Light Directs Zebrafish period2 Expression via Conserved D and E Boxes

A highly conserved promoter module in a vertebrate clock gene confers light-regulated gene expression.


Introduction
Essentially all organisms demonstrate daily rhythms in biochemistry, physiology, and behaviour that are controlled by the circadian clock [1]. Underlying this oscillator is a molecular mechanism consisting of interacting positive and negative feedback loops that function in a cell-autonomous manner [2]. As the free running period of the clock is not precisely 24 h, it must be reset each day by external signals in order to remain synchronized with the photoperiod of the environment [3]. The principal external cue that entrains circadian rhythmicity is light [4]. In mammals, photic signals from the eyes are communicated via the retinohypothalamic tract (RHT) to the master circadian pacemaker in the suprachiasmatic nucleus (SCN) and thereby they synchronize rhythmic neuronal activity [1]. The SCN in turn coordinates an array of peripheral oscillators, hence driving a diversity of rhythmic processes such as locomotor activity, metabolism, and hormonal secretion. One extensively studied SCN-driven process is the rhythmic production of melatonin in the pineal gland. Via a multisynaptic pathway, SCN neurons stimulate the production of melatonin at night by activating cAMP production [5].
In non-mammalian vertebrates, the circadian timing system appears to perceive light through a more distributed photoreceptive system. Thus, outside of the retina, dedicated photoreceptors are also encountered in the pineal gland, the parietal eye, as well as deep brain structures [6]. The pineal gland of non-mammalian vertebrates contains a photoreceptive circadian oscillator that directly drives rhythmic melatonin production. Consequently, cultured pineal glands are capable of autonomously generating a melatonin rhythm that can be entrained by light [6][7][8]. In some species the pineal gland is considered to act as a master circadian pacemaker [6]. Interestingly, in zebrafish, circadian clocks within many tissues and cell lines have also been shown to be directly light entrainable [9][10][11]. The mechanism whereby light directly synchronizes the rhythms of these peripheral clocks is still a mystery.
As demonstrated in several species, period (per) genes, negative components of the circadian oscillator, are involved in its entrainment by light. Per1 and per2 mRNA levels in the rodent SCN are increased by light exposure during the subjective night but not during the subjective day [12]. Moreover, mice carrying mutated per1 or per2 genes do not exhibit the normal advanced and delayed phase shift responses to light pulses [13]. The per2 transcript has also been shown to be light inducible in the chicken pineal gland and SCN [14]. Furthermore, in Xenopus, expression of per2 in the retina and retinal pigment epithelium is light-dependent and is therefore suggested to play a role in circadian entrainment [15].
In zebrafish, the expression pattern of per genes has been studied both in vitro and in vivo. In zebrafish cell lines (PAC-2, Z3), per1 (also termed per4), per2, and per3 transcripts have been shown to exhibit a robust oscillation under light/dark (LD) cycles [9,16,17]. Expression of per1 and per3 seems to be driven by a circadian oscillator since their rhythmic expression persists following transfer to constant darkness (DD) and their daily increase in transcript levels anticipates the light phase [9,16,17]. In contrast, per2 mRNA levels are thought to increase only when cells are exposed to light and its rhythmic expression dampens immediately following transfer to DD. Consistently, in zebrafish embryos, per2 mRNA expression is induced in response to light exposure throughout the body, cranial areas and particularly in the pineal gland [18]. Pineal per2 mRNA levels increase rapidly following ''lights on'', reaching a peak after 3 h while they remain undetectable under DD. Moreover, knock-down analysis has demonstrated that per2 expression is required for the light-induced developmental maturation of the pineal clock [18]. These studies indicate that per2 is involved in the light-input pathway of the circadian clock in zebrafish. Furthermore, the synchronization and onset of circadian rhythmicity in the pineal gland depends on light induction of per2 expression. The mechanism by which light regulates per2 expression is unknown.
With the goal of improving our understanding of the mechanisms underlying light entrainment in the vertebrate circadian clock, we investigated the regulation of zebrafish per2 both in vivo and in vitro. The per2 promoter regulatory region was identified and two distinct transcription factor control mechanisms were shown to contribute to light-driven per2 expression. Our results point to a crosstalk between circadian clock and light-driven regulation as being a key feature of the per2 gene regulation.

Identification of the per2 Promoter
The per2 gene plays a key role in mediating the effects of light on the circadian clock in zebrafish [18]. An important step towards exploring the mechanisms whereby light regulates expression of this gene is to isolate and characterize its promoter. A genomic fragment incorporating the putative promoter (1.8 kb 59 flanking genomic DNA) and the first exon (164 bp) of per2 was subcloned upstream of an EGFP reporter gene creating the 21.8per2:EGFP construct. Microinjection of 21.8per2:EGFP resulted in 94% EGFP-positive embryos, each exhibiting robust transient EGFP expression scattered throughout the body without any apparent tissue specificity as seen under a fluorescent dissecting microscope ( Figure 1). This EGFP expression pattern is consistent with the documented widespread expression of zebrafish per2 mRNA in zebrafish embryos and larvae [18]. In order to identify the minimal promoter required for expression, two additional shorter constructs were tested by microinjection. Embryos injected with 20.9per2:EGFP and 20.43per2:EGFP resulted in 93% and 94% EGFP-positive embryos, respectively. Expression patterns and intensities were similar to those observed with the 21.8per2:EGFPinjected embryos (Figure 1). These results indicate that the 430 bp 59-flanking region minimal promoter is sufficient to drive expression.

Author Summary
Light is the principal signal used by animals to synchronize their circadian clocks with the day/night environment. Central to this vital property is the ability of light to trigger changes in gene expression. However, we still lack a complete understanding of how this occurs. The zebrafish is particularly interesting in this regard since direct light exposure induces the expression of clock genes in most of its tissues and in turn adjusts the phase of the intrinsic clocks. Here, by studying the promoter of one key lightregulated zebrafish clock gene, per2, we have identified a Light Responsive Module (LRM) that is necessary and sufficient for light controlled expression. Interestingly, the LRM is also highly conserved in the per2 genes of other vertebrates that lack widespread light-sensing tissues. In addition, the human LRM can substitute for its zebrafish counterpart to confer direct light regulation of gene expression in zebrafish cells. The LRM contains E-and Dbox enhancers critical for its function. While the E-box is a target of clock regulation, the D-box directs light driven expression. We show that the expression of the D-box binding transcription factor, tef, is itself induced by light and is essential for normal light-induced per2 expression. These results advance our understanding of the mechanisms underlying entrainment by light and how light-regulated clock gene expression has evolved in vertebrates. The per2 Promoter Drives Ubiquitous Expression that is Enhanced in the Pineal Gland Using the Tol2 transposon system [19], a transgenic line, Tg(20.43per2:EGFP)tlv1 containing the 430 bp promoter fragment fused to EGFP, was generated. This line exhibits ubiquitous EGFP expression with enhanced expression in the pineal gland in embryos, larvae, and adults ( Figure 2 and Video S1). This expression pattern is in accordance with previous whole mount in situ hybridization (ISH) analyses for per2 [18]. Tg(20.43per2: EGFP)tlv1 were crossed with Tg(aanat2:mRFP)y164 (kindly provided by Reiko Toyama, NIH), which exhibits red fluorescence exclusively in the pineal gland, to create a double transgenic line. Confocal in vivo analysis reveals co-localized EGFP and mRFP expression in the pineal gland ( Figure S1). These results indicate that per2 is expressed in virtually all cells that contain an autonomous peripheral circadian oscillator and exhibits enhanced expression in the melatonin-producing, master clock-containing, photoreceptor cells of the pineal gland.

The per2 Promoter is Light Responsive In Vivo
To determine whether the 20.43 kb per2 promoter is light responsive in vivo, Tg(20.43per2:EGFP)tlv1 embryos were kept under LD cycles during the first 2 d of development. On the third day of development, entrained embryos were either exposed to 2 h of light at ZT 0 or maintained in DD. Differences in EGFP fluorescence were not observed, most likely due to the stability of the EGFP protein. However, analysis of egfp mRNA using whole mount ISH revealed increased expression of the transgene after light exposure, notably in the pineal gland. Quantification of the pineal signal revealed a 22-fold increase in egfp mRNA levels after a 2 h light pulse with respect to a DD control (n = 16, p,0.05 by one-way ANOVA) ( Figure 3). These results are in agreement with those shown previously by whole mount ISH for the endogenous per2 mRNA [18] and indicate that the 20.43per2 promoter directs light-induced expression in vivo.

The per2 Promoter Is Light-Responsive and Clock-Driven
Light-inducible per2 expression and a directly light-entrainable clock have been encountered in several zebrafish cell lines [9,17]. Stable transfection of these cells with luciferase reporter constructs followed by monitoring expression under various lighting conditions with a live cell bioluminescence assay represents a powerful approach to assess a promoter's light responsiveness [17]. Thus, we chose to stably transfect PAC-2 zebrafish cells with constructs that contain the per2 promoter cloned upstream of the luciferase gene ( Table 1). Expression of the reporter constructs increased during the beginning of the light phase (peaking at ZT 5.360.2 and 6.460.4, respectively) and then decreased during the dark phase ( Figure 4). These results indicate that the minimal per2 promoter is light-responsive in PAC-2 cells. Consistently, transfer to DD leads to a very rapid attenuation of rhythmic expression. Interestingly, under LD conditions the decrease in luciferase activity observed during the cycling anticipates the end of the light phase. In addition, a slight increase in luciferase activity is observed before the beginning of the light phase. Importantly, rhythmicity driven by the minimal promoter was maintained under LL (t = 23.2560.57 h) ( Figure 4B), but not when cells were transferred to DD ( Figure 4A). These results indicate that, in addition to being light driven, the per2 promoter also shows regulation by the circadian clock.  . The per2 minimal promoter is light-induced in vivo: quantification of pineal EGFP mRNA levels. Tg(-0.43per2:EGFP)tlv1 embryos were entrained by exposure to two LD cycles. At the beginning of the third day of development, embryos were either exposed to light or kept in darkness. After 2 h (ZT 2 and CT 2) embryos were fixed and subjected to whole mount ISH for egfp mRNA. Signal intensities, determined using ImageJ software, revealed a 22-fold increase in pineal egfp mRNA levels after a 2 h light pulse (n = 16, p,0.01 by ANOVA) relative to pineal egfp expression in the DD controls. Error bars represent SE. Dorsal views of representative whole mounts are shown below. doi:10.1371/journal.pbio.1000223.g003 Identification of a Light-Responsive Region within the per2 Promoter To define more accurately the light responsive cis-regulatory sequences, a series of deletions of the per2 promoter was prepared. First, a series of constructs carrying partially overlapping deletions within the 20.43per2:Luc wild-type construct was generated. Deletion constructs, stably transfected in PAC-2 cells, were then tested for light-regulated expression. Cells were exposed to LD conditions and the profiles of bioluminescence were compared with those of cells transfected using the wild-type 20.43per2:Luc construct ( Figures 5 and 6). In both Deletion 4 and Deletion 5, the characteristic robust increase in expression observed following ''lights on'' in the wild-type promoter was absent. In addition, in the case of Deletion 4 cycling expression was severely attenuated, while for Deletion 5 the phase of rhythmic expression was significantly shifted ( Figure 5A, Table 1). The remaining deletions did not affect either light-induced expression or the phase of rhythmic expression. These results point to a region of 145 bp (289 to 2233) containing the elements that are necessary for lightresponsiveness.
To better define the light-responsive region, a second series of ,20 bp sub-deletions within this 145 bp region were prepared and tested, in the context of the wild-type 0.43 kb promoter. PAC-2 cells were stably transfected with the sub-deletion constructs and exposed to LD cycles. Sub-deletions 4.3, 4.4, 4.5, and 5.1 resulted in a loss or severe reduction of light-induced expression in LD cycle conditions, whereas other sub-deletions within the wild-type promoter failed to show this effect ( Figures 5B and 6, Table 1). These results delimit the core of the light responsive region to a 67 bp segment (2134 to 2200), which we have termed ''light responsive module'' (LRM).

The LRM Is Sufficient to Drive Light-Dependent Rhythmic Expression
An 87 bp region, containing the LRM core and flanking sequences (2218 to 2132), was inserted, as three copies in tandem, in either the forward or reverse orientation into the TATA-box-containing pLuc-MCS plasmid (Stratagene) creating the 3xLRM(fwd):Luc and 3xLRM(rev):Luc constructs. Mean bioluminescence levels in cells transfected with either 3xLRM(fwd):Luc or 3xLRM(rev):Luc were lower (1%-2.5%) than those of the wildtype per2 promoter. Nevertheless, both LRM constructs presented a light-driven expression pattern that was similar to that of the 20.43per2:Luc construct ( Figure 7, Table 1), indicating that in addition to being necessary for light induction, the LRM also contains enhancer elements that are sufficient to direct light-driven rhythmic expression.

The LRM Sequence and Function Are Evolutionarily Conserved
Given the previous reports of elements of the per2 promoter being evolutionarily conserved in mammalian species [20], we specifically examined whether the LRM core might itself represent an evolutionarily conserved regulatory element. Thus, the LRM core, with its flanking sequences (2184 to 2120), was compared with the per2 promoter sequences of chicken, mice, rat, and human ( Figure 8). Interestingly, in all per2 promoters, a region proximal to the transcription start site that exhibits high sequence identity (60%-63%) with the LRM was identified. Three sequences of interest are present within this conserved region of the per2 promoter. These include (1) an E-box CAYGTG (where Y is a pyrimidine), known to mediate the activity of the positively acting clock component proteins [21], and (2) the sequence CTTATGTAAA that is perfectly conserved among the per2 promoters and where 8/10 bases match the consensus D-box RT(G/T)AYGTAAY (where R is a purine). The D-box is the binding site for bZIP transcription factors of the proline and acid amino acid-rich (PAR) subfamily (DBP; TEF and HLF) and E4BP4. These transcription factors have been implicated in light-regulated phase shifting of the clock and in clock output pathways [22,23]. (3) A CCAAT box known to play a role as a basal promoter element in polII transcribed genes [24].
Given the conservation of the LRM sequence between several vertebrate species, we questioned whether the functionality of LRM is also evolutionarily conserved. A fragment of 55 bp within 20.43per2:Luc, containing the core LRM (including the E-and Dboxes), was replaced by the corresponding 44 bp region of the human per2 promoter (hLRM). PAC-2 cells were stably transfected with the resulting construct, 20.43per2(hLRM):Luc, and the expression profile was compared with that driven by the wildtype construct. Remarkably, expression profiles of the two constructs were similar under all the lighting conditions tested: LD, DD, and LL ( Figure 9, Table 1). These results indicate that, in the context of zebrafish cells, the hLRM can mediate both light and circadian clock-directed expression. The LRM E-and D-boxes Are Essential for the Light Response of the per2 Promoter To determine if any of the evolutionarily conserved putative elements (E-box, D-box, CCAAT-box, as well as an additional conserved CATGG sequence) within the LRM are required for its function, each element was point-mutated and the resulting promoter/reporter constructs were independently tested in stably transfected PAC-2 cells. Cells were then exposed to LD cycles, luciferase activity was monitored, and the expression profile was compared with those of cells transfected using the wild-type 20.43per2:Luc construct ( Figure 10, Table 1). While point mutations disrupting the CCAAT box and the CATGG sequence had no major effect on the expression pattern under LD conditions, mutations of the E-box and D-box, singly or in combination, had a major disruptive effect on light-responsiveness. These results indicate that the E-box and D-box are crucial for LRM function, i.e., light-induction and clock regulation of the per2 promoter.
We next asked whether simply placing a canonical E-box adjacent to a canonical D-box element would be sufficient to confer the light-regulated pattern of expression observed for the per2 promoter. To address this question, we constructed a heterologous promoter/reporter construct based on a canonical E-box (CACGTG) and a D-box (CTTATGTAAA) separated by 17 bp of non-per2 LRM-derived sequence. Four tandem repeats of this artificial module were cloned upstream of a TATA box element, thereby driving transcription of a luciferase reporter gene (4xE/D-box). PAC-2 cells transfected with this construct were then exposed to various lighting regimes and luciferase activity was monitored ( Figure 11, upper panel). Strikingly, a light-driven pattern of expression was observed that closely resembles that of the per2 promoter, demonstrating the importance of E-and Dboxes in LRM function.
In order to dissect the individual contribution of the E-and Dboxes to the expression profile of the 4xE/D construct, we compared its profile with that of constructs containing either multimerized E-box (4xE-box, Figure 11, lower panel) or D-box (6xD-box, Figure 11, middle panel) elements. Consistent with previous results [17], the E-box directs rhythmic expression both in LD and free-running conditions (DD and LL) ( Figure 11). Direct comparison of this expression pattern with that of the 4xE/ D-box construct revealed clear differences: notably in the phase of the rhythms under LD conditions and the persistence of rhythmicity in DD. The D-box reporter, on the other hand, directs a robust light-driven pattern of expression. Under LD cycles, the bioluminescence levels increase 1 h after ''lights on'' and continue to increase until ''lights off'', at which point they start to decline. Immediately following transfer to DD, levels decrease progressively with no cycling. Interestingly, however, upon transfer from LD to LL conditions, expression levels increase with a steplike profile hinting at some regulation by the endogenous circadian clock.

E-and D-Box Binding Factors and Regulation of the LRM
What are the transcriptional control mechanisms targeting the E-and D-box enhancer elements? The E-box enhancer has been widely implicated as the regulatory target of the BMAL/CLOCK heterodimer. To confirm a role for CLOCK and BMAL in the regulation of the LRM via the canonical E-box, 20.43per2:Luc was co-transfected into mammalian COS-7 cells with either an empty vector (pcDNA3.1) or a mixture of zebrafish BMAL2 and CLOCK1 expression vectors ( Figure 12). The COS-7 cell system was selected in order to test transcriptional activation in an endogenous ''clock-free'' environment [25]. Luciferase expression driven by the wild-type per2 promoter was enhanced 5-fold (p,0.001 by three-way ANOVA) in the presence of BMAL/ CLOCK ( Figure 12). However, when the E-box was mutagenized (20.43per2ME:Luc), the observed BMAL/CLOCK activation was 1.7-fold, significantly (p,0.001) lower than the activation of 20.43per2:Luc. These results suggest that the LRM E-box is  Given the striking capacity of the D-box to confer light-driven expression in zebrafish, we chose to focus on how the function of D-box binding transcription factors responds to light. From studies in other vertebrates, bZIP transcription factors of the PAR domain subfamily and E4BP4 are known to bind to D-box elements. One candidate D-box activator, previously identified in zebrafish [26], is Thyrotroph Embryonic Factor (TEF). As a first step to evaluate whether TEF might be linked with light-regulated gene expression in zebrafish cells, we initially studied the effect of light on tef mRNA expression in whole embryos and PAC-2 cells. Whole mount ISH analysis was performed on 50-74 hpf embryos exposed to LD cycles or DD ( Figure 13A). Tef expression appeared to be widely distributed throughout the body (unpublished data) and cranial areas with augmented expression in the pineal gland ( Figure 13A). Interestingly, this pattern of expression is similar to that of per2, suggesting a possible link between these two genes (see also Figures 2 and 3). In addition, under LD cycles, tef levels increased rapidly following ''lights on'', and in the pineal gland and in all other expression regions, reached maximum levels at ZT 2 and declined to undetectable levels by the end of the light phase ( Figure 13A), indicating that tef expression is induced by light. An anticipatory behaviour of tef expression was observed: levels begin to increase before ''lights on'' (ZT22). Moreover, under DD tef mRNA levels exhibit low amplitude cycling, peaking at the beginning of the subjective day, indicating that tef expression is also regulated by the circadian oscillator.
To determine whether tef is light-induced in our experimental zebrafish cell system, qRT-PCR was performed upon cDNA prepared from PAC-2 cells that were kept in darkness or exposed to light for different time periods ( Figure 13B). An increase in tef mRNA levels was observed when cells were exposed to light, reaching its highest levels after 4 h of exposure. Together, these results indicate that tef is up-regulated predominantly by light and partially by the circadian clock. Importantly, the phase of rhythmic tef expression precedes that of per2. These results provide a first clue for TEF being a regulatory factor contributing to the light-driven expression of per2.
We next tested whether TEF can activate the per2 promoter through the LRM D-box. COS-7 cells were co-transfected with the per2 promoter reporter constructs and a TEF expression vector ( Figure 12). Wild-type per2 promoter-driven luciferase expression was enhanced 3.3-fold while a non-significant increase (1.5 fold) was observed when the D-box was mutagenized (20.43per2MD:Luc), suggesting that TEF acts as an activator of per2 via the LRM D-box ( Figure 12). Additionally, when cells were co-transfected with a combination of the BMAL/CLOCK complex along with TEF, an additive effect was observed indicating that both activators, BMAL/ CLOCK and TEF, contribute simultaneously to the LRM activation through the E-box and D-box, respectively.
Aiming to test the ability of TEF to bind the LRM D-box sequences, electrophoretic mobility shift assays (EMSAs) were conducted using reticulocyte lysate-synthesized TEF. A radioactively labelled 50 bp DNA fragment of the LRM sequence was used as a DNA binding probe. Incubation of the LRM probe with in vitro synthesized TEF formed a single DNA-protein complex ( Figure 14, Lane 2). To verify that TEF binds specifically to the D-box element within the LRM, unlabeled double-stranded oligonucleotides were used as competitors. While the LRM (Lane 3) or D-box (Lane 4) acted as effective competitors, an LRM probe containing a mutated D-box (Lane 6), as well as a SP1 (Lane 5) and a single E-box (Lane 7) binding site, was unable to compete. Together our results indicate that TEF is able to specifically bind the D-box element in the LRM core sequence and thereby activate per2 expression.

Tef: A Light-Regulated D-Box-Binding Transcriptional Activator of per2
In order to more directly test whether TEF contributes to lightinduced expression of per2 in vivo, we adopted a knock-down strategy. Morpholino-modified antisense oligonucleotides (MOs) corresponding to tef were injected into transgenic Tg(20.43per2:-EGFP)tlv1 embryos. These morpholino oligonucleotides blocked the splicing of the second intron of the tef gene and thereby introduced a premature stop codon into the tef mRNA ( Figure 15A). Injected embryos were then entrained to two LD cycles. At the beginning of the third day of development, the embryos were either exposed to 2 h of light or kept in darkness. Embryos were then fixed and egfp mRNA expression was assayed by whole mount ISH. In addition, the efficiency of tef (E2I2) MO directed against the splice site was evaluated by RT-PCR (unpublished data). Remarkably, knockdown of tef completely abolished the light-induced expression driven by the per2 promoter ( Figure 15B) but had no effect on the development or morphology of the pineal gland, as indicated by the normal expression pattern of otx5 ( Figure 15C). By contrast, injection of a control MO had no effect on the light-induced expression of the per2 promoter ( Figure 15B). These results indicate that TEF is involved and required in the light-induced pathway of per2.

Discussion
In the last two decades, a great deal has been learned about the components and molecular organization of the circadian clock. However, the pathways by which the circadian oscillator is synchronized by light are still not fully understood. By focusing on the mechanisms whereby light induces the transcription of per2, this study aimed to identify regulatory elements that mediate light induction using the zebrafish as a model. Our in vivo analyses have defined the minimal per2 promoter fragment that is light induced both in transgenic embryos and also in cell lines. In addition to the robust effect of light on per2 expression, experiments under LD and LL conditions revealed that per2 expression is also controlled by a circadian oscillator. An LRM within the promoter was identified and shown to be both necessary and sufficient for light-induced expression and clock regulation. We show that the zebrafish LRM is evolutionarily conserved and indeed can be functionally replaced by the LRM from the human per2 gene. Our mutational analyses indicate that adjacent D-box and E-box elements underlie the LRM function. We demonstrate that the D-box contributes to light-driven transcriptional control while the E-box mediates circadian clock control. The E-box is the target of the BMAL/CLOCK heterodimer, the positive components of the core circadian oscillator, while the bZIP PAR domain transcriptional activator TEF contributes to light-dependent regulation of the D-box.

D-Boxes: Light-Responsive Enhancer Elements in Zebrafish
Our study of the light-regulated per2 promoter has revealed that D-box enhancers direct a light-driven expression pattern in zebrafish cells. This result is important for our general understanding of how transcriptional regulatory mechanisms have evolved in the vertebrate circadian clock. In mouse, D-box binding factors include bZIP transcriptional activators of the PAR domain subfamily (DBP, TEF, HLF) and the E4BP4 repressor. Detailed reverse genetics studies clearly point out that the D-box mediates circadian clock output [27]. CLOCK and BMAL regulate expression of DBP via E-box enhancers in its promoter and thereby drive a high amplitude circadian rhythm of DBP expression that in turn confers circadian clock control on the expression of DBP target genes [28]. Consistently, in mammalian cell culture assays, a D-box reporter construct similar to the 6xDbox described here displays a characteristic clock-regulated rhythmic expression profile [29]. Thus our findings in the zebrafish model, linking D-box regulation predominantly with the light input pathway, points to plasticity in the precise role of this particular gene expression regulatory pathway during the course of evolution. The ancestral genome duplication that occurred during the evolution of the teleost lineage and the resulting extra copies of many zebrafish genes including clock genes implies that there might also be extra copies of D-box binding factors. Thus while our data clearly implicate tef in lightdependent regulation of per2, it will be fascinating to identify and explore the role of additional teleost D-box regulatory factors and compare their function with the mammalian counterparts.

Per2: Light-Dependent Clock Regulation
Previous studies in zebrafish have concluded that transcription of the per2 gene is exclusively light regulated and is not under circadian clock control [9,18,30]. In the current study, we have used a live cell bioluminescence assay of stably transfected zebrafish cell lines that provides us with a high-resolution, realtime image of changes in per2 promoter-driven expression. Using this assay we find that the rhythmic activity of the zebrafish per2 promoter under LD exhibits anticipatory behaviour: expression levels start decreasing before lights off and begin to rise before lights on. Moreover, a circadian rhythm of expression is also is consistent with previous studies demonstrating that the zebrafish circadian clock continues to direct circadian rhythms of gene expression under LL conditions [17,31]. However, it remains unclear whether clock regulation by the D-boxes in LL represents a direct effect of the core clock machinery or might reflect an indirect effect, for example on the sensitivity of the light input pathway. Taken together, these results clearly implicate the circadian oscillator in the regulation of per2. Nevertheless, in contrast to purely E-box-regulated expression, rhythmic expression of per2 is absent under DD, thus pointing to a hierarchic control, where light enables the circadian clock regulation of the per2 promoter.

The LRM Regulatory Mechanism
Recent studies in mouse and chicken addressing the mechanism underlying rhythmic per2 transcription have implicated three signalling pathways. One is the cAMP pathway that may stimulate per2 expression under certain conditions [32,33]. A second pathway involves CLOCK/BMAL that activate the expression of mouse per2 via a non-canonical E-box [34]. A third pathway involves E4BP4 that acts as a suppressor of per2 promoter activity in both chicken and mouse through binding to a D-box [20,22]. This possibly reflects the existence of a feedback between clock outputs and the core clock mechanism itself that is mediated by Dbox binding factors [27]. In the mouse and chicken models, the elements that mediate these activities are located outside of the conserved LRM and also separated from each other in terms of linear DNA sequence. Thus, the E-box is situated in the proximal promoter region of the mouse per2 gene, and the identified D-box is situated in a distal promoter region and first intron of the chicken and mouse per2 genes, respectively. The presence of E-box and D-box sequences within the conserved LRM was mentioned in the mouse and chicken studies, but this D-box was not shown to bind E4BP4.
Importantly, the findings of the chicken and mouse studies concern the effects of light perceived by dedicated photoreceptor cells and then relayed indirectly to responsive cells. However, the findings of the current study reveal the effects of direct exposure of cells to light, made possible by the use of zebrafish cell lines. Although the experimental systems that were used in the chicken and mouse studies failed to assign functionality to the LRM and its constituting elements, the remarkable evolutionary conservation of the LRM and the fact that the human LRM sequence functioned as a light-regulated enhancer in the context of the zebrafish cells suggests that mammalian LRMs may well also play an important regulatory role in entrainment. Alternatively, during evolution, ancestral light-responsive mechanisms may have been subverted to  . The LRM sequence is highly conserved among vertebrates. Multiple alignment of LRM-like sequences derived from chicken, human, mouse, rat, and zebrafish per2 promoters. The conserved sequence is indicated by a yellow background. Studied elements; CATGG sequence, E-box, D-box, and CCAAT-box are bordered. Note that the distance between the E-box and D-box in the zebrafish per2 promoter differs from that of the other species by 11 bp, which is equivalent to one DNA helical turn. Numbers on both ends of the LRMs indicate the distances from the putative transcription start sites. doi:10.1371/journal.pbio.1000223.g008 respond to signals other than light. It would therefore be interesting to determine whether TEF binds the D-box in LRMs of chicken and mice. Our data demonstrate that proximally spaced D-and E-boxes are sufficient to confer the light-and clockregulated expression pattern as shown by the per2 promoter. This expression pattern represents a combination of the regulation exerted by the two separate enhancers. It is likely that the E-box confers the anticipatory rise in expression prior to ''lights on'' and the decline in expression prior to ''lights off'' under LD conditions and the low amplitude cycle observed during the first few cycles after transfer to LL. The D-box confers the strong up-regulation of expression after ''lights on'' and the rapid attenuation of expression in DD conditions as well as contributing to rhythmic expression under LL conditions. We thus propose the following model for the LRM function and regulation by the E-box and Dbox binding factors ( Figure 16). Interaction between the E-and Dbox enhancer elements confers a hierarchic control in which light enables circadian clock regulation. The D-box represents the binding site for a family of bZIP transcription factors; the current results implicate the light-inducible D-box-binding factor TEF as a key player in activating per2 expression. Thus in the proposed model, at the onset of the day, light exposure activates TEF and thereby results in a D-box-mediated transcriptional activation of per2. D-box regulatory factors in turn interact with the CLOCK/ BMAL heterodimer through the proximal E-box. The clockdriven daily changes in CLOCK/BMAL transcriptional activation contribute to the elevation of per2 mRNA levels in the beginning of the light phase and result in the anticipatory reduction of per2 during the light phase. Clock-regulated changes in tef gene expression may further contribute to the rhythms of per2 expression observed under LD and LL conditions. Importantly, previous results have demonstrated that upon blocking of de novo protein synthesis, per2 light-induced expression persists although with a delay in the arrival at peak expression levels [35,36]. This is consistent with TEF already being expressed in the cell prior to light exposure and that light may also serve to activate TEF via post-translational mechanisms.

A General Mechanism for Light-Regulated Gene Expression in Zebrafish?
In the intact animal, per2 expression increases ubiquitously following light exposure and is mainly enhanced in the pineal gland. Furthermore, tef mRNA appears to be widely expressed with high levels in the pineal gland, where its light-induced expression slightly precedes that of per2. The proposed mechanism ( Figure 16) that is based on experiments in cell lines that represent peripheral clock-containing tissues may therefore be applicable to the increased ubiquitous expression. The enhanced expression in the pineal gland is in accordance with the fact that it constitutes a dedicated photoreceptive organ and a central clock tissue in fish. The relevance of the proposed mechanism to light entrainment of the central clock in the pineal gland remains to be determined. Several other genes have been shown to be light inducible in zebrafish cells and tissues. These include the clock gene cry1a, suggesting that light may entrain the circadian clock mechanism by regulating the transcription of multiple clock gene elements. Interestingly however, unlike the case of per2, treatment with the protein synthesis inhibitor, cycloheximide, completely abolishes light-induced cry1a expression [35]. This result implies that there may be more than one mechanism whereby light regulates gene expression. Other non-clock-related genes such as 6-4 DNA photolyase have also been demonstrated to be inducible by acute exposure of zebrafish cells to light [37]. The induction of genes involved in the repair of UV damaged DNA by visible wavelengths of light seems likely to represent part of an adaptive strategy to optimize the cellular response of DNA damage repair during exposure to sunlight when it is most needed. It will therefore be fascinating to determine whether the LRM sequences or even Dbox elements also play a more general role in light-dependent changes in zebrafish cell physiology.
In summary, our study has revealed a novel mechanism that combines light-induced and clock-regulated transcriptional control. The proposed mechanism may lead to a better understanding of the phenomenon of gated light entrainment of the circadian clock and possibly other general photic cellular responses.

Fish Maintenance
Adult zebrafish were raised in a recirculation water system under 12 : 12 h LD cycles at 28uC and fed twice a day. To produce embryos, male and female zebrafish were paired in the evening, and spawning occurred the next day within 1 h after ''lights on''. Embryos were placed in 10 cm Petri dishes with egg water containing methylene blue (0.3 ppm) and raised in a lightcontrolled incubator at 28uC (light intensity, 12 W/m2). To prevent pigmentation, the fish water was supplemented with 0.2 mM phenylthiourea during the first day of development. Tg(20.43per2:EGFP). This construct was prepared as described for 20.43per2:EGFP except for subcloning into BamHI/XhoI-digested pT2ALR150G [38]. This construct was later used for the preparation of transgenic lines utilizing the Tol2 system (see below) [38]. Figure 11. Adjacent E-and D-boxes confer LRM mediated transcription. PAC-2 cells were transiently transfected with a series of heterologous, E-and D-box-containing, luciferase reporter constructs. Transfected cells were then exposed to various lighting conditions as indicated by the black/white bars above the panels, and bioluminescence was monitored using a live-cell automated bioluminescence assay. Transfection results from heterologous constructs incorporating alternating E-and D-boxes (4xE/D-box) as well as multimerized D-boxes (6xD-box) and E-boxes-(4xE-box) are presented. Normalized luciferase activity (%) is plotted on the y-axis of each graph. Time (days) is plotted on the x-axis. doi:10.1371/journal.pbio.1000223.g011

DNA Constructs: Constructs for In Vivo Analyses
Constructs for In Vitro Analyses 21.7per2:Luc. A fragment containing 1,700 bp of the 59 flanking region of per2 was subcloned into pGL3basic (Promega) upstream to the luciferase gene. A fragment containing 1,571 bp of the 59 flanking region and 129 bp of the 59 UTR of the per2 gene was PCR amplified using the Universal GenomeWalker Kit (Clontech) according to the manufacturer's instructions, using the specific primer PER25UTREX1R (59-AGCCTTGCTTCAA-AACAGGTCTCAGTG -39) in combination with the AP1 primer (Clontech). The resulting product was subcloned into the pGEM-T easy vector (Promega) and sequenced. The 1,700 bp fragment was then double digested with NotI, blunt ended by Klenow treatment (New England Biolabs), and then ligated into Sma1-digested pGL3.
20.43per2:Luc. The per2 minimal promoter was subcloned into pGL3basic upstream of the luciferase reporter gene. A fragment containing 431 bp of the 59 flanking region and 164 bp of 59 UTR of the per2 gene was PCR amplified using 20.43per2:EGFP as a template and a set of specific primers: PER2PRO(595)bF, containing a NheI restriction site (59-cgcgctagcaacctattggatcacttcgaggcatcac-39), and PER2EX1bR, containing a BglII restriction site. The PCR product was double-digested with NheI and BglII and ligated into NheI/BglIIdigested AVP-pGL3 [39] to replace the AVP sequence.

Deletion Mutations
20.43per2(D2437/2393):Luc (Deletion 1). A mutation creating a NheI recognition site located at position 2393 in 20.43per2:Luc was accomplished using a set of two complementary primers: Nhe1Del1F (59-cccgcacgtccatgctagctattgtaaaagc-39) and Nhe1Del1R (59-gcttttacaatagctagcatggacgtgcggg-39), which incorporated the desired mutation using the QuikChange Site-directed mutagenesis kit (Stratagene) as instructed by the manufacturer. The construct was then digested using NheI, and re-ligated resulting in a 45 bp deletion of the 59 end of the 430 bp promoter.
All subsequent deletion, sub-deletion, and point mutation constructs were also generated using the QuikChange Sitedirected mutagenesis kit.

Point Mutations
20.43per2-ME:Luc. The E-box located at position 2174 to 2169 was mutated (from CACGTG to CTCGAG). Two Figure 13. Temporal and spatial expression pattern of TEF under LD and DD cycles. (A) During the first 2 d of development, embryos were exposed to LD cycles. During the third and fourth days of development, embryos were kept under LD or under DD, sampled at 4 h intervals (50-74 hpf), and subjected to whole mount ISH for tef (Genebank Accession number U43671). White bars represent light phase, black bars represent dark phase, and gray bars represent subjective day (ZT, zeitgeber time). Red arrows indicate expression in the pineal gland. Tef is expressed throughout the body and cranial areas with augmented expression in the pineal gland and exhibits a circadian expression pattern with higher levels at the beginning of the subjective day (DD). Under LD, tef expression increases before lights on (ZT 2) and the amplitude of rhythmicity increases. (B) PAC-2 cells were maintained for 5 d in DD. Subsequently, total RNA was extracted from cells kept in darkness or exposed to light for different time periods (1, 2, 4, 6, 8 h). Quantification of tef mRNA levels was performed using qRT-PCR. The mRNA levels in each sample are expressed relative to the level of cells kept in DD. Values shown are the mean from three independent cell pools. Error bars represent SE. These results indicate that tef mRNA levels increase following exposure to light, peaking at 4 h of exposure. Statistical analysis was performed by one sample t-test. All light-treated samples showed significantly higher tef mRNA expression levels relative to DD controls.

Recombinant Promoter Constructs
3xLRM(fwd):Luc and 3xLRM(rev):Luc. Three copies of the 87 bp sequence containing the LRM core and flanking sequences (corresponding to nt 2218 to 2132) were inserted in tandem into pLuc-MCS (Stratagene). Using the 20.43per2:Luc plasmid as a template, the 87 bp fragment was PCR amplified using primers XmaI(LRM)R (59-cgccccggggcccgtcagtctttttacataagtgt-39) and XhoI(LRM)F (59-cgcctcgagatttgtccatacgcagcgca-39). The product was subcloned into pGEM-T-easy creating the LRM-pGEM construct. The same fragment was amplified using XhoI(LRM)F and XhoI(LRM)R (59-cgcctcgaggcccgtcagtctttttacataagtgt-39). The PCR product was digested with XhoI and two additional copies were ligated into XhoI digested LRM-pGEM. The fragment containing three copies of the LRM was digested with EcoRI and ligated into EcoRI-digested pLuc-MCS, resulting in clones with inserts in both the forward and reverse orientation.

Heterologous Promoter Reporter Constructs
4xE-box. This E-box reporter construct contains four copies of the E-box sequence from the period4 promoter of zebrafish cloned into the vector pLucMCS. The precise sequence and construction details have been reported elsewhere [17].

In Vivo Transient Expression Assay and Examination of Embryos
Transient expression assays of the EGFP-reporter constructs were performed by microinjection of zebrafish embryos as described previously [41,42]. Following injections, embryos (,300) were incubated in a 10 cm plastic dish at 28uC. Embryos were examined and graded during the light phase on Day 2-5 post-fertilization. Green fluorescence in live embryos was detected as previously described [41]. Since per2 is ubiquitously expressed [18,43], embryos were considered ''positive'' if fluorescence was detected anywhere in the embryo's body. Embryos were housed in 10 cm dishes (,60 embryos per dish) for additional daily observation and validation until Day 5. Results are presented as percentage of EGFP expressing embryos.

Preparation of Transgenic Fish
The transgenic line Tg(-0.43per2:EGFP)tlv1 was generated using the Tol2 system as described [19]. Plasmids were kindly provided by Koichi Kawakami. Briefly, transposase mRNA was synthesized in vitro using mMESSAGE mMACHINE SP6 Kit (Ambion Inc.). Approximately 1 nl of a DNA/RNA solution containing 25 ng/ml of Tg(-0.43per2:EGFP) circular DNA and 25 ng/ml transposase mRNA were injected into each fertilized egg. Founder (F0) fish were crossed and EGFP expressing progeny (F1) were raised to adulthood. F2 progeny from out crossed F1 fish were used.

Morpholino Design and Injection
Gene knock-down experiments were performed using morpholino-modified antisense oligonucleotides (MO; Gene Tools): Gene Tools standard control MO (59-ctcttacctcagttacaatttata-39) and Tef (E2I2) MO (59-agtgttctgttcttacagacctgat-39), which was designed to target the exon 2-intron 2 boundary to interfere with splicing. MO injected embryos (2 nl, 1 mM) were incubated and fixed as described above. Efficiency of tef (E2I2) MO, directed against the splice site, was evaluated by RT-PCR. Uninjected, tef (E2I2) MO and control MO-injected embryos were entrained to two LD cycles and sampled at 50 hpf, after 2 h light exposure or darkness, and total RNA was extracted as previously described [44]. DNA fragments were then PCR-amplified by using primers directed to exons 1 and 4 of tef (TEFa F and TEFa R described above). PCR products were cut out from gel, purified, and sequenced.
Whole Mount ISH Quantification and Statistical Analysis mRNA expression was detected by whole mount ISH as described [18,45]. Embryos/larvae exposed to different photic regimes were fixed overnight in 4% paraformaldehyde and stored in 100% methanol. Whole mount ISH was performed with a dioxygenin (DIG)-labelled probe at a concentration of 1 ng/ml. Detection and documentation of the signal was performed as described [18,45]. The ISH signal, expressed as optical density, was quantified by using ImageJ software (National Institutes of Health, Bethesda, MD, USA) as described [18,45]. Statistical differences in signal intensities between treatments were determined by one-way ANOVA or Kruskal-Wallis test followed by Mann-Whitney test. Results are expressed as mean total optical density 6 standard error.

In Vitro Transient Transfection Assays in COS-7 Cells
COS-7 cells were plated at a density of 3610 4 cells per well in a 24-well plate (Costar) and transfected 24 h later with a mixture containing Lipofectamine Plus (1.25/2.5 ml) reagents (Invitrogen), 10 ng of 20.43per2:Luc, 20.43per2-ME:Luc or 20.43per2-MallE: Luc, and 0.75 mg of a 1:1:1 expression vector mix of zebrafish TEF, BMAL2, and CLOCK1 [16] or empty vector pcDNA 3.1 (Invitrogen) in 50 ml of Vitacell Dulbecco's modified Eagle's medium (ATCC) without fetal bovine serum. On the following day, 0.5 ml of culture medium (Vitacell Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum) was added to each well; cells were harvested 24 h later. Differences in transfection efficiency were controlled by measuring the enzyme activity generated by a co-transfected thymidine kinase promoterdriven Renilla luciferase (RL) plasmid (0.5 ng). Firefly and RL enzyme activities were measured using the Stop and Glo kit (Promega) following the manufacturer's instructions and relative luciferase activity was determined for each well. Results are the Figure 15. TEF is required for the light-induced per2 promoter activation in vivo. Transgenic Tg(20.43per2:EGFP)tlv1 embryos were injected with tef-MO or control MO. Injected and uninjected embryos were entrained for two LD cycles. At the beginning of the third day of development, embryos were either exposed to light or kept in darkness. After 2 h (ZT 2) embryos were fixed and subjected to whole mount ISH for egfp or otx5 mRNA. (A) Tef (E2I2) MO injection altered tef mRNA splicing. Sequence analysis of the PCR products indicate that tef (E2I2) MO caused the integration of the 226 nucleotides of intron 2, which creates a premature stop-codon (at position +1,303), generating a truncated, non-functional protein lacking the proline and acidic amino acid-rich region (PAR), DNA binding domain (DBD), and the basic leucine zipper domain (bZIP). Grey boxes represent exons, and a red box represents the integrated intron 2. Translation start site and integrated stop codon are marked by arrows. (B) Signal intensities, determined using ImageJ software, revealed a ,10-fold increase in pineal egfp mRNA levels after a 2 h light pulse in uninjected (n = 20, p,0.001) and control MO injected (n = 16, p,0.001) embryos (see also Figure 3). Tef-MO-injected embryos (n = 22) showed no significant increase in pineal signal after a 2 h light pulse suggesting that tef is required for the light-induced expression driven by the 20.43per2 promoter. Error bars represent SE. Dorsal views of representative whole mounts are shown below. Red arrows indicate the location of the pineal gland. (C) Pineal zotx5 mRNA levels of tef-MO injected, control-MOinjected, and uninjected transgenic embryos were similar, indicating that the injected MOs did not affect the development of the pineal gland. Statistical analysis was performed by Kruskal-Wallis test followed by Mann-Whitney test. doi:10.1371/journal.pbio.1000223.g015 mean of three independent experiments, each performed in triplicate.

In Vitro Transient Transfection Assays in PAC-2 Cells
PAC-2 cells were transfected using Fugene6 TM transfection reagent according to the manufacturer's instructions with a 4:1 ratio of FugeneHD (in ml):DNA(in mg) (Roche Diagnostics, FugeneHD) and subsequently incubated for 24 h at 25uC prior to the in vivo luciferase assay.

Establishment of Stable PAC-2 Cell Lines
PAC-2 cells [10] were cultivated as previously described [10,17,46]. Cells were transfected with KpnI-linearized luciferase reporter plasmids (listed above) and a neomycin resistance plasmid linearized with EcoRI [pcDNA3,1 His-Myc(A), (Invitrogen)] at a molar ratio of 7:1. Electroporation was performed at 0.29 kV, 960 mF, by using a Gene Pulser apparatus (Bio-Rad). Three days later, Neomycin G-418 (Gibco BRL) was added at a final concentration of 800 ng/ml. During 1 mo of selection, the concentration was gradually reduced to 250 ng/ml, and 100-200 resistant colonies per transfection were obtained. Colonies were trypsinized and propagated as a single pool.

Live Cell Luciferase Assay and Data Analysis
Live cell luciferase assays were performed as previously described [17]. In total, 3610 4 cells were seeded into each well of a 96-well Fluoplate (Nunc Rochester). At least six independent stable or transient transfections were made for each construct. Following the addition of luciferin to the culture medium, bioluminescence was assayed with a Topcount NXT counter (2detector model, Perkin Elmer). Each well was counted for 3 s at intervals of ,30 min. Plates were counted in an uninterrupted cycle. The counter was located in a thermostatically controlled room. During the experiments, cells were maintained in different lighting regimes. During the light phase, plates were illuminated between counting, with a tungsten light source (20 mW/cm 2 ). Each trace represents the mean of at least two independent pools, each plated in a minimum of four wells. SD was also calculated and plotted. All assays were performed at least three times. Period and peak values were calculated as in previous studies [47,48]. The original data were smoothed by an adjacent-averaging method with 4 h running means. The peak was calculated as the highest point of smoothed data, and the period was computed as the mean between the peaks in each cycle expressed as mean 6 SD.

qRT-PCR
Quantitative Real Time PCR analysis was performed using a StepOnePlus Real-Time PCR System (Applied Biosystems). Total RNA was reverse-transcribed into cDNA by using Superscript III Reverse Transcriptase (Invitrogen) with a mix of oligo dT and random primers. qRT-PCR conditions were 15 min at 95uC, then 40 cycles of 15 s at 95uC, 30 s at 60uC. The relative levels of tef mRNA were calculated by the 2-DDCT method. Relative expression levels were normalized to zebrafish b-actin.
EMSA TEF protein was obtained using the expression vector pcDNA3.1-TEF with the TnT T7 Quick Coupled Transcription/ Translation System from Promega according to the manufacturer's instructions. The in vitro translated TEF (0,5 ml) was pre-incubated for 10 min at room temperature in a 20 ml reaction containing 10% glycerol, 50 mM KCl, 10 mM HEPES (pH 7.6), 1 mM dithiothreitol, 0.1 mM EDTA, 5 mM MgCl 2 , 50 ng/ml poly[d(C-G)], and 50 mM Spermidine (Calbiochem). After pre-incubation, the DNA binding mixture was incubated for an additional 15 min with 25,000 cpm of the LRM DNA probe (from 2187 to 2137 of the zebrafish per2 promoter) (59-ggccatgggtgtccacgtgagttgtatgacacacttatgtaaaaagactgac-39) labelled with c 32 PATP using T4 Polynucleotide kinase.
The following competitor DNAs were added during the preincubation step, with a 50-fold molar excess with respect to the LRM probe.
LRM D-box (59-gtatgacacacttatgtaaaaagactgac-39), Sp1 (59attcgatcggggcggggcgagc-39), LRM carrying a mutated D-box (59- Figure 16. Light and clock-directed regulation of the LRM. A model is proposed where the per2 LRM E-and D-boxes represent a convergence point for both clock and light-driven transcription control mechanisms. The core clock mechanism regulates the E-box enhancer (E) via CLOCK (Clk) and BMAL (Bml) heterodimer, while light exposure predominantly drives expression from the D-box by regulation of the D-box binding factor, TEF. In addition to resetting the phase of the clock mechanism (block arrow), light exposure induces transcription via TEF in two possible ways. In one mechanism, light exposure triggers post-translational modification of existing TEF protein. In another mechanism, light induces the de novo expression of the tef gene that in turn contributes to an increase in transcriptional activation by the D-box. Interaction between the E-and D-box binding factors ultimately defines the precise kinetics of light-regulated per2 expression. doi:10.1371/journal.pbio.1000223.g016 ggccatgggtgtccacgtgagttgtatgacacactcctcctagaagactgac-39), and the LRM E-box (59-ggccatgggtgtccacgtgagttgtatg-39). The DNA binding mixtures were loaded and run for approximately 2 h on a 5% Polyacrylamide/Bis-acrylamide (37,5:1) native gel in TEB 0.56 at 200 V after a pre-run at 50 volts for 30 min. The gel was then dried and exposed for autoradiography for at least 3 h. Figure S1 Co-localization of per2-and aanat2-driven expression in the pineal gland. A double transgenic line was generated by crossing Tg(-0.43per2:EGFP)tlv1 with Tg(aa-nat2:mRFP)y164, which exhibits red fluorescence specifically in the melatonin producing photoreceptor cells of the pineal gland. embryos. The per2 promoter drives ubiquitous EGFP expression that is augmented in the pineal gland. Thus per2 promoter drives expression in virtually all peripheral clock-containing cells and expression is enhanced in the master clock located in the pineal gland. Found at: doi:10.1371/journal.pbio.1000223.s002 (8.17 MB AVI)