As the day progresses, the strength of the brain’s global signal fluctuation shows an unexpected decrease, according to a study published on February 18 in the open-access journal PLOS Biology by Csaba Orban and a multi-disciplinary team of scientists from the Faculty of Engineering, Yong Loo Lin School of Medicine and N.1 Institute of Health at the National University of Singapore.
Circadian rhythms govern diverse aspects of physiology including sleep/wake cycles, cognition, gene expression, temperature regulation, and endocrine signaling.
But despite the clear influence of circadian rhythms on physiology, most studies of brain function do not report or consider the impact of time of day on their findings.
To address this gap in knowledge, the team analysed functional magnetic resonance imaging (fMRI) data of approximately 900 subjects who were scanned between 8 am and 10 pm on two different days as part of the Human Connectome Project (HCP; http://www.humanconnectomeproject.org/).
Multiple studies have shown that the brain’s global signal fluctuates more strongly when one is drowsy (e.g. after insufficient sleep), and fluctuates less when one is more alert (e.g. after coffee).
Based on known circadian variation in sleepiness, the authors hypothesized that global signal fluctuation would be lowest in the morning, increase in the mid-afternoon and dip in the early evening.
Instead, they observed a cumulative decrease in global signal fluctuation as the day progressed.
This global decrease was most prominent in visual and somatosensory brain regions, which are known for expressing dynamic fluctuations within individuals over time.
Across the whole brain, time of day was also associated with marked decreases in resting-state functional connectivity – the correlated activity between different brain regions when no explicit task is being performed.
“We were surprised by the size of the overall time-of-day effects, since the global fMRI signal is affected by many factors and there is substantial variation across individuals.
At the present moment we don’t have a good explanation of the directionality of our findings. However, the fact that we also observed slight time-of-day-associated variation in the breathing patterns of participants suggests that we may also need to consider clues outside of the brain to fully understand these effects,” said Csaba Orban, first author of the study.
The negative association between time of day and brain signal fluctuations was strongest in visual and somatosensory regions. Image is credited to Csaba Orban.
Based on the findings, the authors recommend that researchers explicitly report the time of day of fMRI scans and other experimental protocols and measurements, as this could help account for between-study variation in results and potentially even failure to replicate findings.
“We hope these findings will motivate fellow neuroscientists to give more consideration to potential effects of time of day on measures of brain activity, especially in other large-scale studies where subjects are often scanned throughout the day for logistical reasons,” said Thomas Yeo, the study’s senior author.
Circadian rhythms (CR) are endogenous autonomous oscillators of physiological activities resulting 24-h day/night cycles, which allow organisms to adapt to a fluctuating environment (Reppert and Weaver, 2002; Dibner et al., 2010).
The core pacemaker of CR lies in the SCN, which plays crucial roles in maintenance of systemic CR and regulates peripheral tissue clocks through secretion of endogenous regulatory factors (Dibner et al., 2010).
The molecular clock of the CR system, which is present in all cells, is made up of oscillating clock-related proteins that compose TTFLs (Gekakis et al., 1998). The core TTFL is composed of the transcriptional activator proteins CLOCK and BMAL1, and the repressor proteins Period-1 (PER1), PER2, PER3, Cryptochrome-1 (CRY1) and CRY2 (Gekakis et al., 1998).
Other loops are coupled to the core TTFL to maintain oscillation. The first sub-loop is composed of RORs and nuclear REV-ERB receptors. The second sub-loop comprises D-box-related genes, which include DBP, TEF, and HLF (Figure 1; Preitner et al., 2002; Sato et al., 2004). Moreover, recent studies have suggested that circadian regulation is heavily involved in gene expression. A considerable portion (approximately 10%) of genes expressed in cells or tissues have been found to display circadian oscillations, resulting in identification of these genes as “CCGs” (Duffield, 2003; Zhang et al., 2014).
Mammalian tissues and cells have an autonomous circadian oscillator with a period of roughly 24 h. External stimuli are essential for maintaining the appropriate circadian oscillations (Bass and Takahashi, 2010; Roenneberg and Merrow, 2016; Tahara et al., 2017).
In vivo, CR is mainly entrained by environmental signals such as light, food, and arousal stimuli. In the SCN, the circadian clock mainly responds to the LD cycle. In peripheral tissues, CR can be synchronized by food or temperature (Damiola et al., 2000; Hara et al., 2001; Buhr et al., 2010).
Moreover, internal signals such as circulating hormones, cytokines, metabolites, sympathetic nervous activation, and body temperature are significant timing cues that regulate peripheral clocks (Albrecht, 2012; Menaker et al., 2013). In vitro, CR are difficult to observe due to lack of SCN signals. As such, external stimuli should be applied to induce CR in cultured cells or explants (Figure 2).
The field of chronobiology requires detection of features of CR in tissues or cells. Through chronological collection or luminescence monitoring, the basic data resulting from circadian fluctuations in vivo or in vitro can be accurately recorded.
Approaches such as FFT (Moore et al., 2014), JTK-Cycle (Hughes et al., 2010), or autocorrelation (Levine et al., 2002) can be used to analyze the rhythmic features of CR. Subsequently, we introduced several circadian databases established in recent years.
Finally, we summarized in vivo and in vitro factors that can alter CR. In vivo factors include entrainments such as light, food, and temperature, while the in vitro factors include cell density (Noguchi et al., 2013), osmotic pressure, PH, mechanical stimulus, temperature, oxygen concentration, and microorganisms (Haspel et al., 2014). These factors can lead to circadian disruptions, and result in various diseases such as cancer, dysplasia, and metabolic and behavioral disorders (Takahashi et al., 2008).
Circadian Rhythm Synchronizers
Nearly every eukaryote has a cell-autonomous circadian clock which exhibits 24-h physiological oscillations and can be influenced by external timing cues.
These external timing cues, also called “synchronizers,” “zeitgeber,” or ”entraining agents” can reset the body’s circadian clock and place all cells at the same phase of circadian oscillation, a process called circadian rhythm synchronization.
The word “ZT” is used to describe external cues that synchronize CR such as LD cycles or temperature cycles, and the word “CT” is utilized to describe timing without external signals (Li et al., 2017). In the field of chronobiology research, whether in vivo or in vitro, researchers use these entraining agents to synchronize CR of the experimental animals or cells. A variety of factors which act as synchronizers are summarized in Figure 2.
Most organisms acquire the time information through changing light intensity throughout the day to reset their own clock, referred to as “photic entrainment” (Roenneberg and Foster, 1997). In chronobiology experiments, light is often served as a stimulus to induce responses from the circadian clock. The light signaling cascades that entrain the circadian clock are fairly complex.
In vivo, pRGCs receive light information, which is transmitted to the SCN directly through the RHT (Hattar et al., 2002; Ishida et al., 2005; Kalsbeek et al., 2006; van Diepen et al., 2015; Pilorz et al., 2016; Mouland et al., 2017; Astiz et al., 2019), resulting in regulation of the peripheral clocks by the SCN via secreting neurohumoral factors (Mohawk et al., 2012). The light-sensitive photopigment melanopsin (Provencio et al., 2000), is expressed in the pRGCs which mediate a series of responses to light (Hankins et al., 2008; Do and Yau, 2010; Schmidt et al., 2011).
Stimulation of melanopsin causes activation of G protein coupled signaling cascades and the PLC pathway, leading to opening of TRPCs such as TRPC6 and TRPC7, Ca2+ influx, and cellular depolarization (Poletini et al., 2015).
The input pathway to the SCN mainly depends on the monosynaptic RHT. The primary neurotransmitters in the RHT are glutamate, PACAP, SP, and aspartate (Ebling, 1996; Chen et al., 1999; Hannibal, 2002; Golombek et al., 2003; Hannibal and Fahrenkrug, 2004; Fahrenkrug, 2006).
Glutamate, which binds to glutamatergic receptors such as NMDA or AMPA receptors, is the main signal for photic entrainment (Mikkelsen et al., 1993; Golombek and Ralph, 1996; Chambille, 1999; Peytevin et al., 2000; Michel and Colwell, 2001; Paul et al., 2005), and induces increased intracellular Ca2+ concentrations (Kim et al., 2005). PACAP, which activates PAC1 receptor, is also responsible for mediating synchronization to light (Hannibal et al., 1998, 2008; Shen et al., 2000; Colwell et al., 2004; Hashimoto et al., 2006). Ca2+ influx activates a range of kinases including PKA, MAPK, CaMK, PKCA, and PKG (Meijer and Schwartz, 2003).
These kinases are involved in phosphorylation of CREB, which binds to cAMP response elements in promoters, resulting in transcription of clock genes such as PER1 and PER2 (Meijer and Schwartz, 2003). Moreover, transcription factors such as c-FOS (Travnickova-Bendova et al., 2002) and EGR1 (Riedel et al., 2018) also participate in regulation of the SCN clock through optical signals.
Based on RHT innervation and the neurochemical natures of the cells, the SCN is distinguished into two main subdivisions: the VL area (core area) and the DM area (shell area) (Moore et al., 2002).
The VL-SCN, which contains VIP and gastrin-releasing peptide, is located above the optic chiasm, and receives most of its photic input from RHT innervation. In contrast, the DM-SCN region, where neurons contain AVP and calretinin, receives mostly neural signals from the hypothalamus, limbic areas, and the VL-SCN (Golombek and Rosenstein, 2010).
VL-SCN neurons respond to photic stimuli during the subjective night and communicate with DM-SCN neurons through several neurotransmitters such as VIP, GRP, and SP (Best et al., 1999; Berson et al., 2002), resulting in synchronization of the DM-SCN to the expression of these proteins. Therefore, DM-SCN neurons depend on neuropeptide release from the VL-SCN, although they show stronger autonomous oscillations (Moga and Moore, 1997; Meijer and Schwartz, 2003; Antle and Silver, 2005).
Other stimuli can synchronize CR, resulting in “non-photic entrainment.” Arousal stimuli are non-photic entrainments which include social interactions, exercise, restraint stress, and caffeine-induced arousal (Mistlberger and Skene, 2004).
Different species respond differently to arousal stimuli. For example, during the daytime, wheel running or gentle handling cause a robust phase advance in locomotor activity rhythms of
Syrian hamsters, whereas these stimuli do not arouse a similar phase advance in the behavioral rhythms of mice or rats (Antle and Mistlberger, 2000; Mistlberger et al., 2002). However, phase advance in peripheral clocks such as liver and kidney can be induced by restraint stress or running wheel exercise during the intermediate of the light phase in mice (Tahara et al., 2015; Sasaki et al., 2016).
These studies suggested that under LD cycles, peripheral tissues directly respond to behavioral entrainment without influencing the SCN. However, in constant darkness during projected daytime, behavioral rhythms of mice can also be observed under several-hour daily wheel stimulation (Mistlberger, 1991). These results imply that the master clock in mice is primarily impacted by light rather than arousal stimuli.
A review by Golombek and Rosenstein (2010) showed that arousal stimuli transmit to the SCN through two major pathways. The GHT from the thalamic IGL, employs NPY, GABA, and endorphins as neurotransmitters, and plays a role in activation of the Y2 receptor/PKC pathway. The other is the serotonergic median raphe nucleus projection to the SCN, where serotonergic effects are mediated by 5-HT1A/7 receptors and PKA activation (Golombek and Rosenstein, 2010). The mechanisms responsible for behavioral entrainment of peripheral clocks include physiological factors such as glucocorticoids, sympathetic nerves, oxidative stress, hypoxia, PH, cytokines, and temperature (Tahara and Shibata, 2018).
The CR of mammals can be synchronized by light, and thus the central clock for light response, the SCN, is considered a light-entrainable oscillator. Food is a non-photic stimulus that can reset the circadian rhythm. Many studies have confirmed the existence of a FEO, but the anatomical location(s) and molecular timekeeping mechanisms of the FEO have not been determined.
In the case of mammalian research, food is provided and consumed within a few hours. This condition is called time-restricted feeding, or temporal food restriction, and is frequently used to study FAA (Challet et al., 1998).
FAA refers to the output of the FEO, which appears under time-restricted feeding, but disappears under ad libitum access, and reoccurs during the following fasting (Pendergast and Yamazaki, 2018). Mice allowed access to a standard diet ad libitum typically ingest 60–80% of their daily food intake at night.
Feeding-fasting signals reset the circadian clocks in the peripheral tissues by causing periodic availability of many circulating macronutrients (Woods, 2005). For example, ingestion causes release of insulin into the blood and induces clock gene expression in insulin-sensitive tissues (such as liver, adipose, and muscle) (Oike et al., 2014).
Feeding also enhances blood glucose levels, and high glucose concentrations can down-regulate the expression of PER1 and PER2 in fibroblasts (Hirota et al., 2010), and indirectly regulate AMPK, which controls the stability of CRYs (Lamia et al., 2009). Recent evidence showed that the liver may be a communication center of the FAA synchronization signal. Under time-restricted feeding, the liver increased βOHB production through regulation of Cpt1a and Hmgcs2 by Per2, which was then used as a signal to cause animals to anticipate feeding time (Chavan et al., 2016).
In addition, food restriction may alter the anabolic/catabolic cycles of tissues, which may affect the cellular redox state (Rutter et al., 2002) and further influence the circadian feedback loop (Mendoza, 2007).
For instance, NADH and NADPH, classical cofactors for intracellular redox reactions, can promote binding of CLOCK/BMAL1 and NPAS2/BMAL1 dimers to DNA, while their oxidized forms, NAD(P)+, inhibit this binding (Rutter et al., 2001, 2002; Schibler and Naef, 2005). Furthermore, restricted feeding may influence body temperature cycles, which have been confirmed to entrain peripheral clocks (Brown et al., 2002).
Although the central oscillators are not affected by temporal food restriction during the light phase of the day (Damiola et al., 2000; Stokkan et al., 2001), SCN-derived physiological rhythms such as locomotor activity and body temperature can be entrained by caloric restriction (also called hypocaloric diet, characterized by caloric intake reduction to 60% of the animal’s normal daily food intake) under LD cycles (Challet et al., 1997) or time-restricted feeding in constant darkness (Holmes and Mistlberger, 2000). Although the mechanisms by which caloric restriction affects the SCN clock has not been characterized, previous studies suggested that receptors of metabolism-related hormones such as insulin, leptin, and ghrelin, which exist in SCN cells, may involve in synchronizing the SCN (Guan et al., 1997; Zigman et al., 2006).
Moreover, feeding regulates brain structures that project straightly to the SCN such as the DM hypothalamus and the arcuate nucleus through orexin and ghrelin (Akiyama et al., 2004; Mieda et al., 2004; LeSauter et al., 2009; Moriya et al., 2009; Acosta-Galvan et al., 2011). Moreover, glucose influences the neural activity phase in vitro SCN slices (Hall et al., 1997). Further researches are required to evaluate the involvement of these factors and their probable impacts on the SCN.
Temperature is a non-photic synchronizer with a weaker synchronization effect than that of light. Roberto Refinetti found that only 60–80% mice can be synchronized by ambient temperature cycles, and stable entrainment takes longer in response to temperature cycles than LD cycles (Refinetti, 2010).
Though multiple physiological processes rely on ambient temperature, the CR system has a significant feature called temperature compensation, in which circadian oscillations remain resistant to temperature changes, resulting in the period length still maintaining approximately 24 h despite ambient temperature changes (Isojima et al., 2009; Narasimamurthy and Virshup, 2017).
In addition, the SCN clock does not respond to temperature stimuli, whereas cells and tissues outside of the SCN can be synchronized by temperature fluctuations (Brown et al., 2002; Buhr et al., 2010; Ohnishi et al., 2014).
The reason why the SCN is resistant to temperature changes could be that the circadian clocks in the SCN cells have robust couplings. After uncoupling of SCN cells using tetrodotoxin or nimodipine, temperature sensitivity was detected (Buhr et al., 2010). Because temperature is a weak synchronizer, it is not typically used as a zeitgeber in animal experiments.
The phase shift of the clock in response to simulated body temperature fluctuations has been demonstrated in vitro (Brown et al., 2002; Buhr et al., 2010; Saini et al., 2012; Dudek et al., 2017), and the amplitude of circadian gene expression was enhanced by temperature cycles (Dibner et al., 2009; Sporl et al., 2011; Dudek et al., 2017).
Mammalian cells sense temperature fluctuations through a series of temperature-stimulated TRP channel subfamily members called thermo-TRPs, each of which is activated in a narrow thermal range (Poletini et al., 2015). The intrinsic mechanism of temperature effects on CR may be mediated by HSF1 and CIRBPs (Ki et al., 2015). HSF1 is a circadian transcription factor which binds the heat shock element sequence, causing circadian activation of downstream promoters such as Per2 (Kornmann et al., 2007; Reinke et al., 2008; Tamaru et al., 2011). CIRBP-mediated post-transcriptional regulation allows high-amplitude clock genes express, including that of the core clock gene CLOCK (Ki et al., 2015).
In vitro experiments lack the humoral and neuronal factors that can act as entrainments in vivo. As such, cultured cells or explants are desynchronized and circadian oscillations are absent. Thus, to observe endogenous oscillations of cells or explants in vitro, chemical factors with similar effects to in vivo entrainments are required.
Many chemical factors can function as synchronizers. It has been reported that glucocorticoid, a kind of anti-inflammatory hormones released by the adrenal cortex, serves as a vigorous synchronizer of peripheral tissues (Balsalobre et al., 2000a, b; Yamamoto et al., 2005; Segall et al., 2006; So et al., 2009; Cheon et al., 2013).
Glucocorticoids activate GRs, which bind to the GREs on the promoters of core clock genes such as Per1, Per2, and E4bp4, thereby directly activating the core clock cycle, or by activating the transcription factor HNF4alpha, which targets downstream rhythmic genes without GRE elements (Reddy et al., 2007; Cheon et al., 2013). Dex cannot be used for SCN synchronization, because the SCN does not express GRs.
A study performed in 1998 indicated that serum shock could induce the rhythmic expression of Perl, Per2, Reverb-α, Dbp, and Tef in cultured rat fibroblasts (Balsalobre et al., 1998). Thereafter, serum shock was used to synchronize the circadian oscillations of various tissues, immortalized cells, and fibroblast cells.
This study suggested that various factors in the blood could stimulate rhythmic oscillations. Previous reports showed that various factors such as such as EGF (Izumo et al., 2006), calcium (Balsalobre et al., 2000b), glucose (Hirota et al., 2002), PGE2 (Tsuchiya et al., 2005), 1α,25-dihydroxyvitamin D3 (Gutierrez-Monreal et al., 2014), and atomoxetine (O’Keeffe et al., 2012) can synchronize clock genes.
Moreover, serum shock can induce Ser/Thr phosphorylation of CLOCK through the Ca2+-dependent PKC pathway (Shim et al., 2007), or activate the p42/44 MAPK pathway in a manner similar to that observed in response to light pulses (Ginty et al., 1993; Yagita and Okamura, 2000), which may cause resetting of the clock. In addition, a previous report showed that blood-borne signals activate GTPase RhoA, which promotes G-actin polymerization into F-actin, resulting in release of MRTFs into the nucleus, thus regulating the transcription of clock genes (Esnault et al., 2014).
Another common chemical synchronizing agent, Fsk, has a similar mechanism to that of serum. Fsk directly activates AC, which activates CREB through phosphorylation by promoting the synthesis of cAMP and activating PKA indirectly (Yagita and Okamura, 2000).
Mechanical Stimuli and Oxidative/ Hypoxia Stress
A recent study showed that intermittent uniaxial stretching of bone marrow-derived mesenchymal stem cells, DPSCs, and adipose tissue-derived mesenchymal stem cells can reset their CR, resulting in a synchronization effect similar to that observed in response to Dex treatment (Rogers et al., 2017).
Changing media also could reset cellular CR (Yeom et al., 2010; Guenthner et al., 2014). Mechanical stimuli provide researchers with alternate mechanisms to reset the circadian clocks of cells, such as DPSCs, which cannot be synchronized by other methods. However, the mechanisms by which mechanical stimuli induced synchronization are still unclear. It might be involved the RhoA pathway, by which short-duration fluid shear force can induce changes expression levels of clock genes in osteoblasts (Hamamura et al., 2012).
Oxidative or hypoxic stimuli may lead to circadian clock entrainment. In vitro, oxidative stimulation via hydrogen peroxide synchronizes cellular circadian oscillations in dose- and time-dependent manners (Tamaru et al., 2013).
In vivo, phase shifts in peripheral clocks such as those in the kidney and liver are observed after hydrogen peroxide injection (Geerdink et al., 2016). However, whether the master clock responds to oxidative stress is unknown. Oxygen cycles (12-h 5%:12-h 8%) synchronize cellular clocks via a key transcription factor in cultured fibroblasts, HIF1α, which has similar a similar PAS domain to CLOCK and BMAL1 (Adamovich et al., 2017).
Furthermore, at the onset of a 6-h-shifted dark period, hypoxic stimulation (14% O2) for 2 h advanced phases of locomotor activities in response to a new LD cycle (Adamovich et al., 2017). These results suggested that both central and peripheral oscillations can be reset by hypoxic stimulation. Taken together, oxygen signals may induce circadian synchronization in vitro.