0013-7227/92/1312-058s$03.00/0 Endocrinology Copyright (0 1992 by The Endocrine
Vol. 131, No. 2 Prmted ,n U.S.A.
Socety
Feedback Sensitivity of the Rat Hypothalamo-PituitaryAdrenal Axis and Its Capacity to Adjust to Exogenous Corticosterone* SUSAN F. AKANA, KAREN A. SCRIBNERt, MARGARET J. BRADBURY.+, ALISON M. STRACKS, C.-DOMINIQUE WALKER, AND MARY F. DALLMAN Department
of Physiology,
University
of California
San Francisco,
San Francisco,
California
94143-0444
ABSTRACT Chronic stress causing elevated morning (AM) corticosterone (B) concentrations of 2-8 pg B/d1 does not appear to inhibit subsequent activity in the hypothalamic-pituitary-adrenal (HPA) axis, a surprising finding in view of the known depression in AM basal ACTH by only 3 pg B/d1 in adrenalectomized rats. To distinguish between the possibilities that either intact rats are less sensitive to B feedback than adrenalectomized rats, or that chronic stress facilitates responses in the HPA axis, we elevated basal B levels in young male rats with slowrelease B pellets in the absence of stress. Between 4-6 days after implantation of B pellets at three doses that elevated basal AM (diurnal trough) plasma B to approximately 1.2, 4, and 10 *g/d& we studied basal ACTH and B at trough (AM) and peak evening (PM) times of the diurnal cycle, as well as the responses to the stress of restraint and blood collection from the tail at each time of day. We also determined mean daily plasma B, insulin, and glucose from samples collected at six intervals during the day. Adrenal, thymus, and body wts were measured as were transcortin (CBG) and adrenal phenylethanolamineN-methyl transferase activity. Compared to controls implanted with wax pellets, all doses of B inhibited adrenal wt and AM stress responses and tended to inhibit pituitary ACTH content and adrenal phenylethanolamine-N-methyl transferase activity. Inhibition with the middle dose B pellet was close to maximally effective for these endpoints. Plasma glucose and thymus wt were significantly decreased and insulin was significantly increased in the middle and highest B pellet groups, with significantly greater effects at the highest dose. The gain in body wt and transcortin concentrations were significantly decreased only in
the highest dose groups, in which mean daily plasma B was approximately 10 pg/dl, a level that clearly overwhelmed the capacity of the adrenocortical system to respond to any stimulus tested. By contrast, rats with low and middle dose B pellets appeared to adjust HPA axis function by decreasing the peak diurnal increase in B, so that 24-h mean B levels did not differ from control, and were maintained at approximately 5 pg/dl. Both of these groups also had inhibited ACTH responses to stress applied during the diurnal trough (AM). By contrast, neither group had inhibited ACTH responses to stress applied during the diurnal peak (PM). We conclude that: 1) The HPA axis of intact rats is extremely sensitive to exogenous B. Small tonic increases above normal in AM plasma B, up to approximately 4 pg/dl, provoke decreased endogenous B secretion in the PM so that mean daily B levels remain at approximately 5 Fg/dl, an effect that is probably mediated by association of B with mineralocorticoid receptors. 2) The effects of elevated B on B-sensitive peripheral targets required significantly higher B levels than those observed for feedback effects on the HPA axis and are probably mediated by association of B with lower affinity glucocorticoid receptors. 3) Because the ACTH response to stress was inhibited in the AM but not in the PM by the low and middle B doses, B feedback occurs at a neural site that is parallel to the stress pathway and proximal to the CRF/arginine vasopressin neurons. Additionally, we conclude that chronic stress which causes increases of similar magnitude in trough B levels, as those seen with low and middle B doses in these studies also induces facilitation of the HPA axis, allowing continual responsiveness to new stressors. (Endocrinology 131: 585594,1992)
I
in B of this magnitude clearly provide an adequate feedback signal in adrenalectomized rats (12). The increases in CRF and arginine vasopressinmessenger RNA levels (13, 14) and basal diurnal trough ACTH secretion in the AM (12-15) after adrenalectomy can be prevented by treatment at the time of adrenal surgery with a SCpellet of B. Remarkably low amounts of circulating B, about 3 pg/dl, are sufficient to restore hypothalamic neuroendocrine factor messengerRNA expression and basal plasma ACTH in the AM to normal (12-15). Although this concentration of tonic steroid replacement normalizes basal ACTH at the trough of the diurnal cycle, it is sufficient neither to prevent some hypersecretion of ACTH after stressat this time nor to reduce basal ACTH to normal in the evening (PM), the peak of the diurnal rhythm in rats (16). By contrast, elevation of B to 10 pg/dl inhibits both stressresponsesand the PM rise in ACTH (16). These and other findings have led us (12, 17-19) and others (20-22) to propose that basal ACTH secretion in the AM is regulated by association of B with high affinity min-
T IS characteristic of intact rats and other speciesundergoing chronic or repeated stressorsthat, with time, plasma glucocorticoid levels are only slightly elevated, particularly during the trough of the diurnal rhythm (l-l 1). Nonetheless, chronically stressed rats usually exhibit thymus atrophy, evidence for corticosterone (B) secretion that is elevated above normal (1, 8). Normally basal plasma B levels in the morning (AM) are lessthan 1 pg/dl (12); chronic stressmay elevate these values to 2-8 Fg/dl (l-11), with no apparent effect on either the diurnal peak or stresslevels of B. Eleva-
tions
Received February 12, 1992. Address all correspondence and requests for reprints to: Dr. Susan F. Akana, Department of Physiology, Box 0444, University of California San Francisco School of Medicine, San Francisco, California 941430444.
* This work was supported in part by USPHS Grants DK-28172 and HL-38774. t Member of the graduate program in Endocrinology and Scholar of Achievement Reward for College Scientists, Northern California. $ Member of the Neurosciences graduate program of UCSF. 5 Supported by NIH Grant NS-07067. 585
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 22:38 For personal use only. No other uses without permission. . All rights reserved.
586
FEEDBACK
SENSITIVITY
eralocorticoid receptors (MR), whereas control of basal ACTH in the PM and stress-induced ACTH secretion additionally requires association of B with the lower affinity glucocorticoid receptors (GR). Because of the high sensitivity of adrenalectomized rats to the inhibitory effects of B, it is surprising that the elevation in AM levels during chronic stress in intact rats does not appear to act as an effective feedback signal. This could be because the hypothalamo-pituitary-adrenal (HPA) axis of intact rats is not as sensitive to B as that of adrenalectomized animals, or it could be because of the facilitatory effect of stress on subsequent activity in the HPA axis (1, 23, 24). We have recently provided direct evidence that stress acts to facilitate subsequent responsiveness in the HPA axis, and that the facilitation is observed specifically on diurnal peak basal levels in ACTH and B and on stress responsiveness during the trough, but not the peak, of the rhythm (24). To determine whether the slight rise in AM B levels that is commonly observed during conditions of chronic or repeated stress in rats is an inadequate feedback signal, or whether the rise in B is neutralized by a concurrent stressinduced facilitation of subsequent HPA responses, we performed the following experiments. Three groups of rats were provided with SC pellets of B:cholesterol in varying ratios such that 5 days later AM plasma B concentrations were approximately 1.2, 4 and 10 pg/dl, compared to sham levels of less than 1 pg/dl; B levels throughout the day and AM and PM stress responses were measured as were a variety of B-sensitive targets. We hypothesized that: 1) if, in the absence of concurrent stress, the HPA axis were sensitive to levels of B similar to the range of AM values observed under conditions of chronic stress, we should observe compensatory responses of the system attempting to adjust to the excessive morning steroid levels; and, 2) we could find the range of B over which the HPA axis is able to adjust. Portions of these results have been presented in abstract form (25). Materials
IN THE HPA AXIS
Endo. 1992 Vol 131. No 2
to be collected at a later time, the formed clot over the initial nick was wiped away and the initial wound then usually bled freely. The entire blood collection procedure occupied 1 min or less. The blood was promptly drained into chilled plastic tubes containing disodium EDTA (0.3 M; decapitation blood, 500 Nl/tube; tail blood, 20 pi/tube). Plasma separated by centrifugation was used fresh for measurement of glucose, and aliquots were frozen for later assay of ACTH, B, and occasionally transcortin (CBG) and insulin. Anterior pituitaries were collected into 2 ml 0.1 N HCI. Thymuses and adrenals were collected onto filter papersaturated with 0.9% NaCl in closed dishes for later cleaning and weighing In one experiment adrenals were removed and frozen and the activity of adrenal phenylethanolamine-h-methyl transferase (PNMT) was measured subsequently by Dr. Dona Wong, (Stanford University, Stanford, CA) (26).
Experiments In all, seven experiments were performed; the results of five of these are collapsed for presentation of basal results. The effect of B pellet implantation into rats was assessed on: basal AM and PM ACTH and B levels, as well as peripheral B targets; diurnal rhythms in ACTH, B, glucose, and insulin; the ACTH and B responses to restraint stress in the AM and PM; and the effect of an overnight (14 h) fast on AM levels of pituitary ACTH and plasma ACTH and B. Experiments performed in the PM were carried out 2 h or less before lights off, and experiments performed in the AM were completed 2 h or less after lights on in the animal room.
Assays ACTH, B, insulin, and CBG were assayed by RIA, as described previously (27-29). Pituitaries were homogenized and an aliquot of the supernatant was assayed for ACTH (30); another aliquot was used for determination of protein (31). Adrenals were homogenized 1:lOO (wt/ vol) in 20% EtOH:80% 0.154 M NaCI, and aliquots were used for determination of B. Samples were assayed in duplicate whenever possible; however, approximately 20% of the tail nick samples were run only once. Glucose was measured enzymatically using a Glucose-Analyzer 2 (Beckman Instruments, Palo Alto, CA).
Statistics The data were analyzed by regression of and two-way analysis of variance (ANOVA) ate, for repeated measures across time, and hoc testing when P < 0.05 (32). Data are Significance was accepted at P < 0.05.
steroid target on B by onecorrected, when appropriNeuman-Keuls (N-K) postplotted as means + SEM.
and Methods
Young male Sprague-Dawley derived rats (Harlan-Holtzman, Madison, WI, Bantin and Kingman, Fremont, CA, or Simonsen, Gilroy, CA, suppliers) weighing 95-105 g on delivery to UCSF were housed two per hanging basket cage in a light- (12 h) and temperature-controlled room in the UCSF animal care facility. The studies were approved by the UCSF Committee on Animal Research. Two to 3 days after arrival rats were weighed, anesthetized with ether, and fused pellets weighing approximately 100 mg containing 20%, 40%, or 80% B in cholesterol (wt/wt) (16) were imulanted subcutaneously through a small skin incision on’the iower baik (20% B, 40% B, and 80% B). The skin was closed with a clip, and the rats were returned to their cages. Wax pellets weighing approximately 100 mg were implanted into control animals (O%B). Rats were weighed again on day 4, and studies were performed on days 4-6. The rats weighed 110-180 g at this time.
Sample collection Blood samples were collected either from the trunk after decapitation or from the tail by placing the rats in plastic tubes (24), making a superficial cut over a lateral tail vein, and collecting between 70-200 ~1 blood into heparinized capillary tubes. If a second tail blood sample was
Results Effect of B pellets on basal diurnal and on B-sensitive targets
ACTH
and B, on gland wts,
In the AM, 5 days after implantation of pellets, there was a dose-related effect of the concentration of B in the implant and circulating B levels. BasalAM B concentrations in pooled data from five experiments were 0.4 -C0.1 pg/dl in controls with wax pellets (0% B); B concentrations rose progressively with increasing pellet B to 1.2 f 0.1, 4.0 f 0.5, and 10.0 + 1.7 pg/dl in rats bearing the 20%, 40%, and 80% B pellets, respectively. Pooled basal ACTH and B levels in the AM and PM are shown in Fig. 1 (top, ACTH; bottom, B). In the AM (open bars) there was no difference in ACTH between the 0% B and 20% B; ACTH was suppressedin the 40% B and 80% B groups (ANOVA: df = 3,87, F = 9.22, P < 0.001). In the PM (closed bars), in two of the five experiments, mean ACTH
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 22:38 For personal use only. No other uses without permission. . All rights reserved.
FEEDBACK
0
20
% FIG. 1. Plasma ACTH PM (W) as a function from five experiments; of each bar). Asterisks from that of the next
40
B
SENSITIVITY
IN THE
HPA AXIS
80
in Pellet
(top) and B levels (bottom) in the AM (0) and of exogenous B treatment (the data are pooled number of rats per group is indicated at the top indicate that the mean is significantly different lower B treatment group.
was higher in the 20% B group than in the 0% B group with ACTH in the 40% and 80% B groups lower; however, the differences were seldom significant. The ANOVA for pooled PM ACTH was not significant (df = 3,82, F = 2.682, P = 0.051). The data for B were consistent and clear. In the AM, there was no difference between 0% and 20% B, but both were significantly different from the 40% B group which, in turn, was significantly different from the 80% B group (ANOVA: df = 3,87, F = 29.29, P < 0.001). In the PM, B treatment groups differed from each other in roughly the opposite direction from that seen in the AM with decreasing mean levels as the concentration of B in the pellet increased (ANOVA: df = 3,94, F = 8.19, I’ < 0.001; 0% B = 20% B > 40% B = 80% B). Plasma B was significantly lower in the 20% group than in the 0% group in three of the four experiments in which both AM and PM B were measured. When tested, there were significant AM-PM differences in B in the 0% B, 20% B, and 40% B groups, but no differences in the 80% B group (ANOVA followed by N-K). Pituitary ACTH content, adrenal wt, and adrenal PNMT activity are shown in Fig. 2. There were inverse relationships
5 3
3 IO
0
20
40
80
% B in Pellet FIG. 2. Pituitary ACTH (top), adrenal wet wt (middle), activity (bottom) in rats exposed to exogenous B. Pituitary PNMT represent the results from single experiments in were measured; adrenal wt represents the pooled results experiments. Number of rats per group is indicated at the bar. Asterisks indicate that the mean is significantly lower the next lower B treatment group.
and PNMT ACTH and which they from five top of each than that of
between pituitary ACTH concentration, adrenal wt, adrenal PNMT, and B pellet dose. The first significant decreasein pituitary ACTH concentration below 0% B occurred in the 40% B pellet group (ANOVA: df = 3,36, F = 4.91, P = 0.006). Adrenal wt was significantly decreased with each
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 22:38 For personal use only. No other uses without permission. . All rights reserved.
588
FEEDBACK
SENSITIVITY
dose of B, consistently across experiments (ANOVA: df = 3,180, F = 86.58, P < 0.001; N-K, 0% > 20% (P = O.OOl), 20% > 40% (P = O.OOl), 40% > 80% (P = 0.03). Since adrenal weight is largely determined by ACTH, this result suggests strongly that the lowest dose of B provided an effective feedback signal in intact rats. Adrenal PNMT activity, which depends on B secretion from the cortex, also decreased with increasing percent B in the pellets; the first significant decrease in PNMT occurred with 40% B pellets (ANOVA: df = 3,36, F = 11.95, P < 0.001). For all of these target tissues, the inhibitory effect of B was maximal between the 20% B and 40% B pellet groups, Body wt gain over 4 days, thymus wt, and plasma CBG are shown in Fig. 3. Significant reductions in all three variables occurred with the 40-80% B pellets (ANOVA: body wt gain df = 3,182, F = 106.71, P < 0.001; thymus wt df = 3,182, F = 180.06, P < 0.001; CBG df = 3,18, F = 5.36, P = 0.008). By post-hoc N-K analysis, the first significant decrease in thymus wt occurred at the 40% B pellet dose; body wt gain and plasma CBG first significantly declined in the 80% B group. When analyzed by regression of thymus wt on plasma B, the relationship was highly significant and negative (F = 21.1, r2 = 45.7, P < 0.001). When analyzed by regression of insulin on plasma B, this relationship was also highly significant and positive (F = 15.0, r* = 40.6, P < 0.001; not shown). For all of these B-sensitive targets, the maximum effect of B occurred between the 40% B and 80% B groups. Mean plasma B, glucose, and insulin day
levels throughout
IN THE _
HPA
AXIS
Endo. 1992 Vol 1Hl -No 2
50
Jc38
the 24-h
When we estimated mean daily B levels in the preceding experiments by assuming that mean daily B is well approximated by the relationship: [2/3(basal AM B) + l/3 (basal PM B)] (12), we were interested to find that the estimated mean B values averaged 5.4, 4.5, and 5.0 pg/dl for the 0% B, 20% B, and 40% B groups, respectively (Fig. 1). The 80% B group had an average mean daily B of 9.5 pg/dl. To obtain a more precise estimate of the average daily B, groups of rats were killed at 4-h intervals around the clock. The values for the six time points were averaged for each group of rats (0% B, 20% B, 40% B, and 80% B). In this experiment, the 20% B pellet did not elevate plasma B above 1 wg/dl in the AM hours, and peak B levels were the same as in the 0% B group. The 24-h mean results are shown in Fig. 4. For the 0% B, 20% B, and 40% B groups, there was no significant difference in the average daily B concentration: 5.5, 6.0, and 5.4 pg/dl, respectively. As expected, the daily mean B concentration was higher in the 80% B group, averaging 10 pg/dl (ANOVA: df = 3,135, F = 3.36, P = 0.020; no significant differences by N-K). As in the data shown in Figs. 2 and 3, pituitary ACTH and adrenal wt were first significantly inhibited in the 20% B groups (not shown), and the first significant inhibition of thymus weight was in the 40% B group and of body wt gain was in the 80% B group (not shown). Mean daily plasma insulin (Fig. 4, middle panel) was significantly increased in the 80% B pellet groups (ANOVA: df = 3,135, F = 13.50, P < 0.001; N-K: 0% B = 20% B = 40% B < 80% B, P = 0.001). Mean glucose (Fig. 4, bottom panel)
6
6 T
4
6
+
T
0.5 P 0 0.0 - 1111
-
0
20 40 80 96 B in PELLET FIG. 3. Body wt gain (top), thymus wt (middle), and plasma CBG concentration (bottom) in rats exposed to exogenous B. Body wt gain and thymus wt represent the pooled results from five experiments; CBG was measured in a single experiment. Number of rats per group is indicated at the top of each bar. Asterisks indicate that the mean is significantly lower than that of the next lower B treatment group.
decreased with increasing percent B pellets (ANOVA: 3,135, F = 3.32, P = 0.022; no significant difference U AM and PM responses to 30.min
restraint
In the AM, there were clear B-pellet in the ACTH responses to restraint
df = by N-
and two tail nicks
dose-related with two
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 22:38 For personal use only. No other uses without permission. . All rights reserved.
decreases tail blood
FEEDBACK
SENSITIVITY
IN THE HPA AXIS
m
120
-
60
-
0 TIME
589
after
15 AM Restraint
30 (mid
FIG. 5. Plasma ACTH (top) and B (bottom) responses to restraint with tail blood sampling stress applied in the AM. Asterisks indicate that the group response is significantly less than that of the next lower B treatment group (n = 6-8 per group).
FIG. 4. Averaged values from six times of day for plasma B (top), insulin (middle), and glucose (bottom) as a function of exogenous B. By ANOVA all endpoints were significantly affected by B treatment; post-hoc analysis revealed only the insulin levels in the 80% B group to be significantly different from the other groups (*) (n = 31-36 per group).
collections (Fig. 5, top). There were significant group (df = 3,22,F=36.5,P
Vol. 131, No. 2 Prmted ,n U.S.A.
Socety
Feedback Sensitivity of the Rat Hypothalamo-PituitaryAdrenal Axis and Its Capacity to Adjust to Exogenous Corticosterone* SUSAN F. AKANA, KAREN A. SCRIBNERt, MARGARET J. BRADBURY.+, ALISON M. STRACKS, C.-DOMINIQUE WALKER, AND MARY F. DALLMAN Department
of Physiology,
University
of California
San Francisco,
San Francisco,
California
94143-0444
ABSTRACT Chronic stress causing elevated morning (AM) corticosterone (B) concentrations of 2-8 pg B/d1 does not appear to inhibit subsequent activity in the hypothalamic-pituitary-adrenal (HPA) axis, a surprising finding in view of the known depression in AM basal ACTH by only 3 pg B/d1 in adrenalectomized rats. To distinguish between the possibilities that either intact rats are less sensitive to B feedback than adrenalectomized rats, or that chronic stress facilitates responses in the HPA axis, we elevated basal B levels in young male rats with slowrelease B pellets in the absence of stress. Between 4-6 days after implantation of B pellets at three doses that elevated basal AM (diurnal trough) plasma B to approximately 1.2, 4, and 10 *g/d& we studied basal ACTH and B at trough (AM) and peak evening (PM) times of the diurnal cycle, as well as the responses to the stress of restraint and blood collection from the tail at each time of day. We also determined mean daily plasma B, insulin, and glucose from samples collected at six intervals during the day. Adrenal, thymus, and body wts were measured as were transcortin (CBG) and adrenal phenylethanolamineN-methyl transferase activity. Compared to controls implanted with wax pellets, all doses of B inhibited adrenal wt and AM stress responses and tended to inhibit pituitary ACTH content and adrenal phenylethanolamine-N-methyl transferase activity. Inhibition with the middle dose B pellet was close to maximally effective for these endpoints. Plasma glucose and thymus wt were significantly decreased and insulin was significantly increased in the middle and highest B pellet groups, with significantly greater effects at the highest dose. The gain in body wt and transcortin concentrations were significantly decreased only in
the highest dose groups, in which mean daily plasma B was approximately 10 pg/dl, a level that clearly overwhelmed the capacity of the adrenocortical system to respond to any stimulus tested. By contrast, rats with low and middle dose B pellets appeared to adjust HPA axis function by decreasing the peak diurnal increase in B, so that 24-h mean B levels did not differ from control, and were maintained at approximately 5 pg/dl. Both of these groups also had inhibited ACTH responses to stress applied during the diurnal trough (AM). By contrast, neither group had inhibited ACTH responses to stress applied during the diurnal peak (PM). We conclude that: 1) The HPA axis of intact rats is extremely sensitive to exogenous B. Small tonic increases above normal in AM plasma B, up to approximately 4 pg/dl, provoke decreased endogenous B secretion in the PM so that mean daily B levels remain at approximately 5 Fg/dl, an effect that is probably mediated by association of B with mineralocorticoid receptors. 2) The effects of elevated B on B-sensitive peripheral targets required significantly higher B levels than those observed for feedback effects on the HPA axis and are probably mediated by association of B with lower affinity glucocorticoid receptors. 3) Because the ACTH response to stress was inhibited in the AM but not in the PM by the low and middle B doses, B feedback occurs at a neural site that is parallel to the stress pathway and proximal to the CRF/arginine vasopressin neurons. Additionally, we conclude that chronic stress which causes increases of similar magnitude in trough B levels, as those seen with low and middle B doses in these studies also induces facilitation of the HPA axis, allowing continual responsiveness to new stressors. (Endocrinology 131: 585594,1992)
I
in B of this magnitude clearly provide an adequate feedback signal in adrenalectomized rats (12). The increases in CRF and arginine vasopressinmessenger RNA levels (13, 14) and basal diurnal trough ACTH secretion in the AM (12-15) after adrenalectomy can be prevented by treatment at the time of adrenal surgery with a SCpellet of B. Remarkably low amounts of circulating B, about 3 pg/dl, are sufficient to restore hypothalamic neuroendocrine factor messengerRNA expression and basal plasma ACTH in the AM to normal (12-15). Although this concentration of tonic steroid replacement normalizes basal ACTH at the trough of the diurnal cycle, it is sufficient neither to prevent some hypersecretion of ACTH after stressat this time nor to reduce basal ACTH to normal in the evening (PM), the peak of the diurnal rhythm in rats (16). By contrast, elevation of B to 10 pg/dl inhibits both stressresponsesand the PM rise in ACTH (16). These and other findings have led us (12, 17-19) and others (20-22) to propose that basal ACTH secretion in the AM is regulated by association of B with high affinity min-
T IS characteristic of intact rats and other speciesundergoing chronic or repeated stressorsthat, with time, plasma glucocorticoid levels are only slightly elevated, particularly during the trough of the diurnal rhythm (l-l 1). Nonetheless, chronically stressed rats usually exhibit thymus atrophy, evidence for corticosterone (B) secretion that is elevated above normal (1, 8). Normally basal plasma B levels in the morning (AM) are lessthan 1 pg/dl (12); chronic stressmay elevate these values to 2-8 Fg/dl (l-11), with no apparent effect on either the diurnal peak or stresslevels of B. Eleva-
tions
Received February 12, 1992. Address all correspondence and requests for reprints to: Dr. Susan F. Akana, Department of Physiology, Box 0444, University of California San Francisco School of Medicine, San Francisco, California 941430444.
* This work was supported in part by USPHS Grants DK-28172 and HL-38774. t Member of the graduate program in Endocrinology and Scholar of Achievement Reward for College Scientists, Northern California. $ Member of the Neurosciences graduate program of UCSF. 5 Supported by NIH Grant NS-07067. 585
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 22:38 For personal use only. No other uses without permission. . All rights reserved.
586
FEEDBACK
SENSITIVITY
eralocorticoid receptors (MR), whereas control of basal ACTH in the PM and stress-induced ACTH secretion additionally requires association of B with the lower affinity glucocorticoid receptors (GR). Because of the high sensitivity of adrenalectomized rats to the inhibitory effects of B, it is surprising that the elevation in AM levels during chronic stress in intact rats does not appear to act as an effective feedback signal. This could be because the hypothalamo-pituitary-adrenal (HPA) axis of intact rats is not as sensitive to B as that of adrenalectomized animals, or it could be because of the facilitatory effect of stress on subsequent activity in the HPA axis (1, 23, 24). We have recently provided direct evidence that stress acts to facilitate subsequent responsiveness in the HPA axis, and that the facilitation is observed specifically on diurnal peak basal levels in ACTH and B and on stress responsiveness during the trough, but not the peak, of the rhythm (24). To determine whether the slight rise in AM B levels that is commonly observed during conditions of chronic or repeated stress in rats is an inadequate feedback signal, or whether the rise in B is neutralized by a concurrent stressinduced facilitation of subsequent HPA responses, we performed the following experiments. Three groups of rats were provided with SC pellets of B:cholesterol in varying ratios such that 5 days later AM plasma B concentrations were approximately 1.2, 4 and 10 pg/dl, compared to sham levels of less than 1 pg/dl; B levels throughout the day and AM and PM stress responses were measured as were a variety of B-sensitive targets. We hypothesized that: 1) if, in the absence of concurrent stress, the HPA axis were sensitive to levels of B similar to the range of AM values observed under conditions of chronic stress, we should observe compensatory responses of the system attempting to adjust to the excessive morning steroid levels; and, 2) we could find the range of B over which the HPA axis is able to adjust. Portions of these results have been presented in abstract form (25). Materials
IN THE HPA AXIS
Endo. 1992 Vol 131. No 2
to be collected at a later time, the formed clot over the initial nick was wiped away and the initial wound then usually bled freely. The entire blood collection procedure occupied 1 min or less. The blood was promptly drained into chilled plastic tubes containing disodium EDTA (0.3 M; decapitation blood, 500 Nl/tube; tail blood, 20 pi/tube). Plasma separated by centrifugation was used fresh for measurement of glucose, and aliquots were frozen for later assay of ACTH, B, and occasionally transcortin (CBG) and insulin. Anterior pituitaries were collected into 2 ml 0.1 N HCI. Thymuses and adrenals were collected onto filter papersaturated with 0.9% NaCl in closed dishes for later cleaning and weighing In one experiment adrenals were removed and frozen and the activity of adrenal phenylethanolamine-h-methyl transferase (PNMT) was measured subsequently by Dr. Dona Wong, (Stanford University, Stanford, CA) (26).
Experiments In all, seven experiments were performed; the results of five of these are collapsed for presentation of basal results. The effect of B pellet implantation into rats was assessed on: basal AM and PM ACTH and B levels, as well as peripheral B targets; diurnal rhythms in ACTH, B, glucose, and insulin; the ACTH and B responses to restraint stress in the AM and PM; and the effect of an overnight (14 h) fast on AM levels of pituitary ACTH and plasma ACTH and B. Experiments performed in the PM were carried out 2 h or less before lights off, and experiments performed in the AM were completed 2 h or less after lights on in the animal room.
Assays ACTH, B, insulin, and CBG were assayed by RIA, as described previously (27-29). Pituitaries were homogenized and an aliquot of the supernatant was assayed for ACTH (30); another aliquot was used for determination of protein (31). Adrenals were homogenized 1:lOO (wt/ vol) in 20% EtOH:80% 0.154 M NaCI, and aliquots were used for determination of B. Samples were assayed in duplicate whenever possible; however, approximately 20% of the tail nick samples were run only once. Glucose was measured enzymatically using a Glucose-Analyzer 2 (Beckman Instruments, Palo Alto, CA).
Statistics The data were analyzed by regression of and two-way analysis of variance (ANOVA) ate, for repeated measures across time, and hoc testing when P < 0.05 (32). Data are Significance was accepted at P < 0.05.
steroid target on B by onecorrected, when appropriNeuman-Keuls (N-K) postplotted as means + SEM.
and Methods
Young male Sprague-Dawley derived rats (Harlan-Holtzman, Madison, WI, Bantin and Kingman, Fremont, CA, or Simonsen, Gilroy, CA, suppliers) weighing 95-105 g on delivery to UCSF were housed two per hanging basket cage in a light- (12 h) and temperature-controlled room in the UCSF animal care facility. The studies were approved by the UCSF Committee on Animal Research. Two to 3 days after arrival rats were weighed, anesthetized with ether, and fused pellets weighing approximately 100 mg containing 20%, 40%, or 80% B in cholesterol (wt/wt) (16) were imulanted subcutaneously through a small skin incision on’the iower baik (20% B, 40% B, and 80% B). The skin was closed with a clip, and the rats were returned to their cages. Wax pellets weighing approximately 100 mg were implanted into control animals (O%B). Rats were weighed again on day 4, and studies were performed on days 4-6. The rats weighed 110-180 g at this time.
Sample collection Blood samples were collected either from the trunk after decapitation or from the tail by placing the rats in plastic tubes (24), making a superficial cut over a lateral tail vein, and collecting between 70-200 ~1 blood into heparinized capillary tubes. If a second tail blood sample was
Results Effect of B pellets on basal diurnal and on B-sensitive targets
ACTH
and B, on gland wts,
In the AM, 5 days after implantation of pellets, there was a dose-related effect of the concentration of B in the implant and circulating B levels. BasalAM B concentrations in pooled data from five experiments were 0.4 -C0.1 pg/dl in controls with wax pellets (0% B); B concentrations rose progressively with increasing pellet B to 1.2 f 0.1, 4.0 f 0.5, and 10.0 + 1.7 pg/dl in rats bearing the 20%, 40%, and 80% B pellets, respectively. Pooled basal ACTH and B levels in the AM and PM are shown in Fig. 1 (top, ACTH; bottom, B). In the AM (open bars) there was no difference in ACTH between the 0% B and 20% B; ACTH was suppressedin the 40% B and 80% B groups (ANOVA: df = 3,87, F = 9.22, P < 0.001). In the PM (closed bars), in two of the five experiments, mean ACTH
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 22:38 For personal use only. No other uses without permission. . All rights reserved.
FEEDBACK
0
20
% FIG. 1. Plasma ACTH PM (W) as a function from five experiments; of each bar). Asterisks from that of the next
40
B
SENSITIVITY
IN THE
HPA AXIS
80
in Pellet
(top) and B levels (bottom) in the AM (0) and of exogenous B treatment (the data are pooled number of rats per group is indicated at the top indicate that the mean is significantly different lower B treatment group.
was higher in the 20% B group than in the 0% B group with ACTH in the 40% and 80% B groups lower; however, the differences were seldom significant. The ANOVA for pooled PM ACTH was not significant (df = 3,82, F = 2.682, P = 0.051). The data for B were consistent and clear. In the AM, there was no difference between 0% and 20% B, but both were significantly different from the 40% B group which, in turn, was significantly different from the 80% B group (ANOVA: df = 3,87, F = 29.29, P < 0.001). In the PM, B treatment groups differed from each other in roughly the opposite direction from that seen in the AM with decreasing mean levels as the concentration of B in the pellet increased (ANOVA: df = 3,94, F = 8.19, I’ < 0.001; 0% B = 20% B > 40% B = 80% B). Plasma B was significantly lower in the 20% group than in the 0% group in three of the four experiments in which both AM and PM B were measured. When tested, there were significant AM-PM differences in B in the 0% B, 20% B, and 40% B groups, but no differences in the 80% B group (ANOVA followed by N-K). Pituitary ACTH content, adrenal wt, and adrenal PNMT activity are shown in Fig. 2. There were inverse relationships
5 3
3 IO
0
20
40
80
% B in Pellet FIG. 2. Pituitary ACTH (top), adrenal wet wt (middle), activity (bottom) in rats exposed to exogenous B. Pituitary PNMT represent the results from single experiments in were measured; adrenal wt represents the pooled results experiments. Number of rats per group is indicated at the bar. Asterisks indicate that the mean is significantly lower the next lower B treatment group.
and PNMT ACTH and which they from five top of each than that of
between pituitary ACTH concentration, adrenal wt, adrenal PNMT, and B pellet dose. The first significant decreasein pituitary ACTH concentration below 0% B occurred in the 40% B pellet group (ANOVA: df = 3,36, F = 4.91, P = 0.006). Adrenal wt was significantly decreased with each
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 22:38 For personal use only. No other uses without permission. . All rights reserved.
588
FEEDBACK
SENSITIVITY
dose of B, consistently across experiments (ANOVA: df = 3,180, F = 86.58, P < 0.001; N-K, 0% > 20% (P = O.OOl), 20% > 40% (P = O.OOl), 40% > 80% (P = 0.03). Since adrenal weight is largely determined by ACTH, this result suggests strongly that the lowest dose of B provided an effective feedback signal in intact rats. Adrenal PNMT activity, which depends on B secretion from the cortex, also decreased with increasing percent B in the pellets; the first significant decrease in PNMT occurred with 40% B pellets (ANOVA: df = 3,36, F = 11.95, P < 0.001). For all of these target tissues, the inhibitory effect of B was maximal between the 20% B and 40% B pellet groups, Body wt gain over 4 days, thymus wt, and plasma CBG are shown in Fig. 3. Significant reductions in all three variables occurred with the 40-80% B pellets (ANOVA: body wt gain df = 3,182, F = 106.71, P < 0.001; thymus wt df = 3,182, F = 180.06, P < 0.001; CBG df = 3,18, F = 5.36, P = 0.008). By post-hoc N-K analysis, the first significant decrease in thymus wt occurred at the 40% B pellet dose; body wt gain and plasma CBG first significantly declined in the 80% B group. When analyzed by regression of thymus wt on plasma B, the relationship was highly significant and negative (F = 21.1, r2 = 45.7, P < 0.001). When analyzed by regression of insulin on plasma B, this relationship was also highly significant and positive (F = 15.0, r* = 40.6, P < 0.001; not shown). For all of these B-sensitive targets, the maximum effect of B occurred between the 40% B and 80% B groups. Mean plasma B, glucose, and insulin day
levels throughout
IN THE _
HPA
AXIS
Endo. 1992 Vol 1Hl -No 2
50
Jc38
the 24-h
When we estimated mean daily B levels in the preceding experiments by assuming that mean daily B is well approximated by the relationship: [2/3(basal AM B) + l/3 (basal PM B)] (12), we were interested to find that the estimated mean B values averaged 5.4, 4.5, and 5.0 pg/dl for the 0% B, 20% B, and 40% B groups, respectively (Fig. 1). The 80% B group had an average mean daily B of 9.5 pg/dl. To obtain a more precise estimate of the average daily B, groups of rats were killed at 4-h intervals around the clock. The values for the six time points were averaged for each group of rats (0% B, 20% B, 40% B, and 80% B). In this experiment, the 20% B pellet did not elevate plasma B above 1 wg/dl in the AM hours, and peak B levels were the same as in the 0% B group. The 24-h mean results are shown in Fig. 4. For the 0% B, 20% B, and 40% B groups, there was no significant difference in the average daily B concentration: 5.5, 6.0, and 5.4 pg/dl, respectively. As expected, the daily mean B concentration was higher in the 80% B group, averaging 10 pg/dl (ANOVA: df = 3,135, F = 3.36, P = 0.020; no significant differences by N-K). As in the data shown in Figs. 2 and 3, pituitary ACTH and adrenal wt were first significantly inhibited in the 20% B groups (not shown), and the first significant inhibition of thymus weight was in the 40% B group and of body wt gain was in the 80% B group (not shown). Mean daily plasma insulin (Fig. 4, middle panel) was significantly increased in the 80% B pellet groups (ANOVA: df = 3,135, F = 13.50, P < 0.001; N-K: 0% B = 20% B = 40% B < 80% B, P = 0.001). Mean glucose (Fig. 4, bottom panel)
6
6 T
4
6
+
T
0.5 P 0 0.0 - 1111
-
0
20 40 80 96 B in PELLET FIG. 3. Body wt gain (top), thymus wt (middle), and plasma CBG concentration (bottom) in rats exposed to exogenous B. Body wt gain and thymus wt represent the pooled results from five experiments; CBG was measured in a single experiment. Number of rats per group is indicated at the top of each bar. Asterisks indicate that the mean is significantly lower than that of the next lower B treatment group.
decreased with increasing percent B pellets (ANOVA: 3,135, F = 3.32, P = 0.022; no significant difference U AM and PM responses to 30.min
restraint
In the AM, there were clear B-pellet in the ACTH responses to restraint
df = by N-
and two tail nicks
dose-related with two
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 22:38 For personal use only. No other uses without permission. . All rights reserved.
decreases tail blood
FEEDBACK
SENSITIVITY
IN THE HPA AXIS
m
120
-
60
-
0 TIME
589
after
15 AM Restraint
30 (mid
FIG. 5. Plasma ACTH (top) and B (bottom) responses to restraint with tail blood sampling stress applied in the AM. Asterisks indicate that the group response is significantly less than that of the next lower B treatment group (n = 6-8 per group).
FIG. 4. Averaged values from six times of day for plasma B (top), insulin (middle), and glucose (bottom) as a function of exogenous B. By ANOVA all endpoints were significantly affected by B treatment; post-hoc analysis revealed only the insulin levels in the 80% B group to be significantly different from the other groups (*) (n = 31-36 per group).
collections (Fig. 5, top). There were significant group (df = 3,22,F=36.5,P
