Sleep Homeostasis and General Anesthesia: Are Fruit Flies Well Rested after Emergence from Propofol?

THE mechanisms underlying anesthetic-induced loss of consciousness remain nebulous, despite ubiquitous clinical use and the identification of numerous neuronal targets.^1–3  One emerging hypothesis is that anesthetic drugs reversibly induce unconsciousness by acting on endogenous sleep and arousal circuitry.⁴  For example, sleep duration and sensitivity to anesthesia are controlled by a common pathway in the fly brain.⁵  Although general anesthesia and sleep exhibit obvious clinical differences,⁶  numerous shared neurophysiologic features indicate an overlap in neuronal circuitry underlying both states.⁷  Further evidence supporting this overlap originates from the reports that sleep deprivation (SD) enhances anesthetic potency.⁸  Fittingly, pharmacological augmentation of arousal (via monoaminergic stimulation) destabilizes the anesthetic state and precipitates emergence.^9–12 

The convergence between general anesthesia and sleep has led to questions about the extent to which anesthetics may substitute for aspects of natural sleep.¹³  One approach to address this question is to investigate how anesthetics affect various sleep parameters, such as delta power or the homeostatic response to SD. Seminal studies conducted in rodents show that rapid eye movement (REM) sleep debt accrues during exposure to volatile anesthesia, as evidenced by a decreased REM sleep latency and an increased REM sleep time during the active period after emergence.¹⁴  Although the volatile anesthetics fail to fulfill REM sleep requirements, postanesthesia reductions in delta power indicate that these agents substitute for non-REM (NREM) sleep, excluding halothane, which incurs NREM debt.^14–16 

In contrast to these findings, the widely used IV agent propofol satisfies the homeostatic drive for both NREM and REM sleep, perhaps reflecting a unique property of this anesthetic. Preexisting sleep debt is discharged identically under propofol anesthesia as during natural rebound sleep, whereas new sleep debt does not accumulate.^17,18  Although these agent-specific effects on sleep homeostasis likely reflect heterogeneity in the molecular and neuronal mechanisms of action of structurally diverse anesthetic ligands, an interesting question of how different hypnotics might fulfill some homeostatic functions of sleep remains unanswered.

The finding in rodents that propofol satisfies sleep requirements is in need of replication under divergent conditions and in other species to better assess its general validity. Drosophila melanogaster is an established genetic model for elucidating mechanisms of general anesthesia¹⁹  and has been used successfully to identify a variety of polymorphisms that confer either hypersensitivity or resistance to anesthetics.²⁰  To our knowledge, propofol has never been tested in Drosophila due to practical obstacles imposed by its typically IV route of administration. Therefore, we determined whether fruit flies are sensitive to propofol by using a behavioral assay^21,22  and found that flies are effectively anesthetized when propofol is added to the food medium. We first asked whether fruit flies fed propofol are unresponsive to mechanical stimuli as one would expect of truly anesthetized subjects. We next investigated the pharmacokinetic profile of orally ingested propofol in flies to better correlate fly tissue propofol levels with locomotor activity. Finally, we hypothesized that propofol’s ability to satisfy the homeostatic requirement for sleep in rodents should generalize to Drosophila and used an SD paradigm to test this assumption.

Flies were grown on standard dextrose/cornmeal media. Female, Canton-S (wild-type) flies were used exclusively in all experiments. No vertebrate animals were used in these studies.

All experiments were performed using flies maintained at 25°C under 12:12 light:dark conditions. Daily sleep was measured by monitoring locomotor activity in flies by using the Trikinetics Drosophila Activity Monitoring System (DAM2; Trikinetics, USA). Flies 1 to 4 days in age were carbon dioxide anesthetized on a porous polyethylene/acrylic sorting plate (Tyler Research Co., Canada) and loaded individually into glass tubes containing 5% sucrose and 2% agar medium. Activity counts correspond to the number of times a fly physically interrupts an infrared light beam perpendicular to the axis of the glass tube. Sleep in this assay is defined as an activity count of 0 for a minimum of 5 consecutive minutes.²²  Sleep and activity parameters were analyzed by using custom software (Insomniac3; RP Metrix, USA). Results of all behavioral assays were restricted to flies that survived for at least 24 h after the analysis period. The time of death is determined from the time when all activity counts stop varying from 0 for the remainder of the assay.²³ 

For SD experiments, flies were loaded into two DAM2 activity monitors (32 flies in each monitor), and baseline values were collected for 2 to 3 days. Flies were subjected to SD during the dark phase, from zeitgeber time (ZT) 14 to 24 (where ZT 0 to 12 is lights on and 12 to 24 is lights off) or as otherwise indicated. SD was performed by attaching one of the activity monitors to a multitube vortexer (Corning, USA). The other monitor was left undisturbed as a control. The vortexer was controlled by software (Trikinetics) that activated the vortexer for 1 s with interstimulus intervals randomly varying between 2 and 40 s. The strength of the vortexer was set to the minimal intensity that was required to maintain wakefulness for at least 10 h (i.e., the duration of the desired SD). At the end of the SD period, half of the flies from each monitor were transferred to activity tubes containing propofol mixed with food, whereas the other half was transferred to tubes with food containing equivalent concentrations of vehicle (see Preparation and Administration of Propofol). After 6 h, all flies were transferred to activity tubes containing regular food. Results are reported from two to four replicate experiments as indicated.

To determine the arousal threshold of flies fed propofol, we again used the vortexer to stimulate flies at two different intensities, either mild or strong. Mild and strong stimuli were delivered in an alternating pattern, with a 2-h interval between successive strong stimuli. Interstimulus intervals were a minimum of 30 min, such that mild stimulation was applied at 1.5, 3.5, and 5.5 h during propofol exposure, and strong stimulation was introduced at 2, 4, and 6 h after induction of propofol anesthesia. Arousal responses were also monitored for up to 24 h after withdrawal, with the strong stimulus applied as indicated. The percentage of flies responding correspond to the number of sleeping (or anesthetized) flies that were awakened by the vibratory stimuli divided by the total number of flies that were quiescent before stimulation. Sleep or consolidated immobility is stipulated as a minimum of 0 activity counts for at least 5 min before the stimulation. Awakening or arousal by the mild stimulus is defined by having any activity counts occurring within the first minute after stimulation. Arousals in response to the strong stimulus are defined by more than 1 to 2 activity counts within the minute of stimulation to eliminate the movement artifact that occurred at this stimulus strength. In each experiment, arousal responses were measured from 16 to 24 flies for each condition (drug, vehicle, and others). Flies that were already awake at the time of stimulation (i.e., registered greater than 0 activity counts in the 5 min before stimulation) were excluded from this analysis. Flies are considered responsive only if they are also quiescent for a minimum of 5 min before stimulation. Results are reported from four to six replicate experiments as indicated.

A 100 mM stock solution of propofol was prepared by dissolving 9 μl of propofol (2,6-diisopropylphenol, greater than 97%; Sigma-Aldrich, USA) in 491 μl of polyethylene glycol (PEG400). Activity tubes containing various concentrations of propofol were prepared by diluting an appropriate volume of 100 mM propofol stock solution in regular fly food (5% sucrose and 2% agar).

Propofol was administered by individually transferring flies to activity tubes with sucrose food containing various doses of propofol or equivalent amounts of vehicle. Activity data were collected for 3 baseline days during photoentrainment before administering propofol treatment on the fourth day for 6 h from ZT 0 to 6 (where ZT 0 to 12 = lights on and ZT 12 to 24 = lights off) or as otherwise indicated. After propofol treatment, flies were transferred back to freshly prepared activity tubes containing regular sucrose medium, and behavioral data were collected for an additional 3 to 7 days after drug exposure.

Flies were fed 1.0 mM propofol for 6 h from ZT 0 to 6. Flies were collected in Eppendorf tubes over dry ice every 2 h during exposure and up to 48 h afterward. Fly heads were separated from bodies by agitating Eppendorf tubes containing frozen whole flies and manually dividing the heads from bodies on a piece of weighing paper suspended over dry ice. For evaluating the effects of SD on propofol concentrations in flies, flies with and without SD were collected immediately after a 6-h treatment with 1.0 mM propofol at ZT 6. Two other groups of flies (sleep deprived and control) were left to recover until ZT 6 the following day. Flies were manually decapitated, and head weights were estimated by weighing whole flies before and after decapitation. Groups of five fly heads and five whole flies were analyzed independently to determine the propofol concentrations in tissue. The experimenter who conducted the high-performance liquid chromatography (HPLC) measurements was blinded to the condition and time points from which the flies were collected. Results are reported from three replicates.

The tissue from each sample was homogenized in 400 μl of solution containing a 3:1 volume ratio of acetonitrile and 0.02 M phosphase buffer. The homogenate was centrifuged (10,000g at 4° for 20 min), and 25 μl of the supernatant was injected into a Shimadzu HPLC instrument (Shimadzu, Japan).

Calibration curves were run before each batch of samples, and all had correlation coefficients of 0.99 or better. Calibrations curves were prepared as follows: A standard 0.05 mM propofol solution was prepared with methanol, and 2, 5, 10, 15, and 20 μl of solution were injected into the HPLC used to construct the calibration curve by plotting the propofol peak against the known amount of analytes and fitted by linear regression analysis.

The chromatography system consisted of a Gold system (Beckman Coulter, USA) separation module and an FP-2020 plus fluorescence detector (Agilent Technologies, USA). Separation was achieved on a Vydac analytical C18 column. The mobile phase was an isocratic mixture of A (acetonitrile), B (2-propanol), and C (0.02 M NaH₂PO₄ buffer) solution. The flow rate was 1.5 ml/min, and the column temperature was ambient. The fluorescence detector was set at an excitation of 276 nm and an emission of 310 nm.

Samples of flies were inherently randomized through the loading of glass tubes and activity monitors (see Behavioral Assays). Sample sizes were restricted by the size of the DAM2 monitors (32 flies per monitor), by the number of monitors that could be accommodated by the vortexing apparatus, and by the speed in which flies could be manually transferred from regular food to drug in a timely manner, typically two to four monitors per experiment.

To analyze recovery sleep after SD, we calculated net changes in sleep across 12-h periods (day and night) for 2 days after the SD period using an approach similar to that described by Kuo et al.²⁴  In brief, baseline values were subtracted from postdeprivation values in individual flies and averaged across handled control (nondeprived) groups. The average control value was then subtracted from values obtained in individual flies in sleep-deprived groups. The net changes in sleep are, therefore, corrected for effects of time in the assay as well as effects of drug. Baseline correction was not applied in the starvation assay due to increased activity in the nondeprived controls. Results were subjected to one-sample t tests. A Bonferroni correction was applied to P values for the number of tests performed (n = 4; daytime and nighttime points were treated separately). Corrected P values and all other P values described of 0.05 or less were considered significant. A two-way ANOVA was also performed to evaluate an effect of drug treatment as well as time on recovery sleep.

One-way ANOVA followed by Tukey post hoc comparison was used to evaluate the differences in locomotor activity and other sleep parameters among groups exposed to vehicle or varying doses of propofol to determine whether a significant effect of dosage exists. For arousal threshold experiments, results are reported as percent immobile (sleeping/anesthetized) flies responding to stimuli; Kruskal–Wallis tests were used to evaluate the differences among propofol doses with a Bonferroni correction applied to the P value for each of six time points tested. Survival was analyzed by using a Kaplan–Meier estimator followed by a log-rank test. A Cox-proportional hazard regression analysis was performed to evaluate the effects of dose and duration of propofol on survival. All statistical analyses were performed by using open-access software developed by Hammer et al.,²⁵ Paleontological Statistics software (PAST, version 2.17; available at: http://folk.uio.no/ohammer/past/).

To examine the effect of different doses of propofol on Drosophila behavior, we transferred individual flies to activity tubes containing regular sucrose food mixed with varying amounts of propofol. Flies were kept on propofol or an equivalent concentration of vehicle for 6 h from ZT 0 to 6, the first half of the lights-on phase. The highest dose, 3.0 mM propofol, rapidly abolished locomotor activity in all flies (fig. 1A) but killed 66% (21 of 32) of animals treated with this high dose. Exposing flies to two lower, nontoxic doses of propofol during wakefulness also robustly decreased locomotor activity (P < 0.0001 one-way ANOVA, n = 32 vehicle [VEH] vs. n = 41, 39, and 11 flies fed 0.33, 1.0, and 3.0 mM propofol, respectively; fig. 1, A and B). The group of flies fed the lowest dose (0.33 mM propofol) regained baseline activity levels much more rapidly than the groups of flies fed higher doses of propofol (fig. 1A). The emergence period (ZT 7 to 12) occurring after the flies were returned to fresh, drug-free sucrose activity tubes, most clearly separated the effects of different doses on locomotor activity (fig. 1B). Flies fed the lowest dose of propofol showed activity levels during the emergence period that were significantly higher than that in groups fed 1.0 and 3.0 mM propofol (P < 0.0002, Tukey post hoc test). These differences in activity level during emergence from anesthesia putatively reflect the varying depth of anesthesia and residual drug effects after the 6-h ingestion of different propofol doses in the food medium.

As consolidated inactivity in flies lasting more than 5 min is conventionally defined as sleep, another means of evaluating these data quantify the epochs during which flies fail to trigger beam breaks. Although these propofol-fed animals could either be in an extended state of propofol-induced immobility or could merely have exited the anesthetic state and fallen asleep as discussed below, figure 2A demonstrates that inactivity putatively assigned as “sleep” was increased dose-dependently by propofol exposure. The duration of “sleep” bouts also reached maximal levels toward the end of the propofol-feeding period (fig. 2B). When evaluated across the propofol period as well as during emergence, “sleep” bout length was strongly influenced by propofol dose (fig. 2C). Specifically, all three doses of propofol significantly increased “sleep” bout duration during the propofol treatment period (P < 0.00001, one-way ANOVA), with the two higher doses (1.0 and 3.0 mM) producing significantly longer “sleep” bouts than the lower dose (0.33 mM; P < 0.002, Tukey post hoc test). The 0.33 mM dose also increased bout duration relative to the vehicle control (P < 0.02, Tukey post hoc test). During emergence, after all flies were transferred back to regular food (ZT 7 to 12; fig. 2C), bout duration remained increased in flies that had received 1.0 and 3.0 mM relative to lower doses and vehicle controls (P < 0.001, Tukey post hoc test). Together, these data indicate that propofol ingestion by flies dose-dependently reduces locomotor activity with a concomitant increase in consolidated inactivity.

Given that 3.0 mM propofol fed over a 6-h period killed 66% of flies, we attempted to create a deeper anesthetic state by feeding flies high doses of propofol for a shorter duration and wondered whether this might be less toxic. Flies were fed 3.0 or 10 mM propofol or equivalent concentrations of vehicle for 1 or 3 h, starting at lights on, ZT 0, and then transferred back to regular food. Survival outcome was monitored for 7 days. Similar to the 6-h treatment period, high doses of propofol rapidly reduced activity and increased “sleep” as well as bout duration of inactivity. These high-dose propofol effects persisted for more than 12 h irrespective of 1- or 3-h treatment periods (fig. 3, A and B). For all behavioral parameters measured, activity counts (fig. 3, C and D), total “sleep” time (fig. 3, E and F), and bout length (fig. 3, G and H) per 6 h, one-way ANOVA showed significant effects of drug (P < 0.00001, for 1-h fly dosing: n = 47 fed VEH, n = 48 fed 3.0 mM propofol, and n = 48 fed VEH and n = 30 fed 10.0 mM propofol across three experimental replicates; for 3-h fly dosing: n = 47 fed VEH, n = 31 fed 3.0 mM propofol, and n = 48 fed VEH and n = 13 fed 10.0 mM propofol in total across three independent experimental replicates). However, there was no significant effect of dose on behavior within the 1- or 3-h feeding times (P > 0.05, Tukey post hoc test), indicating a possible ceiling effect of propofol as delivered in the food medium.

Striking effects of dose and feeding duration of propofol were detected on survival (fig. 3, I and J). Cox-proportional hazards regression analysis showed significant effects of dose (risk ratio, 2.20; P < 0.0001) and duration of feeding (risk ratio, 2.08; P < 0.0002, n = 191 flies across all propofol-fed groups). As shown in figure 3I, most of the flies that succumbed to 1-h exposure of the drug did so within the first 48 h, while the remaining flies survived for the duration of the assay. Exposing flies to 3 h propofol killed approximately 70% of flies receiving 10 mM propofol and 30% of flies receiving 3 mM propofol (fig. 3J). Few remaining flies succumbed over the next several days. In conclusion, flies show dose-dependent behavioral responses to propofol with a ceiling effect on inactivity at 3.0 mM and increased anesthetic-associated mortality in keeping with high-dose propofol exposure in other species.^26,27 

To further analyze behavioral effects of propofol and to attempt to differentiate between sleeping and anesthetized flies, we assessed the arousal threshold of flies fed nontoxic doses of propofol. We subjected propofol-fed flies to mechanical stimuli using a vortexer and compared their responsiveness to that in naturally sleeping flies. Flies were considered responsive only if they were also quiescent for a minimum of 5 min before mechanical stimuli. Propofol (0, 0.5, and 1.0 mM) was administered during the second half of the night, from ZT 18 to 0, corresponding to a period during which entrained flies would normally be sleeping. Propofol-treated flies were significantly less responsive to vibratory stimuli compared with naturally sleeping flies (fig. 4). Propofol-fed flies were nearly unresponsive to the mild stimuli, with a significant effect detected at ZT 19.5 (P < 0.008, Kruskal–Wallis; n sleeping flies = 79 fed VEH, n = 92 fed 0.5 mM propofol, and n = 101 fed 1.0 mM propofol across six replicates). Propofol robustly reduced responses to strong stimuli at all time points measured (fig. 4; P < 0.02, Kruskal–Wallis, n = 137 fed VEH, n = 119 fed 0.5 mM propofol, and n = 112 fed 1.0 mM propofol in total across six experimental replicates). Although the higher, 1.0 mM dose of propofol almost completely abolished responsiveness and the 0.5 mM dose impaired responsiveness to a lesser degree, these differences were not statistically significant (P > 0.09, Mann–Whitney comparison).

We next used HPLC to evaluate the pharmacokinetics of ingested propofol in whole flies as well as in isolated fly heads. Groups of five female flies fed 1.0 mM propofol were collected at indicated time points during a 6-h propofol treatment and afterward for up to 48 h (fig. 5). The concentration of propofol in homogenized fly tissue was then measured at each time point and normalized to the weight of the tissue samples. Propofol accumulated at higher concentrations in heads than throughout the whole fly, with peak concentrations of 1,623.0 ± 211.3 ng/mg tissue in heads and 243.4 ± 17.3 ng/mg tissue in whole flies (fig. 5) after 6-h exposure. Oral ingestion of propofol caused rapid drug accumulation and, in keeping with propofol’s high lipophilicity,²⁸  displayed the expected slower clearance (fig. 5). The prolonged clearance of oral 1.0 mM propofol correlates with the sustained reduction in locomotor activity observed in flies during the withdrawal period (fig. 1). These data indicate that the absorption and clearance of orally ingested propofol correlates with fly behavior during exposure.

Were propofol anesthesia to substitute for natural sleep, no sleep debt would accrue during propofol exposure and preexisting sleep debts would be discharged. To determine whether sleep pressure is dissipated under propofol anesthesia, we subjected flies to SD for 10 h from ZT 14 to 24 and individually transferred them to activity tubes containing either 1.0 mM propofol or an equivalent concentration of vehicle for 6 h from ZT 0 to 6. Flies were then returned to activity tubes containing fresh food and monitored continuously for up to 7 days. After the 10 h SD, vehicle-fed flies showed the expected period of recovery sleep in the morning (ZT 0 to 6) relative to the nondeprived control (fig. 6A). Net changes in sleep were calculated across 12-h periods in the sleep-deprived groups and normalized to those in corresponding nondeprived controls (see Materials and Methods). The increase in sleep in the vehicle group was significant (P < 0.0001, one-sample t test, Bonferroni corrected; n = 48 flies [SD], and 46 controls across three replicates) for the first 12 h after SD, whereas no significant changes were detected at night or for the subsequent day (fig. 6B). Propofol-fed flies showed significant change in sleep neither during the first 12 h after SD nor at nighttime (P > 0.05, n = 30 flies [SD], and 45 propofol-fed controls across three replicates). However, a significant increase in sleep was detected on the day after propofol treatment (day 2, fig. 6B; P < 0.03, one-sample t test, Bonferroni corrected). Because SD elicited no significant changes in sleep at night in either the vehicle or the propofol-fed groups (data not shown), we focused on daytime sleep only and conducted a two-way ANOVA to evaluate the effects of drug condition and time on recovery sleep. A significant interaction between drug effect and time was detected (F_1,152 = 11.0, P < 0.002), but no effect of drug (P = 0.41). These findings indicate that sleep debt does not dissipate during propofol anesthesia but is instead delayed until after emergence from anesthesia.

A criterion for including flies in analyses of sleep and other behavioral parameters is that they survive for at least 24 h after the analysis period (see Materials and Methods). Approximately 95% or greater survival is common for most behavioral studies in flies. However, we noticed that a much larger proportion of propofol-treated flies that were also subjected to SD did not survive and had to be excluded from our analyses described above (see Behavioral Assays in Materials and Methods). To investigate this issue further, we calculated survival in vehicle- and propofol-fed flies with and without SD (fig. 6C). Although 1.0 mM propofol alone had no significant effect on survival (P = 0.08, log-rank test, n = 46 VEH and propofol-fed flies; fig. 6C, left panel), subjecting flies to 10 h SD before propofol anesthesia significantly decreased survival outcome (P < 0.0001, log-rank test, n = 46 VEH and 44 propofol-fed flies; fig. 6C, right panel). This finding suggests that propofol anesthesia may severely impair restorative properties of sleep.

We next assessed propofol concentrations in flies subjected to SD and compared values to those that were not. Groups of flies were subjected to 10 h SD from ZT 14 to 24 and fed 1.0 mM propofol from ZT 0 to 6. Propofol-fed flies that were sleep deprived were collected at ZT 6, along with a nondeprived control. The remaining flies from each group were allowed to recover for 24 h and collected the next day at ZT 6, 24 h after propofol treatment. Propofol concentration was measured in fly heads by using HPLC, as described in Materials and Methods. Results are shown in figure 6D. All flies fed propofol for 6 h had 1,438.7 ± 267.1 ng/mg tissue in heads (N = 3 replicates, five flies each for SD and control groups), most of which was cleared over the ensuing 24 h, consistent with the result shown in figure 5. However, flies subjected to SD had significantly lower concentrations of propofol (by −45.9 ± 9%) than those in the control group (F_1,8 = 11.14, P < 0.02, two-way ANOVA; fig. 6D). This finding indicates that sleep-deprived flies may have been more sensitive to propofol such that lower amounts of the drug were sufficient to elicit anesthetic effects.

To test whether propofol anesthesia in sleep-deprived flies was equivalent to that in the nondeprived group, we again tested arousal thresholds in flies at different times relative to the propofol or vehicle treatment after SD. We subjected additional groups of flies to 10 h SD and applied strong vibratory stimuli (1-s pulse; see Behavioral Assays in Materials and Methods) during and after the 6-h propofol treatment period. Specifically, stimuli were applied during vehicle or propofol treatment at ZT 4 and 6, during the subsequent nighttime period at ZT 16 and 18, and the following day at ZT 4 and 6. Responses were determined during drug treatment, 10 to 12 h after treatment (during the nighttime), and 24 h later (N = 4 experimental replicates). Results are shown in figure 6E. Kruskal–Wallis tests indicate significant effects during the treatment period (P < 0.007, n = 64 vehicle and 48 propofol controls; 63 vehicle and 33 propofol-fed flies subjected to SD across four experiments), and during the nighttime period (P < 0.04). Mann–Whitney comparisons demonstrate onset of the anesthetic state as evidenced by significantly reduced responsiveness of propofol-treated flies during the drug exposure period relative to the corresponding vehicle group (P < 0.04; fig. 6E, left), but no differences between the propofol-fed sleep-deprived and propofol-fed nondeprived groups or in the vehicle-fed sleep-deprived and vehicle-fed non–sleep-deprived groups, supporting the ability of strong external stimuli to distinguish between sleeping and anesthetized flies. During the nighttime recovery period (fig. 6E, middle), arousal responses of both propofol groups were equivalent to corresponding vehicle groups, but the sleep-deprived vehicle group showed significantly reduced responsiveness when compared with both the nondeprived vehicle and propofol groups (P < 0.04). No significant group effects were detected 24 h after exposure (fig. 6E, right). These findings indicate that SD and control groups were anesthetized by propofol to an equivalent extent and recovered in a manner that was also statistically equivalent. These findings also indicate that the net increase in sleep seen in the propofol-fed sleep-deprived group the next morning, on day 2 (fig. 6B), occurred after flies had emerged from an anesthetized state, supporting the notion that recovery sleep is delayed by propofol anesthesia.

Given that sleep-deprived flies ingested lower amounts of propofol, we next tested whether the decrease in survival in propofol-anesthetized flies was attributed to an impaired access to food. Although flies have free access to food, the combination of forced activity by the mechanical stimulus used to keep flies awake for 10 h and an anesthetized state (despite oral delivery) that followed for 6 h may have reduced the flies’ ability to ingest a sufficient amount of food to support survival. To test this possibility, all flies were placed on nutrient-free medium during the propofol treatment period after 10 h SD from ZT 14 to 24. Interestingly, starvation during the 6-h period after SD blocked recovery sleep and increased locomotor activity in vehicle-fed flies, but not in propofol-fed flies during the 6-h treatment period (fig. 7, A and B). However, the vehicle-fed group that was subjected to SD before the starvation period showed a significant increase in sleep that lasted through the second daytime period of recovery (fig. 7C). The propofol-fed group, in contrast, showed recovery sleep on the second day, similar to that observed without starvation (compare fig. 7C to fig. 6B). Two-way ANOVA showed significant effects of propofol (F_1,216 = 5.98, P < 0.02), as well as time (F_3,216 = 3.05, P < 0.03), with a significant interaction (F_3,216 = 3.08, P < 0.03). Thus, starving flies both delays and prolongs their recovery sleep in response to SD. Propofol-anesthetized flies, in contrast, also showed a delayed recovery sleep, but in a manner that was similar to nonstarved flies, such that the increase in sleep occurred on the second day after SD and after emergence from the anesthesia. If lack of access to the food during propofol anesthesia decreased survival, we would expect that vehicle-treated, starved flies would also show reduced survival after 10 h SD. Instead, we found that starving flies for the 6-h period after SD had no effect on survival in the vehicle-treated flies, but similarly reduced survival in propofol-treated starved flies (fig. 7D). Thus, inadequate access to food does not account for reduced survival in propofol-anesthetized flies after SD.

A shorter SD duration of 6 h (from ZT 18 to 24) also reduced the survival in 1.0 mM propofol-treated flies when compared with vehicle controls (P < 0.00005, log-rank test, n = 31 and 28 flies for sleep-deprived VEH and propofol groups, respectively, across two independent replicates; data not shown). Moreover, 6 h SD produced a significant sleep rebound in vehicle-treated flies on the day after SD with a net increase in 93.6 ± 22.0 min sleep over the 12-h daytime period (P < 0.001, one-sample t test, Bonferroni corrected, n = 28 flies across two replicates). An increase in sleep in the propofol-fed flies was also noted on the second day after SD, but this fell short of significance (60.9 ± 34.4 min, P = 0.088, one-sample t test, n = 14 flies across two replicates). Nonetheless, the reduced survival in the propofol-fed flies support the notion that propofol anesthesia does not satisfy the restorative properties of sleep.

The canonical two-process model of sleep holds that sleep is dissociable into circadian and homeostatic components that together orchestrate sleep–wake cycles in addition to numerous other diurnal neurophysiologic outputs.²⁹  Knowledge of the circadian process has advanced with the identification of various clock genes, such as Period (Per) and Timeless (Tim), which comprise a cell-autonomous transcriptional feedback loop, underlying the core molecular clock in Drosophila.³⁰  Sleep homeostasis, by contrast, is postulated a priori to arise from byproducts of neuronal activity, which accumulate during wakefulness and modulate the duration and intensity of sleep (i.e., sleep pressure). It remains controversial, however, whether a unitary account of sleep homeostasis will emerge paralleling the identification of Per and Tim, due to the complexity and redundancy of a number of sleep regulatory substances, such as adenosine,³¹  numerous cytokines,^32,33  and crossveinless-c³⁴  and sleepless in Drosophila.³⁵  Tung et al.^17,18  reported that propofol anesthesia satisfies the homeostatic requirement for both REM and NREM sleep in rodents, such that preexisting sleep debt dissipates under propofol anesthesia and new debt does not accrue. Although intriguing, this result is in need of replication under disparate conditions and in other species to better assess its general validity and potential relevance to human neurophysiology. Herein, we report that propofol induces a state of general anesthesia in Drosophila and provide behavioral evidence that prior sleep debt does not dissipate during propofol anesthesia, which contrasts with previous findings in rodents.¹⁸  Instead, recovery sleep is delayed until after flies have emerged from anesthesia. Moreover, lethality was increased in flies subjected to SD before propofol anesthesia, suggesting either that temporal suppression of recovery sleep by propofol compromises resilience to SD or that SD increases sensitivity to the adverse side effects of propofol.

We first demonstrate that flies are effectively anesthetized when propofol is added to the food medium as evidenced by the dose-dependent decrease in locomotor activity. The decrease in locomotor activity was limited by the flies’ ability to ingest food, as we noted a ceiling effect on the behavior at high doses, but not on toxicity. One limitation of the locomotor activity assay we used is that flies might cease moving yet remain alert, such that the assay is blind to the fly’s actual arousal level. Or in other words, how can we be certain that flies are not simply motionless rather than anesthetized? The ceiling effect of high-dose propofol is unlikely due to this limitation. Anecdotally, we found that propofol treatment caused a loss of postural reflexes in Drosophila, such that propofol-treated flies were supine and did not attempt to escape during the manual transfer back to fresh activity tubes containing regular food. Nevertheless, to address the possibility that propofol treatment might specifically impair locomotor activity without affecting arousal at lower, less toxic doses, we subjected anesthetized subjects to mechanical stimuli using a vortexer and compared their responsiveness to that in naturally sleeping, control flies. Noxious stimuli are sufficient to rouse sleeping flies, but they fail to rouse propofol-fed flies. As the immobilizing endpoint requires higher doses of propofol than that required for loss of consciousness,³⁶  our data are congruent with the idea that propofol-induced inactivity likely reflects the true induction of an anesthetic state. Although it remains technically unfeasible to directly measure other clinically significant anesthetic endpoints in flies (e.g., analgesia or amnesia), unresponsiveness to external stimuli and cessation of locomotor activity are prima facie tenable indications of the state of general anesthesia.

High-performance liquid chromatography measurements of propofol concentration in fly heads after 6-h exposure to 1.0 mM propofol were slightly greater than predicted in human brain. This estimate assumes that pericerebral fat body occupies approximately 20% of the volume in the fly head,³⁷  that it has a partition coefficient similar to octanol (55-fold greater than brain³⁸ ), and suggests that the actual fly brain propofol levels peak at 97 μg/ml. Given a brain: plasma partition coefficient of 8.2,³⁹  were flies to have plasma, we would predict propofol serum levels corresponding to 11.8 μg/ml. This slightly exceeds the clinical range of 1 to 10 μg/ml in human plasma.^40–43  Considering that propofol takes up to 6 h to reach equilibrium into brain slices,⁴⁴  our HPLC results may cumulatively explain why ingestion of 1 mM propofol leads to consolidated inactivity that is most congruent with a state of anesthesia. This lends credence to our findings that 1 mM propofol-fed flies are not merely sleeping and should not be roused by vibratory stimuli as these tissue levels are predicted to produce surgical plane of propofol anesthesia. Despite millions of years of evolution separating arthropods from mammals, it is remarkable that such strikingly similar doses of propofol elicit anesthesia in flies. Therefore, we suggest that homologous molecular and neuronal targets likely mediate the behavioral effects of propofol in flies.

Using rodents to investigate how propofol alters the response to SD is advantageous because propofol can be delivered intraperitoneally or intravenously, allowing more careful titration to ensure that hypnotic doses of anesthetic are maintained. It is well known that anesthetics produce sedation and depress central nervous system function at lower doses before wholesale loss of consciousness. However, it might be argued that Tung et al.’s result can be rationalized by a transient drop to subhypnotic propofol levels, allowing animals to partially emerge from the anesthetic state, intermittently access sleep, and thereby covertly discharge sleep pressure. Inducing anesthesia by adding propofol to the food medium, the method used in the current study, appears even more susceptible to this objection because oral propofol offers less control over dose than the typical IV route.

However, the pharmacokinetic data directly address this difficulty. Oral propofol is absorbed rapidly in flies, and levels steadily increase during our 6-h exposure. Propofol also slowly clears with detectable amounts remaining in tissue for at least 24 h after removal from the drug (figs. 5 and 6D). These findings are consistent with the arousal assay, where anesthetized flies showed significantly reduced responses to mechanical stimuli applied during nighttime (fig. 4) or daytime (fig. 6E) treatment with propofol. Moreover, ingested propofol preferentially concentrates in neural tissue (on a nanogram propofol per milligram fly tissue basis) rather than in the body of the fly (fig. 5). Intravenously delivered propofol in humans, by contrast, is rapidly cleared from neural tissue and is continually redistributed into other bodily compartments and hence must be administered as a continuous infusion to compensate for its rapid pharmacokinetic profile.⁴⁵  The authors speculate that propofol largely accumulates in neural tissue in Drosophila because (1) ingested propofol passes directly through the proboscis into the head and (2) propofol taken up into circulating hemolymph is pumped directly into the head.⁴⁶  Together, these findings indicate that it is unlikely that the propofol concentration in the central nervous system would fall to subhypnotic concentrations and may partially account for why results of the current study contrast with the previous work of Tung et al.¹⁸ 

If propofol substituted for natural sleep, we would expect animals to revert to normal sleep patterns after emergence from anesthesia regardless of prior sleep history. Instead, flies experience a net increase in sleep 24 h after treatment. Although traces of drug remained in tissue by this time, flies emerged from an anesthetized state as indicated by their responsiveness to mechanical stimuli that was indistinguishable from vehicle-treated controls. To our surprise, 6-h starvation in vehicle-fed flies delayed and extended recovery sleep after a 10-h deprivation. Starvation initially suppresses sleep^47,48  but ultimately causes rebound recovery sleep in adult flies.⁴⁸  The extended recovery sleep in the vehicle-fed group may be explained by the prolonged waking induced by both starvation and SD in the current study. In contrast, propofol-fed flies showed the same pattern of recovery sleep whether they were starved during treatment or not. Our explanation for this is that the propofol-fed flies did not experience waking during the starvation period (and were not further sleep deprived); thus the recovery sleep was solely attributed to the prior nighttime SD.

Although the current findings indicate that propofol does not satisfy the homeostatic need for sleep, whether propofol or other anesthetic drugs produce cognitive or other deficits associated with SD will require further study. The current findings show that lethality in anesthetized flies was increased by SD. To our knowledge, SD does not alter the pharmacokinetics of drugs, but pharmacokinetic changes cannot be entirely ruled out. Propofol may instead antagonize the restorative properties of sleep that are necessary for survival.^49–51  Alternatively, it is likely that as in rodents,⁸  SD increased flies’ sensitivity to propofol as indicated by HPLC measurements showing reduced ingestion of the drug. Future studies should address these issues by evaluating the performance in a learning assay or measuring other physiological parameters associated with sleep such as synaptic scaling.^52,53 

Taken together, our results indicate that flies are effectively anesthetized by the addition of propofol to the food medium and, importantly, establish Drosophila as a suitable genetic model to investigate the mechanisms of propofol anesthesia. These data also indicate that neither sleep debt does dissipate during propofol anesthesia nor does propofol substitute for the restorative aspects of natural sleep in Drosophila.

The authors thank Olivia Lenz, B.S., and Lucas Wittman, Center for Sleep and Circadian Neurobiology (CSCN), Department of Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, for their technical support; and Emma Spikol and Tzu-Hsing Kuo, Ph.D., CSCN, Department of Medicine, Perelman School of Medicine, University of Pennsylvania, for their contributions to early stages of this project.

This work was supported by National Science Foundation grant no. IOS-1025627 and National Institutes of Health (NIH) grant no. R21NS078582 to Dr. Williams and NIH grant no. R01GM088156 to Dr. Kelz; Dr. Naidoo is supported by NIH grant no. P01AG017628. Drs. Meng and Kelz have received research support from the Department of Anesthesiology and Critical Care, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania.

The authors declare no competing interests.