The duration of caffeine treatment plays an essential role in its effect on sleep and circadian rhythm

Abstract Sleep is regulated by the homeostatic system and the circadian clock. Caffeine intake promotes wakefulness in Drosophila. In humans, caffeine is consumed on a daily basis and hence it is important to understand the effect of prolonged caffeine intake on both circadian and homeostatic regulation of sleep. Furthermore, sleep changes with age and the impact of caffeine on age-dependent sleep fragmentation are yet to be understood. Hence in the present study, we examined the effect of short exposure to caffeine on homeostatic sleep and age-dependent sleep fragmentation in Drosophila. We further assessed the effect of prolonged exposure to caffeine on homeostatic sleep and circadian clock. The results of our study showed that short exposure to caffeine reduces sleep and food intake in mature flies. It also enhances sleep fragmentation with increasing age. However, we have not assessed the effect of caffeine on food intake in older flies. On the other hand, prolonged caffeine exposure did not exert any significant effect on the duration of sleep and food intake in mature flies. Nevertheless, prolonged caffeine ingestion decreased the morning and evening anticipatory activity in these flies indicating that it affects the circadian rhythm. These flies also exhibited phase delay in the clock gene timeless transcript oscillation and exhibited either behavioral arrhythmicity or a longer free-running period under constant darkness. In summary, the results of our studies showed that short exposure to caffeine increases the sleep fragmentation with age whereas prolonged caffeine exposure disrupts the circadian clock.


Introduction
Sleep is a well-documented and conserved behavior across kingdoms [1]. Sleep is critical for diverse physiological functions like metabolism, immune response [2][3][4], cognition, and memory formation [5]. Sleep is majorly regulated by two processes namely the circadian clock that regulates the timing of sleep [6] and the homeostatic pathways that modulate the intensity and depth of sleep [1]. This regulation is best understood by the two-process sleep model proposed by Borbély, in 1981 [7]. This model has been successful in simulating the sleep timing and depth in various experimental conditions [7]. Accumulating evidence continues to help us understand the interactions between these two processes. Electrophysiological recordings in the suprachiasmatic nucleus (SCN), the central pacemaker in mammals, reported continuous interaction between the circadian clock and the sleep homeostat indicating that the interplay between these two systems is important in governing sleep [7].
The circadian clock that controls the timing of sleep consists of interlocked transcriptional-translational feedback loop (TTFL) composed of period (per) and timeless (tim) genes which are activated when the positive transcription factors CLOCK (CLK) and CYCLE (CYC) bind to their E-box promoters [8,9]. This transcriptional-translational feedback loop maintains the transcript and protein oscillations that govern the circadian rhythm in Drosophila [10]. Studies in the recent past have unraveled that circadian neuropeptide-pigment-dispersing factor (PDF), diuretic hormone 31, glutamate and GABA A receptor Resistant to dieldrin (RDL) expressed in subsets of circadian clock neurons mediate sleep and wakefulness to occur at the specific time of the day [11][12][13]. Additionally, circadian output molecule WIDE AWAKE (WAKE) upregulates GABA A receptor RDL and reduces the excitability of clock neurons at the day/night transition to promote sleep onset.
The sleep homeostat controls the depth and intensity of sleep [14][15][16]. Sleep depth is highly variable depending on multiple environmental factors and sleep pressure [17]. One of the major aspects that decides the sleep pressure is the prior wakefulness period. An elevation in sleep pressure accounts for the immediate increase in sleep, termed as sleep rebound observed after a period of sleep deprivation. Sleep loss also increases the overall abundance of synaptic proteins and impairs the neuronal homeostasis [18,19]. From behavioral, neurogenetic, and physiological approaches, researchers have dissected the neural circuits involved in homeostatic regulation of sleep. In Drosophila, a cluster of neurons projecting into the dorsal fan-shaped body (dFSB) present in the central brain complex regulates sleep homeostasis [20,21]. Along with the dFSB, mushroom body (MB) and R2 neurons in the ellipsoid body are the important brain centers mediating sleep drive and arousal in Drosophila [17,22].
Caffeine, a psychostimulant used widely for promoting wakefulness, is an antioxidant derived from plant-based sources. In Drosophila, caffeine is shown to increase calcium signaling in the dopaminergic neuron sub-clusters called the protocerebral anterior medial (PAM) neurons [23]. Furthermore, it was also shown that the activity of caffeine required dopaminergic receptor 1 (DopR1) expressed in MBs to cause increased wakefulness in flies [24]. The increased wakefulness observed due to caffeine treatment leads to decreased sleep in flies. As sleep is governed by both the circadian clock and homeostatic system, it is important to understand the effect of caffeine on both of these pathways.
In Drosophila, development influences sleep during the early adult phase leading to a progressive decrease in sleep from young to mature flies, and further, sleep quality decreases with increasing age [17,25,26]. Previously it has been shown that caffeine affects sleep duration and also causes sleep fragmentation [23]. But, as sleep quality is age dependent, the effect of caffeine on sleep may vary with increasing age. Therefore, in our present study, we assessed the effect of caffeine on age-dependent sleep fragmentation in Drosophila. The results of our study showed that short exposure to caffeine reduces sleep in young and old flies. It also increased the age-dependent sleep fragmentation. Furthermore, we also addressed the effect of prolonged caffeine exposure on sleep and circadian clock in Drosophila. Prolonged exposure to caffeine did not reduce the sleep whereas it delayed the phase of core-clock gene timeless transcript oscillation and also resulted in either a longer free-running period or behavioral arrhythmicity in mature flies. Furthermore, prolonged caffeine treatment delayed the pre-adult development and reduced the life span in Drosophila.

Materials and Methodology
Fly stock and maintenance w 1118 flies were obtained from the Bloomington Drosophila Stock Centre (BDSC #5905). All the flies were raised in cornmeal dextrose medium and were maintained under 12 h light: 12 h dark (LD) cycle where lights came on at Zeitgeber Time 00 (ZT 00) and went off at ZT 12 in the Drosophila growth chamber (Percival Scientific, Perry IA) at 25°C temperature with 65 ± 5% humidity.

Short exposure to caffeine
To record the sleep under short exposure to caffeine, 1-, 10-, 20-, and 30-day-old w 1118 male flies were transferred to locomotor activity glass tubes (65 mm length, 5 mm diameter) containing cornmeal dextrose medium for acclimatization 12 h prior to the start of the experiment. These flies were transferred into locomotor activity glass tubes containing cornmeal dextrose medium with different concentrations of caffeine (0.0, 0.5, 0.75, and 1 mg/mL) just before ZT 00 on the day of locomotor activity recording. Sleep was recorded by using the Drosophila Activity Monitors (Trikinetics, USA) for 24 h under LD in a cooled incubator (MIR-154, Panasonic, Japan). On the subsequent day, at ZT 00, these flies were transferred back to fresh cornmeal dextrose medium without caffeine to record their sleep rebound for the next 24 h. Activity counts were measured at every 1 min interval and sleep was defined as 5 min or more of continuous inactivity. Sleep parameters such as total sleep duration, sleep bout length, sleep bout number, sleep latency, and sleep in 1 h were analyzed using Sleep and Circadian Analysis MATLAB Program (SCAMP) [27]. To confirm the results, each experiment was replicated three times with a sample size of 25-32 flies in each replicate. The data obtained from one such biological replicate is used in the results and figures.

Prolonged caffeine exposure
For prolonged caffeine exposure, we reared first instar larvae in cornmeal dextrose medium containing 0.5, 0.75, and 1 mg/mL of caffeine. Freshly emerged w 1118 male flies were also maintained in vials with the respective caffeine-containing medium. To assess the effect of prolonged caffeine exposure on sleep, 10-day-old mature flies under prolonged caffeine treatment were transferred into the locomotor activity glass tubes containing the cornmeal dextrose medium with the respective caffeine concentrations. Sleep for 24 h and sleep rebound was recorded as mentioned in in the previous section.
Furthermore, to assess the effect of prolonged caffeine exposure on the circadian clock, 10-day-old male flies under prolonged caffeine treatment were individually introduced into locomotor activity glass tubes containing standard cornmeal medium with different concentrations of caffeine. To assess the effect of caffeine on clock-mediated activity-rest rhythm, locomotor activity was recorded under LD with caffeine for the first 5 days. Subsequently, their locomotor activity was recorded under constant darkness (DD) with caffeine for 10 days to assess the free-running period. The normalized waveform of activity-rest rhythms under LD was obtained by dividing the activity data collected in every 15 min interval by the total amount of activity during the 24 h multiplied by 100. Anticipation index (AI) for lights-on and lights-off was calculated as the ratio of activity in 3 h just prior to lights-on or lights-off to the activity that occurs 6 h before the transition [28]. The free-running period of activity-rest rhythm was estimated by using the Lomb Scargle (LS) Periodogram of CLOCKLAB, Actimetrics, USA. We used LS periodogram as it has been shown to be the most robust algorithm for analyzing circadian periodicities [29]. For rhythmicity analysis, autocorrelation periodograms were constructed using raw activity count data and the rhythmicity index (RI) was calculated using VANESSA [30]. Flies were considered rhythmic if RI > 0.3, weakly rhythmic if 0.01 > RI < 0.3. Each experiment was repeated thrice with 28-32 flies in each replicate. One such biological replicate is shown in the sleep and locomotor activity plots under LD. Whereas, a sample size of 80-90 flies was used for assessing the free-running period.

Adult stage-specific caffeine treatment
For the adult stage-specific caffeine treatment, freshly emerged w 1118 male flies raised in cornmeal medium were transferred and maintained in vials containing cornmeal medium with different caffeine concentrations (as mentioned above). After 10 days of adult stage-specific caffeine treatment, sleep, sleep rebound, and activity-rest rhythm under LD and DD were recorded by using the protocol mentioned in prolonged caffeine exposure.

CApillary FEeder assay
To study the feeding behavior, we used a modified CApillary FEeder assay (CAFE) [31] using microtips (Details of the modified CAFE protocol are available in Supplementary Figure S4). To measure food intake, 16 male flies were introduced into CAFE assay tubes and the food containing 0.1 M sucrose solution along with 1% Orange-G (Sigma Aldrich) solution was provided. Three such replicates each containing 16 flies were used in this study. To test the effect of short exposure to caffeine on feeding, after 24 h of caffeine treatment, the flies were introduced into CAFE assay tubes containing different concentrations of caffeine along with sucrose/Orange-G solution during the feeding assay. To assess the food intake post-caffeine treatment, flies kept under caffeine-containing medium for 24 h were removed and introduced into CAFE tubes containing only 0.1 M sucrose solution with 1% Orange-G without caffeine during the feeding assay. The feeding was recorded for 3 h starting from ZT 01 and the volume of food intake in each CAFE tube was measured using vernier calipers. Further, the food intake of each fly was calculated by dividing the volume of food intake in each CAFE tube by the number of flies present in that tube. To assess the effect of prolonged and adult stage-specific caffeine treatment on feeding, the food intake was assessed in 10 day-old flies under prolonged and adult stage-specific caffeine treatment respectively (by using the same protocol mentioned for short-term exposure to caffeine).

Total RNA isolation and quantitative real-time PCR
To assess the effect of prolonged caffeine treatment on clock gene transcript oscillation, 10-day-old male flies entrained to LD under cornmeal dextrose medium and under 1 mg/mL of prolonged caffeine treatment were used. Three biological replicates each consisting of 30 male flies were flash frozen using liquid nitrogen with 4 h time intervals from ZT 02 to 22 under LD. To check the expression of genes involved in homeostatic sleep regulation we sampled the flies only at one-time point-ZT 14. The heads were separated and total RNA isolation was performed using TRIzol (Invitrogen) Phenol: Chloroform (Sigma Aldrich) method. The heads were crushed in TRIzol and post incubation, was separated into an aqueous layer using 99.5% chloroform. The aqueous layer was precipitated using Isopropanol (Sigma Aldrich). The RNA was quantified using nanodrop and cDNA was synthesized using Takara PrimeScript 1st strand cDNA synthesis kit (Cat #6110A) using the manufacturer's protocol. Further quantitative real-time PCR was performed using Takara TB green Premix Ex Taq II intercalating agent (Cat #RR820A) and detected using Bio Rad CFX96 Touch Real-Time PCR. The timeless mRNA values are shown after normalization with the house-keeping gene rp49. The primer details are provided in Supplementary  Table S1.

Jonckheere-Terpstra-Kendall analysis for checking gene oscillation
To analyze timeless transcript oscillation we used the JTK algorithm from Metacycle [32] using R-progamming. The mean from technical replicates of each biological replicate was used for analyzing the transcript oscillation. The transcript levels were considered to be oscillating if the p-value was less than .05. For transcripts that showed oscillation, the phase was calculated.

Life span assay
To test the effect of prolonged caffeine treatment on life span, larvae were reared under different caffeine concentrations, and freshly emerged 10 male flies were transferred into vials containing cornmeal dextrose medium with 0.5, 0.75, and 1 mg/mL of caffeine. Three biological replicates each consisting of ten such replicate vials were set up for each caffeine concentration under LD in a cooled incubator (Percival Scientific, Perry IA). Flies were provided with fresh medium on every third day to avoid death due to desiccation. The death that occurred on each day was noted until all flies died and survival curves were analyzed using a log-rank test.

Pupation time assay
Fifty first instar larvae were transferred into the vial containing either non-caffeinated or caffeinated food. Each biological replicate consisting of three such technical replicates was maintained under LD in a cooled incubator (Percival Scientific, Perry IA). Their pupation was assessed after the third instar larval stage with 6 h interval starting from ZT 00 to ZT 18. An average of three biological replicates was used to quantify the pupation time. The percentage of pupation was analyzed by using a Log-rank test and one-way ANOVA followed by post hoc Dunnett's test was used for mean pupation time.

Statistical analysis
One-way ANOVA or two-way ANOVA followed by post hoc Dunnett's or Tukey's HSD multiple comparisons were used respectively when data were normally distributed. For data sets that did not have a normal distribution, the non-parametric Kruskal-Wallis test followed by Dunn's post hoc multiple comparisons was used. The statistical analyses and sleep data analyses were performed using GraphPad PRISM version 9.2.0. Error bars in all the box-whisker plots represent inter quartile range and the error bars in the waveform of sleep, waveform of activity-rest rhythm, and percentage pupariation graph represents standard error of the mean.

Short-term exposure to caffeine reduces sleep in Drosophila
Sleep changes with increasing age in Drosophila. As caffeine reduces sleep, we assessed the effect of short exposure to caffeine on sleep during different ages namely 1-, 10-, 20-, and 30-day old flies. Freshly emerged 1-day-old flies showed a decrease in sleep only at 1 mg/mL of caffeine concentration when compared to the control (Figure 1,A and C). Whereas 10-, 20-, and 30-day-old flies showed a decrease in sleep in all the three caffeine concentrations namely 0.5, 0.75, and 1 mg/mL when compared to the control ( Figure 1, A-C and Supplementary Figure S1, A and B) (statistical details in Table 1).
Furthermore, we observed a significant decrease in sleep from 1 to 10 day old control flies due to sleep ontogeny, as already reported in previous studies [25,33] (Figure 1, A and C). Although sleep decreases during the first 10 days, sleep increases post ~15 days of emergence with respect to 10-day-old flies [25,33]. Similar changes in sleep were observed in our study as well, where 20-and 30-day-old control flies showed increased sleep when compared to 10-day-old control flies (Figure 1, C and Supplementary Figure S1, E). Furthermore, even with 0.5 mg/mL caffeine a similar increase in sleep from days 10 to 20 was observed, whereas this increase in sleep from 10 to 20 was not observed in higher concentrations namely 0.75 and 1 mg/ mL caffeine concentrations. This indicated that higher concentrations of caffeine caused a larger decrease in sleep in older flies ( Figure 1, C and Supplementary Figure S1, F-H) (statistical details in Table 1). In addition, there was no significant difference in sleep between 20-and 30-day-old caffeine-treated flies, when compared to caffeine counterparts, except in 0.75 mg/mL ( Figure 1, C and Supplementary Figure S1, F-H). These results suggest that sleep changes with age and short-term exposure to only higher caffeine concentrations decreases sleep in young flies whereas even lower concentrations of caffeine lead to a significant decrease in sleep in older flies ( Figure 1, C) (statistical details in Table 1).
Furthermore, in 1 and 10 day old flies, there was a phase delay in the siesta onset when compared to the control ( Figure  1, A and B). We considered whether these effects could be due to caffeine that was provided to flies just prior to lights on. We speculated that immediately after the induction of caffeine, flies exhibit increased activity which could be reflected as sleep loss. To verify this, we provided caffeine prior to the start of the night (ZT 12) instead of the day (ZT 00) (night-time induction of caffeine for 24 h). When we provided caffeine during the night there was no phase delay observed during the morning in day 1 old flies (Supplementary Figure S1, C). But interestingly, in day 10 old flies there was a slight phase delay in siesta onset and offset observed when compared to the control (Supplementary Figure S1, D). This indicates that in day 10 old flies, short exposure to caffeine may be affecting the circadian clock. However, we cannot confirm the effect of caffeine on the phase of circadian rhythm because the sleep-wake cycle was recorded for only for 1 day.

Short-term exposure to caffeine increases the sleep fragmentation with age in Drosophila
Caffeine consumption increases sleep fragmentation in flies [34]. To assess if the fragmentation of sleep due to caffeine is dependent on age, we provided short-term exposure of caffeine to different age groups of flies and measured their sleep bout number. More sleep episodes translate to shorter sleep bouts leading to fragmented sleep. Upon assessing sleep bout numbers across the different ages, it was observed that sleep fragmentation is increased in 10-, 20-, and 30-day-old flies when compared to day 1 control flies indicating that sleep fragmentation increases with age ( Figure 1, D and Supplementary Figure S1, I) (Statistical details in Table 2). Furthermore, caffeine consumption did not increase the sleep fragmentation in both 1-and 10-day-old flies ( Figure 1, D). Upon caffeine ingestion, 20-and 30-day-old flies showed higher sleep fragmentation in 0.5 and 1 mg/mL caffeine concentrations when compared to the control group as well as their counterparts from day 1 and day 10 old flies ( Figure 1, D and Supplementary Figure S1, J and L, and Table 2). Whereas, no significant difference was observed in sleep fragmentation between 10-and 20-day-old flies fed with 0.75 mg/mL caffeine (Supplementary Figure S1, K). But a significant increase in sleep fragmentation was observed between 20-and 30-day -old flies fed with 0.75 mg/mL caffeine. These results indicate that short exposure to caffeine increases sleep fragmentation with increasing age in flies.

Short-term exposure to caffeine does not lead to rebound in older flies
To test whether short-term exposure to caffeine affects homeostatic sleep, we assessed the sleep rebound of caffeine-treated flies under various age groups. An increase in sleep upon removal of caffeine was observed in freshly emerged flies fed with 1 mg/ mL caffeine (Figure 1, E and Supplementary Figure S1, M) (statistical details in Table 3). This indicates that homeostatic sleep pathway could be playing a role in caffeine-mediated sleep changes in freshly emerged flies but in older flies, as we did not find any sleep rebound after the removal of caffeine ( Figure 1, E and Supplementary Figure S1, N-P). To test this further, we analyzed the transcript level of genes known to be upregulated upon sleep deprivation namely bruchpilot, synaptotagmin, syntaxin-18, and homer [35,36] in 10-day-old flies treated with 1 mg/mL caffeine for 24 h along with the control. Only 1 mg/mL was used for this transcript analysis as there was no significant difference in sleep rebound of 10-day-old flies across different caffeine concentrations. Surprisingly, none of these genes were upregulated in caffeine-treated day 10 old flies when compared to the control (Supplementary Figure S1, Q). Older flies were not tested in this assay as there was no difference in sleep rebound among 10-, 20-, and 30-day-old flies.

Short-term exposure to caffeine affects food intake in flies
The decrease in sleep observed upon caffeine ingestion was not mediated through the homeostatic pathway in older flies. Previously, it has been shown that starvation-mediated sleep loss is independent of the homeostatic sleep pathway [37]. As caffeine is bitter to taste, we hypothesized that decreased sleep could be observed due to reduced feeding. To analyze this, we assessed the food intake on both day 1-and day 10-old flies after 24 h of short-term exposure to caffeine. Food intake was recorded in the medium containing the respective caffeine  Table 2) (E) Total sleep in min during sleep rebound post short exposure to caffeine of 1-, 10-, 20-, and 30-day-old w1118 flies (statistical details in Table 3). (F) Food intake post 24 h short exposure to caffeine. Food intake was assessed from ZT 01-04 using CAFE with the presence of caffeine in the food during the assay. Decreased feeding was observed upon the presence of caffeine during the CAFE assay (one-way ANOVA followed by Dunnett's multiple comparison control (day 1) vs. 1 mg/mL (day 1) p < .05) (one-way ANOVA followed by Dunnett's multiple comparison control (day 10) vs. 0.75 mg/mL (day 10) p < .01; Control (day 10) vs. 1 mg/mL (day 10) p < .001).
(G) Food intake post short exposure to caffeine from ZT 01-04 using CAFE without caffeine in the liquid food during the assay. Increased feeding was observed in the absence of caffeine during the CAFE assay (one-way ANOVA followed by Dunnett's multiple comparison Control (day 1) vs. 0.75 mg/ mL (day 1) p < .05; Control (day 1) vs. 1 mg/mL (day 1) p < .001) (one-way ANOVA followed by Dunnett's multiple comparison control (day 10) vs. 0.75 mg/mL (day 10) p < .05; control (day 10) vs. 1 mg/mL (day 10) p < .01). The same alphabets indicate that they are not statistically different from each other and different alphabets indicate statistically significant differences. The error bars represent the interquartile range.
concentrations and we observed that the presence of caffeine in the medium caused a reduction in food intake compared to the control flies. In freshly emerged flies only 1 mg/mL flies showed a significant decrease in feeding, coinciding with the observed decrease in sleep (Figure 1, F), whereas in mature 10-day-old flies a significant decrease in feeding was observed in both 0.75 and 1 mg/mL caffeine concentrations (Figure 1, F). Furthermore, feeding increased immediately upon the removal of caffeine post 24 h of caffeine treatment in both day 1 and day 10 old flies. In both day 1 and day 10 old flies, a significant increase in food intake was observed upon removal of 0.75 and 1 mg/mL of caffeine concentration (Figure 1, G). These results suggest that short exposure to caffeine reduces food intake in flies. However, feeding was not measured in 20-and 30-day-old flies. Without these experiments, we cannot determine the impact of reduced The table contains statistical details of the two-way ANOVA followed by Tukey's post hoc details on short term exposure to caffeine on sleep in 1-, 10-, 20-, and 30-day-old flies. The significance is represented with asterisks (*) and non significant comparisons with ns. The table contains statistical details of the two-way ANOVA followed by Tukey's post hoc details on effect of short-term exposure to caffeine on sleep fragmentation in 1-, 10-, 20-, and 30-day-old flies. Rest of the details are the same as in Table 1. food intake on sleep changes observed under short exposure to caffeine.

Prolonged exposure to caffeine does not affect sleep and food intake in flies
Previously it has been shown that prolonged caffeine ingestion has altered effects when compared to short exposures. For example, previous studies showed that prolonged exposure causes tolerance to caffeine, leads to drowsiness, and also decline in cognition when compared to short exposure to caffeine in humans [38][39][40][41].
As the effect of caffeine is dependent on the duration of caffeine intake, we decided to understand its effects on sleep in Drosophila.
To assess the effect of prolonged caffeine exposure on sleep, we reared larvae under cornmeal dextrose medium containing different concentrations of caffeine namely 0.5, 0.75, and 1 mg/mL. Caffeine ingestion during the pre-adult development delayed the pupation time (Supplementary Figure S2, A). The developmental delay was quantified and a significant increase in pupation time was observed in 0.75 and 1 mg/mL concentrations of caffeine (Supplementary Figure S2, A and B). Although caffeine treatment during pre-adult development increased pupation time, we did not observe any lethality during adult emergence. Even after the emergence, the freshly emerged flies were reared in their respective caffeine concentrations and the activity-rest rhythm was recorded under caffeine post-10 days of emergence. Surprisingly, upon prolonged exposure to caffeine, the flies did not exhibit any significant difference in sleep compared to the control (Figure 2, A and B). Although there was no difference in total sleep, there could be an effect on sleep quality. To validate this, we estimated the mean sleep bout length in flies with prolonged caffeine treatment. Prolonged caffeine treatment did not affect the sleep depth and quality (Supplementary Figure S2, C). We also assessed the effect of prolonged caffeine treatment on feeding and found that flies under prolonged caffeine treatment did not exhibit any significant difference in food intake compared to the control (Figure 2, C). Further, we analyzed the food intake on the removal of caffeine post 10 days of adult caffeine treatment. No difference in feeding was observed even after the removal of caffeine (Supplementary Figure S2, D). This indicates that flies with prolonged exposure to caffeine probably developed caffeine tolerance and it abolished the effect of caffeine on sleep and food intake.
Although flies did not exhibit a significant difference in sleep, a visible delay in siesta offset during the transition from day to night was observed in caffeine-fed flies (Figure 2, A). In accordance with that, to understand if caffeine ingestion might cause a delay in sleep onset during the transition from day to night, we quantified the sleep latency of caffeine-treated flies with respect to the control. We did not find any significant difference in sleep latency (Supplementary Figure S2, E) and we further quantified the sleep during the siesta offset, that is,  This indeed showed that flies fed with caffeine had significantly enhanced sleep at ZT 11-12 when compared to the control (Figure 2,D). Furthermore, this delay in siesta offset increased the sleep prior to lights on (ZT 23-24) under 0.75 and 1 mg/mL of caffeine (Supplementary Figure S2, F). To assess if there was a delay in siesta onset, we quantified sleep at ZT 03-04. A significant decrease in sleep was observed at 1 mg/mL that indicates that higher caffeine concentration delayed siesta onset as well (Supplementary Figure S2, G). As the timing of siesta onset and offset was altered, this indicated that it could be due to the effect of caffeine on the circadian clock-mediated evening anticipatory activity. To investigate this, we recorded the locomotory activity-rest rhythm of flies under prolonged caffeine treatment.

Prolonged caffeine exposure alters the anticipatory activity and the free-running periodicity
As we observed a delay in siesta offset, we further examined the effect of prolonged caffeine treatment on activity-rest rhythm of 10-day-old flies under LD. Upon 5 days of activity-rest rhythm The table contains statistical details of the two-way ANOVA followed by Tukey's post hoc details on effect of short-term exposure to caffeine on sleep fragmentation in 1-, 10-, 20-, and 30-day-old flies. Rest of the details are the same as in Table 1. recording under LD, it was observed that flies fed with 0.5 mg/ mL caffeine concentration reduced the activity prior to lights off ( Figure 2, E) (statistical details in Supplementary Table S2). By increasing the caffeine concentration to 0.75 mg/mL, the morning activity prior to lights on was reduced, but no effect on activity prior to lights off was observed (Figure 2, F) (statistical details in Supplementary Table S2). Whereas, by increasing the concentration to 1 mg/mL, activity onset prior to light on and lights off was phase delayed with an increase in the activity after the morning and evening activity peaks (Figure 2, G) (statistical details in Supplementary Table S2). This change in the activity prior to lights-on and lights-off indicates that there could be a change in the clock-mediated morning and evening anticipatory activity. To confirm this, we quantified their morning and evening AI. Flies treated with 1 mg/mL of caffeine showed a significant decrease in both the morning and evening AI when compared to the control (Figure 2, H and I).
Further, to test the effect of prolonged caffeine treatment on the endogenous clock, we recorded their activity-rest rhythm under constant darkness (DD). Upon recording the activity-rest rhythm under DD, we observed that prolonged caffeine consumption either lengthened the free-running period (Figure 3, A-D and H) or resulted in behavioral arrhythmicity when compared to the control (Figure 3, E-G and I). In total, 93.4% of flies showed rhythmicity in control, whereas the percentage of rhythmicity decreased with increasing caffeine concentration. About 81.8% of flies in 0.5 mg/mL, 81.9% of flies in 0.75 mg/mL, and only 37.5% flies in 1 mg/mL of caffeine exhibited rhythmicity (Figure 3, H). To further assess the strength of the rhythmicity, we analyzed the rhythmicity index. With increasing caffeine concentration more flies showed either weaker rhythmicity or they became arrhythmic (Figure 3, I). These results suggest that prolonged caffeine exposure disrupts the free-running rhythm in Drosophila.

Prolonged caffeine exposure causes phase delay in the timeless transcript oscillation
As we observed a significant difference in behavioral rhythm which is associated with the molecular clock, we assessed the effect of prolonged caffeine treatment on the molecular clock. To this end, we measured the mRNA level of timeless under LD across a 24 h period, with 4 h intervals in control and 1 mg/mL caffeine-fed flies. Only 1 mg/mL was used for the above assay because the highest effect was observed with 1 mg/mL caffeine-fed flies. The core-clock gene timeless transcript exhibited a diurnal oscillation with a peak at ZT 12 under LD in control whereas this oscillation was phase delayed by 04 h (ZT 16) in 1 mg/mL of prolonged caffeine treatment (Figure 3, J and Supplementary Table S3). These results indicate that prolonged caffeine treatment affects the molecular clock and causes a phase delay in timeless transcript oscillation.

Effect of prolonged caffeine exposure on circadian clock is largely adult stage specific
To understand whether the effect of prolonged caffeine treatment on circadian clock is development stage mediated or adult stage specific, we provided caffeine treatment only after the flies emerged from the pupae in an adult-stage specific manner. After 10 days of adult stage-specific caffeine treatment, we recorded their activity-rest rhythm with caffeine. In accordance with prolonged caffeine treatment, adult stage-specific caffeine treatment also delayed the siesta offset when compared to the control. This delayed siesta offset led to increased sleep during ZT 11-12 ( Figure 4 A and C) and also prior to lights-on (ZT 23-24) (Supplementary Figure S3, A). Further, adult stage-specific caffeine treatment showed a delay in siesta onset which resulted in a decrease in sleep from ZT 03-04 in 0.75 and 1 mg/mL of caffeine concentrations when compared to the control (Supplementary Figure S3, B). Apart from delaying the siesta offset, a significant difference in sleep latency was observed in flies treated with 1 mg/mL caffeine concentration (Supplementary Figure S3, C). We did not observe any significant difference in sleep quality, which was quantified using mean sleep length (Supplementary Figure  S3, D). Furthermore, adult stage-specific caffeine treatment did not affect the food intake in the presence of caffeine (Figure 4, D). In addition, adult stage-specific caffeine treatment for 10 days post emergence increased total sleep in 0.75 and 1 mg/mL caffeine-fed flies when compared to the control (Figure 4, A and B). This was indeed surprising, as caffeine is known to only decrease sleep or cause tolerance which leads to no effect on sleep.
To further assess the effect of adult stage-specific caffeine treatment on circadian rhythm, we recorded their activity-rest rhythm under LD with caffeine for five consecutive days. In accordance with prolonged caffeine consumption, 0.5 mg/mL of adult stage-specific treatment caused an increased activity after the evening activity peak (Figure 4, E) (statistical details in Supplementary Table S4). Whereas when the concentration was increased to 0.75 mg/mL, apart from observing decreased activity coinciding with lights on and lights off, these flies also exhibited an increased activity after the evening activity peak ( Figure  4, F) (statistical details in Supplementary Table S4). Similar to prolonged caffeine treatment, 1 mg/mL caffeine concentration decreased activity prior to lights on and lights off and increased the activity after the evening activity peak (Figure 4, G) (statistical details in Supplementary Table S4). This altered activity during the evening hours was further quantified using the evening AI, but no significant difference was observed (Figure 4, H).
Furthermore, we also assessed the effect of adult stage-specific caffeine treatment on endogenous clocks by recording their activity under DD for 10 consecutive days and assessed their free-running periodicity. Similar to the prolonged caffeine treatment, adult stage-specific caffeine treatment also led to either longer free-running periodicity (Figure 4, I) or arrhythmicity (Figure 4, J). These results suggest that the effects of prolonged caffeine treatment on a circadian clock are mostly arising from the adult stage whereas, effects on sleep could be influenced by development as well.

Effect of prolonged caffeine exposure on circadian rhythm is not due to aging
In both the prolonged and adult stage-specific caffeine treatment, a decrease in lifespan was observed ( Figure S3, E-F). The average lifespan of a control male fly was close to 75 days with a 50% life expectancy (death rate) to be over 60 days. Whereas, in caffeine-fed flies this life expectancy decreased drastically with 1 mg/mL caffeine-fed flies having the least lifespan (Supplementary Figure S3, E-F). The 50% life expectancy of 1 mg/mL caffeine-fed flies was close to 25-30 days. The LD and DD activity-rest rhythm recording of flies starting with 10-day-old flies took 16 days for completion, by which the age of the fly is close to 25 days, that is, close to the 50% life expectancy of 1 mg/mL caffeine-fed flies. Studies in the past have shown that aging dampens the amplitude of circadian rhythm and also leads to longer free-running period [42,43]. One of the stark observations we made during prolonged caffeine treatment was longer free-running periodicity and arrhythmicity. Therefore, to understand if caffeine-mediated premature aging was affecting the circadian rhythm or if the action of caffeine affected the circadian clock, we designed experiments by recording the activity-rest rhythm from a younger age (day 2 old flies, which were fed with different caffeine concentrations post emergence). In this case, the age of these flies by the time of completion of activity-rest rhythm recording under LD and DD was only 18 days. By 18 days, only less than 10% of fly death was observed in the life span assay of 1 mg/mL caffeine-fed flies (Supplementary Figure S3, E and F) Upon recording activity-rest rhythm with caffeine for five consecutive days (day 2-day 7) under LD starting with 2 day-old younger flies, it was observed that similar to prolonged caffeine treatment, 0.5 mg/mL showed decreased activity prior to lights  Supplementary Table S5). By increasing the concentration to 0.75 mg/mL, flies still showed a decreased activity prior to lights off and increased activity after lights off without having any effect on the morning peak, as observed in prolonged caffeine treatment ( Figure 5, B) (statistical details in Supplementary   Table S5). By increasing the concentration to 1 mg/mL caffeine concentration, a decrease in the activity was observed before and during the evening activity peak and an increase in the activity after the evening activity peak. But surprisingly, under 1 mg/mL caffeine concentration an increase in activity prior to lights on was observed when compared to the control ( Figure 5,   Supplementary Table S5). Although there was an increase in morning activity prior to lights on, we did not find any significant difference in the morning AI compared to the control ( Figure 5D). We further assessed the evening AI and a significant decrease was observed in 1 mg/mL caffeine-fed flies when compared to the control (Figure 5, E). Following the locomotor activity recording under LD for 5 days, the endogenous free-running periodicity was assessed (day 8-day 18). Similar to prolonged caffeine treatment, we observed longer free-running periodicity in caffeine-fed flies when compared to the control ( Figure 5, F). Further, arrhythmicity increased with increased caffeine concentrations when compared to the control ( Figure 5, G). These results further indicate that the effects of prolonged caffeine treatment on endogenous circadian rhythm are not because of premature aging but, are due to the effects of caffeine on the circadian clock.

Discussion
Caffeine, one of the most widely consumed psychostimulants, induces arousal by the activation of dopaminergic 1 receptors (D1R) [24] and causes sleep fragmentation in flies [24]. Physiological processes are known to deteriorate with age, including the sleep:wake cycle [26,44,45]. With age, sleep consolidation deteriorates by increasing sleep during the day-time and also leads to increased wakefulness during the night-time, causing shorter sleep episodes [44,45]. Previous studies have already shown that apart from decreasing sleep, caffeine also causes sleep fragmentation in younger flies (5-to 8-day-old flies) [34]. However, in our present study, we did not observe sleep fragmentation in day 1 and day 10 flies under a short duration of caffeine treatment. The strain used in the previous study Wu et al. was RC1 whereas, we have used w 1118 flies in our present study. RC1 is a caffeine-sensitive strain and the effects of caffeine may be enhanced in these flies. Hence, sleep fragmentation was observed due to caffeine administration in 5-to 8-day-old younger flies. But, we did not find changes in sleep bout number in our studies by using younger 1-and 10-day-old w 1118 flies under different concentrations of caffeine. And further, we observe that caffeine increases the sleep bout number with increasing age (in day 20 and day 30 old flies). An experiment to verify RC1 strain-specific effect on age-dependent sleep fragmentation is to assess the effect of caffeine sleep bout number or sleep bout duration with this caffeine-sensitive strain. Due to the unavailability of the strain, these experiments could not be performed. Nonetheless, our results with w 1118 flies show that with increasing age sleep fragmentation increases, and this fragmentation is elevated upon caffeine ingestion (Figure 1, D). It has been previously shown that sleep loss can lead to the accumulation of reactive oxygen species (ROS) [46,47]. With increasing age, sleep quality reduces which may also lead to ROS accumulation [45]. Furthermore, treating young flies with paraquat, a ROS inducer increases the sleep fragmentation similar to old flies [45,48]. These results indicate that increase in ROS level leads to sleep fragmentation and vice versa. Furthermore, it has been previously shown that cytochrome proteins (CYPs) whose expression levels are known to be repressed by ROS, also had decreased expression upon caffeine ingestion [49]. This indicates that ingestion of caffeine could lead to increased ROS production and that could be implicated in higher sleep fragmentation observed in the present study. Although the molecular basis of sleep fragmentation is yet to be determined, our results show that caffeine increases sleep fragmentation with age in flies. Previously, it has been shown that 24 h of caffeine treatment leads to sleep rebound in female flies [50]. The results of our study showed that short exposure to caffeine results in reduced sleep and feeding in young and old flies. Furthermore, upon removal of caffeine, sleep rebound was observed only in young flies indicating that caffeine may affect the homeostatic sleep pathway in young flies. But no sleep rebound was observed in older flies. It has been previously shown that starvation suppresses sleep without a subsequent homeostatic sleep rebound [50]. As we observed a decrease in feeding as a result of short exposure to caffeine, it is likely that reduced feeding may increase the foraging or the starvation-mediated hyperactivity [51,52] and suppress the sleep without inducing a sleep rebound. The results of our present study and a previous study indicate that decreased food intake and starvation might be the possible additional factors that could play a role in caffeine-mediated sleep loss [50,53,54]. However, with the available results, we cannot determine the extent to which the caffeine-mediated sleep changes are driven by starvation. Freshly emerged flies possess more reserves of fat including the larval fat that plays an important role in survival under starvation stress for younger flies. This excess fat is consumed mostly during the early adult stages of 4-5 days [55,56]. Hence, we speculate that starvation may not have a stronger impact to induce hyperactivity in young flies and thus caffeine-mediated sleep loss may not be largely driven by starvation in these flies. Whereas, older flies are more sensitive to starvation compared to younger adults, and hence the impact of starvation on caffeine-mediated sleep loss may enhance in older flies.
The major component that is causing food restriction in flies is believed to be the bitter taste of caffeine. Taste-sensing neurons or the gustatory receptor Gr93 is known to be involved in sensing caffeine in flies [53]. It has already been shown that the absence of these gustatory receptors (Gr93a 3 , ΔGr66a) still shows a decrease in sleep [54,57]. This indicates that the bitter taste of caffeine is not responsible for starvation-mediated sleep decrease but other post-ingestive mechanisms could mediate the sleep changes. Another way to reduce food restriction by bitter taste was by increasing the sweet components in the food. But, this experimental design introduces more variables into the study and hence this methodology was not included in the present study.
Short exposure to caffeine reduced the feeding in day 1 and day 10 old flies. In day 1 old flies, only 1 mg/mL of flies showed a significant decrease in feeding compared to the control. This raises the possibility that even though the caffeine concentration was high at 1 mg/mL of treatment, the dose of caffeine these flies got may not be the same due to a significant decrease in feeding. However, we did not find any significant difference in feeding across different caffeine concentrations. For example, flies under 1 mg/mL of caffeine concentration did not exhibit any significant difference in feeding compared to 0.5 mg/mL and 0.75 mg/mL of caffeine concentrations. Similarly in day 10 old flies, a significant decrease in feeding was observed at 0.75 and 1 mg/mL of caffeine concentrations compared to the control. Here in the case of day 10 old flies also, no significant difference was observed in feeding across the different caffeine concentrations. Nevertheless, we cannot ensure an equal dose of caffeine across flies in these experiments. Apart from this, another way to assess the differences in dose in the future study is to assess the caffeine metabolites across different caffeine concentrations and check if there is a concentration-dependent increase in these metabolites.
By using proboscis extension response (PER) as a measure for food intake throughout the lifespan of the flies, a previous study has shown that feeding decreases with age especially in the first cannot be determined without further experimentation. With the available results on sleep and feeding in day 1 and day 10 old flies, it is likely that feeding may decrease in older flies in the presence of caffeine which needs to be validated with further studies.
From our results and previous experiments it has been shown that the effects of caffeine are dependent on the duration of caffeine administration [59,60]. For instance, animals rapidly develop tolerance to caffeine-induced locomotor activity under prolonged caffeine administration [61]. We assessed the effect of prolonged caffeine treatment on sleep and feeding. Ten days of prolonged caffeine treatment under different concentrations of caffeine did not decrease sleep or feeding possibly because these flies acquired tolerance to caffeine. But prolonged caffeine treatment affected the morning and evening anticipatory activity and further led to longer free-running period or behavioral arrhythmicity in flies. The effect of prolonged caffeine ingestion on circadian rhythm is not just restricted to flies. It has already been established that caffeine leads to longer free-running periodicity in Neurospora crassa by increasing the conidiation time. This has been hypothesized to be due to the inhibition of cAMP in the cells [62]. Furthermore, through in-vivo and in-vitro studies in mice, it has been shown that caffeine ingestion causes an increased free-running period [63,64]. Taken together our results showed short exposure to caffeine reduces sleep whereas prolonged caffeine treatment does not affect sleep and it disrupts the circadian rhythm in flies.
Previously, it has been shown that caffeine induces arousal by acting on the dopaminergic 1 receptors (D1R) [24], but the effect of caffeine on the molecular circadian clock is not yet well understood. Our study indicates that the prolonged caffeine treatment phase delays the timeless transcript oscillation. Even though the interaction of caffeine with the circadian clock is yet to be understood, the impact of caffeine on the circadian clock at the molecular level is evident from our studies. Similarly, it has been previously reported that caffeine can phase delay the electrical activity rhythm in the isolated SCN from hamster and rat brain [65,66]. Further studies are required to unravel the underlying mechanisms through which the caffeine phase delays the timeless transcript oscillation. control v/s 0.5 mg/mL p < .001; control vs. 0.75 mg/mL p < .001; control v/s 1 mg/mL p < .001), (G) Rhythmicity index of flies under adult stage-specific caffeine treatment in flies in which experiment was started in post 2 days of emergence with concentrations ranging from 0.5 mg/mL−1 mg/mL along with the control. As the caffeine concentration increases more and more flies show decreased rhythmic power. Rest of the experimental details are the same as in Figure 4.
Nevertheless, we cannot exclude the possibility that caffeine acts on additional relevant targets to affect the circadian timing in Drosophila. Previous reports showed that caffeine ingestion causes inhibition of cyclic Adenosine Monophosphate Phosphodiesterase (cAMP-PDE) activity through secondary messenger molecule cAMP and leads to activation of ryanodine receptors [34,67]. Caffeine causes widespread elevation of cAMP throughout the brain [34]. Furthermore, it is also shown that cAMP downstream signaling activation by calcium can directly modulate the speed of the circadian clock independent of the TTFL's [68]. Whether prolonged caffeine treatment affects the phase of the circadian clock through the changes in cAMP level requires further empirical testing. Moreover, the effect of caffeine on behavioral output is dependent not only on the duration of caffeine intake but also on the concentration of caffeine. Higher concentrations of caffeine have been shown to cause longer free-running periodicity when provided for a small duration [59,64].
Apart from this, our study also showed that prolonged caffeine treatment during larval stages delays the pre-adult development in Drosophila. From previous reports, it is known that this caffeine-mediated developmental delay is mostly associated with the changes in the expression of Adenosine receptor (dAdoR) that orchestrates development in Drosophila [69]. Subsequently, it has been shown that the effect of short exposure to caffeine on sleep is not mediated through dAdoR signaling in Drosophila [34,70] unlike mammals [71]. But the impact of caffeine treatment during larval stages on dAdoRs could have an effect on the adult-stage sleep needs to be further investigated. Along with decreased locomotor activity and delayed pre-adult development, prolonged caffeine treatment also reduced the life span in flies. Reduction in life span is majorly associated with increased ROS production [72]. As it has been previously shown that caffeine reduces the expression of genes involved in clearing ROS [49], prolonged caffeine treatment may enhance ROS levels leading to a reduced life span. Furthermore, it is also interesting to understand if premature aging is due to systemic changes or due to the changes in specific tissues or organs in the fly. Previously, it has been shown that there are differences in caffeine action on the nervous system and the circulatory system. Caffeine acts as a vasoconstrictor in the brain and as a vasodilator in the peripheral blood. Hence it is important to understand the differential effect of prolonged caffeine treatment on premature aging of the brain and hemolymph [73]. These effects could be analyzed in different age groups of flies under prolonged caffeine treatment by using aging markers such as ROS, biochemical markers including protein carbonylation, and so on [74].
The prolonged caffeine treatment disrupted the circadian clock and a similar phase delay in activity onset under LD, longer free-running period, and behavioral arrhythmicity was observed when caffeine was provided in an adult stage-specific manner. Although adult stage-specific caffeine treatment phase delays the circadian clock, an increase in sleep duration was also observed. The increase in sleep observed is quite surprising as caffeine has never been shown to cause an increase in sleep ( Figure 4B). In a previous study, adult stage caffeine treatment (defined as chronic caffeine treatment in Wu et al.) for 7 days still decreased sleep in flies. But, in our study, we did not observe this. One of the major reasons associated with this is the use of a different genotype RC1 in the previously reported study and w 1118 flies were used in our study [34]. RC1 is found to be caffeine sensitive and a decrease in sleep has been observed whereas 10 days of adult stage-specific caffeine treatment in w 1118 flies increased the sleep. More experimental evidence is required to decode the differential effect of caffeine treatment duration on sleep. Another interesting observation from our study is that the concentration and duration of caffeine treatment play a critical role in affecting the circadian clock.
It is important to consider whether these findings on effects of caffeine in Drosophila can be translated into human perspectives. The average caffeine consumption of humans is around 400 mg/ day [75]. These concentrations have been prescribed stating to not have any side effects. Although most of the effects of caffeine in this Drosophila study are largely dose dependent, we cannot compare it with humans as the normalizing factor such as the body mass index (BMI) for Drosophila is not known. The differential effect of caffeine concentrations and duration of caffeine intake on tolerance and the side effects in humans have also been established. Although many features of human sleep and circadian clock are observed in Drosophila, this chronic caffeine treatment experimental paradigm in Drosophila cannot be compared with humans consuming coffee on a daily basis but can be used as a model to study side effects observed due to caffeine consumption. Through this study, we provide a good baseline to conduct experiments and understand how short and prolonged caffeine intake may affect the sleep, food intake and circadian clock by using Drosophila as a model organism. In summary, the results of our studies showed that short exposure to caffeine reduces sleep in flies and increases sleep fragmentation with age. While prolonged caffeine treatment does not have an effect on sleep, it affects the morning and evening anticipatory activity in flies. In addition, the prolonged caffeine treatment phase delays the transcript oscillation of the clock gene timeless and disrupts the circadian rhythm in Drosophila.

Supplementary Material
Supplementary material is available at SLEEP Advances online.