André Luis Alves Neves 6

Elucidating the role of the rumen microbiome in cattle feed efficiency and its 1 potential as a reservoir for novel enzyme discovery 2 3 by 4 5 André Luis Alves Neves 6

7 8 9 10 11 12 A thesis submitted in partial fulfillment of the requirements for the degree of 13

14 15 Doctor of Philosophy 16

17 in 18 19 Animal Science 20 21 22 23 24 25 Department of Agricultural, Food and Nutritional Science 26

University of Alberta 27

28 29 30 31 32 33 34 35 36 37 © André Luis Alves Neves, 2019 38

39 40 Abstract 1 2

The rapid advances in omics technologies have led to a tremendous progress in our 3 understanding of the rumen microbiome and its influence on cattle feed efficiency. 4

However, significant gaps remain in the literature concerning the driving forces that 5 influence the relationship between the rumen microbiota and host individual variation, and 6 how their interactive effects on animal productivity contribute to the identification of cattle 7 with improved feed efficiency. Furthermore, little is known about the impact of mRNA- 8 based metatranscriptomics on the analysis of rumen taxonomic profiles, and a strategy 9 for the discovery of lignocellulolytic enzymes through the targeted functional profiling of 10 carbohydrate-active enzymes (CAZymes) remains to be developed. Study 1 investigated 11 the dynamics of rumen microorganisms in cattle raised under different feeding regimens 12

(forage vs. grain) and studied the relationship among the abundance of these 13 microorganisms, host individuality and the diet. To examine host individual variation in 14 the rumen microbial abundance following dietary switches, hosts were grouped based on 15 the magnitude of microbial population shift using log2-fold change (log2-fc) in the copy 16 numbers of bacteria, archaea, protozoa and fungi. Three groups of log2-fc in the bacterial 17 and fungal abundance (Low, log2-fc < -1; Stable, -1 < log2-fc < 1; and High, log2-fc > 1) 18 were identified from the magnitude of change in baseline rumen microbial populations. 19

By monitoring the microbial population shift within the same animal in response to the 20 diet, significant ecological features of rumen microorganisms were identified and shed 21 new light on their dynamic roles in animal feed utilization and individual variation. Study 22

2 compared the outcomes of two methods, Kraken (mRNA based) and a pipeline 23 developed in-house based on Mothur (16S rRNA based), concerning the taxonomic 24

ii profiles (bacteria and archaea) of rumen microbial communities using total RNA 1 sequencing of rumen fluid collected from cattle with different feed conversion ratios (FCR). 2

Both approaches revealed a similar phyla distribution of the most abundant taxa, with 3

Bacteroidetes, Firmicutes, and Proteobacteria accounting for approximately 80% of total 4 bacterial abundance. For bacterial taxa, although 69 genera were commonly detected by 5 both methods, an additional 159 genera were exclusively identified by Kraken. Kraken 6 detected 423 species, while Mothur was not able to assign bacterial sequences to the 7 species level. For archaea, both methods generated similar results only for the 8 abundance of Methanomassiliicoccaceae and Methanobrevibacter ruminantium. 9

Although Kraken enhanced the microbial classification at the species level, identification 10 of bacteria or archaea in the rumen was limited due to a lack of reference genomes for 11 the rumen microbiome. Study 3 investigated the effect of cattle breeds on specific ruminal 12 taxonomic microbial groups and functions associated with FCR, using two genetically 13 related Angus breeds as a model. Total RNA was extracted from rumen content samples 14 collected from purebred Black and Red Angus bulls fed the same forage diet and then 15 subjected to metatranscriptomic analysis. Multivariate discriminant analysis (sPLS-DA) 16 and analysis of composition of microbiomes (ANCOM) were conducted to identify 17 microbial signatures characterizing Black and Red Angus cattle. Although Black and Red 18

Angus are genetically similar, sPLS-DA detected 25 bacterial species and ten functions 19 that differentiated the rumen microbial signatures between those two breeds. ANCOM 20 revealed an association between FCR and breed with Chitinophaga pinensis and 21

Clostridium stercorarium, suggesting that these bacterial species may play a key role in 22 the feed conversion efficiency of forage-fed bulls. Study 4 combined selective pressure 23

iii to enrich the rumen for lignocellulolytic microbes with bioinformatic tools to guide the 1 discovery of unknown CAZymes in the microbiome. It was demonstrated that the rumen 2 microbiome increased the abundance of lignocellulolytic bacteria, such as Fibrobacter 3 succinogens, and a diverse set of CAZymes over time, including 18 uncharacterized 4 members of the family GH11 (xylanases) and three of the family GH45 (endoglucanases). 5

Further experiments confirmed the lignocellulolytic activity of xylanase using such 6 approach. In summary, the data presented in this thesis provide fundamental knowledge 7 on the role of the rumen microbiome in cattle feed efficiency and offers opportunities to 8 further explore the potential of the rumen as a source for novel enzyme discovery. 9

18 19

Preface 1

2 This thesis is an original work by Andre Luis Alves Neves and is part of a 3 collaborative project between Profs. Leluo Guan at the University of Alberta and Prof. Kim 4 H. Ominski of the University of Manitoba. The Veterinary Services and the Animal Care 5 Committee of the University of Manitoba granted ethical approval for all experimental 6 procedures described in Chapters 2, 3, 4 and 5. 7

Chapter 2 of this thesis has been submitted to FEMS Microbiology Ecology as 8 “Dynamics of microbial populations driven by interactions between diet and host shed 9 light on individualized rumen microbiota” by Andre L. A. Neves, Yanghong Chen, Eoin 10 O’Hara, Tim McAllister, Kim H. Ominski and Le Luo Guan. (2019). ALAN was responsible 11 for laboratory and data analysis, as well as manuscript writing. YC performed the 12 laboratory analysis, while LLG, KHO, TM, and EOH revised and contributed to the writing 13 of the manuscript. 14

Chapter 3 of the thesis has been published as Andre L. A. Neves, Fuyong Li, 15 Bibaswan Ghoshal, Tim McAllister and Le Luo (2017). “Enhancing the resolution of rumen 16 microbial classification from metatranscriptomic data using Kraken and Mothur”. Frontiers 17 Microbiology; 8: 2445. doi: 10.3389/fmicb.2017.02445. ALAN and BG executed Kraken, 18 and FL executed the pipeline based on Mothur. ALAN analyzed the data and performed 19 the statistical analysis. ALAN, FL and BG executed the experiment and wrote the 20 manuscript. LLG and TM contributed to the experiment and revised the manuscript. 21

Chapter 4 has been submitted to Animal as “Taxonomic and functional 22 assessment reveals the effect of Angus breed genetics on rumen microbial signatures” 23 by Andre L. A. Neves, Yanhong Chen, Kim-Anh L. Cao, Siddhartha Mandal, Thomas J. 24 Sharpton, Tim McAllister, and Le Luo Guan. ALAN run Kraken, analyzed the data, 25 performed the statistical analysis, and wrote the manuscript. KALC and SM helped ALAN 26 run the sPLS-DA and ANCOM, respectively. TJS helped ALAN run ShotMAP. YC 27 assisted with the laboratory analysis. LLG and TM contributed to the manuscript writing. 28

Dedication 1

2 This thesis is dedicated to my Lord and Savior Jesus Christ, who said “I am the 3 resurrection, and the life: he that believeth in me, though he were dead, yet shall he live. 4 And whosoever liveth and believeth in me shall never die. Believe thou this?”, KJV 5 Bible, John 11.25,26. 6 7 8 I sincerely appreciate your support and guidance: I love you my Lord Jesus! 9

This thesis is also dedicated to my beautiful wife, Euzilene Neves, and our beloved son, 11 Daniel Neves! 12

Thanks Lene for the patience and encouragement during this time, I love you and 14 Daniel! 15

21 22

23 24

Acknowledgements 1 2

First, I would like to express my gratitude to my supervisor Dr. Le Luo Guan for her 3 dedication, mentoring, advice and patience during my Ph.D. program. I am proud to say 4 to everyone that I had an excellent supervisor, who provided me with the best 5 opportunities to learn and grow in various aspects of my professional career and personal 6 life. You may have no idea about the important role you played in my life, Dr. Guan! 7 Throughout these four years, you helped me get rid of all academic fears, and I am 8 passionate about science because you taught me so. Thanks for accepting me as your 9 student in May/2015 and for making me what I am now. I will take your academic and life 10 teachings for the rest of my days! Thank you very much! 11 I also want to thank the funding bodies AITF for the doctorate scholarship and 12 EMBRAPA for the financial support during the program. 13 To my committee members: Drs. Tim McAllister, Michael Steele, and Paul Stothard. 14 Thanks for providing me with constructive suggestions on my projects that contributed to 15 my professional development. Additionally, I want to thank Dr. McAllister for introducing 16 me to Dr. Guan in 2015 and for providing me with resources to improve my English skills 17 when I started writing my first paper in English in 2015. Since then, I have written in a 18 much faster pace! I also appreciated being a student of Drs. Steele and Stothard. I 19 enjoyed all the classes at the UofA, especially those at DRTC! 20 I thank Dr. Lynn McMullen and Dr. Carl Yeoman for taking their time to review my 21 thesis and agreeing to be my arms in length and external examiners. 22 I also thank the R developers and those involved in the elaboration of bioinformatic 23 software for their dedication and provision of amazing tools that were vastly employed in 24 this thesis. Special thanks to Drs. Kim-Anh L. Cao, Siddhartha Mandal, and Thomas J. 25 Sharpton for their collaborations in the statistical and bioinformatic approaches 26 implemented in the Chapters 3 and 4 of this thesis. Also a special thanks to my dear 27 friend Dr. Eóin O'Hara for his critical review of the findings described in this thesis, 28 especially Chapters 2 and 5. 29 I also thank Prof. Joanne Lemieux for opening the doors of her lab to a humble 30 student like me and for helping me perfect the protein work! To Dr. Elena Arutyunova, for 31

vii her incredible help and guidance in the protein purification and enzyme assays. Thanks 1 also to my dear brother Jiangkun Yu for being such an incredible friend and great support 2 24/7! Also thanks to my sister Marisol for joining Jiangkun and me in the journey to study 3 the proteins. I also thank Yanhong for her incredible help in all lab analysis of this thesis 4 and for being such a great friend and a beautiful human being. 5 I also would like to thank my little sister Yang Song, my preferred officemate of all 6 times and one of my best friends, as well as Biba, Anusha, Cameron and Fuyong for the 7 friendship and help with my Ph.D. program. 8 I want to thank all graduate students and staff who are unforgettable people: Urmila, 9 Yan Meng, Ou, Ke, Nilu, Huizeng, Bin, Junhong, Charnpal, Rebecca, Guanxiang, Mimi, 10 Xiao Xie, Jitka, Zhi, Junhua, Tao, Yawei, Duanqin, Diming, Xu Sun, Officemates: Andrea 11 and Andre, and the College Plaza people: Robert, Tiago, Huaigang (“the dark cut man”), 12 Jiyuan, Anahid and Janelle for the friendship and friendly conversations. 13 I would like to express my gratitude to my brothers and sisters at the West 14 Edmonton Christian Assembly, especially those of the Ignite group, for encouraging me 15 during my doctorate: Pastors Jane, Sandra, and J. Hayford and their families, brothers 16 Hayford, Olu, Noel, Mark, Greg, Tim, Art, Marco, sisters Wendy, Linda, Marlene, Joyce, 17 Cindy. I also thank the members of my home church in Brazil for the prayers and 18 encouragement, especially Pastor Luiz Oliveira, sister Benigna, brother Alex, and the 19 Assembly of God at the Aracaju’s airport neighborhood. 20 I would like to thank Humberto Reis, Sava Maria, Mariu, Cristiane, Mariazinha, 21 Jose Araujo, Duda, Cau, Coda and their entire families, Marilene, Eldio, Edilene, 22 Reginaldo, Keniel, Ezinho, Manu, Nilzete, Maria Pereira, Marcela, Eloi, Ivan, Mari, and 23 Tereza for the friendship, encouragement and support from Brazil and Portugal! 24 If I could say thanks to someone else, I would say thanks to my grandparents Zeca 25 and Noemia and my dear aunt Ana Maria for supporting my education since the very 26 beginning and for being great examples of perseverance. 27 Last but not least, I would like to express my gratitude to my father Jorge Neves 28 and my mother Doraci Neves, my brothers Anderson and Anselmo and their families, as 29 well as my sister Erica Neves, for the encouragement of all times and for believing in my 30 dreams. I love you forever! 31

viii

Table of Contents 1 2 Abstract ...... ii 3

Preface ...... v 4

Dedication ……………………………………………………………………………………………… vi 5 Acknowledgements ...... vii 6

Table of contents ...... ix 7

List of tables ...... xiv 8

List of figures ...... xv 9

Abbreviations ...... xvi 10

Chapter 1: Literature Review ...... 1 11 1.1 Introduction ...... 1 12

1.2 Importance of the rumen microbiome to cattle production ...... 3 13

1.2.1 Rumen microbial diversity and functions ...... 3 14

1.2.1.1 Rumen bacteria ...... 4 15

1.2.1.1 Rumen archaea ...... 5 16 1.2.1.1 Rumen protozoa ...... 6 17 1.2.1.1 Rumen fungi ...... 7 18 1.2.2 Feed efficiency and the rumen microbiome ...... 8 19

1.2.3 Rumen microbial contribution to methane emissions ...... 10 20

1.3 Rumen microbiome: a potential reservoir of industrially important microbial enzymes 21 ...... 13 22

1.3.1 CAZymes ...... 14 23

1.3.1.1 Cellulases ...... 15 24

1.3.1.2 Hemicellulases ...... 16 25

1.3.1.2.1 Mannanases ...... 16 26

1.3.1.2.2 Arabinofuranosidases ...... 17 27

1.3.1.2.3 Ferulic acid esterases ...... 17 28

1.3.1.2.4 pCoumaric acid esterases ...... 18 29

1.3.1.2.5 Xylanases ...... 18 30

1.3.1.3 Pectinases ...... 18 1

1.3.1.4 Polyphenol degrading enzymes ...... 18 2

1.4 Molecular approaches to study the rumen microbiome ...... 19 3

1.4.1 Metataxonomics ...... 20 4

1.4.2 Metagenomics ...... 22 5

1.4.3 Metatranscriptomics ...... 24 6

1.4.4 Metaproteomics ...... 26 7

1.4.5 Future perspectives in omics technologies to study the rumen microbiome 8 ...... 27 9

1.4.6 Future perspectives in molecular techniques to investigate lignocellulolytic 10 enzymes ...... 28 11

1.5 Statistical challenges associated with the studies of the rumen microbiome ...... 29 12

1.5.1 Alternative techniques to study microbiome data ...... 31 13

1.5.1.1 ANCOM ...... 31 14

1.5.1.2 MixMC ...... 32 15

1.5.2 Current challenges when comparing results across studies ...... 33 16

1.6 Knowledge gaps, hypotheses, and objectives ...... 34 17

1.7 Literature cited ...... 35 18

Chapter 2: Dynamics of microbial populations driven by interactions between diet and 19 host shed light on individualized rumen microbiota ...... 59 20

2.1 Introduction ...... 59 21

2.2 Materials and methods ...... 60 22

2.2.1 Animal study ...... 60 23

2.2.2 Feed chemical analysis ...... 62 24

2.2.3 VFA analysis ...... 62 25

2.2.4 DNA extraction ...... 63 26

2.2.5 Quantitative real-time PCR analysis ...... 63 27 2.2.6 Stratification of animals according to shift in microbial abundances ...... 64 28 2.2.7 Statistical analysis ...... 64 29 2.3 Results ...... 66 30

2.3.1 Effect of diet on the rumen microbes and phenotypic traits ...... 66 1 2.3.2 Effect of the phenotypic traits on the microbial populations ...... 67 2 2.3.3 Rumen microbial dynamics in response to the interactions of phenotypic traits and 3 the magnitude of change in host individual microbial population ...... 68 4 2.4 Discussion ...... 69 5 2.5 Conclusions ...... 73 6 2.6 Literature cited ...... 74 7 2.7 Tables and figures ...... 80 8

Chapter 3: Enhancing the resolution of rumen microbial classification from 9 metatranscriptomic data using Kraken and Mothur ...... 85 10 3.1 Introduction ...... 85 11

3.2 Materials and methods ...... 86 12

3.2.1 Animal study and sampling ...... 86 13

3.2.2 RNA extraction and sequencing ...... 87 14

3.2.3 Pipeline settings ...... 88 15

3.2.4 Statistical analysis ...... 89 16

3.2.5 Data submission ...... 90 17 3.3 Results ...... 90 18 3.3.1 Taxonomic distribution of the microbial profiles performed by Mothur or Kraken . 90 19 3.3.2 Differences in relative abundances of taxa in H vs. LFCR rumen samples ...... 91 20 3.3.3 Potential interactions between bacteria and archaea detected by Mothur or Kraken 21 ...... 92 22 3.4 Discussion ...... 93 23 3.5 Conclusions ...... 98 24 3.6 Literature cited ...... 99 25 3.7 Tables and figures ...... 107 26

Chapter 4: Taxonomic and functional assessment reveals the effect of Angus breed 27 genetics on rumen microbial signatures ...... 116 28 4.1 Introduction ...... 116 29

4.2 Materials and methods ...... 117 30

4.2.1 Animal trial and sampling ...... 117 31

4.2.2 VFA analysis ...... 118 1 4.2.3 RNA extraction and sequencing ...... 118 2

4.2.4 Bioinformatics analysis ...... 119 3

4.2.5 Statistical analysis ...... 120 4

4.2.6 Data repository resources ...... 121 5 4.3 Results ...... 121 6 4.3.1 Overview of the classification of active bacterial taxa ...... 121 7 4.3.2 Overview of the active microbial functions ...... 122 8 4.3.3 Identification of active microbial and functional signatures across breeds ...... 123 9 4.3.4 Relationship between active bacteria and volatile fatty acids ...... 124 10 4.3.5 Relationship between active bacteria and feed efficiency ...... 124 11 4.4 Discussion ...... 125 12 4.5 Conclusions ...... 130 13 4.6 Literature cited ...... 131 14 4.7 Tables and figures ...... 137 15 Chapter 5: Discovery and targeted functional profiling of novel glycoside hydrolases 16 through selective pressure on the native rumen microbial community ...... 146 17 5.1 Introduction ...... 146 18

5.2 Materials and methods ...... 148 19

5.2.1 Animal trial and sampling ...... 148 20

5.2.2 RNA extraction and sequencing ...... 148 21

5.2.3 Bioinformatic and statistical analysis ...... 149 22

5.2.4 Cloning, protein expression and purification ...... 150 23 5.2.5 Differential scanning fluorimetry (DSF) ...... 151 24 5.2.6 Size-exclusion chromatography ...... 152 25 5.2.7 Kinetic measurements ...... 152 26 5.2.8 Thermal inactivation of Xalanase 1 and thermodynamic analysis ...... 153 27 5.3 Results ...... 154 28 5.3.1 Taxonomic composition of the rumen microbiota revealed by metatranscriptomic 29 sequencing ...... 154 30

xii

5.3.2 Functional capability of the rumen microbiota ...... 154 1 5.3.3 Fibrolytic bacteria and glycoside hydrolases are abundant in the rumen of feed 2 efficient animals ...... 155 3 5.3.4 Targeted functional profiling of GH11 and GH45 families ...... 157 4 5.3.5 Structural analysis of xylanase 1 through homology modelling ...... 158 5 5.3.6 Purification and characterization of oligomeric state of Xylanase 1 ...... 159 6 5.3.7 Kinetic analysis of Xylanase 1 ...... 159 7 5.3.8 Thermal stability of Xylanase 1 ...... 159 8 5.4 Discussion ...... 160 9 5.5 Conclusions ...... 164 10 5.6 Literature cited ...... 164 11 5.7 Tables and figures ...... 171 12 Chapter 6: General Discussion ...... 178 13

6.1 Significance of the research ...... 178 14

6.2 Understanding the interactions between host variation in feed efficiency and the dynamics of 15 the rumen microbial population ...... 179 16

6.3 Understanding the rumen microbiota using mRNA-based metatranscriptomics ...... 181 17

6.4 Associations between microbial signatures, feed efficiency and host genetics ...... 182 18 6.5 Discovery of novel CAZymes in the rumen through selective pressure on the native 19 microbial population and targeted functional profiling ...... 184 20

6.6 Implications, and future directions ...... 186 21

6.7 Literature cited ...... 188 22

Bibliography ...... 192 23

xiii

List of Tables 1 2 3

Table 2.1 Quantification of the copy numbers of microbial populations in the rumen of beef 4 cattle fed forage or grain diets ...... 80 5

Table 3.1 Quantification of taxonomic phylotypes identified by each classification method 6 ...... 107 7

Table 3.2 Differentially abundant bacteria in efficient (low FCR) and inefficient (high FCR) 8 cattle according to the two classification methods ...... 108 9

Table 3.3 Differentially abundant archaea in efficient (low FCR) and inefficient (high FCR) 10 cattle according to the two classification methods ...... 110 11

Table 3.4 Comparison of bacterial and archaeal alpha-diversity indexes between efficient 12 (low FCR) and inefficient (high FCR) cattle according to the two microbial 13 classification methods ...... 112 14

Table 5.1 Kinetic parameters characterizing the thermal inactivation of Xylanase 1 ...... 171 15

xiv

1 List of Figures 2 3 Figure 2.1 Rumen microbial population and phenotypic traits in response to diet changes 81 4

Figure 2.2 Dynamics of rumen bacteria and fungi in response to DMI, starch intake, and 5 average daily gain ...... 82 6

Figure 2.3 Dynamics of rumen bacteria in response to NDF intake and CH4/kg DMI ...... 83 7

Figure 2.4 Feed conversion ratio and feeding time in response to the interactions of NDF 8 intake and the magnitude of change in host microbial population ...... 84 9

Figure 3.1 Flow chart of the pipelines (Mothur and Kraken) presenting software parameters 10 used to analyze the rumen microbiota ...... 113 11

Figure 3.2 Correlation circle plots and relevance networks generated from the output of 12 regularized canonical correlation method applied to rumen bacteria (genera) and 13 archaea (species) classified by Mothur or Kraken ...... 114 14

Figure 4.1 Taxonomic composition and microbial gene families obtained from 15 metatranscriptome data ...... 137 16

Figure 4.2 sPLS-DA results on rumen bacterial species in Black and Red Angus ...... 139 17

Figure 4.3 sPLS-DA results on rumen microbial gene families in Black and Red Angus .. 141 18 Figure 4.4 Heatmap and correlation circle plot generated from the output of regularized 19 canonical correlation (rCC) method for Black and Red Angus ...... 143 20 Figure 4.5 Differentially abundant bacterial species detected in Black and Red Angus when 21 FCR was adjusted to time ...... 145 22 Figure 5.1 Taxonomic composition and microbial gene families obtained from 23 metatranscriptome data in beef cattle ...... 172 24 Figure 5.2 Differentially abundant bacterial species, CAZymes, and microbial functions 25 detected in the rumen microbiome of feed efficient cattle ...... 174 26 Figure 5.3 Abundance of CAZyme members in the rumen microbiome ...... 175 27 Figure 5.4 Purification and characterization of oligomeric state, and kinetic analysis of 28 Xylanase 1 ...... 176 29 Figure 5.5 Thermal stability of Xylanase 1...... 177 30

Abbreviations 1 2

3-NOP – 3-Nitrooxypropanol 3

ANCOM – Analysis of composition of microbiomes 4

ANOVA – Analysis of variance 5

CAZymes - Carbohydrate-active enzymes 6

CH4 – Methane 7

CO2 – Carbon dioxide 8

DMI – Dry matter intake 9

DGGE – Denaturing gradient gel electrophoresis 10

FAO – Food and Agriculture Organization of the United Nations 11

FCR – Feed conversion ratio 12

FDR – False discovery rate 13

FE – Feed efficiency 14

GH – Glycoside hydrolase 15

GHG – Greenhouse gas 16

GWP – Global warming potential 17

H2 – Hydrogen 18

IPCC – United Nations Intergovernmental Panel on Climate Change 19

ITS – Internal transcribed spacer gene 20

MixMC - Multivariate Statistical Framework to Gain Insight into Microbial Communities 21

NDF – Neutral detergent fiber 22

OTU – Operational taxonomic unit 23

xvi

PCA – Principle component analysis 1

PCR – Polymerase chain reaction 2

PICRUSt – Phylogenetic investigation of communities by reconstruction of unobserved 3 states 4

QIIME – Quantitative insights into molecular ecology 5 qPCR – Quantitative polymerase chain reaction 6

RFI – Residual feed intake 7

RIM-DB – Rumen and intestinal methanogens database 8

RIN – RNA integrity number 9 rRNA/DNA – ribosomal RNA/DNA gene 10 sPLS-DA – Sparse partial least square regression – discriminant analysis 11

TMM –Trimmed mean of m-values 12

VFA – Volatile fatty acids 13

14 15 16 17

xvii

Chapter 1 18

Literature review 19

20 1.1 Introduction 21

The agri-food industry contributes to the economic development of many countries 22 through the provision of employment, income and food security. It has been estimated 23 that this industry employs over 1 billion people around the world, accounting for 1 in 3 of 24 all workers of the global workforce (Network 2016). In the domestic scenario, the beef 25 industry is the largest component of the Canadian food processing sector, with annual 26 sales surpassing $28 billion, including exports exceeding $6 billion and direct employment 27 for over 66,000 people (Statistics Canada, 2017). The Canadian beef sector is the 12th 28 largest globally and provides around 1.9% (around 62 million metric tonnes) of the world’s 29 beef supply (USDA, 2017). However, increased global demand for food will continue to 30 rise as the global population will continue to increase to 9.15 billion people by the year 31 2050. Despite the importance of Canada’s beef industry in the world, the dramatic 32 increase in the human population will require a further growth of 70% in food production 33 to meet the global demands for adequate nutrition (Nations 2015, Alexandratos and 34 Bruinsma 2012). 35

However, with the increase in the demand for animal protein, there is a 36 simultaneous increase in greenhouse gases (GHG) produced by the environmental 37 footprint of ruminant agriculture (Huws et al. 2018). On a global scale, it has been shown 38 that ruminants contribute between 9 and 11% of total anthropogenic GHG production 39 (Pickering et al. 2015b), with approximately 44% of ruminant emissions generated in the 40 form of methane (Rojas-Downing et al. 2017). Furthermore, methane emissions represent 41 2-12% energy loss from the gross energy intake in cattle, and this outcome reduces host 42 feed efficiency (Johnson and Johnson 1995). Therefore, mitigating methane emissions 43 from ruminants is a necessary step in the process of reducing the negative impacts of 44 GHG on the environment, and also to improve animal feed efficiency. 45

One fascinating microbial community that has raised much interest is the one that 46 resides in the upper digestive tract of ruminants, termed the rumen microbiome (Shaani 47 et al. 2018). A detailed understanding of the structure and function of the rumen 48 microbiome is central to improving animal feed efficiency and minimizing energy loss as 49 a result of methane emissions, as the rumen microbiota plays a fundamental role in the 50 food digestion and provision of energy for the host (Matthews et al. 2018). In the last 51 decades, the advent of high-throughput sequencing technologies has greatly advanced 52 our understanding of the major microbial populations and functional pathways of the 53 rumen microbiome involved in methane emissions and feed efficiency (Denman et al. 54 2018, Huws et al. 2018). However, little is known about the precise mechanisms of host- 55 microbe interactions in ruminants, and how the dynamics of individual changes in the 56 rumen microbial communities impact the host feed efficiency. Since inter-individual 57 variability and the microbiome are likely to vary simultaneously (Bashiardes et al. 2018), 58 a comprehensive understanding of their interactive effects on the host may create 59 opportunities to enhance animal productivity and alleviate the pressure of the livestock 60 sector on the environment. 61

In addition to its role in feed degradation and host energy provision, the rumen 62 microbiome is a valuable source of novel enzymes with major applications in the 63 biotechnology and biofuels industries (Seshadri et al. 2018, Ribeiro et al. 2016). The 64 rumen microbiome remains a source of valuable bioactives for the biotechnology industry 65 due to the proven ability of the rumen enzymes to break down and release the vast energy 66 stored in the most abundant carbon polymer on the planet, namely lignocellulose (Hess 67 et al. 2011, Gharechahi and Salekdeh 2018, Meng Qi et al. 2011). The importance of 68 discovering novel rumen-derived enzymes capable of converting lignocellulose into 69 biobased alternatives stems from the growing demands for a bioeconomy less dependent 70 on petroleum-derived fuels (Clark et al. 2006). Despite the wealth of knowledge on the 71 rumen microbiome and its relevance as a source for enzyme discovery, limitations related 72 to the analysis of microbiome datasets still exist and they are exacerbated by the vast 73 differences in analytical approaches employed across studies. Thus, standardization of 74 the sample processing workflow - from collection to analysis - to allow more reliable 75 comparisons of results across studies will provide comprehensive insights into the rumen 76

2 microbial composition and its functional capacity, which will ultimately lead to the 77 development of intervention strategies to reduce livestock methane emissions and to 78 discover new enzymes. 79

This literature review consists of four main sections. The first will summarize the 80 rumen microbiome and its microbes, and the contribution of the rumen microbiome to 81 ruminant feed efficiency and methane emissions. The second section relates to the rumen 82 microbiome as a reservoir for novel enzyme discovery. The third section reviews 83 molecular approaches used to investigate the rumen microbiome and to discover rumen 84 enzymes. The fourth section discusses the optimal statistical approaches for analyzing 85 microbiome datasets generated from the molecular technologies. 86

1.2 Importance of the rumen microbiome to cattle production 87

1.2.1 Rumen microbial diversity and functions 88

The ruminant forestomach is composed of four compartments, the rumen, reticulum, 89 omasum, and abomasum. The rumen is the largest compartment and contains a vast 90 array of anaerobic microorganisms (bacteria, archaea, protists, and fungi) that functions 91 in a coordinated fashion to ferment feed into volatile fatty acids (VFA), which are ultimately 92 used as energy source for milk and meat production (Russell and Rychlik 2001). The 93 rumen is the primary site of microbial fermentation and is lined with a filiform and foliate 94 epithelium with an extended surface area that facilitates the absorption of VFAs (Steele 95 et al. 2016). Temperature, pH, buffering capacity, osmotic pressure, and redox potential 96 (Eh) contribute to the maintenance of ruminal homeostasis and create the ideal habitat 97 for the growth of anaerobic and facultative anaerobic microbes (Russell and Rychlik 2001, 98 Weimer 1992). A distinguishing feature of the rumen microbiome compared to other 99 microbiomes (e.g., soil) is that it exhibits a highly anaerobic condition (Eh = -150 to -350 100 mV) that creates the ideal environment for microbial fermentation of feed particles 101 (Russell and Rychlik 2001). This fermentative activity results in a continuous production 102 of gases that are largely removed via eructation (Russell and Rychlik 2001, Weimer 1992). 103 Chemical reactions with reducing agents and a dedicated population of facultative 104 anaerobic microorganisms maintain the anaerobic conditions in the rumen for the whole 105 microbial population by scavenging oxygen taken in with food and drinking water (Weimer 106

1992). This synergistic microbial consortium is made up of cellulolytic, hemicellulolytic, 107 amylolytic, and biohydrogenating species that exhibit a high level of functional 108 redundancy (Firkins and Yu 2015, Hobson and Stewart 1997, Weimer 1992). Enhancing 109 our understanding of this complex microbial community is critical to provide the framework 110 needed to address the current challenges (e.g., mitigating methane emissions and 111 improving feed efficiency) of the ruminant agriculture chain. The next sections will discuss 112 in depth the major microbial members of the rumen microbiome (bacteria, archaea, 113 protozoa, and fungi) and their functions. A more in-depth discussion about the rumen 114 microbiome membership and their functions can be found in Huws et al. (2018) and 115 Matthews et al. (2018). 116

1.2.1.1 Rumen bacteria 117

Bacteria represent the most abundant organisms of the rumen microbiome (density of 118

1010-1011 cells/ml rumen fluid), making up more than 50% of the cell mass (Creevey et 119 al. 2014). Rumen bacteria are dominated by members of the Firmicutes, Bacteroidetes 120 and Proteobacteria phyla, which can account for >90% of the total bacterial population 121 (M. S. McCabe et al. 2015, Henderson et al. 2015, Fouts et al. 2012). Bacteria can be 122 free in the rumen fluid, attached to ingested feed particles (firmly or loosely), adhered to 123 the rumen wall (known as epimural bacteria), or associated with eukaryotic organisms 124 (Stewart et al. 1988, Cheng et al. 1979, Miron et al. 2001, McAllister et al. 1994a). 125 Numerically, the digesta-associated bacteria are the most abundant among the four 126 fractions (McAllister et al. 1994a), followed by the liquid-associated (planktonic) fraction 127 of the rumen fluid, which comprises around 30% of the rumen bacterial population (Millen 128 et al, 2016). The digesta-associated bacteria ferment feed or utilize the end-products of 129 this fermentation, and are dominated by fermentative species. In contrast, the epimural 130 bacteria are often facultative anaerobes, producing urease and scavenging oxygen to 131 assist in the maintenance of the ruminal anaerobic environment (Liu et al. 2016). 132

Early work indicated that at least 22 bacterial species had been characterized by 133 the beginning of the 21st century using culture dependent-methods (Russell and Rychlik 134 2001), but molecular techniques showed that less than 20% of the rumen microbiota can 135

4 be cultured using standard culture-based techniques (Krause et al. 2013, Nocker et al. 136 2007). The assessment of the rumen bacterial diversity in the Ribosomal Database 137 project revealed 13,478 bacterial sequences and the estimated number of bacterial 138 species in the rumen was of approximately 7,000 species (Kim et al. 2011). Recently, a 139 global census of rumen bacterial membership analyzed 742 rumen content samples 140 collected from 32 ruminant species across 35 countries and found that 30 dominant 141 bacterial genera were present in over 90% of animals and represented 89.4% of all 142 generated sequence data (Henderson et al. 2015). This study concluded that the current 143 technologies have already defined the dominant groups of rumen bacteria, and also 144 revealed the existence of a “core” group of rumen microbes, ubiquitous to ruminants 145 worldwide, containing the most abundant ruminal bacterial species, such as Prevotella 146 spp., Butyrivibrio spp., and Ruminococcus spp. (Henderson et al. 2015). However, the 147 taxonomy of ruminal bacteria is continually being updated, and the possibility of 148 discovering unculturable or previously unrecognized ruminal bacteria still exists. 149

1.2.1.2 Rumen archaea 150

The domain archaea is divided into two different kingdoms; Euryarchaeota, consisting of 151 methanogens and extreme halophytes, and Crenarchaeota, consisting of 152 hyperthermophiles and nonthermophiles (Bayley et al. 1999). Rumen methanogens 153 belong exclusively to Euryarchaeota (with a range of 106 to 108 cells per ml) and account 154 for 0.3-3.3% of the ruminal microbial population (Peter H. Janssen and Marek Kirs 2008, 155 Lin et al. 1997). These microbes require a very low redox potential (Eh = -300 mV) to 156 grow and are among the strictest anaerobes on Earth (Sirohi et al. 2010). The assessment 157 of the rumen archaeal community present in the Ribosomal Database project revealed 158 3,516 archaeal sequences and an estimated number of 1,500 species (Kim et al. 2011). 159 Despite this diversity, studies have shown that 90% of rumen archaea belong to the 160 Methanobrevibacter (63.2% of the methanogenic population), Methanomicrobium (7.7%), 161 Methanosphaera (9.8%), Rumen Cluster C (now referred as Methanomassiliicoccaceae, 162 7.4%), and Methanobacterium (1.2%) genera (Peter H. Janssen and Marek Kirs 2008, 163 Patra et al. 2017). Most methanogens (e.g., Methanobrevibacter species) use hydrogen 164 gas as electron donors to reduce CO2 to methane (formate can also be used as an 165

5 electron donor and may contribute to the production of up to 18% of ruminal methane) 166 (Hungate 1967). Other species, such as Methanosphaera stadtmanae, only produce 167 methane through the reduction of methanol with hydrogen, having one of the strictest 168 energy metabolisms of all methanogenic archaea (Peter H. Janssen and Marek Kirs 169 2008). The continuous production of methane carried out by methanogens leads to a low 170 concentration of hydrogen in the rumen and creates a favorable environment for the 171 growth of other species and more efficient fermentation (Peter H. Janssen and Marek Kirs 172 2008). 173

However, methane produced in the rumen is eructed and released into the 174 environment, leading to atmospheric pollution (Matthews et al. 2018). Due to these 175 environmental concerns, research on rumen methanogens has attracted great interest in 176 the last decade (Wallace et al. 2014, Johnson and Johnson 1995). Therefore, 177 manipulating the rumen microbiome to mitigate the activity of methanogens would help 178 reduce the negative impact of methane emissions on the environment and improve the 179 feed efficiency of ruminants. Strategies of mitigating methane emissions in ruminants and 180 the contribution of the rumen microbiome to host feed efficiency are discussed further in 181 sections 1.2.2 and 1.2.3. 182

1.2.1.3 Rumen protozoa 183

Protozoa are found in the range of 104 to 106 cells per ml in the rumen and are 184 accountable for 20-50% of the rumen microbial biomass (McSweeney and Mackie 2012). 185 They are involved in lipid hydrolysis and can produce large amounts of hydrogen via their 186 hydrogenosomes (Tymensen et al. 2012), and thus may contribute to methane production 187 via interspecies hydrogen transfer with methanogens (Hobson and Stewart 1997, S. 188 Kittelmann et al. 2015, Gijzen et al. 1988). Morphological studies have identified over 250 189 ciliate species in a range of ruminant hosts, represented by around 40 genera (Veira 1986, 190 Williams and Coleman 1997). The Entodinium genus is the most abundant protozoan in 191 the rumen, and other common genera include Polyplastron, Epidinium and Eudiplodinium 192 (Kittelmann et al. 2013a, Carberry et al. 2012a, Sylvester et al. 2004). Protozoa rapidly 193 digest feed particles via engulfment, and also predate on bacteria and other small 194 microbes, to ultimately convert them to an iodophilic storage polymer through a vast array 195

6 of enzymes such as amylases and glucosidases (Matthews et al. 2018, McSweeney and 196 Mackie 2012). 197

The contribution of protozoa to rumen fermentation remains controversial in the 198 scientific community. Deliberate removal of protozoa from the rumen, known as 199 defaunation, does not have a detrimental effect on the animal survival but it can affect 200 feed degradation efficiency in the rumen (Newbold et al. 2015). Defaunation can be 201 carried out by several techniques such as chemical removal of protozoa through the use 202 of copper sulphate, calcium peroxide, alcohol ethoxylate, coconut oil, linseed oil or soya 203 oil hydrolysate (Jouany 1995). A recent meta-analysis analyzed 23 in vivo defaunation 204 studies with the goal of determining the function of rumen protozoa and concluded that 205 the removal of protozoa from the rumen decreased the degradation of neutral and acid 206 detergent fibers, indicating that protozoa play important roles in fiber digestion (Newbold 207 et al. 2015). In addition to their functional importance as fibrolytic microbes, protozoa have 208 been associated with methanogenesis since defaunation reduces methane output by 11% 209 (Morgavi et al. 2012b, Newbold et al. 2015). Protozoa-associated methanogens also 210 account for approximately 37% of methanogenesis in ruminants (Finlay et al. 1994), and 211 defaunation could disrupt this symbiotic relationship in order to reduce methane 212 production (Matthews et al. 2018). However, research outcomes suggest that a strategy 213 to eliminate all protozoa in the rumen may not be the best approach to mitigate methane 214 emissions as it can compromise the activity of those protozoa involved in fiber 215 degradation (Newbold et al. 2015), and this can affect the ruminal fermentation stability 216 and consequently host productivity. 217

1.2.1.4 Rumen fungi 218

Anaerobic fungi (103 to 106 zoospores/ml) were firstly described by Colin Orpin in 1975 219 who reported them in the gastrointestinal tract of herbivores, particularly in the rumen and 220 caecum (Orpin 1975). These strictly anaerobic microbes within the phylum 221 Neocallimastigomycota are key players in the degradation of lignocellulosic feedstuffs 222 within the gut of ruminant and non-ruminant herbivores (Gruninger et al. 2014). The 223 anaerobic rumen fungi are currently grouped into eight genera (Neocallimastix, 224 Piromyces, Ontomyces, Buwchfawromyces, Caecomyces, Orpinomyces, Anaeromyces, 225

7 and Cyllamyces) characterized primarily using classical microscopy (Dollhofer et al. 2015). 226 However, next-generation sequencing technology has revealed an immense and 227 undiscovered biodiversity which may have been underappreciated in the past (Koetschan 228 et al. 2014). Each genus is distinguishable by morphological features such as thallus 229 morphology (monocentric vs. polycentric), rhizoids morphology (filamentous vs. bulbous) 230 and zoospore flagellation (monoflagellate vs. polyflagellate) (Ho and Barr 1995, Ozkose 231 et al. 2001). 232

To illustrate the importance of anaerobic fungi in plant cell wall degradation, 233 several experiments provided insight into their contribution to fiber digestion, feed intake, 234 rumen fermentation and rumen metabolism (Gruninger et al. 2014). For example, the 235 removal of anaerobic fungi from the rumen can reduce voluntary feed intake and dry 236 matter degradation, indicating that feed digestion is usually impaired when they are 237 removed from the rumen (Akin et al. 1988, Morrison et al. 1990, Gordon and Phillips 1998). 238 Furthermore, the elimination of anaerobic fungi from the rumen significantly reduced the 239 degradation of dry matter, neutral detergent fiber, acid detergent fiber and the activity of 240 carboxymethylcellulases (Ford et al. 1987, Gordon and Phillips 1993, Gao et al. 2013), 241 further confirming the importance of these microbes for fiber digestion in the rumen. 242

1.2.2 Feed efficiency and the rumen microbiome 243

Feed efficiency is a measure used to determine the relative ability of cattle to convert feed 244 into useable products (beef or milk), and it is a moderately heritable trait in cattle 245 (Cammack et al. 2018, Berry and Crowley 2013). In the past, beef cattle performance 246 was traditionally evaluated using traits that measured total output (i.e., weight gain or lean 247 meat carcass yield) without considering input measurements such as feed intake. As the 248 costs associated with feed increased dramatically to 55-70% of the total costs in beef 249 cattle operations (NRC 2016), the selection for feed efficient cattle is essential for the 250 continuous growth in farm profitability as well as environmental sustainability. Feed 251 conversion ratio (FCR) (gain in body weight per unit of feed consumed) has been 252 considered as one of the standard measures of feed efficiency for individual farmers, and 253 selecting for low FCR (efficient) animals usually results in a corresponding improvement 254 in gross efficiency (e.g., growth rate) (Korver 1988). The popularity of FCR among farmers 255

8 is a reflection of its moderate heritability for gross efficiency (Crews 2005), and this means 256 that selecting for low FCR animals leads to a simultaneous improvement in gross 257 efficiency. 258

However, FCR-based selection strategies result in increased growth rate and 259 mature cow size, leading to increased nutritional requirements to maintain the breeding 260 female that can thus increase the feeding costs (Gunsett 1984). To circumvent this 261 limitation, a metrics called residual feed intake (RFI) has been proposed as an alternative 262 to FCR (Koch et al. 1963). Expected feed intake is calculated by regressing average daily 263 gain and metabolic mid-test weight on standardized feed intake, and the difference of this 264 value from actual feed intake provides a “residual feed intake” or RFI value. Genetically 265 independent of growth, animals may be classified as low-RFI (efficient) or high-RFI 266 (inefficient) (Alemu et al. 2017, Kong et al. 2016, Carberry et al. 2014b, Montanholi et al. 267 2010). The main limitation to adopting the RFI methodology for predicting feed efficiency 268 on a large-scale basis is the inability to accurately measure feed intake in a feedlot setting. 269 However, with advancements in individual feed intake recording systems, such as 270 GrowSafe® Feeding Systems (GrowSafe Systems Ltd., Airdrie, Alberta), accurate feed 271 intake monitoring has been possible, and the use of RFI measures to estimate feed 272 efficiency in cattle has increased in recent years (Thompson 2015). 273

Regardless of the differences in the feed efficiency metrics (FCR vs. RFI) used in 274 beef cattle operations, it has been reported a significant variation in feed efficiency from 275 one animal to the other that is influenced mainly by the feeding behavior, energy 276 metabolism, the rumen microbiota composition, and the genetic background of the host 277 (Cantalapiedra-Hijar et al. 2018). Although all these factors are usually interconnected 278 with each other, studies have revealed that energy metabolism-related factors (at the 279 cellular level) are more relevant to explain individual variations in feed efficiency than 280 digestion-related factors (e.g., feeding behavior) (Cantalapiedra-Hijar et al. 2018). The 281 main finding described in those studies is that feed efficient cattle produce less heat than 282 inefficient animals due to a decreased protein turnover and a higher efficiency of 283 mitochondrial ATP production generated from the digestion and catabolism of 284 carbohydrates, proteins and lipids in the rumen (Kong et al. 2016, Cantalapiedra-Hijar et 285

9 al. 2018). The lower energy metabolic rate observed in feed efficient cattle is independent 286 of changes in feeding behavior (e.g., voluntary feed intake), suggesting that metabolic 287 energy-consuming metrics need to be taken into consideration to explain individual 288 variations in ruminants’ feed efficiency (Cantalapiedra-Hijar et al. 2018, Nkrumah et al. 289 2006). 290

Another important factor contributing to individual variations in feed efficiency is 291 the rumen microbiota as it plays essential roles in feed degradation and host energy 292 provision (Matthews et al. 2018). Several studies have shown linkages between rumen 293 microbial profiles and feed efficiency (Shabat et al. 2016, Guan et al. 2008, Carberry et 294 al. 2012b, Fuyong Li and Le Luo Guan 2017), including the linkages between efficient 295 cattle and the abundances of the bacterial Lachnospiraceae and Veillonellaceae families 296 (Myer et al. 2015a, F. Li and L. L. Guan 2017), and a number of archaeal taxa (Carberry 297 et al. 2014a, Carberry et al. 2014b, F. Li and L. L. Guan 2017). In the attempt to 298 understand the rumen microbiota of feed efficient cattle, a study of the functional activity 299 of the rumen microbial population revealed that efficient cattle (Low RFI) had lower rumen 300 microbial diversity and richness than inefficient cattle (High RFI) (Fuyong Li and Le Luo 301 Guan 2017), demonstrating that the rumen microbiota of efficient animals is less complex 302 and more specialized in harvesting energy from the diet than inefficient animals (Shabat 303 et al. 2016). More recently, Li et al. (2019b) analyzed the rumen microbiota of 709 beef 304 cattle and showed that multiple factors including sex, breed, and diet are responsible for 305 variations in the rumen microbial composition. The authors found that the relative 306 abundance of ~34% of microbial taxa as well as the copy number of total bacteria were 307 associated with feed efficiency and had a heritability estimate ≥ 0.15, suggesting that they 308 are influenced by the host genetic background. 309

As mentioned above, the interactions of the rumen microbiome with the host can 310 lead to individual variations in feed efficiency and understanding their influence on animal 311 production may provide opportunities to create strategies aiming to reduce the impact of 312 ruminant agriculture on the environment as discussed in the next section. 313

1.2.3 Rumen microbial contribution to methane emissions 314

The rumen microbial fermentation represents a significant source of greenhouse gas 315 emissions, with CH4 production contributing to reduce host dietary energy availability by 316 up to 12% (Johnson and Johnson 1995). This outcome has led to the urgent need for the 317 development of strategies to mitigate methane production in the livestock sector while 318 concurrently improving animal feed efficiency to meet the increasing global demand for 319 food. 320

The development of strategies to mitigate methane production by ruminants 321 depends on our understanding of the linkage between methanogens and methane 322 production in the rumen. Hydrogen is a by-product of normal rumen fermentation, and is 323 a regulator of methane production in ruminants (Hegarty et al. 2007), whereby 324 methanogens consume ruminal H2 in the terminal step of carbohydrate fermentation 325 (Deppenmeier 2002). As described previously, hydrogenotrophic methanogenesis is the 326 predominant pathway in the rumen, and is carried out mainly by Methanobrevibacter 327 species, which typically account for over 90% of archaeal 16S rRNA reads (Hristov et al. 328 2012). Species of Methanosphaera, Methanimicrococcus, and Methanobacterium also 329 utilize H2 to produce CH4 (Tapio et al. 2017). The less abundant methylotrophic 330 methanogens include members of the Methanosarcinales, Thermoplasmatales, 331 Methanophaera, and Methanomassillicoccaceae (Tapio et al. 2017, P. H. Janssen and 332 M. Kirs 2008, Poulsen et al. 2013). Due to the important role played by rumen archaea in 333

CH4 production, several strategies have been developed to reduce enteric methane 334 emissions, either by directly targeting the methanogen population or attempting to reduce 335 the substrates for archaeal metabolism. 336

These mitigation strategies include dietary manipulation, plant lipid feeding, 337 synthetic methanogen inhibitor supplementation, defaunation, and genetic selection for 338 reduced methane emitting animals (Hristov et al. 2013, McAllister et al. 2015, Martin et 339 al. 2010, Knapp et al. 2014, Kumar et al. 2014, Cammack et al. 2018, Pickering et al. 340 2015a). A promising compound for reducing ruminal methanogenesis is 3- 341 nitrooxypropanol (3-NOP), which has been shown to reduce methane emissions by up to 342 30% (Hristov et al. 2015, Jayanegara et al. 2018). However, reductions in methane 343 emissions do not always result in a redirection of energy towards enhancing animal 344

11 production efficiency while supplementing with 3-NOP, likely because the biochemical 345 processes involved in VFA and H2 production derived from cellulose breakdown requires 346 energy input (Jayanegara et al. 2018). Thus, the combination of 3-NOP supplementation 347 with phloroglucinol has been proposed to capture the excess of ruminal H2 from 348 methanogenesis and generate valuable metabolites (e.g., propionate) for the host 349 (Martinez-Fernandez et al. 2017). Although promising, it should be noted that the costs 350 of the application of 3-NOP supplementation and the long-term effect of this strategy on 351 the abatement of methane production in ruminants still need to be assessed in more 352 details. 353

Another promising method of reducing ruminal methane emissions per unit product 354 is through selecting genetic traits that increase the general efficiency of production in 355 ruminants (Kamke et al. 2016). This strategy aims at associating genetic traits (e.g., 356 weaning weight, dag score, muscle depth, etc.) with the rumen microbiota in order to 357 predict methane yield phenotypes across ruminant species (Kamke et al. 2016). 358 Researchers have recently characterized the rumen microbiota of sheep and found that 359 some microbial profiles were associated with low methane production in some individuals, 360 which they named as low methane emitting sheep. Species closely related to the lactic 361 acid producing bacteria Sharpaea azabuensis showed an increased abundance in low 362 methane emitting sheep, likely indicating rapid heterofermentative pathways resulting 363 from higher rumen turnover rates. However, H2 producing Ruminoccocaceae, 364 Lachnospiraceae and Verricomicrobia were more prevalent in the rumen of high methane 365 yielding sheep (Kittelmann et al. 2014, Kamke et al. 2016). In other ruminant species like 366 cattle and goats, high methane emitting animals exhibited an increased abundance of 367 Verrucomicrobia and Synergistetes bacteria and a decreased abundance of 368 methanogens (Denman et al. 2015, Wallace et al. 2015, Martinez-Fernandez et al. 2016). 369 Collectively, these findings suggest that the selection for animals that yield less methane 370 per unit of consumed feed based on the rumen microbiota may be beneficial for animal 371 production systems. 372

It is important to note that in addition to concentrating efforts to mitigating methane 373 emissions from ruminants and improving cattle feed efficiency through the understanding 374

12 of the rumen microbiota, researchers have discovered valuable enzymes in the rumen 375 that can be applied in the biotechnology industry (e.g., feed additives), and this topic will 376 be discussed in the next section. 377

1.3 Rumen microbiome: a potential reservoir of industrially important microbial 378 enzymes 379

The success of ruminants in digesting recalcitrant biomass is largely explained by the 380 ability of ruminal microorganisms to metabolize the constituents of plant structural 381 carbohydrates (Russell and Rychlik 2001), as the ruminant genome lacks the genes 382 encoding for the enzymes involved in the breakdown of lignocellulose (Seshadri et al. 383 2018). Structural carbohydrates (plant cell wall) are complex structures composed 384 mainly of lignocellulose – the most abundant organic material on Earth – which is a 385 matrix composed of polysaccharide networks, glycosylated proteins, and lignin (Chafe 386 1969, McNeil et al. 1984). Although abundant in nature, lignocellulose represents a 387 critical barrier for the conversion of plant biomass into feed sources and biofuels owing to 388 the lack of an efficient enzymatic system to deconstruct lignocellulose and release all 389 fermentable sugars it contains (Bader et al. 2010). 390

The lignocellulolytic matrix is comprised of three main components: cellulose, 391 hemicellulose, and lignin. Cellulose is a linear biopolymer of anhydroglucopyranose 392 molecules, connected by β-1,4-glycosidic bonds and represents the most abundant 393 component of the lignocellulose complex, accounting for approximately 20-30% of the dry 394 weight of most plant primary cell walls (Chafe 1969, McNeil et al. 1984). Adjacent 395 cellulose chains are bound by hydrogen bonds, hydrophobic interactions, and Van der 396 Waal’s forces that generate a parallel alignment of crystalline structures known as 397 microfibrils (Zhang et al. 2006). The second most abundant component of lignocellulose 398 is hemicellulose, which consists of mixed polymers of pentoses (including xylose and 399 arabinose), hexoses (mainly mannose, less glucose, and galactose) and sugar acids 400 (Bhatia et al. 2012). Although the composition of hemicelluloses varies considerably 401 depending on the plant source (Saha 2003, Saha 2000), hemicellulose is composed 402 mainly of xylans with a backbone structure of β-1,4-linked xylose residues attached to 403

13 various sidechain molecules such as acetic acid, coumaric acid, glucuronic acid, ferulic 404 acid, and arabinose (McNeil et al. 1984). Lignin is the third main polymer of lignocellulose 405 and possesses a structure comprised of three aromatic alcohols: coniferyl alcohol, sinapyl, 406 and p-coumaryl (McNeil et al. 1984). Compared to cellulose and hemicellulose, lignin is 407 the most recalcitrant structure for rumen microorganisms to degrade and extract 408 metabolizable energy (Himmel et al. 2007, Sanchez 2009). The surface area available 409 for microbial attachment and the retention time of the ingested feed in the rumen are listed 410 as the main obstacles for lignin digestion by rumen microorganisms (Wang and McAllister 411 2002). Thus, solubilization of lignin is a necessary step for increasing digestibility of 412 lignocellulosic compounds through microbial fermentation in the rumen (Weimer et al. 413 1990). 414

Due to the recalcitrant nature of lignocellulose, the degradation of the complex 415 cross-linkage structure intertwined by cellulose, hemicellulose, and lignin requires the 416 synchronized action of a diverse array of rumen microbial enzymes to cleave the 417 numerous bounds within the plant cell wall structure and access specific substrates in the 418 lignocellulosic fiber (Wang and McAllister 2002). Owing to the efficacy of the rumen 419 enzymes to deconstruct lignocellulose, researchers have recommended these 420 biocatalysts to the animal feed industry as feed additives for ruminants (Sehgal et al. 2008) 421 and non-ruminants (poultry and swine) (Theodorou et al. 2007). Potent polysaccharide- 422 degrading enzymes derived from the rumen have also raised the interest of several 423 industries including brewing, food, textile, paper, and biofuel production (Gruninger et al. 424 2014). A broad group of enzymes present in the rumen which are of particular interest to 425 the industry are the carbohydrate-active enzymes (CAZymes), and they will be discussed 426 in the topic of the next section. 427

1.3.1 CAZymes 428

Microbes play a significant role in regulating the biochemical processes of feed digestion 429 in the rumen (Russell and Rychlik 2001) and are valuable sources of enzymes that have 430 various biotechnological applications (Seshadri et al. 2018). The interest in discovering 431 new enzymes for the saccharification of lignocellulose has led to the creation of an 432 extensive database describing CAZymes families of structurally-related catalytic and 433

14 carbohydrate-binding modules (CBMs) that are continually updated to cover all CAZymes 434 across organisms and subfields of glycoscience (Cantarel et al. 2009, Lombard et al. 435 2014). CAZymes and their associated CBMs are classified based on sequence similarity 436 and encompass biocatalysts whose job is to modify and cleave carbohydrates (glycoside 437 hydrolyses (GHs), polysaccharide lyases - PLs, carbohydrate esterases - CEs) and 438 synthesize them (glycosyltransferases) (Cantarel et al. 2009, Lombard et al. 2014). Today, 439 well over 100 GHs families of over 50 CBM families have been described in the 440 Carbohydrate-Active enZymes - CAZy database. In the next sections, important 441 CAZymes will be discussed, with the reader being encouraged to visit the CAZy database 442 for more details about the structure and classification of CAZymes (http://www.cazy.org). 443 444

1.3.1.1 Cellulases 445

Cellulases are family members of the broad group of GHs (e.g., GH5, GH6, GH9, GH45), 446 which have gained interest for a number of biotechnological applications (e.g., treatment 447 of paper pulp) owing to their ability to hydrolyze 1,4 β-D-glycosidic linkages of the 448 cellulose chain (Bayer et al. 1998). Based on the structure and functionality, cellulases 449 have been classified into three groups: 1) Endoglucanases, a group of cellulases that 450 exhibits a deep cleft or groove to accommodate the cellulose chain at any point along its 451 length in order to cleave internal bonds at amorphous sites of new chain ends; 2) 452 Exoglucanases, in contrast, are a group of cellulases that possess the active site in an 453 extended loop that forms a tunnel, though which one of the termini of a cellulose chain 454 can be threaded; and 3) β-glucosidases, which are cellulases that hydrolyze cellobiose 455 to generate two molecules of glucose, and are often associated with the microbial cell 456 surface when cellodextrins are transported into the cell (Bhat and Bhat 1997). 457

In addition to the free enzymes, self-assembled multienzyme complexes known as 458 cellulosomes have been reported in several anaerobic environments including the rumen 459 (Artzi et al. 2016, Bayer et al. 1998). To date, the only rumen species that exhibits an 460 elaborated cellulosomal system is the mesophilic bacterium Ruminoccocus flavefaciens 461 (Dassa et al. 2014). The major difference between free enzymes and cellulosomal 462 enzymes is that the free enzymes usually contain a CBM to guide the catalytic domain to 463

15 the substrate, while the cellulosomal enzymes carry a dockerin domain that incorporates 464 the enzyme into the cellulosomal complex (Artzi et al. 2016). The attachment of the 465 cellulosomal machinery to cellulose chains is achieved via cellulose-binding proteins 466 (Family 3a-CBM) of the scaffoldin subunit, which contains one or more cohesin modules 467 connected to other types of functional modules. The arrangement of the modules on the 468 scaffoldin subunit and the specificity of the cohesins and/or dockerin for their modular 469 counterpart dictate the overall structure of the cellulosome (Artzi et al. 2016, Bayer et al. 470 1998, Doi and Kosugi 2004). Due to their high hydrolytic activity, cellulosomes have a 471 great potential for the degradation of biomass and are the focus of much effort to engineer 472 an effective cellulosomal structure for the conversion of lignocellulose into valuable 473 products, such as biofuels. 474

1.3.1.2 Hemicellulases 475

In contrast to cellulose degradation, the digestion of hemicelluloses poses a different 476 challenge, as this group of polysaccharides includes widely different types of sugars or 477 non-sugar constituents with different types of bonds. Thus, hemicellulases can be divided 478 into two main groups: a) those that cleave the mainchain backbone (e.g., mannanases 479 and xylanases), and b) those that degrade sidechain substituents or short end products 480 (e.g., arabinofuranosidase) (Shallom and Shoham 2003). Thus, the catalytic modules of 481 hemicellulases can be either glycosyl hydrolases that hydrolyze glycosidic bonds, or 482 carbohydrate esterases, which hydrolyze ester linkages of acetate or ferulic acid side 483 groups (Shallom and Shoham 2003). Due to the different structures of the hemicellulose 484 molecule, details of the various hemicellulases needed for their catabolism are given 485 below. 486

1.3.1.2.1 Mannanases 487

Mannan is a component of hemicellulose and consists of β-1,4 linkages between 488 mannose monomers that form the hemicellulose cross-linkages (Hogg et al. 2003). β- 489 Mannanases (e.g., GH5, GH26) hydrolyze mannan-based hemicelluloses and release 490 short β-1,4-manno-oligomers, which can be further hydrolyzed to mannose by the action 491 of β-mannosidases. Deficiency in these enzymes in ruminants (termed β-mannosidosis) 492 causes skeletal abnormalities (Shallom and Shoham 2003), demonstrating the relevance 493

16 of β-mannases for the digestion of mannan in ruminants. Recently, researchers have 494 applied exogenous β-mannanases from Aspergillus niger in the animal feed industry as 495 an additive to improve feed conversion efficiency in dairy cattle (Tewoldebrhan et al. 496 2017). However, the rumen microbiome possesses the ability to degrade mannan as 497 demonstrated in the discovery of a multifunctional glycosyl hydrolase encoded in the 498 genome of the bacterium Prevotella bryantti B14 (Palackal et al. 2007). 499

1.3.1.2.2 Arabinofuranosidases 500

Arabinose is found in conjunction with xylan as a hemicellulose component of plant cell 501 walls, with arabinose units being attached to xylan via alpha-1,2,1,3,1,5 or linked to C2 502 or C3 positions on the arabinoxylan chain (Shallom and Shoham 2003). In the rumen, 503 arabinose units can be cleaved off the xylose backbone by arabinofuranosidases (e.g., 504 GH3, GH43) expressed by rumen bacteria such as Ruminococcus albus (Greve et al. 505 1984). 506

1.3.1.2.3 Ferulic acid esterases 507

Ferulic acid esterase (e.g., CE1) is a group of enzymes that forms a subclass of carboxylic 508 ester hydrolases. These enzymes hydrolyze the bonds between hydroxycinnamates and 509 sugars to release ferulic acid (Rashamuse et al. 2007, Wang and McAllister 2002). In the 510 rumen, these ester bonds are cleaved by ferulic acid esterases encoded in the genome 511 of the rumen fungi Anaeromyces mucronatus (M. Qi et al. 2011) as well as by rumen 512 bacteria such as P. ruminicola (Kabel et al. 2011). 513

1.3.1.2.4 p-Coumaric acid esterases 514 p-Coumaric acid esterase or p-coumaroyl esterase (e.g., CE1) is an essential enzyme for 515 efficient degradation of lignocellulose biomass in the rumen (Wang and McAllister 2002). 516 This enzyme targets the p-Coumaroyl ester bonds that connect lignin to hemicelluloses, 517 releasing p-Coumaric acid. Surprisingly, this type of enzyme is exclusively produced by 518 anaerobic rumen fungi (phylum Neocallimastigomycota) and was not yet described in 519 rumen bacteria (Borneman et al. 1990), further strengthening the ecological role and 520 significance of anaerobic fungi for deconstructing lignocellulose biomass in the rumen 521 (Gruninger et al. 2014). 522

1.3.1.2.5 Xylanases 523

The xylan molecule consists of β-1,4 linked xylopyranosyl residues and contains 524 sidechains with acetyl group and L-arabinofuranosyl residues. Xylanases (e.g., GH5, 525 GH8, GH10, GH11, GH51) are responsible for the hydrolysis of xylan by breaking the 526 glycosidic linkages present in the xylan backbone (Shallom and Shoham 2003). Like the 527 cellulases, the xylanases can be classified into three groups: endoxylanases, β- 528 xylosidases and acetyl xylan esterases (Beg et al. 2001). All three enzymes hydrolyze 529 the xylan molecule, rendering D-xylose sugar (Kosugi et al. 2001). The scientific interest 530 in discovering new xylanases is reflected by the vast number of research papers 531 published in recent years describing numerous xylanases applications in the pulp and 532 paper industries (Beg et al. 2001, El Enshasy et al. 2016), and as exogenous enzyme 533 preparations marketed by the animal feed industry (Wang and McAllister 2002). The 534 rumen harbors a wide range of microbes cable of degrading xylan, including Prevotella 535 spp. (such as P. ruminocola, P. albensis, P. brevis, and P. bryantii) and non-Prevotella 536 spp. (such as Ruminococcus and Fibrobacter) (Russell and Rychlik 2001, Dai et al. 2015). 537

1.3.1.3 Pectinases 538

Pectin exists in the primary cell wall and represents the plant’s first line of defense against 539 dehydration and penetration by phytopathogens. The pectin’s structure is a backbone of 540 alpha-1,4-linked residues of D-galacturonate that is degraded by rumen pectinolytic 541 enzymes (e.g., PL11, GH28) including pectin lyases, polygalacturonases and pectin 542 methylesterases (Wang and McAllister 2002). One of the major pectinolytic bacterial 543 species that inhabits the rumen is Lachnospira multiparus, which produces pectin lyases 544 and pectin methylesterases (Silley 1985, Russell and Rychlik 2001). In addition to that 545 bacterial species, rumen fungi also exhibit pectinolytic enzymes (Orpin and Joblin 1997, 546 Gordon and Phillips 1992). 547

1.3.1.4 Polyphenol degrading enzymes 548

Feed consumed by ruminants contain not only the nutrients required by the host animal 549 for maintenance and production, but also hold naturally occurring plant secondary 550 compounds such as tannins, saponins, phenolic acids and silica that usually cause 551

18 adverse effects on the activity of fibrolytic enzymes (McAllister et al. 1994b, Bae et al. 552 1997). However, some gastrointestinal microbes of ruminants are able to break down 553 tannin-protein complexes through an enzyme known as tannin acyl hydrolase (tannase), 554 which catalyzes the hydrolysis of ester bonds present in gallotannins, complex tannins, 555 and gallic acid esters (Rodríguez et al. 2009, Ramírez et al. 2008). Tannase activity in 556 ruminants has been reported mainly in Streptococcus caprinus (now Streptococcus 557 gallolyticus) (Brooker et al. 1994) and Selenomonas ruminantium (Skene and Brooker 558 1995). These tannninolytic microbes possess tannin-degrading ability to tolerate tannins 559 in feeds, and their selection to promote long-term protection against tannin toxicity can 560 be used to improve the nutritive value of tannin-rich feeds (Goel et al. 2005). 561

It is important to note that a key component to improve our understanding of the 562 rumen microbiome and its role in animal productivity and enzyme discovery is the 563 development of bioinformatic tools specially tailored to overcome the technical challenges 564 of the analysis of massively paralleled, high-throughput sequencing data. The section 1.4 565 of this chapter will discuss in depth the molecular biology techniques designed for 566 investigating the rumen microbiome, its microbes and enzymes. 567

1.4 Molecular approaches to study the rumen microbiome 568

The work of Robert Hungate, outlined in the book “The Rumen and its Microbes” (Hungate 569 1966) formed the foundation for investigations of basic rumen microbial ecology in the 570 context of agricultural production (McCann et al. 2017). Hungate’s roll-tube method called 571 for the use of rumen fluid in growth media to isolate ruminal microbes, and quickly became 572 the most common approach available for rumen microbiologists to discover new species 573 (Bryant and Burkey 1953, Bryant and Robinson 1961). By the 1990s and before the rise 574 of the ‘omics’ approaches, rumen microbiologists perfected culture media techniques to 575 isolate rumen bacteria and redefined our understanding of how the anaerobic microbiota 576 functions (Krause et al. 2013). This pioneering work enabled the identification of at least 577 22 major bacteria and improved the knowledge of the interrelationship between 578 carbohydrate and protein-nitrogen metabolism driven by ruminal microbes (Russell and 579 Rychlik 2001). From the use of the Hungate “roll-tube” technique, modern culture-based 580 approaches were decisive to unveiling the biochemical and physiological activities of 581

19 rumen microbes since the cultured organisms present in the public collections can be 582 studied both in vitro and in vivo (Huws et al. 2018). 583

Our ability to culture rumen microbes was further improved by technologies (e.g., 584 dilution to extinction and removal of reducing agents) that can be used to culture the as 585 yet unculturable rumen microbes (Kenters et al. 2011, Poelaert et al. 2017). Despite the 586 success of cultivation experiments to characterize undescribed rumen microbes, most 587 cultures are not available in every collection, and culture-based methods may also be 588 limited to, or biased toward, strains that are highly abundant and organisms which are of 589 specific interest to research (Zehavi et al. 2018, Huws et al. 2018). Furthermore, it has 590 been demonstrated that the number of microbial species isolated and characterized from 591 the rumen is low (Morgavi et al. 2013), highlighting the importance of molecular biology 592 methods to overcome the limitations of culture-based approaches to study the rumen 593 microbial composition and diversity (Morgavi et al. 2013). As new technology has become 594 available, what initially involved the isolation and culture of strains in the laboratory, has 595 now moved to large-scale sequencing of the ‘total’ detectable rumen microbiota nucleic 596 acids (e.g., metagenomics and metatranscriptomics) (Denman et al. 2018). The 597 comprehensive application of these new technologies in most ruminant microbiota studies 598 has caused significant advances in our understanding of the rumen microbiota diversity 599 and function, and therefore these techniques will be discussed in more details in the 600 following sections. 601

1.4.1 Metataxonomics 602

Metataxonomics (or amplicon sequencing) refers to the high-throughput sequencing 603 analysis of amplified taxonomic marker genes and has been used to characterize the 604 taxonomic composition of microbiota of many ecosystems (Marchesi and Ravel 2015). In 605 ruminant studies, this method has been routinely employed to capture variations in the 606 microbial composition in response to perturbations such as dietary changes, subacute 607 acidosis, and different feed efficiencies and methane production (Petri et al. 2013, 608 McGovern et al. 2017, Henderson et al. 2015, Myer et al. 2015b). Studies of bacterial and 609 archaeal diversity have relied on primer sets that target the 16S rDNA or rRNA gene 610 (Deusch et al. 2015, Matthew Sean McCabe et al. 2015, McGovern et al. 2017), whereas 611

18S rRNA and Internal Transcribed Spacer (ITS) genes have been used for eukaryotic 612 targets, typically protists and fungi (Kittelmann et al. 2013b, Sandra Kittelmann et al. 613 2015). While metataxonomics is a cheap, fast technology to characterize the rumen 614 composition and diversity in a wide range of ruminant hosts (McGovern et al. 2017, Tapio 615 et al. 2016, F. Li et al. 2016, M. S. McCabe et al. 2015, Myer et al. 2015a, Jami et al. 616 2013, Jami and Mizrahi 2012), it is subject to several limitations. These include PCR 617 biases (primer specificity and sensitivity, non-specific annealing, differential amplification 618 specificity of taxonomic groups, artefact formation), poor resolution beyond the genus 619 level, and the fact that amplicon sequencing cannot account for marker gene copy 620 number variations (Firkins and Yu 2015, Poretsky et al. 2014). Another critical step in 621 metataxonomics is the construction and maintenance of updated and representative 622 databases to aid accurate taxonomical assignments of rumen phylotypes. Commonly 623 used databases in rumen metataxonomic studies are the Greengenes (DeSantis et al. 624 2006) and SILVA (Quast et al. 2013) databases for bacteria, RIM-DB (Seedorf et al. 2014) 625 for methanogens, the ureC database (Jin et al. 2017) for ureolytic bacteria, and AF- 626 RefSeq (Paul et al. 2018) for anaerobic fungi. 627

In such analysis, phylogenetic gene investigations are based on the sequence 628 similarity or, more precisely, on evolutionary divergence between defined taxonomic units 629 (Denman et al. 2018). In earlier studies, a value of 97% sequence similarity was used to 630 define a species level rank at least for the full-length 16S rRNA gene (Stackebrandt and 631 Goebel 1994). With the rise of shorter sequences generated from next-generation 632 sequencing technologies, the previously recommended threshold of 97% has now been 633 revised to a suggested value of 99 to 100% sequence similarities to define the operational 634 taxonomic unit (OTU) (or more commonly now, the Amplicon Sequence Variant) 635 (Callahan et al. 2016). However, using a value of 100% is also likely to generate multiple 636 OTUs from the same species, as a single species (or different strains) may possess 637 multiple copies of an identical 16S rRNA gene (Větrovský and Baldrian 2013). The choice 638 of a suitable variable region along with the similarity threshold of 99% to allow for possible 639 polymorphism effects may be a desirable step to mitigate those limitations (Denman et al. 640 2018). The most popular variable region currently being targeted to cover both bacterial 641 and archaeal populations is the V4 region (Kozich et al. 2013), but other researchers have 642

21 suggested that the V1-V3 and V6-V8 regions should be the choice for rumen bacteria and 643 archaea, respectively (Peter H. Janssen and Marek Kirs 2008, Fuyong Li et al. 2016). 644 While metataxonomics provides insight into the composition of the rumen microbiota, this 645 technique cannot assess the microbial function within a given environment, which needs 646 to be performed using other methods as discussed below. Regardless of this limitation, 647 researchers can still predict the rumen metabolic functions from phylogenetic data 648 through programs derived from the PICRUSt tool (Langille et al. 2013) such as CowPi 649 (Wilkinson et al. 2018) although such tools are still subject to the inherent biases and 650 limitations of amplicon sequencing. 651

1.4.2 Metagenomics 652

The concept of metagenomics was first used by Handelsman et al. (1998) to explore the 653 biosynthetic machinery of soil microbiota, and later it became the recommended 654 approach to characterize the potential function of the microbiota directly from their 655 genomes (Marchesi and Ravel 2015). Determining the functional capacity of the rumen 656 microbiota is feasible and can be achieved through random sequencing of all genetic 657 material contained in a sample (metagenomic shotgun sequencing), with the aim of 658 cataloging genes and species (Denman et al. 2018). Metagenomic shotgun sequencing 659 can thus be used to profile the taxonomy, catalogue the functional genes, discover new 660 enzymes and pathways, assemble whole- and fragmented-genomes, and quantify the 661 abundance of functional genomic elements across and between samples (Shinichi 662 Sunagawa et al. 2013, Gupta et al. 2016b, Huws et al. 2018). Metagenomics can provide 663 informative taxonomic and functional profiles using several analytical methods (S. 664 Sunagawa et al. 2013), which include the analysis of informative marker genes (e.g. the 665 16S rRNA gene) or contigs assembly aligned to databases of reference microbial 666 genomes (Gupta et al. 2016a). Several studies to date have employed those analytical 667 approaches in rumen microbiome investigations, including the first functional 668 metagenomic assessment of the rumen microbiome in pre-ruminant calves (Li et al. 2012) 669 and the effect of feed conversion rate and breeds on the structure and functions of the 670 rumen microbiome (Roehe et al. 2016). 671

While shotgun metagenomics allows for the study of uncultivable microbial profiles 672 and has become an essential tool for understanding the genomic potential of the rumen 673 microbiome (Morgavi et al. 2013), limitations still apply to this method. Sample collection 674 protocols, DNA extraction techniques and issues around varying levels of genomic DNA 675 for different species can bias the detection of closely related species or strains that can 676 become inadvertently co-assembled by traditional metagenomic approaches (Denman et 677 al. 2018). To overcome these limitations, binning techniques have been developed to 678 exploit the functional capacity of the rumen microbiota through the construction of 679 complete or near complete microbial genomes directly from metagenomic sequencing 680 data (Tyson et al. 2004). The ability to bin genomes from metagenomes stems from an in 681 silico approach whereby metagenomic assembled contigs are placed in common bins 682 based on the frequency and abundance of nucleotides, and coverage depth within the 683 sample using programs like PhyloPythiaS, GroopM, and MetaBat (Patil et al. 2011, Kang 684 et al. 2015, Imelfort et al. 2014). Completeness and contamination of metagenomic 685 assembled genomes (MAGs) can then be assessed based on the presence of multiple 686 lineage-specific single-copy marker genes using programs like CheckM or PhyloSift 687 (Darling et al. 2014, Parks et al. 2015). These methodologies also allow for assigning 688 taxonomic information to MAGs from single-copy markers genes concatenated in 689 genome-based taxonomy trees (Denman et al. 2018). The utility of MAGs was 690 demonstrated by Svartström et al. (2017), who assembled 99 microbial genomes from 691 the moose rumen, and by Stewart et al. (2018), who assembled 913 microbial genomes 692 from the rumen of cattle. 693

In addition to the applications discussed previously, metagenomics can be used to 694 identify novel enzymes in the rumen. The discovery of enzymes in metagenomes can be 695 accomplished using two different strategies: a) sequence-based metagenomic 696 approaches that look for homologous enzymes in metagenomic datasets, and b) 697 functional metagenomic approaches, in which metagenomic libraries are constructed, 698 and clones are screened for enzyme activity (Lam et al. 2015, Sabree et al. 2009). 699 Sequence-based metagenomic techniques rely on the search for genes that code for a 700 particular enzyme through the design of PCR primers or hybridization probes on 701 conserved regions and motifs of known protein families (Ferrer et al. 2009). In contrast, 702

23 the functional approach is not dependent on previous genomic knowledge and allows for 703 the discovery of novel enzymes with unexpected peptide sequences (exhibiting classical 704 or new activities) that would not be predicted based on DNA sequencing alone (Distaso 705 et al. 2017). 706

The diversity of hydrolytic enzymes in a metagenomic library was screened for 707 hydrolase activity for the first time in 2005 (Ferrer et al. 2005), when a total of 22 clones 708 with hydrolytic activities were identified and characterized in the rumen of a cow. Since 709 then, numerous metagenomic studies have reported the diversity of fibrolytic enzymes in 710 ruminants fed forage diets (Brulc et al. 2009, Hess et al. 2011, Wang et al. 2013). While 711 metagenomic techniques have successfully uncovered genes for lignocellulose 712 breakdown, this method cannot provide further information about transcripts that are 713 actively transcribed during lignocellulose breakdown (Rosnow et al. 2017). As described 714 above, shotgun metagenomics can be used for taxonomic identification, functional 715 characterization and discovery of novel enzymes in the rumen microbiome, but it tends 716 to be biased towards the most abundant genes encoded by the most numerically 717 abundant microbial species, and this outcome may not necessarily reflect the importance 718 of a gene, species or strains residing in the metagenome. 719

1.4.3 Metatranscriptomics 720

Metatranscriptomics refers to the analysis of expressed RNAs by high-throughput 721 sequencing of the corresponding cDNAs, and is the method used to provide information 722 on the regulation and expression profiles of complex microbiomes (Marchesi and Ravel 723 2015). While metagenomic shotgun sequencing and MAGs have been advantageous 724 over metataxonomic techniques, they cannot distinguish whether the genomic content of 725 the microbiota is from viable cells as opposed to active function revealed by RNA-based 726 technologies. Currently, metatranscriptomics is considered a reliable approach to identify 727 metabolically active microbial communities (Franzosa et al. 2014) and to find new 728 functions and/or enzymes that would not be otherwise revealed by metagenomic 729 approaches (Rosnow et al. 2017). Metatranscriptome profiling has become popular in 730 rumen microbiology studies in recent years and was initially used to reveal the snapshot 731 of the composition and relative abundance of active species (bacteria and fungi) involved 732

24 in the lignocellulose breakdown (Meng Qi et al. 2011, Dai et al. 2015). In addition to 733 taxonomic identification, metatranscriptomic techniques have been used to explore the 734 functional capacity of the rumen microbiome to degrade lignocellulose as exemplified by 735 the investigations of novel CAZymes (e.g., glycoside hydrolases, carbohydrate-binding 736 modules, carbohydrate esterases) involved in the ruminal digestion of recalcitrant diets in 737 ruminants (Fuyong Li and Le Luo Guan 2017, Comtet-Marre et al. 2017, Comtet-Marre 738 et al. 2018). 739

Currently, most metatranscriptomic analyses employ techniques to deplete the 740 rRNA sequences with the aim of increasing the number of non-rRNA reads in the datasets, 741 as concentrations of rRNA are not consistently correlated with microbial growth and can 742 differ between closely related taxa (Blazewicz et al. 2013). Another reason for rRNA 743 depletion is that rRNA transcripts are highly abundant (~90% of all the read data) and this 744 may bring issues to the analysis of mRNA sequences (Fuyong Li and Le Luo Guan 2017). 745 Thus, metatranscriptomic analysis requires either computational binning of the rRNA 746 genes to identify and remove ribosomal sequences (using total RNA- or mRNA-based 747 methods) or use of kits (probes) to deplete rRNA pre-sequencing (Huws et al. 2018). In 748 the first alternative, computational programs such as SortMeRNA (Kopylova et al. 2012) 749 can be used to sort the filtered metatranscriptomic reads (total RNA-based methods) into 750 fragments of 16S rRNA for taxonomic identification using pipelines like Mothur (Schloss 751 et al. 2009, Fuyong Li et al. 2016) or to obtain mRNA sequences for microbial 752 classification using software like Kraken (Neves et al. 2017, Wood and Salzberg 2014). 753 The second alternative is the use of kits to remove prokaryotic and eukaryotic rRNAs, but 754 a large variety of kits are required to remove rRNAs from complex microbiomes such as 755 the rumen, and this may be costly and laborious (Huws et al. 2018). Additionally, the time 756 required to perform the removal of ribosomal sequences using the kits may cause partial 757 RNA degradation and thus introduce biases in the downstream analysis (Huws et al. 758 2018). Despite these limitations, bespoke kits used to remove rumen-derived microbial 759 rRNAs were described in a study on dairy cows in France where 18 newly ribosomal 760 depletion probes were designed to remove rumen microbial rRNAs and covered a 761 significant proportion of the rumen bacterial, archaeal, fungal, and protozoal populations 762 enriched for non-rRNA reads (Comtet-Marre et al. 2017). More recently, Li et al. (2019a) 763

25 suggested that mRNA-enriched metatranscriptomics should be used for the study of 764 specific genes and/or metabolic pathways with low expression levels, while total RNA- 765 based metatranscriptomics are best for linking compositional and functional profiles of the 766 rumen microbiota to host phenotypes. 767

While metatranscriptomic analysis is a powerful tool to identify which organisms 768 are present and genes that are expressed in a sample, one major challenge is the de 769 novo assembly of the sequencing data to obtain a more reliable annotation of the 770 expressed genetic content of the metatranscriptome (Davids et al. 2016). In addition to 771 the challenges associated with identifying unique transcripts through assembly, 772 metatranscriptomic analysis does not account for variations in the translation and turnover 773 rates of transcripts transcribed into proteins, and thus functional activities captured by 774 metatranscriptomics are poorly inferred from correlations between the gene content 775 (enzyme abundance) and the transcript expression (transcript abundance) (Rosnow et al. 776 2017). Thus further investigations in protein detection could offer additional information 777 about the microbial activity and protein expression in the rumen. 778

1.4.4 Metaproteomics 779

Metaproteomics is a method used to characterize the entire protein content of 780 environmental samples at a given point in time and was introduced by Wilmes and Bond 781 (2004). The method indiscriminately identifies proteins from the environmental samples 782 (metagenomes) and is performed using liquid-chromatography-based separation 783 techniques coupled to mass spectrometry (Marchesi and Ravel 2015). Inference of 784 proteins from the identified peptides and the determination of the taxonomic origin and 785 function of these proteins can then be achieved using protein alignment tools such as 786 UniPept (Mesuere et al. 2012). This method has been successfully used to detect proteins 787 and to identify physiological responses to changes observed in various environmental 788 conditions such as in soil sediments (Benndorf et al. 2009), rhizosphere (Wu et al. 2011) 789 and human distal gut microbiota (Verberkmoes et al. 2008). However, a limited number 790 of metaproteomic studies have been published for the rumen. The technical issues that 791 have been presented to explain the shortcomings of metaproteomics for predicting the 792 function of rumen proteins include the interference of polyphenolic compounds of the diets 793

26 with protein isolation procedures. Phenolic compounds (e.g., tannins and humic acids) 794 co-precipitate with proteins and alter gel mobility, resulting in unresolved smears rather 795 than distinct protein bands, and thus limit the resolution of the peptides from the rumen 796 microbiome (Snelling and Wallace 2017). 797

Despite these limitations, Snelling and Wallace (2017) identified rumen microbial 798 proteins, such as actin, alpha and beta tubulins, and axonemal isoforms dynein light chain, 799 which are involved in the locomotion of ciliate protozoa. Removal of protozoa from digesta 800 before protein extraction revealed the prokaryotic metaproteome and the results showed 801 a predominance of enzymes of the central metabolism originating from the Firmicutes and 802 Bacteroidetes phyla (Snelling and Wallace 2017). Another possibility is to combine 803 shotgun-metaproteomics with 16S rRNA amplicon-based methods to unveil the 804 metaproteome expressed by the digesta-associated microbiota in cattle undergoing 805 dietary changes. By using this approach, Deusch et al. (2017) identified over 8000 806 bacterial and 350 archaeal proteins in the rumen, further improving our understanding of 807 the complicated interplay among rumen microbes, proteins and their adaptation to various 808 fermentation substrates (e.g., starch, cellulose, hemicellulose). More rumen microbiome 809 research is likely to be published in the upcoming years as bioinformatic and technical 810 progress enhance the metaproteomic coverage and maximize the focus on microbiome- 811 derived peptides. 812

1.4.5 Future perspectives in omics technologies to study the rumen microbiome 813

Each of the previously described technological approaches are powerful tools in 814 assessing the microbial composition and functional potential of the rumen microbiome, 815 but our understanding could be further enhanced if the information generated by those 816 technologies were integrated into multivariate models (Huws et al. 2018). However, this 817 integration is a complicated process owing to the complex and heterogeneous nature of 818 the available datasets generated by a wide variety of omics approaches (Fan et al. 2014). 819 Although challenging, there are instances of success in the literature showing the benefits 820 of integrating a varied array of data types generated from omics technologies in rumen 821 microbiome studies. Hart et al. (2018) compared metaproteomic data with published 822 genomic datasets of the rumen microbiome and improved protein identification that could 823

27 be impossible without integrating the information generated by the different datasets. 824 Advances in statistical methods could also offer more opportunities to integrate large- 825 scale molecular omics datasets and assess microbial interactions at multiple functional 826 levels. 827

Despite these developments, there is still a need to increase the number of 828 representative isolates in the microbial culture collections of rumen origin. The 829 “Hungate1000” project recently sequenced 420 representatives of rumen microbes, and 830 thus provided significant input of data for the scientific community (Seshadri et al. 2018). 831 Notwithstanding all of these efforts, it has been estimated that only 3.6% of the OTUs (61 832 out of 1,698 OTUs) of the sequenced rumen samples have representative isolates in the 833 Ribosomal Database project and only 117 bacterial species (not including different strains) 834 of rumen origin are available from international culture collections (Nordberg et al. 2014, 835 Seshadri et al. 2018, Zehavi et al. 2018, Cole et al. 2014). Thus, it is imperative the 836 adoption of complementary approaches (e.g., single-cell genomics) to sequence the DNA 837 of the whole microbiota in order to study the genomes of as-yet uncultured species from 838 the rumen. Single-cell genomics analysis aims to characterize the genomic variability 839 among individual cells, and thus it could be used to reconstruct cellular ancestries in the 840 form of a lineage tree (Shapiro et al. 2013). In addition to single-cell genomics 841 technologies, metabolomics could be used to profile (qualitatively and quantitatively) the 842 metabolites of a microbial community to prospect novel compounds and metabolic 843 pathways in the rumen microbiome (Yi et al. 2016, Deusch et al. 2017, Deusch et al. 844 2015). These technologies are in constant development, and the continued technical and 845 analytical advances in the field of molecular biology are likely to cause the emergence of 846 new technologies in the coming years. 847

1.4.6 Future perspectives in molecular techniques to investigate lignocellulolytic 848 enzymes 849

While metagenomics and metatranscriptomics have been successful to identify novel 850 enzymes responsible for lignocellulose degradation, these techniques have limitations to 851 evaluate changes in protein abundance, isoform expression, turnover rates, and post- 852 translational modifications and interactions (Larance and Lamond 2015). These 853

28 limitations have been overcome by high-throughput screening methods developed in 854 recent years, and with the progress in robotics and functional assays. In discovery-based 855 studies, quantitative MS and MS/MS measurements and nuclear magnetic resonance 856 spectroscopy are the most popular methods to identify enzymes and metabolites involved 857 in lignocellulose degradation (Rosnow et al. 2017). Efficient screening tools such as 858 fluorescence-activated cell sorting in combination with substrate-induced gene 859 expression have been developed to perform direct measurements of active enzyme in a 860 sample and to identify enzyme functions necessary for lignocellulose breakdown (Distaso 861 et al. 2017). These tools are still in their infancy, and as the new technical and analytical 862 advancements are achieved, we will see a continued breakthrough in our understanding 863 of how microbial enzymes accomplish lignocellulose breakdown. 864

1.5 Statistical challenges associated with the studies of the rumen microbiome 865

Datasets generated using omics technologies discussed in previous sections are 866 inherently compositional, a feature which is known to be problematic and should not be 867 ignored by data analysts (Pawlowsky‐Glahn et al. 2015, Aitchison 1982, Fernandes et 868 al. 2013, Warton et al. 2012, Gloor et al. 2017, Pearson 1897). Compositional data is a 869 type of dataset comprised in a mathematical space known as simplex space, where the 870 features (OTUs, genes, etc.) in each sample hold proportions of a unit varying between 871 0 and 1 (Aitchison 1982). Unlike the simplex space, the Euclidean space does not exhibit 872 constraints between 0 and 1, but can accept any real number along its dimensions 873 (Fernandes et al. 2014). Thus, the analysis of microbiome data requires statistical 874 methods accounting for the simplex structure of compositional datasets and excludes 875 standard statistical techniques (including Pearson correlations, Principal Component 876 Analysis, linear regression, etc.) that use the assumptions of the Euclidian space 877 (Pearson 1897, Aitchison 1982, Lovell et al. 2015, Gloor and Reid 2016, Fernandes et al. 878 2014). Despite these limitations, those traditional statistical methods are still commonly 879 used by the scientific community. 880

The original problem in analyzing compositional data was first identified by 881 Pearson in 1897 (Pearson 1897) when he realized that the count values per feature in a 882 compositional data are not independent, with the value of one feature necessarily 883

29 restricting the value of at least one other feature (Fernandes et al. 2014). This property 884 can lead to negative correlation biases and false univariate inferences in compositional 885 data, which renders invalid any correlation- or covariance-based results (Pearson 1897, 886 Aitchison 1982, Fernandes et al. 2013). An easy analogy to explain this distortion is the 887 “see-saw effect”, in which a change in the abundance of one feature results in a biased 888 correlation between the other features (one goes up, another goes down). In addition to 889 the possibility of obtaining spurious results, investigators should acknowledge that the 890 relationship between absolute abundance in the environment and the relative abundance 891 after sequencing is not equivalent in compositional datasets because the number of reads 892 obtained for a sample is determined by the capacity of the instrument and not by the 893 actual number of molecules of DNA in the environment (Mandal et al. 2015, Gloor and 894 Reid 2016). Therefore, compositional datasets are very different from datasets composed 895 of ordinary numbers that can take any value, and treating high-throughput sequencing 896 data as compositional is rather intuitive if one considers that the number of counts in such 897 datasets reflects the proportion of counts per feature per sample multiplied by the 898 sequencing depth (Fernandes et al. 2014, Gloor et al. 2017). 899

Sequencing depth (the total number of counts observed) between samples is 900 another significant confounder of the analysis, as abundance issues arise around the 901 variation in the number of sequences obtained for each sample. Rarefying or subsampling 902 the read counts of each sample to a defined level across samples excludes lower 903 abundant features (OTUs, genes, etc.) leading to a loss of precision (McMurdie and 904 Holmes 2014). If the researcher chooses to use the entire dataset (without rarefying), 905 s/he must account for the magnitude of sequence depth between samples and usually 906 needs to employ a transformation or scaling method (e.g., trimmed mean of M values – 907 TMM and the median methods) (McMurdie and Holmes 2014, Weiss et al. 2017, 908 Robinson et al. 2010, Anders and Huber 2010, Robinson and Oshlack 2010, Lovell et al. 909 2015, Love et al. 2014). Methods for the identification of differentially abundant OTUs 910 associated with a given phenotype or treatment should not use models that apply Poisson 911 distribution because it is too restrictive to deal with overdispersion (Anders and Huber 912 2010). To address the overdispersion problem, researchers have proposed the use of 913 negative binomial distributions, but this tends to increase the false discovery rate arising 914

30 from the compositional nature of the data (Lovell et al. 2015, Anders and Huber 2010, 915 Gloor et al. 2017). Taken altogether, the data analyst should be careful while analyzing 916 microbiome data as it exhibits a compositional structure that needs to be dealt with 917 appropriately in the statistical analysis. 918

1.5.1 Alternative techniques to study microbiome data 919

As outlined above, data collected from high-throughput sequencing platforms present 920 challenges to ecological and statistical analysis, and to circumvent these issues, 921 alternative statistical methods have been developed to substitute the standard statistical 922 approaches in the analysis of compositional data. The reader is directed to a review by 923 Gloor et al. (2017) who provided a more in-depth discussion of the application of these 924 techniques in the analysis of microbiome data as alternatives to the standard statistical 925 approaches. These alternative methods include initial normalization of the count data 926 using log-ratio transformations (centered or isometric) rather than rarefaction. Another 927 significant change in the analysis steps is the replacement of beta diversity analysis using 928 Aitchison calculations of distances for Bray Curtis. Researchers have also suggested 929 substituting philr transform for the unifrac distance metric when analyzing phylogenetic 930 trees (Silverman et al. 2017), and that beta diversity variance is visualized based on 931 compositional principal component biplot rather than principal co-ordinate plots (Gloor et 932 al. 2017). Correlation analysis to assess the extent to which a pair of random variables 933 are proportional should be performed with appropriate metrics such as the “goodness-of- 934 fit to proportionality” statistic ϕ rather than Pearson or Spearman correlations (Lovell et al. 935 2015). Finally, identification of differentially abundant features and microbial signatures 936 has been advised using ANCOM (Mandal et al. 2015) and MixMC (Cao et al. 2016). 937 These two methods will be briefly discussed below. 938

1.5.1.1 ANCOM 939

The Analysis of Composition of Microbiomes (ANCOM) (Mandal et al. 2015) is a statistical 940 procedure that compares the Aitchison’s log-ratio of the abundance of each taxon with 941 the abundance of all remaining taxa one at a time. Then, differential tests (e.g., Mann- 942 Whitney U, ANOVA, ANOVA with Linear Mixed Effect Models, Friedman, Kruskal-Wallis, 943 Wilcoxon tests) are calculated on each log ratio to reveal differences in the relative 944

31 abundance of a taxon between two ecosystems. If there are “m” taxa, then for each taxon 945 ANCOM performs “m-1” tests and the final significance of each test for a taxon is 946 determined using Benjamini-Hochberg (Benjamini and Hochberg 1995) algorithm to 947 control for false discovery rates. For each taxon, ANCOM counts the number of tests 948 among the m-1 tests and obtains a count random variable W that represents the number 949 of null among the m-1 tests that are rejected. To deal with the sparsity of the data, 950 ANCOM uses an arbitrary pseudo count value of 0.001 to replace the zero counts and 951 calculate the log-ratios. For drawing inferences regarding taxon abundance in the 952 ecosystem, ANCOM has been suggested as a reliable method to control the identification 953 of false positives and has been recently incorporated into the QIIME2 pipeline (Caporaso 954 et al. 2010) (https://qiime2.org). ANCOM was recently implemented in a bioinformatic 955 pipeline developed by our group (Neves et al. 2017) to detect differentially abundant taxa 956 identified by Kraken (Wood and Salzberg 2014) and Mothur (Schloss et al. 2009) in the 957 rumen metatranscriptome. 958

1.5.1.2 MixMC 959

MixMC (Multivariate Statistical Framework to Gain Insight into Microbial Communities) 960 (Cao et al. 2016) is a framework that takes into account the inherent characteristics of 961 microbiome data (sparsity and compositionality) to identify microbial signatures 962 associated with their environment, and it is currently implemented in the package 963 mixOmics (Rohart et al. 2017). In MixMC, the method Sparse Partial Least Square 964 Discriminant Analysis (sPLS-DA) is associated with centered log-ratio (CLR) 965 transformations to project the data from a simplex space to a Euclidian space and 966 includes a multilevel decomposition for repeated measurements designs that are 967 commonly encountered in microbiome studies (Cao et al. 2016). To account for subject 968 variability, the data variance is decomposed into within variation (due to habitat) and 969 between-subject variation. This is an appropriate analytical step towards detecting subtle 970 differences between samples when high inter-subject variability is present due to 971 sampling repeatedly performed on the same subjects and in multiple habitats (Westerhuis 972 et al. 2010, Liquet et al. 2012, Cao et al. 2016). Before the datasets are log-transformed 973 (CLR) and analyzed by the sPLS-DA models, preprocessing and normalization (e.g., total 974

32 sum scaling) steps are performed to account for the sparsity of the dataset and uneven 975 sequencing depths across samples. 976

1.5.2 Current challenges when comparing results across studies 977

While next-generation sequencing resulted in an explosion of publications exploring the 978 rumen microbial diversity and functions in the last decades, interpretation of the data 979 generated across multiple studies are still hampered by the lack of standardization in the 980 bioinformatic and statistical procedures employed by the different research groups. As 981 discussed previously, differences among studies exist for DNA extraction methods, 982 primers, PCR cycling parameters, and downstream bioinformatic analysis which make 983 comparisons across studies problematic and impractical at the moment. One instance of 984 this problem appeared when the rumen microbiome of efficient cattle was compared 985 across studies in order to find consensus microbial genes that could serve as global 986 biomarkers for predicting ruminant feed efficiency and methane emissions. Huws et al. 987 (2018) reported that microbial gene correlations with RFI described by Shabat et al. (2016) 988 overlapped with those of Fuyong Li and Le Luo Guan (2017) only in relation to a lower 989 abundance of genes involved in amino acid metabolism in the rumen of feed efficient 990 animals. However, genes related to methanogenesis did not show a consensus between 991 the datasets of Shabat et al. (2016) and Fuyong Li and Le Luo Guan (2017), indicating 992 that a standardization in the analysis is needed to compare across studies and promote 993 reproducibility of the results. 994

Considering the limitations of the omics technologies and the lack of awareness to 995 use robust statistical tools to analyze compositional data, there is an urgent need for 996 guidelines and practices that standardize rumen microbiome studies from the wet 997 laboratory to the publication stage. Aspects of the analysis that are important to 998 standardize include the methods of OTU picking, the databases (public or customized) 999 and algorithms for taxonomy classification, cutoffs for taxa inclusion/exclusion, and the 1000 statistical methods used to analyze the microbiome data (Goodrich et al. 2017). 1001 Irrespective, only after the standardization of the workflow (from sample collection to the 1002 analysis) is completed, there will be more reliable comparisons of results across studies. 1003

1.6. Knowledge gaps, hypothesis and objectives 1004

The significant advances in omics technologies in recent years has revolutionized our 1005 understanding of the rumen microbiome, its role in feed efficiency, and the degree to 1006 which it is influenced by the host genetic background. However, despite these steps 1007 forward, there remain substantial gaps in our knowledge. The broad aim of this thesis 1008 was to fill the gaps concerning the driving forces that influence the relationship between 1009 the rumen microbiota and host individual variation, and how their interactive effects on 1010 animal productivity contribute to the identification of cattle with improved feed efficiency. 1011 Moreover, this thesis fills the knowledge gaps concerning the impact of mRNA-based 1012 metatranscriptomics on the analysis of rumen taxonomic profiles. Further assessing such 1013 methodologies will contribute to elucidate the link between these taxonomic profiles and 1014 feed efficiency at the RNA level. Another gap in the literature is the lack of a strategy for 1015 the discovery of lignocellulosic enzymes through the targeted functional profiling of 1016 CAZyme families, as the discovery of these biocatalysts is usually performed using 1017 metagenomic screening or metatranscriptomic analysis that do not prioritize the 1018 identification of novel enzymes based on their ecological relevance in the microbiome. 1019 Therefore, further investigations of the microbial and functional dynamics of the rumen 1020 metatranscriptome are needed, and examining how they relate to the ruminant ability to 1021 degrade lignocellulosic biomass will be critical in designing innovative strategies to 1022 discover unknown enzymes for the breakdown, biosynthesis or modification of 1023 lignocellulosic biomass. 1024

The overall hypotheses for this thesis were that the stratification of dietary 1025 responses obtained from the magnitude of change in baseline rumen microbiota can be 1026 used to identify feed efficient cattle, and mRNA-based metatranscriptomic methods can 1027 be applied to characterize the microbial composition, diversity and functional profiles of 1028 the rumen microbiome in cattle with different feed efficiencies. Moreover, I hypothesized 1029 that feed efficiency affects the structure of the rumen microbiota and subsequently the 1030 expression of enzymes associated with lignocellulose degradation. The research 1031 presented in this thesis contributes to our understanding of the dynamics of the rumen 1032 microbiome in cattle undergoing dietary changes and also sheds light on the factors within 1033

34 the microbiome and the host that maximize the degradation of lignocellulosic biomass. 1034 The objectives of this thesis were as follows: 1035

1. To investigate the dynamics of rumen microorganisms in cattle raised under different 1036 feeding regimens and understand the relationship among the abundance of these 1037 microorganisms, host individuality and phenotypic traits; 1038

2. To compare and contrast methodologies (mRNA- vs. total rRNA-based methods) to 1039 assess the taxonomic profiles of the rumen microbiota and to investigate the impact of 1040 the comparative analysis of both analytical approaches on the rumen microbial 1041 classification obtained from cattle exhibiting different feed efficiencies; 1042

3. To characterize active microbial functional signatures differentiating breeds of beef 1043 cattle, and identify specific taxonomic microbial groups and functions associated with feed 1044 efficiency; 1045 4. To identify and validate biological associations (composition and functions) between 1046 the rumen metatranscriptome and feed efficiency in beef cattle, and use this knowledge 1047 to discover new enzymes for lignocellulose degradation. 1048 1049 1.7 Literature Cited 1050

Aitchison, J. (1982). The statistical-analysis of compositional data. Journal of the Royal 1051 Statistical Society Series B-Methodological 44(2), 139-177. 1052 Akin, D.E., Borneman, W.S., and Windham, W.R. (1988). Rumen fungi - morphological 1053 types from georgia cattle and the attack on forage cell-walls. Biosystems 21(3-4), 1054 385-391. doi: 10.1016/0303-2647(88)90037-8. 1055 Alemu, A.W., Vyas, D., Manafiazar, G., Basarab, J.A., and Beauchemin, K.A. (2017). 1056 Enteric methane emissions from low– and high–residual feed intake beef heifers 1057 measured using GreenFeed and respiration chamber techniques12. Journal of 1058 Animal Science 95, 3727-3737. doi: 10.2527/jas.2017.1501. 1059 Alexandratos, N., and Bruinsma, J. (2012). World Agriculture Towards 2030/2050: The 1060 2012 Revision. FAO - Food and Agriculture Organization of the United Nations. 1061 Anders, S., and Huber, W. (2010). Differential expression analysis for sequence count 1062 data. Genome Biology 11(10), R106. doi: 10.1186/gb-2010-11-10-r106. 1063 AOAC (2000). Official Methods of Analysis (17th). Gaithersburg, MD: Association of 1064 Official Analytical Chemists. 1065 Archer, J.A., Richardson, E.C., Herd, R.M., and Arthur, P.F. (1999). Potential for selection 1066 to improve efficiency of feed use in beef cattle: a review. Australian Journal of 1067 Agricultural Research 50(2), 147-162. doi: https://doi.org/10.1071/A98075. 1068

Artzi, L., Bayer, E.A., and Moraïs, S. (2016). Cellulosomes: bacterial nanomachines for 1069 dismantling plant polysaccharides. Nature Reviews Microbiology 15, 83. doi: 1070 10.1038/nrmicro.2016.164. 1071 Bader, J., Mast-Gerlach, E., Popovic, M.K., Bajpai, R., and Stahl, U. (2010). Relevance 1072 of microbial coculture fermentations in biotechnology. Journal of Applied 1073 Microbiology 109(2), 371-387. doi: 10.1111/j.1365-2672.2009.04659.x. 1074 Bae, H.D., McAllister, T.A., Kokko, E.G., Leggett, F.L., Yanke, L.J., Jakober, K.D., et al. 1075 (1997). Effect of silica on the colonization of rice straw by ruminal bacteria. Animal 1076 Feed Science and Technology 65(1), 165-181. doi: https://doi.org/10.1016/S0377- 1077 8401(96)01093-0. 1078 Bashiardes, S., Godneva, A., Elinav, E., and Segal, E. (2018). Towards utilization of the 1079 human genome and microbiome for personalized nutrition. Current Opinion in 1080 Biotechnology 51, 57-63. doi: 10.1016/j.copbio.2017.11.013. 1081 Bates, D., Mächler, M., Bolker, B., and Walker, S. (2015). Fitting Linear Mixed-Effects 1082 Models Using lme4. 2015 67(1), 48. doi: 10.18637/jss.v067.i01. 1083 Bayer, E.A., Chanzy, H., Lamed, R., and Shoham, Y. (1998). Cellulose, cellulases and 1084 cellulosomes. Current Opinion in Structural Biology 8(5), 548-557. doi: 1085 https://doi.org/10.1016/S0959-440X(98)80143-7. 1086 Bayley, D.P., Correia, J.D., Jarrell, K.F., and Thomas, N.A. (1999). Recent Excitement 1087 about the Archaea: The Archaea are valuable for studying basic biological 1088 questions and have novel biotechnology applications. BioScience 49(7), 530-541. 1089 doi: 10.2307/1313474. 1090 Beg, Q., Kapoor, M., Mahajan, L., and Hoondal, G. (2001). Microbial xylanases and their 1091 industrial applications: a review. Applied Microbiology and Biotechnology 56(3), 1092 326-338. doi: 10.1007/s002530100704. 1093 Belanche, A., Abecia, L., Holtrop, G., Guada, J.A., Castrillo, C., de la Fuente, G., et al. 1094 (2011). Study of the effect of presence or absence of protozoa on rumen 1095 fermentation and microbial protein contribution to the chyme1. Journal of Animal 1096 Science 89(12), 4163-4174. doi: 10.2527/jas.2010-3703. 1097 Belanche, A., Doreau, M., Edwards, J.E., Moorby, J.M., Pinloche, E., and Newbold, C.J. 1098 (2012). Shifts in the Rumen Microbiota Due to the Type of Carbohydrate and Level 1099 of Protein Ingested by Dairy Cattle Are Associated with Changes in Rumen 1100 Fermentation. Journal of Nutrition 142(9), 1684-1692. doi: 10.3945/jn.112.159574. 1101 Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate - a practical 1102 and powerful approach to multiple testing. Journal of the Royal Statistical Society 1103 Series B-Methodological 57(1), 289-300. 1104 Benndorf, D., Vogt, C., Jehmlich, N., Schmidt, Y., Thomas, H., Woffendin, G., et al. 1105 (2009). Improving protein extraction and separation methods for investigating the 1106 metaproteome of anaerobic benzene communities within sediments. 1107 Biodegradation 20(6), 737-750. doi: 10.1007/s10532-009-9261-3. 1108 Berry, D.P., and Crowley, J.J. (2013). CELL BIOLOGY SYMPOSIUM: Genetics of feed 1109 efficiency in dairy and beef cattle1. Journal of Animal Science 91(4), 1594-1613. 1110 doi: 10.2527/jas.2012-5862. 1111 Bhat, M.K., and Bhat, S. (1997). Cellulose degrading enzymes and their potential 1112 industrial applications. Biotechnology Advances 15(3), 583-620. doi: 1113 https://doi.org/10.1016/S0734-9750(97)00006-2. 1114

Bhatia, L., Johri, S., and Ahmad, R. (2012). An economic and ecological perspective of 1115 ethanol production from renewable agro waste: a review. Amb Express 2. doi: 1116 10.1186/2191-0855-2-65. 1117 Blazewicz, S.J., Barnard, R.L., Daly, R.A., and Firestone, M.K. (2013). Evaluating rRNA 1118 as an indicator of microbial activity in environmental communities: limitations and 1119 uses. The Isme Journal 7, 2061. doi: 10.1038/ismej.2013.102. 1120 Borneman, W.S., Hartley, R.D., Morrison, W.H., Akin, D.E., and Ljungdahl, L.G. (1990). 1121 Feruloyl and p-coumaroyl esterase from anaerobic fungi in relation to plant cell 1122 wall degradation. Applied Microbiology and Biotechnology 33(3), 345-351. doi: 1123 10.1007/BF00164534. 1124 Brooker, J.D., O'Donovan, L.A., Skene, I., Clarke, K., Blackall, L., and Muslera, P. (1994). 1125 Streptococcus caprinus sp.nov., a tannin-resistant ruminal bacterium from feral 1126 goats. Letters in Applied Microbiology 18(6), 313-318. doi: 10.1111/j.1472- 1127 765X.1994.tb00877.x. 1128 Brulc, J.M., Antonopoulos, D.A., Berg Miller, M.E., Wilson, M.K., Yannarell, A.C., 1129 Dinsdale, E.A., et al. (2009). Gene-centric metagenomics of the fiber-adherent 1130 bovine rumen microbiome reveals forage specific glycoside hydrolases. 1131 Proceedings of the National Academy of Sciences 106(6), 1948-1953. doi: 1132 10.1073/pnas.0806191105. 1133 Bryant, M.P., and Burkey, L.A. (1953). Cultural Methods and Some Characteristics of 1134 Some of the More Numerous Groups of Bacteria in the Bovine Rumen. Journal of 1135 Dairy Science 36(3), 205-217. doi: https://doi.org/10.3168/jds.S0022-0302(53)91482-9. 1136 Bryant, M.P., and Robinson, I.M. (1961). An Improved Nonselective Culture Medium for 1137 Ruminal Bacteria and Its Use in Determining Diurnal Variation in Numbers of 1138 Bacteria in the Rumen. Journal of Dairy Science 44(8), 1446-1456. doi: 1139 https://doi.org/10.3168/jds.S0022-0302(61)89906-2. 1140 Burnham, K.P., Anderson, D.R., and Burnham, K.P. (2002). Model selection and 1141 multimodel inference : a practical information-theoretic approach. New York : 1142 Springer, c2002. 1143 2nd ed. 1144 Callahan, B.J., McMurdie, P.J., Rosen, M.J., Han, A.W., Johnson, A.J.A., and Holmes, 1145 S.P. (2016). DADA2: High-resolution sample inference from Illumina amplicon 1146 data. Nature Methods 13, 581. doi: 10.1038/nmeth.3869 1147 https://www.nature.com/articles/nmeth.3869 - supplementary-information. 1148 Cammack, K.M., Austin, K.J., Lamberson, W.R., Conant, G.C., and Cunningham, H.C. 1149 (2018). Tiny but mighty: The role of the rumen microbes in livestock production. J 1150 Anim Sci. doi: 10.1093/jas/skx053. 1151 Cantarel, B.L., Coutinho, P.M., Rancurel, C., Bernard, T., Lombard, V., and Henrissat, B. 1152 (2009). The Carbohydrate-Active EnZymes database (CAZy): an expert resource 1153 for Glycogenomics. Nucleic Acids Research 37, D233-D238. doi: 1154 10.1093/nar/gkn663. 1155 Cao, K.A.L., Costello, M.E., Lakis, V.A., Bartolo, F., Chua, X.Y., Brazeilles, R., et al. 1156 (2016). MixMC: A Multivariate Statistical Framework to Gain Insight into Microbial 1157 Communities. Plos One 11(8). doi: 10.1371/journal.pone.0160169. 1158

Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, 1159 E.K., et al. (2010). QIIME allows analysis of high-throughput community 1160 sequencing data. Nature Methods 7(5), 335-336. doi: 10.1038/nmeth.f.303. 1161 Carberry, C.A., Kenny, D.A., Han, S., McCabe, M.S., and Waters, S.M. (2012a). Effect of 1162 phenotypic residual feed intake and dietary forage content on the rumen microbial 1163 community of beef cattle. Appl Environ Microbiol 78(14), 4949-4958. doi: 1164 10.1128/aem.07759-11. 1165 Carberry, C.A., Kenny, D.A., Kelly, A.K., and Waters, S.M. (2014a). Quantitative analysis 1166 of ruminal methanogenic microbial populations in beef cattle divergent in 1167 phenotypic residual feed intake (RFI) offered contrasting diets. J Anim Sci 1168 Biotechnol 5(1), 41. doi: 10.1186/2049-1891-5-41. 1169 Carberry, C.A., Kenny, D.A., Sukkyan, H., McCabe, M.S., and Waters, S.M. (2012b). 1170 Effect of phenotypic residual feed intake and dietary forage content on the rumen 1171 microbial community of beef cattle. Applied and Environmental Microbiology 1172 78(14), 4949-4958. 1173 Carberry, C.A., Waters, S.M., Kenny, D.A., and Creevey, C.J. (2014b). Rumen 1174 methanogenic genotypes differ in abundance according to host residual feed 1175 intake phenotype and diet type. Appl Environ Microbiol 80(2), 586-594. doi: 1176 10.1128/aem.03131-13. 1177 Chafe, S.C. (1969). The fine structure of the collenchyma cell wall. Planta 90(1), 12-21. 1178 doi: 10.1007/BF00389292. 1179 Cheng, K.J., McCowan, R.P., and Costerton, J.W. (1979). Adherent epithelial bacteria in 1180 ruminants and their roles in digestive tract function. Am J Clin Nutr 32(1), 139-148. 1181 doi: 10.1093/ajcn/32.1.139. 1182 Ciucci, S., Ge, Y., Durán, C., Palladini, A., Martínez-Sánchez, L.M., Cannistraci, C.V., et 1183 al. (2017). Enlightening discriminative network functional modules behind principal 1184 component analysis separation in differential-omic science studies. Scientific 1185 Reports 7. doi: 10.1038/srep43946. 1186 Clark, J.H., Budarin, V., Deswarte, F.E.I., Hardy, J.J.E., Kerton, F.M., Hunt, A.J., et al. 1187 (2006). Green chemistry and the biorefinery: a partnership for a sustainable future. 1188 Green Chemistry 8(10), 853-860. doi: 10.1039/B604483M. 1189 Cole, J.R., Wang, Q., Fish, J.A., Chai, B., McGarrell, D.M., Sun, Y., et al. (2014). 1190 Ribosomal Database Project: data and tools for high throughput rRNA analysis. 1191 Nucleic Acids Research 42(D1), D633-D642. doi: 10.1093/nar/gkt1244. 1192 Comtet-Marre, S., Chaucheyras-Durand, F., Bouzid, O., Mosoni, P., Bayat, A.R., Peyret, 1193 P., et al. (2018). FibroChip, a Functional DNA Microarray to Monitor Cellulolytic 1194 and Hemicellulolytic Activities of Rumen Microbiota. Frontiers in Microbiology 1195 9(215). doi: 10.3389/fmicb.2018.00215. 1196 Comtet-Marre, S., Parisot, N., Lepercq, P., Chaucheyras-Durand, F., Mosoni, P., 1197 Peyretaillade, E., et al. (2017). Metatranscriptomics Reveals the Active Bacterial 1198 and Eukaryotic Fibrolytic Communities in the Rumen of Dairy Cow Fed a Mixed 1199 Diet. Frontiers in Microbiology 8(67). doi: 10.3389/fmicb.2017.00067. 1200 Creevey, C.J., Kelly, W.J., Henderson, G., and Leahy, S.C. (2014). Determining the 1201 culturability of the rumen bacterial microbiome. Microbial Biotechnology 7(5), 467- 1202 479. doi: 10.1111/1751-7915.12141. 1203

Crews, D.H., Jr., Shannon, N. H., Genswein, B. M. A., Crews, R. E., Johnson, C. M. and 1204 Kendrick, B. A (2005). Genetic parameters for net feed efficiency of beef cattle 1205 measured during postweaning growing versus finishing periods. Proc. West Sec. 1206 Am. Soc. Anim. Sci 54(54), 125-128. 1207 Dai, X., Tian, Y., Li, J., Su, X., Wang, X., Zhao, S., et al. (2015). Metatranscriptomic 1208 Analyses of Plant Cell Wall Polysaccharide Degradation by Microorganisms in the 1209 Cow Rumen. Applied and Environmental Microbiology 81(4), 1375. 1210 Darling, A.E., Jospin, G., Lowe, E., Matsen, F.A.I.V., Bik, H.M., and Eisen, J.A. (2014). 1211 PhyloSift: phylogenetic analysis of genomes and metagenomes. PeerJ 2, e243. 1212 doi: 10.7717/peerj.243. 1213 Davids, M., Hugenholtz, F., Martins dos Santos, V., Smidt, H., Kleerebezem, M., and 1214 Schaap, P.J. (2016). Functional Profiling of Unfamiliar Microbial Communities 1215 Using a Validated De Novo Assembly Metatranscriptome Pipeline. PLOS ONE 1216 11(1), e0146423. doi: 10.1371/journal.pone.0146423. 1217 Denman, S.E., Martinez Fernandez, G., Shinkai, T., Mitsumori, M., and McSweeney, C.S. 1218 (2015). Metagenomic analysis of the rumen microbial community following 1219 inhibition of methane formation by a halogenated methane analog. Frontiers in 1220 Microbiology 6(1087). doi: 10.3389/fmicb.2015.01087. 1221 Denman, S.E., and McSweeney, C.S. (2006). Development of a real-time PCR assay for 1222 monitoring anaerobic fungal and cellulolytic bacterial populations within the rumen. 1223 Fems Microbiology Ecology 58(3), 572-582. doi: 10.1111/j.1574- 1224 6941.2006.00190.x. 1225 Denman, S.E., Morgavi, D.P., and McSweeney, C.S. (2018). Review: The application of 1226 omics to rumen microbiota function. animal 12(s2), s233-s245. doi: 1227 10.1017/S175173111800229X. 1228 Deppenmeier, U. (2002). The unique biochemistry of methanogenesis. Prog Nucleic Acid 1229 Res Mol Biol 71, 223-283. 1230 DeSantis, T.Z., Hugenholtz, P., Larsen, N., Rojas, M., Brodie, E.L., Keller, K., et al. 1231 (2006). Greengenes, a chimera-checked 16S rRNA gene database and 1232 workbench compatible with ARB. Applied and Environmental Microbiology 72(7), 1233 5069-5072. doi: 10.1128/aem.03006-05. 1234 Deusch, S., Camarinha-Silva, A., Conrad, J., Beifuss, U., Rodehutscord, M., and Seifert, 1235 J. (2017). A Structural and Functional Elucidation of the Rumen Microbiome 1236 Influenced by Various Diets and Microenvironments. Frontiers in Microbiology 1237 8(1605). doi: 10.3389/fmicb.2017.01605. 1238 Deusch, S., Tilocca, B., Camarinha-Silva, A., and Seifert, J. (2015). News in livestock 1239 research — use of Omics-technologies to study the microbiota in the 1240 gastrointestinal tract of farm animals. Computational and Structural Biotechnology 1241 Journal 13, 55-63. doi: https://doi.org/10.1016/j.csbj.2014.12.005. 1242 Distaso, M.A., Tran, H., Ferrer, M., and Golyshin, P.N. (2017). "Metagenomic Mining of 1243 Enzyme Diversity," in Consequences of Microbial Interactions with Hydrocarbons, 1244 Oils, and Lipids: Production of Fuels and Chemicals, ed. S.Y. Lee. (Cham: 1245 Springer International Publishing), 245-269. 1246 Doi, R.H., and Kosugi, A. (2004). Cellulosomes: plant-cell-wall-degrading enzyme 1247 complexes. Nature Reviews Microbiology 2, 541. doi: 10.1038/nrmicro925. 1248

Dollhofer, V., Podmirseg, S.M., Callaghan, T.M., Griffith, G.W., and Fliegerova, K. (2015). 1249 Anaerobic Fungi and Their Potential for Biogas Production. Advances in 1250 biochemical engineering/biotechnology 151, 41-61. doi: 10.1007/978-3-319- 1251 21993-6_2. 1252 Duffield, T., Plaizier, J.C., Fairfield, A., Bagg, R., Vessie, G., Dick, P., et al. (2004). 1253 Comparison of techniques for measurement of rumen pH in lactating dairy cows. 1254 Journal of Dairy Science 87(1), 59-66. 1255 Edwards, J.E., McEwan, N.R., Travis, A.J., and Wallace, R.J. (2004). 16S rDNA library- 1256 based analysis of ruminal bacterial diversity. Antonie van Leeuwenhoek (3), 263. 1257 El Enshasy, H.A., Kandiyil, S.K., Malek, R., and Othman, N.Z. (2016). "Microbial 1258 Xylanases: Sources, Types, and Their Applications," in Microbial Enzymes in 1259 Bioconversions of Biomass, ed. V.K. Gupta. (Cham: Springer International 1260 Publishing), 151-213. 1261 Fan, J., Han, F., and Liu, H. (2014). Challenges of Big Data analysis. National Science 1262 Review 1(2), 293-314. doi: 10.1093/nsr/nwt032. 1263 Fernandes, A.D., Macklaim, J.M., Linn, T.G., Reid, G., and Gloor, G.B. (2013). ANOVA- 1264 Like Differential Expression (ALDEx) Analysis for Mixed Population RNA-Seq. 1265 PLOS ONE 8(7), e67019. doi: 10.1371/journal.pone.0067019. 1266 Fernandes, A.D., Reid, J.N.S., Macklaim, J.M., McMurrough, T.A., Edgell, D.R., and 1267 Gloor, G.B. (2014). Unifying the analysis of high-throughput sequencing datasets: 1268 characterizing RNA-seq, 16S rRNA gene sequencing and selective growth 1269 experiments by compositional data analysis. Microbiome 2(1), 15. doi: 1270 10.1186/2049-2618-2-15. 1271 Fernando, S.C., Purvis, H.T., II, Najar, F.Z., Sukharnikov, L.O., Krehbiel, C.R., Nagaraja, 1272 T.G., et al. (2010). Rumen Microbial Population Dynamics during Adaptation to a 1273 High-Grain Diet. Applied and Environmental Microbiology 76(22), 7482-7490. doi: 1274 10.1128/aem.00388-10. 1275 Ferrer, M., Beloqui, A., Timmis, K.N., and Golyshin, P.N. (2009). Metagenomics for 1276 Mining New Genetic Resources of Microbial Communities. Journal of Molecular 1277 Microbiology and Biotechnology 16(1-2), 109-123. doi: 10.1159/000142898. 1278 Ferrer, M., Golyshina, O.V., Chernikova, T.N., Khachane, A.N., Reyes-Duarte, D., 1279 Santos, V.A.P.M.D., et al. (2005). Novel hydrolase diversity retrieved from a 1280 metagenome library of bovine rumen microflora. Environmental Microbiology 1281 7(12), 1996-2010. doi: 10.1111/j.1462-2920.2005.00920.x. 1282 Finlay, B.J., Esteban, G., Clarke, K.J., Williams, A.G., Embley, T.M., and Hirt, R.P. (1994). 1283 Some rumen ciliates have endosymbiotic methanogens. FEMS Microbiology 1284 Letters 117(2), 157-161. 1285 Firkins, J.L., and Yu, Z. (2015). RUMINANT NUTRITION SYMPOSIUM: How to use data 1286 on the rumen microbiome to improve our understanding of ruminant nutrition. J 1287 Anim Sci 93(4), 1450-1470. doi: 10.2527/jas.2014-8754. 1288 Ford, C.W., Elliott, R., and Maynard, P.J. (1987). The effect of chlorite delignification on 1289 digestibility of some grass forages and on intake and rumen microbial activity in 1290 sheep fed barley straw. Journal of Agricultural Science 108, 129-136. 1291 Fouts, D.E., Szpakowski, S., Purushe, J., Torralba, M., Waterman, R.C., MacNeil, M.D., 1292 et al. (2012). Next generation sequencing to define prokaryotic and fungal diversity 1293

in the bovine rumen. PLoS One 7(11), e48289. doi: 1294 10.1371/journal.pone.0048289. 1295 Franzosa, E.A., Morgan, X.C., Segata, N., Waldron, L., Reyes, J., Earl, A.M., et al. (2014). 1296 Relating the metatranscriptome and metagenome of the human gut. Proceedings 1297 of the National Academy of Sciences 111(22), E2329-E2338. doi: 1298 10.1073/pnas.1319284111. 1299 Gao, A., Wang, H., Yang, J., and Shi, C. (2013). "The effects of elimination of fungi on 1300 microbial population and fiber degradation in sheep rumen," in Progress in 1301 Environmental Protection and Processing of Resource, Pts 1-4, eds. X. Tang, W. 1302 Zhong, D. Zhuang, C. Li & Y. Liu.), 224-231. 1303 Gharechahi, J., and Salekdeh, G.H. (2018). A metagenomic analysis of the camel 1304 rumen’s microbiome identifies the major microbes responsible for lignocellulose 1305 degradation and fermentation. Biotechnology for Biofuels 11(1), 216. doi: 1306 10.1186/s13068-018-1214-9. 1307 Gijzen, H.J., Lubberding, H.J., Gerhardus, M.J.T., and Vogels, G.D. (1988). Contribution 1308 of rumen protozoa to fibre degradation and cellulase activity in vitro. FEMS 1309 Microbiology Letters 53(1), 35-43. doi: 10.1111/j.1574-6968.1988.tb02645.x. 1310 Gloor, G.B., Macklaim, J.M., Pawlowsky-Glahn, V., and Egozcue, J.J. (2017). Microbiome 1311 Datasets Are Compositional: And This Is Not Optional. Frontiers in Microbiology 1312 8(2224). doi: 10.3389/fmicb.2017.02224. 1313 Gloor, G.B., and Reid, G. (2016). Compositional analysis: a valid approach to analyze 1314 microbiome high-throughput sequencing data. Canadian Journal of Microbiology 1315 62(8), 692-703. doi: 10.1139/cjm-2015-0821. 1316 Goel, G., Puniya, A.K., Aguilar, C.N., and Singh, K. (2005). Interaction of gut microflora 1317 with tannins in feeds. Naturwissenschaften 92(11), 497-503. doi: 10.1007/s00114- 1318 005-0040-7. 1319 Goodrich, J.K., Davenport, E.R., Clark, A.G., and Ley, R.E. (2017). The Relationship 1320 Between the Human Genome and Microbiome Comes into View. Annual Review 1321 of Genetics 51(1), 413-433. doi: 10.1146/annurev-genet-110711-155532. 1322 Gordon, G.L.R., and Phillips, M.W. (1992). Extracellular pectin lyase produced by 1323 Neocallimastix sp. LM1, a rumen anaerobic fungus. Letters in Applied Microbiology 1324 15(3), 113-115. doi: 10.1111/j.1472-765X.1992.tb00740.x. 1325 Gordon, G.L.R., and Phillips, M.W. (1993). Removal of anaerobic fungi from the rumen 1326 of sheep by chemical treatment and the effect on feed consumption and in-vivo 1327 fiber digestion. Letters in Applied Microbiology 17(5), 220-223. doi: 1328 10.1111/j.1472-765X.1993.tb01451.x. 1329 Gordon, G.L.R., and Phillips, M.W. (1998). The role of anaerobic gut fungi in ruminants. 1330 Nutrition Research Reviews 11(1), 133-168. doi: 10.1079/nrr19980009. 1331 Greve, L.C., Labavitch, J.M., and Hungate, R.E. (1984). alpha-L-arabinofuranosidase 1332 from Ruminococcus albus 8: purification and possible role in hydrolysis of alfalfa 1333 cell wall. Applied and Environmental Microbiology 47(5), 1135-1140. 1334 Grubb, J.A., and Dehority, B.A. (1975). Effects of an abrupt change in ration from all 1335 roughage to high concentrate upon rumen microbial numbers in sheep. Applied 1336 Microbiology 30(3), 404-412. 1337 Gruninger, R.J., Puniya, A.K., Callaghan, T.M., Edwards, J.E., Youssef, N., Dagar, S.S., 1338 et al. (2014). Anaerobic fungi (phylum Neocallimastigomycota): advances in 1339

understanding their taxonomy, life cycle, ecology, role and biotechnological 1340 potential. Fems Microbiology Ecology 90(1), 1-17. doi: 10.1111/1574-6941.12383. 1341 Guan, L.L., Nkrumah, J.D., Basarab, J.A., and Moore, S.S. (2008). Linkage of microbial 1342 ecology to phenotype: correlation of rumen microbial ecology to cattle's feed 1343 efficiency. Fems Microbiology Letters 288(1), 85-91. doi: 10.1111/j.1574- 1344 6968.2008.01343.x. 1345 Gunsett, F.C. (1984). Linear Index Selection to Improve Traits Defined as Ratios. Journal 1346 of Animal Science 59(5), 1185-1193. doi: 10.2527/jas1984.5951185x. 1347 Gupta, A., Kumar, S., Prasoodanan, V.P.K., Harish, K., Sharma, A.K., and Sharma, V.K. 1348 (2016a). Reconstruction of Bacterial and Viral Genomes from Multiple 1349 Metagenomes. Frontiers in Microbiology 7, 469. doi: 10.3389/fmicb.2016.00469. 1350 Gupta, A., Kumar, S., Prasoodanan, V.P.K., Harish, K., Sharma, A.K., and Sharma, V.K. 1351 (2016b). Reconstruction of Bacterial and Viral Genomes from Multiple 1352 Metagenomes. Frontiers in Microbiology 7(469). doi: 10.3389/fmicb.2016.00469. 1353 Hackmann, T.J., and Firkins, J.L. (2015). Maximizing efficiency of rumen microbial protein 1354 production. Frontiers in Microbiology 6. doi: 10.3389/fmicb.2015.00465. 1355 Hall, M.B. (2009). Determination of Starch, Including Maltooligosaccharides, in Animal 1356 Feeds: Comparison of Methods and a Method Recommended for AOAC 1357 Collaborative Study. Journal of Aoac International 92(1), 42-49. 1358 Handelsman, J., Rondon, M.R., Brady, S.F., Clardy, J., and Goodman, R.M. (1998). 1359 Molecular biological access to the chemistry of unknown soil microbes: a new 1360 frontier for natural products. Chemistry & Biology 5(10), R245-R249. doi: 1361 https://doi.org/10.1016/S1074-5521(98)90108-9. 1362 Hart, E.H., Creevey, C.J., Hitch, T., and Kingston-Smith, A.H. (2018). Meta-proteomics of 1363 rumen microbiota indicates niche compartmentalisation and functional dominance 1364 in a limited number of metabolic pathways between abundant bacteria. Scientific 1365 Reports 8(1), 10504. doi: 10.1038/s41598-018-28827-7. 1366 Healey, G.R., Murphy, R., Brough, L., Butts, C.A., and Coad, J. (2017). Interindividual 1367 variability in gut microbiota and host response to dietary interventions. Nutrition 1368 Reviews 75(12), 1059-1080. 1369 Hegarty, R.S., Goopy, J.P., Herd, R.M., and McCorkell, B. (2007). Cattle selected for 1370 lower residual feed intake have reduced daily methane production. J Anim Sci 1371 85(6), 1479-1486. doi: 10.2527/jas.2006-236. 1372 Henderson, G., Cox, F., Ganesh, S., Jonker, A., Young, W., and Janssen, P.H. (2015). 1373 Rumen microbial community composition varies with diet and host, but a core 1374 microbiome is found across a wide geographical range. Sci Rep 5, 14567. doi: 1375 10.1038/srep14567. 1376 Hernandez-Sanabria, E., Goonewardene, L.A., Wang, Z., Durunna, O.N., Moore, S.S., 1377 and Guan, L.L. (2012). Impact of Feed Efficiency and Diet on Adaptive Variations 1378 in the Bacterial Community in the Rumen Fluid of Cattle. Applied and 1379 Environmental Microbiology 78(4), 1203-1214. doi: 10.1128/aem.05114-11. 1380 Hess, M., Sczyrba, A., Egan, R., Kim, T.-W., Chokhawala, H., Schroth, G., et al. (2011). 1381 Metagenomic Discovery of Biomass-Degrading Genes and Genomes from Cow 1382 Rumen. Science 331(6016), 463-467. doi: 10.1126/science.1200387. 1383

Himmel, M.E., Ding, S.-Y., Johnson, D.K., Adney, W.S., Nimlos, M.R., Brady, J.W., et al. 1384 (2007). Biomass Recalcitrance: Engineering Plants and Enzymes for Biofuels 1385 Production. Science (5813). 1386 Ho, Y.W., Abdullah, N., and Jalaludin, S. (1988). Penetrating structures of anaerobic 1387 rumen fungi in cattle and swamp buffalo. Journal of General Microbiology 134(1), 1388 177-182. 1389 Ho, Y.W., and Barr, D.J.S. (1995). Classification of anaerobic gut fungi from herbivores 1390 with emphasis on rumen fungi from malaysia. Mycologia 87(5), 655-677. doi: 1391 10.2307/3760810. 1392 Hobson, P.N., and Stewart, C.S. (1997). The Rumen Microbial Ecosystem. London: 1393 Blackie Academic & Professional. 1394 Hogg, D., Pell, G., Dupree, P., Goubet, F., Martín-Orúe, S.M., Armand, S., et al. (2003). 1395 The modular architecture of Cellvibrio japonicus mannanases in glycoside 1396 hydrolase families 5 and 26 points to differences in their role in mannan 1397 degradation. The Biochemical journal 371(Pt 3), 1027-1043. doi: 1398 10.1042/BJ20021860. 1399 Hristov, A.N., Callaway, T.R., Lee, C., and Dowd, S.E. (2012). Rumen bacterial, archaeal, 1400 and fungal diversity of dairy cows in response to ingestion of lauric or myristic acid. 1401 J Anim Sci 90(12), 4449-4457. doi: 10.2527/jas.2011-4624. 1402 Hristov, A.N., Oh, J., Firkins, J.L., Dijkstra, J., Kebreab, E., Waghorn, G., et al. (2013). 1403 SPECIAL TOPICS — Mitigation of methane and nitrous oxide emissions from 1404 animal operations: I. A review of enteric methane mitigation options. Journal of 1405 Animal Science 91, 5045-5069. doi: 10.2527/jas.2013-6583. 1406 Hristov, A.N., Oh, J., Giallongo, F., Frederick, T.W., Harper, M.T., Weeks, H.L., et al. 1407 (2015). An inhibitor persistently decreased enteric methane emission from dairy 1408 cows with no negative effect on milk production. Proceedings of the National 1409 Academy of Sciences 112(34), 10663-10668. doi: 10.1073/pnas.1504124112. 1410 Hungate, R.E. (1966). The rumen and its microbes. New York and London: Academic 1411 Press. 1412 Hungate, R.E. (1967). Hydrogen as an intermediate in the rumen fermentation. Archiv für 1413 Mikrobiologie 59(1-3), 158-164. doi: 10.1007/BF00406327. 1414 Huws, S.A., Creevey, C.J., Oyama, L.B., Mizrahi, I., Denman, S.E., Popova, M., et al. 1415 (2018). Addressing Global Ruminant Agricultural Challenges Through 1416 Understanding the Rumen Microbiome: Past, Present, and Future. Frontiers in 1417 Microbiology 9(2161). doi: 10.3389/fmicb.2018.02161. 1418 Imelfort, M., Parks, D., Woodcroft, B.J., Dennis, P., Hugenholtz, P., and Tyson, G.W. 1419 (2014). GroopM: an automated tool for the recovery of population genomes from 1420 related metagenomes. PeerJ 2, e603. doi: 10.7717/peerj.603. 1421 IPCC (2006). Guidelines for National Greenhouse Gas Inventories. Japan: Institute for 1422 Global Environmental Strategies (IGES). 1423 Jami, E., Israel, A., Kotser, A., and Mizrahi, I. (2013). Exploring the bovine rumen bacterial 1424 community from birth to adulthood. Isme j 7(6), 1069-1079. doi: 1425 10.1038/ismej.2013.2. 1426 Jami, E., and Mizrahi, I. (2012). Composition and similarity of bovine rumen microbiota 1427 across individual animals. PLoS One 7(3), e33306. doi: 1428 10.1371/journal.pone.0033306. 1429

Janssen, P.H. (2010). Influence of hydrogen on rumen methane formation and 1430 fermentation balances through microbial growth kinetics and fermentation 1431 thermodynamics. Animal Feed Science and Technology 160(1-2), 1-22. doi: 1432 10.1016/j.anifeedsci.2010.07.002. 1433 Janssen, P.H., and Kirs, M. (2008a). Structure of the archaeal community of the rumen. 1434 Appl Environ Microbiol 74(12), 3619-3625. doi: 10.1128/aem.02812-07. 1435 Janssen, P.H., and Kirs, M. (2008b). Structure of the Archaeal Community of the Rumen. 1436 Applied & Environmental Microbiology 74(12), 3619-3625. doi: 1437 10.1128/AEM.02812-07. 1438 Jayanegara, A., Sarwono, K.A., Kondo, M., Matsui, H., Ridla, M., Laconi, E.B., et al. 1439 (2018). Use of 3-nitrooxypropanol as feed additive for mitigating enteric methane 1440 emissions from ruminants: a meta-analysis AU - Jayanegara, Anuraga. Italian 1441 Journal of Animal Science 17(3), 650-656. doi: 10.1080/1828051X.2017.1404945. 1442 Jin, D., Zhao, S., Zheng, N., Bu, D., Beckers, Y., Denman, S.E., et al. (2017). Differences 1443 in Ureolytic Bacterial Composition between the Rumen Digesta and Rumen Wall 1444 Based on ureC Gene Classification. Frontiers in Microbiology 8(385). doi: 1445 10.3389/fmicb.2017.00385. 1446 Johnson, D.E., Ferrell, C.L., and Jenkins, T.G. (2003). The history of energetic efficiency 1447 research: Where have we been and where are we going? Journal of Animal 1448 Science 81(13_suppl_1), E27-E38. doi: 10.2527/2003.8113_suppl_1E27x. 1449 Johnson, K.A., and Johnson, D.E. (1995). Methane emissions from cattle. Journal of 1450 Animal Science 73(8), 2483-2492. 1451 Jouany, J.P. (1995). Rumen microbial metabolism and ruminant digestion: J-P. Jouany 1452 (Ed). 1991. INRA, 147 rue de l'Université, 75338 Paris cedex 07, France. ISBN 2 1453 7380 0345 1. 374 pp. Text in English. 350 FF. Livestock Production Science 44(3), 1454 284. doi: https://doi.org/10.1016/0301-6226(95)90065-9. 1455 Kamke, J., Kittelmann, S., Soni, P., Li, Y., Tavendale, M., Ganesh, S., et al. (2016). 1456 Rumen metagenome and metatranscriptome analyses of low methane yield sheep 1457 reveals a Sharpea-enriched microbiome characterised by lactic acid formation and 1458 utilisation. Microbiome 4(1), 56-56. 1459 Kang, D.D., Froula, J., Egan, R., and Wang, Z. (2015). MetaBAT, an efficient tool for 1460 accurately reconstructing single genomes from complex microbial communities. 1461 PeerJ 3, e1165. doi: 10.7717/peerj.1165. 1462 Kenters, N., Henderson, G., Jeyanathan, J., Kittelmann, S., and Janssen, P.H. (2011). 1463 Isolation of previously uncultured rumen bacteria by dilution to extinction using a 1464 new liquid culture medium. Journal of Microbiological Methods 84(1), 52-60. doi: 1465 https://doi.org/10.1016/j.mimet.2010.10.011. 1466 Kim, M., Yu, Z., and Morrison, M. (2011). Status of the phylogenetic diversity census of 1467 ruminal microbiomes. FEMS Microbiology Ecology 76(1), 49-63. doi: 1468 10.1111/j.1574-6941.2010.01029.x. 1469 Kittelmann, S., Devente, S.R., Kirk, M.R., Seedorf, H., Dehority, B.A., and Janssen, P.H. 1470 (2015a). Phylogeny of Intestinal Ciliates, Including <span class="named- 1471 content genus-species" id="named-content-1">Charonina 1472 ventriculi</span>, and Comparison of Microscopy and 18S rRNA Gene 1473 Pyrosequencing for Rumen Ciliate Community Structure Analysis. Applied and 1474 Environmental Microbiology 81(7), 2433. doi: 10.1128/AEM.03697-14. 1475

Kittelmann, S., Devente, S.R., Kirk, M.R., Seedorf, H., Dehority, B.A., and Janssen, P.H. 1476 (2015b). Phylogeny of Intestinal Ciliates, Including Charonina ventriculi, and 1477 Comparison of Microscopy and 18S rRNA Gene Pyrosequencing for Rumen 1478 Ciliate Community Structure Analysis. Appl Environ Microbiol 81(7), 2433-2444. 1479 doi: 10.1128/aem.03697-14. 1480 Kittelmann, S., Pinares-Patiño, C.S., Seedorf, H., Kirk, M.R., Ganesh, S., McEwan, J.C., 1481 et al. (2014). Two Different Bacterial Community Types Are Linked with the Low- 1482 Methane Emission Trait in Sheep. PLoS ONE 9(7), 1-9. doi: 1483 10.1371/journal.pone.0103171. 1484 Kittelmann, S., Seedorf, H., Walters, W.A., Clemente, J.C., Knight, R., Gordon, J.I., et al. 1485 (2013a). Simultaneous amplicon sequencing to explore co-occurrence patterns of 1486 bacterial, archaeal and eukaryotic microorganisms in rumen microbial 1487 communities. PLoS One 8(2), e47879. doi: 10.1371/journal.pone.0047879. 1488 Kittelmann, S., Seedorf, H., Walters, W.A., Clemente, J.C., Knight, R., Gordon, J.I., et al. 1489 (2013b). Simultaneous Amplicon Sequencing to Explore Co-Occurrence Patterns 1490 of Bacterial, Archaeal and Eukaryotic Microorganisms in Rumen Microbial 1491 Communities. Plos One 8(2). doi: 10.1371/journal.pone.0047879. 1492 Knapp, J.R., Laur, G.L., Vadas, P.A., Weiss, W.P., and Tricarico, J.M. (2014). Invited 1493 review: Enteric methane in dairy cattle production: Quantifying the opportunities 1494 and impact of reducing emissions. Journal of Dairy Science 97(6), 3231-3261. doi: 1495 http://dx.doi.org/10.3168/jds.2013-7234. 1496 Koch, R.M., Swiger, L.A., Chambers, D., and Gregory, K.E. (1963). Efficiency of Feed 1497 Use in Beef Cattle. Journal of Animal Science 22(2), 486-494. doi: 1498 10.2527/jas1963.222486x. 1499 Koetschan, C., Kittelmann, S., Lu, J., Al-Halbouni, D., Jarvis, G.N., Mueller, T., et al. 1500 (2014). Internal Transcribed Spacer 1 Secondary Structure Analysis Reveals a 1501 Common Core throughout the Anaerobic Fungi (Neocallimastigomycota). Plos 1502 One 9(3). doi: 10.1371/journal.pone.0091928. 1503 Kong, R.S., Liang, G., Chen, Y., Stothard, P., and Guan le, L. (2016). Transcriptome 1504 profiling of the rumen epithelium of beef cattle differing in residual feed intake. BMC 1505 Genomics 17, 592. doi: 10.1186/s12864-016-2935-4. 1506 Kopylova, E., Noé, L., and Touzet, H. (2012). SortMeRNA: fast and accurate filtering of 1507 ribosomal RNAs in metatranscriptomic data. Bioinformatics (Oxford, England) 1508 28(24), 3211-3217. doi: 10.1093/bioinformatics/bts611. 1509 Korver, S. (1988). Genetic aspects of feed intake and feed efficiency in dairy cattle: A 1510 review. Livestock Production Science 20(1), 1-13. doi: https://doi.org/10.1016/0301- 1511 6226(88)90049-8. 1512 Kosugi, A., Murashima, K., and Doi, R.H. (2001). Characterization of Xylanolytic Enzymes 1513 in Clostridium cellulovorans: Expression of Xylanase Activity 1514 Dependent on Growth Substrates. Journal of Bacteriology 183(24), 7037-7043. 1515 doi: 10.1128/jb.183.24.7037-7043.2001. 1516 Kozich, J.J., Westcott, S.L., Baxter, N.T., Highlander, S.K., and Schloss, P.D. (2013). 1517 Development of a dual-index sequencing strategy and curation pipeline for 1518 analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. 1519 Applied And Environmental Microbiology 79(17), 5112-5120. doi: 1520 10.1128/AEM.01043-13. 1521

Krause, D.O., Nagaraja, T.G., Wright, A.D.G., and Callaway, T.R. (2013). Board-invited 1522 review: Rumen microbiology: Leading the way in microbial ecology1,2. Journal of 1523 Animal Science 91(1), 331-341. doi: 10.2527/jas.2012-5567. 1524 Kumar, S., Choudhury, P.K., Carro, M.D., Griffith, G.W., Dagar, S.S., Puniya, M., et al. 1525 (2014). New aspects and strategies for methane mitigation from ruminants. Appl 1526 Microbiol Biotechnol 98(1), 31-44. doi: 10.1007/s00253-013-5365-0. 1527 Lam, K.N., Cheng, J., Engel, K., Neufeld, J.D., and Charles, T.C. (2015). Current and 1528 future resources for functional metagenomics. Frontiers in Microbiology 6(1196). 1529 doi: 10.3389/fmicb.2015.01196. 1530 Langille, M.G.I., Zaneveld, J., Caporaso, J.G., McDonald, D., Knights, D., Reyes, J.A., et 1531 al. (2013). Predictive functional profiling of microbial communities using 16S rRNA 1532 marker gene sequences. Nature Biotechnology 31, 814. doi: 10.1038/nbt.2676 1533 https://www.nature.com/articles/nbt.2676 - supplementary-information. 1534 Larance, M., and Lamond, A.I. (2015). Multidimensional proteomics for cell biology. 1535 Nature Reviews Molecular Cell Biology 16, 269. doi: 10.1038/nrm3970 1536 https://www.nature.com/articles/nrm3970 - supplementary-information. 1537 Leahy, S.C., Kelly, W.J., Altermann, E., Ronimus, R.S., Yeoman, C.J., Pacheco, D.M., et 1538 al. (2010). The Genome Sequence of the Rumen Methanogen 1539 Methanobrevibacter ruminantium Reveals New Possibilities for Controlling 1540 Ruminant Methane Emissions. Plos One 5(1). doi: 10.1371/journal.pone.0008926. 1541 Lee, S.S., Ha, J.K., and Cheng, K.J. (2000). Relative contributions of bacteria, protozoa, 1542 and fungi to in vitro degradation of orchard grass cell walls and their interactions. 1543 Applied and Environmental Microbiology 66(9), 3807-3813. doi: 1544 10.1128/aem.66.9.3807-3813.2000. 1545 Li, F., and Guan, L.L. (2017a). Metatranscriptomic profiling reveals linkages between the 1546 active rumen microbiome and feed efficiency in beef cattle. Applied And 1547 Environmental Microbiology. doi: 10.1128/AEM.00061-17. 1548 Li, F., and Guan, L.L. (2017b). Metatranscriptomic Profiling Reveals Linkages between 1549 the Active Rumen Microbiome and Feed Efficiency in Beef Cattle. Appl Environ 1550 Microbiol 83(9). doi: 10.1128/aem.00061-17. 1551 Li, F., Henderson, G., Sun, X., Cox, F., Janssen, P.H., and Guan le, L. (2016a). 1552 Taxonomic Assessment of Rumen Microbiota Using Total RNA and Targeted 1553 Amplicon Sequencing Approaches. Front Microbiol 7, 987. doi: 1554 10.3389/fmicb.2016.00987. 1555 Li, F., Henderson, G., Sun, X., Cox, F., Janssen, P.H., and Guan, L.L. (2016b). 1556 Taxonomic Assessment of Rumen Microbiota Using Total RNA and Targeted 1557 Amplicon Sequencing Approaches. Frontiers in Microbiology 7(987). doi: 1558 10.3389/fmicb.2016.00987. 1559 Li, F., Hitch, T.C.A., Chen, Y., Creevey, C.J., and Guan, L.L. (2019). Comparative 1560 metagenomic and metatranscriptomic analyses reveal the breed effect on the 1561 rumen microbiome and its associations with feed efficiency in beef cattle. 1562 Microbiome 7(1), 6. doi: 10.1186/s40168-019-0618-5. 1563 Li, M., Penner, G.B., Hernandez-Sanabria, E., Oba, M., and Guan, L.L. (2009). Effects of 1564 sampling location and time, and host animal on assessment of bacterial diversity 1565 and fermentation parameters in the bovine rumen. Journal of Applied Microbiology 1566 107(6), 1924-1934. doi: 10.1111/j.1365-2672.2009.04376.x. 1567

Li, R.W., Connor, E.E., Li, C., Baldwin Vi, R.L., and Sparks, M.E. (2012). Characterization 1568 of the rumen microbiota of pre-ruminant calves using metagenomic tools. Environ 1569 Microbiol 14(1), 129-139. doi: 10.1111/j.1462-2920.2011.02543.x. 1570 Li, Z.P., Wright, A.D.G., Si, H.Z., Wang, X.X., Qian, W.X., Zhang, Z.G., et al. (2016c). 1571 Changes in the rumen microbiome and metabolites reveal the effect of host 1572 genetics on hybrid crosses. Environmental Microbiology Reports 8(6), 1016-1023. 1573 doi: 10.1111/1758-2229.12482. 1574 Lin, C., Stahl, D.A., and Raskin, L. (1997). Microbial community structure in 1575 gastrointestinal tracts of domestic animals: comparative analyses using rRNA- 1576 targeted oligonucleotide probes. FEMS Microbiology Ecology 22(4), 281-294. doi: 1577 10.1111/j.1574-6941.1997.tb00380.x. 1578 Liquet, B., Cao, K.-A.L., Hocini, H., and Thiébaut, R. (2012). A novel approach for 1579 biomarker selection and the integration of repeated measures experiments from 1580 two assays. BMC Bioinformatics 13(1), 325. doi: 10.1186/1471-2105-13-325. 1581 Liu, J.h., Zhang, M.l., Zhang, R.y., Zhu, W.y., and Mao, S.y. (2016). Comparative studies 1582 of the composition of bacterial microbiota associated with the ruminal content, 1583 ruminal epithelium and in the faeces of lactating dairy cows. Microbial 1584 Biotechnology 9(2), 257-268. doi: doi:10.1111/1751-7915.12345. 1585 Lombard, V., Golaconda Ramulu, H., Drula, E., Coutinho, P.M., and Henrissat, B. (2014). 1586 The carbohydrate-active enzymes database (CAZy) in 2013. Nucleic Acids 1587 Research 42(D1), D490-D495. doi: 10.1093/nar/gkt1178. 1588 Love, M.I., Huber, W., and Anders, S. (2014). Moderated estimation of fold change and 1589 dispersion for RNA-seq data with DESeq2. Genome biology 15(12), 550-550. doi: 1590 10.1186/s13059-014-0550-8. 1591 Lovell, D., Pawlowsky-Glahn, V., Egozcue, J.J., Marguerat, S., and Bähler, J. (2015). 1592 Proportionality: A Valid Alternative to Correlation for Relative Data. PLoS 1593 Computational Biology 11(3), 1-12. doi: 10.1371/journal.pcbi.1004075. 1594 Malmuthuge, N., Li, M., Chen, Y., Fries, P., Griebel, P.J., Baurhoo, B., et al. (2012). 1595 Distinct commensal bacteria associated with ingesta and mucosal epithelium in 1596 the gastrointestinal tracts of calves and chickens. Fems Microbiology Ecology 1597 79(2), 337-347. doi: 10.1111/j.1574-6941.2011.01220.x. 1598 Mandal, S., Van Treuren, W., White, R.A., Eggesbo, M., Knight, R., and Peddada, S.D. 1599 (2015). Analysis of composition of microbiomes: a novel method for studying 1600 microbial composition. Microbial Ecology in Health and Disease 26, 27663-Article 1601 No.: 27663. doi: 10.3402/mehd.v26.27663. 1602 Marchesi, J.R., and Ravel, J. (2015). The vocabulary of microbiome research: a proposal. 1603 Microbiome 3(1), 31. doi: 10.1186/s40168-015-0094-5. 1604 Martin, C., Morgavi, D.P., and Doreau, M. (2010). Methane mitigation in ruminants: from 1605 microbe to the farm scale. Animal 4(3), 351-365. doi: 1606 10.1017/s1751731109990620. 1607 Martínez, I., Jaehyoung, K., Duffy, P.R., Schlegel, V.L., and Walter, J. (2010). Resistant 1608 Starches Types 2 and 4 Have Differential Effects on the Composition of the Fecal 1609 Microbiota in Human Subjects. PLoS ONE 5(11), 1-11. doi: 1610 10.1371/journal.pone.0015046. 1611 Martinez-Fernandez, G., Denman, S.E., Cheung, J., and McSweeney, C.S. (2017). 1612 Phloroglucinol Degradation in the Rumen Promotes the Capture of Excess 1613

Hydrogen Generated from Methanogenesis Inhibition. Frontiers in Microbiology 1614 8(1871). doi: 10.3389/fmicb.2017.01871. 1615 Martinez-Fernandez, G., Denman, S.E., Yang, C., Cheung, J., Mitsumori, M., and 1616 McSweeney, C.S. (2016). Methane Inhibition Alters the Microbial Community, 1617 Hydrogen Flow, and Fermentation Response in the Rumen of Cattle. Frontiers in 1618 Microbiology 7(1122). doi: 10.3389/fmicb.2016.01122. 1619 Matthews, C., Crispie, F., Lewis, E., Reid, M., O’Toole, P.W., and Cotter, P.D. (2018). 1620 The rumen microbiome: a crucial consideration when optimising milk and meat 1621 production and nitrogen utilisation efficiency AU - Matthews, Chloe. Gut Microbes, 1622 1-18. doi: 10.1080/19490976.2018.1505176. 1623 McAllister, T.A., Bae, H.D., Jones, G.A., and Cheng, K.J. (1994a). Microbial attachment 1624 and feed digestion in the rumen. Journal of Animal Science 72(11), 3004-3018. 1625 McAllister, T.A., Bae, H.D., Yanke, L.J., Cheng, K.J., and Muir, A. (1994b). Effect of 1626 condensed tannins from birdsfoot trefoil on endoglucanase activity and the 1627 digestion of cellulose filter paper by ruminal fungi. Can J Microbiol 40(4), 298-305. 1628 McAllister, T.A., Meale, S.J., Valle, E., Guan, L.L., Zhou, M., Kelly, W.J., et al. (2015). 1629 RUMINANT NUTRITION SYMPOSIUM: Use of genomics and transcriptomics to 1630 identify strategies to lower ruminal methanogenesis. J Anim Sci 93(4), 1431-1449. 1631 doi: 10.2527/jas.2014-8329. 1632 McCabe, M.S., Cormican, P., Keogh, K., O'Connor, A., O'Hara, E., Palladino, R.A., et al. 1633 (2015a). Illumina MiSeq Phylogenetic Amplicon Sequencing Shows a Large 1634 Reduction of an Uncharacterised Succinivibrionaceae and an Increase of the 1635 Methanobrevibacter gottschalkii Clade in Feed Restricted Cattle. PLoS One 10(7), 1636 e0133234. doi: 10.1371/journal.pone.0133234. 1637 McCabe, M.S., Cormican, P., Keogh, K., O’Connor, A., O’Hara, E., Palladino, R.A., et al. 1638 (2015b). Illumina MiSeq Phylogenetic Amplicon Sequencing Shows a Large 1639 Reduction of an Uncharacterised Succinivibrionaceae and an Increase of the 1640 Methanobrevibacter gottschalkii Clade in Feed Restricted Cattle. PLOS ONE 1641 10(7), e0133234. doi: 10.1371/journal.pone.0133234. 1642 McCann, J.C., Elolimy, A.A., and Loor, J.J. (2017). Rumen Microbiome, Probiotics, and 1643 Fermentation Additives. The Veterinary Clinics Of North America. Food Animal 1644 Practice. doi: 10.1016/j.cvfa.2017.06.009. 1645 McGovern, E., McCabe, M.S., Cormican, P., Popova, M., Keogh, K., Kelly, A.K., et al. 1646 (2017). Plane of nutrition affects the phylogenetic diversity and relative abundance 1647 of transcriptionally active methanogens in the bovine rumen. Scientific Reports 1648 7(1), 13047. doi: 10.1038/s41598-017-13013-y. 1649 McMurdie, P.J., and Holmes, S. (2014). Waste Not, Want Not: Why Rarefying Microbiome 1650 Data Is Inadmissible. PLOS Computational Biology 10(4), e1003531. doi: 1651 10.1371/journal.pcbi.1003531. 1652 McNeil, M., Darvill, A.G., Fry, S.C., and Albersheim, P. (1984). Structure and Function of 1653 the Primary Cell Walls of Plants. Annual Review of Biochemistry 53(1), 625-663. 1654 doi: 10.1146/annurev.bi.53.070184.003205. 1655 McSweeney, C., and Mackie, R.I. (2012). Microorganisms and ruminant digestion: state 1656 of knowledge, trends and future prospects. 1657 Mesuere, B., Devreese, B., Debyser, G., Aerts, M., Vandamme, P., and Dawyndt, P. 1658 (2012). Unipept: Tryptic Peptide-Based Biodiversity Analysis of Metaproteome 1659

Samples. Journal of Proteome Research 11(12), 5773-5780. doi: 1660 10.1021/pr300576s. 1661 Miron, J., Ben-Ghedalia, D., and Morrison, M. (2001). Invited Review: Adhesion 1662 Mechanisms of Rumen Cellulolytic Bacteria. Journal of Dairy Science 84(6), 1294- 1663 1309. doi: 10.3168/jds.S0022-0302(01)70159-2. 1664 Montanholi, Y.R., Swanson, K.C., Palme, R., Schenkel, F.S., McBride, B.W., Lu, D., et al. 1665 (2010). Assessing feed efficiency in beef steers through feeding behavior, infrared 1666 thermography and glucocorticoids. Animal 4(5), 692-701. doi: 1667 10.1017/s1751731109991522. 1668 Morgavi, D., Kelly, W.J., Janssen, P.H., and Attwood, G.T. (2013). "Rumen microbial 1669 (meta) genomics and its application to ruminant production".). 1670 Morgavi, D.P., Forano, E., Martin, C., and Newbold, C.J. (2010). Microbial ecosystem and 1671 methanogenesis in ruminants. Animal 4(7), 1024-1036. doi: 1672 10.1017/s1751731110000546. 1673 Morgavi, D.P., Kelly, W.J., Janssen, P.H., and Attwood, G.T. (2012a). Rumen microbial 1674 (meta)genomics and its application to ruminant production. animal 7(s1), 184-201. 1675 doi: 10.1017/S1751731112000419. 1676 Morgavi, D.P., Martin, C., Jouany, J.P., and Ranilla, M.J. (2012b). Rumen protozoa and 1677 methanogenesis: not a simple cause-effect relationship. Br J Nutr 107(3), 388-397. 1678 doi: 10.1017/s0007114511002935. 1679 Morrison, M., Murray, R.M., and Boniface, A.N. (1990). Nutrient metabolism and rumen 1680 microorganisms in sheep fed a poor-quality tropical grass hay supplemented with 1681 sulfate. Journal of Agricultural Science 115, 269-275. 1682 Myer, P.R., Smith, T.P.L., Wells, J.E., Kuehn, L.A., and Freetly, H.C. (2015a). Rumen 1683 Microbiome from Steers Differing in Feed Efficiency. PLOS ONE 10(6), e0129174. 1684 doi: 10.1371/journal.pone.0129174. 1685 Myer, P.R., Smith, T.P.L., Wells, J.E., Kuehn, L.A., and Freetly, H.C. (2015b). Rumen 1686 Microbiome from Steers Differing in Feed Efficiency. Plos One 10(6). doi: 1687 10.1371/journal.pone.0129174. 1688 Nations, U. (2015). World Population Prospects [Online]. [Accessed]. 1689 Network, I.F.C. (2016). "Long-term dairy outlook", in: IFCN - the dairy research network.). 1690 Neves, A.L.A., Li, F., Ghoshal, B., McAllister, T., and Guan, L.L. (2017). Enhancing the 1691 Resolution of Rumen Microbial Classification from Metatranscriptomic Data Using 1692 Kraken and Mothur. Frontiers in Microbiology 8(2445). doi: 1693 10.3389/fmicb.2017.02445. 1694 Newbold, C.J., de la Fuente, G., Belanche, A., Ramos-Morales, E., and McEwan, N.R. 1695 (2015). The Role of Ciliate Protozoa in the Rumen. Front Microbiol 6, 1313. doi: 1696 10.3389/fmicb.2015.01313. 1697 Nocker, A., Burr, M., and Camper, A.K. (2007). Genotypic Microbial Community Profiling: 1698 A Critical Technical Review. Microbial Ecology 54(2), 276-289. doi: 1699 10.1007/s00248-006-9199-5. 1700 Nordberg, H., Cantor, M., Dusheyko, S., Hua, S., Poliakov, A., Shabalov, I., et al. (2014). 1701 The genome portal of the Department of Energy Joint Genome Institute: 2014 1702 updates. Nucleic acids research 42(Database issue), D26-D31. doi: 1703 10.1093/nar/gkt1069. 1704

NRC (2016). Nutrient Requirements of Beef Cattle. Washington, D.C., USA: National 1705 Academy Press. 1706 Olfert, E., Cross, B., and McWilliams, A. (1993). Guide to the Care and Use of 1707 Experimental Steers. Ottawa, ON, Canada: Canadian Council on Animal Care. 1708 Orpin, C.G. (1975). Studies on rumen flagellate neocallimastix-frontalis. Journal of 1709 General Microbiology 91(DEC), 249-262. 1710 Orpin, C.G., and Joblin, K.N. (1997). "The rumen anaerobic fungi," in The Rumen 1711 Microbial Ecosystem, eds. P.N. Hobson & C.S. Stewart. Springer Netherlands), 1712 140-195. 1713 Ozkose, E., Thomas, B.J., Davies, D.R., Griffith, G.W., and Theodorou, M.K. (2001). 1714 Cyllamyces aberensis gen.nov sp.nov., a new anaerobic gut fungus with branched 1715 sporangiophores isolated from cattle. Canadian Journal of Botany-Revue 1716 Canadienne De Botanique 79(6), 666-673. 1717 Palackal, N., Lyon, C.S., Zaidi, S., Luginbühl, P., Dupree, P., Goubet, F., et al. (2007). A 1718 multifunctional hybrid glycosyl hydrolase discovered in an uncultured microbial 1719 consortium from ruminant gut. Applied Microbiology and Biotechnology 74(1), 113- 1720 124. doi: 10.1007/s00253-006-0645-6. 1721 Parks, D.H., Imelfort, M., Skennerton, C.T., Hugenholtz, P., and Tyson, G.W. (2015). 1722 CheckM: Assessing the quality of microbial genomes recovered from isolates, 1723 single cells, and metagenomes. Genome Research 25(7), 1043-1055. doi: 1724 10.1101/gr.186072.114. 1725 Parmar, N.R., Solanki, J.V., Patel, A.B., Shah, T.M., Patel, A.K., Parnerkar, S., et al. 1726 (2014). Metagenome of Mehsani buffalo rumen microbiota: an assessment of 1727 variation in feed-dependent phylogenetic and functional classification. J Mol 1728 Microbiol Biotechnol 24(4), 249-261. doi: 10.1159/000365054. 1729 Patil, K.R., Haider, P., Pope, P.B., Turnbaugh, P.J., Morrison, M., Scheffer, T., et al. 1730 (2011). Taxonomic metagenome sequence assignment with structured output 1731 models. Nature Methods 8, 191. doi: 10.1038/nmeth0311-191 1732 https://www.nature.com/articles/nmeth0311-191 - supplementary-information. 1733 Patra, A., Park, T., Kim, M., and Yu, Z. (2017). Rumen methanogens and mitigation of 1734 methane emission by anti-methanogenic compounds and substances. Journal of 1735 Animal Science and Biotechnology 8(1), 13. doi: 10.1186/s40104-017-0145-9. 1736 Paul, S.S., Yu, Z., Bu, D., Xu, J., and Hyde, K.D. (2018). A phylogenetic census of global 1737 diversity of gut anaerobic fungi and a new taxonomic framework. Fungal Diversity 1738 89(1), 253-266. doi: 10.1007/s13225-018-0396-6. 1739 Pawlowsky‐ Glahn, V., Egozcue, J.J., and Tolosana‐ Delgado, R. (2015). Modelling and 1740 Analysis of Compositional Data. 1741 Pearson, K. (1897). Mathematical contributions to the theory of evolution.—On a form of 1742 spurious correlation which may arise when indices are used in the measurement 1743 of organs. Proceedings of the Royal Society of London 60(359-367), 489-498. doi: 1744 10.1098/rspl.1896.0076. 1745 Petri, R.M., Schwaiger, T., Penner, G.B., Beauchemin, K.A., Forster, R.J., McKinnon, J.J., 1746 et al. (2013). Characterization of the Core Rumen Microbiome in Cattle during 1747 Transition from Forage to Concentrate as Well as during and after an Acidotic 1748 Challenge. Plos One 8(12). doi: 10.1371/journal.pone.0083424. 1749

Pickering, N.K., Oddy, V.H., Basarab, J., Cammack, K., Hayes, B., Hegarty, R.S., et al. 1750 (2015a). Animal board invited review: genetic possibilities to reduce enteric 1751 methane emissions from ruminants. Animal 9(9), 1431-1440. doi: 1752 10.1017/s1751731115000968. 1753 Pickering, N.K., Oddy, V.H., Basarab, J., Cammack, K., Hayes, B., Hegarty, R.S., et al. 1754 (2015b). Animal board invited review: genetic possibilities to reduce enteric 1755 methane emissions from ruminants. Animal : an international journal of animal 1756 bioscience 9(9), 1431-1440. doi: 10.1017/S1751731115000968. 1757 Poelaert, C., Nollevaux, G., Boudry, C., Taminiau, B., Nezer, C., Daube, G., et al. (2017). 1758 Reducing agent can be omitted in the incubation medium of the batch in vitro 1759 fermentation model of the pig intestines. animal 12(6), 1154-1164. doi: 1760 10.1017/S1751731117002749. 1761 Poretsky, R., Rodriguez, R.L., Luo, C., Tsementzi, D., and Konstantinidis, K.T. (2014). 1762 Strengths and limitations of 16S rRNA gene amplicon sequencing in revealing 1763 temporal microbial community dynamics. PLoS One 9(4), e93827. doi: 1764 10.1371/journal.pone.0093827. 1765 Poulsen, M., Schwab, C., Borg Jensen, B., Engberg, R.M., Spang, A., Canibe, N., et al. 1766 (2013). Methylotrophic methanogenic Thermoplasmata implicated in reduced 1767 methane emissions from bovine rumen. Nature Communications 4, 1428. doi: 1768 10.1038/ncomms2432 1769 https://www.nature.com/articles/ncomms2432 - supplementary-information. 1770 Qi, M., Wang, P., O'Toole, N., Barboza, P.S., Ungerfeld, E., Leigh, M.B., et al. (2011a). 1771 Snapshot of the Eukaryotic Gene Expression in Muskoxen Rumen-A 1772 Metatranscriptomic Approach. Plos One 6(5). doi: 10.1371/journal.pone.0020521. 1773 Qi, M., Wang, P., Selinger, L.B., Yanke, L.J., Forster, R.J., and McAllister, T.A. (2011b). 1774 Isolation and characterization of a ferulic acid esterase (Fae1A) from the rumen 1775 fungus Anaeromyces mucronatus. Journal of Applied Microbiology 110(5), 1341- 1776 1350. doi: 10.1111/j.1365-2672.2011.04990.x. 1777 Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., et al. (2013). The 1778 SILVA ribosomal RNA gene database project: improved data processing and web- 1779 based tools. Nucleic Acids Research 41(Database issue), D590-D596. doi: 1780 10.1093/nar/gks1219. 1781 Ramírez, L., Arrizon, J., Sandoval, G., Cardador, A., Bello-Mendoza, R., Lappe, P., et al. 1782 (2008). A New Microplate Screening Method for the Simultaneous Activity 1783 Quantification of Feruloyl Esterases, Tannases, and Chlorogenate Esterases. 1784 Applied Biochemistry and Biotechnology 151(2), 711-723. doi: 10.1007/s12010- 1785 008-8319-8. 1786 Rashamuse, K.J., Burton, S.G., and Cowan, D.A. (2007). A novel recombinant ethyl 1787 ferulate esterase from Burkholderia multivorans. Journal of Applied Microbiology 1788 103(5), 1610-1620. doi: 10.1111/j.1365-2672.2007.03394.x. 1789 Ribeiro, G.O., Gruninger, R.J., Badhan, A., and McAllister, T.A. (2016). Mining the rumen 1790 for fibrolytic feed enzymes. Animal Frontiers 6(2), 20-26. doi: 10.2527/af.2016- 1791 0019. 1792 Rico, D.E., Preston, S.H., Risser, J.M., and Harvatine, K.J. (2015). Rapid changes in key 1793 ruminal microbial populations during the induction of and recovery from diet- 1794

induced milk fat depression in dairy cows. British Journal of Nutrition 114(3), 358- 1795 367. doi: 10.1017/s0007114515001865. 1796 Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010). edgeR: a Bioconductor 1797 package for differential expression analysis of digital gene expression data. 1798 Bioinformatics 26(1), 139-140. doi: 10.1093/bioinformatics/btp616. 1799 Robinson, M.D., and Oshlack, A. (2010). A scaling normalization method for differential 1800 expression analysis of RNA-seq data. Genome Biology 11(3), R25. doi: 1801 10.1186/gb-2010-11-3-r25. 1802 Rodríguez, H., Curiel, J.A., Landete, J.M., de las Rivas, B., de Felipe, F.L., Gómez- 1803 Cordovés, C., et al. (2009). Food phenolics and lactic acid bacteria. International 1804 Journal of Food Microbiology 132(2), 79-90. doi: 1805 https://doi.org/10.1016/j.ijfoodmicro.2009.03.025. 1806 Roehe, R., Dewhurst, R.J., Duthie, C.-A., Rooke, J.A., McKain, N., Ross, D.W., et al. 1807 (2016). Bovine Host Genetic Variation Influences Rumen Microbial Methane 1808 Production with Best Selection Criterion for Low Methane Emitting and Efficiently 1809 Feed Converting Hosts Based on Metagenomic Gene Abundance. Plos Genetics 1810 12(2). doi: 10.1371/journal.pgen.1005846. 1811 Rohart, F., Gautier, B., Singh, A., and Lê Cao, K.-A. (2017). mixOmics: An R package for 1812 ‘omics feature selection and multiple data integration. PLoS Computational Biology 1813 13(11), 1-19. doi: 10.1371/journal.pcbi.1005752. 1814 Rojas-Downing, M.M., Nejadhashemi, A.P., Harrigan, T., and Woznicki, S.A. (2017). 1815 Climate change and livestock: Impacts, adaptation, and mitigation. Climate Risk 1816 Management 16, 145-163. doi: https://doi.org/10.1016/j.crm.2017.02.001. 1817 Romain, G., Thomas, T., and Margaret, E.K.E. (2009). Ecological Conditions Affect 1818 Evolutionary Trajectory in a Predator-Prey System. Evolution (3), 641. doi: 1819 10.1111/j.1558-5646.2008.00559.x. 1820 Romero-Perez, A., Okine, E.K., McGinn, S.M., Guan, L.L., Oba, M., Duval, S.M., et al. 1821 (2014). The potential of 3-nitrooxypropanol to lower enteric methane emissions 1822 from beef cattle. Journal of Animal Science 92(10), 4682-4693. doi: 1823 10.2527/jas.2014-7573. 1824 Rosnow, J.J., Anderson, L.N., Nair, R.N., Baker, E.S., and Wright, A.T. (2017). Profiling 1825 microbial lignocellulose degradation and utilization by emergent omics 1826 technologies. Critical Reviews in Biotechnology 37(5), 626-640. doi: 1827 10.1080/07388551.2016.1209158. 1828 Russell, J.B., and Rychlik, J.L. (2001). Factors that alter rumen microbial ecology. 1829 Science 292(5519), 1119-1122. doi: 10.1126/science.1058830. 1830 Sabree, Z.L., Rondon, M.R., and Handelsman, J. (2009). "Metagenomics," in 1831 Encyclopedia of Microbiology (Third Edition), ed. M. Schaechter. (Oxford: 1832 Academic Press), 622-632. 1833 Saha, B.C. (2000). α-l-Arabinofuranosidases: biochemistry, molecular biology and 1834 application in biotechnology. Biotechnology Advances 18(5), 403-423. doi: 1835 https://doi.org/10.1016/S0734-9750(00)00044-6. 1836 Saha, B.C. (2003). Hemicellulose bioconversion. Journal of Industrial Microbiology and 1837 Biotechnology 30(5), 279-291. doi: 10.1007/s10295-003-0049-x. 1838 Salonen, A., Lahti, L., Salojärvi, J., Holtrop, G., Korpela, K., Duncan, S.H., et al. (2014). 1839 Impact of diet and individual variation on intestinal microbiota composition and 1840

fermentation products in obese men. ISME Journal: Multidisciplinary Journal of 1841 Microbial Ecology 8(11), 2218-2230. doi: 10.1038/ismej.2014.63. 1842 Sanchez, C. (2009). Lignocellulosic residues: Biodegradation and bioconversion by fungi. 1843 Biotechnology Advances 27(2), 185-194. doi: 10.1016/j.biotechadv.2008.11.001. 1844 Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., et al. 1845 (2009). Introducing mothur: Open-Source, Platform-Independent, Community- 1846 Supported Software for Describing and Comparing Microbial Communities. 1847 Applied & Environmental Microbiology 75(23), 7537-7541. doi: 1848 10.1128/AEM.01541-09. 1849 Seedorf, H., Kittelmann, S., Henderson, G., and Janssen, P.H. (2014). RIM-DB: a 1850 taxonomic framework for community structure analysis of methanogenic archaea 1851 from the rumen and other intestinal environments. Peerj 2. doi: 10.7717/peerj.494. 1852 Sehgal, J.P., Jit, D., Puniya, A.K., and Singh, K. (2008). Influence of anaerobic fungal 1853 administration on growth, rumen fermentation and nutrient digestion in female 1854 buffalo calves. Journal of Animal and Feed Sciences 17(4), 510-518. doi: 1855 10.22358/jafs/66678/2008. 1856 Seshadri, R., Leahy, S.C., Attwood, G.T., Teh, K.H., Lambie, S.C., Cookson, A.L., et al. 1857 (2018). Cultivation and sequencing of rumen microbiome members from the 1858 Hungate1000 Collection. Nature Biotechnology 36(4), 359-367. doi: 1859 10.1038/nbt.4110. 1860 Shaani, Y., Zehavi, T., Eyal, S., Miron, J., and Mizrahi, I. (2018). Microbiome niche 1861 modification drives diurnal rumen community assembly, overpowering individual 1862 variability and diet effects. The ISME Journal. doi: 10.1038/s41396-018-0203-0. 1863 Shabat, S.K.B., Sasson, G., Doron-Faigenboim, A., Durman, T., Yaacoby, S., Berg Miller, 1864 M.E., et al. (2016). Specific microbiome-dependent mechanisms underlie the 1865 energy harvest efficiency of ruminants. ISME J. doi: 10.1038/ismej.2016.62. 1866 Shallom, D., and Shoham, Y. (2003). Microbial hemicellulases. Current Opinion in 1867 Microbiology 6(3), 219-228. doi: https://doi.org/10.1016/S1369-5274(03)00056-0. 1868 Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., et al. (2003). 1869 Cytoscape: A software environment for integrated models of biomolecular 1870 interaction networks. Genome Research 13(11), 2498-2504. doi: 1871 10.1101/gr.1239303. 1872 Shapiro, E., Biezuner, T., and Linnarsson, S. (2013). Single-cell sequencing-based 1873 technologies will revolutionize whole-organism science. Nature Reviews Genetics 1874 14, 618. doi: 10.1038/nrg3542. 1875 Silley, P. (1985). A note on the pectinolytic enzymes of Lachnospira multiparus. Journal 1876 of Applied Bacteriology 58(2), 145-149. doi: 10.1111/j.1365-2672.1985.tb01441.x. 1877 Silverman, J.D., Washburne, A.D., Mukherjee, S., and David, L.A. (2017). A phylogenetic 1878 transform enhances analysis of compositional microbiota data. eLife 6, e21887. 1879 doi: 10.7554/eLife.21887. 1880 Sirohi, S.K., Pandey, N., Singh, B., and Puniya, A.K. (2010). Rumen methanogens: a 1881 review. Indian Journal of Microbiology 50(3), 253-262. doi: 10.1007/s12088-010- 1882 0061-6. 1883 Skene, I.K., and Brooker, J.D. (1995). Characterization of Tannin Acylhydrolase Activity 1884 in the Ruminal Bacterium Selenomonas Ruminantium. Anaerobe 1(6), 321-327. 1885 doi: https://doi.org/10.1006/anae.1995.1034. 1886

Snelling, T.J., and Wallace, R.J. (2017). The rumen microbial metaproteome as revealed 1887 by SDS-PAGE. BMC Microbiology 17(1), 9. doi: 10.1186/s12866-016-0917-y. 1888 Sniffen, C.J., and Robinson, P.H. (1987). Microbial growth and flow as influenced by 1889 dietary manipulations. Journal of Dairy Science 70(2), 425-441. 1890 Stackebrandt, E., and Goebel, B.M. (1994). Taxonomic Note: A Place for DNA-DNA 1891 Reassociation and 16S rRNA Sequence Analysis in the Present Species Definition 1892 in Bacteriology. International Journal of Systematic and Evolutionary Microbiology 1893 44(4), 846-849. doi: doi:10.1099/00207713-44-4-846. 1894 Steele, M.A., Penner, G.B., Chaucheyras-Durand, F., and Guan, L.L. (2016). 1895 Development and physiology of the rumen and the lower gut: Targets for improving 1896 gut health. J Dairy Sci. doi: 10.3168/jds.2015-10351. 1897 Stevenson, D.M., and Weimer, P.J. (2007). Dominance of Prevotella and low abundance 1898 of classical ruminal bacterial species in the bovine rumen revealed by relative 1899 quantification real-time PCR. Applied Microbiology and Biotechnology 75(1), 165- 1900 174. doi: 10.1007/s00253-006-0802-y. 1901 Stewart, C.S., Fonty, G., and Gouet, P. (1988). The establishment of rumen microbial 1902 communities. Animal Feed Science and Technology 21(2), 69-97. doi: 1903 https://doi.org/10.1016/0377-8401(88)90093-4. 1904 Stewart, R.D., Warr, A., Watson, M., Auffret, M.D., Dewhurst, R.J., Roehe, R., et al. 1905 (2018). Assembly of 913 microbial genomes from metagenomic sequencing of the 1906 cow rumen. Nature Communications 9(1). doi: 10.1038/s41467-018-03317-6. 1907 Sunagawa, S., Mende, D.R., Zeller, G., Izquierdo-Carrasco, F., Berger, S.A., Kultima, 1908 J.R., et al. (2013a). Metagenomic species profiling using universal phylogenetic 1909 marker genes. 10(12), 1196-1199. doi: 10.1038/nmeth.2693. 1910 Sunagawa, S., Mende, D.R., Zeller, G., Izquierdo-Carrasco, F., Berger, S.A., Kultima, 1911 J.R., et al. (2013b). Metagenomic species profiling using universal phylogenetic 1912 marker genes. Nature Methods 10(12), 1196-1199. doi: 10.1038/nmeth.2693. 1913 Svartström, O., Alneberg, J., Terrapon, N., Lombard, V., de Bruijn, I., Malmsten, J., et al. 1914 (2017). Ninety-nine de novo assembled genomes from the moose (Alces alces) 1915 rumen microbiome provide new insights into microbial plant biomass degradation. 1916 The ISME Journal. doi: 10.1038/ismej.2017.108. 1917 Sylvester, J.T., Karnati, S.K., Yu, Z., Morrison, M., and Firkins, J.L. (2004). Development 1918 of an assay to quantify rumen ciliate protozoal biomass in cows using real-time 1919 PCR. J Nutr 134(12), 3378-3384. 1920 Tajima, K., Aminov, R.I., Nagamine, T., Matsui, H., Nakamura, M., and Benno, Y. (2001). 1921 Diet-dependent shifts in the bacterial population of the rumen revealed with real- 1922 time PCR. Applied and Environmental Microbiology 67(6), 2766-2774. doi: 1923 10.1128/aem.67.6.2766-2774.2001. 1924 Tap, J., Furet, J.-P., Bensaada, M., Philippe, C., Roth, H., Rabot, S., et al. (2015). Gut 1925 microbiota richness promotes its stability upon increased dietary fibre intake in 1926 healthy adults. Environmental Microbiology 17(12), 4954-4964. doi: 10.1111/1462- 1927 2920.13006. 1928 Tapio, I., Shingfield, K.J., McKain, N., Bonin, A., Fischer, D., Bayat, A.R., et al. (2016). 1929 Oral Samples as Non-Invasive Proxies for Assessing the Composition of the 1930 Rumen Microbial Community. PLoS One 11(3), e0151220. doi: 1931 10.1371/journal.pone.0151220. 1932

Tapio, I., Snelling, T.J., Strozzi, F., and Wallace, R.J. (2017). The ruminal microbiome 1933 associated with methane emissions from ruminant livestock. J Anim Sci Biotechnol 1934 8, 7. doi: 10.1186/s40104-017-0141-0. 1935 Tewoldebrhan, T.A., Appuhamy, J.A.D.R.N., Lee, J.J., Niu, M., Seo, S., Jeong, S., et al. 1936 (2017). Exogenous β-mannanase improves feed conversion efficiency and 1937 reduces somatic cell count in dairy cattle. Journal of Dairy Science 100(1), 244- 1938 252. doi: https://doi.org/10.3168/jds.2016-11017. 1939 Theodorou, M.K., Mennim, G., Davies, D.R., Zhu, W.-Y., Trinci, A.P.J., and Brookman, 1940 J.L. (2007). Anaerobic fungi in the digestive tract of mammalian herbivores and 1941 their potential for exploitation. Proceedings of the Nutrition Society 55(3), 913-926. 1942 doi: 10.1079/PNS19960088. 1943 Thompson, S. (2015). The effect of diet type on residual feed intake and the use of 1944 infrared thermography as a method to predict efficiency in beef bulls. M.Sc., 1945 University of Manitoba. 1946 Turnbaugh, P.J., Hamady, M., Yatsunenko, T., Cantarel, B.L., Duncan, A., Ley, R.E., et 1947 al. (2009). A core gut microbiome in obese and lean twins. Nature 457(7228), 480- 1948 484. doi: 1949 http://www.nature.com/nature/journal/v457/n7228/suppinfo/nature07540_S1.html. 1950 Tymensen, L.D., Beauchemin, K.A., and McAllister, T.A. (2012). Structures of free-living 1951 and protozoa-associated methanogen communities in the bovine rumen differ 1952 according to comparative analysis of 16S rRNA and mcrA genes. Microbiology 1953 158(7), 1808-1817. doi: doi:10.1099/mic.0.057984-0. 1954 Tyson, G.W., Chapman, J., Hugenholtz, P., Allen, E.E., Ram, R.J., Richardson, P.M., et 1955 al. (2004). Community structure and metabolism through reconstruction of 1956 microbial genomes from the environment. Nature 428, 37. doi: 1957 10.1038/nature02340 1958 https://www.nature.com/articles/nature02340 - supplementary-information. 1959 Van Soest, P.J., Robertson, J.B., and Lewis, B.A. (1991). Methods for dietary fiber, 1960 neutral detergent fiber, and nonstarch polysaccharides in relation to animal 1961 nutrition. Journal of Dairy Science 74(10), 3583-3597. 1962 Vanwonterghem, I., Evans, P.N., Parks, D.H., Jensen, P.D., Woodcroft, B.J., Hugenholtz, 1963 P., et al. (2016). Methylotrophic methanogenesis discovered in the archaeal 1964 phylum Verstraetearchaeota. Nature Microbiology 1, 16170. doi: 1965 10.1038/nmicrobiol.2016.170 1966 http://www.nature.com/articles/nmicrobiol2016170 - supplementary-information. 1967 Varga, G.A., and Kolver, E.S. (1997). Microbial and animal limitations to fiber digestion 1968 and utilization. The Journal Of Nutrition 127(5 Suppl), 819S-823S. doi: 1969 10.1093/jn/127.5.819S. 1970 Veira, D.M. (1986). The role of ciliate protozoa in nutrition of the ruminant. J Anim Sci 1971 63(5), 1547-1560. 1972 Verberkmoes, N.C., Russell, A.L., Shah, M., Godzik, A., Rosenquist, M., Halfvarson, J., 1973 et al. (2008). Shotgun metaproteomics of the human distal gut microbiota. The 1974 Isme Journal 3, 179. doi: 10.1038/ismej.2008.108 1975 https://www.nature.com/articles/ismej2008108 - supplementary-information. 1976

Větrovský, T., and Baldrian, P. (2013). The Variability of the 16S rRNA Gene in Bacterial 1977 Genomes and Its Consequences for Bacterial Community Analyses. PLOS ONE 1978 8(2), e57923. doi: 10.1371/journal.pone.0057923. 1979 Wallace, R.J., Rooke, J.A., Duthie, C.-A., Hyslop, J.J., Ross, D.W., McKain, N., et al. 1980 (2014). Archaeal abundance in post-mortem ruminal digesta may help predict 1981 methane emissions from beef cattle. Scientific Reports 4. doi: 10.1038/srep05892. 1982 Wallace, R.J., Rooke, J.A., McKain, N., Duthie, C.-A., Hyslop, J.J., Ross, D.W., et al. 1983 (2015). The rumen microbial metagenome associated with high methane 1984 production in cattle. BMC Genomics 16, 1-14. doi: 10.1186/s12864-015-2032-0. 1985 Wang, L., Hatem, A., Catalyurek, U.V., Morrison, M., and Yu, Z. (2013). Metagenomic 1986 Insights into the Carbohydrate-Active Enzymes Carried by the Microorganisms 1987 Adhering to Solid Digesta in the Rumen of Cows. PLoS ONE 8(11), 1-11. doi: 1988 10.1371/journal.pone.0078507. 1989 Wang, S., Giller, K., Kreuzer, M., Ulbrich, S.E., Braun, U., and Schwarm, A. (2017). 1990 Contribution of Ruminal Fungi, Archaea, Protozoa, and Bacteria to the Methane 1991 Suppression Caused by Oilseed Supplemented Diets. Frontiers in Microbiology 1992 8(1864). doi: 10.3389/fmicb.2017.01864. 1993 Wang, Y., and McAllister, T.A. (2002). Rumen Microbes, Enzymes and Feed Digestion- 1994 A Review. Asian-Australas J Anim Sci 15(11), 1659-1676. doi: 1995 10.5713/ajas.2002.1659. 1996 Warton, D.I., Wright, S.T., and Wang, Y. (2012). Distance-based multivariate analyses 1997 confound location and dispersion effects. Methods in Ecology and Evolution 3(1), 1998 89-101. doi: 10.1111/j.2041-210X.2011.00127.x. 1999 Weimer, P.J. (1992). Cellulose degradation by ruminal microorganisms. Critical Reviews 2000 in Biotechnology 12, 189-223. 2001 Weimer, P.J. (2015). Redundancy, resilience, and host specificity of the ruminal 2002 microbiota: implications for engineering improved ruminal fermentations. Frontiers 2003 in Microbiology 6(296). doi: 10.3389/fmicb.2015.00296. 2004 Weimer, P.J., Lopezguisa, J.M., and French, A.D. (1990). Effect of cellulose fine-structure 2005 on kinetics of its digestion by mixed ruminal microorganisms invitro. Applied and 2006 Environmental Microbiology 56(8), 2421-2429. 2007 Weiss, S., Xu, Z.Z., Peddada, S., Amir, A., Bittinger, K., Gonzalez, A., et al. (2017). 2008 Normalization and microbial differential abundance strategies depend upon data 2009 characteristics. Microbiome 5. doi: 10.1186/s40168-017-0237-y. 2010 Westerhuis, J.A., van Velzen, E.J.J., Hoefsloot, H.C.J., and Smilde, A.K. (2010). 2011 Multivariate paired data analysis: multilevel PLSDA versus OPLSDA. 2012 Metabolomics 6(1), 119-128. doi: 10.1007/s11306-009-0185-z. 2013 Wilkinson, T.J., Huws, S.A., Edwards, J.E., Kingston-Smith, A.H., Siu-Ting, K., Hughes, 2014 M., et al. (2018). CowPI: A Rumen Microbiome Focussed Version of the PICRUSt 2015 Functional Inference Software. Frontiers in Microbiology 9(1095). doi: 2016 10.3389/fmicb.2018.01095. 2017 Williams, A.G., and Coleman, G.S. (1997). "The rumen protozoa," in The Rumen 2018 Microbial Ecosystem, eds. P.N. Hobson & C.S. Stewart. (Dordrecht: Springer 2019 Netherlands), 73-139. 2020 Wilmes, P., and Bond, P.L. (2004). The application of two-dimensional polyacrylamide 2021 gel electrophoresis and downstream analyses to a mixed community of prokaryotic 2022

microorganisms. Environmental Microbiology 6(9), 911-920. doi: 2023 doi:10.1111/j.1462-2920.2004.00687.x. 2024 Wolff, S.M., Ellison, M.J., Hao, Y., Cockrum, R.R., Austin, K.J., Baraboo, M., et al. (2017). 2025 Diet shifts provoke complex and variable changes in the metabolic networks of the 2026 ruminal microbiome. Microbiome 5(1), 60. doi: 10.1186/s40168-017-0274-6. 2027 Wood, D.E., and Salzberg, S.L. (2014). Kraken: ultrafast metagenomic sequence 2028 classification using exact alignments. Genome Biology 15(3), R46-R46. doi: 2029 10.1186/gb-2014-15-3-r46. 2030 Wu, L., Wang, H., Zhang, Z., Lin, R., Zhang, Z., and Lin, W. (2011). Comparative 2031 Metaproteomic Analysis on Consecutively Rehmannia glutinosa-Monocultured 2032 Rhizosphere Soil. PLOS ONE 6(5), e20611. doi: 10.1371/journal.pone.0020611. 2033 Yang, T., Wei, Z., Friman, V.-P., Xu, Y., Shen, Q., Kowalchuk, G.A., et al. (2017). 2034 Resource availability modulates biodiversity-invasion relationships by altering 2035 competitive interactions. Environmental Microbiology 19(8), 2984-2991. doi: 2036 10.1111/1462-2920.13708. 2037 Yi, L., Shi, S., Wang, Y., Huang, W., Xia, Z.-a., Xing, Z., et al. (2016). Serum Metabolic 2038 Profiling Reveals Altered Metabolic Pathways in Patients with Post-traumatic 2039 Cognitive Impairments. Scientific Reports 6, 21320. doi: 10.1038/srep21320. 2040 Yu, Z., and Morrison, M. (2004). Improved extraction of PCR-quality community DNA from 2041 digesta and fecal samples. BioTechniques 36(5), 808-812. doi: 2042 10.2144/04365ST04. 2043 Zehavi, T., Probst, M., and Mizrahi, I. (2018). Insights Into Culturomics of the Rumen 2044 Microbiome. Frontiers in Microbiology 9(1999). doi: 10.3389/fmicb.2018.01999. 2045 Zhang, Y.H.P., Himmel, M.E., and Mielenz, J.R. (2006). Outlook for cellulase 2046 improvement: Screening and selection strategies. Biotechnology Advances (5), 2047 452. doi: 10.1016/j.biotechadv.2006.03.003. 2048 Zhou, M., Chung, Y.H., Beauchemin, K.A., Holtshausen, L., Oba, M., McAllister, T.A., et 2049 al. (2011). Relationship between rumen methanogens and methane production in 2050 dairy cows fed diets supplemented with a feed enzyme additive. Journal of Applied 2051 Microbiology 111(5), 1148-1158. doi: 10.1111/j.1365-2672.2011.05126.x. 2052 Zhou, M., Hernandez-Sanabria, E., and Guan, L.L. (2009). Assessment of the Microbial 2053 Ecology of Ruminal Methanogens in Cattle with Different Feed Efficiencies. 2054 Applied and Environmental Microbiology 75(20), 6524-6533. doi: 2055 10.1128/aem.02815-08. 2056 Zhou, M., Hernandez-Sanabria, E., and Guan, L.L. (2010). Characterization of Variation 2057 in Rumen Methanogenic Communities under Different Dietary and Host Feed 2058 Efficiency Conditions, as Determined by PCR-Denaturing Gradient Gel 2059 Electrophoresis Analysis. Applied and Environmental Microbiology 76(12), 3776- 2060 3786. doi: 10.1128/aem.00010-10. 2061 Zhou, M., Hünerberg, M., Beauchemin, K.A., McAllister, T.A., Okine, E.K., and Guan, L.L. 2062 (2012). Individuality of ruminal methanogen/protozoa populations in beef cattle fed 2063 diets containing dried distillers' grain with solubles. Acta Agriculturae 2064 Scandinavica: Section A, Animal Science 62(4), 273-288. doi: 2065 10.1080/09064702.2013.788206. 2066 Zhou, M., Peng, Y.-J., Chen, Y., Klinger, C.M., Oba, M., Liu, J.-X., et al. (2018). 2067 Assessment of microbiome changes after rumen transfaunation: implications on 2068

improving feed efficiency in beef cattle. Microbiome 6(1), 62-62. doi: 2069 10.1186/s40168-018-0447-y. 2070 Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., and Smith, G.M. (2009). "Mixed 2071 Effects Models and Extensions in Ecology with R," in Mixed Effects Models and 2072 Extensions in Ecology with R.). 2073 2074

2075

Chapter 2 2076 Dynamics of microbial populations driven by interactions between diet and host 2077 shed light on individualized rumen microbiota 2078

2079

2.1 Introduction 2080

The bovine rumen harbors a symbiotic community of anaerobic bacteria, archaea, fungi, 2081 and protozoa that plays an essential role in feed degradation and energy provision for the 2082 host. The fermentation of host-indigestible plant biomass by the rumen microbiota 2083 provides up to 70% of the energetic requirements of ruminant animals, mainly through 2084 the production of VFAs (NRC 2016, Russell and Rychlik 2001). While vital to efficient milk 2085 and meat production, rumen microbial fermentation is also a significant source of 2086 greenhouse gas emissions (Huws et al. 2018), with CH4 exhibiting a global warming 2087 potential 25 times greater than that of CO2 (IPCC 2006), as well as reducing host dietary 2088 energy availability by up to 12% (Johnson and Johnson 1995). Microbial composition in 2089 the rumen is primarily affected by the diet and exhibits a large degree of individual 2090 variation that may be a reflection of differences in the host nutrient utilization efficiency 2091 (Henderson et al. 2015). Because individuals may respond differently to dietary changes 2092 (Hernandez-Sanabria et al. 2012), a greater understanding of the drivers that affect the 2093 relationship between the rumen microbiota and host responsiveness to the diet could help 2094 to develop better dietary interventions to meet individual host nutritional requirements 2095 (Healey et al. 2017, Morgavi et al. 2012a). 2096

The success of rumen microbes in digesting recalcitrant feedstuffs depends on the 2097 microbial interactions that occur in the ruminal ecosystem (Wolff et al. 2017, Russell and 2098 Rychlik 2001), which is in turn modulated by at least two possible factors: 1) the 2099 environment (e.g., nutrient availability for microbial growth) (Yang et al. 2017), and 2) the 2100 inter-individual variability of the host organism in terms of feed degradation (Weimer 2101 2015). While nutrient availability is related to niche occupancy (Yang et al. 2017), inter- 2102 individual variability may reveal the effectiveness of the microbiota of one particular 2103 animal versus another in terms of efficiency of feed utilization (Weimer 2015). Individual 2104 animals are heterogeneous regarding the responsiveness to the diet, and thus they 2105

exhibit a ‘personalized’ rumen microbial composition even when fed an identical diet and 2106 raised in the same environment (Brulc et al. 2009, Z. P. Li et al. 2016). Such host 2107 specificity does not appear to be restricted only to bacteria, as the inter-individual variation 2108 has also been observed in the methanogenic archaeal and protozoal communities (Zhou 2109 et al. 2012). Despite the wealth of information explaining how individual hosts control the 2110 rumen microbiota in relation to phenotypic traits (e.g., feed efficiency and CH4 emissions) 2111 (Roehe et al. 2016), little is known regarding the precise mechanisms of host-microbe 2112 interactions in ruminants undergoing dietary changes. Moreover, interactions of host 2113 variation in feed efficiency with rumen microbial populations are not well understood, and 2114 further investigating their interactive effects on animal productivity may contribute to the 2115 identification of feed efficient animals and the reduction of methane emissions. 2116

In this context, several studies have demonstrated that rumen bacteria and 2117 archaea (composition and abundance) play important roles in the feed conversion 2118 efficiency of cattle (Fuyong Li and Le Luo Guan 2017, Neves et al. 2017), suggesting that 2119 the identification of host-specific microbes should be considered as the regulating 2120 targets to improve host feed efficiency. Sequencing technologies were indispensable 2121 tools to quantify transcripts involved in feed efficiency in those studies, but we speculated 2122 that the application of qPCR in the current study could offer an opportunity to assess the 2123 dynamics of the rumen microbiota in a cost-effective manner. Here, we hypothesized that 2124 the stratification of dietary responses obtained from the magnitude of change in baseline 2125 rumen microbiota using qPCR can be used to identify individualized shifts of rumen 2126 microbes in cattle undergoing diet changes. The objectives of this research were (i) to 2127 investigate the dynamics of rumen microorganisms in cattle raised under different feeding 2128 regimens (forage vs. grain) and (ii) understand the relationship among the abundance of 2129 these microorganisms, host individuality and phenotypic traits. 2130

2.2 Materials and methods 2131

2.2.1 Animal study 2132

The experimental procedures described in this study were approved by the Veterinary 2133 Services and the Animal Care Committee, University of Manitoba, Canada. Fifty-nine 2134

purebred Angus bulls (mean age of 249 ± 22 days and average body weight of 313.9 ± 2135 32 kg) were raised in confinement at the Glenlea Research Station, University of 2136 Manitoba, Canada in accordance with the guidelines of the Canadian Council on Animal 2137 Care (Olfert et al. 1993). The bulls were randomly assigned into four pens, and each pen 2138 was bedded with a mixture of barley/flaxseed straw and equipped with GrowSafe® 2139 (GrowSafe Systems Ltd., Airdrie, Alberta) feed bunks and a heated watering bowl. Bulls 2140 were fed forage or grain diets over two experimental periods (Period 1 and 2, each with 2141 an 80-day duration) in a crossover design (Table S1). In the first feeding period (FP1), 2142 two pens (1 and 2) were fed forage-based diets and the remaining two pens (3 and 4) 2143 were fed a grain-based diet on an ad libitum basis, as described in Table S1. Following 2144 FP1, the animals in pen 2 were switched from a forage-based to a grain-based diet, and 2145 the animals in pen 3 were switched from grain to forage diets (Table S1). Bulls in pens 1 2146 and 4 were exclusively fed forage and grain diets, respectively, in both FPs (Table S1). It 2147 is important to mention that a 14-day adaptation period was observed before the 2148 commencement of the second feeding period. Nutritional composition of the forage and 2149 grain diets is listed in Table S2. 2150

Individual feed intakes were measured using the GrowSafe® feeding system to 2151 provide growth and intake data (DMI) needed for the estimation of feed conversion ratio 2152 (FCR), which was used as a measure of feed efficiency in this study. FCR was determined 2153 as a ratio of DMI to average daily gain (ADG) (computed on a biweekly basis) (Montanholi 2154 et al. 2010). Starch and neutral detergent fiber (NDF) contents (Table S2.) of the 2155 respective diets were used to calculate starch and NDF intakes from the daily DMI (NRC 2156

2016). Finally, enteric CH4 emissions were measured in four 24-h periods during Period 2157

1, and in three 24-h periods during Period 2 using the Sulphur hexafluoride (SF6) 2158 technique as described by Thompson (2015). 2159

2.2.1.1 Rumen fluid sampling. Two rumen fluid samples were collected on days 0 and 2160 80 in both FP1 and FP2 using a Geishauser oral probe (Duffield et al. 2004). However, 2161 the downstream analysis was performed only on the rumen samples taken on day 80 in 2162 each period, as a previous statistical analysis (data not shown) revealed a carryover effect 2163 on the rumen microbial abundances between feeding periods likely because the washout 2164

period (adaptation phase) was 14 days (Table S1). Approximately 250 mL of rumen fluid 2165 was collected in each sampling, snap frozen using liquid nitrogen, and then stored at - 2166 80°C awaiting molecular analysis. 2167

2.2.2 Feed chemical analysis 2168

Briefly, the composited and dried feeds were analyzed for DM (dry matter; method 934.01) 2169 as described by AOAC (2000). Both Neutral (NDF; with α-amylase and sodium sulphite) 2170 and acid (ADF) detergent fiber (Van Soest et al. 1991) were quantified using Ankom Fiber 2171 Analyzer (Ankom Technology Corporation, Macedon, NY, USA). Starch was determined 2172 by the -amylase method as described by Hall (2009). A Leco combustion nitrogen (N) 2173 analyzer (FP-428N Determinator, Leco Corporation, St Joseph MI, USA) was used to 2174 measure N content. Crude protein (CP) was calculated as N × 6.25 (NRC 2016). 2175

2.2.3 VFA analysis 2176

Supernatants from rumen fluid samples were obtained after centrifugation at 3,000 × g 2177 for 15 min at 4˚C and mixed with 25% phosphoric acid (4:1; v/v) for the subsequent gas 2178 chromatography (GC) analysis. After adding the internal standard to the samples and 2179 incubating them at -20˚C overnight, they were centrifuged at 19,000 × g for 5 min at 4˚C 2180 and the supernatant was transferred to the GC vials (1.8 mL). Next, 0.8 mL of the sample 2181 was combined with 0.2 mL of 25% phosphoric acid and 0.2 mL of internal standard 2182 solution. Standards (for acetic, propionic, isobutyric, butyric, isovaleric, valeric, and 2183 caproic acids) were prepared by combining 1 mL of standard solution and 0.2 mL of 2184 internal standard solution. The GC analysis was performed using the column Stabilwax- 2185 DA 30 meter (Restek Corp), the head pressure of 7.5 psi, split vent flow of 20 mL/minute, 2186 and injector temperature of 170ºC. 2187

2.2.4 DNA extraction 2188

Total DNA was extracted from the ruminal content samples using the bead beating 2189 method as described by Yu and Morrison (2004). Briefly, frozen rumen content was 2190 thawed on ice and 1g of sample was added to 15 mL falcon tubes, washed in 4 mL of 2191 TN150 buffer (10 mM Tris-HCl [pH 8.0], 150 mM NaCl), and centrifuged at 14,600  g for 2192

5 min at 4C. Thereafter, samples were physically disrupted in a BioSpec Mini Beads 2193

beater 8 (BioSpec, Bartlesville, OK, USA) at 4800 rpm for 3 min, and subjected to a 2194 phenol/chloroform/isoamyl alcohol (25: 24: 1) extraction protocol. The DNA was 2195 precipitated with cold ethanol and dissolved in nuclease-free water (30 μl). Lastly, the 2196 concentration and quality of DNA were measured using Nanodrop® ND-1000 2197 spectrophotometer (Thermo Scientific, Waltham, MA, US). Quantitative real-time PCR 2198 (qPCR) was performed only on DNA samples with a ratio of absorbance at 260nm to 2199 280nm higher than 1.8 and a ratio of 260 nm wavelength absorbance to 230 nm between 2200 2.0 and 2.2. 2201

2.2.5 Quantitative real-time PCR analysis 2202 qPCR was performed using SYBR Green chemistry (Fast SYBR® Green Master Mix; 2203 Applied Biosystems) on a StepOnePlusTM Real-Time PCR System (Applied Biosystems). 2204 The partial bacterial and archaeal 16S rRNA genes (V3-V4 regions) were amplified using 2205 U2F/U2R (5’-ACTCCTACGGGAGGCAG-3’; 5’-GACTACCAGGGTATCTAATCC-3’) 2206 (Stevenson and Weimer 2007) and uniMet1-F/uniMet1-R primer pairs (5'- 2207 CCGGAGATGGAACCTGAGAC-3'; 5’-CGGTCTTGCCCAGCTCTTATTC-3') (Zhou et al. 2208 2009). Protozoa and fungi were amplified using P-SSU-316F/P-SSU-539R (5’- 2209 GCTTTCGWTGGTAGTGTATT-3’; 5’-CTTGCCCTCYAATCGT WCT-3') (Romero-Perez 2210 et al. 2014) and Fungi-F1/Fungi-R1 (5’-GAGGAAGTAAAAGTCGTAACAAGGTTTC-3’; 2211 5’-CAAATTCACAAAGGGTAGGATGATT-3’) (Denman and McSweeney 2006) to target 2212 18S rRNA genes and internal transcribed spacer (ITS), respectively. qPCR experiments 2213 were performed using the following program: 95 °C for 10 min, followed by 40 cycles of 2214 95 °C for 20 s and 62 °C for 1 min for bacteria, and 95 °C for 20 s, followed by 40 cycles 2215 of 95 °C for 3 s and 60 °C for 30s for archaea, protozoa and fungi. A standard curve was 2216 constructed using serial dilutions of plasmid DNA containing the 16S rRNA gene 2217 sequence for bacteria and methanogens, ITS for fungi and 18S rRNA gene for protozoa. 2218 Copy numbers for each standard curve were calculated based on the following equation: 2219 (NL × A × 10-9) / (660 × n), where NL was the Avogadro constant (6.02 × 1023), A was the 2220 molecular weight of DNA molecules (ηg), and n was the length of amplicon (bp) 2221 (Malmuthuge et al. 2012). The copy number of 16S rRNA genes for total bacteria, total 2222 methanogens, and 18S rRNA gene for protozoa and ITS for fungi per sample was 2223

calculated using the equation of Li et al. (2009): (QM × C × DV) / (S × W), where QM was 2224 the quantitative mean of the copy number, C was the DNA concentration of each sample 2225 (ηg/μL), DV was the dilution volume of extracted DNA (μL), S was the DNA amount 2226 subjected to analysis (ηg), and W was the sample weight subjected to DNA extraction (g). 2227 qPCR assay efficiency (E) was determined for each primer by running a set of serial 2228 dilutions of the target species along with the unknowns, all in triplicates. A plot of Ct vs. 2229 Log DNA dilution allowed the efficiency to be calculated as the antilog of the negative 2230 reciprocal of the line slope: E = (10-1/slope - 1) × 100. The data generated from the reactions 2231 were used for further analysis only if the qPCR assay efficiency was within the range of 2232 90% ≤ E ≤ 105%. 2233

2.2.6 Stratification of animals according to shifts in microbial abundances 2234

The total abundance of each microbial group (measured by average copy number without 2235 log transformation) was determined for each bull according to the dietary regimen, as 2236 described previously (see Table S1 and Fig. S6). Groups of animals were created from 2237 the magnitude of change in the microbial abundance based on the microorganisms copy 2238 numbers recorded for each bull at the end of Period 1 (baselines; D80) compared with 2239 the copy numbers recorded at the end of Period 2 (D180) for each sequence of dietary 2240 treatment (Fig. S6). Thus, the formation of the groups was obtained by ranking the 2241 animals according to the change in the microbial population in response to each dietary 2242 treatment (Fig. S6). Log2 fold change (fc) in microbial copy numbers higher than +1 2243 represented the High group and lower than –1 represented the Low group. The log2fc in 2244 the copy numbers between +1 and –1 cut-offs were considered as the Stable group (Fig. 2245 S6). 2246

2.2.7 Statistical analysis 2247

2.2.7.1 Exploring the effect of the diet on the rumen microbes and phenotypic traits 2248

Principal component analysis (Rohart et al. 2017) was initially used to assess the 2249 relationship between rumen microbial abundances (bacteria, protozoa, fungi and archaea) 2250 and eight phenotypic traits (DMI, NDF intake, starch intake, CH4 production, CH4/kg DMI, 2251 FCR, Feeding Time in minutes - DUR, and ADG) according to the dietary regimens 2252

(forage vs. grain). Then, we implemented the PC-corr (Principal Component-Correlation) 2253 algorithm (Ciucci et al. 2017) to generate discriminative functional modules associated 2254 with the diets using p-values of Mann-Whitney tests as evaluators, with each module 2255 displaying a correlation structure between features (e.g., microbes, phenotypic traits) and 2256 the diets. Features were normalized through z-score (centered to have mean 0 and 2257 scaled to have standard deviation 1) or log (logarithm base 10 plus 1 applied to each data 2258 element to avoid problems with 0 values) before building PC-corr networks in Cytoscape 2259 3.6.0 (Shannon et al. 2003). 2260

2.2.7.2 Investigating the relationship between the phenotypic traits and the 2261 microbial population 2262

Two linear mixed effects models (LMMs) were set up using the package lme4 (‘lmer’ 2263 function) (Bates et al. 2015) to investigate the effect of the phenotypic traits on both the 2264 microbial populations (Model 1) and the stratification groups obtained from the shifts in 2265 the microbial abundances as described above (Model 2). The rumen microbial abundance 2266 was log10 – transformed to meet the assumptions of normality, and the phenotypic traits 2267 were scaled (centered to have mean 0 and scaled to have standard deviation 1) to 2268 facilitate model convergence when required. 2269

In the Model 1, the following parameters were used: 2270

풴푖푗푘푙 = µ + 휓푖 + 휋푘 + 휏푙 + 푠푖푗 + 휀푖푗푘푙 2271

th Where, 풴푖푗푘푙 is the rumen microbial abundance for the j subject (j = 1, 2, …ni) in 2272 the ith sequence (i =1, 2, 3, 4) during kth period (j = 1, 2) receiving treatment l (phenotypic 2273 traits measured according to the diets); µ is the overall mean; 휓푖 is the sequence effect 2274 th or carryover effect; 휋푘 is the effect of the k period (k = 1, 2); 휏푙 is the effect of the lth 2275 treatment; 푠푖푗 is the effect of the jth subject in the ith sequence (j = 1, 2, …ni ; i =1, 2, 3, 2276

4); 휀푖푗푘푙 is the within-subject error for the jth subject in the ith sequence and the kth period 2277 receiving treatment l. Here, it was assumed that 휓, 휋, and 휏 are fixed effects, 푠푖푗 is 2278 2 random effect with mean zero and variance 휎s (between-subject variance), and 휀푖푗푘푙 is 2279 the random error with means zero and variances 휎2 (within-subject variance). 2280

In Model 2, the following parametrization was implemented: 2281

풴푖푗 = µ + 풳푖푗 + 풲푖푗 + 풳푖푗 × 풲푖푗 + 훼푖 + 휀푖푗 2282

Where, 풴푖푗 is the phenotypic traits (CH4 production, CH4/kg DMI, FCR, DUR, DMI 2283 th and ADG) for the j observation in the subject/bull i, and i = 1, 2, …ni; µ is the overall 2284 mean; 풳푖푗 is the fixed effect of starch and NDF intake measured when bulls were fed 2285 forage/grain diets; 풲푖푗 represents the fixed effect of the groups determined from the 2286 magnitude of change in the host microbial population (high, stable and low); 풳푖푗 × 풲푖푗 is 2287 the interaction term (fixed effect); 훼푖 is the random effect of bull i; 휀푖푗 is the residual 2288 random error. Here, it was assumed that 훼푖 is normally distributed with mean 0 and 2289 2 2 variance 휎s (between-subject variance), and 휀푖푗 has mean 0 and variance 휎 (within- 2290 subject variance). 2291

The best models were selected using a backward stepwise approach based on 2292 Akaike information criteria (‘drop1’ function) (Burnham et al. 2002), with significance of 2293 fixed effects and their interactions being tested by comparing the models with a likelihood- 2294 ratio test (i.e., chi-squared test) (Romain et al. 2009). Modeling assumptions were also 2295 checked by visually inspecting residual patterns (Zuur et al. 2009). All the statistical 2296 analysis described in this study were performed using R 3.4.2. 2297

2.3 Results 2298

2.3.1 Effect of diet on the rumen microbes and phenotypic traits 2299

Total bacterial abundance ranged from 0.32 ± 0.282 to 5.9 ± 6.80 × 1011 copy numbers 2300 of 16S rRNA genes/mL rumen fluid (Table 2.1). Total methanogen abundance varied from 2301 2.5 ± 1.38 to 7.5 ± 5.97 × 108 copy numbers of 16S rRNA genes/mL rumen fluid, while 2302 the abundances of fungi and protozoa ranged from 0.05 ± 0.062 to 1.1 ± 0.89 × 105 and 2303 from 1.9 ± 3.83 to 10.0 ± 11.2 × 107 copy numbers of ITS/18S rRNA genes/mL rumen 2304 fluid, respectively (Table 2.1). 2305

Then, it was evaluated if a relationship could be established between the 2306 abundances of the four microbial populations and eight phenotypic traits according to the 2307

dietary regimens (forage vs. grain) (Fig. S1). Application of the PC-Corr algorithm showed 2308 that all four (bacteria, methanogens, protozoa, and fungi) were clearly separated across 2309 diets (P <0.01) and emerged as an interconnected module positively correlated to each 2310 other and with a higher abundance in cattle fed forage diets (Figure 2.1). PC-Corr also 2311 differentiated (P <0.01) phenotypic traits (NDF intake, starch intake, ADG, DMI, CH4/kg 2312 DMI, and FCR) depending on the dietary treatments (Figure 2.1). This result showed that 2313 grain diets resulted in a higher DMI, starch intake, and ADG than forage diets (Figure 2.1). 2314 In the network structure, those three phenotypic traits arose as a module that was 2315 negatively correlated with NDF intake, FCR, and CH4/kg DMI, which were features 2316 displaying higher values in cattle fed forage diets (Figure 2.1). 2317

2.3.2 Effect of the phenotypic traits on the microbial populations 2318

2.3.2.1 Response to the intake of DM and starch, and animal performance (ADG) 2319

Our results revealed that the intake of DM and starch affected (P < 0.05) the abundance 2320 of bacteria, reducing it by about 0.2 (±0.06) log10 (Figure 2.2). It should be noted that the 2321 diets influenced (P < 0.05) DMI, with bulls consuming grain diets exhibiting a greater DMI 2322 (9.6 ± 0.19 kg/day) than those fed forage diets (7.4 ± 0.35 kg/day) (Fig. S5). In a similar 2323 fashion, the diets affected (P < 0.05) starch intake, with bulls fed grain diets consuming 2324 more starch (3.2 ± 0.05 kg/day) than those fed forage diets (1.5 ± 0.10 kg/day) (Fig. S5). 2325

It is worth mentioning that DMI influenced (P < 0.05) the host production of CH4, 2326 increasing it by about 8.7 L/day ± 4.52 (Fig. S2). Although starch intake did not influence 2327 the host production of CH4, it affected (P < 0.05) the host production of CH4/kg DMI, 2328 reducing it by about 1.8 L ± 0.34 CH4/kg DMI (Fig. S2). 2329

As to the animal performance, our findings showed that the increase in ADG 2330 influenced (P < 0.05) the abundance of bacteria, lowering it by about 0.1 (±0.05) log10 2331 (Figure 2.2). The diets also affected (P < 0.05) ADG, with bulls fed grain diets gaining 2332 more weight (1.6 ± 0.05 kg/day) than bulls fed forage diets (1.2 ± 0.08 kg/day) (Fig S5). 2333

2.3.2.2 Response to NDF intake, CH4/kg DMI and VFA concentrations 2334

NDF intake and CH4/kg DMI influenced (P < 0.05) the abundance of bacteria, increasing 2335 it by about 0.1 (±0.05) log10 (Figure 2.3). It is important to note that bulls fed forage diets 2336

consumed more (P < 0.05) NDF (3.3 ± 0.10 vs. 1.4 ± 0.06 kg/day) and produced more (P 2337

< 0.05) CH4/kg DMI (23.0 ± 1.01 vs. 19.3 ± 0.7 L CH4/ kg DMI) than bulls fed grain diets 2338

(Fig. S5). NDF intake also affected (P < 0.05) the host production of CH4 and CH4/kg DMI, 2339 increasing them by about 9.3 ± 4.2 L/day and 1.4 ± 0.37 L CH4/kg DMI, respectively (Fig. 2340 S2). As expected, NDF intake affected (P < 0.05) the total VFA and acetic acid 2341 concentrations, increasing them by about 9.0 ± 2.67 and 7.0 ± 1.78 µmol/mL, respectively 2342 (Fig. S4). However, the increase in the total VFA concentrations reduced (P < 0.05) the 2343 abundance of bacteria in about 0.02 (±0.00) log10 (Fig. S4). Similarly, a reduction trend 2344

(P < 0.1) of 0.01 (±0.01) log10 in the bacterial abundance was detected with the increase 2345 in acetic acid concentrations (Fig. S4). Although we did not find any relationship between 2346

NDF intake and propionic acid, bacterial abundance reduced in 0.09 (±0.04) log10 when 2347 rumen concentrations of propionic acid increased (Fig. S4). 2348

2.3.3 Rumen microbial dynamics in response to the interactions of phenotypic 2349 traits and the magnitude of change in host individual microbial population 2350

Since the abundances of methanogens and protozoa did not change significantly with the 2351 dietary treatments, they were not considered during stratification of animals according to 2352 the shift in microbial abundances. Thus, groups of animals were only created according 2353 to the magnitude of change in bacterial and fungal abundances to assess individual 2354 variability driven by the diets, and to identify feed efficient animals within the created 2355 groups (Fig. S6). Our results showed that shifts from forage to grain resulted in a more 2356 pronounced animal-to-animal variation than in individuals that began on grain and then 2357 switched to the forage diets (See Figs. S6B and C). 2358

The magnitude of change values from baseline (D80) to D180 revealed three 2359 distinct groups of animals based on the grouping cutoffs (Low, log2-fc < -1; Stable, -1 < 2360 log2-fc < 1; and High, log2-fc > 1), and the data showed that the interaction between the 2361 stratification of these groups and NDF intake influenced (P<0.05) FCR (Figure 2.4). Yet, 2362 our analysis showed that both NDF and starch intake influenced (P < 0.05) FCR, with 2363 NDF intake increasing FCR by about 0.3 (± 0.08) kg DMI/kg gain, and starch intake 2364 lowering it (P < 0.05) by about 0.2 (± 0.08) kg DMI/kg gain (Fig. S3). 2365

Based on the inter-dependence detected between the stratification of those groups 2366 and NDF intake, the bulls classified in the high and low groups determined from the 2367 change in the bacterial abundance were more (P < 0.05) feed efficient (increase in 0.2 ± 2368 0.09 and 0.1 ± 0.20 kg DMI/kg gain, respectively) than bulls ranked in the stable group 2369 (increase in 0.6 ± 0.14 kg DMI/kg gain) (Figure 2.4). To confirm these findings, the feeding 2370 time was measured in each group and the results indicated that the bulls of the high/low 2371 groups spent less (P < 0.05) time feeding than those of the stable group (Figs. 2.4B). A 2372 trend (P < 0.1) towards an improved feed efficiency in bulls belonging to the high/low 2373 groups compared to the stable group was detected in animals ranked according to the 2374 fungal abundance change (Figure 2.4C). Finally, animals ranked in the high/low groups 2375 according to the fungal abundance change spent less time (P < 0.05) in the feed bunker 2376 than those of the stable group (Figure 2.4D). 2377

2.4 Discussion 2378

Previous studies have demonstrated that microbial community composition and 2379 metabolic potentials in the rumen are remarkably different with respect to nutrient 2380 utilization, even in animals raised under the same diet and management regimens (Brulc 2381 et al. 2009, Z. P. Li et al. 2016). Thus, a better understanding of interactions between the 2382 host and its individualized microbiota is crucial for predicting microbial shifts and 2383 identifying individuals that are either responsive (positive and negative responders) or 2384 resilient to dietary changes (Bashiardes et al. 2018). Despite being widely investigated 2385 using molecular-based approaches, most studies examining diet-driven microbial shifts 2386 in the rumen have generated results that are usually qualitative (presence or absence of 2387 particular microbial taxa) (Belanche et al. 2011) or semi-quantitative (relative abundance 2388 of each microbial taxa) (Comtet-Marre et al. 2017), but not quantitative. To overcome 2389 such limitations, it was defined the “baseline” of the quantified rumen microbiota (detected 2390 by qPCR) in cattle experiencing dietary changes and developed a strategy to identify feed 2391 efficient cattle from the dynamic shift of microbial population abundance. It was also 2392 explored the potential to use this approach to assess differences in cattle phenotypes 2393

(e.g., FCR, CH4 emissions). 2394

To our knowledge, this is the first study that quantified the four groups of rumen 2395 microbes (bacteria, fungi, protozoa, and methanogens) simultaneously and integrated the 2396 microbial data obtained in vivo with environmental factors (e.g., diet components). 2397 Although previous investigations have focused on those microbial groups and their 2398 response to dietary supplementation under in vitro conditions (Wang et al. 2017), most in 2399 vivo studies have concentrated efforts on those microbial groups separately (Tajima et al. 2400 2001, Rico et al. 2015). In the current work, it was monitored the shift in microbial 2401 population abundance within the same animal in response to the diet, as well as the 2402 response of phenotypic traits to the interactions between the groups of microbial 2403 population change (high, low, stable) and the dietary components. Our approach is 2404 analogous to that used by biomedical researchers in which individuals are grouped 2405 according to the magnitude of microbial change to analyze the effect of dietary factors on 2406 microbial composition within the groups of individuals (Martínez et al. 2010, Salonen et 2407 al. 2014, Tap et al. 2015). Here, our results clearly showed that shifts from forage to grain 2408 (Fig. S6B) resulted in a more pronounced animal-to-animal variation in individuals fed the 2409 latter diet, suggesting that differences in microbial abundance drive the resulting variation 2410 in microbial baselines between and within diets, and not necessarily by the presence or 2411 absence of taxa (Wolff et al. 2017). 2412

The decline in the bacterial population density as starch intake increased (Figure 2413 2.2) did not corroborate past studies that suggested that adding more available energy 2414 (starch) to the diet in the form of non-fiber carbohydrates usually favors microbial growth 2415 (Grubb and Dehority 1975, Hackmann and Firkins 2015). This study speculated that 2416 secondary factors (e.g., individual variability in feed passage rate) in addition to the diet 2417 composition might have affected the microbial density. Grain diets usually increase feed 2418 passage rate through the gastrointestinal tract, which can lower bacterial abundance in 2419 the rumen by decreasing the residence time of feed particles and the subsequent 2420 bacterial attachment to the digesta (McAllister et al. 1994a). On the other hand, the 2421 models predicted an increase in the bacterial abundance as NDF intake increased (Figure 2422 2.3A), suggesting that bacteria and likely fungi acted in concert to stimulate fiber digestion 2423 in those animals. It has been reported that fungi can penetrate and physically disrupt the 2424 plant cell wall using an appressorium-like structure (Ho et al. 1988) to increase the surface 2425

area for subsequent bacterial colonization (Lee et al. 2000). However, it is worth 2426 mentioning that the differences in the bacterial abundance response to the intake of NDF 2427 could have been caused by the type of fiber (alfalfa hay, corn silage, alfalfa silage) present 2428 in the feed studied and on their variation (Table S2). Although this study did not 2429 investigate the influence of those different types of fiber on the microbial abundance, it 2430 has been demonstrated that the rate and extent to which fiber components are degraded 2431 depend on the microbial accessibility to feed substrates, which is affected by the physical 2432 and chemical nature of the forage (Varga and Kolver 1997). Another prerequisite to be 2433 considered is the level of NDF in the diet, as it can influence the time of rumination and 2434 consequently the microbial colonization of feed particles (NRC 2016). Therefore, 2435 variations in the content of other diet components that no NDF can affect the time of 2436 rumination and the microbial fermentation of the feed, and consequently the results found 2437 in this study. 2438

Although the bacterial density had a significant influence on CH4 production per kg 2439 DMI (Figure 2.3), PC-Corr revealed a functional module constituted by the four groups of 2440 rumen microbes in cattle fed forage diets (Figure 2.1B), indicating that CH4 production 2441 may depend on the interactions of methanogens (H2-consumer) with bacteria, protozoa, 2442 and fungi (H2-producers) to lower the partial pressure of H2 during the ruminal 2443 fermentation (Morgavi et al. 2012a, Janssen 2010, Morgavi et al. 2010). Surprisingly, this 2444 study did not find any relationship between methanogens/protozoa and starch/NDF intake, 2445 suggesting that protozoa-methanogens associations may have served to protect 2446 methanogens from being washed out of the rumen, as protozoa pass the rumen at a 2447 slower rate than bacteria and fungi (McAllister et al. 1994a). The lack of effect of total 2448 methanogen abundance on CH4 emissions confirms the findings of Zhou et al. (2011) 2449 who reported that the total methanogen numbers did not influence CH4 production, but 2450 rather only the abundance of particular species (e.g., Methanobrevibacter gottschalkii) 2451 were linked to CH4 output. However, our models predicted that the ability of NDF to 2452 increase CH4 (Fig. S2) may be related to the increase in total bacteria abundance when 2453

NDF consumption and CH4/kg DMI increased (Figure 2.3), indicating that increased 2454 bacterial abundance leads to greater H2 production which is subsequently used for 2455 methanogenesis (Janssen 2010, Leahy et al. 2010, Vanwonterghem et al. 2016). It is 2456

important to note that specific genera/species may be critical for feed digestion, even 2457 though in low numbers, which may affect the microbial response to the NDF consumption, 2458

CH4/kg DMI, and consequently the host feed efficiency. 2459

The changes at a single taxa level can be challenging to assess, and it may not 2460 directly contribute to the overall rumen fermentation outcomes (such as VFA production) 2461 due to the fuctional redundancy of the rumen microbiome. It has been reported that 2462 among host-adapted microbes, DNA and RNA abundances would be correlated if many 2463 microbial genes were not differentially regulated and were transcribed at the same 2464 constant rate (Franzosa et al. 2014). Consequently, variation in microbial composition 2465 may not correlate with their metagenomic content (Franzosa et al. 2014) since different 2466 taxa may have the same functions. Although in the past a direct relationship between the 2467 total densities of bacteria and methanogens with feed efficiency traits was not observed 2468 (Zhou et al. 2011, Zhou et al. 2010, Zhou et al. 2009), it is possible that this could be due 2469 to the comparatively poor understanding of the roles of protozoa and fungi in the rumen. 2470 Results of the current study showed that the dynamics of the four groups of microbes is 2471 associated with the variation in the VFA production which is directly linked with host 2472 energy metabolism, feed efficiency and methane emissions. 2473

The approach used to create groups of animals based on the variation in the 2474 abundance of microbial populations revealed that a relationship could be established 2475 between feed conversion efficiency and the groups of microbial change (Figure 2.4). This 2476 approach may be implemented in livestock systems as it reflects how individual hosts 2477 utilize the diet and how each of them responds to the diet based on the microbial shift. 2478 First, it was found that an improved FCR was linked to a higher abundance of bacteria 2479

(log2fc in bacteria > 1), and this result is in line with previous studies showing that bacterial 2480 growth is associated with a better animal performance when followed by the increase in 2481 bacterial N flow and microbial protein synthesis in the rumen (Sniffen and Robinson 1987, 2482 NRC 2016). Second, this study advances current knowledge in this field by demonstrating 2483 that a lower bacterial abundance can also be associated with feed efficient cattle as 2484 observed in bulls exhibiting a lower bacterial abundance (log2fc in bacteria < -1) (Figure 2485 2.4). These results suggest that those animals utilized efficiently the energy from the diet 2486

in order to optimize microbial protein synthesis (Turnbaugh et al. 2009) despite the 2487 decline in their bacterial abundance. In this case, the reduction in bacterial density may 2488 have been accompanied by a decrease in richness as shown by the metatranscriptomic 2489 analysis of cattle with lower feed efficiency (Fuyong Li and Le Luo Guan 2017, Neves et 2490 al. 2017) and a higher dominance in taxonomic composition (e.g., lactate utilizing bacteria) 2491 that favors simpler metabolic networks (e.g. acrylate pathway) (40). 2492

Our interest in the stratification of animals via their magnitude of change in 2493 baseline microbiota following dietary changes is driven by previous observations of 2494 substantial variation in performance across individuals, even when maintained on the 2495 same diet (Z. P. Li et al. 2016, Brulc et al. 2009, Zhou et al. 2018). This research 2496 speculated that such variation could be due to differences in rumen microbial function 2497 and cattle genetic makeup. In this study, we used cattle with extremely similar background 2498 (Red and Black Angus) and aimed to find out at what extent the rumen microbiota 2499 abundance differed when environmental traits (such as diet components) were taken into 2500 consideration. The findings of this study could benefit animal husbandry in two aspects. 2501 First, by understanding the dynamics of the ruminal microbial population in response to 2502 dietary changes, researchers will be able to design better feeding strategies to improve 2503 the rumen function and host performance, since the microbial community plays an 2504 important role in the digestion of feedstuffs (Belanche et al. 2012, Fernando et al. 2010). 2505 Second, our approach could serve as a inexpensive strategy to quickly assess rumen 2506 microbial shifts in cattle populations experiencing dietary changes under field situations 2507 (e. g., feedlots), and to identify feed efficient animals since a relationship between 2508 microorganisms abundance and FCR was found in this study. 2509

2.5 Conclusions 2510

This study showed that the individual responses of the rumen microbiota to the dietary 2511 treatments reflected the interactions between host and the examined phenotypic traits. 2512 The key finding is that the dynamics of the rumen microbial population is intimately 2513 associated with inter-individual variability in the baseline microbiota, confirming that the 2514 host microbiome individuality may play a more pronounced role in gut response than the 2515 dietary change itself. It was also found that bacterial abundance may serve as a useful 2516

proxy measurement to predict changes in feed conversion efficiency and CH4/kg DMI in 2517 cattle. Even though this experiment used simple and inexpensive methods (e.g., ranking 2518 animals based on their microbial population shift using qPCR), the results obtained here 2519 could be used to design better feeding strategies to enhance the rumen function and to 2520 identify cattle with improved feed efficiency based on an individualized microbiota- 2521 targeted feeding approach. By ranking animals according to their microbial response to 2522 the diet, it was showed that individual hosts exhibiting variability in bacterial abundance 2523

(log2fc < -1 or > 1) were more efficient in terms of feed conversion ratio than bulls 2524 presenting a stable variability (-1 > log2fc < 1) in bacterial abundance. 2525

2.6 Literature cited 2526

AOAC (2000). Official Methods of Analysis (17th). Gaithersburg, MD: Association of 2527 Official Analytical Chemists. 2528 Bashiardes, S., Godneva, A., Elinav, E., and Segal, E. (2018). Towards utilization of the 2529 human genome and microbiome for personalized nutrition. Current Opinion in 2530 Biotechnology 51, 57-63. doi: 10.1016/j.copbio.2017.11.013. 2531 Bates, D., Mächler, M., Bolker, B., and Walker, S. (2015). Fitting Linear Mixed-Effects 2532 Models Using lme4. 2015 67(1), 48. doi: 10.18637/jss.v067.i01. 2533 Belanche, A., Abecia, L., Holtrop, G., Guada, J.A., Castrillo, C., de la Fuente, G., et al. 2534 (2011). Study of the effect of presence or absence of protozoa on rumen 2535 fermentation and microbial protein contribution to the chyme1. Journal of Animal 2536 Science 89(12), 4163-4174. doi: 10.2527/jas.2010-3703. 2537 Belanche, A., Doreau, M., Edwards, J.E., Moorby, J.M., Pinloche, E., and Newbold, C.J. 2538 (2012). Shifts in the Rumen Microbiota Due to the Type of Carbohydrate and Level 2539 of Protein Ingested by Dairy Cattle Are Associated with Changes in Rumen 2540 Fermentation. Journal of Nutrition 142(9), 1684-1692. doi: 10.3945/jn.112.159574. 2541 Brulc, J.M., Antonopoulos, D.A., Berg Miller, M.E., Wilson, M.K., Yannarell, A.C., 2542 Dinsdale, E.A., et al. (2009). Gene-centric metagenomics of the fiber-adherent 2543 bovine rumen microbiome reveals forage specific glycoside hydrolases. 2544 Proceedings of the National Academy of Sciences 106(6), 1948-1953. doi: 2545 10.1073/pnas.0806191105. 2546 Burnham, K.P., Anderson, D.R., and Burnham, K.P. (2002). Model selection and 2547 multimodel inference : a practical information-theoretic approach. New York : 2548 Springer, c2002. 2549 2nd ed. 2550 Ciucci, S., Ge, Y., Durán, C., Palladini, A., Martínez-Sánchez, L.M., Cannistraci, C.V., et 2551 al. (2017). Enlightening discriminative network functional modules behind principal 2552 component analysis separation in differential-omic science studies. Scientific 2553 Reports 7. doi: 10.1038/srep43946. 2554 Comtet-Marre, S., Parisot, N., Lepercq, P., Chaucheyras-Durand, F., Mosoni, P., 2555 Peyretaillade, E., et al. (2017). Metatranscriptomics Reveals the Active Bacterial 2556

and Eukaryotic Fibrolytic Communities in the Rumen of Dairy Cow Fed a Mixed 2557 Diet. Frontiers in Microbiology 8(67). doi: 10.3389/fmicb.2017.00067. 2558 Denman, S.E., and McSweeney, C.S. (2006). Development of a real-time PCR assay for 2559 monitoring anaerobic fungal and cellulolytic bacterial populations within the rumen. 2560 Fems Microbiology Ecology 58(3), 572-582. doi: 10.1111/j.1574- 2561 6941.2006.00190.x. 2562 Duffield, T., Plaizier, J.C., Fairfield, A., Bagg, R., Vessie, G., Dick, P., et al. (2004). 2563 Comparison of techniques for measurement of rumen pH in lactating dairy cows. 2564 Journal of Dairy Science 87(1), 59-66. 2565 Fernando, S.C., Purvis, H.T., II, Najar, F.Z., Sukharnikov, L.O., Krehbiel, C.R., Nagaraja, 2566 T.G., et al. (2010). Rumen Microbial Population Dynamics during Adaptation to a 2567 High-Grain Diet. Applied and Environmental Microbiology 76(22), 7482-7490. doi: 2568 10.1128/aem.00388-10. 2569 Franzosa, E.A., Morgan, X.C., Segata, N., Waldron, L., Reyes, J., Earl, A.M., et al. (2014). 2570 Relating the metatranscriptome and metagenome of the human gut. Proceedings 2571 of the National Academy of Sciences 111(22), E2329-E2338. doi: 2572 10.1073/pnas.1319284111. 2573 Grubb, J.A., and Dehority, B.A. (1975). Effects of an abrupt change in ration from all 2574 roughage to high concentrate upon rumen microbial numbers in sheep. Applied 2575 Microbiology 30(3), 404-412. 2576 Hackmann, T.J., and Firkins, J.L. (2015). Maximizing efficiency of rumen microbial protein 2577 production. Frontiers in Microbiology 6. doi: 10.3389/fmicb.2015.00465. 2578 Hall, M.B. (2009). Determination of Starch, Including Maltooligosaccharides, in Animal 2579 Feeds: Comparison of Methods and a Method Recommended for AOAC 2580 Collaborative Study. Journal of Aoac International 92(1), 42-49. 2581 Healey, G.R., Murphy, R., Brough, L., Butts, C.A., and Coad, J. (2017). Interindividual 2582 variability in gut microbiota and host response to dietary interventions. Nutrition 2583 Reviews 75(12), 1059-1080. 2584 Henderson, G., Cox, F., Ganesh, S., Jonker, A., Young, W., Janssen, P.H., et al. (2016). 2585 Rumen microbial community composition varies with diet and host, but a core 2586 microbiome is found across a wide geographical range. Scientific Reports 6. doi: 2587 10.1038/srep19175. 2588 Hernandez-Sanabria, E., Goonewardene, L.A., Wang, Z., Durunna, O.N., Moore, S.S., 2589 and Guan, L.L. (2012). Impact of Feed Efficiency and Diet on Adaptive Variations 2590 in the Bacterial Community in the Rumen Fluid of Cattle. Applied and 2591 Environmental Microbiology 78(4), 1203-1214. doi: 10.1128/aem.05114-11. 2592 Ho, Y.W., Abdullah, N., and Jalaludin, S. (1988). Penetrating structures of anaerobic 2593 rumen fungi in cattle and swamp buffalo. Journal of General Microbiology 134(1), 2594 177-182. 2595 Huws, S.A., Creevey, C.J., Oyama, L.B., Mizrahi, I., Denman, S.E., Popova, M., et al. 2596 (2018). Addressing Global Ruminant Agricultural Challenges Through 2597 Understanding the Rumen Microbiome: Past, Present, and Future. Frontiers in 2598 Microbiology 9(2161). doi: 10.3389/fmicb.2018.02161. 2599 IPCC (2006). Guidelines for National Greenhouse Gas Inventories. Japan: Institute for 2600 Global Environmental Strategies (IGES). 2601

Janssen, P.H. (2010). Influence of hydrogen on rumen methane formation and 2602 fermentation balances through microbial growth kinetics and fermentation 2603 thermodynamics. Animal Feed Science and Technology 160(1-2), 1-22. doi: 2604 10.1016/j.anifeedsci.2010.07.002. 2605 Johnson, K.A., and Johnson, D.E. (1995). Methane emissions from cattle. Journal of 2606 Animal Science 73(8), 2483-2492. 2607 Leahy, S.C., Kelly, W.J., Altermann, E., Ronimus, R.S., Yeoman, C.J., Pacheco, D.M., et 2608 al. (2010). The Genome Sequence of the Rumen Methanogen 2609 Methanobrevibacter ruminantium Reveals New Possibilities for Controlling 2610 Ruminant Methane Emissions. Plos One 5(1). doi: 10.1371/journal.pone.0008926. 2611 Lee, S.S., Ha, J.K., and Cheng, K.J. (2000). Relative contributions of bacteria, protozoa, 2612 and fungi to in vitro degradation of orchard grass cell walls and their interactions. 2613 Applied and Environmental Microbiology 66(9), 3807-3813. doi: 2614 10.1128/aem.66.9.3807-3813.2000. 2615 Li, F., and Guan, L.L. (2017). Metatranscriptomic profiling reveals linkages between the 2616 active rumen microbiome and feed efficiency in beef cattle. Applied And 2617 Environmental Microbiology. doi: 10.1128/AEM.00061-17. 2618 Li, M., Penner, G.B., Hernandez-Sanabria, E., Oba, M., and Guan, L.L. (2009). Effects of 2619 sampling location and time, and host animal on assessment of bacterial diversity 2620 and fermentation parameters in the bovine rumen. Journal of Applied Microbiology 2621 107(6), 1924-1934. doi: 10.1111/j.1365-2672.2009.04376.x. 2622 Li, Z.P., Wright, A.D.G., Si, H.Z., Wang, X.X., Qian, W.X., Zhang, Z.G., et al. (2016). 2623 Changes in the rumen microbiome and metabolites reveal the effect of host 2624 genetics on hybrid crosses. Environmental Microbiology Reports 8(6), 1016-1023. 2625 doi: 10.1111/1758-2229.12482. 2626 Malmuthuge, N., Li, M., Chen, Y., Fries, P., Griebel, P.J., Baurhoo, B., et al. (2012). 2627 Distinct commensal bacteria associated with ingesta and mucosal epithelium in 2628 the gastrointestinal tracts of calves and chickens. Fems Microbiology Ecology 2629 79(2), 337-347. doi: 10.1111/j.1574-6941.2011.01220.x. 2630 Martínez, I., Jaehyoung, K., Duffy, P.R., Schlegel, V.L., and Walter, J. (2010). Resistant 2631 Starches Types 2 and 4 Have Differential Effects on the Composition of the Fecal 2632 Microbiota in Human Subjects. PLoS ONE 5(11), 1-11. doi: 2633 10.1371/journal.pone.0015046. 2634 McAllister, T.A., Bae, H.D., Jones, G.A., and Cheng, K.J. (1994). Microbial attachment 2635 and feed digestion in the rumen. Journal of Animal Science 72(11), 3004-3018. 2636 Montanholi, Y.R., Swanson, K.C., Palme, R., Schenkel, F.S., McBride, B.W., Lu, D., et al. 2637 (2010). Assessing feed efficiency in beef steers through feeding behavior, infrared 2638 thermography and glucocorticoids. Animal 4(5), 692-701. doi: 2639 10.1017/s1751731109991522. 2640 Morgavi, D.P., Forano, E., Martin, C., and Newbold, C.J. (2010). Microbial ecosystem and 2641 methanogenesis in ruminants. Animal 4(7), 1024-1036. doi: 2642 10.1017/s1751731110000546. 2643 Morgavi, D.P., Kelly, W.J., Janssen, P.H., and Attwood, G.T. (2012). Rumen microbial 2644 (meta)genomics and its application to ruminant production. animal 7(s1), 184-201. 2645 doi: 10.1017/S1751731112000419. 2646

Neves, A.L.A., Li, F., Ghoshal, B., McAllister, T., and Guan, L.L. (2017). Enhancing the 2647 Resolution of Rumen Microbial Classification from Metatranscriptomic Data Using 2648 Kraken and Mothur. Frontiers in Microbiology 8(2445). doi: 2649 10.3389/fmicb.2017.02445. 2650 NRC (2016). Nutrient Requirements of Beef Cattle. Washington, D.C., USA: National 2651 Academy Press. 2652 Olfert, E., Cross, B., and McWilliams, A. (1993). Guide to the Care and Use of 2653 Experimental Steers. Ottawa, ON, Canada: Canadian Council on Animal Care. 2654 Rico, D.E., Preston, S.H., Risser, J.M., and Harvatine, K.J. (2015). Rapid changes in key 2655 ruminal microbial populations during the induction of and recovery from diet- 2656 induced milk fat depression in dairy cows. British Journal of Nutrition 114(3), 358- 2657 367. doi: 10.1017/s0007114515001865. 2658 Roehe, R., Dewhurst, R.J., Duthie, C.-A., Rooke, J.A., McKain, N., Ross, D.W., et al. 2659 (2016). Bovine Host Genetic Variation Influences Rumen Microbial Methane 2660 Production with Best Selection Criterion for Low Methane Emitting and Efficiently 2661 Feed Converting Hosts Based on Metagenomic Gene Abundance. Plos Genetics 2662 12(2). doi: 10.1371/journal.pgen.1005846. 2663 Rohart, F., Gautier, B., Singh, A., and Lê Cao, K.-A. (2017). mixOmics: An R package for 2664 ‘omics feature selection and multiple data integration. PLoS Computational Biology 2665 13(11), 1-19. doi: 10.1371/journal.pcbi.1005752. 2666 Romain, G., Thomas, T., and Margaret, E.K.E. (2009). Ecological Conditions Affect 2667 Evolutionary Trajectory in a Predator-Prey System. Evolution (3), 641. doi: 2668 10.1111/j.1558-5646.2008.00559.x. 2669 Romero-Perez, A., Okine, E.K., McGinn, S.M., Guan, L.L., Oba, M., Duval, S.M., et al. 2670 (2014). The potential of 3-nitrooxypropanol to lower enteric methane emissions 2671 from beef cattle. Journal of Animal Science 92(10), 4682-4693. doi: 2672 10.2527/jas.2014-7573. 2673 Russell, J.B., and Rychlik, J.L. (2001). Factors that alter rumen microbial ecology. 2674 Science 292(5519), 1119-1122. doi: 10.1126/science.1058830. 2675 Salonen, A., Lahti, L., Salojärvi, J., Holtrop, G., Korpela, K., Duncan, S.H., et al. (2014). 2676 Impact of diet and individual variation on intestinal microbiota composition and 2677 fermentation products in obese men. ISME Journal: Multidisciplinary Journal of 2678 Microbial Ecology 8(11), 2218-2230. doi: 10.1038/ismej.2014.63. 2679 Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., et al. (2003). 2680 Cytoscape: A software environment for integrated models of biomolecular 2681 interaction networks. Genome Research 13(11), 2498-2504. doi: 2682 10.1101/gr.1239303. 2683 Sniffen, C.J., and Robinson, P.H. (1987). Microbial growth and flow as influenced by 2684 dietary manipulations. Journal of Dairy Science 70(2), 425-441. 2685 Stevenson, D.M., and Weimer, P.J. (2007). Dominance of Prevotella and low abundance 2686 of classical ruminal bacterial species in the bovine rumen revealed by relative 2687 quantification real-time PCR. Applied Microbiology and Biotechnology 75(1), 165- 2688 174. doi: 10.1007/s00253-006-0802-y. 2689 Tajima, K., Aminov, R.I., Nagamine, T., Matsui, H., Nakamura, M., and Benno, Y. (2001). 2690 Diet-dependent shifts in the bacterial population of the rumen revealed with real- 2691

time PCR. Applied and Environmental Microbiology 67(6), 2766-2774. doi: 2692 10.1128/aem.67.6.2766-2774.2001. 2693 Tap, J., Furet, J.-P., Bensaada, M., Philippe, C., Roth, H., Rabot, S., et al. (2015). Gut 2694 microbiota richness promotes its stability upon increased dietary fibre intake in 2695 healthy adults. Environmental Microbiology 17(12), 4954-4964. doi: 10.1111/1462- 2696 2920.13006. 2697 Thompson, S. (2015). The effect of diet type on residual feed intake and the use of 2698 infrared thermography as a method to predict efficiency in beef bulls. M.Sc., 2699 University of Manitoba. 2700 Turnbaugh, P.J., Hamady, M., Yatsunenko, T., Cantarel, B.L., Duncan, A., Ley, R.E., et 2701 al. (2009). A core gut microbiome in obese and lean twins. Nature 457(7228), 480- 2702 484. doi: 2703 http://www.nature.com/nature/journal/v457/n7228/suppinfo/nature07540_S1.html 2704 . 2705 Van Soest, P.J., Robertson, J.B., and Lewis, B.A. (1991). Methods for dietary fiber, 2706 neutral detergent fiber, and nonstarch polysaccharides in relation to animal 2707 nutrition. Journal of Dairy Science 74(10), 3583-3597. 2708 Vanwonterghem, I., Evans, P.N., Parks, D.H., Jensen, P.D., Woodcroft, B.J., Hugenholtz, 2709 P., et al. (2016). Methylotrophic methanogenesis discovered in the archaeal 2710 phylum Verstraetearchaeota. Nature Microbiology 1, 16170. doi: 2711 10.1038/nmicrobiol.2016.170 2712 http://www.nature.com/articles/nmicrobiol2016170 - supplementary-information. 2713 Varga, G.A., and Kolver, E.S. (1997). Microbial and animal limitations to fiber digestion 2714 and utilization. The Journal Of Nutrition 127(5 Suppl), 819S-823S. doi: 2715 10.1093/jn/127.5.819S. 2716 Wang, S., Giller, K., Kreuzer, M., Ulbrich, S.E., Braun, U., and Schwarm, A. (2017). 2717 Contribution of Ruminal Fungi, Archaea, Protozoa, and Bacteria to the Methane 2718 Suppression Caused by Oilseed Supplemented Diets. Frontiers in Microbiology 2719 8(1864). doi: 10.3389/fmicb.2017.01864. 2720 Weimer, P.J. (2015). Redundancy, resilience, and host specificity of the ruminal 2721 microbiota: implications for engineering improved ruminal fermentations. Frontiers 2722 in Microbiology 6(296). doi: 10.3389/fmicb.2015.00296. 2723 Wolff, S.M., Ellison, M.J., Hao, Y., Cockrum, R.R., Austin, K.J., Baraboo, M., et al. (2017). 2724 Diet shifts provoke complex and variable changes in the metabolic networks of the 2725 ruminal microbiome. Microbiome 5(1), 60. doi: 10.1186/s40168-017-0274-6. 2726 Yang, T., Wei, Z., Friman, V.-P., Xu, Y., Shen, Q., Kowalchuk, G.A., et al. (2017). 2727 Resource availability modulates biodiversity-invasion relationships by altering 2728 competitive interactions. Environmental Microbiology 19(8), 2984-2991. doi: 2729 10.1111/1462-2920.13708. 2730 Yu, Z., and Morrison, M. (2004). Improved extraction of PCR-quality community DNA from 2731 digesta and fecal samples. BioTechniques 36(5), 808-812. doi: 2732 10.2144/04365ST04. 2733 Zhou, M., Chung, Y.H., Beauchemin, K.A., Holtshausen, L., Oba, M., McAllister, T.A., et 2734 al. (2011). Relationship between rumen methanogens and methane production in 2735 dairy cows fed diets supplemented with a feed enzyme additive. Journal of Applied 2736 Microbiology 111(5), 1148-1158. doi: 10.1111/j.1365-2672.2011.05126.x. 2737

Zhou, M., Hernandez-Sanabria, E., and Guan, L.L. (2009). Assessment of the Microbial 2738 Ecology of Ruminal Methanogens in Cattle with Different Feed Efficiencies. 2739 Applied and Environmental Microbiology 75(20), 6524-6533. doi: 2740 10.1128/aem.02815-08. 2741 Zhou, M., Hernandez-Sanabria, E., and Guan, L.L. (2010). Characterization of Variation 2742 in Rumen Methanogenic Communities under Different Dietary and Host Feed 2743 Efficiency Conditions, as Determined by PCR-Denaturing Gradient Gel 2744 Electrophoresis Analysis. Applied and Environmental Microbiology 76(12), 3776- 2745 3786. doi: 10.1128/aem.00010-10. 2746 Zhou, M., Hünerberg, M., Beauchemin, K.A., McAllister, T.A., Okine, E.K., and Guan, L.L. 2747 (2012). Individuality of ruminal methanogen/protozoa populations in beef cattle fed 2748 diets containing dried distillers' grain with solubles. Acta Agriculturae 2749 Scandinavica: Section A, Animal Science 62(4), 273-288. doi: 2750 10.1080/09064702.2013.788206. 2751 Zhou, M., Peng, Y.-J., Chen, Y., Klinger, C.M., Oba, M., Liu, J.-X., et al. (2018). 2752 Assessment of microbiome changes after rumen transfaunation: implications on 2753 improving feed efficiency in beef cattle. Microbiome 6(1), 62-62. doi: 2754 10.1186/s40168-018-0447-y. 2755 Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., and Smith, G.M. (2009). "Mixed 2756 Effects Models and Extensions in Ecology with R," in Mixed Effects Models and 2757 Extensions in Ecology with R.). 2758 2759

2760

2761

2762

2763

2764

2765

2766

2767

2768

2769

2.7 Tables and figures 2770

Table 2.1 Quantification of the copy numbers of microbial populations in the rumen of beef cattle fed forage or grain diets 2771

Sequence of dietary treatments4

Microorganisms

Forage - Forage Forage - Grain Grain - Forage Grain - Grain Bacteria1 5.9 ± 6.80 2.5 ± 1.67 6.0 ± 4.24 3.2 ± 2.82 5.8 ± 4.39 1.9 ± 1.46 5.0 ± 3.08 5.7 ± 3.44

× 1011 × 1011 × 1010 × 1010 × 1010 × 1011 × 1011 × 1011

Fungi2 1.2 ± 1.03 9.1± 7.06 1.1 ± 0.89 8.8 ± 8.6 5.0 ± 6.28 2.4 ± 4.59 2.1 ± 3.89 2.6 ± 2.79

× 104 × 103 × 105 × 104 × 103 × 104 × 104 × 104

Methanogens1 3.7 ± 1.71 3.1± 0.93 2.9 ± 1.29 2.5 ± 1.38 3.0 ± 1.68 4.4 ± 2.91 5.4 ± 3.95 7.5 ± 5.97

× 108 × 108 × 108 × 108 × 108 × 108 × 108 × 108

Protozoa3 4.8 ± 7.73 2.9 ± 4.10 7.6 ± 0.10 6.4 ± 0.11 2.6 ± 3.06 1.9 ± 3.83 1.0 ± 1.12 6.8 ± 6.14

× 107 × 107 × 107 × 107 × 107 × 107 × 108 × 107

1Copy number of 16S rRNA (Mean ± SD) ⁄ mL of rumen fluid; 2772

2Copy number of ITS (Mean ± SD) ⁄ mL of rumen fluid; 2773

3Copy number of 18S rRNA (Mean ± SD) ⁄ mL of rumen fluid; 2774

4The hyphen (-) means diet change. 2775

(A) (B) PC - corr > 0 non-centred PCA for norm: LOG PC - corr < 0

Forage Forage group Grain -14 -14

)

8 Grain group

R Cut-off=0.27

-16 5 -16

= -18 -18 p

(

0.0 0.2 0.4 0.6 0.8 -0.2 0.0 0.2 0.4 PC6 ( p < 0.01, AUC = 0.882, AUPR = 0.871 )

0 -0.2 0.0 0.2 0.4

Forage Grain

)

2 . 2

7 0 . 0 PC - corr > 0

PC - corr < 0

1 Forage group

0 -2 -2 <

(

6 Grain group

-4 -4 Cut-off=0.50

0.0 0.2 0.4 -3 0 3 6 PC1 ( p < 0.01, AUC = 0.982, AUPR = 0.966 )

0.4

0.3

0.2

0.1

0.0 -3 0 3 6

Figure 2.1 Rumen microbial population and phenotypic traits in response to diet changes. Linear dimensionality reduction by PC-Corr separated rumen microbes (A) and phenotypic traits (C) recorded in cattle fed forage (black) or grain diets (red). PC-Corr detected differences (Mann-Whitney test, P <0.01) in rumen microbes between forage and grain diets along principal component 6 (PC6), whose loadings were used to build a microbial network at a cut-off (Pearson correlation) of 0.27 (B). Differences (Mann-Whitney test, P <0.01) in phenotypic traits were also observed for forage and grain diets along PC1 and PC6, whose loadings were used for the construction of a network at a cut-off of 0.50 (D).

Y = 10.2 (±0.27) - 0.2 (±0.06) × DMI.std + 0.16 Y = 3.7 (±0.31) - 0.1 (±0.07) × (A) (±0.06) × Carryover + 0.30 (±0.11) × Period (B) DMI.std - 0.01 (±0.07) × P < 0.001 ( χ2 = 12.65) 5.5 Carryover + 0.27 (±0.13) × adjusted- R2 = 0.70 Period

) 2 0 P < 0.1 ( χ = 3.38)

12 g 5.0 ) 2

0 adjusted- R = 0.66

) 4.5

)

o e Diets Diets

(

a Forage Forage

m 11

r 4.0

n t Grain Grain

y 3.5

(

e 10 i

g 3.0

2.5 −2 −1 0 1 2 3 −2 −1 0 1 2 3 (C) DMI.std (kg/DMIday) Y = 10.4 (±0.23) - 0.2 (±0.05) × (D) DMI.stdDMI (kg/day) Starch.std + 0.19 (±0.06) × Y= 10.6 (±0.22) - 0.1 (±0.05) × 12.5 Carryover + 0.14 (±0.08) × Period 12.5 ADG.std + 0.11 (±0.06) × P < 0.001 ( χ2 = 15.01) Carryover + 0.10 (±0.09) × Period

) 2 2 ) 0 P < 0.05 ( χ = 5.52)

adjusted- R = 0.72 0

g 12.0 12.0 2 g adjusted- R = 0.66

) 11.5 ) 11.5

e Diets Diets

(

a Forage a Forage

u 11.0 11.0

t Grain t Grain

C 10.5 10.5

(

10.0 c 10.0

−1 0 1 2 −2 0 2 Starch intake ADG Starch.std Intake (kg/day) ADG.std (kg/day)

Figure 2.2 Dynamics of rumen bacteria and fungi in response to DMI, starch intake, and average daily gain (ADG). Effect of DMI (A and B), starch intake (C) (kg/day), and ADG

(kg/day) on the copy numbers of bacteria and fungi (Log10).

(A) (B)

Y = 10.8 (±0.20) + 0.1 (±0.05) × NDF + Y = 10.6 (±0.23) + 0.1 (±0.05) × CH4/kg DMI + 12.5 0.18 (±0.06) × Carryover – 0.12 (±0.08) × 12.5 0.11 (±0.06) × Carryover + 0.12 (±0.09) × Period Period P < 0.01 ( χ2 = 9.46) P < 0.05 ( χ2 = 4.46) adjusted- R2 = 0.71 adjusted- R2 = 0.68

12.0 12.0

)

11.5 11.5

)

L L Diets Diets

(

a a Forage Forage i

u u 11.0 Grain 11.0 Grain

(

r r 10.5 10.5

10.0 10.0

−1 0 1 2 −2 −1 0 1 2 NDF intake CH4DMI NDF.std Intake (kg/day) CH4/kg DMI.std

Figure 2.3 Dynamics of rumen bacteria in response to NDF intake and CH4/kg DMI. Effect of NDF intake (A) and CH4/kg DMI (B) (kg/day) on the copy numbers of bacteria (Log10).

(A) (B) High Low Stable High Low Stable Y= 163.1 (±3.95) + Y= 166.6 (±7.32) + 5.3 Y= 167.9 (±5.74) + 16.5 10.0 Y= 6.0 (±0.11) + Y= 5.9 (±0.21) + Y= 6.1 (±0.16) + 0.6 250

0.2 (±0.09) × 0.1 (±0.20) × (±0.14) × NDF.std ) 2.7 (±3.43) × (±7.51) × NDF.std (±5.35) × NDF.std

) s NDF.std n NDF.std NDF.std

a u 200

n G Groups Groups

i 7.5

High m High

)

a Low i Low

) ( 150

e d Stable Stable

(

e 5.0

(

R 100 ( Diets Diets

R Forage i Forage

2.5 Grain e Grain

F e 50 P < 0.05 ( χ2 = 8.75) F P < 0.05 ( χ2 = 7.00) adjusted- R2 = 0.51 adjusted- R2 = 0.34 0.0 0

−1 0 1 2 −1 0 1 2 −1 0 1 2 −1 0 1 2 −1 0 1 2 −1 0 1 2 (C) NDF.std IntfNDFake .std(kg/day) NDfNDFF.st.stdd Intake (kg/day) High Low Stable High Low Stable (D) Y= 165.7 (±6.46) + Y= 165.2 (±6.31) + 22.0 10.0 Y= 6.0 (±0.11) + Y= 6.0 (±0.19) + Y= 6.0 (±0.18) + 0.6 250 Y= 163.1 (±3.92) + 6.28 (±6.10) × (±6.36) × NDF.std 0.2 (±0.09) × 0.3 (±0.17) × (±0.18) × NDF.std ) 2.8 (±3.27) × NDF.std

s ) NDF.std NDF.std NDF.std

u a 200

Groups n Groups

G 7.5 i High High

)

a / Low Low

)

( 150

Stable e Stable

(

e 5.0

(

R 100 D Diets Diets

(

Forage i Forage

C 2.5 Grain e Grain

e F 2 50 2 P < 0.1 (χ = 4.71) F P < 0.05 ( χ = 9.20) adjusted- R2 = 0.49 adjusted- R2 = 0.37 0.0 0

−1 0 1 2 −1 0 1 2 −1 0 1 2 −1 0 1 2 −1 0 1 2 −1 0 1 2 NDF.stdfNDF Inta.stdke (kg/day) NDF.std fNDFInta.stdke (kg/day)

Figure 2.4 Feed conversion ratio (FCR) and Feeding Time (in minutes) in response to the interactions of NDF intake and the magnitude of change in host microbial population. Interactive effects of NDF intake (kg/day) and the magnitude of microbial shift (High, Stable, Low) on FCR and Feeding Time when groups of bulls were created from the magnitude of change in bacteria (A and B) and fungi (C and D).

Chapter 3

Enhancing the resolution of rumen microbial classification from metatranscriptomic data using Kraken and Mothur

3.1 Introduction The success of microbiome studies (composition, structure, diversity, and function) is primarily ascribable to the development of bioinformatics tools specially tailored to overcome the technical challenges posed by the analysis of massively paralleled, high-throughput sequencing data (Simon and Daniel, 2011; Siegwald et al., 2017). These bioinformatics tools make use of several techniques (e.g., read mapping, k-mer alignment, and composition analysis) (Piro et al., 2017) and can be categorized into two distinct groups: 1) programs that use all available genome sequences (Lindgreen et al., 2016), also called assignment-first approaches (Siegwald et al., 2017) (e.g., CLARK - Ounit et al., 2015; GOTTCHA - Freitas et al., 2015; KRAKEN - Wood and Salzberg, 2014; MG-RAST - Meyer et al., 2008), and 2) programs that target a set of marker genes (Lindgreen et al., 2016), also known as clustering- first approaches (Siegwald et al., 2017) (e.g., QIIME - Caporaso et al., 2010; MOTHUR - Schloss et al., 2009; MetaPhlAn - Segata et al., 2012; mOTU - Sunagawa et al., 2013). In the assignment-first tools, all reads are assigned to the lowest taxonomy unit (lower common ancestor-LCA) within a reference database based on their annotations, while in the clustering- first approaches the reads are grouped into Operational Taxonomic Units (OTUs) using different OTU picking strategies (closed or open reference) to assign reads to a taxonomic group based on their sequence similarities (Siegwald et al., 2017).

However, most of the above studies are focused on demonstrating how single analytical steps (e.g., sequence pre-processing, OTU clustering or taxonomic assignment) generated by the existing tools impact the microbial classification in real or simulated datasets derived from the Human Microbiome Project (Siegwald et al., 2017). Comparison of methodologies to comprehensively classify the rumen microbiome is lacking which may be in part due to its complexity, as the rumen microbial community consists of bacteria, archaea, protozoa and fungi (Russell and Rychlik, 2001). A recent study by Li et al. (2016a) developed a Mothur (Schloss et al., 2009) based pipeline to assess active rumen microbiota from data generated from total RNA sequencing. Later, the same researchers applied this pipeline to

investigate linkages between the active rumen microbiome (structure and function) and feed efficiency in beef cattle using metatranscriptomics (Li and Guan, 2017). Using the developed mothur-based pipeline for taxonomic assignment, the authors identified that the active microbial taxa differed in the rumen of cattle with differing feed efficiency and suggested that the active rumen microbiome is one of the biological factors that may contribute to variations in feed efficiency of beef cattle (Li and Guan, 2017). There were two steps employed in taxonomic classification by Li et al. (2016a): bacterial sequences belonging to V1-V3 regions were extracted from the aligned Greengenes database, and archaeal sequences belonging to the V6-V8 regions were aligned with a rumen-specific archaeal 16S rRNA gene database (Janssen and Kirs, 2008). Despite the efficacy of this pipeline, it still remains a challenge for researchers to determine which approach (assignment- or clustering-first methods) of taxonomic classification delivers the most realistic representation of rumen microbial ecology.

In the current study, we propose a comparative analysis of the outcomes of Kraken (Wood and Salzberg, 2014) and the pipeline of Li et al. (2016a) with a focus on the biological interpretation of the rumen microbial classification from the perspective of two conceptually different software packages. Unlike the pipeline developed by Li et al. (2016a), Kraken algorithms can make multiple comparisons of single or assembled k-mers against any hypervariable region, providing useful information regarding a particular species detected in a region of the 16S rRNA gene that is different from the targeted internal conserved region initially sequenced (Wood and Salzberg, 2014; Valenzuela-González et al., 2016). Although Kraken algorithms have been originally designed to assign taxonomic identity to short DNA reads (Wood and Salzberg, 2014), studies have shown that Kraken is also useful to provide taxonomic classification for long (up to 1352.1 ± 153.72 bp) metagenomic DNA sequences (Valenzuela-González et al., 2016). Therefore, the objectives of this study were (i) to compare and contrast the pipeline of Li et al. (2016a) and Kraken to assess the taxonomic profiles of rumen bacteria and archaea and (ii) to investigate the impact of the comparative analysis of both analytical approaches on the biological interpretation of the rumen microbial classification obtained from cattle exhibiting different feed efficiencies.

3.2 Materials and methods

3.2.1 Animal study and sampling

The experimental procedures described in this study were approved by the Veterinary Services and the Animal Care Committee, University of Manitoba, Canada, to ensure that animals were cared for in compliance with those ethics. Rumen contents were collected from 12 purebred Angus bulls (mean age of 249 ± 22 days and average body weight of 313.9 ± 32 kg) raised in confinement at the Glenlea Research Station located at the University of Manitoba according to the guidelines of the Canadian Council on Animal Care (CCAC) (Olfert et al., 1993), with bulls being fed a forage diet over two 80-day feeding periods (with a 20-day adaptation in between) as described by Thompson (2015). In the current study, 250 ml of rumen contents (liquid and solid fractions) were collected at the end of the second feeding period using a Geishauser oral probe (Duffield et al., 2004), immediately snap frozen in liquid nitrogen, and stored at -80°C for later processing. The feed intake of individual bulls was recorded using the GrowSafe® feeding system (GrowSafe Systems Ltd., Airdrie, Alberta, CA) and the feed conversion rate (FCR) was calculated as a ratio of dry matter intake to average daily gain (computed on a biweekly basis; Montanholi et al., 2010). The bulls were ranked into two groups: high (n=6) and low (n=6) FCR, with high (H-FCR) and low (L-FCR) standing for inefficient and efficient cattle in terms of diet utilization, respectively.

3.2.2 RNA extraction and sequencing

Total RNA was extracted from rumen samples using the TRIzol protocol based on the acid guanidinium-phenol-chloroform method (Chomczynski and Sacchi, 2006; Béra-Maillet et al., 2009) with the modified procedures described by Li et al. (2016a). Briefly, ~200 mg of rumen sample was subjected to RNA extraction with the addition of 1.5 ml of TRIzol reagent (Invitrogen, Carlsbad, CA, USA), followed by 0.4 ml of chloroform, 0.3 ml of isopropanol, and 0.3 ml of high salt solution (1.2 M sodium acetate, 0.8 M NaCl) for the extraction protocol (Li et al., 2016a). The yield and integrity of the RNA samples were determined using a Qubit 2.0 fluorimeter (Invitrogen, Carlsbad, CA, USA) and Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). RNA samples were subjected to downstream RNA- sequencing only if they exhibited RNA with integrity number (RIN) higher than 7.0. Briefly, total RNA (100 ng) of each sample was used for library construction using the TruSeq RNA sample prep v2 LS kit (Illumina, San Diego, CA, USA) without the mRNA enrichment step (Li et al., 2016a). The quality of libraries was assessed using Agilent 2200 TapeStation (Agilent

Technologies) and Qubit 2.0 fluorimeter (Invitrogen). Finally, cDNA fragments (∼140 bp) were

paired-end (2 X 100 bp) sequenced using an Illumina HiSeq 2000 system at the McGill University and Génome Québec Innovation Centre (Montréal, QC, Canada).

3.2.3 Pipeline settings

A flow chart is shown in Figure 3.1 to present the software parameters used to obtain the microbial classification from either Mothur (Schloss et al., 2009) or Kraken (Wood and Salzberg, 2014) taxonomic assignment strategies. In the pre-processing steps, all fastq- formatted sequences were firstly uploaded into FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ for quality control and removal of ambiguous sequences, and then the software Trimmomatic (version 0.32; Bolger et al., 2014) was used to trim residual artificial sequences, cut bases with quality scores below 20, and remove reads shorter than 50 bp (Li et al., 2016a). After pre-processing, SortMeRNA (version 1.9; Kopylova et al., 2012) was used to sort the filtered reads into fragments of 16S rRNA (for taxonomic identification using Mothur) based on the rRNA reference databases SILVA_SSU (release 119; Quast et al., 2013) and mRNA (for microbial classification using Kraken). In the pipeline developed by Li et al. (2016a), sorted paired-end reads belonging to bacterial and archaeal 16S rRNA were joined to increase the read length by combining the forward and reverse sequences. After the 16S rRNA sequences were enriched, downstream analyses were performed using Mothur (version 1.31.2; Schloss et al., 2009) as described by Kozich et al. (2013) (Figure 3.1). For taxonomic classification, bacterial and archaeal 16S rRNA sequences were aligned with the V1-V3 region-enriched Greengenes database (DeSantis et al., 2006) and the V6-V8 region-enriched rumen-specific archaea database (Janssen and Kirs, 2008, which was updated from Kittelmann et al., 2013), respectively. De novo chimera detection was then conducted using UCHIME (Edgar et al., 2011), and non-chimeric sequences were taxonomically assessed using a naive Bayesian method (Wang et al., 2007). The pipeline developed by Li et al. (2016a) will be referred as Mothur through the rest of the paper.

As for the Kraken pipeline (Wood and Salzberg, 2014), newly developed Perl scripts were used to retrieve all complete genomes of bacteria (5,294) and archaea (209) from NCBI (RefSeq) (May 2016), to build a Kraken standard database (June 2016) based on their annotations at the lowest taxonomic level (Figure 3.1). Ninety-one complete genomes from organisms isolated from the rumen or from ruminant feces or saliva deposited in the

Hungate1000 project were also retrieved from JGI's IMG database (using NCBI Taxon IDs). After downloading the genomes, the script kraken-build (option --build) was used to set the lowest common ancestors (LCAs) in a bacteria-archaea joint database (size: 115G; number of sequences mapped to profiles: 10,174; and time for database construction: 6h33m35s). Thereafter, each pair of mRNA sequences was assembled by MEGAHIT (Li et al., 2015), with the resulting contigs (with average extension of 472.31 ± 31.10 bp) being assigned by Kraken (through k-mer discrimination) to the LCA in the customized standard database for microbial classification (Figure 3.1). Full taxonomic names associated with each classified sequence (separated from unclassified reads using kraken option --preload) and standard ranks (from domain to species) for each taxon were provided by kraken-translate and kraken-mpa-report (Figure 3.1).

3.2.4 Statistical analysis

In this study, a phylotype was considered as classified by both methods if it had at least one count detected in the 12 samples. For comparisons between H-FCR and L-FCR groups, we investigated only bacterial and archaeal profiles with a relative abundance > 0.1% prevalent in at least three samples (3 out 6) to avoid sparsely observed counts, which tend to introduce noise in the analysis (Chen and Li, 2013). The ANCOM procedure (Mandal et al., 2015), which uses an alternative normalization approach called Aitchison’s log-ratio transformation (Aitchison, 1982), was then used to normalize the sequence data and to compare the normalized log ratio of the abundance of each taxon to the abundance of all remaining taxa (Weiss et al., 2017). To deal with zero counts in the datasets, ANCOM used an arbitrary pseudo count value of 0.001 (Mandal et al., 2015). Thereafter, Wilcoxon rank sum tests were calculated on each log ratio to find differences between feed efficiency groups (H-FCR vs. L- FCR) as provided by each classification method (Mothur or Kraken) (Figure 3.1). The p-value of each test were adjusted into false discovery rate (FDR) using the Benjamini-Hochberg algorithm (Benjamini and Hochberg, 1995), and a threshold of FDR lower than 0.15 (Korpela et al., 2016) was applied to determine the significance due to the small sample size of this study. Correlation circle plots and relevance networks for core bacterial genera and archaeal species (with a relative abundance > 0.1% detected in all rumen samples; Li and Guan, 2017) were generated from the output of regularized canonical correlation (rCC) analysis as implemented in the R package mixOmics (Gonzalez et al., 2008) and Cytoscape 3.4.0 (Shannon et al., 2003). Before running rCC analysis, the data was normalized by total sum

scaling (TSS) (dividing each taxon count by the total number of counts in each individual sample to account for uneven sequencing depths across samples) and then transformed by centered log ratio to project the data from a simplex to a Euclidian space (Aitchison, 1982; Mandal et al., 2015; Cao et al., 2016). Then, estimation of regularization parameters (λ1 and λ2) and canonical correlations were calculated using the cross-validation procedure (Gonzalez et al., 2008). Finally, alpha-diversity indexes were calculated using the R package vegan (as provided by each classification method) and compared between FCR groups (H- FCR vs. L-FCR) using paired Wilcoxon signed rank test. All statistical procedures were performed using R 3.3.2 (R Core Team, 2016).

3.2.5 Data submission

The datasets analyzed in this study were submitted to NCBI Sequence Read Archive (SRA) under the accession number PRJNA403833.

3.3 Results

3.3.1 Taxonomic distribution of the microbial profiles performed by Mothur or Kraken

In this study, two bioinformatics approaches, Kraken and a Mothur-based pipeline developed in-house by Li et al. (2016a), were used to obtain taxonomic classifications (bacteria and archaea) of the ruminal microbiota in bulls exhibiting different (P < 0.05) feed efficiencies (average FCR for H-FCR group= 7.64 kg dry matter intake (DMI)/kg gain; average FCR for L-FCR group= 5.71 kg DMI/kg gain; P = 0.008). Taking into consideration the total number of microbial taxa in the samples, Kraken identified a higher number of bacterial and archaeal phylotypes at all taxonomic ranks than Mothur (Table 3.1). At the phylum level, the results of bacterial profiles revealed a similar taxa distribution of the most abundant taxa classified by both methods (Tables 3.1 and 3.2), with Bacteroidetes, Firmicutes, and Proteobacteria being highly abundant and accounting for approximately 80% of the total bacterial community. However, Spirochaetes (4.9%) were the fourth-most abundant taxon identified by Kraken, followed by Verrucomicrobia (2.3%), Actinobacteria (2.1%), Tenericutes (1.9%), and Fibrobacteres (1.2%). In contrast, Fibrobacteres (3.4%) was found to be the fourth-most abundant taxon detected by Mothur, followed by Spirochaetes (2.2%), Verrucomicrobia (1.7%), Tenericutes (0.8%), and Cyanobacteria (0.6%). Although there was some congruency (69 commonly detected taxa) at the most resolvable level (up to genus) of bacteria in between the two pipelines, an additional 159 genera were exclusively identified by Kraken. Genera 90

such as Ruminiclostridium, Lachnoclostridium, and Acholeplasma were uniquely identified by Kraken, whereas Ruminobacter, Coprococcus, YRC22, and Oscillospira were exclusively detected by Mothur. As for the most abundant genera, Kraken revealed Prevotella (33.5%), Treponema (4.1%), Ruminoccocus (4.1%), Ruminiclostridium (3.2%), Bacteroides (3.0%), Butyrivibrio (2.4%) and Clostridium (2.2%) at relatively high abundances, while Mothur identified Prevotella (22.6%), Ruminoccocus (14.6%), Ruminobacter (4.9%), Fibrobacter (4.3%), Treponema (2.4%), and Butyrivibrio (1.2%) as more abundant. It is worth noting that although Mothur could theoretically classify sequences at the species level, it was not able to assign bacterial contigs further than the genus level in the current study. Conversely, Kraken detected 423 species (Tables 3.1 and 3.2) such as Prevotella ruminicola (27.6%), Butyrivibrio proteoclasticus (2.8%), Treponema succinifaciens (2.6%), Ruminiclostridium sp KB18 (2.2%), and Fibrobacter succinogenes (1.8%). A complete list of all bacteria phylotypes (in all taxonomic ranks) classified by Mothur or Kraken is provided in Supplementary Tables 1 and 2, respectively. In addition, the direct comparisons of the bacterial taxonomic assignments obtained from both methods across all samples are included in Supplementary Table 3.

In terms of archaea identification, both methods exhibited similar results on the abundance of Methanomassiliicoccaceae (previously referred to as RCC), which comprised more than 65% of the total archaeal families (Table 3.3). However, the two methods generated significantly different archaeal profiles at the species level, with 7 species being exclusively identified by Kraken and 4 taxa being exclusively detected by Mothur (Tables 3.1 and 3.3). Only Methanobrevibacter ruminantium was commonly detected by the two methods, being the second-most abundant species classified by Mothur and the seventh-most abundant identified by Kraken. A detailed list of archaeal classification (in all taxonomic ranks) for Mothur or Kraken can be found in the Supplementary Tables 1 and 2, respectively, together with the information on the direct comparison of the archaeal taxonomic assignments obtained from both methods across all samples included in Supplementary Table 4.

3.3.2 Differences in relative abundances of taxa in H- vs. L-FCR rumen samples

To evaluate how the above two approaches affect the biological interpretation of bacteria and archaea diversity and community structure, comparisons of rumen microbiota between H- and L-FCR cattle were performed. Differences in microbial abundance between H- and L- FCR datasets were found to be minimal (making up less than 1% of the total microbial

community), regardless of the classification method (Tables 3.2 and 3.3). In this regard, only the family R4-41B (exclusively detected by Mothur) were more (FDR < 0.15) abundant in the rumen of L-FCR bulls, while the family Actinomycetaceae was more (FDR < 0.15) abundant in L-FCR samples classified by Kraken (Table 3.2). Methanococcaceae and Xenorhabdus exhibited a higher (FDR < 0.15) abundance in the rumen of H-FCR bulls when sequences were exclusively classified by Kraken (Tables 3.2 and 3.3).

In addition, alpha-diversity indexes of bacteria (genus level) and archaea (species level) were compared between H- and L-FCR groups to determine how the two pipelines differed in microbial biodiversity estimates. Shannon, Inverse Simpson and Simpson (with rarefy) indexes were higher (P < 0.05, paired Wilcoxon signed rank test) in H-FCR than in L-FCR bulls as shown by both pipelines (Table 3.4). On the other hand, a higher (P < 0.05, paired Wilcoxon signed rank test) archaeal diversity in the H-FCR group was observed only by the Kraken pipeline (Table 3.4).

3.3.3 Potential interactions between bacteria and archaea detected by Mothur or Kraken

To investigate interactions among different taxa classified by Kraken or Mothur, rCC analysis was implemented to identify relationships within and between bacteria and archaea communities. Our results revealed that bacteria and archaea interactions were quite contrasting between the two methods, with the microbial groups exhibiting different correlation outcomes as shown in Figure 3.2. Within bacterial communities, negative correlations between Prevotella, Treponema, Fibrobacter and Ruminobacter, Butyrivibrio, and Ruminoccocus were observed using the Mothur pipeline (Figure 3.2a), while Prevotella and Bacteroides were negatively correlated with Treponema, Fibrobacter and Ruminoccocus when Kraken was used (Figure 3.2c). Associations within archaeal species were also different between the two methods, with Methanobrevibacter gottschalkii and Methanobrevibacter ruminantium being negatively correlated with each other from the Mothur pipeline, and Candidatus Methanoplasma termitum and Candidatus Methanomethylophilus alvus exhibiting negative correlations with each other in the Kraken pipeline (Figures 3.2a and 3.2c). Relevance networks of the associations between bacteria and archaea revealed a positive correlation between Methanobrevibacter ruminantium and Fibrobacter, RFN20, Treponema, and BF311, and a positive correlation between Methanobrevibacter gottschalkii and

Ruminococcus, Butyrivibrio and Succiniclasticum based on the microbial classification by Mothur (Figure 3.2b). On the other hand, the positive correlations were detected between Candidatus Methanoplasma termitum and Prevotella, Porphyromonas, Bacillus, Sphingobacterium, and Moraxella, as well as between Candidatus Methanomethylophilus alvus and Fibrobacter, Eubacterium, and Mageeibacillus in the classification provided by Kraken (Figure 3.2d).

3.4 Discussion

In this study, the comparison of taxonomic outcomes of two pipelines, Mothur (developed by Li et al., 2016) and Kraken (developed by Wood and Salzberg, 2014, and adapted to the conditions of this study), was performed to determine which is a better approach in rumen microbial classification when total RNA-seq data were used. The advent of high-throughput sequencing has greatly advanced our knowledge of the ecology and functional capacity of rumen microbes and their role in converting low-quality and unusable feedstuffs into energy sources for host productivity (McCann et al., 2017). As a result, an assiduous effort has been made to unveil the linkage between the rumen microbiota and phenotypic traits of interest such as feed efficiency (Li and Guan, 2017), enzyme discovery (Qi et al., 2011) and methane emissions (Kittelmann et al., 2014; Shi et al., 2014; Kamke et al., 2016). Metagenomic studies have shown that the host may regulate the microbiota and its metabolic activity in relation to feed efficiency (FCR) through host-microbiome cross talk genes such as TSTA3 (GDP-L- fucose synthetase) and Fucl (L-fucose isomerase), suggesting that the relative abundance of these genes could be used as a predictor for host feed efficiency (Roehe et al., 2016). Although the number of rumen metagenomics and metatranscriptomics studies has grown enormously over the last couple of years (McCann et al., 2017), the functional outcomes and biological interpretation of omics data strongly depend on the computational methods used (Simon and Daniel, 2011; Siegwald et al., 2017). In this study, both Mothur and Kraken pipelines showed the rumen of the bulls to be dominated by Prevotella, Treponema, Ruminoccocus, Fibrobacter, and Butyrivibrio, which are considered as part of a “core bacterial microbiome” (Henderson et al., 2015). In addition to the mutual “core microbiome” shared by the two pipelines at the genus level, Kraken detected a relatively high abundance of 1) Prevotella ruminicola (Supplementary Table 2), which is involved in the ruminal digestion of hemicellulose and pectin (Marounek and Duskova, 1999); 2) Fibrobacter succinogenes (Supplementary Table 2), a gram-negative, fiber degrader species (Suen et al., 2011); and 3)

non-motile species within the Ruminoccocus genus (Supplementary Table 2) that share different niches (La Reau et al., 2016): R. bicirculans, which selectively utilizes hemicelluloses but not cellulose or arabinoxylan (Wegmann et al., 2014), and R. albus, which is capable of digesting cellulose and xylan (Christopherson et al., 2014).

Interestingly, both methods identified about 1% of Cyanobacteria (Supplementary Tables 1 and 2), corroborating the findings of previous studies that have reported low abundances of these oxygenic phototrophic bacteria in the rumen of dairy (Scharen et al., 2017) and beef cattle (Li and Guan, 2017), and of camels (Gharechahi et al., 2015). Cyanobacteria are aerobic bacteria that can perform carbohydrate fermentation in a deficient

N2 concentration (heterocystous) or in a combination of N2 deficiency and anoxic conditions (nonheterocystous) (Nandi and Sengupta, 1998). Although the ruminal environment is widely considered to be anaerobic, significant concentrations of O2 (60 and 100 nmol/min per mL) can be detected in the rumen fluid (Newbold et al., 1996), indicating that the presence of

Cyanobacteria in the rumen may be related to O2 scavenging and sugar fermentation performed under restrict aerobic conditions. It is important to mention that although Cyanobacteria has been widely detected in aqueous and soil environments (Williams et al., 2004; Cruz-Martinez et al., 2009), the identification of this phylum in the mammals’ gut has raised critical questions on what roles these organisms may play in aphotic and anaerobic habitats (Soo et al., 2014) like the rumen. Recent researches have reported that gut Cyanobacteria are highly conserved but their 16S rRNA gene phylogenetic tree differed from the photosynthetic Cyanobacteria, which led to the designation of a new candidate class called Melainabacteria (Soo et al., 2014) whose members are capable of fermenting a range of sugars (e.g., glucose, fructose, sorbitol) into acetate and butyrate in the gut (Di Rienzi et al., 2013). Neither Kraken nor Mothur identified Melainabacteria in the samples, demonstrating that further studies are needed to disentangling its role in the rumen.

However, the two methods (Kraken and Mothur) generated microbial classification at different taxonomic levels for rumen bacteria. To completely understand the function of the rumen microbiota, it is essential to identify organisms at the species level since different species, within the same genus, can have varied functions and niches. The Mothur based method was useful to identify a diverse bacterial microbiota from the RNA-seq datasets, but it was not able to classify any of the bacterial sequences further than the genus level (Tables 3.1 and 3.2). Microbial classification to the species level is a major challenge for clustering-

first approaches based on targeted regional 16S rRNA when short (up to 250 bp) or even longer reads generated from total RNA-seq are used to identify environmental microbes (Xiang et al., 2017). Most existing tools (for bacteria and archaea) lack solid probabilistic- based criteria to evaluate the accuracy of taxonomic assignments to determine the best- matched database hits to distinguish multiple species from the targeted sequence region of the 16S rRNA gene (Xiang et al., 2017). To identify bacteria at the species level, sequencing of full length of 16S rRNA is desired and thus future studies need to increase the sequence length to enhance the resolution for microbial identification. For the Kraken based approach, the reference database was built based on all known microbial genomes and as a result it generated a higher resolution (to the species level) of the rumen microbiota, enabling the program to annotate each microbial sequence to the LCAs (Wood and Salzberg, 2014). In this process, k-mer paths formed by Kraken assign a specific weight to each node (equal to the number of sequences associated with the node's taxon) while increasing the sensitivity of the species classification even if regions (for example, V3-V5) of the 16S rRNA gene were analyzed (Wood and Salzberg, 2014; Valenzuela-González et al., 2016). Consequently, the generation of chimeric trees using short or long input sequences is improbable with Kraken as unlike other programs (such as Ribosomal Database Project classifier and Mothur), it leaves out specific sequences if there is insufficient evidence for classification and they are designated as unclassified (Valenzuela-González et al., 2016). Therefore, inputting short or long environmental sequences (containing most of the 16S hypervariable regions or mRNA sequences) into Kraken may generate a more representative profile of complex microbiomes (Valenzuela-González et al., 2016) like the rumen. However, the lack of reference genomes for rumen microorganisms also limits Kraken. For example, the classification of Xenorhabdus (Table 2) and Xenorhabdus doucetiae (data not shown; relative abundance (%): H-FCR, 0.1 ± 0.10 found in 6 samples; L-FCR, 0%), a motile, gram-negative soil bacterium usually described as being part of entomopathogenic nematode/bacterium symbiotic complex (Furgani et al., 2008) has not been previously reported in amplicon based sequencing (Li et al, 2016a) or metagenomic/metatranscriptome sequencing (Li and Guan, 2017) of rumen contents. The classification of this bacterial species may indicate that Kraken did not properly identify the microbe since the reference genome information was built mostly from all microbial genomes annotated in the NCBI database. However, these organisms may have been actually detected in the rumen since cattle can consume soil, raising the possibility that their detection was transitory. 95

It is noteworthy that Methanobrevibacter (family Methanobacteriaceae) was identified in both databases (Supplementary Tables 1, 2, and 4). This genus has been reported to be the most abundant archaeal population in the rumen based on DNA datasets (Kittelmann et al., 2013; Henderson et al., 2015), but it had a lower abundance than Methanomassiliicoccaceae at the RNA level in this study. This result is consistent with the research conducted by Li et al. (2016a), who reported a predominance of Methanomassiliicoccaceae over Methanobrevibacter in RNA-based datasets when compared to DNA Amplicon-seq outcomes, suggesting that Methanomassiliicoccaceae may be more active in the rumen than Methanobacteriaceae. However, further studies are needed to determine whether the differences in abundance between those two archaeal populations have a methodological influence or are controlled by diet, host animal or management strategies. Unlike bacterial classification, Kraken and Mothur generated contrasting results on archaea identification (Table 3.3), which reflects the divergent taxonomic profiles at the species level. For example, certain archaeal genomes, such as Methanobrevibacter wolinii and Methanobrevibacter woesei, were only found in the rumen-specific archaea database, as the Kraken standard database lacked these complete genomes. However, Kraken was able to detect Candidatus Methanoplasma termitum and Candidatus Methanomethylophilus alvus, which were not identified by Mothur pipeline. Li et al. (2016b) isolated the archaeon ISO4-H5 (member of the order Methanomassiliicoccales) from the sheep rumen and discovered that this archaeal taxon exhibited genome size (1.9 Mb) and GC content (54%) similar to Candidatus Methanoplasma termitum (enriched from the termite gut) and Candidatus Methanomethylophilus alvus (enriched from human feces). These two species encode pathways required for hydrogen-dependent methylotrophic methanogenesis by reduction of methyl substrates, without the ability to oxidize methyl substrates to carbon dioxide (Li et al., 2016b). Thus, it is possible that these microbes reside in the rumen. Future analysis with archaeon ISO4-H5 sequences included in the databases of both pipelines as well as its isolation, culture and characterization may provide further evidence of this possibility.

To further verify how these two methods affected data interpretation, the rumen microbiota of H-FCR and L-FCR bulls were compared based on the taxonomic outcomes generated by the two software packages. Both computational pipelines revealed differences in microbial abundance between H- and L-FCR groups at all taxonomic ranks, with Mothur

exclusively identifying a higher abundance of poorly characterized bacterial phylotypes (e.g., R4-41B) in L-FCR bulls (Table 3.2). It has been reported that the abundance of R4-41B was negatively correlated with production traits over the first 12 weeks postpartum in dairy cows (Lima et al., 2015), suggesting that it may have undesirable impacts on the function of the rumen microbiome of L-FCR cattle. Although Kraken identified a relatively higher abundance of Xenorhabdus in H-FCR bulls (Table 2), this result could be erroneous with further validation needed as described above. However, researchers have enumerated and identified a high number (15.7 x 104 Most Probable Number/g) of chlortetracycline resistant Enterobacteriacea in cattle feces that largely consisted of Xenorhabdus doucetiae (Watanabe et al., 2016). Since antimicrobial agents (e. g., chlortetracycline) are typically administered subtherapeutically to beef cattle (Inglis et al., 2005), our results suggest that H-FCR animals may be more susceptible to harbor chlortetracycline resistant bacteria than L-FCR animals in the event of a therapeutic administration of this antibiotic. Further investigations aiming to evaluate the effects of antimicrobial agents (e. g., chlortetracycline) on the development of antimicrobial resistance in Xenorhabdus recovered from less efficient cattle (H-FCR) are warranted. Kraken also detected a higher (P = 0.09) abundance of Methanococcaceae (relative abundance (%): H-FCR, 13.6 ± 8.96; L-FCR, 4.1 ± 4.85) in the rumen of H-FCR bulls, indicating that Methanococcaceae may play a potential role in the linkages between methanogenesis and reduced feed efficiency in cattle. Although RNA-targeted DNA probes and genomic DNA sequencing have revealed a significant population of this archaeal family residing in the rumen (Janssen and Kirs, 2008) and exhibiting a positive correlation with increased forage content in the diet (Pitta et al., 2016), members of this methanogenic archaea family still need to be cultured from the rumen to test our findings.

Finally, our study demonstrated that both pipelines (Mothur and Kraken) were effective in detecting a lower bacterial diversity in efficient (L-FCR) cattle (Table 3.4), corroborating the recent findings by Li and Guan et al. (2017) and (Shabat et al., 2016) that the rumen microbiota of efficient cattle is less complex and more specialized in harvesting energy from the diet through simpler metabolic networks (e.g., acrylate pathway) than inefficient cattle. However, only Kraken identified a significantly lower diversity in the archaeal community in L- FCR bulls, but this result should be carefully interpreted as many archaea phylotypes classified by Kraken are environmental organisms that have not yet been described in the rumen. For example, the methane-producing archaeon Methanothermococcus okinawensis

(the third-most abundant archaea taxon classified by Kraken, Supplementary Table 2) was first isolated from a deep-sea hydrothermal vent system (Takai et al., 2002), Picrophilus torridus and Acidilobus saccharovorans (the fourth and fifth-most abundant archaea taxa detected by Kraken, Supplementary Table 2) were isolated from a dry solfataric field (Fütterer et al., 2004) and a terrestrial acidic hot spring (Mardanov et al., 2010), respectively. Thus, it is worth mentioning that, in spite of the Kraken’s promising results, this pipeline is severely limited when studying a microbiome that is not well described in its standard database (like the rumen), indicating that Mothur (using a specific archaea database described by Li et al., 2016a) could be more suited for identifying archaeal taxonomic profiles.

3.5 Conclusion

The current study was the first to compare the molecular-phylogenetic outcomes of Mothur and Kraken using transcriptomic sequence data (~140 bp in length) of rumen samples. The Kraken pipeline has been adapted to include reference genomes for rumen specific organisms, which has led to the identification of rumen bacteria at species level and more bacterial phylotypes. However, the results of the archaeal classification as well as some of the bacterial species identified by Kraken should be carefully interpreted as many detected phylotypes have not yet been described in the rumen, highlighting the importance of strengthening the Kraken database through the inclusion of more genomes annotated by single cell sequencing of rumen cultures/isolates to enable a more accurate classification. As to the future directions, new sequenced genomes (410 draft bacterial and archaeal genomes) by Hungate1000 project (JGI database) will be included in the Kraken standard database and the recently developed Rumen and Intestinal Methanogen Database will be used for archaea classification (Seedorf et al., 2014), with the goal of improving the accuracy of the results. It is also proposed the configuration of a joint pipeline using both Kraken and Mothur simultaneously to improve the resolution of taxonomic profiling of the rumen microbiome. This joint pipeline will produce a final rumen microbial profile obtained from the combination of multiple results generated from different bioinformatics tools as outlined by Piro et al. (2017), who published a computational method called MetaMeta that executes and integrates results from six metagenomic analysis tools (CLARK - Ounit et al., 2015; DUDes - Piro et al., 2016; GOTTCHA - Freitas et al., 2015; KRAKEN - Wood and Salzberg, 2014; KAIJU - Menzel et al., 2016; and mOTUs - Sunagawa et al., 2013). If the rumen microbiome datasets are strengthened to the same level as the human databases, the joint pipeline will generate more

sensitive and reliable results than those of the best single profile (generated separately by each tool) (Piro et al. 2017). It is believed that a joint pipeline supported by a collection of tools will be useful to control sources of variation present in any metagenomics/metatranscriptomic analysis (e.g., analytical pipelines, related databases and software parameters), which will ultimately lead to standardized results and more reliable biological interpretations. In addition, although Kraken has improved the taxonomic assessment at species level, the high number of unclassified sequences (65%) suggests a need for identifying the rumen microbes with a more resolved taxonomy assignment. Regardless of the approach undertaken, the only way for improvement is through a continued strengthening of the databases by including additional information of whole genome sequencing of rumen isolates as well as single cell sequencing of unculturable rumen microbes, as the ability to culture rumen microorganisms is still limited.

3.6 Literature cited

Aitchison, J. (1982). The statistical analysis of compositional data. Journal of the Royal Statistical Society Series B-Methodological 44(2), 139-177. Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate - a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B-Methodological 57(1), 289-300. Béra-Maillet, C., Mosoni, P., Kwasiborski, A., Suau, F., Ribot, Y., and Forano, E. (2009). Development of a RT-qPCR method for the quantification of Fibrobacter succinogenes S85 glycoside hydrolase transcripts in the rumen content of gnotobiotic and conventional sheep. Journal Of Microbiological Methods 77(1), 8-16. doi: 10.1016/j.mimet.2008.11.009. Bolger, A.M., Lohse, M., and Usadel, B. (2014). Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30(15), 2114-2120. doi: 10.1093/bioinformatics/btu170. Cao, K.A.L., Costello, M.E., Lakis, V.A., Bartolo, F., Chua, X.Y., Brazeilles, R., et al. (2016). MixMC: A Multivariate Statistical Framework to Gain Insight into Microbial Communities. Plos One 11(8). doi: 10.1371/journal.pone.0160169. Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., et al. (2010). QIIME allows analysis of high-throughput community sequencing data. Nature Methods 7(5), 335-336. doi: 10.1038/nmeth.f.303. Chen, J., and Li, H.Z. (2013). Variable selection for sparse dirichlet-multinomial regression with an application to microbiome data analysis. Annals of Applied Statistics 7(1), 418- 442. doi: 10.1214/12-aoas592. Chomczynski, P., and Sacchi, N. (2006). The single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction: twenty-something years on. Nature Protocols (2), 581.

Christopherson, M.R., Dawson, J.A., Stevenson, D.M., Cunningham, A.C., Bramhacharya, S., Weimer, P.J., et al. (2014). Unique aspects of fiber degradation by the ruminal

ethanologen Ruminococcus albus 7 revealed by physiological and transcriptomic analysis. Bmc Genomics 15. doi: 10.1186/1471-2164-15-1066. Cruz-Martinez, K., Suttle, K.B., Brodie, E.L., Power, M.E., Andersen, G.L., and Banfield, J.F. (2009). Despite strong seasonal responses, soil microbial consortia are more resilient to long-term changes in rainfall than overlying grassland. Isme Journal 3(6), 738-744. doi: 10.1038/ismej.2009.16. DeSantis, T.Z., Hugenholtz, P., Larsen, N., Rojas, M., Brodie, E.L., Keller, K., et al. (2006). Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Applied and Environmental Microbiology 72(7), 5069-5072. doi: 10.1128/aem.03006-05. Di Rienzi, S.C., Sharon, I., Wrighton, K.C., Koren, O., Hug, L.A., Thomas, B.C., et al. (2013). The human gut and groundwater harbor non-photosynthetic bacteria belonging to a new candidate phylum sibling to Cyanobacteria. Elife 2. doi: 10.7554/eLife.01102. Duffield, T., Plaizier, J.C., Fairfield, A., Bagg, R., Vessie, G., Dick, P., et al. (2004). Comparison of techniques for measurement of rumen pH in lactating dairy cows. Journal of Dairy Science 87(1), 59-66. Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., and Knight, R. (2011). UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 27(16), 2194-2200. doi: 10.1093/bioinformatics/btr381. Freitas, T.A.K., Li, P.-E., Scholz, M.B., and Chain, P.S.G. (2015). Accurate read-based metagenome characterization using a hierarchical suite of unique signatures. Nucleic Acids Research 43(10), e69-e69. doi: 10.1093/nar/gkv180. Furgani, G., Böszörményi, E., Fodor, A., Máthé-Fodor, A., Forst, S., Hogan, J.S., et al. (2008). Xenorhabdus antibiotics: a comparative analysis and potential utility for controlling mastitis caused by bacteria. Journal of Applied Microbiology 104(3), 745-758. doi: 10.1111/j.1365-2672.2007.03613.x. Fütterer, O., Angelov, A., Liesegang, H., Gottschalk, G., Schleper, C., Schepers, B., et al. (2004). Genome sequence of Picrophilus torridus and its implications for life around pH 0. Proceedings Of The National Academy Of Sciences Of The United States Of America 101(24), 9091-9096. Gharechahi, J., Zahiri, H.S., Noghabi, K.A., and Salekdeh, G.H. (2015). In-depth diversity analysis of the bacterial community resident in the camel rumen. Systematic and Applied Microbiology 38(1), 67-76. doi: 10.1016/j.syapm.2014.09.004. Gonzalez, I., Dejean, S., Martin, P.G.P., and Baccini, A. (2008). CCA: An R package to extend canonical correlation analysis. Journal of Statistical Software 23(12), 1-14. Henderson, G., Cox, F., Ganesh, S., Jonker, A., Young, W., Janssen, P.H., et al. (2016). Rumen microbial community composition varies with diet and host, but a core microbiome is found across a wide geographical range. Scientific Reports 6. doi: 10.1038/srep19175. Inglis, G.D., McAllister, T.A., Busz, H.W., Yanke, L.J., Morck, D.W., Olson, M.E., et al. (2005). Effects of subtherapeutic administration of antimicrobial agents to beef cattle on the prevalence of antimicrobial resistance in Campylobacter jejuni and Campylobacter hyointestinalis. Applied and Environmental Microbiology 71(7), 3872-3881. doi: 10.1128/aem.71.7.3872-3881.2005. Janssen, P.H., and Kirs, M. (2008). Structure of the archaeal community of the rumen. Applied & Environmental Microbiology 74(12), 3619-3625. doi: 10.1128/AEM.02812-07.

100

Jost L. (2007). Partitioning diversity into independent alpha and beta components. Ecology 88, 2427–2439. 10.1890/06-1736.1 Kamke, J., Kittelmann, S., Soni, P., Li, Y., Tavendale, M., Ganesh, S., et al. (2016). Rumen metagenome and metatranscriptome analyses of low methane yield sheep reveals a Sharpea-enriched microbiome characterised by lactic acid formation and utilisation. Microbiome 4(1), 56-56. Kittelmann, S., Seedorf, H., Walters, W.A., Clemente, J.C., Knight, R., Gordon, J.I., et al. (2013). Simultaneous amplicon sequencing to explore co-occurrence patterns of bacterial, archaeal and eukaryotic microorganisms in rumen microbial communities. Plos One 8(2). doi: 10.1371/journal.pone.0047879.

Kittelmann, S., Pinares-Patino, C.S., Seedorf, H., Kirk, M.R., Ganesh, S., McEwan, J.C., et al. (2014). Two different bacterial community types are linked with the low-methane emission trait in sheep. PLoS ONE 9(7), 1-9. doi: 10.1371/journal.pone.0103171. Kopylova, E., Noé, L., and Touzet, H. (2012). SortMeRNA: fast and accurate filtering of ribosomal RNAs in metatranscriptomic data. Bioinformatics (Oxford, England) 28(24), 3211-3217. doi: 10.1093/bioinformatics/bts611. Korpela K., Salonen A., Virta L. J., Kekkonen R. A., Forslund K., Bork P., et al. (2016). Intestinal microbiome is related to lifetime antibiotic use in Finnish pre-school children. Nature Communication 7:10410. 10.1038/ncomms10410. Kozich, J.J., Westcott, S.L., Baxter, N.T., Highlander, S.K., and 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. La Reau, A.J., Meier-Kolthoff, J.P., and Suen, G. (2016). Sequence-based analysis of the genus Ruminococcus resolves its phylogeny and reveals strong host association. Microbial Genomics 2(12), e000099-e000099. doi: 10.1099/mgen.0.000099. Li, D.H., Liu, C.M., Luo, R.B., Sadakane, K., and Lam, T.W. (2015). MEGAHIT: an ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinformatics 31(10), 1674-1676. doi: 10.1093/bioinformatics/btv033. Li, F., and Guan, L.L. (2017). Metatranscriptomic profiling reveals linkages between the active rumen microbiome and feed efficiency in beef cattle. Applied And Environmental Microbiology. doi: 10.1128/AEM.00061-17. Li, F., Henderson, G., Sun, X., Cox, F., Janssen, P.H., and Guan, L.L. (2016a). Taxonomic Assessment of Rumen Microbiota Using Total RNA and Targeted Amplicon Sequencing Approaches. Frontiers in Microbiology 7(987). doi: 10.3389/fmicb.2016.00987. Li, Y., Leahy, S.C., Jeyanathan, J., Henderson, G., Cox, F., Altermann, E., et al. (2016b). The complete genome sequence of the methanogenic archaeon ISO4-H5 provides insights into the methylotrophic lifestyle of a ruminal representative of the Methanomassiliicoccales. Standards in Genomic Sciences 11. doi: 10.1186/s40793- 016-0183-5. Lima, F.S., Oikonomou, G., Lima, S.F., Bicalho, M.L.S., Ganda, E.K., Filho, J.C.d.O., et al. (2015). Prepartum and postpartum rumen fluid microbiomes: characterization and correlation with production traits in dairy cows. Applied And Environmental Microbiology 81(4), 1327-1337.

101

Lindgreen, S., Adair, K.L., and Gardner, P.P. (2016). An evaluation of the accuracy and speed of metagenome analysis tools. Scientific Reports 6, 19233-19233. doi: 10.1038/srep19233. Mandal, S., Van Treuren, W., White, R.A., Eggesbo, M., Knight, R., and Peddada, S.D. (2015). Analysis of composition of microbiomes: a novel method for studying microbial composition. Microbial Ecology in Health and Disease 26, 27663-Article No.: 27663. doi: 10.3402/mehd.v26.27663. Mardanov, A.V., Svetlitchnyi, V.A., Beletsky, A.V., Prokofeva, M.I., Bonch-Osmolovskaya, E.A., Ravin, N.V., et al. (2010). The genome sequence of the crenarchaeon Acidilobus saccharovorans supports a new order, Acidilobales, and suggests an important ecological role in terrestrial acidic hot springs. Applied And Environmental Microbiology 76(16), 5652-5657. doi: 10.1128/AEM.00599-10. McCann, J.C., Elolimy, A.A., and Loor, J.J. (2017). Rumen Microbiome, Probiotics, and Fermentation Additives. The Veterinary Clinics Of North America: Food Animal Practice. doi: 10.1016/j.cvfa.2017.06.009. Menzel, P., Ng, K.L., and Krogh, A. (2016). Fast and sensitive taxonomic classification for metagenomics with Kaiju. Nature Communications 7, 11257-11257. doi: 10.1038/ncomms11257. Meyer, F., Paarmann, D., D'Souza, M., Olson, R., Glass, E.M., Kubal, M., et al. (2008). The metagenomics RAST server - a public resource for the automatic phylogenetic and functional analysis of metagenomes. BMC Bioinformatics 9. doi: 10.1186/1471-2105- 9-386. Montanholi, Y.R., Swanson, K.C., Palme, R., Schenkel, F.S., McBride, B.W., Lu, D., et al. (2010). Assessing feed efficiency in beef steers through feeding behavior, infrared thermography and glucocorticoids. Animal 4(5), 692-701. doi: 10.1017/s1751731109991522. Marounek, M., and Duskova, D. (1999). Metabolism of pectin in rumen bacteria Butyrivibrio fibrisolvens and Prevotella ruminicola. Letters in applied microbiology 29(6), 429-433. Nandi, R., and Sengupta, S. (1998). Microbial production of hydrogen: An overview. Critical Reviews in Microbiology 24(1), 61-84. doi: 10.1080/10408419891294181. Newbold, C.J., Wallace, R.J., and McIntosh, F.M. (1996). Mode of action of the yeast Saccharomyces cerevisiae as a feed additive for ruminants. British Journal of Nutrition 76(2), 249-261. doi: 10.1079/bjn19960029. Olfert E. D., Cross B. M., McWilliams A. A. (1993). Guide to the Care and Use of Experimental Steers. Ottawa, ON: Canadian Council on Animal Care. Ounit, R., Wanamaker, S., Close, T.J., and Lonardi, S. (2015). CLARK: fast and accurate classification of metagenomic and genomic sequences using discriminative k-mers. BMC Genomics 16(1), 1-13. doi: 10.1186/s12864-015-1419-2. Piro, V.C., Lindner, M.S., and Renard, B.Y. (2016). DUDes: a top-down taxonomic profiler for metagenomics. Bioinformatics (Oxford, England) 32(15), 2272-2280. doi: 10.1093/bioinformatics/btw150. Piro, V.C., Matschkowski, M., and Renard, B.Y. (2017). MetaMeta: integrating metagenome analysis tools to improve taxonomic profiling. Microbiome 5(1), 101-101. doi: 10.1186/s40168-017-0318-y.

102

Pitta, D.W., Indugu, N., Kumar, S., Vecchiarelli, B., Sinha, R., Baker, L.D., et al. (2016). Metagenomic assessment of the functional potential of the rumen microbiome in Holstein dairy cows. Anaerobe 38, 50-60. doi: 10.1016/j.anaerobe.2015.12.003. Qi, M., Wang, P., O'Toole, N., Barboza, P.S., Ungerfeld, E., Leigh, M.B., et al. (2011). Snapshot of the eukaryotic gene expression in muskoxen rumen-a metatranscriptomic approach. Plos One 6(5). doi: 10.1371/journal.pone.0020521. Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., et al. (2013). The SILVA ribosomal RNA gene database project: improved data processing and web- based tools. Nucleic Acids Research 41(Database issue), D590-D596. doi: 10.1093/nar/gks1219. R Core Team (2016). R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing. Roehe, R., Dewhurst, R.J., Duthie, C. A., Rooke, J.A., McKain, N., Ross, D.W., et al. (2016). Bovine Host Genetic Variation Influences Rumen Microbial Methane Production with Best Selection Criterion for Low Methane Emitting and Efficiently Feed Converting Hosts Based on Metagenomic Gene Abundance. Plos Genetics 12(2). doi: 10.1371/journal.pgen.1005846. Russell, J.B., and Rychlik, J.L. (2001). Factors that alter rumen microbial ecology. Science 292(5519), 1119-1122. doi: 10.1126/science.1058830. Scharen, M., Drong, C., Kiri, K., Riede, S., Gardener, M., Meyer, U., et al. (2017). Differential effects of monensin and a blend of essential oils on rumen microbiota composition of transition dairy cows. Journal of Dairy Science 100(4), 2765-2783. doi: 10.3168/jds.2016-11994. Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., et al. (2009). Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Applied & Environmental Microbiology 75(23), 7537-7541. doi: 10.1128/AEM.01541-09. Seedorf, H., Kittelmann, S., Henderson, G., and Janssen, P.H. (2014). RIM-DB: a taxonomic framework for community structure analysis of methanogenic archaea from the rumen and other intestinal environments. Peerj 2. doi: 10.7717/peerj.494. Segata, N., Waldron, L., Ballarini, A., Narasimhan, V., Jousson, O., and Huttenhower, C. (2012). Metagenomic microbial community profiling using unique clade-specific marker genes. Nature Methods 9(8), 811-814. doi: 10.1038/nmeth.2066. Shabat, S.K.B., Sasson, G., Doron-Faigenboim, A., Durman, T., Yaacoby, S., Berg Miller, M.E., et al. (2016). Specific microbiome-dependent mechanisms underlie the energy harvest efficiency of ruminants. ISME J. doi: 10.1038/ismej.2016.62. Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., et al. (2003). Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Research 13(11), 2498-2504. doi: 10.1101/gr.1239303. Shi, W., Kang, D., Froula, J., Fan, C., Deutsch, S., Chen, F., et al. (2014). Methane yield phenotypes linked to differential gene expression in the sheep rumen microbiome. Genome Research 24(9), 1517-1525. doi: 10.1101/gr.168245.113. Siegwald, L., Touzet, H., Lemoine, Y., Hot, D., Audebert, C., and Caboche, S. (2017). Assessment of Common and Emerging Bioinformatics pipelines for targeted metagenomics. PLoS One 12(1), 1-26. doi: 10.1371/journal.pone.0169563.

103

Soo, R.M., Skennerton, C.T., Sekiguchi, Y., Imelfort, M., Paech, S.J., Dennis, P.G., et al. (2014). An Expanded Genomic Representation of the Phylum Cyanobacteria. Genome Biology and Evolution 6(5), 1031-1045. doi: 10.1093/gbe/evu073. Simon, C., and Daniel, R. (2011). Metagenomic Analyses: Past and Future Trends. Applied and environmental microbiology AEM 77(4), 1153-1161. Suen, G., Weimer, P.J., Stevenson, D.M., Aylward, F.O., Boyum, J., Deneke, J., et al. (2011). The complete genome sequence of fibrobacter succinogenes s85 reveals a cellulolytic and metabolic specialist. PLoS One 6. doi: e0018814. Sunagawa, S., Mende, D.R., Zeller, G., Izquierdo-Carrasco, F., Berger, S.A., Kultima, J.R., et al. (2013). Metagenomic species profiling using universal phylogenetic marker genes. Nature Methods 10(12), 1196-1199. doi: 10.1038/nmeth.2693. Takai, K., Inoue, A., and Horikoshi, K. (2002). Methanothermococcus okinawensis sp. nov., a thermophilic, methane-producing archaeon isolated from a Western Pacific deep- sea hydrothermal vent system. International Journal Of Systematic And Evolutionary Microbiology 52(Pt 4), 1089-1095. Thompson, S. (2015). The effect of diet type on residual feed intake and the use of infrared thermography as a method to predict efficiency in beef bulls. Master’s thesis. Winnipeg, MB: University of Manitoba. Valenzuela-González, F., Martínez-Porchas, M., Villalpando-Canchola, E., and Vargas- Albores, F. (2016). Studying long 16S rDNA sequences with ultrafast-metagenomic sequence classification using exact alignments (Kraken). Journal of Microbiological Methods 122, 38-42. doi: 10.1016/j.mimet.2016.01.011. Wang, Q., Garrity, G.M., Tiedje, J.M., and 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: 10.1128/aem.00062-07. Watanabe, K., Horinishi, N., Matsumoto, K., Tanaka, A., and Yakushido, K. (2016). A New Evaluation Method for Antibiotic-Resistant Bacterial Groups in Environment. Advances in Microbiology 6(3), 133-151. doi: 10.4236/aim.2016.63014. Wegmann, U., Louis, P., Goesmann, A., Henrissat, B., Duncan, S.H., and Flint, H.J. (2014). Complete genome of a new Firmicutes species belonging to the dominant human colonic microbiota ('Ruminococcus bicirculans') reveals two chromosomes and a selective capacity to utilize plant glucans. Environmental Microbiology 16(9), 2879- 2890. doi: 10.1111/1462-2920.12217. Weiss, S., Xu, Z.Z., Peddada, S., Amir, A., Bittinger, K., Gonzalez, A., et al. (2017). Normalization and microbial differential abundance strategies depend upon data characteristics. Microbiome 5. doi: 10.1186/s40168-017-0237. Williams, M.M., Domingo, J.W.S., Meckes, M.C., Kelty, C.A., and Rochon, H.S. (2004). Phylogenetic diversity of drinking water bacteria in a distribution system simulator. Journal of Applied Microbiology 96(5), 954-964. doi: 10.1111/j.1365- 2672.2004.02229.x. Wood, D.E., and Salzberg, S.L. (2014). Kraken: ultrafast metagenomic sequence classification using exact alignments. Genome Biology 15(3), R46-R46. doi: 10.1186/gb-2014-15-3-r46. Xiang, G., Huaiying, L., Kashi, R., and Qunfeng, D. (2017). A Bayesian taxonomic classification method for 16S rRNA gene sequences with improved species-level accuracy. BMC Bioinformatics 18, 1-10. doi: 10.1186/s12859-017-1670-4.

104

3.7 Tables and figures Table 3.1 Quantification of taxonomic phylotypes identified by each method Mothur1 Kraken2 Commonly Detected Phylotypes Phylotypes Classified (N°) Unclassified (N°) Classified (N°) (N°) Bacteria Phyla 23 1 26 16 Families 121 66 204 78 Genera 189 135 348 69 Species - - 423 0 Archaea Phyla 1 1 2 1 Families 3 3 7 2 Genera 4 5 8 1 Species 5 6 8 1 1Pipeline to assess the rumen microbiota developed by Li et al. (2016) based on Mothur (Schloss et al., 2009). Clustering-first approaches (such as Mothur) allow the discrimination of unclassified reads (Siegwald et al., 2017). 2Metagenomic sequence classification method developed by Wood and Salzberg (2014). Unlike clustering-first approaches, assignment-first tools (such as Kraken) do not allow the discrimination of unclassified reads (Siegwald et al., 2017).

105

Table 3.2 Differentially abundant bacteria in efficient (low FCR) and inefficient (high FCR) cattle according to the two classification methods1,2,3

Phylotypes Mothur Kraken

High (%) Low (%) High (%) Low (%) Phyla Bacteroidetes 35.0 ± 8.52 43.2 ± 6.98 37.7 ± 9.75 46.1 ± 11.35 Firmicutes 24.8 ± 9.53 19.2 ± 6.71 28.1 ± 7.01 22.6 ± 6.98 Proteobacteria 20.0 ± 4.88 14.6 ± 3.12 15.7 ± 2.59 12.8 ± 2.63 Fibrobacteres 2.50 ± 0.80 4.4 ± 1.28 1.1 ± 0.42 1.3 ± 0.31 Spirochaetes 2.0 ± 0.50 2.5 ± 0.33 4.6 ± 0.60 5.1 ± 1.02 Verrucomicrobia 1.4 ± 0.19 2.1 ± 0.82 2.1 ± 0.42 2.4 ± 0.62

Families Prevotellaceae 18.4 ± 6.73 23.7 ± 6.17 26.9 ± 35.3 ± 13.63 10.48 Ruminococcacea 10.2 ± 4.38 7.73 ± 3.89 8.4 ± 3.01 5.7 ± 3.33 Lachnopiraceae 7.1 ± 3.62 5.5 ± 2.23 7.1 ± 1.65 6.1 ± 1.95 Fibrobacteriacea 2.6 ± 0.83 4.6 ± 1.34 1.8 ± 0.37 1.67 ±0.51 Spirochaetaceae 1.8 ± 0.56 2.3 ± 0.34 4.8 ±0.73 5.3 ±1.21 R4 – 41B 0.02 ± 0.037a 0.13 ± 0.118b - - Actinomycetaceae - - 0.1± 0.03 b 0.2 ±0.04 a Genera Prevotella 20.0 ± 8.28 25.2 ± 7.10 28.8 ± 37.5 ± 14.15 10.75 Ruminococcus 5.9 ± 2.88 4.4 ± 2.56 3.9 ± 1.70 2.7 ± 1.83 Fibrobacter 3.2 ± 1.12 5.5 ± 1.56 1.4 ± 0.56 1.6 ± 0.34 Butyrivibrio 1.0 ± 0.19 1.3 ± 0.68 2.2 ± 0.29 2.5 ± 1.31 Xenorhabdus - - 0.29 ± 0.04 ± 0.036b 0.246a

106

Species Prevotella - - 23.0 ± 9.99 31.6 ± 13.15 ruminicola Butyrivibrio - - 2.6 ± 0.33 2.9 ± 1.44 proteoclasticus Ruminiclostridium - - 2.8 ± 0.98 1.6 ± 1.14 sp KB18 Fibrobacter - - 1.6 ± 0.66 1.8 ± 0.37 succinogens Ruminococcus - - 1.7 ± 0.89 1.0 ± 0.77 albus

1Statistical comparisons were obtained by the application of ANCOM (Mandal et al., 2015) on taxa counts determined by Mothur (OUTs) or Kraken (K-mers), and thus estimators are comparable only between High (n=6) and Low (n=6) FCR cattle as provided by each classification method. 2P values were obtained using Wilcoxon exact test (calculated on the log-ratio matrix; Mandal et al., 2015), and then adjusted to FDR using Benjamini-Hochberg algorithm (Benjamini and Hochberg, 1995). A threshold of FDR < 0.15 was applied to determine significance. Within a row, means with different superscript are statistically different between High and Low FCR cattle for each method (separately). 3Blank spaces indicate that the phylotypes were not detected in the dataset either by Mothur or Kraken.

107

Table 3.3 Differentially abundant archaea in efficient (low FCR) and inefficient (high FCR) cattle according to the two classification methods1,2,3

Phylotypes Mothur Kraken

High (%) Low (%) High (%) Low (%) Families RCC and relatives 73.2 ± 71.9 ± - - 3.77 13.12 Methanomassiliicoccaceae - - 65.5 ± 9.92 67.1 ± 11.29 Methanococcaceae - - 13.6 ± 8.96a 4.1 ± 4.85b Methanobacteriaceae 23.9 ± 24.8 ± 6.0 ± 6.42 7.0 ± 7.52 4.19 13.75 Methanosarcinaceae 0.3 ± 0.35 0.6 ± 0.72 5.6 ± 8.51 10.7 ± 5.32 Genera Candidatus - - 49.0 ± 13.61 55.4 ± 9.51 Methanoplasma Candidatus - - 19.0 ± 7.92 12.8 ± 6.78 Methanomethylophilus Methanosarcina - - 5.1 ± 9.20 11.0 ± 5.81 Methanobrevibacter 21.8 ± 21.8 ± 4.7 ± 5.01 5.3 ± 5.35 3.85 10.53 Methanosphaera 0.8 ± 0.47 1.2 ± 1.57 - - Methanimicrococcus 0.3 ± 0.33 0.6 ± 0.70 - - Species Candidatus - - 51.0 ± 14.27 62.8 ± 11.23 Methanoplasma termitum Candidatus - - 19.7 ± 8.28 14.6 ± 8.09 Methanomethylophilus alvus Methanobrevibacter 14.8 ± 14.7 ± 6.11 - - gottschalkii and relatives 3.60 Methanobrevibacter 3.8 ± 1.70 4.0 ± 4.19 2.4 ± 4.05 1.2 ± 3.14 ruminantium

108

Methanobrevibacter wolinii 0.1 ± 0.14 0.2 ± 0.32 - - and relatives Methanobrevibacter 0.1 ± 0.10 0.05 ± 0.06 - - woesei Methanobrevibacter smithii 0.03 ± 0.10 ± 0.14 - - 0.07

1Statistical comparisons were obtained by the application of ANCOM (Mandal et al., 2015) on taxa counts determined by Mothur (OUTs) or Kraken (K-mer), and thus estimators are comparable only between High (n=6) and Low (n=6) FCR cattle as provided by each classification method. 2P values were obtained using Wilcoxon exact test (calculated on the log-ratio matrix; Mandal et al., 2015), and then adjusted to FDR using Benjamini-Hochberg algorithm (Benjamini and Hochberg, 1995). A threshold of FDR < 0.15 was applied to determine significance. Within a row, means with different superscript are statistically different only between High and Low FCR cattle for each method (separately). 3Blank spaces indicate that the phylotypes were not detected in the dataset either by Mothur or Kraken.

109

Table 3.4 Comparison of bacterial and archaeal alpha-diversity indexes between efficient (low FCR) and inefficient (high FCR) cattle according to the two microbial classification methods1

Bacteria Archaea Indexes Mothur Kraken Mothur Kraken High Low High Low High Low High Low Number of 244.1 ± 239.3 ± 241.1 ± 224.5 ± 8.8 ± 1.33 8.8 ± 5.0 ± 4.5 ± observed 23.88 19.98 28.29 28.37 0.75 0.89 1.05 phylotypes Shannon2 2.78 ± 2.73 ± 3.93 ± 3.51 ± 0.90 ± 0.08 0.91 ± 1.27 ± 1.06 ± 0.12a 0.14b 0.42a 0.62b 0.28 0.19a 0.25b Inverse 9.8 ± 8.7±1.82b 12.7 ± 8.8 ± 1.74 ± 0.13 1.85 ± 2.95 ± 2.30 ± Simpson 1.93a 5.75a 5.60b 0.61 0.89a 0.64b Simpson 0.89 ± 0.88 ± 0.90 ± 0.83 ± 0.42 ± 0.05 0.42 ± 0.64 ± 0.54 ± (with 0.03a 0.03b 0.07a 0.11b 0.15 0.04a 0.05b rarefy) 1Within a row, means with different superscript were different at P < 0.05. Comparison was conducted using paired Wilcoxon signed rank test for bacteria (genus level) and archaea (species level) separately for High and Low FCR cattle as provided by each classification method, and thus estimators between bacterial and archaeal groups are comparable only within each method and between High and Low FCR animals. 2Shannon indices showed in the table are the raw values, and the comparison of Shannon indices between High and Low FCR cattle was based on the exponentially transformed values (Jost, 2007) using paired Wilcoxon signed rank test.

1 2

Figure 3.1 Flow chart of the pipelines (Mothur and Kraken) presenting software 3 parameters used to analyze the rumen microbiota. Part of this figure was adapted from 4 the pipeline published by Li et al. (2016a). 5

111

2 3

Figure 3.2 Correlation circle plots and relevance networks generated from the output of 4 regularized canonical correlation (rCC) method (Total Sum Scaling + Centered Log Ratio) 5 applied to rumen bacteria (genera) and archaea (species) classified by Mothur or Kraken. 6 (a) and (b) show the correlation and network plots of the first two rCC components for 7 Mothur. (c) and (d) represent the correlation and network plots of the first two rCC 8 components for Kraken. In the correlation circle plots, bacteria (X) and archaea (Y) are 9 shown inside a circle of radius 1 centered at the origin, with strongly associated (or 10 correlated) variables being projected in the same direction from the origin. The greater 11 the distance from the origin indicates stronger association. Two circumferences of radius 12 1 and 0.5 are plotted to reveal the correlation structure of the variables (Gonzalez et al., 13 2008). In the relevance networks, red and green edges indicate positive and negative 14

112

correlations respectively, and the sizes of the nodes indicate the mean average 1 abundance. Only bacterial genera and archaeal species with a relative abundance > 0.1% 2 detected in all rumen samples were included in the rCC analysis (Li and Guan, 2017). 3

113

Chapter 4 1

Taxonomic and functional assessment reveals the effect of Angus breed genetics 2 on rumen microbial signatures 3

4 4.1 Introduction 5 The rumen is a complex anaerobic ecosystem inhabited by a symbiotic array of bacterial, 6 archaeal, protozoal, and fungal species that work together to supply protein, vitamins and 7 short-chain organic acids to the ruminant host (Russell and Rychlik 2001). Compared to 8 other groups of microbes, bacteria represent the most genetically diverse and abundant 9 organisms of the rumen microbiome (density of 1010-1011 cells/ml rumen fluid), making 10 up more than 50% of the cell mass (Opdahl et al. 2018, Creevey et al. 2014). Many 11 bacteria species (e.g., Prevotella ruminicola, Ruminococcus albus, and Fibrobacter 12 succinogenes) have been isolated and cultured from the rumen microbial community 13 (Russell and Rychlik 2001), and efforts have been made to reveal functional activities 14 of rumen bacteria with the purpose of enhancing feed efficiency of the host (Fuyong Li 15 and Le Luo Guan 2017). The most important functional activity performed by rumen 16 bacteria involves the digestion and metabolism of plant structural carbohydrates to aid 17 in the breakdown of lignocellulose, as the ruminant host does not produce enzymes 18 involved in plant cell wall digestion (Wang and McAllister 2002). Additionally, rumen 19 bacteria have been associated with rumen fermentation parameters (e.g., volatile fatty 20 acids - VFAs, NH3-N, pH) and host performance (dry matter intake, average daily gain, 21 and feed conversion ratio - FCR) (Zhou et al. 2018), suggesting that the identification 22 of host-specific bacteria can contribute to improve our knowledge of microbial signatures 23 associated with feed efficiency, and this may offer opportunities to enhance the 24 efficiency of digestion in the rumen. 25

Long-standing efforts to increase the efficient use of feedstuffs to enhance 26 ruminant growth have moved producers to adopt feed efficiency indicators like feed 27 conversion ratio (FCR, feed/gain) to monitor phenotypic and genetic variations in host 28 productivity (Crews 2005). More recently, a study showed that the host may regulate the 29 rumen microbiota and its function in relation to FCR through microbial genes such 30

114

as TSTA3 (GDP-l-fucose synthetase) and Fucl (l-fucose isomerase), demonstrating that 1 the relative abundance of these genes could serve as a predictor for host feed efficiency 2 (Roehe et al. 2016). Roehe et al. (2016) also highlighted the effect of genetically distant 3 crossbred breeds (Aberdeen Angus and Limousin) on the rumen microbiota and its 4 microbial genes, suggesting that the host genetics plays a significant role in shaping the 5 composition of the rumen microbiota. Studies have revealed that twins exhibited a 6 different fecal microbial signature despite their genetic similarity (Lee et al. 2011) and 7 variations in animal gain to feed ratios were observed in cattle fed and managed under 8 the same environment although they had a similar genetic background (Wolfger et al. 9 2016). Based on these assumptions, this chapter speculated that variations in the rumen 10 microbiota associated with the genetic makeup of the host could exist in genetically 11 similar breeds. 12

Here, dimensionality reduction techniques (Cao et al. 2016) and 13 metatranscriptomics were used to identify the subtle differences in the active rumen 14 bacterial and functional signatures that characterize the closely related breeds of Black 15 and Red Angus cattle. The genomic differences between these two beef breeds are 16 minimal and attributable to genes (e.g., MC1-R – melanocortin 1 receptor) encoding 17 proteins involved in biological functions of pigmentation (coat color) (Wolfger et al. 2016, 18 McLean and Schmutz 2009). However, Black and Red Angus in terms of genetic 19 diversity and population structure are described as separate breeds in North America 20 (Márquez et al. 2010). Thus, comparisons between them with respect to the taxonomic 21 and functional profiles of their bacterial population can ultimately help uncover microbial 22 signatures that differentiate phenotypes of the Angus breed. Therefore, the objectives 23 of this study were to (a) characterize active bacterial and microbial functional signatures 24 discriminating two breeds of beef cattle (Black vs. Red Angus) fed forage-based diets, 25 and (b) identify specific bacterial groups and functions associated with feed efficiency. 26 27 4.2 Materials and methods 28 4.2.1 Animal trial and sampling 29

115

The experimental procedures described here were reviewed and approved by the 1 University of Manitoba animal care committee. Briefly, rumen contents were collected 2 from 6 (Black Angus= 3; Red Angus= 3) purebred bulls (mean age of 249 ± 22 days and 3 average body weight of 313.9 ± 32 kg) raised in confinement at the Glenlea Research 4 Station (University of Manitoba) according to the guidelines of the Canadian Council on 5 Animal Care. In the current trial, which lasted 180 days, bulls were fed a forage based 6 diet (alfalfa hay, 17.9%; corn silage, 81.7%; limestone, 0.2%; salt, 0.1%; mineral, 0.1%) 7 throughout the experimental period. Then, representative samples (250 ml) of rumen 8 contents were collected over four-time points (0, 80, 100, 180 d) (24 samples in total) 9 using a Geishauser oral probe (Geishauser 1993), immediately snap frozen in liquid 10 nitrogen, and stored at -80°C for three months before RNA extraction. The feed intake of 11 individual bulls was recorded using the GrowSafe® feeding system (GrowSafe Systems 12 Ltd., Airdrie, Alberta, CA), and the FCR was calculated biweekly as a ratio of dry matter 13 intake to average daily gain. 14

4.2.2 VFA analysis 15

Supernatants from rumen fluid samples were obtained after centrifugation at 3,000 × g 16 for 15 min at 4˚C and mixed with 25% phosphoric acid (4:1; v/v) for the subsequent gas 17 chromatography (GC) analysis. After adding the internal standard to the samples and 18 incubating them at -20˚C overnight, they were centrifuged at 19,000 × g for 5 min at 4˚C 19 and the supernatant was transferred to the GC vials (1.8 mL). Next, 0.8 mL of the sample 20 was combined with 0.2 mL of 25% phosphoric acid and 0.2 mL of internal standard 21 solution. Standards (for acetic, propionic, isobutyric, butyric, isovaleric, valeric, and 22 caproic acids) were prepared by combining 1 mL of standard solution and 0.2 mL of 23 internal standard solution. The GC analysis was performed using the column Stabilwax- 24 DA 30 meter (Restek Corp), the head pressure of 7.5 psi, split vent flow of 20 mL/minute, 25 and injector temperature of 170ºC. 26

4.2.3 RNA extraction and sequencing 27

Total RNA was extracted from rumen samples using the TRIzol protocol based on the 28 acid guanidinium-phenol-chloroform method with modifications reported previously 29

116

(Fuyong Li and Le Luo Guan 2017). Approximately 200 mg of rumen sample was 1 subjected to RNA extraction with the addition of 1.5 ml of TRIzol reagent (pH: 4.6; 2 Invitrogen, Carlsbad, CA, USA), followed by 0.4 ml of chloroform (pH: 7.0), 0.3 ml of 3 isopropanol (pH: 7.0), and 0.3 ml of high salt solution (pH: 8.0) (1.2 M sodium acetate, 4 0.8 M NaCl) for the extraction protocol. The yield and integrity of the RNA samples were 5 determined using a Qubit 2.0 fluorimeter (Invitrogen, Carlsbad, CA, USA) and Agilent 6 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). RNA samples were 7 subjected to downstream RNA-sequencing only if they exhibited RNA with integrity 8 number (RIN) higher than 7.0. Total RNA (100 ng) of each sample was used for library 9 construction using the TruSeq RNA sample prep v2 LS kit (Illumina, San Diego, CA, USA) 10 without the mRNA enrichment step. The quality of libraries was assessed using Agilent 11 2200 TapeStation (Agilent Technologies) and Qubit 2.0 fluorimeter (Invitrogen). Finally, 12 cDNA fragments (∼140 bp) were paired-end (2 X 100 bp) sequenced using an Illumina 13

HiSeq 2000 system at the McGill University and Génome Québec Innovation Centre 14 (Montréal, QC, Canada). 15

4.2.4 Bioinformatics analysis 16

Taxonomic annotation of metatranscriptomes was obtained through the software Kraken 17 (Wood and Salzberg 2014) as implemented in a pipeline developed by Neves et al. (2017). 18 Metatranscriptome functional annotations were identified using ShotMAP as described by 19 Nayfach et al. (2015). In summary, all fastq-formatted sequences were firstly analyzed 20 through FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) for quality 21 control, and then the software Trimmomatic (version 0.32) (Bolger et al. 2014) was used 22 to trim residual artificial sequences, cut bases with quality scores below 20, and remove 23 reads shorter than 50 bp. After the pre-processing steps described previously, 24 SortMeRNA (version 1.9) (Kopylova et al. 2012) was used to sort the filtered reads into 25 fragments of mRNA for microbial classification. Prior to performing the microbial 26 classification, a Kraken standard database was built based on all complete genomes of 27 bacteria downloaded from NCBI (RefSeq) plus complete genomes from organisms 28 isolated from the rumen or from ruminant feces or saliva deposited in the Hungate1000 29 project (JGI's IMG database) ) (Neves et al. 2017). Then, each pair of mRNA sequences 30

117

was assembled by MEGAHIT (Li et al. 2015), with the resulting contigs being assigned 1 by Kraken (through k-mer discrimination) to the lowest common ancestor in the 2 customized standard database for bacterial classification. Full taxonomic names 3 associated with each classified sequence (kraken option --preload) and standard ranks 4 (from domain to species) for each taxon were provided by the options kraken-translate 5 and kraken-mpa-report (Neves et al. 2017). 6

Finally, the functions (microbial gene families) were obtained by mapping the 7 assembled reads to the KEGG (Kanehisa et al. 2004) Orthology database using ShotMAP 8 (Nayfach et al., 2015), which is an algorithm based on the aligner RAPSearch2. Bit-score 9 cutoffs (option --class-score 200) for matching contigs to the protein families were 10 selected based on the average read length of each sample as described by Nayfach et 11 al. (2015). The KEGG Orthology (KO) database was chosen because it annotates a large 12 number of bacteria, including many species observed in the rumen microbiome, and 13 covers a wide range of gene families, including metabolic enzymes, and signaling 14 proteins. Average genome size (ags) was not estimated in this study (option --ags-method 15 none) because we used mRNA data to annotate the microbial genes (Nayfach et al., 16 2015). 17

4.2.5 Statistical analysis 18

To avoid sparsely observed counts, which tend to introduce noise in the analysis, we 19 considered bacterial profiles and microbial functions with a relative abundance > 0.05% 20 prevalent in at least 50% of the samples (12 out 24). Next, the filtered datasets were 21 normalized by total sum scaling (TSS) (dividing each taxon/function count by the total 22 number of counts in each individual sample to account for uneven sequencing depths 23 across samples) and then transformed by centered log ratio (CLR) to project the data 24 from a simplex to a Euclidean space (Aitchison 1982) as described in the mixMC 25 multivariate statistical framework (Cao et al. 2016) (R package mixOmics) (Rohart et al., 26 2017). Then, sparse partial least square discriminant analysis (sPLS-DA) (Cao et al. 2016) 27 was applied to identify microbial signatures in Black Angus and Red Angus cattle while 28 handling the sparsity of microbiome datasets and the sampling repeatedly performed on 29 the same subjects (0, 80, 100, and 180 days). sPLS-DA includes a multilevel 30

118

decomposition approach that enables the detection of subtle differences in the signatures 1 when high inter-subject variability is present due to sampling performed repeatedly on the 2 same subjects, and the samples collected from four-time points in each breed were 3 combined for sPLS-DA analysis. The ANCOM procedure (Mandal et al., 2015), which 4 also uses log-ratio transformations (Aitchison 1982), was performed to test whether the 5 features of the filtered datasets were associated with FCR. The zero counts were 6 replaced by ANCOM for an arbitrary pseudo count value of 0.001 and the normalized log- 7 ratio of the abundance of each taxon/function was compared to the abundance of all 8 remaining taxa/functions one at a time. Thereafter, ANOVA was calculated on the log- 9 ratio matrix to find differentially abundant bacterial species and microbial functions 10 between Black and Red Angus, adjusted for FCR and the time points. In ANCOM, the p- 11 values were corrected for false discovery rate (FDR) using the Benjamini-Hochberg 12 algorithm, and a threshold of FDR lower than 0.05 was applied to determine statistical 13 significance. Finally, regularized canonical correlation (rCC) analysis (Gonzalez et al. 14 2008) was used to investigate the relationship between the VFA and bacterial species. 15 Before running rCC analysis, the microbial data were normalized by TSS and then 16 transformed by CLR, as described previously. The estimation of regularization 17 parameters (λ1 and λ2) and canonical correlations were calculated using cross-validation 18 procedures (Gonzalez et al. 2008). All statistical procedures and figures were done in R 19 3.4.2 (R Core Team, 2017) and Python 3.6.0. 20

4.2.6 Data repository resources 21 The datasets analyzed in this study were submitted to NCBI Sequence Read Archive 22 (SRA) under the accession number PRJNA496209. 23

4.3 Results 24 4.3.1 Overview of the classification of active bacterial taxa 25

Assembly of mRNA reads resulting from total RNA sequencing of rumen samples 26 generated a total of 7,330 contigs (with an average extension of 473.31 ± 30.34 bp and 27 N50 of 434 ± 31.04 bp), which were further classified using a taxonomic assignment 28 approach developed by our group. Approximately 53% of the mRNA reads were mapped 29 to the contigs, indicating that they represented a significant proportion of the examined 30

119

rumen metatranscriptome. From the total number of phyla (n = 19) identified in the rumen 1 of both breeds, Bacteroidetes (46.3%), Firmicutes (22.7%), Proteobacteria (14%), 2 Spirochaetes (5%), Verrucomicrobia (2.3%), Tenericutes (2.2%), Actinobacteria (2%), 3 and Fibrobacteres (1.4%) were the most abundant taxa, accounting for approximately 96% 4 of the bacterial population in the rumen of all bulls (Figure 4.1A). A total of 109 families 5 were detected and the most abundant phylotypes were assigned to Prevotellaceae 6 (36.2%), Lachnospiraceae (6.0%), Ruminococcaceae (5.7%), Spirochaetaceae (5.3%), 7 Bacteroidaceae (3.0%), Porphyromonadaceae (2.6%), Enterobacteriaceae (2.1%), 8 Clostridiaceae (2.0%), and Fibrobacteraceae (1.7%) (Figure 4.1A). Our analysis also 9 revealed a total of 114 genera, with Prevotella (41.5%), Treponema (4.6%), Bacteroides 10 (3.5%), Ruminoccocus (3.5%), Ruminiclostridium (2.7%), Butyrivibrio (2.4%), Clostridium 11 (2.2%), and Fibrobacteres (2.0%) being the most abundant in the rumen of all bulls 12 (Figure 4.1A). Moreover, 114 bacterial species were classified in all samples, including 13 Prevotella ruminicola (35.8%), Butyrivibrio proteoclasticus (3.3%), Treponema 14 succinifaciens (2.9%), Fibrobacter succinogenes (1.8%), Ruminiclostridium sp KB18 15 (2.0%) and Ruminococcus albus (1.1%) (Figure 4.1A). A complete list of all bacterial 16 phylotypes is provided in Supplementary Table S1. 17

4.3.2 Overview of the active microbial functions 18

To investigate the functional potential of the rumen microbiota, ShotMAP was used to 19 survey which microbial functions (microbial gene families) were encoded in the 20 microbiome by mapping the mRNA reads to the KEGG Orthology (KO) database. It was 21 identified 109 active microbial functions in the rumen microbiome of the bulls and most of 22 them were associated with ribosome, Calvin cycle, reductive citrate cycle, 23 gluconeogenesis, glycolysis, and citrate cycle modules (Figure 4.1B). Among the various 24 functions encoded by the bacteriome, pyruvate - orthophosphate dikinase (K01006, 25 6.8%), pyruvate-ferredoxin/flavodoxin oxidoreductase (K03737, 3.3%), DNA-directed 26 RNA polymerase subunit beta (KO3046, 3.2%), glyceraldehyde 3-phosphate 27 dehydrogenase (K00134, 2.7%), small subunit ribosomal protein S1 (K02945, 1.7%), 28 phosphoenolpyruvate carboxykinase (ATP) (K01610, 1.4%), succinate dehydrogenase / 29 fumarate reductase, flavoprotein subunit (K00239, 1.4%), and large subunit ribosomal 30

120

protein L1 (K02863, 1.2%) were the most abundant pathways within their respective 1 modules (Figure 4.1B). A complete list of all microbial functions identified in this study is 2 provided in Supplementary Table S2. 3

4.3.3 Identification of active microbial and functional signatures across breeds 4

To gain insight into the bacterial and functional signatures characterizing the rumen 5 microbiome in Black and Red Angus, sPLS-DA was used to identify bacteria species and 6 functions characterizing each breed. Based on the best classification error rate 7 determined through cross-validation approaches two components were selected for the 8 bacterial signature and only one component for the functional signature (Figures S1 and 9 S2). Additionally, comparisons between CLR transformations of counts data with relative 10 abundance data were performed. The results showed the relative abundance data 11 generated only one component as the classification error rate increased after the addition 12 of a second component (Figures S3 and S4). It was evident that some rumen bacterial 13 species of the Black Angus signature overlapped in component 1 in both datasets (CLR 14 vs. Relative Abundance) (Figure 4.2B and Figure S4). However, the CLR transformed 15 data allowed the addition of a second component, improving the outcomes considerably 16 by providing further information for the bacterial signature of Red Angus cattle (Figure 17 4.2B). 18

Following the CLR transformation procedures, a clear separation in bacteria 19 species and functions differentiating the rumen microbiome of Black Angus from Red 20 Angus cattle was identified (Figures 4.2A and 4.3A). Overall, 80% of the bacterial 21 signature selected in component 1 of the sPLS-DA characterized the rumen microbiome 22 of Black Angus (Figures 4.2B and 4.2C), and this bacterial signature included members 23 of the families Chitinophagaceae (Chitinophaga pinensis), Clostridiaceae (Clostridium 24 stercorarium, Clostridium cellulosi, and Clostridium clariflavum), Ruminococcaceae 25 (Ruminoccocus albus and Ruminococcus bicirculans), Bacteroidaceae (Bacteroides 26 salanitronis), Porphyromonadaceae (Parabacteroides distasonis), and 27 Paludibacteraceae (Paludibacter propionicigenes). On the other hand, the component 2 28 of the sPLS-DA revealed that 60% of the bacterial signature was associated with Red 29 Angus, and this bacterial signature was comprised of the following species: Oscillibacter 30

121

valericigenes (Oscillospiraceae), Flavonifractor plautii (Clostridiales), Acidaminococcus 1 fermentans (Acidaminococcaceae), Paenibacillus graminis (Paenibacillaceae), and 2 Prevotella ruminicola (Prevotellaceae) (Figures 4.2B and 4.2C). 3

For the microbial functional signature, the rumen microbiome of Black Angus 4 displayed a diverse set of pathways including genetic information processing (e.g., 5 K02878, K02992), environmental information processing/membrane transport (e.g., 6 K10108), elongation factors for protein biosynthesis (e.g., K02355) and amino acid 7 metabolism (e.g., K01740) (Figure 4.3B and 4.3C). However, the rumen microbiome of 8 Red bulls was mostly enriched with pathways related to carbohydrate metabolism (e.g., 9 K00895, K01785) (Figure 4.3B and 4.3C). A complete list of bacterial species and 10 functions characterizing the rumen microbiome of Black and Red Angus (per component 11 of the sPLS-DA) is presented in Supplementary Table S3. 12

4.3.4 Relationship between active bacteria and volatile fatty acids 13 rCC analysis was implemented to investigate interactions between bacterial taxa and 14 VFAs (Figure 4.4A), as well as to identify interactions among different bacterial species 15 (Figure 4.4B). Associations between bacteria and VFAs revealed a positive correlation 16 between R. albus and total VFA, acetic and propionic acids (Figure 4.4A). Additionally, a 17 positive correlation was observed between P. ruminicola and S. ruminantium and 18 propionic, butyric and valeric acids, as well as between O. valericigenes and propionic, 19 butyric, and valeric acids (Figure 4.4A). On the other hand, negative correlations were 20 detected between C. pinensis and A. fermentans and total VFA, acetic, propionic, butyric 21 and valeric acids, as well as between C. stercorarium and propionic, butyric and valeric 22 acids (Figure 4.4A). Within the bacterial communities, P. ruminicola, S. ruminantium and 23 R. albus were negatively correlated with A. fermentans (Figures 4.4B), while C. 24 stercorarium and C. pinensis were positively correlated with each other and with R. 25 bicirculans, Ruminiclostridium sp KB18 and C. cellulosi (Figure 4.4B). 26

4.3.5 Relationship between active bacteria and feed efficiency 27

To evaluate the linkage among bacteria, functions and feed efficiency (measured as 28 FCR), ANCOM was used to identify taxa and functions differentially (P < 0.05) abundant 29

122

between breeds, adjusted for FCR and the time points. Of the 114 bacteria species 1 classified in all samples, it was found that only two species were associated (P < 0.05) 2 with FCR according to the breeds (Figure 4.5). Chitinophaga pinensis showed a relative 3 abundance 2.5 times higher (P < 0.05) in the rumen of Black Angus compared to Red 4 Angus associated with FCR and the four-time points (Figure 4.5A). Clostridium 5 stercorarium also exhibited a higher (P < 0.05) relative abundance in the rumen of Black 6 Angus compared to Red Angus associated with FCR, but only on the days 0 and 180 of 7 the experimental period (Figures 4.5B). No link between the functional signatures 8 detected by ShotMAP and FCR was found in the rumen of Black and Red Angus cattle. 9 10 4.4 Discussion 11 Until now only a few studies have addressed the question of the breed effect on the rumen 12 microbiome in beef cattle (Roehe et al. 2016, Guan et al. 2008, Z. P. Li et al. 2016). First, 13 PCR-denaturating gradient gel electrophoresis revealed that the rumen microbiome of 14 beef steers exhibiting different feed efficiencies (measured as residual feed intake) 15 clustered according to their breeds (Angus, Charolais, and Hereford-Angus), suggesting 16 that host breed may play a role in shaping the structure of the rumen microbiota (Guan et 17 al. 2008). Second, a metagenomic analysis of the rumen microbiome in Limousin- and 18 Aberdeen Angus-sired cattle showed that the abundance of microbial genes involved in 19 methanogenesis and feed efficiency (measured as FCR) could be used to predict host 20 metabolism, performance, and behavior (Roehe et al. 2016). Third, a metagenomic 21 analysis of the rumen microbiome in crosses between sika deer and elk reported that the 22 rumen microbiota in the hybrids differed from their parents, suggesting a significant effect 23 of host genetics on the rumen microbiome likely caused by vertical transmission of the 24 maternal microbiota (Z. P. Li et al. 2016). Although those three investigations suggested 25 a connection of the host genetics with the rumen microbiome, those studies used DNA- 26 based methods, which do not give a clear assessment of gene expression within the 27 rumen microbiota (Franzosa et al. 2014). More recently, Li et al. (2019a) compared rumen 28 metagenomic and metatranscriptomic datasets of three breeds of beef cattle (Angus, 29 Charolais, and Kinsella composite hybrid) and revealed that the rumen 30 metatranscriptome represents the functional activities of microbes and is more useful than 31

123

metagenomics to show the associations between rumen microbes and host phenotypical 1 traits. 2

However, previous studies using rumen-based metagenomics or 3 metatranscriptomics (Roehe et al. 2016, Guan et al. 2008, Z. P. Li et al. 2016) have 4 ignored that sequencing data hold a compositional structure because of necessary 5 relative proportion transformation (Cao et al. 2016, Mandal et al. 2015, Gloor et al. 2016, 6 Aitchison 1982). Thus, the traditional statistical methods (including Pearson correlations, 7 PCA, PCoA, GLM, partial least square analysis, ANOVA, linear regression) cannot be 8 applied directly in the analysis of relative abundance data because the independence 9 assumption between predictor variables is not met, which may result in spurious results 10 (Cao et al. 2016, Aitchison 1982, Lovell et al. 2015, Ban et al. 2015, Kurtz et al. 2014). A 11 recent study (Weiss et al. 2017) evaluating seven statistical methods for differential 12 abundance testing suggested that a novel methodology (ANCOM) based on log-ratio 13 transformations of count data, as defined by Aitchison (1982), was the most effective 14 approach to control false discovery rates. Similarly, Lê Cao and colleagues developed 15 sPLS-DA based on centered log ratio transformations of count data to identify microbial 16 signatures from diverse microbiomes (Rohart et al. 2017). Here, the transformations of 17 counts data implemented in these statistical methods provided more reliable outcomes 18 for the bacterial signatures than the relative abundance data and enabled a further 19 understanding of the relationships that exist between breed-specific bacterial signatures, 20 host phenotype (Black vs. Red Angus) and feed efficiency at the RNA-level. 21

Microbial signatures (or biomarkers) reflect correspondences between microbial 22 taxa and functions and are widely used to predict host phenotypes in human disease 23 states and forensic diagnostics (Knights et al. 2011, San-Juan-Vergara et al. 2018). In 24 general, it is unlikely that taxa that comprise the core rumen microbiome (including 25 Prevotella, Treponema, Ruminoccocus, Fibrobacter, and Butyrivibrio) will constitute 26 breed-specific bacterial signatures, since these taxa are prevalent in all ruminants 27 (bovines, camelids, caprids, cervids) regardless of sex, age, and breed (Henderson et al. 28 2015). Nonetheless, our results showed a breed effect on the core rumen microbiome for 29 Prevotella ruminicola, since this bacterium exhibited unexpectedly a higher abundance in 30

124

the rumen of Red Angus compared to Black Angus fed the same diet. As both breeds 1 were raised under the same nutritional regimen, it is probable that P. ruminicola was more 2 active in Red Angus to occupy specific ecological niches within the rumen, like the 3 degradation of hemicellulose and pectin (Russell and Rychlik 2001). Thus, we speculate 4 that if P. ruminicola is part of the microbial signature predicting the Red Angus phenotype, 5 then this biological signature may be used as a target for rumen manipulation purposes 6 in that specific breed. However, further investigations on a larger set of animals should 7 be conducted to differentiate the rumen microbiome of Red from Black Angus raised in 8 the same or in different geographical/environmental conditions to assess the robustness 9 of P. ruminicola as a differentiator of these two closely related breeds. 10

Moreover, we observed that Red Angus showed a microbial signature comprised 11 of bacterial phylotypes (e.g., Oscillibacter valericigenes and Flavonifractor plautii) that 12 were not yet described in the rumen microbiota from previous studies. In the gut of pigs, 13 O. valericigenes presented a positive correlation with valerate and butyrate production 14 (Pajarillo et al. 2015), which are short-chain fatty acids associated with improved feed 15 efficiency in beef cattle (Guan et al. 2008). In this study, O. valericigenes also showed a 16 positive correlation with valeric and butyric acids, suggesting that these microbes played 17 a role in the ruminal kinetics of Red Angus cattle, especially in the metabolism of valerate 18 and butyrate. Another microbe we detected in the bacterial signature of Red Angus was 19 Acidaminococcus fermentans. At the metagenomic level, a recent study recovered and 20 assembled the genome of A. fermentans from rumen samples collected from Aberdeen 21 Angus, Limousin, Charolais, and Luing (Stewart et al. 2018), confirming that this 22 bacterium is a ubiquitous inhabitant of the rumen microbiota in beef cattle. Possible 23 activities that A. fermentans perform in the rumen include the utilization of citrate as an 24 energy source to produce hydrogen and hydrogen sulfide and the decrease in the 25 accumulation of tricarballylate (a toxic end-product of ruminal fermentation) by oxidizing 26 trans-aconitate (Cook et al. 1994, Wallace et al. 2015, Stewart et al. 2018), 27 demonstrating that A. fermentans plays important roles in the Red Angus bacterial 28 signature. 29

125

In contrast, the Black Angus bacterial signature exhibited a larger number of 1 bacteria species responsible for the degradation of lignocellulosic biomass. One 2 example is Chitinophaga pinensis, which has been reported to encode a diverse array 3 of glycoside hydrolases (e.g., the GH27 enzyme CpArap27) capable of utilizing plant 4 biomass-derived carbohydrates (Larsbrink et al. 2017a, Larsbrink et al. 2017b). Other 5 instances of lignocellulose-degrading microbes we found in Black Angus bacterial 6 signature included Ruminoccocus albus and Clostridium stercorarium, which are fibrolytic 7 species actively involved in the digestion of cellulose/xylan (Christopherson et al. 2014) 8 and hemicellulose (Schellenberg et al. 2014), respectively. In addition to showing a higher 9 abundance in the rumen of Black Angus compared to Red Angus, C. pinensis and C. 10 stercorarium were the only species directly connected with feed efficiency measurements 11 (FCR) taken across the feeding period, confirming the beneficial effects of these two 12 species on the deconstruction of plant cell wall fibers in forage-fed cattle (Christopherson 13 et al. 2014, Larsbrink et al. 2017a). With its limited sample size, however, our study lacked 14 the power necessary to find differences in feed efficiency (FCR) between Black and Red 15 Angus (Figure S5). Although FCR measurements did not differ significantly between the 16 two breeds, evidence for superior performance of Black Angus compared to Red Angus 17 has been described previously (McLean and Schmutz 2009). However, other studies 18 showed that Red Angus had a better gain to feed ratios than Black Angus, demonstrating 19 that differences in feeding behavior exist between these two genetically similar breeds 20 (Wolfger et al. 2016). Thus, it is necessary to further investigate whether C. pinensis and 21 C. stercorarium would contribute to growth performance in Angus cattle exhibiting 22 different feed efficiencies (e.g., high vs. low FCR) under various feeding regimens to 23 validate our findings. Although three animals per breed may not confer the necessary 24 power to detect differences in FCR with four-time points per animal in the current study, 25 the variations in those two rumen bacteria associated with FCR are valid. However, more 26 animals should be included in the future to support our findings (Figure S5). 27

Understanding the functions of the rumen microbiome is crucial to the 28 development of technologies and practices that support efficient global food production 29 from ruminants (Seshadri et al. 2018). The identification of functional signatures in the 30 current study provided evidence that the functions performed by the rumen microbiota 31

126

depend on breed. The rumen functional signature found in Red Angus was to a large 1 extent related to carbohydrate metabolism, corroborating the findings reported in 2 previous metagenomic studies that showed a high abundance of carbohydrate-related 3 functions in cattle fed forage diets (Wang et al. 2013). In this context, it was not a 4 surprise to detect active pathways involved mainly in the fermentation of glucose to 5 acetate from Acetyl-CoA via the Embden-Meyerhof-Parnas (EMP) pathway (Russell 6 and Rychlik 2001, Gottschalk 1986). However, it was found an enzyme in Red Angus 7 called phosphofructokinase [EC:2.7.1.90, K00895] that can form acetate through an 8 atypical (incomplete) EMP pathway. A recent study has provided evidence that atypical 9 EMP pathways are encoded in the genome of four rumen bacterial species (including P. 10 ruminicola) (Hackmann et al. 2017), which showed a higher abundance in the Red Angus 11 rumen microbiome as discussed previously. According to Hackmann et al. (2017), those 12 bacteria could convert glucose into pyruvate through the supply of phosphofructokinases 13 [EC:2.7.1.90, K00895] rather than from Acetyl-CoA. 14

Unlike Red Angus cattle, the functional signature in Black Angus exhibited an 15 increase in microbial functions associated with protein synthesis during the assembly of 16 the bacterial ribosome (e.g., ribosomal protein L16 [K02878] and ribosomal protein S7 17 [K02992]). The mechanisms of action of proteins L16 have been described in Escherichia 18 coli and Thermus thermophilus HB8 (Anikaev et al. 2016). These studies have shown 19 that mutations in those proteins lead to a significant reduction in cell growth and a 20 decrease in their translation apparatus. Mutations in the protein L16, which participates 21 in the formation of functional sites of the 50S ribosomal subunit (Anikaev et al. 2016), can 22 activate the binding sites for antibiotics in the internal cavity of the bacterial ribosome. In 23 a similar fashion, mutations in the binding sites of conserved regions (30S subunit) of the 24 bacterial RNA in protein S7 can destabilize the correct three-dimensional structure of the 25 assembled ribosome (Wimberly et al. 1997). Although our findings showed that the 26 proteins L16/S7 were more relevant to Black Angus than to Red Angus, further studies 27 exploring relationships between proteins L16/S7 and bacterial growth dynamics in other 28 beef cattle breeds are needed to validate our results. Future directions may also include 29 studying the interactions between proteins L16/S7 and RNA-biding sites with the goal of 30

127

designing antibiotics with novel structures and modes of action to lower the increasing 1 bacterial resistance to commonly used antimicrobials in the beef cattle industry. 2

4.5 Conclusions 3

This study described the active microbial and functional signatures of the rumen 4 microbiome in two closely related breeds using total RNA-seq-based metatranscriptomics. 5 Our results showed that Black and Red Angus exhibited differences in microbial 6 signatures at the functional and compositional levels, indicating that the breed influences 7 the structure of the rumen microbiome and its function. It is noticeable that the bacterial 8 signature in Black Angus was characterized by an increase in fibrolytic species such as 9 Chitinophaga pinensis, Clostridium stercorarium, and Ruminococcus albus, whereas the 10 bacterial signature in Red Angus was composed of poorly characterized species (e.g., 11 Oscillibacter valericigenes and Flavonifractor plautii). Interestingly, FCR was associated 12 with a specific bacterial signature depending on the breed type, suggesting that there is 13 a connection among feed efficiency, active rumen microbiota, and the genetic makeup of 14 the host. In this context, C. pinensis and C. stercorarium were the only species associated 15 with FCR, and this result emphasizes the important role of these species in the feed 16 conversion efficiency of forage-fed bulls. Moreover, our results showed specific functional 17 signatures characterizing each breed (separately), demonstrating that different strategies 18 to modify the rumen microbial pathways need to be designed for Red Angus as compared 19 to Black Angus. While the first showed microbial functions enriched mainly from 20 carbohydrate metabolism, the latter exhibited a more diverse set of pathways, with 21 emphasis on the ribosomal proteins L16 and S7. It is important to mention that the 22 statistical methods employed in this study can also be applied to investigate rumen 23 archaea, fungi, and protozoa in addition to bacteria as reported in the current study. 24 However, the lack of genome sequences representatives of rumen archaeal, fungal, and 25 protozoal populations (especially in the publicly available database - e.g., NCBI) does not 26 allow accurate taxonomic assessment at species level for these organisms by Kraken. 27 Thus, future studies involving the rumen microbiome as a whole are warranted to 28 enhance our knowledge of the symbiotic relationship that exists among rumen 29 microorganisms and how they may differ between these two genetically similar breeds. 30

128

Regardless, our findings highlight how we can use signatures of bacterial taxa and their 1 functions to harness the full potential of the rumen microbiome in Angus cattle. 2 Furthermore, the results of our study can be used as a reference for future investigations 3 related to the manipulation of the ruminal function and selection of beef cattle with high 4 feed efficiency based on specific microbial signatures. 5

4.6 Literature cited 6 Aitchison, J. (1982) 'The statistical analysis of compositional data', Journal of the Royal 7 Statistical Society Series B-Methodological, 44(2), 139-177. 8 Anikaev, A., Isaev, A., Korobeinikova, A., Garber, M. and Gongadze, G. (2016) 'Role of 9 protein L25 and its contact with protein L16 in maintaining the active state of 10 Escherichia coli ribosomes in vivo', Biochemistry (00062979), 81(1), 19-27. 11 Asnicar, F., Weingart, G., Tickle, T. L., Huttenhower, C. and Segata, N. (2015) 'Compact 12 graphical representation of phylogenetic data and metadata with GraPhlAn', 13 PeerJ, 3. 14 Ban, Y., Jiang, H. and An, L. (2015) '- Investigating microbial co-occurrence patterns 15 based on metagenomic compositional data', - 31(- 20), - 3329. 16 Bolger, A. M., Lohse, M. and Usadel, B. (2014) 'Trimmomatic: a flexible trimmer for 17 Illumina sequence data', Bioinformatics (Oxford, England), 30(15), 2114-2120. 18 Cao, K. A. L., Costello, M. E., Lakis, V. A., Bartolo, F., Chua, X. Y., Brazeilles, R. and 19 Rondeau, P. (2016) 'MixMC: A multivariate statistical framework to gain insight into 20 microbial communities', Plos One, 11(8). 21 Christopherson, M. R., Dawson, J. A., Stevenson, D. M., Cunningham, A. C., 22 Bramhacharya, S., Weimer, P. J., Kendziorski, C. and Suen, G. (2014) 'Unique 23 aspects of fiber degradation by the ruminal ethanologen Ruminococcus albus 7 24 revealed by physiological and transcriptomic analysis', BMC Genomics, 15. 25 Cook, G. M., Rainey, F. A., Chen, G., Stackebrandt, E. and Russell, J. B. (1994) 26 'Emendation of the description of Acidaminococcus fermentans, a trans- aconitate- 27 and citrate-oxidizing bacterium', International Journal Of Systematic Bacteriology, 28 44(3), 576-578. 29 Creevey, C. J., Kelly, W. J., Henderson, G. and Leahy, S. C. (2014) 'Determining the 30 culturability of the rumen bacterial microbiome', Microbial Biotechnology, 7(5), 467- 31 479. 32 Crews, D. H., Jr., Shannon, N. H., Genswein, B. M. A., Crews, R. E., Johnson, C. M. and 33 Kendrick, B. A (2005) 'Genetic parameters for net feed efficiency of beef cattle 34 measured during postweaning growing versus finishing periods', Proc. West Sec. 35 Am. Soc. Anim. Sci, 54(54), 125-128. 36 Franzosa, E. A., Morgan, X. C., Segata, N., Waldron, L., Reyes, J., Earl, A. M., 37 Giannoukos, G., Boylan, M. R., Ciulla, D., Gevers, D., Izard, J., Garrett, W. S., 38 Chan, A. T. and Huttenhower, C. (2014) 'Relating the metatranscriptome and 39 metagenome of the human gut', Proceedings of the National Academy of 40 Sciences, 111(22), E2329-E2338. 41

129

Geishauser, T. (1993) 'An Instrument for collection and transfer of ruminal fluid and for 1 administration of water soluble drugs in adult cattle', The Bovine Practitioner, (27), 2 38-42. 3 Gloor, G., Wu, J. R., Pawlowsky-Glahn, V. and Jose Egozcue, J. (2016) '- It's all relative: 4 analyzing microbiome data as compositions', - 26(- 5), - 329. 5 Gonzalez, I., Dejean, S., Martin, P. G. P. and Baccini, A. (2008) 'CCA: An R package to 6 extend canonical correlation analysis', Journal of Statistical Software, 23(12), 1- 7 14. 8 Gottschalk, G. (1986) 'Bacterial metabolism', 9 Guan, L. L., Nkrumah, J. D., Basarab, J. A. and Moore, S. S. (2008) 'Linkage of microbial 10 ecology to phenotype: correlation of rumen microbial ecology to cattle's feed 11 efficiency', Fems Microbiology Letters, 288(1), 85-91. 12 Hackmann, T. J., Tao, J., Ngugi, D. K. and Firkins, J. L. (2017) 'Genomes of rumen 13 bacteria encode atypical pathways for fermenting hexoses to short-chain fatty 14 acids', Environmental Microbiology, 19(11), 4670-4683. 15 Kanehisa, M., Goto, S., Kawashima, S., Okuno, Y. and Hattori, M. (2004) 'The KEGG 16 resource for deciphering the genome', Nucleic Acids Research, 32(Database 17 issue), D277-D280. 18 Knights, D., Parfrey, L. W., Zaneveld, J., Lozupone, C. and Knight, R. (2011) 'Human- 19 associated microbial signatures: examining their predictive value', Cell Host & 20 Microbe, 10(4), 292-296. 21 Kopylova, E., Noé, L. and Touzet, H. (2012) 'SortMeRNA: fast and accurate filtering of 22 ribosomal RNAs in metatranscriptomic data', Bioinformatics (Oxford, England), 23 28(24), 3211-3217. 24 Kurtz, Z. D., Mueller, C. L., Miraldi, E. R., Littman, D. R., Blaser, M. J. and Bonneau, R. 25 A. (2014) 'Sparse and compositionally robust inference of microbial ecological 26 networks', 27 Larsbrink, J., Tuveng, T. R., Pope, P. B., Bulone, V., Eijsink, V. G. H., Brumer, H. and 28 McKee, L. S. (2017a) 'Proteomic insights into mannan degradation and protein 29 secretion by the forest floor bacterium Chitinophaga pinensis', Journal of 30 Proteomics, 156, 63-74. 31 Larsbrink, J., Tuveng, T. R., Pope, P. B., Eijsink, V. G. H., Bulone, V., Brumer, H. and 32 McKee, L. S. (2017b) 'Proteomic data on enzyme secretion and activity in the 33 bacterium Chitinophaga pinensis', Data in Brief, 11, 484-490. 34 Lee, S., Sung, J., Lee, J. and Ko, G. (2011) 'Comparison of the gut microbiotas of healthy 35 adult twins living in South Korea and the United States', Applied and Environmental 36 Microbiology, 77(20), 7433-7437. 37 Li, D. H., Liu, C. M., Luo, R. B., Sadakane, K. and Lam, T. W. (2015) 'MEGAHIT: an ultra- 38 fast single-node solution for large and complex metagenomics assembly via 39 succinct de Bruijn graph', Bioinformatics, 31(10), 1674-1676. 40 Li, F. and Guan, L. L. (2017) 'Metatranscriptomic profiling reveals linkages between the 41 active rumen microbiome and feed efficiency in beef cattle', Applied and 42 Environmental Microbiology. 43 Li, F., Hitch, T. C. A., Chen, Y., Creevey, C. J. and Guan, L. L. (2019) 'Comparative 44 metagenomic and metatranscriptomic analyses reveal the breed effect on the 45

130

rumen microbiome and its associations with feed efficiency in beef cattle', 1 Microbiome, 7(1), 6. 2 Li, Z. P., Wright, A. D. G., Si, H. Z., Wang, X. X., Qian, W. X., Zhang, Z. G. and Li, G. Y. 3 (2016) 'Changes in the rumen microbiome and metabolites reveal the effect of host 4 genetics on hybrid crosses', Environmental Microbiology Reports, 8(6), 1016- 5 1023. 6 Lovell, D., Pawlowsky-Glahn, V., Egozcue, J. J., Marguerat, S. and Bähler, J. (2015) 7 'Proportionality: A Valid Alternative to Correlation for Relative Data', Plos 8 Computational Biology, 11(3), 1-12. 9 Mandal, S., Van Treuren, W., White, R. A., Eggesbo, M., Knight, R. and Peddada, S. D. 10 (2015) 'Analysis of composition of microbiomes: a novel method for studying 11 microbial composition', Microbial Ecology in Health and Disease, 26, 27663-Article 12 No.: 27663. 13 Márquez, G. C., Speidel, S. E., Enns, R. M. and Garrick, D. J. (2010) 'Genetic diversity 14 and population structure of American Red Angus cattle', Journal of Animal 15 Science, 88(1), 59-68. 16 McLean, K. L. and Schmutz, S. M. (2009) 'Associations of melanocortin 1 receptor 17 genotype with growth and carcass traits in beef cattle', Canadian Journal of Animal 18 Science, 89(3), 295-300. 19 Nayfach, S., Bradley, P. H., Wyman, S. K., Laurent, T. J., Williams, A., Pollard, K. S., 20 Sharpton, T. J. and Eisen, J. A. (2015) 'Automated and accurate estimation of gene 21 family abundance from shotgun metagenomes', Plos Computational Biology, 22 11(11), 29p. 23 Neves, A. L. A., Li, F., Ghoshal, B., McAllister, T. and Guan, L. L. (2017) 'Enhancing the 24 resolution of rumen microbial classification from metatranscriptomic data using 25 Kraken and Mothur', Frontiers in Microbiology, 8(2445). 26 Opdahl, L. J., Gonda, M. G. and St-Pierre, B. (2018) 'Identification of Uncultured Bacterial 27 Species from Firmicutes, Bacteroidetes and CANDIDATUS Saccharibacteria as 28 Candidate Cellulose Utilizers from the Rumen of Beef Cows', Microorganisms, 29 6(1). 30 Pajarillo, E. A. B., Chae, J. P., Balolong, M. P., Kim, H. B., Kang, D.-K. and Seo, K.-S. 31 (2015) '- Characterization of the Fecal Microbial Communities of Duroc Pigs Using 32 16S rRNA Gene Pyrosequencing', - 28(- 4), - 591. 33 Roehe, R., Dewhurst, R. J., Duthie, C.-A., Rooke, J. A., McKain, N., Ross, D. W., Hyslop, 34 J. J., Waterhouse, A., Freeman, T. C., Watson, M. and Wallace, R. J. (2016) 35 'Bovine Host Genetic Variation Influences Rumen Microbial Methane Production 36 with Best Selection Criterion for Low Methane Emitting and Efficiently Feed 37 Converting Hosts Based on Metagenomic Gene Abundance', Plos Genetics, 12(2). 38 Rohart, F., Gautier, B., Singh, A. and Lê Cao, K.-A. (2017) 'mixOmics: An R package for 39 ‘omics feature selection and multiple data integration', Plos Computational Biology, 40 13(11), 1-19. 41 Russell, J. B. and Rychlik, J. L. (2001) 'Factors that alter rumen microbial ecology', 42 Science, 292(5519), 1119-1122. 43 San-Juan-Vergara, H., Peña, M., Cadena-Cruz, C., Diaz-Olmos, Y., Hurtado-Gómez, L., 44 Vidal-Orjuela, O., Viasus, D., Cervantes-Acosta, G., Zurek, E., Portnoy, I., Vélez, 45 J. I., Ajami, N. J., Petrosino, J. F., Mogollon, C., Hernández, D., Urruchurtu, I., Di- 46

131

Ruggiero, P., Sanchez-Sit, S., Guardo-García, E. and Torres, N. (2018) 'A 1 Lachnospiraceae-dominated bacterial signature in the fecal microbiota of HIV- 2 infected individuals from Colombia, South America', Scientific Reports, 8(1). 3 Schellenberg, J. J., Verbeke, T. J., Sparling, R., McQueen, P., Krokhin, O. V., Wilkins, J. 4 A., Zhang, X., Alvare, G., Fristensky, B., Levin, D. B., Thallinger, G. G. and 5 Henrissat, B. (2014) 'Enhanced whole genome sequence and annotation of 6 Clostridium stercorarium DSM8532T using RNA-seq transcriptomics and high- 7 throughput proteomics', BMC Genomics, 15(1). 8 Seshadri, R., Leahy, S. C., Attwood, G. T., Teh, K. H., Lambie, S. C., Cookson, A. L., 9 Eloe-Fadrosh, E. A., Pavlopoulos, G. A., Hadjithomas, M., Varghese, N. J., Paez- 10 Espino, D., Perry, R., Henderson, G., Creevey, C. J., Terrapon, N., Lapebie, P., 11 Drula, E., Lombard, V., Rubin, E., Kyrpides, N. C., Henrissat, B., Woyke, T., 12 Ivanova, N. N. and Kelly, W. J. (2018) 'Cultivation and sequencing of rumen 13 microbiome members from the Hungate1000 Collection', Nature Biotechnology, 14 36(4), 359-367. 15 Stewart, R. D., Warr, A., Watson, M., Auffret, M. D., Dewhurst, R. J., Roehe, R., Wiser, 16 A. H., Press, M. O., Langford, K. W., Liachko, I., Snelling, T. J. and Walker, A. W. 17 (2018) 'Assembly of 913 microbial genomes from metagenomic sequencing of the 18 cow rumen', Nature Communications, 9(1). 19 Wallace, R. J., Rooke, J. A., McKain, N., Duthie, C.-A., Hyslop, J. J., Ross, D. W., 20 Waterhouse, A., Watson, M. and Roehe, R. (2015) 'The rumen microbial 21 metagenome associated with high methane production in cattle', BMC Genomics, 22 16, 1-14. 23 Wang, L., Hatem, A., Catalyurek, U. V., Morrison, M. and Yu, Z. (2013) 'Metagenomic 24 Insights into the Carbohydrate-Active Enzymes Carried by the Microorganisms 25 Adhering to Solid Digesta in the Rumen of Cows', Plos One, 8(11), 1-11. 26 Wang, Y. and McAllister, T. A. (2002) 'Rumen microbes, enzymes and feed digestion - A 27 review', Asian-Australas J Anim Sci, 15(11), 1659-1676. 28 Weiss, S., Xu, Z. Z., Peddada, S., Amir, A., Bittinger, K., Gonzalez, A., Lozupone, C., 29 Zaneveld, J. R., Vazquez-Baeza, Y., Birmingham, A., Hyde, E. R. and Knight, R. 30 (2017) 'Normalization and microbial differential abundance strategies depend upon 31 data characteristics', Microbiome, 5. 32 Wimberly, B. T., Ramakrishnan, V. and White, S. W. (1997) 'The structure of ribosomal 33 protein S7 at 1.9 Å resolution reveals a β-hairpin motif that binds double-stranded 34 nucleic acids', Structure, 5(9), 1187-1198. 35 Wolfger, B., Quinn, C., Torres, G. W., Taylor, M. and Orsel, K. (2016) 'Comparison of 36 feeding behavior between black and red Angus feeder heifers', Canadian Journal 37 of Animal Science, 96(3), 404-409. 38 Wood, D. E. and Salzberg, S. L. (2014) 'Kraken: ultrafast metagenomic sequence 39 classification using exact alignments', Genome Biology, 15(3), R46-R46. 40 Zhou, M., Peng, Y.-J., Chen, Y., Klinger, C. M., Oba, M., Liu, J.-X. and Guan, L. L. (2018) 41 'Assessment of microbiome changes after rumen transfaunation: implications on 42 improving feed efficiency in beef cattle', Microbiome, 6(1), 62-62. 43 44 45 46

132

1 4.7 Tables and figures 2 3

(A) Legend:

A: Bacteroidaceae B: Bacteroides C: Porphyromonadaceae D: Prevotellaceae E: Prevotella F: Prevotella ruminicola G: Marivirga tractuosa H: Flavobacteriaceae I: Fibrobacteraceae J: Fibrobacter K: Fibrobacter succinogens L: Clostridiaceae M: Clsotridium N: Lachnospiraceae O: Butyrivibrio P: Butyrivibrio proteoclasticus Q: Ruminococcaceae R: Ruminiclstridium S: Ruminiclsotridium sp KB18 T: Ruminococcus U: Ruminococcus bicirculans V: Natranaerobius thermophilus W: Enterobcteriaceae X: Spirochaetaceae Y: Treponema Z: Treponema brenaborense a: Treponema succinifaciens

(B) Legend:

A: K00134 a: K02931 B: K00615 b: K02945 C: K00927 c: K02982 D: K01624 d: K00615 E: K00239 e: K01624 F: K00239 G: K02111 H: K02112 I: K00134 J: K00927 Modules K: K01610 L: K01624 M: K00134 N: K00927 O: K01624 P: K00134 Q: K00927 R: K00239 S: K03040 T: K03043 U: K03046 V: K00239 W: K01006 X: K03737 Y: K02863 Z: K02886 4 Figure 4.1 Taxonomic composition and microbial gene families obtained from 5 metatranscriptome data in beef cattle. (A) Taxonomic cladogram showing the most 6

133

abundant detected taxa (relative abundance ≥ 0.05% in at least half of the samples). The 1 six rings of the cladogram stand for phylum (innermost), class, order, family, genus, and 2 species (outermost), respectively. The sizes of the circles indicate the mean average 3 abundance of each taxon. (B) Cladogram showing the most abundant microbial functions 4 (relative abundance 0.05% ≥ in at least half of the samples). Two rings of the cladogram 5 stand for module (innermost) and microbial functions obtained from the KEEG database 6 (outermost), respectively. The sizes of the circles indicate the mean average abundance 7 of each function (microbial gene family). Cladograms were built using GraPhlan (Asnicar 8 et al. 2015). 9

134

Figure 4.2 sPLS-DA results on rumen bacterial species in Black and Red Angus. (A) 3 shows a sample plot on the first two sPLS-DA components with 95% confidence level 4

135

ellipse plots and (B) represents the contribution of each species selected on the first and 1 second components, with length of the bar representing the importance of each species 2 to the component (importance from bottom to top). Colors indicate the breed (Black vs. 3 Red Angus) in which the species is most abundant. (C) represents a hierarchical 4 clustering (Euclidean distance, Ward linkage) of the selected species from sPLS-DA 5 results. In the heatmap legend, the red and blue colors indicate strong positive and 6 negative correlations respectively, whereas yellow or green indicate weaker correlation 7 values. On the left-hand side of the heatmap, clusters of the biological samples collected 8 according to time are colored in orange, blue, dark green and violet to signify the time 9 points 0, 80, 100 and 180 days, respectively. 10

136

137

Figure 4.3 sPLS-DA results on rumen microbial gene families in Black and Red Angus. 1 (A) shows a sample plot on the first sPLS-DA component with 95% confidence level 2 ellipse plots and (B) represents the contribution of each function (microbial gene family) 3 selected on the first component, with length of the bar representing importance of each 4 microbial gene family to the component (importance from bottom to top). Colors indicate 5 the breed (Black vs. Red Angus) in which the gene family is most abundant. (C) 6 represents the hierarchical clustering (Euclidean distance, Ward linkage) of the selected 7 gene families from sPLS-DA results. In the heatmap legend, the red and blue colors 8 indicate strong positive and negative correlations respectively, whereas yellow or green 9 indicate weaker correlation values. On the left-hand side of the heatmap, clusters of the 10 biological samples collected according to time are colored in orange, blue, dark green 11 and violet to signify the time points 0, 80, 100 and 180 days, respectively. 12

14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

138 Bacterial Species, sPLSDA comp 1 − 2 Bacterial Species, sPLSDA comp 1 − 2

4 4

91ZD3L 91ZD3L

91ZD4L 91ZD4L

2 2

91ZD2L 91ZD2L

162ZD4L 162ZD4L

162ZD2L (a) 17ZD3H (b)162ZD2L 17ZD3H

r r Component 1 Component 2

a Bacterial Species, sPLSDA compa 1 − 2

v v Contribution on comp 2

Contribution on comp 1

l 60ZD2L 4 VFA, rCCA comp 1 - 2 l 60ZD2L

(A) 91ZD1L 91ZD1L 17ZD2H Breeds [Clostridium] cellulosi 17ZD2H Breeds

% 1.0 % Staphylococcus aureus

Black Black

: 20LZD4L : Bacteria 20LZD4L

e 91ZD3L e Red Red

t t VFA Ruminococcus bicirculans

i 20LZD2L i 20LZD2L

r 3 r

v 0 1 91ZD4L v 0

− 2 − Legend:

X 162ZD3L 2 162ZD3LClostridium bor nimense 6 Psychroflexus torquis 91ZD2L 1. Total VFA 2. Acetic 60ZD3L 17ZD1H162ZD4L 60ZD3L 17ZD1H 0.5 3. BacillusPro thurpingiensisionic 162ZD2L 126D1L 17ZD3H 126D1L

a 4.

v Valeric

. l 60ZD2L

e 91ZD1L 17ZD2H Breeds 5. Butyric

4 7

162ZD1L60ZD4L Black162ZD1L60ZD4L Oscillibacter valericigenes 5 : 20LZD4L Capnocytophaga haemolytica

6. Ruminococcus albus

e Red Breeds t Breeds

i 20LZD2L

2 v 0 7. Prevotella ruminicola

− 126D2L 126D2L Black Black X 162ZD3L t 60ZD1L 7 60ZD1L8. Selenomonas ruminantium Ruminococcus alb us Red n Red

2 60ZD3L 17ZD1H

e 126D1L 9. Oscillibacter valericigenes

t 20LZD3L 17ZD4H Block 20LZD3L 17ZD4H

n 10 n 8 126D4L9 126D4L e 162ZD1L60ZD4L 10. Clostridium stercorarium

o 20LZD1L 20LZD1L n 0.0 X Acidaminococcus fermentans

p 126D2L 11Igna. Rvibacterumiumi nalboumcoccus bicirculans p 60ZD1L11

m Y 126D3L 17ZD4H 126D3L

m 20LZD3L o 126D4L 12. Ruminiclostridium sp KB18 20LZD1L

o 13 126D3L 13. Clostridium cellulosi C 14 12 15 [Clostridium] stercor arium −2 −2 −2 14. Flavonifractor plautii 15. Chitinophaga pinensis 16. Acidaminococcus fermentans Olsenella uli Chitinophaga pinensis

-0.5 −4 −2 0 2 16 X−variate 1: 7% expl. var 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.8

X−variate 1: 7% expl. var Color key X−variate 1: 7% expl. var Breeds

-1.0 Black Red -1.0 -0.5 0.0 0.5 1.0 140 Component 1 ** P < 0.05 −3.65 −1.83 0 1.83 3.65 Component 1 Black_17ZD3H_100 130 Black_20LZD4L_180 x Black_17ZD2H_80

d Black_20LZD2L_80

I Red_91ZD3L_100 (B)

y Red_91ZD4L_180

t 120 i Red_91ZD2L_80

Color key s

)

r (A) Black_126D1L_0 s

v Black_17ZD1H_0

Black_17ZD4H_180 n

D 110 i

(

-0.47 -0.24 0 0.24 0.47 Black_126D2L_80

h Black_20LZD3L_100

l Black_20LZD1L_0

D A Black_126D4L_180

I 100

_ r Clostridium stercorarium Black_126D3L_100

Red_60ZD2L_80 e

s Red_91ZD1L_0 r i Chitinophaga pinensis

B Bacterial Species,F 90 sPLSDA comp 1 − 2 Bacterial Species, sPLSDA comp 1 − 2 Red_162ZD4L_180 Ruminococcus bicirculans Red_162ZD2L_80 Red_162ZD3L_100 4 4 Clostridium cellulosi Red_60ZD3L_100 80 Red_60ZD1L_0 Red_60ZD4L_180 Ruminiclostridium sp KB18 Red_162ZD1L_0

s s s

s s

e a

e n

s c

Black Red l

n a

a i

Oscillibacter valericigenes l

q l

n i

i Breeds c

]

u l

t Flavonifractor plautii g

h a

]

Acidaminococcusa fermentans

o 91ZD3L a 91ZD3L

p s

[

O Ruminococcus albus R

C Prevotella ruminicola 91ZD4L 91ZD4LBacterial Species Selenomonas ruminantium 2 2

c 91ZD2L i 91ZD2L

P 162ZD4L 162ZD4L

162ZD2L (a) 17ZD3H (b)162ZD2L 17ZD3H

r (B) r Component 1 Component 2

a Bacterial Species, sPLSDA compa 1 − 2

v v Contribution on comp 2

Contribution on comp 1

l 60ZD2L 4 VFA, rCCA comp 1 - 2 l 60ZD2L

(A) 91ZD1L 91ZD1L 17ZD2H Breeds [Clostridium] cellulosi 17ZD2H Breeds

% 1.0 % Staphylococcus aureus

Black Black

: 20LZD4L : Bacteria 20LZD4L

e 91ZD3L e Red Red

t t VFA Ruminococcus bicirculans

a 20LZD2L a 20LZD2L

r 3 r

v 0 1 91ZD4L v 0

− 2 − Legend:

X 162ZD3L 2 162ZD3LClostridium bornimense 6 Psychroflexus torquis 91ZD2L 1. Total VFA 2. Acetic 60ZD3L 17ZD1H162ZD4L 60ZD3L 17ZD1H 0.5 3. BacillusPro thurpingiensisionic 162ZD2L 126D1L 17ZD3H 126D1L

a 4.

v Valeric

. l 60ZD2L

e 91ZD1L 17ZD2H Breeds 5. Butyric

4 7

162ZD1L60ZD4L Black162ZD1L60ZD4L Oscillibacter v alericigenes 5 : 20LZD4L Capnocytophaga haemolytica

6. Ruminococcus albus

e Red Breeds t Breeds

a 20LZD2L

2 v 0 7. Prevotella ruminicola

− 126D2L 126D2L Black Black X 162ZD3L t 60ZD1L 7 60ZD1L8. Selenomonas ruminantium Ruminococcus albus Red n Red

2 60ZD3L 17ZD1H

e 126D1L 9. Oscillibacter valericigenes

t 20LZD3L 17ZD4H Block 20LZD3L 17ZD4H

n 10 n 8 126D4L9 126D4L e 162ZD1L60ZD4L 10. Clostridium stercorarium

o 20LZD1L 20LZD1L n 0.0 X Acidaminococcus f ermentans

p 126D2L 11Igna. Rvibacterumiumi nalboumcoccus bicirculans p 60ZD1L11

m Y 126D3L 17ZD4H 126D3L

m 20LZD3L o 126D4L 12. Ruminiclostridium sp KB18 20LZD1L

o 13 126D3L 13. Clostridium cellulosi C 14 12 15 [Clostridium] stercorarium −2 −2 −2 14. Flavonifractor plautii 15. Chitinophaga pinensis 16. Acidaminococcus fermentans Olsenella uli Chitinophaga pinensis

-0.5 −4 −2 0 2 16 X−variate 1: 7% expl. var 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.8

X−variate 1: 7% expl. var Color key X−variate 1: 7% expl. var Breeds

-1.0 Black Red -1.0 -0.5 0.0 0.5 1.0 140 Component 1 ** P < 0.05 −3.65 −1.83 0 1.83 3.65 Component 1 Black_17ZD3H_100 1 130 Black_20LZD4L_180 x Black_17ZD2H_80

d Black_20LZD2L_80

I Red_91ZD3L_100 2 (B)

y Red_91ZD4L_180

t 120 i Red_91ZD2L_80

Color key s

)

r Black_126D1L_0 s

e a 3 v Black_17ZD1H_0

Black_17ZD4H_180 n

D 110 i

(

-0.47 -0.24 0 0.24 0.47 Black_126D2L_80

h Black_20LZD3L_100

l Black_20LZD1L_0

D A Black_126D4L_180

I 100

_ r Clostridium stercorarium Black_126D3L_100

Red_60ZD2L_80 e

s Red_91ZD1L_0 r i Chitinophaga pinensis

B F 90 Red_162ZD4L_180 Ruminococcus bicirculans Red_162ZD2L_80 Red_162ZD3L_100 Clostridium cellulosi Red_60ZD3L_100 80 Red_60ZD1L_0 139 Red_60ZD4L_180 Ruminiclostridium sp KB18 Red_162ZD1L_0

s s s

s s

e a

e n

s c

Black Red l

n a

a i

Oscillibacter valericigenes l

q l

n i

i Breeds c

]

u l

g Flavonifractor plautii t

h a

]

Acidaminococcus fermentans t

t C

[

O Ruminococcus albus R

C Prevotella ruminicola Bacterial Species Selenomonas ruminantium

Figure 4.4. Heatmap and correlation circle plot generated from the output of regularized 1 canonical correlation (rCC) method. (A) shows the correlations between VFAs (total VFA, 2 acetic, butyric, propionic, and valeric acids) and rumen bacterial species in the first two 3 rCC components. The color key indicates the correlation values among variables. (B) 4 shows the correlation circle plot, where VFAs and bacteria are shown inside a circle of 5 radius 1 centered at the origin, with strongly associated (or correlated) variables being 6 projected in the same direction from the origin. Two circumferences of radius 1 and 0.5 7 were plotted to reveal the correlation structure of the variables. 8 9 10 11 12 13 14 15 16 17 18

140

1 2 Figure 4.5 Differentially abundant bacterial species detected in Black and Red Angus 3 when FCR (Feed consumed kg/Gain kg) was adjusted to time (in days). Unadjusted raw 4 average relative abundance and standard errors of (A) Chitinophaga pinensis, (B) 5 Clostridium stercorarium, and (C) Ruminococcus albus detected when FCR (kg/day) was 6 adjusted to time (in days). Statistical procedures were performed by ANCOM (calculated 7 on the log-ratio matrix; FDR, q < 0.1). 8

141

1 Chapter 5 2

Discovery and targeted functional profiling of novel glycoside hydrolases 3 through selective pressure on the native rumen microbial community 4

5.1 Introduction 7

Comprised mainly of cellulose, hemicellulose, and lignin, lignocellulose is a major 8 constituent of plant cell walls and is the most abundant organic polymer on Earth (McNeil 9 et al. 1984). However, the suitability of lignocellulose as a substrate for the synthesis of 10 biobased products is limited by its resilience to enzymatic digestion (Chafe 1969, McNeil 11 et al. 1984). A specialized group of enzymes, known as carbohydrate-active enzymes 12 (CAZymes), are required for the efficient degradation of lignocellulosic biomass in order 13 to access the carbon sources within the lignocellulosic matrix (Cantarel et al. 2009, 14 Lombard et al. 2014). CAZymes are a widespread and structurally diverse set of enzymes 15 involved in the breakdown, biosynthesis or modification of lignocellulose and can be 16 produced by microorganisms that inhabit various microbiomes including the mammalian 17 gut (Lombard et al. 2014). The potent lignocellulolytic capability of CAZymes is conferred 18 by an array of enzymatic catalysts, usually grouped into families according to amino acid 19 sequence similarity (Lombard et al. 2014), which facilitate the degradation of complex 20 polymers into simple sugars. The major CAZymes that degrade carbohydrate polymers 21 are glycoside hydrolases - GHs, polysaccharide lyases - PL and polysaccharide 22 monooxygenases (Munoz-Munoz et al. 2017). GHs are the most abundant CAZymes and 23 are widely employed in biotechnological and biomedical settings (Cantarel et al. 2009). 24 Yet, the discovery of novel GHs from microbial communities has been a challenging task 25 due to the complexity and diversity of CAZyme families present in microbial habitats, 26 which result in many as-yet-uncharacterized members in these families. 27

Microbial communities are dynamic and can evolve novel CAZymes according to 28 the external environment and the nature of substrates available for metabolism (Wilkens 29

142

et al. 2017). This attribute enables us to use selective pressure to enrich the microbial 1 environment with substrates that favor the growth of microbes that express the desired 2 biocatalyst. One fascinating microbial community that has garnered much interest due to 3 its phylogenetic diversity is the one residing in the forestomach of ungulates, termed 4 rumen microbiome (Shaani et al. 2018, Russell and Rychlik 2001). This microbial 5 ecosystem is an ideal model for the discovery of novel enzymes as it is contained within 6 a unique animal host and can be easily manipulated through dietary interventions to gain 7 insight into the dynamics and functions of the microbiota (Weimer 2015). Rapid advances 8 in omics technologies has facilitated the discovery of thousands of genes encoding 9 novel CAZymes in the rumen microbiome during the past two decades (Brulc et al. 2009, 10 Hess et al. 2011, Wang et al. 2013, Svartström et al. 2017, Gharechahi and Salekdeh 11 2018, Meng Qi et al. 2011, Comtet-Marre et al. 2017, Fuyong Li and Le Luo Guan 2017). 12 However, only a handful of these candidate enzymes have entered industrial application, 13 mainly due to annotation mistakes (sequence/activity incoherence) in publicly databases 14 (Fernández-Arrojo et al. 2010, Ferrer et al. 2016). This technical issue is further 15 exacerbated by the fact that individual proteins within a protein family can have vastly 16 diverse functions (Franzosa et al. 2015, Schnoes et al. 2009, Levin et al. 2017). 17

Targeted functional profiling of a microbial community (Kaminski et al. 2015) 18 following metatranscriptomic sequencing may overcome these limitations as it allows the 19 compilation of a de novo database of marker peptides derived from reference proteins of 20 interest. This was recently demonstrated in the discovery of a novel biomarker of host- 21 microbial symbiosis in the human gut (Levin et al. 2017). However, the complexity and 22 diversity of CAZymes in microbial habitats has made targeted functional profiling of 23 CAZymes extremely difficult, and consequently there remains substantial gaps in our 24 knowledge of the functions of uncharacterized members of the CAZyme families. 25

Here, we adopted a selective pressure approach to enrich the rumen ecosystem 26 of Angus bulls fed a forage-rich diet for microbes capable of degrading lignocellulose, in 27 order to facilitate mining of the rumen metatranscriptome for metabolically active CAZyme 28 families. Next, we employed the targeted functional profiling of these families to capture 29 ecological information (abundance and distribution) of functionally distinct members of 30

143

enriched CAZyme families to guide the discovery of unknown enzymes. By screening 775 1 uncharacterized members of the GH11 family and 409 of the GH45 family, we identified 2 18 putative xylanases (GH11) and three putative endoglucanases (GH45). These 3 outcomes highlight the usefulness of this strategy for uncovering novel rumen enzymes 4 for the breakdown, biosynthesis or modification of lignocellulosic biomass. 5

5.2 Materials and methods 6

5.2.1 Animal trial and sampling 7

The experimental procedures described here were reported previously (Thompson 2015), 8 with the study protocol being reviewed and approved by the University of Manitoba animal 9 care committee. Briefly, rumen contents were collected from 12 purebred bulls (mean age 10 of 249 ± 22 days and average body weight of 313.9 ± 32 kg) raised in confinement at the 11 Glenlea Research Station (University of Manitoba) according to the guidelines of the 12 Canadian Council on Animal Care (CCAC) (CCAC 1993). In the current trial, bulls were 13 fed a forage based diet (Supplementary Table 1) throughout the experimental period (180 14 days). Then, representative samples (250 ml) of rumen contents (liquid and solid fractions) 15 were collected over four-time points (0, 80, 100, 180 d) (48 samples in total) using a 16 Geishauser oral probe (Duffield et al. 2004), immediately snap frozen in liquid nitrogen, 17 and stored at -80°C for further processing. The feed intake of individual bulls was 18 recorded using the GrowSafe® feeding system (GrowSafe Systems Ltd., Airdrie, Alberta, 19 CA) to calculate feed conversion rate (FCR), which is a ratio of dry matter intake to 20 average daily gain (computed on a biweekly basis (Montanholi et al. 2010)). The bulls 21 were then ranked into two groups: high and low FCR, with high (H-FCR) and low (L-FCR) 22 standing for inefficient and efficient cattle in terms of diet utilization, respectively. 23

5.2.2 RNA extraction and sequencing 24

Total RNA extraction and sequencing protocols were described by Fuyong Li and Le Luo 25 Guan (2017). In summary, total RNA was extracted from rumen samples using the TRIzol 26 protocol based on the acid guanidinium-phenol-chloroform method (Chomczynski and 27 Sacchi 2006, Béra-Maillet et al. 2009) with modifications reported previously (Fuyong Li 28 and Le Luo Guan 2017). Approximately 200 mg of rumen sample was subjected to RNA 29

144

extraction with the addition of 1.5 ml of TRIzol reagent (Invitrogen, Carlsbad, CA, USA), 1 followed by 0.4 ml of chloroform, 0.3 ml of isopropanol, and 0.3 ml of high salt solution 2 (1.2 M sodium acetate, 0.8 M NaCl) for the extraction protocol. Total RNA (100 ng) of 3 each sample was used for library construction using the TruSeq RNA sample prep v2 LS 4 kit (Illumina, San Diego, CA, USA) without the mRNA enrichment step. Finally, cDNA 5 fragments (∼140 bp) were paired-end (2 X 100 bp) sequenced using an Illumina HiSeq 6

2000 system. 7

5.2.3 Bioinformatic and statistical analysis 8

After checking the quality of fastq-formatted sequences using FastQC 9 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), the software Trimmomatic 10 (version 0.32; 58) was used to trim residual artificial sequences, cut bases with quality 11 scores below 20, and remove reads shorter than 50 bp (Bolger et al. 2014). Then, 12 SortMeRNA (Kopylova et al. 2012) (version 1.9) was used to sort the filtered reads into 13 fragments of mRNA for microbial classification. Prior to performing the taxonomic 14 classification, a Kraken (Wood and Salzberg 2014) standard database was built based 15 on all complete genomes of bacteria downloaded from NCBI plus complete genomes from 16 organisms isolated from the rumen or from ruminant feces or saliva deposited in the 17 Hungate1000 project (JGI's IMG database) (Neves et al. 2017). Then, each pair of mRNA 18 sequences was assembled by MEGAHIT (Li et al. 2015), with the resulting contigs being 19 assigned by Kraken (through k-mer discrimination) to the lowest common ancestor in a 20 customized standard database for microbial classification. For gene annotation, the 21 contigs were submitted to MG-RAST (Meyer et al. 2008a) where they were de-replicated 22 and quality checked using the methods described by Gomez-Alvarez et al. (Gomez- 23 Alvarez et al. 2009) and Cox et al. (Cox et al. 2010). The per-base coverage depth across 24 all contigs was calculated by mapping raw reads from each sample against the 25 assembled contigs using BBMap v35.92 with the parameters ‘kfilter=22, subfilter=15 and 26 maxindel=80’ (https://sourceforge.net/projects/bbmap/). Within MG-RAST, the contigs 27 were annotated using subsystems technology (Aziz et al. 2008) with maximum e-value of 28 10-5, minimum percent identity of 60, and minimum alignment length of 30 as the 29 parameter settings (Xu et al. 2015). Translated non-rRNA sequences from the mRNA- 30

145

enriched RNA sample were submitted to a local version of dbCAN (Yin et al. 2012) to 1 annotate sequences for the presence of CAZymes with a cutoff e-value < 10-5. 2 ShortBRED was then used to determine the abundance of distinct members of CAZymes 3 of interest in the metatranscriptomic dataset, with the sequences being grouped at a 4 specified amino acid similarity threshold of 85% identity to detect non-redundant 5 representative matches (Kaminski et al. 2015). UniRef90 (downloaded on March 2017) 6 was used as the comprehensive protein reference catalog to annotate the representative 7 short peptides (markers) within the CAZyme family (Suzek et al. 2015). The theoretical 8 atomic models of the markers quantified in the previous step were constructed using I- 9 TASSER (Zhang 2008, Yang et al. 2014). Multiple threading alignments of the markers 10 were generated by the meta-server LOMETS (Wu and Zhang 2007) to identify template 11 structures from the Protein Data Bank (PDB) library, followed by structural assembly and 12 refinement steps, with subsequent reconstruction of the atomic models (Li and Zhang 13 2009). Protein-ligand binding sites of the homology models were verified with the I- 14 TASSER associated COACH package (Yang et al. 2013). Read counts classified by 15 kraken (microbial taxonomic assignment), MG-RAST (functions) and dbCAN (CAZyme 16 families) were subjected to differential abundance analysis (L-FCR vs. H-FCR) using 17 edgeR (functions: glmFit to fit the negative binomial generalized log-linear models and 18 glmLRT to conduct likelihood ratio tests for coefficient contrasts) (Robinson et al. 2010, 19 Robinson and Smyth 2008, McCarthy et al. 2012). The trimmed mean of M-values (TMM) 20 method was used to normalize the data and minimize the log-fold change between 21 samples (Bullard et al. 2010). All P values were corrected for a false discovery rate (FDR) 22 of 0.05 using the Benjamin-Hochberg algorithm (Benjamini and Hochberg 1995), and 23 FDR-corrected P values <0.05 were considered as significant. Cladograms were 24 generated by GraPhlAn (Asnicar et al. 2015) and the heat trees were built using the 25 package metacodeR (Foster et al. 2017). All statistical procedures were done in R 3.4.2 26 (R Core Team, 2017) and Python 3.6.0. 27

5.2.4 Cloning, protein expression and purification 28

The sequence of the gene of putative xylanase 1 selected from candidates obtained in 29 the previous bioinformatic analysis was cloned in pET43.1a (ligation at Xho1 and BamH1 30

146

sites) using commercial services (Genscript, NJ, USA). The protein expression vector 1 was then transformed into chemically competent E. coli Rosetta-gami™ 2 DE3 cells 2 (Millipore, Ontario, Canada), with single colonies being grown in Luria broth medium 3 supplemented with 100 μg/μL of ampicillin (Amresco, Solon, OH) at 37°C until reaching 4

OD600 of 0.5-0.6. Overexpression was induced with 0.4 mM IPTG, and the cultures were 5 grown for 8 h at 24°C. Cells were then harvested by centrifugation at 6000 × g for 15 min 6 at 4°C. The cell pellets were resuspended in phosphate-buffer (pH 7.4), containing 1mM 7 PMSF (phenylmethylsulfonyl fluoride) and lysed using Emulsiflex (Avestin, Ottawa, 8 Canada) at a pressure of 206.8 MPa. The unbroken cells and cell debris were pelleted 9 by centrifugation at 17,000 × g for 30 min. The supernatant was then incubated with 10 subtilisin resin (Profinity exact Expression Technology, Bio-RAD, USA) for 1 h at 4oC, and 11 unbound proteins were washed away with phosphate buffer. The protein of interest was 12 eluted by incubating the resin overnight with elution buffer (pH 7.2, 0.1M Sodium 13 phosphate and 0.1M Sodium fluoride) at 4°C, and then dialyzed (Spectra/Por membrane 14 tubing, Vol/Length: 1 ml/cm) against McDougall’s buffer, pH 7.0 (McDougall 1948) and 15 concentrated using 10,000 MWCO concentrators (Millipore, USA) to 0.4-1 mg/ml. The 16 concentrated protein samples were aliquoted, flash‐frozen and stored at −80°C. Protein 17 concentration was determined by colorimetric detection and quantification of total protein 18 using the Pierce BCA protein assay kit (Thermo-Fisher Scientific) with the bovine serum 19 albumin as standard. The purified protein was then visualized by SDS-PAGE gel. 20

5.2.5 Differential scanning fluorimetry (DSF) 21

To investigate the effect of pH on protein stability, DFS assay of Xylanase 1 in different 22 buffers (100 mM Sodium acetate buffer: pH 5.0, 6.0; 100 mM Tricine buffer: pH 7.0, 8.0, 23 and 9.0; McDougall’s buffer: pH 6.0, 7.0, and 8.0) spanning the interval of pH 3–8 was 24 performed. Xylanase 1 in a final concentration of 5 µM was mixed with SyproOrange dye 25 (Thermo Fisher Scientific, USA). Prior to use, the dye stock was diluted 1:50 (100X) in 26 water and used immediately while protecting from light to reduce photobleaching. The 27 optimal dilution of dye in the assay was determined empirically with a 5X dilution for the 28 final assay. The thermal denaturation assay was performed in a total volume of 40 µl. All 29 samples were run in duplicates. The thermal scan was conducted from 25 to 95°C, at 30

147

0.5 °C/min (ViiA 7 Real-Time PCR System, ThermoFisher). The melting point (Tm) was 1 calculated by fitting the raw fluorescence data over the temperature using the Boltzmann 2 equation in GraphPad Prism program (GraphPadPrizm 7 for Windows, GraphPad 3 Software, USA). 4

5.2.6 Size-exclusion chromatography 5

The oligomeric state and homogeneity of Xylanase 1 was determined by size-exclusion 6 chromatography on Superdex 75 (10/30) column (GE Healthcare, Canada), equilibrated 7 with McDougall’s buffer, pH 6. Molar mass of the protein peak was calculated using a 8 logarithmic interpolation of elution volumes (Ve) using a gel filtration LMW calibration kit 9

(GE Healthcare, Pittsburgh,USA) containing 1) blue dextran 2000 (V0), 2) thyroglobulin 10 (670 kDa), 3) g-globulin (158 kDa), 4) ovalbumin (44 kDa), 5) myoglobulin (17 kDa), and 11 6) vitamin B12 (1.3 kDa). 12

5.2.7 Kinetic measurements 13

Xylanase 1 activity was determined by measuring the quantity of reducing sugars (xylose, 14 molecular weight: 150 g/mol) released from xylan (Beachwood xylan, Megazyme) by the 15 dinitrosalicylic acid (DNS) method (Miller 1959). Before kinetic calculations, all the 16 parameters for the assay were optimized. The minimal concentration of the enzyme that 17 produced a linear dependence of generated product with the time was chosen, as well as 18 the minimal time of reaction within the linear part of the curve. For kinetic measurements 19 Xylan was incubated at 40°C with activity buffer - McDougall’s buffer (McDougall 1948), 20 pH 6) for 10 min for equilibration and then the purified Xylanase was added and the 21 reaction was performed for 10min. The final concentration of enzyme was fixed at 0.05 22 μM, and the final concentration of Xylan varied (0, 0.88, 1.75, 3.5, 7.0, 15.0, and 30.0 23 mg/ml). The total volume of reaction was 200 μl. The samples with the same 24 concentrations of substrate but without enzyme addition were treated the same way 25 and were used as negative controls. After adding 300 μl of DNS reagent to stop the 26 reaction, the samples were boiled for 5 minutes, and then put on ice, following the 27 measurement under absorbance at 540 nm using a plate reader (SpectraMax M3). All 28 reactions were performed in duplicates. The plots of the reaction velocity against the 29

148

corresponding substrate concentration were fitted with Michaelis–Menten equation 1

(v0 = kcat[E]0[S]0/([S]0 + KM)) using GraphPad Prism program (GraphPadPrizm 7 for 2 Windows, GraphPad Software, USA). 3

5.2.8 Thermal inactivation of Xalanase 1 and thermodynamic analysis 4

Thermal inactivation assay was performed at 25, 40, 50 and 60oC. The samples, 5 containing 0.5µM of Xalanase1 in McDougall’s buffer (pH 6.0) were incubated at the 6 specified temperatures. Then, 20µl aliquot was taken out at each time point and stored 7 on ice until the activity measurements were performed as described above, using 0.05µM 8 of Xalanase 1 and 30 mg/ml of xylan. A non-heated enzyme was used as positive control 9 and its activity was taken as 100%. 10

Enzyme inactivation over time was described as a first-order reaction: 11

ln A/A0 = -kt (1) 12 where A – activity at time t, A0 – initial activity at time zero, k is inactivation rate constant 13 at the tested temperature (min-1) and t is time (min). k values were calculated from linear 14 regression analysis of the natural logarithm of residual activity versus incubation time and 15 replotted in Arrhenius plot. Activation energy (Ea) was calculated using the slope of 16 Arrhenius plot according to Eq. 2 17

ln(k) = -Ea/RT+c (2) 18 where R is the gas constant (8.314 J mol-1 K-1) and T is the absolute temperature. 19

The half-life of Xalanase 1 (t1/2 in min), defined as time after which activity is reduced to 20 one-half of its initial value was determined according to 21

t1/2 = ln (2)/k (3) 22

The D-value is the time (min) needed to reduce the initial activity to 90%. It is inversely 23 related to k-values and mathematically expressed as 24

D = ln(10)/k (4) 25

The values of Gibbs free energy (ΔGo, kJ mol-1), enthalpy (ΔHo, kJ mol-1), and entropy 26 ΔSo (J mol-1K-1) were determined as 27

149

o ΔG = -RTln(kh/kbT) (5) 1

ΔHo = Ea-RT (6) 2

ΔSo = (ΔHo- ΔGo)/T (7) 3

-34 Where h is the Plank constant (6.626 x 10 Js) and kb is the Boltzmann constant (1.38 5 x 10-23 JK-1). All experiments were performed in duplicates. 6

5.3 Results 7

5.3.1 Taxonomic composition of the rumen microbiota revealed by 8 metatranscriptomic sequencing 9

Assembly of mRNA reads resulting from total RNA sequencing of rumen contents yielded 10 on average 7,627 contigs per sample (with an average extension of 474.3 ± 26.67 bp and 11 N50 of 462.6 ± 27.99 bp), which were further classified using a taxonomic assignment 12 strategy as described in chapter 3 (Neves et al. 2017). Approximately 51% of the quality 13 filtered mRNA reads were mapped to the assembled contigs, indicating that they 14 represented a significant proportion of the metatranscriptome (Supplementary Data 1). 15 Of the 20 phyla identified in the dataset, the majority of sequences were assigned to 16 Bacteroidetes (45%), followed by Firmicutes (23%), Proteobacteria (14%), Spirochaetes 17 (5.0%), Verrucomicrobia (2.3%), Actinobacteria (2.2%), Tenericutes (2.1%), and 18 Fibrobacteres (1.4%) (Figure 5.1a; Supplementary Data 2; and Supplementary Figure 1). 19

5.3.2 Functional capability of the rumen microbiota 20

To examine the functional potential of the microbial community associated with the 21 degradation of lignocellulose, mRNA transcripts (assembled reads) were mapped against 22 the publicly available Subsystems database using MG-RAST (Meyer et al. 2008b), which 23 resulted in the detection of 1205 functions (ranging from the most detailed, level 4, to the 24 least detailed category, level 1) (Supplementary Data 3). Central carbohydrate 25 metabolism (including glycolysis/gluconeogenesis, glyoxylate cycle, pyruvate metabolism 26 and pentose phosphate pathway) and protein biosynthesis were the most abundant 27 functional categories, representing 10% and 33% of the annotated reads, respectively 28

150

(Supplementary Figure 2). In the polysaccharides- and monosaccharides-related 1 categories (level 3), the cellulosome and xylose utilization systems comprised 0.39% and 2 0.72% of the total annotated reads, respectively (Figure 5.1b). 3

Assembled metatranscriptomic contigs were also aligned against the CAZyme 4 database (Yin et al. 2012) to obtain more in-depth information regarding the carbohydrate 5 enzymes in the dataset. A total of 6904 unique alignments were recovered, dominated by 6 GHs (42.8% of the total CAZyme matches) and carbohydrate-binding modules - CBMs 7 (33.2%), followed by glycosyltransferases- GTs (9.5%), carbohydrate esterases - CEs 8 (6.5%), Dockerin (2.7%), S-layer homology domains - SLHs (2.6%), PL (2.0%), Cohesin 9 (0.5%), and auxiliary activities – AA6 (0.2%) (Figure 5.1c). 10

The rumen metatranscriptome of forage-fed bulls was notably enriched with GH 11 catalytic modules that provided 2914 hits belonging to 61 families (Supplementary Data 12 4). Cellulases (GH5, GH9, GH45, and GH48) and hemicellulases (GH8, GH10, GH11, 13 GH26, GH28, GH53) exhibited a high representation (25% of the total CAZyme matches) 14 in the metatranscriptome, indicating that they were actively involved in the degradation of 15 plant cell wall components contained in the forage diets. A wide variety of non-catalytic 16 proteins known as CBMs were highly represented (2297 hits) and were predicted to 17 partake in interactions with various substrates such as cellulose (e.g., CBM1, CBM2, 18 CBM3, CBM6, CBM13, CBM16, CBM44), xylan (e.g., CBM4, CBM22, CBM37), starch 19 (e.g., CBM20, CBM26), and chitin (e.g., CBM50) (Supplementary Data 4). Other 20 important classes of CAZymes frequently encountered in our dataset were CEs (e.g., 21 CE1, CE2, CE3, CE4, CE7, CE12) and PLs (e.g., PL1, PL9, PL11). The presence of 22 accessory modules (>400 hits in this study) commonly found in bacterial cellulosome- 23 associated structures (AAs, dockerins, and cohesins) and the SLH domain, provided 24 additional evidence of active cellulosome-mediated plant cell-wall degradation employed 25 by rumen microorganisms (Supplementary Data 4). 26

5.3.3 Fibrolytic bacteria and glycoside hydrolases are abundant in the rumen of 27 feed efficient animals 28

To facilitate the mining of active CAZyme families in the rumen metatranscriptome, we 29 compared the rumen microbial population in feed efficient cattle with those that were less 30

151

effective at digesting the same diet. To achieve this, we divided our experimental bulls 1 into two groups based on feed efficiency ranking (measured by feed conversion ratio, 2 FCR): 1) efficient or low feed conversion rate (L-FCR) and 2) inefficient or high-FCR (H- 3 FCR). The FCR of the two groups of animals was statistically divergent (P < 0.01), with 4 L-FCR bulls exhibiting a feed conversion efficiency 22% lower than H-FCR bulls 5 (Supplementary Figure 3). 6

Although there was no difference in the bacterial diversity between the feed 7 efficiency groups (Supplementary Figure 4), the data showed that fibrolytic bacteria and 8 GHs were abundant in the rumen of feed efficient animals. Of the 115 species analyzed 9 in all samples (Supplementary Data 2), Fibrobacter succinogens (a cellulose degrader 10 species (Suen et al. 2011)) was the only species affected by feed efficiency, showing a 11 nearly 0.5-log2-fold change increase (P < 0.05) in L-FCR relative to H-FCR (Figure 5.2a). 12 Our results also showed that the relative abundance of Fibrobacter succinogens and 13 other common plant cell wall degraders (Butyrivibrio proteoclasticus and 14 th Ruminiclostridium sp KB18) exhibited > 4-log2-fold change increase on the 180 day 15 relative to day 0 in L-FCR compared to H-FCR (Figure 2b). Prevotella ruminicola exhibited 16 th a 0.6-log2-fold change increase on the 180 day relative to day 0 in L-FCR compared to 17 H-FCR (Figure 5.2b). 18

Next, we examined whether the functional potential of the microbial community 19 could be linked to the feed efficiency groups, with the aim of finding genes involved in the 20 degradation of lignocellulosic biomass (Figures 5.1d-e). Of the 1205 features analyzed in 21 all samples (Supplementary Data 3), we could not confirm genes for lignocellulosic 22 biomass degradation differentiating the rumen microbiome of L-FCR from H-FCR. Instead, 23 we found ammonia assimilation functions mediated by aspartate-ammonia ligases (EC: 24 6.3.1.1) (Ricard et al. 2006) and motor organelles, which propel the rotating flagella to 25 enable bacteria to move towards favorable environments (Sowa and Berry 2008), at a 26 higher abundance in the rumen of L-FCR compared to H-FCR (Figure 5.3d). However, 27 genes involved in degradation of di- and oligosaccharides, which includes cellulose (e.g., 28 Cellobiose phosphorylase - EC: 2.4.1.20) and xylose utilization (e.g., Endo-1,4-훽- 29

152

th xylanase), exhibited a >5-log2-fold change increase in their abundance on the 180 day 1 relative to day 0 in L-FCR compared to H-FCR (Figure 5.2e; Supplementary Data 5). 2

To further investigate whether feed efficiency affected CAZymes, we then 3 analyzed all CAZymes families in our dataset in relation to the FCR groups (Figure 5.2c; 4 Supplementary Data 4). As observed in the results of the functional profiles, we did not 5 detect CAZymes differences between L-FCR and H-FCR; however, we found a consistent 6 th 1.9-log2-fold change increase in the relative abundance of GH13 on the 180 day relative 7 to day 0 in L-FCR compared to H-FCR (Figure 5.2c). GH13, the most abundant CAZyme 8 family in the present study with 6.8% of the total matches, is a well-represented CAZyme 9 family in the rumen (Hess et al. 2011) and the largest glycoside hydrolase family, acting 10 mainly on the catalysis of 훼-glucoside linkages encapsulated in starch and glycogen. 11 More importantly, we found that GH11 (endo-훽-1,4-xylanase - EC 3.2.1.8), GH45 12 (endoglucanase: EC 3.2.1.4), and CBMs connected to cellulose degradation (CBM79) 13 exhibited >2.5-log2-fold change increase in their relative abundance in L-FCR compared 14 to H-FCR on day 180 relative to day 0 (Figure 5.2c). 15

5.3.4 Targeted functional profiling of GH11 and GH45 families 16

To gain insight into functionally distinct members of GH11 (xylanases) and GH45 17 (endoglucanases), we applied ShortBRED (Kaminski et al. 2015) to screen those families 18 against a de novo protein reference database built from UniProt, and then profile their 19 abundance and distribution in the rumen metatranscriptome. By screening 775 20 uncharacterized members of the family GH11 and 409 of the family GH45, we identified 21 18 putative xylanases (GH11) and three putative endoglucanases (GH45) 22 (Supplementary Data 6). 23

In this study, bacteria and eukaryotic organisms represented the major sources 24 of the identified xylanases and endoglucanases (Figure 5.3b). While only two genes 25 were sourced by known rumen microbes such as Fibrobacter succinogens (100% 26 identity; UniProt IDs: C9RR38 and D9SBI1) and uncultured rumen ciliates (UniProt ID: 27 G5DDC1; 70.5% identity with Epidinium caudatum, Supplementary Figure 5), the vast 28 majority of enzymes matched bacteria and fungi strains found in other environments 29 (e.g., soil), indicating that they are unknown in the rumen microbiome (Figure 3b). 30

153

5.3.5 Structural analysis of Xylanase 1 through homology modelling 1

To further investigate key active sites and the tertiary conformation of xylanase 1 (the 2 most broadly distributed and abundant xylanase in our rumen metatranscriptome dataset), 3 we constructed and compared a homology model of xylanase 1 with crystal structures 4 deposited on the Protein Data Bank (PDB) using I-TASSER (Zhang 2008, Yang et al. 5 2014). Of the 33 crystal structures reported for the family GH11 in PDB and CAZyme 6 databases, the top homology model of xylanase 1 showed 57% sequence identity with 7 the crystal structure of a xylanase family 11 (PDB ID: 1h4hB) encoded by Bacillus 8 agaradhaerens (Supplementary Figure 6; Supplementary Table 2). The confidence score 9 (C-scores) of the top model for xylanase 1 was 1.67, indicating good quality of the 10 predicted homology model (C-score is typically in the range [-5, 2], where scores of higher 11 values signify a model with high confidence). Superimposing the homology model of 12 xylanase 1 onto the crystal structure of Bacillus agaradhaerens xylanase resulted in 13 global structural alignment scores (TM-scores) of 0.95 (TM-score >0.5 indicates a model 14 of correct topology) and root-mean-square deviation (RMSD) of the TM-aligned residues 15 of 0.47 Å (Supplementary Figure 6; Supplementary Table 2), confirming that the model 16 was in agreement with the crystal structure of Bacillus agaradhaerens xylanase. 17

Xylanase 1 exhibited the 훽 jelly-roll fold typical of GH xylanases, with the concave 18 antiparallel 훽 sheet being constituted of nine 훽 strands (훽2, 훽3, 훽6, 훽8, 훽9, 훽10, 훽11, 훽12, 19 훽14) and the antiparallel convex sheet comprising six 훽 strands (훽1, 훽4, 훽5, 훽7, 훽13, 20 훽15). The 훼 – helixes were found in the loops connecting 훽 strands 6 and 7 and 훽 strands 21 13 and 14 (Figure 5.3c). The structure of xylanase 1 resembles the shape of a right hand 22 (Törrönen et al. 1994) with the “fingers” at the top of the palm, comprising the loops 23 connecting 훽1 to 훽2, 훽3 to 훽4, 훽5 to 훽6, 훽14 to 훽15 and 훽7 to 훽8 and the “thumb”, 24 consisting of the loop connecting 훽11 to 훽12, at the right-hand side of the molecule 25 (Vardakou et al. 2008). Multiple functional annotations of xylanase 1 was investigated 26 with the I-TASSER associated COACH package (Yang et al. 2013), which showed the 27 ligand 1,2-Deoxy-2-Fluoro-Xylopyranose (DFX) (PDB ID: 1c5iA) (Joshi et al. 2000) 28 docked in the predicted substrate-binding cleft (Figure 5.3d). 29

154

Following the procedures described above, we also constructed homology models 1 (TM scores ≥ 0.7) for other xylanases and endoglucanases found in the rumen 2 metatranscriptome (Supplementary Figures 6 and 7; Supplementary Table 2), which 3 provides a basis for understanding the molecular functions of these enzymes and guide 4 future structure determination. 5

5.3.6 Purification and characterization of oligomeric state of Xylanase 1 6

The purification resulted in a highly pure enzyme with an expected molecular weight of 7 23 kDa (Figure 5.4a). Gel-filtration analysis revealed that Xylanase 1 was eluted as a 8 single homogeneous peak with a calculated molecular weight of 25 kDa, suggesting that 9 Xylanase 1 exists in a monomeric state in solution (Figure 5.4b). 10

5.3.7 Kinetic analysis of Xylanase 1 11

The pH curve exhibited a standard bell-shaped curve with an optimum pH of 6.0, with the 12 enzymatic activity being still high at pH 7.0 and decreasing only at pH 8.0 (Figure 5.4c). 13 The catalytic parameters for the activity of Xylanase 1 towards xylan determined by the 14 Michaelis-Menten equation were as follows: a) cleavage rate and catalytic efficiency of 15 480s-1 and 872 M-1s-1, respectively, and b) Km of 8.7 ± 0.9 mg/ml (Figure 5.4d). 16

5.3.8 Thermal stability of Xylanase 1 17

To assess favorable conditions for the thermal stability of Xylanase 1, a range of pHs and 18 different buffers were screened. As observed in Figure 5D, the melting temperatures (Tm 19 values) at pHs 6.0 and 7.0 in all buffer systems provided the most stable environment for 20 Xylanase 1 (Figure 5.5a). Additionally, the residual activity of Xylanase 1 was evaluated 21 at 25, 40, 50, and 60oC by thermal inactivation assays in a time- and temperature- 22 dependent manner (Table 1; Figures 5.5 b-c). The semi-log plots of the residual activity 23 versus heating time were linear at all temperatures studied (Figures 5.5 b-c), 24 demonstrating that inactivation of Xylanase 1 is a simple first-order monophasic process. 25

Inactivation rate constants (Kd) for each temperature were obtained from the 26 slopes of the Arrhenius equation (Figure 5.5c), with Kd values increasing ~10-fold per 27 10oC, suggesting a high degree of irreversible denaturation (Table 1). Additionally, the 28

155

results showed that Xylanase 1 was stable at 25 oC and had activity loss of only 10% after 1 1 hour of incubation (D value of 1354 min) (Table 1). At 40 oC, 35% of activity remained 2 after 1 hour of incubation, and the D value was of 127 min. The loss of activity with a D 3 value of 2.5 min was observed after incubating the enzyme at 60 oC (Table 1). 4

The thermodynamic parameters of inactivation including the Gibbs free energy 5 change (ΔG), the enthalpy change (ΔH), and the entropy change (ΔS) were also 6 assessed to understand the enzyme’s behavior at each step of the heat-induced 7 denaturation process (Table 1). The value of ΔH was 150.52 kJ mol-1 (25 oC), and this 8 result was independent of temperature, assuming that there is no change in the enzyme 9 heating capacity. The fact that ΔH values were positive indicates that enzyme inactivation 10 is an endothermic process. The results also showed that ΔG value reduced significantly 11 from 78.64 to 70.75 kJ mol-1 when the incubation temperature increased from 25 to 60 12 oC, showing that protein destabilization followed the rise in the temperature. Moreover, 13 the values of ΔS were positive meaning that there is an increase in the molecule disorder 14 during the exposure to higher temperatures and that unfolding is a rate-limiting step for 15 thermal inactivation (Table 1). 16

5.4 Discussion 17

The rumen microbiome has proven to be a valuable reservoir of microbial proteins with 18 application in the biotechnological industry, most notably those involved in the hydrolysis 19 of lignocellulosic biomass (CAZymes) (Seshadri et al. 2018). Identifying novel candidate 20 proteins from the rumen microbiome (and other host sites) has typically been performed 21 via comparison to sequence data, but this does not capture the nuanced differences in 22 the functions of individual proteins within each family of CAZymes, due to vast structural 23 diversity among members (e.g., GHs). In addition, such an approach to identify novel 24 enzymes is also time-consuming. To circumvent these limitations, the present study 25 employed and adapted existing bioinformatic tools to detect novel glycoside hydrolase 26 enzyme sequences in the rumen metatranscriptome of Angus bulls fed a forage diet. 27 Targeted functional profiling (Kaminski et al. 2015) using predictive computational 28 analyses was used to discover a broadly distributed novel ruminal GH, xylanase 1. 29 Application of this approach on a larger scale may be a critical tool for further discovery 30

156

of novel CAZymes within the rumen and other host-associated microbiomes, which may 1 have critical functions in feed digestion and host health, as well as applications in the 2 biotechnology industry. 3

While the targeted functional profiling (Kaminski et al. 2015) has been used to 4 examine uncharacterized enzymes found in the human microbiome (Levin et al. 2017), 5 these studies have not profiled CAZymes in microbial communities. Similarly, although 6 metagenomic and metatranscriptomic screening have been widely applied to discover 7 CAZymes in various microbiomes (Ferrer et al. 2005, Ferrer et al. 2016, Hess et al. 2011, 8 Comtet-Marre et al. 2017), to our knowledge the combination of selective pressure on the 9 native microbial community with the targeted functional profiling has not been applied to 10 quantitative metatranscriptomic analysis in order to accelerate the discovery of new 11 enzymes. To facilitate the identification of metabolically active CAZymes in the rumen 12 metatranscriptome, we adopted a selective pressure approach to enrich the rumen 13 ecosystem of bulls fed a forage diet for microbes capable of degrading lignocellulose. The 14 selective advantage to a microbe in the ruminal ecosystem arises in one of two ways: 1) 15 increased dietary abundance of a specific substrate (selective pressure), favoured by the 16 microbe or 2) niche specialization by a microbe (the lack of competition for the substrate 17 of choice) (Weimer 1998). In the present study, the bulls were fed with a forage (and thus 18 lignocellulose-rich) diet, and their rumens were dominated by microbes (e.g., 19 Bacteroidetes, Firmicutes, Spirochaetes, and Fibrobacteres) and CAZymes (e.g., GH5, 20 GH9, GH45, GH11) known to be responsible for digesting complex and recalcitrant plant 21 polymers. These findings reflect those of previous studies investigating the rumen 22 microbiome of forage-fed animals (Comtet-Marre et al. 2017, Dai et al. 2015, Gharechahi 23 and Salekdeh 2018, Svartström et al. 2017), and confirm that offering animals a forage- 24 rich diet selectively enriches the rumen microbiome for microbes and enzymes involved 25 in plant cell wall hydrolysis. As well as the GH families listed above, we also noted the 26 abundance of the SLH domain in our data, which is to the best of our knowledge the first 27 time its existence has been documented in the rumen metatranscriptome of cattle, having 28 been previously characterized in the camel rumen (Gharechahi and Salekdeh 2018). The 29 SLH domain is part of a large multi-enzyme complex known as cellulosome (Artzi et al. 30 2016, Bayer et al. 1998), and its presence in the bovine rumen lends further credence to 31

157

the hypothesis that feeding animals high-forage diets over a prolonged period of time 1 selectively enriches the rumen with specialized fiber-degrading enzymes. 2

In comparing the rumen microbiome among cattle divergent for feed efficiency, we 3 found that cellulose- and hemicellulose-degrading species (Fibrobacter succinogenes, 4 Butyrivibrio proteoclasticus and Ruminiclostridium sp KB18) and specific CAZymes 5 (GH13, GH11, GH45, CBM79) were more abundant in the metatranscriptome of efficient 6 cattle. This suggests that the rumen microbiome of efficient cattle can harbor increased 7 numbers of fibrolytic species and enzymes when undergoing a long-term forage feeding. 8 This is an important finding and provides more robust evidence of a causal relationship 9 between the rumen microbes and feed efficiency than has previously been reported 10 (Fuyong Li and Le Luo Guan 2017). 11

Having characterized the presence of enzyme families which we expected to be 12 enriched in the forage-fed rumen, we then applied this strategy to identify enzymes of 13 unknown functions in the GH11 and GH45 families, which were one of the most enriched 14 CAZyme families in the rumen microbiome of the feed efficient cattle. Although the 15 activities of certain GHs have been investigated in the rumen (Vardakou et al. 2008, Jones 16 et al. 2017), a large number of GH11 and GH45 members are as yet uncharacterized, 17 and are typically excluded in analyses of the rumen proteome. GH11, unlike other 18 xylanase families (e.g., GH10), represents only endo-훽-1,4-xylanases whose function is 19 to cleave 훽-1,4-xylosidic bonds between xylose monomers, whereas endo-1,4-훽- 20 glucanases of the family GH45 play a role in the hydrolysis of the 1,4-훽-D-glucan chain. 21 Although these enzymes act on lignocellulosic substrates (xylose and cellulose), they 22 exhibited different abundances and distributions in the rumen, suggesting that they may 23 perform distinct activities within the GH11 and GH45 families (Figure 5.3a). Taking this 24 ecological context into consideration, 18 previously uncharacterized xylanases (GH11) 25 and three uncharacterized endoglucanases (Gh45) were found in the rumen, especially 26 Xylanase 1 that was the most broadly distributed, widespread and abundant enzyme 27 (Figure 5.3a). Crucially, Xylanase 1 was detected in the metatranscriptome of all animals, 28 indicating that it has a core function or functions within the rumen. Kinetic analysis and 29 thermodynamic experiments confirmed that the abundant, uncharacterized GH11 is an 30

158

active and stable enzyme capable of degrading xylan, which is the most common 1 hemicellulose and considered the second most abundant biopolymer in the plant kingdom 2 (Ebringerová and Heinze 2000, Stephen 1983). 3

The homology model of xylanase 1 revealed a similar structural fold and catalytic 4 residues as reported for 1h4hB, but it showed different rearrangements in the loops that 5 sculpture the active site, which can impact substrate binding and consequently enzyme 6 activity and function. Further investigations showed that Xylanase 1 had two predicted 7 catalytic residues, E92 (catalytic nucleophile) and E182 (catalytic acid-base), located on 8 훽 strands 9 and 14 (Figures 5.3c-d). The location of these residues adopts a similar 9 conformation in GH11 xylanases and is entirely consistent with the catalytic apparatus of 10 a retaining glycoside hydrolase, which hydrolyzes glycosidic bonds by a double 11 displacement mechanism (Davies and Henrissat 1995). In Xylanase 1, the residue in 12 close spatial proximity to E182 (the catalytic acid-base) is N180, which characterize 13 enzymes that display non-acidic pH optimum (Joshi et al. 2000). Our experiments 14 confirmed that Xylanase 1 possessed a pH optimum similar to the rumen pH of forage- 15 fed cattle (~pH 6.0) (Holtshausen et al. 2013), and thus it is likely that this enzyme plays 16 a key role in the digestion of xylan in the rumen. 17

The discovery of new enzymes to break down lignocellulose into simple sugars is 18 urgently needed, as market demands for enzymes with applications in industrial 19 processes and animal feed are quickly increasing as a result of the cumulative impacts 20 of feedstocks on the environment. Unfortunately, a considerable amount of agricultural 21 residuals are underutilized due to the lack of an effective enzymatic system to degrade 22 lignocellulose and release its constituent sugars for fermentation. The current study 23 highlights the usefulness of combining selective pressure on the native rumen microbial 24 community with the targeted functional profiling performed by well-established algorithms 25 (Kaminski et al. 2015, Neves et al. 2017, Wood and Salzberg 2014, Meyer et al. 2008a, 26 Yin et al. 2012). This approach revealed not only the diversity of bacteria and genes 27 associated with efficient plant cell wall digestion but also facilitated the characterization 28 of novel CAZyme family members, which may be critical in feed degradation. Applying 29 this strategy and its underlying concepts in microbial ecology, nutrition, and bioinformatics, 30

159

we discovered previously uncharacterized hemicellulases and endoglucanases. The 1 demonstration of the xylanolytic capacity of the most abundant and conserved member 2 of the GH11 family (xylanase 1) present in the rumen metatranscriptome validates the 3 power of this strategy in discovering lignocellulolytic enzymes from microbial 4 environments. However, like any other enzyme, the structural basis of these new 5 enzymes must be investigated in more details to consolidate their status as suitable 6 candidates for the industrial enzyme market. 7

5.5 Conclusions 8

In summary, this study has demonstrated a robust pipeline for the discovery and 9 characterization of novel CAZymes in the rumen microbiome. Furthermore, we provide 10 evidence that the host feed efficiency phenotype may be mediated though evolution of 11 the rumen microbiome in response to dietary pressure. This proof-of-concept approach 12 may have many applications outside of animal agriculture, particularly in the recovery of 13 novel microbial enzymes for use in the biotechnology sector. It may be adapted to any 14 microbial environment for the discovery of CAZymes of interest, provided that the targeted 15 microbiome is easy to manipulate and facilitates enrichment for the microbes of interest. 16 It is likely that in the future, many more lignocellulolytic enzymes with high efficiency will 17 be discovered through this strategy as knowledge of the precise factors which drive 18 microbial community shifts increases. 19

5.6 Literature cited 20

Artzi, L., Bayer, E.A., and Moraïs, S. (2016). Cellulosomes: bacterial nanomachines for 21 dismantling plant polysaccharides. Nature Reviews Microbiology 15, 83. doi: 22 10.1038/nrmicro.2016.164. 23 Asnicar, F., Weingart, G., Tickle, T.L., Huttenhower, C., and Segata, N. (2015). Compact 24 graphical representation of phylogenetic data and metadata with GraPhlAn. Peerj 25 3. doi: 10.7717/peerj.1029. 26 Aziz, R.K., Bartels, D., Best, A.A., DeJongh, M., Disz, T., Edwards, R.A., et al. (2008). 27 The RAST Server: Rapid Annotations using Subsystems Technology. BMC 28 Genomics 9(1), 75. doi: 10.1186/1471-2164-9-75. 29 Bayer, E.A., Chanzy, H., Lamed, R., and Shoham, Y. (1998). Cellulose, cellulases and 30 cellulosomes. Current Opinion in Structural Biology 8(5), 548-557. doi: 31 https://doi.org/10.1016/S0959-440X(98)80143-7. 32

160

Benjamini, Y., and Hochberg, Y. (1995). CONTROLLING THE FALSE DISCOVERY 1 RATE - A PRACTICAL AND POWERFUL APPROACH TO MULTIPLE TESTING. 2 Journal of the Royal Statistical Society Series B-Methodological 57(1), 289-300. 3 Béra-Maillet, C., Mosoni, P., Kwasiborski, A., Suau, F., Ribot, Y., and Forano, E. (2009). 4 Development of a RT-qPCR method for the quantification of Fibrobacter 5 succinogenes S85 glycoside hydrolase transcripts in the rumen content of 6 gnotobiotic and conventional sheep. Journal Of Microbiological Methods 77(1), 8- 7 16. doi: 10.1016/j.mimet.2008.11.009. 8 Bolger, A.M., Lohse, M., and Usadel, B. (2014). Trimmomatic: a flexible trimmer for 9 Illumina sequence data. Bioinformatics (Oxford, England) 30(15), 2114-2120. doi: 10 10.1093/bioinformatics/btu170. 11 Brulc, J.M., Antonopoulos, D.A., Berg Miller, M.E., Wilson, M.K., Yannarell, A.C., 12 Dinsdale, E.A., et al. (2009). Gene-centric metagenomics of the fiber-adherent 13 bovine rumen microbiome reveals forage specific glycoside hydrolases. 14 Proceedings of the National Academy of Sciences 106(6), 1948-1953. doi: 15 10.1073/pnas.0806191105. 16 Bullard, J.H., Purdom, E., Hansen, K.D., and Dudoit, S. (2010). Evaluation of statistical 17 methods for normalization and differential expression in mRNA-Seq experiments. 18 BMC Bioinformatics 11(1), 94. doi: 10.1186/1471-2105-11-94. 19 Cantarel, B.L., Coutinho, P.M., Rancurel, C., Bernard, T., Lombard, V., and Henrissat, B. 20 (2009). The Carbohydrate-Active EnZymes database (CAZy): an expert resource 21 for Glycogenomics. Nucleic Acids Research 37, D233-D238. doi: 22 10.1093/nar/gkn663. 23 CCAC (1993). "Guide to the care and use of experimental steers," in Canadian Council 24 on Animal Care. 2nd ed. ed (Ottawa: Ottawa, Ontario, Canada). 25 Chafe, S.C. (1969). The fine structure of the collenchyma cell wall. Planta 90(1), 12-21. 26 doi: 10.1007/BF00389292. 27 Chomczynski, P., and Sacchi, N. (2006). The single-step method of RNA isolation by acid 28 guanidinium thiocyanate-phenol-chloroform extraction: twenty-something years 29 on. Nature Protocols (2), 581. 30 Clark, J.H., Budarin, V., Deswarte, F.E.I., Hardy, J.J.E., Kerton, F.M., Hunt, A.J., et al. 31 (2006). Green chemistry and the biorefinery: a partnership for a sustainable future. 32 Green Chemistry 8(10), 853-860. doi: 10.1039/B604483M. 33 Comtet-Marre, S., Parisot, N., Lepercq, P., Chaucheyras-Durand, F., Mosoni, P., 34 Peyretaillade, E., et al. (2017). Metatranscriptomics Reveals the Active Bacterial 35 and Eukaryotic Fibrolytic Communities in the Rumen of Dairy Cow Fed a Mixed 36 Diet. Frontiers in Microbiology 8(67). doi: 10.3389/fmicb.2017.00067. 37 Cox, M.P., Peterson, D.A., and Biggs, P.J. (2010). SolexaQA: At-a-glance quality 38 assessment of Illumina second-generation sequencing data. BMC Bioinformatics 39 11(1), 485. doi: 10.1186/1471-2105-11-485. 40 Dai, X., Tian, Y., Li, J., Su, X., Wang, X., Zhao, S., et al. (2015). Metatranscriptomic 41 Analyses of Plant Cell Wall Polysaccharide Degradation by Microorganisms in the 42 Cow Rumen. Applied and Environmental Microbiology 81(4), 1375. 43 Davies, G., and Henrissat, B. (1995). Structures and mechanisms of glycosyl hydrolases. 44 Structure 3(9), 853-859. doi: https://doi.org/10.1016/S0969-2126(01)00220-9. 45

161

Duffield, T., Plaizier, J.C., Fairfield, A., Bagg, R., Vessie, G., Dick, P., et al. (2004). 1 Comparison of techniques for measurement of rumen pH in lactating dairy cows. 2 Journal of Dairy Science 87(1), 59-66. 3 Ebringerová, A., and Heinze, T. (2000). Xylan and xylan derivatives – biopolymers with 4 valuable properties, 1. Naturally occurring xylans structures, isolation procedures 5 and properties. Macromolecular Rapid Communications 21(9), 542-556. doi: 6 doi:10.1002/1521-3927(20000601)21:9<542::AID-MARC542>3.0.CO;2-7. 7 Fernández-Arrojo, L., Guazzaroni, M.-E., López-Cortés, N., Beloqui, A., and Ferrer, M. 8 (2010). Metagenomic era for biocatalyst identification. Current Opinion in 9 Biotechnology 21(6), 725-733. doi: https://doi.org/10.1016/j.copbio.2010.09.006. 10 Ferrer, M., Golyshina, O.V., Chernikova, T.N., Khachane, A.N., Reyes-Duarte, D., 11 Santos, V.A.P.M.D., et al. (2005). Novel hydrolase diversity retrieved from a 12 metagenome library of bovine rumen microflora. Environmental Microbiology 13 7(12), 1996-2010. doi: 10.1111/j.1462-2920.2005.00920.x. 14 Ferrer, M., Martínez-Martínez, M., Bargiela, R., Streit, W.R., Golyshina, O.V., and 15 Golyshin, P.N. (2016). Estimating the success of enzyme bioprospecting through 16 metagenomics: current status and future trends. Microbial Biotechnology 9(1), 22- 17 34. doi: doi:10.1111/1751-7915.12309. 18 Foster, Z.S.L., Sharpton, T.J., and Grunwald, N.J. (2017). Metacoder: An R package for 19 visualization and manipulation of community taxonomic diversity data. Plos 20 Computational Biology 13(2). doi: 10.1371/journal.pcbi.1005404. 21 Franzosa, E.A., Hsu, T., Sirota-Madi, A., Shafquat, A., Abu-Ali, G., Morgan, X.C., et al. 22 (2015). Sequencing and beyond: integrating molecular 'omics' for 23 microbial community profiling. Nature Reviews Microbiology 13, 360. doi: 24 10.1038/nrmicro3451. 25 Gharechahi, J., and Salekdeh, G.H. (2018). A metagenomic analysis of the camel 26 rumen’s microbiome identifies the major microbes responsible for lignocellulose 27 degradation and fermentation. Biotechnology for Biofuels 11(1), 216. doi: 28 10.1186/s13068-018-1214-9. 29 Gomez-Alvarez, V., Teal, T.K., and Schmidt, T.M. (2009). Systematic artifacts in 30 metagenomes from complex microbial communities. The Isme Journal 3, 1314. 31 doi: 10.1038/ismej.2009.72. 32 Hess, M., Sczyrba, A., Egan, R., Kim, T.-W., Chokhawala, H., Schroth, G., et al. (2011). 33 Metagenomic Discovery of Biomass-Degrading Genes and Genomes from Cow 34 Rumen. Science 331(6016), 463-467. doi: 10.1126/science.1200387. 35 Holtshausen, L., Schwartzkopf-Genswein, K.S., and Beauchemin, K.A. (2013). Short 36 Communication: Ruminal pH profile and feeding behaviour of feedlot cattle 37 transitioning from a high-forage to a high-concentrate diet. Canadian Journal of 38 Animal Science 93(4), 529-533. doi: 10.4141/cjas2013-073. 39 Jones, D.R., Uddin, M.S., Gruninger, R.J., Pham, T.T.M., Thomas, D., Boraston, A.B., et 40 al. (2017). Discovery and characterization of family 39 glycoside hydrolases from 41 rumen anaerobic fungi with polyspecific activity on rare arabinosyl substrates. 42 Journal of Biological Chemistry 292(30), 12606-12620. doi: 43 10.1074/jbc.M117.789008. 44 Joshi, M.D., Sidhu, G., Pot, I., Brayer, G.D., Withers, S.G., and McIntosh, L.P. (2000). 45 Hydrogen bonding and catalysis: a novel explanation for how a single amino acid 46

162

substitution can change the ph optimum of a glycosidase11Edited by M. F. 1 Summers. Journal of Molecular Biology 299(1), 255-279. doi: 2 https://doi.org/10.1006/jmbi.2000.3722. 3 Kaminski, J., Gibson, M.K., Franzosa, E.A., Segata, N., Dantas, G., and Huttenhower, C. 4 (2015). High-Specificity Targeted Functional Profiling in Microbial Communities 5 with ShortBRED. Plos Computational Biology 11(12). doi: 6 10.1371/journal.pcbi.1004557. 7 Kopylova, E., Noé, L., and Touzet, H. (2012). SortMeRNA: fast and accurate filtering of 8 ribosomal RNAs in metatranscriptomic data. Bioinformatics (Oxford, England) 9 28(24), 3211-3217. doi: 10.1093/bioinformatics/bts611. 10 Levin, B.J., Huang, Y.Y., Peck, S.C., Wei, Y., Campo, A.M.D., Marks, J.A., et al. (2017). 11 A prominent glycyl radical enzyme in human gut microbiomes metabolizes trans- 12 4-hydroxy-L-proline. Science 355(6325), 595-+. doi: 10.1126/science.aai8386. 13 Li, D.H., Liu, C.M., Luo, R.B., Sadakane, K., and Lam, T.W. (2015). MEGAHIT: an ultra- 14 fast single-node solution for large and complex metagenomics assembly via 15 succinct de Bruijn graph. Bioinformatics 31(10), 1674-1676. doi: 16 10.1093/bioinformatics/btv033. 17 Li, F., and Guan, L.L. (2017). Metatranscriptomic profiling reveals linkages between the 18 active rumen microbiome and feed efficiency in beef cattle. Applied And 19 Environmental Microbiology. doi: 10.1128/AEM.00061-17. 20 Li, Y., and Zhang, Y. (2009). REMO: A New Protocol to Refine Full Atomic Protein Models 21 from C-alpha Traces by Optimizing Hydrogen-Bonding Networks. Proteins 76(3), 22 665-676. doi: 10.1002/prot.22380. 23 Lombard, V., Golaconda Ramulu, H., Drula, E., Coutinho, P.M., and Henrissat, B. (2014). 24 The carbohydrate-active enzymes database (CAZy) in 2013. Nucleic Acids 25 Research 42(D1), D490-D495. doi: 10.1093/nar/gkt1178. 26 McCarthy, D.J., Chen, Y., and Smyth, G.K. (2012). Differential expression analysis of 27 multifactor RNA-Seq experiments with respect to biological variation. Nucleic 28 Acids Research 40(10), 4288-4297. doi: 10.1093/nar/gks042. 29 McNeil, M., Darvill, A.G., Fry, S.C., and Albersheim, P. (1984). Structure and Function of 30 the Primary Cell Walls of Plants. Annual Review of Biochemistry 53(1), 625-663. 31 doi: 10.1146/annurev.bi.53.070184.003205. 32 Meyer, F., Paarmann, D., D'Souza, M., Olson, R., Glass, E., Kubal, M., et al. (2008a). 33 The metagenomics RAST server – a public resource for the automatic 34 phylogenetic and functional analysis of metagenomes. BMC Bioinformatics 9(1), 35 386. doi: 10.1186/1471-2105-9-386. 36 Meyer, F., Paarmann, D., D'Souza, M., Olson, R., Glass, E.M., Kubal, M., et al. (2008b). 37 The metagenomics RAST server - a public resource for the automatic phylogenetic 38 and functional analysis of metagenomes. Bmc Bioinformatics 9. doi: 39 10.1186/1471-2105-9-386. 40 Montanholi, Y.R., Swanson, K.C., Palme, R., Schenkel, F.S., McBride, B.W., Lu, D., et al. 41 (2010). Assessing feed efficiency in beef steers through feeding behavior, infrared 42 thermography and glucocorticoids. Animal 4(5), 692-701. doi: 43 10.1017/s1751731109991522. 44 Munoz-Munoz, J., Cartmell, A., Terrapon, N., Henrissat, B., and Gilbert, H.J. (2017). 45 Unusual active site location and catalytic apparatus in a glycoside hydrolase 46

163

family. Proceedings of the National Academy of Sciences. doi: 1 10.1073/pnas.1701130114. 2 Neves, A.L.A., Li, F., Ghoshal, B., McAllister, T., and Guan, L.L. (2017). Enhancing the 3 Resolution of Rumen Microbial Classification from Metatranscriptomic Data Using 4 Kraken and Mothur. Frontiers in Microbiology 8(2445). doi: 5 10.3389/fmicb.2017.02445. 6 Qi, M., Wang, P., O'Toole, N., Barboza, P.S., Ungerfeld, E., Leigh, M.B., et al. (2011). 7 Snapshot of the Eukaryotic Gene Expression in Muskoxen Rumen-A 8 Metatranscriptomic Approach. Plos One 6(5). doi: 10.1371/journal.pone.0020521. 9 Ricard, G., McEwan, N.R., Dutilh, B.E., Jouany, J.-P., Macheboeuf, D., Mitsumori, M., et 10 al. (2006). Horizontal gene transfer from Bacteria to rumen Ciliates indicates 11 adaptation to their anaerobic, carbohydrates-rich environment. BMC Genomics 7, 12 22-22. doi: 10.1186/1471-2164-7-22. 13 Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010). edgeR: a Bioconductor 14 package for differential expression analysis of digital gene expression data. 15 Bioinformatics 26(1), 139-140. doi: 10.1093/bioinformatics/btp616. 16 Robinson, M.D., and Smyth, G.K. (2008). Small-sample estimation of negative binomial 17 dispersion, with applications to SAGE data. Biostatistics 9(2), 321-332. doi: 18 10.1093/biostatistics/kxm030. 19 Russell, J.B., and Rychlik, J.L. (2001). Factors that alter rumen microbial ecology. 20 Science 292(5519), 1119-1122. doi: 10.1126/science.1058830. 21 Schnoes, A.M., Brown, S.D., Dodevski, I., and Babbitt, P.C. (2009). Annotation Error in 22 Public Databases: Misannotation of Molecular Function in Enzyme Superfamilies. 23 PLOS Computational Biology 5(12), e1000605. doi: 24 10.1371/journal.pcbi.1000605. 25 Seshadri, R., Leahy, S.C., Attwood, G.T., Teh, K.H., Lambie, S.C., Cookson, A.L., et al. 26 (2018). Cultivation and sequencing of rumen microbiome members from the 27 Hungate1000 Collection. Nature Biotechnology 36(4), 359-367. doi: 28 10.1038/nbt.4110. 29 Shaani, Y., Zehavi, T., Eyal, S., Miron, J., and Mizrahi, I. (2018). Microbiome niche 30 modification drives diurnal rumen community assembly, overpowering individual 31 variability and diet effects. The ISME Journal. doi: 10.1038/s41396-018-0203-0. 32 Sowa, Y., and Berry, R.M. (2008). Bacterial flagellar motor. Quarterly Reviews of 33 Biophysics 41(2), 103-132. doi: 10.1017/S0033583508004691. 34 Stephen, A.M. (1983). The Polysaccharides. San Diego, CA, USA: Academic. 35 Stuart, D.I., Courcelle, E., M√©toz, F., and Gouet, P. (1999). ESPript: analysis of multiple 36 sequence alignments in PostScript. Bioinformatics 15(4), 305-308. doi: 37 10.1093/bioinformatics/15.4.305. 38 Suen, G., Weimer, P.J., Stevenson, D.M., Aylward, F.O., Boyum, J., Deneke, J., et al. 39 (2011). "The Complete Genome Sequence of Fibrobacter succinogenes S85 40 Reveals a Cellulolytic and Metabolic Specialist". (United States, North America: 41 Public Library of Science). 42 Suzek, B.E., Wang, Y., Huang, H., McGarvey, P.B., Wu, C.H., and the UniProt, C. (2015). 43 UniRef clusters: a comprehensive and scalable alternative for improving sequence 44 similarity searches. Bioinformatics 31(6), 926-932. doi: 45 10.1093/bioinformatics/btu739. 46

164

Svartström, O., Alneberg, J., Terrapon, N., Lombard, V., de Bruijn, I., Malmsten, J., et al. 1 (2017). Ninety-nine de novo assembled genomes from the moose (Alces alces) 2 rumen microbiome provide new insights into microbial plant biomass degradation. 3 The ISME Journal. doi: 10.1038/ismej.2017.108. 4 Thompson, S. (2015). The effect of diet type on residual feed intake and the use of 5 infrared thermography as a method to predict efficiency in beef bulls. M.Sc., 6 University of Manitoba. 7 Törrönen, A., Harkki, A., and Rouvinen, J. (1994). Three-dimensional structure of endo- 8 1,4-beta-xylanase II from Trichoderma reesei: two conformational states in the 9 active site. The EMBO Journal 13(11), 2493-2501. 10 Vardakou, M., Dumon, C., Murray, J.W., Christakopoulos, P., Weiner, D.P., Juge, N., et 11 al. (2008). Understanding the Structural Basis for Substrate and Inhibitor 12 Recognition in Eukaryotic GH11 Xylanases. Journal of Molecular Biology 375(5), 13 1293-1305. doi: https://doi.org/10.1016/j.jmb.2007.11.007. 14 Wang, L., Hatem, A., Catalyurek, U.V., Morrison, M., and Yu, Z. (2013). Metagenomic 15 Insights into the Carbohydrate-Active Enzymes Carried by the Microorganisms 16 Adhering to Solid Digesta in the Rumen of Cows. PLoS ONE 8(11), 1-11. doi: 17 10.1371/journal.pone.0078507. 18 Weimer, P.J. (1998). Manipulating ruminal fermentation: a microbial ecological 19 perspective. Journal of Animal Science 76(12), 3114-3122. doi: 20 10.2527/1998.76123114x. 21 Weimer, P.J. (2015). Redundancy, resilience, and host specificity of the ruminal 22 microbiota: implications for engineering improved ruminal fermentations. Frontiers 23 in Microbiology 6(296). doi: 10.3389/fmicb.2015.00296. 24 Wilkens, C., Busk, P.K., Pilgaard, B., Zhang, W.-J., Nielsen, K.L., Nielsen, P.H., et al. 25 (2017). Diversity of microbial carbohydrate-active enzymes in Danish anaerobic 26 digesters fed with wastewater treatment sludge. Biotechnology for Biofuels 10(1), 27 158. doi: 10.1186/s13068-017-0840-y. 28 Wood, D.E., and Salzberg, S.L. (2014). Kraken: ultrafast metagenomic sequence 29 classification using exact alignments. Genome Biology 15(3), R46-R46. doi: 30 10.1186/gb-2014-15-3-r46. 31 Wu, S., and Zhang, Y. (2007). LOMETS: A local meta-threading-server for protein 32 structure prediction. Nucleic Acids Research 35(10), 3375-3382. doi: 33 10.1093/nar/gkm251. 34 Xu, B., Xu, W., Li, J., Dai, L., Xiong, C., Tang, X., et al. (2015). Metagenomic analysis of 35 the Rhinopithecus bieti fecal microbiome reveals a broad diversity of bacterial and 36 glycoside hydrolase profiles related to lignocellulose degradation. BMC Genomics 37 16(1), 174. doi: 10.1186/s12864-015-1378-7. 38 Yang, J., Roy, A., and Zhang, Y. (2013). Protein–ligand binding site recognition using 39 complementary binding-specific substructure comparison and sequence profile 40 alignment. Bioinformatics 29(20), 2588-2595. doi: 10.1093/bioinformatics/btt447. 41 Yang, J., Yan, R., Roy, A., Xu, D., Poisson, J., and Zhang, Y. (2014). The I-TASSER 42 Suite: protein structure and function prediction. Nature Methods 12, 7. doi: 43 10.1038/nmeth.3213. 44

165

Yin, Y.B., Mao, X.Z., Yang, J.C., Chen, X., Mao, F.L., and Xu, Y. (2012). dbCAN: a web 1 resource for automated carbohydrate-active enzyme annotation. Nucleic Acids 2 Research 40(W1), W445-W451. doi: 10.1093/nar/gks479. 3 Zhang, Y. (2008). I-TASSER server for protein 3D structure prediction. BMC 4 Bioinformatics 9(1), 40. doi: 10.1186/1471-2105-9-40. 5 6

7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

166

5.7 Tables and figures 1 2 Table 5.1. Kinetic parameters characterizing the thermal inactivation of Xylanase 1 3 ΔSo o o Temperature -1 ΔH ΔG Kd (min ) t1/2 (min) D (min) - (oC) (kJ∙mol-1) (kJ∙mol-1) (J∙mol 1∙K-1)

25 0.0017 407.73 1354.46 150.52 78.64 0.24 40 0.0180 38.51 127.92 150.40 76.58 0.24 50 0.2400 2.89 9.59 150.31 72.16 0.24 60 0.92 0.75 2.50 150.23 70.75 0.24

Kd, inactivation rate constant; t1/2, half-time (i.e., the time after which activity is reduced to 5 one-half of the initial value); D, the time required to reduce enzymatic activity to 10% of 6 its orginal value; ΔHo, activation enthalpy; ΔGo, activation free-energy barrier; ΔSo, 7 activation entropy of thermal denaturation. 8

9 10 11 12 13 14 15 16

167

A: Clostridiaceae B: Clostridium A: CE11 a C: Lachnospiraceae c B: CE1 D: Butyrivibrio C: PL3 E: Butyrivibrio proteoclasticus D: GH13 F: Ruminococcaceae E: GH10 G: Ruminiclostridium F: GH11 H: Ruminiclostridium sp KB18 G: GH45 I: Ruminococcus H: GH77 J: Ruminococcus bicirculans I: GH57 K: Ruminococcus albus J: GH32 L: Natranaerobius thermophilus K: GH43 M: Spirochaetaceae L: GH25 N: Treponema M: GH5 O: Treponema brennaborense Glycogen_metabolism N: GH3 P: Treponema succinifaciens CAZymes Edges O: GH9 Q: Flavobacteriaceae P: GH133 R: Prevotellaceae Alpha-Amylase_locus_in_Streptocococcus Q: Dockerin S: Prevotella 0.0721 R: SLH

Cellulosome e T: Prevotella fusca S: GT35 U: Prevotella dentalis g 0.3680 T: GT2

V: Prevotella enoeca t U: GT4

n V: GT5 WA:c ePtyrle-CvooAt_efelrlma ernutamtioin_itco_oBlautyrate Polysaccharides Fermentations:_MixXed: _Parceidvotella denticola e 0.8940 W: CBM37

Y: Porphyromonadaceae Lactate_utilization c X: CBM79 Z: Bacteroidaceae r Y: CBM50

e 1.6500 a: Bacteroides Z: CBM13 b: Marivirga tractuosa p a: CBM26 c: Enterobacteriaceae _ 2.6300 b: CBM20 d: Fibrobacteraceae Propionyl-CoA_to_Succinyl-CoA_Module n c: CBM44

a Acetone_Butanol_Ethanol_Syntheesi:s FibrFobeacrtmer entation d: CBM61 Organic acids e 3.8400 f: Fibrobacter succinogens e: CBM48

m f: CBM4 5.2800

Acetoin,_butanediol_metabolism Glycogen_metabolism Sugar alcohols Ethanolamine_utilization Edges Xylose_utilization Alpha-Amylase_locus_in_Streptocococcus 0.0721 b CarbohyCdellulrosaometes e g 0.3680

a L-Arabinose_utilization CO2 fixation t Acetyl-CoA_fermentation_to_Butyrate Polysaccharides Calvin-Benson_cycle n Fermentations:_Mixed_acid One-carbon Metabolism e 0.8940

Lactate_utilization c

Monosaccharides r

e 1.6500 Serine-glyoxylate_cycle

_ 2.6300

n D-Galacturonate_and_D-Glucuronate_Utilization Pyruvate_AlanineP_rSoeprinoen_yInl-tCerocoAn_vteors_iSonusccinyl-CoA_Module Nodes Acetone_Butanol_Ethanol_Synthesis Fermentation a

Organic acids e ) 3.8400 L-rhamnose_utilization Peripheral_Glucose_Catabolism_Pathways 0.0721 1.00

e ( 5.2800

g Pyruvate:ferredoxin_oxidoreductase e

Maltose_and_Maltodextrin_Utilization c 0.3680 2.61 Pyruvate_metabolism_I:_anaplerotic_reactions,_PEP a Acetoin,_butanediol_metabolism t

Sugar alcohols n Ethanolamine_utilization a

Central carbohydrate metabolism e 0.8940 7.44

d Xylose_utilization s

Di- and oligosaccharides c

u Glyoxylate_bypass o Carbohydrates e 1.6500 15.50

Entner-Doudoroff_Pathway p

A Beta-Glucoside_Metabolism CO2 fixation n

L-Arabinose_utilization _ Calvin-Benson_cycle 2.6300 26.80

e One-carbon Metabolism n

Lactose_and_Galactose_Uptake_and_Utilization v

a Monosaccharides i

e 3.8400 41.30 Serine-glyoxylate_cycle Glycolysis_and_Gluconeogenesis a Fructooligosaccharides(FOS)_and_Raffinose_Utilization Pentose_phosphate_pathway l

D-Galacturonate_and_D-Glucuronate_Utilization Pyruvate_Alanine_Serine_Interconversions R 5.2800 Nodes 59.00 L-rhamnose_utilization Peripheral_Glucose_Catabolism_Pathways 0.0721 1.00

Pyruvate:ferredoxin_oxidoreductase g 0.3680 2.61

Maltose_and_Maltodextrin_Utilization a

Pyruvate_metabolism_I:_anaplerotic_reactions,_PEP t

Central carbohydrate metabolism e 0.8940 7.44

Di- and oligosaccharides c

Glyoxylate_bypass o e 1.6500 15.50

Entner-Doudoroff_Pathway p

Beta-Glucoside_Metabolism n _ 2.6300 26.80

n Lactose_and_Galactose_Uptake_and_Utilization

Glycolysis_and_Gluconeogenesis e 3.8400 41.30 Fructooligosaccharides(FOS)_and_Raffinose_Utilization Pentose_phosphate_pathway m 5.2800 59.00 1

Figure 5.1. Taxonomic composition and microbial gene families obtained from 3 metatranscriptome data in beef cattle. a Taxonomic cladogram showing the most 4 abundant detected taxa (relative abundance ≥ 0.1% in at least half of the samples). The 5 six rings of the cladogram stand for phylum (innermost), class, order, family, genus, and 6 species (outermost), respectively. The sizes of the circles indicate the mean average 7 abundance of each taxon. b Heat tree displaying the functional capability of the rumen 8

168

microbiota. Each node represents the functional categories (up to three levels) and the 1 edges determine where each node fits in the functional hierarchy. Node diameter (and 2 colors) indicate the relative abundance of the functions at level 3. c Cladogram showing 3 the most abundant CAZymes. The sizes of the circles indicate the mean average 4 abundance of each CAZyme family. 5

169

Figure 5.2. Differentially abundant bacterial species (a-b), CAZymes (c), and microbial 5 functions (d-e) detected in the rumen microbiome of feed efficient cattle. Features were 6 significant (P < 0.05) between L-FCR and H-FCR by the trimmed mean of M-values 7 method implemented in edgeR. In the heat tree, node diameter (and colors) indicate the 8 log2fc in the functional categories at level 4. 9

170

a Endo-1,4-beta-xylanase (1)

Endo-1,4-beta-xylanase (2)

s Endo-1,4-beta-xylanase (8)

Endoglucanase (2) t

Endo-1,4-beta-xylanase (9) e

Endo-1,4-beta-xylanase (7) 5

Endo-1,4-beta-xylanase (6) H

Endo-1,4-beta-xylanase (10)

Endoglucanase (3) a

Endo-1,4-beta-xylanase (3) 1

Endo-1,4-beta-xylanase (5) G

Endo-1,4-beta-xylanase (4)

Endoglucanase (1)

Samples

b c d Endo-1,4-beta-xylanase (1)

0.30691 Xylanase 1 rumen ciliate/UniprotID:G5DDC1 b1 b2 b3 b4 b5 b6 a1 1h4hB TT TT 0.09309 1 10 20 30 40 50 Xylanase 10 uncultured fungus / UniprotID: A0A0A1CM37 xylanase1 T LT NN AS G KI DG LDYE LWKD YG N.T SM NL YG GG KF DC SW SSI NN ALFR IGKK W DC T KT WD 0.18061 1h4hB I VT DN SI G NH DG YDYE FWKD SG GSG TM IL NH GG TF SA QW NNV NN ILFR KGKK F NE T QT HQ Xylanase 2 uncultured eukaryote / UniprotID:A0A0A8LFJ4 0.12407 Xylanase 3 Penicillium brasilianum/ UniprotID:A0A0F7TQZ8

0.15156 Xylanase 7 Aspergillus flavus/UniprotID:A0A0D9MS98/A0A2P2HMK2 b7 b8 b9 b10 1h4hB TT 0.12561 Xylanase 8 Penicillium griseofulvum/UniprotID:A0A135LSW2 60 70 80 90 100 110 xylanase1 Q LG SI V V KYG V D YQPNGN SYLCVYGWT RS PL I EYYIV ESWG TWRPPG GT S RG K VTVDGGT 0.18746 Xylanase 4 Calocera cornea / UniprotID: A0A165DD01 1h4hB Q VG NM S I NYG A N FQPNGN AYLCVYGWT VD PL V AYYIV DSWG NWRPPG AT P KG T ITVDGGT 0.18884 Xylanase 9 Actinobacteria bacterium/ UniprotID: A0A0N1H0F9 0.17515 Didymozoophaga oligospora/ UniprotID: G1X0J7 b11 b12 b13 a2 b14 Xylanase 5 1h4hB TT 0.18441 120 130 140 150 160 170 uncultured eukaryote / UniprotID: A0A0A8LF36 Xylanase 6 xylanase1 YD VY VT DR INQPSI DG DT TFKQ FWSVR TE K KTSG SISV DK HF SAW T SM G L K LG LMYE A SL 1h4hB YD IY ET LR VNQPSI KG IA TFKQ YWSVR RS K RTSG TISV SN HF RAW E NL G M N MG KMYE V AL

0.34907 Endoglucanase 1 Pseudogymnoascus sp. VKM F-4515/UniprotID: A0A094CQ25 0.28777 Endoglucanase 2 Fibrobacter succinogenes/ UniprotID: C9RR38 b15 b16 0.28955 1h4hB TT Endoglucanase 3 Fibrobacter succinogenes/ UniprotID: D9SBI1 180 190 200 xylanase1 NVEGYQSSG KA A IY QN D V IGG...... 1h4hB TVEGYQSSG SA N VY SN T L RINGNPLSTI

Figure 5.3. Abundance of CAZyme members in the rumen microbiome. a Heatmap 4 showing the abundance and distribution of the 13 most abundant members of the GH11 5 and GH45 families quantified according to ShortBRED (Kaminski et al. 2015). b 6 Phylogenetic tree of xylanases and endoglucanases detected in the rumen 7 metatranscriptome generated by the neighbor-joining method. c Sequence alignment of 8 xylanase 1 with 1h4hB, with the residues involved in substrate-binding labeled with a red 9 star as shown by ESPript (Stuart et al. 1999). d Theoretical 3D structure of xylanase 1 10 bound to 1,2-Deoxy-2-Fluoro-Xylopyranose (DFX) (PDB ID: 1c5iA) generated by I- 11 TASSER (Zhang 2008, Yang et al. 2014). 12

171

a b

c d

Figure 5.4. Purification and characterization of oligomeric state, and kinetic analysis of 2 Xylanase 1. SDS-PAGE analysis (a), oligomeric state characterization (b), pH curve (c) 3 and Michaelis-Menten plot (d) of Xylanase 1. 4

172

a McDougall’s buffer

b c

Figure 5.5. Thermal stability of Xylanase 1. Tm determined by differential scanning 3 fluorimetry in different buffers (a) and residual activity (%) of Xylanase 1 at different 4 temperatures (b) with the respective Arrhenius plot (c). 5

173

Chapter 6 2

General Discussion 3

6.1 Significance of the research 4

The advance in omics technologies has led to tremendous progress in our understanding 5 of the rumen microbiome and its influence on host feed efficiency. However, significant 6 gaps remain in the literature concerning the interactions between host variation in feed 7 efficiency and the dynamics of rumen microbial populations. Channeling efforts towards 8 investigating these interactions offers an opportunity to enhance cattle productivity, since 9 the rumen microbiota plays a significant role in the host feed efficiency. While steps 10 forward have been made in this regard (Belanche et al. 2012, Fernando et al. 2010, 11 Roehe et al. 2016), little is known about the driving forces that influence the relationship 12 between the rumen microbiota and host individual variation, and how their interactive 13 effects on animal productivity contribute to the identification of cattle with improved feed 14 efficiency. To address this, Chapter 2 of this thesis examined the interactions between 15 host variation in feed efficiency and the dynamics of the rumen microbial population, 16 contributing to a broader effort to enhance the rumen function and improve cattle feed 17 efficiency. 18

An intrinsic relationship exists between feed efficiency and microbial functions in 19 the rumen, as revealed by total RNA-based techniques recently developed to investigate 20 the rumen metatranscriptome (Fuyong Li and Le Luo Guan 2017). However, this 21 methodology may be limited as total rRNA-based metatranscriptomics may 22 underestimate the abundance of lowly expressed transcripts (Li et al. 2019a). Thus, in- 23 depth analysis of the associations between rumen taxonomic profiles and feed efficiency 24 using other approaches (such as mRNA-based methods) are urgently needed in order to 25 enhance the resolution of the rumen microbial classification. Chapter 3 filled this 26 knowledge gap and assessed the rumen microbial classification from metatranscriptomic 27 data using mRNA-based methods, improving our understanding of the linkage between 28 rumen taxonomic profiles and feed efficiency. 29

174

Although the role of the rumen microbiome in the host feed efficiency has been 1 widely investigated using metatranscriptomic analysis (Fuyong Li and Le Luo Guan 2017, 2 Comtet-Marre et al. 2017), little is known regarding variations in microbial signatures that 3 could be targeted to improve feed efficiency. Chapter 4 employed statistical approaches 4 that allowed the identification of breed-specific microbial signatures by capturing within- 5 and between-individual variation observed in closely related breeds. The findings 6 presented in Chapter 4 suggested that the identification of rumen microbial signatures in 7 genetically similar breeds may benefit cattle production by providing further possibilities 8 to target specific taxa and functions and to discover enzymes that can be used to 9 maximize feed digestion in the rumen. 10

The rumen is a potent reservoir of microbial enzymes and the role of these 11 biocatalysts (CAZymes) in the feed digestion has been extensively studied using omics 12 technologies in order to discover new enzymes with applications in the biotechnology 13 industry (Hess et al. 2011, Meng Qi et al. 2011, Seshadri et al. 2018). However, the data 14 described in the literature are limited to describe CAZyme family classifications and their 15 relative abundance in the rumen, and albeit this information is necessary, it does not give 16 details on the functions of distinct members of these families and their association with 17 host feed efficiency. To address this gap, Chapter 5 used a targeted functional profiling 18 approach to identify distinct members of CAZyme families of interest and pinpointed 19 enzymes of unknown functions from their abundance, enzymatic activity, and distribution 20 in the rumen of feed efficient cattle. Overall, the data presented in this thesis contribute 21 to our fundamental understanding concerning the role of the rumen microbiome in cattle 22 feed efficiency (Chapters 2, 3 and 4) and provide opportunities to further explore the 23 potential of the rumen as a reservoir for novel enzyme discovery (Chapter 5). 24

6.2 Understanding the interactions between host variation in feed efficiency and 25 the dynamics of the rumen microbial population 26

Understanding the interactions between the host and the microbiota is crucial for 27 predicting microbial shifts and identifying individuals that are either responsive or resilient 28 to dietary changes (Bashiardes et al. 2018). Most rumen studies to date have generated 29 results that are usually qualitative (Belanche et al. 2011) or semi-quantitative (Comtet- 30

175

Marre et al. 2017), but not quantitative, which may not be sufficient to capture the dynamic 1 changes of the rumen microbiota according to the host responsiveness to the diet. 2 Defining the “baseline” of the quantified rumen microbiota in cattle undergoing dietary 3 changes (Chapter 2) revealed that animals grouped according to the magnitude of rumen 4 microbial changes reflected the individual responses to the dietary interventions. To our 5 knowledge, this is the first study that validated the relationship between feed efficiency 6 and host individual variation in the rumen microbial abundance following the dietary 7 switches. 8

The interest in the stratification of animals via the magnitude of change in baseline 9 microbiota following dietary changes is driven by previous observations of variations in 10 performance across individuals, even when maintained on the same diet (Z. P. Li et al. 11 2016, Brulc et al. 2009, Zhou et al. 2018). The approach used in Chapter 2 to group 12 animals based on the individual variation in the abundance of microbial populations 13 showed a relationship with feed efficiency (measured as FCR). The outcomes of Chapter 14 2 suggested that individual hosts exhibiting variability in bacterial abundance above or 15 below a certain cut-off (log2fc < -1 or > 1) were more efficient in terms of FCR than those 16 presenting a stable variability in the bacterial abundance. In practice, this approach is 17 likely to be more effective when the diets shift from forage to grain than from grain to 18 forage, as the former sequence of dietary treatments caused a larger variation in microbial 19 baselines between the diets. 20

Yet, the results presented in Chapter 2 are further limited by the absence of 21 microbial diversity data characterizing each group of log2fc in rumen microbes, and this 22 should be prioritized in future studies in order to get insight into changes in specific 23 genera/species that may be critical for feed digestion. Moreover, samples were only 24 available at two-time points prior to the determination of the baseline rumen microbiota, 25 and thus future investigations should include more sampling points to clearly define the 26 precise baseline rumen microbiota that is necessary to obtain the stratification of animals 27 undergoing dietary changes. One might also argue that variations in the content of dietary 28 components (e.g., NDF) could influence the effectiveness of microbial fermentation of the 29 feed, and consequently the results found in Chapter 2. In this case, it might be beneficial 30

176

to focus future research efforts towards analyzing the effects of different types of dietary 1 fibers on the inter-individual variability of the rumen microbiota in terms of feed 2 degradation, as the type of fibers affects the time of rumination and the microbial 3 fermentation of the feed (NRC 2016). Whatever the limitations might be, these data 4 showed that the dynamics of the rumen microbial population is intimately associated with 5 inter-individual variability in the baseline microbiota, and these findings could be used to 6 design better feeding strategies to enhance the rumen function and to identify cattle with 7 improved feed efficiency. 8

6.3 Understanding the rumen microbiota using mRNA-based metatranscriptomics 9

A recent study by Fuyong Li et al. (2016) developed a Mothur (Schloss et al. 2009) based 10 pipeline to assess active rumen microbiota in total RNA sequencing datasets. Despite the 11 efficacy of this pipeline to investigate linkages between the active rumen microbiome 12 (structure and function) and feed efficiency in beef cattle (Fuyong Li and Le Luo Guan 13 2017), it still remained a challenge for researchers to determine which approach (total 14 RNA or mRNA-based methods) of taxonomic classification delivered the most realistic 15 representation of the rumen microbial community. In chapter 3, we adapted the Kraken 16 pipeline (Wood and Salzberg 2014) to analyze mRNA sequences via reference genomes 17 of rumen microorganisms. The comparative analysis of Mothur and Kraken revealed that 18 both pipelines showed the rumen being dominated by Prevotella, Treponema, 19 Ruminoccocus, Fibrobacter, and Butyrivibrio, which are considered as part of a “core 20 bacterial microbiome” (Henderson et al. 2015). However, the Kraken pipeline classified 21 rumen microbes beyond the genus level when compared with Mothur, and this resulted 22 in additional information of the species and their functions in the rumen. 23

Overall, the Kraken based approach generated a higher resolution of the rumen 24 microbiota because the reference database used to annotate each microbial sequence 25 to the lowest common ancestral (Wood and Salzberg 2014) was built based on all known 26 microbial genomes present in the NCBI database (Neves et al. 2017). Although Kraken 27 enhanced the microbial classification at the species level, identification of archaea in the 28 rumen was challenging due to a lack of archaeal reference genomes for the rumen 29 microbiome. These issues highlight the importance of further strengthening the Kraken 30

177

database through the inclusion of more genomes to enable a more accurate classification. 1 Another issue the needs to be addressed is the high number of sequences unclassified 2 (65%) by Kraken, which points to a need for further research on poorly studied microbial 3 groups. Two possible causes for the high percentage of unclassified sequences were a) 4 the limited number of rumen species deposited in JGI at the time of the data analysis and 5 b) the lack of sequencing depth of the total RNA approach used in the study. Thus, there 6 is a need to include new genomes and/or metagenome assembled genomes from the 7 rumen microbiome in the customized Kraken database as well as a deeper sequence 8 depth if we use mRNA to improve the results of the data analysis in future studies.. 9 Regardless of the approach undertaken in the future, the only way for improvement is 10 through continued strengthening of the databases by including additional information of 11 whole-genome sequencing of rumen isolates, as the ability to culture rumen 12 microorganisms is still limited. 13

6.4 Associations between microbial signatures, feed efficiency and host genetics 14

Chapter 4 built on the bioinformatic pipeline developed as part of Chapter 3. Taking 15 advantage of the analytical capabilities outlined in that previous work, we determined the 16 taxonomic profiles of the rumen microbiome of Angus cattle, which are known for their 17 superior performance and beef quality (McLean and Schmutz 2009, Wolfger et al. 2016). 18 The animal effects that can impact feed efficiency include feeding behavior, energy 19 metabolism, the rumen microbiota, and the genetic background of the host 20 (Cantalapiedra-Hijar et al. 2018). Although these factors are usually interconnected with 21 each other, research questions regarding the associations between the genetic 22 background of the host with the rumen microbiota as well as with feed efficiency have 23 raised much interest in recent years (Roehe et al. 2016). One of these questions is to 24 address the current lack of information regarding the contributions of the genetic makeup 25 of the host (e.g., genetically similar breeds of Black and Red Angus) to variations in 26 specific microbial signatures that could be targeted to improve ruminal function and feed 27 efficiency. Black and Red Angus cattle were the chosen candidates to study variations in 28 microbial signatures in Chapter 4 because they are genetically similar and comparisons 29 between them with respect to the taxonomic and functional profiles could help uncover 30

178

microbial signatures related to feed efficiency of the Angus breed, which comprises a 1 large proportion of the beef cattle population in North America. 2

In Chapter 4, the functional potential of the rumen microbiota was investigated 3 through innovative statistical models (sPLS-DA and ANCOM) to account for the 4 compositional aspect of the microbiome data and to discover microbial signatures that 5 characterized the closely related breeds of Black and Red Angus. To the best of our 6 knowledge, these are the first data to provide a further understanding of the relationships 7 that exist between breed-specific microbial signatures (bacterial species and microbial 8 functions) and host phenotype (Black vs. Red Angus) at the RNA-level. Despite the close 9 genetic similarities between Red and Black Angus, Chapter 4 revealed unique bacterial 10 phylotypes and functions that differentiated the rumen microbiome of each breed. It was 11 found that rumen fibrolytic species and a diverse set of microbial gene families were more 12 abundant in Black Angus, whereas poorly-characterized bacterial species and specific 13 microbial genes (e.g., carbohydrate pathways) were more predominant in the rumen of 14 Red Angus. 15

With its limited sample size, however, this study lacked the power necessary to 16 find differences in feed efficiency (FCR) between Black and Red Angus. Thus, care 17 should be taken while interpreting the data outlined in Chapter 4 as the power of this study 18 may not support the relatively strong statement of ‘genetic effects/influences on the rumen 19 microbiome’. Power analysis showed that the minimal sample size to detect a FCR effect 20 size of 1.5 in samples collected from Black and Red Angus over four-time points, with a 21 power of 0.9 and p-value <0.5 is of 9 animals per breed (Figure S5), indicating that this 22 number of replicates needs to be included in future experiments to get more accurate 23 results. Despite these issues, it was revealed associations between microbial signatures 24 (Chitinophaga pinensis and Clostridium stercorarium) and FCR adjusted to the sampling 25 time across breeds, suggesting that these bacterial species may play a key role in the 26 feed conversion efficiency of forage-fed bulls. Further studies with the sample size 27 mentioned previously might offer a deeper insight into the microbial signatures that are 28 related to host genetics and feed efficiency. Moreover, the inclusion of other breeds might 29 be beneficial to elucidate the host genetic influence on the rumen microbiome. Finally, 30

179

the microbial signatures identified in Chapter 3 should be validated using other 1 technologies (such as qPCR) to confirm their contribution to growth performance in Angus 2 cattle reared in other geographical locations and exhibiting different feed efficiencies. In 3 this context, it was attempted to design primers to target the microbial signatures 4 identified in this study (e.g., Chitinophaga pinensis and Clostridium stercorarium). 5 However, the primers were not successful in amplifying the targeted species due to the 6 lack of representative isolates of the rumen microbiota that could be used as templates 7 to test the specificity of the primer pairs for those species. 8

6.5 Discovery of novel CAZymes in the rumen through selective pressure on the 9 native microbial population and targeted functional profiling 10

The idea of discovering CAZymes through selective pressure on the rumen native 11 microbial population and targeted functional profiling (Chapter 5) was conceptualized 12 from the relationship that exists between the rumen microbiota and host feed efficiency 13 (Chapters 2 and 4). The starting point for the development of this strategy was based on 14 the hypothesis that lignocellulolytic microbes (and their encoded enzymes) are more 15 abundant in the rumen of feed efficient cattle compared to inefficient animals fed the same 16 diet. Previous research supported that hypothesis by showing that selective pressure to 17 enrich the rumen microbial environment for the desired biocatalyst could be achieved 18 when specific substrates are included in the diet in order to favor the growth of microbes 19 specialized at expressing the enzymes of interest (Weimer 1998). This concept was 20 incorporated in Chapter 5, and the data showed that the relative abundance of bacterial 21 species was altered after the long-term forage feeding used to enrich the rumen 22 microbiome for lignocellulolytic microbes according to the groups of feed efficiency (L- 23 FCR and H-FCR). Of the 115 species analyzed, Fibrobacter succinogens, a cellulose 24 degrader species (Suen et al. 2011), was the only species affected by the feed efficiency 25 groups; it showed a nearly 0.5-log2-fold change increase (P < 0.05) in L-FCR relative to 26 H-FCR. Our results also showed that the relative abundance of Fibrobacter succinogens 27 and other common plant cell wall degraders (Butyrivibrio proteoclasticus and 28 th Ruminiclostridium sp KB18) exhibited > 4-log2-fold change increase on the 180 day 29 relative to day 0 in L-FCR compared to H-FCR, confirming the enrichment of the rumen 30

180

microbiome for lignocellulolytic microbes after the long-term forage feeding. To further 1 investigate whether feed efficiency affects CAZymes, we then analyzed all CAZymes 2 families in our dataset in relation to the FCR groups. Chapter 5 showed that GH11 (endo- 3 훽-1,4-xylanase - EC 3.2.1.8), GH45 (endoglucanase: EC 3.2.1.4), and CBMs connected 4 to cellulose degradation (CBM79) exhibited >2.5-log2-fold change increase in their relative 5 abundance in L-FCR compared to H-FCR after the long-term feeding. Taken together, 6 our results indicated that the use of a single type of fibrous diets promotes the increase 7 in the relative abundance of fibrolytic microbes and lignocellulolytic enzymes in the rumen 8 of feed efficient cattle over time. 9

An exciting aspect of the findings outlined in Chapter 5 is that they are not limited 10 in classifying CAZyme families and determining their abundance in the rumen as 11 observed in previous research (Fuyong Li and Le Luo Guan 2017, Hess et al. 2011, Meng 12 Qi et al. 2011). Instead, it showed the importance of assessing individual members of 13 CAZyme families to guide the discovery of new enzymes and get a comprehensive picture 14 of their ecological distribution in the rumen microbiome. This is particularly relevant by the 15 fact that individual proteins within a protein family can have vastly diverse functions 16 (Franzosa et al. 2015, Schnoes et al. 2009, Levin et al. 2017), and this information may 17 be ignored when the focus is only on the identification of the protein families in the 18 microbiome. I recognize that the identification of CAZyme families followed by the 19 screening of genes encoding novel CAZymes has led to the discovery of thousands of 20 putative enzymes in metagenomic or metatranscriptomic datasets (Brulc et al. 2009, 21 Hess et al. 2011, Wang et al. 2013, Svartström et al. 2017, Gharechahi and Salekdeh 22 2018, Meng Qi et al. 2011, Comtet-Marre et al. 2017, Fuyong Li and Le Luo Guan 2017), 23 but then a question arises: “why only a handful of these candidate enzymes have entered 24 industrial application?”. This is a serious issue that needs to be carefully considered by 25 the researchers interested in exploring the potential of microbiomes for enzyme discovery. 26 I came to the conclusion that the problem stems from annotation mistakes 27 (sequence/activity incoherence) in publicly databases (Fernández-Arrojo et al. 2010, 28 Ferrer et al. 2016). Actually, the literature cites many of these examples like the one 29 reported by Jiménez et al. (2012) who discovered a novel cold-tolerant esterase using 30 metagenomics approaches, but this protein was annotated in the database as a MarR 31

181

family transcriptional regulator! These outcomes indicate that database entries are not 1 sufficiently reliable. 2 In Chapter 5, the targeted functional profiling method (Kaminski et al. 2015) was 3 applied to screen members of CAZyme families against a de novo protein reference 4 database and profile their abundance and distribution in the rumen microbiome without 5 relying on the annotations of external databases. By screening 775 uncharacterized 6 members of the family GH11 and 409 of the family GH45, Chapter 5 identified 18 putative 7 xylanases (GH11) and three putative endoglucanases (GH45). Although these enzymes 8 act on lignocellulosic substrates (xylose and cellulose), they exhibited different 9 abundances and distributions in the rumen metatranscriptome, suggesting that they 10 may perform distinct activities within the GH11 and GH45 families. Later, it was 11 investigated key active sites and the tertiary conformation of GH11 and GH45 enzymes 12 and prioritized the enzymatic characterization of xylanase 1 (the most broadly distributed 13 and abundant xylanase in our rumen metatranscriptome dataset). Several experiments 14 involving cloning and protein expression and purification of xylanase 1, as well as 15 differential scanning fluorimetry, size-exclusion chromatography, kinetic measurements, 16 and thermodynamic calculations were performed to characterize that enzyme. The 17 biochemical proof of the xylanolytic activity of the most locally abundant and widespread 18 member of GH11 family using homology modeling and enzymatic assays have 19 demonstrated the usefulness of the strategy outlined in Chapter 5 and strengthened our 20 knowledge of the role played by this previously uncharacterized enzyme. However, 21 further studies are needed to elucidate the crystal structure of xylanase 1 and test the 22 enzymatic activity of all putative xylanases and endoglucanases identified in Chapter 5. 23

6.6 Implications and future directions 24

Despite the limitations described above, this thesis provides fundamental knowledge 25 concerning the role of the rumen microbiome in cattle feed efficiency and offers 26 opportunities to further explore the potential of the rumen as a source for novel enzyme 27 discovery. Chapter 2 demonstrated how the interactions between the diet and host 28 variation in the abundance of rumen microbes could be used to assess feed efficiency in 29 ruminants, and these data could be beneficial to animal husbandry in two ways. First, by 30

182

understanding the dynamics of the ruminal microbial population in response to dietary 1 changes, researchers will be able to design better feeding strategies to improve the 2 rumen function and host performance, since the microbial community plays an important 3 role in the digestion of feedstuffs (Belanche et al. 2012, Fernando et al. 2010, Roehe et 4 al. 2016). Second, our approach could serve as a cheap strategy to quickly assess rumen 5 microbial shifts in cattle populations undergoing dietary changes under field situations (e. 6 g., feedlots), and to identify feed efficient or inefficient animals under the same dietary 7 management since we found a relationship between FCR and the abundance of rumen 8 microorganisms. The animals could then be managed differently for the most profitable 9 way to the farmers. However, the data presented in Chapter 2 are limited by the lack of 10 microbial diversity information for each group of log2fc in rumen microbes, and this should 11 be prioritized in future experiments as discussed previously. 12

The studies outlined in Chapters 3 and 4 developed a pipeline to explore the 13 taxonomic and functional characteristics of the rumen microbiome and applied novel 14 statistical approaches as alternatives to the traditional analytical methods discussed in 15 Chapter 1. The pipeline described in Chapter 3 will offer opportunities to analyze 16 metatranscriptomic sequence data (~140 bp in length) of rumen samples by providing 17 further possibilities to enhance the resolution of the rumen microbial classification. 18 Chapter 4, in turn, showed results on the link between the rumen microbiome, breed 19 differences, and FCR and is of interest for the scientific community because of the 20 statistical methodologies that were applied in combination with metatranscriptomics. The 21 future directions for Chapter 3 include the inclusion of 913 bacterial and archaeal 22 genomes (Stewart et al. 2018) in the Kraken database and the development of a pipeline 23 combining the approaches of Fuyong Li et al. (2016) and Neves et al. (2017) for a more 24 inclusive and representative classification of the rumen microbiome. For Chapter 4, the 25 main point to be considered in future experiments includes improvements in the design 26 of the study to enhance the power of the analysis and the reproducibility of the results. 27

Chapter 5 developed a strategy to discover new enzymes through the application 28 of selective pressure on the rumen native population and the targeted functional profiling 29 of CAZyme families. This strategy can uncover previously unappreciated enzymes and 30

183

this might offer further possibilities for meeting increasing market demands for novel 1 biocatalysts with applications in animal feed industries. An interesting aspect to be 2 explored in the future using the strategy outlined in Chapter 5 is the implementation of a 3 feeding intervention that includes recalcitrant diets (e.g., sugar cane bagasse) at a rate 4 as high as possible without compromising the animal health. This approach will 5 encourage the evolution of microbes specialized at digesting lignocellulose, and it is likely 6 that many other lignocellulolytic enzymes with high efficiency will be discovered through 7 this strategy. In addition, such an approach may also be applied to discover new enzymes 8 with biotechnology implications such as biofuels. 9

6.7 Literature cited 10

Bashiardes, S., Godneva, A., Elinav, E., and Segal, E. (2018). Towards utilization of the 11 human genome and microbiome for personalized nutrition. Current Opinion in 12 Biotechnology 51, 57-63. doi: 10.1016/j.copbio.2017.11.013. 13 Belanche, A., Abecia, L., Holtrop, G., Guada, J.A., Castrillo, C., de la Fuente, G., et al. 14 (2011). Study of the effect of presence or absence of protozoa on rumen 15 fermentation and microbial protein contribution to the chyme1. Journal of Animal 16 Science 89(12), 4163-4174. doi: 10.2527/jas.2010-3703. 17 Belanche, A., Doreau, M., Edwards, J.E., Moorby, J.M., Pinloche, E., and Newbold, C.J. 18 (2012). Shifts in the Rumen Microbiota Due to the Type of Carbohydrate and Level 19 of Protein Ingested by Dairy Cattle Are Associated with Changes in Rumen 20 Fermentation. Journal of Nutrition 142(9), 1684-1692. doi: 10.3945/jn.112.159574. 21 Brulc, J.M., Antonopoulos, D.A., Berg Miller, M.E., Wilson, M.K., Yannarell, A.C., 22 Dinsdale, E.A., et al. (2009). Gene-centric metagenomics of the fiber-adherent 23 bovine rumen microbiome reveals forage specific glycoside hydrolases. 24 Proceedings of the National Academy of Sciences 106(6), 1948-1953. doi: 25 10.1073/pnas.0806191105. 26 Comtet-Marre, S., Parisot, N., Lepercq, P., Chaucheyras-Durand, F., Mosoni, P., 27 Peyretaillade, E., et al. (2017). Metatranscriptomics Reveals the Active Bacterial 28 and Eukaryotic Fibrolytic Communities in the Rumen of Dairy Cow Fed a Mixed 29 Diet. Frontiers in Microbiology 8(67). doi: 10.3389/fmicb.2017.00067. 30 Fernández-Arrojo, L., Guazzaroni, M.-E., López-Cortés, N., Beloqui, A., and Ferrer, M. 31 (2010). Metagenomic era for biocatalyst identification. Current Opinion in 32 Biotechnology 21(6), 725-733. doi: https://doi.org/10.1016/j.copbio.2010.09.006. 33 Fernando, S.C., Purvis, H.T., II, Najar, F.Z., Sukharnikov, L.O., Krehbiel, C.R., Nagaraja, 34 T.G., et al. (2010). Rumen Microbial Population Dynamics during Adaptation to a 35 High-Grain Diet. Applied and Environmental Microbiology 76(22), 7482-7490. doi: 36 10.1128/aem.00388-10. 37 Ferrer, M., Martínez-Martínez, M., Bargiela, R., Streit, W.R., Golyshina, O.V., and 38 Golyshin, P.N. (2016). Estimating the success of enzyme bioprospecting through 39

184

metagenomics: current status and future trends. Microbial Biotechnology 9(1), 22- 1 34. doi: doi:10.1111/1751-7915.12309. 2 Franzosa, E.A., Hsu, T., Sirota-Madi, A., Shafquat, A., Abu-Ali, G., Morgan, X.C., et al. 3 (2015). Sequencing and beyond: integrating molecular 'omics' for 4 microbial community profiling. Nature Reviews Microbiology 13, 360. doi: 5 10.1038/nrmicro3451. 6 Gharechahi, J., and Salekdeh, G.H. (2018). A metagenomic analysis of the camel 7 rumen’s microbiome identifies the major microbes responsible for lignocellulose 8 degradation and fermentation. Biotechnology for Biofuels 11(1), 216. doi: 9 10.1186/s13068-018-1214-9. 10 Henderson, G., Cox, F., Ganesh, S., Jonker, A., Young, W., Janssen, P.H., et al. (2015). 11 Rumen microbial community composition varies with diet and host, but a core 12 microbiome is found across a wide geographical range. Scientific Reports 6. doi: 13 10.1038/srep19175. 14 Hess, M., Sczyrba, A., Egan, R., Kim, T.-W., Chokhawala, H., Schroth, G., et al. (2011). 15 Metagenomic Discovery of Biomass-Degrading Genes and Genomes from Cow 16 Rumen. Science 331(6016), 463-467. doi: 10.1126/science.1200387. 17 Jiménez, D.J., Montaña, J.S., Álvarez, D., and Baena, S. (2012). A novel cold active 18 esterase derived from Colombian high Andean forest soil metagenome. World 19 Journal of Microbiology and Biotechnology 28(1), 361-370. doi: 10.1007/s11274- 20 011-0828-x. 21 Kaminski, J., Gibson, M.K., Franzosa, E.A., Segata, N., Dantas, G., and Huttenhower, C. 22 (2015). High-Specificity Targeted Functional Profiling in Microbial Communities 23 with ShortBRED. Plos Computational Biology 11(12). doi: 24 10.1371/journal.pcbi.1004557. 25 Levin, B.J., Huang, Y.Y., Peck, S.C., Wei, Y., Campo, A.M.D., Marks, J.A., et al. (2017). 26 A prominent glycyl radical enzyme in human gut microbiomes metabolizes trans- 27 4-hydroxy-L-proline. Science 355(6325), 595-+. doi: 10.1126/science.aai8386. 28 Li, F., and Guan, L.L. (2017). Metatranscriptomic profiling reveals linkages between the 29 active rumen microbiome and feed efficiency in beef cattle. Applied And 30 Environmental Microbiology. doi: 10.1128/AEM.00061-17. 31 Li, F., Henderson, G., Sun, X., Cox, F., Janssen, P.H., and Guan, L.L. (2016a). 32 Taxonomic Assessment of Rumen Microbiota Using Total RNA and Targeted 33 Amplicon Sequencing Approaches. Frontiers in Microbiology 7(987). doi: 34 10.3389/fmicb.2016.00987. 35 Li, F., Hitch, T.C.A., Chen, Y., Creevey, C.J., and Guan, L.L. (2019). Comparative 36 metagenomic and metatranscriptomic analyses reveal the breed effect on the 37 rumen microbiome and its associations with feed efficiency in beef cattle. 38 Microbiome 7(1), 6. doi: 10.1186/s40168-019-0618-5. 39 Li, Z.P., Wright, A.D.G., Si, H.Z., Wang, X.X., Qian, W.X., Zhang, Z.G., et al. (2016b). 40 Changes in the rumen microbiome and metabolites reveal the effect of host 41 genetics on hybrid crosses. Environmental Microbiology Reports 8(6), 1016-1023. 42 doi: 10.1111/1758-2229.12482. 43 McLean, K.L., and Schmutz, S.M. (2009). Associations of melanocortin 1 receptor 44 genotype with growth and carcass traits in beef cattle. Canadian Journal of Animal 45 Science 89(3), 295-300. doi: 10.4141/CJAS08094. 46

185

Neves, A.L.A., Li, F., Ghoshal, B., McAllister, T., and Guan, L.L. (2017). Enhancing the 1 Resolution of Rumen Microbial Classification from Metatranscriptomic Data Using 2 Kraken and Mothur. Frontiers in Microbiology 8(2445). doi: 3 10.3389/fmicb.2017.02445. 4 NRC (2016). Nutrient Requirements of Beef Cattle. Washington, D.C., USA: National 5 Academy Press. 6 Piro, V.C., Matschkowski, M., and Renard, B.Y. (2017). MetaMeta: integrating 7 metagenome analysis tools to improve taxonomic profiling. Microbiome 5(1), 101- 8 101. doi: 10.1186/s40168-017-0318-y. 9 Qi, M., Wang, P., O'Toole, N., Barboza, P.S., Ungerfeld, E., Leigh, M.B., et al. (2011). 10 Snapshot of the Eukaryotic Gene Expression in Muskoxen Rumen-A 11 Metatranscriptomic Approach. Plos One 6(5). doi: 10.1371/journal.pone.0020521. 12 Roehe, R., Dewhurst, R.J., Duthie, C.-A., Rooke, J.A., McKain, N., Ross, D.W., et al. 13 (2016). Bovine Host Genetic Variation Influences Rumen Microbial Methane 14 Production with Best Selection Criterion for Low Methane Emitting and Efficiently 15 Feed Converting Hosts Based on Metagenomic Gene Abundance. Plos Genetics 16 12(2). doi: 10.1371/journal.pgen.1005846. 17 Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., et al. 18 (2009). Introducing mothur: Open-Source, Platform-Independent, Community- 19 Supported Software for Describing and Comparing Microbial Communities. 20 Applied & Environmental Microbiology 75(23), 7537-7541. doi: 21 10.1128/AEM.01541-09. 22 Schnoes, A.M., Brown, S.D., Dodevski, I., and Babbitt, P.C. (2009). Annotation Error in 23 Public Databases: Misannotation of Molecular Function in Enzyme Superfamilies. 24 PLOS Computational Biology 5(12), e1000605. doi: 25 10.1371/journal.pcbi.1000605. 26 Seshadri, R., Leahy, S.C., Attwood, G.T., Teh, K.H., Lambie, S.C., Cookson, A.L., et al. 27 (2018). Cultivation and sequencing of rumen microbiome members from the 28 Hungate1000 Collection. Nature Biotechnology 36(4), 359-367. doi: 29 10.1038/nbt.4110. 30 Stewart, R.D., Warr, A., Watson, M., Auffret, M.D., Dewhurst, R.J., Roehe, R., et al. 31 (2018). Assembly of 913 microbial genomes from metagenomic sequencing of the 32 cow rumen. Nature Communications 9(1). doi: 10.1038/s41467-018-03317-6. 33 Suen, G., Weimer, P.J., Stevenson, D.M., Aylward, F.O., Boyum, J., Deneke, J., et al. 34 (2011). "The Complete Genome Sequence of Fibrobacter succinogenes S85 35 Reveals a Cellulolytic and Metabolic Specialist". (United States, North America: 36 Public Library of Science). 37 Svartström, O., Alneberg, J., Terrapon, N., Lombard, V., de Bruijn, I., Malmsten, J., et al. 38 (2017). Ninety-nine de novo assembled genomes from the moose (Alces alces) 39 rumen microbiome provide new insights into microbial plant biomass degradation. 40 The ISME Journal. doi: 10.1038/ismej.2017.108. 41 Wang, L., Hatem, A., Catalyurek, U.V., Morrison, M., and Yu, Z. (2013). Metagenomic 42 Insights into the Carbohydrate-Active Enzymes Carried by the Microorganisms 43 Adhering to Solid Digesta in the Rumen of Cows. PLoS ONE 8(11), 1-11. doi: 44 10.1371/journal.pone.0078507. 45

186

Weimer, P.J. (1998). Manipulating ruminal fermentation: a microbial ecological 1 perspective. Journal of Animal Science 76(12), 3114-3122. doi: 2 10.2527/1998.76123114x. 3 Wolfger, B., Quinn, C., Torres, G.W., Taylor, M., and Orsel, K. (2016). Comparison of 4 feeding behavior between black and red Angus feeder heifers. Canadian Journal 5 of Animal Science 96(3), 404-409. doi: 10.1139/cjas-2014-0161. 6 Wood, D.E., and Salzberg, S.L. (2014). Kraken: ultrafast metagenomic sequence 7 classification using exact alignments. Genome Biology 15(3), R46-R46. doi: 8 10.1186/gb-2014-15-3-r46. 9 Zhou, M., Peng, Y.-J., Chen, Y., Klinger, C.M., Oba, M., Liu, J.-X., et al. (2018). 10 Assessment of microbiome changes after rumen transfaunation: implications on 11 improving feed efficiency in beef cattle. Microbiome 6(1), 62-62. doi: 12 10.1186/s40168-018-0447-y. 13

187

Bibliography 1

Aitchison, J. (1982) 'The statistical analysis of compositional data', Journal of the Royal 2 Statistical Society Series B-Methodological, 44(2), 139-177. 3 4 Akin, D. E., Borneman, W. S. and Windham, W. R. (1988) 'RUMEN FUNGI - 5 MORPHOLOGICAL TYPES FROM GEORGIA CATTLE AND THE ATTACK ON 6 FORAGE CELL-WALLS', Biosystems, 21(3-4), 385-391. 7 8 Alemu, A. W., Vyas, D., Manafiazar, G., Basarab, J. A. and Beauchemin, K. A. (2017) 9 'Enteric methane emissions from low– and high–residual feed intake beef heifers 10 measured using GreenFeed and respiration chamber techniques12', Journal of 11 Animal Science, 95, 3727-3737. 12 13 Alexandratos, N. and Bruinsma, J. (2012) 'World Agriculture Towards 2030/2050: The 14 2012 Revision', FAO - Food and Agriculture Organization of the United Nations. 15 16 Anders, S. and Huber, W. (2010) 'Differential expression analysis for sequence count 17 data', Genome Biology, 11(10), R106. 18 19 Anikaev, A., Isaev, A., Korobeinikova, A., Garber, M. and Gongadze, G. (2016) 'Role of 20 protein L25 and its contact with protein L16 in maintaining the active state of 21 Escherichia coli ribosomes in vivo', Biochemistry (00062979), 81(1), 19-27. 22 23 AOAC (2000) Official Methods of Analysis (17th), Gaithersburg, MD: Association of 24 Official Analytical Chemists. 25 26 Archer, J. A., Richardson, E. C., Herd, R. M. and Arthur, P. F. (1999) 'Potential for 27 selection to improve efficiency of feed use in beef cattle: a review', Australian 28 Journal of Agricultural Research, 50(2), 147-162. 29 30 Artzi, L., Bayer, E. A. and Moraïs, S. (2016) 'Cellulosomes: bacterial nanomachines for 31 dismantling plant polysaccharides', Nature Reviews Microbiology, 15, 83. 32 33 Asnicar, F., Weingart, G., Tickle, T. L., Huttenhower, C. and Segata, N. (2015) 'Compact 34 graphical representation of phylogenetic data and metadata with GraPhlAn', 35 PeerJ, 3. 36 37 Aziz, R. K., Bartels, D., Best, A. A., DeJongh, M., Disz, T., Edwards, R. A., Formsma, K., 38 Gerdes, S., Glass, E. M., Kubal, M., Meyer, F., Olsen, G. J., Olson, R., Osterman, 39 A. L., Overbeek, R. A., McNeil, L. K., Paarmann, D., Paczian, T., Parrello, B., 40 Pusch, G. D., Reich, C., Stevens, R., Vassieva, O., Vonstein, V., Wilke, A. and 41 Zagnitko, O. (2008) 'The RAST Server: Rapid Annotations using Subsystems 42 Technology', BMC Genomics, 9(1), 75. 43 44

188

Bader, J., Mast-Gerlach, E., Popovic, M. K., Bajpai, R. and Stahl, U. (2010) 'Relevance 1 of microbial coculture fermentations in biotechnology', Journal of Applied 2 Microbiology, 109(2), 371-387. 3 4 Bae, H. D., McAllister, T. A., Kokko, E. G., Leggett, F. L., Yanke, L. J., Jakober, K. D., Ha, 5 J. K., Shin, H. T. and Cheng, K. J. (1997) 'Effect of silica on the colonization of rice 6 straw by ruminal bacteria', Animal Feed Science and Technology, 65(1), 165-181. 7 8 Ban, Y., Jiang, H. and An, L. (2015) '- Investigating microbial co-occurrence patterns 9 based on metagenomic compositional data', - 31(- 20), - 3329. 10 11 Bashiardes, S., Godneva, A., Elinav, E. and Segal, E. (2018) 'Towards utilization of the 12 human genome and microbiome for personalized nutrition', Current Opinion in 13 Biotechnology, 51, 57-63. 14 15 Bates, D., Mächler, M., Bolker, B. and Walker, S. (2015) 'Fitting Linear Mixed-Effects 16 Models Using lme4', 2015, 67(1), 48. 17 18 Bayer, E. A., Chanzy, H., Lamed, R. and Shoham, Y. (1998) 'Cellulose, cellulases and 19 cellulosomes', Current Opinion in Structural Biology, 8(5), 548-557. 20 21 Bayley, D. P., Correia, J. D., Jarrell, K. F. and Thomas, N. A. (1999) 'Recent Excitement 22 about the Archaea: The Archaea are valuable for studying basic biological 23 questions and have novel biotechnology applications', BioScience, 49(7), 530-541. 24 25 Beg, Q., Kapoor, M., Mahajan, L. and Hoondal, G. (2001) 'Microbial xylanases and their 26 industrial applications: a review', Applied Microbiology and Biotechnology, 56(3), 27 326-338. 28 29 Belanche, A., Abecia, L., Holtrop, G., Guada, J. A., Castrillo, C., de la Fuente, G. and 30 Balcells, J. (2011) 'Study of the effect of presence or absence of protozoa on 31 rumen fermentation and microbial protein contribution to the chyme1', Journal of 32 Animal Science, 89(12), 4163-4174. 33 34 Belanche, A., Doreau, M., Edwards, J. E., Moorby, J. M., Pinloche, E. and Newbold, C. 35 J. (2012) 'Shifts in the Rumen Microbiota Due to the Type of Carbohydrate and 36 Level of Protein Ingested by Dairy Cattle Are Associated with Changes in Rumen 37 Fermentation', Journal of Nutrition, 142(9), 1684-1692. 38 39 Benjamini, Y. and Hochberg, Y. (1995) 'Controlling the false discovery rate - a practical 40 and powerful approach to multiple testing', Journal of the Royal Statistical Society 41 Series B-Methodological, 57(1), 289-300. 42 43 Benndorf, D., Vogt, C., Jehmlich, N., Schmidt, Y., Thomas, H., Woffendin, G., 44 Shevchenko, A., Richnow, H.-H. and von Bergen, M. (2009) 'Improving protein 45 extraction and separation methods for investigating the metaproteome of 46

189

anaerobic benzene communities within sediments', Biodegradation, 20(6), 737- 1 750. 2 3 Béra-Maillet, C., Mosoni, P., Kwasiborski, A., Suau, F., Ribot, Y. and Forano, E. (2009) 4 'Development of a RT-qPCR method for the quantification of Fibrobacter 5 succinogenes S85 glycoside hydrolase transcripts in the rumen content of 6 gnotobiotic and conventional sheep', Journal Of Microbiological Methods, 77(1), 8- 7 16. 8 9 Berry, D. P. and Crowley, J. J. (2013) 'CELL BIOLOGY SYMPOSIUM: Genetics of feed 10 efficiency in dairy and beef cattle1', Journal of Animal Science, 91(4), 1594-1613. 11 12 Bhat, M. K. and Bhat, S. (1997) 'Cellulose degrading enzymes and their potential 13 industrial applications', Biotechnology Advances, 15(3), 583-620. 14 15 Bhatia, L., Johri, S. and Ahmad, R. (2012) 'An economic and ecological perspective of 16 ethanol production from renewable agro waste: a review', Amb Express, 2. 17 18 Blazewicz, S. J., Barnard, R. L., Daly, R. A. and Firestone, M. K. (2013) 'Evaluating rRNA 19 as an indicator of microbial activity in environmental communities: limitations and 20 uses', The ISME Journal, 7, 2061. 21 22 Bolger, A. M., Lohse, M. and Usadel, B. (2014) 'Trimmomatic: a flexible trimmer for 23 Illumina sequence data', Bioinformatics (Oxford, England), 30(15), 2114-2120. 24 25 Borneman, W. S., Hartley, R. D., Morrison, W. H., Akin, D. E. and Ljungdahl, L. G. (1990) 26 'Feruloyl and p-coumaroyl esterase from anaerobic fungi in relation to plant cell 27 wall degradation', Applied Microbiology and Biotechnology, 33(3), 345-351. 28 29 Brooker, J. D., O'Donovan, L. A., Skene, I., Clarke, K., Blackall, L. and Muslera, P. (1994) 30 'Streptococcus caprinus sp.nov., a tannin-resistant ruminal bacterium from feral 31 goats', Letters in Applied Microbiology, 18(6), 313-318. 32 33 Brulc, J. M., Antonopoulos, D. A., Berg Miller, M. E., Wilson, M. K., Yannarell, A. C., 34 Dinsdale, E. A., Edwards, R. E., Frank, E. D., Emerson, J. B., Wacklin, P., 35 Coutinho, P. M., Henrissat, B., Nelson, K. E. and White, B. A. (2009) 'Gene-centric 36 metagenomics of the fiber-adherent bovine rumen microbiome reveals forage 37 specific glycoside hydrolases', Proceedings of the National Academy of Sciences, 38 106(6), 1948-1953. 39 40 Bryant, M. P. and Burkey, L. A. (1953) 'Cultural Methods and Some Characteristics of 41 Some of the More Numerous Groups of Bacteria in the Bovine Rumen', Journal of 42 Dairy Science, 36(3), 205-217. 43 44

190

Bryant, M. P. and Robinson, I. M. (1961) 'An Improved Nonselective Culture Medium for 1 Ruminal Bacteria and Its Use in Determining Diurnal Variation in Numbers of 2 Bacteria in the Rumen', Journal of Dairy Science, 44(8), 1446-1456. 3 4 Bullard, J. H., Purdom, E., Hansen, K. D. and Dudoit, S. (2010) 'Evaluation of statistical 5 methods for normalization and differential expression in mRNA-Seq experiments', 6 Bmc Bioinformatics, 11(1), 94. 7 8 Burnham, K. P., Anderson, D. R. and Burnham, K. P. (2002) Model selection and 9 multimodel inference : a practical information-theoretic approach, New York : 10 Springer, c2002. 11 2nd ed. 12 13 Callahan, B. J., McMurdie, P. J., Rosen, M. J., Han, A. W., Johnson, A. J. A. and Holmes, 14 S. P. (2016) 'DADA2: High-resolution sample inference from Illumina amplicon 15 data', Nature Methods, 13, 581. 16 17 Cammack, K. M., Austin, K. J., Lamberson, W. R., Conant, G. C. and Cunningham, H. C. 18 (2018) 'Tiny but mighty: The role of the rumen microbes in livestock production', J 19 Anim Sci. 20 21 Cantalapiedra-Hijar, G., Abo-Ismail, M., Carstens, G. E., Guan, L. L., Hegarty, R., Kenny, 22 D. A., McGee, M., Plastow, G., Relling, A. and Ortigues-Marty, I. (2018) 'Review: 23 Biological determinants of between-animal variation in feed efficiency of growing 24 beef cattle', Animal, 12(s2), s321-s335. 25 26 Cantarel, B. L., Coutinho, P. M., Rancurel, C., Bernard, T., Lombard, V. and Henrissat, 27 B. (2009) 'The Carbohydrate-Active EnZymes database (CAZy): an expert 28 resource for Glycogenomics', Nucleic Acids Research, 37, D233-D238. 29 30 Cao, K. A. L., Costello, M. E., Lakis, V. A., Bartolo, F., Chua, X. Y., Brazeilles, R. and 31 Rondeau, P. (2016) 'MixMC: A multivariate statistical framework to gain insight into 32 microbial communities', Plos One, 11(8). 33 34 Caporaso, J. G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F. D., Costello, E. 35 K., Fierer, N., Peña, A. G., Goodrich, J. K., Gordon, J. I., Huttley, G. A., Kelley, S. 36 T., Knights, D., Koenig, J. E., Ley, R. E., Lozupone, C. A., McDonald, D., Muegge, 37 B. D., Pirrung, M., Reeder, J., Sevinsky, J. R., Turnbaugh, P. J., Walters, W. A., 38 Widmann, J., Yatsunenko, T., Zaneveld, J. and Knight, R. (2010) 'QIIME allows 39 analysis of high-throughput community sequencing data', Nature Methods, 7(5), 40 335-336. 41 42 Carberry, C. A., Kenny, D. A., Han, S., McCabe, M. S. and Waters, S. M. (2012a) 'Effect 43 of phenotypic residual feed intake and dietary forage content on the rumen 44 microbial community of beef cattle', Appl Environ Microbiol, 78(14), 4949-58. 45 46

191

Carberry, C. A., Kenny, D. A., Kelly, A. K. and Waters, S. M. (2014a) 'Quantitative analysis 1 of ruminal methanogenic microbial populations in beef cattle divergent in 2 phenotypic residual feed intake (RFI) offered contrasting diets', J Anim Sci 3 Biotechnol, 5(1), 41. 4 5 Carberry, C. A., Kenny, D. A., Sukkyan, H., McCabe, M. S. and Waters, S. M. (2012b) 6 'Effect of phenotypic residual feed intake and dietary forage content on the rumen 7 microbial community of beef cattle', Applied and Environmental Microbiology, 8 78(14), 4949-4958. 9 10 Carberry, C. A., Waters, S. M., Kenny, D. A. and Creevey, C. J. (2014b) 'Rumen 11 methanogenic genotypes differ in abundance according to host residual feed 12 intake phenotype and diet type', Appl Environ Microbiol, 80(2), 586-94. 13 14 CCAC (1993) 'Guide to the care and use of experimental steers' in Canadian Council on 15 Animal Care, 2nd ed. ed., Ottawa: Ottawa, Ontario, Canada. 16 17 Chafe, S. C. (1969) 'The fine structure of the collenchyma cell wall', Planta, 90(1), 12-21. 18 19 Cheng, K. J., McCowan, R. P. and Costerton, J. W. (1979) 'Adherent epithelial bacteria 20 in ruminants and their roles in digestive tract function', Am J Clin Nutr, 32(1), 139- 21 48. 22 23 Chomczynski, P. and Sacchi, N. (2006) 'The single-step method of RNA isolation by acid 24 guanidinium thiocyanate-phenol-chloroform extraction: twenty-something years 25 on', Nature Protocols, (2), 581. 26 27 Christopherson, M. R., Dawson, J. A., Stevenson, D. M., Cunningham, A. C., 28 Bramhacharya, S., Weimer, P. J., Kendziorski, C. and Suen, G. (2014) 'Unique 29 aspects of fiber degradation by the ruminal ethanologen Ruminococcus albus 7 30 revealed by physiological and transcriptomic analysis', BMC Genomics, 15. 31 32 Ciucci, S., Ge, Y., Durán, C., Palladini, A., Martínez-Sánchez, L. M., Cannistraci, C. V., 33 Gerl, M. J., Jiménez-Jiménez, V., Wang, Y., Sales, S., Shevchenko, A., 34 Androutsellis-Theotokis, A., Poser, S. W., Herbig, M., Otto, O. and Guck, J. (2017) 35 'Enlightening discriminative network functional modules behind principal 36 component analysis separation in differential-omic science studies', Scientific 37 Reports, 7. 38 39 Clark, J. H., Budarin, V., Deswarte, F. E. I., Hardy, J. J. E., Kerton, F. M., Hunt, A. J., 40 Luque, R., Macquarrie, D. J., Milkowski, K., Rodriguez, A., Samuel, O., Tavener, 41 S. J., White, R. J. and Wilson, A. J. (2006) 'Green chemistry and the biorefinery: a 42 partnership for a sustainable future', Green Chemistry, 8(10), 853-860. 43 44 Cole, J. R., Wang, Q., Fish, J. A., Chai, B., McGarrell, D. M., Sun, Y., Brown, C. T., Porras- 45 Alfaro, A., Kuske, C. R. and Tiedje, J. M. (2014) 'Ribosomal Database Project: data 46

192

and tools for high throughput rRNA analysis', Nucleic Acids Research, 42(D1), 1 D633-D642. 2 3 Comtet-Marre, S., Chaucheyras-Durand, F., Bouzid, O., Mosoni, P., Bayat, A. R., Peyret, 4 P. and Forano, E. (2018) 'FibroChip, a Functional DNA Microarray to Monitor 5 Cellulolytic and Hemicellulolytic Activities of Rumen Microbiota', Frontiers in 6 Microbiology, 9(215). 7 8 Comtet-Marre, S., Parisot, N., Lepercq, P., Chaucheyras-Durand, F., Mosoni, P., 9 Peyretaillade, E., Bayat, A. R., Shingfield, K. J., Peyret, P. and Forano, E. (2017) 10 'Metatranscriptomics Reveals the Active Bacterial and Eukaryotic Fibrolytic 11 Communities in the Rumen of Dairy Cow Fed a Mixed Diet', Frontiers in 12 Microbiology, 8(67). 13 14 Cook, G. M., Rainey, F. A., Chen, G., Stackebrandt, E. and Russell, J. B. (1994) 15 'Emendation of the description of Acidaminococcus fermentans, a trans- aconitate- 16 and citrate-oxidizing bacterium', International Journal Of Systematic Bacteriology, 17 44(3), 576-578. 18 19 Cox, M. P., Peterson, D. A. and Biggs, P. J. (2010) 'SolexaQA: At-a-glance quality 20 assessment of Illumina second-generation sequencing data', Bmc Bioinformatics, 21 11(1), 485. 22 23 Creevey, C. J., Kelly, W. J., Henderson, G. and Leahy, S. C. (2014) 'Determining the 24 culturability of the rumen bacterial microbiome', Microbial Biotechnology, 7(5), 467- 25 479. 26 27 Crews, D. H., Jr., Shannon, N. H., Genswein, B. M. A., Crews, R. E., Johnson, C. M. and 28 Kendrick, B. A (2005) 'Genetic parameters for net feed efficiency of beef cattle 29 measured during postweaning growing versus finishing periods', Proc. West Sec. 30 Am. Soc. Anim. Sci, 54(54), 125-128. 31 32 Dai, X., Tian, Y., Li, J., Su, X., Wang, X., Zhao, S., Liu, L., Luo, Y., Liu, D., Zheng, H., 33 Wang, J., Dong, Z., Hu, S. and Huang, L. (2015) 'Metatranscriptomic Analyses of 34 Plant Cell Wall Polysaccharide Degradation by Microorganisms in the Cow 35 Rumen', Applied and Environmental Microbiology, 81(4), 1375. 36 37 Darling, A. E., Jospin, G., Lowe, E., Matsen, F. A. I. V., Bik, H. M. and Eisen, J. A. (2014) 38 'PhyloSift: phylogenetic analysis of genomes and metagenomes', PeerJ, 2, e243. 39 40 Dassa, B., Borovok, I., Ruimy-Israeli, V., Lamed, R., Flint, H. J., Duncan, S. H., Henrissat, 41 B., Coutinho, P., Morrison, M., Mosoni, P., Yeoman, C. J., White, B. A. and Bayer, 42 E. A. (2014) 'Rumen Cellulosomics: Divergent Fiber-Degrading Strategies 43 Revealed by Comparative Genome-Wide Analysis of Six Ruminococcal Strains', 44 Plos One, 9(7), e99221. 45 46

193

Davids, M., Hugenholtz, F., Martins dos Santos, V., Smidt, H., Kleerebezem, M. and 1 Schaap, P. J. (2016) 'Functional Profiling of Unfamiliar Microbial Communities 2 Using a Validated De Novo Assembly Metatranscriptome Pipeline', Plos One, 3 11(1), e0146423. 4 5 Davies, G. and Henrissat, B. (1995) 'Structures and mechanisms of glycosyl hydrolases', 6 Structure, 3(9), 853-859. 7 8 Denman, S. E., Martinez Fernandez, G., Shinkai, T., Mitsumori, M. and McSweeney, C. 9 S. (2015) 'Metagenomic analysis of the rumen microbial community following 10 inhibition of methane formation by a halogenated methane analog', Frontiers in 11 Microbiology, 6(1087). 12 13 Denman, S. E. and McSweeney, C. S. (2006) 'Development of a real-time PCR assay for 14 monitoring anaerobic fungal and cellulolytic bacterial populations within the 15 rumen', Fems Microbiology Ecology, 58(3), 572-582. 16 17 Denman, S. E., Morgavi, D. P. and McSweeney, C. S. (2018) 'Review: The application of 18 omics to rumen microbiota function', Animal, 12(s2), s233-s245. 19 20 Deppenmeier, U. (2002) 'The unique biochemistry of methanogenesis', Prog Nucleic Acid 21 Res Mol Biol, 71, 223-83. 22 23 DeSantis, T. Z., Hugenholtz, P., Larsen, N., Rojas, M., Brodie, E. L., Keller, K., Huber, T., 24 Dalevi, D., Hu, P. and Andersen, G. L. (2006) 'Greengenes, a chimera-checked 25 16S rRNA gene database and workbench compatible with ARB', Applied and 26 Environmental Microbiology, 72(7), 5069-5072. 27 28 Deusch, S., Camarinha-Silva, A., Conrad, J., Beifuss, U., Rodehutscord, M. and Seifert, 29 J. (2017) 'A Structural and Functional Elucidation of the Rumen Microbiome 30 Influenced by Various Diets and Microenvironments', Frontiers in Microbiology, 31 8(1605). 32 33 Deusch, S., Tilocca, B., Camarinha-Silva, A. and Seifert, J. (2015) 'News in livestock 34 research — use of Omics-technologies to study the microbiota in the 35 gastrointestinal tract of farm animals', Computational and Structural Biotechnology 36 Journal, 13, 55-63. 37 38 Distaso, M. A., Tran, H., Ferrer, M. and Golyshin, P. N. (2017) 'Metagenomic Mining of 39 Enzyme Diversity' in Lee, S. Y., ed. Consequences of Microbial Interactions with 40 Hydrocarbons, Oils, and Lipids: Production of Fuels and Chemicals, Cham: 41 Springer International Publishing, 245-269. 42 43 Doi, R. H. and Kosugi, A. (2004) 'Cellulosomes: plant-cell-wall-degrading enzyme 44 complexes', Nature Reviews Microbiology, 2, 541. 45 46

194

Dollhofer, V., Podmirseg, S. M., Callaghan, T. M., Griffith, G. W. and Fliegerova, K. (2015) 1 'Anaerobic Fungi and Their Potential for Biogas Production', Advances in 2 biochemical engineering/biotechnology, 151, 41-61. 3 4 Duffield, T., Plaizier, J. C., Fairfield, A., Bagg, R., Vessie, G., Dick, P., Wilson, J., Aramini, 5 J. and McBride, B. (2004) 'Comparison of techniques for measurement of rumen 6 pH in lactating dairy cows', Journal of Dairy Science, 87(1), 59-66. 7 8 Ebringerová, A. and Heinze, T. (2000) 'Xylan and xylan derivatives – biopolymers with 9 valuable properties, 1. Naturally occurring xylans structures, isolation procedures 10 and properties', Macromolecular Rapid Communications, 21(9), 542-556. 11 12 El Enshasy, H. A., Kandiyil, S. K., Malek, R. and Othman, N. Z. (2016) 'Microbial 13 Xylanases: Sources, Types, and Their Applications' in Gupta, V. K., ed. Microbial 14 Enzymes in Bioconversions of Biomass, Cham: Springer International Publishing, 15 151-213. 16 17 Fan, J., Han, F. and Liu, H. (2014) 'Challenges of Big Data analysis', National Science 18 Review, 1(2), 293-314. 19 20 Fernandes, A. D., Macklaim, J. M., Linn, T. G., Reid, G. and Gloor, G. B. (2013) 'ANOVA- 21 Like Differential Expression (ALDEx) Analysis for Mixed Population RNA-Seq', 22 Plos One, 8(7), e67019. 23 24 Fernandes, A. D., Reid, J. N. S., Macklaim, J. M., McMurrough, T. A., Edgell, D. R. and 25 Gloor, G. B. (2014) 'Unifying the analysis of high-throughput sequencing datasets: 26 characterizing RNA-seq, 16S rRNA gene sequencing and selective growth 27 experiments by compositional data analysis', Microbiome, 2(1), 15. 28 29 Fernández-Arrojo, L., Guazzaroni, M.-E., López-Cortés, N., Beloqui, A. and Ferrer, M. 30 (2010) 'Metagenomic era for biocatalyst identification', Current Opinion in 31 Biotechnology, 21(6), 725-733. 32 33 Fernando, S. C., Purvis, H. T., II, Najar, F. Z., Sukharnikov, L. O., Krehbiel, C. R., 34 Nagaraja, T. G., Roe, B. A. and DeSilva, U. (2010) 'Rumen Microbial Population 35 Dynamics during Adaptation to a High-Grain Diet', Applied and Environmental 36 Microbiology, 76(22), 7482-7490. 37 38 Ferrer, M., Beloqui, A., Timmis, K. N. and Golyshin, P. N. (2009) 'Metagenomics for 39 Mining New Genetic Resources of Microbial Communities', Journal of Molecular 40 Microbiology and Biotechnology, 16(1-2), 109-123. 41 42 Ferrer, M., Golyshina, O. V., Chernikova, T. N., Khachane, A. N., Reyes-Duarte, D., 43 Santos, V. A. P. M. D., Strompl, C., Elborough, K., Jarvis, G., Neef, A., Yakimov, 44 M. M., Timmis, K. N. and Golyshin, P. N. (2005) 'Novel hydrolase diversity retrieved 45

195

from a metagenome library of bovine rumen microflora', Environmental 1 Microbiology, 7(12), 1996-2010. 2 3 Ferrer, M., Martínez-Martínez, M., Bargiela, R., Streit, W. R., Golyshina, O. V. and 4 Golyshin, P. N. (2016) 'Estimating the success of enzyme bioprospecting through 5 metagenomics: current status and future trends', Microbial Biotechnology, 9(1), 22- 6 34. 7 8 Finlay, B. J., Esteban, G., Clarke, K. J., Williams, A. G., Embley, T. M. and Hirt, R. P. 9 (1994) 'Some rumen ciliates have endosymbiotic methanogens', Fems 10 Microbiology Letters, 117(2), 157-161. 11 12 Firkins, J. L. and Yu, Z. (2015) 'RUMINANT NUTRITION SYMPOSIUM: How to use data 13 on the rumen microbiome to improve our understanding of ruminant nutrition', J 14 Anim Sci, 93(4), 1450-70. 15 16 Ford, C. W., Elliott, R. and Maynard, P. J. (1987) 'THE EFFECT OF CHLORITE 17 DELIGNIFICATION ON DIGESTIBILITY OF SOME GRASS FORAGES AND ON 18 INTAKE AND RUMEN MICROBIAL ACTIVITY IN SHEEP FED BARLEY STRAW', 19 Journal of Agricultural Science, 108, 129-136. 20 21 Foster, Z. S. L., Sharpton, T. J. and Grunwald, N. J. (2017) 'Metacoder: An R package for 22 visualization and manipulation of community taxonomic diversity data', Plos 23 Computational Biology, 13(2). 24 25 Fouts, D. E., Szpakowski, S., Purushe, J., Torralba, M., Waterman, R. C., MacNeil, M. 26 D., Alexander, L. J. and Nelson, K. E. (2012) 'Next generation sequencing to define 27 prokaryotic and fungal diversity in the bovine rumen', PLoS One, 7(11), e48289. 28 29 Franzosa, E. A., Hsu, T., Sirota-Madi, A., Shafquat, A., Abu-Ali, G., Morgan, X. C. and 30 Huttenhower, C. (2015) 'Sequencing and beyond: integrating molecular 31 'omics' for microbial community profiling', Nature Reviews Microbiology, 32 13, 360. 33 34 Franzosa, E. A., Morgan, X. C., Segata, N., Waldron, L., Reyes, J., Earl, A. M., 35 Giannoukos, G., Boylan, M. R., Ciulla, D., Gevers, D., Izard, J., Garrett, W. S., 36 Chan, A. T. and Huttenhower, C. (2014) 'Relating the metatranscriptome and 37 metagenome of the human gut', Proceedings of the National Academy of 38 Sciences, 111(22), E2329-E2338. 39 40 Gao, A., Wang, H., Yang, J. and Shi, C. (2013) 'The effects of elimination of fungi on 41 microbial population and fiber degradation in sheep rumen' in Tang, X., Zhong, W., 42 Zhuang, D., Li, C. and Liu, Y., eds., Progress in Environmental Protection and 43 Processing of Resource, Pts 1-4, 224-231. 44 45

196

Geishauser, T. (1993) 'An Instrument for collection and transfer of ruminal fluid and for 1 administration of water soluble drugs in adult cattle', The Bovine Practitioner, (27), 2 38-42. 3 4 Gharechahi, J. and Salekdeh, G. H. (2018) 'A metagenomic analysis of the camel rumen’s 5 microbiome identifies the major microbes responsible for lignocellulose 6 degradation and fermentation', Biotechnology for Biofuels, 11(1), 216. 7 8 Gijzen, H. J., Lubberding, H. J., Gerhardus, M. J. T. and Vogels, G. D. (1988) 'Contribution 9 of rumen protozoa to fibre degradation and cellulase activity in vitro', FEMS 10 Microbiol Lett, 53(1), 35-43. 11 12 Gloor, G., Wu, J. R., Pawlowsky-Glahn, V. and Jose Egozcue, J. (2016) '- It's all relative: 13 analyzing microbiome data as compositions', - 26(- 5), - 329. 14 15 Gloor, G. B., Macklaim, J. M., Pawlowsky-Glahn, V. and Egozcue, J. J. (2017) 16 'Microbiome Datasets Are Compositional: And This Is Not Optional', Frontiers in 17 Microbiology, 8(2224). 18 19 Gloor, G. B. and Reid, G. (2016) 'Compositional analysis: a valid approach to analyze 20 microbiome high-throughput sequencing data', Canadian Journal of Microbiology, 21 62(8), 692-703. 22 23 Goel, G., Puniya, A. K., Aguilar, C. N. and Singh, K. (2005) 'Interaction of gut microflora 24 with tannins in feeds', Naturwissenschaften, 92(11), 497-503. 25 26 Gomez-Alvarez, V., Teal, T. K. and Schmidt, T. M. (2009) 'Systematic artifacts in 27 metagenomes from complex microbial communities', The ISME Journal, 3, 1314. 28 29 Gonzalez, I., Dejean, S., Martin, P. G. P. and Baccini, A. (2008) 'CCA: An R package to 30 extend canonical correlation analysis', Journal of Statistical Software, 23(12), 1- 31 14. 32 33 Goodrich, J. K., Davenport, E. R., Clark, A. G. and Ley, R. E. (2017) 'The Relationship 34 Between the Human Genome and Microbiome Comes into View', Annual Review 35 of Genetics, 51(1), 413-433. 36 37 Gordon, G. L. R. and Phillips, M. W. (1992) 'Extracellular pectin lyase produced by 38 Neocallimastix sp. LM1, a rumen anaerobic fungus', Letters in Applied 39 Microbiology, 15(3), 113-115. 40 41 Gordon, G. L. R. and Phillips, M. W. (1993) 'REMOVAL OF ANAEROBIC FUNGI FROM 42 THE RUMEN OF SHEEP BY CHEMICAL TREATMENT AND THE EFFECT ON 43 FEED CONSUMPTION AND IN-VIVO FIBER DIGESTION', Letters in Applied 44 Microbiology, 17(5), 220-223. 45 46

197

Gordon, G. L. R. and Phillips, M. W. (1998) 'The role of anaerobic gut fungi in ruminants', 1 Nutrition Research Reviews, 11(1), 133-168. 2 3 Gottschalk, G. (1986) 'Bacterial metabolism', 4 5 Greve, L. C., Labavitch, J. M. and Hungate, R. E. (1984) 'alpha-L-arabinofuranosidase 6 from Ruminococcus albus 8: purification and possible role in hydrolysis of alfalfa 7 cell wall', Applied and Environmental Microbiology, 47(5), 1135-1140. 8 9 Grubb, J. A. and Dehority, B. A. (1975) 'EFFECTS OF AN ABRUPT CHANGE IN RATION 10 FROM ALL ROUGHAGE TO HIGH CONCENTRATE UPON RUMEN MICROBIAL 11 NUMBERS IN SHEEP', Applied Microbiology, 30(3), 404-412. 12 13 Gruninger, R. J., Puniya, A. K., Callaghan, T. M., Edwards, J. E., Youssef, N., Dagar, S. 14 S., Fliegerova, K., Griffith, G. W., Forster, R., Tsang, A., McAllister, T. and 15 Elshahed, M. S. (2014) 'Anaerobic fungi (phylum Neocallimastigomycota): 16 advances in understanding their taxonomy, life cycle, ecology, role and 17 biotechnological potential', Fems Microbiology Ecology, 90(1), 1-17. 18 19 Guan, L. L., Nkrumah, J. D., Basarab, J. A. and Moore, S. S. (2008) 'Linkage of microbial 20 ecology to phenotype: correlation of rumen microbial ecology to cattle's feed 21 efficiency', Fems Microbiology Letters, 288(1), 85-91. 22 23 Gunsett, F. C. (1984) 'Linear Index Selection to Improve Traits Defined as Ratios', Journal 24 of Animal Science, 59(5), 1185-1193. 25 26 Gupta, A., Kumar, S., Prasoodanan, V. P. K., Harish, K., Sharma, A. K. and Sharma, V. 27 K. (2016a) 'Reconstruction of Bacterial and Viral Genomes from Multiple 28 Metagenomes', Front Microbiol, 7, 469. 29 30 Gupta, A., Kumar, S., Prasoodanan, V. P. K., Harish, K., Sharma, A. K. and Sharma, V. 31 K. (2016b) 'Reconstruction of Bacterial and Viral Genomes from Multiple 32 Metagenomes', Frontiers in Microbiology, 7(469). 33 34 Hackmann, T. J. and Firkins, J. L. (2015) 'Maximizing efficiency of rumen microbial protein 35 production', Frontiers in Microbiology, 6. 36 37 Hackmann, T. J., Tao, J., Ngugi, D. K. and Firkins, J. L. (2017) 'Genomes of rumen 38 bacteria encode atypical pathways for fermenting hexoses to short-chain fatty 39 acids', Environmental Microbiology, 19(11), 4670-4683. 40 41 Hall, M. B. (2009) 'Determination of Starch, Including Maltooligosaccharides, in Animal 42 Feeds: Comparison of Methods and a Method Recommended for AOAC 43 Collaborative Study', Journal of Aoac International, 92(1), 42-49. 44 45

198

Handelsman, J., Rondon, M. R., Brady, S. F., Clardy, J. and Goodman, R. M. (1998) 1 'Molecular biological access to the chemistry of unknown soil microbes: a new 2 frontier for natural products', Chemistry & Biology, 5(10), R245-R249. 3 4 Hart, E. H., Creevey, C. J., Hitch, T. and Kingston-Smith, A. H. (2018) 'Meta-proteomics 5 of rumen microbiota indicates niche compartmentalisation and functional 6 dominance in a limited number of metabolic pathways between abundant bacteria', 7 Scientific Reports, 8(1), 10504. 8 9 Healey, G. R., Murphy, R., Brough, L., Butts, C. A. and Coad, J. (2017) 'Interindividual 10 variability in gut microbiota and host response to dietary interventions', Nutrition 11 Reviews, 75(12), 1059-1080. 12 13 Hegarty, R. S., Goopy, J. P., Herd, R. M. and McCorkell, B. (2007) 'Cattle selected for 14 lower residual feed intake have reduced daily methane production', J Anim Sci, 15 85(6), 1479-86. 16 17 Henderson, G., Cox, F., Ganesh, S., Jonker, A., Young, W. and Janssen, P. H. (2015) 18 'Rumen microbial community composition varies with diet and host, but a core 19 microbiome is found across a wide geographical range', Sci Rep, 5, 14567. 20 21 Henderson, G., Cox, F., Ganesh, S., Jonker, A., Young, W., Janssen, P. H. and Global 22 Rumen Census, C. (2015) 'Rumen microbial community composition varies with 23 diet and host, but a core microbiome is found across a wide geographical range', 24 Scientific Reports, 6. 25 26 Hernandez-Sanabria, E., Goonewardene, L. A., Wang, Z., Durunna, O. N., Moore, S. S. 27 and Guan, L. L. (2012) 'Impact of Feed Efficiency and Diet on Adaptive Variations 28 in the Bacterial Community in the Rumen Fluid of Cattle', Applied and 29 Environmental Microbiology, 78(4), 1203-1214. 30 31 Hess, M., Sczyrba, A., Egan, R., Kim, T.-W., Chokhawala, H., Schroth, G., Luo, S., Clark, 32 D. S., Chen, F., Zhang, T., Mackie, R. I., Pennacchio, L. A., Tringe, S. G., Visel, 33 A., Woyke, T., Wang, Z. and Rubin, E. M. (2011) 'Metagenomic Discovery of 34 Biomass-Degrading Genes and Genomes from Cow Rumen', Science, 331(6016), 35 463-467. 36 37 Himmel, M. E., Ding, S.-Y., Johnson, D. K., Adney, W. S., Nimlos, M. R., Brady, J. W. 38 and Foust, T. D. (2007) 'Biomass Recalcitrance: Engineering Plants and Enzymes 39 for Biofuels Production', Science, (5813). 40 41 Ho, Y. W., Abdullah, N. and Jalaludin, S. (1988) 'PENETRATING STRUCTURES OF 42 ANAEROBIC RUMEN FUNGI IN CATTLE AND SWAMP BUFFALO', Journal of 43 General Microbiology, 134(1), 177-182. 44 45

199

Ho, Y. W. and Barr, D. J. S. (1995) 'CLASSIFICATION OF ANAEROBIC GUT FUNGI 1 FROM HERBIVORES WITH EMPHASIS ON RUMEN FUNGI FROM MALAYSIA', 2 Mycologia, 87(5), 655-677. 3 4 Hobson, P. N. and Stewart, C. S. (1997) The Rumen Microbial Ecosystem, London: 5 Blackie Academic & Professional. 6 7 Hogg, D., Pell, G., Dupree, P., Goubet, F., Martín-Orúe, S. M., Armand, S. and Gilbert, 8 H. J. (2003) 'The modular architecture of Cellvibrio japonicus mannanases in 9 glycoside hydrolase families 5 and 26 points to differences in their role in mannan 10 degradation', The Biochemical journal, 371(Pt 3), 1027-1043. 11 12 Holtshausen, L., Schwartzkopf-Genswein, K. S. and Beauchemin, K. A. (2013) 'Short 13 Communication: Ruminal pH profile and feeding behaviour of feedlot cattle 14 transitioning from a high-forage to a high-concentrate diet', Canadian Journal of 15 Animal Science, 93(4), 529-533. 16 17 Hristov, A. N., Callaway, T. R., Lee, C. and Dowd, S. E. (2012) 'Rumen bacterial, 18 archaeal, and fungal diversity of dairy cows in response to ingestion of lauric or 19 myristic acid', J Anim Sci, 90(12), 4449-57. 20 21 Hristov, A. N., Oh, J., Firkins, J. L., Dijkstra, J., Kebreab, E., Waghorn, G., Makkar, H. P. 22 S., Adesogan, A. T., Yang, W., Lee, C., Gerber, P. J., Henderson, B. and Tricarico, 23 J. M. (2013) 'SPECIAL TOPICS — Mitigation of methane and nitrous oxide 24 emissions from animal operations: I. A review of enteric methane mitigation 25 options', Journal of Animal Science, 91, 5045-5069. 26 27 Hristov, A. N., Oh, J., Giallongo, F., Frederick, T. W., Harper, M. T., Weeks, H. L., Branco, 28 A. F., Moate, P. J., Deighton, M. H., Williams, S. R. O., Kindermann, M. and Duval, 29 S. (2015) 'An inhibitor persistently decreased enteric methane emission from dairy 30 cows with no negative effect on milk production', Proceedings of the National 31 Academy of Sciences, 112(34), 10663-10668. 32 33 Hungate, R. E. (1966) The rumen and its microbes, New York and London: Academic 34 Press. 35 36 Hungate, R. E. (1967) 'Hydrogen as an intermediate in the rumen fermentation', Archiv 37 für Mikrobiologie, 59(1-3), 158-164. 38 39 Huws, S. A., Creevey, C. J., Oyama, L. B., Mizrahi, I., Denman, S. E., Popova, M., Muñoz- 40 Tamayo, R., Forano, E., Waters, S. M., Hess, M., Tapio, I., Smidt, H., Krizsan, S. 41 J., Yáñez-Ruiz, D. R., Belanche, A., Guan, L., Gruninger, R. J., McAllister, T. A., 42 Newbold, C. J., Roehe, R., Dewhurst, R. J., Snelling, T. J., Watson, M., Suen, G., 43 Hart, E. H., Kingston-Smith, A. H., Scollan, N. D., do Prado, R. M., Pilau, E. J., 44 Mantovani, H. C., Attwood, G. T., Edwards, J. E., McEwan, N. R., Morrisson, S., 45 Mayorga, O. L., Elliott, C. and Morgavi, D. P. (2018) 'Addressing Global Ruminant 46

200

Agricultural Challenges Through Understanding the Rumen Microbiome: Past, 1 Present, and Future', Frontiers in Microbiology, 9(2161). 2 3 Imelfort, M., Parks, D., Woodcroft, B. J., Dennis, P., Hugenholtz, P. and Tyson, G. W. 4 (2014) 'GroopM: an automated tool for the recovery of population genomes from 5 related metagenomes', PeerJ, 2, e603. 6 7 IPCC (2006) Guidelines for National Greenhouse Gas Inventories, Agriculture, forestry 8 and other land use, Japan: Institute for Global Environmental Strategies (IGES). 9 10 Jami, E., Israel, A., Kotser, A. and Mizrahi, I. (2013) 'Exploring the bovine rumen bacterial 11 community from birth to adulthood', Isme j, 7(6), 1069-79. 12 13 Jami, E. and Mizrahi, I. (2012) 'Composition and similarity of bovine rumen microbiota 14 across individual animals', PLoS One, 7(3), e33306. 15 16 Janssen, P. H. (2010) 'Influence of hydrogen on rumen methane formation and 17 fermentation balances through microbial growth kinetics and fermentation 18 thermodynamics', Animal Feed Science and Technology, 160(1-2), 1-22. 19 20 Janssen, P. H. and Kirs, M. (2008) 'Structure of the archaeal community of the rumen', 21 Appl Environ Microbiol, 74(12), 3619-25. 22 23 Janssen, P. H. and Kirs, M. (2008) 'Structure of the Archaeal Community of the Rumen', 24 Applied & Environmental Microbiology, 74(12), 3619-3625. 25 26 Jayanegara, A., Sarwono, K. A., Kondo, M., Matsui, H., Ridla, M., Laconi, E. B. and 27 Nahrowi (2018) 'Use of 3-nitrooxypropanol as feed additive for mitigating enteric 28 methane emissions from ruminants: a meta-analysis AU - Jayanegara, Anuraga', 29 Italian Journal of Animal Science, 17(3), 650-656. 30 31 Jiménez, D. J., Montaña, J. S., Álvarez, D. and Baena, S. (2012) 'A novel cold active 32 esterase derived from Colombian high Andean forest soil metagenome', World 33 Journal of Microbiology and Biotechnology, 28(1), 361-370. 34 35 Jin, D., Zhao, S., Zheng, N., Bu, D., Beckers, Y., Denman, S. E., McSweeney, C. S. and 36 Wang, J. (2017) 'Differences in Ureolytic Bacterial Composition between the 37 Rumen Digesta and Rumen Wall Based on ureC Gene Classification', Frontiers in 38 Microbiology, 8(385). 39 40 Johnson, D. E., Ferrell, C. L. and Jenkins, T. G. (2003) 'The history of energetic efficiency 41 research: Where have we been and where are we going?', Journal of Animal 42 Science, 81(13_suppl_1), E27-E38. 43 44 Johnson, K. A. and Johnson, D. E. (1995) 'Methane emissions from cattle', Journal of 45 Animal Science, 73(8), 2483-2492. 46

201

1 Jones, D. R., Uddin, M. S., Gruninger, R. J., Pham, T. T. M., Thomas, D., Boraston, A. 2 B., Briggs, J., Pluvinage, B., McAllister, T. A., Forster, R. J., Tsang, A., Selinger, 3 L. B. and Abbott, D. W. (2017) 'Discovery and characterization of family 39 4 glycoside hydrolases from rumen anaerobic fungi with polyspecific activity on rare 5 arabinosyl substrates', Journal of Biological Chemistry, 292(30), 12606-12620. 6 7 Joshi, M. D., Sidhu, G., Pot, I., Brayer, G. D., Withers, S. G. and McIntosh, L. P. (2000) 8 'Hydrogen bonding and catalysis: a novel explanation for how a single amino acid 9 substitution can change the ph optimum of a glycosidase11Edited by M. F. 10 Summers', Journal Of Molecular Biology, 299(1), 255-279. 11 12 Jouany, J. P. (1995) 'Rumen microbial metabolism and ruminant digestion: J-P. Jouany 13 (Ed). 1991. INRA, 147 rue de l'Université, 75338 Paris cedex 07, France. ISBN 2 14 7380 0345 1. 374 pp. Text in English. 350 FF', Livestock Production Science, 15 44(3), 284. 16 17 Kabel, M. A., Yeoman, C. J., Han, Y., Dodd, D., Abbas, C. A., de Bont, J. A. M., Morrison, 18 M., Cann, I. K. O. and Mackie, R. I. (2011) 'Biochemical Characterization and 19 Relative Expression Levels of Multiple Carbohydrate Esterases of the Xylanolytic 20 Rumen Bacterium', Applied and Environmental Microbiology, 77(16), 5671-5681. 21 22 Kaminski, J., Gibson, M. K., Franzosa, E. A., Segata, N., Dantas, G. and Huttenhower, 23 C. (2015) 'High-Specificity Targeted Functional Profiling in Microbial Communities 24 with ShortBRED', Plos Computational Biology, 11(12). 25 26 Kamke, J., Kittelmann, S., Soni, P., Li, Y., Tavendale, M., Ganesh, S., Janssen, P. H., 27 Shi, W., Froula, J., Rubin, E. M. and Attwood, G. T. (2016) 'Rumen metagenome 28 and metatranscriptome analyses of low methane yield sheep reveals a Sharpea- 29 enriched microbiome characterised by lactic acid formation and utilisation', 30 Microbiome, 4(1), 56-56. 31 32 Kanehisa, M., Goto, S., Kawashima, S., Okuno, Y. and Hattori, M. (2004) 'The KEGG 33 resource for deciphering the genome', Nucleic Acids Research, 32(Database 34 issue), D277-D280. 35 36 Kang, D. D., Froula, J., Egan, R. and Wang, Z. (2015) 'MetaBAT, an efficient tool for 37 accurately reconstructing single genomes from complex microbial communities', 38 PeerJ, 3, e1165. 39 40 Kenters, N., Henderson, G., Jeyanathan, J., Kittelmann, S. and Janssen, P. H. (2011) 41 'Isolation of previously uncultured rumen bacteria by dilution to extinction using a 42 new liquid culture medium', Journal Of Microbiological Methods, 84(1), 52-60. 43 44 Kim, M., Yu, Z. and Morrison, M. (2011) 'Status of the phylogenetic diversity census of 45 ruminal microbiomes', Fems Microbiology Ecology, 76(1), 49-63. 46

202

1 Kittelmann, S., Devente, S. R., Kirk, M. R., Seedorf, H., Dehority, B. A. and Janssen, P. 2 H. (2015) 'Phylogeny of Intestinal Ciliates, Including <span class="named- 3 content genus-species" id="named-content-1">Charonina 4 ventriculi</span>, and Comparison of Microscopy and 18S rRNA Gene 5 Pyrosequencing for Rumen Ciliate Community Structure Analysis', Applied and 6 Environmental Microbiology, 81(7), 2433. 7 8 Kittelmann, S., Devente, S. R., Kirk, M. R., Seedorf, H., Dehority, B. A. and Janssen, P. 9 H. (2015) 'Phylogeny of Intestinal Ciliates, Including Charonina ventriculi, and 10 Comparison of Microscopy and 18S rRNA Gene Pyrosequencing for Rumen 11 Ciliate Community Structure Analysis', Appl Environ Microbiol, 81(7), 2433-2444. 12 13 Kittelmann, S., Pinares-Patiño, C. S., Seedorf, H., Kirk, M. R., Ganesh, S., McEwan, J. 14 C. and Janssen, P. H. (2014) 'Two Different Bacterial Community Types Are 15 Linked with the Low-Methane Emission Trait in Sheep', Plos One, 9(7), 1-9. 16 17 Kittelmann, S., Seedorf, H., Walters, W. A., Clemente, J. C., Knight, R., Gordon, J. I. and 18 Janssen, P. H. (2013a) 'Simultaneous amplicon sequencing to explore co- 19 occurrence patterns of bacterial, archaeal and eukaryotic microorganisms in 20 rumen microbial communities', PLoS One, 8(2), e47879. 21 22 Kittelmann, S., Seedorf, H., Walters, W. A., Clemente, J. C., Knight, R., Gordon, J. I. and 23 Janssen, P. H. (2013b) 'Simultaneous Amplicon Sequencing to Explore Co- 24 Occurrence Patterns of Bacterial, Archaeal and Eukaryotic Microorganisms in 25 Rumen Microbial Communities', Plos One, 8(2). 26 27 Knapp, J. R., Laur, G. L., Vadas, P. A., Weiss, W. P. and Tricarico, J. M. (2014) 'Invited 28 review: Enteric methane in dairy cattle production: Quantifying the opportunities 29 and impact of reducing emissions', J Dairy Sci, 97(6), 3231-3261. 30 31 Knights, D., Parfrey, L. W., Zaneveld, J., Lozupone, C. and Knight, R. (2011) 'Human- 32 associated microbial signatures: examining their predictive value', Cell Host & 33 Microbe, 10(4), 292-296. 34 35 Koch, R. M., Swiger, L. A., Chambers, D. and Gregory, K. E. (1963) 'Efficiency of feed 36 use in beef cattle', Journal of Animal Science, 22(2), 486-494. 37 38 Koetschan, C., Kittelmann, S., Lu, J., Al-Halbouni, D., Jarvis, G. N., Mueller, T., Wolf, M. 39 and Janssen, P. H. (2014) 'Internal Transcribed Spacer 1 Secondary Structure 40 Analysis Reveals a Common Core throughout the Anaerobic Fungi 41 (Neocallimastigomycota)', Plos One, 9(3). 42 43 Kong, R. S., Liang, G., Chen, Y., Stothard, P. and Guan le, L. (2016) 'Transcriptome 44 profiling of the rumen epithelium of beef cattle differing in residual feed intake', 45 BMC Genomics, 17, 592. 46

203

1 Kopylova, E., Noé, L. and Touzet, H. (2012) 'SortMeRNA: fast and accurate filtering of 2 ribosomal RNAs in metatranscriptomic data', Bioinformatics (Oxford, England), 3 28(24), 3211-3217. 4 5 Korver, S. (1988) 'Genetic aspects of feed intake and feed efficiency in dairy cattle: A 6 review', Livestock Production Science, 20(1), 1-13. 7 8 Kosugi, A., Murashima, K. and Doi, R. H. (2001) 'Characterization of Xylanolytic Enzymes 9 in Clostridium cellulovorans: Expression of Xylanase Activity 10 Dependent on Growth Substrates', Journal of Bacteriology, 183(24), 7037-7043. 11 12 Kozich, J. J., Westcott, S. L., Baxter, N. T., Highlander, S. K. and Schloss, P. D. (2013) 13 'Development of a dual-index sequencing strategy and curation pipeline for 14 analyzing amplicon sequence data on the MiSeq Illumina sequencing platform', 15 Applied and Environmental Microbiology, 79(17), 5112-5120. 16 17 Krause, D. O., Nagaraja, T. G., Wright, A. D. G. and Callaway, T. R. (2013) 'Board-invited 18 review: Rumen microbiology: Leading the way in microbial ecology1,2', Journal of 19 Animal Science, 91(1), 331-341. 20 21 Kumar, S., Choudhury, P. K., Carro, M. D., Griffith, G. W., Dagar, S. S., Puniya, M., 22 Calabro, S., Ravella, S. R., Dhewa, T., Upadhyay, R. C., Sirohi, S. K., Kundu, S. 23 S., Wanapat, M. and Puniya, A. K. (2014) 'New aspects and strategies for methane 24 mitigation from ruminants', Appl Microbiol Biotechnol, 98(1), 31-44. 25 26 Kurtz, Z. D., Mueller, C. L., Miraldi, E. R., Littman, D. R., Blaser, M. J. and Bonneau, R. 27 A. (2014) 'Sparse and compositionally robust inference of microbial ecological 28 networks', 29 30 Lam, K. N., Cheng, J., Engel, K., Neufeld, J. D. and Charles, T. C. (2015) 'Current and 31 future resources for functional metagenomics', Frontiers in Microbiology, 6(1196). 32 33 Langille, M. G. I., Zaneveld, J., Caporaso, J. G., McDonald, D., Knights, D., Reyes, J. A., 34 Clemente, J. C., Burkepile, D. E., Vega Thurber, R. L., Knight, R., Beiko, R. G. and 35 Huttenhower, C. (2013) 'Predictive functional profiling of microbial communities 36 using 16S rRNA marker gene sequences', Nature Biotechnology, 31, 814. 37 38 Larance, M. and Lamond, A. I. (2015) 'Multidimensional proteomics for cell biology', 39 Nature Reviews Molecular Cell Biology, 16, 269. 40 41 Larsbrink, J., Tuveng, T. R., Pope, P. B., Bulone, V., Eijsink, V. G. H., Brumer, H. and 42 McKee, L. S. (2017a) 'Proteomic insights into mannan degradation and protein 43 secretion by the forest floor bacterium Chitinophaga pinensis', Journal of 44 Proteomics, 156, 63-74. 45 46

204

Larsbrink, J., Tuveng, T. R., Pope, P. B., Eijsink, V. G. H., Bulone, V., Brumer, H. and 1 McKee, L. S. (2017b) 'Proteomic data on enzyme secretion and activity in the 2 bacterium Chitinophaga pinensis', Data in Brief, 11, 484-490. 3 4 Leahy, S. C., Kelly, W. J., Altermann, E., Ronimus, R. S., Yeoman, C. J., Pacheco, D. M., 5 Li, D., Kong, Z. H., McTavish, S., Sang, C., Lambie, S. C., Janssen, P. H., Dey, D. 6 and Attwood, G. T. (2010) 'The Genome Sequence of the Rumen Methanogen 7 Methanobrevibacter ruminantium Reveals New Possibilities for Controlling 8 Ruminant Methane Emissions', Plos One, 5(1). 9 10 Lee, S., Sung, J., Lee, J. and Ko, G. (2011) 'Comparison of the gut microbiotas of healthy 11 adult twins living in South Korea and the United States', Applied and Environmental 12 Microbiology, 77(20), 7433-7437. 13 14 Lee, S. S., Ha, J. K. and Cheng, K. J. (2000) 'Relative contributions of bacteria, protozoa, 15 and fungi to in vitro degradation of orchard grass cell walls and their interactions', 16 Applied and Environmental Microbiology, 66(9), 3807-3813. 17 18 Levin, B. J., Huang, Y. Y., Peck, S. C., Wei, Y., Campo, A. M. D., Marks, J. A., Franzosa, 19 E. A., Huttenhower, C. and Balskus, E. P. (2017) 'A prominent glycyl radical 20 enzyme in human gut microbiomes metabolizes trans-4-hydroxy-L-proline', 21 Science, 355(6325), 595-+. 22 23 Li, D. H., Liu, C. M., Luo, R. B., Sadakane, K. and Lam, T. W. (2015) 'MEGAHIT: an ultra- 24 fast single-node solution for large and complex metagenomics assembly via 25 succinct de Bruijn graph', Bioinformatics, 31(10), 1674-1676. 26 27 Li, F. and Guan, L. L. (2017) 'Metatranscriptomic Profiling Reveals Linkages between the 28 Active Rumen Microbiome and Feed Efficiency in Beef Cattle', Appl Environ 29 Microbiol, 83(9). 30 31 Li, F. and Guan, L. L. (2017) 'Metatranscriptomic profiling reveals linkages between the 32 active rumen microbiome and feed efficiency in beef cattle', Applied and 33 Environmental Microbiology. 34 35 Li, F., Henderson, G., Sun, X., Cox, F., Janssen, P. H. and Guan le, L. (2016) 'Taxonomic 36 Assessment of Rumen Microbiota Using Total RNA and Targeted Amplicon 37 Sequencing Approaches', Front Microbiol, 7, 987. 38 39 Li, F., Henderson, G., Sun, X., Cox, F., Janssen, P. H. and Guan, L. L. (2016) 'Taxonomic 40 Assessment of Rumen Microbiota Using Total RNA and Targeted Amplicon 41 Sequencing Approaches', Frontiers in Microbiology, 7(987). 42 43 Li, F., Hitch, T. C. A., Chen, Y., Creevey, C. J. and Guan, L. L. (2019a) 'Comparative 44 metagenomic and metatranscriptomic analyses reveal the breed effect on the 45

205

rumen microbiome and its associations with feed efficiency in beef cattle', 1 Microbiome, 7(1), 6. 2 3 Li, F., Li, C., Chen, Y., Liu, J., Zhang, C., Irving, B., Fitzsimmons, C., Plastow, G. and 4 Guan, L. L. (2019b) 'Host genetics influence the rumen microbiota and heritable 5 rumen microbial features associate with feed efficiency in cattle', Microbiome, 7(1), 6 92. 7 8 Li, M., Penner, G. B., Hernandez-Sanabria, E., Oba, M. and Guan, L. L. (2009) 'Effects 9 of sampling location and time, and host animal on assessment of bacterial diversity 10 and fermentation parameters in the bovine rumen', Journal of Applied 11 Microbiology, 107(6), 1924-1934. 12 13 Li, R. W., Connor, E. E., Li, C., Baldwin Vi, R. L. and Sparks, M. E. (2012) 14 'Characterization of the rumen microbiota of pre-ruminant calves using 15 metagenomic tools', Environ Microbiol, 14(1), 129-39. 16 17 Li, Y. and Zhang, Y. (2009) 'REMO: A New Protocol to Refine Full Atomic Protein Models 18 from C-alpha Traces by Optimizing Hydrogen-Bonding Networks', Proteins, 76(3), 19 665-676. 20 21 Li, Z. P., Wright, A. D. G., Si, H. Z., Wang, X. X., Qian, W. X., Zhang, Z. G. and Li, G. Y. 22 (2016) 'Changes in the rumen microbiome and metabolites reveal the effect of host 23 genetics on hybrid crosses', Environmental Microbiology Reports, 8(6), 1016- 24 1023. 25 26 Lin, C., Stahl, D. A. and Raskin, L. (1997) 'Microbial community structure in 27 gastrointestinal tracts of domestic animals: comparative analyses using rRNA- 28 targeted oligonucleotide probes', Fems Microbiology Ecology, 22(4), 281-294. 29 30 Liquet, B., Cao, K.-A. L., Hocini, H. and Thiébaut, R. (2012) 'A novel approach for 31 biomarker selection and the integration of repeated measures experiments from 32 two assays', Bmc Bioinformatics, 13(1), 325. 33 34 Liu, J. h., Zhang, M. l., Zhang, R. y., Zhu, W. y. and Mao, S. y. (2016) 'Comparative 35 studies of the composition of bacterial microbiota associated with the ruminal 36 content, ruminal epithelium and in the faeces of lactating dairy cows', Microb 37 Biotechnol, 9(2), 257-268. 38 39 Lombard, V., Golaconda Ramulu, H., Drula, E., Coutinho, P. M. and Henrissat, B. (2014) 40 'The carbohydrate-active enzymes database (CAZy) in 2013', Nucleic Acids 41 Research, 42(D1), D490-D495. 42 43 Love, M. I., Huber, W. and Anders, S. (2014) 'Moderated estimation of fold change and 44 dispersion for RNA-seq data with DESeq2', Genome Biology, 15(12), 550-550. 45 46

206

Lovell, D., Pawlowsky-Glahn, V., Egozcue, J. J., Marguerat, S. and Bähler, J. (2015) 1 'Proportionality: A Valid Alternative to Correlation for Relative Data', Plos 2 Computational Biology, 11(3), 1-12. 3 4 Malmuthuge, N., Li, M., Chen, Y., Fries, P., Griebel, P. J., Baurhoo, B., Zhao, X. and 5 Guan, L. L. (2012) 'Distinct commensal bacteria associated with ingesta and 6 mucosal epithelium in the gastrointestinal tracts of calves and chickens', Fems 7 Microbiology Ecology, 79(2), 337-347. 8 9 Mandal, S., Van Treuren, W., White, R. A., Eggesbo, M., Knight, R. and Peddada, S. D. 10 (2015) 'Analysis of composition of microbiomes: a novel method for studying 11 microbial composition', Microbial Ecology in Health and Disease, 26, 27663-Article 12 No.: 27663. 13 14 Marchesi, J. R. and Ravel, J. (2015) 'The vocabulary of microbiome research: a proposal', 15 Microbiome, 3(1), 31. 16 17 Márquez, G. C., Speidel, S. E., Enns, R. M. and Garrick, D. J. (2010) 'Genetic diversity 18 and population structure of American Red Angus cattle', Journal of Animal 19 Science, 88(1), 59-68. 20 21 Martin, C., Morgavi, D. P. and Doreau, M. (2010) 'Methane mitigation in ruminants: from 22 microbe to the farm scale', animal, 4(3), 351-65. 23 24 Martínez, I., Jaehyoung, K., Duffy, P. R., Schlegel, V. L. and Walter, J. (2010) 'Resistant 25 Starches Types 2 and 4 Have Differential Effects on the Composition of the Fecal 26 Microbiota in Human Subjects', Plos One, 5(11), 1-11. 27 28 Martinez-Fernandez, G., Denman, S. E., Cheung, J. and McSweeney, C. S. (2017) 29 'Phloroglucinol Degradation in the Rumen Promotes the Capture of Excess 30 Hydrogen Generated from Methanogenesis Inhibition', Frontiers in Microbiology, 31 8(1871). 32 33 Martinez-Fernandez, G., Denman, S. E., Yang, C., Cheung, J., Mitsumori, M. and 34 McSweeney, C. S. (2016) 'Methane Inhibition Alters the Microbial Community, 35 Hydrogen Flow, and Fermentation Response in the Rumen of Cattle', Frontiers in 36 Microbiology, 7(1122). 37 38 Matthews, C., Crispie, F., Lewis, E., Reid, M., O’Toole, P. W. and Cotter, P. D. (2018) 39 'The rumen microbiome: a crucial consideration when optimising milk and meat 40 production and nitrogen utilisation efficiency AU - Matthews, Chloe', Gut Microbes, 41 1-18. 42 43 McAllister, T. A., Bae, H. D., Jones, G. A. and Cheng, K. J. (1994a) 'Microbial attachment 44 and feed digestion in the rumen', Journal of Animal Science, 72(11), 3004-3018. 45 46

207

McAllister, T. A., Bae, H. D., Yanke, L. J., Cheng, K. J. and Muir, A. (1994b) 'Effect of 1 condensed tannins from birdsfoot trefoil on endoglucanase activity and the 2 digestion of cellulose filter paper by ruminal fungi', Can J Microbiol, 40(4), 298- 3 305. 4 5 McAllister, T. A., Meale, S. J., Valle, E., Guan, L. L., Zhou, M., Kelly, W. J., Henderson, 6 G., Attwood, G. T. and Janssen, P. H. (2015) 'RUMINANT NUTRITION 7 SYMPOSIUM: Use of genomics and transcriptomics to identify strategies to lower 8 ruminal methanogenesis', J Anim Sci, 93(4), 1431-49. 9 10 McCabe, M. S., Cormican, P., Keogh, K., O'Connor, A., O'Hara, E., Palladino, R. A., 11 Kenny, D. A. and Waters, S. M. (2015) 'Illumina MiSeq Phylogenetic Amplicon 12 Sequencing Shows a Large Reduction of an Uncharacterised Succinivibrionaceae 13 and an Increase of the Methanobrevibacter gottschalkii Clade in Feed Restricted 14 Cattle', PLoS One, 10(7), e0133234. 15 16 McCabe, M. S., Cormican, P., Keogh, K., O’Connor, A., O’Hara, E., Palladino, R. A., 17 Kenny, D. A. and Waters, S. M. (2015) 'Illumina MiSeq Phylogenetic Amplicon 18 Sequencing Shows a Large Reduction of an Uncharacterised Succinivibrionaceae 19 and an Increase of the Methanobrevibacter gottschalkii Clade in Feed Restricted 20 Cattle', Plos One, 10(7), e0133234. 21 22 McCann, J. C., Elolimy, A. A. and Loor, J. J. (2017) 'Rumen Microbiome, Probiotics, and 23 Fermentation Additives', The Veterinary Clinics Of North America. Food Animal 24 Practice. 25 26 McCarthy, D. J., Chen, Y. and Smyth, G. K. (2012) 'Differential expression analysis of 27 multifactor RNA-Seq experiments with respect to biological variation', Nucleic 28 Acids Research, 40(10), 4288-4297. 29 30 McDougall, E. I. (1948) 'Studies on ruminant saliva. 1. The composition and output of 31 sheep's saliva', Biochemical Journal, 43(1), 99-109. 32 33 McGovern, E., McCabe, M. S., Cormican, P., Popova, M., Keogh, K., Kelly, A. K., Kenny, 34 D. A. and Waters, S. M. (2017) 'Plane of nutrition affects the phylogenetic diversity 35 and relative abundance of transcriptionally active methanogens in the bovine 36 rumen', Scientific Reports, 7(1), 13047. 37 38 McLean, K. L. and Schmutz, S. M. (2009) 'Associations of melanocortin 1 receptor 39 genotype with growth and carcass traits in beef cattle', Canadian Journal of Animal 40 Science, 89(3), 295-300. 41 42 McMurdie, P. J. and Holmes, S. (2014) 'Waste Not, Want Not: Why Rarefying Microbiome 43 Data Is Inadmissible', Plos Computational Biology, 10(4), e1003531. 44 45

208

McNeil, M., Darvill, A. G., Fry, S. C. and Albersheim, P. (1984) 'Structure and Function of 1 the Primary Cell Walls of Plants', Annual Review of Biochemistry, 53(1), 625-663. 2 3 McSweeney, C. and Mackie, R. I. (2012) Microorganisms and ruminant digestion: state 4 of knowledge, trends and future prospects, Comission on genetic resources for 5 food and agriculture 6 7 Mesuere, B., Devreese, B., Debyser, G., Aerts, M., Vandamme, P. and Dawyndt, P. 8 (2012) 'Unipept: Tryptic Peptide-Based Biodiversity Analysis of Metaproteome 9 Samples', Journal of Proteome Research, 11(12), 5773-5780. 10 11 Meyer, F., Paarmann, D., D'Souza, M., Olson, R., Glass, E., Kubal, M., Paczian, T., 12 Rodriguez, A., Stevens, R., Wilke, A., Wilkening, J. and Edwards, R. (2008a) 'The 13 metagenomics RAST server – a public resource for the automatic phylogenetic 14 and functional analysis of metagenomes', Bmc Bioinformatics, 9(1), 386. 15 16 Meyer, F., Paarmann, D., D'Souza, M., Olson, R., Glass, E. M., Kubal, M., Paczian, T., 17 Rodriguez, A., Stevens, R., Wilke, A., Wilkening, J. and Edwards, R. A. (2008b) 18 'The metagenomics RAST server - a public resource for the automatic 19 phylogenetic and functional analysis of metagenomes', Bmc Bioinformatics, 9. 20 21 Miller, G. L. (1959) 'Use of Dinitrosalicylic Acid Reagent for Determination of Reducing 22 Sugar', Analytical Chemistry, 31(3), 426-428. 23 24 Miron, J., Ben-Ghedalia, D. and Morrison, M. (2001) 'Invited Review: Adhesion 25 Mechanisms of Rumen Cellulolytic Bacteria', Journal of Dairy Science, 84(6), 26 1294-1309. 27 28 Montanholi, Y. R., Swanson, K. C., Palme, R., Schenkel, F. S., McBride, B. W., Lu, D. 29 and Miller, S. P. (2010) 'Assessing feed efficiency in beef steers through feeding 30 behavior, infrared thermography and glucocorticoids', Animal, 4(5), 692-701. 31 32 Morgavi, D., Kelly, W. J., Janssen, P. H. and Attwood, G. T. (2013) 'Rumen microbial 33 (meta) genomics and its application to ruminant production', 184. 34 35 Morgavi, D. P., Forano, E., Martin, C. and Newbold, C. J. (2010) 'Microbial ecosystem 36 and methanogenesis in ruminants', Animal, 4(7), 1024-1036. 37 38 Morgavi, D. P., Kelly, W. J., Janssen, P. H. and Attwood, G. T. (2012a) 'Rumen microbial 39 (meta)genomics and its application to ruminant production', Animal, 7(s1), 184- 40 201. 41 42 Morgavi, D. P., Martin, C., Jouany, J. P. and Ranilla, M. J. (2012b) 'Rumen protozoa and 43 methanogenesis: not a simple cause-effect relationship', Br J Nutr, 107(3), 388- 44 97. 45 46

209

Morrison, M., Murray, R. M. and Boniface, A. N. (1990) 'NUTRIENT METABOLISM AND 1 RUMEN MICROORGANISMS IN SHEEP FED A POOR-QUALITY TROPICAL 2 GRASS HAY SUPPLEMENTED WITH SULFATE', Journal of Agricultural Science, 3 115, 269-275. 4 5 Munoz-Munoz, J., Cartmell, A., Terrapon, N., Henrissat, B. and Gilbert, H. J. (2017) 6 'Unusual active site location and catalytic apparatus in a glycoside hydrolase 7 family', Proceedings of the National Academy of Sciences. 8 9 Myer, P. R., Smith, T. P. L., Wells, J. E., Kuehn, L. A. and Freetly, H. C. (2015a) 'Rumen 10 Microbiome from Steers Differing in Feed Efficiency', PLoS One, 10(6), e0129174. 11 12 Myer, P. R., Smith, T. P. L., Wells, J. E., Kuehn, L. A. and Freetly, H. C. (2015b) 'Rumen 13 Microbiome from Steers Differing in Feed Efficiency', Plos One, 10(6). 14 15 Nations, U. (2015) 'World Population Prospects', [online], available]. 16 17 Nayfach, S., Bradley, P. H., Wyman, S. K., Laurent, T. J., Williams, A., Pollard, K. S., 18 Sharpton, T. J. and Eisen, J. A. (2015) 'Automated and accurate estimation of gene 19 family abundance from shotgun metagenomes', Plos Computational Biology, 20 11(11), 29p. 21 22 Network, I. F. C. (2016) Long-term dairy outlook. 23 24 Neves, A. L. A., Li, F., Ghoshal, B., McAllister, T. and Guan, L. L. (2017) 'Enhancing the 25 resolution of rumen microbial classification from metatranscriptomic data using 26 Kraken and Mothur', Frontiers in Microbiology, 8(2445). 27 28 Newbold, C. J., de la Fuente, G., Belanche, A., Ramos-Morales, E. and McEwan, N. R. 29 (2015) 'The Role of Ciliate Protozoa in the Rumen', Front Microbiol, 6, 1313. 30 31 Nkrumah, J. D., Okine, E. K., Mathison, G. W., Schmid, K., Li, C., Basarab, J. A., Price, 32 M. A., Wang, Z. and Moore, S. S. (2006) 'Relationships of feedlot feed efficiency, 33 performance, and feeding behavior with metabolic rate, methane production, and 34 energy partitioning in beef cattle', Journal of Animal Science, 84(1), 145-153. 35 36 Nocker, A., Burr, M. and Camper, A. K. (2007) 'Genotypic Microbial Community Profiling: 37 A Critical Technical Review', Microbial Ecology, 54(2), 276-289. 38 39 Nordberg, H., Cantor, M., Dusheyko, S., Hua, S., Poliakov, A., Shabalov, I., Smirnova, 40 T., Grigoriev, I. V. and Dubchak, I. (2014) 'The genome portal of the Department 41 of Energy Joint Genome Institute: 2014 updates', Nucleic Acids Research, 42 42(Database issue), D26-D31. 43 44 NRC (2016) Nutrient Requirements of Beef Cattle, 8th Revised Edition ed., Washington, 45 D.C., USA: National Academy Press. 46

210

1 Olfert, E., Cross, B. and McWilliams, A. (1993) Guide to the Care and Use of Experimental 2 Steers, Ottawa, ON, Canada: Canadian Council on Animal Care. 3 4 Opdahl, L. J., Gonda, M. G. and St-Pierre, B. (2018) 'Identification of Uncultured Bacterial 5 Species from Firmicutes, Bacteroidetes and CANDIDATUS Saccharibacteria as 6 Candidate Cellulose Utilizers from the Rumen of Beef Cows', Microorganisms, 7 6(1). 8 9 Orpin, C. G. (1975) 'STUDIES ON RUMEN FLAGELLATE NEOCALLIMASTIX- 10 FRONTALIS', Journal of General Microbiology, 91(DEC), 249-262. 11 12 Orpin, C. G. and Joblin, K. N. (1997) 'The rumen anaerobic fungi' in Hobson, P. N. and 13 Stewart, C. S., eds., The Rumen Microbial Ecosystem, Springer Netherlands, 140- 14 195. 15 16 Ozkose, E., Thomas, B. J., Davies, D. R., Griffith, G. W. and Theodorou, M. K. (2001) 17 'Cyllamyces aberensis gen.nov sp.nov., a new anaerobic gut fungus with branched 18 sporangiophores isolated from cattle', Canadian Journal of Botany-Revue 19 Canadienne De Botanique, 79(6), 666-673. 20 21 Pajarillo, E. A. B., Chae, J. P., Balolong, M. P., Kim, H. B., Kang, D.-K. and Seo, K.-S. 22 (2015) '- Characterization of the Fecal Microbial Communities of Duroc Pigs Using 23 16S rRNA Gene Pyrosequencing', - 28(- 4), - 591. 24 25 Palackal, N., Lyon, C. S., Zaidi, S., Luginbühl, P., Dupree, P., Goubet, F., Macomber, J. 26 L., Short, J. M., Hazlewood, G. P., Robertson, D. E. and Steer, B. A. (2007) 'A 27 multifunctional hybrid glycosyl hydrolase discovered in an uncultured microbial 28 consortium from ruminant gut', Applied Microbiology and Biotechnology, 74(1), 29 113-124. 30 31 Parks, D. H., Imelfort, M., Skennerton, C. T., Hugenholtz, P. and Tyson, G. W. (2015) 32 'CheckM: Assessing the quality of microbial genomes recovered from isolates, 33 single cells, and metagenomes', Genome Research, 25(7), 1043-1055. 34 35 Parmar, N. R., Solanki, J. V., Patel, A. B., Shah, T. M., Patel, A. K., Parnerkar, S., Kumar, 36 J. I. and Joshi, C. G. (2014) 'Metagenome of Mehsani buffalo rumen microbiota: 37 an assessment of variation in feed-dependent phylogenetic and functional 38 classification', J Mol Microbiol Biotechnol, 24(4), 249-61. 39 40 Patil, K. R., Haider, P., Pope, P. B., Turnbaugh, P. J., Morrison, M., Scheffer, T. and 41 McHardy, A. C. (2011) 'Taxonomic metagenome sequence assignment with 42 structured output models', Nature Methods, 8, 191. 43 44

211

Patra, A., Park, T., Kim, M. and Yu, Z. (2017) 'Rumen methanogens and mitigation of 1 methane emission by anti-methanogenic compounds and substances', Journal of 2 Animal Science and Biotechnology, 8(1), 13. 3 4 Paul, S. S., Yu, Z., Bu, D., Xu, J. and Hyde, K. D. (2018) 'A phylogenetic census of global 5 diversity of gut anaerobic fungi and a new taxonomic framework', Fungal Diversity, 6 89(1), 253-266. 7 8 Pawlowsky‐ Glahn, V., Egozcue, J. J. and Tolosana‐ Delgado, R. (2015) Modelling and 9 Analysis of Compositional Data. 10 11 Pearson, K. (1897) 'Mathematical contributions to the theory of evolution.—On a form of 12 spurious correlation which may arise when indices are used in the measurement 13 of organs', Proceedings of the Royal Society of London, 60(359-367), 489-498. 14 15 Petri, R. M., Schwaiger, T., Penner, G. B., Beauchemin, K. A., Forster, R. J., McKinnon, 16 J. J. and McAllister, T. A. (2013) 'Characterization of the Core Rumen Microbiome 17 in Cattle during Transition from Forage to Concentrate as Well as during and after 18 an Acidotic Challenge', Plos One, 8(12). 19 20 Pickering, N. K., Oddy, V. H., Basarab, J., Cammack, K., Hayes, B., Hegarty, R. S., 21 Lassen, J., McEwan, J. C., Miller, S., Pinares-Patino, C. S. and de Haas, Y. 22 (2015a) 'Animal board invited review: genetic possibilities to reduce enteric 23 methane emissions from ruminants', animal, 9(9), 1431-40. 24 25 Pickering, N. K., Oddy, V. H., Basarab, J., Cammack, K., Hayes, B., Hegarty, R. S., 26 Lassen, J., McEwan, J. C., Miller, S., Pinares-Patiño, C. S. and de Haas, Y. 27 (2015b) 'Animal board invited review: genetic possibilities to reduce enteric 28 methane emissions from ruminants', Animal : an international journal of animal 29 bioscience, 9(9), 1431-1440. 30 31 Piro, V. C., Matschkowski, M. and Renard, B. Y. (2017) 'MetaMeta: integrating 32 metagenome analysis tools to improve taxonomic profiling', Microbiome, 5(1), 101- 33 101. 34 35 Poelaert, C., Nollevaux, G., Boudry, C., Taminiau, B., Nezer, C., Daube, G., Schneider, 36 Y. J., Portetelle, D., Théwis, A. and Bindelle, J. (2017) 'Reducing agent can be 37 omitted in the incubation medium of the batch in vitro fermentation model of the 38 pig intestines', Animal, 12(6), 1154-1164. 39 40 Poretsky, R., Rodriguez, R. L., Luo, C., Tsementzi, D. and Konstantinidis, K. T. (2014) 41 'Strengths and limitations of 16S rRNA gene amplicon sequencing in revealing 42 temporal microbial community dynamics', PLoS One, 9(4), e93827. 43 44 Poulsen, M., Schwab, C., Borg Jensen, B., Engberg, R. M., Spang, A., Canibe, N., 45 Højberg, O., Milinovich, G., Fragner, L., Schleper, C., Weckwerth, W., Lund, P., 46

212

Schramm, A. and Urich, T. (2013) 'Methylotrophic methanogenic Thermoplasmata 1 implicated in reduced methane emissions from bovine rumen', Nat Commun, 4, 2 1428. 3 4 Qi, M., Wang, P., O'Toole, N., Barboza, P. S., Ungerfeld, E., Leigh, M. B., Selinger, L. B., 5 Butler, G., Tsang, A., McAllister, T. A. and Forster, R. J. (2011) 'Snapshot of the 6 Eukaryotic Gene Expression in Muskoxen Rumen-A Metatranscriptomic 7 Approach', Plos One, 6(5). 8 9 Qi, M., Wang, P., Selinger, L. B., Yanke, L. J., Forster, R. J. and McAllister, T. A. (2011) 10 'Isolation and characterization of a ferulic acid esterase (Fae1A) from the rumen 11 fungus Anaeromyces mucronatus', Journal of Applied Microbiology, 110(5), 1341- 12 1350. 13 14 Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Peplies, J. and 15 Glöckner, F. O. (2013) 'The SILVA ribosomal RNA gene database project: 16 improved data processing and web-based tools', Nucleic Acids Research, 17 41(Database issue), D590-D596. 18 19 Ramírez, L., Arrizon, J., Sandoval, G., Cardador, A., Bello-Mendoza, R., Lappe, P. and 20 Mateos-Díaz, J. C. (2008) 'A New Microplate Screening Method for the 21 Simultaneous Activity Quantification of Feruloyl Esterases, Tannases, and 22 Chlorogenate Esterases', Applied Biochemistry and Biotechnology, 151(2), 711- 23 723. 24 25 Rashamuse, K. J., Burton, S. G. and Cowan, D. A. (2007) 'A novel recombinant ethyl 26 ferulate esterase from Burkholderia multivorans', Journal of Applied Microbiology, 27 103(5), 1610-1620. 28 29 Ribeiro, G. O., Gruninger, R. J., Badhan, A. and McAllister, T. A. (2016) 'Mining the rumen 30 for fibrolytic feed enzymes', Animal Frontiers, 6(2), 20-26. 31 32 Ricard, G., McEwan, N. R., Dutilh, B. E., Jouany, J.-P., Macheboeuf, D., Mitsumori, M., 33 McIntosh, F. M., Michalowski, T., Nagamine, T., Nelson, N., Newbold, C. J., 34 Nsabimana, E., Takenaka, A., Thomas, N. A., Ushida, K., Hackstein, J. H. P. and 35 Huynen, M. A. (2006) 'Horizontal gene transfer from Bacteria to rumen Ciliates 36 indicates adaptation to their anaerobic, carbohydrates-rich environment', BMC 37 Genomics, 7, 22-22. 38 39 Rico, D. E., Preston, S. H., Risser, J. M. and Harvatine, K. J. (2015) 'Rapid changes in 40 key ruminal microbial populations during the induction of and recovery from diet- 41 induced milk fat depression in dairy cows', British Journal of Nutrition, 114(3), 358- 42 367. 43 44

213

Robinson, M. D., McCarthy, D. J. and Smyth, G. K. (2010) 'edgeR: a Bioconductor 1 package for differential expression analysis of digital gene expression data', 2 Bioinformatics, 26(1), 139-140. 3 4 Robinson, M. D. and Oshlack, A. (2010) 'A scaling normalization method for differential 5 expression analysis of RNA-seq data', Genome Biology, 11(3), R25. 6 7 Robinson, M. D. and Smyth, G. K. (2008) 'Small-sample estimation of negative binomial 8 dispersion, with applications to SAGE data', Biostatistics, 9(2), 321-332. 9 10 Rodríguez, H., Curiel, J. A., Landete, J. M., de las Rivas, B., de Felipe, F. L., Gómez- 11 Cordovés, C., Mancheño, J. M. and Muñoz, R. (2009) 'Food phenolics and lactic 12 acid bacteria', International Journal of Food Microbiology, 132(2), 79-90. 13 14 Roehe, R., Dewhurst, R. J., Duthie, C.-A., Rooke, J. A., McKain, N., Ross, D. W., Hyslop, 15 J. J., Waterhouse, A., Freeman, T. C., Watson, M. and Wallace, R. J. (2016) 16 'Bovine Host Genetic Variation Influences Rumen Microbial Methane Production 17 with Best Selection Criterion for Low Methane Emitting and Efficiently Feed 18 Converting Hosts Based on Metagenomic Gene Abundance', Plos Genetics, 12(2). 19 20 Rohart, F., Gautier, B., Singh, A. and Lê Cao, K.-A. (2017) 'mixOmics: An R package for 21 ‘omics feature selection and multiple data integration', Plos Computational Biology, 22 13(11), 1-19. 23 24 Rojas-Downing, M. M., Nejadhashemi, A. P., Harrigan, T. and Woznicki, S. A. (2017) 25 'Climate change and livestock: Impacts, adaptation, and mitigation', Climate Risk 26 Management, 16, 145-163. 27 28 Romain, G., Thomas, T. and Margaret, E. K. E. (2009) 'Ecological Conditions Affect 29 Evolutionary Trajectory in a Predator-Prey System', Evolution, (3), 641. 30 31 Romero-Perez, A., Okine, E. K., McGinn, S. M., Guan, L. L., Oba, M., Duval, S. M., 32 Kindermann, M. and Beauchemin, K. A. (2014) 'The potential of 3-nitrooxypropanol 33 to lower enteric methane emissions from beef cattle', Journal of Animal Science, 34 92(10), 4682-4693. 35 36 Rosnow, J. J., Anderson, L. N., Nair, R. N., Baker, E. S. and Wright, A. T. (2017) 'Profiling 37 microbial lignocellulose degradation and utilization by emergent omics 38 technologies', Critical Reviews in Biotechnology, 37(5), 626-640. 39 40 Russell, J. B. and Rychlik, J. L. (2001) 'Factors that alter rumen microbial ecology', 41 Science, 292(5519), 1119-1122. 42 43 Sabree, Z. L., Rondon, M. R. and Handelsman, J. (2009) 'Metagenomics' in Schaechter, 44 M., ed. Encyclopedia of Microbiology (Third Edition), Oxford: Academic Press, 45 622-632. 46

214

1 Saha, B. C. (2000) 'α-l-Arabinofuranosidases: biochemistry, molecular biology and 2 application in biotechnology', Biotechnology Advances, 18(5), 403-423. 3 4 Saha, B. C. (2003) 'Hemicellulose bioconversion', Journal of Industrial Microbiology and 5 Biotechnology, 30(5), 279-291. 6 7 Salonen, A., Lahti, L., Salojärvi, J., Holtrop, G., Korpela, K., Duncan, S. H., Date, P., 8 Farquharson, F., Johnstone, A. M., Lobley, G. E., Louis, P., Flint, H. J. and de Vos, 9 W. M. (2014) 'Impact of diet and individual variation on intestinal microbiota 10 composition and fermentation products in obese men', ISME Journal: 11 Multidisciplinary Journal of Microbial Ecology, 8(11), 2218-2230. 12 13 San-Juan-Vergara, H., Peña, M., Cadena-Cruz, C., Diaz-Olmos, Y., Hurtado-Gómez, L., 14 Vidal-Orjuela, O., Viasus, D., Cervantes-Acosta, G., Zurek, E., Portnoy, I., Vélez, 15 J. I., Ajami, N. J., Petrosino, J. F., Mogollon, C., Hernández, D., Urruchurtu, I., Di- 16 Ruggiero, P., Sanchez-Sit, S., Guardo-García, E. and Torres, N. (2018) 'A 17 Lachnospiraceae-dominated bacterial signature in the fecal microbiota of HIV- 18 infected individuals from Colombia, South America', Scientific Reports, 8(1). 19 20 Sanchez, C. (2009) 'Lignocellulosic residues: Biodegradation and bioconversion by fungi', 21 Biotechnology Advances, 27(2), 185-194. 22 23 Schellenberg, J. J., Verbeke, T. J., Sparling, R., McQueen, P., Krokhin, O. V., Wilkins, J. 24 A., Zhang, X., Alvare, G., Fristensky, B., Levin, D. B., Thallinger, G. G. and 25 Henrissat, B. (2014) 'Enhanced whole genome sequence and annotation of 26 Clostridium stercorarium DSM8532T using RNA-seq transcriptomics and high- 27 throughput proteomics', BMC Genomics, 15(1). 28 29 Schloss, P. D., Westcott, S. L., Ryabin, T., Hall, J. R., Hartmann, M., Hollister, E. B., 30 Lesniewski, R. A., Oakley, B. B., Parks, D. H., Robinson, C. J., Sahl, J. W., Stres, 31 B., Thallinger, G. G., Van Horn, D. J. and Weber, C. F. (2009) 'Introducing mothur: 32 Open-Source, Platform-Independent, Community-Supported Software for 33 Describing and Comparing Microbial Communities', Applied & Environmental 34 Microbiology, 75(23), 7537-7541. 35 36 Schnoes, A. M., Brown, S. D., Dodevski, I. and Babbitt, P. C. (2009) 'Annotation Error in 37 Public Databases: Misannotation of Molecular Function in Enzyme Superfamilies', 38 Plos Computational Biology, 5(12), e1000605. 39 40 Seedorf, H., Kittelmann, S., Henderson, G. and Janssen, P. H. (2014) 'RIM-DB: a 41 taxonomic framework for community structure analysis of methanogenic archaea 42 from the rumen and other intestinal environments', PeerJ, 2. 43 44

215

Sehgal, J. P., Jit, D., Puniya, A. K. and Singh, K. (2008) 'Influence of anaerobic fungal 1 administration on growth, rumen fermentation and nutrient digestion in female 2 buffalo calves', Journal of Animal and Feed Sciences, 17(4), 510-518. 3 4 Seshadri, R., Leahy, S. C., Attwood, G. T., Teh, K. H., Lambie, S. C., Cookson, A. L., 5 Eloe-Fadrosh, E. A., Pavlopoulos, G. A., Hadjithomas, M., Varghese, N. J., Paez- 6 Espino, D., Perry, R., Henderson, G., Creevey, C. J., Terrapon, N., Lapebie, P., 7 Drula, E., Lombard, V., Rubin, E., Kyrpides, N. C., Henrissat, B., Woyke, T., 8 Ivanova, N. N. and Kelly, W. J. (2018) 'Cultivation and sequencing of rumen 9 microbiome members from the Hungate1000 Collection', Nature Biotechnology, 10 36(4), 359-367. 11 12 Shaani, Y., Zehavi, T., Eyal, S., Miron, J. and Mizrahi, I. (2018) 'Microbiome niche 13 modification drives diurnal rumen community assembly, overpowering individual 14 variability and diet effects', The ISME Journal. 15 16 Shabat, S. K. B., Sasson, G., Doron-Faigenboim, A., Durman, T., Yaacoby, S., Berg 17 Miller, M. E., White, B. A., Shterzer, N. and Mizrahi, I. (2016) 'Specific microbiome- 18 dependent mechanisms underlie the energy harvest efficiency of ruminants', ISME 19 J. 20 21 Shallom, D. and Shoham, Y. (2003) 'Microbial hemicellulases', Current Opinion in 22 Microbiology, 6(3), 219-228. 23 24 Shannon, P., Markiel, A., Ozier, O., Baliga, N. S., Wang, J. T., Ramage, D., Amin, N., 25 Schwikowski, B. and Ideker, T. (2003) 'Cytoscape: A software environment for 26 integrated models of biomolecular interaction networks', Genome Research, 27 13(11), 2498-2504. 28 29 Shapiro, E., Biezuner, T. and Linnarsson, S. (2013) 'Single-cell sequencing-based 30 technologies will revolutionize whole-organism science', Nature Reviews Genetics, 31 14, 618. 32 33 Silley, P. (1985) 'A note on the pectinolytic enzymes of Lachnospira multiparus', Journal 34 of Applied Bacteriology, 58(2), 145-149. 35 36 Silverman, J. D., Washburne, A. D., Mukherjee, S. and David, L. A. (2017) 'A phylogenetic 37 transform enhances analysis of compositional microbiota data', Elife, 6, e21887. 38 39 Sirohi, S. K., Pandey, N., Singh, B. and Puniya, A. K. (2010) 'Rumen methanogens: a 40 review', Indian Journal of Microbiology, 50(3), 253-262. 41 42 Skene, I. K. and Brooker, J. D. (1995) 'Characterization of Tannin Acylhydrolase Activity 43 in the Ruminal Bacterium Selenomonas Ruminantium', Anaerobe, 1(6), 321-327. 44 45

216

Snelling, T. J. and Wallace, R. J. (2017) 'The rumen microbial metaproteome as revealed 1 by SDS-PAGE', BMC Microbiology, 17(1), 9. 2 3 Sniffen, C. J. and Robinson, P. H. (1987) 'Microbial growth and flow as influenced by 4 dietary manipulations', Journal of Dairy Science, 70(2), 425-441. 5 6 Sowa, Y. and Berry, R. M. (2008) 'Bacterial flagellar motor', Quarterly Reviews of 7 Biophysics, 41(2), 103-132. 8 9 Stackebrandt, E. and Goebel, B. M. (1994) 'Taxonomic Note: A Place for DNA-DNA 10 Reassociation and 16S rRNA Sequence Analysis in the Present Species Definition 11 in Bacteriology', International Journal Of Systematic And Evolutionary 12 Microbiology, 44(4), 846-849. 13 14 Steele, M. A., Penner, G. B., Chaucheyras-Durand, F. and Guan, L. L. (2016) 15 'Development and physiology of the rumen and the lower gut: Targets for 16 improving gut health', J Dairy Sci. 17 18 Stephen, A. M. (1983) The Polysaccharides, San Diego, CA, USA: Academic. 19 20 Stevenson, D. M. and Weimer, P. J. (2007) 'Dominance of Prevotella and low abundance 21 of classical ruminal bacterial species in the bovine rumen revealed by relative 22 quantification real-time PCR', Applied Microbiology and Biotechnology, 75(1), 165- 23 174. 24 25 Stewart, C. S., Fonty, G. and Gouet, P. (1988) 'The establishment of rumen microbial 26 communities', Animal Feed Science and Technology, 21(2), 69-97. 27 28 Stewart, R. D., Warr, A., Watson, M., Auffret, M. D., Dewhurst, R. J., Roehe, R., Wiser, 29 A. H., Press, M. O., Langford, K. W., Liachko, I., Snelling, T. J. and Walker, A. W. 30 (2018) 'Assembly of 913 microbial genomes from metagenomic sequencing of the 31 cow rumen', Nature Communications, 9(1). 32 33 Stuart, D. I., Courcelle, E., M√©toz, F. and Gouet, P. (1999) 'ESPript: analysis of multiple 34 sequence alignments in PostScript', Bioinformatics, 15(4), 305-308. 35 36 Suen, G., Weimer, P. J., Stevenson, D. M., Aylward, F. O., Boyum, J., Deneke, J., 37 Drinkwater, C., Ivanova, N. N., Mikhailova, N., Chertkov, O., Goodwin, L. A., 38 Currie, C. R., Mead, D. and Brumm, P. J. (2011) 'The Complete Genome 39 Sequence of Fibrobacter succinogenes S85 Reveals a Cellulolytic and Metabolic 40 Specialist', 41 42 Sunagawa, S., Mende, D. R., Zeller, G., Izquierdo-Carrasco, F., Berger, S. A., Kultima, J. 43 R., Coelho, L. P. and Arumugam, M. (2013) 'Metagenomic species profiling using 44 universal phylogenetic marker genes', 10(12), 1196-9. 45 46

217

Sunagawa, S., Mende, D. R., Zeller, G., Izquierdo-Carrasco, F., Berger, S. A., Kultima, J. 1 R., Coelho, L. P., Arumugam, M., Tap, J., Nielsen, H. B., Rasmussen, S., Brunak, 2 S., Pedersen, O., Guarner, F., de Vos, W. M., Wang, J., Li, J., Doré, J., Ehrlich, S. 3 D. and Stamatakis, A. (2013) 'Metagenomic species profiling using universal 4 phylogenetic marker genes', Nature Methods, 10(12), 1196-1199. 5 6 Suzek, B. E., Wang, Y., Huang, H., McGarvey, P. B., Wu, C. H. and the UniProt, C. (2015) 7 'UniRef clusters: a comprehensive and scalable alternative for improving sequence 8 similarity searches', Bioinformatics, 31(6), 926-932. 9 10 Svartström, O., Alneberg, J., Terrapon, N., Lombard, V., de Bruijn, I., Malmsten, J., Dalin, 11 A.-M., El Muller, E., Shah, P., Wilmes, P., Henrissat, B., Aspeborg, H. and 12 Andersson, A. F. (2017) 'Ninety-nine de novo assembled genomes from the moose 13 (Alces alces) rumen microbiome provide new insights into microbial plant biomass 14 degradation', The ISME Journal. 15 16 Sylvester, J. T., Karnati, S. K., Yu, Z., Morrison, M. and Firkins, J. L. (2004) 'Development 17 of an assay to quantify rumen ciliate protozoal biomass in cows using real-time 18 PCR', J Nutr, 134(12), 3378-84. 19 20 Tajima, K., Aminov, R. I., Nagamine, T., Matsui, H., Nakamura, M. and Benno, Y. (2001) 21 'Diet-dependent shifts in the bacterial population of the rumen revealed with real- 22 time PCR', Applied and Environmental Microbiology, 67(6), 2766-2774. 23 24 Tap, J., Furet, J.-P., Bensaada, M., Philippe, C., Roth, H., Rabot, S., Lakhdari, O., 25 Lombard, V., Henrissat, B., Corthier, G., Fontaine, E., Doré, J. and Leclerc, M. 26 (2015) 'Gut microbiota richness promotes its stability upon increased dietary fibre 27 intake in healthy adults', Environmental Microbiology, 17(12), 4954-4964. 28 29 Tapio, I., Shingfield, K. J., McKain, N., Bonin, A., Fischer, D., Bayat, A. R., Vilkki, J., 30 Taberlet, P., Snelling, T. J. and Wallace, R. J. (2016) 'Oral Samples as Non- 31 Invasive Proxies for Assessing the Composition of the Rumen Microbial 32 Community', PLoS One, 11(3), e0151220. 33 34 Tapio, I., Snelling, T. J., Strozzi, F. and Wallace, R. J. (2017) 'The ruminal microbiome 35 associated with methane emissions from ruminant livestock', J Anim Sci 36 Biotechnol, 8, 7. 37 38 Tewoldebrhan, T. A., Appuhamy, J. A. D. R. N., Lee, J. J., Niu, M., Seo, S., Jeong, S. and 39 Kebreab, E. (2017) 'Exogenous β-mannanase improves feed conversion efficiency 40 and reduces somatic cell count in dairy cattle', Journal of Dairy Science, 100(1), 41 244-252. 42 43 Theodorou, M. K., Mennim, G., Davies, D. R., Zhu, W.-Y., Trinci, A. P. J. and Brookman, 44 J. L. (2007) 'Anaerobic fungi in the digestive tract of mammalian herbivores and 45

218

their potential for exploitation', Proceedings of the Nutrition Society, 55(3), 913- 1 926. 2 3 Thompson, S. (2015) The effect of diet type on residual feed intake and the use of infrared 4 thermography as a method to predict efficiency in beef bulls, unpublished thesis 5 University of Manitoba. 6 7 Törrönen, A., Harkki, A. and Rouvinen, J. (1994) 'Three-dimensional structure of endo- 8 1,4-beta-xylanase II from Trichoderma reesei: two conformational states in the 9 active site', The EMBO Journal, 13(11), 2493-2501. 10 11 Turnbaugh, P. J., Hamady, M., Yatsunenko, T., Cantarel, B. L., Duncan, A., Ley, R. E., 12 Sogin, M. L., Jones, W. J., Roe, B. A., Affourtit, J. P., Egholm, M., Henrissat, B., 13 Heath, A. C., Knight, R. and Gordon, J. I. (2009) 'A core gut microbiome in obese 14 and lean twins', Nature, 457(7228), 480-484. 15 16 Tymensen, L. D., Beauchemin, K. A. and McAllister, T. A. (2012) 'Structures of free-living 17 and protozoa-associated methanogen communities in the bovine rumen differ 18 according to comparative analysis of 16S rRNA and mcrA genes', Microbiology, 19 158(7), 1808-1817. 20 21 Tyson, G. W., Chapman, J., Hugenholtz, P., Allen, E. E., Ram, R. J., Richardson, P. M., 22 Solovyev, V. V., Rubin, E. M., Rokhsar, D. S. and Banfield, J. F. (2004) 'Community 23 structure and metabolism through reconstruction of microbial genomes from the 24 environment', Nature, 428, 37. 25 26 Van Soest, P. J., Robertson, J. B. and Lewis, B. A. (1991) 'Methods for dietary fiber, 27 neutral detergent fiber, and nonstarch polysaccharides in relation to animal 28 nutrition', Journal of Dairy Science, 74(10), 3583-3597. 29 30 Vanwonterghem, I., Evans, P. N., Parks, D. H., Jensen, P. D., Woodcroft, B. J., 31 Hugenholtz, P. and Tyson, G. W. (2016) 'Methylotrophic methanogenesis 32 discovered in the archaeal phylum Verstraetearchaeota', Nature Microbiology, 1, 33 16170. 34 35 Vardakou, M., Dumon, C., Murray, J. W., Christakopoulos, P., Weiner, D. P., Juge, N., 36 Lewis, R. J., Gilbert, H. J. and Flint, J. E. (2008) 'Understanding the Structural 37 Basis for Substrate and Inhibitor Recognition in Eukaryotic GH11 Xylanases', 38 Journal Of Molecular Biology, 375(5), 1293-1305. 39 40 Varga, G. A. and Kolver, E. S. (1997) 'Microbial and animal limitations to fiber digestion 41 and utilization', The Journal Of Nutrition, 127(5 Suppl), 819S-823S. 42 43 Veira, D. M. (1986) 'The role of ciliate protozoa in nutrition of the ruminant', J Anim Sci, 44 63(5), 1547-60. 45 46

219

Verberkmoes, N. C., Russell, A. L., Shah, M., Godzik, A., Rosenquist, M., Halfvarson, J., 1 Lefsrud, M. G., Apajalahti, J., Tysk, C., Hettich, R. L. and Jansson, J. K. (2008) 2 'Shotgun metaproteomics of the human distal gut microbiota', The ISME Journal, 3 3, 179. 4 5 Větrovský, T. and Baldrian, P. (2013) 'The Variability of the 16S rRNA Gene in Bacterial 6 Genomes and Its Consequences for Bacterial Community Analyses', Plos One, 7 8(2), e57923. 8 9 Wallace, R. J., Rooke, J. A., Duthie, C.-A., Hyslop, J. J., Ross, D. W., McKain, N., de 10 Souza, S. M., Snelling, T. J., Waterhouse, A. and Roehe, R. (2014) 'Archaeal 11 abundance in post-mortem ruminal digesta may help predict methane emissions 12 from beef cattle', Scientific Reports, 4. 13 14 Wallace, R. J., Rooke, J. A., McKain, N., Duthie, C.-A., Hyslop, J. J., Ross, D. W., 15 Waterhouse, A., Watson, M. and Roehe, R. (2015) 'The rumen microbial 16 metagenome associated with high methane production in cattle', BMC Genomics, 17 16, 1-14. 18 19 Wang, L., Hatem, A., Catalyurek, U. V., Morrison, M. and Yu, Z. (2013) 'Metagenomic 20 Insights into the Carbohydrate-Active Enzymes Carried by the Microorganisms 21 Adhering to Solid Digesta in the Rumen of Cows', Plos One, 8(11), 1-11. 22 23 Wang, S., Giller, K., Kreuzer, M., Ulbrich, S. E., Braun, U. and Schwarm, A. (2017) 24 'Contribution of Ruminal Fungi, Archaea, Protozoa, and Bacteria to the Methane 25 Suppression Caused by Oilseed Supplemented Diets', Frontiers in Microbiology, 26 8(1864). 27 28 Wang, Y. and McAllister, T. A. (2002) 'Rumen microbes, enzymes and feed digestion - A 29 review', Asian-Australas J Anim Sci, 15(11), 1659-1676. 30 31 Warton, D. I., Wright, S. T. and Wang, Y. (2012) 'Distance-based multivariate analyses 32 confound location and dispersion effects', Methods in Ecology and Evolution, 3(1), 33 89-101. 34 35 Weimer, P. J. (1992) 'Cellulose degradation by ruminal microorganisms', Critical Reviews 36 in Biotechnology, 12, 189-223. 37 38 Weimer, P. J. (1998) 'Manipulating ruminal fermentation: a microbial ecological 39 perspective', Journal of Animal Science, 76(12), 3114-3122. 40 41 Weimer, P. J. (2015) 'Redundancy, resilience, and host specificity of the ruminal 42 microbiota: implications for engineering improved ruminal fermentations', Frontiers 43 in Microbiology, 6(296). 44 45

220

Weimer, P. J., Lopezguisa, J. M. and French, A. D. (1990) 'EFFECT OF CELLULOSE 1 FINE-STRUCTURE ON KINETICS OF ITS DIGESTION BY MIXED RUMINAL 2 MICROORGANISMS INVITRO', Applied and Environmental Microbiology, 56(8), 3 2421-2429. 4 5 Weiss, S., Xu, Z. Z., Peddada, S., Amir, A., Bittinger, K., Gonzalez, A., Lozupone, C., 6 Zaneveld, J. R., Vazquez-Baeza, Y., Birmingham, A., Hyde, E. R. and Knight, R. 7 (2017) 'Normalization and microbial differential abundance strategies depend upon 8 data characteristics', Microbiome, 5. 9 10 Westerhuis, J. A., van Velzen, E. J. J., Hoefsloot, H. C. J. and Smilde, A. K. (2010) 11 'Multivariate paired data analysis: multilevel PLSDA versus OPLSDA', 12 Metabolomics, 6(1), 119-128. 13 14 Wilkens, C., Busk, P. K., Pilgaard, B., Zhang, W.-J., Nielsen, K. L., Nielsen, P. H. and 15 Lange, L. (2017) 'Diversity of microbial carbohydrate-active enzymes in Danish 16 anaerobic digesters fed with wastewater treatment sludge', Biotechnology for 17 Biofuels, 10(1), 158. 18 19 Wilkinson, T. J., Huws, S. A., Edwards, J. E., Kingston-Smith, A. H., Siu-Ting, K., Hughes, 20 M., Rubino, F., Friedersdorff, M. and Creevey, C. J. (2018) 'CowPI: A Rumen 21 Microbiome Focussed Version of the PICRUSt Functional Inference Software', 22 Frontiers in Microbiology, 9(1095). 23 24 Williams, A. G. and Coleman, G. S. (1997) 'The rumen protozoa' in Hobson, P. N. and 25 Stewart, C. S., eds., The Rumen Microbial Ecosystem, Dordrecht: Springer 26 Netherlands, 73-139. 27 28 Wilmes, P. and Bond, P. L. (2004) 'The application of two-dimensional polyacrylamide gel 29 electrophoresis and downstream analyses to a mixed community of prokaryotic 30 microorganisms', Environmental Microbiology, 6(9), 911-920. 31 32 Wimberly, B. T., Ramakrishnan, V. and White, S. W. (1997) 'The structure of ribosomal 33 protein S7 at 1.9 Å resolution reveals a β-hairpin motif that binds double-stranded 34 nucleic acids', Structure, 5(9), 1187-1198. 35 36 Wolff, S. M., Ellison, M. J., Hao, Y., Cockrum, R. R., Austin, K. J., Baraboo, M., Burch, K., 37 Lee, H. J., Maurer, T., Patil, R., Ravelo, A., Taxis, T. M., Truong, H., Lamberson, 38 W. R., Cammack, K. M. and Conant, G. C. (2017) 'Diet shifts provoke complex and 39 variable changes in the metabolic networks of the ruminal microbiome', 40 Microbiome, 5(1), 60. 41 42 Wolfger, B., Quinn, C., Torres, G. W., Taylor, M. and Orsel, K. (2016) 'Comparison of 43 feeding behavior between black and red Angus feeder heifers', Canadian Journal 44 of Animal Science, 96(3), 404-409. 45 46

221

Wood, D. E. and Salzberg, S. L. (2014) 'Kraken: ultrafast metagenomic sequence 1 classification using exact alignments', Genome Biology, 15(3), R46-R46. 2 3 Wu, L., Wang, H., Zhang, Z., Lin, R., Zhang, Z. and Lin, W. (2011) 'Comparative 4 Metaproteomic Analysis on Consecutively Rehmannia glutinosa-Monocultured 5 Rhizosphere Soil', Plos One, 6(5), e20611. 6 7 Wu, S. and Zhang, Y. (2007) 'LOMETS: A local meta-threading-server for protein 8 structure prediction', Nucleic Acids Research, 35(10), 3375-3382. 9 10 Xu, B., Xu, W., Li, J., Dai, L., Xiong, C., Tang, X., Yang, Y., Mu, Y., Zhou, J., Ding, J., Wu, 11 Q. and Huang, Z. (2015) 'Metagenomic analysis of the Rhinopithecus bieti fecal 12 microbiome reveals a broad diversity of bacterial and glycoside hydrolase profiles 13 related to lignocellulose degradation', BMC Genomics, 16(1), 174. 14 15 Yang, J., Roy, A. and Zhang, Y. (2013) 'Protein–ligand binding site recognition using 16 complementary binding-specific substructure comparison and sequence profile 17 alignment', Bioinformatics, 29(20), 2588-2595. 18 19 Yang, J., Yan, R., Roy, A., Xu, D., Poisson, J. and Zhang, Y. (2014) 'The I-TASSER Suite: 20 protein structure and function prediction', Nature Methods, 12, 7. 21 22 Yang, T., Wei, Z., Friman, V.-P., Xu, Y., Shen, Q., Kowalchuk, G. A. and Jousset, A. 23 (2017) 'Resource availability modulates biodiversity-invasion relationships by 24 altering competitive interactions', Environmental Microbiology, 19(8), 2984-2991. 25 26 Yi, L., Shi, S., Wang, Y., Huang, W., Xia, Z.-a., Xing, Z., Peng, W. and Wang, Z. (2016) 27 'Serum Metabolic Profiling Reveals Altered Metabolic Pathways in Patients with 28 Post-traumatic Cognitive Impairments', Scientific Reports, 6, 21320. 29 30 Yin, Y. B., Mao, X. Z., Yang, J. C., Chen, X., Mao, F. L. and Xu, Y. (2012) 'dbCAN: a web 31 resource for automated carbohydrate-active enzyme annotation', Nucleic Acids 32 Research, 40(W1), W445-W451. 33 34 Yu, Z. and Morrison, M. (2004) 'Improved extraction of PCR-quality community DNA from 35 digesta and fecal samples', Biotechniques, 36(5), 808-812. 36 37 Zehavi, T., Probst, M. and Mizrahi, I. (2018) 'Insights Into Culturomics of the Rumen 38 Microbiome', Frontiers in Microbiology, 9(1999). 39 40 Zhang, Y. (2008) 'I-TASSER server for protein 3D structure prediction', Bmc 41 Bioinformatics, 9(1), 40. 42 43 Zhang, Y. H. P., Himmel, M. E. and Mielenz, J. R. (2006) 'Outlook for cellulase 44 improvement: Screening and selection strategies', Biotechnology Advances, (5), 45 452. 46

222

1 Zhou, M., Chung, Y. H., Beauchemin, K. A., Holtshausen, L., Oba, M., McAllister, T. A. 2 and Guan, L. L. (2011) 'Relationship between rumen methanogens and methane 3 production in dairy cows fed diets supplemented with a feed enzyme additive', 4 Journal of Applied Microbiology, 111(5), 1148-1158. 5 6 Zhou, M., Hernandez-Sanabria, E. and Guan, L. L. (2009) 'Assessment of the Microbial 7 Ecology of Ruminal Methanogens in Cattle with Different Feed Efficiencies', 8 Applied and Environmental Microbiology, 75(20), 6524-6533. 9 10 Zhou, M., Hernandez-Sanabria, E. and Guan, L. L. (2010) 'Characterization of Variation 11 in Rumen Methanogenic Communities under Different Dietary and Host Feed 12 Efficiency Conditions, as Determined by PCR-Denaturing Gradient Gel 13 Electrophoresis Analysis', Applied and Environmental Microbiology, 76(12), 3776- 14 3786. 15 16 Zhou, M., Hünerberg, M., Beauchemin, K. A., McAllister, T. A., Okine, E. K. and Guan, L. 17 L. (2012) 'Individuality of ruminal methanogen/protozoa populations in beef cattle 18 fed diets containing dried distillers' grain with solubles', Acta Agriculturae 19 Scandinavica: Section A, Animal Science, 62(4), 273-288. 20 21 Zhou, M., Peng, Y.-J., Chen, Y., Klinger, C. M., Oba, M., Liu, J.-X. and Guan, L. L. (2018) 22 'Assessment of microbiome changes after rumen transfaunation: implications on 23 improving feed efficiency in beef cattle', Microbiome, 6(1), 62-62. 24 25 Zuur, A. F., Ieno, E. N., Walker, N. J., Saveliev, A. A. and Smith, G. M. (2009) 'Mixed 26 Effects Models and Extensions in Ecology with R' in Mixed Effects Models and 27 Extensions in Ecology with R. 28 29 30 31

223