Sleep deprivation (SD) can suppress cell proliferation in the hippocampal dentate gyrus of adult male rodents, suggesting that sleep may contribute to hippocampal functions by promoting neurogenesis. However, suppression of cell proliferation in rats by the platform-over-water SD method has been attributed to elevated corticosterone (Cort), a potent inhibitor of cell proliferation and nonspecific correlate of this procedure. We report here results that do not support this conclusion. Intact and adrenalectomized (ADX) male rats were subjected to a 96-h SD using multiple- and single-platform methods. New cells were identified by immunoreactivity for 5-bromo-2′-deoxyuridine (BrdU) or Ki67 and new neurons by immunoreactivity for BrdU and doublecortin. EEG recordings confirmed a 95% deprivation of rapid eye movement (REM) sleep and a 40% decrease of non-REM sleep. Cell proliferation in the dentate gyrus was suppressed by up to 50% in sleep-deprived rats relative to apparatus control or home cage control rats. This effect was also observed in ADX rats receiving continuous low-dose Cort replacement via subcutaneous minipumps but not in ADX rats receiving Cort replacement via drinking water. In these latter rats, Cort intake via water was reduced by 60% during SD; upregulation of cell proliferation by reduced Cort intake may obscure inhibitory effects of sleep loss on cell proliferation. SD had no effect on the percentage of new cells expressing a neuronal phenotype. These results demonstrate that the Cort replacement method is critical for detecting an effect of SD on cell proliferation and support a significant role for sleep in adult neurogenesis.
- cell proliferation
- rapid eye movement sleep
the functions of mammalian sleep and adult hippocampal neurogenesis remain unresolved. Both have been implicated in neuroplasticity and are disrupted by factors, such as stress, depression, and aging, that can impair hippocampus-dependent learning and memory (7, 9, 26, 30, 43). These associations invite a hypothesis that sleep and hippocampal neurogenesis may be causally linked, such that sleep-dependent processes either directly or indirectly promote the production of new neurons in the hippocampus. If so, then proliferation, survival, or maturation of new hippocampal neurons should be suppressed by deprivation of sleep or its component stages. For such a result to speak to the function of sleep, it must be shown to be independent of nonspecific consequences of sleep deprivation procedures, such as elevated stress hormones, social isolation, or activity restriction, any of which could suppress neurogenesis (19, 30, 44, 51). If sleep loss does not affect neurogenesis independent of nonspecific factors, then the hypothesis is falsified.
Recent studies show that cell proliferation or maturation in the adult rodent hippocampus is not inhibited by sleep deprivation procedures lasting 24 h or less (30, 40, 49) but is significantly suppressed by sleep deprivation or fragmentation procedures lasting 3 days or more (11, 14, 15, 17, 47). These procedures are particularly disruptive of rapid eye movement sleep (REMS), and REMS deprivation (RSD), with only partial reduction of non-rapid eye movement sleep (NREMS), can inhibit cell proliferation with equivalent potency (12, 31). However, the effects of RSD have been attributed to nonspecific stress, based on the observation that 72-h RSD, by the classic single platform-over-water method, significantly suppresses cell proliferation in intact rats but not in adrenalectomized (ADX) rats provided low-dose corticosterone (Cort) replacement in drinking water (31).
While this latter result appears to refute a hypothesis ascribing to normal sleep (particularly REMS) a role in the production of new brain cells, the sleep deprivation, Cort replacement, and blood sampling methods used in that study raise interpretive issues. In nocturnal rats, 70–80% of daily water intake occurs at night. Consequently, ADX rats provided Cort replacement via drinking water should ingest three to four times more Cort at night than during the day (21), and a single daytime blood sample, as in the Mirescu et al. (31) study, should fail to detect this and should provide values of Cort that are lower than samples taken during the night, because of the short half-life of blood Cort (42). Furthermore, during RSD on a platform over water, the daily rhythm of water intake is strongly attenuated (45) and rats may drink from the pool, thus reducing Cort intake relative to home cage control rats. Reduced Cort intake in RSD rats, sustained over 72 h, may increase cell proliferation sufficiently to offset an inhibitory effect of RSD, thereby eliminating group differences that would otherwise be apparent between control and sleep-deprived ADX rats.
We show here that 96 h of RSD, using either the classic single platform-over-water method or, to avoid social isolation and activity restriction, the multiple-platform method (27, 46, 50), significantly suppresses cell proliferation in intact rats and in ADX rats with low-dose Cort replacement provided by subcutaneous minipumps. We also confirm the finding of Mirescu et al. (31) that suppression of cell proliferation by RSD, relative to home cage controls, is absent in ADX rats provided Cort replacement via drinking water, but further show that this is likely an artifact of the Cort replacement method, because ADX rats in the control condition ingested nearly threefold more Cort per day than ADX rats during RSD. In this article, we refer to platform sleep deprivation as RSD because REMS was much more severely reduced (∼95%) than NREMS (∼40%), but a role for total sleep time, sleep continuity, or NREMS in the antineurogenic effects of this procedure is not ruled out.
MATERIALS AND METHODS
A summary of experiments and groups in this study is provided in Table 1. All experiments and procedures were approved by the Simon Fraser University Animal Care Committee.
Experiment 1. Effect of 96-h RSD via Multiple-Platform Method on Cell Proliferation and Cort in Intact Rats
In the classic single-platform (inverted flowerpot) RSD method used by Mirescu et al. (31), rats are individually confined to a water-filled chamber containing a small platform 1 cm above water level. The characteristic loss of muscle tone during REMS causes the rat to awaken or else dip into the water. NREMS is at least partly spared because it does not require muscle atonia. This method is known to be stressful and also entails social isolation and restriction of activity (19, 44, 51). To avoid these potential confounds, we first used the multiple-platform method to examine the effect of 96-h RSD on cell proliferation. By this method, rats are maintained in stable social groups of six individuals and are confined to a large chamber containing water and 20 small platforms that permit locomotion and NREMS but are too small to permit REMS.
Subjects and apparatus.
Male Long-Evans rats [300–400 g, n = 42; Charles River (Quebec)] were housed in Plexiglas cages in groups of six social littermates. A 12:12-h light-dark (LD) cycle and constant room temperature were maintained. Food and water were available ad libitum. All rats were handled each day for 1 wk before the start of experiments. The rats were then randomly assigned to the experimental conditions and weighed daily.
The multiple-platform apparatus consisted of a large pool (45 cm tall, 150-cm diameter) bisected with a stainless steel grid and filled to a depth of 30 cm with 23°C water. Half of the pool contained 20 inverted flowerpot platforms, separated by at least 8 cm and distributed over a total area of 3.56 m2. The top of each platform was 6.5 cm in diameter and 1 cm above the water. The other half of the pool served as an apparatus control (AC) environment. A wire mesh floor was placed over the platforms 1 cm over the water, thereby exposing the rats to similar environmental conditions but permitting both NREMS and REMS. Overhead infrared motion sensors were used to measure locomotor activity in each group. Food and water were provided by overhead food dispensers and water bottles. The rats were weighed each day at lights-on, during which the pool water was replaced.
Experiment 1a. EEG validation.
The multiple-platform RSD method has been validated polysomnographically but not so extensively as the single-platform method (27, 46, 50). Therefore, to confirm that multiple platforms of our configuration prevent REMS, and to characterize the effects of the procedure on daily rhythms of wake and NREMS, a group of six rats were anesthetized with isoflurane for stereotaxic placement of EEG and EMG electrodes (Plastics One). EEG electrodes were stainless steel jeweler's screws threaded into holes drilled through the skull at the following stereotaxic coordinates, in millimeters relative to bregma: parietal cortex, 1.0 posterior, 2.5 lateral; hippocampus, 4.3 posterior, 2.0 lateral, 2.3 ventral; frontal cortex, 3.0 anterior, 2.0 lateral; occipital cortex, 1.0 anterior to lambda, 3.5 lateral. Slow wave (1–3 Hz) and spindle (10–15 Hz) EEG activity were obtained by fast Fourier transforms of differential recordings from the electrodes positioned over the parietal and occipital cortexes. Hippocampal theta rhythms (5–9 Hz) were obtained by differential recordings from two electrodes positioned over the dorsal hippocampus and, as a reference, the frontal cortex. EMG was recorded from two subcutaneous wire electrodes implanted between the occipital bone and the neck muscle. A screw was also placed on the skull above the cerebellum to provide additional anchor support. The pins from all six electrodes were connected to a protective plastic head cap. The entire electrode assembly was then insulated and bonded to the skull with dental acrylic.
After a week of recovery from surgery, the rats were handled 15 min/day for an additional week, during which test recordings were made to habituate the animals to the recording apparatus. Recordings were conducted in Plexiglas chambers (37 × 35 × 53 cm) within electrically shielded and sound-attenuating enclosures (model E3125AA-3 Animal Chest, Grason-Stadler, West Concord, MA). Recording cables were connected to a commutator (Plastics One) to allow free movement of the animal. An overhead passive infrared motion detector (model 49-426, RadioShack) was used to measure locomotor activity. After this adaptation period, the rats were housed in the recording chambers for 1 baseline day, 4 days of experimental conditions, and 1 recovery sleep day. On each recording day, the rats were removed from their recording chambers for 30 min at zeitgeber time 2 (ZT2, i.e., 2 h after light onset, where light onset is ZT0 by convention) for food, water, and bedding change. To accommodate the recording cables, the rats had to be tested individually in a smaller three-platform water chamber. Rats normally would not rest on more than two platforms at a time; thus the three-platform version was considered suitable for confirming RSD electrophysiologically. Two rats were subjected to 96-h RSD, two served as ACs (platforms were covered by a wire mesh floor, permitting REMS), and two served as home cage controls (CCs, regular cages with bedding). These rats were not used for cell proliferation or Cort measurements.
Electrophysiological signals were amplified and band-pass filtered (0.3–35 and 30–300 Hz for EEG and EMG, respectively) by a polysomnograph (Grass model 9, Grass Instruments) and then digitized (sampling rate of 250 Hz) and stored on a computer using SleepSign data acquisition software (Kissei Comptec) for off-line analysis. Behavioral states were scored according to previously published criteria in 10-s epochs, with each epoch classified as whatever state was predominant (34).
Experiment 1b. Cell proliferation and Cort.
Three social groups of six rats received 1-h habituation sessions in the multiple-platform apparatus 4–7 days before the experiment. The groups were then randomly assigned to RSD, AC, and CC conditions, to be run in parallel. The procedures began at ZT2 and lasted 96 h. All rats were weighed each day at ZT2, when pool water was flushed and replaced. At hour 94, all rats received an intraperitoneal injection of 5-bromo-2′-deoxyuridine (BrdU, 200 mg/kg), an analog of the nucleoside thymidine that is incorporated into DNA during the S phase of the cell cycle. This dose was chosen based on past studies (4, 31). Two hours after BrdU administration, the rats were overdosed with pentobarbital sodium. Cardiac blood samples were collected, and the adrenal glands were extracted and weighed. The rats were then perfused transcardially with phosphate-buffered saline (PBS) and 4% paraformaldehyde. The brains were extracted and stored at 4°C for 24 h in 4% paraformaldehyde followed by an additional incubation, for cryoprotection, in a 15% sucrose-PBS solution until equilibration.
Experiment 1c. Cell phenotyping and Cort.
Three additional groups of six rats were assigned to 96-h RSD, AC, and CC procedures, with BrdU injections at hour 94 as in experiment 1b. To examine whether RSD alters the proportion of new cells developing a neuronal phenotype, these rats were allowed to survive for an additional 96 h of recovery sleep in the home cages, which is the amount of time required for doublecortin (DCX), an endogenous marker of immature neurons, to be expressed maximally (3). The rats were then euthanized and perfused, and brains were extracted for immunocytochemical processing as described above. The proportion of BrdU-immunoreactive (+) cells double labeled with DCX was determined as described below.
Experiment 2. Effect of 96-h RSD via Multiple-Platform Method on Cell Proliferation in ADX Rats With Low-Dose Cort Minipumps
The results of experiments 1b and 1c showed that, by contrast with the control procedures, RSD via the multiple-platform method suppressed hippocampal cell proliferation and elevated serum Cort. To determine whether suppression of proliferation is mediated by stress-induced Cort release, 18 additional Long-Evans rats [300–400 g, Charles River (Quebec)] were anesthetized with isoflurane and adrenalectomized bilaterally. Osmotic minipumps (model 2ML2, Alzet) for constant low-dose Cort replacement were implanted subcutaneously. Cort was dissolved in polyethylene glycol 400 at a concentration providing 40 μg·kg−1·h−1, which has been shown to produce blood plasma levels at 3, 7, and 14 days comparable to intact, unstressed rats (52). After a week recovery, the rats were assigned to the 96-h RSD, AC, or CC conditions as in experiment 1. To maintain sodium:potassium balance, the drinking bottles and pool were filled with a 0.9% saline solution. All rats received an intraperitoneal BrdU injection at hour 94 and were euthanized at hour 96 for perfusions and cardiac blood sampling. The brains were harvested and processed for BrdU immunocytochemistry, and serum was frozen for Cort assays.
Experiment 3. Effect of RSD via Single-Platform Method on Cort Intake in ADX Rats Provided Replacement Cort in Drinking Water
Experiment 2 showed that RSD inhibits cell proliferation in ADX rats with Cort clamped via minipumps. At this point in our study, Mirescu et al. (31) reported that 72-h RSD, by the single-platform method, inhibited cell proliferation in intact rats but not ADX rats receiving Cort replacement via drinking water. To determine whether the Cort replacement method used in that study might result in reduced Cort intake in RSD rats relative to CC rats, thereby offsetting an inhibitory effect of RSD on proliferation, we measured total daily Cort intake in ADX rats with the methods of Mirescu et al. (31).
The rat strain and methods of Cort replacement and RSD (including start time, duration, and platform method) were based on the study of Mirescu et al. (31). Accordingly, male Sprague-Dawley rats [n = 5, 250–300 g, Charles River (Quebec)] were adrenalectomized bilaterally and provided low-dose Cort replacement via drinking water (25 μg/ml in 0.9% NaCl). After a week recovery, Cort water intake was measured twice daily, at lights-on and lights-off (LD 12:12), for 4 days. The rats were then subjected to 98-h RSD with the single-platform method (Supplemental Fig. S1, e–g). At ZT7 (7 h after lights-on), individual rats were placed on a 6.5-cm-diameter platform 1 cm above the water in a circular container (28.5 cm high, 29-cm diameter, 9 cm of water). Food and drinking water containing Cort were available in an overhead dispenser and bottle, respectively. Water intake was measured at lights-on and lights-off each day. Rats were weighed and water changed at lights-on. It should be noted that, as in the study of Mirescu et al. (31), there was no AC condition.
Experiment 4. Effect of RSD via Single-Platform Method on Cell Proliferation in ADX Rats: Comparison of Cort Replacement Methods
Experiment 3 confirmed, using a within-subjects design, that ADX rats provided Cort via drinking water ingest nearly threefold more Cort in the CC condition than during RSD on a single platform. We next examined whether suppression of cell proliferation in ADX rats is dependent on the Cort replacement method. To ensure that suppression of cell proliferation as measured by incorporation of exogenous BrdU into DNA during S phase was not secondary to changes in bioavailability of BrdU or duration of S phase, the brain tissue was also processed for Ki67, an endogenous marker for dividing cells that is expressed at all phases of the cell cycle except G0 and early G1 following G0.
Male Sprague-Dawley rats [n = 24, 250–300 g, Charles River (Quebec)] were adrenalectomized bilaterally. Half of the rats received Cort replacement via subcutaneous minipumps and half via drinking water. In each of these groups of 12 rats, 6 served as home cage controls and 6 were subjected to 98-h RSD by the-single platform method, beginning at ZT7 as in the study of Mirescu et al. (31). At hour 96, all rats received an injection of BrdU. At hour 98, all rats were euthanized and perfused, and the brains were harvested for immunocytochemical processing.
Experiment 5. Dose-Response Effects of Cort in Low Physiological Range on Cell Proliferation and Comparison With RSD
When data from experiments 2 and 4 were pooled, ADX rats with low-dose Cort replacement via minipump exhibited a weak trend for increased serum Cort after RSD, relative to control rats. Group means were in the low physiological range and differed by ∼2.5 μg/dl in both experiments. To determine whether this small, unexpected difference may have contributed to group differences in cell proliferation, a two-point Cort dose-response experiment was conducted.
Adrenalectomized male Sprague-Dawley rats (n = 21, 250–300 g) were purchased from Charles River (Quebec). The rats were received 4 days after adrenalectomy and were immediately placed on low-dose Cort replacement via drinking water (25 μg/ml) for 1 wk, during which body weight and fluid intake were measured daily and the rats habituated to handling. All of the rats were then anesthetized with isoflurane, and osmotic minipumps containing Cort (40 μg·kg−1·h−1) were implanted subcutaneously. At ZT7 on day 3 after surgery, blood samples were taken by tail bleeding. Three days after that, the rats were assigned to single-dose, double-dose, or RSD groups (n = 7 each) and anesthetized with isoflurane. Rats in the double-dose Cort group received new minipumps (80 μg·kg−1·h−1). Rats in the single-dose Cort group received sham surgery, and rats in the RSD group received anesthesia only. At ZT7, the RSD rats were placed in single-platform chambers and subjected to 98-h RSD as above. The single- and double-dose groups remained in their home cages. At hour 96, all rats received an intraperitoneal injection of BrdU (200 mg/kg). At hour 98, tail blood was collected rapidly, and the rats were euthanized and perfused for harvesting of brains.
All blood samples were stored overnight at 4°C and centrifuged at 3,200 rpm for 10 min, and the serum was extracted and stored at −20°C until radioimmunoassay. Serum Cort concentrations were assayed as previously described (25) with a commercial kit according to the manufacturer's instructions (MP Biomedicals, Orangeburg, NY). The antiserum cross-reacts 100% for Cort, 2.3% for deoxycorticosterone, 0.47% for testosterone, 0.17% for progesterone, and 0.05% for aldosterone. The minimum detectable Cort concentration was 0.7 μg/dl, and the intra- and interassay coefficients of variation were <7.1% and <6.1%, respectively. Antiserum was obtained from MP Biomedicals, tritiated Cort tracer from Mandel Scientific, and Cort for standards from Sigma-Aldrich. Dextran-treated charcoal (Fisher Scientific) was used to absorb free Cort after incubation.
All brains were sectioned on a cryostat at 40-μm intervals throughout the entire hippocampus. For BrdU peroxidase immunocytochemistry, tissue was incubated in the following solutions at room temperature: 0.6% H2O2 in distilled water for 30 min; 2 N HCl 37°C bath for 30 min; 0.5 M borate buffer (pH 8.5) for 10 min; 3% normal horse serum with 0.1% Triton X-100 in Tris-buffered saline (TBS) for 30 min; mouse primary antibody against BrdU (1:200; Roche) overnight; mouse IgG biotinylated secondary antibody (1:100; Vector) for 4 h; avidin-biotin horseradish peroxidase (1:50; ABC Elite Kit, Vector) for 1.5 h; 0.05% diaminobenzidine (DAB; Sigma-Aldrich) with 0.15% H2O2 for 5 min. Unless otherwise specified, sections were rinsed in TBS (3 × 10 min) between each step. The sections were then mounted onto slides, counterstained with cresyl violet, and coverslipped with Permount (Fisher Scientific).
For Ki67 immunostaining, the tissue was incubated at room temperature and rinsed 3 × 10 min in 0.1 M PBS between each of the following steps: 0.3% H2O2 for 25 min; primary antibody solution containing rabbit anti-Ki67 (1:1,000; Novocastra), 0.5% Triton X-100, and 1% normal goat serum diluted in PBS and incubated overnight; secondary antibody, goat anti-rabbit (1:1,000; Vector) for 1 h; avidin-biotin horseradish peroxidase (1:100, ABC Elite Kit; Vector) for 40 min; 0.05% DAB (Sigma-Aldrich) with 0.15% H2O2 for 10 min.
BrdU/DCX immunofluorescent double labeling.
Brains from rats permitted to survive for 96 h after BrdU injections were sectioned at 40 μm. Sections throughout the entire dentate gyrus were labeled with fluorescent antibodies to BrdU and DCX by the following steps, with tissue rinsed 3 × 10 min in PBS between each step: 3% normal donkey serum (NDS) with 0.1% Triton X-100 in PBS for 30 min; goat primary antibody against DCX (1:100; Santa Cruz) for 24 h at 4°C; 3% NDS with 0.3% Triton X-100 in PBS for 30 min. The remaining steps were performed in the dark: secondary antibody against goat Alexa 488 (1:200; Invitrogen) in 3% NDS with 0.1% Triton X-100 for 4 h; 4% paraformaldehyde for 10 min, then rinsing in 0.9% NaCl twice for 5 min, followed by 2 N HCl 37°C bath for 30 min; 3% NDS with 0.3% Triton X-100 for 30 min; primary antibody against BrdU (1:250; Oxford Biotech) in 3% NDS with 0.1% Triton X-100 for 18 h at 4°C; 3% NDS with 0.3% Triton X-100 for 30 min; secondary antibody against rat Cy3 (1:200; Jackson Immunoresearch) in 3% NDS with 0.1% Triton X-100 for 4 h. The sections were then mounted onto slides, coverslipped with 2.5% polyvinyl alcohol-1,4-diazabicyclo[2.2.2]octane (PVA-DABCO), and stored at 4°C.
For each experiment, an experimenter blind to the treatment conditions counted the number of BrdU+ cells unilaterally in every 6th section or the number of Ki67+ cells bilaterally in every 12th section, using a ×100 objective on a Nikon Elipse 600 light microscope. The number and level of hippocampal sections were equivalent across groups. Cell counts were obtained from the granule cell layer/subgranular zone and the hilus separately. Cells were considered BrdU+ or Ki67+ if they were intensely stained and exhibited normal size and a round or oval cell body. Total cell counts were estimated by multiplying by 6 the observed counts from every 6th section unilaterally or every 12th section bilaterally. The total area in which cells were counted within the dentate gyrus was measured with NIH ImageJ. The volume of the dentate gyrus was calculated with Cavalieri's principle. For cell phenotyping, 50 BrdU+ cells from at least four sections per brain were analyzed on a Leica DM 6000B fluorescence microscope and verified with a Zeiss LSM 410 confocal microscope with a ×63 objective. Double labeling for BrdU and DCX was examined throughout the z-axis in 1-μm steps. The ratio of BrdU+ cells colabeled with DCX was then calculated.
Group mean differences in total cell counts were evaluated by t-tests or ANOVA with Tukey or Bonferroni post hoc tests as appropriate. All means are reported ±SE. The cell count data were also analyzed as mean number of cells/section. These results did not differ from the total cell counts; thus to avoid redundancy and be consistent with related studies (e.g., Ref. 31) only the total cell count data are reported.
Experiment 1. RSD by Multiple-Platform Method in Intact Rats
Experiment 1a. Polysomnographic confirmation of 96-h RSD.
Rats confined to a modified version of the multiple-platform apparatus for 96 h exhibited a 95% reduction of REMS [F(2,9) = 14.47, P = 0.028], a 40% reduction of NREMS [F(2,9) = 1.40, P = 0.372; Fig. 1, A–D], and a 45% reduction of slow wave activity (1–3 Hz) in NREMS [F(2,9) = 1.60, P = 0.19]. By contrast, rats in the AC condition showed a 39% decrease in REMS and no decrease of either NREMS or slow wave activity. The daily rhythms of wake, NREMS, and locomotor activity were strongly attenuated in the RSD rats, but not in the AC rats (Fig. 1, E and F). During 24 h of recovery sleep, RSD rats relative to CC rats showed a marked rebound in total amount of REMS (+100%) but modest, nonsignificant increases of NREMS time (12%) and slow wave activity (6%; data not shown).
Experiment 1b. Cell proliferation (BrdU staining).
In 96-h RSD rats perfused 2 h after BrdU injections, the number of BrdU+ cells in the dentate gyrus (granule cell layer and subgranular zone) was reduced by 46% and 52%, respectively, by comparison with the AC and CC groups [F(2,15) = 15.89, P = 0.0001; Fig. 2, A–C, Fig. 3A]. A small number of BrdU+ cells were also observed in the hilus, representing 12% of the total in each of the three groups when combined with granule cell layer and subgranular zone cell counts. Hilus cell counts exhibited a similar decrease in the RSD group relative to the CC (−49%) and AC (−43%) groups [F(2,15) = 6.54, P = 0.009; data not shown]. There were no group differences in the volume of the dentate gyrus [F(2,15) = 1.13, P = 0.35] or the granule cell layer/subgranular zone [F(2,15) = 0.08, P = 0.92].
Experiment 1c. Cell differentiation (DCX staining).
To determine whether RSD affects the proportion of new cells developing a neuronal phenotype, the experiment was repeated and the rats were allowed 4 days of recovery sleep after BrdU administration. The brains were then stained for both BrdU and DCX (Fig. 2). The majority of BrdU+ cells in the granule cell layer and subgranular zone also expressed DCX, and this proportion did not differ between the RSD group (76 ± 8%) and either the AC (81 ± 3%) or the CC (81 ± 4%) group [F(2,15) = 0.06, P = 0.94]. In the hilus, the majority of BrdU+ cells did not express DCX, and there were again no group differences in the proportion of double-labeled cells [35 ± 17%, 39 ± 8%, and 40 ± 16% in RSD, AC, and CC groups, respectively; F(2,15) = 0.03, P = 0.97].
Because stress inhibits cell proliferation, several correlates of stress were examined. Relative to CC rats, RSD rats did not exhibit increased adrenal weights [experiment 1b: F(2,15) = 0.84, P = 0.45; experiment 1c: F2,15 = 0.38, P = 0.69] but did exhibit a fourfold increase in serum Cort sampled at hour 96 of RSD, 2 h after lights-on [experiment 1b: F(2,15) = 7.06, P = 0.007; experiment 1c: data not collected because of 4-day recovery after RSD; Fig. 3C]. RSD rats also exhibited a 5–9% decrease in body weight relative to the control groups over the 4-day procedure [experiment 1b: F(2,60) = 34.82, P < 0.0001; experiment 1c: F(2,120) = 48.15, P < 0.0001]. AC rats showed a Cort increase and a body weight decrease of approximately half this magnitude relative to CC rats.
Experiment 2. RSD by Multiple-Platform Method in ADX Rats
Because of tissue loss, the number of brains available for analysis was reduced to four, two, and two in the RSD, AC, and CC groups, respectively. Nonetheless, significant group differences were evident. The AC and CC groups exhibited similar cell counts (Fig. 3B); thus the data were pooled for statistical comparison with the RSD group. This confirmed a significant suppression of cell proliferation in RSD ADX rats by an amount comparable to the suppression evident in RSD intact rats [∼45%; F(2,5) = 19.04, P = 0.005; Fig. 3B]. Comparisons with experiment 1b (Fig. 3A) revealed a marked increase of cell proliferation in the ADX control rats (AC and CC combined = 7,638 ± 590 BrdU+ cells) compared with the intact control rats [AC and CC combined = 3,276 ± 149 BrdU+ cells; t(14) = 11.85, P < 0.0001].
Analysis of serum samples revealed slightly, albeit nonsignificantly, higher Cort levels in the RSD group relative to the AC and CC groups combined [F(2,5) = 2.41, P = 0.19; Fig. 3D]. Inspection of the distribution of Cort values revealed that the range in ADX rats (3.07–9.17 μg/dl) fell within the range exhibited in experiment 1b by intact control rats (1.44–12.71 μg/dl) sampled at the same time of day, early in the lights-on period, when Cort is at its daily minimum. Among the RSD ADX rats, the two with the highest Cort levels (7.35 and 9.17 μg/dl) had BrdU+ counts of 4,962 and 3,036 (mean = 3,999), whereas the two with the lowest Cort values (3.07 and 3.35 μg/dl) both had BrdU+ counts of 4,338. Also, two RSD ADX rats had Cort values virtually identical to those of two control ADX rats (∼3 μg/dl) yet exhibited 49% fewer BrdU+ cells. Thus, even when RSD and control ADX rats are closely matched for Cort level, BrdU+ cell counts are nonetheless much lower in the RSD rats.
Similar to intact rats, ADX rats exhibited a significant decline in body weight over the 96-h RSD [F(2,60) = 81.7, P < 0.0001].
Experiment 3: Cort Intake via Drinking Water in ADX Rats
Given the discrepancy between the cell proliferation results of experiment 2 and those of Mirescu et al. (31), we examined whether the method of Cort replacement might be a critical factor. We first tested whether providing ADX rats with Cort in drinking water might result in different Cort levels in the RSD and CC conditions. ADX rats received Cort replacement via drinking water (25 μg/ml Cort in 0.9% saline), and water intake was measured twice daily, at lights-on and lights-off, during 4 days in the home cage and during 4 days of RSD by the single-platform method, as in the study of Mirescu et al. (31). In the CC condition, water intake was 89 ± 1% nocturnal in a group of intact rats (data not shown) and 83 ± 4% nocturnal in ADX rats receiving Cort in the water. During RSD, nocturnality of fluid intake in ADX rats decreased to 62 ± 11% [1-tailed paired t(4) = 2.15, P = 0.045]. Daytime Cort intake did not differ between conditions, but nighttime intake was much greater in the CC condition, and total daily Cort intake was 2.75 times higher in the CC condition relative to the RSD condition [2.2 ± 0.25 vs. 0.8 ± 0.38 mg/day; t(4) = 7.17, P = 0.002; Fig. 4A]. Thus, as predicted, when replacement Cort is provided in drinking water there is a large difference in daily Cort intake between RSD and CC conditions. Why fluid intake from the drinking bottle is reduced in ADX rats during RSD is uncertain but likely reflects at least in part the ready availability of drinking water in the pool, freshened daily and only 1 cm below the platform.
Experiment 4. Cort Replacement Method and Cell Proliferation After RSD
To directly test whether the Cort replacement method is critical for detecting an effect of RSD on cell proliferation, ADX rats received Cort replacement either via drinking water or via minipump and were assigned to CC or 98-h RSD (single platform) groups, with deprivation beginning at ZT7 (as in Ref. 31). Cort intake measurements in ADX rats receiving Cort via drinking water confirmed the results of experiment 3. Intake was 78 ± 2% nocturnal in the CC group and 57 ± 5% nocturnal in the RSD group [1-tailed independent t(10) = 3.54, P = 0.003]. Daytime Cort intake again did not differ between groups, but, relative to RSD rats, the CC rats consumed 2.86 times more Cort at night [t(10) = 5.67, P = 0.0001] and 2.27 times more Cort overall [t(10) = 4.01, P = 0.001; Fig. 4B].
Consistent with Mirescu et al. (31), BrdU+ cell counts did not differ between the CC and RSD groups in ADX rats receiving Cort via drinking water (Fig. 5A). By contrast, BrdU+ cell counts were significantly reduced by RSD in ADX rats receiving Cort replacement via minipumps [F(3,17) = 16.38, P = 0.0001]. Cell counts in this latter group were equivalent to counts in RSD and CC rats receiving Cort via drinking water. Similar results were obtained with the endogenous protein Ki67 as a marker for new cells [F(3,19) = 20.56, P = 0.0001; Fig. 5B], indicating that the reduction of BrdU labeling was not due to an effect of RSD on the bioavailability of BrdU or on the duration of S phase of the cell cycle. Ki67+ cell counts were higher than BrdU+ cell counts, consistent with expression of Ki67 through most of the cell cycle. These results confirm both Mirescu et al. (31) and experiment 2 above and indicate that the different results across studies can be explained by the different Cort replacement methods.
Consistent with experiment 1, there were no group differences in the volume of the granule cell layer/subgranular zone [F(3,17) = 0.30, P = 0.82] or the hilus [F(3,17) = 0.73, P = 0.54]. Consistent with experiment 2, RSD ADX rats with Cort replacement via minipumps exhibited a slightly higher mean serum Cort level than CC ADX rats [3.25 ± 0.32 vs. 5.92 ± 1.50 μg/dl; t(7) = 2.43, P = 0.05].
Experiment 5. Cort Dose Response and RSD
Despite the use of a minipump to clamp Cort in the low physiological range in ADX rats, serum Cort was increased by both multiple (experiment 2)- and single (experiment 4)-platform RSD, by contrast with control groups [pooled means = 5.82 ± 0.98 vs. 2.83 ± 0.28 μg/dl, respectively; t(15) = 3.38, P = 0.004]. To determine whether a ∼3 μg/dl difference in the low physiological range might explain the inhibitory effect of RSD on cell proliferation, three groups of ADX rats received Cort replacement via minipump at the standard concentration (40 μg·kg−1·h−1). After 1 wk, minipumps in one group were replaced with a double concentration (80 μg·kg−1·h−1). BrdU was administered 96 h later with the same timeline as in experiment 4. By comparison with the standard dose, the double-dose pumps elevated serum Cort by 2.7 μg/dl on average [1-tailed t(12) = 1.47, P = 0.09; Fig. 6B]. However, the single- and double-dose groups showed no difference in BrdU+ cell counts (P = 0.98; Fig. 6A). The RSD group run in parallel exhibited a significant suppression of BrdU+ cell counts [F(2,18) = 6.42, P = 0.008] by 25% (P = 0.01) and 24% (P = 0.02) relative to the single-dose and double-dose CC groups, respectively, despite a Cort level equivalent to that of the double-dose Cort replacement group (Fig. 6B). There were no group differences in the volume of the granule cell layer/subgranular zone [F(2,18) = 0.51, P = 0.60] or the hilus [F(2,18) = 0.34, P = 0.71].
The results of this study make several substantive contributions toward clarifying the effects of sleep deprivation on adult hippocampal neurogenesis. First, the results confirm that depriving rats of REMS, with a modest reduction of NREMS, for >24 h strongly suppresses cell proliferation in the dentate gyrus and show that this is independent of rat strain (Long-Evans and Sprague-Dawley), RSD method (multiple and single platforms), and proliferation marker (BrdU and Ki67). Second, the results establish that this effect persists in ADX rats on constant, low-dose Cort replacement and thus is not secondary to stress-induced Cort release. Third, the results provide a plausible explanation for the diametrically opposite results and conclusions of another recent study that also used adrenalectomy to evaluate the role of Cort in suppression of cell proliferation by RSD (31).
RSD Can Suppress Hippocampal Cell Proliferation Independently of Cort
Inhibitory effects of stress on hippocampal cell proliferation are mediated by elevated glucocorticoids (30). Sleep deprivation procedures can be stressful (see, e.g., Refs. 17, 28, 29, 31); therefore any effects of sleep loss on cell proliferation may be secondary to elevated Cort. Consistent with this hypothesis, Mirescu et al. (31) reported that the ∼40% suppression of cell proliferation observed in intact rats after a 3-day single-platform RSD did not occur in ADX rats. However, it is likely that the RSD and control rats in that study were not matched for Cort exposure, despite adrenalectomy. Low-dose Cort replacement was provided via drinking water, and although the control and RSD rats exhibited similar blood Cort levels, this comparison was based on a single blood sample drawn 7 h after lights-on. A single daytime blood sample should greatly underestimate total daily Cort exposure, given that fluid intake, and thus Cort intake, is 70–80% nocturnal, and given that the half-life of total and free blood Cort is <30 min (42). Furthermore, if fluid and thus Cort intake are reduced at night during RSD, then there may be significant group differences in total Cort intake that cannot be detected by a single daytime blood sample. Group differences in total Cort exposure, sustained over three or four consecutive days, could obscure group differences in cell proliferation caused by sleep loss.
Our results are consistent with this interpretation and confirm that the method of Cort replacement is crucial for detecting an effect of RSD on cell proliferation. Relative to the control condition, RSD inhibited cell proliferation in ADX rats receiving Cort via minipump but not in ADX rats receiving Cort in drinking water. ADX rats receiving Cort in water exhibited an ∼60% decrease in Cort consumption during RSD relative to the control condition. We hypothesize that reduced Cort intake during RSD may upregulate cell proliferation by an amount sufficient to offset, and thereby obscure, downregulation by sleep loss.
The interpretation that sustained decreases in Cort can upregulate hippocampal cell proliferation is supported by previous results (5, 20, 33, 35, 53) and by a comparison of BrdU+ cell counts in intact control rats of experiment 1 (Fig. 3A) and ADX control rats of experiment 2 (Fig. 3B). Our conclusion that RSD can inhibit cell proliferation independently of Cort is bolstered by the results of two other recent studies (11, 12). One study included extensive pretraining on a treadmill apparatus and reported suppression of cell proliferation by RSD with no evidence of elevated Cort, albeit from a single midday blood sample (12). A second study showed that sleep fragmentation by intermittent treadmill activation inhibits cell proliferation similarly in intact and ADX rats (11). Together with the results of the present study, these data provide compelling evidence that RSD, and possibly other sleep disruptions, negatively regulates hippocampal cell proliferation independently of Cort.
A curiosity of the ADX experiments was the observation of a small increase (∼2.5–3.0 μg/dl) of Cort following RSD, despite the use of minipumps to provide constant low-dose Cort replacement. Similar Cort increases have been reported in ADX rats subjected to acute stress (10) and may reflect a decreased rate of Cort metabolism. We were able to induce a similar magnitude Cort increase with a double-dose minipump but saw no effect on cell proliferation. We can infer from this result that the 40–50% decreases in cell proliferation observed in RSD groups relative to control groups in experiments 1–3 were not due to small differences in Cort. We can also infer that for group differences in Cort to explain the apparent lack of effect of RSD on cell proliferation when Cort replacement was provided via drinking water (experiment 5), the differences must have been >2.7 μg/dl.
In addition to suppression of cell proliferation, RSD was also associated with reduced body weight gain relative to control groups. This has been consistently reported in other RSD studies using a variety of RSD procedures (24, 31, 46, 50). Where measured, weight loss has been associated with increased rather than decreased food intake, indicating increased energy expenditure. The similar levels of weight loss in our intact and ADX rats indicates that increased energy expenditure during RSD, like the suppression of cell proliferation, is not dependent on elevated Cort.
Role of NREMS
The platform RSD methods used in this study are effective but not entirely selective for REMS, as they also partially suppress NREMS and attenuate slow wave activity in NREMS. This raises the possibility that disruption of NREMS, or of interactions between REMS and NREMS, mediates or contributes to suppression of cell proliferation. A role for NREMS cannot be ruled out without selective NREMS or slow wave sleep deprivation experiments, procedures that are complicated by the difficulty in sustaining normal levels of REMS without NREMS. One argument against a critical role for NREMS rests on the minimal compensatory increase of NREMS after the RSD procedure. In our EEG validation study, RSD rats relative to control rats exhibited a substantial rebound increase of REMS during the first 24 h of recovery sleep but little change in either NREMS duration or NREMS slow wave activity, a presumed marker of NREMS intensity. Evidently, the degree of NREMS restriction caused by RSD is not sufficient to generate a compensatory response during recovery sleep, suggesting that it is physiologically insignificant. Other work has shown that partial NREMS deprivation in apparatus control rats does not induce the morbidity and mortality effects associated with total sleep deprivation or selective near-total RSD (38). Finally, a recent study has shown that suppression of cell proliferation by a treadmill RSD method correlates significantly with loss of REMS but not NREMS (12).
Consistent with Mirescu et al. (31), we did not see an effect of RSD on differentiation of new cells after BrdU administration, because the percentage of cells double labeled for BrdU and DCX, a marker for immature neurons, did not differ across RSD and control groups. By contrast, 96 h of total sleep deprivation has been reported to reduce both the number of new cells and the percentage of new cells exhibiting a neuronal phenotype 3 wk after BrdU administration (14). Thus there may be effects of combined loss of NREMS and REMS that extend beyond those associated here with selective loss of REMS.
Mechanisms by Which RSD Affects Cell Proliferation
The mechanisms by which sleep disruption affects cell proliferation in the dentate gyrus are unknown. Adult neurogenesis is regulated by multiple hormone and neurotransmitter systems (1, 2), some of which also regulate or are regulated by sleep and are affected by sleep loss. Some proneurogenic systems afferent to the hippocampus exhibit enhanced activity during REMS [e.g., acetylcholine (6, 32)], while other proneurogenic hippocampal inputs are quiescent in NREMS and virtually silent in REMS [e.g., serotonin and norepinephrine (20, 23, 36)]. Similarly, sleep disruption may decrease or increase levels of neurogenic growth factors, depending on the factor and duration of deprivation (see, e.g., Refs. 8, 16, 18, 48). Nonuniform changes of neurogenic factors during sleep and sleep deprivation raise the question of whether a neurogenic effect of sleep, suggested by the inhibitory effect of sleep loss on cell proliferation, reflects a special role for the sleep state as a time for cell proliferation (e.g., when one or more regulatory factors are high or low) or whether sleep promotes neurogenesis indirectly, by maintaining optimal functioning of neural or endocrine systems that mediate neurogenic effects of waking experience. There is so far little direct evidence that cell proliferation occurs exclusively or predominantly during sleep. Evidence consistent with the idea includes the observation of increased BrdU labeling following injections near the end of the daily sleep period in rats (13) and the observation of increased levels of cell proliferation a week after a 3-day REMS deprivation (31). In both cases, cell proliferation is increased after a period of elevated sleep, as a consequence of the daily sleep-wake cycle or in compensation for sleep loss (37). However, both observations are open to other interpretations. Other studies have failed to see a daily rhythm of proliferation, although these differed methodologically in one or more ways (19, 22, 49). It remains possible that cell proliferation does not occur preferentially during sleep but that sleep is essential for normal functioning of brain systems that regulate hippocampal cell proliferation, which may occur in any behavioral state. A similar indirect mechanism may underlie effects of sleep loss on hippocampal memory functions.
Perspectives and Significance
The results of this study are consistent with other work demonstrating that deprivation of REMS has effects on hippocampal structure and function that are not explained by elevated adrenal glucocorticoids (28, 41). These results raise the possibility that chronic sleep disruptions associated with lifestyle, adverse life events, aging, sleep disorders, neurodegenerative diseases, or other conditions may attenuate production of new hippocampal neurons, thereby compromising hippocampal adaptive plasticity and associated memory or other functions. While speculative, a causal link between sleep and hippocampal neurogenesis may underlie the reduced hippocampal volume observed in humans with chronic insomnia (39).
This work was supported by operating and equipment grants from the Natural Sciences and Engineering Research Council of Canada and the Canadian Institutes for Health Research. L. A. M. Galea is a Michael Smith Foundation for Health Research Senior Scholar.
We thank Dr. Melissa Holmes for assistance with a preliminary study, Dr. Mary Dallman for advice on corticosterone replacement, and Dr. Christine Carson for assistance with confocal microscopy.
↵* A. D. Mueller and M. S. Pollock contributed equally to this work.
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