The ontogeny of circadian clock gene expression during mouse fetal development

Abstract

The circadian clock in the suprachiasmatic nucleus and peripheral tissues functions to regulate key physiological and cellular systems in a cycle approximating 24 h. Understanding the ontogeny of the circadian clock mechanism during mammalian development is incomplete. Accordingly, we used the mouse as a model and a previously published RNAseq dataset to determine when expression of core genes regulating the circadian clock increase in transcript abundance in fetal and postnatal brain, heart, liver, and kidney. Transcripts for all six core genes examined (Clock, Bmal1, Per1, Per2, Cry1, and Cry2) were identified in all tissues and time points. For brain, there was a small increase in Clock at E13.5 and a larger increase at E18.5. Similarly, Bmal1 transcript abundance increased slightly at E14.5 and to a greater degree at E17.5, Per2 increased slightly at E15.5 and remained constant until after birth and Cry2 increased slightly at E18.5. For liver, transcript abundance of most genes increased at the end of gestation, with increases observed for Clock at E18.5, and for Bmal1, Per2, and Cry2 at E17.5. The genes whose expression increased before birth in heart was Clock (at E18.5), and Per2 (at E15.5) and the genes whose expression increased before birth in kidney were Clock (E18.5), Bmal1 (E15.5) and Per1 (P18.5). For all tissues, there were further increases in transcript abundance for most genes in the postnatal period with the exception of Cry1 (all tissues) and Per1 (liver and kidney). Results support the idea that the organization of the molecular clock is more advanced for the fetal brain and liver than for fetal heart and kidney. Furthermore, based on changes in gene expression, components of the molecular clock continued to mature after birth.

Graphical abstract

Highlights

• •Ontogeny of the molecular clock in brain, heart, liver and kidney was assessed

• •Core clock gene expression increased in late gestation or after birth

• •Maturation of clock gene expression in the fetus was greatest for brain and liver

1. Introduction

The molecular circadian clock has evolved across the animal lineage to orchestrate pathways necessary to maintain homeostasis in a changing environment [1,2]. One rotation of the transcription/translation feedback loop that drives the molecular circadian clock mechanism occurs approximately every 24 h [[3], [4], [5]]. In adults, molecular circadian clocks function in peripheral organs such as the lung, heart, liver, kidney, and skeletal muscle [[6], [7], [8], [9], [10], [11], [12]]. The suprachiasmatic nucleus (SCN) located within the hypothalamus has an endogenous molecular circadian clock entrained by the light: dark cycle [6,13,14]. Peripheral clocks can exhibit both SCN-dependent and independent rhythmicity [[15], [16], [17]].

Understanding the ontogeny of the circadian clock mechanism during mammalian development is incomplete. Several lines of evidence indicate that the circadian clock mechanism is not functional in the preimplantation embryo. There is a continuous decline in transcript abundance for the core circadian clock genes (BMAL1, CLOCK, CRY1, CRY2, PER1, and PER2) during this time for several species [[18], [19], [20]], there are no intronic reads in transcripts of core clock genes in the bovine embryo [20], BMAL1, CLOCK, and CRY1 disappear from blastomere nuclei of the mouse embryo after the four to eight cell stage [18], and gene expression in the mouse embryos is not affected by knockdown of Cry1 [18]. CLOCK-BMAL1 complexes can inhibit the segmentation clock required for somatogenesis [21] and it has been speculated that the circadian clock must remain suppressed in the embryo until after the completion of the body plan [22].

The circadian clock mechanism in the fetal SCN develops late in gestation in all species examined and gradually increases in functionality into the postnatal period as neuron networks develop [23]. For example, Per2 expression in cultured fetal SCN from mice exhibited some rhythmicity at embryonic day 13 (E13) (the earliest day examined) and the amplitude and robustness of rhythms increased at E15 and E17 [24]. Similarly, daily rhythms in Per1 but not Per2 expression in the mouse SCN were observed at E18 [24]. In another study, protein abundance for core clock genes in the SCN was lower at E18 than at postnatal day 10 (P10) [25].

The clock mechanism in peripheral tissues also appears to emerge in late gestation with full maturation occurring after birth. Rhythmic expression of a bioluminescent reporter construct consisting of NR1D1-luciferase (NR1D1 is a negative regulator of core clock genes) in whole mouse fetuses was apparent at E18.5 but the rhythm did not develop fully in terms of amplitude and acrophase until P20 [5]. Measurement of transcript abundance for several genes important for clock function at 4 h intervals in the kidney indicated circadian variation in gene expression at E20 but continued development of circadian rhythmicity through 4 weeks after birth [26]. In cultured fetal hearts, circadian oscillations in Per2 expression were not observed at E10 but had developed by E18 [27]. In cultured fetal kidneys, Per2 expression showed rhythmicity at E17.5 but not earlier [28]. In the same experiment, Per2 rhythmicity could be seen for bladder, heart, lung, adrenal and liver at E18.5.

Here, we used the mouse as a model to examine changes in the transcript abundance of circadian clock genes from mid gestation to the early postnatal period in four tissues – brain, heart, liver, and kidney. The assumption was that increased expression of core clock genes is indicative for development of capacity of the tissue to have a functional molecular clock. Based on previous results [14,27,28], it was hypothesized that the capacity for a functional molecular clock develops between E10 and E18 and that maturation of the clock continues into the postnatal period.

2. Materials and methods

We mined the RNAseq dataset E-MTAB-6798 [29] deposited in the European Bioinformatics Institute Biostudies database (https://www.ebi.ac.uk/biostudies/arrayexpress/studies/E-MTAB-6798) to determine transcript abundance for Bmal1, Clock, Cry1, Cry2, Per1, and Per2 in brain, heart, liver, and kidney tissues across mouse development during the fetal and early postnatal periods. Tissues were collected from fetuses of RjOrl:Swiss/CD-1 outbred mice (Mus musculus). Whole brain, heart, and liver were collected from one male and two female mice at E10.5. Whole brain, heart, liver and kidney were collected from one female and three males at E11.5. Forebrain/cerebrum, heart, liver, and kidney were collected from two male and two female mice for the remainder of the development stages [29]. Stages of development included E10.5, E11.5, E12.5, E13.5, E14.5, E15.5, E16.5, E17.5, E18.5, P0, P3, P7, P14, P28, and P63. The authors did not specify if individuals came from the same or different litters.

Raw data of single-end RNA-seq were retrieved. First, adaptors and low-quality reads were trimmed from fastq files using trim_galore (v0.6.10) with the following parameters -q 25 --phred33 --length 25 --stringency 2 (https://github.com/FelixKrueger/TrimGalore). The clean reads were mapped to mouse genome (GRCm39) and annotation (M35) from GENCODE by Salmon in mapping-based mode (v1.10.1) [30]. A full decoy-aware transcriptome file was built by using the entire genome as the decoy sequence to generate index transcriptome. Quantification of counts was executed in single-end mode. Counts from all samples were extracted, compiled and summarized to gene level in R using Bioconductor package tximeta (v1.24.0) [31]. To generate counts per million (CPM) values, count matrix were imported to edgeR (v4.2.2) [32]. Genes with very low transcript abundance were filtered out to reduce computational burden and noise, thus improving the accuracy and efficiency of differential expression analysis with meaningful expression changes. Normalization across the samples were done with intrinsic TMM model in edgeR, followed by estimation of dispersion. The CPM values were called with the function cpm in edgeR. Six circadian clock genes (Bmal1, Clock, Bmal1, Per1, Per2, Cry1 and Cry2) were selected for analysis.

The effect of stage on circadian clock gene transcript abundance for each of the four fetal tissues was tested by analysis of variance using the GLM procedure of SAS (SAS v 9.4; SAS Institute Inc., Cary, NC, USA). Differences between individual developmental days were identified using a Tukey post-hoc test when the main effect of stage of development was significant.

3. Results

Transcripts for all six core genes in the circadian molecular clock were identified in four tissue types at each developmental time point. Transcript abundance was significantly affected by day of collection for each gene. There were some differences among genes and tissues in the timing of increases in transcript abundance for each gene (Table 1). Specific patterns of transcript abundance in brain, heart, liver, and kidney for each gene are described below.

Table 1.

Timing of the increases in expression of core clock genes during fetal development and the postnatal period.

Gene	Tissue	First increase relative to E10.5–11.5a	Second increase in fetal perioda	First increase in postnatal periodb
Clock	Brain	E13.5	E18.5	P14
	Heart	E17.5	none	P14
	Liver	E18.5	none	P28
	Kidney	E18.5	none	P14
Bmal1	Brain	E14.5	E17.5	P0
	Heart	none	none	P14
	Liver	E17.5	none	P28
	Kidney	E15.5	none	P14
Per1	Brain	none	none	P14
	Heart	none	none	P14
	Liver	none	none	none
	Kidney	P18.5	none	none
Per2	Brain	E15.5	none	P14
	Heart	E16.5	none	P63
	Liver	E17.5	none	P63
	Kidney	none	none	P14
Cry1	Brain	none	none	none
	Heart	none	none	none
	Liver	none	none	none
	Kidney	none	none	none
Cry2	Brain	E18.5	none	P14
	Heart	none	none	P63
	Liver	E17.5	none	P63
	Kidney	none	none	P63

^aAn increase was defined as the first or second significant difference relative to E10.5 and E11.5 or, for kidney, E11.5.

^bAn increase was defined as the first significant difference relative to E18.5.

3.1. Clock

Results are shown in Fig. 1. Developmental patterns in gene expression for brain, liver and kidney were similar to each other with transcript abundance remaining low throughout most of the fetal period. As compared to expression at E10.5 and E11.5, there was a slight increase in transcript abundance in brain at E13.5 but transcript abundance remained low in the brain through most of gestation. Expression of Clock increased at E18.5 compared to all other stages. After birth, transcript abundance remained constant until an increase at P14 that was sustained through P63. Amounts of mRNA for Clock in heart increased at E17.5 and then again at P14. Transcript abundance in the liver remained low until a slight increase at E18.5 and a further increase at P28. Amounts of Clock mRNA in kidney remained low throughout the fetal period, with a slight increase at E18.5. Values remained low after birth until an increase at P14 and a further increase at P28.

Fig. 1.

Comparison of Clock expression across stages of fetal and postnatal developmental stages for brain, heart, liver, and kidney. Gene expression is expressed as counts per million (CPM). The statistical effect of stage is represented by the p values on each graph. Data are least-squares means ± SEM. Means with different letters differ (P < 0.05) as determined by Tukey's test.

3.2. Bmal1

Transcript abundance for Bmal1 remained low for most of fetal development for all tissues examined (Fig. 2). There was a slight increase in Bmal1 transcript abundance at E14.5 and again at E 17.5 in brain,at E17.5 in liver, and at E15.5 in kidney but amounts remained low throughout the fetal period for heart and kidney. Transcript abundance was elevated at P0 and again at P28 in brain, P14 and again at P 28 n heart, P28 in liver and P14 and again at P28 in kidney.

Fig. 2.

Comparison of Bmal1 expression across stages of fetal and postnatal developmental stages for brain, heart, liver, and kidney. Gene expression is expressed as counts per million (CPM). The statistical effect of stage is represented by the p values on each graph. Data are least-squares means ± SEM. Means with different letters differ (P < 0.05) as determined by Tukey's test.

3.3. Per1

Transcript abundance of Per1 remained low throughout fetal development (Fig. 3). There was no increase in transcript abundance for Per1 in the fetal period for brain, heart and liver and the first increase in transcript abundance in kidney was at E18.5. After birth, the first increase in Per1 expression in brain was at P14, with an additional increase at P63. The first postnatal increase in heart was at P14 with another increase at P63. There was no significant increase relative to E18.5 in the postnatal period for liver or kidney.

Fig. 3.

Comparison of Per1 expression across stages of fetal and postnatal developmental stages for brain, heart, liver, and kidney. Gene expression is expressed as counts per million (CPM). The statistical effect of stage is represented by the p values on each graph. Data are least-squares means ± SEM. Means with different letters differ (P < 0.05) as determined by Tukey's test.

3.4. Per2

Results are shown in Fig. 4. A slight increase in Per2 transcript abundance in the fetal brain was observed at E15.5 compared to E10.5-E11.5 and values remained constant thereafter until P14 with a subsequent increase at P28. For heart, there was a slight increase in Per2 transcript abundance at E16.5 and a further increase at P63. Abundance of Per2 transcripts in liver increased at E17.5 and again at P63. Amounts of Per2 transcripts in kidney were low until P14 with a further increase at P63.

Fig. 4.

Comparison of Per2 expression across stages of fetal and postnatal developmental stages for brain, heart, liver, and kidney. Gene expression is expressed as counts per million (CPM). The statistical effect of stage is represented by the p values on each graph. Data are least-squares means ± SEM. Means with different letters differ (P < 0.05) as determined by Tukey's test.

3.5. Cry1

Unlike for other genes, Cry1 transcript abundance did not increase during fetal development (Fig. 5). In fact, as compared to values at E10.5–11.5, there was a decline in Cry1 mRNA from E12.5 to 17.5 in brain, and from E12.5 to E16.5 in liver. Furthermore, transcript abundance did not increase at any point during postnatal life for any tissue.

Fig. 5.

Comparison of Cry1 expression across stages of fetal and postnatal developmental stages for brain, heart, liver, and kidney. Gene expression is expressed as counts per million (CPM). The statistical effect of stage is represented by the p values on each graph. Data are least-squares means ± SEM. Means with different letters differ (P < 0.05) as determined by Tukey's test.

3.6. Cry2

Transcript abundance of Cry2 remained nearly constant throughout fetal development with a slight increase at E18.5 in brain and at E17.5 in liver (Fig. 6). Thereafter, Cry2 transcript abundance increased at P14 and again at P63 in brain at and at P63 in heart, liver and kidney.

Fig. 6.

Comparison of Cry2 expression across stages of fetal and postnatal developmental stages for brain, heart, liver, and kidney. Gene expression is expressed as counts per million (CPM). The statistical effect of stage is represented by the p values on each graph. Data are least-squares means ± SEM. Means with different letters differ (P < 0.05) as determined by Tukey's test.

4. Discussion

The overall picture from examination of developmental changes in transcript abundance for the core clock genes in brain, heart, liver and kidney is one where changes in gene expression do not occur for the most part until late gestation or after birth. This study also supports the idea that the organization of the molecular clock is more advanced for the fetal brain and liver than for fetal heart and kidney.

In the brain, there was a small increase in Clock at E13.5 and a larger increase at E18.5. Similarly, Bmal1 transcript abundance increased slightly at E14.5 and to a greater degree at E17.5, Per2 increased slightly at E15.5 and remained constant until after birth and Cry2 increased slightly at E18.5. There are reports of rhythmicity of cultured SCN as early as between E13 and E15.5 [14,[33], [34], [35]]. Likewise, transcript abundance of most genes increased at the end of gestation in the liver, with increases observed for Clock at E18.5, and for Bmal1, Per2, and Cry2 at E17.5. The only genes whose transcript abundance increased before birth in heart was Clock (at E17.5), and Per2 (at E16.5). Similarly, the genes that experienced an increase in expression in the kidney during the fetal period were Clock (at E18.5), Bmal1 (at E15.5) and Per1 (E18.5).

The fact that tissues were collected at one or more unknown time points makes it impossible to identify when circadian rhythmicity is first established. Nonetheless, findings are consistent with other experiments indicating formation of a functional clock near the end of gestation for the SCN [5,14,24]. Despite the current observations that there were limited changes in expression of core clock genes during late gestation in heart and kidney, there is other experimental evidence for a functional clock in late gestation for both heart [27] and kidney [26,28]. Thus, it is possible that tissues can develop the capacity for a circadian clock in a tissue-specific manner prior to birth.

Based on changes in gene expression, another finding of the current study was that the molecular clock continues to mature after birth because there was an increase in transcript abundance for most genes at a time between P14 and P63. Transcript abundance in the postnatal period could have been affected by the light: dark cycle (the time of day of tissue collection was not reported) but, even with this caveat, the increase in transcript abundance at P14 or later for most genes is indicative of postnatal maturation of the clock mechanism. Similar results indicating maturation of the core clock mechanism during the postnatal period have been obtained for the SCN [25], [[33], [34], [35], [36]], heart [37], liver [38], and kidney [26].

The notion that an animal may have the capacity for a functional clock prior to birth has implications for fetal medicine, especially in the context of preterm birth. The molecular switch that initiates circadian clock transcription during development is unknown. Conceivably, light could regulate fetal circadian rhythms directly, via fetal perception of light. There is evidence for fetal perception of light in late gestation in mice [39] and humans [40]. Use of cycled lighting in neonatal intensive care units to mimic a diurnal environment improved outcomes such as increased weight gain, shortened hospital stay, sleep improvement, stabilized vital signs, and behavioral stress [[41], [42], [43], [44]]. Studies to examine if preterm infants have the capacity for a functional circadian clock mechanism and consequences of disruptions in the molecular clock for the neonate is warranted.

One gene that did not fit the general pattern of ontogeny was Cry1. During the fetal period, transcript abundance either did not change or, for brain and liver, went down from E12.5 until E17.5 or E18.5. Moreover, expression declined in the late postnatal period. Cry1 and Cry2 are both important for control of circadian rhythmicity, but it is necessary to knock out both genes to eliminate circadian rhythmicity [45]. Thus, it is likely that the molecular clock can function even without a developmental increase in Cry1 expression. One explanation for the decrease in transcript abundance for Cry1 in mid gestation or after birth would be changes in activity of regulatory systems regulating Cry1 expression. One candidate is the AMPK/mTOR pathway since CRY1 can be degraded by AMPK-mediated proteasomal degradation or mTOR-mediated autophagy [[45], [46], [47], [48], [49]].

One of the aspects of the design of sample collection was that variation existed in the relative number of male and female pups at various developmental stages and any variation due to dam was not controlled for. As a result, some of the observed variation in gene expression could reflect the influence of fetal sex or dam.

In conclusion, these results are consistent with the idea that the circadian clock mechanism develops during the late fetal period in a tissue-specific manner. Moreover, data support the concept that the clock mechanism is not fully established at birth but rather undergoes several weeks of maturation. It will be important to confirm these conclusions through experimentation to determine circadian oscillations in gene expression relative to key zeitgebers such as light-dark cycle and feeding schedule.

Data Statement

The dataset used in this study was previously deposited in the European Bioinformatics Institute Biostudies database under the accession E-MTAB-6798 (https://www.ebi.ac.uk/biostudies/arrayexpress/studies/E-MTAB-6798).

Author contributions

DLS, LZ, and PJH - conceptualization, investigation, formal analysis; DLS and LZ – methodology; DLS - visualization, writing original draft; PJH – supervision and funding acquisition; DLS, LZ, and PJH – writing, reviewing and editing.

Funding

Research was supported by the L.E. ‘Red’ Larson Endowment.

Declaration of competing interest

The authors declare there are no conflicts of interest.

Data availability

data are publically available.