Non-circadian expression masking clock-driven weak transcription rhythms in U2OS cells

Abstract

U2OS cells harbor a circadian clock but express only a few rhythmic genes in constant conditions. We identified 3040 binding sites of the circadian regulators BMAL1, CLOCK and CRY1 in the U2OS genome. Most binding sites even in promoters do not correlate with detectable rhythmic transcript levels. Luciferase fusions reveal that the circadian clock supports robust but low amplitude transcription rhythms of representative promoters. However, rhythmic transcription of these potentially clock-controlled genes is masked by non-circadian transcription that overwrites the weaker contribution of the clock in constant conditions. Our data suggest that U2OS cells harbor an intrinsically rather weak circadian oscillator. The oscillator has the potential to regulate a large number of genes. The contribution of circadian versus non-circadian transcription is dependent on the metabolic state of the cell and may determine the apparent complexity of the circadian transcriptome.

Figure 1. Genome-wide binding sites of the circadian transcription regulators BMAL1 CLOCK and CRY1.

(A) Venn diagram showing numbers and percentage of individual and overlapping binding sites of BMAL1, CLOCK, and CRY1. Percentages represent the fraction of the CRBSs over the total number of sites for the 3 proteins (B) UCSC browser views of BMAL1 (top), CLOCK (middle) and CRY1 (bottom) occupancy at the promoter of CRY2. Regions detected as binding sites of the individual transcription regulators are indicated by colored bars. Black bars indicate common binding sites of the three regulators. (C) CRBSs cluster near TSSs. Histograms of positions of all CRBSs are shown for a window of ±20 kb around TSSs with a bin size of 200 bp. 352 CRBSs are enriched ±1 kb around TSS. (D) Histogram of the distance between two consecutive CRBSs (black) in comparison to a set of 3040 random sites (red). (E) Pie chart showing the percentage of binding sites in a genomic region (black) and the contribution of the region (%) to the genome (grey). (F) Genomic annotation of the CRBSs to promoter (−1 kb to TSS), 5′UTR, exon, intron, 3′UTR, and intergenic region.

Figure 2. Heat map view of 24 h cycling genes.

Rhythmic genes with CRBSs (A) and without CRBSs (B) were ordered by phase of expression. Temperature entrained U2OS cells were released into constant conditions (37°C) and RNA expression levels were analyzed in 3 h intervals over a period of two days.

Figure 3. Temporal expression profiles of rhythmic and non-rhythmic genes.

RNA levels of U2OS cells were analyzed in 3(0–45 h) relative to the average were calculated and plotted on a log2 scale versus the circadian time (CT). For each gene up to 10 probes were spotted on the arrays. Light blue lines correspond to expression profiles based on individual probes. The black triangles and the fitted dark blue sine curve correspond to the median of the data. Light and dark areas in the background indicate subjective day and night, respectively. Examples are shown for genes that fall in various categories. (A) Clock genes with CRBSs. (B) Clock-genes without CRBSs. (C) Rhythmic genes with CRBSs. (D) Rhythmic genes without CRBSs. (E) Genes not expressed in a rhythmic fashion. AGAP11 harbors a highly enriched binding site for BMAL1, CLOCK, and CRY1, CHP has a CRY1 binding site. In both genes the CRBSs were close to the TSS. CSNK2B and KDELR1 do not have a CRBS in a window of ±20 kb.

Figure 4. mRNA expression phase and correlations with CRBSs.

Analysis of temporal expression profiles of 5708 expressed genes in two microarray replicates identified 118 common genes with circadian expression rhythms. Data from array A is shown. (A) The amplitudes of expressed genes (n = 5708) were plotted versus the circadian phase. The 118 rhythmic genes are indicated by black symbols. The shade of blue corresponds to the rhythmicity of a gene (light blue = low 24 h-rhythmicity and/or phase undefined; dark blue = high 24 h-rhythmicity and highly reliable phase). (B) The amplitudes of the 118 rhythmic genes are plotted against the phase. Genes with CRBSs are shown with black diamonds, the other rhythmic genes are displayed with gray diamonds. Core circadian clock genes are indicated. (C) Phase distribution of rhythmic genes with and without CRBSs.

Figure 5. Expression analysis of genes with CRBSs in their promoters.

(A) UCSC browser views of BMAL1 (top), CLOCK (middle) and CRY1 (bottom) occupancy at the promoters of ATG3, EIF5A2, and SCN5A. Binding sites of the individual transcription regulators are indicated by colored bars. Black bars indicate common binding sites of the three regulators. (B) Temporal expression profiles (microarray analysis) of ATG3, EIF5A2, and SCN5A, which have strong CRBSs in their promoters. (C, D) Around the clock qRT-PCR analysis of preRNA (C) and mRNA (D) levels of ATG3, EIF5A2, and SCN5A. qRT-PCR analysis was carried out with intron- and exon-specific probes, respectively, of ATG3, EIF5A2, and SCN5A using RNA of time courses A and B. Expression levels of HPRT1 were used for normalization.

Figure 6. Rhythmic transcription of genes with CRBSs in their promoter is masked by high baseline expression.

Robust low amplitude transcription rhythms revealed by long-term luminescence measurements of stable U2OS cell lines expressing the destabilized luciferase2P (luc2P) under control of the promoters of ATG3, EIF5A2, and SCN5A. BMAL1-luc, PER2-luc2, and GAPDH-luc2P cell lines are shown for control. Luminescence of synchronized cultures was measured at 30 min intervals. Raw expression data of the indicated reporter genes for day 1 to day 5 after synchronization are shown in the left panels. The middle panels show the 24 h moving average of the luciferase activity to estimate the non-rhythmic contribution and the right panels show the de-trended data to estimate the rhythmic contribution. The black and grey traces show expression profiles in the absence and presence of 200 nM rapamycin, respectively.