An Integrated Study on Chicken Gut Microbiome Associated with Diets and Feed Utilization

Using Microarray and Illumina Sequencing

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Deng Pan

Graduate Program in Animal Sciences

The Ohio State University

2014

Dissertation Committee:

Dr. Zhongtang Yu, Advisor

Dr. Hua Wang

Dr. Michael Lilburn

Dr. Thaddeus Ezeji

Copyrighted by

Deng Pan

2014

Abstract

The gastrointestinal (GI) tract of chicken harbors a complex and dynamic microbiome that is almost exclusively composed of bacteria. This bacterial community is considered as an essential component which contributes to the wellbeing of animal host in a wide range of aspects, especially nutrition and disease resistance. The overall objective of the studies presented here was to better understand the role gut microbiome plays in host growth performance and disease resistance by investigating the diversity of gut microbiome and identifying bacteria with different relative abundance in chickens with different feed utilization efficiencies and fed with different diets.

In the first study (Chapter 3), the relationship between gut microbiome and host feed utilization efficiency was investigated. Poultry Intestinal Tract Chip version 2 (PITChip2), a poultry specific phylogenetic microarray was used to characterize gut microbiome structures in 24 male Cobb 500 broiler chickens with low feed conversion ratio (FCR) and 24 chickens with high FCR. No clear separation in the gut microbiome structure was observed between good- and bad-performance chickens as indicated by hierarchical clustering (HCL) and principal component analysis (PCA). As indicated by significant difference (p≤0.05) in relative abundance between high FCR and low FCR chickens, 100 phylotypes were found to correlate with bird performance, with 48 of them showing greater than 3-fold differences. Five phylotypes were more abundant in good- performance low FCR chickens. Three of them represent species of Bacteroides,

ii

Peptococcus, and Ruminococcaceae. The other two can only be classified as Bacteria.

These five phylotypes may serve as indicators of feed utilization efficiency in chicken and also as potential probiotic candidates or targets of enhancement by prebiotics to improve growth performance of broilers.

In the second study (Chapter 4), PITChip3, the latest version of PITChip, was developed.

It has 1,204 customized oligonucleotide probes designed based on the V1-V3 region of bacterial 16S rRNA and are able to detect 62 genus-level phylotypes, 662 species-level phylotypes, and 34 pathogens. Each PITChip3 slide has six microarrays and every probe has 4 replicates randomly located on each microarray. The optimal hybridization temperature of PITChip3 was determined to be 48℃. At the optimal hybridization temperature, PITChip3 has a linear detection range from 3×107 to 3×1010 copies of a target per hybridization reaction. The utility of PITChip3 was tested using metagenomic

DNA samples extracted from the cecal content samples of broilers fed either corn-based or wheat-based diet. Clear separations between chickens fed with different diets and between chickens at two different ages were observed on PCA plots. Eighteen species- level phylotypes, including Akkermansia muciniphila, Clostridium perfringens, and

Corynebacterium variabile, along with 2 genus-level phylotypes Bacteroides and

Escherichia/Shigella/Salmonella, were further identified with different relative abundance in chickens fed different diets. Such difference in gut microbiome revealed by

PITChip3 may help to better understand the interactions between gut microbiome and diet and the benefits and risk associated with the two common diets.

iii

In the last study (Chapter 5), the difference in gut microbiome between chickens fed with corn- or wheat-based diet was further investigated using Illumina sequencing. Two hundred 1-day-old straight run Cobb 500 broiler chicks were randomly assigned to one of two dietary treatments. Each dietary treatment had 20 replicate pens with 5 birds per pen.

At 14 and 35 days of age, ileal mucosa and cecal content were collected from one bird per pen. Metagenomic DNA was extracted from those samples and Illumina MiSeq platform was used to sequence the V1-V3 hypervariable region (2×300bp paired-end) of bacterial 16S rRNA gene. A phylotype-based analysis was performed and no clear difference in microbial diversity between the two dietary treatments was observed. Yet we were able to identify phylotypes that have different relative abundance in 14-day-old chickens fed with the two different diets. One phylotype representing unclassified members in the family Ruminococcaceae and one representing unclassified Bacteria had higher relative abundance in the cecal content of chickens fed with corn-based diet, whereas a phylotype classified as Escherichia/Shigella had higher relative abundance in chickens fed with wheat-based diet. It is known that chickens fed with corn-based diet perform better and are less likely to have Clostridium perfringens induced necrotic enteritis (NE). Further studies on the phylotypes identified may help to find non- antibiotic alternatives to enhance growth and protect chickens from disease such as NE.

Taken together, the series of studies provided a comprehensive characterization of gut microbiome in chickens fed with corn- and wheat-based diets and also in chickens with different feed utilization efficiencies. These studies expanded our knowledge on the potential role gut microbiome plays in host growth performance and disease resistance, iv and established a poultry specific phylogenetic microarray which can serve as a powerful tool for the study of poultry gut microbiome.

v

Acknowledgements

Foremost, I would like to express my sincere gratitude to my advisor Dr. Zhongtang Yu for his support and the guidance during my Ph.D. study at OSU. I appreciate him giving me the opportunity to advance my education. His immense knowledge and inspiration helped me in all the time of my research and dissertation writing. Besides my advisor, I would also like to thank Drs. Hua Wang, Michael Lilburn, and Thaddeus Ezeji for their service on both my candidacy and dissertation committee. The insightful suggestions from my committee were of great value to my research.

I also extend my gratitude to the former and present members of Dr. Yu’s lab: Jill

Stiverson, Lingling Wang, Yueh-Fen Li, Shan Wei, Dr. Paul, Minseok Kim, Tansol Park,

Amlan Patra, Wen Lv, Hao Wu, Elmerson Ferreira de Jesus, Jackie Gano, Bethany

Denton, Gabriella Cobellis, and anyone else I may have missed, for their willingness to share knowledge and experience with me, as well as their support and friendship for the past three years. In particular, I am grateful to Lingling Wang, Yueh-Fen Li, and Shan

Wei for their helps in troubleshooting and discussion.

Last but not the least I would like to thank my beloved wife Fangfei for her unconditional love and support. I would also like to thank my family for being a constant source of support and this dissertation would certainly not have existed without them.

vi

Vita

2009...... B.S. Animal Sciences, Zhejiang University, China

2011...... M.S. Animal Sciences, Purdue University

2011 to present ...... Graduate Research Associate, Department of

...... Animal Sciences, The Ohio State University

Publications

Pan, D., & Yu, Z. (2013). Intestinal microbiome of poultry and its interaction with host and diet. Gut Microbes, 5(1), 108-119.

Fields of Study

Major Field: Animal Sciences

Focus: Poultry Intestinal Microbial Ecology

vii

Tabel of Contents

Abstract ...... ii

Acknowledgments...... vi

Vita ...... vii

List of Tables ...... xiii

List of Figures ...... xiv

Chapter 1: Introduction ...... 1

Chapter 2: Literature Review ...... 5

2.1 Intestinal microbiome of poultry ...... 5

2.2 Interactions between gut microbiome and poultry host ...... 8

2.2.1 Nutritional interactions ...... 8

2.2.2 Microbiome affects intestinal morphology and physiology ...... 13

2.2.3 Microbiome and immunity ...... 15

2.3 Interactions between gut microbiome and diet ...... 20

viii

2.3.1 Dietary Components Affect Gut Microbiome ...... 20

2.3.2 Antibiotic growth promoters ...... 22

2.3.3 Prebiotics ...... 23

2.4 Interactions among avian gut microbes ...... 24

2.4.1 Competition for nutrient and attachment site ...... 24

2.4.2 Production of bacteriostatic and bactericidal substances ...... 27

2.4.3 Horizontal gene transfer ...... 28

2.4.4 Probiotics ...... 29

2.4.5 Poultry litter microorganisms influence gut microbiome ...... 31

2.5 Methods for studying microbiome diversity ...... 32

2.5.1 Cultivation-based methods ...... 32

2.5.2 Cultivation-independent methods ...... 34

2.6 Bioinformatic and phylogenetic analyses of microbiome studies ...... 40

2.6.1 Measurement of microbial diversity ...... 40

2.6.2 Tools to carry out bioinformatic and phylogenetic analysis of microbiome

studies ...... 42

2.7 Summary ...... 44 ix

Chapter 3: Comparative analysis of gut microbiome between broilers with low and high feed conversion ratios ...... 46

3.1 Abstract ...... 46

3.2 Introduction ...... 47

3.3 Experimental Procedures ...... 49

3.3.1 Sample collection and DNA extraction ...... 49

3.3.2 Sample preparation and labeling ...... 50

3.3.3 Microarray hybridization and washing of microarray slides ...... 51

3.3.4 Signal detection and data analysis ...... 52

3.4 Results and Discussion ...... 53

3.4.1 Chickens with high and low FCR have similar cecal microbiome structure ..53

3.4.2 A universal core microbiome was identified for all 48 chickens ...... 55

3.4.3 Identification of phylotypes associated with FCR ...... 56

3.5 Conclusion ...... 60

Chapter 4: Development of PITChip3 and comparative analysis of gut microbiome of broilers fed corn- or wheat-based diet ...... 68

4.1 Abstract ...... 68

x

4.2 Introduction ...... 69

4.3 Experimental Procedures ...... 71

4.3.1 Oligonucleotide probe design ...... 71

4.3.2 Fabrication of PITChip3 ...... 74

4.3.3 Sample collection and DNA extraction ...... 75

4.3.4 Sample preparation and labeling ...... 75

4.3.5 Microarray hybridization and washing of microarray slides ...... 76

4.3.6 Signal detection and data analysis ...... 77

4.3.7 Linear detection range and optimal hybridization temperature determination79

4.4 Results and Discussion ...... 80

4.4.1 PITChip3 design ...... 80

4.4.2 Optimal hybridization temperature and linear detection range determination 82

4.4.3 Microarray data summary ...... 84

4.4.4 Impact of dietary grain on cecal microbiome structure of chickens ...... 84

4.4.5 Succession of cecal microbiome ...... 88

4.5 Conclusion ...... 89

xi

Chapter 5: Identification of chicken intestinal bacteria associated with corn-based or wheat-based diet...... 96

5.1 Abstract ...... 96

5.2 Introduction ...... 97

5.3 Experimental Procedures ...... 99

5.3.1 Birds and Diet ...... 99

5.3.2 Sample collection and DNA extraction ...... 100

5.3.3 High throughput sequencing and analysis of sequencing data ...... 101

5.4 Results and Discussion ...... 102

5.4.1 Summary of birds’ bodyweight data ...... 102

5.4.2 Summary of the sequence data ...... 102

5.4.3 Comparison of microbiome diversity ...... 104

5.4.4 Identification of phylotypes with differential relative abundance in different

dietary groups ...... 107

5.5 Conclusion ...... 110

Chapter 6: General discussion ...... 121

References ...... 125

xii

List of Tables

Table 5.1 Dietary composition ...... 112

Table 5.2 Predominant phylotypes at each taxonomic rank ...... 113

Table 5.3 Alpha diversity measurements ...... 114

xiii

List of Figures

Figure 2.1 Conceptual model of the interactions among gut microbiome, avian host, diet, and litter microbiome ...... 45

Figure 3.1 A hierarchical tree generated from HCL analysis ...... 62

Figure 3.2 PCA plot showing the grouping of 48 chickens based on the structure of their gut microbiome ...... 63

Figure 3.3 Universal core microbiome of the 48 chickens ...... 64

Figure 3.4 Sixteen predominant phylotypes of the universal core microbiome ...... 65

Figure 3.5 One hundred phylotypes differing in abundance (p≤0.05) between birds with low or high FCR...... 66

Figure 3.6 Phylotypes showing greater than 3-fold differences in mean relative abundances between good- and poor-performance birds...... 67

Figure 4.1 Linear detection range of PITChip3 ...... 90

Figure 4.2 A Venn diagram showing the number of phylotypes shared among different groups of chickens ...... 91

Figure 4.3 PCA plots showing the grouping of chickens based on the structure of their gut microbiome at 14 (A) and 35 (B) days of age ...... 92

xiv

Figure 4.4 Phylotypes with different abundance in 14-day-old chickens ...... 93

Figure 4.5 Phylotypes with different abundance in 35-day-old chickens ...... 94

Figure 4.6 PCA plots showing the grouping of chickens fed with corn-based diet (A) and wheat-based diet (B) based on the structure of their gut microbiome ...... 95

Figure 5.1 Average pen weight ...... 115

Figure 5.2 Dendrogram showing similarity among ileal mucosa samples ...... 116

Figure 5.3 Dendrogram showing similarity among cecal content samples ...... 117

Figure 5.4 PCoA plot showing the grouping of chickens based on the structure of their ileal mucosa microbiome ...... 118

Figure 5.5 PCoA plot showing the grouping of chickens based on the structure of their cecal content microbiome ...... 119

Figure 5.6 Three phylotypes with differential (p≤0.05; q≤0.05) relative abundance in cecal content of 14-day-old chickens fed with different diet ...... 120

xv

Chapter 1: Introduction

The gastrointestinal (GI) tract of poultry is densely populated with various microbes, mostly bacteria, which play critical roles in host nutrition and health (Johansson, 1948).

Functioning as an interface between the host and the feed ingested, members of gut microbiome hydrolyze and then ferment indigestible carbohydrates, producing short chain fatty acids (SCFAs) which serve as extra energy and carbon sources to be utilized by the host (Van der Wielen et al., 2000; Hooper et al., 2002; Koutsos and Arias, 2006;

Tellez et al., 2006). In addition, gut microbiome as a whole helps the development of GI tract and educate birds’ immune system and reduces risk of enteric diseases, assuring the functionality of the digestive system (Furuse and Okumura, 1994; Gabriel and Mallet,

2006; Brisbin et al., 2008; Crhanova et al, 2011).

In modern broiler production industries, feed utilization efficiency is of critical concern for broiler producers. It is observed that even when chickens with similar genetics are raised together and fed the same diet, feed conversion ratios (FCR) can be quite variable among individual bird (Stanley et al., 2012). Since gut microbiome affects host nutrition in various aspects, the variation in FCR could be caused, at least partially, by differences in gut microbiome structure. In order to explore such possibility, we used a poultry specific phylogenetic microarray (PITChip2) designed by Wei (2013) to investigate the gut microbiome of chickens with high and low FCR and look for groups of bacteria that are potentially associated with host feed utilization efficiency (Chapter 3).

1

As a robust and comprehensive analytic tool, PITChip2 enabled comparative analysis of poultry microbial ecosystems at a broad scope. However, through the use of PITChip2, we realized that it has several limitations. First, probes on PITChip2 were designed from the V3 region (approximately 200bp) of bacterial 16S rRNA gene sequence, limiting its ability to accurately detect and identify bacteria present in chicken’s gut. Second, approximately 40% of the probes on PITChip2 either are shared probes or multi-target probes which, in contrast to unique probes, can hybridize to multiple targets. These probes are included when no unique probe could be designed for a target to increase detection range. However, the ‘ambiguous’ results from these probes make it is difficult to pin point to specific bacteria that are detected. Last, on PITChip2 only one probe was designed for each target, making it difficult to discriminate cross-hybridization induced false positives from true positives. Therefore, we optimized PITChip2 and developed

PITChip3 by designing multiple probes per target over a longer region of 16S rRNA genes (Chapter 4). We also tested the utility of PITChip3 using metagenomic DNA samples recovered from cecal content of 16 broiler chickens fed with either corn-based or wheat-based diet.

Both corn and wheat are widely used to formulate diet in broiler industry. Compared to wheat-based diet, chickens fed with corn-based diet usually have better growth performance and lower risk of necrotic enteritis caused by virulent Clostridium perfringens. Previous studies have suggested that corn has higher nutritional value whereas wheat contains high levels of indigestible non-starch polysaccharides which interfere with the mixing of digestive enzymes and substrates, thus negatively affect the 2 growth performance of chickens (Wang et al., 2005; Slominski, 2011). These studies, however, did not consider the potential roles gut microbiome could play in chickens’ growth. Although the chemical difference between corn and wheat may be the major reason explaining the difference in growth performance, it was hypothesized that the different performance might be partially caused by difference in members of the gut microbiome in chickens fed with the two cereal grains. Another difference in these two diets is that wheat-based diet predisposes young chickens to Clostridium perfringens induced necrotic enteritis (NE), while incidence of NE is less frequently seen in young birds fed with corn-based diet (Annett et al., 2002; Jia et al., 2009). Proposed wheat- associated predisposing factors include the high level of non-starch polysaccharides, which leads to increased digesta viscosity, decreased digesta passage rate, and a decline in nutrient digestibility (Choct et al., 1996; Timbermont et al., 2011), and potential C. perfringens proliferation promoting factors in digested wheat (Immerseel et al., 2004).

Again, these studies gave little attention to the gut microbiome. Extensive interactions, such as competition, cooperation, and antagonism, exist among different members of the gut microbiome (Pan and Yu, 2013). It is likely that wheat and corn first affect the populations of some members of gut microbiome, which in turn affect the proliferation of

C. perfringens in the gut. To investigate if the difference in growth performance and NE resistance observed in chickens fed with corn or wheat can be explained by the changes in gut microbiome, Illumina sequencing was used to characterize bacterial community in the gut and identify bacteria associated with different diets (Chapter 5).

3

Collectively, the series of studies presented here are aimed at providing a comprehensive characterization of gut microbiome in chickens fed with different diets and also in chickens with different feed utilization efficiencies. These studies have expanded our knowledge on the potential role gut microbiome plays in host growth performance and disease resistance, which may help to develop alternative strategies that can replace

AGPs in poultry industry.

4

Chapter 2: Literature Review

2.1 Intestinal microbiome of poultry

The GI tract of poultry (e.g. chicken, turkey, and duck) consists of esophagus, crop, proventriculus, gizzard, small intestines (duodenum, jejunum, and ileum), cecum, colon, and cloaca. Relative to body length, the poultry GI tract is much shorter than that of mammalian animals. As such, the digesta passes through the entire GI tract faster in poultry than in mammalians. Although diet and feeding can have an effect on passage rate, the average whole tract transit time is less than 3.5 hours (Hughes, 2008). Such a short retention time selects bacteria that can adhere to the mucosal layer and/or grow fast.

On the other hand, the ceca, which are two blind pouches, have a rather slow passage rate, which makes them ideal habitats for a diverse microbiome that has considerable effect on host nutrition and health. The cecal microbiome is indeed the most studied intestinal microbiome of poultry.

The cecum of both chickens and turkeys harbors a complex microbiome, which is almost exclusively composed of bacteria (Wei et al., 2013b). Early cultivation-based studies revealed low abundances of lactobacilli (>104/g colony forming units, CFUs) and clostridia (102-104/g CFUs) in the small intestines and high abundance (1010-1011/g microscope counts) of anaerobic bacteria in the cecum of chickens (Barnes et al., 1972;

5

Salanitro et al., 1974). Identified bacteria included anaerobic Gram-negative cocci, facultative anaerobic cocci, and streptococci. Peptostreptococcus, Propionibacterium,

Eubacterium, Bacteroides, and Clostridium were the major genera that were recovered from cecum by cultivation. Between 20-60% of the total cecal bacteria could be cultivated depending on the media used (Barnes et al., 1972; Salanitro et al., 1974; Lu et al., 2003). Temporal changes were also observed as chicken aged (Barnes et al., 1972).

The first cultivation-based study on the intestinal microbiome of domesticated turkeys was reported in 1983 (BEDBURY and Duke, 1983). Most (77%) of the microbes were

Gram-positive rods, followed by Gram-negative rods (14%), and Gram-positive cocci

(9%). Bacteria of Eubacterium, Lactobacillus, Peptostreptococcus, Escherichia coli,

Propionibacterium, and Bacteroides were isolated as predominant microorganisms.

Although only revealing a limited number and diversity of bacteria, these early studies laid the foundation for microbiological studies of the intestinal microbiome in poultry.

Because of its high information content, universal distribution, and conservative nature, small subunit ribosomal RNA (16S rRNA) is an ideal phylogenetic marker for establishing distant relationships among bacteria, and using 16S rRNA gene-targeted analyses as a cultivation-independent tool to classify microorganisms and investigate microbial phylogenetic relationship has gained wide acceptance since the mid-1980’s

(Murray, 1984; Lane et al., 1985). In early 2000’s, 16S rRNA gene-targeted analyses started to play important roles in studying poultry gut microbiome. By using random cloning and terminal restriction fragment length polymorphism (T-RFLP), Gong et al.

(2002a; 2002b) reported that Lactobacilli, Enterococcus cecorum, and butyrate- 6 producing bacteria were the largest groups detected in ileum and ceca of broiler chickens.

In addition, they found that bacterial populations in the ileum were different from those in the cecum, and bacterial populations present in the mucosa and lumen were also different. Another study at the same period revealed that, while certain phylogenetic subgroups were common to all birds analyzed, there also existed differences among the cecal microbiome of individual birds reared under very similar conditions (Zhu et al.,

2002). By using oligonucleotide fingerprinting of rRNA genes (OFRG) and dot blot quantification, Scupham et al. (2008) characterized and compared the cecal mirobiome of domestic and wild turkeys. The results of their study indicated that although the two groups had similar levels of community richness and evenness at high taxonomic levels, considerable divergence at low taxonomic levels were found in the ceca of domestic birds from those of wild birds, probably due to high-density production practices. Despite being widely used in poultry gut microbiome studies, the 16S rRNA gene-based microbial profiling techniques used in these studies can only provide limited information on poultry gut microbiome as only a small number of bacteria can be practically analyzed

(Wei, 2013).

Recently, 16S rRNA gene-based high-throughput analyses, such as next-generation sequencing (NGS), make it possible to comprehensively characterize the intestinal microbiome of poultry, and has greatly expanded our knowledge on the bacterial diversity present in the intestinal tract, particularly the cecum, of chickens and turkeys

(Wet et al., 2013). Through phylogenetic and statistical analysis of 16S rRNA gene sequences recovered from intestinal microbiome of both chickens and turkeys, a global 7 bacterial census was created for poultry intestinal microbiome (Wei et al., 2013b).

Although this census is not complete, it serves as a phylogenetic framework for the bacterial diversity in the intestinal microbiome of both chickens and turkeys. In total, 13 phyla of bacteria were found, but Firmicutes, Bacteroidetes, and Proteobacteria accounted for most (>90%) of the intestinal bacteria of chickens and turkeys. More than

900 species-equivalent operational taxonomic units (OTUs, defined at 0.03 phylogenetic distance) were found in chicken, and these OTUs represent 117 established genera of bacteria. For turkey, the census contained nearly 500 OTUs of bacteria within 69 existing genera. The most predominant genera found in both chicken and turkey were Clostridium,

Ruminococcus, Lactobacillus, and Bacteroides, but with different distribution between the two bird species. Chickens and turkeys have distinct intestinal microbiomes, sharing only 16% similarity at species-equivalent level. Genetic and other factors (e.g., diet, digesta passage rate, and rearing environment) may be attributable to the difference in intestinal microbiome composition between chicken and turkey.

2.2 Interactions between gut microbiome and poultry host

Extensive interactions occur between poultry host and its gut microbiome (Fig. 2.1).

These interactions are manifested particularly through exchange of nutrients, modulation of host gut morphology, physiology, and immunity.

2.2.1 Nutritional interactions

8

Most readily digestible dietary carbohydrates are digested and absorbed by the host in the proximal gut, leaving indigestible carbohydrates and residual digestible carbohydrates to bacteria residing the distal gut (Hooper et al., 2002). Many intestinal bacteria can hydrolyze indigestible dietary polysaccharides, oligosaccharides, and disaccharides to their compositional sugars, which can then be fermented by intestinal bacteria, yielding short chain fatty acids (SCFAs), primarily acetate, propionate, and butyrate. The SCFAs can be utilized by the host as energy and carbon source (Van der Wielen et al., 2000;

Hooper et al., 2002; Koutsos and Arias, 2006; Tellez et al., 2006). Such fermentation can be observed in most part of the avian gut (from crop to cecum) but primarily takes place in the cecum, which is densely populated with bacteria (Rehman et al., 2007). The above fermentation increases as young birds grow. Cecal acetate, propionate, and butyrate are undetectable in 1-day-old broilers. As the cecal microbiome becomes established, these

SCFAs reach high concentrations in 15-day-old broilers and remain stable afterwards

(Van der Wielen et al., 2000). In the cecum, SCFAs are absorbed across the epithelium by passive diffusion and enter a variety of metabolic pathways (Hooper et al, 2002). A previous study has provided evidence that SCFAs, especially butyrate, can serve as an important energy source for intestinal epithelial cells (Pryde et al., 2002). In addition, it is reported that SCFAs can regulate intestinal blood flow, stimulate enterocyte growth and proliferation, regulate mucin production, and affect intestinal immune responses (Hooper et al., 2002; Sanderson, 2004; Tellez et al., 2006).

Gut bacteria also contribute to host nitrogen metabolism. In birds, the intestinal and urogenital tracts meet at the cloaca where urine mixes with feces. Some urine may travel 9 to the ceca due to the retrograde peristalsis in the rectum (Denbow, 2000). Cecal bacteria can then catabolize uric acid to ammonia, which can be absorbed by the host and used to synthesize a few amino acids such as glutamine (Vispo and Karasov, 1997). Some of the dietary nitrogen is incorporated into bacterial cellular proteins. Therefore, gut bacteria themselves can be a source of amino acids (Metges, 2000). However, the majority of these bacterial proteins are lost to the host with the excretion of feces because most of the intestinal bacteria in birds reside in the cecum which does not have the ability to digest and absorb protein. Utilization of bacterial proteins is possible when chickens are housed on hard floors, where coprophagy (ingestion of feces) can occur and bacterial proteins can be digested and absorbed in proximal gut (Vispo and Karasov, 1997; Koutsos and

Arias, 2006).

A recent study demonstrated in vitro that the chicken intestinal microbiome required simple sugars and peptides for balanced growth whereas human intestinal microbiome preferred polysaccharides and proteins (Lei et al., 2012). Chicken microbiome also produced greater concentrations of SCFAs than human microbiome. Given the shorter digestive tract and faster digesta transit in poultry than in mammalian animals, more sugars and peptides may be available in the intestines of poultry than in the colon of human, which in turn selected an intestinal microbiome adapted to simple sugars and peptides.

Gut microbiome of poultry may also serve as a vitamin (especially B vitamins) supplier to its host (Hooper et al., 2002; LeBlanc et al., 2013). Similar as bacterial protein, most of

10 the vitamins synthesized by gut bacteria are excreted with feces because they cannot be absorbed in the cecum (Hooper et al., 2002). However, coprophagic birds may benefit from bacterial vitamin synthesis. This is evidenced by a greater vitamin requirement by chickens housed in wire cages, where coprophagy is prevented, than by chickens raised on hard floors (Vispo and Karasov, 1997).

In a reciprocal manner, birds can also provide some nutrients to intestinal bacteria. For instance, mucins produced by goblet cells of the gut are important sources of carbon, nitrogen, and energy for some commensal and pathogenic bacteria (Hooper et al., 2002;

Derrien et al., 2010). Few reports are available on mucin-utilizing bacteria of poultry origin, but studies on other animal species showed that a variety of bacteria can degrade mucins, including some species of Bifidobacterium (Ruas-Madiedo et al., 2008; Killer and Marounek, 2011), Bacteroides (Hooper et al., 2002), and Akkermansia muciniphila

(Derrien et al., 2008). These bacteria are able to attach to the mucus layer and secrete specific enzymes for mucin degradation (Derrien et al., 2010). Although mucin degradation by these bacteria has not been demonstrated in poultry yet, members of these species have been found in the gut of poultry, and it is reasonable to assume that some of the intestinal bacteria can and do degrade mucins in birds. The mucus layer of GI tract serves as a protective barrier for attached bacteria, and the constantly replenished mucin is an excellent source of nutrient for some gut bacteria. The ability to attach to and utilize mucin enables mucin-utilizing bacteria to outcompete other species on the surface of the mucus layer. As a result, these bacteria play an important role in enteric disease and health. 11

Despite the fact that birds and its intestinal inhabitants both benefit from the host-microbe nutrient exchange, some of the intestinal bacteria are sometimes found to compete with the host for nutrients. Gut microbiome has evolved with the host towards a symbiotic relationship, and in healthy birds direct competition for nutrients is limited, as most digestible nutrients are absorbed by the host in the small intestine, where bacterial density is low and bacterial utilization of nutrient is suppressed due to the low pH and short retention time (Rehman et al., 2007). However, when bacteria overgrow in the small intestine under certain circumstances, nutrients are captured and utilized by bacteria before normal absorption by host can take place (Fan and Sellin, 2009). In humans and mice, some intestinal bacteria can deconjugate bile acids thereby suppressing lipid digestion by the host (Miyata et al., 2011; Kuribayashi et al., 2012). Clostridium perfringens, streptococci, and some of the bifidobacteria and lactobacilli isolated from chickens are able to deconjugate bile acids, but it remains to be determined to what extent bacterial deconjugation of bile acids decreases lipid digestion in chicken.

In modern broiler production industries, feed represents the major portion of production cost. Efficiency in converting feed into body mass is thus of critical concern for broiler producers. Because gut microbiome plays such an important role in feed digestion and absorption, attentions have been drawn to the associations between gut microbiome and host feed utilization efficiency. By using microbial profiling on broiler chickens across various feeding trails, Torok et al. (2011) were able to identify groups of bacteria that are potentially associated with broiler growth performance. Recently, more comprehensive analyses using a NGS technology also revealed certain bacteria that might be associated 12 with growth performance of broiler chickens (Stanley et al., 2012; Stanley et al., 2013).

Future studies are needed to determine if these bacteria are the cause or consequence of variations in feed utilization efficiency.

2.2.2 Microbiome affects intestinal morphology and physiology

The early post-hatch period is a critical stage for poultry growth and health as the new hatchling switches its nutrient source from the yolk to carbohydrate- and protein-based diet (Gilbert et al., 2010; Cheled-Shoval et al., 2011). In order to accommodate the rapid transition of nutrient source, the digestive organs of newly hatched poults undergo both anatomical and physiological changes and are the most rapidly developing organs during the early post-hatch period (Uni et al., 1999). The rapidly developed intestinal tract provides an ideal niche for microbial colonization. In the meantime, gut microbiome also plays an important role in intestinal development. Previous studies using germ-free (GF) chickens indicated that, comparing with conventional birds, the small intestine and cecum of GF birds had a reduced weight and a thinner wall (Furuse and Okumura, 1994; Gabriel and Mallet, 2006). It has been suggested that SCFAs increases enterocyte growth and proliferation, which may partially explain the stimulating effect on intestinal growth by gut microbiome (Le Blay et al., 2000; Blottiere et al., 2003; Fukunaga et al., 2003). This premise was supported by the study of Muramatsu et al. (1993) who reported that feeding fermentable carbohydrates, which can stimulate microbial fermentation and consequently

SCFAs production, increased the gut weight in chicken.

13

Gut microbiome can also affect intestinal morphology of poultry. Intestinal villi are shorter and the crypts are shallower in GF birds or birds colonized with a low load of bacteria than in conventionally-raised birds (Gabriel and Mallet, 2006; Forder et al.,

2007). Dietary supplementation of three different probiotic species (Lactobacillus acidophilus, Bacillus subtilis, and Saccharomyces cerevisiae) also increased villus height in duodenum and villus height:crypt depth ratio in ileum of broilers (Chae et al., 2012).

Similarly, it has been reported that dietary inclusion of prebiotics (e.g. fructooligosaccharide and mannanoligosaccharide) or fermented feed (e.g. fermented cottonseed, soybean, and rapeseed meal) also result in increased villus height and villus height:crypt depth ratio in the small intestine of chicken (Sonmez and Eren, 1999; Xu et al., 2003; Feng et al., 2007; Chiang et al., 2010; Sun et al., 2013). Such morphology alterations are not likely a direct effect of these dietary supplements, but an indirect effect through the manipulation of gut microbiome structure (Xu et al., 2003). Intestinal morphology change can also be an outcome of infections caused by enteric pathogens.

For instance, chickens with Eimeria spp./C. perfringens-induced necrotic enteritis had significantly reduced villus heights and villus height:crypt depth ratio in comparison to unchallenged controls or challenged chickens fed zinc bacitracin/monensin (Golder et al.,

2011). Fasina et al. (2010) also demonstrated that mock-challenged chicks had significantly greater villus height, villus area, crypt depth, and villus height:crypt depth ratio than chicks challenged with Salmonella Typhimurium.

The activity of intestinal digestive enzymes can be affected by gut microbiome as well.

Compared with GF chickens, conventional birds had greater activity of intestinal alkaline 14 phosphatase (Palmer and Rolls, 1983). Diets that can induce changes in gut microbiome structure may also influence intestinal digestive enzyme activity. For instance, the activities of amylase and protease are elevated in broilers fed diets containing fermented cottonseed meal or fructooligosaccharides (Xu et al., 2003; Sun et al., 2013). Feeding broilers with fermented soybean meal instead of unfermented soybean meal increased the activities of protease, trypsin and lipase (Feng et al., 207). It was concluded that these diets stimulate certain bacteria (e.g. Bifidobacterium and Lactobacillus) that can increase digestive enzyme activity, while suppressing some bacteria (e.g. Escherichia coli) that can either impair digestive enzyme secretion by damaging the villus and microvillus of mucosa or secrete proteolytic enzyme to degrade digestive enzymes (Xu et al., 2003).

2.2.3 Microbiome and immunity

Colonization with microorganisms in the poultry gut occurs immediately after hatch and microbial succession follows until eventual establishment of a complex and dynamic microbiome (Brisbin et al., 2008). Digestive tract is the most important reservoir of microorganisms and extensive interaction between these non-self cells and host immune system takes place in the GI tract.

The inner surface of avian gut is coated with a gel-like mucus layer which is formed from mucin glycoprotein secreted by the goblet cells (Forder et al., 2012). This layer of mucin consists of an outer loose layer in which microorganisms can colonize and an inner compact layer which repels most bacteria (Hansson and Johansson, 2010). As a component of the intestinal mucosal innate immune system, the mucus layer prevents gut

15 microorganisms from penetrating into the intestinal epithelium and serves as the first line of defense against infection (Brisbin et al., 2008). A good example is the different pathogenicity of Campylobacter jejuni in chicken and human. In vitro studies have shown that C. jejuni is able to adhere and invade both chicken and human intestinal epithelial cells (Grant et al., 1993; Byrne et al., 2007). However, C. jejuni does not cause disease in chicken even though the chicken gut is heavily populated with this bacterium, whereas ingestion of C. jejuni-contaminated food may lead to severe gastroenteritis in human (Hermans et al., 2012). It has been shown that the chicken intestinal mucus is able to attenuate C. jejuni virulence by inhibiting its ability to adhere and invade intestinal epithelial cells (Byrne et al., 2007; Alemka et al., 2010b), whereas the human mucus- adherent E12 cell line was found to enhance C. jejuni adhesion and invasion (Alemka et al., 2010a). Thus, it has been suggested that the difference in intestinal mucus layer between chicken and human may contribute to the distinct pathogenesis of C. jejuni seen in these two hosts (Alemka et al., 2012). Another study reported that sialylated mucin is more abundant in conventional chicks while sulfated mucin is more predominant in birds with a low bacterial load (Forder et al., 2007). Such change in mucin composition can be observed as early as 4 days post-hatch and indicates a potential role gut microbiome plays in regulating the establishment of mucus layer (Forder et al., 2007). By using a chicken necrotic enteritis model (coccidial infection followed by C. perfringens inoculation),

Collier et al. (2008) also showed that infection with Eimeria acervulina and E. maxima enhanced host mucogenesis, which benefits the growth of mucolytic bacteria C. perfringens. Interestingly, as the severity of necrotic enteritis becomes greater, the

16 expression of mucin gene (e.g. MUC2) decreases (Forder et al, 2012). Such decline in mucogenesis is probably due to the severe necrosis of the intestinal mucosa which results in extensive shedding of goblet cells.

Another important component of the innate immune system that functions in the avian gut is the antimicrobial peptides present on the intestinal epithelial surface (Brisbin et al.,

2008). In poultry, the most important and well-studied antimicrobial peptides are β- defensins. They are small cationic peptides produced by avian macrophages, heterophils, and epithelial cells, and they can kill various intestinal pathogens by disrupting cell membrane permeability, which leads to cell lysis (Jenssen et al., 2006; Derache et al.,

2009). Brisbin et al. (2008) indicated that Salmonella infection increased the expression of β-defensin genes in chicken, whereas administration of probiotics prior to Salmonella inoculation resulted in a decline in the gene expression of β-defensins. However, in a study conducted by Derache et al. (2009), in vitro infection with live Salmonella

Enteritidis did not increase the expression of β-defensin genes in avian epithelial cell cultures. A possible explanation for this discrepancy is that the increase in β-defensin gene expression after in vivo Salmonella challenge was due to the recruitment of heterophils to the gut in response to Salmonella infection. Interestingly, in their study,

Derache et al. (2009) found that avian epithelial cells responded differently to live and heat-inactivated Salmonella Enteritidis: the expression of β-defensin gene AvBD2 in epithelial cells was increased after incubation with heat-inactivated Salmonella

Enteritidis. Such finding indicates that live Salmonella Enteritidis may be able to block

17 the induction of β-defensin gene expression in epithelial cells by a yet unknown mechanism and use this mechanism as a strategy to prevent itself from being eliminated by host immunity. Such a strategy may subsequently facilitate Salmonella Enteritidis to adhere to and invade the intestinal epithelium.

The cellular component of the avian innate system, such as macrophages and heterophils, also protects host from enteric infection. These cells can be found in peripheral circulation and the lamina propria. When intestinal microorganisms breach the intestinal epithelial barrier these immune cells are recruited to the site of infection, where they kill the invaders using a variety of strategies, such as phagocytosis and oxidative burst

(Brisbin et al., 2008). The post-hatch colonization of avian gut by commensal microorganisms typically leads to a mild inflammation, which in turn results in macrophage and heterophil infiltration into the lamina propria (Crhanova et al., 2011). In addition, increased influx of macrophages and heterophils to the lamina propria and villus epithelium can be observed in chickens infected with enteric pathogens such as

Salmonella Typhimurium and Salmonella Enteritidis (Van Immerseel et al., 2002; Fasina et al., 2010). Although leukocyte infiltration is a defense mechanism against microbial infection during acute inflammatory response, it is worth noting that some pathogens are able to take advantage of this defense mechanism and use it to facilitate its pathogenicity.

For instance, Salmonella is known as an intracellular pathogen which is able to survive and replicate in some host cells such as macrophages (Buchmeier and Heffron, 1991;

Cirillo et al., 1998). The influx of macrophages to lamina propria and villus epithelium may therefore help spreading the pathogen to other organs and causing systemic infection. 18

The interaction between gut microbiome and host innate immune system can leads to subsequent adaptive immune response. B cells and T cells, which elicit antibody- mediated and cell-mediated immune responses, respectively, are the two primary types of lymphocytes that are of fundamental importance in the adaptive immune system. In avian gut, B cells and T cells can be found in organized lymphoid tissues (e.g. cecal tonsils,

Peyer’s patches, and the bursa of Fabricius) and in more dispersed areas such as lamina propria and epithelium (Bar-Shira et al., 2003; Brisbin et al., 2008). It has been shown that manipulation of gut microbiome through administration of probiotics can influence antibody-mediated immune response. Birds receiving probiotics containing L. acidophilus, Bifidobacterium bifidum, and Streptococcus faecalis showed enhanced systemic antibody response to sheep red blood cells (Haghighi et al., 2005). In addition, intestinal IgG reactive to tetanus toxoid, and serum IgG and IgM reactive to tetanus toxoid and C. perfringens alpha-toxin were also increased in chickens fed the same probiotic product (Haghighi et al., 2006). Other studies suggest that various strains of lactobacilli have a stimulating effect on antibody-mediated response in chicken and such effect is dependent on the strain of Lactobacillus used and the type (layer- or meat-type) and age of the chicken (Koenen et al., 2004; Brisbin et al., 2011). However, it remains to be elucidated how probiotics enhance antibody-mediated immune response. It is speculated that probiotics can stimulate the production of Th2 cytokines (e.g. IL-4 and

IL-10), which may subsequently enhance the immune response mediated by antibody

(Haghighi et al., 2005).

19

Besides antibody-mediated response, cell-mediated immune response is also found to be affected by gut microbiome. By using germ-free, conventional, and gnotobiotic chickens,

Mwangi et al. (2010) demonstrated that enteric microbiome complexity had a dramatic influence on the gut T cell repertoire. Brisbin et al. (2012) reported that various

Lactobacillus species had the capacity to induce differential cytokine expression in T cells of chicken cecal tonsils which could contribute to intestinal homeostasis. In addition, it has been shown that after being challenged with Salmonella Typhimurium, broiler chickens treated with probiotics containing L. acidophilus, Bifidobacterium bifidum, and

Streptococcus faecalis had a significant decrease in gene expression of IL-12 and IFN-γ, which are important cytokines in cell-mediated response against intracellular pathogens, in cecal tonsil (Haghighi et al., 2008). It should be noted that besides pathogens and probiotic strains, commensal bacteria, at least some of them, may also affect the immune response. Future studies are needed to determine the types of such commensal bacteria and their importance to immune response in poultry.

2.3 Interactions between gut microbiome and diet

2.3.1 Dietary Components Affect Gut Microbiome

Diet has the greatest potential impact on the intestinal microbiome in poultry as dietary components that escape host digestion and absorption serve as the substrates for the growth of intestinal bacteria (Fig. 2.1). One of the most important impacts stems from the

20 use of wheat-, barley-, or rye-based diets. These diets contain high levels of indigestible, water-soluble, non-starch polysaccharides, favor the proliferation of C. perfringens and predispose young chicks to necrotic enteritis, whereas diets poor in non-starch polysaccharides, such as corn-based diets, do not (Annett et al., 2002; Jia et al., 2009). It has been suggested that high level of non-starch polysaccharides leads to increased digesta viscosity, decreased digesta passage rate, and a decline in nutrient digestibility, which in turn favors the growth of C. perfringens (Chot et al., 1996; Timbermont et al.,

2011). When compared to corn-based diet, wheat-based diets also affect a number of other bacteria (Torok et al., 2008; Shakouri et al., 2009). Even a small variation in dietary cereal grain composition can potentially affect the intestinal bacteria at strain level as demonstrated by Hammons et al. (2010) who showed that a standard corn-soybean ration favored Lactobacillus agilis type R5, whereas a ration high in wheat middlings favored L. agilis type R1. The source and level of dietary protein may also affect gut microbiome. It has been demonstrated that unlike soybean meal, which is widely used as a source of protein in poultry industry, fermented cottonseed meal as a protein source increases the population of lactobacilli and decreases the number of coliforms in cecum of broiler chickens (Sun et al., 2013). Diet with high percentages of animal protein (e.g. fishmeal) favors the growth of C. perfringens in the hind-gut of chicken and is considered as one of the predisposing factors of necrotic enteritis (Drew et al., 2004). In addition, it has been reported that C. perfringens was more abundant in the ileum of broiler chickens fed diet with animal fat (a mixture of lard and tallow) than chickens fed diet with soy oil,

21 indicating that gut microbiome can also be influenced by dietary fat source (Knarreborg et al., 2002).

Various feed additives in poultry diet can influence gut microbiome and some of them are used to modulate intestinal microbiome to reduce enteric pathogens. Dietary enzymes, such as xylanase and β-glucanase, increase intestinal lactic acid bacteria and decrease the population of adverse and pathogenic bacteria such as E. coli (Rodríguez et al., 2012).

Dietary supplementation with xylanase and β-glucanase can also offer chickens some protection against necrotic enteritis as the enzymes break down the non-starch polysaccharides in the diet and reduce the digesta viscosity (Engberg et al., 2004;

McDevitt et al., 2006; Owens et al., 2008). Dietary inclusion of some plant-derived essential oils has also been used to protect chickens from enteric disease. For instance, plant-derived trans-cinnamaldehyde and eugenol were shown to be effective at reducing

Salmonella Enteritidis colonization in 20-day-old broiler chickens (Kollanoor-Johny et al.,

2012). In addition, it has been demonstrated that a blend of essential oils containing thymol, carvacrol, eugenol, curcumin, and piperin reduced the colonization and proliferation of C. perfringens in the gut of broiler chickens (Mitsch et al., 2004).

2.3.2 Antibiotic growth promoters

Another class of feed additives that has drastic effect on intestinal microbiome is antibiotic growth promoters (AGPs) (Fig. 2.1). AGPs are a group of dietary antibiotics used at sub-therapeutic levels to improve feed efficiency, increase animal growth, and maintain animal health (Danzeisen et al., 2011). Dietary inclusion of AGPs has been 22 practiced in food animal industry for more than 50 years (Danzeisen et al., 2011; Lin et al., 2013). Although the precise mode of action of AGPs still remains to be elucidated, it is widely accepted that the growth-promoting effect of AGPs is primarily brought about through modulation of intestinal microbiome (Dibner and Richards, 2005). Adverse and pathogenic bacteria in the GI tract of chicken, such as E. coli, Salmonella spp., and C. perfringens, compete with the host for nutrient and may also damage the intestinal epithelium, which adversely affect the digestion and absorption function of the host

(Gunal et al., 2006). Inclusion of AGPs in poultry diet can inhibit the growth of enteric pathogens, reduce the incidence of disease and promote growth of the birds. However, due to the growing concern over widespread antibiotic resistance, there is a trend towards abolishing the use of AGPs. Most AGPs are banned in the European Union, and the

United States has started to reduce the use of AGPs, with a possible ban on AGPs in the not-so-distant future (Van Immerseel et al., 2004a). A negative outcome of banning

AGPs is potential increase in incidence of disease in chicken. For instance, after the

AGPs ban, C. perfringens-induced necrotic enteritis has become one of the most noticeable, emerging diseases of broiler chickens in Europe (Van Immerseel et al., 2009).

Therefore, non-antibiotic alternatives which can control disease and promote growth of chicken are of great interest.

2.3.3 Prebiotics

Prebiotics are indigestible food ingredients which benefits the host animal by serving as a substrate for one or several beneficial bacteria present in the intestine (Fig. 2.1), granting

23 these beneficial bacteria proliferative advantages over other bacteria (Van Immerseel et al., 2004a; Dahiya et al., 2006; Tellez et al., 2006). Most prebiotics are polysaccharides such as galactooligosaccharides (GOS) and fructooligosaccharides (FOS). It has been reported that dietary inclusion of GOS favored the growth of bifidobacteria in the GI tract of broiler chickens (Jung et al., 2008). Inclusion of FOS in an alfalfa molting diet significantly decreased cecal Salmonella Enteritidis counts in laying hens (Donalson et al.,

2008). Dietary supplementation with FOS also decreased C. perfringens and E. coli, and increased Lactobacillus diversity in chicken gut (Kim et al., 2011).

Mannanoligosaccharides (MOS) is another prebiotic used in poultry industry. In addition to stimulating beneficial bacteria, MOS can also block pathogen binding to mannan receptors on the mucosal surface, thus hampering the attachment to and colonization of intestinal epithelia by certain pathogenic bacteria, particularly Salmonella Typhimurium

(Spring et al., 2000).

2.4 Interactions among avian gut microbes

As in other microbiome, different members of the GI microbiome can have different interactions, such as competition, cooperation, and antagonism (Fig. 2.1). The interactions among different bacteria that are important to poultry production are overviewed.

2.4.1 Competition for nutrient and attachment site

24

Although the avian GI tract is an ideal habitat for microorganisms, it does not support unrestricted microbial growth or proliferation due to the limited availability of nutrient and space therein. Therefore, competition for these resources (i.e. nutrient and attachment site) among microorganisms is a common phenomenon in intestinal ecosystem (Soler et al., 2010). A good example is the competition for zinc among GI microbes. Zinc is an essential trace element required by both eukaryotic and prokaryotic cells and is involved in various cellular functions, such as enzymatic reactions and gene expression (Berg and

Shi, 1996; Gielda and DiRita, 2012). Under low-zinc conditions, C. jejuni uses the high- affinity ZnuABC transporter to bring zinc into cell (Davis et al., 2009). In a recent study,

Gielda and DiRita showed that both a wild-type C. jejuni strain and a znuABC- mutant strain of C. jejuni were able to colonize limited-mirobiota chicks at similar efficiencies, but only the wild-type C. jejuni strain was able to colonize conventional chicks (Gielda and DiRita, 2012). In the same study, it was also shown that the zinc level in cecal content was significantly lower in the conventional chicks than in the limited-microbiota chicks, suggesting that under low zinc conditions, C. jejuni lacking the high-affinity zinc uptake system was outcompeted by other bacteria present in the GI tract. The ZnuABC transporter system is found not only in C. jejuni but also in some pathogenic bacteria (e.g.

Salmonella Typhimurium and E. coli), making it a potential target for development of broad-spectrum antimicrobials (Patzer and Hantke, 2000; Campoy et al., 2002; Gielda and DiRita, 2012).

In order to cause infections in birds, enteric pathogens need to first attach to and then breach the intestinal epithelial barrier (Lan et al., 2005). In healthy birds, the commensal 25 bacterial communities in the GI tract colonize intestinal mucosa and form a layer of bacteria covering the mucosal surface. By occupying a diverse array of adhering niches along the GI tract, this layer of dense and complex microbial communities can effectively block the attachment and subsequent colonization by most invading enteric pathogens

(Lan et al., 2005; Lawley and Walker, 2013). This phenomenon is called “competitive exclusion” (Gabriel and Mallet, 2006). The GI tract of newly hatched chick is sterile, but is immediately colonized by microorganisms present in the surrounding environment

(Brisbin et al., 2008). In the wild, the GI tract of the new hatchling is rapidly colonized by members of the gut microbiome from its mother’s feces and is therefore protected from pathogen invasion (Lutful Kabir, 2009). However, in commercial poultry production the chicks are hatched in incubators and have no contact with hens. The surrounding environment is therefore relatively clean and usually has a microbial community distinct from the microbiome in a healthy adult chicken’s gut, which may lead to a delay in normal colonization and succession of intestinal microbiome (Dahiya et al., 2006; Lutful Kabir, 2009). Enteric pathogens in the environment may thus have a greater opportunity to attach to and breach intestinal mucosal layer and cause infection in new hatchlings as a result of the absence of a normal gut microbiome. This may partially explain why newly hatched chicks are particularly vulnerable to enteric infections such as necrotic enteritis (Lan et al., 2005; Dahiya et al., 2006). In order to protect newly hatched chicks from enteric disease, competitive exclusion cultures have been used by poultry producers to help newly hatched chicks to rapidly establish a healthy gut microbiome.

Competitive exclusion cultures are suspensions of intestinal contents obtained from

26 healthy adult birds (Nisbet, 2002). By oral administration to newly hatched poults, competitive exclusion cultures have been shown to be effective in protecting new hatchings from being infected by some pathogens such as Salmonella and C. perfringens

(Craven et al., 1999; Hollister et al., 1999; McReynolds et al., 2007).

2.4.2 Production of bacteriostatic and bactericidal substances

Another widely used strategy for some bacteria to gain competitive advantages is to produce bacteriostatic or bactericidal substances hostile to competitors. Previous studies have shown that lactic acid and other SCFAs produced by various commensal bacteria are inhibitory to certain pathogens. For instance, in vitro studies have shown that lactic acid bacteria ferment carbohydrates present in chickens’ feed and produce lactic acid, which lowers the pH in the surrounding environment and inhibits the growth of certain pathogens such as E. coli, Salmonella Typhimurium, and C. perfringens (Hinton et al.,

1992; Murry et al., 2004). An in vivo study also demonstrated a negative correlation between concentrations of SCFAs (in particular acetate, propionate, and butyrate) and abundance of the family Enterobacteriaceae in broilers’ ceca (Van der Wielen et al.,

2000). Such a negative correlation was further substantiated by an in vitro study conducted by the same researchers. It was proposed that, in addition to lowering extracellular pH, SCFAs in undissociated form can diffuse freely across the bacterial cell membrane into the cell, where they dissociate, lowering the intracellular pH that inhibits some essential enzymes or metabolism (Van der Wielen et al., 2000; Van Immerseel et al., 2004b; Van Immerseel et al., 2006).

27

Certain bacteria can also produce bacteriocins to selectively inhibit the growth of other bacteria. Bacteriocins are a group of antimicrobial peptides produced by bacteria and archaea (Dobson et al., 2012). Various strains of Lactobacillus salivarius isolated from chicken GI tract can produce bacteriocins which are inhibitory to some Gram-negative and Gram-positive bacteria such as Salmonella Enteritidis and C. jejuni (Stem et al., 2006;

Svetoch et al., 2011; Messaoudi et al., 2012). Bacteriocins produced by strains of

Enterococcus faecium, Pediococcus pentosaceus, and Bacillus subtilis isolated from broiler chicken are able to inhibit C. perfringens and Listeria monocytogenes (Teo and

Tan, 2005; Shin et al., 2008). In addition, it has been shown that several strains of E. faecium produce bacteriocins against the oocysts of poultry Eimeria spp. (Strompfova et al., 2010). The inhibitory effect on various adverse bacteria and pathogens makes bacteriocin production a frequently considered trait in selection of probiotics.

Nevertheless, it is worth noting that a variety of pathogenic bacteria (e.g. Staphylococcus aureus) also produce bacteriocin effective against competing bacteria (Wladyka et al.,

2012).

2.4.3 Horizontal gene transfer

Horizontal gene transfer is ‘the non-genealogical transmission of genetic material from one organism to another’ (Goldenfeld and Woese, 2007). It is mediated by processes such as conjugation, transformation, and transduction and is an effective mechanism which contributes to bacterial diversification and facilitates bacterial adaptation to new environments (Boto, 2010; Holmgren, 2010). In modern poultry industry, litter, which

28 contains bacteria excreted from chickens or turkey, is often used for multiple growth cycles. Once developed in the GI tract, antibiotic resistant bacteria can accumulate in the litter and recycle between the litter and GI tract over multiple growth cycles (Fig. 2.1).

Such a practice can greatly increase the incidence of horizontal transfer of resistance genes and may contribute to the wide spread of antimicrobial resistance among adverse and pathogenic bacteria and is thus of particular interest (Dhanarani et al., 2009;

Wimalarathna et al., 2013; You et al., 2013). In addition, virulence genes can also be exchanged among poultry enteric pathogens, increasing the recipient’s pathogenicity

(Johnson et al., 2010). The predominant commensal intestinal microorganisms usually possess certain traits which enable them to outcompete other bacteria (especially adverse and pathogenic bacteria) and survive in the GI tract. These traits, however, may be acquired by pathogens via horizontal gene transfer, making those pathogens more competitive. On the other hand, commensal bacteria may also become pathogenic to the poultry host by obtaining virulence factors from pathogens (Van Reenen and Dicks,

2011). Therefore, caution should be taken when using direct-fed microbials such as probiotics.

2.4.4 Probiotics

Probiotics are live microbial feed supplement used by livestock and poultry producers to protect animals from enteric pathogen infection and improve animal health (Dahiya et al.,

2006; Kabir, 2009). The mode of action of probiotics can vary depending on the traits of the specific probiotic strains/species used, but most probiotics benefit the host through

29 the following mechanisms: (1) inhibition of colonization by and proliferation of pathogenic bacteria through competition for nutrient and attachment site (Lan et al., 2005;

Lawley and Walker, 2013), (2) production of bacteriostatic and bactericidal substances against pathogens (Van der Wielen et al., 2000; Murry et al., 2004), (3) neutralizing enterotoxins (Knap et al., 2010), (4) enhancing gut barrier function (Yang et al., 2012), and (5) enhancing host immunity (Ng et al., 2009; Yang et al., 2012). The effect of different probiotics on chicken gut microbiome has been extensively investigated.

Several lactobacilli strains have been shown to decrease the population of Salmonella,

Campylobacter and some other non-beneficial bacterial groups in chicken gut (Pascual et al., 1999; Nakphaichit et al., 2011). Molnár et al. (2011) reported that dietary supplementation of Bacillus subtilis significantly decreased E. coli population in the ileum of chicken. Another study demonstrated that spores of Bacillus licheniformis could prevent C. perfringens-induced necrotic enteritis in broiler chickens (Knap et al., 2010).

Some strains of Clostridium butyricum are also potential probiotics that can be used in poultry production. This was demonstrated by Yang et al. (2012) who showed that C. butyricum HJCB998 significantly decreased cecal Salmonella and C. perfringens population while increasing Lactobacillus and Bifidobacterium populations in the cecum.

The protective effect of multispecies probiotics has also been investigated. A multispecies probiotics containing Enterococcus faecium, Bifidobacterium animalis,

Pediococcus acidilactici, L. salivarius, and Lactobacillus reuteri isolated from chicken gut decreased cecal coliform population (Mountzouris et al., 2010). Ghareeb et al. also demonstrated that multispecies probiotics containing E. faecium, P. acidilactici, L.

30 salivarius, and L. reuteri significantly reduced cecal colonization by C. jejuni, indicating that probiotic products can also be used to improve food safety by reducing the population of human pathogens, such as C. jejuni, in chicken (Ghareeb et al., 2012).

2.4.5 Poultry litter microorganisms influence gut microbiome

During their growth cycle, chickens continuously take up microorganisms from the surrounding environment. Poultry litter, the bedding material used in chicken houses, is usually mixed with chicken excreta and thus harbors a complex microbial community

(mostly intestinal bacteria), and is thus of a potential impact on chicken gut microbiome

(Fig. 2.1). Reusing litter for several growth cycles before a thorough clean-out is a management practice commonly used by poultry producers to reduce production cost and to help alleviate the challenges faced in litter disposal (Coufal et al., 2006). Reuse of poultry litter influences the microbial community resident in the litter, which may in turn affect chicken gut microbiome. In a recent study, Cressman et al. (2010) demonstrated that more environmental bacteria were found in fresh litter, while more bacteria of intestinal origin resided in reused litter. It was also found in the same study that the ileal mucosal microbiome of chickens reared on fresh litter was dominated by Lactobacillus spp., whereas a group of unclassified Clostridiales were the dominating bacteria in chickens reared on reused litter. It was also reported that microorganisms in reused poultry litter can function as competitive exclusion culture and delay ileal mucosal colonization by C. perfringens during early post-hatch period (Wei et al., 2013a). On the other hand, reused litter may also harbor disease-causing microorganisms from the

31 previous flock and thus serves as a source of pathogens to the subsequent flock (Stanley et al., 2004).

2.5 Methods for studying microbiome diversity

2.5.1 Cultivation-based methods

Before the era of -omics, studying microbiome diversity has been heavily relying on cultivation-based approaches. Morphological, developmental, and nutritional characteristics revealed by microbial cultures are traditionally used to classify microorganisms (Lane et al., 1985). The development of anaerobic culture techniques has further expanded our knowledge on microorganisms in the GI tract as gut microbiome is composed of mostly anaerobic bacteria (Zoetendal et al., 2004; Faber and Bäumler, 2014).

Indeed, early cultivation-based studies on GI microbes of chickens and turkeys have laid the foundation for studies on intestinal microbiome in poultry. However, when compared to cultivation-independent techniques, cultivation-based approaches suffer several drawbacks which make them less used in studying complex microbiome. Firstly, and most importantly, the majority of microorganisms present in the nature cannot be cultured with current cultivation techniques. It is estimated that our planet hosts 7 million prokaryotic species, but so far only 0.1% of the existing prokaryotes have been cultured

(Alain and Querellou, 2009). Compared to other microbial ecosystems, the culturable fraction of poultry cecal bacteria is relatively high (20-60%), but still the majority of cecal bacteria cannot be studied with current cultivation approaches (Barnes et al., 1972; 32

Salanitro et al., 1974; Lu et al., 2003; Zoetendal et al., 2004). Reasons for non- culturability include limited knowledge in the complex biotic and abiotic conditions required for the growth of microbes, lack of advanced technologies to mimic natural growth environment for microbes, nutritional and environmental stress caused by cultivation procedures, inability to simulate interactions among microbes and with host cells, cell death caused by accelerated substrate, and over growth of weed-like microorganisms (Zoetendal et al., 2004; Alain and Querellou, 2009; Prakash et al., 2013).

Secondly, cultivation-based approaches are biased for the growth of specific microorganisms. Artificial homogenous medium favors the growth of microorganisms with short generation time and also microorganisms that can efficiently utilize the substrate present in the medium, while suppressing the growth of others (Nocker et al.,

2007). Thus, microbial richness and diversity revealed by cultivation-based approaches are not an accurate representation of the actual in situ microbiome. Third, conventional cultivation methods are laborious and time consuming. For some microorganisms, adaptations to laboratory growth conditions may take a very long time as transition from a natural habitat to a synthetic medium is a quite stressful event (Alain and Querellou,

2009). In addition, some microorganisms, such as Mycobacterium tuberculosis, grow extremely slow even within their natural habitat (Hett and Rubin, 2008). The long incubation period makes it impractical to use cultivation-based approaches to perform microbiome study given the large numbers of different members present in the microbial ecosystem.

33

Although cultivation-based approaches have these intrinsic limitations and currently the majority of microbiome studies use cultivation-independent omics-based approaches, a comprehensive understanding of the biology and ecology of microorganisms will partially rely on cultivation-dependent techniques as the physiology and metabolism of microorganisms cannot be fully understood without cultivation (Prakash et al., 2013).

Efforts are now being made in refining culture media, reproducing cell-cell interactions, and developing advanced cultivation techniques such as high-throughput cultivation and in situ cultivation (Alain and Querellou, 2009). The continuous development in microbial cultivation techniques will help revealing more hidden secrets of the microbial world.

2.5.2 Cultivation-independent methods

The limitations of conventional cultivation-based approaches make cultivation- independent molecular approaches desirable in microbiome studies. For decades, marker gene-based methods have been used to investigate microbial communities in different environments. Because it is phylogenetically conserved, universally distributed, and not laterally transferred, the gene encoding 16S rRNA has been predominantly used as a phylogenetic marker by 16S rRNA gene-targeted approaches, such as cloning libraries, real-time PCR, 16S rRNA gene fingerprinting, NGS, and phylogenetic microarray, all of which have been widely used for microbiome studies (Gentry et al., 2006; Nocker et al.,

2007).

Construction of 16S rRNA gene cloning libraries, followed by sequencing of the cloning libraries, is a conventional cultivation-independent approach for microbiome studies. By

34 comparing the sequences of the cloned amplicons with sequences archived in public databases, such as Ribosomal Database Project (RDP) and GenBank, this method can identify both already known species and novel species that are either culturable or unculturable (Zoetendal et al., 2004; Nocker et al., 2007). However, because this method is laborious and cost-ineffective, researchers can only afford to sequence limited numbers of clones. In addition, this method does not provide quantitative information. Therefore, construction and sequencing of 16S rRNA gene cloning libraries is no longer the preferred or primary approach to investigate species richness and evenness of complex microbial communities (e.g. gut microbiome of animals and human).

Real-time PCR assay is another cultivation-independent approach widely used in microbiome studies. By measuring the increase in fluorescence signal generated during

PCR amplification (either from a fluorochrome added to the PCR mixture or from fluorochrome attached to probes), amplification of target sequences is monitored in real time during the assay (Albuquerque et al., 2009). Compared to other techniques, real-time

PCR is extremely sensitive and can detect targets in very low concentration, which makes it a powerful tool for microbial detection and diagnostics (Zoetendal et al., 2004). Real- time PCR is also the most reliable quantitative method to quantify individual groups

(species, or a higher taxonomic unit). However, similar with cloning libraries, real-time

PCR is not suitable for investigating complex microbiome as the number of targets is limited by cost and time. In addition, because primers are designed from already known sequences, real-time PCR assay does not provide the opportunity for novel species discovery. 35

16S rRNA gene fingerprinting refers to a set of molecular techniques that are used for rapid comparison of microbial communities. The frequently used 16S rRNA gene fingerprinting approaches include denaturing gradient gel electrophoresis (DGGE), temperature gradient gel electrophoresis (TGGE), single strand conformation polymorphism (SSCP), and terminal-restriction fragment length polymorphism (T-RFLP).

DGGE and TGGE separate small PCR products on acrylamide gel by their denaturing behavior, which is differentiated primarily by the GC content of the PCR products, under chemical denaturation gradient and temperature gradient, respectively (Muyzer et al.,

1993; Muyzer and Smalla, 1998). SSCP is an electrophoretic method that differentiates single strands prepared from PCR products by their secondary structure, which is determined by sequence diversity (Lee et al., 1996). DGGE, TGGE, and SSCP have a powerful discriminative ability and are able to differentiate small DNA fragments difference by a single base (Nocker et al., 2007). In fact, they were initially used to detect mutation in clinical research before being applied to microbiome studies (Zoetendal et al.,

2004). T-RFLP physically differentiates targets based solely on the fragment length of restriction-digested and fluorescent labeled PCR products (Liu et al., 1997). It has a resolution at single base pair and can thus be used for phylogenetic assignment by direct reference to public sequence databases (Dunbar et al., 2001). However, often it does not allow unambiguous identification of microbes. All these 16S rRNA gene fingerprinting techniques, compared to cloning libraries and real-time PCR, can provide a broader view over complex microbial communities such as gut microbiome. However, they are only semi-quantitative and they do not provide detailed sequence information to definitively

36 identify the microbes represented by the bands or peaks. Subsequent analysis is required to identify targets of interest.

NGS is a set of sequencing technologies which, in contrast to first-generation sequencing techniques such as Sanger sequencing, share the following advantages: (1) preparation of

NGS libraries uses a cell free system (i.e. no bacterial cloning), (2) high-throughput parallel sequencing reactions are able to produce millions of sequencing reads, (3) base interrogation is performed in parallel and does not require electrophoresis, and (4) fast and cost-effective (Van Dijk et al., 2014). 454 pyrosequencing and Illumina sequencing are two NGS platforms frequently used in microbial community studies. Released on

2005, 454 pyrosequencing was the first NGS developed (Margulies et al., 2005). It uses a sequencing by synthesis approach, in which pyrophosphate released during nucleotide incorporation is converted by sulfurylase into ATP, which is in turn used by luciferase to oxidize luciferin and lead to light emission (Buermans and den Dunnen, 2014). For each run, 454 pyrosequencing is able to generate approximately 1 million reads with a length of 500-800 bp. The long read lengths used to be an advantage of 454 pyrosequencing over other NGS technologies. However, some competitor platforms are now able to produce similar read lengths, making them more cost efficient than 454 pyrosequencing

(Buermans and den Dunnen, 2014). In addition, as the light intensities do not always reflect the actual homopolymer length, 454 pyrosequencing can generate a lot of artifactual sequences, which makes it difficult to interpret 454 pyrosequencing data

(Quince et al., 2009). Illumina sequencing was commercialized a year after 454 pyrosequencing was released. During sequencing, four nucleotides are labelled with 37 different fluorescent labels which also serve as reversible terminators, and one nucleotide is read at a time so there is no homopolymer issue for Illumina sequencing (Van Dijk et al., 2014; Buermans and den Dunnen, 2014). Illumina sequencing is considered as an ultra-high-throughput approach (Caporaso et al., 2012). For each run, approximately 3 billion reads can be generated from HiSeq2000/2500 platform and approximately 25 million reads can be produced from MiSeq platform. In addition, the MiSeq platform is able to produce 2×300 paired-end reads, which is similar to the reads generated from 454 pyrosequencing (Buermans and den Dunnen, 2014). Since its first release a decade ago, the fast developing NGS technologies have revolutionized the study of microbial community, granting researchers the ability to study microbiome at a broad scope never possible before. However, NGS technologies have several inherent limitations that may considerably compromise assessment of the microbial community. First, multiple rounds of PCR are required for library preparation. However, PCR is known to produce selective biases and PCR artifacts such as chimeras, mutations, and heteroduplex molecules

(Reysenbach et al., 1992; Qiu et al., 2001; Kurata et al, 2004). Therefore, the frequencies of individual NGS reads do not necessarily reflect the relative abundance of the microbes represented by each sequence read (Brugère et al., 2009). Second, all the NGS sequencing technologies themselves also produce artificial sequences, and it can be difficult to identify and remove all the artificial sequences. Third, although bioinformatists have developed a series of improved software tools to handle the huge datasets as well as mitigate the negative impacts of artifactual sequences generated by

38

PCR and sequencing processes, it is still time-consuming and resource-demanding (e.g. computing power) to analysis sequencing data.

Phylogenetic microarray is another molecular technique that enables high-throughput analysis of microbial ecosystems. It was first developed to simultaneously analyze expression of thousands of genes (Scott et al., 2006). Now a single microarray chip can have hundreds of thousand probes, which can provide semi-quantitative information on the microbial ecosystem of much greater diversity (DeSantis et al., 2007). Compared to

NGS technologies, microarray has been shown to be advantageous for several reasons in comparative analysis of complex microbiomes. First, microarray can provide a uniformed, robust assessment (linear range up to >5 logs) of individual populations of complex microbiomes (Brodie et al, 2006). Second, nucleic acid samples can be directly applied on a microarray chip. Thus potential PCR bias is not likely to compromise the assessment of microbiomes. Third, microarray chips can be custom fabricated or printed in house at relatively low cost (less than $250 per microarray chip), updated quickly when new probes need to be added, and analysis of microarray data is much less resource demanding and time consuming. To date, phylogenetic microarrays have been developed for the assessment of human GI microbiome as well as complex microbiomes in other environments, and have gained recognition and acceptance as a highly efficient phylogenetic fingerprinting tool (DeSantis et al., 2007; Rajilić-Stojanović et al., 2009;

Kang et al., 2010; Tu et al., 2014). The main limitations of phylogenetic microarray include: (1) inability to detect novel sequences as all the probes are designed from already known sequences, (2) difficulty in designing a unique probe for every member in 39 a complex microbial community due to the highly conserved nature of 16S rRNA gene sequences and limited probe lengths (<70 bp), and (3) potential cross-hybridization and hybridization biases (e.g. some sequences hybridize more efficiently than others). These issues may compromise the interpretation of microarray data (Zoetendal et al., 2004;

Sekirov et al., 2010). In addition, as the cost of NGS technologies keeps dropping, cost- effectiveness may no longer be an advantage of phylogenetic microarray over other approaches.

2.6 Bioinformatic and phylogenetic analyses of microbiome studies

2.6.1 Measurement of microbial diversity

Measurement of diversity can help understanding the structure of microbial communities in a wide range of habitats and the compositional changes of microbial communities in response to external stimuli (e.g. changes in environmental conditions, treatments), and is thus a routine analysis in most microbiome studies. Depends on the focus of the study, one may be interested in the assessment of microbial diversity within a given community

(i.e. alpha diversity) or how the diversity of two or more microbial communities differ

(i.e. beta diversity), or both (Whittaker, 1972). Alpha diversity is usually characterized using species richness (i.e. the total number of species detected and predicted in the given microbial community) and species evenness (i.e. the relative abundance of the species). It can be calculated/represented using several methods/indices with some focusing on species richness, considering only the presence or absence of species, while others, in 40 addition to the presence or absence, also taking the relative abundance of each species into account, covering both species richness and evenness (Lozupone and Knight, 2008).

To estimate species richness in a microbial community, rarefaction curves and indices, such as Chao1 and abundance-based coverage estimators (ACE), are frequently used

(Bent and Forney, 2008; Lozupone and Knight, 2008). Rarefaction curves represent a function with cumulative number of observed OTUs as the dependent variable and sampling effort as independent variable. Species richness can be estimated from the rarefaction curves by using curve-fitting methods (Hughes et al., 2001). Chao1 and ACE are nonparametric estimators which estimate species richness by adding a correction factor to the observed number of species or OTUs. The correction factor in Chao1 is calculated based on the number of singletons (OTUs each represented by only one sequence) and doubletons (OTUs each represented by only two individual sequences). In contrast, the correction factor in ACE takes into account of all species or OTUs each represented by up to 10 individual sequences. Both Chao1 and ACE, however, tend to underestimate species richness at low sample sizes (Hughes et al., 2001). To estimate alpha diversity with consideration of both species richness and species evenness,

Shannon’s index and Simpson’s index are two approaches widely used (Lozupone and

Knight, 2008). Shannon’s index measures the entropy of a microbial community, whereas

Simpson’s index measures the probability that any two microorganisms sampled will be the same phylotype (Bent and Forney, 2008).

Initially applied as a measurement of changes in diversity of macroorganisms across environmental gradients or along transects (Whittaker, 1960), beta diversity is now 41 widely used to compare diversity between two or more microbial communities. Similar with alpha diversity, beta diversity can be assessed using qualitative approaches (i.e. approaches that only consider the presence/absence of species) and quantitative approaches (i.e. approaches that take species abundance into account). The Unique

Fraction metric (UniFrac) is an approach which measures the phylogenetic distance between different communities/samples based on the phylogenetic tree constructed from the species present in the communities/samples being compared. It is a qualitative approach because identical sequences do not contribute to the branch length to the tree

(Lozupone and Knight, 2005). In contrast, weighted UniFrac, which is a variant of the

UniFrac algorithm, is a quantitative measurement of beta diversity as it weights the branches in the phylogenetic tree and detects changes in abundance of each lineage

(Lozupone and Knight, 2008).

All these methods/indices are powerful approaches to characterize alpha and beta diversities of microbial communities. However, no single method/index can cover all aspects of diversity. Therefore, in an effort to comprehensively understand microbiome diversity, these methods/indices are usually used in combination.

2.6.2 Tools to carry out bioinformatic and phylogenetic analysis of microbiome studies

During the past ten years, the fast developing NGS technologies have boosted microbiome studies to a very fast pace and unraveled many hidden secrets of the microbial world. However, because these high-throughput technologies are able to generate millions and even billions sequence reads per run, analyzing sequencing data

42 from NGS technologies has always been a challenging task. Numerous computer programs have been built to carry out sequence analysis tasks such as trimming and screening sequences, sequence alignment, distance matrix calculation, OTU picking, tree building, and diversity calculation. Without these computational tools, it is almost impossible to analyze and interpret NGS sequencing data. However, most of these tools are specified in one task, thus completing the sequence analysis pipeline will require a number of programs, which is time consuming and inconvenient. To overcome this limitation, Schloss et al. (2009) developed Mothur, a single piece of software which combines previous tools (e.g. NAST, SINA, DNADIST, and UniFrac) and can perform various sequence analysis tasks on multiple operating systems (e.g. Windows,

Unix/Linux, and Mac OS). Another widely used software package for sequence analysis is ‘quantitative insights into microbial ecology’ (QIIME) developed by Caporaso et al.

(2010). It supports a wide range of sequence analysis and also visualization of data, but currently it only runs on Unix/Linux and Mac OS. Both Mothur and QIIME, however, are resource-demanding. The performance of these programs is directly affected by the number and speed of CPUs, as well as the size of memory. As NGS technologies are developing fast, more and longer reads can be generated during a single run. For example,

Illumina MiSeq platform is able to generate approximately 25 million 2×300 paired-end reads each run. Using current version of Mothur or QIIME to analyze sequence data generated from such ultra-high-throughput platforms is a daunting a task. Clearly, new tools or improvement in current tools is desperately needed.

43

2.7 Summary

Gut microbiome is now recognized as an essential component of the intestinal ecosystem and is considered as a forgotten organ, which contributes to the wellbeing of animal host in a wide range of aspects, especially nutrition and disease resistance (O’Hara and

Shanahan, 2006). Manipulations of gut microbiome through dietary and managerial interventions have been used by poultry producers to enhance bird growth and reduce the incidence of disease. Undeniably, however, these alternatives to AGPs cannot completely replace AGPs in poultry diets as AGPs are still the most effective and cost-efficient strategy to promote bird growth and control disease. Thanks to the fast developing high- throughput technologies such as NGS and phylogenetic microarray, we now have the unprecedented opportunity to gain a comprehensive knowledge of poultry gut microbiome and its interaction with host and diet, which can provide the knowledge base needed to develop more effective alternative strategies that may eventually completely replace AGPs in modern poultry production.

44

Figure 2.1: Conceptual model of the interactions among gut microbiome, avian host, diet, and litter microbiome.

45

Chapter 3: Comparative analysis of gut microbiome between broilers with low and high

feed conversion ratios

3.1 Abstract

In this study we investigated the relationship between gut microbiome and host feed utilization efficiency using PITChip2, a poultry specific phylogenetic microarray. Ninety- six male Cobb 500 broiler chickens were used in this study and they were ranked from 1 to 96 based on their feed conversion ratio (FCR). On day 25 post-hatch the birds were euthanized. Metagenomic DNA extracted from cecal content collected from 24 birds on each end of the FCR spectrum was used for microbial analysis using PITChip2. There was no clear separation in the gut microbiome structure between high-performance (low

FCR) and low-performance (high FCR) chickens as indicated by the hierarchical clustering (HCL) analysis or principal component analysis (PCA). Sixty-nine phylotypes of bacteria were found to exist universally in all 48 birds, with 16 phylotypes were identified as predominant, including phylotypes in Ruminococcaceae, Faecalibacterium,

Oscillibacter, Bacteroidetes, and Lactobacillales. One hundred phylotypes were found to correlate with bird performance (p≤0.05), with 48 of them showing a greater than 3-fold difference in relative abundance between birds with high and low FCR. Three phylotypes representing species of Bacteroides, Peptococcus, and Ruminococcaceae, and two unclassified phylotypes were found to be more abundant in high-performance than in 46

low-performance chickens. These five phylotypes may serve as bacterial indicators of feed utilization efficiency and potential probiotic candidates or targets of prebiotics for improvement of growth performance of broilers.

3.2 Introduction

The gastrointestinal (GI) tract of poultry is populated with microorganisms, mostly bacteria, which closely and intensively interact with the host and ingested feed (Pan and

Yu, 2013). Most readily digestible dietary carbohydrates are digested by host-secreted digestive enzymes and absorbed by the host in the proximal gut, leaving indigestible carbohydrates and residual digestible carbohydrates to bacteria residing the distal gut

(Hooper et al., 2002). Indigestible dietary carbohydrates can be hydrolyzed and then fermented by intestinal bacteria, yielding short chain fatty acids (SCFAs), which can be utilized by the host as energy and carbon source (Van der Wielen et al., 2000; Hooper et al., 2002; Koutsos and Arias, 2006; Tellez et al., 2006). Such fermentation can be observed in most part of the avian gut (from crop to cecum) but primarily takes place in the cecum, which is densely populated with bacteria (Rehman et al., 2007). In addition to the carbohydrate metabolism, it has been reported that intestinal bacteria also contribute to host nitrogen metabolism (Pan and Yu, 2013).

In modern broiler production industries, feed represents the major portion of production cost. Efficiency in converting feed into body mass is thus of critical concern for broiler producers. Because intestinal bacteria can provide nutrients to the host from otherwise 47

poorly utilized dietary substrates, attentions have been drawn to the associations between intestinal bacteria and host feed utilization efficiency. By using microbial profiling on broiler chickens across various feeding trails, Torok et al. (2011) were able to identify groups of bacteria that are potentially associated with broiler growth performance.

However, since the GI tract of chickens harbors a complex microbiome (more than 900 species-equivalent operational taxonomic units [OTUs, defined at 0.03 phylogenetic distance] as reported by Wei et al. [2013]), a comprehensive assessment of the associations between intestinal bacteria and host feed utilization efficiency requires high- throughput techniques which can, rather than focusing on specific members of intestinal bacteria, take a broader view on the total microbiome in the GI tract of chicken.

Phylogenetic microarray is a molecular biology technique that enables simultaneous detection and semi-quantification of individual members of microbial ecosystems. A single microarray chip can have hundreds of thousand probes, which can provide semi- quantitative information on the population of complex microbial ecosystem (DeSantis et al., 2007). To date, phylogenetic microarrays have been developed for the assessment of human GI microbiome as well as complex microbiomes in other environments, and have gained recognition and acceptance as a highly efficient phylogenetic fingerprinting tool

(DeSantis et al., 2007; Rajilić-Stojanović et al., 2009; Kang et al., 2010). To our knowledge, however, very little effort has been made in designing microarrays for poultry gut microbiome. Our lab has developed two versions of phylogenetic microarrays for poultry, namely Poultry Intestinal Tract Chip (PITChip) 1 and 2 (Wei, 2013). The objectives of this study was to investigate possible associations between feed utilization 48

efficiency and cecal microbiome in broiler chickens with high or low FCR, and to identify bacteria that are indicative of feed utilization efficiency in broiler chickens using

PITChip2, which was designed based on the V3 region of bacterial 16S rRNA gene sequences of poultry origin and contains 1,962 OTU-specific probes, 105 genus-specific probes, and 83 probes targeting pathogens.

3.3 Experimental Procedures

3.3.1 Sample collection and DNA extraction

Metagenomic DNA samples collected from broiler chickens have been used in a previous study (Stanley et al. 2013). Briefly, male Cobb 500 broiler chickens were raised in a rearing pen in a temperature-controlled room. Broiler grower diet was formulated to meet or exceed National Research Council guidelines for broiler chickens (NRC, 1994) and was fed to birds ad libitum. At 13 days of age, 96 birds were transferred in pairs to 48 metabolism cages followed by transferring to individual metabolism cages at 15 days of age. Body weight and feed intake were recorded at 18 and 25 days of age. On day 25 post-hatch the birds were euthanized and fresh cecal content from each bird was collected.

Metagenomic DNA was extracted from the cecal content samples using the repeated bead beating plus column (RBB+C) method developed by Yu and Morrison (2004).

The feed conversion ratio (FCR; g feed eaten/g weight gain) over the 7-day experimental period was calculated for each bird using the body weight and feed intake data recorded

49

at 18 and 25 days of age. Chickens were ranked from 1 to 96 based on their FCR.

Metagenomic DNA samples from 24 birds each at the high and low ends of the FCR spectrum were used for microbial analysis using PITChip2.

3.3.2 Sample preparation and labeling

From each metagenomic DNA sample 16S rRNA gene was amplified by PCR using a universal bacterial primer set T7/27F (5’-

TCTAATACGACTCACTATAGGGAGAGTTTGATCMTGGCTCAG-3’, T7 promoter region is the underlined region) and 1525R (5’-AAGGAGGTGWTCCARCC-3’) as described previously (Devereux and Willis, 1995; Kang et al., 2010). PCR was performed with 33 cycles (denaturation, 95℃ for 30s; annealing, 55℃ for 45s; and extension, 72℃ for 90s) using a PTC-100 thermocycler (MJ Research, Waltham, MA). Amplicons were confirmed by agarose gel (1%) electrophoresis, purified using a QIAquick PCR purification kit (QIAGEN, Hilden, Germany) and quantified using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE). The same amount of each purified PCR product (100ng) was then used as template for in vitro transcription using the MEGAScript T7 in vitro transcription kit (Ambion, Austin, TX). Synthesized cRNA was again confirmed by agarose gel (1%) electrophoresis, purified using the MEGAClear kit (Ambion, Foster City, CA) and quantified using a NanoDrop ND-1000 spectrophotometer. Equal amount (500ng) of the purified cRNA was then labeled with

Cy5 fluorescent dye at 37℃ for 1 hour using the Label IT µArray Cy5 reagent (Mirus,

Madison, WI). The labeled cRNA was purified to remove the free Cy5 dye using the

50

MEGAClear kit and stored at -80℃ until microarray hybridization. All subsequent microarray experiment procedures (e.g. hybridization and washing) after the Cy5 labeling step were performed in dark to avoid light bleaching of the light-sensitive Cy5 dye.

3.3.3 Microarray hybridization and washing of microarray slides

The hybridization was performed following manufacturer's protocol provided by

MYcroarray. Briefly, for each microarray 60 μL hybridization solution was prepared with

18 μL 20×SSPE buffer (3M NaCl, 20mM EDTA, 118.2mM NaH2PO4, and 81.8mM

Na2HPO4), 6 μL formamide, 0.6μL 1% bovine serum albumin, 0.6μL 1% tween-20, 0.6

μL Control-oligos provided by the manufacturer, 1 μL Cy5-labeled internal controls, and

33.2 μL Cy5-labeled cRNA. The hybridization solution was incubated at 65℃ for 5 minutes, then immediately cooled on ice for another 5 minutes. Each microarray was applied with 54 μL hybridization solution. Microarray slide was placed in an Agilent

Hybridization Chamber (Agilent, Santa Clara, CA) and the hybridization was carried out in an HB-1000 hybridization oven (UVP, Upland, CA) at 45℃ for 20 hours with rotation speed set at 10 rpm.

The washing of the microarray slide following hybridization was performed per manufacturer’s protocol. Right before hybridization, 1×SSPE buffer and 0.25×SSPE buffer were placed in the hybridization oven along with the microarray slide so that they remained at hybridization temperature need in the subsequent washing steps. The microarray slide was first washed in 1×SSPE buffer for 5 minutes, followed by washing

51

in 0.25×SSPE buffer for 30 seconds. The microarray slide was then dried for 10 min using a high speed microarray centrifuge (Arrayit, Sunnyvale, CA).

3.3.4 Signal detection and data analysis

Microarray slides were scanned using a GenePix 4000B microarray scanner (Molecular

Devices, Sunnyvale, CA) with oversaturation tolerance set to 0.01%, photomultiplier sensitivity set at auto-PMT gain, and resolution set to 5 µm. The information of each probe spot on the scanned images was extracted by fitting the gene allocation list (GAL) file provided by the manufacturer which carries the annotation information of each spot on the microarray. The scanned image was then quantified using the GenePix pro 6.0 software (Axon Instruments, Union City, CA). Probe spots of negative detection were flagged out by the auto-alignment function of the GenePix software. Probe spots with oversaturation, bad shape, or suspected contamination were flagged manually.

The GenePix extracted results of each array were exported as a GenePix Results Format

(GPR) file. ExpressConverter (version 2.1) of the TM4 Microarray Suite (Saeed et al.,

2006) was used to convert each GPR file to an annotation (ANN) file and a multiexperiment viewer (MEV) file with median signal intensity of each spot after subtracting the background kept for downstream analysis. The converted data was inputted to Ginkgo (version 1.01, J. Craig Venter Institute, Rockville, MD) and the replicated spots in the array were consolidated using the standard “In-Slide Replicates

Analysis” option of the Ginkgo software.

52

Normalization of microarray data was performed based on the signal intensities of internal control probes. Normalized data were imported into MeV program (version 4.8.1) within the TM4 Microarray Suite for statistical analysis. Heat maps were generated based on the log2 value of the median intensities. Clustering of samples was performed using hierarchical clustering (HCL) in the MeV program with Manhattan distance metric.

Average linkage was used to determine cluster-to-cluster distances when constructing the hierarchical tree. Principal component analysis (PCA) was conducted using the MeV program to identify clusters of samples with similar cecal microbiome structure.

Clustering Affinity Search Technique (CAST) in MeV was used to cluster the hybridized probes according to the pattern similarity. The standard “TTEST” option of the MeV program was used to identify the hybridized probes that were significantly (p≤0.05) correlated with bird performance (i.e. high and low FCR).

3.4 Results and Discussion

3.4.1 Chickens with high and low FCR have similar cecal microbiome structure

The heat map and hierarchical tree generated from the HCL analysis (Fig. 3.1) showed that the 48 samples were not grouped based on the FCR grouping (low or high), suggesting that the overall cecal microbiome structure of good performance (i.e. low FCR) chickens was similar with that of poor performance (i.e. high FCR) chickens. This finding was further supported by the result of the PCA analysis. As principal component

1 and 2 only accounted for 10.73% and 4.85% of total variance, respectively, the samples 53

on the PCA plot were randomly scattered and showed no clear separation between good- and poor-performance birds, again indicating that there was no significant difference in the cecal microbiome structure between good- and poor-performance birds (Fig. 3.2).

Considering all chickens were raised together in a rearing pen until 13 days of age, they may establish compositionally similar gut microbiome by picking up microorganisms from the surrounding environment. Our results are in agreement with study conducted by

Stanley et al. (2012) who found that, based on the PCA analysis of chicken cecal microbiome diversity, there was no strong separation between chickens with high and low FCR. Interestingly, however, in a follow up study conducted by the same group in

2013, a full separation between high and low FCR birds was observed on the PCA plot

(Stanley et al., 2013). It is worth noting that in the study conducted at 2012, Stanley et al. picked the top and bottom 24 birds on the FCR spectrum while in the later study conducted in 2013, 12 birds at each end of the FCR spectrum was selected. Such different sampling strategies may partially explain the inconsistent PCA analysis results of those two studies. In our study, we analyzed samples from 24 high-FCR birds and 24 low-FCR birds, therefore the divergence between our results and those of Stanley et al.’s (2013) may also be explained by the different sampling strategies, even though both studies use samples from the same feeding trial. It should also be noted that although both studies are based on the analysis of bacterial 16S rRNA gene sequences, Stanley et al. (2013) used

454 pyrosequencing to study microbiome diversity while in this study we used microarray (i.e. PITChip2). Microarray probes are designed from already known sequences and are thus not able to detect sequences that do not have corresponding

54

probes. If novel bacterial species and/or known bacterial species which cannot be detected by PITChip2 were contributing to the difference in cecal microbiome structure between high FCR birds and low FCR birds, such contribution will be missed in a microarray study but will likely be captured in a 454 pyrosequencing study. Moreover,

PITChip2 probes were designed based on the V3 region of bacterial 16S rRNA gene. In contrast, V1-V3 region was sequenced in Stanley et al.’s study (Stanley et al., 2013). The different hypervariable regions used may affect OUT clustering and thus affect assessment of microbiome diversity. On the other hand, analysis of microbiome using

454 pyrosequencing can have reproducibility issue (Kim and Yu, 2014). Putting all these together, PITChip2 and 454 pyrosequencing may not produce the same results and direct comparison between PITChip2 and 454 pyrosequencing results should be done with caution.

3.4.2 A universal core microbiome was identified for all 48 chickens

The HCL and PCA analysis did not show clear separation between high and low FCR chickens, but they did show that each individual chicken in our study harbors a unique gut microbiome as no identical pattern in HCL heat map (Fig. 3.1) and sample overlaps in PCA plot (Fig. 3.2) was observed. Indeed, gut microbiome can be affected by so many factors including stochastic ones. Although the chickens used in this study had similar genetics and were raised together and fed the same diet, particular bacteria species present in one particular chicken may not necessarily be found in other chickens.

However, a group of bacteria, which are located at the top rows on the HCL heat map

55

(Fig. 3.1), appears to be present in all the birds. To further identify this group of bacteria,

CAST analysis was performed using the Mev program, and 69 phylotypes were found in all 48 the birds (Fig. 3.3). Among these 69 phylogypes, 16 were identified as predominant, including phylotypes in Ruminococcaceae, Faecalibacterium, Oscillibacter,

Bacteroidetes, and Lactobacillales (Fig. 3.4). This group of phylotypes is known as universal core microbiome due to its universal distribution among hosts (Rajilić-

Stojanović et al., 2009). The universal distribution of these phylotypes indicated that they are ubiquitous in the GI environment. In addition, this group of bacteria, especially the 16 predominant phylotypes, is likely to benefit from the interactions with host and diet which in turn grant it the ability to colonize and thrive in the gut of chickens. Since gut microbiome has evolved with the host towards a symbiotic relationship, these phylotypes may also be functionally important to the chicken host, playing key roles such as providing the host with nutrient extracted from the indigestible fraction of the feed and protecting host from pathogen invasion (Pan and Yu, 2013). It is worth noting that, however, this group of phylotypes may not be universally distributed in all chickens. The composition of gut microbiome can be influenced by factors such as locations, diets, and breeds of chicken. Even different investigational tools (e.g. 454 pyrosequencing vs. microarray) may reveal different universal core microbiome.

3.4.3 Identification of phylotypes associated with FCR

Although in a microbiome-based perspective no clear difference was observed between the high- and the low-FCR birds, we were still able to identify individual phylotypes that

56

correlated with bird performance. By using t-test in Mev, 100 phylotypes were found to correlate with bird performance (p≤0.05; Fig. 3.5), with 48 phylotypes showed greater than 3-fold differences in relative abundance between good- and poor-performance birds

(Fig. 3.6). Among these 48 phylotypes, five phylotypes were found to be more abundant in good-performance low FCR chickens. Two of them could be classified to genus level

(i.e. unclassified Bacteroides and unclassified Peptococcus), one of them was classified to family level (i.e. unclassified Ruminococcaceae), and the other two could only be classified as Bacteria. Such findings are supported by Stanley et al. (2012) who reported that in cecal microbiome the abundance of four OTUs, with three belonging to the order of Clostridiales and the fourth belonging to the genus of Bacteroides, is negatively correlated with the FCR of broiler chickens.

Bacteroides was reported to be one of the major genera in the cecum of chicken (Lu et al.,

2003; Gong et al., 2007). While some Bacteroides spp. are pathogenic and some may have negative impact on host performance, generally this genus has a mutualism relationship with host (Xu and Gordon, 2003; Wexler, 2007; Torok et al., 2010). With numerous genes encoding starch binding proteins and starch hydrolysis enzymes (i.e. glycosyl hydrolases), Bacteroides spp. have a powerful starch utilization system, which grants them the ability to utilize a wide range of carbohydrates available in the gut. In fact, one species in this genus, namely Bacteroides thetaiotaomicron, has more glycosyl hydrolases than any other prokaryotes that has been sequenced (Xu et al., 2004). SCFAs produced from fermentation of carbohydrate by Bacteroides can be absorbed by the host by passive diffusion through the cecal epithelium, enabling the host to extract energy 57

from the indigestible fraction of the diet, resulting in increased feed utilization efficiency and thus lower FCR of the birds (Hooper et al, 2002; Wexler, 2007). In addition, by breaking down polysaccharides and releasing simple sugars, Bacteroides spp. may benefit the growth and proliferation of commensal and beneficial bacteria that cannot efficiently utilize polysaccharides, which may in turn improve health and increase the performance of chicken (Sonnenburg et al., 2004).

Previous studies suggested that Peptococcus is one of the predominant genera found in human large intestine and might be important in the functioning of the host (Salminen et al., 1998; Kolida et al., 2000; Isolauri et al., 2002). This genus is also reported to be found in the chicken cecal mucosa but only represent a small fraction of mucosal microbiome in the ceca (Gong et al., 2007). Although Peptococcus is commonly found in gut microbiome, its role on host health and nutrition has not been extensively studied. To our knowledge, we are the first to report that there is association between the abundance of Peptococcus and performance of chicken host. However, there is little evidence from previous studies to support a cause-and-effect relationship between the abundance of this genus and chicken performance. Early studies have revealed limited information on the metabolism characteristics of this genus, but clear connection between those metabolism characteristics and host performance could not be made (Rogosa, 1971). In fact, it is reported that the metabolism traits of Peptococcus vary among different species or even strains in this genus, making it harder to identify potential traits that may affect host performance (Ezaki et al., 1983). However, since the phylotype we detected here can only be classified to genus level, it might represent a novel species or species has not 58

been extensively studied. Further studies are needed to ascertain whether this

Peptococcus sp. can indeed affect or be affected by host feed utilization efficiency.

Members of the genus Ruminococcus are usually found in the rumen and similar habitats, playing important roles as major resistant polysaccharides (e.g. cellulose) degraders

(Krause et al., 1999). Some species of this genus are fibrolytic bacteria and their superb capability to degrade cellulose and xylan is achieved by their ability to adhere to and hydrolyze cellulose (Morrison and Miron, 2000; Ohara et al., 2000). As one of the major genera found in the gut of chicken and turkey, Ruminococcus spp. may help degrade the otherwise indigestible cellulose and hemicellulose present in the diet and thus increase the feed utilization efficiency of the birds. Supporting this view, Stanley et al. (2013) reported that four OTUs, with closest culturable isolate being Ruminococcus sp. str.

16442, were positively correlated with the apparent metabolizable energy (AME), which is another commonly used parameter to measure performance, of broiler chickens. While a chicken with high AME does not necessary mean that it will have a low FCR as not all energy extracted from diet will be used for body growth, their study did suggest that

Ruminococcus spp. can provide extra energy to the host which may result in a low FCR value of the host.

For the two phylotypes classified only as Bacteria, little can be said about their metabolism characteristics and thus the relationship between these two phylotypes and the FCR of host remains a mystery. However, the public sequence databases are

59

expanding and annotation of sequences is updated frequently. The role or importance of these two phylotypes in chicken feed utilization may be unraveled in the future.

3.5 Conclusion

In the current study we have identified five phylotypes that were found more abundant in the ceca of birds with higher feed utilization efficiency as indicated by low FCR. While association does not mean causality, the current study has suggested the direction of future researches. If any of these identified phylotypes are repeatedly shown to increase with efficient feed utilization in chickens, intervention strategies (e.g. feeding probiotics or prebiotics) can then be designed to increase the abundance of such phylotypes, which may in turn benefit the growth of the birds. Such intervention strategies may serve as alternatives for currently widely used in-feed antibiotic growth promoters (AGP) and help control the spreading of antibiotic resistance among bacteria residing in food animals.

While next-generation sequencing is becoming more affordable and being widely used in microbiome studies, the current study showed that our phylogenetic microarray,

PITChip2, can serve as an alternative tool that enables high-throughput and comprehensive analysis of the intestinal microbiome in chicken’s GI tract. However, probes on PITChip2 were designed from the V3 region of bacterial 16S rRNA gene sequence, limiting its ability to reliably detect and identify bacteria present in chicken’s gut. Phylogenetic microarray with probes designed from longer region of bacterial 16S 60

rRNA gene sequences, such as the V1-V3 region, is likely to generate more accurate and reliable results, which are critically needed in future studies to investigate microbial interactions taking place in the GI tract of chicken.

61

Figure 3.1: A hierarchical tree generated from HCL analysis. Heat map was generated based on the log2 value of the ratio between the mean fluorescence intensities of each group of birds and the internal standard, and displayed using the standard Green-Black- Red scheme. Good, low FCR good performance chicken; Poor, high FCR poor performance chicken.

62

6

3

Figure 3.2: PCA plot showing the grouping of 48 chickens based on the structure of their gut microbiome.

a 63

Figure 3.3: Universal core microbiome of the 48 chickens. Heat map was generated based on the log2 value of the ratio between the mean fluorescence intensities of each group of birds and the internal standard, and displayed using the standard Green-Black- Red scheme. Good, low FCR good performance chicken; Poor, high FCR poor performance chicken.

64

Figure 3.4: Sixteen predominant phylotypes of the universal core microbiome. Heat map was generated based on the log2 value of the ratio between the mean fluorescence intensities of each group of birds and the internal standard, and displayed using the standard Green-Black-Red scheme. Good, low FCR good performance chicken; Poor, high FCR poor performance chicken.

65

Figure 3.5: One hundred phylotypes differing in abundance (p≤0.05) between birds with low or high FCR. Heat map was generated based on the log2 value of the ratio between the mean fluorescence intensities of each group of birds and the internal standard, and displayed using the standard Green-Black-Red scheme. Good, low FCR good performance chicken; Poor, high FCR poor performance chicken.

66

Figure 3.6: Phylotypes showing greater than 3-fold differences in mean relative abundances between good- and poor-performance birds. Solid and open bars indicate more abundance in poor- and good-performance chickens, respectively.

67

Chapter 4: Development of PITChip3 and comparative analysis of gut microbiome of

broilers fed corn- or wheat-based diet

4.1 Abstract

In this chapter we presented the development of PITChip3, a poultry specific phylogenetic microarray, which is intended to be used as a comparative analytical tool to comprehensively study bacterial community in poultry GI tract. Designed based on the

V1-V3 region of bacterial 16S rRNA gene sequences of poultry origin and with up to three probes targeting the same target, the probes on PITChip3 are able to simultaneously detect 62 genus-level phylotypes and 662 species-level phylotypes, as well as 34 pathogens. Each PITChip3 slide has six microarrays and every probe has 4 replicates randomly located on each microarray. To test its utility, PITChip3 was used to study the gut microbiome of broilers fed either corn-based or wheat-based diet. PCA analysis showed clear separation between chickens fed with the two different diets and also between chickens at two different ages. Twenty phylotypes (either species-level or genus- level) were further identified with different abundance in chickens fed different diets.

Akkermansia muciniphila was found inversely associated with bodyweight, while

Bacteroides was more abundant in chickens fed with wheat-based diet irrespective of diet.

Escherichia/Shigella/Salmonella was also more abundant in chickens fed with wheat- based diet at 14 days of age. Such difference in gut microbiome revealed by PITChip3 68

may help to better understand the interactions between gut microbiome and diet and the risk associated with the two common diets.

4.2 Introduction

The gastrointestinal (GI) tract of chickens and turkeys, especially cecum, harbors a complex microbiome, which is almost exclusively composed of bacteria (Wei et al.,

2013). The microbiome in the GI tract functions as an interface between the host and the feed ingested and contributes to the wellbeing of animal host in a range of aspects, especially nutrition and disease resistance (O’Hara et al., 2006; O’Hara and Shanahan,

2006), and it is thus of great interest to researchers. A comprehensive understanding of poultry gut microbiome will help develop new dietary or managerial interventions that can enhance bird growth, maximize host feed utilization, and protect birds from enteric diseases caused by pathogenic bacteria.

Early studies on poultry gut microbiome had been primarily dependent on cultivation- based methods (Barnes et al., 1972; Salanitro et al., 1974; BEDBURY and Duke, 1983).

However, these cultivation-based studies revealed only a very small part of the poultry gut microbiome, as most intestinal bacteria cannot be cultured using laboratory media.

Nevertheless, these early studies laid the foundation for microbiological studies of the intestinal microbiome in poultry. In the early 2000’s, cultivation-independent 16S rRNA gene-targeted techniques started to be used in investigating poultry gut microbiome, and these techniques have greatly helped expand our knowledge on the bacterial diversity 69

present in the intestinal tract of poultry ever since (Gong et al., 2002a; Gong et al., 2002b;

Zhu et al., 2002). PCR-denaturing gradient gel Eeectrophoresis (DGGE) and cloning and sequencing library are the most commonly used 16S rRNA gene-based molecular techniques used in studying poultry intestinal microbiome (Gong et al., 2007; Zhou et al.,

2007; Scupham et al., 2008; Cressman et al., 2010). These techniques, however, can only provide limited information on poultry gut microbiome as only a small number of bacteria can be practically analyzed (Wei, 2013). Comprehensive characterization of poultry gut microbiome requires more powerful techniques or technologies.

Phylogenetic microarray is such a molecular technique because a single microarray chip can have hundreds of thousand probes, which can provide semi-quantitative information on members of microbial ecosystems with great diversity (DeSantis et al., 2007).

Compared to other high-throughput techniques such as 454 DNA pyrosequencing, microarray has been shown to be advantageous for several reasons in comparative analysis of complex microbiomes. First, microarray can provide a uniformed, robust assessment of individual populations of complex microbiomes (Brodie et al, 2006).

Second, nucleic acid, preferably, samples can be directly hybridized with the microarray chip, avoiding inherent PCR bias, which has been shown to compromise the assessment of microbiomes. Last but not least, microarray chips can be custom fabricated or printed in house at relatively low cost (less than $250 per microarray chip), updated quickly when new probes need to be added, and analysis of microarray data is much less resource demanding.

70

To date, phylogenetic microarrays have been developed for the assessment of human GI microbiome as well as complex microbiomes in other environments, and have gained recognition and acceptance as a highly efficient phylogenetic fingerprinting tool

(DeSantis et al., 2007; Rajilić-Stojanović et al., 2009; Kang et al., 2010). To our knowledge, however, very little effort has been made in designing microarrays for poultry GI microbiome. Our lab has developed two versions of a phylogenetic microarray specifically for poultry, namely Poultry Intestinal Tract Chip (PITChip) 1 and 2 (Wei,

2013). The objective of this study was to further develop the PITChip and improve its utility in analyzing poultry gut microbiome. The utility of the new version, PITChip3, was tested using metagenomic DNA recovered from cecal content of broiler chickens fed with either wheat-based or corn-based diet.

4.3 Experimental Procedures

4.3.1 Oligonucleotide probe design

Three sources of bacterial 16S rRNA gene sequences were used to construct a sequence dataset for the design of oligonucleotide probes for PITChip3: 1) 3,922 bacterial 16S rRNA gene sequences of poultry origin (chicken and turkey) retrieved from the GenBank and the RDP database (Release 10, Update 32) using search terms such as ‘poultry’,

‘chicken(s)’, ‘chick(s)’, ‘broiler(s)’, and ‘turkey’; 2) 4,095 bacterial 16S rRNA gene sequences (V1-V3 region) from chicken gut microbiome kindly provided by our collaborator at CSIRO Livestock Industries, Australia; 3) bacterial 16S rRNA gene 71

sequences (V1-V3 region) from broiler chickens generated using 454 FLX Titanium pyrosequencing (sequences deposited at the MG-RAST server [Meyer et al., 2008] with

ID:9131 and 9132) from another two projects in our lab (unpublished). The sequences from these sources were combined into a single dataset and subjected to quality screening to eliminate those sequences of low quality. The Mothur program (Schloss et al., 2009) was used to dereplicate identical sequences, filter out sequences with ambiguous bases, and remove sequences shorter than 400 bp. Chimeric sequences were detected and removed using UCHIME (Edgar et al., 2011) and DECIPHER (Wright et al., 2012). The remaining sequences were then aligned against the SILVA alignment reference database that was incorporated into Mothur. Based on the alignment, only sequences that contain the V1-V3 hypervariable region were retained, and the V1-V3 region were delineated and sliced out for subsequent probe design.

Oligonucleotide probes at two taxonomy levels (genus- and species-equivalent) were designed for PITChip3. In order to design probes at genus level, sequences that passed the above mentioned quality check procedures were first classified using RDP Classifier

(Wang et al., 2007). Sequences that cannot be classified at genus level (with confidence threshold set to 70%) were not used for genus-level probe design. Genera with only one sequence were also excluded from subsequent design of genus-level probes. Sequences used for genus-level probe design were aligned against the SILVA alignment reference database using Mothur. The alignment was imported to Geneious Basic version 5.6.6

(Kearse et al., 2012) and one consensus sequence for each genus was generated. The probe design program Picky version 2.20 (Chou et al., 2004) was used to design probes 72

from these consensus sequences. Sequences not used for genus-level probe design were imported into Picky as nontarget so that the probes designed will not hybridize to these sequences. Picky was set to design oligonucleotide probes with sizes of 20-30 nt and with the minimum temperature difference (i.e. minimum safe difference between a probe's target melting temperature and its highest nontarget melting temperature) set to 7℃. In addition, Picky was instructed to design three oligonucleotide probes targeting different regions of each target sequence, and only unique probes (probes that can only hybridize to one target) were selected. For genera that Picky failed to design unique probes from consensus sequences, a distance matrix with Jukes-Cantor correction was calculated using ARB (Ludwig et al., 2004). The distance matrix was then used to generate a representative sequence for each genus using Mothur. Representative sequences from these genera were then imported into Picky to design unique oligonucleotide probes.

Species-level probes were designed following similar procedures as for genus-level probe design. Briefly, ARB was used to produce a distance matrix, which was then used to cluster sequences into species-equivalent operational taxonomic units (OTUs, defined at

0.04 phylogenetic distance). Unique oligonucleotide probes were first designed based on the consensus sequences of each OTU. For OTUs that had no unique probes designed from consensus sequences, additional probes were designed based on the representative sequences. The parameters in Picky used for species-level probe design were identical to those used for genus-level probe design.

73

All the pathogen probes on PITChip2 were also included on PITChip3. Designed from the 16S rRNA gene or genes encoding virulence factors using Picky, these probes target poultry pathogens, including pathogenic bacteria, virus and parasite, which have their

16S rRNA or virulence genes archived in GenBank.

In addition to the aforementioned probes, PITChip3 has positive and negative control probes and also internal control probes, the latter of which was used for microarray data normalization. The probe EUB338 (5’-GCTGCCTCCCGTAGGAGT-3’) which targets most bacteria was used as positive control probe on PITChip3 (Banerjee et al., 2002).

Negative control probe was designed from human mitochondrial 16S rRNA gene sequence using Picky. In order to design internal control probes, 4 randomly generated

DNA sequences (1,600 nt) were imported into Picky as targets, while the sequences used for genus- and species-level probe design were imported into Picky as nontarget, thus assuring the designed internal control probes will not hybridize to the targets of genus- and species-level probes. Picky designed several oligonucleotide probes from the 4 randomly generated DNA sequences, from which 6 probes were used on PITChip3.

These 6 internal control probes were further tested using the probe match tool in RDP to ensure that they do not hybridize to any bacterial 16S rRNA gene sequences archived in

RDP.

4.3.2 Fabrication of PITChip3

The PITChip3 slides were custom-fabricated by MYcroarray (Ann Arbor, MI). There are

6 arrays on each slide with each array consists of a grid of 56 columns by 94 rows for a

74

total of 5,264 spots, among which 264 spots are reserved for internal quality control, leaving 5,000 spots available for the customized oligonucleotide probes. Each probe has

4 replicates occupying 4 spots randomly located on each microarray and each spot has approximately 1.5×109 probes.

4.3.3 Sample collection and DNA extraction

Intestinal microbial samples from another study (unpublished) in our lab were used to test

PITChip3. Briefly, 200 straight run Cobb 500 broiler chicks at age of one day were randomly assigned to two different diets (wheat-based diets vs. corn-based diets).

Chickens were housed in 40 floor pens (n=5 birds per pen) with 20 replicate pens per dietary treatment. One bird per pen was randomly picked for sampling at 14, 20, and 35 days of age. Ileal mucosa and cecal content samples were collected from the birds selected and frozen at -80℃ until DNA extraction. The birds were handled and cared of following a protocol approved by the Institutional Animal Care and Use Committee.

Cecal content samples collected from 4 birds per treatment at 14 and 35 days of age were used for microarray experiments. Metagenomic DNA was extracted from each sample using the repeated bead beating plus column (RBB+C) method developed by Yu and

Morrison (2004). Extracted DNA was stored at -20℃ until use.

4.3.4 Sample preparation and labeling

From each metagenomic DNA sample 16S rRNA gene was amplified by PCR using universal bacterial primer set T7/27F (5’-

75

TCTAATACGACTCACTATAGGGAGAGTTTGATCMTGGCTCAG-3’, T7 promoter region is the underlined region) and 1525R (5’-AAGGAGGTGWTCCARCC-3’) as described previously (Devereux and Willis, 1995; Kang et al., 2010). PCR was performed with 33 cycles (denaturation, 95℃ for 30 s; annealing, 55℃ for 45 s; and extension, 72℃ for 90 s) using a PTC-100 thermocycler (MJ Research, Waltham, MA). Amplicons were confirmed by agarose gel electrophoresis, purified using QIAquick PCR purification kit

(QIAGEN, Hilden, Germany) and quantified using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE). The same amount of each purified PCR product (100 ng) was then added as template for in vitro transcription using the MEGAScript T7 in vitro transcription kit (Ambion, Austin, TX). Synthesized cRNA was again confirmed by agarose gel electrophoresis, purified using the MEGAClear kit

(Ambion, Foster City, CA) and quantified using a NanoDrop ND-1000 spectrophotometer. Equal amount (500 ng) of the purified cRNA was then labeled with

Cy5 fluorescent dye at 37℃ for 1 hour using the Label IT µArray Cy5 reagent (Mirus,

Madison, WI). The labeled cRNA was purified to remove the free Cy5 dye using the

MEGAClear kit and stored at -80℃ until microarray hybridization. All subsequent microarray experiment procedures (e.g. hybridization and washing) after the Cy5 labeling step were performed in dark to avoid light bleaching of the light-sensitive Cy5 dye.

4.3.5 Microarray hybridization and washing of microarray slides

The hybridization was performed following manufacturer's protocol provided by

MYcroarray. Briefly, for each array 60 μL hybridization solution was prepared with 18

76

μL 20×SSPE buffer (3M NaCl, 20mM EDTA, 118.2mM NaH2PO4, and 81.8mM

Na2HPO4), 6 μL formamide, 0.6 μL 1% bovine serum albumin, 0.6 μL 1% tween-20, 0.6

μL Control-oligos provided by the manufacturer, 0.6 μL Cy5-labeled internal controls, and 26 μL Cy5-labeled cRNA. The hybridization solution was incubated at 65℃ for 5 minutes, then immediately cooled on ice for another 5 minutes. Each array was applied with 54 μL hybridization solution. Microarray slide was placed in Agilent Hybridization

Chamber (Agilent, Santa Clara, CA) and the hybridization was carried out in an HB-1000 hybridization oven (UVP, Upland, CA) at predetermined optimal hybridization temperature for 20 h with rotation speed set at 10 rpm.

The washing of microarray following hybridization was performed per manufacturer’s protocol. Right before hybridization, 1×SSPE buffer and 0.25×SSPE buffer were placed in the hybridization oven along with the microarray slide so that they remained at hybridization temperature in the subsequent washing steps. The microarray slide was first washed in 1×SSPE buffer for 5 minutes, followed by washing in 0.25×SSPE buffer for

30 seconds. The microarray slide was then dried for 10 min using a high speed microarray centrifuge (Arrayit, Sunnyvale, CA).

4.3.6 Signal detection and data analysis

Microarray slides were scanned using a GenePix 4000B microarray scanner (Molecular

Devices, Sunnyvale, CA) with oversaturation tolerance set to 0.01%, photomultiplier sensitivity set at auto-PMT gain, and resolution set to 5 µm. The information of each probe spot on the scanned images was extracted by fitting the gene allocation list (GAL)

77

file provided by the manufacturer which carries the annotation information of each spot on the microarray. The scanned image was then quantified using GenePix pro 6.0 software (Axon Instruments, Union City, CA). Probe spots of negative detection were flagged out by the auto-alignment function of the GenePix software. Probe spots with oversaturation, bad shape, or suspected contamination were flagged manually.

The GenePix extracted results of each array were exported as a GenePix Results Format

(GPR) file. ExpressConverter version 2.1 of the TM4 Microarray Suite (Saeed et al.,

2006) was used to convert each GPR file to an annotation (ANN) file and a multiexperiment viewer (MEV) file. From the MEV file, the median signal intensity of each spot after subtracting the background noise was used for downstream analysis. The converted data was entered into Ginkgo version 1.01 (J. Craig Venter Institute, Rockville,

MD) and the replicated spots in the array were consolidated using the standard “In-Slide

Replicates Analysis” option of the Ginkgo software.

The microarray data were normalized based on the signal intensities of internal control probes. Normalized data were imported into MeV program (version 4.8.1) within the

TM4 Microarray Suite for statistical analysis. Heat maps were generated based on the log2 values of the median intensities. Clustering of samples was performed using hierarchical clustering (HCL) in the MeV program with a Manhattan distance metric.

Average linkage was used to determine cluster-to-cluster distances when constructing the hierarchical tree. Principal component analysis (PCA) was conducted using the MeV program to identify clusters of samples with similar cecal microbiome structure. The

78

standard “TTEST” option of the MeV program was used to identify the hybridized probes that were significantly (p≤0.05) correlated with treatments.

4.3.7 Linear detection range and optimal hybridization temperature determination

In order to determine the probe specificity, optimal hybridization temperature, and linear detection range, the reverse complementary oligonucleotides of the 6 internal control probes and 20 other probes (10 genus-level probes, 8 species-level probes, and 2 pathogen probes) were selected and synthesized by Alpha DNA (Montreal, Quebec,

Canada). These 26 oligonucleotides were first labeled using the Label IT µArray Cy5 kit.

The labeling procedure was modified based on manufacturer's protocol so that most of the oligonucleotides (20-30 nt) can be labeled by Cy5 fluorescent dye. Briefly, 500 ng oligonucleotides were labeled in a 50 µL labeling reaction. The Cy5 dye used in each reaction was increased from the recommended 2 µL to 6 µL. The incubation time was increased from 1 hour to 2 hours. The labeled oligonucleotides were purified using the

Oligo Clean-Up and Concentration Kit (Norgen, Thorold, Ontario, Canada). NanoDrop

ND-1000 spectrophotometer was then used to quantify both the oligonucleotides and the

Cy5 dye. It was estimated that there was one Cy5 label every 20nt. Since the sizes of oligonucleotide targets range from 20-30nt, it is safe to say most of the oligonucleotides were labeled by at least 1 Cy5 fluorescent dye.

The hybridization was performed following procedures described above. In order to determine linear detection range, Cy5-labeled targets for the 6 internal control probes, namely C1, C2, C3, C4, C5, C6, were serially diluted and applied to each array at 0.5, 5,

79

50, 500, 5000, and 50000 pg, respectively. Each array was also hybridized with the other

20 synthetic oligonucleotide targets, 109 copies each, to determine the specificity of the probes. Microarray hybridization was carried out in an HB-1000 hybridization oven

(UVP, Upland, CA) at 42℃ for 20h with rotation speed set at 24 rpm. The hybridization experiment was repeated at 45℃, 48℃, and 52℃ to determine optimal hybridization temperature.

4.4 Results and Discussion

4.4.1 PITChip3 design

Bacterial 16S rRNA gene sequences from three sources were combined into a single dataset and subjected to a series quality check and screening processes, leaving 7,492 partial (V1-V3 hypervariable region) 16S rRNA gene sequences for subsequent genus- and species-level probe design. From the 7,492 16S rRNA gene sequences, 4,402 sequences could be classified at genus level using RDP Classifier and were used to design genus-level probes based on either the consensus or the representative sequences of each genus. After removing 50 singletons (genera with only one sequence), 4,352 sequences representing 90 genera were used to design genus-level probes based on consensus sequence from each genus. In total 107 unique probes were designed for 62 genera, with 64 probes designed from consensus sequences targeting 34 genera and 43 probes designed from representative sequences targeting another 28 genera. Picky was not able to design unique probes for 28 genera. To our knowledge, it is almost impossible 80

to design unique probes for every genus of the 90 genera without having cross- hybridization issues as the 16S rRNA gene sequences of some genera (e.g. genera in the same family) are too similar with each other. Therefore, PITChip3 will not be able to analyze those 28 genera. Another approach to design genus-level probes would be to use genus-equivalent OTUs defined at 0.05 phylogenetic distance. The OTU0.05 approach may allow design of probes for more genera than the approach we used here. However, the genus-equivalent OTUs are not the same as the genera classified in the RDP database.

AS such, the probes designed using the OTU0.05 approach do not target the established genera. The current approach was used because we want PITChip3 to be able to analyze the established genera whose physiology and ecology are known.

For Species-level probes design, the 7,492 16S rRNA gene sequences were clustered into

1,739 species-equivalent OTUs at 0.04 phylogenetic distance. The conventional 0.03 phylogenetic distance (97% similarity) was not used as it was suggested by Kim et al.

(2011) that 0.04 phylogenetic distance will allow more accurate phylogenetic analysis when species-equivalent OTUs clustering was based on V1-V3 hypervariable region. In total, 1,006 unique probes were designed for 662 species-equivalent OTUs. Almost two thirds of the species-equivalent OTUs had no unique probes designed from Picky. As mentioned previously it is not feasible to design unique probes for all of the sequences without having cross-hybridization issues because the sequences used for probe design are quite similar. Although in the current study the hypervariable regions of the 16S rRNA gene sequences were used for probe design, when comparing to other genes they

81

are still highly conserved, especially those sequences belonging to the same genus or family.

On PITChip3 there are 83 probes targeting the 16S rRNA genes or virulence genes of 34 known poultry pathogens including pathogenic bacteria (e.g. Clostridium perfringens), viruses (e.g. avian adenovirus), and parasites (e.g. Eimeria maxima). Identification of pathogens using oligonucleotide probes targeting non-16S rRNA gens is applicable only when metagenomic DNA or metatranscriptome are analyzed.

Taken together, in total there are 1,204 customized oligonucleotide probes on PITChip3, including 107 genus-level probes targeting 62 genera, 1,006 species-level probes targeting 662 species-equivalent OTUs, and 83 pathogen-specific probes targeting 34 pathogens, plus 8 control probes (i.e. positive and negative control and internal control probes). The sequence information about the probes and their targets is available upon request.

4.4.2 Optimal hybridization temperature and linear detection range determination

When performing microarray hybridization, if hybridization temperature is too low, cross-hybridization is likely to occur, resulting in false positives. On the other hand, when hybridization temperature is too high, hybridization between target and probe may not occur even with a perfect match and false negatives may result (Kim, 2011). The optimal hybridization temperature would be the one at which least false positives and least false negatives result. In order to determine the optimal hybridization temperature for

PITChip3, synthetic oligonucleotide targets for 20 probes (10 genus-level probes, 8 82

species-level probes, and 2 pathogen probes) were labeled with Cy5 dye and hybridized to PITChip3 at 42℃, 45℃, 48℃, and 52℃ for 20 hours. When the hybridization temperature was set to 42℃ and 45℃, there were 4 and 2 false positives detected on the array, respectively. No false negatives were found at those two temperatures. Neither false positives nor false negatives were detected at 48℃. At 52℃, there were no false positives but 3 false negatives. Therefore, based on the 20 synthetic oligonucleotide targets tested, 48℃ was selected as the optimal hybridization temperature for PITChip3.

It is worth noting that the selected 20 probes only make up a very small fraction (<2%) of the probes on PITChip3. Optimal hybridization temperature determined based on these

20 probes may not necessarily be optimal for all the other probes on PITChip3.

Theoretically we could order synthetic oligonucleotide targets for all 1204 probes and use them to determine optimal hybridization temperature. However, it is impractical to do so.

It should also be noted that although at optimal hybridization temperature the Picky designed probes do not likely cross-hybridize to non-target within V1-V3 region, some probes may hybridize to non-target 16S rRNA sequence at locations beyond the V1-V3 region (Kim, 2011). Such a possibility cannot be excluded based just on the 20-30nt synthetic oligonucleotide targets tested here. All microarrays, especially phylochips, have these potential limitations, but these limitations should not affect comparative analysis of related or similar microbiome samples.

The linear detection range of PITChip3 at 48℃ was determined using serially diluted and

Cy5-labeled oligonucleotide targets for the 6 internal control probes. These 6 oligonucleotides, namely C1, C2, C3, C4, C5, C6, were applied to the array at 0.5, 5, 50, 83

500, 5,000, and 50,000 pg (approximately 3×107 to 3×1012 copies), respectively. The linear detection range was determined based on the log10 value of median signal intensity of each probe-target hybrid (Fig. 4.1). Signal intensity reached plateau when the copy number of target was 3×1011. When the copy number of targets ranged from 3×107 to 3×1010, it had a linear relationship with log10 signal intensity, indicating PITChip3 has a linear detection range of three orders of magnitude (107-1010 copies per array). The linear detection is narrower than that of quantitative real-time PCR, and PITChip3 is not as sensitive as qPCR. However, it allows much more comprehensive analysis of at both genus and species-equivalent levels.

4.4.3 Microarray data summary

Across all treatments, PITChip3 was able to detect 419 targets (or phylotypes), including one target for positive control (i.e. total bacteria), 44 targets for genus-level probes, 345 targets for species-level probes, and 29 pathogens. At 14 days of age, birds fed with corn- and wheat-based diet had 188 and 242 targets detected, respectively. At 35 days of age,

230 and 198 targets were detected from chickens fed with corn- and wheat-based diet, respectively. A Venn diagram was constructed to show the distribution of detected targets among different groups of chicken (Fig. 4.2).

4.4.4 Impact of dietary grain on cecal microbiome structure of chickens

Based on PCA analysis of the PITChip3 data, birds fed corn-based diet had distinct cecal microbiome from birds fed wheat-based diet at 14 and 35 days of ages (Fig. 4.3). This is not surprising because diet has the greatest potential impact on the intestinal microbiome 84

in poultry as dietary components that escape host digestion and absorption serve as the substrates for the growth of bacteria residing in the gut. Depending on the nature of feed ingredients, a diet may selectively enhance the growth and proliferation of some bacteria while having no effect or even adverse effects on others (Pan and Yu, 2013). Corn-based diet may favor the growth and proliferation of a group of bacteria which does not thrive in the GI tract of chickens fed with wheat-based diet, and vice versa.

The PCA analysis revealed difference in the cecal microbiome structure of chickens fed corn- and wheat-based diet, but it did not provide any information on what phylotypes were contributing to such difference. T-test in MeV was therefore used to further identify phylotypes with different abundance in chickens fed different diet. At 14 days of age, eight species-level phylotypes and two genus-level phylotypes were found more abundant in chickens fed wheat-based diet, whereas two species-level phylotypes were more abundant in corn-based diet (Fig. 4.4). At 35 days of age, seven species-level and one genus-level phylotypes were more abundant in birds fed wheat-based diet, while two species-level and one genus-level phylotypes were less abundant in wheat-based diet group (Fig. 4.5).

Only three species-level phylotypes, namely Akkermansia muciniphila, Clostridium perfringens, and Corynebacterium variabile can be classified to species level. A. muciniphila is the type species of the relatively new genus Akkermansia, which is proposed a decade ago by Derrien et al. (2004). Commonly found in the GI tract of human, this species is known as a mucin degrader and has been shown to be inversely

85

associated with body weight in humans and rodents (Derrien et al., 2008; Lyra et al.,

2010; Everard et al., 2013). In the current study we found that, at 14 days of age, A. muciniphila was more abundant (p≤0.05) in chickens fed wheat-based diet. At 35 days of age, however, no statistical difference was found in the abundance of A. muciniphila between chickens fed the two different diets. It is worth noting that, at both ages, A. muciniphila was present in only one (out of four) bird in the corn-based diet group, while all birds in the wheat-based diet group had this phylotype. Considering that chickens fed with corn-based diet had greater body weight than those fed with wheat-based diet

(please refer to Chapter 5 for body weight information) at both ages, it is reasonable to speculate that A. muciniphila might have inverse association with body weight of chickens fed corn- or wheat-based diets. Previous studies on mice suggested that A. muciniphila is involved in the control of host gut barrier function and physiological homeostasis during obesity and type 2 diabetes (Everard et al., 2013). To explain why in the current study A. muciniphila was found more abundant in wheat-fed chicken and to elucidate the potential connection between A. muciniphila and chicken body weight, however, will require further studies.

C. perfringens is the causative organism of necrotic enteritis (NE), a common and costly enteric disease in young chickens at approximately 4 weeks of age (Van Immerseel et al.,

2004a; Timbermont et al., 2011). This species is ubiquitous in the environment and commonly found in the GI tract of healthy birds (Si et al., 2007). In the current study C. perfringens was detected in all 16 chickens yet none of them showed symptoms of NE.

Indeed, the presence of C. perfringens alone does not necessarily lead to the development 86

of NE in chicken. Other predisposing factors are required for the disease to take place.

Previous studies indicated that wheat-based diets, which contain high levels of indigestible, water-soluble, non-starch polysaccharides, favor the proliferation of C. perfringens and predispose young chickens to NE, whereas diets poor in non-starch polysaccharides, such as corn-based diets, do not (Annett et al., 2002; Jia et al., 2009).

Supporting this view, the current study showed that at 14 days post-hatch, chickens fed with wheat-based diet had more C. perfringens in the ceca than birds fed with corn-based diet. However, no difference in the abundance of C. perfringens was observed at 35 days of age.

In the current study, Corynebacterium variabile was detected in almost all birds but difference in its abundance between corn- and wheat-fed chickens was only evident at 35 days post-hatch. Corynebacterium spp. are commonly found in chicken litter (Schefferle,

1966), therefore it is not surprising to have them detected in chicken GI tract. However, it remains to be determined why C. variabile was more abundant in birds fed with corn- based diet at 35 days of age but not at 14 days of age.

Two genus-level phylotypes, namely Bacteroides and Escherichia/Shigella/Salmonella cluster, were found to be associated with the diets. At both ages, Bacteroides were more abundant in chickens fed with wheat-based diet. With numerous genes encoding polysaccharide binding proteins and hydrolytic enzymes, Bacteroides spp. have a powerful carbohydrates utilization system, which grants them the ability to utilize a wide range of carbohydrates available in the gut (Xu and Gordon, 2003). Comparing to corn-

87

based diet, the higher content of indigestible, water-soluble, non-starch polysaccharides present in wheat-based diet may serve as substrate for the growth of Bacteroides spp., allowing them to out-compete other bacteria that cannot utilize these non-starch polysaccharides.

Members of Escherichia/Shigella/Salmonella are common dwellers in chicken GI tract even though at low abundance. While the majorities are commensal bacteria, some of them are pathogenic and can cause diseases in chickens and human. We found that

Escherichia/Shigella/Salmonella were more abundant in chickens fed with wheat-based diet at 14 days of age, whereas at 35 days of age, Escherichia/Shigella/Salmonella were found to be more abundant chickens fed with corn-based diet. The underlying mechanism of such divergence at 14 and 35 days of age, however, remains to be elucidated.

4.4.5 Succession of cecal microbiome

Age is another factor that affects gut microbiome structure. Colonization with microorganisms in the poultry gut occurs immediately after hatch and is followed by microbial succession from a transient community to a complex and climax microbiome

(Lu et al., 2003; Brisbin et al., 2008). While the establishment of a typical microbiome in the small intestine takes approximately two weeks, a typical cecal microbiome takes up to

30 days to develop (Amit-Romach et al., 2004). In the current study, when cecal microbiome of the chickens fed the same diet were compared at the two ages, a clear separation was observed between chickens at 14 and 35 days of age, suggesting that birds at these two ages had distinct cecal microbiome structures (Fig. 4.6).

88

4.5 Conclusion

Based on the previous two versions of PITChip, here we developed the third version

(PITChip3) of the poultry specific phylogenetic microarray. Compared to PITChip2,

PITChip3 has several improvements/advantages: (1) instead of using only V3 region, probes on PITChip3 were designed from V1-V3 region, which can provide more accurate classification, (2) up to three unique probes were designed for a single target which may alleviate cross-hybridization and false-positive or false negative issues, (3) no shared or multiple targets probes were used, therefore no ambiguous result will be generated. The utility of PITChip3 was tested using a small number of samples from the study in Chapter

5, and it has shown to be a powerful yet easy to use high-throughput analytical tool for poultry gut microbiome study. A few groups (genera and species) of bacteria were found to be associated with corn or wheat-based diets. These phylotypes may be partially responsible for the differences in growth performance and NE resistance seen in chickens fed with corn- and wheat-based diets.

89

Figure 4.1: Linear detection range of PITChip3. Targets for 6 internal controls, from left to right, had copy numbers ranging from 3×107 to 3×1012. The log10 median signal intensity is shown as mean±SD. PITChip3 has a linear detection range from 3×107 to 3×1010.

90

Figure 4.2: A Venn diagram showing the number of phylotypes shared among different groups of chickens. D14 and D35 reprensent birds at 14 and 35 days of age, respectively.

91

Figure 4.3: PCA plots showing the grouping of chickens based on the structure of their gut microbiome at 14 (A) and 35 (B) days of age.

92

Figure 4.4: Phylotypes with different abundance in 14-day-old chickens. Heat map (A) was generated based on the log2 value of the ratio between the mean fluorescence intensities of each phylotypes and the internal standard and displayed using the standard Green-Black-Red scheme. The mean relative abundance (shown as mean±SE) of each phylotype for each diet (B) is also represented as the log2 ratio between the mean fluorescence intensities of each phylotypes and the internal standard.

93

Figure 4.5: Phylotypes with different abundance in 35-day-old chickens. Heat map (A) was generated based on the log2 value of the ratio between the mean fluorescence intensities of each phylotypes and the internal standard and displayed using the standard Green-Black-Red scheme. The mean relative abundance (shown as mean±SE) of each phylotype for each diet (B) is also represented as the log2 ratio between the mean fluorescence intensities of each phylotypes and the internal standard.

94

Figure 4.6: PCA plots showing the grouping of chickens fed with corn-based diet (A) and wheat-based diet (B) based on the structure of their gut microbiome.

95

Chapter 5: Identification of chicken intestinal bacteria associated with corn-based or

wheat-based diet

5.1 Abstract

When comparing to chickens fed with wheat-based diet, chickens fed with corn-based diet grow better and are less susceptible to necrotic enteritis (NE) induced by Clostridium perfringens. Gut microbiome affected by different diets may be partially responsible for the difference in growth rate and susceptibility to NE. In this chapter, we investigated the difference in gut microbiome between chickens fed with corn-based or wheat-based diet.

Two hundred 1-day-old straight run Cobb 500 broiler chicks were randomly assigned to one of two dietary treatments. Each dietary treatment had 20 replicate pens with 5 birds per pen. Ileal mucosa and cecal content were collected from one bird per pen at 14 and 35 days of age. Metagenomic DNA was extracted from those samples and Illumina sequencing of the V1-V3 hypervariable region of bacterial 16S rRNA gene amplicons was performed. A phylotype-based analysis was performed on the sequence data and the results showed no clear difference in microbial diversity between two dietary treatments.

Two phylotypes, with one representing unclassified members in the family

Ruminococcaceae and the other representing unclassified Bacteria, were more abundant in the cecal content of 14-day-old chickens fed with corn-based diet and less abundant in those fed with wheat-based diet. In contrast, one phylotype classified as 96

Escherichia/Shigella was more abundant in 14-day-old chickens fed with wheat-based diet than in those fed with corn-based diet. Further studies on these phylotypes may help to find non-antibiotic alternatives to enhance growth and protect chickens from NE.

5.2 Introduction

Corn and wheat are two major cereal grains used in broiler industry worldwide. It is well accepted that corn has higher nutritional value than wheat and thus chickens fed with corn-based diet usually have better growth performance than chickens fed with wheat- based diet (Jia et al., 2009; Rodríguez et al., 2012). It is suggested that wheat-based diet contains high levels of indigestible, water-soluble, non-starch polysaccharides which raise the viscosity in the gastrointestinal (GI) tract and interfere with the contact between digestive enzymes and substrates, thus negatively affecting the growth performance of chickens (Wang et al., 2005; Slominski, 2011).

Another difference in these two diets is that chickens fed with wheat-based diet are generally more susceptible to necrotic enteritis (NE), a bacterial disease caused by virulent Clostridium perfringens. NE is the most common and costly bacterial disease in modern broiler and turkey industry as it can affect as much as 40% of the commercial broiler flocks and cost as high as $0.05 per broiler chicken in the US (Kaldhusal and

Lovland, 2000; Van der Sluis, 2000). According to the USDA Poultry Production and

Value Summary (USDA, 2013), the total number of broiler produced in the USA in 2013 was 8.52 billion, for which NE could cost the poultry industry approximately $426 97

million in 2013. Previous studies suggested that high level of non-starch polysaccharides leads to increased digesta viscosity, decreased digesta passage rate, and a decline in nutrient digestibility, which in turn favors the growth of C. perfringens (Choct et al.,

1996; Timbermont et al., 2011). Other studies postulated that there exist anti-clostridial factors in digested corn or proliferation promoting factors in digested wheat (Immerseel et al., 2004). In addition, redox potential, which can be affected by feed composition, has also been shown to have an effect on the growth of C. perfringens (Tabatabai and Walker,

1970).

To date, however, studies investigating why chickens fed with corn- and wheat-based diet differ in growth performance and NE susceptibility have placed the focus almost entirely on chemical and physical factors. Very little attention has been given to the chicken GI microbiome. The microbiome in the GI tract plays an important role in host nutrition and health as it functions as an interface between the host and the feed ingested (O’Hara et al.,

2006). Chemical and physical attributes of diets can modify the structure of the GI microbiome in chickens. In return, the GI microbiome can affect the chemical and physical environment in the GI tract (Pan and Yu, 2013). Thus, it is tempting and reasonable to speculate that chickens fed wheat- or corn-based diets will have a GI microbiome with different composition, which may, directly or indirectly, affect the growth performance of chicken and the proliferation of C. perfringens in chicken GI tract.

It is estimated that more than 900 bacterial species reside in the GI tract of chicken (Wei et al., 2013). A comprehensive characterization of such a complex microbiome requires

98

high-throughput techniques, which, compared to conventional microbial profiling methods, can provide a detailed view of the total microbiome in the GI tract of chicken.

In this study we sequenced the V1-V3 region of bacterial 16S rRNA genes from chicken gut microbiome using the Illumina MiSeq platform to characterize the bacterial populations from birds fed with either wheat- or corn-based diets.

5.3 Experimental Procedures

5.3.1 Birds and Diet

Two hundred straight run Cobb 500 broiler chicks were randomly assigned to one of two dietary treatments (wheat-based diets vs. corn based diets). The composition of the two diets is shown in Table 5.1. Each dietary treatment had 20 replicate pens with 5 birds per pen at the start of the study. The feeding experiment was conducted at the OARDC

Poultry Research Center, Wooster, OH. All the pens in this facility were separated by 24- inch-high plastic barriers to prevent litter sharing between pens. Within each pen, there was a 3-ft2 section of open concrete floor without litter so that daily water and feed management could be performed without stepping onto the litter. Water was provided by

Plasson bell-shaped drinkers and feed was provided by trough-type feeders ad libitum.

All daily management activities (feed, water) were conducted by treatment, and disposable shoes covers were used before entering each pen to minimize potential cross contamination among different treatments.

99

5.3.2 Sample collection and DNA extraction

At 14, 20, and 35 days of age, all birds were weighed to determine the average pen weight. One bird per pen was then randomly picked for sampling. The weight of birds picked was determined before euthanasia by cervical dislocation (14 days of age) or CO2 inhalation (20 and 35 days of age). Ileal mucosa and cecal content samples were collected from the birds as described previously by Cressman et al. (2010). Briefly, the ileum was excised between Meckel’s diverticulum and ileocecal junction. Ileal digesta were gently rinsed out with sterile saline. The ileum was then slit lengthwise and the mucosa was scraped off with the blunt side of a scalpel. The ceca from selected birds were excised and milked to collect the cecal contents. Both ileal mucosa and cecal content samples were kept on dry ice immediately after collection. Upon arrival at our laboratory on main campus, the samples were transferred to -80℃ until DNA extraction. All birds were handled and cared of following a protocol approved by the Institutional Animal Care and

Use Committee.

Metagenomic DNA was extracted from ileal mucosa and cecal content samples using the repeated bead beating plus column (RBB+C) method developed by Yu and Morrison

(2004). The DNA quality was accessed using agarose gel (1.0%) electrophoresis, and

DNA concentrations were determined using a NanoDrop ND-1000 spectrophotometer

(Thermo Scientific, Wilmington, DE). An aliquot of each DNA extract was taken and the

DNA concentration was adjusted to 100 ng/µl using Tris-EDTA (TE) buffer (pH 8.0).

DNA samples were stored at -20℃ until sequencing.

100

5.3.3 High throughput sequencing and analysis of sequencing data

Illumina sequencing of the V1-V3 hypervariable region of 16S rRNA gene amplicons was performed at the Molecular and Cellular Imaging Center (MCIC) of OARDC,

Wooster, OH, using 300bp paired-end MiSeq platform. The Mothur (v1.33.3) program was used to analyze the sequencing data following the Mothur MiSeq SOP (accessed 22

July 2014) with a few modifications (Kozich et al., 2013). Briefly, barcodes and primers in each read were trimmed and the paired reads were then assembled into the full-length amplicons. Base-calling at the overlapping region was based on the quality scores. The assembled sequences were then screened to remove low-quality sequences. Only sequences with no ambiguous base, maximum of 8 homopolymers, and with length more than 450 bases were retained for downstream analysis. The screened sequences were aligned using the NAST algorithm with the Silva v102 reference set of sequences, and pre-clustered allowing 1 difference for every 100bp of sequence. Potential chimeric sequences were identified and filtered using the Mothur implementation of the UCHIME algorithm (Edgar et al., 2011). Sequences were then classified using the RDP classifier

(Mothur-formatted version of training set 9) with a bootstrap value of 80% and non- bacterial sequences were removed. Due to the large size of sequence data and the limited computing resources, sequences were separated into two groups according to the sample types (i.e. ileal mucosa and cecal content) before further analysis. Sequences were binned into genus-level phylotypes based on their classification. The alpha diversity was assessed using Chao1 and ACE richness estimates, as well as Shannon and invSimpson diversity estimators. Distance matrix generated using the Yue-Clayton theta (ThetaYC) 101

similarity coefficient was used for the assessment of beta diversity (Yue and Clayton,

2005). Dendrograms based on Unweighted Pair Group Method with Arithmetic mean

(UPGMA) algorithm were constructed and principal coordinate analysis (PCoA) was performed to describe the similarity/dissimilarity among samples. Both alpha and beta diversity analyses were performed on rarefied data subsampled based on the smallest samples size in each group. Phylotypes with differential relative abundance in different groups were identified using the Mothur implementation of the Metastats method (White et al., 2009).

5.4 Results and Discussion

5.4.1 Summary of birds’ bodyweight data

At 14 and 35 days of age, the average bodyweight of the chickens in the same pen was calculated and statistical comparison using Student’s t-test was performed between chickens in the different dietary groups (Fig. 5.1). The average bodyweights in both groups are in line with the breeders expectations. At both ages, chickens fed with the corn-based diet were heavier (p≤0.05) than those fed with the wheat-based diet. Similar results were also found when comparison was done using the bodyweights of birds picked for sampling (data not shown).

5.4.2 Summary of the sequence data

102

In total 152 samples (75 ileal mucosa samples and 77 cecal content samples) were used in analyzing the gut microbiome. From these samples, 17,487,185 sequences were obtained after trimming primers and assembling paired reads. The initial screening process reduced the number of sequences to 12,828,515. After removing of chimeric and non- bacterial sequences, we obtained a total of 10,684,413 sequences, with 8,050,520 sequences from ileal mucosa samples and 2,633,893 sequences from cecal content samples. Sequences from ileal mucosa and cecal content samples were clustered into 99 and 110 genus-level phylotypes, respectively. RDP classification showed that 97.8% of the sequences from ileal mucosa can be classified into 7 phyla, whereas 8 phyla were identified representing 97.6% of sequences from cecal content samples. A detailed list of predominant phylotypes at each taxonomic rank is presented in Table 5.2. Firmicutes was the predominant phylum in both sample types, followed by Bacteroidetes, Proteobacteria, and Actinobacteria. These phyla accounted for more than 97.0% of total sequences in both the ileal mucosa and cecal content samples, which agrees with previously reported data (Wei et al., 2013b). For the ileal mucosa samples, more than 80% the sequences assigned to Firmicutes cannot be classified further to deeper taxonomic ranks, indicating the limited knowledge we currently have in the bacterial community residing in chicken ileal mucosa. Lactobacillus was the most predominant genus found in the ileal mucosa samples, followed by Clostridium cluster XI. All the other identified genera each have a relative abundance less than 1%. The predominance of Lactobacillus in the ileal mucosa samples is in agreement with previously reported data (Gong et al., 2007; Cressman et al.,

2010; Wei, 2013). Interestingly, in the studies conducted by Gong et al. (2007) and

103

Cressman et al. (2010), Lactobacillus was found to have higher relative abundances (38% and 75%, respectively) than that found in our study (10.4%), whereas the relative abundance of Lactobacillus reported by Wei (2013) was similar (11%) to our study.

Considering that studies of Gong et al. and Cressman et al. used cloning libraries while the current study and Wei’s study used NGS (Illumina MiSeq and 454 pyrosequencing, respectively), the difference in methodologies might partially present a possible explanation for such divergence seen in the relative abundance of Lactobacillus. The most predominant genera present in the cecal microbiome were Faecalibacterium and

Lachnospiracea incertae sedis, which accounted for 49.0% and 4.2% of the total sequences obtained from the cecal content samples, respectively. Similar result was reported by Cressman et al. (2010), but with Lachnospiracea incertae sedis being the most predominant genera followed by Faecalibacterium. Again, the different methodologies used (Illumina MiSeq vs. cloning library) might be partially responsible for such divergence. Other less predominant genera identified in cecal content included

Lactobacillus, Alistipes, Bifidobacterium, Clostridium cluster XlVa, Bacteroides, and

Escherichia/Shigella. These genera accounted for 11.6% of total sequences of the cecal content samples.

5.4.3 Comparison of microbiome diversity

The alpha diversity for each group of samples was investigated using Chao1, ACE,

Shannon and invSimpson and the results are summarized in Table 5.3. Overall, these diversity measurements had higher values for the cecal content samples than for the ileal

104

mucosa samples, suggesting that the cecal microbiome is more diverse than the microbiome associated with lieal mucosa. The digesta passes through the GI tract faster in poultry than in mammalians as the poultry GI tract is much shorter than that of mammalian animals. Previous studies suggested that the average whole tract transit time is less than 3.5 hours (Hughes, 2008). With such a short retention time, not all bacteria but only those that adhere to the mucosal layer and/or have a short generation time can thrive in the upper GI tract including ileum. On the other hand, the ceca, which are two blind pouches, have a rather slow passage rate, which makes them ideal habitats for a diverse microbiome. When comparing within the same sample type, we found that ileal mucosa in chickens at 35 days of age harbors a more diverse microbiome than those at 14 days of age. While previous studies have reported that a typical microbiome in chicken small intestine is established within 2 weeks (Amit-Romach et al., 2004), our study suggested that the microbiome associated with ileal mucosa may continue to develop after two weeks of age. To our surprise, no apparent difference in these diversity measurements was found between the chickens fed the two different diets. However, alpha diversity measurements only reflect overall diversity and differences in relative abundance of some species or groups are not necessarily indicated by these measurements.

UPGMA-based dendrograms were constructed using ThetaYC-based distance matrix to describe the similarity among samples of the same type. For ileal mucosa samples, chickens at 14 days of age seemed to group together while chickens at 35 days of age formed another group (Fig. 5.2). However, no clear clustering was observed based on the 105

two different diets. For cecal content samples, although a few samples in the same group were clustered together at the terminal nodes, in general no clear grouping was observed based on either the age of the birds or the diets fed to the birds (Fig. 5.3). It was also observed that, the ileal mucosa samples collected from the 35-day-old chicken in pen 31

(Wheat.35.M.31) and the cecal content from the 35-day-old chicken in pen 46

(Wheat.35.C.46) seemed to be different from all other samples and was clustered by themselves. A closer look at these two samples revealed that the sample Wheat.35.M.31 had a much higher proportion (79.2%) of sequences that cannot be classified to any known phylum when compared to other samples (average relative abundance of unclassified sequences <2%), whereas sample Wheat.35.C.46 contributed 87.0% of sequences from the phylum Actinobacteria found in all cecal content samples (data not shown). It is noteworthy that interpretation of dendrograms should be done with caution, as they are sensitive to outliers and small perturbations in the data (Milligan and Cheng,

1996; Park et al., 2009). In addition, dendrogram is merely a description of the results of the clustering algorithm. Different algorithm used will lead to different dendrograms.

Hierarchical structure indicated by dendrogram may not actually exist in the data.

The similarity among samples was further interrogated using principal coordinate analysis (PCoA) based on ThetaYC distance matrix. With the first principle coordinate 1

(PC1) and PC2 explaining about 63% of total variation, ileal mucosa samples collected from 14-day-old chickens were separated from those collected at 35 days of age along

PC1 (Fig. 5.4), whereas no clear separation was observed between chickens fed with wheat and chickens fed with corn, which is in agreement with the clustering shown on the 106

dendrogram. Clustering of samples was also observed on the PCoA plot of cecal content samples but does not correspond to either diet or age of chickens (Fig. 5.5).

Due to the large amount of reads produced by MiSeq platform and the limitation in computing recourses, we were only able to perform phylotype-based analysis but not

OTU-based analysis. Phylotype-based analysis relies upon reference taxonomic outlines to classify sequences and then bin sequences into different groups based on their classification (Schloss and Westcott, 2011). Compared to OTU-based analysis, phylotype-based analysis is far less demanding on computing power and time. It is also less sensitive to sequencing errors, which is a common problem faced by NGS technologies (Kunin et al, 2010). However, most taxonomic references only provide classification down to genus level. At this taxonomic resolution, information at species level is masked. While diversity analysis at genus level indicated no difference in microbiome composition between the two dietary groups, it is possible that difference can be observed if analysis was done at species level. Update in taxonomic references and development in bioinformatics tools that is less computing power demanding may help revealing the hidden secrets at species level in the future.

5.4.4 Identification of phylotypes with differential relative abundance in different dietary groups

Although both the alpha and the beta diversity analyses did not reveal apparent difference in gut microbiome between the two dietary groups, it is possible that diet only affected a small number of phylotypes which did not cause differences on the dendrograms or the

107

PCoA plots. Therefore, Metastats was performed with 1000 permutations to test if such phylotypes exist. In order to control the false positive rate and false discovery rate we set both the p-value and q-value threshold of significance to 0.05 (Huang et al., 2011; Looft et al., 2014). Only those phylotypes whose p-values and q-values both fell below 0.05 were considered as significantly different between the two dietary groups. We noticed that some phylotypes reported as being significant were present in only 1 or 2 samples.

We feel that in such cases statistically significant may not suggest biological significance.

In other words, if corn-based or wheat-based diet really causes an increase in population of one phylotype, we would like to see that particular phylotypes is present in sufficient numbers of samples in the corresponding group. Therefore, in addition to p-value and q- value, only those phylotypes present in >50% of samples in at least one group (corn- based or wheat-based diet) were considered as significant.

No phylotype in the ileal mucosa samples was found with differential relative abundance between the two different dietary groups. Three phylotypes was identified as significant in the cecal content samples of chickens at 14 days of age (Fig. 5.6). One of them represented sequences classified to the family Ruminococcaceae without further classification at genus level. This phylotype was more abundant in birds fed with corn- based diet than in birds fed with wheat-based diet. Since birds fed with corn-based diet had better growth than birds fed with wheat-based diet, the relative abundance of this phylotype may be associated with chickens’ performance. Supporting this view, in the study described in Chapter 3, a phylotype classified as unclassified Ruminococcaceae was found with a greater relative abundance in chickens with better performance (low 108

FCR) than in chickens with high FCR. Members in the Ruminococcaceae family, such as

Ruminococcus spp., are powerful cellulose and xylan degraders which may help degrade the otherwise indigestible polysaccharides present in the diet and thus enhance the growth of the birds (Stanley et al., 2013).

Another phylotype was also found to be more abundant in birds fed corn-based diet.

However, this phylotype can only be classified down to phylum level. Different from

OTU-based analysis, in which OTUs classified down to only phylum level still represent groups of sequences at species-equivalent level (i.e. >97% similarity among all sequences in one OTU), in phylotype-based studies the phylotype classified as unclassified Bacteria represents a collection of all sequences that cannot be classified at deeper taxonomy ranks (i.e. class). Sequences in this phylotype could represent bacteria in different species, genera, families, orders, classes, and even different phyla. Therefore, it is difficult to interpret the result regarding this particular phylotype. Further study is needed.

The last phylotype with differential relative abundance between the two different dietary groups was classified as Escherichia/Shigella. Different from the other two, this phylotype was found to be more abundant in chickens fed with wheat-based diet, which is in agreement with the PITChip3 result presented in Chapter 4. The average relative abundance of this phylotype in birds fed wheat-based diet was approximately 4.0%, whereas in birds fed with corn-based diet this phylotype had an average relative abundance less than 0.1%. Members of the Escherichia/Shigella cluster usually do not possess traits beneficial to the host. Most of them are commensal bacteria, while a few,

109

such as avian pathogenic Escherichia coli (APEC), are pathogenic to chicken host. Even the commensal members in this cluster may cause infectious diseases when the birds are immunosuppressed or the gut barrier is violated (Nakazato et al., 2009). Higher relative abundance of this phylotype in chickens fed with wheat-based diet might have negatively affected the gut function of those birds, which in turn slowed the growth of chickens fed with wheat-based diet.

It is widely accepted that wheat-based diet favors the proliferation of C. perfringens whereas corn-based diet does not (Annett et al., 2002; Jia et al., 2009). In the study described in Chapter 4, the PITChip3 results indicated that C. perfringens was more abundant in chickens fed wheat-based diet than in chickens fed with corn-based diet at 14 days of age. However, our study provided no information on the relative abundance of C. perfringens as sequences can only be classified to genus level using phylotype-based analysis described above. Considering the samples used in the PITChip3 study were a subset of samples in this study, it is possible that the relative abundance of C. perfringens were different in the two dietary treatments. However, further study is needed to ascertain such possibility.

5.5 Conclusion

Given the increasingly high cost of corn due to the competition from the bioehanol industry, there is increasing interest in utilizing higher levels of wheat than what has been used in past years. However, when comparing to chickens fed with corn-based diet, 110

chickens fed with wheat-based diet grow slower and are more susceptible to NE.

Currently, the primary nutritional approach to enhance growth and reduce risk of NE in chickens is the incorporation of sub-therapeutic dietary antibiotics (Dahiya et al., 2006).

However, due to the growing concern over increased development of antibiotic resistance, there is a trend towards the abolishment of the use of in-feed antibiotic growth promoters

(AGP). In the current study 3 phylotypes were found to have different relative abundance in different dietary groups. Although we were not able to provide information on these phylotypes at deeper taxonomic ranks, further studies on these phylotypes may help to find non-antibiotic alternatives, such as probiotics and prebiotics, to enhance growth and protect chickens from NE.

111

Table 5.1: Dietary composition

Ingredient Wheat-based Diet, % Corn-based Diet, %

Ground wheat 65.90 NA

Ground corn NA 59.40

Soybean meal, 48% 24.30 33.30

Blended fat 5.50 3.00

Dicalcium phosphate, 18.5% 1.80 1.80

Ground limestone 1.10 1.20

Akey Turkey Starter Premix 0.50 0.50

Salt 0.35 0.40

DL-methionine 0.27 0.30

L-lysine 0.28 0.10

112

Table 5.2: Predominant phylotypes at each taxonomic rank*

Phylum Class Order Family Genus

Firmicutes (97.6%)** Bacilli (10.7%) Lactobacillales (10.7%) Lactobacillaceae (10.4%) Lactobacillus (10.4%) Proteobacteria (0.2%) Clostridia (8.1%) Clostridiales (8.1%) Peptostreptococcaceae (7.3%) Clostridium_XI (7.2%) Bacteroidetes (<0.05%) Gammaproteobacteria (0.2%) Oceanospirillales (0.1%) Clostridiaceae_1 (0.4%) Clostridium_sensu_stricto (0.3%) Ileal Actinobacteria (<0.05%) Bacteroidia (<0.05%) Enterobacteriales (0.1%) Ruminococcaceae (0.1%) Faecalibacterium (0.1%) Planctomycetes mucosa (<0.05%) Flavobacteria (<0.05%) Bacteroidales (<0.05%) Halomonadaceae (0.1%) Halomonas (0.1%) Verrucomicrobia (<0.05%) Sphingobacteria (<0.05%) Bifidobacteriales (<0.05%) Enterococcaceae (0.1%) Enterococcus (0.1%) Acidobacteria (<0.05%) Actinobacteria (<0.05%) Alteromonadales (<0.05%) Enterobacteriaceae (0.1%) Escherichia_Shigella (0.1%) Erysipelotrichia (<0.05%) Erysipelotrichales (<0.05%) Lachnospiraceae (<0.05%) Alistipes (<0.05%)

Negativicutes (<0.05%) Selenomonadales (<0.05%) Rikenellaceae (<0.05%) Bifidobacterium (<0.05%) Epsilonproteobacteria (<0.05%) Campylobacterales (<0.05%) Bifidobacteriaceae (<0.05%) Shewanella (<0.05%)

11

3 Firmicutes (90.7%) Clostridia (83.7%) Clostridiales (83.7%) Ruminococcaceae (56.8%) Faecalibacterium (49.0%)

Lachnospiracea_incertae_sedis Bacteroidetes (4.0%) Bacteroidia (4.0%) Bacteroidales (4.0%) Lachnospiraceae (21.6%) (4.2%) Actinobacteria (1.5%) Bacilli (4.0%) Lactobacillales (4.0%) Lactobacillaceae (4.0%) Lactobacillus (4.0%) Cecal Proteobacteria (1.3%) Actinobacteria (1.5%) Bifidobacteriales (1.4%) Rikenellaceae (3.0%) Alistipes (3.0%) Verrucomicrobia content (<0.05%) Gammaproteobacteria (1.3%) Enterobacteriales (1.3%) Bifidobacteriaceae (1.4%) Bifidobacterium (1.3%) TM7 (<0.05%) Erysipelotrichia (1.2%) Erysipelotrichales (1.2%) Enterobacteriaceae (1.3%) Clostridium_XlVa (1.3%) Lentisphaerae (<0.05%) Negativicutes (0.5%) Selenomonadales (0.5%) Erysipelotrichaceae (1.2%) Bacteroides (1.0%) Epsilonproteobacteria Tenericutes (<0.05%) (<0.05%) Coriobacteriales (0.1%) Bacteroidaceae (1.0%) Escherichia_Shigella (1.0%) Verrucomicrobiae (<0.05%) Campylobacterales (<0.05%) Veillonellaceae (0.5%) Clostridium_XVIII (0.8%) Verrucomicrobiales Alphaproteobacteria (<0.05%) (<0.05%) Peptostreptococcaceae (0.1%) Pseudoflavonifractor (0.7%) * The top 10 predominant phylotypes in each taxonomic rank is presented

** The percentages indicate the relative abundance of each phylotypes 62

113

Table 5.3: Alpha diversity measurements

Observed phylotypes Chao1 ACE Shannon invSimpson Corn14 6.07±3.86* 6.99±4.55 8.5±6.32 0.3±0.36 1.26±0.47 Ileal Wheat14 5.67±2.71 6.48±3.39 8.64±5.9 0.2±0.24 1.12±0.2 Mucosa Corn35 14.11±7.04 17.27±9.49 19.8±10.82 0.59±0.5 1.62±0.95 Wheat35 14.32±6.56 17.34±9.22 22.08±13.05 0.78±0.5 1.87±1.01 Corn14 28.21±4.79 33.37±5.34 36.32±4.71 1.63±0.48 3.78±2.01 Cecal Wheat14 25.69±4.63 30.2±6.12 33.11±7.12 1.46±0.47 3.2±1.72 content Corn35 32.07±5.74 38.4±6.96 41.56±8.59 1.58±0.54 3.36±1.59 Wheat35 28.9±5.37 34.93±6.94 37.62±7.86 1.39±0.54 2.99±1.56 * Subsampling was performed 1000 times to each sample and diversity measurements were calculated for each subsample. The numbers reported here are shown as mean±SD

114

Figure 5.1: Average pen weight. Data is shown as mean±SE.

115

Figure 5.2: Dendrogram showing similarity among ileal mucosa samples. Sample ID is arranged as “Diet.Age.Sample type.Pen#”, with M stands for ileal mucosa.

116

Figure 5.3: Dendrogram showing similarity among cecal content samples. Sample ID is arranged as “Diet.Age.Sample type.Pen#”, with C stands for cecal conent. 117

11

8

Figure 5.4: PCoA plot showing the grouping of chickens based on the structure of their ileal mucosa microbiome.

62

118

11

9

Figure 5.5: PCoA plot showing the grouping of chickens based on the structure of their cecal content microbiome.

62

119

1

20

Figure 5.6: Three phylotypes with differential (p≤0.05; q≤0.05) relative abundance in cecal content of 14-day-old

chickens fed with different diet. Data is shown as mean±SE.

62

120

Chapter 6: General discussion

In the old days, people used to believe that all bacteria in the gut are bad. Now it is widely recognized and accepted that bacteria in the GI tract are of fundamental importance to host as they benefit the host in a wide range of aspects, especially nutrition and disease resistance. Improving poultry growth and health by manipulation of the gut microbiome through dietary and managerial interventions has been practiced in poultry industry for decades. Currently, AGPs is still the most effective nutritional approach to enhance growth and reduce risk of disease in chickens. However, due to the growing concern over widespread antibiotic resistance, there is a trend towards abolishing the use of AGPs worldwide, and the United States is likely to follow this trend and ban AGPs by

2016. Several non-antibiotic approaches have been developed and used but none of them can be as cost-effective as APGs. Clearly, further knowledge of chicken gut microbiome is needed. However, due to the fact that most bacteria in the gut cannot be cultured and also their community is complex and individual populations are large and variable, gaining a comprehensive knowledge of gut microbiome is never an easy task. Thanks to the fast developing 16S rRNA gene-targeted high-throughput approaches such as phylogenetic microarray and NGS, we now have the opportunity to study gut microbiome at a broad scope never before possible.

As the first and only phylogenetic microarrays specific for poultry, the PITChips we developed have proved their utility as robust, fast, and cost-effective analytic tools for

121

poultry gut microbiome studies. Compared to NGS technologies, PITChips have several advantages in comparative analysis of poultry gut microbiomes. First, PITChips can provide a uniformed, robust assessment of individual populations of poultry gut microbiomes. Second, total rRNA, which reflects both the diversity and metabolic status of gut microbiome, can be directly labeled and analyzed using PITChips. Thus potential

PCR bias is not likely to compromise the assessment of microbiomes. Third, analysis of data generated from PITChips is far less resource demanding and time consuming. Fourth,

PITChips can be updated quickly when annotations of existing probes need to be changed or new probes need to be added. Last, PITChips can be custom fabricated or printed in house at relatively low cost (less than $250 per chip). Designed based on the previous two versions, PITChip3, the latest version of PITChips, are advantageous for three reasons. First, the probes on PITChip3 were designed based on the V1-V3 region of bacterial 16S rRNA gene, whereas PITChip1 and 2 used only V3 region. Second,

PITChip3 only included unique probes, thus no ambiguous results from shared and multi- target probes are expected. Third, each target has up to three unique probes designed, minimizing potential false positive issues caused by cross-hybridization. Admittedly, however, PITChip3 is still far away from being perfect. A major drawback is that, as a trade-off to alleviate cross-hybridization issue and assure probe specificity, specific probes cannot be designed for a significant portion of the sequences of poultry origin.

Future researches should focus on this issue, and increasing the length and number of high quality sequences may be one solution.

122

Recently, Illumina sequencing has gained more popularity than other NGS methods (e.g.

454 pyrosequencing) as with current technologies the number of reads Illumina sequencing platforms generate far exceeds that of 454 pyrosequencing systems yet both platforms produce reads with similar length. However, the increase in the number of reads makes analysis a challenging task. In our case, for example, Illumina sequencing with MiSeq platform generated too many reads to analyze with OTU-based approaches using current version of Mothur. We tried QIIME as well but without success. We noticed that for both Mothur and QIIME, OTU clustering was the step where they failed to accomplish. Indeed, as the number of reads increases, the size of distance matrix increases exponentially. With more than 10 million sequences prior to distance matrix construction, sequence analysis with OTU-based approaches is currently out of reach.

Therefore, a phylotype-based approach was utilized to analysis our data. As the analytic software/programs develop quickly and computers with more RAM and CPUs become affordable, OTU-based analysis on large scale sequence data may be feasible in the future.

By using high-throughput phylogenetic microarray (PITChip2 and 3) and Illumina sequencing, the three studies presented above investigated gut microbiome in chickens with different growth performance and fed different diets. The basic rationale behind these studies is that if we can recreate the gut microbiome structures seen in chickens with better growth performance and disease resistance, we may be able to improve growth and disease resistance in other chickens. Our studies served as the first step towards understanding the association between gut microbiome and chicken growth performance/disease resistance. By identifying phylotypes associated with feed 123

utilization efficiency and diets, we revealed the potential inter-connection between gut microbiome and host growth performance and health. Although association does not necessarily indicate a cause-and-effect relationship, our studies pointed out the direction for future studies by narrowing down the area of focus to only a few phylotypes rather than the whole gut microbiome. Next step would be to test and confirm the cause-and- effect relationship between identified phylotypes and growth performance/disease resistance. The last step would be to test the effectiveness of the manipulation, by non- antibiotic approaches such as probiotics and prebiotics, of the identified phylotypes on altering chicken growth performance and disease resistance. Through this series of studies we may be able to find alternative approaches that improve disease resistance and also growth performance in chickens, even when they are fed low nutritional value diets such as wheat-based diet.

124

References

Alain, K., & Querellou, J. (2009). Cultivating the uncultured: Limits, advances and future

challenges. Extremophiles, 13(4), 583-594.

Albuquerque, P., Mendes, M. V., Santos, C. L., Moradas-Ferreira, P., & Tavares, F.

(2009). DNA signature-based approaches for bacterial detection and identification.

Science of the Total Environment, 407(12), 3641-3651.

Alemka, A., Clyne, M., Shanahan, F., Tompkins, T., Corcionivoschi, N., & Bourke, B.

(2010a). Probiotic colonization of the adherent mucus layer of HT29MTXE12

cells attenuates Campylobacter jejuni virulence properties. Infection and

Immunity, 78(6), 2812-2822. doi:10.1128/IAI.01249-09; 10.1128/IAI.01249-09

Alemka, A., Whelan, S., Gough, R., Clyne, M., Gallagher, M. E., Carrington, S. D., &

Bourke, B. (2010b). Purified chicken intestinal mucin attenuates Campylobacter

jejuni pathogenicity in vitro. Journal of Medical Microbiology, 59(Pt 8), 898-903.

doi:10.1099/jmm.0.019315-0; 10.1099/jmm.0.019315-0

Alemka, A., Corcionivoschi, N., & Bourke, B. (2012). Defense and adaptation: The

complex inter-relationship between Campylobacter jejuni and mucus. Frontiers in

Cellular and Infection Microbiology, 2, 15. doi:10.3389/fcimb.2012.00015;

10.3389/fcimb.2012.00015

125

Amit-Romach, E., Sklan, D., & Uni, Z. (2004). Microflora ecology of the chicken

intestine using 16S ribosomal DNA primers. Poultry Science, 83(7), 1093-1098.

Annett, C., Viste, J., Chirino-Trejo, M., Classen, H., Middleton, D., & Simko, E. (2002).

Necrotic enteritis: Effect of barley, wheat and corn diets on proliferation of

Clostridium perfringens type A. Avian Pathology, 31(6), 598-601.

Banerjee, A., Yasseri, M., & Munson, M. (2002). A method for the detection and

quantification of bacteria in human carious dentine using fluorescent in situ

hybridisation. Journal of Dentistry, 30(7), 359-363.

Bar-Shira, E., Sklan, D., & Friedman, A. (2003). Establishment of immune competence

in the avian GALT during the immediate post-hatch period. Developmental and

Comparative Immunology, 27(2), 147-157.

Barnes, E. M., Mead, G., Barnuml, D., & Harry, E. (1972). The intestinal flora of the

chicken in the period 2 to 6 weeks of age, with particular reference to the

anaerobic bacteria. British Poultry Science, 13(3), 311-326.

BEDBURY, H. P., & Duke, G. (1983). Cecal microflora of turkeys fed low or high fiber

diets: Enumeration, identification, and determination of cellulolytic activity.

Poultry Science, 62(4), 675-682.

Bent, S. J., & Forney, L. J. (2008). The tragedy of the uncommon: Understanding

limitations in the analysis of microbial diversity. The ISME Journal, 2(7), 689-

695.

126

Berg, J. M., & Shi, Y. (1996). The galvanization of biology: A growing appreciation for

the roles of zinc. Science (New York, N.Y.), 271(5252), 1081-1085.

Blottiere, H. M., Buecher, B., Galmiche, J. P., & Cherbut, C. (2003). Molecular analysis

of the effect of short-chain fatty acids on intestinal cell proliferation. The

Proceedings of the Nutrition Society, 62(1), 101-106. doi:10.1079/PNS2002215

Boto, L. (2010). Horizontal gene transfer in evolution: Facts and challenges.

Proceedings.Biological Sciences / the Royal Society, 277(1683), 819-827.

doi:10.1098/rspb.2009.1679; 10.1098/rspb.2009.1679

Brisbin, J. T., Gong, J., & Sharif, S. (2008). Interactions between commensal bacteria and

the gut-associated immune system of the chicken. Animal Health Research

Reviews / Conference of Research Workers in Animal Diseases, 9(1), 101-110.

doi:10.1017/S146625230800145X; 10.1017/S146625230800145X

Brisbin, J. T., Gong, J., Orouji, S., Esufali, J., Mallick, A. I., Parvizi, P., . . . Sharif, S.

(2011). Oral treatment of chickens with lactobacilli influences elicitation of

immune responses. Clinical and Vaccine Immunology : CVI, 18(9), 1447-1455.

doi:10.1128/CVI.05100-11; 10.1128/CVI.05100-11

Brisbin, J. T., Parvizi, P., & Sharif, S. (2012). Differential cytokine expression in T-cell

subsets of chicken caecal tonsils co-cultured with three species of Lactobacillus.

Beneficial Microbes, 3(3), 205-210. doi:10.3920/BM2012.0014;

10.3920/BM2012.0014

127

Brodie, E. L., Desantis, T. Z., Joyner, D. C., Baek, S. M., Larsen, J. T., Andersen, G.

L., . . . Firestone, M. K. (2006). Application of a high-density oligonucleotide

microarray approach to study bacterial population dynamics during uranium

reduction and reoxidation. Applied and Environmental Microbiology, 72(9),

6288-6298. doi:72/9/6288 [pii]

Brugère, J., Mihajlovski, A., Missaoui, M., & Peyret, P. (2009). Tools for stools: The

challenge of assessing human intestinal microbiota using molecular diagnostics.

Expert. Rev. Mol. Diagn. 9:353-365.

Buchmeier, N. A., & Heffron, F. (1991). Inhibition of macrophage phagosome-lysosome

fusion by Salmonella Typhimurium. Infection and Immunity, 59(7), 2232-2238.

Buermans, H., & den Dunnen, J. (2014). Next generation sequencing technology:

Advances and applications. Biochimica Et Biophysica Acta (BBA)-Molecular

Basis of Disease, 1932-1941.

Byrne, C. M., Clyne, M., & Bourke, B. (2007). Campylobacter jejuni adhere to and

invade chicken intestinal epithelial cells in vitro. Microbiology (Reading,

England), 153(Pt 2), 561-569. doi:10.1099/mic.0.2006/000711-0

Campoy, S., Jara, M., Busquets, N., Perez De Rozas, A. M., Badiola, I., & Barbe, J.

(2002). Role of the high-affinity zinc uptake znuABC system in Salmonella

enterica serovar Typhimurium virulence. Infection and Immunity, 70(8), 4721-

4725.

128

Caporaso, J. G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F. D., Costello, E.

K., . . . Gordon, J. I. (2010). QIIME allows analysis of high-throughput

community sequencing data. Nature Methods, 7(5), 335-336.

Caporaso, J. G., Lauber, C. L., Walters, W. A., Berg-Lyons, D., Huntley, J., Fierer, N., . . .

Bauer, M. (2012). Ultra-high-throughput microbial community analysis on the

illumina HiSeq and MiSeq platforms. The ISME Journal, 6(8), 1621-1624.

Chae, B., Ingale, S., Kim, J., Kim, K., Sen, S., Lee, S., . . . Kwon, I. (2012). Effect of

dietary supplementation of probiotics on performance, caecal microbiology and

small intestinal morphology of broiler chickens. Animal Nutrition and Feed

Technology, 12(1), 1-12.

Cheled-Shoval, S. L., Amit-Romach, E., Barbakov, M., & Uni, Z. (2011). The effect of in

ovo administration of mannan oligosaccharide on small intestine development

during the pre- and posthatch periods in chickens. Poultry Science, 90(10), 2301-

2310. doi:10.3382/ps.2011-01488; 10.3382/ps.2011-01488

Chen, Y. A., Chou, C. C., Lu, X., Slate, E. H., Peck, K., Xu, W., . . . Almeida, J. S.

(2006). A multivariate prediction model for microarray cross-hybridization. BMC

Bioinformatics, 7, 101. doi:1471-2105-7-101 [pii]

Chiang, G., Lu, W., Piao, X., Hu, J., Gong, L., & Thacker, P. (2010). Effects of feeding

solid-state fermented rapeseed meal on performance, nutrient digestibility,

129

intestinal ecology and intestinal morphology of broiler chickens. Asian-

Aust.J.Anim.Sci, 23(2), 263-271.

Choct, M., Hughes, R., Wang, J., Bedford, M., Morgan, A., & Annison, G. (1996).

Increased small intestinal fermentation is partly responsible for the anti-nutritive

activity of non-starch polysaccharides in chickens. British Poultry Science, 37(3),

609-621.

Chou, H. H., Hsia, A. P., Mooney, D. L., & Schnable, P. S. (2004). Picky: Oligo

microarray design for large genomes. Bioinformatics (Oxford, England), 20(17),

2893-2902. doi:10.1093/bioinformatics/bth347 [doi]

Cirillo, D. M., Valdivia, R. H., Monack, D. M., & Falkow, S. (1998). Macrophage-

dependent induction of the Salmonella pathogenicity island 2 type III secretion

system and its role in intracellular survival. Molecular Microbiology, 30(1), 175-

188.

Collier, C. T., Hofacre, C. L., Payne, A. M., Anderson, D. B., Kaiser, P., Mackie, R. I., &

Gaskins, H. R. (2008). Coccidia-induced mucogenesis promotes the onset of

necrotic enteritis by supporting Clostridium perfringens growth. Veterinary

Immunology and Immunopathology, 122(1-2), 104-115.

doi:10.1016/j.vetimm.2007.10.014

130

Coufal, C., Chavez, C., Niemeyer, P., & Carey, J. (2006). Measurement of broiler litter

production rates and nutrient content using recycled litter. Poultry Science, 85(3),

398-403.

Craven, S., Stern, N., Cox, N., Bailey, J., & Berrang, M. (1999). Cecal carriage of

Clostridium perfringens in broiler chickens given mucosal starter culture™.

Avian Diseases, , 484-490.

Cressman, M. D., Yu, Z., Nelson, M. C., Moeller, S. J., Lilburn, M. S., & Zerby, H. N.

(2010). Interrelations between the microbiotas in the litter and in the intestines of

commercial broiler chickens. Applied and Environmental Microbiology, 76(19),

6572-6582. doi:10.1128/AEM.00180-10 [doi]

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. Infection and

Immunity, 79(7), 2755-2763. doi:10.1128/IAI.01375-10; 10.1128/IAI.01375-10

Dahiya, J., Wilkie, D., Van Kessel, A., & Drew, M. (2006). Potential strategies for

controlling necrotic enteritis in broiler chickens in post-antibiotic era. Animal

Feed Science and Technology, 129(1), 60-88.

Danzeisen, J. L., Kim, H. B., Isaacson, R. E., Tu, Z. J., & Johnson, T. J. (2011).

Modulations of the chicken cecal microbiome and metagenome in response to

anticoccidial and growth promoter treatment. PloS One, 6(11), e27949.

131

Davis, L. M., Kakuda, T., & DiRita, V. J. (2009). A Campylobacter jejuni znuA

orthologue is essential for growth in low-zinc environments and chick

colonization. Journal of Bacteriology, 191(5), 1631-1640. doi:10.1128/JB.01394-

08; 10.1128/JB.01394-08

Denbow, D. (2000). Gastrointestinal anatomy and physiology. Sturkie's Avian

Physiology, 5, 299-325.

Derache, C., Esnault, E., Bonsergent, C., Le Vern, Y., Quere, P., & Lalmanach, A. C.

(2009). Differential modulation of beta-defensin gene expression by Salmonella

Enteritidis in intestinal epithelial cells from resistant and susceptible chicken

inbred lines. Developmental and Comparative Immunology, 33(9), 959-966.

doi:10.1016/j.dci.2009.03.005; 10.1016/j.dci.2009.03.005

Derrien, M., Vaughan, E. E., Plugge, C. M., & de Vos, W. M. (2004). Akkermansia

muciniphila gen. nov., sp. nov., a human intestinal mucin-degrading bacterium.

International Journal of Systematic and Evolutionary Microbiology, 54(Pt 5),

1469-1476. doi:10.1099/ijs.0.02873-0 [doi]

Derrien, M., Collado, M. C., Ben-Amor, K., Salminen, S., & de Vos, W. M. (2008). The

mucin degrader Akkermansia muciniphila is an abundant resident of the human

intestinal tract. Applied and Environmental Microbiology, 74(5), 1646-1648.

doi:10.1128/AEM.01226-07

132

Derrien, M., van Passel, M. W., van de Bovenkamp, J. H., Schipper, R. G., de Vos, W.

M., & Dekker, J. (2010). Mucin-bacterial interactions in the human oral cavity

and digestive tract. Gut Microbes, 1(4), 254-268. doi:10.4161/gmic.1.4.12778

DeSantis, T. Z., Brodie, E. L., Moberg, J. P., Zubieta, I. X., Piceno, Y. M., & Andersen,

G. L. (2007). High-density universal 16S rRNA microarray analysis reveals

broader diversity than typical clone library when sampling the environment.

Microbial Ecology, 53(3), 371-383.

Devereux, R., & Willis, S. G. (1995). Amplification of ribosomal RNA sequences.

Molecular microbial ecology manual (pp. 277-287) Springer.

Dhanarani, T. S., Shankar, C., Park, J., Dexilin, M., Kumar, R. R., & Thamaraiselvi, K.

(2009). Study on acquisition of bacterial antibiotic resistance determinants in

poultry litter. Poultry Science, 88(7), 1381-1387. doi:10.3382/ps.2008-00327;

10.3382/ps.2008-00327

Dibner, J., & Richards, J. (2005). Antibiotic growth promoters in agriculture: History and

mode of action. Poultry Science, 84(4), 634-643.

Dobson, A., Cotter, P. D., Ross, R. P., & Hill, C. (2012). Bacteriocin production: A

probiotic trait? Applied and Environmental Microbiology, 78(1), 1-6.

doi:10.1128/AEM.05576-11; 10.1128/AEM.05576-11

Donalson, L., McReynolds, J., Kim, W., Chalova, V., Woodward, C., Kubena, L., . . .

Ricke, S. (2008). The influence of a fructooligosaccharide prebiotic combined

133

with alfalfa molt diets on the gastrointestinal tract fermentation, Salmonella

Enteritidis infection, and intestinal shedding in laying hens. Poultry Science,

87(7), 1253-1262.

Drew, M., Syed, N., Goldade, B., Laarveld, B., & Van Kessel, A. (2004). Effects of

dietary protein source and level on intestinal populations of Clostridium

perfringens in broiler chickens. Poultry Science, 83(3), 414-420.

Dunbar, J., Ticknor, L. O., & Kuske, C. R. (2001). Phylogenetic specificity and

reproducibility and new method for analysis of terminal restriction fragment

profiles of 16S rRNA genes from bacterial communities. Applied and

Environmental Microbiology, 67(1), 190-197. doi:10.1128/AEM.67.1.190-

197.2001 [doi]

Edgar, R. C., Haas, B. J., Clemente, J. C., Quince, C., & Knight, R. (2011). UCHIME

improves sensitivity and speed of chimera detection. Bioinformatics (Oxford,

England), 27(16), 2194-2200. doi:10.1093/bioinformatics/btr381 [doi]

Engberg, R. M., Hedemann, M. S., Steenfeldt, S., & Jensen, B. B. (2004). Influence of

whole wheat and xylanase on broiler performance and microbial composition and

activity in the digestive tract. Poultry Science, 83(6), 925-938.

Everard, A., Belzer, C., Geurts, L., Ouwerkerk, J. P., Druart, C., Bindels, L. B., . . . Cani,

P. D. (2013). Cross-talk between Akkermansia muciniphila and intestinal

epithelium controls diet-induced obesity. Proceedings of the National Academy of

134

Sciences of the United States of America, 110(22), 9066-9071.

doi:10.1073/pnas.1219451110 [doi]

Ezaki, T., Yamamoto, N., Ninomiya, K., SUZUKI, S., & Yabuuchi, E. (1983). Transfer

of Peptococcus indolicus, Peptococcus asaccharolyticus, Peptococcus prevotii,

and Peptococcus magnus to the genus Peptostreptococcus and proposal of

Peptostreptococcus tetradius sp. nov. International Journal of Systematic

Bacteriology, 33(4), 683-698.

Faber, F., & Bäumler, A. J. (2014). The impact of intestinal inflammation on the

nutritional environment of the gut microbiota. Immunology Letters,

Fan, X., & Sellin, J. H. (2009). Review article: Small intestinal bacterial overgrowth, bile

acid malabsorption and gluten intolerance as possible causes of chronic watery

diarrhoea. Alimentary Pharmacology & Therapeutics, 29(10), 1069-1077.

doi:10.1111/j.1365-2036.2009.03970.x; 10.1111/j.1365-2036.2009.03970.x

Fasina, Y. O., Hoerr, F. J., McKee, S. R., & Conner, D. E. (2010). Influence of

Salmonella enterica serovar Typhimurium infection on intestinal goblet cells and

villous morphology in broiler chicks. Avian Diseases, 54(2), 841-847.

Feng, J., Liu, X., Xu, Z. R., Wang, Y. Z., & Liu, J. X. (2007). Effects of fermented

soybean meal on digestive enzyme activities and intestinal morphology in broilers.

Poultry Science, 86(6), 1149-1154.

135

Forder, R. E., Howarth, G. S., Tivey, D. R., & Hughes, R. J. (2007). Bacterial modulation

of small intestinal goblet cells and mucin composition during early posthatch

development of poultry. Poultry Science, 86(11), 2396-2403.

doi:10.3382/ps.2007-00222

Forder, R. E., Nattrass, G. S., Geier, M. S., Hughes, R. J., & Hynd, P. I. (2012).

Quantitative analyses of genes associated with mucin synthesis of broiler chickens

with induced necrotic enteritis. Poultry Science, 91(6), 1335-1341.

doi:10.3382/ps.2011-02062; 10.3382/ps.2011-02062

Fukunaga, T., Sasaki, M., Araki, Y., Okamoto, T., Yasuoka, T., Tsujikawa, T., . . .

Bamba, T. (2003). Effects of the soluble fibre pectin on intestinal cell

proliferation, fecal short chain fatty acid production and microbial population.

Digestion, 67(1-2), 42-49. doi:69705

Furuse, M., & Okumura, J. (1994). Nutritional and physiological characteristics in germ-

free chickens. Comparative Biochemistry and Physiology.Part A, Physiology,

109(3), 547-556.

Gabriel, I. L., M, & Mallet, S. G., JF. (2006). Microflora of the digestive tract: Critical

factors and consequences for poultry. World's Poultry Science Journal, 62(3),

499-512.

Gentry, T., Wickham, G., Schadt, C., He, Z., & Zhou, J. (2006). Microarray applications

in microbial ecology research. Microbial Ecology, 52(2), 159-175.

136

Ghareeb, K., Awad, W. A., Mohnl, M., Porta, R., Biarnes, M., Bohm, J., & Schatzmayr,

G. (2012). Evaluating the efficacy of an avian-specific probiotic to reduce the

colonization of Campylobacter jejuni in broiler chickens. Poultry Science, 91(8),

1825-1832. doi:10.3382/ps.2012-02168; 10.3382/ps.2012-02168

Gielda, L. M., & DiRita, V. J. (2012). Zinc competition among the intestinal microbiota.

Mbio, 3(4), e00171-12. doi:10.1128/mBio.00171-12; 10.1128/mBio.00171-12

Gilbert, E. R., Williams, P. M., Ray, W. K., Li, H., Emmerson, D. A., Wong, E. A., &

Webb, K. E.,Jr. (2010). Proteomic evaluation of chicken brush-border membrane

during the early posthatch period. Journal of Proteome Research, 9(9), 4628-4639.

doi:10.1021/pr1003533; 10.1021/pr1003533

Goldenfeld, N., & Woese, C. (2007). Biology's next revolution. Nature, 445(7126), 369.

doi:10.1038/445369a

Golder, H. M., Geier, M. S., Forder, R. E. A., Hynd, P. I., & Hughes, R. J. (2011). Effects

of necrotic enteritis challenge on intestinal micro-architecture and mucin profile.

British Poultry Science, 52(4), 500-506.

Gong, J., Forster, R. J., Yu, H., Chambers, J. R., Sabour, P. M., Wheatcroft, R., & Chen,

S. (2002a). Diversity and phylogenetic analysis of bacteria in the mucosa of

chicken ceca and comparison with bacteria in the cecal lumen. FEMS

Microbiology Letters, 208(1), 1-7.

137

Gong, J., Forster, R. J., Yu, H., Chambers, J. R., Wheatcroft, R., Sabour, P. M., & Chen,

S. (2002b). Molecular analysis of bacterial populations in the ileum of broiler

chickens and comparison with bacteria in the cecum. FEMS Microbiology

Ecology, 41(3), 171-179.

Gong, J., Si, W., Forster, R. J., Huang, R., Yu, H., Yin, Y., . . . Han, Y. (2007). 16S

rRNA gene-based analysis of mucosa-associated bacterial community and

phylogeny in the chicken gastrointestinal tracts: From crops to ceca. FEMS

Microbiology Ecology, 59(1), 147-157.

Grant, C. C., Konkel, M. E., Cieplak, W.,Jr, & Tompkins, L. S. (1993). Role of flagella in

adherence, internalization, and translocation of Campylobacter jejuni in

nonpolarized and polarized epithelial cell cultures. Infection and Immunity, 61(5),

1764-1771.

Gunal, M., Yayli, G., Kaya, O., Karahan, N., & Sulak, O. (2006). The effects of antibiotic

growth promoter, probiotic or organic acid supplementation on performance,

intestinal microflora and tissue of broilers. International Journal of Poultry

Science, 5(2), 149-155.

Haghighi, H. R., Gong, J., Gyles, C. L., Hayes, M. A., Sanei, B., Parvizi, P., . . . Sharif, S.

(2005). Modulation of antibody-mediated immune response by probiotics in

chickens. Clinical and Diagnostic Laboratory Immunology, 12(12), 1387-1392.

doi:10.1128/CDLI.12.12.1387-1392.2005

138

Haghighi, H. R., Gong, J., Gyles, C. L., Hayes, M. A., Zhou, H., Sanei, B., . . . Sharif, S.

(2006). Probiotics stimulate production of natural antibodies in chickens. Clinical

and Vaccine Immunology : CVI, 13(9), 975-980. doi:10.1128/CVI.00161-06

Haghighi, H. R., Abdul-Careem, M. F., Dara, R. A., Chambers, J. R., & Sharif, S. (2008).

Cytokine gene expression in chicken cecal tonsils following treatment with

probiotics and Salmonella infection. Veterinary Microbiology, 126(1-3), 225-233.

doi:10.1016/j.vetmic.2007.06.026

Hammons, S., Oh, P. L., Martínez, I., Clark, K., Schlegel, V. L., Sitorius, E., . . . Walter,

J. (2010). A small variation in diet influences the Lactobacillus strain composition

in the crop of broiler chickens. Systematic and Applied Microbiology, 33(5), 275-

281.

Hansson, G. C., & Johansson, M. E. (2010). The inner of the two Muc2 mucin-dependent

mucus layers in colon is devoid of bacteria. Gut Microbes, 1(1), 51-54.

doi:10.4161/gmic.1.1.10470

Hermans, D., Pasmans, F., Messens, W., Martel, A., Van Immerseel, F., Rasschaert,

G., . . . Haesebrouck, F. (2012). Poultry as a host for the zoonotic pathogen

Campylobacter jejuni. Vector Borne and Zoonotic Diseases (Larchmont, N.Y.),

12(2), 89-98. doi:10.1089/vbz.2011.0676; 10.1089/vbz.2011.0676

139

Hett, E. C., & Rubin, E. J. (2008). Bacterial growth and cell division: a Mycobacterial

perspective. Microbiology and Molecular Biology Reviews : MMBR, 72(1), 126-

56, table of contents. doi:10.1128/MMBR.00028-07 [doi]

Hinton Jr, A., Corrier, D., & DeLoach, J. (1992). In vitro inhibition of Salmonella

Typhimurium and escherichia coli O157: H7 by an anaerobic gram-positive

coccus isolated from the cecal contents of adult chickens. Journal of Food

Protection, 55

Hollister, A., Corrier, D., & Nisbet, D. (1999). Effects of chicken-derived cecal

microorganisms maintained in continuous culture on cecal colonization by

Salmonella Typhimurium in turkey poults. Poultry Science, 78(4), 546-549.

Holmgren, L. (2010). Horizontal gene transfer: You are what you eat. Biochemical and

Biophysical Research Communications, 396(1), 147-151.

doi:10.1016/j.bbrc.2010.04.026; 10.1016/j.bbrc.2010.04.026

Hooper, L. V., Midtvedt, T., & Gordon, J. I. (2002). How host-microbial interactions

shape the nutrient environment of the mammalian intestine. Annual Review of

Nutrition, 22(1), 283-307.

Huang, S., Yang, F., Zeng, X., Chen, J., Li, R., Wen, T., . . . Xu, J. (2011). Preliminary

characterization of the oral microbiota of chinese adults with and without

gingivitis. BMC Oral Health, 11, 33-6831-11-33. doi:10.1186/1472-6831-11-33

[doi]

140

Hughes, J. B., Hellmann, J. J., Ricketts, T. H., & Bohannan, B. J. (2001). Counting the

uncountable: Statistical approaches to estimating microbial diversity. Applied and

Environmental Microbiology, 67(10), 4399-4406.

Hughes, R. J. (2008). Relationship between digesta transit time and apparent

metabolisable energy value of wheat in chickens. British Poultry Science, 49(6),

716-720.

Immerseel, F. V., Buck, J. D., Pasmans, F., Huyghebaert, G., Haesebrouck, F., &

Ducatelle, R. (2004). Clostridium perfringens in poultry: An emerging threat for

animal and public health. Avian Pathology, 33(6), 537-549.

Isolauri, E., da Costa Ribeiro, H., Gibson, G., Saavedra, J., Saliminen, S., Vanderhoof, J.,

& Varavithya, W. (2002). Functional foods and probiotics: Working group report

of the first world congress of pediatric gastroenterology, hepatology, and nutrition.

Journal of Pediatric Gastroenterology and Nutrition, 35, S106-S109.

Jenssen, H., Hamill, P., & Hancock, R. E. (2006). Peptide antimicrobial agents. Clinical

Microbiology Reviews, 19(3), 491-511. doi:10.1128/CMR.00056-05

Jia, W., Slominski, B., Bruce, H., Blank, G., Crow, G., & Jones, O. (2009). Effects of diet

type and enzyme addition on growth performance and gut health of broiler

chickens during subclinical Clostridium perfringens challenge. Poultry Science,

88(1), 132-140.

141

Johansson, K. R. (1948). Intestinal microflora of chickens. University of Wisconsin-

Madison.

Johnson, T. J., Thorsness, J. L., Anderson, C. P., Lynne, A. M., Foley, S. L., Han, J., . . .

Singer, R. S. (2010). Horizontal gene transfer of a ColV plasmid has resulted in a

dominant avian clonal type of Salmonella enterica serovar Kentucky. PloS One,

5(12), e15524. doi:10.1371/journal.pone.0015524; 10.1371/journal.pone.0015524

Jung, S., Houde, R., Baurhoo, B., Zhao, X., & Lee, B. (2008). Effects of galacto-

oligosaccharides and a bifidobacteria lactis-based probiotic strain on the growth

performance and fecal microflora of broiler chickens. Poultry Science, 87(9),

1694-1699.

Kabir, S. (2009). The role of probiotics in the poultry industry. International Journal of

Molecular Sciences, 10(8), 3531-3546.

Kaldhusdal, M., & Lovland, A. (2000). The economical impact of Clostridium

perfringens is greater than anticipated. World Poult, 16(8), 50-51.

Kaldhusdal, M., Schneitz, C., Hofshagen, M., & Skjerve, E. (2001). Reduced incidence

of Clostridium perfringens-associated lesions and improved performance in

broiler chickens treated with normal intestinal bacteria from adult fowl. Avian

Diseases, , 149-156.

Kang, S., Denman, S. E., Morrison, M., Yu, Z., Dore, J., Leclerc, M., & McSweeney, C.

S. (2010). Dysbiosis of fecal microbiota in crohn's disease patients as revealed by

142

a custom phylogenetic microarray. Inflammatory Bowel Diseases, 16(12), 2034-

2042.

Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S., . . .

Drummond, A. (2012). Geneious basic: An integrated and extendable desktop

software platform for the organization and analysis of sequence data.

Bioinformatics (Oxford, England), 28(12), 1647-1649.

doi:10.1093/bioinformatics/bts199 [doi]

Killer, J., & Marounek, M. (2011). Fermentation of mucin by bifidobacteria from rectal

samples of humans and rectal and intestinal samples of animals. Folia

Microbiologica, 56(2), 85-89. doi:10.1007/s12223-011-0022-4; 10.1007/s12223-

011-0022-4

Kim, G., Seo, Y., Kim, C., & Paik, I. (2011). Effect of dietary prebiotic supplementation

on the performance, intestinal microflora, and immune response of broilers.

Poultry Science, 90(1), 75-82.

Kim, M. S. (2011). An Integrated Investigation of Ruminal Microbial Communities using

16S rRNA Gene-Based Techniques (Doctoral dissertation, The Ohio State

University).

Kim, M., Morrison, M., & Yu, Z. (2011). Evaluation of different partial 16S rRNA gene

sequence regions for phylogenetic analysis of microbiomes. Journal of

Microbiological Methods, 84(1), 81-87.

143

Kim, M., & Yu, Z. (2014). Variations in 16S rRNA-based microbiome profiling between

pyrosequencing runs and between pyrosequencing facilities. Journal of

Microbiology, 52(5), 355-365.

Knap, I., Lund, B., Kehlet, A. B., Hofacre, C., & Mathis, G. (2010). Bacillus

licheniformis prevents necrotic enteritis in broiler chickens. Avian Diseases, 54(2),

931-935.

Knarreborg, A., Simon, M. A., Engberg, R. M., Jensen, B. B., & Tannock, G. W. (2002).

Effects of dietary fat source and subtherapeutic levels of antibiotic on the bacterial

community in the ileum of broiler chickens at various ages. Applied and

Environmental Microbiology, 68(12), 5918-5924.

Koenen, M. E., Kramer, J., van der Hulst, R., Heres, L., Jeurissen, S. H., & Boersma, W.

J. (2004). Immunomodulation by probiotic lactobacilli in layer- and meat-type

chickens. British Poultry Science, 45(3), 355-366.

doi:10.1080/00071660410001730851

Kolida, S., Tuohy, K., & Gibson, G. R. (2000). The human gut flora in nutrition and

approaches for its dietary modulation. Nutrition Bulletin, 25(3), 223-231.

Kollanoor-Johny, A., Mattson, T., Baskaran, S. A., Amalaradjou, M. A., Babapoor, S.,

March, B., . . . Schreiber, D. (2012). Reduction of Salmonella enterica serovar

Enteritidis colonization in 20-day-old broiler chickens by the plant-derived

144

compounds trans-cinnamaldehyde and eugenol. Applied and Environmental

Microbiology, 78(8), 2981-2987.

Koutsos, E., & Arias, V. (2006). Intestinal ecology: Interactions among the

gastrointestinal tract, nutrition, and the microflora. The Journal of Applied Poultry

Research, 15(1), 161-173.

Kozich, J. J., Westcott, S. L., Baxter, N. T., Highlander, S. K., & Schloss, P. D. (2013).

Development of a dual-index sequencing strategy and curation pipeline for

analyzing amplicon sequence data on the MiSeq illumina sequencing platform.

Applied and Environmental Microbiology, 79(17), 5112-5120.

doi:10.1128/AEM.01043-13 [doi]

Krause, D. O., Dalrymple, B. P., Smith, W. J., Mackie, R. I., & McSweeney, C. S. (1999).

16S rDNA sequencing of Ruminococcus albus and Ruminococcus flavefaciens:

Design of a signature probe and its application in adult sheep. Microbiology

(Reading, England), 145 ( Pt 7)(Pt 7), 1797-1807.

Kunin, V., Engelbrektson, A., Ochman, H., & Hugenholtz, P. (2010). Wrinkles in the rare

biosphere: Pyrosequencing errors can lead to artificial inflation of diversity

estimates. Environmental Microbiology, 12(1), 118-123.

Kurata, S., Kanagawa, T., Magariyama, Y., Takatsu, K., Yamada, K., Yokomaku, T., &

Kamagata, Y. (2004). Reevaluation and reduction of a PCR bias caused by

145

reannealing of templates. Applied and Environmental Microbiology, 70(12),

7545-7549. doi:70/12/7545 [pii]

Kuribayashi, H., Miyata, M., Yamakawa, H., Yoshinari, K., & Yamazoe, Y. (2012).

Enterobacteria-mediated deconjugation of taurocholic acid enhances ileal

farnesoid X receptor signaling. European Journal of Pharmacology, 697(1-3),

132-138. doi:10.1016/j.ejphar.2012.09.048; 10.1016/j.ejphar.2012.09.048

Lan, Y., Verstegen, M., Tamminga, S., & Williams, B. (2005). The role of the

commensal gut microbial community in broiler chickens. World's Poultry Science

Journal, 61(01), 95-104.

Lane, D. J., Pace, B., Olsen, G. J., Stahl, D. A., Sogin, M. L., & Pace, N. R. (1985).

Rapid determination of 16S ribosomal RNA sequences for phylogenetic analyses.

Proceedings of the National Academy of Sciences of the United States of America,

82(20), 6955-6959.

Lawley, T. D., & Walker, A. W. (2013). Intestinal colonization resistance. Immunology,

138(1), 1-11. doi:10.1111/j.1365-2567.2012.03616.x; 10.1111/j.1365-

2567.2012.03616.x

Le Blay, G., Blottiere, H. M., Ferrier, L., Le Foll, E., Bonnet, C., Galmiche, J. P., &

Cherbut, C. (2000). Short-chain fatty acids induce cytoskeletal and extracellular

protein modifications associated with modulation of proliferation on primary

146

culture of rat intestinal smooth muscle cells. Digestive Diseases and Sciences,

45(8), 1623-1630.

LeBlanc, J. G., Milani, C., de Giori, G. S., Sesma, F., van Sinderen, D., & Ventura, M.

(2013). Bacteria as vitamin suppliers to their host: A gut microbiota perspective.

Current Opinion in Biotechnology, 24(2), 160-168.

doi:10.1016/j.copbio.2012.08.005; 10.1016/j.copbio.2012.08.005

Lee, D. H., Zo, Y. G., & Kim, S. J. (1996). Nonradioactive method to study genetic

profiles of natural bacterial communities by PCR-single-strand-conformation

polymorphism. Applied and Environmental Microbiology, 62(9), 3112-3120.

Lei, F., Yin, Y., Wang, Y., Deng, B., Yu, H. D., Li, L., . . . Wang, X. (2012). Higher-

level production of volatile fatty acids in vitro by chicken gut microbiotas than by

human gut microbiotas as determined by functional analyses. Applied and

Environmental Microbiology, 78(16), 5763-5772.

Lin, J., Hunkapiller, A. A., Layton, A. C., Chang, Y., & Robbins, K. R. (2013). Response

of intestinal microbiota to antibiotic growth promoters in chickens. Foodborne

Pathogens and Disease, 10(4), 331-337.

Liu, W. T., Marsh, T. L., Cheng, H., & Forney, L. J. (1997). Characterization of

microbial diversity by determining terminal restriction fragment length

polymorphisms of genes encoding 16S rRNA. Applied and Environmental

Microbiology, 63(11), 4516-4522.

147

Looft, T., Allen, H. K., Cantarel, B. L., Levine, U. Y., Bayles, D. O., Alt, D. P., . . .

Stanton, T. B. (2014). Bacteria, phages and pigs: The effects of in-feed antibiotics

on the microbiome at different gut locations. The ISME Journal, 8, 1566-1576.

Lozupone, C., & Knight, R. (2005). UniFrac: A new phylogenetic method for comparing

microbial communities. Applied and Environmental Microbiology, 71(12), 8228-

8235. doi:71/12/8228 [pii]

Lozupone, C. A., & Knight, R. (2008). Species divergence and the measurement of

microbial diversity. FEMS Microbiology Reviews, 32(4), 557-578.

Lu, J., Idris, U., Harmon, B., Hofacre, C., Maurer, J. J., & Lee, M. D. (2003). Diversity

and succession of the intestinal bacterial community of the maturing broiler

chicken. Applied and Environmental Microbiology, 69(11), 6816-6824.

Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H., Yadhukumar, . . . Schleifer,

K. H. (2004). ARB: A software environment for sequence data. Nucleic Acids

Research, 32(4), 1363-1371. doi:10.1093/nar/gkh293 [doi]

Lutful Kabir, S. M. (2009). The role of probiotics in the poultry industry. International

Journal of Molecular Sciences, 10(8), 3531-3546. doi:10.3390/ijms10083531;

10.3390/ijms10083531

Lyra, A., Lahtinen, S., Tiihonen, K., & Ouwehand, A. (2010). Intestinal microbiota and

overweight. Beneficial Microbes, 1(4), 407-421.

148

Margulies, M., Egholm, M., Altman, W. E., Attiya, S., Bader, J. S., Bemben, L. A., . . .

Chen, Z. (2005). Genome sequencing in microfabricated high-density picolitre

reactors. Nature, 437(7057), 376-380.

McDevitt, R., Brooker, J., Acamovic, T., & Sparks, N. (2006). Necrotic enteritis; a

continuing challenge for the poultry industry. World's Poultry Science Journal,

62(2), 221-248.

McReynolds, J., Moore, R., McElroy, A., Hargis, B., & Caldwell, D. (2007). Evaluation

of a competitive exclusion culture and megan vac 1 on Salmonella Typhimurium

colonization in neonatal broiler chickens. The Journal of Applied Poultry

Research, 16(3), 456-463.

Messaoudi, S., Kergourlay, G., Dalgalarrondo, M., Choiset, Y., Ferchichi, M., Prevost,

H., . . . Dousset, X. (2012). Purification and characterization of a new bacteriocin

active against Campylobacter produced by Lactobacillus salivarius SMXD51.

Food Microbiology, 32(1), 129-134. doi:10.1016/j.fm.2012.05.002;

10.1016/j.fm.2012.05.002

Metges, C. C. (2000). Contribution of microbial amino acids to amino acid homeostasis

of the host. The Journal of Nutrition, 130(7), 1857S-64S.

Meyer, F., Paarmann, D., D'Souza, M., Olson, R., Glass, E. M., Kubal, M., . . . Edwards,

R. A. (2008). The metagenomics RAST server - a public resource for the

149

automatic phylogenetic and functional analysis of metagenomes. BMC

Bioinformatics, 9, 386-2105-9-386. doi:10.1186/1471-2105-9-386 [doi]

Milligan, G. W., & Cheng, R. (1996). Measuring the influence of individual data points

in a cluster analysis. Journal of Classification, 13(2), 315-335.

Mitsch, P., Zitterl-Eglseer, K., Kohler, B., Gabler, C., Losa, R., & Zimpernik, I. (2004).

The effect of two different blends of essential oil components on the proliferation

of Clostridium perfringens in the intestines of broiler chickens. Poultry Science,

83(4), 669-675.

Miyata, M., Yamakawa, H., Hamatsu, M., Kuribayashi, H., Takamatsu, Y., & Yamazoe,

Y. (2011). Enterobacteria modulate intestinal bile acid transport and homeostasis

through apical sodium-dependent bile acid transporter (SLC10A2) expression.

The Journal of Pharmacology and Experimental Therapeutics, 336(1), 188-196.

doi:10.1124/jpet.110.171736; 10.1124/jpet.110.171736

Molnár, A. K., Podmaniczky, B., Kurti, P., Tenk, I., Glavits, R., Virag, G., & Szabo, Z.

(2011). Effect of different concentrations of Bacillus subtilis on growth

performance, carcase quality, gut microflora and immune response of broiler

chickens. British Poultry Science, 52(6), 658-665.

doi:10.1080/00071668.2011.636029; 10.1080/00071668.2011.636029

150

Morrison, M., & Miron, J. (2000). Adhesion to cellulose by Ruminococcus albus: A

combination of cellulosomes and Pil-proteins? 1. FEMS Microbiology Letters,

185(2), 109-115.

Mountzouris, K., Tsitrsikos, P., Palamidi, I., Arvaniti, A., Mohnl, M., Schatzmayr, G., &

Fegeros, K. (2010). Effects of probiotic inclusion levels in broiler nutrition on

growth performance, nutrient digestibility, plasma immunoglobulins, and cecal

microflora composition. Poultry Science, 89(1), 58-67.

Muramatsu, T., Takasu, O., & Okumura, J. (1993). Research note: Fructose feeding

increases lower gut weights in germ-free and conventional chicks. Poultry

Science, 72(8), 1597-1600.

Murray, R. (1984). The higher taxa, or, a place for everything. Bergey's Manual of

Systematic Bacteriology, 1, 31-34.

Murry Jr, A., Hinton Jr, A., & Morrison, H. (2004). Inhibition of growth of Escherichia

coli, Salmonella Typhimurium, and Clostridia perfringens on chicken feed media

by Lactobacillus salivarius and Lactobacillus plantarum. International Journal of

Poultry Science, 3(9), 603-607.

Muyzer, G., de Waal, E. C., & Uitterlinden, A. G. (1993). Profiling of complex microbial

populations by denaturing gradient gel electrophoresis analysis of polymerase

chain reaction-amplified genes coding for 16S rRNA. Applied and Environmental

Microbiology, 59(3), 695-700.

151

Muyzer, G., & Smalla, K. (1998). Application of denaturing gradient gel electrophoresis

(DGGE) and temperature gradient gel electrophoresis (TGGE) in microbial

ecology. Antonie Van Leeuwenhoek, 73(1), 127-141.

Mwangi, W. N., Beal, R. K., Powers, C., Wu, X., Humphrey, T., Watson, M., . . . Smith,

A. L. (2010). Regional and global changes in TCRalphabeta T cell repertoires in

the gut are dependent upon the complexity of the enteric microflora.

Developmental and Comparative Immunology, 34(4), 406-417.

doi:10.1016/j.dci.2009.11.009; 10.1016/j.dci.2009.11.009

Nakazato, G., Campos, T. A. d., Stehling, E. G., Brocchi, M., & Silveira, W. D. d. (2009).

Virulence factors of avian pathogenic Escherichia coli (APEC). Pesquisa

Veterinária Brasileira, 29(7), 479-486.

Nakphaichit, M., Thanomwongwattana, S., Phraephaisarn, C., Sakamoto, N.,

Keawsompong, S., Nakayama, J., & Nitisinprasert, S. (2011). The effect of

including Lactobacillus reuteri KUB-AC5 during post-hatch feeding on the

growth and ileum microbiota of broiler chickens. Poultry Science, 90(12), 2753-

2765.

Ng, S., Hart, A., Kamm, M., Stagg, A., & Knight, S. (2009). Mechanisms of action of

probiotics: Recent advances. Inflammatory Bowel Diseases, 15(2), 300-310.

152

Nisbet, D. (2002). Defined competitive exclusion cultures in the prevention of

enteropathogen colonisation in poultry and swine. Antonie Van Leeuwenhoek,

81(1-4), 481-486.

Nocker, A., Burr, M., & Camper, A. K. (2007). Genotypic microbial community profiling:

A critical technical review. Microbial Ecology, 54(2), 276-289.

NRC. (1994). Nutrient requirements of poultry. National Research Council. National

Academy Press, Washington.

Ohara, H., Karita, S., Kimura, T., Sakka, K., & Ohmiya, K. (2000). Characterization of

the cellulolytic complex (cellulosome) from Ruminococcus albus. Bioscience,

Biotechnology, and Biochemistry, 64(2), 254-260.

O'Hara, A. M., O'Regan, P., Fanning, Á., O'Mahony, C., MacSharry, J., Lyons, A., . . .

Shanahan, F. (2006). Functional modulation of human intestinal epithelial cell

responses by bifidobacterium infantis and lactobacillus salivarius. Immunology,

118(2), 202-215.

O'Hara, A. M., & Shanahan, F. (2006). The gut flora as a forgotten organ. EMBO

Reports, 7(7), 688-693.

Owens, B., Tucker, L., Collins, M., & McCracken, K. (2008). Effects of different feed

additives alone or in combination on broiler performance, gut microflora and ileal

histology. British Poultry Science, 49(2), 202-212.

153

Palmer, M. F., & Rolls, B. A. (1983). The activities of some metabolic enzymes in the

intestines of germ-free and conventional chicks. The British Journal of Nutrition,

50(3), 783-790.

Pan, D., & Yu, Z. (2013). Intestinal microbiome of poultry and its interaction with host

and diet. Gut Microbes, 5(1), 0-1.

Park, P. J., Manjourides, J., Bonetti, M., & Pagano, M. (2009). A permutation test for

determining significance of clusters with applications to spatial and gene

expression data. Computational Statistics & Data Analysis, 53(12), 4290-4300.

Pascual, M., Hugas, M., Badiola, J. I., Monfort, J. M., & Garriga, M. (1999).

Lactobacillus salivarius CTC2197 prevents Salmonella Enteritidis colonization in

chickens. Applied and Environmental Microbiology, 65(11), 4981-4986.

Patzer, S. I., & Hantke, K. (2000). The zinc-responsive regulator zur and its control of the

znu gene cluster encoding the ZnuABC zinc uptake system in Escherichia coli.

The Journal of Biological Chemistry, 275(32), 24321-24332.

doi:10.1074/jbc.M001775200

Prakash, O., Shouche, Y., Jangid, K., & Kostka, J. E. (2013). Microbial cultivation and

the role of microbial resource centers in the omics era. Applied Microbiology and

Biotechnology, 97(1), 51-62.

154

Pryde, S. E., Duncan, S. H., Hold, G. L., Stewart, C. S., & Flint, H. J. (2002). The

microbiology of butyrate formation in the human colon. FEMS Microbiology

Letters, 217(2), 133-139.

Qiu, X., Wu, L., Huang, H., McDonel, P. E., Palumbo, A. V., Tiedje, J. M., & Zhou, J.

(2001). Evaluation of PCR-generated chimeras, mutations, and heteroduplexes

with 16S rRNA gene-based cloning. Applied and Environmental Microbiology,

67(2), 880-887. doi:10.1128/AEM.67.2.880-887.2001 [doi]

Quince, C., Lanzén, A., Curtis, T. P., Davenport, R. J., Hall, N., Head, I. M., . . . Sloan,

W. T. (2009). Accurate determination of microbial diversity from 454

pyrosequencing data. Nature Methods, 6(9), 639.

Rajilić-Stojanović, M., Heilig, H. G., Molenaar, D., Kajander, K., Surakka, A., Smidt, H.,

& De Vos, W. M. (2009). Development and application of the human intestinal

tract chip, a phylogenetic microarray: Analysis of universally conserved

phylotypes in the abundant microbiota of young and elderly adults.

Environmental Microbiology, 11(7), 1736-1751.

Rehman, H. U., Vahjen, W., Awad, W. A., & Zentek, J. (2007). Indigenous bacteria and

bacterial metabolic products in the gastrointestinal tract of broiler chickens.

Archives of Animal Nutrition, 61(5), 319-335.

155

Reysenbach, A. L., Giver, L. J., Wickham, G. S., & Pace, N. R. (1992). Differential

amplification of rRNA genes by polymerase chain reaction. Applied and

Environmental Microbiology, 58(10), 3417-3418.

Riddell, C., & Kong, X. (1992). The influence of diet on necrotic enteritis in broiler

chickens. Avian Diseases, , 499-503.

Rodríguez, M. L., Rebolé, A., Velasco, S., Ortiz, L. T., Treviño, J., & Alzueta, C. (2012).

Wheat-and barley-based diets with or without additives influence broiler chicken

performance, nutrient digestibility and intestinal microflora. Journal of the

Science of Food and Agriculture, 92(1), 184-190.

Rogosa, M. (1971). Peptococcaceae, a new family to include the gram-positive,

anaerobic cocci of the genera Peptococcus, Peptostreptococcus, and

Ruminococcus. International Journal of Systematic Bacteriology, 21(3), 234-237.

Ruas-Madiedo, P., Gueimonde, M., Fernandez-Garcia, M., de los Reyes-Gavilan, C. G.,

& Margolles, A. (2008). Mucin degradation by bifidobacterium strains isolated

from the human intestinal microbiota. Applied and Environmental Microbiology,

74(6), 1936-1940. doi:10.1128/AEM.02509-07; 10.1128/AEM.02509-07

Saeed, A. I., Bhagabati, N. K., Braisted, J. C., Liang, W., Sharov, V., Howe, E. A., . . .

Quackenbush, J. (2006). [9] TM4 microarray software suite. Methods in

Enzymology, 411, 134-193.

156

Salanitro, J. P., Fairchilds, I. G., & Zgornicki, Y. D. (1974). Isolation, culture

characteristics, and identification of anaerobic bacteria from the chicken cecum.

Applied Microbiology, 27(4), 678-687.

Salminen, S., Bouley, C., Boutron, M., Cummings, J., Franck, A., Gibson, G., . . .

Rowland, I. (1998). Functional food science and gastrointestinal physiology and

function. British Journal of Nutrition, 80(S1), S147-S171.

Sanderson, I. R. (2004). Short chain fatty acid regulation of signaling genes expressed by

the intestinal epithelium. The Journal of Nutrition, 134(9), 2450S-2454S.

Schefferle, H. E. (1966). Coryneform bacteria in poultry deep litter. Journal of Applied

Bacteriology, 29(1), 147-160.

Schloss, P. D., Westcott, S. L., Ryabin, T., Hall, J. R., Hartmann, M., Hollister, E. B., . . .

Weber, C. F. (2009). Introducing mothur: Open-source, platform-independent,

community-supported software for describing and comparing microbial

communities. Applied and Environmental Microbiology, 75(23), 7537-7541.

doi:10.1128/AEM.01541-09 [doi]

Schloss, P. D., & Westcott, S. L. (2011). Assessing and improving methods used in

operational taxonomic unit-based approaches for 16S rRNA gene sequence

analysis. Applied and Environmental Microbiology, 77(10), 3219-3226.

doi:10.1128/AEM.02810-10 [doi]

157

Scott, K. P., Martin, J. C., Campbell, G., Mayer, C. D., & Flint, H. J. (2006). Whole-

genome transcription profiling reveals genes up-regulated by growth on fucose in

the human gut bacterium "Roseburia inulinivorans". Journal of Bacteriology,

188(12), 4340-4349. doi:188/12/4340 [pii]

Scupham, A. J., Patton, T. G., Bent, E., & Bayles, D. O. (2008). Comparison of the cecal

microbiota of domestic and wild turkeys. Microbial Ecology, 56(2), 322-331.

Sekirov, I., Russell, S. L., Antunes, L. C., & Finlay, B. B. (2010). Gut microbiota in

health and disease. Physiological Reviews, 90(3), 859-904.

doi:10.1152/physrev.00045.2009 [doi]

Shakouri, M., Iji, P., Mikkelsen, L., & Cowieson, A. (2009). Intestinal function and gut

microflora of broiler chickens as influenced by cereal grains and microbial

enzyme supplementation. Journal of Animal Physiology and Animal Nutrition,

93(5), 647-658.

Shin, M. S., Han, S. K., Ji, A. R., Kim, K. S., & Lee, W. K. (2008). Isolation and

characterization of bacteriocin-producing bacteria from the gastrointestinal tract

of broiler chickens for probiotic use. Journal of Applied Microbiology, 105(6),

2203-2212. doi:10.1111/j.1365-2672.2008.03935.x; 10.1111/j.1365-

2672.2008.03935.x

Si, W., Gong, J., Han, Y., Yu, H., Brennan, J., Zhou, H., & Chen, S. (2007).

Quantification of cell proliferation and alpha-toxin gene expression of

158

Clostridium perfringens in the development of necrotic enteritis in broiler

chickens. Applied and Environmental Microbiology, 73(21), 7110-7113.

doi:AEM.01108-07 [pii]

Slominski, B. (2011). Recent advances in research on enzymes for poultry diets. Poultry

Science, 90(9), 2013-2023.

Soler, J. J., Martín-Vivaldi, M., Peralta-Sánchez, J. M., & Ruiz-Rodríguez, M. (2010).

Antibiotic-producing bacteria as a possible defence of birds against pathogenic

microorganisms. Open Ornithology Journal, 3, 93-100.

Sonmez, G., & Eren, M. (1999). Effects of supplementation of zinc bacitracin, mannan

oligosaccharide, and probiotic into the broiler feeds on morphology of the small

intestine. Vet.Fac.Dergisi Uludag Univ, 18, 125-138.

Sonnenburg, J. L., Angenent, L. T., & Gordon, J. I. (2004). Getting a grip on things: How

do communities of bacterial symbionts become established in our intestine?

Nature Immunology, 5(6), 569-573.

Spring, P., Wenk, C., Dawson, K., & Newman, K. (2000). The effects of dietary

mannaoligosaccharides on cecal parameters and the concentrations of enteric

bacteria in the ceca of Salmonella-challenged broiler chicks. Poultry Science,

79(2), 205-211.

Stanley, D., Denman, S. E., Hughes, R. J., Geier, M. S., Crowley, T. M., Chen, H., . . .

Moore, R. J. (2012). Intestinal microbiota associated with differential feed

159

conversion efficiency in chickens. Applied Microbiology and Biotechnology,

96(5), 1361-1369.

Stanley, D., Geier, M. S., Denman, S. E., Haring, V. R., Crowley, T. M., Hughes, R. J., &

Moore, R. J. (2013). Identification of chicken intestinal microbiota correlated with

the efficiency of energy extraction from feed. Veterinary Microbiology, 164(1-2),

85-92. doi:10.1016/j.vetmic.2013.01.030 [doi]

Stanley, V., Gray, C., Daley, M., Krueger, W., & Sefton, A. (2004). An alternative to

antibiotic-based drugs in feed for enhancing performance of broilers grown on

Eimeria spp.-infected litter. Poultry Science, 83(1), 39-44.

Stanley, D., Keyburn, A. L., Denman, S. E., & Moore, R. J. (2012). Changes in the caecal

microflora of chickens following Clostridium perfringens challenge to induce

necrotic enteritis. Veterinary Microbiology, 159(1), 155-162.

Stern, N. J., Svetoch, E. A., Eruslanov, B. V., Perelygin, V. V., Mitsevich, E. V.,

Mitsevich, I. P., . . . Seal, B. S. (2006). Isolation of a Lactobacillus salivarius

strain and purification of its bacteriocin, which is inhibitory to Campylobacter

jejuni in the chicken gastrointestinal system. Antimicrobial Agents and

Chemotherapy, 50(9), 3111-3116. doi:10.1128/AAC.00259-06

Strompfova, V., Laukova, A., Marcinakova, M., & Vasilkova, Z. (2010). Testing of

probiotic and bacteriocin-producing lactic acid bacteria towards Eimeria sp.

Polish Journal of Veterinary Sciences, 13(2), 389-391.

160

Sun, H., Tang, J. W., Yao, X. H., Wu, Y. F., 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.

Tropical Animal Health and Production, 45(4), 987-993. doi:10.1007/s11250-

012-0322-y; 10.1007/s11250-012-0322-y

Svetoch, E. A., Eruslanov, B. V., Levchuk, V. P., Perelygin, V. V., Mitsevich, E. V.,

Mitsevich, I. P., . . . Stern, N. J. (2011). Isolation of Lactobacillus salivarius 1077

(NRRL B-50053) and characterization of its bacteriocin, including the

antimicrobial activity spectrum. Applied and Environmental Microbiology, 77(8),

2749-2754. doi:10.1128/AEM.02481-10; 10.1128/AEM.02481-10

Tabatabai, L. B., & Walker, H. W. (1970). Oxidation-reduction potential and growth of

clostridium perfringens and pseudomonas fluorescens. Applied Microbiology,

20(3), 441-446.

Tellez, G., Higgins, S., Donoghue, A., & Hargis, B. (2006). Digestive physiology and the

role of microorganisms. The Journal of Applied Poultry Research, 15(1), 136-144.

Teo, A. Y., & Tan, H. M. (2005). Inhibition of Clostridium perfringens by a novel strain

of Bacillus subtilis isolated from the gastrointestinal tracts of healthy chickens.

Applied and Environmental Microbiology, 71(8), 4185-4190.

doi:10.1128/AEM.71.8.4185-4190.2005

161

Timbermont, L., Haesebrouck, F., Ducatelle, R., & Van Immerseel, F. (2011). Necrotic

enteritis in broilers: An updated review on the pathogenesis. Avian Pathology,

40(4), 341-347.

Torok, V. A., Ophel-Keller, K., Loo, M., & Hughes, R. J. (2008). Application of methods

for identifying broiler chicken gut bacterial species linked with increased energy

metabolism. Applied and Environmental Microbiology, 74(3), 783-791.

Torok, V., Ophel-Keller, K., Mikkelsen, L., Perez-Maldonado, R., Balding, K.,

MacAlpine, R., & Hughes, R. (2010). Characterisation of gut bacteria associated

with broiler performance across various australian feeding trials. Proceedings of

the 21st Annual Australian Poultry Science Sumposium, Sydney, New South

Wales, 1-3rd February 2010. 195-198.

Torok, V. A., Hughes, R. J., Mikkelsen, L. L., Perez-Maldonado, R., Balding, K.,

MacAlpine, R., . . . Ophel-Keller, K. (2011). Identification and characterization of

potential performance-related gut microbiotas in broiler chickens across various

feeding trials. Applied and Environmental Microbiology, 77(17), 5868-5878.

doi:10.1128/AEM.00165-11 [doi]

Tu, Q., He, Z., Li, Y., Chen, Y., Deng, Y., Lin, L., . . . Wu, L. (2014). Development of

HuMiChip for functional profiling of human microbiomes. PloS One, 9(3),

e90546.

162

Uni, Z., Noy, Y., & Sklan, D. (1999). Posthatch development of small intestinal function

in the poult. Poultry Science, 78(2), 215-222.

USDA. (2013). Poultry-Production and Value 2013 Summary. United States Department

of Agriculture, National Agricultural Statistics Service.

Van der Sluis, W. (2000). Clostridial enteritis is an often underestimated problem. World

Poultry, 16(7), 42-43.

Van der Wielen, P. W., Biesterveld, S., Notermans, S., Hofstra, H., Urlings, B. A., & van

Knapen, F. (2000). Role of volatile fatty acids in development of the cecal

microflora in broiler chickens during growth. Applied and Environmental

Microbiology, 66(6), 2536-2540.

Van Dijk, E. L., Auger, H., Jaszczyszyn, Y., & Thermes, C. (2014). Ten years of next-

generation sequencing technology. Trends in Genetics, 30(9), 418-426.

Van Immerseel, F., De Buck, J., De Smet, I., Mast, J., Haesebrouck, F., & Ducatelle, R.

(2002). Dynamics of immune cell infiltration in the caecal lamina propria of

chickens after neonatal infection with a Salmonella Enteritidis strain.

Developmental and Comparative Immunology, 26(4), 355-364.

Van Immerseel, F., De Buck, J., Pasmans, F., Huyghebaert, G., Haesebrouck, F., &

Ducatelle, R. (2004a). Clostridium perfringens in poultry: An emerging threat for

animal and public health. Avian Pathology, 33(6), 537-549.

163

Van Immerseel, F., Fievez, V., de Buck, J., Pasmans, F., Martel, A., Haesebrouck, F., &

Ducatelle, R. (2004b). Microencapsulated short-chain fatty acids in feed modify

colonization and invasion early after infection with Salmonella Enteritidis in

young chickens. Poultry Science, 83(1), 69-74.

Van Immerseel, F., Russell, J. B., Flythe, M. D., Gantois, I., Timbermont, L., Pasmans,

F., . . . Ducatelle, R. (2006). The use of organic acids to combat Salmonella in

poultry: A mechanistic explanation of the efficacy. Avian Pathology : Journal of

the W.V.P.A, 35(3), 182-188. doi:10.1080/03079450600711045

Van Immerseel, F., Rood, J. I., Moore, R. J., & Titball, R. W. (2009). Rethinking our

understanding of the pathogenesis of necrotic enteritis in chickens. Trends in

Microbiology, 17(1), 32-36.

Van Reenen, C. A., & Dicks, L. M. (2011). Horizontal gene transfer amongst probiotic

lactic acid bacteria and other intestinal microbiota: What are the possibilities? A

review. Archives of Microbiology, 193(3), 157-168. doi:10.1007/s00203-010-

0668-3; 10.1007/s00203-010-0668-3

Vispo, C., & Karasov, W. H. (1997). The interaction of avian gut microbes and their host:

An elusive symbiosis. Gastrointestinal microbiology (pp. 116-155) Springer.

Wang, Z., Qiao, S., Lu, W., & Li, D. (2005). Effects of enzyme supplementation on

performance, nutrient digestibility, gastrointestinal morphology, and volatile fatty

164

acid profiles in the hindgut of broilers fed wheat-based diets. Poultry Science,

84(6), 875-881.

Wang, Q., Garrity, G. M., Tiedje, J. M., & Cole, J. R. (2007). Naive bayesian classifier

for rapid assignment of rRNA sequences into the new bacterial taxonomy.

Applied and Environmental Microbiology, 73(16), 5261-5267. doi:AEM.00062-

07 [pii]

Wei, S. (2013). Towards a Better Understanding of Poultry Intestinal Microbiome

through Metagenomic and Microarray Studies (Doctoral dissertation, The Ohio

State University).

Wei, S., Gutek, A., Lilburn, M., & Yu, Z. (2013a). Abundance of pathogens in the gut

and litter of broiler chickens as affected by bacitracin and litter management.

Veterinary Microbiology,

Wei, S., Morrison, M., & Yu, Z. (2013b). Bacterial census of poultry intestinal

microbiome. Poultry Science, 92(3), 671-683.

Wexler, H. M. (2007). Bacteroides: The good, the bad, and the nitty-gritty. Clinical

Microbiology Reviews, 20(4), 593-621. doi:20/4/593 [pii]

White, J. R., Nagarajan, N., & Pop, M. (2009). Statistical methods for detecting

differentially abundant features in clinical metagenomic samples. PLoS

Computational Biology, 5(4), e1000352.

165

Whittaker, R. H. (1960). Vegetation of the siskiyou mountains, oregon and california.

Ecological Monographs, 30(3), 279-338.

Whittaker, R. H. (1972). Evolution and measurement of species diversity. Taxon, , 213-

251.

Wimalarathna, H. M., Richardson, J. F., Lawson, A. J., Elson, R., Meldrum, R., Little, C.

L., . . . Sheppard, S. K. (2013). Widespread acquisition of antimicrobial resistance

among Campylobacter isolates from UK retail poultry and evidence for clonal

expansion of resistant lineages. BMC Microbiology, 13(1), 160.

doi:10.1186/1471-2180-13-160

Wladyka, B., Wielebska, K., Wloka, M., Bochenska, O., Dubin, G., Dubin, A., & Mak, P.

(2012). Isolation, biochemical characterization, and cloning of a bacteriocin from

the poultry-associated Staphylococcus aureus strain CH-91. Applied

Microbiology and Biotechnology, doi:10.1007/s00253-012-4578-y

Wright, E. S., Yilmaz, L. S., & Noguera, D. R. (2012). DECIPHER, a search-based

approach to chimera identification for 16S rRNA sequences. Applied and

Environmental Microbiology, 78(3), 717-725. doi:10.1128/AEM.06516-11 [doi]

Xu, J., & Gordon, J. I. (2003). Honor thy symbionts. Proceedings of the National

Academy of Sciences of the United States of America, 100(18), 10452-10459.

doi:10.1073/pnas.1734063100 [doi]

166

Xu, Z. R., Hu, C. H., Xia, M. S., Zhan, X. A., & Wang, M. Q. (2003). Effects of dietary

fructooligosaccharide on digestive enzyme activities, intestinal microflora and

morphology of male broilers. Poultry Science, 82(6), 1030-1036.

Xu, J., Chiang, H. C., Bjursell, M. K., & Gordon, J. I. (2004). Message from a human gut

symbiont: Sensitivity is a prerequisite for sharing. Trends in Microbiology, 12(1),

21-28.

Yang, C. M., Cao, G. T., Ferket, P. R., Liu, T. T., Zhou, L., Zhang, L., . . . Chen, A. G.

(2012). Effects of probiotic, Clostridium butyricum, on growth performance,

immune function, and cecal microflora in broiler chickens. Poultry Science, 91(9),

2121-2129. doi:10.3382/ps.2011-02131; 10.3382/ps.2011-02131

You, Y., Hilpert, M., & Ward, M. J. (2013). Identification of Tet45, a tetracycline efflux

pump, from a poultry-litter-exposed soil isolate and persistence of tet(45) in the

soil. The Journal of Antimicrobial Chemotherapy, doi:10.1093/jac/dkt127

Yu, Z., & Morrison, M. (2004). Improved extraction of PCR-quality community DNA

from digesta and fecal samples. Biotechniques, 36(5), 808-813.

Yue, J. C., & Clayton, M. K. (2005). A similarity measure based on species proportions.

Communications in Statistics-Theory and Methods, 34(11), 2123-2131.

Zengler, K., Toledo, G., Rappe, M., Elkins, J., Mathur, E. J., Short, J. M., & Keller, M.

(2002). Cultivating the uncultured. Proceedings of the National Academy of

167

Sciences of the United States of America, 99(24), 15681-15686.

doi:10.1073/pnas.252630999 [doi]

Zhang, J., Finney, R. P., Clifford, R. J., Derr, L. K., & Buetow, K. H. (2005). Detecting

false expression signals in high-density oligonucleotide arrays by an in silico

approach. Genomics, 85(3), 297-308.

Zhou, H., Gong, J., Brisbin, J., Yu, H., Sanei, B., Sabour, P., & Sharif, S. (2007).

Appropriate chicken sample size for identifying the composition of broiler

intestinal microbiota affected by dietary antibiotics, using the polymerase chain

reaction-denaturing gradient gel electrophoresis technique. Poultry Science,

86(12), 2541-2549.

Zhu, X. Y., Zhong, T., Pandya, Y., & Joerger, R. D. (2002). 16S rRNA-based analysis of

microbiota from the cecum of broiler chickens. Applied and Environmental

Microbiology, 68(1), 124-137.

Zoetendal, E. G., Collier, C. T., Koike, S., Mackie, R. I., & Gaskins, H. R. (2004).

Molecular ecological analysis of the gastrointestinal microbiota: A review. The

Journal of Nutrition, 134(2), 465-472.

168