2. Neuroendocrine systems

2.1 The hypothalamic-pituitary-adrenocortical axis

2.1.1. General organization

The HPA axis has the classical architecture of the major neuroendocrine systems (Mormède, 1995). The main active hormone of the axis in pigs is cortisol, a cholesterol-derived steroid synthesized in the fascicular zone of the adrenal cortex under the control of the pituitary hormone ACTH (adrenocorticotropic hormone) and released in the general circulation to reach its receptors in tissues. ACTH is synthesized by specialized cells of the anterior pituitary gland (corticotrophs) and its release is triggered by the coordinated action of two neuropeptides, the corticotropin-releasing hormone (CRH) and vasopressin (AVP), that are synthesized in specialized neurons of the paraventricular nucleus of the hypothalamus (PVN) and released in the capillary bed of the median eminence from where they reach the pituitary directly via the hypothalamic-pituitary portal circulation. The PVN receives numerous inputs from other hypothalamic nuclei (these inputs carry metabolic and nycthemeral signals), from the brain stem (in relation with neural inputs from the periphery), from the subfornical organ (that monitors blood plasma composition) and from the limbic system (that generates signals related to the emotional state). This multiplicity of signals converging to the PVN explains the sensitivity of the HPA axis to a wide range of stimuli from both internal and external origin. On the other hand, cortisol exerts a negative feedback on the axis by acting on the pituitary corticotrophs, the PVN and higher levels in the central nervous system. This feedback action of cortisol participates in the return of the HPA axis activity to basal levels after stimulation.


2.1.2. Physiology of the HPA axis

Due to its lipophilic nature, most circulating cortisol (approx. 90%) is bound to proteins, principally albumin and corticosteroid-binding globulin (CBG), a specialized glycoprotein that binds cortisol with high affinity, and regulates its bioavailability (Gayrard et al., 1996). The free fraction can easily cross biologic membranes (that are permeable to lipid-soluble compounds), including the blood-brain barrier and cellular membranes. Cortisol interacts with intracellular receptors – mineralocorticoid receptors (MR) and glucocorticoid receptors (GR) – that, upon activation by their ligand, translocate to the cell nucleus to activate or inhibit gene expression (transcription factors) (Perreau et al., 1999). In the periphery, aldosterone is the primary hormone activating the mineralocorticoid receptor. Aldosterone is released by the adrenal cortex under the influence of the renin-angiotensin system, but also ACTH. In the, tissues involved in water and electrolytes metabolism, and responsive to aldosterone (kidney, salivary glands and colon), the mineralocorticoid receptor is protected from cortisol by a specific enzyme, the 11β-hydroxysteroid dehydrogenase (11βHSD) that metabolizes cortisol into its inactive derivative cortisone (Stewart and Krozowski, 1999).

Detailed information on the metabolic effects of cortisol that are numerous and complex can be found in Sapolsky et al. (2000). Altogether, cortisol has catabolic activity – proteolytic and lipolytic – in peripheral tissues and anabolic activity in liver, including gluconeogenesis and protein synthesis (McMahon et al., 1988). Since cortisol also reduces the entrance of glucose into cells, it increases blood glucose and insulin secretion (the latter is also increased by an action of cortisol on ANS function in the hypothalamus), leading to the storage of energy as fat in the adipose tissue, if the energy is not used in the stress response, by behavioural adjustments for instance. The net effect is an increase of fat depots at the expense of tissue proteins (e.g. from muscle, bone) that may not be too interesting in the long range, at least for the efficiency of pork production (Devenport et al., 1989). Cortisol also increases food intake by an action on the brain so that, as it is frequently the case in homeostatic regulations, the increase of energy availability is a coordinated process via peripheral and central mechanisms (Tempel and Leibowitz, 1994).

The activity of the HPA axis is highly variable. First, the secretion of cortisol is pulsatile, with a periodicity of approx. 90 minutes. Although this feature is well documented in several species like humans (Follenius et al., 1987), bovine (Thun et al., 1981) or sheep (Fulkerson and Tang, 1979), it has not been described in pigs, and we got no experimental evidence that it is indeed the case. The second source of variability is the diurnal cycle that is genetically determined and synchronized by light. In diurnal species (including pigs), the peak activity can be measured in the morning and the trough during the evening and the night (Favre and Moatti, 1977; Ruis et al., 1997; Désautés et al., 1999; Hay et al., 2000). This is indeed an important factor to be taken into consideration in experimental and clinical studies, since the difference between morning and evening levels of cortisol in plasma, saliva, and urine – just to cite the most important biological fluids used in these studies – is very large (figures 1 and 2). The third source of variability comes from food intake that activates the HPA axis. Surprisingly enough, although the meal-induced release of cortisol has been described in humans many years ago (Follenius et al., 1982) and confirmed in several species (Honma et al., 1984; Garcia-Belenguer et al., 1993), it has not been specifically investigated in pigs, but it can be found in experimental data (Ruis et al., 1997; Hay et al., 2000; Geverink et al., 2003; figures 1 and 2).


2.1.3. Response to acute stimulations

The acute response of the HPA axis can be studied by monitoring the release of ACTH and cortisol as well as the effects of cortisol such as the increase of blood glucose levels or the changes in leucocytes counts. Since the assay of plasma cortisol levels is easily feasible in any biology laboratory, it has become the golden standard to evaluate the stress response and plasma cortisol levels are frequently equated to the level of stress, explicitly or implicitly. Several comments must be made to replace blood cortisol levels more accurately among the different measures available to assess pig welfare.

The knowledge of the HPA axis physiology shows that cortisol levels may be increased by several environmental factors, such as meals or physical exercise (Brandenberger et al., 1984; Luger et al., 1988), that are not necessarily considered as stress factors. Salivary cortisol levels are also higher in pigs raised in an enriched (vs. barren) environment (De Jong et al., 1998; de Groot et al., 2000; De Jong et al., 2000). Therefore, any change measured in cortisol levels should be interpreted in the context of its physiological roles.

Cortisol levels are exquisitely sensitive to many environmental factors. The mere exposition of an animal to a novel environment is sufficient to increase blood cortisol to its highest possible levels (Mormède and Dantzer, 1978). Although there is no semantic problem to qualify novel environment exposure as a stress factor – and indeed any novel stimulus is a potential challenge – it does not necessarily compromise welfare, and many would agree that novelty frequently adds salience to life, especially in a generally boring environment as offered by intensive farming. One dimension of the response that cortisol levels do not evaluate correctly is its intensity. Dose-response studies show that the increase of plasma ACTH levels is much more graded with stimulus intensity, as mimicked experimentally by injections of CRH at increasing doses (Oelkers et al., 1988, in humans; Zhang et al., 1990, in pigs; Veissier et al., 1999, in calves). This reflects both the extreme sensitivity of the adrenal cortex to detect and amplify the ACTH signal, and the rapid saturation of the response with increasing ACTH concentrations. Therefore, measuring ACTH and cortisol in the same plasma samples allows a better coverage of the whole range of response intensities that should help in the evaluation of stimulus strength, an important dimension in the assessment of welfare.

The use of blood samples to assess HPA axis activity is not without problem, especially in pigs from which blood is difficult to collect without physical contact with the animal and in most instances, without a strong restraint that can by itself activate the HPA axis, not only in the handled animal, but also in the pen- and room-mates. Alternative biological fluids have therefore been considered. Saliva can be collected easily since pigs are very keen to chew anything including cotton buds out of which saliva can be obtained. Saliva being primarily an ultra-filtrate of blood plasma, salivary cortisol reflects the free fraction of plasma cortisol after metabolism by 11βHSD that is highly active in the salivary glands. Therefore it follows the same pattern as plasma cortisol, but at much lower levels (approx. 10%) that require a very sensitive and specific (due to the presence of cortisone) assay (Cook et al., 1996). On the other hand, the interest of saliva to measure other parameters related to stress and welfare processes is limited. Catecholamine contents do not reflect general activity of the ANS but rather local regulation. Another interesting fluid is urine that is a major excretory pathway of cortisol and its metabolites (including cortisone), and many other biological products. The first interest of using urine is that it can be collected with minimal trouble to the animal, although the investigator is dependent upon its good will to obtain urine when spontaneously voided. A few tricks are available or under evaluation to help Mother Nature, such as dipping the paws of piglets in water to initiate urination. An important characteristic of urine is that it integrates the excretion of biological compounds over the time between two successive emissions. This is eventually a handicap to study short duration processes but is an invaluable help when studying long term changes or individual differences in basal activity of the HPA axis, since it removes short term variations that usually introduce noise into measures made in plasma or saliva. Finally, many other parameters can be measured in urine, including catecholamines and their metabolites, which allows a more comprehensive investigation of the neuroendocrine adaptive processes. We have developed the analytical techniques to measure corticosteroids and catecholamines in pig urine samples (Hay and Mormède, 1997a, b) and these techniques have been validated in a number of studies aiming at the study of genetic variation and/or stress studies (Hay and Mormède, 1998; Hay et al., 2000, 2001; Mormède et al., 2004; Foury et al., 2005a). More work is under way to compare different breeding systems (Pol et al., 2002; Foury et al., 2005b; Colson et al., 2006; Lebret et al., 2006) and it should be extended to epidemiological studies in relationships with other parameters to be discussed later.


2.1.4. Chronic changes

Even if the triggering stimulus is maintained, plasma cortisol levels usually decline after the acute response. This does not mean that the HPA axis is back to basal functioning since several indices show that the activity of the system has changed. Each level of the axis (hypothalamus, anterior pituitary, adrenal cortex) is subjected to opposite influences, trophic via their respective stimulating inputs (such as CRH to the pituitary or ACTH to the adrenal cortex) and inhibitory via corticosteroid hormones. Among the changes induced by chronic activation of the HPA axis and well documented in laboratory animals (Gertz et al., 1987; Mormède et al., 1990), we can cite weight loss (the result of the catabolic effect of cortisol and catecholamines), the shrinkage of thymus (a tissue rich in MR and a sensitive indicator of chronic cortisol action), the proliferation of the corticotrope cells in the anterior pituitary (a trophic effect of CRH), an inhibition of ACTH synthesis (by cortisol) and a reduction of the feedback affect of GR agonists on ACTH release, an increase of the size of the adrenal glands and of the response of the adrenals to ACTH (a trophic effect of ACTH). This resetting of the HPA axis at a different level of activity, that Selye (1956) described as the stage of resistance, is also known as allostasis (McEwen, 1998), and specific protocols are necessary to demonstrate these changes induced by a sustained activation of the HPA axis.

Plasma or salivary cortisol levels are not very informative. Although they can be slightly elevated over basal levels, these changes are difficult to detect without catheterization and/or multiple sampling, as compared to spontaneous variations or the effects of blood sampling itself (Barnett et al., 1988). Furthermore, the effect of stress is not constant over the day, and increased cortisol levels have been seen mostly at night, when they are usually low (Barnett et al., 1981; Janssens et al., 1995b). Measurement of cortisol levels in urine may be more sensitive to detect these small changes since rapid variations are buffered over time, but this hypothesis has still to be documented.

The best studied change is the sensitization of the adrenal cortex response to ACTH by chronic stimulation. Prolonged chain tethering, confinement, high densities increase the release of cortisol induced by a standard dose of ACTH (Rampacek et al., 1984; Meunier-Salaün et al., 1987; Von Borell and Ladewig, 1989; Janssens et al., 1994). Similarly, the cortisol response to various stimuli acting at higher levels of the axis is also increased (CRH, insulin-induced hypoglycemia for instance, figure 3), although the ACTH response is frequently lowered or unchanged, due to the chronically enhanced feedback of cortisol on the pituitary (Janssens et al., 1995a).

Another classical test is named the dexamethasone suppression test (DST) and was developed initially to explore the HPA axis function in depressive patients (Carroll et al., 1981). Dexamethasone is a synthetic compound with glucocorticoid activity and inhibits strongly the secretion of cortisol by acting on pituitary ACTH release. Since the feedback activity of glucocorticosteroids is reduced when the HPA axis is chronically activated, such as in depressed patients, it ‘escapes’ dexamethasone suppression. It was shown to be the case in pigs when space was restricted (Meunier-Salaün et al., 1987).

When the increased activity of the HPA axis is documented, we still have to interpret the results in the context of environmental influences on HPA axis activity and individual differences, especially when no longitudinal data are available from the same animals, such as in epidemiological studies. First, environmental factors may have a strong influence on HPA axis activity, as described earlier. The study of the influence of temperature and humidity is rather impressive in this respect (Marple et al., 1972; figure 4). Second, large individual variations in HPA axis activity have been documented in pigs. In a pioneering series of studies, Hennessy and collaborators have shown that the adrenal response to ACTH is an individual characteristic, reproducible across successive testing, and inheritable (Hennessy et al., 1988; Zhang et al., 1990; Zhang et al., 1992). The intensity of the adrenal response to ACTH was found to be negatively correlated with body weight and growth rate (Hennessy and Jackson, 1987). Several lines of evidence confirm the genetic influences on HPA axis activity. Large differences can be found between porcine breeds (Bergeron et al., 1996; Désautés et al., 1997; Weiler et al., 1998; Mormède et al., 2004) and divergent genetic selection on the basis of the HPA axis response to stress could be made in different species (Edens and Siegel, 1975; Pottinger and Carrick, 2001), but no such data are available yet in pigs. Individual variations can also arise from environmental influences, either during pregnancy and early post-natal life or as a result of previous experience. These early influences have been extensively studied in laboratory animals and offer new approaches to control for emotional reactivity, including neuroendocrine responses (Meaney et al., 1991) but more studies are necessary to validate these findings in pigs (Weaver et al., 2000; Otten et al., 2001; Tuchscherer et al., 2002).

In conclusion, the HPA axis is primarily involved in the regulation of energy fluxes in the body, and is therefore sensitive to various environmental stimuli challenging the energy balance of the body, such as nycthemeral influences, food intake and temperature regulation. Furthermore, plasma cortisol levels are exquisitely sensitive to a whole range of stimuli, which do not need to be intense to maximally activate cortisol release, and altogether known as stressors. Therefore, monitoring plasma cortisol levels may give valuable information about the status of the animal with regard to its internal and external environment but must be interpreted with caution in terms of welfare. For long lasting stimulations, dynamic testing of the HPA axis (stimulation or suppression tests) are necessary to demonstrate the resetting of the system as an adaptation to sustained stimulation, and the data have to be interpreted in the context of environmental influences and individual variation of genetic or acquired origin.


2.2. The autonomic nervous system

2.2.1. General organization

The ANS regulates the function of all the internal organs of the body including the cardiovascular, respiratory and digestive systems, and energy fluxes. In stress responses, the orthosympathetic part (catecholaminergic) of the system is primarily involved, although the parasympathetic (cholinergic) part may also be activated, for instance in the effect of stress on the digestive tract. Noradrenaline is released at the nerve endings of the sympathetic nervous system in the target organs, and both adrenaline and noradrenaline are released in the general circulation from the medullary part of adrenal glands. The differential activity of both catecholamines on their receptors and the differential distribution of the different receptor types and subtypes in tissues shape the specificity of their actions. A comprehensive understanding of the role of the ANS in adaptation / stress processes can be found in the cornerstone paper of Walter B. Cannon published in 1935. As a response to cold for instance, noradrenaline is preferentially released. It induces peripheral vasoconstriction (energy saving mechanism) and energy mobilization, primarily from fat tissues. As a response to hypoglycemia on the other hand, adrenaline is released from the adrenal medulla as an effective mechanism to release glucose from liver and muscle glycogen. In case of acute, non specific challenge, such as in response to a strong emotional stimulus, a general activation occurs with an increase of blood pressure and cardiac pulse rate, respiratory frequency, energy mobilization (with an increase in plasma free fatty acids and glucose).


2.2.2. Assessment of acute sympathetic responses

The release of catecholamines can be monitored directly from blood plasma (Fernández et al., 1994, 1995). There are several limitations to this approach. Plasma levels of catecholamines are extremely sensitive to handling and the response can be detected within seconds, since the transmission of information, via neurons, is extremely fast, as compared to the HPA axis for instance (there is a time lag of approx. three minutes between stimulus application and the increase of cortisol levels in plasma). It is therefore illusory to measure basal catecholamine levels in plasma collected by direct venous puncture, and chronic catheter must be implanted for experimental purposes. On the other hand, the assay techniques to measure catecholamines in blood necessitate specific equipment and know-how and are not easily accessible to any biology laboratory. This approach is therefore limited to physiological investigations. Measuring catecholamines and their metabolites in urine (Hay and Mormède, 1997a) alleviates the problem of sample collection and rapid variations, but has obviously not the same time resolution. However, considering the time scale of the processes involved in welfare studies, this resolution may be largely sufficient.

In many cases, acute sympathetic responses have been studied by monitoring its physiological effects such as heart rate, blood pressure, plasma glucose and free fatty acid levels. The interest of such a parameter like heart rate is that it can be continuously monitored at distance by remote transmission of the physiological signal or stored by simple portable devices. It therefore allows a very sharp second to second analysis of the response to complex stimuli like transportation for instance. However, heart rate is susceptible not only to emotional / stress factors, but also to various influences like locomotion, physical activity or food intake (Villé et al., 1993; Talling et al., 1996; Webster and Jones, 1998). Therefore data will have to be interpreted in this context. On the other hand, metabolic variables are frequently used in studies related to handling of animals before slaughter (mixing, transportation, duration and conditions of lairage, duration of food withdrawal). Plasma glucose and free fatty acids levels reflect the balance between the mobilization of energy stores and the use of energetic metabolites, primarily by muscular activity. Lactic acid levels reflect the intensity of anaerobic metabolism. These metabolic parameters are frequently associated with the measurement of circulating activity of enzymes of intracellular origin, such as transaminases and creatine kinase (CK) that reflect cell suffering. CK has been largely used to detect stress susceptibility in pigs (Guise et al., 1998; Warriss et al., 1998 a,b; Pérez et al., 2002; Foury et al., 2005b; Faucitano and Geverink, in this book).


2.2.3. Metabolic adaptation, stress, welfare and ANS

The data collected recently in the study of the response of piglets to early weaning illustrates several critical aspects of the use of physiological parameters to assess welfare (Hay et al., 2001). The selection of hyperprolific sows results sometimes in an excessive number of piglets that can be saved, when there is no possibility of adoption by other dams, by early weaning at 5 days after ingestion of colostrum. One question is whether this practice is acceptable in terms of animal welfare since these young animals are not prepared to ingest significant levels of solid food before the 3rd-4th week when the piglet is fed by the dam. Indeed, early weaning induces an early and long-lasting reduction in growth rate. Urine was collected from the piglets before weaning and at regular intervals over the two following weeks. Cortisol levels were elevated in urine the day after weaning but were back to control levels at day 5 after weaning. This activation of the HPA axis by weaning has been shown in several studies using plasma cortisol levels, whatever the age at weaning. Since it is only short lasting, we could conclude that, although early weaning induces a typical stress response, stress is short lasting and the animal is adapted after a few days. However, the measurement of catecholamine levels in urine gives a completely different view of adaptation processes. Indeed, early weaning induces an early and profound reduction of the levels of noradrenaline, that do not return to control levels before the end of the second week after weaning, as well as a delayed (measurable on the ninth day after weaning) but sustained reduction of adrenaline levels. How can we interpret these changes? By considering the metabolic roles of catecholamines. As shown before, noradrenaline produces heat by burning lipids. The noradrenergic system is therefore activated by cold to keep core temperature constant, and by excess caloric intake to regulate body weight. In case of food shortage, the switch off of the noradrenergic system is an energy saving mechanism (Stefanovic et al., 1970). Indeed, the reduction of metabolic heat production is compensated for by behavioural adaptation mechanisms since the early weaned piglets spend more time under the heating lamp than control animals (Orgeur et al., 2001). On the other hand, adrenaline is able to mobilize glycogen stores and its sustained release during the first days after weaning is therefore adapted to the reduction of food intake in order to maintain blood glucose levels. What may be conclusions with regard to the assessment of welfare? That cortisol gives a limited and therefore biased view of adaptive processes. Although cortisol excretion levels were back to control levels a few days after early weaning, the profound and long lasting changes in catecholamine excretion levels are indicative of a sustained taxing of adaptive processes, primarily related to the deficit of food intake. The reader is invited to read Mormède and Hay (2003) for a more complete discussion.

The second example illustrating the interest of monitoring the ANS via the measure of catecholamine excretion in urine comes from the study of breed differences in the response to preslaughter stress (Mormède et al., 2004). Since this experiment was designed primarily to study molecular characteristics of muscles in relationship with meat quality, the animals were managed with as little stress as possible. They were not mixed from different pens, the truck was driven smoothly, and the animals were handled gently. Urine was collected on farm (basal conditions), upon the arrival in the slaughterhouse in the evening (after 10 hours of transportation from the breeding farm) and the next morning before slaughter. Data for the Large White and Duroc pigs are shown in figure 5 and compared to the values measured in urine collected from the bladder after slaughter in a group of approx. 300 animals from an F2 intercross between these two breeds, handled and slaughtered in commercial conditions 50 km away from the breeding unit (Foury et al., 2005a). First, these data illustrate the breed differences in neuroendocrine functioning (both basal and in response to stimulation). Second, although urinary cortisol levels are in the same range after transportation in optimal conditions and after slaughter of animals handled more conventionally, large differences are measured for adrenaline and more so for noradrenaline that is hardly increased by smooth handling and transportation of the animals.

These two examples show that taking into consideration the ANS response to challenges widens our understanding of adaptive processes.

The importance of metabolic factors. Although we showed previously their influence in the regulation of HPA axis activity, they are even more important for the functioning of the ANS (see also Hay et al., 2000, about the influence of meal). One noticeable point to stress out is the specificity of metabolic influences on adrenaline and noradrenaline. It shows that the ANS is not activated as a whole single system but differentially according to the specific action of adrenaline and noradrenaline.

The differential sensitivity of cortisol, adrenaline and noradrenaline as measures of stress. The experiment on transportation (Mormède et al., 2004) confirms that the HPA axis is highly sensitive and does not discriminate stimuli with different intensity. The noradrenaline response appears to be the least sensitive, adrenaline being intermediate. Therefore monitoring the excretion of catecholamines gives not only qualitative but also quantitative information about the adaptive response.


2.2.4. Chronic changes

The knowledge of allostatic processes in the ANS is rather limited in pigs. A well-documented effect of chronic stress in laboratory animals is the increase of enzymatic activities in the synthetic pathway of catecholamines, including tyrosine hydroxylase, the first and limiting step, and PNMT (phenylethanolamine N-methyl transferase) that catalyzes the transformation of noradrenaline into adrenaline in the adrenal medulla (Mormède et al., 1990). A few studies only have been done in pigs (Stanton and Mueller, 1976; Roberts et al., 1996). The study of heart rate variability is a possible approach to study sympathetic vs parasympathetic tone (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996; Villé et al., 1993).