Expression of the gene for Dec2, a basic helix–loop–helix transcription factor, is regulated by a molecular clock system

Abstract

Dec2, a member of the basic helix–loop–helix superfamily, is a recently confirmed regulatory protein for the clockwork system. Transcripts of Dec2, as well as those of its related gene Dec1, exhibit a striking circadian oscillation in the suprachiasmatic nucleus, and Dec2 inhibits transcription from the Per1 promoter induced by Clock/Bmal1 [Honma, Kawamoto, Takagi, Fujimoto, Sato, Noshiro, Kato and Honma (2002) Nature (London) 419, 841–844]. It is known that mammalian circadian rhythms are controlled by molecular clockwork systems based on negative-feedback loop(s), but the molecular mechanisms for the circadian regulation of Dec2 gene expression have not been clarified. We show here that transcription of the Dec2 gene is regulated by several clock molecules and a negative-feedback loop. Luciferase and gel retardation assays showed that expression of Dec2 was negatively regulated by binding of Dec2 or Dec1 to two CACGTG E-boxes in the Dec2 promoter. Forced expression of Clock/Bmal1 and Clock/Bmal2 markedly increased Dec2 mRNA levels, and up-regulated the transcription of the Dec2 gene through the CACGTG E-boxes. Like Dec, Cry and Per also suppressed Clock/Bmal-induced transcription from the Dec2 promoter. Moreover, the circadian expression of Dec2 transcripts was abolished in the kidney of Clock/Clock mutant mice. These findings suggest that the Clock/Bmal heterodimer enhances Dec2 transcription via the CACGTG E-boxes, whereas the induced transcription is suppressed by Dec2, which therefore must contribute to its own rhythmic expression. In addition, Cry and Per may also modulate Dec2 transcription.

INTRODUCTION

A variety of organisms have circadian rhythms that control daily rhythms of physiology and behaviour, enabling them to adapt to recurring environmental conditions [1–3]. In mammals, the SCN (suprachiasmatic nucleus) of the hypothalamus acts as the master pacemaker that is essential for the generation of circadian rhythms and entrainment to the 24 h day. Light signals are perceived by the retina and transmitted to the SCN via the retinohypothalamic tract; synchronized oscillators in the SCN are transduced to peripheral oscillators through the output pathway.

Molecular and genetic studies have revealed that transcriptional regulation of multiple clock genes is crucial for the generation of circadian rhythms [4,5]. The molecular clock is composed of autoregulatory feedback loops containing both positive and negative components. Clock and Bmal1, bHLH (basic helix–loop–helix) and PAS domain-containing transcription factors, form a heterodimer which activates transcription of the Per and Cry genes by binding to their CACGTG E-boxes [6,7]. A Per/Cry heterodimer suppresses its own transcription by interacting directly with Clock/Bmal1, although Per/Cry does not itself have DNA-binding capacity [8]. Thus the feedback loop generates rhythmic expression of Per, Cry and other genes. In addition to this core loop, several transcription factors, which are expressed in a circadian fashion in the SCN and in the peripheral tissues, are reported to affect clock gene transcription, suggesting that multiple feedback loops interact with the core loop. Rev-Erbα, for example, an orphan nuclear receptor, is up-regulated by Clock/Bmal1 through CACGTG E-boxes in its promoter [9], and the increased levels of Rev-Erbα protein repress Bmal1 transcription by binding to its response elements in the Bmal1 promoter [9,10]. The existence of such an interlocked feedback loop may be necessary for the maintenance of stable and precise circadian rhythms.

Dec1 (also called Stra13 [11] or Sharp-2 [12]) and Dec2 are structurally related to the Hes and Hey family proteins [13,14], with Sharp-1 being a minor or artificial frame-shift mutant of Dec2 [12,14]. While the Dec, Hes and Hey families share similar bHLH and Orange domains, Dec (unlike Hes and Hey) lacks the C-terminal WRPW or YRPW motif. We recently identified Dec1 and Dec2 as circadian regulatory genes [15,16]. Transcripts of Dec2 and Dec1 in the SCN oscillated in a circadian fashion under conditions of LD (light–dark) and DD (constant darkness), with peaks early in and in the middle of the subjective day respectively [15]. The phase of circadian rhythms for Dec2 and Dec1 was similar to that for Per1, Per2, Per3 [17], Cry1 [18], Dbp [19] and Rev-Erbα [9]. Since Dec2 and Dec1 repressed Clock/Bmal1-induced transactivation of the Per1 promoter through direct protein–protein interaction and/or competition for E-boxes [15], Dec2 is probably a component of the mammalian clock system.

In the present study, we characterized the promoter of the Dec2 gene in order to clarify the transcriptional mechanisms involved in the circadian-dependent expression of this gene. We show that the transcription of Dec2 is up-regulated by Clock/Bmal through two CACGTG E-boxes in the Dec2 promoter, and down-regulated by its own product through binding to the E-boxes. Furthermore, Cry and Per suppressed the Clock/Bmal1-mediated transactivation of the Dec2 promoter. These findings suggest that Dec2 expression is controlled by the clock genes, and that the autoregulatory feedback loop of Dec2 transcription works together with the core loop to control circadian-dependent gene expression in mammals.

EXPERIMENTAL

Cell culture

NIH3T3 cells and C2C12 cells were supplied by the cell bank of the Institute of Physical and Chemical Research (Tsukuba, Japan) and maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum, penicillin (100 units/ml), streptomycin (100 μg/ml) and amphotericin (250 ng/ml). Human umbilical vein endothelial cells were obtained from Bio-Whittaker Inc. (Walkersville, MD, U.S.A.) and cultured in M199 medium supplemented with 20% (v/v) fetal bovine serum, 60 μg/ml endothelial cell growth supplement (Collaborative Biomedical, Bedford, MA, U.S.A.) and 50 mg/ml heparin, as described previously [20].

Isolation of the mDec2 (mouse Dec2) gene

A mouse genomic DNA bacterial artificial chromosome library was screened by PCR using mDec2 specific primers: 5′-ACAGGACAGAAACCTCCAAATC-3′ and 5′-TCTTTCAGCTGAGCAATGCATTC-3′. Positive genomic clones were analysed further for restriction mapping by Southern blot analysis. A 12 kb HindIII fragment which hybridized to the mDec2 cDNA probe was subcloned into the HindIII site of pBluescript (Stratagene, La Jolla, CA, U.S.A.). The entire nucleotide sequence was determined by using the BigDye terminator cycle sequencing kit with an ABI Prime 310 DNA sequencer (both from PE Applied Biosystems, Redwood, CA, U.S.A.). The fragment contained the complete Dec2 coding region constituting five exons and 8 kb of the upstream region.

Luciferase reporter plasmid constructions

A 3.2 kb 5′-upstream fragment (−3171 to −83; +1 indicates the translation initiation site) of the mDec2 gene was amplified by PCR using a forward primer (5′-GGATCCACTGAACCATCTCTCCAACCCTAA-3′) and a reverse primer (5′-GGATCCGTGCGTCTCCAGGCTGTCTCGCTCT-3′), and ligated to the pGEM-T Easy vector (Promega, Madison, WI, U.S.A.). After confirming the sequence, the fragment was then subcloned into the BglII site of the promoter-less luciferase reporter plasmid pGL3-Basic vector (Promega), and named m-3171. Deletion constructs (p-1596-Luc, p-1388-Luc, p-795-Luc and p-303-Luc) were derived from m-3171 by digestion with restriction endonucleases (XhoI, ApaI, SmaI and PmaCI respectively), followed by ligation (see Figure 3).

Figure 3. Enhancement of Dec2 promoter activity by Clock/Bmal1 and Clock/Bmal2.

(A) Comparison of nucleotide sequences of mouse and human Dec2 gene upstream regions. Nucleotides are numbered beginning from the translation initiation site. The transcription initiation site of the human Dec2 gene is shown by an arrow. The two CACGTG E-box elements present in both sequences are indicated by underlining. (B) Deletion analysis of mDec2 promoter activity. Schematic diagrams of various Dec2 promoter constructs are shown on the left, and the names of the plasmids are listed in the middle. The locations of the CACGTG E-boxes are indicated by closed boxes. Each construct was co-transfected with expression vectors for Clock (Cl), Bmal1 (Bm1) and/or Bmal2 (Bm2) into NIH3T3 cells. The total amount of transfected DNA was adjusted to a constant value with an empty vector, and promoter activity was normalized to the Renilla luciferase activity of a co-transfected internal control plasmid (phRL-TK). The promoter activity of p-1596-Luc in the absence of expression vector is given a value of 1. All data presented are means±S.D. for four different experiments.

pTK-Luc was obtained by subcloning a DNA fragment of the herpes simplex virus TK (thymidine kinase) region from plasmid phRL-TK into the HindIII–BglII site of pGL3-Basic vector. A 53 bp construct which contained three copies of the distal CACGTG E-box site (E-box1) with flanking sequence linked in tandem was made by annealing oligonucleotides 5′-CTAGTCCCGGCACGTGACCCGCCCGGCACGTGACCCGCCCGGCACGTGACCCG-3′ and 5′-TCGACGGGTCACGTGCCGGGCGGGTCACGTGCCGGGCGGGTCACGTGCCGGGA-3′. A 52 bp construct which contained three copies of the proximal CACGTG E-box site (E-box2) with flanking sequence linked in tandem was made by annealing oligonucleotides 5′-CTAGTTCCGCACGTGAGCTGTTCCGCACGTGAGCTGTTCCGCACGTGAGCTG-3′ and 5′-TCGACAGCTCACGTGCGGAACAGCTCACGTGCGGAACAGCTCACGTGCGGAA-3′. These fragments were ligated into the NheI and XhoI sites of pTK-Luc upstream of the TK promoter (pE1-TK-Luc and pE2-TK-Luc).

Expression plasmid constructions

Cry1 and Per1 expression vectors were generously provided by M. Ikeda (Research Center for Genomic Medicine, Saitama Medical School, Saitama, Japan) and H. Tei (Laboratory of Functional Genomics, Human Genome Center, Institute of Medical Science, University of Tokyo, Tokyo, Japan) [21]. Expression vectors for Clock, Bmal1, Per2, Cry2 and Bmal2 were described previously [15,22]. The coding region of mDec2 was obtained by RT-PCR (reverse transcription–PCR) with primers 5′-GGATCCAGCCATTGAACATGGACGAAGGAAT-3′ and 5′-GGCACGCTTTAGAGGACGTTTGAA-3′, and subsequently subcloned into the pcDNA3.1 expression vector (Invitrogen, Carlsbad, CA, U.S.A.). The coding region of mDec1 was obtained by RT-PCR with primers 5′-CTGCTCGCCGCATCATGGAACGGAT-3′ and 5′-CCAGAGTTTAGTCTTTGGTTTCTAAG-3′, and subcloned into the pcDNA3.1 vector. All constructs were verified by sequence analysis.

Transient transfection and luciferase assay

NIH3T3 cells were seeded at 1×10⁴ cells/well in 24-well plates, and transfected with plasmids on the next day. The phRL-TK vector (0.5 ng) was co-transfected for normalization, and the total amount of DNA per well was adjusted by adding pcDNA3.1 vector. At 24 h after transfection, cells were harvested to determine luciferase activity using the Dual-Luciferase Reporter Assay System (Promega). All experiments were repeated at least twice, and the results from representative experiments (n>3) are shown with S.D.s.

Electrophoretic mobility shift assay

mDec1 and mDec2 were synthesized using the TNT Coupled Reticulocyte Lysate System (Promega). For preparation of the probes, the E-box1 oligonucleotides (5′-AGGCTCCCGGCACGTGACCCGCT-3′ and 5′-CTGGAGCGGGTCACGTGCCGGGA-3′) and the E-box2 oligonucleotides (5′-GGTACGTTCCGCACGTGAGCTGG-3′ and 5′-GCACCCAGCTCACGTGCGGAACG-3′) were annealed and then end-labelled with [³²P]dCTP using DNA polymerase I Klenow fragment (TAKARA, Kyoto, Japan). The ³²P-labelled probe (10⁴ c.p.m.) was incubated for 20 min at room temperature in a buffer containing 10 mM Tris/HCl (pH 8.0), 50 mM NaCl, 5 mM MgCl₂, 0.5 mM dithiothreitol, 10% (v/v) glycerol and 0.1 μg/μl poly(dI-dC) in the presence of the in vitro-translated products. For competition experiments, the in vitro-synthesized protein was incubated with competitors at 100-fold excess for 15 min at room temperature before adding the probe. The protein–DNA complexes were run on 6% (w/v) polyacrylamide/TBE (Tris/borate/EDTA) gels and visualized by autoradiography.

RNA isolation and Northern blot analysis

Human umbilical vein endothelial cells were infected with adenovirus AdCMV.GFP, AdCMV.CLOCK or AdCMV.CLIF/BMAL2, as described previously [22]. Total RNA was prepared with the RNeasy kit (QIAGEN, Valencia, CA, U.S.A.) 48 h after adenovirus infection and subjected to Northern blot analysis. The probe for human Dec2 cDNA has been described previously [14].

RT-PCR analysis

First-strand cDNA was synthesized using ReverTra Ace (TOYOBO, Osaka, Japan) with 1 μg of total RNA. PCR was performed using an aliquot of first-strand cDNA as a template under standard conditions with Klentaq polymerase (Clontech, Palo Alto, CA, U.S.A.) for 28 cycles for mDec2. For normalization of RNA loading, RT-PCR of GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was also performed in each RT-PCR reaction as an internal control (25 cycles). The pairs of oligonucleotides 5′-ATGCTCGACAGGCTTAGGACA-3′ and 5′-GTGTGAGCTGAGACATGAAAC-3′ for mDec2, and 5′-GCTTCACCACCTTCTTGATG-3′ and 5′-GTCAAGGCCGAGAATGGGAA-3′ for GAPDH, were used as primers for the PCR. The PCR products were separated on 1% (w/v) agarose gels.

Quantitative real-time RT-PCR analysis

Male wild-type and Clock/Clock mutant C57/BL6 mice (2 months old) were maintained under a 12 h/12 h LD cycle for 2 weeks before the day of the experiment. Mice kept under LD or DD conditions were killed by decapitation at different times of the day. Animals were cared for according to the Guidelines for the Care and Use of Laboratory Animals (NIH) in the Hokkaido University Graduate School of Medicine. Total RNA was extracted from the kidneys of the mice (three mice at each time point) at the indicated time points and reverse-transcribed with ReverTra Ace. Quantitative real-time RT-PCR analysis was performed using an ABI PRISM 7900HT Sequence Detection System instrument and software (PE Applied Biosystems, Redwood, CA, U.S.A.) as described [23]. The synthesized first-strand cDNA was amplified using the specific primers 5′-ATTGCTTTACAGAATGGGGAGCG-3′ and 5′-AAAGCGCGCGAGGTATTGCAAGAC-3′ for mDec2. The amplified cDNA was quantified using 6FAM-CGACTTGGATGCGTTCCACTCGG-TAMRA.

RESULTS

Induction of mDec2 mRNA by Clock/Bmal

Both Clock/Bmal1 and Clock/Bmal2 heterodimers act as positive components in the clockwork system, and are involved in the transcriptional regulation of a number of circadian genes. To investigate whether Dec2 gene expression is up-regulated by Clock/Bmal, we overexpressed Clock and/or Bmal2 in endothelial cells using recombinant adenovirus vectors. As shown in Figure 1, co-expression of Clock and Bmal2 induced Dec2 mRNA expression, which robustly increased in a manner dependent on multiplicity of infection, whereas infection with Clock-, Bmal2- or GFP (green fluorescent protein)-expressing virus alone did not induce Dec2 mRNA expression. Thus expression of Clock/Bmal2 was sufficient to stimulate Dec2 gene expression, indicating that Dec2 is a target gene for Clock/Bmal in living cells.

Figure 1. Induction of Dec2 mRNA expression by Clock/Bmal2.

Human umbilical vein endothelial cells were infected with adenovirus expressing GFP, Clock and/or Bmal2 at the indicated multiplicity of infection (M.O.I.). Total RNA was isolated 48 h after infection and subjected to Northern blot analysis using ³²P-labelled human Dec2 cDNA as a probe. The blot was also hybridized with 18 S rRNA to normalize for loading.

To examine further whether Dec2 expression is indeed controlled by Clock/Bmal in vivo, we compared the expression pattern of Dec2 in Clock/Clock mutant mice [24] with that in wild-type mice (Figure 2). In wild-type mice, Dec2 transcripts in the kidney exhibited a robust circadian rhythm, with peaks in the middle of the subjective day under both LD and DD conditions. In contrast, rhythmic Dec2 mRNA expression was severely blunted in Clock/Clock mutant mice under both LD and DD conditions, indicating that the rhythmic expression of Dec2 depends on Clock in vivo.

Figure 2. Reduction of Dec2 mRNA expression in the kidneys of Clock/Clock mutant mice.

Relative mRNA levels of Dec2 in the kidneys of wild-type mice (○, •) and Clock/Clock mutant mice (▵, ▴) were determined by quantitative real-time RT-PCR under LD (left) and DD (right) conditions. The horizontal bar at the bottom of the left panel represents the light (white)/dark (black) cycle. Circadian time indicates the corresponding time in DD: grey and black bars in the right panel represent subjective day and night respectively. All data presented are means±S.D. for three different experiments.

Sequencing and promoter activity of the upstream region in the mDec2 gene

It has been proposed that a CACGTG E-box element is necessary for Clock/Bmal1-mediated transcriptional activation [6]. To characterize the molecular mechanism of Dec2 gene expression, we isolated and sequenced the mDec2 gene, and compared its upstream region with that of the human Dec2 gene (Figure 3A). There was no typical TATA box upstream of the coding region, and alignment of the approx. 1.5 kb mouse upstream region with the corresponding region from the human Dec2 gene revealed some striking similarities, along with some interruption by deletions or insertions. The human Dec2 and mDec2 promoters each contain two CACGTG E-boxes, which are found at similar locations (at −1495 and −306 in the mDec2 gene).

Enhancement of Dec2 promoter activity by Clock/Bmal

To determine whether these putative Clock/Bmal1 binding sites in the Dec2 promoter are functional, we performed reporter gene assays in NIH3T3 cells using a series of 5′ deletions of the mDec2–luciferase constructs. As shown in Figure 3(B), the promoter activity of the mDec2 gene sequence p-1596-Luc was comparable with that of the simian virus 40 promoter. Deletion of the mDec2 gene sequence from −1596 to −1388, or from −1388 to −795, had little effect on promoter activity, but deletion from −795 to −303 resulted in a 75% decrease in promoter activity, indicating the presence of positively acting elements in the region. Similar results were obtained in C2C12 cells and 10T1/2 cells (not shown). We also examined the effects of Clock/Bmal1 or Clock/Bmal2 on Dec2 gene promoter activity. The promoter activities of p-1596-Luc (which contains two CACGTG sequences) and of p-1388-Luc and p-795-Luc (which contain one CACGTG sequence) were up-regulated by both Clock/Bmal1 and Clock/Bmal2 (Figure 3B), whereas Clock, Bmal1 and Bmal2 alone had no effect on the promoter activities (results not shown). Furthermore, Clock/Bmal-induced up-regulation was not observed with the p-303-Luc reporter, which contains no CACGTG sequence.

These results suggest that a proximal E-box (E-box2), at least, is functional for induction of Dec2 transcription. To confirm the involvement of the CACGTG E-box in Clock/Bmal-mediated transactivation, we performed reporter gene assay using pE2-TK-Luc, in which three repeats of E-box2 and its flanking sequence were ligated into a luciferase reporter plasmid upstream of the TK minimal promoter. As shown in Figure 4, Clock/Bmal1 increased the promoter activities of the reporter gene, indicating that Clock/Bmal1 up-regulates Dec2 transcription through the CACGTG E-box.

Figure 4. Inhibition of Clock/Bmal1-induced Dec2 promoter activity by Cry and Per.

pE2-TK-Luc and an expression vector for Cry1, Cry2, Per1 or Per2 were transfected into NIH3T3 cells with or without Clock (Cl) and Bmal1 (Bm1) expression vectors. Promoter activities in the absence of expression vectors are given a value of 1. All data presented are means±S.D. for four different experiments.

Effects of Cry and Per on Dec2 promoter activity

The above findings suggest that the rhythmic expression of Dec2 is positively regulated by Clock/Bmal. Since Clock/Bmal-mediated transcriptional activation of the Per gene is suppressed by Cry1, Cry2, Per1 and Per2 [8], we examined whether these clock gene products also regulate the expression of Dec2. Cry1, Cry2, Per1 or Per2 alone had no effect on the basal promoter activity of pE2-TK-Luc, but Cry1 and Cry2 abolished the Clock/Bmal1-induced promoter activity, and Per1 and Per2 also decreased the Clock/Bmal1-induced promoter activity, but to a lesser extent (Figure 4).

Suppression of Dec2 promoter activity by Dec2 and Dec1

Stra13/Dec1 negatively autoregulates its gene expression [25] and also suppresses transcription from some artificial promoters by binding to CACGTG E-box sequences [26]. However, whether Dec2 negatively autoregulates its gene expression has remained unknown. To address this question, we examined the expression level of endogenous mDec2 mRNA in C2C12 cells by RT-PCR analysis with oligonucleotides located on the 3′ non-coding region of the mDec2 cDNA, which can be distinguished from exogenous Dec2 mRNA. As shown in Figure 5(A), the forced expression of Dec2 down-regulated basal and Clock/Bmal1-induced endogenous mDec2 mRNA expression. We also examined the effects of Dec2 and Dec1 expression on the promoter activity of the mDec2 gene by reporter gene assays. Both Dec2 and Dec1 repressed basal and Clock/Bmal1-induced Dec2 promoter activity in a dose-dependent manner, and Dec2 seemed to be more potent than Dec1 (Figure 5B).

Figure 5. Repression of Dec2 promoter activity by Dec2 and Dec1.

(A) C2C12 cells were transfected with expression vectors for Dec2, Clock (Cl) and/or Bmal1 (Bm1). At 48 h after transfection, total RNA was prepared to examine the expression levels of endogenous Dec2 mRNA by RT-PCR analysis. (B) Dose-dependent repression of Dec2 promoter activity by Dec2 and Dec1. NIH3T3 cells were transfected with 50 ng of the p-1596-Luc construct and the indicated amounts of mDec2 or mDec1 expression vector, with or without Clock (Cl) and Bmal1 (Bm1) expression vectors. The total amount of transfected DNA was adjusted to a constant value with an empty vector. Promoter activities in the absence of expression vectors are given a value of 1. (C) Effects of HDAC inhibitors on Dec2-mediated transcriptional repression. The p-1596-Luc reporter construct was co-transfected with or without 10 ng of Dec2 expression vector. TSA (trichostatin A; 100 nM) or Scriptaid (8 μM) was added 3 h after transfection, and incubation was continued for 21 h. Luciferase activity is expressed as percentage of that in the absence of Dec2. (D) Involvement of E-box sites in Dec2- and Dec1-mediated transcriptional repression. An artificial promoter construct (pE1-TK-Luc or pE2-TK-Luc) containing three repeats of E-box1 or E-box2 was used as a reporter plasmid for the luciferase assay. Expression vectors were co-transfected into NIH3T3 cells. Promoter activities in the absence of expression vectors are given a value of 1. All data presented are means±S.D. for four different experiments; *P<0.02; **P<0.005.

HDAC (histone deacetylase) is often involved in transcriptional repression by bHLH transcription factors, and some actions of Stra13/Dec1 or Sharp-1/Dec2 are suppressed by HDAC inhibitors [25,27]. We therefore examined the effects of the HDAC inhibitors trichostatin A and Scriptaid on Dec2-mediated repression. Treatment with trichostatin A or Scriptaid partly restored the Dec2-mediated repression for the p-1596-Luc reporter, indicating that Dec2 suppresses its own expression at least partly through an HDAC-dependent mechanism (Figure 5C).

Since Dec2 has high structural similarity to Dec1 in the bHLH region, which is involved in DNA binding and homo- and hetero-dimerization, Dec2 might also bind to the CACGTG E-boxes in the Dec2 promoter to repress its own expression. To examine this hypothesis, we performed reporter gene assays using pE1-TK-Luc or pE2-TK-Luc. Both Dec2 and Dec1 repressed the basal and Clock/Bmal1-stimulated promoter activities of these reporter genes, whereas Dec2 and Clock/Bmal1 had little effect on the promoter activity of pTK-Luc, which contains no CACGTG sequence (Figure 5D). Thus the transcription of Dec2 was positively regulated by Clock/Bmal1 and negatively regulated by both Dec2 and Dec1 via common CACGTG E-boxes.

Dec2 and Dec1 bind to E-boxes in the Dec2 promoter

To determine whether Dec2 and Dec1 bind directly to the CACGTG E-box elements of the Dec2 promoter, we carried out an electrophoretic mobility shift assay. Double-stranded oligonucleotides E-box1 (W1) and E-box2 (W2) were radiolabelled and used as probes. Incubation of each probe with in vitro-translated mDec2 yielded a shifted band (Figures 6B and 6C, lane 2). This binding was specific, since the shifted bands were successfully competed by a 100-fold excess of unlabelled probe (Figures 6B and 6C, lane 3), but not by mutated probes M1 and M2 (Figures 6B and 6C, lane 4). We also confirmed that Dec1, as well as Dec2, can bind to these E-box elements (Figures 6B and 6C, lanes 5–7). These findings suggest that Dec2 and Dec1 negatively regulate Dec2 gene expression at least partly by direct binding to the CACGTG E-box in the Dec2 promoter.

Figure 6. Both Dec2 and Dec1 bind to two CACGTG E-box sites in the Dec2 promoter.

(A) Sense-strand sequences of double-stranded oligonucleotides used as probes and competitors in electrophoretic mobility shift assays. The position of each E-box is underlined, and the mutant nucleotides are shown in lower-case letters. ³²P-labelled oligonucleotides encompassing the E-box1 (B) or E-box2 (C) sites were incubated with reticulocyte lysates in the absence (lane 1; negative control) or presence of Dec proteins. For competition assays, a 100-fold excess of unlabelled oligonucleotides was added. W1, wild-type oligonucleotide identical to the E-box1 probe; M1, mutant oligonucleotide of E-box1 probe; W2, wild-type oligonucleotide identical to the E-box2 probe; M2, mutant oligonucleotide of E-box2 probe.

DISCUSSION

We had shown previously that Dec2 mRNA is expressed in the SCN in a circadian fashion [15], and that the phase of the circadian rhythm for Dec2 is similar to that for Per1 and Cry1, whose gene expressions are known to be controlled by the clock-work system. The Clock/Bmal1 heterodimer, a positive regulator of the clock genes, induces expression of Per1 and Cry1, whose protein products form a complex that suppresses Clock/Bmal1-mediated transactivation. In the present study, we showed that Dec2 expression is also regulated by the clock genes. The circadian expression of Dec2 mRNA was abolished in the kidneys of Clock/Clock mutant mice under LD and DD conditions, which is in accordance with our observation in the SCN of Clock/Clock mutant mice [28]. In living cells, Dec2 transcription was up-regulated by Clock/Bmal1 and Clock/Bmal2. Bmal2 is expressed at high levels in some tissues, including blood vessels [29], and binds to Clock to form a heterodimer for gene activation [22]. A luciferase assay showed that Clock/Bmal1 and Clock/Bmal2 increased the promoter activities of pE2-TK-Luc and p-1596-Luc reporters, whereas Clock, Bmal1 and Bmal2 alone had no effect on the promoter activities. These findings suggest that both Clock/Bmal1 and Clock/Bmal2 directly regulate Dec2 gene expression through interactions with CACGTG sequences. In addition, co-transfection of Cry or Per suppressed the Clock/Bmal1-mediated transactivation of the Dec2 promoter.

We also show here that Dec2 could interact with its own promoter through the CACGTG E-boxes, and thereby repress its own expression. These findings, taken together, indicate that Clock/Bmal up-regulates Dec2 expression, and that the increased Dec2 protein product may inversely down-regulate its own expression. Thus the expression of Dec2 is probably controlled by a negative-feedback mechanism, which may contribute to its oscillatory expression, since autofeedback regulation is a common molecular mechanism for the clockwork system in a variety of organisms.

In accordance with the similar circadian profiles of Dec, Per and Cry mRNA levels, the transcriptional regulatory mechanism involved in Dec2 expression may resemble that of Per and Cry expression. All Dec, Per and Cry genes are activated by Clock/Bmal and repressed by their own products through CACGTG E-boxes in their promoter regions. However, Dec2 binds directly to the CACGTG E-box, whereas Per and Cry have no DNA-binding capacity. Dec2 can compete with Clock/Bmal for DNA binding, whereas Per and Cry have to bind to Clock/Bmal in order to inhibit transcriptional activity [5]. In addition, Dec2 can bind to Bmal1 [15], which may also contribute to the repression of Clock/Bmal1-mediated transactivation.

The present study revealed that mDec1/Stra13, as well as mDec2, repressed the promoter activity of the mDec2 gene, with mDec2 apparently more potent than mDec1 in suppressing transcription from the mDec2 promoter. These findings are consistent with a suppression of transcription from the human Dec2 promoter by human Dec1 [30]. In addition, Stra13/Dec1 and Sharp-1/Dec2 suppressed Dec1 promoter activity [31,32]. Thus Dec1 and Dec2 not only repress their own transcription, but each also mutually represses the expression of the other. Accordingly, mice lacking Stra13/Dec1 showed up-regulation of Dec2/Sharp-1 [33].

Dec2 and Dec1 show a similar circadian rhythm in the SCN and other tissues [15,34]. It has recently been reported that Dec1/Stra13 gene expression is up-regulated by Clock/Bmal, and that the induced activity is suppressed by Dec1 and Dec2 [31,33]. Thus the circadian expression of both Dec2 and Dec1 seems to be controlled by similar mechanisms. In addition, the Dec proteins have a similar bHLH domain, which may be required for binding to CACGTG E-boxes. These findings show a functional redundancy between these family members in the circadian system. In fact, mice lacking Stra13/Dec1 exhibited up-regulation of Dec2/Sharp-1 and showed no significant changes in the expression patterns of clock genes, although these mice did show changes in expression levels of some genes, including several clock-controlled genes [33]. It is likely that Dec2 can compensate for the function of Dec1; however, Dec1 expression is enhanced by a light pulse of 30 min in a phase-dependent manner, while Dec2 expression is not [15], suggesting that the expression of these genes is regulated by different mechanisms in circadian light entrainment.

Besides being circadian clock genes, Dec(s) play roles in cell proliferation and/or differentiation. mDec1/Stra13, for example, inhibits cell proliferation and serum deprivation-induced apoptosis [25,35]; overexpression of mDec1/Stra13 induces differentiation of nerve cells [11] and chondrocytes [36,37]; and mDec1/Stra13-deficient mice exhibit ineffective elimination of activated T and B cells [38]. Although it is unclear how these functions are associated with the rhythmical expression of Dec genes, a number of genes are controlled by circadian genes in a tissue-specific manner, and numerous biological phenomena, including cell proliferation and apoptosis, are associated with circadian rhythms. For example, Per2-mutant mice, which are deficient in circadian clock function, show cancer formation with an increase in c-myc expression, suggesting that Per2 plays a part in tumour suppression [39]. Interestingly, mDec1/Stra13 also represses the expression of c-myc [25], and is expressed abundantly in carcinomas [30,35], whereas Dec2 is highly expressed in adjacent normal tissue [30]. Further studies are needed in order to establish the relationship between the circadian control of Dec gene expression and cell proliferation/differentiation. The identification of downstream target genes and analysis of mice lacking Dec2 should help in determining the physiological functions of Dec2. Furthermore, analysis of mice lacking both Dec1 and Dec2 will be required to elucidate the role of Decs in circadian rhythms.