Defining the role of Drosophila lateral neurons in the control of circadian rhythms in motor activity and eclosion by targeted genetic ablation and PERIOD protein overexpression

Fly strains

Drosophila cultures were maintained on a 12-h light–dark cycle on corn meal–yeast–agar medium at 25 °C and 50% relative humidity. All the fly lines carried the white mutation on the X chromosome. The alleles of the X-linked genes disconnected and period were disco¹ and per⁰¹ (associated with the forked marker for the former). To make the upstream activating sequence (UAS)-per constructs, a per cDNA of 3.9 kb beginning at the SpeI site (46 bp upstream from the ATG) and ending at position 4230 in the mRNA (134 bp upstream of the HindIII site in the 3′ untranslated region (UTR) was cloned into the pUAST vector by XhoI and XbaI sites. Two independent lines were obtained by microinjection and used to generate five additional lines by P-element mobilization, including the third chromosomal lines UAS-per16 and UAS-per19. The UAS-bax construct used to ablate cells by activating cell death is described elsewhere (Gaumer et al., 2000). Briefly, an EcoRI-XbaI fragment of murine bax cDNA containing the full-length coding region was ligated into pUAST, the resulting plasmid was microinjected and several independent insertions were obtained, comprising the UAS-bax1 insertion on the third chromosome that was used in this study. The UAS-TeTxLC (TeTxLC, tetanus toxin light chain) construct allows GAL4-induced synthesis of the tetanus toxin light chain which blocks synaptobrevin-dependent neurotransmitter release (Sweeney et al., 1995). Two lines of the active toxin (tntG, tntH) that are strongly activated by gal4 drivers and a line bearing an insertion coding for a mutant inactive protein (tntV1-A) were used (all second chromosomal lines), and have been described previously (Sweeney et al., 1995). A UAS-lacZ reporter was used in our initial disco¹ studies, but was then replaced by a UAS-gfp (GFP-A65T) for better sensitivity (see text). The two lines carry insertions on the second chromosome and were provided by the Bloomington stock centre.

Behavioural analyses

Behavioural analyses were carried out in LD 12 : 12 h conditions and in constant darkness (DD) at 20 °C. Temperature was controlled via a Temperatel processor (SLEL, France), reducing fluctuations around the set point to < 0.3 °C, in both light-on and light-off conditions. Light was provided by cool white fluorescent lamps. The locomotor activity of young males (1–7 days) was followed with Drosophila activity monitors (Trikinetics Inc., Waltham, USA). For LD 12 : 12 analysis, flies were left in LD for 1–2 days before starting to collect data. For DD analysis, flies were first entrained in LD 12 : 12 for 2–4 days and recorded in DD for 6–8 days. LD and DD average activity plots were generated according to Hamblen-Coyle et al. (1992) and Wheeler et al. (1993). Circadian rhythmicity in DD was tested by χ² periodogram analysis as described in Hamblen et al. (1986). Only flies with a period of between 15 and 32 h (a range that includes all known rhythmic clock mutants with altered periods) and a power (height of the periodogram peak above the 5% significance threshold) > 20 (with filter on) were considered circadian rhythmic. To assess the rhythmicity of eclosion, ≈ 15 males and 15 females were introduced into 150-mL vials, with a separable bottom part containing standard medium. Eggs were laid for 10–11 days at 20 °C in LD 12 : 12 conditions. When the top part of the vials was covered with pupae, the adults were discarded and the food-containing bottom of the vials was replaced with a clean empty one. Lights were switched off when the first dark pupae appeared. Newly eclosed flies were emptied from the vials every 4 h, under red safelight illumination, for 3–4 days. For each time point, the number of eclosed flies was calculated with the formula n= sample weight/average fly weight (average fly weight = 1 mg), because pilot experiments had shown that the eclosion curves obtained by counting and by weighing the flies were indistinguishable.

Immunocytochemistry and GFP fluorescence

All experiments were performed on whole-mounted adult or larval brains. CNSs from third instar larvae and adults were dissected into phosphate-buffered saline (PBS), fixed in 4% formaldehyde in PBS for 2 h at room temperature (RT) and washed for 6 × 10 min in OX-PBS (PBS and Thimerosal) with 0.3% Triton X-100 (OX-PBS-T). Tissues were pretreated with OX-PBS with 1% Triton X-100 for 10 min at RT and blocked in OX-PBS-T containing 10% normal goat serum (NGS; Sigma #G9023) for 1 h at RT. Then they were incubated at 4 °C with rabbit antifull-length PER serum (Stanewsky et al., 1997) that had been adsorbed against per⁰¹ adult acetone powder, at 1 : 15 000 in OX-PBS-T containing 5% NGS (OX-PBS-T-N) for at least 72 h, rabbit anticrab β-PDH serum (Dircksen et al., 1987) at 1 : 5000 in OX-PBS-T-N for 48 h, or anti-TeTxLC monoclonal antibody (J. Thierer & H. Niemann, unpublished data) at 1 : 2000 in OX-PBS-T-N for 48 h. After washing for 6 × 10 min in OX-PBS-T, samples were labelled with Texas red-conjugated goat antirabbit IgG antibody (Cappel #55675; Costa Mesa, USA) at 1 : 1000 in OX-PBS-T-N for 3 h at RT or overnight at 4 °C. They were washed for 6 × 10 min in OX-PBS, then mounted in Mowiol. To detect GFP expression, CNSs were dissected into saline buffer, fixed, washed and mounted. Preparations were examined immediately using an epifluorescence microscope (Zeiss Axioplan2) and a cooled digital camera (Diagnostic Instruments SPOT2, USA). Fluorescence intensity was quantified from digital images using the following formula: I = 100 × (S − B)/B, where S is the average fluorescence intensity (which was never saturating on a 0–255 scale) of the labelled cells and B is the average intensity in regions adjacent to the cells. It thus represents the percentage fluorescence above background.

Immunoblotting

Protein extracts were made from ≈ 20 heads for each time-point. Frozen heads were homogenized in 100 µL ice-cold extraction buffer in mm: Tris pH 7.5, 50; NaCl, 100; NaF, 50; EDTA, 5; [4-(2-aminoethyl)-benzenesulfonylfluoride (AEBSF), 1; plus Triton X-100, 1%; aprotinine, 1%; chymostatin, pepstatin, leupeptin and antipapain, 1 µg/mL each]. After centrifugation, supernatant fractions containing 50 µg of proteins were mixed with SDS sample buffer, boiled, and analysed by protein immunoblot as previously described (Edery et al., 1994). The following modifications were made: rabbit anti-PER antibody was adsorbed against per⁰¹ adult head acetone powder and incubated with a per⁰¹ head immunoblot before being used at a 1 : 10 000 dilution. Anti-PER was detected with an antirabbit IgG horseradish peroxydase-conjugated antibody (Amersham) diluted 1 : 3000.

The gal1118 enhancer trap line is mostly expressed in the PDF-expressing lateral neurons of the central brain

The gal1118 line was generated in the course of a P-gal4 enhancer-trap screen aimed at isolating genes expressed in the central brain (Boquet et al., 1999). The gal1118 insertion is located in cytological position 82D-E on the third chromosome and does not correspond to any known gene involved in circadian rhythms. Flies carrying one or two copies of the gal1118 insertion display a 1-h lengthening of their circadian period (Table 1). The period change is lost by excision of the P element (A. Lamouroux & F. Rouyer, unpublished results), indicating that it is a consequence of the insertion, although it is not known whether it is due to the disruption of the genomic sequence or to the production of large amounts of GAL4 protein that may alter the functioning of the gal1118-expressing cells.

gal1118 expression was studied in the adult brain. Using UAS-gfp or UAS-lacZ as reporters, no difference could be detected between expression at Z9 and ZT22 (ZT, Zeitgeber time) in a wild-type background nor between wild-type and per⁰¹ or tim⁰¹ mutants (data not shown). In w;UAS-gfp/+;gal1118/+ flies, the gal1118 expression pattern closely resembles the PDF one (Helfrich-Förster & Homberg, 1993; Helfrich-Förster, 1997), namely two small sets of cell bodies in the lateral brain with projections into the optic lobe medulla, toward the superior protocerebrum and through the posterior optic tract (Fig. 1A and B). However, gal1118 is not expressed in the few PDF-expressing cells of the tritocerebrum (PDF-Tri in Fig. 1B) nor in the PDF neurons of the abdominal ganglion (PDF-Ab, not shown). gal1118 cells are observed at the surface of the medulla in young adults (gme for gal1118 medulla, Fig. 1A), but the signal strongly decreases in the ageing fly (not shown).

In the lateral brain of w;UAS-gfp/+;gal1118/+ flies, gal1118‘w;UAS-gfp;gal118/+’ expression is detected in 3–8 small cells (s-LN_v-like in Table 2) and 3–6 large cells (l-LN_v-like in Table 2) (Fig. 1A, C and F and Table 2). Although the number of s-LN_v-like cells is often difficult to estimate accurately because of the presence of numerous gal1118-labelled neuronal processes in this region, it is larger on average than the number of PDF s-LN_vs (5 vs. 3.5, Table 2) whereas the number of l-LN_v-like cells is similar to the number of PDF l-LN_vs (4 vs. 3.8, Table 2; see also Helfrich-Förster, 1997). To identify these different cells, double-labelling experiments were performed with GFP fluorescence and PDF (Fig. 1C–E) or PER (Fig. 1F–H) immunocytochemistry. These experiments show that gal1118 is strongly expressed in the PDF-positive and PER-positive s- and l-LN_vs (Table 2). In agreement with more gal1118- than PDF-expressing s-LN_v-like cells, double labelings show that gal1118 is also expressed (although less strongly) in 1–4 s-LN_v-like cells that do not express PDF, whereas no cells staining for PDF only could be observed (Fig. 1C–E and Table 2). gal1118 cells that do not express PER were also identified (Fig. 1F–H and Table 2) and because there are no PER-negative PDF-positive cells (Helfrich-Förster, 1995), we assume that the PER-negative gal1118 cells are the same cells as the PDF-negative gal1118 cells. These additional s-LN_v-like cells revealed by gal1118 send projections that follow the s-LN_vs dorsal projections but stop shortly to arborize without reaching the dorsal brain (Fig. 1K–M), whereas each of the projections going further dorsally is both GFP- and PDF-labelled in w;UAS-gfp/+;gal1118/+ brains (Fig. 1K–M, and data not shown). A single PER-positive LN_v-like cell (PER-only cell in Table 2 and Fig. 1F–H), which could be the same cell as the PDF-negative PER-expressing fifth LN described in the larval brain (Kaneko et al., 1997), does not express gal1118. This cell accounts for the excess of PER- vs. PDF-expressing cells in the s-LN_v-like group (4.2 vs. 3.5 in Table 2).

As shown in Fig. 1G–H and Table 2, other PER-expressing cells of the lateral brain are the 5–8 LN_ds (Ewer et al., 1992) that do not express PDF (Helfrich-Förster, 1995). These cells appear to be gal1118-negative in most of the w;UAS-gfp/+;gal1118/+ brains as shown in Fig. 1A and F. However, some brains show very weak gal1118 expression in ≈ 50% of the LN_ds as illustrated in Fig. 1I and J and Table 2 (see also below). No gal1118 expression could be detected outside the brain on sections of adult heads (using UAS-lacZ) or whole white pupae (using UAS-gfp) although weak expression could be easily missed in the whole pupae sections (data not shown).

In order to detect cells weakly expressing gal1118, we also analysed gal1118 cells in flies carrying two copies of the gal1118 and UAS-gfp insertions (w;UAS-gfp;gal1118). In addition to the strong LN_v labelling described above, w;UAS-gfp;gal1118 brains reveal the weak gal1118 expression in the PER-expressing LN_ds (Fig. 2A, C–E). Several other groups of cells weakly expressing gal1118 can be detected in the homozygous w;UAS-gfp;gal1118 flies. Ten to fifteen cells termed gd1 (for gal1118 dorsal 1) are located in the same region as the previously reported PER-expressing DN1 and DN2 neurons, close to the ending of the LN_vs dorsal projections (Fig. 2B and F and Kaneko et al., 1997; Kaneko, 1998). Double labelings indicate that some of the cells in the gd1 group are PER-expressing DN1s, whereas two slightly more ventral PER-positive cells are likely to be the PER-expressing DN2s (Fig. 2F–H). A pair of gal1118 cells, termed gd2, are located laterally to the gd1 cells not far from the PER-expressing DN3 cluster (Fig. 2B and F and Kaneko et al., 1997; Kaneko, 1998), but do not express PER (Fig. 2F–H). The two other characterized groups of gal1118 cells are a small group of large-cell-body neurons located at the surface of the central dorsal brain and projecting into the pars intercerebralis (gpi), and ≈ 20 cells located underneath the calyces of the mushroom bodies (gmbc). These two groups of cells were found to display variable levels of gal1118 expression but did not express PER (not shown; see also Kaneko et al., 1997; Kaneko, 1998) nor PDF (see Fig. 1B and Helfrich-Förster & Homberg, 1993; Helfrich-Förster, 1997), as several other cells scattered through the brain (Fig. 2A and B and data not shown).

In the brain of w;UAS-gfp/+;gal1118/+ third instar larvae, gal1118 strongly labelled four dorsally projecting neurons in each brain hemisphere (Fig. 3A and Table 2), that appear similar to the four PDF-expressing neurons (larval LNs) (Fig. 3B, Table 2) previously identified in the larval brain (Helfrich-Förster, 1997). Weak staining was also often observed in two dorsolateral cells termed gd2^L (Fig. 3A and C). As in the adult, there was no gal1118 expression in the PDF-Ab neurons (Helfrich-Förster, 1997) at the tip of the ventral ganglion (Fig. 3A and B). Double labelling experiments showed that the cells strongly expressing gal1118 are the PDF larval LNs (Fig. 3C–E) that persist through metamorphosis to become the adult s-LN_vs (Helfrich-Förster, 1997). Homozygous w;UAS-gfp;gal1118 larval brains were also analysed to detect cells with weaker gal1118 expression. In addition to the four larval PDF-expressing LNs, four other PDF-negative gal1118 cells were detected in each hemisphere of the homozygous brains (Fig. 3F), that include the two gd2^L cells faintly detected in heterozygous brains. A single neuron (gd1^L) located dorsally to the LNs axon terminals sends projections into the mid-brain that finally go to the corpora cardiaca of the ring gland (CC) through the interhemispheric region (Fig. 3F, J and K). The two gd2^L neurons also project to the CC through the same path (Fig. 3F, J and K) and look very much like the adult gd2 cells (Fig. 2B and F). A fourth gal1118 neuron (named EH; see below) is located close to the interhemispheric region and sends projections toward the dorsal brain and to the CC (Fig. 3F, J and K). Double labelling experiments with anti-eclosion hormone (EH) antibody (Fig. 3I) show that this cell is the previously described EH-synthesizing neuron (McNabb et al., 1997). Interestingly, the dorsal projections of the EH neuron arborize close to the LNs dorsal projection terminals (Fig. 3I and K). Double labelling with anti-PER antibody showed that besides the LNs, none of the larval gal1118 neurons (namely gd1^L, gd2^L and EH) expressed PER (Fig. 3F–H).

Lateral neurons are present in the majority of adult disco¹ flies but fail to express PER

Lack of PER and PDF staining in previous studies has shown that PER and PDF do not accumulate in the brains of the vast majority of disco¹ flies (Zerr et al., 1990; Helfrich-Förster & Homberg, 1993; Helfrich-Förster, 1998). We used the strong gal1118 expression in the PDF-expressing LN_vs to test whether those neurons were still present in the disco¹ brain. The study was initiated with a UAS-lacZ reporter gene. No LN-like labelling was detected in any disco¹ larval brain (in the presence of one copy each of gal1118 and UAS-lacZ, data not shown), and 62% of the adult brains displayed stained cells (e.g. Figure 4C and D). Using a UAS-gfp reporter resulted in much greater sensitivity in both larvae (Fig. 4A and B) and adults (compare Fig. 4C and D with Fig. 4E and F), and the latter reporter was used for cell characterization and counting.

In the larva, 10/13 wdisco¹f brains carrying one copy of gal1118 and either one or two copies of UAS-gfp displayed one or more putative LNs (Table 3; see Fig. 4 legend for the identification of putative larval LNs). In contrast, neither PDF (10/10 brains) nor PER (20/20 brains at ZT21) staining was ever detected in the putative LNs of disco¹ larval brains (data not shown). In the adult, 83% of the disco¹ brains displayed at least one putative LN (Table 3; see Fig. 4 legend for the identification of putative adult LNs). These putative LNs were not labelled with the anti-PDF antiserum, which on the other hand clearly revealed the PDF-Tri neurons (7/7 brains, data not shown), as previously reported in earlier studies (Helfrich-Förster & Homberg, 1993; Helfrich-Förster, 1998).

Only 34 of 101 behaviourally tested disco¹ flies (34%), carrying one or two copies of the gal1118 insertion, were classified as rhythmic, with a mean period 2.5–3.7 h shorter than the corresponding controls (Table 1). Two of these rhythmic flies had a clearly distinct phenotype, with a robust circadian rhythmicity indistinguishable from wild-type, whereas the remaining 32 rhythmic flies had much weaker (compare powers) short-period rhythms (Table 3). In LD 12 : 12, the evening activity peak of the disco¹ flies was advanced compared to the controls (Fig. 5A and D), consistent with a fast-running clock. That peak was observed even for the individuals which were subsequently arrhythmic in DD (not shown), as described previously (Dushay et al., 1989; Hardin et al., 1992; Wheeler et al., 1993; Helfrich-Förster, 1998). The distribution of the average activity over a 24-h window shifts from nonrandom during the first 2 days in DD (Fig. 5E) to random during the next two days (Fig. 5F), indicating that the disco¹ flies transiently keep a synchronized rhythmic behaviour.

Samples of the tested disco¹ individuals belonging to different behavioural categories were then compared for brain anatomy. We found the same proportion of brains without dorsally projecting gal1118-positive neurons among arrhythmic and weak short-period rhythmic disco¹ flies (Table 3 and Fig. 4E and F). In contrast, such dorsally projecting neurons were seen in both of the two flies with robust rhythmicity (see Table 3). One of them was obviously of the connected phenotype (Steller et al., 1987), with close-to-wild-type anatomy (Table 3) but reduced gal1118-driven GFP intensity (not shown). The other one was a UAS-lacZ-bearing fly of the unconnected phenotype, whose contracted gal1118 expression pattern nevertheless included dorsal projections (Fig. 4D).

We wondered whether the short-period residual rhythmicity of the disco¹ flies in constant darkness could be related to PER cycling in some central brain neurons. Although gal1118-positive putative LNs are observed in the majority of disco¹ brains (see above), no PER was ever detected at circadian time (CT8; 16 brains) or CT22 (18 brains) in those flies as illustrated in Fig. 4G–I. In contrast, PER was detected at CT22 but not at CT8 in one or two groups of dorsal neurons, at locations very similar to those found in the wild-type and with a similar staining difference between the two time points (compare Fig. 4G and H with Fig. 6E and F). This indicates that PER is cycling in a circadian manner in the DNs of disco¹ flies.

Genetic ablation of the PDF LN_vs strongly alters activity rhythms but does not prevent PER cycling in other PER-expressing neurons

The gal1118 line was used to express the pro-apoptosis gene bax in the lateral neurons. bax is a mammalian gene that also induces apoptosis in Drosophila (Gaumer et al., 2000). gal1118-driven bax expression resulted in 85% lethality, mostly at embryonic stages, and half of the surviving adults had shrivelled wings. The finding of ≈ 70% lethality in flies ablated for the EH neurons (McNabb et al., 1997) suggests that the killing of these cells in the w;;gal1118/UAS-bax flies is likely to explain most of this effect, but other gal1118-expressing neurons may also be involved. Surviving individuals with expanded or shrivelled wings did not show major differences in their viability, at least up to 10–15 days of age, and were used for both the anatomical and behavioural analyses.

To control for the effect of ablation, we analysed the brain of flies carrying a UAS-gfp transgene in addition to the UAS-bax one (w;UAS-gfp/+;gal1118/UAS-bax). No gross brain rearrangement was observed either in larval or adult brains. In third instar larvae, a complete ablation of gal1118 cells was observed (Fig. 6A and Table 4). In agreement with the ablation of the four PDF- and gal1118-expressing LNs, anti-PER antibody staining at CT22 revealed a single PER-expressing cell in the LN region (Fig. 6C and Table 4), whereas five PER-expressing LNs were observed in the wild-type larval brain (Fig. 6D; see also Fig. 3 and Table 2). As expected from the nonoverlapping gal1118- and PER-expression patterns in the larval dorsal brain (Fig. 3F–H), PER-expressing cells were detected in 4/5 brains (see Fig. 6C).

In adult brains, we observed the complete disappearance of the gal1118-driven GFP expression in the LN_vs region (Fig. 6B and Table 4), whereas gal1118 cells at the periphery of the medulla were observed in 16/33 brains (Fig. 6B). In a minority of the brains, very few GFP-positive cells were observed in the region of the LN_ds (Table 4). PER expression was then characterized in the brains of flies with and without gal1118 cells. As illustrated in Fig. 6F, control brains contain several groups of neurons expressing PER at CT22: the LN_vs and LN_ds in the lateral region and the DNs in the dorsal protocerebrum (Siwicki et al., 1988; Zerr et al., 1990; Ewer et al., 1992; Frisch et al., 1994). The latter group has been subdivided into three subgroups: the central DN1, two slightly more ventral DN2 cells and the lateral DN3 cluster (Kaneko et al., 1997; Kaneko, 1998). We could not readily distinguish the DN2 cells among the larger DN1 group in many brains, but the more lateral DN3 group could easily be identified and analysed separately from the DN1 + DN2 set. In w;UAS-gfp/+;gal1118/UAS-bax brains, as well, four groups of PER-expressing cells were detected at CT22 (Fig. 6H and J). In agreement with the presence of a single gal1118-negative PER-positive LN_v-like cell and with the excess of PER-labelled vs. GFP-labelled LN_ds in w;UAS-gfp/+;gal1118/+ brains (see Fig. 1 and Table 2), a single PER-expressing LN_v-like and two or three PER-expressing LN_ds were labelled in w;UAS-gfp/+;gal1118/UAS-bax brains (Fig. 6H and Table 4). In the dorsal brain, large groups of central DNs and laterally located DN3 were found (Fig. 6J and Table 4), as expected from the only partial overlap between gal1118 dorsal cells (gd1) and PER-expressing DN1 + DN2 group and from the absence of gal1118 expression in the PER-expressing DN3 (see Fig. 2F–H).

PER temporal expression was analysed in constant darkness by comparing PER staining at CT9 (Fig. 6E, G and I) and CT22 (Fig. 6F, H and J). In control brains, PER cycles in the LNs (both LN_vs and LN_ds) as previously reported (Zerr et al., 1990), but also in the central DNs and in the DN3 group with a phase similar to the one of the LNs (Fig. 6E and F; see also Fig. 9E). In the w;UAS-gfp/+;gal1118/UAS-bax brains, PER levels were found to cycle with a phase similar to the wild-type in all of the nonablated PER-expressing neurons: the LN_ds and the single LN_v-like cell as well as the DN1 + DN2 and DN3 (Fig. 6G–J and Table 4).

The locomotor activity rhythm of the w;;gal1118/UAS-baxadult flies was tested in DD and 83% of the flies were found to be arrhythmic (Table 1). Similarly to disco¹, the rhythmic flies showed a shortening of the period, with a mean value of 22.5 h (± 0.4 h) and a low power (Table 1). The genetic ablation of the PDF-expressing LN_vs with the hid and reaper apoptosis genes of Drosophila has been recently reported and led to similar findings (Renn et al., 1999). In LD 12 : 12 conditions, the w;;gal1118/UAS-bax flies showed a clear anticipation of the light-to-dark transition with a shift of the evening activity peak ≈ 2 h earlier than the controls (compare Fig. 5G with Fig. 5A and J). As for disco¹, a clear lights-off anticipation was detected in LD for the majority of the flies, regardless of their behaviour in DD (not shown). Conversely to the disco¹ (Fig. 5E) and pdf⁰¹ (Renn et al., 1999) flies, no clear rhythm was displayed in average activity plots during the first two days in DD (Fig. 5H, I, and Fig. 5B, C, K and L for the controls).

Gal1118-driven PER overexpression abolishes activity and eclosion rhythms

The above results indicate that the absence of PER (and PDF) expression in the LNs of the disco¹ flies or the ablation of the PDF-expressing LN_vs in the w;;gal1118/UAS-bax flies lead to an advanced evening peak in LD and a mostly arrhythmic locomotor activity in DD with a weak short-period rhythm for a minority of the individuals. We wondered whether similar effects would be produced by blocking PER oscillations with saturating levels of PER in gal1118 flies carrying a UAS-per cDNA construct.

Two transgenic lines bearing independent UAS-per insertions were generated and tested for their rhythmic behaviour under DD and LD conditions with one or two copies of the gal1118 driver (Table 1). gal1118-driven per expression suppressed activity rhythms in a PER dose-dependent manner, as the number of arrhythmic flies strongly increases between heterozygous w;;gal1118UAS-per/++ and homozygous w;;gal1118 UAS-per. For both UAS-per16 and UAS-per19 insertions, almost complete arrhythmicity was observed in homozygous w;;gal1118UAS-per (Table 1). In heterozygous w;;gal1118 UAS-per/++ individuals, a clear effect is seen with the UAS-per19 insertion whereas almost no effect was observed with the UAS-per16 insertion, which has lower levels of PER in the gal1118 cells (not shown). In contrast to the disco¹ and w;;gal1118/UAS-bax flies, no short-period weak rhythm could be detected in the PER-overexpressing lines. However, anticipation of the lights-off transition could be observed in LD 12 : 12 conditions, even for the strongly arrhythmic w;;gal1118UAS-per16 (Fig. 7G) and w;gal1118 UAS-per19 (Fig. 7M) homozygous genotypes, as opposed to the complete absence of such anticipation in the arrhythmic per⁰¹ mutant (Fig. 7S), as previously described (Wheeler et al., 1993). The anticipation of lights-off is rather poor with the UAS-per19 insertion but very robust with the lower PER-expressing UAS-per16. Conversely to the disco¹ and LN_v-ablated flies, no phase advance was observed for the anticipation of lights-off in LD (Fig. 7G and M). In the same line, no weak short-period rhythm could be detected in constant darkness (Table 1), even during the first days in DD (Fig. 7H, I and N, O).

Eclosion rhythms were also analysed for the two lines overexpressing PER under gal1118 control. Arrhythmic eclosion was observed for both heterozygous w;;gal1118 UAS-per19/++ and homozygous w;;gal1118 UAS-per19 (Fig. 8). Homozygous w;;gal1118 UAS-per16 were also arrhythmic whereas heterozygous w;;gal1118 UAS-per16/++ were not (not shown), in agreement with the lower GAL4-dependent PER expression observed with the UAS-per16 insertion. As for activity, the differences between heterozygous and homozygous flies for the UAS-per16 insertion suggest that the suppression of the eclosion rhythm depends upon PER dose.

PER levels oscillate in the DN3 dorsal neurons and the eye photoreceptors of arrhythmic flies overexpressing PER under gal1118 control

We performed an analysis of PER expression in the brains of the behaviourally arrhythmic w;;gal1118 UAS-per16 flies in DD to verify that PER cycling would be abolished in the gal1118-expressing neurons and to test whether changes of PER expression could be detected in the other PER-expressing cells. In agreement with the behavioural results, high and constant PER expression was detected in the LN_vs (Fig. 9C–E), whereas PER levels cycled in the control flies (Fig. 9E). As expected from the gal1118-driven GFP expression pattern in the homozygous w;UAS-gfp;gal1118 genotype (see Fig. 2), lower level noncycling PER expression seems to occur in the LN_ds (Fig. 9C and D), and in dorsal cells (gd1) overlapping with the DN1 + DN2 PER-expressing cells (Fig. 9A and B). The proximity and therefore frequent overlap between the strongly PER-expressing l-LN_vs and the weakly expressing LN_ds prevented us from unambiguously quantifying PER in the LN_ds. Although strong PER oscillations were observed in the DN1 + DN2 group of the control brains (Fig. 9E), the overlap between the DN1 + DN2 group and the gd1 cells also prevented us from unambiguously quantifying PER in this region. However, PER expression could be analysed in the DN3 and showed strong oscillations in both the w;;gal1118 UAS-per16 brains and the controls (Fig. 9A, B and E). In contrast, larval DN3 do not show PER cycling (Kaneko et al., 1997), suggesting that larval and adult DN3s are different cells or change their PER expression during metamorphosis.

PER oscillations have been shown to occur in the eyes of the arrhythmic disco¹ mutant (Zerr et al., 1990; Hardin et al., 1992) and in the photoreceptors of per⁰¹ flies carrying a per construct that constitutively express per mRNA in these cells only, indicating that the eye photoreceptors contain an autonomous oscillator (Cheng & Hardin, 1998). Furthermore, the constitutive overexpression of per in the eye photoreceptors has no effects on the rhythmic behaviour and PER oscillations in a per⁺ brain (Zeng et al., 1994). To test further the independence of the eye clock with respect to the brain, we have analysed the expression of PER that emanates mainly from the eye photoreceptors in the arrhythmic w;;gal1118UAS-per16 flies, which have strong noncycling PER expression in the LN_vs (see above). Anti-PER antibody staining on head sections showed that the eye photoreceptors represent the major contribution to head PER expression in these flies, as is the case for the wild-type flies (data not shown), allowing us to use head protein extracts to analyse PER expression from the eye. Western blots of heads were made from w;;gal1118 UAS-per16 and control w;;UAS-per16 flies in DD conditions and showed that PER cycled with similar overall levels and a peak around CT18 in both genotypes (Fig. 10).

Expression of the tetanus toxin light chain in the LN_vs does not alter rhythmic behaviour

Knowing that the LNs are involved in both activity and eclosion rhythms, we wondered whether gal1118-driven expression of the tetanus toxin light chain, which should block the synaptic output from the LNs, might alter rhythmicity. Flies bearing gal1118 and two different UAS-tnt insertions known to give strong GAL4-dependent expression (Sweeney et al., 1995) were generated. They were verified for the presence of the tetanus toxin light chain in the gal1118 neurons by immunohistochemistry using anti-TeTxLC antibody (not shown). Activity rhythms of flies expressing the TeTxLC under gal1118 control were tested in both LD and DD conditions. In either conditions, flies carrying one copy of gal1118and one copy of UAS-tntG or UAS-tntH appear to be unaffected compared to w;;gal1118/+controls (Table 1 and Fig. 11). In flies carrying two copies of gal1118 and two copies of either one of the two UAS-tnt insertions, no strong effect could be seen when comparing them with w;;gal1118 and w;UAS-tnt, or with a w;UAS-tntV1-A;gal1118 line expressing an inactive form of the toxin (Sweeney et al., 1995) (Table 1 and Fig. 11). A slight increase in the variability of the individual circadian periods of the w;UAS-tntH;gal1118 line is suggested by the higher SEM value of tau (0.2 vs. 0.1 for the w;UAS-tntG;gal1118 line and controls; see Table 1), and may reflect a looser control of the period in this genotype. Indeed, a weakening of the rhythmicity occurs at the end of the 6-day period of DD analysis when the activity is plotted by fractions of 2 days (compare days 1–4, Fig. 11N, with days 5–6, Fig. 11P). This long-term effect has been seen only with the UAS-tntH insertion, in the presence of two copies of gal1118 and UAS-tntH. No similar weakening of the rhythm was found for the homozygous w;UAS-tntG;gal1118 line during a 15-day experiment (not shown).