteria Bacteroidetes DNA (ng µl-1 ) Bacteria (g-1) DNA (ng µl-1 ) Bacteria (g-1) NTC 9 (±3) × 10-6 NA BD BD Carrier 45.6 (±5.9) 1.9 (± 0.3)10 25.5 (±1.8) 1.1 (±0.1)10 DMSO 43.9 (±3.2) 1.9 (± 0.1)10 25.2 (±1.9) 1.1 (±0.1)10 DPIX 60.1 (±8.6) 2.5 (± 0.4)10 42.5 (±6.5) 1.8 (±0.3)10 Metronidazole 78.3 (±11.8) 3.3 (± 0.5)10 39.6 (±5.7) 1.7 (±0.2)10 P5 37.3 (±6.1) 1.6 (± 0.3)10 20.1 (±3.0) 8.5 (±0.1)9

163 Appendix

Table 6.3: Average of Bacteroidetes populations and non-Bacteroidetes recovered immediately from the faeces of ‘2mM treatment’ groups detected by qPCR (5 rats per group). ‘Non-Bacteroidetes’ were estimated by subtracting the total bacteria from the Bacteroidetes group.

Bacteroides/total Rat group Non-Bacteroidetes (%) Bacteria (%) Carrier 56.0 44.0 DMSO 57.3 42.7 DPIX 71.0 29.0 Metronidazole 50.1 49.9 P5 53.8 46.2

After 2 weeks, the total detected bacteria from rat faeces was similar, ranging from 1.4- 2.7 ×1010 per gram wet weight, the highest of which was in metronidazole treatment (Table 6.4). Total bacteria in DPIX treatment was identical to other groups (2.5 × 1010 g-1), although Bacteroidetes populations ranged from 73% to 47% (or 0.6 × 1010 g-1) (Tables 6.4 and 6.5). The inflated number of bacteria 3.3 × 1010 g-1 in the metronidazole treatment (which was 1.7 fold higher than non-treated or DMSO) observed immediately after treatment had reduced to 2.9 × 1010 g-1 by two weeks (Tables 6.4 and 6.5). Changes in populations, in particular a doubling of detected bacteria in metronidazole treated rats in respect to DMSO treatment, were taken to represent a metronidazole-induced perturbation of bacterial populations of the rat intestine, so a portion of the 16S rDNA from samples was amplified and sequenced (454 pyrosequenced).

Table 6.4: Total bacteria and Bacteroidetes detected in rat faeces two weeks after ‘2mM treatment’ by qPCR. No template control (NTC) contained qPCR reagents without DNA template. No template control (NTC) contained qPCR reagents without DNA [the fluorescent threshold of the NTC was 37.6 (±0.3)]. BD=below detection.

Bacteria Bacteroidetes DNA (ng µl-1 ) Bacteria (g-1) DNA (ng µl-1 ) Bacteria (g-1) NTC 4 (±1) × 10-6 NA BD BD Carrier 45.5 (±3.7) 1.9 (± 0.2)10 26.2 (±2.32) 1.1 (±0.1)10 DMSO 64.0 (±8.6) 2.7 (± 0.4)10 32.1 (±3.10) 1.4 (±0.1)10 DPIX 60.1 (±8.6) 2.5 (± 0.4)10 28.3 (±6.28) 1.2 (±0.3)10 Metronidazole 68.2 (±7.0) 2.9 (± 0.3)10 28.2 (±2.84) 1.2 (±0.1)10 P5 32.0 (±8.3) 1.4 (± 0.4)10 17.3 (±2.07) 0.7 (±0.1)10

164 Appendix

After quality filtering and chimera removal 67% (156,870 out of 233,259) of sequenced amplicons remained (Table 6.6) with an average length of 405 bp. Regions conserved in all amplicons were ‘masked’ (removed), because they did not inform taxonomic mapping, leaving an average of 274 bp (range 255-290 bp) for identification with the Greengenes reference 16S rDNA taxonomy database. 10,725 unique sequences were found, classifying to 1,346 OTUs and 707 OTUs at 3% and 5% similarity respectively. An average of ≈6,300 sequences/sample was obtained. None of the known contaminants in the QIAamp DNA Stool Mini Kit DNA extraction kit, were detected in the samples (Salter et al., 2014).

Table 6.5: Average Bacteroidetes populations and non-Bacteroidetes recovered from the faeces two weeks after 2mM treatment using qPCR. ‘Non-Bacteroidetes’ were calculated by extrapolation from total bacteria minus Bacteroidetes group.

Bacteroides/total Rat group Non-Bacteroidetes (%) Bacteria (%) Carrier 57.6 42.4 DMSO 50.2 49.8 DPIX 47.2 52.8 Metronidazole 41.3 58.7 P5 54.1 45.9

165 Appendix

Table 6.6: raw sequences from ‘2mM treatment’ of rats and the number of sequences that passed filtering, chimera checking and alignment (‘Aligned’). See Section 5.2.4 for details of how the sequences were processed.

Treatment Rat Raw Count Filtered Aligned 1 9,362 7,805 6,716 2 10,197 8,610 7,018 Carrier 3 9,615 8,017 6,498 4 10,445 8,708 7,341 5 10,180 8,569 7,056 1 10,545 9,065 6,803 2 7,672 6,347 4,990 DMSO 3 10,502 8,877 6,976 4 8,973 7,201 5,873 5 9,125 7,648 6,205 1 11,029 9,129 7,620 2 9,681 8,043 6,372 DPIX 3 8,292 6,886 4,847 4 8,385 7,009 5,422 5 8,859 7,439 5,192 1 8,624 7,459 5,686 2 8,399 7,036 5,589 M 3 8,956 7,393 5,501 4 9,493 7,788 6,570 5 8,371 7,123 5,461 1 9,548 7,916 6,387 2 9,454 7,978 7,055 P5 3 10,378 8,763 7,466 4 9,477 8,071 7,083 5 7,697 6,504 5,143 Total 233,259 195,384 156,870

The Bacteroidetes (average 55.3%) and Firmicutes (average 42.7%) dominated samples, with six other phyla (Proteobacteria, Actinobacteria, Deferribacteres and Lentisphaerae (Table 6.7). A small number of sequences (214) passed quality control filters, mapped to domain Bacteria and were not mapped to a known phylum. The Firmicutes consisted of two groups, mostly Clostridiales (Ruminococcaceae 15.5%; Lachnospiraceae 7.6%) which were generally poorly resolved, for example one clade of sequences were unidentified past Order (10.1%). The Bacilli from Firmicutes contained Lactobacillus reuteri (0.3%) and Lactobacillus spp. (1.3%). The Bacteroidetes were more diverse, consisting mostly of S24-7 (37.3%), Prevotella (10.0%), Bacteroides (4.3%) and Rikenellaceae (1.3%) (Tables 6.7 and 6.8).

166 Appendix

Table 6.7: Summary (in percent) of phylum-level classification of pyrosequencing reads from rat faeces immediately after ‘2mM’ treatment. The standard error of the mean from 5 rats are shown in brackets. Sequences from five phyla (Actinobacteria, Deferribacteres, Lentisphaerae and sequences that were identified as Bacterial, but were unclassified at the Phylum resolution), totalling 0.1-0.3% of reads, are not shown.

Phylum Carrier DMSO DPIX M P5 Average Bacteroidetes 51.0 (±8.5) 63.2 (±3.6) 52.0 (±5.2) 59.7 (±4.7) 50.4 (±5.0) 55.3 Firmicutes 47.5 (±8.7) 35.0 (±3.5) 46.2 (±5.4) 37.9 (±4.9) 46.8 (±5.3) 42.7 Proteobacteria 1.3 (±0.2) 1.6 (±0.2) 1.6 (±0.2) 2.0 (±0.3) 2.7 (±0.4) 1.8 Total 99.8 99.8 99.8 99.7 99.9

The average number of Bacteroidetes was similar with qPCR results, though unlike the qPCR results, the highest population of Bacteroidetes was detected in DMSO treatment and not DPIX (Table 6.7). The Proteobacteria were highest in the P5 treatment, with a 2-3 fold increase in Desulfovibrio, a genus known to use fumarate reductase and sharing 53% homology to the fumarate reductase of B. thetaiotaomicron (Appendix Table A4). However, the total Proteobacterial population was small (2.7%) and the Firmicutes and Bacteroidetes were similar in P5 treatment to untreated rats. The metronidazole treatment did have a slightly low count of Firmicutes (38%) and a relatively high count of Proteobacteria (2%), however, the populations of bacteria in this treatment were similar to DMSO treatment. None of the changes in genera were assessed as significant by a metastats implementation of the student’s T-test. Since the number of bacteria at the genera level were similar, it was concluded that variance in the data was more likely biological and not due to chemical perturbations of the bacterial communities.

167 Appendix

Table 6.8: Summary of bacterial 16S rDNA from the faeces of rats immediately after a 5 day ‘2mM treatment’. M=metronidazole.

Carrier DMSO DPIX M P5 Total Bacterial Clade Av. SEM Av. SEM Av. SEM Av. SEM Av. SEM Av Bacteroidetes 51.0 8.5 63.2 3.6 52.0 5.2 59.7 4.7 50.4 5.0 55.3 Bacteroidales (unclassified family) 0.5 0.2 1.2 0.1 0.8 0.1 0.7 0.1 1.3 0.2 0.9 Bacteroidaceae Bacteroides spp. 4.3 1.4 5.3 0.9 5.0 1.1 5.2 0.9 5.4 0.5 5.0 Bacteroides acidifaciens 0.7 0.2 2.0 0.3 1.7 0.4 1.1 0.1 1.4 0.3 1.4 Bacteroides caccae 0.5 0.2 0.2 0.2 0.0 0.0 0.2 0.1 0.5 0.1 0.3 Bacteroides ovatus 0.0 0.0 0.0 0.0 0.0 0.0 0.2 0.1 0.0 0.0 0.0 Bacteroides uniformis 0.6 0.3 0.4 0.1 0.2 0.1 0.3 0.1 0.3 0.1 0.4 Paraprevotellaceae Prevotella spp. 3.3 0.6 4.6 1.1 4.1 1.6 4.9 0.8 2.8 0.6 4.0 Porphyromonadaceae Parabacteroides spp. 0.5 0.1 0.9 0.1 0.7 0.2 0.6 0.1 0.6 0.1 0.6 Parabacteroides (unclassified species) 0.3 0.1 0.6 0.0 0.5 0.1 0.4 0.1 0.4 0.1 0.4 Parabacteroides distasonis 0.2 0.1 0.3 0.1 0.2 0.1 0.2 0.1 0.2 0.0 0.2 Prevotellaceae Prevotella spp. 4.8 1.3 7.0 0.9 2.9 1.1 10.2 1.7 5.1 0.9 6.0 Rikenellaceae (unclassified genus) 1.1 0.2 1.4 0.1 1.6 0.4 1.0 0.2 1.5 0.5 1.3 S24-7 (unclassified genus) 36.5 5.0 42.7 2.4 36.9 2.5 37.0 2.2 33.6 4.5 37.3 Firmicutes 47.5 8.7 35.0 3.5 46.2 5.4 37.9 4.9 46.8 5.3 42.7 Lactobacillaceae 1.5 0.3 1.7 0.6 0.9 0.2 0.9 0.5 1.3 0.5 1.3 Lactobacillus reuteri 0.3 0.1 0.4 0.2 0.2 0.1 0.2 0.1 0.3 0.1 0.3 Clostridiales 45.9 9.0 33.1 3.7 45.3 5.3 36.8 4.7 45.3 5.2 41.3 Clostridiales (unclassified genus) 9.3 1.2 9.7 2.0 11.7 3.0 7.2 1.2 12.6 2.1 10.1 Christensenellaceae 0.2 0.1 0.4 0.1 0.3 0.1 0.3 0.0 0.2 0.0 0.3 Clostridiaceae 0.5 0.1 1.2 0.3 1.1 0.3 0.6 0.2 1.5 0.8 1.0 Clostridium spp. 0.2 0.1 0.4 0.2 0.9 0.3 0.4 0.1 0.4 0.1 0.5 SMB53 spp. 0.2 0.1 0.8 0.3 0.3 0.2 0.2 0.1 1.1 0.8 0.5 Lachnospiraceae 8.0 1.9 4.1 0.7 10.6 1.6 8.5 2.4 6.9 1.5 7.6 Lachnospiraceae (unclassified genus) 1.8 0.6 1.3 0.5 5.4 1.3 2.7 1.0 2.3 0.6 2.7 Blauta 0.4 0.2 0.2 0.1 0.1 0.0 0.3 0.1 0.2 0.1 0.2 Coprococcus 0.8 0.2 0.4 0.1 1.5 0.3 0.9 0.2 0.5 0.2 0.8 Dorea 0.2 0.1 0.3 0.2 0.1 0.0 0.1 0.0 0.2 0.0 0.2 Ruminococcus gnavus 0.3 0.1 0.1 0.0 0.4 0.2 0.4 0.1 0.2 0.0 0.3 Lachnospiraceae (unclassified genus)* 4.3 1.6 1.8 0.3 3.0 0.7 4.1 1.5 3.3 0.9 3.3 Mogibacteriaceae 0.5 0.1 0.4 0.1 0.2 0.0 0.4 0.1 0.4 0.1 0.4 Mogibacteriaceae (unclassified genus) 0.4 0.1 0.4 0.1 0.2 0.0 0.4 0.1 0.4 0.1 0.4 Peptococcaceae 0.1 0.1 0.2 0.1 0.1 0.0 0.2 0.1 0.3 0.1 0.2 Ruminococcaceae 13.8 0.6 12.1 1.3 19.3 3.5 15.6 3.0 16.9 1.4 15.5 Ruminococcaceae (unclassified genus) 2.6 0.4 3.0 0.8 3.5 1.0 6.0 1.8 4.7 0.5 3.9 Oscillospira spp. 7.1 1.4 5.6 1.6 11.0 2.6 7.1 1.3 9.1 1.6 8.0 Ruminococcus spp. 3.8 0.9 3.4 0.8 4.6 1.2 2.2 0.6 2.8 0.5 3.3 Proteobacteria 1.3 0.2 1.6 0.2 1.6 0.2 2.0 0.3 2.7 0.4 1.8 Alcaligenaceae Sutterella spp. 0.7 0.3 0.8 0.0 0.7 0.2 1.0 0.3 1.3 0.2 0.9 Desulfovibrionaceae 0.4 0.2 0.6 0.1 0.6 0.2 0.7 0.3 1.1 0.2 0.7 Desulfovibrio spp. 0.3 0.1 0.5 0.1 0.5 0.1 0.5 0.3 1.0 0.2 0.6 Actinobacteria 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.1 Deferribacteres 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Lentisphaerae Victivallaceae 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Unclassified Phylum 0.1 0.0 0.1 0.0 0.1 0.1 0.3 0.1 0.0 0.0 0.1

*This clade of Lachnospiraceae (mean 4.3%) was sufficiently divergent (more than 3% dissimilar) to the other clade of Lachnospiraceae (mean 1.8%)

168 Appendix

6.3.2 Optimising P5 and metronidazole treatment concentrations

From pyrosequencing results, it was apparent that initial 2 mM treatment was insufficient to kill susceptible nanaerobes. A thorough culture based assessment of the microbiology after using ‘high-dose’ metronidazole in the rat gastrointestinal tract detected decreases in nanaerobic cocci and other nanaerobes from the colonic mucosa which were not observed in faeces (Mikelsaar and Siigur, 1992). In an initial expeirment using similar ‘high-dose’ five-day metronidazole protocol, no difference in daily faecal texture or consistency was observed over six days. Also, 3-4 log changes in the viable count of susceptible nanaerobes were not detected on kanamycin and vancomycin containing BHI medium (results not shown). These results confirmed that an examination the colonic mucosa was warranted. To achieve these results, Mikelsaar and Siigur administered high concentrations of metronidazole (70 mg kg-1 day-1, equating to a total of 4.9 g per day for a 70 kg human). For 5 rats weighing 100 g, a similar protocol requires around 800 mg of P5. Sufficient P5 was synthesised for these experiments (see Table 6.9).

Rats require an initial one week acclimatisation before experiments can begin. During this period, the weight of rats within each treatment group was similar (Fig. 6.2). After five days administering ‘250 mM’ treatments, the percent weight increase over baseline between groups was similar [metronidazole (94.5%), carrier (92.4%), DPIX (88.1%), and P5 (72.1%), Fig. 6.2]. Although growth rate of P5 treatment was the lowest, this group also had the lowest growth rate before P5 was administered (Fig. 6.2). In all groups, weight gain plateaued by around 10 days. Strikingly during surgery, it was noted the caecum of metronidazole treated rats was distended, doubling in size/volume (4.63 ±0.3g) compared with other treatment groups (Fig. 6.3). The caecum size/volume was similar in other groups: carrier control (2.12 ±0.2 g), DPIX (2.12 ±0.1g) and P5 (2.18 ±0.2) (Fig. 6.3). The caecum of metronidazole-treated rats also had a less pronounced vasculature (Fig. 6.3). No other differences in gross anatomy of rats were observed.

Table 6.9: The total amount of daily ‘250 mM’ treatments given for five days to rats weighing ≈80 g. Rats were 3-4 weeks old at time of treatment administration. Treatments were administered at 12 hour intervals in 150 µL of CMC with 250 mM DPIX, metronidazole or P5.

Treatment mg day-1 mg kg-1 day-1 Equivalent for a 70 kg human g-1 day-1 Deuteroporpyrhin IX 14.6 182.3 12.8 Metronidazole 4.3 53.5 3.7 P5 16.6 207.3 14.5

169 Appendix

100 CARRIER 100 DPIX 80 80 60 40 60

g baseline(%) Weight (g) Weight g 20 % %

40 0 Weight increase from from increase Weight 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

100 100 M P5 80 80 60 40 60

g baseline(%)

Weight (g) Weight 20 g % %

40 0 from increase Weight 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 Time (d) Time (d)

Figure 6.2: Rat weight during a 7-day acclimatisation followed by a 5-day treatment with CMC (carrier), DPIX, metronidazole (M) and P5. Treatments were administered twice daily on days 7-12. Rats were culled on day 13 for further examination. Rats were 3-4 weeks old at day 0. Error bars show standard error of the mean from five rats within each treatment group.

Anatomical changes at a finer resolution were next examined. In relation to the carrier control, the large intestine of the metronidazole, DPIX and P5 treatments all exhibited an increase in lymphoid follicles, particularly adjacent to the crypts. The metronidazole treatment also exhibited goblet cell hyperplasia, and an increase in infiltrating leukocytes (see Fig. 6.4). Due to time constraints an exploration into the microbiology associated with colonic mucosa was not possible. Nevertheless, these results provide impetus to assess microbiological as well as key changes in host epithelial expression, discussed further in future directions (Chapter IV).

170 Appendix

A

B

C

D

Figure 6.3: Caecum of rats on day 13 immediately after a 5-day treatment regime. Treatment groups included metronidazole (A), P5 (B), carrier (C) and DPIX (D). The numbers at the bottom of the figure are in cm.

171 Appendix

C DPIX

* ** **

* * *

M P5

Figure 6.3: Representative sections from the colonic mucosa of rats after a 5-day high dose of carrier (‘C’), DPIX, metronidazole and P5. Scale bar=50 M. Slight and pronounced goblet cell hyperplasia are marked with one or two blue asterisks respectively. Infiltrating leukocytes are marked with yellow asterisks.

6.4 DISCUSSION

Chapter VI used a rat gastrointestinal model to test the predicted metabolism of P5 based on nitroimidazole metabolism and b-cytochrome formation outlined in Chapter III. Real time PCR comparisons of total bacteria in ‘2mM treatment’ and Bacteroidetes populations indicated that insufficient metronidazole was administered to rats to broadly alter the number of susceptible

172 Appendix bacteria. The Bacteroidetes were chosen for assessment since they represent a large portion of the rat faecal bacteria and the majority of the genus within this clade are sensitive to metronidazole. However, when bacterial populations were assessed by 454 pyrosequencing, the number of nanaerobic bacteria susceptible to metronidazole did not appreciably change, indicating treatment concentrations were too low. The concentration of administered metronidazole was increased accordingly, as described below.

For initial ‘2mM treatment’ sequencing, the Greengenes 16S rDNA taxonomy was chosen to map sequences to, since it is known to contain candidate phyla (new phyla recalcitrant to culturing). Indeed, using the RDP reference 16S rDNA taxonomy, many more reads mapped to unknown phyla (5%). However, a significant portion of the reads mapped to genera S24-7 (average 37.3%), whereas using the RDP reference, a large number of these reads mapped to unclassified Porphyromonadaceae (29%) and Barnsiella (4.1% a Porphyromonadaceae). Using the RDP classification, no significant changes were detected between treatment groups of these clades, so the discrepancy in mapping sequences between RDP and Greengenes classifications was not checked with in silico suites such as BLAST. Otherwise, the Firmicutes were mapped with a similar taxonomic distribution and like the Greengenes 16S rDNA taxonomy reference, the resolution of the RDP database was poor for this clade.

As mentioned, current molecular profiling studies of bacteria in the faeces of rats are limited by severe lack of sequencing depth or by short pyrosequencing reads. Whilst faecal estimations of bacteria are a rough approximation, pyrosequencing of the ‘2mM treatment’ experiments established that previously unidentified clades (Paraprevotellaceae, Rikenellaceae, Barnsiella/S24-7 and Oscillospira spp. reside in rats. Importantly, this experiment also established that rat faecal Bacteroides spp. populations are lower than in human counterparts. Nevertheless, 454 sequencing and real time PCR are sufficiently sensitive to detect more than a 100-fold reduction in this clade. Also, there was a surprisingly high variation (2-3 fold) in the number of major bacterial clades of rats housed in one cage. This variation could be due to environmental factors (diet, extent of coprophagy), handling of microbes (DNA extraction) and molecular bias (PCR amplification and sequencing). Using the improved DNA extraction method may help to determine if the apparent variability within rat groups was biological or an artefact of the original DNA extraction. Ultimately, establishing the types of bacteria and variability within these clades will establish a sound basis for the concentrated administration of P5.

Increasing metronidazole treatment concentration from ‘250 mM’ (a daily dose of 53.5 mg kg-1 metronidazole, which was a 170-fold increase over initial ’2mM’ treatments) had minimal

173 Appendix effect on the weight of treated rats. However, a profound impact on the gastrointestinal tract was observed, namely a distended caecum that was double the volume of other treatment groups. The caecum of metronidazole-treated rats also had a less pronounced vasculature. No other differences in gross anatomy of rats were observed. Sections of the large intestine of the metronidazole, DPIX and P5 treatments all exhibited an increase in lymphoid follicles, particularly adjacent to the crypts. The metronidazole treatment also exhibited goblet cell hyperplasia. Due to time constraints, the microbiology associated with such changes remains to be detailed. Nevertheless, rats exhibited no obvious changes in grooming or sociability when administered with a concentrated dose of P5, demonstrating that even with high doses, there is no apparent toxicity. Profound changes in tissue architecture warrant further investigation, as described in future directions (Chapter IV).

174 Appendix

6.5 Appendix Chapter VI References

BROOKS, S. P. J., MCALLISTER, M., SANDOZ, M. & KALMOKOFF, M. L. 2003. Culture-independent phylogenetic analysis of the faecal flora of the rat. Canadian Journal of Microbiology, 49, 589-601. CLAESSON, M. J., WANG, Q., O'SULLIVAN, O., GREENE-DINIZ, R., COLE, J. R., ROSS, R. P. & O'TOOLE, P. W. 2010. Comparison of two next-generation sequencing technologies for resolving highly complex microbiota composition using tandem variable 16S rRNA gene regions. Nucleic acids research, gkq873. HALEBIAN, S., HARRIS, B., FINEGOLD, S. M. & ROLFE, R. D. 1981. Rapid method that aids in distinguishing Gram-positive from Gram-negative anaerobic bacteria. Journal of Clinical Microbiology, 13, 444-448. HUSE S.M., WELCH D.M., H.G., M. & M.L., S. 2010. Ironing out the wrinkles in the rare biosphere through improved OTU clustering. Environ Microbiol, 12, 1889-1898. INOUE, R. & USHIDA, K. 2003. Vertical and horizontal transmission of intestinal commensal bacteria in the rat model. FEMS Microbiology Ecology, 46, 213-219. JORDAN, D. C. 1961. Effect of vancomycin on the synthesis of the cell wall mucopeptide of Staphylococcus aureus. Biochemical and Biophysical Research Communications, 6, 167-170. KIANOUSH, N., ADLER, C. J., NGUYEN, K.-A. T., BROWNE, G. V., SIMONIAN, M. & HUNTER, N. 2014. Bacterial Profile of Dentine Caries and the Impact of pH on Bacterial Population Diversity. PLoS ONE, 9, e92940. METZKER, M. L. 2010. Sequencing technologies—the next generation. Nature reviews genetics, 11, 31-46. MIKELSAAR, M. & SIIGUR, U. 1992. Metronidazole and the intestinal microecology of rats. Microbial ecology in health and disease, 5, 139-146. MISUMI, M. & TANAKA, N. 1980. Mechanism of inhibition of translocation by kanamycin and viomycin: a comparative study with fusidic acid. Biochemical and biophysical research communications, 92, 647-654.

NADKARNI, M. A., BROWNE, G. V., CHHOUR, K. L., BYUN, R., NGUYEN, K. A., CHAPPLE, C. C., JACQUES, N. A. & HUNTER, N. 2012. Pattern of distribution of Prevotella species/phylotypes associated with healthy gingiva and periodontal disease. European Journal of Clinical Microbiology & Infectious Diseases, 31, 2989-2999.

175 Appendix

NADKARNI, M. A., MARTIN, F. E., JACQUES, N. A. & HUNTER, N. 2002. Determination of bacterial load by real-time PCR using a broad-range (universal) probe and primers set. Microbiology- Sgm, 148, 257-266. ROGOSA, M., FITZGERALD, R. J., MACKINTOSH, M. E. & BEAMAN, A. J. 1958. IMPROVED MEDIUM FOR SELECTIVE ISOLATION OF VEILLONELLA. Journal of Bacteriology, 76, 455-456. SALTER, S. J., COX, M. J., TUREK, E. M., CALUS, S. T., COOKSON, W. O., MOFFATT, M. F., TURNER, P., PARKHILL, J., LOMAN, N. J. & WALKER, A. W. 2014. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC biology, 12, 87. SCHLOSS, P. D. & WESTCOTT, S. L. 2011. Assessing and Improving Methods Used in Operational Taxonomic Unit-Based Approaches for 16S rRNA Gene Sequence Analysis. Applied and Environmental Microbiology, 77, 3219-3226. UMEZAWA, H., UEDA, M., MAEDA, K., YAGISHITA, K., KONDO, S., OKAMI, Y., UTAHARA, R., OSATO, Y., NITTA, K. & TAKEUCHI, T. 1957. Production and isolation of a new antibiotic: kanamycin. The Journal of antibiotics, 10, 181. WAITZ, J. A., MOSS, E. L., DRUBE, C. G. & WEINSTEIN, M. J. 1972. Comparative Activity of Sisomicin, Gentamicin, Kanamycin, and Tobramycin. Antimicrobial Agents and Chemotherapy, 2, 431- 437. WHITE, J. R., NAGARAJAN, N. & POP, M. 2009. Statistical methods for detecting differentially abundant features in clinical metagenomic samples. PLoS Comput Biol, 5, e1000352. WLODARSKA, M., WILLING, B., KEENEY, K., MENENDEZ, A., BERGSTROM, K., GILL, N., RUSSELL, S., VALLANCE, B. & FINLAY, B. 2011. Antibiotic treatment alters the colonic mucus layer and predisposes the host to exacerbated Citrobacter rodentium-induced colitis. Infection and immunity, 79, 1536-1545.

176