News & Views | Published:

Cellular clockwork

Nature Genetics volume 32, pages 559560 (2002) | Download Citation


A new study identifies a cell-autonomous two-hour clock in diverse types of mammalian cells in culture. The two-hour oscillations depend on negative feedback of a gene product, Hes1, on transcription of its own gene—a regulatory mechanism that is identical to the circadian clock. Although a similar two-hour clock is required for vertebrate somitogenesis, the physiological role of the two-hour clock in cells remains unclear.

It is well established that rhythms of gene expression in pacemaker neurons regulate sleep–wake cycles in flies and mammals1. But these molecular circadian clocks are not limited to neurons, and even cultured fibroblast and hepatoma cell lines keep circadian time with oscillations in gene expression that measure increments of approximately 24 hours2. A study by Hiromi Hirata and colleagues recently published in Science3 has identified another molecular clock present in fibroblast, myoblast and neuroblastoma cell lines—a two-hour clock. Both the circadian and two-hour clocks in cultured cells are independent of the cell cycle, and they are probably independent of one another. The results of Hirata et al.3 provide new insight into the regulation of molecular clocks and raise once again the unresolved question of their purpose.

Negative feedback

The circadian and two-hour clocks both rely on a negative feedback mechanism in which gene products downregulate transcription of their own genes, resulting in rhythmic gene expression. Current models of the mammalian circadian clock1 involve rhythmic expression of the Period (Per) and Cryptochrome (Cry) gene families. Transcription of Per and Cry is activated by the Clock/Arntl (also called Bmal1 or Mop3) heterodimer binding the Per and Cry promoters. Accumulation of Per and Cry is delayed by roughly 4–6 hours relative to the respective RNAs, but ultimately Per/Cry heterodimers repress further Per and Cry transcription by binding Clock/Arntl and preventing transcriptional activation (see figure). Loss of function mutations in Clock, Arntl or the Per or Cry gene families stop the circadian clock.

Marking time. Conceptually similar molecular clocks. Models depicting the 24-hour circadian clock (left panel) and the two-hour clock (right panel) as negative feedback loops. Transcription of the repressors Per and Cry (left) and Hes1 (right) is activated by Clock/Arntl and an unidentified protein X, respectively. The repressor proteins feed back to repress transcription of the genes that encode them to close the loop. Regulated proteolysis of the repressors allows transcription to start again, thus beginning another cycle. Image: Katie Ris

A similar feedback loop exists for the two-hour clock described by Hirata et al.3. Their study was inspired by previous findings of rhythmic expression of genes such as that encoding the bHLH transcriptional repressor Hes1, and other members of the Notch signaling pathway, during the embryonic development of the somites, mesodermal precursors that give rise to skeleton and muscle (reviewed in ref. 4). A repeating wave of expression of these genes progresses from the back to the front of the pre-somitic mesoderm (PSM) every two hours, which parallels the appearance of new somites as they emerge from the anterior end of the PSM at two-hour intervals.

Shocking the clock

As was described for the fibroblast circadian clock2, Hirata et al.3 found that the two-hour clock could be started with a serum shock applied to mammalian cells in culture, and this induced up to six cycles of Hes1 expression. These oscillations had a period of two hours, identical to Hes1 RNA oscillations in the PSM. In vitro oscillations could also be induced in Notch-expressing cells by exposing them to cells expressing Delta, a Notch ligand. Again, this parallels the PSM clock, where the oscillating Hes1 expression requires activation of the Notch signaling pathway5.

Hes1 levels also oscillate with a two-hour period in serum-induced cells, but this protein oscillation is delayed by roughly 15 minutes relative to Hes1 RNA3. Two features of Hes1 make it ideal for implementing a two-hour oscillation: Hes1 RNA and Hes1 protein have short half-lives of approximately 20 minutes3, and Hes1 binds the Hes1 promoter and represses transcription6. Thus, Hes1 represses Hes1 transcription, but is rapidly degraded, allowing resumption of Hes1 transcription, re-synthesis of Hes1, and so on (see figure). Hirata et al.3 tested the requirement of Hes1 oscillations for Hes1 RNA rhythms by preventing Hes1 levels from oscillating. Constitutively high levels of Hes1 protein caused constitutively low levels of Hes1 RNA, and constitutively low levels of Hes1 protein caused constitutively high levels of Hes1 RNA.

The different periods of the circadian and two-hour clocks may reflect the different half-lives of the RNAs and proteins involved and the presence of additional rate-limiting steps in the circadian clock, such as the regulation of Per/Cry nuclear entry by protein phosphorylation (reviewed in ref. 1). Nevertheless, the findings of Hirata et al.3 suggest that molecular clocks share a common feature: they all use negative feedback loops of transcription and repression.

Protein X

In the circadian clock, Clock and Arntl RNA and protein levels also exhibit circadian oscillations. But Per and Cry RNAs and protein levels are rhythmic even in a mutant mouse with constitutively high Clock and Arntl levels7. Oscillating levels of the repressors Per and Cry seem to be required for circadian clock function, whereas oscillation of the activators Clock and Arntl is not. Hirata et al.3 speculate that sustained oscillation of Hes1 would require another cycling gene and its product. Because the circadian clock can oscillate when only the repressors oscillate, it is possible that either a cycling or a constitutive activator of Hes1 transcription constitutes the additional element necessary to sustain the two-hour clock. As activation of the Notch pathway can activate Hes1 transcription in vitro, and Notch signaling is required for cyclic expression of Hes1 in the PSM in vivo5, a member of the Notch signaling pathway may be the missing transcriptional activator X (see figure).

Can the Hes1 RNA oscillation explain the PSM clock and somitogenesis? Probably not. Although Hes1 homozygous-null mutant mice die perinatally, somitogenesis appears to occur normally8. But mice lacking Hes7, a close relative of Hes1, have severe defects in somitogenesis9. Like Hes1, expression of Hes7 is rhythmic in the PSM, and Hes7 can directly repress transcription from the Hes7 promoter10. Thus, Hes7 RNA and protein oscillations in the PSM probably rely on a negative feedback loop analogous to the Hes1 loop identified by Hirata et al.3 in cultured cells. Any model of the PSM clock must be rich enough to account for features such as the spatially oriented propagation of the two-hour repeating wave of gene expression—which suggests the involvement of intercellular communication, and a model may require incorporation of cyclic expression of other members of the Notch signaling pathway such as Lunatic fringe11.

Hirata et al.3 have convincingly established a molecular mechanism for a cell-autonomous two-hour clock, and the ability to study this question in cell culture is a great breakthrough for further mechanistic studies. It is yet to be determined, however, why diverse mammalian cell types have a two-hour molecular clock. Similarly, we do not understand why fibroblasts possess a cell-autonomous circadian clock. Further work is required to understand the physiological roles of these clocks, and how these clocks control downstream molecular and physiological processes.


  1. 1.

    & Nature Rev. Genet. 2, 702–715 (2001).

  2. 2.

    , & Cell 93, 929–937 (1998).

  3. 3.

    et al. Science 298, 840–843 (2002).

  4. 4.

    & Nature Rev. Genet. 2, 835–845 (2001).

  5. 5.

    et al. Development 127, 1421–1429 (2000).

  6. 6.

    et al. J. Biol. Chem. 269, 5150–5156 (1994).

  7. 7.

    et al. Cell 110, 251–260 (2002).

  8. 8.

    et al. Genes Dev. 9, 3136–3148 (1995).

  9. 9.

    et al. Genes Dev. 15, 2642–2647 (2001).

  10. 10.

    , , & Genes Cells 6, 175–185 (2001).

  11. 11.

    , , & Curr. Biol. 8, 979–982 (1998).

Download references

Author information


  1. Department of Biology, New York University, 100 Washington Square East, New York, New York 10003, USA. or

    • Michael N. Nitabach
    •  & Justin Blau


  1. Search for Michael N. Nitabach in:

  2. Search for Justin Blau in:

About this article

Publication history



Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing