Resetting of Circadian Time in Peripheral Tissues by Glucocorticoid Signaling

See allHide authors and affiliations

Science  29 Sep 2000:
Vol. 289, Issue 5488, pp. 2344-2347
DOI: 10.1126/science.289.5488.2344


In mammals, circadian oscillators reside not only in the suprachiasmatic nucleus of the brain, which harbors the central pacemaker, but also in most peripheral tissues. Here, we show that the glucocorticoid hormone analog dexamethasone induces circadian gene expression in cultured rat-1 fibroblasts and transiently changes the phase of circadian gene expression in liver, kidney, and heart. However, dexamethasone does not affect cyclic gene expression in neurons of the suprachiasmatic nucleus. This enabled us to establish an apparent phase-shift response curve specifically for peripheral clocks in intact animals. In contrast to the central clock, circadian oscillators in peripheral tissues appear to remain responsive to phase resetting throughout the day.

Daily rhythms in gene expression, physiology, and behavior persist under constant conditions and must, therefore, be driven by self-sustained biological oscillators called circadian clocks [for reviews, see (1,2)]. Circadian clocks can count time only approximately and must be adjusted every day by the photoperiod in order to be in harmony with the outside world. In mammals, light signals perceived by the retina are transmitted directly to the suprachiasmatic nucleus (SCN) via the retino-hypothalamic tract (3). The SCN, located in the ventral part of the hypothalamus, is thought to contain the master pacemaker, which synchronizes all overt rhythms in physiology and behavior (4).

In most systems, circadian oscillations rely on a negative feedback loop in gene expression that involves multiple clock genes. In Drosophila, the repertoire of essential clock genes includes period (per),timeless (tim), clock(clk), cycle (cyc),doubletime (dbt), cryptochrome(cry), and vrille (vrl) (1, 5). During the past few years, one or more mammalian homologs to all of these genes have been uncovered. These include Per1, Per2, and Per3,Tim, Clock, Bmal1, Tau,Cry1 and Cry2, and E4bp4(1, 5, 6).

Molecular oscillators may exist in most peripheral cells ofDrosophila (7), zebrafish (8), and mammals (9, 10). In Drosophila and zebrafish, the peripheral clocks can be entrained directly by light (11–14). In mammals, it is thought that the phase of these peripheral timekeepers is reset by signals regulated by the SCN pacemaker (10). Because serum induces circadian gene expression in cultured rat-1 fibroblasts (9), one or more blood-borne factors must stimulate signal transduction pathways that influence the molecular oscillators in peripheral cells. Glucocorticoid hormones are particularly attractive candidates, because (i) they are secreted in daily cycles (15) and (ii) the glucocorticoid receptor (GR) is expressed in most peripheral cell types, but not in SCN neurons (16, 17). The second point would be consistent with glucocorticoids as entraining signals, given that circadian gene expression has a different phase angle in the SCN and in peripheral tissues (18).

We first investigated whether a 1-hour treatment with the glucocorticoid dexamethasone (19) can induce circadian gene expression in rat-1 fibroblasts (Fig. 1). This treatment rapidly activates Per1 expression (Fig. 1A) but, in contrast to a serum shock (9), does not activate Per2 mRNA levels above the slightly induced levels seen after treatment with the solvent alone (ethanol at 0.001%). We followed the accumulation pattern of circadian mRNAs during the 44 hours after the dexamethasone treatment (Fig. 1B). Similar to serum and unlike the solvent alone (Fig. 1C), dexamethasone elicits robust circadian gene expression of the clock genes Per1, Per2, Per3, andCry1 and the clock-controlled genes Rev-erbα and Dbp.

Figure 1

Dexamethasone induces circadian gene expression in rat-1 fibroblasts grown in tissue culture. At the indicated time, whole cell RNA was extracted and examined by ribonuclease protection. The antisense RNA probes are indicated at the right hand side of the panels. TBP mRNA (a transcript that is not induced by dexamethasone) served as an internal control. Y (yeast RNA only) served as a negative control. (A) Induction of Per1 expression by dexamethasone (Dex). EtOH, solvent treatment. (B) Induction of circadian gene expression by dexamethasone. (C) Injection of the solvent alone does not induce circadian gene expression in vitro.

Next, we wished to determine whether dexamethasone can also affect peripheral oscillators in vivo. To discriminate between direct actions of this glucocorticoid analog on peripheral cell types and indirect actions via the SCN, we examined whether the absence of GR expression in the SCN of rats (16, 17) also applies to mice. To this end, we hybridized mouse coronal brain sections (20) to a GR antisense RNA probe (21). Although strong to moderate signals for GR mRNA can be observed for most brain regions (Fig. 2), few if any silver grains could be detected over the two SCNs (Fig. 2B) irrespective of the time of day at which the mice were killed (22). This suggests that GR mRNA does not accumulate to physiologically important levels in mouse SCN neurons. Hence, the effects of dexamethasone on circadian gene expression in peripheral tissues are unlikely to be mediated via the central SCN pacemaker. This hypothesis is confirmed by the inability of dexamethasone to shift the phase of the SCN (23).

Figure 2

GR mRNA does not accumulate to detectable levels in the SCN. Coronal brain sections containing the SCN were hybridized in situ to a 0.5 Kb 35S-labeled GR antisense RNA probe spanning exons 6 through 9. After hybridization, sections were exposed to film for 1 week (A) and then were coated with Kodak NTB2 autoradiographic emulsion (KODAK Gmbh, Stuttgart, Germany) and exposed for 1 month at 4°C (B). After development of the emulsion, sections were colored with cresyl violet to stain nuclei (C). (A) Autoradiography of a coronal section harvested at ZT6. (B) Dark-field photograph of the in situ hybridized hypothalamus region (hybridization signals appear as reduced silver grains in white). VLTN, ventrolateral thalamic nucleus; PVHN, Paraventricular hypothalamic nucleus. (C) Light-field photograph of the section presented in (B), showing cresyl violet–stained cell nuclei. Similar results were obtained at all other times examined (ZT2, ZT10, ZT14, ZT18, and ZT22) (22). Magnification of (B) and (C) is threefold that of (A).

Next, we measured the effect of dexamethasone on circadian liver gene expression in vivo. Dexamethasone 21-phosphate transiently induces Per1 expression at all examined time points (24). If mice were injected before or during zenith levels (ZTs) at which Per1 mRNA accumulation normally reaches such levels (ZT6 to ZT14), injection resulted in prolonged and/or stronger expression of Per1 (e.g., injections at ZT21 or ZT8). However, if the injections were performed after Per1 mRNA peak expression, a new surge of Per1 expression was induced (see injection at ZT14 in Fig. 3A). Dexamethasone injection also resulted in transient phase shifts. For example, when dexamethasone was injected at ZT14, Per1expression was considerably delayed on the following day (Fig. 3A), although, in these experiments, circadian Per1 gene expression was somewhat confounded by the dexamethasone-induced transient burst of Per1 expression. To determine dexamethasone-induced phase shifts more accurately, we recorded the high-amplitude expression cycles of Dbp andRev-erbα, two genes that are not induced transiently by dexamethasone.

Figure 3

Dexamethasone induces phase shifts in circadian gene expression in the periphery but not in the SCN. Dexamethasone-phosphate (Dex, 400 μg/ml in PBS) or solvent (PBS with 1.5 μl/ml of ethanol) were injected at different times into mice (2 μg/g body weight). Animals were killed after the times indicated. In each experiment, a TBP antisense RNA probe was included as an internal control (22) for a transcript whose cellular concentration remains constant throughout the day. (A) CircadianPer1 expression in liver after dexamethasone injection. Compared with the stimulation of Per1 expression observed in vitro, the activation observed in vivo was delayed by 2 to 4 hours (Fig. 1A) (22). (B) The levels of DBP and Rev-erbα mRNAs were determined by ribonuclease protection experiments. (C) The signals obtained in ribonuclease protection assays for DBP and TBP transcripts were determined by scanning the gels The relative levels of DBP transcripts are given as DBP/TBP mRNA signal ratios on they axis. Dotted lines indicate peak expression; differences indicate phase shift. (D) Dexamethasone-induced phase-shifts in the expression of DBP and Rev-erbα transcripts were recorded after injections at ZT1, ZT3, ZT5, ZT8, ZT10, ZT11, ZT14, ZT18, ZT21, and ZT23. Open circles, DBP mRNA; filled circles, Rev-erbα mRNA. For aesthetic reasons, the values obtained at ZT5 and ZT8 are shown at the beginning and the end of the PRC.

To examine the response of peripheral oscillators around the clock, we injected mice with dexamethasone at 10 time points, and we monitored the temporal DBP and Rev-erbα mRNA accumulation profiles for 36 to 60 hours. The data obtained for four different times of dexamethasone injection (ZT1, ZT5, ZT14, and ZT21) are compared with the data obtained with a control injection of solvent at ZT21 (Fig. 3, B and C). The control injections did not affect circadian gene expression at ZT1, ZT3, ZT5, ZT8, and ZT21 (22). On the basis of these data, we estimated that strong phase delays were observed when dexamethasone was injected at ZT14 and ZT21, whereas injection of dexamethasone at ZT1 resulted in a phase advance (Fig. 3, B and C). In contrast, injection at ZT5 caused little if any change in the phase angle of DBP and Rev-erbα mRNA expression (Fig. 3, B and C). An apparent phase-shift response curve (aPRC) (25) for liver was obtained from these data by plotting the amplitudes of positive and negative phase shifts as a function of the circadian time of the dexamethasone injection (Fig. 3D).

The aPRC for liver oscillators shows an important difference from PRCs established for the SCN pacemaker. The PRCs obtained for the central clock consist of three time windows: a dead zone of about 12 hours during which no resetting is observed, and two windows of about 7 to 5 hours during which phase advances and phase delays are observed (26). In contrast, no extended dead zone can be observed in the aPRC of liver, indicating that the phase of peripheral oscillators can be changed throughout the 24-hour day.

To test whether dexamethasone can act on peripheral oscillators directly by cell-autonomous mechanisms, we examined the effect of dexamethasone on circadian gene expression in mutant mice (GRAlfpCre) with a hepatocyte-specific inactivation of the GR gene (27). These mice do not accumulate GRs in hepatocytes and biliary duct cells, which constitute the vast majority of liver cells (27). Five hours after dexamethasone injection (ZT21), Per1 mRNA accumulation is strongly elevated in kidney and heart of both wild-type and mutant mice (Fig. 4A). However, Per1 expression is only activated in the liver of wild-type mice and remains low in GRAlfpCre mice. Hence, a functional GR is necessary for the activation of Per1 transcription after dexamethasone injection. A similar conclusion can be drawn for dexamethasone-induced phase shifting. DBP mRNA accumulation was delayed by about 3 to 4 hours in liver, kidney, and heart of dexamethasone-injected wild-type mice and in kidney and heart of dexamethasone-injected GRAlfpCremice (Fig. 4B). However, the phase of DBP mRNA accumulation was not affected in the liver of these mutant animals. Thus, the dexamethasone-mediated phase shifting of peripheral oscillators is a tissue-autonomous process.

Figure 4

The GR is required for dexamethasone-inducedPer1 expression and phase shifting. (A) Wild-type mice or GRAlfpCre mice with a liver-specific disruption of the GR gene were injected at ZT21 with dexamethasone-phosphate (Dex) or solvent as described in the legend to Fig. 3. Per1 and TBP mRNA accumulations were determined 5 hours after injection in liver, kidney, and heart. The low level of Per1 transcripts in the liver of the mutant mouse is probably contributed by nonparenchymal liver cells (fibroblasts, Kupffer cells, endothelial cells, etc). (B) Animals were treated as in (A), and DBP and TBP mRNA levels were determined in liver, kidney, and heart at different times.

The rhythmic secretion of glucocorticoids and their ability to phase shift peripheral clocks makes them valid candidates for signals establishing the link between the SCN pacemaker and peripheral oscillators. Yet, the phases of accumulation cycles for the mRNAs encoding DBP (Fig. 4B) and Per1, Per2, Per3, Cry1, and Rev-erbα (22) are the same in livers of GRAlfpCre mutant and wild-type mice. Therefore, glucocorticoids cannot be the only signals setting the phase of peripheral clocks. That these hormones do, however, play a role in the entrainment of peripheral oscillators is suggested by recent experiments in which peripheral oscillators were uncoupled from the SCN pacemaker by restricted feeding (28).

By exploring glucocorticoid signaling, which does not affect the central circadian pacemaker in the SCN, we have determined that peripheral oscillators can be phase delayed or phase advanced during the entire 24-hour day. This phase-shifting behavior would be expected for slave oscillators that are synchronized by a master pacemaker because they should remain responsive to phase-resetting signals from the SCN at all times.

  • * Present address: Center for Neurobiology and Behavior, 722 West 168th Street, Research Annex, New York, NY 10032, USA

  • To whom correspondence should be addressed. E-mail: ueli.schibler{at}


Stay Connected to Science

Navigate This Article