Investigation of Genomic Regions Affecting Ketosis in Dairy Cattle By

Investigation of Genomic Regions Affecting Ketosis in Dairy Cattle

by Riani Ananda Nunes Soares

A Thesis presented to The University of Guelph

In partial fulfilment of requirements for the degree of Master of Science in Animal Biosciences

Guelph, Ontario, Canada © Riani Ananda Nunes Soares, September, 2020 ABSTRACT

INVESTIGATION OF GENOMIC REGIONS AFFECTING KETOSIS IN DAIRY CATTLE

Riani Ananda N. Soares Advisor: University of Guelph, 2020 Dr. E James Squires Co-advisor: Dr. Flavio Schenkel

Ketosis is a severe metabolic disease caused by negative energy balance (NEB) that leads to decreased milk production. Thus, the main objectives were to investigate literature information on the genetic background of NEB, subclinical and clinical ketosis through a systematic review and meta-analysis and to identify genomic regions associated with subclinical and clinical ketosis in first and later lactations, and also for subclinical ketosis in two periods in early lactation (5 to

21 DIM and 22 to 45 DIM) by performing genome wide association studies (GWAS). The weighted single-step GBLUP (wssGBLUP) method was used for GWAS to identify genetic regions associated with the traits. The genes identified within the top-20 genomic windows were used for functional analyses. Biological understanding of this metabolic disease will help enhance accuracy of selection in the Canadian dairy cattle breeding programs and help with the development of biomarkers for early diagnosis and prevention of ketosis.

iii

ACKNOWLEDGEMENTS

First, I would like to thank my advisor, Dr. James Squires, who trusted me and gave me the opportunity to pursue my master’s degree in Canada, starting a new chapter of my life. Thank you for always having your door open, for trusting me with new ideas and for making me challenge myself every step of the way. I am tremendous grateful for all your support and for everything I learned from you.

I also want to thank the members of my advisory committee, Dr. Flavio Schenkel and Dr.

Todd Duffield, for all your help along this journey. I am grateful for having the opportunity to meet Dr. Schenkel. You are an inspiring person and a professional everyone should have as a role model. I appreciate all your support since the first day I went to the Department of Animal

Biosciences, even when I needed to have a surgery and your kind words made this process smoother. Thank you for always been available and for being a great person.

I can only say thank you one million times to Dr. Giovana Vargas. Without her, I would not be where I am today. You helped me put this thesis together, and you were with me every step of the way, teaching me, giving me advices, cheering for my success, and even sharing our best times in the department. I found an amazing mentor and a great friend. Most of all, thank you for being so patient and for teaching me how to be a better person and a better professional.

Dr. Angela Canovas, I am grateful for the opportunity you gave me to be your TA, for making me grow as a researcher, for all the knowledge shared, for the patience, and for helping me to become a better professional. Thank you for the conversations and for the trust placed in me.

At this time, I can only be grateful, and I can not thank enough the friends I made here and helped me go through the challenges of living abroad. In special, I would like to say thank you to

Malane Muniz, Hinayah Rojas, Tatiane Chud, Ivan Campos, Gerson Oliveira, Lucas

Alcantara, Pablo Fonseca, Amelia Soares, Samla Marques, Mariana Berton, Bruna Mion,

Bianca Kato, Larissa Costa. You were my family in Canada!

I also want to express my sincere gratitude to my lab mates, Christine Bone and Jocelyn

Cameron, who I had the pleasure of working alongside and to have the best times in the lab and in our trip to Texas. I have learned the best jokes from you! Special thanks to Yanping Lou, for all your help in the lab and for always being available and with a smile on your face.

I thank my parents, Rivaldo Soares and Nivani Soares, for all the times they put me first in their lives, abdicating your dreams to give me the opportunity to pursue mine. For allowing me to grow away from you, and even so, to be present even when I was away. You taught me the importance of being an honest person and being tankful for what I have in life. Your trust and love made me get here. I especially thank my mother for always believing in me and for being such a strong woman that makes me want to fight for the best always. To my brothers, Rivaldo Filho and

Rian Soares, for forgiving all the absence and for always helping me when I felt lost. Thank you for your unconditional love and support always. It was very difficult to spend so much time away from you.

To the Nunes family, for all their understanding and support, believing that one day everything would be all right. You have been my “saudade” for all those years away. Thank you for every encouragement and for every word of motivation, for every time I called crying and you consoled me, for every joke and for every hug in our reunions. With a unique and special way, you helped me on this journey. I know that wherever I go, I will never be alone, because I have you.

I need to acknowledge Shakeel Ashraf for being with me during the best and worst moments of this journey. I am grateful for meeting you here, and for sure you made it more special.

Thank you for always encouraging me to seek the best, for rooting for my happiness, for being my best friend and for being my accomplice at all times.

To those who were always my companions, Thais Magalhaes, Herbeson Martins, Eden

Souza, Jadilson Mariano, Lais Maia, and Verenna Barros, thank you for being part of this achievement. Defying distance, time, and logic, we have become closer than ever. I would also like to thank Felipe Ivo for these seven years of friendship and for never let me feel alone in

Canada.

It is not possible to give thanks without thanking God. Who gives me the strength to believe and move forward, fighting for my goals and believing in a better future. Thank you for all the opportunities and the signs that made me get where I am, for guiding and protecting me.

Finally, I thank all those people who, directly or indirectly, were part of my life during this degree and those who contributed to the accomplishment of this work. vi

TABLE OF CONTENTS

ABSTRACT ...... ii2

ACKNOWLEDGEMENTS ...... i3ii

TABLE OF CONTENTS ...... v6i

LIST OF TABLES ...... i1x

LIST OF FIGURES ...... x1i

LIST OF ABBREVIATIONS ...... xi1 ii

CHAPTER ONE: General Considerations ...... 1 Introduction ...... 1 Objectives ...... 2 Literature Review...... 3 Overview of Ketosis ...... 3 Ketosis Categorization ...... 4 Etiology and Pathogenesis ...... 5 Epidemiology ...... 8 Diagnosis and Clinical Findings ...... 9 Impact of the disease ...... 11 Treatment ...... 11 Genetic control of Ketosis ...... 14 Genome-wide Association Studies ...... 15 Conclusion ...... 17 References ...... 17

CHAPTER TWO: Differential gene expression in dairy cows under negative energy balance and ketosis: A systematic review and meta-analysis ...... 31 Abstract ...... 31 Introduction ...... 33 Material and Methods ...... 35 Search of the Published Literature ...... 35 Metabolic Pathway Analysis...... 36 Venn Diagrams and Enrichment Analysis ...... 37

vii

QTL Enrichment Analysis ...... 38 Results ...... 38 Functional Enrichment Analysis ...... 42 QTL Enrichment Analysis ...... 43 Discussion ...... 43 Conclusions ...... 51 Acknowledgments...... 52 References ...... 52 Tables and Figures ...... 65

CHAPTER THREE: Genome-wide association study and functional analyses of clinical and subclinical ketosis in Holstein cattle ...... 73 Abstract ...... 73 Introduction ...... 74 Materials and Methods ...... 76 Animals and dataset description ...... 76 Genotypic data and quality control ...... 77 De-regressed EBV ...... 77 Genome-wide association analyses...... 78 Functional annotation analyses ...... 79 Results ...... 80 Genome-wide association analyses...... 80 Function annotation and gene network analyses ...... 81 Discussion ...... 83 Clinical ketosis in first lactation (CK1) ...... 84 Clinical ketosis in later lactations (CK2) ...... 86 Subclinical ketosis in first lactation (SCK1) ...... 88 Subclinical ketosis in later lactations (SCK2) ...... 90 Overlapping regions and genes among the traits ...... 93 Functional annotation...... 95 Conclusions ...... 97 Acknowledgments...... 98 References ...... 98 vii i

Tables and Figures ...... 123 Supplementary material ...... 138

CHAPTER FOUR: Genome-wide association analysis for subclinical ketosis in two periods in early lactation in Holstein dairy cattle ...... 145 Abstract ...... 145 Introduction ...... 147 Materials and Methods ...... 148 Animals and dataset description ...... 148 Variance Component Estimation ...... 149 Genotypic data and quality control ...... 149 De-regressed EBV ...... 150 Genome-wide association studies (GWAS)...... 150 Functional annotation analyses ...... 152 Results ...... 152 Estimation of genetic parameters ...... 152 Genome-wide association analyses (GWAS) ...... 152 Functional annotation...... 153 Discussion ...... 154 Estimation of genetic parameters ...... 154 Candidate genes from GWAS analysis ...... 155 For subclinical ketosis in first lactation from 5 to 21 DIM (SCK1.1) ...... 155 For subclinical ketosis in first lactation from 22 to 45 DIM (SCK1.2) ...... 156 For subclinical ketosis in later lactations from 5 to 21 DIM (SCK2.1) ...... 157 For subclinical ketosis for later lactations from 22 to 45 DIM (SCK2.2) ...... 159 Overlapping and differences among traits ...... 160 Conclusions ...... 165 Acknowledgments...... 166 References ...... 166 Tables and Figures ...... 186 Supplementary material ...... 198

CHAPTER FIVE: General Conclusions ...... 206 References ...... 210 ix

LIST OF TABLES Table 1. Significant SNPs and their position within the start and end coordinates of each differentially expressed gene obtained through systematic review ...... 65 Table 2. Significant pathways, genes and markers involved in ketosis from genome-wide association studies ...... 67 Table 3. Significant differentially expressed genes with significant SNPs in published GWAS for NEB, subclinical and clinical ketosis, chromosome, SNP identification, SNP position, and number of SNPs harbored by the genes included in the Illumina bovine SNP60 BeadChip ...... 67 Table 1. Top-20 sliding windows explaining the highest proportion of genetic variance for clinical ketosis 1 (CK1)...... 123 Table 2. Top-20 sliding windows explaining the highest proportion of genetic variance for clinical ketosis 2 (CK2)...... 124 Table 3. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis 1 (SCK1)...... 124 Table 4. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis 2 (SCK2)...... 125 Table 5. Annotated genes within the top 20 sliding windows associated with subclinical and clinical ketosis obtained by wssGBLUP analyses...... 128 Table 6. Pathways and their genes obtained for clinical ketosis 1 (CK1)...... 132 Table 7. Pathways and their genes obtained for clinical ketosis 2 (CK2)...... 133 Table 8. Pathways and their genes obtained for subclinical ketosis 1 (SCK1)...... 133 Table 9. Pathways and their genes obtained for subclinical ketosis 2 (SCK2)...... 134 Supplementary table 1. Enriched molecular function, cellular component and biological process with Fisher's test for clinical ketosis 1 (CK1)...... 138 Supplementary table 2. Enriched molecular function, cellular component and biological process with Fisher's test for clinical ketosis 2 (CK2)...... 139 Supplementary table 3. Enriched molecular function, cellular component and biological process with Fisher's test for subclinical ketosis 1 (SCK1)...... 141 Supplementary table 4. Enriched molecular function, cellular component and biological process with Fisher's test for subclinical ketosis 2 (SCK2)...... 142 Table 1. Variance components and heritability estimates for SCK1.1, SCK1.2, SCK2.1 and SCK2.2 traits in Holstein cattle...... 186 Table 2. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis for first lactation from 5 to 21 DIM (SCK1.1)...... 187 Table 3. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis for first lactation from 22 to 45 DIM (SCK1.2)...... 188

Table 4. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis for later lactations (2 to 5) from 5 to 21 DIM (SCK2.1)...... 189 Table 5. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis for later lactations (2 to 5) from 22 to 45 DIM (SCK2.2)...... 190 Table 6. Annotated genes within the top-20 sliding windows associated with subclinical ketosis obtained from the wssGBLUP analyses...... 193 Supplementary Table 1. Enriched Molecular Function, Cellular component, and Biological process with Fisher's test for subclinical ketosis for first lactation from 5 to 21 DIM (SCK1.1)...... 198 Supplementary Table 2. Enriched Molecular Function, Cellular component, and Biological process with Fisher's test for subclinical ketosis for first lactation from 22 to 45 DIM (SCK1.2)...... 198 Supplementary Table 3. Enriched Molecular Function, Cellular component, and Biological process with Fisher's test for subclinical ketosis for later lactations (2 to 5) from 5 to 21 DIM (SCK2.1)...... 200 Supplementary Table 4. Enriched Molecular Function, Cellular component, and Biological process with Fisher's test for subclinical ketosis for later lactations (2 to 5) from 22 to 45 DIM (SCK2.2)...... 200 Supplementary Table 5. Pathways obtained for subclinical ketosis for first lactation from 5 to 21 DIM (SCK1.1)...... 202 Supplementary Table 6. Pathways obtained for subclinical ketosis for first lactation from 22 to 45 DIM (SCK1.2)...... 203 Supplementary Table 7. Pathways obtained for subclinical ketosis for later lactations (2 to 5) from 5 to 21 DIM (SCK2.1)...... 203 Supplementary Table 8. Pathways obtained for subclinical ketosis for later lactations (2 to 5) from 22 to 45 DIM (SCK2.2)...... 204

LIST OF FIGURES

Figure 1. Venn diagrams depicting genes differentially expressed in liver of dairy cows under NEB, subclinical ketosis (SCK) and clinical ketosis (CK). A) Depicting the genes found in transcriptomic studies. B) Depicting the genes found in proteomic studies...... 69

Figure 2. Venn diagrams depicting differentially expressed genes detected using gene expression analysis in dairy cattle under negative energy balance (NEB), subclinical and clinical ketosis conditions. A) Number of common genes among transcriptomic techniques. B) Number of common genes between transcriptomic and proteomic studies...... 70

Figure 3. Number of common differentially expressed genes detected by different techniques in dairy cattle under NEB and clinical ketosis conditions. A) Genes detected in the clinical ketosis condition. B) Genes detected in the NEB condition...... 70

Figure 4. Network integration of candidate genes related to NEB, subclinical and clinical ketosis in dairy cattle. The colorful pie charts show the biological function associated (FDR <0.05) with the genes presented in the network, and the grey color in the pie charts indicate that these genes were not associated with any of the analyzed biological functions. The light grey stripes inside the pies indicate the 14 differentially expressed genes used as input for the network...... 71

Figure 5. Pie chart for 2,000 annotated QTL that overlap with the 14 differentially expressed genes identified for NEB, subclinical and clinical ketosis in dairy cattle. The different colors in pie chart show the phenotypic trait classes associated with the QTL and the numbers are the corresponding QTL frequencies...... 71

Figure 6. Specific QTL categories associated with “Health” QTL class out of 2,000 annotated QTL that overlap with the 14 differentially expressed genes identified for NEB, subclinical and clinical ketosis in dairy cattle...... 72

Figure 1. Genetic variance explained by 20-SNP sliding windows for clinical ketosis 1 (CK1). Chromosome number is shown on the horizontal axis...... 126

Figure 2. Genetic variance explained by 20-SNP sliding windows for clinical ketosis 2 (CK2). Chromosome number is shown on the horizontal axis...... 127

Figure 3. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis 1 (SCK1). Chromosome number is shown on the horizontal axis...... 127

Figure 4. Genetic variance explained by 20-SNP windows for subclinical ketosis 2 (SCK2). Chromosome number is shown on the horizontal axis...... 128

Figure 4. Gene network analysis for clinical ketosis 1 (CK1)...... 135

Figure 5. Gene network analysis for clinical ketosis 2 (CK2)...... 136

xii

Figure 6. Gene network analysis for subclinical ketosis 1 (SCK1)...... 137

Figure 7. Gene network analysis for subclinical ketosis 2 (SCK2)...... 138

Figure 1. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis for first lactation from 5 to 21 DIM (SCK1.1). Chromosome number is shown on the horizontal axis. 191

Figure 2. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis for first lactation from 22 to 45 DIM (SCK1.2). Chromosome number is shown on the horizontal axis...... 191

Figure 3. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis for later lactations (2 to 5) from 5 to 21 DIM (SCK2.1). Chromosome number is shown on the horizontal axis...... 192

Figure 4. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis for later lactations (2 to 5) from 22 to 45 DIM (SCK2.2). Chromosome number is shown on the horizontal axis...... 192

Figure 5. Main metabolic pathways associated to SCK1.1, SCK1.2, SCK2.1 and SCK2.2. .... 193

xiii

LIST OF ABBREVIATIONS

AMS Automatic Milking Systems BHB β-hydroxybutyrate CDN Canadian Dairy Network Chr Chromosome CK1 Clinical Ketosis for fist lactation CK2 Clinical Ketosis for later lactations DA Displaced Abomasum dEBV De-regressed Estimated Breeding Value DIM Days in Milk DMI Dry Matter Intake DNA Deoxyribonucleic acid EBV Estimated Breeding Value ER Endoplasmic reticulum FFA Free Fatty Acid GALLO Genomic Annotation in Livestock for Positional Candidate LOci GBLUP Genomic best linear unbiased prediction GEBV Genomic Breeding Value GH Growth Hormone GO Gene Ontology GWAS Genome Wide Association Study HWE Hardy-Weinberg Equilibrium KEGG Kyoto Encyclopedia of Genes and Genomes KET Ketosis LD Linkage Disequilibrium LDL Low-density Lipoproteins MAF Mino Allele Frequency MIR Mid-Infrared Spectroscopy NCBI National Center for Biotechnology Information NEB Negative Energy Balance NEFA Non-esterified fatty acid PPI Protein-Protein Interaction QC Quality Control qPCR Quantitative Polymerase Chain Reaction RNA-Seq RNA sequencing SCK1 Subclinical Ketosis for fist lactation SCK2 Subclinical Ketosis for later lactations SCK1.1 Subclinical Ketosis for fist lactation, 1st DIM period SCK1.2 Subclinical Ketosis for fist lactation, 2nd DIM period SCK2.1 Subclinical Ketosis for later lactations, 1st DIM period SCK2.2 Subclinical Ketosis for later lactations, 2nd DIM period SNP Single Nucleotide Polymorphism

xiv

QTL Quantitative Trait Loci VLDL Very low-density Lipoproteins wssGBLUP Weighted single step GBLUP

CHAPTER ONE: General Considerations

Introduction

Among the production diseases that affect the health of dairy cows is ketosis, a metabolic disease that directly interferes with the peak of milk production. In Canada, there is a great interest in studying ketosis, especially to understand the biochemical mechanisms by which energy deficiency affects productivity and the genetic factors that lead some cows to develop ketosis year after year (Duffield, 2000; Duffield et al., 2009; Kroezen et al., 2018; Nayeri et al., 2019).

Most of the time, the cows are naturally capable of recovering from negative energy balance (NEB) and do not develop subclinical or clinical ketosis, and it is the cow’s ability to adapt to NEB that determines the occurrence, severity and duration of this period. However, by the time this occurs, production could have been affected. Thus, preventing ketosis is the best way to avoid economic losses since, on average, each case of subclinical and clinical ketosis can result in loss of $203 and $289, respectively (McArt et al., 2015, Gohary et al., 2016). For this, genomic data is a potential aid to accurately select animals for economically important traits, such as less susceptibility to diseases. Through genome-wide association studies (GWAS) and other genomic data analyses, regions of interest in the genome can reveal possible candidate genes and variants

(Ku et al., 2010; McCarthy et al., 2008).

Hence, the literature on ketosis and its genetic components will be reviewed here. The etiology and pathogenesis of ketosis in dairy cattle as well as other aspects of this disease will be discussed to comprehend its mechanisms more clearly and how it affects productivity.

Furthermore, the potential genetic control of ketosis through breeding programs and the applications of genomics technologies to potentially reduce ketosis will be examined.

Objectives

The main objective of this research was to study clinical and subclinical ketosis in a genomic perspective and determine genomic regions with a greater effect on ketosis susceptibility expression in dairy cattle through association analyses.

Chapter 2 evaluates the pattern of differential gene expression in liver of cows in negative energy balance (NEB), and under subclinical and clinical ketosis through a systematic review and meta-analysis of gene expression and genome-wide association studies (GWAS) results.

Chapter 3 use the de-regressed breeding values (dEBVs) as pseudo-phenotypes to perform genome-wide association studies (GWAS) for clinical and subclinical (BHB concentrations) ketosis for first and later lactations in Holstein cattle in order to find regions of the genome that could explain the genetic variability for these traits. Regions of the genome that are found to be significantly associated with the four traits are further investigated in a functional analysis to narrow down possible candidate genes.

Chapter 4 focuses on calculating variance components and association analysis for subclinical ketosis (BHB concentrations) in Holstein cattle for 5 to 21 DIM and 22 to 45 DIM in first and later lactations using dEBVs calculated from BHB concentrations. Next, genomic regions found to be associated with the traits on the GWAS are further investigated in a functional analysis to create a list of possible candidate genes.

Chapter 5 presents a general discussion of the results obtained in the previous chapters.

This chapter also summarizes the future opportunities for the genomic selection for resistance to clinical and subclinical ketosis in Holstein cattle and the opportunities for future research projects with these traits.

Literature Review

Overview of Ketosis

The phase of a cow's life when she is most susceptible to disease is the transition period

(Drackley, 1999). Some authors differ on what they consider to be the transition period. Some define it as the last three weeks before parturition to the three weeks after parturition (Block & Co,

2010; Drackley, 1999; Grummer, 1995), while others characterize it as starting at dry-off and extending until at least one month after calving. Regardless of its exact length, all agree that the cow experiences substantial metabolic, hormonal, immunologic, and physiologic changes during this period (Van Dorland et al., 2009).

The demands of growing the calf, the body’s preparation to start lactation, constraints on the intake of dry matter, and the greater predisposition to mobilize body fat are all factors that can put the animal into negative energy balance (NEB). Therefore, the body needs to utilize other sources of energy to meet this deficit. Primarily the fat reserves are mobilized to support this energy demand, which causes weight loss. Thus, the cow's body enters a stage of hypoglycaemia, stimulating the mobilization of free fatty acids (FFA) and glycerol from fat deposits and then the production of ketone bodies starts as a compensation for the lack of glucose for the tissues (Leek

& Reece, 2014).

The adaptation of ruminants to NEB is challenging because they utilize non-carbohydrate compounds in gluconeogenesis to meet their energy needs (Andersson, 1988; Sun et al., 2014).

Cows that fail to have a good adaption may, therefore, experience metabolic disturbances, such as ketosis. Ketosis is a serious metabolic disease caused by an energetic disequilibrium due to the high energy demand for milk production and to maintain their body functions. This disease mainly affects high yielding and lactating cows, and is defined by higher concentrations of ketone bodies

– acetone, acetoacetic acid, and beta-hydroxybutyrate (BHB) – in blood, urine and milk due to the 3

incomplete metabolism of lipids (Andrews et al., 2004; Vicente et al., 2014) and it is one of the most frequent metabolic diseases in dairy cows, with high economic consequences to the industry

(Duffield, 2000; González & Silva, 2017). Although the changes that occur during the transition period have been extensively studied, ketosis is still a prevalent disease in dairy herds, and a significant problem for the cow and producers.

Ketosis Categorization

Ketosis was formerly classified into three theorized types: Type I, Type II and Butyric Acid

Silage, with each type having a different etiology (Herdt, 2000; Oetzel, 2007). Type I is characterized as a spontaneous form, with the highest risk period lasting from three to six weeks post-calving, when the glucose demand exceeds the gluconeogenesis capacities of the liver. Thus, gluconeogenic pathways are maximally stimulated, but the supply of glucose precursors is insufficient to allow maximum production of glucose. Type I ketosis presents with high levels of

BHB, non-esterified fatty acids (NEFA) and liver gluconeogenesis, while glucose and insulin blood levels are low stimulating rapid entry of NEFAs into hepatic mitochondria. In this type, the animal is insulin-dependent with a good prognosis, and there is no pathology in the liver (Herdt,

2000).

For Type II, large amounts of NEFAs are transported to the liver, but gluconeogenesis and ketogenesis are not maximally stimulated. In this case, overly fat cows are at higher risk, and they present a fatty liver because of the high amounts of NEFA overwhelming their liver. The levels of

BHB, NEFA, glucose and insulin are high, while gluconeogenesis is low and resistant to insulin thus mitochondrial uptake of NEFA is not as active as in type I ketosis. The risk period lasts from one to two weeks after the start of lactation, and the animal presents a favorable prognosis

(González & Silva, 2017; Herdt, 2000). In addition, this type of ketosis is generally associated

with the appearance of other diseases such as displaced abomasum (DA) and mastitis (Andrews et al., 2004). Thirdly, the Butyric type is characterized by feeding cows silage with high concentrations of butyric acid (daily doses of over 50 to 100g) (Andersson, 1988). Approximately

75% of the additional ruminal butyrate that the cow can not metabolize is released as BHB into the blood (Sakha et al., 2007). This type shows variable pathology in the liver, high levels of BHB, normal or high level of NEFA, and variable levels for glucose, insulin and gluconeogenesis. The prognosis and the period of risk are also variable; however, particularly in early lactation cows, the animals do not eat the forage well because of abnormal fermentation thus presenting lowered dry matter intake which is associated with poorly fermented forage and becomes a major factor in causing ketosis (González & Silva, 2017; Herdt, 2000).

Another form of classification is according to the extent of clinical signs where ketosis can either be classified as clinical or subclinical ketosis (Duffield, 2000). Clinical ketosis is characterized by an increased concentration of ketone bodies in blood, urine, or milk and recognizable symptoms such as a lack of appetite and rapid weight loss (Itle et al., 2015).

Subclinical ketosis is defined as an excess of ketone bodies in blood and other fluids, which increases the risk of development of other diseases and unfavorable outcomes (Andersson, 1988;

Duffield, Lissemore, McBride, & Leslie, 2009). Clinical symptoms are difficult to identify in dairy cows; therefore, the measurement of ketone bodies in the fluids is a meaningful result to distinguish these two types of ketosis.

Etiology and Pathogenesis

Several factors may influence the appearance of ketosis in cows, including feeding too much silage with high butyric acid content, a poor-quality forage, sudden changes in feeding, and fat accumulation in the liver. This is because fat cows consume less dry matter at the beginning of

lactation and have more fat to mobilize, thus are more likely to have ketosis (Andrews et al., 2004;

Sakha et al., 2007).

Other characteristics of the pathology are the reduction of blood sugar, accumulation of fat in the liver, lack of appetite, increase of free fatty acids in blood plasma and a reduction in the level of hepatic glycogen: all characteristics associated with the energy deficit caused by milk production during lactation (Andrews et al., 2004).

Ketosis usually occurs within a few days or weeks after calving, with the highest incidence being in the second week postpartum, typically around 11 or 12 days in milk (DIM). The fact that some cows develop ketosis year after year suggests that some animals are more likely to develop ketosis than others, probably because of genetic causes (Shaw, 1956).

During the transition period, there is a decline of cow`s food intake (Bertics et al., 1992).

Both the decrease in dry matter intake and the endocrine changes that occur at the end of gestation and the beginning of lactation, influence the cow’s metabolism (Grummer, 1995). Due to these conditions and having a high demand of energy, the cows initiate the mobilization of body reserves to supply the energetic demand. Hypoglycaemia will then lead to an increase in free fatty acid

(FFA) levels by the mobilization of fat stores (Bergman, 1971), caused by the reduction of the sensitivity of the fatty tissue to insulin, along with the reduction of the plasma insulin concentration

(Doepel et al., 2002). However, insulin reduction does not appear to affect glucose uptake by the mammary gland (Kronfeld et al., 1963). This way, due to homeorrhetic mechanisms, the lactating cow prioritizes the delivery of nutrients towards the mammary gland (Bauman & Currie, 1980).

An increase in growth hormone concentration (GH) during early lactation stimulates hepatic gluconeogenesis and creates an insulin resistance, which prevents glucose utilization by the liver, muscle, or adipose tissue and stimulates lipolysis, which mobilizes fatty acids (mainly NEFA) for

milk fat synthesis or to be used as an energy source to some extent in the postpartum cow

(Wankhade et al., 2017).

The use of body fat for the production of energy leads to the formation of NEFAs that will be completely oxidized, generating the metabolite acetyl coenzyme A that can be used to generate energy by the Krebs cycle (Esposito et al., 2014). However, there is a limit to the amount of fatty acid that can be oxidized to completion by the tricarboxylic acid cycle of the liver. When this limit is reached, triglycerides accumulate within the hepatocytes, and the acetyl-coenzyme A that is not incorporated into the tricarboxylic acid cycle is converted to acetoacetate and β-hydroxybutyrate

(Goff & Horst, 1997). Consequently, blood ketone concentrations increase, and cows may even excrete these ketones in the urine and milk (Andrews et al., 2004). The increase of ketone bodies combined with lipolysis and mobilization of fatty acids leads to hyperketonemia and acid-base balance alterations (Foss-Freitas & Foss, 2003).

An important factor in determining ketosis is the severe carbohydrate deficiency that develops when a susceptible animal attempts to maintain milk production in the presence of a marked deficiency in glucose supply (Baird, 1982) and impaired gluconeogenesis, resulting in hypoglycaemia (Goff & Horst, 1997).

In the scenario described above, if the NEFA increase in plasma is sufficiently high, this contributes to the development of fatty liver, which in itself contributes to other problems, such as predisposition of cows to clinical ketosis and displaced abomasum in the postpartum period

(Grummer, 1995). Increased fat in the liver along with decreased glycogen can lead to liver damage

(Littledike et al., 1981). The animal begins to show reduced weight, milk production, and in some cases, there are symptoms of the nervous system.

Several studies reported that NEFA and BHB are responsible for a suppression of immune responsiveness that consequently leads to the occurrence of metabolic and infectious diseases, exposing the cow to greater risks. Therefore, ketosis is often accompanied by several other complications, such as metritis and retained placenta (Bertoni et al., 2008; Bionaz et al., 2007;

Sordillo & Mavangira, 2014). This makes diagnosis of ketosis difficult, although it is extremely important to know the cause of the disease so that the treatments are effective (Shaw, 1956).

Epidemiology

The incidence of clinical ketosis in North America varies greatly and depends on many factors including the method of identification. A range of 2 to 15% incidence of clinical ketosis was identified by Divers & Peek (2008) and Duffield (2000), but Duffield et al. (1997) and Radostits et al.(2007) reported a prevalence of 0.2 to 10%. McArt et al. (2012) found an average incidence of both clinical and subclinical ketosis of 43% of cows during the first two weeks of lactation. The incidence of the subclinical ketosis in the United States and Canada was reported through milk test between 12.1% and 61%, which is an indicator of variation (Dohoo & Martin, 1984; McLaren et al., 2006).

Different studies conducted in Ontario, Canada over the years showed different prevalence proportions. Duffield et al. (1997) determined a prevalence of 14.1% of subclinical ketosis according to BHB concentrations, whereas a prevalence of 12% was reported by Geishauser et al.

(2000) in a study also based on serum BHB. In cases of subclinical ketosis, the drop in milk production could reach 1 to 1.5 liters per day (Duffield et al., 1997), provoking considerable economic losses.

Regarding herd management practices, one of the main risk factors is the automatic milking systems (AMS) which was shown to be associated with increased within-herd prevalence of

ketosis (Tatone et al., 2017). Other risk factor associated with ketosis are older age at first calving for primiparous animals especially if exceeding 24 months, and for subclinical ketosis dry-period duration exceeding 72 days (Tatone et al., 2017; Vanholder et al., 2015). In addition, parity number also affects the incidence of ketosis. Vanholder et al. (2015) showed that there is a higher risk of developing subclinical ketosis in parity 2 whereas for clinical ketosis parity 3 or higher. However, other studies have shown increased risk with increased parity (Berge and Vertenten, 2014; Suthar et al., 2013).

Diagnosis and Clinical Findings

Depending on the type of ketosis, its diagnosis is based on a number of factors, including clinical signs; time in lactation; determination of ketone bodies in the blood, urine and milk; hypoglycemia; and ketonemia (Duffield et al., 2009; Gordon et al., 2013; Riet-Correa et al., 2002).

The disease is characterized by weight loss, reduced milk production, and in some cases there are effects on the nervous system (Herdt, 2000). In the nervous form of ketosis, acute neurological signs are observed, such as excessive salivation, abnormal chewing movements, body licking, ataxia, motor incoordination, apparent blindness, hyperesthesia, which is excessive physical sensitivity, and tetanic seizure, along with decreased levels of glucose and calcium in the serum (Reddy et al., 2014).

It is very difficult to determine whether an animal is exhibiting clinical symptoms of ketosis without a routine monitoring program in place because it may be quite challenging to identify clinical signs amongst a group of animals. The breath of the cow may have a smell of acetone

(which not every individual is capable of detecting), and the cow’s stool is frequently firm and dry. There are typically no alterations in rectal temperature, heart rate and respiration (Andrews et al., 2004). The classification between clinical and subclinical ketosis is primarily due to the

presence of clinical symptoms and BHB levels in the blood, and distinguishing this difference is not so clear (Duffield et al., 2009; Gordon, LeBlanc, & Duffield, 2013). Gordon et al. (2013) examined animals with high levels of ketonemia that showed no clinical symptoms and animals with low levels that presented clear signs of illness.

In ruminants, the normal blood glucose concentration is 2.2 to 3.3 mmol/L (Mair et al.,

2016). Reddy et al. (2014) performed biochemical analyses in ketotic cows that showed 1.6 mmol/L and 0.5 mmol/L concentrations of glucose and calcium in the blood, respectively. Levels below the threshold for blood glucose concentration (2.2 to 3.3 mmol/L) may be indicative of ketosis and clinical symptoms may be evident, but it is worth emphasizing that the severity of the disease depends on the duration and degree of hypoglycemia.

The most used method of diagnosis of ketosis in dairy cattle in research is a blood test for

BHB, but urine and milk BHB tests are the most utilized in the field. It is necessary to perform tests regularly, especially during the first few weeks after calving. A benchmark test is measuring

BHB concentrations in blood as an indicator of ketosis presence, and the thresholds reported are between 1-1.4 mmol/L during postpartum for subclinical ketosis. Animals with values equal to or greater than 1.4 mmol/L BHB are three times more predisposed to developing clinical ketosis

(Duffield et al., 1997, 2009; Iwersen et al., 2013; Rollin, Berghaus, Rapnicki, Godden, & Overton,

2010).

Due to the difficulty in recognizing clinical symptoms and diagnosing the disease, misrepresentations of the number of cows with subclinical ketosis or misdiagnosed clinical ketosis make it difficult to address the problem. This way, it is necessary to be careful when diagnosing ketosis so appropriate treatment can be administered.

Impact of the disease

A significant part of the cattle health and resistance to diseases is due to genetics, and a genetic correlation exists between ketosis and other traits. Heringstad et al. (2005) performed a genetic analysis of clinical mastitis, milk fever, ketosis, and retained placenta in Norwegian Red cows and found that the largest estimate of genetic correlations was between ketosis and milk fever

(0.40 ± 0.055). Therefore, indirect selection for a health trait correlated with ketosis may improve the accuracy of selection (Berry, Bermingham, Good, & More, 2011). On the other hand, selection for production traits may be affected by selection for ketosis resistance because they are negatively correlated provoking economic losses, making it necessary to better understand the genes involved.

There is an association of ketosis with other diseases, such as displaced abomasum and metritis. In general, these diseases are correlated with intensified culling at a period or price that may not be commercially desirable and also pose a financial loss (Fetrow et al., 2006; Gröhn et al., 1998). Early lactation cows also tend to have increased incidence of displaced abomasum (DA;

>8%) (Oetzel, 2007). High genetic correlation between DA and clinical ketosis (0.86) was found by Koeck et al. (2014). According to McArt et al. (2012), animals tested positive for subclinical ketosis from 3 to 5 DIM had a chance of developing DA 6.1 times higher than animals only testing positive at 6 DIM or later. Besides being correlated with DA, cows diagnosed with ketosis are also more likely to develop metritis (Duffield et al., 2009). Dohoo & Martin (1984) stated that subclinical ketosis was more likely to be found in cows experiencing metritis than in healthy cows.

Treatment

Restoring normal metabolism for milk production and normal appetite is the main focus of ketosis treatment. Several treatments have been developed to restore normal metabolism, such as

dextrose, propylene glycol, glucocorticoids and B vitamins, all with a therapeutic function of increasing the glucose supply and gluconeogenic intermediates for the animal (Baird, 1982;

Gordon et al., 2013a; Gordon, Duffield, et al., 2017; Gordon, LeBlanc, et al., 2017; Herdt &

Emery, 1992).

The first treatment option discussed here is the administration of dextrose or glucose intravenously, especially in cows with acute ketosis that exhibit nervous signs due to hypoglycaemia. The purpose of this therapy is to increase glucose levels and thereby provide the required glucose for the synthesis of lactose, to reduce the mobilization of body reserves. The dose typically administered is 500 mL of 50% dextrose, increasing the blood glucose concentration up to eight times, but the decrease in BHB levels caused by the administration of dextrose is of short duration, less than 24 hours (Gordon et al., 2013; Sakai et al., 1996; Wagner & Schimek, 2010).

However, this treatment should only be used as a short-term measure and not alone, perhaps associating with oral administration of glucose precursors such as propylene glycol (Gordon et al.,

2013; Wagner & Schimek, 2010).

The second treatment discussed here is propylene glycol, which is administered orally and serves as a glucose precursor to the animal because it generates propionate or it is directly absorbed in the rumen, serving as a gluconeogenesis enhancer (Gordon et al., 2013). After the administration of this treatment, insulin levels increase significantly, at around 15 to 30 minutes, and continue to increase gradually to 90 minutes and maintain the level for up to two hours (Miyoshi et al., 2001;

Studer et al., 1993). This spike in insulin helps decrease fat breakdown and hepatic ketone body production. Propylene glycol was beneficial in reducing the disease, improving milk yield and reproductive performance (McArt et al., 2011). However, depending on the way the propylene glycol is administered, as an oral drench or feed additives, the outcome might not be the same

because with the last option the spike of insulin does not happen (McArt et al., 2011; Nielsen &

Ingvartsen, 2004).

Besides the propylene glycol, another option to treat ketosis is the use of glucocorticoids.

These steroid hormones, such as dexamethasone, can be administered to simulate the increase of insulin and glucose blood levels, and they can therefore prolong the transient hyperglycemia

(Herdt & Emery, 1992; Jorritsma et al, 2004). However, these hormones can decrease the production of milk and do not impact disease incidence (Tatone et al., 2016). This method should be therefore used with caution: overdose can induce the appearance of symptoms, such as loss of appetite (Andrews et al., 2004; Gordon et al., 2013). Thus, considering these possible side effects and the lack of evidence of efficiency, this treatment is not the most indicated one for ketosis

(Tatone et al., 2016).

The last treatment method discussed here is the use of B vitamins. Blood and liver vitamin

B levels are reduced postpartum, so adding vitamins of this class as a dietary supplement can be used as a treatment for ketosis (Gordon et al., 2013; Gordon et al., 2017). The administration of class B vitamins in cows with ketosis has variable effects, such as increased milk and blood sugar production and decreased ketone bodies in the blood, but their efficacy has not yet been proven

(Gordon et al., 2013; Scott et al., 2011). One of the B vitamins used as a treatment for ketosis is niacin, or vitamin B3. This is an antilipolytic agent which induces the increase of blood glucose and insulin levels and can be administered together with propylene glycol for better efficacy. In addition, another is biotin, or vitamin B7, which is a coenzyme that is a component of gluconeogenesis. This decreases levels of triacylglycerol and hepatic NEFA, which may improve the percentage of fat in the milk of lactating cows, but it does not decrease BHB levels. Another option is cobalamin (vitamin B12) that generally has reduced levels in postpartum cows. This is

an essential cofactor in propionate metabolism, and is usually co-administered with cobalt (Gordon et al., 2017; Radostits et al., 2007; Scott et al., 2011). However, the use of B vitamins alone is thus not an effective treatment for ketosis.

The choice of the best treatment should be based on the symptoms and should aim to improve the quality of the animals’ life, such as the restoration of the normal appetite of the cows.

Genetic control of Ketosis

For years, the main focus of dairy cow breeding programs had been on increasing efficiency in milk production without considering ketosis incidence, which has made the animals more susceptible to metabolic imbalances in order to meet the demands often above physiological capacity (Veerkamp & Beerda, 2003). However, genetic evaluation for metabolic diseases, including ketosis, have been implemented in several countries, including in Canada (Beavers &

Van Doormaal, 2016).

Improving ketosis resistance in cows presents some problems such as the need for accurate diagnosis, records of phenotype and the low heritability of the disease (0.01 to 0.28) (Heringstad et al., 2005; Kadarmideen et al., 2000; Pryce et al., 2016; Uribe et al., 1995). Heritability estimates for clinical ketosis ranging from 0.02 to 0.16 and for subclinical ketosis ranging from 0.04 to 0.17 have been reported (Lee et al., 2016; Pryce et al., 2016). The subjective nature of this trait's diagnosis influences ketosis’ low heritability estimates, since BHB, an indicator trait for ketosis, has a relatively higher heritability of 0.14 to 0.28 (van der Drift et al., 2012). In addition, since many countries record disease on a voluntary basis, not all dairy producers maintain a consistency in the records and of those who do that, only a small percentage have satisfactory records for a correct evaluation (Koeck et al., 2012). Additionally, the information kept for ketosis diagnosis is

based on clinical ketosis, which could underrepresent the disease (Kelton et al., 1998; Parker

Gaddis et al., 2018).

Therefore, aiming to create a genetic evaluation system for disease traits that would be used across Canada, a National Health Project to help with herd management was created in 2007 as collaboration between the Canadian Dairy Network (CDN; Guelph, ON, Canada) and

Canadian DHI organizations (Koeck et al., 2012). This initiative led to the development and implementation of a metabolic disease resistance index in Canada, which was introduced in 2016 by CDN (Beavers & Van Doormaal, 2016).

Genome-wide Association Studies

Molecular markers are numerous and are distributed throughout the entire genome

(Griffiths et al., 2016). Since the 1980s, several types of molecular markers have been used in genomic studies, including single nucleotide polymorphisms (SNPs) (Xia et al., 2014). SNPs are characterized by the substitution, addition, or deletion of a nucleotide. However, for this variation to be considered a SNP, it has to occur in at least 1% of the population (Borem & Caixeta, 2006).

Molecular marker-assisted selection has been implemented and currently focuses on genomic selection (Meuwissen et al., 2001), which has advantages over traditional selection

(Ibanez-Escriche & Simianer, 2016). This is expanding in livestock breeding due to the possibility of genotyping thousands of SNPs at one time in equipment such as high-density genotyping chips.

Thus, SNP chips have brought great advances for genetic studies, enabling analysis such as genome-wide association study (GWAS) and making it possible to include genomic information generated in genetic improvement schemes (Caetano, 2009).

In the GWAS, the entire genome is scanned to identify genetic markers associated with

Quantitative trait loci (QTL) in linkage disequilibrium (LD) with causative mutations in genes that influence traits of economic interest (Lin & Musunuru, 2018). Moreover, the understanding of the biological and physiological mechanisms of genes involved with the expression of the phenotype of interest is aided by GWAS, which allows the identification of variations in DNA that are associated with a given trait through statistical analysis (Ku et al., 2010).

This type of genomic analysis was initially developed for human studies, but has been applied to different animal species (McCarthy et al., 2008; Santana et al., 2014). In dairy cattle,

GWAS have been used as an important approach for quantitative trait loci (QTL) mapping studies

(Cole et al., 2011). These association studies are useful tools for detecting and identifying genomic regions of interest, including those associated with disease resistance, which would ultimately help selecting animals less susceptible to diseases. However, discovering an associated SNP marker may only mean genetic linkage to the causal mutation, so not implying causality. Moreover, further investigation and fine mapping of the areas around associated SNPs are generally needed, mainly because GWAS are liable to type I (false positive) and type II errors (false negative) (Chan et al.,

2009; Goddard & Hayes, 2009; Ziegler et al., 2008).

GWA studies have used different strategies to overcome limitations, such as assuming that all SNP come from the same normal distribution. The weighted single-step GBLUP (wssGBLUP) was proposed by Wang et al. (2012) as an alternative method of the ssGBLUP. This method combines genotype, phenotype and pedigree information together in one step to estimate the effects of SNPs. This way, SNPs are simultaneously considered with the phenotypes of animals with or without genotype information available and weights are estimated for each SNP (Wang et al., 2012).

A limited number of GWAS results available in the literature indicate the existence of genomic regions in the bovine genome associated with ketosis (clinical and subclinical). A specific

SNP (rs109896020) on BTA 5 was found as significantly associated with ketosis in dairy cattle

(Klein et al., 2019). Nayeri et al. (2019) conducted a GWAS and identified a novel region on

BTA20 significantly associated with ketosis in Canadian dairy Holstein cattle. Kroezen et al.

(2018) identified six candidate genes related to ketosis metabolism and Parker Gaddis et al. (2018) found few genes involved in lipid metabolism and immune response to be associated with ketosis.

The relatively low heritability estimates for both subclinical and clinical of ketosis and the results from these previous studies showing only regions/genes with modest effect on subclinical and clinical ketosis indicate that these traits are polygenic in nature. Thus, the identification of further candidate genes for clinical and subclinical ketosis may contribute for a more efficient selection of animals less susceptible to these diseases.

Conclusion

Ketosis in dairy cattle is a significant problem for producers and cows, as it causes economic losses and decrease the cow’s quality of life. Ketosis is a multifactorial trait with low heritability, which shows the need for better understanding of the genetic bases of this disease. The search for polymorphic genes, whose alleles are associated with different levels of resistance to the development of this disease is of extreme importance to help select less susceptible animals.

In this sense, genome-wide association studies are a powerful strategy in the detection of genomic regions and genes involved with this metabolic disease.

References

Andersson, L. (1988). Subclinical ketosis in dairy cows. 4, 233–251.

Andrews, A. H., Blowey, R. W., Boyd, H., & Eddy, R. G. (2004). Bovine Medicine Diseases and

Husbandry of Cattle (Second). Blackwell Science.

Baird, G. D. (1982). Primary Ketosis in the High-Producing Dairy Cow: Clinical and Subclinical

Disorders, Treatment, Prevention, and Outlook. Journal of Dairy Science, 65(1), 1–10.

https://doi.org/10.3168/jds.S0022-0302(82)82146-2

Bauman, D. E., & Bruce Currie, W. (1980). Partitioning of Nutrients During Pregnancy and

Lactation: A Review of Mechanisms Involving Homeostasis and Homeorhesis. Journal

of Dairy Science, 63(9), 1514–1529. https://doi.org/10.3168/jds.S0022-0302(80)83111-0

Beavers, L. and Van Doormaal, B. (2016). Selection for Increased Resistance to Metabolic Diseases. Canadian Dairy Network (CDN). Berge, A. C., and G. Vertenten. 2014. A field study to determine the prevalence, dairy herd management systems, and fresh cow clinical conditions associated with ketosis in western European dairy herds. J.Dairy Sci. 97: 2145-2154. Bergman, E. N. (1971). Hyperketonemia-Ketogenesis and Ketone Body Metabolism. Journal of

Dairy Science, 54(6), 936–948. https://doi.org/10.3168/jds.S0022-0302(71)85950-7

Bertics, S. J., Grummer, R. R., Cadorniga-Valino, C., & Stoddard, E. E. (1992). Effect of

Prepartum Dry Matter Intake on Liver Triglyceride Concentration and Early Lactation.

75(7), 9.

Bertoni, G., Trevisi, E., Han, X., & Bionaz, M. (2008). Effects of Inflammatory Conditions on

Liver Activity in Puerperium Period and Consequences for Performance in Dairy Cows.

Journal of Dairy Science, 91(9), 3300–3310. https://doi.org/10.3168/jds.2008-0995

Bionaz, M., Trevisi, E., Calamari, L., Librandi, F., Ferrari, A., & Bertoni, G. (2007). Plasma

Paraoxonase, Health, Inflammatory Conditions, and Liver Function in Transition Dairy

Cows. Journal of Dairy Science, 90(4), 1740–1750. https://doi.org/10.3168/jds.2006-445

Block, E., & Co, D. (2010). Transition Cow Research – What Makes Sense Today? 24.

Borem, A., & Caixeta, E. T. (2006). Marcadores Moleculares (2nd ed.). UFV.

Caetano, A. R. (2009). Marcadores SNP: Conceitos básicos, aplicações no manejo e no

melhoramento animal e perspectivas para o futuro. Revista Brasileira de Zootecnia,

38(spe), 64–71. https://doi.org/10.1590/S1516-35982009001300008

Chan, E. K. F., Hawken, R., & Reverter, A. (2009). The combined effect of SNP-marker and

phenotype attributes in genome-wide association studies. Animal Genetics, 40(2), 149–

156. https://doi.org/10.1111/j.1365-2052.2008.01816.x

Cole, J. B., Wiggans, G. R., Ma, L., Sonstegard, T. S., Jr, T. J. L., & Crooker, B. A. (2011).

Genome-wide association analysis of thirty one production, health, reproduction and

body conformation traits in contemporary U.S. Holstein cows. BMC Genomics, 12, 408.

Divers, T. J., & Peek, S. F. (2008). Rebhun’s Diseases of Dairy Cattle (2nd ed.). MO:Elsevier.

Doepel, L., Lapierre, H., & Kennelly, J. J. (2002). Peripartum Performance and Metabolism of

Dairy Cows in Response to Prepartum Energy and Protein Intake. Journal of Dairy

Science, 85(9), 2315–2334. https://doi.org/10.3168/jds.S0022-0302(02)74312-9

Dohoo, I. R., & Martin, S. W. (1984). Subclinical ketosis: Prevalence and associations with

production and disease. Canadian Journal of Comparative Medicine: Revue Canadienne

De Medecine Comparee, 48(1), 1–5.

Drackley, J. K. (1999). Biology of Dairy Cows During the Transition Period: The Final Frontier?

Journal of Dairy Science, 82(11), 2259–2273. https://doi.org/10.3168/jds.S0022-

0302(99)75474-3

Duffield, T. F. (2000). Subclinical Ketosis in Lactating Dairy Cattle. Veterinary Clinics of North

America: Food Animal Practice, 16(2), 231–253. https://doi.org/10.1016/S0749-

0720(15)30103-1

Duffield, T. F., Kelton, D. F., Leslie, K. E., Lissemore, K. D., & Lumsden, J. H. (1997). Use of

test day milk fat and milk protein to detect subclinical ketosis in dairy cattle in Ontario.

The Canadian Veterinary Journal = La Revue Veterinaire Canadienne, 38(11), 713–718.

Duffield, T. F., Lissemore, K. D., McBride, B. W., & Leslie, K. E. (2009). Impact of

hyperketonemia in early lactation dairy cows on health and production. Journal of Dairy

Science, 92(2), 571–580. https://doi.org/10.3168/jds.2008-1507

Esposito, G., Irons, P. C., Webb, E. C., & Chapwanya, A. (2014). Interactions between negative

energy balance, metabolic diseases, uterine health and immune response in transition

dairy cows. Animal Reproduction Science, 144(3–4), 60–71.

https://doi.org/10.1016/j.anireprosci.2013.11.007

Fetrow, J., Nordlund, K. V., & Norman, H. D. (2006). Invited Review: Culling: Nomenclature,

Definitions, and Recommendations. Journal of Dairy Science, 89(6), 1896–1905.

https://doi.org/10.3168/jds.S0022-0302(06)72257-3

Foss-Freitas, M. C., & Foss, M. C. (2003). Cetoacidose diabética e estado hiperglicêmico

hiperosmolar. 36, 389–393.

Geishauser, T., Leslie, K., Tenhag, J., & Bashiri, A. (2000). Evaluation of Eight Cow-Side

Ketone Tests in Milk for Detection of Subclinical Ketosis in Dairy Cows. 83(2), 4.

Goddard, M. E., & Hayes, B. J. (2009). Mapping genes for complex traits in domestic animals

and their use in breeding programmes. Nature Reviews Genetics, 10(6), 381–391.

https://doi.org/10.1038/nrg2575

Goff, J. P., & Horst, R. L. (1997). Physiological Changes at Parturition and Their Relationship to

Metabolic Disorders. Journal of Dairy Science, 80(7), 1260–1268.

https://doi.org/10.3168/jds.S0022-0302(97)76055-7

González, F. H., & Silva, S. C. (2017). Introdução à Bioquímica Clínica Veterinária (Third).

UFRGS Editora.

Gordon, Jessica L., LeBlanc, S. J., & Duffield, T. F. (2013a). Ketosis Treatment in Lactating

Dairy Cattle. Veterinary Clinics of North America: Food Animal Practice, 29(2), 433–

445. https://doi.org/10.1016/j.cvfa.2013.03.001

Gordon, Jessica L., LeBlanc, S. J., & Duffield, T. F. (2013b). Ketosis Treatment in Lactating

Dairy Cattle. Veterinary Clinics of North America: Food Animal Practice, 29(2), 433–

445. https://doi.org/10.1016/j.cvfa.2013.03.001

Gordon, J.L., Duffield, T. F., Herdt, T. H., Kelton, D. F., Neuder, L., & LeBlanc, S. J. (2017).

Effects of a combination butaphosphan and cyanocobalamin product and insulin on

ketosis resolution and milk production. Journal of Dairy Science, 100(4), 2954–2966.

https://doi.org/10.3168/jds.2016-11925

Gordon, J.L., LeBlanc, S. J., Kelton, D. F., Herdt, T. H., Neuder, L., & Duffield, T. F. (2017).

Randomized clinical field trial on the effects of butaphosphan-cyanocobalamin and

propylene glycol on ketosis resolution and milk production. Journal of Dairy Science,

100(5), 3912–3921. https://doi.org/10.3168/jds.2016-11926

Griffiths, A. J., Miller, J. H., Suzuki, D. T., Lewontin, R. C., Gelbart, W. M., Griffiths, A. J.,

Miller, J. H., Suzuki, D. T., Lewontin, R. C., & Gelbart, W. M. (2016). An Introduction

to Genetic Analysis (11th ed.). W. H. Freeman.

Gröhn, Y. T., Eicker, S. W., Ducrocq, V., & Hertl, J. A. (1998). Effect of Diseases on the

Culling of Holstein Dairy Cows in New York State. Journal of Dairy Science, 81(4),

966–978. https://doi.org/10.3168/jds.S0022-0302(98)75657-7

Grummer, R. R. (1995a). Impact of changes in organic nutrient metabolism on feeding the

transition dairy cow. Journal of Animal Science, 73(9), 2820.

https://doi.org/10.2527/1995.7392820x

Grummer, R. R. (1995b). Impact of changes in organic nutrient metabolism on feeding the

transition dairy cow. Journal of Animal Science, 73(9), 2820–2833.

Hayes, B. J., Chamberlain, A. J., McPARTLAN, H., Macleod, I., Sethuraman, L., & Goddard,

M. E. (2007). Accuracy of marker-assisted selection with single markers and marker

haplotypes in cattle. Genetical Research, 89(4), 215–220.

https://doi.org/10.1017/S0016672307008865

Herdt, T. H., & Emery, R. S. (1992). Therapy of diseases of ruminant intermediary metabolism.

The Veterinary Clinics of North America. Food Animal Practice, 8(1), 91–106.

Herdt, Thomas H. (2000). Ruminant Adaptation to Negative Energy Balance. Veterinary Clinics

of North America: Food Animal Practice, 16(2), 215–230. https://doi.org/10.1016/S0749-

0720(15)30102-X

Heringstad, B., Chang, Y. M., Gianola, D., & Klemetsdal, G. (2005). Genetic Analysis of

Clinical Mastitis, Milk Fever, Ketosis, and Retained Placenta in Three Lactations of

Norwegian Red Cows. Journal of Dairy Science, 88(9), 3273–3281.

https://doi.org/10.3168/jds.S0022-0302(05)73010-1

Ibanez-Escriche, N., & Simianer, H. (2016). Animal breeding in the genomics era. 6, 4.

Itle, A. J., Huzzey, J. M., Weary, D. M., & von Keyserlingk, M. A. G. (2015). Clinical ketosis

and standing behavior in transition cows. Journal of Dairy Science, 98(1), 128–134.

https://doi.org/10.3168/jds.2014-7932

Iwersen, M., Klein-Jöbstl, D., Pichler, M., Roland, L., Fidlschuster, B., Schwendenwein, I., &

Drillich, M. (2013). Comparison of 2 electronic cowside tests to detect subclinical ketosis

in dairy cows and the influence of the temperature and type of blood sample on the test

results. Journal of Dairy Science, 96(12), 7719–7730. https://doi.org/10.3168/jds.2013-

7121

Jorritsma, R., Thanasak, J., Houweling, M., Noordhuizen, J. P. T. M., & Muller, K. E. (2004).

Effects of a single dose of dexamethasone-21—Isonicotinate on the metabolism of

heifers in early lactation. Veterinary Record, 155(17), 521–523.

https://doi.org/10.1136/vr.155.17.521

Kadarmideen, H. N., Thompson, R., & Simm, G. (2000). Linear and threshold model genetic

parameters for disease, fertility and milk production in dairy cattle. Animal Science,

71(03), 411–419. https://doi.org/10.1017/S1357729800055338

Kelton, D. F., Lissemore, K. D., & Martin, R. E. (1998). Recommendations for recording and

calculating the incidence of selected clinical diseases of dairy cattle. Journal of Dairy

Science, 81(9), 2502–2509. https://doi.org/10.3168/jds.S0022-0302(98)70142-0

Klein, S.-L., Scheper, C., Brügemann, K., Swalve, H. H., & König, S. (2019). Phenotypic

relationships, genetic parameters, genome-wide associations, and identification of

potential candidate genes for ketosis and fat-to-protein ratio in German Holstein cows.

Journal of Dairy Science, 102(7), 6276–6287. https://doi.org/10.3168/jds.2019-16237

Koeck, A., Jamrozik, J., Schenkel, F. S., Moore, R. K., Lefebvre, D. M., Kelton, D. F., &

Miglior, F. (2014). Genetic analysis of milk β-hydroxybutyrate and its association with

fat-to-protein ratio, body condition score, clinical ketosis, and displaced abomasum in

early first lactation of Canadian Holsteins. Journal of Dairy Science, 97(11), 7286–7292.

https://doi.org/10.3168/jds.2014-8405

Koeck, A., Miglior, F., Kelton, D. F., & Schenkel, F. S. (2012). Health recording in Canadian

Holsteins: Data and genetic parameters. Journal of Dairy Science, 95(7), 4099–4108.

https://doi.org/10.3168/jds.2011-5127

Kroezen, V., Schenkel, F. S., Miglior, F., Baes, C. F., & Squires, E. J. (2018). Candidate gene

association analyses for ketosis resistance in Holsteins. Journal of Dairy Science, 101(6),

5240–5249. https://doi.org/10.3168/jds.2017-13374

Kronfeld, D. S., Mayer, G. P., Robertson, J. M., & Raggi, F. (1963). Depression of Milk

Secretion during Insulin Administration1. 46(6), 559–563.

Ku, C. S., Loy, E. Y., Pawitan, Y., & Chia, K. S. (2010). The pursuit of genome-wide association

studies: Where are we now? Journal of Human Genetics, 55, 195–206.

Leek, B. F., & Reece, W. O. (2014). Digestão no Estômago do Ruminante. In Dukes, fisiologia

dos animais domésticos (12th ed., pp. 404–437). Guanabara Koogan.

Lin, J., & Musunuru, K. (2018). From Genotype to Phenotype: A Primer on the Functional

Follow-Up of Genome-Wide Association Studies in Cardiovascular Disease. Circulation:

Genomic and Precision Medicine, 11(2). https://doi.org/10.1161/CIRCGEN.117.001946

Littledike, E. T., Young, J. W., & Beitz, D. C. (1981). Common Metabolic Diseases of Cattle:

Ketosis, Milk Fever, Grass Tetany, and Downer Cow Complex. Journal of Dairy

Science, 64(6), 1465–1482. https://doi.org/10.3168/jds.S0022-0302(81)82715-4

Mair, B., Drillich, M., Klein-Jöbstl, D., Kanz, P., Borchardt, S., Meyer, L., Schwendenwein, I.,

& Iwersen, M. (2016). Glucose concentration in capillary blood of dairy cows obtained

by a minimally invasive lancet technique and determined with three different hand-held

devices. BMC Veterinary Research, 12. https://doi.org/10.1186/s12917-016-0662-3

McArt, J. A. A., Nydam, D. V., & Oetzel, G. R. (2012). Epidemiology of subclinical ketosis in

early lactation dairy cattle. Journal of Dairy Science, 95(9), 5056–5066.

https://doi.org/10.3168/jds.2012-5443

McArt, J. A. A., Nydam, D. V., Ospina, P. A., & Oetzel, G. R. (2011). A field trial on the effect

of propylene glycol on milk yield and resolution of ketosis in fresh cows diagnosed with

subclinical ketosis. Journal of Dairy Science, 94(12), 6011–6020.

https://doi.org/10.3168/jds.2011-4463

McArt, J. A. A., Nydam, D. V., & Overton, M. W. (2015). Hyperketonemia in early lactation

dairy cattle: A deterministic estimate of component and total cost per case. Journal of

Dairy Science, 98(3), 2043–2054. https://doi.org/10.3168/jds.2014-8740

McCarthy, M. I., Abecasis, G. R., Cardon, L. R., Goldstein, D. B., Little, J., Ioannidis, J. P. A., &

Hirschhorn, J. N. (2008). Genome-wide association studies for complex traits:

Consensus, uncertainty and challenges. Nature Reviews Genetics, 14.

McLaren, C. J., Lissemore, K. D., Duffield, T. F., Leslie, K. E., Kelton, D. F., & Grexton, B.

(2006). The relationship between herd level disease incidence and a return over feed

index in Ontario dairy herds. 47, 7.

Miyoshi, S., Pate, J. L., & Palmquist, D. L. (2001). Effects of propylene glycol drenching on

energy balance, plasma glucose, plasma insulin, ovarian function and conception in dairy

cows. Animal Reproduction Science, 68(1–2), 29–43.

Nayeri, S., Schenkel, F., Fleming, A., Kroezen, V., Sargolzaei, M., Baes, C., Cánovas, A.,

Squires, J., & Miglior, F. (2019). Genome-wide association analysis for β-

hydroxybutyrate concentration in Milk in Holstein dairy cattle. BMC Genetics, 20(1), 58.

https://doi.org/10.1186/s12863-019-0761-9

Nielsen, N. I., & Ingvartsen, K. L. (2004). Propylene glycol for dairy cows. Animal Feed Science

and Technology, 115(3–4), 191–213. https://doi.org/10.1016/j.anifeedsci.2004.03.008

Oetzel, G. R. (2007). Herd-Level Ketosis – Diagnosis and Risk Factors. Preconference Seminar

7C: Dairy Herd Problem Investigation Strategies: Transition Cow Troubleshooting, 67–

91.

Parker Gaddis, K. L., Megonigal, J. H., Clay, J. S., & Wolfe, C. W. (2018). Genome-wide

association study for ketosis in US Jerseys using producer-recorded data. Journal of

Dairy Science, 101(1), 413–424. https://doi.org/10.3168/jds.2017-13383

Pryce, J. E., Parker Gaddis, K. L., Koeck, A., Bastin, C., Abdelsayed, M., Gengler, N., Miglior,

F., Heringstad, B., Egger-Danner, C., Stock, K. F., Bradley, A. J., & Cole, J. B. (2016).

Invited review: Opportunities for genetic improvement of metabolic diseases. Journal of

Dairy Science, 99(9), 6855–6873. https://doi.org/10.3168/jds.2016-10854

Radostits, O. M., Gay, C. C., Hinchcliff, K. W., & Constable, P. D. (2007). Veterinary Medicine:

A text book of the disease of cattle, horses, sheep, pigs and goats (10th ed.). Elsevier.

Riet-Correa, F., Riet-Correa, G., & Schild, A. L. (2002). Importância do exame clínico para o

diagnóstico das enfermidades do sistema nervoso em ruminantes e eqüídeos. Pesquisa

Veterinária Brasileira, 22(4), 161–168. https://doi.org/10.1590/S0100-

736X2002000400006

Rollin, E., Berghaus, R. D., Rapnicki, P., Godden, S. M., & Overton, M. W. (2010). The effect

of injectable butaphosphan and cyanocobalamin on postpartum serum β-hydroxybutyrate,

calcium, and phosphorus concentrations in dairy cattle. Journal of Dairy Science, 93(3),

978–987. https://doi.org/10.3168/jds.2009-2508

Sakai, T., Hamakawa, M., & Kubo, S. (1996). Glucose and xylitol tolerance tests for ketotic and

healthy dairy cows. Journal of Dairy Science, 79(3), 372–377.

https://doi.org/10.3168/jds.S0022-0302(96)76374-9

Sakha, M., Ameri, M., Sharifi, H., & Taheri, I. (2007). Bovine Subclinical Ketosis in Dairy

Herds in Iran. Veterinary Research Communications, 31(6), 673–679.

https://doi.org/10.1007/s11259-007-0053-0

Santana, M. H. A., Utsunomiya, Y. T., Neves, H. H. R., Gomes, R. C., Garcia, J. F., Fukumasu,

H., Silva, S. L., Leme, P. R., Coutinho, L. L., Eler, J. P., & Ferraz, J. B. S. (2014).

Genome-wide association study for feedlot average daily gain in Nellore cattle (Bos

indicus). J. Anim. Breed. Genet., 7.

Scott, P. R., Penny, C. D., & Macrae, A. I. (2011). Scott, P. R., Penny, C. D., Macrae, A. I.

(2011) “Metabolic Diseases” in Scott, P. R. et al., Cattle Medicine, Manson Publishing,

247-257. In Cattle Medicine (pp. 247–257). Manson Publishing.

Shaw, J. C. (1956). Ketosis in Dairy Cattle. A Review. Journal of Dairy Science, 39(4), 402–

434. https://doi.org/10.3168/jds.S0022-0302(56)94765-8

Sordillo, L. M., & Mavangira, V. (2014). The nexus between nutrient metabolism, oxidative

stress and inflammation in transition cows. Animal Production Science, 54(9), 1204.

https://doi.org/10.1071/AN14503

Studer, V. A., Grummer, R. R., Bertics, S. J., & Reynolds, C. K. (1993). Effect of Prepartum

Propylene Glycol Administration on Periparturient Fatty Liver in Dairy Cows. Journal of

Dairy Science, 76(10), 2931–2939. https://doi.org/10.3168/jds.S0022-0302(93)77633-X

Sudhakara Reddy, B., Reddy, B. S. S., Reddy, Y. V. P., & Vennkatasivakumar, R. (2014a).

Nervous form of ketosis in cows and its treatment. International Journal of Biological

Research, 2(2), 143. https://doi.org/10.14419/ijbr.v2i2.3591

Sudhakara Reddy, B., Reddy, B. S. S., Reddy, Y. V. P., & Vennkatasivakumar, R. (2014b).

Nervous form of ketosis in cows and its treatment. International Journal of Biological

Research, 2(2), 143. https://doi.org/10.14419/ijbr.v2i2.3591

Sun, L. W., Zhang, H. Y., Wu, L., Shu, S., Xia, C., Xu, C., & Zheng, S. (2014). 1H-Nuclear

magnetic resonance-based plasma metabolic profiling of dairy cows with clinical and

subclinical ketosis. Journal of Dairy Science, 97, 1552–1562. http://dx.doi.org/

10.3168/jds.2013-6757

Suthar, V. S., J. Canelas-Raposo, A. Deniz, and W. Heuwieser. 2013. Prevalence of subclinical ketosis and relationships with postpartum disease in European dairy cows. J. Dairy Sci. 96: 2925–2938. Tatone, E. H., Duffield, T. F., Capel, M. B., DeVries, T. J., LeBlanc, S. J., & Gordon, J. L.

(2016). A randomized controlled trial of dexamethasone as an adjunctive therapy to

propylene glycol for treatment of hyperketonemia in postpartum dairy cattle. Journal of

Dairy Science, 99(11), 8991–9000. https://doi.org/10.3168/jds.2016-11358

Tatone, E. H., Duffield, T. F., LeBlanc, S. J., DeVries, T. J., & Gordon, J. L. (2017).

Investigating the within-herd prevalence and risk factors for ketosis in dairy cattle in

Ontario as diagnosed by the test-day concentration of β-hydroxybutyrate in milk. Journal

of Dairy Science, 100(2), 1308–1318. https://doi.org/10.3168/jds.2016-11453

Uribe, H. A., Kennedy, B. W., Martin, S. W., & Kelton, D. F. (1995). Genetic Parameters for

Common Health Disorders of Holstein Cows. Journal of Dairy Science, 78(2), 421–430.

https://doi.org/10.3168/jds.S0022-0302(95)76651-6

Van Dorland, H. A., Richter, S., Morel, I., Doherr, M. G., Castro, N., & Bruckmaier, R. M.

(2009). Variation in hepatic regulation of metabolism during the dry period and in early

lactation in dairy cows. Journal of Dairy Science, 92(5), 1924–1940.

https://doi.org/10.3168/jds.2008-1454

Vanholder, T., Papen, J., Bemers, R., Vertenten, G., & Berge, A. C. B. (2015). Risk factors for

subclinical and clinical ketosis and association with production parameters in dairy cows

in the Netherlands. Journal of Dairy Science, 98(2), 880–888.

https://doi.org/10.3168/jds.2014-8362

Veerkamp, R. F., & Beerda, B. (2003). Effects of genetic selection for milk yield on energy

balance, levels of hormones, and metabolites in lactating cattle, and possible links to

reduced fertilityq. Livestock Production Science, 19.

Vicente, F., Rodríguez, M. L., Martínez-Fernández, A., Soldado, A., Argamentería, A., Peláez,

M., & de la Roza-Delgado, B. (2014). Subclinical Ketosis on Dairy Cows in Transition

Period in Farms with Contrasting Butyric Acid Contents in Silages. The Scientific World

Journal, 2014, 1–4. https://doi.org/10.1155/2014/279614

Wagner, S. A., & Schimek, D. E. (2010). Evaluation of the effect of bolus administration of 50%

dextrose solution on measures of electrolyte and energy balance in postpartum dairy

cows. American Journal of Veterinary Research, 71(9), 1074–1080.

https://doi.org/10.2460/ajvr.71.9.1074

Wang, H., Misztal, I., Aguilar, I., Legarra, A., & Muir, W. M. (2012). Genome-wide association

mapping including phenotypes from relatives without genotypes. Genetics Research,

94(2), 73–83. https://doi.org/10.1017/S0016672312000274

Wankhade, P. R., Manimaran, A., Kumaresan, A., Jeyakumar, S., Ramesha, K. P., Sejian, V.,

Rajendran, D., & Varghese, M. R. (2017). Metabolic and immunological changes in

transition dairy cows: A review. Veterinary World, 10(11), 1367–1377.

https://doi.org/10.14202/vetworld.2017.1367-1377

Xia, J. H., Wan, Z. Y., Ng, Z. L., Wang, L., Fu, G. H., Lin, G., Liu, F., & Yue, G. H. (2014).

Genome-wide discovery and in silico mapping of gene-associated SNPs in Nile tilapia.

Aquaculture, 432, 67–73. https://doi.org/10.1016/j.aquaculture.2014.04.028

Ziegler, A., König, I. R., & Thompson, J. R. (2008). Biostatistical Aspects of Genome-Wide

Association Studies. Biometrical Journal, 50(1), 8–28.

https://doi.org/10.1002/bimj.200710398

CHAPTER TWO: Differential gene expression in dairy cows under negative energy balance and ketosis: A systematic review and meta-analysis

R.A.N. Soares1, G. Vargas1, M.M.M. Muniz1, M.A.M. Soares2, A. Cánovas1; F. Schenkel1#, and

E. J. Squires1#

(Accepted for publication at the Journal of Dairy Science)

1 University of Guelph, Department of Animal Biosciences, Center for Genetic Improvement of

Livestock, University of Guelph, Guelph, Canada, N1G 2W1

2 Federal Rural University of Rio de Janeiro, Seropédica-RJ, Brazil, 23897-000

# These authors had equal contribution as senior authors

* Corresponding author: Riani Ananda Nunes Soares

50 Stone Rd E Guelph, ON N1G 2W1, c/o Department of Animal Biosciences

519-993-8724 [email protected]

Abstract

Development of ketosis in high producing dairy cows contributes to several animal health issues and highlights the need for a better understanding of the genetic basis of metabolic diseases.

To evaluate the pattern of differential gene expression in the liver of cows under negative energy balance (NEB), and under subclinical and clinical ketosis, a meta-analysis of gene expression and genome-wide association studies (GWAS) results was performed. An initial systematic review identified 118 articles based on the keywords cow, liver, negative energy balance, ketosis, expression, qPCR, microarray, proteomic, RNA-Seq, and GWAS. After further screening for only peer-reviewed and pertinent articles for gene expression during NEB, clinical and subclinical

ketosis (considering plasma levels of β-hydroxybutyrate, BHB), 20 articles were included in the analysis. From the systematic review, 430 significant SNPs identified by GWAS were assigned to genes reported in gene expression studies by considering chromosome and base pair positions in the ARS-UCD 1.2 bovine assembly. Venn diagrams were created to integrate the data obtained in the systematic review, and Gene Ontology enrichment analysis was carried out using official gene names. A QTL enrichment analysis was also performed to identify potential positional candidate loci. Twenty-four significant SNPs were located within the coordinates of differentially expressed genes located on chromosomes 2, 3, 6, 9, 11, 14, 27, and 29. Three significant metabolic pathways were associated with NEB, subclinical and clinical ketosis. In addition, two important genes,

PPARA (Peroxisome Proliferator Activated Receptor Alpha) and ACACA (Acetyl-CoA

Carboxylase alpha), were identified, which were differentially expressed in the three metabolic conditions. The PPARA gene is involved in the regulation of lipid metabolism and fatty liver disease, and the ACACA gene gene encodes an enzyme that catalyzes the carboxylation of acetyl-

CoA to malonyl-CoA, which is a rate-limiting step in fatty acid synthesis. Gene network analysis revealed co-expression interactions among 34 genes associated with functions involving fatty acid transport and fatty acid metabolism. For the annotated QTL, nine QTL were identified for ketosis.

The genes FN1 (fibronectin 1) and PTK2 (protein tyrosine kinase 2), which are mainly involved in cell adhesion and formation of extracellular matrix constituents, were enriched for QTL previously associated with the trait “ketosis” on chromosome 2 and for the trait “milk iron content” on chromosome 14, respectively. This integration of gene expression and GWAS data provides an additional understanding of the genetic background of NEB and subclinical and clinical ketosis in dairy cattle. Thus, it is a useful approach to identify biological mechanisms underlying these metabolic conditions in dairy cattle.

Keywords: β-hydroxybutyrate, functional enrichment analysis, gene expression, GWAS, meta- analysis

Introduction

During the initial lactation period, dairy cows experience a period of negative energy balance (NEB) caused by insufficient dietary intake to meet the animal’s requirements for maintenance and milk production (Leek & Reece, 2014; Seymour et al., 2019). As a result, cows must mobilize adipose reserves to meet these requirements and maintain a high level of milk production; consequently, they lose about 60% or more of their body fat in the first weeks after calving (Kadokawa & Martin, 2006; Renate & Cernescu, 2009). Dairy cows that are unable to adapt to NEB have an increased risk of developing metabolic diseases, such as subclinical or clinical ketosis and fatty liver. Ketosis is a major metabolic disease of dairy cows in early lactation and is defined as the presence of increased concentrations of ketone bodies such as acetoacetate,

β-hydroxybutyrate (BHB) and acetone, and a concurrent low level of glucose in the blood

(Andersson, 1988).

These metabolic diseases are a common problem in dairy farms and may lead to significant economic losses. At least 50% of dairy cows undergo a temporary period of subclinical ketosis in the first month of lactation (Esposito et al., 2014; Biswal et al., 2016). McArt et al. (2012) reported an average incidence of 43% of ketosis in North America during the first two weeks of lactation.

On average, each case of subclinical and clinical ketosis can result in losses of $203 and $289, respectively, because of decreased milk production, extended interval from calving to conception,

and increased risk of displaced abomasum (Geishauser et al., 1998; Geishauser et al., 2001; McArt et al., 2015; Gohary et al., 2016).

Despite NEB being essentially universal among high yielding dairy cows during the peripartum period, some animals are naturally capable of recovering and do not develop subclinical or clinical ketosis. The NEB condition causes breakdown of triglycerides within the adipose tissue, resulting in the release of non-esterified fatty acids (NEFA). The liver plays a crucial role in removing a large portion of NEFA by metabolism to ketone bodies or re- esterification to produce triglycerides. In addition, skeletal muscle has an essential role in adaptation to NEB by using fat-derived fuels, such as NEFA and ketone bodies (Emery et al.,

1992; Reynolds et al., 2013).

Previous genetic studies reported heritability estimates for clinical ketosis ranging from

0.02 to 0.16 and for subclinical ketosis ranging from 0.04 to 0.17 (Lee et al., 2016; Pryce et al.,

2016). The low heritability estimates for this trait may be due to the subjective nature of its diagnosis, since BHB, an indicator trait for ketosis, has a relatively higher heritability of 0.14 to

0.28 (van der Drift et al., 2012). To better understand this metabolic problem, several studies have investigated the biological processes, metabolic pathways, candidate genes and gene networks that influence NEB, subclinical and clinical ketosis in dairy cows by performing gene expression and genome-wide association studies (GWAS) (McCabe et al., 2012; Ha et al., 2015; Tetens et al.,

2015; Huang et al., 2019; Nayeri et al., 2019). The relationship among patterns of gene expression that occur as the clinical condition progresses from NEB to ketosis could be better understood by connecting information of the main metabolic pathways associated with this process. Thus, the objective of this study was to carry out a systematic review and meta-analysis from previously

published studies to identify consistent biological information across gene expression and GWAS results for NEB, subclinical and clinical ketosis in dairy cattle.

Material and Methods

Search of the Published Literature

A systematic review of NEB, subclinical and clinical ketosis in dairy cattle was carried out as described by Fonseca et al. (2018) to identify significant SNPs located within genomic regions overlapping with candidate genes that are differentially expressed under these metabolic conditions. Different transcriptomic techniques for gene expression studies were considered, including microarray, quantitative polymerase chain reaction (qPCR), and RNA-Sequencing

(RNA-Seq). In addition, differentially expressed gene products detected by proteomic studies were also considered. The keywords “cow”, “liver”, “negative energy balance”, “ketosis”, “expression”,

“qPCR”, “microarray”, “proteomic”, “RNA-Seq”, and “GWAS” were entered into three different databases: google scholar (https://scholar.google.com.br/), PubMed

(https://www.ncbi.nlm.nih.gov/pubmed/) and JANE (http://jane.biosemantics.org/) to perform this search.

The databases contained references in all languages, but our search was limited to complete references written in English. The first screening was performed on titles and abstracts of all papers to assure their relevance, and if the information was not available or was judged irrelevant to the research question of gene expression in NEB, subclinical and clinical ketosis, the reference was excluded. Only peer-reviewed and pertinent articles were obtained in full and assessed. Some additional criteria were applied to this step, such as only references on dairy cows were selected.

Also, gene expression data from three transcriptomic techniques (microarray, qPCR, or RNA-Seq)

and proteomic studies were kept. The metabolic condition of the cow had to be clearly stated, and only information from cows that did not receive any treatment to reverse NEB was kept. For articles with relevant information, but which did not clearly express the animal's clinical condition, plasma BHB levels were adopted as a criterion. Blood levels less than or equal to 0.6 mM were classified as control animals, concentrations higher than or equal to 1.2 mM were classified as subclinical ketosis and for values greater than or equal to 3.0 mM, as clinical ketosis (Zhu et al.

2019).

Metabolic Pathway Analysis

The pathways associated with NEB, subclinical, and clinical ketosis were investigated through a permutation analysis approach, proposed by Cabrera et al. (2012). For this, 430 significant SNPs obtained from four GWAS (Supplemental Table S1) were aligned against the locations of genes identified in gene expression studies using the ARS-UCD 1.2 bovine reference genome assembly. Thus, only SNPs located within differentially expressed genes were retained.

The SNPs were then ordered according to chromosomal location. According to Cabrera et al.

(2012), this approach can establish the significance level of pathways, considering the linkage disequilibrium (LD) and the clustering of functionally related genes over the genome.

For each permutation, gene P-values were calculated using Fisher’s combination test, followed by the hypergeometric test to identify significant pathways (Cabrera et al., 2012). This process was run over 10,000 times to obtain the test statistic distribution under the null hypothesis that the genes in a pathway are not more associated with the phenotype than are the non-pathway genes. A comparison-wise threshold of 0.0002 (0.05/225) was used based on Bonferroni correction and considering a database of 225 pathways obtained from the Kyoto Encyclopedia of Genes and

Genome database (KEGG.db, Kanehisa et al., 2012) from Bioconductor (Carlson, 2016). The analyses were performed using the genomicper package (Cabrera et al., 2016) in R (R

Development Core Team, 2019).

Venn Diagrams and Enrichment Analysis

Venn diagrams were created using the official gene names obtained from the systematic review to integrate the available information. The Venn diagram tool available at

“http://bioinformatics.psb.ugent.be/software” was used to calculate the intersections between gene lists detected from the articles analyzed here (Cánovas et al., 2014). Among them, information from different transcriptomic and proteomic studies to identify genes that were differentially expressed in the liver of cows under NEB, subclinical and clinical ketosis conditions was included.

The GeneMANIA software (www.genemania.org) implemented as a plug-in on the

Cytoscape platform (Warde-Farley, 2010) was used to investigate possible interactions among the genes detected in the permutation analysis carried out in the pathway analysis. This analysis searches for related genes on publicly available biological datasets and classifies the links in the network based on their relationship, such as co-expression, physical interaction, genetic interaction, shared protein domains, co-localization, and pathway. Furthermore, this analysis uses a pie graphic to show functions associated with genes in the network and their FDR and coverage

(as number of genes annotated with that function in the network versus number of genes annotated with that function in the genome). This gene network analysis could provide a better understanding of the genetic architecture of complex polygenic traits.

QTL Enrichment Analysis

From the above analyses, a list of SNPs associated with differentially expressed genes were used for gene and quantitative trait loci (QTL) annotation using R (Version 1.1.463.; R Core Team,

2013) and the GALLO R package for genomic functional annotation of positional candidate loci in livestock (https://github.com/pablobio/GALLO). The .gtf format file corresponding to the bovine gene annotation and the .gff format file with the QTL information from Animal QTL

Database (Hu et al., 2013) were used for gene and QTL annotation using the ARS-UCD1.2 bovine genome assembly coordinates, respectively. In addition, to identify the potential effects of polymorphisms in cattle traits, QTL enrichment analysis was performed. Single nucleotide polymorphisms associated to any of the six trait classes, e.g. ‘Reproduction’, ‘Milk’, ‘Production’,

‘Exterior’, ‘Meat and Carcass’, and ‘Health’, were considered for QTL enrichment analysis.

Results

A total of 118 articles were initially identified from the systematic review after applying the keywords into the databases. After screening titles and abstracts and removing duplicate articles, 76 unique articles were available. Following this, 33 articles were excluded as they did not meet the criterion of gene expression for NEB, subclinical and clinical ketosis. After this screening, 43 articles were kept for the next step. Further review was performed, and only articles including expression data from microarray, qPCR, RNA-Seq studies, and proteomic studies with clearly stated metabolic conditions were kept, leaving 20 articles considered appropriate for analysis (Supplemental Table S2).

From the 20 articles kept for further analyses, 3,216 differentially expressed genes were identified when considering the three metabolic conditions (NEB, subclinical and clinical ketosis),

and the transcriptomic and proteomic techniques. A total of 156 duplicate genes were removed from the gene list, and the 3,060 remaining genes were used to perform the subsequent analyses.

Then, they were ordered by techniques, irrespective of the metabolic condition, resulting in the following number of genes per technique: a) transcriptomic techniques: microarray (N= 1,812); qPCR (N= 89); RNA-Seq (N= 1,035), and b) proteomics (N=124 genes).

From the 430 significant SNPs obtained from published GWAS, 24 of them were located within 14 differentially expressed genes (Table 1). This set of 24 SNPs are located on chromosomes 2, 3, 6, 9, 11, 14, 27, and 29 and were selected from three different studies, which investigated milk levels of prognostic ketosis biomarkers (Tetens et al., 2015), candidate genes for ketosis resistance (Kroezen et al., 2018), and β-hydroxybutyrate concentration in milk (Nayeri et al., 2019) in dairy cattle.

From the permutation analysis, three significant metabolic pathways associated with 3

SNPs (rs41588443, rs378674931 and rs42197374) within 3 genes out of the 14 differentially expressed genes were identified for NEB, subclinical and clinical ketosis, namely “Glyoxylate and dicarboxylate metabolism” (bta00630), “Metabolic pathways” (bta01100), and “Peroxisome”

(bta04146) (Table 2). The hydroxy acid oxidase 2, long chain (HAO2), the acyl-CoA synthetase long chain family member 1 (ACSL1), and pyruvate carboxylase (PC) genes were involved in these pathways. The HAO2 gene was involved in all the significant pathways.

Venn diagrams showed that ten common genes detected by published transcriptomic analysis were differentially expressed in the liver of dairy cows under NEB, subclinical and clinical ketosis (Figure 1A). This includes the ACSL1 (acyl-CoA synthetase long chain family member 1) gene, ACACA (acetyl-CoA carboxylase alpha), FABP1 (fatty acid binding protein 1), PPARA

(peroxisome proliferator activated receptor alpha), CPT2 (carnitine palmityl transferase 2), and the

CPT1A (carnitine palmityl transferase 1A), DGAT1 (diacylglycerol O-acyltransferase 1), BAX

(BCL2 associated X, apoptosis regulator), TP53 (tumor protein P53), and the HMGCS1 (3- hydroxy-3-methylglutaryl-CoA synthase 1) genes. The highest number of common differentially expressed genes were observed between animals in NEB and clinical ketosis (N=133), whereas only 6 common genes were observed between animals under NEB and subclinical ketosis. These genes were p21 (cyclin-dependent kinase inhibitor 1), ACADL (acyl-CoA dehydrogenase long chain), MDM2 (MDM2 proto-oncogene), CASP3 and CASP9 (caspase 3 and 9), and BCL2 (BCL2 apoptosis regulator) (Supplemental Table S3). Figure 1B shows the number of expressed genes in the liver of dairy cows detected by proteomic studies. The GSTA1 (glutathione S-transferase alpha

1) gene was the only one common gene among cows under NEB, subclinical, and clinical ketosis conditions.

Figure 2A displays Venn diagrams showing the distribution of unique and common differentially expressed genes in dairy cows under NEB, subclinical and clinical ketosis detected by different transcriptomic techniques (microarray, qPCR, and RNA-Seq). There were 14 common genes among the three techniques. Some of these genes, named HR (HR lysine demethylase and receptor corepressor), NR4A1 (nuclear receptor subfamily 4 group A member 1), and PPARA

(peroxisome proliferator activated receptor alpha) are involved in DNA-binding transcription factor activity and transcription corepressor activity. The CPT1A and CPT1B genes are involved in the transport of long-chain fatty acids into the mitochondria for utilization as an energy source.

The CD36 (CD36 molecule) and FADS2 (fatty acid desaturase 2) genes have roles in lipid metabolism, ion binding, and oxidoreductase activities. The APOA1 (apolipoprotein A1),

ACADVL (acyl-CoA dehydrogenase very long chain), and SLC27A2 (solute carrier family 27

Member A2) genes are involved in long-chain fatty acid-CoA ligase activity and lipid binding. In

addition, the LPIN3 (lipin 3), ANGPTL4 (angiopoietin-like 4), PC (pyruvate carboxylase), and

GPX3 (glutathione peroxidase 3) genes are related to transcription coactivator, enzyme inhibitor, and ligase and biotin carboxylase activities.

The highest number of commonly expressed genes (N=125) was obtained between microarray and RNA-Seq techniques. For qPCR and microarray techniques, 28 common genes were found (Supplemental Table S3).

Figure 2B shows the number of unique and common differentially expressed genes identified in transcriptomic studies and proteomic studies. A total of 47 genes were detected in both transcriptomic and proteomic studies. Transcriptomic techniques identified a higher number of unique genes (N=2,693) than proteomics techniques (N=77).

When analyzing which genes were identified as differentially expressed during clinical ketosis by different techniques, a much larger number of genes were detected using microarrays than using qPCR and RNA-Seq approaches (Figure 3A). However, 63.6% of the differentially expressed genes identified by qPCR were also detected in studies that used a microarray technique.

The RNA-Seq method had the smallest sample set (N =26), but 42.3% of the detected genes were also identified as differentially expressed in studies that used microarray or qPCR approaches. No common genes were detected by all three techniques.

Figure 3B shows the number of differentially expressed genes in cows under the NEB condition detected different techniques. The PPARA gene was the only gene detected by all three techniques, i.e., microarray, qPCR, and RNA-Seq. This gene acts in the Peroxisome pathway, which is related to ketosis, and could be a key gene in ketosis development. Additionally, FABP1

(fatty acid binding protein 1) and CPT1A (carnitine palmitoyltransferase 1A) are important genes in the regulation of lipid metabolism that were detected using microarray and qPCR techniques.

The genes ACACA and ACSL1 were also identified by more than one technique (qPCR and RNA-

Seq).

Functional Enrichment Analysis

Figure 4 presents the gene network for the 14 differentially expressed genes with significant SNPs for NEB, subclinical and clinical ketosis (Table 1). The functional enrichment analysis identified 20 additional genes linked via various interaction types (co-expression, metabolic pathway, shared protein domains, co-localization, predicted physical interactions and/or genetic interactions) to the 14 differentially expressed genes. However, only co-expression links were highlighted on the network in Figure 4 to provide a clearer view of the genes that showed correlated expression levels for NEB, subclinical and clinical ketosis in dairy cattle.

The 20 additional linked genes did not harbor any significant SNPs from the published

GWAS. The six genes (G3BP2, CPT1C, CPT1B, FABP6, CHRNB1, and PPARD) arranged in a column in the right side of Figure 4 were associated with the 14 differentially expressed genes

(Table 1) via different biological functions related to ketosis development in dairy cattle (Figure

4). In addition, out of the 34 genes in Figure 4, 25 genes harbored SNPs included in the Illumina bovine SNP60 (60K, 57,798 SNPs) BeadChip (Illumina, Inc., San Diego, CA, USA), as shown in

Table 3.

The 34 genes in Figure 4 were mainly involved in seven significant functions (FDR < 0.05): fatty acid transport, long-chain fatty acid transport, lipid transport, lipid localization, fatty acid metabolic process, long-chain fatty acid-CoA ligase activity, and fatty acid transmembrane transport. These genes might be good potential candidate genes to enhance cow’s resistance to the development of ketosis.

QTL Enrichment Analysis

A total of 2,000 annotated QTL from the cattle QTL database (Hu et al., 2013), overlapped with the 14 differentially expressed genes harboring 24 SNPs (Table 1). These QTL pertained to six different phenotypic classes (Production, Health, Milk, Exterior, Reproduction, and Meat and

Carcass). From these annotated QTL, 85.15% were associated with the “Milk” class of QTL and

6.4% with the class “Health” (Figure 5). The differentially expressed genes FN1 and PTK2 were enriched for QTL previously described for the traits “Ketosis” on chromosome 2 and “Milk iron content” on chromosome 14, respectively. Regarding the QTL health class, from the 2,000 annotated QTL, 0.45% of the QTL were partly associated with ketosis (Figure 6).

Discussion

Gene expression and GWAS analyses are two complementary approaches to determine the genetic background of traits of interest. Here, potential candidate genes associated with NEB, subclinical, and clinical ketosis were identified by combining the results obtained from a systematic review of genomic, transcriptomic, and proteomic studies. From the 14 differentially expressed genes identified by this study (ALB, CPT1A, PC, UGT2B10, SHROOM3, FN1, G3BP2,

ANXA3, PTK2, ACSL1, BIRC6, CSN1S1, HAO2 and RASSF6), the FN1 (fibronectin 1), ACSL1

(acyl-CoA synthetase long chain family member 1), PC (pyruvate carboxylase) and CPT1A

(carnitine palmityl transferase 1A) genes were previously reported to be associated with all three metabolic conditions. As described by Graber et al. (2010), an increased expression of ACSL1,

PC, and CPT1A genes, associated with an increase in the fatty Acid β-oxidation in dairy cows, may occur due to an elevated level of NEFA during negative energy balance postpartum.

The ACSL1 protein activates fatty acids to acyl-CoA, and it is involved in fatty acid transport across the plasma membrane, providing substrates for metabolic pathways that catabolize fatty acids or synthesize complex lipids (de Jong et al., 2007; Li et al., 2009). Ellis et al. (2010,

2011) studied the metabolic and tissue-specific regulation of acyl-CoA metabolism in mice. They reported the tissue-dependent effect of ACSL1 on fatty acid metabolism in the liver and the severe inhibition of fatty acid oxidation in muscle and adipose tissues that occurs when this gene is not expressed. Zhu et al. (2019) studied the expression patterns of hepatic genes involved in lipid metabolism in dairy cows with subclinical or clinical ketosis. They reported that the expression of the ACSL1 gene was significantly higher in both subclinical and clinical ketosis than in the control group. Furthermore, these authors suggested that hepatic fatty acid uptake, transport, and activation are significantly increased in cows with subclinical ketosis and clinical ketosis. Li et al.

(2012) also reported that the expression level of ACSL1 was significantly increased in cows with subclinical ketosis, with the highest gene expression value (FClog2= +5.7) for subclinical ketosis.

Subclinically affected cows have an increased chance of developing clinical ketosis and are less fertile than animals with normal serum ketone body concentrations (Rutherford et al., 2016).

The CPT1A gene has been associated with ketosis in dairy cattle. The carnitine palmitoyltransferase (CPT) system is responsible for the transport of fatty acids into the mitochondria (Cobb and Dukes, 1998). The CPT1 is one of the three enzymes that compose the

CPT system and has a crucial role in the regulation of long-chain fatty acids metabolism (McGarry and Brown, 1997). Previous studies investigated the activity and function of CPT1 in ruminants at the various stages of lactation in an attempt to better understand metabolic changes during ketosis and hepatic lipidosis (Knapp, 1990; Mizutani et al., 1999). Aiello et al. (1984) showed that the higher gluconeogenesis and ketogenesis may be associated with higher activity of CPT1 in early

lactation in dairy cows, which is possibly due to greater NEB. Knapp (1990) investigated nonketotic, nonlactating, and lactating dairy cows and reported no difference in CPT1 activity.

Similar results were shown by Mizutani et al. (1999), who reported no difference in CPT1 activity in cows in early, mid, and late lactation. Dann and Drackley (2005), investigating how prepartum nutrient intake and postpartum health status affects hepatic CPT1 activity, reported that alterations of this gene may not be involved in the etiology of ketosis in dairy cows. Kroezen et al. (2018) identified four SNPs located within the CPT1A gene associated with the first lactation estimated breeding value (EBV) for ketosis.

The PC gene is highly expressed in adipose tissue and is involved in fatty acid synthesis by providing oxaloacetate for conversion into citrate, which is cleaved to form oxaloacetate and acetyl-CoA (Ballard and Hanson, 1967). In the liver, the lack of oxaloacetate prevents the oxidation of acetyl-CoA produced from pyruvate and fatty acids and the excess of acetyl-CoA results in hepatic ketone body synthesis (De Vivo et al., 1977). Thus, pyruvate carboxylase produces oxaloacetate from pyruvate, which is a critical anaplerotic reaction that affects the capacity of the tricarboxylic acid cycle and supports the cataplerotic reactions of gluconeogenesis and fatty acid synthesis (Oliver et al., 2002). The PC gene plays a crucial role in hepatic gluconeogenesis to ensure a high level of glucose during early lactation for milk production (Baird et al., 1968; Donkin, 1999; Aschenbach et al., 2010). Previous studies highlighted the importance of gluconeogenesis in supporting the immediate metabolic requirements of dairy cows at calving

(Bell, 1995; Danfaer et al., 1995). Weber et al. (2013) reported an increase of hepatic expression of PC immediately at parturition. The deficiency of PC is associated with an increase in the ratio of acetoacetate and 3-hydroxybutyrate, resulting in a low NADH to NAD ratio inside the mitochondria.

The FN1 gene encodes a glycoprotein component of the extracellular matrix, which is associated with cell signaling and lipid metabolism (Loor et al., 2007). This protein has been shown to be differentially expressed in the liver of ketotic cows, but the interaction mechanism of FN1 with ketosis has not been established (Loor et al., 2007; Buitenhuis et al., 2013).

In addition to the four genes described above (CPT1A, PC, FN1, and ACSL1), other genes can be highlighted as potentially associated with NEB and subclinical and clinical ketosis, namely the CSN1S1 (alpha S1-casein) and HAO2 genes. The gene CSN1S1 encodes alpha S1-casein, one of the main proteins in milk, has two significant SNPs within its promoter region, which were associated with milk yield traits in German Angeln dairy cattle in a quantitative trait loci (QTL) mapping study (Sanders et al., 2006). Although most literature shows that the expression of this gene occurs in the epithelial cells of the mammary gland, Vordenbaumen et al. (2010) collected evidence that this gene can be expressed in human monocytes and has immunomodulatory properties. Alpha S1-casein has also been shown to exhibit chaperone-like activity (Bhattacharyya and Das 1999), which makes possible that it can stabilize other milk caseins. Thus, the protein encoded by the CSN1S1 gene has functions that go beyond nutritional aspects, and this gene may be expressed in other cells of the body in stressful situations, which needs to be further investigated.

The HAO2 gene product in humans is localized in the peroxisomes and is involved in the oxidation of long-chain fatty acids, such as 2-hydroxypalmitate, and is highly expressed in the liver and kidney (Jones et al., 2000). As mentioned earlier, ketosis has been associated with the development of fatty liver in cows. This state happens when the hepatic uptake of NEFA exceeds its oxidation (Drackley et al., 2001). The increase in the expression of the HAO2 gene in cows with clinical ketosis is consistent with the metabolic situation of these animals, but it may still not

be enough to help reverse this state (Schaff et al., 2012). The understanding of the fluctuations of the expression HAO2 gene in the development of ketosis needs to be better investigated by further experimental studies.

A permutation approach was used to identify the metabolic pathways associated with the

14 differentially expressed genes investigated. For all genes, patterns of pathways associated with different types of ketosis and NEB were identified. In the glyoxylate cycle, key enzymes isocitrate lyase and malate synthase convert isocitrate, and acetyl-CoA into succinate and malate. Thus, the glyoxylate cycle acts as a link between catabolic activities and biosynthetic capacities, enabling cells to utilize fatty acids or C2-units, such as acetate or ethanol as a carbon source. The Glyoxylate and dicarboxylate metabolism pathway was previously associated with the production of ketone bodies due to a deficit in the utilization of glucose, resulting in fatty acid oxidation and decomposition in mice (Song et al., 2015).

One of the processes involved in the Glyoxylate and dicarboxylate metabolism pathway is

Pyruvate metabolism, which is associated with the production of formate. A previous study found decreased levels of formic acid in dairy cows with clinical ketosis, indicating that the generation of formate from pyruvic acid decreased. Instead, the pyruvic acid was converted to acetyl-CoA, which resulted in ketogenesis and elevated blood ketone content (Wang et al., 2016). The HAO2 gene was present in all three significant pathways reported in this study, highlighting its possible association with NEB, subclinical and clinical ketosis.

The metabolic pathways found significantly associated with ketosis in this study were consistent with their biological role. Ketosis is a major metabolic disease, which is often observed when the gluconeogenic capacity in the liver fails to meet the increased tissue demands for glucose in dairy cattle (Aschenbach et al., 2010). Thus, the rapid adaptation of key metabolic pathways in

the liver is important for the cows to make a proper transition to support lactation. This study also raised other questions that may be investigated in future studies. For instance, the expression pattern of the 14 genes highlighted in this study could be further investigated to determine if they would have the same expression pattern in animals that normally recover from the NEB status compared to those animals that fail to recover.

The biological aspects involved in the pathogenesis of bovine ketosis is not completely understood. However, it is well known that there is a combination of intense adipose tissue mobilization and high glucose demand, mainly in early lactation (Weber et al., 2013). Adipose tissue mobilization is accompanied by high blood serum concentrations of non-esterified fatty acids (NEFA) (Adewuyi et al., 2005). The peroxisome pathway includes fatty acid-oxidation metabolism and is considered an alternate pathway for hepatic oxidation of NEFA, showing a considerable capacity for B-oxidation in liver tissues of cows (Drackley et al., 2001). The presence of an elevated concentration of NEFA in hepatic tissue in postpartum dairy cows is one of the more obvious biochemical features of ketosis (Emery et al., 1992; Reynolds et al., 2013). Grum et al.

(1996) demonstrated that esterification of palmitate by liver slices increased significantly at 1 d after calving compared to 21 d before or 21 d after calving, suggesting that additional fat fed to cows during the prepartum period does not decrease plasma NEFA concentrations. However,

Douglas et al. (1998, 2002) found that the reduction in liver triglycerides was largely attributable to the decreased dry matter intake (DMI) in cows fed with fat during the dry period, resulting in a decrease in peripartal NEFA concentrations in plasma.

Once absorbed by the liver, fatty acids are β-oxidized to acetyl-CoA, which will undergo complete oxidation through the tricarboxylic acid cycle, incomplete oxidation through ketogenesis, or synthesis of triglycerides and packing as very-low density lipoprotein for export

from the liver, or storage as liver lipids (Grummer, 1993). When the levels of acetyl-CoA exceed the capacity of the tricarboxylic acid cycle, there is increased production of ketone bodies and deposition of triglycerides. Thus, fatty acid metabolism in the liver and the development of ketosis are closely related (Grummer, 1993).

Based on Venn intersections, ten differentially expressed genes were found in common for

NEB, subclinical and clinical ketosis in liver tissue of dairy cows (ACSL1, ACACA, FABP1,

HMGCS1, CPT2, CPT1A, DGAT1, TP53, BAX, and PPARA; Figure 1). All these genes are involved in the pathway of lipid metabolism regulated by PPARA (Peroxisome Proliferator-

Activated Receptor Alpha). The highest number of commonly expressed genes (N=133) was obtained between NEB and clinical ketosis, since these are the two most explored metabolic states in terms of gene expression. NEB is a common status among dairy cows in the first few weeks of lactation, but most of the cows recover from this period through metabolic adaptation (Sundrum,

2015). Thus, an important goal is to be better understand how the expression of these genes in animals that naturally recover from the NEB differs from the expression in those animals that do not recover. Unfortunately, the available data in the literature was limited to allow for a better understanding of these differences.

The interaction gene-gene network (Figure 4) performed for the 14 differentially expressed genes revealed that these genes were co-expressed with 20 other genes. This analysis revealed that the ACSL1 gene showed the highest number of co-expression interactions (seven links). This gene was co-expressed with seven other genes, and among them, the genes CPT1A, PC, ACACB and

HAO2 are known to play critical roles in fatty acid metabolic processes (Zhu et al., 2019). In addition, the ACSL5 gene was co-expressed with the SLCO1B3 and ABCC3 genes. Therefore, the

results show that the ACSL family has an important connective role in the network, suggesting that it might play a central role in the occurrence of ketosis.

Besides the 430 significant SNPs from published GWAS, SNPs from the 60K Illumina panel were also investigated for the 34 genes in the network in Figure 4. In total, 25 genes in the network harbored 89 SNPs included in the 60K Illumina SNP panel (Table 3). From the 89 SNPs, the genes CPT1A and ACSL1 harbored 6 and 4 SNPs, respectively. The ACSL1 gene was co- expressed with the CPT1A gene. These two genes are involved in the oxidation of fatty acids and ketone body synthesis; as previously reported in this study, the expression of ACSL1 and CPT1A genes was significantly increased for subclinical ketosis in dairy cows (Zhu et al., 2019). Ellis et al. (2010) reported an 80% lower ACSL specific activity in adipose-specific ACSL1 knockout mice

(Acsl1A−/−) than in the control group (Acsl1flox/flox)), which showed that this gene has different expression patterns and activity between ruminants and non-ruminants. The identification of SNPs specifically associated with the studied traits highlights the importance of investigating these key genes. These SNPs are associated with important pathways (Table 2), which are directly related to biological mechanisms (i.e., fatty acid metabolism) potentially involved in the pathogenesis of ketosis. Thus, these SNPs might be potential candidates to be further investigated for their roles in ketosis susceptibility since these SNPs were located in coding regions of key genes that are differentially expressed in the liver of dairy cows under NEB, subclinical or clinical ketosis.

As expected for a trait with a low heritability and a highly polygenic architecture, such as ketosis, for the ten main genes, there was a considerable number of genes involved in the network, with several connections. The results from gene network analysis suggest the potential association of the identified genes with fatty acid metabolism underlying NEB, subclinical and clinical ketosis.

In this study, the genes FN1 and PTK2 (protein tyrosine kinase 2) were enriched for QTL previously described for ketosis and milk iron content, respectively, which is interesting because, as previously discussed, the interaction mechanism of FN1 with ketosis has not been established despite its known function in lipid metabolism in ketotic cows (Loor et al., 2007). The FN1 gene was enriched for QTL at 103kb on chromosome 2 harboring one SNP (rs109315364), while the

PTK2 gene was enriched for QTL around 28-29kb on chromosome 14, harboring six SNPs

(rs132789965, rs109558046, rs137202573, rs108968192, rs136634846, rs137438227). These observations suggest that these genes could be considered as candidate genes for further investigation.

To our knowledge, this study is the first to assess the overlap between published gene expression and GWAS results for NEB, subclinical and clinical ketosis in dairy cattle.

Conclusions

The findings in this investigation suggest that meta-analysis of gene expression studies combined with genome-wide association studies can contribute to a better understanding of the genetic background of negative energy balance, subclinical and clinical ketosis in dairy cattle. This investigation also identified the scarcity of studies on differentially expressed genes in the liver of cows during the transition period. Among the three metabolic conditions, the state of subclinical ketosis had the fewest candidate genes reported in the literature. This study highlighted 14 genes

(ACSL1, ALB, ANXA3, BIRC6, CPT1A, CSN1S1, FN1, G3BP2, HAO2, PC, PTK2, RASSF6,

SHROOM3 and UGT2B10) which harbor significant polymorphisms in reported genome-wide association studies; these polymorphisms have a great potential to be used as molecular markers in animal breeding programs for selection for cows less susceptible to ketosis. In addition, the

genes FN1 and PTK2 were enriched for QTL for ketosis and should be considered as candidate genes for further investigation. These findings may improve the understanding of negative energy balance and ketosis in dairy cows, which could enhance selection decisions and help with the development of biomarkers for early diagnosis and prevention of ketosis.

Acknowledgments

This work was supported by funding from the DairyGen Council of the Canadian Dairy Network and a Collaborative Research and Development grant from the Natural Science and Engineering

Research Council of Canada. Drs. F. Schenkel and G. Vargas thank the financial support in main part by Agriculture and Agri-Food Canada, and by additional contributions from Dairy Farmers of

Canada, the Canadian Dairy Network and the Canadian Dairy Commission under the Agri-Science

Clusters Initiative. As per the research agreement, aside from providing financial support, the funders have no role in the design and conduct of the studies, data collection and analysis, or interpretation of the data. Researchers maintain independence in conducting their studies, own their data, and report the outcomes regardless of the results. The decision to publish the findings rests solely with the researchers.

References

Adewuyi, A. A., E. Gruys, and F. J. C. M. Van Eerdenburg. 2005. Non esterified fatty acids

(NEFA) in dairy cattle. A review. Vet Q, 27:(3):117-126.

https://doi.org/10.1080/01652176.2005.9695192.

Aiello R.J., Kenna T.M., and Herbein J.H., Hepatic gluconeogenesis and ketogenic

interrelationships in the lactating cow. J. Dairy Sci. 1984; 67: 1707-1715.

Akbar, H.; M. Bionaz, B. Carlson, S. L. Rodriguez-Zas, R.E. Everts, H. A. Lewin, J. K. Drackley,

and J. J. Loor. 2013. Feed restriction, but not L-carnitine infusion, alters the liver

transcriptome by inhibiting sterol synthesis and mitochondrial oxidative phosphorylation

and increasing gluconeogenesis in mid-lactation dairy cows. J Dairy Sci. 96:2201–2213.

doi: 10.3168/jds.2012-6036.

Andersson, L. 1988. Subclinical ketosis in dairy cows. Vet Clin North Am Food Anim Pract.

4(2):233-51. https://doi-org.subzero.lib.uoguelph.ca/10.1016/S0749-0720(15)31046-X.

Aschenbach, J.R., N. B. Kristensen, S. S. Donkin, H. M. Hammon, and Penner, G.B. 2010.

Gluconeogenesis in dairy cows: the secret of making sweet milk from sour dough. IUBMB

Life. 62(12):869-77. https://doi.org/10.1002/iub.400.

Baird, G.D.; K. G. Hibbitt, G. D. Hunter, P. Lund, M. Stubbs, H. A. Krebs. 1968. Biochemical

aspects of bovine ketosis. Biochem J. 107(5): 683–689. https://doi: 10.1042/bj1070683.

Ballard, F.J., and R. W. Hanson. 1967. The citrate cleavage pathway and lipogenesis in rat adipose

tissue: replenishment of oxaloacetate. J Lipid Res. 8:73–79.

Bhattacharyya, J., and K. P. Das. 1999. Molecular Chaperone-like Properties of an Unfolded

Protein, alpha(s)-Casein. J Biol Chem. 274:15505–15509. DOI: 10.1074/jbc.274.22.15505

Bell, A.W. 1995. Regulation of organic nutrient metabolism during transition from late pregnancy

to early lactation. J Anim Sci. 73:2804–2819. https://doi.org/10.2527/1995.7392804x.

Biswal S., D. C. Nayak, and K. K. Sardar. 2016. Prevalence of ketosis in dairy cows in milk shed

areas of Odisha state, India, Vet World, 9(11):1242-1247. https://

https://dx.doi.org/10.14202/vetworld.2016.1242-1247.

Buitenhuis, A.J., U. K. Sundekilde, N. A. Poulsen, H. C. Bertram, L. B. Larsen, and P. Sorensen.

2013. Estimation of genetic parameters and detection of quantitative trait loci for

metabolites in Danish Holstein milk. J Dairy Sci. 96:3285–3295. https://doi-

org.subzero.lib.uoguelph.ca/10.3168/jds.2012-5914.

Cabrera, C.P., P. Navarro, J. E. Huffman, A. F. Wright, C. Hayward, H. Campbell, J. F. Wilson,

I. Rudan, N. D. Hastie, V. Vitart, and C. S. Haley. 2012. Uncovering networks from

genome-wide association studies via circular genomic permutation. G3 (Bethesda).

2(9):1067-75. https://doi:10.1534/g3.112.002618.

Cánovas, A., A., K. L. Reverter, R. L. DeAtley, M. L. Ashley, M. R. S. Colgrave, A. Fortes, S.

Islas-Trejo, L. Lehnert, G. Porto-Neto, G. Rincón, G. A. Silver, W. M. Snelling, J. F.

Medrano, M. G. Thomas . 2014. Multi-tissue omics analyses reveal molecular regulatory

networks for puberty in composite beef cattle. PLoS One. 9. doi:

10.1371/journal.pone.0102551

Carlson, M.R., H. Pagès, S. Arora, V. Obenchain, and M. Morgan. 2016. Genomic annotation

resources in R/Bioconductor. Methods Mol Biol. 1418:67-90. https://doi:10.1007/978-1-

4939-3578-9_4.

Cobb, J., and I. Dukes. 1998. Recent advances in the development of agents for the treatment of

type 2 diabetes. Annu Rep Med Chem, 33:213-222.

Danfaer, A., V. Tetens, and N. Agergaard. 1995. Review and an experimental study on the

physiological and quantitative aspects of gluconeogenesis in lactating ruminants. Comp

Bio-chem Physiol. 111B:201–210. https://doi-org.subzero.lib.uoguelph.ca/10.1016/0305-

0491(94)00242-M.

Dann, H. M. and Drackley J. K. Carnitine palmitoyltransferase I in liver of periparturient dairy

cows: effects of prepartum intake, postpartum induction of ketosis, and periparturient

disorders. J Dairy Sci. 2005 Nov;88(11):3851-9. doi: 10.3168/jds.S0022-0302(05)73070-

Douglas, G. N., J. K. Drackley, T. R. Overton, and H. G. Bateman. 1998. Lipid metabolism and

production by Holstein cows fed control or high fat diets at restricted or ad libitum intakes

during the dry period. J. Dairy Sci. 81(Suppl. 1):295. (Abstr.)

Douglas, G. N., J. Rehage, A. D. Beaulieu, A. O. Bahaa, and J. K. Drackley. 2002. Peripartal

changes in fatty acid profiles of blood, adipose tissue, and liver of dairy cows can be

modulated by diet. J. Dairy Sci. 85(Suppl. 1):185. (Abstr.) de Jong, H., A. C. Neal, R. A. Coleman, and T. M. Lewin. 2007. Ontogeny of mRNA expression

and activity of long-chain acyl-CoA synthetase (ACSL) isoforms in Mus musculus heart.

Biochim. Biophys Acta. 1771(1):75-82. https://doi:10.1016/j.bbalip.2006.11.007.

De Vivo, D.C., M. W. Haymond, M. P. Leckie, Y. L. Bussman, D. B. Jr, McDougal, and A. S.

Pagliara. 1977. The clinical and biochemical implications of pyruvate carboxylase

deficiency. J Clin Endocrinol Metab. 45:1281–96. https://doi.org/10.1210/jcem-45-6-

1281.

Donkin, S.S. 2010. Role of the endocrine pancreas in animal metabolism, growth and performance.

S.G. Pierzynowski, R. Zabielski (Eds.), Biology of the Pancreas in Growing Animals,

Elsevier, Ellis J, Li L, Wu P, Koves T, Ilkayeva O, et al. (2010) Adipose acyl-CoA

synthetase-1 (ACSL1) directs fatty acids towards β-oxidation and is required for cold

thermogenesis. Cell Metabolism. 12: 53–64.

Drackley, J. K., Overton, T. R., and G. N. Douglas. 2001. Adaptations of glucose and long-chain

fatty acid metabolism in liver of dairy cows during the periparturient period. J. Dairy Sci.

84:E100–E112. https://doi.org/10.3168/jds.S0022-0302(01)70204-4.

Du, X., T. Shen, H. Wang, X. Qin, D. Xing, Q. Ye, Z. Shi, Z. Fang, Y. Zhu, Y. Yang, Z. Peng,

C. Zhao, B. Lv, X. Li, G. Liu, and X. Li. 2018. Adaptations of hepatic lipid metabolism

and mitochondria in dairy cows with mild fatty liver. J Dairy Sci. 101:9544–9558.

https://doi.org/10.3168/jds.2018-14546.

Ellis, J., L. Li, P. Wu, T. Koves, O. Ilkayeva, R. D. Stevens, S. M. Watkins, D. M. Muoio, and R.

A. Coleman. 2010. Adipose acyl-CoA synthetase-1 (ACSL1) directs fatty acids towards β-

oxidation and is required for cold thermogenesis. Cell Metab. 12(1):53–64.

https://doi:10.1016/j.cmet.2010.05.012.

Ellis, J.M., S. M. Mentock, M. A. DePetrillo, T. R. Koves, S. Sen, S. M. Watkins, D. M. Muoio,

G. W. Cline, H. Taegtmeyer, G. I. Shulman, M. S. Willis, and R. A. Coleman. 2011. Mouse

cardiac acyl coenzyme a synthetase 1 deficiency impairs fatty acid oxidation and induces

cardiac hypertrophy. Mol Cell Biol. 31(6):1252-62. doi:10.1128/MCB.01085-10

Emery, R.S., J. S. Liesman, and T. H. Herdt. 1992. Metabolism of long chain fatty acids by

ruminant liver. J Nutr. 122(3 Suppl):832-7. https://doi.org/10.1093/jn/122.suppl_3.832.

Esposito, G., P. C. Irons, E. C. Webbb, and A. Chapwanya. 2014. Interactions between negative

energy balance, metabolic diseases, uterine health and immune response in transition dairy

cows. Anim Reprod Sci. 144:60–71. https://doi.org/10.1016/j.anireprosci.2013.11.007.

Fonseca, P.A.S., F.C. dos Santos, S. Lam, A. Suárez-Vega, F. Miglior, F. S. Schenkel, L. A. F.

Diniz, S. Id-Lahoucine, M. R. S. Carvalho, and A. Cánovas. 2018. Genetic mechanisms

underlying spermatic and testicular traits within and among cattle breeds: systematic

review and prioritization of GWAS results. J. Anim. Sci. 96:4978-4999.

https://doi:10.1093/jas/sky382.

Geishauser, T., K. Leslie, T. Duffield, D. Sandals, and V. Edge. 1998. The association between

selected metabolic parameters and left displaced abomasum in dairy cows. Zentralbl

Veterinärmed A. 45:499–511.

Geishauser, T., D. Leslie, D. Kelton, and T. Duffield. 2001. Monitoring for subclinical ketosis in

dairy herds. Compend Contin Educ Pract Vet 23:65- 71.

Gohary, K., M. W. Overton, M. Von Massow, S. J. LeBlanc, K. D. Lis-semore, and T. F. Duffield.

2016. The cost of a case of subclinical ketosis in Canadian dairy herds. Can Vet J. 57:728–

732.

Graber, M., S. Kohler, T. Kaufmann, M. G. Doherr, R. M. Bruckma-ier, and H. A. van Dorland.

2010. A field study on characteristics and diversity of gene expression in the liver of dairy

cows during the transition period. J Dairy Sci. 93:5200–5215.

https://doi.org/10.3168/jds.2010-3265.

Grum, D. E., J. K. Drackley, R. S. Younker, D. W. LaCount, and J. J. Veenhuizen. 1996. Nutrition

during the dry period and hepatic lipid metabolism of periparturient dairy cows. J. Dairy

Sci. 79:1850–1864.

Grummer, R.R. 1993. Etiology of lipid-related metabolic disorders in periparturient dairy cows. J

Dairy Sci. 76(12):3882-3896. https://doi.org/10.3168/jds.S0022-0302(93)77729-2.

Ha, N-T., J. J. Gross, A. van Dorland, J. Tetens, G. Thaller, M. Schlather, R. Bruckmaier, and H.

Simianer. 2015. Gene-based mapping and pathway analysis of metabolic traits in dairy

cows. PLoS One. 10:e0122325. https://doi:10.1371/journal.pone.0122325.

Ha, N. T., C. Drögemüller, C. Reimer, F. Schmitz-Hsu, R. M. Bruckmaier, H. Simianer, and J. J.

Gross. 2017. Liver transcriptome analysis reveals important factors involved in the

metabolic adaptation of the transition cow. J Dairy Sci, 100(11):9311-9323.

https://doi.org/10.3168/jds.2016-12454.

Hammon, H.M., G. Stürmer, F. Schneider, A. Tuchscherer, H. Blum, T. Engelhard, A. Genzel, R.

Staufenbiel, and W. Kanitz. 2009. Performance and metabolic and endocrine changes with

emphasis on glucose metabolism in high-yielding dairy cows with high and low fat content

in liver after calving. J. Dairy Sci. 92:1554–1566. https://doi.org/10.3168/jds.2008-1634.

Huang, H., J. Cao, Q. Hanif, Y. Wang, Y. Yu, S. Zhang, and Y. Zhang. 2019. Genome‐wide

association study identifies energy metabolism genes for resistance to ketosis in Chinese

Holstein cattle. Anim Genet. 50(4):376-380. https://doi:10.1111/age.12802

Jones, J.M., J. C. Morrell, and S. J. Gould. 2000. Identification and characterization of HAOX1,

HAOX2, and HAOX3, three human peroxisomal 2-hydroxy acid oxidases. J Biol Chem.

75(17):12590-7. http://doi:10.1074/jbc.275.17.12590.

Kadokawa, H., and B. G. Martin. 2006. A new perspective on Management of reproduction in

dairy cows: the need for detailed metabolic information, an improved selection index and

extended lactation. J. Reprod. Dev., 52: 161-168. https://doi:10.1262/jrd.17088

Kanehisa, M., S. Goto, Y. Sato, M. Furumichi, and M. Tanabe. 2012. KEGG for integration and

interpretation of large-scale molecular data sets. Nucleic Acids Res. 40:D109-14.

https://doi:10.1093/nar/gkr988.

Knapp, J. R. 1990. Lactation ketosis: Evaluation of current concepts. Ph.D. Diss., University of

California, Davis.

Kroezen, V., F. S. Schenkel, F. Miglior, C. F. Baes, and E. J. Squires. 2018. Candidate gene

association analyses for ketosis resistance in Holsteins. J Dairy Sci. 101(6):5240-5249.

https://doi.org/10.3168/jds.2017-13374.

Lee, S., K. Cho, M. Park, T. Choi, S. Kim, and C. Do. 2016. Genetic parameters of milk β-

hydroxybutyric acid and acetone and their genetic association with milk production traits

of Holstein cattle. Asian-Australas J Anim Sci, 29(11), 1530.

https://doi:10.5713/ajas.16.0310.

Li, L.O., J. M. Ellis, H. A. Paich, S. Wang, N. Gong, G. Altshuller, R. J. Thresher, T. R. Koves, S.

M. Watkins, D. M. Muoio, G. W. Cline, G. L. Shulman, and R. A. Coleman. 2009. Liver-

specific loss of long chain acyl-CoA synthetase-1 decreases triacylglycerol synthesis and

beta-oxidation and alters phospholipid fatty acid composition. J Biol Chem. 284(41):

27816-27826. https://doi:10.1074/jbc.M109.022467.

Li, P., X. B. Li, S. X. Fu, C. C. Wu, X. X. Wang, G. J. Yu, M. Long, Z. Wang, and G. W. Liu.

2012. Alterations of fatty acid β-oxidation capability in the liver of ketotic cows. J Dairy

Sci, 95(4):1759-1766. https://doi.org/10.3168/jds.2011-4580.

Leek, B.F., and W. O. Reece. 2014. Digestão no Estômago do Ruminante. In Dukes, fisiologia dos

animais domésticos (12th ed., pp. 404–437). Guanabara Koogan.

Loor, J.J., R. E. Everts, M. Bionaz, H. M. Dann, D. E. Morin, R. Oliveira, R., S. L. Rodriguez-

Zas, J. K. Drackley, and H. A. Lewin. 2007. Nutrition-induced ketosis alters metabolic and

signaling gene networks in liver of periparturient dairy cows. Physiol. Genomics 32:105–

116. https://doi:10.1152/physiolgenomics.00188.2007.

Lu, J., E. A. Fernandes, A. E. P Cano, J. Vinitwatanakhun, S. Boeren, T. van Hooijdonk, A. van

Knegsel, J. Vervoort, and K. A. Hettinga. 2013. Changes in Milk Proteome and

Metabolome Associated with DryPeriod Length, Energy Balance, and Lactation Stage in

Post parturient Dairy Cows. J Proteome Res. 5;12(7):3288-96. https://doi:

10.1021/pr400130.

McArt, J. A. A., D. V. Nydam, and G. R. Oetzel. 2012. Epidemiology of subclinical ketosis in

early lactation dairy cattle. J Dairy Sci. 95(9), 5056–5066.

https://doi.org/10.3168/jds.2012-5443.

McArt, J. A. A., D. V. Nydam, and M. W. Overton. 2015. Hyperketonemia in early lactation dairy

cattle: A deterministic estimate of component and total cost per case. J Dairy Sci. 98:2043–

2054. https://doi.org/10.3168/jds.2014-8433.

McCabe, M., S. Waters, D. Morris, D. Kenny, D. Lynn, and C. Creevey. 2012. RNA-seq analysis

of differential gene expression in liver from lactating dairy cows divergent in negative

energy balance. BMC Genomics. 13:193. https://doi:10.1186/1471-2164-13-193.

McCarthy, S. D., M. Waters, A. Kenny, G. Diskin, R. Fitzpatrick, J. Patton, D. C. Wathes, and D.

G. Morris. 2010. Negative energy balance and hepatic gene expression patterns in high-

yieldingdairy cows during the early postpartum period: a global approach. Physiol

Genomics. 42A: 188–199. https://doi:10.1152/physiolgenomics.00118.2010.

McGarry J.D. and Brown N.F. 1997. The mitochondrial carnitine palmitoyltransferase system.

From concept to molecular analysis. Eur. J. Biochem. 244: 1-14

McGarry, J.D., and D. W. Foster. 1980. Regulation of hepatic fatty acid oxidation and ketone body

production. Annu Rev Biochem. 49:395-420.

https://doi:10.1146/annurev.bi.49.070180.002143.

Mizutani H., Sako T., Toyoda Y., Kawabata T., Urumuhang N., Koyama H., Motoyoshi S.

Preliminary studies on hepatic carnitine palmitoyltransferase in dairy cattle with or without

fatty liver. Nayeri, S., F. Schenkel, A. Fleming, V. Kroezen, M. Sargolzaei, C. Baes, A.

Cánovas, J. Squires, and F. Miglior. 2019. Genome-wide association analysis for β-

hydroxybutyrate concentration in Milk in Holstein dairy cattle. BMC Genetics, 20(1), 58.

https://doi.org/10.1186/s12863-019-0761-9.

Owen, O. E., S. C. Kalhan, and R. W. Hanson. 2002. The key role of anaplerosis and cataplerosis

for citric acid cycle function. J Biol Chem, 277(34), 30409-30412. https://doi:

10.1074/jbc.R200006200.

Pryce, J.E., K. P. Gaddis, A. Koeck, C. Bastin, M. Abdelsayed, N. Gengler, F. Miglior, B.

Heringstad, C. Egger-Danner, and K. Stock. 2016. Invited review: Opportunities for

genetic improvement of metabolic diseases. J Dairy Sci. 99: 6855-6873.

https://doi:10.3168/jds.2016-10854.

R Core Team, 2019. R: A language and environment for statistical computing. R. Foundation for

Statistical Computing, Vienna. R Core Team.

Renate, K., and H. Cernescu. 2009. Effects of negative energy balance on reproduction in dairy

cows. Lucrări Stiinłifice Medicină Veterinară, XLII(2).

Reynolds, C.K., P. C. Aikman, B. Lupoli, D. J. Humphries, and D. E. Beever. 2003. Splanchnic

metabolism of dairy cows during the transition from late gestation through early lactation.

J Dairy Sci. 86(4):1201-17. https://doi.org/10.3168/jds.S0022-0302(03)73704-7.

Rutherford, A. J., G. Oikonomou, and R. F. Smith.2016. The effect of subclinical ketosis on

activity at estrus and reproductive performance in dairy cattle. J Dairy Sci. 99(6);4808-

4815. https://doi.org/10.3168/jds.2015-10154.

Sanders, K., J. Bennewitz, N. Reinsch, G. Thaller, E.-M. Prinzenberg, C. Kühn C, E. Kalm . 2006.

Characterization of the DGAT1 mutations and the CSN1S1 promoter in the German Angeln

dairy cattle population. J Dairy Sci. 89(8):3164-74.https:// 10.3168/jds.S0022-

0302(06)72590-5.

Schäff, C., S. Börner, S. Hacke, U. Kautzsch, D. Albrecht, H. M. Hammon, M. Röntgen, and B.

Kuhla. 2012. Increased Anaplerosis, TCA Cycling, and Oxidative Phosphorylationin the

Liver of Dairy Cows with Intensive Body Fat Mobilization during Early Lactation. J

Proteome Res. 11(11):5503-14. https://doi:10.1021/pr300732n.

Seymour D. J., A. Cánovas, C. F. Baes, T. C. S. Chud, V. R. Osborne, J. P. Cant, L. F. Brito, B.

Gredler-Grandl, R. Finocchiaro, R. F. Veerkamp, Y. de Haas, and F. Miglior. 2019. Invited

review: Determination of large-scale individual dry matter intake phenotypes in dairy

cattle. J Dairy Sci. 102(9):7655-7663. https://doi.org/10.3168/jds.2019-16454.

Song, L., H. Liu, Y. Wang, Y. Wang, J. Liu, Z. Zhou, H. Chu, P. Zhuang, and Y. Zhang. 2015.

Application of GC/MS-based metabonomic profiling in studying the therapeutic effects of

Huangbai–Zhimu herb-pair (HZ) extract on streptozotocin-induced type 2 diabetes in mice.

J Chromatogr B Analyt Technol Biomed Life Sci. 997:96-104.

https://doi.org/10.1016/j.jchromb.2015.05.003.

Sundrum, A. 2015. Metabolic disorders in the transition period indicate that the dairy cows’ ability

to adapt is overstressed. Animals, 5(4), 978-1020. https://doi:10.3390/ani5040395.

Tetens, J., C. Heuer, I. Heyer, M. S. Klein, W. Gronwald, W. Junge, P. J. Oefner, G. Thaller, and

N. Krattenmacher. 2015. Polymorphisms within the APOBR gene are highly associated

with milk levels of prognostic ketosis biomarkers in dairy cows. Physiol Genomics.

47(4):129-137. https://doi:10.1152/physiolgenomics.00126.2014. van der Drift, S., K. Van Hulzen, T. Teweldemedhn, R. Jorritsma, M. Nielen, H. Heuven. 2012.

Genetic and nongenetic variation in plasma and milk β-hydroxybutyrate and milk acetone

concentrations of early-lactation dairy cows. J Dairy Sci. 95: 6781-6787.

https://doi:10.3168/jds.2012-5640.

Vordenbäumen, S., A. Braukmann, K. Petermann, A. Scharf, E. Bleck, A.a von Mikecz, J. Jose

and M. Schneider. 2011. Casein α s1 Is Expressed by Human Monocytes and Upregulates

the Production of GM-CSF via p38 MAPK. J Immunol. 186(1):592-601. https://doi:

10.4049/jimmunol.1001461.

Xu, C., and Z. Wang. 2008. Comparative proteomic analysis of livers from ketotic cows. Vet Res

Commun. 32:263–273. https://doi:10.1007/s11259-007-9028-4.

Zhou, C., C. Li, W. Cai, S. Liu, H. Yin, S. Shi, Q. Zhang, and S. Zhang. 2019. Genome-wide

association study for milk protein composition traits in a Chinese Holstein population using

a single-step approach. Front Genet. 10:72. https://doi:10.3389/fgene.2019.00072.

Zhu, Y., G. Liu, X. Du, Z. Shi, M. Jin, X. Sha, X. Li, Z. Wang, and X. Li. 2019. Expression

patterns of hepatic genes involved in lipid metabolism in cows with subclinical or clinical

ketosis. J Dairy Sci. 102:1725-1735. https://doi.org/10.3168/jds.2018-14965.

Wang, Y.,Y. Gao, C. Xia, H. Zhang, W. Qian, and Y. Cao. 2016. Pathway analysis of plasma

different metabolites for dairy cow ketosis. Ital J Anim Sci. 15:545-551.

https://doi.org/10.1080/1828051X.2016.1180643.

Warde-Farley, D., S. L. Donaldson, O. Comes, K. Zuberi, R. Badrawi, P. Chao, M. Franz, C.

Grouios, F. Kazi, C. T. Lopes, A. Maitland, S. Mostafavi, J. Montojo, Q. Shao, G.

Wright, G. D. Bader, and Q. Morris. 2010. The GeneMANIA prediction server: biological

network integration for gene prioritization and predicting gene function. Nucleic Acids

Res. 38, W214–20. https://doi: 10.1093/nar/gkq537.

Weber, C., C. Hametner, A. Tuchscherer, B. Losand, E. Kanitz, W. Otten, H. Sauerwein, R.

M. Bruckmaier, F. Becker, W. Kanitz, and H. M. Hammon. 2013. Hepatic gene

expression involved in glucose and lipid metabolism in transition cows: Effects of fat

mobilization during early lactation in relation to milk performance and metabolic changes.

J. Dairy Sci. 96:5670-5681. https://doi.org/10.3168/jds.2012-6277.

Tables and Figures Table 1. Significant SNPs and their position within the start and end coordinates of each differentially expressed gene obtained through systematic review TRANSCRIPTOMICS Metabolic Technique Gene symbol1 FC (log2)2 Reference SNP3 Chr4 Pos(bp)5 condition FN1 -2.51 rs109315364 2 103,403,451 rs109901151 6 88,494,442 ALB 4.43 rs110246305 6 88,498,400 G3BP2 -1.47 rs109793149 6 90,557,327 ANXA3 -1.73 rs136734574 6 93,308,071 Microarray Loor et al., 2007 rs132789965 14 2,857,000 rs109558046 14 2,909,929 Clinical Ketosis rs137202573 14 2,915,391 PTK2 -1.26 rs108968192 14 2,916,658 rs136634846 14 2,936,478 rs137438227 14 2,940,147 PC 1.60 rs42197374 29 44,864,841 rs378674931 27 15,179,505 qPCR ACSL1 0.50 Zhu et al., 2019 rs42194370 29 46,165,677 CPT1A -0.60 rs42194371 29 46,165,505 1.7 Zhu et al., 2019 ACSL1 rs378674931 27 15,179,505 5.7 Li et al., 2012 Subclinical Ketosis qPCR rs42194370 29 46,165,677 CPT1A 1.50 Zhu et al., 2019 rs42194371 29 46,165,505 0.84 rs41600541 6 84,446,558 UGT2B10 rs42197374 29 44,864,841 PC 2.0 Ha et al., 2017 rs43338548 6 88,650,827 NEB6 RNA-Seq7 RASSF6 -1.10 rs110360281 6 88,668,261 McCabe et al., rs42608667 6 91,422,949 SHROOM3 1.10 2012 rs42606224 6 91,447,170

ACSL1 1.28 rs378674931 27 15,179,505 rs110571223 11 15,050,414 BIRC6 -1.04 rs43672311 11 15,081,731 Akbar et al., 2013 1.5 PC Hammon et al., rs42197374 29 44,864,841 5.2 2009 qPCR8 McCarthy et al., 3.4 rs42194370 29 46,165,677 CPT1A 2010 rs42194371 29 46,165,505 2.2 Du et al., 2018 PROTEOMICS Metabolic condition Gene symbol FC (log2) Reference SNP Chr Pos (bp) HAO2 0.19 rs41588443 3 23,737,990 Clinical Ketosis Schaff et al., 2012 rs109901151 6 88,494,442 ALB 1.30 rs110246305 6 88,498,400 Subclinical Ketosis CSN1S1 0.25 Xu et al., 2008 rs110899052 6 87,145,250 1Gene symbol = HGNC gene name 2FC (log2) = Fold change 3SNP = single nucleotide polymorphism 4Chr = chromosome 5Pos (bp) = Position in base pairs 6NEB = negative energy balance 7RNA-Seq = RNA-Sequencing 8qPCR = quantitative polymerase chain reaction.

Table 2. Significant pathways, genes and markers involved in ketosis from genome-wide association studies Pathway SNP1 Chr2 Pos (bp)3 Gene symbol4

Glyoxylate and dicarboxylate rs41588443 3 23737990 HAO2 metabolism

rs41588443 3 23737990 HAO2

Metabolic pathways rs378674931 27 15179505 ACSL1

rs42197374 29 44864841 PC

rs41588443 3 23737990 HAO2 Peroxisome rs378674931 27 15179505 ACSL1

1SNP = single nucleotide polymorphism 2Chr = chromosome 3Pos (bp) = Position in base pairs 4Gene symbol = HGNC gene name

Table 3. Significant differentially expressed genes with significant SNPs in published GWAS for NEB, subclinical and clinical ketosis, chromosome, SNP identification, SNP position, and number of SNPs harbored by the genes included in the Illumina bovine SNP60 BeadChip

Gene symbol1 Chr2 SNP identification3 Pos (bp)4 Number of SNPs5

RXRB 23 - - -

FN1 2 rs109315364 103403451 2

rs109901151 88494442 ALB 6 1 rs110246305 88498400

FABP6 5 - - -

G3BP2 6 rs109793149 90557327 2

ANXA3 6 rs136734574 93308071 1

rs132789965 2857000 PTK2 14 16 rs109558046 2909929

rs137202573 2915391

rs108968192 2916658

rs136634846 2936478

rs137438227 2940147

PC 29 rs42197374 44864841 2

ACACB 17 - - 1

UGT2B10 6 rs41600541 84446558 1

rs43338548 88650827 RASSF6 6 2 rs110360281 88668261

rs42608667 91422949 SHROOM3 6 7 rs42606224 91447170

ACSL5 26 - - 3

rs110571223 15050414 BIRC6 11 10 rs43672311 15081731

HAO2 3 rs41588443 23737990 1

CSN1S1 6 rs110899052 87145250 1

ACSL1 27 rs378674931 15179505 4

rs42194370 46165677 CPT1A 29 6 rs42194371 46165505

EDIL3 5 - - 10

CPT1C 18 - - -

CPT1B 5 - - -

ACSL6 7 - - 2

ACSL3 2 - - 1

ABCC3 19 - - 5

TRIP6 25 - - -

APBB1IP 13 - - 1

CHRNG 2 - - -

CHRNB1 19 - - -

PPARG 22 - - 4

PPARD 23 - - 3

SLCO1A2 5 - - 1

SLCO1B3 5 - - 2

SLCO1B1 12 - - -

SLCO10A1 14 - - -

1Gene symbol = HGNC gene name 2Chr = chromosome 3SNP = single nucleotide polymorphism 4Pos (bp) = SNP position in base pairs 5 Number of SNPs included in the Illumina bovine SNP60 BeadChip

CHAPTER THREE: Genome-wide association study and functional analyses of clinical and subclinical ketosis in Holstein cattle

R.A.N. Soares1*, G. Vargas1, T. Duffield2, F. Schenkel1#, and E. J. Squires1#

(Prepared for submission to Journal of Dairy Science)

1 Department of Animal Biosciences, Center for Genetic Improvement of Livestock, University of

Guelph, Guelph, Ontario, Canada, N1G 2W1

2 Department of Population Medicine, Ontario Veterinary College, University of Guelph, Guelph,

Ontario, Canada, N1G 2W1

# These authors had equal contribution as senior authors

* Corresponding author: Riani Ananda Nunes Soares

50 Stone Rd E Guelph, ON N1G 2W1, c/o Department of Animal Biosciences

519-993-8724 [email protected]

Abstract

Ketosis affects high yielding cows and it is one of the most frequent metabolic diseases in dairy cows, characterized by high concentrations of ketone bodies in blood, urine and milk, and causing high economic losses. The search for polymorphic genes, whose alleles have different effects on resistance to developing the disease, is of extreme importance to help select less susceptible animals. The aims of this study were to identify genomic regions associated with clinical and subclinical ketosis (β-hydroxybutyrate (BHB) concentration) in North American

Holstein dairy cattle and to investigate these regions to identify candidate genes and metabolic pathways associated with these traits. To achieve this, a genome-wide association study (GWAS)

was performed for four traits: Clinical Ketosis Lactation 1 (CK1), Clinical Ketosis Lactation 2 to

5 (CK2), Subclinical Ketosis Lactation 1 (SCK1), and Subclinical Ketosis Lactation 2 to 5 (SCK2).

The estimated breeding values (EBVs) from 77,277 cows and 7,704 bulls were de-regressed and used as pseudo-phenotypes in the GWAS. The top-20 genomic regions explaining the largest proportion of the genetic variance were investigated for putative genes associated with the traits through functional analyses. Regions of interest were identified on chromosomes 2, 5 and 6 for

CK1, 3, 6 and 7 for CK2, 1, 2 and 12 for SCK1 and 20, 11 and 25 for SCK2. The highlighted genes potentially related to clinical and subclinical ketosis included ACAT2 and IFG1. Enrichment analysis of the list of candidate genes for clinical and subclinical ketosis showed molecular functions and biological processes involved in fatty acid metabolism, lipid metabolism and inflammatory response in dairy cattle. Several genomic regions and SNPs related to susceptibility to ketosis in dairy cattle that were previously described in other studies were confirmed. In addition, the novel genomic regions identified in this study aid to characterize the more important genes and pathways that explain the susceptibility to clinical and subclinical ketosis.

Introduction

Increased milk production in high yielding cows is associated with increased health issues, such as metabolic diseases. Because the cow is unable to satisfy the requirement for energy early in lactation, it enters in negative energy balance (NEB). Some cows naturally recover from this state, but others do not recover and develop metabolic diseases, such as ketosis (Andersson, 1988;

L. W. Sun et al., 2014). Clinical and subclinical ketosis are defined by the presence of clinical symptoms and β-hydroxybutyrate (BHB) levels in the blood, respectively (Duffield et al., 2009;

Gordon et al., 2013).

Being one of the most frequent metabolic diseases in dairy cows, ketosis causes high economic losses as the result of decreased productivity and costs of treatment. Due to the dairy industry's focus on the genetic improvement of milk production, energy demands will increase, and cows will continue to present with ketosis (Duffield, 2000; Esposito et al., 2014). In cases of subclinical ketosis, the drop in milk production could reach 1 to 1.5 liters per day (Duffield et al.,

1997). Thus, preventing ketosis is one of the most effective strategies to avoid economic losses associated with milk production since, on average, each case of subclinical and clinical ketosis can result in loss of $203 and $289, respectively (Geishauser et al., 1998; Gohary et al., 2016; McArt et al., 2015; Steeneveld et al., 2020).

The clinical symptoms of this metabolic disease include weight loss, loss of appetite and decreased milk yield, along with increased levels of ketone bodies in the blood and milk (Duffield et al., 2009; Gordon, LeBlanc, & Duffield, 2013). Many misrepresentations of subclinical ketosis or misdiagnosed clinical ketosis occur because of the difficulty in diagnosing the condition, which makes it difficult to resolve the issue. In addition, ketosis in high producing dairy cows is a multifactorial trait with low heritability. Heritability estimates for clinical ketosis ranging from

0.02 to 0.16 and for subclinical ketosis ranging from 0.04 to 0.17 have been reported (S. Lee et al.,

2016; Pryce et al., 2016), which shows the need to better understand the genetic origin of this problem (Parker Gaddis et al., 2018). The search for polymorphic genes, whose alleles are associated with different levels of resistance to the development of the disease, is of extreme importance to help selecting for less susceptible animals. In this sense, genome-wide association

(GWA) studies are a powerful strategy for detecting genomic regions involved with metabolic processes associated with this trait (Ku et al., 2010).

The aims of this study were: 1) to conduct a genome-wide association study (GWAS) to screen for genetic variants that may influence clinical and subclinical ketosis (BHB concentrations) susceptibility in North American Holstein dairy cattle, and 2) to pinpoint candidate genes and pathways with biologically relevant associations with clinical and subclinical ketosis susceptibility.

Materials and Methods

Animals and dataset description

The dataset used in this study was provided by the Canadian Dairy Network (CDN), a member of Lactanet (Guelph, Ontario, Canada, www.lactanet.ca) and contained pedigree, genotypes, and estimated breeding values (EBVs) for cows and bulls.

Comparing the first lactation with later lactations, previous studies found that cows respond to NEB differently and, for this reason, the first and later (from 2 to 5) lactations were considered as different traits (Chapinal et al., 2012; Jamrozik et al., 2016). This way, four traits were defined:

Clinical Ketosis Lactation 1 (CK1), Clinical Ketosis Lactation 2 to 5 (CK2), Subclinical Ketosis

Lactation 1 (SCK1), and Subclinical Ketosis Lactation 2 to 5 (SCK2). Milk BHB measurements were predicted by mid-infrared spectroscopy (MIR) in routine DHI milk analyses, and phenotypes for clinical ketosis were recorded either by farmers as part of the Canadian health recording system or by veterinarians via Valacta (a member of Lactanet). A total of 84,981 animals (77,277 cows and 7,704 bulls) with EBVs for CK1, CK2, SCK1, and SCK2 provided by CDN were used for the analyses. The pedigree consisted of 329,958 individuals traced back 7 generations.

Genotypic data and quality control

A total of 9,161 Canadian Holstein animals were genotyped with the Illumina bovine

SNP60 (60K; 57,798 SNPs) BeadChip (Illumina, Inc., San Diego, CA, USA). These SNPs have passed the standard quality control used by CDN. A sample of 652 Canadian Holstein cows were also genotyped with a low-density array containing 998 putative ketosis markers (1K; 998 SNPs).

These 998 markers are a combination of novel and previous reported SNP markers that are located within or near candidate genes related to ketosis (Kroezen et al., 2018). The 60K and 1K in the

652 cows were combined to create a new panel (61K SNP). Genotype imputation from the 60K to

61K panel was performed for the remaining animals using the FImpute V2.2 software (Sargolzaei et al., 2014). The algorithm in FImpute is based on the combination of family and population methods to reconstruct haplotypes (Sargolzaei et al., 2014). To estimate the accuracy of imputation, animals were randomly selected from the reference population (61K panel) and their genotypes were masked down to the 60K panel. The squared correlation between imputed genotypes and true genotypes (푟2) was used as a measure of imputation accuracy (Browning &

Browning, 2009) and it was 푟2 = 0.98.

After imputation, a quality control was performed using the BLUPF90 software (Misztal et al., 2018). The genotyping quality control (QC) filtered out markers with minor allele frequency

(MAF) lower than 1% and with a heterozygosity that deviates by more than 15% from the expected value under Hardy-Weinberg equilibrium (HWE). After quality control, 57,957 SNPs remained for further analyses.

De-regressed EBV

De-regressed EBVs (dEBVs) computed following Garrick et al. (2009) were used as response variables for genomic prediction. This method results in dEBVs containing only

information for each animal along with their descendant`s information because it eliminates shrinkage in the EBV and adjusts for ancestral information.

Genome-wide association analyses

The SNP effects were estimated using the weighted single-step method (wssGBLUP) proposed by Wang et al. (2012). The wssGBLUP model can be described as:

∗ 풚 = ퟏµ + 풁풂풂 + 풆

where 풚∗ is the vector of pseudo phenotypes (dEBVs); 1 is a vector of ones, µ is the overall mean;

풁풂 is an incidence matrix that relates animals to dEBVs; 풂 is the vector of direct additive genetic

2 2 effects and 풆 is the vector of random residuals. It was assumed that a~N(0, H휎푎 ) and e~N(0, R휎푒 ),

2 where H is the relationship matrix based on genomic and pedigree information, 휎푎 is the additive genetic variance, R is a diagonal matrix, whose elements account for the differences in the

2 reliabilities of the dEBVs, and 휎푒 is the residual variance. The inverse of H matrix can be defined as (Aguilar et al., 2010):

−1 −1 0 0 푯 = 푨 + [ −1 −1] 0 푮 − 푨22

where A is the numerator relationship matrix based on pedigree for all animals; 푨ퟐퟐ is the numerator relationship matrix based on pedigree for genotyped animals only; and G is the genomic relationship matrix for genotyped animals.

The solutions of SNP effect estimates (풖̂) were obtained using the formula: 풖̂ =

푫풁′[풁푫풁′]−1풂̂, where: D is a diagonal matrix with weights for SNPs; Z is an incidence matrix of 78

genotyped for each locus, Z’ is the transpose of Z matrix, and 풂̂ is the vector of dEBVs of genotyped animals. The 풖̂ vector and the D matrix were derived as follows (Wang et al., 2012):

1. In the first iteration 푡 = 0 and 푫풕 = 푰, where 푡 is the iteration number, 푫풕 is matrix D at iteration

푡, and I is an identity matrix.

2. Calculate the SNP effects at iteration 푡(풖̂(풕)).

∗ ∗ 2 3. Recalculate the diagonal elements of D to obtain 푫(풕+ퟏ) as: 푑푖(푡+1) = 푢̂푖(푡)2푝푖(1 − 푝푖) for all

th SNPs, where 푝푖 is the allele frequency of the reference allele of the i SNP.

∗ ∗ 4. Normalize 푫(풕+ퟏ) = (푡푟(푫(ퟎ))/푡푟(푫(풕+ퟏ)))푫(풕+ퟏ).

5. 푡 = 푡 + 1.

6. Exit, or loop to step 2.

The SNP effects and weights were iteratively recomputed over two iterations. The iteration process increases the weight of SNPs with large effects and decreases those with small effects, resulting in an increasing proportion of variance explained by the remaining markers. SNP effects were calculated using POSTGSF90 and sliding windows of 20 adjacent SNPs (Misztal, 2019).

Functional annotation analyses

The top-20 genomic regions explaining the largest proportion of genetic variance were investigated for putative genes associated with subclinical and clinical ketosis. The ENSEMBL

Biomart Martview application (Ensembl Genes 97, http://www.ensembl.org/biomart/martview) was used for identification of the genes located within those windows as well as nucleotide and protein sequences of these genes, using the ARS-UCD1.2 assembly as the reference map. All unique sequences were aligned against the sequences in the National Center for Biotechnology

Information (NCBI, www.ncbi.nlm.nih.gov) non-redundant (nr) database using the BLASTp algorithm from the Blast2GO software (Götz et al., 2008). Sequences with a BLAST match were

mapped and annotated. Gene Ontology (GO) analysis was performed considering the three GO categories: biological process, molecular function, and cellular component. Gene Ontology significance levels (P<0.05) were computed following Fisher’s exact test for multiple testing in

Blast2GO, comparing the list of positional candidate genes vs the functions from the reference

(background genes) using the bovine genome assembly ARS-UCD1.2 (release 97). Metabolic pathways associated with significant sequences were identified using the Kyoto Encyclopedia of

Genes and Genome (KEGG) (Kanehisa et al., 2016; Kanehisa & Goto, 2000). Additionally, a gene network analysis was performed using STRING database (Szklarczyk et al., 2019), to identify specific protein-protein interactions and “key nodes” were selected from proteins with degree ≥4,

Betweenness Centrality (BC) ≥ 0.05 and Closeness Centrality (CC) ≥ 0.2.

Results

Genome-wide association analyses

The proportion of variance explained by the top-20 sliding windows associated with clinical ketosis 1 (CK1) are presented in Table 1, clinical ketosis 2 (CK2) are presented in Table

2, subclinical ketosis 1 (SCK1) are presented in Table 3, and subclinical ketosis 2 (SCK2) are presented in Table 4. The highest proportion of variance observed using wssGBLUP analyses were found on chromosomes 6 for CK1 and CK2, and on chromosomes 2 and 20 for SCK1 and SCK2, respectively. The window that explained the highest proportion of variance overlapped for both traits (CK1 and CK2). The sum of the proportion of variance explained by the top-20 genomic windows was 16.7%, 18.0%, 19.2% and 13.4% for CK1, CK2, SCK1, and SCK2, respectively.

The genetic variance explained by the SNPs in the GWAS analyses are shown as

Manhattan plots for CK1 (Figure 1), CK2 (Figure 2), SCK1 (Figure 3) and SCK2 (Figure 4). The

results indicated important genomic regions on chromosomes 2, 5, 6, 11, 17, and 23 for CK1, 3,

6, 7, 15 and 20 for CK2, 1, 2, 3, 12 and 23 for SCK1, which explained more than 1% of the total genetic variance. For SCK2, none of the windows explained more than 1% of the genetic variance.

Function annotation and gene network analyses

The annotated genes within the top-20 sliding windows associated with the four traits using wssGBLUP analyses are presented in Table 5. For CK1, 114 annotated genes were found within

19 genomic windows identified by the GWAS. For CK2, 126 genes were found within the top-20 genomic windows. For SCK1, a total of 145 genes were identified within 18 genomic windows.

For SCK2, the highest number of annotated genes was found, i.e. 166 genes within 16 genomic windows.

Gene ontology enrichment analyses was performed for all candidate genes for the four traits to identify genes involved in known metabolic pathways. For CK1 (Supplementary Table 1),

29 GO terms were found, whereas 60 GO terms were associated with CK2 (Supplementary Table

2). For SCK1 (Supplementary Table 3) and SCK2 (Supplementary Table 4), 31 and 62 GO terms were found, respectively.

Based on the functional annotation results, significant pathways were found for subclinical and clinical ketosis. Highlighted pathways included synthesis and degradation of ketone bodies, biosynthesis of unsaturated fatty acids (Table 10), fatty acid elongation (Tables 10; 13), fatty acid degradation (Tables 10; 12; 13), arachidonic acid metabolism (Tables 11; 12; 13), and pyruvate

metabolism (Tables 10; 12). The ACAT2 gene was involved in several pathways for CK1 and

SCK1 (Tables 10 and 12).

The relationships among the genes in the network were determined by co-expression, gene fusion, protein homology, gene-neighborhood, and gene co-occurrence. A network containing 97 nodes and 61 edges between them (PPI enrichment P = 2.2e−16; Figure 5) was constructed for CK1.

Based on degree ≥4, BC ≥0.05 and CC ≥0.2, six key nodes were identified, with Interleukin 10

(IL10) having the highest degree (6), BC (0.72) and CC (0.69) being the most interconnected and powerful to disseminate information through the network. A network containing 120 nodes and

114 edges between them (PPI enrichment P = 7.11e−15; Figure 6) was constructed for CK2.

Apolipoprotein 1 (APOA1) was identified as a key node having the highest degree (6), BC (0.31) and CC (0.24). A network containing 135 nodes and 123 edges between them (PPI enrichment P

= 1.1e−15; Figure 7) was constructed for SCK1. With degree of 8, BC of 0.26 and CC of 0.22,

Ribosomal protein S10 (RPS10) was identified as a key node of this network. A network containing

164 nodes and 186 edges between them (PPI enrichment P = 1.0e−16; Figure 8) was constructed for SCK2. Sulfatase Modifying Factor 2 (SUMF2) is a key node in this network with degree of 10,

BC of 0.07 and CC of 0.21.

Discussion

Advances in genomics and systems biology can aid in the understanding the genomic background of clinical and subclinical ketosis. The proportions of genetic variance explained by the top-20 sliding windows identified in the GWAS confirmed the polygenic nature of these traits.

The window explaining the largest proportion of the genetic variance for CK1 and CK2 was located in a region on chromosome 6 (1.69% and 2.35%, respectively), with CK2 having four windows on the same chromosome. This chromosome was also present in the top-20 genomic windows for SCK1 and SCK2. Parker Gaddis et al. (2018) reported that the largest proportion of the genetic variance for ketosis was associated with a region on chromosome 6 located at 56.1 Mb.

In addition, Nayeri et al. (2019) reported SNPs on chromosome 6 (rs109930261, rs109793149, rs110899052) to be associated with milk BHB concentrations in dairy cattle. Regions on chromosome 6 identified in this study were consistent with the ones found by Nayeri et al. (2019) for CK2 (56Mb and 78.7-98.6Mb) and for SCK2 (87-93Mb). These findings validate the association of chromosome 6 with clinical and subclinical ketosis previously reported in the literature. Thus, chromosome 6 harbours regions of interest for both clinical and subclinical ketosis.

Additional chromosomes with regions explaining high proportions of genetic variance for

CK1 include chromosomes 2, 5, 11 and 23. Furthermore, of the top-20 windows explaining the highest genetic variance, three windows were present on chromosome 3, which is a region with genes known to affect the concentration of non-esterified fatty acids (NEFA).

Regions on chromosome 20 were found to be associated with clinical and subclinical ketosis for later lactations. For CK2, five windows were associated with regions on chromosome

20, while SCK2 presented three windows on this chromosome. These regions were previously related to pathways associated with ketosis, and genes within these regions were associated with 83

inflammatory response (Nayeri et al., 2019). Nayeri et al. (2019) reported for the first time a region on chromosome 20 (55–63 Mb) for SCK2. For SCK1, the window that explains the highest genetic variance is located on chromosome 2, which has been previously associated with genes related to insulin resistance (Clempson et al., 2012).

Many of the genes located within the regions studied here pertain to at least one common function group and they will be discussed below. For instance, several genes were involved in insulin metabolism, lipid metabolism, inflammatory response, and immune system.

Clinical ketosis in first lactation (CK1)

An important gene associated with CK1 was the acetyl-CoA acetyltransferase 2 (ACAT2), which is an enzyme responsible for fatty acid beta-oxidation, lipid metabolism and synthesis and degradation of ketone bodies (Buhman et al., 2000; Puchalska & Crawford, 2017). The expression of this gene was reported to be lowered in the liver tissues of cows with ketosis, suggesting impaired liver function in utilization of fatty acids (Xu et al., 2008; Xu & Wang, 2008). Thus, reduced concentrations of ACAT2 may be caused by a negative feedback from elevated concentrations of ketone bodies in ketotic cows. The ACAT2 was found to be related with IGF1, providing evidence that ketone bodies and insulin resistance can be related (Farrés et al., 2010).

Indoleamine 2,3-Dioxygenase 1 (IDO1) and Indoleamine 2,3-Dioxygenase 2 (IDO2) are the enzymes responsible for degradation of the essential amino acid tryptophan to kynurenine

(Schröcksnadel et al., 2006). The ratio between kynurenine and tryptophan was shown to be increased in unhealthy cows because of their metabolic imbalance, and this has been suggested as an indicator of immune regulation and IDO activity (Huber et al., 2016; Okamoto et al., 2007). In addition, during early pregnancy in cattle, IDO expression is increased accompanied by a decrease

in the tryptophan/kynurenine ratio, which is considered an important component in the resolution of inflammation (Groebner et al., 2011).

The gene Phospholipase B1 (PLB1) is present on chromosome 11 and it is related to lipid digestion and absorption. This gene encodes the membrane-associated phospholipase B1, which cleaves fatty acid residues from the phospholipids (Okada et al., 2014). The gene GDP-Mannose

4,6-Dehydratase (GMDS) produces GDP-fucose during the de novo pathway of L-fucose synthesis and it was highly expressed in milk of transition cows (Wickramasinghe et al., 2011). C2 Calcium

Dependent Domain Containing 5 (C2CD5) is involved in GLUT4 trafficking in adipocytes and it is required for insulin-stimulated glucose transport (Xie et al., 2011; Zhou et al., 2018). In addition, this gene was shown to be highly expressed in the hypothalamus of mice and its level decreases in obesity and in the presence of saturated fatty acids (Gavini et al., 2020). ST8 Alpha-N-acetyl- neuraminide alpha-2,8-Sialyltransferase 1 (ST8SIA1) is known to regulate the synthesis of GD3, which is the most abundant ganglioside in early lactation bovine and human milk, and it showed a higher expression in transition milk (Rueda, 2007; Wickramasinghe et al., 2011).

Adenylate cyclase 8 (ADCY8) mediates the effects of glucagon-like peptide 1 (GLP-1) and plays an important role in insulin secretion (Delmeire et al., 2003). It is an enzyme required for the activation of glucose-induced signalling pathways in β-cells and for glucose tolerance via regulation of islet insulin secretion. Thus, the reduction of ADCY8 is related to damaged secretory capacity of β-cells in diabetes (Raoux et al., 2015). In addition, this gene is expressed in the brain and controls energy homeostasis and nutrition (Z. Liu et al., 2007).

Clinical ketosis in later lactations (CK2)

The main function of apolipoproteins is the transport of lipids throughout the body and in the regulation of energy metabolism and the genes apolipoprotein A1 (APOA1), apolipoprotein A4

(APOA4) and apolipoprotein A5 (APOA5) were found in regions associated with CK2. An association to ketosis and other metabolic diseases has been established to decreased serum concentrations of APOA1 and other apolipoproteins (İleri-Büyükoğlu et al., 2009; Oikawa &

Katoh, 2002). Although APOA1 is a component of high-density lipoprotein, Kroezen et al. (2018) detected a SNP within the gene APOA1 which is associated with ketosis and established a relationship between this apolipoprotein and lipid homeostasis. In addition, this gene has been associated with cholesterol metabolism, a pathway that should be further studied when considering the pathogenesis of ketosis.

Along with APOA1 and APOA5, Apolipoprotein C3 (APOC3) and fatty acid-binding protein 2 (FABP2) are also involved in the pathway of regulation of lipid metabolism by peroxisome proliferator-activated receptor alpha (PPARalpha). Furthermore, FABP2 is also involved in fatty acid metabolism and it was found in animal intestine and liver tissues (Belaguli et al., 2007; Q. Wang et al., 2005). Also, a polymorphism of the FABP2 gene was associated with insulin resistance (Chiu et al., 2001). The FABP2 interacts with metabolic enzymes impacting fatty acid utilization (Glatz et al., 1997) and its function postpartum is increased in transition dairy cows to overcome NEB (van Dorland et al., 2009).

Solute carrier family 37 member 4 (SLC37A4), also known as glucose-6-phosphate translocase (G6PT), is involved in the transportation of G6P from the cytoplasm into the ER lumen.

It was mapped to chromosome 11 found to be associated with regions for CK2. In the liver and kidney, the activity of this complex is required to maintain blood glucose homeostasis (Bartoloni

& Antonarakis, 2004). The G6PT is abundantly expressed in gluconeogenic organs and when 86

defective it results in inadequate glucose production, leading to an abnormal storage of glycogen

(Chou et al., 2015).

N-deacetylase and N-sulfotransferase 3 (NDST3) gene encodes a monomeric bifunctional enzyme that catalyzes the N-deacetylation and N-sulfation of N-acetylglucosamine residues in heparan sulfate and heparin, which are the initial chemical modifications required for the biosynthesis of the functional oligosaccharide sequences. Oligosaccharides are low molecular weight carbohydrates, intermediate in nature between simple sugars and polysaccharides (Weijers et al., 2008). The NDST3 gene is involved in the Glycosaminoglycan biosynthesis - heparan sulfate

/ heparin pathway. The basic problem in ketosis is a deficiency of glucose (sugar) in the blood and body tissues. NDST3 deficiency is associated with small changes in high density lipoprotein and total cholesterol and altered anxiety related behavior in mice (Pallerla et al., 2008).

The genes pyrroline-5-carboxylate reductase 1 (PYCR1), asparagine synthetase

(glutamine-hydrolyzing) (ASNS), and phenylalanine hydroxylase (PAH) are involved in the biosynthesis of amino acids. The PYCR1 gene is a key gene in proline biosynthesis and catalyzes the NADH-dependent conversion of pyrroline-5-carboxylate (P5C) to L-proline (Zeng et al.,

2017). This gene is involved in intracellular amino acid metabolism and it was found to be upregulated during lactation (Dai et al., 2018). Furthermore, it was shown to have an influence in insulin resistance, by the elevated circulating levels of proline in subjects with insulin resistance

(Tai et al., 2010; Zhuang et al., 2019). The ASNS gene catalyzes the transfer of an amino group to aspartic acid to form asparagine, which can then be converted to oxoaloacetate, a rate limiting compound in gluconeogenesis (Richards & Kilberg, 2006). The PAH gene is mainly expressed in liver and kidney and codes for the phenylalanine hydroxylase enzyme (Lichter-Konecki et al.,

1999). This enzyme is key in the phenylalanine metabolic pathway and it is responsible for the

first step in processing phenylalanine to tyrosine, which are compatible gluconeogenic and ketogenic amino acids (Y. Wang et al., 2016; Woo et al., 1983). The concentration of tyrosine was found to be lower in the blood plasma of dairy cows with clinical ketosis compared to subclinical ketosis (Y. Wang et al., 2016). Previous studies have also described these three genes as upregulated, to some degree, in the lactating bovine mammary gland (Ouattara et al., 2016).

Genes related to the immune system located in the top 20 genomic regions associated with

CK2 included actin, cytoplasmic 2 (ACTG1), ras-related C3 botulinum toxin substrate 3 (RAC3),

C-X-C motif chemokine receptor 5 (CXCR5) and adenylate cyclase 2 (ADCY2). The ACTG1 is a cytoskeletal protein involved cell movement, which has been reported in several human diseases

(Luo et al., 2014). A study evaluating the immune system in neonatal calves showed that ACTG1 was expressed and categorized as antimicrobial (He et al., 2018). The RAC3 is a member of the p160 nuclear receptor coactivator family and it is known as a nuclear receptor coactivator (H. Li et al., 1997) specifically expressed in the nervous system (Malosio et al., 1997). The CXCR5 is known to be expressed on CD4+ follicular T helper cells in lymph nodes (Lipp & Müller, 2003).

The expression of CXCR5 was impacted during parasitic infection in resistant cattle (R. W. Li et al., 2011). The ADCY2 is essential in accelerating phosphor-acidification and the synthesis and breakdown of glycogen (Rexroad et al., 2001; Yuan & López Bernal, 2007). The ADCY2 has been associated with immune response and inflammatory diseases (Killeen et al., 2014; Walker et al.,

2012).

Subclinical ketosis in first lactation (SCK1)

Solute carrier family 26 member 6 (SLC26A6), also known as putative anion transporter-1

(PAT-1), codes for a protein involved in transporting chloride, oxalate, sulfate and bicarbonate. It

is associated with diabetic ketoacidosis which is marked by the triad of the uncontrolled hyperglycemia, acidosis, and ketosis. Also, this gene is related to a phenotype of increased resistance towards insulin (F.-M. Zhang et al., 2019).

Aldehyde dehydrogenase 9 family member A1 (ALDH9A1) gene has high expression in adult liver, skeletal muscle, and kidney (Lin et al., 1996). This superfamily of enzymes (ALDHs) consists of enzymes which detoxify aldehydes produced in various metabolic pathways to the corresponding carboxylic acids (largely distributed in nature and are intermediates in the degradation pathways of amino acids, fats, and carbohydrates) (Singh et al., 2013).

The gene membrane bound O-acyltransferase domain containing 1 (MBOAT1) is part of the membrane-bound O-acetyltransferase superfamily. The encoded transmembrane protein is an enzyme that transfers organic compounds, usually fatty acids, to hydroxyl groups of protein targets in membranes (Hofmann, 2000). The MBOAT members participate in diverse biological processes, including neutral lipid biosynthesis, membrane lipid remodelling, and developmental processes

(Dauwerse et al., 2007).

Genes involved in fatty acid synthesis and degradation were identified in SCK1, named acyl-CoA oxidase 3 (ACOX3) and solute carrier family 25 member 20 (SLC25A20). The ACOX3 is involved in fatty acid oxidation and the degradation of the branched-chain fatty acids. Genes encoding enzymes involved in fatty acid oxidation are relevant because they could reduce the influx of NEFA (Schäff et al., 2013). The SLC25A20 gene encodes the CACT protein. Defective

CACT results in decreased carnitine/acylcarnitine transport and impaired fatty acid β-oxidation.

The clinical effects of such changes may include hypoglycaemia, hyperammonaemia, cardiomyopathy, hepatic failure and encephalopathy (Palmieri, 2014). Wu et al. (2020) indicated

SLC25A20 had higher relative expression level when comparing ketotic cows pre and postpartum.

An aminohydrolase enzyme, glutaminase (GLS), catalyses the conversion of the amino acid glutamine into glutamate and plays a key part in maintaining acid-base homeostasis (Lieu et al., 2020).

Subclinical ketosis in later lactations (SCK2)

The gene phosphorylase kinase catalytic subunit gamma 1 (PHKG1) encodes a phosphorylase kinase catalytic subunit that activates the cascade of glycogen breakdown (Rubin et al., 2012). The deficiency of this gene induces a glycogen breakdown deficiency, which can cause excess glycogen in the liver and muscle (Ma et al., 2014). In addition, PHKG1 was associated with dry matter intake in lactating dairy cows (Hardie et al., 2017).

Ras association domain family member 6 (RASSF6) induces apoptosis and has enhanced ability to kill cells when Ras is activated, and it also controls the degree of inflammatory response

(Allen et al., 2007). Nayeri et al. (2019) identified the RASSF6 gene on chromosome 6 as associated with subclinical ketosis for later lactations in Canadian Holstein dairy cattle, corroborating the results of this study.

The F-box proteins are ubiquitylated and degraded via the proteasome or ubiquitinated by other ubiquitin E3 ligases (Chen et al., 2013; Y. Liu et al., 2015). The F-Box and leucine rich repeat protein 19 (FBXL19) has an anti-inflammatory effect and regulates actin cytoskeleton dynamics for their ubiquitination and degradation by targeting cell membrane receptors and small

GTPases (Dong et al., 2014; Zhao et al., 2012). The F-box and leucine rich repeat protein 7

(FBXL7) play a role in phosphorylation-dependent ubiquitination of proteins, which regulates the immune response and controls cell proliferative activity and viability (Park et al., 2014). Tiezzi et al. (2015) identified the FBXL7 gene in a window explaining one of the highest variances

associated with clinical mastitis in first parity US Holstein dairy cattle and Forde et al. (2012) showed that there was an increase in endometrial gene expression in pregnant heifers. Therefore, these genes might be associated with changes in the endometrium during early pregnancy and inflammation factors.

The C-X-C motif chemokine ligand 8 (CXCL8) is involved in inflammatory response and mediates neutrophil attraction and activation acting in the site of infection (Harada et al., 1994).

Paape et al. (2002) reported that to protect the mammary gland from infection, it is necessary a localize and rapidly move neutrophils to the area. The cytokine CXCL8 has an important role in this process because it activates the recruitment of neutrophils to inflammation sites enhancing phagocytosis and it was shown to have increased production during mastitis (Fonseca et al., 2009;

J.-W. Lee et al., 2006; Ryman et al., 2013). In addition, Nayeri et al. (2019) identified a highly significant SNP (rs109793149) on this gene on BTA6) at around 90.5 Mb for SCK1. The result of this study agrees with previous researches reporting important interactions between metabolic disease and immune response in dairy cow (Esposito et al., 2014; Wathes et al., 2009).

Selenoprotein I (SELENOI) encodes a multi-pass transmembrane protein associated to the production of phosphatidylethanolamine, which is involved in the regulation of lipid metabolism.

A previous study reported that feeding a selenium deficient diet or the targeted removal of selenium causes complete loss of seleno proteins expression, resulting in increased plasma cholesterol concentration along with an increase in APOE levels in mouse (Sengupta et al., 2008).

Earlier studies have demonstrated the role of apolipoprotein E (APOE) in many steps of lipoprotein homeostasis such as production, delivery, and utilization of cholesterol in the body (Huang et al.,

2007). In the plasma, APOE is associated with very-low-density lipoproteins (VLDL) (Huang et

al., 2007); it is also a high affinity ligand for the low-density lipoprotein (LDL) receptor

(Yamamoto & Ryan, 2009).

Molybdenum cofactor synthesis 2 (MOCS2) is involved in the biosynthesis of molybdenum cofactor (MoCo), which participates in the activity of aldehyde oxidase, sulfite oxidase and xanthine dehydrogenase (Schwarz & Mendel, 2006). The MOCS2 catalyzes the conversion of the precursor Z into the organic moiety of molybdenum co-factor and then forms the active site of a terminal catalyser (sulphite oxidase; SUOX) in the oxidative degradation of the amino acids cysteine and methionine (Drögemüller et al., 2010). This gene was upregulated in lactating mammary gland tissue (+3.413, 0.03) (Suchyta et al., 2003).

The gene hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 7

(HSD3B7) encodes an enzyme that acts on the early stages of cholesterol bile acid synthesis. It is involved in the absorption of fat and fat-soluble vitamins since bile acids promote bile flow

(Russell, 2003; Shea et al., 2007). This gene is mainly expressed in liver and in humans, it is associated with a recessive form of neonatal liver failure when a mutation inactive this gene. In mouse, it changes the types of bile acids synthesized by the liver (Cheng et al., 2003; Shea et al.,

2007).

The genes hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit alpha (HADHA) and beta (HADHB) are involved in fatty acid synthesis and degradation. The

HADHA encodes the alpha subunit while the gene HADHB encodes the beta subunit of the mitochondrial trifunctional protein, which catalyzes the last three steps of mitochondrial beta- oxidation of long chain fatty acids (Maeyashiki et al., 2017; Puchalska & Crawford, 2017). The

NEFA can be either esterified or oxidized in mitochondria and peroxisomes in the liver to produce acetyl-CoA and the mobilization of fat from body stores builds the transition of NEFA into the

liver (Drackley et al., 2005). Thus, genes involved in mitochondrial fatty acid oxidation could impact the influx of NEFA into the liver (Schäff et al., 2013). The mitochondrial beta-oxidation of fatty acids is a key source of energy for the liver and other tissues, especially during NEB. In addition, mutations in these genes could impact the mitochondrial tri-functional protein resulting in trifunctional protein deficiency. Costa et al. (2018) reported that a restricted feeding strategy for bulls provoked a downregulated expression of HADHA.

Overlapping regions and genes among the traits

For the first lactation, two genes (SLIT2 and PBX1) were identified in both clinical and subclinical ketosis (CK1 and SCK1), while 42 genes were found to be in common between clinical and subclinical ketosis for later lactations (CK2 and SCK2). The overlap of regions and genes is expected due to the existence of genetic correlation between lactations. The genes found in common between lactations stand out for being identified as having a biological significance or being expressed in target tissues for ketosis, as they are related to insulin resistance and lipid metabolism.

Insulin is a key metabolic hormone (Herdt, 2000) and an indicator of energy status (Blum et al., 1983). Insulin acts in a feedback loop where NEFA and ketone bodies stimulate insulin secretion and insulin then acts to suppress the formation of more ketone bodies (Herdt, 2000).

Early lactating dairy cows typically have reduced blood insulin levels to support lactation. Healthy

NEB-related diseases such as ketosis have been linked to an insulin-resistant condition in near- parturition dairy cows (Gross et al., 2011).

Insulin like growth factor 1 (IGF1) was found to be associated with clinical and subclinical ketosis for later lactations (CK2 and SCK2). The IGF1 concentration has been proposed as an

early biomarker of ketosis (Piechotta et al., 2015). This gene is expressed in many organs of the body and can have an influence on proliferation, differentiation, and metabolic activities. It has been reported that expression of IGF1 has a positive effect on ovarian oestradiol production

(Fenwick et al., 2008; Jorritsma et al., 2003).

For clinical and subclinical ketosis in later lactations, the gene DCXR was identified. This well-conserved gene is a member of the short-chain dehydrogenase/reductase superfamily, which may play a role the urinate cycle of glucose metabolism. DCXR is considered as a moonlighting protein as it is involved in different processes in the organism and is mainly expressed in the liver and kidney (Ebert et al., 2015; S.-K. Lee et al., 2013). Those processes include: the regulation of the osmotic state of the cellular environment (Nakagawa et al., 2002); the detoxification of the cellular environment (Asami et al., 2006); and cell to cell interaction in the skin (Cho-Vega et al.,

2007). This gene is located on chromosome 19 and is involved in fatty acid and carbohydrate metabolism (Ebert et al., 2015; Palombo et al., 2018), specifically in “Arachidonic acid metabolism” and “Metabolism of xenobiotics by cytochrome P450”.

The synthesis and secretion of lipids are essential steps in milk fat formation. This complex biological process is controlled in part by gene expression (Bionaz et al., 2012). Fatty acid synthase

(FASN) is a multifunctional protein which catalyzes long-chain fatty acid synthesis and plays a central role in mammalian de novo lipogenesis (Matsumoto et al., 2012). Lipogenesis is typically decreased around parturition, in part due to low plasma insulin concentrations, and thus there is lower mRNA abundance for FASN in the liver (Claycombe et al., 1998; Graber et al., 2010).

However, X. Sun et al. (2019) showed that the amount of mRNA for FASN was higher in cows with clinical ketosis than in healthy cows. Zhu et al. (2019) showed that cows with clinical ketosis had increased gene expression of FASN than cows with subclinical ketosis; this can be justified by

elevated fatty acid β-oxidation in cows with subclinical ketosis, which provides them with a better way of addressing fatty acid overload without the need to promote lipid synthesis. In addition, the bovine FASN gene is considered a candidate gene for a milk‐fat content QTL (Roy et al., 2006).

LIM homeobox transcription factor 1 alpha (LMX1A) was found associated with CK1, CK2 and SCK1. This transcription factor is involved in the development of dopamine (DA)-producing neurons (Friling et al., 2009). LMX1A was also associated with milk yield, fat yield and protein yield in buffaloes (Du et al., 2019) and it was found in QTL regions associated with antibody response of dairy cows to Ostertagia ostertagi (Twomey et al., 2019).

For CK2 and SCK2, phosphate cytidylyltransferase 2, ethanolamine (PCYT2) gene is the main regulatory enzyme in biosynthesis of phosphatidylethanolamine (the most abundant lipid on the cytoplasmic layer of cellular membranes) from ethanolamine and diacylglycerol by the CDP- ethanolamine Kennedy pathway (Gibellini & Smith, 2010). The major physiological consequences of deletion of one PCYT2 allele include development of symptoms of the metabolic syndrome such as elevated lipogenesis and lipoprotein secretion, liver steatosis, and insulin resistance (Fullerton et al., 2009; Pavlovic & Bakovic, 2013).

Functional annotation

For further functional annotation, enrichment analysis of the list of candidate genes for clinical and subclinical ketosis was performed. The results showed molecular functions and biological processes that are associated with fatty acid metabolism, synthesis and degradation of ketone bodies, lipid metabolism and inflammatory response in dairy cattle. For CK1, 29 GO terms were found, whereas 60 GO terms were associated with CK2. For SCK1 and SCK2, 31 and 62 GO terms were found, respectively.

The GO enrichment analysis performed in this study revealed that different annotated genes are enriched in many GO terms. From these significant GO terms, several terms including

GO:0008152 (metabolic process), GO:0044238 (primary metabolic process), GO:0043170

(macromolecule metabolic process) and GO:0009893 (positive regulation of metabolic process) were related to metabolic processes.

The gene-gene network analysis was very effective in the investigation of shared biological processes among the candidate genes. The most informative networks for this study were the networks grouped around the genes IL10 for CK1, apolipoprotein A1 (APOA1) for CK2, RPS10 for SCK1 and SUMF2 for SCK2. IL10 is an anti-inflammatory cytokine that regulates the immune system by preventing inflammatory pathology (Couper et al., 2008). Conditions such as decreased milk yield and lack of appetite in ketotic cows in early lactation might be related to the presence of a systemic inflammation which are usually associated with greater pro-inflammatory cytokines levels in dairy cows (Kasimanickam et al., 2013; G. Zhang et al., 2016). The failure to successfully overcome the transition period is associated with an inflammatory state in cows that might contribute to development of ketosis and insulin resistance, altering normal liver function (Loor et al., 2007). Heiser et al. (2015) showed that IL10 expression increased during the first week after calving. The IL10 also has a regulatory role on pro-inflammatory cytokines which was demonstrated in mice (Zhong et al., 2006).

Metabolic pathways associated with the genomic regions for clinical and subclinical ketosis were investigated. An important metabolic pathway is Pyruvate metabolism, which is associated with the synthesis of formate. Formate production can be altered by formic acid levels, and as a substitute pyruvic acid can be transformed into acetyl-CoA. This conversion results in increased level of ketone bodies in the blood, as seen in a clinically ketotic dairy cows (Y. Wang

et al., 2016). The ACAT2 gene was present in the pyruvate metabolism pathway and several other significant pathways reported in this study for CK1 and SCK1, highlighting its possible association with clinical and subclinical ketosis for first lactation.

The ACAT2 gene is also associated with the Glyoxylate and dicarboxylate pathway. As a source of carbon, cells use fatty acids by converting isocitrate and acetyl-CoA into succinate and malate during the glyoxylate cycle. Glyoxylate and dicarboxylate was formerly associated with the formation of ketone bodies due to a deficiency in the use of glucose resulting in oxidation and decomposition of fatty acids in mice (Song et al., 2015). This metabolic pathway was one of the pathways that was greatly affected in lactating cows.

Conclusions

This study reported genomic regions associated with both clinical ketosis and subclinical ketosis and indicated potential candidate genes associated with them. The window that explained the highest proportion of variance for CK1 and CK2 was found on chromosome 6. This chromosome was also present in the top-20 genomic windows for SCK1 and SCK2, which is known to harbour important genes related to ketosis. Regions on chromosome 20 were found to be associated with all traits, and genes within these regions were associated with inflammatory response. Highlighted genes were FASN and ACAT2. Enrichment analyses highlighted important

Gene Ontology terms (GO:0008152 (metabolic process), GO:0044238 (primary metabolic process), GO:0043170 (macromolecule metabolic process) and GO:0009893 (positive regulation of metabolic process)), and pathways (Pyruvate metabolism and Glyoxylate and dicarboxylate pathway) that may be involved in fatty acid metabolism, lipid metabolism, and inflammatory response in dairy cattle. The study of the gene-gene network among the candidate genes was very successful in researching mutual biological processes among them, with the most informative

networks being clustered around the genes IL10 for CK1, APOA1 for CK2, RPS10 for SCK1 and

SUMF2 for SCK2. Several genomic regions and SNPs related to susceptibility to ketosis in dairy cattle that were previous described in the literature were confirmed. Such results enhance the genetic understanding of clinical and subclinical ketosis in dairy cows, which could improve selection decisions in the future. However, further studies are necessary to confirm the roles of newly identified SNPs and genes by this study in the susceptibility to ketosis.

Acknowledgments

Canadian Dairy Network (Lactanet) is acknowledged for providing the data used in this study.

The authors thank The Natural Sciences and Engineering Research Council of Canada (NSERC) for financial support thorough a Collaborative Research and Development (CRD) grant in partnership with DairyGen Council of the Canadian Dairy Network. Dr. Schenkel and Dr. Vargas thank Agriculture and Agri-Food Canada, and by additional contributions from Dairy Farmers of

Canada, the Canadian Dairy Network, and the Canadian Dairy Commission under the Agri-

Science Clusters Initiative. As per the research agreement, aside from providing financial support, the funders have no role in the design and conduct of the studies, data collection and analysis or interpretation of the data. Researchers maintain independence in conducting their studies, own their data, and report the outcomes regardless of the results. The decision to publish the findings rests solely with the researchers.

References

Aguilar, I., Misztal, I., Johnson, D. L., Legarra, A., Tsuruta, S., & Lawlor, T. J. (2010). Hot

topic: A unified approach to utilize phenotypic, full pedigree, and genomic information

for genetic evaluation of Holstein final score. Journal of Dairy Science, 93(2), 743–752.

https://doi.org/10.3168/jds.2009-2730

Allen, N. P. C., Donninger, H., Vos, M. D., Eckfeld, K., Hesson, L., Gordon, L., Birrer, M. J.,

Latif, F., & Clark, G. J. (2007). RASSF6 is a novel member of the RASSF family of

tumor suppressors. Oncogene, 26(42), 6203–6211.

https://doi.org/10.1038/sj.onc.1210440

Andersson, L. (1988). Subclinical Ketosis in Dairy Cows. Veterinary Clinics of North America:

Food Animal Practice, 4(2), 233–251. https://doi.org/10.1016/S0749-0720(15)31046-X

Asami, J., Odani, H., Ishii, A., Oide, K., Sudo, T., Nakamura, A., Miyata, N., Otsuka, N., Maeda,

K., & Nakagawa, J. (2006). Suppression of AGE Precursor Formation Following

Unilateral Ureteral Obstruction in Mouse Kidneys by Transgenic Expression of α-

Dicarbonyl/ L -Xylulose Reductase. Bioscience, Biotechnology, and Biochemistry,

70(12), 2899–2905. https://doi.org/10.1271/bbb.60311

Bartoloni, L., & Antonarakis, S. E. (2004). The human sugar-phosphate/phosphate exchanger

family SLC37. Pflügers Archiv, 447(5), 780–783. https://doi.org/10.1007/s00424-003-

1105-0

Belaguli, N. S., Zhang, M., Rigi, M., Aftab, M., & Berger, D. H. (2007). Cooperation between

GATA4 and TGF-beta signaling regulates intestinal epithelial gene expression. American

Journal of Physiology. Gastrointestinal and Liver Physiology, 292(6), G1520-1533.

https://doi.org/10.1152/ajpgi.00236.2006

Bionaz, M., Periasamy, K., Rodriguez-Zas, S. L., Hurley, W. L., & Loor, J. J. (2012). A Novel

Dynamic Impact Approach (DIA) for Functional Analysis of Time-Course Omics

Studies: Validation Using the Bovine Mammary Transcriptome. PLOS ONE, 7(3),

e32455. https://doi.org/10.1371/journal.pone.0032455

Blum, J. W., Kunz, P., Leuenberger, H., Gautschi, K., & Keller, M. (1983). Thyroid hormones,

blood plasma metabolites and haematological parameters in relationship to milk yield in

dairy cows. Animal Science, 36(1), 93–104. https://doi.org/10.1017/S0003356100039982

Browning, B. L., & Browning, S. R. (2009). A Unified Approach to Genotype Imputation and

Haplotype-Phase Inference for Large Data Sets of Trios and Unrelated Individuals.

American Journal of Human Genetics, 84(2), 210–223.

https://doi.org/10.1016/j.ajhg.2009.01.005

Buhman, K. K., Accad, M., Novak, S., Choi, R. S., Wong, J. S., Hamilton, R. L., Turley, S., &

Farese, R. V. (2000). Resistance to diet-induced hypercholesterolemia and gallstone

formation in ACAT2-deficient mice. Nature Medicine, 6(12), 1341–1347.

https://doi.org/10.1038/82153

Chapinal, N., Leblanc, S. J., Carson, M. E., Leslie, K. E., Godden, S., Capel, M., Santos, J. E. P.,

Overton, M. W., & Duffield, T. F. (2012). Herd-level association of serum metabolites in

the transition period with disease, milk production, and early lactation reproductive

performance. Journal of Dairy Science, 95(10), 5676–5682.

https://doi.org/10.3168/jds.2011-5132

Chen, B. B., Coon, T. A., Glasser, J. R., McVerry, B. J., Zhao, J., Zhao, Y., Zou, C., Ellis, B.,

Sciurba, F. C., Zhang, Y., & Mallampalli, R. K. (2013). A combinatorial F box protein

directed pathway controls TRAF adaptor stability to regulate inflammation. Nature

Immunology, 14(5), 470–479. https://doi.org/10.1038/ni.2565

100

Cheng, J. B., Jacquemin, E., Gerhardt, M., Nazer, H., Cresteil, D., Heubi, J. E., Setchell, K. D.

R., & Russell, D. W. (2003). Molecular genetics of 3beta-hydroxy-Delta5-C27-steroid

oxidoreductase deficiency in 16 patients with loss of bile acid synthesis and liver disease.

The Journal of Clinical Endocrinology and Metabolism, 88(4), 1833–1841.

https://doi.org/10.1210/jc.2002-021580

Chiu, K. C., Chuang, L.-M., Chu, A., & Yoon, C. (2001). Fatty acid binding protein 2 and

insulin resistance. European Journal of Clinical Investigation, 31(6), 521–527.

https://doi.org/10.1046/j.1365-2362.2001.00837.x

Chou, J. Y., Jun, H. S., & Mansfield, B. C. (2015). Type I glycogen storage diseases: Disorders

of the glucose-6-phosphatase/glucose-6-phosphate transporter complexes. Journal of

Inherited Metabolic Disease, 38(3), 511–519. https://doi.org/10.1007/s10545-014-9772-x

Cho-Vega, J. H., Tsavachidis, S., Do, K.-A., Nakagawa, J., Medeiros, L. J., & McDonnell, T. J.

(2007). Dicarbonyl/L-xylulose reductase: A potential biomarker identified by laser-

capture microdissection-micro serial analysis of gene expression of human prostate

adenocarcinoma. Cancer Epidemiology, Biomarkers & Prevention: A Publication of the

American Association for Cancer Research, Cosponsored by the American Society of

Preventive Oncology, 16(12), 2615–2622. https://doi.org/10.1158/1055-9965.EPI-07-

0684

Claycombe, K. J., Jones, B. H., Standridge, M. K., Guo, Y., Chun, J. T., Taylor, J. W., &

Moustaïd-Moussa, N. (1998). Insulin increases fatty acid synthase gene transcription in

human adipocytes. American Journal of Physiology-Regulatory, Integrative and

Comparative Physiology, 274(5), R1253–R1259.

https://doi.org/10.1152/ajpregu.1998.274.5.R1253

101

Clempson, A. M., Pollott, G. E., Brickell, J. S., & Wathes, D. C. (2012). Associations between

bovine IGFBP2 polymorphisms with fertility, milk production, and metabolic status in

UK dairy cows. Animal Biotechnology, 23(2), 101–113.

https://doi.org/10.1080/10495398.2011.650775

Costa, A. S. H., Costa, P., Alves, S. P., Alfaia, C. M., Prates, J. A. M., Vleck, V., Cassar-Malek,

I., Hocquette, J.-F., & Bessa, R. J. B. (2018). Does growth path influence beef lipid

deposition and fatty acid composition? PLOS ONE, 13(4), e0193875.

https://doi.org/10.1371/journal.pone.0193875

Couper, K. N., Blount, D. G., & Riley, E. M. (2008). IL-10: The master regulator of immunity to

infection. Journal of Immunology (Baltimore, Md.: 1950), 180(9), 5771–5777.

https://doi.org/10.4049/jimmunol.180.9.5771

Dai, W.-T., Zou, Y.-X., White, R. R., Liu, J.-X., & Liu, H.-Y. (2018). Transcriptomic profiles of

the bovine mammary gland during lactation and the dry period. Functional & Integrative

Genomics, 18(2), 125–140. https://doi.org/10.1007/s10142-017-0580-x

Dauwerse, J. G., de Vries, B. B. A., Wouters, C. H., Bakker, E., Rappold, G., Mortier, G. R.,

Breuning, M. H., & Peters, D. J. M. (2007). A t(4;6)(q12;p23) translocation disrupts a

membrane-associated O -acetyl transferase gene ( MBOAT1) in a patient with a novel

brachydactyly–syndactyly syndrome. European Journal of Human Genetics, 15(7), 743–

751. https://doi.org/10.1038/sj.ejhg.5201833

Delmeire, D., Flamez, D., Hinke, S. A., Cali, J. J., Pipeleers, D., & Schuit, F. (2003). Type VIII

adenylyl cyclase in rat beta cells: Coincidence signal detector/generator for glucose and

GLP-1. Diabetologia, 46(10), 1383–1393. https://doi.org/10.1007/s00125-003-1203-8

102

Dong, S., Zhao, J., Wei, J., Bowser, R. K., Khoo, A., Liu, Z., Luketich, J. D., Pennathur, A., Ma,

H., & Zhao, Y. (2014). F-box protein complex FBXL19 regulates TGFβ1-induced E-

cadherin down-regulation by mediating Rac3 ubiquitination and degradation. Molecular

Cancer, 13, 76. https://doi.org/10.1186/1476-4598-13-76

Drackley, J. K., Dann, H. M., Douglas, N., Guretzky, N. A. J., Litherland, N. B., Underwood, J.

P., & Loor, J. J. (2005). Physiological and pathological adaptations in dairy cows that

may increase susceptibility to periparturient diseases and disorders. Italian Journal of

Animal Science, 4(4), 323–344. https://doi.org/10.4081/ijas.2005.323

Drögemüller, C., Tetens, J., Sigurdsson, S., Gentile, A., Testoni, S., Lindblad-Toh, K., & Leeb,

T. (2010). Identification of the bovine Arachnomelia mutation by massively parallel

sequencing implicates sulfite oxidase (SUOX) in bone development. PLoS Genetics,

6(8). https://doi.org/10.1371/journal.pgen.1001079

Du, C., Deng, T., Zhou, Y., Ye, T., Zhou, Z., Zhang, S., Shao, B., Wei, P., Sun, H., Khan, F. A.,

Yang, L., & Hua, G. (2019). Systematic analyses for candidate genes of milk production

traits in water buffalo (Bubalus Bubalis). Animal Genetics, 50(3), 207–216.

https://doi.org/10.1111/age.12739

Duffield, T.F. (2000). Subclinical Ketosis in Lactating Dairy Cattle. Veterinary Clinics of North

America: Food Animal Practice, 16(2), 231–253. https://doi.org/10.1016/S0749-

0720(15)30103-1

Duffield, T.F., Lissemore, K. D., McBride, B. W., & Leslie, K. E. (2009). Impact of

hyperketonemia in early lactation dairy cows on health and production. Journal of Dairy

Science, 92(2), 571–580. https://doi.org/10.3168/jds.2008-1507

103

Duffield, Todd F, Kelton, D. F., Leslie, K. E., Lissemore, K. D., & Lumsden, J. H. (1997). Use

of test day milk fat and milk protein to detect subclinical ketosis in dairy cattle in

Ontario. 38, 6.

Ebert, B., Kisiela, M., & Maser, E. (2015). Human DCXR - another “moonlighting protein”

involved in sugar metabolism, carbonyl detoxification, cell adhesion and male fertility?

Biological Reviews of the Cambridge Philosophical Society, 90(1), 254–278.

https://doi.org/10.1111/brv.12108

Esposito, G., Irons, P. C., Webb, E. C., & Chapwanya, A. (2014). Interactions between negative

energy balance, metabolic diseases, uterine health and immune response in transition

dairy cows. Animal Reproduction Science, 144(3–4), 60–71.

https://doi.org/10.1016/j.anireprosci.2013.11.007

Farrés, J., Pujol, A., Coma, M., Ruiz, J. L., Naval, J., Mas, J. M., Molins, A., Fondevila, J., &

Aloy, P. (2010). Revealing the molecular relationship between type 2 diabetes and the

metabolic changes induced by a very-low-carbohydrate low-fat ketogenic diet. Nutrition

& Metabolism, 7, 88. https://doi.org/10.1186/1743-7075-7-88

Fenwick, M. A., Llewellyn, S., Fitzpatrick, R., Kenny, D. A., Murphy, J. J., Patton, J., & Wathes,

D. C. (2008). Negative energy balance in dairy cows is associated with specific changes

in IGF-binding protein expression in the oviduct. Reproduction (Cambridge, England),

135(1), 63–75. https://doi.org/10.1530/REP-07-0243

Fonseca, I., Silva, P. V., Lange, C. C., Guimarães, M. F. M., Weller, M. M. D. C. A., Sousa, K.

R. S., Lopes, P. S., Guimarães, J. D., & Guimarães, S. E. F. (2009). Expression profile of

genes associated with mastitis in dairy cattle. Genetics and Molecular Biology, 32(4),

776–781. https://doi.org/10.1590/S1415-47572009005000074

104

Forde, N., Duffy, G. B., McGettigan, P. A., Browne, J. A., Mehta, J. P., Kelly, A. K., Mansouri-

Attia, N., Sandra, O., Loftus, B. J., Crowe, M. A., Fair, T., Roche, J. F., Lonergan, P., &

Evans, A. C. O. (2012). Evidence for an early endometrial response to pregnancy in

cattle: Both dependent upon and independent of interferon tau. Physiological Genomics,

44(16), 799–810. https://doi.org/10.1152/physiolgenomics.00067.2012

Friling, S., Andersson, E., Thompson, L. H., Jönsson, M. E., Hebsgaard, J. B., Nanou, E.,

Alekseenko, Z., Marklund, U., Kjellander, S., Volakakis, N., Hovatta, O., Manira, A. E.,

Björklund, A., Perlmann, T., & Ericson, J. (2009). Efficient production of mesencephalic

dopamine neurons by Lmx1a expression in embryonic stem cells. Proceedings of the

National Academy of Sciences, 106(18), 7613–7618.

https://doi.org/10.1073/pnas.0902396106

Fullerton, M. D., Hakimuddin, F., Bonen, A., & Bakovic, M. (2009). The development of a

metabolic disease phenotype in CTP:phosphoethanolamine cytidylyltransferase-deficient

mice. The Journal of Biological Chemistry, 284(38), 25704–25713.

https://doi.org/10.1074/jbc.M109.023846

Garrick, D. J., Taylor, J. F., & Fernando, R. L. (2009). Deregressing estimated breeding values

and weighting information for genomic regression analyses. Genetics Selection

Evolution, 41(1), 55. https://doi.org/10.1186/1297-9686-41-55

Gavini, C. K., Cook, T. M., Rademacher, D. J., & Mansuy-Aubert, V. (2020). Hypothalamic C2-

domain protein involved in MC4R trafficking and control of energy balance. Metabolism:

Clinical and Experimental, 102, 153990. https://doi.org/10.1016/j.metabol.2019.153990

105

Geishauser, T., Leslie, K., Kelton, D., & Duffield, T. (1998). Evaluation of Five Cowside Tests

for Use with Milk to Detect Subclinical Ketosis in Dairy Cows. Journal of Dairy Science,

81(2), 438–443. https://doi.org/10.3168/jds.S0022-0302(98)75595-X

Gibellini, F., & Smith, T. K. (2010). The Kennedy pathway—De novo synthesis of

phosphatidylethanolamine and phosphatidylcholine. IUBMB Life, 62(6), 414–428.

https://doi.org/10.1002/iub.337

Glatz, J. F. C., van Nieuwenhoven, F. A., Luiken, J. J. F. P., Schaap, F. G., & van der Vusse, G.

J. (1997). Role of membrane-associated and cytoplasmic fatty acid-binding proteins in

cellular fatty acid metabolism. Prostaglandins, Leukotrienes and Essential Fatty Acids,

57(4–5), 373–378. https://doi.org/10.1016/S0952-3278(97)90413-0

Gohary, K., Overton, M. W., Von Massow, M., LeBlanc, S. J., Lissemore, K. D., & Duffield, T.

F. (2016). The cost of a case of subclinical ketosis in Canadian dairy herds. The

Canadian Veterinary Journal, 57(7), 728–732.

Gordon, J. L., LeBlanc, S. J., & Duffield, T. F. (2013). Ketosis Treatment in Lactating Dairy

Cattle. Veterinary Clinics of North America: Food Animal Practice, 29(2), 433–445.

https://doi.org/10.1016/j.cvfa.2013.03.001

Götz, S., García-Gómez, J. M., Terol, J., Williams, T. D., Nagaraj, S. H., Nueda, M. J., Robles,

M., Talón, M., Dopazo, J., & Conesa, A. (2008). High-throughput functional annotation

and data mining with the Blast2GO suite. Nucleic Acids Research, 36(10), 3420–3435.

https://doi.org/10.1093/nar/gkn176

Graber, M., Kohler, S., Kaufmann, T., Doherr, M. G., Bruckmaier, R. M., & van Dorland, H. A.

(2010). A field study on characteristics and diversity of gene expression in the liver of

106

dairy cows during the transition period. Journal of Dairy Science, 93(11), 5200–5215.

https://doi.org/10.3168/jds.2010-3265

Groebner, A. E., Schulke, K., Schefold, J. C., Fusch, G., Sinowatz, F., Reichenbach, H. D., Wolf,

E., Meyer, H. H. D., & Ulbrich, S. E. (2011). Immunological mechanisms to establish

embryo tolerance in early bovine pregnancy. Reproduction, Fertility, and Development,

23(5), 619–632. https://doi.org/10.1071/RD10230

Gross, J., van Dorland, H. A., Schwarz, F. J., & Bruckmaier, R. M. (2011). Endocrine changes

and liver mRNA abundance of somatotropic axis and insulin system constituents during

negative energy balance at different stages of lactation in dairy cows. Journal of Dairy

Science, 94(7), 3484–3494. https://doi.org/10.3168/jds.2011-4251

Harada, A., Sekido, N., Akahoshi, T., Wada, T., Mukaida, N., & Matsushima, K. (1994).

Essential involvement of interleukin-8 (IL-8) in acute inflammation. Journal of

Leukocyte Biology, 56(5), 559–564. https://doi.org/10.1002/jlb.56.5.559

Hardie, L. C., VandeHaar, M. J., Tempelman, R. J., Weigel, K. A., Armentano, L. E., Wiggans,

G. R., Veerkamp, R. F., Haas, Y. de, Coffey, M. P., Connor, E. E., Hanigan, M. D.,

Staples, C., Wang, Z., Dekkers, J. C. M., & Spurlock, D. M. (2017). The genetic and

biological basis of feed efficiency in mid-lactation Holstein dairy cows. Journal of Dairy

Science, 100(11), 9061–9075. https://doi.org/10.3168/jds.2017-12604

He, Z., Fischer, A., Song, Y., Steele, M., & Guan, L. L. (2018). Genome wide transcriptome

analysis provides bases on colonic mucosal immune system development affected by

colostrum feeding strategies in neonatal calves. BMC Genomics, 19(1), 635.

https://doi.org/10.1186/s12864-018-5017-y

107

Heiser, A., McCarthy, A., Wedlock, N., Meier, S., Kay, J., Walker, C., Crookenden, M. A.,

Mitchell, M. D., Morgan, S., Watkins, K., Loor, J. J., & Roche, J. R. (2015). Grazing

dairy cows had decreased interferon-γ, tumor necrosis factor, and interleukin-17, and

increased expression of interleukin-10 during the first week after calving. Journal of

Dairy Science, 98(2), 937–946. https://doi.org/10.3168/jds.2014-8494

Herdt, T. H. (2000). Ruminant Adaptation to Negative Energy Balance. Veterinary Clinics of

North America: Food Animal Practice, 16(2), 215–230. https://doi.org/10.1016/S0749-

0720(15)30102-X

Hofmann, K. (2000). A superfamily of membrane-bound O-acyltransferases with implications

for wnt signaling. Trends in Biochemical Sciences, 25(3), 111–112.

https://doi.org/10.1016/s0968-0004(99)01539-x

Huang, Z. H., Luque, R. M., Kineman, R. D., & Mazzone, T. (2007). Nutritional regulation of

adipose tissue apolipoprotein E expression. American Journal of Physiology.

Endocrinology and Metabolism, 293(1), E203-209.

https://doi.org/10.1152/ajpendo.00118.2007

Huber, K., Dänicke, S., Rehage, J., Sauerwein, H., Otto, W., Rolle-Kampczyk, U., & von

Bergen, M. (2016). Metabotypes with properly functioning mitochondria and anti-

inflammation predict extended productive life span in dairy cows. Scientific Reports,

6(1), 24642. https://doi.org/10.1038/srep24642

İleri-Büyükoğlu, T., Şahinduran, Ş., Sezer, K., & Güldür, T. (2009). Evaluation of changes in

serum lipoprotein and apolipoprotein patterns in cows with ketosis. American Journal of

Veterinary Research, 70(5), 563–570. https://doi.org/10.2460/ajvr.70.5.563

108

Jamrozik, J., Koeck, A., Kistemaker, G. J., & Miglior, F. (2016). Multiple-trait estimates of

genetic parameters for metabolic disease traits, fertility disorders, and their predictors in

Canadian Holsteins. Journal of Dairy Science, 99(3), 1990–1998.

https://doi.org/10.3168/jds.2015-10505

Jorritsma, R., Wensing, T., Kruip, T. A. M., Vos, P. L. A. M., & Noordhuizen, J. P. T. M.

(2003). Metabolic changes in early lactation and impaired reproductive performance in

dairy cows. Veterinary Research, 34(1), 11–26. https://doi.org/10.1051/vetres:2002054

Kanehisa, M., & Goto, S. (2000). KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic

Acids Research, 28(1), 27–30.

Kanehisa, M., Sato, Y., Kawashima, M., Furumichi, M., & Tanabe, M. (2016). KEGG as a

reference resource for gene and protein annotation. Nucleic Acids Research, 44(D1),

D457–D462. https://doi.org/10.1093/nar/gkv1070

Kasimanickam, R. K., Kasimanickam, V. R., Olsen, J. R., Jeffress, E. J., Moore, D. A., &

Kastelic, J. P. (2013). Associations among serum pro- and anti-inflammatory cytokines,

metabolic mediators, body condition, and uterine disease in postpartum dairy cows.

Reproductive Biology and Endocrinology: RB&E, 11, 103. https://doi.org/10.1186/1477-

7827-11-103

Killeen, A. P., Morris, D. G., Kenny, D. A., Mullen, M. P., Diskin, M. G., & Waters, S. M.

(2014). Global gene expression in endometrium of high and low fertility heifers during

the mid-luteal phase of the estrous cycle. BMC Genomics, 15(1).

https://doi.org/10.1186/1471-2164-15-234

109

Kroezen, V., Schenkel, F. S., Miglior, F., Baes, C. F., & Squires, E. J. (2018). Candidate gene

association analyses for ketosis resistance in Holsteins. Journal of Dairy Science, 101(6),

5240–5249. https://doi.org/10.3168/jds.2017-13374

Ku, C. S., Loy, E. Y., Pawitan, Y., & Chia, K. S. (2010). The pursuit of genome-wide association

studies: Where are we now? Journal of Human Genetics, 55, 195–206.

Lee, J.-W., Bannerman, D. D., Paape, M. J., Huang, M.-K., & Zhao, X. (2006). Characterization

of cytokine expression in milk somatic cells during intramammary infections with

Escherichia coli or Staphylococcus aureus by real-time PCR. Veterinary Research, 37(2),

219–229. https://doi.org/10.1051/vetres:2005051

Lee, S., Cho, K., Park, M., Choi, T., Kim, S., & Do, C. (2016). Genetic Parameters of Milk β-

Hydroxybutyric Acid and Acetone and Their Genetic Association with Milk Production

Traits of Holstein Cattle. Asian-Australasian Journal of Animal Sciences, 29(11), 1530–

1540. https://doi.org/10.5713/ajas.16.0310

Lee, S.-K., Son, L. T., Choi, H.-J., & Ahnn, J. (2013). Dicarbonyl/l-xylulose reductase (DCXR):

The multifunctional pentosuria enzyme. The International Journal of Biochemistry &

Cell Biology, 45(11), 2563–2567. https://doi.org/10.1016/j.biocel.2013.08.010

Li, H., Gomes, P. J., & Chen, J. D. (1997). RAC3, a steroid/nuclear receptor-associated

coactivator that is related to SRC-1 and TIF2. Proceedings of the National Academy of

Sciences of the United States of America, 94(16), 8479–8484.

Li, R. W., Rinaldi, M., & Capuco, A. V. (2011). Characterization of the abomasal transcriptome

for mechanisms of resistance to gastrointestinal nematodes in cattle. Veterinary Research,

42(1), 114. https://doi.org/10.1186/1297-9716-42-114

110

Lichter-Konecki, U., Hipke, C. M., & Konecki, D. S. (1999). Human phenylalanine hydroxylase

gene expression in kidney and other nonhepatic tissues. Molecular Genetics and

Metabolism, 67(4), 308–316. https://doi.org/10.1006/mgme.1999.2880

Lieu, E. L., Nguyen, T., Rhyne, S., & Kim, J. (2020). Amino acids in cancer. Experimental &

Molecular Medicine, 52(1), 15–30. https://doi.org/10.1038/s12276-020-0375-3

Lin, S. W., Chen, J. C., Hsu, L. C., Hsieh, C.-L., & Yoshida, A. (1996). Human γ-

Aminobutyraldehyde Dehydrogenase (ALDH9): CDNA Sequence, Genomic

Organization, Polymorphism, Chromosomal Localization, and Tissue Expression.

Genomics, 34(3), 376–380. https://doi.org/10.1006/geno.1996.0300

Lipp, M., & Müller, G. (2003). Shaping up adaptive immunity: The impact of CCR7 and

CXCR5 on lymphocyte trafficking. Verhandlungen Der Deutschen Gesellschaft Fur

Pathologie, 87, 90–101.

Liu, Y., Lear, T., Zhao, Y., Zhao, J., Zou, C., Chen, B. B., & Mallampalli, R. K. (2015). F-box

protein Fbxl18 mediates polyubiquitylation and proteasomal degradation of the pro-

apoptotic SCF subunit Fbxl7. Cell Death & Disease, 6, e1630.

https://doi.org/10.1038/cddis.2014.585

Liu, Z., Yan, S. F., Walker, J. R., Zwingman, T. A., Jiang, T., Li, J., & Zhou, Y. (2007). Study of

gene function based on spatial co-expression in a high-resolution mouse brain atlas. BMC

Systems Biology, 1(1), 19. https://doi.org/10.1186/1752-0509-1-19

Loor, J. J., Everts, R. E., Bionaz, M., Dann, H. M., Morin, D. E., Oliveira, R., Rodriguez-Zas, S.

L., Drackley, J. K., & Lewin, H. A. (2007). Nutrition-induced ketosis alters metabolic

and signaling gene networks in liver of periparturient dairy cows. Physiological

Genomics, 32(1), 105–116. https://doi.org/10.1152/physiolgenomics.00188.2007

111

Luo, Y., Kong, F., Wang, Z., Chen, D., Liu, Q., Wang, T., Xu, R., Wang, X., & Yang, J. Y.

(2014). Loss of ASAP3 destabilizes cytoskeletal protein ACTG1 to suppress cancer cell

migration. Molecular Medicine Reports, 9(2), 387–394.

https://doi.org/10.3892/mmr.2013.1831

Ma, J., Yang, J., Zhou, L., Ren, J., Liu, X., Zhang, H., Yang, B., Zhang, Z., Ma, H., Xie, X.,

Xing, Y., Guo, Y., & Huang, L. (2014). A Splice Mutation in the PHKG1 Gene Causes

High Glycogen Content and Low Meat Quality in Pig Skeletal Muscle. PLOS Genetics,

10(10), e1004710. https://doi.org/10.1371/journal.pgen.1004710

Maeyashiki, C., Oshima, S., Otsubo, K., Kobayashi, M., Nibe, Y., Matsuzawa, Y., Onizawa, M.,

Nemoto, Y., Nagaishi, T., Okamoto, R., Tsuchiya, K., Nakamura, T., & Watanabe, M.

(2017). HADHA, the alpha subunit of the mitochondrial trifunctional protein, is involved

in long-chain fatty acid-induced autophagy in intestinal epithelial cells. Biochemical and

Biophysical Research Communications, 484(3), 636–641.

https://doi.org/10.1016/j.bbrc.2017.01.159

Malosio, M. L., Gilardelli, D., Paris, S., Albertinazzi, C., & de Curtis, I. (1997). Differential

expression of distinct members of Rho family GTP-binding proteins during neuronal

development: Identification of Rac1B, a new neural-specific member of the family. The

Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 17(17),

6717–6728.

Matsumoto, H., Inada, S., Kobayashi, E., Abe, T., Hasebe, H., Sasazaki, S., Oyama, K., &

Mannen, H. (2012). Identification of SNPs in the FASN gene and their effect on fatty

acid milk composition in Holstein cattle. Livestock Science, 144(3), 281–284.

https://doi.org/10.1016/j.livsci.2011.12.003

112

McArt, J. A. A., Nydam, D. V., & Overton, M. W. (2015). Hyperketonemia in early lactation

dairy cattle: A deterministic estimate of component and total cost per case. Journal of

Dairy Science, 98(3), 2043–2054. https://doi.org/10.3168/jds.2014-8740

Misztal, I., Lourenco, D., Aguilar, I., Legarra, A., & Vitezica, Z. (2018). Manual for BLUPF90

family of programs. 142.

Nakagawa, J., Ishikura, S., Asami, J., Isaji, T., Usami, N., Hara, A., Sakurai, T., Tsuritani, K.,

Oda, K., Takahashi, M., Yoshimoto, M., Otsuka, N., & Kitamura, K. (2002). Molecular

characterization of mammalian dicarbonyl/L-xylulose reductase and its localization in

kidney. The Journal of Biological Chemistry, 277(20), 17883–17891.

https://doi.org/10.1074/jbc.M110703200

Nayeri, S., Schenkel, F., Fleming, A., Kroezen, V., Sargolzaei, M., Baes, C., Cánovas, A.,

Squires, J., & Miglior, F. (2019). Genome-wide association analysis for β-

hydroxybutyrate concentration in Milk in Holstein dairy cattle. BMC Genetics, 20(1), 58.

https://doi.org/10.1186/s12863-019-0761-9

Oikawa, S., & Katoh, N. (2002). Decreases in serum apolipoprotein B-100 and A-I

concentrations in cows with milk fever and downer cows. Canadian Journal of

Veterinary Research, 66(1), 31–34.

Okada, Y., Diogo, D., Greenberg, J. D., Mouassess, F., Achkar, W. A. L., Fulton, R. S., Denny,

J. C., Gupta, N., Mirel, D., Gabriel, S., Li, G., Kremer, J. M., Pappas, D. A., Carroll, R.

J., Eyler, A. E., Trynka, G., Stahl, E. A., Cui, J., Saxena, R., … Plenge, R. M. (2014).

Integration of Sequence Data from a Consanguineous Family with Genetic Data from an

Outbred Population Identifies PLB1 as a Candidate Rheumatoid Arthritis Risk Gene.

PLoS ONE, 9(2), e87645. https://doi.org/10.1371/journal.pone.0087645

113

Okamoto, T., Toné, S., Kanouchi, H., Miyawaki, C., Ono, S., & Minatogawa, Y. (2007).

Transcriptional regulation of indoleamine 2,3-dioxygenase (IDO) by tryptophan and its

analogue: Down-regulation of the indoleamine 2,3-dioxygenase (IDO) transcription by

tryptophan and its analogue. Cytotechnology, 54(2), 107–113.

https://doi.org/10.1007/s10616-007-9081-4

Ouattara, B., Bissonnette, N., Duplessis, M., & Girard, C. L. (2016). Supplements of vitamins B9

and B12 affect hepatic and mammary gland gene expression profiles in lactating dairy

cows. BMC Genomics, 17(1), 640. https://doi.org/10.1186/s12864-016-2872-2

Paape, M., Mehrzad, J., Zhao, X., Detilleux, J., & Burvenich, C. (2002). Defense of the bovine

mammary gland by polymorphonuclear neutrophil leukocytes. Journal of Mammary

Gland Biology and Neoplasia, 7(2), 109–121. https://doi.org/10.1023/a:1020343717817

Pallerla, S. R., Lawrence, R., Lewejohann, L., Pan, Y., Fischer, T., Schlomann, U., Zhang, X.,

Esko, J. D., & Grobe, K. (2008). Altered Heparan Sulfate Structure in Mice with Deleted

NDST3 Gene Function. The Journal of Biological Chemistry, 283(24), 16885–16894.

https://doi.org/10.1074/jbc.M709774200

Palmieri, F. (2014). Mitochondrial transporters of the SLC25 family and associated diseases: A

review. Journal of Inherited Metabolic Disease, 37(4), 565–575.

https://doi.org/10.1007/s10545-014-9708-5

Palombo, V., Milanesi, M., Sgorlon, S., Capomaccio, S., Mele, M., Nicolazzi, E., Ajmone-

Marsan, P., Pilla, F., Stefanon, B., & D’Andrea, M. (2018). Genome-wide association

study of milk fatty acid composition in Italian Simmental and Italian Holstein cows using

single nucleotide polymorphism arrays. Journal of Dairy Science, 101(12), 11004–

11019. https://doi.org/10.3168/jds.2018-14413

114

Park, Y., Jin, H., Aki, D., Lee, J., & Liu, Y.-C. (2014). The ubiquitin system in immune

regulation. Advances in Immunology, 124, 17–66. https://doi.org/10.1016/B978-0-12-

800147-9.00002-9

Parker Gaddis, K. L., Megonigal, J. H., Clay, J. S., & Wolfe, C. W. (2018). Genome-wide

association study for ketosis in US Jerseys using producer-recorded data. Journal of

Dairy Science, 101(1), 413–424. https://doi.org/10.3168/jds.2017-13383

Pavlovic, Z., & Bakovic, M. (2013). Regulation of Phosphatidylethanolamine Homeostasis—

The Critical Role of CTP:Phosphoethanolamine Cytidylyltransferase (Pcyt2).

International Journal of Molecular Sciences, 14(2), 2529–2550.

https://doi.org/10.3390/ijms14022529

Piechotta, M., Mysegades, W., Ligges, U., Lilienthal, J., Hoeflich, A., Miyamoto, A., &

Bollwein, H. (2015). Antepartal insulin-like growth factor 1 and insulin-like growth

factor binding protein 2 concentrations are indicative of ketosis in dairy cows. Journal of

Dairy Science, 98(5), 3100–3109. https://doi.org/10.3168/jds.2014-8885

Pryce, J. E., Parker Gaddis, K. L., Koeck, A., Bastin, C., Abdelsayed, M., Gengler, N., Miglior,

F., Heringstad, B., Egger-Danner, C., Stock, K. F., Bradley, A. J., & Cole, J. B. (2016).

Invited review: Opportunities for genetic improvement of metabolic diseases. Journal of

Dairy Science, 99(9), 6855–6873. https://doi.org/10.3168/jds.2016-10854

Puchalska, P., & Crawford, P. A. (2017). Multi-dimensional roles of ketone bodies in fuel

metabolism, signaling, and therapeutics. Cell Metabolism, 25(2), 262–284.

https://doi.org/10.1016/j.cmet.2016.12.022

Raoux, M., Vacher, P., Papin, J., Picard, A., Kostrzewa, E., Devin, A., Gaitan, J., Limon, I., Kas,

M. J., Magnan, C., & Lang, J. (2015). Multilevel control of glucose homeostasis by

115

adenylyl cyclase 8. Diabetologia, 58(4), 749–757. https://doi.org/10.1007/s00125-014-

3445-z

Rexroad, C. E., Bennett, G. L., Stone, R. T., Keele, J. W., Fahrenkrug, S. C., Freking, B. A.,

Kappes, S. M., & Smith, T. P. (2001). Comparative mapping of BTA15 and HSA11

including a region containing a QTL for meat tenderness. Mammalian Genome: Official

Journal of the International Mammalian Genome Society, 12(7), 561–565.

https://doi.org/10.1007/s0033500-20028

Richards, N. G. J., & Kilberg, M. S. (2006). Asparagine synthetase chemotherapy. Annual

Review of Biochemistry, 75, 629–654.

https://doi.org/10.1146/annurev.biochem.75.103004.142520

Roy, R., Ordovas, L., Zaragoza, P., Romero, A., Moreno, C., Altarriba, J., & Rodellar, C. (2006).

Association of polymorphisms in the bovine FASN gene with milk-fat content. Animal

Genetics, 37(3), 215–218. https://doi.org/10.1111/j.1365-2052.2006.01434.x

Rubin, C.-J., Megens, H.-J., Martinez Barrio, A., Maqbool, K., Sayyab, S., Schwochow, D.,

Wang, C., Carlborg, Ö., Jern, P., Jørgensen, C. B., Archibald, A. L., Fredholm, M.,

Groenen, M. A. M., & Andersson, L. (2012). Strong signatures of selection in the

domestic pig genome. Proceedings of the National Academy of Sciences of the United

States of America, 109(48), 19529–19536. https://doi.org/10.1073/pnas.1217149109

Rueda, R. (2007). The role of dietary gangliosides on immunity and the prevention of infection.

The British Journal of Nutrition, 98 Suppl 1, S68-73.

https://doi.org/10.1017/S0007114507832946

116

Russell, D. W. (2003). The Enzymes, Regulation, and Genetics of Bile Acid Synthesis. Annual

Review of Biochemistry, 72(1), 137–174.

https://doi.org/10.1146/annurev.biochem.72.121801.161712

Ryman, V. E., Nickerson, S. C., Kautz, F. M., Hurley, D. J., Ely, L. O., Wang, Y. Q., &

Forsberg, N. E. (2013). Effect of dietary supplementation on the antimicrobial activity of

blood leukocytes isolated from Holstein heifers. Research in Veterinary Science, 95(3),

969–974. https://doi.org/10.1016/j.rvsc.2013.09.009

Sargolzaei, M., Chesnais, J. P., & Schenkel, F. S. (2014). A new approach for efficient genotype

imputation using information from relatives. BMC Genomics, 15(1), 478.

https://doi.org/10.1186/1471-2164-15-478

Schäff, C., Börner, S., Hacke, S., Kautzsch, U., Sauerwein, H., Spachmann, S. K., Schweigel-

Röntgen, M., Hammon, H. M., & Kuhla, B. (2013). Increased muscle fatty acid oxidation

in dairy cows with intensive body fat mobilization during early lactation. Journal of

Dairy Science, 96(10), 6449–6460. https://doi.org/10.3168/jds.2013-6812

Schröcksnadel, K., Wirleitner, B., Winkler, C., & Fuchs, D. (2006). Monitoring tryptophan

metabolism in chronic immune activation. Clinica Chimica Acta; International Journal

of Clinical Chemistry, 364(1–2), 82–90. https://doi.org/10.1016/j.cca.2005.06.013

Schwarz, G., & Mendel, R. R. (2006). Molybdenum Cofactor Biosynthesis and Molybdenum

Enzymes. Annual Review of Plant Biology, 57(1), 623–647.

https://doi.org/10.1146/annurev.arplant.57.032905.105437

Sengupta, A., Carlson, B. A., Hoffmann, V. J., Gladyshev, V. N., & Hatfield, D. L. (2008). Loss

of housekeeping selenoprotein expression in mouse liver modulates lipoprotein

117

metabolism. Biochemical and Biophysical Research Communications, 365(3), 446–452.

https://doi.org/10.1016/j.bbrc.2007.10.189

Shea, H. C., Head, D. D., Setchell, K. D. R., & Russell, D. W. (2007). Analysis of HSD3B7

knockout mice reveals that a 3α-hydroxyl stereochemistry is required for bile acid

function. Proceedings of the National Academy of Sciences, 104(28), 11526–11533.

https://doi.org/10.1073/pnas.0705089104

Singh, S., Brocker, C., Koppaka, V., Chen, Y., Jackson, B. C., Matsumoto, A., Thompson, D. C.,

& Vasiliou, V. (2013). Aldehyde dehydrogenases in cellular responses to

oxidative/electrophilic stress. Free Radical Biology & Medicine, 56, 89–101.

https://doi.org/10.1016/j.freeradbiomed.2012.11.010

Steeneveld, W., Amuta, P., Soest, F. J. S. van, Jorritsma, R., & Hogeveen, H. (2020). Estimating

the combined costs of clinical and subclinical ketosis in dairy cows. PLOS ONE, 15(4),

e0230448. https://doi.org/10.1371/journal.pone.0230448

Suchyta, S. P., Sipkovsky, S., Halgren, R. G., Kruska, R., Elftman, M., Weber-Nielsen, M.,

Vandehaar, M. J., Xiao, L., Tempelman, R. J., & Coussens, P. M. (2003). Bovine

mammary gene expression profiling using a cDNA microarray enhanced for mammary-

specific transcripts. Physiological Genomics, 16(1), 8–18.

https://doi.org/10.1152/physiolgenomics.00028.2003

Sun, L. W., Zhang, H. Y., Wu, L., Shu, S., Xia, C., Xu, C., & Zheng, J. S. (2014). 1H-Nuclear

magnetic resonance-based plasma metabolic profiling of dairy cows with clinical and

subclinical ketosis. Journal of Dairy Science, 97(3), 1552–1562.

https://doi.org/10.3168/jds.2013-6757

118

Sun, X., Wang, Y., Loor, J. J., Bucktrout, R., Shu, X., Jia, H., Dong, J., Zuo, R., Liu, G., Li, X.,

& Li, X. (2019). High expression of cell death-inducing DFFA-like effector a (CIDEA)

promotes milk fat content in dairy cows with clinical ketosis. Journal of Dairy Science,

102(2), 1682–1692. https://doi.org/10.3168/jds.2018-15439

Szklarczyk, D., Gable, A. L., Lyon, D., Junge, A., Wyder, S., Huerta-Cepas, J., Simonovic, M.,

Doncheva, N. T., Morris, J. H., Bork, P., Jensen, L. J., & Mering, C. von. (2019).

STRING v11: Protein–protein association networks with increased coverage, supporting

functional discovery in genome-wide experimental datasets. Nucleic Acids Research,

47(Database issue), D607–D613. https://doi.org/10.1093/nar/gky1131

Tai, E. S., Tan, M. L. S., Stevens, R. D., Low, Y. L., Muehlbauer, M. J., Goh, D. L. M.,

Ilkayeva, O. R., Wenner, B. R., Bain, J. R., Lee, J. J. M., Lim, S. C., Khoo, C. M., Shah,

S. H., & Newgard, C. B. (2010). Insulin resistance is associated with a metabolic profile

of altered protein metabolism in Chinese and Asian-Indian men. Diabetologia, 53(4),

757–767. https://doi.org/10.1007/s00125-009-1637-8

Tiezzi, F., Parker-Gaddis, K. L., Cole, J. B., Clay, J. S., & Maltecca, C. (2015). A genome-wide

association study for clinical mastitis in first parity US Holstein cows using single-step

approach and genomic matrix re-weighting procedure. PloS One, 10(2), e0114919.

https://doi.org/10.1371/journal.pone.0114919

Twomey, A. J., Berry, D. P., Evans, R. D., Doherty, M. L., Graham, D. A., & Purfield, D. C.

(2019). Genome-wide association study of endo-parasite phenotypes using imputed

whole-genome sequence data in dairy and beef cattle. Genetics Selection Evolution,

51(1), 15. https://doi.org/10.1186/s12711-019-0457-7

119

van Dorland, H. A., Richter, S., Morel, I., Doherr, M. G., Castro, N., & Bruckmaier, R. M.

(2009). Variation in hepatic regulation of metabolism during the dry period and in early

lactation in dairy cows. Journal of Dairy Science, 92(5), 1924–1940.

https://doi.org/10.3168/jds.2008-1454

Walker, C. G., Littlejohn, M. D., Mitchell, M. D., Roche, J. R., & Meier, S. (2012). Endometrial

gene expression during early pregnancy differs between fertile and subfertile dairy cow

strains. Physiological Genomics, 44(1), 47–58.

https://doi.org/10.1152/physiolgenomics.00254.2010

Wang, H., Misztal, I., Aguilar, I., Legarra, A., & Muir, W. M. (2012). Genome-wide association

mapping including phenotypes from relatives without genotypes. Genetics Research,

94(2), 73–83. https://doi.org/10.1017/S0016672312000274

Wang, Q., Li, H., Liu, S., Wang, G., & Wang, Y. (2005). Cloning and tissue expression of

chicken heart fatty acid-binding protein and intestine fatty acid-binding protein genes.

Animal Biotechnology, 16(2), 191–201. https://doi.org/10.1080/10495390500276882

Wang, Y., Gao, Y., Xia, C., Zhang, H., Qian, W., & Cao, Y. (2016). Pathway analysis of plasma

different metabolites for dairy cow ketosis. Italian Journal of Animal Science, 15(3),

545–551. https://doi.org/10.1080/1828051X.2016.1180643

Wathes, D. C., Cheng, Z., Chowdhury, W., Fenwick, M. A., Fitzpatrick, R., Morris, D. G.,

Patton, J., & Murphy, J. J. (2009). Negative energy balance alters global gene expression

and immune responses in the uterus of postpartum dairy cows. Physiological Genomics,

39(1), 1–13. https://doi.org/10.1152/physiolgenomics.00064.2009

120

Weijers, C. A. G. M., Franssen, M. C. R., & Visser, G. M. (2008). Glycosyltransferase-catalyzed

synthesis of bioactive oligosaccharides. Biotechnology Advances, 26(5), 436–456.

https://doi.org/10.1016/j.biotechadv.2008.05.001

Wickramasinghe, S., Hua, S., Rincon, G., Islas-Trejo, A., German, J. B., Lebrilla, C. B., &

Medrano, J. F. (2011). Transcriptome Profiling of Bovine Milk Oligosaccharide

Metabolism Genes Using RNA-Sequencing. PLOS ONE, 6(4), e18895.

https://doi.org/10.1371/journal.pone.0018895

Woo, S. L., Lidsky, A. S., Güttler, F., Chandra, T., & Robson, K. J. (1983). Cloned human

phenylalanine hydroxylase gene allows prenatal diagnosis and carrier detection of

classical phenylketonuria. Nature, 306(5939), 151–155. https://doi.org/10.1038/306151a0

Wu, Z.-L., Chen, S.-Y., Qin, C., Jia, X., Deng, F., Wang, J., & Lai, S.-J. (2020). Clinical

Ketosis-Associated Alteration of Gene Expression in Holstein Cows. Genes, 11(2), 219.

https://doi.org/10.3390/genes11020219

Xie, X., Gong, Z., Mansuy-Aubert, V., Zhou, Q. L., Tatulian, S. A., Sehrt, D., Gnad, F., Brill, L.

M., Motamedchaboki, K., Chen, Y., Czech, M. P., Mann, M., Krüger, M., & Jiang, Z. Y.

(2011). C2 domain-containing phosphoprotein CDP138 regulates GLUT4 insertion into

the plasma membrane. Cell Metabolism, 14(3), 378–389.

https://doi.org/10.1016/j.cmet.2011.06.015

Xu, C., & Wang, Z. (2008). Comparative proteomic analysis of livers from ketotic cows.

Veterinary Research Communications, 32(3), 263–273. https://doi.org/10.1007/s11259-

007-9028-4

Xu, C., Wang, Z., Liu, G., Li, X., Xie, G., Xia, C., & Zhang, H. you. (2008). Metabolic

Characteristic of the Liver of Dairy Cows during Ketosis Based on Comparative

121

Proteomics. Asian-Australasian Journal of Animal Sciences, 21(7), 1003–1010.

https://doi.org/10.5713/ajas.2008.70392

Yamamoto, T., & Ryan, R. O. (2009). Domain Swapping Reveals That Low Density Lipoprotein

(LDL) Type A Repeat Order Affects Ligand Binding to the LDL Receptor. The Journal

of Biological Chemistry, 284(20), 13396–13400.

https://doi.org/10.1074/jbc.M900194200

Yuan, W., & López Bernal, A. (2007). Cyclic AMP signalling pathways in the regulation of

uterine relaxation. BMC Pregnancy and Childbirth, 7 Suppl 1, S10.

https://doi.org/10.1186/1471-2393-7-S1-S10

Zeng, T., Zhu, L., Liao, M., Zhuo, W., Yang, S., Wu, W., & Wang, D. (2017). Knockdown of

PYCR1 inhibits cell proliferation and colony formation via cell cycle arrest and apoptosis

in prostate cancer. Medical Oncology (Northwood, London, England), 34(2), 27.

https://doi.org/10.1007/s12032-016-0870-5

Zhang, F.-M., Tian, S.-X., Geng, Y., Wei, C.-L., Li, N., Zhang, X.-G., & Xie, J.-P. (2019). Novel

SLC26A6 gene polymorphism rs184187143 is associated with diabetic ketoacidosis of

gestational diabetes. European Review for Medical and Pharmacological Sciences,

23(17), 7526–7531. https://doi.org/10.26355/eurrev_201909_18868

Zhang, G., Hailemariam, D., Dervishi, E., Goldansaz, S. A., Deng, Q., Dunn, S. M., & Ametaj,

B. N. (2016). Dairy cows affected by ketosis show alterations in innate immunity and

lipid and carbohydrate metabolism during the dry off period and postpartum. Research in

Veterinary Science, 107, 246–256. https://doi.org/10.1016/j.rvsc.2016.06.012

Zhao, J., Wei, J., Mialki, R. K., Mallampalli, D. F., Chen, B. B., Coon, T., Zou, C., Mallampalli,

R. K., & Zhao, Y. (2012). F-box protein FBXL19-mediated ubiquitination and

122

degradation of the receptor for IL-33 limits pulmonary inflammation. Nature

Immunology, 13(7), 651–658. https://doi.org/10.1038/ni.2341

Zhong, J., Deaciuc, I. V., Burikhanov, R., & de Villiers, W. J. S. (2006). Lipopolysaccharide-

induced liver apoptosis is increased in interleukin-10 knockout mice. Biochimica Et

Biophysica Acta, 1762(4), 468–477. https://doi.org/10.1016/j.bbadis.2005.12.012

Zhou, Q. L., Song, Y., Huang, C.-H., Huang, J.-Y., Gong, Z., Liao, Z., Sharma, A. G., Greene,

L., Deng, J. Z., Rigor, M. C., Xie, X., Qi, S., Ayala, J. E., & Jiang, Z. Y. (2018).

Membrane Trafficking Protein CDP138 Regulates Fat Browning and Insulin Sensitivity

through Controlling Catecholamine Release. Molecular and Cellular Biology, 38(8).

https://doi.org/10.1128/MCB.00153-17

Zhu, Y., Liu, G., Du, X., Shi, Z., Jin, M., Sha, X., Li, X., Wang, Z., & Li, X. (2019). Expression

patterns of hepatic genes involved in lipid metabolism in cows with subclinical or clinical

ketosis. Journal of Dairy Science, 102(2), 1725–1735. https://doi.org/10.3168/jds.2018-

14965

Zhuang, J., Song, Y., Ye, Y., He, S., Ma, X., Zhang, M., Ni, J., Wang, J., & Xia, W. (2019).

PYCR1 interference inhibits cell growth and survival via c-Jun N-terminal kinase/insulin

receptor substrate 1 (JNK/IRS1) pathway in hepatocellular cancer. Journal of

Translational Medicine, 17. https://doi.org/10.1186/s12967-019-2091-0

Tables and Figures

Table 1. Top-20 sliding windows explaining the highest proportion of genetic variance for clinical ketosis 1 (CK1).

Chr Position (bp) Var (%) 6 39470632:40182771 1.70 5 87558801:88246615 1.48 2 108277094:109019023 1.27 23 50731493:51335705 1.08 11 71130818:71946048 1.07

123

17 42736277:43493467 1.02 4 78071301:79231632 0.97 12 68852955:69746467 0.88 2 125665173:126315385 0.78 9 96043531:96090673 0.77 23 48617540:49291906 0.69 3 32800931:32888689 0.68 19 49744592:50605756 0.64 27 34061853:35143952 0.62 14 9361706:9949694 0.61 16 4263770:4831223 0.51 20 50585802:51392649 0.49 3 110042206:110866523 0.48 3 3596795:4313521 0.48 7 32543025:33469201 0.47 Chr: chromosome; Position (bp): starting and ending coordinates of the window; Var: % genetic variance explained by the SNPs within the window.

Table 2. Top-20 sliding windows explaining the highest proportion of genetic variance for clinical ketosis 2 (CK2).

Chr Position (bp) Var (%) 6 39825526:40456680 2.35 20 8811738:9562025 1.37 7 38298547:38608111 1.32 3 3639255:4343859 1.31 15 29240225:30156144 1.23 15 26653665:27503740 1.10 6 5063081:6358918 1.04 5 2143145:2940529 0.91 20 65741860:66319602 0.89 9 17170909:19212422 0.77 20 35757652:36278126 0.74 4 14277959:15633760 0.66 6 38929469:39797151 0.57 19 50691143:51263493 0.55 18 59412224:60980891 0.54 20 59086224:59794504 0.54 5 65581391:66435941 0.53 5 66446463:66777438 0.52 6 7070264:7925817 0.52 20 8151139:8771290 0.50 Chr: chromosome; Position (bp): starting and ending coordinates of the window; Var: % genetic variance explained by the SNPs within the window.

Table 3. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis 1 (SCK1).

124

Chr Position (bp) Var (%) 2 77688697:78437805 2.43 23 36786202:37974716 2.22 3 3043960:3781801 1.52 1 100122413:101049645 1.15 2 78507223:79522994 1.11 12 59129931:60463147 1.08 14 1147948:1315998 0.97 6 104960055:116222948 0.91 28 25778236:26552286 0.89 23 43155060:43481899 0.75 5 13068223:14695474 0.70 22 50423075:51293138 0.67 19 47674203:48522690 0.66 13 6382837:7531220 0.63 8 67492932:68136948 0.62 20 56196377:56292004 0.61 10 96631113:97342537 0.60 23 8282432:9205114 0.59 14 26094070:26691608 0.57 18 6592522:7571766 0.57 Chr: chromosome; Position (bp): starting and ending coordinates of the window; Var: % genetic variance explained by the SNPs within the window.

Table 4. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis 2 (SCK2).

Chr Position (bp) Var (%) 20 65741860:66319602 0.95 11 71946048:72911805 0.89 25 30067534:30543167 0.85 19 50660229:51233023 0.82 5 65743164:66456344 0.77 21 14564126:14847447 0.74 9 5795323:6181422 0.73 20 3442790:4360799 0.69 17 18266121:18862410 0.69 14 8996524:9624604 0.63 6 87840175:88893733 0.58 6 5063081:6358918 0.58 2 77688697:78437805 0.58 25 26903532:27969929 0.56 8 29018683:29865913 0.56 19 51263493:52029384 0.55 11 73053087:73269991 0.54 4 77978159:79139483 0.54 20 25229625:26352669 0.52 4 49216223:49813182 0.519

125

Chr: chromosome; Position (bp): starting and ending coordinates of the window; Var: % genetic variance explained by the SNPs within the window.

Figure 1. Genetic variance explained by 20-SNP sliding windows for clinical ketosis 1 (CK1). Chromosome number is shown on the horizontal axis.

126

Figure 2. Genetic variance explained by 20-SNP sliding windows for clinical ketosis 2 (CK2). Chromosome number is shown on the horizontal axis.

Figure 3. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis 1 (SCK1). Chromosome number is shown on the horizontal axis.

127

Figure 4. Genetic variance explained by 20-SNP windows for subclinical ketosis 2 (SCK2). Chromosome number is shown on the horizontal axis.

Table 5. Annotated genes within the top 20 sliding windows associated with subclinical and clinical ketosis obtained by wssGBLUP analyses.

Traits Chr Position (bp) Genes 6 39470632:40182771 SLIT2, bta-mir-218-1 5 87558801:88246615 ETNK1, C2CD5, ST8SIA1, CMAS 2 108277094:109019023 - SERPINB6, SERPINB9, SERPINB1, 23 50731493:51335705 RF00003, WRNIP1, MYLK4, GMDS PLB1, FOSL2, BABAM2, RF02271, 11 71130818:71946048 RBKS, MRPL33 17 42736277:43493467 CTSO, RF00026, TDO2, ASIC5 CK1 4 78071301:79231632 MRPL32, PSMA2, C4H7orf25, GLI3 DCT, RF00001, TGDS, GPR180, 12 68852955:69746467 RF00004, SOX21 IFI6, FGR, AHDC1, WASF2, 2 125665173:126315385 RF00026, GPR3, CD164L2, MAP3K6, SYTL1, TMEM222 ACAT2, TCP1, RF00401, RF00429, 9 96043531:96090673 MRPL18, PNLDC1 23 48617540:49291906 F13A1, NRN1, FARS2 3 32800931:32888689 KCNA2

128

B3GNTL1, TBCD, bta-mir-2344, ZNF750, RF00045, FN3KRP, RAB40B, WDR45B, FOXK2, NARF, 19 49744592:50605756 CYBC1, HEXD, OGFOD3, SECTM1A, bta-mir-2345, SECTM1, CD7, CSNK1D PLEKHA2, HTRA4, TM2D2, ADAM9, 27 34061853:35143952 ADAM32, ADAM2, ADAM3A, ADAM18, RF00026, IDO1, IDO2 14 9361706:9949694 ADCY8 IKBKE, RASSF5,EIF2D, DYRK3, 16 4263770:4831223 MAPKAPK2, IL10, IL19, IL20, IL24, FCMR, PIGR, FCAMR 20 50585802:51392649 CDH12 CLSPN, C3H1orf216, PSMB2, TFAP2E, NCDN, KIAA0319L, ZMYM4, SFPQ, 3 110042206:110866523 ZMYM1, ZMYM6, TMEM35B, DLGAP3, SMIM12 3 3596795:4313521 RXRG, LMX1A, PBX1 7 32543025:33469201 PRR16 SLIT2, bta-mir-218-1, PACRGL, 6 39825526:40456680 KCNIP4 TNPO1, ZNF366, PTCD2, MRPS27, 20 8811738:9562025 MAP1B UNC5A, HK3, UIMC1, ZNF346, 7 38298547:38608111 RF00026, FGFR4 3 3639255:4343859 LMX1A, PBX1 ARCN1, PHLDB1, bta-mir-12011, TREH, DDX6, RF00001, CXCR5, BCL9L, UPK2, FOXR1, CCDC84, RPS25, TRAPPC4, SLC37A4, 15 29240225:30156144 HYOU1, VPS11, HMBS, DPAGT1, C2CD2L, HINFP, ABCG4, NLRX1, PDZD3, CCDC153, CBL, MCAM, RNF26, C1QTNF5, MFRP, USP2, THY1 CK2 RF00285, BUD13, ZPR1, APOA5, 15 26653665:27503740 APOA4, APOC3, APOA1 PDE5A, FABP2, C6H4orf3, USP53, 6 5063081:6358918 RF00001, MYOZ2, SYNPO2 5 2143145:2940529 TRHDE 20 65741860:66319602 ADYC2 MEI4, RF00100, IRAK1BP1, PHIP, 9 17170909:19212422 RF00026, LCA5 20 35757652:36278126 LIFR, EGFLAM RF01887, RF01888, DLX6, RF01778, 4 14277959:15633760 DLX5, SDHAF3, TAC1, ASNS, RF00322, C1GALT1, COL28A1 6 38929469:39797151 SLIT2 CCDC57, FASN, DUS1L, GPS1, 19 50691143:51263493 RFNG, DCXR, RAC3, LRRC45, CENPX, ASPSCR1, bta-mir-2346,

129

NOTUM, MYADML2, PYCR1, MAFG, SIRT7, PCYT2, bta-mir-2347, NPB, ANAPC11, ALYREF, ARHGDIA, P4HB, PP1R27, MCRIP1, GCGR, MRPL12, HGS, ARL16, CCDC137, OXLD1, PDE6G, TSPAN10, NPLOC4, FAAP100, FSCN2, ACTG1, bta-mir-3533 ZNF677, ZNF331, MGC139164, 18 59412224:60980891 bta-mir-371, NLRP12 20 59086224:59794504 RF00026, DNAH5 GNPTAB, DRAM1, WASHC3, 5 65581391:66435941 NUP37, PARPBP , PMCH, IGF1 5 66446463:66777438 RF00026, PAH, ASCL1 6 7070264:7925817 NDST3 RF00026, TMEM174, TMEM171, 20 8151139:8771290 FCHO2, TNPO1 2 77688697:78437805 - 23 36786202:37974716 CDKAL1, E2F3, RF00001, MBOAT1 UCK2, TMCO1, ALDH9A1, MGST3, 3 3043960:3781801 LRRC52, RXRG, LMX1A 1 100122413:101049645 ZBBX, RF00026 2 78507223:79522994 RF00612, GYPC, GLS, STAT1 12 59129931:60463147 RF00026 ZC3H3, MAFA, RHPN1, TOP1MT, 14 1147948:1315998 ZNF696 PSAPL1, bta-mir-2452, AFAP1, ABLIM2, bta-mir-95, SH3TC1, HTRA3, ACOX3, TRMT44, CPZ, HMX1, ADRA2C, RF00001, LRPAP1, DOK7, HGFAC, 6 104960055:116222948 MSANTD1, HTT, bta-mir-2451, bta-mir-2450a, SCK1 bta-mir-2450b, bta-mir-2450d, GRK4, NOP14, MFSD10, ADD1, RF00026, SH3BP2, TNIP2, FAM193A, RNF4 TACR2, TSPAN15, NEUROG3, FAM241B, COL13A1, H2AFY2, 28 25778236:26552286 AIFM2, TYSND1, SAR1A, PPA1, NPFFR1, LRRC20 23 43155060:43481899 GFOD1, TBC1D7, PHACTR1 RF00026, SLC6A15, LRRIQ1, 5 13068223:14695474 TSPAN19 UBA7, INKA1, CDHR4, IP6K1, GMPPB, RNF123, AMIGO3, MST1, APEH, BSN, DAG1, NICN1, AMT, 22 50423075:51293138 TCTA, RHOA, GPX1, RF00026, USP4, C22H3orf62, CCDC36, C22H3orf84, KLHDC8B, CCDC71, LAMB2, USP19, QARS, QRICH1,

130

IMPDH2, NDUFAF3, MIR191, bta- mir-425, DALRD3, WDR6, P4HTM, ARIH2, SLC25A20, PRKAR2A, IP6K2, NCKIPSD, CELSR3, SLC26A6, TMEM89, UQCRC1, COL7A1 TANC2, CYB561, ACE, KCNH6, DCAF7, TACO1, MAP3K3, LIMD2, STRADA, CCDC47, DDX42, FTSJ3, 19 47674203:48522690 PSMC5, SMARCD2, bta-mir-10173, TCAM1, GH1, CD79B, SCN4A, ERN1, RF00289, RF00407, RF00598, TEX2 SPTLC3, ISM1, TASP1, ESF1, 13 6382837:7531220 NDUFAF5, SEL1L2 8 67492932:68136948 - 20 56196377:56292004 MYO10 10 96631113:97342537 FLRT2 HMGA1, NUDT3, RF00322, RPS10, PACSIN1, SPDEF, ILRUN, RF00001, 23 8282432:9205114 RF00156, RF00066, SNRPC, UHRF1BP1, TAF11, ANKS1A, TCP11, SCUBE3 14 26094070:26691608 RAB2A, CHD7 18 6592522:7571766 DYNLRB2, CDYL2 20 65741860:66319602 ADCY2 SUPT7L, GPN1, ZNF512, C11H2orf16, FNDC4, IFT172, KRTCAP3, NRBP1, PPM1G, ZNF513, SNX17, EIF2B4, GTF3C2, MPV17, 11 71946048:72911805 TRIM54, DNAJC5G, SLC30A3, CAD, ATRAID, SLC5A6, TCF23, PRR30, PREB, ABHD1, CGREF1, KHK, EMILIN1, AGBL5, TMEM214, MAPRE3, RF00091, DPYSL5, CENPA, SLC35F6, KCNK3 25 30067534:30543167 - CCDC57, FASN, DUS1L, GPS1, RFNG, DCXR, RAC3, SCK2 LRRC45, CENPX, ASPSCR1, bta-mir-2346, NOTUM, MYADML2, PYCR1, MAFG, SIRT7, 19 50660229:51233023 PCYT2, bta-mir-2347, NPB, ANAPC11, ALYREF, ARHGDIA, P4HB, PPP1R27, MCRIP1, GCGR, MRPL12, HGS, ARL16, CCDC137, OXLD1, PDE6G, TSPAN10, NPLOC4, FAAP100 WASHC3, NUP37, PARPBP, PMCH, 5 65743164:66456344 IGF1, RF00026 21 14564126:14847447 ST8SIA2, SLCO3A1 9 5795323:6181422 -

131

FBXW11, STK10, EFCAB9, UBTD2, 20 3442790:4360799 SH3PXD2B NAA15, NDUFC1, MGARP, ELF2, 17 18266121:18862410 NOCT, RF00026 14 8996524:9624604 KCNQ3, HHLA1, EFR3A SNORD42, COX18, ANKRD17, ALB, 6 87840175:88893733 AFP, AFM, RASSF6, 7SK, CXCL8 PDE5A, FABP2, C6H4orf3, USP53, 6 5063081:6358918 RF00001, MYOZ2, SYNP02 2 77688697:78437805 - ZNF629, BCL7C, CTF1, FBXL19, ORAI3, SETD1A, HSD3B7, STX1B, STX4, ZNF668, ZNF646, PRSS53, VKORC1, BCKDK, KAT8, PRSS8, PRSS36, FUS, PYCARD, TRIM72, ITGAM, ITGAD, COX6A2, ARMC5, TGFB1I1, 25 26903532:27969929 SLC5A2, C25H16orf58, AHSP, SEPT14, ZNF713, MRPS17, NIPSNAP2, PSPH, CCT6A, RF00414, RF00398, SUMF2, PHKG1, CHCHD2, NUPR2, VKORC1L1, GUSB TTC39B, RF00001, FREM1, CER1, 8 29018683:29865913 ZDHHC21, NFIB BAHCC1, RF00100, SLC38A10, NDUFAF8, 19 51263493:52029384 TEPSIN, CEP131, AATK, bta-mir- 338, BAIAP2, CHMP6, RPTOR, RF00003 OTOF, DRC1, SELENOI, 11 73053087:73269991 bta-mir-2284y-1, DGRF3, HADHB, HADHA 4 77978159:79139483 MRPL32, PSMA2, C4H7orf25, GLI3 NDUFS4, FST, MOCS2, ITGA2, 20 25229625:26352669 ITGA1, PELO NRCAM, PNPLA8, NME8, SFRP4, 4 49216223:49813182 EPDR1 Chr: chromosome; Position (bp): starting and ending coordinates of the window; Genes: ENSEMBL symbol of annotated genes using the Bos taurus ARS-UCD1.2 assembly.

Table 6. Pathways and their genes obtained for clinical ketosis 1 (CK1). Trait Pathways Gene symbol CK1 Purine metabolism ADCY8 Tryptophan metabolism TDO2 ACAT2 IDO1 IDO2 Aminoacyl-tRNA biosynthesis FARS2

132

Melanogenesis ADCY8 DCT Fatty acid elongation Valine, leucine and isoleucine degradation Biosynthesis of unsaturated fatty acids Pyruvate metabolism Terpenoid backbone biosynthesis Propanoate metabolism ACAT2 Fatty acid degradation Lysine degradation Synthesis and degradation of ketone bodies Glyoxylate and dicarboxylate metabolism Butanoate metabolism Alpha-Linolenic acid metabolism PLB1 T cell receptor signaling pathway IL10 Amino sugar and nucleotide sugar metabolism GMDS Fructose and mannose metabolism Glycerophospholipid metabolism PLB1 ETNK1 Pentose phosphate pathway RBKS Tyrosine metabolism DCT

Table 7. Pathways and their genes obtained for clinical ketosis 2 (CK2). Trait Pathways Gene symbol CK2 Purine metabolism PDE5A Metabolism of xenobiotics by cytochrome P450 DCXR Arachidonic acid metabolism Arginine and proline metabolism PYCR1 Glycosaminoglycan biosynthesis - heparan sulfate NDST3 / heparin Nicotinate and nicotinamide metabolism SIRT7 Starch and sucrose metabolism TREH Alanine, aspartate and glutamate metabolism ASNS Folate biosynthesis PAH

Table 8. Pathways and their genes obtained for subclinical ketosis 1 (SCK1). Trait Pathways Gene symbol SCK1 Purine metabolism IMPDH2

133

Drug metabolism - other enzymes MGST3 UCK2 IMPDH2 Glycolysis/Gluconeogenesis ALDH9A1 Arginine and proline metabolism ALDH9A1 PYCR1 Alanine, aspartate and glutamate metabolism GLS ASNS GPX1 Arachidonic acid metabolism DCXR Amino sugar and nucleotide sugar metabolism GMPPB Fructose and mannose metabolism Fatty acid degradation ACAT2 Glycerophospholipid metabolism MBOAT1 Metabolism of xenobiotics by cytochrome P450 DCXR Folate biosynthesis PAH Starch and sucrose metabolism TREH Glycine, serine and threonine metabolism PSPH Pyruvate metabolism ACAT2

Table 9. Pathways and their genes obtained for subclinical ketosis 2 (SCK2). Trait Pathways Gene symbol SCK2 Lysine degradation SETD1A Nicotinate and nicotinamide metabolism SIRT7 Purine metabolism PDE5A ADYC2 Fatty acid elongation HADHB Valine, leucine and isoleucine degradation Fatty acid degradation HADHA HADHB Arachidonic metabolism DCXR Metabolism of xenobiotics by cytochrome P450 DCXR Pentose and glucuronate interconversions GUSB Oxidative phosphorylation COX6A2 NDUFC1 DUFS4 Ether lipid metabolism SELENOI Phosphonate and phosphinate metabolism Glycerophospholipid metabolism SELENOI PCYT2 Glycosaminoglycan degradation Drug metabolism - other enzymes GUSB Porphyrin and chlorophyll metabolism

134

Other types of O-glycan biosynthesis RFNG Butanoate metabolism HADHA Folate biosynthesis MOCS2 Glycine, serine and threonine metabolism PSPH Fructose and mannose metabolism KHK Arginine and proline metabolism PYCR1

Figure 4. Gene network analysis for clinical ketosis 1 (CK1).

135

Figure 5. Gene network analysis for clinical ketosis 2 (CK2).

136

Figure 6. Gene network analysis for subclinical ketosis 1 (SCK1).

137

Figure 7. Gene network analysis for subclinical ketosis 2 (SCK2).

Supplementary material Supplementary table 1. Enriched molecular function, cellular component and biological process with Fisher's test for clinical ketosis 1 (CK1).

GO Terms Enriched process Description FDR P-value

138

Oxidoreductase activity, acting on single donors with incorporation of molecular GO:0016702 Molecular Function 4.9E-2 1.3E-4 oxygen, incorporation of two atoms of oxygen GO:0004833 Molecular Function Tryptophan 2,3-dioxygenase activity 1.7E-4 4.2E-8 GO:0033754 Molecular Function Idoleamine 2,3-dioxygenase activity 1.1E-2 1.2E-5 DNA-binding transcription factor activity, GO:0000981 Molecular Function 3.5E-2 8.5E-5 RNA polymerase II-specific GO:0043167 Molecular Function Ion binding 4.3E-2 1.1E-4 GO:0005623 Cellular Component Cell 2.3E-2 4.8E-5 GO:0044464 Cellular Component Cell part 2.1E-2 3.8E-5 GO:0005622 Cellular Component Intracellular 3.0E-2 6.7E-5 GO:0044424 Cellular Component Intracellular part 2.2E-2 4.4E-5 GO:0043226 Cellular Component Organelle 5.0E-3 3.6E-6 GO:0043227 Cellular Component Membrane-bounded organelle 7.1E-3 6.1E-3 GO:0043229 Cellular Component Intracellular organelle 9.3E-3 9.4E-6 GO:0043231 Cellular Component Intracellular Membrane-bounded organelle 3.2E-2 7.4E-5 GO:0005634 Cellular Component Nucleus 8.0E-3 7.3E-6 GO:0044249 Biological Process Cellular biosynthetic process 2.1E-2 3.9E-5 GO:1901576 Biological Process Organic substance biosynthetic process 1.5E-2 2.0E-5 GO:0044238 Biological Process Primary metabolic process 6.7E-5 6.5E-9 Cellular nitrogen compound biosynthetic GO:0044271 Biological Process 4.6E-2 1.2E-4 process GO:0043170 Biological Process Macromolecule metabolic process 1.1E-2 1.4E-5 GO:0031323 Biological Process Regulation of cellular metabolic process 2.2E-2 4.4E-5 Regulation of nitrogen compound GO:0051171 Biological Process 2.1E-2 3.7E-5 metabolic process GO:0009059 Biological Process Macromolecule biosynthetic process 3.5E-2 8.3E-5 GO:0080090 Biological Process Regulation of primary metabolic process 2.7E-2 5.95E-5 GO:0048522 Biological Process Positive regulation of cellular process 2.3E-2 4.75E-5 GO:0009893 Biological Process Positive regulation of metabolic process 9.5E-3 1.0E-5 GO:0044260 Biological Process Cellular macromolecule metabolic process 8.4E-3 8.1E-6 Regulation of macromolecule metabolic GO:0060255 Biological Process 2.1E-2 3.9E-5 process Positive regulation of macromolecule GO:0010604 Biological Process 2.0E-2 3.2E-5 metabolic process GO:0021631 Biological Process Optic nerve morphogenesis 3.2E-2 7.3E-5 FDR: False discovery rate; P-value<0.05.

Supplementary table 2. Enriched molecular function, cellular component and biological process with Fisher's test for clinical ketosis 2 (CK2).

GO Terms Enriched process Description FDR P-value GO:0003824 Molecular Function Catalytic activity 1.6E-5 1.08E-8 GO:0005488 Molecular Function Binding 5.6E-9 1.1E-12 GO:0005515 Molecular Function Protein binding 7.1E-5 5.5E-8 GO:0043167 Molecular Function Ion binding 4.5E-4 5.2E-7 GO:0008241 Molecular Function Peptidyl-dipeptidase activity 3.4E-2 8.4E-5 GO:0003953 Molecular Function NAD+ nucleosidase activity 3.4E-2 8.4E-5 GO:0005623 Cellular Component Cell 2.2E-10 2.1E-14

139

GO:0044464 Cellular Component Cell part 2.2E-10 1.2E-14 GO:0005622 Cellular Component Intracellular 9.7E-8 3.3E-11 GO:0044424 Cellular Component Intracellular part 4.9E-8 1.4E-11 GO:0005737 Cellular Component Cytoplasm 6.0E-5 4.1E-8 GO:0044444 Cellular Component Cytoplasm part 4.5E-4 5.2E-7 GO:0044422 Cellular Component Organelle part 1.3E-2 2.5E-5 GO:0044446 Cellular Component Intracellular organelle part 1.7E-2 3.4E-5 GO:0043226 Cellular Component Organelle 4.9E-8 1.3E-11 GO:0043227 Cellular Component Membrane-bounded 2.4E-7 9.2E-11 organelle GO:0043229 Cellular Component Intracellular organelle 4.8E-7 2.1E-10 GO:0043231 Cellular Component Intracellular membrane- 2.6E-6 1.5E-9 bounded organelle GO:0005634 Cellular Component Nucleus 2.9E-4 2.8E-7 GO:0008152 Biological Process Metabolic process 1.8E-6 9.6E-10 GO:0071704 Biological Process Organic substance metabolic 8.0E-5 7.0E-8 process GO:006807 Biological Process Nitrogen compound 8.0E-5 6.9E-8 metabolic process GO:0044281 Biological Process Small molecule metabolic 2.6E-2 6.0E-5 process GO:0044238 Biological Process Primary metabolic process 7.1E-5 5.4E-8 GO:0019222 Biological Process Regulation of metabolic 1.2E-2 2.3E-5 process GO:0044237 Biological Process Cellular metabolic process 1.6E-6 7.7E-10 GO:0043170 Biological Process Macromolecule metabolic 1.2E-2 2.4E-5 process GO:1901564 Biological Process Organonitrogen compound 2.6E-3 4.2E-6 metabolic process GO:0051171 Biological Process Regulation of nitrogen 4.1E-2 1.2E-4 compound metabolic process GO:0080090 Biological Process Regulation of primary 3.5E-2 9.2E-5 metabolic process GO:0031323 Biological Process Regulation of cellular 2.0E-2 4.3E-5 metabolic process GO:0044260 Biological Process Cellular macromolecule 4.1E-2 1.2E-4 metabolic process GO:0060255 Biological Process Regulation of macromolecule 3.4E-2 8.2E-5 metabolic process GO:0065007 Biological Process Biological regulation 1.1E-4 1.0E-7 GO:0009987 Biological Process Cellular process 3.1E-9 4.4E-13 GO:0050789 Biological Process Regulation of biological 3.1E-4 3.1E-7 process GO:0071840 Biological Process Cellular component 3.3E-4 3.5E-7 organization or biogenesis GO:0048519 Biological Process Negative regulation of 3.7E-2 1.0E-4 biological process GO:0050794 Biological Process Regulation of cellular process 9.9E-4 1.3E-6 GO:0016043 Biological Process Cellular component 1.5E-3 2.0E-6 organization GO:0044085 Biological Process Cellular component 1.2E-2 2.3E-5 biogenesis

140

GO:0048518 Biological Process Positive regulation of 1.8E-3 2.7E-6 biological process GO:0048523 Biological Process Negative regulation of 2.3E-2 5.2E-5 cellular process GO:0048522 Biological Process Positive regulation of cellular 1.7E-3 2.5E-6 process GO:0051128 Biological Process Regulation of cellular 1.9E-3 3.0E-6 component organization GO:0022607 Biological Process Cellular component assembly 3.6E-2 9.4E-5 GO:0044087 Biological Process Regulation of cellular 1.8E-2 3.7E-5 component biogenesis GO:0051091 Biological Process Positive regulation of 3.5E-2 9.1E-5 developmental process GO:0048856 Biological Process Anatomical structure 3.7E-2 1.0E-4 development GO:0050793 Biological Process Regulation of developmental 7.4E-4 9.0E-7 process GO:0051239 Biological Process Regulation of multicellular 4.3E-3 7.3E-6 organismal process GO:007275 Biological Process Multicellular organism 3.6E-2 9.6E-5 development GO:0050886 Biological Process Endocrine process 9.1E-3 1.6E-5 GO:2000026 Biological Process Regulation of multicellular 8.1E-4 1.0E-6 organism development GO:0048731 Biological Process System development 3.5E-2 8.9E-5 GO:0060429 Biological Process Epithelium development 2.3E-2 5.3E-5 GO:0045777 Biological Process Positive regulation of blood 1.6E-3 2.2E-6 pressure GO:0003044 Biological Process Regulation of systemic 2.3E-2 4.9E-5 arterial blood pressure mediated by a chemical signal GO:0003084 Biological Process Positive regulation of 1.9E-3 2.9E-6 systemic arterial blood GO:0001990 Biological Process Regulation of systemic 1.2E-2 2.2E-5 arterial blood pressure by hormone FDR: False discovery rate; P-value<0.05.

Supplementary table 3. Enriched molecular function, cellular component and biological process with Fisher's test for subclinical ketosis 1 (SCK1).

GO Terms Enriched process Description FDR P-value UDP-N-acetylglucosamine-lypososomal- GO:0003976 Molecular Function enzyme N- 2.0E-2 4.6E-6 acetylglucosaminephosphotransferase activity GO:0005488 Molecular Function Binding 4.4E-4 2.1E-8 GO:0097159 Molecular Function Organic cyclic compound binding 3.9E-2 4.9E-5 GO:0036094 Molecular Function Small molecule binding 2.4E-2 1.8E-5 GO:1901265 Molecular Function Nudeoside phosphate binding 4.2E-2 5.7E-5 GO:0000166 Molecular Function Nucleotide binding 4.2E-2 5.7E-5 GO:0043167 Molecular Function Ion binding 2.7E-2 2.3E-5

141

GO:0043168 Molecular Function Anion binding 2.4E-2 1.3E-5 GO:0097367 Molecular Function Carbohydrate derivative binding 2.4E-2 1.5E-5 GO:0005515 Molecular Function Protein binding 2.4E-2 1.9E-5 GO:0043226 Cellular Component Organelle 2.4E-2 1.8E-5 GO:0032991 Cellular Component Protein-containing complex 3.9E-2 4.7E-5 GO:0005622 Cellular Component Intracellular 2.1E-2 7.0E-6 GO:0044424 Cellular Component Intracellular part 2.0E-2 4.9E-6 GO:0005737 Cellular Component Cytoplasm 2.4E-2 1.5E-5 GO:0043229 Cellular Component Intracellular organelle 2.1E-2 8.9E-6 GO:0032501 Biological Process Multicellular organismal process 4.8E-2 6.7 E-5 GO:0032502 Biological Process Developmental process 2.8E-2 2.7 E-5 GO:0009987 Biological Process Cellular process 1.5 E-2 2.2 E-6 GO:0048856 Biological Process Anatomical structure development 2.1 E-2 9.3 E-6 GO:0048869 Biological Process Cellular developmental process 2.4 E-2 1.6 E-5 GO:0016043 Biological Process Cellular component organization 3.9 E-2 4.9 E-5 GO:0071704 Biological Process Organic substance metabolic process 4.9 E-2 7.5 E-5 GO:0044238 Biological Process Primary metabolic process 2.8 E-2 2.7 E-5 GO:0001834 Biological Process Trophectodermal cell proliferation 2.4 E-2 1.4 E-5 GO:0048731 Biological Process System development 4.8E-2 6.9 E-5 GO:0007399 Biological Process Nervous system development 3.6 E-2 3.8 E-5 GO:0030154 Biological Process Cell differentiation 2.1 E-2 7.0 E-6 GO:0048468 Biological Process Cell development 3.0E-3 2.9E-7 GO:0030182 Biological Process Neuron differentiation 3.9E-2 4.8E-5 GO:0001754 Biological Process Eye photoreceptor cell differentiation 4.8E-2 7.2E-5 FDR: False discovery rate; P-value<0.05.

Supplementary table 4. Enriched molecular function, cellular component and biological process with Fisher's test for subclinical ketosis 2 (SCK2).

GO Terms Enriched process Description FDR P-value GO:0003824 Molecular Function Catalytic activity 1.4E-3 1.5E-6 GO:0005488 Molecular Function Binding 2.2E-4 2.0E-7 GO:0016491 Molecular Function Oxidoreductase activity 4.3E-2 1.2E-4 GO:0016614 Molecular Function Oxidoreductase activity, acting on CH-OH 4.6E-3 6.7E-6 group of donors GO:0098634 Molecular Function Cell-matrix adhesion mediator activity 4.3E-3 6.0E-6 GO:0016900 Molecular Function Oxidoreductase activity, acting on CH-OH 1.6E-2 3.1E-5 group of donors, disulfide as acceptor GO:0050038 Molecular Function L-xylulose reductase (NADP+) acitivity 1.6E-2 3.1E-5 GO:0047057 Molecular Function Vitamin-K-epoxide reductase (warfarin- 1.6E-2 3.1E-5 sensitive) activity GO:0016509 Molecular Function Long-chain-3-hydroxyacyl-CoA 1.6E-2 3.1E-5 dehydrogenase activity GO:0032991 Cellular Component Protein-containing complex 1.7E-3 1.9E-6 GO:0098636 Cellular Component Protein complex involved in cell adhesion 9.2E-4 9.3E-7 GO:0008305 Cellular Component Integrin complex 6.1E-4 5.6E-7 GO:0005623 Cellular Component Cell 1.3E-7 2.4E-11 GO:0044464 Cellular Component Cell part 2.0E-7 4.9E-11 GO:0005622 Cellular Component Intracellular 2.0E-7 5.8E-11

142

GO:0044424 Cellular Component Intracellular part 1.3E-7 2.5E-11 GO:0043226 Cellular Component Organelle 1.8E-6 9.6E-10 GO:0005737 Cellular Component Cytoplasm 4.2E-8 2.0E-12 GO:0044444 Cellular Component Cytoplasmic part 1.1E-6 5.0E-10 GO:0044422 Cellular Component Organelle part 1.2E-5 8.7E-9 GO:0031090 Cellular Component Organelle membrane 8.7E-4 8.4E-7 GO:0043227 Cellular Component Membrane-bounded organelle 5.6E-6 3.5E-9 GO:0044446 Cellular Component Intracellular organelle part 1.5E-5 1.1E-8 GO:0043229 Cellular Component Intracellular organelle 1.1E-6 5.5E-10 GO:0043231 Cellular Component Intracellular membrane-bounded organelle 1.1E-4 8.9E-8 GO:0043228 Cellular Component Non-membrane-bounded organelle 4.3E-2 1.3E-4 GO:0043232 Intracellular non-membrane-bounded Cellular Component 4.3E-2 1.3E-4 organelle GO:0005739 Cellular Component Mitochondrion 4.3E-2 1.3E-4 GO:0050896 Biological Process Response to stimulus 2.7E-2 6.4E-5 GO:0048585 Biological Process Negative regulation of response to stimulus 7.8E-3 1.2E-5 GO:0023057 Biological Process Negative regulation of signaling 1.6E-2 2.8E-5 GO:0009968 Biological Process Negative regulation of signal transduction 8.2E-3 1.3E-5 GO:0065007 Biological Process Biological regulation 2.0E-3 2.6E-6 GO:0009987 Biological Process Cellular process 9.4E-8 9.1E-12 GO:0050789 Biological Process Regulation of biological process 2.1E-2 4.7E-5 GO:0050794 Biological Process Regulation of cellular process 4.3E-2 1.3E-4 GO:0010648 Biological Process Negative regulation of cell communication 1.6E-2 2.8E-5 GO:0048518 Biological Process Positive regulation of biological process 8.7E-3 1.4E-5 GO:0048522 Biological Process Positive regulation of cellular process 2.5E-2 5.8E-5 GO:0008152 Biological Process Metabolic process 8.5E-7 2.9E-10 GO:0071704 Biological Process Organic substance metabolic process 9.6E-7 3.7E-10 GO:006807 Biological Process Nitrogen compound metabolic process 2.0E-2 4.4E-5 GO:0044237 Biological Process Cellular metabolic process 5.6E-6 3.5E-9 GO:0044238 Biological Process Primary metabolic process 7.2E-6 4.9E-9 GO:0055114 Biological Process Oxidation-reduction process 1.2E-2 2.1E-5 GO:1901575 Biological Process Organic substance catabolic process 4.3E-2 1.3E-4 GO:0043170 Biological Process Macromolecule metabolic process 1.9E-3 2.3E-6 GO:0044260 Biological Process Cellular macromolecule metabolic process 1.8E-3 2.0E-6 GO:0043412 Biological Process Macromolecule modification 1.9E-2 4.1E-5 GO:0019538 Biological Process Protein metabolic process 4.1E-2 1.1E-4 GO:1904564 Biological Process Organonitrogen compound metabolic process 2.3E-2 5.1E-5 GO:0044267 Biological Process Cellular protein metabolic process 1.0E-2 1.6E-5 GO:0036211 Biological Process Protein modification process 1.6E-2 3.3E-5 GO:0006464 Biological Process Cellular protein modification process 1.6E-2 3.3E-5 GO:0018214 Biological Process Protein carboxylation 3.5E-2 9.4E-5 GO:0017187 Biological Process Peptidyl-glutamic acid carboxylation 3.5E-2 9.4E-5 GO:0044281 Biological Process Small molecule metabolic process 3.1E-3 4.0E-6 GO:0005997 Biological Process Xylulose metabolic process 3.5E-2 9.4E-5 GO:0051179 Biological Process Localization 3.3E-3 4.4E-6 GO:0051234 Biological Process Establishment of localization 2.8E-2 6.8E-5 GO:0006810 Biological Process Transport 3.0E-2 7.4E-5 GO:0046337 Biological Process Phosphatidylethanolamine metabolic process 2.6E-2 6.0E-5 FDR: False discovery rate; P-value<0.05.

143

144

CHAPTER FOUR: Genome-wide association analysis for subclinical ketosis in two periods in early lactation in Holstein dairy cattle

R.A.N. Soares1*, G. Vargas1, T. Duffield2, F. Schenkel1#, and E. J. Squires1#

(Prepared for submission to Journal of Dairy Science)

1 Department of Animal Biosciences, Center for Genetic Improvement of Livestock, University of

Guelph, Guelph, Ontario, Canada, N1G 2W1

2 Department of Population Medicine, Ontario Veterinary College, University of Guelph, Guelph,

Ontario, Canada, N1G 2W1

# These authors had equal contribution as senior authors

* Corresponding author: Riani Ananda Nunes Soares

50 Stone Rd E Guelph, ON N1G 2W1, c/o Department of Animal Biosciences

519-993-8724 [email protected]

Abstract

One way of identifying subclinical ketosis, a metabolic disease in dairy cows, is by measuring the amount of ketone bodies, i.e. β-hydroxybutyric acid (BHB), since symptoms are not present. The aim of this study was to estimate genetic parameters and variance components for subclinical ketosis (BHB concentration) and to identify genomic regions associated with subclinical ketosis in North American Holstein cattle. Since the first two weeks after the start of lactation have been shown to be the period of highest risk for the development of subclinical ketosis, the genome-wide association study (GWAS) was performed for subclinical ketosis in two periods, from 5 to 21 days in milk in first (SCK1.1) and later (SCK2.1) and from 22 to 45 days in

145

milk in first (SCK1.2) and later (SCK2.2) lactations, to investigate genetic differences among these periods. Estimated breeding values were deregressed (dEBV) and taken as pseudo phenotypes in the GWAS. After the GWAS, the top-20 genomic regions explaining the largest proportion of genetic variance were investigated for putative genes associated with subclinical ketosis through functional analyses. Several potential genomic regions were identified on chromosomes 11, 16 and 17 for SCK1.1, 9, 16 and 26 for SCK1.2, 1, 9 and 24 for SCK2.1, 1, 17 and 20 for SCK2.2.

The genes highlighted overall were ACSL3, which transcriptionally regulates lipogenic genes, and

ADCY8, which plays an important role in insulin secretion. Zinc Finger Proteins were heavily present in the window explaining the highest genetic variation for SCK1.1, which indicates that regulation of gene expression plays an important role in the first period of early lactation.

Additionally, featured genes for SCK1.1 in the top windows were PXMP2, which is involved in the Peroxisome pathway, and GOLGA3, which has a function in transporter activity. Several genomic regions and SNPs previously reported in other studies as linked to subclinical ketosis susceptibility in dairy cattle were also identified in this study in addition to novel regions not previously reported. Enrichment analyses highlighted key Gene Ontology terms (GO:0033500

(carbohydrate homeostasis), GO:0048306 (calcium-dependent protein binding), GO:0046332

(SMAD binding) and GO:0003924 (GTPase activity)). In addition, important metabolic pathways related to fatty acid metabolism were also identified. A higher presence of genes linked to metabolic pathways were found for the second period of early lactation compared to the first period for both first and later lactations. This study offers new insights into the fundamental molecular mechanism of subclinical ketosis in cattle and highlights promising candidate genes for susceptibility to ketosis.

146

Introduction

In intensive milk production systems, cows experience a significant increase in energy demand associated with a drop in dry matter intake and a greater predisposition to conserve body fat, which favors the negative energy balance (NEB), especially in the first months of lactation, predisposing the cow to various metabolic diseases and its consequences (Duffield, 2000; Esposito et al., 2014). One of the main metabolic diseases that high yielding dairy cows develop during the transition period is ketosis. Ketosis is characterized by high serum concentration of nonesterified fatty acids (NEFAs) and ketone bodies, including β-hydroxybutyrate (BHB), in the blood and low concentration of glucose (Duffield et al., 2009). Dairy cows become highly susceptible to developing ketosis during early lactation (Loor et al., 2007), rarely occurring in late gestation. For instance, McArt et al.(2012) found an incidence of 43% of cows with elevated BHB during the first two weeks of lactation, with a peak occurring at 5 days in milk (DIM). Tatone et al. (2017) identified that cows had a higher risk of the first diagnosis of ketosis when the examination was performed within the first 14 days postpartum.

Ketosis may occur in a clinical and a subclinical form being differentiated by levels of

BHB and presence of clinical symptoms, thus the early detection of this disease is essential to reduce costs (Geishauser et al., 2000; McArt et al., 2015). Duffield et al. (1997) observed that, in cases of subclinical ketosis, the drop in production could reach 1 to 1.5 liters per day.

Subclinical ketosis is a multifactorial trait with low heritability ranging from 0.04 to 0.17

(Parker Gaddis et al., 2018; Pryce et al., 2016). Thus, it is important to understand the genetic basis of this disease. To help select less susceptible animals, the search for single nucleotide polymorphisms (SNPs) associated with different degrees of resistance to the development of ketosis is of critical importance. In this context, the genome-wide association (GWA) studies are

147

an effective technique for detecting genomic regions involved in the biological mechanisms underlying the studied trait (Ku et al., 2010; Nayeri et al., 2019).

Many studies have investigated genomic regions associated with susceptibility to clinical and subclinical ketosis in first and later lactations; however, significant genomic regions and genes harboring SNPs have not been investigated by using DIM to split the early lactation in two periods.

The first period would include the first two weeks of lactation when the risk of developing subclinical ketosis has been shown to be the highest (Duffield et al., 1997; Tatone et al., 2017) and the second period which would include the next three weeks. The aim of this study was 1) to estimate genetic parameters and variance components for subclinical ketosis (β-hydroxybutyrate;

BHB concentration) in North American Holstein dairy cattle in two periods in the start of lactation

(5-21 DIM and 22-45 DIM), 2) to conduct a GWAS to identify genomic regions associated with subclinical ketosis in the two periods of the lactation, and 3) to investigate potential candidate genes and metabolic pathways that could play a role in subclinical ketosis susceptibility.

Materials and Methods

Animals and dataset description

The datasets used in this study were provided by the Canadian Dairy Network (CDN), a member of Lactanet (Guelph, Ontario, Canada, www.lactanet.ca) and contained pedigree, genotypes, and phenotype information (milk BHB levels), which was predicted via mid-infrared spectroscopy (MIR) in routine DHI milk analyses.

Cows might respond differently to NEB depending on the DIM and number of lactations.

In this study, subclinical ketosis were evaluated from first and later lactations (2 to 5) recorded between 5 and 45 DIM (Chapinal et al., 2012; Jamrozik et al., 2016). This way, four traits were defined: Subclinical ketosis in first (SCK1.1) and later (SCK2.1) lactations from 5 to 21 DIM, and

148

subclinical ketosis in first (SCK1.2) and later (SCK2.2) lactations from 22 to 45 DIM. A total of

730,942 animals with records for SCK1.1, 826,120 for SCK1.2, 606,390 for SCK2.1, and 644,870 for SCK2.2 were included in the analyses. The pedigree consisted of 4,098,142 individuals traced back 7 generations.

Variance Component Estimation

A univariate animal model was fit for milk BHB at 5-21 DIM or for milk BHB at 22-45 DIM to estimate variance components and heritabilities. In matrix notation, the model was as follow:

풚 = ퟏµ + 푿풃 + 풁ퟏ풉 + 풁ퟐ풂 + 풆 where 풚 is the vector of observations for SCK1.1, SCK2.1, SCK1.2, or SCK2.2, µ is the overall mean; b is a vector of all fixed effects (including herd, year-season of measurement, age of cow- season-parity of measurement and DIM in the corresponding period as a covariate), h is a vector of herd-year effects; 풂 is the vector of animal additive genetic effects; 풆 is the vector of random residuals; 1 is a vector of ones, and X and 풁풊 are incidence matrices relating fixed effects and random effects to y.

Genotypic data and quality control

A total of 9,787 Holstein cows were genotyped with the Illumina bovine SNP60 (60K,

57,798 SNPs) BeadChip (Illumina, Inc., San Diego, CA, USA). These SNPs have passed the standard quality control (QC) used by CDN. A sample of 652 Canadian Holstein cows were also genotyped with a low-density array containing 998 putative ketosis markers (1K, 998 SNPs).

These 998 markers are a combination of novel and previous reported SNP markers that are located within or near candidate genes related to ketosis (Kroezen et al., 2018). The 60K and 1K in the

149

652 cows were combined to create a new panel (61K SNP). Genotype imputation from the 60K to

61K panel was performed for the remaining animals using the FImpute V2.2 software (Sargolzaei et al., 2014). The algorithm in the FImpute is based on the combination of family and population technique to reconstruct haplotypes (Sargolzaei et al., 2014). To estimate the accuracy of imputation, animals were randomly selected from the reference population (61K panel) and their genotypes were masked down to the 60K panel. The squared correlation between imputed genotypes and true genotypes (푟2) was used as a measure of imputation accuracy (Browning &

Browning, 2009) and it was r2= 0.98.

After imputation, a quality control was performed using the BLUPF90 software (Misztal,

2019). The genotyping quality control filtered out markers with minor allele frequency (MAF) lower than 1% and with a heterozygosity that deviates by more than 15% from the expected value under Hardy-Weinberg equilibrium (HWE). After quality control, 57,957 SNPs remained for further analyses.

De-regressed EBV

De-regressed EBVs (dEBVs) computed following Garrick et al. (2009) were used as response variables for genomic prediction. This method results in dEBVs containing only information for each animal with their own and the descendant`s information because it eliminates shrinkage in the EBV and adjusts for ancestral information.

Genome-wide association studies (GWAS)

The SNP effects were estimated using the weighted single-step method (wssGBLUP) proposed by Wang et al. (2012). The wssGBLUP model can be described as:

∗ 풚 = µ + 풁풂풂 + 풆

150

∗ where 풚 is the vector of pseudo phenotypes (dEBVs); µ is a vector of the overall mean; 풁풂 is an incidence matrix that relates animals to dEBVs; 풂 is the vector of direct additive genetic effects

2 2 and 풆 is the vector of random residuals. It was assumed that a~N(0, H휎푎 ) and e~N(0, R휎푒 ), where

2 H is the relationship matrix based on genomic and pedigree information, 휎푎 is the additive genetic variance, R is an diagonal matrix, whose elements account for the differences in the reliabilities of

2 the de-regressed EBVs, and 휎푒 is the residual variance. The inverse of H matrix can be defined as

(Aguilar et al., 2010):

−ퟏ −ퟏ ퟎ ퟎ 푯 = 푨 + [ −ퟏ −ퟏ] ퟎ 푮 − 푨ퟐퟐ

The solutions of SNP effect estimates (풖̂) were obtained using the formula: 풖̂ =

푫풁′[풁푫풁′]−ퟏ풂̂, where: D is a diagonal matrix with weights for SNPs; Z is an incidence matrix of genotyped for each locus, Z’ is the transpose of Z matrix, and 풂̂ is the vector of dEBVs of genotyped animals. The 풖̂ vector and the D matrix were derived as follows (Wang et al., 2012):

1. In the first iteration 푡 = 0 and 푫풕 = 푰, where 푡 is the iteration number, 푫풕 is matrix D at iteration

푡, and I is an identity matrix.

2. Calculate the SNP effects at iteration 푡(풖̂(풕)).

∗ 2 3. Recalculate the diagonal elements of D to obtain as: 푫(풕+ퟏ) = 푢̂푖(푡)2푝푖(1 − 푝푖) for all SNPs,

th where 푝푖 is the allele frequency of the reference allele of the i SNP. 151

∗ ∗ 4. Normalize 푫(풕+ퟏ) = (풕풓(푫(ퟎ))/풕풓(푫(풕+ퟏ)))푫(풕+ퟏ).

5. 푡 = 푡 + 1.

6. Exit, or loop to step 2. The SNP effects and weights were iteratively recomputed over two iterations. The iteration process increases weight of SNP with large effects and decrease those with small effects, resulting in an increasing proportion of variance explained by the remaining markers.

SNP effects were calculated using POSTGSF90 (Misztal et al., 2018) and sliding windows of 20 adjacent SNPs.

Functional annotation analyses

Genomic regions explaining the largest proportion of genetic variance were investigated for BHB using the GALLO R package (Version 1.0.; R Core Team, 2013) for genomic functional annotation of positional candidate loci in livestock (https://github.com/pablobio/GALLO). Genes within the top-20 windows were annotated, and the list of candidate genes was used to identify

Gene Ontology (GO) terms and to search metabolic pathways in the Kyoto Encyclopedia of Genes and Genomes (KEGG). The ARS-UCD1.2 assembly was used as the reference map.

Results

Estimation of genetic parameters

Table 1 shows variance components and heritability estimates. Heritability estimates ranged from 0.14 (SCK2.2) to 0.20 (SCK1.2), being higher for first lactation (~0.20) compared to later lactations (~0.14), but very similar for the two periods, from 5 to 21 DIM and from 22 to 45

DIM.

Genome-wide association analyses (GWAS)

The proportions of genetic variance explained by the top-20 sliding windows for SCK1.1,

SCK1.2, SCK2.2, and SCK2.2 are presented in Table 2, Table 3, Table 4, and Table 5, respectively.

152

The highest proportion of genetic variance observed using wssGBLUP analyses were found on chromosomes 17 for SCK1.1, 16 for SCK1.2, chromosome 9 for SCK2.1 and chromosome 20 for

SCK2.2. The sum of the proportion of variances explained by the top-20 genomic windows were

20.9%, 18.0%, 18.9% and 16.8% for SCK1.1, SCK1.2, SCK2.1, and SCK2.2, respectively.

The genetic variance explained by the SNPs in the GWAS analyses are shown as

Manhattan plots for SCK1.1 (Figure 1), SCK1.2 (Figure 2), SCK2.1 (Figure 3) and SCK2.2

(Figure 4). The results pointed important genomic windows on chromosomes 11, 13, 16 and 17 for SCK1.1; on chromosomes 9, 16, 24 and 26 for SCK1.2; on chromosomes 1, 9 and 24 for

SCK2.1; and on chromosomes 1, 5, 17 and 20 for SCK2.2, which individually explained more than 1% of the total genetic variance.

Functional annotation

The annotated genes within the top-20 sliding windows associated with the four traits from wssGBLUP analyses are presented in Table 6. For SCK1.1, 123 annotated genes were found within the top-20 genomic windows. For SCK1.2 136 genes were identified within the top-20 genomic windows. For SCK2.1, a total of 120 genes were located within the top-20 genomic windows. For

SCK2.2, a total of 87 genes was found within the top-20 genomic windows.

Supplementary Tables 1, 2, 3, and 4 present the gene ontology (GO) enrichment analyses performed for all candidate genes and traits to investigate genes involved in known metabolic pathways. For SCK1.1, 10 GO terms were found, whereas 30 GO terms were associated with

SCK1.2. For SCK2.1, 15 GO terms were found and, for SCK2.2, 43 GO terms were identified.

Based on the functional annotation results, significant pathways were found for all traits and are shown in Supplementary Tables 5, 6, 7 and 8. Highlighted pathways included fatty acid

153

metabolism, fatty acid degradation, fatty acid biosynthesis, arachidonic acid metabolism, and pyruvate metabolism.

Discussion

Previous studies suggested that the prevalence of subclinical ketosis varies across parity and stage of lactation in cows (Duffield et al., 1997; Toni et al., 2011; van der Drift et al., 2012).

In this study, variance components and heritability were estimated for Subclinical ketosis for first lactation in the periods from 5 to 21 DIM (SCK1.1) and from 22 to 45 DIM (SCK1.2), and for later lactations in the periods from 5 to 21 DIM (SCK2.1) and from 22 to 45 DIM (SCK2.2) in

Holstein cattle. For these same traits, single-step GWAS analyses were performed. The results of the GWAS were further investigated by functional analyses to identify candidate genes and genetic variants. In addition, metabolic pathways were analyzed to reveal potential biological relationship.

Estimation of genetic parameters

Heritability estimates ranged from 0.14 (SCK2.2) to 0.20 (SCK1.2), being higher for first lactation (~0.20) compared to later lactations (~0.14), but very similar for the two periods considered, from 5 to 21 DIM and from 22 to 45 DIM. These estimates are close to those reported in the literature. Estimates of heritability for milk BHB concentration were previously reported with a significant range from 0.14 to 0.29 (Koeck et al., 2014). For first lactation and 30 days in milk, S. Lee et al. (2016) reported a heritability of 0.10 for milk BHB. Ranaraja et al. (2018) reported an estimated heritability of 0.14 for first parity and lower estimates for the three subsequent parities (from 0.11 to 0.09). Therefore, in general the results suggest that these traits are under moderate genetic control and should respond to selection.

154

Candidate genes from GWAS analysis

The high number of windows explaining a small proportion of the genetic variance for the traits indicates that there are no QTL with large effects, therefore, the susceptibility of Holsteins to ketosis appears to be influenced by a large number of genes with small effects across the genome. The largest peak was identified on chromosome 17 for SCK1.1, which explained 3.53% of the genetic variance, and the second largest on chromosome 9 for SCK2.1 explaining 3.10% of the variance. For SCK1.2 and SCK2.2, the window explaining the highest proportion of variance explained 1.60% and 1.53%, respectively. The windows explaining the highest proportion of the genetic variance when considering 5 to 21 DIM in both first and later lactations explained almost double of the proportion explained when considering 22 to 45 DIM period. Windows on chromosome 1 were found for all traits, on chromosome 19 for SCK1.1, SCK2.1 and SCK2.2, on chromosome 6 for SCK1.1 and SCK2.1, and on chromosome 20 for SCK1.2 and SCK2.2. Regions on chromosomes 6 and 20 have been shown to be associated to ketosis before (Nayeri et al., 2019;

Parker Gaddis et al., 2018) and genes within regions on chromosome 20 were previously associated with inflammatory response (Nayeri et al., 2019).

For subclinical ketosis in first lactation from 5 to 21 DIM (SCK1.1)

An important gene that mapped to a candidate window for SCK1.1 was SLC19A3 (Solute

Carrier Family 19 Member 3), which was shown to act as a biotin and thiamine transporter

(Nabokina & Said, 2004; Vlasova et al., 2005). Mutations in SLC19A3 gene are associated with pyruvate oxidation defects, lactate elevation, and increased levels of leucine and valine (Sperl et al., 2015; W.-Q. Zeng et al., 2005).

The PSMD12 (Proteasome 26S Subunit, Non-ATPase 12) gene was mapped to a candidate window on chromosome 19 and it is associated with protein catabolism and aspartate

155

aminotransferase which indicates liver function (Kim et al., 2011; Pihlajamäki et al., 2009). This gene was also associated with the proteasome pathway, which acts in the inhibition of cytokine production by liver cells (Kim et al., 2011; Richert et al., 2006). Loor et al. (2011) indicated that the expression of the PSMD12 gene was affected by mastitis-causing pathogens and associated with a protein ubiquitination pathway specific to the early or late response to intramammary infection in several ruminant species or bovine specific responses.

The PXMP2 (Peroxisomal Membrane Protein 2) gene was mapped to the candidate window on chromosome 17, which explained 3.53% of the genetic variance, and encodes the

PMP2 that, in the membrane of peroxisomes, is the most abundant protein component (Alexson et al., 1985; Rokka et al., 2009). This protein acts as a non-selective channel which transfer solutes across the peroxisomal membrane and allows α- and β-oxidation products to exit (Rokka et al.,

2009). The peroxisome pathway, in which the PXMP2 gene is involved, has significant β-oxidation capability in cow's liver tissue (Drackley et al., 2001). Disturbances of lipid metabolism in the mammary fat pad occur during reduced expression of PXMP2 (Vapola et al., 2014).

The GOLGA3 (golgin subfamily A member 3) gene is a Golgi complex-associated protein implicated in protein trafficking, apoptosis, positioning of the Golgi, and spermatogenesis, and it was mapped to a window on chromosome 17. This gene was suggested to have a role in disease resistance in goats (Cecchi et al., 2017).

For subclinical ketosis in first lactation from 22 to 45 DIM (SCK1.2)

Highlighted genes for SCK1.2 were ACADSB (acyl-CoA dehydrogenase, short/branched chain) on chromosome 26 in a window explaining 1.42% of the genetic variance, and ALDH1A1

(Aldehyde dehydrogenase 1 family, member A1) on chromosome 8. The ACADSB gene is a member of the acyl-CoA dehydrogenase family of enzymes that catalyse the dehydrogenation of

156

acyl-CoA derivatives in the fatty acid metabolism and is important for the metabolism of amino acids in mitochondria (Alfardan et al., 2010). Jiang et al. (2018) showed that ACADSB gene overexpression could significantly increase intracellular triglyceride levels and is associated with milk fat metabolism. The ALDH1A1 is associated with important pathways for ketosis such as pyruvate metabolism, fatty acid metabolism, propanoate metabolism, arginine and proline metabolism, glycolysis/gluconeogenesis and tryptophan metabolism (McCarthy et al., 2010).

These genes were downregulated in the liver of Holstein heifers in periods of food restriction

(Doelman et al., 2012).

The genes HADHA (hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit alpha) and HADHB (beta) were also found in an association study for subclinical ketosis in later lactations (data not published). These genes impact the influx of NEFA into the liver and mutations in these genes could impact the mitochondrial tri-functional protein resulting in trifunctional protein deficiency (Drackley et al., 2005; Schäff et al., 2013). The HADHAB and

HADHB are involved in fatty acid synthesis and degradation pathway (Maeyashiki et al., 2017).

The PDHX (Pyruvate Dehydrogenase Complex Component X), found in a window on chromosome 15, catalyzes the conversion of pyruvate into acetyl CoA and is associated with body weight and body mass index (Fox et al., 2007). In humans, mutations on this gene are associated with pyruvate dehydrogenase complex deficiency (Patel et al., 2012). Two SNPs in the PDHX gene were associated with citrate and fatty acid synthesis in cattle (Cánovas et al., 2013).

For subclinical ketosis in later lactations from 5 to 21 DIM (SCK2.1)

The genes RIMS1 (Regulating Synaptic Membrane Exocytosis 1) and KCNQ5 (Potassium

Voltage-Gated Channel Subfamily Q Member 5) were found in more than one window for

SCK2.1. RIMS1 is known to regulate calcium channels with voltage gated and exocytosis of the

157

synaptic vesicle (Sisodiya et al., 2007; Han & Peñagaricano, 2016). Polymorphisms in the RIMS1 gene were associated with impaired insulin secretion, and this gene has also been implicated with granule docking (Andersson et al, 2012; Gandasi et al., 2018). The gene KCNQ5 is a member of the KCNQ potassium channel gene family and is primarily expressed in the brain and skeletal muscle (Schroeder et al., 2000).

The DCXR (Dicarbonyl and L-Xylulose Reductase) gene is involved in several processes within the organism and is particularly expressed in the liver and kidney (Ebert et al., 2015; S.-K.

Lee et al., 2013). In addition, DCXR is involved in the fatty acid composition and in carbohydrate metabolism (Ebert et al., 2015; Palombo et al., 2018).

The FASN (Fatty acid synthase) gene catalyzes long-chain fatty acid synthesis and plays a central role in mammalian de novo lipogenesis (Matsumoto et al., 2012). Mutations in this gene affect the percentage of C14:0 in both bovine adipose tissue and milk fat (C. A. Morris et al.,

2007). Zhu et al. (2019) showed that cows with subclinical ketosis had decreased gene expression of FASN; this can be justified by elevated fatty acid β-oxidation in cows with subclinical ketosis.

In addition, Loor et al. (2007) showed that ketosis is associated with downregulation of de novo fatty acid synthesis.

Proline biosynthesis involves the PYCR1 (pyrroline-5-carboxylate reductase 1) gene which is responsible for catalysing the NADH-dependent conversion of pyrroline-5-carboxylate (P5C) to L-proline (T. Zeng et al., 2017). The PCYCR1 is associated with the metabolism of intracellular amino acids and has been found to be upregulated during lactation (Dai et al., 2018). This gene also influences insulin resistance, which has been demonstrated by high circulating proline levels in subjects with insulin resistance (Tai et al., 2010; Zhuang et al., 2019).

158

The PCYT2 (phosphate cytidylyltransferase 2, ethanolamine) gene is the major regulatory enzyme in ethanolamine phosphatidylethanolamine and diacylglycerol biosynthesis via the

Kennedy CDP-ethanolamine pathway (Gibellini & Smith, 2010). The greatest physiological consequences of removing one PCYT2 allele include the development of metabolic syndrome symptoms such as elevated lipogenesis and lipoprotein secretion, hepatic steatosis, and insulin resistance. The main physiological implications of deleting one PCYT2 allele include the emergence of metabolic syndrome symptoms such as elevated lipogenesis and lipoprotein secretion, hepatic steatosis, and insulin resistance (Fullerton et al., 2009; Pavlovic & Bakovic,

2013).

For subclinical ketosis for later lactations from 22 to 45 DIM (SCK2.2)

The family Acyl-coenzyme A dehydrogenases (ACAD) catalyzes fatty acid α,β- dehydrogenation (Crane & Beinert, 1956; Nguyen et al., 2002). The ACAD8 (Acyl-CoA

Dehydrogenase Family Member 8) gene encodes an enzyme called isobutyrylcoenzyme dehydrogenase (IBD) toward the gluconeogenic pathway, which acts in the conversion of isobutyryl-CoA to methylacrylyl-CoA in the metabolism of valine (Ghisla & Thorpe, 2004; Roe et al., 1998). Comparing lactating to nonlactating sows, this gene has been shown to have an increased expression in the liver of lactating sows (Rosenbaum et al., 2012). Moreover, this gene was associated with growth and the beef tenderness in cattle (H. Li et al., 2007).

Another important gene for SCK2.2 was the gene ADCY8 (Adenylate cyclase 8) which plays a critical role in the secretion of insulin and is responsible for mediating glucagon-like peptide 1 (GLP-1) activity (Delmeire et al., 2003). It is an enzyme required for activation of glucose-induced signalling pathways in β-cells as well as for glucose tolerance via islet insulin

159

secretion regulation. Furthermore, ADCY8 reduction is linked to damaged β-cell secretory capacity in diabetes (Raoux et al., 2015).

The PSMA4 (Proteasome 20S Subunit Alpha 4) and PSMB6 (Proteasome 20S Subunit Beta

6) are genes that encode α and β proteasome components, respectively (Gomes, 2013). Both genes were down-regulated in higher intra-muscular fat muscle in pork and have been identified in immunoproteosome pathway (Hamill et al., 2012; Vahey et al., 2010). Ibeagha-Awemu et al.

(2016) have identified associations of SNPs in PSMA4 with fatty acid traits. The gene PSMB6 has been shown to be significant in the presentation of MHC class I molecular antigen, and its knockdown has been shown to reduce cellular viability (L. Sun et al., 2013; X. Wu et al., 2011).

A variant of CYC1 (Cytochrome c1) was associated with Mitochondrial Complex III

Deficiency (Anastasio et al., 2016). This gene was found in QTL intervals identified based on single-trait GWAS for milk yield, fat yield and protein yield and it was associated with the oxidative phosphorylation pathway (Cai et al., 2020; Grala et al., 2013). The SDHB (Succinate

Dehydrogenase Complex Iron Sulfur Subunit B) gene has also been associated with the oxidative phosphorylation pathway, as well as with mitochondrial dysfunction and butanoate metabolism pathways (Forde et al., 2012; D. G. Morris et al., 2009).

Overlapping and differences among traits

Six genes found to be associated with SCK1.2, SCK2.1 and SCK2.2 were mapped to a common candidate window (50Kb) on chromosome 24. From these, highlighted genes were ME2

(Malic Enzyme 2), SMAD4 (SMAD Family Member 4) and MEX3C (Mex-3 RNA Binding Family

Member C).

The enzyme coded by the ME2 gene is widely found in insulin cell lines and pancreatic islets and is responsible for catalysing the oxidative decarboxylation of malate to pyruvate that is

160

involved in the synthesis of glutamate (Hasan et al., 2015; Hassel & Bråthe, 2000). The mitochondrial ME2 is associated with insulin release, responding to elevated amino acids and providing sufficient pyruvate when glucose is restricted (Pongratz et al., 2007). This gene was differentially expressed in mammary tissue after intramammary infection with E. coli or S. uberis

(Chen et al., 2015). The SMAD4 gene codes for a member of a protein family that work as intracellular mediator for the transforming growth factor beta (TGF-β) superfamily (K.-B. Lee et al., 2014). It is a co-smad which regulates other smad’s pathways and it is involved in myostatin signaling, cell differentiation, growth and apoptosis (Attisano & Wrana, 2000; Zhou et al., 1998).

Another highlighted gene was MEX3C which participates in cellular growth by up- regulating insulin-like growth factor 1 (IGF1) expression, and in mice IGF1 deficiency in developing bone causes growth retardation (Y. Jiao et al., 2012; Yan Jiao et al., 2012). A mutation in mice MEX3C was shown to improve glucose and lipid profiles (Han et al., 2014). Furthermore, this gene is associated with energy balance regulation and plays a role in ubiquitination of proteins, immune response, growth, and adiposity (Y. Jiao et al., 2012; Kuniyoshi et al., 2014; Zhao et al.,

2016). It is possible to notice genes in common among all traits except for SCK1.1, and this might be explained by the fact that in the first lactation, specifically the first DIM period, the most active mechanism is the regulation of gene expression, while in the other traits, genes involved in the metabolic pathway may play a greater role, and perhaps this explains the fact that there are no genes in common among all traits.

For SCK1.1, the window explaining the highest proportion of the genetic variance was located on chromosome 17 and genes involved in regulation of gene expression were found in that region. Zinc Finger Proteins were heavily present in this genomic region. The genes were ZNF605

(Zinc Finger Protein 605), ZNF26 (Zinc Finger Protein 26), ZNF84 (Zinc Finger Protein 84),

161

ZNF140 (Zinc Finger Protein 140), ZNF891 (Zinc Finger Protein 891), ZNF10 (Zinc Finger

Protein 10), and ZNF268 (Zinc Finger Protein 268) and among their related pathways is Gene

Expression. Parker Gaddis et al. (2018) found the gene ZNF638 (Zinc Finger Protein 638), which is from the same family, to be associated with ketosis in Jerseys. Studies have shown that one to two weeks after the start of lactation is the period of highest risk for the development of subclinical ketosis (Duffield et al., 1997; Tatone et al., 2017), and this might be related to the expression of regulatory genes, such as the ones mentioned above. In addition, genes involved in DNA methylation, such as MBD3L1 (Methyl-CpG Binding Domain Protein 3 Like 1), might suggest a period of initial adaptation occurring in this first period of DIM, since the animal has not experienced NEB before, therefore, this might explain the highest incidence of subclinical ketosis in the 5 to 21 DIM period.

Another commonality found between SCK1.1 and SCK1.2 was the genes ACSL3 (Acyl-

CoA Synthetase Long Chain Family Member 3) and KCNE4 (Potassium Voltage-Gated Channel

Subfamily E Regulatory Subunit 4). The Acyl-CoA synthetases (ACSLs) family produces fatty

Acyl-CoA esters from free long-chain fatty acids (Coleman et al., 2002). The ACSL3 has a substrate specificity for palmitate and it is predominantly expressed in endoplasmic reticulum (ER) and lipid droplets (Grevengoed et al., 2014; Soupene & Kuypers, 2008). The gene has been shown to transcriptionally regulate lipogenic genes and it was down-regulated in the liver when de novo lipogenesis is reduced (Bu et al., 2009; Ellis et al., 2015). In addition, a study showed that ACSL3 was implicated in the synthesis of phosphatidylcholine, which is essential for the assembly of very low density lipoproteins (VLDL) (Yao & Ye, 2008). This gene was only found for traits in first lactation for both periods of DIM. The KCNE4 gene is involved in ion transport and modulates

162

voltage-gated potassium channels (Grunnet et al., 2003). It was up-regulated in cows with high genetic merit for fertility (Moran et al., 2017).

The traits SCK2.1 and SCK2.2 had 11 genes in common, among them the genes RERGL

(RERG Like), SEPTIN9 (Septin 9) and PIK3C2G (Phosphatidylinositol-4-Phosphate 3-Kinase

Catalytic Subunit Type 2 Gamma). The RERGL gene is ubiquitously expressed, even in adipose tissue, with no known role established. However, protein alignment suggests significant sequence homology with the ras superfamily member RERG (RAS-like, estrogen-regulated, growth inhibitor) which encodes a small GTPase (Liu & Zhang, 2017; Uhlén et al., 2015). Also related to

Ras-like GTPase/GTP-binding domain, septins are a cytoskeletal protein family (Mostowy &

Cossart, 2012). In humans, septin proteins were found to build cage structures around bacterial pathogens acting in the immobilization of the bacteria (Mostowy et al., 2010). Methylation and decreased expression of the SEPTIN9 gene has been associated with liver fibrosis (Y. Wu et al.,

2017). The PIK3C2G gene has been shown to be associated with type 2 diabetes-related phenotypes (Daimon et al., 2008). In addition, copy number variants (CNV) at PIK3C2G were associated with hyperlipidemia and myocardial infarction (Shia et al., 2011). This gene has also been found in IL-4 signaling, virus entry via endocytic pathways and natural killer cell signaling pathways (Rinner et al., 2017). For SCK1.1 and SCK2.1, only U6 (spliceosomal RNA) was found in common, meaning that 122 genes and 119 genes are unique for each period, respectively. That is, the first DIM period itself did not show much in common between the first and later lactations.

Meanwhile, for SCK1.2 and SCK2.2, i.e. the second DIM period, no genes were found to be in common, with 136 and 87 genes being unique to each trait, respectively.

The genes discussed in previous sections for SCK1.1 and SCK2.1 were found in the top windows explaining the genetic variance for each trait. For instance, the genes PXMP2 and

163

GOLGA3 are in the first top window for SCK1.1, which explained 3.53% of the genetic variance, and RIMS1 and KCNQ5 are in the first top window explaining 3.10% of the genetic variance for

SCK2.1. Therefore, genes with functions associated with the development of ketosis were found in windows explaining the highest genetic variance for the period from 5 to 21 DIM, which might related with the higher incidence of ketosis in this period compared to 22 to 45 DIM period.

Based on the functional annotation results, significant pathways were found for subclinical ketosis and are summarized in Figure 5. An important metabolic pathway was fatty acid metabolism found to be associated with all traits (Supplementary tables 5, 6, 7 and 8). It is known that livers of ketotic cows possess high concentrations of acetyl-CoA, and the determination of its concentration is due to the steps in the fatty acid metabolism (Xu et al., 2008; Zhu et al., 2019).

Another highlighted metabolic pathway were Fatty acid biosynthesis and Arachidonic acid metabolism associated with SCK1.1, SCK1.2 and SCK2.1. Intermediates in fatty acid biosynthesis have the ability to supress or stimulate ketone body production (Holtenius & Holtenius, 1996).

Arachidonic acid is identified in the liver and its metabolism is a pathway associated with lipid metabolism (H.-Z. Sun et al., 2018). For SCK1.1, SCK1.2 and SCK2.2, the pathway Fatty acid degradation was found while Fatty acid elongation was only found for SCK1.2. The expression of enzymes involved in fatty acid degradation was shown to be decreased in liver tissues of ketotic cows (Xu et al., 2008). The pathway Pyruvate metabolism was associated with the traits SCK1.2,

SCK2.1 and SCK2.2 and it is related to the conversion of pyruvic acid to acetyl-CoA, contributing to ketogenesis and increased blood ketone content (Wang et al., 2016). In addition, the Peroxisome pathway, an alternative hepatic oxidation pathway for NEFA, was found for SCK1.1, SCK1.2 and

SCK2.1, and it is associated with an increased for β-oxidation in cow's liver tissue (Drackley et al., 2001). Another important pathway associated with SCK1.1 and SCK1.2 is Insulin resistance.

164

Animals with ketosis normally present insulin resistance and adipose tissue present insulin resistance as a response to lactation (Hayirli, 2006; Mills et al., 1986).

Conclusions

Heritability estimates reported in this study support breeding efforts for lowering incidence of subclinical ketosis in dairy cattle, as all four traits analyzed showed a sizable heritability, which was very similar in both periods studied in the start of the lactation, but higher for first lactation compared to subsequent lactations. In total, for all traits, 22 windows were found describing more than 1% of the genetic variation. Based on these windows, a list of potential candidate genes for subclinical ketosis was generated, which will be useful for future functional studies. Zinc Finger Proteins were heavily present on the first top window explaining the highest genetic variance of subclinical ketosis in the first DMI period (5 to 21 days) in first lactation cows and suggests that regulation of gene expression plays an important role in this first DIM period. In addition, highlighted genes in the top windows for subclinical ketosis in the first DMI period (5 to

21 days) in first and later lactations were PXMP2 and GOLGA3, and RIMS1 and KCNQ5, respectively. Compared to the first DIM period in both first and later lactations, the second DMI period (22 to 45 days) showed a higher presence of genes related to metabolic pathways. Several candidate genes showed activity in relation to cell structure, immune function, and ketone bodies synthesis and degradation. Overall highlighted genes were ACSL3, which transcriptionally regulates lipogenic genes, and ADCY8, which plays a role in the secretion of insulin. The metabolic pathway analysis provided insight into the biology of disease resistance. The pathway fatty acid metabolism was associated to all traits and has been shown to be important in the establishment of subclinical ketosis. Therefore, the findings of this study provide new insights into the underlying

165

molecular mechanism of ketosis and help to better understand the genetic architecture that underpins complex traits and diseases.

Acknowledgments

Canadian Dairy Network (Lactanet) is acknowledged for providing the data used in this study.

Canada, the Canadian Dairy Network and the Canadian Dairy Commission under the Agri-

References

Aguilar, I., Misztal, I., Johnson, D. L., Legarra, A., Tsuruta, S., & Lawlor, T. J. (2010). Hot

topic: A unified approach to utilize phenotypic, full pedigree, and genomic information

for genetic evaluation of Holstein final score. Journal of Dairy Science, 93(2), 743–752.

https://doi.org/10.3168/jds.2009-2730

Alexson, S. E., Fujiki, Y., Shio, H., & Lazarow, P. B. (1985). Partial disassembly of

peroxisomes. The Journal of Cell Biology, 101(1), 294–304.

https://doi.org/10.1083/jcb.101.1.294

166

Alfardan, J., Mohsen, A.-W., Copeland, S., Ellison, J., Keppen-Davis, L., Rohrbach, M., Powell,

B. R., Gillis, J., Matern, D., Kant, J., & Vockley, J. (2010). Characterization of new

ACADSB gene sequence mutations and clinical implications in patients with 2-

methylbutyrylglycinuria identified by newborn screening. Molecular Genetics and

Metabolism, 100(4), 333–338. https://doi.org/10.1016/j.ymgme.2010.04.014

Andersson, S. A., Olsson, A. H., Esguerra, J. L., Heimann, E., Ladenvall, C., Edlund, A., ... & Ling, C. (2012). Reduced insulin secretion correlates with decreased expression of exocytotic genes in pancreatic islets from patients with type 2 diabetes. Molecular and cellular endocrinology, 364(1-2), 36-45. Attisano, L., & Wrana, J. L. (2000). Smads as transcriptional co-modulators. Current Opinion in

Cell Biology, 12(2), 235–243. https://doi.org/10.1016/S0955-0674(99)00081-2

Browning, B. L., & Browning, S. R. (2009). A Unified Approach to Genotype Imputation and

Haplotype-Phase Inference for Large Data Sets of Trios and Unrelated Individuals.

American Journal of Human Genetics, 84(2), 210–223.

https://doi.org/10.1016/j.ajhg.2009.01.005

Bu, S. Y., Mashek, M. T., & Mashek, D. G. (2009). Suppression of long chain acyl-CoA

synthetase 3 decreases hepatic de novo fatty acid synthesis through decreased

transcriptional activity. The Journal of Biological Chemistry, 284(44), 30474–30483.

https://doi.org/10.1074/jbc.M109.036665

Cánovas, A., Rincón, G., Islas-Trejo, A., Jimenez-Flores, R., Laubscher, A., & Medrano, J. F.

(2013). RNA sequencing to study gene expression and single nucleotide polymorphism

variation associated with citrate content in cow milk. Journal of Dairy Science, 96(4),

2637–2648. https://doi.org/10.3168/jds.2012-6213

167

Cecchi, F., Russo, C., Iamartino, D., Galiero, A., Turchi, B., Fratini, F., Degl’Innocenti, S.,

Mazza, R., Biffani, S., Preziuso, G., & Cantile, C. (2017). Identification of candidate

genes for paratuberculosis resistance in the native Italian Garfagnina goat breed. Tropical

Animal Health and Production, 49(6), 1135–1142. https://doi.org/10.1007/s11250-017-

1306-8

Chapinal, N., Leblanc, S. J., Carson, M. E., Leslie, K. E., Godden, S., Capel, M., Santos, J. E. P.,

Overton, M. W., & Duffield, T. F. (2012). Herd-level association of serum metabolites in

the transition period with disease, milk production, and early lactation reproductive

performance. Journal of Dairy Science, 95(10), 5676–5682.

https://doi.org/10.3168/jds.2011-5132

Chen, X., Cheng, Z., Zhang, S., Werling, D., & Wathes, D. C. (2015). Combining Genome Wide

Association Studies and Differential Gene Expression Data Analyses Identifies Candidate

Genes Affecting Mastitis Caused by Two Different Pathogens in the Dairy Cow. Open

Journal of Animal Sciences, 05(04), 358–393. https://doi.org/10.4236/ojas.2015.54040

Coleman, R. A., Lewin, T. M., Van Horn, C. G., & Gonzalez-Baró, M. R. (2002). Do long-chain

acyl-CoA synthetases regulate fatty acid entry into synthetic versus degradative

pathways? The Journal of Nutrition, 132(8), 2123–2126.

https://doi.org/10.1093/jn/132.8.2123

Crane, F. L., & Beinert, H. (1956). On the mechanism of dehydrogenation of fatty acyl

derivatives of coenzyme A. II. The electron-transferring flavoprotein. The Journal of

Biological Chemistry, 218(2), 717–731.

168

Dai, W.-T., Zou, Y.-X., White, R. R., Liu, J.-X., & Liu, H.-Y. (2018). Transcriptomic profiles of

the bovine mammary gland during lactation and the dry period. Functional & Integrative

Genomics, 18(2), 125–140. https://doi.org/10.1007/s10142-017-0580-x

Daimon, M., Sato, H., Oizumi, T., Toriyama, S., Saito, T., Karasawa, S., Jimbu, Y., Wada, K.,

Kameda, W., Susa, S., Yamaguchi, H., Emi, M., Muramatsu, M., Kubota, I., Kawata, S.,

& Kato, T. (2008). Association of the PIK3C2G gene polymorphisms with type 2 DM in

a Japanese population. Biochemical and Biophysical Research Communications, 365(3),

466–471. https://doi.org/10.1016/j.bbrc.2007.10.180

Delmeire, D., Flamez, D., Hinke, S. A., Cali, J. J., Pipeleers, D., & Schuit, F. (2003). Type VIII

adenylyl cyclase in rat beta cells: Coincidence signal detector/generator for glucose and

GLP-1. Diabetologia, 46(10), 1383–1393. https://doi.org/10.1007/s00125-003-1203-8

Doelman, J., Cao, H., Purdie, N. G., Kim, J. J. M., Swanson, K. C., Osborne, V. R., Tey, J., Ali,

A., Feng, Z., Karrow, N. A., & Cant, J. P. (2012). Transcript profiling of the ruminant

liver indicates a unique program of transcriptional regulation of ketogenic enzymes

during food restriction. Comparative Biochemistry and Physiology Part D: Genomics

and Proteomics, 7(3), 303–310. https://doi.org/10.1016/j.cbd.2012.06.002

Drackley, J. K., Dann, H. M., Douglas, N., Guretzky, N. A. J., Litherland, N. B., Underwood, J.

P., & Loor, J. J. (2005). Physiological and pathological adaptations in dairy cows that

may increase susceptibility to periparturient diseases and disorders. Italian Journal of

Animal Science, 4(4), 323–344. https://doi.org/10.4081/ijas.2005.323

Duffield, T.F. (2000). Subclinical Ketosis in Lactating Dairy Cattle. Veterinary Clinics of North

America: Food Animal Practice, 16(2), 231–253. https://doi.org/10.1016/S0749-

0720(15)30103-1

169

Duffield, T.F., Lissemore, K. D., McBride, B. W., & Leslie, K. E. (2009). Impact of

hyperketonemia in early lactation dairy cows on health and production. Journal of Dairy

Science, 92(2), 571–580. https://doi.org/10.3168/jds.2008-1507

Duffield, Todd F, Kelton, D. F., Leslie, K. E., Lissemore, K. D., & Lumsden, J. H. (1997). Use

of test day milk fat and milk protein to detect subclinical ketosis in dairy cattle in

Ontario. 38, 6.

Ebert, B., Kisiela, M., & Maser, E. (2015). Human DCXR - another “moonlighting protein”

involved in sugar metabolism, carbonyl detoxification, cell adhesion and male fertility?

Biological Reviews of the Cambridge Philosophical Society, 90(1), 254–278.

https://doi.org/10.1111/brv.12108

Ellis, J. M., Bowman, C. E., & Wolfgang, M. J. (2015). Metabolic and Tissue-Specific

Regulation of Acyl-CoA Metabolism. PLoS ONE, 10(3).

https://doi.org/10.1371/journal.pone.0116587

Esposito, G., Irons, P. C., Webb, E. C., & Chapwanya, A. (2014). Interactions between negative

energy balance, metabolic diseases, uterine health and immune response in transition

dairy cows. Animal Reproduction Science, 144(3–4), 60–71.

https://doi.org/10.1016/j.anireprosci.2013.11.007

Forde, N., Mehta, J. P., Minten, M., Crowe, M. A., Roche, J. F., Spencer, T. E., & Lonergan, P.

(2012). Effects of Low Progesterone on the Endometrial Transcriptome in Cattle1.

Biology of Reproduction, 87(5). https://doi.org/10.1095/biolreprod.112.103424

Fox, C. S., Heard-Costa, N., Cupples, L. A., Dupuis, J., Vasan, R. S., & Atwood, L. D. (2007).

Genome-wide association to body mass index and waist circumference: The Framingham

170

Heart Study 100K project. BMC Medical Genetics, 8 Suppl 1, S18.

https://doi.org/10.1186/1471-2350-8-S1-S18

Fullerton, M. D., Hakimuddin, F., Bonen, A., & Bakovic, M. (2009). The development of a

metabolic disease phenotype in CTP:phosphoethanolamine cytidylyltransferase-deficient

mice. The Journal of Biological Chemistry, 284(38), 25704–25713.

https://doi.org/10.1074/jbc.M109.023846

Gandasi, N. R., Yin, P., Omar-Hmeadi, M., Laakso, E. O., Vikman, P., & Barg, S. (2018). Glucose-dependent granule docking limits insulin secretion and is decreased in human type 2 diabetes. Cell metabolism, 27(2), 470-478. Geishauser, T., Leslie, K., Tenhag, J., & Bashiri, A. (2000). Evaluation of Eight Cow-Side

Ketone Tests in Milk for Detection of Subclinical Ketosis in Dairy Cows. Journal of

Dairy Science, 83(2), 296–299. https://doi.org/10.3168/jds.S0022-0302(00)74877-6

Gessner, D. K., Schlegel, G., Ringseis, R., Schwarz, F. J., & Eder, K. (2014). Up-regulation of

endoplasmic reticulum stress induced genes of the unfolded protein response in the liver

of periparturient dairy cows. BMC Veterinary Research, 10(1), 46.

https://doi.org/10.1186/1746-6148-10-46

Ghisla, S., & Thorpe, C. (2004). Acyl-CoA dehydrogenases. European Journal of Biochemistry,

271(3), 494–508. https://doi.org/10.1046/j.1432-1033.2003.03946.x

Gibellini, F., & Smith, T. K. (2010). The Kennedy pathway—De novo synthesis of

phosphatidylethanolamine and phosphatidylcholine. IUBMB Life, 62(6), 414–428.

https://doi.org/10.1002/iub.337

Gomes, A. V. (2013). Genetics of Proteasome Diseases. Scientifica, 2013.

https://doi.org/10.1155/2013/637629

171

Grevengoed, T. J., Klett, E. L., & Coleman, R. A. (2014). Acyl-CoA Metabolism and

Partitioning. Annual Review of Nutrition, 34, 1–30. https://doi.org/10.1146/annurev-nutr-

071813-105541

Grunnet, M., Rasmussen, H. B., Hay-Schmidt, A., Rosenstierne, M., Klaerke, D. A., Olesen, S.-

P., & Jespersen, T. (2003). KCNE4 Is an Inhibitory Subunit to Kv1.1 and Kv1.3

Potassium Channels. Biophysical Journal, 85(3), 1525–1537.

https://doi.org/10.1016/S0006-3495(03)74585-8

Hamill, R. M., McBryan, J., McGee, C., Mullen, A. M., Sweeney, T., Talbot, A., Cairns, M. T.,

& Davey, G. C. (2012). Functional analysis of muscle gene expression profiles associated

with tenderness and intramuscular fat content in pork. Meat Science, 92(4), 440–450.

https://doi.org/10.1016/j.meatsci.2012.05.007

Han, Y., & Peñagaricano, F. (2016). Unravelling the genomic architecture of bull fertility in Holstein cattle. BMC genetics, 17(1), 143. Hasan, N. M., Longacre, M. J., Stoker, S. W., Kendrick, M. A., & MacDonald, M. J. (2015).

Mitochondrial Malic Enzyme 3 Is Important for Insulin Secretion in Pancreatic β-Cells.

Molecular Endocrinology, 29(3), 396–410. https://doi.org/10.1210/me.2014-1249

Hassel, B., & Bråthe, A. (2000). Neuronal Pyruvate Carboxylation Supports Formation of

Transmitter Glutamate. The Journal of Neuroscience, 20(4), 1342–1347.

https://doi.org/10.1523/JNEUROSCI.20-04-01342.2000

Hayirli, A. (2006). The Role of Exogenous Insulin in the Complex of Hepatic Lipidosis and

Ketosis Associated with Insulin Resistance Phenomenon in Postpartum Dairy Cattle.

Veterinary Research Communications, 30(7), 749–774. https://doi.org/10.1007/s11259-

006-3320-6

172

Holtenius, P., & Holtenius, K. (1996). New Aspects of Ketone Bodies in Energy Metabolism of

Dairy Cows: A Review. Journal of Veterinary Medicine Series A, 43(1–10), 579–587.

https://doi.org/10.1111/j.1439-0442.1996.tb00491.x

Ibeagha-Awemu, E. M., Peters, S. O., Akwanji, K. A., Imumorin, I. G., & Zhao, X. (2016). High

density genome wide genotyping-by-sequencing and association identifies common and

low frequency SNPs, and novel candidate genes influencing cow milk traits. Scientific

Reports, 6(1), 31109. https://doi.org/10.1038/srep31109

Jamrozik, J., Koeck, A., Kistemaker, G. J., & Miglior, F. (2016). Multiple-trait estimates of

genetic parameters for metabolic disease traits, fertility disorders, and their predictors in

Canadian Holsteins. Journal of Dairy Science, 99(3), 1990–1998.

https://doi.org/10.3168/jds.2015-10505

Jiang, P., Fang, X., Zhao, Z., Yu, X., Sun, B., Yu, H., & Yang, R. (2018). The effect of

short/branched chain acyl-coenzyme A dehydrogenase gene on triglyceride synthesis of

bovine mammary epithelial cells. Archives Animal Breeding, 61(1), 115–122.

https://doi.org/10.5194/aab-61-115-2018

Jiao, Y., George, S. K., Zhao, Q., Hulver, M. W., Hutson, S. M., Bishop, C. E., & Lu, B. (2012).

Mex3c Mutation Reduces Adiposity and Increases Energy Expenditure. Molecular and

Cellular Biology, 32(21), 4350–4362. https://doi.org/10.1128/MCB.00452-12

Jiao, Yan, Bishop, C. E., & Lu, B. (2012). Mex3c regulates insulin-like growth factor 1 (IGF1)

expression and promotes postnatal growth. Molecular Biology of the Cell, 23(8), 1404–

1413. https://doi.org/10.1091/mbc.E11-11-0960

173

Kim, H.-Y., Byun, M.-J., & Kim, H.-B. (2011). A replication study of genome-wide CNV

association for hepatic biomarkers identifies nine genes associated with liver function.

BMB Reports, 44(9), 578–583. https://doi.org/10.5483/BMBRep.2011.44.9.578

Koeck, A., Jamrozik, J., Schenkel, F. S., Moore, R. K., Lefebvre, D. M., Kelton, D. F., &

Miglior, F. (2014). Genetic analysis of milk β-hydroxybutyrate and its association with

fat-to-protein ratio, body condition score, clinical ketosis, and displaced abomasum in

early first lactation of Canadian Holsteins. Journal of Dairy Science, 97(11), 7286–7292.

https://doi.org/10.3168/jds.2014-8405

Kroezen, V., Schenkel, F. S., Miglior, F., Baes, C. F., & Squires, E. J. (2018). Candidate gene

association analyses for ketosis resistance in Holsteins. Journal of Dairy Science, 101(6),

5240–5249. https://doi.org/10.3168/jds.2017-13374

Ku, C. S., Loy, E. Y., Pawitan, Y., & Chia, K. S. (2010). The pursuit of genome-wide association

studies: Where are we now? Journal of Human Genetics, 55, 195–206.

Kuniyoshi, K., Takeuchi, O., Pandey, S., Satoh, T., Iwasaki, H., Akira, S., & Kawai, T. (2014).

Pivotal role of RNA-binding E3 ubiquitin ligase MEX3C in RIG-I–mediated antiviral

innate immunity. Proceedings of the National Academy of Sciences, 111(15), 5646–5651.

https://doi.org/10.1073/pnas.1401674111

Lee, K.-B., Zhang, K., Folger, J. K., Knott, J. G., & Smith, G. W. (2014). Evidence Supporting a

Functional Requirement of SMAD4 for Bovine Preimplantation Embryonic

Development: A Potential Link to Embryotrophic Actions of Follistatin1. Biology of

Reproduction, 91(3). https://doi.org/10.1095/biolreprod.114.120105

Lee, S., Cho, K., Park, M., Choi, T., Kim, S., & Do, C. (2016). Genetic Parameters of Milk β-

Hydroxybutyric Acid and Acetone and Their Genetic Association with Milk Production

174

Traits of Holstein Cattle. Asian-Australasian Journal of Animal Sciences, 29(11), 1530–

1540. https://doi.org/10.5713/ajas.16.0310

Lee, S.-K., Son, L. T., Choi, H.-J., & Ahnn, J. (2013). Dicarbonyl/l-xylulose reductase (DCXR):

The multifunctional pentosuria enzyme. The International Journal of Biochemistry &

Cell Biology, 45(11), 2563–2567. https://doi.org/10.1016/j.biocel.2013.08.010

Li, C., Cai, W., Zhou, C., Yin, H., Zhang, Z., Loor, J. J., Sun, D., Zhang, Q., Liu, J., & Zhang, S.

(2016). RNA-Seq reveals 10 novel promising candidate genes affecting milk protein

concentration in the Chinese Holstein population. Scientific Reports, 6.

https://doi.org/10.1038/srep26813

Li, H., Xu, S., Gao, X., & Ren, H. (2007). Structure of the Bovine ACAD8 Gene and the

Association of Its Polymorphism with the Production Traits. Journal of Genetics and

Genomics, 34(4), 315–320. https://doi.org/10.1016/S1673-8527(07)60033-2

Liu, H. Y., & Zhang, C. J. (2017). Identification of differentially expressed genes and their

upstream regulators in colorectal cancer. Cancer Gene Therapy, 24(6), 244–250.

https://doi.org/10.1038/cgt.2017.8

Loor, J. J., Everts, R. E., Bionaz, M., Dann, H. M., Morin, D. E., Oliveira, R., Rodriguez-Zas, S.

L., Drackley, J. K., & Lewin, H. A. (2007). Nutrition-induced ketosis alters metabolic

and signaling gene networks in liver of periparturient dairy cows. Physiological

Genomics, 32(1), 105–116. https://doi.org/10.1152/physiolgenomics.00188.2007

Loor, J. J., Moyes, K. M., & Bionaz, M. (2011). Functional Adaptations of the Transcriptome to

Mastitis-Causing Pathogens: The Mammary Gland and Beyond. Journal of Mammary

Gland Biology and Neoplasia, 16(4), 305–322. https://doi.org/10.1007/s10911-011-9232-

175

Maeyashiki, C., Oshima, S., Otsubo, K., Kobayashi, M., Nibe, Y., Matsuzawa, Y., Onizawa, M.,

Nemoto, Y., Nagaishi, T., Okamoto, R., Tsuchiya, K., Nakamura, T., & Watanabe, M.

(2017). HADHA, the alpha subunit of the mitochondrial trifunctional protein, is involved

in long-chain fatty acid-induced autophagy in intestinal epithelial cells. Biochemical and

Biophysical Research Communications, 484(3), 636–641.

https://doi.org/10.1016/j.bbrc.2017.01.159

Maga, G., & Hubscher, U. (2003). Proliferating cell nuclear antigen (PCNA): A dancer with

many partners. Journal of Cell Science, 116(Pt 15), 3051–3060.

https://doi.org/10.1242/jcs.00653

Matsumoto, H., Inada, S., Kobayashi, E., Abe, T., Hasebe, H., Sasazaki, S., Oyama, K., &

Mannen, H. (2012). Identification of SNPs in the FASN gene and their effect on fatty

acid milk composition in Holstein cattle. Livestock Science, 144(3), 281–284.

https://doi.org/10.1016/j.livsci.2011.12.003

McArt, J. A. A., Nydam, D. V., & Oetzel, G. R. (2012). Epidemiology of subclinical ketosis in

early lactation dairy cattle. Journal of Dairy Science, 95(9), 5056–5066.

https://doi.org/10.3168/jds.2012-5443

McArt, J. A. A., Nydam, D. V., & Overton, M. W. (2015). Hyperketonemia in early lactation

dairy cattle: A deterministic estimate of component and total cost per case. Journal of

Dairy Science, 98(3), 2043–2054. https://doi.org/10.3168/jds.2014-8740

McCarthy, S. D., Waters, S. M., Kenny, D. A., Diskin, M. G., Fitzpatrick, R., Patton, J., Wathes,

D. C., & Morris, D. G. (2010). Negative energy balance and hepatic gene expression

patterns in high-yielding dairy cows during the early postpartum period: A global

176

approach. Physiological Genomics, 42A(3), 188–199.

https://doi.org/10.1152/physiolgenomics.00118.2010

Mills, S. E., Beitz, D. C., & Young, J. W. (1986). Evidence for Impaired Metabolism in Liver

during Induced Lactation Ketosis of Dairy Cows. Journal of Dairy Science, 69(2), 362–

370. https://doi.org/10.3168/jds.S0022-0302(86)80414-3

Misztal, I., Lourenco, D., Aguilar, I., Legarra, A., & Vitezica, Z. (2018). Manual for BLUPF90

family of programs. 142.

Moran, B., Butler, S. T., Moore, S. G., MacHugh, D. E., & Creevey, C. J. (2017). Differential

gene expression in the endometrium reveals cytoskeletal and immunological genes in

lactating dairy cows genetically divergent for fertility traits. Reproduction, Fertility and

Development, 29(2), 274–282. https://doi.org/10.1071/RD15128

Morris, C. A., Cullen, N. G., Glass, B. C., Hyndman, D. L., Manley, T. R., Hickey, S. M.,

McEwan, J. C., Pitchford, W. S., Bottema, C. D. K., & Lee, M. A. H. (2007). Fatty acid

synthase effects on bovine adipose fat and milk fat. Mammalian Genome, 18(1), 64–74.

https://doi.org/10.1007/s00335-006-0102-y

Morris, D. G., Waters, S. M., McCarthy, S. D., Patton, J., Earley, B., Fitzpatrick, R., Murphy, J.

J., Diskin, M. G., Kenny, D. A., Brass, A., & Wathes, D. C. (2009). Pleiotropic effects of

negative energy balance in the postpartum dairy cow on splenic gene expression:

Repercussions for innate and adaptive immunity. Physiological Genomics, 39(1), 28–37.

https://doi.org/10.1152/physiolgenomics.90394.2008

Mostowy, S., Bonazzi, M., Hamon, M. A., Tham, T. N., Mallet, A., Lelek, M., Gouin, E.,

Demangel, C., Brosch, R., Zimmer, C., Sartori, A., Kinoshita, M., Lecuit, M., & Cossart,

177

P. (2010). Entrapment of Intracytosolic Bacteria by Septin Cage-like Structures. Cell

Host & Microbe, 8(5), 433–444. https://doi.org/10.1016/j.chom.2010.10.009

Mostowy, S., & Cossart, P. (2012). Septins: The fourth component of the cytoskeleton. Nature

Reviews Molecular Cell Biology, 13(3), 183–194. https://doi.org/10.1038/nrm3284

Moyes, K. M., Drackley, J. K., Morin, D. E., Bionaz, M., Rodriguez-Zas, S. L., Everts, R. E.,

Lewin, H. A., & Loor, J. J. (2009). Gene network and pathway analysis of bovine

mammary tissue challenged with Streptococcus uberis reveals induction of cell

proliferation and inhibition of PPARγ signaling as potential mechanism for the negative

relationships between immune response and lipid metabolism. BMC Genomics, 10(1),

542. https://doi.org/10.1186/1471-2164-10-542

Nabokina, S. M., & Said, H. M. (2004). Characterization of the 5′-regulatory region of the

human thiamin transporter SLC19A3: In vitro and in vivo studies. American Journal of

Physiology-Gastrointestinal and Liver Physiology, 287(4), G822–G829.

https://doi.org/10.1152/ajpgi.00234.2004

Nayeri, S., Schenkel, F., Fleming, A., Kroezen, V., Sargolzaei, M., Baes, C., Cánovas, A.,

Squires, J., & Miglior, F. (2019). Genome-wide association analysis for β-

hydroxybutyrate concentration in Milk in Holstein dairy cattle. BMC Genetics, 20(1), 58.

https://doi.org/10.1186/s12863-019-0761-9

Nguyen, T. V., Andresen, B. S., Corydon, T. J., Ghisla, S., Abd-El Razik, N., Mohsen, A.-W. A.,

Cederbaum, S. D., Roe, D. S., Roe, C. R., Lench, N. J., & Vockley, J. (2002).

Identification of isobutyryl-CoA dehydrogenase and its deficiency in humans. Molecular

Genetics and Metabolism, 77(1–2), 68–79. https://doi.org/10.1016/S1096-

7192(02)00152-X

178

Palombo, V., Milanesi, M., Sgorlon, S., Capomaccio, S., Mele, M., Nicolazzi, E., Ajmone-

Marsan, P., Pilla, F., Stefanon, B., & D’Andrea, M. (2018). Genome-wide association

study of milk fatty acid composition in Italian Simmental and Italian Holstein cows using

single nucleotide polymorphism arrays. Journal of Dairy Science, 101(12), 11004–

11019. https://doi.org/10.3168/jds.2018-14413

Parker Gaddis, K. L., Megonigal, J. H., Clay, J. S., & Wolfe, C. W. (2018). Genome-wide

association study for ketosis in US Jerseys using producer-recorded data. Journal of

Dairy Science, 101(1), 413–424. https://doi.org/10.3168/jds.2017-13383

Patel, K. P., O’Brien, T. W., Subramony, S. H., Shuster, J., & Stacpoole, P. W. (2012). The

spectrum of pyruvate dehydrogenase complex deficiency: Clinical, biochemical and

genetic features in 371 patients. Molecular Genetics and Metabolism, 105(1), 34–43.

https://doi.org/10.1016/j.ymgme.2011.09.032

Paunesku, T., Mittal, S., Protić, M., Oryhon, J., Korolev, S. V., Joachimiak, A., & Woloschak, G.

E. (2001). Proliferating cell nuclear antigen (PCNA): Ringmaster of the genome.

International Journal of Radiation Biology, 77(10), 1007–1021.

https://doi.org/10.1080/09553000110069335

Pavlovic, Z., & Bakovic, M. (2013). Regulation of Phosphatidylethanolamine Homeostasis—

The Critical Role of CTP:Phosphoethanolamine Cytidylyltransferase (Pcyt2).

International Journal of Molecular Sciences, 14(2), 2529–2550.

https://doi.org/10.3390/ijms14022529

Pihlajamäki, J., Boes, T., Kim, E.-Y., Dearie, F., Kim, B. W., Schroeder, J., Mun, E., Nasser, I.,

Park, P. J., Bianco, A. C., Goldfine, A. B., & Patti, M. E. (2009). Thyroid Hormone-

179

Related Regulation of Gene Expression in Human Fatty Liver. The Journal of Clinical

Endocrinology & Metabolism, 94(9), 3521–3529. https://doi.org/10.1210/jc.2009-0212

Pongratz, R. L., Kibbey, R. G., Shulman, G. I., & Cline, G. W. (2007). Cytosolic and

Mitochondrial Malic Enzyme Isoforms Differentially Control Insulin Secretion. Journal

of Biological Chemistry, 282(1), 200–207. https://doi.org/10.1074/jbc.M602954200

Pryce, J. E., Parker Gaddis, K. L., Koeck, A., Bastin, C., Abdelsayed, M., Gengler, N., Miglior,

F., Heringstad, B., Egger-Danner, C., Stock, K. F., Bradley, A. J., & Cole, J. B. (2016).

Invited review: Opportunities for genetic improvement of metabolic diseases. Journal of

Dairy Science, 99(9), 6855–6873. https://doi.org/10.3168/jds.2016-10854

Ranaraja, U., Cho, K., Park, M., Kim, S., Lee, S., & Do, C. (2018). Genetic parameter estimation

for milk β-hydroxybutyrate and acetone in early lactation and its association with fat to

protein ratio and energy balance in Korean Holstein cattle. Asian-Australasian Journal of

Animal Sciences, 31(6), 798–803. https://doi.org/10.5713/ajas.17.0443

Raoux, M., Vacher, P., Papin, J., Picard, A., Kostrzewa, E., Devin, A., Gaitan, J., Limon, I., Kas,

M. J., Magnan, C., & Lang, J. (2015). Multilevel control of glucose homeostasis by

adenylyl cyclase 8. Diabetologia, 58(4), 749–757. https://doi.org/10.1007/s00125-014-

3445-z

Richert, L., Liguori, M. J., Abadie, C., Heyd, B., Mantion, G., Halkic, N., & Waring, J. F.

(2006). Gene expression in human hepatocytes in suspension after isolation is similar to

the liver of origin, is not affected by hepatocyte cold storage and cryopreservation, but is

strongly changed after hepatocyte plating. Drug Metabolism and Disposition: The

Biological Fate of Chemicals, 34(5), 870–879. https://doi.org/10.1124/dmd.105.007708

180

Rinner, B., Gandolfi, G., Meditz, K., Frisch, M.-T., Wagner, K., Ciarrocchi, A., Torricelli, F.,

Koivuniemi, R., Niklander, J., Liegl-Atzwanger, B., Lohberger, B., Heitzer, E., Ghaffari-

Tabrizi-Wizsy, N., Zweytick, D., & Zalaudek, I. (2017). MUG-Mel2, a novel highly

pigmented and well characterized NRAS mutated human melanoma cell line. Scientific

Reports, 7. https://doi.org/10.1038/s41598-017-02197-y

Roe, C. R., Cederbaum, S. D., Roe, D. S., Mardach, R., Galindo, A., & Sweetman, L. (1998).

Isolated isobutyryl-CoA dehydrogenase deficiency: An unrecognized defect in human

valine metabolism. Molecular Genetics and Metabolism, 65(4), 264–271.

https://doi.org/10.1006/mgme.1998.2758

Rokka, A., Antonenkov, V. D., Soininen, R., Immonen, H. L., Pirilä, P. L., Bergmann, U.,

Sormunen, R. T., Weckström, M., Benz, R., & Hiltunen, J. K. (2009). Pxmp2 Is a

Channel-Forming Protein in Mammalian Peroxisomal Membrane. PLoS ONE, 4(4).

https://doi.org/10.1371/journal.pone.0005090

Rosenbaum, S., Ringseis, R., Hillen, S., Becker, S., Erhardt, G., Reiner, G., & Eder, K. (2012).

Genome-wide transcript profiling indicates induction of energy-generating pathways and

an adaptive immune response in the liver of sows during lactation. Comparative

Biochemistry and Physiology. Part D, Genomics & Proteomics, 7(4), 370–381.

https://doi.org/10.1016/j.cbd.2012.09.001

Sargolzaei, M., Chesnais, J. P., & Schenkel, F. S. (2014). A new approach for efficient genotype

imputation using information from relatives. BMC Genomics, 15(1), 478.

https://doi.org/10.1186/1471-2164-15-478

Schäff, C., Börner, S., Hacke, S., Kautzsch, U., Sauerwein, H., Spachmann, S. K., Schweigel-

Röntgen, M., Hammon, H. M., & Kuhla, B. (2013). Increased muscle fatty acid oxidation

181

in dairy cows with intensive body fat mobilization during early lactation. Journal of

Dairy Science, 96(10), 6449–6460. https://doi.org/10.3168/jds.2013-6812

Schroeder, B. C., Hechenberger, M., Weinreich, F., Kubisch, C., & Jentsch, T. J. (2000). KCNQ5, a novel potassium channel broadly expressed in brain, mediates M-type currents. Journal of Biological Chemistry, 275(31), 24089-24095. Shia, W.-C., Ku, T.-H., Tsao, Y.-M., Hsia, C.-H., Chang, Y.-M., Huang, C.-H., Chung, Y.-C.,

Hsu, S.-L., Liang, K.-W., & Hsu, F.-R. (2011). Genetic copy number variants in

myocardial infarction patients with hyperlipidemia. BMC Genomics, 12(Suppl 3), S23.

https://doi.org/10.1186/1471-2164-12-S3-S23

Sisodiya, S. M., Thompson, P. J., Need, A., Harris, S. E., Weale, M. E., Wilkie, S. E., ... & Gerrelli, D. (2007). Genetic enhancement of cognition in a kindred with cone–rod dystrophy due to RIMS1 mutation. Journal of medical genetics, 44(6), 373-380. Soupene, E., & Kuypers, F. A. (2008). Mammalian Long-Chain Acyl-CoA Synthetases.

Experimental Biology and Medicine (Maywood, N.j.), 233(5), 507–521.

https://doi.org/10.3181/0710-MR-287

Sperl, W., Fleuren, L., Freisinger, P., Haack, T. B., Ribes, A., Feichtinger, R. G., Rodenburg, R.

J., Zimmermann, F. A., Koch, J., Rivera, I., Prokisch, H., Smeitink, J. A., & Mayr, J. A.

(2015). The spectrum of pyruvate oxidation defects in the diagnosis of mitochondrial

disorders. Journal of Inherited Metabolic Disease, 38(3), 391–403.

https://doi.org/10.1007/s10545-014-9787-3

Sun, H.-Z., Wang, D.-M., Liu, H.-Y., & Liu, J.-X. (2018). Metabolomics Integrated with

Transcriptomics Reveals a Subtle Liver Metabolic Risk in Dairy Cows Fed Different

Crop By-products. PROTEOMICS, 18(16), 1800122.

https://doi.org/10.1002/pmic.201800122

182

Sun, L., Ye, Y., Sun, H., Yu, J., Zhang, L., Sun, Y., Zhang, D., Ma, L., Shen, B., & Zhu, C.

(2013). Identification of Proteasome Subunit Beta Type 6 (PSMB6) Associated with

Deltamethrin Resistance in Mosquitoes by Proteomic and Bioassay Analyses. PLoS ONE,

8(6). https://doi.org/10.1371/journal.pone.0065859

Tai, E. S., Tan, M. L. S., Stevens, R. D., Low, Y. L., Muehlbauer, M. J., Goh, D. L. M.,

Ilkayeva, O. R., Wenner, B. R., Bain, J. R., Lee, J. J. M., Lim, S. C., Khoo, C. M., Shah,

S. H., & Newgard, C. B. (2010). Insulin resistance is associated with a metabolic profile

of altered protein metabolism in Chinese and Asian-Indian men. Diabetologia, 53(4),

757–767. https://doi.org/10.1007/s00125-009-1637-8

Tatone, E. H., Duffield, T. F., LeBlanc, S. J., DeVries, T. J., & Gordon, J. L. (2017).

Investigating the within-herd prevalence and risk factors for ketosis in dairy cattle in

Ontario as diagnosed by the test-day concentration of β-hydroxybutyrate in milk. Journal

of Dairy Science, 100(2), 1308–1318. https://doi.org/10.3168/jds.2016-11453

Timperio, A. M., D’Alessandro, A., Pariset, L., D’Amici, G. M., Valentini, A., & Zolla, L.

(2009). Comparative proteomics and transcriptomics analyses of livers from two different

Bos taurus breeds: “Chianina and Holstein Friesian.” Journal of Proteomics, 73(2), 309–

322. https://doi.org/10.1016/j.jprot.2009.09.015

Uhlén, M., Fagerberg, L., Hallström, B. M., Lindskog, C., Oksvold, P., Mardinoglu, A.,

Sivertsson, Å., Kampf, C., Sjöstedt, E., Asplund, A., Olsson, I., Edlund, K., Lundberg, E.,

Navani, S., Szigyarto, C. A.-K., Odeberg, J., Djureinovic, D., Takanen, J. O., Hober, S.,

… Pontén, F. (2015). Tissue-based map of the human proteome. Science, 347(6220).

https://doi.org/10.1126/science.1260419

183

Vahey, M. T., Wang, Z., Kester, K. E., Cummings, J., Heppner, Jr, D. G., Nau, M. E., Ofori‐

Anyinam, O., Cohen, J., Coche, T., Ballou, W. R., & Ockenhouse, C. F. (2010).

Expression of Genes Associated with Immunoproteasome Processing of Major

Histocompatibility Complex Peptides Is Indicative of Protection with Adjuvanted RTS,S

Malaria Vaccine. The Journal of Infectious Diseases, 201(4), 580–589.

https://doi.org/10.1086/650310

Vapola, M. H., Rokka, A., Sormunen, R. T., Alhonen, L., Schmitz, W., Conzelmann, E., Wärri,

A., Grunau, S., Antonenkov, V. D., & Hiltunen, J. K. (2014). Peroxisomal membrane

channel Pxmp2 in the mammary fat pad is essential for stromal lipid homeostasis and for

development of mammary gland epithelium in mice. Developmental Biology, 391(1), 66–

80. https://doi.org/10.1016/j.ydbio.2014.03.022

Veerkamp, R. F., Oldenbroek, J. K., Van Der Gaast, H. J., & Werf, J. H. J. V. D. (2000). Genetic

Correlation Between Days Until Start of Luteal Activity and Milk Yield, Energy Balance,

and Live Weights. Journal of Dairy Science, 83(3), 577–583.

https://doi.org/10.3168/jds.S0022-0302(00)74917-4

Vlasova, T. I., Stratton, S. L., Wells, A. M., Mock, N. I., & Mock, D. M. (2005). Biotin

Deficiency Reduces Expression of SLC19A3, a Potential Biotin Transporter, in

Leukocytes from Human Blood. The Journal of Nutrition, 135(1), 42–47.

Wang, H., Misztal, I., Aguilar, I., Legarra, A., & Muir, W. M. (2012). Genome-wide association

mapping including phenotypes from relatives without genotypes. Genetics Research,

94(2), 73–83. https://doi.org/10.1017/S0016672312000274

Weaver, S. R., Prichard, A. S., Maerz, N. L., Prichard, A. P., Endres, E. L., Hernández-

Castellano, L. E., Akins, M. S., Bruckmaier, R. M., & Hernandez, L. L. (2017). Elevating

184

serotonin pre-partum alters the Holstein dairy cow hepatic adaptation to lactation. PLoS

ONE, 12(9). https://doi.org/10.1371/journal.pone.0184939

Wu, X., Wang, Y., & Sun, Y. (2011). Molecular characterization, expression analysis and

association study with immune traits of porcine PSMB6 gene. Molecular Biology

Reports, 38(8), 5465–5470. https://doi.org/10.1007/s11033-011-0866-3

Wu, Y., Bu, F., Yu, H., Li, W., Huang, C., Meng, X., Zhang, L., Ma, T., & Li, J. (2017).

Methylation of Septin9 mediated by DNMT3a enhances hepatic stellate cells activation

and liver fibrogenesis. Toxicology and Applied Pharmacology, 315, 35–49.

https://doi.org/10.1016/j.taap.2016.12.002

Xu, C., Wang, Z., Liu, G., Li, X., Xie, G., Xia, C., & Zhang, H. you. (2008). Metabolic

Characteristic of the Liver of Dairy Cows during Ketosis Based on Comparative

Proteomics. Asian-Australasian Journal of Animal Sciences, 21(7), 1003–1010.

https://doi.org/10.5713/ajas.2008.70392

Yao, H., & Ye, J. (2008). Long chain acyl-CoA synthetase 3-mediated phosphatidylcholine

synthesis is required for assembly of very low density lipoproteins in human hepatoma

Huh7 cells. The Journal of Biological Chemistry, 283(2), 849–854.

https://doi.org/10.1074/jbc.M706160200

Zeng, T., Zhu, L., Liao, M., Zhuo, W., Yang, S., Wu, W., & Wang, D. (2017). Knockdown of

PYCR1 inhibits cell proliferation and colony formation via cell cycle arrest and apoptosis

in prostate cancer. Medical Oncology (Northwood, London, England), 34(2), 27.

https://doi.org/10.1007/s12032-016-0870-5

Zeng, W.-Q., Al-Yamani, E., Acierno, Jr., J. S., Slaugenhaupt, S., Gillis, T., MacDonald, M. E.,

Ozand, P. T., & Gusella, J. F. (2005). Biotin-Responsive Basal Ganglia Disease Maps to

185

2q36.3 and Is Due to Mutations in SLC19A3. American Journal of Human Genetics,

77(1), 16–26.

Zhao, F., Wei, C., Zhang, L., Liu, J., Wang, G., Zeng, T., & Du, L. (2016). A genome scan of

recent positive selection signatures in three sheep populations. Journal of Integrative

Agriculture, 15(1), 162–174. https://doi.org/10.1016/S2095-3119(15)61080-2

Zhou, S., Buckhaults, P., Zawel, L., Bunz, F., Riggins, G., Le Dai, J., Kern, S. E., Kinzler, K.

W., & Vogelstein, B. (1998). Targeted deletion of Smad4 shows it is required for

transforming growth factor β and activin signaling in colorectal cancer cells. Proceedings

of the National Academy of Sciences of the United States of America, 95(5), 2412–2416.

Zhu, Y., Liu, G., Du, X., Shi, Z., Jin, M., Sha, X., Li, X., Wang, Z., & Li, X. (2019). Expression

patterns of hepatic genes involved in lipid metabolism in cows with subclinical or clinical

ketosis. Journal of Dairy Science, 102(2), 1725–1735. https://doi.org/10.3168/jds.2018-

14965

Zhuang, J., Song, Y., Ye, Y., He, S., Ma, X., Zhang, M., Ni, J., Wang, J., & Xia, W. (2019).

PYCR1 interference inhibits cell growth and survival via c-Jun N-terminal kinase/insulin

receptor substrate 1 (JNK/IRS1) pathway in hepatocellular cancer. Journal of

Translational Medicine, 17. https://doi.org/10.1186/s12967-019-2091-0

Tables and Figures

Table 1. Variance components and heritability estimates for SCK1.1, SCK1.2, SCK2.1 and SCK2.2 traits in Holstein cattle. Variance Components2 Traits1 Heritability (h2) 2 2 흈 a 흈 e SCK1.1 0.0529 0.2020 0.194

186

SCK1.2 0.0478 0.1729 0.202 SCK2.1 0.0402 0.2138 0.149 SCK2.2 0.0398 0.2385 0.143 1 SCK1.1 = Subclinical ketosis in first lactation from 5 to 21 DIM; SCK1.2 = Subclinical ketosis in first lactation from 22 to 45 DIM; SCK2.1 = Subclinical ketosis in later lactations from 5 to 21 DIM; SCK2.2 = Subclinical ketosis in later lactations from 22 to 45 DIM. 2 2 2 흈 a: additive genetic variance, 흈 e: residual variance

Table 2. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis for first lactation from 5 to 21 DIM (SCK1.1). Chr Position (bp) Var (%) 17 43794669:44564877 3.53 16 65783105:66583504 2.06 11 93586086:95028686 1.93 13 55867844:56453556 1.58 1 118667764:118717516 0.97 3 110001115:110693717 0.94 2 123446559:124279927 0.89 12 59129931:60463147 0.87 11 92317695:93302992 0.84 6 47776901:47939531 0.77 25 40419340:41015047 0.76 28 42572614:43308248 0.75 15 46647993:47771121 0.72 1 26719210:27781845 0.69 2 111092245:111491176 0.63 2 115355532:115875702 0.60

187

3 92457948:92955100 0.59 4 40575998:41925904 0.58 8 64766892:66234558 0.57 19 48855388:49594540 0.55 Chr: chromosome; Position (bp): starting and ending coordinates of the window; Var: % genetic variance explained by the SNPs within the window.

Table 3. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis for first lactation from 22 to 45 DIM (SCK1.2). Chr Position (bp) Var (%) 16 76114559:76866839 1.60 9 86155276:86962793 1.51 26 42329789:43149049 1.42 24 50175324:50983655 1.33 1 83857598:84911107 1.06 20 55277748:55926147 0.96 29 49697006:50182655 0.94 16 42293112:43231366 0.91 5 66493596:66806781 0.89 11 73256269:73523402 0.87 12 57881493:58857555 0.81 8 48532734:49616613 0.81 2 111111674:111658962 0.66 20 45897606:46474260 0.65 10 42491695:43821719 0.63 1 122562289:123068200 0.61 1 133833175:134802487 0.59 3 93380590:94226523 0.58

188

2 125511462:126222480 0.58 15 64914539:65678820 0.52 Chr: chromosome; Position (bp): starting and ending coordinates of the window; Var: % genetic variance explained by the SNPs within the window.

Table 4. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis for later lactations (2 to 5) from 5 to 21 DIM (SCK2.1). Chr Position (bp) Var (%) 9 11652908:12571145 3.10 1 17263605:18512267 1.43 1 115422912:115822652 1.42 24 50173729:50833720 1.25 15 43476234:43858386 1.19 5 91757353:93138660 1.17 9 12621882:13548524 1.17 19 49744592:50605756 0.86 6 86465400:87369626 0.74 6 100180758:100589251 0.73 22 32601141:33572970 0.68 19 57189095:57596864 0.66 22 24392724:25555193 0.62 9 10782504:11496361 0.57 19 54230885:54701019 0.56 5 109331643:109808607 0.55 14 8996524:9624604 0.55 19 50691143:51263493 0.55 9 51310266:51935099 0.55 23 17944024:19204806 0.50

189

Chr: chromosome; Position (bp): starting and ending coordinates of the window; Var: % genetic variance explained by the SNPs within the window.

Table 5. Top-20 sliding windows explaining the highest proportion of genetic variance for subclinical ketosis for later lactations (2 to 5) from 22 to 45 DIM (SCK2.2). Chr Position (bp) Var (%) 20 36496636:36932715 1.53 1 109341277:110076693 1.31 17 62985424:63518156 1.15 5 114222945:115346239 1.10 1 17235751:18439746 1.05 4 73571440:74539360 1.05 20 8771290:9539935 0.98 5 91646669:92415826 0.98 22 31415213:31988767 0.87 12 33603872:34072962 0.80 14 9427743:10036916 0.76 11 71130818:71946048 0.65 19 54230885:54701019 0.64 15 83207058:83864395 0.62 27 44266304:45240208 0.59 5 92527188:93441439 0.59 9 76209166:77454491 0.55 17 11106946:12127579 0.52 24 50173729:50833720 0.52 17 33039443:34015241 0.50 Chr: chromosome; Position (bp): starting and ending coordinates of the window; Var: % genetic variance explained by the SNPs within the window.

190

Figure 1. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis for first lactation from 5 to 21 DIM (SCK1.1). Chromosome number is shown on the horizontal axis.

Figure 2. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis for first lactation from 22 to 45 DIM (SCK1.2). Chromosome number is shown on the horizontal axis.

191

Figure 3. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis for later lactations (2 to 5) from 5 to 21 DIM (SCK2.1). Chromosome number is shown on the horizontal axis.

Figure 4. Genetic variance explained by 20-SNP sliding windows for subclinical ketosis for later lactations (2 to 5) from 22 to 45 DIM (SCK2.2). Chromosome number is shown on the horizontal axis.

192

Figure 5. Main metabolic pathways associated to SCK1.1, SCK1.2, SCK2.1 and SCK2.2.

Table 6. Annotated genes within the top-20 sliding windows associated with subclinical ketosis obtained from the wssGBLUP analyses.

Traits Chr Position (bp) Genes

17 43794669:44564877 ZNF605, ZNF26, ZNF84, ZNF140, ZNF891, ZNF10, ZNF268, MBD3L1, ANHX, CHFR, GOLGA3, ANKLE2, PGAM5, PXMP2, POLE, U6

16 65783105:66583504 EDEM3, NIBAN1, RNF2, TRMT1L, SWT1, IVNS1ABP

11 93586086:95028686 OR1J2, OR1N2, OR1N1, OR1Q1, OR1B1, OR1L1, OR1L3, OR5C1, OR1K1, PDCL, RC3H2, SNORD90, ZBTB6, ZBTB26, RABGAP1, GPR21, SCK1.1 STRBP, U6, CRB2, DENND1A

13 55867844:56453556 -

1 118667764:118717516 TM4SF4

3 110001115:110693717 CLSPN, C3H1orf216, PSMB2, TFAP2E, NCDN, KIAA0319L, ZMYM4, SFPQ, ZMYM1, ZMYM6

2 123446559:124279927 PTPRU

12 59129931:60463147 U6

193

11 92317695:93302992 GGTA1, DAB2IP, NDUFA8, MORN5, LHX6, RBM18, MRRF, PTGS1, bta- mir-10175

6 47776901:47939531 -

25 40419340:41015047 CARD11, GNA12, AMZ1, BRAT1, bta-mir-11980, IQCE, TTYH3, LFNG, bta-mir-12029, GRIFIN, CHST12, bta-mir-12019, EIF3B, SNX8, NUDT1, MRM2, MAD1L1

28 42572614:43308248 FRMPD2, MAPK8, ARHGAP22, WDFY4, LRRC18, VSTM4

15 46647993:47771121 APBB1, SMPD1, bta-mir-2316, CAVIN3, CCKBR, CNGA4, FAM160A2, C15H11orf42, OR52W1, OR52B2, OR56A1, OR56A5, OR52E4, U6, OR52N1

1 26719210:27781845 -

2 111092245:111491176 ACSL3, KCNE4, U6

2 115355532:115875702 COL4A3, MFF, TM4SF20, AGFG1, bta-mir-2285ao-3, SLC19A3

3 92457948:92955100 GLIS1, DMRTB1, LRP8

4 40575998:41925904 -

8 64766892:66234558 NR4A3, STX17, ERP44, INVS, TEX10, CNTNAP3, 7SK

19 48855388:49594540 SMURF2, KPNA2, C17orf58, BPTF, NOL11, U6, PSMD12

16 76114559:76866839 ZBTB41, CRB1, bta-mir-2284n, DENND1B, C1orf53, LHX9

9 86155276:86962793 UST, TAB2, ZC3H12D, PPIL4, GINM1, KATNA1, LATS1, NUP43, PCMT1

SCK1.2 26 42329789:43149049 HTRA1, bta-mir-10225b, bta-mir- 10225a, SPADH2, SPADH1, C26H10orf120, PSTK, IKZF5, ACADSB, U6, HMX3, HMX2, BUB3, U6

24 50175324:50983655 MAPK4, MRO, ME2, ELAC1, SMAD4, U1, MEX3C, U6

194

1 83857598:84911107 MCF2L2, LAMP3, MCCC1, DCUN1D1, ATP11B

20 55277748:55926147 BASP1

29 49697006:50182655 CTSD, IFITM10, DUSP8, MOB2, TOLLIP

16 42293112:43231366 UBIAD1, MTOR, ANGPTL7, EXOSC10, SRM, MASP2, TARDBP, CASZ1, PEX14, DFFA, CORT, CENPS, PGD

5 66493596:66806781 U6, PAH, ASCL1, U1

11 73256269:73523402 HADHB, HADHA, GAREM2, RAB10, KIF3C

12 57881493:58857555 SLITRK1

8 48532734:49616613 ZFAND5, TMC1, ALDH1A1, ANXA1

2 111111674:111658962 ACSL3, KCNE4, U6

20 45897606:46474260 -

10 42491695:43821719 RPS29, LRR1, RPL36AL, MGAT2, DNAAF2, bta-mir-6517, POLE2, KLHDC1, KLHDC2, NEMF, ARF6, bta-mir-10171, VCPKMT, SOS2, U6, L2HGDH, U1, DMAC2L, CDKL1, MAP4K5, ATL1, SAV1, NIN, ABHD12B, PYGL

1 122562289:123068200 -

1 133833175:134802487 U6, EPHB1, KY, CEP63

3 93380590:94226523 SCP2, ECHDC2, ZYG11A, ZYG11B, COA7, SHISAL2A, GPX7, TUT4, PRPF38A, ORC1, bta-mir-2285bc, CC2D1B, ZFYVE9

2 125511462:126222480 PPP1R8, SCARNA1, STX12, FAM76A, IFI6, FGR, AHDC1, WASF2, U6, GPR3, CD164L2, MAP3K6, SYTL1, TMEM222, WDTC1, SLC9A1, TENT5B

15 64914539:65678820 ABTB2, CAT, ELF5, EHF, APIP, PDHX, CD44

SCK2.1 9 11652908:12571145 RIMS1, KCNQ5

195

1 17263605:18512267 U6

1 115422912:115822652 MBNL1

24 50173729:50833720 MAPK4, MRO, ME2, ELAC1, SMAD4, U1, MEX3C

15 43476234:43858386 SCUBE2, NRIP3, U3, TMEM9B, ASCL3, C15H11orf16, AKIP1, DENND2B

5 91757353:93138660 PIK3C2G, RERGL

9 12621882:13548524 KCNQ5, KHDC3L, OOEP, CGAS, MTO1, EEF1A1, SLC17A5, U6, CD109, U6

19 49744592:50605756 B3GNTL1, TBCD, bta-mir-2344, ZNF750, SNORA73, FN3KRP, RAB40B, WDR45B, FOXK2, NARF, CYBC1, HEXD, OGFOD3, SECTM1A, bta-mir-2345, SECTM1, bta-mir-2345, CD7, CSNK1D

6 86465400:87369626 SLC4A4, GC, NPFFR2

6 100180758:100589251 -

22 32601141:33572970 TAFA4, TAFA1, U6

19 57189095:57596864 RPL38

22 24392724:25555193 CNTN6, U6

9 10782504:11496361 RIMS1

19 54230885:54701019 SEPTIN9

5 109331643:109808607 PEX26, U6, TUBA8, CDC42EP1, LGALS2, GGA1, SH3BP1, PDXP, LGALS1, NOL12, bta-mir-2438, TRIOBP, bta-mir-2285ca, GCAT, GALR3, ANKRD54, bta-mir-12001, bta-mir-658, EIF3L, MICALL1, C5H22orf23, POLR2F, SOX10

14 8996524:9624604 KCNQ3, HHLA1, EFR3A

19 50691143:51263493 CCDC57, FASN, DUS1L, GPS1, RFNG, RAC3, DCXR, LRRC45, CENPX, ASPSCR1, bta-mir-2346, NOTUM, MYADML2, PYCR1, MAFG, SIRT7, PCYT2, bta-mir-2347, NPB, ANAPC11, ALYREF,

196

ARHGDIA, P4HB, PPP1R27, MCRIP1, GCGR, MRPL12, HGS, ARL16, CCDC137, OXLD1, PDE6G, TSPAN10, NPLOC4, FAAP100, FSCN2, ACTG1, bta-mir-3533

9 51310266:51935099 -

23 17944024:19204806 SUPT3H, RUNX2, CLIC5

20 36496636:36932715 U6, GDNF, WDR70

1 109341277:110076693 RSRC1, SHOX2, 7SK, VEPH1

17 62985424:63518156 ACADS, SPPL3, HNF1A, C17H12orf43, OASL, ANKRD13A, GIT2, TCHP, GLTP, TRPV4, FAM222A

5 114222945:115346239 MPPED1, EFCAB6, SULT4A1, SAMM50, PARVB, PARVG, SHISAL1, RTL6, PRR5

1 17235751:18439746 U6

4 73571440:74539360 STEAP1, STEAP2, CFAP69, FAM237B, GTPBP10

20 8771290:9539935 TNPO1, ZNF366, PTCD2, MRPS27, MAP1B

SCK2.2 5 91646669:92415826 PIK3C2G, RERGL

22 31415213:31988767 MITF

12 33603872:34072962 ATP8A2, U6, NUP58, MTMR6, AMER2

14 9427743:10036916 ADCY8

11 71130818:71946048 PLB1, FOSL2, BABAM2, RBKS, MRPL33

19 54230885:54701019 SEPTIN9

15 83207058:83864395 MS4A7, MS4A14, MS4A5, MS4A1, MS4A13, bta-mir-6519, NCAPD3, VPS26B, THYN1, ACAD8, GLB1L3, GLB1L2, B3GAT1

27 44266304:45240208 ZNF385D

5 92527188:93441439 LMO3

197

9 76209166:77454491 NHSL1, CCDC28A, ECT2L, REPS1, ABRACL, HECA, TXLNB, CITED2, U5

17 11106946:12127579 TTC29, POU4F2, 7SK, SLC10A7, REELD1

24 50173729:50833720 MAPK4, MRO, ME2, ELAC1, SMAD4, U1, MEX3C

17 33039443:34015241 ANKRD50 Chr: chromosome; Position (bp): starting and ending coordinates of the window; Genes: ENSEMBL symbol of annotated genes using the Bos taurus ARS-UCD1.2 assembly.

Supplementary material

Supplementary Table 1. Enriched Molecular Function, Cellular component, and Biological process with Fisher's test for subclinical ketosis for first lactation from 5 to 21 DIM (SCK1.1).

GO Terms Enriched process Description P-value

GO:0007600 Biological Process sensory perception 0.003014215

GO:0051606 Biological Process detection of stimulus 0.007291555

GO:0006979 Biological Process response to oxidative stress 0.035852626

GO:0061053 Biological Process somite development 0.038420303

GO:0098732 Biological Process macromolecule deacylation 0.047178093

GO:0099023 Cellular Component tethering complex 0.021135654

GO:0005793 Cellular Component endoplasmic reticulum-Golgi intermediate 0.034329376 compartment

GO:0031300 Cellular Component intrinsic component of organelle 0.04588196 membrane

GO:0004984 Molecular Function olfactory receptor activity 0.000890325

GO:0042826 Molecular Function histone deacetylase binding 0.003607858

P-value<0.05

Supplementary Table 2. Enriched Molecular Function, Cellular component, and Biological process with Fisher's test for subclinical ketosis for first lactation from 22 to 45 DIM (SCK1.2).

GO Terms Enriched process Description P-value

GO:0030855 Biological Process epithelial cell differentiation 0.000853049

198

GO:0051259 Biological Process protein complex oligomerization 0.001891055

GO:0048732 Biological Process gland development 0.003128574

GO:0032787 Biological Process monocarboxylic acid metabolic process 0.003468634

GO:0043588 Biological Process skin development 0.00751176

GO:0045165 Biological Process cell fate commitment 0.011451648

GO:0008544 Biological Process epidermis development 0.012060613

GO:0060191 Biological Process regulation of lipase activity 0.022008114

GO:0008213 Biological Process protein alkylation 0.028504125

GO:0030029 Biological Process actin filament-based process 0.035794216

GO:0061008 Biological Process hepaticobiliary system development 0.040104341

GO:0051321 Biological Process meiotic cell cycle 0.040884606

GO:0001667 Biological Process ameboidal-type cell migration 0.042663349

GO:0061458 Biological Process reproductive system development 0.042663349

GO:0043086 Biological Process negative regulation of catalytic activity 0.044125524

GO:0032989 Biological Process cellular component morphogenesis 0.045253126

GO:0051186 Biological Process cofactor metabolic process 0.045347575

GO:0022406 Biological Process membrane docking 0.046499139

GO:0042303 Biological Process molting cycle 0.046499139

GO:0044770 Biological Process cell cycle phase transition 0.046802482

GO:0015850 Biological Process organic hydroxy compound transport 0.047453829

GO:0042579 Cellular Component microbody 0.011257029

GO:0098687 Cellular Component chromosomal region 0.019623309

GO:0016614 Molecular Function oxidoreductase activity 0.004938145

GO:0048306 Molecular Function calcium-dependent protein binding 0.018675769

GO:0016903 Molecular Function oxidoreductase activity 0.019917162

GO:0016684 Molecular Function oxidoreductase activity 0.029521763

GO:0048037 Molecular Function cofactor binding 0.035942561

GO:1990837 Molecular Function sequence-specific double-stranded DNA 0.044303386 binding

GO:0047485 Molecular Function protein N-terminus binding 0.047530559

P-value<0.05

199

Supplementary Table 3. Enriched Molecular Function, Cellular component, and Biological process with Fisher's test for subclinical ketosis for later lactations (2 to 5) from 5 to 21 DIM (SCK2.1).

GO Terms Enriched process Description P-value

GO:0097435 Biological Process supramolecular fiber organization 0.006679588

GO:0072524 Biological Process pyridine-containing compound metabolic 0.011604105 process

GO:0000959 Biological Process mitochondrial RNA metabolic process 0.015717264

GO:0051493 Biological Process regulation of cytoskeleton organization 0.020931706

GO:0030029 Biological Process actin filament-based process 0.024714244

GO:0006766 Biological Process vitamin metabolic process 0.027866216

GO:0010638 Biological Process positive regulation of organelle 0.037017545 organization

GO:0008544 Biological Process epidermis development 0.048551804

GO:0099080 Cellular Component supramolecular complex 0.048714305

GO:0016614 Molecular Function oxidoreductase activity 0.004352978

GO:0019842 Molecular Function vitamin binding 0.004688778

GO:0019001 Molecular Function guanyl nucleotide binding 0.013555534

GO:0003924 Molecular Function GTPase activity 0.018770301

GO:0048037 Molecular Function cofactor binding 0.030922647

GO:0001882 Molecular Function nucleoside binding 0.043864929

P-value<0.05

Supplementary Table 4. Enriched Molecular Function, Cellular component, and Biological process with Fisher's test for subclinical ketosis for later lactations (2 to 5) from 22 to 45 DIM (SCK2.2).

GO Terms Enriched process Description P-value

GO:0061008 Biological Process hepaticobiliary system development 0.001233848

GO:0030522 Biological Process intracellular receptor signaling pathway 0.002199689

GO:0048729 Biological Process tissue morphogenesis 0.002330266

GO:0007498 Biological Process mesoderm development 0.002873308

200

GO:0003682 Molecular Function chromatin binding 0.003122603

GO:0061448 Biological Process connective tissue development 0.003437503

GO:0060485 Biological Process mesenchyme development 0.003905243

GO:0007507 Biological Process heart development 0.004562933

GO:0043627 Biological Process response to estrogen 0.005232661

GO:0009887 Biological Process animal organ morphogenesis 0.005783163

GO:0000785 Cellular Component Chromatin 0.00637377

GO:0048736 Biological Process appendage development 0.006889036

GO:0007422 Biological Process peripheral nervous system development 0.008640569

GO:0016627 Molecular Function oxidoreductase activity 0.011266515

GO:0060537 Biological Process muscle tissue development 0.01132791

GO:0033500 Biological Process carbohydrate homeostasis 0.011586793

GO:0048863 Biological Process stem cell differentiation 0.011890356

GO:0048598 Biological Process embryonic morphogenesis 0.012027729

GO:0001763 Biological Process morphogenesis of a branching structure 0.012198494

GO:0046332 Molecular Function SMAD binding 0.012823567

GO:0003712 Molecular Function transcription coregulator activity 0.012968787

GO:0048732 Biological Process gland development 0.015762334

GO:0010817 Biological Process regulation of hormone levels 0.017804689

GO:0033218 Molecular Function amide binding 0.020352289

GO:0045444 Biological Process fat cell differentiation 0.020569304

GO:0098876 Biological Process vesicle-mediated transport to the plasma 0.021559348 membrane

GO:0009725 Biological Process response to hormone 0.021817142

GO:0001501 Biological Process skeletal system development 0.0249003

GO:0000228 Cellular Component nuclear chromosome 0.030695204

GO:0140030 Molecular Function modification-dependent protein binding 0.032144753

GO:0009743 Biological Process response to carbohydrate 0.032163318

GO:0006325 Biological Process chromatin organization 0.03233375

GO:0044815 Cellular Component DNA packaging complex 0.032505737

GO:0001655 Biological Process urogenital system development 0.033042123

201

GO:0005667 Cellular Component transcription factor complex 0.040370933

GO:0061061 Biological Process muscle structure development 0.040981252

GO:0042330 Biological Process Taxis 0.041866429

GO:0035270 Biological Process endocrine system development 0.042258386

GO:0008643 Biological Process carbohydrate transport 0.043325918

GO:0051169 Biological Process nuclear transport 0.043539802

GO:0045597 Biological Process positive regulation of cell differentiation 0.044493264

GO:0032940 Biological Process secretion by cell 0.046375135

GO:0048017 Biological Process inositol lipid-mediated signaling 0.049939503

P-value<0.05

Supplementary Table 5. Pathways obtained for subclinical ketosis for first lactation from 5 to 21 DIM (SCK1.1).

Trait Pathways

SCK1.1 Regulation of lipolysis in adipocytes

Adipocytokine signaling pathway

Peroxisome

Fatty acid biosynthesis

cAMP signaling pathway

Glycosaminoglycan biosynthesis

Other types of O-glycan biosynthesis

Vitamin digestion and absorption

Glycosphingolipid biosynthesis

Non-alcoholic fatty liver disease (NAFLD)

Fatty acid degradation

Type II diabetes mellitus

Fatty acid metabolism

PPAR signaling pathway

Arachidonic acid metabolism

Th1 and Th2 cell differentiation

Inflammatory mediator regulation of TRP channels

202

T cell receptor signaling pathway

Insulin resistance

Insulin signaling pathway

Supplementary Table 6. Pathways obtained for subclinical ketosis for first lactation from 22 to 45 DIM (SCK1.2).

Trait Pathways

SCK1.2 Fatty acid degradation

Fatty acid metabolism

Valine, leucine, and isoleucine degradation

Peroxisome

Fatty acid elongation

Tryptophan metabolism

Adipocytokine signaling pathway

Fatty acid biosynthesis

PPAR signaling pathway

Insulin resistance

Biosynthesis of unsaturated fatty acids

Propanoate metabolism

Insulin signaling pathway

Folate biosynthesis

Pyruvate metabolism

Lysine degradation

Biosynthesis of amino acids

Arachidonic acid metabolism

Glucagon signaling pathway

Supplementary Table 7. Pathways obtained for subclinical ketosis for later lactations (2 to 5) from 5 to 21 DIM (SCK2.1).

Trait Pathways

SCK2.1 Vitamin B6 metabolism

203

Fatty acid biosynthesis

Lysosome

Pentose and glucuronate interconversions

Pyruvate metabolism

Glycine, serine, and threonine metabolism

Arginine and proline metabolism

Fatty acid metabolism

Purine metabolism

Biosynthesis of amino acids

B cell receptor signaling pathway

Arachidonic acid metabolism

Peroxisome

TGF-beta signaling pathway

Pyrimidine metabolism

Phosphatidylinositol signaling system

Glucagon signaling pathway

Glycerophospholipid metabolism

Insulin signaling pathway

Supplementary Table 8. Pathways obtained for subclinical ketosis for later lactations (2 to 5) from 22 to 45 DIM (SCK2.2).

Trait Pathways

SCK2.2 Mineral absorption

Valine, leucine, and isoleucine degradation

Mitophagy

Inflammatory mediator regulation of TRP

channels

Vitamin digestion and absorption

Signaling pathways regulating pluripotency of

stem cells

Maturity onset diabetes of the young

204

Pentose phosphate pathway

Butanoate metabolism

alpha-Linolenic acid metabolism

Pyruvate metabolism

Fatty acid degradation

Fatty acid metabolism

Regulation of lipolysis in adipocytes

Glycerolipid metabolism

Insulin secretion

Glycerophospholipid metabolism

Wnt signaling pathway

Purine metabolism

205

CHAPTER FIVE: General Conclusions

After calving, cows enter a state of negative energy balance (NEB) because they need energy to both maintain their bodies functioning and also energy to produce milk. At this stage, a body tissue mobilisation starts to produce energy. However, fatty acids are produced as a by- product of burning fat and the body accumulates ketone bodies which causes on the liver, kidneys, and brain. This NEB is worsened by the increasing demand of the industry for high producing dairy cows and the not so effective treatments available. For instance, this problem brings a cost of approximately CAD $300 per case (Geishauser et al., 1998; Gohary et al., 2016; McArt et al.,

2015; Steeneveld et al., 2020). Therefore, this thesis aimed to study clinical and subclinical ketosis to better understand the genetic contribution to these traits by determining genomic regions with a greater effect on ketosis expression in dairy cattle.

Before analyzing the data available for genomic association analysis, it was necessary to understand what was available in the literature and how the reported results could help on deciphering the genetic association of the traits studied in this thesis. To date, studies on differential gene expression have not revealed the full picture of the genetic basis of NEB and ketosis. Thus, in Chapter 2, the pattern of differential gene expression in liver of cows in NEB, and under subclinical and clinical ketosis was evaluated through a systematic review and meta- analysis of gene expression and genome-wide association studies (GWAS). In total, 118 articles were selected. Once all the papers initially included in the systematic review were screened, 20 papers were included in the study. The results highlighted 14 differentially expressed genes

(ACSL1, ALB, ANXA3, BIRC6, CPT1A, CSN1S1, FN1, G3BP2, HAO2, PC, PTK2, RASSF6,

SHROOM3 and UGT2B10) in the liver of cows under NEB, subclinical ketosis and clinical ketosis, harboring 24 significant polymorphisms in reported GWAS. Additionally, in the three metabolic

206

conditions, two important genes, PPARA and ACACA, were identified as differentially expressed.

Three significant metabolic pathways (“Glyoxylate and dicarboxylate metabolism”, “Metabolic pathways”, and “Peroxisome”) were associated with NEB, subclinical and clinical ketosis. Gene network analysis revealed co-expression interactions among 34 genes associated with functions involving fatty acid transport and metabolism. These genes might be good potential candidate genes to enhance cow’s resistance to the development of ketosis. The genes FN1 and PTK2 were enriched for QTL previously associated with ketosis on chromosome 2 and with milk iron content on chromosome 14, respectively. The results suggest that meta-analysis of gene expression studies, using qPCR, microarray, RNA-Seq and proteome techniques, combined with GWA studies can contribute to a better understanding of the genetic background of NEB, subclinical and clinical ketosis in dairy cattle, which could enhance selection decisions and help with the development of biomarkers for early diagnosis and prevention of ketosis. With that understanding obtained from the available literature, additional inquiries arose for the dataset to be analyzed, which were explored in chapter 3.

In Chapter 3, the main research query was: In view of this association of genomic regions with the traits, what would be the genetic variation explained by the main regions of the genome?

Furthermore, there was an interest in comparing the results found for clinical and subclinical ketosis in first and later lactations to assess the genetic association between with these two groups of lactations, treated as different traits. To address these issues, de-regressed breeding values

(dEBVs) were used as pseudo-phenotypes to perform a GWAS for clinical and subclinical ketosis for first and later lactations (2 to 5) in Holstein cattle to identify genomic regions that could clarify the genetic background of these traits. In addition, this study validated the 998 markers from

Kroezen et al. (2018), which are a combination of novel and previous reported SNP markers that

207

are located within or near candidate genes related to ketosis. In a functional analysis, regions of the genome that were found to be substantially linked to the four traits were further investigated to lessen the list of potential candidate genes. Highlighted genes were FASN and ACAT2.

Enrichment analyses found important gene ontology terms that are involved in fatty acid metabolism, lipid metabolism, and inflammatory response in dairy cattle. Gene-gene network analysis among the candidate genes provided very informative networks, which could help in defining shared biological processes among them. These results enhance the genetic understanding of clinical and subclinical ketosis in dairy cows, which could improve selection decisions in the future. The results of this study led to a new research question investigated in Chapter 4.

The major research question in Chapter 4 was: Is there any difference when looking at subclinical ketosis in a period very early in the lactation (5 to 21 DIM) compared to a later period

(22 to 45 DIM), as cows face a higher risk of developing ketosis within the first two weeks of lactation (Duffield et al., 1997; Tatone et al., 2017). Thus, to answer this question, this chapter concentrated on estimating variance components and GWAS analysis for subclinical ketosis (BHB concentrations) in Holstein cattle considering the periods of 5 to 21 DIM and 22 to 45 DIM as different traits within first and later lactations (2 to 5) using deregressed estimated breeding values derived from BHB concentrations. Then, to construct a list of potential candidate genes, the 20- top genomic windows were analyzed. Genomic regions found to be correlated with the traits on the GWAS were further examined in functional analysis. The findings of the current study raised the question of how these processes lead to the development of ketosis in the 14 days postpartum lactation. The results emphasized the importance of potential candidate genes for subclinical ketosis that regulated the synthesis and degradation of cell structure, immune function, and ketone bodies, and will be readily testable in future functional studies. The genes highlighted were ACSL3,

208

which transcriptionally regulates lipogenic genes, and ADCY8, which plays an important role in insulin secretion. Additionally, featured genes in the top window for SCK1.1 were PXMP2 and

GOLGA3, and for SCK2.1 were RIMS1 and KCNQ5. These genes might not clearly show an association with the phenotype, however they are located in windows explaining the highest genetic variation of subclinical ketosis in the period from 5 to 21 DIM, which may be associated with the higher incidence of ketosis in this period relative to 22 to 45 DIM period and should be further investigated. Furthermore, genes in the family Zinc Finger Proteins were found in the window explaining the highest genetic variation for SCK1.1, which might indicate that the regulation of gene expression plays an important role in the first DIM period compared to the second period, which showed a greater presence of genes linked to metabolic pathways. Reported metabolic pathways provide insight into the biology of resistance to ketosis These findings add to the knowledge about the genetic background of subclinical ketosis and thus may contribute in the future to improved selection of cows that are less susceptible to ketosis.

This thesis investigated the genetic architecture underlying the onset of clinical and subclinical ketosis by analyzing available literature in a meta analysis and performing association studies and downstream in silico functional analyses. This thesis confirmed genes and SNPs that were previously identified as associated with sub-clinical and clinical ketosis, as well as suggested new polymorphisms in candidate genes with known function in important metabolic pathways.

The limitations found were mainly due to the challenges imposed by the nature of the traits studied, which have low heritability levels. In addition, the diagnosis to define which animals really have clinical ketosis is not straightforward, while subclinical ketosis is indirectly assessed by BHB levels in the blood or milk. Another limitation was the scarcity of papers in the literature to compare the results to, especially when subclinical ketosis was analyzed in two DIM periods in

209

early lactation. Despite these limitations, together the results of this thesis make it possible to recognize key genes, which will make it possible to define future studies that will lead to an ever greater understanding of this undesired metabolic condition for the dairy industry and that affects animal welfare. Future research projects could try to identify real causal mutations. For example, in vitro cell culture or gene expression analysis can be used in animals with different genotypes.

Gene expression studies can also be performed on different tissues of animals with different genotypes, to understand the orchestration of gene expression in the tissues where they are expressed.

References

Duffield, T. F., Kelton, D. F., Leslie, K. E., Lissemore, K. D., & Lumsden, J. H. (1997). Use of

test day milk fat and milk protein to detect subclinical ketosis in dairy cattle in Ontario.

The Canadian Veterinary Journal = La Revue Veterinaire Canadienne, 38(11), 713–718.

Geishauser, T., D. Leslie, D. Kelton, and T. Duffield. 2001. Monitoring for subclinical ketosis in

dairy herds. Compend Contin Educ Pract Vet 23:65- 71.

Gohary, K., M. W. Overton, M. Von Massow, S. J. LeBlanc, K. D. Lis-semore, and T. F. Duffield.

2016. The cost of a case of subclinical ketosis in Canadian dairy herds. Can Vet J. 57:728–

732.

Kroezen, V., Schenkel, F. S., Miglior, F., Baes, C. F., & Squires, E. J. (2018). Candidate gene

association analyses for ketosis resistance in Holsteins. Journal of Dairy Science, 101(6),

5240–5249. https://doi.org/10.3168/jds.2017-13374

210

McArt, J. A. A., Nydam, D. V., & Overton, M. W. (2015). Hyperketonemia in early lactation

dairy cattle: A deterministic estimate of component and total cost per case. Journal of

Dairy Science, 98(3), 2043–2054. https://doi.org/10.3168/jds.2014-8740

Steeneveld, W., Amuta, P., Soest, F. J. S. van, Jorritsma, R., & Hogeveen, H. (2020). Estimating

the combined costs of clinical and subclinical ketosis in dairy cows. PLOS ONE, 15(4),

e0230448. https://doi.org/10.1371/journal.pone.0230448

211