SEARCH

SEARCH BY CITATION

Keywords:

  • allopregnanolone;
  • corticotropin-releasing factor;
  • endogenous opioids;
  • noradrenaline;
  • parturition;
  • prenatal stress;
  • vasopressin

Abstract

  1. Top of page
  2. Abstract
  3. Early pregnancy
  4. Late pregnancy
  5. Parturition
  6. Lactation
  7. Conclusion
  8. Acknowledgements
  9. References

Over the past 40 years, it has been recognised that the maternal hypothalamic-pituitary-adrenal (HPA) axis undergoes adaptations through pregnancy and lactation that might contribute to avoidance of adverse effects of stress on the mother and offspring. The extent of the global adaptations in the HPA axis has been revealed and the underlying mechanisms investigated within the last 20 years. Both basal, including the circadian rhythm, and stress-induced adrenocorticotrophic hormone and glucocorticoid secretory patterns are altered. Throughout most of pregnancy, and in lactation, these changes predominantly reflect reduced drive by the corticotropin-releasing factor (CRF) neurones in the parvocellular paraventricular nucleus (pPVN). An accompanying profound attenuation of HPA axis responses to a wide variety of psychological and physical stressors emerges after mid-pregnancy and persists until the end of lactation. Central to this suppression of stress responsiveness is reduced activation of the pPVN CRF neurones. This is consequent on the reduced effectiveness of the stimulation of brainstem afferents to these CRF neurones (for physical stressors) and of altered processing by limbic structures (for emotional stressors). The mechanism of reduced CRF neurone responses to physical stressors in pregnancy is the suppression of noradrenaline release in the PVN by an up-regulated endogenous opioid mechanism, which is induced by neuroactive steroid produced from progesterone. By contrast, in lactation suckling the young provides a neural stimulus that dampens the HPA axis circadian rhythm and reduces stress responses. Reduced noradrenergic input activity is involved in reduced stress responses in lactation, although central prolactin action also appears important. Such adaptations limit the adverse effects of excess glucocorticoid exposure on the foetus(es) and facilitate appropriate metabolic and immune responses.

Pregnancy and lactation have been shown over the last 40 years to be physiological states in which hypothalamic-pituitary-adrenal (HPA) axis responses to stressors are markedly attenuated (1, 2). These phenomena provide an unequalled opportunity to understand natural mechanisms that reduce stress responses, and the prospect of new therapies for stress-related disorders.

The (HPA) axis comprises the corticotropin-releasing factor (CRF) neurones in the parvocellular paraventricular nucleus (pPVN), which also variably produce vasopressin and project to the external zone of the median eminence. These neurosecretory neurones release their peptides into the primary capillary plexus of the hypothalamic-hypophysial portal system to act respectively on the CRF1 and V1b receptors on the corticotrophs in the anterior pituitary gland (3, 4). The consequent stimulation of secretion of corticotropin [adrenocorticotrophic hormone (ACTH); a product of pro-opiomelanocortin (POMC)], leads to increased synthesis and secretion of glucocorticoid (cortisol in humans and other species; corticosterone in rodents) by the adrenal cortex. Glucocorticoids have powerful actions on metabolism and immune mechanisms (5, 6). The HPA axis is regulated by tonic glucocorticoid feedback (7) [involving mineralocorticoid receptors (MR) in the hippocampus, and glucocorticoid receptors (GR) in the brain and corticotrophs], by metabolic signals (8) (including from adipose tissue), and the circadian clock in the suprachiasmatic nuclei (9).

Stressors, stimuli that signal actual or threatened harm or disturbance of homeostasis from external or internal causes, over-ride these control mechanisms to excite the pPVN CRF neurones. Multiple brain pathways process different types of stressors, most simply categorised as emotional (or psychological) or physical. The respective predominantly rostral (glutamatergic) or caudal (noradrenergic) pathways converge onto the pPVN CRF/vasopressin neurones, interacting with GABA inputs (10). Rapid glucocorticoid feedback contributes to terminating a stress response (7, 11, 12). This is important because chronic exposure to high levels of glucocorticoids is damaging.

Intuitively, it could be expected that optimal outcomes of pregnancy and lactation would be supported by minimising the exposure of the mother to stressors. Failing such protection, a supplementary strategy is reduced maternal stress responses, in particular attenuation of HPA axis responses, while advantageous behavioural responses are maintained. Within the past 10–20 years, detailed studies of the central mechanisms regulating the HPA axis have shown how its activity is altered and how stress responses are reduced in pregnancy (13) and lactation (14). There are three aspects to this phenomenon: (i) the altered basal activity of the HPA axis in pregnancy or lactation upon which stressors act; (ii) the level in the stress processing pathways at which the impact of a stressor is reduced; and (iii) the conditions in pregnancy or lactation that lead to the attenuated HPA responses.

Early pregnancy

  1. Top of page
  2. Abstract
  3. Early pregnancy
  4. Late pregnancy
  5. Parturition
  6. Lactation
  7. Conclusion
  8. Acknowledgements
  9. References

Maternal HPA axis in early pregnancy: basal activity

In early pregnant rats, during the early light phase when HPA axis activity is at the circadian nadir, the secretion of ACTH and corticosterone is similar to that observed in virgins (15–19). Similarly, the expression of CRF and vasopressin genes in the pPVN, for GR and MR mRNAs in the PVN and hippocampus (20), and for POMC mRNA and ACTH content in the anterior pituitary (21) are unaltered in early pregnancy.

However, compared with pre-pregnancy, there is a striking alteration in the circadian pattern of basal HPA axis activity, even before implantation. In rats, circulating peak levels of ACTH and corticosterone in the late light phase are reduced by day 2 compared with dioestrus (22). As pregnancy proceeds, the circadian pattern remains suppressed, and the mesor level of corticosterone declines further until mid-pregnancy (day 10), when it begins to increase again (2, 22, 23). The suppression of the diurnal increase in ACTH secretion is more striking and is sustained throughout pregnancy (22). In women, early pregnancy salivary cortisol levels are also typically lower than in late pregnancy, and differences between nadir and peak (early light phase in humans) become greater as cortisol levels increase through pregnancy (24, 25). Lower glucocorticoid secretion in early pregnancy may facilitate implantation because miscarriage is associated with elevated salivary cortisol levels in women at 1–3 weeks post-conception compared with women with ongoing gestation (26).

Maternal HPA axis in early pregnancy: responses to stress

HPA axis responses to stressors in rodents have mostly been analysed during the morning phase of the circadian cycle, and these responses in early pregnancy are not different from those in virgin animals. Rodents exposed to stress at the time of implantation exhibit increased ACTH secretion, even though the risk of pregnancy failure is high due to inhibition of progesterone secretion and action (27–29). Furthermore, in the first half of pregnancy in rodents, HPA axis secretory responses to acute physical (water immersion) and emotional (restraint) stressors remain similar to those of nonpregnant females, although CRF release may already be reduced (13, 18). In rats, exposure to chronic social stress during the dark phase still elicits a corticosterone response in the first half of gestation (19). By contrast, the experience of ‘chronic stressful life events’ during early pregnancy in women blunts salivary cortisol levels in morning (peak) samples, without affecting evening (nadir) levels (24).

In summary, there are coordinated adaptations even in early- to mid-pregnancy that moderate the circadian activity of the HPA axis, but with little effect on its responses to stress. This may mean that activation of the maternal HPA axis in early pregnancy by stress might jeopardise the pregnancy.

Late pregnancy

  1. Top of page
  2. Abstract
  3. Early pregnancy
  4. Late pregnancy
  5. Parturition
  6. Lactation
  7. Conclusion
  8. Acknowledgements
  9. References

From mid-pregnancy onwards, there are further, more substantial, adaptations that lead to altered control of the HPA axis and hyporesponsiveness to stressors. The mechanisms underlying reduced HPA axis responses to stress in late pregnancy have been investigated extensively in the rat.

Maternal HPA axis in late pregnancy: basal activity

In women, assessment of maternal HPA axis function is problematic because the placenta also produces HPA axis peptides (30, 31); however, this is not the case in rodents. In the rat, late pregnancy is associated with reduced basal activity of the HPA axis (Fig. 1). In the pPVN, expression of CRF and vasopressin mRNAs is decreased (20), concomitant with reduced CRF content in the median eminence (21). In the anterior pituitary, mRNA expression for the ACTH precursor, POMC (21), and CRF type 1 (13) and V1b (21) receptor binding is reduced. Enhanced glucocorticoid negative feedback may underpin attenuated basal HPA activity at the end of pregnancy: GR mRNA expression in the hippocampal dentate gyrus is up-regulated in late pregnancy (20). Furthermore, activity of 11β-hydroxysteroid dehydrogenase type 1 (which acts as a reductase, reactivating glucocorticoids from their inert metabolites) is increased in both the PVN and anterior pituitary (20), presumably increasing local glucocorticoid production and enhancing negative feedback mechanisms.

image

Figure 1.  Changes in basal hypothalamic-pituitary-adrenal (HPA) axis activity in late pregnancy and lactation in the rat. Late pregnancy: the circadian rhythm in adrenocorticotrophic hormone (ACTH) and corticosterone (CORT) secretion is suppressed and increased circulating corticosterone binding globulin (CBG) reduces the amount of free corticosterone. Corticotropin-releasing factor (CRF) and arginine vasopressin (AVP) mRNA expression in the parvocellular division of the paraventricular nucleus (pPVN) and CRF content in the median eminence are reduced and in the anterior pituitary, levels of pro-opiomelanocortin (POMC) mRNA are also decreased. Activity of the enzyme 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD-1) is increased in the PVN and anterior pituitary, which is expected to increase levels of local corticosterone to negatively feedback onto the HPA axis. In vitro, anterior pituitary corticotrophs display increased sensitivity to CRF and vasopressin, despite reduced CRF and V1b receptor expression; hence, in vivo reduced CRF/AVP release underlies reduced ACTH and corticosterone secretion. Lactation: suckling increases basal activity of the HPA axis in lactation, indicated by increased circulating trough and peak ACTH and increased trough but decreased peak corticosterone levels. Anterior pituitary corticotrophs are less sensitive to CRF, but more sensitive to AVP in lactation, despite unaltered levels of CRF and V1b receptor binding compared with nonpregnant rats. Unlike in late pregnancy, POMC mRNA expression levels in the anterior pituitary are not different from those observed in the non-pregnant state. CRF mRNA expression in the pPVN is markedly reduced in lactation, with a compensatory increase in AVP mRNA expression.

Download figure to PowerPoint

The plasma ACTH-corticosterone relationship is altered in late pregnancy in the rat. Despite suppression of the normal circadian rhythm in ACTH secretion in the second half of pregnancy, the circadian variation in corticosterone secretion is maintained, with a relatively small reduction in the peak (22). This may reflect increased sensitivity of the adrenal gland to ACTH by the action of increased circulating estrogen (32). Daily integrated circulating corticosterone levels increase progressively from mid-gestation until term in the rat (22). Metabolic clearance of corticosterone is sustained at pre-pregnancy levels throughout gestation (17), although the total amount of free (unbound) corticosterone declines in pregnancy, consistent with an increase in circulating corticosterone binding globulin (CBG) (33). In mice, total plasma corticosterone levels increase dramatically in the second half of pregnancy (34); however, as is the case in rats, physiologically available corticosterone is reduced due to greatly elevated levels of circulating CBG in late pregnancy (35).

Maternal HPA axis in late pregnancy: responses to stress

The responsiveness of the HPA axis to stressors is progressively attenuated during late pregnancy in rats (13, 18, 36), mice (35) and women (37). In the rat, HPA hyporesponsiveness manifests between days 10–15 of gestation and is maintained until term, through parturition and into lactation (discussed below). This adaptation acts to buffer the impact of stress and should reduce foetal exposure to excess glucocorticoid, so minimising the risk of detrimental glucocorticoid programming (38).

Suppressed HPA axis responses to a wide range of stressors in late pregnancy are indicated by reduced ACTH (Table 1) (13, 36, 39–46) and corticosterone secretion. The stressors tested include exposure to the elevated plus maze (13), forced swimming (13), immobilisation (47) and immune challenge (36) in rats, and novel environment and forced swimming in mice (35). In late pregnant women, exogenously administered CRF fails to increase ACTH or corticosterone secretion (48) and suppressed salivary cortisol responses are observed following exposure to the cold pressor test (49).

Table 1.   Responsivity of the Hypothalamic-Pituitary-Adrenal Axis to Different Stressors in Late Pregnancy and Mid-Lactation.
Stress paradigmLate pregnancyMid-lactation
  1. Summary of peak ACTH responses, expressed as a percentage of the response in nonpregnant female rats following exposure to a range of different stress stimuli during late pregnancy (day 21) and mid-lactation (days 9–12). In each case where data are presented, the response was significantly lower in the late pregnant and lactating rats compared with nonpregnant rats. NA, not available. CCK, cholecystokinin.

Psychological
 Elevated plus maze (5 min)32% (13)25% (13)
 NoiseNA0% (39)
 Restraint (30 min)34% (40)NA
Combined
 Forced swimming (90 s)42% (13)27% (13)
 Social defeat (30 min)27% (41)44% (42)
Physical
 Systemic CCK (20 μg/kg i.v.)37% (43)NA
 Systemic endotoxin (lipopolysaccharide)20% (36) (1 μg/kg i.v.)56% (44) (200 μg/rat i.p.)
 Interleukin-1β (500 ng/kg i.v.) 4% (36)NA
 Neuropeptide Y (5 μg/rat i.c.v.)7% (45)NA
 Orexin-A (500 ng/rat i.c.v.)0% (46)NA

Mechanisms of reduced HPA axis responses to stress in late pregnancy

Adaptations at the anterior pituitary, hypothalamus and other brain regions contribute to reduced HPA axis responses to stress in late pregnancy (Fig. 2). CRF (13) and vasopressin (21) are individually less effective in stimulating ACTH secretion in late pregnant rats but, given together, CRF and vasopressin stimulate ACTH secretion similarly in virgin and late pregnant rats (21). This normal response to combined exogenous CRF and vasopressin in vivo is explained by in vitro studies in which CRF-evoked ACTH secretion by corticotrophs from late pregnant rats shows greater augmentation by vasopressin. Together, these in vivo and in vitro studies suggest that a lack of coproduced vasopressin underpins stress hyporesponsiveness in late pregnancy (21). Furthermore, in both rats (47) and mice (35), pPVN CRF and vasopressin neurones display reduced stress-induced activation compared with nonpregnant animals, as reflected by reduced CRF, vasopressin and immediate early gene mRNA expression.

image

Figure 2.  Mechanisms of reduced stress-induced hypothalamic-pituitary-adrenal (HPA) axis activity in late pregnancy and lactation in the rat. In both late pregnancy and lactation, stressors stimulate corticotrophin-releasing factor (CRF) and arginine vasopressin (AVP) neurones less strongly than in virgin rats, reflected by reduced stimulation of CRF and AVP biosynthesis in the parvocellular division of the paraventricular nucleus (pPVN). CRF and AVP release at the median eminence is also attenuated in late pregnant and lactating rats, compared with nonpregnant rats, resulting in reduced stimulation of gene transcription for the adrenocorticotrophic hormone (ACTH) precursor, pro-opiomelanocortin (POMC) in the anterior pituitary, reduced ACTH release from the anterior pituitary and hence reduced corticosterone secretion from the adrenal cortex. In late pregnancy, there is reduced afferent drive to the pPVN CRF/AVP neurones from limbic forebrain regions that are involved in processing psychological/emotional stressors. Endogenous opioids presynaptically inhibit noradrenergic brainstem inputs to the PVN that are involved in signalling to the CRF neurones following exposure to physical stressors (e.g. immune challenge, forced swimming and systemic cholecystokinin) and there is reduced α1 receptor subunit mRNA expression. Induction of inhibitory opioid tone in pregnancy appears to result from increased levels of allopregnanalone (AP), the neuroactive steroid progesterone metabolite. In lactation, there is also reduced afferent drive to the CRF/AVP neurones in the pPVN from both limbic and brainstem regions. In particular, there is reduced α1 receptor-mediated sensitivity to noradrenergic drive in lactation. The suckling stimulus from the pups plays a pivotal role in maintaining HPA axis hyporesponsivess in lactation, although the precise mechanism is not clear, it is likely that prolactin is involved.

Download figure to PowerPoint

At least in rats, this is not due to enhanced rapid glucocorticoid feedback inhibition (20). Hence, corticosterone is less effective in suppressing ACTH secretion following pharmacological adrenalectomy in late pregnant rats, indicating a reduced sensitivity to rapid glucocorticoid feedback (20). Rather, it is a result of reduced afferent drive from limbic forebrain (47) and brainstem (36) stressor processing networks that provide excitatory input to the PVN. For example, brainstem noradrenergic neurones in the nucleus tractus solitarii (NTS; A2 cell group) and ventrolateral medulla (A1 cell group) are activated by physical stressors and mediate HPA axis responses via a direct excitatory input to the CRF neurones in the pPVN; however, in late pregnancy, in contrast to virgin rats, forced swimming (43) and cytokine challenge [systemic interleukin (IL)-1β] (36) fail to evoke noradrenaline release in the PVN and hence do not activate the HPA axis. This is accompanied by a loss of effectiveness of an α1 antagonist on ACTH secretion under basal or stress conditions, and decreased α1 mRNA expression in the PVN (43).

Endogenous inhibitory opioid mechanisms

This HPA hyporesponsiveness appears to be a consequence of up-regulated central inhibitory opioid mechanisms because systemic administration of the opioid receptor antagonist, naloxone, restores ACTH secretory responses to cytokine challenge (36), forced swimming (50) and systemic cholecystokinin (43) in late pregnant rats. Opioids evidently act at the level of the PVN, presumably presynaptically on noradrenergic terminals because naloxone infused directly into the PVN reinstates a noradrenaline response, at least following IL-1β administration (36). The origin of the opioids that restrain the responses of CRF neurones in pregnancy is likely to be the NTS neurones: mRNA expression for pro-enkephalin-A and μ-opioid receptors are up-regulated in the NTS in late pregnancy (36). If these NTS neurones are the same population activated by physical stressors, such as IL-1β, this would afford a mechanism through which noradrenergic drive from the brainstem to the PVN could be selectively inhibited in late pregnancy. By contrast, enhanced endogenous opioid inhibition does not underlie the reduced HPA axis responses to stress during late pregnancy in the mouse (35), and the mechanisms involved in HPA axis hyporesponsiveness in this species are unclear.

Inducers of HPA axis hyporesponsiveness

Several studies have sought candidate inducers of pregnancy-related adaptations in HPA axis activity in rat models. For example, increased levels of central oxytocin (51) and circulating oestrogen and progesterone (52) do not appear to play a role in the induction or maintenance of HPA axis hyporesponsiveness in late pregnancy. However, the progesterone neurosteroid metabolite, allopregnanolone, is involved. Circulating and central levels of allopregnanolone are elevated in pregnancy as a result of increased progesterone secretion (53). Blocking allopregnanolone production with finasteride (an inhibitor of 5α-reductase, a key enzyme in the conversion of progesterone into allopregnanolone) restores HPA axis responses to systemically administered IL-1β in late pregnant rats, whereas mimicking pregnancy in virgin rats with allopregnanolone treatment attenuates the response (54). Allopregnanolone appears to depend upon the actions of endogenous opioids to exert its suppressive effects on HPA axis activity because allopregnanolone induces opioid tone over ACTH responses to immune challenge in virgin rats. Moreover, allopregnanolone treatment up-regulates opioid expression in the brainstem of virgin rats by a similar proportion to that seen in late pregnancy (54). The mechanism by which allopregnanolone regulates opioid expression has yet to be investigated, but may involve an interaction with GABAA receptors (55, 56). Furthermore, a role for allopregnanolone in HPA axis hyporesponsiveness during late pregnancy in other species remains to be elucidated.

In summary, HPA axis hyporesponsiveness in late pregnancy is driven by allopregnanolone acting in part at least through central opioid mechanisms. The strong attenuation of HPA axis responses to stressors in pregnancy is likely to provide a first-line defence to protect the foetuses from any adverse programming by glucocorticoids, but may also be involved in immune system adaptations and in metabolic rebalancing.

Parturition

  1. Top of page
  2. Abstract
  3. Early pregnancy
  4. Late pregnancy
  5. Parturition
  6. Lactation
  7. Conclusion
  8. Acknowledgements
  9. References

Maternal HPA axis at term

In several species (including rodents, sheep, pigs, and humans), maternal ACTH and/or cortisol levels are increased on the day of parturition (13, 32, 35, 57–62). Despite this, the HPA axis is inhibited during the birth process in rodents (35, 63). The usual circadian pattern of ACTH and corticosterone secretion is suppressed on the day of expected parturition in rats and mice, and the hormone concentrations decline to their lowest levels during labour and delivery, independent of time of day (35, 63). In the pig, neither vaginal cervical dilatation nor births appear to induce cortisol secretion (61, 64).

The lack of maternal HPA axis activation in parturition does not exclude the process from being a stressor. Dystocia can stimulate HPA axis activity (63), while the lack of a HPA axis response in normal parturition is evidently due to continued or exaggerated endogenous opioid inhibition because naloxone administration greatly enhances HPA axis hormone secretion during rat parturition (63). This may not generalise to other species because opioids do not appear to inhibit the HPA axis in parturient pigs (62).

In women, circulating HPA axis hormone levels are increased during the dilatation and expulsion stages (32, 65–68). However, elevated circulating CRF, ACTH and glucocorticoid hormone levels in women at term are not necessarily indicative of maternal HPA axis activation. The human placenta and endometrium synthesise and secrete CRF which drives foetal ACTH and cortisol secretion increasingly towards term, whereas the placenta is also a source of ACTH (69). Placental CRF has complex effects including a role in the onset of labour (70), providing a placental clock signal for the timing of birth onset (71, 72). Similarly, in the sheep, interaction between the placenta and foetal HPA axis has a crucial role in initiating parturition (73).

Maternal HPA axis at parturition: exposure to stress

Although stressors disrupt parturition, through sympathetic activation (35), and/or inhibition of oxytocin secretion (74–76), HPA axis activity remains suppressed. Thus, in rodents and pigs, exposure to stressors such as airpuff startle or space restriction does not increase ACTH or corticosterone/cortisol secretion (35, 61, 77); in pigs, this lack of response to imposed stress at the time of parturition might be mediated by opioid inhibition (61).

Central oxytocin is an inhibitor of HPA axis activity in virgin rats (51, 77), but not in pregnancy or lactation (51) (see below), and although oxytocin release within the brain increases at birth (78), increased oxytocin inhibition does not explain attenuation of the HPA axis perinatally in rats (77).

In summary, it is evident that, in some species (rat and pig), hypoactivity of the HPA axis persists through parturition and HPA responses to extraneous stressors are suppressed in parturition. In rats and pigs, endogenous opioids can inhibit the HPA axis through parturition. In women, circulating levels of HPA axis hormones increase during parturition; however, this might reflect placental-foetal mechanisms rather than maternal hormone secretion.

Lactation

  1. Top of page
  2. Abstract
  3. Early pregnancy
  4. Late pregnancy
  5. Parturition
  6. Lactation
  7. Conclusion
  8. Acknowledgements
  9. References

Lactation is also a state of HPA axis hyporesponsiveness but, in contrast with pregnancy, this is maintained by the suckling of the young. Suckling is a powerful neuroendocrine stimulus that also excites prolactin and oxytocin secretion, essential for milk production, and inhibits gonadotropin secretion. Furthermore, suckling rapidly increases ACTH and corticosterone secretion (79). The functional significance of these HPA responses to suckling are likely related to important actions of glucocorticoid on the cells that produce milk (80, 81) and actions on metabolism, favouring energy flow for milk production (82).

By contrast, other stimuli considered as stressors [(i) physical: exposure to ether, footshock, forced swimming, hypertonic saline injection, lipopolysaccharide (LPS) injection; (ii) emotional: elevated plus maze, noise stress, social stress], stimulate the HPA axis much less effectively in lactation in rats (1, 13, 14, 39, 42, 79, 83–90), but this is not consistent in studies of mice (35, 91). In sheep, cortisol secretion in response to isolation, restraint stress or exposure to a dog barking is also reduced in lactation (92, 93). In lactating women, reduced HPA axis responses have been found to a physical exercise or cold stressor, but not to a social stressor (Trier Social Stress Test) (49, 94, 95), unless breast-feeding occurs shortly beforehand (96, 97). Furthermore, multiparous but not primiparous breast-feeding women show reduced cortisol responses to emotional stressors (98).

Maternal HPA axis activity in lactation: basal activity

The changes in HPA axis stress responsiveness are superimposed on a state of increased basal HPA axis activity in lactation (99). Circulating ACTH concentration is increased by post-natal day 4, with a marked increase at the circadian nadir, persisting until at least day 10, with increased peak ACTH levels (22, 79). Circulating corticosterone shows a reduced circadian peak on day 4, which is maintained until at least day 10, but with elevated trough levels (22). One study found higher corticosterone levels than in virgin rats, but failed to find increased ACTH levels (100). Basal salivary cortisol level is increased in breast-feeding women 8 weeks post partum (49).

Anterior pituitary corticotrophs and pPVN CRF/vasopressin neurones

The maximal ACTH secretory response to CRF is decreased and the response to vasopressin is increased in lactation (13, 101) (Fig. 1). Changes in corticotroph sensitivity to vasopressin and CRF are not due to altered levels of CRF and V1b receptor binding in the anterior pituitary (101). Furthermore, corticotroph POMC mRNA expression (100, 102) and adrenal sensitivity to ACTH are not altered in lactation (84). It is likely that reduced peak HPA axis activity in lactation is a result of reduced CRF production and release by pPVN neurones. Hence, apart from one study that reported increased CRF mRNA expression in the pPVN in lactation (103), CRF mRNA expression in the pPVN has been found to be markedly reduced in lactation (39, 44, 88, 100, 104).

The reduced ACTH response to CRF is unlikely to be due to lack of pPVN vasopressin production as in late pregnancy (21) because pPVN vasopressin mRNA expression is greater than in virgins (88, 100, 104, 105) and there is greater colocalisation of vasopressin with CRF (104). Thus, the drive for increased ACTH secretion in lactation might be provided by vasopressin.

Reduced pPVN CRF mRNA expression in lactation is consistent with greater negative feedback signal from the increased corticosterone levels, perhaps accounting for the relatively low ACTH levels reported in one study (100). However, sensitivity of ACTH secretion to negative feedback by corticosterone is not increased in lactation (79) and the suppressed corticosterone response to stress indicates that enhanced rapid feedback cannot explain the attenuated ACTH response (89).

Importance of suckling

The increased basal HPA activity in lactation is not affected by inducing an increase in calorie intake and, thus, is not a response to metabolic drain (79). Dependence of increased basal HPA activity on the continued presence of the litter is demonstrated by finding that blood ACTH and corticosterone levels begin to decrease within 3.5 h of pup removal, and are at virgin levels within 14–24 h (79, 100). Conversely, return of the pups after separation from the mother for 4 h, which rapidly initiates a nursing bout, stimulates maternal ACTH secretion (79), and more strongly if the pups have been exposed to a noxious stimulus when separated (86). Separation of pups for 48 h partially restores the attenuated ACTH response to CRF (101), which may partly explain the stimulation of the HPA axis when the mother and pups are reunited. It has been argued that maternal separation experiments may not represent the effects of uninterrupted undisturbed nursing, and that the important effect of ongoing nursing is to attenuate circadian input to the HPA axis (90).

In women, morning cortisol secretion is lower in breast-feeding multiparous but not primiparous women (106), and a nursing bout reduces ACTH and cortisol secretion (107, 108), although the awakening cortisol response is not acutely reduced by breast-feeding (109).

Evidently, the increased HPA basal activity during the circadian nadir is driven by suckling, and does not account for the reduced stress responsiveness seen in lactation. The neural pathway through which suckling activates the pPVN CRF/vasopressin neurones has not been defined, although a histaminergic link from the tuberomammillary nucleus, involving H1 receptors is important (110).

Maternal HPA axis activity in lactation: responses to stress

The strong ACTH and corticosterone secretory responses to resumption of suckling after a short (4 h) period of separation (79) indicate that reduced responses to classical stressors are a consequence of altered central stressor processing rather than changes in corticotroph or adrenal cortex responsiveness to stimulation. Moreover, feedback regulation of ACTH responses to stress by glucocorticoid is unaltered in lactation (79).

CRF neurones in the pPVN

That reduced responsiveness of the pPVN CRF/vasopressin neurones or their inputs to stressors in lactation accounts for the reduced HPA axis responses to stressors is evident from studies using markers of neuronal activation (Fos immunocytochemistry, or c-fos or NGFI-B mRNA, CRF hnRNA or mRNA, vasopressin mRNA or proenkephalin-A mRNA in situ hybridisation studies) (14, 44, 47, 90, 111). Such studies have shown that, in lactating rats, pPVN CRF neurones are stimulated less by stressors (Fig. 2). In response to restraint stress, lactating rats show less c-fos mRNA expression in the PVN than virgin rats (47). Further studies have shown attenuated increases in the pPVN in CRF receptor-1 and vasopressin mRNA levels that rapidly follow restraint stress in lactating rats compared with virgins (103). The increased production of vasopressin by pPVN CRF neurones (104) indicates that vasopressin, although it is a weaker secretagogue than CRF (112), may have a more important role in stimulating ACTH secretion in lactation.

Stress processing regions

Some stressor-processing brain regions (i.e. rostral brain regions involved in emotional stressor processing) (113) are also activated less by stressors in lactating rats. Of several regions involved in stressor processing, reduced c-fos mRNA expression in lactating rats was found only in the medial amygdala, ventral lateral septum and cingulate cortex (47).

In lactation, basal CRF mRNA expression in the bed nucleus of the stria terminalis is increased selectively in the dorsolateral region, and reduced in the central nucleus of amygdala (114). Functionally, central administration of CRF in virgin rats activates immediate early gene expression in several brain areas involved in emotional stress responses, including PVN neurones but, in lactating rats, CRF fails to activate several limbic areas as well as PVN neurones (111).

Mechanisms of reduced HPA axis responses to stress in lactation

Noradrenergic input

Further evidence for reduced stimulation by inputs to pPVN neurones comes from analysis of the brainstem noradrenergic (A1/2) inputs to the pPVN CRF neurones (115, 116). These projections have a major role in mediating activation of the HPA axis by stressors, especially physical stressors (43). Evidently, the noradrenergic input to pPVN CRF neurones is less important in stimulating HPA axis responses in lactation. Basal release of noradrenaline in the PVN tends to be reduced in lactating rats (116) and neurotoxin lesion of the noradrenergic input to the PVN reduces HPA axis responses to swim stress in virgin but not lactating rats (115). Also, the HPA axis in lactating rats is insensitive to stimulation by central administration of a noradrenergic α1 agonist that strongly stimulates the HPA axis in virgins, and this attenuation in lactating rats is likely due to reduced sensitivity of the pPVN neurones, as revealed by in vitro electrophysiological recording, albeit of unidentified PVN neurones (105). Furthermore, an α1 antagonist decreased the ACTH response in virgins but had no effect in lactating rats (116). However, in mid-lactation, α1 radioligand binding in the pPVN is not consistently altered, although α1D subtype mRNA level is selectively increased in the pPVN, but α2 receptor binding is reduced, unless the pups are removed for several hours (116). Central α2 receptor antagonist administration does not increase ACTH secretion with swim stress in lactating rats in contrast with virgins but, instead, has the reverse effect (116). Hence, there is a suckling-dependent decrease in effectiveness of the noradrenergic input to pPVN CRF neurones in lactation, involving reduced α1 receptor function. The mechanisms leading to this reduced effectiveness are unknown.

Several peptides are coexpressed in A1/2 noradrenergic neurones, as discussed above in relation to opioid restraint of HPA axis responses to stress. Another potentially important peptide is prolactin releasing peptide (PrRP), which is coexpressed in A1/2 neurones that project to the PVN (117). This peptide stimulates CRF neurones, and hence the HPA axis, in response to a range of stressors (118–120). PrRP expression is positively regulated by oestrogen (119, 121) and negatively regulated by suckling: PrRP mRNA increases within 4 h of separation from the pups (119). Whether suckling interacts with stress responses of PrRP neurones is not clear but failure of the PrRP-expressing A1/2 subset of neurones to respond to stress might contribute to attenuated HPA stress responses in lactation.

Suckling input

The suckling stimulus, or at least the presence of the pups, is of paramount importance (79). The effect of the presence of the pups depends upon the valence of the stressor: in early, but not late, lactation a stressor that threatens the pups (e.g. a conspecific male intruder or predator odour) stimulates a maternal HPA axis response, but only if the pups are present (122). Hence, this stress response appears to be contingent on fear for the safety of the pups, and lack of this response in late lactation may relate to reduced CRF mRNA expression in the amygdala (122). By contrast, exposure to strobe light repeatedly in pregnancy and lactation effectively activates the HPA axis in lactating rats, without the pups present, although only in the afternoon (123). Furthermore, in sheep, the cortisol response to emotional stressors is reduced in lactation, but more markedly if the lamb is still present and it is fully suppressed if the lamb is able to suckle (93, 124). As mentioned above, in lactating women, a recent breast feed is a necessary condition to reveal reduced HPA axis responses to an emotional stressor (96, 97), but a breast feed does not suppress the cortisol response to 35% CO2 inhalation challenge (125).

Sex steroids

The low level of oestrogen in lactation may contribute to the reduced glucocorticoid response to stressors because oestrogen enhances adrenocortical sensitivity to ACTH, as seen in pregnancy (126, 127). However, in virgin rats, ovariectomy with or without oestrogen or progesterone treatment does not replicate the suppressed HPA axis response to central α1 agonist administration that is seen in lactation (105), nor does such treatment in intact rats to simulate pregnancy levels and steroid withdrawal at birth alter HPA axis stress responses (52).

Oxytocin

In lactating women, reduced ACTH and cortisol secretion following suckling is associated with increased oxytocin secretion (107, 108). In humans, systemic oxytocin inhibits ACTH secretion (128), but whether oxytocin acts similarly on the anterior pituitary in lactating women is not known. The possible role of oxytocin released in the brain during suckling (78), from centrally-projecting parvocellular oxytocin neurone axon terminals or from magnocellular oxytocin neurone dendrites in the PVN (129, 130), has been critically tested. Initial studies indicated that centrally-released oxytocin could reduce HPA axis responses to stressors (131). However, although central infusion of an oxytocin antagonist increases basal and stressor-stimulated ACTH and corticosterone secretion in virgin rats, it has no effect in lactating rats (51). Moreover, attenuation of HPA axis responses to restraint stress by central oxytocin requires high levels of oestrogen, and oestrogen levels are low in lactation (132). Hence, centrally released oxytocin may act to reduce stress responses in other conditions (133, 134), but not in lactation.

Prolactin

Early studies suggested a correlation between suckling-stimulated prolactin secretion and reduced HPA axis responses in lactation (85, 135). Circulating prolactin from anterior pituitary lactotrophs can enter the brain, and the receptors in the choroid plexus that mediate this are up-regulated in lactation (136). Prolactin mRNA has been detected in brain (137), and stress stimulates both prolactin release in the PVN and hypothalamic prolactin mRNA expression (138); however, the neuronal source of this prolactin is not clear. Long- and short-form prolactin receptor mRNAs are expressed in the PVN (139, 140), and central infusion of prolactin in virgin rats reduces the pPVN CRF and c-fos mRNA and Fos protein responses and the ACTH secretory response to stress (141, 142), suggesting a putative role for prolactin in suppressed HPA axis responses in lactation.

Moreover, prolactin but not oxytocin infusion into the PVN reduces cortisol secretion in response to an emotional stressor in sheep, lactating or not (92). The possibility that prolactin acts in the brain to inhibit pPVN CRF neurone responses to stressors in lactation is supported by studies in rats that satisfy several criteria: prolactin receptor in the PVN, and elsewhere, is induced in lactation (143); central infusion of prolactin receptor antisense oligonucleotide substantially reverses the reduced ACTH response to a mild emotional stressor in lactating rats, and hypothalamic prolactin mRNA levels are increased in lactation (144, 145); furthermore, suckling increases prolactin release in the PVN and increases prolactin mRNA expression in the hypothalamus (138).

Influence of genotype

Lactating rats bred for ‘high’- or ‘low anxiety’ phenotype show differential suppression of responses to a social challenge (aggressive defence against an intruder), with the high-anxiety phenotype showing pPVN neurone activation and increased ACTH secretion, with no responses in the low-anxiety rats (146). The differences in emotionality between these rat strains are attributable to a single nucleotide polymorphism in the regulatory element of the vasopressin gene and consequent over-expression (147), which might underlie the greater HPA axis response in the high-anxiety rats. Alternatively the greater release of oxytocin in the PVN in these high-anxiety rats during the social stress might enhance pPVN CRF neurone responses (148).

The high-anxiety genotype and experience of stress in pregnancy interact such that exposure of high- but not low-anxiety mothers to a social stressor in pregnancy increases their HPA axis responses to mild stress in lactation (149). Furthermore, when the female offspring of high-anxiety mothers exposed to stress in pregnancy are lactating, they show loss of the normal reduction in HPA axis stress responses, and disturbance of maternal behaviour (150). These findings might relate to genetic and experiential predisposition to mood disturbances in post partum women (151).

In summary, in lactation, suckling the young over-rides the circadian rhythm control of HPA axis activity, and maintains a steady enhanced level of glucocorticoid important for milk production. Recent suckling is also a powerful suppressor of HPA axis responses to stress. Attenuated responsiveness to stressors of neural inputs to pPVN CRF neurones underlies the reduced HPA axis stress responses in lactation. This is presumably a consequence of a central effect of suckling, not yet defined, but brain prolactin may be important. A consequence of the attenuated circadian peak in corticosterone secretion and the attenuated HPA axis stress response should be to limit transfer of corticosterone in the milk, thereby giving the offspring some protection against adverse neonatal programming.

Conclusion

  1. Top of page
  2. Abstract
  3. Early pregnancy
  4. Late pregnancy
  5. Parturition
  6. Lactation
  7. Conclusion
  8. Acknowledgements
  9. References

Overall, in pregnancy, parturition and lactation, the circadian variation in HPA axis activity is dampened, whereas responses to stressors are attenuated. This will smooth fluctuations in glucocorticoid levels and actions, limiting the catabolic effects that may otherwise follow surges in secretion and aiding steady energy supply to the foetuses or offspring. The HPA adaptations are similarly expected to contribute to limiting glucocorticoid delivery to the young, hence protecting against adverse programming. The mechanisms underpinning these changes are in the brain, but are induced by actions of hormones in pregnancy and by suckling in lactation. In pregnancy, a mechanism involving the induction of an inhibitory opioid tone by a neuroactive steroid has been proposed to explain the attenuation of HPA axis responses, but this has not yet been shown to apply across the spectrum of stressors. In lactation, suckling clearly provides a neural stimulus that regulates HPA axis activity, but details of the mechanism have yet to be established.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Early pregnancy
  4. Late pregnancy
  5. Parturition
  6. Lactation
  7. Conclusion
  8. Acknowledgements
  9. References

Research in our laboratory is supported by the Biotechnology and Biological Sciences Research Council (P.J.B., J.A.R.), the Society for Endocrinology, the Medical Research Council and the Wellcome Trust (A.J.D.). Collaborations are supported by the British Council/Komitet Badan Naukowych (State Committee for Scientific Research, Poland; P.J.B., J.A.R.), and the European Union (A.J.D.).

References

  1. Top of page
  2. Abstract
  3. Early pregnancy
  4. Late pregnancy
  5. Parturition
  6. Lactation
  7. Conclusion
  8. Acknowledgements
  9. References