Adaptive changes in basal and stress-induced HPA activity in lactating and post-lactating female rats.

Lactation represents a period of marked adaptation of the hypothalamo-pituitary-adrenal HPA axis. We characterized basal and stress-induced HPA activity during lactation and experimental weaning using dynamic blood sampling in rats. Pulsatile and diurnal corticosterone release occurred at all reproductive stages studied (virgin; day 10 of lactation; 3 and 14 days after experimental weaning on day 10 of lactation). However, in lactating rats the diurnal peak was significantly reduced, resulting in a flattened rhythm, and three days after weaning, basal HPA activity was markedly suppressed: the number of pulses and underlying basal levels of corticosterone were reduced and the diurnal rise phase delayed. Marked changes in the HPA response to 10 min noise stress also occurred at these times: being completely absent in lactating animals, but restored and highly prolonged in early weaned animals. Injection of methylprednisolone (2 mg, iv) was used to determine whether changes in fast glucocorticoid suppression correlated with these adaptive changes. Methylprednisolone induced a rapid suppression of corticosterone in virgin animals, but this effect was markedly attenuated in lactating and early weaned animals and was accompanied by significant changes in relative expression of hippocampal glucocorticoid and mineralocorticoid receptor mRNA. All effects were reversed or partially reversed 14 days after experimental weaning. Thus, the presence of the pups has an important influence on regulation of the HPA axis, and while postpartum adaptations are reversible, acute weaning evokes marked reorganisation of basal and stress-induced HPA activity.

The hypothalamo–pituitary–adrenal (HPA) axis regulates the level of circulating corticosteroids that, in turn, play a key role in metabolic homeostasis. The HPA axis has particular significance during pregnancy and lactation when the metabolic demands placed on the mother are increased, and specific adaptations of both basal and stress-induced HPA activity occur across the reproductive cycle to fulfill the needs of the offspring (1–4). Studies in rats have shown that lactation is associated with a flattening of the diurnal rhythm of secretion (5, 6), such that there is a rise in the nadir levels of corticosterone (eg, 7, 8) and a decrease in the peak evening levels. This change in HPA activity may serve several roles in the lactating rat. First, it may provide a more constant level of glucocorticoids required to cope with the increased metabolic demands, such as those associated with galactopoesis. Second, since glucocorticoids can freely enter the maternal milk and influence the offspring, the stabilization of levels may prevent neonatal exposure to varying glucocorticoid levels that are known to have long term programming effects (9, 10).

Frequent automated sampling of plasma corticosterone levels has demonstrated that HPA activity comprises a pulsatile (ultradian) pattern, with pulses of secretory activity occurring approximately once every 60–90 min, the changing amplitude of which determines the circadian rhythm (eg, 11, 12). In both rats and humans this pattern of activity has been shown to alter with pathophysiological demand. For example, we have shown changes in amplitude and frequency of pulses in response to chronic stress of adjuvant-induced arthritis (13) and to early life exposure to infective agents (14). In the present studies we investigated whether changes in pulsatile activity contribute to alterations in basal corticosterone levels both during lactation and in the period following removal of the suckling litter.

In addition to changes in basal HPA activity, a remarkable and consistent finding is that the HPA response to stress is markedly attenuated during lactation (1–4, 15). Early studies by Levine and colleagues showed that responses to footshock and ether stress were reduced in lactating rats (16, 17), and subsequent studies have shown attenuated responses to a wide variety of stresses. These include both psychological stresses—such as noise stress (18), conditioned footshock (19), forced swimming (20–22), and restraint (23, 24)—and physical stresses, such as intraperitoneal injection of NaCl (25) or lipopolysaccharide (26), or exposure to ether vapor (23). While this stress hyporesponsiveness is consistent in rodents, the situation is less clear in primates (15). Free-ranging lactating macaque monkeys show a similar response to the stress of capture as nonlactating animals (27), while in women, lactation-related stress hyporesponsiveness is seen in response to physical exercise (28) but not to breathing 35% CO2 (29). Furthermore, whether there is an attenuated response to the Trier Social Stress Test appears to depend on parity (30, 31) or whether the mother has recently breastfed (32).

In rats, stress hyporesponsiveness has its onset sometime towards the end of gestation (24, 33, 34), possibly coinciding with changing levels of ovarian steroids (35). Indeed, steroid treatment of ovariectomized virgin rats to simulate the changes in progesterone and estrogen following luteolysis produces a similar hyporesponsive state (36). While the onset of stress-hyporesponsiveness occurs within a defined time window in late gestation and is maintained by the presence of the pups, few studies have addressed the recovery of the stress response after experimental removal of the litter. It has been shown that the HPA response to ether remains suppressed 24 hours after removal of the litter (7) and that the accumulation of CRH mRNA in the paraventricular nucleus (PVN) following intraperitoneal injection of hypertonic NaCl only becomes fully restored 48–72 h after separation of the dam from the litter (25).

In the current study we characterized basal and stress-induced HPA activity at different stages of the postpartum period, particularly examining the change following experimental weaning. We also examined whether the adaptive changes relate in any way to corticosteroid receptor expression or fast inhibitory actions of exogenous corticosteroid administration (37, 38).

Materials and Methods

Animals and cannulation

All procedures were carried out in accordance with the U.K. Animals (Scientific Procedures) Act, 1986. Studies were performed on female Sprague-Dawley rats (Bantin and Kingman, Hull, United Kingdom) maintained on a 14-hour light:10-hour dark illumination cycle (lights on at 05:00 h). All animals were provided with wood chippings and a small amount of shredded paper for bedding. To obtain groups of lactating, weaned, and postlactating animals, we mated virgin rats and housed them singly at the end of pregnancy in order that they could give birth in their home cage. Virgin animals remained group-housed until the time of surgery. Cannulation was performed five days prior to sampling as described previously (11), with a liquid swivel allowing access to all parts of the home cage. The timing of studies relative to lactation and weaning are given in Table 1.

Table 1.

Timing of Various Procedures in Relation to Lactation and Weaning

Group	Surgery	Pups Removed (08:00 h)	Study 1: Start of 24-h Sampling (06:00 h)	Study 1: Noise Stress (08:00 h)	Study 2: Glucocorticoid Feedback (15:00 h)	Study 3: Tissue Collection (08:00–10:00 h)
Virgin	Any estrous stage	n/a	Day 5 after surgery	Day 6 after surgery	Day 5 after surgery	Day 6 after surgery
Lactating	Day 4 of lactation	n/a	Day 9 of lactation	Day 10 of lactation	Day 9 of lactation	Day 10 of lactation
Weaned	Day 7 of lactation	Day 10 of lactation	Day 2 after weaning	Day 3 after weaning	Day 2 after weaning	Day 3 after weaning
Post-lactation	Day 8 after weaning	Day 10 of lactation	Day 13 after weaning	Day 14 after weaning	Day 13 after weaning	Day 14 after weaning

Study 1: diurnal and stress-induced corticosterone secretion

For collection of blood samples animals were attached to an automated blood sampling system (11, 39) at 18:00 h prior to commencing sampling. A simple flush cycle was initiated to keep the cannula patent. The collection of blood samples for the measurement of corticosterone concentrations commenced automatically at 06:00 h on the day of study and continued every 10 minutes. At 08:00 h on day 2 (26 h after initiating sampling), a white-noise generator was activated and rats exposed to 114 dB for 10 minutes. Sampling then continued for a further 120 minutes.

Behavioral analysis

To examine whether episodes of corticosterone secretion were associated with specific interactions with the litter, we obtained a 5-hour period of video recording from the lactating group between 06:00 h and 11:00 h during the 24-hour sampling period and coded it for behavior. This time period matched the observation period around noise stress on the subsequent day. The time the dam was on the nest and periods of feeding, grooming, and exploratory rearing were coded, as were any milk ejection-related events. The latter comprised discrete events in which the litter displayed signs of coordinated stretching to obtain milk or the dam exhibited an accentuated arched back posture (kyphosis) at the same time as increased pup activity. Behavior was also examined during the period of noise stress in all groups. Each animal was video recorded for a period of 30 minutes, commencing 10 minutes before the onset of the noise. Behavior was coded every minute by recording 1) total activity (time spent moving; maximum 60 sec) and 2) exploration (number of rearing events).

Study 2: fast glucocorticoid suppression

To examine changes in fast glucocorticoid suppression of HPA, we cannulated separate groups of animals at times shown in Table 1 and acute injection of methylprednisolone performed as described elsewhere (38). Five days after surgery, automated sampling was initiated at 15:00 h at a frequency of once every 5 minutes. The initial 1 h of samples was discarded to avoid any nonspecific release and measurements commenced at 16:00 h. Sampling during late light phase was expected to be near the peak of the diurnal rhythm of corticosterone secretion (cf. Figure 1). Injections of methylprednisolone (2 mg methylprednisolone sodium succinate; Solu-Medrone, Pharmacia and Upjohn Ltd., Milton Keynes, United Kingdom) or 0.9% NaCl (a group of virgin animals only) were performed at 17:00 h by briefly disconnecting the iv cannula from the swivel and injecting through the sampling cannula. This procedure was completed between two samples and caused minimal disturbance to the animals. Sampling continued for a further 60 minutes after injection.

Figure 1.

Diurnal rhythms of corticosterone secretion in groups of virgin rats (A), rats on day 9 of lactation (B), weaned dams 2 days after pup removal (C), or postlactating dams 13 days after pup removal following 10 days of lactation (D). Values are mean ± SE of 6–11 animals per group and have been derived by combining successive groups of six samples to obtain an average value for every hour. In panels B–D the mean values from the virgin group have been displayed for comparison (gray line). The shaded area indicates the dark phase of the illumination cycle.

Corticosterone measurements and analysis

Total plasma corticosterone concentrations were measured directly in plasma by radioimmunoassay using a citrate buffer at pH 3.0 to denature the binding globulin (4 μl of diluted plasma fraction diluted in 100 μl buffer), antiserum kindly supplied by Professor G. Makara (Institute of Experimental Medicine, Budapest, Hungary) and [125I]-corticosterone (High Wycombe, United Kingdom) with a specific activity of 2–3 mCi/μg. The assay had a limit of detection of 5 ± 1 ng/ml (n = 20) and cross-reactivity of the assay with methylprednisolone was less than 0.4% (38).

Study 3: mineralocorticoid receptor (MR) and glucocorticoid receptor (GR) gene expression

Parallel groups of noncannulated animals were killed by decapitation between 08:00 and 10:00 h on the comparable days to those shown for noise stress in Table 1. The brains were rapidly dissected, frozen on dry ice, and stored at −80ºC prior to analysis of gene expression. Cryostat sections (12 μm) were cut to include the dorsal hippocampus and hypothalamic PVN. Three sections separated by 50 μm were collected for each animal. MR and GR mRNA expression levels were analyzed using antisense 35S-labeled riboprobes. The probes were kindly donated by J. Seckl (University of Edinburgh) and the method used was as previously described (40). All sections for a given probe were hybridized in the same reaction and exposed to autoradiographic film together with a series of 35S standards for 6 days. The films were subject to densitometric analysis using public access software Image (http://rsb.info.nih.gov/nih-image). The integrated optical density (area detected above threshold × mean optical density within this area) was measured for each area of the hippocampus and the PVN; the mean value for a given animal being determined from all sections analyzed. Reproductive status was unknown to those performing the in situ analysis.

Data analysis

The Pulsar program (41) was used to analyze number of pulses, pulse amplitude, pulse length, and mean baseline (ie, the mean value of samples that were not considered to contribute to a pulse) as previously described (11). For determination of diurnal variations in pulse characteristics the 24-hour cycle was divided into four 6-hour periods commencing at 06:00 h and analyzed separately for each animal. ANOVA and post hoc Tukey's test were used to compare endocrine and behavioral data between the different experimental groups. Hourly diurnal hormone levels were calculated from the mean of samples falling within that time period. The data for mRNA were expressed as the percentage difference from the virgin group. ANOVA and post hoc Fischer's test were used to compare all values (hormone levels, pulse characteristics, behavioral measures, and gene expression) between the experimental groups.

Results

Study 1: ultradian and diurnal rhythms of corticosterone

A significant diurnal rhythm of corticosterone levels was detected across all groups (P < .001, F = 9.560, df = 23, 541) with highest values during the period spanning the end of the light phase and the beginning of the dark phase (Figure 1). This rhythm was produced from an underlying variation in baseline secretion and from an ultradian pattern of corticosterone release, with hormone release occurring as a series of pulses that could be detected both during lactation and after experimental weaning (Figure 2). Overall analysis of the characteristics of this pulsatile corticosterone release indicated that the diurnal pattern was principally derived from changes in the number of pulses at different times of the day (Figure 3A; P < .001, F = 12.224, df = 3, 92) and from a change in the underlying baseline hormone levels (Figure 3C; P < .001, F = 20.15, df = 3, 92), but not from differences in either the amplitude (Figure 3B) or length of pulses (data not shown).

Figure 2.

Examples of ultradian patterns of corticosterone secretion in a virgin rat (A), a rat on day 9 of lactation (B), a weaned dam 2 days after pup removal (C), and a postlactating dam 13 days after removing the litter following 10 days of lactation (D). The shaded area indicates the dark phase of the illumination cycle.

Figure 3.

The characteristics of the pulsatile pattern of corticosterone release seen in virgin rats, rats on day 9 of lactation, weaned dams 2 days after pup removal, or postlactating dams 13 days after removing the litter following 10 days of lactation. The data are derived from Pulsar analysis following division of the 24-hour cycle into four 6-hour blocks; 06:00–12:00 hours (open bars), 12:00–18:00 hours (light gray bars), 18:00–24:00 hours (solid bars), and 24:00–06:00 hours (dark gray bars). The dark phase of the daily cycle ran from 19:00 to 05:00 hours. The data shown are the mean + SE of 6–11 animals per group for the number of pulses detected (A), mean pulse amplitude (B), and average baseline corticosterone secretion after exclusion of pulsatile release (C). See text for statistical comparisons.

Comparison between the groups showed significant differences in the diurnal patterns of corticosterone release (Figure 1; P < .01, F = 45.88, df = 3, 541). Within the virgin group, post hoc comparisons showed that hourly average corticosterone varied significantly over the 24-hour period: levels rose between 11:00 hours to a peak at 16:00 hours before falling again over the remainder of the 24-hour period (Figure 1A), with the greatest decline occurring between 22:00 hours and 03:00 hours. Analysis of the number of pulses in the four 6-hour time periods (Figure 3A) showed a significantly greater number were detected during both 12:00–18:00 hours and 18:00–24:00 hours than during 24:00–06:00 hours or 06:00–12:00 hours (P < .001, F = 2.95, df = 3, 28). Similarly the baseline levels of corticosterone varied significantly over the four 6-hour time blocks, being highest at 18:00–24:00 hours and lowest at 06:00–12:00 hours (P < .02, F = 4.04, df = 3, 28). Although pulse amplitude also showed some tendency to vary across the day (Figure 3B), this did not reach statistical significance, even using one-way ANOVA within this group (P = .12, F = 2.08, df = 3, 28).

While a significant diurnal variation in corticosterone release was also seen during lactation, this was associated with a flattening of the rhythm (Figure 1B). Although early morning values were comparable with those seen in the virgin group and a rise in corticosterone levels commenced in parallel in the two groups (Figure 1B), the peak values at 16:00 hours were significantly lower in lactating rats (P < .05). This flattened rhythm was also reflected in the underlying pulse characteristics: in contrast to the virgin group, the number of detected pulses did not vary significantly over the 24-hour cycle (Figure 3A). However, the baseline hormone release continued to show a diurnal variation, being significantly greater at 18:00–24:00 hours than at any of the other time periods measured (Figure 3C) (P < .001, F = 12.20, df = 3, 23). The ultradian pattern of hormone release seen in the lactating group did not appear to be related to any one aspect of specific maternal activity (please see Supplemental Figure 1 published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org), there being no clear association between any of the measured behaviors (time on the nest, milk ejection activity, feeding, or exploration [rearing]) and the timing or amplitude of pulses.

Very marked changes in corticosterone release were seen after experimental weaning. Post hoc tests revealed that corticosterone levels were significantly suppressed throughout the 24-hour period in these animals when compared with any of the other groups (Figure 1C). Although these animals did show a pulsatile pattern of hormone release (Figure 2C), post hoc tests showed that overall the number of pulses was significantly lower than any other group (Figure 3A), pulse amplitude was significantly lower than that in the virgin controls (Figure 3B), and baseline levels were significantly lower than those seen in both the lactating and postlactating groups (Figure 3C). Despite these overall lower levels, analysis within this group revealed the presence of significant diurnal variations in the number of pulses (Figure 3A; P < .01, F = 5.08, df = 3, 27) and baseline hormone levels (Figure 3C; P < .01, F = 5.49, df = 3, 27), suggesting that the diurnal drive to pulsatility was quickly re-established following removal of the suckling litter. However, there was a marked delay in the onset of the diurnal rise in corticosterone levels: in the virgin group this began between 11:00 and 12:00 hours, and in the weaned animals the rise was not evident until 15:00–16:00 hours (Figure 1C).

By 2 weeks following removal of the litter, the diurnal and ultradian rhythms of corticosterone were comparable with those seen in the virgin controls in terms of mean hormone levels (Figure 1D) and pulse characteristics (Figures 2D and 3). However, post hoc tests on the diurnal profile of hormone release did show a significant difference between the two groups, mainly due to the continued presence of a phase shift in the diurnal rise, which occurred between 13:00 and 14:00 hours, earlier than in the weaned animals but later than in the virgin controls (Figure 1).

Study 1: stress-induced HPA activity and behavior

In virgin animals, noise stress caused a rapid and significant increase in plasma corticosterone concentrations (Figure 4A; P < .001, F = 17.665, df = 17, 161): levels were significantly elevated above all basal measurements within 10 minutes of the onset of the stress, reached a peak of 177 ± 21 ng/ml after 20 minutes, before rapidly returning to baseline (Figure 4A). In contrast, no significant change in plasma corticosterone levels was seen in the lactating group following the noise stress (Figure 4B). In the experimentally weaned group the initial response to noise was similar to the virgin group—overall a significant elevation of plasma corticosterone levels was seen (P < .01, F = 3.824, df = 17, 155), being significantly raised within 10 minutes of stress onset and peaking at 20 minutes with a value of 160 ± 32 ng/ml. However, in this group, levels remained significantly elevated above all basal measurements for a further 60 minutes and only slowly declined towards baseline (Figure 4C). In contrast, the response to noise stress in the postlactating group had returned to a pattern that was indistinguishable from that seen in the virgin controls (Figure 4D).

Figure 4.

Corticosterone response to the psychological stress of white noise (110 dB × 10 min; shaded bar) in virgin rats (A), rats on day 10 of lactation (B), weaned dams 3 days after pup removal (C), or postlactating dams 14 days after pup removal following 10 days of lactation (D). Values are mean ± SE of 6–11 animals per group. *P < .05 compared to all corticosterone values prior to initiation of the noise.

Noise stress also evoked behavioral responses from the animals. Overall, there was a significant increase in total activity levels in response to the noise (Figure 5, A–D; P < .005, F = 6.288, df = 3, 68), and this response did not differ significantly between the different stages of lactation or after weaning. Most groups displayed an immediate increase in activity following onset of the noise, which was maintained in the 10-minute period after the noise (Figure 5, A, C, D). However, even though not statistically different from the other groups, the pattern in the lactating group (Figure 5B) had a different profile—the level of activity in the prenoise period tended to be greater than in the other groups, so that the increase evoked by noise was smaller, and in the postnoise period, activity levels fell as the dams directed their attention to the pups. Analysis of stress-induced exploratory behavior showed an immediate and significant increase in the number of rearing events in all groups when compared with the control period (Figure 5, E–H; P < .01, F = 13.848, df = 3, 68), which after the end of the noise, declined to low levels in most groups or to complete cessation of rearing in the lactating group (Figure 5F). Noise-induced rearing did not vary significantly with the stage of lactation, indicating that lactating animals were able to react to the stimulus despite the lack of an HPA response.

Figure 5.

The effect of noise stress of total activity levels (panels A–D) and rearing behavior (panels E–H) in virgin rats (A, E), rats on day 10 of lactation (B, F), weaned dams 3 days after pup removal (C, G), or postlactating dams 14 days after pup removal following 10 days of lactation (D, H). The values are the mean + SE of the behavior during 1-minute time periods, commencing 10 minutes prior to the period of white noise (110 dB) for 10 minutes, and continuing for 10 minutes after cessation of the stimulus. Dashed lines mark the start and end of the noise stress. The shaded blocks in panels A–D show the mean activity levels during the three 10-minute periods.

When sampled during the late light phase, iv injection of methylpredisolone caused a rapid decrease in basal plasma corticosterone levels, becoming significantly lower than the average preinjection level by 40 minutes after treatment and continuing to be suppressed throughout the remaining 2-hour sampling period (Figure 6A, solid symbols). The time taken for corticosterone levels to fall to half of preinjection levels (t½) was calculated at 9.5 ± 4.0 min. Corticosterone levels were not significantly affected by injection of saline vehicle (Figure 6A, open symbols). The response to methylprednisolone was markedly attenuated in the lactating group, with no significant decline below preinjection levels during the 120 minutes of sampling (Figure 6B). Animals that had been weaned from their litter showed a similar pattern of response to lactating animals, ie, no significant suppression. However, by 2 weeks after removing the litter, the suppressive effect of methylprednisolone was again evident (Figure 6D), although a significant effect was not detected until 60 minutes after injection and the t½ was slower (17 ± 4 min), suggesting that the loss of inhibition seen in the lactating and weaned groups had not been entirely reversed.

Figure 6.

Hypothalamo–pituitary–adrenal response to acute injection of methylprednisolone (2 mg, iv). A, Responses to saline (open symbols; n = 8) or methylprednisolone (solid symbols; n = 6) in virgin animals. B–D, Comparison of the responses on day 9 of lactation (B, n = 6), and postlactating rats on day 2 (C, n = 6) and day 13 (D, n = 5) after removing the litter following 10 days of lactation. In order to compare response profiles and compensate for group differences in basal corticosterone levels, the hormone values have been normalized within each animal to the mean value over the 120 minutes prior to injection. In each panel the normalized baseline value of 1 is indicated as a horizontal line and the timing of the injection as the vertical dashed line. *Period over which all data points in the methylprednisolone groups are significantly below preinjection baseline (P < .05).

Study 3: MR and GR mRNA expression

Animals in all groups displayed high levels of MR mRNA in all subfields of the dorsal hippocampus but in all areas, expression varied significantly between the different stages of lactation and after weaning (Cornu Ammonis [CA], P < .007; CA2, P < .001; CA3, P < .001; dentate gyrus [DG], P < .001). Post hoc tests showed that MR levels were significantly reduced during lactation and remained lower in the postlactating animals (Figure 7, A and B). GR mRNA expression was detected in the CA1 and DG (but not in CA2 and CA3) and in both of these subfields, expression also varied significantly with the phase of lactation and weaning (CA1 P < .011; DG P < .012). However, in both subfields, expression was highest in the experimentally weaned group, although this difference only reached significance in the DG where levels were significantly higher than in the postlactating group (Figure 7, C and D). When the relative ratio of MR:GR was calculated (Figure 7, E and F), a significant variation was seen over the stages of lactation and weaning within the DG (P < .003), being significantly lower in both the lactating and early weaned groups in comparison with virgin animals. A similar variation in CA1 did not reach significance. GR mRNA levels in the PVN did not vary significantly across any groups (P = .34, data not shown).

Figure 7.

Histograms showing the expression of mineralocorticoid receptor (MR) (A, B) and glucocorticoid receptor (GR) (C, D) mRNA and their relative ratios (E, F) in the CA1 subfield (A, C, E) and dentate gyrus (B, D, F) of the dorsal hippocampus of virgin rats (open columns), rats on day 10 of lactation (solid columns), and postlactating rats either 3 days (dark gray columns) or 14 days (light gray columns) after removing the litter following 10 days of lactation. Note that mRNA values are expressed in arbitrary units in relation to the mean expression level in the virgin group. Values are mean + SE of 6–11 animals per group. *P < .05 vs virgin group; P < .05 vs postlactating group; +P < .05 vs control group.

Discussion

These data demonstrate the marked changes in HPA axis activity during the course of lactation and following experimental weaning, and build on work demonstrating that the dynamic patterns of HPA activity (diurnal, pulsatile, and stress-induced) show reversible changes in response to physiological stimuli, which may have adaptive significance. Lactating animals displayed a flattening of the diurnal pattern of basal corticosterone secretion and marked suppression of stress reactivity, which results in more stable (unvarying) corticosteroid levels that may be important for maintaining metabolic activity and minimizing the programming effects on the suckling young. While these adaptive changes were largely reversed within 2 weeks of removal of the young, in the early weaning period, HPA axis activity showed a marked suppression of basal corticosterone secretion and the re-emergence of stress responsiveness, which was characterized by a highly prolonged pattern of activation. These adaptive changes may, in part, arise from a change in the relative expression of central corticosteroid receptor mRNA and the attenuation of fast glucocorticoid feedback.

Changes in pulsatile and diurnal HPA activity

There has been considerable interest in the regulation of the HPA axis during the reproductive cycle, both because of the importance of corticosteroids in programming of the neonatal nervous system (9, 10) and because of the remarkable attenuation of stress-induced HPA activation (1–4, 18–26). The present data extend earlier observations that lactation is associated with a flattening of the diurnal rhythm of basal corticosterone secretion. Previously, single-point sampling has shown a rise in the nadir levels of corticosterone (eg, 7, 8), a decrease in the peak evening levels (5, 6), or both. The current detailed characterization of the underlying dynamics of pulsatile secretion show that this flattening is not due to a loss of the underlying pulsatile pattern, but rather the number of detected pulses fails to show significant variation across the day, as it does in virgin animals. One reason for this might be that during lactation the drive to the HPA axis may have a greater dependency on the persistent sensory cues from the litter, and this may override the drive from diurnal inputs. Studies using a variety of paradigms in which the pups are separated and reunited with the dams have shown that the onset of suckling is stimulatory to HPA activity (eg, 5), a response that is contingent upon pup-generated cues (42, 43). However, large increases in pup-induced HPA activity do not occur naturally when the mother regulates the frequency and duration of the suckling bouts. Consistent with this we found that neither milk-ejection events nor the onset of individual suckling episodes show any direct relationship to the individual pulses of corticosterone secretion. Nevertheless, the presence of the litter is essential for maintaining HPA activity, as shown by the changes following experimental weaning.

The dependence of basal HPA activity on the presence of the litter is demonstrated by the fact that ACTH and corticosterone levels begin to decrease within 3.5 hours of removal of the suckling litter (5), and it has been assumed that basal HPA function reverts to normal within a rapid time frame (24–48 h) after weaning (5, 8, 15). However, the present data show that during this early postlactating period, basal corticosterone secretion is far from normal. While a clear diurnal pattern of secretion was restored within 3 days of weaning, the overall level of corticosterone is markedly suppressed in the absence of the litter. Interestingly, the greatest effect occurs during the late light phase, leading to a marked phase delay in the diurnal rise in corticosterone (cf. Figure 1, A and C), and a residual effect is still seen during this time period 2 weeks after weaning (Figure 1D). While the underlying mechanisms of these adaptive changes in afferent input remain to be determined, it is likely that these changes in diurnal patterns of HPA activity are a reflection of the modulation of circadian systems that occur during lactation (44) and could involve modulation of the circadian generator in the suprachiasmatic nucleus. Furthermore, a potential contribution may come from alteration of the noradrenergic drive that underlies diurnal HPA activity, since there has been shown to be both a reduction of noradrenergic input, which appears in late gestation (45) and persists through lactation (46), and a reduction in HPA sensitivity to α-1 adrenergic activation in the lactating rat (47).

Adaptive changes in stress responsiveness

It has been consistently observed that, in rodents, HPA responses to stress are highly attenuated during lactation (1–4, 18–26), although this hyporesponsiveness may be reversed for certain stressors that represent a threat to the offspring (48). The present data support our previous report (18) that the corticosterone response to noise stress is completely abolished in lactation. Loud white noise is one of the few stresses that has been applied in the presence of the suckling litter (thereby overcoming the possible confound of acute withdrawal of pups), and the preservation of normal behavioral reactivity (ie, stress-induced exploration) suggests that the HPA hyporesponsiveness is not the result of insensitivity to the sensory stimulus. Nevertheless, measures of stress-induced immediate-early gene (24, 26, 49, 50) or CRH (25) mRNA expression in the PVN have indicated that the strength of afferent stimulation of the HPA axis is attenuated during lactation. Indeed, since stress-induced release of oxytocin (51, 52) and prolactin (53) are also markedly reduced in lactation, it is likely that attenuation of a common stress network underlies the hyporesponsive state. Furthermore, since the present data show that basal pulsatile corticosterone secretion is largely unaffected, it is unlikely that stress hyporesponsiveness arises from downstream blockade of the HPA axis.

It has been assumed that the lactation-related attenuation of neuroendocrine and behavioral responses to stress can rapidly revert to normal after removal of the pups (15). However, very few studies have directly examined recovery of neuroendocrine responses to stress after weaning (25, 54, 55). The current data show that, while restoration to a prepregnancy state does occur after 14 days, there are marked changes arising from acute withdrawal of the litter. Particularly notable is the marked dissociation between basal and stress-induced HPA activity, in that while basal corticosterone levels are highly suppressed, the response to the relatively brief stimulus of 10 minutes noise stress involves highly prolonged HPA activation (significant elevation for >80 min). We have previously observed this persistent activation using a paradigm in which animals were exposed to two periods of noise stress (one on day 7–10 of lactation and the other 3 days later after removing the pups) (18). The present data using animals exposed to only a single noise stress show that this highly prolonged response is not due to any sensitization from a previous stress. This pattern of response is particularly notable as the sampling conditions and noise stress protocol have been applied to wide variety of pathophysiological states that modify HPA responses to stress (eg, adjuvant-induced arthritis [13], chronic hypothyroidism [56], neonatal programing [14], intra-uterine growth retardation [57], gonadectomy [12]), but this is the only instance of such prolonged responses, suggesting that a unique set of factors underlie this response. Even when the noise stimulus is extended to 30 minutes the corticosterone levels does not persist and start to decline as soon as the noise stimulus ends (eg, 38). While others have reported a slightly prolonged autonomic responses to stress after experimental weaning (42), this apparent HPA hyperresponsive state was not associated with a greater behavioral response (either number of rearings or overall activity) compared with virgin or postlactating groups, suggesting that this may be directly related to the mechanisms terminating HPA axis activation. Of the various mechanisms suggested to be responsible for modulating the termination of stress-induced HPA activity, the most widely cited is corticosteroid negative feedback.

Corticosteroid feedback

Variation in corticosteroid negative feedback is one mechanism that may underlie adaptive changes in HPA activity. It has been reported previously that corticosteroid negative feedback is intact (5, 7, 25) or even increased (54) in lactating rats. The parallel time course and magnitude of CRH mRNA accumulation following adrenalectomy of virgin and lactating rats (25), and the levels of ACTH achieved after adrenalectomy and constant corticosterone replacement (5), have been cited as evidence that the sensitivity to negative feedback is not significantly altered. However, the clamping or removal of corticosterone levels used in these studies may itself lead to a normalization of feedback mechanisms that might be otherwise modulated during lactation. In contrast to the reports in rats, studies in humans using exogenous glucocorticoid administration suggest that negative feedback may be reduced over the reproductive cycle. Both during pregnancy (58) and lactation (59) there is a reduction in the ability of dexamethasone (a GR agonist) to suppress circulating levels of cortisol. When tested during the first 2 weeks postpartum, plasma cortisol levels 17 hours after a single oral dose of dexamethasone were found to be significantly higher than those measured in nonpregnant women (59).

We examined the rapid component of negative feedback using a protocol of injecting methylprednisolone that has been shown to cause suppression of basal HPA activity and blunting of response to noise stress (37, 38). Using this approach, we found that during lactation and the early postlactating period there was a significant attenuation of this rapid HPA suppression by methylprednisolone. Importantly, this effect was reversed by 2 weeks postlactation when HPA function is largely restored. Our characterization of the effect of methylprednisolone has shown a component of the response is due to a central (suprapituitary) site of action (38). We considered whether this attenuation of negative feedback might arise from changes in corticosteroid receptor expression, since Meaney and colleagues (60) had reported that the level of binding of [3H]dexamethasone in the soluble fraction prepared from the hippocampus was reduced in lactation. Although it will act on both GR and MR, methylprednisolone has a preferential selectivity for GR, and recent characterization of the rapid feedback effect suggests the involvement of both GR (38) and MR (37). Our data showed that there was no significant fall in GR mRNA in lactation while there was small increase after experimental weaning, an effect that may arise from the fact that hippocampal GR mRNA is under negative regulation by circulating corticosteroids (eg, 61, 62) and might, therefore, be triggered by the dramatic decline in circulating corticosterone. In contrast, compared with virgin animals, MR mRNA levels were reduced in the lactating and postlactating groups, and the relative expression of the two receptor mRNAs showed a decline in MR:GR ratio in the lactating and weaned animals, particularly in the dentate gyrus. Interestingly, a similar change in relative expression of MR:GR mRNA has been reported in the DG of late pregnant (day 21) rats, although this arises from a small decline in MR mRNA and a significant increase in GR mRNA (34). This change in corticosterone receptor expression in pregnancy was also shown to be associated with a slowing of the rapid negative feedback of ACTH after chemical adrenalectomy. Although the relationship between corticosteroid receptor mRNA expression and rapid negative feedback is not yet established, it is tempting to speculate that this relative change in hippocampal corticosteroid receptor expression may underlie the attenuation of fast negative feedback observed during lactation and weaning, and indeed may explain the slower turn-off of the corticosterone response to noise stress in the weaned animals.

In conclusion, we have shown that lactation and experimental weaning are associated with major changes in the dynamic regulation of HPA activity. Despite assumptions of a rapid restoration of normal patterns of regulation after weaning, our data show that in the early period following withdrawal of the litter, both basal and stress-induced HPA activity show marked adaptation of regulatory mechanisms with a dissociation between these modes of secretion (ie, basal hypoactivity and stress hyperactivity). It is proposed that the loss of stimulatory drive from the suckling stimulus combined with the persistent attenuation of rapid negative feedback may be important factors underlying this dynamic process of reorganization.