.............................................................. Morning and evening peaks of activity rely on different clock neurons of the Drosophila brain Brigitte Grima*, Elisabeth Che ´ lot*, Ruohan Xia & Franc ¸ ois Rouyer Institut de Neurobiologie Alfred Fessard (NGI, CNRS UPR 2216), Centre National de la Recherche Scientifique, av. de la terrasse, 91198 Gif-sur-Yvette, France * These authors contributed equally to this work ............................................................................................................................................................................. In Drosophila, a ‘clock’ situated in the brain controls circadian rhythms of locomotor activity. This clock relies on several groups of neurons that express the Period (PER) protein, including the ventral lateral neurons (LN v s), which express the Pigment-dis- persing factor (PDF) neuropeptide, and the PDF-negative dorsal lateral neurons (LN d s) 1 . In normal cycles of day and night, adult flies exhibit morning and evening peaks of activity 1,2 ; however, the contribution of the different clock neurons to the rest– activity pattern remains unknown. Here, we have used targeted expression of PER to restore the clock function of specific subsets of lateral neurons in arrhythmic per 0 mutant flies. We show that PER expression restricted to the LN v s only restores the morning activity, whereas expression of PER in both the LN v s and LN d s also restores the evening activity. This provides the first neuronal bases for ‘morning’ and ‘evening’ oscillators in the Drosophila brain. Furthermore, we show that the LN v s alone can generate 24 h activity rhythms in constant darkness, indicating that the morning oscillator is sufficient to drive the circadian system. Many animals exhibit a bimodal activity distribution in the 24 h day/night cycle that is imposed by the rotation of the Earth 3 . Two circadian oscillators have been proposed to explain bimodal activity rhythms, and this theory has gained support from the finding of split free-running rhythms that have been reported in various organisms 2,4–6 . It was shown recently that different regions of the mammalian suprachiasmatic nucleus (SCN) control morning and evening electrophysiological oscillations in vitro 7 , but no specific neuronal groups dedicated to morning or evening behaviours have been reported so far. The morning and evening activity bouts displayed by Drosophila in 12/12 h light/dark (LD) conditions are controlled by the circadian clock, as inferred from the loss of anticipatory activity of both light transitions (lights-on morning transition and lights-off evening transition) in per 0 mutants 1 (Fig. 1). Rhythmic behaviour is con- trolled by the brain, in which six groups of PER-expressing neurons have been described in the adult, including three subsets of lateral neurons (LNs) and three subsets of dorsal neurons (DNs) 1,8 (Fig. 2a). In LD conditions, PER levels cycle similarly in these neuronal groups, showing a peak in the early morning and a trough at the end of the day 9,10 . The two subsets of LN v s (except one of the five small LN v s (s-LN v s)) express the PDF neuropeptide 11 (Fig. 2a), and their contribution to the rhythmic behaviour has been studied in pdf 0 mutants 12 ) and LN v -ablated flies 12,13 . Flies devoid of func- tional LN v s do not display lights-on anticipatory activity but show a robust phase-advanced evening peak 12,13 , as shown here for pdf 0 mutants (Fig. 1). These data indicate that PDF-expressing LN v s are required for the morning activity but are dispensable for the evening activity. In constant darkness (DD), both pdf 0 and LN v -ablated flies become largely arrhythmic (although a variable proportion of the flies display a weak, short period rhythmicity), indicating that PDF- expressing LN v s are required for robust 24 h free-running rhythms 12,13 . To understand the contribution of the different neur- onal groups to rhythmic behaviour, we undertook to express PER specifically in clock neuron subsets of per 0 flies and assay activity rhythms in LD and DD conditions. It has been shown that moderate levels of constitutive expression of per in per 0 flies can restore low- amplitude PER protein cycling in the eye 14 . More importantly, robust rhythmic behaviour could be restored in per 0 flies with pan-neuronal elav-Gal4-driven per expression 15 . We first verified that constitutive per expression in the brain clock neurons could restore a near-wild-type behaviour. A cryptochrome (cry)-Gal4 line, which we have previously described as showing Gal4-driven UAS- gfp expression in the six clock neuronal groups of the adult brain 16 , was used to drive a UAS-per transgene in per 0 flies. Indeed, UAS-per expression in cry-expressing neurons restores a complete LD beha- viour to per 0 flies, showing both morning and evening activity bouts (Fig. 1). We then tested the contribution of the PDF-expressing LN v s by driving per expression using a pdf-Gal4 transgene. Pdf- Gal4-driven UAS-per transcription restores PER protein expression specifically in the LN v s of per 0 flies, and reveals cycling of PER levels similar to the wild type, with high levels at Zeitgeber time (ZT) 0 and no staining at ZT12 (Fig. 2b; see also Supplementary Table S1). Flies with LN v -restricted PER exhibited a clear morning activity that anticipated the lights-on transition, but no anticipatory evening activity (Fig. 1). We then used the Mz520-Gal4 line (see Methods), whose expression in the adult brain is also restricted to the two subsets of PDF-expressing LN v s (Fig. 2c; see also Supplementary Table S1). Mz520-Gal4-driven per expression in a per 0 background shows PER cycling in the LN v s, with a ZT0 peak and a ZT12 trough. As expected from the PER expression results, these flies exhibited the same LD profile as flies with pdf-Gal4-driven per expression, characterized by the loss of the lights-off anticipation (Fig. 1). Therefore PER expression in the PDF-expressing LN v s is sufficient to restore the morning activity peak but not the evening activity peak. We then used two Gal4 lines expressed in various neuronal groups of the adult brain, including subsets of PER-expressing neurons. In both cases, Gal4-driven UAS-per expression in a per 0 background resulted in the accumulation of PER protein only in the clock-cell- restricted fraction of the Gal4 expression pattern (Fig. 2c, d; see also Supplementary Table S1). This is in agreement with the expression patterns of per-Gal4 transgenes being much broader than the protein expression pattern 8 . It indicates that PER protein is unable to accumulate, at least to detectable levels, in neurons that do not express it in the wild type. No restoration of morning or evening anticipatory activity was observed in flies expressing per under the control of the C929-Gal4 insertion (Fig. 1). C929-Gal4 expression includes the large LN v s (l-LN v s) as the only clock cells 17 , and results in accumulation of PER only in the l-LN v s in a per 0 background (Fig. 2b; see also Supplementary Table S1). This result indicates that the large PDF neurons do not contribute to the anticipation of light transitions or require the small ones to do so. The Mai179-Gal4 transgene is expressed in several groups of neurosecretory cells of the larval brain, including the PDF neurons 18 . In the adult brain, Mai179-Gal4 is expressed in the five s-LN v s (comprising four PDF-positive and one PDF-negative cell), a small number of l-LN v s and three–four of the six PDF-negative LN d s (Fig. 2d; see also Supplementary Table S1). As for the previous Gal4 lines, expression in the dorsal neurons was never observed. Driving a UAS-per transgene with Mai179-Gal4 in a per 0 background results in PER accumulation in the s-LN v s (occasionally in one l-LN v ) and LN d s, with high levels at ZT0 and no labelling at ZT12 (Fig. 2d; see also Supplementary Table S1). The comparison of Mai179-Gal4 with pdf-Gal4 and Mz520-Gal lines allows us to address the contribution of the PDF-negative LN d s (namely those expressing Mai179-Gal4) to the LD behaviour. A very different behaviour was observed in the Mai179-Gal4 (or pdf-Gal4/Mai179-Gal4) flies, which displayed a robust anticipation of both lights-on and lights-off transitions (Fig. 1). PER expression in the LN v s and LN d s therefore restores morning and evening activity bouts in letters to nature NATURE | VOL 431 | 14 OCTOBER 2004 | www.nature.com/nature 869 ©2004 Nature Publishing Group