A noncanonical E-box enhancer drives mouse Period2 circadian oscillations in vivo

Abstract

The mouse Period2 (mPer2) locus is an essential negative-feedback element of the mammalian circadian-clock mechanism. Recent work has shown that mPer2 circadian gene expression persists in both central and peripheral tissues. Here, we analyze the mouse mPer2 promoter and identify a circadian enhancer (E2) with a noncanonical 5′-CACGTT-3′ E-box located 20 bp upstream of the mPer2 transcription start site. The E2 enhancer accounts for most circadian transcriptional drive of the mPer2 locus by CLOCK:BMAL1, is a major site of DNaseI hypersensitivity in this region, and is constitutively bound by a transcriptional complex containing the CLOCK protein. Importantly, the E2 enhancer is sufficient to drive self-sustained circadian rhythms of luciferase activity in central and peripheral tissues from mPer2-E2::Luciferase transgenic mice with tissue-specific phase and period characteristics. Last, genetic analysis with mutations in Clock and Bmal1 shows that the E2 enhancer is a target of CLOCK and BMAL1 in vivo.

The mammalian circadian system is composed of a hierarchy of robustly rhythmic central and peripheral oscillators (1–3). An ensemble of coupled oscillators in the suprachiasmatic nucleus (SCN) of the hypothalamus is entrained by daily light input from the visual system, and neural and humoral output signals from the SCN coordinate the phase of independent circadian oscillators in peripheral tissues throughout the organism (1, 4). The clock mechanism is similar in cells of the SCN and periphery and consists of a network of transcriptional/translational feedback loops (5–7). In the primary feedback loop, transcription is driven by the bHLH-PAS proteins CLOCK and BMAL1 (or MOP3), which heterodimerize and initiate transcription of three Period genes (in mice; mPer1, mPer2, and mPer3) and two Cryptochrome genes (mCry1 and mCry2) (8–14). The PER and CRY proteins then negatively feedback to repress transcription at their own promoters by acting on the CLOCK:BMAL1 complex (9, 11, 15, 16). The primary feedback loop is modulated by a second feedback loop composed of the two retinoic acid-related orphan receptors (RORs), REV-ERBα and RORa (17, 18), which drive a circadian rhythm in Bmal1 transcription (11, 19).

Of the three Period genes, mPer2 plays a dominant role (14, 20). However, in contrast to mPer1 (8, 9, 21), very little is known regarding the transcriptional regulation of mPer2 (10, 22, 23) because of the absence of canonical 5′-CACGTG-3′ E-box elements (24) within the 5′ upstream regulatory region. Here, we report a comprehensive analysis of the circadian transcriptional drive of the mouse Per2 locus.

Materials and Methods

DNA Constructs. A 3.3-kb EcoRI/XbaI fragment was isolated from a Per2-positive bacterial artificial chromosome clone (1) and sub-cloned into the pGL3-Basic luciferase reporter vector. The primers used to generate truncated promoter constructs are given in the supporting information.

Cell Culture. HepG2 cells were grown in DMEM supplemented with 10% FBS containing penicillin/streptomycin. Cells were plated the day before transfection at 2 × 10⁵ cells per well in six-well plates and transfected with 50 ng of reporter, 100 ng of pCMV-β-gal for normalization, and pcDNA3.1 mClock and mBmal1 constructs (25) by using Lipofectamine Plus. At 48 h after transfection, cells were lysed, and luminescence was measured as described in the supporting information, which is published on the PNAS web site.

Transgenic Mice. Transgenic mice were generated as described (26). See supporting information for details. Explant culture and bioluminescence experiments were performed as reported (1).

DnaseI Hypersensitive Site Mapping. Nuclei from Hepa1–6 cells and C57BL/6J female mice were isolated as described (27, 28) with modifications (see supporting information).

Chromatin immunoprecipitation (ChIP) assays were performed as described (11), or with modifications, as described in the supporting information.

Supporting Information Primers used to generate truncated promoter constructs, additional materials and methods, identification of transcription initiation sites of mPer2, and Western blot analysis of mouse liver nuclei can be found in Supporting Materials and Methods and Figs. 6 and 7, which are published as supporting information on the PNAS web site.

Results

Identification of an E-box Enhancer in the mPer2 5′ Regulatory Region. The mPer2 locus resides on the midportion of chromosome 1, spans ≈40 kb, and contains 23 exons (Fig. 1A). Transcription produces a mature mRNA of ≈7.5 kb (29). To investigate transcriptional regulation of mPer2, we identified the transcription start site by 5′ RACE and found a major site of initiation, which we designated as base pair +1 (Fig. 6). A 3.3-kb fragment covering base pairs -1,128 to +2,129 was then used to analyze this region further (Fig. 1B, pGL-6). Five truncated constructs were generated by deleting regions in the 5′ direction (Fig. 1B, pGL-5 to pGL-1). In transient transfection assays using HepG2 cells, we assessed the ability of CLOCK and BMAL1 to drive luciferase expression from putative E-boxes contained in the five constructs. Deletion of a previously identified E-box in intron 1 (Fig. 1B, E1, pGL-5) (10) failed to cause a decline in CLOCK:BMAL1 activation compared with the complete 3.3-kb fragment (pGL-6). Transcriptional activation by CLOCK:BMAL1 was lower in pGL-4 compared with that of pGL-5, suggesting that E2a (5′-CACGTT-3′) is a putative but noncanonical E-box element. When compared with pGL-4, pGL-3 showed only a slightly higher rate of transcriptional activation by Clock/Bmal1 cotransfection. Still, all constructs responded significantly to Clock/Bmal1 cotransfection, with the exception of pGL-1. These results strongly suggested the existence of one or more active E-boxes in the region of base pairs -141 through +386. Also, upon deletion of base pairs +386 to +1,060, a significant increase in the basal activities of reporter constructs pGL-2 and pGL-1 was apparent, indicating the presence of cis-acting repressor elements within the region. Analysis of the sequence of base pairs -141 to +386 by the matinspector analysis program revealed the presence of a single sequence similar to the aryl hydrocarbon receptor nuclear transporter (ARNT) homodimer and Hif1α:ARNT binding site (24, 30) 5′-CACGTT-3′, at base pairs -20 to -15 on the negative strand (5′-AACGTG-3′), corresponding to a noncanonical E-box candidate E2 (Figs. 1B and 2A).

Fig. 1.

An E-box enhancer in the mPer2 promoter drives CLOCK/BMAL1-mediated transcriptional activation. (A) Schematic representation of the mPer2 locus. The open box represents exon 1 of mPer2, and the arrow indicates the transcription start site. Ovals indicate the following putative E-box enhancer sites: E1 (+2,066), 5′-CACGTG-3′; E2a (+1,560), 5′-CACGTT-3′; E2 (-20), 5′-CACGTT-3′;E3(-2,816), 5′-AACGTG-3′;E4(-3,271), 5′-AACGTG-3′; and E5 (-4,111), 5′-CACGTT-3′. The position of probes and the restriction enzymes used for DNaseI hypersensitive site mapping (C) are indicated. (B) Truncated mPer2-Luc reporter constructs with various deletions in the mPer2 5′ regulatory region. An equimolar amount of each mutant reporter construct (pGL3-Basic) was transiently transfected into HepG2 cells with (white bar, 200 ng each) and without (black bar) Clock and Bmal1 expression constructs. Ovals indicate the following E-box candidates: E1 (+2,066), 5′-CACGTG-3′; E2a (+1,560), 5′-CACGTT-3′; and E2 (-20), 5′-CACGTT-3′. Numbers given to the right of each construct indicate the end position of the insert from the transcription start site. Two-way ANOVA revealed significant main effects of construct (F = 121.07, P < 10^-4) and Clock/Bmal1 cotransfection (F = 223.7, P < 10^-4), as well as a significant interaction between construct and Clock/Bmal1 cotransfection (F = 106.3, P < 10^-4). Each value is the mean ± SEM of three replicates from a single assay. The results are representative of at least three independent experiments. (C) DNaseI hypersensitive sites in the mPer2 5′ flanking region. Nuclei were isolated from samples of equivalent cell density (Hepa1–6 cells) or tissue mass (C57BL/6J liver; ZT 12) and were treated with 8, 16, and 24 units of DNaseI or no DNaseI as control (C). Probes A and B (B) were used to visualize the 3.9-kb full-length XbaI genomic DNA digestion product and the DNaseI hypersensitive sites (1 and 2) located between base pairs -300 to +30 and at +1,500, respectively. Arrowheads indicate the putative E-box enhancer site corresponding to the approximate location of E2.

Fig. 2.

Analysis of the mPer2 E2 enhancer. (A) Sequence alignment of putative mouse Per2 E-box elements with the corresponding region of human Per2. Among the six putative mouse E-box sequences, only E2 and its flanking regions are well conserved with the human Per2 locus. The conserved E2 E-box sequences (5′-CACGTT-3′) from mouse and human Per2 are shown in red; the five additional E-boxes and flanking sequence in mPer2 are shown in blue. Dotted lines indicate the E-box flanking sequence that is partially conserved with the CLOCK:BMAL1 M34 core binding site (31). (B) Putative mPer2 E-box elements (containing 200 bp of flanking sequence of E1, E2a, E2, E3, E4, and E5) were cloned into the pGL3-Promoter vector (pE1, pE2a, pE2, pE3, pE4, and pE5), and 50 ng of each reporter construct was transfected into HepG2 cells with (white bar) or without (black bar) 200 ng each of Clock and Bmal1 cDNA expression construct to test for E-box function. Only pE2 showed significant transactivation by CLOCK:BMAL1. Two-way ANOVA revealed a main effect of construct (F = 334.9, P < 10^-4), a main effect of Clock/Bmal1 cotransfection (F = 491.6, P < 10^-4), and an interaction between construct and Clock/Bmal1 cotransfection (F = 381.4, P < 10^-4). Each value is the mean ± SEM of three replicates from a single assay. The results shown are representative of at least three independent experiments. (C) The 210-bp mPer2 E2 putative enhancer was cloned into the pGL3-Basic and pGL3-Promoter (SV40) vectors (pGL3Basic-E2 and pGL3Promoter-E2, respectively). Also, E2 in the reverse orientation was cloned into the pGL3-Promoter vector (pGL3Promoter-RevE2). The E2 sequence, 5′-CACGTT-3′, was mutated to E2mut 5′-GCTAGT-3′ (mutE) by using site-directed mutagenesis in pGL3Basic-E2, pGL3Promoter-E2, and pGL3Promoter-RevE2 to make pGL3Basic-mutE2, pGL3Promoter-mutE2, and pGL3Promoter-mutRevE2, respectively. White bars represent cotransfection of Clock and Bmal1 cDNA expression constructs with E2 constructs; black bars represent controls with no Clock and Bmal1 cDNA construct cotransfection. None of the mutant E2-containing reporters were transactivated by Clock/Bmal1 cotransfection (right half of chart). Each value is the mean ± SEM of three replicates from a single assay. The results are representative of at least three independent experiments. Although it did not contain the SV40 promoter sequence, pGL3Basic-E2 exhibited a capacity to drive transcription initiation and transactivation by CLOCK:BMAL1, similar to the SV40-containing construct (pGL3Promoter-E2). The asterisk indicates the construct used subsequently to generate E2 enhancer transgenic animals. (D and E) The E2 enhancer sequence in the pGL-6 and pGL2 constructs (Fig. 1B) was mutated to 5′-GCTAGT-3′ by using site-directed mutagenesis to create pGL-6-mutE2 and pGL-2-mutE2, respectively. White bars represent cotransfection of Clock and Bmal1 cDNA expression constructs; black bars represent reporter constructs alone. Compared with pGL-6 and p-GL2, the pGL6-mutE2 and pGL2-mutE2 constructs were not significantly transactivated by Clock/Bmal1 cotransfection. Each value is the mean ± SEM of three replicates from a single assay. The results shown are representative of at least three independent experiments.

The candidate E-box information was independently confirmed by using DNaseI hypersensitive site-mapping assays (27, 28). Nuclei were isolated from Hepa1–6 cells and from liver of C57BL/6J female mice at zeitgeber time (ZT) 12. Five strong DNaseI hypersensitive sites were found between base pairs -300 and +30 (Fig. 1C). The first strong band in Fig. 1C Left and the second band in Fig. 1C Right are consistent with the E2 site (Fig. 1C, arrows). An additional hypersensitive site was located around base pair +1,500, which corresponds to the E2a region. Based on the series of deletion mutant reporter experiments and DNaseI hypersensitive analyses, we identified the E2 site at base pairs -20 to -15 as a primary E-box candidate.

Sequence analysis revealed the following three additional E-box candidates upstream of base pair -1,128: E3, E4, and E5 (Figs. 1A and 2A). Among these E-boxes, only E2 and its flanking region are well conserved at the orthologous human Per2 locus on chromosome 2q37.3. Also, the E2 sequence is conserved at 12 of the 14 bases forming the CLOCK:BMAL1 M34 core binding site 5′-(G/T)G(A/G)ACACGTGACCC-3′ (Fig. 2A) (31).

Next, we examined the ability of CLOCK:BMAL1 to activate transcription from each of the putative E-box elements in isolation (pE1 to pE5). Only the E2 sequence was highly responsive to Clock/Bmal1 (Fig. 2B). As expected for a circadian-relevant E-box, the activation of E2 by CLOCK:BMAL1 was completely inhibited by mCry1 or mCry2 cotransfection (data not shown). E2a, which we identified as a potential active E-box (Fig. 1 B and C), was not activated by Clock/Bmal1 (Fig. 2B).

To test whether the E2 site can act as an enhancer, we assayed its dependence on orientation, and we found that it was not affected by reverse orientation (Fig. 2C). We next performed site-directed mutagenesis and changed the 5′-CACGTT-3′ sequences in a number of E2 constructs. As anticipated, CLOCK:BMAL1-dependent activation was abolished in all three mutant constructs (Fig. 2C). Indeed, mutation of E2 alone in the large pGL-6 construct (Fig. 1B), which contains the E2a and E1 sequences as well as a putative 5′-CATGTG-3′ E-box at base pair -160 (23), caused a substantial reduction (>85%) of CLOCK:BMAL1 activation (Fig. 2D). Similar results were obtained in experiments with pGL-1, pGL-2, and pGL-2mutE2 (Fig. 2E), suggesting that the residual activation observed in pGL-6-mutE2 and pGL-2mutE2 results from the presence of the 5′-CATGTG-3′ E-box at base pair -160. These findings confirm our results, showing that the E2 element is the primary active cis element in the mPer2 promoter.

ChIP Assays. We next examined whether endogenous CLOCK complex binds to the E2 enhancer sequence in vivo. Liver nuclei from C57BL/6J mice were collected every 6 h in light/dark (LD) conditions, and every 3 h on the second day of exposure to constant darkness over one circadian cycle. Integrity of collected nuclei was monitored by Western blotting the nuclear extracts for mPER2 and CLOCK. Consistent with previous work (11), mPER2 showed robust circadian oscillations in abundance and phosphorylation, whereas CLOCK exhibited modest circadian oscillations in both (Fig. 7). We used anti-CLOCK antibody to immunoprecipitate formaldehyde-crosslinked chromatin from C57BL/6J mouse liver nuclei (11). As a positive control for ChIP using anti-CLOCK antibody, we used an E-box from mPer1 (11). Because of the presence of large DNA fragments (≈2–5 kb) after sonication (Fig. 3A), E1, which is located ≈2 kb downstream from E2, was also detected after PCR in our initial ChIP analyses. To resolve this problem, we modified the procedure to obtain DNA fragments ≈1 kb in size, and only E2 was detected in the immunoprecipitated samples using these revised conditions (Fig. 3B). Next, we performed real-time PCR with samples collected over one circadian cycle in constant darkness to determine whether there was a rhythm in CLOCK complex binding to the E2 element. Quantitative PCR detected CLOCK bound to E2 at all times examined (Fig. 3C); no significant difference in immunoprecipitated CLOCK levels was evident (data not shown). It is clear from our data that CLOCK binds to E2 constitutively over the circadian cycle (Fig. 3C). Quantitative PCR with samples collected in LD conditions produced similar results (data not shown). This result is consistent with previously reported results showing constitutive CLOCK and BMAL1 binding to mPer1 E-boxes (11). We conclude that although E2 is a noncanonical E-box, it effectively binds CLOCK:BMAL1 both in vitro and in vivo, whereas E1, which is a canonical E-box, does not significantly bind CLOCK:BMAL1 under either condition.

Fig. 3.

ChIP assay for E2. (A) Agarose gel electrophoresis after sonication of chromatin isolated at ZT 6 and 12 from mouse liver nuclei. Note that the resulting DNA fragments are generally >3kb (Left). ZT 6 and 12 samples were combined for ChIP (Start). PCR analysis of the Start (input genomic DNA) with primers to E1, E2, and E5 reveals the presence of all three E-boxes. PCR analysis of ChIP with CLOCK antibody reveals the presence of E1 and E2 only (Right). (B) In the presence of SDS, sonication yielded DNA fragments of ≈1 kb. Immunoprecipitation with CLOCK antibody, followed by PCR, reveals the presence of E2 alone. (C) CLOCK was bound to E2 constitutively over the circadian cycle. The data were normalized to the input control and plotted as a percentage relative to the highest value (100%). The results are given as mean ± SEM of three experiments.

mPer2 E2 Enhancer Drives the Self-Sustained Circadian Expression of Luciferase in Transgenic Mice. To validate our molecular results, we generated transgenic mice by using the pGL3Basic-E2 construct (Fig. 2C, asterisk) to test whether a 210-bp fragment containing the E2 enhancer sequence is sufficient to drive the rhythmic expression of luciferase in the organism. Of 10 E2 transgenic mouse founder lines (mPer2-E2::Luc), 4 lines showed a circadian rhythm of luciferase expression. Two of these lines, E2–1 and E2–29, were chosen for further photomultiplier tube (PMT) bioluminescence measurements. Southern blot analysis revealed that the E2–1 line contained two to three copies of the transgene, whereas the E2–29 line contained ≈10–12 copies (data not shown). After entrainment of the mPer2-E2::Luc transgenic animals to a 12:12 LD cycle for at least 7 days, SCN and peripheral tissues were removed and placed in static cultures containing media supplemented with luciferin substrate, as described (1). By using PMT analysis, we successfully monitored the luminescence signal produced by the 210-bp E2 fragment in SCN, cornea, lung, and pituitary tissues for 7 days (Fig. 4A). As expected, a robust and sustained circadian rhythm of luminescence was evident in SCN tissue over the 7-day period, consistent with similar studies using mPer1 promoter fragments to drive luciferase expression in transgenic animals (25, 32). Intriguingly, similar to our data obtained from mPer2^Luc knock-in animals (1), peripheral tissues from the mPer2-E2::Luc transgenic mice also expressed a robust self-sustained circadian rhythm of luminescence, which continued for 7 days, a phenomenon not observed in the aforementioned mPer1::Luc transgenic studies. In particular, note that the observation that the circadian rhythm of luciferase expression in all tissues for which measurements were obtained mirrored the reported phase of the endogenous mPer2 RNA rhythm (11, 33, 34).

Fig. 4.

Real-time analysis of circadian expression of luciferase driven by the mPer2 E2 enhancer in transgenic mice. (A) Records of bioluminescence recorded from static SCN, pituitary, cornea, and lung tissue explants from mPer2-E2::Luc transgenic mice. A robust circadian oscillation in bioluminescence is evident in all four tissues for 7 days and correlates with the normal phase observed for mPer2 RNA rhythms. (B) Phase data for central and peripheral circadian oscillators in mPer2-E2::Luc transgenic animals compared with mPer2^Luc knock-in animal phase data (1). The peak of the circadian oscillation was determined during the interval between 12 and 36 h in culture. The average times ± SEM of peaks were plotted against the time of last lights-on. The LD cycle to which the mPer2-E2::Luc transgenic (n = 6; 3 heterozygous males each, from lines E2–1 and E2–29) and mPer2^Luc knock-in (n = 4) animals were exposed before tissue removal is shown above the plot (white and black bars). The phase of luciferase rhythmic expression in these tissues mirrors that of mPer2 RNA expression observed in other studies. Data for SCN, cornea, pituitary, and lung are shown. ○, mPer2-E2::Luc transgenic animals; •, mPer2^Luc knock-in animals. (C) Period plot of luciferase expression in central and peripheral tissues of mPer2-E2::Luc transgenic mice compared with mPer2^Luc knock-in animal period data (1). Mean period ± SEM of SCN, cornea, pituitary, and lung are shown. ○, mPer2-E2::Luc transgenic animals; •, mPer2^Luc knock-in animals.

mPer2^Luc Knock-in Animals as a Translation-Monitoring System, mPer2-E2::Luc Transgenic Animals as a Transcriptional Monitoring System. To examine further the phase relationships in luminescence between SCN and peripheral tissues, we compared the peak luciferase activity in tissues from the mPer2-E2::Luc transgenic animals with that observed in the mPer2^Luc knock-in animals (1) (Fig. 4B). In SCN tissue from the mPer2-E2::Luc transgenic animals, luciferase bioluminescence is highest late in the day, whereas it peaks ≈2–4 h later in SCN from the mPer2^Luc knock-in animals. This finding is consistent with our interpretation that the luciferase activity in the mPer2-E2::Luc transgenic animals reflects the endogenous mPer2 RNA rhythm, whereas in the knock-in animals, luminescence reports the daily mPER2 protein oscillation. In addition to the delay between mPer2 transcription and translation in the SCN, we also observe a similar phase delay in mPer2 expression in peripheral tissues for both the mPer2-E2::Luc transgenic animals and the mPer2^Luc knock-in animals, together yielding a significant main effect of genotype on luminescence peak time [ANOVA, F(1, 32) = 5.31, P = 0.028; Tukey–Kramer post hoc comparison, P ≤ 0.05]. Furthermore, there is a main effect of tissue such that in all tissues luminescence peak time differs significantly [ANOVA, F(3, 32) = 244.5, P = 0; Duncan's post hoc comparison, P ≤ 0.05]. By using different methods, others have shown an ≈3- to 6-h delay in mPER1 and mPER2 protein rhythms relative to the respective mRNA rhythms in free-running and entrained conditions in SCN and peripheral tissues (11, 33). Although the function of this delay is not known, a similar lag between transcription and translation has also been observed for the period gene in Drosophila (35, 36). Interestingly, a specific region in the 3′ untranslated region of mPer1 has been implicated in the repression of translation of mPER1 protein (37). Whether a related mechanism is involved in the delay between mPer2 transcription and translation remains to be determined.

As shown in Fig. 4C, period differences are evident in the mPer2-E2::Luc transgenic animals and in the knock-in animals for each tissue examined. The period differences observed from tissue to tissue in the mPer2-E2::Luc and mPer2^Luc knock-in animals reflect tissue-specific differences in clock gene expression. Within individual tissues, the period differences between the mPer2 RNA and protein rhythms are <1 h and do not attain statistical significance. However, from one tissue to another, the period differences may approach 2 h (e.g., cornea versus pituitary) and are statistically significant. Cornea and lung differ from each other and from all other tissues, whereas SCN and pituitary differ significantly only from cornea and lung but not from each other [ANOVA, F(3, 32) = 10.43, P = 0.00006; Duncan's post hoc comparison, P ≤ 0.05].

Circadian Expression of Luciferase in mPer2-E2::Luc Transgenic Mice Is Abolished in Homozygous Clock and Bmal1 Mutant Animals. To examine the effects of Clock and Bmal1 on the circadian oscillation of luciferase expression driven by the mPer2 E2 enhancer, mPer2-E2::Luc transgenic animals (line E2–1) were crossed to Clock mutant (38) and Bmal1 knockout mice (39). As expected, the circadian oscillation of luminescence was abolished in mPer2-E2::Luc^Clock(m/m) mutant SCN (Fig. 5A). Both mPer2-E2::Luc^Clock(m/+) and WT littermate SCN exhibit a circadian oscillation of luminescence. However, the luciferase expression from mPer2-E2::Luc^Clock(m/+) SCN is reduced compared with that observed in WT littermates. Consistent with behavioral phenotypes observed in previous studies (38), and the lengthened circadian period of the spontaneous firing rate in dispersed SCN cells (40), the period of the luciferase rhythm was lengthened by ≈1 h in mPer2-E2::Luc^Clock(m/+) SCN (Fig. 5C). No difference in period was observed in mPer2-E2::Luc^Bmal1(-/+) and WT littermate SCN cultures (Fig. 5 B and C). This result is consistent with the recessive behavioral circadian phenotype reported for the Bmal1 knockout, because the wheel-running period of Bmal1 heterozygous knockout mice does not differ from that of WT mice (39). Phase plots of the SCN luminescence data from the mPer2-E2::Luc^Clock(m/+) animals and WT littermates shows that the peak phase of the heterozygotes was significantly delayed (P = 0.031, Student's t test) (Fig. 5D). A similar phase delay was not observed in cultures of SCN from mPer2-E2::Luc^Bmal1(-/+) animals compared with WT littermates.

Fig. 5.

Effects of Clock and Bmal1 mutations on circadian expression of luciferase in mPer2-E2::Luc transgenic mice. (A) Bioluminescence recorded from SCN explants from mPer2-E2::Luc transgenic mice on the Clock mutant genetic background (E2–1-Clock). Robust circadian oscillations are evident in both Clock WT (+/+) and heterozygous (m/+) SCN. However, E2–1-Clock heterozygotes displayed a longer period and a delayed phase (refer to Fig. 5 B and C) than WT tissue. Circadian oscillations of bioluminescence were abolished in E2–1-Clock homozygotes (m/m). (B) Bioluminescence recorded from SCN explants from mPer2-E2::Luc transgenic mice on the Bmal1 knockout genetic background (E2–1-Bmal1). Robust circadian oscillations in bioluminescence are evident in both Bmal1 WT (+/+) and heterozygous (-/+) SCN and did not differ significantly. Circadian oscillations of bioluminescence were abolished in E2–1-Bmal1 homozygotes (-/-). (C) Period analysis of E2–1-Clock and E2–1-Bmal1. The circadian rhythms in bioluminescence were significantly different between E2–1-Clock WT (+/+) and heterozygous (m/+) animals (period, 23.56 and 24.48 h, respectively; P = 0.015), whereas E2–1-Bmal1 WT (+/+) and heterozygous (-/+) mice were not significantly different from each other (period, 24.00 and 23.94 h, respectively; NS). (D) Phase data for E2–1-Clock and E2–1-Bmal1. The peak of the circadian oscillation was determined during the interval between 12 and 36 h in culture. The average times ± SEM of peaks were plotted against the time of last lights-on. The LD cycle to which the animals were exposed before tissue removal is shown above the plot (white and black bar). The peak phases were significantly different between E2–1-Clock WT (Upper, blue circle) and heterozygous (Upper, red circle) animals (peak at 37.38 and 39.44, respectively; P = 0.031). The peaks of the circadian oscillation in E2–1-Bmal1 WT (Lower, blue circle) and heterozygous (Lower, red circle) mice were not significantly different from each other (peak, 34.84 and 35.02, respectively; NS).

Discussion

One of our goals is to understand how daily changes in the transcription and translation of clock genes lead to the generation of circadian rhythms in mammals. Recently, we developed a mouse knockin PERIOD2::LUCIFERASE fusion-protein reporter system to study the dynamics of mPER2 protein oscillations in real time. The mPer2^Luc system demonstrated that, contrary to previous reports, peripheral tissues exhibit self-sustained, rather than damped, circadian oscillations (1). To understand the mechanism of the robust self-sustained circadian oscillation that we observed in mPer2^Luc knock-in mice further, we have studied the regulation of the mPer2 gene with the aim of identifying cis-regulatory elements mediating transcriptional activation by means of CLOCK:BMAL1 heterodimers. However, unlike the mPer1 locus (21), we did not find any canonical E-boxes in the mPer2 promoter region, with the exception of E1 in the first intron (+2,066 from the transcription start site). Interestingly, we found a noncanonical 5′-CACGTT-3′ E-box enhancer (referred to here as E2) ≈20 bp upstream of the transcription initiation site that alone is sufficient to drive transactivation in cell transfection experiments with CLOCK:BMAL1. This finding is supported by the occurrence of a DNaseI hypersensitive site in the E2 region, indicative of the presence of a transacting factor bound to the cis-acting regulatory sequence. ChIP experiments also demonstrated that a native complex containing the CLOCK protein is bound to the E2 enhancer in cells. Also important, the mouse E2 5′-CACGTT-3′ E-box and >30 bp of flanking sequence share perfect identity with sequence in the 5′ regulatory region of the human Per2 gene. The E2 enhancer sequence is also conserved at 12 of 14 bases forming the CLOCK:BMAL1 M34 core binding site (Fig. 2A) (31).

After we confirmed that E2 is a functional cis-acting element for CLOCK:BMAL1 binding in vitro and in vivo, we generated reporter transgenic mouse lines with a 210-bp E2-containing fragment to validate its function as a circadian cis-acting element. We were able to monitor circadian rhythms of luciferase from both SCN and peripheral tissues that reflect the endogenous mPer2 RNA oscillation. Last, genetic experiments confirm that the E2 enhancer is a target of CLOCK and BMAL1 because mutations in these transcriptional activators lead to a loss of circadian oscillation in the SCN of E2 enhancer transgenic mice.

A recent microarray analysis of 127 genes exhibiting a circadian pattern of expression and for which at least 10 kb of 5′ upstream human and mouse sequence information was available, revealed that only nine genes, including Per2, contained a CLOCK:BMAL1 5′-CACGTGA-3′ binding site in both species (41). Our present results for the E2 sequence argue strongly that additional noncanonical E-box sequences may be identified in genes under circadian control but that they lack the canonical 5′-CACGTG-3′ sequence. Perhaps screening the 5′ upstream regions of such genes using the longer M34 core sequence would be more successful in yielding circadian-relevant cis-acting elements. In contrast to the mPer1 E-boxes, which act additively to activate maximal expression (21), the mPer2 E2 E-box identified here is more similar to the Drosophila period E-box (42, 43). Groups studying Drosophila per have used methods similar to ours by creating lacZ and luciferase reporter systems driven by the per promoter region and identified a 5′-CACGTG-3′-containing 69-bp sequence sufficient to generate rhythmic per transcription and proper spatial expression. In a similar manner, not only was the E2 enhancer identified in this study able to drive strong circadian rhythms of bioluminescence in peripheral tissue with a phase delay relative to the SCN rhythm, the circadian rhythm in the peripheral tissues persisted in a self-sustained manner for >7 days in culture. Also, when we analyzed the circadian phase for the bioluminescence data from the mPer2-E2::Luc transgenic and mPer2^Luc knock-in animals, the peak of mPer2 RNA rhythm in the SCN occurred ≈3 h before the mPER2 protein peak. Period analysis revealed that tissues from mPer2-E2::Luc transgenic mice have equivalent periods with the mPer2^Luc knock-in animals. We did not observe this robust rhythm in mPer1::Luciferase transgenic mouse peripheral tissues. Analysis of three different circadian reporter systems (mPer1::Luc and mPer2-E2::Luc transgenic mice, as well as mPer2^Luc knock-in mice) suggest that the persistence of circadian oscillations in peripheral tissues is likely due to a difference in mPer1 and mPer2 regulatory elements. In mPer1, the proximal promoter region is not sufficient to drive circadian oscillations in peripheral tissues, whereas it is sufficient in mPer2. It may be interesting to compare mPer1 and mPer2 luciferase knock-in mice to determine whether, in mPer1, distal regulatory elements are necessary for circadian oscillations in peripheral tissues.