To understand how female sex hormones influence homeostatic mechanisms of sleep, we studied the effects of estradiol (E2) replacement on c-Fos immunoreactivity in sleep/wake-regulatory brain areas after sleep deprivation (SD) in ovariectomized rats. Adult rats were ovariectomized and implanted subcutaneously with capsules containing 17β-E2 (10.5 μg; to mimic diestrous E2 levels) or oil. After 2 wk, animals with E2 capsules received a single subcutaneous injection of 17β-E2 (10 μg/kg; to achieve proestrous E2 levels) or oil; control animals with oil capsules received an oil injection. Twenty-four hours later, animals were either left undisturbed or sleep deprived by “gentle handling” for 6 h during the early light phase, and killed. E2 treatment increased serum E2 levels and uterus weights dose dependently, while attenuating body weight gain. Regardless of hormonal conditions, SD increased c-Fos immunoreactivity in all four arousal-promoting areas and four limbic and neuroendocrine nuclei studied, whereas it decreased c-Fos labeling in the sleep-promoting ventrolateral preoptic nucleus (VLPO). Low and high E2 treatments enhanced the SD-induced c-Fos immunoreactivity in the laterodorsal subnucleus of the bed nucleus of stria terminalis and the tuberomammillary nucleus, and in orexin-containing hypothalamic neurons, with no effect on the basal forebrain and locus coeruleus. The high E2 treatment decreased c-Fos labeling in the VLPO under nondeprived conditions. These results indicate that E2 replacement modulates SD-induced or spontaneous c-Fos expression in sleep/wake-regulatory and limbic forebrain nuclei. These modulatory effects of E2 replacement on neuronal activity may be, in part, responsible for E2's influence on sleep/wake behavior.
- tuberomammillary nucleus
- ventrolateral preoptic nucleus
there is a growing interest in understanding the role that female sex hormones play in the regulation of sleep in women. Changes in hormonal levels across the menstrual cycle, in pregnancy, and during menopause have been associated with changes in sleep patterns and sleep EEG, and often with the emergence of sleep disturbances [reviewed in (19, 51)]. Hormone replacement therapies providing estradiol (E2) alone or combined with progesterone have been reported to improve sleep quality in some postmenopausal women under baseline sleep conditions (54, 62, 76). The effectiveness of such treatments, however, remains controversial (36). It is also unclear whether hormone replacement therapy benefits recovery sleep after sleep deprivation (SD; Ref. 37). Given the evidence that chronic, partial SD is pervasive in industrialized societies (3) and the common use of hormone treatments for women in a variety of situations (31, 61), it is important to clarify how hormone replacement might influence the ability to recover lost sleep.
Immunohistochemical detection of c-Fos levels has been used as a marker of neuronal activation in many studies, including studies of neural responses to SD (reviewed in Refs. 7, 13, 65). Because most of these studies have used male rodents, little is known about how the presence of E2 in female rodents might affect SD-induced c-Fos expression. Both E2-concentrating cells (60) and neurons expressing estrogen receptors (ER)α or β (4, 70, 72) are found in many brain regions, which show increased c-Fos labeling after SD (reviewed in Refs. 7 and 13), providing a mechanism for circulating E2 to affect c-Fos responses to SD. In addition, E2 treatment can modulate c-Fos expression in some of these “SD-responsive” areas under baseline conditions or in response to challenges, such as stress (27, 59, 63, 77). A recent study from our laboratory showed no effects of a low or high dose of E2 on c-Fos immunoreactivity in any of eight brain areas examined, following 3 h of SD by forced locomotion (45). However, SD was performed 9 days after ovariectomy in that study, and residual E2 in vehicle-treated animals might have affected the results.
In the present study, we reinvestigated the possibility that E2 replacement alters neuronal c-Fos responses to SD in rats, with a focus on four well-established wake/arousal-promoting areas: the substantia innominata-magnocellular basal nucleus (SI-MBN) of the basal forebrain, the perifornical lateral hypothalamic area (PeF-LH), the tuberomammillary nucleus, and the locus coeruleus (35, 66). For comparison, c-Fos immunoreactivity was also examined in the VLPO, thought to be involved in promoting sleep (69, 74). The distributions of ER in the rat brain have been mapped; ERα and/or β are present at low to moderate concentrations in all of the sleep/wake-regulatory regions mentioned above (22, 40, 70, 72). In addition, four limbic or neuroendocrine nuclei in the forebrain that are known to contain high ER concentrations (4, 70) were examined. Ovariectomized rats were implanted subcutaneously with an E2-releasing capsule to produce stable levels of E2 comparable to diestrous levels, or with an E2-releasing capsule plus a single subcutaneous injection of E2 serum to produce a transient increase in E2 level similar to that occurring during proestrus. Single-label immunohistochemistry for c-Fos was used to examine the patterns of neuronal activation after 6 h of total SD by “gentle handling,” which involves presenting new objects and mild sensory stimuli to rats without directly touching them. We also double labeled c-Fos and orexin (also known as hypocretin) because hypothalamic orexin neurons of male rats have been shown to increase c-Fos immunoreactivity after SD (20, 21, 44, 50). Preliminary results of this study have been published in abstract form (10, 11).
MATERIALS AND METHODS
Adult female Wistar rats (Charles River Canada, St. Constant, QC, Canada), 214 to 228 g in body weight at the time of surgery, were housed initially in standard polypropylene cages in a temperature-controlled (23 ± 1°C) room on a light-dark cycle with 12 h of light daily (lights on at 0700). Rat chow and water were available ad libitum. All protocols for animal handling were approved by the Dalhousie University Committee on Laboratory Animals in accordance with the guidelines of the Canadian Council on Animal Care.
Ovariectomy and E2 capsule implantation.
Rats were ovariectomized bilaterally under anesthesia (60 mg/kg ip ketamine, 3.2 mg/kg ip xylazine, and 0.6 mg/kg ip acepromazine). A ventral midline incision was made through the skin and muscle layers, and the uterine horns were pulled out of the abdominal cavity. The oviducts were then clamped with forceps, and the ovaries were removed. The uterine horns were replaced inside the abdomen and the muscle layers were sutured. A capsule containing either the major female estrogen, 17β-E2 (10.5 μg in 0.06 ml sesame oil; Catalog No. E8875; Sigma-Aldrich, St. Louis, MO) or sesame oil (0.06 ml; Sigma-Aldrich) was inserted subcutaneously lateral to the incision in each animal. After capsule implantation, skin layers were closed and stapled. Following surgery, animals received an injection of the analgesic ketoprofen (5 mg/kg sc) and were kept warm until they emerged from anesthesia.
Capsules made of Silastic brand silicon tubing (1.6 mm inner diameter, 3.2 mm outer diameter, 35 mm in length; Dow Corning, Midland, MI) were used for chronic administration of E2. Silastic implants are commonly used for chronic administration of sex hormones; after an initial rise over 24 h after implantation, serum concentrations of sex hormones decline and remain at fairly stable levels over the next several weeks (5, 29, 43). We used the same amount of E2 and the same Silastic capsule material used by Dubal et al. (17, 18), who showed that this amount of E2 and delivery system resulted in low and steady levels of serum E2 that were equivalent to diestrous levels over 8 days after implantation. Capsules were closed at both ends with silicon sealant (Dow Corning) and stored in oil or the same concentration of E2 for 1–3 days for equilibration before use. During surgery, capsules were rinsed in 95% ethanol for 10–15 min and then in saline just before implantation.
Two weeks after ovariectomy and capsule implantation, animals with E2 capsules received a single subcutaneous injection of oil (Low E2 group) or 17β-E2 (10 μg/kg in oil; High E2 group) at 1400 (i.e., 18 h before the start of SD; see below for SD procedures). This E2 dose and the timing of injection were selected to achieve relatively high levels of serum E2, equivalent to proestrus levels (29), during the SD period. We used an additional injection of E2 instead of implanting a capsule containing a high concentration of E2 to avoid chronic exposure to high levels of E2 and to mimic the transiently high E2 levels that occur during proestrus. Control animals with oil capsules received a subcutaneous injection of oil (Oil group). All E2 or oil injections were given in a volume of 0.1 ml/100 g of body weight. To acclimatize animals to the injection procedure, animals were handled with a mock injection (no needles used) daily for 3 days prior to E2 or oil injection.
Three days before SD, animals were transferred to a room where they were housed singly in a clear Plexiglas cage (40 × 30 × 40 cm) placed inside an individual experimental chamber equipped with a fan and an incandescent bulb providing the same lighting schedule as in the colony room. A transparent window in the chamber's front door allowed observation of the animal. Food and water were available ad libitum. On the day of SD, animals were either left undisturbed (control) or were sleep-deprived for 6 h starting 1–2 h after lights on (0800-0900). The SD starting times were staggered over a 1-h period to allow enough time for perfusion between animals after SD. SD was achieved as gently as possible by stimulation that did not involve direct contact with the animal, such as opening the doors of their cages, introducing new objects (plastic items such as tubes, cups and toys of different shapes) into their cages, gently shuffling their cage bedding, and applying mild acoustic stimuli. The stimuli were used only when the rats showed behavioral signs of sleepiness, that is, when they were immobile with eyes closed or started adopting a sleeping posture. Rats were not disturbed when they were spontaneously awake, that is, while grooming, feeding, drinking, or immobile with eyes open. This deprivation procedure was chosen because it allows the animal to display a relatively normal pattern of wake behaviors between interventions; in particular, this procedure does not involve forced locomotion or adjustment of postures as occurs when treadmills or rotating wheels are used. The number of interventions required to keep the animals awake was counted during the last hour of the SD period to provide a measure of sleep drive/pressure.
The behavior of nondeprived controls was monitored visually during the last 2 h of the SD period and scored every minute as active wake (eyes open and motor activity present), quiet wake (eyes open but no motor activity), or sleep (eyes closed and no body movement). A group of three to four animals, including both sleep-deprived rats and nondeprived controls with different hormone treatments were tested simultaneously.
Perfusion and tissue collection.
At the end of SD or control period, rats were given an overdose of a mixture of ketamine/xylazine/acepromazine (see Ovariectomy and E2 capsule implantation). Blood samples were collected by cardiac puncture, left to coagulate for 20–30 min at room temperature, and then centrifuged at 3,000 rpm for 10 min. Serum was collected and samples were kept frozen at −80°C until RIA for E2 levels. Immediately after blood collection, rats were perfused intracardially with 50 ml of 0.1 M phosphate-buffered saline (pH 7.4) followed by 200 ml of 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) at room temperature. Brains were removed, postfixed in the same fixative solution overnight at 4°C, and cryoprotected in 30% sucrose in 0.1 M phosphate buffer until they sank. In addition, uteri were collected from most of the animals (at least 7 animals per hormonal treatment groups; Table 1) and kept in the fixative solution. After fat was trimmed off, a 2-cm length of each uterine horn along with a 0.5-cm length of the cervix to which the uterine horn is attached was drained and weighed to provide an additional measure of E2 treatment efficacy.
Brains were cut into 40-μm coronal sections on a freezing microtome. Sections were collected in five serial sets in 0.05 M Tris-buffered saline (TBS, pH 7.4). Endogenous peroxidase activity was quenched in 10% methanol and 30% hydrogen peroxide for 5 min. Immunohistochemistry for single labeling of c-Fos protein and for double labeling of c-Fos and orexin B was conducted as previously reported (12). Briefly, sections were incubated with a rabbit polyclonal anti-c-Fos antibody (1:15,000; Catalog No. PC38, Lot # D09803; Oncogene Research Products, Cambridge, MA) for either 2 days at 4°C or overnight at room temperature (there was no difference in the quality of c-Fos staining between the two procedures). This antibody was raised against amino acid residues 4–17 of human c-Fos protein that are conserved in rats, and it stained a band of ∼55 kDa on Western blot, corresponding to the apparent molecular weight of c-Fos (manufacturer's technical information). Sections were then incubated for 60 min with a biotinylated donkey anti-rabbit IgG (1:1,000, Jackson ImmunoResearch Laboratories, West Grove, PA) and reacted for 1 h with the avidin-biotin-horseradish peroxidase complex (ABC; Vector Laboratories, Burlingame, CA). Sections were then developed in 0.02% diaminobenzidine (DAB) in the presence of 0.65% nickel ammonium sulfate and hydrogen peroxide (0.006%) to produce a black-purple reaction product in c-Fos-immunoreactive (ir) cellular nuclei.
For double-labeling of c-Fos and orexin B, sections through the PeF-LH were additionally immunostained using a goat anti-orexin B antibody (1:1,000; sc-8071, C-19, Santa Cruz Biotechnology, Santa Cruz, CA). This antibody was raised against a 19-amino acid peptide whose sequence is located within amino acids 60–100 of human prepro-orexin precursor (personal communication from Technical Services, Santa Cruz Biotechnology), and its specificity was confirmed previously (12). Sections were incubated for 60 min in a biotinylated donkey anti-goat IgG (1:1,000; Jackson ImmunoResearch Laboratories), followed by ABC, and reacted with DAB without nickel to yield a brown cytoplasmic product in orexin-ir neurons. Four to eight brains from different experimental groups were processed simultaneously. After staining, sections were mounted on gelatin-coated slides, dehydrated, and cover-slipped.
An experimenter who was blind to experimental conditions conducted cell (profile) counts in nine brain structures. These included five regions known to be implicated in behavioral state control: VLPO [0.3–0.8 mm posterior (P) to bregma]; SI-MBN (P1.3-P1.6); PeF-LH (P2.8-P3.1); tuberomammillary nucleus, dorsal and ventral divisions (P3.8-P4.3); and locus coeruleus [P9.7-P9.8; (58)]. These regions, with the exception of the VLPO, show increased c-Fos labeling during spontaneous wakefulness and in response to 2–6 h of SD (reviewed in Ref. 13). Counts were also obtained from three areas rich in ER α and β: the laterodorsal subnucleus (also called the oval nucleus) of the bed nucleus of the stria terminalis (BSTLD; P0.2-P0.3), anteroventral periventricular nucleus (P0.2-P0.4), and medial nucleus of the amygdala [P2.5-P3.1; (4, 70)]. In addition, c-Fos labeling was examined in the paraventricular hypothalamic nucleus (P1.6-P1.9) as an indirect measure of stress induced by the SD procedure, as described previously (68).
Two cell counting methods were used for different brain regions. Cell counting in the anteroventral periventricular nucleus, SI-MBN, PeF-LH, and ventral division of the tuberomammillary nucleus (TMN) was carried out manually on an Olympus BH-2 microscope at ×125 or ×250. In all other brain regions, c-Fos-ir cells were counted automatically by setting thresholds on photographic images as follows. First, sections were photographed digitally with an AxioCam camera directly mounted on a Zeiss Axiovert 200 microscope (Carl Zeiss, Oberkochen, Germany). All images were taken using the same microscope and camera settings. Captured images were then imported into Adobe Photoshop 6.0 (Adobe Systems, San Jose, CA) and converted to grayscale mode. The background level of the tissue was normalized for each image by setting the c-Fos-ir nucleus that was visually judged to be the darkest as the “black” point, and a point outside the tissue or within a blood vessel as the “white” point. Images were then imported into Scion Image (Alpha 220.127.116.11, Scion, http://www.scioncorp.com). A threshold gray level for positive c-Fos staining was set for each image such that c-Fos-ir nuclei that were judged visually to be above the background level of the tissue would be counted. The background level was variable from section to section, but pilot analyses in the VLPO showed that the threshold levels were strongly and positively correlated with the normalized background levels of the tissue (r = 0.97, P < 0.0001, n = 10 randomly selected sections), indicating that the procedure used to set a threshold visually for each image was appropriate for obtaining cell counts.
For quantitative analysis, counting boxes were used for six of the nine areas examined (BSTLD, VLPO, SI-MBN, medial amygdala, PeF-LH, and tuberomammillary nucleus; Fig. 1). The sizes and placements of the boxes used for the BSTLD, SI-MBN, and PeF-LH have been described previously in detail (12). For the VLPO (Fig. 1A), an equilateral triangle with 200-μm sides (42) was placed so that its medial border was 100 μm lateral to the optic chiasm, and the midpoint of the ventral segment of the box intersected the ventral surface of the brain. This triangle encompassed the small cluster of c-Fos-ir cells visible in the VLPO of nonsleep-deprived rats (42). For the medial amygdala (Fig. 1C), a box (1,000 μm × 500 μm) was positioned so that its dorsomedial corner was 50 μm from the intersection of the optic tract with the base of the brain, and the medial segment of the box was parallel to the optic tract. Two boxes were used for cell counts in the TMN (Fig. 1, D and E): one was located in the dorsal division (250 μm × 250 μm), with the medial segment of the box abutting the dorsal roof of the third ventricle and the midpoint of the medial segment of the box centered between the midline and the lateral surface of the third ventricle (Fig. 1D); the second box (400 μm × 100 μm) was placed in the ventral division, with the lateral segment of the box abutting the base of the brain, and the dorsolateral corner of the box positioned at 100 μm below the intersection between the base of the brain and a horizontal line tangential to the ventral edge of the fornix (Fig. 1E). The sizes and placements of the dorsal and ventral boxes in the TMN were determined on the basis of the distribution of histaminergic neurons that were identified in sections immunolabeled for histidine decarboxylase in a different rat.
Cell counting in the other three areas (anteroventral periventricular nucleus, paraventricular nucleus, and locus coeruleus) was conducted within the boundaries of these regions directly under the microscope. The borders of the two hypothalamic nuclei were delineated using a brain atlas (58). The histological texture of these regions was revealed under maximal phase contrast by adjusting the diaphragm aperture and the position of the condenser. The locus coeruleus was delineated according to the distribution of noradrenergic neurons identified in sections immunostained for tyrosine hydroxylase in a different rat. To account for variations in the area of sampling for these three brain regions, the density of c-Fos-ir neurons in these regions was first calculated by dividing the number of c-Fos-ir cells by the sectional area. This value was then multiplied by the mean area across all animals to yield an estimate of the number of c-Fos-ir cells for each brain region.
All c-Fos-ir nuclei were counted bilaterally in either one section (BSTLD, VLPO and anteroventral periventricular nucleus, dorsal and ventral divisions of the tuberomammillary nucleus) or two sections 160 μm apart (paraventricular nucleus, SI-MBN, medial amygdala, and locus coeruleus) that contained the largest nuclear areas. Similarly, in the PeF-LH, single c-Fos-ir nuclei, single orexin-ir neurons, and double-labeled neurons were counted bilaterally in the section containing the largest number of orexin-ir neurons, and the counts were averaged in each animal.
Plasma E2 concentration was determined by a technician who was unaware of treatment conditions, using a double-antibody commercial RIA kit (Product Code: KE2D1; Inter Medico, Markham, ON, Canada). For each animal, two aliquots of 200-μl samples were assayed for E2, and the two values were averaged. The detection limit of the assay was 5 pg/ml.
Statistical tests were conducted using Statview 5.0 (SAS Institute, Cary, NC) or SPSS 14.0 (SPSS, Chicago, IL). Body weights were analyzed using one-way repeated-measures ANOVA with E2 Treatment (Oil, Low E2, and High E2) as the between-subjects factor. Uterus weights were analyzed using one-way ANOVA with E2 treatment as the factor. Behavioral measures (active wake, quiet wake, and sleep) in nondeprived controls, and the numbers of interventions required to prevent sleep in sleep-deprived animals, were each analyzed using a one-way ANOVA with E2 treatment as the factor. Tukey-Kramer post hoc analyses were used for multiple comparisons when ANOVA revealed a significant main effect or interaction.
Because a number of serum E2 values fell below the detection limit of the assay (5 pg/ml), nonparametric Kruskal-Wallis tests were performed with all values <5 pg/ml tied at the lowest rank (28). Nonparametric Dunn post hoc tests were used to determine group differences (83, p. 225). The nonparametric Spearman's rank correlation coefficients were used to determine whether serum E2 levels were correlated with other physiological variables or cell counts.
Cell counts were transformed by converting to log (X+1) to equalize variances as required. For each brain structure, two-way ANOVA was used with E2 treatment and sleep condition (SD and control) as factors, followed by Tukey post hoc tests where appropriate. Pearson correlation coefficients were used to examine the relationship between cell counts and physiological variables (except for serum E2 levels). P values less than 0.05 were considered statistically significant.
Results were obtained from a total of 41 ovariectomized rats, of which 11 had received an oil capsule plus an oil injection (Oil group), 17 had received an E2 capsule (10.5 μg) plus an oil injection (Low E2 group), and 13 had received an E2 capsule plus an E2 injection (10 μg/kg; High E2 group). Of these 41 animals, 20 were sleep-deprived for a period of 6 h starting at 0800-0900, and 21 served as nondeprived controls.
Serum E2 levels, uterus weights, and body weights.
The endocrine measures from the three hormonal treatment groups are presented in Table 1, including serum E2 levels, uterus weights, and body weights. As expected, there were no significant differences in any of these measures between sleep-deprived and nondeprived animals in each hormonal treatment group (all P > 0.14); therefore, these data were combined for each E2 treatment group. Administration of E2 resulted in a dose-dependent increase in serum E2 levels [Kruskal-Wallis test: H(2) corrected for ties = 16.35, P < 0.001; n = 40], ranging from a median of ≤5 pg/ml in the Oil group, to 9 pg/ml in the Low E2 group, and 16 pg/ml in the High E2 group. The difference between High E2 and Oil was significant (P < 0.05; Table 1). Consistent with the serum E2 levels, E2 treatments increased uterus weights in a dose-dependent manner [F(2,25) = 80.99, P < 0.0001], with significant differences between any two of the three groups (all P < 0.01; Table 1); the results were similar when uterus weights were normalized to total body weights (Table 1). Both E2 treatments decreased body weight gain over the 15-day period from surgery to the SD experiment, compared with oil treatment [F(2,25) = 4.03, P < 0.05, for the Time × E2 Treatment interaction] but, unlike uterus weights, with no significant difference between the two E2 doses (Table 1).
A moderate positive correlation was found between serum E2 levels and normalized uterus weights using ranks (Spearman's rank correlation coefficient ρ = 0.637; P < 0.001; n = 28). Conversely, moderate negative correlations were found between the ranks of both serum E2 levels and body weight gains (ρ = −0.444; P < 0.025; n = 28), and between the ranks of both normalized uterus weights and body weight gains (ρ = −0.564; P < 0.01; n = 28).
Sleep-wake states during the last 2 h of the nondeprivation period.
The analysis of behavioral measures indicated that nondeprived animals were mostly (67–74% of time) asleep during the last 2-h period before perfusion (i.e., 1200-1400 or 1300-1500), regardless of hormone treatment condition (Table 2). The percentages of time spent in active wake, quiet wake, or sleep in the last 2 h did not differ among the three hormonal treatment groups (all P > 0.50; Table 2).
Numbers of interventions during the last hour of the SD period.
All rats appeared behaviorally awake during the SD period. The number of interventions required to keep the animals awake increased over the course of the SD period, and it was highest during the last hours. There were no statistically significant differences among the hormonal treatment groups in the numbers of interventions required during the last hour of the SD period [F(2, 13) < 1; NS; Table 2].
The numbers of neurons immunolabeled for c-Fos in five nuclei of the sleep/wake regulatory system and in four limbic and neuroendocrine nuclei of the forebrain that are rich in ER are summarized in Table 3 and illustrated in Figs. 2–6. For the tuberomammillary nucleus, the overall patterns of c-Fos labeling did not differ between the dorsal and ventral divisions [F(1,29) < 1], so cell counts from the two divisions were combined in each animal. In nondeprived animals, only small numbers of c-Fos-ir cells were found in most brain regions (Table 3). Overall, the effects of SD were seen consistently in all brain regions examined, whereas effects of E2 treatments and the modulation of SD effects by E2 treatments were seen only in some brain regions.
As shown in Table 3 and Figs. 3–6, regardless of hormonal condition, the number of c-Fos-ir cells in sleep-deprived animals was higher than in nondeprived controls in all brain regions examined except the VLPO (for statistics, see Table 3). The regions showing increased c-Fos labeling included the four wake/arousal-promoting nuclei studied: the SI-MBN, PeF-LH, tuberomammillary nucleus, locus coeruleus, and the limbic and neuroendocrine nuclei examined: the BSTLD, anteroventral periventricular and paraventricular nuclei, and medial amygdala. In contrast, in the VLPO, SD resulted in a lower number of c-Fos-ir cells compared with control animals (Table 3, Fig. 2A).
A significant main effect of E2 treatment on c-Fos immunoreactivity was found in three of the nine regions examined, including the BSTLD, the tuberomammillary nucleus, and the orexin field of the PeF-LH (see below; Table 3). In the BSTLD (Figs. 2B and 4B), the number of c-Fos-ir cells was higher in both Low E2 and High E2 groups compared with the Oil group (P < 0.05 vs. Oil), with no significant difference between Low E2 and High E2. Similarly, c-Fos labeling in the TMN was higher in Low E2 compared with Oil (P < 0.05), with a similar trend in High E2 vs. Oil (P = 0.053).
A significant interaction effect between E2 treatment and sleep condition was found in both the TMN and VLPO, suggesting that the effects of SD on c-Fos labeling in these two areas were modulated by hormonal treatment. In the tuberomammillary nucleus, the increase in c-Fos labeling following SD was enhanced by E2 treatment (Table 3). As shown in Fig. 5A, SD caused a nonsignificant change (+76% relative to controls) in c-Fos immunoreactivity in Oil-treated rats, whereas in Low E2 and High E2 rats, SD increased c-Fos labeling significantly by 230% and 530%, respectively, relative to controls (P < 0.05 for both comparisons). The number of c-Fos-ir cells in sleep-deprived rats was greater in Low E2 than in Oil animals (+104%; P < 0.05; Figs. 2C and 5A), and a similar trend was found for High E2 vs. Oil (+70%; P = 0.053), with no significant difference between Low E2 and High E2 animals.
In the VLPO, the decrease in c-Fos labeling following SD appeared to be modulated by E2 treatment. As shown in Fig. 5B, there was a nonsignificant decrease in the number of c-Fos-ir cells after SD in the Oil-treated group (67% decrease compared with nondeprived controls); the lack of statistical significance is likely due to the small number of animals used (n = 2–3 per group) and the large within-group variability. The decrease of c-Fos immunoreactivity after SD was significant in Low E2 (70% decrease compared with corresponding controls; P < 0.05; Fig. 5B), whereas in High E2, SD did not affect c-Fos labeling. This latter result may be because the number of c-Fos-ir cells in the control animals was substantially lower (by 66%) in the controls for the High E2 than in Low E2 groups (P < 0.05).
For c-Fos labeling in the PeF-LH area, we used additional orexin immunolabeling to identify neurochemical phenotypes of c-Fos-ir neurons. Fig. 2D shows examples of c-Fos immunoreactivity in orexin-ir and nonorexin-ir neurons in the PeF-LH in a sleep-deprived, high E2-treated rat. Double-labeling results indicated that c-Fos immunoreactivity was present in both orexin-ir and nonorexin-ir neurons in both sleep-deprived and nondeprived animals. As shown in Table 3 and Fig. 6A, the number of orexin-ir neurons that were c-Fos-ir in sleep-deprived animals was higher than in nondeprived controls by 620%, irrespective of hormonal treatment. However, the number of double-labeled neurons increased with E2 treatment regardless of sleep condition, with higher numbers in High E2 than in Oil or Low E2 (P < 0.05 vs. High E2), and no significant differences between Low E2 and Oil (Fig. 6B). In general, orexin-ir neurons that were c-Fos-ir were more abundant in the medial than the lateral part of the orexin field.
The total number of orexin-ir neurons was, not surprisingly, unaffected by sleep conditions. It did, however, vary across E2 treatment groups for unknown reasons (Table 3), with a significantly lower number in Low E2 (means ± SE, 139 ± 7) relative to both Oil (176 ± 13) and High E2 groups (193 ± 7; both P < 0.05 vs. Low E2). We, therefore, calculated the number of orexin-ir neurons expressing c-Fos as a percentage of the total number of orexin-ir neurons in each animal. Regardless of E2 treatment conditions, SD increased the percentage of double-labeled neurons relative to their nondeprived controls (Fig. 6C). The proportion of orexin-ir cells showing c-Fos-ir was also affected significantly by E2 treatment condition (Fig. 6D), with larger percentages in High E2 compared with Oil (P < 0.05) and no significant differences between Low E2 and either Oil or High E2.
Similar to orexin-ir neurons, nonorexin-ir neurons showed more (+268%) c-Fos labeling in sleep-deprived animals than in nondeprived controls (Fig. 6E), and this increase was not affected by E2 treatment (Table 3). Unlike for orexin-ir neurons, however, the number of nonorexin-ir neurons that were c-Fos-positive was not affected significantly by hormonal treatment (Fig. 6F).
Correlations between c-Fos immunoreactivity and endocrine measures.
Correlations were evaluated between the number of c-Fos-ir cells in each brain region studied and each of the following variables: serum E2 levels, uterus weights, and body weight gain, in both sleep-deprived and nondeprived groups, using either Pearson's or Spearman's correlation coefficients. In the PeF-LH, correlations were also calculated specifically for orexin-ir neurons.
No significant correlations were found between c-Fos-ir cell numbers and these variables in any of the brain regions studied in nondeprived controls. In the sleep-deprived group, however, moderate, but statistically significant, correlations were found in three of the nine brain regions examined; namely, the PeF-LH (orexin neurons), TMN and BSTLD (Fig. 7). Specifically, for orexin-ir neurons, there were positive correlations between the number of c-Fos-ir neurons and both serum E2 levels (ρ = 0.48, P < 0.05; n = 18) and normalized uterus weights (r = 0.68; P < 0.01; n = 14; Fig. 7A). For the tuberomammillary nucleus and the BSTLD, the number of c-Fos-ir cells was positively correlated with normalized uterus weights (respectively, r = 0.61; P < 0.025; and r = 0.71; P < 0.01; n = 14 for each region; Fig. 7, B and C), but not with serum E2 levels. In none of these three brain regions were there significant correlations between c-Fos labeling and body weight gain.
We found that 6 h of SD during the early light phase produced an increase in the number of c-Fos-ir neurons in nearly all brain regions studied in ovariectomized rats, regardless of E2 treatment. These regions included wake/arousal-promoting nuclei, namely, the SI-MBN, orexin-containing PeF-LH, tuberomammillary nucleus, and locus coeruleus, as well as a number of limbic and neuroendocrine nuclei rich in ER, including the BSTLD, anteroventral periventricular nucleus, and paraventricular hypothalamic nucleus, and medial amygdala. The c-Fos responses to SD in several of these regions were modulated by circulating E2 dose dependently. Specifically, the numbers of c-Fos-ir cells after SD in the BSTLD, tuberomammillary nucleus, and orexin-containing PeF-LH were positively correlated with E2-induced increases in uterus weight (normalized to body weight). Interestingly, in the VLPO, an area implicated in the regulation of sleep, SD decreased c-Fos immunoreactivity when E2 levels were low or intermediate; at higher E2 levels, basal c-Fos was decreased. These results show that SD generally had more robust and broader effects on c-Fos immunoreactivity than did E2 treatments and that the effects of E2 treatments were selective to enhancing c-Fos responses to SD dose-dependently in the orexin-containing PeF-LH, tuberomammillary nucleus, and BSTLD, while decreasing c-Fos labeling in the VLPO in the nondeprived and high E2 condition.
Endocrine responses to E2 treatments.
The E2 replacement procedures used in this study produced a dose-dependent increase in serum E2 levels at the time of SD experiments, with median levels ranging from ≤5 to 9 to 16 pg/ml in rats in the Oil, Low E2, and High E2 groups, respectively. These median serum E2 levels in both E2-treated groups were lower than expected and, in fact, some animals in both Low and High E2 groups had undetectable E2 levels. It is possible that the capsules in these animals failed to produce a steady release of E2 at the intended levels. When the values below the detectable level were excluded, the ranges of detectable serum E2 levels in the Oil, Low E2, and High E2 groups were 6–9, 9–25, and 8–64 pg/ml, and the latter two correspond to the range of levels detected during diestrus (7–20 pg/ml on average) and proestrus (up to 50 pg/ml) of intact female rats (29, 73).
Despite some variability, the low E2 treatment was effective in increasing uterine weights and in attenuating body weight gains, which are well-documented effects of E2 (25, 38, 81). An additional dose of E2 24 h before perfusion in the High E2 group caused a further increase in uterine weight, but no further change in body weight gain, presumably because of the limited time available.
Behavioral responses to E2 treatments.
Behavioral observations of nondeprived animals indicated that E2 administration had no effect on sleep and wake states during a 2-h observation period in the light phase: all animals appeared to be asleep most of the time. Previous studies reported that E2 replacement increased wakefulness and decreased sleep (especially rapid eye movement sleep) most robustly during the dark phase (8, 46, 57). The brief observation period during the light phase in this study may not have been adequate to reveal any behavioral effects. Studies using longer recordings of sleep/wake states (14, 15) are needed to fully evaluate the behavioral effect of E2 replacement. The absence of any behavioral differences among groups during this interval, however, makes it less likely that differences in c-Fos levels are related to differences in behavior, rather than in hormonal status.
Behavioral responses to SD.
All rats in the sleep-deprived condition appeared to be awake during the entire deprivation period. E2 replacement did not have a significant effect on the number of interventions required during the last hour of deprivation to maintain wakefulness (roughly one every 2 min). This finding suggests that E2 treatment probably did not affect the buildup of sleep drive/pressure during deprivation under the current experimental design.
SD-induced increase in c-Fos immunoreactivity.
Independent of hormonal treatment condition, 6 h of SD strongly increased c-Fos labeling in 8/9 brain regions examined, suggesting that neurons in these areas were activated by forced wakefulness. These regions included wake/arousal-promoting nuclei, as well as limbic and neuroendocrine nuclei rich in ER. Increases in c-fos mRNA in the hypothalamus and pons after 6 h of SD have been reported in intact female rats, whose stages of the estrous cycle were not documented (55). The pattern of SD-induced c-Fos immunoreactivity in the brain regions examined in the present study is similar to that reported in male rats (6, 68).
We also found that both orexin and nonorexin neurons in the PeF-LH showed increased c-Fos labeling after SD, consistent with previous findings in male rats (20, 21, 44, 50). In agreement with these findings, electrophysiological and microdialysis evidence indicates that orexin neurons become more active during both spontaneous and forced wakefulness (41, 48, 75, 82), and selective stimulation of orexin neurons causes EEG activation and increased wakefulness (1).
SD procedures in general, even those involving only gentle handling, are intrinsically stressful. It is, therefore, necessary to consider the possibility that stress contributed to the observed c-Fos labeling, particularly in the paraventricular hypothalamic nucleus, medial amygdala, and locus coeruleus, which are known to express c-Fos in response to stressful stimuli (reviewed in Ref. 2). The levels of stress during SD, often assessed by measuring circulating levels of adrenal stress hormones, were not examined in the present study. One way to control for potential stress confound would be to include animals that are subjected to the same number of interventions during the dark phase, when animals are spontaneously awake most of the time. The use of procedures known to be associated with high levels of stress (e.g., restraint) might also be useful for comparison. Without these additional control procedures, we cannot exclude the possibility that some changes in c-Fos after SD in the present study may reflect mild stress associated with the manipulation, in addition to effects of sleep loss. Nonetheless, to the degree to which stress responses contributed to c-Fos increases in the aforementioned nuclei, it appears that these were not affected by E2 levels.
SD-induced c-Fos immunoreactivity in some brain regions is modulated by E2 treatments.
The E2 treatments in this study increased c-Fos-positive labeling in a subset of brain areas studied: the BSTLD, tuberomammillary nucleus, and orexin-containing PeF-LH. E2's modulatory effect on c-Fos labeling in these three areas was positively correlated with E2-induced changes in (normalized) uterus weight after SD, but not under nondeprived conditions. Other brain areas showed an increase in c-Fos labeling in response to SD, but not to E2 administration. These areas included the anteroventral periventricular nucleus, paraventricular hypothalamic nucleus, SI-MBN, medial amygdala, and locus coeruleus.
These results contrast with some findings in previous studies using E2 replacement in ovariectomized rats. An increase in basal (i.e., unstimulated) c-Fos labeling was found in the tuberomammillary nucleus during the early light phase at 24 h after two subcutaneous injections of 5 μg and then 10 μg of E2 per animal (27), and in the anteroventral periventricular nucleus 48 h after a single injection of the same dose of E2 as used in the present study (10 μg/kg; Ref. 34; see also Ref. 59). Conversely, swimming stress-induced c-Fos immunoreactivity (but not basal level) in the paraventricular hypothalamic nucleus and medial amygdala (plus several other forebrain regions) was reduced following daily subcutaneous injections of 10 μg of E2 per animal for 1 wk (63); serum E2 levels were not measured. Reductions in c-Fos labeling by E2 treatments were also found in ovariectomized rats subjected to restraint stress (9, 24, 77). Differences between the results of previous studies and the present study may be attributable to a number of factors: the dose of E2 (physiological vs. supraphysiological), the pattern of E2 administration (acute vs. chronic), the time interval between E2 treatment and c-Fos immunohistochemistry, and the behavioral manipulations used.
There are a number of mechanisms by which E2 replacement and SD could interact to enhance c-Fos immunoreactivity. E2 may enhance c-Fos synthesis by directly activating nuclear ERα, which can increase the transcription rate of the c-fos gene (23, 33, 47, 56, 79). ERα (and β) is present in variable densities in the BSTLD, tuberomammillary nucleus, and PeF-LH (22, 70, 71), which are the areas that showed enhanced SD-induced c-Fos labeling following E2 replacement. Membrane receptor-initiated actions of E2, such as the rapid activation of kinases and calcium-regulated element binding protein, could also increase c-Fos transcription (78). It is, however, difficult to reconcile these actions of E2 with the absence of E2 effects on basal c-Fos levels.
An alternative mechanism may involve an increase in glutamatergic neurotransmission to SD-responsive neurons. Previous studies have shown that 4 h of SD by gentle handling are sufficient to induce long-term potentiation of glutamatergic synapses onto orexin neurons in male mice (64) and that activation of ionotropic glutamate receptors is a potent inducer of c-Fos expression in the brain (30). E2 treatment has also been shown to increase hypothalamic levels of NMDA-NR2D receptor mRNA (80) and of AMPA-GluR1 and GluR2/3 receptor proteins (16) in ovariectomized rats. Colocalization of these glutamate receptors and ERα mRNAs has been found in single hypothalamic neurons in ovariectomized mice (39) and intact female rats (16), respectively. Recently, activation of membrane-associated ER by E2 has been shown to enhance glutamate release in the hypothalamus of female rat pups (67). Taken together, these findings suggest that SD and E2 replacement may have an additive (and possibly synergistic) stimulatory effect on glutamatergic neurotransmission. Thus, the lack of E2 effect at baseline (i.e., the midpoint of the rest phase) may simply reflect the absence of excitatory drive onto SD-responsive neurons, whereas increased glutamatergic synaptic activity as a result of SD could be further potentiated by E2, thereby leading to greater increases in c-Fos labeling.
This explanation, however, cannot account for the lack of E2 effects on c-Fos immunoreactivity in the other brain areas (anteroventral periventricular nucleus, SI-MBN, paraventricular nucleus, medial amygdala, and locus coeruleus), in which c-Fos labeling increased in response to SD but not to E2 administration. These areas display variable densities of ERα and/or β (4, 70). The absence of E2 effects in these regions may be due to a ceiling effect if SD alone induced maximal expression of c-Fos protein. Additional studies are needed to clarify the mechanisms by which E2 replacement can affect cellular responses to SD in specific brain regions.
c-Fos labeling in the VLPO.
In contrast to all other regions studied, SD decreased c-Fos immunoreactivity in the sleep-promoting VLPO, as previously observed in male rats (26, 49, 59, 69). Under nondeprived conditions, transiently high levels of E2 in the High E2 group decreased baseline c-Fos labeling during the light phase when animals were mostly asleep. These low basal c-Fos levels after the high E2 regimen may have precluded the detection of any further decrease after SD. In any case, the inhibition of basal c-Fos labeling in the VLPO by high levels of E2 is consistent with a recent study by Hadjimarkou et al. (27), who found a reduction in basal c-Fos labeling in this nucleus during the light phase at 24 h after subcutaneous injections of 5 μg and then 10 μg of E2 per animal. In contrast, however, Peterfi et al. (59) found an increase in basal c-Fos labeling in the same nucleus during the light phase in intact female rats in estrus or diestrus, compared with ovariectomized rats, as well as at 24 h following a single intraperitoneal injection of a very high dose of E2 (1,000 μg/kg). The relations between circulating E2 levels and c-Fos activation in the VLPO appear to be complex, but these results, at least, indicate that E2 can act at the sleep-promoting VLPO to modulate sleep/wake states.
Both direct and indirect mechanisms might mediate the decrease in basal c-Fos immunoreactivity in the VLPO observed in the present experiment after a high dose of E2. E2 may reduce c-Fos synthesis through the direct activation of ER β in VLPO neurons. The VLPO of ovariectomized rats contains low levels of ERβ mRNA (70, 71), and ERβ activation can reduce the transcription rate of the c-fos gene in cell lines (47). Alternatively, high E2 treatment may indirectly reduce VLPO activation by inhibiting lipocalin-type prostaglandin D2 synthase (L-PGDS), an enzyme that catalyzes the production of prostaglandin D2, a possible sleep-promoting factor. Prostaglandin D2 has been suggested to activate VLPO neurons by increasing adenosine release in the VLPO (32). E2 profoundly depressed L-PGDS mRNA expression in the VLPO 24 h after treatment (27, 52, 53).
Perspectives and Significance
We found that physiological E2 replacements following ovariectomy increased SD-induced, but not basal, levels of c-Fos immunoreactivity in PeF-LH orexin neurons and in the tuberomammillary nucleus and BSTLD. In contrast, E2 treatments decreased c-Fos labeling in the VLPO only under nondeprived conditions. These results suggest that the levels of activity in PeF-LH orexin neurons and in the tuberomammillary nucleus, VLPO, and BSTLD change depending on the serum levels of E2, and that the modulatory effect of E2 replacement varies depending on sleep/wake behavior and brain regions. Whether and how E2-dependent changes in the pattern of c-Fos labeling are related to the behavioral and EEG responses to SD are of particular importance, and we are currently investigating this question. A better understanding of the effects of E2 replacement on neuronal activity in the sleep/wake system and on the regulation of sleep/wake states in ovariectomized rats may have implications for a better understanding of the consequences of sleep loss in postmenopausal women who use hormone replacement therapy.
This work was supported by a Canadian Institutes of Health Research grant (MOP-67085). EMC was a recipient of a 2006 Heinish IWK Summer Student Award.
We thank Joan Burns for assistance in ovariectomy, Diane Wilkinson for conducting estradiol RIA, and Jessica Wilson for cell counting in the locus coeruleus. We also thank Drs. Michael Wilkinson and Jennifer Stamp for helpful discussions during the course of this experiment.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society