Preservation of ∼12-h ultradian rhythms of gene expression of mRNA and protein metabolism in the absence of canonical circadian clock

Abstract

Introduction: Besides the ∼24-h circadian rhythms, ∼12-h ultradian rhythms of gene expression, metabolism and behaviors exist in animals ranging from crustaceans to mammals. Three major hypotheses were proposed on the origin and mechanisms of regulation of ∼12-h rhythms, namely, that they are not cell-autonomous and controlled by a combination of the circadian clock and environmental cues, that they are regulated by two anti-phase circadian transcription factors in a cell autonomous manner, or that they are established by a cell-autonomous ∼12-h oscillator. Methods: To distinguish among these possibilities, we performed a post hoc analysis of two high temporal resolution transcriptome dataset in animals and cells lacking the canonical circadian clock. Results: In both the liver of BMAL1 knockout mice and Drosophila S2 cells, we observed robust and prevalent ∼12-h rhythms of gene expression enriched in fundamental processes of mRNA and protein metabolism that show large convergence with those identified in wild-type mice liver. Bioinformatics analysis further predicted ELF1 and ATF6B as putative transcription factors regulating the ∼12-h rhythms of gene expression independently of the circadian clock in both fly and mice. Discussion: These findings provide additional evidence to support the existence of an evolutionarily conserved 12-h oscillator that controls ∼12-h rhythms of gene expression of protein and mRNA metabolism in multiple species.

Keywords: Drosophila S2 cell; X-box binding protein 1 (XBP1); mRNA metabolism; proteostasis; ultradian and circadian rhythms.

FIGURE 1

Prevalent ∼12-h rhythms of gene expression in the liver of BMAL1 knockout mice. (A) Distribution of periods uncovered from all oscillations by the eigenvalue/pencil method. (B) Heatmap of ∼12-h rhythms of gene expression uncovered by the eigenvalue/pencil method. (C) Number of genes cycling with different periods uncovered by the RAIN method, with different FDR cut-off. (D) Heatmap of ∼12-h rhythms of gene expression ranked by different FDR cut-off. (E) Venn diagram comparing ∼12-h transcriptome uncovered by the eigenvalue/pencil and RAIN (FDR cut-off of 0.15) method. p-value calculated by Chi-square test. (F) GO analysis showing enriched biological pathways of ∼12-h genes revealed by different methods using all mice genes as background.

FIGURE 2

Preservation of ∼12-h gene program and functionality between wild-type and BMAL1 knockout mice. (A) Heatmap of 1,162 commonly found ∼12-h rhythm gene expression in wild-type, BMAL1 knockout and XBP1 LKO mice. (B) Venn diagram comparing ∼12-h rhythm genes in wild-type and BMAL1 knockout mice. p-value is calculated by Chi-square test. (C) Relative amplitude of ∼12-h oscillations for 1,162 genes in wild-type and BMAL1 knockout mice. (D) Heatmap of 667 commonly found ∼12-h rhythm gene expression in wild-type, BMAL1 knockout and XBP1 LKO mice (FDR cut-off of 0.15). (E) GO analysis showing enriched biological pathways of ∼12-h genes found in wild-type or in both wild-type and BMAL1 knockout mice using all mice genes as background.

FIGURE 3

Many hepatic genes are under dual circadian clock and 12-h oscillator control. (A) Heatmap of 2,450 commonly found circadian gene expression (common between wild-type and XBP1 LKO mice) in wild-type, BMAL1 knockout and XBP1 LKO mice. (B) GO analysis showing enriched biological pathways of common ∼12-h (between wild-type and BMAL1 knockout mice) and circadian (between wild-type and XBP1 LKO mice) using hepatically expressed genes as background. (C) Venn diagram illustrating a total of 141 genes under robust dual clock control. (D) Eigenvalue/pencil deconvolution of Cct3 and Tomm40l temporal gene expression profiles in wild-type, BMAL1 knockout and XBP1 LKO mice. Gray line in each graph illustrates simulated temporal gene expression profile by addition of all superimposed oscillations.

FIGURE 4

ELF1, KLF7 and ATF6B are putative transcription regulators of 12-h oscillator. (A) Motif analysis of promoter regions of common ∼12-h rhythm genes identified in wild-type and BMAL1 knockout mice by either method. (B,C) Expression of different transcription factors in wild-type, BMAL1 knockout and XBP1 LKO mice. Period identified by the eigenvalue/pencil method, and q value (FDR) for having 12-h rhythms of gene expression via the RAIN method were shown for each gene in each genotype.

FIGURE 5

∼12-h rhythms of gene expression are present in S2 cells. (A) Distribution of periods uncovered from all oscillations by the eigenvalue/pencil method. (B) Heatmap of ∼12-h (top) and circadian rhythms (bottom) of gene expression uncovered by the eigenvalue/pencil method. (C) Number of genes cycling with different periods uncovered by the RAIN method, by p-value or FDR cut-off. (D) Heatmap of ∼12-h rhythms of gene expression with FDR cut-off of 0.15 via RAIN. (E) GO analysis of ∼12-h and circadian rhythms uncovered by different methods/FDR cut-off in S2 cells using all fly genes as background. (F) Representative expression of selective genes. Period identified by the eigenvalue/pencil method, and q value (FDR) for having ∼12-h rhythms of gene expression via the RAIN method were shown for each gene.

FIGURE 6

∼12-h rhythms of transcription regulators in S2 cell. Representative expression of selective transcription regulators in S2 cell. Period identified by the eigenvalue/pencil method, and q value (FDR) for having ∼12-h rhythms of gene expression via the RAIN method were shown for each gene.