Independent Photoreceptive Circadian Clocks Throughout Drosophila

See allHide authors and affiliations

Science  28 Nov 1997:
Vol. 278, Issue 5343, pp. 1632-1635
DOI: 10.1126/science.278.5343.1632


Transgenic Drosophila that expressed either luciferase or green fluorescent protein driven from the promoter of the clock geneperiod were used to monitor the circadian clock in explanted head, thorax, and abdominal tissues. The tissues (including sensory bristles in the leg and wing) showed rhythmic bioluminescence, and the rhythms could be reset by light. The photoreceptive properties of the explanted tissues indicate that unidentified photoreceptors are likely to contribute to photic signal transduction to the clock. These results show that autonomous circadian oscillators are present throughout the body, and they suggest that individual cells in Drosophilaare capable of supporting their own independent clocks.

Circadian oscillators have been localized in several organisms. For example, the suprachiasmatic nucleus (SCN) is important for mammalian rhythms (1), whereas Iguana iguana has at least three independent oscillators: the retina, parietal eye, and pineal gland (2). Sparrows show activity rhythms that can be altered by lesioning the pineal gland (3); this operation reveals the influence of other oscillators on the bird's behavior. The brain controls behavioral rhythms in moth (4) andDrosophila (5, 6), whereas sperm release in the moth is controlled by an independent oscillator (7). Recently, free-running rhythms have been demonstrated in theDrosophila ring gland (8) and Malpighian tubules (9). At a molecular level, two differentDrosophila clock genes, period (per) (10) and timeless (tim) (11), have been identified. The transcripts and proteins from both of these genes cycle daily in abundance, and both genes are needed to maintain a biological clock (12).

In mammals, all demonstrated clock input comes from the eye (13), although the exact photoreceptor is unknown (14). Photoreceptors within the brain can mediate rhythms in birds (15) and insects (4, 16).Drosophila that lack all known photoreceptive organs can still transmit light information to the clock (17), indicating the existence of unidentified circadian photoreceptors within the animal.

Functional per in the head of the fly has been explicitly shown to be essential for one output: behavior (6). Eclosion (10) and transcriptional control (18) have been shown to require per, but without any specific localization of the message or its protein. PER protein has been detected in many tissues throughout the animal (19). Although specific functions outside the head are unknown, per in these tissues may be needed for localized clock function. Theseper-dependent oscillators could be targets for signals from the head; however, light sensitivity is a characteristic of a fully autonomous oscillator that can receive stimuli from the environment, transmit this information to the oscillator, then use the oscillator to affect downstream clock-controlled functions.

per-driven bioluminescent oscillations occur in livingper-luc Drosophila (per is fused to the luciferase gene luc) (20-22). These rhythms are entrainable by light and free-run in constant darkness (20). To examine the circadian autonomy of Drosophila tissues, we monitored rhythmic bioluminescence from cultured dissociated body segments (head, thorax, or abdomen) from per-luc animals (23, 24); per-driven green fluorescent protein (GFP) was used concurrently as a bright spatial expression marker.

Each of the three segments are capable of rhythmic bioluminescence (Fig. 1) in light-dark (LD) conditions. Changing the conditions to constant darkness (DD) resulted in a gradual decrease in amplitude. The cultures were able to reentrain to a new LD cycle, where the new onset of light occurred 6 hours later than the free-running subjective dawn. Reentrainment occurred within one cycle, with the main bioluminescent peak falling about 20 hours after lights-on, just as it did in the initial LD cycle. The waveform and phase of the rhythms from all three segments were nearly identical, and there was very little noise in the individual traces, especially when compared with whole-animal records (20-22). Also, there was no evidence of the second peak of bioluminescence that was previously reported in whole-animal studies (20, 21), indicating that this feature likely arises from a whole-animal physiological-bioluminescent phenomenon rather than as a direct feature of per transcription.The proboscis and antenna (Fig.2A) expressed per-driven GFP (23, 25); the bright green signal was easily distinguishable from the yellow autofluorescence [compare (26)]. Like the whole-body segments, the proboscis (Fig.2B) and antennae (Fig. 2C) are also capable of maintaining oscillations in LD and DD (although, as in whole-body segments, rhythmicity damps in DD). GFP expression was also detected at the near-single-cell level in legs (Fig. 2D) and wings (Figs. 2F and 3B). GFP was present in these tissues in chemosensory cells (Fig. 3) (27), which are also found in the proboscis and antennae (27). The leg (Fig. 2E) and wing oscillators (Fig. 2F) showed rhythmic per-driven bioluminescence. As with body segments, the tissue oscillators are capable of rapid resetting in response to light. Other tissues throughout the body that displayed rhythmicity include the eyes, Malpighian tubules, and testes (28); the ovaries, though, did not display any appreciable cycling (28, 29).

Figure 1

Bioluminescence rhythms in cultured body segments. per-driven GFP expression can be seen throughout the whole fly (A). Individual (B) heads, (C) thoraxes, and (D) abdomens were individually cultured in LD and DD and monitored for bioluminescence expression levels. Of the cultured segments 79% (221/279) demonstrated rhythms in LD; 59% (130/222) demonstrated at least two cycles in DD, and 82% (182/222) became arrhythmic within four cycles in DD. Filled bars, darkness; open bars, light; gray bars subjective light. CPS, counts per second.

Figure 2

Bioluminescent rhythms in distinct tissues. Expression of per-driven GFP by (A) the proboscis (blue arrow) and antennae (white arrow). (B) Single probosci and (C) antennae were cultured individually and were rhythmic in LD and DD. (D) Expression of per-driven GFP in the leg. (E) Rhythmic per-driven bioluminescence in the leg. (F) Expression of per-driven GFP in the whole wing. (G) Rhythmic per-driven bioluminescence in the wing: 94% (159/169) of the cultures exhibited rhythms in LD; 82% (137/167) persisted for at least two cycles in DD; and 18% (30/166) became arrhythmic within four cycles in DD. Filled bars, darkness; open bars, light; gray bars subjective light.

Figure 3

GFP expression in wing chemosensory cells. (A) Detail of an anterior wing margin. Red arrows show individual chemosensory cells. (B) Superimposition ofper-driven GFP fluorescence on the same wing margin. Signal co-localizes with the basal cells of the sensory bristles, as well as fluorescence from chemosensory cells on the opposite side of the wing margin (blue arrows).

Our imaging and analysis of per-gal4; UAS-GFP flies showed broad per expression throughout the fly (Fig. 1A), supporting previous reports of widespread per expression (19). Several lines of evidence confirm that our fluorescent images (Figs. 1A, 2A, 2D, and 3B) do not represent ectopic expression: multiple staining studies with per-lacZ fusions and antibodies to PER have shown per expression similar to that shown here (19, 30); our bioluminescent studies in culture show that all of the examined tissues that express fluorescence also express cycling per-driven bioluminescence (Figs. to 3); and previous single-photon imaging (20) [reconfirmed in recent bioluminescence imaging (28)] showedper-luc bioluminescence in the legs and wings, as well as other parts of the animal. Although the fluorescence pattern does not exactly reflect that of native per expression (as assayed immunohistochemically), inconsistencies with this line occur when fluorescence is not seen where per expression has otherwise been detected (notably the eye; see Fig. 1A). We have at no time observed external GFP fluorescence where per has not been otherwise detected.

Because there are numerous oscillators in the fly, each of these oscillators are photoreceptive, and per is expressed in single cells, we hypothesize that the Drosophila clock can operate at a cell-autonomous level. Each clock cell may be capable of photoreceptivity and endogenous rhythm maintenance to some extent. Independent clock cells within a circadian tissue can also explain the overall arrhythmicity of a tissue after several days in DD. Whole animals free-run with a range of near-circadian, but different, periods [as monitored by bioluminescence (21)]. In LD conditions, single cells are resynchronized every 12 hours by light transitions; however, cells free-running with a range of periods similar to whole animals will mathematically yield net arrhythmicity in a matter of days. Single-cell monitoring of cells in a tissue as it becomes arrhythmic will be necessary to tell whether the arrhythmicity is due to cell asynchronicity [compare (31)], or the gradual “winding-down” of the clock itself at a cellular level.

In constant conditions, RNA oscillations in the adult body damp in DD with about the same kinetics reported here, but that rhythmicity persists in the head (29). The brain is the source of control for locomotor behavior (5, 6), and such behavior oscillates for weeks in constant conditions (32). Although the brain rhythms do not apparently damp over several days in constant conditions, other parts of the head do (the proboscis, eyes, and antennae; Fig. 2, B and C). The asynchrony within these tissues, which are on the surface of the head, most likely masks the synchronized brain rhythmicity deep within the head.

The function or functions of per outside the head are unknown (19). A particularly interesting feature of the GFP expression pattern is its labeling of chemosensory cells. This pattern specifically identifies structures at the base of chemoreceptor bristles in the proboscis, antennae, anterior wing margins, and legs (Figs. 1A, 2A, 2D, 2F, and 3) (27). Moreover,per-driven bioluminescence in these tissues is rhythmic (Fig. 2, B, C, E, and G), showing that there is a functional clock in these cells. These cells are spatially independent of each other (on the basis of noncontiguous fluorescence) and are capable of cycling and entraining without an attached head. The presence of a functional clock implies circadian regulation of chemosensory sensitivity, analogous to the circadian regulation of sensitivity thresholds in luminance (33) and pain (34) reported in mammals. Although the evidence for clock control of sensory thresholds in the fly is still circumstantial, the presence of independent clocks along with examples of similar rhythmic phenomena in other systems indicates a central role for per-dependent clock functions in tissues outside the head.

Every known oscillating tissue in the fly has shown the capacity for light perception. Also, each dissociated segment is rhythmic with the same phase and waveform. This raises the possibility that the head, which was previously believed to be the master oscillator in the fly, does not coordinate all rhythms throughout the animal. In this case, light, which has the potential to affect all parts of the fly simultaneously, serves as the master coordination signal. The asynchrony of independent clock cells over several days of free-run should not be of practical concern to the whole animal because flies in the wild almost always have an environmental light cycle. This control mechanism could reasonably be extended to other animals (although it will likely be more complicated in higher eukaryotes). The mouse circadian gene, Clock, is expressed throughout the animal (35), indicating that mammals may have oscillators throughout the body; also, a mammalian homolog of Drosophila per has been recently identified (36) and found to be localized throughout the body. Like the fly, much of the evidence for a central oscillator in mammals has come from observing a single output: behavior. Similarly, the simple interpretation of a master controlling oscillator may need to be revised with the closer examination of multiple outputs and isolated multiple oscillators.


View Abstract

Navigate This Article