---

We know how disrupted our days can be after a sleepless night or during jet lag while our internal body clock adjusts to a new time zone. Although we still do not understand the restorative power of sleep, we have a relatively good understanding of the internal clocks that drive rhythmic sleep/wake cycles. These clocks drive circadian rhythms of approximately 24 hours that persist even in constant darkness. Circadian clocks operate in many tissues apart from the brain, leading to daily rhythms that include changes in blood pressure and the timing of cell division (1).

The fruit fly Drosophila also has circadian rhythms of rest and activity. Because its rest period shows general characteristics of sleep, such as immobility and reduced responses to external stimuli (2, 3), Drosophila seems to be an excellent model for studying the regulation of sleep/wake cycles. Genetic screens in Drosophila led to the identification of the first circadian mutants (4) and the subsequent cloning of the first circadian clock gene from any species (5, 6). Different mutations in this gene, called period (per), caused short (19-hour) or long (29-hour) days or a complete loss of rhythms altogether (4). Thirty years later, human clock gene "mutants" were identified: An allele of a human Per gene was associated with behavioral rhythms of 23 hours, putting an affected individual's clock constantly ahead of local time (7). Circadian rhythmicity is probably the complex behavior that we understand the best at molecular and neuronal levels. But how does an internal clock drive behavior? And what does this clock consist of in the first place?

Although it was initially a surprise that mutations in a single gene could affect behavior so profoundly, the identification of per was only the first of many studies that helped describe the molecular clock that drives circadian rhythms in Drosophila. The initial clock genes identified form a single negative feedback loop: The transcription factors Clock and Cycle activate expression of the per and timeless (tim) genes, and Per and Tim proteins then feed back to inhibit Clock/Cycle activity, thereby further repressing per and tim expression (8). This generates molecular rhythms: The RNA and protein levels of per and tim oscillate, peaking once per day. Highly related genes function in the mammalian clock.

Because forward genetic screens for clock genes were reaching saturation, we initiated a screen for rhythmically expressed genes that would encode either novel clock components or outputs from the clock, linking intracellular rhythms in gene expression to whole-animal behavior. We identified vrille (9) and Par Domain Protein 1 (Pdp1) (10), which encode transcription factors with almost identical DNA binding domains. Like per and tim, vrille and Pdp1 are direct transcriptional targets of Clock and Cycle and are rhythmically expressed, although vrille RNA and protein levels peak a few hours before those of Pdp1. Genetic experiments revealed that increasing Vrille or decreasing Pdp1 levels slowed the clock, suggesting that Vrille and Pdp1 have opposite molecular functions. But what was their precise role? We found that Vrille directly represses and Pdp1 directly activates Clock transcription. Because the Clock protein itself activates vrille and Pdp1 transcription, we had identified the components of a second clock feedback loop that generates rhythmic Clock expression in antiphase to per, tim, vrille, and Pdp1 expression (10). This second feedback loop presumably makes the molecular clock more robust and probably regulates the transcription of output genes expressed just before dawn.

Approximately 100 neurons in the adult Drosophila brain have molecular clocks, and these drive highly predictable behavioral rhythms of flies for weeks in constant darkness. But how is the ticking of an intracellular clock in so few neurons translated into time-of-day information for an animal? Counterintuitively, our studies of inputs to the clock may help answer this question (11). It is well known that light resets the molecular clocks in the behaviorally relevant pacemaker neurons, with corresponding changes to the timing of behavioral rhythms (8). However, unlike most other neurons, there was little evidence that pacemaker neurons could transmit rapid signals. We found that circadian pacemaker neurons in Drosophila larvae not only receive light from the visual system to reset the clock, but are also an essential part of a neural circuit leading to rapid light avoidance. Larvae lacking pacemaker neurons were unable to avoid the light and were essentially as "blind" as larvae lacking photoreceptors. Furthermore, the molecular clock in the pacemaker neurons determines the visual sensitivity of larvae: Mutations in Clock or cycle, the transcriptional activators, made larvae supersensitive to light, whereas larvae mutant for the repressors (per or tim) had greatly reduced light sensitivity (11).

Most sensory neurons have their own internal clock that confers circadian rhythms in sensitivity. However, the larval visual system does not have a clock and relies on pacemaker neurons as a circadian filter for visual outputs. Although the ecological significance of our results is not completely clear, our results show that the transcriptional state of the clock in the pacemaker neurons determines how well they transmit an input signal. We (and others) have proposed that the membrane excitability of pacemaker neurons is determined by the internal clock and thus shows a daily rhythm (12-14). This would explain our larval behavioral data: The resting membrane potential would be relatively depolarized in a Clock or cycle mutant background, allowing the threshold for action-potential firing to be reached relatively easily; in contrast, the membrane would be hyperpolarized in a per or tim mutant, and a stronger input would be needed to reach the threshold for firing. Thus, the transcriptional rhythms driven by the molecular clock in pacemaker neurons presumably lead to 24-hour rhythms in neuronal firing rates even in constant darkness. Larval light avoidance offers a simple 15-min assay to identify mutants in pacemaker neuron physiology. Coupling this with a transcriptional profile of pacemaker neurons should allow us to find key clock outputs that are rhythmically expressed and to determine firing rates of clock neurons. Which loop will they be controlled by? Only time will tell.

References

1. 1. S. M. Reppert, D. R. Weaver, Nature 418, 935 (2002).
2. J. C. Hendricks et al., Neuron 25, 129 (2000).
3. P. J. Shaw, C. Cirelli, R. J. Greenspan, G. Tononi, Science 287, 1834 (2000).
4. R. J. Konopka, S. Benzer, Proc. Natl. Acad. Sci. U.S.A. 68, 2112 (1971).
5. T. A. Bargiello, M. W. Young, Proc. Natl. Acad. Sci. U.S.A. 81, 2142 (1984).
6. P. Reddy et al., Cell 38, 701 (1984).
7. K. L. Toh et al., Science 291, 1040 (2001).
8. R. Stanewsky, J. Neurobiol. 54, 111 (2003).
9. J. Blau, M. W. Young, Cell 99, 661 (1999).
10. S. A. Cyran et al., Cell 112, 329 (2003).
11. E. O. Mazzoni, C. Desplan, J. Blau, Neuron 45, 293 (2005).
12. M. N. Nitabach, J. Blau, T. C. Holmes, Cell 109, 485 (2002).
13. M. N. Nitabach, V. Sheeba, D. Vera, J. Blau, T. C. Holmes, J. Neurobiol. 62, 1 (2005).
14. C. M. Pennartz, M. T. de Jeu, N. P. Bos, J. Schaap, A. M. Geurtsen, Nature 416, 286 (2002).