1 Department of Pediatrics, University of Michigan, 1150 Medical Center Drive, Ann Arbor, Ml 48109-9550, USA; " Center for Neuroscience, Department of Psychiatry, University of California, Davis, Davis, California, 95616, USA
Glossary: Maternal deprivation: separation of mother and infant during the stress-hyporesponsive period, which needs to last for at least 8h for immediate and persistent effects on the neuroendocrine regulation of the hypothalamic-pituitary-adrenal axis.
Stress-hyporesponsive period: a period of reduced adrenal corticosterone and pituitary adrenocorticotrophic hormone release in response to stress lasting in the rat from postnatal days 4-14.
Based primarily on the pioneering work of the late Hans Selye, the stress response has become somewhat synonymous with the release of hormones from the pituitary and adrenal glands. Thus, in most adult mammals stimuli presumed to be stressful result in a systematic release of adrenocorticotrophic hormone (ACTH) and the subsequent secretion of glucocorticoids from the adrenal. This simplistic view of the pituitary-adrenal axis as first described by Selye has been elaborated on extensively. Thus, the regulation of the so-called stress hormone clearly involves specific peptides synthesized and stored in the brain (i.e., corticotropin-releasing factor ((CRF) and arginine vasopressin (AVP)) and brain-derived neurotransmitters (i.e., noradrenaline). Thus the brain must be included as a critical stress-responsive system. However, the sequence of responses observed consistently in the adult are in many ways very different in the developing organism. Abundant evidence indicate that the rules that govern the activity of the hypothalamic-pituitary-adrenal (HPA) axis in the adult are very different in the neonate. This is best appreciated in rodent. Thus, in this chapter, the ontogeny and regulation of the rodent HPA is discussed. In addition, developmental aspects of the human HPA axis during the first years of life are reviewed.
In 1950, a report appeared that first indicated that the neonatal response to stress deviated markedly from that observed in the adult rodents and thus, created a field of inquiry that has persisted for over four decades. Using depletion of adrenal ascorbic acid as the indicator of the stress response, Jailer reported that the neonate did not show any response to stress (Jailer, 1950). By the early 1960s, Shapiro placed a formal label on this phenomenon and designated it as the "stress nonresponsive period"
*Corresponding author: E-mail: [email protected]
(SNRP) (Shapiro et al., 1962). It is important to note that for the most part the basis for this description was the inability of the rat pup to show significant elevations of corticosterone (CORT) following stress. There was one study that received little attention at the time but did raise important questions concerning the validity of the notion of an SNRP. In that study, in addition to exposing the pup to stress and demonstrating a lack of CORT response, another group was injected with adrenocorticoid hormone (ACTH) (Levine et al., 1967). These pups also failed to elicit a CORT response, which indicated that one of the factors contributing to the SNRP could be a decreased sensitivity of the adrenal to ACTH.
Therefore, it was conceivable that other components of the HPA axis might be responsive to stress. The resolution of this question was dependent on the availability of relatively easy and inexpensive procedures for examining other components of the HPA axis. The methodological break-through, which altered most of the endocrinology and had a major impact on our understanding the ontogeny of the stress response, was the development of radio-immuneassay (RIA) procedures.
The initial impact of the RIA was to change the designation of this developmental period from the SNRP to the "stress-hyporesponsive period" (SHRP). This change was a result of studies that showed a small but significant rise in CORT when measured by RIA (Sapolsky and Meaney, 1986). Although the response of the adrenal was reduced markedly during the SHRP, the adrenal is capable of releasing small amounts of CORT when exposed to certain types of stress.
When investigators began to examine other components of the neonates' HPA axis it became apparent that the SHRP is still a valid concept. However, in order to confront this question, we will examine the development of several components of the HPA axis. These include the adrenal, the pituitary, and the brain.
SHRP, the adrenal, and corticosterone
It is generally agreed that in response to most stressors the neonate fails to elicit adrenocortical response, or does so minimally (Walker et al., 2002). There are several features that characterize the function of the pup's adrenal. The first and most obvious characteristic of the adrenal function during SHRP is that basal levels of CORT are considerably lower than that observed immediately following parturition and that these low basal levels continue to predominate between postnatal days 4-14. Further, numerous investigators have reported that the neonate can elicit a significant increase in plasma CORT levels (Walker et al., 2002). However, invariably the magnitude of the response is small compared to older pups that are outside the SHRP and of course to the adult. Thus, whereas the reported changes in CORT levels following stress in the adult can at times exceed 50pg/dl, rarely does the infant reach levels that exceed lO^g/dl during the SHRP. These levels are reached only under special circumstances, which shall be described later. Thus, the ability of the neonatal adrenal to secrete CORT seems to be impaired markedly. Morphological, biochemical, and molecular biological studies suggest that the development of the adrenal cortex is in part responsible for this phenomenon. Chromaffin cells in the adrenal medulla and maternal factors are also important (see Section "Adrenal Sensitivity").
The mature adrenal cortex in the rodent consists of three concentric steroidogenic zones that are morphologically and functionally distinct: the zona glomerulosa (ZG), the zona intermedia, and the zona fasciculata (ZF)/reticularis (ZR). The ZG, ZF/ZR have unique expression of specific steroidogenic enzymes that defines the specific steroid produced by each zone (Parker et al., 2001). Thus, cytochrome P450 aldosterone synthase (P450aldo) is produced within the glomerulosa to produce the mineralocor-ticoid aldosterone, whereas P450 11 P-hydroxylase (P45011 (3) defines the glucocorticoid producing zona fasciculata/reticularis. In many mammalian species the development of the adrenal cortical layers and steroidogenic enzyme synthesis primarily occur during fetal life (Parker et al., 2001). However, cells expressing P45011P clearly resolve into their cortical layer by the third day after birth (Mitani et al., 1997).
The development of adrenal cortical zones are closely related to the development of the chromaffin cells of the adrenal medulla (Bornstein and Ehrhart-Bornstein, 2000). As shown by Bornstein and co-workers, a variety of regulatory factors produced and released by the adrenal medulla play an important role in modulating adrenocortical function. Isolated adrenocortical cells loose the normal capacity to produce glucocorticoids, whereas culture of adrenocortical cells with chromaffin cells causes marked upregulation of P450 enzymes and the steroidogenic regulatory protein (StAR), which mediates the transport of cholesterol to the inner mitochondrial membrane where steroidogenesis occurs (Bornstein and Ehrhart-Bornstein, 2000). On the 18th day of fetal life, cells containing tyrosine hydroxylase (TH), the initial and rate-limiting enzyme of catecholamine synthesis, and a marker for adrenal medullary cells, are found intermingled with cortical cells expressing
P45011p in the area that is later defined as the ZF/ ZR. However, the adrenal medulla becomes a well-defined morphological region at the end of the first week of life (Pignatelli et al., 1999), at a midpoint in the SHRP. Within this period and until PND 29, the TH enzymatic activity increases (Lau et al., 1987). It is during this time that most of the adrenocortical cellular proliferation activity is observed, but limited to the outer cortex: ZG and ZF. Studies that utilized a specific antibody that recognizes antigens found specifically in these cortical cells of the rat adrenal (IZAgl and Ag2) showed faint ZF immunostain-ing on the first day of postnatal life. A progressive increase in staining was observed until 18-20 days postnatally. Taken together, these data suggest that the limited adrenocortical activity in the infant rat is greatly due to the maturity of the steroidogenic enzymatic pathways of the adrenal during the SHRP (see Fig. 2). In addition, there is evidence that suggests that the autonomic nervous system through the adrenal medulla is also an important contributor to the regulation of adrenocortical development through paracrine activity (Pignatelli et al., 1999).
There is an important caveat in making the assumption that the reduced level of CORT following stress indicates a reduction in biological activity. CORT exists in the circulation in two forms, bound and unbound. The large majority of CORT in the adult is bound to cortisol-binding protein (CBG) and other plasma binding proteins. Only a small fraction exists in the free form, which is considered to be the biologically active form. Following stress, CBG is somewhat decreased, making more of the circulating CORT available as free CORT (Fleshner et al., 1995; Tannenbaum et al., 1997). Another aspect of the SHRP in rodents is the relative absence of CBG during the SHRP (Henning, 1978). Thus, although the absolute values of CORT, which normally include both bound and unbound hormone, are very low in the absence of CBG the actual fraction of CORT that is available in the free form for binding to corticosteroid receptors may actually be higher than is observed in the adult. There are few data on free CORT in the neonate following stress or ACTH
administration. However, in one study at postnatal day 12, the ratio of free versus total corticosterone is much higher in the neonatal rat than in the adult (Henning, 1978). Further, the clearance of CORT from the circulation is significantly slower than the pup (Van Oers et al., 1998). Therefore, as a consequence, CORT is available for a more prolonged period. The biologically active CORT has a more prolonged period of time to exert its effects in the periphery and the brain.
Although there appear to be rate-limiting factors that act developmentally to limit the secretion of CORT in the neonate, evidence indicates that the adrenal is actively suppressed during the SHRP. It has been extensively documented that certain aspects of the rodent maternal behavior play an important role in regulating the neonate HPA axis. In particular, two specific components of the dam's caregiving activities seem to be critical; licking/stroking and feeding. Numerous studies have demonstrated that feeding is in part responsible for the downregulation of the pups' capacity to both secrete and clear CORT from the circulation (Suchecki et al., 1993; Van Oers et al., 1999). Thus, removing the mother from the litter for 24 h results in a significantly higher basal level and a further increase in the secretion of CORT following stress or administration of ACTH. The authors have postulated that one of the consequences of maternal deprivation is to increase the sensitivity of the adrenal to ACTH (Rosenfeld et al, 1992). This has been demonstrated in several ways. (1) Significantly lower doses of ACTH are required to induce the adrenal to secrete CORT. (2) Although the levels of ACTH are equivalent between deprived and nondeprived pups under certain experimental conditions, the levels of CORT are greater in deprived pups. (3) Studies indicate that following mild stress (injection of isotonic saline) there is an increase in c-fos gene expression in the adrenal cortex of the deprived neonate, whereas the nondeprived pup exhibited almost no detectable levels of c-fos mRNA (Okimoto et al, 2002). If maternally deprived pups are provided with food during the period of maternal deprivation, both basal and stress levels of
CORT no longer differ from mother-reared pups. Although a clear mechanism has not been elucidated, it is possible that the gastrointestinal-mediated activity of the autonomic nervous system may regulate this phenomenon.
At this time the physiological consequences of these changes in the exposure to high levels of CORT in the deprived pup are not known. Studies have shown that exposure to high levels of glucocorticoids during development have profound long-term effects on the developing brain (Bohn, 1984). It should be noted, however, that many of these studies used pharmacological doses of adrenal steroids and, in many cases, used hormones that were atypical for the rat (Cortisol, dexamethasone). With the availability of the maternal deprivation model, only recently has it been possible to achieve elevated levels of CORT that are generated endogenously by the pup.
Evidence of reduced clearance of CORT was obtained in a study that examined the ontogeny of negative feedback regulation (Van Oers et al., 1998). The technique employed to study negative feedback in the neonate was to adrenalectomize (ADX) the pup and to measure ACTH following ADX. Pups were tested with and without CORT replacement. When deprived pups were implanted with the identical dose of CORT, their CORT levels were invariably higher than those observed in nondeprived pups. This was interpreted as indicating that clearance was reduced as a consequence of reduced blood flow resulting from 24 h of fasting. Maternal deprivation therefore alters the pattern of exposure to CORT as a function of elevated CORT levels following deprivation that persist in the circulation and presumably in the brain of the developing pup due to reduced rates of clearance.
The concept of an absolute SHRP regarding the response of the pituitary following stress in the neonate is much more problematic. Whether the pituitary can show an increase in ACTH in response to stress is dependent on numerous factors. Among these are the age of the neonate, the type of stress imposed, and, once again, maternal factors (Walker et al., 1991; Walker and Dallman, 1993). The early findings concerning the stress response of the pituitary suggested that there was a deficiency in the neonates' capacity to synthesize ACTH. Thus as a result, the pup should exhibit a reduction in the magnitude of the ACTH stress response. However, sufficient data indicate that the pituitary of the neonate does have the capacity to synthesize and release ACTH that resembles the adult response. What seems to discriminate the neonate from the adult is that for the pup the response of the pituitary is much more stimulus-dependent (Walker et al., 1991). Further, the ability to terminate the stress response is also not fully developed and does not mature until quite late in development (Vazquez and Akil, 1993a). Perhaps the earliest demonstration that the neonate can indeed mount an ACTH response to at least some types of challenges was a study that challenged neonates with an injection of endotoxin throughout the period from birth to weaning (Witek-Janusek, 1998). At all ages the neonate exhibited a significant elevation of ACTH that, beginning day 5, was equivalent to the adult. Of interest is that although there was a robust ACTH response, the CORT response was reduced markedly from day 5 and did not begin to approach adult values until about day 15. The difficulty with this study is that very large doses approaching the lethal sensitivity of young rats to bacterial endotoxin (0.5-30 mg/kg) were used. It has been reported more recently that administration of IL-ip elicited an ACTH response in pups as early as day 6 postnatal (Levine et al., 1994). The peak of the response followed a similar time course to that of the adult, although the magnitude of the response was significantly lower earlier in development. The reduced response in day 6 neonates cannot, however, be interpreted as a reduction in the neonate's capacity to produce ACTH. Three hours following ADX, a robust increase in ACTH occurs as early as day 5, presumably due to the absence of a CORT negative feedback signal (van Oers et al., 1998). This magnitude of the ACTH response is as great as that seen in older neonates at day 18, which are well out of the SHRP.
It has been reported that the neonate does show a significant increase in ACTH in response to a variety of different stimuli in an "adult-like manner." It is noteworthy for each stimulus examined that appears to be an idiosyncratic time course that is dependent on the age and the type of stimulus (Walker et al., 1991). Regardless, it is apparent that the capacity for a pituitary response is present early in development. Under some circumstances the pup can show a greater ACTH response early in development than later. Following treatment with kainic acid the ACTH response of day 12 pups exceeded that of day 6 and day 18 neonates (Kent et al., 1996). The largest ACTH response to /V - m e t h y 1 -1) - a s p a r t a t e (NMDA) was at day 6. However, mother-reared pups failed to respond to milder perturbations. Brief periods of maternal separation, exposure to novelty, injections of isotonic saline, and restraint for 30 min, all failed to elicit an ACTH response in normally reared pups until they escaped from the SHRP (Suchecki et al., 1993).
Why do neonates discriminate between different classes of stimuli, whereas older pups that have escaped from the SHRP, and adults appear to respond in a similar manner regardless of the stress-inducing stimulus? Several hypotheses could account for this phenomenon. First it could simply be a matter of stimulus intensity. Thus, the neonate may be less responsive to stimuli of lower intensities and may therefore require a more intense stressor to activate the neuroendocrine cascade that eventually leads to the release of ACTH. Second, it has been well documented that different stimuli activate distinct neural pathways that lead to the release of CRF and thus ACTH. It is conceivable that the neural pathways that regulate the response to different classes of stimuli mature differently (Sawchenko et al., 2000), and thus if a particular stimulus activates a pathway, which matures early in development then it is likely that a pituitary response will be manifest. Stimuli that threaten survival, such as severe infection (endotoxin exposure) or hypoglycemia, may fit this category. However, if the regulating pathways are developing more slowly, such as stimuli that require some level of associative processing, these stimuli may not be able to be processed neuronally and produce the neuroendocrine cascade required to activate the pituitary.
A third factor appears to contribute to the reduced capacity of the mother-reared pup to respond to milder stress-inducing procedures. The role of mothers caregiving activities on the developing adrenal was discussed earlier. Evidence shows that maternal factors can also actively inhibit the release of ACTH (Suchecki et al., 1993; Van Oers et al., 1998; van Oers et al., 1998). Pups deprived of maternal care for 24 h sowed an increase in ACTH following an injection of saline as early as day 6. The effects of maternal deprivation were even more apparent at days 9 and 12. Although in subsequent experiments the response at day 6 was not reliable, significant increases in ACTH were replicated in 9- and 12-day old neonates. These increases are also observed following 30 min of restraint. In contrast, nondepri-ved pups failed to show an acute release of ACTH following stress. Whereas feeding was required in order to reduce the sensitivity of the adrenal, ano-genital stroking can reverse the increased ACTH secretion following deprivation (Suchecki et al., 1993). Thus, different components of the mother's behavior appear to be involved in regulating different components of the endocrine stress response. In pups that were stroked and fed, both ACTH and CORT are suppressed (see Fig. 1). In pups that received only stroking, ACTH was downregulated but CORT was still elevated. These data would suggest that the dam's behavior was actively inhibiting the neuroendocrine cascade that ultimately results in the peripheral endocrine responses to stress. Thus, the capacity to respond is present early in development but is only observable if the maternal inhibitory factors are not present. It is important to note, however, that although the pup can be induced to show an endocrine response to stress during the SHRP, maternal inhibition is not the only rate-limiting factor. If one examines the body of data on the ACTH responses in pups, what emerges is that even when the infant responds to mild stress during the SHRP, the magnitude of the response is always considerably lower in the SHRP than that of the older pups (day 18) and adults (Dent et al., 2000a).
It has been concluded that during the SHRP the neural pathways regulating the ACTH response to these milder stimuli are not as yet mature or that
Was this article helpful?