Nuclear Receptor Rev-erbα Is a Critical Lithium-Sensitive Component of the Circadian Clock

See allHide authors and affiliations

Science  17 Feb 2006:
Vol. 311, Issue 5763, pp. 1002-1005
DOI: 10.1126/science.1121613


Lithium is commonly used to treat bipolar disorder, which is associated with altered circadian rhythm. Lithium is a potent inhibitor of glycogen synthase kinase 3 (GSK3), which regulates circadian rhythm in several organisms. In experiments with cultured cells, we show here that GSK3β phosphorylates and stabilizes the orphan nuclear receptor Rev-erbα, a negative component of the circadian clock. Lithium treatment of cells leads to rapid proteasomal degradation of Rev-erbα and activation of clock gene Bmal1. A form of Rev-erbα that is insensitive to lithium interferes with the expression of circadian genes. Control of Rev-erbα protein stability is thus a critical component of the peripheral clock and a biological target of lithium therapy.

Genetic and biochemical analysis reveals that a 24-hour circadian rhythm is present throughout the animal kingdom (13). In mammals, circadian rhythm is a fundamental regulatory factor for many aspects of behavior and physiology, including sleepwake cycles, blood pressure, body temperature, and metabolism (13). Disruption in circadian rhythms leads to increased incidence of many diseases, such as cancer and mental illness (1, 3). Bipolar disorder in particular is associated with disturbed circadian rhythm (4).

Cells throughout the body also display 24-hour rhythms (3, 5). These are entrained by signals from a central clock located in the suprachiasmatic nucleus (SCN) of the hypothalamus, which is reset daily by light (3). Cellular rhythms are generated and maintained through interconnected transcriptional feedback of clock genes (3, 6). The cycle starts when two bHLHPAS domain proteins, BMAL1 and CLOCK, heterodimerize to activate a number of clock genes including Per1, Per2, Cry1, and Cry2.As a negative feedback loop, PER and CRY accumulate in the cytosol and then translocate into the nucleus. Once inside the nucleus, the PER-CRY complex inhibits its own transcription by binding to BMAL-CLOCK (3, 68). An additional negative feedback loop requires the transcription repression function of the orphan nuclear receptor Rev-erbα, which represses the transcription of Bmal1 during circadian night and is responsible for rhythmic expression of the Bmal1 gene (911). Rev-erba itself is activated by BMAL1-CLOCK and thereby represents the link between the positive and negative loops of the circadian clock (9).

Posttranslational modifications also play an essential role in resetting the clock (2, 3, 12). Phosphorylation of PER by casein kinase Iϵ leads to its ubiquitination and proteasomal degradation and, therefore, controls the period length in mammals as well as Drosophila (13). Mutation in Shaggy, the Drosophila homolog of glycogen synthase kinase 3β (GSK3β), lengthens the circadian period (12, 14), similar to the mammalian effects of lithium (15, 16), a potent and selective GSK3 inhibitor (17). We observed that the amino terminus of Rev-erbα is serine-rich, with several potential GSK3β phosphorylation sites (fig. S1A). Remarkably, in human 293T embryonic kidney cells, reduced expression of GSK3β by small interfering RNA (siRNA) led to a near complete loss of endogenous Rev-erbα protein (Fig. 1A), as well as ectopically expressed Flag epitope–tagged Rev-erbα (fig. S2). This effect was posttranscriptional, as Rev-erba mRNA was increased dramatically by loss of GSKβ (Fig. 1B), consistent with the known function of Rev-erbα protein to repress its own gene expression (18). Bmal1, the key circadian target of Rev-erbα, was also markedly induced in cells lacking GSK3β (Fig. 1B). A similar reduction in Rev-erbα protein was observed when a dominant-negative form of GSK3β was introduced into HepG2 cells by adenovirus-mediated delivery (fig. S3). These results suggest that GSK3β activity is required for stabilization of Rev-erbα protein.

Fig. 1.

GSK3β regulates Rev-erbα protein amount and function. (A) Immunoblots of extracts from human 293T cells transfected with siRNA vector for either β-galactosidase (β-gal) control or human GSK3β. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as loading control. (B) 293T cells transfected with β-gal or GSK3β siRNA vector were analyzed for Rev-erba and Bmal1 mRNA as described in methods (in the SOM). Shown is the mean ± SD of three experiments. *P < 0.05 versus control siRNA. (C) Immunoblots of extracts of mouse NIH3T3 cells subjected to serum shock (exposed to medium containing 50% horse serum at time 0 then switched to medium containing 0.5% bovine serum 2 hours later). (D) Chromatin immunoprecipitation (ChIP) for Rev-erbα and the corepressors N-CoR/SMRT at the mouse Bmal1 promoter in serum-shocked NIH3T3 cells. Rabbit IgG was used as nonspecific control. (E) Effect of serum shock on Bmal1 and Reverba mRNA in NIH3T3 cells.

Next, we asked whether modulating GSK3β activity influences Rev-erbα–mediated clock gene regulation. GSK3β is a constitutively active kinase that is inhibited by phosphorylation on serine 9 by multiple signaling pathways (19). We found that serum shock, which synchronizes circadian oscillations in cultured NIH3T3 mouse fibroblasts (5, 20), led to the immediate and robust phosphorylation of GSK3β at serine 9 (Fig. 1C). Remarkably, the level of Rev-erbα protein plummeted during this time and recovered when the cells were returned to serum-free medium and GSK3β phosphorylation abated. The changes in the cellular Rev-erbα protein were reflected by the cyclic occupancy of the Bmal1 gene promoter by the nuclear repressor corepressor (N-CoR) that is recruited by Rev-erbα (Fig. 1D), accompanied by transient induction of Bmal1 and Rev-erba mRNAs (Fig. 1E).

GSK3β is also inhibited by lithium (17). Treatment of 293T cells with 20 mM LiCl dramatically down-regulated ectopically expressed Flag epitope–tagged Rev-erbα without reducing the mRNA level (Fig. 2A; fig. S4). By contrast, endogenous β-catenin was stabilized by LiCl as expected (21). The destabilizing effect of lithium on Rev-erbα was prevented by treatment of the cells with MG132, an inhibitor of the 26S proteasome (Fig. 2A). Inhibition of proteasome activity also prevented loss of Reverbα due to siRNA inhibition of GSK3β (fig. S2A). Furthermore, we detected polyubiquitination of Flag–Rev-erbα (Fig. 2B), indicating that inhibition of GSK3β targets Rev-erbα for degradation by the ubiquitin-dependent proteasome pathway. Lithium treatment also reduced the association of Rev-erbα and the corepressor complex with the Bmal1 promoter, while increasing histone acetylation (Fig. 2C) and gene expression (Fig. 2D).

Fig. 2.

Lithium reduces Rev-erbα protein amount and function. (A) Immunoblots of extracts from 293T cells transfected with Flag epitope–tagged Rev-erbα, and exposed to 20 mM LiCl and/or MG132. β-Catenin served as positive control for lithium effect on inhibition of GSK3β, and Ran guanosine triphosphatase (Ran GTPase) served as loading control. (B) Immunoprecipitation with Flag-specific antibody was performed from 293T cell extracts expressing Flag epitope–tagged Rev-erbα treated with or without MG132. The ubiquitin-conjugated Reverbα protein was detected by immunoblotting. (C) ChIP assay comparing the occupancy of endogenous Rev-erbα, N-CoR/SMRT, HDAC3, and acetylated histone (Ac-H3) at the human Bmal1 promoter of 293T cells treated with or without LiCl (20 mM) for 16 hours. (D) Effect of LiCl (20 mM, 16 hours) on Bmal1 gene expression in 293T cells. (E) Effect of LiCl (1 mM, 72 hours) on Reverbα levels in 293T cells. Heat shock protein Hsp90 served as loading control. (F) Effect of LiCl (1 mM, 72 hours) on Bmal1 gene expression in 293T cells. For RNA analyses, shown are the mean ± SD of three experiments. *P < 0.05 versus control treatment.

Although we have used lithium at a concentration of 20 mM to maximally inhibit GSK3β, chronic lithium therapy for patients with bipolar disorder aims for a serum concentration of ∼1 mM (22). We therefore treated 293T cells with 1 mM LiCl for 72 hours and observed a marked reduction of Rev-erbα protein (Fig. 2E) and induction of Bmal1 gene expression (Fig. 2F). Thus, degradation of Rev-erbα occurs at a clinically relevant concentration of lithium.

Within the Rev-erbα N terminus, serine 55 and serine 59 are located in a GSK3β consensus site that is identical in human, mouse, and rat (fig. S1B). Mutation of both amino acids to negatively charged aspartate (S55D/S59D, here shortened to 55/59SD) stabilized the protein (fig. S5). The 55/59SD mutant had a longer half-life and was resistant to lithium-induced degradation (Fig. 3, A and B). An in vitro kinase assay comparing the wild type (WT) and the 55/59SD mutant as substrate for GSK3β confirmed that these two serine residues were required for phosphorylation of Rev-erbα by GSK3β (Fig. 3C).

Fig. 3.

Rev-erbα is stabilized by GSK3β-mediated phosphorylation of serines 55 and 59. (A) Immunoblots of 293T cells transfected with Flag-tagged Rev-erbα (WT or 55/59SD mutant), exposed to 25 μg/ml cycloheximide (CHX) for 2 hours. (B) Immunoblots of 293T cells transfected with Flag-tagged Rev-erbα and treated with or without LiCl (20 mM) and MG132 (1μM). (C) In vitro phosphorylation of Rev-erbα by GSK3β. HeLa cells were stably transfected with expression vector for WT or 55/59SD Flag-tagged Rev-erbα. Flag-tagged Rev-erbα protein was immunoprecipitated and incubated with [γ-32P]ATP (adenosine triphosphate) and recombinant GSK3β as in the SOM. Proteins were resolved on 12% SDS–polyacrylamide gel electrophoresis (SDS-PAGE) gel and analyzed by autoradiography (top) and immunoblot (bottom).

We hypothesized that early inactivation of GSK3β, causing Rev-erbα degradation and leading to Bmal1 induction, is a critical step for synchronization of rhythmic expression of clock genes in NIH3T3 cells. To test this, we established stable NIH3T3 cell lines expressing green fluorescent protein (GFP, control), WT Rev-erbα, or 55/59SD Rev-erbα. Bmal1 was induced by serum shock in GFP and WT cells, but not in cells expressing the 55/59SD form of Rev-erbα (Fig. 4A). Moreover, Bmal1 induction by lithium treatment was also absent in cells expressing the 55/59SD mutant (Fig. 4B). Lithium treatment also caused a significant change in the expression pattern of the clock gene Per2 and the circadian output gene Dbp (23) in GFP and WT Rev-erbα–expressing cells, but not in cells expressing the Rev-erbα 55/59SD mutant (fig. S6).

Fig. 4.

Control of circadian gene expression by regulation of Rev-erbα protein stability. NIH3T3 cells stably expressing GFP, WT, or 55/59SD Rev-erbα were established. (A) Induction of mouse Bmal1 gene expression in cells exposed to 50% horse serum for 2 hours. (B) Bmal1 gene expression in cells exposed to 20 mM LiCl for 16 hours. (C) Bmal1 gene expression in cells exposed to 50% horse serum for 2 hours, then switched to 0.5% horse serum for 72 hours. Shown are the mean ± SD of three experiments. *P < 0.05 versus control treatment.

We performed further analysis to determine whether Rev-erbα degradation is required for the generation and maintenance of oscillatory gene expression over several circadian cycles. Remarkably, expression of the degradation-resistant 55/59SD form, but not WT Rev-erbα, severely dampened the oscillatory expression of Bmal1 over three circadian cycles following serum shock (Fig. 4C). Thus, GSK3β-dependent regulation of Rev-erbα is important for synchronizing and maintaining the peripheral clock.

The GSK3β homolog Shaggy regulates the length of the circadian period in Drosophila (14), and mammalian GSK3β enzymatic activity oscillates with a 24-hour period both in SCN and liver (16, 24). Moreover, a recent report identified a single nucleotide polymorphism within the GSK3β promoter that is associated with age at onset in bipolar depression (25). Circadian targets of GSK3β are beginning to be elucidated, with a recent study showing that GSK3β affects nuclear entry of mPER2 (16). Our results demonstrate that GSK3β-dependent stabilization of Rev-erbα maintains Bmal1 in a repressed state and, more importantly, we have shown that the inability to degrade Rev-erbα is sufficient to prevent the onset of circadian gene oscillation. One or more components in serum that mimic external endocrine and/or neural cues induce phosphorylation and inactivation of GSK3β, resulting in the degradation of Reverbα and initiating the cycle of gene expression (fig. S7).

Lithium influences the circadian clock in humans, and circadian rhythms are altered in patients with bipolar disorder for which lithium is a common therapy (26). Here, we have shown that degradation of Rev-erbα is a critical target for lithium regulation of circadian gene expression. Intriguingly, histone deacetylase activity of the N-CoR/HDAC3 (histone deacetylase 3) corepressor is inhibited by valproic acid (27, 28), another mood stabilizer that modulates circadian rhythm (29), which suggests that this therapy may also target repression of clock genes by Rev-erbα. Given the toxicity of existing therapies, we suggest that novel approaches targeting Rev-erbα degradation may have potential in the treatment of bipolar and circadian disorders.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article