EFFECTS OF DIETARY FIBER SOURCES ON THE GASTRO-INTESTINAL MICROBIOTA, FERMENTATION METABOLITES, AND LISTERIA MONOCYTOGENES IN VIVO AND IN VITRO By Ryan Walker Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Human Nutrition – Doctor of Philosophy A DISSERTATION 2019 By Ryan Walker ABSTRACT EFFECTS OF DIETARY FIBER SOURCES ON THE GASTRO-INTESTINAL MICROBIOTA, FERMENTATION METABOLITES, AND LISTERIA MONOCYTOGENES IN VIVO AND IN VITRO Dietary fiber consumption influences the gastro-intestinal microbiota, gastro- intestinal function, and health. This research investigated the effects of dietary fiber sources on bacterial composition, short-chain fatty acid (SCFA) production, gastro- intestinal barrier function, and Listeria monocytogenes challenge in mice and human colon cells. L. monocytogenes causes life-threatening illness in humans and animals. Some dietary fibers protect animals against illness in models of infection, while others enhance infection. Therefore, it is unclear if consuming certain dietary fiber sources confers protection against foodborne illnesses. The objectives of this research were: 1) describe how dietary fiber sources affect bacterial composition in vivo and in vitro; 2) quantify SCFAs produced by bacterial fermentation; 3) determine if fiber sources differentially affect L. monocytogenes infection in vivo; and 4) determine if bacterial metabolites promote or inhibit L. monocytogenes infection by affecting in vitro epithelial barrier function. In this project, dietary fiber sources differentially affected gastro-intestinal bacterial composition in vivo and in vitro, L. monocytogenes infection in mice, and enhanced barrier integrity in vitro. The overall results of this project demonstrate that dietary fiber sources differentially influence certain gastro-intestinal bacterial populations, measures of diversity, and that resulting compositional changes are important in the pathogenicity of L. monocytogenes. Additionally, fermentation end products enhance gastro-intestinal barrier function in this in vitro model, which may be an important factor in the prevention of enteric infection in humans. In summary, both the gastro-intestinal microbiota and its metabolites are important factors for maintaining gastro-intestinal health. Copyright by RYAN WALKER 2019 This accomplishment is dedicated to the United States Marine Corps and my daughter, Syailendra. I am forever grateful to the Marine Corps for developing me into the disciplined and committed person I am today. Syailendra, you have been my inspiration while completing this work. Thank you. Semper fidelis. v ACKLOWLEDGEMENTS I am fortunate and grateful for the opportunity to learn from my advisor, Dr. Les Bourquin, and to undertake this academic training. My dream to become a scientist was ignited as a teenager, and I took a very non-traditional path to complete this work and to reach this professional accomplishment. I would like to sincerely thank him for his sage advice, patience, open mind, and friendship. We share many professional interests and ideas, and you have taught and inspired me to integrate many of them in ways that are useful. I am also grateful for the advice and mentorship of my guidance committee: Dr. Jenifer Fenton, Dr. Linda Mansfield, and Dr. James Pestka. Thank you for your encouragement, helpful suggestions, advice, technical support, and for challenging me during this work. I am also grateful for my friendship with Dr. Loan Cao. You were always one-step ahead of me in terms of our research timelines. I am grateful for your technical advice. To all of the undergraduates who helped with this research, thank you. Most importantly, I would like to thank my family. To my wife, Verna: thank you for leaving paradise and joining me in Michigan. My life was forever changed when we met in Malang, and I am blessed to have you in my life. To my sister, Vanessa: thank you for being supportive. You are a great sister and aunt. vi TABLE OF CONTENTS LIST OF TABLES ..................................................................................... x LIST OF FIGURES ................................................................................. xii KEY TO ABBREVIATIONS .................................................................... xiv CHAPTER 1: INTRODUCTION, CHAPTER SUMMARIES, SIGNIFICANCE AND LITERATURE REVIEW .................................................................... 1 INTRODUCTION ................................................................................................. 1 CHAPTER SUMMARIES ..................................................................................... 3 SIGNIFICANCE.................................................................................................... 5 LITERATURE REVIEW ....................................................................................... 7 Dietary fiber and microbial fermentation metabolites .................................. 7 Gastro-intestinal microbiota ......................................................................... 11 L. monocytogenes and bacterial infection models ...................................... 15 Gastro-intestinal barrier function ............................................................... 20 REFERENCES .................................................................................................... 25 CHAPTER 2: EFFECT OF DIETARY FIBER SOURCES ON MICROBIAL POPULATIONS AND LISTERIA MONOCYTOGENES IN MICE ............... 36 ABSTRACT ......................................................................................................... 36 INTRODUCTION .............................................................................................. 38 MATERIALS AND METHODS .......................................................................... 41 Mouse model and experimental design ....................................................... 41 Phosphate buffered saline and L. monocytogenes challenge protocol ....... 43 DNA preparation for 16S rRNA gene sequencing ........................................ 45 16S rRNA gene sequence processing ............................................................ 46 16S rRNA gene sequence analysis ................................................................ 46 Statistical analysis ........................................................................................ 47 RESULTS ............................................................................................................ 49 Infection classification and exclusion criteria.............................................. 49 Mouse weights .............................................................................................. 49 Cecum and cecum contents as percent of body weight................................ 51 Infection status of L. monocytogenes challenged mice ............................... 53 Bacterial diversity of mice ............................................................................ 58 Phyla relative abundances ............................................................................ 64 Genera relative abundances ......................................................................... 67 Correlation analysis of cecum bacterial taxa with cecum L. monocytogenes counts ............................................................................. 70 Correlation analysis of cecum bacterial taxa with cecum Shannon diversity ......................................................................................... 70 DISCUSSION...................................................................................................... 72 APPENDICES .................................................................................................... 82 APPENDIX A: MOTHUR MISEQ WORKFLOW ........................................ 83 vii APPENDIX B: MOUSE DATA BY CAGE ..................................................... 85 REFERENCES .................................................................................................. 100 CHAPTER 3: EFFECT OF DIETARY FIBER SOURCES ON MICROBIAL POPULATIONS AND SHORT-CHAIN FATTY ACID PRODUCTION IN VITRO ............................................................................................. 108 ABSTRACT ....................................................................................................... 108 INTRODUCTION ............................................................................................. 110 MATERIALS AND METHODS .........................................................................113 Research participants ..................................................................................113 Study design .................................................................................................113 Human fecal sample collection .................................................................. 114 Anaerobic dilution solution preparation ..................................................... 115 Semi-defined anaerobic medium preparation ........................................... 116 Gas volume determination .......................................................................... 117 Preparation of fermentation end product aliquots and bacterial pellets .. 118 pH determination ....................................................................................... 118 SCFA gas chromatography analysis ........................................................... 118 DNA preparation for 16S rRNA gene sequencing ...................................... 119 16S rRNA gene sequencing ......................................................................... 120 16S rRNA gene sequence analysis .............................................................. 121 Statistical analysis ...................................................................................... 121 RESULTS .......................................................................................................... 123 Dietary assessment of energy and nutrient intake ..................................... 123 Gas volume after 24 h dietary fiber source fermentation .......................... 125 pH of fermentation end products ............................................................... 126 SCFA composition of fermentation end products...................................... 127 Associations between gas volume, pH, and SCFAs .................................... 130 16S rRNA gene sequencing and bacterial diversity ................................... 130 Bacterial composition of 24 h fermentations ............................................. 137 Correlation analysis of predominant phyla and genera with SCFAs ......... 141 DISCUSSION.................................................................................................... 144 APPENDICES ................................................................................................... 155 APPENDIX A: GAS CHROMATOGRAPHY CALIBRATION STANDARDS .............................................................................................. 156 APPENDIX B: MOTHUR MISEQ WORKFLOW ....................................... 157 APPENDIX C: MOCK COMMUNITY COMPOSITION ............................. 159 REFERENCES .................................................................................................. 160 CHAPTER 4: EFFECT OF IN VITRO FERMENTATION END PRODUCTS ON EPITHELIAL CELL BARRIER FUNCTION AND APICAL LISTERIA MONOCYTOGENES CHALLENGE ........................................................ 167 ABSTRACT ....................................................................................................... 167 INTRODUCTION ............................................................................................. 170 MATERIALS AND METHODS ........................................................................ 173 viii Cell culture .................................................................................................. 173 Monolayer resistance of HT29-MTX-E12 cells .......................................... 174 FITC-D4 permeability of HT29-MTX-E12 cells ......................................... 175 Mucin quantification .................................................................................. 175 L. monocytogenes inoculum preparation .................................................. 176 Monolayer resistance of L. monocytogenes challenged cells .................... 177 FITC-D4 permeability of L. monocytogenes challenged cells ................... 177 Mucin quantification in L. monocytogenes challenged cells ..................... 177 Enumeration of cell-associated and internalized L. monocytogenes in cell culture ............................................................................................. 178 Cytotoxicity ................................................................................................. 179 Statistical analysis ...................................................................................... 179 RESULTS .......................................................................................................... 181 Effects of fiber fermentation end products on cell monolayer integrity ... 181 Effects of fermentation end products on L. monocytogenes ..................... 185 Correlation analysis of fermentation end product SCFA composition with monolayer barrier integrity ................................................................ 187 Correlation analysis of bacterial taxa with monolayer barrier integrity ... 188 Correlation analysis of L. monocytogenes counts with SCFAs ................. 190 DISCUSSION.................................................................................................... 192 APPENDICES .................................................................................................. 200 APPENDIX A: CELL CYTOTOXICITY ....................................................... 201 APPENDIX B: FERMENTATION END PRODUCT CONCENTRATION DETERMINATION ASSAYS .................................... 202 APPENDIX C: MUC5AC PRELIMINARY ASSAY..................................... 205 REFERENCES ................................................................................................. 206 CHAPTER 5: SUMMARY AND FUTURE STUDIES ................................ 212 SUMMARY ....................................................................................................... 212 FUTURE DIRECTIONS ................................................................................... 214 Study 1: Gastro-intestinal composition of enteric infection cases ............. 214 Study 2: Gastro-intestinal redox potential and diet .................................. 215 Study 3: Effect of dietary fiber consumption on mouse mucus layer and barrier integrity .................................................................................. 217 REFERENCES ................................................................................................. 220 ix Table 2.1 AIN-93G and modified AIN-93G diet compositions .................................... 42 LIST OF TABLES Table 2.2 Shannon diversity index by diet and digesta sources ................................... 59 Table 2.3 Effective number of species by diet and digesta sources .............................. 61 Table 2.4 Relative abundances (%) of small intestine digesta phyla by diet ............... 65 Table 2.5 Relative abundances (%) of cecum digesta phyla by diet ............................. 66 Table 2.6 Relative abundances (%) of colon digesta phyla by diet ............................... 67 Table 2.7 Relative abundances (%) of small intestine digesta genera by diet ............. 68 Table 2.8 Relative abundances (%) of cecum digesta genera by diet ........................... 69 Table 2.9 Relative abundances (%) of colon digesta genera by diet............................. 70 Table 2.10A Shannon diversity index by diet and digesta sources ............................... 94 Table 2.11A Relative abundances (%) of small intestine digesta phyla by diet ............ 94 Table 2.12A Relative abundances (%) of cecum digesta phyla by diet ......................... 95 Table 2.13A Relative abundances (%) of colon digesta phyla by diet........................... 95 Table 2.14A Relative abundances (%) of small intestine digesta genera by diet ......... 96 Table 2.15A Relative abundances (%) of cecum digesta genera by diet ....................... 97 Table 2.16A Relative abundances (%) of colon digesta genera by diet ....................... 98 Table 2.17A Correlation analysis of cecum taxa with cecum Shannon diversity ......... 99 Table 2.18A Correlation analysis of cecum taxa with cecum L. monocytogenes counts ............................................................................................................................ 99 Table 3.1 Anaerobic dilution solution composition .................................................... 116 Table 3.2 Semi-defined anaerobic medium composition ............................................ 117 Table 3.3 Participant energy and nutrient intake of the 24 h period preceding fecal sample collection ......................................................................................................... 124 x Table 3.4 SCFA composition of feces and 24 h in vitro fermentations by fiber source ........................................................................................................................... 129 Table 3.5 Relative abundances (%) of predominant phyla by feces and fiber source ........................................................................................................................... 138 Table 3.6 Relative abundances (%) of predominant genera by feces and fiber source ........................................................................................................................... 141 Table 3.7 Correlation analysis between phyla relative abundances (%) and SCFA concentration ............................................................................................................... 142 Table 3.8 Correlation analysis between genera relative abundances (%) and SCFA concentration ............................................................................................................... 143 Table 3.9A Gas chromatography calibration curve standards ................................... 156 Table 3.10A Mock community composition ............................................................... 159 Table 4.1 Short-chain fatty composition of fermentation end products from five dietary fibers source fermented for 24 h by human fecal inoculum ................... 174 Table 4.2 Correlation analysis of SCFA concentration and cell monolayer integrity ....................................................................................................................... 188 Table 4.3 Correlation of relative abundances (%) of phyla identified in the production of fermentation end products by 16S rRNA gene sequencing with monolayer integrity ............................................................................................. 189 Table 4.4 Correlation of relative abundances (%) of genera identified in the production of fermentation end products by 16S rRNA gene sequencing with monolayer integrity ............................................................................................. 190 Table 4.5 Correlation analysis of SCFA concentration and L. monocytogenes counts .......................................................................................................................... 191 xi LIST OF FIGURES Figure 2.1 Experimental timeline.................................................................................. 43 Figure 2.2 Mouse weights by diet and age ....................................................................50 Figure 2.3 Mouse weights by sex and age ..................................................................... 51 Figure 2.4 Mouse wet cecum proportion by diet .......................................................... 52 Figure 2.5 Mouse wet cecum proportion by sex ........................................................... 53 Figure 2.6 Positive L. monocytogenes samples in challenged mice by diet ................. 54 Figure 2.7 Cecum L. monocytogenes counts of challenged mice by diet and sex ........ 55 Figure 2.8 Cecum digesta L. monocytogenes counts of challenged mice by diet ........ 56 Figure 2.9 Spleen L. monocytogenes counts of challenged mice by diet and sex........ 57 Figure 2.10 Spleen L. monocytogenes counts of challenged mice by diet ................... 58 Figure 2.11 NMDS plot of Θ-YC distances of small intestine digesta by diet ............... 62 Figure 2.12 NMDS plot of Θ-YC distances of cecum digesta by diet ............................ 63 Figure 2.13 NMDS plot of Θ-YC distances of colon digesta by diet ............................. 64 Figure 2.14A Mouse weights by diet and age ................................................................ 85 Figure 2.15A Mouse weights by sex and age ................................................................ 86 Figure 2.16A Mouse wet cecum proportion by diet ...................................................... 87 Figure 2.17A Mouse wet cecum proportion by sex ...................................................... 88 Figure 2.18A Positive L. monocytogenes samples in challenged mice by diet ............ 89 Figure 2.19A Cecum L. monocytogenes counts of challenged mice by diet and sex .. 90 Figure 2.20A Cecum digesta L. monocytogenes counts of challenged mice by diet ... 91 Figure 2.21A Spleen L. monocytogenes counts of challenged mice by diet and sex ... 92 Figure 2.22A Spleen L. monocytogenes counts of challenged mice by diet ................ 93 xii Figure 3.1 Gas volume by fiber source after 24 h fermentation ................................. 125 Figure 3.2 pH by fiber source after 24 h fermentation ............................................... 126 Figure 3.3 Shannon diversity index of 24 h fermentations by donor excluding feces ............................................................................................................................. 132 Figure 3.4 Shannon diversity index of 24 h fermentations by donor including feces ............................................................................................................................. 132 Figure 3.5 NMDS plot of Θ-YC distances by fermentation fiber source .................... 134 Figure 3.6 NMDS plot of Θ-YC distances by feces and fermentation fiber source .... 135 Figure 3.7 NMDS plot of bacterial communities by donor excluding fecal samples ........................................................................................................................ 136 Figure 3.8 NMDS plot of bacterial communities by donor including fecal samples ........................................................................................................................ 137 Figure 4.1 Cell monolayer TEER relative to DMEM by fermentation fiber source ... 182 Figure 4.2 Absolute TEER by fermentation fiber source ........................................... 183 Figure 4.3 FITC-D4 flux relative to DMEM by fermentation fiber source ................ 184 Figure 4.4 Absolute FITC-D4 flux by fermentation fiber source ............................... 185 Figure 4.5 Cell-associated L. monocytogenes counts by fermentation fiber source ........................................................................................................................... 186 Figure 4.6 Internalized L. monocytogenes counts by fermentation fiber source ...... 187 Figure 4.7A Cytotoxicity of fermentation end products in unchallenged HT29-MTX-E12 cells after 48 h exposure .................................................................. 201 Figure 4.8A Preliminary evaluation of fermentation end product dilutions on TEER response ........................................................................................................... 202 Figure 4.9A Preliminary evaluation of fermentation end product dilutions on FITC-D4 flux response ............................................................................................... 203 Figure 4.10A Preliminary cytotoxicity evaluation of fermentation end product dilutions ...................................................................................................................... 204 Figure 4.11A Preliminary MUC5AC assay .................................................................. 205 xiii AIN-93G AMOVA ASA24 BHI CaCl2-H20 CAZymes CFU CoCl2-6H20 DMEM DNA EDTA ENS ESPC FeSO4-7H2O FITC-D4 FOS GOS GPCR GRAS HCl HDAC IL-6 KEY TO ABBREVIATIONS American Institute of Nutrition-93G Analysis of Molecular Variance National Cancer Institute’s Automated Self-Administered 24-Hour Dietary Assessment Brain Heart Infusion Calcium Chloride Dihydrate Carbohydrate-active Enzymes Colony Forming Unit Cobalt(II) Chloride Hexahydrate Dulbecco's Modified Eagle Medium Deoxyribonucleic Acid Ethylenediaminetetraacetic Acid Effective Number of Species Estimated Standard Plate Count Ferrous Sulfate Heptahydrate Fluorescein Isothiocyanate-Dextran 4 Fructo-oligosaccharide Galacto-oligosaccharide G-protein coupled receptors Generally Recognized As Safe Hydrochloric Acid Histone Deacetylase Interleukin-6 xiv InlA InlB IQR K2HPO4 KH2PO4 LLO Ln LS LSD MgCl-6H20 MnCl2-4H20 MOX MUC Na2CO3 NaCl NaOH (NH4)2SO4 NMDS OTU PBS pH rRNA SCFA TEER Internalin-A Internalin-B Interquartile Range Dipotassium Hydrogenphosphate Potassium Dihydrogen Phosphate Listeriolysin O Natural Logarithm Least Square Fisher’s Least Significant Difference Magnesium Chloride Hexahydrate Manganese(II) Chloride Tetrahydrate Modified Oxford Mucin Sodium Carbonate Sodium Chloride Sodium Hydroxide Ammonium Sulfate Non-metric Multidimensional Scaling Operational Taxonomic Unit Phosphate Buffered Saline Potential Hydrogen Ribosomal Ribonucleic Acid Short-chain Fatty Acid Transepithelial Electrical Resistance xv XOS ZnSO4-7H2O Θ-YC Xylo-oligosaccharide Zinc Sulfate Heptahydrate Yue and Clayton Θ xvi CHAPTER 1: INTRODUCTION, CHAPTER SUMMARIES, SIGNIFICANCE AND LITERATURE REVIEW INTRODUCTION Diet strongly influences the gastro-intestinal microbiota responsible for harvesting energy from dietary fiber sources and other undigested nutrients transiting the human gastro-intestinal tract (Walker et al., 2011; David et al., 2014; De Filippo et al., 2017). Dietary fibers are plant and analogous carbohydrates resistant to digestion by humans and other mammals. After consumption, they reach the large intestine undigested and are by metabolized by micro-organisms. They include polysaccharides, oligosaccharides, lignin, and other plant-associated substances. Dietary fiber consumption can lead to improvements in laxation, blood cholesterol levels, and blood glucose homeostasis (DeVries, 2001); and is protective against cardiovascular disease, type II diabetes, and colorectal cancer (Anderson et al., 2009; Threapleton et al., 2013; Cho et al., 2013; World Cancer Research Fund and American Institute for Cancer Research, 2011). In humans, bacterial fermentation of dietary fiber to short-chain fatty acids (SCFA), which are then absorbed and metabolized, accounts for about 10% of dietary energy (McNeil, 1984). Diets rich in fiber sources are thought to protect against enteric bacterial infection in humans (De Filippo et al., 2010), and fiber-rich food relieves clinical symptoms in children with enteric infections (Rabbani, 2009). Certain dietary fiber sources such as inulin, xylo-oligosaccharide (XOS), and galacto- oligosaccharide (GOS) protect animals from bacterial infection (Buddington et al., 2002; Ebersbach et al., 2010). On the other hand, certain fibers sources such as fructo- oligosaccharide (FOS), XOS, inulin, and apple pectin promote pathogenesis in animals (Petersen et al., 2009; Ebersbach et al., 2010). One hypothesis for these disparate effects 1 is that certain dietary fibers modulate the gut microbiota composition to a so-called healthy state, which is associated with decreased intestinal permeability and markers of inflammation (Cani et al., 2009). Listeria monocytogenes is a Gram-positive environmental organism found in high moisture soils and pastures, as well as food processing environments. It rarely causes serious illness in healthy humans, but is responsible for the deadly foodborne illness, listeriosis, in high-risk groups (Hoelzer et al., 2012). The intersection of dietary fiber metabolism by gastro-intestinal bacteria and L. monocytogenes infection in vivo and in vitro is the focus of the present work. This research is a systematic study evaluating the efficacy of dietary fibers to modulate gastro-intestinal bacterial composition, bacterial fermentation metabolites, gastro-intestinal barrier function, and potential susceptibility to listeriosis. Whether dietary fiber sources promote or inhibit listeriosis remains to be determined. The objectives of this research were to evaluate the effects of gum arabic, arabinoxylan cellulose, guar gum and XOS on gastro-intestinal bacterial composition, barrier function, SCFA composition, and L. monocytogenes infection in vivo and in vitro. I hypothesized that these dietary fiber sources, due to differences in structure and carbohydrate composition, would have differential effects on gastro-intestinal bacterial composition, SCFA concentrations and composition, barrier function, and L. monocytogenes association in in vivo and in vitro models of listeriosis. 2 CHAPTER SUMMARIES Chapter 1 presents the motivation behind my research in Dr. Les Bourquin’s laboratory, followed by a thorough review of the literature related to dietary fiber, SCFAs, the gastro-intestinal microbiota, L. monocytogenes, and gastro-intestinal barrier function. Chapter 2 presents the findings of in vivo research modeling the effects of dietary fiber sources on mouse gastro-intestinal bacterial composition, L. monocytogenes challenge, and relationships between gastro-intestinal bacterial composition and L. monocytogenes. The working hypothesis of this research is that dietary fiber sources differentially affect mouse gastro-intestinal bacterial composition, and these differences influence L. monocytogenes pathogenesis in a model of human listeriosis. To test this hypothesis, mice were fed five dietary fiber sources for 2 wk and then challenged phosphate buffered saline or L. monocytogenes. After 7 d, mice were sacrificed, gastro-intestinal bacterial 16S ribosomal ribonucleic acid (rRNA) genes were sequenced, and L. monocytogenes was enumerated in digesta and internal organs. This research presents evidence that dietary fiber sources differentially affect bacteria composition, diversity measures, and L. monocytogenes infection in vivo. Chapter 3 presents the findings of in vitro dietary fiber fermentation using human feces as the bacterial inoculum source. The working hypothesis of this research is that dietary fiber sources differentially affect human gastro-intestinal bacterial composition and SCFAs produced by bacterial fermentation. To test this hypothesis, bacteria from feces of healthy human volunteers were used to ferment five dietary fiber sources in vitro for 24 h. 16S rRNA bacterial genes from feces and fermentation bioreactors were used to detect bacterial compositional changes, and SCFAs were 3 analyzed by gas chromatography. This research demonstrated that dietary fiber sources differentially affect the composition of human gastro-intestinal bacteria and the SCFAs produced by fermentation, but have a limited effect on bacterial diversity. Chapter 4 presents the findings of in vitro cell culture experiments, which were designed to evaluate the effect of fermentation end products on gastro-intestinal barrier function. The working hypothesis of this research is that barrier function is differentially affected by the SCFA composition of fermentation end products, and differences in barrier function will affect L. monocytogenes association with and internalization in human colon cells. To test this hypothesis, HT29-MTX-E12 cells were seeded on Transwell inserts and treated with fermentation end products produced in the work presented in chapter 3. This research presents evidence that fermentation end products differentially affect barrier function, but there is limited evidence this provides protection against L. monocytogenes challenge in vitro. Chapter 5 summarizes the findings in chapters 2-4 and serves as the overall conclusion of this research. Within this chapter, I briefly propose new research directions based on this project’s findings. These future projects will enhance our understanding of the complex interactions between the gut microbiota and host, which prelude future studies investigating health-promoting roles of dietary substrates. These studies will also provide a strong foundation for exploring applications related to food safety and human health. 4 SIGNIFICANCE Many diseases of Western populations are associated with reduced dietary fiber intake (Burkitt et al., 1974), and diets rich in fiber sources are protective against cardiovascular disease, type II diabetes, and colorectal cancer (Threapleton et al., 2013; Cho et al., 2013; World Cancer Research Fund and American Institute for Cancer Research, 2011). The U.S. recommended adequate intake for fiber is 14 g/1000 kcal, or about 28 g/d and 34 g/d for adult women and men, respectively (U.S. Department of Health and Human Service, U.S. Department of Agriculture, 2015). Most Americans do not have adequate fiber intake (Hoy and Goldman, 2014). Dietary fiber sources influence gastro-intestinal bacterial composition, and fiber- rich foods are believed to be protective against certain enteric pathogens in humans (Rabbani et al., 2009; De Filippo et al., 2010). Foodborne illnesses affect more than 48 million Americans annually (Centers for Disease Control and Prevention, 2018). L. monocytogenes is a deadly foodborne pathogen affecting 1,600 people and causing approximately 260 deaths annually in the U.S. It is the third leading cause of foodborne related deaths and has a high fatality rate (Centers for Disease Control and Prevention, 2016). The high fatality rate of listeriosis makes it an important public health concern (World Health Organization, 2018). Select dietary fiber sources protect animals in models of bacterial infection (Buddington et al., 2002; Ebersbach et al., 2010), while others promote infection (Petersen et al., 2009; Ebersbach et al., 2010). This work focused on select dietary fiber sources and their fermentation end products, the relationship of these to gastro-intestinal bacteria, and their impact on gastro-intestinal barrier function. This research is the first comprehensive investigation into the relationship of select dietary fiber sources to bacterial composition, SCFA 5 production, gastro-intestinal barrier integrity, and L. monocytogenes in vivo and in vitro infection models. These studies are important not only for evaluating this model of foodborne infection, but also other gastro-intestinal disease states sharing similar states of dysbiosis. This research provides new data on how dietary fiber sources affect bacterial composition and foodborne infection. Finally, this work preludes future studies investigating the gastro-intestinal microbiota provides a strong foundation for exploring the microbiota for food safety and human health applications. 6 LITERATURE REVIEW Dietary fiber and microbial fermentation metabolites Dietary fibers are plant and analogous carbohydrates resistant to small intestine digestion and absorption by humans, but are subjected to bacterial fermentation in the large intestine. They include polysaccharides, oligosaccharides, lignin, and other plant- associated substances (DeVries, 2001). In addition to their structural classifications, dietary fibers are often classified as insoluble or soluble, based on their solubility in water. Cellulose and hemicelluloses are for the most part insoluble and provide bulk to digesta and decrease digesta transit time. Soluble fibers, such as gums, pectins, oligosaccharides, and β-glucans increase transit time without providing bulk (Eastwood, 1992; Makki et al., 2018). Notably, soluble fiber sources have more complete fermentation than insoluble fibers, as a general rule (Slavin, 2013). The non-digestible dietary fiber sources used in this research were gum arabic, arabinoxylan, cellulose, guar gum, and XOS. Gum arabic is a highly soluble and fermentable fiber composed of arabinose, galactose, rhamnose, methyl glucuronic acid, and glucuronic acid. The structure consists of a β(1,3) galactose backbone with 1,6-linked branches of galactose and arabinose. Rhamnose, methyl glucuronic acid, and glucuronic acid are terminal units on side chains. It is sourced from the acacia tree as sap and typically used in processed foods at a rate of 1-3%. In food applications, it is used in beverages to enhance mouth feel, and in baked goods to improve texture and stability (Baray, 2009). Gum arabic has been evaluated for prebiotic potential in humans, and daily consumption is associated with increases of beneficial fecal bacteria such as bifidobacteria and lactobacilli (Calame et al., 7 2008). In vitro gum arabic fermentation leads to the enrichment of bifidobacteria and lactobacilli; and the production of acetate, propionate, and butyrate (Alarifi et al., 2018). Arabinoxylan is a fiber source with soluble and insoluble fractions commonly sourced from wheat and other grasses. It is classified as a hemicellulose and its structure consists of β(1,4) linked xyloses as the backbone with arabinose units linked at the second and third carbon (Kellow and Walker, 2018). It is a viscous fiber, and is often used in the baking industry to enhance water absorption in dough (Biliaderis et al., 1995). Research has linked arabinoxylan to improved glycemic response due to its high viscosity. In vitro arabinoxylan fermentation leads to the enrichment of bifidobacteria, lactobacilli, eubacteria, bacteroides, and clostridia; and the production of lactate, acetate, propionate, and butyrate (Hughes et al., 2007). Cellulose is an insoluble β(1,4) glucose polymer with a linear structure. It is the most abundant structural component of plant cell walls (Sinnott, 2007). Purified cellulose is generally not fermentable in humans and non-ruminant mammals (Cummings, 1984; Bourquin et al., 1993; Flint et al., 2012b). However, cellulolytic bacteria have been isolated from humans and have membership in the genera: Ruminococcus, Clostridium, Eubacterium, and Bacteroides (Flint et al., 2012b). In food applications, cellulose is used as an emulsifier, stabilizer, bulking agent, and thickener. Powered or microcrystalline cellulose increases digesta viscosity and has clinical application for enteral nutrition products to alleviate diarrhea (Takahashi et al., 2005; Takahashi, 2009). Guar gum is a soluble and highly fermentable fiber composed of the monomers galactose and mannose. The structure consists of a β(1,4) mannose backbone with branches of galactose linked through 1,6 linkages (Stewart and Slavin, 2006; Mudgil et 8 al., 2014). It is sourced from the shrub, Cyamopsis tetragonolobus, and is typically used in processed foods due to its ability to form highly viscous gels in cold water (Mudgil et al., 2014). Partially hydrolyzed guar gum has been used clinically in enteral nutrition products to promote normal laxation and reduce diarrhea (Slavin and Greenberg, 2003). In vitro guar gum fermentation produces acetate, propionate, and butyrate (Khan and Edwards, 2005). XOS is a soluble and fermentable oligosaccharide consisting of xylose monomers β(1,4) linked and having a degree of polymerization between 2-10. It is produced through hydrothermal and enzymatic processing of beach wood, corncobs, or other lignocellulosic material rich in xylan (Kumar, 2012). Food applications are generally related to its use as a prebiotic or as an aid to reduce available calories. It can be used as a mild laxative, but high doses are associated with diarrhea (Moure et al., 2006). XOS supplementation modifies human gastro-intestinal bacterial composition by reducing pro-inflammatory bacteria from the genera Dialister and Slackia (Yang et al., 2015), and enriching bifidobacteria (Fehlbaum et al., 2018). In vitro XOS fermentation produces acetate, propionate, and butyrate (Fehlbaum et al., 2018). The dietary fibers used in the present study are differentiated by their monomer carbohydrate composition, structure, and solubility. Utilization of fiber sources for energy by gastro-intestinal micro-organisms depends on their expression of carbohydrate degrading enzymes, also known as carbohydrate-active enzymes (CAZymes). Four classes of these enzymes are glycoside hydrolases, polysaccharide lyases, carbohydrate esterases, and a fourth that degrades lignin (Hamaker and Tuncil, 2014). The major products of bacterial fiber metabolism are SCFAs. 9 SCFAs are volatile organic acids with six or fewer saturated carbons. The major SCFAs produced from dietary fiber fermentation are acetate, propionate, and butyrate. These SCFAs account for 95% of the SCFAs produced in humans (den Besten et al., 2013). Molar ratios produced by in vitro fermentation vary by substrate source (Bourquin et al., 1993), but on average have a molar ratio of 60:20:20 in human digesta (den Besten et al., 2013). Biosynthesis pathways include: the Wood-Ljungdahl pathway, acrylate pathway, succinate pathway, propanediol pathway, and the classical pathway (condensation of two acetyl-CoA molecules) (Koh et al., 2016). Isobutyrate, valerate, isovalerate, hexanoate, and heptanoate are also found in human digesta in low concentrations (Zhao et al., 2006). Other minor products and intermediates include lactate, succinate, formate, butanediol, butanol, and acetone (Müller, 2001). Total large intestine luminal content of SCFAs ranges from 20-140 mM, and approximately 95% is absorbed by passive diffusion or active transport with bicarbonate exchange in gastro- intestinal epithelial cells in humans (den Besten et al., 2013). Total daily SCFA production is estimated to be 400-600 mmol/d (Verbeke et al., 2015). Acetate is a substrate for cholesterol synthesis (Hellman et al., 1953). Propionate is a utilized by the liver for glucose production and inhibits cholesterol synthesis from acetate (Wolever et al., 1991). Butyrate is a preferred energy source of colonocytes (Clausen and Mortensen, 1995). In addition to these functions, SCFAs are also signaling molecules. Butyrate and propionate are both histone deacetylase (HDAC) inhibitors (Koh et al., 2016). Butyrate in particular has anti-inflammatory and anti-cancer properties. For example, by inhibiting histone deacetylation, butyrate inhibits colorectal cancer cell growth and causes apoptosis (Koh et al., 2016). Acetate, propionate, and butyrate are also ligands 10 for the G-protein coupled receptors (GPCR): GPR41, GPR43, and GPR109A. Activation of these receptors is linked to intestinal homeostasis by affecting inflammation pathways and immune responses (Koh et al., 2016). Gastro-intestinal microbiota The human gastro-intestinal tract is home to more than 1014 micro-organisms from the domains Archaea, Bacteria, and Eukaryota. This collection of life that outnumbers human host cells by 10-fold is collectively known as the gastro-intestinal microbiota. The human microbiota is represented by more than 2100 bacterial species from 12 phyla (Thursby and Juge, 2017). Nearly 1000 bacterial species have been cultured from the gut (Rajilic-Stojanovic and de Vos, 2014). Many of these gastro- intestinal inhabitants are strict anaerobes and have an important symbiotic relationship with their host (Thursby and Juge, 2017). For example, the microbiota influences gastro-intestinal development and immune function by direct interaction with tissues and cells (Di Mauro et al., 2013). Microbial composition of the gastro-intestinal tract determines the metabolic activity and subsequent catabolism of dietary fibers sources (Flint et al., 2012a). Environmental factors, like diet, have significant impacts on bacterial composition and diversity more so than host genetics (De Filippo et al., 2010; David et al., 2014; De Filippo et al., 2017). Bacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria are the predominant human-associated bacterial phyla identified though 16S rRNA gene sequencing and account for more than 93% of isolates (Flint et al., 2012b; Thursby and Juge, 2017). Firmicutes is generally the predominant phylum in humans followed by Bacteroidetes (Martens et al., 2014). Another important phylum, Verrucomicrobia, 11 represents about 3% of human fecal bacteria (Flint et al., 2012b). Patterns of how dietary substrates affect human gastro-intestinal bacterial composition are inconsistent, and this is likely due to the uniqueness of individual microbiotas and methodological differences in 16S rRNA gene sequencing. However, consumption of low-fiber diets is consistently associated with reduced bacterial diversity and SCFA concentration in digesta (Sonnenburg and Sonnenburg, 2014; Makki et al., 2018). Low bacterial diversity is associated with gastro-intestinal dysbiosis (Nguyen et al., 2015; Zeng et al., 2017). Members of Bacteroidetes are Gram-negative organisms representing approximately 7000 species within the classes Bacteroidia, Cytophagia, Flavobactera, and Sphingobacteria. They are inhabitants of the gastro-intestinal tracts of humans, animals and nearly all-environmental habitats (Thomas et al., 2011). Members of Bacteroidetes are primary fermenters, express numerous CAZymes, and are known to degrade more than a dozen different complex carbohydrates (Martens et al., 2009; Martens et al., 2014; Flint et al., 2012b; Sun and O’Riordan, 2013). The relationship between host and Bacteroidetes is generally described as symbiotic, but some members are pathogens (Martens et al., 2009). The genus Bacteroides, in particular, degrades many different dietary fiber sources (Flint et al., 2012a). Predominate human gastro- intestinal families include Bacteroidaceae, Prevotellaceae, Porphyromonadaceae, and Rikenellaceae. Species from these families are important producers of succinate, acetate, and propionate (Rajilic-Stojanovic and de Vos, 2014). Members of Firmicutes are Gram-positive organisms and important classes include: Bacilli, Clostridia, Erysipelotrichi, and Negativicutes (Rajilic-Stojanovic and de Vos, 2014). Members of Firmicutes are secondary fermenters and metabolize a wide variety of complex carbohydrates and some are thought to be “keystone” gastro- 12 intestinal organisms due to their ability to use organic acids as substrates for the production of other SCFAs (Flint et al., 2012a; Sum and O’Riordan, 2013). Firmicutes is the dominant gastro-intestinal phylum in healthy humans and accounts for 50-80% of the microbiota (Rajilic-Stojanovic and de Vos, 2014). The families Lachnospiraceae and Ruminococcaceae are highly abundant inhabitants of the human large intestine, and members of Firmicutes are significant contributors to butyrate production (Flint et al., 2012a). Members of Proteobacteria are Gram-negative organisms and important classes include: Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, Deltaproteobacteria, and Epsilonproteobacteria (Garrity et al., 2005). It is the largest bacterial phylum and contains a number of well-known enteric human pathogens such as Escherichia coli, Salmonella, and Campylobacter from the families Enterobacteriaceae and Campylobacteraceae (Moon et al., 2018). Proteobacteria members are facultative or obligate anaerobes and are some of the first gastro-intestinal colonizers of humans early in life, but only represent about 1% of the microbiota (Rajilic-Stojanovic and de Vos, 2014). Members of Proteobacteria can metabolize carbohydrates, lipids, and proteins, which is thought to contribute to functional metabolic variation of the gastro-intestinal microbiota (Bradley and Pollard, 2017; Moon et al., 2018). Members of Actinobacteria are Gram-positive and commonly found in soil, but are suited for a wide range of environments, including the human gastro-intestinal tract, albeit at low abundance. Members are from six classes, which include: Actinobacteria, Acidimicrobiia, Coriobacteriia, Nitriliruptoria, Rubrobacteria, and Thermoleophilia. They are clinically important organisms due to their production of approximately two- 13 thirds of all naturally produced antibiotics. Most members of Actinobacteria are aerobes but there are exceptions (Barka et al., 2016). Members of genus Bifidobacterium are anaerobes and produces lactate and acetate (Rajilic-Stojanovic and de Vos, 2014). Bifidobacterium is an important early gastro-intestinal resident of infants due to its ability to metabolize human milk oligosaccharides (Flint et al., 2012b; Barka et al., 2016). Verrucomicrobia is commonly isolated from human and animals. Two species have been identified: Akkermansia muciniphila and Prosthecobacter fluviatilis; however, the latter has sparsely been detected while the former is the predominantly detected Verrucomicrobia species in human digesta (Rajilic-Stojanovic and de Vos, 2014). A. muciniphila is a Gram-negative and obligate anaerobe that utilizes gastro- intestinal mucus as a nutrient source (Derrien et al., 2004; Thursby and Juge, 2017). Verrucomicrobia represents about 3% of human gastro-intestinal bacteria (Flint et al., 2012b) and is associated with protection against metabolic disorders associated with inflammation (Cani and de Vos, 2017). Mouse models are commonly used in gastro-intestinal microbiota research because of their low cost, high reproduction rate, similar gastro-intestinal anatomy compared to humans, reproducibility of experiments, and abundant supply of digesta material (Nguyen et al., 2015). However, there are important physiological and anatomical differences between mice and humans. For example, gastro-intestinal transit time for mice is 6-7 h while in humans it is 14-76 h (Hugenholtz and de Vos, 2018). Mice also have a functional cecum, which is the primary site of bacterial fermentation, whereas in humans the colon is the primary site of fermentation (Hugenholtz and de Vos, 2018; Nguyen et al., 2015). The colon of adult mice is 14 cm in length and smooth. 14 The adult human colon is approximately 105 cm in length, and divided into three sections: ascending, transverse, and descending. The colon mucus layer in humans is also thicker and has a faster growth rate than in mice (Nguyen et al., 2015; Hugenholtz and de Vos, 2018). Nonetheless, despite these differences, the gastro-intestinal tract microbiota of mice and humans are both dominated by Bacteroidetes, Firmicutes, and Proteobacteria; share approximately 80 genera; and are 80-95% functionally similar (Hugenholtz and de Vos, 2018; Xiao et al., 2015). Furthermore, mouse and human gastro-intestinal bacteria share more than 300 carbohydrate active enzymes (Xiao et al., 2015). However, relative abundances of genera varies considerably between mice and humans. For example, humans have high abundances of Prevotella, Faecalibacterium, and Ruminococcus. Conversely, mice have high abundances of Lactobacillus, Alistipes, and Turicibacter (Nguyen et al., 2015). In summary, dietary fiber sources strongly influence the gastro-intestinal microbiota due to specialized expression of carbohydrate degrading enzymes. The microbiota is an important producer of SCFAs from the fermentation of dietary fiber sources. Mice are an important tool for modeling the effects of dietary substrates on gastro-intestinal bacteria composition. L. monocytogenes and bacterial infection models L. monocytogenes is a facultative intracellular Gram-positive bacillus from the phylum Firmicutes and is responsible for the deadly foodborne illness listeriosis. It is a common environmental organism found in high moisture soils and pastures. In healthy humans, infection is generally mild and self-limiting (Hoelzer et al., 2012). L. monocytogenes is also an asymptomatic gastro-intestinal resident in up to 10% of 15 humans (Buchanan et al., 2017). Serious infections affect pregnant women, children, the elderly, and the immune compromised resulting in sepsis, meningoencphalitis, or abortion in pregnant women (Pizarro-Cerda et al., 2012). The estimated LD5o for humans is 1.9 x 106 CFU and onset of illness can take several weeks; however, doses as low as 102-104 CFU can lead to serve illness in high-risk groups (Food and Agriculture Organization, the World Health Organization, 2004; Buchanan et al., 2017; Radoshevich and Cossart, 2018). Foods, especially ready-to-eat foods with extended refrigeration, are responsible for most infections (Buchanan et al., 2017). In the U.S., approximately 1,600 cases of invasive listeriosis occurs annually, leading to about 260 deaths (Centers for Disease Control and Prevention, 2019). Listeriosis is the third leading cause of foodborne related deaths (Centers for Disease Control and Prevention, 2016). The high fatality rate makes it a significant public health concern (World Health Organization, 2018). Animal models have been important for modeling dose and infection mechanisms of L. monocytogenes in humans. The gastro-intestinal tract is the primary site of invasion for L. monocytogenes. In the intestine, it invades M-cells of Peyer’s patches (Hoelzer et al., 2012); however, L. monocytogenes can invade almost all adherent cell types (Portnoy et al., 2002). One important mode of human infection is through interaction between the E-cadherin epithelial cell surface protein and the L. monocytogenes invasion protein internalin-A (InlA). L. monocytogenes appears to preferentially adhere to and invade gastro-intestinal villus extrusion sites with InlA without directly disrupting tight junctions (Pentecost et al., 2006). A second receptor protein involved in infection is internalin-B (InlB), which interacts with the hepatocyte growth factor receptor Met. InlB interacts with mouse and human Met receptors 16 (Pizarro-Cerda et al., 2012). L. monocytogenes also utilizes an InlA independent invasion mechanism using the pore-forming listeriolysin O (LLO) toxin that leads to its cellular entry in in vitro models (Pizarro-Cerda et al., 2012). In phagocytes, cellular entry occurs by phagocytosis (Radoshevich and Cossart, 2018). Recently, a new invasion mechanism was described and involves the interaction between Listeria adhesion protein and heat shock protein 60. This interaction leads to epithelial barrier disruption and L. monocytogenes translocation in mice (Drolia et al., 2018; Drolia and Bhunia, 2019). After cell entry, L. monocytogenes escapes vacuoles though the action of phospholipase A, phospholipase B, and LLO (Radoshevich and Cossart, 2018). It then produces the actin assembly-inducing protein, which enables it to use host actin to propel itself through the cytoplasm to neighboring cells where it begins a new infection cycle and ultimately spreads to the mesenteric lymph nodes, liver, and spleen (Pizarro- Cerda et al., 2012; Hoelzler et al., 2012). After reaching the circulatory system, the liver is the primary entrapment site. Infection is generally cleared by neutrophils, macrophages, and natural killer cells (Hoelzler et al., 2012). Oral transmission of L. monocytogenes is generally not as efficient in rodents compared to humans; however, oral infection of BALB/cBy/J mice with a dose of 107 CFU leads to invasion of large intestine cells and systemic migration (Bou Ghanem et al. 2012; Bou Ghanem et al., 2013). In this model, peak infection occurs 5 d after oral inoculation and L. monocytogenes is typically detectable for 8 d to 2 wk (Bou Ghanem et al., 2012; Bou Ghanem et al., 2014). The strain used in this research, L. monocytogenes EGD-e serotype 1/2a, was first isolated from rabbits (Murray et al., 1926). This strain, along with other lineage I 17 strains, account for approximately 63% of sporadic human cases of listeriosis, 93% of human epidemic cases, and 42% of animal cases (Jeffers et al., 2001). Gastro-intestinal microbial composition affects L. monocytogenes infection in mice. Germfree mice orally challenged with 100 CFU of L. monocytogenes develop diarrhea, lose weight, and have high cecum and colon counts. Specific pathogen free mice inoculated with as many as 5 x 105 CFU of L. monocytogenes do not show signs of listeriosis. The resident microbiota is hypothesized to interfere with pathogen proliferation through direct and indirect mechanisms (Zachar and Savage, 1979). Examples of direct mechanisms are the production of antimicrobial compounds, competition for nutrients, or steric hindrance of adhesion sites (Sun and O’Riordan, 2013). An indirect mechanism is the slowing of gastric motility. For example, peristalsis is slower in germfree mice compared to conventional mice; therefore, mice with a stable microbiota benefit from decreased digesta transit time and decreased pathogen exposure (Zachar and Savage, 1979; Abrams and Bishop, 1967). L. monocytogenes outgrowth may occur in the gastro-intestinal tract during microbiota dysbiosis, and the extent of imbalance and host susceptibility are factors influencing active infection or symptom-free carrier status (Zachar and Savage, 1979). L. monocytogenes infection also causes intestinal cells to secrete mucus. In vivo and in vitro models of infection result in goblet cell mucus release, which prevents the organism from infecting cells by steric hindrance (Liévin-Le Moal et al., 2005). Mucus release occurs through the action of the pore-forming toxin LLO (Coconnier et al., 1998). Mucus was also associated with decreased in vitro pathogen adhesion in a similar experiment with L. monocytogenes and Escherichia coli (Laparra and Sanz, 2009). In addition to their barrier function, mucus glycoproteins are an energy substrate for 18 gastro-intestinal bacteria much like dietary fiber sources. One organism thought to be beneficial and capable of metabolizing mucus is A. muciniphila (Reunanen et al., 2015), and its impact on gastro-intestinal homeostasis is an active area of research. Dietary fiber sources have been used as interventions in animal models of bacterial infection with varying results. In an experiment simulating L. monocytogenes systemic infection, 28% of B6C3F1 mice fed diets containing 10% cellulose died within 14 days. None of the mice fed diets containing 10% inulin died, and feeding diets containing 10% oligofructose resulted in fewer deaths than cellulose. Similarly, fewer mice fed inulin died after being challenged with Salmonella enterica serovar Typhimurium. Oligofructose and inulin were hypothesized to cause changes in the gastro-intestinal micro-organism population and its functional metabolic characteristics rather than directly inhibiting growth of the enteric pathogens (Buddington et al., 2002). In another L. monocytogenes infection study, guinea pigs were fed diets containing 10% XOS, GOS, inulin, apple pectin, or polydextrose prior to oral challenge with L. monocytogenes. Resistance to infection improved in guinea pigs fed XOS or GOS and decreased in animals fed inulin or apple pectin. The authors speculated that XOS and GOS stimulated the growth of resident bifidobacteria and lactobacilli, which can metabolize XOS and GOS to produce lactic and acetic acids (Ebersbach et al., 2010). However, other researcher have hypothesized that organic acids and other bacterial metabolites cause gastro-intestinal inflammation and mucosal barrier dysfunction (Ten Bruggencate et al., 2005). In a study investigating the effect of dietary fiber consumption on severity of S. enterica serovar Typhimurium infection in mice, FOS and XOS-fed mice had higher S. Typhimurium liver counts than cornstarch-fed mice. Additionally, the acute phase 19 response protein, haptoglobin, which is a biomarker of bacterial translocation, was approximately three times higher in serum of mice fed XOS and FOS compared to the control diet (Petersen et al. 2009). Recently, Desai et al. (2016) found dietary fiber deprivation in mice results in degradation of colon mucus and promotes increased epithelial access to the mouse pathogen Citrobacter rodentium. In summary, L. monocytogenes infection is rare in healthy humans, but is highly fatal for high-risk groups. Dietary fiber sources differentially affect models of bacterial infection. The gastro-intestinal microbiota is strongly influenced by dietary fiber sources, and is an important pathogenesis factor affecting infection responses. Gastro-intestinal barrier function The gastro-intestinal epithelium is the site of nutrient absorption and serves as a physical barrier protecting the host from foreign particles, such as micro-organisms. The human small intestine has three distinct segments - the duodenum, jejunum, and ileum - and is the primary site of nutrient absorption (Edelblum and Turner, 2015). The human large intestine also has three distinct segments - the ascending, transverse, and descending colon - and is an important site for water absorption, electrolyte exchange, vitamin production, and bacterial fermentation (Azzouz and Sharma, 2018). Gastro- intestinal cell types include progenitor stem cells, absorptive enterocytes, enteroendrocine cells, antimicrobial secreting Paneth cells, and mucus secreting goblet cells (Edelblum and Turner, 2015). These cells form the protective polarized monolayer. Monolayer integrity is a function of several transmembrane proteins forming connections between neighboring cells. The complexes linking neighboring cells are tight junctions, gap junctions, adherens junctions, and desmosomes (Ulluwishewa et al, 20 2011). Tight junctions consist of four types of transmembrane proteins: claudins, occludin, junctional adhesion molecule, and tricellulin. Intracellular portions of these proteins interact with scaffold proteins called zonula occludens (Suzuki, 2013). Claudin proteins are the structural components of tight junctions, while occludin proteins regulate intermembrane and paracellular diffusion (Ulluwishewa et al, 2011). Transcellular transport such as that occurring with nutrients, electrolytes, immune factors, and water occurs across apical and basolateral membranes via passive diffusion, facilitated diffusion, and vesicle trafficking. Paracellular transport occurs between neighboring epithelial cells, and tight junctions are considered the gatekeepers of exchange (Edelblum and Turner, 2015). SCFAs differentially affect gastro-intestinal barrier function. Suzuki et al. (2008) demonstrated in rat cecum in situ loops, T84 and Caco-2 cells that physiological concentrations of acetate and propionate improve barrier function, which was evaluated by transepithelial electrical resistance (TEER) and Lucifer yellow migration. Additionally, by decreasing pH of the incubation medium the beneficial effect observed with acetate further improved due to its enhanced cellular diffusion. On the other hand, butyrate did not affect barrier function. In a more recent study, butyrate and dietary fiber fermentation end products improved barrier function in HT29-MTX-E12 cells treated with deoxynivalenol. Paracellular permeability decreased and TEER values increased with exposure to 5-10 mM butyrate, whereas no effect was observed at concentrations of 0.1 mM or 50-100 mM (Nielsen et al., 2018). This study supports evidence from a Caco-2 model where 2 mM butyrate promoted barrier function, but 8 mM disrupted barrier function, which was evaluated by TEER and inulin permeability. The hypothesized mechanism of action 21 is redistribution of tight junction proteins, zonula occludens-1 and occludin, from the cytoplasm to the cell membrane due to activation of the AMP-activated protein kinase (AMPK) pathway (Peng et al., 2007; Peng et al., 2009). Sodium butyrate also increases expression of the tight-junction protein claudin-1 in cdx2-IEC cells, and decreases monolayer permeability at concentrations of 2 and 4 mM (Wang et al., 2012). In another Caco-2 model, the composition of acetate, propionate, and butyrate produced by in vitro fermentation varied by carbohydrate source and differentially affected TEER values during the 48-h exposure period. Barrier function was positively correlated with total SCFAs, acetate, and butyrate concentration. Pure SCFA treatments with less than 100 mM total SCFA improved barrier function; however, 150 and 200 mM concentrations were cytotoxic and caused barrier disruption of the Caco-2 monolayer. Finally, 40 mM SCFA solutions containing 20 or 50% butyrate were protective against barrier disruption and repaired lipopolysaccharide and tumor necrosis factor-α induced damage (Chen et al., 2017). These studies demonstrate the barrier enhancing effects of pure SCFAs and fermentation end products at normal physiological concentrations. Extreme low and high concentrations of SCFAs either have no effect or resulted in diminished barrier function. In general, butyrate has consistently had the strongest influence on improving barrier integrity. Importantly, butyrate affects the assembly of the tight-junction proteins in a way that maintains barrier integrity (Peng et al., 2007; Peng et al., 2009; Wang et al., 2012). However, new evidence shows propionate, when delivered in drinking water, also improves barrier function in a mouse model of colitis by affecting tight junction protein expression (Tong et al., 2016). 22 In addition to cell surface proteins contributing to barrier function, mucus secretions are also an important barrier defense factor, especially in the large intestine (Johansson et al., 2015). Mucus is a family of high molecular weight glycoproteins produced by goblet cells (Cornick et al., 2015). The mucus layer consists of an inner cell adherent layer and an outer secretory layer. In humans there are 21 mucins with the nomenclature MUC1-MUC21. Membrane bound mucins include: MUC1, MUC3A, MUC3B, MUC4, MUC12, MUC13, MUC15, MUC16, and MUC17. Secreted mucins include MUC2, MUC5AC, MUC5B, MUC6, and MUC7 (Perez-Vilar and Hill, 1999; McGuckin et al., 2011; Cornick et al., 2015). Functions of the other mucins have yet to be described. Animals have analogous mucus glycoproteins (Perez-Vilar and Hill, 1999), and mucins secreted by humans and mice have similar gene structure and gastro- intestinal expression (Escande et al., 2004). Mucin secretions can be either basal, which results in regular release of single mucin granules, or they can be large exocytosis events when cells are exposed to bacteria, bacterial metabolites, toxins, or cytokines (Cornick et al., 2015). Dietary fiber sources affect mucus production in vivo and in vitro. In rats, soluble and insoluble fiber sources increase mucus production in the small intestine without affecting goblet cell proliferation (Hino et al., 2012). In rats fed non-digestible fiber sources, the large intestine mucus layer is thicker compared to rats fed a fiber-free diet. Additionally, mucus thickness is positively correlated with total SCFAs and acetate, but negatively correlated with the proportion of butyrate found in digesta (Hedemann et al., 2009). Dietary fibers also interact with small intestine epithelial cells to stimulate mucus secretion with higher levels of mucin glycoproteins with sialic acid than sulfate. Mucins having lower levels of sulfate conjugation are thought to be more easily 23 degraded by intestinal bacteria (Hino et al., 2013; Hino et al., 2012; Deplancke and Gaskins, 2001). A hypothesized indirect mechanism for mucus stimulation is through SCFAs ability to increase mucin gene expression by enhanced histone acetylation and methylation, and through the activator protein 1 signaling cascade. Burger-van Paassen et al. (2009) demonstrated that low levels of butyrate (1 mM) stimulated mucin gene expression, but higher levels (5-15 mM) returned expression to baseline levels in LS174T goblet cells. Additionally, 1-5 mM propionate increased mucin gene expression, but propionate levels greater than 15 mM reduced expression. Acetate also increased mucin expression when cells were exposed to 5-15 mM concentrations. SCFAs but not fiber sources influence mucin secretion in rats when delivered to in situ colon loops. For example, 5 mM butyrate increased mucin release in perfused rat colons, whereas 100 mM reduced the mucin response. On the other hand, 5 mM acetate had no effect on mucin expression, whereas 100 mM acetate increased it (Barcelo et al., 2000). Gastro-intestinal mucus is an important barrier defense factor affecting L. monocytogenes pathogenicity. L. monocytogenes adheres to mucins with internalin proteins. Internalin-mucin binding has potentially important consequences related to infection. One mucin glycoprotein, MUC2, has been hypothesized to have opposing effects and may inhibit L. monocytogenes from interacting with epithelial cells but it has also been hypothesized to promote retention of L. monocytogenes near epithelial cells thereby enabling adhesion to epithelial cells (Lindén et al., 2008). In summary, dietary fiber sources and SCFAs differentially affect gastro- intestinal barrier function by affecting tight junctions and mucus secretions. Differences in responses may have a beneficial or inhibitory effect on L. monocytogenes association with gastro-intestinal epithelial cells. 24 REFERENCES 25 REFERENCES Abrams GD, Bishop J. Effect of the Normal Microbial Flora on Gastrointestinal Motility. Proc Soc Exp Biol Med. 1967;301–4. Alarifi S, Bell A, Walton G. In vitro fermentation of gum acacia – impact on the faecal microbiota. Int J Food Sci Nutr. Informa UK Ltd.; 2018;69:696–704. Anderson JW, Baird P, Davis RH, Ferreri S, Knudtson M, Koraym A, Waters V, Williams CL. Health benefits of dietary fiber. Nutr Rev. 2009;67:188–205. Azzouz L, Sharma S. Physiology, Large Intestine [Internet]. Treasure Island, FL: StatsPearls Publishing; 2018 [cited 2019 May 1]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK507857 Baray S. Acacia Gum. In: Cho S, Samuel P, editors. Fiber Ingredients: Food Applications and Health Benefits. Boca Raton: CRC Press; 2009. p. 122. Barcelo A, Claustre J, Moro F, Chayvialle JA, Cuber JC, Plaisancié P. Mucin secretion is modulated by luminal factors in the isolated vascularly perfused rat colon. Gut. 2000;46:218–24. Barka EA, Vatsa P, Sanchez L, Gaveau-vaillant N, Jacquard C, Klenk H, Clément C, Ouhdouch Y, van Wezel P. Taxonomy, Physiology, and Natural Products of Actinobacteria. Microbiol Mol Biol Rev. 2016;80:1–44. Biliaderis CG, Izydorczyk MS, Rattan O. Effect of arabinoxylans on bread-making quality of wheat flours. Food Chem. 1995;53:165–71. Bou Ghanem EN, Jones GS, Myers-Morales T, Patil PD, Hidayatullah AN, D’Orazio SEF. InlA Promotes Dissemination of Listeria monocytogenes to the Mesenteric Lymph Nodes during Food Borne Infection of Mice. PLoS Pathog. 2012;8. Bou Ghanem EN, Myers-Morales T, D’Orazio SE. A mouse model of food borne Listeria monocytogenes infection. Curr Protoc Microbiol. 2014;31:1–21. Bou Ghanem E, Myers-Morales T, Jones G, D’Orazio S. Oral transmission of listeria monocytogenes in mice via ingestion of contaminated food. J Vis Exp. 2013;75:1–8. Bourquin LD, Titgemeyer EC, Fahey GC. Vegetable Fiber Fermentation by Human Fecal Short-Chain Fatty Acid Production during In Vitro Bacteria: Cell Wall Polysaccharide Disappearance and Fermentation and Water-Holding Capacity of Unfermented Residues. J Nutr. 1993;123:860–9. 26 Bradley PH, Pollard KS. Proteobacteria explain significant functional variability in the human gut microbiome. Microbiome. 2017;5:1–23. Buchanan RL, Gorris LGM, Hayman MM, Jackson TC, Whiting RC. A review of Listeria monocytogenes: An update on outbreaks, virulence, dose-response, ecology, and risk assessments. Food Control. Elsevier Ltd; 2017;75:1–13. Buddington KK, Donahoo JB, Buddington RK. Dietary Oligofructose and Inulin Protect Mice from Enteric and Systemic Pathogens and Tumor Inducers. J Nutr. 2002;132:472– 7. Burger-van Paassen N, Vincent A, Puiman PJ, van der Sluis M, Bouma J, Boehm G, van Goudoever JB, van Seuningen I, Renes IB. The regulation of intestinal mucin MUC2 expression by short-chain fatty acids: implications for epithelial protection. Biochem J. 2009;420:211–9. Burkitt DP, Walker ARP, Painter NS. Dietary Fiber and Disease. JAMA. 1974;229:1068– 74. Calame W, Weseler AR, Viebke C, Flynn C, Siemensma AD. Gum arabic establishes prebiotic functionality in healthy human volunteers in a dose-dependent manner. Br J Nutr. 2008;100:1269–75. Cani PD, Possemiers S, Van de Wiele T, Guiot Y, Everard A, Rottier O, Geurts L, Naslain D, Neyrinck A, Lambert DM, Muccioli GG, Delzenne NM. Changes in gut microbiota control inflammation in obese mice through a mechanism involving GLP-2-driven improvement of gut permeability. Gut. 2009;58:1091–103. Cani PD, de Vos WM. Next-Generation Beneficial Microbes: The Case of Akkermansia muciniphila. Frontiers Microbiol. 2017;8:1–8. Centers for Disease Control and Prevention. Foodborne Germs and Illnesses [Internet]. Food Safety. 2018 [cited 2019 May 13]. Available from: https://www.cdc.gov/foodsafety/foodborne-germs.html Centers for Disease Control and Prevention. Listeria (Listeriosis) [Internet]. 2019 [cited 2019 Apr 17]. Available from: https://www.cdc.gov/listeria/index.html Centers for Disease Control and Prevention. Listeria (Listeriosis) [Internet]. People at Risk. 2016 [cited 2019 Apr 30]. Available from: https://www.cdc.gov/listeria/risk.html Chen T, Kim CY, Kaur A, Lamothe L, Shaikh M, Keshavarzian A, Hamaker BR. Dietary fibre-based SCFA mixtures promote both protection and repair of intestinal epithelial barrier function in a Caco-2 cell model. Food Funct. Royal Society of Chemistry; 2017;8:1166–73. 27 Cho SS, Qi L, Fahey GC, Klurfeld DM. Consumption of cereal fiber, mixtures of whole grains and bran, and whole grains and risk reduction in type 2 diabetes, obesity, and cardiovascular disease. Am J Clin Nutr. 2013;98:594–619. Clausen MR, Mortensen PB. Kinetic studies on colonocyte metabolism of short chain fatty acids and glucose in ulcerative colitis. Gut. 1995;37:684–9. Coconnier NE, Dlissi E, Robard M, Laboisse CL. Listeria monocytogenes Stimulates Mucus Exocytosis in Cultured Human Polarized Mucosecreting Intestinal Cells through Action of Listeriolysin O. Infect Immun. 1998;66:3673–81. Cornick S, Tawiah A, Chadee K. Roles and regulation of the mucus barrier in the gut. Tissue Barriers. 2015;3:e982426-1-e982426-15. Cummings JH. Cellulose and the human gut. Gut. 1984;25:805–10. David LA, Maurice CF, Carmody RN, Gootenberg DB, Button JE, Wolfe BE, Ling AV, Devlin AS, Varma Y, Fischbach MA, Biddinger SB, Dutton RJ, Turnbaugh PJ. Diet rapidly and reproducibly alters the human gut microbiome. Nature. Nature Publishing Group; 2014;505:559–63. De Filippo C, Cavalieri D, Di Paola M, Ramazzotti M, Poullet JB, Massart S, Collini S, Pieraccini G, Lionetti P. Impact of diet in shaping gut microbiota revealed by a comparative study in children from Europe and rural Africa. Proc Natl Acad Sci. 2010;107:14691–6. De Filippo C, Paola M Di, Ramazzotti M, Albanese D, Pieraccini G, Banci E, Miglietta F, Cavalieri D, Collado MC. Diet, Environments, and Gut Microbiota. A Preliminary Investigation in Children Living in Rural and Urban Burkina Faso and Italy. Front Microbiol. 2017;8:1–14. den Besten G, Eunen K Van, Groen AK, Venema K, Reijngoud D, Bakker BM. The role of short-chain fatty acids in the interplay between diet, gut microbiota, and host energy metabolism. J Lipid Res. 2013;54:2325–40. Deplancke B, Gaskins HR. Microbial modulation of innate defense: Goblet cells and the intestinal mucus layer. Am J Clin Nutr. 2001;73. Derrien M, Vaughan EE, Plugge CM, de Vos WM. Akkermansia muciniphila gen. nov., sp. nov., a human intestinal mucin-degrading bacterium. Int J Syst Evol Microbiol. 2004;54:1469–76. Desai MS, Seekatz AM, Koropatkin NM, Stappenbeck TS, Martens EC. Article A Dietary Fiber-Deprived Gut Microbiota Degrades the Colonic Mucus Barrier and Enhances Pathogen Article A Dietary Fiber-Deprived Gut Microbiota Degrades the Colonic Mucus Barrier and Enhances Pathogen Susceptibility. Cell. 2016;167:1339–53. 28 DeVries JW. The definition of dietary fibre. Cereal Foods World. 2001;46:112–29. Di Mauro A, Neu J, Riezzo G, Raimondi F, Martinelli D, Francavilla R, Indrio F. Gastrointestinal function development and microbiota. Ital J Pediatr. 2013;39:1–7. Drolia R, Bhunia AK. Crossing the Intestinal Barrier via Listeria Adhesion Protein and Internalin A. Trends Microbiol. Elsevier Ltd; 2019;27:408–25. Drolia R, Tenguria S, Durkes AC, Turner JR, Bhunia AK, Drolia R, Tenguria S, Durkes AC, Turner JR, Bhunia AK. Listeria Adhesion Protein Induces Intestinal Epithelial Barrier Dysfunction for Bacterial Article Listeria Adhesion Protein Induces Intestinal Epithelial Barrier Dysfunction for Bacterial Translocation. Cell Host Microbe. Elsevier Inc.; 2018;23:470–484.e7. Eastwood MA. The Physiological Effect of Dietary Fiber: An Update. Annu Rev Nutr. 1992;12:19–35. Ebersbach T, Jørgensen JB, Heegaard PM, Lahtinen SJ, Ouwehand AC, Poulsen M, Frøkiaer H, Licht TR. Certain dietary carbohydrates promote Listeria infection in a guinea pig model, while others prevent it. Int J Food Microbiol. 2010; 140:218–24. Edelblum KL, Turner JR. Epithelial Cells: Structure, Transport, and Barrier Function. In: Mestecky J, Strober W, Russel MW, Kelsall BL, Cheroutre H, Lambrecht BN, editors. Mucosal Immunology. 4th ed. Kidlington and Waltham: Elsevier; 2015. p. 187–210. Escande F, Porchet N, Berigaud A, Petitprez D, Aubert J-P, Buisine M-P. The mouse secreted gel-forming mucin gene cluster. Biochim Biophys Acta. 2004;1676:240–50. Fehlbaum S, Prudence K, Kieboom J, Heerikhuisen M, Broek T Van Den, Schuren FHJ, Steinert RE, Raederstorff D. In Vitro Fermentation of Selected Prebiotics and Their Effects on the Composition and Activity of the Adult Gut Microbiota. Int J Mol Sci. 2018;19:1–16. Flint HJ, Karen P, Louis P, Duncan SH. The role of the gut microbiota in nutrition and health. Nat Rev Gastroenterol Hepatol. Nature Publishing Group; 2012a;9:577–89. Flint HJ, Scott KP, Duncan SH, Louis P, Forano E, Flint HJ, Scott KP, Duncan SH, Louis P, Forano E, et al. Microbial degradation of complex carbohydrates in the gut. Gut Microbes. 2012b;3:289–306. Food and Agriculture Organization, The World Health Organization. Risk assessment of Listeria monocytogenes in ready-to-eat foods. Geneva; 2004. Garrity G, Bell J, Lilburn T. Volume Two The Proteobacteria Part C The Alpha-, Beta-, Delta-, and Epsilonproteobacteria. In: Brenner D, Krieg N, Stanley J, Garrity G, editors. Bergey’s Manual of Systematic Bacteriology. Second. New York: Springer; 2005. p. 1– 1194. 29 Hamaker B, Tuncil Y. A Perspective on the Complexity of Dietary Fiber Structures and Their Potential Effect on the Gut Microbiota. J Mol Biol. 2014;426:3838–50. Hedemann MS, Theil PK, Bach Knudsen KE. The thickness of the intestinal mucous layer in the colon of rats fed various sources of non-digestible carbohydrates is positively correlated with the pool of SCFA but negatively correlated with the proportion of butyric acid in digesta. Br J Nutr. 2009;102:117–25. Hellman L, Rosenfeld RS, Gallagher TF. Cholesterol Synthesis from C14-Acetate in Man. J Clin Invest. 1953;33:142–9. Hino S, Sonoyama K, Bito H, Kawagishi H, Aoe S, Morita T. Low-Methoxyl Pectin Stimulates Small Intestinal Mucin Secretion Irrespective of Goblet Cell Proliferation and Is Characterized by Jejunum Muc2 Upregulation in Rats. J Nutr. 2013;143:34–40. Hino S, Takemura N, Sonoyama K, Morita A, Kawagishi H, Aoe S, Morita T. Small Intestinal Goblet Cell Proliferation Induced by Ingestion of Soluble and Insoluble Dietary Fiber Is Characterized by An Increase in Sialylated Mucins in Rats. J Nutr. 2012;142:1429–36. Hoelzer K, Pouillot R, Dennis S. Animal models of listeriosis: a comparative review of the current state of the art and lessons learned. Vet Res. 2012;43:1–27. Hoy MK, Goldman JD. Dietary Fiber Intake of the U.S. Population. 2014. Hugenholtz F, de Vos WM. Mouse models for human intestinal microbiota research: a critical evaluation. Cell Mol Life Sci. Springer International Publishing; 2018;75:149–60. Hughes SA, Shewry PR, Gibson GR, Sanz ML, Rastall RA. In Vitro Fermentation by Human Fecal Microflora of Wheat Arabinoxylans. J Agric Food Chem. 2007;55:4589– 95. Jeffers GT, Bruce JL, McDonough PL, Scarlett J, Boor KJ, Wiedmann M. Comparative genetic characterization of Listeria monocytogenes isolates from human and animal listeriosis cases. Microbiology. 2001;147:1095–104. Johansson MEV, Jakobsson HE, Holmen-Larsson J, Svensson F, Backhed F, Gunnar C. Hansson. Normalization of Host Intestinal Mucus Layers Article Normalization of Host Intestinal Mucus Layers Requires Long-Term Microbial Colonization. Cell Host Microbe. 2015;18:582–92. Kellow NJ, Walker KZ. Authorised EU health claim for arabinoxylan. Foods, Nutrients and Food Ingredients with Authorised EU Health Claims, Volume 3. Elsevier Ltd; 2018. 201-218 p. Khan KM, Edwards CA. In vitro fermentation characteristics of a mixture of Raftilose and guar gum by human faecal bacteria. Eur J Nutr. 2005;44:371–76. 30 Koh A, De Vadder F, Kovatcheva-Datchary P, Bäckhed F. From dietary fiber to host physiology: Short-chain fatty acids as key bacterial metabolites. Cell. 2016;165:1332–45. Kumar G. A Review on Xylooligosaccharides. Int Res J Pharm. 2012;3:71–4. Laparra JM, Sanz Y. Comparison of in vitro models to study bacterial adhesion to the intestinal epithelium. Lett Appl Microbiol. 2009;49:695–701. Liévin-Le Moal V, Servin AL, Coconnier-Polter MH. The increase in mucin exocytosis and the upregulation of MUC genes encoding for membrane-bound mucins induced by the thiol-activated exotoxin listeriolysin O is a host cell defence response that inhibits the cell-entry of Listeria monocytogenes. Cell Microbiol. 2005;7:1035–48. Lindén SK, Bierne H, Sabet C, Png CW, Florin TH, McGuckin MA, Cossart P. Listeria monocytogenes internalins bind to the human intestinal mucin MUC2. Arch Microbiol. 2008;190:101–4. Makki K, Deehan EC, Walter J, Backhed F. The Impact of Dietary Fiber on Gut Microbiota in Host Health and Disease. Cell Host Microbe. 2018;23:705–15. Martens EC, Kelly AG, Tauzin AS, Brumer H. The Devil Lies in the Details: How Variations in Polysaccharide Fine-Structure Impact the Physiology and Evolution of Gut Microbes. J Mol Biol. Elsevier Ltd; 2014;426:3851–65. Martens EC, Koropatkin NM, Smith TJ, Gordon JI. Complex Glycan Catabolism by the Human Gut Microbiota: The Bacteroidetes Sus-like Paradigm. J Biol Chem. 2009;284:24673–7. McGuckin MA, Lindén SK, Sutton P, Florin TH. Mucin dynamics and enteric pathogens. Nat Publ Gr. Nature Publishing Group; 2011;9:265–78. McNeil NI. The contribution of the large intestine to energy supplies in man. Am J Clin Nutr. 1984;39:338–42. Moon CD, Cookson AL, Young W, Maclean PH, Bermingham EN. Metagenomic insights into the roles of Proteobacteria in the gastrointestinal microbiomes of healthy dogs and cats. 2018;1–20. Moure A, Gullón P, Domínguez H, Parajó JC. Advances in the manufacture, purification and applications of xylo-oligosaccharides as food additives and nutraceuticals. Process Biochem. 2006;41:1913–23. Mudgil D, Barak S, Khatkar BS. Guar gum: Processing, properties and food applications - A Review. J Food Sci Technol. 2014;51:409–18. Müller V. Bacterial Fermentation. Encycl Life Sci. 2001;1–7. 31 Murray EGD, Webb RA, Swann MBR. A Disease of Rabbits Characterised by a Large Mononuclear Leucocytosis, Caused by a Hitherto Undescribed Bacillus Bacterium monocytogenes(n.sp.). J Pathol Bacteriol. 1926;29:407–39. Nguyen TLA, Vieira-Silva S, Liston A, Raes J. How informative is the mouse for human gut microbiota research? Dis Model Mech. 2015;8:1–16. Nielsen DSG, Jensen BB, Theil PK, Nielsen TS, Knudsen KEB, Purup S. Effect of butyrate and fermentation products on epithelial integrity in a mucus-secreting human colon cell line. J Funct Foods. Elsevier; 2018;40:9–17. Peng L, Li Z-R, Green RS, Holzman IR, Lin J. Butyrate Enhances the Intestinal Barrier by Facilitating Tight Junction Assembly via Activation of AMP-Activated Protein Kinase in Caco-2 Cell Monolayers. J Nutr. 2009;139:1619–25. Peng L, He Z, Chen W, Holzman IR, Lin J. Effects of butyrate on intestinal barrier function in a caco-2 cell monolayer model of intestinal barrier. Pediatr Res. 2007;61:37– 41. Pentecost M, Otto G, Theriot JA, Amieva MR. Listeria monocytogenes invades the epithelial junctions at sites of cell extrusion. PLoS Pathog. 2006;2:0029–40. Perez-Vilar J, Hill RL. The Structure and Assembly of Secreted Mucins. J Biol Chem. 1999;274:31751–5. Petersen A, Heegaard PMH, Pedersen AL, Andersen JB, Sørensen RB, Frøkiær H, Lahtinen SJ, Ouwehand AC, Poulsen M, Licht TR. Some putative prebiotics increase the severity of Salmonella enterica serovar Typhimurium infection in mice. BMC Microbiol. 2009;9:1–11. Pizarro-Cerda J, Kuhbacher A. Entry of Listeria monocytogenes in Mammalian Epithelium Cells- An Updated View. Cold Spring Harb Perspect Med. 2012;1–18. Portnoy DA, Auerbuch V, Glomski IJ. The cell biology of Listeria monocytogenes infection: the intersection of bacterial pathogenesis and cell-mediated immunity. Journal Cell Biol. 2002;158:409–14. Rabbani GH, Ahmed S, Hossain I, Islam R, Marni F, Akhtar M, Majid N. Green Banana Reduces Clinical Severity of Childhood Shigellosis. Pediatr Infect Dis J. 2009;28:420–5. Radoshevich L, Cossart P. Listeria monocytogenes: towards a complete picture of its physiology and pathogenesis. Nature. Nature Publishing Group. 2018;16:32–46. Rajilic-Stojanovic M, de Vos WM. The first 1000 cultured species of the human gastrointestinal microbiota. FEMS Microbiol Rev. 2014;38:996–1047. 32 Reunanen J, Kainulainen V, Huuskonen L, Ottman N, Belzer C, Huhtinen H, de Vos WM. Akkermansia muciniphila Adheres to Enterocytes and Strengthens the Integrity of the Epithelial Cell Layer. Appl Environ Microbiol. 2015;81:3655–62. Sinnott M. Primary Structure and Conformation of Oligosaccharides and Polysaccharides. Carbohydrate Chemistry and Biochemistry Structure and Mechanism. Cambridge: The Royal Society of Chemistry; 2007. p. 194. Slavin J. Fiber and Prebiotics: Mechanisms and Health Benefits. Nutrients. 2013;5:1417–35. Slavin JL, Greenberg NA. Partially hydrolyzed guar gum. Nutrition. 2003;19:549–52. Sonnenburg ED, Sonnenburg JL. Perspective Starving our Microbial Self: The Deleterious Consequences of a Diet Deficient in Microbiota-Accessible Carbohydrates. Cell Metab. Elsevier Inc.; 2014;20:779–86. Stewart ML, Slavin JL. Molecular weight of guar gum affects short-chain fatty acid profile in model intestinal fermentation. Mol Nutr Food Res. 2006;50:971–6. Sun Y, O’Riordan M. Regulation of Bacterial Pathogenesis by Intestinal Short-Chain Fatty Acids. 1st ed. Advances in Applied Microbiology. Elsevier Inc.; 2013. 93-118 p. Suzuki T. Regulation of intestinal epithelial permeability by tight junctions. Cell Mol Life Sci. 2013;70:631–59. Suzuki T, Yoshida S, Hara H. Physiological concentrations of short-chain fatty acids immediately suppress colonic epithelial permeability. Br J Nutr. 2008;100:297–305. Takahashi T. Cellulose. In: Sungsoo S, Samuel P, editors. Fiber Ingredients: Food Applications and Health Benefits. Boca Raton: CRC Press; 2009. p. 263–82. Takahashi T, Karita S, Ogawa N, Goto M. Crystalline Cellulose Reduces Plasma Glucose Concentrations and Stimulates Water Absorption by Increasing the Digesta Viscosity in Rats. J Nutr. 2005;135:2405–10. Ten Bruggencate SJM, Bovee-Oudenhoven IMJ, Lettink-Wissink MLG, Van der Meer R. Dietary fructooligosaccharides increase intestinal permeability in rats. J Nutr. 2005;135:837–42. Thomas F, Hehemann J-H, Rebuffet E, Cczjzek M, Michel G. Environmental and gut Bacteroidetes: the food connection. Front Microbiol. 2011;2:1–16. Threapleton DE, Greenwood DC, Nykjaer C, Woodhead C, Cade JE. Dietary fibre intake and risk of cardiovascular disease: systematic review and meta-analysis. BMJ. 2013;347:1–12. 33 Thursby E, Juge N. Introduction to the human gut microbiota. Biochem J. 2017;474:1823–36. Tong L, Wang Y, Wang Z, Liu W, Sun S, Li L, Su D, Zhang L. Propionate Ameliorates Dextran Sodium Sulfate-Induced Colitis by Improving Intestinal Barrier Function and Reducing Inflammation and Oxidative Stress. Front Pharmacol. 2016;7:1–9. U.S. Department of Health and Human Services, U.S. Department of Agriculture. 2015 – 2020 Dietary Guidelines for Americans. 8th ed. 2015. Ulluwishewa D, Anderson RC, McNabb WC, Moughan PJ, Wells JM, Roy NC. Regulation of Tight Junction Permeability by Intestinal Bacteria and Dietary Components. J Nutr. 2011;141:769–76. Verbeke KA, Boobis AR, Chiodini A, Edwards CA, Franck A, Kleerebezem M, Nauta A, Raes J, Van Tol EAF, Tuohy KM. Towards microbial fermentation metabolites as markers for health benefits of prebiotics. Nutr Res Rev. 2015;28:42–66. Walker AW, Ince J, Duncan SH, Webster LM, Holtrop G, Ze X, Brown D, Stares MD, Scott P, Bergerat A, et al. Dominant and diet-responsive groups of bacteria within the human colonic microbiota. ISME J. Nature Publishing Group. 2011;5:220–30. Wang H-B, Wang P-Y, Wang X, Wan Y-L, Lui Y-C. Butyrate Enhances Intestinal Epithelial Barrier Function via Up-Regulation of Tight Junction Protein Claudin-1 Transcription. Dig Dis Sci. 2012;57:3126–35. Wolever T, Spadafora P, Eshuis H. Interaction between colonic acetate and propionate in humans. Am J Clin Nutr. 1991;53:681–7. World Cancer Research Fund, American Institute for Cancer Research. Continuous Update Project Report. Food, Nutrition, Physical Activity, and the Prevention of Colorectal Cancer. 2011. World Health Organization. Listeriosis [Internet]. Fact Sheet. 2018 [cited 2019 Apr 26]. Available from: https://www.who.int/mediacentre/factsheets/listeriosis/en/ Xiao L, Feng Q, Liang S, Sonne SB, Xia Z, Qiu X, Li X, Long H, Zhang J, Zhang D, et al. A catalog of the mouse gut metagenome. Nat Biotechnol. 2015;33:1103–8. Yang J, Summanen PH, Henning SM, Hsu M, Lam H, Huang J, Tseng CH, Dowd SE, Finegold SM, Heber D, Li Z. Xylooligosaccharide supplementation alters gut bacteria in both healthy and prediabetic adults: A pilot study. Front Physiol. 2015;6:1–11. Zachar Z, Savage DC. Microbial interference and colonization of the murine gastrointestinal tract by Listeria monocytogenes. Infect Immun. 1979;23:168–74. 34 Zeng M, Inohara N, Nunez G. Mechanisms of inflammation-driven bacterial dysbiosis in the gut. Nature. 2017;10:18–26. Zhao G, Nyman M, Jönsson JÅ. Rapid determination of short-chain fatty acids in colonic contents and faeces of humans and rats by acidified water-extraction and direct- injection gas chromatography. Biomed Chromatogr. 2006;20:674–82. 35 CHAPTER 2: EFFECT OF DIETARY FIBER SOURCES ON MICROBIAL POPULATIONS AND LISTERIA MONOCYTOGENES IN MICE ABSTRACT Background: Dietary fiber consumption influences the host microbiota, gastro- intestinal function, and health. This research investigated the effects of feeding five dietary fibers sources to mice on gastro-intestinal bacterial composition, Listeria monocytogenes infection in mice, and the relationship between L. monocytogenes and bacterial composition. Objective: The objectives of this study were as follows: 1) determine the gastro- intestinal bacterial composition of mice fed select dietary fiber sources with and without L. monocytogenes challenge; 2) enumerate L. monocytogenes in challenged mice; and 3) relate L. monocytogenes counts to gastro-intestinal bacterial composition. Methods: For 3 wk, BALB/CBy/J mice were fed modified AIN-93G diets containing 10% of the following dietary fiber sources: gum arabic, arabinoxylan, cellulose, guar gum, or xylo-oligosaccharide (XOS). During the final week, mice were orally challenged with phosphate buffered saline (PBS) or L. monocytogenes serotype 1/2a. Seven days post challenge, mice were sacrificed and L. monocytogenes recovered from mouse cecum digesta, spleen, and livers were serially diluted and plated on modified Oxford (MOX) agar for enumeration. Mouse digesta bacterial composition was determined by 16S rRNA gene sequencing. Data were analyzed using standard parametric and nonparametric statistical tests. 36 Results: Mouse weight did not differ by fiber source or L. monocytogenes infection status at study conclusion (P ≥ 0.05). XOS-fed mice had a higher infection rate than mice fed other fiber sources (P = 0.025), and the highest cecum digesta and spleen counts (P < 0.05) of L. monocytogenes. Bacterial diversity was lowest in the cecum and colon digesta of XOS-fed mice. Gum arabic and XOS-fed mice digesta were enriched with Verrucomicrobia (Akkermansia) (P < 0.05). Verrucomicrobia (Akkermansia) tended (P < 0.1) to be higher in the cecum digesta of L. monocytogenes-positive mice and was higher in colon digesta (P < 0.05). Verrucomicrobia (Akkermansia) was positively associated with cecum L. monocytogenes counts (P = 0.005). Oscillibacter and Ruminiclostridium were negatively associated with cecum L. monocytogenes counts (P < 0.05). Conclusion: Dietary fiber sources differentially affect bacterial composition, diversity measures, and L. monocytogenes in mice. XOS supplementation, relative to gum arabic, arabinoxylan, cellulose, and guar gum, shifts gastro-intestinal bacterial composition and decreases diversity in a way that enhances the survival and invasion of L. monocytogenes in BALB/CBy/J mice. Keywords: dietary fiber, Listeria monocytogenes, microbiota, 16S rRNA, xylo- oligosaccharide 37 INTRODUCTION Foodborne infections affect more than 48 million Americans annually (Centers for Disease Control and Prevention, 2018). L. monocytogenes is a common environmental organism that rarely causes serious illness in healthy humans (World Health Organization, 2018), but certain populations like the elderly, pregnant women, children, and those with immune deficiencies are vulnerable to invasive infection. In the U.S., approximately 1,600 cases of invasive listeriosis occurs annually, leading to about 260 deaths (Centers for Disease Control and Prevention, 2019). This high fatality rate makes it a significant public health concern (World Health Organization, 2018). Dietary fiber sources, when fed to animals, have mixed effects on severity of bacterial infection and host survival; and certain dietary fiber sources are protective whereas others enhance infection (Buddington et al., 2002; Ten Bruggencate et al., 2005; Petersen et al. 2009; Ebersbach et al., 2010; Hryckowian et al., 2018). Consuming foods rich in non- digestible carbohydrates improves clinical symptoms of children with enteric bacterial infection (Rabbani et al., 2009). Recent advances in high-throughput genetic sequencing revealed humans consuming fiber-rich diets have distinct gastro-intestinal bacterial compositions thought to be protective against certain enteric infections compared to those with lower fiber intake (De Filippo et al., 2010). Therefore, the question of how certain dietary fiber sources affect gastro-intestinal bacterial composition and enteric foodborne pathogens is still unresolved. Dietary fibers are plant and analogous carbohydrates resistant to digestion by humans and mammals and are important substrates for microbial fermentation in the gastro-intestinal tract. They include polysaccharides, oligosaccharides, lignin, and other plant-associated substances (DeVries, 2001; Williams et al., 2017). Dietary substrates 38 can enrich certain bacterial taxa due to differences in their expression of bacterial enzymes, such as polysaccharidases, glycosidases, proteases, and peptidases (Flint et al., 2012; Hamaker and Tuncil, 2014; Williams et al., 2017). The non-digestible dietary fiber sources used in this research were gum arabic, arabinoxylan, cellulose, guar gum, and XOS; and were selected due to differences in their carbohydrate composition and structures. Gum arabic is a highly soluble and fermentable non-starch polysaccharide, sourced from the Acacia tree. It is composed of arabinose, galactose, rhamnose, methyl glucuronic acid, and glucuronic acid. Its structure consists of a β(1,3) galactose backbone with branches of galactose and arabinose (Baray, 2009). Arabinoxylan is a hemicellulose with soluble and insoluble fractions and is commonly sourced from wheat and other grasses. It has a β(1,4) linked xylose backbone with arabinose side units (Kellow and Walker, 2018). Cellulose is an insoluble β(1,4) glucose polymer and is the most abundant structural component of plant cell walls (Sinnott, 2007). Purified cellulose has limited fermentation in humans and non-ruminant mammals (Cummings, 1984; Bourquin et al., 1993; Flint et al., 2012). Guar gum is sourced from the shrub, Cyamopsis tetragonolobus, and is a soluble and fermentable non-starch polysaccharide with a β(1,4) mannose backbone and branches of galactose (Stewart and Slavin, 2006; Mudgil et al., 2014). XOS is a highly soluble and fermentable oligosaccharide with β(1,4) linked xylose monomers having a degree of polymerization between 2-10 (Kumar, 2012). How dietary fiber sources alter the gastro-intestinal microbiota and affect L. monocytogenes in an oral challenge model of infection has important implications for human health. In the present study, we hypothesized that feeding diets containing different dietary fiber sources would lead to differences in gastro-intestinal bacterial 39 composition and measures of diversity. We also hypothesized that fiber sources would differentially affect L. monocytogenes in the BALB/CBy/J mouse model of human gastro-intestinal infection. The objectives of this study were: 1) determine the gastro- intestinal bacterial composition of mice fed select dietary fiber sources with and without L. monocytogenes challenge; 2) enumerate L. monocytogenes in challenged mice; and 3) relate L. monocytogenes counts to gastro-intestinal bacterial composition. 40 MATERIALS AND METHODS Mouse model and experimental design An equal number of 6-wk old female and male (n = 150) BALB/CBy/J mice (Jackson Laboratory, Bar Harbor, ME) were housed (1-2 mice per cage) in cages in a room with constant humidity (30-70%), temperature (20-24 °C), and a 12 h day:night cycle. Mice were randomly assigned to one of five modified American Institute of Nutrition-93G (AIN-93G) (Reeves et al., 1993) diets containing 10% of the following dietary fiber sources: gum arabic, arabinoxylan, cellulose, guar gum, or XOS. Experimental diet compositions and the original AIN-93G formula are referenced in Table 2.1. Mice were initially fed the 10% cellulose experimental diet for 2 wk (Figure 2.1), which served to normalize the gastro-intestinal bacterial composition across all treatments. Cellulose also served as the control diet in this experiment because of its low degree of fermentation in mice (Cummings, 1984; Kimura et al., 2004; Flint et al., 2012). After the 2-wk acclimation period, mice were fed their randomly assigned experimental diet for 3 wk. During the final week of the experiment, mice were challenged with phosphate buffered saline (PBS) or L. monocytogenes. The treatment design was a 5 x 2 x 2 factorial with five diets, two sexes, and two challenge treatments. Mice had constant access to reverse osmosis water and were fed ad libitum. Mice were weighed weekly from the time of receipt to the day of sacrifice. This study was approved by the Michigan State University All-University Committee on Animal Use and Care. 41 Table 2.1 AIN-93G and modified AIN-93G diet compositions Ingredient (g/kg) diet Cornstarch Casein Maltodextrin Sucrose Soybean oil Cellulose Test fiber Mineral mix Vitamin mix L-cystine Choline6 t-BHQ7 Energy8 AIN- 93G 397.49 200.00 132.00 100.00 70.00 50.00 35.00 10.00 3.00 2.50 0.014 3760 Gum arabic1 329.84 200.00 132.00 100.00 70.00 Arabino- xylan2 314.67 200.00 132.00 100.00 70.00 Cellulose3 Guar gum4 329.84 200.00 132.00 100.00 70.00 347.49 200.00 132.00 100.00 70.00 117.65 35.00 10.00 3.00 2.50 0.014 3726 132.82 35.00 10.00 3.00 2.50 0.014 3694 100.00 35.00 10.00 3.00 2.50 0.014 3580 117.65 35.00 10.00 3.00 2.50 0.014 3746 XOS5 347.49 200.00 96.87 100.00 70.00 135.14 35.00 10.00 3.00 2.50 0.014 3729 1Gum arabic, 85% purity, 2% protein 2Arabinoxylan, 75.29% purity, 6.12% protein 3Cellulose, powdered 4Guar gum, 85% purity, 1% fat, 4% protein 5XOS, 74.22% purity, 25.72% maltodextrin, 0.06% ash 6Choline = Choline bitartrate 7t-BHQ = t-butylhydroquinone 8Energy value of diet (kcal/kg); energy value of individual ingredients (kcal/g): casein (3.58), sucrose (4.0), cornstarch (3.6), maltodextrin (3.8), soybean oil (9.0), mineral mix (0.88), vitamin mix (3.87), L-cystine (4.0), cellulose (0.0), gum arabic (2.0), arabinoxylan (2.0), guar gum (2.0), XOS (1.5) (Reeves et al., 1993; Dyets Inc., 1993; Phillips, 1998; Roberfroid and Slavin, 2000) 42 Figure 2.1 Experimental timeline1 1Mouse age (wk) depicted in timeline Phosphate buffered saline and L. monocytogenes challenge protocol After 2 wk of feeding the experimental diets, half of the mice (n = 75) were orally challenged with autoclaved PBS (n = 75) while the other half (n = 75) were orally challenged with 1.2 x 107 CFU of L. monocytogenes serovar 1/2a (strain ATCC BAA-679 / EGD-e) (American Type Culture Collection, Rockville, MD) grown in brain heart infusion (BHI) broth using a modified protocol developed by Bou Ghanem et al. (2013). Prior to L. monocytogenes or PBS challenge, all mice were tested for presence/absence of L. monocytogenes by plating diluted fresh fecal samples (1 fecal pellet/100 μL PBS) on MOX agar plates (NEOGEN, Lansing, MI). The L. monocytogenes inoculum was propagated from an isolated colony grown on BHI agar. The isolated colony was incubated at 30 °C in 25 mL of BHI broth for 24 h. 43 Then, 5 mL of the growth culture was added to 250 mL of BHI broth in a 500 mL Pyrex media storage bottle. After 8 h of growth, the culture was thoroughly mixed and 1 mL aliquots were frozen at -80 °C. L. monocytogenes concentration was determined by warming frozen aliquots to room temperature, serially diluting with PBS, then plating on BHI agar. The inoculum contained 5.4 x 107 CFU/mL. Frozen 1 mL inoculum aliquots were thawed at room temperature on the day of challenge and centrifuged to pellet the bacterial mass. Medium was aspirated and replaced with 27 μL of autoclaved PBS to achieve a concentration of 2.0 x 106 CFU/uL. Prior to challenge, mice were fasted for 12 h. Individual mice were placed in 2 L plastic beakers and given a piece of puffed-rice cereal with 5 μL PBS or 5 μL of the 2.0 x 106 CFU/μL L. monocytogenes inoculum, with the latter being the established dose required for intestinal infection. The actual inoculum dose determined by MOX plating was 1.2 x 107 CFU. The LD50 for this mouse model is 5 x 109 CFU (Bou Ghanem et al., 2013). After mice consumed the puffed-rice cereal, they were returned to cages and fed their respective test diet for 7 d. During the dark cycle of day seven, mice were euthanized by measured CO2 inhalation and cervical dislocation. Immediately following euthanization, cecums including contents were weighed. Then cecum contents, spleens and livers were stored on ice prior to performing serial dilutions in PBS for L. monocytogenes enumeration. Small intestine, cecum, and colon digesta used for isolating genomic deoxyribonucleic acid (DNA) for 16S rRNA gene sequencing were stored at -80 °C. Throughout the experiment mice were observed for distress associated with L. monocytogenes infection. In the event mice experienced stress, the following protocol was implemented: mice were isolated and monitored for weight loss, body conditioning, 44 and feed intake. Weight loss of 20% or greater and/or a body condition score of BC1 (Ullman-Culleré and Foltz, 1999) would result in mice being humanly euthanized by measured CO2 inhalation and cervical dislocation. No mice died due to the challenge protocol; however, three mice died due to factors unrelated to the experiment. Following the euthanization protocol, liver, spleen, and cecum digesta were diluted 1:10 (w/v) in autoclaved PBS. Livers were a homogenized with an Ultra-Turrax tissue homogenizer for 30 sec at 60% power (IKA, Wilmington, NC). Spleens were homogenized by rolling a 500 mL Pyrex storage bottle 20x over sterile Whirl pack bags containing isolated spleens. Following homogenization, samples were serially diluted with PBS and plated in duplicate for L. monocytogenes enumeration on MOX agar. Plates were incubated at 37 °C for 48 h. Plates with counts between 25-250 CFU were considered countable. Plates with fewer than 25 CFU were counted as estimated standard plate counts. DNA preparation for 16S rRNA gene sequencing The DNA Stool Mini Kit (QIAGEN, Germantown, MD) was utilized for purification of DNA from small intestine, cecum, and colon digesta. The protocol for pathogen detection with lysis temperature of 95 °C was selected to limit bias against Gram-positive organisms. Following DNA extraction, purified DNA was stored at -20 °C. Samples for 16S rRNA gene sequencing were submitted to the Research Technology Support Facility Genomics Core at Michigan State University. In total, 441 samples were submitted for further processing and analysis, which included 147 samples each from small intestine, cecum, and colon contents. 45 16S rRNA gene sequence processing Amplicon libraries of the V4 hyper-variable region of the 16S rRNA gene were created using the dual indexed Illumina library primers 515F/806R as described by Kozich et al. (2013). Individual polymerase chain reaction (PCR) products were normalized using SequalPrep DNA normalization plates (Invitrogen, Waltham, MA), and all DNA recovered from these plates were pooled. The pooled libraries were then purified with AmpureXP magnetic beads. Pooled libraries were quality checked and quantified using a combination of Qubit dsDNA HS, Agilent 2100 Bioanalyzer High Sensitivity DNA (Agilent, Santa Clara, CA) and Kapa Biosystems Illumina Library Quantification qPCR (Roche, Indianapolis, IN) assays. The pool was loaded onto a standard MiSeq v2 flow cell (Illumina, San Diego, CA) and sequenced in a 2 x 250 base pair paired-end format with a 500 cycle v2 reagent cartridge. Custom sequencing and index primers were added to appropriate slots of the reagent cartridge as described in Kozich et al. (2013). Base calling was performed by Illumina Real Time Analysis (RTA) version 1.18.64 and output of RTA was de-multiplexed and converted to FastQ file format with Illumina Bcl2fastq version 1.8.4 16S rRNA gene sequence analysis Phylotype operational taxonomic unit (OTU) analysis of 16S rRNA gene sequences was performed as described by Kozich et al. (2013) with slight modification. The sequence and taxonomy alignment references used were Silva version 132 (Quast et al., 2013; Kozich et al., 2013). Ambiguous bases and sequences longer than 260 base pairs were removed. Chimeric sequences were identified and removed using VSEARCH. Samples were subsampled to 4983 sequences. The workflow used in this experiment is 46 located in APPENDIX A. Mock bacterial community analysis was not performed in this experiment. Statistical analysis Mouse weights were compared using repeated measures analysis of variance (ANOVA), and Least Square (LS) means were compared using Fisher’s Least Significant Difference (LSD) method. Mouse cecum and contents as percent body weight are were compared using the Kruskal-Wallis with Dunn’s procedure or Wilcoxon Signed-Rank tests. The likelihood ratio chi-square test was performed to examine the relationship between diet and L. monocytogenes positive:negative status in challenged mice, and LS means of positive:negative ratios were compared using Fisher’s LSD method. L. monocytogenes plate counts were analyzed using a generalized estimating equation method with negative binomial distribution with log link function, and LS means were compared by Fisher’s LSD method. Bacterial relative abundance data and Shannon diversity were compared using the Kruskal-Wallis and Wilcoxon Signed-Rank tests. Predominant genera were selected by screening those with mean relative abundances greater than 1% averaged across all samples. Only genera with P-values less than 0.0001, after controlling for false discovery using the Benjamini and Hochberg method, were included for analysis, and only genera with median relative abundance greater than 1% across treatments are reported (R, R Core Team, Vienna, Austria). Pairwise comparisons were made using Dunn’s procedure. β-diversity data were analyzed using Analysis of Molecular Variance (AMOVA) with pairwise comparison made using the Bonferroni correction procedure in mothur version 1.41.1 (Kozich et al., 2013). Data were analyzed with individual mice as the experimental unit in order to 47 analyze 16S rRNA β-diversity data. Located within APPENDIX B are results with mouse cage as the experimental unit. Unless otherwise noted, all statistical analyses were performed using SAS, version 9.4 (SAS Institute Inc., Cary, NC) and GraphPad Software (La Jolla, CA). In all tests, significance was determined when (P < 0.05). Results of parametric tests are presented as LS means ± SEM and nonparametric tests are presented as medians and interquartile ranges (IQR). 48 RESULTS Infection classification and exclusion criteria Challenged mice were declared L. monocytogenes positive if any one of their cecum digesta, spleen, or liver MOX agar plates resulted in the growth of characteristic L. monocytogenes colonies. The first 25 mice were excluded from analysis for liver L. monocytogenes counts due to missing data and difference in liver serial dilution plating protocol for the first 25 mice. Mice with positive L. monocytogenes cecum or spleen samples were included for analysis of those respective samples. Mouse weights To examine the effect of fiber sources on mouse growth (Figure 2.2), mice were weighed weekly from the day of receipt to the day of sacrifice. A significant diet x age interaction was observed for mouse weight (P < 0.001); this was largely driven by reduced weights of mice consuming gum arabic-containing diets. Diet was not statistically significant (P = 0.73); however, age was statistically significant (P < 0.001). Mice weighed (19.2 ± 0.2 g) during the first week of the experiment and (24.2 ± 0.3 g) at the time of sacrifice. At age 9 wk, which corresponded to the first weighing period after experimental diet initiation, mice fed guar gum (23.2 ± 0.5 g) weighed more than those fed gum arabic (21.4 ± 0.7 g). Mouse weight was significantly influenced by sex (P < 0.001), with male mice weighing more than females during each week of the experiment (Figure 2.3). A significant sex x age interaction also was observed for mouse weight (P < 0.001). 49 Mouse weights by infection status were not different at the time of sacrifice (P = 0.17). Figure 2.2 Mouse weights by diet and age1,2 1Data are LS means ± SEM, n = 22-30 per dietary treatment 2*Mice fed gum arabic and guar gum were different at age 9 wk (P < 0.05) 50 Figure 2.3 Mouse weights by sex and age1,2 1Data are LS means ± SEM, n = 63 per sex 2Males weighed more than females for all age periods (P < 0.05) Cecum and cecum contents as percent of body weight Cecum and cecum contents as a percent of body weight (wet cecum proportion) represents a proxy for comparing substrate fermentation in mice (Zhou et al., 2009). Wet cecum proportion (Figure 2.4) was significantly influenced by diet (P < 0.001). Wet cecum proportion was lowest in mice fed cellulose (0.89%, 0.19%) compared to all other diets. Mice fed XOS had a higher wet cecum proportion (2.78%, 0.40%) compared to mice fed guar gum (1.90%, 0.36%) and cellulose (0.89%, 0.19%). Mice fed gum arabic (2.79%, 0.49%) had a higher wet cecum proportion compared to mice fed guar gum (1.90%, 0.36%) and cellulose (0.89%, 0.19%). Wet cecum proportion was significantly influenced by sex (P = 0.028) (Figure 2.5), and female wet cecum proportion (2.37%, 51 0.94%) was greater than in males (2.03%, 0.90%). Wet cecum proportion was not influenced by L. monocytogenes infection status (P = 0.70). Figure 2.4 Mouse wet cecum proportion by diet1,2 1Data are presented as box and whisker plots (5-95%), n = 22-30 per dietary treatment 2Diets without a common letter are different (P < 0.05) 52 Figure 2.5 Mouse wet cecum proportion by sex1,2 1Data are presented as box and whisker plots (5-95%), n = 63 per sex 2Sexes without a common letter are different (P < 0.05) Infection status of L. monocytogenes challenged mice L. monocytogenes was isolated from cecum digesta, spleen, and livers of challenged mice (Figure 2.6). L. monocytogenes was generally isolated well from cecum digesta, and there was moderate migration to mouse spleens; however, due to the low number of positive liver samples, statistical tests could not be performed on liver counts. To examine the effect of diet on L. monocytogenes positive:negative status, a maximum likelihood ratio chi-square test was performed. The relationship between diet and positive:negative status was significant (P = 0.025). Positive L. monocytogenes status and diet were not independent, and more XOS-fed mice tested positive (93%) compared to mice fed diets containing the other fiber sources (average of 51%). 53 Figure 2.6 Positive L. monocytogenes samples in challenged mice by diet1,2,3 1Total positive is not necessarily cumulative of cecum, spleen and liver samples 2Data presented as a frequency plot, n = 11-16 per dietary treatment 3*Denotes positive:negative ratio difference between mice fed XOS and the other diets (P < 0.05) L. monocytogenes cecum digesta counts (Figure 2.7) were significantly influenced with an interaction between diet and sex (P < 0.001). Cecum digesta counts were also significantly influenced by diet (P < 0.001); however, counts by sex were not different (P = 0.41). Counts in male mice fed XOS (Ln 15.0 ± 0.6 CFU/mg) and female mice fed cellulose (Ln 13.5 ± 0.8 CFU/mg) were greater than all other diet and sex combinations. Male mice fed cellulose (Ln 6.8 ± 0.6 CFU/mg) had lower cecum digesta L. monocytogenes counts compared to all diet and sex combinations except male mice fed arabinoxylan (Ln 6.9 ± 0.6 CFU/mg) and female mice fed guar gum (Ln 8.3 ± 0.6 CFU/mg). 54 Averaged across both sexes (Figure 2.8), mice fed XOS had the highest cecum digesta L. monocytogenes counts (Ln 12.6 ± 0.4 CFU/mg) compared to all other diets. Mice fed arabinoxylan (Ln 7.8 ± 0.5 CFU/mg) had the lowest cecum digesta L. monocytogenes counts compared to all other diets except guar gum (Ln 9.1 ± 0.4 CFU/mg). Figure 2.7 Cecum L. monocytogenes counts of challenged mice by diet and sex1,2,3 1Data presented as (Ln LS mean ± SEM CFU/mg Cecum Digesta) ESPC 2LS means without common letter are different (P < 0.05), n = 5-8 per treatment 3Dotted line represents limit of detection (Ln 200 CFU/mg) 55 Figure 2.8 Cecum digesta L. monocytogenes counts of challenged mice by diet1,2,3 1Data presented as (Ln LS mean ± SEM CFU/mg Cecum Digesta) ESPC 2LS means without common letter are different (P < 0.05), n = 11-15 per dietary treatment 3Dotted line represents limit of detection (Ln 200 CFU/mg) L. monocytogenes spleen counts (Figure 2.9) were significantly influenced with an interaction between diet and sex (P < 0.001). Spleen counts were also significantly influenced by diet (P = 0.001); but were not significantly different by sex (P = 0.42). Male mice fed XOS (Ln 10.4 ± 0.5 CFU/mg) and female mice fed guar gum (Ln 9.6 ± 0.5 CFU/mg) had the highest spleen L. monocytogenes counts compared to all other diet and sex combinations. Averaged across both sexes (Figure 2.10), mice fed XOS (Ln 8.6 ± 0.4 CFU/mg) and guar gum (Ln 8.1 ± 0.4 CFU/mg) had the highest spleen L. monocytogenes counts compared to mice consuming all other diets. 56 Figure 2.9 Spleen L. monocytogenes counts of challenged mice by diet and sex1,2,3 1Data presented as (Ln LS mean ± SEM CFU/mg spleen) ESPC 2LS means without common letter are different (P < 0.05), n = 5-8 per treatment 3Dotted line represents limit of detection (Ln 200 CFU/mg) 57 Figure 2.10 Spleen L. monocytogenes counts of challenged mice by diet1,2,3 1Data presented as (Ln LS mean ± SEM CFU/mg spleen) ESPC 2LS means without common letter are different (P < 0.05), n = 11-16 per dietary treatment 3Dotted line represents limit of detection (Ln 200 CFU/mg) Bacterial diversity of mice To examine how dietary fiber sources affect bacterial composition in this in vivo model, bacterial 16S rRNA genes were sequenced using the MiSeq Illumina platform. A total of 16,611,865 unpaired raw clusters were obtained from sequencing with 71.9% ≥ Q30 and 85.4% passing filtration. After sequence curation, small intestine digesta samples contained 1,855,186 sequences belonging to 556 unique phylotype OTUs, cecum digesta samples contained 2,545,756 sequences belonging to 178 unique phylotype OTUs, and colon digesta samples contained 2,510,296 sequences belonging to 191 unique phylotype OTUs from 12 phyla. After subsampling to 4983 sequences, OTU representative coverage across all digesta sources was greater than 99.8%. 58 To examine the relationship between dietary fiber source and α-diversity within small intestine, cecum, and colon digesta, Shannon diversity indices were compared (Table 2.2). Shannon diversity was influenced by diet within two of three digesta sources, which were the cecum (P < 0.001) and colon (P < 0.001). Mice fed cellulose (2.35) had the greatest median cecum digesta α-diversity compared to all other diets. Mice fed XOS (1.22) had the lowest cecum digesta α-diversity compared to all other diets. Likewise, mice fed cellulose (2.17) had the greatest colon digesta α-diversity compared to all other diets, and mice fed XOS (1.24) had the lowest colon α-diversity diversity compared to all diets. Small intestine Shannon diversity was not influenced by infection status (P = 0.982). Cecum Shannon diversity by infection status had a trending (P = 0.08) difference with α-diversity of negative mice (1.83, 0.54) greater than positive mice (1.67, 0.76); however, cecum Shannon diversity was not correlated with cecum L. monocytogenes counts (r = 0.132, P = 0.291). Shannon diversity was lower in the colon digesta of positive mice (1.60, 0.63) compared to negative mice (1.82, 0.44) (P = 0.026). Shannon diversity was not influenced by sex in small intestine, cecum, or colon digesta (P ≥ 0.05). Table 2.2 Shannon diversity index by diet and digesta sources1,2,3 Mouse Diet Gum arabic Arabinoxylan Cellulose Guar gum XOS Small Intestine 1.35 (0.34) 1.24 (0.63) 1.13 (0.58) 1.26 (0.77) 1.36 (0.18) Cecum 1.62c (0.19) 1.84bc (0.31) 2.35a (0.16) 1.92b (0.23) 1.22d (0.21) Colon 1.74b (0.39) 1.80b (0.19) 2.17a (0.16) 1.86b (0.41) 1.24c (0.23) 1Data presented as median (IQR) 2Diets without a common letter within a column are different (P < 0.05) 3Small Intestine (n = 19-30), Cecum (n = 22-30), Colon (n = 22-30) per dietary treatment 59 Since Shannon diversity is a non-linear index, the effective number of species (ENS) was determined to compare diversity directly. ENS is calculated using the formula: ENS = eH, where H = Shannon diversity. This converts Shannon diversity into a richness metric (Table 2.3) where all species are in equal proportion while maintaining the diversity of the original samples (Jost et al., 2010). The ENS of cecum digesta was influenced by diet (P < 0.001). Mice fed cellulose had the greatest median ENS (10.53) compared to all other diets, which means those digesta samples were 1.5 to 3.1 times more diverse than cecum digesta samples of mice fed the other dietary fiber sources. XOS-fed mice had the lowest diversity (3.39) compared to all other diets, which was 1.5 to 3.1 times less diverse than mice fed the other dietary fiber sources. Likewise, the ENS of colon digesta was influenced by diet (P < 0.001). Again, cellulose-fed mice had the greatest ENS (8.72) compared to all other diets, which translated to colon digesta samples 1.4 to 2.5 times more diverse than those from mice fed the other dietary fiber sources. XOS-fed mice had the lowest ENS (3.45) compared to all other diets. This means the bacterial communities in XOS-fed mice were 1.6 to 2.5 times less diverse mice fed the other dietary fiber sources. 60 Table 2.3 Effective number of species by diet and digesta sources1,2,3 Mouse Diet Gum arabic Arabinoxylan Cellulose Guar gum XOS Small Intestine 3.85 (1.40) 3.45 (2.55) 3.09 (1.77) 3.54 (3.04) 3.91 (0.72) Cecum 5.07c (0.99) 6.29bc (1.93) 10.53a (1.67) 6.83b (1.56) 3.39d (0.72) Colon 5.69b (2.17) 6.08b (1.17) 8.72a (1.36) 6.41b (2.41) 3.45c (0.80) 1Data presented as median (IQR) 2Diets without a common letter within a column are different (P < 0.05) 3Small Intestine (n = 19-30), Cecum (n = 22-30), Colon (n = 22-30) per dietary treatment To examine how dietary fiber sources affect bacterial community β-diversity, Yue and Clayton Θ (Θ-YC) distances were calculated for small intestine, cecum, and colon digesta bacterial communities by diet. The non-metric multidimensional scaling (NMDS) plot of Θ-YC distances by dietary fiber source in small intestine digesta (Figure 2.11), cecum (Figure 2.12), and colon (Figure 2.13) indicated bacterial communities cluster separately by diet. Statistical tests (AMOVA) revealed the centroids of bacterial communities were significantly influenced by diet in small intestine (P < 0.001), cecum (P < 0.001) and colon (P < 0.001) digesta. In small intestine digesta, the centroids of mice fed arabinoxylan differed from those fed gum arabic and XOS. The centroids of mice fed gum arabic differed from those fed cellulose, guar gum, and XOS. The centroids of mice fed XOS also differed from the centroids of cellulose and guar gum fed mice. Centroids of bacterial communities in small intestine digesta were not influenced by sex (P = 0.507) or infection status (P = 0.628). 61 In cecum digesta, all centroids of bacterial communities by diets were different from each other except arabinoxylan and guar gum. Centroids of cecum digesta bacterial communities by sex (P = 0.13) and infection status (P = 0.475) were not different. In colon digesta, all centroids of bacterial communities by diet were different from each other except arabinoxylan and guar gum. Centroids of colon digesta bacterial communities differed by sex (P = 0.014) but not infection status (P = 0.257). Figure 2.11 NMDS plot of Θ-YC distances of small intestine digesta by diet1,2 1Data presented as a 3D plot of bacterial communities by diet, n = 19-30 per dietary treatment 2X axis = Axis 1, Y axis = Axis 2, Z axis = Axis 3; axes are for ordination only 62 Figure 2.12 NMDS plot of Θ-YC distances of cecum digesta samples by diet1,2 1Data presented as a 3D plot of bacterial communities by diet, n = 22-30 per dietary treatment 2X axis = Axis 1, Y axis = Axis 2, Z axis = Axis 3; axes are for ordination only 63 Figure 2.13 NMDS plot of Θ-YC distances of colon digesta by diet1,2 1Data presented as a 3D plot of bacterial communities by diet, n = 22-30 per dietary treatment 2X axis = Axis 1, Y axis = Axis 2, Z axis = Axis 3; axes are for ordination only Phyla relative abundances Bacteroidetes, Verrucomicrobia, Firmicutes, and Actinobacteria represented 99.13 ± 0.27% of small intestine digesta (Table 2.4) sequences, 99.09 ± 0.19% of cecum digesta (Table 2.5) sequences, and 99.20 ± 0.15% of colon digesta (Table 2.6) sequences. Other phyla detected included Deinococcus-Thermus, Chloroflexi, Proteobacteria, Tenericutes, Acidobacteria, Gemmatimonadetes, Patescibacteria, and Omnitrophicaeota. 64 In small intestine digesta (Table 2.4), Bacteroidetes (P < 0.001) was greatest in mice fed arabinoxylan (64.61%), cellulose (68.93%), and guar gum (60.76%) compared to mice fed gum arabic (0.70%) and XOS (31.83%). Verrucomicrobia (P < 0.001) was greatest in mice fed XOS (39.87%) and gum arabic (59.20%). Firmicutes (P < 0.001) was greatest in mice fed gum arabic (32.61%). Actinobacteria (P < 0.001) was greatest in mice fed XOS (5.46%) and guar gum (2.95%). Abundances of these phyla were not influenced by sex (P ≥ 0.05) or infection status (P ≥ 0.05). Table 2.4 Relative abundances (%) of small intestine digesta phyla by diet1,2 Phylum Cellulose Arabino- Bacteroidetes Verrucomicrobia Firmicutes Actinobacteria Gum arabic 0.70b (12.46) 59.20a (18.60) 32.61a (12.62) 1.18b (1.40) xylan 64.61a (22.10) 15.97b (9.16) 13.90b (10.02) 1.85ab (5.20) 68.93a (29.82) 6.42b (15.71) 16.72b (14.75) 0.98b (2.21) Guar gum 60.76a (17.92) 20.07b (10.05) 13.45b (14.30) 2.95a (4.84) XOS 31.83b (26.59) 39.87a (16.60) 17.42b (17.38) 5.46a (4.53) 1Data presented as median (IQR), n = 19-30 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) In cecum digesta (Table 2.5), Bacteroidetes (P < 0.001) was greatest in mice fed arabinoxylan (53.26%), cellulose (44.91%), guar gum (45.60%), and XOS (48.69%). Verrucomicrobia (P < 0.001) was greatest in mice fed gum arabic (40.92%) and XOS (34.82%). Firmicutes (P < 0.001) was greatest in mice fed gum arabic (54.76%) and cellulose (48.16%), while XOS-fed mice had the lowest (12.67%) abundance. Actinobacteria (P < 0.001) was greatest in mice fed XOS (0.98%) and guar gum (0.46%). Abundances of these phyla were not influenced by sex (P ≥ 0.05). There was a trend (P = 65 0.096) for Verrucomicrobia abundance to be different between L. monocytogenes positive (21.24%, 30.48%) and negative (10.83%, 25.97%) mice. No other phyla abundances were influenced by infection status (P ≥ 0.05). Table 2.5 Relative abundances (%) of cecum digesta phyla by diet1,2 Phylum Cellulose Arabino- Bacteroidetes Verrucomicrobia Firmicutes Actinobacteria Gum arabic 0.60b (5.28) 40.92a (19.59) 54.76a (18.98) 0.07b (0.20) xylan 53.26a (12.94) 7.37bc (6.94) 38.75b (6.46) 0.08b (0.14) 44.91a (11.32) 2.51c (4.66) 48.16ab (9.75) 0.04b (0.08) Guar gum 45.60a (9.51) 10.20b (9.85) 41.48b (14.22) 0.46a (0.48) XOS 48.69a (8.85) 34.82a (12.52) 12.67c (6.52) 0.98a (0.58) 1Data presented as median (IQR), n = 22-30 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) In colon digesta (Table 2.6), Bacteroidetes (P < 0.001) was greater in mice fed arabinoxylan (54.08%), cellulose (55.95%), guar gum (51.57%), and XOS (55.48%) compared to gum arabic (0.56%). Verrucomicrobia (P < 0.001) was greatest in mice fed gum arabic (41.96%) and XOS (29.64%), and lowest in mice fed arabinoxylan (7.26%), cellulose (5.02%), and guar gum (11.53%). Firmicutes (P < 0.001) was greatest in mice fed gum arabic (53.91%), while mice fed XOS (13.25%) had the lowest abundance. Actinobacteria (P < 0.001) was greatest in mice fed XOS (1.57%) and guar gum (0.64%). Verrucomicrobia (P = 0.020) was greater in female (19.34%, 28.68%) than male mice (12.73%, 18.48%). There was a trend for Actinobacteria abundance (P = 0.078) to be lower in female mice (0.28%, 0.82%) compared to that in male mice (0.43%, 1.20%). Verrucomicrobia abundance (P = 0.025) was greater in L. monocytogenes positive mice 66 (22.64%, 26.61%) compared to negative mice (12.75%, 18.88%). Firmicutes abundance (P = 0.025) was greater in L. monocytogenes negative mice (37.44%, 13.41%) than in positive mice (33.49%, 24.94). Bacteroidetes and Actinobacteria abundances were not influenced by infection status (P ≥ 0.05). Table 2.6 Relative abundances (%) of colon digesta phyla by diet1,2 Phylum Cellulose Bacteroidetes Verrucomicrobia Firmicutes Actinobacteria Gum arabic 0.56b (8.83) 41.96a (22.50) 53.91a (20.91) 0.12b (0.28) Arabino- xylan 54.08a (5.16) 7.26b (6.04) 35.92b (8.32) 0.18b (0.20) Guar gum 51.57a (10.05) 11.53b (7.24) 35.60b (13.30) 0.64a (0.90) XOS 55.48a (12.68) 29.64a (10.46) 13.25c (7.75) 1.57a (0.98) 55.95a (7.26) 5.02b (6.76) 37.09b (7.73) 0.09b (0.24) 1Data presented as median (IQR), n = 22-30 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) Genera relative abundances To examine the relationship between diet and bacterial compositional changes in digesta, relative abundances of predominant genera from the digesta sources were compared. Five genera represent statistically significant (P < 0.001) genera in small intestine digesta (Table 2.7). Mice fed arabinoxylan, cellulose, and guar gum had the greatest relative abundance of Muribaculaceae genomosp. Gum arabic and XOS-fed mice had the greatest relative abundance of Akkermansia. Dubosiella was greatest in gum arabic, guar gum, and XOS-fed mice. Bifidobacterium was greatest in arabinoxylan, 67 guar gum, and XOS-fed mice. One member of Lachnospiraceae genera was lowest in XOS fed mice. Table 2.7 Relative abundances (%) of small intestine digesta genera by diet1,2 Genus Muribaculaceae genomosp. Akkermansia Arabino- Gum arabic 0.54b (12.48) 59.20a (18.60) 15.95a (12.62) 0.67b (1.12) 4.46a (8.73) xylan 63.40a (23.60) 15.97b (9.16) 0.47b (0.83) 1.84a (5.19) 2.68a (2.84) Cellulose Guar gum 60.42a (19.21) 20.07b (10.05) 6.42a (6.92) 2.72a (4.87) 1.11a (1.38) 68.93a (30.14) 6.42b (15.71) 0.06b (0.10) 0.46b (2.29) 1.44a (1.81) XOS 31.81b (26.59) 39.87a (16.60) 10.27a (14.63) 4.76a (4.23) 0.02b (0.02) Dubosiella Bifidobacterium Lachnospiraceae uncultured 1Data presented as median (IQR), n = 19-30 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) Eight genera represent statistically significant (P < 0.001) genera in cecum digesta (Table 2.8). Mice fed gum arabic and cellulose had the lowest relative abundance of Muribaculaceae genomosp. Meanwhile, gum arabic and XOS-fed mice had the greatest relative abundance of Akkermansia. Two Lachnospiraceae genera were lowest in XOS fed mice. Oscillibacter was greatest in mice fed arabinoxylan and guar gum. Alistipes was greatest in cellulose and guar gum-fed mice. Two Ruminiclostridium genera were greatest in mice fed arabinoxylan, cellulose, and guar gum. 68 Gum arabic 0.60b (5.28) 40.92a (19.59) 29.78a (31.44) 0.46b (2.43) 0.00c (0.02) 0.00c (0.02) 0.36b (1.02) 0.02b (0.42) xylan 49.59a (14.11) 7.37bc (6.94) 5.82b (6.08) 6.54a (6.46) 7.14a (4.29) 2.99b (3.13) 2.23a (2.27) 3.15a (1.77) Cellulose Guar gum 42.12a (8.65) 10.20b (9.85) 5.89b (4.50) 6.80a (18.74) 7.20a (4.29) 5.00ab (4.44) 2.43a (1.14) 1.39a (2.17) 32.55b (13.14) 2.51c (4.66) 7.47ab (3.31) 9.29a (4.52) 2.15b (1.18) 14.21a (11.48) 3.63a (1.99) 2.47a (0.96) XOS 48.66a (8.61) 34.82a (12.52) 0.02c (0.10) 0.00b (0.02) 0.00c (0.02) 0.00c (0.02) 0.00b (0.00) 0.00b (0.00) Lachnospiraceae uncultured Lachnospiraceae NK4A136 group Oscillibacter Alistipes Ruminiclostridium 9 Ruminiclostridium Table 2.8 Relative abundances (%) of cecum digesta genera by diet1,2 Genus Muribaculaceae genomosp. Akkermansia Arabino- 1Data presented as median (IQR), n = 22-30 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) Seven genera represent statistically significant (P < 0.001) genera in colon digesta (Table 2.9). Mice fed gum arabic had the lowest relative abundance of Muribaculaceae genomosp. Meanwhile, gum arabic and XOS-fed mice had the greatest relative abundance of Akkermansia. Two Lachnospiraceae genera were lowest in XOS- fed mice. Alistipes was greatest in cellulose fed mice. Oscillibacter was greatest in arabinoxylan and guar gum-fed mice. One member of Ruminiclostridium was greatest in arabinoxylan, cellulose and guar gum-fed mice. 69 Gum arabic 0.55c (8.27) 41.96a (22.50) 30.73a (32.65) 0.55b (2.21) 0.00c (0.02) 0.00c (0.02) 0.36cd (1.22) Arabino -xylan 50.35ab (7.39) 7.27b (6.04) 5.38ab (5.26) 7.73a (7.59) 3.25b (2.81) 5.10a (2.41) 1.97ab (1.36) 42.18b (17.86) 5.02b (6.76) 4.88ab (1.91) 8.00a (6.06) 15.18a (7.34) 1.62b (0.72) 2.61a (1.10) Guar gum 44.73b (9.99) 11.53b (7.24) 3.99b (4.43) 10.98a (14.53) 5.70b (4.23) 4.47a (3.51) 1.38bc (1.08) XOS 54.67a (12.68) 29.64a (10.45) 0.02c (0.08) 0.00b (0.02) 0.00c (0.02) 0.00c (0.02) 0.00d (0.02) Lachnospiraceae uncultured Lachnospiraceae NK4A136 group Alistipes Oscillibacter Ruminiclostridium 9 Table 2.9 Relative abundances (%) of colon digesta genera by diet1,2 Genus Muribaculaceae genomosp. Akkermansia Cellulose 1Data presented as median (IQR), n = 22-30 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) Correlation analysis of cecum bacterial taxa with cecum L. monocytogenes counts Verrucomicrobia (Akkermansia) relative abundance was weakly, positively correlated with cecum L. monocytogenes counts (r = 0.340, P = 0.005). Actinobacteria relative abundance and cecum L. monocytogenes counts had a weak, positive trending association (r = 0.240, P = 0.052). Oscillibacter (r = -0.264, P = 0.032) and Ruminiclostridium relative abundance (r = -0.312, P = 0.011) was weakly, negatively correlated with cecum L. monocytogenes counts. Lachnospiraceae NK4A136 group relative abundance had a trending negative (r = -0.227, P = 0.067) association with cecum L. monocytogenes counts. Correlation analysis of cecum bacterial taxa with cecum Shannon diversity Bacteroidetes relative abundance and Shannon diversity were weakly negatively correlated (r = -0.192, P = 0.031). Verrucomicrobia (Akkermansia) relative abundance 70 and Shannon diversity were moderately, negatively correlated (r = -0.657, P < 0.001). Firmicutes relative abundance and Shannon diversity were moderately, positively correlated (r = 0.616, P < 0.001). Actinobacteria relative abundance and Shannon diversity were weakly, negatively correlated (r = -0.376, P < 0.001). 71 DISCUSSION Dietary fiber sources were evaluated for their ability to affect mouse gastro- intestinal bacterial composition, L. monocytogenes infection, and to understand relationships between these responses. The BALB/CBy/J mouse is susceptible to wild- type L. monocytogenes EGD-e and is effective for modeling large intestine infection (Bou Ghanem et al., 2012; Bou Ghanem et al., 2013). Dietary fiber sources, when fed to animals, have mixed effects on severity of bacterial infection and host survival, and certain dietary fiber sources are protective whereas others enhance infection (Buddington et al., 2002; Ten Bruggencate et al., 2005; Petersen et al. 2009; Ebersbach et al., 2012; Hryckowian et al., 2018). Our results confirm that feeding diets containing different fiber sources differentially affect mouse gut bacterial composition, diversity measures, and L. monocytogenes counts in mouse cecum digesta and spleens. These effects were observed using a dietary fiber dose comparable to U.S. the Dietary Guidelines for Americans dietary fiber recommended intake (25-34 g/day) for people aged 14-30 y (Buddington et al., 2002; U.S. Department of Health and Human Services and U.S. Department of Agriculture, 2015). We assessed L. monocytogenes challenge by plating cecum digesta and mouse spleens on MOX agar 7 d post challenge. We chose to enumerate L. monocytogenes from cecum digesta because high counts were found in mouse cecums in a previous study (Zachar and Savage, 1979). The time period for detection was selected because peak L. monocytogenes infection occurs after 5 d, but fecal shedding is observed for up to 8 d in this model (Bou Ghanem et al., 2012). Plate counts revealed that dietary fiber sources differentially affected L. monocytogenes infection rate, with more XOS-fed mice (93%) testing positive for L. monocytogenes in either cecum digesta, spleen, or liver 72 compared to an average of (51%) for the other four dietary fiber treatments. XOS-fed mice also had the highest L. monocytogenes counts in cecum digesta and spleens. In a similar study, mice fed diets containing 10% XOS or 10% fructo-oligosaccharide (FOS) had increased Salmonella enterica serovar Typhimurium liver, spleen, and mesenteric lymph node counts compared to those fed a cornstarch control diet (Petersen et al., 2009). Contrary to our observations, Ebersbach et al. (2010) found that diets supplemented with 10% XOS or 10% galacto-oligosaccharide (GOS) improved guinea pig resistance to L. monocytogenes. However our results are not directly comparable due to differences in animal model and L. monocytogenes challenge strains, which both affect infection response (Hoelzer et al., 2012). It is important to note that L. monocytogenes cannot ferment xylose (Orsi and Wiedmann, 2016). Therefore, XOS is unlikely to contribute to L. monocytogenes enrichment in the present study. Dietary factors affect gastro-intestinal bacterial composition in mice and humans (Murphy et al., 2010; Holscher, 2017). We believe the gastro-intestinal microbiota to be an important factor differentiating the L. monocytogenes counts we observed. For example, germfree mice are highly susceptible to infection when orally challenged with 100 CFU of L. monocytogenes, while specific pathogen free mice can be challenged with up to 5 x 107 CFU of L. monocytogenes without obvious disease symptoms (Zachar and Savage, 1979). To understand how dietary fiber sources affect bacterial diversity and composition, we sequenced bacterial 16S rRNA genes from small intestine, cecum, and colon digesta from unchallenged and challenged mice. Dietary fiber sources did not differentiate small intestine Shannon diversity indices, but diversity was differentiated by fiber source in cecum and colon digesta. This finding is not surprising. In mammals, bacterial abundance increases from the small 73 intestine to the large intestine, and the cecum is an important site for dietary substrate fermentation in mice (Donaldson et al., 2016; Nguyen et al., 2015; Tropini et al., 2017). Additionally, small intestine transit time is faster than colon digesta transit, and bacterial nutrient metabolism differs by gastro-intestinal segment. In the small intestine, simple carbohydrates and proteins are preferred energy sources; while in the large intestine, complex carbohydrate such as non-digestible fibers and host glycoproteins are preferred (Tropini et al., 2017). We observed substrate-specific differences in cecum and colon digesta Shannon diversity. Mice fed XOS had the lowest bacterial diversity, whereas mice fed cellulose had the highest diversity. Diversity was also lower in mice fed gum arabic, arabinoxylan, and guar gum. All of the fiber sources except cellulose were readily fermentable, which was confirmed by the wet cecum proportion data that served as a proxy for fiber fermentation (Savage and Dubos, 1968; Remesy et al., 1992). Our Shannon diversity results are consistent with Liu et al. (2016) who observed decreased bacterial diversity in mice fed diets containing 10% FOS or 10% inulin compared to 5% or 10% cellulose. NMDS analysis and AMOVA statistical tests revealed bacterial communities clustered by dietary fiber source in all digesta samples. Bacterial communities did not cluster by sex in small intestine or cecum digesta, but did for colon digesta. Estrogens are metabolized by the gut microbiota (Baker et al., 2017), suggesting estrogen compounds may affect colon bacterial composition due to the influence of longer transit time. Bacterial communities did not cluster by L. monocytogenes positive:negative infection status in small intestine, cecum, or colon digesta samples. This confirms our assumption that diet would be the main factor affecting bacterial community structure, and not our challenge protocol. This is in agreement with Petersen et al. (2010) who 74 observed that S. enterica serovar Typhimurium challenge did not affect bacterial composition of mouse digesta sources. Overall, infection status did not infuence bacterial composition or diversity measures, and fiber source in the diet was the distinguishing factor for these observed differences. Low gastro-intestinal bacterial diversity is associated with dysbiosis, which is an imbalance in bacterial composition and changes to functional microbial metabolism (Nguyen et al., 2015; Zeng et al., 2017). Conversely, a “normal” gastro-intestinal microbiota stabilizes the community structure and resists compositional changes, which has been coined “colonization resistance” (Pickard et al., 2017). At the phylum level, the most dramatic compositional changes were observed with gum arabic and XOS-fed mice, which both had high relative abundance of Verrucomicrobia (Akkermansia) in all digesta sources. Similarly, Liu et al. (2016) observed feces of mice fed 10% FOS were enriched with Verrucomicrobia (17.0%) compared to mice fed 5% cellulose (0.08%). In mice, Akkermansia is an important Verrucomicrobia member in the gut. Its relative abundance is < 2% in mice (Xiao et al., 2015). In healthy humans, Bacteroidetes, Firmicutes, Actinobacteria and Proteobacteria account for 93.5% of bacterial sequences (Thursby and Juge, 2017) while Verrucomicrobia accounts for 3% of sequences (Geerlings et al., 2018). The high relative abundance of Verrucomicrobia in mice fed XOS and gum arabic is striking and important considering mouse and human gut microbiota are comparable at the phylum level, with Firmicutes and Bacteroidetes both highly abundant and dominantly represented phyla (Hugenholtz and de Vos, 2018). While both gum arabic and XOS-fed mice had high abundance of Verrucomicrobia, the relative abundances of Bacteroidetes, Firmicutes, and Actinobacteria differed substantially. Gum arabic-fed mouse digesta had very low 75 abundance of Bacteroidetes, but the highest abundance of Firmicutes. On the other hand, XOS-fed mice had a similar abundance of Bacteroidetes compared to the other treatments, low abundance of Firmicutes, and higher abundance of Actinobacteria. Gastro-intestinal bacteria rapidly respond to changes in dietary substrates, and dietary fiber sources are selective nutrient sources enabling certain bacterial taxa to thrive due to difference in expression of glycoside hydrolases, polysaccharide lyases, and carbohydrate esterases. Members of Firmicutes and Actinobacteria are more responsive to dietary fiber substrates even though they express fewer fiber degrading enzymes (Makki et al., 2018). At phylum level, we confirmed that feeding diets rich in different fiber sources selectively enrich certain phyla and provide new evidence for directional shifts. Notably, feeding XOS-rich diets enriches Verrucomicrobia and Actinobacteria in mice. We examined changes in genus-level bacterial abundances and observed an increase in quantity of statistically different and important genera from the small intestine to the large intestine. This finding is not surprising because the large intestine is the primary site of bacterial fermentation and we expected fiber sources to more profoundly affect bacterial composition where fermentation occurs (Donaldson et al., 2016; Nguyen et al., 2015; Tropini et al., 2017). We consider cecum digesta sources the most intriguing for deeper compositional analysis because of the corresponding L. monocytogenes plate counts from cecum digesta. Corresponding with the phylum level differences we observed, the genera of mice fed gum arabic and XOS were notably different than the other treatments. Important differences in the genera of XOS-fed mice include lower relative abundances of Lachnospiraceae, Alistipes, Oscillibacter, and Ruminiclostridium. Gum arabic-fed 76 mice had low abundances of Muribaculaceae, Alistipes, Oscillibacter, and Ruminiclostridium; but high abundance of an uncultured member of Lachnospiraceae. Muribaculaceae members belong to the phylum Bacteroidetes. These Gram-negative organisms, previously identified as Bacteroidales family S24-7, are common inhabitants of the mouse gut and have the capacity to digesta a wide variety of carbohydrates based on their genomes (Ormerod et al., 2016; Lagkouvardos et al., 2019). Alistipes is a member of the family Rickenellaceae within phylum Bacteroidetes. Its members are Gram-negative obligate anaerobes and four species have been identified (Könönen et al., 2015). Alistipes is associated with frail mice (Langille et al., 2014). Oscillibacter is a member of the family Ruminococcaceae within phylum Firmicutes. High abundance is associated with decreased gastro-intestinal barrier function and decreased tight junction protein expression in mice fed a high fat diet (Lam et al., 2012). Higher abundances of Alistipes and Oscillibacter are associated with frailty and poor health in elderly humans (Claesson et al., 2012). Ruminiclostridium is a member of the family Ruminococcaceae within the phylum Firmicutes. In a colitis model, Ruminococcaceae was associated with decreased enterocyte injury and decreased colon proinflammatory cytokine secretion in mice fed non-digestible soluble corn fiber (Valcheva et al., 2015). Lachnospiraceae is a family in phylum Firmicutes and has high gastro-intestinal representation in herbivores (Eren et al., 2015). The most notable shift in genus-level relative abundance occurred with Akkermansia, which is a member of the phylum Verrucomicrobia. In our study, feeding gum arabic or XOS enriched the mouse gut with Akkermansia. A. muciniphila is a mucin degrader present in humans and mice (Derrien et al., 2004; Derrien et al., 2008; Xiao et al., 2015). Some have found a positive association between Akkermansia and 77 gastrointestinal barrier function (Everard et al., 2013; Reunanen et al., 2015). However, Desai et al. (2016) found that dietary fiber deprivation of mice results in degradation of colon mucus and promotes increased epithelial access to the mouse pathogen Citrobacter rodentium. The mucus layer is rich in glycoproteins and can serve as an alternative energy source for the gastro-intestinal microbiota when dietary fibers are absent (Desai et al., 2016). In another study, A. muciniphila presence in S. enterica Typhimurium infected mice caused increased intestinal inflammation and disturbance of the mucus layer and resulted in significant Salmonella outgrowth (Ganesh et al., 2013). These studies support the hypothesis that degradation of the host mucus layer is a pathogenicity factor. Our finding of elevated Akkermansia in mice fed XOS reveals an important association between diet and listeriosis. Verrucomicrobia (Akkermansia) was positively associated with cecum L. monocytogenes counts. L. monocytogenes positive mice also tended to have greater Verrucomicrobia relative abundance than negative mice in cecum digesta, and its relative abundance was greater in colon digesta of positive mice. Our observations suggest that high Akkermansia relative abundance may provide ideal conditions for L. monocytogenes association with gastro-intestinal epithelial cells. One theory explaining this hypothesis is that Akkermansia degrades the protective mucus layer of the gastro-intestinal tract, thereby making epithelial cells accessible for attachment and invasion by L. monocytogenes. In humans, L. monocytogenes invades gastro-intestinal epithelial cells by the interaction of the internalin-A protein (InlA) with E-cadherin. In mice, this mechanism is less efficient due its low affinity to murine E- cadherin. However, Bou Ghanem et al. (2012) demonstrated the BALB/CBy/J mouse is susceptible to L. monocytogenes and InlA was not required to establish intestinal 78 infection. New evidence also suggests L. monocytogenes utilizes Listeria adhesion protein (LAP) to bind to heat shock protein 60 (Hsp60), which causes intestinal barrier dysfunction and increased translocation in mice (Drolia et al., 2018). In vitro dietary fiber fermentation with added mucin enriches mucin-degrading bacteria (Tran et al., 2016). L. monocytogenes causes intestinal cells to secrete mucins, which is though to inhibit infection by steric hindrance (Liévin-Le Moal et al., 2005; Laparra and Sanz, 2009). Our results suggest that high Akkermansia relative abundance provides ideal conditions for L. monocytogenes association with gastro-intestinal epithelial cells, potentially though enhanced degradation of the protective mucus layer. This hypothesis is supported by Schroeder et al. (2018) who observed that bacterial composition rapidly changed in mice fed a western style diet. This composition change resulted in decreased bacterial diversity and increased gut mucus bacterial-sized bead penetration. Moreover, mucus layer deterioration was prevented when mice were supplemented with inulin or bifidobacteria. These observations demonstrate certain dietary fiber sources and bacterial species affect barrier function differentially. We show XOS feeding is associated with higher L. monocytogenes cecum counts, and XOS may enhance invasive bacterial infections by enriching mucus-degrading bacteria. In the present study, we demonstrated that feeding diets rich in different dietary fiber sources alter mouse gastro-intestinal bacterial composition and diversity measures, and differentially affects L. monocytogenes gastro-intestinal cecum counts and migration to the spleen. We recognize our study lacked a true control and are unable to determine if dietary fiber sources promote or limit L. monocytogenes infection in susceptible mice. However, without dietary fiber, which is a fecal bulking agent, we would not have had adequate digesta material for experimental assays. Furthermore, 79 feeding a fiber-free diet would not be considered “normal” in nature. We did not evaluate gastro-intestinal mucus layer thickness or mechanisms for the effects we observed. However, dietary fibers fermentation results in the production of SCFAs. Dietary fibers and SCFAs both stimulate mucus production (Makki et al., 2018). Another limitation of the study was the absence of SCFA compositional data for mouse digesta sources. SCFAs are modulators of gastro-intestinal barrier function by influencing tight junction proteins (Peng et al., 2007; Peng et al., 2009). Detecting substrate-specific differences in SCFA profiles and relating them to our bacterial compositional data would provide evidence for potential mechanisms of action of fermentation metabolites on gastro-intestinal barrier function. Despite these weaknesses, this study confirms that discrete carbohydrate structures selectively enrich certain bacterial taxa (Hamaker and Tuncil, 2014). It is challenging to identify potentially beneficial or detrimental bacterial taxa associated with the gastro-intestinal phase of listeriosis in mixed culture without corresponding metabolomic data; however, our sequencing results confirm bacterial composition to be an important factor affecting susceptibility to listeriosis. Verrucomicrobia, Bacteroidetes, and Actinobacteria were negatively associated with Shannon diversity, while Firmicutes was positively associated. This leads us to believe that members of Firmicutes are important for maintaining stability and diversity, thereby preventing gastro-intestinal dysbiosis due to changes in diet or pathogen exposure. On one hand, Verrucomicrobia (Akkermansia) was associated with higher L. monocytogenes counts. On the other hand, Lachnospiraceae, Alistipes, Oscillibacter, and Ruminiclostridium contributed to relatively greater bacterial diversity and this appears to be protective against listeriosis, as evidenced by their higher relative abundances in mice fed 80 arabinoxylan, cellulose, and guar gum. The negative association we observed between L. monocytogenes cecum counts and Oscillibacter, Ruminiclostridium, and Lachnospiraceae NK4A136 group relative abundances further strengthens this conclusion. While the BALB/CBy/J mouse is an effective model of human listeriosis, we propose future bacterial 16S rRNA gene sequencing studies to be done with fecal samples derived from known human cases of listeriosis. Sequencing data from known cases will further aid in understanding bacterial compositional patterns associated with invasive listeriosis. Furthermore, we believe that feeding diets enriched in highly fermentable fiber sources or semi-purified fiber sources to be associated with decreased bacterial diversity. Therefore, additional investigation using our model with slowly fermentable fiber sources, such as resistant starches or whole foods, and their effects on listeriosis is warranted. 81 APPENDICES 82 APPENDIX A. MOTHUR MISEQ WORKFLOW pcr.seqs(fasta=silva.nr_v132.align, start=11894, end=25319, keepdots=F, processors=8) system(mv silva.nr_v132.pcr.align silva.v132.fasta) summary.seqs(fasta=silva.v132.fasta) make.contigs(file=current) summary.seqs(fasta=current) screen.seqs(fasta=current, group=current, maxambig=0, maxlength=260) unique.seqs(fasta=current) count.seqs(name=current, group=current) summary.seqs(count=current) align.seqs(fasta=current, reference=silva.v132.fasta) summary.seqs(fasta=current, count=current) screen.seqs(fasta=current, count=current, summary=current, start=1968, end=11550, maxhomop=8) summary.seqs(fasta=current, count=current) filter.seqs(fasta=current, vertical=T, trump=.) unique.seqs(fasta=current, count=current) pre.cluster(fasta=current, count=current, diffs=2) chimera.vsearch(fasta=current, count=current, dereplicate=t) remove.seqs(fasta=current, accnos=current) classify.seqs(fasta=current, template=silva.v132.fasta, taxonomy=silva.nr_v132.tax, method=knn, numwanted=1) remove.lineage(fasta=current, count=current, taxonomy=current, taxon=Chloroplast- Mitochondria-unknown-Archaea-Eukaryota) phylotype(taxonomy=current) 83 make.shared(list=current, count=current, label=1) classify.otu(list=current, count=current, taxonomy=current, cutoff=80, label=1) rename.file(taxonomy=current, shared=current) count.groups(shared=current) sub.sample(shared=current size=4983) rarefaction.single(shared=current, calc=sobs, freq=100) summary.single(shared=current, calc=nseqs-coverage-sobs-invsimpson-shannon, subsample=4983) dist.shared(shared=current, calc=thetayc-jclass, subsample=4983) pcoa(phylip=current) nmds(phylip=current, mindim=3, maxdim=3) amova(phylip=current, design=variable) 84 APPENDIX B: MOUSE DATA BY CAGE Figure 2.14A Mouse weights by diet and age1,2,3 1Data are LS means ± SEM, n = 13-16 per dietary treatment 2Mouse weights different by age period (P < 0.05) 3Data compiled by cage 85 Figure 2.15A Mouse weights by sex and age1,2,3 1Data are LS means ± SEM, n = 34-39 per sex 2Male mice weighed more than female mice for each age period (P < 0.05) 3Data compiled by cage 86 Figure 2.16A Mouse wet cecum proportion by diet1,2,3 1Data are presented as box and whisker plots (5-95%), n = 13-16 per dietary treatment 2Diets without a common letter are different (P < 0.05) 3Data compiled by cage 87 Figure 2.17A Mouse wet cecum proportion by sex1,2,3 1Data are presented as box and whisker plots (5-95%), n = 34-39 per sex 2Sexes are not different (P ≥ 0.05) 3Data compiled by cage 88 Figure 2.18A Positive L. monocytogenes samples in challenged mice by diet1,2,3 1Total positive is not necessarily cumulative of cecum, spleen and liver samples. 2Data presented as a frequency plot, n = 7-9 per dietary treatment 3Data compiled by cage 89 Figure 2.19A Cecum L. monocytogenes counts of challenged mice by diet and sex1,2,3,4 1Data presented as (Ln LS mean ± SEM CFU/mg Cecum Digesta) ESPC 2LS means without common letter are different (P < 0.05), n = 7-9 per treatment 3Dotted line represents limit of detection (Ln 200 CFU/mg) 4Data complied by cage 90 Figure 2.20A Cecum digesta L. monocytogenes counts of challenged mice by diet1,2,3,4 1Data presented as (Ln LS Mean ± SEM CFU/mg Cecum Digesta) ESPC 2LS means without common letter are different (P < 0.05), n = 7-9 per dietary treatment 3Dotted line represents limit of detection (Ln 200 CFU/mg) 4Data complied by cage 91 Figure 2.21A Spleen L. monocytogenes counts of challenged mice by diet and sex1,2,3,4 1Data presented as (Ln LS mean ± SEM CFU/mg Cecum Digesta) ESPC 2LS means without common letter are different (P < 0.05), n = 7-9 per treatment 3Dotted line represents limit of detection (Ln 200 CFU/mg) 4Data complied by cage 92 Figure 2.22A Spleen L. monocytogenes counts of challenged mice by diet1,2,3,4 1Data presented as (Ln LS Mean ± SEM CFU/mg Cecum Digesta) ESPC 2LS means without common letter are different (P < 0.05), n = 7-9 per dietary treatment 3Dotted line represents limit of detection (Ln 200 CFU/mg) 4Data complied by cage 93 Table 2.10A Shannon diversity index by diet and digesta sources1,2,3,4 Mouse Diet Gum arabic Arabinoxylan Cellulose Guar gum XOS Small intestine Cecum Colon 1.32 (0.28) 1.29 (0.48) 1.11 (0.64) 1.46 (0.45) 1.32 (0.19) 1.59bc (0.12) 1.86b (0.39) 2.39a (0.09) 1.92b (0.12) 1.22c (0.20) 1.72b (0.22) 1.81b (0.22) 2.15a (0.13) 1.84b (0.34) 1.26c (0.21) 1Data presented as median (IQR) 2Diets without a common letter within a column are different (P < 0.05) 3Small intestine (n = 12-16), Cecum (n = 13-16), Colon (n = 13-16) per dietary treatment 4Data complied by cage Table 2.11A Relative abundances (%) of small intestine digesta phyla by diet1,2,3 Phylum Bacteroidetes Cellulose Arabino- XOS Gum arabic 3.71c (12.59) 59.03a (20.96) 30.94a (15.26) 0.93c (1.03) xylan 64.38a (35.01) 14.37b (8.70) 14.75b (15.00) 3.29ab (5.76) 62.45a (24.51) 8.83b (17.25) 16.96ab (18.37) 1.03bc (2.16) Guar gum 58.00ab (14.87) 21.74b (10.55) 14.46b (8.96) 4.33a (3.96) 31.69bc (30.60) 37.99a (14.45) 17.14ab (19.19) 6.02a (5.02) Verrucomicrobia Firmicutes Actinobacteria 1Data presented as median (IQR), n = 12-16 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) 3Data compiled by cage 94 Table 2.12A Relative abundances (%) of cecum digesta phyla by diet1,2,3 Phylum Cellulose Arabino- Bacteroidetes Verrucomicrobia Firmicutes Actinobacteria Gum arabic 2.41b (6.36) 42.91a (17.94) 53.04a (16.00) 0.06b (0.16) xylan 52.11a (13.54) 8.14b (5.31) 38.75a (9.09) 0.16b (0.21) 43.53a (7.01) 2.51b (3.69) 50.27a (8.41) 0.03b (0.10) Guar gum 45.78a (10.09) 10.14b (9.43) 37.51a (15.91) 0.48a (0.63) XOS 52.04a (9.80) 35.76a (13.90) 12.56b (7.02) 1.08a (7.02) 1Data presented as median (IQR), n = 13-16 per dietary treatment 2Dietary fiber sources without a common letter within a row are different (P < 0.05) 3Data compiled by cage Table 2.13A Relative abundances (%) of colon digesta phyla by diet1,2,3 Phylum Cellulose Arabino- Bacteroidetes Verrucomicrobia Firmicutes Actinobacteria Gum arabic 3.83b (8.26) 44.17a (17.30) 54.90a (12.70) 0.09b (0.18) xylan 53.09a (4.75) 8.21b (4.68) 35.69ab (11.03) 0.21b (0.18) Guar gum 53.22a (12.18) 11.96b (10.64) 33.79b (14.15) 0.73a (0.82) XOS 55.61a (12.18) 29.30a (10.84) 12.06c (7.53) 1.63a (1.18) 55.52a (7.79) 6.47b (8.03) 35.57ab (4.92) 0.11b (0.25) 1Data presented as median (IQR), n = 13-16 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) 3Data compiled by cage 95 Table 2.14A Relative abundances (%) of small intestine digesta genera by diet1,2,3 Genus Muribaculaceae genomosp. Akkermansia Cellulose Arabino- Gum arabic 3.68c (12.59) 59.03a (20.96) 15.63a (12.37) 0.66b (0.82) 3.24a (8.33) xylan 62.91a (35.42) 14.37b (8.70) 0.57b (1.13) 3.28a (5.69) 2.81a (1.45) Guar gum 57.31ab (15.57) 21.74b (10.55) 6.52a (5.98) 4.20a (4.03) 1.32a (1.25) XOS 31.67bc (20.68) 37.99a (14.50) 10.13a (15.05) 4.68a (4.67) 0.01b (0.01) 61.81a (25.24) 8.83b (17.25) 0.06b (0.16) 0.47b (2.08) 1.75a (1.45) Dubosiella Bifidobacterium Lachnospiraceae uncultured 1Data presented as median (IQR), n = 12-16 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) 3Data compiled by cage 96 Table 2.15A Relative abundances (%) of cecum digesta genera by diet1,2,3 Genus Muribaculaceae genomosp. Akkermansia Cellulose Arabino- Guar gum 41.11ab (10.20) 10.14b (9.43) 5.62a (4.41) 6.02a (16.36) 4.67ab (3.00) 6.84a (3.79) 2.67a (0.86) 1.31ab (1.95) XOS 52.02a (9.79) 35.76a (13.90) 0.02b (0.09) 0.01b (0.02) 0.00cd (0.02) 0.01b (0.02) 0.00b (0.01) 0.00c (0.01) 30.73bc (12.80) 2.51b (3.69) 7.49a (4.60) 10.46a (5.84) 14.09a (13.83) 2.15b (1.34) 3.61a (2.02) 2.48a (0.87) Lachnospiraceae uncultured Lachnospiraceae NK4A136 group Alistipes Oscillibacter Ruminiclostridium 9 Ruminiclostridium Gum arabic 2.41c (6.36) 42.91a (17.94) 27.05a (30.68) 1.20b (1.96) 0.00d (0.01) 0.00b (0.01) 0.48b (0.86) 0.02bc (0.36) xylan 48.89a (13.42) 8.14b (5.31) 6.74a (4.49) 6.73a (5.73) 2.71bc (1.87) 7.03a (4.64) 2.35a (1.72) 3.57a (1.86) 1Data presented as median (IQR), n = 13-16 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) 3Data compiled by cage 97 Table 2.16A Relative abundances (%) of colon digesta genera by diet1,2,3 Genus Muribaculaceae genomosp. Akkermansia Arabino- Lachnospiraceae uncultured Lachnospiraceae NK4A136 group Alistipes Oscillibacter Ruminiclostridium 9 1Data presented as median (IQR), n = 13-16 per dietary treatment 2Diets without a common letter within a row are different (P < 0.05) 3Data compiled by cage Cellulose Guar gum 46.65ab (7.93) 11.96b (10.64) 3.47a (5.11) 7.69a (13.98) 4.83ab (3.48) 4.68ab (3.36) 1.29ab (0.88) 40.96bc (17.49) 6.47b (8.03) 4.91a (1.44) 7.56a (6.14) 15.18a (9.73) 1.53bc (0.52) 2.49a (1.08) XOS 55.59a (11.38) 29.30a (10.84) 0.02b (0.06) 0.01b (0.02) 0.00c (0.02) 0.01c (0.01) 0.00c (0.02) Gum arabic 3.81c (8.26) 44.17a (17.30) 32.07a (32.06) 1.28b (1.71) 0.01c (0.02) 0.00c (0.01) 0.34bc (0.99) xylan 49.44ab (6.70) 8.21b (4.68) 6.53a (3.57) 8.92a (5.89) 3.89b (2.88) 5.23a (2.26) 2.27a (1.12) 98 Table 2.17A Correlation analysis of cecum taxa with cecum Shannon diversity1,2,3 r -0.21 -0.70 0.64 -0.43 P 0.076 < 0.0001 < 0.0001 0.0001 r 0.29 0.29 -0.24 -0.27 P 0.065 0.065 0.13 0.09 Phylum/Genus Bacteroidetes Verrucomicrobia Firmicutes Actinobacteria Phylum/Genus Verrucomicrobia Akkermansia Oscillibacter Ruminiclostridium 1Data presented as Spearman correlation coefficients, n = 73 2Signfigance determined when (P < 0.05) 3Data compiled by cage Table 2.18A Correlation analysis of cecum taxa with cecum L. monocytogenes counts1,2,3 1Data presented as Spearman correlation coefficients, n = 40 2Signfigance determined when (P < 0.05) 3Data compiled by cage 99 REFERENCES 100 REFERENCES Baker JM, Al-Nakkash L, Herbst-Kralovetz MM. Maturitas Estrogen – gut microbiome axis: Physiological and clinical implications. Maturitas. Elsevier; 2017;103:45–53. Baray S. Acacia Gum. In: Cho S, Samuel P, editors. Fiber Ingredients: Food Applications and Health Benefits. Boca Raton: CRC Press; 2009. p. 122. Bou Ghanem EN, Jones GS, Myers-Morales T, Patil PD, Hidayatullah AN, D’Orazio SEF. InlA Promotes Dissemination of Listeria monocytogenes to the Mesenteric Lymph Nodes during Food Borne Infection of Mice. PLoS Pathog. 2012;8. Bou Ghanem E, Myers-Morales T, Jones G, D’Orazio S. Oral transmission of listeria monocytogenes in mice via ingestion of contaminated food. J Vis Exp. 2013;75:1–8. Bourquin LD, Titgemeyer EC, Fahey GC. Vegetable Fiber Fermentation by Human Fecal Short-Chain Fatty Acid Production during In Vitro Bacteria: Cell Wall Polysaccharide Disappearance and Fermentation and Water-Holding Capacity of Unfermented Residues. J Nutr. 1993;123:860–9. Buddington KK, Donahoo JB, Buddington RK. Dietary Oligofructose and Inulin Protect Mice from Enteric and Systemic Pathogens and Tumor Inducers. J Nutr. 2002;132:472– 7. Centers for Disease Prevention and Control. Estimates of Foodborne Illness in the United States [Internet]. Burden of Foodborne Illnesses in the United States. 2018 [cited 2019 Apr 26]. Available from: https://www.cdc.gov/foodborneburden/pdfs/scallan-estimated-illnesses-foodborne- pathogens.pdf Centers for Disease Control and Prevention. Listeria(Listeriosis) [Internet]. 2019 [cited 2019 Apr 17]. Available from: https://www.cdc.gov/listeria/index.html Claesson MJ, Jeffery IB, Conde S, Power SE, O’Connor EM, Cusack S, Harris HMB, Coakley M, Lakshminarayanan B, Sullivan OO, Fitzgerald GF, Deane J, O’Connor, M, Harnedy N, O’Connor K, O’Mahony D, van Sinderen D, Wallace M, Brennan L, Stanton C, Marchesi J, Fitzgerald A, Shanahan F, Hill C, Ross RP, O’Toole PW. Gut microbiota composition correlates with diet and health in the elderly. Nature. 2012;488:178–85. Cummings JH. Cellulose and the human gut. Gut. 1984;25:805–10. De Filippo C, Cavalieri D, Di Paola M, Ramazzotti M, Poullet JB, Massart S, Collini S, Pieraccini G, Lionetti P. Impact of diet in shaping gut microbiota revealed by a comparative study in children from Europe and rural Africa. Proc Natl Acad Sci. 2010;107:14691–6. 101 Derrien M, Collado MC, Ben-Amor K, Salminen S, de Vos WM. The Mucin Degrader Akkermansia muciniphila is an Abundant Resident of the Human Intestinal Tract. Appl Environ Microbiol. 2008;74:1646–8. Derrien M, Vaughan EE, Plugge CM, de Vos WM. Akkermansia muciniphila gen. nov., sp. nov., a human intestinal mucin-degrading bacterium. Int J Syst Evol Microbiol. 2004;54:1469–76. Desai MS, Seekatz AM, Koropatkin NM, Stappenbeck TS, Martens EC. Article A Dietary Fiber-Deprived Gut Microbiota Degrades the Colonic Mucus Barrier and Enhances Pathogen Article A Dietary Fiber-Deprived Gut Microbiota Degrades the Colonic Mucus Barrier and Enhances Pathogen Susceptibility. Cell. 2016;167:1339–53. DeVries JW. The definition of dietary fibre. Cereal Foods World. 2001;46:112–29. Donaldson GP, Lee SM, Mazmanian SK. Gut biogeography of the bacterial microbiota. Nat Rev Micriobiol. 2016;14:20–32. Drolia R, Tenguria S, Durkes AC, Turner JR, Bhunia AK, Drolia R, Tenguria S, Durkes AC, Turner JR, Bhunia AK. Listeria Adhesion Protein Induces Intestinal Epithelial Barrier Dysfunction for Bacterial Article Listeria Adhesion Protein Induces Intestinal Epithelial Barrier Dysfunction for Bacterial Translocation. Cell Host Microbe. Elsevier Inc.; 2018;23:470–484.e7. Dyets Inc. DYET #110700 AIN-93G Purified Rodent Diet [Internet]. 1993. Available from: https://dyets.com/wp-content/uploads/2015/04/D110700-AIN-93G-Purified- Rodent-Diet.pdf Ebersbach T, Jørgensen JB, Heegaard PM, Lahtinen SJ, Ouwehand AC, Poulsen M, Frøkiaer H, Licht TR. Certain dietary carbohydrates promote Listeria infection in a guinea pig model, while others prevent it. Int J Food Microbiol. Elsevier B.V.; 2010;140:218–24. Eren AM, Sogin ML, Morrison HG, Vineis JH, Fisher JC, Newton RJ, Mclellan SL. A single genus in the gut microbiome reflects host preference and specificity. ISME J. Nature Publishing Group; 2015;9:90–100. Everard A, Belzer C, Geurts L, Ouwerkerk JP, Druart C, Bindels LB, Guiot Y. Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity. Proc Natl Acad Sci. 2013;110. Flint HJ, Karen P, Louis P, Duncan SH. The role of the gut microbiota in nutrition and health. Nat Rev Gastroenterol Hepatol. Nature Publishing Group; 2012;9:577–89. Ganesh BP, Klopfleisch R, Loh G, Blaut M. Commensal Akkermansia muciniphila Exacerbates Gut Inflammation in Salmonella Typhimurium-Infected Gnotobiotic Mice. PLoS One. 2013;8:1–15. 102 Geerlings SY, Kostopoulos I, de Vos WM, Belzer C. Akkermansia muciniphila in the Human Gastrointestinal Tract: When, Where, and How? Microorganisms. 2018;6:1–26. Hamaker B, Tuncil Y. A Perspective on the Complexity of Dietary Fiber Structures and Their Potential Effect on the Gut Microbiota. J Mol Biol. 2014;426:3838–50. Hoelzer K, Pouillot R, Dennis S. Animal models of listeriosis: a comparative review of the current state of the art and lessons learned. Vet Res. 2012;43:1–27. Holscher HD. Dietary fiber and prebiotics and the gastrointestinal microbiota. Gut Microbes. Taylor & Francis; 2017;8:172–84. Hryckowian AJ, Treuren W Van, Smits SA, Davis NM, Gardner JO, Bouley DM, Sonnenburg JL. Microbiota-accessible carbohydrates suppress Clostridium difficile infection in a murine model. Nat Microbiol. Springer US; 2018;3:662–9. Hugenholtz F, de Vos WM. Mouse models for human intestinal microbiota research: a critical evaluation. Cell Mol Life Sci. Springer International Publishing; 2018;75:149–60. Jost L, DeVries P, Walla T, Greeney H, Chao A, Ricotta C. Partitioning diversity for conservation analyses analyses. Divers Distrib. 2010;16:65–76. Kellow NJ, Walker KZ. Authorised EU health claim for arabinoxylan. Foods, Nutrients and Food Ingredients with Authorised EU Health Claims, Volume 3. Elsevier Ltd; 2018. 201-218 p. Kimura Y, Nagata Y, Buddington RK. Nutrient Interactions and Toxicity Some Dietary Fibers Increase Elimination of Orally Administered Polychlorinated Biphenyls but Not That of Retinol in Mice 1. J Nutr. 2004;134:135–42. Könönen E, Song Y, Rautio M, Finegold SM. Alistipes. Bergey’s Manual of Systematics of Archaea and Bacteria. John Wiley & Sons, Inc.; 2015. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a Dual- Index Sequencing Strategy and Curation Pipeline for Analyzing Amplicon Sequence Data on the MiSeq. Appl Environ Microbiol. 2013;79:5112–20. Kumar G. A Review on Xylooligosaccharides. Int Res J Pharm. 2012;3:71–4. Lagkouvardos I, Lesker TR, Hitch TCA, Gálvez EJC, Smit N, Neuhaus K, Wang J, Baines JF, Abt B, Stecher B, Overman J, Strowig T, Clavel T. Sequence and cultivation study of Muribaculaceae reveals novel species, host preference, and functional potential of this yet undescribed family. Microbiome. Microbiome; 2019;7:1–15. 103 Lam Y, Ha C, Campbell C, Mitchell A, Dinudom A, Oscarsson J, Cook D, Hunt N, Caterson I, Holmes A, Storlien LH. Increased Gut Permeability and Microbiota Change Associate with Mesenteric Fat Inflammation and Metabolic Dysfunction in Diet-Induced Obese Mice. PLoS One. 2012;7:1–10. Langille MGI, Meehan CJ, Koenig JE, Dhanani AS, Rose RA, Howlett SE, Beiko RG. Microbial shifts in the aging mouse gut. Microbiome. 2014;2:1–12. Laparra JM, Sanz Y. Comparison of in vitro models to study bacterial adhesion to the intestinal epithelium. Lett Appl Microbiol. 2009;49:695–701. Liévin-Le Moal V, Servin AL, Coconnier-Polter MH. The increase in mucin exocytosis and the upregulation of MUC genes encoding for membrane-bound mucins induced by the thiol-activated exotoxin listeriolysin O is a host cell defence response that inhibits the cell-entry of Listeria monocytogenes. Cell Microbiol. 2005;7:1035–48. Liu T, Cephas KD, Holscher HD, Kerr KR, Mangian HF, Tappenden KA, Swanson KS. Nondigestible Fructans Alter Gastrointestinal Barrier Function, Gene Expression, Histomorphology, and the Microbiota Profiles of Diet-Induced Obese C57BL / 6J Mice. Journal Nutr. 2016;146:949–56. Makki K, Deehan EC, Walter J, Backhed F. The Impact of Dietary Fiber on Gut Microbiota in Host Health and Disease. Cell Host Microbe. 2018;23:705–15. Mudgil D, Barak S, Khatkar BS. Guar gum: Processing, properties and food applications - A Review. J Food Sci Technol. 2014;51:409–18. Murphy EF, Cotter PD, Healy S, Marques TM, O’Sullivan O, Fouhy F, Clarke SF, O’Toole PW, Quigley EM, Stanton C, O’Doherty RM, Shanahan F. Composition and energy harvesting capacity of the gut microbiota: relationship to diet, obesity and time in mouse models. Gut. 2010;59:1635–42. Nguyen TLA, Vieira-Silva S, Liston A, Raes J. How informative is the mouse for human gut microbiota research? Dis Model Mech. 2015;8:1–16. Ormerod KL, Wood DLA, Lachner N, Gellatly SL, Daly JN, Parsons JD, Dal’Molin CGO, Palfreyman RW, Nielsen LK, Cooper MA, Morrison M, Hansbro PM, Hugenholtz P. Genomic characterization of the uncultured Bacteroidales family S24-7 inhabiting the guts of homeothermic animals. Microbiome; 2016;4:1–17. Orsi RH, Wiedmann M. Characteristics and distribution of Listeria spp., including Listeria species newly described since 2009. Applied Microbiology and Biotechnology. 2016;2009:5273–87. Peng L, Li Z-R, Green RS, Holzman IR, Lin J. Butyrate Enhances the Intestinal Barrier by Facilitating Tight Junction Assembly via Activation of AMP-Activated Protein Kinase in Caco-2 Cell Monolayers. J Nutr. 2009;139:1619–25. 104 Peng L, He Z, Chen W, Holzman IR, Lin J. Effects of butyrate on intestinal barrier function in a caco-2 cell monolayer model of intestinal barrier. Pediatr Res. 2007;61:37– 41. Petersen A, Bergstrom A, Andersen J, Hansen M, Lahtinen S, Wilcks A, Licht T. Analysis of the intestinal microbiota of oligosaccharide fed mice exhibiting reduced resistance to Salmonella infection. Benef Microbes. 2010;1:271–81. Petersen A, Heegaard PMH, Pedersen AL, Andersen JB, Sørensen RB, Frøkiær H, Lahtinen SJ, Ouwehand AC, Poulsen M, Licht TR. Some putative prebiotics increase the severity of Salmonella enterica serovar Typhimurium infection in mice. BMC Microbiol. 2009;9:1–11. Phillips GO. Acacia gum (Gum Arabic): a nutritional fibre; metabolism and calorific value. Food Addit Contam. 1998;15:251–64. Pickard JM, Zeng MY, Caruso R, Núñez G. Gut microbiota: Role in pathogen colonization , immune responses , and inflammatory disease. Immunol Rev. 2017;279:70–89. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Glo FO, Yarza P. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41:590–6. Rabbani GH, Ahmed S, Hossain I, Islam R, Marni F, Akhtar M, Majid N. Green Banana Reduces Clinical Severity of Childhood Shigellosis. Pediatr Infect Dis J. 2009;28:420–5. Reeves PG, Nielsen FH, Fahey GC. Committee Report AIN-93 Purified Diets for Laboratory Rodents: Final Report of the American Institute of Nutrition Ad Hoc Writing Committee on the Reformulation of the AIN-76A Rodent Diet. J Nutr. 1993;123:1939– 51. Remesy C, Behr SR, Levrat M-A, Demigne C. Fiber Fermentability in the Rat Cecum and its Physiological Consequences. Nutr Res. 1992;12:1235–44. Roberfroid M, Slavin J. Nondigestible Oligosaccharides. Crit Rev Food Sci Nutr. 2000;40:461–80. Reunanen J, Kainulainen V, Huuskonen L, Ottman N, Belzer C, Huhtinen H, de Vos WM. Akkermansia muciniphila Adheres to Enterocytes and Strengthens the Integrity of the Epithelial Cell Layer. Appl Environ Microbiol. 2015;81:3655–62. Savage DC, Dubos R. Alterations In The Mouse Cecum And Its Flora Produced By Antibacterial Drugs. J Exp Med. 1968;1:97–110. 105 Schroeder BO, Birchenough GMH, Arike L, Sta M, Johansson ME V, Arike L, Johansson MEV, Hansson GC, Backhed F. Bifidobacteria or Fiber Protects against Diet-Induced Microbiota-Mediated Colonic Mucus Deterioration. Cell Host Microbe. 2018;23:27–40. Sinnott M. Primary Structure and Conformation of Oligosaccharides and Polysaccharides. Carbohydrate Chemistry and Biochemistry Structure and Mechanism. Cambridge: The Royal Society of Chemistry; 2007. p. 194. Stewart ML, Slavin JL. Molecular weight of guar gum affects short-chain fatty acid profile in model intestinal fermentation. Mol Nutr Food Res. 2006;50:971–6. Ten Bruggencate SJM, Bovee-Oudenhoven IMJ, Lettink-Wissink MLG, Van der Meer R. Dietary fructooligosaccharides increase intestinal permeability in rats. J Nutr. 2005;135:837–42. Thursby E, Juge N. Introduction to the human gut microbiota. Biochem J. 2017;474:1823–36. Tran THT, Boudry C, Everaert N, Thewis A, Portetelle D, Daube G, Nezer C, Taminiau B, Bindelle J. Adding mucins to an in vitro batch fermentation model of the large intestine induces changes in microbial population isolated from porcine feces. FEMS Microbiol Ecol. 2016;92:1–13. Tropini C, Earle KA, Huang KC, Sonnenburg JL. Review The Gut Microbiome: Connecting Spatial Organization to Function. Cell Host Microbe. Elsevier Inc.; 2017;21:433–42. U.S. Department of Health and Human Services, U.S. Department of Agriculture. 2015 – 2020 Dietary Guidelines for Americans. 2015 – 2020 Dietary Guidelines for Americans. 8th ed. 2015. Ullman-Culleré MH, Foltz CJ. Body Condition Scoring: A Rapid and Accurate Method for Assessing Health Status in Mice. Lab Anim Sci. 1999;49:319–23. Valcheva R, Hotte N, Gillevet P, Sikaroodi M, Thiessen A, Madsen KL. Soluble Dextrin Fibers Alter the Intestinal Microbiota and Reduce Proinflammatory Cytokine Secretion in Male IL-10 – Deficient Mice. J Nutr. 2015;2060–6. Williams BA, Grant LJ, Gidley MJ, Mikkelsen D. Gut Fermentation of Dietary Fibres: Physico-Chemistry of Plant Cell Walls and Implications for Health. Int J Mol Sci. 2017;18:1–25. World Health Organization. Listeriosis [Internet]. Fact Sheet. 2018 [cited 2019 Apr 26]. Available from: https://www.who.int/mediacentre/factsheets/listeriosis/en/ 106 Xiao L, Feng Q, Liang S, Sonne SB, Xia Z, Qiu X, Li X, Long H, Zhang J, Zhang D, Liu C, Fang Z, Chou J, Glanville J, Hao Q, Kotowska D, Colding C, Licht TR, Wu D, Yu J, Sung JJY, Liang Q, Li J, Jia H, Lan Z, Tremaroli V, Dworxynski P, Nielsen HB, Backhead F, Dore J, Le Chatelier E, Ehrlich SD, Lin JC, Arumugam M, Wang J, Madsen L, Kristiansen K. A catalog of the mouse gut metagenome. Nat Biotechnol. 2015;33:1103–8. Zachar Z, Savage DC. Microbial interference and colonization of the murine gastrointestinal tract by Listeria monocytogenes. Infect Immun. 1979;23:168–74. Zeng M, Inohara N, Nunez G. Mechanisms of inflammation-driven bacterial dysbiosis in the gut. Nature. 2017;10:18–26. Zhou J, Martin RJ, Tulley RT, Raggio AM, Shen L, Mccutcheon K, Keenan MJ. Failure to ferment dietary resistant starch in specific models of obesity results in no body fat loss. J Agric Food Chem. 2009;57:8844–51. 107 CHAPTER 3: EFFECT OF DIETARY FIBER SOURCES ON MICROBIAL POPULATIONS AND SHORT-CHAIN FATTY ACID PRODUCTION IN VITRO ABSTRACT Background: Dietary factors influence human gastro-intestinal bacterial composition, and non-digestible substrates are differentially metabolized due to variations in bacterial enzyme expression. Bacterial fermentation of dietary fiber sources results in the production of short-chain fatty acids (SCFA), which are saturated fatty acids with six or fewer carbons known for beneficial effects on gastro-intestinal health. This research expands existing knowledge of dietary fiber sources effects on the human gastro- intestinal microbiota and fermentation metabolites. Objective: The objectives of this study were as follows: 1) quantify SCFAs produced during in vitro fermentation of five fiber sources using human colonic bacteria; 2) determine the bacterial compositions in human fecal inoculum and changes in these bacterial populations during in vitro fermentation of different fiber sources; and 3) relate SCFA production to bacterial composition. Methods: Gum arabic, arabinoxylan, cellulose, guar gum, and xylo-oligosaccharide (XOS) were analyzed for their potential to change the composition of bacterial populations and fermentation end products in vitro using fecal samples from healthy human volunteers (n = 6) as inocula. Fiber sources were fermented for 24 h at 37 °C. SCFAs were analyzed by gas chromatography. Bacterial composition was determined by 16S rRNA gene sequencing. Data were analyzed using standard parametric and nonparametric statistical tests. 108 Results: Cellulose fermentation in vitro and consequent SCFA production was very limited compared to the other four fiber sources (P < 0.05). Bacterial diversity measures were not differentiated during fermentation of the different fiber sources (P ≥ 0.05), nor were the relative abundances of Bacteroidetes, Firmicutes, and Verrucomicrobia. Relative abundances of Proteobacteria, Actinobacteria, and nine Firmicutes genera were altered by fermentation of different fiber sources (P < 0.05). Specific bacterial taxa were significantly correlated both positively and negatively with SCFA concentrations (P < 0.05). Conclusion: Dietary fiber sources rapidly affect the bacterial composition of human colon bacterial phyla with low representation, as well as genera within the phylum Firmicutes. In vitro modeling using human fecal bacteria is useful for identifying potential beneficial bacterial genera; however, it may not suitable for understanding the complex interaction between the immune system, gut microbiota, and fermentation end products. Keywords: dietary fiber, fermentation, short-chain fatty acids, microbiota 109 INTRODUCTION Diet and other environmental factors are important modulators of the gastro- intestinal microbiota (Rothschild et al., 2018). Humans consuming diets rich in dietary fiber sources have distinct bacterial communities compared to those consuming Western style diets (De Filippo et al., 2010). Diets low in fiber sources can lead to decreased gastro-intestinal bacterial diversity and a reduction in beneficial bacterial metabolites (Makki et al, 2018). Numerous others have investigated in vitro fermentation metabolites of dietary fiber sources (Bourquin et al., 1992; Rose et al., 2010; Chen et al., 2017; Pham et al., 2018). To our knowledge, few studies have evaluated the effects of the semi-purified dietary fiber sources on bacterial populations using 16S ribosomal ribonucleic acid (rRNA) gene sequencing in concordance with SCFA analysis in an in vitro human colon model. In vitro batch fermentation is one approach for modeling human colonic fermentation of dietary substrates in a controlled manner, and is useful for quantitatively assessing changes in bacterial composition and metabolic end products without the ethical constraints of in vivo studies (Wang et al., 2019). Resident micro-organisms of the human gastro-intestinal tract are estimated to number 1014 and outnumber human cells as much as 10-fold (Thursby and Juge, 2017). Recent advances in genetic sequencing, namely 16S rRNA gene sequencing, have allowed researchers to investigate the co-evolving relationship between the microbiota and humans. Dietary factors are at the forefront of investigation, and dietary fiber sources and their effect on the human microbiota is the focus of this research. Dietary fibers are defined by the American Association of Cereal Chemists as the “edible parts of plants or analogous carbohydrates that are resistant to digestion and 110 absorption in the human small intestine with complete or partial fermentation in the large intestine. Dietary fiber includes polysaccharides, oligosaccharides, lignin, and associated plant substances” (DeVries, 2001). In the U.S., recommended dietary fiber intake for adults is 25 – 34 g/day depending on age and gender (U.S. Department of Health and Human Services, U.S. Department of Agriculture, 2015). Average U.S. intake is low for males (18 g/day) and females (15 g/day), and only 5% of the adult population meets current recommendations (Hoy and Goldman, 2014; Quagliani and Felt- Gunderson, 2017). The human health benefits of diets rich in fiber sources are many-fold and include improved laxation, management of blood cholesterol and blood glucose; and reduced risk of cardiovascular disease, type II diabetes, and colorectal cancer (Abrams and Bishop, 1967; DeVries, 2001; Anderson et al., 2009; Threapleton et al., 2013; World Cancer Research Fund and American Institute for Cancer Research, 2011). Additionally, recent research has identified SCFA concentration as a potential marker of gastro- intestinal heath (Verbeke et al., 2015). Dietary fiber fermentation by the gastro-intestinal microbiota results in the production of SCFAs, which are volatile organic acids with six or fewer saturated carbons. The major SCFAs produced from dietary fiber fermentation are acetate, propionate, and butyrate, which account for 95% of the SCFAs produced in humans (den Besten et al., 2013). The composition of SCFAs is dependent on dietary substrate (Bourquin et al., 1992; Martens et al., 2014; Verbeke et al., 2015) and microbial composition (Thursby and Juge, 2017), but on average these three SCFAs have a molar ratio of 60:20:20 in human digesta (den Besten et al., 2013). Isobutyrate, valerate, isovalerate, hexanoate, and heptanoate are also found in human digesta in low 111 concentrations (Zhao et al., 2006). Other minor fermentation intermediates and end products include: lactate, succinate, formate, butanediol, butanol, and acetone (Müller, 2001). Total colon luminal content of SCFAs ranges from 20-140 mM and approximately 95% is absorbed by passive diffusion or active transport by gastro- intestinal epithelial cells (den Besten et al., 2013). Total daily SCFA production is estimated to be 400-600 mmol/d in humans (Verbeke et al., 2015). The dietary fiber sources used in this research were gum arabic, arabinoxylan, cellulose, guar gum, and XOS. Numerous researchers have investigated in vitro fermentations of other dietary fiber sources, such as resistant starch, inulin, fructo- oligosaccharide (FOS), and galacto-oligosaccharide (GOS). However, few studies have examined the effect of fiber sources selected in the present study on bacterial compositional changes (Wang et al., 2019). Since bacterial composition and their metabolites are highly dependent on dietary fiber sources, in vitro modeling is important for understanding the effects of select dietary fiber sources and their potential impact on human health. We hypothesized that bacterial fermentation of these different fiber sources would produce differing SCFA profiles and alter colonic bacterial composition. Additionally, we hypothesized that SCFA profiles to be related to bacterial composition. The objectives of this study were as follows: 1) quantify SCFAs produced during in vitro fermentation of five fiber sources using human colonic bacteria; 2) determine the bacterial compositions in human fecal inoculum and changes in these bacterial populations during in vitro fermentation of different fiber sources; and 3) relate SCFA production to bacterial composition. 112 MATERIALS AND METHODS Research participants An equal number of healthy adult female and males (n = 6) were recruited from the local population to supply fecal samples, which were used as bacterial inoculum sources for 24 h in vitro fermentations of dietary fiber sources. Research participants were recruited via Michigan State University Communication Arts and Sciences’ Paid Research Pool system and compensated $50/each for their fecal sample donation. Eligible participants were identified as healthy adults who maintained their routine dietary pattern, did not prescribe to diets restricting energy or specific food groups, did not use antibiotics during the previous 3-mo period, and had an unremarkable history of gastro-intestinal related diseases. Research subjects completed one 24 h dietary recall using the National Cancer Institute’s Automated Self-Administered 24-Hour (ASA24) Dietary Assessment Tool. The dietary assessment covered the 24 h period immediately prior to fecal sample submission. This study was conducting according to the guidelines in the Declaration of Helsinki, and all procedures research involving human subjects were approved by the Michigan State University Human Research Protection Program. Written informed consent was obtained from all research participants. Study design Five semi-purified dietary fiber sources were obtained commercially and were gum arabic, arabinoxylan, cellulose, guar gum, and XOS. Each of the five fiber sources was subjected to in vitro fermentation by fecal inoculum prepared from the six donor sources. In addition, the fiber sources were fermented with a composite fecal inoculum prepared by combining that from each of the 113 six fecal donors. Substrates were fermented in duplicate in sealed serum vials for 24 h in a water bath maintained at 37 °C. Duplicate vials containing no fiber sources were also incubated for 24 h for each fecal inoculum. The total assay consisted of 84 serum vials. The dietary fiber substrates, 300 mg each, were hydrated in 25 mL of semi- defined anaerobic medium in sealed 250 mL glass Wheaton serum vials in which atmospheric air was replaced with 99.8% pure CO2 delivered through an in-line filter (Whatman HEPA-Vent, GE Healthcare, Chicago, IL). Each vial was aseptically injected with 1 mL of freshly prepared cysteine HCl-H20 solution to maintain anaerobiosis. Vials were stored overnight at 2-6 °C. After overnight hydration, fermentation vials were aseptically injected with 4 mL of freshly prepared fecal inoculum to initiate fermentation. Serum vials were maintained at 37 °C in a water bath with periodic shaking. After 24 h, gas volume was determined by displacement in a 60 mL syringe, and aliquots were removed for determining pH. Additional aliquots were collected for SCFA analysis by gas chromatography and stored at -80 °C. Bacterial cell pellets were collected for 16S rRNA gene sequencing and stored at -80 °C. Human fecal sample collection Research participants were given specimen collection pans (Nuns Hat Specipan Collection Unit) and sealable plastic bags in a non-descript cardboard box prior to sample collection. Participants were instructed to collect fecal samples in sealable plastic bags, evacuate air, and transport at room temperature immediately to our research lab on the campus of Michigan State University. Fecal samples were processed according to the procedures described by Bryant and Burkey (1953) and Bourquin et al. 114 (1992). Briefly, feces were diluted 1:10 in anaerobic dilution solution and blended for 60 sec at room temperature in a laboratory stomacher (Seward, Islandia, NY) with atmospheric air replaced with in-line filtered CO2. A composite fecal slurry was made by taking equal fecal masses from each donor and diluting 1:10 in anaerobic dilution solution. Blended feces were filtered through four layers of cheesecloth and sealed in 250 mL glass Wheaton serum vials (Sigma-Aldrich, St. Louis, MO) with atmospheric air replaced with in-line filtered CO2. All materials used in this procedure were autoclaved, dry sterilized, or sterilized and individually wrapped by the manufacturer. Anaerobic dilution solution preparation The anaerobic dilution solution (Table 3.1) was prepared, with slight modification, according to the procedures described by Bryant and Burkey (1953) and Bourquin et al. (1992). Wet and dry components were added to a 1 L volumetric flask and filled to capacity with purified distilled water (Barnstead International, Dubuque, IA), The anaerobic dilution solution was transferred to Pyrex media storage bottles and filled to 50% capacity. Containers were sealed and the dilution solution was autoclaved for 30 min. After cooling to room temperature, the solution was purged with 99.8% pure CO2 delivered through an in-line filter to adjust pH to 6.7 to 6.8. Prepared solutions were stored at 2-6 °C in sealed Pyrex media storage bottles until use. 115 Table 3.1 Anaerobic dilution solution composition Concentration Component Solution A1 Solution B2 Trace mineral solution3 Resazurin solution4 Distilled water Na2CO3 Cysteine HCl-H2O 1Composition (g/L): NaCl, 5.4; KH2PO4, 2.7; CaCl2-H20, 0.16; MgCl- 6H20, 0.12; MnCl2-4H20, 0.06; CoCl2-6H20, 0.06; (NH4)2SO4, 5.4 2Composition: K2HPO4, 2.7 g/L 3Composition (mg/L): EDTA (disodium salt), 500; FeSO4-7H2O, 200; ZnSO4-7H2O, 10; MnCl2-H2O, 3; H3PO4, 30; CoCl2-6H2O, 20; CuCl2-2H2O, 1; NiCl2-6H2O, 2; NaMoO4-2H2O, 3 4Resazurin, 1 g/L in purified distilled water mL/L 330 330 10 1.0 329 g/L 3.0 0.5 Semi-defined anaerobic medium preparation The semi-defined anaerobic medium (Table 3.2) was prepared and used the same day of preparation according to the procedures described by Bourquin et al. (1992) with slight modification. Wet and dry components, except the vitamin mixture and cysteine HCl-H20 solution, were added to a 1 L volumetric flask and filled with purified distilled water. The medium was transferred to a 5 L round-bottom flask and a butyl rubber stopper was secured with steel baling wire. The medium was autoclaved for 30 min. After cooling to room temperature, the sterile-filtered (0.22 μM filter system, Sigma- Aldrich, St. Louis, MO) vitamin mixture was added, and the medium was purged with 99.8% pure CO2 delivered through an in-line filter to adjust pH to 6.7 to 6.8. 116 Table 3.2 Semi-defined anaerobic medium composition Concentration mL/L 330 330 10 20 2.5 0.4 1.0 267.5 38.5 g/L 0.5 0.5 4.0 Component Solution A1 Solution B2 Trace mineral solution3 Vitamin mixture4 Hemin solution5 SCFA mixture6 Resazurin solution7 Distilled water Cysteine HCl-H20 solution8 Yeast extract Trypticase Na2CO3 1Composition (g/L): NaCl, 5.4; KH2PO4, 2.7; CaCl2-H20, 0.16; MgCl-6H20, 0.12; MnCl2-4H20, 0.06; CoCl2-6H20, 0.06; (NH4)2SO4, 5.4 2Composition: K2HPO4, 2.7 g/L 3Composition (mg/L): EDTA (disodium salt), 500; FeSO4-7H2O, 200; ZnSO4-7H2O, 10; MnCl2-H2O, 3; H3PO4, 30; CoCl2-6H2O, 20; CuCl2-2H2O, 1; NiCl2-6H2O, 2; NaMoO4-2H2O, 3 4Composition (mg/L): thiamin-HCl, 100; pantothenic acid, 100; niacin, 100; pyridoxine, 100; riboflavin, 100; folate, 2.5; biotin, 2.5; p-aminobenzoic acid, 5; vitamin B-12, 0.25; phylloquinone solution, 10 mL/L. Phylloquinone solution composition: 5 mL/L in ethanol 5Hemin, 500 mg/L in 0.01 mol/L NaOH 6SCFA mixture contained 250 mL/L each of n-valerate, isovalerate, isobutyrate, and DL- alpha-methyl butyrate 7Resazurin, 1 g/L in purified distilled water 8Cysteine HCl-H20 in purified distilled water: 13 mg/mL; mixture autoclaved 15 min in sealed serum vials Gas volume determination Fermentation vessels were removed from the 37 °C water bath and placed in a room temperature water bath 30 min prior to the end of the 24 h fermentation period to normalize to room temperature. Gas volume was measured by using a 60 mL plastic syringe with Luer-lok tip (Becton Dickinson, Franklin Lakes, NJ) and 18-gauge needle to 117 puncture the butyl rubber serum stopper and the release gas into the syringe body (Noack et al., 2013). Fermentation replicates were averaged prior to statistical analysis. Preparation of fermentation end product aliquots and bacterial pellets After determining gas volume, fermentation end products were transferred into 50 mL centrifuge tubes and centrifuged at 2300 x g for 30 minutes. Aliquots were removed for pH determination and subsequent SCFA analysis. Fermentation end products not assayed immediately were stored at -80 °C. Cell pellets from the 50 mL centrifuge tubes were collected for 16S rRNA gene sequencing. Cell pellets were weighed and transferred using sterile single-use plastic spatulas. Cell masses of approximately 200 mg from each vial were transferred into autoclaved 2.0 mL centrifuge tubes and stored at -80 °C. pH determination Potential hydrogen of each fermentation vial was determined with an Oakton 11 series pH meter (Oakton Instruments, Vernon Hills, IL). Fermentation replicates were averaged prior to statistical analysis. SCFA gas chromatography analysis SCFAs from in vitro fermentations (n = 35) and fecal samples (n = 6) were quantified using a Thermo Trace 1310 gas chromatograph coupled to a flame ionization detector with a Thermo TG-WAXMS A Gas Chromatography Column (0.32 mm diameter, 30 m length, and 0.25 μm film thickness) (Thermo Fisher Scientific, Waltham, MA). Operating conditions and stock standards were prepared according to 118 the procedures described by Zhao et al. (2006). Briefly, stock standards were prepared for acetic acid, propionic acid, n-butyric acid at concentrations of 400 mM; n-valeric acid and isovaleric acid standards were prepared at a concentration of 200 mM; 100 mM for isobutyric acid, 50 mM m for n-caproic acid and 15 mM for n-heptanoic acid. Calibration curves were generated from four stock standard solutions (APPENDIX A, Table 3.8A). A 2-ethylbutyric acid solution containing 12% formic acid was prepared as the internal standard stock solution. Frozen fecal samples were thawed and diluted 1:10 in water and homogenized for 3 min. Fecal slurry pH was adjusted to 2-3 with 5 M HCl, and then centrifuged for 2o min. The internal standard was added to the final supernatant at a concentration of 1 mM prior to gas chromatography injection. SCFAs were identified on chromatograms by their retention times. Calibration curves were generated for each compound relative to the internal standard. In all cases, individual SCFA concentrations were corrected by subtracting the corresponding donor blank SCFA values. Fermentation replicates were averaged for statistical analysis. DNA preparation for 16S rRNA gene sequencing One group of 24-h in vitro fermentation replicates (n = 42) and the original and composite fecal samples (n = 7) were utilized for bacterial 16S rRNA gene sequencing. The DNA Stool Mini Kit (QIAGEN, Germantown, MD) was utilized for isolation and purification of genomic deoxyribonucleic acid (DNA). The protocol for pathogen detection with lysis temperature of 95 °C was selected to limit bias against Gram- positive organisms. Purified DNA was stored at -20 °C until further use. Genomic DNA yield was determined by a Qubit 2.0 Flurometer (Invitrogen, Waltham, MA) and was normalized to a final concentration of 5 ng/uL with 119 RNase/DNase-free water (Fisher BioReagents, Waltham, MA). Purified DNA was tested for amplification by polymerase chain reaction (PCR). The V4 region of 16S rRNA gene was amplified with the HotStarTaq Plus Master Mix kit (QIAGEN, Germantown, MD) and 515F (5’ – GTG CCA GCM GCC GCG GTA A – 3’) and 806R (5’ – GGA CTA CHV GGG TWT CTA AT – 3’) primers (Integrated DNA Technologies, Coralville, IA) using the protocol described by Caporaso et al. (2011). Polymerase chain reactions (PCR) contained 10 μL HotStarTaq Plus Master Mix, 1 μL each of the forward and reverse primers (10 μM initial concentration), 5 μL genomic DNA (5 ng/μL), and 3 μL RNase- free water. Polymerase chain reaction cycle temperatures and times were 94 °C for 3 min to denature template DNA; and 35 cycles of 94 °C for 45 s, 50 °C for 60 s, and 72 °C for 90 s for amplification. The final extension was 10 min at 72 °C. Target DNA banding at approximately 300 - 350 base pairs was visualized on a 1% agarose gel with ethidium bromide. Genomic DNA from 42 fermentation samples, 7 fecal samples, and 1 mock community (ATCC MSA-1002, American Type Cell Culture, Manassas, VA) were submitted for 16S rRNA gene sequencing by the Research Technology Support Facility Genomics Core at Michigan State University. 16S rRNA gene sequencing Amplicon libraries for V4 region of the 16S rRNA gene were created using the dual indexed Illumina library primers, 515F/806R, as described by Kozich et al. (2013). PCR products were normalized using SequalPrep DNA normalization plate (Invitrogen, Waltham, MA) and the product recovered as a pool. The pool was quality checked and quantified using a combination of Qubit dsDNA HS, Agilent 2100 Bioanalyzer High Sensitivity DNA (Agilent, Santa Clara, CA) and Kapa Biosystems Illumina Library 120 Quantification qPCR (Roche, Indianapolis, IN) assays. The pool was loaded onto a standard MiSeq v2 flow cell (Illumina, San Diego, CA) and sequenced in a 2 x 250 base pair paired-end format using a v2 500-cycle reagent cartridge. Custom sequencing and index primers complementary to the 515F/806R sequences used for PCR were added to appropriate wells of the reagent cartridge as described by Kozich et al. (2013). Base calling was performed by Illumina Real Time Analysis (RTA) version 1.18.54 and output of the RTA was demultiplexed and converted to FastQ file format with Illumina Bcl2fastq version 2.19.1. 16S rRNA gene sequence analysis Phylotype operational taxonomic unit (OTU) analysis of 16S rRNA gene sequences was performed as described Kozich et al. (2013) with slight modification. The sequence and taxonomy alignment references utilized were Silva version 132 (Quast et al., 2013; Kozich et al., 2013). Ambiguous bases and sequences longer than 260 base pairs were removed. Chimeric sequences were identified and removed using VSEARCH. Samples were subsampled to 112,238 sequences, which reflected the minimum number of sequences from the sample population. The workflow used in this experiment is located in APPENDIX A. Mock community composition is located in APPENDIX B (Table 3.9A). Mock community analysis was performed using the One Codex platform (One Codex, San Francisco, CA). Statistical analysis Dietary assessment data were analyzed by a one-way t-test for each sex against current age-specific U.S. Dietary Guidelines for Americans (U.S. Department of Health 121 and Human Services, U.S. Department of Agriculture, 2015). Differences in energy and nutrient intake between sexes were analyzed by unpaired t-tests. Gas volume and pH of 24 h fermentation end products were compared using the Kruskal-Wallis test with Dunn’s test for pairwise comparisons. SCFAs were compared using the Kruskal-Wallis test with Dunn’s test for pairwise comparisons. SCFAs were compared to the original fecal samples using the Kruskal-Wallis test, with the fecal samples as the control comparison. Spearman correlation analysis was used to examine the relationship between pH and gas volume; pH and total SCFAs; and gas volume and total SCFAs of 24 h fermentation end products. Relative abundance data and α-diversity were compared using the Kruskal-Wallis with Dunn’s test for pairwise comparisons or Wilcoxon Signed-Rank tests. Predominant genera were selected by screening those with mean relative abundances greater than 1% across all samples. Only genera with P-values less than 0.05 were included for analysis and medians are reported (R, R Core Team, Vienna, Austria). β-diversity data were analyzed using Analysis of Molecular Variance (AMOVA) with pairwise comparison made using the Bonferroni correction procedure in mothur version 1.41.1 (Kozich et al., 2013). Spearman correlation analysis was used to examine the relationship between relative abundance of significant taxa with SCFAs concentration of 24 h fermentation end products. Unless otherwise noted, all statistical analyses were performed using SAS, version 9.4 (SAS Institute Inc., Cary, NC) and GraphPad Software (La Jolla, CA). In all tests, significance was determined when P < 0.05. Results of parametric tests are presented as LS means/means ± SEM and nonparametric test results are presented as medians with interquartile range (IQR), which appears within text as (median, IQR). 122 RESULTS Dietary assessment of energy and nutrient intake The average ages of female and male participants were 23 y and 35 y, respectively. Average daily fiber intake was 21.7 g and 26.2 g for females and males, respectively. Energy and nutrient intakes (Table 3.3) for the 24 h period preceding fecal sample collection were not different between females and males in any instance (P ≥ 0.05). Female participants consumed 282.7 g of total carbohydrates, while male participants consumed 254.9 g; both were significantly greater (P < 0.05) than gender and age specific recommendations (130 g). 123 TABLE 3.3 Participant energy and nutrient intake of the 24 h period preceding fecal sample collection1,2 Age (y) Food items reported # of supplements Energy (kcal) Protein (g) Protein (% kcal) Carbohydrate (g) Carbohydrate (% kcal) Dietary fiber (g) Added sugars (% kcal) Total fat (% kcal) Saturated fat (% kcal) Linoleic acid (g) α-linolenic acid (g) Calcium (mg) Iron (mg) Magnesium (mg) Phosphorus (mg) Potassium (mg) Sodium (mg) Zinc (mg) Copper (μg) Selenium (μg) Vitamin A (mg) RAE Vitamin E (mg) AT Vitamin D (IU) Vitamin C (mg) Thiamin (mg) Riboflavin (mg) Niacin (mg) Vitamin B-6 (mg) Vitamin B12 (μg) Choline (mg) Vitamin K (μg) Folate (μg) DFE 1Values are mean ± SEM, n = 3 for each sex 2*Denotes difference from current U.S. Dietary Guidelines for Americans (P < 0.05) Male 34.7 ± 3.4 16.7 ± 1.9 0.3 ± 0.3 2598.1 ± 267.3 125.1 ± 30.2 19.0 ± 3.2 254.9 ± 11.8* 40.5 ± 6.0 26.2± 4.4 5.6 ± 1.2 42.4 ± 4.8 10.8 ± 1.2 33.4 ± 11.1 3.0 ± 1.7 1097.3 ± 168.3 36.7 ± 9.9 609.4 ± 149.5 2276.0 ± 373.5 4218.2 ± 609.9 6807.0 ± 2604.0 25.7 ± 12.4 3599.0 ± 1636.6 171.8 ± 61.6 1846.6 ± 790.6 60.5 ± 25.8 624.9 ± 471.3 133.0 ± 51.3 4.2 ± 1.5 4.0 ± 1.4 57.5 ± 24.5 4.3±1.5 10.6 ± 5.5 385.0 ± 134.9 217.4 ± 126.9 1347.7 ± 354.8 Female 22.7 ± 0.3 18.0 ± 5.0 2.0± 1.2 2290.8 ± 192.4 105.6 ± 29.7 17.8 ± 3.6 282.7 ± 15.3* 50.3 ± 6.4 21.7 ± 6.6 18.2 ± 7.0 33.0 ± 2.8 13.1 ± 2.3 14.1 ± 2.4 1.1 ± 0.2 1504.4 ± 221.4 24.2 ± 3.1 415.7 ± 136.3 1742.3 ± 389.6 3233.3 ± 905.7 4332.7 ± 952.5 17.0 ± 5.5 2162.0 ± 885.0 197.0 ± 63.9 1091.9 ± 541.1 22.1 ± 10.2 1402.2 ± 708.6 158.7 ± 61.1 19.2 ± 16.0 19.8 ± 16.5 47.8 ± 12.4 19.6 ± 16.3 356.5 ± 326.7 380.0 ± 134.5 153.1 ± 70.4 845.1 ± 155.5 124 Gas volume after 24 h dietary fiber source fermentation Differences in gas volume after 24 h fermentation were evaluated using syringe barrel displacement (Figure 3.1). Median gas volume differed by fiber source (P < 0.001). XOS (69.5 mL, 11.0 mL) fermentations produced more gas than gum arabic (53.0 mL, 17.0 mL) and cellulose (0 mL, 0 mL) fermentations. The gas volume of cellulose fermentations was also less than guar gum (62.0 mL, 4.5 mL) fermentations. There were no differences in gas volume by sex (P = 0.39) or donor (P = 0.67). Figure 3.1 Gas volume by fiber source after 24 h fermentation1,2 1Data presented as box and whisker plot (5-95%), n = 7 for each fiber source 2Fiber sources without common letters are different (P < 0.05) 125 pH of fermentation end products At the completion of the 24 h fermentation period, pH (Figure 3.2) differed by fiber source (P < 0.001). XOS (5.08, 0.18) fermentations produced the lowest median pH, which differed from arabinoxylan (6.11, 0.18) and cellulose (7.56, 0.27). Cellulose fermentations had the highest pH, which also differed from guar gum (5.38, 0.25). There were no differences in pH by sex (P = 0.92) or donor (P = 0.99). Figure 3.2 pH by fiber source after 24 h fermentation1,2 1Data presented as box and whisker plot (5-95%), n = 7 for each fiber source 2Fiber sources without common letters are different (P < 0.05) 126 SCFA composition of fermentation end products The SCFAs detected in quantifiable concentrations after blank adjustment were acetate, propionate, and butyrate; total SCFAs presented in Table 3.4 is the sum of the three individual SCFAs. Across all fiber sources, median concentration of total SCFAs, acetate, propionate, and butyrate were 110.00 mM, 59.38 mM, 19.00 mM, and 23.41 mM, respectively. Median acetate, propionate, and butyrate molar ratios were 61.86%, 15.60%, and 20.15%. Total SCFAs differed by fiber source (P < 0.001). Total SCFAs of XOS (131.11 mM) and arabinoxylan (117.13 mM) fermentations were greater than cellulose (2.43 mM). Acetate concentration differed by fiber source (P < 0.001), with acetate concentration of cellulose (1.91 mM) fermentations less than all other fiber sources except guar gum (51.02 mM). Propionate concentration differed by fiber source (P < 0.001). Propionate concentration from cellulose (0.35 mM) fermentations was less than arabinoxylan (21.34 mM) and guar gum (29.25 mM) fermentations. Butyrate concentration differed by fiber source (P < 0.001). Butyrate concentration from cellulose (0.48 mM) fermentations was less than guar gum (29.53 mM) and XOS (44.20 mM) fermentations. XOS fermentation also produced more butyrate than gum arabic (14.09 mM). Acetate molar ratio differed by fiber source (P < 0.001). The acetate molar ratios of cellulose (76.61%) and gum arabic (64.97%) fermentations were greater than guar gum (47.69%). Propionate molar ratios differed by fiber source (P < 0.001). The propionate molar ratio of guar gum (26.67%) fermentations was greater than cellulose (10.92%) and XOS (14.26%) fermentations. Butyrate molar ratios differed by fiber source (P < 0.001). The butyrate molar ratio of XOS (33.26%) fermentations was greater than cellulose (12.47%) and gum arabic (14.86%) fermentations. Additionally, the 127 butyrate molar ratio of guar gum (27.57%) fermentations was greater than cellulose (12.47%) fermentations. Concentrations of individual and total SCFAs were not influenced by sex (P ≥ 0.05) or donor (P ≥ 0.05). We compared SCFA composition of the original fecal samples to that in 24 h fermentation end products. Total SCFA concentration differed between feces and fiber source (P < 0.001), with total SCFAs of feces (137.15 mM) greater than that in cellulose (2.43 mM) fermentations. Acetate concentrations differed between feces and fiber source (P<0.001), with fecal acetate concentrations (75.54 mM) greater than that in cellulose (1.91 mM) fermentations. Propionate concentrations differed between feces and fiber source (P < 0.001), with the fecal propionate concentrations (20.27 mM) greater than that in cellulose (0.35 mM) fermentations. Butyrate concentrations differed between feces and fiber source (P < 0.001), with fecal butyrate concentration (29.87 mM) greater than cellulose (0.48 mM) fermentations. Propionate molar ratios differed between feces and fiber source (P < 0.001). The propionate molar ratio of feces (22.31%) was greater than that in cellulose (10.92%) fermentations. 128 129 Associations between gas volume, pH, and SCFAs There was a strong, negative correlation between pH and gas volume (r = -0.811, P < 0.001). Potential hydrogen and total SCFAs were strongly, negatively correlated (r = -0.834, P < 0.001). Gas volume and total SCFAs were moderately, positively correlated (r = 0.688, P < 0.001). 16S rRNA gene sequencing and bacterial diversity To examine the relationship between fermentation fiber source and bacterial community diversity, and to detect changes from original fecal samples, 16S rRNA genes were sequenced using the MiSeq Illumina platform. A total of 18,626,234 raw clusters were obtained from sequencing with 80.0% ≥ Q30 and 81.4% passing filtration. Twenty true positives from the 20 mock community organisms were detected resulting in a true positives score of 100%. The relative abundance score was 61%, and the false positive score 99% due to low-level Cronobacter detection. True positives represented 97.46% of reads, false positives 0.25%, and unclassified reads 2.29% resulting in an overall score of 87%. After sequence curation, genus-level classification resulted in mock community members correctly identified to 18 genera with average relative abundances of 5.55 ± 0.94% and representing 99.94% of sequences. The feces and 24 h fermentation samples contained 7,273,646 total sequences belonging to 295 unique phylotype OTUs from 13 phyla. The mean number of sequences/sample was 173,182 ± 4392. There were 154 phylotype OTUs with 100 or more sequences, and 141 phylotype OTUs with fewer than 100 sequences. After subsampling to 112,238 sequences, representative coverage across all samples was greater than 99.9%, and 99.4 ± 0.17% of sequences corresponded to members of the following phyla: Bacteroidetes, Firmicutes, Proteobacteria, 130 Actinobacteria, and Verrucomicrobia. Phyla with low sequence representation were: Cyanobacteria, Tenericutes, Lentisphaerae, Epsilonbacteraeota, Fusobacteria, Deinococcus-Thermus, Patescibacteria, and Dependentiae. To examine how dietary fiber sources influence α-diversity of human colonic bacteria in vitro, Shannon diversity indices were compared. Median Shannon diversity across all fiber sources was (2.70, 0.65). Shannon diversity was not different among fermentations of fiber sources (P = 0.85) or when fiber sources were compared to fecal samples (P = 0.82). However, Shannon diversity of fermentations were influenced by fecal sample donor (Figure 3.3) (P = 0.001). Donor 3 (3.40, 0.13) fermentations were more diverse than those of donor 1 (2.35, 0.03), donor 2 (2.08, 0.40), and donor 5 (2.35, 0.11). Donors were differentiated by their relative abundances of phyla Bacteroidetes, Firmicutes, Verrucomicrobia. Shannon diversity was not influenced by inoculum donor sex (P = 0.42). With the inclusion of fecal samples in the analysis (Figure 3.4), Shannon diversity again differed by fecal sample donor (P < 0.001). Donor 3 (3.34, 0.14) diversity was greater than donor 1 (2.36, 0.15), donor 2 (2.23, 0.41), and donor 5 (2.36, 0.11). 131 Figure 3.3 Shannon diversity index of 24 h fermentations by donor excluding feces1,2 1Data presented as box and whisker plot (5-95%), n = 5 for each donor 2Donors not sharing a common letter are different (P < 0.05) Figure 3.4 Shannon diversity index of 24 h fermentations by donor including feces1,2 1Data presented as box and whisker plot (5-95%), n = 6 for each donor 2Donors not sharing a common letter are different (P < 0.05) 132 To examine if dietary fibers affect bacterial community β-diversity, Yue and Clayton Θ (Θ-YC) distances were calculated for 24 h fermentations by fiber source. The non-metric multidimensional scaling (NMDS) plot of Θ-YC distances by fiber source (Figure 3.5) indicated bacterial communities cluster separately by fiber source. The centroids of bacterial communities were significantly different among fiber source fermentations (P = 0.046). However, after Bonferroni correction (P ≥ 0.005) no centroids were significantly different. The NMDS plot of Θ-YC distances by feces and fiber source (Figure 3.6) indicated that bacterial communities cluster separately. The centroids of bacterial communities were significantly different by sample (P = 0.033); however, after Bonferroni correction (P ≥ 0.0033) no centroids were significantly different. The NMDS plot of Θ-YC distances of fermentations by donor (Figure 3.7) indicaed bacterial communities clustered separately by donor. The centroids of bacterial communities were significantly influenced by donor (P < 0.001). Donor 1 and donor 3 centroids were different (P = 0.001). The centroids of donor 2 and donor 3 were different (P = 0.001). The centroids of donor 3 and the fecal sample composite (Donor Mix) were different (P = 0.002). The NMDS plot of Θ-YC distances of fermentations and feces by donor (Figure 3.8) indicated that bacterial communities clustered separately by donor. The centroids of bacterial communities were significantly influenced by donor (P < 0.001). Donor 1 and donor 3 centroids were different (P = 0.002). The centroids of donor 1 and donor 4 were different (P = 0.004). The centroids of donor 2 and donor 3 were different (P = 0.001). The centroids of donor 3 and the donor 6 were different (P = 0.001). The 133 centroids of donor 3 and the fecal sample composite (Donor Mix) were different (P = 0.001). Figure 3.5 NMDS plot of Θ-YC distances by fermentation fiber source1,2,3 1Data presented as a 3D plot of bacterial communities by fiber source, n = 7 for each fiber source 2X axis = Axis 1, Y axis = Axis 2, Z axis = Axis 3; axes are for ordination only 3No centroids significantly different after Bonferroni correction (P ≥ 0.005) 134 Figure 3.6 NMDS plot of Θ-YC distances by feces and fermentation fiber source1,2,3 1Data presented as a 3D plot of bacterial communities by feces and fiber source, n = 7 for each fiber source and feces 2X axis = Axis 1, Y axis = Axis 2, Z axis = Axis 3; axes are for ordination only 3No centroids significantly different after Bonferroni correction (P ≥ 0.0033) 135 Figure 3.7 NMDS plot of bacterial communities by donor excluding fecal samples1,2,3 1Data presented as a 3D plot of bacterial communities by fecal sample donor without feces, n = 5 for each donor 2X axis = Axis 1, Y axis = Axis 2, Z axis = Axis 3; axes are for ordination only 3Donor 3 different from donor 1, donor 2, and donor mix (P < 0.002) 136 Figure 3.8 NMDS plot of bacterial communities by donor including fecal samples1,2,3 1Data presented as a 3D plot of bacterial communities by fecal sample donor with feces, n = 6 for each donor 2X axis = Axis 1, Y axis = Axis 2, Z axis = Axis 3; axes are for ordination only 3Donor 3 different from donor 1, donor 2, donor 6, and donor mix; donors 1 and 4 different (P < 0.002) Bacterial composition of 24 h fermentations The predominant phyla in the present study were Bacteroidetes, Firmicutes, Proteobacteria, Actinobacteria, and Verrucomicrobia (Table 3.5), and these phyla accounted for 99.4 ± 0.17% of all sequences. Relative abundances of Bacteroidetes (P = 137 0.32), Firmicutes (P = 0.15), and Verrucomicrobia (P = 0.68) were not influenced by fiber source. However, relative abundances of Proteobacteria (P =0.013) and Actinobacteria were significantly influenced by fiber source (P = 0.020). Proteobacteria relative abundance was greater in cellulose (8.51%) than arabinoxylan (4.40%) fermentations. Actinobacteria relative abundance was lower with cellulose (0.09%) than XOS (1.73%) fermentations. To examine the ability of fiber sources to change bacterial composition, phyla relative abundances were compared between fermentation fiber source and the original fecal samples. Relative abundances of Bacteroidetes (P = 0.17), Firmicutes (P = 0.053), Verrucomicrobia (P = 0.79), and Proteobacteria (not significant after Dunn’s procedure) were not different. The relative abundance of Actinobacteria differed between original fecal material and fermentation fiber sources (P < 0.001), with its relative abundance greater in XOS (1.73%) fermentations compared to feces (0.09%). Table 3.5 Relative abundances (%) of predominant phyla by feces and fiber source1,2,3 Phylum Arabino- Feces 1Data presented as median (IQR), n = 7 for each fiber source and feces 2Fiber sources without a common letter with rows are different (P < 0.05) 3*Denotes difference between feces and fiber source (P < 0.05) 138 Bacteroidetes Firmicutes Proteobacteria Actinobacteria Verrucomicrobia 50.96 (10.56) 44.10 (6.66) 7.75 (4.72) 0.09 (0.06) 0.03 (0.18) Gum arabic 48.06 (5.71) 43.07 (11.01) 7.16ab (5.38) 0.14ab (0.12) 0.02 (0.04) xylan 49.44 (19.30) 45.25 (19.49) 4.40b (3.40) 0.18ab (0.31) 0.01 (0.01) Cellulose Guar gum 46.85 (22.58) 48.35 (20.97) 4.99ab (2.05) 0.23ab (0.33) 0.01 (0.02) 39.06 (26.86) 49.19 (25.81) 8.51a (5.27) 0.09b (0.19) 0.02 (0.19) XOS 32.58 (10.91) 60.66 (11.04) 4.62ab (1.56) 1.73a* (1.09) 0.01 (0.06) To examine the relationship between fiber source and changes to higher taxonomic rank classification, relative abundances of predominant genera (Table 3.6) from 24 h fermentations were compared. The relative abundances of Faecalibacterium (P < 0.001), Agathobacter (P < 0.001), Roseburia (P< 0.001), Ruminococcaceae UCG-002 (P = 0.004), Erysipelotrichaceae UCG-003 (P = 0.001), Fusicatenibacter (P < 0.001), Blautia (P < 0.001), Lachnospiraceae genomosp. (P<0.001), and an uncultured member of Lachnospiraceae (P<0.001) were significantly influenced by fiber source. Faecalibacterium was lowest with arabinoxylan (2.55%) fermentations compared to gum arabic (11.28%) and cellulose (12.49%). Agathobacter was higher with XOS (9.99%) fermentations compared to guar gum (0.44%) and cellulose (0.99%) fermentations. Roseburia was lower with cellulose (0.55%) fermentations compared to all other fiber sources except gum arabic (1.29%) and arabinoxylan (2.58%). Ruminococcaceae UCG-002 was greater with cellulose (2.09%) fermentations than XOS (0.22%). Erysipelotrichaceae UCG-003 was highest with XOS (5.28%) fermentations compared to all fiber sources except guar gum (1.21%). Fusicatenibacter was lowest with gum arabic (0.06%) and cellulose (0.17%) fermentations. Blautia was higher with XOS (3.92%) fermentations than cellulose (0.32%) and guar gum (0.48%) fermentations. Lachnospiraceae genomosp. was greater with XOS (2.89%) fermentations than gum arabic (0.83%), cellulose (0.84%), and guar gum (1.06%) fermentations. An uncultured member of Lachnospiraceae was higher with XOS (2.04%) fermentations than gum arabic (0.72%) and guar gum (0.71%) fermentations. To examine the ability of fiber sources to rapidly change bacterial composition, genera relative abundances of 24 h fermentations were compared with donor fecal 139 samples. The relative abundances of Faecalibacterium (P < 0.001), Agathobacter (P = 0.003), Roseburia (P < 0.001), Ruminococcaceae UCG-002 (P < 0.001), Erysipelotrichaceae UCG-003 (P < 0.001), Fusicatenibacter (P < 0.001), Blautia (P < 0.001), and Lachnospiraceae uncultured (P<0.001) were different between feces and 24 h fermentations. Faecalibacterium was higher with gum arabic (11.28%) and cellulose (7.37%) fermentations compared to feces (3.69%). Agathobacter increased with XOS (9.99%) fermentations when compared to feces (1.55%). Roseburia increased with guar gum (12.37%) fermentations when compared to feces (1.06%). Ruminococcaceae UCG-002 was greater in feces (2.57%) and decreased with XOS (0.22%) fermentations. Erysipelotrichaceae UCG-003 increased with guar gum (1.21%) and XOS (5.28%) fermentations when compared to feces (0.18%). Fusicatenibacter increased with arabinoxylan (1.63%), guar gum (2.03%), and XOS (2.52%) fermentations when compared to feces (0.12%). Blautia was higher after arabinoxylan (1.57%) and XOS (3.92%) fermentations when compared to feces (0.12%). Finally, Lachnospiraceae uncultured was higher with XOS (2.04%) fermentations relative to feces (0.57%). 140 Table 3.6 Relative abundances (%) of predominant genera by feces and fiber source1,2,3 Genus Faecalibacterium Agathobacter Roseburia Ruminococcaceae UCG-002 Erysipelotrichaceae UCG-003 Fusicatenibacter Blautia Feces Gum arabic 11.28ab* (8.10) 2.02abc (3.32) 1.29ab (0.95) 0.57ab (0.79) 0.21b (0.68) 0.06c (0.10) 0.58ab (1.68) 0.83b (0.65) 0.72b (0.49) 3.69 (3.03) 1.55 (1.87) 1.06 (0.62) 2.57 (4.09) 0.18 (0.25) 0.12 (0.19) 0.12 (0.25) 0.57 (0.35) 0.57 (0.35) Arabino- xylan 2.55c (2.81) 5.11ab (3.72) 2.58ab (5.46) 0.93ab (0.83) 0.21b (0.79) 1.63ab* (3.63) 1.57ab* (0.97) 1.39ab (0.89) 1.08ab (0.45) Cellulose Guar gum 7.37abc (7.30) 0.44c (1.41) 12.37a* (8.20) 0.81ab (0.96) 1.21ab* (2.01) 2.03a* (1.35) 0.48b (0.69) 1.06b (0.80) 0.71b (0.30) 12.49a* (6.87) 0.99bc (1.43) 0.55b (0.35) 2.09a (2.12) 0.21b (0.40) 0.17bc (0.14) 0.32b (0.21) 0.84b (0.77) 1.20ab (0.87) XOS 3.91bc (1.64) 9.99a* (10.22) 2.64a (4.76) 0.22b* (0.16) 5.28a* (7.36) 2.52ab* (2.28) 3.92a* (1.34) 2.89a (1.83) 2.04a* (0.99) Lachnospiraceae genomosp. Lachnospiraceae uncultured 1Data presented as median (IQR) 2Fiber sources without a common letter with rows are different (P < 0.05), n = 7 for each fiber source and feces 3*Denotes difference between feces and fiber source (P < 0.05) Correlation analysis of predominant phyla and genera with SCFAs Within (Table 3.7 and Table 3.8) are results of correlation analyses. Actinobacteria relative abundance was moderately, positively associated with total SCFAs (r = 0.406, P = 0.015) and moderately, positively associated with butyrate (r = 0.515, P = 0.002). Firmicutes relative abundance was weakly, positively associated with butyrate (r = 0.346, P = 0.042). Proteobacteria relative abundance was moderately, negatively associated with butyrate (r = -0.508, P = 0.002). All of the bacterial genera examined by correlation analysis were members of Firmicutes. Faecalibacterium was moderately, negatively correlated with total SCFA, 141 acetate, and butyrate. Agathobacter was weakly to moderately, positively associated with total SCFA, acetate, and butyrate. Roseburia was moderately, positively associated with propionate and moderately, positively associated with butyrate. Ruminococcaceae UCG-002 was weakly to moderately, negatively associated with total SCFA, acetate, and butyrate. Erysipelotrichaceae UCG-003 was moderately positively associated with butyrate. Fusicatenibacter was moderately, positively associated with propionate and butyrate. Blautia was moderately, positively associated with total SCFA, acetate, and butyrate. Lachnospiraceae genomosp. was weakly to moderately, positively associated with total SCFA, acetate, and butyrate. Finally, Lachnospiraceae uncultured was weakly, negative associated with propionate. Table 3.7 Correlation analysis between phyla relative abundances (%) and SCFA concentration1,2 Phylum Relative Abundance (%) Bacteroidetes Propionate (mM) Butyrate (mM) Total SCFA (mM) -0.070, P = 0.689 0.099, P = 0.573 -0.311 P = 0.069 0.406, P = 0.015 -0.255, P = 0.139 Acetate (mM) 0.102, P = 0.559 -0.090, P = 0.608 -0.146, P = 0.402 0.281, P = 0.102 -0.291, P = 0.090 0.034, P = 0.846 -0.017, P = 0.924 -0.177, P = 0.309 0.183, P = 0.292 -0.272, P = 0.115 -0.309, P = 0.071 0.346, P = 0.042 -0.508, P = 0.002 0.515, P = 0.002 0.001, P = 0.996 Firmicutes Proteobacteria Actinobacteria Verrucomicrobia 1Data presented as Spearman correlation coefficients, n = 35 2Signfigance determined when (P < 0.05) 142 Agathobacter Roseburia Ruminococcaceae UCG-002 Erysipelotrichaceae UCG-003 Fusicatenibacter Blautia Total SCFA (mM) -0.642, P < 0.001 0.519, P = 0.001 0.292, P = 0.088 -0.537, P = 0.009 0.277, P = 0.107 0.287, P = 0.094 0.709, P < 0.001 0.541, P < 0.001 0.245, P = 0.156 -0.544, P < 0.001 0.422, P = 0.012 0.158, P = 0.365 -0.390, P = 0.021 0.009, P = 0.959 0.167, P = 0.338 0.638, P < 0.001 0.376 P = 0.026 0.142, P = 0.417 Butyrate (mM) -0.553, P < 0.001 0.370, P = 0.029 0.633, P < 0.001 -0.606, P < 0.001 0.624, P < 0.001 0.469, P = 0.005 0.462, P = 0.005 0.588, P < 0.001 0.104, P = 0.552 -0.229, P = 0.186 -0.132, P = 0.450 0.700, P < 0.001 -0.262, P = 0.129 0.183, P = 0.292 0.528, P = 0.001 0.062, P = 0.722 -0.002, P = 0.992 -0.344, P = 0.043 Table 3.8 Correlation analysis between genera relative abundances (%) and SCFA concentation1,2 Genus Relative Abundance (%) Faecalibacterium Propionate (mM) Acetate (mM) Lachnospiraceae genomosp. Lachnospiraceae uncultured 1Data presented as Spearman correlation coefficients, n = 35 2Signfigance determined when (P < 0.05) 143 DISCUSSION The approach taken in this research was to model the effects of fermentation of semi-purified dietary fiber sources on colonic bacterial composition and their fermentation metabolites in a 24 h period. Fecal sample donors represented healthy Americans who recently consumed a nutritionally adequate diet in terms of energy and nutrients. Our results demonstrate that batch fermentation is useful for modeling substrate-specific changes in bacterial composition and resulting bacterial fermentation products. In the present study, fermentation of all of the dietary fiber sources, except cellulose, produced gas and decreased pH of the anaerobic medium. The range of pH values (4.93-7.88) and gas volumes (0 - 72.5 mL) are indicative of bacterial metabolic activity. As expected, total SCFAs were positively correlated with gas volume and negatively correlated with pH. Carbon dioxide and hydrogen are the major gases produced by fermentation of dietary fiber by human colonic bacteria (Cummings et al., 2001). While we anticipated limited SCFA production from cellulose (Cummings, 1984), we expected to observe differences in SCFAs between the other dietary fiber sources due to differences in their structure and carbohydrate monomer composition (Hamaker and Tuncil, 2014). Apart from differences with cellulose, one important observation was that butyrate concentration and its molar ratio was higher with XOS than with gum arabic fermentation. Additionally, acetate molar ratios differed between gum arabic and guar gum fermentation, with the latter being lower. However acetate concentrations did not differ between the two. We chose to stop fermentation at 24 h to limit potential adverse effects of low pH on bacterial metabolism and SCFA production (Edwards et al., 1985), 144 which may have led to incomplete substrate fermentation and limited our ability to detect more differences in SCFA composition between fiber sources. For example, Kaur et al. (2011) observed higher SCFA production with 48 h compared to 24 h in vitro fermentation with both slowly and rapidly fermentable fiber sources. The increased butyrate production of XOS fermentation, or rather decreased production with gum arabic, is notable because butyrate, in addition to being a preferred energy source of colon cells, inhibits inflammation through the nuclear factor- κB and peroxisome proliferator-activated receptor-γ pathways. Additionally, butyrate stimulates gastro-intestinal cell mucus production and decreases gut permeability in vitro, which is thought to aid in gastro-intestinal barrier defense (Canani et al., 2011). These observations suggest that XOS, and not gum arabic, may have a protective effect on barrier function when only considering butyrate. We also hypothesized that fermentation of the dietary fiber sources would produce SCFA profiles different from the fecal samples used for inoculation. Total SCFA, acetate, propionate, and butyrate productions were considerably lower with cellulose fermentation compared to fecal inoculums. However, SCFA profiles resulting from fermentation of the other fiber sources were not different from feces. These data demonstrate that highly fermentable fibers behave similarly in vitro when fermented by human colonic bacteria, and this is reflected in overall median molar ratios of acetate (62), propionate (16), and butyrate (20), which closely approximates average ratios (60:20:20) for these fatty acids in human digesta (den Besten et al., 2013). Next we examined the effect of fermentation of the different dietary fiber sources on bacterial diversity. Dietary fiber sources did not affect within (data not shown) or between sample bacterial diversity (Figures 3.5-3.6). Sasaki et al. (2018) similarly 145 reported bacterial diversity was not influenced by dietary fiber sources. However, we did observe differences in relative abundances of bacterial populations at the phyla level, which Sasaki et al. (2018) and Chung et al. (2016) did not observe in vitro. We included the fiber substrates at a level of 1% (w/v) in the fermentations in this experiment. Conversely, Sasaki used prebiotic concentrations of 0.2% (w/v). Additionally, Chung used a continuous, pH controlled model with a duration of 12 d, and noted bacterial diversity initially decreased and then stabilized after 3 d. These are notable experimental differences that reflect the effect of study design on measured responses, and makes direct comparisons among different studies somewhat challenging. Nonetheless, our results support the “discrete structure” hypothesis of Hamaker and Tuncil (2014) whereby unique carbohydrate structures favor the proliferation of certain bacterial taxa. In the present study, Proteobacteria and Actinobacteria were differentially affected by fiber source. However, Bacteroidetes, Firmicutes, and Verrucomicrobia did not differ by fiber source but were influenced by fecal sample donor. Proteobacteria increased and Actinobacteria decreased during cellulose fermentation, while the opposite was observed for XOS when comparing fiber sources. However, only XOS increased Actinobacteria relative abundance when compared to fecal samples. In healthy humans, Bacteroidetes, Firmicutes, Actinobacteria and Proteobacteria account for 93.5% of bacterial sequences (Thursby and Juge, 2017) while members of Verrucomicrobia account for 3% of sequences (Geerlings et al., 2018). These phyla represented 99.4% of sequences in the present study. Recently, the Human Microbiome Project Consortium (2012) noted that bacterial metabolic pathways are stable even with changes to community structure. Even more so, there exists high inter-individual bacterial community variation, but an individual’s bacterial community is stable over 146 time. This is consistent with the present study, in which we observed differences in bacterial diversity between fecal sample donors (Figure 3.7), but not between fermentation of fiber sources (Figure 3.5). Therefore, it is not surprising that relative abundances of the most abundant human phyla, Bacteroidetes and Firmicutes, were not influenced after exposure to fermentable fiber sources. Members of Bacteroidetes have on average 137 glycoside hydrolases and polysaccharide lyases, whereas, Firmicutes members have 40 such enzymes (Thursby and Juge, 2017). Variation in carbohydrate enzymes expression enables members of these phyla to either be generalists or specialists in carbohydrate polymer degradation, but perhaps the overall genetic redundancy in metabolic enzyme expression of these phyla explains why relative abundance differences were not observed by dietary fiber source (Bradley and Pollard, 2017). Additionally, phylum Verrucomicrobia was unaffected by dietary fiber source in the present study. Akkermansia is a known mucus forager found in human digesta from phylum Verrucomicrobia (Geerlings et al., 2018). Since our model did not contain mucus other than that potentially originating from fecal inoculums, this finding is not surprising. The difference in Proteobacteria relative abundance between cellulose (8.51%) and arabinoxylan (4.40%) fermentations is notable. Members of Proteobacteria are capable of metabolizing a wide range of carbohydrates, lipids, and proteins, which is thought to contribute to the functional metabolic variation of the gut microbiota more so than Firmicutes and Bacteroidetes (Bradley and Pollard, 2017; Moon et al., 2018;). Protein fermentation metabolites are ammonia, biogenic amines, branched-chain fatty acids, phenols, and sulfur containing metabolites (Williams et al., 2017). Our anaerobic fermentation medium contained yeast extract and trypticase as nitrogen sources, and 147 these substrates in the absence of a fermentable fiber substrate may be responsible for the observed difference in Proteobacteria relative abundance. It is important to note that excessive accumulation of protein fermentation metabolites has been linked to gastro-intestinal disease such as colorectal cancer and ulcerative colitis (Williams et al., 2017). Proteobacteria is also the largest bacterial phylum and contains a number of well- known enteric human pathogens such as Escherichia coli, Salmonella, and Campylobacter (Moon et al., 2018). Although below our threshold for detecting statistical differences, our samples contained the following Enterobacteriaceae members: Escherichia-Shigella, Citrobacter, Klebsiella, Enterobacter, Serratia, Raoultella, Kluyvera, Hafnia-Obesumbacterium, Erwinia, Dickeya, and Cronobacter. Higher Proteobacteria relative abundance is linked to gastro-intestinal inflammation, Crohn’s disease, and inflammatory bowel diseases (Rizzatti et al., 2017). One hypothesized mechanism for Proteobacteria “blooming” is that inflamed gastro- intestinal tissue has reduced capacity for β-oxidation, which increases gut lumen oxygen thus providing an environmental niche enabling proliferation of facultative anaerobic members of Enterobacteriaceae and decreasing obligate anaerobe abundance. These phyla-level shifts are associated with gastro-intestinal dysbiosis (Winter and Bäumler, 2014; Rizzatti et al., 2017). We also observed differences in Actinobacteria relative abundance. Specifically, Actinobacteria was greater with XOS (1.73%) than cellulose (0.09%) fermentations. Predominant Actinobacteria members were Bifidobacterium and Collinsella; however, their relative abundances were below our threshold for statistical testing. Moniz et al. (2015) observed an increase in Bifidobacterium with in vitro XOS fermentation. We also observed and increase in Actinobacteria relative abundance with XOS fermentation 148 (1.73%) relative to fecal inoculums (0.09%). Notably, Actinobacteria relative abundance was positively associated with butyrate concentration. These results indicate that Actinobacteria members may preferentially metabolize XOS, and fermentation of other fiber sources is associated with higher butyrate production. Bifidobacteria are especially well suited for further investigation due to their preferential metabolism of oligosaccharides, including XOS (Rivière et al., 2016). Next we examined changes in genus-level bacterial abundances and observed important differences in nine genera, all belonging to phylum Firmicutes. We hypothesized that fermentation of the selected dietary fiber sources would differentially affect bacterial composition, and bacterial composition resulting from fiber fermentation would differ from the fecal inoculums. Relative abundances of Faecalibacterium, Agathobacter, Roseburia, Ruminococcaceae UCG-002, Erysipelotrichaceae UCG-003, Fusicatenibacter, Blautia, and an uncultured member of Lachnospiraceae were influenced by fiber source fermentation when compared to donor feces, suggesting these genera are involved in substrate fermentation and their growth is selectively stimulated (or suppressed in the case of Ruminococcaceae UCG-002) by fermentation of specific fiber sources. These genera, in addition to Lachnospiraceae genomosp. were also influenced by fiber source further highlighting their selective stimulation by dietary fiber source or fermentation metabolites. The reasons for observing differences in genera relative abundance by fiber source are difficult to untangle in mixed culture and we recognize there is likely significant cross feeding among bacteria in our model. Therefore, the differences observed cannot be independently explained by selective stimulation associated with fermentation of dietary fiber substrates. Furthermore, species vary in their metabolic 149 capacity (Reichardt et al., 2011), and attempting to infer function from genus level classification is highly speculative. Additionally, Hibbing et al. (2010) noted numerous pure and mixed culture microbe studies have revealed many diverse strategies microbes that employ to impair or kill competitors, which adds another layer of complexity to understanding microbial cross-talk. Therefore, this study presents observations on how selected fiber sources affect bacterial genera composition. However, we chose to analyze differences between fiber sources and feces, which focuses our analysis on the response of bacteria to fiber sources from one vantage point. We observed a 3-fold increase in Faecalibacterium relative abundance with fermentation of gum arabic and cellulose. The only known species is Faecalibacterium prausnitzii, (Duncan et al., 2002), which is associated with decreased inflammation and gastro-intestinal health (Laursen et al., 2017; Louis and Flint, 2017). For example, F. prausnitzii has anti-inflammatory effects in in vitro and in vivo colitis models, and low abundance is associated with Crohn’s disease (Sokol et al., 2008). Agathobacter relative abundance increased approximately 6-fold with XOS. The only known species are, Agathobacter rectalis, which was previously identified as Eubacterium rectale, and A. ruminis (Rosero et al., 2016). Agathobacter typically is highly abundant in the human colon, and has long been associated with colon health due to its butyrate production (Flint et al., 2012). Our study also revealed a nearly 12-fold increase in Roseburia during guar gum fermentation. Roseburia intestinalis is a primary degrader of β-mannans in mixed culture and mice (La Rosa et al., 2019). Roseburia colonization of mice has been linked to decreased gastro-intestinal permeability through the upregulation of tight junction genes claudin-3 and claudin-4 (Kasahara et al., 2018). 150 On the other hand, we observed a decrease in Ruminococcaceae UCG-002 during XOS fermentation. Ruminococcaceae UCG-002 is an uncultured member of the family Ruminococcaceae. Ruminococcaceae produce butyrate and all are obligate anaerobes (Rainey 2009), but little else is known about this uncultured organism. Based on metagenome prediction it is thought to produce butyrate from acetate, lactate, and succinate (Esquivel-Elizondo et al., 2017). We found that Erysipelotrichaceae UCG-003 increased 7 to 79-fold during fermentation of guar gum and XOS. Erysipelotrichaceae UCG-003 is an uncultured member of the family Erysipelotrichaceae, with some members aerobes and others facultative anaerobes (Verbarg et al., 2014). Higher abundances are found in patients with colorectal cancer and inflammatory bowel diseases, and it is considered a robust microbial marker of ulcerative colitis (Wingfield et al., 2018). Meta-analysis has also linked an unclassified member of Erysipelotrichaceae to gastro-intestinal inflammation (Mancabelli et al., 2017). Fusicatenibacter relative abundance increased 14 to 21-fold during fermentation of arabinoxylan, guar gum, and XOS. Fusicatenibacter is a member of the family Lachnospiraceae, whose members are obligate anaerobes and are commonly found in human feces. The only known species is Fusicatenibacter saccharivorans (Takada et al., 2013). Decreased F. saccharivorans relative abundance has been observed in patients with ulcerative colitis (Rapozo, Bernardazzi, and de Souza, 2017). Blautia relative abundance increased 13 to 33-fold during arabinoxylan and XOS fermentations. Blautia members are from the family Lachnospiraceae. There are 13 anaerobe species, and it is considered a major taxonomic group in humans and herbivores due to their ability to degrade carbohydrates (Liu et al., 2008; Eren et al., 151 2015; Durand et al., 2017). Blautia relative abundance has been positively correlated associated with Crohn’s disease (Juste et al., 2013). Finally, an uncultured member of Lachnospiraceae increased 4-fold during XOS fermentation. A recent meta-analysis identified an unclassified member of Lachnospiraceae to be associated with gastro-intestinal health (Mancabelli et al., 2017). Esquivel-Elizondo et al. (2017) predicted Lachnospiraceae to be involved in acetate production and/or production of butyrate through acetate conversion based on metagenome analysis. Collectively, these results confirm that bacterial composition is rapidly changed through dietary fiber source intervention, but overall structure, measured by within and between sample diversity, is stable. At the phylum level, Actinobacteria and Proteobacteria relative abundances were differentially affected by fiber source. However, these phyla represent only a small faction of the genera detected by sequencing and genera relative abundances were below our threshold for statistical testing. On the other hand, nine genera from phylum Firmicutes were rapidly changed during fermentation by different fiber sources, but fiber sources did not significantly influence the overall relative abundance of phylum Firmicutes. This suggests specific genera are influenced by fiber source. This study serves as a foundation for future work, and one of its strengths is the inclusion of dietary assessment data. While we recognize one 24 h assessment period does not provide insight into habitual energy and nutrient intake of research participants, it does provide a snapshot, which has rarely been reported in similar studies. On the other hand, there are limitations to this study. One limitation to our in vitro model is the absence of immune responses, which are dynamically responsive to 152 the gut microbiota and SCFAs. Another limitation is the potential wide variation in dietary habits of the participants. While we excluded participants who followed strict diets excluding major food groups or those modifying their diets to lose weight, a more practical approach may be to select participants who do adhere to specific dietary patterns in order to understand how dietary fiber sources might affect the microbiota of humans with a specific dietary pattern. How the fiber sources used in the present study affect human health remains to be more fully resolved. The SCFAs profiles observed after fermentation of these substrates were more similar than different, with the notable exception of cellulose. We believe that potential gastro-intestinal health benefits associated with consumption of these dietary fiber sources will not be dependent only on SCFAs, but also likely arise as a consequence of complex immune responses between microbes and host. We observed nine genera from Firmicutes to be differentially affected by fiber source in this study. Based on our results and existing evidence, a potential strategy to improve gastro- intestinal health is to increase relative abundances of Faecalibacterium, Agathobacter, Roseburia, Fusicatenibacter, and certain members of Lachnospiraceae; and decrease relative abundances of Erysipelotrichaceae UCG-003 and Blautia. XOS has been associated with increases in bifidobacteria (Moniz et al., 2015). In this study, we also observed an increase in Actinobacteria relative abundance, which is the phylum to which bifidobacteria belong. However, XOS also increased relative abundances of Erysipelotrichaceae UCG-003 and Blautia. These genera have been associated with diminished gastro-intestinal health, and therefore, more consideration should be given to XOS-associated bacterial compositional changes before prematurely classifying it as a prebiotic (Aachary and Prapulla, 2011; Lin et al., 2016). However, we 153 do identify gum arabic and guar gun as potentially important modulators of beneficial genera like Faecalibacterium and Roseburia. In the food industry, gum arabic is used in beverages to enhance mouth feel and in baked goods to improve texture and stability (Baray, 2009). Guar gum forms highly viscous gels in cold water and has numerous processed food applications (Mudgil et al., 2014). These fiber sources are not typically consumed from whole foods, thus their human health and potential economic value may be limited to use as dietary fiber supplements. Studies involving human diet supplementation are a logical next step for understanding potential benefits of these fiber sources on the gut microbiota, and both gum arabic and guar gum are Generally Recognized as Safe (GRAS) food substances by the U.S. Food and Drug Administration (Select Committee on GRAS Substances, 1973A; Select Committee on GRAS Substances, 1973B). In conclusion, this study supports the “discrete structure” hypothesis of Hamaker and Tuncil (2014) whereby unique carbohydrate structures favor the proliferation of certain bacterial taxa. 154 APPENDICES 155 APPENDIX A: GAS CHROMATOGRPAHY CALIBRATION STANDARDS Table 3.9A Gas chromatography calibration curve standards Component Acetic acid Propionic acid Isobutyric acid Butyric acid Isovaleric acid Valeric acid Hexanoic acid Heptanoic acid Standard 1 (mM) 40.0 20.0 5.0 1.0 1.0 1.0 0.5 0.5 Standard 2 (mM) 8.0 4.0 1.0 0.2 0.2 0.2 0.1 0.1 Standard 3 (mM) 1.60 0.80 0.20 0.04 0.04 0.04 0.02 0.02 Standard 4 (mM) 0.320 0.160 0.04 0.008 0.008 0.008 0.004 0.004 156 APPENDIX B: MOTHUR MISEQ WORKFLOW pcr.seqs(fasta=silva.nr_v132.align, start=11894, end=25319, keepdots=F, processors=8) system(mv silva.nr_v132.pcr.align silva.v132.fasta) summary.seqs(fasta=silva.v132.fasta) make.contigs(file=current) summary.seqs(fasta=current) screen.seqs(fasta=current, group=current, maxambig=0, maxlength=260) unique.seqs(fasta=current) count.seqs(name=current, group=current) summary.seqs(count=current) align.seqs(fasta=current, reference=silva.v132.fasta) summary.seqs(fasta=current, count=current) screen.seqs(fasta=current, count=current, summary=current, start=1968, end=11550, maxhomop=8) summary.seqs(fasta=current, count=current) filter.seqs(fasta=current, vertical=T, trump=.) unique.seqs(fasta=current, count=current) pre.cluster(fasta=current, count=current, diffs=2) chimera.vsearch(fasta=current, count=current, dereplicate=t) remove.seqs(fasta=current, accnos=current) classify.seqs(fasta=current, template=silva.v132.fasta, taxonomy=silva.nr_v132.tax, method=knn, numwanted=1) remove.lineage(fasta=current, count=current, taxonomy=current, taxon=Chloroplast- Mitochondria-unknown-Archaea-Eukaryota) remove.groups(count=current, fasta=current, taxonomy=current, groups=variable) 157 phylotype(taxonomy=current) make.shared(list=current, count=current, label=1) classify.otu(list=current, count=current, taxonomy=current, cutoff=80, label=1) rename.file(taxonomy=current, shared=current) count.groups(shared=current) sub.sample(shared=current size=112238) rarefaction.single(shared=current, calc=sobs, freq=100) summary.single(shared=current, calc=nseqs-coverage-sobs-invsimpson-shannon, subsample=112238) dist.shared(shared=current, calc=thetayc-jclass, subsample=112238) pcoa(phylip=current) nmds(phylip=current, mindim=3, maxdim=3) amova(phylip=current, design=variable) 158 APPENDIX C: MOCK COMMUNITY COMPOSITION ATCC Code ATCC 17978 ATCC 17982 ATCC 10987 ATCC 8482 ATCC 15703 ATCC 35702 ATCC 11828 ATCC BAA-816 ATCC 47077 ATCC 700926 ATCC 700392 ATCC 33323 ATCC BAA-335 ATCC 33277 ATCC 9027 ATCC 17029 ATCC BAA-1556 ATCC 12228 ATCC BAA-611 ATCC 700610 Table 3.10A Mock community composition1 Organism Acinetobacter baumannii Actinomyces odontolyticus Bacillus cereus Bacteroides vulgatus Bifidobacterium adolescentis Clostridium beijerinckii Cutibacterium acnes Deinococcus radiodurans Enterococcus faecalis Escherichia coli Helicobacter pylori Lactobacillus gasseri Neisseria meningitidis Porphyromonas gingivalis Pseudomonas aeruginosa Rhodobacter sphaeroides Staphylococcus aureus Staphylococcus epidermidis Streptococcus agalactiae Streptococcus mutans 1All organisms have equal relative abundance (5%) 159 REFERENCES 160 REFERENCES Aachary AA, Prapulla SG. Xylooligosaccharides (XOS) as an Emerging Prebiotic: Microbial Synthesis, Utilization, Structural Characterization, Bioactive Properties, and Applications. Compr Rev Food Sci Food Saf. 2011;10:2–16. Abrams GD, Bishop J. Effect of the Normal Microbial Flora on Gastrointestinal Motility. Proc Soc Exp Biol Med. 1967;301–4. Anderson JW, Baird P, Davis RH, Ferreri S, Knudtson M, Koraym A, Waters V, Williams CL. Health benefits of dietary fiber. Nutr Rev. 2009;67:188–205. Baray S. Acacia Gum. In: Cho S, Samuel P, editors. Fiber Ingredients: Food Applications and Health Benefits. Boca Raton: CRC Press; 2009. p. 122. Bourquin LD, Titgemeyer EC, Garleb K, Fahey GC. Short-chain fatty acid production and fiber degradation by human colonic bacteria: effects of substrate and cell wall fractionation procedures. J Nutr. 1992;122:1508–20. Bradley PH, Pollard KS. Proteobacteria explain significant functional variability in the human gut microbiome. Microbiome. 2017;1–23. Bryant M, Burkey L. Cultural Methods and Some Characteristics of Some of the More Numerous Groups of Bacteria in the Bovine Rumen. J Dairy Sci. 1953;36:205–217. Canani RB, Costanzo M Di, Leone L, Pedata M, Meli R, Calignano A, Canani RB, Costanzo M Di. Potential beneficial effects of butyrate in intestinal and extraintestinal diseases. World J Gastroenterol. 2011;17:1519–28. Caporaso JG, Lauber CL, Walters WA, Berg-lyons D, Lozupone CA, Turnbaugh PJ, Fierer N, Knight R. Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proc Natl Acad Sci. 2010;108:4516–22. Chen T, Kim CY, Kaur A, Lamothe L, Shaikh M, Keshavarzian A, Hamaker BR. Dietary fibre-based SCFA mixtures promote both protection and repair of intestinal epithelial barrier function in a Caco-2 cell model. Food Funct. Royal Society of Chemistry; 2017;8:1166–73. Chung WSF, Walker AW, Louis P, Parkhill J, Vermeiren J, Bosscher D, Duncan SH, Flint HJ. Modulation of the human gut microbiota by dietary fibres occurs at the species level. BMC Biology. 2016;1–13. Cummings JH. Cellulose and the human gut. Gut. 1984;25:805–10. 161 Cummings JH, Macfarlane GT, Englyst HN. Prebiotic digestion and fermentation. Am J Clin Nutr. 2001;73(suppl):415S–420S. De Filippo C, Cavalieri D, Di Paola M, Ramazzotti M, Poullet JB, Massart S, Collini S, Pieraccini G, Lionetti P. Impact of diet in shaping gut microbiota revealed by a comparative study in children from Europe and rural Africa. Proc Natl Acad Sci. 2010;107:14691–6. den Besten G, Eunen K Van, Groen AK, Venema K, Reijngoud D, Bakker BM. The role of short-chain fatty acids in the interplay between diet, gut microbiota, and host energy metabolism. 2013;54:2325–40. DeVries JW. The definition of dietary fibre. Cereal Foods World. 2001;46:112–29. Duncan SH, Hold GL, Harmsen HJM, Stewart CS, Flint HJ. Growth requirements and fermentation products of Fusobacterium prausnitzii, and a proposal to reclassify it as Faecalibacterium prausnitzii gen. nov., comb. nov. Int J Syst Evol Microbiol. 2002;52:2141–6. Durand GA, Pham T, Ndongo S, Ibrahima S, Armstrong N, Fournier P-E, Raoult D, Million M. Anaerobe Blautia massiliensis sp. nov., isolated from a fresh human fecal sample and emended description of the genus Blautia. Anaerobe. 2017;43:47–55. Edwards CA, Duerden BI, Read NW. The Effects of pH on Colonic Bacteria Grown in Continuous Culture. J Med Microbiol. 1985;19:169–80. Eren AM, Sogin ML, Morrison HG, Vineis JH, Fisher JC, Newton RJ, Mclellan SL. A single genus in the gut microbiome reflects host preference and specificity. ISME J. Nature Publishing Group; 2015;9:90–100. Esquivel-Elizondo S, Ilhan ZE, Garcia-Peña EI, Krajmalnik-Brown R. Insights into Butyrate Production in a Controlled Fermentation System via Gene Predictions. mSystems. 2017;2:1–13. Flint HJ, Scott KP, Duncan SH, Louis P, Forano E. Microbial degradation of complex carbohydrates in the gut. Gut Microbes. 2012;3:289–306. Geerlings SY, Kostopoulos I, de Vos WM, Belzer C. Akkermansia muciniphila in the Human Gastrointestinal Tract: When, Where, and How? Microorganisms. 2018;6:1–26. Hamaker B, Tuncil Y. A Perspective on the Complexity of Dietary Fiber Structures and Their Potential Effect on the Gut Microbiota. J Mol Biol. 2014;426:3838–50. Hibbing ME, Fuqua C, Parsek MR, Peterson SB. Bacterial competition: surviving and thriving in the microbial jungle. Nat Rev Microbiol 2010. 2010;8:15–25. Hoy MK, Goldman JD. Dietary Fiber Intake of the U.S. Population. 2014. 162 Human Microbiome Project Consortium. Structure, function and diversity of the healthy human microbiome. Nature. Nature Publishing Group; 2012;486:207–14. Juste C, Kreil DP, Beauvallet C, Guillot A, Vaca S, Carapito C, Mondot S, Sykacek P, Sokol H, Blon F, et al. Bacterial protein signals are associated with Crohn’s disease. Gut. 2014;63:1566–77. Kasahara K, Krautkramer KA, Org E, Romano KA, Kerby RL, Vivas EI, Mehrabian M, Denu JM, Backhead F, Lusis AJ, Rey F. Interactions between Rosburia intestinalis and diet modulate atherogenesis in a murine model. Nat Microbiol. Springer US; 2018;3:1461–71. Kaur A, Rose DJ, Rumpagaporn P, Patterson JA, Hamaker BR. In Vitro Batch Fecal Fermentation Comparison of Gas and Short‐Chain Fatty Acid Production Using “Slowly Fermentable” Dietary Fibers. J Food Sci. 2011;76:H137-42. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a Dual- Index Sequencing Strategy and Curation Pipeline for Analyzing Amplicon Sequence Data on the MiSeq. 2013;79:5112–20. La Rosa SL, Leth ML, Michalak L, Hansen ME, Pudlo NA, Glowacki R, Pereira G, Workman CT, Arntzen MØ, Pope PB, et al. The human gut Firmicute Roseburia intestinalis is a primary degrader of dietary β-mannans. Nat Commun. Springer US;10:1–14. Laursen MF, Laursen P, Larnkjær A, Mølgaard C, Michaelsen KF, Frøkiær H, Bahl I. Faecalibacterium Gut Colonization Is Accelerated by Presence of Older Siblings. mSphere. 2017;2:1–6. Lin S-H, Chou L-M, Chien Y-W, Chang J-S, Lin C-I. Prebiotic Effects of Xylooligosaccharides on the Improvement of Microbiota Balance in Human Subjects. Gastroenterol Res Pract. Hindawi Publishing Corporation; 2016;1–6. Liu C, Finegold SM, Song Y, Lawson PA. Reclassification of Clostridium coccoides, Ruminococcus hansenii, Ruminococcus hydrogenotrophicus, Ruminococcus luti, Ruminococcus productus and Ruminococcus schinkii as Blautia coccoides gen. nov., comb. nov., Blautia hansenii comb. nov., Blautia hydroge. Int J Syst Evol Microbiol. 2008;58:1896–902. Louis P, Flint HJ. Formation of propionate and butyrate by the human colonic microbiota. Environmental Microbiol. 2017;19:29–41. Makki K, Deehan EC, Walter J, Backhed F. The Impact of Dietary Fiber on Gut Microbiota in Host Health and Disease. Cell Host Microbe. 2018;23:705–15. 163 Mancabelli L, Milani C, Lugli GA, Turroni F, Cocconi D, Sinderen D Van, Ventura M. Identification of universal gut microbial biomarkers of common human intestinal diseases by meta-analysis. FEMS Microbiol Ecol. 2017;1–10. Martens EC, Kelly AG, Tauzin AS, Brumer H. The Devil Lies in the Details: How Variations in Polysaccharide Fine-Structure Impact the Physiology and Evolution of Gut Microbes. J Mol Biol. Elsevier Ltd; 2014;426:3851–65. Moniz P, Ling A, Duarte LC, Kolida S, Rastall RA, Pereira H, Carvalheiro F. Assessment of the bifidogenic effect of substituted xylo-oligosaccharides obtained from corn straw. Carbohydr Polym. Elsevier Ltd.; 2016;136:466–73. Moon CD, Cookson AL, Young W, Maclean PH, Bermingham EN. Metagenomic insights into the roles of Proteobacteria in the gastrointestinal microbiomes of healthy dogs and cats. MicrobiologyOpen. 2018;7:1–20. Mudgil D, Barak S, Khatkar BS. Guar gum: Processing, properties and food applications - A Review. J Food Sci Technol. 2014;51:409–18. Müller V. Bacterial Fermentation. Encycl Life Sci. 2001;1–7. Noack J, Timm D, Hospattankar A, Slavin J. Fermentation Profiles of Wheat Dextrin, Inulin and Partially Hydrolyzed Guar Gum Using an in Vitro Digestion Pretreatment and in Vitro Batch Fermentation System Model. Nutrients. 2013;5:1500–10. Pham VT, Seifert N, Richard N, Raederstorff D, Steinert R, Prudence K, Mohajeri MH. The effects of fermentation products of prebiotic fibres on gut barrier and immune functions in vitro. PeerJ. 2018;e5288. Quagliani D, Felt-Gunderson P. Closing America’s Fiber Intake Gap: Communication Strategies From a Food and Fiber Summit. Am J Lifestyle Med. 2017;80–5. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Glo FO, Yarza P. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. 2013;41:590–6. Rainey F. Family VIII. Ruminococcaceae fam. nov. In: Paul De Vos, George M. Garrity DJ, Noel R. Krieg, Wolfgang Ludwig FAR, Whitman K-HS and WB, editors. Bergey’s Manual of Systematic Bacteriology. 2nd ed. Dordrecht Heidelberg London New York: Springer; 2009. p. 1016–1043. Rapozo DCM, Bernardazzi C, de Souza H. Diet and microbiota in inflammatory bowel disease: The gut in disharmony. World J Gastroenterol. 2017;23:2124–40. Reichardt N, Barclay AR, Weaver LT, Morrison DJ. Use of Stable Isotopes To Measure the Metabolic Activity of the Human Intestinal Microbiota. Appl Environ Microbiol. 2011;77:8009–14. 164 Rivière A, Selak M, Lantin D, Leroy F, De Vuyst L. Bifidobacteria and Butyrate- Producing Colon Bacteria: Importance and Strategies for Their Stimulation in the Human Gut. Frontiers Microbiol. 2016;7. Rizzatti G, Lopetuso LR, Gibiino G, Binda C, Gasbarrini A. Proteobacteria: A Common Factor in Human Diseases. Biomed Res Int. Hindawi; 2017;1–7. Rose D, Patterson J, Hamaker B. Structural Differences among Alkali-Soluble Arabinoxylans from Maize (Zea mays), Rice (Oryza sativa), and Wheat (Triticum aestivum) Brans Influence Human Fecal Fermentation Profiles. J Agric Food Chem. 2010;58:493–9. Rosero JA, Killer J, Sechovcova H, Mrazek J, Benada O, Fliegerova K, Havlik J, Kopecny J. Reclassification of Eubacterium rectale (Hauduroy et al. 1937) Prevot 1938 in a new genus Agathobacter gen. nov., and description of Agathobacter ruminis sp. nov., isolated from the rumen contents of sheep and cows. Int J Syst Evol Microbiol. 2016;768–73. Rothschild D, Weissbrod O, Barkan E, Kurilshikov A, Korem T, Zeevi D, Costea PI, Godneva A, Kalka IN, Bar N, Shilo S, Lador D, Vila AV, Zmora N, Pevsner-Fisher M, Israeli D, Kosower, N, Malka G, Wolf BC, Avnit-Sagi T, Lotan-Pompan, M, Weinberger A, Halpern Z, Carmi S, Fu J, Wijmenga C, Zhernakova A, Elinav E, Segal E. Environment dominates over host genetics in shaping human gut microbiota. Nature. Nature Publishing Group; 2018;555:210–5. Sasaki D, Sasaki K, Ikuta N, Yasuda T, Fukuda I. Low amounts of dietary fibre increase in vitro production of short-chain fatty acids without changing human colonic microbiota structure. Sci Rep. Springer US; 2018;1–7. Select Committee on GRAS Substances. Select Committee on GRAS Substances (SCOGS) Opinion: Gum Arabic [Internet]. 21 CFR Section: 184.1330 1973A. Available from: http://wayback.archive- it.org/7993/20171031064431/https://www.fda.gov/Food/IngredientsPackagingLabelin g/GRAS/SCOGS/ucm260422.htm Select Committee on GRAS Substances. Select Committee on GRAS Substances (SCOGS) Opinion: Guar Gum [Internet]. 21 CFR Section: 184.1339 1973B. Available from: http://wayback.archive- it.org/7993/20171031063759/https://www.fda.gov/Food/IngredientsPackagingLabelin g/GRAS/SCOGS/ucm260421.htm Sokol H, Pigneur B, Watterlot L, Lakhdari O, Bermudez-Humaran LG, Gratadoux J, Blugeon S, Bridonneau C, Furet J-P, Corthier G, Grangette C, Vasquez N, Pochart P, Trugnan G, Thomas G, Blottiere HM, Dore J, Marteau P, Seksik P, Langella P. Faecalibacterium prausnitzii is an anti-inflammatory commensal bacterium identified by gut microbiota analysis of Crohn disease patients. Proc Natl Acad Sci. 2008;105:16731–16736. 165 Takada T, Kurakawa T, Tsuji H, Nomoto K. Fusicatenibacter saccharivorans gen. nov., sp. nov., isolated from human faeces. Int J Syst Evol Microbiol. 2013;3691–6. Thursby E, Juge N. Introduction to the human gut microbiota. Biochem J. 2017;474:1823–36. Threapleton DE, Greenwood DC, Nykjaer C, Woodhead C, Cade JE. Dietary fibre intake and risk of cardiovascular disease: systematic review and meta-analysis. BMJ. 2013;347:1–12. U.S. Department of Health and Human Services, U.S. Department of Agriculture. 2015 – 2020 Dietary Guidelines for Americans. 8th ed. 2015. Verbarg S, Goker M, Scheuner C, Schumann P, Stackebrandt E. The Families Erysipelotrichaceae emend., Coprobacillaceae fam. nov., and Turicibacteraceae fam. nov. In: Rosenberg E, DeLong EF, Lory S, Stackebrandt E, Thompson F, editors. The Prokaryotes. 4th ed. Berlin: Springer-Verlag Berlin Heidelberg; 2014. p. 79–105. Verbeke KA, Boobis AR, Chiodini A, Edwards CA, Franck A, Kleerebezem M, Nauta A, Raes J, Van Tol EAF, Tuohy KM. Towards microbial fermentation metabolites as markers for health benefits of prebiotics. Nutr Res Rev. 2015;28:42–66. Wang M, Wichienchot S, He X, Fu X, Huang Q. In vitro colonic fermentation of dietary fibers: Fermentation rate, short-chain fatty acid production and changes in microbiota. Trends Food Sci Technol. Elsevier; 2019;88:1–9. Williams BA, Grant LJ, Gidley MJ, Mikkelsen D. Gut Fermentation of Dietary Fibres: Physico-Chemistry of Plant Cell Walls and Implications for Health. 2017. Wingfield B, Coleman S, McGinnity T, Bjourson A. Robust Microbial Markers for Non- Invasive Inflammatory Bowel Disease Identification. IEEE/ACM Trans Comput Biol Bioinforma. 2018;1545–5963. Winter SE, Bäumler AJ. Why related bacterial species bloom simultaneously in the gut: Principles underlying the “like will to like” concept. Cell Microbiol. 2014;16:179–84. World Cancer Research Fund, American Institute for Cancer Research. Continuous Update Project Report. Food, Nutrition, Physical Activity, and the Prevention of Colorectal Cancer. 2011. Zhao G, Nyman M, Jönsson JA. Rapid determination of short-chain fatty acids in colonic contents and faeces of humans and rats by acidified water-extraction and direct- injection gas chromatography. Biomed Chromatogr. 2006;20:674–82. 166 CHAPTER 4: EFFECT OF IN VITRO FERMENTATION END PRODUCTS ON EPITHELIAL CELL BARRIER FUNCTION AND APICAL LISTERIA MONOCYTOGENES CHALLENGE ABSTRACT Background: Invasive infection by Listeria monocytogenes is highly fatal in young children, pregnant women, the elderly, and immune compromised individuals. Dietary fiber sources improve gastro-intestinal barrier function through their fermentation end products, namely short-chain fatty acids (SCFA), which may be beneficial for preventing listeriosis. This research expands existing knowledge on the effects of bacterial derived fermentation end products on gastro-intestinal barrier function in an in vitro model of the gastro-intestinal phase of L. monocytogenes cell invasion. Objective: The objectives of this study were: 1) evaluate the effects of fermentation end products on epithelial cell barrier integrity in vitro; 2) determine if fermentation end products affect L. monocytogenes association with HT29-MTX-E12 cells; 3) relate barrier integrity to fermentation end product SCFA composition; and 4) relate bacterial taxa identified during the in vitro production of fermentation end products to barrier integrity to identify potentially beneficial bacteria. Methods: Bacterial derived fermentation end products from gum arabic, arabinoxylan, cellulose, guar gum, and xylo-oligosaccharide (XOS) were analyzed for the potential to affect the barrier integrity of polarized HT29-MTX-E12 cells after 48 h treatment. Barrier integrity was evaluated by transepithelial electrical resistance (TEER) and fluorescein isothiocyanate-dextran 4 (FITC-D4) migration. Human colon cells were challenged with L. monocytogenes to determine if differences in barrier integrity 167 resulting from exposure to differing fermentation end product compositions affect pathogen association and internalization. Data were analyzed using standard parametric and nonparametric statistical tests. Results: All fermentation end products increased cell monolayer resistance and decreased FITC-D4 migration when compared to cells only exposed to DMEM. Relative monolayer resistance of cells treated with gum arabic (252.47 ± 9.63%) and arabinoxylan (260.48 ± 9.63%) fermentation end products were greater than that in cells treated with cellulose (149.38 ± 9.63%), guar gum (222.83 ± 9.63%), or XOS (183.70 ± 9.63%) fermentation end products (P < 0.05). A similar, but inverse trend was observed with FITC-D4 transport (P < 0.05). Cell associated L. monocytogenes counts tended to be influenced (P = 0.068) by fermentation end product fiber source; and internalized L. monocytogenes counts were higher in cells exposed to cellulose compared to guar gum and XOS fermentation end products (P < 0.05). Total SCFAs, acetate, and propionate were weakly to moderately positively associated with monolayer resistance (P < 0.05). Bacteroidetes was positively associated with barrier integrity, and Firmicutes was negatively associated (P < 0.05). Conclusion: Gum arabic, arabinoxylan, cellulose, guar gum, and XOS fermentation end products improved barrier function in this in vitro model. However, exposure to cellulose and XOS derived end products resulted in muted responses. Higher L. monocytogenes internalization occurred with cells exposed to cellulose-derived fermentation end products. Human colonic bacteria poorly ferment cellulose, and 24 h in vitro fermentation of cellulose produces low SCFA concentrations. Low SCFA 168 concentration exposure may provide more favorable conditions for cell invasion by L. monocytogenes. Keywords: dietary fiber, fermentation, short-chain fatty acids, Listeria monocytogenes, HT29-MTX-E12 169 INTRODUCTION The gastro-intestinal epithelial monolayer provides a physical barrier against invasive bacterial pathogens like L. monocytogenes. Although this Gram-positive environmental organism rarely causes serious illness in healthy humans, certain groups like the elderly, pregnant women, children, and those with immune deficiencies are vulnerable to invasive infection. In the U.S., approximately 1,600 cases of invasive listeriosis occur annually, leading to about 260 deaths (Centers for Disease Control and Prevention, 2019). Dietary fiber sources, when fed to animals, have mixed effects on severity of bacterial infection and host survival, and certain dietary fiber sources are protective whereas others enhance infection (Buddington et al., 2002; Ten Bruggencate et al., 2005; Petersen et al. 2009; Ebersbach et al., 2012; Hryckowian et al., 2018). Humans and other mammals lack genes encoding proteins required for digesting dietary fiber sources. Therefore, most dietary fibers pass through the small intestine undigested and are metabolized by micro-organisms in the colon by fermentation, resulting in the production of short-chain fatty acids (SCFA) (DeVries, 2001). Gastro- intestinal bacterial composition and carbohydrate structure influence fermentation end product composition due to the bacterial enzymes required for metabolizing complex structures into their respective monomers and metabolites (Flint et al., 2012; Hamaker and Tuncil, 2014). Acetate, propionate, and butyrate are the most abundantly produced SCFAs by bacterial fermentation in human in vitro models (Bourquin et al., 1992; Bourquin et al., 1993). SCFAs and fermentation intermediates differentially affect epithelial permeability by improving or diminishing barrier integrity as a function of SCFA composition and concentration (Suzuki et al., 2008; Nielsen et al., 2018; Peng et al., 170 2007; Peng et al., 2009; Chen et al., 2017). One hypothesized mechanism is the redistribution of tight junction proteins from the cytoplasm to the cell membrane (Peng et al., 2007; Peng et al., 2009). In addition to cell surface proteins contributing to epithelial barrier function, mucin secretions can also influence the potential for enteric infections by contributing to bacteria expulsion (Cornick et al., 2015). Dietary fiber sources differentially affect mucin production in vivo and in vitro. A hypothesized mechanism for this effect is through the stimulation of mucin gene expression by SCFAs. However, SCFA composition and concentration are factors influencing the response (Barcelo et al., 2000; Burger-van Paassen et al., 2009). L. monocytogenes adheres to cells with internalin proteins, and the internalin-A (Inl-A) protein adheres to the epithelial adherens junction protein E-cadherin of gastro- intestinal cells. E-cadherin is not typically accessible apically. However, tight junction remodeling, such as that occurring during extrusion of senescent cells, exposes E- cadherin (Pentecost et al., 2006). Other disturbances, such as inflammation, reduce cell- to-cell adhesion and impair gastro-intestinal barrier function (Schulzke et al., 2009). Mucus glycoproteins may inhibit L. monocytogenes from interacting with epithelial cells, but have also been hypothesized to promote retention of L. monocytogenes near epithelial cells thereby enabling its adhesion (Liévin-Le Moal et al., 2005; Lindén et al., 2008). In order to understand how bacterial produced metabolites affect gastro- intestinal barrier function, we studied the in vitro effects of fermentation end products from gum arabic, arabinoxylan, cellulose, guar gum, and (XOS) using the well- established L. monocytogenes and HT29-MTX-E12 cell challenge model. The objectives of this research are as follows: 1) evaluate the effects of fermentation end products on 171 epithelial cell barrier integrity in vitro; 2) determine if fermentation end products affect L. monocytogenes association with HT29-MTX-E12 cells; 3) relate barrier integrity to fermentation end product SCFA composition; and 4) relate bacterial taxa identified during the in vitro production of fermentation end products to barrier integrity to identify potentially beneficial bacteria. 172 MATERIALS AND METHODS Cell culture HT29-MTX-E12 human colon cells (passages 66-68, Sigma-Aldrich, St. Louis, MO) were seeded at a density of 5.7-5.8 x 105 cells/insert onto polyester Transwell inserts with a membrane growth area of 1.12 cm2 and 0.4 μM pores (Corning, Corning, NY). Cells were routinely grown on Dulbecco’s Modified Eagle Medium (DMEM) with glucose (25 mM), GlutaMax (4 mM), sodium bicarbonate (44 mM), phenol red (Gibco, ThermoFisher, Waltham, MA), and supplemented with 10% (v/v) heat-inactivated (56 °C for 30 min) fetal bovine serum (FBS) (Sigma-Aldrich, St. Louis, MO), and 1% (v/v) penicillin/streptomycin solution (10,000 units/mL of penicillin and 10,000 μg/mL of streptomycin) (Gibco, Sigma-Aldrich, St. Louis, MO). Cells were maintained at 37 °C in an incubator with an atmosphere of 5% CO2 and 95% air and constant humidity. Culture medium was changed every other day. On the 2oth day of culture, culture medium was replaced with the previously described medium without antibiotics. On the 21st day of culture, sterile-filtered (Nalgene 0.2 μM polyethersulfone filter system, ThermoFisher, Waltham, MA) experimental culture medium containing bacterial fermentation end products was added and cells were incubated for 48 h. Fermentation end products (Table 4.1) were produced from in vitro fermentation of gum arabic, arabinoxylan, cellulose, guar gum, and XOS. An equal volume of each individual fermentation end product was added to DMEM with glucose (25 mM), GlutaMax (4 mM), sodium bicarbonate (44 mM), and 10% FBS. This resulted in the experimental cell culture media containing 5% FBS and a 50% concentration of the fermentation end products. In all medium changes, 0.5 and 1.0 mL were added to the apical and basolateral chambers, respectively. 173 Prior to seeding, cells were grown in DMEM with 10% FBS and passaged with 0.25% Trypsin/EDTA (Gibco, ThermoFisher, Waltham, MA) as needed in T225 flasks (ThermoFisher, Waltham, MA). Cells were counted using a hemocytometer, and live/dead cells distinguished after staining with trypan blue. Total SCFA6 (mM) 114.0 120.5 4.2 106.5 136.6 Acetate6 Propionate6 Butyrate6 Table 4.1 Short-chain fatty composition of fermentation end products from five dietary fibers source fermented for 24 h by human fecal inoculum1,2,3,4,5,6 SCFA Source Gum arabic1 Arabinoxylan2 Cellulose3 Guar gum4 XOS5 1Gum arabic, 85% purity, 2% protein 2Arabinoxylan, 75.29% purity, 6.12% protein 3Cellulose, powdered 4Guar gum, 85% purity, 1% fat, 4% protein 5XOS, 74.22% purity 25.72% maltodextrin, 0.06% ash 6Mean SCFA composition from 24 h fermentation of each fiber source (n = 7) (mM) 18.8 20.7 0.4 29.2 19.2 (mM) 18.4 25.0 0.7 28.5 43.8 (mM) 76.8 74.7 3.1 48.8 73.6 Monolayer resistance of HT29-MTX-E12 cells A total of 74 Transwell inserts were utilized in this experiment: 70 representing duplicate fermentation end products (n = 35), duplicate negative controls (DMEM with 5% FBS), and duplicate positive controls (DMEM with 5% FBS, 2.5 mM EDTA) (Cajnko et al., 2015). Basal medium was changed to antibiotic-free medium 24 h prior to the addition of experimental cell culture medium. After 24 h, 1 mL of DMEM with glucose (25 mM), GlutaMax (4 mM), sodium bicarbonate (44 mM), and 5% FBS was added the basolateral chamber. Then, 0.5 mL of the experimental cell culture medium was added to the apical chamber and cells were incubated for 48 h at 37 °C. After 48 h, cells were 174 normalized to room temperature for 30 min prior to cell monolayer integrity evaluation by trans-epithelial electrical resistance (TEER) using a Millicell ERS meter (EMD Millipore Co., Billerica, MA). Only cells with TEER values > 200 Ωcm2 were used for subsequent resistance readings. TEER readings were performed in duplicate for each well according to manufacturer instructions. FITC-D4 permeability of HT29-MTX-E12 cells After recording TEER values, the experimental cell culture medium in the apical chamber was gently aspirated and replaced with 0.5 mL of DMEM with glucose (25 mM), GlutaMax (4 mM), sodium bicarbonate (44 mM), 5% FBS, and FITC-D4 (0.25 mM) (Sigma-Aldrich, St. Louis, MO). Cells were returned to the incubator and maintained at 37 °C. After 2 h, 100 uL of medium from each basolateral chamber was transferred to a 96-well fluorescence measurement plate. Fluorescence was determined by duplicate scans with a CytoFluor II microplate reader (PerSeptive Biosystems) with an excitation wavelength set at 485 nm and emission wavelength set at 530 nm (Chen et al., 2017). Mucin quantification A total of 74 Transwell inserts were utilized in this experiment: 70 representing duplicate fermentation end products (n = 35), duplicate negative controls (DMEM with 5% FBS), and duplicate positive controls (DMEM with 5% FBS, 15 ng/mL Interleukin-6; Sigma-Aldrich, St. Louis, MO) (Smirnova et al., 2001). The same cell culture protocol described previously was used for this procedure. After 48 h of exposure to the experimental cell culture medium, apical chamber medium was aspirated and transferred to 1.7 mL micro centrifuge tubes. The monolayer was rinsed with 0.5 mL of 175 sterile PBS to remove adhered mucus, aspirated, and recovered in corresponding microcentrifuge tubes (Coconnier et al., 1998). The supernatant was centrifuged at 1000 x g for 15 minutes. The supernatant was then removed and transferred to a new 1.7 mL micro centrifuge tube for storage at -80 °C. Human mucin-5AC was quantified with the human mucin-5AC ELISA kit (Cusabio, Wuhan, China) according to manufacturer instructions. Mucin-5AC was selected due to its high expression in the HT29-MTX-E12 cell line compared to other mucin glycoproteins (Navabi et al., 2013). L. monocytogenes inoculum preparation The L. monocytogenes serovar 1/2a (strain ATCC BAA-679 / EGD-e) (American Type Culture Collection, Rockville, MD) inoculum was prepared from an isolated colony grown on brain heart infusion (BHI) agar. The isolated colony was incubated at 37 °C in 250 mL of BHI broth in a 500 mL Pyrex media storage bottle. After 24 h of growth, the mixture was thoroughly mixed and 35-mL aliquots were frozen at -80°C. To determine viability after freeze-thaw, a sample of the growth culture was serially diluted, then plated in duplicate on BHI agar and determined to contain 1.4 x 107 CFU/mL after 24 h incubation at 37 °C. On the day of the experiment, frozen 35 mL inoculum aliquots were thawed at room temperature and centrifuged at 1200 x g for 15 min to pellet the bacterial mass. Medium was aspirated from each aliquot and replaced with 4 mL of DMEM with 5% FBS to achieve a theoretical concentration of 1.2 x 108 CFU/mL. All aliquots were pooled, and the final pooled inoculum was serially diluted with PBS and plated on BHI to determine L. monocytogenes concentration for each respective experiment. The target MOI was 50:1 (Kühbacher et al., 2013; Drolia et al., 2018). 176 Monolayer resistance of L. monocytogenes challenged cells A total of 74 Transwell inserts were utilized in this experiment: 70 representing duplicate fermentation end products (n = 35), duplicate negative controls (DMEM with 5% FBS), and duplicate positive controls (DMEM with 5% FBS, 2.5 mM EDTA). The identical protocol to unchallenged cells was followed through the initial 48 h incubation period. After 48 h, the experimental cell culture medium was removed from the apical chamber and replaced with 0.5 mL L. monocytogenes inoculum and incubated for 90 min at 37 °C. The inoculum contained 1.9 x 108 CFU/mL L. monocytogenes. After 90 min, TEER was evaluated using the previously described protocol. The HT29-MTX-E12 monolayer density was 1.3 x 106 cells/well. This resulted in a multiplicity of infection of 73:1. FITC-D4 permeability of L. monocytogenes challenged cells After recording TEER values, the experimental cell culture medium from the apical chamber was gently aspirated and replaced with 0.5 mL DMEM (5% FBS) with FITC-D4 (0.25 mM). Fluorescence was measured using the previously described protocol. Mucin quantification in L. monocytogenes challenged cells A total of 74 Transwell inserts were utilized in this experiment: 70 representing duplicate fermentation end products (n = 35), duplicate negative controls (DMEM with 5% FBS), and duplicate positive controls (DMEM with 5% FBS, 15 ng/mL Interleukin-6, Sigma-Aldrich, St. Louis, MO). The same cell culture protocol described previously was used for this procedure. After 48 h, the experimental cell culture medium was removed 177 from the apical chamber and replaced with 0.5 mL L. monocytogenes inoculum and incubated for 90 min at 37 °C. After 90 min, the same protocol for collecting and quantifying human mucin-5AC was followed as previously described. The inoculum contained 1.9 x 108 CFU/mL L. monocytogenes. The HT29-MTX-E12 monolayer density was 1.3 x 106 cells/well. This resulted in a multiplicity of infection of 73:1. Enumeration of cell-associated and internalized L. monocytogenes in cell culture A total of 144 Transwell inserts were utilized in this experiment: 70 x 2 representing duplicate fermentation end products (n = 35) and duplicate negative controls (DMEM with 5% FBS). The identical protocol to unchallenged cells was followed through the initial 48 h incubation period. After 48 h, medium was aspirated from the apical chamber and replaced with 0.5 mL L. monocytogenes inoculum and incubated for 90 min at 37 °C. The inoculum contained 8.7 x 107 CFU/mL L. monocytogenes. The HT29-MTX-E12 monolayer density was 1.3 x 106 cells/well. This resulted in a multiplicity of infection of 33:1. Cell-associated and internalized L. monocytogenes was determined following the protocol described by Coconnier et al. (1998). To determine cell-associated L. monocytogenes, cells were incubated for 90 min at 37 °C. After 90 min, cells were rinsed 3x with 0.5 mL of sterile PBS. Then, 0.5 mL of autoclaved purified distilled water as added and cells were incubated for 30 min to lyse the HT29-MTX-E12 monolayer. Then, 100 μL of cell lysate was serially diluted in PBS, plated in duplicate on modified Oxford (MOX) agar plates (NEOGEN, Lansing, MI), and incubated at 37 °C for 48 h. To determine internalized L. monocytogenes, cells were incubated for 90 min at 37 °C. After 90 min, cells were rinsed twice with 0.5 mL of sterile PBS. Then, 0.5 mL of 178 DMEM with 5% FBS and gentamicin (50 μg/mL, Sigma-Aldrich, St. Louis, MO) was added and cells were incubated for 60 min to kill non-internalized L. monocytogenes. After 60 min, the monolayer was rinsed once with 0.5 mL of sterile PBS. Then, 0.5 mL of autoclaved purified distilled water as added and cells were incubated for 30 min to lyse the HT29-MTX-E12 monolayer. Then, 100 μL of cell lysate was serially diluted in PBS, plated in duplicate on MOX agar, and incubated at 37 °C for 48 h. Plates with counts between 25-250 CFU were considered countable. Plates with fewer than 25 CFU were counted as estimated standard plate counts. Cytotoxicity Any cell cytotoxicity caused by fermentation end products was assessed after the 48 h incubation period using the Pierce Lactate Dehydrogenase Cytotoxicity Assay Kit (ThermoFisher, Waltham, MA) according to manufacturer instructions. Absorbance was measured at 490 nm and 680 nm on an EnSpire multimode plate reader (PerkinElmer, Waltham, MA). Lactate dehydrogenase activity was determined by subtracting the 680 nm absorbance value from the 490 nm absorbance value before calculating % cell death using the formula % cytotoxicity = experimental LDH release/ maximum LDH release (Bucior et al., 2014). Statistical analysis TEER and FITC-D4 data were compared using analysis of variance (ANOVA), and Least Square (LS) means were compared using Fisher’s Least Significant Difference (LSD) method. L. monocytogenes plate counts were natural log transformed and analyzed using a generalized estimating equation method with identity link function; LS means were compared using Fisher’s LSD method. Spearman correlation analysis was 179 used to examine the relationship between SCFA composition of fermentation end products and cell monolayer integrity; relative abundance of significant bacterial taxa (Chapter 3) and cell monolayer integrity; and SCFA composition and L. monocytogenes counts. All statistical analyses were performed using SAS, version 9.4 (SAS Institute Inc., Cary, NC) and GraphPad Software (La Jolla, CA). In all tests, significance was determined when (P < 0.05). 180 RESULTS Effects of fiber fermentation end products on cell monolayer integrity The effect of fermentation end products from five dietary fiber sources on epithelial barrier function was modeled in HT29-MTX-E12 cells. TEER was measured after 48 h exposure and no difference in resistance was observed between L. monocytogenes unchallenged and challenged cells (P ≥ 0.05). Across all treatments, average TEER (2262 ± 70 Ωcm2) indicated the monolayers were intact. For cells treated with only DMEM (Figure 4.1), TEER was influenced by fermentation end product fiber source (P < 0.001), and all fermentation end products improved TEER relative to DMEM. Monolayer resistance of cells treated with gum arabic (252.47 ± 9.63%) or arabinoxylan (260.48 ± 9.63%) fermentation end products were greater than that of cells treated with cellulose (149.38 ± 9.63%), guar gum (222.83 ± 9.63%), and XOS (183.70 ± 9.63%) fermentation end products. The absolute TEER response (Figure 4.2) of HT29-MTX-E12 cells differed by fermentation end product fiber source (P < 0.001). Cell monolayer resistance in cells treated with gum arabic (2669 ± 103 Ωcm2) or arabinoxylan (2762 ± 103 Ωcm2) fermentation end products was greater than in cells treated with cellulose (1573 ± 103 Ωcm2), guar gum (2359 ± 103 Ωcm2) or XOS (1947 ± 103 Ωcm2) fermentation end products. Cellulose fermentation end products produced the smallest improvement in barrier resistance among all treatments. Guar gum and XOS fermentation products produced intermediate responses. 181 Figure 4.1 Cell monolayer TEER relative to DMEM by fermentation fiber source1,2 1Data presented as LS means ± SEM, n = 14 for each fiber source 2Fermentation end product fiber sources without a common letter are different (P < 0.05) 182 Figure 4.2 Absolute TEER by fermentation fiber source1,2 1Data presented as LS means ± SEM, n = 14 for each fiber source 2Fermentation end product fiber sources without a common letter are different (P < 0.05) FITC-D4 flux was measured after 2 h of incubation, following 48 h substrate exposure, and no difference in barrier function was observed between L. monocytogenes unchallenged and challenged cells (P ≥ 0.05). FITC-D4 flux across all treatments averaged (128.24 ± 1.10 FU). FITC-D4 flux (Figure 4.3) relative to cells treated with only DMEM was influenced by fermentation end product fiber source (P < 0.001), and all fermentation end products decreased FITC-D4 flux relative to DMEM. Relative FITC- D4 flux of cells treated with gum arabic (-39.09 ± 0.97%), arabinoxylan (-39.22 ± 0.97%), or guar gum (-37.29 ± 0.97%) fermentation end products were less than cellulose (-34.67 ± 0.97%) or XOS (-31.60 ± 0.97%) fermentation end products. The absolute FITC-D4 flux response (Figure 4.4) of HT29-MTX-E12 cells was also influenced by fermentation end product fiber source (P < 0.001). Cells treated with 183 gum arabic (122.7 ± 1.93 FU) or arabinoxylan (122.4 ± 1.93 FU) fermentation end products had less FITC-D4 migration across the polarized layer than cells treated with cellulose (131.8 ± 1.93 FU) or XOS (137.9 ± 1.93 FU) fermentation end products. XOS fermentation end products produced the largest flux response among all treatments. Cellulose fermentation and guar gum end products produced an intermediate response. Figure 4.3 FITC-D4 flux relative to DMEM by fermentation fiber source1,2 1Data presented as LS means ± SEM, n = 14 for each fiber source 2Fermentation end product sources without a common letter are different (P < 0.05) 184 Figure 4.4 Absolute FITC-D4 flux by fermentation fiber source1,2 1Data presented as LS means ± SEM, n = 14 for each fiber source 2Fermentation end product sources without a common letter are different (P < 0.05) Effects of fermentation end products on L. monocytogenes The effect of fermentation end products on L. monocytogenes association were evaluated by plating serial dilutions of HT29-MTX-E12 lysates after 90 min challenge following the 48 h fermentation end product treatment period. Fermentation end product by fiber source tended (P = 0.068) to influence cell-associated L. monocytogenes counts (Figure 4.5), and internalized L. monocytogenes counts (Figure 4.6) were significantly different (P = 0.012) for cells treated with fermentation end products from different fiber sources. Cell-associated counts were greatest in cells treated with arabinoxylan (Ln 15.52 ± 0.37 CFU/mL) compared to cellulose (Ln 14.25 ± 0.37 CFU/mL), guar gum (Ln 14.15 ± 0.37 CFU/mL), or XOS (Ln 14.22 ± 0.37 CFU/mL) fermentation end products. Internalized L. monocytogenes estimated standard plate counts (ESPC) were greatest in cells treated with gum arabic (Ln 7.60 ± 0.50 CFU/ml), 185 arabinoxylan (Ln 7.14 ± 0.50 CFU/ml), or cellulose (Ln 7.60 ± 0.50 CFU/ml) fermentation end products. Cells treated with XOS fermentation end products (Ln 5.30 ± 0.50 CFU/ml) had the lowest internalized L. monocytogenes counts. Figure 4.5 Cell-associated L. monocytogenes counts by fermentation fiber source1,2 1Data presented as (Ln LS means ± SEM CFU/ml L. monocytogenes), n = 7 for each fiber source 2Fermentation end product fiber sources without a common letter are different (P < 0.1) 186 Figure 4.6 Internalized L. monocytogenes counts by fermentation fiber source1,2 1Data presented as (Ln LS means ± SEM CFU/ml L. monocytogenes), n = 7 for each fiber source 2Fermentation fiber sources without a common letter are different (P < 0.05) Correlation analysis of fermentation end product SCFA composition with monolayer barrier integrity Correlation analysis was conducted to examine the relationship between monolayer integrity and SCFAs in this model (Table 4.2). Total SCFA concentration was weakly, positively correlated with absolute TEER and relative TEER. Acetate concentration was moderately, positively correlated with absolute TEER and relative TEER. Propionate was moderately, positively correlated with absolute TEER and relative TEER. No measures of cell permeability were significantly correlated with butyrate concentration (P ≥ 0.05). However, there were trends (P < 0.1) for weak, 187 positive correlations between absolute FITC-D4 and relative FITC-D4 flux with butyrate concentration. Table 4.2 Correlation analysis of SCFA concentration and cell monolayer integrity1,2 0.514, 0.243, P = 0.037 P < 0.001 Acetate (mM) Total SCFA (mM) 0.250, Cell Permeability Absolute TEER Relative TEER Absolute FITC-D4 Relative FITC-D4 1Data presented Spearman correlation coefficients, n = 70 2Signfigance determined when (P < 0.05) -0.108, P = 0.371 -0.099, P = 0.413 0.094, P = 0.441 P = 0.043 P < 0.001 0.100, P = 0.408 0.512, Propionate (mM) 0.417, P < 0.001 0.408, P < 0.001 -0.189, P = 0.118 -0.186, P = 0.120 Butyrate (mM) 0.038, P = 0.758 0.027, P = 0.824 0.217, P = 0.072 0.218, P = 0.069 Correlation analysis of bacterial taxa with monolayer integrity Correlation analysis was conducted to examine the relationship between cell monolayer integrity in this model and the relative abundances of important bacterial phyla identified by 16S rRNA gene sequencing (Chapter 3) were correlated to examine their relationship in this model (Table 4.3). Bacteroidetes was moderately, positively correlated with relative and absolute TEER; and weakly, negatively correlated with relative and absolute FITC-D4 flux. Firmicutes was weakly, negatively correlated with relative and absolute TEER, and weakly, positively correlated with relative and absolute FITC-D4 flux. Verrucomicrobia was weakly, negatively correlated with relative and absolute TEER. Correlation analysis was also conducted to examine the relationship of the relative abundances of important bacterial genera identified by 16S rRNA gene 188 sequencing (Chapter 3) with cell monolayer integrity in this model system (Table 4.4). Faecalibacterium was weakly, negatively correlated with relative and absolute TEER. Erysipelotrichaceae UCG-003 was weakly, negatively correlated with relative and absolute TEER, and weakly, positively correlated with relative and absolute FITC-D4 flux. Lachnospiraceae genomosp. was weakly, positively correlated with relative and absolute FITC-D4 flux. Lachnospiraceae uncultured was weakly, negatively correlated with relative and absolute TEER. Table 4.3 Correlation of relative abundances (%) of phyla identified in the production of fermentation end products by 16S rRNA gene sequencing with monolayer integrity1,2 Genus Absolute TEER Relative to DMEM 0.400, P < 0.001 -0.362, P = 0.002 -0.032, P = 0.792 -0.128, P = 0.290 -0.337, P = 0.004 FITC-D4 Relative to DMEM -0.270, P = 0.024 0.300 P = 0.012 -0.157, P = 0.194 0.196, P = 0.105 -0.059, P = 0.631 TEER 0.393, P < 0.001 -0.358, P = 0.002 -0.041, P = 0.733 -0.112, P = 0.355 -0.334, P = 0.005 Absolute FITC-D4 -0.265, P < 0.027 0.297, P = 0013 -0.164, P = 0.176 0.194, P = 0.108 -0.054, P = 0.654 Bacteroidetes Firmicutes Proteobacteria Actinobacteria Verrucomicrobia 1Data presented as Spearman correlation coefficients, n = 70 2Signfigance determined when (P < 0.05) 189 Faecalibacterium Agathobacter Roseburia Ruminococcaceae UCG-002 Erysipelotrichaceae UCG-003 Fusicatenibacter Blautia TEER Relative to DMEM -0.250, P = 0.037 0.068, P = 0.576 0.142, P = 0.240 0.067, P = 0.580 -0.217, P = 0.071 0.190, P = 0.116 0.200, P = 0.096 FITC-D4 Relative to DMEM -0.032, P = 0.794 0.126, P = 0.297 0.010, P = 0.934 -0.105, P = 0.386 0.297, P = 0.013 0.083, P = 0.494 0.143, P = 0.237 0.302, P = 0.011 TEER -0.250, P = 0.037 0.067, P = 0.583 0.161, P = 0.183 0.059, P = 0.626 -0.211, P = 0.080 0.188, P = 0.118 0.200, P = 0.096 -0.120, P = 0.322 -0.271, P = 0.023 Absolute FITC-D4 -0.029, P = 0.814 0.130, P = 0.283 0.016, P = 0.892 -0.100, P = 0.408 0.308, P = 0.010 0.093, P = 0.442 0.143, P = 0.239 0.298, P = 0.012 0.193, P = 0.109 Table 4.4 Correlation of relative abundances (%) of genera identified in the production of fermentation end products by 16S rRNA gene sequencing with monolayer integrity1,2 Genus Absolute Lachnospiraceae genomosp. Lachnospiraceae uncultured 1Data presented as Spearman correlation coefficients, n = 70 2Signfigance determined when (P < 0.05) -0.129, P = 0.286 -0.285, P = 0.017 P = 0.104 0.196, Correlation analysis of L. monocytogenes counts with SCFAs SCFA concentrations were correlated with L. monocytogenes counts to examine their relationship in this model (Table 4.5). Cell associated L. monocytogenes counts were not correlated with SCFAs. However, internalized counts were moderately, negatively associated with total SCFAs and butyrate concentrations (P < 0.05). 190 Table 4.5 Correlation analysis of SCFA concentration and L. monocytogenes counts1,2 Total SCFA (mM) L. monocytogenes Counts Cell Associated (Ln CFU/ml) Internalized (Ln CFU/ml) 1Data presented Spearman correlation coefficients, n = 35 P = 0.400 -0.233, P = 0.178 Acetate (mM) 0.147, -0.017, P = 0.925 -0.352, P = 0.038 Propionate (mM) -0.017, P = 0.923 -0.165, P = 0.345 Butyrate (mM) -0.101, P = 0.562 -0.420, P = 0.012 191 DISCUSSION The HT29-MTX-E12 human cell line has been widely used to model in vitro L. monocytogenes infection (Coconnier et al., 1998; Lievin-Le Moal et al., 2005). This colon cell line expresses tight-junction proteins and secrets mucus (Martinez-Maqueda et al., 2015), which are both important factors affecting L. monocytogenes entry into gastro-intestinal cells (Lievin-Le Moal et al., 2005; Drolia and Bhunia, 2019). We examined the effects of fermentation end products from five commercially available dietary fiber sources on gastro-intestinal barrier function in L. monocytogenes challenged and unchallenged HT29-MTX-E12 cells. While our fermentation end products contained a range of SCFAs compositions (Table 4.1), incubation with them increased resistance of the polarized HT29-MTX-E12 monolayer and decreased FITC- D4 flux after 48 h exposure regardless of L. monocytogenes challenge. These observations support results from Coconnier et al. (1998) who observed no difference in TEER between L. monocytogenes infected and uninfected HT29-MTX-E12 cells. Across all treatments TEER averaged (2262 ± 70 Ωcm2), and with HT29-MTX-E12 cells, the monolayer is considered intact with TEER values > 200 Ωcm2 (Martinez-Maqueda et al., 2015). Additionally, fermentation end products were not cytotoxic (APPENDIX A, Figure 4.7A). Our results generally agree with Chen et al. (2017), who observed that in vitro fermentation end products from fructo-oligosaccharide, sorghum arabinoxylan, corn arabinoxylan, and raw potato starch increased Caco-2 monolayer resistance; however, both of their arabinoxylan treatments were somewhat cytotoxic. In humans, normal proximal colon SCFA concentration is 70-140 mM and decreases to 20-70 mM in the distal colon (Topping and Clifton, 2001). Total SCFAs in the experimental treatments ranged from 2 – 68 mM. We chose to use a 50% 192 concentration of fermentation end products (Table 4.1) to approximate normal colon SCFA concentrations as closely as possible while including 5% FBS to limit the potential for detachment of cells from Transwell membranes. HT29-MTX-E12 cells grown in media without fetal calf serum have decreased growth, reduced ability to form an adherent cell layer, and abnormal mucin production (Navabi et al., 2013). Preliminary TEER (APPENDIX B, Figure 4.8A), FITC-D4 (APPENDIX B, Figure 4.9A), and cytotoxicity (APPENDIX B, Figure 4.10A) assays were performed and a dose-dependent response was observed in these parameters as the concentration of fermentation end products was increased, with the maximum response observed at 50% fermentation end product concentration. There was no evidence of cytotoxicity when the 50% concentration of fermentation end products was used (APPENDIX B, Figure 4.10A). We did not evaluate the effects of pure, individual SCFAs in this model as other have, but do present the initial SCFA compositions of our fermentation end products, and we believe SCFAs are likely bioactive compounds in our model. For example, cell culture medium containing 5% fermentation end products or sodium butyrate (1-10 mM) improved HT29-MTX-E12 monolayer barrier performance. On the other hand, lower (0.1 mM) and higher (50-100 mM) concentrations of sodium butyrate did not improve barrier function (Nielsen et al., 2018). Similarly, Peng et al. (2007) observed that low levels of butyrate (2 mM) improved Caco-2 cell barrier function, whereas 8 mM butyrate diminished function. For the treatments used in this experiment, the butyrate concentration ranged from 0.35-21.9 mM, with cellulose and XOS having the lowest and highest concentrations, respectively. Cells treated with cellulose and XOS fermentation end products had the smallest improvement in barrier integrity. Recently, Pham et al. (2018) 193 demonstrated that monolayer permeability decreased in Caco-2/HT29-MTX-E12 co- culture apically treated with a 20-fold dilution of oat β-glucan (28% or 94%) 24 h fermentation end products compared to corresponding 0 h fermentation end product treatments. We observed a weak to moderately positive association between barrier function and total SCFA, acetate, and propionate concentrations. Butyrate concentration tended to be positively associated with FITC-D4 flux, indicating that higher concentrations of butyrate were associated with diminished barrier integrity, which supports observations by Peng et al. (2007). We believe improvement in barrier function is likely due to a synergistic combination of acetate, propionate, and butyrate rather than individual SCFAs. Butyrate enhances barrier integrity by affecting tight junction assembly through the AMP-activated protein kinase (AMPK) pathway (Peng et al., 2009). However, other signaling pathways may be affected and their physiological effect on barrier function has yet to be fully described. Other potentially important signaling pathways regulating barrier function involve G-protein coupled receptors. For example, the receptor GRP43 is highly expressed by immune cells, enterocytes, and colonocytes, and acetate and propionate have higher activity than longer chain fatty acids (Husted et al., 2017; Ulven, 2012). On the other hand, propionate and butyrate have greater activity than acetate with GPR41 (Ulven, 2012). A third receptor, GPR109A, is expressed by enterocytes and butyrate is a ligand involved in stabilizing the intestinal barrier (Husted et al., 2017). Next, we correlated relative abundances of bacterial phyla, sequenced in a previous study (Chapter 3), with barrier function. Bacteroidetes was positively associated with barrier integrity, while Firmicutes was negatively associated. To identify potential bacterial genera associated with increased or decreased barrier function, we 194 correlated relative abundances of statistically significant bacterial genera, all which were members of the phylum Firmicutes. We found higher relative abundances of Faecalibacterium, Erysipelotrichaceae UCG-003, Lachnospiraceae uncultured, Lachnospiraceae genomosp. to be associated with diminished barrier function. The only known species of Faecalibacterium is F. prausnitzii, which is associated with decreased inflammation and gastro-intestinal health (Duncan et al., 2002; Sokol et al., 2008; Laursen et al., 2017; Louis and Flint, 2017). Erysipelotrichaceae UCG-003 is an uncultured member of the family Erysipelotrichaceae. Ten genera and 12 species make up this family of rod-shaped anaerobes and microaerophiles. Some strains are pathogens for mammals and birds, and higher abundances are found in patients with gastro-intestinal diseases such as colorectal cancer, inflammatory bowel diseases, and ulcerative colitis (Verbarg et al., 2014; Kaakoush, 2015; Wingfield et al., 2018). Members of the family Lachnospiraceae are all anaerobes (Rainey, 2015). A recent meta-analysis identified an unclassified member of Lachnospiraceae to be associated with gastro-intestinal health (Mancabelli et al., 2017). Our observations contradict the positive gastro-intestinal health associations others have noted for members of Faecalibacterium and Lachnospiraceae. We recognize the limitation in correlating phylum and genus-level bacterial classification with in vitro modeled responses of barrier function and are cautious in drawing a firm conclusion for that reason and because bacteria were removed by filtration of the fermentation end products. Recently, Franco-de-Moraes et al. (2017) observed that strict vegetarians had higher fecal Bacteroidetes abundance and lower markers of inflammation. Our data suggest shifts in the relative abundances of Bacteroidetes and Firmicutes may have 195 differing effects on barrier integrity depending on which phylum is dominantly represented, and fermentation end product composition may be an important factor. Next we evaluated the effects of fermentation end products on L. monocytogenes association. Cell-associated L. monocytogenes, which consisted of external and internalized bacterial cells, tended to be influenced by the different treatments. Internalized L. monocytogenes counts were significantly influenced by fermentation end product fiber source, albeit the internalized levels of L. monocytogenes were low. The lowest internalized L. monocytogenes counts (Figure 4.6) were recovered from cells treated with XOS fermentation end products. We are unaware of other in vitro models evaluating the effects of fermentation end products on L. monocytogenes association. However, other researchers have demonstrated that inulin, oligofructose, XOS, and galacto-oligosaccharide are protective against L. monocytogenes and Salmonella enterica serovar Typhimurium infection in animal models of infection (Buddington et al., 2002; Ebersbach et al., 2012). On the other hand, Petersen et al. (2009) observed increased severity of S. enterica serovar Typhimurium infection in mice fed XOS and FOS. Previously, we observed a higher rate of infection in L. monocytogenes challenged mice fed 10% XOS, as well as higher cecum and spleen counts (Chapter 2). We are cautious in drawing a firm conclusion on differences in L. monocytogenes association due to our internalized counts being estimated standard plate counts, but we highlight increased L. monocytogenes internalization with cellulose derived fermentation end products because cellulose has limited fermentation by human colonic bacteria (Cummings, 1984). Cells treated with cellulose-derived fermentation end products were exposed to the lowest levels of SCFAs and had reduced barrier resistance when compared to cells treated with end products from fermentation of gum arabic, 196 arabinoxylan, guar gum, or XOS. This is consistent with the negative association between internalized L. monocytogenes counts and, total SCFAs and butyrate that we observed. However, we found no significant correlations between barrier integrity responses and L. monocytogenes counts (data not shown), which is surprising. On the other hand, butyrate and FITC-D4 flux tended to be positively associated, which means that as butyrate concentration increased, permeability increased. We observed greater permeability in cells treated with XOS fermentation end products, which contained the highest concentration of butyrate, but these cells had the lowest level of L. monocytogenes internalization. These observations support our conclusion there is likely a synergistic effect of fermentation metabolites on barrier function rather than individual SCFAs, and low SCFA exposure may be responsible for higher L. monocytogenes internalization. We believe disruption of the epithelial cell barrier to be an important factor influencing L. monocytogenes cell invasion due to the variety of virulence factors it expresses. For example, one invasion mechanism is dependent on the internalin (InlA) virulence factor, which attaches to the adherens junction protein E-cadherin (Lievin-Le Moal et al., 2005). Wild-type L. monocytogenes EGD enters E-cadherin expressing HT29-MTX-E12 cells in an InlA-dependent mechanism. (Lievin-Le Moal et al., 2005). We demonstrate the concentration (Table 4.5) and source of SCFAs (Figure 4.6) to be important factors affecting the internalization of L. monocytogenes. How fermentation metabolites affect tight junction expression was not measured in this study, and additional investigation of the effects of varying fermentation end product concentrations on L. monocytogenes invasion is warranted. 197 Mucin was not evaluated as planned due to low levels of detection in a preliminary assay (APPENDIX B, Figure 4.11A). Pham et al. (2018) observed that XOS fermentation end products affected the mucin response in HT-29MTX-E12 cells. However, the effects they observed were modest and varied by fecal sample donor. They also did not observe differences in mucin production between other fermentation end products they evaluated (Pham et al., 2018), which leads us to believe that differences in L. monocytogenes association are likely not due to any mucus response in our model. In the present study, we recognize there are some weaknesses related to our experimental design. First, we only investigated the effects of bacterial produced fermentation metabolites on physical barrier response by measuring TEER and FITC- D4 flux. We acknowledge that complex immune signaling, bacterial metabolites, and bacterial cross talk are important factors regulating gastro-intestinal homeostasis and inhibition of L. monocytogenes invasion, and these that were not evaluated in our model. We also chose one dose of fermentation end products to evaluate, where others have observed dose-dependent responses of purified SCFAs on varying response variables related to barrier function. However, we also consider this a strength because we demonstrated that end product arising from fermentation of different fiber sources had differential effects on gastro-intestinal barrier function. In conclusion, barrier integrity improved with exposure to fermentation end products regardless of fiber source. We observed small, but significant, differences in HT29-MTX-E12 internalization of L. monocytogenes due to end products from fermentation of different fiber sources and a trend for a difference in total L. monocytogenes counts. Therefore, we believe fiber source to be an important factor influencing fermentation end product composition, and certain compositions may have 198 greater potential to limit invasion of human colon cells by L. monocytogenes and potentially other invasive pathogens. We did not evaluate mechanisms for the responses we observed, but hypothesize our fermentation end products influence accessibility to tight junction proteins, which are important for preventing L. monocytogenes internalization through the InlA-dependent mechanism. In Caco-2 cells, butyrate facilitates assembly of tight junction proteins by activation of the AMPK pathway (Peng et al., 2009). Sodium butyrate, sodium propionate and sodium acetate increase oxygen consumption by Caco-2 cells but at varying degrees, which stabilizes hypoxia-inducible factor (HIF), which is linked to enhanced barrier function (Kelly et al., 2015). Future research investigating the effects of fiber sources and their fermentation end products on these pathways in in vivo infection models is warranted. 199 APPENDICES 200 APPENDIX A: CELL CYTOXICTY Figure 4.7A Cytotoxicity of fermentation end products in unchallenged HT29-MTX- E12 cells after 48 h exposure1 1Data presented as means ± SEM relative (%) to Lysis Buffer (+) control, n = 7 for each fiber sources 201 APPENDIX B: FERMENTATION END PRODUCT CONCENTRATION DETERMINATION ASSAYS Figure 4.8A Preliminary evaluation of fermentation end product dilutions on TEER response1,2 1Data are individual values of TEER responses for each fermentation end product concentration 2Fermentation end products concentrations in DMEM (5% FBS): 6.25%, 12.50%, 25%, and 50% 202 Figure 4.9A Preliminary evaluation of fermentation end product dilutions on FITC-D4 flux response1,2 1Data are individual values of FITC-D4 flux for each fermentation end product dilution 2Fermentation end products concentrations in DMEM (5% FBS): 6.25%, 12.50%, 25%, and 50% 203 Figure 4.10A Preliminary cytotoxicity evaluation of fermentation end product dilutions1,2 1Data are individual values (%) cytotoxicity relative to lysis buffer 2Fermentation end products concentrations in DMEM (5% FBS): 6.25%, 12.50%, 25%, and 50% 204 APPENDIX C: MUC5AC PRELIMINARY ASSAY Figure 4.11A Preliminary MUC5AC assay1,2,3,4 1Data presented as individual values of preliminary MUC5AC ELISA assay 2Standard curve R2 = 0.977 3Assay detection range: 15.6-1000 ng/mL 4Assay sensitivity: 3.9 ng/mL 205 REFERENCES 206 REFERENCES Barcelo A, Claustre J, Moro F, Chayvialle JA, Cuber JC, Plaisancié P. Mucin secretion is modulated by luminal factors in the isolated vascularly perfused rat colon. Gut. 2000;46:218–24. Bourquin LD, Titgemeyer EC, Garleb K, Fahey GC. Short-chain fatty acid production and fiber degradation by human colonic bacteria: effects of substrate and cell wall fractionation procedures. J Nutr. 1992;122:1508–20. Bourquin LD, Titgemeyer EC, Fahey GC. Vegetable Fiber Fermentation by Human Fecal Short-Chain Fatty Acid Production during In Vitro Bacteria: Cell Wall Polysaccharide Disappearance and Fermentation and Water-Holding Capacity of Unfermented Residues. J Nutr. 1993;123:860–9. Bucior I, Tran C, Engel J. Assessing Pseudomonas Virulence Using Host Cells. In: Alain Filloux, Ramos J-L, editors. Methods in Molecular Biology. New York: Springer; 2014. p. 741–55. Buddington KK, Donahoo JB, Buddington RK. Dietary Oligofructose and Inulin Protect Mice from Enteric and Systemic Pathogens and Tumor Inducers. J Nutr. 2002;132:472– 7. Burger-van Paassen N, Vincent A, Puiman PJ, van der Sluis M, Bouma J, Boehm G, van Goudoever JB, van Seuningen I, Renes IB. The regulation of intestinal mucin MUC2 expression by short-chain fatty acids: implications for epithelial protection. Biochem J. 2009;420:211–9. Cajnko MM, Marusic M, Kisovec M, Rojko N, Bencina M, Caserman S, Anderluh G. Listeriolysin O Affects the Permeability of Caco-2 Monolayer in a Pore-Dependent and Ca2+ -Independent Manner. PLoS One. 2015;1–21. Centers for Disease Control and Prevention. Listeria(Listeriosis) [Internet]. 2019 [cited 2019 Apr 17]. Available from: https://www.cdc.gov/listeria/index.html Chen T, Kim CY, Kaur A, Lamothe L, Shaikh M, Keshavarzian A, Hamaker BR. Dietary fibre-based SCFA mixtures promote both protection and repair of intestinal epithelial barrier function in a Caco-2 cell model. Food Funct. Royal Society of Chemistry; 2017;8:1166–73. Coconnier NE, Dlissi E, Robard M, Laboisse CL. Listeria monocytogenes Stimulates Mucus Exocytosis in Cultured Human Polarized Mucosecreting Intestinal Cells through Action of Listeriolysin O. Infect Immun. 1998;66:3673–81. 207 Cornick S, Tawiah A, Chadee K. Roles and regulation of the mucus barrier in the gut. Tissue Barriers. 2015;3:e982426-1-e982426-15. Cummings JH. Cellulose and the human gut. Gut. 1984;25:805–10. DeVries JW. The definition of dietary fibre. Cereal Foods World. 2001;46:112–29. Drolia R, Bhunia AK. Crossing the Intestinal Barrier via Listeria Adhesion Protein and Internalin A. Trends Microbiol. Elsevier Ltd; 2019;27:408–25. Drolia R, Tenguria S, Durkes AC, Turner JR, Bhunia AK, Drolia R, Tenguria S, Durkes AC, Turner JR, Bhunia AK. Listeria Adhesion Protein Induces Intestinal Epithelial Barrier Dysfunction for Bacterial Article Listeria Adhesion Protein Induces Intestinal Epithelial Barrier Dysfunction for Bacterial Translocation. Cell Host Microbe. Elsevier Inc.; 2018;23:470–484. Duncan SH, Hold GL, Harmsen HJM, Stewart CS, Flint HJ. Growth requirements and fermentation products of Fusobacterium prausnitzii, and a proposal to reclassify it as Faecalibacterium prausnitzii gen. nov., comb. nov. Int J Syst Evol Microbiol. 2002;52:2141–6. Ebersbach T, Jørgensen JB, Heegaard PM, Lahtinen SJ, Ouwehand AC, Poulsen M, Frøkiaer H, Licht TR. Certain dietary carbohydrates promote Listeria infection in a guinea pig model, while others prevent it. Int J Food Microbiol. Elsevier B.V.; 2010;140:218–24. Flint HJ, Scott KP, Duncan SH, Louis P, Forano E. Microbial degradation of complex carbohydrates in the gut. Gut Microbes. 2012;3:289–306. Franco-de-Moraes AC, Almeida-Pititto B de, Fernandes G da R, Gomes EP, Pereira A da C, Ferreira SRG. Worse inflammatory profile in omnivores than in vegetarians associates with the gut microbiota composition. Diabetol Metab Syndr. BioMed Central; 2017;9:1–8. Hamaker B, Tuncil Y. A Perspective on the Complexity of Dietary Fiber Structures and Their Potential Effect on the Gut Microbiota. J Mol Biol. 2014;426:3838–50. Hryckowian AJ, Treuren W Van, Smits SA, Davis NM, Gardner JO, Bouley DM, Sonnenburg JL. Microbiota-accessible carbohydrates suppress Clostridium difficile infection in a murine model. Nat Microbiol. Springer US; 2018;3:662–9. Husted AS, Trauelsen M, Rudenko O, Hjorth SA, Schwartz TW. Review GPCR-Mediated Signaling of Metabolites. Cell Metab. Elsevier Inc.; 2017;25:777–96. Kaakoush NO. Insights into the Role of Erysipelotrichaceae in the Human Host. Front Cell Infect Microbiol. 2015;5:1–4. 208 Kelly CJ, Zheng L, Taylor CT, Colgan SP. Tissue Barrier Function Short Article Crosstalk between Microbiota-Derived Short-Chain Fatty Acids and Intestinal Epithelial HIF Augments Tissue Barrier Function. Cell Host Microbe. Elsevier Inc.; 2015;17:662–71. Kühbacher A, Cossart P, Pizarro-Cerdá J. Internalization assays for Listeria monocytogenes. Methods in Molecular Biology. Humana Press/Springer Imprint; 2013. p. 167–78. Laursen MF, Laursen P, Larnkjær A, Mølgaard C, Michaelsen KF, Frøkiær H, Bahl I. Faecalibacterium Gut Colonization Is Accelerated by Presence of Older Siblings. mSphere. 2017;2:1–6. Liévin-Le Moal V, Servin AL, Coconnier-Polter MH. The increase in mucin exocytosis and the upregulation of MUC genes encoding for membrane-bound mucins induced by the thiol-activated exotoxin listeriolysin O is a host cell defence response that inhibits the cell-entry of Listeria monocytogenes. Cell Microbiol. 2005;7:1035–48. Lindén SK, Bierne H, Sabet C, Png CW, Florin TH, McGuckin MA, Cossart P. Listeria monocytogenes internalins bind to the human intestinal mucin MUC2. Arch Microbiol. 2008;190:101–4. Louis P, Flint HJ. Formation of propionate and butyrate by the human colonic microbiota. Environmental Microbiol. 2017;19:29–41. Mancabelli L, Milani C, Lugli GA, Turroni F, Cocconi D, Sinderen D Van, Ventura M. Identification of universal gut microbial biomarkers of common human intestinal diseases by meta-analysis. FEMS Microbiol Ecol. 2017;1–10. Martínez-Maqueda D, Miralles B, Recio I. HT29 Cell Line. In: Verhoeckx K, Cotter P, López-Expósito I, Kleiveland C, Lea T, Mackie A, Requena T, Swiatecka D, Wichers H, editors. The Impact of Food Bioactives on Health: in vitro and ex vivo models. Cham: Springer International Publishing; 2015. p. 113–24. Navabi N, Mcguckin MA, Linden SK. Gastrointestinal Cell Lines Form Polarized Epithelia with an Adherent Mucus Layer when Cultured in Semi-Wet Interfaces with Mechanical Stimulation. PLoS One. 2013;8:1–15. Nielsen DSG, Jensen BB, Theil PK, Nielsen TS, Knudsen KEB, Purup S. Effect of butyrate and fermentation products on epithelial integrity in a mucus-secreting human colon cell line. J Funct Foods. Elsevier; 2018;40:9–17. Peng L, Li Z-R, Green RS, Holzman IR, Lin J. Butyrate Enhances the Intestinal Barrier by Facilitating Tight Junction Assembly via Activation of AMP-Activated Protein Kinase in Caco-2 Cell Monolayers. J Nutr. 2009;139:1619–25. 209 Peng L, He Z, Chen W, Holzman IR, Lin J. Effects of butyrate on intestinal barrier function in a caco-2 cell monolayer model of intestinal barrier. Pediatr Res. 2007;61:37– 41. Pentecost M, Otto G, Theriot JA, Amieva MR. Listeria monocytogenes invades the epithelial junctions at sites of cell extrusion. PLoS Pathog. 2006;2:0029–40. Petersen A, Heegaard PMH, Pedersen AL, Andersen JB, Sørensen RB, Frøkiær H, Lahtinen SJ, Ouwehand AC, Poulsen M, Licht TR. Some putative prebiotics increase the severity of Salmonella enterica serovar Typhimurium infection in mice. BMC Microbiol. 2009;9:1–11. Pham VT, Seifert N, Richard N, Raederstorff D, Steinert R, Prudence K, Mohajeri MH. The effects of fermentation products of prebiotic fibres on gut barrier and immune functions in vitro. PeerJ. 2018;e5288. Rainey FA. Lachnospiraceae fam. nov. In: Whitman WB, editor. Bergey’s Manual of Systematics of Archaea and Bacteria. John Wiley & Sons, Inc.; 2015. p. 1–2. Schulzke JD, Ploeger S, Amasheh M, Fromm A, Zeissig S, Troeger H, Richter J, Bojarski C, Schumann M, Fromm M, et al. Epithelial Tight Junctions in Intestinal Inflammation. New York Acad Sci. 2009;1165:294–300. Smirnova M, Kiselev S, Birchall JP, Pearson JP. Up-regulation of mucin secretion in HT29-MTX cells by the pro-inflammatory cytokines tumor necrosis factor-alpha and interleukin-6. Eurpopean Cytokine Netw. 2001;12:119–25. Sokol H, Pigneur B, Watterlot L, Lakhdari O, Bermudez-Humaran LG, Gratadoux J, Blugeon S, Bridonneau C, Furet J-P, Corthier G, Grangette C, Vasquez N, Pochart P, Trugnan G, Thomas G, Blottiere HM, Dore J, Marteau P, Seksik P, Langella P. Faecalibacterium prausnitzii is an anti-inflammatory commensal bacterium identified by gut microbiota analysis of Crohn disease patients. Proc Natl Acad Sci. 2008;105:16731–36. Suzuki T, Yoshida S, Hara H. Physiological concentrations of short-chain fatty acids immediately suppress colonic epithelial permeability. Br J Nutr. 2008;100:297–305. Ten Bruggencate SJM, Bovee-Oudenhoven IMJ, Lettink-Wissink MLG, Van der Meer R. Dietary fructooligosaccharides increase intestinal permeability in rats. J Nutr. 2005;135:837–42. Topping DL, Clifton PM. Short-Chain Fatty Acids and Human Colonic Function: Roles of Resistant Starch and Nonstarch Polysaccharides. Physiol Rev. 2001;81:1031–64. Ulven T, Smith NJ, Chang V. Short-chain free fatty acid receptors FFA2 / GPR43 and FFA3 / GPR41 as new potential therapeutic targets. Frontiers Endocrinol. 2012;3:1–9. 210 Verbarg S, Goker M, Scheuner C, Schumann P, Stackebrandt E. The Families Erysipelotrichaceae emend., Coprobacillaceae fam. nov., and Turicibacteraceae fam. nov. In: Rosenberg E, DeLong EF, Lory S, Stackebrandt E, Thompson F, editors. The Prokaryotes. 4th ed. Berlin: Springer-Verlag Berlin Heidelberg; 2014. p. 79–105. Wingfield B, Coleman S, McGinnity T, Bjourson A. Robust Microbial Markers for Non- Invasive Inflammatory Bowel Disease Identification. IEEE/ACM Trans Comput Biol Bioinforma. 2018;1545–5963. 211 SUMMARY CHAPTER 5: SUMMARY AND FUTURE STUDIES Americans are not consuming adequate dietary fiber (Hoy and Goldman, 2014). Consumption of dietary fiber strongly influences gastro-intestinal bacterial composition (Walker et al., 2011; David et al., 2014; De Filippo et al., 2017). The five major factors for gastro-intestinal health are: 1) effective digestion and absorption of food; 2) absence of gastro-intestinal disease; 3) a normal and stable gut microbiota; 4) a functioning immune system; and 5) well-being status (Bischoff, 2011). The results from this project support the idea that a normal and stable gut microbiota and its fermentation metabolites are important factors for gastro-intestinal health and the prevention of enteric infections. We demonstrated drastic bacterial compositional changes are associated with enhanced L. monocytogenes infection in mice (Chapter 2). The digesta of mice fed XOS were enriched with Verrucomicrobia (Akkermansia) and Actinobacteria, and had low gastro-intestinal bacterial diversity as a result of this compositional shift. Additionally, Verrucomicrobia (Akkermansia) was positively associated with cecum L. monocytogenes counts. These fiber source-dependent bacterial compositional changes observed in vivo were confirmed in vitro, albeit with different and muted responses (Chapter 3). Conversely, bacterial diversity and SCFA compositions in vitro were more similar than different when compared by fiber source. This reflects a limitation of in vitro modeling. However, despite the muted response in vitro, XOS fermentation increased the relative abundances of genera (Erysipelotrichaceae and Blautia) associated with diminished gastro-intestinal health, while gum arabic and guar gum increased relative abundances of beneficial genera (Faecalibacterium and Roseburia). 212 The hypothesis that gastro-intestinal barrier function is differentially affected by fermentation end product fiber source was confirmed (Chapter 4). All fiber source fermentation products improved barrier function in vitro. However, XOS and cellulose produced muted responses, but this did not lead to meaningful differences in L. monocytogenes association in HT29-MTX-E12 cells. This leads us to conclude that in vivo models are more suitable for understanding complex diet and microbiota interactions, although in vitro models maybe useful for investigating small-scale, designed bacterial community metabolic studies. Overall, this project indicates that XOS leads to enhanced L. monocytogenes infection in mice, but this is independent of the effects of fermentation end products on barrier function, based on the in vitro results. We also demonstrated that low-level SCFA exposure may enhance L. monocytogenes association with gastro-intestinal epithelial cells in vitro. We believe that XOS, as a rapidly fermentable fiber, alters the gastro-intestinal microbiota in a way that shifts bacterial composition favoring mucus layer degradation and thereby enhancing epithelial cell contact with potential microbial pathogens such as L. monocytogenes. In conclusion, the gastro-intestinal microbiota is sensitive to changes in fiber source availability and dramatic compositional shifts of the gastro- intestinal microbiota may be an important pathogenicity factor enhancing enteric infections. 213 FUTURE DIRECTIONS Study 1: Gastro-intestinal composition of enteric infection cases The results of chapters 2 and 3 demonstrated that dietary substrates affect gastro-intestinal bacteria composition. The in vivo results of chapter 2 demonstrated that habitual dietary fiber intake has a more robust influence on bacterial composition than short-term exposure in vitro. Chapters 2 and 4 demonstrated that fiber sources, either directly or from the exposure to fermentation metabolites, affect L. monocytogenes counts, invasion, and migration. Determining how dietary substrates affect bacterial composition and enteric infections in humans is challenging. One obstacle is the wide variation of human diets. A second obstacle is the obvious ethical limitation of challenging humans with pathogens. Therefore, I am proposing a new approach to understand the complex interaction between diet, gastro-intestinal bacterial composition, and enteric infections. This approach includes collecting fecal samples from enteric infection cases and dietary assessment data with the aid of healthcare practitioners. Foodborne illnesses affect more than 48 million Americans annually (Centers for Disease Control and Prevention, 2018). Examples of reportable bacterial infections include Salmonella, Listeria, Shigella, Vibrio, and Shiga-toxin producing Escherichia coli. Healthcare professionals report these illnesses to county and city health departments, and states voluntarily report nationally notifiable illnesses to the Centers for Disease Control and Prevention (Centers for Disease Control and Prevention, 2015). Collaboration with county health departments providing primary care would be essential for collecting fecal samples and dietary assessment data from suspect cases. From a practical standpoint, Salmonella, Campylobacter, and Shigella infections are the most intriguing for investigation due to their high number of 214 laboratory-confirmed cases (Centers for Disease Control and Prevention, 2018). Fecal samples would be subjected to 16S rRNA gene sequencing. This would enable us to understand if there are gastro-intestinal compositions or populations common for these enteric infections in ill humans. Detailed dietary assessment data gathered during patient assessment may provide evidence that certain dietary habits are associated with increased risk of illness due to changes in the gastro-intestinal microbiota. Nutritional deficiencies due to illness or poor nutrition are risk factors for foodborne infections (Lund and O’Brien, 2011), but it is unknown if certain human dietary habits also pose a risk. However, De Filippo et al. (2010) demonstrated that fiber-rich diets are associated with lower relative abundances of potential human pathogens in human digesta. Recently, Decuypere et al. (2016) determined in a proof-of-concept study that 16S rRNA genes could be sequenced from the blood of children with severe febrile illness. Additionally, Salmonella, Shigella, and E. coli were species detected in blood samples. These results provide promising experimental evidence, and further support this proposed experimental approach. Study 2: Gastro-intestinal redox potential and diet The results of chapter 2 demonstrated that low gastro-intestinal bacterial diversity may be an important pathogenicity factor enhancing enteric infections. We found mice that were fed XOS had a higher L. monocytogenes infection rate, and the highest cecum and spleen counts of L. monocytogenes. Mice fed XOS also had the lowest bacterial diversity, and L. monocytogenes-positive mice tended to have lower bacterial diversity in cecum digesta, and bacterial diversity was lower in colon digesta of positive mice. Low bacterial diversity is associated with gastro-intestinal dysbiosis 215 (Nguyen et al., 2015; Zeng et al., 2017). Dysbiosis that occurs with gastro-intestinal inflammation results in enrichment of Proteobacteria, specifically the expansion of facultative anaerobic Enterobacteriaceae in mice (Winter and Bäumler, 2014; Reese et al., 2018). Administration of antibiotics to mice also increases relative abundances of Enterobacteriaceae (Winter and Bäumler, 2014) and rapidly increases gastro-intestinal redox potential (Reese et al., 2018). Redox potential is thought to be shaped by bacterial composition due to differences in metabolism (Reese et al., 2018). Diet affects gut redox potential in mice, and high protein diets leads to an imbalance between gastro-intestinal antioxidants and oxidants (Gu et al., 2008). Additionally, undernourished children have positive fecal redox potentials (Million and Raoult, 2018). Understanding how diet affects redox potential, gastro-intestinal composition, and enteric infections is an emerging area of nutrition research. Previously, Rabbani et al. (2009) demonstrated that supplementation with green bananas rich in resistant starch improved clinical symptoms in children infected with Shigella. Furthermore, humans consuming diets rich in dietary fiber sources have distinct bacterial communities compared to those consuming Western style diets (De Filippo et al., 2010). Inflammatory gastro-intestinal diseases are linked to consumption of Western-style diets (Statovci et al., 2017). Hintze et al. (2012) developed a Westernized diet for rodents, which provides a standardized and more relevant approach for modeling the human diet. These studies justify investigating how different states of malnutrition affect gastro-intestinal redox potential and bacterial composition, and models of enteric infection or inflammatory diseases could be appropriately adapted in experimental animals. 216 Furthermore, to expand the idea that dietary factors affect redox potential and gastro-intestinal bacterial composition, phytochemicals such as anthocyanins, which have antioxidant, anti-inflammatory and anticarcinogenic activity (Pojer et al., 2013) are an important class of dietary substrates that warrant exploration. Recently, Igwe et al. (2019) noted few have investigated the effects of anthocycanins on the gastro-intestinal microbiota. Their 6-study meta-analysis revealed that gastro-intestinal bacteria exposed to anthocyanins in vitro and in vivo were enriched in bifidobacteria and inhibited the human pathogen Clostridium histolyticum. As antioxidant and anti-inflammatory agents, it would be interesting to determine if anthocyanins affect gastro-intestinal microbial composition by modulating the redox potential of the gut. To test this, mice would be fed a Western-style diet containing varying concentrations of either whole- foods rich in anthocyanins or purified compounds. Fecal samples would be subjected to 16S rRNA gene sequencing. Fecal redox potential would be measured according to the method developed by Khan (2013). In summary, diet is an important modulator of the gastro-intestinal microbiota. How different states of malnutrition or dietary substrates affect redox potential and gastro-intestinal bacterial composition is an area of research warranting deeper investigation. The proposed study designs can be applied to understand interactions between gastro-intestinal redox potential, the microbiota, enteric infections, and inflammation related pathologies. Study 3: Effect of dietary fiber consumption on mouse mucus layer and barrier integrity In chapter 2 of this project, we found that mice fed XOS had a higher L. monocytogenes infection rate, and the highest cecum and spleen counts of L. 217 monocytogenes. Gum arabic and XOS-fed mice had digesta enriched with the mucus degrading bacteria, Akkermansia, which tended to be higher in the cecum digesta of L. monocytogenes-positive mice and was higher in colon digesta of L. monocytogenes- positive mice. Akkermansia relative abundance was also positively associated with cecum L. monocytogenes counts. We did not measure mucus layer thickness, gastro- intestinal barrier function, or short-chain fatty acid composition of digesta in this project. Therefore, it would be interesting to conduct a similar study measuring these variables to more fully describe the effects of fiber sources on these factors, which may be important for affecting L. monocytogenes epithelial cell association and subsequent invasion. In a previous study, Hino et al. (2012) demonstrated that soluble and insoluble fibers increase mucus production in rats. In another study, rats fed fiber sources had a thicker large intestine mucus layer than rats fed a fiber-free diet (Hedemann et al., 2009). Short-chain fatty acids stimulate mucin gene expression in vitro (Burger-van Paassen et al., 2009). Therefore, it is important to understand the complex interactions between fiber sources on gastro-intestinal bacterial composition (especially mucus degrading bacteria), mucus production, and enteric infections. Gastro-intestinal mucus may be an important barrier defense factor affecting L. monocytogenes pathogenicity. L. monocytogenes adheres to mucins with internalin proteins. Mucin glycoproteins have been hypothesized to inhibit L. monocytogenes from interacting with epithelial cells, but it has also been hypothesized to promote retention of L. monocytogenes near epithelial cells thereby enabling adhesion to epithelial cells (Lindén et al., 2008). In chapter 4 of this project, we demonstrated that fermentation end products improve in vitro barrier function. To model this in animals, a non-digestible tracer such 218 as Cr-ethylenediaminetetraacetic acid or fluorescein isothiocyanate-dextran 4 could be utilized similarly to how barrier integrity was measured in chapter 4 of this project (Gonzalez-Gonzalez et al., 2019). Another approach would be to use Ussing chambers to measure permeability across gastro-intestinal segments isolated from animals. These experiments would provide in vivo evidence on how dietary fiber sources affect gastro- intestinal barrier integrity. In summary, the approach described here would enable assessment of dietary fiber sources on gastro-intestinal barrier function in vivo, and further expand the results from this project. 219 REFERENCES 220 REFERENCES Bischoff SC. ’Gut health’: a new objective in medicine? BMC Med. 2011;9:1–14. Burger-van Paassen N, Vincent A, Puiman PJ, van der Sluis M, Bouma J, Boehm G, van Goudoever JB, van Seuningen I, Renes IB. The regulation of intestinal mucin MUC2 expression by short-chain fatty acids: implications for epithelial protection. Biochem J. 2009;420:211–9. Centers for Disease Control and Prevention. Foodborne Germs and Illnesses [Internet]. Food Safety. 2018 [cited 2019 May 13]. Available from: https://www.cdc.gov/foodsafety/foodborne-germs.html Centers for Disease Prevention and Conrol. How to Report a Foodborne Illness - Health Departments [Internet]. 2015 [cited 2019 Jun 5]. Available from: https://www.cdc.gov/foodsafety/outbreaks/investigating-outbreaks/report- illness/health-dept.html David LA, Maurice CF, Carmody RN, Gootenberg DB, Button JE, Wolfe BE, Ling A V, Devlin AS, Varma Y, Fischbach MA, Biddinger SB, Dutton RJ, Turnbaugh PJ. Diet rapidly and reproducibly alters the human gut microbiome. Nature. Nature Publishing Group; 2014;505:559–63. De Filippo C, Cavalieri D, Di Paola M, Ramazzotti M, Poullet JB, Massart S, Collini S, Pieraccini G, Lionetti P. Impact of diet in shaping gut microbiota revealed by a comparative study in children from Europe and rural Africa. Proc Natl Acad Sci. 2010;107:14691–6. De Filippo C, Paola M Di, Ramazzotti M, Albanese D, Pieraccini G, Banci E, Miglietta F, Cavalieri D, Collado MC. Diet, Environments, and Gut Microbiota. A Preliminary Investigation in Children Living in Rural and Urban Burkina Faso and Italy. Front Microbiol. 2017;8:1–14. Decuypere S, Meehan CJ, Puyvelde S Van, Block T De, Maltha J, Palpouguini L, Tahita M, Tinto H, Jacobs J, Deborggraeve S. Diagnosis of Bacterial Bloodstream Infections: A 16S Metagenomics Approach. PLoS Neglected Trop Dis. 2016;1–12. González-González M, Díaz-Zepeda C, Eyzaguirre-Velásquez J, González-Arancibia C, Bravo JA, Julio-Pieper M. Investigating Gut Permeability in Animal Models of Disease. Front Phy. 2019;9:1–10. Gu C, Shi Y, Le G. Effect of Dietary Protein Level and Origin on the Redox Status in the Digestive Tract of Mice. Int J Mol Sci. 2008;9:464–75. 221 Hedemann MS, Theil PK, Bach Knudsen KE. The thickness of the intestinal mucous layer in the colon of rats fed various sources of non-digestible carbohydrates is positively correlated with the pool of SCFA but negatively correlated with the proportion of butyric acid in digesta. Br J Nutr. 2009;102:117–25. Hino S, Takemura N, Sonoyama K, Morita A, Kawagishi H, Aoe S, Morita T. Small Intestinal Goblet Cell Proliferation Induced by Ingestion of Soluble and Insoluble Dietary Fiber Is Characterized by An Increase in Sialylated Mucins in Rats. J Nutr. 2012;142:1429–36. Hintze KJ, Benninghoff AD, Ward RE. Formulation of the Total Western Diet (TWD) as a Basal Diet for Rodent Cancer Studies. J Agric Food Chem. 2012;60:6736–42. Hoy MK, Goldman JD. Dietary Fiber Intake of the U.S. Population. 2014. Igwe E, Charlton K, Probst Y, Kent K, Netzel M. A systematic literature review of the effect of anthocyanins on gut microbiota populations. J Hum Nutr Diet. 2019;32. Khan M. Novel physiological and metabolic insights into the beneficial gut microbe Faecalibacterium prausnitzii: from carbohydrates to current. Groningen: s.n.; 2013. Lindén SK, Bierne H, Sabet C, Png CW, Florin TH, McGuckin MA, Cossart P. Listeria monocytogenes internalins bind to the human intestinal mucin MUC2. Arch Microbiol. 2008;190:101–4. Lund BM, O’Brien SJ. The Occurrence and Prevention of Foodborne Disease in Vulnerable People. Foodborne Pathog Dis. 2011;8:961–73. Million M, Raoult D. Linking gut redox to human microbiome. Hum Microbiome J. Elsevier; 2018;10:27–32. Nguyen TLA, Vieira-Silva S, Liston A, Raes J. How informative is the mouse for human gut microbiota research? Dis Model Mech. 2015;8:1–16. Pojer E, Mattivi F, Johnson D, Stockley CS. The Case for Anthocyanin Consumption to Promote Human Health: A Review. Compr Rev Food Sci Food Saf. 2013;12:483–508. Rabbani GH, Ahmed S, Hossain I, Islam R, Marni F, Akhtar M, Majid N. Green Banana Reduces Clinical Severity of Childhood Shigellosis. Pediatr Infect Dis J. 2009;28:420–5. Reese AT, Cho EH, Klitzman B, Nichols SP, Wisniewski NA, Villa MM, Durand HK, Jiang S, Midani FS, Nimmagadda SN, O’Connell TM, Wright JP, Deshusses, David LA. Antibiotic-induced changes in the microbiota disrupt redox dynamics in the gut. eLIFE. 2018;1–22. 222 Statovci D, Aguilera M, MacSharry J, Melgar S. The impact of western diet and nutrients on the microbiota and immune response at mucosal interfaces. Front Immunol. 2017;8:1–21. Walker AW, Ince J, Duncan SH, Webster LM, Holtrop G, Ze X, Brown D, Stares MD, Scott P, Bergerat A, McIntosh F, Johnstone AM, Lobley GE, Parkhill J, Flint HJ. Dominant and diet-responsive groups of bacteria within the human colonic microbiota. ISME J. Nature Publishing Group; 2011;5:220–30. Winter SE, Bäumler AJ. Dysbiosis in the inflamed intestine. Gut Microbes. 2014;5:71–3. Zeng M, Inohara N, Nunez G. Mechanisms of inflammation-driven bacterial dysbiosis in the gut. Nature. 2017;10:18–26. 223