Circadian rhythm displays endogenous oscillation of approximately 24-h period in various biological processes. This rhythm is entrained by diverse environmental cues such as light/dark cycle and fluctuation of temperature (Rensing and Ruoff, 2002; Wright et al., 2013). By coordinating the external time cue, organisms are able to maintain normal physiology and behavior at the appropriate time of the day. Even without the external cues, the internal clock can maintain the circadian rhythm per se. In mammals, the master clock residing in the suprachiasmatic nucleus (SCN) of the anterior hypothalamus serves as the central peacemaker and synchronizes the peripheral clocks through humoral and neuronal cues (Honma, 2018).

The mammalian molecular clock network consists of two loops, namely core and auxiliary (or stabilizing) loops. In the core loop, the positive elements, circadian locomotor output cycles kaput (CLOCK) and brain and muscle arnt-like protein-1 (BMAL1), heterodimerize and initiate the transcription of target genes including Periods (Per), Cryptochromes (Cry), Rev-erbα, RORα, and clock-controlled genes (CCGs) by acting on the E-boxes in their promoter regions. The heterodimer of PERIOD (PER) and CRYPTOCHROME (CRY) gives negative feedback to CLOCK and BMAL1 heterodimer complex and represses their own transcription. In the auxiliary loop, REV-ERVα and retinoic acid receptor-related orphan receptor α (RORα) regulate ROR/REV-ERB response elements (ROREs) in the Bmal1 promoter region. RORα positively regulates the transcription of Bmal1, while REV-ERVα negatively controls it respectively (Takahashi, 2017). This molecular time-keeping system is necessary for the maintenance of the circadian rhythms of physiological and neural functions (Refinetti, 2016).

Recently, compelling evidence indicates that gut microbiota plays an important role in controlling the development, physiology, and behaviors of the host (Cryan and Dinan, 2012; Kim et al., 2017; Sgritta et al., 2019). Gut microbiota is defined as an ecological community of symbiotic and pathogenic microorganisms in the gastrointestinal tract (Bäckhed et al., 2005). In fact, gut microbiota produces a variety of metabolic compounds related to neurotransmitters and signaling molecules that may influence physiological consequences including the circadian rhythm of the host (Ge et al., 2017; Leone et al., 2015; Parkar et al., 2019). For instance, circadian oscillation of Bmal1 transcript in the small intestine of antibiotics-induced microbiota depleted mice was down-regulated by the dysbiosis of gut microbiota (Mukherji et al., 2013) and the circadian pattern of Per2 mRNA expression in the hepatic organoid changed upon treatment with various metabolites from gut microbiota (Leone et al., 2015). These results strongly suggest that gut microbiota can influence the host circadian rhythm.

It is of note that small chemical compounds exclusively produced by gut microbiota could act as chemical messengers and influence the physiological regulation of the host. Gut microbiota digests dietary carbohydrates, thereby producing many compounds that cannot be broken down by the host metabolism (Jones, 2014; Makki et al., 2018). Interestingly, the genus Clostridium, one of the major microbiomes in the gut microbiota, produces a unique set of metabolites to achieve chemical communication with the host (Liang and Fitzgerald, 2017; Thaiss et al., 2014). Among their metabolites, the biochemical pathways of three metabolites, 4-hydroxyl-phenylpropionic acid (4-OH-PPA) and phenylpropionic acid (PPA) are recently identified to be exclusively produced by Clostridium sporogenes (Dodd et al., 2017; Elsden et al., 1976). We presumed that C. sporogenes-derived metabolites, 4-OH-PPA and PPA, could affect physiological functions such as the circadian rhythm of the host. To test this hypothesis, the effect of 4-OH-PPA and PPA on the expression of circadian clock genes in genetically engineered mouse embryonic fibroblast (MEF) cell lines expressing luciferase (Luc) reporter under the control of Per2 gene and Bmal1 promoter was examined using a real-time bioluminescence monitoring technique. The effect of 4-OH-PPA on PER2 oscillation profiles in several neural and peripheral tissues derived from Per2::Luc knockin mouse was also examined.

Cell culture

MEF cells were generated from Per2::LucSV knockin (Yoo et al., 2017) and Bmal1-Luc (Nakajima et al., 2010) transgenic mice. In Per2::LucSV knockin MEF cells, luciferase fused with endogenous PER2 protein was expressed, and in Bmal1-Luc MEF cells, the expression of luciferase was under the control of Bmal1 promoter (Fig. 1A). MEF cells were incubated in the CO₂ incubator (Thermo Fisher Scientific, USA) at 37°C and 5% CO₂. The cells were cultured in Dulbecco’s modified Eagle’s medium (Gibco, USA) supplemented with 10% fetal bovine serum (Gibco) and 1% Antibiotic-Antimycotic (Gibco). Cells were subcultured at 80% confluence once every 2 days.

Real-time recording of circadian gene-regulated bioluminescence in vitro

Real-time bioluminescence recording was conducted according to a previously described method with minor modification (Lee et al., 2016). A day before the real-time recording, MEF cells were seeded in 35 mm dish (Fisher Scientific, USA) at 50% confluence. After 24 h, culture media was changed to synchronization media with 200 nM dexamethasone (Sigma-Aldrich, USA) to synchronize circadian rhythms of cell population (So et al., 2009) for 2 h. To record the real-time bioluminescence, synchronized cells were cultured in recording media with 200 μM D-luciferin (Promega, USA) and culture dishes were sealed with parafilm (Sigma-Aldrich). Next, the dishes were placed in Kronos, a real-time bioluminescence recording device (ATTO, Japan), at 37°C and 5% CO₂, and luciferase (Luc) activity in each dish was measured for 1 min every 10 min for 3 to 5 days. For the dose-response effect of 4-OH-PPA and PPA, the cells were treated with four different doses, 0.125 mM, 0.5 mM, 1 mM, and 2 mM of 4-OH-PPA (Sigma-Aldrich) and PPA (Sigma-Aldrich) using two different administration modes, either immediately after the nadir or the peak of the oscillation (Fig. 1B). To verify the effect of 4-OH-PPA and PPA, their precursors, tyrosine (Sigma-Aldrich) and phenylalanine (Sigma-Aldrich), were examined initially as the control experiments.

Animal care and organotypic slice culture

Per2::Luc knockin mice (Yoo et al., 2004) were maintained in a specific pathogen free (SPF) condition at 23°C under 12:12 light/dark cycle with standard mouse chow and water. All animal procedures were approved by the Institutional Animal Care and Use Committee of laboratory animal resource center in Daegu Gyeongbuk Institute of Science and Technology (DGIST-IACUC-19070201-01). Organotypic culture was prepared by Stoppini method (Stoppini et al., 1991) with minor modifications. Briefly, heterozygous Per2::Luc neonatal mice between postnatal days 5 to 7 were anesthetized on ice and several tissues were removed for culture. The excised brain was coronally sectioned at a thickness of 400 μm using the VT1000S vibratome (Leica Biosystem, Germany), and the SCN and hippocampus were used in this experiment. Peripheral tissues including liver and small intestine were trimmed and placed onto the culture insert (Merck, Germany) in a 24 well plate (SPL Life Sciences, Korea) with pre-warmed culture media (minimal essential media [MEM] supplemented with 25% horse serum [Gibco], 25% Gey’s balanced salt solution [Sigma-Aldrich], 36 mM glucose and 100 U/ml Antibiotic-Antimycotic). The cultures were incubated at 37°C and 5% CO₂. Next day, the slice culture media were replaced and during stabilization for 2 weeks, the slice culture media were replaced once every 3 days.

For real-time bioluminescence recording ex vivo, a method previously described by Koo et al. (2015) was used. After the preparation, the tissue was cultured in synchronization media with 1 μM final concentration of dexamethasone for 2 h except for the SCN. After 2 h of synchronization, the tissue was transferred to pre-warmed recording media (culture media containing 200 μM D-Luc). The dishes were sealed with parafilm and placed in Kronos device as described above.

Statistical analysis

The analysis of circadian rhythm after the administration of metabolites or vehicle was performed using FFT-NLLS function from the online BioDare2 analysis platform (https://www.biodare2.ed.ac.uk) (Zielinski et al., 2014). Three peaks of PER2 circadian expression and a single peak of Bmal1 circadian expression after administration of drug was used for analysis. The change (Δ) in the amplitude and period of respective oscillatory cycles of Per2 and Bmal1 genes was defined as a relative difference based on the control values. Statistical significance in the period and amplitude of real-time bioluminescence data was evaluated by one-way ANOVA followed by post hoc Dunnett’s test. On the other hand, the significance of phase response was additionally calculated by two-way ANOVA followed by post hoc Bonferroni’s test to compare the effects of the different modes of administration (nadir vs peak). Ex vivo real-time bioluminescence data were analyzed by unpaired two-tailed Student’s t-test. Experiments were replicated at least 4 to 6 times, and the data on the amplitude and period of clock gene oscillation profiles are presented as mean ± SEM. A P value < 0.05 indicated statistical significance.

Effects of 4-OH-PPA and PPA on real-time oscillation profiles of PER2 and Bmal1 circadian gene expression in vitro

To investigate if C. sporogenes-derived two metabolites, 4-OH-PPA and PPA, can affect the circadian profiles of clock gene expression, we used a real-time bioluminescence recording method to monitor PER2 expression in the Per2::LucSV knockin MEF cells (Yoo et al., 2017). As the biochemical pathways synthesizing 4-OH-PPA and PPA in C. sporogenes were recently characterized (Dodd et al., 2017), identifying tyrosine and phenylalanine as the precursors of 4-OH-PPA and PPA, respectively. Therefore, we initially conducted control experiments using the precursors of the metabolites. As shown in Fig. 2, these two amino acids did not alter PER2 oscillation. Furthermore, we examined the effect of 4-OH-PPA and PPA on PER2 oscillation at four different doses (0.125 mM, 0.5 mM, 1 mM, and 2 mM). Because organisms change internal status under the control of circadian rhythm and may react differently, we adopted two different administration modes, either immediately after the nadir or the peak of circadian oscillation. 4-OH-PPA increased the amplitude of PER2 circadian expression in a dose-dependent manner regardless of the administration modes (Figs. 3A and 3B). The changes (Δ) in period and amplitude of four consecutive cycles were calculated based on the value of vehicle group (Table 1). Higher doses of 4-OH-PPA increased the period significantly except when 1 mM 4-OH-PPA was treated immediately after the peak of PER2 oscillation. The circadian oscillation of PER2 expression showed higher amplitude in the 4-OH-PPA treatment groups when compared with vehicle groups in both nadir and peak administration modes (Table 1).

We also determined Bmal1 circadian expression by applying real-time bioluminescence recording technique to MEF derived from Bmal1-Luc mice (Nakajima et al., 2010). The amplitude of Bmal1 circadian rhythm was increased by the four different doses of 4-OH-PPA at the first circadian oscillation and damped at the second cycle (Figs. 3C and 3D). Even though the period was not affected at any of the doses of 4-OH-PPA on Bmal1 oscillation, 4-OH-PPA significantly increased the amplitude of Bmal1 oscillation at the higher doses (1 mM and 2 mM) (Table 1).

Similar to 4-OH-PPA, the period of PER2 circadian expression was lengthened by the higher doses (1 mM and 2 mM) of PPA at both nadir and peak of its oscillation. However, the period of Bmal1 rhythmicity was significantly lengthened only with the administration of 2 mM PPA immediately after the nadir due to more damping at the second peak. In addition, the administration of PPA at the four different doses increased the amplitude of PER2 and Bmal1 oscillation in a dose-dependent manner (Table 2, Fig. 4). The increase in amplitude was greater with 4-OH-PPA than with PPA (Please compare Tables 1 and 2).

We also examined the phase shift of PER2 oscillation induced by 4-OH-PPA and PPA with the two administration modes. Depending on the mode of administration, 4-OH-PPA (Fig. 5A) and PPA (Fig. 5B) revealed different phase shifts in a dose-dependent manner. The peak phase of PER2 oscillation was delayed with treatment administered immediately after the nadir, whereas the peak phase was advanced with treatment administered immediately after the peak.

Effect of 4-OH-PPA on circadian oscillation of PER2 in organotypic cultures ex vivo

According to the in vitro results, the 4-OH-PPA caused greater alterations in PER2 oscillation when compared with PPA treatment, especially at the peak time point. We proceeded to examine if 4-OH-PPA can affect PER2 circadian oscillation ex vivo. We used four different tissue regions derived from Per2::Luc knockin mice: the SCN containing the master clock, hippocampus (one of the brain local clocks), liver, and small intestine (two other peripheral clocks). In comparison with the vehicle, 4-OH-PPA at 2 mM increased PER2 oscillation in the SCN tissue when administered at the peak, resulting in higher amplitude and lengthened period of PER2 oscillation (Fig. 6A). In the hippocampus, 4-OH-PPA treatment elicited a significant increase in the amplitude, but not the period of PER2 (Fig. 6B). In the two peripheral tissues (liver and small intestine), 4-OH-PPA treatment significantly increased the amplitude and lengthened the period of PER2 oscillation (Figs. 6C and 6D). These ex vivo results demonstrated that the rhythmicity of PER2 circadian oscillation consistently enhanced the amplitude in the four tissue regions examined. Increased amplitude was accompanied by the increased overall expression of PER2::LUC in the brain oscillators, the SCN and hippocampus, but not in the peripheral oscillators, the liver and small intestine. The increased amplitude in the peripheral oscillators was due to the difference in the bioluminescence of the peak and the nadir rather than the increase of basal bioluminescence throughout the oscillation. It appeared that 4-OH-PPA exhibited a tissue-specific modulation of the circadian rhythm of each tissue.

In the present study, we demonstrated that two metabolites of C. sporogenes, 4-OH-PPA and PPA, regulated the circadian oscillation of Per2 and Bmal1 genes in the MEF cells. First, 4-OH-PPA and PPA increased the amplitude of both PER2 and Bmal1 oscillation in a dose-dependent manner with a significant increase at higher concentrations (1 mM and 2 mM). The period of PER2 oscillation was lengthened by 4-OH-PPA and PPA, and the phase response in PER2 circadian oscillation differed significantly depending on the mode of treatment. In addition, the organotypic cultures of the SCN, hippocampus, liver, and small intestine derived from Per2::Luc knockin mice revealed that the effect of 4-OH-PPA on circadian rhythm in the MEF cells can be reproduced at the tissue level to a certain extent. In fact, 2 mM 4-OH-PPA affected the circadian rhythm of all of the tissue regions by increasing the amplitude and lengthening the period of PER2 oscillation.

The dosage of metabolites used in the present study (0.125-2 mM) was within the physiological range because endogenous concentrations of 4-OH-PPA and PPA in human blood are around 0.5 mM according to human metabolome database (Wishart et al., 2018). It appears that metabolites produced by C. sporogenes can affect the host’s molecular circadian clock at a physiologically relevant level.

Even though gut microbiota is not directly exposed to external time cues, such as light/darkness cycle, it has been known that gut microbiota including clostridium exhibits diurnal fluctuation that is influenced by feeding rhythms and metabolic profiles over the course of time of the day, are revealed recently by gut microbiome studies (Liang and Fitzgerald, 2017; Thaiss et al., 2014). In addition, antibiotics treatment disturbed the compositional changes in gut microbiota leading to physiological consequences such as gut motility (Ge et al., 2017). Considering the daily fluctuation in the concentration of metabolites produced by Clostridium, it is important to assess circadian profiles depending on the different responses to circadian time points. As shown in Fig. 5, the two modes of administration (nadir and peak) of the drugs yielded different phase shift of circadian oscillation in the MEF cells. The phase shift is important because it can determine when to exert certain physiology correctly by advance or delay (Eastman et al., 2015). In fact, it has been known that short-chain fatty acid (SCFA), one of the gut microbial metabolites, and dietary fiber-containing diets regulate the phase of the host’s circadian rhythm in the peripheral tissues (Tahara et al., 2018).

The overall robustness of PER2 oscillation pattern in the SCN as a master clock appears to be similar to those of the local clocks in other tissues examined. Throughout the recording, the brain oscillators showed increased average bioluminescence with higher basal level after the administration of 4-OH-PPA. However, the peripheral oscillators had the tendency to maintain or lower the basal level. It appeared that there might be differential tissue-specificity in the central and peripheral clock machineries, and this is consistent with the tissue-specific robustness of circadian oscillation repeatedly noticed in the central and local clocks (Buhr et al., 2010; Yoo et al., 2004). Furthermore, it is worth noting that the gut-produced metabolites of C. sporogenes may be uptaken into either the blood circulation or the nerve terminals of the nerve system (Abot et al., 2018; Krishnan et al., 2015), transported either into the peripheral tissues or central nervous system (CNS), and involved in the homeostatic regulation of neural functions. Although there is not much information about 4-OH-PPA and PPA, recent researches have been conducted on which receptors are involved to transfer the signal of several representative molecules from the gut microbiota, such as SCFAs (Chen et al., 2019; Lund et al., 2018). Following signaling cascades of these receptors, we may find a clue to figure out how 4-OH-PPA and PPA control expression of circadian genes. Since several papers recently report that the signals from gut microbiota are transferred to the brain through the vagus nerve system (Bonaz et al., 2018; Sgritta et al., 2019), a study using vagotomy would be worthwhile to investigate the physiological significance of C. sporogenes-derived metabolites. Thus, further studies are needed to elucidate the mechanism of action in the microbiota-gut-brain axis.

In conclusion, the present study demonstrated that novel function of 4-OH-PPA and PPA produced by C. sporogenes playing an important role in modulating the circadian clock machinery in the peripheral tissues.

This work was supported by the Ministry of Science, ICT and Future Planning through the National Research Foundation of Korea (NRF-2017R1A2A1A05001351) and DGIST Startup Fund Program (2019010078).

Changes in period and amplitude of circadian oscillation after 4-OH-PPA treatment

Changes in period and amplitude of circadian oscillation after PPA treatment