Reorganization of circadian activity and the pacemaker circuit under novel light regimes

Abstract

Many environmental features are cyclic, with predictable changes across the day, seasons and latitudes. Additionally, anthropogenic, artificial-light-induced changes in photoperiod or shiftwork-driven novel light/dark cycles also occur. Endogenous timekeepers or circadian clocks help organisms cope with such changes. The remarkable plasticity of clocks is evident in the waveforms of behavioural and molecular rhythms they govern. Despite detailed mechanistic insights into the functioning of the circadian clock, practical means to manipulate activity waveform are lacking. Previous studies using a nocturnal rodent model showed that novel light regimes caused locomotor activity to bifurcate such that mice showed two bouts of activity restricted to the dimly lit phases. Here, we explore the generalizability of these findings and leverage the genetic toolkit of Drosophila melanogaster to obtain mechanistic insights into this unique phenomenon. We find that dim scotopic illumination of specific durations induces circadian photoreceptor CRYPTOCHROME-dependent activity bifurcation in male flies. We show circadian reorganization of the pacemaker circuit, wherein the ‘evening’ neurons regulate the timing of both bouts of activity under novel light regimes. Our findings indicate that such environmental regimes can be exploited to design light cycles, which can ease the circadian waveform into synchronizing with challenging conditions.

1. Introduction

Cyclic geophysical factors have shaped the evolution of organisms across millennia. Periodic changes in light and temperature across a daily and seasonal scale are among the most apparent abiotic rhythmic phenomena. They are associated with rhythmic physiological and behavioural processes in almost all life forms. It is thought that organisms have evolved internal mechanisms called circadian clocks to appropriately synchronize biological processes to environmental cycles. However, cyclic environmental inputs are variable across seasons and latitudes. Such variation demands that the circadian clock and its output rhythms be flexible to generate appropriate changes in the behavioural or physiological rhythm waveform. This lability in clock function is likely to be adaptive. The response of activity waveform to photoperiodic changes, jet lag, shiftwork or even daylight-saving time indicates that the circadian system strikes a delicate balance between rigidity and plasticity. This would allow the system to be labile enough to respond to changes but also maintain robust timing in the face of minor environmental fluctuations. Despite detailed mechanistic insights into circadian clock functioning, an understanding of the practical means by which rhythm waveforms can be efficiently manipulated is lacking. This would require a better understanding of the principles and factors governing waveform plasticity of biological rhythms.

Among all the behavioural rhythms exhibited by metazoans, daily rhythmicity in locomotion has garnered significant interest owing to its apparent utility in providing insights into waveform regulation. While the waveform of locomotor activity is considered a reasonably close reflection of the state of the underlying circadian machinery, factors such as startle responses or timing of food availability are also known to influence it [1–3]. Thus, it is crucial to understand clock and non-clock factors regulating activity waveform. Even though waveform plasticity has received limited attention, a few studies have used novel light regimes to probe the waveform plasticity of locomotor activity in rodents [4,5]. Through these, a unique phenomenon has been described in Siberian and Syrian hamsters exposed to light regimes with two alternating bright and dim light phases (LD^imLD^im) within 24 h. Under this regime, the activity waveform comprises two segregated bouts restricted to the dim light phases. This phenomenon is distinct from the activity pattern of the same organisms under standard LD 12 : 12 conditions and is termed ‘bifurcation’ or behavioural decoupling. Dim scotopic illumination was found to be critical for activity bifurcation since a similarly timed regime with darkness in the scotophases (LD^arkLD^ark) produced only a single bout of activity across 24 h [6].

At the cellular level, such bifurcation of locomotor activity was correlated with antiphasic oscillations of both PER1 and BMAL1 between the core and the shell of the SCN in mice [7,8]. It is thought that such behavioural plasticity reflects the inherent rigidity and plasticity of the circadian system and depends on the ability of dim scotopic illumination to influence coupling within the circadian circuit [6,9,10]. Through this series of studies using the model system of nocturnal rodents, for which the neuronal network dynamics are known to a considerable extent, the authors unravel factors contributing to waveform flexibility, namely dim scotopic illumination, duration of light phases and the reorganization of the neuronal network. Considering that the above studies are limited to nocturnal rodents, it is impossible to generalize how circadian systems, behaviours or physiological processes respond to such environmental challenges unless studied across taxa, especially those that occupy different ecological and temporal niches.

In a previous study, Drosophila melanogaster flies were subjected to a light regime with alternating light and dark phases (LD^arkLD^ark) to probe the flexibility of activity waveform and oscillator or network interactions [11]. This study demonstrated that similar to rodents, fly activity also does not bifurcate under LD^arkLD^ark. Instead, the waveform is similar to that under a long photoperiod, albeit with a less-prominent evening peak. This observation and the fact that dim scotopic illumination was crucial for activity bifurcation in rodents led to the following questions. (i) Would dim scotopic illumination induce bifurcation of activity waveform in flies similar to nocturnal rodents? (ii) If so, what aspects of the light regime contribute to it—duration of the light phases, wavelength and/or intensity? We reasoned that our studies using the fly could unravel common principles governing circadian oscillator decoupling and waveform plasticity. Additionally, we leveraged the genetic toolkit of D. melanogaster to examine the role of photoreception, circadian clock genes and the cellular components which house the oscillator(s) regulating the activity pattern under novel light regimes.

We exposed male D. melanogaster to a series of light regimes based on those previously described for rodents. We used alternating 70 lux and 1 lux of light intensities (LD^imLD^im) across 24 h. Throughout the manuscript, the light intensities provided during the photophases are referred to as bright light and the intensities provided during the scotophases are referred to as dim light. This was repeated over 5–7 cycles, and activity waveforms were characterized. We find that D. melanogaster undergoes activity bifurcation under certain combinations of durations of the light and dimly lit phases. Additionally, we reveal a role for the Drosophila circadian photoreceptor CRYPTOCHROME (CRY) and canonical circadian clock genes in determining the waveform under this regime. Our study also uncovers unique features of the fly circadian network, which suggest that specific subsets previously described as ‘evening’ cells can modulate the phasing of both bouts when activity is bifurcated under this novel regime.

2. Results

(a). Locomotor activity bifurcation in Drosophila melanogaster

Drosophila melanogaster has been previously shown to exhibit heightened activity under moonlit conditions when subjected to light : moonlight (LM) 12 : 12 regimes [12]. Thus, we first assayed the activity of flies under LD^im 12 : 2 (with 70 lux: 1 lux), wherein we observed a large evening bout, a reduced morning bout and increased nocturnal activity consistent with the previous findings (electronic supplementary material, figure S1a). In the study on nocturnal rodents, LD^arkLD^ark 7 : 5 : 7 : 5 and LD^imLD^im 7 : 5 : 7 : 5 was used [5]. To keep the window of time conducive to activity in a diurnal organism consistent with this, in a previous study we characterized fly activity under LD^arkLD^ark using 5 : 7 : 5 : 7 [11]. Since this activity pattern was already characterized, we introduced dim scotopic illumination and used LD^imLD^im 5 : 7 : 5 : 7 to specifically test whether introducing dim light would induce and maintain stable bifurcation, similar to rodents [4]. This would enable us to attribute the activity pattern observed to dim scotopic illumination alone. The flies were under LD 12 : 12 for the first few days. Then, they were shifted to LD^imLD^im 5 : 7 : 5 : 7. Interestingly, we observed bifurcation—the activity was restricted to and symmetrically distributed between the two dim light phases, with very little activity during the bright light phases (figure 1). This is unlike LD 12 : 12 or LD^im 12 : 12 (electronic supplementary material, figure S1a). To verify that such bifurcation and dim light preference is not unique to a particular strain, we assayed activity/rest under the LD^imLD^im regime for flies of three different strains—one from a well-studied laboratory population with a large standing genetic variation called controls, described in a previous study [11], and common inbred laboratory ‘wild-type’ strains of Canton-S and w¹¹¹⁸ (figure 1a). The batch actogram of the controls (figure 1b) shows the activity distribution under LD 12 : 12 followed by transfer to LD^imLD^im 5 : 7 : 5 : 7, under which they appear to bifurcate their activity. We used the bifurcation index (BI) to quantify the pattern observed [5] (for details, see §4). This index will only assume a value of 1 if all the activity is symmetrically distributed between the two scotophases (dim light phases) with no activity during the photophases. A value significantly higher than 0.5 indicates that most activity is restricted to the scotophases and symmetrically distributed, thus indicative of a bifurcated activity pattern. For all three strains assayed under this regime, the values of the BI were significantly higher than 0.5 by one-sample t-tests (figure 1c). Also, a sudden startle or masked response was observed at the bright-to-dim light transition. These observations across strains show that the effect of exposure to LD^imLD^im on activity waveform is similar between rodents and the fruit fly (i.e. bifurcation or behavioural decoupling).

Figure 1.

Activity bifurcation in D. melanogaster. (a) Activity profiles averaged over cycles and across individuals of the control populations along with the inbred lines of w¹¹¹⁸ and Canton-S under LD^imLD^im 5 : 7 : 5 : 7 (for each strain n > 50). Error bands are s.e.m. The yellow and white bars at the top indicate bright (70 lux) and dim light (1 lux), respectively. (b) Batch actogram of flies (n > 50) from control populations exposed to LD 12 : 12 with 70 lux light intensity followed by LD^imLD^im 5 : 7 : 5 : 7 with 70 lux of light during the photophase alternating with 1 lux of dim light. Yellow shading indicates 70 lux, while grey shading is for complete darkness. In the following cycles of LD^imLD^im, yellow indicates 70 lux, while non-shaded zones indicate 1 lux. (c) BI of Canton-S, w¹¹¹⁸ and control fly strains (for each strain n > 50). The BI was averaged over cycles for each individual and then across individuals. BI is significantly higher than 0.5 by one-sample t‐test for all three strains (p < 0.0001). Error bars are s.d.

(b). Activity bifurcation in D. melanogaster occurs only under specific combinations of bright : dim light durations

To assess the properties of the light regime influencing activity bifurcation, we screened a series of LD^imLD^im regimes since such a peculiar waveform where activity is highly restricted to the scotophases may be produced only by a particular sequence and duration of bright and dim light cycles. We tested the following hypotheses.

Hypothesis 1: symmetrical light regimes facilitate waveform bifurcation. We reasoned that LD^imLD^im 5 : 7 : 5 : 7 is a symmetrical regime, wherein there are two consecutive short light-cycles/T-cycles in 24 h and, thus, can be considered entrainment to 12 h days (LD^im 5 : 7). We examined whether the activity could entrain with a bifurcated pattern to other combinations of T-cycles which we termed ‘asymmetrical’ since the alternate T-cycles were not similar. Under the regime LD^imLD^im 5 : 5 : 5 : 9, we observed that activity bifurcated into the two scotophases (figure 2b). Hence, we negated this hypothesis and concluded that the activity also bifurcates under asymmetrical light regimes, indicating synchronization to 24 h cycles.

Figure 2.

Activity bifurcation depends on scotophase duration. (a–d) Activity profiles of flies from control populations of D. melanogaster under four different light regimes. The regime for each profile is indicated at the top, with the yellow and white bars indicating the bright and dim light durations, respectively. Profiles have been averaged over cycles and across individuals (for each regime n > 84); error bands are s.e.m. (e) Batch actogram of Canton-S flies exposed to LD^imLD^im, yellow indicates 70 lux, while non-shaded zones indicate 1 lux. The flies were first exposed to LD^imLD^im 5 : 7 : 5 : 7 for three cycles and then to LD^imLD^im 5 : 3 : 7 : 9 for the subsequent three cycles to visualize the change in activity waveform from a bifurcated one to a non-bifurcated one as a function of the light regime. (f) BI of flies under each of the four LD^imLD^im regimes. The asterisk indicates that the value was significantly lower than 0.5 by a one-sample t‐test. In contrast, the absence of an asterisk indicates that BI was significantly higher than 0.5. For each of the four comparisons, p < 0.0001, n > 84, and error bars are s.d.

Hypothesis 2: bifurcation occurs when the total duration of dim light is greater than that of bright light. The total duration of dim versus the total duration of bright light could also be a factor that induces bifurcation. This would be akin to the observation of specific waveforms only under certain photoperiods; for example under a long photoperiod, a diminished morning bout and a prominent evening bout of activity are observed [13]. In both regimes described thus far, the total duration of dim light was greater than that of bright light (14 versus 12 h). Therefore, we next imposed LD^imLD^im 5 : 5 : 7 : 7, which has equal durations of total bright and dim light (12 h each). The observation of bifurcated activity under this regime indicates that the total duration of dim light need not be greater than that of bright light for such a pattern to be observed (figure 2c). So far in our study, the regimes used had a total dim light duration equal to or greater than the total bright light. To test the necessity of such a ratio, we provided LD^imLD^im 7 : 5 : 7 : 5 (10 h of dim light as opposed to 14 h of bright light). A bifurcated activity waveform was also observed under this regime (electronic supplementary material, figure S1b).

Hypothesis 3: a critical minimum scotophase duration is required for bifurcation of activity. We exposed the flies to LD^imLD^im 5 : 3 : 7 : 9, where the total duration of bright and dim light was equal (12 h). Here, the durations of each of the photophases are the same as the previous regime (LD^imLD^im 5 : 5 : 7 : 7). However, in this regime, we introduced a shorter duration (3 h) for the first scotophase. We found that the activity did not bifurcate (figure 2d). We also exposed Canton-S flies to LD^imLD^im 5 : 7 : 5 : 7 and then transferred them to 5 : 3 : 7 : 9 to confirm and longitudinally visualize the effect of the regime on the waveform (figure 2e). Thus, we find that there is a threshold duration of the scotophase (> 3 h) that facilitates activity bifurcation.

Bifurcation indices under the above light regimes were estimated, and the value was significantly lower than 0.5 only for LD^imLD^im 5 : 3 : 7 : 9 (figure 2f). Using this series of regimes, we found that there is a requirement of a critical minimum duration of the scotophase and that activity bifurcation can also occur under asymmetrical regimes and even under regimes with a total dim light duration less than the bright light duration. In the case of nocturnal rodents as well, specific photoperiodic requirements were crucial for stable induction of behavioural decoupling [14]. Thus, our results highlight that along with dim scotopic illumination, threshold durations of scotophases and photophases could also be a conserved factor determining the induction of such behavioural decoupling.

(c). Bifurcated activity restricted to dim light is mediated by cryptochrome in the circadian clock neurons

Our experiments indicated that dim scotopic illumination of a specific duration is crucial for activity bifurcation. Thus, factors involved in dim light detection would play a significant role in regulating activity waveforms under these conditions. Fruit flies have three major pathways for photoreception: HB eyelets, compound eyes and the cell-autonomous photoreceptor CRYPTOCHROME (CRY) expressed in subsets of the circadian pacemaker neurons. CRY is known to sense lower light intensities, compound eyes moderate light intensities, while HB eyelets sense higher light intensities [15,16].

Given the involvement of CRY in dim light detection [17], we assayed the activity of a cry null mutant, cry⁰¹, under one of the bifurcation inducing regimes, LD^imLD^im 5 : 7 : 5 : 7. The genetic background of cry⁰¹ fly line is w¹¹¹⁸ and the same has been used as the background control. As seen from the activity profiles (figure 3a, top panel), cry⁰¹ mutants did not display bifurcation (electronic supplementary material, figure S2a shows the batch actogram of cry⁰¹ flies under LD^imLD^im 5 : 7 : 5 : 7). We next assayed activity under LD^imLD^im 5 : 7 : 5 : 7 with 700 lux of bright light and 10 lux of dim light (figure 3a, bottom panel). Thus, we increased the overall light intensity while maintaining the same contrast. These absolute light intensities fall within the detection range by HB eyelets and compound eyes [15,16]. Even under the high light intensities, activity did not bifurcate in the mutants (figure 3a, bottom panel), although the activity amplitude was higher. Further, we observed greater suppression of activity during the photophases under higher light intensity than regimes with 70 lux of bright light. The activity waveform of the cryptochrome mutants was quantified, and the BI was significantly lower than 0.5 (figure 3b). These results indicate that this bona fide circadian photoreceptor is required for bifurcation.

Figure 3.

Role of CRY in activity bifurcation. (a) Average activity profiles of cry⁰¹ and its background control w¹¹¹⁸ under LD^imLD^im 5 : 7 : 5 : 7 with the bright light of 70 lux, dim light of 1 lux (upper), and bright light of 700 lux with dim light of 10 lux (lower). For both the strains under the two light intensities, n > 20. Error bands are s.e.m. (b) Bifurcation indices of cry mutant and its background control, under the regimes with 70:1 lux of bright: dim light and 700:10 lux of bright: dim light. The asterisks over the bars indicate that the BI is significantly lower than 0.5 by a one-sample t‐test (p < 0.0001). For all strains, n > 20. Error bars are s.d. (c) Activity profiles of flies with cry knocked down in the circadian clock neurons and their respective parental controls under LD^imLD^im 5 : 7 : 5 : 7. Profiles of fly lines with two different cryRNAi constructs on the second and third chromosomes are depicted in the upper (VDRC-KK 103414) and lower (VDRC-GD 738) panels, respectively. For each of the genotypes, n > 25. (d) Bifurcation indices of experimental flies and their respective parental controls. Error bars are s.d., and asterisks indicate BI significantly lower than 0.5 for the two experimental genotypes. BI was significantly higher than 0.5 for the parental controls by one-sample t-tests (p < 0.0001 for all genotypes except clk>cryRNAi, for which p = 0.0031). For each genotype, n > 25.

CRY is also expressed in tissues other than the circadian circuit, like the compound eyes [18,19]. Therefore, using the RNAi approach, we knocked down cry expression specifically in the circadian clock neurons using the Clock856GAL4 driver combined with UAS-dicer and assayed activity under LD^imLD^im 5 : 7 : 5 : 7 (70 and 1 lux of bright and dim light, respectively). It is evident from the activity profiles that no bifurcation occurred in the flies with cry knocked down in the clock neurons (figure 3c). Two independent transgenic constructs, one from VDRC-GD and one from VDRC-KK RNAi collections, inserted in chromosomes 3 and 2, respectively, yielded similar results (figure 3c, upper and lower panels). The bifurcation indices for the experimental flies were significantly lower than 0.5, while parental controls (UAS controls for both RNAi constructs and Clock856GAL4 with UAS-dicer) exhibited bifurcation indices significantly higher than 0.5 (figure 3d). Flies with cryptochrome knocked down using timA3GAL4 were also assayed under constant light for rhythmicity to verify the fly lines (electronic supplementary material, figure S2b–d) since flies lacking CRY are expected to be rhythmic under constant light, while control flies should be arrhythmic [20,21].

It is noteworthy that the activity profiles of cry⁰¹ flies are distinct from that of flies with cryptochrome knocked down in the circadian neurons, suggesting that CRY in the compound eyes could also be contributing to the waveform observed under LD^imLD^im regimes.

Thus, these experiments suggest that cell-autonomous CRY-mediated photoreception in circadian pacemaker neurons is necessary for dim light-restricted bifurcation of locomotor activity. This is in line with the observation that even light intensities within the detection range of compound eyes and HB eyelets (700 : 10 lux) did not result in bifurcation.

(d). Activity of circadian clock mutants bifurcates with altered waveform

To assess the contribution of circadian clock genes to the bifurcated waveform, we used null mutants of the components of the primary transcription–translation feedback loop in flies—period, timeless, Clock and cycle. Null mutants of period and timeless are known to show rhythms under synchronizing conditions of LD 12 : 12 owing to startle responses to light–dark transitions [20,22]. Clock and cycle null mutants also show startle responses, but their activity rhythm is weak [22,23]. A previous report showed that under LM 12 : 12 with moonlight intensity of 0.03 lux, clock mutants exhibit low-light induced activity such that an increase in activity is observed post the dusk transition and a high level of nocturnal activity is sustained under moonlight [24]. We thus expected startle responses and dim light-induced activity to be intact in clock mutants and aimed to assess the contribution of the circadian genes in regulating the waveform under LD^imLD^im 5 : 7 : 5 : 7.

We exposed flies with period, timeless, Clock and cycle null mutations and their respective background controls to LD^imLD^im 5 : 7 : 5 : 7 and examined their profiles and quantified the BI (figure 4). Startle responses to bright-dim transitions were observed in the mutant flies in addition to which mutants also exhibit sustained activity under dim light, consistent with the activity pattern observed under LM 12 : 12 [24] (figure 4a–d). However, the overall activity waveform was altered compared to their control genotypes (figure 4a–d). A major difference is a secondary peak of activity observed within each bout in the wild-type background controls which is not observed in the clock mutants (marked by grey arrows in figure 4). We calculated the mean phase of activity of each of the two bouts to quantify this difference in the waveform of the clock mutants and controls. For all the clock mutants tested, the mean phase was significantly advanced compared to their respective controls. This is primarily owing to the absence of the secondary peak in each of the two activity bouts in the clock mutants. This change in the overall shape of the waveform suggests that clock genes contribute significantly to the timing of activity peaks and offsets under LD^imLD^im.

Figure 4.

Role of circadian clock genes in the regulation of activity waveform. (a–d) Activity profiles with s.e.m. error bands of the indicated clock gene mutants with their respective controls (n > 12). The dotted lines mark the mean phase of activity within each of the scotophases for the clock mutant and its respective genetic background control. The values of the mean phase of activity in zeitgeber time along with s.e.m. are mentioned as inset in the activity profiles. The mean phases of period, clock and cycle mutants were significantly advanced compared to Canton-S (Welch’s ANOVA, followed by Dunnett’s T3 post hoc comparisons, p < 0.005). The mean phase of timeless mutant was also significantly advanced compared to w¹¹¹⁸ (Welch’s t‐test, p < 0.0001). (e) Bifurcation indices of all the genotypes. The values of each of the clock mutants were compared with the respective background controls. An asterisk indicates that the genotype had significantly lower BI than its control genotype. Error bars are s.d. The values for the period, clock and cycle mutants and their background control were compared using a one-way ANOVA (p < 0.05, n > 12) followed by Tukey’s post hoc comparisons and the timeless mutant was compared to its control genotype using an unpaired t‐test (p < 0.0001, n > 15). The bifurcation indices for all the six indicated genotypes were significantly greater than 0.5 by one-sample t‐test, p < 0.001.

The above results suggest that the canonical components of the molecular circadian clock and its cell-autonomous photoreceptor play critical roles in determining the waveform under novel light regimes. This led us to investigate the contribution of the circadian clock circuit and its components in enabling such a distinct pattern of rhythmic behaviour.

(e). The evening cell group (LNds) governs the phases of both activity bouts

In the mouse model system, activity bifurcation was found to be correlated with antiphasic oscillations between the core and shell regions of the suprachiasmatic nucleus for the circadian proteins PER1 and BMAL1 [7,8]. This suggests that the circadian pacemaker neuronal circuit undergoes reorganization, and its components are likely decoupled under these novel light regimes. We aimed to use the genetic toolkit of D. melanogaster to understand the contributions of components of the pacemaker circuit in enabling this altered waveform of activity. Under LD 12 : 12, flies exhibit two bouts of activity centred around the dawn and dusk transitions. Previous work has shown that the morning and evening bouts of activity are strongly modulated by subsets of circadian neurons, often referred to as morning (M) and evening (E) cells, respectively [25,26]. The M group mainly comprises the large and small ventral lateral neurons (l-LNvs and s-LNvs) expressing the neuropeptide pigment dispersing factor (PDF). On the other hand, the canonical E cells primarily comprise the dorsal lateral neurons and a small ventral lateral neuron lacking the neuropeptide PDF (CRY^+ve LNds and 5th s-LNv) [27].

To examine the contribution of the M and E neurons to the waveform observed under novel light regimes, we overexpressed DBT ^s in subsets of the canonical circadian pacemaker circuit. Using this approach, we could alter a parameter of the molecular clock (i.e. speed) within specific groups of neurons. As expected, both activity bouts were advanced under LD 12 : 12 in flies, with the allele being expressed in all the circadian neurons (electronic supplementary material, figure S3a). Only the morning bout of activity advanced when DBT ^s was expressed in the PDF^+ve neurons using the pdfGAL4 driver (electronic supplementary material, figure S3b), while only the evening bout advanced when DBT ^s was expressed in the LNds using the LNdGAL4 driver (electronic supplementary material, figure S3c), which are significant components of the evening group of neurons.

Each genotype was then exposed to LD^imLD^im 5 : 7 : 5 : 7. We observed that both activity bouts were advanced in flies with DBT ^s expression in the circadian neurons compared to the control genotypes, indicating the regulation of the waveform by the circadian circuit (figure 5a). The effect was pronounced, especially for the peaks and the offsets of the two activity bouts. This was in line with our previous result, which indicated that the onset/dim light-induced activity is non-clock mediated. There were no differences in the activity waveform of the flies with only the morning cells expressing DBT ^s (figure 5b). Interestingly, the expression of DBT ^s only in the LNds/evening cells caused both bouts to advance (figure 5c). We used the centre of mass (CoM) to estimate the mean phase of activity within each scotophase. The value would thus indicate the mean time around which the activity is centred within each scotophase. Expression of DBT ^s in all the circadian cells (Clk > DBT ^s) resulted in a significantly advanced mean phase of activity compared to its controls (UAS-DBT ^s and ClkGAL4) (figure 5a). However, there was no difference in the mean phase of flies with a faster clock in M cells (pdf > DBT ^s) relative to their parental controls (UAS-DBT ^s and pdfGAL4) (figure 5b). In contrast, the expression of DBT ^s only in E cells (LNd > DBT ^s) was sufficient to significantly advance the mean phase of both activity bouts compared to its control genotypes (UAS-DBT ^s and LNdGAL4) (figure 5c). Together, these results indicate that the circadian clocks in the evening cells of the pacemaker circuit modulate this unique behavioural pattern.

Figure 5.

E cells govern the activity phases of the bifurcated activity bouts. On the left are activity profiles of flies with DBT ^s expressed in circadian clock neurons or subsets of the circuit with their respective controls. Profiles have been averaged over days and then across individuals and are depicted with s.e.m. error bands, for each of the genotypes, n > 26. Yellow and white bars on the top indicate bright and dim light, respectively. On the right are the mean phases of activity of each of the two bouts in scotophases (dim light). The phase estimated is the CoM, which considers the entire bout of activity during each scotophase. Phases of individual flies are circularly plotted with zeitgeber time on the circular axis, and the length of the vector indicates the consolidation of the mean phase values across individuals. For all genotypes, n > 26. Red arrows in the circular plots indicate that the experimental genotype’s activity phase significantly differs from parental controls (Welch’s ANOVA, post hoc Dunett’s T3 multiple comparisons, p < 0.001). (a) Activity profiles and mean phases of flies with all circadian clock neurons expressing the DBT ^s. (b) Activity profiles and mean phases of flies with only the PDF^+ve circadian neurons expressing the DBT ^s. (c) Activity profiles and phases of flies with only the dorsolateral (LNds) circadian neurons expressing DBT ^s.

In summary, under LD 12 : 12, the respective neuronal groups regulate morning and evening bouts. However, under LD^imLD^im, the evening neurons regulate the timing of both bouts of activity. Thus, we found evidence for circadian reorganization under LD^imLD^im compared to LD 12 : 12 using this approach.

3. Discussion

The waveform of biological rhythms is altered in response to changes in the environmental cycles. Methods for waveform manipulation could be of tremendous use for practical purposes, such as easing the system to synchronize to challenging regimes like shiftwork or resynchronization during jet lag or photoperiodic changes. The waveform of biological rhythms has also long served as a proxy to gauge the state of the underlying circuitry across model systems, including D. melanogaster. Analysing the waveform has also allowed the formation and testing of hypotheses regarding the wiring and coupling within the circadian circuit. The morning and evening oscillator model is a prime example [28]. Here, we characterized Drosophila activity rhythm under novel light regimes to systematically study the extent of plasticity of the activity waveform and its underlying circuitry.

We find that dim scotopic illumination (LD^imLD^im) induces and maintains a bifurcated activity pattern in male D. melanogaster. Activity is consolidated in the dim light phases with little to no activity during the photophases. This indicates that the effect of scotopic illumination, leading to behavioural decoupling, is similar in mice and flies. Previous work with fruit flies shows that they prefer dimmer light intensities and exhibit increased nocturnal activity under dim nighttime illumination [12,29]. However, the use of novel light regimes revealed a phenomenon, wherein flies exhibit activity exclusively in the dimly lit phases (figure 1a). Furthermore, our results indicate that the duration of the scotophases is crucial to the induction of bifurcation. A light regime with a very short duration of the first scotophase did not induce activity bifurcation, suggesting that a threshold duration exists for behavioural decoupling to occur (figure 2d). This is like the finding in rodents wherein specific durations of the photophases and scotophases induced activity bifurcation. While our study demonstrates that in flies, too, there are threshold durations for inducing bifurcation, we cannot yet pinpoint a value for that threshold duration. Nevertheless, our results reveal that certain factors contributing to activity bifurcation are conserved across flies and rodents. This bifurcated activity pattern, once adopted, is highly stable across cycles, indicating that particular light regimes can help ease the system to arrive at such an equilibrium. While specific combinations of the durations are critical for inducing activity bifurcation, the spectral quality of light could also be crucial to enabling such synchronization. Another aspect would be to assess the impact on physiology upon exposure to such light regimes. Even though we did not conduct such a study, it is warranted since it can inform us about the practicality of using these regimes for waveform manipulation.

In mice, the bifurcated activity pattern was correlated with the reorganization of the expression pattern of the circadian clock proteins in the suprachiasmatic nucleus of mice [8]. Using the genetic toolkit of the fly model system, we gained insights into the mechanistic components regulating activity bifurcation. In both model systems, dim scotopic illumination was crucial for inducing activity bifurcation. Our study revealed that the dim light photoreceptor CRY is necessary for this behaviour (figure 3a). Additionally, we show that its function in the circadian neurons is essential for activity bifurcation (figure 3c). Although not quantitatively comparable, it is noteworthy that when cry is absent in the brain and the compound eyes (cry⁰¹), the activity onset is tied to the onset of dim light; however, when it is absent only in the clock neurons, the activity is aligned to a photophase and appears like that under a long photoperiod. This observation hints that CRY may function differently in the pacemaker neurons versus in the compound eyes, where other photoreceptors are also present. Thus, while compound eyes may contribute to the observed waveform, cell-autonomous CRY-mediated photoreception within the circadian pacemaker neurons is necessary for activity bifurcation (figure 3c,d). Despite the presence of multiple photoreception pathways in flies, the role of a dedicated circadian photoreceptor in the pacemaker circuit is crucial for adopting such a distinct temporal niche. Hence, in mammals, it would be interesting to know if dim light detecting components in the rod cells or melanopsin in ipRGCs are responsible for the pattern observed.

A previous study assessing activity waveform under LM 12 : 12 demonstrated that moonlight-induced activity persisted even in circadian mutants [24]. Under LD^imLD^im regimes, dim light-induced activity is intact in flies with mutations in the canonical clock genes. However, the activity waveform of these clock mutants is markedly different from their background controls (figure 4), indicating that the clock regulates the phase of the activity peak and offset of the two bouts. These observations in circadian mutants suggest that while dim light-induced activity is independent of the clock, the timing of activity, particularly its peak and offset, is determined by canonical clock genes (figure 4). While these canonical clock genes are rhythmically expressed across the circadian pacemaker circuit, the M and E cells have distinct roles in determining the state of the circuit and, thus, the waveform. The behavioural decoupling observed in our study was also accompanied by an alteration in the dynamics of the circadian network compared to LD 12 : 12. Under this novel regime, we find that the E cells regulate the phasing of both activity bouts. Previous studies have demonstrated that the activity waveform of D. melanogaster exhibits remarkable overlap with the state of the circuit. For instance, under a long photoperiod, the morning peak is reduced to a startle, and a prominent evening peak is observed in accordance with the E cells determining the phasing of activity under these conditions [30]. Similarly, under LM 12 : 12, a prominent evening peak is observed and rescuing the clock only in the E neurons (Mai positive, pdf negative) by overexpressing period in a period null background is sufficient for wild-type like activity pattern under LM 12 : 12 [31]. The latter study also found that under LM 12 : 12, two groups within the E neurons were desynchronized in these flies. The 5th sLNv and 1 of the CRY^+ve LNds formed one group, and two other CRY^+ve LNds formed the other group. Analyses of the connectomic data also indicate the presence of two groups within the canonical E cells (i.e. the 5th sLNv and 1 of the CRY^+ve LNds forming one group and two other CRY^+ve LNds forming another group) [32,33]. Here, for the first time, we report that under the novel light regime LD^imLD^im, the phasing of two symmetrical bouts of activity is driven by evening neurons. We hypothesize that this regime has decoupled the E cellular group completely, with each group regulating one bout of activity. This warrants further investigation using cell-specific drivers such that only particular subsets of the evening neurons can be targeted separately to study their role in timing each of the two bouts of activity. Though challenging to execute, such an experiment may provide further insights into the reorganization of the circadian network when exposed to novel light regimes.

In the past decade, these novel light regimes of LD^arkLD^ark and LD^imLD^im have been used to probe the extent of plasticity of rodent activity waveform [34]. Interestingly, like rodents, flies also do not exhibit bifurcation under LD^arkLD^ark [11]. In the current study, we find that dim scotopic illumination and threshold durations of the scotophases and photophases are common factors regulating activity bifurcation in both model systems. We also find that circadian organization is altered compared to LD 12 : 12. Our study reveals that the evening cells can regulate the phases of two comparable bouts of activity. Thus, we uncover a hitherto unknown degree of plasticity of the circadian circuit using these novel light regimes. These regimes have, therefore, proved helpful in probing the plasticity of the circadian system across taxa and have led to insights at the behavioural and neuronal levels.

A combination of features of the light regime, like the ratio of light intensities and specific durations of the scotophases and photophases, can enable the system to synchronize to unconventional light regimes uniquely. However, the temporal niches of the organisms assayed under these novel light regimes are nocturnal (rodents) or crepuscular (D. melanogaster). In both cases, activity was found to be restricted to dim light. The circadian rhythms and the underlying neuronal pacemaker circuitry of these two model systems have been extensively investigated. There is evidence, at behavioural and circuit levels, for the existence of two oscillators and that they can decouple to a certain extent. Hence, it is essential to conduct such studies across organisms (from different taxa with varying temporal niches and/or circuitry) to delineate the utility of such regimes for waveform manipulation. A recent report also suggests that there is sexual dimorphism in the properties of the circadian network in D. melanogaster [35]. This indicates that there could be differences in the coupling within the circadian network between male and female flies. Thus, with additional information regarding the circadian network in females, assaying female activity under LD^imLD^im could reveal aspects of their circadian organization and the effect of this regime on females. This would further help delineate the practical use of such regimes across species and between the sexes within a species.

Taken together, it is evident that certain factors inducing waveform bifurcation are conserved between flies and rodents. This opens up the possibility of designing light regimes to ease the activity waveform to arrive at such an equilibrium. Thus, such studies can result in an understanding of factors and principles governing waveform plasticity and help arrive at a means to manipulate activity waveform for practical purposes such as shiftwork or intercontinental travel.

4. Material and methods

(a). Fly strains

The strains used in this study include outbred populations of D. melanogaster (fly populations referred to as control [11]) and inbred strains of Canton-S and w¹¹¹⁸. The controls are four distinct long-term laboratory populations of D. melanogaster maintained in large numbers in cages (~1200 flies per cage) [36]. Thus, they are outbred ‘wild-type’ lines and have been used to characterize the activity waveform under LD^imLD^im regimes in addition to the inbred strains. The mutant and transgenic lines used are shown in table 1 and table 2.

Table 1.

Detailed genotypes of the mutants and transgenic fly lines used. The notation used throughout this study is also mentioned along with the source for acquiring the line. The last column lists the references or stock centre IDs of the fly lines.

fly line	notation used	source	reference/ID
w1118;;	w1118	Bloomington Drosophila Stock Centre (BDSC)	BDSC ID-5095
w;;cry01	cry01	obtained from Jeffrey Hall, Brandeis University (backcrossed into the w1118 mentioned above)	[21]
per01;;	per01	BDSC	BDSC ID-80928
w;tim01;	tim01	BDSC (backcrossed into the w1118 mentioned above)	BDSC ID-80922
;;Clkjrkst[1]	Clkjrk	BDSC	BDSC ID-24515
;;cyc01	cyc01	BDSC	BDSC ID-80929
pdfGAL4	pdfGal4	obtained from Todd Holmes, UC Irvine	[37]
Clock856GAL4	ClkGal4	obtained from Orie Shafer, ASRC, CUNY	[38]
Dvpdf-GAL4,pdf-GAL80/ ; R78G02- GAL4,pdf-GAL80/+	LNdGal4	obtained from Daniel Cavanaugh, Loyola University	[39]
tim(A3)GAL4	timGal4	obtained from Todd Holmes, UC Irvine	[40]
yw;;UAS-DBTs-myc/sb	UAS-DBTs	obtained from Orie Shafer, ASRC, CUNY	[41]
;UAS-cryRNAi;	UAS- cryRNAi(II)	VDRC	VDRC ID-7238
;;UAS-cryRNAi	UAS-cryRNAi(III)	VDRC	VDRC ID-105172
;;UAS-dicer−2	UAS-dcr	BDSC	BDSC ID-24651

Table 2.

Detailed genotypes of the experimental transgenic fly lines used in the study. These lines were obtained by crossing the transgenic fly lines listed in table 1, and the shorthand/notation used in this study for each of the fly lines has been tabulated alongside.

experimental transgenic lines	notation used
;Clk856Gal4/UAS-cryRNAi;UAS-dcr	Clk > cryRNAi (II)
;Clk856Gal4;UAS-dcr/UAS-cryRNAi	Clk > cryRNAi (III)
;Clk856Gal4;UAS-DBTs-myc	Clk > DBTs
;PdfGal4;UAS-DBTs-myc	Pdf > DBTs
;LNdGal4;UAS-DBTs-myc	LNd > DBTs

(b). Locomotor rhythm recording

Three- to five-day-old virgin male flies were housed in locomotor tubes with corn food medium and placed in Drosophila Activity Monitors (DAM system by Trikinetics) to measure their activity counts. Assays were carried out in light-proof metal boxes placed in cubicles with an ambient temperature of 25 ± 0.5°C. Light intensity was adjusted using a combination of 210 and 298 neutral-density Lee filters over white LED strips fitted inside the boxes. Light intensities were adjusted using a LI-COR light meter.

Flies were maintained under LD 12 : 12, 70 lux : 0 lux cycles, then shifted to LD^imLD^im regimes. The first bright light/70 lux phase is referred to as photophase 1 (PP1), and its onset coincides with that of the photophase of the previous LD cycle. The first scotophase with dim light is referred to as scotophase 1 (SP1), and the following bright light and dim light phases are photophase 2 (PP2) and scotophase 2 (SP2), respectively. Four regimes have been used, with varying durations of each photophase or scotophase. In each case, the transfer from LD 12 : 12 to LD^imLD^im was such that the start of PP1 coincided with the start of the 12 h photophase of the previous day. Thus, the start of the first photophase is regarded as zeitgeber time 00 (ZT00).

(c). Data analyses

Activity profiles for all experiments have been averaged over days and across individuals. In all cases, flies were first maintained under LD 12 : 12 and then transferred to LD^im or LD^imLD^im regimes. Average profiles were calculated by using the days when stable phases were attained and transients had subsided after a change in the light regime (i.e. the first 2 days after the transfer to a new regime have been eliminated when calculating the average activity profiles. Activity profiles were obtained by averaging over five cycles for each individual and then across individuals.

The bifurcation symmetry index has been used previously to quantify the bifurcated activity patterns observed in rodents [5]. Since we too observed activity restricted to the scotophases, we have used the same measure to assess the degree of symmetry in activity division between successive scotophases:

$${\displaystyle \mathrm{b}\mathrm{i}\mathrm{f}\mathrm{u}\mathrm{r}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}\left(\mathrm{B}\mathrm{I}\right)=\frac{2\times \mathrm{m}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{s}\left(\mathrm{S}\mathrm{P}1\mathrm{v}\mathrm{s}\mathrm{S}\mathrm{P}2\right)}{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{d}\mathrm{a}\mathrm{i}\mathrm{l}\mathrm{y}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}}.}$$

The total number of activity counts in each scotophase (SP1, SP2) and the daily total activity was calculated for each cycle. The activity counts in the lesser-active scotophase are doubled to provide an objective measure of bifurcation ranging from 0 to 1. If activity is perfectly bifurcated between the successive scotophases, the index should be a value of 1, BI = 1. The index enables quantification of the symmetry in the distribution of activity between the two bouts. Any activity occurring in the photophases would be accounted for in the denominator and will not contribute to the numerator and the BI value would be low. Additionally, since the formula uses the minimum activity of the two bouts, if there is an uneven distribution of activity between the two bouts, the additional activity in one of the scotophases would be accounted for in the denominator but not the numerator and this would also lower the value of BI.

The circular mean phase of activity within the scotophases was also quantified for certain genotypes to measure the centrality of activity on the time axis [42]. It considers and quantifies the overall activity waveform.

All the activity profiles and bar graphs have been plotted using GraphPad Prism v. 8. Statistical tests reported have also been performed using GraphPad Prism v. 8.

Ethics

This work did not require ethical approval from a human subject or animal welfare committee.

Data accessibility

Data tables listing the values calculated along with experimental replicates and fly numbers are available in Dryad [43].

Supplementary material is available online [44].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

P.N.S.: conceptualization, data curation, formal analysis, investigation, methodology, validation, visualization, writing—original draft, writing—review and editing; V.S.: conceptualization, funding acquisition, project administration, resources, supervision, writing—review and editing.

Both authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

We would like to acknowledge financial support through intramural funding from the Jawaharlal Nehru Centre for Advanced Scientific Research (JNCASR).