Circadian Regulation of cAMP Response Element-mediated Gene Expression in the Suprachiasmatic Nuclei*

A program of stringently-regulated gene expression is thought to be a fundamental component of the circadian clock. Although recent work has implicated a role for E-box-dependent transcription in circadian rhythmicity, the contribution of other enhancer elements has yet to be assessed. Here, we report that cells of the suprachiasmatic nuclei (SCN) exhibit a prominent circadian oscillation in cAMP response element (CRE)-mediated gene expression. Maximal reporter gene expression occurred from late-subjective night to mid-subjective day. Cycling of CRE-dependent transcription was not observed in other brain regions, including the supraoptic nucleus and piriform cortex. Levels of the phospho-active form of the transcription factor CREB (P-CREB) varied as a function of circadian time. Peak P-CREB levels occurred during the mid- to late-subjective night. Furthermore, photic stimulation during the subjective night, but not during the subjective day, triggered a marked increase in CRE-mediated gene expression in the SCN. Reporter gene experiments showed that activation of the p44/42 mitogen-activated protein kinase signaling cascade is required for Ca2+-dependent stimulation of CRE-mediated transcription in the SCN. These findings reveal the CREB/CRE transcriptional pathway to be circadian-regulated within the SCN, and raise the possibility that this pathway provides signaling information essential for normal clock function.

In mammals, the suprachiasmatic nuclei (SCN) 1 of the hypothalamus contain a circadian oscillator that functions as the major biological clock (1)(2)(3). The biorhythm generated by the SCN allows an organism to predict and coordinate its daily physiological processes to an approximate 24-h period. If the SCN are lesioned, there is a loss of physiological and behavioral circadian rhythms (4,5). The most effective regulator of the endogenous clock is light; endogenous clock rhythmicity is entrained to the environmental light cycle by photic cues conveyed from the eyes to the SCN via the retinohypothalamic tract (RHT) (6). For example, if an animal receives a light flash during the dark phase of the day/night light cycle, the circadian rhythm is reset, or phase-shifted (7). Light-induced phaseshifting results from the synaptic release of glutamate from the RHT onto the SCN (8 -10). Entrainment of the clock by light is thought to involve changes in gene expression. In support of this, several studies have shown that immediate early gene induction is triggered by light (11)(12)(13).
Recent work has revealed important information about specific proteins and transcriptional events essential for circadian rhythmicity. For example, CLOCK and BMAL1 proteins heterodimerize to form a transcription factor that binds the E-box enhancer element, resulting in mper1 gene expression (14). In Drosophila, PER/TIM dimers negatively regulate CLOCKdependent transcription (15), thus forming a negative feedback loop. Mutations of any of these genes disrupts circadian rhythmicity (16 -21), suggesting that this transcriptional loop is essential for normal clock function. Given these results, it is likely that the E-box enhancer element regulates the rhythmic expression of other genes within the SCN. However, the transcriptional activation of several circadian-regulated genes (vasopressin, brain-derived neurotrophic factor, Fos; Refs. [22][23][24] requires complex interactions of several different classes of enhancer elements (25)(26)(27), suggesting the involvement of different transcription pathways in circadian gene regulation within the SCN. The elucidation of these pathways will provide valuable insight into the series of coordinated transcriptional events underlying circadian rhythmicity.
Ostensibly, transcriptional pathways that contribute to SCN rhythmicity should have the capacity to integrate signaling information from a variety of stimuli, as well as possess properties that allow for its stringent regulation. One candidate is the CREB/CRE transcriptional pathway. This pathway has been shown to be activated by multiple kinases, including protein kinase A (PKA), Ca 2ϩ /calmodulin-dependent kinase, and mitogen-activated protein kinase (MAPK) (28 -31). The CREB/CRE transcriptional pathway also has the capacity to integrate the activation of multiple signaling pathways and the strength of signal into striking variations in downstream gene transcription (32)(33)(34)(35). Furthermore, CRE-mediated transcription can be rapidly repressed through a myriad of mechanisms, including inducible early cAMP repressor induction, phosphatase activation, or as a result of CREB heterodimerization with inhibitory transcription factors (32,36). These unique functional properties led us to explore whether this transcriptional pathway plays a role in circadian rhythmicity. Toward this end, we used a mouse CRE-␤-galactosidase transgenic reporter strain to monitor CRE-mediated transcription in vivo.

EXPERIMENTAL PROCEDURES
Treatments-A description of the generation of the CRE-␤-galactosidase reporter mouse strain is found in Ref. 33 dark (L/D) cycle for at least 7 days before being transferred to constant darkness (D/D). Light intensity during L was ϳ400 lux. Food and water were available ad libitum throughout the experiment. Circadian activity rhythms were monitored using motion sensors connected to Vital-View software (Mini-Mitter Co., Sunriver, OR).
For immunohistochemical analysis of CRE/␤-galactosidase rhythmicity, mice (n ϭ 3-6/time point) kept in D/D for 6 days were sacrificed at 4-h intervals over a 24-h period. For the Western analysis of CRE/ ␤-galactosidase rhythmicity, mice (4/time point) kept in D/D for 6 days were sacrificed at 6-h intervals over a 24-h period. For photic stimulation experiments, animals were exposed to light (400 lux) at CT 16.5, CT 22.5, or CT 6. Mice were then returned to constant darkness. Eight hours after light treatment, animals were sacrificed and brain slices were processed for ␤-galactosidase immunoreactivity.
Tissue Collection-Cervical dislocation followed by decapitation allowed for the rapid removal of the brain, which was immediately placed in ice-cold oxygenated Dulbecco's modified Eagle's medium (pH 7.4) and cut into 400-m coronal sections with a vibratome. For P-CREB experiments, animals were sacrificed and their brains were removed under dim red illumination Ͻ 10 lux. For immunohistochemistry, sections were placed in a 6% formaldehyde/phosphate-buffered saline (PBS) solution for 4 -6 h at room temperature. Sections were then cryoprotected with 30% sucrose for at least 12 h. Thin (35-40 m) sections were cut through the SCN using a sliding microtome. For Western analysis, 400-m coronal brain sections were quick-frozen onto glass coverslips. The SCN and lateral hypothalamus were then excised with the use of a dissecting microscope. Tissue was stored at Ϫ70°C.
Immunohistochemistry-For ␤-galactosidase immunolabeling, freefloating sections were blocked for 2 h in 1% normal goat serum and 10% bovine serum albumin in PBS with 0.1% Triton X-100 (PBST). After blocking, sections were incubated overnight at 4°C with an affinitypurified polyclonal ␤-galactosidase antibody raised in rabbit (1:1000 final dilution, Cappel) in PBST, and 2.5% bovine serum albumin. The tissue was then incubated for 6 h with a lissamine-rhodamine-conjugated secondary antibody raised in goat and directed against rabbit IgG (2 g/ml final dilution, Jackson Laboratories) in PBST containing 2.5% bovine serum albumin. To ensure minimal immunolabeling variability for endogenous rhythm experiment, all samples were processed concurrently.
For immunolabeling against the Ser-133 phosphorylated form of CREB, free-floating sections were initially blocked as described above, except that PBST also contained NaF (1 mM). Tissue was then incubated overnight with phospho-specific CREB antibody (1:500, New England Biolabs). On Western blots, this affinity-purified antibody specifically recognizes the phosphorylated form of CREB and the phosphorylated forms of CREB-related proteins ATF-1 and CREM. The tissue was then incubated for 6 h with a fluorescein-conjugated secondary antibody raised in goat and directed against rabbit IgG (4 g/ml final dilution, Cappel), then stained with an Alexa-488-conjugated antifluorescein antibody raised in rabbit (2 g/ml final dilution, Molecular Probes). Data obtained with this technique were confirmed by an alternative technique that employed a biotin-linked secondary antibody directed against the IgG domain of the P-CREB antibody, followed by the administration of a streptavidin-conjugated lissamine-rhodamine fluorophore. Although results with this technique were consistent with those obtained using the Alexa-488 tertiary antibody, we noticed a high level of nonspecific labeling. Trial experiments showed that the extraordinary sensitivity of the ABC immunostaining technique (Vector Laboratories) did not allow for subtle differences in the levels of P-CREB immunoreactivity to be observed. All samples were processed concurrently.
For immunolabeling against map-2, tissue was incubated with a mouse monoclonal anti-map-2 antibody (1:1000 final dilution, Sigma), then labeled with Cy-2-conjugated goat anti-mouse IgG secondary antibody (1 g/ml final dilution, Jackson Laboratories). After each antibody treatment, the tissue was washed six times (Ͼ5 min/wash) in PBST. Tissue was mounted using Gelmount (Biomedia). For nuclear staining, tissue was labeled with the DNA-specific dye syto-13 (1 nM final dilution, Molecular Probes) for 15 min, then washed six times in PBST.
Data Collection and Analysis-Images of the sections were collected using a Bio-Rad MRC-600 scanning laser confocal microscope equipped with a krypton-argon laser. SCN images were collected at a 1.2ϫ amplification using a 10ϫ Nikon objective. Data analysis was performed using Metamorph software (Universal Imaging). Semiquantitative densitometric analysis of immunoreactivity was used to quantify endogenous rhythmicity of CRE-regulated reporter gene expression and P-CREB rhythmicity. A 140 (x axis) ϫ 200 (y axis) pixel oval was placed over the digitized SCN image to determine mean fluorescent intensity. A basal level of background fluorescence was determined for each section by acquiring a mean fluorescent signal from the lateral hypothalamic area. This value was then subtracted from the value for the corresponding SCN. Imaging and data analysis were performed "blind." The amplitude of the signal was determined by dividing the peak CT fluorescent intensity signal from the trough CT fluorescent intensity signal. An analogous procedure was used for analysis of other nuclei. For light flash experiments, conventional counting was used to determine the number of immunoreactive cells. The sum of immunopositive cells from three central SCN sections was averaged and expressed as the total number of immunoreactive cells per nuclei. To identify maximal variations in signal intensity, the sensitivity of the confocal microscope was set differently for image collection from endogenous rhythm experiments as compared with light stimulation experiments.
Tissue Culture-Initially, whole neonatal rat brain was cut into 400 -500-m coronal slices using a tissue chopper. A polished 22-gauge needle was then used to punch out the SCN. The tissue was immediately placed in sterile dissociation medium (DM: 90 mM Na 2 SO 4 , 30 mM K 2 SO 4 , 16 mM MgCl 2 , 0.25 mM CaCl 2 , 32 mM HEPES, 0.01% phenol red, 1 mM kynurenic acid (Sigma), pH 7.7), washed three times, and then finely minced. Next, the tissue was incubated for 30 min in DM containing 100 units of papain latex (Worthington) and 4.5 mg of cysteine (Sigma) at 37°C. After removal of the proteolytic solution, the tissue was washed in standard tissue culture medium (minimal essential medium (Life Technologies, Inc.), 5% fetal bovine serum, 100 units/ml penicillin/streptomycin, and 6 g/liter glucose), then triturated into a single cell suspension. Cells were then pelleted, washed twice, then plated in 48-well dishes (Costar) coated with high molecular mass (Ͼ540 kDa, Sigma) poly-D-lysine. Cells cultures were maintained at 37°C and 5% CO 2 in a Napco 6100 incubator. Cytosine arabinofuranoside (8 M) was added to the tissue culture medium on day 2. For Western analysis, the region containing the SCN was excised from postnatal day 1 rat brain. Using the optic chiasm for orientation, an ϳ1-mm cube of tissue was removed. Tissue was washed, triturated, and plated as described above.
Transfections-SCN cells (1 ϫ 10 4 cells/well) were transfected with DOSPER (Roche Molecular Biochemicals). On day 5 in culture, cells were treated with a complex of 0.6 g of DNA and 6 l of DOSPER, as described by the manufacturer, in 100 l of minimal essential medium. After 6 h, the DNA-DOSPER complex was replaced with conditioned tissue culture medium. In some experiments pcDNA3.1-LacZ (Invitrogen) was added (4 ng/well) for transfection efficiency normalization. Transfection efficiency typically did not vary by more than 10% for quadruplicate determinations. The following plasmids have been described previously: dominant-negative MEK S222A (37), dominant-negative PKA, and the CRE-luciferase construct (38).
Reporter Assays-After removal of media, 150 l of cell lysis buffer (0.2% Triton X-100, 4 mM ATP, 6 mM MgCl 2 , 100 mM potassium phosphate, pH 7.8) was added to each well. Following one freeze/thaw cycle, luciferase activity was measured, as described in Ref. 39, and ␤-galactosidase activity was assayed, as described in Ref. 30, using a luminometer (Berthold).
Western Blots-Excised brain tissue was resuspended in 100 l of buffer H (50 mM ␤-glycerophosphate, 1.5 mM EGTA, 0.1 mM Na 3 VO 4 , 1 mM dithiothreitol, 10 g/ml aprotinin, 2 g/ml pepstatin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, pH 7.4) and sonicated for 15 s. 100 l of 5ϫ sample buffer was then added, and the samples were heated to 90°C for 10 min. Extracts were then vortexed (20 s) and centrifuged (8 min at 13,000 ϫ g). 40 l of extract was loaded onto an 8% SDS-PAGE gel and electrophoresed using standard procedures. For cultured SCN neurons, agonist-treated cells were lysed in hot (90°C) 2.5ϫ sample buffer (80 l/dish). After vortexing and centrifugation (7 min at 13,000 ϫ g), 30 l of extract was loaded onto a 12% SDS-PAGE gel and electrophoresed using standard procedures. Once transblotted, membranes (Immobilon P: Millipore) were blocked with 10% powdered milk in PBS. Membrane were then incubated overnight at 4°C in PBST with primary rabbit antibody against ␤-galactosidase (1:500, 5Prime 3 3Prime Inc.) or P-CREB (1:1000, New England Biolabs), or phospho-p44/42-(P-ERK) specific antibody (1:1000 final dilution, New England Biolabs). Membranes were then treated with a goat anti-rabbit IgG alkaline phosphatase-conjugated secondary antibody (1:2000, Cappel). Immunoreactivity was developed using the Western-star alkaline phosphatase detection system (Tropix). A second cultured SCN cell membrane and the same ␤-galactosidase membrane were then probed with a monoclonal anti-CREB antibody (1:1000, Santa Cruz), and labeled with a rabbit anti-mouse IgG antibody conjugated to horseradish peroxidase. The Renaissance chemiluminescence detection reagent (NEN Life Science Products) was used to visualize immunoreactivity. Membranes were washed six times with a 5% milk/PBST solution after each antibody treatment.

Circadian Oscillation in CRE-mediated Gene Expression-If
the CRE enhancer element plays a role in endogenous clock rhythmicity, then its regulation must fulfill several criteria. In the absence of photic cues, gene expression mediated by activation at the CRE must by regulated in a rhythmic manner. Furthermore, this rhythmicity should be observed within the SCN, and have a period close to 24 h. To address whether CRE-mediated gene expression shows circadian oscillations in the SCN, animals transgenic for the CRE-␤-galactosidase reporter (33) were entrained to a 12-h L/D cycle, then placed in total darkness (D/D). Under this condition, circadian rhythmicity is controlled by the endogenous pacemaker. After 6 days in D/D, animals were sacrificed every 4 h over a 24-h cycle and coronal sections containing the SCN were labeled immunohistochemically for the expression of the reporter gene. Quantitation revealed a significant (analysis of variance: p Ͻ 0.001, F ϭ 7.16) circadian variation in CRE-mediated gene expression in the SCN (Fig. 1A). Maximal gene expression was observed from early-to mid-subjective day (Fig. 1B). Levels of the reporter gene dropped markedly from mid-subjective day to mid-subjective night, then rose during late-subjective night. Significant variations in CRE-mediated gene expression were not observed in other hypothalamic nuclei (supraoptic, Fig. 1C) or in other brain regions (piriform cortex and primary motor cortex, data not shown).
Circadian regulation of CRE-mediated transcription was also analyzed by Western blot. As above, transgenic mice were sacrificed over the circadian cycle. Within the SCN, the expression of the CRE-driven ␤-galactosidase reporter construct oscillated over the circadian cycle. Relatively high levels of reporter were observed during late-subjective night and earlysubjective day ( Fig. 2A). Reporter expression within the lateral hypothalamus did not significantly fluctuate over a 24-h period (Fig. 2B). Probing the same membranes revealed that levels of the transcription factor CREB did not significantly vary as a function of circadian time in either the SCN or the lateral hypothalamus. Together, these data reveal the presence of an endogenous, circadian, oscillation of CRE-dependent transcription localized to the SCN.
One necessary step proximal to the induction of CRE-mediated gene expression is the phosphorylation of CREB at Ser-133. Given the circadian oscillation in CRE-mediated transcription, one might expect circadian variations in the phosphorylation state of CREB in the SCN. To address this question, coronal SCN sections from mice sacrificed at different circadian times were labeled for P-CREB immunoreactivity (Fig. 3A). As with CRE-mediated gene expression, levels of P-CREB immunoreactivity exhibited significant (analysis of variance: p Ͻ 0.005, F ϭ 4.33) circadian oscillations under free-running conditions (Fig. 3B). Maximal P-CREB signal was observed from CT 18 -22. Levels of P-CREB declined during subjective day, then began to rise with onset of subjective night. As with CRE-mediated reporter gene expression, P-CREB rhythmicity occurred throughout the dorso-ventral axis of the SCN.
Photic Activation of CRE-dependent Transcription-The ability of light to phase-delay the clock and overt rhythmicity during early-subjective night, to phase-advance the clock during late-subjective night, and to have no effect during subjective day has been well characterized (7). These phase-shifting effects of light result from the synaptic release of glutamate from the RHT and the subsequent activation of ionotropic glutamate receptors in the SCN (8 -10). Within the SCN, photic stimulation during subjective night also induces the expression of a number of immediate early genes (11)(12)(13). Although these data suggest a connection between photic stimulation, glutamate release, gene induction, and phase regulation, the inducible transcriptional pathways activated have not been determined. A clue to the transcriptional pathways that may be critical for phase-shifting was provided by the finding that photic stimulation triggers phosphorylation of CREB at Ser-133 (40). Although suggestive, this observation does not show that CRE-induced transcription occurred, since CREB phosphorylation at Ser-133 is necessary, but not sufficient to trigger CRE-mediated transcription (32,(41)(42)(43)(44)(45). To assess the inducibility of CRE-dependent transcription, reporter mice were entrained to a 12-h L/D cycle, then allowed to free-run under D/D conditions for 5 days. Animals were then exposed to light (400 lux) of varying durations during either the subjective day or the subjective night, and sacrificed 8 h later for analysis of CREmediated transcription. A 60-min light exposure at CT 16.5, a time at which light phase-delays activity rhythms (7), elicited a robust increase in CRE-mediated gene expression relative to control animals not exposed to light (Fig. 4, A and B). The induction of CRE-mediated transcription was primarily localized to the retinoreceptive, ventrolateral region of the SCN. A 60-min light exposure at CT 22.5 (a time point shown to phaseadvance activity rhythms; Ref. 7), also triggered CRE-mediated gene expression (Fig. 4, A and B). A densitometric comparison of the ventrolateral SCN revealed that light induced an ϳ4-fold increase in CRE-mediated gene expression relative to peak levels driven by the endogenous clock under D/D conditions. To assess whether light-mediated gene expression was regulated in a phase-restricted manner, animals were exposed to light during the subjective day. A 60-min light treatment during mid-subjective day (CT 6) did not significantly increase CREmediated gene expression (Fig. 4, A and B).
If CRE-dependent transcription plays a critical role in the phase-shifting effects of light, then very brief light treatments known to phase-shift the clock should also trigger CRE-mediated gene expression. To address this question, mice were exposed to light for only 5 min during subjective night. Compared with control transgenic mice, light treatment increased CRE-mediated gene expression in a highly localized band of cells within the retinoreceptive region of the SCN (Fig. 4C).
Signaling to the CRE-To assess the contribution of different signaling pathways to activation of CRE-mediated transcription, we cultured SCN neurons from P1 rat and transfected them with a CRE-regulated reporter construct. Neurons obtained from a coronal punch of the SCN (Fig. 5A) stained positive for the neuron-specific antigen MAP-2 (Fig. 5B) and were readily transfected (Fig. 5C). Treatment with the adenylyl cyclase activator forskolin (5 M) triggered an ϳ6 -8-fold increase in CRE-mediated transcription (Fig. 5D). K ϩ (50 mM) or NMDA (50 M, plus glycine: 2 M) elicited an ϳ 2-4-fold increase in CRE-mediated transcription (Fig. 5D). Ca 2ϩ influx resulting from depolarization-mediated opening of voltage-activated ion channels is the likely mechanism for K ϩ stimulation of CRE-mediated transcription since administration of the voltage-activated Ca 2ϩ channel blocker isradipine (2 M) blocked CRE-dependent transcription evoked by high K ϩ (data not shown). Likewise, NMDA stimulation of CRE-mediated transcription is very likely initiated by Ca 2ϩ influx.
Identification of signaling pathways that couple to CRE-dependent transcription in the SCN should provide mechanistic insights as to how phase-regulated transcriptional activation is conferred. The MAPK (28), the calmodulin-dependent protein kinase (30), and PKA (29) signaling pathways couple receptor stimulation to CRE-dependent transcription. To assess the contribution of the MAPK cascade, SCN neurons were cotransfected with the CRE-reporter and a dominant-negative interfering form of MEK, an upstream activator of MAPK. This completely inhibited K ϩ -and NMDA-stimulated gene expression (Fig. 5E). Similar results were obtained with PD 98059, a specific inhibitor of MEK (data not shown). Forskolin-activation of CRE-dependent transcription was slightly reduced by disruption of the MAPK signaling pathway. Co-expression of the dominant-negative regulatory subunit of PKA (D/N PKA) blocked Ca 2ϩ as well as cAMP-dependent transcription (Fig.  5E). These results suggest that MAPK and PKA signaling are both required for Ca 2ϩ -dependent CRE-regulated transcription in SCN cells.
Immunoblotting cultured SCN cell extracts showed that potassium (40 mM) depolarization as well as forskolin (5 M) triggered CREB phosphorylation at Ser-133 (Fig. 6A: lane K, potassium; lane F, forskolin; lane C, control). In addition, glutamate (20 M) administration elicited CREB phosphorylation (lane G, glutamate). This suggests that either Ca 2ϩ influx or cAMP can trigger CREB phosphorylation in the SCN. Pretreatment with the specific MEK inhibitor, PD 98059 (75 M), markedly attenuated potassium stimulation of P-CREB, as well as reducing forskolin-and glutamate-evoked CREB phosphorylation. Equal amounts of extract were also probed for CREB expression. Cell extracts were also probed with an antibody that detects the activated (dually phosphorylated Thr-202 and Tyr-204) form of the extracellular signal-regulated kinases erk-1 and erk-2 (here, collectively referred to as ERK). All three agonists triggered robust ERK phosphorylation (Fig. 6B). Pretreatment with PD 98059 blocked agonist-induced ERK phosphorylation.
Pairing of Ca 2ϩ and cAMP signals resulted in robust, synergistic, activation of CRE-mediated transcription in the SCN. Specifically, the coadministration of forskolin (5 M) and K ϩ (50 mM) triggered an ϳ3-fold potentiation of reporter expression (Fig. 5F). To determine whether circadian-regulated genes are modulated in a similar manner, we assessed transcriptional activation driven by the CRE-containing vasoactive intestinal peptide promoter. Vasoactive intestinal peptide is found in high levels in SCN (46) and has been shown to regulate clock rhythmicity (47). Stimulation with either forskolin (5 M) or K ϩ (50 mM) elicited similar levels of reporter expression as the CRE construct (Fig. 5F). Pairing of forskolin and K ϩ synergistically stimulated reporter expression to a level that was similar to that observed with the CRE-driven reporter (Fig. 5F). Together, these results reveal that Ca 2ϩ -and cAMPdependent signaling pathways act with robust synergy to potentiate CRE-dependent transcription in SCN cells. Since Ca 2ϩ can activate the MAPK and PKA pathways, it may function as a signal integrator for regulation of SCN function. DISCUSSION The results presented here show that the SCN exhibit a prominent circadian oscillation in CRE-mediated gene expression in dark-adapted animals. Furthermore, photic stimulation during the subjective night triggered CRE-dependent transcription, whereas light treatment during the subjective day was not effective. Transient transfection experiment revealed that the ERK/MAPK pathway activity is essential for Ca 2ϩ -dependent stimulation of CRE-mediated transcription in SCN cells. Together, these results provide the first evidence linking the CREB/CRE transcriptional pathway to endogenous timing mechanisms.
Photic Stimulation of CRE-mediated Transcription-There is a large body of evidence suggesting that the phase-shifting effects of light result from new protein synthesis. However, the enhancer elements mediating transcriptional activation have not been thoroughly characterized. Our data suggest that the CRE may play a central role in the ability of light to activate gene expression and, in turn, phase-shift the clock. In support of this, we show that light induces CRE-mediated gene expression. Additionally, the finding that a brief 5-min light treatment triggers a highly localized induction of CRE-mediated gene expression correlates with a duration of light shown to phase-shift overt activity rhythms (11). The general pattern of CRE-mediated gene expression, both temporally and anatomically, parallels the induction of several immediate early genes thought to be involved in phase-shifting the clock (11)(12)(13). Given that the promoters for these immediate early genes (including c-fos, junB, and NGFI-B) contain at least one CRE (48 -50), it is reasonable to hypothesize that the CRE may play a central role in mediating the ability of light to trigger immediate early gene induction. Our results showing that lightinduces CRE-regulated gene expression in a phase-restricted manner are consistent with work showing that light-induced CREB phosphorylation is restricted to the subjective night (40). These results suggest that stringent regulation of the CREB/ CRE-transcriptional pathway during the day may be a critical element that confers the phase-restricted phase-shifting effects of light.
Signaling Pathways-A role for Ca 2ϩ in photic entrainment of the clock has been suggested by the finding that lightinduced phase-shifts require NMDA receptor activation (8). Given the evidence identifying a transcriptional component to light-induced phase shifts, we assessed the signaling mechanisms that couple Ca 2ϩ to gene expression in the SCN. In primary cultures of SCN neurons, we found that increasing cytosolic Ca 2ϩ , either through high K ϩ or NMDA administration, resulted in enhanced CRE-dependent transcription. In addition, the co-activation of Ca 2ϩ and cAMP pathways resulted in a robust synergistic activation of CRE-dependent transcription. Interestingly, besides glutamate, RHT nerve terminals also express PACAP (51), a peptide capable of stimu-

FIG. 2. Western analysis of CRE-mediated gene expression.
After CRE-␤-galactosidase transgenic mice were maintained for 6 days in D/D, tissue was isolated at different circadian times and analyzed for ␤-galactosidase expression. A, SCN tissue showed a marked variation in ␤-galactosidase levels as a function of circadian time. CREB levels probed from the same membrane were consistent over circadian time. B, significant circadian variations in ␤-galactosidase or CREB were not observed in tissue from the lateral hypothalamus.
lating cAMP production. Light-induced release of transmitters capable of stimulating Ca 2ϩ and cAMP pathways may be important for robust activation of the CREB/CRE transcriptional pathway in the SCN. Along these lines, we have observed that modest stimulation of cAMP signaling pathways that, alone, was unable to increase CRE-mediated transcription, potently augmented Ca 2ϩ -dependent CRE-mediated transcription in the SCN. 2 CRE-mediated gene expression is regulated by a variety of cellular stimuli acting through a number of different kinase 2 K. Obrietan and D. R. Storm, unpublished observation. cascades (28 -31). Our work shows that Ca 2ϩ stimulation of CRE-mediated transcription was dependent upon activation of the MAPK signaling pathway. Cotransfection with a dominantnegative interfering form of MEK or treatment with the MEK inhibitor PD 98059 blocked Ca 2ϩ -stimulated gene expression. Coupling of Ca 2ϩ to activation of the MAPK cascade has been shown to be dependent upon an enhancement of Ras-GTPase catalytic activity (52). Ca 2ϩ -dependent Ras activation is triggered by a variety of signaling intermediates, including calmodulin kinases (42,53), Src (54), Ras-GRF (55), and the epidermal growth factor receptor (56). The MAPK pathway gains access to the nucleus via the activation-dependent nuclear translocation of ERK. ERK has been shown to triggered CREB phosphorylation through activation of RSK 1, 2, and 3, all of which are CREB kinases (57). A requirement for MAPK activity was also revealed by the observation that Ca 2ϩ -induced CREB phosphorylation was attenuated by the MEK inhibitor PD 98059. In addition, forskolin-stimulated CREB phosphorylation was reduced by PD 98059, indicating that the cAMP-dependent signaling pathway acts, in part, via activation of the MAPK cascade. Along these lines, cAMP has been shown to activate the MAPK cascade in hippocampal and cortical neurons (31).
Within the SCN, elevated cytosolic Ca 2ϩ has been shown to trigger CREB phosphorylation through a mechanism requiring the production of nitric oxide (58). Recently, Ca 2ϩ -dependent nitric oxide activation was shown to elicit ERK phosphorylation in neuronal cultures (59), thus providing a pathway by which light-induced nitric oxide production could trigger sequential MAPK activation. Although other transcription factors may be activated by increased cytosolic Ca 2ϩ , it is intriguing to note that nitric oxide synthetase antagonism blocks both glutamate-induced CREB-phosphorylation and glutamate-in-duced phase-shifts (58,60). This suggests a strong correlation between clock phase-shifting and the activation of a transcriptional pathway involved in triggering CRE-mediated gene expression. Taken together, the results presented here provide a mechanism by which glutamate receptor stimulation leads to CRE-dependent transcription.
Endogenous Rhythmicity-Rhythmic transcription appears to be central to maintaining circadian timekeeping. For example, a mutated form of the putative transcription factor CLOCK abolishes circadian activity rhythms under D/D conditions (16). Our data show that CRE-mediated gene expression is regulated in a circadian manner under free-running conditions. Levels of reporter protein began to rise during the late subjective night and peaked during mid-subjective day. Given the approximately 6 h between transcription and maximal reporter expression, one may deduce that induction of CRE-mediated gene expression is restricted to the subjective night, and possibly the early subjective morning. This result indicates that the phase-dependent regulation of endogenous CRE-mediated gene rhythmicity overlaps with the phase dependence of light inducible CRE-mediated gene expression, suggesting that a similar signaling mechanism may govern both processes.
It is unclear why the light-evoked stimulation of CRE-mediated gene expression was greater than the peak in reporter expression resulting from endogenous pacemaker activity. Possible explanations may include more robust activation of signaling pathways by light or synergism between signaling pathways that are activated by light and the endogenous clock. Interestingly, we recently reported that light triggers MAPK activation in the SCN, and that MAPK activity is regulated in a circadian manner in the SCN under D/D conditions (61).
Recent work performed in Drosophila has revealed a robust circadian oscillation in CRE-mediated gene expression and an FIG. 4. Light induces phase-dependent stimulation of CRE-mediated gene expression in the SCN. Initially, mice transgenic for the CRE-regulated reporter gene construct were placed on a 12-h L/D cycle, then transferred to D/D. After 5 days in total darkness, animals were exposed to white light (60 min, 400 lux) at different circadian times: early subjective night (CT 16.5), late subjective night (CT 22.5), or mid-subjective day (CT 6). Eight hours after light exposure, animals were sacrificed and their brains were removed and immunocytochemically processed for reporter expression. A, photic stimulation during the subjective night, but not during the subjective day, triggered CRE-mediated gene expression. Of note, the late subjective night group (CT 22.5) was sacrificed at CT 6.5 and the mid-subjective day group (CT 6) was sacrificed at CT 14. Given this, the observed CT variation in immunoreactive cells in control groups (shown as black bars in B) corresponds to the variations shown in Fig. 1. Also, note that different confocal settings were used to maximize variations in light flash versus endogenous rhythm experiments. B, quantitation of CRE-mediated gene expression for the three circadian times assayed. Error bars denote S.E. The number above each time point refers to the number of animals assayed. ** ϭ p Ͻ 0.0001, two-tailed Student's t test. C, relative to a control animals, a 5-min light treatment at CT 16.5 induced a highly localized expression of the CREregulated reporter gene.
interdependence between rhythmic CRE-dependent transcription and period oscillations, indicating that the CRE transcriptional pathway is a component of the circadian clock (74). Further work in mammalian systems may reveal a similar interaction between the CRE transcriptional pathway and period homolog rhythmicity.
Circadian variations in the phosphorylation state of the transcription factor CREB at Ser-133 were also observed. The peak in the P-CREB rhythm preceded the reporter gene peak by approximately 6 h, an expected time lag for transcriptional activation and maximal protein expression. The circadian P-CREB rhythm does not result from CREB oscillations, since levels of CREB in the SCN were stable at subjective day versus night time points. This result suggests that circadian oscillations in P-CREB result from circadian fluctuations in the activation state of CREB kinases or phosphatases.
It is unclear how rhythmic CRE-mediated gene expression is maintained under free-running conditions. However there are several plausible explanations. Given that extracellular membrane receptor-mediated signaling events regulate CRE-mediated gene expression, one may expect to observe circadian changes in the level of extracellular transmitters capable of eliciting CRE-mediated gene expression. In support of this idea, circadian variations in the concentrations of excitatory amino acids have been observed within the region of the SCN under free-running conditions, and in slice preparations (62)(63)(64). Circadian oscillations in CRE-dependent transcription also may be a result of an inherent rhythmic transcriptional program of SCN pacemaker cells. Another possibility is that the amount or ratio of CREB heterodimerization partners within the SCN varies over the circadian cycle. In support of this idea, CREM-deficient mice do not express circadian locomotor activity. 3 The circadian expression of a variety of genes within the SCN may result from circadian CRE-dependent transcription. For example, the promoters for several peptides that show circadian oscillations at the mRNA or protein level in the SCN, including vasopressin, somatostatin, and, during development, vasoactive intestinal peptide (22,65,66), contain one or more CREs (67)(68)(69). Interestingly, these peptides have the capacity to both modulate CRE-dependent transcription, either positively or negatively (70 -72), and to alter rhythmicity when added to the SCN (47,73). Conceivably, the circadian expression of these proteins could be generated by temporally overlapping feedback loops that either activate or inhibit CREmediated transcription. Based on the results presented here, we propose that the CRE transcriptional pathway plays an important role in orchestrating the series of transcriptional events essential for both endogenous clock rhythmicity and the ability of light to phase-shift the clock.