The Zebrafish Period2 Protein Positively Regulates the Circadian Clock through Mediation of Retinoic Acid Receptor (RAR)-related Orphan Receptor α (Rorα)*

Background: Null mutants for zebrafish period2 were generated to elucidate its functions. Results: Both locomotor activity and expression of key circadian clock genes are disrupted in the period2 mutant zebrafish. Conclusion: Period2 is essential for zebrafish circadian regulation. Significance: Period2 plays a positive role in the zebrafish circadian clock by enhancing bmal1b expression through binding to nuclear receptor Rorα. We report the characterization of a null mutant for zebrafish circadian clock gene period2 (per2) generated by transcription activator-like effector nuclease and a positive role of PER2 in vertebrate circadian regulation. Locomotor experiments showed that per2 mutant zebrafish display reduced activities under light-dark and 2-h phase delay under constant darkness, and quantitative real time PCR analyses showed up-regulation of cry1aa, cry1ba, cry1bb, and aanat2 but down-regulation of per1b, per3, and bmal1b in per2 mutant zebrafish, suggesting that Per2 is essential for the zebrafish circadian clock. Luciferase reporter assays demonstrated that Per2 represses aanat2 expression through E-box and enhances bmal1b expression through the Ror/Rev-erb response element, implicating that Per2 plays dual roles in the zebrafish circadian clock. Cell transfection and co-immunoprecipitation assays revealed that Per2 enhances bmal1b expression through binding to orphan nuclear receptor Rorα. The enhancing effect of mouse PER2 on Bmal1 transcription is also mediated by RORα even though it binds to REV-ERBα. Moreover, zebrafish Per2 also appears to have tissue-specific regulatory roles in numerous peripheral organs. These findings help define the essential functions of Per2 in the zebrafish circadian clock and in particular provide strong evidence for a positive role of PER2 in the vertebrate circadian system.

cadian clocks (37). In addition, light can entrain the circadian clock to impact the cell cycle and DNA damage repairs in zebrafish (33,38). Although the transcription/translationbased loops are also thought to operate in zebrafish, there are notable differences in circadian regulation between zebrafish and mammals; for instance, at the transcription level, clock and ror␣ oscillate in numerous tissues in zebrafish but not in mice (29), and csnk1␦, which encodes for casein kinase ␦, also oscillates in zebrafish pineal gland but not in mice (39), emphasizing the necessity and importance to investigate the zebrafish circadian clock to obtain a full understanding of the vertebrate circadian mechanisms.
Mammalian Period2 as a canonical component of the circadian clock plays important roles in the circadian clock (21,40), sleeping (41), metabolism (42), and carcinogenesis (43). In humans, a PER2 missense mutation abolishes phosphorylation by CK1␦ (44) and results in familial advanced sleep phase syndrome (45). Mouse Per2 is a light-responded gene, and its circadian phase and amplitude of expression in the suprachiasmatic nuclei can be altered by different light/dark cycles (46). Zebrafish per2 is also a light-regulated gene, and its expression is significantly damped under constant darkness (32). Using transgenic fish and stably transfected cell line-based assays, a light-responsive module composed of D-box and E-box motifs within the per2 promoter was identified (32). In addition, zebrafish per2 is required for expression of the clock-controlled arylalkylamine N-acetyltransferase 2 (aanat2) (47,48) encoding the rate-limiting enzyme for melatonin synthesis (48).
Despite all this important progress concerning zebrafish per2, there have been no stable genetic mutants for zebrafish per2, which are critical for determining its roles in zebrafish circadian regulation as well as other life processes. Recently, genome-editing tools, zinc finger nuclease, transcription activator-like effector nuclease (TALEN), and CRISPR/Cas9 have been developed to generate site-specific DNA double strand breaks that trigger the endogenous nonhomologous end joining DNA repair pathway to induce indel mutations in targeted genes in numerous species including zebrafish (49 -51). Here we have successfully generated two lines of per2-null mutants with TALEN. Characterization of per2 mutant zebrafish showed that Per2 is essential for maintaining rhythmicity of zebrafish locomotor activities and expression of circadian clock genes as well as a circadian clock-controlled gene. We also determined that Per2 plays dual roles in the zebrafish circadian clock, not only repressing expression of E-box-containing genes but also promoting expression of RORE-containing genes. The positive role of PER2 in Bmal1 expression is conserved from zebrafish to mice with the difference that zebrafish Per2 enhances bmal1b expression through its binding of Ror␣ rather than Rev-erb␣, whereas the enhancement of Bmal1 expression by mouse PER2 still requires ROR␣ mediation even though it binds to REV-ERB␣ rather than ROR␣. Zebrafish Per2 seems to have distinct regulatory functions in the different peripheral organs. These results should help elucidate the essential functions of Per2 in the zebrafish circadian clock and provide critical evidence for a positive role of PER2 in the vertebrate circadian system.

EXPERIMENTAL PROCEDURES
Zebrafish Maintenance and Embryo Production-Zebrafish wild-type AB strain and per2 mutant lines were raised on a 14-h/10-h light/dark (LD) cycle at 28°C in the Soochow University Zebrafish Facility according to standard protocols (52). Embryos were produced by pair mating, maintained in culture dishes, and used for experiments at specified stages.
TALEN Construction and Microinjection-TALEN target sites were designed using the web tool TALE-NT (53). TALEN expression vectors were constructed using the "unit assembly" method with Sharkey-AS and Sharkey-R forms of FokI cleavage domains as described previously (54). Briefly, TALEN expression vectors were linearized by NotI and used as templates for capped TALEN mRNA synthesis with the SP6 mMESSAGE mMACHINE kit (Ambion). Capped TALEN mRNAs encoding each monomer were simultaneously microinjected into onecell zebrafish embryos.
Analysis of Mutagenesis Frequencies and Identification of per2 Mutants-The injected embryos were maintained in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl 2 , 0.33 mM MgSO 4 ) at 28.5°C. The genomic DNAs of five groups (five embryos each) were extracted at 2 days postfertilization. A 259base pair (bp) DNA fragment containing the per2 target site was amplified by PCR (primers used are listed in Table 1). 7 l of the PCR product was digested using BslI (New England Biolabs) at 55°C for 3 h. The intensities of cleaved and uncleaved bands were quantified with NIH ImageJ and used for estimating mutagenesis frequencies. The uncleaved bands were recovered after gel electrophoresis and cloned into pMD-19T (Takara) for sequencing analysis. To identify germ line-transmitted mutations, the microinjected founder (F 0 ) embryos were raised to adulthood. The F 0 fish were then outcrossed with wild-type zebrafish to produce F 1 . From each cross, 10 F 1 embryos were collected for genomic DNA extraction and enzymatic digestion (data not shown). Siblings of the F 1 embryos that carry heritable mutations were raised to adulthood, and an individual F 1 fish was reidentified via PCR amplification and sequencing with fin clipped DNAs. Homozygous per2 mutant fish were generated by crossing of the male and female fish carrying the same mutation. Two per2-null mutant lines were established (see Fig. 1D).
Behavioral Analysis for Zebrafish-Locomotor activity analysis was performed as described previously with some modifications (39). On the 4th day postfertilization, larvae were singly placed into each well of the 48-well plate. Locomotor activities of larvae were monitored and recorded for 4 consecutive days using an automated video tracking system (Videotrack, View-Point Life Sciences, Montreal, Canada) and analyzed with Zebralab3.10 software (ViewPoint Life Sciences). Locomotor activities were measured from day 5 to day 8 postfertilization as the total distance moved by one larva during 10-min time windows. The data are presented as a moving distance average for each group (n ϭ 24). The period length of each larval locomotor trace was retrieved by a fit to a damped cosine curve using non-linear least square fitting with the CellulaRhythm R script (55). Statistical analysis of period length differences between the treatment groups was performed with one-way analysis of variance followed by Dunnett's posttest comparing each sample group with the control group (55).
RNA Extraction and Quantitative Real Time PCR (qRT-PCR)-Total RNAs were extracted from ϳ30 larvae of homozygous per2 or wild type at 4-h intervals from 120 to 148 h postfertilization under LD or DD conditions and from adult organs including the brain, muscle, heart, and liver under LD using TRIzol (Invitrogen) reagent, respectively. qRT-PCR was performed in an ABI StepOnePlus instrument with the SYBR Green detection system (Invitrogen). PCR thermal profiles were 40 cycles of 95°C for 10 s and 60°C for 30 s. Experiments were performed in triplicates, each with at least two different biological samples for corresponding genotypes and developmental stages. All results were normalized to the expression level of the housekeeping gene ␤-actin. qRT-PCR results are shown as a relative expression level calculated using the 2Ϫ⌬⌬CT method. p values were analyzed with one-way analysis of variance test or Student's t test. All primers used are listed in Table 1.
DNA Constructs-A 405-bp fragment of the zebrafish aanat2 promoter region containing one EЈ-box (Ϫ76 bp) and a 1,700-bp fragment of zebrafish bmal1b promoter region containing two RORE boxes (Ϫ22 and ϩ18 bp) were isolated and cloned into the luciferase reporter pGL4.17 vector (Promega). The resulting DNA constructs were named as aanat2-luc and bmal1b-luc, respectively. Full-length cDNAs of zebrafish bmal1b and clock1a have been cloned into pcDNA3.1 previously (56). Full-length cDNAs of zebrafish per2, rev-erb␣, ror␣a, and ror␣b genes were PCR-amplified from zebrafish embryonic/larval cDNAs and cloned into the pCMV-HA vec-tor or the pcDNA3.1-Myc/His expression vector, respectively. The stop codon TAA sequence at the 3Ј-ends of each cDNA was removed to fuse with the Myc/His tag in the pcDNA3.1-Myc/His vector. All DNA constructs were confirmed by DNA sequencing. The primers used are listed in Table 1.
DNA Site-directed Mutagenesis-The mutagenized DNA vectors were constructed by PCR-based site-directed mutagenesis. PCR was performed with KOD Plus DNA polymerase (Toyobo). DpnI (New England Biolabs) restriction enzymetreated PCR products were transformed into Escherichia coli. Positive clones were selected and verified by sequencing.
Cell Transfection and Luciferase Reporter Assays-Human embryonic kidney (HEK) 293T cells were used for cell transfection assays. HEK 293T cells were cultured in DMEM containing 10% serum and penicillin-streptomycin in 24-well plates. Transfection was done with Lipofectamine Plus (Invitrogen) according to the manufacturer's instructions. Reporter gene assays were performed with the Dual-Luciferase reporter assay system (Promega) using 100 ng each for bmal1b-luc, ror␣a, ror␣b, and rev-erb␣ and 200 ng for per2. 2 ng of pRL-TK vector was added for control, and pCDNA3.1-Myc/His was added to bring to the same amount. Each experiment was conducted in triplicate.
Co-immunoprecipitation Assay and Western Blotting-One day before transfection, HEK 293T cells were seeded in 10-cm Petri dishes. Thirty-four hours following transfection, cells were lysed in radioimmune precipitation assay buffer with protease inhibitor (Sigma). Lysates were released with protein G-Sepharose beads (GE Healthcare) and then incubated with GTGTCCGATTCAGTCATGCC Transgenic fish identification CTGGTTGTTTCTGTCAGGCC rabbit polyclonal anti-HA antibodies (Protech) or anti-His antibodies (Protech). After washing five times, the precipitates were resuspended in SDS-PAGE sample buffer, boiled for 3 min, and resolved by 8% SDS-PAGE followed by Western blot analysis using mouse monoclonal anti-His antibody or anti-HA antibody (Protech). Immunoreactive bands were detected by ECL reagents (Biological Industries).
Generation of bmal1b-luc Transgenic Fish and in Vivo Measurement of Bioluminescence Rhythms-The vector bmal1b-luc was linearized by KpnI digestion and microinjected into one-to two-cell zebrafish embryos. Injected embryos were raised to adulthood and individually bred to wild-type fish or pairwise bred to each other. Transgenic progeny were identified by PCR using a pair of primers (Table 1). Transgenic embryos or larval fish were placed individually in a well of a 96-well with 200 l of Holtfreter solution (7.0 g of NaCl, 0.4 g of sodium bicarbonate, 0.2 g of CaCl 2 , and 0.1 g of KCl (pH 7.0) in 2 liters of double distilled H 2 O) aerated overnight and containing 0.5 mM D-luciferin potassium salt (BBI). The monitoring of bioluminescence was performed with a Luminoskan Ascent microplate luminometer (Thermo), and data analysis was performed according to the protocol described previously (55).
Chromatin Immunoprecipitation (ChIP) Assays-ChIP assays were performed according to the manufacturer's protocol (Millipore's ChIP assay kit). Briefly, a group of 200 capped per2 mRNA-injected larvae and control injected larvae at 5 days postfertilization was collected and cross-linked in 2% formaldehyde at room temperature for 30 min, and then a 1 ⁄ 10 volume of 1.25 M glycine was added to stop cross-linking followed by PBS washes (three times, each for 10 min). We used purified rabbit or mouse IgG (Invitrogen) as a negative control. ChIP PCRs were performed using primers flanking the E-boxes or RORE sites as well as primers not flanking the E-boxes or the RORE sites in the promoter regions of annat2 or bmal1b as controls. Primers used for the ChIP PCR are listed in Table 1.
Statistical Analysis-Groups of data are presented as mean Ϯ S.E. We performed statistical analyses with analysis of variance or the unpaired two-tailed Student's t test. All statistical analyses were performed using SPSS 16.0 software, and p Ͻ 0.05 was regarded as a statistically significant difference.

Generation of Zebrafish per2
Mutants-Using the software TALE-NT (53), we selected a TALEN pair targeting the second exon (containing the start codon ATG) of zebrafish per2, and within the targeted 57-bp fragment, there is a BslI restriction site for evaluating mutagenesis efficiency and subsequent mutant identification (Fig. 1A). We then used the "unit assembly" method (54) to construct the two arms of the per2 TALEN. The capped mRNAs of the two TALEN arms were microinjected into one-cell embryos at a concentration of 250 pg. To evaluate the mutagenesis efficiency, a 259-bp genomic DNA fragment containing the target site was PCR-amplified from five groups of injected embryos and a control group of embryos (five embryos each). Enzymatic digestion of the PCR-amplified fragments with BslI showed that the efficiencies of the five groups are 10.5, 18.9, 20, 11.2, and 10.2%, respectively, with an average of 14.2% (Fig. 1B). We cloned the uncleaved PCR frag-ments into the sequencing vector pMD-19T. Single clone sequencing revealed six different types of indel mutations in the per2 TALEN target site in F 0 (Fig. 1C). The siblings of these microinjected F 0 embryos were raised to adulthood. Out-crossing F 0 fish with wild-type fish produced F 1 embryos. Two of 15 fish examined with PCR amplification and BslI digestion of DNAs extracted from their F 1 embryos were found to carry heritable mutations (Fig. 1D). DNA sequencing showed that one fish carried a 11-bp deletion and the other carried a 13-bp insertion, which both result in frameshift mutations: the 11-bp deletion mutation might encode a truncated protein with only 160 amino acids, and the 13-bp insertion mutation might encode a truncated protein with only 167 amino acids (Fig. 1E). The homozygous mutant fish with the 11-bp deletion were used primarily for all subsequent experiments.

Disrupted Rhythmicity of Locomotor Activities and Altered Expression of Key Circadian Clock Genes and a Circadian
Clock-controlled Gene in per2 Mutant Zebrafish-Locomotor activities of zebrafish larvae exhibit robust circadian rhythmicity and peak during the subjective day (57). To determine whether Per2 affects locomotor activity rhythms, behavior analyses were performed for per2 mutant and wild-type larvae starting at 96 h postfertilization under LD and DD conditions. Under the LD condition, the locomotor activities of per2 mutant larvae were significantly reduced compared with wild types ( Fig. 2A); specifically there was an ϳ30% reduction in the total moving distance during the 3 days examined (Fig. 2B). Under the DD condition, per2 mutant larvae displayed an approximately 2-h phase delay (Fig. 2C) and an ϳ1.1-h lengthened periodicity (Fig. 2D). These results indicate that the locomotor activity rhythms are disrupted in per2 knock-out fish.
We also examined expression of key circadian clock genes and a circadian clock-controlled gene in per2 mutant fish. Zebrafish possess four per genes, per1a and per1b (co-orthologs of mammalian Per1) and per2 and per3 (single orthologs of mammalian Per2 and Per3) (58). Under the LD condition, per2 exhibited robust oscillation in wild types but still oscillates with much damped amplitude in per2 mutant zebrafish, whereas under the DD condition, per2 became arrhythmic in both wildtype and per2 mutant zebrafish (Fig. 2, E and F). Compared with wild types, per1b and per3 were significantly down-regulated in per2 mutant fish under both LD and DD conditions (Fig. 2, E and F), suggesting that Per2 plays a positive role in regulation of these two zebrafish per genes. Similar to per genes, cry genes are negative regulators in the transcription/translational feedback loop (59). Although mice have two Cry genes (59), zebrafish have six cry genes (31,56,60,61). Under both LD and DD, cry1aa, cry1ba, and cry1bb were primarily up-regulated in per2 mutant fish (Fig. 2, E and F), suggesting that Per2 plays a repressive role in regulation of these cry genes. Intriguingly, bmalb gene, one of the two co-orthologs of mouse Bmal1 (62), was down-regulated in per2 mutant under both LD and DD conditions (Fig. 2, E and F), suggesting that Per2 plays a positive role in bmalb expression. The other two bmal genes (bmal1a and bmal2) also had disrupted expression patterns in per2 mutant under both LD and DD conditions (Fig. 2, E and F).
Melatonin plays an important role in the endogenous circadian clock system in vertebrates (63). Melatonin rhythms are generated by the oscillating rate-limiting enzyme aralkylamine N-acetyltransferase (AANAT) in the pineal gland (47,64). Mammals have only a single Aanat gene that is expressed in both the pineal gland and the retina (65). In zebrafish, there are two copies of aanat genes, aanat1 and aanat2. Although aanat1 is expressed only in the retina, aanat2 is expressed in the pineal gland and in the retina at relatively lower levels (66). aanat2 was significantly up-regulated under both LD and DD conditions in per2 mutant fish (Fig. 2, E and F), suggesting that Per2 plays a repressive role in regulation of aanat2.
We also characterized another null per2 mutant line carrying the 13-bp insertion and found that it displays a similar behavioral phenotype (data not shown) and altered gene expression (Fig. 2, E and F) like the 11-bp deletion mutant line. These results demonstrate that like mammalian Per2 zebrafish per2 is also essential for the zebrafish circadian clock. The targeted fragment was PCR-amplified from pooled genomic DNAs of five embryos microinjected with capped TALEN mRNAs at a concentration of 250 ng and then digested with BslI. The uncleaved and cleaved PCR products are indicated. Mutagenesis efficiencies were estimated by the ratios of intensities of uncleaved bands and the sum of cleaved bands and uncleaved bands quantified with NIH ImageJ software. WT, wild type; M, marker. C, types of indel mutations in the per2 TALEN target site shown by representative sequencing results of the uncleaved PCR fragments. D, screening of heritable mutants. F 0 founder fish were out-crossed with wild-type fish to produce F 1 , and the DNAs extracted from F 1 embryos were used for identifying mutant fish. Two of 15 fish were found to carry heritable mutations (lanes 4 and 10). E, two mutated fish lines. One has an 11-bp deletion, the other has a 13-bp insertion (upper), and both are frameshift mutations that result in truncated proteins (lower). AA, amino acids.

Dual Roles of Per2 Are Mediated by E-boxes and RORE Boxes
in Zebrafish-Our qRT-PCR analysis showed that aanat2 is up-regulated but bmal1b is down-regulated in per2 mutant fish (Fig. 2, E and F), indicating that Per2 represses aanat2 expres-sion but enhances bmal1b expression in zebrafish. To investigate the role of Per2 in zebrafish aanat2 expression, a 405-bp aanat2 promoter containing one E-box was isolated and cloned into the pGL4.17 vector (Fig. 3A). The full-length cDNAs of . Behavioral assays were done with an automated video tracking system (Videotrack) and analyzed with Zebralab3.10 software (see "Experimental Procedures" for details). E and F, disrupted expression of key circadian clock genes and a circadian clock-controlled gene in per2 mutant fish. Shown are qRT-PCR analyses of key circadian clock genes (per1b, per2, per3, cry1aa, cry1ba,  cry1bb, bmal1a, bmal1b, and bmal2) and a circadian clock-controlled gene (aanat2) in wild-type and per2 mutant fish under LD (E) and DD conditions (F). WT is shown in black, 11-bp deletion mutant is shown in red, and 13-bp insertion mutant is shown in yellow. RNA extraction was done with TRIzol, and qRT-PCR experiments were performed with an ABI StepOnePlus instrument and the SYBR Green detection system (see "Experimental Procedures" for details). *, p Ͻ 0.05; **, p Ͻ 0.01. Error bars represent S.E.  FEBRUARY 13, 2015 • VOLUME 290 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4373 per2, bmal1b, and clock1a were cloned into the pcDNA3. 1-Myc/His vector, respectively. In vitro cell transfection showed that aanat2 is significantly up-regulated by a combination of Clock1a and Bmal1b but repressed by Per2, and Per2 could no longer inhibit aanat2 when the E-box in the aanat2 promoter was mutated (Fig. 3C). In addition, ChIP assays showed that Per2 can bind to the E-box in the aanat2 promoter (Fig. 4E). These results demonstrate that zebrafish Per2 can repress the gene expression through the E-box elements, which is consistent with mouse PER2 function (67).

A Positive Role of Per2 in Zebrafish Circadian Regulation
Similar to the two RORE boxes in the mouse Bmal1 promoter (68), we also identified the two RORE motifs in the zebrafish bmal1b promoter region (Fig. 3B). To determine the role of Per2 in zebrafish bmal1b expression, a 1.7-kb promoter of bmal1b harboring the two RORE motifs was isolated and cloned into the pGL4.17 vector. Due to the third round of teleost genome duplication (58,62,69), zebrafish have two copies of ror␣ genes, ror␣a and ror␣b (70). The full-length cDNAs of rev-erb␣, ror␣a, and ror␣b were PCR-amplified and cloned into the pcDNA3.1-Myc/His vector, respectively. Cell transfection assays showed that these two ror␣ genes can significantly activate bmal1b expression, and in particular, the activation activity of ror␣a is ϳ2-fold higher than that of ror␣b (Fig. 3D). In addition, Rev-erb␣ can outcompete Ror␣a or Ror␣b to inhibit bmal1b expression in a dosage-dependent manner (Fig.  4E). These results demonstrate that like what happens in mammals Ror␣ activates bmal1b expression, whereas Rev-erb␣ represses it in zebrafish.
To determine how Per2 regulates bmal1b expression, we also conducted co-transfection experiments with per2, ror␣a, ror␣b, or rev-erb␣. Results showed that Per2 can enhance bmal1b expression via RORE motifs (Fig. 3D), and the enhancing effects of Per2 on bmal1b expression are also dosage-dependent (Fig. 3F). To examine the roles of the two RORE motifs in bmal1b expression, we mutagenized them in the bmal1b promoter. Luciferase assays showed that the inhibitory effect of Rev-erb␣, the activating effect of Ror␣a, and the enhancing effect of Per2 on bmal1b expression are all abolished (Fig. 3D) without the two RORE motifs (Fig. 3D). These results indicated that the two RORE motifs are required for Per2 to enhance bmal1b expression in zebrafish (Fig. 3D).
To further distinguish the roles of these two RORE motifs in bmal1b expression, we mutagenized them individually. Results showed that although both the individually mutated RORE motifs can result in the reduction of Ror␣-mediated bmal1b expression RORE1 appears to be especially important for Ror␣mediated activation, whereas RORE2 is necessary for enhancement of the level of activation (Fig. 3, G and F).

. Enhancement of bmal1b expression by Per2 in vivo.
A, strong rhythmic expression of bioluminescence driven by the bmal1b promoter in the wild-type fish. B, weak but still rhythmic expression of bioluminescence driven by the bmal1b promoter in per2 mutant fish. C, no rhythmic expression of bioluminescence driven by the bmal1b promoter with two mutated RORE motifs in the wild-type fish. D, no rhythmic expression of bioluminescence driven by the bmal1b promoter with two mutated RORE motifs in per2 mutant fish. We generated four stable transgenic zebrafish lines, normal bmal1b-luc in wild-type fish, normal bmal1b-luc in per2 mutant fish, mutated bmal1b-luc in wild-type fish, and mutated bmal1b-luc in per2 mutant fish. The experiments were done with stable F 1 transgenic fish lines. E, Per2 binds to the RORE motifs in the bmalb promoter and the E-box in the aanat2 promoter as shown by ChIP assays. Capped per2 mRNAs were microinjected into one-cell embryos, and ChIP assays were done with larvae at 96 (ZT0) and 108 h postfertilization (ZT12). F and G, rescue of annat2 and bmal1b expression by capped wild-type per2 mRNAs. Microinjection of capped wild-type per2 mRNAs rescues expression of annat2 (F) and bmal1b (G) in per2 mutant fish. Quantitative RT-PCR analysis was done with microinjected per2 mutant larvae and control larvae at 108 h postfertilization (ZT12). *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. Error bars represent S.E. bmal1b displays robust oscillation and in per2 mutant Tg (1.7bmal1b-luc) per2Ϫ/Ϫ transgenic fish bmal1b exhibits much dampened oscillation (Fig. 4, A and B) in both wild-type Tg (1.7bmal1b-ROREmut-luc) AB and per2 mutant Tg (1.7bmal1b-ROREmut:luc) per2Ϫ/Ϫ transgenic fish bmal1b shows no oscillation (Fig. 4, C and D). These results indicate that Per2 and RORE motifs are crucial for bmal1b rhythmic expression in vivo.
To determine whether Per2 binds to the RORE motifs in the bmal1b promoter region, capped mRNAs of per2 were microinjected into one-cell embryos for ChIP assays. The results showed that Per2 can bind to RORE motifs in the bmal1b promoter at ZT0 and ZT12 in vivo (Fig. 4E), demonstrating that the enhancing effect of Per2 on bmal1b expression depends upon the RORE motifs in the bmal1b promoter.
We also performed rescue experiments by microinjecting capped wild-type per2 mRNAs into per2 mutant embryos. qRT-PCR results showed that wild-type per2 mRNAs indeed can down-regulate aanat2 but up-regulate bmal1b in per2 mutant fish (Fig. 4, F and G), indicating that the TALEN-induced mutation is responsible for its phenotypes.
Per2 Enhances bmal1b Expression through Ror␣ Rather than Rev-erb␣ in Zebrafish-To determine whether Per2 enhances bmal1b expression through Ror␣a, Ror␣b, or Rev-erb␣ in zebrafish, we conducted cell transfection experiments. The results showed that Per2 alone cannot reverse the inhibitory effect of Rev-erb␣ on bmal1b expression in the absence of Ror␣ (Fig. 5A) and that Per2 can enhance Ror␣a-or Ror␣b-mediated bmal1b expression (Fig. 5B). We also performed co-immunoprecipitation (co-IP) assays. The results showed that Per2 can directly bind to Ror␣a rather than to Rev-erb␣ (Fig. 4, C and D). Taken together, these results showed that Per2 can enhance bmal1b expression through Ror␣a or Ror␣b rather than through Rev-erb␣ in zebrafish.
In mice, the N-terminal LXXLL motif of PER2 was shown to be able to bind to REV-ERB␣ (24). We observed that zebrafish Per2 also possesses the same LXXLL motif at its N terminus (Fig. 5E). Does zebrafish Per2 bind to Ror␣a through its N-ter- C and D, Per2 binds to Ror␣a rather than Rev-erb␣ in zebrafish as shown by co-IP assays. C, the left panel shows that HA-tagged Per2 could not pull down His-tagged Rev-erb␣, and the right panel shows that HA-tagged Per2 did pull down His-tagged Ror␣a. D, the left panel shows that His-tagged Rev-erb␣ could not pull down HA-tagged Per2, and the right panel shows that His-tagged Ror␣a did pull down HA-tagged Per2. E, the conserved protein-protein interaction motif LXXLL (yellow) between zebrafish Per2 and mouse PER2. F, Per2 with mutated LXXLL motif cannot bind to Ror␣a as shown by co-IP assays. G, Per2 with mutated LXXLL motif cannot enhance Ror␣a-mediated transcription of bmal1b as shown by in vitro cell transfection. ***, p Ͻ 0.001; NS, not significant. Error bars represent S.E. minal LXXLL motif? To address this question, we mutagenized the Per2 LXXLL motif and performed co-IP and transfection assays. The results showed that Per2 with the mutated LXXLL motif neither binds to Ror␣a (Fig. 5F) nor enhances Ror␣amediated bmal1b transcription (Fig. 5G), further supporting the notion that the positive role of zebrafish Per2 in bmal1b expression is fulfilled through its binding to Ror␣a rather than to Rev-erb␣.
Enhancement of Bmal1 Expression by PER2 Is Evolutionally Conserved-Our results showed that Per2 enhances bmal1b expression through binding to Ror␣ in zebrafish. Although previous studies have shown that Bmal1 expression also was down-regulated in the Per2 knock-out mouse (21,40,43), PER2 directly interacts with the REV-ERB␣ in mice, and overexpression of Per2 can increase Bmal1 expression in cultured cells (24), the mechanisms underlying how mouse PER2 regulates BMAL1 have not been carefully examined. To delineate how PER2 enhances Bmal1 expression in mice, we performed cell transfection experiments. Indeed, the results showed that mouse ROR␣ can activate Bmal1 expression, whereas mouse REV-ERB␣ can repress ROR␣-mediated Bmal1 expression, and PER2 can enhance ROR␣-mediated Bmal1 expression (Fig.  6A). In addition, in the case of the mutated RORE motifs, all these effects were abolished (Fig. 6A). Moreover, PER2 can significantly enhance Bmal1 transcription with the presence of ROR␣ (Fig. 6B); however, PER2 cannot reverse the inhibitory effect of REV-ERB␣ on Bmal1 transcription without ROR␣ (Fig. 6C), suggesting that although PER2 interacts with REV-ERB␣ the enhancing activity of PER2 on Bmal1 expression is still mediated by ROR␣ in mice, which is different from the notion that mouse PER2 binds to REV-ERB␣ to enhance Bmal1 expression as implicated previously (24).
To elucidate whether the roles of zebrafish Per2/mouse PER2 in regulation of bmal1b or Bmal1 are conserved between mice and zebrafish, zebrafish Per2 was co-transfected with mouse Bmal1-luc and ROR␣. The results showed that mouse PER2 can enhance zebrafish Ror␣a-or Ror␣b-mediated bmal1b expression (Fig. 6D). Similarly, zebrafish Per2 can enhance mouse ROR␣-mediated Bmal1 expression (Fig. 6E). These results supported the conservative function of the Per2/PER2 proteins in enhancing expression of bmal1b/Bmal1 in zebrafish and mice.
We also examined the role of Rev-erb␣ in this system by co-transfection experiments. The results showed that neither can zebrafish Per2 enhance mouse ROR␣-mediated mouse Bmal1 expression nor can mouse PER2 enhance zebrafish Ror␣-mediated zebrafish bmal1b expression (Fig. 6, F and G). The cell transfection experiments also showed that even though zebrafish Rev-erb␣ can significantly inhibit mouse Bmal1 expression (Fig. 6H) without the presence of ROR␣ and mouse REV-ERB␣ also can significantly inhibit zebrafish bmal1b expression without the presence of Ror␣ (data not show) zebrafish Rev-erb␣ cannot outcompete mouse ROR␣ to repress Bmal1 expression. However, the mouse REV-ERB␣ can outcompete zebrafish Ror␣a or Ror␣b to repress bmal1b expression (Fig. 6, F and H), suggesting that mouse REV-ERB␣ has evolved stronger inhibitory abilities than zebrafish Rev-erb␣.
Tissue-specific Circadian Regulation by Zebrafish Per2-Zebrafish per2 is expressed extensively in numerous tissues/ organs (32,71). To investigate the per2 functions in different tissues/organs, we compared expression patterns of several canonical circadian clock genes including per2 itself in different organs of wild-type and per2 mutant zebrafish including the brain, muscle, heart, and liver. Results showed marked downregulation of per1b and per2 in all four organs of the per2 mutant fish (Fig. 7). In contrast, the expression of cry1ba was significantly up-regulated in the brain but down-regulated in the muscle, heart, and liver of per2 mutant fish (Fig. 7). The expression of bmal1b exhibited a significant phase delay in the muscle and was down-regulated in the brain, heart, and liver of per2 mutant fish. Intriguingly, the four circadian clock genes per1b, per2, cry1ba, and bmal1b were all down-regulated in the heart and liver of per2 mutant fish (Fig. 7, C and D), implicating that Per2 is critical for maintaining circadian regulation in the heart and liver and in turn impacts functions of these two important peripheral organs. These results showed that although different circadian clock genes display distinct rhythmic expression patterns in the same organ the same circadian clock gene exhibits distinct rhythmic expression patterns in a different organ, and these genes were all significantly disrupted in the per2 mutant fish (Fig. 7), thereby implicating that Per2 may play distinct regulatory roles in different zebrafish peripheral organs/tissues.

DISCUSSION
Per2 Is Essential for the Zebrafish Circadian Clock-Zebrafish have recently figured as an excellent circadian model and contributed to our understanding of vertebrate circadian rhythmicity (30 -32). Genetic analysis of the zebrafish circadian clock, however, has lagged behind largely due to difficulties in obtaining stable genetic circadian mutants (14,31,72). Here we used TALEN, a fast and convenient genome-editing tool, to generate null mutants for zebrafish per2. Analysis of per2 mutant zebrafish helps us to ascertain the roles of Per2 in the zebrafish circadian clock. Per2Ϫ/Ϫ knock-out mice display a short circadian period and arrhythmicity under DD (21), and a human PER2 mutation results in ϳ4-h phase advance under LD FIGURE 7. Tissue-specific regulatory roles of Per2 in zebrafish peripheral tissues/organs. Wild-type and per2 mutant fish were sacrificed at 120 days postfertilization, and the organs/tissues of the brain, muscle, heart, and liver were dissected out and collected for RNA extraction at 4-h intervals under LD for a total of consecutive 24 h. Each sample contains tissues/organs from at least two female fish and two male fish. Shown is qRT-PCR analysis of expression levels of four circadian clock genes (per1b, per2, cry1ba, and bmal1b) in the brain (A), muscle (B), heart (C), and liver (D) of wild-type and per2 mutant adult fish. Data represent mean Ϯ S.E. (error bars) (n ϭ 2-3). (45). Our per2 mutant zebrafish displayed reduced activities under LD (Fig. 2, A and B) and a 2-h phase delay and a 1.1-h prolonged period under DD (Fig. 2, C and D), which are completely different from the phenotypes of the zebrafish per1b insertional null mutant that displays hyperactivities under LD and a 2-h phase advance under DD (73). Hence, like mammalian Per2, zebrafish per2 also is critical for maintaining fish locomotor rhythmicity. The fact that both zebrafish per2 and per1b mutant fish still are not completely arrhythmic under DD strongly suggests that zebrafish per genes resemble mouse Per genes in that they are partially redundant and can compensate for the other's loss for maintaining locomotor rhythmicity (74). However, the intricate roles of Per2 in regulating activity rhythms differ between fish and mammals and likely represent different stages of Per2 functions during evolution from fish to mammals.
We also found that rhythmic expression patterns as well as phases of key circadian genes are disrupted in per2 mutant fish (Fig. 2, E and F). For instance, both per1b and per3 were significantly down-regulated in per2 mutant fish under both LD and DD, suggesting that Per2 plays a positive role in regulating these two per genes (Fig. 2, E and F), which diametrically differs from the repressive role of Per1b in regulating other per genes; i.e. all three other per genes are significantly up-regulated in the per1b-null mutant fish (73). Although Per1 knock-out mice have no effects on rhythmic expression of Per1 or Per2, both Per1 and Per2 are significantly down-regulated in Per2 knockout mice (21,40). Thus both zebrafish Per2 and mouse PER2 share the conservative function of positively regulating per genes. Together, the disrupted locomotor behaviors and altered expression of circadian clock genes in per2 mutant indicate that per2 is an essential component in the zebrafish circadian clock.
Dual Roles of Per2 in the Zebrafish Circadian Clock-In the Drosophila negative feedback loop, per and tim act as negative regulators (75). Similarly, it has long been thought that mammalian Per1, Per2, Cry1, and Cry2 proteins as negative factors form heterodimers to repress CLOCK-BMAL1-mediated transcription (2,59,76). However, there is evidence that PER2 might also play a positive role in mammalian circadian regulation (21,40,77). Although PER2 is not a transcriptional factor, it can directly bind to numerous nuclear receptors to control mammalian physiological processes (24,78). Our results showed that zebrafish Per2 not only represses aanat2 expression through E-box but also enhances bmal1b expression through RORE. The repressive role of Per2 is in line with the traditional notion that Per2 and Cry form a heterodimer that represses Clock-Bmal heterodimer-mediated transcription (Fig. 3C) (2, 59), whereas the enhancing role of Per2 is fulfilled by its binding to Ror␣ nuclear receptor (Fig. 3D).
The Positive Role of Zebrafish Per2 and Mouse PER2 in Circadian Regulation Is Fulfilled through Mediation of ROR␣ or Ror␣-Even though previous studies have shown that mouse PER2 can increase Bmal1expression (21,24,43), the exact mechanisms are not clear. Our detailed cell transfection assays showed that although mouse PER2 binds to REV-ERB␣ (24) it still requires ROR␣ mediation to enhance BMAL1 expression (Fig. 6, A-C). Our studies also showed down-regulation of RORE-containing bmal1b in per2 mutant zebrafish (Fig. 2, E  and F), and zebrafish Per2 positively regulated bmal1b expression through RORE (Fig. 3D). In particular, co-IP experiments showed that Per2 directly interacts with Ror␣a through the LXXLL motif conserved between zebrafish Per2 and mouse PER2 (Fig. 5E).
The roles of ROR␣, REV-ERB␣, and PER2 in regulation of Bmal1 are highly conserved between zebrafish and mice but with a difference. Although zebrafish Ror␣a/Ror␣b or mouse ROR␣ can bind RORE to activate bmal1b or Bmal1 expression (Fig. 8, A, panel 1, and B, panel 1), zebrafish Rev-erb␣ or mouse REV-ERB␣ can outcompete Ror␣ or ROR␣ to bind to RORE to repress bmal1b or Bmal1 expression (Fig. 8, A, panel 2, and B,  panel 2), and zebrafish Per2 or mouse PER2 alone cannot reverse the repressive effects of Rev-erb␣ or REV-ERB␣ on bmal1b or Bmal1 expression without the presence of zebrafish Ror␣ or mouse ROR␣ (Fig. 8, A, panel 3, and B, panel 3), zebrafish Per2 or mouse PER2 alone can significantly enhance Ror␣-mediated bmal1b expression or ROR␣-mediated Bmal1 expression without the presence of Rev-erb␣ or REV-ERB␣ (Fig. 8, A, panel 4, and B, panel 4), which are all conserved in zebrafish and mice. The difference is that zebrafish Per2 binds to Ror␣ rather than Rev-erb␣ to enhance bmal1b expression (Fig 8A, panel 4), whereas mouse PER2 enhances Bmal1 expression still through ROR␣ mediation, but it binds to REV-ERB␣ rather than ROR␣ (Fig. 8B, panel 5). These results revealed that zebrafish Per2 or mouse PER2 enhances bmal1b or Bmal1 expression through mediation of Ror␣ or ROR␣. A recent study showed that mouse PER1 and PER2 but not PER3 can inhibit CRY-mediated transcriptional repression by preventing CRY from being recruited to the CLOCK-BMAL1 complex, providing evidence that PER2 and PER1 have positive roles in the mammalian circadian clock (79), which complements our promoter analysis of zebrafish per2 and mouse Per2 in support of PER2 as a positive factor in vertebrate circadian regulation. Hence, the positive role of PER2 in regulation of Bmal1 expression is conserved from zebrafish to mice, which is at odds with the prevailing notion of PER2 as a negative circadian factor (1,2,59).
Alignment of the amino acid sequence revealed that numerous functional motifs are highly conserved in zebrafish Per2 and mouse PER2 (data not shown), thereby supporting the notion that their roles in regulation of bmal1b or Bmal1 are conserved. We also examined the possible roles of zebrafish Per2 and Rev-erb␣ in mice and vice versa and found that although mouse PER2 can enhance zebrafish Ror␣-mediated bmal1b expression (Fig. 8C, panel 1) so can zebrafish Per2 enhance mouse ROR␣-mediated Bmal1 transcription (Fig. 8D,  panel 1); mouse PER2 cannot reverse the inhibitory effect of zebrafish Rev-erb␣ on bmal1b expression in the presence of Ror␣a, nor can zebrafish Per2 reverse the inhibitory effect of mouse REV-ERB␣ on Bmal1 expression in the presence of ROR␣ (Fig. 8, C, panel 2, and D, panel 2). The difference is that although mouse REV-ERB␣ can outcompete zebrafish Ror␣ to bind to RORE to repress zebrafish bmal1b expression (Fig. 8C,  panel 3) zebrafish Rev-erb␣ cannot (Fig. 8D, panel 3).
Because zebrafish and tetrapods including mammals shared a common ancestor more than 300 million years ago (58,80), it is fascinating that most aspects of the functions of zebrafish Per2 and mouse PER2 are still highly conserved (for instance, both can positively regulate expression of bmal1b or Bmal1). However, it is not unexpected that zebrafish Per2 and mouse PER2 have evolved divergent functions; i.e. zebrafish Per2 binds to Ror␣ to exert its enhancing effects on bmal1b expression, whereas mouse PER2 has evolved to bind to REV-ERB␣ instead but still requires ROR␣ mediation to exert its enhancing effects on Bmal1 expression (Fig. 8). The functions of zebrafish Rev-erb␣ and mouse REV-ERB␣ are also mostly conserved but show divergence in that mouse REV-ERB␣ has evolved a stronger ability to outcompete zebrafish Ror␣ to repress zebrafish bmal1b expression (Fig. 8). Therefore, it is necessary and important to study the functions of zebrafish circadian clock genes because it provides comparative perspectives for how the functions of these key circadian clock genes have evolved.
In conclusion, our analysis of the per2-null mutants generated by TALEN showed that the rhythms of locomotor behaviors and expression of core circadian genes and one circadian clock controlled-gene expression are disrupted in per2 mutant fish, indicating that per2 is essential for the zebrafish circadian clock. We determined that zebrafish per2 plays both positive and negative roles in circadian regulation, and in particular, Per2 positively regulates bmal1b expression by directly binding to Ror␣. Mouse PER2 enhances Bmal1 expression still through ROR␣ mediation even though it has evolved to bind to REV-ERB␣ instead. Moreover, zebrafish Per2 appears to have tissuespecific functions in the peripheral circadian clocks. These results help define the Per2 functions in the zebrafish circadian clock and provide invaluable evidence for a positive role of PER2 in the vertebrate circadian system.