|Home | About | Journals | Submit | Contact Us | Français|
Circadian clocks in the brain are organized as coupled oscillators that integrate seasonal cues to time daily behaviors. In Drosophila, the PIGMENT DISPERSING FACTOR (PDF) neuropeptide expressing morning (M) and non-PDF evening (E) cells are coupled cell groups important for morning and evening behavior, respectively. Depending on day length, either M- (short days) or E-cells (long days) dictate both morning and evening phase, a phenomenon we term network hierarchy. To examine the role of PDF in light-dark conditions, we examined flies lacking both the PDF receptor (PDFR) and the circadian photoreceptor CRYPTOCHROME (CRY). We found that subsets of E-cells exhibit molecular oscillations antiphase to those of wild-type flies, single cry mutants, or single Pdfr mutants, demonstrating a potent role for PDF in light-mediated entrainment specifically in the absence of CRY. Moreover, evening behavioral phase is more strongly reset by PDF(+) M-cells in the absence of CRY. Based on our findings, we propose that CRY can gate PDF signaling to determine behavioral phase and network hierarchy.
Many organisms exhibit 24-hour, or circadian, rhythms, in their behavior and physiology These rhythms are driven by circadian clocks that are synchronized to the day-night cycle via environmental cues such as light and temperature. Many animals, including the fruit fly Drosophila, exhibit bimodal rest-activity profiles with clock-regulated activity peaks that anticipate dawn and dusk (termed morning and evening anticipation).
In Drosophila, these behavioral rhythms arise from well-conserved transcriptional feedback loops in which the basic helix-loop-helix heterodimer CLOCK/CYCLE (CLK/CYC) activates expression of PERIOD (PER) and TIMELESS (TIM) . PER and TIM proteins accumulate in the cytoplasm and then translocate into the nucleus and suppress the transcriptional activities of CLK/CYC. Rhythmic expression and subcellular distribution of PER and TIM serve as indicators of the amplitude, period and phase of the molecular clockwork.
Circadian behavior is driven by coupled oscillators that are organized in an interconnected network of 150 pacemaker neurons clustered into seven groups: large and small ventral lateral neurons (l- and s-LNvs), dorsal lateral neurons (LNds), three groups of dorsal neurons (DN1, DN2, and DN3), as well as the lateral posterior neurons (LPNs) [2, 3]. The l-LNvs and four of the five s-LNvs express a neuropeptide, PIGMENT DISPERSING FACTOR (PDF). Mutations in the genes Pdf or Pdf receptor (Pdfr), or ablation of PDF neurons all result in severely reduced morning anticipation but intact evening anticipation [4–7]. In addition, a functional clock in the PDF(+) LNvs is sufficient for morning anticipation . Thus, these PDF(+) cells are often referred to as morning cells (M-cells). In contrast, ablation of cells that do not express PDF, the PDF(−) s-LNv, the LNds, and subsets of DN1s and DN3s result in a loss of evening anticipation while retaining morning anticipation and thus are designated evening cells (E-cells) . While this framework is useful, M- and E-cells may also mediate E and M behavior respectively, perhaps reflecting the coupling of these oscillators [9–11].
These anticipatory activities adjust to seasonal changes in day length or photoperiod. The pacemaker network switches control between M- and E-cells in response to different photoperiods. In short photoperiods (10 hours light: 14 hours dark; as in winter), M-cells largely set the timing of morning and evening locomotor behavior . In constant darkness (DD), M-cells reset non-PDF oscillators and set circadian period . In long photoperiods (as in summer), E-cells set the timing of morning and evening behavior . In constant light (LL), in certain photoreceptor mutants or in flies over-expressing certain core clock components, E-cells drive behavioral rhythms [12, 14, 15]. We term this light/photoperiod-dependent switch in the identity of the master pacemaker, M/E network hierarchy. It has been suggested that light activates the output of E-cells and inhibits the output of M-cells . However, the molecular mechanisms underlying M/E network hierarchy remains to be elucidated.
Both visual and nonvisual pathways confer light information to entrain pacemaker neurons . The nonvisual photoreceptor CRYPTOCHROME (CRY) is activated by blue light and subsequently causes rapid TIM degradation to reset the molecular clockwork . In the absence of CRY, most pacemaker neurons can be entrained likely via the visual system . How these different photoreceptors contribute to M/E network hierarchy remains unclear.
Here we demonstrate that PDFR and CRY collaborate to set the phase and amplitude of circadian oscillators in light-dark cycles. While flies lacking PDFR or CRY alone have relatively minor effects on LD entrainment, flies lacking both PDFR and CRY exhibit E-cell clocks and behavior that cycle antiphase to those of wild-type flies or in some cases, fail to cycle. PDF(+) M-cells influence the phase of both morning and evening behavior in the absence, but not in the presence, of CRY under standard 12 hour light: 12 hour dark conditions. We propose that the balance between PDFR and CRY activities in E-cells determine network hierarchy and may signal day length information within the circadian network to set behavioral phase.
To examine the role for PDF signaling in light-mediated entrainment, we examined LD behavior in flies mutant for both Pdfr and cry (Pdfrhan5304;cryb). These flies carry a null allele of Pdfr (Pdfrhan5304)  and a strong hypomorphic allele of cry [17–19]. We reasoned that in the absence of the cell autonomous CRY photoreceptor, entrainment would be more dependent on network interactions. Wild-type, Pdfrhan5304, Pdfrhan5304;cryb/+ and cryb flies show robust evening anticipation in LD, with Pdfr mutants displaying an advanced phase as previously observed (Figure 1A–D). In contrast, Pdfrhan5304;cryb double mutant flies do not exhibit evening anticipation (Figure 1E).
We also examined whether the presence or absence of PDF neurons had a similar effect in cry mutants. To selectively ablate PDF neurons, we expressed the proapoptotic gene hid under the control of a GAL4 driven Upstream Activating Sequence (UAShid). hid expression was driven by GAL4 under the control of the Pdf promoter (PdfGAL4). Previous studies have shown consistent ablation of PDF neurons using this approach [4, 9]. As observed in Pdfr mutants, PdfGAL4/UAShid and PdfGAL4/UAShid;cryb/+ flies exhibit robust and phase-advanced evening anticipation (Figure S1A–D). On the other hand, no evening anticipation is observed in pdfGAL4/UAShid;cryb flies (Figure S1E).
Because the lights-on response in LD can mask some of the clock-driven morning behavior, we assessed morning behavior on the first day of DD (DD1) after LD entrainment. Pdfrhan5304;cryb/+ and PdfGAL4/UAShid;cryb/+ flies show much reduced or undetectable morning behavior, similar to Pdfrhan5304 and PdfGAL4/UAShid controls (Figure 1F–I and S1F–I). Surprisingly, we observed a morning activity peak around the time of subjective lights-on but no apparent evening peak in either Pdfrhan5304;cryb or PdfGAL4/UAShid;cryb flies (Figure 1J and S1J). These data indicate that Pdfr and cry double mutants display a profound change in light-dark behavior with a loss of evening but a gain of morning behavior.
The lack of evening anticipation in Pdfrhan5304;cryb flies indicates that either the core molecular clock in E-cells or the output of E-cell clocks to drive behavior is disrupted. To assay core clock phase and amplitude, we examined molecular oscillations of the core clock component, PER, in circadian pacemaker neurons. Here we performed PER immunolabeling in Pdfrhan5304;cryb and cryb flies at four time points in LD. Consistent with published observations, cryb flies display robust oscillations in the s-LNvs, LNds, and DN1s . Of note, we find the LNds exhibit reduced amplitude oscillations relative to wild-type (data not shown; [16, 17]). In addition, the l-LNvs do not show any significant oscillation in cryb flies as previously reported [16, 17]. Pdfrhan5304 mutants display robust oscillations with reduced peak PER levels in both the l-LNvs and LNds . While there are differences between Pdfr and cry mutants, such as in the l-LNv, DN2 [16, 20], and perhaps in LNd amplitude, both Pdfr and cry mutants display intact rhythms with phase similar to wild-type in the s-LNv, LNd and DN1 .
In Pdfrhan5304;cryb flies, we find that PER oscillations are strongly affected in subsets of pacemaker neurons. In a subset of E-cells, the LNds and the PDF(−) s-LNv (the latter is identified by its lack of PDF immunolabeling), PER oscillations are approximately 12 hours out of phase relative to the cryb control and Pdfr mutants (Figure 2A and 2C) . Considering that Pdfrhan5304 flies do not exhibit morning anticipation, the most likely explanation for the robust morning anticipation observed in Pdfrhan5304;cryb flies is that this behavior is actually driven by the antiphase LNd and PDF(−) s-LNv clocks. Given that we observe similar morning behavior in PdfGAL4/UAShid;cryb flies, it virtually excludes the possibility that the PDF neurons are driving this behavior.
In addition, PER oscillation is largely abolished in the DN1s of Pdfrhan5304;cryb flies, while PER oscillation in these cells are intact in Pdfrhan5304 and cryb single mutants (Figure 2B and 2C; ). We examined the standard deviation of PER intensities within DN1s of a given brain as a measure of desynchrony. If DN1s within a single brain are desynchronized then you would expect to see high standard deviations in PER levels within a brain compared to a brain in which DN1 PER oscillations/levels are synchronous. However, we did not observe a larger within-cluster standard deviation among the DN1s of Pdfrhan5304;cryb compared to cryb flies (Table S1). Therefore we believe that lack of PER oscillation in the DN1s is likely the result of a damped PER oscillation at the single cell level in the DN1 group, rather than desynchrony among individual DN1s. In addition, we observe constitutive trough levels, not intermediate levels as would be expected for purely desynchronized oscillators (Figure 2B and 2C). These results demonstrate that the net effect of PDFR signaling is to increase PER levels in the DN1s. Lack of both PDFR and CRY appears to have a modest effect on the amplitude of PER oscillations in the s-LNvs as well (Figure 2A and 2C). Taken together, these data suggest that in the absence of CRY, PDFR plays an important role in setting the phase of [PDF(−) s-LNv and LNd], sustaining (DN1) or amplifying (s-LNv) the oscillation of the molecular clock.
Our data indicate an important role for PDF signaling in photic entrainment. We hypothesized that PDF signaling may affect non-photic entrainment pathways or affect core clocks in ways that would be evident using nonphotic signals. To address this issue, we tested behavior of Pdfrhan5304;cryb flies, as well as Pdfrhan5304 and cryb flies under temperature entrainment conditions. Flies exhibit anticipatory behavior under temperature cycles, reflecting clock entrainment [21–23]. Using 12h/12h 29 ° C/21 ° C thermophase/cryophase (TC) temperature cycles in DD, a locomotor activity peak anticipating warm to cool transition in wild-type flies is evident (Figure S2A). A previous study has demonstrated that E-cell clocks are sufficient for this evening activity peak under temperature cycles . Unlike what is observed in light entrainment, we do not observe any substantial difference in the phase of evening behavior between Pdfrhan5304;cryb, Pdfrhan5304, cryb and wild-type flies entrained to temperature (Figure S2A–F). Thus lack of PDF and/or CRY signaling does not appear to alter circadian locomotor behavior during temperature entrainment. This suggests that PDFR is required specifically for light-mediated entrainment of the evening oscillator.
The visual system plays an important role in entrainment to light-dark cycles especially in the absence of CRY . We observe that GMR-hid/+;cryb flies, in which the visual system (i.e., the compound eyes, ocelli, and H-B eyelet) is ablated by hid expression , entrain poorly to light-dark cycles with wide inter-individual variation in behavioral phase (Figure S3A–F). When these variant individuals are averaged together, little or no rhythm is evident in GMR-hid/+;cryb flies (Figure S3G–J). This is consistent with a previous study  and suggests that pacemaker neurons, including evening oscillators are entrained to light-dark cycles mainly via CRY and visual system mediated pathways.
Given the role of the visual system in entrainment to light-dark cycles, we asked whether loss of the visual system in Pdfr mutants also resulted in disruption in entrained behavior comparable to Pdfr and cry double mutants. We examined circadian locomotor behavior of Pdfrhan5304;GMR-hid/+ flies. In LD, Pdfrhan5304;GMR-hid/+ flies exhibit robust evening anticipation that is phase advanced, comparable to Pdfrhan5304 flies but dissimilar from the behavior of Pdfrhan5304;cryb (Figure S4 and and1E).1E). Subtle differences between Pdfrhan5304;GMR-hid/+, Pdfrhan5304 and GMR-hid/+ flies suggest that PDF and the visual system may have functions independent of each other. These results demonstrate that when only the CRY pathway is present, lack of PDF signaling (Pdfrhan5304;GMR-hid/+) does not lead to severely altered behavior, while on the other hand, when only the visual pathway is present, lack of PDF signaling (Pdfrhan5304;cryb) results in antiphase evening behavior. Given the important role of both the visual system and PDF in CRY-independent entrainment, it is possible that PDF may participate in visual system entrainment in cryb mutants.
Prior studies had shown that the influence of the M-cell clock in evening behavior is dependent on photoperiod . Given the dramatic alteration of phase and amplitude in the LNds and DN1s, respectively, in flies lacking both PDFR and CRY, we hypothesized that CRY may balance PDF signaling in determining M/E network hierarchy. If true, PDF(+) M-cells would more strongly reset E-cell clocks in cry mutants than in wild-type flies.
To address this question, we tested wild-type and cryb flies expressing RNAi targeting the regulatory beta subunit of CK2 (CK2betaRNAi) under 12L:12D conditions . CK2 is a protein kinase critical to the timing of the Drosophila circadian clocks [25, 26]. Expression of CK2betaRNAi in PDF(+) M-cells in wild-type background results in long periods in DD (Table 1), similar to reported partial loss-of-function alleles of CK2beta . Nonetheless in LD, altering the M-cell clock does not significantly alter the timing of evening behavior (Figure 3A and 3B), suggesting that manipulating M-cell clocks does not alter E-cell clocks in 12L:12D conditions, similar to reported results under 12L:12D and 14L:10D . However, when expressing CK2betaRNAi in cryb background we now observe a significant phase-delay of evening behavior in LD (p<0.001), in particular, in the offset of evening behavior relative to cryb and PdfGAL4-UASCK2betaRNAi/+ flies (Figure 3B–D; Table 1). We find that on DD1, the phase of morning behavior is delayed in both PdfGAL4-UASCK2betaRNAi/+ and PdfGAL4-UASCK2betaRNAi/+;cryb relative to wild-type and cryb, consistent with the predicted effects of CK2betaRNAi on M-cell clocks (Figure 3E–H). The delay in the evening peak cannot be attributed to differences in circadian period as PdfGAL4-UASCK2betaRNAi/+ flies actually have a longer period than PdfGAL4-UASCK2betaRNAi/+;cryb flies (Table 1). These results demonstrate that the ability of M-cells to reset the phase of evening behavior under 12L:12D conditions is evident only in the absence of cry.
Here we demonstrate an unexpectedly large role for PDFR signaling in synchronization (or entrainment) of circadian pacemaker neurons to light-dark cycles. Flies mutant for both Pdfr and cry, in contrast to single mutants for either gene, display radically altered LD behavior and molecular oscillations. We find that the PDF(+) M-cells set the phase of evening anticipation in LD in cry mutants, but not in wild-type flies, indicating a CRY-dependent change in oscillator dominance. We hypothesize that CRY may gate PDF regulation of E-cell phase, providing a framework for photoperiod dependent changes in oscillator dominance.
It is intriguing that some E-cells exhibit antiphase activity and molecular rhythms in the absence of both CRY and PDF signaling. Antiphase behavior is evident under light, but not temperature, entrainment. As in flies, both visual and nonvisual photoreceptors participate in circadian entrainment in mammals . Mice with only a residual amount of rod function exhibit a switch from nocturnal to diurnal activity rhythms accompanied by reversal of clock gene expression in the suprachiasmatic nucleus (SCN) . It is believed that this switch of temporal niche is a result of light entrainment via rod pathways, which is known for its role in night (dim light) vision . We propose that in the absence of PDF and CRY signaling, the visual system uses novel mechanisms to entrain the LNds and the PDF(−) s-LNv to a novel phase.
The reciprocal influence of PDF(+) M-cells and non-PDF E-cells on each other as well as morning and evening behavior depends on photoperiod . Based on our results, lack of CRY mimics the effect of less light (or short photoperiod) on the hierarchy of the pacemaker network. Changing the pace of M-cells changes the phases of both morning and evening anticipation in cry mutants, whereas in wild-type flies, changing the pace of M-cells only changes the phase of morning anticipation, the latter consistent with published literature . It has been observed that the visual system suppresses M-cell output that drives LL locomotor rhythms . Our data taken together with these data suggest the presence of two distinct pathways operating in LL or long days: an output pathway that is suppressed by the visual system and an M to E circuit that is regulated by CRY. In fact, PDF neurons have been shown in multiple cases to influence the period of E-cell driven LL locomotor rhythms [12, 14, 15], indicating that an M to E circuit is intact in LL. Our data suggests that CRY directly or indirectly regulates this pathway.
To summarize, we demonstrate that PDF signaling and CRY collaborate to regulate the phase and amplitude of E-cell clock in LD. PDF signaling may be modulating visual system mediated light entrainment of E-cell clock. Moreover, lack of CRY enhances the ability of M-cells to phase-reset evening behavior. These findings provide a framework for understanding how photoperiod may determine oscillator dominance in the circadian network. We propose that in long days, light-activated CRY function renders E-cell clocks relatively insensitive to PDF signaling, whereas in short days, light-activated CRY function is more temporally restricted thus allowing PDF signaling to impact clock phase. Thus, the balance between these dual pathways, light-mediated PDFR activity from M-cells and CRY signaling, may dictate network hierarchy under different photoperiods.
Flies were reared on cornmeal/agar medium at 22–25° C. All entrainment for molecular and behavioral testing is conducted at 25 °C except where indicated. y w or w (for temperature entrainment experiments) are used as wild-type strains. w and y w show similar behavior under temperature cycles . cryb , Pdf01 , and PdfGAL4 , and lines were obtained from Michael Rosbash and have been maintained in the lab for nearly ten years. Pdfrhan5304 was obtained from Jaesob Kim . UAShid  and GMR-hid  were obtained from Bloomington Stock Center.
Locomotor activity levels of male flies were monitored using Trikinetics Activity Monitors (Waltham, MA) for up to 5 days of 12 hr light: 12 hr dark conditions (LD) followed by 7 days of constant darkness (DD). For temperature entrainment male flies were monitored for 7–12 days of 12 hr 29° C: 12 hr 21°C in DD and then released into constant temprature of 25°C in DD. For LD analyses (Figure 1, Figure 3, S1, and S4), activity levels from each fly were normalized and averaged within genotypes over 4 days, as described previously . For DD1 analyses (Figure 1, Figure 3, and S1), activity levels were normalized and averaged over the first day of DD. For LD5 through DD4 analysis (Figure S3), activity levels were normalized and averaged over the last day of LD and first four days of DD. For temperature entrainment analyses (Figure S2), activity levels from each fly were normalized and averaged within genotypes over 4 days.
To calculate time of offset of evening anticipation in LD (Table 1), we determined the largest two-hour decrease in normalized average activity for each fly over the first six hours after lights-off. The time designation refers to the end point of the maximal activity decrease, as averaged among individual flies in each genotype.
Male flies were entrained for at least 5 days at 25°C. Flies were anesthetized with CO2 and dissected in 3.7% formaldehyde diluted in PBS. After fixing for 30 minutes at room temperature, the brains were rinsed 2 times in PBS and incubated in PBS with 1% Triton for 10 minutes at room temperature. The brains were then incubated with 5–10% goat or donkey serum diluted in PBT (PBS with 0.3% Triton) for 30 minutes at room temperature, followed by overnight incubation of 1:500 rat anti-PDF and 1:4000 rabbit anti-PER (both antibodies are generous gifts from M. Rosbash) in PBT containing 5–10% goat or donkey serum at 4°C. After several PBT rinses, the brains were incubated with 1:500 goat-anti-rat AlexaFluor488 (Amersham) for PDF immunostaining and 1:500 goat-anti-rabbit AlexaFluor 594 (Amersham) for PER immunostaining in PBT overnight at 4°C. Final rinses in PBT and PBS were followed by mounting in 80% glycerol diluted in PBS. All slides were coded as to sample identity and remained so until the numerical analysis stage. PDF and PER-stained specimens were photographed with 60× oil lens on a Nikon Eclipse 800 laser scanning confocal microscope. For a given experiment the microscope, laser, and filter settings were held constant. PER immunostaining was quantified from digitally projected Z stacks using ImageJ (NIH). PER-stained soma was outlined to obtain average pixel intensity. On each projection image an unstained area was used for background subtraction. All background-subtracted intensity measurements within a condition (time and genotype) were averaged. For display in Figure 2, all PER immunostaining images were equally adjusted for brightness and contrast to more easily visualize cell groups and brain borders. This does not alter the quantification. To combine experiments, background subtracted measurements were scaled to ZT1 of the control genotype in that experiment. Statistical analysis was conducted in STATISTICA and Excel.
To analyze synchrony among DN1s within a brain, we examined PER intensities within a single hemisphere and calculated the standard deviation among individual DN1s for that hemisphere. For a given genotype and time point, we then calculated the average standard deviation. A higher standard deviation indicates variation between DN1s within a hemisphere which in turn reflects desynchrony.
We thank M. Rosbash for PER and PDF antibodies as well as various fly stocks. We thank Bloomington Stock Center for fly stocks. We also thank H. Purdy for help with PER immunostaining quantification and B. Chung for help with statistical analysis. This work was supported by National Institute of Health grants R01NS059042, R01NS052903, R01MH067870 to R.A.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.