The dramatic increase in productivity achieved in the second half of the 20th century has resulted in concomitant reduction in animal fitness, which increases susceptibility to infectious diseases, metabolic diseases, locomotion problems and subfertility (Koolhaas and Van Reenen, Reference Koolhaas and Van Reenen2016; Friggens et al., Reference Friggens, Blanc, Berry and Puillet2017; Colditz, Reference Colditz2018; Gabai et al., Reference Gabai, Amadori, Knight and Werling2018). ‘Fitness’ can be defined as the ability to transfer the genes that an individual, a genotype, a population or a species carries, although this definition is the subject of extensive discussion and debate (Barker, Reference Barker, Van der Werf, Graser, Frankham and Gondro2009). Two main features contribute to fitness: reproductive success and survival rate. In the context of farm animals, the term fitness has been often used as a synonym of ‘robustness’, which refers to traits such as fertility, disease resistance, health and longevity (Koolhaas and Van Reenen, Reference Koolhaas and Van Reenen2016). According to Friggens et al. (Reference Friggens, Blanc, Berry and Puillet2017) ‘robustness’ can be defined as the ability to continue activities that favour animal fitness despite environmental constraints.
Within the same population, behavioral and physiological responses to identical conditions may differ among individual animals, possibly leading to differences in health and reproductive success outcomes. Modern farm animals exhibit individual variation in the way they respond to environmental challenges and can express differential behavioral and physiological response patterns (Koolhaas and Van Reenen, Reference Koolhaas and Van Reenen2016; Colditz, Reference Colditz2018). Animal brains generate expectations derived from previous experiences, rather than operating merely as a stimulus-response network, and this view is applicable to both external and internal environment. Therefore, the farm environment can be very complex, and an animal needs to be able to predict and control the cognitive, affective, physical and immunological components of the environment in order to survive, thrive and reproduce (Colditz, Reference Colditz2018). It is worth noting that the affective and mental state has an important implication for animal fitness (or robustness) and welfare (Koolhaas and Van Reenen, Reference Koolhaas and Van Reenen2016; Colditz, Reference Colditz2018). An inability to predict and control this complex environment may result in ethological and physiological changes generating a stress response (Del Giudice et al., Reference Del Giudice, Buck, Chaby, Gormally, Taff, Thawley, Vitousek and Wada2018).
The endocrine, immune and central nervous systems are responsible for coordinating appropriate behavioral and physiological responses to environmental challenges (Koolhaas and Van Reenen, Reference Koolhaas and Van Reenen2016; Colditz, Reference Colditz2018; Gabai et al., Reference Gabai, Amadori, Knight and Werling2018). Recent data obtained from laboratory animals and human studies highlights the presence of bidirectional interactions between the nervous system, metabolic organs and the immune system. Peripheral nerves and specialized structures within the nervous system can detect immune stimuli, which allows the brain to orchestrate complex behavioral and physiological responses (Soto-Tinoco et al., Reference Soto-Tinoco, Guerrero-Vargas and Buijs2016). Inflammatory processes affect social behavior, leading to changes that may help the individual to adapt to the social environment during times of sickness. Conversely, adverse social behavior affects aspects of immunity and inflammatory state, predisposing the body to situations in which injury and infection will be more likely (Eisenberger et al., Reference Eisenberger, Moieni, Inagaki, Muscatell and Irwin2017).
The concept of an integrated whole organism regulatory network needs to be characterized in dairy animals. In addition, a better understanding is needed of how environmental conditions affect genetic and phenotypic links between temperament, immune function, metabolic performance, affective state, resilience and welfare. Future animal breeding strategies should focus on balancing high production levels with optimal health status and welfare. In this context, the discovery, validation and contextual implementation of biomarkers will contribute to our understanding of physiological processes and regulatory mechanisms used by animals to adapt to new conditions. Biomarkers may also be used to monitor health status and welfare and to establish new management and breeding strategies.
In this article, we review the definition of biomarker and summarize the major achievements in the establishment of health and fitness biomarkers in dairy cows and small ruminants. This review mainly focuses on physiological biomarkers used to monitor animal responses to environmental disturbances. It also highlights future avenues for research in this field, allowing us to present a timely and impactful positioning document to the scientific community.
Scientific definition of biomarker
Biomarkers, also called molecular markers, biochemical markers or signature molecules, are biological molecules used to understand a physiological process or diagnose an abnormal process or a disease. In healthcare, biomarkers are used to diagnose diseases, assess their progression and monitor responses to new treatments, combination treatments and repositioned drugs. In 2001, the Biomarkers Definitions Working Group defined a biomarker as ‘a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention’ (Biomarkers Definitions Working Group, 2001). Different scientific disciplines have refined the original definition provided by the Biomarkers Definitions Working Group. In this paper we present a biomarker definition that is relevant to the major stakeholders in dairy production science.
Emergence of new analytical technologies to improve dairy production
Over the last years, research in the dairy production science field has progressed immensely. Indeed, research in animal and dairy science has moved from a traditional cause-effect perspective, to a broader and more comprehensive view on several fundamental fields such as physiology, health, productivity and welfare, among others. The use of bioinformatics and high-throughput technologies, namely ‘omics’ techniques such as proteomics and metabolomics, has allowed researchers to perform different experiments in the dairy science field in a more consistent and comprehensive manner. Proteomics can be defined as the analytical science that studies proteomes and their functions. The proteome refers to the proteins that exist in a given cell, tissue, organ, organism, population or the proteins that are produced and secreted by cells to form the extracellular matrix and soluble components of biological fluids. Metabolomics is the analytical science that studies the metabolites present within an organism, cell, or tissue. Metabonomics is a subset of metabolomics and is defined as the quantitative measurement of the multiparametric metabolic responses to pathophysiological stimuli or genetic modification, with particular emphasis on the elucidation of differences in population groups due to genetic modification (Ramsden, Reference Ramsden and Ramsden2009). For further details on the subject, kindly refer to our previous reviews on the subject (Eckersall et al., Reference Eckersall, de Almeida and Miller2012; Ferreira et al., Reference Ferreira, Bislev, Bendixen and Almeida2013; Almeida et al., Reference Almeida, Bassols, Bendixen, Bhide, Ceciliani, Cristobal, Eckersall, Hollung, Lisacek, Mazzucchelli, McLaughlin, Miller, Nally, Plowman, Renaut, Rodrigues, Roncada, Staric and Turk2015). The use of these two omics technologies has been crucial in dairy sciences and several proteins and metabolites have been associated to specific study topics. Thus, studies about tolerance to seasonal weight loss in dairy goats (Hernández-Castellano et al., Reference Hernández-Castellano, Argüello, Almeida, Castro and Bendixen2015), insulin resistance in transition dairy cows (Zachut, Reference Zachut2015), response to seasonal heat stress in late pregnant dairy cows (Zachut et al., Reference Zachut, Kra, Livshitz, Portnick, Yakoby, Friedlander and Levin2017) and amino acid composition of milk replacer on muscle metabolites (Yu et al., Reference Yu, Matzapetakis, Valent, Saco, De Almeida, Terré and Bassols2018) have been published over the last few years. Moreover, as newborn survival is an important factor affecting fitness, assessment of colostrum quality and intake represents an important area of investigation in both large and small ruminants (Banchero et al., Reference Banchero, Milton, Lindsay, Martin and Quintans2015; Buczinski and Vanderveerd, Reference Buczinski and Vanderveerd2016). Therefore, the establishment of biomarkers of adequate colostrum quality and intake related to newborn health status is an area that will grow in importance. Examples in this field include for instance the works of Hernández-Castellano et al. (Reference Hernández-Castellano, Almeida, Ventosa, Coelho, Castro and Argüello2014a, Reference Hernández-Castellano, Almeida, Castro and Argüello2014b, Reference Hernández-Castellano, Argüello, Almeida, Castro and Bendixen2015) that used proteomics to establish markers of proper colostrum intake and consequently adequate immune status in lambs.
The application of omics technologies in dairy science and their association with the establishment of biomarkers is of particular interest from commercial and welfare perspectives. However, omics-based research requires specific analytical instrumentation, specialized reagents and technical knowledge which is uncommon for researchers that traditionally work in this area (Almeida et al., Reference Almeida, Bassols, Bendixen, Bhide, Ceciliani, Cristobal, Eckersall, Hollung, Lisacek, Mazzucchelli, McLaughlin, Miller, Nally, Plowman, Renaut, Rodrigues, Roncada, Staric and Turk2015; Boschetti et al., Reference Boschetti, Hernández-Castellano and Righetti2019). These facts may explain the difficulties of omics-based studies in farm animals compared to human medicine and preclinical animal models (e.g. rats and mice), where omics approaches are very commonly employed and the relevant databases are well-established. This situation is even more complex in studies performed in sheep and goats, as the reduced size of databases makes the identification and quantification of proteins and metabolites very challenging (Soares et al., Reference Soares, Franco, Pires, Ventosa, Palhinhas, Koci, Martinho de Almeida and Varela Coelho2012; Almeida et al., Reference Almeida, Bassols, Bendixen, Bhide, Ceciliani, Cristobal, Eckersall, Hollung, Lisacek, Mazzucchelli, McLaughlin, Miller, Nally, Plowman, Renaut, Rodrigues, Roncada, Staric and Turk2015). As previously described, the number of entries in publicly available databases for farm animals is extremely low compared to humans and animal models. Furthermore, homology searches may present some difficulties as there are no ruminant model species. Therefore, new research needs to be done to establish more complete and more extensive databases for farm animals.
Despite all these technical challenges and the practical difficulties associated with biomarker identification and verification, establishing biomarkers for applied ‘field and farm’ applications in animal and dairy sciences is extremely relevant to our community. This requires efficient integration of researchers, veterinarians, extension professionals, legislators, dairy farmers and industry professionals.
Stress biomarkers
Stress occurs when a physiological control system detects a state of real or presumptive threat to the animal's homeostasis or a failure to control a fitness-critical variable internal or external to the organism (Ralph and Tilbrook, Reference Ralph and Tilbrook2016; Del Giudice et al., Reference Del Giudice, Buck, Chaby, Gormally, Taff, Thawley, Vitousek and Wada2018). Unambiguously defining stress is problematic because many factors can act as stressors, and the stress response is multifaceted and multidimensional and possesses an individual component (Romero et al., Reference Romero, Platts, Schoech, Wada, Crespi, Martin and Buck2015). Animals respond to stress by activating a wide array of behavioral and physiological responses, collectively referred to as the ‘stress response’. In animal husbandry, stress is often associated with poor health or hampered welfare. The definition of a stressful condition has mostly relied on glucocorticoids measurement. However, glucocorticoids only represent one stress-response system, the hypothalamus-pituitary-adrenocortical (HPA) axis. Therefore, it seems that this type of response needs to be combined with biomarkers of other physiological systems in order to provide a holistic characterization of the stress response.
Monitoring the HPA axis using different biological matrices
The HPA axis is responsible for initiation of stress responses in all vertebrates, and corticosteroids are widely recognized as stress biomarkers (Mormède et al., Reference Mormède, Andanson, Aupérin, Beerda, Guémené, Malmkvist, Manteca, Manteuffel, Prunet, van Reenen, Richard and Veissier2007; Cook, Reference Cook2012). However, laboratories running cortisol assays find that results are often contradictory. In cattle, several ‘social’ and ‘physical’ stressors can influence the animal's physiological systems. How the individual perceives a specific stressor affects both the intensity and duration of the stress (Mormède et al., Reference Mormède, Andanson, Aupérin, Beerda, Guémené, Malmkvist, Manteca, Manteuffel, Prunet, van Reenen, Richard and Veissier2007; Bova et al., Reference Bova, Chiavaccini, Cline, Caitlin, Matheny, Muth, Voelz, Kesler and Memili2014).
Stressors activate the HPA axis through different pathways and several neuronal circuits from different brain districts (hippocampus, amygdala and the medial prefrontal cortex) converge in the hypothalamus, which secretes the corticotrophin releasing hormone (CRH), the primary hypothalamic factor that trigger HPA axis activation. CRH stimulates corticotrophin (ACTH) release by the pituitary gland that triggers the release of glucocorticoids from the adrenal cortex. Arginine vasopressin (AVP) is another factor that can potentiate CRH effects. The glucocorticoid feedback inhibition plays a prominent role in regulating the magnitude and duration of glucocorticoid release and their physiological and behavioral effects (Smith and Vale, Reference Smith and Vale2006; Mormède et al., Reference Mormède, Andanson, Aupérin, Beerda, Guémené, Malmkvist, Manteca, Manteuffel, Prunet, van Reenen, Richard and Veissier2007; Jankord and Herman, Reference Jankord and Herman2008; McEwen et al., Reference McEwen, Bowles, Gray, Hill, Hunter, Karatsoreos and Nasca2015). Comprehensive studies about HPA axis regulation in dairy animals are still lacking, and most available information is based on studies performed in laboratory animals, or in the relationship between stress and reproduction in farm animals (Ghuman et al., Reference Ghuman, Morris, Spiller, Smith and Dobson2010; Ralph et al., Reference Ralph, Lehman, Goodman and Tilbrook2016).
Considering the complexity of the HPA axis regulation and the variability in the stress response, the suitability of cortisol as a stress biomarker is controversial (Mormède et al., Reference Mormède, Andanson, Aupérin, Beerda, Guémené, Malmkvist, Manteca, Manteuffel, Prunet, van Reenen, Richard and Veissier2007; Otovic and Hutchinson, Reference Otovic and Hutchinson2014; Ralph and Tilbrook, Reference Ralph and Tilbrook2016). Cortisol concentrations increase rapidly in blood when animals are subjected to a specific stressor. As cortisol is commonly measured in plasma or serum samples, immobilization of the animal for blood collection may already trigger the stress response. Moreover, glucocorticoids participate in metabolic regulation independent of stressors, and their release is pulsatile and follows a circadian pattern (Lefcourt et al., Reference Lefcourt, Bitman, Kahl and Wood1993; Marinelli et al., Reference Marinelli, Trevisi, Da Dalt, Merlo, Bertoni and Gabai2007; Ogino et al., Reference Ogino, Matsuura, Yamazaki, Irimajiri, Suzuki, Kushibiki, Singu, Kasuya, Hasegawa and Hodate2014; Otovic and Hutchinson, Reference Otovic and Hutchinson2014). All these factors are significant sources of variability and act as confounding factors. Consequently, it is unlikely that blood cortisol, particularly measured once daily, can be used for detecting cows under stressful conditions. As glucocorticoids can be measured in several biological matrices, less invasive procedures may contribute to avoid stressful conditions at sampling (Mormède et al., Reference Mormède, Andanson, Aupérin, Beerda, Guémené, Malmkvist, Manteca, Manteuffel, Prunet, van Reenen, Richard and Veissier2007; Cook, Reference Cook2012).
Salivary cortisol reflects the free hormone fraction measured in plasma (Negrao et al., Reference Negrao, Porcionato, de Passillé and Rushen2004). In cattle, there is a time lag of approximately 10 min to reach a salivary peak in cortisol compared to plasma (Hernandez et al., Reference Hernandez, Thierfelder, Svennersten-Sjaunja, Berg, Orihuela and Lidfors2014). However, this finding is still controversial (Schwinn et al., Reference Schwinn, Knight, Bruckmaier and Gross2016). A positive correlation between plasma and salivary cortisol is found during an ACTH challenge, but basal cortisol concentrations in saliva are poorly correlated to cortisol in blood during different feeding and drinking procedures (Schwinn et al., Reference Schwinn, Knight, Bruckmaier and Gross2016). It is possible that only acute and intense stress or pain can affect salivary cortisol, while mild or chronic stress and circadian variation can only be observed in plasma. These observations should be considered when salivary cortisol is used for monitoring HPA axis.
Besides saliva, circulating cortisol can accumulate in matrices such as milk, urine, feces and hair over different time intervals. Therefore, cortisol measured in those matrices gives an average representation of the HPA activity during those periods. Milk may be considered as a preferred medium in dairy cows to point out short-term stimulation of the HPA axis (Cook, Reference Cook2012) and this can be measured in real-time, especially in intensive dairy farms that employ robotic milking. Milk cortisol represents the average plasma cortisol concentrations between two consecutive milkings. Social challenges such as group relocation can be detected within herds by measuring milk cortisol (Pošćić et al., Reference Pošćić, Gabai, Stefanon, Da Dalt and Sgorlon2017), suggesting that milk cortisol measurement can detect short- and mid-term HPA axis activation. However, several factors (e.g. herd, breed, somatic cell count) not directly related to the stress exposure can affect milk cortisol concentrations (Sgorlon et al., Reference Sgorlon, Fanzago, Guiatti, Gabai, Stradaioli and Stefanon2015). In the case of intense stressor exposure (ATCH or transportation), cortisol concentrations in milk reflect those in plasma, but are dependent upon the duration of the stimulus and the consequent elevation in blood cortisol, suggesting that cortisol can freely cross the milk-blood barrier following a concentration gradient. Consequently, if an animal is exposed to a mild or moderate stressor and has sufficient time to recover before milking, milk cortisol concentrations may not be able to reflect that specific perturbation (Verkerk et al., Reference Verkerk, Phipps and Matthews1996).
Urine cortisol represents the average free hormone filtered by the kidney between two consecutive emptyings of the bladder, providing that the total volume of urine produced in that period can be measured (Cook, Reference Cook2012). Feces is a potentially useful biological matrix to measure glucocorticoids. Glucocorticoids are continuously secreted into the intestine by bile secretions. Some of these glucocorticoids are further metabolized by the intestinal microbiota into diverse glucocorticoid metabolites (GM) such as 11-oxoaetiocholanolone, this being the most abundant metabolite in ruminants (Mostl et al., Reference Mostl, Maggs, Schrotter, Besenfelder and Palme2002). Fecal GM may be useful biomarkers for long-term adrenocortical activity if the non-adrenal sources of variation can be controlled or the animals are used as their own controls. Fecal GM concentrations can be affected by several processes including conjugation in the liver, intestinal transit time, bacterial metabolism, type of food and procedures of sample collection and preservation (Cook, Reference Cook2012). The measurement of fecal glucocorticoids requires a solvent extraction which can be only performed in well-equipped laboratories. Therefore, it is difficult to apply this method for monitoring large herds at the farm level.
Circulating cortisol can be incorporated in growing hair. Therefore, cortisol concentrations in hair are representative of cumulative cortisol exposure over a period of several weeks or months and can be successfully measured in humans (Stalder and Kirschbaum, Reference Stalder and Kirschbaum2012). Promising examples of hair cortisol measurements in cattle have been obtained in experimental settings. Hair cortisol concentrations were higher in ACTH-treated heifers on weeks 2 and 3 but not on week 6 after challenge and were affected by hair color and age (del Rosario González-de-la-Vara et al., Reference del Rosario González-de-la-Vara, Valdez, Lemus-Ramirez, Vázquez-Chagoyán, Villa-Godoy and Romano2011). However, Tallo-Parra et al. (Reference Tallo-Parra, Carbajal, Monclus, Manteca and Lopez-Bjar2018) did not find any correlation between hair cortisol concentrations and age. Hair location and the sampling method (clipping vs. plucking) affects cortisol concentrations. Indeed, cortisol concentrations are higher in hair sampled from the tail than from the head or shoulders. Similarly, cortisol concentrations are higher in samples collected by clipping (Moya et al., Reference Moya, Schwartzkopf-Genswein and Veira2013) than those collected by plucking. These authors suggested that clipping hair from the tail seems the most suitable method for measuring cortisol concentrations in hair. Braun et al. (Reference Braun, Michel, Baumgartner, Hassig and Binz2017) and Tallo-Parra et al. (Reference Tallo-Parra, Carbajal, Monclus, Manteca and Lopez-Bjar2018) were not able to detect any increase in hair cortisol concentrations following cows’ illness, while others found a significant rise in response to illness in cows (Comin et al., Reference Comin, Peric, Corazzin, Veronesi, Meloni, Zufferli, Cornacchia and Prandi2013; Burnett et al., Reference Burnett, Madureira, Silper, Tahmasbi, Nadalin, Veira and Cerri2015). The lack of standardization of the methodology is a critical issue that can explain the differences observed among studies and limits the exploitation in the field of this methodology for the assessment of stress and welfare. Rapidly growing hair can take-up cortisol from the circulation more efficiently than hair obtained from a previously unshorn area. Additionally, growing hair samples can provide a better retrospective picture of the HPA axis activation related to a mild stressor such as approaching to parturition (Braun et al., Reference Braun, Michel, Baumgartner, Hassig and Binz2017). Hair cortisol concentrations show a circannual variation with the higher concentrations observed in early summer and in cold-temperate regions (Uetake et al., Reference Uetake, Morita, Sakagami, Yamamoto, Hashimura and Tanaka2018).
As described by Ralph and Tilbrook (Reference Ralph and Tilbrook2016), assessing health status and welfare by measuring glucocorticoids is not precise enough as glucocorticoids are also released in non-stressful related situations such as pleasure, excitement and arousal. The stress response is mediated by the activation of several circuits in the brain (Reeder and Kramer, Reference Reeder and Kramer2005; Smith and Vale, Reference Smith and Vale2006; Romero et al. Reference Romero, Platts, Schoech, Wada, Crespi, Martin and Buck2015), which can affect other physiological systems and consequently animal behavior. Therefore, it is important to understand which mechanisms activate the HPA axis, which receptors are involved in the stressor identification, and what effects glucocorticoids exert in both the brain and target tissues (Ralph and Tilbrook, Reference Ralph and Tilbrook2016). Based on these facts, studying the downstream effects of glucocorticoids may contribute to better define the stress response and identify animals under stressful conditions.
Adrenal androgens
Several studies in humans and laboratory animals have evidenced the role of the adrenal androgens dehydroepiandrosterone (DHEA) and dehydroepiandrosterone sulfate (DHEAS) in the stress response. Both DHEA and DHEAS are mainly synthesized in the adrenal glands and gonads, and there is evidence that they are synthesized also in the brain. The function of these hormones is still unclear, but available data suggest their antagonistic effects on glucocorticoids and a role in the stress response (Maninger et al., Reference Maninger, Wolkowitz, Reus, Epel and Mellon2009). In humans, DHEA and DHEAS release can be differently affected by psychosocial stress. For instance, increased circulating DHEA and DHEAS levels were observed in response to an acute psychosocial stressor (Lennartsson et al., Reference Lennartsson, Kushnir, Bergquist and Jonsdottir2012). Conversely, exposure to prolonged stress did not affect circulating DHEA concentrations, while it seemed to negatively affect the capacity to produce DHEAS in response to a subsequent acute stress (Lennartsson et al., Reference Lennartsson, Theorell, Kushnir, Bergquist and Jonsdottir2013).
Only a few published studies have explored DHEA and DHEAS release in cattle in response to stressors. In young beef bulls, 9-hours transportation induced a significant rise in blood cortisol, a decrease in DHEA and an increase in the cortisol/DHEA ratio (Buckham Sporer et al., Reference Buckham Sporer, Weber, Burton, Earley and Crowe2008). Almeida et al. (Reference Almeida, Weber, Burton and Zanella2008) observed a significant decrease in serum DHEA in lame cows, while Fustini et al. (Reference Fustini, Galeati, Gabai, Mammi, Bucci, Baratta, Accorsi and Formigoni2017) observed increased plasma DHEA in cows kept in overcrowded conditions during the last month before parturition. Cows with metritis displayed higher plasma DHEA concentrations only in subjects with low white blood cells count (Gundlach et al., Reference Gundlach, Feldmann, Gundelach, Gil, Siebert, Hoedemaker and Schmicke2017).
The stressors considered in the above-mentioned papers are different in nature and duration and may affect DHEA release in different ways. For example, in cattle ACTH is a poor secretagogue for DHEA and the adrenal contribution to circulating DHEA seems to be quite modest compared to humans. In addition, placenta from cows in late pregnancy seems to be a major source of circulating DHEA (Marinelli et al., Reference Marinelli, Trevisi, Da Dalt, Merlo, Bertoni and Gabai2007).
Therefore, the characterization of DHEA release affected by the nature (inflammatory, physical or social) and persistency of specific stressors over time may provide new information to improve the definition of the stress phenotype in dairy animals.
Sympathetic nervous system
The sympathetic nervous system is another important player during the stress response. When the sympathetic system is activated and catecholamines (in particular adrenaline) are released, a plethora of physiological adaptations are triggered to prepare the animals for the ‘fight or flight response’: heart rate (HR) increases, blood flow is redistributed and metabolic fuels (glucose and fatty acids) are mobilized. However, the sympathetic nervous system has been poorly studied in farmed animals, possibly due to the complexity of measuring catecholamines. Their very short half-life (approximately 1–2 m) means that measuring catecholamines requires particular care in sample collection and storage, and complex pre-analytical steps (Peaston and Weinkove, Reference Peaston and Weinkove2004), all of which makes them unsuitable as biomarkers for monitoring stress in large groups of animals under field conditions.
Sympathetic nervous system activity can be monitored indirectly by assessing physiological variables regulated by catecholamines, such as reticulum-rumen motility, HR and superficial body temperature. The development of sensors that automatically measure physiological parameters, such as rumen and heart functions, can potentially help the early detection of undesirable health or welfare conditions (Caja et al., Reference Caja, Castro-Costa and Knight2016). In healthy cows, for instance, the frequency of reticular contractions is significantly lower in stressed animals (Braun and Rauch, Reference Braun and Rauch2008). Therefore, the continuous measurement of reticulum-rumen motility by ruminal boluses can assess the autonomous nervous system (ANS) activity (Caja et al., Reference Caja, Castro-Costa and Knight2016). Measurement of HR has also been used as a good indicator of the physiological response to acute stress (Kovács et al., Reference Kovács, Jurkovich, Bakony, Szenci, Póti and Tőzsér2014).
Changes in superficial body temperature are influenced by changes in blood flow and metabolism and can be accurately measured using infrared thermography (IRT). However, superficial body temperature may be affected by the anatomical site where IRT is measured (Salles et al., Reference Salles, da Silva, Salles, Roma, El Faro, Bustos Mac Lean, Lins de Oliveira and Martello2016). In cattle, the superficial eye temperature can be considered as an indicator of ANS activity as it reflects changes in the capillary blood flow of the conjunctiva, which is under ANS control (McManus et al., Reference McManus, Tanure, Peripolli, Seixas, Fischer, Gabbi, Menegassi, Stumpf, Kolling, Dias and Costa2016). Interestingly, the exogenous stimulation of the HPA axis alone is not associated with any modifications in eye temperature, while psychological/cognitive stressors (i.e. catheterization procedures) can increase cortisol concentrations and eye temperature (Stewart et al., Reference Stewart, Webster, Verkerk, Schaefer, Colyn and Stafford2007). Adrenaline infusion can decrease eye temperature in calves, associated with increased blood cortisol concentrations, suggesting that changes in eye temperature are mediated by the sympathetic nervous system (Stewart et al., Reference Stewart, Webster, Stafford, Schaefer and Verkerk2010). In addition, IRT may be used for detecting variations in metabolism, inflammatory processes and diseases, and reproduction (McManus et al., Reference McManus, Tanure, Peripolli, Seixas, Fischer, Gabbi, Menegassi, Stumpf, Kolling, Dias and Costa2016). However, limitations to the use of IRT are related to environmental (temperature, humidity, sunlight exposure) and animal's coat conditions (color, dirtiness).
Neurotrophins
Neutrophins are considered putative biomarkers for chronic stress. These molecules are actively investigated in laboratory animals and humans as potential biomarkers of stress and psychiatric disorders (Bath et al., Reference Bath, Schilit and Lee2013). Unfortunately, there is no published information in the literature about the involvement of neurotrophins in stress responses in domestic animals.
One of the most investigated neurotrophins is the brain-derived growth factor (BDNF). This molecule is a dimeric protein that participates in neuronal development and plasticity. It is also found in the brain areas regulating mood, emotion and cognition. Additionally, BDNF is implicated in the development of stress-associated pathologies in humans. As BDNF can cross the blood–brain barrier, high blood concentrations of this molecule are highly correlated with BDNF levels in cerebrospinal fluid (Fernandes et al., Reference Fernandes, Berk, Turck, Steiner and Gonçalves2014). In rodent models, several stressors (e.g. maternal separation, social defeat, social isolation, acute or chronic restraint) can affect BDNF expression throughout animal life (Bath et al., Reference Bath, Schilit and Lee2013). In rats, increased plasma BDNF concentrations are detected in response to both acute and chronic stress (Saruta et al., Reference Saruta, Kondo, Sato, Shiiki, Tsukinoki and Sato2010a, Reference Saruta, Lee, Shirasu, Takahashi, Sato, Sato and Tsukinoki2010b). In the rat, the submandibular gland seems to be the major source of circulating BDNF during acute immobilization stress (Saruta et al., Reference Saruta, Kondo, Sato, Shiiki, Tsukinoki and Sato2010a). Conversely, circulating BDNF concentrations are poorly influenced by BDNF secreted by the submandibular glands during chronic stress, since increased circulating BDNF concentrations were observed in both sialoadenectomized and non-sialoadenectomized rats (Saruta et al., Reference Saruta, Lee, Shirasu, Takahashi, Sato, Sato and Tsukinoki2010b). The involvement of neurotrophins in the stress response should be investigated in dairy animals, as these molecules are promising biomarkers that may be suitable for assessing animal welfare.
Pheromones
Pheromones are molecules released in the environment by most mammals. They can elicit behavioral and physiological responses (reproduction, maternal care, aggression and alarm) in conspecific animals. In terrestrial mammals, pheromones are mainly volatile molecules, although proteins and peptides with pheromonal or odorant binding protein properties have been detected in several species (Brennan and Keverne, Reference Brennan and Keverne2004), including cattle (Japaridze et al., Reference Japaridze, Senda, Nozaki, Yanagida, Suzuki, Ganzorig, Kushi, Kida, Urashima, Bruckmaier and Fukuda2012). Pheromones can be classified as ‘releasers’ if they trigger immediate short-term responses, or ‘primers’ if they trigger medium to long-term changes in behavior or physiology. Another class of pheromones, named ‘signallers’, carry information about the identity of an individual (Swaney and Keverne, Reference Swaney and Keverne2009). Since animals can perceive the environment through ‘deciphering’ these chemosignals, altered pheromone sensing may contribute to poor welfare.
In cows, pheromones are released through body fluids (vaginal fluids, urine, saliva, feces and milk). A pheromonal message can be delivered either by a single compound or a mixture of compounds, whose proportion is crucial for conveying specific reproductive and social signals (Archunan et al., Reference Archunan, Rajanarayanan, Karthikeyan and Mucignat-Caretta2014). This may pose limitations to pheromone detection. In dairy animals, pheromones are used in ‘pheromone therapies’ to induce favorable effects such as estrus stimulation or to appease agitated animals. Pheromones could also be used as biomarkers to detect estrus (Archunan et al., Reference Archunan, Rajanarayanan, Karthikeyan and Mucignat-Caretta2014) and therefore increase fertility rates. In the future, estrous pheromones could be detected by electronic noses (Wiegerinck et al., Reference Wiegerinck, Setkus, Buda, Borg-Karlson, Mozuraitis and de Geed2011; Cave et al., Reference Cave, Wickiser and Mitropoulos2019), providing relevant information to farmers.
A class of pheromones that may be worth investigating in farm animals are those related to the alarm status. Alarm pheromones are secreted by animals in threatening situations to inform other members of the same species of perceived danger. So far, alarm pheromones have been identified in insects and aquatic organisms (i.e. invertebrates and fish), even though behavioral responses ascribed to alarm pheromones have also been observed in ungulates (Hauser et al., Reference Hauser, Wiergowski, Kaliszan, Gos, Kernbach-Wighton, Studniarek, Jankowski and Namiesnik2011). They are highly volatile compounds, characterized by high propensity for spread and penetration and by a short duration of action. (Hauser et al., Reference Hauser, Wiergowski, Kaliszan, Gos, Kernbach-Wighton, Studniarek, Jankowski and Namiesnik2011). Information about alarm pheromones in ungulates is scarce, although novel findings may become available in the near future, as detection of these molecules by electronic noses can inform the farmer or the veterinarian about conditions of fear, uneasiness or pain.
Biomarkers in dairy sheep and goats
Small ruminants represent above 3–5% of the total milk produced worldwide. In Mediterranean countries, most milk produced by small ruminants is transformed in valuable and high-quality dairy products, namely cheeses. Compared to dairy cows, the small ruminant dairy field is poorly studied and the establishment of biomarkers in these species is no exception. One of the reasons is the multiplicity of small ruminant production systems around the world. In contrast, in developed countries, the dairy cattle sector is fundamentally based in one type of animal, the Holstein-Friesian dairy cow, and a production system that favors high yields per animal and, consequently, has to supply adequate environmental, nutritional and management practices to maximize animal performance. Based on the economic importance of small ruminants in the production of highly valuable and high-quality dairy products, increased research effort in strategic topics related to small ruminants are suggested. These strategies may be grouped into different major categories as summarized in Fig. 1.
Fig. 1. Schematic representation of the major areas of research for the establishment of biomarkers in dairy small ruminants. They include the classification and characterization of dairy production systems, particularly highlighting contrasting intensive and extensive production systems; product certification; seasonal weight loss tolerance and finally colostrum management. The picture shows dairy Chios sheep in a milking parlor in an intensive milk production farm near Nicosia (Republic of Cyprus).
One of these strategies is related to the different production systems, particularly extensive vs. intensive production systems. In small ruminant species, extensive production systems are the most common worldwide. These systems are characterized by the use of natural pastures with seasonal variations in the ewe and goat's live weights arising due to variation in the availability of pasture. On the contrary, intensive systems imply the use of high-quality forages and high amounts of concentrate. The extensive vs. intensive management is perhaps the most important aspect regarding dairy small ruminant research and the use of biomarkers. Indeed, the definition of biomarkers to differentiate both systems may have important implications in the future, mainly related to the sustainability of the production system and animal welfare. Examples of this type of research already exist in the literature, such as the studies performed in Israel on high yielding Awassi and Assaf sheep breeds (Pollott and Gootwine, Reference Pollott and Gootwine2004; Gootwine, Reference Gootwine2011). These breeds have been selected to be reared in intensive production systems. In these studies, molecular genetics markers were used to better understand how the Awassi breed was created during domestication. Authors have also uncovered differences in its genetic structure compared to other breeds, particularly highlighting genetic differences between high and low yielding populations.
Extensive small ruminant dairy production systems are very dependent on local ecosystems. Rainfall pattern defines growth and nutritional value of the pasture, with a consequential impact on animal performance. Seasonal Weight Loss (SWL) tolerance is one of the most important aspects in this area. During pasture scarcity due to restricted rainfall, live weight is significantly reduced in grazing animal, affecting animal performance. The selection of breeds or certain genetic groups that are particularly well-adapted to SWL confers significant commercial advantage, so biomarkers of tolerance to SWL are of particular importance as they could provide tools to improve animal production. In dairy goat production, research has compared Majorera (tolerant to SWL) and Palmera (susceptible to SWL) dairy goat breeds from the Canary Islands (Spain). This research involved the quantification of production levels in these two breeds subjected to nutritional stress (Lérias et al., Reference Lérias, Hernández-Castellano, Morales-delaNuez, Araújo, Castro, Argüello, Capote and Almeida2013). This study also allowed the establishment of different molecular markers of tolerance in blood (Lerias et al., Reference Lerias, Pena, Hernández-Castellano, Capote, Castro, Argüello, Araujo, Saco, Bassols and Almeida2015), some of them classic biomarkers such as creatinine, urea, non-esterified fatty acids (NEFAs), cholesterol, IGF-1 and T3. Similarly, this experiment also investigated differences between breeds on the mammary gland secretory tissue from proteomics (Cugno et al., Reference Cugno, Parreira, Ferlizza, Hernández-Castellano, Carneiro, Renaut, Castro, Argüello, Capote, Campos and Almeida2016; Hernández-Castellano et al., Reference Hernández-Castellano, Argüello, Almeida, Castro and Bendixen2015), metabolomics (Palma et al., Reference Palma, Hernández-Castellano, Castro, Argüello, Capote, Matzapetakis and de Almeida2016) and fatty acid (Palma et al., Reference Palma, Alves, Hernández-Castellano, Capote, Castro, Argüello, Matzapetakis, Bessa and de Almeida2017) perspectives. The last two studies also involved analysis and the establishment of biomarkers in milk. Biomarkers of tolerance to SWL include for instance cadherin-13, collagen alpha-1, nidogen-2, clusterin and protein s100-A8 that were found to have higher expression in the SWL-tolerant breed than in the SWL-susceptible breed (Hernández-Castellano et al., Reference Hernández-Castellano, Argüello, Almeida, Castro and Bendixen2015). Furthermore, these results indicated that the susceptible breed had higher expression of apoptosis related proteins.
As mentioned above, extensive production is the most common system used while raising small ruminants. However, the use of intensive productions systems for raising small ruminants has increased over recent decades. Such intensification has led to an increased occurrence of metabolic diseases, including acidosis, foot rot or mastitis, among others. The establishment of biomarkers to prevent or detect these metabolic diseases in sheep and goat will be relevant for farmers. In addition, the use of genetic biomarkers will contribute to the establishment of more resistant or tolerant populations to these metabolic diseases within certain breeds. In order to reach these goals, it is necessary to define reference values for well-known biomarkers of stress in small ruminants. For instance, Miglio et al. (Reference Miglio, Moscati, Scoccia, Maresca, Antognoni and Felici2018) established reference values for different acute phase proteins in Lacaune sheep. In another study performed by Banos et al. (Reference Banos, Bramis, Bush, Clark, McCulloch, Smith, Schulze, Arsenos, Hume and Psifidi2017), the authors established several genetic markers of mastitis resistance in dairy Chios sheep.
As can be observed, further studies need to be performed in small ruminants to determine biomarkers specifically suitable for sheep and goats. The growth and the specific needs of the sector, combined with novel genetics and omics-based tools, may contribute with important information in the near future.
Future prospects
Further studies will need to identify welfare biomarkers and develop accessible, inexpensive and sensitive analytical or non-invasive techniques to measure them. The development of automated monitoring technologies will provide a more precise quantification of animal responses to environmental perturbations (Caja et al., Reference Caja, Castro-Costa and Knight2016; Friggens et al., Reference Friggens, Blanc, Berry and Puillet2017). In addition, biomarkers of robustness can be used to design effective breeding strategies to select more robust and resilient animals (Friggens et al., Reference Friggens, Blanc, Berry and Puillet2017).
Stress cannot be measured with a single biomarker but requires the use of a combination of physiological and behavioral indicators. The hypothesis of non-invasively monitoring HPA axis activation at the farm level has not been tested so far, although the development of technologies for measuring cortisol in milk may not be a limitation (Daems et al., Reference Daems, Lu, Delport, Marien, Orbie, Aernouts, Adriaens, Huybrechts, Saeys, Spasic and Lammertyn2017). Milk cortisol concentrations can change as a result of several genetic and environmental factors (Sgorlon et al., Reference Sgorlon, Fanzago, Guiatti, Gabai, Stradaioli and Stefanon2015; Pošćić et al., Reference Pošćić, Gabai, Stefanon, Da Dalt and Sgorlon2017), which require more thorough analysis and mathematical modeling before this biomarker could be exploited in the field. Finally, it should be emphasized that the information available in this field for small ruminants is limited compared to cows. Similarly, differences between the two bovine dairy species, Bos taurus and Bos indicus and in other bovine dairy breeds such as the Jersey or the Ayrshire could also be explored as most of the results available in the literature are obtained using the Holstein-Friesian breed. Therefore, more research is suggested in both small ruminants and non-mainstream bovine dairy breeds.
Acknowledgements
This review is based upon work from COST Action FA1308 DairyCare (DairyCare, 2016), supported by COST (European Cooperation in Science and Technology). COST is a pan-European intergovernmental framework. Its mission is to enable breakthrough scientific and technological developments leading to new concepts and products and thereby contributing to strengthening Europe's research and innovation capacities.
