Disturbed Clockwork Resetting in Sharp-1 and Sharp-2 Single and Double Mutant Mice

Background The circadian system provides the basis to anticipate and cope with daily recurrent challenges to maintain the organisms' homeostasis. De-synchronization of circadian feedback oscillators in humans causes ‘jet lag’, likely contributes to sleep - , psychiatric - , metabolic disorders and even cancer. However, the molecular mechanisms leading to the disintegration of tissue-specific clocks are complex and not well understood. Methodology/Principal Findings Based on their circadian expression and cell culture experiments, the basic Helix-Loop-Helix (bHLH) transcription factors SHARP-1(Dec2) and SHARP-2(Stra13/Dec1) were proposed as novel negative regulators of the molecular clock. To address their function in vivo, we generated Sharp-1 and Sharp-2 single and double mutant mice. Our experiments reveal critical roles for both factors in regulating period length, tissue-specific control of clock gene expression and entrainment to external cues. Light-pulse experiments and rapid delays of the light-dark cycle (experimental jet lag) unravel complementary functions for SHARP-1 and SHARP-2 in controlling activity phase resetting kinetics. Moreover, we show that SHARP-1 and 2 can serve dual functions as repressors and co-activators of mammalian clock gene expression in a context-specific manner. This correlates with increased amplitudes of Per2 expression in the cortex and liver and a decrease in the suprachiasmatic nucleus (SCN) of double mutant mice. Conclusions/Significance The existence of separate mechanisms regulating phase of entrainment, rhythm amplitude and period length has been postulated before. The differential effects of Sharp-deficiency on rhythmicity and behavioral re-entrainment, coupled to tissue-dependent regulatory functions, provide a new mechanistic basis to further understand the complex process of clock synchronizations.


Introduction
Circadian rhythms in mammals are orchestrated by a pacemaker residing in the suprachiasmatic nucleus (SCN) of the hypothalamus [1]. The endogenous period length of this pacemaker deviates from the 24 h solar cycle and the clock therefore needs to be constantly reset to maintain synchrony with external time, a process termed entrainment. Light, transmitted to the SCN via the retino-hypothalamic tract, is the main entrainment stimulus (or Zeitgeber) for the core circadian oscillator [2]. Autonomous extra-SCN clocks in the forebrain and peripheral organs are synchronized via humoral and neuronal pathways and control the tissue and cell-type specific rhythmic expression of target genes [2,3]. The oscillating expression of these clockcontrolled genes is thought to provide the basis for an organism to anticipate and cope with daily recurrent challenges. Internal desynchronization of peripheral and central oscillators causes 'jet lag' upon time-zone travelling and severely affects shift workers [4]. Genetic alterations in circadian genes have been shown to be linked to the familial advanced sleep phase syndrome (FASPS) [5,6] and may contribute to psychiatric disorders [7][8][9] and even cancer [10,11], highlighting the pivotal role of the circadian timing system in the maintenance of healthy body function.
Based on their rhythmic mRNA expression in the SCN and forebrain structures, the bHLH transcription factors SHARP-1(DEC2) and SHARP-2(DEC1) have been suggested as additional negative regulators of the mammalian clock [12]. They both bind to canonical E-Box cis-elements and repress CLOCK/BMAL1mediated reporter gene expression in cell culture [12]. Sharp-2 mRNA expression in the SCN was shown to be coupled to light stimulation [12] and neuronal activity in the cortex [13]. Analysis of Sharp-2/Dec1 single mutant mice provided evidence for a role as a circadian output regulator in the periphery and in modulating phase of clock gene expression [14,15].

Results
In this study, we characterize the circadian system of mouse mutants lacking functional Sharp-1 and/or Sharp-2 genes. The targeted inactivation of Sharp-2/Stra13 in mice has been described recently [16]. To inactivate Sharp-1, we replaced exon 3 coding for the DNA binding domain by a neomycin resistance cassette ( Figure 1A). Genomic PCR and southern blot analysis proved successful targeting in ES cells and in mice ( Figure 1B,C). RT-PCR with primers located in exon 1 and 4 revealed three aberrantly spliced mRNA species that might potentially lead to two different truncated protein fragments lacking any characterized functional domain (Supplementary Figure S1A-D). These truncated fragments were cloned in expression plasmids and appear to be non-functional ( Figure S1E).
We assessed circadian clock function by monitoring wheelrunning behavior of Sharp single and double mutants. Mice of all genotypes entrained normally to a 12 hr light/12 hr dark (LD) cycle ( Figure S2 and data not shown). In constant darkness (DD), circadian activity rhythms remained stable without significant differences in free-running period lengths (t) in single homozygous null mutants ( Figure 1D,E). However, when Sharp gene dosage was further lowered, we observed a significant lengthening of t (up to 0.5 h) in mice with only one or no functional Sharp allele left ( Figure 1D,E and Figure S2) indicating a high degree of functional redundancy between both genes. The loss of one Sharp gene resulted in a broadened t distribution indicating an elevated variability in pacemaker regulation while in the double mutants t was consistently shifted towards higher values ( Figure 1E).
Because Sharp-2 expression in the SCN is rapidly activated after nocturnal light exposure, and remains elevated in the SCN in constant light conditions [12], we speculated that the Sharps could also be involved in the light synchronization of the SCN clock by light. Using a standard Aschoff type II paradigm, 15 min light pulses were given at Zeitgeber time (ZT)14, running-wheel activity was monitored after subsequent release into DD and the resulting phase shifts were calculated by comparison of activity onsets of light-pulsed animals and non-pulsed controls. Similar to what was seen for the period length in DD, Sharp-1 deficiency did not significantly affect phase shifting behavior when compared to WT controls ( Figure 2A). We observed, however, a markedly pronounced prolonged resetting kinetics (transition phase) in Sharp-2-deficient animals (data not shown). To further substantiate this finding, we tested WT and S2 2/2 animals in an Aschoff type I setup applying a light pulse 2 hrs after activity onset under DD conditions (Circadian time (CT)14) when phase shift transients are more easily assessable. In this paradigm total phase shift amplitudes were again comparable between genotypes but S2 2/2 mice needed consistently longer to complete the transition phase ( Figure S3). We further reduced the gene dosage analyzing mice with only one functional Sharp-2 (S1 2/2 S2 +/2 ) or Sharp-1 (S1 +/2 S2 2/2 ) allele and homozygous double-deficient animals (S1/2 2/2 ). S1 2/2 S2 +/2 mice did not show a significant change in phase shifts (6869 min) compared to WT animals (83610 min) ( Figure 2A). However, mice lacking both Sharp-2 and one Sharp-1 wild-type allele(s) as well as double mutants displayed marked reduction in phase shift responses (2769 min and 1965 min, respectively; Figure 2A). Although we again detected a functional redundancy of both Sharp genes in this paradigm, Sharp-2 loss-of-function cannot be fully compensated for by only one Sharp-1 wild-type allele. The amount and direction of photic activity phase shifts depends on the time of light exposure and involves differential molecular signaling mechanisms. To get a first idea about the circadian dependence of phase responses in Sharp mutant animals we also exposed all animals to light pulses at ZT22 (phase advancing) and ZT10 (no phase shift in wild-type mice). For both time points no significant differences were detected between wild-type and all Sharp single and double mutant animals (data not shown) indicating that the shape of the photic phase response curve is preserved in the absence of Sharp-1 and/or Sharp-2 and that the influence of both genes on clock resetting is restricted to phase delays.
The resetting of the SCN clock by light -and with it the activity phase -is thought to be the direct consequence of transcriptional activation of light-responsive genes in the SCN [2,3,17]. The prototypical immediate-early-genes c-Fos and c-Jun as well as Per1 are known to be rapidly activated in the SCN after nocturnal light exposure [18]. We therefore microdissected SCN tissue from frozen coronal brain sections of mice that either received a 15 min light stimulus at ZT14 and were sacrificed 30 and 60 min later or were sacrificed under dim red light at ZT14 without prior light treatment ( Figure 2B). The samples of three WT and three S1/ 2 2/2 mice were pooled and subjected to quantitative RT-PCR analysis, monitoring Per1, c-Fos and c-Jun mRNA levels ( Figure 2C-E). We observed elevated base levels of expression for all three genes already at ZT14 indicating alterations in the diurnal regulation of these genes in the absence of SHARP-1 and -2 ( Figure 2C). In contrast, light induction of both c-Fos and Per1 was significantly blunted both 30 ( Figure 2D) and 60 min ( Figure 2E) after light exposure in Sharp double mutant animals. These data correlate with the attenuated behavioral light responses observed in these mice and indicate a direct influence of the SHARP proteins on circadian gene regulation in the SCN.
We next investigated the impact of Sharp-deficiency on clockwork resetting linked to non-SCN oscillators [3,4]. Npas2deficient mice are characterized by blunted expression rhythms of Per2 in the cerebral cortex [19] and adaptation to rapidly shifted LD cycles (jet lag) is accelerated when compared to WT controls [20]. Given the overlapping expression patterns of Npas2 and Sharp-1 and -2 [12,13,19] and the proposed negative function of the SHARP proteins in the mammalian clock [12], we hypothesized that these animals should also show a related -but inverted -jet lag phenotype. Therefore, Sharp-deficient mice were subjected to a rapid 8 h delay of the LD cycle mimicking a flight from central Europe to Arizona. In WT animals, activity phase resetting to the new LD cycle occurred within 3 days after the shift ( Figure 2F and Figure S4), in accordance with published data [20,21]. In contrast, both Sharp single mutants displayed significantly slowed behavioral re-entrainment to the delayed LD cycle (two-way ANOVA for S1 2/2 , p,0.0005; S2 2/2 , p,0.0005) ( Figure 2F). With respect to the resetting kinetics, Sharp-1 mutants appeared to be less affected than Sharp-2 mutants ( Figure 2F). S1 2/2 S2 +/2 , S1 +/2 S2 2/2 and double mutants (S1/2 2/2 ) were also significantly delayed in their adaptation to a delayed LD cycle (S1 2/2 S2 +/2 , p,0.0005; S1 +/ 2 S2 2/2 , p,0.0005; S1 2/2 S2 2/2 , p,0.005) ( Figure 2G-I and Figure S4). As judged by ANOVA, the additional loss of functional Sharp-1 alleles, however, did not significantly alter the resetting kinetics when compared to S2 2/2 mice, although the phase delay in particular during the first days was markedly reduced ( Figure 2H-I). Thus, in contrast to the other phenotypes described so far, Sharp-2 function seems to be essential for a rapid resetting to an 8 h delayed LD cycle and cannot be compensated for by Sharp-1. These effects appear to be tied to tissue-dependent function of SHARP proteins, e.g. in the forebrain clock.
To follow up on the dual mode of SHARP protein function in more detail, we used mammalian two-hybrid assays ( Figure S7) and monitored the subcellular localization of NPAS2 upon co-transfection with either BMAL1, SHARP-1 or SHARP-2 encoding expression plasmids ( Figure S8). These and previous experiments (Figure 4 and Figure S5) reveal that SHARP-1 and -2 most likely do not interact via the HLH domain with BMAL1, NPAS2 and CLOCK to either repress or co-activate as heterodimeric complexes clock gene expression in an E-Box dependent manner. Quantification of the effects of SHARP-1 and -2 reporter assays show that the co-activator function is dependent on BMAL1/NPAS2 or BMAL1/ CLOCK complexes but is E-Box independent ( Figure S9). We thus generated SHARP-1 and -2 constructs lacking the DNA binding domains (SHARP-1^b and SHARP-2^b) and assessed their function in reporter gene assays. In HEK293 and PC12 cells, the SHARP-1^b and SHARP-2^b proteins could not repress E-Box mediated reporter gene transcription when transfected seperately ( Figure 4H,I). However, SHARP-1^b and SHARP-2^b further stimulated the NPAS2/BMAL1 or CLOCK/BMAL1 driven reporter gene expression (up to 4-fold in PC12 cells) ( Figure 4L-M), with one exception. In HEK293 cells, SHARP-2^b did not enhance NPAS2/ BMAL1 activities ( Figure 4J) but, unlike SHARP-1 ( Figure 4A), did not display any repressive effect.

Discussion
Taken together, our data show that SHARP-1 and -2 act both as E-Box binding dependent transcriptional repressors and as coactivators of NPAS2-CLOCK/BMAL1 in reporter gene assays which correlates with the differential Per2 gene regulation in vivo. In the absence of functional NPAS2-CLOCK/BMAL1 heterodimers, SHARP-1 and -2 have an inhibitory function, which is in accordance with published data [12,26,27]. However, in the presence of functional NPAS2-CLOCK/BMAL1 heterodimers, SHARP-1 and -2 likely act either as directly E-Box binding repressors or as co-activators of transcription depending on the context and the nature of E-Box binding complexes ( Figure 4N) and may thus regulate the amplitude of clock gene expression in different tissues ( Figure 4O).
The activity of SHARP-1 and -2 on clock gene transcription is thus principally different to that of CRY1 and CRY2 [30]. Deletion of both Cry1 and Cry2 alleles in mice leads to a loss of rhythmicity in running wheel activity in DD and both genes are therefore essential negative components of the core clock of the SCN [31]. In contrast to the Cry genes, our analyses show that SHARP-1 and -2 are essentially involved in the fine-tuning of circadian functions such as period control, clockwork resetting and entrainment in the SCN as well as in clocks operating in other tissues such as the forebrain or liver. A similar effect on regulation of clock gene amplitude has been described for Drosophila mutants lacking cwo [28,29,32,33].
Our in vivo and cell culture analyses suggest a tissue/cell typedependent function of SHARP proteins in the circadian clock. Sharp-1 and Sharp-2 are expressed in the SCN as well as in other tissues and brain regions where an extensive overlap with NPAS2 expression domains is apparent [12,13,19]. The phenotypes of Sharp mutants in the jet lag paradigm and activity periods most likely represent the cumulative net effect of altered SHARP-1 and/or SHARP-2 functions integrated from different circadian oscillators, including the SCN pacemaker, where CLOCK dominates, as well as other CNS timers like the forebrain clock, where NPAS2 is essential. This tissue-specific clockwork composition appears as the key to SHARP functions. Further molecular and behavioral studies are required to dissect the cellular and molecular mechanisms that may be modulated by SHARP-1/2 in different cell types or tissues. The mechanism that determines the context-dependent switch between a repressing versus activating mode of SHARP function remains to be determined. However, our data with SHARP constructs lacking only the basic domain implies that this may occur post-translationally regulating the E-Box binding. The existence of separate mechanisms regulating phase of entrainment, rhythm amplitude and period length of the circadian system has been postulated before [6,34]. The differential effects of Sharp-deficiency on free-running rhythmicity and behavioral re-entrainment, coupled to tissue-dependent SHARP regulatory functions, provide a new mechanistic basis to support this concept. An improved mechanistic insight of clock desynchronization processes is likely to be relevant for several human diseases with altered 'clocks'.

Gene targeting and homologous recombination in mouse ES cells
Gene targeting in mouse embryonic stem cells was performed as already described [35]. Briefly, a l mouse 129Sv genomic library (Stratagene, La Jolla, CA) was screened with a full-length rat SHARP-1 cDNA, and several clones containing the five-exon gene [13], including at least 3 kb of 59-and 5 kb of 39-flanking sequences each, were isolated. A gene-targeting vector for positivenegative selection was constructed by subcloning the neomycin resistance (neoR) gene (modified by PCR from pMCneo; Stratagene) as a XbaI-PstI fragment into a linearized vector pKS (Stratagene). A XbaI fragment inluding 4 kb 59 upstream and parts of intron 2 was placed 39 of the neoR cassette. A NotI-PstI fragment containing 39 sequences of the SHARP-1 gene was placed 59 to the neoR cassette. For negative selection, the HSV-tk gene (modified by PCR from pMC1TK; Stratagene) was inserted into the NotI-SacII linearized construct. The targeting construct lacked the DNA binding domain located in exon 3 but transcription of the mutant gene could be monitored at the RNA level using a sense neoR specific probe. R1 mouse embryonic stem (ES) cells were cultured on mitomycin-treated primary mouse embryonic fibroblasts (feeders). For electropora-tion (240 V, 500 mF; Gene Pulser; BioRad, Munich, Germany), ,10 7 ES cells were suspended in 0.8 ml of PBS containing 50 mg of the linearized (ScaI) targeting construct. Transfected cells were cultured on gelatinized dishes (Falcon) in the presence of 10 3 U/ ml leukemia inhibitory factor (GIBCO), and selection with 300 mg/ml G418 (GIBCO) and 2 mM ganciclovir (Syntex) was started after 48 and 72 hr, respectively. Between 8 and 10 d after electroporation, double-resistant colonies were picked and trypsinized. Approximately one third of each colony was plated onto feeder cells. The remaining cells of eight clones each were pooled, suspended in 50 ml of water, heat-lysed at 95uC for 10 min, and digested with 10 mg of proteinase K at 55uC for 30 min.
To identify homologous recombinants, we used aliquots of the pooled lysates to PCR amplify a 1.2 kb genomic fragment. Two primers located in exon 5 39 of the targeting construct at position A (S1-KOA, 59-TGTCGGTGTCCGTGTCGTGT-39) and position B (S1-KOB, 59-AGACTTCCGTGGCTCTTAGAAC-39) in combination with one primer located in position C of the targeting construct (corresponding to the 59 end of the neoR cassette) (neo6, 59-GCAATCCATCTTGTTCAATGGC-39) were used (locations depicted in Figure 1A). In a 50 ml reaction, 'nested' PCRs were performed with denaturation at 94uC (30 sec), annealing at 56uC (45 sec), and extension at 72uC (60 sec) for 19 cycles (using S1-KOA and neo6) followed by 35 cycles with primers S1-KOB and neo6. Amplification products were visualized on ethidium bromide-stained gels and verified by Southern blot analysis. Correct targeting was also confirmed by Southern blotting of ES cell DNA (10 mg), hybridized to a 59 genomic probe (see location in Figure 1A). Microinjection of selected ES cells into C57Bl/6J blastocysts was performed by standard procedures. Highly chimeric males were crossed to C57Bl/6J females.
For routine analysis, offspring genotype was determined by PCR on tail-biopsy DNA with primers S1-s (59-ACCTACAAGT-TACCGCACAG-39), S1-as (59-TTTCTCCAAATGCCCCA-GTG-39), and neo6. The wild-type allele yielded in a single 420 bp product amplified with S1-s and S1-as and the mutant allele in a single product of 350 bp amplified with neo6 and S1-as. (a-f) Reporter gene assays using a firefly luciferase reporter construct driven by a herpes simplex thymidine-kinase (TK) minimal promoter and three upstream clustered CACGTG E-Box elements and BMAL1, NPAS2 and SHARP-1 or SHARP-2 expression plasmids performed in HEK293 cells (a,d), PC12 cells (b,e) and primary cultured mouse cortical neurons (c,f) as indicated. (a,d) SHARP-1 and -2 repress basal and NPAS2/BMAL1 (a) and CLOCK/BMAL1 (d) enhanced promoter activity in HEK293 cells. Co-transfection of NPAS2 or CLOCK with BMAL1 encoding plasmids activates the reporter gene more than 8-and 30-fold, respectively. SHARP-1 and -2 efficiently repress NPAS2/BMAL1 and CLOCK/BMAL1 mediated reporter gene activity. (b,e) SHARP-1 represses basal and NPAS2/BMAL1 (b) and CLOCK/BMAL1 (e) enhanced promoter activity in PC12 cells whereas co-transfection of SHARP-2 encoding plasmids has no effect on NPAS2/BMAL1 mediated luciferase activation (b) but further stimulates CLOCK/BMAL1 activities by almost two-fold (e). (c,f) Reporter gene assays performed in primary cultured neurons. (c) SHARP-1 and -2 repress basal promoter activity in reporter gene assays performed in primary cultured neurons, but coactivate NPAS2/BMAL1 mediated reporter activation. Co-transfection of both BMAL1 and NPAS2 leads to an increase of luciferase activity by approximately 6-fold and SHARP-1 and -2 further enhance this activation by a factor of two to at least 12-fold over basal promoter activity. (f) SHARP-1 represses CLOCK/BMAL1 activity and SHARP-2 stimulates CLOCK/BMAL1 activity by more than two-fold.

SHARP-1 and -2 mutant mice
The experiments were performed with cohorts of male mice aged 3-6 months, respectively. Wild-type (WT) and Sharp-1 and -2 mutant mice were obtained from offspring of double heterozygous breeding pairs. Parental single heterozygous mice were independently backcrossed to C57Bl/6J for more than ten generations. All experimental animals were housed in the same ventilated soundattenuated rooms under a 12 h light/12 h dark schedule at an ambient temperature of 21uC with food and water available ad libitum. All experiments were performed blinded for genotype and were conducted in accordance with NIH principles of laboratory animal care and were approved by the Government of Lower Saxony, Germany.

Locomotor activity monitoring and circadian phenotype analysis
Mice housing and handling were performed as described [36,37]. Activity data were recorded and evaluated using the ClockLab data collection and analysis software (Actimetrics, IL). Prior to all experiments, animals were entrained to a 12 hrs light: 12 hrs dark (LD) cycle for at least 2 weeks. For LD-DD transitions, lights were turned off at the end of the last light phase and not turned back on the next day. Total activity in LD and constant darkness (DD) and relative light phase activity in LD were assessed over a time of 5 consecutive days. Rhythmicity and endogenous period length (t) in DD were analyzed by x 2 periodogram analysis over an interval of 5 days. For LD phase-shifting experiments (experimental jet lag) animals were, after 2 weeks of stable entrainment to a first LD cycle, subjected to an elongated night phase followed by an 8 hrs backward shifted LD cycle. Adaptation time was calculated by comparing activity phase onset times on the days after the LD shift as determined by the ClockLab software.
To analyze acute phase shifting by single nocturnal light pulses both Aschoff type I and II paradigms were used [38]. For type I individual period lengths and activity onsets were determined for each animal in DD and 15 min light pulses (400 lx) applied at two circadian hours (t/12) after activity onset (CT14) before releasing the animals back into DD. Phase shifts were calculated as the time difference between two regression lines through the activity onsets before and after the pulse on the day of the light exposure. The first days after the light pulse were excluded from the analysis and classified as transition time. For the type II protocol animals were entrained to an LD cycle, 15 min light pulses were applied two (ZT14) or 10 hrs (ZT22) after ''lights off'' and animals released into DD. Phase shifts were determined as the difference between two regression lines through onsets before and after the pulse on the day of the light exposure as compared to animals of the same genotype which were released into DD without extra light treatment. Again, the first days in DD were not used for the calculation because animals were thought to be in transition between both states.

Tissue Sampling for RT-PCR
For qRT-PCR, frontal cortex (bregma 0 to 22 mm) was isolated using a 'rodent brain matrix' 1 mm coronal slicer (ASI Instruments, Warren, MI) and pooled from three mice of each genotype. Microdissection of SCN tissue was performed according to published protocols [39]. In brief, brains from WT and S1/22/ 2 mice were harvested 30 and 60 min after 15 min light pulse given at ZT14 and were frozen on dry ice (n = 3 for each genotype and timepoint). Cryosections (15 mm) were obtained at the level of the SCN and sections were dip-stained with thionin (0.01% in 5% acetic acid). This protocol was shown not to affect RNA integrity [39]. The SCN was identified by the Nissl-morphology and pure SCN tissue was collected by laser-capture microdissection (Molecular Devices, CA; Veritas system) from at least 10 subsequent sections and RNA was pool-isolated with the RNA micro kit according to protocols given by the manufacturer (Qiagen, Hilden, Germany). cDNA synthesis and qRT-PCR were essentially performed as described [39].

Statistical Analysis
The behavioral data are presented as mean values 6SEM, qRT-PCR and Luciferase assays as means 6SD as indicated. Pairwise comparisons were evaluated by two tailed t-tests, repeated measures by two-way ANOVA with the factors time6 genotype followed by Bonferroni post-hoc tests. A p value ,0.05 was considered significant. Figure S1 Characterization of the mutated Sharp-1 locus. (A) Schematic drawing of the Sharp-1 gene structure, CDS depicted as dark blue, 59 and 39 UTR in light blue. In the mutated Sharp-1 allele, exon3 (encoding the DNA binding domain) is replaced by a neomycin-resistance conferring cassette (neoR). RT-PCR primers located in exon 1 (Ex1s) and 4 (Ex4as) were used to detect aberrantly spliced transcripts, the primer pair amplifying exon 3 containing transcripts (Ex3s with Ex5as) was used to control for the absence of WT transcripts in the SHARP-1 mutant. (B) RT-PCR analysis with brain cDNA obtained from wild-type (+/+), Sharp-1 heterozygous (+/2) and homozygous (2/2) mutant mice. With the primer pairs Ex1s and Ex4as the expected 335 bp long product is amplified, in the hetero-and homozygous mutant additional aberrant products of increased size (.1.3 kB) are detected (left). With the primer pairs Ex3s and Ex5as, no product is amplified in homozygous mutants, cDNA input was checked with Gapdh primers (right). Three aberrantly PCR products (A, B, C) were detected in homozygous sharp-1 mutant cDNA (higher magnification depicted at the very right). (C) The PCR products A, B, C were subloned and sequenced. The analysis of the sequences revealed that transcripts corresponding to aberrant splice variant A may potentially result in an ORF of 49 aa lacking any known functional or homology region. Transcript A is not spliced at the junction from exon 2 to intron 2 and runs into a stop codon after seven residues. Transcripts B and C are unspliced already at exon 1 leading to an ORF of 21aa purely consisting of exon 1 encoded residues. (D) Putative aberrant protein variants from the mutated sharp-1 allele depicted as bar graphs. If expressed, Protein A (49 aa) will be comprised of exon 1 and 2 containing residues including a few additional aberrant amino acids, Protein B, C (21 aa) only contains the residues encoded by exon 1 followed by an immediate stop codon. (E) Reporter gene assay comparing the effect of the potentially expressed aberrant proteins A and B, C and wild-type SHARP-1 on an E-Box containing reporter. Whereas full-length SHARP-1 efficiently repressed reporter gene activation, neither constructs encoding aberrant products A or B, C showed any effect. RLUs = refererence light units. (Data represent mean values 6SD, n = 6 replicates). Found at: doi:10.1371/journal.pone.0002762.s001 (7.17 MB TIF) Figure S2 Normal rhythmicity in LD and DD. Representative actograms showing wheel-running activities of wild-type (WT), Sharp-1 (S12/2), Sharp-2 (S22/2), and double null-mutant (S1/22/2) mice in a 12 h light 12 h dark (LD) cycle and in complete darkness (DD). Lights-off phases are indicated by a grey, lights-on periods by a white background. Mice of all genotypes exhibit a robust circadian activity in LD, peaking during the first hours of darkness. In DD, rhythmic activity is retained with stepwise shifted wheel running onsets. running activities of WT, S12/2, S22/2, S12/2S2+2, S1+/ 2S22/2 and S12/2S22/2 mutant mice exposed to an 8 h delayed LD cycle, respectively. Representative activity plots are depicted for each genotype; arrows indicate time of LD shift. Found at: doi:10.1371/journal.pone.0002762.s004 (0.47 MB TIF) Figure S5 Repression of E-Box driven reporter gene. (A-C) Reporter gene assays with a firefly luciferase reporter construct driven by a herpes simplex thymidine-kinase (TK) minimal promoter and three upstream clustered CACGTG E-Box elements and BMAL1, NPAS2 and SHARP-1 or SHARP-2 expression plasmids performed in HEK293 cells (A), PC12 cells (B) and primary cultured mouse cortical neurons (C) as indicated. In all cell types, SHARP-1 and -2 significantly repress basal reporter gene activity irrespective of the co-transfection of BMAL1, NPAS2 or CLOCK encoding expression constructs. RLUs = refererence light units (Data represent mean values 6SD, n = 6 replicates). Found at: doi:10.1371/journal.pone.0002762.s005 (0.64 MB TIF) Figure S6 Transcriptional repression in CHO cells. SHARP-1 and -2 repress basal and BMAL1/NPAS2 enhanced promoter activity in reporter gene assays performed in Chinese hamster ovary (CHO) cells. CHO cells were transfected with a luciferase reporter construct driven by a herpes simplex thymidine-kinase (TK) minimal promoter and three upstream clustered CACGTG E-Box elements and BMAL1, NPAS2 and SHARP-1 or SHARP-2 expression plasmids as indicated. In these cells, co-transfection of BMAL1 and NPAS2 encoding plasmids activate the reporter gene approximately 2.5-and 1.5-fold, respectively. When both BMAL1 and NPAS2 are co-transfected, reporter activity is increased more than 5-fold. Under all conditions, SHARP-1 represses reporter gene activity below the basal activity. SHARP-2 mediated repression appears to be less efficient in this context RLUs = refererence light units (Data represent mean values 6SD, n = 6 replicates). EYFP-NPAS2, the enhanced yellow fluorescent protein (EYFP) is fused N-terminally to full-length NPAS2 that contains a bHLH and two PAS domains. BMAL1, full-length BMAL1 also harbours a bHLH and two PAS domains. SHARP-1 and SHARP-2, full length SHARP-1 and -2 encoding constructs including the bHLH, orange and C-terminal (CTD) domains. (B) Representative microscopic pictures depicting the EYFP fluorescence of EYFP-NPAS2 when transfected separately or in combination with BMAL1, SHARP-1 or SHARP-2 encoding constructs in COS1 cells. (C) Quantitative analyses of the subcellular localizations (white bars, nucleus; grey bars, both nucleus and cytoplasm (N/C); black bars, cytoplasm) of EYFP-NPAS2. Whereas BMAL1 efficiently alters the EYFP-NPAS2 fluorescence from an almost exclusive cytoplasmic to a nuclear localization, SHARP-1 and -2 do not show any effect in this assay. Data are given as percentage of means, n = 4 with more than 20 microscopic fields (each with at least 3-4 cells) analyzed. Scale bar = 50 mm. Found at: doi:10.1371/journal.pone.0002762.s008 (2.66 MB TIF) Figure S9 SHARP-1 and -2 are E-Box-independent transcriptional co-activators of BMAL1/NPAS2 or BMAL1/CLOCK. (A-C) Reporter gene assays monitoring the transcriptional effect of SHARP-1 and -2 using an E-Box independent assay in CHO cells. (A) Schematic domain structure of constructs used. Gal4-BMAL1, the DNA binding domain of Gal4 was fused N-terminally to fulllength BMAL1. NPAS2 and CLOCK, full-length NPAS2 and CLOCK expression constructs, respectively. VP16-NPAS2 and VP16-CLOCK, the strong transactivation domain VP16 was fused N-terminally to full-length NPAS2 and CLOCK expression constructs, respectively. SHARP-1 and SHARP-2, expression constructs encoding full length SHARP-1 and -2, respectively. (B) Schematic drawing of the assay principle. Gal4-BMAL1/NPAS2 or CLOCK complexes are monitored with a Gal4-dependent reporter plasmid (G5-Luci) in the absence and presence of SHARP-1 or SHARP-2. (C) Gal4-BMAL1 activates along with exogenous NPAS2 or CLOCK basal reporter gene activity more than 8-fold above basal promoter activity (small inset figure). Additional co-transfection with SHARP-1 and SHARP-2 strongly elevates reporter gene transcription. SHARP-1 further stimulates Gal4-BMAL1/VP16-NPAS2 and VP16-CLOCK activities strongly, whereas SHARP-2 co-activation is less pronounced. RLUs = refererence light units (Data represent mean values 6SD, n = 6 replicates). Found at: doi:10.1371/journal.pone.0002762.s009 (0.61 MB TIF)