J Comp Physiol A (1998) 182: 435 ± 453
Ó Springer-Verlag 1998
ORIGINAL PAPER
C. Helfrich-FoÈrster
Robust circadian rhythmicity of Drosophila melanogaster requires the presence of lateral neurons: a brain-behavioral study of disconnected mutants Accepted: 17 September 1997
Abstract Mutations at the disconnected (disco) locus of Drosophila melanogaster disrupt neural cell patterning in the visual system, leading to the loss of many optic lobe neurons. Drosophila's presumptive circadian pacemaker neurons ± the dorsal and ventral lateral neurons ± are usually among the missing cells, and most disco ¯ies are behaviorally arrhythmic. In this study, I show that ventral lateral neurons (LNvs) are occasionally present and provoke robust circadian rhythmicity in disco mutants. Of 357 individual disco ¯ies four animals with robust circadian rhythmicity were found. All four retained LNvs together with terminals in the superior protocerebrum. Residual or bi-circadian rhythmicity was found in about 20% of all ¯ies; the remaining ¯ies were completely arrhythmic. One of the ¯ies with residual rhythmicity and two of the arrhythmic ¯ies also had some LNvs stained. However, these ¯ies lacked the LNv ®bers in the superior protocerebrum. The results suggest that the presence of single LNvs is sucient to provoke robust circadian rhythmicity in locomotor activity if the LNv terminals reach the superior protocerebrum. The presence of residual or bi-circadian rhythmicity in 20% of the ¯ies without LNvs indicates that also other cells contribute to the rhythmic control of locomotor activity. Key words Circadian pacemakers á Clocks á Insects á Locomotor activity á Pigment-dispersing hormone Abbreviations DD continuous constant darkness á disco disconnected á DN dorsal neuron á LD light-dark cycle á LN lateral neuron á LNd, dorsal lateral neuron á LNv ventral lateral neuron á per period gene á PER period protein á PDH pigment-dispersing hormone á tim timeless gene á TIM timeless protein C. Helfrich-FoÈrster Botanisches Institut, Auf der Morgenstelle 1, D-72076 TuÈbingen, Germany Tel.: +49-7071-2976162; Fax: +49-7071-296155 e-mail:
[email protected]
Introduction Internal circadian clocks organize the temporal order of physiological states and activity levels in most organisms. They comprise circadian pacemakers which generate an endogenous rhythm of about 24 h, entrainment pathways which synchronize the internal clock to the environmental light-dark cycles (LDs), and output pathways to eector organs. An important step towards understanding the functioning of circadian clocks in multicellular organisms is to identify the pacemaker cells and their input and output pathways. In Drosophila, the period (per) and timeless (tim) genes and their gene products ± period protein (PER) and timeless protein (TIM) ± are essential components of the circadian clock (reviews Hall 1995, 1996; Hardin and Siwicki 1995; Seghal et al. 1996; Rosbash et al. 1996). PER and TIM show daily oscillations in their abundance. With the help of an in vivo luciferase assay, these oscillations were recently demonstrated for the amount of PER even in living individual ¯ies under free-running conditions (Brandes et al. 1996; Plautz et al. 1997; Stanewsky et al. 1997). The observed oscillations stem from the entity of per-expressing cells. Immunocytochemical studies have shown that PER is located in several types of cells of the brain: in the photoreceptor cells, many glia cells and in two dierent groups of neurons, namely the lateral neurons (LNs) and the dorsal neurons (DNs) (Siwicki et al. 1988; Zerr et al. 1990). Not all per-expressing cells are equally important for behavioral rhythmicity (review: HelfrichFoÈrster 1996). An accumulating body of evidence points to the LNs as the site of the circadian pacemakers that control activity. Some evidence comes from developmental studies: Drosophila's clock appears to work from the ®rst instar larvae onward (Brett 1955; Seghal et al. 1992). Of the per-expressing cells only some LNs are present at that early developmental time (Helfrich-FoÈrster 1997a; Kaneko et al. 1997) and only these show a persistent cycling
436
in the amount of PER throughout metamorphosis (Kaneko et al. 1997). The most compelling evidence for the LNs as circadian pacemakers comes from studies involving genetic mosaics (Ewer et al. 1992) and transgenic ¯ies (Frisch et al. 1994) which express per only in subsets of the PER-positive cells. Robust rhythmicity could be observed only in ¯ies with per expression in the LNs. Further evidence comes from studies of anatomical brain mutants (Helfrich 1986; Helfrich-FoÈrster and Homberg 1993), the most notable example of which is disconnected (disco) (Dushay et al. 1989; Zerr et al. 1990; Hardin et al. 1992). Under freerunning conditions disco mutants are virtually arrhythmic for both eclosion and locomotor activity (Dushay et al. 1989). Due to a failure in optic lobe development, many neurons degenerate and are absent in disco adults (Steller et al. 1987; Tix et al. 1989; Lee et al. 1991). The LNs are apparently among the missing neurons, because they could not be stained by anti-PER, whereas all other per-expressing cells were normally revealed in disco (Zerr et al. 1990). Although this suggests a causal relationship between arrhythmicity and absence of the LNs in disco mutants, it has yet to be veri®ed. Furthermore, it is unknown whether all LNs are equally important for rhythmic activity, and how many LNs are necessary to provoke rhythmicity. The present study addresses these questions. Two clusters of LNs can be distinguished: the dorsal LNs (LNds) and the ventral LNs (LNvs) (Ewer et al. 1992; Frisch et al. 1994). The LNvs are also marked by an antiserum against the crustacean pigment-dispersing hormone (PDH) (Helfrich-FoÈrster 1995) and have been morphologically characterized in detail with the help of this antiserum (Helfrich-FoÈrster and Homberg 1993). In most disco mutants the LNvs are completely absent (Helfrich-FoÈrster and Homberg 1993; Helfrich-FoÈrster 1997a). However, in rare cases single LNvs were detected by PDH-immunocytochemistry (Helfrich-FoÈrster and Engelmann 1995), and occasionally rhythmic disco mutants were found (Dushay et al. 1989). A possible correlation between rhythmicity and presence of LNvs was investigated by testing the rhythmic behavior of many disco mutants and performing PDH-immunocytochemistry on their brains later. As shown in the present study, disco ¯ies with robust circadian rhythmicity had immunostained LNvs projecting into the superior protocerebrum. This supports the idea that the LNvs are pacemaker neurons, and indicates that a single LNv is sucient to drive the rhythm of locomotor activity in the animal.
Materials and methods Strains The disco stock tested was w1118 disco1 (Steller et al. 1987). The rhythms of white-eyed ¯ies are indistinguishable from those of redeyed ¯ies (Dushay et al. 1989), and the white-eyed stock w1118 served as controls for activity recordings. The ¯ies were raised on
standard cornmeal-agar-molasses-yeast medium at 20 °C under a 12:12 LD. Disco phenotype Disco mutations have a severe defect in the optic lobe development (Steller et al. 1987; Tix et al. 1989; Lee et al. 1991). The compound eyes of these mutants are usually disconnected from the optic lobes. As a result, mutants have optic lobes with severe morphological abnormalities (Steller et al. 1987). Two dierent phenotypes exist. Most mutants have no apparent connections between the retinula cells of the compound eye and the optic lobe. In these ¯ies only a tiny rudiment of the optic lobe is present, the brain shows the ``unconnected'' phenotype (see left brain hemisphere in Fig. 2B). In a minority of ¯ies photoreceptor axons innervate the optic lobes (``connected'' phenotype). In these cases the optic lobes have almost normal size, but are still grossly disorganized (see right brain hemisphere in Fig. 2B). The LNs are usually absent in both the ``connected'' and ``unconnected'' phenotype of disco: they could not be revealed with anti-PER (Zerr et al. 1990) or with anti-PDH (Helfrich-FoÈrster and Homberg 1993; Helfrich-FoÈrster 1997a), whereas other PER- or PDH-immunoreactive cells are marked normally in both disco phenotypes. Recording of locomotor activity rhythms The recording device used for monitoring locomotor activity rhythms of the ¯ies is shown in Fig. 1. The ¯ies were individually con®ned to small photometer cuvettes. To record activity an infrared light beam passed through the rear end of each cuvette, whereas food and water was available at the front end. Due to this separation, activity related to food intake or other small movements did not contribute to the recordings; only genuine running activity was monitored. An interruption of the infrared light beam by a ¯y produced a signal which was recorded by a microprocessor. A computer program monitored whether the ¯y had been active in a 4-min time span (i.e., whether it had moved into or out of the light beam). Activity resulted in a scan value of 1, no activity in a value of 0. Fifteen such activity scans were obtained per hour for each ¯y. A more detailed description of the system is given by Engelmann et al. (1996). Experimental design Male and female ¯ies (ratio 1:1) were recorded at 20 1 °C. They were ®rst monitored for about 5 days in a 12:12 LD. Light intensity during the light phase was 1000 lx. The lights were then shut o and the ¯ies were kept under a dim red light provided by a red ¯uorescence tube (Philips TL 20W/25A 032) with primary red cinemoid ®lter (Rank Strand). Light intensity was 6 á 10)8 W cm)2). This light could not be seen by the ¯ies and had no in¯uence on locomotor activity rhythms, but facilitated weekly controls of water supply. The ¯ies were recorded for 15±20 days under continuous constant darkness (DD) and then sacri®ced in order to perform PDH-immunohistology on their brains (see below). disco ¯ies that died before day 15 and whose brains could consequently not be immunostained were discarded from the analysis. Flies whose brains were damaged during the immunohistological procedure were also discarded. In the ®rst experiments, 64 ¯ies (48 disco and 16 w1118 controls) could be monitored simultaneously; later the recording device was expanded and allowed 128 simultaneous recordings (96 disco and 32 w1118 controls). During one experimental run some data were lost due to a recording failure while the ¯ies were under LD. Thus the entrainment to the LD could not be judged unequivocally for 54 disco ¯ies and 26 controls. Nevertheless, their freerun behavior was analyzed and is included in the data.
437
Analysis of circadian rhythms At the end of an experiment the raw data from each recorded ¯y were transformed to actograms and linear time-activity plots by special programs (J. Schuster, W. Hellrung, unpublished). These allowed visual judgment of the activity pattern of individual ¯ies. Whereas actograms displayed the raw, un®ltered activity data of each ¯y, linear time-activity plots showed the data smoothed by a non-recursive digital ®lter in the second canonical form (Lacroix 1980) to make the raw signal more compact and to facilitate the visual judgment of entrainment and freerun. The behavior of the ¯ies under LD and DD was analyzed by periodogram analysis combined with a v2 test with 5% signi®cance level (Sokolove and Bushell 1978, modi®ed by A. Diez-Noguera) on the raw data. The periodogram shows the amplitude ( power) of periodicities in the time series for all periods of interest (in most cases 4- to 40-h periods in 10-min steps; in some cases periods from 14 h to 50 h were analyzed). In order to compare the powers of dierent ¯ies, powers were expressed in percent according to the calculated 5% signi®cance level. The periodograms were calculated over all days in LD or DD, if not stated otherwise. During LD, a ¯y was regarded as well entrained when its phase relationship to the LD was rather stable and periodogram analysis showed a 24-h peak with a power exceeding 50%. It was regarded to be weakly entrained when periodogram analysis revealed a signi®cant peak at 24 h that was 50% or smaller. A ¯y was regarded as not synchronized when no signi®cant 24-h peak occurred in the periodogram. To compare the activity pattern of dierent entrained ¯ies under LD, an average day was calculated for each ¯y. To compare the LD behavior of dierent ¯y groups, the individual average days of several ¯ies were combined to an overall average day (cf. Fig. 4). During DD, a ¯y was regarded as ``simple and robust rhythmic'', when visual inspection revealed a rather stable period throughout the recording interval and the periodogram showed a discrete de®nable peak with a power exceeding 20% (width of peak >1 h, cf. Figs. 2A, 7). The rhythm was de®ned as circadian when
Fig. 1 Recording device for monitoring locomotor activity rhythms of Drosophila melanogaster. The ¯ies were kept separately and visually isolated from each other in photometer cuvettes. The cuvettes had a large hole covered with a ®ne net in order to allow circulation of air, and a small hole on the opposite side to permit access to water. Water was provided by a strip of absorbent sponge the ends of which were placed in water containers. A small piece of sugar in each cuvette served as food. The open ends of eight cuvettes were closed by an envelope of Plexiglas (only shown for the undermost cuvette row). The closed ends were inserted into the recording device in such a way that infrared light beams passed through each cuvette (see upper row; LED infrared light emitting diodes; PT phototransistors). Four cuvette rows were placed above each other and formed one recording unit consisting of 32 channels. Each recording unit was electrically shielded by a metal case. The phototransistors were connected to a comparator type of interface card which processed the interruptions of the infrared light beams. Eight such comparators (stemming from two recording units) were connected to a multiplexer (MP), and two such MPs to a peripheral processor unit (PPU). The activity scans of all ¯ies were stored in the PPU and were transferred hourly to a host computer (PC) where they were saved on a diskette. A radio transmitter-based clock served as a time reference its period ranged between 17 h and 34 h, and as bi-circadian when the period had values between 35 h and 50 h. A ¯y was classi®ed as ``complex rhythmic'' when the periodogram revealed a signi®cant but unstable or weak period (power 20% or lower, width of peak >1 h) when rhythmicity was only visible for few cycles, or when several signi®cant peaks were revealed by periodogram analysis (cf. Fig. 7). It was classi®ed as arrhythmic when no rhythmicity could be revealed by visual inspection and no signi®cant period in the circadian range was detected by periodogram analysis or when there were several narrow ``spikes'' that just reached or barely exceeded the 5% level (power
