Introduction

Circadian rhythm occurring over a 24-h period has been well established in humans. The daily rythmic changes are exhibited in various physiological processes such as sleep-activity, appetite, secretion of hormones, and metabolism [1–3]. The mammalian circadian system is composed of two elements: the central Clock system, located in the suprachiasmatic nucleus of the anterior hypothalamus [4] and acting as the “master” oscillator and generator of the body’s circadian rhythm under the strong influence of the light/dark input from the eye, [1, 5, 6] and the peripheral clock system distributed in all organs and tissues and being under the regulation of the central Clock, by an as yet unknown mechanism(s) [1, 6, 7]. Both clocks communicate with each other and generate circadian rhythmicity by the coordinated activation/inactivation of self-oscillating transcription factors.

To date at least 10 mammalian Clock-related genes have been identified Period-1 (Per-1), Period-2 (Per-2), Period-3 (Per- 3), circadian locomotor output cycles kaput (CLOCK), Cryptochome 1 (CRY1), cryptochrome 2 (CRY2), the transcription factors aryl hydrocarbon receptor nuclear translocator-like (ARNTL or BMAL1), CASENKINASE 1, Timeless (TIM) and retinoic acid-related orphan nuclear receptor (ROR) [4, 8]. In mammals the circadian clock regulation is made up of two interlocking, regulatory feedback loops. The loops rely on positively regulating genes such as CLOCK and BMAL1 and negatively regulating genes such as Per-1, Per-2, CRY1, CRY2, and TIM [9, 10]. The BMAL1 and the CLOCK activate also the period genes (Per-1, -3) and cryptochrome (CRY1-2) genes. After accumulation in the cytoplasm Per/CRY complexes relocate into the nucleus to inactivate CLOCK/BMAL1 transactivation, thereby down-regulating their own expression [9, 10]. RORα is a core part of the clock machinery that positively regulates the expression of BMAL1 [9, 10].

Circadian clock system seems to exert multiple actions upon organs and tissues, among them immune system [4, 11]. Both the circadian clock system and the stress-responsive HPA axis are fundamental for survival and appear to interact with each other [4]. The master clock is responsible for the circadian rhythm of circulating adrenocorticotropic hormone (ACTH) and cortisol [12, 13]. In humans, circulating cortisol concentrations are tightly regulated by the central components of the HPA axis and fluctuate naturally in an ultradian and circadian fashion [12, 13]. Interestingly, circadian rhythm transcription factor CLOCK found to acetylate multiple lysine residues in the hinge region of the human glucocorticoids receptors (hGR), and suppress the hGR-induced transcriptional activity [14]. Thus, this peripheral CLOCK-mediated circadian acetylation of the hGR may function as a target-tissue, gene specific counter regulatory mechanism to the actions of diurnally fluctuating cortisol, effectively decreasing tissue sensitivity to glucocorticoids (GCs) in the morning and increasing it at night [12, 13]. Moreover, there are interesting data regarding the presence of another GRE-dependent gene the GILZ (Glucocorticoid-Induced Leucine Zipper) which plays an important role in GCs immunomodulation and is up-regulated by GCs, mainly in lymphoid organs [15, 16].

Polyglandular autoimmune syndrome (PAS) is characterized by the co-existence of autoimmune diseases in more than one endocrine glands. Particularly in PAS type ΙΙΙ, autoimmune thyroiditis occurs with another organ-specific autoimmune disease such as immune-mediated diabetes mellitus (DM), pernicious anemia, vitiligo, alopecia, celiac disease, hypogonadism, sarcoidosis, Sjogren’s syndrome, myasthenia gravis, rheumatoid arthritis, atrophic gastritis without involving the adrenal cortex [17–19].

In the present study we investigated the daily patterns of 6 circadian Clock-related genes expression (CLOCK, BMAL1, ROR, Per-1, -2, -3), as well as of GR-a and GILZ genes in patients with PAS type III, at two different time-points (morning and evening). We also studied the diurnal cortisol levels through two time-points (morning and evening) and its correlation to the Clock-related genes expression, in an attempt to elucidate if there is any association between a disturbed daily pattern in the expression of the aforementioned genes and PAS type III.

Material and methods

Studied population

Medical records from the Endocrine unit of the Department of Pathophysiology and First Department of Internal Medicine, Laiko Hospital, National and Kapodistrian University of Athens were retrieved by two independent investigators for the identification of patients with PAS type III. Inclusion criteria were the clinical and biochemical documentation of PAS type III. Actually, PAS type III, is characterized by autoimmune thyroiditis along with another organ-specific autoimmune disease, apart from adrenal insufficiency [19]. Recent morning cortisol levels ≥ 18 μg/dl or a normal response to Synacthen test were used to ensure the integrity of the HPA axis [20, 21]. We excluded patients with abnormalities of the HPA axis, patients on any kind of steroid treatment and patients with psychiatric diseases. Nineteen patients diagnosed with PAS type III (5 males, mean age: 51 ± 13 years old) were finally enrolled and 12 healthy volunteers (4 males, mean age: 49.8 ± 12.9 years old) matched for age, sex and body mass index (BMI) in whom the presence of an unsuspected underlying autoimmune disease was excluded based on clinical history and their medical records provided by their physician, clinical examination (absence of clinical evidence of concomitant autoimmune endocrine diseases) as well as biochemical and hormonal analysis including thyroid-stimulating hormone (TSH), ACTH, cortisol, complete blood count and vitamin B12 measurement.

The study was approved by the Laiko hospital on the Ethics of Human Research according to Helsinki declaration and written informed consent were obtained from all patients by a single trained observer.

Blood and sample collection

Venipuncture was carried out from all the participants at 08:00 a.m and 08:00 p.m after a 12 h overnight fasting and a normal 8 h sleep. Blood sampling was performed concomitantly for biochemical and hormonal analyses as well as peripheral blood mononuclear cells (PBMCs) isolation in a supine position after 20 min rest at two different time points (8:00 a.m. and 8:00 p.m.) for patients and controls. All patients and controls were admitted to the Endocrine Unit on the day of study and anthropometrics data were obtained by two trained observers. Biochemical tests for plasma fasting glucose, serum cholesterol and triglycerides levels were performed in the Clinical Chemistry Laboratory of the “Aghia Sophia” Children’s Hospital.

PBMCs isolation

PBMCs from 19 patients with PAS type III and 12 healthy controls were examined for the expression of the 8 studied genes using qRT-PCR to elucide gene expression profiles by comparing relative mRNA levels (ΔΔCt method) in the morning and in the evening. Blood samples from patients and controls were collected from May 2015 throughout February 2016.

PBMCs were isolated from whole blood by density gradient centrifugation using Biocoll Seperating Solution (l6115). Briefly, anticoagulated blood (15 ml) was diluted with an equal volume of phosphate-buffered saline (PBS, pH 7.4) and applied over seperating solution. The tubes were centrifuged at 1500 × rpm for 30 min The cells interface layer was carefully removed and washed twice with PBS-1% of Fetal bovine serum (FBS). Pellets were re-suspended in BD pharma lysis buffer (Cat.No.555899) and incubated for 15 min in a dark chamber to lyse contaminating RBCs. PBMCs isolated from blood were stored at –80 for maximum up to a week. RNA and protein for the majority of the samples were extracted within 7 days after the storage of PBMCs at –80.The cell viability of PBMC was assessed by trypan blue exclusion test before storage.

Real-time quantitative (qRT-PCR) analysis

The mRNA expression of the GR-a, the Clock-related genes and the glucocorticoid-responsive genes (GILZ) were determined by qRT-PCR. Relative changes in genes mRNA expression were assessed by the standard 2(-Delta Delta C(T)) (ΔΔCt method) [22]. Relative mRNA expression in the morning and the evening was compared for patients and controls separately as well as between patients and controls. The ratio (R pm/am) of the relative mRNA expression of the 8 studied genes of patients with PAS type III compared to controls was also estimated in order to study their circadian pattern. The ratio of serum cortisol pm/am (F pm/am), as well as the ratio of relative mRNA levels (R pm/am), were additionally compared between patients and controls. Total RNA was isolated using RNeasy Plus Mini Kit (74134,QIAGEN) according to the manufacturer’s instructions. Concentration of all RNA samples was quantified using Nanodrop ND-2000 spectrophotomer (Thermo Fisher Scientific, Waltham, MA). Complementary DNA (cDNA) was reverse transcribed using the iScript cDNA synthesis kit (170–8891,Bio-Rad).

Real-time PCR with SYBR Green detection was conducted on the BIORAD CFX96 Touch RT-PCR. All reactions were carried out in triplicates. Each q-PCR run included a negative control containing all reaction’s components but template cDNA to detect contamination or non-specific amplification. cDNA from MCF-7 cells was used as an internal control to correct the inter-assay variation for samples run on different plates. The relative fold change was calculated using the 2(-Delta Delta C(T)) [22]. Primer sets used to amplify the genomic region of reference and target genes are included in the supplementary data (Supplementary Table 1) [12]. Amplification efficiencies and correlation coefficients (R 2) of each primer were evaluated using the slopes of the standard curves obtained by serial dilutions. We used β-actin and GAPDH as housekeeping genes in 8 samples to normalize mRNA levels of different genes. As no significant difference was observed in our results normalized with either of these two housekeeping genes, gene expression was normalized to actin.

Western blot analysis

For protein extraction, PBMC pellets were ressuspended in cell lysis buffer (Cell Signaling, Cat.No. 9803) supllemented with phenylmethylsulphonyl fluoride (PMSF). Protein concentration was determined with Bradford assay [23]. Protein lysates (30 mg approximately) were loaded and separated in 8% polyacrylamide-SDS gels. Proteins were then electrophoretically transferred onto nitrocellulose membranes and blocked in 5% nonfat milk in PBS-T for 1 h at room temprature. Membranes were incubated with total GR antibody (3660, Cell Signaling; 1:1000) and anti-actin antibody (MAB1501, Millipore, 1:5000) at 4 °C over-night. After three cycles of washing membranes were probed with secondary HRP-conjugated anti rabbit (AP132P, Millipore; 1:2500) or anti mouse-antibody (31430, Thermo Scientific; 1:2500). The blots were developed using enhanced chemiluminescence solution (170–5061, BioRad). Signals were densitometrically measured by ImageJ software and normalized to β-actin Protein from GR-α transfected Cos cells was used as a positive control.

Biochemical and hormonal analysis

Blood samples from patients and controls were collected from May 2015 throughout February 2016 at two different time points (8:00 a.m. and 8:00 p.m.) in a supine position after 20 min of rest. Serum levels of cortisol, TSH and plasma ACTH were measured on an Immulite 2000 analyzer (Siemens Healthcare Diagnostics Products Ltd., Llanberis, Gwynedd LL55 4EL United Kingdom UK) using two-site chemiluminescent immunometric assays with analytical sensitivities for cortisol 0.20 μg/dl, for TSH 0.004 μIU/mL and for ACTH 5 pg/mL. The intra-assay and inter-assay precision CVs for cortisol have a range of 5.2–9.4%, for TSH of 3.8–12.5% and for ACTH of 6.1–10.0%. All measurements were performed in the same laboratory the Clinical Chemistry Laboratory of the ‘Aghia Sophia’ Children’s Hospital in Athens.

Statistical analysis

All the data are reported as mean ± standard errors (STDEV). Non-parametric parameters were analyzed with Mann–Whitney test for the comparison between patients and controls while Wilcoxon test was performed for the comparison of non-parametric paired values in the same group. Student t-test was applied to evaluate the difference of parametric parameters between the groups. For correlation analysis we used Spearmen rank correlation coefficient test. All tests were 2-sided with statistical significance set at 0.05 and all computation were made using PRISM 7.

Results

Studied population

The clinical and laboratory characteristics of the studied population are shown in Table 1. Ten patients had cortisol levels at 08:00 < 18 μg/dl and underwent a Synacthen test that was found to be normal. All patients had autoimmune thyroiditis; the presence of other autoimmune diseases needed to confirm the diagnosis is shown in Table 1. Although patients with PAS type III present elevated BMI, cholesterol and fasting blood glucose compared to controls these differences in their metabolic profile was not statistically different between patients and controls (Table 1).

Table 1 Characteristics of the studied groups
Full size table

Blood hormone analysis

Patients and controls presented signifficanlty higher cortisol and ACTH levels in the morning compared to the evening indicating that the activity of HPA axis fluctuates in a circadian fashion (Table 2). Comparison of the two groups showed no statistically significant difference of cortisol, ACTH, and TSH hormones levels (Table 3).

Table 2 Comparison of daily variation (a.m. vs. p.m.) of hormones levels in patients and controls and the daily variation of the relative mRNA expression of the eight studied genes in the patients and the control group
Full size table
Table 3 Correlations among the ratio of mRNA levels (R p.m./a.m.) of GR-a, CLOCK, BMAL1, PER-1-3, ROR, GILZ with the ratio of cortisol levels (F p.m./a.m.)
Full size table

Daily pattern of the relative mRNA expresssion (morning vs. evening) of the Clock-related, GR-a and GILZ genes in patients and controls

In patients, transcripts of BMAL1,CLOCK, and Per-1 displayed a time-dependent variation pattern with significantly higher relative mRNA expression in the evening compared to the morning (p = 0.03, p = 0.029, p = 0.013, respectively) (Figs. 1a, c, e). Transcripts of ROR and GR-a were also upregulated in the evening compared to the morning without however statistical significant variations (p = 0.8 and p = 0.1, respectively), (Table 2). Transcripts of Per -2, -3 and GILZ showed similar daily pattern with absence of circadian rhythm variation (p = 0.79, p = 0.9 and p = 0.86, respectively) (Table 2).

Fig. 1
figure 1

Variation pattern (8 a.m. vs. 8 p.m.) of the relative mRNA levels (fold increase) of CLOCK, BMAL1 and Per-1 in patients a, c, e and controls b, d, f

Full size image

In the control group among the eight genes, only transcript of Per-3 displayed a circadian pattern with significantly higher relative mRNA expression in the morning compared to the evening (p = 0.03). Transcripts of GR-a, BMAL1 (Fig. 1b), CLOCK (Fig.1d), ROR, GILZ, Per-1 (Fig. 1f), and Per-2 showed no statistical significant circadian rhythm variation (p = 0.67, p = 0.14, p = 0.52,p = 0.12, p = 0.09, p = 0.6 and p = 0.66 respectively) (Table 2).

Additionally, comparison of the relative mRNA expression of the CLOCK, BMAL1, GR-a and the GILZ genes in the morning between patients and controls showed that patients had statistical significant lower mRNA levels of the BMAL-1 (p = 0.04) and marginally lower levels of the CLOCK and GILZ (p = 0.05 for both genes) in the morning compared to controls. GR-a expression in the morning didn’t differ significantly between patients and controls (p = 0.7). Comparison of the relative mRNA expression of the aforementioned genes in the evening revealed that patients had statistically significant higher CLOCK mRNA levels (p = 0.04) and lower GR-a mRNA levels (p = 0.02) compared to controls (Fig. 1, supplementary data)

Comparison of the daily pattern of the Clock-related, GR-a and GILZ genes between patients and controls

The R p.m./a.m. of the BMAL-1, the Per-3, the GR-a and the CLOCK genes transcripts differed statistically between patients and controls –although for Per-3 and GR-a marginally (p = 0.033, p = 0.05, p = 0.05 and p = 0.018, respectively) (Figs. 2a–d). In these four genes, the R p.m./a.m. was statistically significant higher in patients compared to controls. For the rest of the studied genes, ROR, GILZ, Per-1, and Per-2 the R p.m./a.m. showed no statistical significant difference in their circadian pattern between patients and controls (Table 2).

Fig. 2
figure 2

Ratios (R p.m./a.m.) of the relative mRNA levels of the BMAL1 a, Per-3 b, GR-a c, and CLOCK d in patients compared to controls

Full size image

Western blot analysis of daily variation of GR-a protein levels in patients and controls

As it is shown in Fig. 2c there was a marginal significant difference in the circadian pattern of GR-a expression between patients and controls. Western Blot analysis showed that the patients exhibiting significantly higher R p.m./a.m. of GR-a protein levels compared to controls (p < 0.05), in accordance with the mRNA GR-a levels (Fig. 3a). Figure 3b shows representative results of the morning and evening protein expression of GR-a in patients and controls.

Fig. 3
figure 3

a R p.m./a.m. of GR-a protein levels in patients and controls (* < 0.05) b. Representative western blot analysis of the expression GR-a in four patients (P) and four controls c in the morning (p.m.) and in the evening (a.m.)

Full size image

Correlation analysis of cortisol circadian variation (F p.m./a.m.) with the circadian variation (R p.m./a.m.) of the relative mRNA levels of the Clock-related, GR-a and GILZ genes

The ratio (F p.m./a.m.) of cortisol daily variation demonstated a significant positive correlation with the daily variation of R pm/am of the relative mRNA levels of GILZ in patients group [(r, 95%CI) = 0.46(–0.04–0.8)], (p = 0.002) (Table 3). No significant correlations were found between cortisol daily variations and the other genes expression. In controls, the ratio of cortisol daily variation demonstrated negative correlation with the daily variation of the ratio R p.m./a.m. of the GR-a gene’s expression [r (95%CI) = −0.7(−0.9 to 0.19)], (p = 0.01), while this correlation was abolished in patients. Data are shown in Table 3.

Discussion

In this study we investigated the circadian pattern—as estimated by two different time points (morning and evening)- of 6 Clock-related genes (CLOCK, BMAL1, Per-1, 2, 3), GR-a and GILZ in patients with PAS type III and in healthy controls. Our data showed that patients with PAS type III have a statistical significant time-dependent variation pattern for CLOCK, BMAL1 and Per-1 with higher expression in the evening compared to the morning. When we compared circadian patterns (R p.m./a.m.) between patients and controls, we found significant differences of the daily variation of GR-a, CLOCK, BMAL1 and Per-3. All these four genes were significantly upregulated in the evening in patients compared to controls, leading to a reversed circadian pattern. Despite this reversed circadian rhythm for CLOCK and BMAL1, circadian pattern of cortisol secretion was not disrupted in patients.

Present data show that cortisol levels are tightly regulated by the HPA axis and that probably circadian alteration in the secretion of the two principal regulators of the Clock system is not sufficent to disturb the normal circadian rhythm of cortisol secretion, in our patients [8, 24]. It is known that CLOCK mediates GR-a acetylation in healthy subjects and it is upregulated in the morning increasing acetylation of GR-a [12]. This effect attenuates the transcriptional activity of the GR-a which is stimulated by cortisol [12]. GILZ transcription is directly transactivated by the GCs in a GRE-dependent manner.

Since patients with PAS type III, have higher transcript levels of CLOCK gene in the evening compared to morning, we would expect a higher acetylation of GR-a in the evening, compared to morning, and lower expression of GILZ. However, neither GILZ nor GR-a genes expression, present any significant circadian variation between morning and evening in patients. Comparing CLOCK and BMAL1 mRNA expression between patients and controls we found significant lower expression of morning CLOCK (p = 0.05) and BMAL1(p = 0.04) genes in patients compared to controls, which would expect to result in increased morning sensitivity to GCs in patients. Accordingly, there was a higher expression of evening CLOCK gene in patients compared to controls suggesting decreased evening sensitivity to GCs. However, these were not mirrored by the daily pattern of GILZ expression which is a known bioassay of glucocorticoid sensitivity. A possible explanation is that although, GILZ expression is mainly controlled by GCs it is also modulated by a variety of cytokines which could be differentialy secreted in PAS [25]. If such an altered sensitivity to GCs could be causally related to the disease, remains to be further explored.

Data from other studies in human PBMCs of healthy volunteers have shown that mRNA expression of CLOCK and BMAL1 was higher in the morning compared to the evening suggesting that they are under circadian regulation [12]. In contrast the mRNA expression of the other auxiliary negative regulators Per-1 and CRY1 in human PBMCs were higher in the evening than in the morning, further indicating that they were under reciprocal circadian regulation opposed to that of CLOCK/ BMAL1 [12]. The mRNA expression of RORa, a component of the auxiliary loop of the circadian clock transcriptional system, also demonstrated the same circadian rhythmicity as Per-1 and CRY1 in human PBMCs of healthy subjects. This circadian rhythm of the Clock-related genes was also confirmed in human dermal fibroblasts [12, 26–28]. However other studies have shown that in PMBCs of healthy subjects mRNA expression of Per-1, 2, 3 and CRY 1, 2 genes was higher in the morning compared to the evening whereas BMAL1 mRNA expression was higher in the evening [11]. In our study, in PBMCs from healthy subjects, only transcripts levels of Per-3 showed circadian pattern with higher levels in the morning. These conflicting data indicate that there are yet some controversies in the litterature about the exact circadian regulation of clock genes.

Alteration of the expression of Clock-related genes has been already studied in malignancies and other autoimmune diseases. In PMBCs of patients with chronic myeloid leukemia mRNA expression of Per -1, 2, 3 and CRY 2 presented circadian variation however different from those of healthy subjects with higher levels in the evening [11]. Disruption of Clock-relatedgene expression has been implicated in other autoimmune diseases such as diabete type 1 in mice [29, 30] as well as in human islets of patients with diabetes type 1 [31], in patients with rheumatoid arthritis [32] and in patients with infammatory bowel dieases [33].

In the present study the major positive regulators of clock system BMAL1 and CLOCK showed a reversed circadian rhythm in patients with PAS III with increased relative mRNA levels in the evening and lower in the morning. mRNA levels of the negative regulator Per-1 maintained the circadian rhythm described before, with higher levels in the evening acting thus as counterbalance in the stimulation of the clock system by CLOCK and BMAL1 in the evening in patients. Comparison with healthy subjects showed that circadian rhythm for GR-a, CLOCK, BMAL1 and Per-3 was significantly different from patient with PAS type III. In patients with PAS type III, the ratio R p.m./a.m. of transcript levels of GR-a, CLOCK, BMAL1 and Per-3 showed reverse circadian rhythm compared to controls with overexpression in the evening compared to the morning.

PAS type III is characterized by chronic inflammatory infiltration by lymphocytes with some macrophages, natural killer (NK) cells, and plasma cells of the affected organs. It is associated with immune-response of the class-II loci of the HLA complex and the cytotoxic T lymphocyte antigen-4 (CTLA-4) locus as well as the presence of autoantibodies reacting to targeted tissue-specific antigens [19, 34]. The infiltrating lymphocytes are of both B and T lineages, while the T-cell population includes both the CD4 + and CD8 + subsets. It is clear, however, that both genetic and environmental factors are involved [35].

Our increased knowledge of the molecular and cellular immune mechanisms in the pathophysiology of PAS could explain the role of clock genes. In healthy subjects B and T cell counts as well as cytokines levels present a circadian variation across the day with peak generally observed at night [36]. CD4 + T cell responses are regulated by an intrinsic cellular circadian oscillator capable of driving rhythmic CD4 + T cell immune responses [35]. Serum levels and in vitro production of interferon (IFN), tumor necrosis factor (TNF)-a, interleukin (IL)-1, IL-2, IL-6, and IL-12 were all shown to present a rhythm in humans, with a peak generally observed at night or in the early morning in humans [36, 37]. Daily variations of nuclear factor (NF) kappa-B(kB) mediated transcription involves the participation of CLOCK, which is found in protein complexes with the p65 subunit of NF-kB [38–40]. This mechanism is independent of the usual heterodimerization of CLOCK with BMAL1 [35]. GILZ interacts also with the NF-kB [39, 40].

Furthermore, ROR genes have gained a significant amount of attention over the past few years due to their essential role in the development of T-helper (TH) 17 cells [41]. Mice deficient in both RORα and RORγ completely lack TH 17 cells and are resistant to the development of autoimmune diseases [42]. Similarly to other peripheral tissues, Per-1, 2 and CLOCK/BMAL1 genes in NK cells present higher levels at day and night [40] while in rat-derived RNK16 NK cells, knockdown of Per-2 or BMAL1 differentially alters gene expression of IFN-g, TNF-a, granzyme B, and perforin [40]. All the above data demonstrate that differential expression of Clock-related genes can perturb the immune cells subpopulation and their function, possibly resulting in autoimmune diseases.

Limitation of our study is the small number of patients with PAS III as well as the only two time points that PBMCs were collected. The well-characterized patient and control group and a sufficient number of candidate Clock-related genes that were analyzed, are among the strengths. In addition, to our knowledge this is the first time that Clock system was studied in patients with polyglandular autoimmune syndrome and specifically in PAS type III.

In conclusion, the findings of the present study suggest that there is an abberant expression of Clock-related genes in patients with PAS type III compared to healthy controls. Disturbance of the daily pattern expression of the Clock-related genes in patients with PAS type ΙΙΙ indicates possible association with the pathogenesis of the disease, although a cause-effect relation remains to be elucidated. Expanding our knowledge about how circadian clocks regulate inflammatory responses may provide novel therapeutic strategies for the prevention or even treatment of such autoimmune syndrome focusing mainly in the restoration of circadian rhythms.