Suprachiasmatic nuclei have glucocorticoid receptors that can activate signaling pathways but do not regulate their circadian rhythms.

Experiments were conducted with adult male and female Wistar:Han rats (Institute of Physiology of the Czech Academy of Sciences) and mPer2^Luc mice [28]. Animals were kept under a regime with 12 h of light and 12 h of darkness (LD12:12, lights on at 06:00 a.m., designated Zeitgeber time 0) at 21°C ± 2°C with free access to food (standard diet) and water. Female rats or mice were mated with males, and those with positive sperm in vaginal smears (assigned as embryonic day E0) were housed individually. The dams and their pups were maintained undisturbed under LD12:12 until weaning. Adult rats and pups from 3 days of age were sacrificed under deep anesthesia by injections of a mixture of ketamine (Vétoquinol, s.r.o., Czech Republic; 120 mg/kg) and xylazine (Bioveta a.s., Czech Republic; 12 mg/kg). Mice of all ages and rat pups up to 1–2 days of age were sacrificed via rapid cervical dislocation.

To detect the effect of DEX on the PER2‐driven bioluminescence in the SCN of mPer2^Luc mice ex vivo, brains were collected from fetuses at embryonic day 17 (E17), from pups at postnatal days (P)1/2, P3, P5, P10, and from adults (mean 6‐month‐old ±2 months). SCN explants were prepared and treated for bioluminescence monitoring as described below.

For the detection of 24 h gene expression profiles, 5 rats were sacrificed every 4 h during the 24‐h cycle to collect the brains of pups on postnatal days (P)2, P10, P20, and P28. To collect brains from the fetuses at E19, 5 fetuses from one pregnant rat were sacrificed at each time point.

To demonstrate the acute effects of dexamethasone (DEX) in vivo, adult male rats were injected with 0.5 mL dexamethasone (DEX, 1 mg/kg i.p.) or vehicle (phosphate‐buffered saline) at 9 a.m. (ZT3) and sacrificed 1, 2, and 4 h after the injection to collect brains (n = 4–5/time point). The dose of DEX was selected based on previous studies [19], and the timing of injection was chosen to coincide with low levels of endogenous GCs in rats [29]. Whole brains were immediately frozen in dry ice and stored at −80°C. For RT qPCR, brains were cryosectioned into 20‐μm thick coronal sections containing the medial part of the rostro‐caudal extent of the SCN as visualized with cresyl violet staining (Sigma‐Aldrich, St. Louis, MI, USA). The SCN was precisely separated bilaterally using a laser microdissector (LMD6000, Leica), as we have previously described elsewhere [30].

The brains of E17‐P3 mice were cut into 300‐μm coronal sections using a tissue chopper, while brains from P5, P10 and adult mice were immediately sectioned into 250‐μm coronal sections in cold Earle's Balanced Salt Solution (EBSS, Sigma) using a vibratome (Leica, Germany). The slice containing the medial SCN region was dissected and immediately placed on Millicell Culture Inserts (Merck) inside 35‐mm Petri dishes containing 1 mL of recording medium. Explants were kept in air‐buffered recording media containing DMEM (Sigma) supplemented with 2% B27 supplement (Thermo Fisher), 0.1 mM D‐Luciferin (Biosynth, Switzerland), 100 U/mL penicillin, 100 μg/mL streptomycin and 2 mM GlutaMAX (Thermo Fisher). The explants were placed in the LumiCycle apparatus (Actimetrics, Wilmette, IL, USA) to monitor bioluminescence at the tissue levels. Explants were treated with either 100 nM DEX (1 μL of 0.1 mM DEX in 1% ethanol/1 mL of recording medium) or vehicle (1% ethanol) 3–5 days after the addition of fresh recording medium. Some explants were treated repeatedly after washout and at least 7 days of resting to reduce the number of sacrificed animals. To inhibit glia, DL‐fluorocitrate barium salt (FLC, Sigma) was used according to Paulsen, Contestabile et al. [31] to prepare a fresh 7 mM FLC stock; 7 μL of FLC or VEH (identically prepared stock without FLC) was added to 1 mL of recording medium 48 h before the treatment with DEX. To block the Na⁺ channels, 1 μM tetrodotoxin (Sigma) was present in the recording medium before treatment with DEX.

To image the PER2 luminescence signal at the microscopic level, 14 explants were placed in the motorized luminescence microscope Luminoview LV200 (Olympus, Japan) with LUCPLFLN40X objective (Olympus) and the EMCCD camera ImageEM X2 (Hamamatsu, Japan) water cooled by Minichiller 280 (Huber, Germany); luminescence signal was imaged every hour with an exposure time of 10 min and 250 electromagnetic gain. One micromole of tetrodotoxin (Sigma) was present in the recording medium to block the Na⁺ channels; to avoid affecting the position of the explant above the objective, instead of adding DEX (n = 7) or VEH (n = 7) to the medium after 3 days of imaging, 5 μL of pre‐diluted DEX (200 nM) or VEH was dropped directly on top of the explant.

Lumicycle Analysis (Actimetrics, USA) was used to analyze the raw bioluminescence data. Data were baseline‐subtracted with the running 24‐h average prior to fitting to a sine wave to calculate the amplitude, period and phase shift before and after treatments. Treatment‐induced phase shifts in the bioluminescence rhythms were quantified by fitting a sine curve to the first at least 3 complete circadian cycles of a 24‐h running average baseline‐subtracted rhythm and then extrapolating beyond the time of the treatment (reflecting the original phase). The resulting absolute phase shift was calculated as a difference between the extrapolated sine curve and the post‐treatment luminescence trace (reflecting the new phase). The shift was assigned as a phase advance (+) or a phase delay (−). The phase response curve (PRC) was constructed by plotting the calculated phase shift as a function of the time of the treatment normalized to the endogenous period in vitro and expressed relative to the trough (treatment time, tt0) or peak (tt12) of the rhythm. For statistical comparisons between DEX and VEH PRCs, the data for the phase shifts between tt0 and tt13.3, and tt13.3 and tt24 were fitted with linear regression curves and their slopes were compared by F‐test. The phase transition curve (PTC) was constructed by plotting the peak of the first full cycle after the treatment (y, new phase) as a function of the peak of the extrapolated sine curve (x, old phase). The PTC data were plotted as a tetraplot for clarity. CellSens Dimensions (Olympus) was used to acquire LV200 images and export them as 16‐bit TIFFs. Cosmic rays were removed by pixel‐wise subtraction of consecutive TIFFs. Whole SCN were manually traced in Fiji ImageJ and algorithmically subdivided into approximately cell‐sized regions of interest (ROIs, on average cca. 500/explant). Data were exported as mean intensity signal and xy coordinates were analyzed after detrending and denoising using a custom modified Python script based on the per2py package [32].

The adult mice and rats were deeply anesthetized with a mixture of ketamine (Vétoquinol, s.r.o., Czech Republic; 120 mg/kg) and xylazine (Bioveta a.s., Czech Republic; 12 mg/kg) and transcardially perfused with phosphate buffer (PBS) followed by paraformaldehyde (4% in PBS, Sigma). Brains were removed and postfixed in 4% PFA for 12 h at 4°C and cryoprotected in 20% sucrose‐PBS overnight at 4°C. Coronal 30‐μm‐thick sections were cut with a freezing microtome (CM1850, Leica, Germany) and free‐floating sections were processed for immunofluorescence methods as previously described [33, 34]. After the standard antigen retrieval procedure in citrate buffer (0.3% in PBS), the sections were incubated in 2% normal goat serum (abcam ab7481) for 60 min. The primary polyclonal antibodies applied overnight at 4°C were as follows: GR (M‐20) 1:100, rabbit, Santa Cruz sc‐1004; FOX2 1:500, mouse, abcam ab57154; GFAP 1:1000, chicken, abcam ab4674; ELAVL4 1:600, mouse, Millipore MABN153. The secondary antibodies (incubation for 1 h) were goat anti‐rabbit, 1:1000, Alexa Fluor 594, Thermo Fisher A‐11037; goat anti‐mouse, 1:1000, Alexa Fluor 488, Thermo Fisher A‐11001; goat anti‐chicken, 1:1000, Cy3, Abcam ab97145. The images were taken with the Leica SP8 WLL MP laser scanning confocal microscope.

Dissected rat SCN tissues were collected in a microfuge tube containing RLT buffer from the RNeasy Micro kit (Qiagen, Valencia, USA) and stored until RNA isolation. RNA was isolated using the RNeasy Micro kit (Qiagen); the whole elution volume was then reverse‐transcribed using a High Capacity cDNA RT Kit (ThermoFisher). Diluted cDNA was then amplified on LightCycler480 (Roche) using SYBR Select qPCR Master Mix (ThermoFisher) as described previously [30]. Reverse transcription‐quantitative polymerase chain reaction (RT qPCR) method was used to detect Nr3c1 (Forward: CAGAGAATGTCTCTACCCTG, Reverse: CTTAGGAACTGAGGAGAGAAG) mRNA levels in the SCN (3–5 replicates/time point). Relative quantity was calculated using Livak's ΔΔCt method against the reference gene Tbp (for 24 h expression profiles, For: CATCATGAGAATAAGAGAGCC, Rev.: GGATTGTTCTTCACTCTTGG) or B2m (for the acute effect of DEX, For: CGCTCGGTGACCGTGATCTTTCTG, Rev.: CTGAGGTGGGTGGAACTGAGACACG). All samples were analyzed in the same RT qPCR run. The reference gene Tbp was used for 24 h expression profiles designed to detect developmental changes in the SCN, as its expression was stable during development. The reference gene B2m was used to detect the acute effects of DEX in the SCN of adult animals.

RT qPCR profiles of Nr3c1 gene expression were analyzed by two‐way ANOVA with Šidák's multiple comparisons test to compare E19 and P28 samples (Figure 1). Period and amplitude of PER2 luminescence in adult SCN explants were compared by the Mann–Whitney test (Figure 2D–F). SCN genes in rats injected with DEX or VEH were measured by RT qPCR and analyzed by two‐way ANOVA with Šidák's multiple comparisons test (Figure 4). Amplitude of SCN explants treated with FLC/TTX and then with either DEX or VEH was compared before and after the treatment by Wilcoxon matched‐pair test (Figure 6B,C,G), while the partial PRCs were fitted with linear regression and the slopes were compared by F‐test (Figure 6H). Immediate amplitude increase in individual cell‐sized rois measured in 9 DEX‐ and 9 VEH‐and‐TTX‐treated explants was analyzed by the Mann–Whitney test (Figure 6J). p < 0.05 was required to report significance. Statistical tests implemented in Prism 7 (Graphpad, USA) and SciPy stats Python module were used.

We used RT qPCR to analyze Nr3c1 expression profiles in microdissected SCN samples during the developmental stages of E19, P2, P10, P20, and P28 (after the end of weaning) (Figure 1). Surprisingly, we found a gradual increase in Nr3c1 expression during ontogenesis, with the lowest mRNA levels of the GR gene at E19 and the highest levels at P28 (two‐way ANOVA comparison of E19 vs. P28, age effect p < 0.0001, time effect p = 0.4, interaction effect p = 0.03).

Since we found that GR levels in the SCN increase rather than decrease during development (as previously reported), we investigated the responses of the circadian clock in the SCN to the glucocorticoid analog dexamethasone (DEX, 100 nM) during five distinct perinatal developmental stages and in adulthood. To this end, we constructed phase response curves (PRCs) by analyzing the phase shifts induced by DEX treatments of SCN explants from E17, P1–2, P3, P5 to P10 at various time points to cover a 24‐h interval relative to the peak/trough of the PER2‐bioluminescence rhythm (Figure 2). At E17 (Figure 2A), the SCN clock was sensitive to the manipulation itself (VEH treatment), but DEX induced phase advances at the time of PER2‐bioluminescence decline, whereas VEH induced significant phase delays (tt16‐20). The sensitivity window was consistent with the results of our previous study in which we used a different method to prepare the SCN explants [26]. The PRC for P1‐2 explants showed responses to both DEX and VEH that were randomly spread over a 24‐h interval (Figure 2B). The magnitudes of phase shifts induced by both VEH and DEX progressively decreased at postnatal days (P)3 and P5 (Figure 2C,D), with little to no detectable phase shift until P10 (Figure 2E). The results showed that the SCN clock gradually gains resistance to DEX during the early postnatal period.

Next, we focused in more detail on the responses of the adult SCN clock to DEX. Construction of the full PRC confirmed that DEX did not induce significant phase shifts in the adult SCN clock (Figure 3A–C). There were also no significant persistent effects on period (Figure 3D) or amplitude (Figure 3E) measured over 3 days post‐treatment. Interestingly, DEX caused a rapid temporary increase in amplitude within 24 h after DEX treatment (Figure 3F).

The finding of acute DEX‐induced increase in the amplitude of the PER2‐bioluminescence rhythm in the adult SCN prompted us to test whether GR can activate canonical GC signaling pathways or affect clock gene expression in vivo. We injected adult rats with either DEX (1 mg/kg i.p.) or VEH at ZT3 and detected the expression of selected genes in microdissected SCN samples 1, 2, or 4 h later (Figure 4). We found a highly significant increase in the expression of Serum and glucocorticoid‐regulated kinase 1 (Sgk1, already 1 h after injection, two‐way ANOVA with Šidák's post hoc test, p < 0.0001) and Glucocorticoid‐induced leucine zipper protein (Gilz/Tsc22d3, 2 h after injection, p = 0.0015), both targets of GR transactivation. We also analyzed the expression of the clock genes Dec1, Per1, Per2, Cry1, Rora, Rev‐ErbA and Dbp. Only Dec1 (p < 0.0001) and Rora (p = 0.0028) responded to DEX injection after 4 h, while the other clock genes measured were not significantly affected. In contrast to Sgk1 and Gilz, mRNA levels of both Dec1 and Rora decreased by approximately 50% in response to GR activation, indicating a possibility of a noncanonical mechanism of GR action on the SCN clock.

We used immunofluorescence to spatially and phenotypically identify cells expressing the GR protein in the adult SCN. Since the experiments were performed both in rats and mice, we processed brains from both species. The results showed that GR is expressed in most cells of the adult SCN. We co‐localized GR with neuronal (ELVAL4 in rats, FOX2 in mice) and glial (GFAP in both species) markers. Confocal immunofluorescence images of the adult SCN (Figure 5, left for a rat, right for a mouse) confirmed the presence of GR immunoreactivity in both glia (yellow arrows) and neurons (white arrows).

To investigate whether glia play a role in the resistance of the SCN clock to GC, we tested the effect of an inhibitor of glial metabolism (50 μM fluorocitrate, FLC) on the parameters of the circadian clock in the adult SCN of mPer2^Luc mice. As expected, the impairment of glial metabolism by FLC decreased the amplitude of circadian oscillations compared to VEH (Figure 6A). However, this did not affect the response to DEX, as we found a comparable decrease in amplitude after treatment with VEH (Figure 6B, Wilcoxon paired test, p = 0.03) or DEX (Figure 6C, p = 0.03). The result suggests that glia do not play a dominant role in the previously observed effect on PER2 amplitude.

To reduce the intercellular coupling among individual neuronal oscillators, we treated the adult SCN explants from mPer2^Luc mice with Na⁺ channel inhibitor (1 μM tetrodotoxin, TTX) (Figure 6D–H). As expected, TTX acutely decreased the amplitude of the overall PER2‐bioluminescence rhythm analyzed in Lumicycle (Figure 6D). On the third day after the TTX application, we treated the explants with vehicle (Figure 6E) or 100 nM DEX (Figure 6F) and analyzed the effects on the amplitude (Figure 6G), measured 3 days before and after the treatment. Due to a progressive dampening of oscillations in the adult SCN with largely electrically uncoupled neurons, we detected a significant difference in amplitude before and after the treatment with VEH (Figure 6G left, Wilcoxon pair test, p < 0.0005); however, treatment with DEX prevented the dampening, because there was no significant difference in amplitude before and after the treatment (Figure 6G right, p = 0.18). Therefore, we analyzed the phase response to DEX in electrically uncoupled SCN by constructing the full PRC (Figure 6H) for 100 nM DEX (black squares) and VEH (gray circles). Both compounds produced phase shifts with directions and magnitudes dependent on the timing of the treatment. The effect of VEH on the phase of the TTX‐treated SCN suggests weakening of the robustness of the oscillator. We compared the resulting PRCs for the DEX and VEH by fitting regression lines approx. corresponding to the first and second half of the subjective night (determined by the transition points set at tt 4.5 and 13.5 h) (Figure 6H). The statistical comparison of the slopes revealed that in contrast to VEH, the DEX treatments induced significantly larger phase delays (F‐test, p = 0.0003) but not phase advances (p = 0.6) in electrically uncoupled adult SCN.

To reveal the dynamics of the response to DEX in TTX‐treated adult SCN at the single‐cell level, we used luminescence microscopy (Olympus LV200) and analyzed PER2 levels in approximately cell‐sized regions of interest (rois) recorded every 1 h for 72 h before explants were temporarily treated with either 200 nM DEX or VEH (a drop on top of the explant, Figure 6I). While TTX reduced their amplitude, the majority of rois remained significantly rhythmic and largely synchronized with each other. The recording confirmed a transient increase in the amplitude of the rhythm in individual rois in DEX‐ treated but not in VEH‐treated explants (Figure 6J, n = 9 explants/group, Mann–Whitney test, p < 0.0001). The mild effect is likely due to only a transient activation of GR, because in contrast to previous experiments the drop of 200 nM DEX is rapidly diluted in the medium below the tissue insert (eventually leading to a 200‐fold reduction from the initial concentration).

Altogether, the results show that TTX further increases the sensitivity of the adult SCN to DEX.