The effects of chronic sleep restriction on the hypothalamic-pituitary-adrenal axis and its interaction with abstinence from opioid use

Abstract

Study objectives

The hypothalamic-pituitary-adrenal (HPA) axis is critical in regulating responses to physiological and psychological disturbances. Chronic sleep restriction (SR) interacts with the HPA axis in ways that are poorly delineated. The present study evaluated how chronic SR alters pituitary and adrenal function. Chronic SR was studied both alone and in a model of opioid use disorder as a potential cause of HPA axis abnormalities during abstinence.

Methods

After established self-administration of oxycodone or a saline control, male and female rats were either chronically sleep-restricted or allowed to sleep ad libitum for five weeks to permit changes in phenotype to manifest. Tests of pituitary and adrenal function were then carried out using acute CRH and dexamethasone-ACTH stimulation testing.

Results

Sexual dimorphisms were prominent in the effects of chronic SR on the HPA axis which did not vary by prior opioid exposure. There were essentially no abnormalities in the HPA axis that were due to prior opioid exposure alone. In male SR rats, basal corticosterone concentrations decreased, ACTH responses to stimulation were enhanced, and ACTH suppression by dexamethasone was reduced. In female SR rats, the corticosterone response to CRH-stimulated ACTH release peaked early. Both male and female SR rats consumed more food relative to body weight than comparison rats, indicating homeostatic disruption that is known to require HPA axis mediation.

Conclusions

Chronic SR interferes with HPA axis dynamics in sexually dimorphic ways that are expected to differentially affect SR-induced pathophysiology and disease risks. Chronic SR caused the HPA axis abnormalities observed during abstinence, providing a biological linkage between two hypothesized risk factors in vulnerability to drug taking and relapse that demonstrate sexual dimorphisms.

Graphical Abstract

Graphical Abstract.

Statement of Significance.

The hypothalamic-pituitary-adrenal (HPA) axis regulates an array of critical functions, such as fuel management and responses to physiological and psychogenic stressors. Fundamental gaps in knowledge include how chronic sleep restriction alters HPA axis regulation. The present study in rats provides increased knowledge of the loci and mechanisms of altered HPA axis activity resulting from chronic sleep restriction, both as a frequently encountered condition and as a potential cause of co-morbid HPA axis abnormalities in opioid use disorder during the abstinence period. The findings are expected to translate to humans because of common mechanisms of regulation of both sleep and the HPA axis, advancing understanding of how chronic sleep restriction results in a myriad of disease risks and vulnerabilities.

Introduction

Chronic sleep restriction results in insidious disease processes and a variety of disease risks [1–3]. Here we focus on the effects of chronic sleep restriction on adrenal and pituitary sensitivities and negative feedback within the hypothalamic-pituitary-adrenal (HPA) axis; an essential homeostatic regulatory system for a vast array of life processes, such as fuel management and adaptive responses to stress [4–6]. We furthermore studied chronic sleep restriction during abstinence from opioid use as one scenario where sleep restriction may be posing a disease risk by virtue of causing changes to the HPA axis.

Evidence to date, obtained mostly in males, suggests chronic sleep restriction results in a blunting of activity in the hypothalamic-pituitary-adrenal (HPA) axis, rather than overactivation as commonly assumed, as described below. We have previously reported that chronic sleep restriction in male rats results in reduced basal corticosterone levels without affecting adrenal weights [7]. Using restraint as a highly reproducible and standardized acute stressor, chronic sleep restriction attenuated the increase in adrenocorticotropic hormone (ACTH) and corticosterone in male rats and prevented the full nadir in ACTH and corticosterone during recovery from restraint in female rats, indicating a sexually dimorphic phenomenon of potential clinical significance [8]. In a different sleep restriction paradigm, the ACTH response to restraint stress in male rats was attenuated by eight days of sleep restriction despite elevated basal corticosterone during sleep loss procedures [9]. In mice, sleep deprivation produced by a modified platform technique, which is arguably stressful, attenuated the corticosterone response to restraint stress in both sexes [10]. The available evidence therefore indicates attenuation of components of the HPA axis, especially in male rats which have been studied more than female rats.

Alterations in HPA axis regulation during sleep deficiency in humans are comparable to those found in animal models. In virtually all subjects in many studies over decades, a reliable finding is that endogenous corticosteroids are not elevated during sleep deprivation [11–38], despite widespread assumptions to the contrary. Small increases in plasma or serum cortisol concentrations and/or free salivary cortisol in sleep-restricted humans have been detected in a few studies during certain times of day [39–41], but are considered to be within the normal daily range [42, 43]. Chronically short sleep duration is associated with a flattened diurnal cortisol pattern observed in both a naturalistic study of cohorts, of which ~75% were men [44], and experimentally after six days of slightly less than four hours of sleep per night in men [34]. The ACTH response to stimulation by corticotropin-releasing hormone (CRH) was attenuated during experimentally limited sleep duration of four hours per night for two days in men [45]. Furthermore, the cortisol response to ACTH stimulation was also attenuated [45]. Overall, these outcomes suggest that responses by the pituitary or adrenal cortex are less than expected given disrupted homeostasis caused by deficiency of sleep processes [7, 46] or, in themselves, are causes of disrupted homeostasis. We therefore hypothesized that chronic sleep restriction changes pituitary or adrenal sensitivities to their normal controllers.

Chronic sleep restriction has long been suspected of perpetuating vulnerability to relapse in opioid use disorder [47–57]. During abstinence from opioid use, sleep latency is increased, total sleep time is decreased and/or fragmented by wakefulness, and sleep efficiency is poor [47, 55, 57–63]. This chronic sleep deficiency is co-morbid with HPA axis dynamics that are altered in ways that are specific among drugs of abuse and relapse vulnerability [64–66]. Studies of the HPA axis in human opioid abstinence have reported altered glucocorticoid negative feedback [64] and increased pituitary sensitivity to CRH [67], described in the Discussion. Modifications to the HPA axis during abstinence that are caused by chronic sleep restriction, such as decreased plasma glucocorticoids observed in male rats, might provide a biological basis to vulnerability in drug seeking. Experimental reduction in corticosterone by adrenalectomy or by inhibiting adrenal glucocorticoid synthesis with metyrapone is associated with increased reinstatement of opioid seeking by either the stimuli in the drug environment or by stress-induced reinstatement [68]. While clinical and experimental data overall indicate that opioids suppress the HPA axis [69], the mechanisms underlying increased opioid seeking following glucocorticoid inhibition are unknown, and the phenomenon is inconsistent with the well-described role of central CRH in promoting drug seeking [70, 71]. Additionally, rats self-administer ACTH_4-10 which does not stimulate the adrenal cortex, indicating ACTH/melanocortin (MSH)-like peptide effects on brain reinforcement mechanisms or receptor sites such as endogenous opiate receptors [72, 73]. In a clinical population of residential treatment for prescription opioid use disorder, individuals whose cortisol and total sleep time did not normalize by 60-72 days of study were more likely than others to relapse after discharge from treatment [57]. We therefore hypothesized that chronic sleep restriction causes or exacerbates HPA axis abnormalities during abstinence from opioid use rather than being an incidental co-morbidity.

In the present study, male and female rats self-administered oxycodone or served as vehicle-treated controls and then entered a prolonged period of sleep reduction and fragmentation or control conditions. This intentional reduction of sleep was below maintenance requirements for physical and mental health in humans [74] and physical health in rats, indicated by marked abnormalities in endocrine and metabolic parameters [7, 46]. The sleep reduction was only modestly more stringent than the endogenous 4.6 h per night and poor sleep efficiency documented in opioid-dependent men during abstinence [55]. After five weeks, to allow for both abstinence and the effects of chronic sleep restriction to be manifest, we employed stimulation tests of pituitary and adrenal function by means of CRH and dexamethasone-ACTH stimulation tests, respectively. The purpose of the acute CRH stimulation test was to examine the functional responsivity of the ACTH-secreting pituitary corticotrophs. This challenge test can distinguish between disorders of the hypothalamus from those of the pituitary by directly stimulating the pituitary with the hypothalamic peptide CRH [75]. In addition, the adrenal response of glucocorticoid release in response to the stimulated ACTH release can be determined. CRH stimulation testing has been used in many clinical settings, such as to study the effects on the HPA axis of sex and age [76], clinical states suggestive of hypofunctioning of the HPA axis, such as chronic fatigue syndrome [77], and for diagnostic procedures for ACTH dependent Cushing’s [78]. Dexamethasone is a potent inhibitor of ACTH synthesis and release with high affinity for the glucocorticoid receptor, and therefore useful for establishing the presence of a perturbation in the negative feedback control of the pituitary-adrenal axis and for diagnosing hypercortisolism [78]. Stimulation of the adrenal glands with exogenous ACTH during DEX suppression allows the determination of adrenocortical sensitivity to ACTH [79]. The results indicate that chronic sleep restriction results in sexually dimorphic modifications to adrenal and pituitary sensitivities both with and without prior opioid exposure.

Methods

Animals and experimental conditions

Procedures were carried out in accordance with protocols for animal care and use approved by the IACUC at The Medical College of Wisconsin (#6988). Subjects were 32 male and 32 female Sprague-Dawley rats obtained from Envigo (Madison, WI) that were 132 (6 SD) d old at the start of the baseline period prior to opioid self-administration. Rats were housed throughout the study under a 12:12 light:dark cycle with lights on at 0600 h. Rats were fed ad libitum a standard laboratory diet (LabDiet 5001) and provided with water treated by reverse osmosis. Food and water intake and body weight were measured every other day and analyzed for days 33–36 to coincide with the CRH and DEX-ACTH testing.

Surgery to implant a jugular catheter for rat drug self-administration was performed according to established protocols [80, 81]. Rats were allotted both five days of recovery after the surgery followed by eight days of baseline monitoring. During these periods the catheters were flushed twice daily, six days per week, with heparinized saline containing an antibiotic (100 mg cefazolin/mL). Rats were then assigned to same-sex opioid drug and saline vehicle groups in a quasi-randomized manner. The successive experimental periods of pre-oxycodone baseline, oxycodone self-administration, and oxycodone abstinence with and without chronic sleep restriction are provided in Figure 1.

Figure 1.

Timeline of experimental procedures. Pre SA was composed of surgery for catheter implantation, 7 d for recovery from surgery and 8 d of baseline monitoring, followed by 10 d of training the self-administration (SA) of sucrose pellets. Rats then self-administered opioid or vehicle by cued lever presses for 10 d according to fixed-ratio schedules. Rats next entered the abstinence phase and were acclimated to the sleep restriction apparatuses. During the next 37 days rats were sleep restricted (SR) or served as ambulation controls (AC). The four experimental groups per sex studied during abstinence were (1) saline treatment, AC, (2) saline treatment, SR, (3) prior opioid treatment, AC, and (4) prior opioid treatment, SR. During days 12–16 of SR/AC conditions, rats were tested for opioid drug seeking in a companion study (not reported here). Provocative testing of the HPA axis was carried out by a CRH stimulation test on day 33 or 34 and a DEX-ACTH stimulation test on day 36 or 37. N = 8/group/sex.

Oxycodone self-administration

Self-administration (SA) training of rats by operant conditioning was conducted during 2-hour sessions within a two-week period following baseline. Each operant chamber (31.5 cm × 25.4 cm × 26.7 cm, Med Associates, Fairfax, VT, USA) was equipped with one lever on each side of the front panel. Rats were trained to press the lever paired with a cue light to dispense a sucrose pellet. Pressing the active lever resulted in reward delivery, while inactive lever presses had no effect. The side of the active lever was counterbalanced within each group. Upon meeting criteria for this Fixed Ratio (FR)-1 schedule (i.e., ≥50 rewards, ≥70% of all lever presses on the active lever for three consecutive days), an FR-2 schedule was implemented that required two-level presses on the cued lever to obtain a sucrose pellet. Upon meeting the criteria for the FR-2 schedule (≥50 rewards earned, ≥70% of total lever presses on the active lever within one day), the rats were not tested again until the last day of the two-week training period as a reminder session. Three days separated this training from oxycodone self-administration.

Self-administration of intravenous oxycodone (0.1 mg/kg/infusion; Medisca, Plattsburgh, NY, USA) or saline vehicle was achieved by pressing on the cued lever, as above [80, 82]. The schedules were FR-2 on days 1–5 and FR-4 on days 6–10 with no testing during the intervening two days. The infusion volume and rate of drug delivery were adjusted for rats receiving oxycodone, based on the individual rat body weights (e.g. a 300 g rat received a 40 µL infusion over 1.6 seconds). A ten-second timeout period followed each infusion during which the lever presses were recorded but not counted toward the next infusion. After completion of oxycodone SA, catheter patency was checked (~18 hours following the final session with an infusion of Brevital [9 mg/kg iv]). Only one catheter was nonpatent in a saline-treated rat. This rat remained in the study, given the inertness of the saline and otherwise paired experimental conditions. Rats were then assigned to either be sleep-restricted (SR, N = 16 per sex) or serve as ambulatory controls (AC, N = 16 per sex). Group assignment was matched by oxycodone intake for those that received oxycodone infusions and was matched by the number of total saline infusions for rats that received saline infusions.

Chronic sleep restriction or control conditions

After oxycodone SA, rats were abstinent throughout the remainder of the study. Rats were individually housed in Bergmann-Rechtschaffen apparatuses which are large-open aired chambers that provide freedom of movement and allow for normal behaviors [46, 83]. Each rat was allowed five days post oxycodone or saline SA for acclimation to the apparatus. Chronic sleep restriction was then produced using a validated modification of the Bergmann-Rechtshaffen paradigm [46]. The housing platform rotated for 6 seconds at variable intervals to create brief ambulation requirements that were incompatible with consolidated sleep. This results in a 35% reduction in total sleep time and increased sleep fragmentation [46]. Every five days of sleep restriction was followed by two days of sleep ad libitum intended for chronicity of repeated exposure during the course of 37 days. The duration allowed for the development of subclinical physiological changes while avoiding the later onset of clinical pathology and morbidity [46]. The ambulation control conditions were similar to chronic sleep restriction except that the ambulation requirements were consolidated into periods that allowed for longer opportunities to obtain uninterrupted sleep [46]. Maintenance of the experimental conditions for sleep restriction and ambulation controls was repeatedly verified daily. Ambient temperature controlled by thermostatically controlled heat lamps maintained at 27.4 (0.5 SD) ºC. Each rat was handled daily for two min.

In a companion study, opioid drug seeking was measured during abstinence on Days 12–16 of sleep restriction and control conditions, as described in Supplementary Materials and Data–Methods. The measurements of drug seeking were conducted nearly three weeks before provocative testing of the HPA axis, and therefore are not considered germane here. The timeframe had been chosen to investigate differences in opioid seeking earlier during chronic sleep restriction compared with a later timeframe of days 27–34 which did not reveal group differences [84].

CRH and dexamethasone-ACTH stimulation tests of function of the pituitary and adrenals in vivo

For both stimulation tests, blood was sampled by the tail clip method, which is rapid, reliable, and minimally stressful [85–87]. Blood was collected into EDTA-coated plastic microcapillary tube assemblies that drained into a centrifuge tube during centrifugation. Plasma was stored at -20°C until batch assay.

CRH stimulation test

Pituitary ACTH release was stimulated by the administration of CRH on Day 33 or 34 of chronic sleep restriction or ambulation control conditions during abstinence. Blood samples were obtained at 0900 hour (Time 0) followed by administration of CRH (10 µg CRF [human, rat] acetate salt/kg BW iv; Bachem, CAS Number: 86784-80-7). Blood samples were obtained at 15, 30, and 60 min after CRH administration [88–90].

DEX-ACTH stimulation test

On day 36 or 37 of chronic sleep restriction or ambulation control conditions during abstinence, the endogenous corticosterone level was suppressed by administration of dexamethasone (dexamethasone 21-phosphate disodium salt [Sigma-Aldrich], 250 µg/kg sc) at 0600 hour. Blood samples were obtained 120 min later at 0800 hour (Time 0) to determine ACTH and corticosterone levels during dexamethasone suppression. ACTH_1–39 (1 µg/kg sc, ACTH trifluoroacetate salt (mouse, rat), Bachem, CAS Number 77465-10-2) was then administered to determine stimulated corticosterone release at 15, 30, and 60 minutes after administration of ACTH during dexamethasone suppression [79, 91–93].

Plasma ACTH and corticosterone were determined by direct radioimmunoassays as described and validated previously [94]. Briefly, the plasma ACTH immunoassay used a rabbit anti-porcine ACTH antibody and human synthetic ACTH_1-39 as standards, and an I¹²⁵- tracer (MP Biomedicals, Orangeburg, NJ; catalog #07106101, RRID:AB_2783719, https://scicrunch.org/resolver/AB_2783719). The assay has 100% cross-reactivity with ACTH_1-24 which is shared with the rat ACTH sequence and < 1% cross-reactivity with other POMC fragments. The plasma corticosterone immunoassay used a rabbit anti-corticosterone-BSA antibody and I¹²⁵-corticosterone tracer (MP Biomedicals; catalog #07120102, RRID:AB_2783720, https://scicrunch.org/resolver/AB_2783720). It has <1% cross-reactivity with other known endogenous rat corticosteroids. The intraassay and interassay coefficients of variation (CVs) are 3%–6% for ACTH and 4%–10% for corticosterone.

Statistical analysis

Of the 256 determinations of each ACTH and corticosterone, 11 data points were excluded or extrapolated as follows. ACTH concentrations for one female AC rat in both stimulation tests were omitted because of an assay insensitivity anomaly specific to this rat. In a different female AC rat, a single missing value for basal ACTH in the CRH stimulation test was assigned the group average. For one male SR rat, the 60-minutes values for each ACTH and corticosterone in the DEX-ACTH stimulation test were dubious, pointing to an unexplained deviation from protocol, and replaced by the group averages.

Box-Cox analysis was performed to select appropriate outcome transformations for the modeling to improve the additivity of the effects and normality of the residuals. Three complementary analyses were performed for each experiment: (1) the peak values were compared, (2) the area-under-the-response curve was compared, and (3) the full time-course of the response was analyzed. The area under the curve (AUC) was computed using the AUCmin method, computing the area under the curve but above the minimal value for each animal using the trapezoid method [95]. For the peak and AUC analyses, a 2 × 2 × 2 ANOVA model was fitted with treatment (opioid vs control), condition (SR vs AC), and sex (male vs female) as fully crossed fixed effects. To simplify interpretation, a backward Akaike Information Criterion (AIC)-based model selection was used. All the main effects were forced into the model, but interactions that increased the AIC the most when removed were eliminated one-by-one starting with a three-way, then two-way interactions. In the full timecourse analysis of the responses, a repeated measures ANOVA was fitted using a mixed effects linear model with a random-effects for animal and fully interacting fixed effects of treatment, condition, and sex. Targeted post-hoc comparisons of “simple effects,” i.e. levels of one factor within fixed combinations of all other factors were made based on the model, with single-step multivariate t-distribution adjustment for multiple testing within each set of simple contrasts. Because of the large number of contrasts evaluated, only the major comparisons are shown in the figures and described in the text and elaborated in Supplemental Materials and Data. The time of the peak measurement was computed for each animal and compared between the SR and AC conditions using the exact Wilcoxon test. These analyses were performed using R 4.3.1. p < .05 was considered statistically significant and values are reported as mean ± SE unless otherwise indicated.

Results

Adrenal gland and body weights

Body weights (BW) of rats at the start of abstinence were not different for those that self-administered oxycodone compared with the saline-treated groups (males: vehicle treated, 456 [13 SE] g, oxycodone SA, 448 [11 SE] g; females: vehicle treated, 262 [4 SE] g, oxycodone SA, 261 [5 SE] g, N = 8 per group). At the time of CRH and DEX-ACTH stimulation tests, average body weights on days 33–36 were 7.5% greater in male AC rats than male SR rats (AC: 436 [6 SE] g, SR: 403 [7 SE] g, p < .003, N = 16 inclusive of vehicle treatment and prior opioid SA), but not different between female AC and SR rats (AC: 251 [4 SE] g; SR: 249 [4 SE] g, N = 16 inclusive of vehicle-treatment and prior opioid SA). Adrenal weights normalized to body weight were significantly higher in female than male rats (1.75-fold difference, p < .001, Supplementary Materials and Data, Figure S1), and slightly higher overall in SR compared with AC rats considering both sexes together (SR/AC ratio: 1.09, p = .009). The interaction of opioid treatment with SR or AC was not statistically significant.

CRH stimulation test

Prior to CRH stimulation, basal ACTH at Time 0 was significantly higher in female SR rats compared with male SR rats (ratio: 1.82; p = .017), reflecting the otherwise strong tendency for females to have higher basal ACTH values than males, but basal ACTH did not differ among same-sex groups (Supplementary Materials and Data, Figure S2 and Table S1). Correspondingly, basal corticosterone concentrations were greater in female than male rats regardless of sleep status or prior opioid exposure (range in ratios: 2-fold to 3.8-fold, all p < .001, Figure 2). At Time 0, basal corticosterone was lower in male SR rats than male AC rats in both vehicle-treated and opioid-treated groups (ratio: 0.62 and 0.69, p < .001 and p = .006, respectively, Figure 2), whereas female SR and AC rats did not differ. Despite the lower basal plasma corticosterone in male SR rats, basal ACTH did not distinguish same-sex groups with or without an opioid history (see Supplementary Materials, Figure S2, Table S1).

Figure 2.

Basal plasma corticosterone in male (left panel) and female (right panel) sleep-restricted (SR) rats compared to ambulation controls (AC) during the abstinence phase of prior opioid exposure (Opioids) or vehicle (Control) treatment. Individual values are displayed in dot plots with a box showing the median and range of the 25th to 75th percentiles. (a) p < .001 and p = .006 for male AC-SR differences within Control or Opioid treatments, respectively, and (b) p < .001 for the overall main effect of sex indicated by the vertical line with caps located at the means. N = 8 per group per sex.

The CRH-stimulated ACTH response, expressed as the area under the curve (AUC), averaged 1.8-fold higher in female rats than in male rats (p < .001; Figure 3), consistent with expectations. The sexual dimorphism was modified by chronic SR—significantly higher stimulated ACTH in male SR rats compared with male AC rats (either with and without prior opioid exposure, both ratios = 1.4, p = .002), and a tendency to be lower in female SR rats than in female AC rats, regardless of opioid history (ratios SR/AC: females, 0.86, p = .28, Figure 3). The ACTH ratios and concentrations at each timepoint are shown in Supplementary Materials and Data, Table S1 and Figure S3. These data suggest that male rats under conditions of chronic sleep restriction have significantly increased pituitary sensitivity to CRH regardless of opioid history, an effect not found in SR females.

Figure 3.

Plasma ACTH and corticosterone responses to stimulation by CRH in male (left panel) and female (right panel) rats that were chronically sleep restricted (SR) or served as ambulation controls (AC) during the abstinence phase of prior opioid exposure (Opioids) or vehicle (Control) treatments. CRH was administered in a bolus dose after blood sampling for basal plasma levels measured at 15, 30, and 60 minutes post-injection. The area under the curve (AUC) is expressed as individual values in dot plots with a box showing the median and 25th to 75th percentiles. (a) p < .002 for the within-sex comparisons indicated by brackets above male treatment groups. (b) p < .001 for the main effect of sex, indicated by the vertical bracket with caps located at the means. (c) p = .034 for pairwise comparisons (adjusted for multiple comparisons) of corticosterone concentration in AC males vs. AC females in each opioid- and vehicle-treated group, which was not found in SR males vs. SR females. N = 8 per group per sex.

Very similar to the AUC calculations, analyses of the peak ACTH response to CRH stimulation showed the same patterns of sexual dimorphism and effects of chronic SR. The increase in the peak ACTH response in the SR males relative to AC males, and the tendency toward an attenuated response in SR in females, indicated by nonsignificant differences from AC females, are faithfully reflected by the AUC. The ACTH peak for all male AC rats and most other male and female rats occurred at 30 min. Otherwise there was no discernable numeric difference in timing among the groups, with the peak occurring at 15 minute for 8% of subjects and at 60 minute for 27% of subjects.

The AUC for stimulated corticosterone release differed by sex at each post-CRH timepoint, with concentrations in females averaging 1.7-fold (0.2 SD) that of males in pairwise comparisons (all p < .02). However, significant differences between males and females in the AUC for corticosterone were found in AC rats but not in SR rats, indicating an attenuation of the sex difference in the presence of SR (interaction of sex × SR, NS, Figure 3). Concentrations and pairwise comparisons at individual timepoints are provided in Supplementary Materials and Data, Table S1 and Figure S3. Overall, the corticosterone results for the AUC indicate normal adrenal sensitivity in response to stimulated ACTH, unaffected by treatments under study (Figure 3). The correlation between peak ACTH and the corticosterone concentration AUC was weak within each group which likely reflects a ceiling effect of maximum corticosterone release. The timing of the corticosterone peak in response to CRH-stimulated ACTH was earlier in SR females compared with AC females, regardless of opioid history (female AC vs SR: with opioid exposure, p = .039; without opioid exposure, p = .046, exact Wilcoxon test) (Figure 4).

Figure 4.

The time of the peak corticosterone response to CRH stimulation in male (left panel) and female (right panel) rats that were chronically sleep restricted (SR) or served as comparison ambulation controls (AC) both with or without a prior history of opioid self-administration (Opioid and Control). Data are expressed as the percentage of animals with peak corticosterone concentrations at 15, 30, or 60 min after CRH stimulation. (a) p < 0.05 for female SR-AC differences in corticosterone peak timing with and without an opioid history. N = 8 per group per sex.

DEX-ACTH stimulation test

Endogenous plasma ACTH concentrations, measured during dexamethasone suppression and before exogenous ACTH stimulation, differed between SR and AC rats; SR rats had 1.3-fold more ACTH than AC rats (p = .023, Figure 5). Pairwise comparisons revealed both male and female SR groups, inclusive of opioid exposure or vehicle treatment, had higher dexamethasone-suppressed endogenous ACTH concentrations than their same-sex AC counterparts (1.3-fold, all p = .023).

Fig. 5.

Plasma ACTH (top panel) and corticosterone (bottom panel) after 2 hours of dexamethasone suppression but before exogenous ACTH administration in male (left) and female (right) rats that were chronically sleep-restricted (SR) or served as ambulation controls (AC). Values are plotted with boxes overlaying the 25th to- 75th percentile with the horizontal line representing the median. (a) p = .023, pairwise comparisons of endogenous ACTH in SR and AC same-sex groups (adjusted for multiple testing). (b) p < .001 for a main effect of sex on corticosterone concentrations, indicated by the vertical bracket centered at the means. N = 8 per group per sex.

The dexamethasone-suppressed plasma corticosterone concentrations at Time 0 varied by sex with concentrations in females that were 3.6-fold higher than in males (p < .001, Figure 5). Pairwise comparisons of corticosterone indicate an attenuation of the female–male ratio in AC and SR rats with an opioid history compared to without an opioid history (F/M, AC or SR: vehicle-treated, 5.4 ratio, p < .0001; opioid history, 2.4 ratio, p < .076).

The peak ACTH after ACTH stimulation during dexamethasone suppression was a significant predictor of the corticosterone response; the ratios of peak ACTH to peak corticosterone or AUC were 1.5-fold and 1.8-fold, respectively (both p < .001). The peak plasma corticosterone response after ACTH stimulation during dexamethasone suppression was sexually dimorphic, with concentrations in females that were 2.3-fold those of males (p < .001, Fig. 6). Results of analyses of the plasma corticosterone AUC revealed similar outcomes to the analysis of the peak corticosterone concentration. The concentrations in corticosterone across time are shown in Supplementary Materials and Data, Table S2 and Figure S4. Pairwise comparisons of corticosterone at individual timepoints provide further clarity by revealing that chronic SR in vehicle-treated rats significantly enhanced the female-male difference in corticosterone after dexamethasone-ACTH stimulation (chronic SR, F/M: vehicle-treated, 3.6, 4.2, 3.5, all p < .04; prior opioid treatment, 2.1, 1.4, 1.2, at 15, 30 and 60 minutes, respectively). This enhancement of the male-female dimorphism in vehicle-treated SR rats was not apparent in AC rats with or without an opioid history (AC, F/M: vehicle-treated, 2.7, 3.0, 1.9; opioid history, 2.5, 2.1, 1.2. at 15, 30 and 60 minutes, respectively).

Figure 6.

Peak plasma corticosterone responses to stimulation by ACTH during dexamethasone suppression in male (left panel) and female (right panel) rats that were chronically sleep restricted (SR) or served as ambulation controls (AC) during the abstinence phase of prior opioid exposure (Opioids) or vehicle (Control) treatments. ACTH was administered in a bolus dose after 120 minutes of dexamethasone suppression. Blood was sampled for plasma levels measured at 15, 30, and 60 minutes post injection. Peak corticosterone values were adjusted for the basal levels prior to ACTH injection and expressed on a log scale. Individual values are displayed as dot plots with a box area containing the values from the 25th to 75th percentile with a horizontal line for the median. (a) p < .001 for the overall male-female difference. Interactions between sex and treatment or AC/SR by sex or treatment were not significantly different. N = 8 per group per sex.

The peak plasma corticosterone concentration occurred at 15 minutes for nearly all rats, except for one female and one male SR rat at 30 minutes. The 60-minutes adjusted corticosterone concentrations did not differ by sex, opioid exposure, or SR/AC in pairwise comparisons, indicating that there was no detection of a delay in a return toward basal amounts, although variability was high (Supplementary Materials, Table S2 and Fig. S4).

Associations with food and water intake

Food and water intakes showed significantly different interactions with SR compared with AC by sex (p = .012 and.008, respectively). Pairwise comparisons revealed how the differences are manifested. Both food intake and food intake/BW were increased in SR rats relative to AC rats in same-sex comparisons of both opioid abstinence and vehicle treatment (all p < .001, Figure 7). Furthermore, the ratio of food intake to body weight was higher in female SR rats than male SR rats in each opioid abstinence and vehicle control (both p = .001, Figure 7). Female SR rats consumed more water than female AC rats (p = .043, see Supplementary Materials and Data, Figure S5). Ratios of food intake for SR rats relative to AC rats were 1.3-fold for males and 1.6-fold for females (all p < .001). Prior exposure to opioids did not affect food or water consumption.

Figure 7.

Food intake relative to body weight in male (left panel) and female (right panel) sleep restricted (SR) rats compared to ambulation controls (AC) during the abstinence phase of prior opioid exposure (Opioids) or vehicle (Control) treatment. Individual values are displayed in dot plots with a box showing the median and range of the 25th to 75th percentiles. (a) p ≤ .001 for same-sex bracketed comparisons. (b) P = .001 for SR male-female bracketed comparisons. N = 7–8 per group per sex.

In subsequent exploratory analyses, the interaction of SR or AC by sex in food intake, food/BW, and water intake each predict the ACTH peak response to CRH stimulation (food intake, p = .029; food/BW, p = .006; water intake p = .002). The interactions between opioid history and these metabolic variables were not predictive. The peak in corticosterone after dexamethasone suppression-ACTH stimulation was predicted by male vs female differences in food or water intake or food/BW (all p < .001), but not by SR-AC or by opioid history; this is expected based on the lack of group differences.

Discussion

The purpose of this study was to test the hypothesis that chronic sleep restriction changes pituitary or adrenal function compared to ambulatory control conditions in male and female rats. We evaluated the extent to which these changes cause or exacerbate HPA axis abnormalities during opioid abstinence because chronic sleep restriction is implicated in the risk of relapse, possibly via HPA axis modifications. Major findings discussed below are (1) lack of differences in the HPA axis or metabolic parameters in comparisons of prior opioid history and no prior opioid history, (2) marked sexual dimorphisms in basal and stimulated plasma ACTH and plasma corticosterone concentrations that are consistent with the literature, (3) the dynamics of the sexual dimorphisms are different from control comparisons in the presence of chronic sleep restriction, (4) increased pituitary sensitivity to CRH simulation in male SR rats, but not female SR rats, (5) attenuated suppression of ACTH by dexamethasone in the presence of SR in both sexes compared to controls, (6) earlier corticosterone responses to CRH stimulation in female SR rats, (7) increased food consumption relative to body weight in both male and female SR rats, (8) food intake and food/BW in male and female rats have predictive power for the ACTH response to CRH stimulation, (9) changes in pituitary and adrenal function during opioid abstinence were principally attributable to SR.

Sexual dimorphism in the HPA axis

The sexual dimorphism of basal and stimulated ACTH and corticosterone concentrations was apparent in nearly every comparison. Among them, the ACTH response to stimulation by CRH was enhanced in female AC and SR rats compared with males. This is an expected finding [96]. For example, female lambs respond to CRH stimulation with higher ACTH and cortisol levels than males [97]. In humans, the ACTH response to CRH stimulation is greater in women compared with men [98]. The prevailing explanation for females having higher plasma corticosterone concentrations than males is that females have higher circulating levels of corticosteroid binding globulin (CBG) levels as well as differences in its physiological corticosteroid affinity, driven by estrogen [99–101]. The actions of sex hormones on sexually differentiated substrates in the HPA axis are believed to account for why pathophysiological and pathopsychological conditions affect one sex more than the other [99]. Within this context, our study showed that sleep restriction interacted with the sexes differently. Chronic SR enhanced CRH-stimulated ACTH in males and tended to reduce it in females (Figure 3), indicating that pituitary sensitivity is affected by SR in a sexually dimorphic manner. Numerically, chronic SR increased the magnitude of the sexual dimorphism in the corticosterone response to DEX-ACTH stimulation; the ratios of female-to-male corticosterone at each post-ACTH timepoint were higher in vehicle-treated SR rats (ratio of 3.5–4.2) compared with lower ratios in vehicle-treated AC rats (ratios of 1.9–3.0). Therefore, chronic sleep restriction modifies different pituitary and adrenal responses by sex.

The male phenotype of the HPA axis in chronic sleep restriction

The basal corticosterone concentration prior to CRH stimulation was significantly lower in male SR rats compared with male AC rats, consistent with our earlier report [46]. The present findings indicate that the lower corticosterone concentration in male SR rats compared with male AC rats occurs without group differences in basal ACTH concentration. The lower basal corticosterone concentration in male SR rats does not appear due to an insensitivity to endogenous ACTH because of a normal response to stimulated ACTH after CRH injection. Nor does the lower basal corticosterone concentration appear due to an inability for de novo glucocorticoid production by the adrenal; the DEX-ACTH stimulation test resulted in normal peak and AUC corticosterone responses not different from AC male rats. Both the low basal corticosterone concentrations and a diminished corticosterone response to restraint stress found in our earlier study [8] suggest a decrease in adrenal sensitivity and reactivity to stress in males in the presence of SR. The normal basal ACTH concentrations in the face of low basal corticosterone suggest increased negative feedback sensitivity to glucocorticoids in males.

In male SR rats, the ACTH response to CRH was augmented, demonstrating increased pituitary sensitivity. We speculate that increased pituitary sensitivity is also indicated by ACTH concentrations that were not suppressed by dexamethasone to the extent observed in same-sex AC rats. While there are many potential explanations for this latter result (e.g. changes to the blood-brain barrier [102]), attenuated ACTH suppression by dexamethasone may reflect increased central drive, altered pituitary and/or central glucocorticoid sensitivity, changes to receptor activation patterns, and synergism between neuropeptide actions with co-localized neurotransmitters [102, 103]. Prime candidates are glucocorticoid and mineralocorticoid receptors and vasopressin [103]. Vasopressin amplifies CRH signaling at the level of the pituitary [103, 104]. We previously reported increased hypothalamic PVN corticotropin-releasing factor (CRH) and arginine vasopressin (Avp) mRNA expression, as well as increased plasma copeptin in male SR rats [8], indicating a potential role for magnocellular vasopressin in the control of HPA axis during sleep restriction in males.

The female phenotype of the HPA axis in chronic sleep restriction

In female SR rats, ACTH concentrations during dexamethasone suppression (before exogenous ACTH administration) were higher than in female AC rats. On this one parameter, SR females resembled SR males although the same changes in gene expression in the prior study did not apply to females, suggesting other influences on the pituitary. Many mRNAs in the anterior pituitary and adrenal are lower in females compared to males, but they were unaffected by SR [8]. Furthermore, unlike in males, neither ACTH peak concentrations nor the AUC were enhanced in female SR rats. The early time of peak corticosterone after CRH stimulation of ACTH in most female SR rats suggests an intrinsic change in sensitivity at the level of the adrenal. Candidate mechanisms include increased adrenal sensitivity per se and/or priming by a variety of other peripheral factors. Among putative mediators are estrogen, sympathetic nervous system tone, and inflammatory factors [103, 105, 106], which are changed by sleep restriction in sex-specific ways [107–109].

In our earlier study, female SR rats failed to achieve control-level concentrations of corticosterone on the downward curve after restraint stress [8]. Both the early time of the peak corticosterone response to CRH-ACTH stimulation in the present study and the prolonged corticosterone response after restraint stress in our prior study are consistent with increased capacity or tendency to respond to a stressor, referred to as stress reactivity [110]. Females differ from males in both stress reactivity and tendency toward opioid relapse in both humans and laboratory animals [111–116]. The present results indicate that the physiological components of stress reactivity are modified by sleep restriction. This has been reported human insomnia disorder in a study composed of mainly women [117], but not in a study that mixed the sexes [118].

A role for increased energy expenditure in HPA axis regulation that is modified by chronic sleep restriction

Chronic sleep restriction increases metabolic demand and appetite in both our animal model [7, 46] and human experimental studies [119–122]. We did not find an effect of opioid exposure per se on food intake. The literature is mixed regarding the effects of opioids on food intake [123]. Both male and female SR rats increased food intake relative to body weight regardless of prior opioid exposure, significantly more than same-sex AC rats. Additionally, female SR rats consumed more water than female AC rats and increased food intake relative to body weight more than male SR rats. Our prior studies of calorimetry showed that the food consumed by SR rats is indeed metabolized and not lost in waste products [46]. Exploratory analyses indicate that food intake and food/BW predict the peak ACTH concentration after CRH stimulation in SR-AC comparisons, but not the peak corticosterone concentration after DEX-ACTH stimulation. These results support the speculation that metabolic factors play a role in the sensitivity of the pituitary to both upstream influences and glucocorticoid feedback during chronic sleep restriction. This is especially apparent in light of the fact that glucocorticoid concentrations are not increased during chronic SR, unlike other states of energy deficiency or biological stress, such as food restriction [124, 125] or cold exposure [126, 127]. In chronic food restriction in rats, for example, corticosterone increases 3- to 25-fold, depending on the severity [124]. In humans, anorexia causes hypercortisolism [128]. In a human study that combined calorie deficiency and physical strain, cortisol concentrations were lower in a no-sleep group compared with a partial-sleep group [129], indicating a strong inhibitory effect of SR on the HPA axis. In the present results, nonelevated corticosterone during an energy deficit (indicated by increased ratios of food intake to body weight) in both male and female SR rats could directly reflect either HPA axis abnormalities that cause disrupted homeostasis or be an adaptive, self-adjusting, hierarchical response to gain homeostasis, possibly accompanied with maladaptive features. Given the central role that the HPA axis plays in regulating many homeostatic systems in the body, the adaptations and maladaptation occurring as a result of changes in the HPA axis likely play a role in numerous disease risks resulting from chronic sleep restriction, such as cardiovascular disease [130], osteoporosis [131–134], and inflammatory diseases [135, 136].

HPA axis abnormalities resulting from chronic SR resemble those in human opioid abstinence

Chronic SR in the present study resulted in HPA axis dynamics similar to those observed in human abstinence from opioid use for which chronic SR is a co-morbidity, suggesting subclinical commonalities. Among the similarities are increased sensitivity to glucocorticoid negative feedback, as observed in male SR rats, and increased pituitary sensitivity to upstream factors observed in both male and female SR rats. As long as one year after methadone treatment has been discontinued, increased glucocorticoid negative feedback inhibition and/or decreased endogenous opioid inhibition of the HPA axis can be detected by metyrapone testing [64]. Increased glucocorticoid negative feedback inhibition of ACTH is also indicated by significantly lower afternoon plasma ACTH and/or cortisol levels in methadone-maintained former heroin users after dexamethasone suppression compared with normal volunteers [64]. CRH stimulation tests in men under methadone maintenance resulted in increased ACTH responses without a difference in plasma corticosterone response compared to normal volunteers [67], indicating increased pituitary sensitivity to CRH. To be considered is that studies of human abstinence enrolled mostly men or combined the sexes in the results, whereas the present results are sexually dimorphic. Furthermore, studies of HPA axis regulation during opioid abstinence usually rely on subjects who are in methadone treatment, which reverses HPA axis abnormalities resulting from chronic heroin use [137]. It is therefore not established whether methadone maintenance, prior heroin use, or comorbid sleep disturbances are the responsible circumstances in human relapse to opioid use. However, in the present study, all but one significant effect of ACTH and corticosterone responses to CRH and DEX-ACTH stimulation tests in SR rats without an opioid history were present in SR rats with an opioid history. The exception was a lower female-male ratio in the corticosterone response to DEX-ACTH stimulation in opioid exposed rats compared with non-opioid exposed rats. Therefore, SR can be an etiology of HPA axis abnormalities during opioid abstinence. This conclusion does not discount the probability of other real-life circumstances causing or modifying the HPA axis and/or interacting with sleep loss to affect outcomes of drug relapse.

Changes to the HPA axis are considered meaningful for opioid use disorder

A role for chronic sleep restriction in drug-seeking remains undemonstrated—we did not find an effect of chronic SR during abstinence on increasing drug-seeking behavior in an antecedent study with the same timeframe and identical protocols [84]. Male and female rats showed reliable drug seeking during both extinction and footshock reinstatement testing during days 32–37 after the final oxycodone SA session, but within-sex comparisons did not differ based on sleep-restricted vs. ambulation control conditions [84]. The model of opioid drug seeking that we employed in the prior study [84] is considered an optimal first approach to behavioral quantification of drug seeking during abstinence. However, different paradigms are employable to measure states and conditions that are determinants of drug seeking, such as changes to motivation, decision making, perseverance, or propensity to escalate drug dose intake [138].

On the one hand, it is possible that the changes in HPA axis regulation during chronic SR do not facilitate drug seeking. Mechanisms involved in the activation and inhibition of the HPA axis in response to threats to homeostasis are believed to be relayed directly to PVN CRH neurons from sensory organs [103]. By comparison, responses to psychogenic stimuli involve sensory information relayed to multiple limbic structures that project to the PVN, resulting in an integrated stress response to perceived external danger [103]. The overall observable HPA axis response can be comparable between systemic and psychogenic causes, but not be functionally the same [103]. Therefore, the changes observed in pituitary and adrenal function in the present study might not manifest as a change in the behavior that we measured. On the other hand, vulnerability to relapse is an elusive phenomenon for which the biological underpinnings are expected to be subtly expressed. The SR-induced changes to HPA axis observed in the present results are expected to alter reward pathways, dopamine-dependent responses, neural networks, and neuroinflammatory pathways, among others—all of which are associated with opioid addiction and relapse [139–144]. Appetitive drives, such as those demonstrated by remarkably high food and/or water intake in SR rats, are believed to facilitate the appetitive drive of drug seeking because of overlapping neurobiological mechanisms [145]. Drug-seeking behavior has been associated with both extrahypothalamic and hypothalamic corticotropin releasing factor (CRF) actions [146], which were observed in the present results in stimulation tests to measure pituitary sensitivity. Furthermore, decreased basal corticosterone concentrations in male SR rats resonate with interventions that reduce corticosterone (metyrapone and adrenalectomy) and are associated with increased opioid seeking in male rats [68].

Limitations

The post-CRH and post-DEX/ACTH timepoints of 15, 30, and 60 min allowed the determinations of pituitary and adrenal sensitivities, as planned. Additional information could theoretically have been gained by post-90 and post-120 timepoints to observe the full declining portion of the curves after stimulation [85, 86]. The declining portion of the curve could provide additional information on group differences in steroid clearance and/or changes in glucocorticoid negative feedback sensitivity [147–149].

The extent to which experimental chronic sleep restriction in an animal model interferes with the same mechanisms as endogenously driven sleep disruption in humans is undetermined. It may be duly noted that comparative data in human laboratory studies is also collected by instrumental means, i.e., forced awakenings and forced wakefulness. This limitation is weighed against the benefit of selective sleep restriction in our experimental model to specifically examine the HPA axis by sex and treatment.

Specific durations of abstinence or chronic sleep restriction may be needed for observable changes in opioid drug seeking. We selected a duration of chronic sleep restriction that was long enough to allow for adaptive/maladaptive hormonal changes to become manifest but short enough to preclude obvious disease [7, 46]. The corresponding duration of abstinence might not have captured a critical window of vulnerability to increased drug seeking.

Sex differences in sleep patterns during sleep disturbances are believed to play a role in sex differences in comorbidities [150]. It is possible therefore that sexual dimorphisms in HPA axis function during chronic sleep restriction are relatable to sex differences in sleep architecture or the composition of sleep stages obtained during chronic sleep restriction and their associated underlying mechanisms. This linkage is unexplored in the present study but may hold potential for elucidating sex-specific mechanisms that mediate outcomes [151].

Conclusions

The present study revealed that chronic SR changed the sexually dimorphic phenotypes of HPA axis regulation. The profile in chronic SR in males indicates increased pituitary sensitivity to hypothalamic and/or extrahypothalamic stimulating factors in the brain on ACTH release and increased sensitivity to glucocorticoid negative feedback concurrent with below-normal basal corticosterone. The profile in chronic SR in females indicates increased adrenal sensitivity, indicated by rapid peak corticosterone responses to ACTH after CRH stimulation without elevated or attenuated basal corticosterone concentrations. These outcomes corresponded to metabolic adaptations for which the HPA axis is well-known to regulate and modulate responses to energy utilization and physiological deficits that are expected to be biologically significant in disease risks. One disease risk we studied here was relapse to drug seeking after a history of opioid use. Vulnerability to relapse is considered to have a psychogenic, rather than systemic, basis that implicates the HPA axis. For all intents and purposes, all HPA axis abnormalities observed during abstinence were those resulting from chronic sleep restriction. While yet unproven, modifications to the HPA axis by chronic sleep restriction are expected to be integral to vulnerability to drug seeking because of overlapping neurobiological pathways between homeostatic processes in response to physiological imbalances and behavior requiring processive/anticipatory responses to stressors [103].

Funding

Research reported in this publication was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number [R01HL150523]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Additional support and facilities were provided by the Department of Medicine at the Medical College of Wisconsin.

Data Availability

All group data analyzed for HPA axis regulation are included in this article, its supplementary material, and Figshare, a data repository.