Molecular Mechanisms of the Biological Clock in Cultured Fibroblasts

See allHide authors and affiliations

Science  13 Apr 2001:
Vol. 292, Issue 5515, pp. 278-281
DOI: 10.1126/science.1059542


In mammals, the central circadian pacemaker resides in the hypothalamic suprachiasmatic nucleus (SCN), but circadian oscillators also exist in peripheral tissues. Here, using wild-type andcryptochrome (mCry)–deficient cell lines derived from mCry mutant mice, we show that the peripheral oscillator in cultured fibroblasts is identical to the oscillator in the SCN in (i) temporal expression profiles of all known clock genes, (ii) the phase of the various mRNA rhythms (i.e., antiphase oscillation of Bmal1 and mPer genes), (iii) the delay between maximum mRNA levels and appearance of nuclear mPER1 and mPER2 protein, (iv) the inability to produce oscillations in the absence of functionalmCry genes, and (v) the control of period length by mCRY proteins.

In the mouse, the core oscillator of the master circadian clock in the SCN is composed of interacting positive and negative transcription-translation feedback loops (1–3), which involve three homologs of theDrosophila gene period (mPer1,mPer2, and mPer3), two cryptochrome genes (mCry1 and mCry2), and the transcriptional activator genes Clock and Bmal1 (1, 2,4). A key step in this feedback loop is the shutdown of CLOCK- and BMAL1-driven transcription by mCRY proteins (4). To keep pace with the solar day-night cycle, the master clock can be entrained by light received through photoreceptors in the retina (5). Molecular oscillators also exist in peripheral tissues, where they cycle with a 6- to 8-hour delay with respect to the central pacemaker (6–8). In contrast toDrosophila and zebrafish, mammalian peripheral clocks do not directly respond to light but are synchronized by the SCN by neuronal and/or humoral signals (9). In vitro, brief treatment of cultured cells with various compounds [serum, forskolin, 12-O-tetradecanoylphorbol 13-acetate (TPA), adenosine 3′,5′-monophosphate (cAMP), or dexamethasone] induces rhythmic expression of the clock genes Per1, Per2, andCry1 and the circadian transcription factor genedbp for two to three cycles (6, 10–12).

To investigate whether the molecular makeup of the peripheral oscillator in cultured fibroblasts resembles that of the core oscillator in the SCN, we determined the expression profiles of all known clock genes in cultured rat-1 fibroblasts over a period of 3 days (13). To trigger the oscillations, we used the vasocontracting peptide endothelin-1 (ET-1) (14), which activates the protein kinase C–mitogen-activated protein kinase cascade and cAMP response element–binding protein (CREB) phosphorylation (15). This treatment induces a rapid, robust increase in Per1 and Per2 gene expression, followed by a sharp reduction in corresponding mRNA levels and subsequent synchronous cycling of Per1, Per2,Per3, and dbp mRNAs (Fig. 1) (16, 17). Also, robust cycling of Bmal1 mRNA was observed, with mRNA levels accumulating antiphase to Per and dbp mRNA cycles. Clock mRNA levels were constant at all time points examined. In addition, Cry1 expression showed rhythmicity, peaking 4 to 8 hours after Per mRNAs (16). These data demonstrate that ET-1 can induce circadian gene expression in cultured rat-1 cells and that the temporal expression patterns of Per, Bmal1,Cry1, and dbp genes (all rhythmically expressed) as well as the Clock gene (constitutively expressed) match those in the SCN (1, 18). Casein kinase Iɛ(CKIɛ) and Cry2 genes did not show apparent rhythmic expression in rat-1 cells, a finding consistent with the observation that in the SCN CKIɛ is constitutively expressed (19) and cycling of mouse Cry2 is weak (18) or not detectable (20).

Figure 1

Temporal expression profiles of clock genes in rat-1 fibroblasts after ET-1 treatment: Quantification of temporal changes in Per2, Bmal1, andClock mRNAs. Basal levels of each mRNA (at time point 0) were arbitrarily set to 100. Results shown are means ± SEM;n = number of experiments.

Next, we analyzed by immunocytochemistry the PER1 and PER2 protein expression profiles in these cells. Nuclear staining occurred 26 to 28 hours after treatment, indicating that mPER1 and mPER2 protein cycles follow the rhythm of Per1 and Per2 mRNA expression with a 4- to 8-hour delay (Fig. 2), as in the SCN (21). In addition, pronounced PER1 and PER2 nuclear staining was found 1.5 hours after ET-1 treatment (Fig. 2) (22), suggesting that ET-1 causes rapid synthesis of PER1 and PER2 and translocation of these proteins into the nucleus. This nuclear PER2 may up-regulate Bmal1 expression and down-regulate Per gene expression 4 hours after ET-1 treatment (Fig. 1) (16). The latter effect may also involve the CRY proteins.

Figure 2

Temporal PER1 and PER2 protein expression profiles in ET-1–treated rat-1 fibroblasts and comparison with corresponding mRNA expression profiles. (A) Immunofluorescence showing accumulation of PER1 and PER2 proteins in nuclei of ET-1–treated rat-1 fibroblasts. (B) Percentages of cells positive for antibodies to PER1 and PER2 (counted in 100 to 200 DAPI-stained nuclei) at the indicated times. For comparison, relative mRNA levels of Per1 and Per2 are shown as broken lines. Results shown are means ± SEM (three independent experiments).

We next used spontaneously immortalized (mutant) mouse embryonic fibroblasts (MEFs) from wild-type andmCry1 –/– mCry2 –/– mice to test the role of mCry genes in the fibroblast clock (23, 24). Treatment of wild-type MEFs with ET-1 resulted in a temporary induction of mPer1 gene expression within 1 hour, followed by synchronous cycling of mPer1 anddbp mRNA (Fig. 3A) (17). Four hours after stimulation, increasedBmal1 mRNA levels were observed (Fig. 3A), which most likely requires synthesis and nuclear translocation of the mPER2 protein and subsequent rhythmic expression of Bmal1 mRNA antiphase tomPer1 and dbp. Thus, as in rat-1 cells, ET-1 can induce circadian gene expression in MEFs. In marked contrast, ET-1 treatment ofmCry1 –/– mCry2 –/– MEFs did not result in rhythmic expression of mPer1,Bmal1, or dbp genes (Fig. 3, B and C). Instead, as in the SCN,mCry1 –/– mCry2 –/– MEFs showed continuously accumulating mPer1 mRNA and low levels of Bmal1 mRNA, respectively. The absence of mCRY proteins also resulted in constant high expression of the dbp gene. BecausemCry1 –/– mCry2 –/– MEFs express ET-A receptor mRNA (25), the lack of rhythmic gene expression in these cells is unlikely to result from improper activation of signal transduction pathways, but rather is caused by the absence of mCRY proteins. Interestingly,mCry1 –/– mCry2 –/–cells retain the ability to respond to ET-1 treatment or a serum shock with instantaneous induction of mPer1 and mPer2gene expression (16). Thus, as in the SCN and peripheral tissues in intact animals, mCry genes are indispensable for generation of molecular rhythm in stimulated cultured mouse fibroblasts.

Figure 3

Clock gene expression in ET-1–stimulated wild-type and mCry mutant MEFs. (A and B) Temporal mRNA expression patterns formPer1, dbp, and Bmal1 after ET-1 treatment of wild-type MEFs (A) andmCry1 –/– mCry2 –/– MEFs (B), as determined by Northern blot analysis. (C) Quantification of temporal changes in mPer1, dbp, and Bmal1 mRNAs in andmCry1 –/– mCry2 –/– (red lines) and wild-type cells. Data shown were confirmed in two independent wild-type MEF lines and three independentmCry1 –/– mCry2 –/– MEF lines, respectively. Basal levels of each mRNA (at time point 0) were arbitrarily set to 100. (D and E) Temporaldbp mRNA expression pattern inmCry1 –/– and mCry2 –/–MEFs after the stimulation. Asterisks indicate peaks of rhythmically expressed dbp mRNA. (F) Quantification of temporal changes in dbp mRNA inmCry1 –/– and mCry2 –/–cells. Results shown are means ± SEM (n = 3). Peak levels of dbp mRNA were arbitrarily set to 100.

To investigate whether the periodicity of peripheral clocks is an intrinsic property of the peripheral oscillator or whether it is instigated by cues from the SCN, we have measured temporal expression patterns of the dbp gene in immortalized MEFs frommCry1 and mCry2 single-mutant mice, known for their short (τ = 22.5 hours) and long (τ = 24.6 hours) free-running periodicity of locomotor activity, respectively (18,20, 23). The periodicity of dbp mRNA oscillation inmCry1 –/– MEFs, although weak, is about 2 to 4 hours shorter than in mCry2 –/– MEFs (Fig. 3, D and E). This indicates that mCRY-mediated control over the pace of biological clockwork is not restricted to the central pacemaker in the SCN, but holds for circadian oscillators in any mammalian tissue.

Finally, we measured DBP protein oscillation patterns in serum shock–stimulated MEFs. Robust oscillation of nuclearly localized DBP was observed in wild-type andmCry2 –/– MEFs (Fig. 4) (26). InmCry1 –/– cells, nuclear DBP levels remained high after a brief initial nadir. This finding not only confirms the unexpected pattern of dbp gene expression in these cells but also emphasizes the weakness of mCry2-mediated oscillations. As expected on the basis of constant high levels of dbpmRNA, nuclei ofmCry1 –/– mCry2 –/–cells were positive at any time. For all cell lines tested, the appearance of nuclear DBP largely coincided with the (ET-1–mediated)dbp mRNA expression profile. These findings suggest that, as in the SCN, DBP is rapidly synthesized and translocated into the nucleus (27).

Figure 4

Temporal DBP protein expression profile in wild-type and mCry mutant MEFs. (A) Immunofluorescence study showing nuclear DBP protein (green) at 12-hour intervals after stimulation of wild-type and mCry1mutant cells with 50% horse serum. As an internal control, cells were stained for the nonoscillating nuclear p62 protein, a component of the basal transcription factor TFIIH (red nuclei). (B) Percentages of nuclei positive for antibody to DBP (counted in 150 to 200 nuclei positive for p62 mAb), measured at 4-hour intervals. Time points were analyzed in a blind fashion, and the results were confirmed in three independent experiments (and in two independent wild-type MEF lines).

Taken together, our data indicate that the molecular makeup of the peripheral circadian oscillator in cultured fibroblasts is similar to that of the master oscillator in the SCN. The same set of circadian genes is assembled into positive and negative transcription-translation feedback loops. The mRNA expression profiles for these circadian genes display an “SCN-like” temporal expression profile as well as phase relationship, and, at least for mPER1, mPER2, and DBP, the delay between onset of transcription and nuclear appearance of the corresponding gene product is comparable to that in the SCN. Moreover, the homozygous inactivation of one or both mCry genes—known to accelerate, retard, or even abolish the biological clock in the SCN (18, 20, 23)—affects the peripheral oscillator to a similar extent. Thus, the peripheral oscillator in immortalized cultured fibroblasts constitutes a bona fide in vitro model for the molecular oscillator in the SCN, and could potentially allow the use of skin fibroblasts as a means of identifying clock gene defects in patients with circadian disorders.

Although peripheral clocks in the intact mouse possess some degree of autonomy, as is evident from the uncoupling of entrainment of peripheral and master clocks by glucocorticoid administration or restricted feeding (6–8), they differ from the master clock in the SCN in one important aspect. Unlike in cultured SCN slices, rhythmic clock gene expression in cultured peripheral organs/tissues and fibroblasts is dampened after a number of days (9). Because, as we have shown, the molecular makeup of the core oscillator of master and peripheral clocks is identical, the mechanism that allows the master clock to keep on ticking remains to be identified.

  • * To whom correspondence should be addressed. E-mail: okamurah{at}


View Abstract

Navigate This Article