The circadian coordination of cell biology

The molecular clockwork of circadian rhythms: Different mechanisms to create 24-h oscillations

Common properties of different clock systems

Circadian oscillators are self-sustained, with an ∼24-h rhythm under constant conditions, yet possess plasticity to integrate intracellular and environmental changes to temporally regulate a diverse set of cellular functions and optimize adaptation (see text box). Entrainment by external cues (called zeitgebers) such as light, food, or temperature allow clock-controlled processes to be tuned to actual environmental conditions. Circadian oscillators are also temperature compensated, allowing them to function with ∼24-h periodicity within a wide range of temperatures. Despite their common properties, clock mechanisms across the phylogenetic kingdoms use different modalities to generate ∼24-h oscillations (Fig. 1).

Open in a separate window

Figure 1.

Molecular oscillators that generate circadian rhythms in multiple organisms. Circadian rhythms are generated by autonomous molecular oscillators that cycle between different activation states within a period of ∼24 h. (A) In cyanobacteria, the KaiABC complex, made of KaiA, KaiB, and a hexamer of KaiC (green), constitute the molecular oscillator. The complex transitions between a phosphorylated and unphosphorylated state within a period of 24 h. (B) In fungi, plants, and animals, the circadian oscillator is based on a TTFL. In mammals, the TTFL involves the activation of Per, Cry, Rev-Erb, and Ror transcription by the CLOCK/BMAL1 heterodimer. Upon translation, PER and CRY proteins form a complex that is imported in the nucleus and suppresses CLOCK/BMAL transcriptional activity. ROR and REV-ERB proteins can activate or repress the transcription of BMAL1, respectively. ROR/REV-ERB also act on the transcription of other clock components and refine their phase of expression. The alternating waves of activation and repression, coupled with the short t_1/2 of their respective mRNA and proteins, generate circadian oscillations of clock components. These circadian transcriptional regulators also act on other genes to produce transcriptional oscillation. (C) In bacteria, archaea, and eukaryotes, antioxidant enzymes called PRX cycle between the oxidized (S-S disulfide bond) and reduced (SH thiol bond) state; this cycle can function as a circadian oscillator (see main text for details). CCG, clock-controlled gene.

Diurnal and circadian rhythms

Events and processes that repeat themselves within a period of ∼24 h are called diurnal rhythms. If they persist in constant condition (i.e., without a signal from the environment), they are called circadian rhythms. Circadian rhythms are born from the activity of an internal biological clock, or “body clock.” The signals that provide environmental cues that can be integrated by the circadian clock are called zeitgebers.

Clock and oscillator

A circadian oscillator is a biochemical entity (a protein or a network of proteins) that oscillates between different states within a period of 24 h. In a cell, these oscillations are self-sustained and cell autonomous, which constitute fundamental properties that characterize circadian rhythms. The activity of the cellular oscillators is orchestrated by the master clock, a pacemaker located in the SCN in the hypothalamus.

Period, amplitude, and phase

When looking at the circadian oscillation of a given process, the mathematical parameters that characterize a cosine wave can be defined. Hence, the period of the oscillator is the exact length of time from one peak of activity to the next, the amplitude is half the height from the lowest to the highest point of activity, and the phase is the time of the peak using the time at which the light turns on as the reference time (circadian time 0).

Gating and phase-locking

When multiple oscillators are present in a cell (e.g., the circadian oscillator and cell cycle), they can interact in various ways. When the activity of one oscillator is gated by another oscillator, certain outputs from the first oscillator can only happen at specific phases (or gates) of the activity of the second. Two phase-locked oscillators oscillate at the same frequency with a constant relationship between their phases.

Transcriptional regulation: The proximal output of the circadian clock

In cyanobacteria, the KaiABC oscillator regulates global changes in chromosomes topology, which in turn drives the rhythmic expression of most genes (Woelfle and Johnson, 2006). Similarly, in mammals, up to 50% of the genome shows circadian modulation (Panda et al., 2002; Storch et al., 2002; Fig. 2), at least in one organ, even though the percentage of oscillating RNA transcripts varies between tissues (from 16% in the liver to 3% in the hypothalamus; Zhang et al., 2014). But only 10 mouse genes, encoding the oscillator components or their direct targets, oscillate in all organs, thus suggesting circadian clock components interact with tissue specific factors to drive locus-specific circadian gene expression (Zhang et al., 2014).

Open in a separate window

Figure 2.

Many steps of gene expression have a circadian rhythm of activity. “Omics” technologies, primarily in mouse hepatocytes, show that circadian rhythms are involved in almost all of the steps of gene expression, from transcription to posttranslational modifications of proteins. Between 3 and 16% of the transcriptome displays circadian rhythms. Several mechanisms have been involved in this regulation. The recruitment of clock transcription factors (TFs) associated with tissue-specific transcription factors is circadian. As a result, the activity of the RNA polymerase II (RNA Pol II) is circadian. The architecture of the chromatin as well as histone modifications also show circadian rhythms. The sirtuin family of deacetylases is involved in this regulation. Even in absence of rhythmic transcription, some mRNA oscillate daily owing to general circadian rhythms in mRNA processing. Splicing factor expression and splicing activity varies with the time of day. PolyA tail editing (as by nocturnin) or binding to RBPs affecting mRNA stability as well as modifications by noncoding RNAs also are circadian. Translation also peaks at a certain time of the day. In hepatocytes, this happens in sync with feeding, when the cellular energy level is high. Cyclical changes in both the activation of the translation initiation complex and biogenesis of ribosome contributes to circadian translation. Finally, at the proteome level, posttranslational modifications like phosphorylation (P), poly-ADP-ribosylation (R), and glucose-N-acetylation (O-GlcNAc) oscillate daily. In hepatocytes, they are tuned to cellular energetics because the donors for these posttranslational modifications are directly affected by metabolic activity. asRNA, antisense RNA; lincRNA, long intergenic noncoding RNA.

Owing to its ease of access and relevance in metabolism, the mouse liver is used extensively to study circadian gene expression at the cistrome (the cis-acting DNA elements bound by transcription factors), DNA methylome, chromatin modifications (via chromatin immunoprecipitation sequencing), and transcriptome levels. Several studies (DiTacchio et al., 2011; Koike et al., 2012; Le Martelot et al., 2012) revealed daily chromatin level rhythms, along with rhythms in promoter occupancy of circadian transcription factors RNA polymerase II, CREB-binding protein, and P300. This suggests that circadian clock components closely interact with histone modification enzymes and with the transcription machinery and that these interactions drive daily rhythms in transcription of their target genes. In addition, these studies provide insight into the complexity of circadian regulation, some of which are discussed in the next paragraph.

The cistromes of each oscillator components are overlapping, but not identical. In addition to shared targets, individual clock components, by their interaction with other transcription factors, can exert circadian regulation of different sets of targets to cater to tissue-specific or metabolic needs. For example, CRY proteins repress CLOCK/BMAL1 activity at E-boxes, and they also repress the transcriptional activity of glucocorticoid receptors at promoters harboring glucocorticoid response elements (Lamia et al., 2011). Similarly, REV-ERB proteins compete for binding to ROR response elements to drive rhythmic expression of clock components while they are tethered to the promoters of metabolic genes through interaction with the histone deacetylase–nuclear receptor corepressor complex (Yin and Lazar, 2005). Even the core clock activators CLOCK/BMAL1 interact with SIRT1 and SIRT6 in a locus-specific manner to regulate the expression of different sets of metabolically relevant genes (Masri et al., 2014).

Circadian rhythms in chromatin modification suggest interactions between the clock components and enzymes that write, read, or erase chromatin modification marks. Accordingly, histone lysine methylases and demethylases and histone acetyltransferases and deacetylases interact with the circadian clock components (Yin and Lazar, 2005; DiTacchio et al., 2011; Masri et al., 2014) and participate in circadian chromatin modifications. These chromatin-modifying enzymes use cofactors or substrates including S-adenosyl methionine, α-ketoglutarate, acetyl coenzyme A, and NAD⁺, whose concentration in the liver shows daily fluctuations and is also affected by feeding or fasting (Eckel-Mahan et al., 2012; Hatori et al., 2012; Peek et al., 2013).

mRNA processing and its relationship with the circadian clock

A comparison of chromatin marks of transcription, nascent RNA, and processed mRNA has revealed that numerous transcripts are not transcribed in a circadian manner; instead, their mature mRNAs show circadian rhythmicity (Koike et al., 2012; Atger et al., 2015). This is because the circadian clock in mammals rhythmically regulates RNA processing and degradation (Kojima et al., 2011). This renewed interest in posttranscriptional mRNA regulation has led to an examination of the roles of antisense RNAs, miRNA, RNA splicing factors, RNA-binding proteins (RBP), and cis-acting elements in modulating the circadian transcriptome (Kojima et al., 2011; Lim and Allada, 2013; Fig. 2).

In the mouse liver, splicing displays a diurnal rhythm and, in some cases, can be responsible for the cycling of mature mRNA (McGlincy et al., 2012). In fact, many, splicing factors themselves display a circadian rhythm (McGlincy et al., 2012). One of the well-characterized RBPs, nocturnin, is a Poly(A) tail shortening deadenylase, which is under circadian control in mice (Wang et al., 2001). Conversely, the transcript levels of some of the clock genes are modulated by RBPs. Circadian clock and feeding/fasting rhythms drive the daily cycle of core body temperature. Simulated body temperature rhythm can drive daily rhythms in cold-inducible RBP, which confers robustness to CLOCK expression by binding and stabilizing CLOCK RNA (Morf et al., 2012). Several dozen miRNAs in the mouse liver show circadian rhythms in abundance (Vollmers et al., 2012). These oscillating miRNAs can potentially drive rhythms in degradation of target RNAs. Accordingly, in liver-specific DICER knockout mice, miRNAs are largely undetectable (Chen et al., 2013). In the absence of their potential action on RNA degradation, the time of peak level of several mRNAs and the peak or trough levels (amplitude) of some rhythmic transcripts are affected, thus indicating a role of miRNA in fine-tuning circadian rhythms (Du et al., 2014). Similarly, recently discovered antisense RNAs of clock components Per2, Bmal1, and Rev-ErbA, or antisense RNAs of several other rhythmic mRNAs (Vollmers et al., 2012; Zhang et al., 2014) likely impose additional layers of regulation of rhythmic gene expression.

Translation: Another layer of circadian control

Does the circadian transcriptome give rise to circadian proteome? Unbiased qualitative as well as quantitative analyses of mouse liver proteome have revealed that up to 20% of soluble proteins have diurnal patterns of abundance (Reddy et al., 2006; Mauvoisin et al., 2014; Robles et al., 2014). Although initial comparison of circadian transcriptome and proteome from the whole liver found limited overlap, recent studies using subcellular fractionation of the mouse liver and quantitative proteomics of the purified mitochondria has revealed much higher correlation between cycling transcripts and proteins that encode mitochondria components (Neufeld-Cohen et al., 2016). However, this study also reconfirmed earlier observations that many cycling proteins do not have a concomitant oscillation of their mRNA. Diurnal rhythm in overall ribosome biogenesis (Jouffe et al., 2013) and activation of the translation initiation complex (Jouffe et al., 2013) or in preferential recruitment of certain mRNAs to ribosomes and translation (Atger et al., 2015; Jang et al., 2015; Janich et al., 2015) have been observed in the mouse liver and may explain cycling protein level in the absence of cycling mRNAs (Fig. 2).

Circadian rhythmicity in the nucleus

Primary functions of the nucleus, including transcription and DNA repair, show circadian rhythms (Fig. 3, blue panels). In concert with thousands of loci undergoing circadian transcription in the liver (Zhang et al., 2014), the spatial organization of the chromosomes show a circadian rhythm (Aguilar-Arnal et al., 2013). Additionally, the nuclear envelope (NE) regulates the spatial organization of the genome and transcription by interacting with transcription factors and chromatin-modifying enzymes (Stancheva and Schirmer, 2014). The disruption of nuclear lamina proteins LMNB1, LBR, and MAN1 affects the period of CLOCK protein oscillations in both Drosophila cells and human U2OS cells, and MAN1 binds to Bmal1 promoter. This suggests that NE proteins play a role in the circadian control of transcription (Lin et al., 2014).

The nucleus hosts various DNA repair mechanisms. Nucleotide excision repair (NER) responds to single-strand DNA damage and UV-induced DNA adducts. Both the activity and efficiency of this repair pathway exhibits circadian rhythmicity (Kang et al., 2009), as do the protein levels of a key coordinator of NER, xeroderma pigmentosum A (Kang et al., 2010; Bee et al., 2015). Three different repair pathways target DNA double-stranded breaks. All three initially recognize such breaks via the kinase ataxia telangiectasia mutated, which coimmunoprecipitates with the clock protein PER1 (Gery et al., 2006). Per1 expression levels also affect cell sensitivity to DNA-damaging agents (Gery et al., 2006). Thus, the circadian clock influences transcription, the DNA damage response, and the general architecture of the nucleus.

Circadian rhythmicity in the ER

The ER is the first organelle in the secretory pathway that properly sorts and folds ∼40% of the proteome that transits through this pathway (Fig. 3, red panels; Uhlén et al., 2015). The liver has various secretory functions, and hepatic protein secretion has diurnal oscillations (Reddy et al., 2006; Mauvoisin et al., 2014; Robles et al., 2014). The protein components of the secretory pathway, including ribosome docking, ER chaperones, and key components of both ER and Golgi vesicle formation and trafficking, display circadian oscillations that are in sync with liver secretion (Robles et al., 2014). In addition, proper protein folding requires the maintenance of an appropriate redox state in the ER, especially for disulfide bond formation. Circadianly expressed PRX4 plays a central role in hydrogen peroxides detoxification and oxidative protein folding (Zhu et al., 2014).

When protein-folding demand exceeds folding capacity, the unfolded protein response (UPR) is activated in the ER to prevent proteotoxicity. Proteins involved in the ER stress response include sensors (ATF6 and the endoribonuclease IRE1a) and effectors (XBP1, ATF4, and ATF6; Oakes and Papa, 2015). ATF4 is a transcriptional target of CLOCK (Koyanagi et al., 2011), and XBP1 splicing by IRE1a follows a 12-h rhythm that is coordinated by the clock machinery (Cretenet et al., 2010). The UPR itself is linked to the circadian clock. For instance, in vitro, Per1 mRNA is a target of the endoribonuclease IRE1a (Pluquet et al., 2013), and ATF4 is required for the circadian transcription of the Per2 gene in mouse embryonic fibroblasts (Koyanagi et al., 2011). Temporal coordination of protein folding, unfolding response, and secretion by the circadian clock likely reduces proteotoxicity.

Circadian rhythmicity in the mitochondria

Several nutrient and energy metabolic pathways are under circadian regulation, and the redox state impacts the circadian clock (Fig. 3, green panels). Mitochondria, being the hubs for both energy metabolism and redox regulation, show reciprocal regulation with the circadian clock. Mitochondria oxidative capacities display a diurnal oscillation (Bray et al., 2008; Peek et al., 2013). This diurnal rhythm of mitochondria function likely results from direct transcriptional regulation of a large number of nuclear-encoded mitochondrial genes by the core oscillator components (Koike et al., 2012; Jacobi et al., 2015). The expression of these mitochondrial genes is well correlated with daily rhythms in the respective mitochondria proteins (Neufeld-Cohen et al., 2016). Even at metabolite level, the rhythmic production of NAD⁺, which results from the rhythmic transcription of the rate-limiting enzyme through the salvage pathway (nicotinamide phosphoribosyltransferase), couples the circadian clock to mitochondrial oxidative capacity. The activity of key fatty acid oxidation (FAO) enzymes is controlled by acetylation. The deacetylase SIRT3 can modulate the activity of these enzymes, thus driving a circadian rhythm in their activity (Peek et al., 2013).

Mitochondria are also dynamic organelles that undergo fusion, fission, mitophagy, and biogenesis (Chan, 2012) to maintain their functional homeostasis. Diurnal changes in mitochondrial architecture, observed by EM and fluorescent markers, are abolished in Bmal1-deficient mouse hepatocytes (Jacobi et al., 2015). Some of these structural changes may be directly driven by transcriptional regulation of genes involved in mitochondrial fission and mitophagy, with higher expression occurring in the fed state (Jacobi et al., 2015). In the absence of Bmal1, mitochondria are defective and abnormally swollen (Jacobi et al., 2015).

The mitochondrial responses to oxidative stress and hypoxia are also connected to the circadian clock. Indeed, hypoxia-inducible factor 1α expression, a key player in the hypoxic response, is diurnal, and cells’ sensitivity to oxidative stress is time dependent (Magnone et al., 2015).

Damaged mitochondria can be removed by mitophagy, and the cellular mitochondrial stock can be replenished through mitochondrial biogenesis. Peroxisome proliferator–activated receptor α coactivator 1α (PGC1α) is the master regulator of mitochondrial biogenesis and transcribed in a circadian manner in many metabolic organs. In heart-specific Bmal1-deficient mice, PGC1α expression is reduced in myocytes. As a result, mitochondrial number and morphology are altered, and these animals develop congestive heart failure (Kohsaka et al., 2014). Moreover, PGC1α controls Bmal1 expression, and PGC1α-deficient animals have a disrupted circadian profile of locomotor activity (Liu et al., 2007). Thus, there is extensive bidirectional coupling between the circadian clock and mitochondrial dynamics and function.

Bidirectional coupling between the cell cycle and circadian clock

During cell-cycle checkpoints, control mechanisms screen for DNA damage and, if necessary, initiate specific DNA-damage repair responses before the cell cycle can resume. In the previous section, we discussed the interplay between DNA repair activation and the circadian clock. In this study, we focus on connections between cell division and the circadian clock. In unicellular organisms that can divide more than once in 24 h, such as the cyanobacteria Synechococcus elongates, division only occurs at a certain phase of the circadian clock cycle. This phenomenon is called circadian “gating” (Mori et al., 1996). The number of gates varies with the speed of cell division (Yang et al., 2010). Another way that the circadian clock can influence cell division is by phase-locking. Two oscillating systems are phase-locked when they oscillate at the same frequency and with a constant relationship between their phases. For example, in single-cell cultures of NIH3T3, the cell cycle oscillations and the circadian clock oscillations are phase-locked. In unsynchronized cells, the cell cycle and the circadian clock are phase-locked in a 1:1 ratio of one cell division per 24 h (Bieler et al., 2014; Feillet et al., 2014). However, when the circadian clocks of the cells are synchronized with dexamethasone, subsets of cells display different ratios of phase-locking (Feillet et al., 2014). These results suggest that the coupling between the clock and the cell cycle is flexible and can adapt to environmental changes.

In vivo, the interaction between the cell cycle and the clock has been illustrated in regenerative tissues in various organisms. The clock is required for intestinal homeostasis in flies after chemical damage (Karpowicz et al., 2013) and in rodents (Mukherji et al., 2013). In rodents, the circadian clock gates the division of epidermal progenitor cells at different stages of the regenerative cycle (Lin et al., 2009; Plikus et al., 2013).

Whether in a gated or phased-locked model, the cell cycle and the circadian clock interact, possibly through bidirectional regulation (Feillet et al., 2014). Various molecular mechanisms underlie this coupling. The circadian clock controls the transcription of several key cell cycle checkpoint regulators, such as the cyclin-dependent kinase inhibitors p21 by Rev-Erb (Gréchez-Cassiau et al., 2008), p16 by Nono and Per (Kowalska et al., 2013), and the activity of the mitosis checkpoint kinase CDK1 by the clock-associated kinase Wee1 (Matsuo et al., 2003). The cell cycle might also reciprocally control the circadian clock. The cell cycle regulator p53 can modulate Per2 expression by blocking the access of the BMAL1–CLOCK complex to its promoter (Miki et al., 2013).

The relationship between the circadian clock and senescence could be explained by the interaction between clock proteins and the mediators of senescence, namely p53 and RB-p16^INK4A pathways (Kowalska et al., 2013; Miki et al., 2013). For instance, reactive oxygen species (ROS)–induced senescence is increased in Bmal1 knockout fibroblasts (Khapre et al., 2011).

The biological significance of the interaction between the circadian clock and the cell cycles likely reflects the necessity to gate the cell cycle to specific times of the day. In cyanobacteria, the presence of a circadian clock system oscillating in sync with the light/dark cycle during the growth phase confers a survival advantage (Ouyang et al., 1998). Current hypotheses include the timing of DNA replication away from DNA-damaging agents such as UV from solar radiation or more likely away from ROS produced by metabolic activity (Geyfman et al., 2012; Stringari et al., 2015). In mammals, the physiological and evolutionary relevance of this coupling is likely to be more complicated and very difficult to test.

This connection between the cell cycle and the circadian clock in dividing cells raises questions about the nature of this interaction in stem cells, differentiating cells, terminally differentiated nondividing cells, and senescence.

Is clock silencing a property of stemness?

Mice that lack any circadian clock components are grossly normal at birth (van der Horst et al., 1999; Kondratov et al., 2006), indicating that the circadian system is dispensable for early differentiation. Pluripotent embryonic stem cells do not exhibit circadian oscillation in vitro (Yagita et al., 2010). As embryonic stem cells begin to differentiate, circadian oscillations in transcription appear. When differentiated cells are reprogrammed to induced pluripotent stem, circadian oscillations disappear (Yagita et al., 2010). In vivo evidence from mice suggests that independent cellular clock systems develop during embryogenesis but that a fully mature and synchronized circadian clock only appears after birth (Sládek et al., 2007; Dolatshad et al., 2010). If circadian oscillations are silenced in stem cells, it could reflect that certain factors required to maintain stemness might also functionally dampen the clock. This remains to be shown, but could shed light on the role of the circadian clock in stemness and differentiation.

The circadian clock is also required for the differentiation of some cell populations. For instance, in vivo studies in mice suggest that REV-ERB and RORγ are necessary for the lineage specification of T helper 17 cells (Yu et al., 2013). Bmal1 regulates adipocyte differentiation (Guo et al., 2012), and the adipocyte-specific deletion of Bmal1 leads to obesity in mice (Paschos et al., 2012). In mice, the colony-forming ability of bone marrow granulomonocytes varies with the time of day and is sustained in vitro (Bourin et al., 2002). The rhythm of bone marrow cell cycle and differentiation is sustained in suprachiasmatic nucleus (SCN)–ablated animals (Filipski et al., 2004), suggesting that a cell-autonomous circadian clock controls the proliferation and differentiation capabilities of bone marrow progenitors.

Research in S. Panda’s laboratory is supported by National Institutes of Health grants DK091618 and EY016807, the Leona M. and Harry B. Helmsley Charitable Trust grant 2012-PG-MED002, the Glenn Foundation for Medical Research, and the American Federation for Aging Research grant M14322. A. Chaix is supported by an American Diabetes Association mentor-based Postdoctoral Fellowship (7–12-MN-64). A. Zarrinpar received support from National Institutes of Health grant K08 DK102902 and the American Association for the Study of Liver Diseases Emerging Liver Scholars Award.

The authors declare no competing financial interests.