Functional D-box sequences reset the circadian clock and drive mRNA rhythms

To develop genome-wide mapping of DBP-binding sites and E4BP4-binding sites, the ChIP DNA fragments were prepared from E4bp4-KO mouse and littermate control livers at ZT12 for the DBP-ChIP and at ZT24 for the E4BP4-ChIP. The ChIP samples were subjected to deep sequencing, which yielded 70–100 million tags in each sample (Supplementary Data 1). These tags were mapped onto the mouse genome, and the peak calling program MACS2 identified 6066 DBP-binding sites at ZT12 and 3064 E4BP4-binding sites at ZT24 in the control livers (Supplementary Data 1). The E4BP4-ChIP signals at the 3064 E4BP4-binding sites were significantly reduced in E4bp4-KO livers (wild type: 16.0 ± 12.4, E4bp4-KO: 1.8 ± 0.2, p = 1.3 × E-111, two-sided Student’s t test). Among the 3064 E4BP4-binding sites, DBP-binding signals were detected at 1490 sites, which we defined as DBP/E4BP4-common sites. When the 1490 common sites were compared with the previous E4BP4-ChIP data¹⁰, E4BP4-binding signals were detected at 1284 sites (1284/1490 = 86.2%), indicating high reliability of the current ChIP-Seq data. Typically, strong peaks of DBP-ChIP tags at ZT12 and E4BP4-ChIP tags at ZT24 were detected at the Per1 TSS region (Fig. 1c), consistent with the ChIP-PCR analysis (Fig. 1d). Notably, our ChIP-Seq data identified a so-far unidentified DBP-binding site in the Per1 −4.2 kb region, where no significant E4BP4-binding signal was detected (Fig. 1c, d; p = 0.49, two-sided Student’s t test). A similar DBP-preference site was detected in the intron 1 region (+2.7 kb region) of the Per2 gene locus, in which a DBP/E4BP4-common site was located in the TSS region (Supplementary Fig. 1c, d).

Among the 1277 rhythmic genes, 359 genes were E4bp4-dependent rhythmic genes that showed robust expression rhythms (p < 0.01) in the control but lost their rhythmicities (p ≥ 0.05) in the E4bp4-KO livers (Supplementary Fig. 4a). Among the E4bp4-dependent rhythmic genes, Marveld1 and Wee1 showed expression peaks at ZT12-16, whereas their expression levels were constantly high in the E4bp4-KO livers (Fig. 4b). These data are consistent with the expression rhythm of the E4BP4 repressor peaking at ZT0-2 (Fig. 1a, b). ChIP-Seq and ChIP-PCR analyses demonstrated rhythmic binding of DBP and E4BP4 to the promoter region of Marveld1 and to the intronic region of Wee1 (Fig. 4c, d). It is noted that D-box#2 (TTATGTAA) was found at around the DBP/E4BP4-common sites in the Marveld1 and Wee1 gene loci. Gene ontology (GO) analysis showed that the E4bp4-dependent rhythmic genes were enriched with genes involved in metabolic pathways (Supplementary Fig. 4b). It has also been reported that E4bp4-KO mice suffer from inflammatory diseases of the intestine^14–16. The current study will provide clues to understanding E4bp4-mediated transcriptional regulation in other physiological processes.

To examine whether E4BP4 induction is required for the acid-evoked phase resetting, we isolated E4bp4-KO/PER2::Luc MEFs, in which circadian rhythms can be examined by real-time monitoring of the bioluminescence from Luciferase fused with endogenous PER2 protein. In the control PER2::Luc MEFs, we observed apparent phase shifts, when pHo of the cultured media was shifted from 7.0 to 6.6 (Fig. 5a) several hours after the trough time of the bioluminescence rhythms (Fig. 5d), as was previously observed in rat-1 fibroblasts²⁰. On the other hand, faint phase shifts were induced by the acidification several hours after the peak time (Supplementary Fig. 5c). Phase response curves and phase transition curves showed phase-dependent phase shifts and revealed type-0 resetting of cellular rhythms in response to the acidification, respectively (Fig. 5e, f, left). Importantly, the acidification-induced phase shifts were almost completely blocked in the E4bp4-KO/PER2::Luc MEFs (Fig. 5d–f, right), indicating an essential role of the acute induction of E4BP4 in acid-evoked phase resetting. The present work revealed physiological roles of D-box-mediated transcriptional regulation, which is important for non-photic phase control in the peripheral clock.

The animal experiments were approved by the animal ethics committee of the University of Tokyo. C57BL/6J mice were individually housed in cages with free access to food and water. E4bp4-KO mice were kind gifts from A. Thomas Look (Children’s Hospital Boston, Harvard Medical School). In the E4bp4-KO mice, the exon 2 of E4bp4 gene was replaced by a neomycin cassette, as previously described³⁶. Mice were reared in 12-h light:12-h dark cycles in a light-tight chamber at a constant temperature (23 ± 1 °C). PER2::LUC knock-in mice³⁷ were used for monitoring bioluminescence rhythms. Wheel-running activity rhythms were monitored and analyzed with Clocklab software (Actimetrics) developed on MatLab (Mathworks), as previously described³⁸.

We generated anti-DBP and anti-E4BP4 antibodies in rabbits, now commercially available as anti-DBP (MBL, PM079) and anti-E4BP4 antibodies (MBL, PM097). We also used anti-rhodopsin 1D4³⁹, anti-TBP (Santa Cruz Biotechnology, sc-421) and anti-beta-actin (Sigma, A2228). In immunoblot analysis, the bound primary antibodies were detected by horseradish peroxidase-conjugated anti-rabbit or anti-mouse IgG antibody (Kirkegaard & Perry Laboratories).

The nuclear proteins were isolated as previously described^40,41. Mouse tissue (1 g, wet weight) was washed with ice-cold PBS and homogenized at 4 °C with 9 ml of ice-cold buffer A (10 mM HEPES-NaOH, 10 mM KCl, 0.1 mM EDTA, 1 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride (PMSF), 4 μg/ml aprotinin, 4 μg/ml leupeptin, 50 mM NaF, and 1 mM Na₃VO₄; pH 7.8). The homogenate was centrifuged twice (700 × g, 5 min each), and the precipitate was resuspended in 2 ml of ice-cold buffer C (20 mM HEPES-NaOH, 400 mM NaCl, 1 mM EDTA, 5 mM MgCl₂, 2% glycerol, 1 mM DTT, 1 mM PMSF, 4 µg/ml aprotinin, 4 µg/ml leupeptin, 50 mM NaF, and 1 mM Na₃VO₄; pH 7.8). After gentle mixing at 4 °C for 30 min, the suspension was centrifuged twice (21,600 × g, 30 min each), and the final supernatant was used as the “nuclear extract”.

ChIP analysis was prepared as described previously¹¹ with minor modifications. Livers were isolated at two time points, ZT12 and ZT24 (n = 3), from E4bp4-KO and the WT littermate mice. They were rinsed with ice-cold PBS and were homogenized with ice-cold buffer A (10 mM HEPES-NaOH, 10 mM KCl, 0.1 mM EDTA, 1 mM DTT, 1 mM PMSF, 4 μg/ml aprotinin, 4 μg/ml leupeptin, 50 mM NaF, and 1 mM Na₃VO₄; pH 7.8). The homogenate was centrifuged twice (700 × g, 5 min each), and the precipitate (nuclear fraction) was cross-linked by 1% formaldehyde in buffer A for 10 min at 25 °C. The cross-linking reaction was stopped by addition of 125 mM glycine (final concentration). The sample was then centrifuged (700 × g, 5 min), and the nuclei pellet was washed twice with buffer A and resuspended in IPB2 buffer (20 mM HEPES-NaOH, 137 mM NaCl, 1 mM EDTA, 5% glycerol, 1% Triton X-100, 1.67 mM MgCl₂, 1 mM DTT, 1 mM PMSF, 4 μg/ml aprotinin, 4 μg/ml leupeptin, 50 mM NaF, and 1 mM Na₃VO₄; pH 7.8) supplemented with 1% SDS. The sample was then sonicated 16 times for 20 s each at intervals of 40 s (Branson Sonifier 450; set at 50% duty cycle, five output). The supernatant was diluted in IPB2 (final 0.1% SDS), and snap-frozen in liquid nitrogen. After thawing, the sample was centrifuged at 20,000 × g for 10 min at 4 °C, and the supernatant was incubated with anti-DBP antibody, anti-E4BP4 antibody, or an irrelevant antibody 1D4 while being gently rotated for 2 h at 4 °C. Protein G-coupled magnetic beads (Dynabeads, Dynal) were added to the mixture, followed by gentle rotation for 1 h at 4 °C. The beads were washed sequentially with the following buffers by using DynaMag-2 magnet: (i) IPB2 buffer; (ii) IPB2 buffer supplemented with 500 mM NaCl; (iii) TE buffer (10 mM Tris-HCl, 1 mM EDTA; pH 8.0) supplemented with 0.25 M LiCl, 1% NP-40, and 1% deoxycholate; and (iv) TE buffer. Finally, the beads were treated with 500 µl of the elution buffer (1% SDS, 0.1 M NaHCO₃) and gently rotated for 30 min at room temperature, and the eluate was mixed with 20 µl of 5 M NaCl and incubated overnight at 65 °C. The de-cross-linked sample was then mixed with 10 μl of 0.5 M EDTA, 20 μl of 1 M Tris-HCl (pH 6.5) and 2 μl of 10 mg/ml Proteinase K, and the mixture was incubated for 2 h at 45 °C. The DNA was purified by extraction with phenol-chloroform-isoamyl alcohol (25:24:1) and subjected to ethanol precipitation. The final precipitate was used as the ChIP sample.

The DBP-ChIP (at ZT12, n = 2) and E4BP4-ChIP samples (at ZT24, n = 2) prepared from E4bp4-KO mice and the WT littermates were sequenced on a HiSeq 3000 sequencer (36 bp, single end). The input samples prior to the immunoprecipitation were also subjected to the deep sequencing as controls. The sequence tags were mapped to the mouse genome by using Bowtie (v1.2.1) with parameter setting of “−a –best –strata -m 1 −p 4”⁴². BAM files of biological duplicates were merged using the “samtools merge” command. Peak calling was performed for the merged ChIP samples versus the merged input using MACS2 (v2.1.1) with default parameters⁴³. The mapped tags were visualized by using Integrative Genomics Viewer⁴⁴.

The equation that calculates [SD of AUC] from [appearance counts] was mathematically derived as follows. Let W be the size of the analyzed window where k-mer sequences are sought at around ChIP-peak positions. If a k-mer sequence appears only once at a random position within the window, its coordinate follows the uniform distribution U(0, W), whose variance is known to be W²/12. Because (i) AUC is calculated by subtracting W/2 from the coordinate and (ii) constant subtraction does not affect variance of probability distributions, variance of AUC is also W²/12 if the appearance count is 1. Next, assume that a k-mer sequence appears C times at random positions within the window. The variance of the sum of their coordinates becomes CW²/12, because variance of sum of random variables that follow the same probability distribution is proportional to the numbers of the variables. Then, because AUC is calculated by dividing the sum of their coordinates by C and subtracting W/2, the variance of AUC is (CW²/12)/C² = W²/12 C, if the appearance count is C. Finally, we obtain [SD of AUC] by taking the square root of the variance:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[{\mathrm{SD}}\, {\mathrm{of}}\, {\mathrm{AUC}}] = \frac{[{\mathrm{window}}\, ({\mathrm{W}})]}{{\sqrt{12}} \times {\sqrt {[{\mathrm{appearance}}\, {\mathrm{count}}\, ({\mathrm{C}})]}}}$$\end{document}[SDofAUC]=[window(W)]12×[appearancecount(C)]

In MOCCS version 2 (abbreviated as MOCCS2), the “MOCCS2 score” of each sequence was defined as a relative value of AUC normalized by the SD at its appearance count: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{MOCCS2}}\, {\mathrm{score}} = \frac{\mathrm{[AUC]}}{{[{\mathrm{SD}}\, {\mathrm{of}}\, {\mathrm{AUC}}]}}$$\end{document} MOCCS2 score = [AUC] [ ] SD of AUC

In this study, W was set to 250 + 1 – (k/2), because k-mer sequences do not appear at the end of the 250-bp windows. If k = 8, [SD of AUC] was 71.303/C^0.5, as shown in the Result section. If k = 6, [SD of AUC] was 71.591/C^0.5. The MOCCS2 is freely available via https://github.com/yuifu/moccs↗.

RNA-Seq analysis was performed as previously described⁴⁵ with minor modifications. The total RNA was prepared from livers of E4bp4-KO/PER2::Luc mice and the littermate PER2::Luc mice at six time points throughout the day (ZT0, 4, 8, 12, 16, and 20; n = 2) by using the TRIzol reagent (Invitrogen) and RNeasy mini kit (QIAGEN) according to the manufacturer’s protocol. Poly(A)-tailed RNA was isolated from the total RNA as the manufacturer’s protocol, and was sequenced on a HiSeq 3000 (36 bp, single end). The mouse genome sequence was obtained from UCSC Genome Browser (mm10, http://genome.ucsc.edu/↗). The annotated gene models (GRCm38) were taken from Ensembl (release 95, http://www.ensembl.org/↗). Hisat2 (v2.1.0) was used for mapping RNA-Seq data with parameter setting of “-p 4 —dta -q -x”⁴⁶. The expression level of each gene was quantified as fragments per kilobase of exon per million fragments (FPKM) by using both StringTie (v1.3.4) with parameter setting of “-e -G”⁴⁷ and Ballgown (v2.12.0) with default parameters⁴⁸. A gene was defined as “expressed” if the average of FPKM values of the 12 samples (6 time points, duplicate) in E4bp4-KO or control mice was higher than 1.0. A gene was defined as “rhythmic” if JTK cycle program⁴⁹ detected any circadian rhythmicity with p < 0.05.

For quantitative PCR (qPCR) analysis, the ChIP samples were subjected to real-time PCR (Applied Biosystems) using GoTaq Master Mix (Promega) with gene-specific primers (Supplementary Table). For qRT-PCR, the total RNA samples prepared at ZT0, 4, 8, 12, 16, and 20 were reverse transcribed by Go Script Reverse Transcriptase (Promega) with both an anchored (dT)15 primer and a random oligo primer. The cDNA samples were subjected to the qPCR analysis with gene-specific primers (Supplementary Table). 1 1

HEK293T17 cells in 24-well plates were transiently transfected by using polyethylenimine (Polysciences, #24765) with 100 ng Flag-DBP/pSG5 or Flag-E4BP4/pSG5 in combination with 10 ng of firefly luciferase reporter plasmids and 0.5 ng of a Renilla luciferase plasmid (pRL-SV40) as an internal control. The total amount of DNA was adjusted to 410.5 ng by adding the empty expression plasmid pSG5. A triple tandem repeat of one of D-box sequences was inserted into a BglII site of a firefly luciferase reporter plasmid (pGL3N) as previously described¹¹. The inserted sequences were shown in Supplementary Table 1. The transfected cells were collected 36 h after the transfection and subjected to the dual luciferase assay according to the manufacturer’s protocol with the aid of a fluorescence plate reader (Promega GloMax). Internal control was used to normalize the transfection efficiency.

Cellular bioluminescence rhythms were monitored as described previously⁵⁰ with minor modifications. In brief, MEFs were prepared from PER2::LUC knock-in mice³⁷. The MEFs were maintained at 37 °C under 5% CO₂, 95% air in Dulbecco’s modified Eagle’s medium (SIGMA) supplemented with 25 units/ml penicillin, 25 µg/ml streptomycin, and 10% fetal bovine serum. PER2::Luc MEFs were plated on 35-mm dishes (1.0 × 10⁶ cells/dish) and cultured at 37 °C under 5% CO₂. After 24 hr, the cells were treated with 0.1 µM (final) dexamethasone (Dex) for 2 h, and then the media were replaced by a recording media (phenol-red free Dulbecco’s modified Eagle’s medium (SIGMA) supplemented with 10% fetal bovine serum, 3.5 g/l glucose, 25 U/ml penicillin, 25 µg/ml streptomycin, 0.1 mM luciferin, and 10 mM HEPES-NaOH; pH 7.0). The bioluminescence signals of the cultured cells were recorded continuously for 5–10 days at 37 °C in air with Dish Type Luminescencer, Kronos (Atto, AB-2500 or AB-2550) or LumiCycle (Actimetrics).

For acid treatment, extracellular pH (pHo) was shifted from 7.0 to 6.6 by adding a minimal volume of 1 M HCl solution to the cultured media as previously described²⁰. In control experiments, the same volume of water was added (H₂O). Circadian time (CT) 0 was defined as the time points of the troughs of the bioluminescence signal waveforms. The phase shifts were calculated from the time of peaks and troughs of the bioluminescence rhythms.

Further information on research design is available in thelinked to this article. Nature Research Reporting Summary

Supplementary Information Description of additional supplementary items Reporting Summary Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Supplementary Data 4 Supplementary Data 5