In Vivo Initiation of Clock Gene Expression Rhythmicity in Fetal Rat Suprachiasmatic Nuclei

In mammals, the suprachiasmatic nuclei of the hypothalamus (SCN) are a site of the principal pacemaker [1], which drives overt behavioral, physiological, hormonal, biochemical and molecular rhythms (for review, see [2]). These rhythms repeat with a period of approximately 24 h and are thus termed “circadian”. Overt circadian rhythms can be detected in the SCN themselves, namely in metabolic [3] and electrical [4] activity. At the molecular level, overt circadian rhythms manifest in spontaneous expressions of genes related to neuronal activity, e.g., immediate early gene c-fos and arginine vasopressin (Avp) [5]–[7]. In adults, the SCN drives these overt rhythms via a cell-autonomous molecular clockwork (for review, see [8]). This clock mechanism is based on the rhythmic expression of so-called clock genes. The rhythm is endogenously generated and maintained due to the mechanism in which the clock gene protein products feed back onto promoters of the clock genes and activate or inhibit their transcription. Clock proteins CLOCK an BMAL1 serve as transcriptional activators that switch on transcription of clock genes Per1,2, Cry1,2 and Nr1d1 (Rev-erbα). PER1,2 and CRY1,2 proteins serve as inhibitors of the CLOCK:BMAL1 activated transcription and NR1D1 directly inhibits Bmal1 expression. As a result, in the SCN, the expression of Per1,2, Cry1,2, Nr1d1 peaks during the subjective day, whereas that of Bmal1 peaks during the subjective night. Importantly, the presence of circadian variation in transcript levels is conditional for clock “ticking”. The expression of Per1 and Per2, but not of other clock genes, is activated by light at night [9], [10], which represents a mechanism for entrainment of the molecular clock by the light/dark cycle. The SCN is a multi-cellular clock whose properties are dependent on a synaptic web ensuring communication among individually rhythmic cells. This unique feature of the SCN provides the central clock with its robustness, plasticity and ability to adjust to changes in external environment [11], [12]. The molecular mechanism is also present in other neuronal and non-neuronal cells of the body, i.e., so-called peripheral oscillators. Importantly, the peripheral clocks are not directly synchronized by light but via by not yet fully recognized local signals and systemic signals derived from the SCN [13]. Systemic signals set the phases of the peripheral oscillators relatively to the SCN by modulating the expression of some of the clock genes of the peripheral clocks [14].

During ontogenesis, SCN morphology and rhythmicity develop gradually (for review, see [15]). In the rat, gestation takes 22 days and the nuclei are formed from the embryonic day (E) 14 through E17 from the specialized zone of the ventral diencephalic germinal epithelium as a component of the periventricular cell groups [16]. Whereas neurogenesis is completed at approximately E17, SCN morphology is not yet complete because the individual SCN neurons are not mutually interconnected by synapses. Synaptogenesis progresses slowly during the late prenatal stage between E19 and parturition, when only few synapses appear in the SCN; however, their number increases rapidly postnatally, with a marked rise from postnatal day (P)4 to P10 [16].

Despite its immaturity, intrinsic rhythmicity is already present in the SCN structure in late stages of embryogenesis. Clear day-night oscillation in metabolic activity was detected in the fetal rat SCN already from E19 through E21 by monitoring 2-deoxyglucose uptake [17]. However, rhythmicity in Avp expression is not present before E20 [18], [19], and rhythmic firing rates in SCN neurons appear only at E22 [20]. To determine if these rhythms are driven by the fetal SCN clock, daily profiles of clock gene expression were measured during the prenatal period by in situ hybridization. At E19, an early stage in rat SCN development, no significant circadian rhythmicity was detected in Per1, Per2, Cry1, Bmal1 expression [21]. One day later, at E20, rhythmic expression of Per1 and Per2 was reported [22], [23]. The robustness of clock oscillation further develops postnatally, in correlation with synaptogenesis; in the rat, the amplitude of clock gene expression rhythms increased up to P5–P10 [19], [21]. In mice, the gestation takes 18 days and a daily rhythm of Per1, but not of Per2, expression was detected at E17 in the SCN. The rhythm in Per2 expression was not present at P3 and it appeared only at P6 [24]. The aim of the present study was to test the hypothesis that at the early fetal SCN development, in vivo oscillation is driven by maternal rhythmic cues before an autonomous clock mechanism is established. To test this hypothesis, we compared the daily profiles of clock gene transcript levels (Per2, Nr1d1 and Bmal1) in the fetal SCN at two developmental stages, namely at E19 and E21. Additionally, daily expression profiles of c-fos, a marker of neuronal activity, and Avp and Vip, two main SCN neurotransmitters, were measured. In contrast to previous studies, which used in situ hybridization to detect the transcript levels, in this study we used a more sensitive quantitative RT-PCR method. In the fetal SCN, most of the transcript levels were found to be relatively low and, therefore, there was a possibility that their circadian variation could be missed especially in case when the background staining of the in situ hybridization probe was higher. Using the quantitative RT-PCR overcomes this methodological difficulty and provides more precise detection of daily variations in the transcript levels. The results revealed that in vivo, circadian rhythms in gene expression are present in the fetal SCN before the functional molecular clock develops.

Our results revealed a gradual development of circadian oscillations in individual clock gene expression in the SCN during late fetal development in vivo. These data suggest that the fetal clock does not operate at E19 but is functional at E21, immediately before parturition. In this study, transcript levels were quantified in laser-dissected SCN tissue by RT PCR, which allowed us to study the developing rhythms in more detail and with higher resolution. Hence, weak rhythms or rhythms of low-abundance transcripts could be revealed. These results are in agreement with our previous data, in which the transcript levels in the fetal SCN were detected using in situ hybridization [19], [21], a less sensitive method.

Whereas the molecular core clock components Per2 and Bmal1 were not expressed rhythmically at E19, expression became rhythmic at E21. In contrast, the clock gene Nr1d1 was rhythmically expressed at both E19 and E21. The presence of the rhythm confirms that the lack of rhythmicity in Per2 and Bmal1 expression at E19 did not occur because the profiles of individual fetuses were rhythmic but not mutually synchronized. Per2 encodes the PER2 protein, which is negative element of the transcriptional-translational feedback loop (for review, see [8]); the rhythm of its availability is necessary for the rhythmic suppression of CLOCK:BMAL1- activated E-box mediated gene transcription. The absence of rhythmic expression in Per2 and Bmal1 suggests that the expression of other clock genes is neither repressed nor activated rhythmically at E19; thus, the clock likely does not operate autonomously. This hypothesis is supported by the results of our previous study; we observed constitutive expression of not only Per2 and Bmal1 but also of other canonical clock genes, namely Per1 and Cry1, in the SCN at E19 [21]. NR1D1 represents a core clock component of the molecular clock which is also rhythmically activated by CLOCK:BMAL1 via an E-box on its promoter. NR1D1 suppresses Bmal1 expression during the subjective day [25]. The rhythmicity of its expression may be driven not only by the clock mechanism, but also by other pathways because its promoter region contains various response elements [26], [27]. NR1D1 represents a formally classified orphan receptor whose ligands (heme, carbon monoxide) have been identified [28], [29] and thus, daily oscillations in these ligands, driven by maternal metabolic cues, might rhythmically activate NR1D1. Because of an auto-feedback loop in which NR1D1 binds its own promoter [30], the imposed rhythm in NR1D1 activity may drive rhythm in Nr1d1 expression, independent of fetal clock silence. In support of this scenario, Nr1d1 rhythmicity independent of rhythm in other canonical clock genes has been observed in the intestine clocks of behaviorally arrhythmic adults rats maintained in LL [31].

Therefore, our results favor the hypothesis that the clock does not operate in the SCN at E19 in vivo; however some genes are expressed rhythmically due to rhythmic maternal drive [15]. Indeed, rhythmicity in expression of genes related to cellular activity, namely c-fos and Avp, was present already at E19. Interestingly, transcription of these two genes is activated by similar signaling pathways; both are regulated by phosphorylation of CREB (cAMP response element binding protein) because their promoters contain a responsive element (CRE) which binds P-CREB [32], [33]. This pathway represents a common signal responding to various stimuli. Therefore, these stimuli may be derived from maternal rhythmic cues, which elevate P-CREB levels rhythmically in SCN cells and thus drive rhythmic expression of these genes. In support of this hypothesis, at E19 the circadian rhythmic expression of both genes peaked at the same time. Moreover, these peaks correspond to peak of Nr1d1 expression rhythm, which suggests that various synchronous maternal stimuli might be involved in activation of these genes. Additionally, this finding further supports the hypothesis that Nr1d1 expression may be driven by the maternal cues at E19 (see above). This may allow us to speculate that the initiation of clock oscillatory mechanisms is in fact related to the Nr1d1 expression rhythm. It is unlikely that the rhythm in Nr1d1 expression is conditional for the initiation of the fetal clock function because the clock develops autonomously even in fetuses of arrhythmic mothers without a functional clock [34]. It could, however, be responsible for proper phasing of the developing fetal clock, which is born in phase with the maternal circadian clock [15]. Nr1d1 oscillation has already been suggested to play a role in entrainment of circadian clocks in adults [35], [36]. In accordance with this hypothesis, the maternally driven rhythmicity in the fetal SCN should be abolished in arrhythmic mothers. However, the gene expression profiles in this study resulted from population data (i.e., each time point of the profile represents fetuses of different mother) and the absence of rhythms in the fetal SCN could also be due to mutual desynchrony among the arrhythmic mothers. Therefore, the role of Nr1d1 in maternal entrainment of the fetal SCN should be tested using other methodological approaches. In vivo recording of the bioluminescence driven by Per1 reporter revealed daily variation in the rat fetal peripheral tissues already at E19 [37], however, such recording could not detect gene expression directly from the fetal brain and the individual nuclei.

At E21, the circadian clock appears already operational in vivo because significant rhythms in clock gene expression were detected. The functional clock likely drives expression of clock-controlled genes, namely Avp. In addition to CRE, the Avp promoter also contains an E-box sequence which binds CLOCK:BMAL1 and is thus rhythmically driven by the clock mechanism [38]. Therefore, multiple pathways originating from both maternal cues and the functional fetal clock are likely to drive Avp expression at E21. As a result of the mechanistic switch, the phase of Avp rhythm was significantly shifted and the amplitude of the rhythm increased at E21 relative to E19. The amplitude and robustness of the circadian signal further increases during the postnatal period up to P5–P10 [19], [21].

Like c-fos and Avp, P-CREB pathway may also regulate Vip expression [39]. Vip expression was rhythmic at both E19 and E21, although the rhythm was very weak compared to other evaluated rhythmic genes. Therefore, the weak circadian rhythm in Vip expression is present before the clock mechanism is fully functional. VIP is an important mediator among SCN cells and is responsible for increasing the robustness of their oscillations [40]. In the fetal rat central nervous system, VIP receptors are abundant already by E11 [41]. Activation of VIP receptors leads to elevation of P-CREB levels transcription of genes containing CRE in their promoter. Therefore, the above mentioned c-fos and Avp expression rhythms observed at E19 may be related to Vip rhythmicity via this mechanism. Additionally, VIP may play a role in synchronizing SCN neurons during the developmental period when the synaptic web is lacking. However, the amplitude of Vip expression rhythm is very low in the fetal SCN and it is thus not clear whether has any functional relevance.

During fetal SCN development, transcript levels changed between E19 and E21. The initially high expression levels of Per2 and Bmal1, i.e., genes which were constitutively expressed at E19, declined with fetal age in correlation with the initiation of circadian regulation. In contrast, expression levels of genes that were rhythmically expressed at E19 (Nr1d1, c-fos, Avp and Vip) increased with fetal age, in correlation with clock development. These findings are in accordance with results of our previous study, in which clock gene expression in the SCN was detected by in situ hybridization [19], [21]; of the genes examined in the previous study (Per1, Per2, Cry1, Bmal1 and Clock), Per2 and Bmal1 were expressed with at the highest levels at E19. The results of the present study confirm that this previous finding was indeed due to differences in transcript levels and was not related to a methodological problems related with the properties of the probes used for the in situ hybridization. Apparently, these data also suggest that with the initiation of circadian control, rhythmicity is generated via suppression of high Per2 and Bmal1 transcription (this study, [21]) and induction of low Per1 and Cry1 transcription [21].

In conclusion, our results support the hypothesis that the rat SCN circadian clock develops in vivo only after E19 and becomes functional just before birth. Before the SCN clock is operational during fetal development, various rhythms could be detected in the SCN that are likely driven by the maternal rhythmic milieu. Nr1d1 represents the first rhythmically expressed clock gene, being rhythmic before the fetal clock begins to operate, likely due its sensitivity to the maternal cues. Therefore, based on our data, Nr1d1 rhythmicity may represent one of the possible mechanisms for setting the fetal molecular clock with maternal cues. Nevertheless, the role of Nr1d1 in maternal entrainment of the fetal SCN has to be proven.

The authors thank Peter Ergang for his help with using the laser dissection technique.

The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper.

This work was supported by Czech Science Foundation Grant No. P303121108 and by research project RV0 67985823. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.