Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2012;10(11):e1001442.
doi: 10.1371/journal.pbio.1001442. Epub 2012 Nov 27.

Genome-wide RNA polymerase II profiles and RNA accumulation reveal kinetics of transcription and associated epigenetic changes during diurnal cycles

Gwendal Le Martelot  1 , Donatella Canella, Laura Symul, Eugenia Migliavacca, Federica Gilardi, Robin Liechti, Olivier Martin, Keith Harshman, Mauro Delorenzi, Béatrice Desvergne, Winship Herr, Bart Deplancke, Ueli Schibler, Jacques Rougemont, Nicolas Guex, Nouria Hernandez, Felix Naef; CycliX Consortium
Collaborators, Affiliations

Genome-wide RNA polymerase II profiles and RNA accumulation reveal kinetics of transcription and associated epigenetic changes during diurnal cycles

Gwendal Le Martelot et al. PLoS Biol. 2012.

Abstract

Interactions of cell-autonomous circadian oscillators with diurnal cycles govern the temporal compartmentalization of cell physiology in mammals. To understand the transcriptional and epigenetic basis of diurnal rhythms in mouse liver genome-wide, we generated temporal DNA occupancy profiles by RNA polymerase II (Pol II) as well as profiles of the histone modifications H3K4me3 and H3K36me3. We used these data to quantify the relationships of phases and amplitudes between different marks. We found that rhythmic Pol II recruitment at promoters rather than rhythmic transition from paused to productive elongation underlies diurnal gene transcription, a conclusion further supported by modeling. Moreover, Pol II occupancy preceded mRNA accumulation by 3 hours, consistent with mRNA half-lives. Both methylation marks showed that the epigenetic landscape is highly dynamic and globally remodeled during the 24-hour cycle. While promoters of transcribed genes had tri-methylated H3K4 even at their trough activity times, tri-methylation levels reached their peak, on average, 1 hour after Pol II. Meanwhile, rhythms in tri-methylation of H3K36 lagged transcription by 3 hours. Finally, modeling profiles of Pol II occupancy and mRNA accumulation identified three classes of genes: one showing rhythmicity both in transcriptional and mRNA accumulation, a second class with rhythmic transcription but flat mRNA levels, and a third with constant transcription but rhythmic mRNAs. The latter class emphasizes widespread temporally gated posttranscriptional regulation in the mouse liver.

PubMed Disclaimer

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Figure 1
Figure 1. Pol II, H3K4me3, and H3K36me3 genomic profiles of core circadian clock genes measured around the clock.
(A) The density profiles of Pol II (red), H3K4me3 (green), and H3K36me3 (blue) are indicated for the Bmal1 gene, which spans 107 kb on chromosome 7, with the thin lines above the profiles indicating the position-specific temporal maxima. The gene structure (RefSeq transcripts) is shown below the panel. The dashed lines starting with a circle and the arrows represent minima and maxima, respectively, of gene body Pol II occupancy (red), promoter H3K4me3 occupancy (green), and gene body H3K36me3 occupancy (blue), as estimated by cosine fits (Materials and Methods). Maximal Pol II, H3K4me3, and H3K36me3 densities are reached at ZT21, ZT23, and ZT2. (B) As in (A), but for the RevErbα (Nr1d1) gene, which spans 7.3 kb on chromosome 11. Maximal Pol II, H3K4me3, and H3K36me3 densities are reached at ZT6, ZT9, and ZT9. Temporal animations of these profiles are provided as supplemental movies. Similar profiles for the circadian Per1 gene and constitutive Tbp gene are shown in Figure S1.
Figure 2
Figure 2. Genomic profiles of Pol II, H3K4me3, and H3K36me3 densities around transcription start sites (TSSs) and polyadenylation sites (PASs) at ZT2.
(A) Average signals over 11,217 genes with nonoverlapping TSSs (see Materials and Methods) for each mark around the TSSs: Pol II (red), H3K4me3 (green), H3K36me3 (blue), and input chromatin (gray). (B) As in (A) but around the PAS. (C) Average Pol II signals over transcripts split by quartile, based on the level of expression as measured by microarrays. Each quartile is represented by a distinct color shading from light (lowest quartile in mRNA expression) to dark (upper quartile in mRNA expression). (D) As in (C) but for the PAS. (E–F) Profiles of input chromatin. Note that the depletion at the PAS only partially explains the dips in panels (B) and (D). Vertical axes have arbitrary units, but the scales on the left and right panels can be compared for the same marks.
Figure 3
Figure 3. Temporal Pol II occupancies at promoters and in gene bodies oscillate with similar phases.
(A) Pol II occupancy at ZT2 in promoters (mean in ±1 kb regions around TSSs) versus gene bodies (mean over the regions from +1 kb to the PAS) for all genes in logarithmic scale. Color intensity indicates population density. One transcription unit per gene is shown (the selection is based on H3K4me3 and Pol II signals at promoters and PAS, as described in Materials and Methods). Two main populations can be distinguished: one with low Pol II occupancy in both promoter and gene body regions (lower left cloud), corresponding to silent or weakly transcribed genes, and one with higher Pol II occupancy within promoter regions and, to a lower extent, gene body regions (fainter cloud shifted slightly up and to the right). The bimodality of the promoter signal is clearly seen in the projection (histogram above the horizontal axis), whereas the signal in the gene body (elongating polymerases) has a lower dynamic range (histogram on the vertical axis shown in panel B). The cross sign, also shown in panels C and D, indicates background levels estimated from the lower maxima of the histograms. (B) Pol II occupancy at ZT2 in the first 1 kb window after the PAS (PAS1K) versus gene bodies (as in A). The two measures show high correlation, but PAS1K has a larger dynamic range (see Figure 2B). In (A) and (B), data are shown for ZT2, but all time points looked virtually identical. (C and D) Temporal profiles of Pol II promoter/gene body occupancy ratios for core clock genes (C) and selected output genes (D). Left, temporal profiles for promoters (red), gene bodies (brown), and PAS (orange) together with cosine fits. Right, the data from the left panels for the promoter and gene body are plotted against each other in the coordinate system of panel A. ZT times are color-coded (see color bar). Cross signs indicate background levels. Note logarithmic scale on axes. (E) Genome-wide analysis showing that Pol II occupancies at promoters and in gene bodies co-vary in time. Each line shows the average orientation and amplitude of changes during a diurnal cycle for genes in regions on a grid. The nonbinned plot is shown in Figure S3C.
Figure 4
Figure 4. H3K4me3 and H3K36me3 marks vary during the diurnal cycle with reduced amplitude as compared to Pol II occupancy.
(A) H3K4me3 promoter levels versus Pol II promoter occupancy at ZT2. Two populations can be identified from the densities: silent (or weakly active) promoters (lower left cloud) and active promoters (fainter cloud shifted above the diagonal and to the right). Bimodality in both signals is clearly seen in the projections (histograms). The cross sign, also shown in panels D and E, indicates background levels estimated from the lower maxima of the histograms. (B) H3K36me3 levels (quantified over the most 3′-proximal 40% of gene bodies) versus Pol II body occupancy at ZT2. Two populations can be identified from the densities: silent (or weakly transcribed) genes (lower left cloud) and transcribed genes. (C) H3K36me3 levels as in (B) versus H3K4me3 promoter levels at ZT2. This comparison shows the two classes most clearly, indicating that the large majority of genes harboring H3K4me3 marks are transcribed. In (A–C), data are shown for ZT2, but all time points looked identical. (D–E) Temporal profiles of H3K4me3 and H3K36me3 marks, and promoter Pol II occupancy for some core clock genes (D) and selected output genes (E). Left, temporal profile for promoter Pol II occupancy (red), H3K4me3 marks (green), and H3K36me3 marks (blue) together with cosine fits. Right, the cosine fits for Pol II promoter occupancy and H3K4me3 plotted against each other in the coordinates of panel A. ZT times are color-coded (see color bar). Crosses indicate background levels. Note that levels of H3K4me3 remain relatively high at the troughs of transcription (as measured by Pol II density). (F) Genome-wide temporal analysis showing that H3K4me3 modifications at promoters show compressed amplitudes compared to Pol II promoter occupancy (compare with Figure 3E). Each line shows the average orientation and amplitude of changes during a diurnal cycle for genes in regions of a grid. The nonbinned plot is shown in Figure S5.
Figure 5
Figure 5. Temporal relationships of Pol II, H3K4me3, H3K36me3 profiles, and mRNA accumulation in mouse liver.
(A) Phase histograms for cyclic genes. A selection of 284 genes (p<0.004, FDR = 0.3) showing cyclic patterns in all marks (see Materials and Methods) were fitted with a cosine function and the phase (peak time of the fit) was computed. These phases show a bimodal distribution for Pol II occupancy in promoters and gene bodies with maxima around ZT9 and ZT21, as well as in mRNA accumulation with a phase delay of approximately 3 h. (B) Phases for the same genes are represented in pairs, with color shade indicating p value (lower p values are darker) for the 24-h rhythm of the Pol II promoter signal. Relative to the phase of Pol II in promoters, we find high concordance for Pol II occupancy phases in gene bodies, an average delay of 1.3 h for H3K4me3 phases, and more spread H3K36me3 and mRNA phases with an average delay of about 3 h. Colored lines are mean-square regressions with intercepts corresponding to the average delays, as indicated in color. The thin dashed lines indicate ±2 h delays. (C) Temporal cross-correlation analysis. Using the same gene selection, we applied Fourier interpolation to obtain a continuous time trace (see Figure 3C,D and Figure 4D,E) and computed average cross-correlations between each mark and the corresponding Pol II promoter trace. Pol II occupancies in promoters and gene bodies are well-correlated and simultaneous, and H3K4me3 lags by about 1 h on average, whereas mRNA and H3K36me3 are phase-delayed by about 3 h. The same figure is shown for a more permissive selection (Figure S8, n = 752, p<0.018, FDR = 0.5).
Figure 6
Figure 6. Amplitude and phase relationships between Pol II signals and mRNA accumulation identify posttranscriptional regulation in mRNA accumulation.
(A) Relative amplitudes (maximum minus minimum, divided by twice the mean after background subtraction; see Materials and Methods) of oscillations in Pol II promoter signals and mRNA accumulation identify rhythmic mRNAs with relative amplitudes comparable to that of transcription (class 1, gray, 675 genes), long-lived transcripts with damped mRNA rhythms (class 2, orange, 668 genes), and mRNAs where posttranscriptional regulation increases rhythmic amplitude (class 3, red, 217 genes). Light gray genes are all genes that cycle robustly in either Pol II or mRNA accumulation (3,446 genes). Fus, Tfrc, and Spon2 are representative of class 3 and Rdbp of class 2 (see panels F–I for qRT-PCR validations). The few values larger than 1 are due to low signals when background subtraction makes trough values negative. (B) Half-lives of the three classes taken from NIH-3T3 fibroblasts show a significant difference (TukeyHSD p value <10−6 for class 2 versus class 1, and class 2 versus class 3). (C) Delays in peak mRNA accumulation versus peak Pol II promoter loading for the union of class 1, class 2, and class 3 genes. The dark gray region delimits the range predicted for a model with constant half-lives (0 h delay for very short-lived up to 6 h for very long-lived transcripts). (D–E) For the same genes the ratio of relative amplitudes (B = relative amplitude of Pol II, b = relative amplitude of mRNA) is plotted against the phase delay, together with the prediction for a constant half-life (red line). The trend (in D, median is the blue line and 25% and 75% percentiles are shown as light blue shading) shows that the ratio is centered on one at short delays and decreases for larger delays. The scatter plot (E) highlights genes for which transcript accumulation is explained by a constant half-life (dark gray area represents short and light gray long half-lives), and genes where nonconstant half-lives either suppress (light orange) or enhance (light red) amplitudes in mRNA accumulation. Triangles show core circadian clock genes. (F–I) Transcription and mRNA accumulation for representative genes. Comparison of (i) mRNA levels as measured by gene expression arrays, (ii) promoter Pol II occupancy as measured by ChIP-Seq, and (iii) pre-mRNA and (iv) mRNA levels as measured by qRT-PCR with intronic and exonic probes, respectively. Symbols and lines indicate measurements and cosine fits, respectively. Open symbols and dashed lines show qRT-PCR data (cDNA from n = 4 animals where pooled) with circles for the pre-mRNA and triangles for the mRNA. Continuous lines and filled symbols represent Pol II ChIP-seq (circles) and mRNA Affymetrix data (triangles). Each temporal profile has been scaled to an arbitrary mean for visualization. Pre-mRNA levels closely follow Pol II promoter occupancy, as expected (given the short half-lives of pre-mRNAs). Fus and Spon2 (F and H) show higher amplitude in mRNA compared to transcription; Tfrc (G) is transcribed at similar rates around the clock but shows rhythmic mRNA accumulation; Rdbp (I) shows rhythmic transcription but dampened mRNA accumulation.
See this image and copyright information in PMC

Comment in

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Bass J, Takahashi JS (2010) Circadian integration of metabolism and energetics. Science 330: 1349–1354. - PMC - PubMed
    1. Kornmann B, Schaad O, Bujard H, Takahashi J, Schibler U (2007) System-driven and oscillator-dependent circadian transcription in mice with a conditionally active liver clock. PLoS Biol 5: e34 doi:10.1371/journal.pbio.0050034. - DOI - PMC - PubMed
    1. Lamia KA, Storch KF, Weitz CJ (2008) Physiological significance of a peripheral tissue circadian clock. Proc Natl Acad Sci U S A 105: 15172–15177. - PMC - PubMed
    1. Huang W, Ramsey KM, Marcheva B, Bass J (2011) Circadian rhythms, sleep, and metabolism. J Clin Invest 121: 2133–2141. - PMC - PubMed
    1. Rudic RD, McNamara P, Curtis AM, Boston RC, Panda S, et al. (2004) BMAL1 and CLOCK, two essential components of the circadian clock, are involved in glucose homeostasis. PLoS Biol 2: e377 doi: 10.1371/journal.pbio.0020377. - DOI - PMC - PubMed

Publication types

MeSH terms

Grants and funding

This work was financed by CycliX, a grant from the Swiss SystemsX.ch (www.systemsx.ch) initiative evaluated by the Swiss National Science Foundation, Sybit, the SystemsX.ch IT unit, the University of Lausanne, the University of Geneva, the Ecole Polytechnique Fédérale de Lausanne (EPFL), and Vital-IT. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

LinkOut - more resources