HAIR CORTISOL: SAMPLING METHODOLOGY AND ASSOCIATIONS WITH HEALTH AND REPRODUCTION IN DAIRY COWS by Tracy Anne Burnett B.Sc., The University of British Columbia, 2011 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Applied Animal Biology) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) October 2014 Tracy Anne Burnett, 2014

ii

Abstract Dairy cattle are often challenged with stressful practices and conditions. Cortisol is often used as a biomarker to detect stress. Hair is a promising new medium to detect long-term changes of circulating cortisol. This thesis investigated methodologies for the collection and processing of hair for cortisol analysis, and determined associations of hair cortisol concentrations with health disorders and fertility in lactating Holstein cows.

First, we investigated the effects of hair colour, sampling location, and processing method on the amount of cortisol extracted from hair samples of 18 black and white Holstein dairy cows. Second, we investigated the associations between hair cortisol with clinical and subclinical disease, and reproductive success. Hair samples were collected from the tail switch of lactating Holstein cows to determine the effects of clinical disease and fertility (n = 64), or subclinical disease (n = 54). White hair had greater cortisol concentrations than black hair (Geometric Mean [95% CI]) (7.8 [6.8, 9.2] vs. 4.2 [3.6, 5.0] pg/mg). When only white samples were analyzed, hair from the tail switch had more cortisol than the shoulder (11.0 [7.6, 16.0] vs. 6.2 [4.2, 9.2] pg/mg). Processing with a ball mill yielded greater concentrations of extracted cortisol than when using scissors (10.4 [5.8, 18.8] vs. 4.7 [2.6, 8.4] pg/mg). In Holsteins, the tail switch is always white and grows faster making it an ideal location for measuring hair cortisol. Animals with clinical disease presented higher hair cortisol concentrations than clinically healthy animals (8.8 [7.8, 9.9] vs. 10.7 [9.6, 12.0] pg/mg); however, animals

iii

diagnosed with subclinical disease did not differ (11.5 [9.7, 13.7] vs. 11.3 [9.6, 13.3] pg/mg for healthy and subclinical groups, respectively). Multiparous cows that became pregnant by 100 days postpartum had lower hair cortisol concentrations at 42 and 84 DIM. Overall, using standard and consistent methods to sample, cortisol in hair offers important insights into long-term changes of circulating cortisol. Hair cortisol concentrations appear to be associated with clinical disorders and have a direct association with pregnancy outcomes; however, hair cortisol concentrations may not be suited to differentiate situations of stress with lower magnitudes, such as subclinical disease.

iv

Preface All of the work presented in this thesis was conducted at the University of British Columbias Dairy Education and Research Centre in Agassiz, BC. The projects and associated methods within this thesis were approved by the University of British Columbias Animal Care Ethics Committee [certificate #A11-0319]. A version of the material in Chapter 2 has been accepted for publication: Burnett, T.A., A.M.L. Madureira, B.F. Silper, A. Nadalin, A. Tahmasbi, D.M. Veira, R.L.A. Cerri. 2014. Short communication: Factors affecting hair cortisol concentrations in lactating dairy cows. J. Dairy Sci. In Press. Additionally, a version of the material in Chapter 3 has been submitted for publication: Burnett, T.A., A.M.L. Madureira, B.F. Silper, A. Nadalin, A. Tahmasbi, D.M. Veira, R.L.A. Cerri. 2014. Relationship of concentrations of cortisol in hair with health, plasma metabolites and reproductive parameters in dairy cows. Both manuscripts were co-authored by my main supervisor, R.L.A. Cerri, my supervisory committee member, D.M. Veira, and four colleagues, A.M.L. Madureira, B.F. Silper, A. Nadalin, A. Tahmasbi. Cerri and Veira supervised, helped with concept formation and interpreting material, as well as editing and providing comments on manuscript drafts. A.M.L. Madureira, B.F. Silper, A. Nadalin, and A. Tahmasbi were involved with data collection, processing, and analysis. I was the lead investigator, responsible for all major areas of concept formation, material interpretation, data collection and processing, statistical analysis, and manuscript composition.

v

Table of Contents

Abstract ................................................................................................................................................... ii

Preface .................................................................................................................................................... iv

Table of Contents ................................................................................................................................... v

List of Tables ........................................................................................................................................ vii

List of Figures ..................................................................................................................................... viii

List of Abbreviations .......................................................................................................................... ix

Acknowledgements .............................................................................................................................. x

Chapter 1: Introduction ..................................................................................................................... 1 1.1 Introduction to stress ............................................................................................................................. 1 1.2 The stress response ................................................................................................................................. 2 1.2.1 Acute stress versus chronic stress ................................................................................................................ 4 1.3 Regulation of the neuroendocrine stress response ..................................................................... 5 1.3.1 Hypothalamic-pituitary-adrenal axis ........................................................................................................... 6 1.3.2 Sympathetic nervous system ........................................................................................................................... 7 1.4 Biological responses to glucocorticoid production ...................................................................... 8 1.4.1 Effects of stress on the immune system ...................................................................................................... 9 1.4.2 Effects of stress on reproduction ................................................................................................................ 11 1.5 How is stress measured? ..................................................................................................................... 13 1.5.1 Using blood to measure cortisol .................................................................................................................. 15 1.5.2 Using saliva to measure cortisol .................................................................................................................. 15 1.5.3 Using faeces to measure cortisol ................................................................................................................. 16 1.5.4 Using hair to measure cortisol ..................................................................................................................... 17 1.6 Important concepts when using hair for the measurement of stress .................................. 18 1.6.1 How is cortisol incorporated into hair? ................................................................................................... 18 1.6.2 What factors affect hair cortisol concentrations? ................................................................................ 19 1.7 Objectives and hypotheses ................................................................................................................. 20

Chapter 2: Factors Affecting Hair Cortisol Concentrations in Lactating Dairy Cows .. 22 2.1 Introduction ............................................................................................................................................. 22 2.2 Materials and methods ......................................................................................................................... 24 2.2.1 Animals and housing ........................................................................................................................................ 25 2.2.2 Hair processing and analysis ........................................................................................................................ 27 2.2.3 Statistical analyses ............................................................................................................................................ 28 2.3 Results and discussion ......................................................................................................................... 29

Chapter 3: Relationship of Concentrations of Cortisol in Hair with Health, Plasma Metabolites and Reproductive Parameters in Dairy Cows .................................................. 37 3.1 Introduction ............................................................................................................................................. 37 3.2 Materials and methods ......................................................................................................................... 39

vi

3.2.1 Animals and housing ........................................................................................................................................ 39 3.2.2 Blood sampling and analysis ......................................................................................................................... 42 3.2.3 Hair sampling and analysis ............................................................................................................................ 43 3.2.4 Uterine cytology and ultrasonography ..................................................................................................... 44 3.2.5 Statistical analyses ............................................................................................................................................ 45 3.3 Results ....................................................................................................................................................... 46 3.4 Discussion ................................................................................................................................................. 51

Chapter 4: General Discussion ....................................................................................................... 66 4.1 Summary ................................................................................................................................................... 66 4.2 Limitations ............................................................................................................................................... 71 4.3 Future directions ................................................................................................................................... 74

Chapter 5: Conclusions ..................................................................................................................... 76

References ............................................................................................................................................ 77

vii

List of Tables Table 2.1: Effect of hair colour, sampling location and processing method on concentrations of cortisol extracted from hair samples collected from black and white Holstein cows ................................................................................................................................................. 33 Table 3.1: Comparative statistics (Geometric Mean [95% CI]) for the Subclinical endometritis (Endo) and Control groups of Experiment 2 ........................................................ 58 Table 3.2: Pearson correlation coefficients found between hair cortisol, plasma haptoglobin, plasma ceruloplasmin, percent neutrophils, milk yield and cervix endometrium diameter in Holstein dairy cows (Experiment 2) .............................................. 59

viii

List of Figures Figure 2.1: Hair sampling locations: a) shoulder, hair located at the scapula; b) top line, hair located at the dorsal side of the vertebrae; c) hip, hair located over the femur-ischium junction; and d) tail switch, hair located at the distal end of the tail ..................................... 34 Figure 2.2: Hair growth rates at different body locations (Mean SE) collected from black and white Holstein cows ............................................................................................................................ 35 Figure 2.3: Effect of parity and DIM on concentrations of cortisol (Mean SE) extracted from hair samples collected from the tail switch of 37 clinically healthy lactating cows (18 primiparous and 19 multiparous) at 0, 21, 42, 84 and 126 DIM ..................................... 36 Figure 3.1: Schematic of the experimental and sampling design of Experiment 1 and Experiment 2 .................................................................................................................................................. 60 Figure 3.2: Effect of DIM on cortisol concentrations (Geometric Mean SEM) extracted from hair samples collected at 0, 21, 42, 84 and 126 DIM from all cows in Experiment 1 ............................................................................................................................................................................... 61 Figure 3.3: Effect of disease on concentrations of cortisol (Geometric Mean SEM) extracted from hair samples from multiparous animals in Experiment 1 .......................... 62 Figure 3.4: Hair cortisol concentrations (Geometric Mean SEM) of multiparous cows that were diagnosed as clinically diseased or clinically healthy by 126 DIM .............................. 63 Figure 3.5. Scatter plots representing Pearson correlations between hair cortisol and blood BHBA (r = 0.10; P = 0.46), hair cortisol and blood glucose (r = 0.26; P = 0.07) and glucose and BHBA (r = -0.26; P = 0.07) concentrations at 21 DIM in Experiment 1 ....... 64 Figure 3.6: Hair cortisol concentrations (Geometric Mean SEM) of multiparous (panel A) and primiparous cows (panel B) that were diagnosed non-pregnant or pregnant by 100 DIM ...................................................................................................................................................................... 65

ix

List of Abbreviations ACTH - Adrenocorticotropin hormone AVP - Arginine vasopressin BCS - Body condition score BHBA - Beta-hydroxybutyrate acid CI Confidence interval CRH - Corticotrophin-releasing hormone DIM Days in milk FSH- Follicle-stimulating hormone GnRH Gonadotropin-releasing hormone HPA - Hypothalamic-pituitary-adrenal HPG - Hypothalamic-pituitary-gonadal LH- Luteinizing hormone NEFA - Non esterified fatty acids PMNT - Phenylethanolamine-N-methyltransferase SNS - Sympathetic nervous system TMR Total mixed ration

x

Acknowledgements First and foremost, I would like to thank Ronaldo Cerri and Douglas Veira for teaching me all I know about research and dairy cattle. Their passion and encouragement has pushed me intellectually and nurtured my aspiration to continue my studies in dairy cattle research. I am extremely thankful for Nelson Dinn who introduced me to the world of dairy cattle in the summer of 2009, and very patiently (and calmly) taught me the ins and outs of working on a dairy farm and sparked my love for cows. A special thank you to Audrey Nadalin for all the lab work she did for me and keeping me stay sane while I tediously cleaned and processed hair samples for hours. To my family, Mom, Dad, Christen and Karen, for their support through this time, particularly for their invaluable advice and engulfing me in a cloud of Burnett laughter. To Joo, for always knowing how to make me happy, supporting me, and going on amazing adventures with me all around the world! And to my friends, Karen, Jillian, Liz, and Chelsea, for being understanding and always by my side no matter the distance.

Finally, I cannot express how much my time at the UBC Dairy has meant to me; I have met the most amazing people at that farm. Most importantly Joo, who without the farm I would have never met. Augusto, Ruby, Heather and Maria, who have all been the best of friends - from early mornings to unlucky night checks that went just a little too long, they were all there making even the worst situations fun. To the original Repro Team (Augusto, Bruna, Maria, and Cristiano), thank you for all the amazing memories! And last but not least, thank you to the farmers, Ted, Brad, Barry, Bill, Andrew, Eric and Hendrik for all the help, and for always covering my back when it needed covering ;)

1

Chapter 1: Introduction

1.1 Introduction to stress The study of stress has been a topic of interest for many years in both humans and in non-human animals. The word stress has been used loosely in popular magazines and newspapers but within the scientific literature the term has a more strict definition; however, even within the scientific field, different researchers have their own definition of stress. Unlike most diseases, stress has no distinct prognosis, which makes it much more difficult to define (Moberg and Mench, 2000). Broom and Johnson (1993) defined stress as an environmental effect on an individual which overtaxes its control systems and reduces its fitness or appears likely to do so, while Mormde et al. (2007) defined stress as a general term used to describe environmental factors soliciting adaptation mechanisms and the response to these challenges. In this thesis, stress will be defined as a biological response elicited when an individual perceives a threat to its homeostasis (Moberg and Mench, 2000).

Although the word stress is commonly thought to imply a negative event, it is not inherently bad. Activities such as exercise and courtship are events that have the biological effects of stress but may be interpreted by certain individuals as exhilarating or rewarding (Broom and Johnson, 1993). This phenomenon is due to stress being contextual and highly dependent on how stimuli are perceived, and if the biological response elicited is warranted for that specific stimuli or not. Stress that is perceived as a positive condition is defined as eustress, while stress that is perceived as a negative condition and causes a reduction of well-being is defined as distress (Trevisi and Bertoni, 2009); for the remainder

2

of this thesis, the word stress will be used in reference to distress. Moreover, the stress response is also seen as beneficial due to its ability to create a state of adaptation in an animal where they have increased vigilance and blood flow to the extremities and vital organs that allow the animal to remove itself from harmful situations quickly (Broom and Johnson, 1993). 1.2 The stress response Stress results from first, a stimulus leading to an imbalance of homeostasis, termed a stressor, followed by the corresponding defence reaction, termed the stress response (Mstl and Palme, 2002). The stress response is inherently very important to animals because it allows for the perception of external stimuli that are threatening to its homeostasis, and allows the body to make necessary physiological and metabolic changes required to cope with the demand of a homeostatic challenge and defend itself against the threat (Moberg, 1985; Miller and OCallaghan, 2002). Moberg (1985) initially created a model of animal stress that was categorized into three general stages: stressor recognition, biological defence against the stressor, and the consequences of the stress response. The consequences resulting from the stress response are what characterize the stress as a positive or negative event. Stressor recognition is initiated from the central nervous system, where stimuli are assessed to determine if they are a threat and may cause a significant challenge to the animal, and thus considered a stressor. A very broad range of stressors are present in the natural world and may include anything from changes in temperature, to pain, or the presence of a predator. Once a stressor has been perceived as

3

such, whether this perception is appropriate or not, the animal mounts a biological response. The stress response has four main divisions: behavioural, autonomic nervous system, immune, and neuroendocrine (Moberg, 2000). The behavioural response usually consists of actions that have the purpose of removing the animal from the stressful stimuli, such as avoiding a predator or taking refuge in the shade to decrease its internal temperature. This type of response is usually the first reaction of an animal, and although it may not be suitable or effective for certain stressors, it is thought to be the most biologically economic response in comparison with the other divisions of the stress response (Moberg, 2000). Behavioural responses may vary greatly and even be inhibited in captivity and thus may be hard to assess in certain situations (Moberg, 2000). The second line of defense for an animal is the autonomic nervous system response, which is the basis for the commonly known fight or flight response. It is a rapid and specific response where the sympathetic and parasympathetic systems alter many biological systems such as the cardiovascular system, gastrointestinal system, exocrine glands and adrenal medulla (Moberg, 2000). The autonomic nervous system response results in many quick and dramatic reactions, such as increased heart rate and blood pressure, in response to acute stress situations; however, these responses last for quite a short period of time making them quite difficult and impractical to accurately measure (Moberg, 1985). The immune system is also negatively impacted by stress due to the suppression of immune competence during stressful events (Moberg, 2000). The link between the nervous

4

system and the immune system is thought to be through different mediators such as glucocorticoids, catecholamines and some endorphins (Dunn, 1988). Although immune suppression is measurable, there is limited research on how this information can be interpreted (Moberg, 2000). The effects stress has on the immune system are discussed in more detail below. Contrary to the other divisions of the stress response, the neuroendocrine system response has long lasting effects on the body and thus is much easier to measure. Stress causes activation of the hypothalamic-pituitary-adrenal (HPA) axis and the sympathetic nervous system (SNS) (Miller and OCallaghan, 2002), which stimulates the secretion of many stress mediators. These substances may compromise processes that are controlled by the pituitary, such as reproduction and metabolism, and thus negatively impact the animals fitness (Moberg, 2000).

1.2.1 Acute stress versus chronic stress Stress can be described in two general conditions in relationship to the duration of the consequent responses: acute and chronic. Acute stress is defined as a short term and low magnitude condition that generally allows for a quick and complete recovery of homeostasis (Trevisi and Bertoni, 2009). Acute stress responses may cause a shift of resources towards functions necessary for survival, such as arousal, vigilance, increased oxygenation and nutrition of the brain, heart and skeletal muscles, and subsequently down regulate functions unnecessary for survival at that specific moment such as eating, growth, and reproduction (Chrousos, 2009). Overall, the acute stress response is thought to be

5

intrinsically beneficial for animals as it is essential for the survival of animals in the wild (Sapolsky, 2000). On the contrary, chronic stress is characterised as a sustained stress response and/or the excessive secretion of stress mediators such as cortisol, epinephrine and norepinephrine (Chrousos, 2009; Matteri et al., 2000); the condition is also characterized as an inability to adapt and recover homeostasis for an extended period of time (Trevisi and Bertoni, 2009). Chronic stress can occur when the body experiences too many stressors or repeated exposure to the same stressor, such that the body is under a continuous stress response and unable to activate a relaxation response, eventually resulting in an overexposure of stress hormones (Trevisi and Bertoni, 2009). Although dependent on the stressor, repeated exposure to the same stressor may also result in the loss of a stress response to that specific stressor, which is usually a result of desensitization (Miller and OCallaghan, 2002). Overall, without the ability to adapt and recover homeostasis, increased production of stress hormones can cause many behavioural and somatic disorders (Chrousos, 2009).

1.3 Regulation of the neuroendocrine stress response Two endocrinological systems are activated during stress events that mediate the stress response: the HPA axis and the SNS. Activation of the HPA axis and SNS is required to be able to achieve the optimal homeostasis, or eustasis, and is done by causing many physiological changes during stress events (Miller and OCallaghan, 2002). Many different hormones and neurotransmitters are released into the body during the stress response

6

such as cortisol, vasopressin, oxytocin, epinephrine, and norepinephrine (Chrousos, 2009; Matteri et al., 2000). Generally, the HPA axis is responsible for the production of glucocorticoids (e.g.: cortisol) while the SNS is responsible for the production of catecholamines (e.g.: epinephrine) (Matteri et al., 2000).

1.3.1 Hypothalamic-pituitary-adrenal axis The three key hormones secreted with activation of the HPA axis: corticotrophin-releasing hormone (CRH), adrenocorticotropin hormone (ACTH) and a glucocorticoid that is specific for each species, in cattle this hormone is cortisol. The HPA axis is controlled through a negative feed back loop, where the end product, cortisol, inhibits the production of the initiating substance, CRH. The initiating step of the neuroendocrine stress response is the synthesis of CRH from neurons located at the paraventricular nucleus in response to internal or external stimuli (Miller and OCallaghan, 2002). The CRH travels from the paraventricular nucleus until the median eminence where it then travels through the hypothalamo-hypophyseal portal system to the anterior lob of the pituitary (Aguilera, 1998). Upon the arrival of CRH, ACTH is synthesized and secreted from the anterior pituitary into the blood system where it stimulates the secretion of cortisol from the cortex of the adrenal gland. In conjunction with CRH expression, arginine vasopressin (AVP) is also secreted from the CRH neurons that are responsible for ACTH secretion; the ACTH response from either CRH or AVP alone is much lower than when both peptides are secreted together (Scott and Dinan, 1998). The ratio of CRH to AVP that is secreted to the anterior pituitary has been shown to change depending on different stimuli and stress

7

paradigms (Scott and Dinan, 1998), such that chronic stress situations lead to a greater increase in AVP than CRH (Aguilera, 1998). The increased ratio of AVP to CRH found during chronic stress is thought to be necessary so that animals can still respond to novel stimuli during a time of desensitization that can be found with repeated exposure to the same stressor (Aguilera, 1998; Scott and Dinan, 1998).

1.3.2 Sympathetic nervous system The SNS stress response is characterized as the fight or flight response, which quickly increases the activity and readiness of the animal. Epinephrine and norepinephrine are two important catecholamines that are produced by the activation of the SNS during stressful events (Romero and Butler, 2007). In contrast to glucocorticoids, which take 20 30 min to have effects on the body, catecholamines are fast acting a characteristic that is crucial during emergency situations (Romero and Butler, 2007). Once an animal perceives a stimulus as threatening, epinephrine and norepinephrine are released from both the adrenal medulla and nerve terminals that are part of the SNS (Romero and Butler, 2007). The synthesis of epinephrine in the adrenal medulla is controlled by the production of glucocorticoids (Wurtman, 2002). Glucocorticoids travel from the adrenal cortex to the medulla via the portal vascular system within the adrenal gland where they induce the production of phenylethanolamine-N-methyltransferase (PNMT), which is the rate-limiting enzyme needed for the synthesis of epinephrine (Wurtman, 2002). Biological responses from epinephrine and norepinephrine include decreased visceral activity and digestion, and increased blood flow to the brain, visual acuity, gas exchange efficiency in the lungs,

8

break down of glycogen to glucose, increased heart rate and vasodilation in muscles (Romero and Butler, 2007). Overall, the SNS biological response increases bodily functions, such as alertness, that may benefit the animal during an acute emergency, and decrease others, such as digestion, that will not benefit them at that specific time; although these responses are valuable during acute stress, they become overtaxing when persistent (Romero and Butler, 2007) 1.4 Biological responses to glucocorticoid production The biological responses to glucocorticoid production can be summarized as five broad actions: increasing blood glucose concentrations, altering behaviour, inhibiting growth, inhibiting reproduction and modulating the immune system (Romero and Butler, 2007). The role of these biological responses is thought to function as a method to help the animal recover from a stressor, and down regulate systems whose functions are not mandatory at that time. Increased glucose availability provides energy for the body tissues for the increased metabolic demands associated with an emergency situation (Miller and OCallaghan, 2002). Elevation of blood glucose is done indirectly by increasing the breakdown of glycogen and by stimulating gluconeogenesis; additionally glucocorticoids reduce the uptake of glucose by targeting tissues thus increasing the amount of glucose available for tissues that are required to respond to the stressor (e.g.: muscles) (Romero and Butler, 2007). The specific mechanism on how glucocorticoids affect behaviour is unknown; although the behavioural response is not known to be specific to a stressor, it is reportedly dependent upon how the stressor is presented and perceived by the animal

9

(Romero and Butler, 2007). Glucocorticoids are known to inhibit growth by blocking the secretion of growth hormone from the pituitary, decreasing the sensitivity of target cells to growth hormone and inhibiting overall protein synthesis in the body (Romero and Butler, 2007). The effects of glucocorticoids are not easily seen in the over all growth of an animal unless there is prolonged exposure, such as during chronic stress. The modulations in which glucocorticoids have on the immune system and reproduction are discussed in detail below.

1.4.1 Effects of stress on the immune system Dysfunction of the HPA axis has been shown to negatively impact animal health (Chrousos, 2009). Chronic stress-induced HPA activity has been associated with reduced feed intake, negative energy balance, and increased metabolic rate (Rushen et al., 2010); however, most health disorders have been related to immunosuppression. Although the acute stress response has been associated with the enhancement of the immune system through release of blood leukocytes from the blood to organs such as lymph nodes, skin and bone marrow (Dhabhar and McEwen, 1997; Sapolsky, 2000), many researchers agree that chronic stress has detrimental effects on the immune system (Dhabhar and McEwen, 1997; Chrousos, 2009). It is suggested that the initial enhancement of the immune system is a defensive response to a stressor, and the succeeding immunosuppressive response is to prevent an overreaction to these defensive responses (Rushen et al., 2010). The immunosuppressive effects of increased concentrations of glucocorticoids have been seen in many farm animals such as in dairy cattle and pigs (Hopster et al., 1998; Tuchscherer et

10

al., 2004). Generally, glucocorticoids impede the immune system by inhibiting the action of cytokines and other mediators that promote immune and inflammatory responses (Rushen et al., 2010). In dairy cattle, increased concentrations of glucocorticoids have been reportedly linked to the depression of anti-body concentrations (through either suppression of antibody production or increased antibody catabolism), inducing neutrophilia (migration of neutrophils out of tissues and into blood), and reducing the phagocytic functions in monocytes and macrophages (Roth, 1985). In humans, there is evidence that chronic stress can induce immune dysfunction reducing the activation and proliferation of T and B cells (Sapolsky et al., 2000) and switching the immune response from T1 mediated (cellular) to T2 mediated (humoral), which may cause the body to be vulnerable to certain infections (Chrousos, 2009). Comin et al. (2013) reported that cows that were clinically compromised (e.g.: laminitis, metritis or mastitis) had higher concentrations of cortisol in hair than those that had an absence of disease. Many researchers have reported that intramammary infections increase cortisol concentrations in dairy cattle (Huszenicza et al., 2004). Additionally, Lavon et al. (2008) reported that animals administrated E. coli endotoxin (using both intramammary and IV injection) showed a significant increase in plasma cortisol when compared to control animals suggesting that an inflammatory response may induce increases in cortisol. Although it is hard to determine the causal relationship between infection and cortisol concentrations (i.e.: whether infection causes increases in cortisol concentrations, or vice versa), there is mounting evidence suggesting that clinical disease may initiate a stress response in animals.

11

1.4.2 Effects of stress on reproduction The effects that stress has on reproduction in humans have been thoroughly studied; examples of these stressors include infectious diseases, psychiatric disorders, surgical trauma, and strenuous exercise (Rivier and Rivest, 1991). In cows, stressors such as isolation, transport and restraint have been found to interfere with endocrine events preceding ovulation and thus resulting in ovulation failure (Moberg, 2000). Stress affects the hypothalamus-pituitary-gonadal (HPG) axis at the hypothalamus (inhibiting GnRH secretion), at the pituitary gland (interfering with gonadotropin-releasing hormone (GnRH) induced luteinizing hormone (LH) release) and, to a lesser extent, at the gonads (altering the stimulatory effect of gonadotropins on sex steroid secretion) (Rivier and Rivest, 1991). Short-term and long-term stressors may affect reproduction differently: short-term stressors have been shown to have neutral or inhibitory effects in dairy cattle (von Borell et al., 2007), while detrimental effects on reproduction generally appear to result from long-term stress (Tilbrook et al., 2000). Short-term stressors have been shown to have stimulatory effects on reproduction with certain species (e.g.: increased LH), but within the dairy cattle literature stimulatory effects have rarely been reported.

High concentrations of cortisol have been shown to suppress pulsatile GnRH secretion, and subsequently reduce LH pulse frequency (von Borell et al., 2007); it is not known if the suppression of GnRH secretion is primarily due to the high concentrations of cortisol or if it is due to high concentrations of CRH (Moberg, 2000). Due to disruptions in GnRH and LH pulses, stress has detrimental effects on ovarian cyclicity, reduced estradiol

12

production by growing follicles, and often results in ovulation failure (Dobson et al., 2003; Smith et al., 2003). Reduced estradiol production also has negative effects on estrous behaviour leading to poor estrous detection and timing of artificial insemination (Dobson and Smith, 2000). Furthermore, glucocorticoids have been shown to exert an inhibitory effect on the GnRH neuron, the pituitary gonadotroph, and the gonads, and render target tissues resistant to gonadal hormones (Charmandari et al., 2005). It has also been hypothesized that stress may have an effect on the feedback actions of inhibin, since glycoprotein is a major feedback regulator of the secretion of follicle-stimulating hormone (FSH) (Tilbrook et al., 2000). Many researchers have shown that chronic stress due to lameness has detrimental effects on dairy cattle fertility. Walker et al. (2008) reported that cattle experiencing chronic lameness have altered estrous behaviour and hypothesized this may be due to disruptions of the pulsatile patterns of GnRH that leads to insufficient gonadotropin support at the ovary, resulting in low progesterone concentrations, and subsequently causing decreased estrous behaviour. Progesterone priming in the luteal phase prior to estrus has been suggested to increase estradiol receptors in the hypothalamus, making it more sensitive to estradiol, resulting in stronger estrous expression (Walker et al., 2008). Moreover, Dobson and Esslemont (2002) reported that bouts of clinical mastitis or infection in the uterus have negative effects on fertility in dairy cattle. Additionally, extended ACTH administration has been shown to create follicular cysts in dairy cattle, due to the inhibition of LH but not of FSH (Dobson et al., 2000). Furthermore, chronic stress can lead to such low LH pulses that the estrus cycle can not be completed and the cow enters

13

anestrus (Dobson and Smith, 2000) 1.5 How is stress measured? Of all the stress mediators, cortisol is the most common physiological parameter used to measure stress, and has therefore been validated as a reliable measure of the stress response in animals such as cows (Christison and Johnson, 1972), goats (Aoyama et al., 2008), horses (Visser et al., 2008), pigs (Turner et al., 2005), and sheep (Smith and Dobson, 2002). The preference for the quantification of glucocorticoids, rather than catecholamines, is a result of the fast clearance rate of catecholamines, making them much more difficult to measure in comparison to glucocorticoids (Romero and Butler, 2007). Cortisol can be analyzed from many different mediums, most notably from the blood (Lefcourt et al., 1993), saliva (Negro et al., 2004) and faeces (Mstl et al., 2002), and more recently from hair (Comin et al, 2011). Each of these media have been individually demonstrated in cattle, and have been shown to represent HPA axis activation over different periods of time. Blood, saliva, and faeces are all mediums useful for measuring cortisol produced as a result of acute stress responses but are not able to capture long-term increases in circulating cortisol; the use of hair to measure cortisol as an indicator of chronic stress responses has recently been a research topic of interest.

Despite the wide use of cortisol concentrations as a marker of the stress response, interpretation can be quite complex due to the diurnal and pulsatile patterns of cortisol secretion, thus leading to substantial individual variation within the animal and throughout the day; cortisol concentrations are highest in the morning and diminish as the day

14

progresses (Thun et al., 1981; Lefcourt et al., 1993). Circadian rhythms of cortisol are driven mainly by the innate biologic clock, which is an internal timekeeping system that allows organisms to anticipate and prepare for changes within the environment and thus behave appropriately for the time of day (Chung et al., 2011). Ultradian rhythms of cortisol are produced by the pulsatile fashion that cortisol is secreted from the adrenal gland (Lefcourt et al., 1993). The pulsatile nature of cortisol release is important to consider when measuring chronic stress, as several samples per day would be necessary to control for the variations of cortisol throughout the day (Rushen et al., 2010); when measuring acute stress, it is important to control for the time of day the sample is collected. Handling and restraint of dairy cattle needed for blood and saliva collection has been shown to rapidly increase plasma cortisol and consequently may confound sampling results (Cook et al., 2000), thus feedback-free sampling methodologies are desirable. It is also noteworthy to mention that the release of cortisol during stress responses has been shown to be stressor dependant (Pacak and Palkovits, 2001), and does not occur with every stressor (Broom and Johnson, 1993). For example, in comparison with control mice, immobilization was shown to greatly increase plasma ACTH, norepinephrine and epinephrine in relatively equal amounts, pain was shown to increase all three with norepinephrine increasing almost twice as much as the others, and cold stress was shown to increase norepinephrine but not ACTH or epinephrine (Pacak and Palkovits, 2001). Although the stress responses varied in hormone profile between stressors, they were shown to be consistent within the same stressor.

15

1.5.1 Using blood to measure cortisol The use of blood as a medium to measure cortisol has been widely used within dairy cattle research to demonstrate the stress responses associated with common practices such as handling (Cook et al., 2000), transportation (Lay et al., 1996), regrouping (Friend et al., 1977), dehorning (Stafford and Mellor, 2005), and overstocking (Friend et al., 1979). The analysis of plasma for cortisol represents the concentration of the free and bonded fraction of cortisol that is present in the blood at the time of sample collection (Mormde et al., 2007). Although the use of blood is very practical and well validated, it has many disadvantages; in addition to being affected by handling (Cook et al., 2000), and circadian rhythms of cortisol release (Thun et al., 1981), repeated blood sampling has also been shown to increase cortisol concentrations (Hopster et al., 1999). Under certain experimental designs, sampling concerns may be alleviated using remote sampling procedures (e.g.: a very long catheter), and by standardizing the time of day samples are collected (Negro et al., 2004).

1.5.2 Using saliva to measure cortisol Sampling of saliva is considered to be a less invasive method in comparison to blood sampling (Mstl and Palme, 2002); however, some researchers suggest it is not completely stress-free because restraint is still needed for adult cows (Negro et al., 2004). Cortisol enters the saliva by passive transfer, is unaffected by salivary flow rate (Negro et al., 2004), and is detectable without delay in plasma cortisol concentrations (Mormde et al.,

16

2007; Yates et al., 2010). In contrast with plasma cortisol, salivary cortisol only represents the free fraction of cortisol resulting in lower overall concentrations of cortisol in saliva than plasma; consequently samples must be concentrated before extraction and sometimes may be too low to be detected by saliva cortisol test kits (Negro et al., 2004). During stress events the free fraction of cortisol increases more than the bonded fraction, thus some researchers suggest that measuring the free fraction of cortisol is a better measure of the stress response than measuring both the free and bonded fractions together, as done in plasma cortisol analysis (Mormde et al., 2007). Like plasma cortisol, salivary cortisol can be measured at a fixed time before and after an imposed stress, which is advantageous if a specific stressor is to be tested (Negro et al., 2004). Sampling saliva can be disadvantageous in free moving animals because, although less affected than plasma, the effects of restraint and circadian rhythms may still confound results (Mstl and Palme, 2002).

1.5.3 Using faeces to measure cortisol In contrast to plasma and salivary cortisol concentrations, faecal cortisol metabolite concentrations are less responsive to small changes in diurnal fluctuations of circulating cortisol, and reflect cortisol concentrations from 10 to 12 hr prior to faecal collection (Mstl et al., 2002); faecal sampling is reported to be a feedback-free method of measuring HPA axis activation as it is a less invasive procedure and is easier to collect than saliva and plasma (Mstl and Palme, 2002). Faecal cortisol metabolite analysis represents an integrated hormone profile of HPA activity and is not solely a point measure like when

17

analysing from saliva and blood (Sheriff et al., 2010). Due to rapid digestion of glucocorticoids, faecal cortisol metabolite concentrations must be measured instead of true cortisol concentrations (Palme et al., 1996); 11, 17-dioxoandrostanes, a group of glucocorticoid metabolites, have been found to parallel cortisol concentrations found in blood (Mstl et al., 2002). A disadvantage of using faeces to measure the stress response is that the time period between the application of a stressor and collection of the faeces is not easily controlled because it relies heavily on the passage rate of digesta (Negro et al., 2004).

1.5.4 Using hair to measure cortisol Sampling concerns and the lack of a reliable method for measuring chronic stress in farm animals has contributed to the need for non-invasive cortisol sampling methods which can reflect long-term changes in cortisol (Moberg and Mench, 2000); this pressure has resulted in the development of methodologies for the use of hair cortisol as a measure of chronic stress. Hair assays have been used for numerous purposes such as DNA extraction (Morin et al., 1994), drug tests, and measurement of trace metals and sex hormones (Wheeler et al., 1998). Hair cortisol has been demonstrated as a measure of chronic stress in rhesus macaques (Davenport et al., 2008), rock hyraxes (Koren et al., 2008), dairy cows (Comin et al., 2011), and dairy calves (Gonzlez de la Vara et al., 2011). Hair cortisol concentrations represent the accumulation of the free fraction of cortisol from the blood as the hair grows; the slow incorporation of cortisol into hair provides a long-term endocrine profile that is a measure of the hormonal activity of the animal averaged

18

over a chosen period (Accorsi et al., 2008; Moya et al., 2013). Circadian variations and acute stress responses do not affect hair cortisol concentrations, and thus hair appears to be an ideal medium to measure chronic activation of the HPA axis (Moya et al., 2013). With the emergence of hair cortisol as a new measure of chronic stress, standardized methodology for lactating dairy cows needs to be formulated such that values can be easily compared between animals and over experiments. Moreover, the relationship between cortisol concentrations in hair and common stressors (i.e.: lameness, metritis, heat stress, etc.) that affect dairy cows under current management systems are not yet clear.

1.6 Important concepts when using hair for the measurement of stress As using hair as a medium to measure cortisol concentrations is quite new in the field of dairy cattle science, it is important to understand how cortisol is incorporated into the hair and which factors may affect the incorporation and extraction of cortisol from hair. These topics are discussed below.

1.6.1 How is cortisol incorporated into hair? Several mechanisms have been proposed as to how cortisol is incorporated into hair. The most common hypothesis, which has been used to explain the incorporation of drugs into human hair, is that the free fraction of cortisol enters the hair at the medulla of the hair shaft through passive diffusion from the blood (Pragst and Balikova, 2006). It has also been hypothesized that cortisol may be deposited into the completed hair shaft by sebum or

19

sweat (Pragst and Balikova, 2006); however, extensive research has not been done to confirm this theory (Russell et al., 2012). The free fraction of cortisol is diffused from the blood capillaries, which bathe the hair follicle, into the growing cells within the length of the follicle, between the basement membrane and the end of the keratinization zone; once hair has completed the keratinization process, substances from the blood can no longer diffuse into the hair shaft (Pragst and Balikova, 2006). As the hair grows, the cells that make up the hair shaft are moved further and further way from the follicle, creating a map of endocrinological events (Russell et al., 2012). Therefore, the rate of hair growth can be used to associate substances found within a particular segment of hair to a specific time period; it is important to note that since there is hair between the follicle and surface of the skin, the collected hair sample has a time delay that is directly related to hair growth rate and the depth of the hair follicle (Russell et al., 2012). 1.6.2 What factors affect hair cortisol concentrations? With this emerging measure of chronic stress, appropriate hair cortisol sampling methodology must be developed and validated, including ideal hair colour, body location and processing methods for hair samples. Bennett and Hayssen (2010) reported cortisol concentrations differed depending on hair colour in dogs, where black hair contained the least amount of cortisol compared to blonde and agouti coloured hair. There has been limited research on the relationship of hair cortisol and hair colour in dairy cattle. However, a study by Gonzlez-de-la-Vara et al. (2011) reported that white hair contained higher cortisol concentrations when compared to black hair.

20

There have been reports that suggest that the location on the body may affect the amount of cortisol found within hair samples, but previous publications using dairy cattle, hair has been sampled from different locations, including the forehead (Comin et al., 2011), ribs (Gonzalez-de-la-Vara et al., 2011), and shoulder (Maiero et al., 2005). Macbeth et al. (2010) reported that hair collected from the neck of grizzly bears had significantly higher cortisol concentrations than hair collected from the shoulder, rump, and abdomen. Furthermore, recent research in beef cattle has shown that cortisol concentrations are highest in hair collected from the tail switch compared with the head and shoulder, while hair from the neck and hip had higher concentrations compared with the shoulder (Moya et al., 2013). Additionally, there have been different methods used for processing hair samples prior to cortisol extraction. Two main methods reported in the literature are mincing the hair into small pieces using scissors (Accorsi et al., 2008; Koren et al., 2002) and the use of a ball mill to create a powder from the hair (Moya et al., 2013). There has been no research published comparing the effects of different processing methods on the concentrations of cortisol that can be extracted from hair of dairy cows.

1.7 Objectives and hypotheses This thesis reports research investigating methodology for the collection and processing of hair from lactating Holstein dairy cows for hair cortisol analysis, and associations between hair cortisol concentrations and reproductive success, clinical and subclinical disease. We hypothesized that hair cortisol concentrations would be affected by

21

hair colour, sampling location, and processing method. In addition, we hypothesized that animals with greater hair cortisol concentrations will have poorer fertility and a higher prevalence of disease.

22

Chapter 2: Factors Affecting Hair Cortisol Concentrations in Lactating

Dairy Cows1

2.1 Introduction Cortisol is a key mediator of the stress response in animals, produced through the activation of the HPA axis during a stressful event. Activation of the HPA axis can cause many physiological changes during acute stress events (i.e. short term or low magnitude increases in cortisol) and promotes a shift of resources towards adaptive functions such as arousal, vigilance, and increased oxygenation and nutrition of the brain, heart and skeletal muscles. As a consequence, it may down regulate non-adaptive functions such as eating, growth, and reproduction (Chrousos, 2009). Many common farm practices such as handling (Cook et al., 2000), transportation (Lay et al., 1996), regrouping (Friend et al., 1977), dehorning (Stafford and Mellor, 2005), and overstocking (Friend et al., 1979) induce acute stress responses in dairy cattle. Prolonged or excessive activation of the HPA axis may result in chronic stress that can cause additional behavioural (Chrousos and Gold, 1992) and somatic disorders (Chrousos et al., 1998; Charmandari et al., 2005). In dairy cattle, long-term increases in glucocorticoid concentrations have been linked to immunosuppression (Roth, 1985), and reduced fertility (Dobson and Esslemont, 2002). Comin et al. (2013) reported that cows that were clinically compromised (e.g.: laminitis, 1 A version of this chapter has been accepted for publication: Burnett, T.A., A.M.L. Madureira, B.F. Silper, A. Nadalin, A. Tahmasbi, D.M. Veira, R.L.A. Cerri. 2014. Short communication: Factors affecting hair cortisol concentrations in lactating dairy cows. J. Dairy Sci. In Press.

23

metritis or mastitis) had greater concentrations of cortisol than those that had an absence of disease. Cortisol has been used extensively as a physiological measure of acute stress response in animals (Moberg and Mench, 2000). Cortisol can be analyzed from many different media, most notably from the blood, saliva and faeces; however, handling and restraint of dairy cattle has also been shown to rapidly increase plasma cortisol, leading to confounding results (Cook et al., 2000). In addition, circulating cortisol follows a diurnal pattern in a pulsatile fashion, leading to substantial individual animal variation depending on the time of day (Thun et al., 1981). Although fecal samples are less responsive to small changes in circulating cortisol, and reflect cortisol concentrations from 10 to 12 hr prior to fecal collection, they still cannot be practically used to measure the long-term changes that are characteristic of chronic stress (Mstl et al., 2002). Sampling concerns and the lack of a reliable method for measuring chronic stress in farm animals, has contributed to the need for non-invasive cortisol sampling methods that can reflect long-term increases in cortisol (Moberg and Mench, 2000). One promising sampling method is the use of hair to assess cortisol levels; this method has been used in rhesus macaques (Davenport et al., 2008), dairy cows (Comin et al., 2011), and dairy calves (Gonzlez-de-la-Vara et al., 2011). Hair cortisol concentrations have been found to have a positive correlation with chronic stress found in humans in different adverse circumstances and uncertain health, such as hospitalized neonates (Yamada et al., 2007), chronic pain in adults (Van Uum et al., 2008), acute myocardial infarction in men (Pereg et al., 2011) and Cushings syndrome (Thomson et al., 2010).

24

Bennett and Hayssen (2010) reported that cortisol concentrations differed depending on hair colour in dogs, where black hair contained the least amount of cortisol compared to blonde and agouti coloured hair. There has been some evidence that the location on the body where the hair sample originates from may affect cortisol concentrations. Macbeth et al. (2010) found that hair collected from the neck of grizzly bears had significantly higher cortisol concentrations than hair collected from the shoulder, rump, and abdomen. In addition, research with beef cattle has shown that cortisol concentrations were highest when hair was collected from the tail switch (Moya et al., 2013). Furthermore, different methods have been used to process hair samples prior to analysis. The two main processing methods used include mincing the hair into small pieces using scissors (Accorsi et al., 2008; Koren et al., 2002) and using a ball mill to powder the hair (Moya et al., 2013). The aim of this study was to determine the effect of hair colour and sampling location on concentrations of cortisol in lactating Holstein dairy cows, and to identify the most effective processing method for cortisol extraction from hair. The hypothesis was that white-coloured hair would yield greater concentrations of cortisol and that a location effect would be evident. Furthermore, more finely processed hair, such as by using a ball mill, would have greater amounts of cortisol extracted from each sample.

2.2 Materials and methods This project was conducted at the University of British Columbias Dairy Education and Research Centre in Agassiz, BC from March 2011 to October 2011 in Experiment 1 (n =

25

18), July 2011 to January 2012 in Experiment 2 (n = 12), and March 2011 to March 2012 (n = 37) in Experiment 3. The average temperature and photoperiod during the entire experimental period were 10.5 C and 14:38:43 (hh:mm:ss) hr light/d, 17.8 C and 14:38:47 hr light/d, 7.4 C and 9:44:27 hr light/d, and 4.9 C and 9:44:59 hr light/d for the spring, summer, fall and winter seasons, respectively. All procedures were carried out in accordance with the University of British Columbias Animal Care Committee. In addition, animals used in this experiment were cared for as outlined by the guidelines provided by the Canadian Council of Animal Care.

2.2.1 Animals and housing All animals within this study were housed indoors within the same free stall barn equipped with deep sand bedded stalls. Animals were milked twice daily at 0500 and 1500 hr and fed a total mixed ration (TMR) twice daily at approximately 0700 and 1600 hr. The TMR was formulated to meet or exceed the requirements of Holstein cows weighing 620 kg, producing 45 kg/d of 3.5% fat corrected milk (NRC, 2001); the animals had ad libitum access to both TMR and water. An initial set of animals consisted of 18 lactating Holstein dairy cows were used to test the effect of hair colour, sampling location and processing method on hair cortisol concentrations. The animals had an average of 3.2 1.9 (Mean SD) lactations (3 primiparous and 15 multiparous), producing 11,720 2,214 kg of milk, and an average of 163.0 123.9 days in milk (DIM). All 18 dairy cows were sampled at 4 different locations: shoulder, top line, hip and tail switch (Figure 2.1); samples of both black and white hair

26

were taken at each location when available with the exception of the tail switch because Holstein cattle only have white hair at this location. Hair collected from the body of the animal was carefully clipped with electric clippers (blade size 40, Oster, Rye, NY; 2 Speed Clipper, Andis, Sturtevant, WI). Hair from the tail switch was collected with scissors, due to its coarse nature, and the hair material closest (2.5 cm of hair) to the skin was kept for analysis. In both cases, the hair was cut as close to the skin as possible and stored at room temperature in the dark, in dry paper envelopes until processed. An additional 12 dairy cows were used to determine the rate at which hair grows in three different areas of the body: shoulder, hip, and tail switch. Only white hair was collected from each region. Hair from each region was cut as close as possible to the skin using electric clippers, for the rump and shoulder, or surgical scissors for the tail switch. Leftover hair stubble was dyed using black hair dye for a minimum of 30 min and washed off with cold water. Three weeks later, using a permanent red marker, hair was marked at the surface of the skin and plucked out using rat tooth tweezers. Hair length was measured using a light table and a jewellers eyepiece equipped with a small ruler; 10 to 12 hairs/ location/cow were measured to the nearest 0.1 mm. The amount of hair found between the dyed black hair and the red marker was defined as the growth length. A third set of animals was collected to test for the effect of parity and DIM. Hair samples were collected from the tail switch of 37 lactating cows (18 primiparous and 19 multiparous) with no history of health disorders at 0, 21, 42, 84 and 126 DIM. All samples were processed using a ball mill following the previously described hair collection protocol. The sequential collection of hair samples occurred every 3 wk from calving until 126 DIM

27

to test if samples could be properly collected and analyzed over short-term intervals from the same animal. 2.2.2 Hair processing and analysis In preparation for analysis, samples were cleaned thoroughly of dirt, dander and hairs of the wrong colour by hand using tweezers. The samples were then washed with warm water for no longer than 3 min and let dry for a minimum of 1 d until completely dry. To prepare the hair samples used to test the effects of hair colour, sampling location and processing method, two 250 mg of hair sample were weighed and washed in two 5 ml washes of isopropanol of 3 min each, as suggested by Davenport et al. (2006). Both samples were left to dry for 5 d and one was finely cut (less than 3 mm) using a pair of surgical scissors, while the other was ground in 10 ml stainless steel milling cups with a 12 mm stainless steel ball in a Retsch Mixer Mill MM400 ball mill (Retsch, Hannover, Germany) for 5 min at a frequency of 30 repetitions/s. Hair samples collected to test the effect of parity and DIM were prepared for analysis in the same manner as previously described; however, all samples were ground using the ball mill method only. Ground hair samples were then stored in the dark in glass jars at room temperature.

Extraction of hair cortisol was performed following the procedure of Koren et al. (2002), with modifications. We took 20 0.2 mg of clean, dried and ground bovine hair and weighed it into 7 ml glass scintillation vials. Then, 1 ml of HPLC-grade methanol (EMD Chemicals, Darmstdat, Germany) was added. The vials were tightly capped, gently swirled to ensure the hair came into contact with the methanol, and sonicated for 30 min. The

28

samples were incubated overnight (New Brunswick Scientific Incubator Shaker, Edison, NJ) at 100 rpm and 50 oC to extract the steroids and 0.8 ml of the original volume of methanol was pipetted into a 2 ml microcentrifuge tube and evaporated at 45 oC under a stream of ultrapure nitrogen gas. The samples were reconstituted in 100 ul of the phosphate buffered saline supplied with the assay kit. Hair cortisol was analyzed using a commercially available assay kit designed for salivary cortisol (Salimetrics Expanded Range, High Sensitivity 1-E3002, State College, PA). The minimal concentration of cortisol that can be distinguished from the 0 standard using this kit is 0.003 ug/dl. Samples were aliquoted into wells in duplicate (25 ul), and absorbance measured using a wavelength of 450 mm in a microplate plate reader (Biorad xMark, Hercules, CA). The average inter- and intra-assay CV was 7.4% and 3.5%, respectively.

2.2.3 Statistical analyses The key hypothesis for study 1 was that hair colour would have an effect (black > white) on hair cortisol concentrations. The acceptable difference () was defined as 0.50 (50%) with a one-sided test ( = 0.05) and power (1-) = 0.8. Using these assumptions, at least 10 cows were required (Minitab 17; Minitab Inc., State College, PA). Prior to statistical analyses, all data were checked for normality using probability distribution plots. Statistical analyses were performed using SAS (version 9.4; SAS Institute Inc., Cary, NC). To test for differences in cortisol concentrations for hair colour, location on the body, and processing method, as well as the analysis for sequential collections (parity

29

and DIM) data were analyzed by ANOVA using the MIXED procedure. Differences between processing methods only used samples of white hair. The hair growth measurements were pooled within animal and analyzed by ANOVA using the GLM procedure. 2.3 Results and discussion The overall hair cortisol concentrations measured within this experiment (5.7 1.7 pg/mg; Mean SD) were less than reported for lactating dairy cattle by Gonzlez-de-la-Vara et al. (2011) (12.15 1.85 pg/mg), but greater than those reported by Comin et al. (2011) (2.5 0.10 pg/mg) from dairy cattle and Moya et al. (2013) (2.35 0.176 pg/mg) from beef cattle. Values obtained from this study may be greater than previously found by Comin et al. (2011) due to the difference in hair processing method used. Furthermore, the values reported here were greatly different than those found in dairy heifer calves (114.5 14.43 pg/mg; Gonzlez-de-la-Vara et al., 2011), and male rhesus macaques (110.3 10.2 pg/mg; Davenport et al., 2006), suggesting that developmental stages and species greatly affect the amount of cortisol found in hair.

A difference in cortisol concentrations was found for hair of different colours (P < 0.001; Table 2.1). White hair had greater concentrations of cortisol than black hair. This difference was found irrespective of the location and the analysis excluded the data from the tail switch because of the lack of black hair. Similarly, Gonzlez-de-la-Vara et al. (2011) also described black hair as having approximately half the concentration of cortisol than white hair from dairy cows. In addition, differences in hair cortisol affected by colour have been reported in dogs, where yellow hair had greater concentrations of cortisol than agouti

30

and black hair from German shepherds (Bennett and Hayssen, 2010). Reasons for this difference are not well understood but could be related to mechanisms associated with melanocyte development (Slominski et al., 2004), differentiation (Roulin et al., 2008), or simply a question of physical space within the hair shaft as white or yellow hair have less pigmentation. Hair colour may have an effect on skin temperature, and therefore increase the blood flow through the microvasculature, compromising the dermal papilla and root sheaths of hair follicles; however, animals in this study were housed indoors, away from direct sunlight. Although it has been suggested that season may effect hair cortisol concentrations (Comin et al., 2011) by changing the rate of hair growth and the hair growth cycle (Courtois et al., 1996), the effects of photoperiod and temperature has not been shown in cattle. Concentrations of cortisol from hair of both colours did not differ among the shoulder, top line and hip (P = 0.36; Table 2.1). When only white was included in the analysis in order to include the tail switch, a difference between the tail switch and shoulder was found (P = 0.01), but not between the tail switch and the hip or top line (Table 2.1). These results for white hair are consistent with a recent study performed on beef cattle by Moya et al. (2013) where it was also reported that hair from the tail switch had greater concentrations of cortisol than the head and shoulder, whereas hair from the neck and hip had greater concentrations of cortisol compared with the shoulder. In this study, the hair from the top line and hip were not different than from the shoulder. The causes for the marked differences between the tail switch and shoulder are unclear; however, it has been suggested that the increased hair growth rate found at the tail switch may be the driving factor for increased cortisol concentrations (Moya et al., 2013).

31

Furthermore, Moya et al. (2013) reported that hair from the tail switch had a stronger correlation to both salivary and fecal cortisol concentrations than hair from the head, neck, shoulder, and hip. The processing method of the hair significantly affected the amount of cortisol extracted from each sample. Samples processed with a ball mill had greater amount of cortisol extracted from each sample (P < 0.001) compared with surgical scissors (Table 2.1). The increase in cortisol extracted from hair is likely caused by the increased surface area created by grinding the hair into much smaller pieces. Processing methods have not previously been directly compared for cortisol extraction from hair samples from dairy cows, but studies using surgical scissors have reported overall smaller concentrations of cortisol (Comin et al., 2011; Comin et al., 2013). The rate of hair growth is affected by location. The growth rate of the hair from the tail switch is over ten times faster (P < 0.001) than the rate of hair growth at the hip and shoulder, whereas the hip and shoulder have similar hair growth rates (Geometric Mean [95% CI]) (0.51 [0.45, 0.60] vs. 0.04 [0.03, 0.05] vs. 0.03 [0.03, 0.04] mm/d; Figure 2.2). The hair growth results from the tail switch were similar to values reported by Fisher et al. (1985) and Schwertl et al. (2003) from beef cattle. Resampling intervals of biological significance in lactating dairy cows, such as during the transition period, can only be achieved with the growth rate found at the tail switch. Parity only tended to affect concentrations of cortisol in hair, where multiparous cows had slightly greater concentrations than primiparous cows (9.6 [9.1, 10.2] vs. 8.9 [8.4, 9.4]; P = 0.10). Such a small difference is unlikely to indicate a clear parity effect, although it

32

is common belief that primiparous cows may suffer more from stress than multiparous cows. Within this portion of the experiment, hair was only collected from cows with no sign of clinical disease throughout the sampling period. In a sister study (Chapter 2), it was observed that parity had a more pronounced effect, even when only including clinically healthy animals. In addition, as DIM increased, a decrease in hair cortisol concentrations was observed (P < 0.001) (Figure 2.3), indicating that hair cortisol may be a valuable indicator of changes in stress during the weeks following parturition. No interaction between parity and DIM was found (P > 0.16). The colour of the hair, sampling location and processing method largely affects the amount of cortisol extracted from hair. This study suggests that the tail switch is the ideal location for hair sampling from black and white Holsteins for analysis of cortisol due to the consistent white colour hair, greater growth rates and easier access. In addition a ball mill should be used to maintain consistent particle sizes among samples as well as to improve cortisol extraction. The creation of a more standardized method for hair sample collection for the analysis of cortisol in dairy cattle will improve the interpretation and consistency of results in different studies aiming to compare chronic stress in lactating dairy cows.

33

Table 2.1: Effect of hair colour1, sampling location1 and processing method on concentrations of cortisol extracted from hair samples collected from black and white Holstein cows. Hair Cortisol Concentration Mean [95% CI] (pg/mg) P-value Colour Black 4.2 [3.6, 5.0] < 0.001 White 7.8 [6.8, 9.2] Location (black and white hair) Shoulder 4.8 [1.4, 5.6] 0.36 Top line 5.1 [1.5, 6.0] Hip 5.8 [1.6, 6.9] Location (only white hair) Shoulder 6.2 [4.2, 9.2]a

0.03 Top line 8.9 [5.8, 13.8]ab Hip 9.5 [6.3, 14.3]ab Tail switch 11.0 [7.6, 16.0]b Processing method Ball mill 10.4 [5.8, 18.8] < 0.001 Surgical scissors 4.7 [2.6, 8.4] ab Values with different superscripts in the same column are significantly different (P < 0.05). 1 All comparisons of colour and location were done using the ball mill processing method.

34

Figure 2.1: Hair sampling locations: a) shoulder, hair located at the scapula; b) top line, hair located at the dorsal side of the vertebrae; c) hip, hair located over the femur-ischium junction; and d) tail switch, hair located at the distal end of the tail.

Domicoolka/ Dollar Photo Club

c

d

b a

35

Figure 2.2: Hair growth rates at different body locations (Mean SE) collected from black and white Holstein cows. Different superscripts (a,b) indicate significant differences (P < 0.001).

0 4 8

12

Tail Hip Shoulder Hair Gr

owth Rate (m

m/2

1d)

Hair Sampling Location

a

b b

36

Figure 2.3: Effect of parity and DIM on concentrations of cortisol (Mean SE) extracted from hair samples collected from the tail switch of 37 clinically healthy lactating cows (18 primiparous and 19 multiparous) at 0, 21, 42, 84 and 126 DIM. There was a tendency for greater hair cortisol concentrations in multiparous cows (P = 0.10). Days in milk affected concentrations of cortisol in hair (P < 0.001); d 0 and 21 had elevated concentrations of cortisol compared with the other collections. No interaction between parity and DIM was found (P > 0.16).

4 6 8

10 12 14

0 20 40 60 80 100 120

Cortisol Con

centration

(pg/mg)

Days in Milk

Multiparous Primiparous

37

Chapter 3: Relationship of Concentrations of Cortisol in Hair with Health,

Plasma Metabolites and Reproductive Parameters in Dairy Cows2

3.1 Introduction Chronic stress, characterized as a sustained stress response or the excessive secretion of stress mediators, has been shown to impair the immune response and reproductive function in animals (Chrousos, 2009; Matteri et al., 2000). In cattle, several reports have shown greater concentrations of cortisol in plasma in different scenarios involving clinical diseases and inflammatory responses. Comin et al. (2013) reported greater concentrations of cortisol in plasma from clinically compromised (e.g.: laminitis, metritis or mastitis) cows compared with those with an absence of disease. Additionally, Lavon et al. (2008) observed that animals administrated with E. coli endotoxin showed an increase in cortisol in plasma suggesting that an inflammatory response induces increases in cortisol. Some researchers have reported that an inflammatory infection caused by E. coli endotoxins administered into the uterine lumen has negative impacts on reproduction which suggests this may be due to the release of cortisol as a result of the inflammation (Lopez-Diaz and Bosu, 1992; Huszenicza et al., 2004). However, it is unclear if subclinical states of common disorders in dairy cows such as subclinical endometritis and mastitis trigger the same corticotrophin-releasing hormone (CRH) responses from the hypothalamus. In addition, increased concentrations of cortisol are known to cause immunosuppressive effects in dairy cattle and swine (Hopster et al., 1998; Tuchscherer et 2 A version of this chapter has been submitted for publication: Burnett, T.A., A.M.L. Madureira, B.F. Silper, A. Nadalin, A. Tahmasbi, D.M. Veira, R.L.A. Cerri. 2014. Relationship of concentrations of cortisol in hair with health, plasma metabolites and reproductive parameters in dairy cows.

38

al., 2004) and are linked to decreases in anti-body concentrations and reductions in the phagocytic functions in monocytes and macrophages (Roth, 1985). Chronic and acute stress can trigger disruptions in reproductive function. Chronic lameness in cattle, for example, is linked to disruptions in the pulsatile pattern of GnRH and consequent impairment of estrous behaviour expression (Walker et al., 2008). Acute stress, characterized as short term and normally measured through the blood, saliva and faeces, has detrimental effects on ovarian cyclicity by disrupting the normal release of hormones from the hypothalamus-pituitary-gonadal (HPG) axis that controls reproduction (Dobson et al., 2003; Smith et al., 2003). Stressors such as isolation, transport and restraint (acute causes) have been found to interfere with the endocrine events preceding ovulation and thus resulting in ovulation failure (Moberg, 2000). Most of the literature in dairy cattle concerning cortisol measurements has been done in plasma (Forslund et al., 2010), saliva (Negro et al., 2004) or faeces (Huzzey et al., 2011). Alternatively, the measurement of cortisol in hair has recently received attention and has the advantage of being a non-invasive procedure; however no research has demonstrated the dynamic changes on how disease (clinical and subclinical) events and reproductive state modify the concentrations of cortisol in hair over time. In addition, the relationship between cortisol in hair, energy metabolites and acute phase proteins is currently unclear and can provide valuable information about the validity of hair cortisol tests. The aim of this study was to determine the associations between cortisol in hair, as a measure of chronic stress, with clinical and subclinical health disorders, and

39

reproductive success in dairy cows. Furthermore, this study determined the association between cortisol in hair with energy metabolites in blood (glucose and BHBA) and acute phase proteins in plasma (ceruloplasmin and haptoglobin). It was hypothesized that (a) the concentration of cortisol found in the hair of dairy cows has a negative relationship with the health and reproductive success of dairy cows, and (b) there is a positive association of hair cortisol concentrations with concentrations of BHBA, glucose, ceruloplasmin and haptoglobin in plasma during the transition period. 3.2 Materials and methods This experiment was conducted between June 2011 and March 2012 at the University of British Columbias Dairy Education and Research Centre, Agassiz, BC. All procedures were approved by the Animal Care Committee of the University of British Columbia. Animals used in this experiment were cared for as outlined by the guidelines provided by the Canadian Council of Animal Care (2009). 3.2.1 Animals and housing A total of 118 high producing Holstein dairy cows were used in this study which consisted of two experiments (Experiment 1: n = 64; Experiment 2: n = 54). Cows produced a mean SD of 12,236 2,219 kg of milk (305-d mature-equivalent yield) and a range of body condition score (BCS) from 1.75 3.25 at 30 3 DIM. All animals were housed in the same naturally ventilated wooden-framed barn with a free stall design, equipped with deep sand bedded stalls. Animals were milked twice daily at 0500 and 1500 hr with automatic

40

milking machines. Fresh TMR was delivered twice daily at approximately 0700 and 1600 hr. The TMR was formulated following the NRC guidelines (NRC, 2001) to meet or exceed the requirements of a 620 kg Holstein cow producing 40 kg/d of 3.5% fat corrected milk; the animals had ad libitum access to both TMR and water. During both experimental periods, various potential forms of stressors were recorded, including retained placenta, clinical hypocalcaemia, displaced abomasum, bouts of clinical mastitis and metritis, surgical procedures, and milk yield (305-d mature-equivalent yield). All animals were body condition scored at 30 3 DIM on a 5-point scale from thin (1) to obese (5) by increments of 0.25 as outlined by Edmonson et al. (1989). Animals were later classified as Thin (BCS < 2.75), Average (BCS = 2.75) or Moderate (BCS > 2.75). All health and production information was collected with the assistance of the dairy herd personnel and the herd veterinarian, and recorded using the on-farm Dairy Comp 305 software (Valley Agricultural Software, Tulare, CA).

Experiment 1 Sixty-four Holstein dairy cows were enrolled in Experiment 1, including 20 primiparous and 44 multiparous cows. Each cow had hair sampled from the tail switch on the day of calving, and was resampled every 3 wk (21 3 DIM) until 126 3 DIM; however, only samples on 0, 21, 42, 84 and 126 DIM were subsequently analyzed. Blood samples were collected at 0, 21, and 42 DIM and analyzed for glucose and BHBA. A visual schematic of the sampling schedule can be found in Figure 3.1. Glucose and BHBA were used to investigate possible correlations of these metabolites and negative energy balance during

41

the transition period with hair cortisol samples. Cows were submitted to the regular reproductive program of the Dairy Centre that consisted of a pre-synchronization with the administration of two injections of PGF2 14 d apart at 30 3 and 44 3 DIM and artificial insemination (AI) performed upon spontaneous estrus from 50 until 100 DIM; if estrus was not detected by 100 DIM, cows were submitted to the ovsynch program (Pursley et al., 1997). At the time of each pre-synchronization injection animals were examined by per rectum palpation using ultrasonography for cyclicity, diameter of the inner layer of the cervix and asymmetry of the diameter of the endometrium between both uterine horns and pregnancy diagnosis at 40 7 d after AI; a more detailed description is included below. Animals were classified as being cyclic if they had a presence of at least one corpus luteum at the time of either pre-synchronization injection. Estrous detection was carried out using a combination of visual observation and the use of automated activity monitors (Heatime; SCR Engineering Ltd., Netanya, Israel). Data on the reproductive success and episodes of health disorders of the animals were recorded. Animals were classified as Healthy if they had an absence of clinical disease during the entire experimental period, or as Clinically Diseased if they were diagnosed with one or more clinical disease event. Sampling methodologies are detailed below. Experiment 2 A different set of 54 Holstein dairy cows was used for Experiment 2, consisting of 24 primiparous and 30 multiparous cows. Animals were chosen retrospectively by parity and divided into two groups according to their health status: diagnosed with subclinical endometritis (Endo) or healthy (control). All animals were enrolled at 30 3 DIM. Samples

42

of hair and blood were collected and a uterine health diagnosis was obtained from each cow using a cytobrush and ultrasonography (Figure 3.1). Subclinical endometritis was positively diagnosed when the proportion of neutrophils on the uterine smear was > 18 % in relation to the total number of cells (Kasimanickam et al., 2004); all animals in the Endo group were also diagnosed with some uterine luminal fluid (ecogenic or non-ecogenic fluid) by the ultrasound examination. Uterine smears with < 18% of neutrophils and absence of ecogenic fluid in the lumen of the uterus were classified as healthy and placed in the Control group; all animals in the Control group had an absence of clinical disease. Blood samples were analysed for acute phase proteins, haptoglobin and ceruloplasmin, and hair was analysed for cortisol. Sampling methodologies are detailed below. 3.2.2 Blood sampling and analysis All blood samples were drawn by puncture of the median coccygeal vein or artery utilizing untreated Vacutainer tubes (Becton Dickinson Vacutainer Systems, Rutherford, NJ) for Experiment 1 and K2EDTA Vacutainer tubes (Becton Dickinson Vacutainer Systems, Rutherford, NJ) for Experiment 2. For Experiment 1, whole blood was immediately analyzed for glucose (Precision Xtra blood glucose kit; Abbott Diabetes Care, Alameda, CA) and BHBA (Precision Xtra blood ketone kit; Abbott Diabetes Care, Alameda, CA) concentrations using the procedures described by Iwersen et al. (2009). For Experiment 2, analyses for haptoglobin and ceruloplasmin were performed using the commercially available PHASE Haptoglobin kit (Tridelta Development Ltd., TP801, Maynooth, CK,

43

Ireland) and using a method outlined by Demetriou et al., (1974) determined by the rate of p-phenylenediamine oxidation, respectively.

3.2.3 Hair sampling and analysis All hair samples collected for this study were harvested from the tail switch as described by Burnett et al. (2014). For the first hair collection in either experiment, the hair was cut as close to the skin as possible using surgical scissors, and the 3 cm closest to the skin was kept for analysis. For Experiment 1, all subsequent hair samples were harvested from the same region and the total regrown hair was collected and used for analysis; only samples at 0, 21, 42, 84, 126 DIM were analysed. For Experiment 2, only one sample was collected at 30 3 DIM. Once the hair was collected, it was stored at room temperature in dry, non-translucent envelopes in a dark room until analysis. Hair samples were cleaned, dried and ground according to the procedures outlined in Burnett et al. (2014). Grinding was performed on 250 mg of sample in a 10 ml stainless steel milling cup with a 12 mm stainless steel ball in a Retsch Mixer Mill MM400 ball mill (Retsch, Hann, Germany) for 5 min at a frequency of 30 reps/sec. Ground hair samples were stored in the dark in glass jars at room temperature. Extraction of hair cortisol was performed using methanol following the procedures of Burnett et al. (2014), which were adapted from Koren et al. (2002). Hair cortisol concentrations were determined using a commercially available assay kit designed for salivary cortisol (Salimetrics Expanded Range, High Sensitivity 1-E3002, State College, PA), as previously performed by Davenport et al. (2006) and Bennett and Hayssen (2010). The

44

sensitivity of the cortisol kit was 0.01 ug/dl. Samples were analyzed for cortisol content following the instructions supplied by the manufacturer. Duplicate 25 ul samples were aliquoted into wells, and absorbance of the samples was read at 450 nm using a microplate plate reader (Biorad xMark, Hercules, CA). Results were calculated using the microplate reader software using a 4 parameter logistic curve fit. Results were converted from ug/dl to pg/mg. The average intra assay CV between samples on the plates was 3.51%.

3.2.4 Uterine cytology and ultrasonography Uterine health monitoring was performed by the method of uterine cytology and ultrasound. Cytological examination of the endometrium was performed by way of a cytobrush (VWR Canlab, Mississauga, ON). The vulva was cleaned of any debris and the cytobrush, threaded onto a steal rod and covered with an outer steel tube, was inserted into the vagina and through the cervix. With contact of the uterine wall, the cytobrush was extended from inside the steel tube and immediately rotated to collect a sample. The cytobrush was then retracted inside the steel tube prior to removal from the uterus so as to not contaminate the sample. Once retracted, the cytobrush was rolled onto a glass microscope slide to make an endometrial smear. Smears were air-dried and stained using a Romanowsky stain (Diff-Quick, Fisher Diagnostics, Middletown, VA). The slides were examined to determine the percent neutrophils. To do so, three sections of 100 cells each were counted for each slide, and the total number of neutrophils and epithelial endometrial cells were tallied. Subclinical endometritis was defined when the proportion of neutrophils

45

was 18% in relation to the total number of the cells (Kasimanickam et al., 2004; Sheldon et al., 2006). After the uterine sample collection, ultrasonography was performed on the uterus via rectal palpation using an ultrasonographic machine (Aloka SSD-500, Aloka Co Ltd., Wallingford, CT) equipped with a 7.5 MHz linear transducer. The following uterine health measures that have previously been demonstrated to be associated with subclinical endometritis (Cerri et al., 2012) were measured: diameter of the inner layer of the ce