ABSTRACT BALLOU, ANNE LAEL. Probiotic Modulation of the Avian Microbiome Releases Systemic Signals That Alter the Lymphocyte

ABSTRACT

BALLOU, ANNE LAEL. Probiotic Modulation of the Avian Microbiome Releases Systemic Signals that Alter the Lymphocyte Transcriptome. (Under the direction of Dr. Matthew Koci).

The gastrointestinal (GI) microbiome plays an important role in animal development, health, and performance. Previous research in our laboratory has demonstrated probiotic supplementation alters host intestinal development, mucosal and systemic immune responses, and increases ATP production in circulating leukocytes. The ability to influence these diverse systems could be a powerful tool for promoting health and maximizing performance of food animals. However, to achieve this level of control I need to identify the specific members of the microbiome involved, understand where they reside in the host, and understand what governs their colonization. The goal of this dissertation research was to address this knowledge gap. Over the course of several distinct, yet interconnected studies I investigated the kinetics of microbial colonization of the chicken cecum from hatch to maturity, examined the impact of early exposure to distinct bacteria on long-term population dynamics, compared the microbial differences along the entire length of the gastrointestinal tract, studied role different diets have on the microbial populations in different regions of the gut, and developed an in vitro system to characterize probiotic-induced signals disseminated in the blood that regulate host immunity.

From these studies I learned the microbiome of growing chicks starts off with only a few members (Enterobacteriaceae), but develops rapidly, adding complexity and diversity.

First increasing the numbers of Firmicutes around day 7 until it reaches maturity by day 28.

As the microbiome develops, the influence of the treatments becomes more apparent; however, treatments that influence colonization early in life before the microbiome is established appear to have lost lasting effects. Additionally, taxonomic differences may not be the best predictor of biologically relevant changes in the microbiomes as predicted metagenomic content suggest fewer functional differences despite the strong taxonomic differences.

Evaluation of the crop, gizzard, duodenum, jejunum, ileum, and cecum of chickens demonstrated GI location was the strongest predictor of similarity among samples, influencing the microbiome more than diet or probiotic treatment. Of the locations evaluated, the cecum had the highest phylogenetic diversity, but the ileal microbiome had more treatment-related changes, and more predicted metagenomic differences. Further, the ileum of probiotic-fed animals had reductions in bacteria associated with inflammation and increased potential for production of anti-inflammatory butyrate. This suggests changes in ileal microbiota are more closely related to changes in host physiology, highlighting the need for a better understanding of how treatments influence changes in the microbiome along the

GI.

Finally, I hypothesized that previously observed changes in immune energy metabolism were regulated by a probiotic-stimulated factor in the serum. To test this I developed an in vitro cell culture assay and demonstrated serum isolated from probiotic-fed animals can augment the amount of ATP/cell in vitro. The specific serum factor still has not been identified; however, the results of this study suggest it is a compound under 30 kDa in size, lipid or associated with lipids, and based on the response of the evaluated cell lines, its effect is specific to immune cells. Transcriptomic analysis of an avian T-cell line (CU205) stimulated with serum from probiotic-treated animals demonstrates gene expression changes consistent with an increase in cell survival and T-helper cell differentiation as compared to the control. This research has added to our understanding of how probiotics alter the GI microbiome to affect host immune function. It indicates that probiotic-directed alteration of the ileal microbiome during development could result in the upregulation of a serum compound that stimulates circulating immune cells in a variety of ways. Continued study of this system will reveal how these ileal changes lead to the lymphocyte-stimulating serum signal, as well as other immune functions impacted by this probiotic.

by Anne Lael Ballou

A dissertation submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Functional Genomics

Raleigh, North Carolina

2017

APPROVED BY:

______Matthew Koci Hsiao-Ching Liu Committee Chair

______Hosni Hassan Jack Odle

BIOGRAPHY

Anne Ballou is a native of Louisville, Kentucky. She attended the University of Kentucky, where she received a B.S. in Animal Science, followed by an M.S. in Animal Science. After a brief internship with the Mote Marine Laboratory, she decided to pursue a Ph.D. in

Functional Genomics at North Carolina State University. This has allowed Anne to apply modern genomics principals to solving animal health challenges facing the food animal industry. She intends to continue in her support of food animal agriculture in her professional career and through service focused on agriculture and science policy.

ACKNOWLEDGMENTS

I would like to thank the many supporters, mentors, and collaborators who contributed to my growth and to the success of this research project.

 Dr. Matthew Koci guided and fostered my training, providing an example of tireless

and dedicated mentorship. None of this would have been possible without him.

 Drs. W. James Croom, Hosni Hassan, Hsiao-Ching Liu, and Jack Odle provided

valuable feedback, encouragement, and assistance during my research and writing.

 Rebecca-Ayme Hughes, Rizwana Ali, Mary Mendoza, Brian Troxell, Amy Savage,

Chris Ellis, and Aaron Oxendine assisted with the collection, analysis, and

interpretation of data.

 I was supported in part by NIFA National Needs Fellowship 2009-38420-05026,

NIFA Predoctoral Fellowship 2016-67011-25168, and a generous gift from Star-

Labs/Forage Research.

Countless others not mentioned also contributed their time and energy to support me during this process. Many thanks to all.

iii

TABLE OF CONTENTS

LIST OF TABLES ...... vii LIST OF FIGURES ...... viii CHAPTER 1: LITERATURE REVIEW ...... 1 INTRODUCTION ...... 2 ROLE OF MICROBES IN DEVELOPMENT OF THE GI ...... 3 Kinetics of gut colonization ...... 3 Impact of bacteria on GI morphology and growth...... 5 Mucosal immune development ...... 7 Microbial niches with in the gastrointestinal tract ...... 8 ROLE OF MICROBES IN FUNCTION AND HEALTH ...... 10 Interactions between diet, the microbiome, and digestive function ...... 10 Metabolism ...... 14 Mucosal immune activity ...... 15 Systemic immune activity ...... 17 PRIMALAC USE AS A MODEL FOR THE STUDY OF THE AVIAN MICROBIOME IN IMMUNE FUNCTION ...... 19 REFERENCES ...... 22 CHAPTER 2: DEVELOPMENT OF THE CHICK MICROBIOME: HOW EARLY EXPOSURE INFLUENCES FUTURE MICROBIAL DIVERSITY ...... 32 ABSTRACT ...... 33 INTRODUCTION ...... 34 MATERIAL AND METHODS ...... 36 Animals and Treatments ...... 36 Sample Collection...... 37 DNA isolation and 16S sequencing ...... 37 Sequence data analysis ...... 38 RESULTS...... 40 Microbiome composition and complexity change rapidly with age...... 40 Age more influential in microbiome development than treatment ...... 41 Treatments alter microbial composition and rate of development ...... 41

Treatment with live bacteria affects abundance of taxa not associated with treatment . 42 Treatment-induced changes in microbiome diversity lead to predicted changes in abundance of functional gene families ...... 43 DISCUSSION ...... 44 ACKNOWLEDGMENTS ...... 50 REFERENCES ...... 52 CHAPTER 3: DIFFERENTIAL EFFECT OF DIET AND PROBIOTICS ON MICROBIAL POPULATIONS ALONG THE AVIAN GASTROINTESTINAL TRACT...... 72 ABSTRACT ...... 73 IMPORTANCE ...... 74 INTRODUCTION ...... 75 MATERIALS AND METHODS ...... 77 Experimental Design ...... 77 Sample Collection...... 78 DNA Isolation and 16S Sequencing ...... 78 Sequence Data Analysis ...... 79 Accession numbers ...... 80 RESULTS...... 81 Evaluation of microbial profiles along the digestive tract...... 81 Diet and probiotic affect microbiome composition in multiple locations...... 83 Diet and probiotic interact to alter potential metagenomic content in the ileum...... 84 DISCUSSION ...... 85 Analysis of microbiome composition by location ...... 86 The ileal microbiome responds to diet and probiotic treatment ...... 88 The cecal microbiome responds to diet more than probiotic ...... 90 CONCLUSION ...... 92 ACKNOWLEDGMENTS ...... 93 REFERENCES ...... 94 CHAPTER 4: PROBIOTICS ALTER ENERGY CONSUMPTION AND GENE EXPRESSION IN CHICKEN LYMPHOCYTES ...... 113 ABSTRACT ...... 114

INTRODUCTION ...... 116 MATERIALS AND METHODS ...... 117 Animals ...... 117 Primary peripheral blood mononuclear cells (PBMC) isolation ...... 118 Serum collection and processing ...... 118 Cell lines ...... 120 Cellular ATP quantification in cell lines ...... 120 RNA Isolation and Sequencing ...... 121 Transcriptomic Analysis ...... 122 RESULTS AND DISCUSSION ...... 123 Circulating PBMC have more ATP/cell in PB-treated animals ...... 123 In vitro lymphocyte cell lines grown in serum from PB-treated animals have more ATP ...... 123 Physical and biochemical characterization of serum ...... 125 Stimulation of CU205 lymphocytes with serum from PB-treated animals alters gene expression ...... 128 CONCLUSIONS ...... 132 REFERENCES ...... 134 CHAPTER 5: FURTHER STUDIES...... 159 REFERENCES ...... 164

LIST OF TABLES

Table 1.1. Selected commercial probiotic products ...... 31 Table 2.1. Similarity percentage analysis (SIMPER) of treatment groups at day 14a...... 56 Table 2.2. Similarity percentage analysis (SIMPER) of treatment groups at day 28a...... 57 Table S2.1. Diet Composition ...... 58 Table S2.2. Number of samples included in analysis for each treatment and time-point ..... 58 Table S2.3. Taxa removed from data set prior to PIRCRUSt analysis, day 14 ...... 59 Table S2.4. Taxa removed from dataset prior to PICRUSt analysis, day 28 ...... 61 Table 3.1. Comparison of NRC-compliant Diets.a ...... 99 Table S3.1. P-values of treatment effects by location ...... 100 Table S3.2. P-values for location effects ...... 101 Table S4.1. Differentially expressed genes between day 2 CON and PB whole serum ...... 138

vii

LIST OF FIGURES

Figure 2.1. As the cecal microbiome develops, the dominant taxa shift from Gram negative to Gram positive bacteria...... 63 Figure 2.2. Age is the dominant factor in the composition of the microbiome...... 64 Figure 2.4. At both day 14 and day 28, the VP group shows the greatest divergence in predicted metagenomic content...... 69 Figure 2.5. Comparison of taxonomic and gene-group abundance trends at day 14 and day 28...... 71 Figure 3.1. GI Location is the primary variable by which microbial samples cluster...... 102 Figure 3.2. Phylogenetic diversity is greatest in the cecum...... 103 Figure 3.3. Location-specific communities respond to dietary and probiotic treatments. ... 104 Figure 3.4. Community composition changes distally, reflecting changes in input and function...... 106 Figure 3.5. “Candidatus Savagella” levels drop in the ileum of probtiotic-treated birds. ... 108 Figure 3.6. Samples cluster by location in both functional and 16S gene analysis...... 110 Figure 3.7. Treatments are associated with predicted functional changes in the ileum...... 111 Figure S3.1. Abundance of butanoate metabolism genes is increased in the PB supplement...... 112 Figure S4.1. Generation of complete serum samples from molecular weight fractions...... 148 Figure 4.1. Peripheral blood mononuclear cells (PBMC) collected from chickens fed a probiotic-supplemented diet have more ATP/cell than PBMC from control animals ...... 149 Figure 4.2. Avian lymphocyte cell lines increase ATP following stimulation by serum from probiotic-fed animals...... 150 Figure S4.2. Coomassie staining of SDS-PAGE-resolved mass fractionated and delipidated serum...... 152 Figure 4.3. Delipidation of serum from PB-treated animals results in lower ATP/cell relative to delipidated CON serum...... 153 Figure 4.4. PB-treated animal serum components below 30 kDa stimulate ATP production in CU205 cells...... 154 Figure 4.5. Serum from PB-treated animals alters gene expression of CU205 relative to serum from CON-treated animals...... 155 Figure 4.6. Differentially expressed genes between cells incubated in serum from CON- or PB-treated animals...... 156 Figure 4.7. Predicted functional changes lymphocytes following stimulation with serum from Probiotic-treated animals...... 158

viii

CHAPTER 1: LITERATURE REVIEW

INTRODUCTION

For centuries, people have used bacteria and their metabolic products in their fight against pathogens. In many ways civilization is based on bacterial fermentation.

Fermentation of foods and drinks, as well as silage for livestock, preserves foods that would otherwise be susceptible to spoilage and pathogen growth (1, 2). As our understanding of the microbial world grows, so does our appreciation of its importance in our nutrition and health.

Animals and humans alike are susceptible to many changes in their gut bacteria (3–6), and preserving a healthy equilibrium in the gut may reduce pathogen and stress-related morbidity.

However, there is a large gap in our understanding of the complex relationship between the microbiome and the host and to what extent that relationship can be manipulated to benefit the host. The benefits of understanding and controlling this relationship have tremendous promise for disease prevention, improved digestion and absorption, and may even extend to endocrine and nervous function (7, 8).

There are several popular treatments for supporting animal health and productivity, including probiotics, prebiotics, synbiotics, as well as enzymes and essential oils. Probiotics are live bacteria intended to confer some benefit to the host, though the nature of the claimed benefit varies by product. Prebiotics are nutrients that are unavailable to the host, but may be digested or used by the microbial community. They include saccharides such as mannan oligosaccharides, galactose oligosaccharides, or more complex nutrients like yeast cell walls

(9). Synbiotic products combine both probiotics and prebiotics. A full discussion of all these treatments is beyond the scope of this literature review, and the majority of this review will

focus on probiotics as a method for altering microbial populations. Most probiotics are orally administered bacterial or yeast products, commonly monocultures or mixed cultures of lactic-acid producing bacteria (LAB) such as Bacillus, Lactobacillus, Bifidobacterium, or

Enterococcus. See Table 1.1 for a selection of common probiotic products for a variety of species.

One of the major barriers to effective use of probiotics is the complexity of the gastrointestinal (GI) environment and its inhabitants. There are a variety of probiotic bacteria, with different mechanisms of action. They interact with their host through both direct contact via interactions with microbe-specific host receptors, as well as through their secreted proteins and metabolites. Microbes also affect their environment through consumption and release of dietary nutrients. The relative effectiveness of probiotic species is dependent on several factors, including host species, host health status, diet, and constituents of the established gut microbial populations. Our current understanding of the gut microbiome, its regulation and development, how it affects the host, and the role of probiotics as a driver of microbiome-mediated animal health, are explored in this review.

ROLE OF MICROBES IN DEVELOPMENT OF THE GI

Kinetics of gut colonization

In conventionally raised animals exposed to a diverse array of microbes from their environment and diet, the microbiome and GI develop in tandem from birth/hatch to maturity. Studies in mice reveal that use of antibiotics early in life hinder the development of microbial diversity and alter the composition of the microbiome extending out several

months beyond cessation of treatment (10). Metagenomic analysis of these altered microbial communities also revealed functional differences related to antibiotic treatment, suggesting that early bacterial exposure may have long-term impacts on GI function. Studies evaluating the impact of dietary factors such as fat and fiber show similar findings (11, 12).

The development of stable microbial populations in the avian gut occurs rapidly. In the chicken, the cecal microbiome appears to stabilize around day 21, with rapid changes occurring during the first 7-10 days of life. Oakley et al (2014) found that the ceca of newly hatched chicks were low in absolute numbers of bacteria, but they were proportionally high in Clostridium. By day 7 Flavonifractor, Pseudoflavonifractor, and a Lachnospiraceae were the dominant taxa, and day 21 was characterized by high Faecalibacterium. By day 42, the cecal microbiome was more diverse, with several abundant members, but no single dominant genus; day 42 bacteria included Faecalibacterium, Oscillospira, Dorea and Alistipes, all of which are Clostridiales or Bacteroidetes genera (13). Yin and associates found that normal colonization and inoculation with synthetic bacterial communities derived from chemostat- grown cecal populations both produced GI microbiota rich in Lachnospiraceae, Bacteroides,

Ruminococcaeae, and Prevotellaceae (14). In other studies, inoculation of chicks with chemostat-grown cecal microbes protected against Salmonella Typhimurium colonization better than the naturally developing microbiome (15, 16). Improved understanding of how the GI and microbiome develop, and how probiotics, antibiotics, and other microbiome- altering treatments affect long-term composition will be important in promoting and maintaining GI health.

Impact of bacteria on GI morphology and growth.

There are numerous physiological features of gut development that are aberrant or absent in gnotobiotic animals, and may therefore be susceptible to manipulation by probiotics. Germ-free mice show enlarged ceca, and several histomorphometric differences from their conventional counterparts (17). Absence of gut microbes reduces villus surface area by 30% (18). On the cellular level, this reduction in villus size could be due to slower division in intestinal crypts leading to fewer crypt cells overall. Banasaz et al. determined that establishing a monoculture of Lactobacillus rhamnosus GG in germ-free mice was enough to restore or exceed normal cell division and crypt number (19, 20).

Supplementation with a probiotic including Lactobacillus acidiophilus and Lactobacillus casei increased villus height and size in conventionally raised poultry (21).

Barrier function may also be adversely affected by abnormal colonization;

Bacteroides thetaiotaomicron, E. coli, as well as several Lactobacillus species, have been shown to strengthen tight junctions through the maintenance of desmosomes and occludins, upregulating ZO-1 and ZO-2 proteins (22, 23). While poor barrier integrity can increase passive uptake of some nutrients, it can also result in loss of electrochemical gradient, ions, water, and other nutrients if the concentration gradient is unfavorable.

Certain probiotics may alleviate pathogen growth and inflammation at the mucosa by modulating the viscosity of digesta (24, 25). Rye and wheat increase the viscosity of intestinal contents due to their high non-starch polysaccharide (NSP) content. This can result in overgrowth of opportunistic bacteria such as Clostridium perfringens, increasing mucosal inflammation and junction leakage (26, 27). This high viscosity also has implications for

whole animal development and health as it impairs absorption of vitamins and minerals enough to cause bone mineralization problems leading to leg deformities in turkeys and broiler chickens (25). Several microbial species have enzymes capable of breaking down these NSP, decreasing digesta viscosity and resolving mucosal inflammation, malabsorption- related bone mineralization problems (24, 25).

Another aspect of GI physiology affected by microbial colonization is the production of mucus, particularly mucin-2. In conventionally raised animals, the mucus layer provides protection to enterocytes, and helps filter many types of bacteria while allowing for efficient movement of nutrients to the apical surface (28, 29). Germ-free mice and chicks have fewer goblet cells and a thinner mucus layer (17, 30). Addition of a probiotic consortium of lactic acid bacteria to the diet resulted in a thicker and more complete mucus layer (31).

Microbiome-related changes in digestion-association gene expression during development may have long-term implications in GI function. For example, inoculation of newly hatched chicks with cecal derived microbial populations alters gene expression in the ileum relative to conventionally colonized birds (14). Genes upregulated included many related to nutrient and ion transport, as well as genes involved in WNT signaling, a major developmental regulator. This suggests a role for gut bacteria in the development of mature

GI physiology and function. Other studies have shown a variety of probiotics administered early in life can stimulate immune gene expression and humoral immune activity (32–34).

Mucosal immune development

Much of what we know regarding the role of bacteria in GI immune development comes from gnotobiotic mice. Mouse studies dating back over 40 years demonstrate that certain gut colonizers may be necessary for normal gut-associated lymphoid tissue (GALT) development. For example, Peyer’s patches, aggregated lymphoid nodules in the intestine, are smaller in gnotobiotic models (35, 36). Monocolonization of mice with Bifidobacterium infantis restored normal IgE-mediated oral tolerance relative to gnotobiotic controls (37).

Some aspects of GALT development may be time sensitive. Reconstitution of young gnotobiotic mice’s microbiome restores normal oral tolerance, while reconstitution in adults does not (37).

Postnatal or posthatch colonization of the gut also appears to be important for generating a balance between pro- and anti-inflammatory immune responses in the gut. This balance is necessary for homeostasis between the host and its resident microbes (38, 39).

Few studies have evaluated the immune development in germ-free chicks, but these support the general findings of similar mouse studies. GALT development is stunted in the absence of microbes (40), and serum IgG and IgA levels were much lower in gnotobiotic chicks than in their conventional counterparts, as were intestinal immunoglobulin producing cells (41).

Studies evaluating the effect of probiotic supplementation in poultry support the ability of the chicken microbiome to affect gene expression of immune genes in the gut. In the ileum, anti-inflammatory IL-10 increased in the probiotic group, while pro-inflammatory

IL-12RB2, and NFKB decreased (34). Further, evaluation of intra-epithelial lymphocytes

(IEL) in the duodenum and jejunum of broilers fed the same probiotic from day of hatch

revealed higher proportions of several T-cell subsets, potentially conferring increased protection against intestinal pathogens (42).

Microbial niches with in the gastrointestinal tract

The role of location-specific GI physiology in development of a stable microbial community is not yet well understood. In the chicken, as in all animals with substantial hindgut fermentation, the cecum houses the most diverse and dense microbial population.

The small intestine microbial population is less diverse than the cecum, but more susceptible to dietary influences and passage rate. It contains fewer bacteria per gram of contents than a normal cecum (43). However, as the small intestine is the major absorptive region (44), changes in microbial populations could have massive impacts on host function and development. Research into the importance of bacterial composition in these locations tends to rely on a binary model involving research comparing and contrasting gnotobiotic and conventional animals. To best utilize the microbiome to promote health, it is vital to understand the contribution of individual taxa to the whole community.

Few high-throughput microbiome studies in poultry have evaluated populations outside the cecum, but this is beginning to change. Ranjitkar et al (2016) evaluated the avian

GI microbiome across both age and intestinal location. The broad conclusion was that

Lactobacillus was the dominant taxon in all locations except the cecum from 8 days of age until 36 days of age. The crop was almost exclusively populated by Lactobacillus, while the gizzard contained about 60% Lactobacillus. Other gizzard species include several

Clostridiaceae taxa, and variable levels of Proteobacteria and Actinobacteria. Despite being

seen in other chicken studies, Erysipelotrichales, Actinobacteria, and Bacillales were low or absent in the ileum; this suggests variability of the normal small intestinal microbiome or alternatively potential differences due to sampling methodologies (43, 45). In the cecum,

Lachnospiraceae, Ruminococcaceae, and Rikenellaceae were dominant taxa. These are commonly found taxa in the chicken microbiome, though their abundance appears to vary from study to study (13, 45, 46). Lachnospiraceae declined in abundance from day 8 to day

21, while Rikenellaceae abundance increased, primarily through the genus Alistipes. Since this study did not evaluate multiple small intestinal locations but instead used the ileum to represent the entire small intestine, the duodenal and jejunal microbiomes are not well- characterized (47).

In addition to longitudinal differences in the GI microbiome, transverse analysis of the microbiome demonstrates that the bacteria associated with the mucosa can be very different from the feed or lumen associated population (48, 49). In both the small intestine and colon of rhesus macaques, the mucosa-associated microbial community is enriched in aerobic and facultative anaerobic species, while the luminal community is higher in obligate anaerobes (49). Additionally, the mucosa has higher levels of gram negative taxa, while the lumen is high in Firmicutes (50). With the increasing number of chicken-focused microbiome studies, meta-analysis will soon allow scientists to gain a more complete picture of the normal composition of the avian microbiome across geographical regions, breeds, ages, and diets. With this information, we will be better able to correlate experimental variables with changes in microbiome composition/function and host function.

ROLE OF MICROBES IN FUNCTION AND HEALTH

Interactions between diet, the microbiome, and digestive function

The diet can both provide a steady supply of microbes for colonization as well as drive microbiome composition through nutrient availability. Different feed components carry with them the bacteria native to their original environment, and any they have incorporated during processing and transport. However, the presence or absence of preferred nutrients will strongly impact which ingested and commensal populations thrive and which are excluded from the GI ecosystem (51, 52). For example, nondigestible starch supports the growth of fermentative bacteria in multiple host species, including swine and humans (53,

54). The byproducts of starch fermentation by these bacteria are often beneficial to the host, and include short chain fatty acids (SCFA), which can have anti-pathogenic and anti- inflammatory effects (55, 56). These SCFA are also thought to stimulate the release of enteroendocrine molecules such as peptide YY, and glucagon-like peptide 1, involved in satiety (52).

Several studies in both humans and mice show the effects of dietary changes on microbiome composition (11, 57, 58). These studies typically present comparisons between diets that differ significantly in energy level or composition. For example, humans eating either isocaloric vegan or meat+dairy diets displayed substantial changes in their microbiome composition relative to each other and to the subjects’ baseline profiles (57). These changes extended beyond phylogenetic profile and into the functional profile of the gut; secondary bile salts (a product of microbial metabolism) were increased in the animal diet group, and several microbial markers of gut inflammation were higher, as well (11).

As dietary carbohydrates are the primary energy source of the chicken, as well as its microbiome, the impact of carbohydrate types and inclusion levels on microbial composition is of considerable interest. Barley, rye and wheat are higher in non-starch polysaccharides

(NSP) than corn. NSP are correlated with growth of Clostridium perfringens, a causative agent in necrotizing enteritis. The inclusion of the probiotic Lactobacillus reuteri expressing heterologous β-glucanase has been reported to break down the NSP in a barley-based diet, improving feed efficiency and potentially removing one of the primary etiological factors in necrotizing enteritis (59). Another study showed few effects of corn:wheat ratios and ensiling on microbial composition or bile salt profile in broiler chickens (47). Addition of short and medium-chain fatty acids to the diet of broilers had no effect on microbiome composition in the cecum at 7, 21, and 42 days (13), but it is possible that these additives are consumed prior to reaching the cecal microbiome.

Few studies have evaluated the interactions between dietary fat and protein sources and the microbiome in poultry, but there are indications that these macronutrients play a role in microbial composition, as well. Replacing varying levels of soy bean meal with fermented cottonseed meal increases levels of Lactobacilli while decreasing coliform bacteria in the cecum (60). Comparison of the guts of chickens fed either soy protein concentrate or fishmeal reveal higher C perfringens growth in fishmeal groups. The fishmeal diet was shown to have higher levels of glycine and methionine, and may support the growth of C. perfringens more than the amino acid profile of soy-based diets (61). Evaluation of the impact of dietary fat source (soy vs lard and tallow) indicates that use of soy fat over animal based fat may reduce levels of C. perfringens (62). Studies in swine with dietary poly-

unsaturated fatty acids (PUFA) demonstrate they can alter the microbiome (63), however a similar study in chickens did not showed minimal changes to the microbiome (64). More research is necessary to determine the role of proteins and lipids on microbiome composition, and whether variations in these macronutrients may impact host function through the microbiome.

It is not surprising that changes in dietary composition leads to shifts in the microbial composition. Fiber, protein, fat, vitamins and minerals all demand different digestive strategies from the host that may change the luminal environment (65–67). Not only can diet affect host GI physiology and through it the microbiome, the GI microbiome exerts an effect on the host through the products of bacterial digestion and metabolism. Hungate and others have studied the role of foregut and hindgut bacteria in digestion of resistant starch and cellulose for decades (68–70). Not only do GI bacteria release certain nutrients for host consumption, they can stimulate the production of host factors such as enzymes and bile salts

(47, 71–73).

Several Lactobacillus species were evaluated singly or in combination, for their ability to alter secreted amylase, protease, and lipase activity in the GI. Of 12 strains tested, all combinations and dosages increased amylase activity, but none measurably affected lipase or protease activity. This increase in amylase activity was accompanied by improved feed efficiency, and a higher growth rate (72). A similar study testing Bacillus coagulans NJ0516 also reported increased amylase and protease activity in the duodenum relative to control fed animals (73).

Ranjitkar et al (2016) demonstrated that increasing levels of Lactobacillus salivarius and various clostridial species increase the rate of bile acid deconjugation in poultry, a process that may impact lipid absorption in the ileum. Degirolama et al (2014) demonstrated that supplementation with the probiotic consortium VSL#3 stimulated bile acid levels in the lumen of mice, as well as bile salt hydrolase activity, responsible for deconjugation of bile acids (71). The mechanism involved in increased bile acid secretion in this case appears to be VSL#3-related downregulation of the enterohepatic farnesoid X receptor-fibroblast growth factor 15 (FXR-FGF15) pathway (71).

There is a lack of studies evaluating the role of probiotic treatments and microbiome composition on expression of other digestive enzymes in poultry. However, decreased brush border enzyme activity in Eimeria-infected chickens does demonstrate that microbes are capable of altering enzyme expression in poultry (74), but to what extent commensal and probiotic organisms routinely do so is still unknown. In other species, there is considerable evidence that microbial colonization affects the expression and density of vital brush-border membrane enzymes. This is particularly apparent in mammals, where transition from milk to solid foods is often preceded by a shift in both microbial composition and levels of lactase, sucrase, gamma-glutamyltranspeptidase, and other digestive enzymes (75). These changes are accompanied by a shift away from facultative anaerobes such as E coli and streptococci toward obligate anaerobes that characterize more mature microbiomes (Bacteroides and

Clostridium) (76). Upregulation of digestive enzymes through bacterial stimulation could improve digestive efficiency. As several probiotic species improve feed efficiency in poultry

through unknown mechanisms (72, 77, 78), the impact of probiotics and other gut microbes on host enzyme activity should be explored further.

Metabolism

Poultry scientists are just beginning to identify microbial mediators of host gene and protein expression. For example, supplementation of broilers with Enterococcus faecium, a popular probiotic in poultry products, is associated with improved meat quality and yield in broilers, potentially through changes in the muscle and hepatic proteomes; several antioxidants were increased in these tissues, as were genes related to glycolysis and other energy metabolism pathways (79, 80).

In mouse studies, germ-free animals are known to be resistant to diet-induced obesity, requiring more nutrients than conventional counterparts do (81). This cannot entirely be accounted for by the loss of microbial enzymes for the breakdown of carbohydrates not otherwise available to the host. Germ-free animals on a high-fat, high-sugar diet show similar difficulties in efficiently utilizing the available nutrients (82). Studies in humans and mice revealed that certain microbial profiles increase circulating levels of leptin, glucose, and insulin, and hepatic lipogenesis increases following conventionalization (81, 83). Further research into these phenomena will improve our understanding of how to use GI microbiomes to impact host functions of interest.

Mucosal immune activity

Multiple studies have described how altering the microbiome affects GALT, while others have focused on how GALT regulate the microbiome and prevent invasion (33, 41,

42, 84, 85). The relationship between host immune tissues and GI microbes is multifactorial and sensitive to a number of genetic and environmental factors.

Many of the interactions between host and microbial cells are mediated by host receptors that recognize a broad array of microbial cell components (microbe-associated molecular patterns; MAMPs), such as flagellin, glycoproteins, and lectins (86). These pattern recognition receptors (PRRs) are expressed on a variety of cell types, particularly antigen-presenting immune cells like dendritic cells and macrophages, and some enterocytes; they include groups of receptors such as toll-like receptors (TLR), nod-like receptors (NLR), and C-type lectin receptors (86, 87). The signaling pathways activated by interactions between MAMPs and PRRs can be activating or inhibitory, and lead to a variety of different immune responses (88). There is growing evidence that some commensal and symbiotic bacteria have MAMPs that promote a tolerant host phenotype, and some have termed these molecular patterns SAMPs (symbiotic-associated molecular patterns) (89). Bacteriodes fragilis polysaccharide A, for example, interacts with TLR2 on regulatory T-cells (Tregs) to promote tolerance and inhibit TH17 responses (85, 89).

Other bacteria, particularly Clostridia species, are able to stimulate the development of Tregs as well (84, 85, 89, 90), while attachment of segmented filamentous bacteria (SFB), notably Candidatus Savagella, in the ileum has been shown to stimulate differentiation of pro-inflammatory TH17 type cells (91). Colonization of germ-free mice with a monoculture

of SFB is enough to induce autoimmune encephalomyelitis in susceptible models (5, 92), and persistent TH17 responses are associated with autoimmune disorders such as multiple sclerosis, rheumatoid arthritis, and inflammatory bowel disease (93). Addition of probiotic bacteria to the chicken microbiome alters gene expression in both the ileum and the cecum: in the ileum, a combination of probiotic species increased anti-inflammatory IL-10, while pro-inflammatory IL-12RB2 and NF-κB decrease (34). Research with Lactobacillus acidiophilus increased expression of interferon α (IFN-α), IFN-γ, IL-18, MyD88, and other immune-related genes in the cecal tonsils of chickens (94).

These changes are consistent with an increased TH2 response in the ileum, but an increased TH1 response in the cecum. The functional results of this may be more IgA- mediated immunity in the ileum, while T-cell mediated and pro-inflammatory immune activity increases in the cecum. In support of this, several researchers have demonstrated that probiotic supplementation increases IgA in the small intestine (34, 95, 96). In contrast,

Crhanova et al (2011) demonstrated that Salmonella Enteritidis infection shifted cecal gene expression from a TH1 phenotype characterized by IFN-γ and iNOS to a TH17 response characterized by high levels of IL-17 (97).

There are many other examples of probiotic bacteria activating or inhibiting expression of major regulators and cytokines of innate immunity, such as NF-κB, Akt, interleukin 10 (IL-10), IL-12, transforming growth factor β (TGF-β), tumor necrosis factor

(TNF), and others (98–101). The varied signaling pathways involved in host response to probiotics and other microbes even include recognition of Lactobacillus gasseri nucleotides by murine B-cells (102). The different response to these treatments seen between the ileum

and cecum suggests that each intestinal location may respond independently and even contradictorily to microbes.

Systemic immune activity

The immune system mounts robust humoral (antibody) responses to bacterial proteins, and in recent years, it has become apparent that some bacteria are capable of exerting an adjuvant-like effect. Studies in chickens have shown that Lactobacillus and

Bifidobacterium-based probiotics can promote antigen-specific humoral immune responses against other antigens (33, 34). Additionally, chicks stimulated with a commercial probiotic containing Lactobacillus acidophilus, Bifidobacterium bifidum, and Streptococcus faecalis increased production of anti-C. perfringens natural antibodies in the blood and mucosa (103).

These antibodies are generally reactive to a broad range of antigens, and provide protection against certain pathogens prior to maturation of the humoral response in young birds (104).

It is thought that the B-cell subset responsible for producing natural antibodies (B-1 cells) are stimulated by commensal and probiotic bacteria through TLR and other PRRs, independent of the normal T-cell mediated pathways (96).

The ability of bacteria to stimulate immune responses against other microbes is not limited to the humoral immune response. Hand et al (2012) showed that Toxoplasma gondii infection in mice stimulates the production of CD4+ T-cells that develop into long-lived memory cells. These populations were not only reactive to T gondii antigens; T-cell populations were generated against antigens derived from the commensal population, providing an avenue for temporary infections to play a long-term role in the composition of

the microbiome (105). There are other examples of the microbiome and microbial treatments affecting cell-mediated immune responses, both in the mucosa and systemically (106, 107).

For example, feeding broiler chickens a Lactobacillus-based probiotic consortium prior to challenging with Eimeria acervulina resulted in more intra-epithelial T-cells in the duodenum and jejunum and fewer oocytes than in control animals (42). Lactobacillus and

Saccharomyces have been shown to increase CD4 and CD8 T-cell numbers as a proportion of total lymphocytes in the jejunum and duodenum, but do not stimulate B-cells in the same manner (77).

Colonic Tregs increase in mice given Clostridium species, suggesting indigenous

Clostridia may be necessary for the generation and maintenance of immune tolerance toward commensals. Further, these Clostridia are associated with high levels of TGF-β in the lamina propria, an important factor in the generation of colonic Tregs (90). Lactobacillus

(reuteri and casei), and Bifidobacterium breve stimulate Treg differentiation and proliferation in the murine GI through different contact-dependent mechanisms involving intraepithelial dendritic cells (99, 108).

Colonic murine Tregs can also be stimulated in both germ free and conventional models with SCFA. As major byproducts of anaerobic fermentation by many hindgut bacteria, SCFA play a role in colonic/cecal pH and serve as an energy source to enterocytes.

Feeding poultry prebiotic polysaccharides and Bacillus cereus increases both antibody titer against Newcastle disease virus, and circulating T-cell numbers at 21 and 42 days of age (109). Newcastle disease virus, primarily a respiratory virus, is an example of a

systemic disease that may be mitigated by preventative treatments aimed at the intestinal microbiome.

Though the impact of probiotics on avian natural killer (NK) and natural killer T-cells

(NKT) has not been evaluated, these cell types are reactive to microbial signals in other model species, and represent other immune cell types outside the GI that respond to changes in the GI microbiome. NK-cells are largely a part of the innate immune response, and are responsible for the cytotoxic killing of tumorous and virus-infected cells. NKT-cells are often seen as bridges between innate and active cell-mediated immunity, as they express a limited number of T-cell receptors with an affinity for glycolipids (110). NKT-cells increase in colonic and lung mucosal tissues in germ free mice, contributing to inflammatory bowel disease and allergic asthma relative to conventionally colonized animals (111). The accumulation of NKT-cells in these tissues can be prevented by neonatal colonization, but the proinflammatory NKT-driven response cannot be rescued by colonizing adult mice (111).

In conventional mice, oral inoculation with Butyricum fibrisolvens MDT-1 increases both

NK and NKT-cells in the spleen by 3.5 and 4-fold, respectively (112).

PRIMALAC USE AS A MODEL FOR THE STUDY OF THE AVIAN MICROBIOME

IN IMMUNE FUNCTION

The goal of this PhD project was to better understand how the avian gut microbiome influences immune function. A probiotic product, PrimaLac (Lactobacillus acidophilus,

Lactobacillus casei, Bifidobacterium bifidum, Enterococcus faecium), has been used to drive these changes. PrimaLac supplementation has been shown to affect the physiology of the

developing avian GI in several ways. Villus height and perimeter were increased, and the number of observed goblet cells in was increased, as well (21). In addition to these morphological effects, it was observed that the mucous layer appeared to be more consistent in probiotic-treated animals, and there were fewer SFB attached to the ileal surface. Further evaluation of the probiotic-fed GI showed a number of gene expression changes in the ileum.

The probiotic increased IL-10 mRNA in the ileum and reduced IL-12RB2, NFκB, CD28

(34). These and other changes in transcription implicate the IL-27 pathway, and an increase in TH2 activity over TH1 (113, 114).

Along with gene expression changes in the GI, supplementation of broiler chickens with PrimaLac resulted in a faster humoral response to an antigen challenge. Chickens immunized with sheep red blood cells (SRBC) had detectable anti-SRBC serum IgG 7 days before immunized control animals (34). Intestinal IgA also increased following probiotic supplementation. Though there was no change in whole body energy utilization between treatment groups as measured by O2 consumption, probiotic-fed animals used less O2 in the ileum and while increasing O2 consumption in the thymus (32, 34). Additionally, their circulating immune cells have been shown to produce and consume more adenosine triphosphate (ATP) than the cells of untreated animals (34).

These results demonstrate the potential for the microbiome to interact with systemic host immune function, but to fully understand the mechanisms involved we need a better understanding of the avian microbiome, how it develops, how the microbiome differs by GI locations, and how each of these are impacted by manipulation by probiotic and/or other dietary treatments. This dissertation reports the progress made in understanding how a

probiotic alters the avian microbiome and changes host lymphocyte gene expression. This and future studies will enable us to improve animal health by manipulating the microbiome.

REFERENCES

1. McGovern PE, Zhang J, Tang J, Zhang Z, Hall GR, Moreau RA, Nuñez A, Butrym ED, Richards MP, Wang C, Cheng G, Zhao Z, Wang C. 2004. Fermented beverages of pre- and proto-historic China. Proc Natl Acad Sci USA.

2. Miles M. 1895. Silos, Ensilage and Silage: A Practical Treatise on the Ensilage of Fodder Corn. Orange Judd Company.

3. Fung I, Garrett JP-D, Shahane A, Kwan M. 2012. Do Bugs Control Our Fate? The Influence of the Microbiome on Autoimmunity. Curr Allergy Asthma Rep 12:511–519.

4. Ezendam J, van Loveren H. 2008. Lactobacillus casei Shirota administered during lactation increases the duration of autoimmunity in rats and enhances lung inflammation in mice. Br J Nutr 99:83–90.

5. Wu H-J, Ivanov II, Darce J, Hattori K, Shima T, Umesaki Y, Littman DR, Benoist C, Mathis D. 2010. Gut-Residing Segmented Filamentous Bacteria Drive Autoimmune Arthritis via T Helper 17 Cells. Immunity 32:815–827.

6. Ahn J, Sinha R, Pei Z, Dominianni C, Wu J, Shi J, Goedert JJ, Hayes RB, Yang L. 2013. Human gut microbiome and risk for colorectal cancer. J Natl Cancer Inst 105:1907–1911.

7. Neuman H, Debelius JW, Knight R, Koren O. 2015. Microbial endocrinology: the interplay between the microbiota and the endocrine system. FEMS Microbiol Rev.

8. Lyte M. 2013. Microbial Endocrinology in the Microbiome-Gut-Brain Axis: How Bacterial Production and Utilization of Neurochemicals Influence Behavior. PLoS Pathog 9.

9. Santin E, Maiorka A, Macari M, Grecco M, Sanchez JC, Okada TM, Myasaka AM. 2001. Performance and Intestinal Mucosa Development of Broiler Chickens Fed Diets Containing Saccharomyces cerevisiae Cell Wall. J Appl Poult Res 10:236–244.

10. Nobel YR, Cox LM, Kirigin FF, Bokulich NA, Yamanishi S, Teitler I, Chung J, Sohn J, Barber CM, Goldfarb DS, Raju K, Abubucker S, Zhou Y, Ruiz VE, Li H, Mitreva M, Alekseyenko AV, Weinstock GM, Sodergren E, Blaser MJ. 2015. Metabolic and metagenomic outcomes from early-life pulsed antibiotic treatment. Nat Commun 6.

11. 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. 2014. Diet rapidly and reproducibly alters the human gut microbiome. Nature 505:559–563.

12. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, Gordon JI. 2006. An obesity- associated gut microbiome with increased capacity for energy harvest. Nature 444:1027–31.

13. Oakley BB, Buhr RJ, Ritz CW, Kiepper BH, Berrang ME, Seal BS, Cox NA. 2014. Successional changes in the chicken cecal microbiome during 42 days of growth are independent of organic acid feed additives. BMC Vet Res 10.

14. Yin Y, Lei F, Zhu L, Li S, Wu Z, Zhang R, Gao GF, Zhu B, Wang X. 2009. Exposure of different bacterial inocula to newborn chicken affects gut microbiota development and ileum gene expression. ISME J 4:367–376.

15. Corrier DE, Nisbet DJ, Hollister AG, Scanlan CM, Hargis BM, Deloach JR. 1993. Development of Defined Cultures of Indigenous Cecal Bacteria to Control Salmonellosis in Broiler Chicks. Poult Sci 72:1164–1168.

16. Nisbet DJ, Corrier DE, DeLoach JR. 1993. Effect of Mixed Cecal Microflora Maintained in Continuous Culture and of Dietary Lactose on Salmonella typhimurium Colonization in Broiler Chicks. Avian Dis 37:528–535.

17. Sommer F, Bäckhed F. 2013. The gut microbiota — masters of host development and physiology. Nat Rev Microbiol 11:227–238.

18. Gordon HA, Bruckner-Kardoss E. 1961. Effect of normal microbial flora on intestinal surface area. Am J Physiol 201:175–178.

19. Alam M, Midtvedt T, Uribe A. 1994. Differential Cell Kinetics in the Ileum and Colon of Germfree Rats. Scand J Gastroenterol 29:445–451.

20. Banasaz M, Norin E, Holma R, Midtvedt T. 2002. Increased enterocyte production in gnotobiotic rats mono-associated with Lactobacillus rhamnosus GG. Appl Environ Microbiol 68:3031–3034.

21. Chichlowski M, Croom WJ, Edens FW, McBride BW, Qiu R, Chiang CC, Daniel LR, Havenstein GB, Koci MD. 2007. Microarchitecture and spatial relationship between bacteria and ileal, cecal, and colonic epithelium in chicks fed a direct-fed microbial, PrimaLac, and salinomycin. Poult Sci 86:1121–1132.

22. Hooper LV, Wong MH, Thelin A, Hansson L, Falk PG, Gordon JI. 2001. Molecular Analysis of Commensal Host-Microbial Relationships in the Intestine. Science 291:881–884.

23. Ulluwishewa D, Anderson RC, McNabb WC, Moughan PJ, Wells JM, Roy NC. 2011. Regulation of Tight Junction Permeability by Intestinal Bacteria and Dietary Components. J Nutr 141:769–776.

24. Latorre JD, Hernandez-Velasco X, Kogut MH, Vicente JL, Wolfenden R, Wolfenden A, Hargis BM, Kuttappan VA, Tellez G. 2014. Role of a Bacillus subtilis Direct-Fed Microbial on Digesta Viscosity, Bacterial Translocation, and Bone Mineralization in Turkey Poults Fed with a Rye- Based Diet. Front Vet Sci 1.

25. Tellez G, Latorre JD, Kuttappan VA, Kogut MH, Wolfenden A, Hernandez-Velasco X, Hargis BM, Bottje WG, Bielke LR, Faulkner OB. 2014. Utilization of rye as energy source affects bacterial translocation, intestinal viscosity, microbiota composition, and bone mineralization in broiler chickens. Front Genet 5.

26. Cooper KK, Songer JG. 2009. Necrotic enteritis in chickens: A paradigm of enteric infection by Clostridium perfringens type A. Anaerobe 15:55–60.

27. McClane BA. 2001. The complex interactions between Clostridium perfringens enterotoxin and epithelial tight junctions. Toxicon: Off J Int Soc Toxinology 39:1781–1791.

28. Moncada DM, Kammanadiminti SJ, Chadee K. 2003. Mucin and Toll-like receptors in host defense against intestinal parasites. Trends Parasitol 19:305–311.

29. Sherman P, Fleming N, Forstner J, Roomi N, Forstner G. 1987. Bacteria and the mucus blanket in experimental small bowel bacterial overgrowth. Am J Pathol 126:527.

30. Cheled-Shoval SL, Gamage NSW, Amit-Romach E, Forder R, Marshal J, Kessel AV, Uni Z. 2014. Differences in intestinal mucin dynamics between germ-free and conventionally reared chickens after mannan-oligosaccharide supplementation. Poult Sci 93:636–644.

31. Johansson MEV, Larsson JMH, Hansson GC. 2011. The two mucus layers of colon are organized by the MUC2 mucin, whereas the outer layer is a legislator of host–microbial interactions. Proc Natl Acad Sci USA 108:4659–4665.

32. Chichlowski M, Croom J, McBride BW, Daniel L, Davis G, Koci MD. 2007. Direct-Fed Microbial PrimaLac and Salinomycin Modulate Whole-Body and Intestinal Oxygen Consumption and Intestinal Mucosal Cytokine Production in the Broiler Chick. Poult Sci 86:1100–1106.

33. Haghighi HR, Gong J, Gyles CL, Hayes MA, Sanei B, Parvizi P, Gisavi H, Chambers JR, Sharif S. 2005. Modulation of Antibody-Mediated Immune Response by Probiotics in Chickens. Clin Diagn Lab Immunol 12:1387–1392.

34. Qiu R, Croom J, Ali RA, Ballou AL, Smith CD, Ashwell CM, Hassan HM, Chiang C-C, Koci MD. 2012. Direct fed microbial supplementation repartitions host energy to the immune system. J Anim Sci 90:2639–2651.

35. Maeda Y, Noda S, Tanaka K, Sawamura S, Aiba Y, Ishikawa H, Hasegawa H, Kawabe N, Miyasaka M, Koga Y. 2001. The Failure of Oral Tolerance Induction is Functionally Coupled to the Absence of T Cells in Peyer’s Patches under Germfree Conditions. Immunobiology 204:442–457.

36. Pollard M, Sharon N. 1970. Responses of the Peyer’s Patches in Germ-Free Mice to Antigenic Stimulation. Infect Immun 2:96–100.

37. Sudo N, Sawamura S, Tanaka K, Aiba Y, Kubo C, Koga Y. 1997. The requirement of intestinal bacterial flora for the development of an IgE production system fully susceptible to oral tolerance induction. J Immunol Baltim Md 1950 159:1739–1745.

38. Eberl G. 2005. Inducible lymphoid tissues in the adult gut: recapitulation of a fetal developmental pathway? Nat Rev Immunol 5:413–420.

39. Bouskra D, Brézillon C, Bérard M, Werts C, Varona R, Boneca IG, Eberl G. 2008. Lymphoid tissue genesis induced by commensals through NOD1 regulates intestinal homeostasis. Nature 456:507–510.

40. Thorbecke GJ, Gordon HA, Wostman B, Wagner M, Reyniers JA. 1957. Lymphoid Tissue and Serum Gamma Globulin in Young Germfree Chickens. J Infect Dis 101:237–251.

41. Parry SH, Allen WD, Porter P. 1977. Intestinal immune response to E. coli antigens in the germ-free chicken. Immunology 32:731.

42. Dalloul RA, Lillehoj HS, Shellem TA, Doerr JA. 2003. Enhanced mucosal immunity against Eimeria acervulina in broilers fed a Lactobacillus-based probiotic. Poult Sci 82:62–66.

43. Oakley BB, Lillehoj HS, Kogut MH, Kim WK, Maurer JJ, Pedroso A, Lee MD, Collett SR, Johnson TJ, Cox NA. 2014. The chicken gastrointestinal microbiome. FEMS Microbiol Lett 360:100–112.

44. Engberg RM, Hedemann MS, Jensen BB. 2002. The influence of grinding and pelleting of feed on the microbial composition and activity in the digestive tract of broiler chickens. Br Poult Sci 43:569–579.

45. Wei S, Morrison M, Yu Z. 2013. Bacterial census of poultry intestinal microbiome. Poult Sci 92:671–683.

46. Danzeisen JL, Kim HB, Isaacson RE, Tu ZJ, Johnson TJ. 2011. Modulations of the Chicken Cecal Microbiome and Metagenome in Response to Anticoccidial and Growth Promoter Treatment. PLoS ONE 6:e27949.

47. Ranjitkar S, Lawley B, Tannock G, Engberg RM. 2016. Bacterial Succession in the Broiler Gastrointestinal Tract. Appl Environ Microbiol 82:2399–2410.

48. Gong J, Forster RJ, Yu H, Chambers JR, Sabour PM, Wheatcroft R, Chen S. 2002. Diversity and phylogenetic analysis of bacteria in the mucosa of chicken ceca and comparison with bacteria in the cecal lumen. FEMS Microbiol Lett 208:1–7.

49. Yasuda K, Oh K, Ren B, Tickle TL, Franzosa EA, Wachtman LM, Miller AD, Westmoreland SV, Mansfield KG, Vallender EJ, Miller GM, Rowlett JK, Gevers D, Huttenhower C, Morgan XC. 2015. Biogeography of the Intestinal Mucosal and Lumenal Microbiome in the Rhesus Macaque. Cell Host Microbe 17:385–391.

50. Chen W, Liu F, Ling Z, Tong X, Xiang C. 2012. Human Intestinal Lumen and Mucosa- Associated Microbiota in Patients with Colorectal Cancer. PLOS ONE 7:e39743.

51. Belanche A, Doreau M, Edwards JE, Moorby JM, Pinloche E, Newbold CJ. 2012. Shifts in the Rumen Microbiota Due to the Type of Carbohydrate and Level of Protein Ingested by Dairy Cattle Are Associated with Changes in Rumen Fermentation. J Nutr 142:1684–1692.

52. Haenen D, Zhang J, Silva CS da, Bosch G, Meer IM van der, Arkel J van, Borne JJGC van den, Gutiérrez OP, Smidt H, Kemp B, Müller M, Hooiveld GJEJ. 2013. A Diet High in Resistant Starch Modulates Microbiota Composition, SCFA Concentrations, and Gene Expression in Pig Intestine. J Nutr 143:274–283.

53. Gibson GR, Beatty ER, Wang X, Cummings JH. 1995. Selective stimulation of bifidobacteria in the human colon by oligofructose and inulin. Gastroenterology 108:975–982.

54. Jensen BB, Jørgensen H. 1994. Effect of dietary fiber on microbial activity and microbial gas production in various regions of the gastrointestinal tract of pigs. Appl Environ Microbiol 60:1897–1904.

55. Segain J, de la Bletiere DR, Bourreille A, Leray V, Gervois N, Rosales C, Ferrier L, Bonnet C, Blottiere H, Galmiche J. 2000. Butyrate inhibits inflammatory responses through NFκB inhibition: implications for Crohn’s disease. Gut 47:397–403.

56. Zhou ZY, Packialakshmi B, Makkar SK, Dridi S, Rath NC. 2014. Effect of butyrate on immune response of a chicken macrophage cell line. Vet Immunol Immunopathol 162:24–32.

57. Zimmer J, Lange B, Frick J-S, Sauer H, Zimmermann K, Schwiertz A, Rusch K, Klosterhalfen S, Enck P. 2012. A vegan or vegetarian diet substantially alters the human colonic faecal microbiota. Eur J Clin Nutr 66:53–60.

58. Muegge BD, Kuczynski J, Knights D, Clemente JC, González A, Fontana L, Henrissat B, Knight R, Gordon JI. 2011. Diet Drives Convergence in Gut Microbiome Functions Across Mammalian Phylogeny and Within Humans. Science 332:970–974.

59. Yu B, Liu JR, Hsiao FS, Chiou PWS. 2008. Evaluation of Lactobacillus reuteri Pg4 strain expressing heterologous β-glucanase as a probiotic in poultry diets based on barley. Anim Feed Sci Technol 141:82–91.

60. Sun H, Tang J, Yao X, Wu Y, Wang X, Feng J. 2013. Effects of dietary inclusion of fermented cottonseed meal on growth, cecal microbial population, small intestinal morphology, and digestive enzyme activity of broilers. Trop Anim Health Prod 45:987–993.

61. Drew MD, Syed NA, Goldade BG, Laarveld B, Van Kessel AG. 2004. Effects of dietary protein source and level on intestinal populations of Clostridium perfringens in broiler chickens. Poult Sci 83:414–420.

62. Knarreborg A, Simon MA, Engberg RM, Jensen BB, Tannock GW. 2002. Effects of Dietary Fat Source and Subtherapeutic Levels of Antibiotic on the Bacterial Community in the Ileum of Broiler Chickens at Various Ages. Appl Environ Microbiol 68:5918–5924.

63. Andersen AD, Mølbak L, Thymann T, Michaelsen KF, Lauritzen L. 2011. Dietary long-chain n-3 PUFA, gut microbiota and fat mass in early postnatal piglet development—exploring a potential interplay. Prostaglandins Leukot Essent Fatty Acids 85:345–351.

64. Geier MS, Torok VA, Allison GE, Ophel-Keller K, Gibson RA, Munday C, Hughes RJ. 2009. Dietary omega-3 polyunsaturated fatty acid does not influence the intestinal microbial communities of broiler chickens. Poult Sci 88:2399–2405.

65. Cummings JH, Wiggins HS, Jenkins DJ, Houston H, Jivraj T, Drasar BS, Hill MJ. 1978. Influence of diets high and low in animal fat on bowel habit, gastrointestinal transit time, fecal microflora, bile acid, and fat excretion. J Clin Invest 61:953–963.

66. Gallaher DD, Franz PM. 1990. Effects of Corn Oil and Wheat Brans on Bile Acid Metabolism in Rats. J Nutr 120:1320–1330.

67. Snook JT, Meyer JH. 1964. Response of Digestive Enzymes to Dietary Protein. J Nutr 82:409– 414.

68. Hamlin LJ, Hungate RE. 1956. CULTURE AND PHYSIOLOGY OF A STARCH- DIGESTING BACTERIUM (BACTEROIDES AMYLOPHILUS N. SP.) FROM THE BOVINE RUMEN1. J Bacteriol 72:548–554.

69. Hungate RE. 1950. THE ANAEROBIC MESOPHILIC CELLULOLYTIC BACTERIA 1. Bacteriol Rev 14:1–49.

70. Hungate RE. 2013. The Rumen and Its Microbes. Elsevier.

71. Degirolamo C, Rainaldi S, Bovenga F, Murzilli S, Moschetta A. 2014. Microbiota Modification with Probiotics Induces Hepatic Bile Acid Synthesis via Downregulation of the Fxr-Fgf15 Axis in Mice. Cell Rep 7:12–18.

72. Jin LZ, Ho YW, Abdullah N, Jalaludin S. 2000. Digestive and Bacterial Enzyme Activities in Broilers Fed Diets Supplemented with Lactobacillus Cultures. Poult Sci 79:886–891.

73. Wang Y, Gu Q. 2010. Effect of probiotic on growth performance and digestive enzyme activity of Arbor Acres broilers. Res Vet Sci 89:163–167.

74. Allen PC. 1987. Physiological Responses of Chicken Gut Tissue to Coccidial Infection: Comparative Effects of Eimeria acervulina and Eimeria mitis on Mucosal Mass, Carotenoid Content, and Brush Border Enzyme Activity. Poult Sci 66:1306–1315.

75. Kozakova H, Kolinska J, Lojda Z, Rehakova Z, Sinkora J, Zakostelecka M, Splichal I, Tlaskalova-Hogenova H. 2006. Effect of bacterial monoassociation on brush-border enzyme activities in ex-germ-free piglets: comparison of commensal and pathogenic Escherichia coli strains. Microbes Infect Inst Pasteur 8:2629–2639.

76. Hooper LV, Littman DR, Macpherson AJ. 2012. Interactions Between the Microbiota and the Immune System. Science 336:1268–1273.

77. Bai SP, Wu AM, Ding XM, Lei Y, Bai J, Zhang KY, Chio JS. 2013. Effects of probiotic- supplemented diets on growth performance and intestinal immune characteristics of broiler chickens. Poult Sci 92:663–670.

78. Stanley D, Hughes RJ, Geier MS, Moore RJ. 2016. Bacteria within the Gastrointestinal Tract Microbiota Correlated with Improved Growth and Feed Conversion: Challenges Presented for the Identification of Performance Enhancing Probiotic Bacteria. Front Microbiol 7.

79. Zheng A, Luo J, Meng K, Li J, Zhang S, Li K, Liu G, Cai H, Bryden WL, Yao B. 2014. Proteome changes underpin improved meat quality and yield of chickens (Gallus gallus) fed the probiotic Enterococcus faecium. BMC Genomics 15:1167.

80. Zheng A, Luo J, Meng K, Li J, Bryden WL, Chang W, Zhang S, Wang LXN, Liu G, Yao B. 2016. Probiotic (Enterococcus faecium) induced responses of the hepatic proteome improves metabolic efficiency of broiler chickens (Gallus gallus). BMC Genomics 17:89.

81. Bäckhed F, Manchester JK, Semenkovich CF, Gordon JI. 2007. Mechanisms underlying the resistance to diet-induced obesity in germ-free mice. Proc Natl Acad Sci U S A 104:979–984.

82. Backhed F, Ding H, Wang T, Hooper LV, Koh GY, Nagy A, Semenkovich CF, Gordon JI. 2004. The gut microbiota as an environmental factor that regulates fat storage. Sci STKE 101:15718.

83. Kau AL, Ahern PP, Griffin NW, Goodman AL, Gordon JI. 2011. Human nutrition, the gut microbiome and the immune system. Nature 474:327–336.

84. Atarashi K, Tanoue T, Oshima K, Suda W, Nagano Y, Nishikawa H, Fukuda S, Saito T, Narushima S, Hase K, Kim S, Fritz JV, Wilmes P, Ueha S, Matsushima K, Ohno H, Olle B, Sakaguchi S, Taniguchi T, Morita H, Hattori M, Honda K. 2013. Treg induction by a rationally selected mixture of Clostridia strains from the human microbiota. Nature 500:232–236.

85. Round JL, Mazmanian SK. 2010. Inducible Foxp3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proc Natl Acad Sci 107:12204–12209.

86. Lebeer S, Vanderleyden J, De Keersmaecker SCJ. 2010. Host interactions of probiotic bacterial surface molecules: comparison with commensals and pathogens. Nat Rev Microbiol 8:171–184.

87. Hausmann M, Kiessling S, Mestermann S, Webb G, Spöttl T, Andus T, Schölmerich J, Herfarth H, Ray K, Falk W, Rogler G. 2002. Toll-like receptors 2 and 4 are up-regulated during intestinal inflammation. Gastroenterology 122:1987–2000.

88. Medzhitov R. 2007. Recognition of microorganisms and activation of the immune response. Nature 449:819–826.

89. Round JL, Lee SM, Li J, Tran G, Jabri B, Chatila TA, Mazmanian SK. 2011. The Toll-like receptor pathway establishes commensal gut colonization. Science 332:974–977.

90. Atarashi K, Tanoue T, Shima T, Imaoka A, Kuwahara T, Momose Y, Cheng G, Yamasaki S, Saito T, Ohba Y, Taniguchi T, Takeda K, Hori S, Ivanov II, Umesaki Y, Itoh K, Honda K. 2011. Induction of colonic regulatory T cells by indigenous Clostridium species. Science 331:337–341.

91. Ivanov II, Atarashi K, Manel N, Brodie EL, Shima T, Karaoz U, Wei D, Goldfarb KC, Santee CA, Lynch SV, Tanoue T, Imaoka A, Itoh K, Takeda K, Umesaki Y, Honda K, Littman DR. 2009. Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell 139:485–498.

92. Lee YK, Menezes JS, Umesaki Y, Mazmanian SK. 2010. Proinflammatory T-cell responses to gut microbiota promote experimental autoimmune encephalomyelitis. Proc Natl Acad Sci 108:4615–4622.

93. Littman DR, Rudensky AY. 2010. Th17 and Regulatory T Cells in Mediating and Restraining Inflammation. Cell 140:845–858.

94. Brisbin JT, Zhou H, Gong J, Sabour P, Akbari MR, Haghighi HR, Yu H, Clarke A, Sarson AJ, Sharif S. 2008. Gene expression profiling of chicken lymphoid cells after treatment with Lactobacillus acidophilus cellular components. Dev Comp Immunol 32:563–574.

95. Galdeano CM, Perdigón G. 2006. The Probiotic Bacterium Lactobacillus casei Induces Activation of the Gut Mucosal Immune System through Innate Immunity. Clin Vaccine Immunol 13:219–226.

96. Macpherson AJ, Gatto D, Sainsbury E, Harriman GR, Hengartner H, Zinkernagel RM. 2000. A Primitive T Cell-Independent Mechanism of Intestinal Mucosal IgA Responses to Commensal Bacteria. Science 288:2222–2226.

97. Crhanova M, Hradecka H, Faldynova M, Matulova M, Havlickova H, Sisak F, Rychlik I. 2011. Immune Response of Chicken Gut to Natural Colonization by Gut Microflora and to Salmonella enterica Serovar Enteritidis Infection. Infect Immun 79:2755–2763.

98. Hooper LV, Stappenbeck TS, Hong CV, Gordon JI. 2003. Angiogenins: a new class of microbicidal proteins involved in innate immunity. Nat Immunol 4:269–273.

99. Smits HH, Engering A, van der Kleij D, de Jong EC, Schipper K, van Capel TMM, Zaat BAJ, Yazdanbakhsh M, Wierenga EA, van Kooyk Y, Kapsenberg ML. 2005. Selective probiotic bacteria induce IL-10–producing regulatory T cells in vitro by modulating dendritic cell function through dendritic cell–specific intercellular adhesion molecule 3–grabbing nonintegrin. J Allergy Clin Immunol 115:1260–1267.

100. Vanderpool C, Yan F, Polk DB. 2008. Mechanisms of probiotic action: Implications for therapeutic applications in inflammatory bowel diseases. Inflamm Bowel Dis 14:1585–1596.

101. Zhang L, Li N, Caicedo R, Neu J. 2005. Alive and dead Lactobacillus rhamnosus GG decrease tumor necrosis factor-alpha-induced interleukin-8 production in Caco-2 cells. J Nutr 135:1752– 1756.

102. Kitazawa H, Ueha S, Itoh S, Watanabe H, Konno K, Kawai Y, Saito T, Itoh T, Yamaguchi T. 2001. AT oligonucleotides inducing B lymphocyte activation exist in probiotic Lactobacillus gasseri. Int J Food Microbiol 65:149–162.

103. Haghighi HR, Gong J, Gyles CL, Hayes MA, Zhou H, Sanei B, Chambers JR, Sharif S. 2006. Probiotics Stimulate Production of Natural Antibodies in Chickens. Clin Vaccine Immunol 13:975–980.

104. Berland R, Wortis HH. 2002. Origins and functions of b-1 cells with notes on the role of cd5. Annu Rev Immunol 20:253–300.

105. Hand TW, Santos LMD, Bouladoux N, Molloy MJ, Pagán AJ, Pepper M, Maynard CL, Elson CO, Belkaid Y. 2012. Acute Gastrointestinal Infection Induces Long-Lived Microbiota-Specific T Cell Responses. Science 337:1553–1556.

106. Koenen ME, Kramer J, Hulst R van der, Heres L, Jeurissen SHM, Boersma WJA. 2004. Immunomodulation by probiotic lactobacilli in layer- and meat-type chickens. Br Poult Sci 45:355–366.

107. Noujaim JC, Filho RLA, Lima ET, Okamoto AS, Amorim RL, Neto RT. 2008. Detection of T Lymphocytes in Intestine of Broiler Chicks Treated with Lactobacillus spp. and Challenged with Salmonella enterica Serovar Enteritidis. Poult Sci 87:927–933.

108. Jeon SG, Kayama H, Ueda Y, Takahashi T, Asahara T, Tsuji H, Tsuji NM, Kiyono H, Ma JS, Kusu T, Okumura R, Hara H, Yoshida H, Yamamoto M, Nomoto K, Takeda K. 2012. Probiotic Bifidobacterium breve Induces IL-10-Producing Tr1 Cells in the Colon. PLoS Pathog 8:e1002714.

109. Li SP, Zhao XJ, Wang JY. 2009. Synergy of Astragalus polysaccharides and probiotics (Lactobacillus and Bacillus cereus) on immunity and intestinal microbiota in chicks. Poult Sci 88:519–525.

110. Bendelac A, Savage PB, Teyton L. 2007. The Biology of NKT Cells. Annu Rev Immunol 25:297–336.

111. Olszak T, An D, Zeissig S, Vera MP, Richter J, Franke A, Glickman JN, Siebert R, Baron RM, Kasper DL, Blumberg RS. 2012. Microbial Exposure During Early Life Has Persistent Effects on Natural Killer T Cell Function. Science 336:489–493.

112. Ohkawara S, Furuya H, Nagashima K, Asanuma N, Hino T. 2005. Oral Administration of Butyrivibrio fibrisolvens, a Butyrate-Producing Bacterium, Decreases the Formation of Aberrant Crypt Foci in the Colon and Rectum of Mice. J Nutr 135:2878–2883.

113. Couper KN, Blount DG, Riley EM. 2008. IL-10: The Master Regulator of Immunity to Infection. J Immunol 180:5771–5777.

114. Trinchieri G. 2003. Interleukin-12 and the regulation of innate resistance and adaptive immunity. Nat Rev Immunol 3:133–146.

Table 1.1 Selected commercial probiotic products

Product (Company) Intended Host Constituent Microbes Enterococcus faecium, Pediococcus acidilactici, Bifidobacterium animalis, Lactobacillus salivarius, Lactobacillus PoultryStar (Biomin) Poultry reuteri Bifidobacterium longum, Bifidobacterium infantis and Bifidobacterium breve, Lactobacillus acidophilus, Lactobacillus casei, Lactobacillus bulgaricus, VSL#3 (VSL Lactobacillus plantarum,Streptococcus Pharmaceuticals) Humans salivarius subspecies thermophilus) Lactobacillus acidophilus LA-5, TruBiotic (Bayer) Humans Bifidobacterium animalis BB-12 Lactobacillus acidophilus DDS-1, Bifidobacterium lactis, Lactobacillus plantarum,Lactobacillus casei, Lactobacillus rhamnosus, Lactobacillus brevis, Dr. Mercola Complete Bifidobacterium longum, Streptococcus (Mercola) Humans thermophilus, Bifidobacterium bifidum Ruminants, Enviva PRO (Danisco) Swine, Poultry 3 Bacillus subtilis strains Lactobacillus acidophilus strain NP51, Probionibacterium freudenreichii strain Bovamin (Chr Hansen) Ruminants NP24 PoultriMax (Chr Hansen) Poultry Lactobacillus animalis Horses, Yea-Sacc (Alltech) Ruminants Saccharomyces cerevisiae Alterion (Adisseo) Poultry Bacillus subtilis Lactobacillus acidophilus, Probionibacterium freudenreichii, Lactobacillus casei, Lactobacillus lactis, Pediococcus cerevisiae, Enterococcus faecium, Bacillus subtilis, Bacillus XP DFM (Diamond-V) Ruminants licheniformis

CHAPTER 2: DEVELOPMENT OF THE CHICK MICROBIOME: HOW EARLY

EXPOSURE INFLUENCES FUTURE MICROBIAL DIVERSITY1

1 AL Ballou, RA Ali, MA Mendoza, JC Ellis, HM Hassan, WJ Croom, MD Koci. Frontiers in Veterinary Science. 2016

ABSTRACT

The concept of improving animal health through improved gut health has existed in food animal production for decades; however, only recently have we had the tools to identify microbes in the intestine associated with improved performance. Currently, little is known about how the avian microbiome develops or the factors that affect its composition. To begin to address this knowledge gap, the present study assessed the development of the cecal microbiome in chicks from hatch to 28 days of age with and without a live Salmonella vaccine and/or probiotic supplement; both are products intended to promote gut health. The microbiome of growing chicks develops rapidly from days 1-3, and the microbiome is primarily Enterobacteriaceae, but Firmicutes increase in abundance and taxonomic diversity starting around day 7. As the microbiome continues to develop, the influence of the treatments becomes stronger. Predicted metagenomic content suggests that functionally, treatment may stimulate more differences at day 14, despite the strong taxonomic differences at day 28. These results demonstrate that these live microbial treatments do impact the development of the bacterial taxa found in the growing chicks; however, additional experiments are needed to understand the biochemical and functional consequences of these alterations.

INTRODUCTION

Increasing evidence in multiple species demonstrate the impact gut microbes have on intestinal function, digestion, host metabolism, and immune function (1,2). While the food animal industry has employed various methods to control and augment the bacteria in the gut for decades, this has been done with little understanding of the complexity of the microbial populations and their association with animal health. The advent of microbiome analysis will allow for better use of these products and the rational design of new therapies to promote animal health and performance. An estimated $585 million/year is spent globally on interventions to manage disease in food animals (3); many of these diseases are intestinal in nature (4), and the indirect costs of these intestinal diseases are far greater. The application of modern nucleotide sequencing and associated bioinformatics techniques to the avian gastrointestinal microbiome will lead to breakthroughs in our understanding of digestive processes, host metabolic regulation, immune function, and intestinal dysfunction and pathology. Collectively, increased understanding of the host-microbiome relationship, and the development of techniques to improve these interactions, could reduce the prevalence of food borne pathogens. In order to effectively apply modern microbial ecology research techniques and elucidate the manner in which the avian intestinal microbiome interacts with the host genome, it is imperative to develop a comprehensive understanding of how the avian microbiome develops under different physiological states and management practices.

There is a dearth of information available on the development and definition of a normal avian gut microbiome. Recent investigations have begun to identify species commonly seen in adult chickens, but little is known about the intermediate and developing

community (5–7). Furthermore, there is a paucity of information of the effects of treatments that target the gut environment on the development of the intestinal microbiome of chickens.

This impairs our ability to understand how these gut-targeted treatments interact with each other and the host, and how they might affect gut activity and health. A better understanding of these interactions will allow for the rational use of bacterial groups to promote specific host responses.

The goal of this study was to understand the ontogeny of the chicken intestinal microbiome, and how commonly used live bacterial treatments influence this dynamic microbial community. Specifically, we included two live bacterial products currently used in the industry that are intended to improve animal health through manipulation of the host microbiota. We used a live attenuated Salmonella enterica, serovar Typhimurium vaccine

(Salmune®, CEVA Biomune) and a probiotic feed supplement comprised of Lactobacillus acidophilus, Lactobacillus casei, Enterococcus faecium, and Bifidobacterium bifidium

(PrimaLac, Star Labs). We hypothesized that the species richness of the microbiome would increase rapidly, and that the addition of live bacterial treatments would alter the development of microbial diversity and the composition of the microbiome. The results from this study demonstrated exposing developing chickens to individual or combined bacterial regimens leads to treatment-specific microbial populations. These populations continue to diverge with age, even in animals receiving only a one-time dose of the Salmonella

Typhimurium vaccine at day of hatch. Predicted metagenomic content in these populations suggest changes in potential microbial metabolic activity and microbe-derived signaling

molecules; however these changes were less numerous than the taxonomic changes seen in the same populations.

MATERIAL AND METHODS

Animals and Treatments

Two hundred one-day-old female commercial white leghorn laying type chickens (W-

36, Hy-line International) were assigned to one of 4 treatments (50 chicks/treatment) in a 2x2 factorial design. The four groups were designated as follows: Diluent-Control (DC); Diluent-

Probiotic (DP); Vaccine-Control (VC); and Vaccine-Probiotic (VP). Animals received either a one-time dose of a live, attenuated Salmonella Typhimurium (ST) spray vaccine

(Salmune®, Ceva Biomune, Lenexa, KS; Vaccine group) or a sham vaccination consisting of the vaccine diluent, water (Diluent group). The vaccine and diluent spray were administered as recommended by the manufacturer. These treatment groups were further divided into two dietary groups; one group (Control) was fed a standard corn-soybean starter diet

(Supplemental Table 1) and the probiotic group was fed an identical starter diet supplemented with 0.1% (w/w) of the probiotic PrimaLac® (Lactobacillus acidophilus,

Lactobacillus casei, Enterococcus faecium, Bifidobacterium bifidium; Star Labs Inc.,

Clarksdale, MO; Probiotic group). Probiotic pre-mix was added to the probiotic groups’ feed prior to the experiment and animals in all groups were fed ad libitum for four weeks.

Animals in all groups were housed in 934-1-WP isolators (L. H. Leathers Inc, Athens

GA) climate-controlled HEPA-filtered isolation units. The animals were maintained and

euthanized under an approved protocol from the North Carolina State University Institutional

Animal Care and Use Committee (OLAW #A3331-01).

Sample Collection

Six chickens from each treatment group were euthanized (CO2 followed by cervical dislocation) on days 0, 1, 3, 7, 14 and 28 and the contents of one cecal lobe collected and maintained on ice. At early timepoints, some chicks yielded minimal or no cecal digesta; these are noted in Supplemental Table 2. The cecal samples were weighed and diluted with

600 μl of 30% glycerol in PBS for storage at -80° C.

DNA isolation and 16S sequencing

DNA was isolated from each cecal sample using the MO BIO Power Soil kit (MO

BIO, Carlsbad, CA) with the following modifications: a 10-minute, 65° C incubation step was added and samples were then homogenized for 45s at 5100 RPM using garnet bead- containing tubes and a Precellys 24 homogenizer (Precellys, Montigny-le-Bretonneux,

France).

DNA recovered from the extraction process was quantified using a NanoDrop 2000 spectrophotometer (NanoDrop, Wilmington, DE) and 10 ng from each sample were aliquoted into 96-well plates in a random order. Some animals contained only small amounts of cecal digesta, particularly at days 0-3, resulting in very small amounts of DNA for some samples.

DNA from these samples was included in the sequencing process, despite the possibility of poor quality sequencing (Supplemental Table 2). MiSeq library preparation and 151x151

paired-end sequencing (Illumina, San Diego, CA) were performed by the Argonne National

Laboratory Institute for Genomics and Systems Biology Next Generation Sequencing Core using a protocol and primers recommended and previously described by the Earth

Microbiome project and others. Primers used spanned the V4 region of the 16S rRNA gene

(515F: GTGYCAGCMGCCGCGGTAA, 806R: GGACTACHVGGGTWTCTAAT) (8).

This primer set is commonly used to evaluate the microbiome community across a variety of fields, and is well validated in several models, including the chicken (8–11). Studies estimating microbial composition using V4 report diversity measurements comparable to those obtained with full-length 16S sequences (12).

Sequence data analysis

The unpaired raw sequencing reads were paired and filtered using EA-Utils (13).

Paired reads were processed using the QIIME suite of tools (v 1.8.0) (14); barcode matching and quality filtering were conducted prior to picking operational taxonomic units (OTUs).

The 16S sequencing process did not yield equal sequence coverage for all samples, and some samples had very low sequence coverage. Samples with low sequence coverage or consistently poor quality were excluded from analysis. Additionally, some ceca from early time points contained little to no recoverable digesta. Consequently, a small number of samples from different time points were removed at this stage (Supplemental Table 2).

OTUs were picked using an open-reference protocol. Briefly, sequences were grouped into

OTUs based on 97% sequence identity using uclust and the Greengenes reference database

(15,16). OTUs that failed to match to the database were reclustered, resampled and re-

compared to the database; in this way, new reference sequences are compared to the database in order to minimize the number of excluded sequences. Finally, OTUs that failed to align to any sequences in the reference database are de novo clustered. Representative sequences from each OTU were picked and assigned taxonomy using the uclust consensus taxonomy assigner. During this process, sequences with high identity (> 97%) were grouped into the same OTU, and are reported at the lowest level of taxonomic identification common to all sequences (17,18). Sequence coverage was normalized across samples in each analysis.

Taxonomic assignments, and alpha and beta diversity metrics were generated using QIIME and Primer-E (v6.1.16; Primer-E LTD, Ivybridge, UK). Principal coordinate analysis (PCoA) plots used in this study were generated in Primer-E using the Bray-Curtis distance metric

(19).

Permutational multivariate analysis of variance (PERMANOVA) was conducted using the PERMANOVA+ add-on to Primer-E. Main and pair-wise tests were conducted using up to 1000 permutations of residuals under a reduced model. Similarity percentage analysis (SIMPER) of taxonomic groups between treatment groups and times was made in

Primer-E using Bray-Curtis distances. Analysis of similarity (ANOSIM) tests were conducted using Primer-E. Tests were conducted using up to 1000 permutations and the

Spearman rank correlation method. A global test statistic (R) was generated for each treatment; the rank similarities between and within treatments were calculated and compared.

The global R statistic is a measure of the strength of a treatment group’s association with microbiome composition, with 1 being the strongest association and 0 being no association.

Metagenomic inferences from the 16S amplicon data were made using the QIIME suite of tools (14,17,18,15), PICRUSt (20) and KEGG (21); statistics and visualization of functional data were depicted using STAMP (22). Closed-reference OTU-picking protocols were used to identify 16S sequences belonging to annotated genomes. Briefly, sequences were grouped into OTUs based on 97% sequence identity using uclust and the Greengenes reference database. Representative sequences from each OTU were picked and assigned taxonomy using the uclust consensus taxonomy assigner. PICRUSt and KEGG were used to generate a list of functional genes predicted to be present in the sample and to organize these genes into gene pathways. Using STAMP, heatmaps were generated displaying differences in gene group abundance at each time point. In order to minimize the number of treatment- based differences that may not be biologically relevant, analysis was limited to those differences with an effect size greater than 0.7 as calculated by STAMP (eta-squared method)

(22). Storey’s FDR correction was applied to all comparisons between treatments (23).

Nearest neighbor hierarchical clustering was used to group each sample according to abundance of gene groups in question.

RESULTS

Microbiome composition and complexity change rapidly with age

16S rRNA sequence analysis of the microbiome from the ceca of untreated animals

(DC) demonstrated a microbiome with low diversity in days 0 and 1, dominated by

Enterobacteriaceae and to a lesser extent Enterococcus (Figure 1). The number of OTUs detected in the microbiome increased significantly (P < 0.05) by day 3 (data not shown). This

increase in bacterial richness starts with Ruminococcaceae groups during the first week of life and continues with other Firmicutes. By day 14, and extending through day 28,

Ruminococcus and other Firmicutes outnumber Enterobacteriaceae (Figure 1).

Age more influential in microbiome development than treatment

Principal coordinate analysis of samples across all time points and treatment groups reveals that the effect of animal age on community composition was larger than that of bacterial treatment (Figure 2A). The ANOSIM-generated global test statistics for time

(R=0.67) and the treatments (Vaccine R=0.361, Probiotic R=0.317) demonstrate the relative impact of each on the community. At time points 0 – 7, the samples show large within time point variability. At day 28, within time point variability is decreased and samples are tightly clustered in the PCoA plot (Figure 2A). Community analysis of cecal samples across time points and treatment groups show that Gram negative bacteria (Proteobacteria) dominate at early time points, whilst Gram positive Firmicutes, especially Clostridia taxa become more prominent with age (Figure 2B).

Treatments alter microbial composition and rate of development

Analyses of microbial populations were conducted within each time point (days 1, 7,

14, and 28) to assess the impact of treatment on composition and richness of the microbiome independent of age. No differences in microbial composition were detected at day 1, but significant differences in cecal microbiome composition were observed among all 4 treatment groups by day 7 (Figure 3A). A PERMANOVA showed all 4 treatment groups are

distinct in composition at days 7, 14, and 28 (P < 0.05). A comparison of taxonomic richness

(alpha diversity) among treatment groups at days 1, 7, 14, and 28 was made using rarefaction plots. The treatment groups show similar levels of unique taxa at day 1; however, at days 7 and 14, probiotic groups tend to have fewer unique taxa (P < 0.1 at day 7, P < 0.05 at day

14). Interestingly, there was no significant difference in alpha diversity at day 28 (Figure

3B).

Treatment with live bacteria affects abundance of taxa not associated with treatment

Similarity percentage analysis (SIMPER) conducted between treatment groups at days 14 and 28 indicates that the differences between treatment groups can largely be attributed to changes in the most abundant order, Clostridiales, including Lachnospiraceae and Ruminococcaceae genera (Tables 1 and 2). MANOVA was used to identify differentially abundant taxa between treatment groups, with an FDR correction made to account for multiple comparisons. At day 14, DC animals harbored a significantly higher proportion of Enterobacteriaceae (Table 1) with 16% Enterobacteriaceae as compared to 3-

9% in the other treatments. Lactobacillus was significantly increased in the DP group relative to DC. At day 14, 83% of significantly different taxa were Firmicutes, and 63% were Clostridia.

Analysis of the taxonomic groups represented in each treatment at day 28 indicate that most significant changes occur in the Firmicutes phylum, including differences in the abundance of Ruminococcaceae, Lachnospiraceae, and Peptostreptococcaceae (Table 2).

Lactobacillus is also increased in the DP group relative to DC. Eighty-one percent of all significantly different taxa at day 28 were in the Order Clostridiales.

Treatment-induced changes in microbiome diversity lead to predicted changes in abundance of functional gene families

Estimates were made of the functional changes that may occur in the cecal microbiome following treatment using closed-reference OTU-picking and PICRUSt. Gene groups targeted for statistical analysis had an FDR-corrected P < 0.01, and an effect size of

0.7 or higher. At day 14, samples cluster primarily by probiotic treatment, and the VP group is most distinct from other treatments. The combination of ST and probiotic treatments increases the expected proportion of genes related to environment-sensing; two-component systems, bacterial motility, chemotaxis, and flagellar component assembly genes were predicted to increase. DC, DP, and VC groups have relatively higher abundance of genes related to amino acid metabolism, DNA repair and replication, and translation (Figure 4A).

The differences between DC, DP, and VC groups were minor, but the DP group displayed the lowest expected abundance of two-component system and bacterial motility genes.

Fewer gene groups met the inclusion criteria at day 28, and the total relative abundance of included gene groups was lower than at day 14. At day 28, DP and VP treatment groups display higher predicted levels of genes related to one carbon metabolism, terpenoid synthesis, and translation proteins (Figure 4B). DC and DP groups had higher proportions of fatty acid metabolism, drug metabolism, and signal transduction gene pathways.

Relative abundance tables of taxa and predicted gene groups were used to generate area charts of between-treatment changes in taxa and gene groups. Vaccine and Probiotic groups differ from the DC group taxonomically at both day 14 and day 28. Gene group abundance shows less treatment-specific variability (Figure 5).

DISCUSSION

Little is known about the development of the microbiome in young birds, and how it is affected by different stimuli (7). The goal of this study was to characterize the healthy developing microbiome in chickens and understand how commonly used bacterial treatments intended to improve or maintain health would affect this process. This is important in the food animal industry as there are numerous feed additives intended to improve animal health, either directly or indirectly through improving gut health. However; the mechanisms by which these amendments work are poorly understood. Most claim to enhance health and performance via manipulation of the host intestinal microbiome, but the mechanism of action has been studied in very few of these products (24,25).

In the present study, we administered two commonly used live bacterial treatments applied in poultry production to enhance intestinal health and function. According to the manufacturer, the live Salmonella Typhimurium vaccine used here is intended to prevent colonization of the gut and internal organs by multiple types of Salmonella including

Heidelberg, Typhimurium, Hadar, Kentucky, and Enteritidis (26). Similarly, the probiotic used here is intended to maintain healthy microbiota balance in the gut (27). We investigated to what extent these health-promoting treatments affect the microbiome of young chicks.

We found that the post-hatch intestinal microbiome has low diversity dominated by

Gram negative bacteria, particularly Enterobacteriaceae, which includes Salmonella,

Klebsiella, Proteus, and E coli. During the first week of life there is a shift to a much more diverse community comprised of a wide variety of Gram positive bacteria, mainly within the

Clostridiales group, resulting in a correspondingly smaller proportion of Gram negative bacteria (Figure 1). The proportion of Gram negative bacteria in the cecum at day 28 is less than 6%, and it is almost entirely Enterobacteriaceae.

Data from the present study suggest a microbiome more affected by age than treatment (Figure 2A). Irrespective of treatment, all groups show a sharp decline in

Enterobacteriaceae with age, including the vaccinated groups, where levels of

Enterobacteriaceae would be expected to increase following Salmonella (member of the

Enterobacteriaceae family) vaccination. Nor does addition of a Firmicutes-based probiotic product stimulate more rapid conversion to a Firmicutes-dominated microbiome (Figure 2B) in probiotic-fed animals. Day-old birds begin with a gut colonized by few bacterial species at a concentration several orders of magnitude lower than mature animals (28,29), so it is likely that the primary driver of age-dependent increase in complexity is bacteria colonizing a previously empty niche. However, diet can also play a major role in the composition of the microbiome and exerts an influence on the developing and mature gut (30,31). Studies by

Sergeant, et al characterizing the microbiome of wheat-fed chickens reported Megamonas and Negativicutes as more abundant in their adult birds, while the Firmicutes most commonly seen in this trial, Lachnospiraceae and Ruminococcaceae, were less abundant

(32). The effect of gut development on the intestinal microbiome is more difficult to

quantify; though studies of germ free and gnotobiotic mice clearly demonstrate the microbiome is essential to the development of a fully-functioning gut (33,34), whether the developmental stage of the gut is a variable influencing the development of the microbiome is less clear.

Despite the strong relationship between age and composition of the microbiome, the bacterial treatments included in this study did affect the microbiome. Principal coordinate analysis of the four treatment groups at days 1, 7, 14, and 28 illustrate the impact of both vaccination and probiotic supplementation on the microbiome (Figure 3A). Despite the fact that the vaccine is only administered on day 0, global R statistics demonstrate the impact of

ST on the composition of the microbiome is on par with that of the continuously fed probiotic at days 14 (vaccine = 0.802, probiotic = 0.882) and 28 (vaccine = 0.697, probiotic =

0.705). The magnitude of the effect of the one-time ST inoculation is nearly as great as that of the continuously fed probiotic despite low levels of the ST-containing taxonomic group

Enterobacteriaceae after day 7, suggesting early colonizers influence the relative abundance of the microbiome despite being transient themselves. While little is known about the long- term effects of early microbiome perturbation, some studies support this idea (35,36).

To understand the impact of treatment at the taxonomic level, SIMPER was conducted to identify species contributing to differences between treated and untreated animals. Most of the differences between DC and treated animals at days 14 and 28 involve

Lachnospiraceae, Ruminococcaceae, and other Clostridiales (Tables 3 and 4). Though abundance of Lactobacillus in the DP and VP groups is higher at day 14 (P < 0.05), the magnitude of the increase in VP over DC is not great enough for Lactobacillus to be a major

source of dissimilarity. There is little microbiological evidence that the bacterial products applied in this study interfere with each other; levels of Lactobacillus are not significantly lower in the VP group than the DP group, and VP animals remain ST-positive at day 28 (data not shown). There are signs of treatment interaction; however, ST vaccination decreases abundance of the group Clostridiales other, and the combination of probiotic and vaccine results in the strongest difference from the DP group; 11% reduced to 1% of the identified bacteria, perhaps indicating a synergy between the two treatments that makes the cecum a more hostile environment for this group of bacteria.

Changes in taxonomic diversity are the most used indicator to infer changes in microbiological activity, but it is becoming apparent that many of the functions of a normal microbiome can be carried out by a number of microbial groups (37,38). Therefore, understanding how treatments affect taxonomic abundance may not provide us with a complete understanding of how they impact healthy and diseased guts, or develop therapies that target the predominant cause of gut dysbiosis; a change in function. In its entirety, the chicken gut is estimated to be colonized by as many as 1013 microbes, and they have a combined genetic potential far in excess of the ~20,000 genes identified in the chicken genome (39,40). PICRUSt uses the 16S rRNA genes obtained during sequencing to infer the presence of functional genes known or predicted to be associated with those 16S sequences.

At day 14, there were predicted increases in the VP group in genes related to motility, flagellar assembly, chemotaxis, and two-component system. In contrast, VP microbiomes displayed lower abundance of many protein and energy metabolism genes, as well as genes related to DNA replication and protein translation (Figure 4A). Supporting the taxonomic

data suggesting that the microbiome is still equilibrating at day 14 (Figures 2 and 3), the functional changes at day 28 are both fewer and less dramatic (Figure 5). The VP group exhibits the most variation of the 4 treatments, and suggests changes in a few cellular metabolism pathways. Interestingly, the effect of probiotic supplementation and its interaction with the vaccine appears to stimulate more functional changes than the vaccine group alone. At days 14 and 28 the DP and VP groups were more likely to have either the highest or lowest levels of any given gene group.

A possible contributor to the lack of more dramatic functional diversity between treatment groups at day 28 could be limitations inherent to this technique and its application to avian microbial communities. One of these is its reliance on sequenced and annotated genomes. Though comparisons between PICRUSt results and metagenomics data from the same samples have shown the predictive value of PICRUSt analysis is very good (20,41),

16S genes without a confident phylogenetic assignment cannot be used as marker genes.

Because of this, about 15% of the 16S sequences were filtered out at day 14 and over 20% at day 28. This number of unknown or uncharacterized sequences may be higher in the avian microbiome than in the human or murine microbiome, as the databases used in this process were all developed based on mammalian microbiota; chicken-specific microbes that may be important in this system could be excluded from analysis because they are not part of the 16s and/or KEGG databases. However, this analytical technique has been successfully used on avian microbiomes in the past (6). While the difference in excluded taxa between time points is not large, it is possible that the bacteria excluded from analysis are active in the community; evaluation of those taxa excluded from PICRUSt analysis indicates that some

are differentially abundant between treatment groups (Supplemental Tables 3 and 4). These bacteria could play a significant role in the activity of the microbiome. Bacteria falling under the Clostriadiales Other group were consistently higher in the DC group relative to other treatments, and could represent an unmeasured source of functional differences between treatment groups. However, their metagenomic contribution to the community cannot be known without further characterization of their genome.

The relative lack of functional gene differences at day 28 could also be an indication that despite continuing taxonomic differences, the microbiome in each treatment is converging toward a similar metabolic pattern. Conservation of function across a variety of microbial profiles has been described in other studies, and extreme dysregulation of the microbiome may be required before severe or protracted functional changes occur (37).

Figure 5 illustrates this concept; while the bacterial treatments applied in this study affect both the taxonomic and inferred metagenomic composition of the microbiome, even statistically significant changes in function gene content are minor when compared to the taxonomic changes seen in the same animals.

In the present study, the chickens were all free of visible disease or stress, and it is possible their gut microbiota were functioning in their optimal range with or without treatment. It is also important to note these birds were not given a pathogen challenge or other stressor of any kind. The addition of vaccinations or probiotics to a chicken with a dysbiotic gut microbiome might yield more significant functional changes. Recent studies demonstrated that exposure of mice to antibiotics at an early age can have a deleterious effect on the diversity of the microbiome for several months following treatment (42). This study

showed no such effect from probiotic or vaccination. The value of select dietary treatments and management practices in poultry production may be their ability to increase the speed at which a disturbed or stunted microbiome is able to return to a normal functional state.

In conclusion, one-time oral inoculation with a live ST strain and daily ingestion of a probiotic feed supplement both alter the microbiome of growing chicks. These differences persisted throughout the study, and are centered on changes in the abundance of core microbes present in all treatment groups. The results of this trial suggest that common bacterial treatments such as probiotics and bacterial vaccines affect the taxonomic composition of the microbiome, but only have transient or small effects on the function and activity of the microbiome under non-stressed growth conditions. In contrast, as has been seen in other studies (7), age played a major role in the composition and richness of the bacterial community. Major shifts from day of hatch to day 14 centered on the early dominance of Enterobacteriaceae, followed by a transition to Firmicutes-dominated ceca.

Future studies will focus on understanding the functional and phylogenetic parameters of a normal developing microbiome, and to evaluate the effect of treatments like these on that normal range of microbial profiles.

ACKNOWLEDGMENTS

The authors wish to thank Dr. Amy Savage, North Carolina State University for helpful discussions and input related to data analysis. AB was supported in part by NIFA

National Needs Fellowship 2009-38420-05026, and generous gift from Star-Labs/Forage

Research, Inc. Additional support from NIFA Animal Health project NC07077.

Microbiome 16S data deposited with Qiita (http://qiita.ucsd.edu/; study 10291).

REFERENCES

1. Mohan B, Kadirvel R, Natarajan A, Bhaskaran M. Effect of probiotic supplementation on growth, nitrogen utilisation and serum cholesterol in broilers. Br Poult Sci (1996) 37:395–401.

2. Roos NM de, Katan MB. Effects of probiotic bacteria on diarrhea, lipid metabolism, and carcinogenesis: a review of papers published between 1988 and 1998. Am J Clin Nutr (2000) 71:405–411.

3. Marangon S, Busani L. The use of vaccination in poultry production. Rev Sci Tech Int Off Epizoot (2007) 26:265–274.

4. Saif YM. Diseases of Poultry. Wiley (2003).

5. Oakley BB, Lillehoj HS, Kogut MH, Kim WK, Maurer JJ, Pedroso A, Lee MD, Collett SR, Johnson TJ, Cox NA. The chicken gastrointestinal microbiome. FEMS Microbiol Lett (2014)n/a–n/a.

6. Waite DW, Taylor MW. Characterizing the avian gut microbiota: membership, driving influences, and potential function. Front Microbiol (2014) 5.

7. Oakley BB, Buhr RJ, Ritz CW, Kiepper BH, Berrang ME, Seal BS, Cox NA. Successional changes in the chicken cecal microbiome during 42 days of growth are independent of organic acid feed additives. BMC Vet Res (2014) 10.

8. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, Owens SM, Betley J, Fraser L, Bauer M, et al. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J (2012) 6:1621–1624.

9. David LA, Maurice CF, Carmody RN, Gootenberg DB, Button JE, Wolfe BE, Ling AV, Devlin AS, Varma Y, Fischbach MA, et al. Diet rapidly and reproducibly alters the human gut microbiome. Nature (2014) 505:559–563.

10. Hale VL, Tan CL, Knight R, Amato KR. Effect of preservation method on spider monkey (Ateles geoffroyi) fecal microbiota over 8 weeks. J Microbiol Methods (2015) 113:16–26.

11. Zhao L, Wang G, Siegel P, He C, Wang H, Zhao W, Zhai Z, Tian F, Zhao J, Zhang H, et al. Quantitative genetic background of the host influences gut microbiomes in chickens. Sci Rep (2013) 3:1163.

12. Youssef N, Sheik CS, Krumholz LR, Najar FZ, Roe BA, Elshahed MS. Comparison of Species Richness Estimates Obtained Using Nearly Complete Fragments and Simulated

Pyrosequencing-Generated Fragments in 16S rRNA Gene-Based Environmental Surveys. Appl Environ Microbiol (2009) 75:5227–5236.

13. Aronesty E. ea-utils : Command-line tools for processing biological sequencing data. (2011) Available at: http://code.google.com/p/ea-utils

14. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Peña AG, Goodrich JK, Gordon JI, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods (2010) 7:335–336.

15. Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinforma Oxf Engl (2010) 26:2460–2461.

16. DeSantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K, Huber T, Dalevi D, Hu P, Andersen GL. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl Environ Microbiol (2006) 72:5069–5072.

17. Caporaso JG, Bittinger K, Bushman FD, DeSantis TZ, Andersen GL, Knight R. PyNAST: a flexible tool for aligning sequences to a template alignment. Bioinforma Oxf Engl (2010) 26:266–267.

18. McDonald D, Price MN, Goodrich J, Nawrocki EP, DeSantis TZ, Probst A, Andersen GL, Knight R, Hugenholtz P. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J (2012) 6:610– 618.

19. Clarke K, Gorley R. PRIMER v6 PRIMER-E Ltd. Plymouth UK (2001)

20. Langille MGI, Zaneveld J, Caporaso JG, McDonald D, Knights D, Reyes JA, Clemente JC, Burkepile DE, Vega Thurber RL, Knight R, et al. Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences. Nat Biotechnol (2013) 31:814–821.

21. Kanehisa M, Goto S. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res (2000) 28:27–30.

22. Parks DH, Tyson GW, Hugenholtz P, Beiko RG. STAMP: statistical analysis of taxonomic and functional profiles. Bioinformatics (2014) 30:3123–3124.

23. Storey JD, Tibshirani R. Statistical significance for genomewide studies. Proc Natl Acad Sci (2003) 100:9440–9445.

24. Lee K, Lillehoj HS, Siragusa GR. Direct-Fed Microbials and Their Impact on the Intestinal Microflora and Immune System of Chickens. J Poult Sci (2010) 47:106–114.

25. Neal-McKinney JM, Lu X, Duong T, Larson CL, Call DR, Shah DH, Konkel ME. Production of Organic Acids by Probiotic Lactobacilli Can Be Used to Reduce Pathogen Load in Poultry. PLoS ONE (2012) 7:e43928.

26. Compendium NA. Compendium of Veterinary Products, 9th Edition 2006. North American Compendiums, Incorporated (2006).

27. PrimaLac. PrimaLac Direct-fed Microbials. (2015) Available at: http://www.primalac.com

28. Barnes EM, Impey CS, Cooper DM. Manipulation of the crop and intestinal flora of the newly hatched chick. Am J Clin Nutr (1980) 33:2426–2433.

29. Lev M, Briggs C a. E. The Gut Flora of the Chick. I. the Flora of Newly Hatched Chicks. J Appl Bacteriol (1956) 19:36–38.

30. Filippo CD, Cavalieri D, Paola MD, 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) 107:14691–14696.

31. Turnbaugh PJ, Bäckhed F, Fulton L, Gordon JI. Diet-Induced Obesity Is Linked to Marked but Reversible Alterations in the Mouse Distal Gut Microbiome. Cell Host Microbe (2008) 3:213–223.

32. Sergeant MJ, Constantinidou C, Cogan TA, Bedford MR, Penn CW, Pallen MJ. Extensive Microbial and Functional Diversity within the Chicken Cecal Microbiome. PLoS ONE (2014) 9:e91941.

33. Tlaskalová-Hogenová H, Štěpánková R, Kozáková H, Hudcovic T, Vannucci L, Tučková L, Rossmann P, Hrnčíř T, Kverka M, Zákostelská Z, et al. The role of gut microbiota (commensal bacteria) and the mucosal barrier in the pathogenesis of inflammatory and autoimmune diseases and cancer: contribution of germ-free and gnotobiotic animal models of human diseases. Cell Mol Immunol (2011) 8:110–120.

34. Willing BP, Van Kessel AG. Intestinal microbiota differentially affect brush border enzyme activity and gene expression in the neonatal gnotobiotic pig. J Anim Physiol Anim Nutr (2009) 93:586–595.

35. Dominguez-Bello MG, Blaser MJ, Ley RE, Knight R. Development of the Human Gastrointestinal Microbiota and Insights From High-Throughput Sequencing. Gastroenterology (2011) 140:1713–1719.

36. Jakobsson HE, Jernberg C, Andersson AF, Sjölund-Karlsson M, Jansson JK, Engstrand L. Short-Term Antibiotic Treatment Has Differing Long-Term Impacts on the Human Throat and Gut Microbiome. PLoS ONE (2010) 5:e9836.

37. Consortium THMP. Structure, function and diversity of the healthy human microbiome. Nature (2012) 486:207–214.

38. Kurokawa K, Itoh T, Kuwahara T, Oshima K, Toh H, Toyoda A, Takami H, Morita H, Sharma VK, Srivastava TP, et al. Comparative Metagenomics Revealed Commonly Enriched Gene Sets in Human Gut Microbiomes. DNA Res (2007) 14:169–181.

39. Burt DW. Chicken genome: Current status and future opportunities. Genome Res (2005) 15:1692–1698.

40. Salanitro JP, Blake IG, Muirehead PA, Maglio M, Goodman JR. Bacteria isolated from the duodenum, ileum, and cecum of young chicks. Appl Environ Microbiol (1978) 35:782.

41. McHardy IH, Goudarzi M, Tong M, Ruegger PM, Schwager E, Weger JR, Graeber TG, Sonnenburg JL, Horvath S, Huttenhower C, et al. Integrative analysis of the microbiome and metabolome of the human intestinal mucosal surface reveals exquisite inter-relationships. Microbiome (2013) 1:17.

42. Nobel YR, Cox LM, Kirigin FF, Bokulich NA, Yamanishi S, Teitler I, Chung J, Sohn J, Barber CM, Goldfarb DS, et al. Metabolic and metagenomic outcomes from early-life pulsed antibiotic treatment. Nat Commun (2015) 6.

Table 2.1. Similarity percentage analysis (SIMPER) of treatment groups at day 14a.

Average abundance % Contribution b Phylum Class Order Family Genus DC DP to dissimilarity d Firmicutes Bacilli Lactobacillales Lactobacillaceae Lactobacillus 1% 16% * 19.4 c Proteobacteria Gammaproteobacteria Enterobacteriales Enterobacteriaceae 16% 3% * 15.65 Firmicutes Clostridia Clostridiales Lachnospiraceae Ruminococcus 25% 26% 15.27 Firmicutes Clostridia Clostridiales Ruminococcaceae 19% 14% * 9.01 Firmicutes Clostridia Clostridiales Ruminococcaceae Oscillospira 11% 15% 8.58

DC VC Firmicutes Clostridia Clostridiales Lachnospiraceae 8% 16% * 15.4 Firmicutes Clostridia Clostridiales 4% 12% * 14.57 Firmicutes Clostridia Clostridiales Ruminococcaceae 19% 13% * 13.61 Firmicutes Clostridia Clostridiales Lachnospiraceae Ruminococcus 25% 24% 13.28 Proteobacteria Gammaproteobacteria Enterobacteriales Enterobacteriaceae 16% 9% * 13.12

DC VP Firmicutes Clostridia Clostridiales Lachnospiraceae 8% 39% * 25.58 Firmicutes Clostridia Clostridiales Lachnospiraceae Ruminococcus 25% 2% * 18.69 Firmicutes Clostridia Clostridiales Ruminococcaceae 19% 3% * 13.46 Proteobacteria Gammaproteobacteria Enterobacteriales Enterobacteriaceae 16% 5% * 9.07 Firmicutes Clostridia Clostridiales 4% 14% * 8.95 aTop 5 taxa shown for each comparison bPercent contribution to total dissimilarity between treatment groups under comparison cIf a sequence matches more than one possible taxon, classification stops at the next highest level dIndicates significance at P<0.05

Table 2.2. Similarity percentage analysis (SIMPER) of treatment groups at day 28a.

Average abundance % Contribution b Phylum Class Order Family Genus DC DP to dissimilarity c Firmicutes Clostridia Clostridiales Lachnospiraceae Ruminococcus 12% 19% 16.65 d Firmicutes Clostridia Clostridiales Ruminococcaceae 11% 18% * 12.94 e Firmicutes Clostridia Clostridiales Other Other 11% 5% * 9.57 Firmicutes Clostridia Clostridiales Ruminococcaceae Oscillospira 12% 8% * 9.22 Firmicutes Clostridia Erysipelotrichales Erysipelotrichaceae 2% 5% 7.73

DC VC Firmicutes Clostridia Clostridiales 12% 21% * 15.41 Firmicutes Clostridia Clostridiales Lachnospiraceae 15% 10% * 13.49 Firmicutes Clostridia Clostridiales Other Other 11% 4% * 12.26 Firmicutes Clostridia Clostridiales Ruminococcaceae Ruminococcus 5% 9% 7.53 Firmicutes Clostridia Clostridiales Lachnospiraceae Ruminococcus 12% 12% 7.45

DC VP Firmicutes Clostridia Clostridiales 12% 36% * 32.27 Firmicutes Clostridia Clostridiales Lachnospiraceae 15% 5% * 14.03 Firmicutes Clostridia Clostridiales Other Other 11% 1% * 13.05 Firmicutes Clostridia Clostridiales Ruminococcaceae 11% 16% * 6.85 Firmicutes Clostridia Clostridiales Lachnospiraceae Ruminococcus 12% 11% 5.11 aTop 5 taxa shown for each comparison bPercent contribution to total dissimilarity between treatment groups under comparison cIf a sequence matches more than one possible taxon, classification stops at the next highest level dIndicates significance at P ≤ 0.05 e”Other” indicates the sequence in question has not been assigned to a taxonomic group at that level

Table S2.1. Diet Composition

Feed component %

Ground Corn 62.66

Soybean Meal 28.4

Limestone 1.2

Dicalcium Phosphate (18.5% P) 1.45

Lysine HCl (78.5%) 0.24

DL-Methionine 0.2

Salt 0.4

Poultry TM (TM-90) 0.2

Choline Chloride (60%) 0.05

Vitamin Premix (NCSU-90) 0.1

Poultry Fat (Pet Food Grade) 5

Table S2.2. Number of samples included in analysis for each treatment and time-point

Treatment DC DP VC VP 0 5b 4b 0ab 1b 1 5b 4b 5b 2b 3 6 6 5b 5a Time-point 7 5b 6 6 6 14 6 5a 6 5b 28 6 5a 6 6 aSamples were removed due to low sequence coverage bSamples unavailable or removed due to low concentration/quality DNA

Table S2.3. Taxa removed from data set prior to PIRCRUSt analysis, day 14

Order Family Genus Vaccine-Probiotic Diluent-Control Vaccine-Control Diluent-Probiotic FDR P-valuea Clostridiales Other Other 0.041214068b 0.064510661 0.015479525 0.015402587 0.017 Coriobacteriales Coriobacteriaceae Eggerthella 6.04E-06 4.22E-06 0.00022303 0 0.017 Clostridiales Lachnospiraceae Other 0.011904706 0.008327372 0.022081509 0.00564618 0.017 Clostridiales Clostridiaceae Other 0.006184172 0.001117249 0.001332672 0.003856649 0.017 Bacillales Other Other 0 0 0 0.000271699 0.017 Clostridiales Ruminococcaceae Other 0.001114086 0.000300223 0.000643586 0.00058587 0.031 Bacillales Bacillaceae Other 5.08E-06 1.49E-05 0 0.015280536 0.072 Bacillales Bacillaceae Bacillus 0 0 0 0.001512628 0.072 Bacillales Planococcaceae Other 0 0 0 0.000587725 0.072 Bacillales Planococcaceae Planomicrobium 0 0 0 0.000115094 0.072 Bacillales Planococcaceae Rummeliibacillus 0 0 0 4.46E-05 0.072 Turicibacterales Turicibacteraceae Turicibacter 0 0 0 0.000108341 0.072 Rickettsiales Mitochondria Other 6.04E-06 0 2.90E-05 0 0.072 Thiotrichales Thiotrichaceae B46 0 3.17E-05 4.89E-06 0 0.072 Lactobacillales Other Other 0 1.84E-05 9.09E-06 0.000199248 0.072 Clostridiales Peptostreptococcaceae Other 0 7.43E-05 3.43E-05 0.000249114 0.072 Other Other Other 0.000492427 0.001540406 0.000857901 0.001166976 0.072 Enterobacteriales Enterobacteriaceae Other 0.002747429 0.002343169 0.001742816 0.000487024 0.114 Lactobacillales Enterococcaceae Other 2.77E-05 9.34E-06 3.32E-05 0.000111409 0.282 Other Other Other 1.22E-05 1.90E-05 0 0 0.419 Other Other Other 1.81E-05 0 0 0 0.493 Rhizobiales Brucellaceae Other 2.21E-05 0 0 0 0.493 Lactobacillales Leuconostocaceae 0 1.49E-05 2.25E-05 0 0.509

Table S2.3 Continued Clostridiales Clostridiaceae 3.02E-05 5.83E-06 5.17E-05 8.88E-05 0.509 Bacteroidales [Barnesiellaceae] Other 0 9.50E-06 0 0 0.509 Other Other Other 0 0 9.81E-06 0 0.509 Clostridiales Clostridiaceae Sarcina 0 1.17E-05 0 0 0.509 Clostridiales Lachnospiraceae Lachnospira 0 0 1.47E-05 0 0.509 Other Other Other 0 1.17E-05 9.09E-06 0 0.663 Other Other Other 0 9.92E-06 9.09E-06 0 0.663 aFalse-discovery-corrected P-values from Kruskal-Wallis tests. bValues represent relative abundance as a proportion of 1.

Table S2.4. Taxa removed from dataset prior to PICRUSt analysis, day 28

Order Family Genus Diluent-Control Diluent-Probiotic Vaccine-Control Vaccine-Probiotic FDR P-valueb Coriobacteriales Coriobacteriaceae Eggerthella 0b 0 0.001462938 0.000180123 0.008 Clostridiales Other Other 0.108313679 0.052878584 0.043422939 0.012453526 0.011 Enterobacteriales Enterobacteriaceae Other 0.000790157 0.001483962 0.001049698 0.000143583 0.176 Clostridiales Clostridiaceae Other 0.003119151 0.001172701 0.00253042 0.002301304 0.276 Clostridiales Ruminococcaceae Other 0.002158875 0.00093703 0.00156716 0.00147708 0.276 Enterobacteriales Enterobacteriaceae Citrobacter 4.23E-05 1.01E-05 0 0 0.276 Bacillales Bacillaceae Other 0 0.000680933 0 1.12E-05 0.276 Clostridiales Lachnospiraceae Butyrivibrio 1.94E-05 3.10E-05 0 0 0.276 Bacillales Bacillaceae Bacillus 2.59E-05 0 0 0 0.280 Clostridiales Peptostreptococcaceae Other 6.82E-06 1.60E-05 3.80E-05 0.000155064 0.307 Clostridiales Lachnospiraceae Other 0.014636359 0.014745188 0.017493251 0.011349562 0.325 Bacillales Other Other 0 2.14E-05 0 3.22E-05 0.345 Clostridiales Clostridiaceae Alkaliphilus 0 0 0 9.41E-06 0.358 Clostridiales Lachnospiraceae Lachnospira 2.22E-05 1.78E-05 0 0 0.358 Rickettsiales Mitochondria Other 3.59E-05 1.33E-05 4.71E-06 3.39E-05 0.450 Other Other Other 3.27E-05 9.05E-05 0 5.83E-05 0.458 Other Other Other 0 1.01E-05 0 0 0.458 Rubrobacterales Rubrobacteraceae Rubrobacter 1.36E-05 0 0 0 0.458 Bacillales Bacillaceae Anaerobacillus 3.31E-05 0 0 0 0.458 Bacillales Planococcaceae Planomicrobium 0 1.25E-05 0 0 0.458 Lactobacillales Other Other 0 2.14E-05 0 0 0.458 Clostridiales Lachnospiraceae Clostridium 0 1.60E-05 0 0 0.458

Table S2.4 Continued Erysipelotrichales Erysipelotrichaceae Other 1.36E-05 0 0 0 0.458 Burkholderiales Alcaligenaceae Sutterella 1.25E-05 0 0 0 0.458 Desulfobacterales Desulfobulbaceae 0 1.07E-05 0 0 0.458 Enterobacteriales Enterobacteriaceae Klebsiella 1.36E-05 0 0 0 0.458 Enterobacteriales Enterobacteriaceae Proteus 0 1.60E-05 0 0 0.458 Lactobacillales Enterococcaceae Other 6.40E-05 3.74E-05 1.88E-05 1.50E-05 0.473 Bacillales Planococcaceae Other 0 0 0 7.62E-06 0.532 Bacillales Planococcaceae Rummeliibacillus 0 0 0 1.52E-05 0.532 Clostridiales Ruminococcaceae Ethanoligenens 0 0 0 8.82E-06 0.532 Other Other Other 1.23E-05 0 0 7.62E-06 0.583 Clostridiales Lachnospiraceae Anaerostipes 0 2.50E-05 0 8.82E-06 0.583 Bacillales Planococcaceae 0 1.60E-05 0 1.90E-05 0.583 Pseudanabaenales Pseudanabaenaceae Leptolyngbya 1.32E-05 1.25E-05 0 8.82E-06 0.739 Other Other Other 0.001976425 0.00209155 0.001295505 0.001626973 0.840 aFalse-discovery-corrected P-values from Kruskal-Wallis tests. bValues represent relative abundance as a proportion of 1.

Figure 2.1. As the cecal microbiome develops, the dominant taxa shift from Gram negative to Gram positive bacteria. A heatmap of taxonomic groups present in untreated (DC) samples over time was generated with Qiime. The composition of the microbiome in DC animals was evaluated to identify trends in the development of the normal microbiome over time. There is a consistent decrease in the proportion of Enterobacteriaceae and Enterococcus over time, and an increase in levels of Clostridiales groups like Ruminococcus and Oscillospira. Sequence coverage was normalized to 16,260 reads/sample.

Figure 2.2. Age is the dominant factor in the composition of the microbiome. (A) Principal coordinate analysis of samples was conducted using Primer-E and samples were labeled based on age. Samples are clustered on two axes based on a multi-dimensional analysis of their sequence diversity and abundance. As the animals age, their microbiome increases in complexity but decreases in variability between samples, even between treatment groups. The effect of age was stronger than the effect of vaccine or probiotic. Differences between all time points were significant at permutational P-value < 0.05, with the exception of day 0 vs days 1 and 3. Coordinate loading for each principal coordinate shows the primary taxonomic groups contributing to each axis. Each data point represents a sample in the appropriate time point, and samples from all treatment groups are included in the analysis. (B) The phylum and class of sequences with an average relative abundance of 1% or greater are displayed by time point and treatment. All treatment groups started with high levels of Gammaproteobacteria which shifted with age into a Firmicutes-dominated community with large numbers of Clostridia. Taxonomy assignments were generated with QIIME, and PCoA plots were generated with Primer-E. Sequence coverage was normalized to 16,260 reads/sample.

Figure 2.3. Principal coordinate analysis and rarefaction analyses demonstrate the impact of the treatments over time. (A) Principal coordinate analysis generated with Primer-E demonstrate the effect of treatments at 1, 7, 14, and 28 days of age. There are no significant treatment differences at day 1. By day 7, treatment groups are statistically different based on PERMANOVA tests. Treatment groups cluster visually at day 14 and 28. All treatment groups at days 7, 14, and 28 were different at permutational P < 0.05. (B) Rarefaction of observed species (unique OTUs) at individual time points was conducted using QIIME, and demonstrate the rapid development of taxonomic diversity. Treatment groups show similar diversity at Day 1. At day 7, DP and VP tend to have lower diversity than VC (P = 0.078 and 0.054, respectively). At day 14, DP and VP diversity is significantly lower than DC and VC diversity (P < 0.05). By day 28, community diversity is similar between treatments. Sequence coverage was normalized for each time point individually: day 1 (16,577 reads/sample), day 7 (16,668 reads/sample), day 14 (20,263 reads/sample), day 28 (20,263 reads/sample).

Figure 2.4. At both day 14 and day 28, the VP group shows the greatest divergence in predicted metagenomic content. PICRUSt was used to generate a list of genes inferred to be present in the samples, their relative abundance, and the gene pathways with which they are associated. A heatmap was generated with STAMP. Samples were clustered using a nearest neighbor metric, and pathways were colored based on their percent abundance relative to all measured genes. All listed gene groups are significantly different between treatment groups with an effect size > 0.70 and FDR-corrected P < 0.01. The combination of vaccine and probiotic treatments stimulates changes in several gene groups and pathways at day 14 (A), namely increases in chemotaxis, two-component system, flagellar assembly, and bacterial motility. Decreases in the VP group include metabolic processes such as amino acid metabolism, DNA replication, and protein translation. (B) There are fewer significantly different pathways at day 28 and changes are largely related to cell metabolism.

Figure 2.5. Comparison of taxonomic and gene-group abundance trends at day 14 and day 28. Relative abundance of taxonomic groups and predicted abundance of functional gene groups demonstrate the stability of gene group abundances relative to changes in taxonomic groups. Relative abundance tables for taxa and gene group were assembled and used to generate area charts of all samples at days 14 and 28. The relative abundance of every identified taxonomic or functional gene group is shown for each sample; (A) predicted gene groups at day 14, (B) taxonomic groups at day 14, (C) predicted gene groups at day 28, (D) taxonomic groups at day 28. At day 14, there are clear taxonomic differences between treatments (A), and smaller changes in the abundance of a few gene groups (B). At day 28, taxonomic changes (C) are accompanied by few visible changes in gene group abundance (D).

CHAPTER 3: DIFFERENTIAL EFFECT OF DIET AND PROBIOTICS ON

MICROBIAL POPULATIONS ALONG THE AVIAN GASTROINTESTINAL

TRACT.2

2 AL Ballou, RA Ali, MD Koci. Applied and Environmental Microbiology. Submitted

ABSTRACT

As food and bacteria are consumed by the host, it moves through the various segments of the gastrointestinal (GI) tract. Each region has a unique role in digestion and absorption. The response of the microbes in these different regions to stimuli can affect the host as well as the microbial populations in subsequent regions. As the microbiome field moves from associating changing tax with host phenotypes, towards identifying specific bacterial-host interactions responsible for functional differences it is critical that we understand how the microbiome changes along the GI tract. To begin to address this, we collected digesta from six locations (crop, gizzard, duodenum, jejunum, ileum, and cecum) along the gastrointestinal tract of chickens receiving one of two diets with or without a probiotic. Analysis of 16S rDNA sequences revealed that GI location was the strongest predictor of similarity among samples, influencing the microbiome more than diet or probiotic treatment. Of the locations evaluated, the cecum had the highest phylogenetic diversity. However, the ileal microbiome had more treatment-related changes than the cecum. Furthermore, in silico metagenomics (PICRUSt) analysis suggested changes in the ileum had a greater potential for functional differences. The ileum of probiotic-fed animals had reductions in bacteria associated with inflammation and increased potential for production of the anti-inflammatory short chain fatty acid, butyrate. These data suggest changes in ileal microbiota are more closely related to changes in host physiology. This highlights the need for a better understanding of how specific treatments influence changes in the microbiome along the GI.

IMPORTANCE

The gut is often discussed as one organ, but functionally it is segmented into regions with distinct roles in digestion and absorption. These different regions result in distinct environments for bacteria to live. Changes in the distinct bacterial populations in any region can affect the host. Understanding the regional differences in these communities is critical to our ability to use the microbiome to maximize host health. In this study, we evaluated the microbial populations of six gut segments in chickens eating one of two diets, with or without a probiotic. Gut location played a stronger role in the makeup of the microbial population than the diets or probiotic; however, when examined individually, the ileum had the most probiotic-induced changes with the greatest potential for affecting the host. As we work to understand how probiotics affect host health, these results demonstrate our attention should focus on host-microbe interactions in the ileum.

INTRODUCTION

Every animal relies on their gut microbiome for normal function. The microbiome has the ability to regulate immune function, gastrointestinal (GI) development, metabolism, and more (1–4). Many questions remain, however, regarding how and where gut bacteria exert an effect on the host, and this lack of information currently limits our ability to determine the mechanics of probiotic-host interactions. Most microbiome-focused studies evaluate the microbe-rich feces or colon, but this likely over represents the organisms in this final region of the GI tract. The avian GI changes considerably from region to region, reflecting the physiological role of each segment. In the crop, consumed food is stored exposed to amylase and other salivary enzymes. As the process of digestion begins in the mildly acidic (pH of 4-5) environment, fermentative bacteria also bloom (5, 6). Food then moves quickly through the proventriculus where gastric acids and enzymes are secreted, followed by the gizzard, where the combination of muscular contractions and gastric acid further breaks down the particles in an environment with activated enzymes such as trypsin and chymotrypsin and a pH of 3.5-4 (7). The bulk of digestion and absorption occurs in the pH-neutral small intestine (SI), a region also rich in immune cells (7, 8). The cecum is the primary site of fermentation in the chicken, and supports as much as 10x more anaerobic bacteria than the other GI regions (7). Host-microbiome interactions in the SI and upper GI could have important impact on nutrition, GI health, and immune function. Understanding the microbial differences between regions of the GI tract is critical to our ability to regulate the host microbiome to boost health and productivity.

There are an ever-increasing number of products being used in humans and animals to regulate host health, through the augmentation of the microbiome. The most common approach has been through the use of probiotics; however, our understanding of where in the

GI tract these products exhibit their effect is limited. The body of research related to probiotics in poultry is growing, and probiotics in poultry have been shown to stimulate immune function, GI development, and performance (9–13). Previous studies in our laboratory demonstrated the addition of a specific probiotic consortium alters gut development and immune function in the small intestine and upper GI (10, 11, 14).

Additionally, this probiotic upregulates systemic immune activity, suggesting that changes in the small intestine may impact the whole animal (10). Further study of these probiotics species will demonstrate how they interact with the microbiome and host in these locations, enabling us to improve animal health through the microbiome.

The scientists and animal producers using these probiotics prefer a variety of diets based on availability, desired outcome, and economics. Not only do these diets come with nutritional variations that may impact microbial growth in the GI, they are also accompanied by diverse groups of bacteria associated with their individual components. These populations may contribute to the microbiome in some regions, but compete with the established populations in others. Several studies in humans and mice have shown diet to be an important factor in microbiome development and composition (15–17). These changes extended beyond phylogenetic profile impacting host gene expression and digestive activity, as well (17). However, not all dietary changes provoke a response in the microbiome (18).

Diet-based effects may be both specific to the GI location, and dependent on the nature of the diet.

The purpose of this study was to provide insight into regulation of microbiome composition through dietary treatments and host physiology. To accomplish this, we evaluated the impact of two nutritionally complete diets and a probiotic product on chickens and their microbiomes at several GI locations. The results suggest the microbiome of each major GI segment reflects its physiological environment, and that the small intestine is an important region for host-microbe interactions.

MATERIALS AND METHODS

Experimental Design

Twenty-four 1-day-old mixed sex broiler-type chicks were randomly assigned to 1 of

2 diets and 1 of 2 probiotic groups. Diets were a corn-soybean based diet obtained from

Purina (Diet A; LabDiet®, St. Louis, MO; Table 1), and a corn-soybean diet obtained from the NC State University feed mill (Diet B; Table 1). Within each diet, 6 animals received either the diet alone (CON) or the diet with the addition of 0.3% probiotic (PB). The probiotic, PrimaLac® contains Lactobacillus acidophilus, Lactobacillus casei, Enterococcus faecium, and Bifidobacterium bifidium (Star Labs Inc., Clarksdale, MO). It was added to the probiotic groups’ feed prior to the experiment (approximately 3x105 CFU/gram feed).

Animals in all groups were fed ad libitum for four weeks. Animals in the same treatment group were cohoused in pens on mesh floors. All animals were maintained and euthanized

under an approved protocol from the North Carolina State University Institutional Animal

Care and Use Committee (OLAW #D16-00214).

Sample Collection

Six chickens from each treatment group were euthanized on day 28 and the contents of 6 GI locations (crop, gizzard, duodenum, jejunum, ileum, and cecum) were sampled.

Digesta were placed into microfuge tubes and frozen at -80°C prior to DNA isolation. The crop was subsampled, while the low volume of the gizzard necessitated collecting the entirety of its contents. Duodenal samples were collected at the mid-point of the duodenal loop, jejunal samples were collected at the mid-point between the end of the duodenal loop and Meckel’s diverticulum, and ileal samples were collected from the mid-point between

Meckel’s diverticulum and the ileo-cecal junction. Cecal samples were expressed from the top of cecum only.

DNA Isolation and 16S Sequencing

DNA was isolated using the MO BIO Power Soil kit (MO BIO, Carlsbad, CA) with the following modifications: a 10-minute, 65° C incubation step was added and samples were homogenized for 45s at 5100 RPM using the provided garnet bead-containing tubes and a

Precellys 24 homogenizer (Precellys, Montigny-le-Bretonneux, France). DNA was quantified using a NanoDrop 2000 spectrophotometer (NanoDrop, Wilmington, DE).

Library preparation and MiSeq 151x151 paired-end sequencing (Illumina, San Diego, CA)

were performed by the Argonne National Laboratory Institute for Genomics and Systems

Biology Next Generation Sequencing Core using a protocol and primers recommended and previously described by the Earth Microbiome project and others (19). Primers spanned the

V4 region of the 16S rRNA gene (515F: GTGYCAGCMGCCGCGGTAA, 806R:

GGACTACHVGGGTWTCTAAT). Studies estimating microbial composition using the V4 region report diversity measurements comparable to those obtained with full-length 16S sequences

(20).

Sequence Data Analysis

The unpaired raw sequencing reads were paired and filtered using EA-Utils (Aronesty

E. 2011. ea-utils : Command-line tools for processing biological sequencing data.). Paired reads were processed using the QIIME suite of tools (v 1.8.0) (21) and barcode matching and quality filtering were conducted prior to picking operational taxonomic units (OTUs). OTUs were picked using an open-reference protocol. Prior to further analysis, samples were rarefied to 11,900 sequences/sample; samples with low sequence coverage (below 11,900 sequences/sample; 4 out of 129 samples) or low sequence quality (1 ileal sample) were excluded from analysis. Taxonomic assignments, alpha and beta diversity metrics, and statistical analysis were generated using QIIME and Primer-E (Clarke K, Gorley R. 2001.

PRIMER v6 PRIMER-E Ltd., Ivybridge, UK) as described (21, 22) as well as the Phyloseq R package (23).

The figures used in this study were generated in R. PCoA (principal coordinates analysis) was conducted using Bray-Curtis distances in the Phyloseq package. Permutational multivariate analysis of variance (PERMANOVA) was conducted using the PERMANOVA+ add-on to Primer-E. Main and pair-wise tests were conducted using up to 1000 permutations of residuals under a reduced model. The Shannon diversity index for each location and treatment was generated using samples rarified to 11900 sequences/sample. The compare_alpha_diversity.py script from QIIME was used to conduct nonparametric two- sample t-tests on the Shannon diversity estimates between sample locations using 999 permutations, employing a false discovery rate (FDR) correction.

Metagenomic inferences from 16S amplicon data were made using the QIIME suite of tools (21, 24–26), PICRUSt (27) and KEGG (28). Statistics and visualization of functional data were depicted using STAMP (29), as described (22). Benjamini-Hochberg

FDR correction was applied to all comparisons between treatments, and significance is defined at q ≤ 0.05. Principal coordinates analysis plots and extended error bar plots were generated at the 3rd level of the KEGG hierarchy (gene groups) (28).

Accession numbers

The sequences generated for this study have been submitted to the Sequence Read Archive, and are available for download. Accession numbers are sequential from SAMN06029764-

SAMN06029892. SRA study SRP095024.

RESULTS

Evaluation of microbial profiles along the digestive tract.

In order to obtain an understanding of how location and dietary treatments impact microbiome composition, 16S rDNA sequence data (6 GI locations x 6 animals x 4 treatments) were analyzed for differences in alpha and beta diversity (Fig 1 and 2). Principal coordinates analysis (PCoA) and PERMANOVA of the whole dataset revealed greater within-location similarity than within-treatment similarity (Fig 1A; Table S1, Table S2). For example, the cecum, gizzard, and crop populations were highly similar from one animal to another, regardless of treatment. Samples from the small intestinal locations (duodenum, jejunum, and ileum) were more variable within location than the cecum, gizzard, and crop.

Samples from each diet were also evaluated for their microbial composition and these samples clustered together, separate from digesta samples, while the PB supplement samples clustered with the gizzard samples (Fig 1A).

Treatment-based clustering of samples was more apparent when each location was examined individually, eliminating the effect of location (Fig 1B-H). The microbial composition of the feed samples was not measurably affected by 0.3% PB supplementation.

However, the composition of the PB supplement was notably different than that of either feed (Fig 1B). Addition of PB to diet B had minimal impact on the overall community at each location (see Table S1), while A:PB was significantly different from A:CON in the ileum and cecum (Fig 1G and H, Table S1). Comparison also revealed differences based on diet, rather than PB status (Fig 1C, D, G, and H; Table S1).

Changes in taxonomic richness were evaluated by location and treatment. Based on the Shannon diversity index, the avian cecum had highest diversity (Fig 2) of any location, followed by the PB supplement. The small intestine locations showed similar richness estimates.

Approximately 20% of the sequences identified in the PB supplement were

Lactobacillus, while Streptophyta, likely derived from the plant material in the probiotic’s carrier material, makes up another 19%. These and a diverse assortment of 12 other taxa made up 95% of the 16S sequences found in the probiotic (data not shown). Bacillus, parent class to Enterococcus faecium, comprised about 2% of the supplement, and very few

Bifidobacterium sequences were recovered, either in the supplement or in the GI samples.

The phylogenetic composition of the microbiome was further evaluated at each GI location in order to determine specific taxonomic changes related to changes in the GI environment (Fig 3). The dominant phyla in both feeds were Cyanobacteria and

Proteobacteria. The crop contained high levels of Firmicutes, while the gizzard was more similar to the feed profile, with high levels of Cyanobacteria and Proteobacteria (Fig 3A).

The crop, small intestine and cecum were dominated by Firmicutes, with varying levels of

Cyanobacteria and Proteobacteria (Fig3A-C). The cecum also contained moderate levels of

Bacteroides (14-31%), which did not make up a major portion of the microbiome in any other location except the B:PB ileum (Fig 3F and G). Though Firmicutes were abundant in most GI segments, Clostridiales dominated in the cecum, while Lactobacillales made up the majority of Firmicutes OTUs in the small intestine and crop (Fig 4).

Diet and probiotic affect microbiome composition in multiple locations.

Statistical analysis of the taxonomic and predicted functional composition of the microbiome was conducted by location and treatment in order to determine which locations experienced treatment-specific changes. Figure 5 illustrates the effect of diet and PB on the abundance of “Candidatus Savagella” (previously grouped with “Candidatus Arthromitus”) in the gut (30). The majority of “Candidatus Savagella” sequences were located in the ileum, with a small fraction in the jejunum (Fig 5A). Levels of “Candidatus Savagella” were over 2-fold higher in the A:CON group than in the B:CON group, and addition of PB reduced “Candidatus Savagella” in the ileum in both diets (Fig 5A). PB treatment also increased total Ruminococcaeae, Lachnospiraceae, and Peptostreptococcaceae (Fig 5B).

Ileal Lactobacillus abundance was variable, ranging from 21-81% (data not shown).

Diet affected the microbiome of some gut locations. Cecal samples from Diet A showed higher levels of Lachnospiraceae (P=0.002) than Diet B, including Blautia, Dorea, and other undefined genera. Irrespective of diet, PB decreased an uncharacterized

Clostridiales group, dropping it from 21 to 12% (data not shown). In the A:PB group,

Lactobacillus was significantly higher (P=0.05) at 15% of the total, compared with 1-3% in the other treatments. Rikenellaceae showed an opposite trend in the A:PB treatment; it dropped relative to A:CON, from 26-31% to 14% of total sequences (P=0.11) (data not shown).

Diet and probiotic interact to alter potential metagenomic content in the ileum.

The genomic potential of each sample was evaluated to identify possible changes in functional capacity due to treatment. The PICRUSt-derived inferred metagenomic content of each sample is summarized in Figure 6. PCoA scaling is highly similar between phylogenetic and metagenomic analysis (Fig 1A and 6A). The cecum maintained a high level of functional similarity between all samples while also being distinct from other locations (Fig

6A). Two functional gene groups in the cecum were impacted by treatment; expression of glycine/serine/threonine metabolism genes and protein folding/processing genes were decreased in the A:PB group relative to A:CON (Fig 6B).

In the ileum, 15 gene groups were affected by the interaction of diet and PB (Fig 7).

No differences were identified between CON and PB groups in Diet B (data not shown). The potential for increased expression of glycolysis/gluconeogenesis, ion-coupled transporters, benzoate metabolism, and butanoate metabolism was seen in the A:PB group, while bacterial secretion systems, branch chain amino acid degradation, and phenylpropanoid biosynthesis were higher in A:CON (Fig 7B).

The probiotic supplement metagenome was different from that of the feed samples in over 100 gene groups. Overall percent of sequences related to metabolism declines in the supplement relative to both diets (50.4% vs 51.7% and 52.7%, not shown), while sequences matching genes in the subset of carbohydrate metabolism increase in relative abundance

(9.2% vs 8.0% and 8.2%, not shown). Of the 7 gene groups in the carbohydrate metabolism category significantly different between the supplement and diets, propionate and butyrate

metabolism genes were two of the most affected groups, both increasing by 2-fold or more

(Fig S1).

DISCUSSION

Across plant and animal systems, the correlation of certain microorganisms with improved growth, development, and disease resistance has driven research to mining these commensal organisms for new therapies. To do this we need to be able to differentiate between correlation and causation with respect to microbiota and host changes. As the host microbiome is perturbed, one can observe changes in dozens if not hundreds of taxa.

However, many of these are likely second and third order changes. As digestion and absorption continues in the upper GI, it will change the nutrients available for bacteria in the lower GI. To realize the therapeutic possibilities of the microbiome, we need to identify where in the host the microbiome is affected as well as begin to appreciate what taxonomic changes are most responsible for changes in host physiology.

Previous studies in our laboratory, using the same commercial probiotic, demonstrated changes in gut development, increased immune response, increased turnover of

ATP in circulating immune cells, energy partitioning within the host (10, 11, 14), changes in bacterial populations (22), as well as changes in host mucosal gene expression (10).

Curiously, the effect on mucosal gene expression varied by intestinal location. In the ileum of birds fed the probiotic we observed a gene expression profile associated with decreased inflammation and a Th-2 type cytokine profile (10); while the cecum of probiotic-fed animals

had an expression profile associated with inflammation (unpublished observation). To begin to understand how the probiotic was affecting the microbiota we initially examined the development of the cecal microbiome over time and how it correlated with changes in host physiology (22). These studies demonstrated changes in several taxa; however, these changes, while statistically significant and interesting, failed to explain concurrent changes in host physiology.

In the current study, we assayed for changes in the microbiome all along the GI to better appreciate the impact the probiotic has on entire ecosystem, and how changes in one section of the GI may influence other regions. In addition, we included two nutritionally complete basal diets. This produced two distinct “control” microbial populations which were then changed by the probiotic treatment. This provided us with the ability to generalize our results across multiple diet sources.

Analysis of microbiome composition by location

Relative to the cecum, little has been published about the composition of the crop, gizzard, and proximal SI, and any similarities they might have to one another or the biome of feed stocks (18, 31). Without this baseline information on the normal microbiome of the upper GI and SI, it is difficult to develop therapies that would target the microbiomes of these regions. Phylum-level evaluation of the microbiome in each location would suggest a high level of similarity between the crop, small intestine, and cecum (Fig 3). Figure 4 provides more insight into how microbial communities change longitudinally along the

digestive tract. Despite being the first gastric compartment and a holding location for newly ingested feed, the crop maintains a Lactobacillus-based microbial community distinct from the feed-derived taxa. The gizzard appears to be much more influenced by diet-derived microbes (Fig 3A-C, Fig 4). However, previous evaluations of the gizzard microbiome did not report high levels of Cyanobacteria, instead listing Lactobacillus and other Firmicutes as dominant taxa (32, 33). The discrepancy may be due to diet source, sequencing, and/or analysis protocols. If Cyanobacteria (likely chloroplasts from plant material) sequences were filtered from the gizzard samples, Lactobacillus would be the most abundant genera in the gizzard, as in the crop and SI. Unfortunately, the microbiomes of the diets in these earlier studies were not reported so it is difficult to draw any conclusions. As no evaluation of diets from these other studies were available, it is not possible to ascertain the cause of the differences.

Though there are no statistical differences between SI locations, analysis of the ileum suggests it experiences less variation in microbial composition than the duodenum and jejunum. The SI is characterized by high levels of Firmicutes (58-61% Lactobacillus, 9-23% various Clostridiales genera) with declining levels of Cyanobacteria (8%-23%, Fig 3D-F).

Most evaluations of avian and mammalian SI microbes list Lactobacillus as the most abundant genus in the region, and these bacteria are thought to play an important role in pathogen resistance by regulating pH and producing various bacteriocins (34).

Similar studies evaluating the avian cecal microbiome consistently list Bacteroidetes and Clostridia as high abundance members, with Prevotella and Proteobacteria levels high

in some studies (18, 33, 35, 36). Though Clostridiales account for an average of 60% of total sequences, Prevotella was not a major member of the cecum in this experiment. Fellow

Bacteroidales taxa Rikenellaceae was present with a range of 15-30%. Lactobacillus levels varied between 1.5-15% (data not shown), in sharp contrast to the dominance of

Lactobacillus in the SI. The distinct compositional differences between the cecum, gizzard, crop, and small intestines evidenced by this (Table S2; Fig 1A) and other studies suggest that it would be prudent for researchers to identify and evaluate the locations likely to be impacted by treatment variables, rather than using cecal or fecal samples as representatives

(18, 33, 36). Even within the SI, regions may respond differently to treatment (37).

The ileal microbiome responds to diet and probiotic treatment

Currently, the cecum and large intestine are the most heavily-sampled GI regions for microbial ecology studies (35, 36, 38). We are learning much about how these microbial communities are formed and how they communicate with their hosts. The avian ileum, immediately anterior to the ceca, is remarkably different in composition than the cecum, and based on other studies may more closely resemble fecal communities (39). In this study, the ileum and the cecum were the only locations that mounted a consistent response to the probiotic (Table S1).

The single largest treatment effect in the ileum was a sharp reduction in “Candidatus

Savagella” abundance with PB supplementation (Fig 5A). This genus was previously identified morphologically as segmented filamentous bacteria (SFB). Histological evaluation

of PB-treated animals from previous studies have shown visible reduction in SFB colonization in the ileum of growing birds consistent with this observation (11). While these bacteria are common in the ileum, and not considered pathogenic, excessive colonization may stimulate damaging and energetically expensive inflammatory responses in otherwise healthy subjects. Monocultures of SFB inoculated into gnotobiotic animals stimulate TH17- type responses in the ileum, and can generate experimental autoimmune disease in susceptible models (40, 41). In contrast to “Candidatus Savagella”, several other taxa within the Clostridiales group (Lachnospiraceae, Ruminococcaceae, and

Peptostreptococcaceae) were increased after treatment (Fig 5B). The significance of these increases in the ileum with PB treatment is unclear. Lachnospiraceae and Ruminococcaeae contain many butyrate producers, and are common commensals in the cecum and colon of a variety of animals (34, 42). Increasing Lachnospiraceae and Ruminococcaeae could indicate an increase in ileal short chain fatty-acid levels, which are associated with reduced inflammation in humans and mice (43, 44).

Despite clear PB-associated changes in microbiome composition, levels of

Lactobacillus, the primary genus in the PB supplement, were variable (21-81%) and not clearly responsive to treatment; Lactobacillus did increase between A:CON and A:PB, but

Lactobacillus was high (80%) in B:CON and levels actually decrease to 50% in the PB group. Despite this, the changes seen in response to PB treatment could support the health of the ileal environment by reducing potentially inflammatory bacteria, while increasing

fermentative bacteria. These results are in line with previous studies demonstrating this probiotic improves GI morphology and decreases host inflammatory genes in the ileum (10).

The cecal microbiome responds to diet more than probiotic

The impact of changes in the cecal microbiome on the host are less clear. Shifts in abundance, due to either diet or PB, occurred in bacteria that are commonly found in the cecum. Diet A showed higher levels of Lachnospiraceae than Diet B samples, and irrespective of diet, PB decreased an uncharacterized Clostridiales group by nearly 50%.

Increases in beneficial fermenters such as Lachnospiraceae in Diet A over Diet B may be the result of higher levels of wheat germ and alfalfa meal in A, as several studies on the role of fiber type and level show (45, 46). Despite PB-related increases in Lachnospiraceae and

Ruminococcaceae in the ileum, cecal populations do not appear to be sensitive to PB treatment. A possible explanation for this is the relatively low levels of PB taxa in the cecum

(Lactobacillus ranges from 1-15%, relative to the 58-61% range in the ileum). Lactobacillus levels are typically much higher in the SI than cecum independent of probiotic treatment, and much of the research done with this probiotic suggests it exerts more influence on the upper

GI and SI rather than the cecum (10, 11, 14).

Probiotics affect imputed microbiome activity

In an attempt to identify microbial signals capable of stimulating host immune cells, we conducted an in silico prediction of the functional capacity of the microbiome in the

ileum and cecum. We expected taxonomic changes in the community would alter its functional capacity by adding certain genes to the metagenome and removing others.

Multidimensional scaling (PCoA) of the metagenome is very similar in structure to that of the 16S data (compare Fig 1A to Fig 6A), however unlike the cecum-specific 16S PCoA (Fig

1H), the cecal metagenome shows no clustering or statistical difference based on treatment

(Fig 6B), in agreement with similar studies evaluating 16S vs metagenomic response to dietary treatments (31). Only two gene groups were significantly different between any treatment group. The A:PB group showed a modest reduction over A:CON in genes associated with amino acid metabolism and protein folding and processing (data not shown).

This result demonstrates the level of functional redundancy in most microbiomes: there were several taxonomic changes such that there may have been expected functional changes, but with a large amount of overlap in the genomes of many bacteria. A change in taxonomy may only result in minor changes in metagenomics content, an effect that has been noted before

(22).

In contrast, a similar survey of ileal samples reveals more meaningful changes in metagenomic content between A:CON and A:PB (Fig 7A). The changes include increases in genes for glycolysis and gluconeogenesis, ion transporters, and butanoate metabolism (Fig

7B). The increase in butanoate-related genes is the most intriguing result relative to the probiotic used in this study. This probiotic reduces ileal mucosal expression of NF-κB and

IL-12RB (markers of inflammation), while increasing the anti-inflammatory cytokine IL-10

(10). Studies on the impact of butanoate (butyrate) on intestinal tissues report that it

suppresses NF-κB, while butyrate-treated human cell lines express less IL-12 and more IL-10

(47, 48). Further studies will be necessary to determine the role of butyrate in the host response to PB, and whether any increase in butyrate production is a direct result of the PB or a downstream effect of treatment mediated by other microbes. This predicted increase in butyrate genes may be due to the increased abundance of butyrate producing

Lachnospiraceae and Ruminococcaceae taxa seen with PB supplementation (Fig 5B) (49).

However, the PB supplement contains far greater potential for butyrate metabolism (Fig S1) than either feed alone, indicating the effect could also stem directly from the bacteria present in the supplement. There is currently no information in the literature regarding the production of butyrate by the species in this product, however one popular Lactobacillus probiotic product VSL#3 is known to stimulate butyrate in mice (50).

CONCLUSION

The microbial composition and environment of each intestinal location are strong factors in host function and gene expression. It is the sum total of these changes that determine how a host responds to treatments targeting the microbiome. Based on the results of this study, animals may benefit from probiotic supplementation even in the absence of digestive stress. Probiotic-treated animals showed reduced “Candidatus Savagella” and increased potential for butyrate in the ileum, both of which may contribute to less inflammation even in healthy animals. Although we can only speculate why in this study, the source of diet had an impact on microbes at some points along the GI tract. This makes

clear the extent to which even subtle differences in diet preparation or composition may lead to broader effects in the animal. With this in mind, diet composition and microbiome should be a consideration in any study involving gut microbial populations. Further study of this probiotic’s effects on the host and microbiome will focus on the ileum to identify specific microbes and microbe-derived metabolites associated with host physiological and gene expression changes, as well as the biological significance of these host changes.

ACKNOWLEDGMENTS

The authors wish to thank Drs. Amy Savage and Rob Dunn of North Carolina State

University, and Dr. J. Christopher Ellis of In Silico LLC for helpful discussions and input related to data analysis. ALB was supported in part by NIFA predoctoral fellowship 2016-

67011-25168 and NIFA National Needs Fellowship 2009-38420-05026. This work was also supported in part by NIFA Animal Health projects NC07080 and NC07077, as well as an unrestricted gift from Star Labs/Forage Research, Inc.

REFERENCES

1. Backhed F, Ding H, Wang T, Hooper LV, Koh GY, Nagy A, Semenkovich CF, Gordon JI. 2004. The gut microbiota as an environmental factor that regulates fat storage. Sci STKE 101:15718.

2. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, Gordon JI. 2006. An obesity-associated gut microbiome with increased capacity for energy harvest. Nature 444:1027–31.

3. Atarashi K, Tanoue T, Oshima K, Suda W, Nagano Y, Nishikawa H, Fukuda S, Saito T, Narushima S, Hase K, Kim S, Fritz JV, Wilmes P, Ueha S, Matsushima K, Ohno H, Olle B, Sakaguchi S, Taniguchi T, Morita H, Hattori M, Honda K. 2013. Treg induction by a rationally selected mixture of Clostridia strains from the human microbiota. Nature 500:232–236.

4. Round JL, Mazmanian SK. 2010. Inducible Foxp3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proc Natl Acad Sci 107:12204– 12209.

5. Fuller R. 1977. The importance of Lactobacilli in maintaining normal microbial balance in the crop. Br Poult Sci 18:85–94.

6. Hinton A, Corrier DE, Deloach JR. 1992. In Vitro Inhibition of Salmonella typhimurium and Escherichia coli 0157:H7 by an Anaerobic Gram-positive Coccus Isolated from the Cecal Contents of Adult Chickens. J Food Prot 55:162–166.

7. Engberg RM, Hedemann MS, Jensen BB. 2002. The influence of grinding and pelleting of feed on the microbial composition and activity in the digestive tract of broiler chickens. Br Poult Sci 43:569–579.

8. Cornes JS. 1965. Number, size, and distribution of Peyer’s patches in the human small intestine. Gut 6:225–229.

9. Dalloul RA, Lillehoj HS, Shellem TA, Doerr JA. 2003. Enhanced mucosal immunity against Eimeria acervulina in broilers fed a Lactobacillus-based probiotic. Poult Sci 82:62–66.

10. Qiu R, Croom J, Ali RA, Ballou AL, Smith CD, Ashwell CM, Hassan HM, Chiang C- C, Koci MD. 2012. Direct fed microbial supplementation repartitions host energy to the immune system. J Anim Sci 90:2639–2651.

11. Chichlowski M, Croom WJ, Edens FW, McBride BW, Qiu R, Chiang CC, Daniel LR, Havenstein GB, Koci MD. 2007. Microarchitecture and spatial relationship between bacteria and ileal, cecal, and colonic epithelium in chicks fed a direct-fed microbial, PrimaLac, and salinomycin. Poult Sci 86:1121–1132.

12. Wang Y, Gu Q. 2010. Effect of probiotic on growth performance and digestive enzyme activity of Arbor Acres broilers. Res Vet Sci 89:163–167.

13. Haghighi HR, Gong J, Gyles CL, Hayes MA, Zhou H, Sanei B, Chambers JR, Sharif S. 2006. Probiotics Stimulate Production of Natural Antibodies in Chickens. Clin Vaccine Immunol 13:975–980.

14. Chichlowski M, Croom J, McBride BW, Daniel L, Davis G, Koci MD. 2007. Direct- Fed Microbial PrimaLac and Salinomycin Modulate Whole-Body and Intestinal Oxygen Consumption and Intestinal Mucosal Cytokine Production in the Broiler Chick. Poult Sci 86:1100–1106.

15. Zimmer J, Lange B, Frick J-S, Sauer H, Zimmermann K, Schwiertz A, Rusch K, Klosterhalfen S, Enck P. 2012. A vegan or vegetarian diet substantially alters the human colonic faecal microbiota. Eur J Clin Nutr 66:53–60.

16. Muegge BD, Kuczynski J, Knights D, Clemente JC, González A, Fontana L, Henrissat B, Knight R, Gordon JI. 2011. Diet Drives Convergence in Gut Microbiome Functions Across Mammalian Phylogeny and Within Humans. Science 332:970–974.

17. 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. 2014. Diet rapidly and reproducibly alters the human gut microbiome. Nature 505:559–563.

18. Ranjitkar S, Lawley B, Tannock G, Engberg RM. 2016. Bacterial Succession in the Broiler Gastrointestinal Tract. Appl Environ Microbiol 82:2399–2410.

19. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, Owens SM, Betley J, Fraser L, Bauer M, Gormley N, Gilbert JA, Smith G, Knight R. 2012. Ultra- high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J 6:1621–1624.

20. Youssef N, Sheik CS, Krumholz LR, Najar FZ, Roe BA, Elshahed MS. 2009. Comparison of Species Richness Estimates Obtained Using Nearly Complete Fragments and Simulated Pyrosequencing-Generated Fragments in 16S rRNA Gene- Based Environmental Surveys. Appl Environ Microbiol 75:5227–5236.

21. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Peña AG, Goodrich JK, Gordon JI, Huttley GA, Kelley ST, Knights D, Koenig JE, Ley RE, Lozupone CA, McDonald D, Muegge BD, Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ, Walters WA, Widmann J, Yatsunenko T, Zaneveld J, Knight R. 2010. QIIME allows analysis of high-throughput community sequencing data. Nat Methods 7:335–336.

22. Ballou AL, Ali RA, Mendoza MA, Ellis JC, Hassan HM, Croom WJ, Koci MD. 2016. Development of the Chick Microbiome: How Early Exposure Influences Future Microbial Diversity. Vet Infect Dis 2.

23. McMurdie PJ, Holmes S. 2013. phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLOS ONE 8:e61217.

24. Caporaso JG, Bittinger K, Bushman FD, DeSantis TZ, Andersen GL, Knight R. 2010. PyNAST: a flexible tool for aligning sequences to a template alignment. Bioinforma Oxf Engl 26:266–267.

25. Edgar RC. 2010. Search and clustering orders of magnitude faster than BLAST. Bioinforma Oxf Engl 26:2460–2461.

26. McDonald D, Price MN, Goodrich J, Nawrocki EP, DeSantis TZ, Probst A, Andersen GL, Knight R, Hugenholtz P. 2012. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J 6:610– 618.

27. Langille MGI, Zaneveld J, Caporaso JG, McDonald D, Knights D, Reyes JA, Clemente JC, Burkepile DE, Vega Thurber RL, Knight R, Beiko RG, Huttenhower C. 2013. Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences. Nat Biotechnol 31:814–821.

28. Kanehisa M, Goto S. 2000. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res 28:27–30.

29. Parks DH, Tyson GW, Hugenholtz P, Beiko RG. 2014. STAMP: statistical analysis of taxonomic and functional profiles. Bioinformatics 30:3123–3124.

30. Thompson CL, Mikaelyan A, Brune A. 2013. Immune-modulating gut symbionts are not “Candidatus Arthromitus.” Mucosal Immunol 6:200–201.

31. Oakley BB, Lillehoj HS, Kogut MH, Kim WK, Maurer JJ, Pedroso A, Lee MD, Collett SR, Johnson TJ, Cox NA. 2014. The chicken gastrointestinal microbiome. FEMS Microbiol Lett 360:100–112.

32. Rehman HU, Vahjen W, Awad WA, Zentek J. 2007. Indigenous bacteria and bacterial metabolic products in the gastrointestinal tract of broiler chickens. Arch Anim Nutr 61:319–335.

33. Yeoman CJ, Chia N, Jeraldo P, Sipos M, Goldenfeld ND, White BA. 2012. The microbiome of the chicken gastrointestinal tract. Anim Health Res Rev 13:89–99.

34. Bjerrum L, Engberg RM, Leser TD, Jensen BB, Finster K, Pedersen K. 2006. Microbial community composition of the ileum and cecum of broiler chickens as revealed by molecular and culture-based techniques. Poult Sci 85:1151–1164.

35. Wei S, Morrison M, Yu Z. 2013. Bacterial census of poultry intestinal microbiome. Poult Sci 92:671–683.

36. Oakley BB, Buhr RJ, Ritz CW, Kiepper BH, Berrang ME, Seal BS, Cox NA. 2014. Successional changes in the chicken cecal microbiome during 42 days of growth are independent of organic acid feed additives. BMC Vet Res 10.

37. Smirnov A, Perez R, Amit-Romach E, Sklan D, Uni Z. 2005. Mucin dynamics and microbial populations in chicken small intestine are changed by dietary probiotic and antibiotic growth promoter supplementation. J Nutr 135:187.

38. Yeoman CJ, White BA. 2014. Gastrointestinal tract microbiota and probiotics in production animals. Annu Rev Anim Biosci 2:469–486.

39. Danzeisen JL, Clayton JB, Huang H, Knights D, McComb B, Hayer SS, Johnson TJ. 2015. Temporal Relationships Exist Between Cecum, Ileum, and Litter Bacterial Microbiomes in a Commercial Turkey Flock, and Subtherapeutic Penicillin Treatment Impacts Ileum Bacterial Community Establishment. Front Vet Sci 2.

40. Wu H-J, Ivanov II, Darce J, Hattori K, Shima T, Umesaki Y, Littman DR, Benoist C, Mathis D. 2010. Gut-Residing Segmented Filamentous Bacteria Drive Autoimmune Arthritis via T Helper 17 Cells. Immunity 32:815–827.

41. Lee YK, Menezes JS, Umesaki Y, Mazmanian SK. 2010. Proinflammatory T-cell responses to gut microbiota promote experimental autoimmune encephalomyelitis. Proc Natl Acad Sci 108:4615–4622.

42. Miquel S, Martín R, Rossi O, Bermúdez-Humarán L, Chatel J, Sokol H, Thomas M, Wells J, Langella P. 2013. Faecalibacterium prausnitzii and human intestinal health. Curr Opin Microbiol 16:255–261.

43. Manokas T, Fromkes JJ, Sundaram U. 2000. Effect of chronic inflammation on ileal short-chain fatty acid/bicarbonate exchange. Am J Physiol Gastrointest Liver Physiol 278:G585-590.

44. Frank DN, Amand ALS, Feldman RA, Boedeker EC, Harpaz N, Pace NR. 2007. Molecular-phylogenetic characterization of microbial community imbalances in human inflammatory bowel diseases. Proc Natl Acad Sci 104:13780–13785.

45. Holscher HD, Caporaso JG, Hooda S, Brulc JM, Fahey GC, Swanson KS. 2015. Fiber supplementation influences phylogenetic structure and functional capacity of the human intestinal microbiome: follow-up of a randomized controlled trial. Am J Clin Nutr 101:55–64.

46. Hooda S, Boler BMV, Serao MCR, Brulc JM, Staeger MA, Boileau TW, Dowd SE, Fahey GC, Swanson KS. 2012. 454 Pyrosequencing Reveals a Shift in Fecal Microbiota of Healthy Adult Men Consuming Polydextrose or Soluble Corn Fiber. J Nutr 142:1259–1265.

47. Segain J, de la Bletiere DR, Bourreille A, Leray V, Gervois N, Rosales C, Ferrier L, Bonnet C, Blottiere H, Galmiche J. 2000. Butyrate inhibits inflammatory responses through NFκB inhibition: implications for Crohn’s disease. Gut 47:397–403.

48. Säemann MD, Böhmig GA, Österreicher CH, Burtscher H, Parolini O, Diakos C, Stöckl J, Hörl WH, Zlabinger GJ. 2000. Anti-inflammatory effects of sodium butyrate on human monocytes: potent inhibition of IL-12 and up-regulation of IL-10 production. FASEB J 14:2380–2382.

49. Vital M, Howe AC, Tiedje JM. 2014. Revealing the Bacterial Butyrate Synthesis Pathways by Analyzing (Meta)genomic Data. mBio 5:e00889-14.

50. Yadav H, Lee J-H, Lloyd J, Walter P, Rane SG. 2013. Beneficial Metabolic Effects of a Probiotic via Butyrate-induced GLP-1 Hormone Secretion. J Biol Chem 288:25088– 25097.

Table 3.1. Comparison of NRC-compliant Diets.a

Diet Ab Diet Bc NRCd

Crude Protein (%) 21.6 23 20-23

Crude Fat (%) 4.1 6 -e

Calcium (%) 0.92 0.9 -

Methionine (%) 0.46 0.558 0.38-0.5

Lysine (%) 1.22 1.29 1.1

Metabolizable Energy (kcal/g) 3.02 3.15 3.0-3.2 aBoth diets composed of ~1.4:1 corn:soybean within the NRC recommendations for broiler chicks bLab animal diet purchased from Purina Mills. cFeed produced at the NC State research feed mill dNational Research Council recommended nutritional requirements2 eNo recommendation given

Table S3.1. P-values of treatment effects by location

PERMANOVA P-valuesa

Comparison Crop Gizzard Duodenum Jejunum Ileum Cecum

Diet B CONb, Diet B PBc 0.541 0.414 0.889 0.062 0.174 0.296

Diet B CON, Diet A CON 0.113 0.224 0.103 0.152 0.027 0.141

Diet B CON, Diet A PB 0.224 0.038 0.723 0.088 0.254 0.012

Diet B PB, Diet A CON 0.012 0.09 0.083 0.55 0.05 0.219

Diet B PB, Diet A PB 0.189 0.024 0.637 0.234 0.093 0.004

Diet A CON, Diet A PB 0.354 0.174 0.376 0.736 0.005 0.004 a Permutational Multivariate Analysis of Variance bControl cProbiotic

100

Table S3.2. P-values for location effects

Comparison PERMANOVAa P-value

Crop, Gizzard 0.001 Crop, Jejunum 0.76 Crop, Cecum 0.001 Crop, Duodenum 0.001 Crop, Ileum 0.026 Crop, Feed 0.001 Gizzard, Jejunum 0.001 Gizzard, Cecum 0.001 Gizzard, Duodenum 0.001 Gizzard, Ileum 0.001 Gizzard, Feed 0.046 Jejunum, Cecum 0.001 Jejunum, Duodenum 0.182 Jejunum, Ileum 0.363 Jejunum, Feed 0.001 Cecum, Duodenum 0.001 Cecum, Ileum 0.001 Cecum, Feed 0.001 Duodenum, Ileum 0.554 Duodenum, Feed 0.001 Ileum, Feed 0.001 a Permutational Multivariate Analysis of Variance

101

H A Cecum Duodenum Gizzard Jejunum 0.2 Cecum Tissue Crop Feed Ileum 0.1 0.6 0.0 -0.1

-0.2 PCoA2 [14.1%] PCoA2 -0.3 -0.6 -0.4 -0.2 0.0 0.2 0.3 PCoA1 [20.2%] G 0.50 Ileum 0.25

0.0 0.00

PCoA2 [25.8%] PCoA2 -0.25

-0.50 -0.25 0.00 0.25

PCoA2 PCoA2 [21.2%] PCoA1 [27.9%] -0.3 F 0.4 Jejunum 0.2

0.0 -0.6

-0.2 PCoA2 [19.8%] PCoA2 -0.50 -0.25 0.00 0.25 -0.4 -0.50 -0.25 0.00 0.25 PCoA1 [28.7%] PCoA1 [35.1%] B C D E 0.25 Feed Crop 0.2 Gizzard Duodenum 0.00 0.2 Treatment 0.1 0.00 Feed A:CON 0.0 Feed A:PB -0.25 0.0 Feed B:CON Feed B:PB -0.1 -0.2 -0.25 PB Supplement -0.50

-0.2

PCoA2 [35.3%] PCoA2 [19.8%] PCoA2 [20.2%] PCoA2 [22.7%] PCoA2 -0.4 -0.75 -0.3 -0.4 -0.2 0.0 0.2 -0.25 0.00 0.25 0.50 -0.3 -0.2 -0.1 0.0 0.1 0.2 -0.2 0.0 0.2 0.4 PCoA1 [61.6%] PCoA1 [49.2%] PCoA1 [43.3%] PCoA1 [26.8%] Treatment A:CON A:PB B:CON B:PB Figure 3.1. GI Location is the primary variable by which microbial samples cluster. Chickens were fed 2 diets, with or without PB supplementation for 28 days. Digesta was collected from the feed, crop, gizzard, duodenum, jejunum, ileum, and cecum and microbial diversity assessed by 16S rDNA sequencing. Principal coordinates analysis (PCoA) was conducted using Bray-Curtis distances, first with all samples colored by location, then within each location colored by treatment. P-values were derived from Permutational multivariate analysis of variance (PERMANOVA). (A) All locations comparisons are different except Jejunum:Ileum, Jejunum:Duodenum, Jejunum:Crop, and Duodenum:Ileum. PCoA of the sample data demonstrates relative differences in microbial diversity between locations. Samples from all 4 treatment groups are included in each location. (B-H) Samples from each location were evaluated to illustrate differences between treatment groups by location. A = Diet A Purina Mills diet, B = NC State feed mill diet, PB = probiotic (0.3% w/w), CON = no probiotic.

102

5 ab Treatment A:CON A:PB 4 B:CON b a ab ab B:PB Feed A:CON

Shannon 3 Feed A:PB Feed B:CON Feed B:PB PB Supplement 2 a

Feed Crop Gizzard Duodenum Jejunum Ileum Cecum Richness by Intestinal Location

Figure 3.2. Phylogenetic diversity is greatest in the cecum. Microbial DNA was isolated from the feed, crop, gizzard, duodenum, jejunum, ileum, and cecum and assessed by 16S rDNA sequencing. The Shannon diversity estimate for each treatment group in each location was calculated using data normalized to 11900 sequences/sample. Outliers are shown as points outside boxes. Diet A = Purina Mills diet, Diet B = NC State feed mill diet, PB = probiotic (0.3% w/w), CON = no probiotic. PB Supplement = probiotic product alone. Locations with different superscripts are different, FDR-corrected P<0.05. Significance was determined using nonparametric T-tests.

103

Figure 3.3. Location-specific communities respond to dietary and probiotic treatments. Microbial DNA was isolated from the feed, crop, gizzard, duodenum, jejunum, ileum, and cecum and assessed by 16S rDNA sequencing. Samples from each location were grouped by treatment and the top 10 relatively abundant phyla were evaluated for each treatment group. Stacked bars represent all OTUs arranged from most abundant (bottom) to least (top), with OTUs colored by phylum. In every location, Firmicutes, Cyanobacteria, Bacteroidetes, and Proteobacteria are most abundant. The gizzard (C) and cecum (G) were phylogenetically distinct from the crop and intestines (D-F); the gizzard contained higher levels of feed- derived Cyanobacteria and Proteobacteria (A), in contrast to the remaining locations, where Firmicutes were dominant. In the cecum, Bacteroides and Tenericutes were more abundant in all treatment groups than in other locations.

104

105

Figure 3.4. Community composition changes distally, reflecting changes in input and function. Data were normalized to 11900 sequences/sample and samples were combined by location. The 100 most abundant OTUs were ordered within location by Bray-Curtis distances, and reported at the order level. Streptophyta and Rickettsiales from the feed are seen in the upper GI and SI, but not cecum. Clostridiales increase distally, while Lactobacillus decrease.

106

Bacteroidales Clostridiales Clostridiales Bacteroidales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Bacteroidales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Bacillales Clostridiales RF39 Clostridiales Clostridiales Clostridiales Clostridiales RF39 Clostridiales Clostridiales Enterobacteriales Abundance Lactobacillales Rhizobiales Sphingomonadales 65536 Enterobacteriales Rickettsiales Streptophyta 4096 Streptophyta Rickettsiales Rickettsiales

Order 256 Streptophyta Streptophyta Streptophyta Rickettsiales 16 Streptophyta Streptophyta Lactobacillales 1 Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Lactobacillales Clostridiales Clostridiales Clostridiales Erysipelotrichales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Clostridiales Desulfovibrionales Clostridiales

Feed Crop Ileum Cecum Gizzard Jejunum Duodenum Sample Location

107

Figure 3.5. “Candidatus Savagella” levels drop in the ileum of probtiotic-treated birds. (A) Samples were normalized to 11,900 sequences/sample and “Candidatus Savagella” abundance was measured in each treatment group across six locations. Stacked bars represent the number of “Candidatus Savagella” sequences in each treatment, colored by location. “Candidatus Savagella” decreases significantly (P<0.05) between CON and PB groups. (B) The composition of the 100 most abundant OTUs from the order Clostridiales was evaluated in the ileum. Clostridiales were divided into families (see top label), and further divided within family by genus (see legend). Clostridiales includes Clostridiaceae, of which “Candidatus Savagella” is the primary member in this experiment, as well as hindgut fermenters Lachnospiraceae and Ruminococcaeae. Each pair of stacked bar charts represents one family of Clostridiales and the OTUs within each bar chart are colored by genus (see legend). Gray represents OTUs unassigned at the genus level (defaults to family). Clostridiaceae were significantly decreased (P<0.05), however, other families presented here were not significantly different between treatments.

108

109

Cecum Duodenum Gizzard Jejunum A Tissue B Crop Feed Ileum Treatment A:CON A:PB B:CON B:PB

0.3 0.10

0.2

0.05 0.1

0.0 0.00

PCoA2 [9.3%] PCoA2 [25.8%]

-0.1 -0.05

-0.25 0.00 0.25 -0.3 -0.2 -0.1 0.0 0.1 PCoA1 [55.8%] PCoA1 [76%] Figure 3.6. Samples cluster by location in both functional and 16S gene analysis. PICRUSt (phylogenetic investigation of communities by reconstruction of unobserved states) analysis was conducted on the whole dataset as well as subsets, generating a list of functional genes and their predicted changes based on 16s rDNA content. (A) Bray-Curtis distances between samples were used to generate a PCoA plot of all samples. As with taxonomic-based PCoA, location is the strongest determinant of functional profile. Increased variability in the PICRUSt analysis results in far fewer significant differences between treatment groups (not shown). (B) PCoA of cecal samples shows no differences based on treatment; only two functional gene groups were affected by treatment in the cecum (not shown).

110

Treatment A:CON A:PB A B

0.2 Benzoate degradation

Butanoate metabolism

Dioxin degradation

Glycan biosynthesis and metabolism 0.1 Glycolysis / Gluconeogenesis

Nucleotide metabolism

One carbon pool by folate

0.0 Other ion-coupled transporters PCoA2 [12.4%] Phenylpropanoid biosynthesis

Secretion system

Synthesis and degradation of ketone bodies -0.1 Transcription related proteins

Valine, leucine and isoleucine degradation

-0.4 -0.2 0.0 0.2 0.00 0.03 0.06 0.09 PCoA1 [67.7%] Abundance Figure 3.7. Treatments are associated with predicted functional changes in the ileum. PICRUSt analysis of ileal samples reveals changes in several gene groups between the Diet A CON and Diet A PB groups. (A) PCoA of ileal samples from Diet A CON and Diet A PB samples demonstrates that the functional gene composition of the treatment groups is dependent on diet treatment more than probiotic treatment. (B) After individual KEGG orthologs were compiled into functional groups, predicted metagenomics content from the Diet A treatments was evaluated on an individual gene group basis; individual functional groups affected by probiotic treatment are organized by alphabetical order. Functional groups are broad categorical groups containing genes associated with a set of pathways and/or functions. All shown predicted changes are statistically significant at an FDR- corrected q-value less than 0.05 using Welch’s uncorrected T-test. Similar analysis of interactions between the Diet B treatments did not reveal any significant alterations. Abundance=percent of total genes/treatment.

111

Butanoate Genes in Feed Treatment P=3.58e-5 A:CON Feed

0.2 A:PB Feed

B:CON Feed

0.1 B:PB Feed

PB Supplement Proportion of Sequences(%) of Proportion

A:PB Feed B:PB Feed A:CON Feed B:CON Feed PB Supplement Treatment

Figure S3.1. Abundance of butanoate metabolism genes is increased in the PB supplement. Feed and supplement samples were evaluated for changes in metagenomic content using PICRUSt. Bacteria associated with the PB supplement have over 5x abundance of butanoate metabolism genes.

112

CHAPTER 4: PROBIOTICS ALTER ENERGY CONSUMPTION AND GENE

EXPRESSION IN CHICKEN LYMPHOCYTES

113

ABSTRACT

The gastrointestinal (GI) microbiome plays an important role in the development and function the GI and the immune system. Probiotics provide promising treatments for improving animal health, but their modes of action are still largely unknown. Previous research in our laboratory has demonstrated supplementation with a lactic acid bacteria probiotic product alters host energy partitioning in immune tissues, including increased ATP production and consumption in circulating leukocytes. These changes were associated with a more rapid antibody response to antigen. To better understand the communication between probiotics and the immune system our laboratory has developed an in vitro model to characterizing factors involved. We hypothesized that the changes observed in immune tissue activity and energy metabolism were regulated by a probiotic-stimulated factor in the serum. In the current study we examined the ability of serum isolated from probiotic-fed animals to augment ATP production in vitro. Chicken cell lines were cultured for 4 days in media supplemented with serum from treated and control birds. Hepatocytes cultured in serum from probiotic-fed birds showed no response, while macrophages showed decreased

ATP concentrations (15%; P<0.05) and B and T-cell lines had higher concentrations of ATP

(12% and 43%; P < 0.05) compared to controls. Transcriptomic analysis of the avian T-cell cell line CU205 following incubation with serum from probiotic-treated animals was consistent with an increase in cell survival and differentiation. Collectively, the results of this study suggest probiotic supplementation upregulates a soluble signal in circulation capable of regulating lymphocyte function. Further studies will evaluate these and other

114

potential compounds as the source of probiotic-associated increases in immune cell activity, and seek to determine their origin.

115

INTRODUCTION

While many scientists have explored the impact of the gut microbiome on host metabolism, digestion, and intestinal dysregulation (1, 2), only in the last decade have they given much attention to how the gut microbiome regulates immune activity, especially in food animals (3–5). This makes gut microbiome-host interactions an untapped well for new approaches to improve animal health. Multiple studies have described how altering the microbiome affects gut-associated lymphoid tissues (GALT; (6–9) and it is becoming clear that the microbiome also affects systemic immune function (10–12). However, we know little about how the complex microbiome communicates with the immune system.

We have used a Lactobacillus-based probiotic as a driver of microbial change in the avian gut to better understand how the microbiome impacts immune function. In the course of our research, this probiotic has been shown to affect both the host and the microbiome in multiple GI locations (13–15). In the ileum, the probiotic increases anti-inflammatory IL-10, while pro-inflammatory IL-12RB2 and NF-κB decrease (15). At the same time, segmented filamentous bacteria, associated with TH17-type inflammatory responses, decline (13).

Changes to immune function were also observed outside the gut, as probiotic-fed animals mounted an antigen-specific antibody responses at least 7 days faster than in control animals

(15). The ability of Lactobacillus and Bifidobacterium-based probiotics to promote antigen- specific humoral immune responses has been reported by numerous groups (8, 10, 16).

Study of the metabolic activity of these animals shows the thymus of probiotic-treated animals consumes more oxygen even in the absence of an immune challenge, indicating a

116

higher rate of energy consumption in immune tissues. This increase in thymus respiration is offset by a decrease in respiration in the small intestine, resulting in no change in respiration in the animal as a whole (17). Further evaluation of immune cells has shown that peripheral blood mononuclear cells (PBMC) from these animals produce and consume more ATP per unit time than PBMC from control animals (15). This research suggests that the probiotic in question stimulates circulating lymphocytes and that the increased energetic cost of this stimulation is offset by a compensatory decrease in energy consumed by the GI (15, 17). It is unclear at this time how the probiotic communicates with the host to stimulate these cells, and whether other facets of the immune response are affected beyond antibody production.

The purpose of this research was to evaluate how the probiotic stimulates systemic immune cells, and to better understand what these changes mean for the animal. To begin to address this, we evaluated the ability of serum from probiotic- and control-fed animals to stimulate chicken lymphocytes. Further, we evaluated the transcriptome of these cells in order to characterize their response to probiotic treatment. This research expands our understanding of how probiotics may be used to benefit not only the gut, but the whole animal.

MATERIALS AND METHODS

Animals

One-day-old mixed sex broiler-type chickens were randomly assigned to either a regular diet (CON) or the CON diet with the addition of 0.3% probiotic (PB; PrimaLac®;

117

Lactobacillus acidophilus, Lactobacillus casei, Enterococcus faecium, Bifidobacterium bifidium; Star Labs Inc., Clarksdale, MO), resulting in the addition of approximately 3x105 colony-forming units (CFU) of probiotic bacteria/gram feed. All animals were fed ad libitum for four weeks. Animals were maintained and euthanized as described (15) under an approved protocol from the North Carolina State University Institutional Animal Care and

Use Committee (OLAW #D16-00214).

Primary peripheral blood mononuclear cells (PBMC) isolation

PBMC were isolated from the whole blood of 28 day old chickens (5 animals/treatment group) by collection into heparinized tubes followed by density centrifugation through Ficoll-Paque (density = 1.073 g/ml; GE Healthcare Life Sciences,

Pittsburgh, PA), as previously described (18). Briefly, 2 ml of blood was diluted with 2 ml phosphate-buffered saline (PBS), which was layered over 4 ml Ficoll-Paque in glass tubes.

Tubes were centrifuged at 400 x g for 40 minutes at room temperature. PBMC were collected from the interface between the plasma and Ficoll-Paque layers.

Serum collection and processing

Whole blood was collected from 6 animals per treatment at 14, 21, and 28 days of age. The blood was incubated at room temperature for 4-6 hours to allow clotting, then centrifuged at 2500 x g for 15 minutes at 4°C. Serum was collected following centrifugation and stored at -20°C. Aliquots of the whole serum obtained from CON- or PB-treated animals

118

was pooled by equal volumes within treatment group prior to use in in vitro assays. Subsets of pooled serum were treated with a combination of butanol and di-isopropyl ether (BDIPE;

35:65 v/v; Fisher Scientific, Hampton, NH) to remove lipid components. Two volumes of

BDIPE were added for each volume of serum, and samples were incubated at 4 °C for 60 min with rocking, as described (19), to extract unesterified fatty acids, triglycerides, cholesterol, and phospholipids. Following incubation, the aqueous phase was transferred to a new tube for later use. The organic phase containing the extracted lipids was discarded.

Alternatively, subsets of serum from each timepoint were subjected to molecular weight (MW) size fractionation using 15ml Vivaspin ultrafiltration tubes fitted with polyethersulfone membranes (MW cutoffs of 5 kDa, 10 kDa, and 30 kDa), as directed

(Sartorius, Goettingen, Germany). Serum samples from the CON- and PB-treated groups were added to individual tubes and centrifuged at 3220 x g for 60-120 minutes, as needed to achieve a ratio of 20% volume in upper fractions, and 80% in lower fractions based on volumetric graduations on the filter tubes. Specific pathogen free (SPF) chicken serum

(Sigma-Aldrich, St. Louis, MO), used for normal cell culture, was also fractionated in order to generate reciprocal fractions for treatment serum fractions (Fig S4.1). For example, the lower-MW PB serum fraction was combined with the upper-MW SPF serum fraction (80% and 20% v/v respectively) or the upper-MW PB serum fraction was added to the lower-MW

SPF fraction (20% and 80% v/v respectively) in order to restore the full complement of normal proteins to the serum sample.

119

Cell lines

Chicken lymphoblastoid T-cell line, Cu205, (20) and lymphoblastoid B-cell line,

RP9, (21) were maintained in RPMI supplemented with 1% fetal bovine serum (FBS), 10%

SPF chicken serum, and 2 mM L-glutamine at 41 ° C, 6% CO2. Chicken hepatocarcinoma cell line, LMH, (22) was maintained in DMEM with 10% chicken serum and 1% FBS at

37°C and 6% CO2. Macrophage cell line, MQ-NCSU, (23) was maintained in LM Hahn’s media supplemented with 10% FBS, 8% chicken serum, 2mM L-glutamine, and 10 µM 2- mercaptoethanol at 37°C and 6% CO2.

Cellular ATP quantification in cell lines

PBMC and cell lines were assayed for differences in ATP content. PBMC were evaluated immediately following isolation and quantification, while cell lines were cultured for 2 or 4 days in media supplemented with 10% whole or fractionated serum pooled from

CON- or PB-treated animals. For both PBMC and cell lines intracellular ATP levels were quantified using an ATP-dependent luciferase assay CellTiterGlo (Promega, Madison, WI) following the manufacturer’s instructions. Briefly, cells were counted using a hemocytometer and seeded into 6-10 replicate wells of a 96-well plate. Cell density at plating was 100,000 cells/well for PBMC and 5000 cells/well for all cell lines. Following plating of PBMC or 2-4 days of incubation of cell lines, cells were treated with CellTiterGlo reagent (1:1 v/v), shaken at 120 RPM at room temperature for 2 minutes and incubated at room temperature for 10 minutes. Luminescence was then measured as relative light units

120

(RLU) using a Thermo Fluoroskan Ascent FL (Thermo Scientific, Waltham, MA). RLU values were converted to ATP concentration using an ATP standard curve 0-250 nmol/well.

Assays with avian cell lines were conducted following a 4-day incubation with serum or processed serum. The suspension cell lines (CU205 and RP9 cells) have a growth rate 2-4x that of the adherent cell lines (LMH and MQ-NCSU) and were not contact inhibited; therefore at 48 hours these cells were recounted and passaged into wells at 5000 cells/well.

To account for potential differences in cell growth between wells of cell lines an additional cell titration assay was conducted on replicate wells at the time ATP was measured. Briefly, the CyQuant NF (Thermo Scientific, Waltham, MA) reagent was added 1:1 to 6-10 replicate wells in each treatment and to a serial 2-fold dilution of cells. After a 45 min incubation at

37°C, mean fluorescence intensity (MFI; 485/520nm ex/em) was recorded for each well

(Fluoroskan Ascent FL, Thermo Scientific, Waltham, MA) in order to generate a cell/well average for each treatment. Results reported as RLU or ATP pmol per cell.

RNA Isolation and Sequencing

CU205 cells were cultured with CON whole serum, PB whole serum, Con <30 kDa serum fraction, or PB <30 kDa serum fraction. Following 2 and 4 days of incubation cells were collected and total RNA was isolated as directed for the NucleoSpin RNA isolation kit

(Machery-Nagel, Duren, Germany). Ribosomal RNA was removed from total RNA, and libraries were generated from the remaining RNA using Illumina TruSeq kits (San Diego,

CA). Library preparation and sequencing were performed by the Genomic Sciences

121

Laboratory at North Carolina State University across 2 lanes of an Illumina HiSeq. Each sample was sequenced on both lanes, and the two files merged prior to analysis. Duplicate fastq files for each sample from each HiSeq lane were merged to yield a single file per sample, and quality-based filtering and trimming was performed using FASTQC (24). Group quality metrics were generated using MultiQC (25). The merged raw sequences for each sample are available through the Sequence Read Archive (project accession SRP095299).

Transcriptomic Analysis

The Tuxedo protocol (26) was used to aligned and analyze sequences, with the exception that HISAT2 (27) with default parameters was used for spliced alignments in place of TOPHAT. The output was used to assemble transcripts and generate FPKM estimates with CuffLinks. Sequences were assembled against the Gallus_gallus-5.0 genome assembly

(GenBank assembly GCF_000002315.4). CuffDiff and CummeRbund were used for differential abundance analysis and visualization, as previously described (26, 28, 29).

Network-based analysis of gene expression changes was conducted with Ingenuity

Pathway Analysis (IPA; Qiagen, Venlo, Netherlands). Differentially expressed genes were evaluated for their impact on known and predicted gene pathways, networks, and functions.

Briefly, CuffDiff gene expression data was input into the software package, and filtered based on a minimum FPKM of 5, and a maximum FDR-corrected q-value of 0.05. IPA’s

Core Analysis function was used to compare the experimental dataset with gene expression values from primary and immortal lymphocyte literature. Gene networks and functions

122

identified by IPA were significantly different at p < 0.05. Predicted activation or inhibition of functions and upstream regulators have an IPA-assigned activation z-score of 1.5 or higher, unless otherwise stated (30).

RESULTS AND DISCUSSION

Circulating PBMC have more ATP/cell in PB-treated animals

There are numerous reports, across multiple host species, of probiotics affecting immune function; however, the mechanisms involved are not understood. Previous research in our laboratory demonstrated immune tissues from PB-treated animals have elevated oxygen consumption and PBMC produce and consume more ATP than CON-treated animals

(15). This suggests probiotic-mediated enhanced immune responses could be due, at least in part, to increased energy available for immune cell activity. The current study was designed to begin to understand the mechanism behind this phenomenon. Initially, PBMC were isolated from 28 day old CON and PB-treated broiler chickens and assayed for ATP content immediately after isolation. PBMC from PB-treated animals generate 18% more ATP- dependent luminescence than cells from control animals (P < 0.01, Fig 4.1). These results were consistent with previously reported probiotic-induced change in PBMC ATP (15).

In vitro lymphocyte cell lines grown in serum from PB-treated animals have more ATP

The observation that consumption of a probiotic supplemented diet changes the amount of ATP in parental immune cells leads to the question; how is the signal

123

communicated from the intestine to immune cells in circulation? We hypothesized that the

ATP-stimulating signal was a soluble factor in the blood. To test this, we assayed the ability of serum from PB-treated and CON-treated animals to alter ATP levels in avian cells in vitro.

Using a panel of cell lines representing different host tissues; LMH (hepatoctyes), MQ-

NCSU (macrophages), and RP9 (B-cell-like) and CU205 (T-cell-like) cells, we assayed for increases in ATP per cell following 2 and 4 days of incubation with serum from PB-treated animals as compared to control. Each were grown in media containing 10% v/v serum from either PB-treated or CON-treated animals in place of their normal SPF chicken or fetal bovine maintenance serum.

The response to serum was consistent between day 2 and day 4 of incubation, but difference between cells treated with serum from PB-treated animals and CON-treated animals was more pronounced at day 4 with the exception of LMH cells which showed no treatment-dependent ATP changes at either time point (Fig 4.2). There was no consistent difference in proliferation between treatment groups in any cell type. The MQ-NCSU cells grown in serum from PB-treated animals had lower ATP per cell, as compared with control, while both lymphocyte cell lines (CU205 and RP9) grown in serum from PB-treated animals were found to have increased ATP per cell (Fig 4.2) as compared to those grown in serum from CON-treated animals. These results suggest there is some soluble factor or factors capable of modulating the amount of ATP/cell of immune cells in the serum of PB-treated animals. This effect appears to be specific to immune cells, as there was no change in ATP levels in the LMH cell line. However, the effect on immune cells appears to be dependent on

124

the cell type. It is unclear if the differences between the macrophage cell line and the two lymphocyte cell lines represent cell-specific responses to the same signal or the presence of multiple cell-type-specific factors in the serum. While the differential response of immune cell subsets is important to understand, for the purposes of this study we chose to focus on the increase in ATP in the Cu205 cell line. The macrophage response will be the focus of subsequent studies.

To ensure that the ATP-modulating effect observed was related to probiotic supplementation and not some other aspect of our experimental design, the Cu205 cells were assayed for changes in ATP using serum from PB- and CON-treated animals from different experiments, different genetic backgrounds (broiler-type and layer-type), as well as using different basal diets. In each of these cases the serum from PB-treated animals induced an increase in ATP/cell as compared to serum from CON-treated animals (data not shown).

These results suggest the phenomenon is linked to the use of this probiotic.

Physical and biochemical characterization of serum

The observation that serum from PB-treated animals can modulate the level of ATP in lymphocyte cell lines leads to the question: how? Initial experiments assayed for differences in glucose showed no significant difference between the serum from PB-treated and control animals (data not shown). Additionally, the amount of glucose in the basal culture media far exceeded that added by the serum. Likewise, one would expect an increase in glucose or other energy source to have a similar impact on all cells. Given that serum

125

from PB-treated animals appears to have different effects on lymphocytes, macrophages and hepatocytes, we hypothesized this factor is more immune specific. Unfortunately, there are thousands of proteins, hormones, and metabolites in serum that may regulate immune responses (31, 32).

To begin to identify the factor(s) responsible for the change in lymphocyte ATP, we subjected the serum to a series of treatments designed to remove or enrich for the active compound. To determine if the active component(s) could be lipid-based, organic solvents were used to extract lipids from day 28 serum. Cell viability was normal following incubation in BDIPE-treated serum, and delipidation appeared to have no effect on serum proteins when compared to SPF serum (Fig S4.2). Incubation of CU205 cells with BDIPE- treated serum from CON- and PB-treated animals produced lower ATP/cell in the cells exposed to PB-treated serum as compared to control (P < 0.05; Fig 4.3).

Alternatively, serum from both CON-treated and PB-treated animals was subjected to physical separation based on molecular weight. In order to provide a complete serum profile to cells in culture, SPF serum was also fractionated, and used to complement fractions from experimental serum (Fig S4.1). Fractions above and below 30 kDa were tested in CU205 cultures for 4 days. ATP levels per cell were 31% higher (P < 0.01) in the under 30 kDa PB treatment than the under 30 kDa CON fractions (Fig 4.4), suggesting the serum factors involved in the increased ATP/cell fall below this molecular weight. Comparison of ATP in cells grown in experimental serum above 30 kDa showed no difference between treatment groups (Fig 4.4). Subsequent attempts to fractionate the serum using membranes with

126

molecular cutoffs of 5 and 10 kDa demonstrated no difference between CON- and PB-treated fractions (data not shown). This suggests there may be multiple serum components of different molecular weights needed to stimulate ATP levels. Alternatively, analysis of the fractionated sera by SDS-PAGE suggest the proteins around a stated molecular cutoff maybe retained in the membrane (Fig S4.2). This is consistent with the manufacturer’s recommendations to use a membrane with a molecular cutoff 50% smaller than the protein to be collected in the upper fraction. It should also be noted that there were no differences in whole serum or fractions proteins bands among CON-treated, PB-treated, or SPF serum visible by SDS-PAGE (Fig S4.2).

Collectively, the results of the mass fractionation and delipidation suggests the probiotic-induced compound stimulating lymphocyte ATP levels is a lipoprotein under 30 kDa. Alternatively, the compound could associate with lipids in serum and is therefore extracted, or inactivated by the solvent treatment. Finally, the possibility of a lipid component and <30 kDa protein component working in conjunction cannot be ruled out.

Further study will be required to determine which scenario is correct.

There are several cytokines in this size range, including transforming growth factor-β

(TGF-β), interleukin-10 (IL-10), IL-2, and IL-7 (Gallus gallus proteome, http://www.uniprot.org/proteomes/UP000000539). There are also numerous lipids known to have immune-specific effects (33–35). Some prostaglandins, for example, are known to stimulate the intracellular signaling molecule cyclic AMP and affect gene and cell surface expression of a variety of immune genes (33). While there are numerous serum molecules

127

that could be involved in the lymphocyte response to probiotics, few have been evaluated for their ability to stimulate lymphocyte energy metabolism. However, there is evidence that some cytokines can regulate lymphocytes through energy metabolism; stimulation of CD28 and IL-7R are both known to increase glucose uptake and glycolytic activity in T cells, increasing survival and proliferation (36, 37).

Though there are several host-derived serum components that could account for metabolic response to probiotic treatment, the limited information on the active serum components available at this time does not exclude the possibility of microbe-derived serum metabolites. In a study of conventional and germ-free mice, it was shown that 145 of roughly 4000 serum metabolites were unique to conventionally colonized animals, with many more shared metabolites affected quantitatively (38). This indicates that normal microbiomes can play a significant role in serum composition.

Stimulation of CU205 lymphocytes with serum from PB-treated animals alters gene expression

To continue to characterize the probiotic-induced changes to immune cells and understand how the PB-treated serum affects lymphocytes, we attempted to characterize the metabolic and transcriptomic response of the cells to the PB-treated serum. Initial experiments assayed for cellular changes related to energy metabolism, specifically an increase in uptake of glucose and an increase in the numbers of mitochondria per cell. These demonstrated no consistently significant change in either parameter (data not shown). This

128

was surprising as an increase in ATP/cell has served as our biomarker for the probiotic- induced change in CU205 cells.

To better understand the global effect the serum from PB-treated animals might have on lymphocytes, the CU205 chicken T-cells were stimulated with serum from PB-treated and

CON-treated animals for 2 and 4 days of incubation, using whole and under 30 kDa fractionated serum. Three replicate samples from each treatment at two time points were sequenced, resulting in an average of 18,729,286 sequences/sample following application of a quality score cutoff of ≥25. Mean read length was 122 bp. Differential expression analysis revealed that cells were more responsive to treatment at 2 days (231 genes changed with whole serum; Table S4.1) than at 4 days (84 changes with whole serum; Fig 4.5), and 2 days of incubation with the lower MW PB-treatment fraction stimulated far more gene expression differences than 2 days with whole serum (1120 vs 231; Fig 4.5A and B). This relationship was not apparent at day 4; samples from PB-treatment < 30 kDa treatment showed only 30 differentially expressed genes compared with CON-treatment < 30 kDa, while a comparison between whole serum groups showed 84 differences (Fig 4.5C and D). Analysis of the differentially expressed genes at day two showed that the whole and fractionated serum gene sets had 151 genes in common (65% of all significant genes in the whole serum comparison).

Of these shared significant genes, 97% (146) changed in the same direction between whole and fractionated serum comparisons.

Interestingly, expression of annotated genes related to oxidative phosphorylation and glycolysis showed no pattern of change. None of the evaluated genes were different at day 2

129

or 4 in the whole serum samples. An increase in expression of 3 genes (COX17, VPS9D1,

NDUFB2) out of 75 involved in oxidative phosphorylation was noted at day 2 in samples grown in fractionated serum; however, the changes were insufficient to predict a change in oxidative phosphorylation activity (data not shown). While both of these ATP-generating pathways can be regulated transcriptionally, both are subject to a complex array of feedback mechanisms and also experience post transcriptional and post translational regulation via ubiquitination, phosphorylation, binding of effector proteins, and substrate availability (39,

40).

Of the 231 genes differentially regulated following 2 days of incubation with whole

CON- and PB-treated serum, many are directly related to immune function (Table S4.1).

Examination of those with the greatest fold-change (Fig 4.6A) include a collection of 13 genes that representing the cytoskeleton, transferase enzymes, cell proliferation regulators, and transcription factors regulating cell differentiation. However, most (15 out of 28, Fig

4.6A) of these highly differentially expressed genes lack functional annotation, and cannot be evaluated for their role in probiotic-stimulated lymphocyte activity.

To contextualize these changes and identify patterns of gene change for further evaluation, the gene expression data from 2 days post whole serum treatment was analyzed using the Ingenuity Pathway Analysis (IPA) platform. Based on the pattern of significant changes in gene expression (Table S4.1), IPA predicted decreases in activity of interferon-γ

(IFN-γ) and transcription factor special AT-rich sequence binding protein-1 (SATB1), and increased activity of transcription factor STAT3, IL-1β, and TGF-β (Fig 4.6B).

130

To assess these predictions, we first examined how well the RNA transcription data supported these conclusions. SATB1, STAT3, and IL-1β were not affected by treatment with either whole or fractionated serum. IFN-γ expression was numerically but not significantly decreased (29% lower) by treatment with whole serum; however, it was reduced by 60% (q <

0.05) in cells stimulated with <30 kDa PB-treated serum fraction compared with <30 kDa

CON-treated serum. TGF-β expression (specifically TGFB1 and TGFB3) is not affected by treatment, however TGF-β receptor 2 (TGFBR2) expression was increased by 30% (q =

0.02) following incubation with fractionated PB-treated serum.

The increase in TGFBR2 and the predicted increase in TGF-β mediated signaling by

IPA analysis suggested this signaling pathway maybe involved. TGF-βR2 can increase T- cell expansion and energy metabolism (36, 41, 42). TGF-β is a 25 kDa secreted protein that is involved in many immune cell processes, including cell growth, proliferation, differentiation, and regulation of inflammatory responses (43). TGF-β is a strong regulator of

T-cells, but the large number of costimulatory molecules and signaling pathways involved in its interactions with T-cells mean that its impact on T-cells is highly dependent on other factors (44). When accompanied by co-stimulation of CD28, TGF-β1 supports the survival and expansion of a variety of T-cell subsets, and may be involved the gene expression changes supporting those functions (41, 42). However, preliminary experiments to demonstrate a difference in levels of active TGF-β were inconclusive (data not shown).

Additional network-based analysis of the day 2 whole serum samples revealed 3 major putative functional changes related to immune cell survival and development were

131

upregulated following stimulation with PB-treated serum (Fig 4.7). Gene expression changes support increased numbers of thymocytes, and development and proliferation of T-helper cells. Observed gene changes that support these predicted upregulated functions include increased ICOS, IL-17R, transcription factor RORC, and T-cell co-receptor CD3E, and decreased expression of IRF 1 and 8 (Fig 4.7). These results indicate the probiotic may stimulate differences in differentiation and proliferation of specific T-cell subsets that could affect host immune function in measurable ways, such as promoting humoral responses or cell-mediated responses (7, 10).

CONCLUSIONS

Studies of PrimaLac and similar probiotic products have shown promising effects on avian GI physiology, growth, and feed efficiency, but their role in immune function is still poorly understood. In this study, we demonstrated that probiotics stimulate the upregulation of a serum signal that results in higher levels of ATP in lymphocytes in vitro. Further, this serum component is less than 30 kDa in weight, and appears to hydrophobic or lipid in nature. The gene expression profiles of these lymphocytes reveal many changes following stimulation with probiotic serum, but the incomplete annotation of the chicken genome means 17 of the top 30 differentially expressed genes cannot be adequately evaluated.

Analysis of annotated genes suggests this factor works to increase lymphocyte survival, expansion and differentiation. The results of this study suggest probiotic supplementation upregulates a signal in circulation that stimulates changes in lymphocyte function. The

132

increase in intracellular ATP in these cells is likely in support of these activities. Further analysis of transcriptional changes and associated physiological response to probiotic stimulation will improve our ability to evaluate the probiotic-host relationship in vivo, and thus, promote systemic and gastrointestinal health.

133

REFERENCES

1. Roos NM de, Katan MB. 2000. Effects of probiotic bacteria on diarrhea, lipid metabolism, and carcinogenesis: a review of papers published between 1988 and 1998. Am J Clin Nutr 71:405–411.

2. Mohan B, Kadirvel R, Natarajan A, Bhaskaran M. 1996. Effect of probiotic supplementation on growth, nitrogen utilisation and serum cholesterol in broilers. Br Poult Sci 37:395–401.

3. van Baarlen P, Wells JM, Kleerebezem M. 2013. Regulation of intestinal homeostasis and immunity with probiotic lactobacilli. Trends Immunol 34:208–215.

4. Jeon SG, Kayama H, Ueda Y, Takahashi T, Asahara T, Tsuji H, Tsuji NM, Kiyono H, Ma JS, Kusu T, Okumura R, Hara H, Yoshida H, Yamamoto M, Nomoto K, Takeda K. 2012. Probiotic Bifidobacterium breve Induces IL-10-Producing Tr1 Cells in the Colon. PLoS Pathog 8:e1002714.

6. Atarashi K, Tanoue T, Oshima K, Suda W, Nagano Y, Nishikawa H, Fukuda S, Saito T, Narushima S, Hase K, Kim S, Fritz JV, Wilmes P, Ueha S, Matsushima K, Ohno H, Olle B, Sakaguchi S, Taniguchi T, Morita H, Hattori M, Honda K. 2013. Treg induction by a rationally selected mixture of Clostridia strains from the human microbiota. Nature 500:232–236.

7. Dalloul RA, Lillehoj HS, Shellem TA, Doerr JA. 2003. Enhanced mucosal immunity against Eimeria acervulina in broilers fed a Lactobacillus-based probiotic. Poult Sci 82:62–66.

8. Haghighi HR, Gong J, Gyles CL, Hayes MA, Sanei B, Parvizi P, Gisavi H, Chambers JR, Sharif S. 2005. Modulation of Antibody-Mediated Immune Response by Probiotics in Chickens. Clin Diagn Lab Immunol 12:1387–1392.

9. Round JL, Mazmanian SK. 2010. Inducible Foxp3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proc Natl Acad Sci 107:12204– 12209.

10. Haghighi HR, Gong J, Gyles CL, Hayes MA, Zhou H, Sanei B, Chambers JR, Sharif S. 2006. Probiotics Stimulate Production of Natural Antibodies in Chickens. Clin Vaccine Immunol 13:975–980.

134

11. Hand TW, Santos LMD, Bouladoux N, Molloy MJ, Pagán AJ, Pepper M, Maynard CL, Elson CO, Belkaid Y. 2012. Acute Gastrointestinal Infection Induces Long-Lived Microbiota-Specific T Cell Responses. Science 337:1553–1556.

12. Parry SH, Allen WD, Porter P. 1977. Intestinal immune response to E. coli antigens in the germ-free chicken. Immunology 32:731.

13. Ballou AL, Ali RA, Mendoza MA, Ellis JC, Hassan HM, Croom WJ, Koci MD. 2016. Development of the Chick Microbiome: How Early Exposure Influences Future Microbial Diversity. Vet Infect Dis 2.

14. Chichlowski M, Croom WJ, Edens FW, McBride BW, Qiu R, Chiang CC, Daniel LR, Havenstein GB, Koci MD. 2007. Microarchitecture and spatial relationship between bacteria and ileal, cecal, and colonic epithelium in chicks fed a direct-fed microbial, PrimaLac, and salinomycin. Poult Sci 86:1121–1132.

15. Qiu R, Croom J, Ali RA, Ballou AL, Smith CD, Ashwell CM, Hassan HM, Chiang C- C, Koci MD. 2012. Direct fed microbial supplementation repartitions host energy to the immune system. J Anim Sci 90:2639–2651.

16. Berland R, Wortis HH. 2002. Origins and functions of b-1 cells with notes on the role of cd5. Annu Rev Immunol 20:253–300.

17. Chichlowski M, Croom J, McBride BW, Daniel L, Davis G, Koci MD. 2007. Direct- Fed Microbial PrimaLac and Salinomycin Modulate Whole-Body and Intestinal Oxygen Consumption and Intestinal Mucosal Cytokine Production in the Broiler Chick. Poult Sci 86:1100–1106.

18. Qureshi MA, Yu M, Saif YM. 2000. A Novel “Small Round Virus” Inducing Poult Enteritis and Mortality Syndrome and Associated Immune Alterations. Avian Dis 44:275–283.

19. Cham BE, Knowles BR. 1976. A solvent system for delipidation of plasma or serum without protein precipitation. J Lipid Res 17:176–181.

20. Schat KA, Pratt WD, Morgan R, Weinstock D, Calnek BW. 1992. Stable Transfection of Reticuloendotheliosis Virus-Transformed Lymphoblastoid Cell Lines. Avian Dis 36:432–439.

21. Okazaki W, Witter RL, Romero C, Nazerian K, Sharma JM, Fadly A, Ewert D. 1980. Induction of lymphoid leukosis transplant able tumours and the establishment of lymphoblastoid cell lines. Avian Pathol J WVPA 9:311–329.

135

22. Kawaguchi T, Nomura K, Hirayama Y, Kitagawa T. 1987. Establishment and Characterization of a Chicken Hepatocellular Carcinoma Cell Line, LMH. Cancer Res 47:4460–4464.

23. Qureshi MA, Miller L, Lillehoj HS, Ficken MD. 1990. Establishment and characterization of a chicken mononuclear cell line. Vet Immunol Immunopathol 26:237–250.

24. Andrews S. 2010. FastQC: A quality control tool for high throughput sequence data. Ref Source.

25. Ewels P, Magnusson M, Lundin S, Käller M. 2016. MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32:3047–3048.

26. Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR, Pimentel H, Salzberg SL, Rinn JL, Pachter L. 2012. Differential gene and transcript expression analysis of RNA- seq experiments with TopHat and Cufflinks. Nat Protoc 7:562–578.

27. Kim D, Langmead B, Salzberg SL. 2015. HISAT: a fast spliced aligner with low memory requirements. Nat Methods 12:357–360.

28. Goff LA, Trapnell C, Kelley D. 2012. CummeRbund: visualization and exploration of Cufflinks high-throughput sequencing data. R Package Version 2.

29. Zhou N, Lee WR, Abasht B. 2015. Messenger RNA sequencing and pathway analysis provide novel insights into the biological basis of chickens’ feed efficiency. BMC Genomics 16:195.

30. Krämer A, Green J, Pollard J, Tugendreich S. 2013. Causal analysis approaches in Ingenuity Pathway Analysis. Bioinformatics btt703.

31. Pieper R, Gatlin CL, Makusky AJ, Russo PS, Schatz CR, Miller SS, Su Q, McGrath AM, Estock MA, Parmar PP, Zhao M, Huang S-T, Zhou J, Wang F, Esquer-Blasco R, Anderson NL, Taylor J, Steiner S. 2003. The human serum proteome: display of nearly 3700 chromatographically separated protein spots on two-dimensional electrophoresis gels and identification of 325 distinct proteins. Proteomics 3:1345–1364.

32. Psychogios N, Hau DD, Peng J, Guo AC, Mandal R, Bouatra S, Sinelnikov I, Krishnamurthy R, Eisner R, Gautam B, Young N, Xia J, Knox C, Dong E, Huang P, Hollander Z, Pedersen TL, Smith SR, Bamforth F, Greiner R, McManus B, Newman JW, Goodfriend T, Wishart DS. 2011. The Human Serum Metabolome. PLOS ONE 6:e16957.

136

33. Cabral GA. 2005. Lipids as bioeffectors in the immune system. Life Sci 77:1699–1710.

34. Claesson H-E, Odlander B, Jakobsson P-J. 1992. Leukotriene B4 in the immune system. Int J Immunopharmacol 14:441–449.

35. Nataraj C, Thomas DW, Tilley SL, Nguyen M, Mannon R, Koller BH, Coffman TM. 2001. Receptors for prostaglandin E2 that regulate cellular immune responses in the mouse. J Clin Invest 108:1229–1235.

36. Frauwirth KA, Riley JL, Harris MH, Parry RV, Rathmell JC, Plas DR, Elstrom RL, June CH, Thompson CB. 2002. The CD28 Signaling Pathway Regulates Glucose Metabolism. Immunity 16:769–777.

37. Wofford JA, Wieman HL, Jacobs SR, Zhao Y, Rathmell JC. 2008. IL-7 promotes Glut1 trafficking and glucose uptake via STAT5-mediated activation of Akt to support T-cell survival. Blood 111:2101–2111.

38. Wikoff WR, Anfora AT, Liu J, Schultz PG, Lesley SA, Peters EC, Siuzdak G. 2009. Metabolomics analysis reveals large effects of gut microflora on mammalian blood metabolites. Proc Natl Acad Sci 106:3698–3703.

39. Berg JM, Tymoczko JL, Stryer L. 2002. The Glycolytic Pathway Is Tightly Controlled.

40. Hüttemann M, Lee I, Pecinova A, Pecina P, Przyklenk K, Doan JW. 2008. Regulation of oxidative phosphorylation, the mitochondrial membrane potential, and their role in human disease. J Bioenerg Biomembr 40:445.

41. Gunnlaugsdottir B, Maggadottir SM, Ludviksson BR. 2005. Anti-CD28-induced co- stimulation and TCR avidity regulates the differential effect of TGF-β1 on CD4+ and CD8+ naïve human T-cells. Int Immunol 17:35–44.

42. Sung JL, Lin JT, Gorham JD. 2003. CD28 co-stimulation regulates the effect of transforming growth factor-β1 on the proliferation of naı̈ ve CD4+ T cells. Int Immunopharmacol 3:233–245.

43. Massague J. 1990. The Transforming Growth Factor-beta Family. Annu Rev Cell Biol 6:597–641.

44. Li MO, Wan YY, Sanjabi S, Robertson A-KL, Flavell RA. 2006. TRANSFORMING GROWTH FACTOR-β REGULATION OF IMMUNE RESPONSES. Annu Rev Immunol 24:99–146.

137

Table S4.1. Differentially expressed genes between day 2 CON and PB whole serum

CON log2 fold Gene FPKM PL FPKM change q Value Gene Name ENSGALG00000019934 2.02 0.00 0.012 ENSGALG00000029546 0.00 0.89 0.005 ENSGALG00000034543 0.00 1.73 0.005 ENSGALG00000042697 0.00 36.34 0.005 ENSGALG00000002064 12.18 0.75 -4.02 0.005 ENSGALG00000036172 0.72 4.73 2.71 0.005 TFPI2 4.35 17.23 1.99 0.005 tissue factor pathway inhibitor 2 TMEM18 3.13 11.44 1.87 0.005 transmembrane protein 18 DYNLRB1 11.94 37.22 1.64 0.005 eukaryotic translation elongation factor 1 delta KRT18 4.32 1.43 -1.59 0.005 keratin 18, type I ENSGALG00000016677 30.69 90.99 1.57 0.005 ENSGALG00000030199 7.06 20.54 1.54 0.005 ENSGALG00000036313 7.55 2.83 -1.42 0.026 B4GALNT3 2.49 6.56 1.40 0.005 beta-1,4-N-acetyl-galactosaminyl transferase 3 ENSGALG00000038255 5.03 1.96 -1.36 0.009 PTCHD3 0.93 2.11 1.18 0.038 patched domain containing 3 ENSGALG00000037595 4.34 9.77 1.17 0.005 RPS6KA2 2.56 5.73 1.16 0.005 ribosomal protein S6 kinase, 90kDa, polypeptide 2 ENSGALG00000014750 2.00 4.45 1.16 0.014 COL22A1 0.52 1.14 1.12 0.024 collagen, type XXIII, alpha 1 TSC22D1 3.13 6.73 1.11 0.005 TSC22 domain family, member 1 SULT1B 3.17 6.69 1.08 0.040 sulfotransferase family, cytosolic, 1B

138

Table S4.1 Continued SOX5 1.92 4.04 1.07 0.005 SRY (sex determining region Y)-box 5 ENSGALG00000022802 2.81 1.34 -1.06 0.005 DUSP1 4.19 8.63 1.04 0.012 dynein, light chain, roadblock-type 1 ENSGALG00000041883 8.52 17.32 1.02 0.009 CRIP1 65.53 132.62 1.02 0.005 cytochrome P450, family 7, subfamily B, polypeptide 1 ENSGALG00000039214 3.99 8.03 1.01 0.005 ENSGALG00000042035 0.92 1.80 0.97 0.009 ENSGALG00000026383 13.29 25.63 0.95 0.005 MREG 2.36 4.55 0.95 0.032 melanoregulin PLEKHA8 8.61 4.48 -0.94 0.005 pleckstrin homology domain containing, family A member 8 CYTIP 10.17 19.44 0.93 0.005 DAB2 interacting protein TET3 1.30 2.47 0.92 0.005 tet methylcytosine dioxygenase 3 MST1 7.65 14.50 0.92 0.005 macrophage stimulating 1 ENSGALG00000043390 1.29 0.69 -0.91 0.043 GEMIN6 18.42 9.87 -0.90 0.005 gem nuclear organelle associated protein 6 BHLHE40 14.35 26.52 0.89 0.005 basic helix-loop-helix family member e40 TST 9.86 18.13 0.88 0.005 thiosulfate sulfurtransferase DMD 0.51 0.94 0.88 0.014 DnaJ homolog, subfamily A, member 1 (DNAJA1) COL23A1 7.91 14.50 0.87 0.005 coactosin-like F-actin binding protein 1 COTL1 3.11 5.69 0.87 0.012 carboxypeptidase D GRIN2C 1.30 2.36 0.86 0.014 glutamate receptor, ionotropic, N-methyl D-aspartate 2C PKHD1 0.66 1.18 0.84 0.005 polycystic kidney and hepatic disease 1 (autosomal recessive) HTRA1 13.08 7.35 -0.83 0.005 HtrA serine peptidase 1 SNRPG 333.58 187.93 -0.83 0.005 small nuclear ribonucleoprotein polypeptide G

139

Table S4.1 Continued IL7R 10.25 18.17 0.83 0.005 interleukin 7 receptor FLI1 7.97 14.09 0.82 0.005 Fli-1 proto-oncogene, ETS transcription factor CD38 12.19 21.49 0.82 0.005 CD3d molecule, delta (CD3-TCR complex) OASL 72.23 40.98 -0.82 0.005 2'-5'-oligoadenylate synthetase-like SNAP91 5.99 10.55 0.82 0.005 synaptosomal-associated protein, 91kDa ARL15 10.08 5.78 -0.80 0.034 ADP ribosylation factor like GTPase 15 BFSP1 2.44 4.24 0.80 0.022 beaded filament structural protein 1 RGS1 50.14 86.90 0.79 0.005 regulator of G-protein signaling 1 GSN 3.46 6.00 0.79 0.005 gelsolin TMEM144 5.17 8.90 0.78 0.028 transmembrane protein 144 ENSGALG00000032428 27.60 16.11 -0.78 0.009 IFI6 44.21 25.99 -0.77 0.005 interferon, alpha-inducible protein 6 LAPTM5 18.56 31.34 0.76 0.005 lysosomal protein transmembrane 5 LGALS3 16.67 28.07 0.75 0.009 lectin, galactoside-binding, soluble, 3 TSPAN13 20.61 12.37 -0.74 0.005 tetraspanin 13 ICOS 22.50 37.44 0.73 0.005 inducible T-cell co-stimulator ZFPM1 2.98 1.79 -0.73 0.005 zinc finger protein, FOG family member 1 TMEM41B 27.46 16.54 -0.73 0.029 transmembrane protein 41B PAX9 5.35 3.22 -0.73 0.048 paired box 9 ENSGALG00000006644 9.02 5.46 -0.72 0.005 PTK2 11.64 7.08 -0.72 0.005 protein tyrosine kinase 2 SLC24A3 6.08 3.71 -0.71 0.043 solute carrier family 24 (sodium/potassium/calcium exchanger), member 3 IRF7 61.68 37.74 -0.71 0.005 interferon regulatory factor 7 ITM2C 18.79 30.70 0.71 0.005 integral membrane protein 2C

140

Table S4.1 Continued SLCO4C1 13.20 21.50 0.70 0.005 solute carrier organic anion transporter family, member 4C1 ENSGALG00000042008 95.13 154.81 0.70 0.005 RNF213 6.98 4.31 -0.70 0.005 ring finger protein 213 ENSGALG00000014754 26.68 42.98 0.69 0.005 RNF157 20.29 12.65 -0.68 0.005 ring finger protein 157 JMJD6 12.95 8.10 -0.68 0.005 jumonji domain containing 6 GC 58.93 37.05 -0.67 0.005 group-specific component (vitamin D binding protein) DENND2A 9.86 6.21 -0.67 0.005 dehydrogenase/reductase (SDR family) X-linked (DHRSX) ENSGALG00000032281 7.43 4.68 -0.67 0.024 IL17A 124.82 197.97 0.67 0.005 interleukin 17A SVIP 22.83 36.04 0.66 0.005 small VCP/p97-interacting protein TUSC3 7.24 11.40 0.66 0.029 tumor suppressor candidate 3 MOV10 18.59 11.81 -0.66 0.005 Mov10, Moloney leukemia virus 10, homolog ENSGALG00000016477 4.60 7.23 0.65 0.014 VAV3 1.72 2.70 0.65 0.040 vav 3 guanine nucleotide exchange factor DDHD1 8.18 5.24 -0.64 0.012 2,4-dienoyl CoA reductase 1, mitochondrial ENSGALG00000006034 7.29 11.27 0.63 0.005 MRPS14 54.93 35.73 -0.62 0.005 mitochondrial ribosomal protein S14 ENSGALG00000037932 12.42 19.05 0.62 0.017 C1S 7.91 5.17 -0.62 0.012 coiled-coil domain containing 28A ENSGALG00000029935 6.79 10.35 0.61 0.024 CCR8 8.72 13.24 0.60 0.009 chaperonin containing TCP1 subunit 4 ENSGALG00000033573 23.80 36.13 0.60 0.012 PHTF2 8.84 5.86 -0.59 0.005 putative homeodomain transcription factor 2

141

Table S4.1 Continued ENSGALG00000033266 21.09 13.98 -0.59 0.009 GRAP 38.01 57.05 0.59 0.005 GRB2-related adaptor protein NUDT14 19.32 29.01 0.59 0.029 nudix DHRSX 33.87 50.80 0.58 0.005 Dystrophin MYO1E 5.47 8.20 0.58 0.017 myosin IE ENSGALG00000041164 251.31 375.72 0.58 0.005 ADAP1 30.16 20.18 -0.58 0.028 ArfGAP with dual PH domains 1 CYP7B1 9.48 6.36 -0.58 0.046 cytohesin 1 interacting protein WWC3 14.77 9.92 -0.57 0.005 WWC family member 3 C17orf62 12.54 18.64 0.57 0.009 complement component 1, s subcomponent ENSGALG00000042001 33.04 22.25 -0.57 0.005 GINS2 21.42 14.45 -0.57 0.017 GINS complex subunit 2 ENSGALG00000021399 3.14 2.12 -0.57 0.043 CCDC28A 11.28 7.70 -0.55 0.049 coiled-coil domain containing 50 SYNE3 5.14 7.53 0.55 0.034 spectrin repeat containing, nuclear envelope family member 3 PLEKHA2 14.26 20.82 0.55 0.005 pleckstrin homology domain containing A2 KHDRBS3 20.15 13.80 -0.55 0.005 KH domain containing, RNA binding, signal transduction associated 3 PXK 25.06 17.18 -0.54 0.005 PX domain containing serine/threonine kinase like AASDHPPT 11.51 16.79 0.54 0.049 aminoadipate-semialdehyde dehydrogenase-phosphopantetheinyl transferase AZIN2 11.22 16.36 0.54 0.012 arginine decarboxylase HSP90AA1 1395 958.91 -0.54 0.005 heat shock protein 90kDa alpha (cytosolic), class A member 1 DAB2IP 6.58 9.57 0.54 0.012 DDHD domain containing 1 PAFAH2 8.51 12.35 0.54 0.043 platelet-activating factor acetylhydrolase 2 TIMM10 65.65 95.12 0.54 0.045 translocase of inner mitochondrial membrane 10 homolog

142

Table S4.1 Continued RPL5 1198.8 1732.81 0.53 0.005 ribosomal protein L5 HSPA5 304.53 210.69 -0.53 0.005 heat shock 70kDa protein 5 (glucose-regulated protein, 78kDa) ENSGALG00000029349 19.04 27.49 0.53 0.005 MYOF 9.43 13.57 0.53 0.005 myoferlin FLNB 77.98 112.18 0.52 0.005 filamin B, beta FAM173A 35.10 50.45 0.52 0.005 family with sequence similarity 173 member A DNAJA1 103.44 72.27 -0.52 0.005 dedicator of cytokinesis 10 KLHDC4 196.10 137.06 -0.52 0.005 kelch domain containing 4 ADAMTSL2 8.74 6.13 -0.51 0.026 ADAMTS-like 2 CD3D 214.39 305.79 0.51 0.005 CD3e molecule MYL12A 9.68 13.81 0.51 0.005 myosin, light chain 12A, regulatory, non-sarcomeric+F145:F158 GNA11 5.71 4.01 -0.51 0.045 guanine nucleotide binding protein (G protein), alpha 11 (Gq class) RORC 69.74 99.31 0.51 0.005 RAR-related orphan receptor C PREP2 30.01 21.07 -0.51 0.014 protein phosphatase 1 regulatory subunit 3E ENSGALG00000035870 35.64 25.11 -0.51 0.005 ARHGDIB 54.91 77.79 0.50 0.005 Rho GDP dissociation inhibitor MGMT 11.30 15.97 0.50 0.029 O-6-methylguanine-DNA methyltransferase NCMAP 58.95 83.03 0.49 0.005 noncompact myelin associated protein ENSGALG00000040401 47.11 66.34 0.49 0.009 BOK 43.65 61.42 0.49 0.005 BCL2-related ovarian killer CPD 6.34 4.53 -0.49 0.042 cysteine rich protein 1 SETD6 13.39 9.57 -0.48 0.043 SET domain containing 6 CCDC80 24.74 34.51 0.48 0.005 chemokine (C-C motif) receptor 8 (CCR8) RET 12.65 17.61 0.48 0.005 ret proto-oncogene

143

Table S4.1 Continued UBE2E2 80.00 57.46 -0.48 0.022 ubiquitin-conjugating enzyme E2E 2 ENSGALG00000042326 14.78 20.56 0.48 0.026 TEX264 10.08 7.27 -0.47 0.049 testis expressed 264 SCLY 17.92 24.84 0.47 0.022 selenocysteine lyase ENSGALG00000032377 126.71 175.45 0.47 0.024 ENSGALG00000000720 30.98 22.41 -0.47 0.005 GLIPR2 11.49 15.81 0.46 0.049 GLI pathogenesis-related 2 NABP1 38.58 28.05 -0.46 0.009 nucleic acid binding protein 1 RPL22L1 3511.2 4825.70 0.46 0.005 ribosomal protein L22-like 1 OGG1 23.11 31.72 0.46 0.028 8-oxoguanine DNA glycosylase C12orf75 79.65 109.33 0.46 0.005 chromosome 12 open reading frame 75 PTGFRN 26.31 19.19 -0.46 0.017 prostaglandin F2 receptor inhibitor EYA1 39.16 53.50 0.45 0.005 eyes absent homolog 1 ENSGALG00000010203 17.08 23.33 0.45 0.012 QSER1 7.02 5.15 -0.45 0.014 glutamine and serine rich 1 FOLH1 11.63 15.84 0.45 0.020 folate hydrolase (prostate-specific membrane antigen) 1 KIAA1324L 55.48 40.78 -0.44 0.005 KIAA1324-like SLC17A5 12.24 9.02 -0.44 0.020 solute carrier family 17 (acidic sugar transporter), member 5 IRF1 120.29 88.71 -0.44 0.005 interferon regulatory factor 1 IL2RA 74.91 101.58 0.44 0.029 interleukin 2 receptor, alpha SPTB 7.45 5.50 -0.44 0.028 spectrin, beta, erythrocytic LARP1B 13.49 9.97 -0.44 0.028 La ribonucleoprotein domain family, member 1B GSTM2 63.01 46.59 -0.44 0.043 glutathione S-transferase mu 2 ENSGALG00000026152 126.76 93.84 -0.43 0.005

144

Table S4.1 Continued MAP4K4 14.59 19.69 0.43 0.020 mitogen-activated protein kinase kinase kinase kinase 4 TMEM64 71.02 95.77 0.43 0.009 transmembrane protein 64 ETV6 17.77 23.95 0.43 0.005 ets variant 6 PPP1R3E 42.45 31.54 -0.43 0.009 protein phosphatase 1, regulatory subunit 3E RAC2 147.76 198.82 0.43 0.005 rho family, small GTP binding protein Rac2 ENSGALG00000039012 14.14 10.52 -0.43 0.017 MTR 29.65 39.85 0.43 0.014 5-methyltetrahydrofolate-homocysteine methyltransferase DECR1 333.78 448.32 0.43 0.005 DENN/MADD domain containing 2A KPTN 17.39 23.35 0.43 0.026 kaptin (actin binding protein) FGGY 41.41 55.58 0.42 0.046 FGGY carbohydrate kinase domain containing GRAMD1C 30.38 40.76 0.42 0.009 GRAM domain containing 1C HSPH1 77.03 57.42 -0.42 0.005 heat shock 105kDa/110kDa protein 1 PLEKHB1 27.59 20.63 -0.42 0.022 pleckstrin homology domain containing, family B (evectins) member 1 RGS9 128.15 170.63 0.41 0.005 regulator of G-protein signaling 9 FAM13B 36.91 27.72 -0.41 0.032 family with sequence similarity 13, member B MYOZ2 45.64 60.73 0.41 0.022 myozenin 2 TRPC6 30.81 23.19 -0.41 0.017 transient receptor potential cation channel, subfamily C, member 6 WIPI1 201.35 267.20 0.41 0.022 WD repeat domain, phosphoinositide interacting 1 TBC1D19 89.43 118.64 0.41 0.005 TBC1 domain family member 19 CCDC50 36.28 47.95 0.40 0.014 coiled-coil domain containing 80 ENSGALG00000037342 23.53 17.84 -0.40 0.040 ATP11A 13.90 10.54 -0.40 0.034 ATPase, class VI, type 11A SAMSN1 35.78 47.11 0.40 0.012 SAM domain, SH3 domain and nuclear localization signals 1 PPFIA1 36.05 27.40 -0.40 0.020 PTPRF interacting protein, alpha 1

145

Table S4.1 Continued BCO2 117.88 154.17 0.39 0.014 beta-carotene oxygenase 2 ENSGALG00000019325 98.42 75.30 -0.39 0.014 ENSGALG00000042590 80.22 61.42 -0.39 0.005 IL18R1 36.97 28.38 -0.38 0.036 interleukin 18 receptor 1 SLC41A3 58.36 75.97 0.38 0.026 solute carrier family 41 member 3 MFNG 89.61 68.88 -0.38 0.038 MFNG O-fucosylpeptide 3-beta-N-acetylglucosaminyltransferase ENSGALG00000029056 31.22 24.01 -0.38 0.043 TRUB1 82.50 63.53 -0.38 0.043 TruB pseudouridine (psi) synthase family member 1 GTF2H3 24.11 31.30 0.38 0.046 general transcription factor IIH subunit 3 IRF8 122.64 94.59 -0.37 0.024 interferon regulatory factor 8 TFRC 57.20 44.20 -0.37 0.043 guanine nucleotide binding protein (G protein) CHORDC1 47.22 36.53 -0.37 0.042 collagen, type XXII, alpha 1 BRCA2 36.88 47.64 0.37 0.012 breast cancer 2 ARHGAP15 91.74 118.43 0.37 0.043 Rho GTPase activating protein 15 ENSGALG00000008704 10.46 13.49 0.37 0.014 HSPA8 1773.6 1378.02 -0.36 0.020 heat shock 70kDa protein 8 HIRA 46.56 36.24 -0.36 0.029 histone cell cycle regulator GNPTAB 30.60 23.84 -0.36 0.024 N-acetylglucosamine-1-phosphate transferase, alpha and beta subunits NFU1 111.63 143.16 0.36 0.045 NFU1 iron-sulfur cluster scaffold ENSGALG00000041161 30.73 23.97 -0.36 0.043 ALCAM 49.31 38.51 -0.36 0.032 activated leukocyte cell adhesion molecule H2AFV 606.70 776.11 0.36 0.026 H2A histone family, member V CD3E 155.64 199.07 0.36 0.020 cysteine and histidine-rich domain (CHORD) containing 1 EEF1D 51.52 65.72 0.35 0.043 eukaryotic translation elongation factor 1 delta

146

Table S4.1 Continued HSPA4L 272.62 213.76 -0.35 0.026 heat shock 70kDa protein 4-like PI4K2B 45.63 35.84 -0.35 0.043 Phosphatidylinositol 4-kinase type 2-beta HSP90B1 779.65 612.29 -0.35 0.026 heat shock protein 90kDa beta (Grp94), member 1 FYCO1 45.78 58.10 0.34 0.038 FYVE and coiled-coil domain containing 1 ENSGALG00000006564 185.43 146.14 -0.34 0.045 HSPA9 348.35 274.98 -0.34 0.026 heat shock protein family A (Hsp70) member 9 CD109 80.22 101.47 0.34 0.024 CD38 molecule CCT4 696.23 551.34 -0.34 0.036 CD109 molecule JCHAIN 138.22 174.40 0.34 0.040 joining chain of multimeric IgA and IgM LDHA 387.87 488.60 0.33 0.022 lactate dehydrogenase A GALNT1 126.74 159.17 0.33 0.029 UDP-N-acetyl-α-D-galactosamine:polypeptide N-acetylgalactosaminyltransferase IL18 1650.1 2071.86 0.33 0.043 interleukin 18 SOX6 236.15 296.46 0.33 0.029 SRY (sex determining region Y)-box 6 DOCK10 125.44 156.12 0.32 0.049 dual specificity phosphatase 1 SNX9 214.97 173.22 -0.31 0.045 sorting nexin 9

147

Figure S4.1. Generation of complete serum samples from molecular weight fractions. Serum samples from CON and PB treated animals, and from SPF chicken serum were fractionated using mass-limited centrifugation filters (30, 10, 5 kDa). Samples were centrifuged until 80% of the original volume had passed through the filter. Reconstituted complete serum was generated by adding 1 volume of the high molecular weight (MW) fraction from CON- or PB-treated serum samples (upper fraction) to 4 volumes of the low MW fraction from fractionated SPF serum. Conversely, 1 volume of the high molecular weight (MW) fraction from SPF serum samples was added to 4 volumes of the low MW fraction from fractionated CON- and PB-treated serum.

148

PBMC

2 3.34 2.81

1 RLU/10,000 cells

0 P = 0.037 Control Probiotic

Figure 4.1. Peripheral blood mononuclear cells (PBMC) collected from chickens fed a probiotic-supplemented diet have more ATP/cell than PBMC from control animals PBMC were isolated from 28-day old chickens fed either a control diet or the control diet supplemented with probiotic. Cells from 5 animals per treatment group were counted, aliquoted 1x105 cells/well with 4 wells/animal. ATP was quantified using an ATP-dependent luciferase assay. Light production was recorded as relative light units (RLU) for each well. Replicate wells for each animal were averaged to determine RLU/cell. Individual animal averages were then used to calculate the treatment average RLU/cell (colored bars) and standard deviation (error bars). RLU is reported in the center of each bar as RLU/10,000 cells. P-values were generated using two-tailed T-tests.

149

Figure 4.2. Avian lymphocyte cell lines increase ATP following stimulation by serum from probiotic-fed animals. Serum collected from PB- and CON-treated animals was added to growth media (10% v/v) in place of the normal maintenance SPF chicken serum. Cells (CU205, RP9, LMH, and MQ-NCSU ) were seeded at 5,000 cells/well, in replicate wells (n=24 wells/treatment) of a 96-well plate. Following 4 days of incubations cells were assayed for intracellular ATP content (n=15 wells/treatment) and cell number/well (n=9 wells/treatment). ATP standard curve (0-250 nmol/well) was generated in serum-free growth media and a cell type specific standard curve was generated from serially diluted, hemocytometer-enumerated stock cells. Data are representative of at least 3 experiments. Colored bars represent the average ATP pmol/cell with value reported in the center of each bar. Error bars represent the standard deviation of replicate wells in each treatment. P-values were generated using two-tailed T-tests.

150

CU205 Lymphocytes RP9 Lymphocytes 5 1.5 4

1.0 3

1.4 4.44 2 3.98

0.5 0.978 ATP (pmol)/cell ATP (pmol)/cell 1

0.0 0 P = 9.68e-05 P = 0.00757 Control Probiotic Control Probiotic

LMH Hepatocytes MQ-NCSU Macrophages 5

3 4

3 2 3.41 4.47 2.82 2 3.87

1 ATP (pmol)/cell ATP (pmol)/cell 1

0 0 P = 0.0623 P = 0.00125 Control Probiotic Control Probiotic

151

Figure S4.2. Coomassie staining of SDS-PAGE-resolved mass fractionated and delipidated serum. Serum samples subjected to size fractionation (lanes 1. PB-treated serum >10 kDa fraction, 2. PB-treated serum <10 kDa fraction, 3. CON-treated serum >10 kDa fraction, 4. CON-treated serum <10 kDa fraction, 5. PB-treated serum <5 kDa fraction, 6. PB-treated serum >5 kDa fraction, 7. CON-treated serum <5 kDa fraction, 8. CON-treated serum >5 kDa fraction, 9. CON-treated serum >30 kDa fraction, 10. CON-treated serum <30 kDa fraction, 11. PB-treated serum >30 kDa fraction), BDIPE treatment (lanes 12. CON- treated serum and 13. PB-treated serum), or SPF control (lane 14) were loaded 20 μg protein/lane with the exception of lower molecular weight fractions (lanes 2, 4, 5, 7, and 10) where total protein concentration was too low, in which case maximum amount of sample (15 ul) was added to the well. Lane M = molecular weight ladder. Samples were resolved using an 18% polyacrylamide gel under reducing conditions.

152

Delipidation 0.8

0.6

0.4 0.751 0.622

0.2 ATP (pmol)/cell

0.0 P = 3.75e-16 Control Probiotic

Figure 4.3. Delipidation of serum from PB-treated animals results in lower ATP/cell relative to delipidated CON serum. Lipids were removed from serum from PB- and CON- treated animals using butanol and di-isopropyl ether (BDIPE) extraction. Delipidated serum was used to replace the normal SPF chicken serum component of CU205 cell media (10% v/v). The CU205 cells were seeded at 5,000 cells/well, in replicate wells (n=24 wells/treatment) of a 96-well plate. Following 4 days of incubation cells were assayed for intracellular ATP content (n=15 wells/treatment) and cell number/well (n=9 wells/treatment). ATP standard curve (0-250 nmol/well) was generated in serum-free growth media and a cell type specific standard curve was generated from serially diluted, hemocytometer-enumerated stock cells. Data are representative of at least 3 experiments. Colored bars represent the average ATP pmol/cell with value reported in the center of each bar. Error bars represent the standard deviation of replicate wells in each treatment. P-values were generated using two- tailed T-tests.

153

Under 30 kDa Over 30 kDa

1.5 1.2

1.0 0.8

1.47 1.19 1.17 1.12

0.5 0.4

ATP (pmol)/cell ATP (pmol)/cell

0.0 0.0 P = 2.89e-20 P = 0.27 Control Probiotic Control Probiotic

Figure 4.4. PB-treated animal serum components below 30 kDa stimulate ATP production in CU205 cells. CON-treated, PB-treated, and SPF maintenance serum were size fractionated centrifugation into >30 kDa and <30 kDa fractions. Serum components under and above 30 kDa from PB- and CON-treated animals were mixed with their reciprocal fraction from SPF serum to reconstitute a complete serum profile (Fig S1). This serum was then used to replace the normal SPF chicken serum component of CU205 cell media (10% v/v). CU205 cells were seeded at 5,000 cells/well, in replicate wells (n=24 wells/treatment) of a 96-well plate. Following 4 days of incubations cells were assayed for intracellular ATP content (n=15 wells/treatment) and cell number/well (n=9 wells/treatment). ATP standard curve (0-250 nmol/well) was generated in serum-free growth media and a cell type specific standard curve was generated from serially diluted, hemocytometer-enumerated stock cells. Data are representative of at least 3 experiments. Colored bars represent the average ATP pmol/cell with value reported in the center of each bar. Error bars represent the standard deviation of replicate wells in each treatment. P-values were generated using two- tailed T-tests.

154

Figure 4.5. Serum from PB-treated animals alters gene expression of CU205 relative to serum from CON-treated animals. CU205 cells were grown in T-25 flasks in media containing whole (A and C) or molecular-weight fractionated serum (B and D) for 2 (A and B) and 4 days (C and D). mRNA was isolated and sequenced from 3 replicate T-25s for each treatment and time point. Differentially expressed genes were identified between CON and PB-treated cells in each condition using CuffLinks. Significance was determined as an FDR- corrected q-value ≤ 0.05. The CummeRbund R package was used to generate volcano plots comparing gene expression between CON and PB samples at each time point, with whole and fractionated serum. Total number of significant differentially expressed genes are denoted on each plot.

155

Figure 4.6. Differentially expressed genes between cells incubated in serum from CON- or PB-treated animals. Transcriptomic analysis of CU205 cells after incubation with whole serum from CON- or PB-treated animals for 2 day identified 231 differentially expressed genes. (A) Genes and predicted gene found to be significantly different at FDR-corrected q ≤ 0.05 and with a minimum 2-fold change are depicted in the heatmap reported in fragments per kilobase of transcript per million mapped reads (FPKM) for each listed gene. (B) Network analysis of all altered genes was conducted using Ingenuity pathway analysis (IPA); listed genes were significantly different at FDR-corrected q ≤ 0.05. Predicted changes are shown in orange (increased) and blue (decreased), while measured changes are shown in red (increased) and green (decreased). Evaluation of potential regulators of measured gene changes reveal several predicted up- and down-regulated upstream regulators.

156

157

Figure 4.7. Predicted functional changes lymphocytes following stimulation with serum from Probiotic-treated animals. Network analysis of altered genes was conducted using Ingenuity pathway analysis (IPA). Listed genes were significantly different at FDR- corrected q ≤ 0.05. Predicted changes are shown in orange (increased) and blue (decreased), while measured changes are shown in red (increased) and green (decreased). Several survival-related functions were predicted to significantly increase (z-score > 1.5) following stimulation with probiotic.

158

CHAPTER 5: FURTHER STUDIES

159

The studies described in this dissertation were done with the ultimate goal of learning enough about the microbiome and its effects on host health that we can control this relationship to prevent and treat disorders that would normally reduce growth, health, or production. There are numerous examples of the microbiome exerting positive and negative effects on its host (1–6), but currently, we are very limited in our ability to do this intentionally. This research improved our understanding of the complexity of the microbiome, the factors that regulate its development and homeostasis, and how the host responds to changes in microbial composition.

We know the microbiome of growing chicks starts with very few species, but develops into a complex, Firmicutes-heavy community in the first weeks of life. However, the changes in composition seen during development or following treatment may be less important than functional changes that occur. There are indications that the ileum is the site of most probiotic-related taxonomic and functional changes. The ilea of probiotic-fed animals had reductions in bacteria associated with inflammation, specifically “Candidatus

Savagella,” as well as increased potential for production of butyrate. This highlights the ileum as an important region for communication between the host and microbiome. These evaluations of the avian microbiome were in support of understanding how probiotics communicate with the host through the microbiome, but we also continued our study of the host immune response. Probiotic treatment upregulates a serum component of currently unknown origin that increases ATP levels in lymphocytes. This serum factor may also be responsible for previous results suggesting increased humoral activity and higher energy

160

consumption in the thymus (7). The effect of the factor on T-cell gene expression indicates the probiotic stimulates cell survival and differentiation, and a number of candidate molecules were identified that could be involved in the lymphocyte response to probiotic treatment, including TGF-β, IL-7, IFN-γ, and IL-1β.

Overall, it appears as though probiotic-directed changes in the microbiome, likely in the ileum, stimulate the release of a signal in the serum that is capable of stimulating a number of transcriptional changes in lymphocytes. The source of this signal is currently unknown and could be of microbe or host origin (8, 9). The impact of the serum signal can be observed as an increase in ATP in both primary cells and cell lines exposed to serum. The increase in ATP may be a result of several functional changes suggested by transcriptional data.

Following this research, we are left with new directions to pursue, as well as several questions that were left unanswered despite our efforts. First and foremost, what is the serum factor responsible for stimulating lymphocytes? Though we narrowed down the list of the thousands of known serum proteins, lipids, and metabolites (below 30 kDa, with a lipid or hydrophobic component, capable of stimulating immune cells), we are still left with a large number of potential candidates of both human and microbial origin. Further, identification of the serum factor may provide insight into which cells will be activated and how.

This leads into a second, biologically significant question: what phenotypic changes result in lymphocytes following probiotic stimulation, and what benefit does this phenotype confer on the animal in vivo? Several probiotic products and other microbiome changes

161

affect immune function (3, 10–12), and this particular one is known to stimulate humoral immune activity (13). The fact that T-cells and perhaps macrophages appear to be sensitive to the serum factor suggests there are other in vivo effects that have yet to be observed.

Characterization of circulating leukocytes may reveal changes in the proportion of different immune cell subsets, and introducing an immune challenge to probiotic-treated animals may reveal other ways in which the probiotic affects the immune response in vivo. Expanding the types of immune cells evaluated in vitro may result in the discovery of other immune cells affected by the probiotic.

While the identity of the serum factor may also provide an answer as to its origin, it is likely that we will still be left uncertain as to which members of the microbiome are responsible for stimulating this effect. Several taxa respond to probiotic treatment

(Lachnospiraceae, Ruminococcaceae, SFB) in multiple locations. Defining the contribution of these and other microbes is not a small task, but will improve our understanding not only of this probiotic, but also how these common members of the avian microbiome impact the animal in the absence of probiotic treatment.

Other questions worth pursuing include: which members of this probiotic consortium

(L. acidophilus, L. casei, E. faecium, and B. bifidum) are necessary to stimulate lymphocyte energy metabolism and activity? With several indications that butyrate production could increase as a result of probiotic supplementation (increased butyrate producers and higher proportion of butyrate metabolism genes), and local and systemic gene changes consistent with the effects of butyrate (↓ NF-κB and IL-12, ↑ IL-10), can supplementation with butyrate

162

stimulate a response similar to that seen with this probiotic? Do these probiotic species produce high levels of butyrate? Sequencing the genomes of probiotic isolates may reveal this or other strain-specific traits that could account for some of the effects seen in this research.

While our understanding of this probiotic, and through it how the microbiome regulates host immune activity, has increased tremendously throughout this process, the answers to these and other questions will be invaluable in the future. They will enable scientists and producers to pinpoint bacterial species with the traits necessary to stimulate desired host responses, improving animal health and productivity.

163

REFERENCES

1. Ahn J, Sinha R, Pei Z, Dominianni C, Wu J, Shi J, Goedert JJ, Hayes RB, Yang L. 2013. Human gut microbiome and risk for colorectal cancer. J Natl Cancer Inst 105:1907–1911.

2. Lee SH, Lillehoj HS, Dalloul RA, Park DW, Hong YH, Lin JJ. 2007. Influence of Pediococcus-Based Probiotic on Coccidiosis in Broiler Chickens. Poult Sci 86:63–66.

3. Round JL, Mazmanian SK. 2010. Inducible Foxp3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proc Natl Acad Sci 107:12204– 12209.

4. Lee YK, Menezes JS, Umesaki Y, Mazmanian SK. 2010. Proinflammatory T-cell responses to gut microbiota promote experimental autoimmune encephalomyelitis. Proc Natl Acad Sci 108:4615–4622.

5. Chichlowski M, Croom J, McBride BW, Daniel L, Davis G, Koci MD. 2007. Direct- Fed Microbial PrimaLac and Salinomycin Modulate Whole-Body and Intestinal Oxygen Consumption and Intestinal Mucosal Cytokine Production in the Broiler Chick. Poult Sci 86:1100–1106.

6. Bai SP, Wu AM, Ding XM, Lei Y, Bai J, Zhang KY, Chio JS. 2013. Effects of probiotic-supplemented diets on growth performance and intestinal immune characteristics of broiler chickens. Poult Sci 92:663–670.

7. Qiu R, Croom J, Ali RA, Ballou AL, Smith CD, Ashwell CM, Hassan HM, Chiang C- C, Koci MD. 2012. Direct fed microbial supplementation repartitions host energy to the immune system. J Anim Sci 90:2639–2651.

8. Pieper R, Gatlin CL, Makusky AJ, Russo PS, Schatz CR, Miller SS, Su Q, McGrath AM, Estock MA, Parmar PP, Zhao M, Huang S-T, Zhou J, Wang F, Esquer-Blasco R, Anderson NL, Taylor J, Steiner S. 2003. The human serum proteome: display of nearly 3700 chromatographically separated protein spots on two-dimensional electrophoresis gels and identification of 325 distinct proteins. Proteomics 3:1345–1364.

9. Psychogios N, Hau DD, Peng J, Guo AC, Mandal R, Bouatra S, Sinelnikov I, Krishnamurthy R, Eisner R, Gautam B, Young N, Xia J, Knox C, Dong E, Huang P, Hollander Z, Pedersen TL, Smith SR, Bamforth F, Greiner R, McManus B, Newman JW, Goodfriend T, Wishart DS. 2011. The Human Serum Metabolome. PLOS ONE 6:e16957.

164

10. Dalloul RA, Lillehoj HS, Shellem TA, Doerr JA. 2003. Enhanced mucosal immunity against Eimeria acervulina in broilers fed a Lactobacillus-based probiotic. Poult Sci 82:62–66.

11. Lee K, Lillehoj HS, Siragusa GR. 2010. Direct-Fed Microbials and Their Impact on the Intestinal Microflora and Immune System of Chickens. J Poult Sci 47:106–114.

12. Haghighi HR, Gong J, Gyles CL, Hayes MA, Zhou H, Sanei B, Chambers JR, Sharif S. 2006. Probiotics Stimulate Production of Natural Antibodies in Chickens. Clin Vaccine Immunol 13:975–980.

13. Qiu R, Croom J, Ali RA, Ballou AL, Smith CD, Ashwell CM, Hassan HM, Chiang C- C, Koci MD. 2012. Direct fed microbial supplementation repartitions host energy to the immune system. J Anim Sci 90:2639–2651.

165