Introduction

The functional and neural basis of sleep is highly conserved from invertebrates through mammals [1,2]. In many cases, powerful genetics in relatively simple model systems, including the fruit fly, Drosophila melanogaster, have allowed for the identification of novel genes and neural mechanisms that have informed our understanding of human sleep [3,4]. However, most work in these models have studied total sleep duration. Therefore, a lack of understanding of the mechanisms underlying sleep quality and broader changes in physiology associated with sleep in non-mammalian models represents a significant gap in our knowledge. In mammals, slow-wave sleep is associated with reduced metabolic rate [5–7]. Growing evidence suggests that many physiological changes associated with mammalian sleep are conserved in flies, including a reduction in whole body metabolic rate [8,9]. The diverse physiological changes associated with sleep, including changes in body temperature, reduced metabolic rate, and synaptic homeostasis, are thought to be critical for sleep’s rejuvenate properties [10–12].

Flies, like mammals, exhibit distinct electrophysiological patterns that correlate with wake and rest [13–15]. We have identified sleep-associated reductions in metabolic rate in flies that are consistent with those that occur in mammals [8,16]. In addition, flies display all the behavioral hallmarks of sleep, including an extended period of behavioral quiescence, rebound following deprivation, increased arousal threshold, and species-specific posture [17,18]. Behavioral tracking systems and software are available for high-throughput detection and analysis of fly sleep using infrared monitoring or video tracking [19,20]. Sleep in Drosophila is typically defined as 5 minutes or more of behavioral quiescence, as this correlates with other behavioral and physiological characteristics that define sleep [8,14,21]. For example, sleep bouts lasting ~10 minutes or longer are associated with increased arousal threshold and low-frequency oscillations in brain activity. These findings are supported by computational analysis modeling sleep pressure [14,21,22]. These analyses suggest the presence of light and deep sleep in flies; however, the genetic and neural basis for these different types of sleep is poorly understood.

The Nf1 gene encodes a large protein that functions as a negative regulator of Ras signaling and mediates pleiotropic cellular and organismal function [23,24]. NF1 mutations in humans cause a disorder called neurofibromatosis type 1, characterized by benign tumors of the nervous system (neurofibromas), as well as increased susceptibility to neurocognitive deficits (e.g., attention-deficit/hyperactivity disorder, autism spectrum disorder, visuospatial memory impairments;[23]. In addition, mutation of Nf1 is associated with dysregulated sleep and circadian rhythms [25,26]. Drosophila deficient for Nf1 recapitulate many of these phenotypes and are widely used as a model to investigate the role of Nf1 in regulation of cellular and neural circuit function [27]. Furthermore, Drosophila Nf1 mutations lead to dysregulated circadian function and shortened sleep [25]. Here, we examine the effects of Nf1 on sleep-dependent changes in metabolic rate and measues of sleep depth.

We find that flies lacking Nf1 fail to suppress metabolic rate during prolonged sleep bouts, revealing a disruption of sleep-dependent changes in metabolic rate. Furthermore, multiple behavioral measurements suggest sleep depth is disrupted in Nf1 mutant flies, including the presence of sleep fragmentation and reduced arousal threshold. Genetic and pharmacological analysis suggest Nf1 modulates GABA signaling to regulate sleep depth and sleep-dependent changes in metabolic rate. Therefore, these findings suggest that Nf1 is a critical regulator of sleep-metabolism interactions, and the conserved molecular and phenotypic nature of Nf1 mutants raises the possibility that these findings may be relevant to the complex pathologies in humans afflicted with NF1.

Results

To examine the effects of Nf1 on sleep and activity, we compared sleep of control flies to nf1^P1 mutants that harbor a near-total deletion in the Nf1 locus [28]. Sleep was reduced during the day and night in nf1^P1 mutants compared to controls (Fig 1A), consistent with previous literature [29,30]. Sleep duration in nf1^P1 heterozygous flies did not differ from controls, indicating that the phenotype is recessive (Fig 1A). The average number of sleep bouts was increased during the night in nf1^P1 flies, while the average bout length was reduced compared to control and heterozygote flies during both the day and night, suggesting that loss of Nf1 results in sleep fragmentation (Fig 1B and 1C). In addition to the loss of sleep, the average velocity of activity during waking periods (waking activity) is elevated, suggesting that loss of nf1^P1 also results in hyperactivity (S1A Fig). These findings suggest Nf1 promotes sleep duration, consolidation of sleep bouts, and modulates waking activity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Loss of Nf1 increases sleep fragmentation and decreases sleep depth.

(A-C) Sleep traits of nf1^P1 mutants, heterozygotes, and their respective control. A. There is a significant effect of genotype on sleep duration (two-way ANOVA: F_2,172 = 80.70, P<0.0001). Compared to control and heterozygote flies, nf1^P1 mutants sleep significantly less during the day (+, P<0.0001; het, P<0.0001) and night (+, P<0.0001; het, P<0.0001). B. There is a significant effect of genotype on bout number (two-way ANOVA: F_2,172 = 21.15, P<0.0001). Compared to control and heterozygote flies, nf1^P1 mutants have significantly more sleep bouts during the night (+, P<0.0001; het, P<0.0001), with no difference during the day (+, P<0.0512; het, P<0.0562). C. There is a significant effect of genotype on bout length (two-way ANOVA: F_2,172 = 53.88, P<0.0001). Compared to control and heterozygote flies, nf1^P1 mutants have significantly lower bout lengths during the day (+, P<0.0201; het, P<0.0001), and night (+, P<0.0001; het, P<0.0001). N = 26–32. (D-F) Sleep traits of pan-neuronal Nf1^RNAi knockdown flies and their respective controls.D. There is a significant effect of genotype on sleep duration (two-way ANOVA: F_2,312 = 46.27, P<0.0001). Compared to controls, pan-neuronal knockdown of Nf1 significantly reduces sleep during the day (nsyb/+, P<0.0001; Nf1^RNAi/+, P<0.0001) and night (nsyb/+, P<0.0001; Nf1^RNAi/+, P<0.0004). E. There is a significant effect of genotype on bout number (two-way ANOVA: F_2,312 = 7.728, P<0.0005). Compared to controls, pan-neuronal knockdown of Nf1 significantly increases bout number during the night (nsyb/+, P<0.0001; Nf1^RNAi/+, P<0.0001), with no difference during the day (nsyb/+, P<0.6476; Nf1^RNAi/+, P<0.9943). F. There is a significant effect of genotype on bout length (two-way ANOVA: F_2,312 = 19.68, P<0.0001). Compared to controls, pan-neuronal knockdown of Nf1 significantly reduces bout length during the day (nsyb/+, P<0.0178; Nf1^RNAi/+, P<0.0253) and night (nsyb/+, P<0.0001; Nf1^RNAi/+, P<0.0001). N = 50–55. (G-N). The Drosophila Arousal Tracking (DART) was used to probe arousal threshold and reactivity measurements. This system records fly movement while simultaneously controlling mechanical stimuli via a digital analog converter (DAC). All measurements included were taken from sleeping flies and determined hourly, starting at ZT0. G. Mechanical stimuli of increasing strength were used to assess arousal threshold in all sleeping flies and was determined by fly movement within 15 sec of stimulus delivery. H. There is a significant effect of genotype on arousal threshold (REML: F_2,94 = 37.62, P<0.0001; N = 29–35). Compared to control and heterozygote flies, arousal threshold significantly decreases in nf1^P1 mutants and occurs during the day (+, P<0.0001; het, P<0.0001) and night (+, P<0.0001; het, P<0.0001). N = 30–35. I. There is a significant effect of genotype on arousal threshold (REML: F_2,98 = 7.795, P<0.0007; N = 25–40). Compared to controls, pan-neuronal knockdown of Nf1 has mixed effects on arousal threshold during the day (nsyb/+, P<0.2138; Nf1^RNAi/+, P<0.0040) and significantly decreases arousal threshold during the night (nsyb/+, P<0.0271; Nf1^RNAi/+, P<0.00062). N = 25–40. J. Mechanical stimuli at maximum intensity was used to measure reactivity as a function of time spent asleep. A fly was considered reactive if it moved within 60 sec of stimulus delivery. K. Linear regression of daytime reactivity as a function of time asleep in nf1^P1 mutants, heterozygotes, and their respective control. The intercepts of each regression line are significantly different from each other (F_2,2064 = 34.77, P<0.0001). L. Linear regression of nighttime reactivity as a function of time asleep in nf1^P1 mutants, heterozygotes, and their respective control. The intercepts of each regression line are significantly different from each other (F_2,2102 = 57.05, P<0.0001). N = 58–70. M. Linear regression of daytime reactivity as a function of time asleep in pan-neuronal Nf1^RNAi knockdown flies and their controls. The intercepts of each regression line are significantly different from each other (F_2,1287 = 53.12, P<0.0001). N. Linear regression of nighttime reactivity as a function of time asleep in pan-neuronal Nf1^RNAi knockdown flies and their controls. The intercepts of each regression line are significantly different from each other (F_2,1564 = 4.887, P<0.0077). N = 36–41. For sleep and arousal threshold measurements, the median (solid line), as well as the 25th and 75th percentiles (dotted lines) are shown. For reactivity, error bars indicate ± SEM. The P-values in each panel indicates whether the slope of the regression line is significantly different from zero. White background indicates daytime, while gray background indicates nighttime. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.

https://doi.org/10.1371/journal.pgen.1011049.g001

To determine if the sleep and activity phenotypes of Nf1 are due to loss of function in neurons, we selectively knocked down Nf1 by expressing Nf1^RNAi under the control of the pan-neuronal driver nsyb-GAL4. Sleep was reduced and fragmented in flies upon pan-neuronal knockdown of Nf1 in neurons (nsyb-GAL4/Nf1^RNAi) compared to flies harboring either transgene alone (Fig 1D–1F). Waking activity was also elevated with neuron-specific knockdown of Nf1 (S1B Fig). To validate that these differences were not due to off-target effects of RNAi, we next confirmed these findings using an independently derived RNAi line [31]. We again found that pan-neuronal knockdown of Nf1 significantly decreased sleep duration, while sleep fragmentation during the night and waking activity increased significantly (S2A–S2D Fig). Therefore, pan-neuronal knockdown of Nf1 fully recapitulates the mutant phenotype, suggesting that Nf1 functions in neurons to regulate sleep.

To further investigate the role of Nf1 on sleep consolidation, we analyzed activity patterns using a Markov model that predicts sleeping and waking propensity, indicators of sleep depth [22]. In both nf1^P1 mutants and nsyb-GAL4/Nf1^RNAi flies, loss of Nf1 increases the propensity to wake, while sleep propensity is reduced or remains unchanged S2E and S2F and S3 Figs). Therefore, the phenotypes of both pan-neuronal RNAi knockdown of Nf1 and genetic mutants further support the notion that Nf1 promotes sleep and prevents sleep fragmentation.

Mounting evidence suggests that flies, like mammals, possess distinct sleep stages comprised of light and deep sleep [21,32,33]. Increased arousal threshold, the phenomenon where a stronger stimulus is required to induce movement, is a key hallmark of sleep that is conserved across phyla [34]. Longer nighttime sleep bouts are associated with elevated arousal threshold, suggesting that sleep intensity increases during longer sleep bouts [32,34]. To determine whether sleep depth is disrupted in Nf1 deficient flies, we used the Drosophila Arousal Tracking (DART) system. We first implemented a paradigm that provides sleeping flies with increasing levels of vibration stimuli to determine the magnitude of the stimulus required to awaken the fly, a metric that is independent of time spent asleep (Fig 1G; [32] Arousal threshold was reduced during the day and the night in nf1^P1 mutant flies compared to wild-type controls and heterozygotes, revealing reduced sleep depth associated with loss of Nf1 (Fig 1H). Similarly, arousal threshold was reduced during the day and night in flies with pan-neuronal knockdown of Nf1 (nsyb-GAL4/Nf1^RNAI) compared to flies harboring either transgene alone (Fig 1I). Sleep duration was also reduced in nf1^P1 mutants and upon pan-neuronal knockdown of Nf1 in this system (S4A and S4B Fig). Together, these findings reveal that neuronal Nf1 is required for normal arousal threshold.

To determine whether loss of Nf1 impacts the propensity of stimuli to awaken flies across individual sleep bouts, we measured the reactivity of flies to a fixed vibration stimulus and calculated their responsiveness as a function of time spent asleep prior to stimulus onset (Fig 1J). In control and heterozygous nf1^P1 flies, there was no effect of time spent asleep on reactivity during the day (Fig 1K). However, reactivity was significantly reduced during the night, as the slope of their respective regression lines differed significantly from zero (Fig 1L). nf1^P1 mutants had no effect on time spent asleep during the night or the day on reactivity (Fig 1K and 1L). Flies with pan-neuronal knockdown of Nf1 maintained high levels of reactivity across sleep bouts of up to 40 minutes, phenocopying nf1^P1 mutants (Fig 1M and 1N). Therefore, Nf1 is required for sleep duration-dependent changes in arousal threshold.

In both flies and mammals, sleep is associated with reduced metabolic rate [8,9,35–39]. To determine the effect of Nf1 on sleep-dependent modulation of metabolic rate, we measured metabolic rate in awake and sleeping flies using the Sleep and Activity Metabolic Monitor (SAMM) system. This system uses indirect calorimetry to measure CO₂ release, while simultaneously measuring activity via counting infrared beam crosses (Fig 2A; In agreement with previous findings, sleep was reduced in nf1^P1 mutants in the SAMM system, and the total metabolic rate (VCO₂) was elevated during the day and the night compared to controls (S5A and S5B Fig; [40]). Similar effects were observed upon pan-neuronal knockdown of Nf1 (S5C and S5D Fig). To specifically examine the effects of sleep on CO₂ output, we compared the overall CO₂ output during waking and sleep. We found that CO₂ output was significantly higher in Nf1 mutant flies during both waking and sleeping, and was consistent during the day and night (S6A and S6B Fig). Pan-neuronal knockdown of Nf1 (nsyb-GAL4/Nf1^RNAi) similarly resulted in significantly higher CO₂ output during both waking and sleeping (S6C and S6D Fig). This systematic dissection of CO₂ output into sleep/waking states suggests that Nf1 is required for the maintenance of metabolic rate.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Loss of Nf1 disrupts metabolic regulation of sleep.

The Sleep and Metabolic Monitoring (SAMM) system was used to measure metabolic rate as a function of time spent asleep. A. Overview of the SAMM system. This system records activity while simultaneously measuring CO₂ production, thereby enabling sleep and activity metrics to be paired with CO₂ output. B. Linear regression of daytime CO₂ output as a function of time asleep in nf1^P1 mutants, heterozygotes, and their respective control. There is no significant difference between the slopes of each regression line (ANCOVA with time asleep as the covariate: F_2,532 = 2.988, P<0.0512). N = 26–27. C. Linear regression of nighttime CO₂ output as a function of time asleep in nf1^P1 mutants, heterozygotes, and their respective control. The slopes of each regression line are significantly different from each other (F_2,538 = 3.847, P<0.0219). D. Linear regression of daytime CO₂ output as a function of time asleep in pan-neuronal Nf1^RNAi knockdown flies and their controls. There is no significant difference between the slopes of each regression line (F_2,805 = 1.001, P<0.3680). E. Linear regression of nighttime CO₂ output as a function of time asleep in pan-neuronal Nf1^RNAi knockdown flies and their controls. The slopes of each regression line are significantly different from each other (F_2,822 = 5.625, P<0.0037). N = 30–44. Error bars indicate ± SEM. The P-values in each panel indicates whether the slope of the regression line is significantly different from zero. White background indicates daytime, while gray background indicates nighttime. *p<0.05; **p<0.01; ****p<0.0001.

https://doi.org/10.1371/journal.pgen.1011049.g002

To directly test whether sleep-metabolism interactions are disrupted by the loss of Nf1, we measured CO₂ output over the length of a sleep bout. Metabolic rate was reduced during longer sleep bouts in control flies during the night, but did not change in nf1^P1 mutants, while there was no effect of sleep on metabolic rate during the day (Fig 2B and 2C). Similarly, pan-neuronal knockdown of Nf1 (nsyb-GAL4/Nf1^RNAi) abolished nighttime sleep-dependent changes in metabolic rate, while there was no effect of sleep on metabolic rate during the day (Fig 2D and 2E). These findings reveal a critical role for neuronal Nf1 in sleep-dependent changes in metabolic rate. Taken together, loss of Nf1 results in sleep fragmentation, reduced arousal threshold, and loss of sleep-dependent changes in metabolic rate, suggesting that Nf1 is required for flies to enter deep sleep.

While Nf1 is broadly expressed throughout the brain, its function has been linked to the modulation of GABA signaling during the formation of associative memories [41]. In Drosophila, the GABA_A receptor Resistant to dieldrin (Rdl) is expressed in numerous populations of sleep-regulating neurons in the brain as well as in the ampula of the intestine (Fig 3A; [42–44]. To examine whether Nf1 functions in GABA_A receptor neurons, we selectively knocked down Nf1 by expressing Nf1^RNAi under the control of Rdl-GAL4 and then measured its effect on sleep. Flies with Nf1 knockdown in GABA_A receptor neurons (Rdl-GAL4/Nf1^RNAi) slept less than control flies harboring either transgene alone (Fig 3B). Sleep was fragmented in Rdl-GAL4/Nf1^RNAi flies, with increased bout number, reduced bout length, and an increased propensity to wake (Figs 3C and 3D and S7). Further, when sleep was measured in the DART and SAMM systems, knockdown of Nf1 in GABA_A receptor neurons similarly reduced sleep duration (S8A and S8B Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. GABA_A receptor neurons mediate sleep depth via Nf1.

GABA_A receptor neurons were targeted using the Rdl-GAL4 driver. A. The expression pattern of Rdl-expressing neurons is visualized with GFP in the brain (left) and the gut (right). For the brain, background staining is NC82 antibody (magenta); scale bar = 100μm. For the gut, background staining is DAPI (blue); scale bar = 1000μm. (B-D). Sleep traits of Nf1^RNAi knockdown flies and their respective controls. B. There is a significant effect of genotype on sleep duration (two-way ANOVA: F_2,362 = 110.4, P<0.0001). Compared to controls, knockdown of Nf1 in Rdl-expressing neurons significantly reduces sleep and occurs during the day (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0001) and night (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0001). C. There is a significant effect of genotype on bout number (two-way ANOVA: F_2,362 = 115.6, P<0.0001). Compared to controls, knockdown of Nf1 in Rdl-expressing neurons significantly increases bout number and occurs during the day (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0001) and night (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0001). D. There is a significant effect of genotype on bout length (two-way ANOVA: F_2,362 = 37.97, P<0.0001). Compared to controls, knockdown of Nf1 in Rdl-expressing neurons significantly reduces bout length and occurs during the day (Rdl/+, P<0.0068; Nf1^RNAi/+, P<0.0001) and night (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0001). N = 57–66. (E-G) Measurements of arousal threshold and reactivity in Nf1^RNAi knockdown flies and their respective controls using the DART system. E. There is a significant effect of genotype on arousal threshold (REML: F_2,103 = 13.76, P<0.0001; N = 22–38). Compared to controls, knockdown of Nf1 in Rdl-expressing neurons significantly decreases arousal threshold and occurs during the day (Rdl/+, P<0.0003; Nf1^RNAi/+, P<0.0011) and night (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0014). N = 29–41. F. Linear regression of daytime reactivity as a function of time asleep in Nf1^RNAi knockdown flies and their controls. The intercepts of each regression line are significantly different from each other (F_2,1771 = 44.12, P<0.0001). G. Linear regression of nighttime reactivity as a function of time asleep in Nf1^RNAi knockdown flies and their controls. The intercepts of each regression line are significantly different from each other (F_2,1948 = 64.06, P<0.0001). N = 49–61. (H-J) Measurements of metabolic rate in Nf1^RNAi knockdown flies and their respective controls using the SAMM system. H. There is a significant effect of genotype on metabolic rate (two-way ANOVA: F_2,188 = 55.60, P<0.0001). Compared to controls, knockdown of Nf1 in Rdl-expressing neurons significantly increases CO₂ output and occurs during the day (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0001) and night (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0001). I. Linear regression of daytime CO₂ output as a function of time asleep in pan-neuronal Nf1^RNAi knockdown flies and their controls. The slopes of each regression line are significantly different from each other (F_2,698 = 11.17, P<0.0001). J. Linear regression of nighttime CO₂ output as a function of time asleep in pan-neuronal Nf1^RNAi knockdown flies and their controls. The slopes of each regression line are significantly different from each other (F_2,756 = 13.64, P<0.0001). N = 30–35. For violin plots, the median (solid line) as well as 25th and 75th percentiles (dotted lines) are shown. For reactivity and metabolic rate measurements, error bars indicate ± SEM. The P-values in each panel indicate whether the slope of the regression line is significantly different from zero. White background indicates daytime, while gray background indicates nighttime. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.

https://doi.org/10.1371/journal.pgen.1011049.g003

To verify that Rdl functions in the brain, we knocked down Nf1 in flies that also contained the tsh-GAL80 transgene, that blocks GAL4 activity in the enteric nervous system [45,46]. The inclusion of tsh-GAL80 removed GFP-reporter expression in the gut, but left expression of Rdl-GAL4 in the central brain (S9A and S9B Fig). Reactivity during sleep long nighttime sleep bouts remained elevated in tsh-GAL80;Rdl-GAL4/Nf1^RNAi flies, supporting the notion that the phenotype observed in flies with Rdl knockdown is not derived by loss of Nf1 in the gut (S9C and S9D Fig). Further, the expression pattern of Rdl-GAL4 was similar in control flies and those harboring Nf1^RNAi suggesting the UAS-Nf1^RNAi transgene is being faithfully expressed in Rdl-GAL4 positive neurons (S10 Fig). Therefore, knockdown of Nf1 in GABA_A receptor neurons of the brain phenocopies pan-neuronal knockdown of Nf1, suggesting that GABA-sensitive neurons contribute to the sleep abnormalities of Nf1 mutant flies.

It is possible that the effects on sleep duration and sleep depth are regulated by shared or distinct populations of neurons. Therefore, we sought to determine whether loss of Nf1 in GABA_A receptor neurons also impacts arousal threshold and sleep-dependent changes in metabolic rate. Similar to pan-neuronal knockdown, arousal threshold was reduced in Rdl-GAL4/Nf1^RNAi flies, and these flies do not decrease reactivity during long nighttime sleep bouts (Fig 3E–3G). The metabolic phenotypes of Nf1 mutant and pan-neuronal knockdown flies were also present in flies upon knockdown of Nf1 in GABA_A receptor neurons. First, total metabolic rate was significantly increased, phenocopying pan-neuronal loss of Nf1 (Fig 3H). Knockdown of Nf1 in GABA_A receptor neurons similarly resulted in significantly higher CO₂ output during both waking and sleeping states (S8C and S8D Fig). In addition, knockdown of Nf1 in GABA_A receptor neurons abolished sleep-dependent changes in metabolic rate during the day and night (Fig 3I and 3J). Therefore, Nf1 is required in GABA_A neurons to regulate sleep duration, arousal threshold, and sleep-dependent changes in metabolic rate.

In Drosophila, sleep loss has been associated with shortened lifespan [47–49]. However, a number of sleep mutants do not impact lifespan, and flies are able to survive prolonged periods of sleep deprivation [50–52]. To examine whether the disrupted sleep of Nf1-defficient flies impacts their lifespan, we measured the effects of loss of Nf1 on longevity in individually housed flies. Lifespan was significantly reduced in nf1^P1 mutant flies, as well as pan-neuronal knockdown (nsyb-GAL4/Nf1^RNAi) or GABA_A-receptor specific knockdown (Rdl-GAL4/Nf1^RNAi) of Nf1, compared to their respective controls (Figs 4A and S11A and S11B). These findings suggest that the loss of Nf1 affects sleep duration and sleep quality and results in a significantly reduced lifespan.

We next sought to measure the functional consequences of loss of Nf1. In Drosophila and mammals, chronic sleep loss is associated with deficiencies in gut homeostasis, that can result in death [49]. To measure gut integrity, flies were fed blue dye and then assayed for gut permeability [53,54]. In control flies, gut permeability remains intact in young (5 days) and aged (20 days) flies (Fig 4B and 4C). However, in nf1^P1 mutants and flies with pan-neuronal knockdown of Nf1, gut permeability significantly increased in aged flies compared to controls (S11C and S11D Fig). Similarly, knockdown of Nf1 in GABA_A receptor neurons (Rdl-GAL4/Nf1^RNAi) significantly increased intestinal permeability in aged flies (Fig 4C). Together, these findings reveal that neuronal loss of Nf1, as well as selective loss in GABA_A receptor neurons, results in increased gut permeability that has been associated with aging.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Loss of Nf1 in GABA_A receptor neurons reduces longevity and promotes aging-associated phenotypes.

A. Compared to controls, knockdown of Nf1 in Rdl-expressing neurons significantly decreases longevity (Log-Rank test: χ² = 253.4, d.f. = 2, P<0.0001). N = 57–70. (B,C) The Smurf assay was used to measure intestinal barrier dysfunction. B. Representative images depicting non-Smurf (left) and Smurf flies (right). C. There is a significant effect of genotype on intestinal permeability (two-way ANOVA: F_2,35 = 29.45, P<0.0001). Knockdown of Nf1 in Rdl-expressing neurons does not change intestinal barrier dysfunction in 5d flies (Rdl/+, P<0.0565; Nf1^RNAi/+, P<0.0648), but significantly increases in 20d flies (Rdl/+, P<0.0001; Nf1^RNAi/+, P<0.0001). N = 6–8. (D-F) ROS was measured in 5d and 20d flies by quantifying oxidized DHE. Scale bar = 500μm. D Oxidized DHE was measured in 5d control and Nf1 knockdown flies. E. There is no significant difference in oxidized DHE signal intensity in 5d flies (one-way ANOVA: F_2,53 = 3.367, P<0.0520). N = 16–22. F. Oxidized DHE was measured in 20d control and Nf1 knockdown flies. G. Knockdown of Nf1 in Rdl-expressing neurons significantly increases oxidized DHE signal intensity in 20d flies (one-way ANOVA: F_2,57 = 25.76, P<0.0001). N = 10–28. The median (solid line) as well as 25th and 75th percentiles (dotted lines) are shown. ****p<0.0001.

https://doi.org/10.1371/journal.pgen.1011049.g004

It has previously been reported that chronic sleep deprivation induces the generation of reactive oxygen species (ROS) that underlies reduced lifespan, and this can be rescued by antioxidant feeding [49]. These findings suggest that low sleep quality negatively impacts the health and longevity of Drosophila. To determine if loss of Nf1 impairs gut function, we measured ROS levels in the gut in young (5 days) and aged (20 days) flies. Pan-neuronal knockdown of Nf1 (nsyb-GAL4/Nf1^RNAi) led to increased ROS levels in both young and aged flies compared to controls (S12 Fig). Furthermore, ROS levels were also elevated in aged flies upon knockdown of Nf1 in GABA_A receptor neurons (Fig 4D–4G). To test whether antioxidant feeding in Nf1^RNAi flies would promote survival and restore gut homeostasis, we added either Lipoic Acid or Melatonin to standard Drosophila media and then measured longevity and ROS in the gut. We found that antioxidant feeding had no effect on longevity, intestinal permeability, or ROS in Rdl-GAL4/Nf1^RNAi flies (S13 Fig). These findings suggest that under the conditions we used there is no effect of antioxidants on ROS and longevity. Overall, these findings support the notion that reduced gut function contributes to the reduced lifespan associated with the loss of Nf1 and suggests that Nf1 may act upstream of the protective effects of these antioxidants on lifespan and gut function.

Given the connection between decreased sleep quality and aging traits, we tested the hypothesis that sleep induction in Nf1^RNAi flies would promote survival and restore gut homeostasis. To pharmacologically induce sleep, we added Gaboxadol, a GABA_A Receptor agonist that has been widely used to increases sleep in Drosophila [34,55,56], to the standard fly media and then measured sleep depth and aging traits. We found that, similar to control flies, feeding Rdl-GAL4/Nf1^RNAi flies Gaboxadol increased both daytime and nighttime sleep, but was unable to restore arousal threshold (S14A–S14D Fig). Interestingly, Gaboxadol feeding across the lifespan extended longevity in control flies, but had no effect on Rdl-GAL4/Nf1^RNAi flies (S14E–S14G Fig). Gaboxadol feeding also had no effect on intestinal permeability or ROS (S14H–S14J Fig). These results support the notion that the reduction in lifespan and gut homeostasis are not due to reduced sleep duration, but may derive from the loss of sleep quality as measured by arousal threshold.

In flies, diet potently impacts both sleep and metabolic regulation [57–61], and long-term dietary restriction has been shown to extend lifespan [62–64]. Given the sleep, metabolic, and gut homeostasis phenotypes of Nf1-defficient flies, we sought to determine the impacts of dietary restriction on these diverse phenotypes. Flies were fed a standard food (SF) diet or a dietary restriction diet (DR), that differs only in the amount of yeast extract [65]. We first examined whether dietary restriction can restore the lifespan of flies with knockdown of Nf1 in GABA_A receptor neurons. We found that in comparison to control food, DR significantly extends lifespan of Nf1-deficient flies, although it does not fully rescue to control levels (Fig 5A). We also measured the effects of DR on intestinal permeability and ROS. We found that DR rescued gut permeability and elevated ROS levels in Rdl-GAL4/Nf1^RNAi flies (Fig 5B–5D). This rescue of lifespan and aging traits by DR was achieved without any changes to sleep duration or sleep depth; dietary restricted Rdl-GAL4/Nf1^RNAi flies had significantly reduced arousal threshold and did not decreased reactivity during long nighttime sleep bouts (Fig 5E–5H), consistent with Rdl-GAL4/Nf1^RNAi flies on the SF diet. These findings suggest that dietary restriction is protective against the negative impacts of Nf1 deficiency, despite having no effect on sleep duration or sleep quality.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Dietary restriction in Nf1-deficient flies extends longevity and restores aging-associated phenotypes but has no effect on sleep depth.

Flies were fed either a control, standard food (SF) diet or a dietary restricted (DR) diet. A. DR significantly extends longevity in flies with knockdown of Nf1 in Rdl-expressing neurons (Log-Rank test: χ² = 263.1, d.f. = 3, P<0.0001). N = 41–55. B. DR restores intestinal permeability in 20d flies with knockdown of Nf1 in Rdl-expressing neurons (one-way ANOVA: F_3,40 = 19.82, P<0.0001). N = 10–12. (C,D) ROS was measured in 20d flies by quantifying oxidized DHE. C. Oxidized DHE was measured in 20d control and Nf1 knockdown flies fed either an SF or DR dietary regimen. Scale bar = 500μm. D. DR restores oxidized DHE signal intensity in 20d flies with knockdown of Nf1 in Rdl-expressing neurons (one-way ANOVA: F_3,59 = 9.332, P<0.0001). N = 11–22. E. There is a significant effect of genotype on sleep duration (two-way ANOVA: F_3,220 = 26.46, P<0.0001). DR has no effect on sleep duration in flies with knockdown of Nf1 in Rdl-expressing neurons during the day (P<0.6963) or the night (P<0.6145). N = 20–34. (F-H) Measurements of arousal threshold and reactivity using the DART system. F. There is a significant effect of genotype on arousal threshold (REML: F_3,339 = 16.66, P<0.0001; N = 22–38). DR has no effect on arousal threshold in flies with knockdown of Nf1 in Rdl-expressing neurons during the day (P<0.6594) or the night (P<0.7728). N = 41–46. G. Linear regression of daytime reactivity as a function of time asleep. The intercepts of each regression line are not significantly different from each other (F_3,990 = 1.088, P<0.3532). H. Linear regression of nighttime reactivity as a function of time asleep. The intercepts of each regression line are significantly different from each other (F_3,994 = 38.89, P<0.0001). N = 20–34. For sleep and gut measurements, the median (solid line) as well as 25th and 75th percentiles (dotted lines) are shown. For survival and reactivity measurements, error bars indicate ± SEM. The P-values in each panel indicate whether the slope of the regression line is significantly different from zero. White background indicates daytime, while gray background indicates nighttime. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.

https://doi.org/10.1371/journal.pgen.1011049.g005

Discussion

Clinical evidence reveals Neurofibromin type 1 (NF1) to be critical for regulating diverse biological functions, as humans afflicted with neurofibromatosis type 1 have behavioral manifestations including a high co-morbidity with ADHD, autism, learning impairments, and sleep disruption [23,26,66–70]. Studies in mammalian models have revealed a robust role for Nf1 in sleep and metabolic regulation, raising the possibility that they contribute to the complex systems in humans [71,72]. In Nf1-deficient Drosophila, metabolic rate is chronically elevated, sleep is shortened, and circadian rhythms are dysregulated in Nf1 mutants [25,29,30,40,73–75], suggesting deep evolutionary conservation of Nf1 function. The findings that Nf1 plays a conserved role in regulating both sleep and metabolic rate raises the possibility that Nf1 is a critical integrator of these processes and that it may play a role in various forms of metabolic dysfunction that are associated with sleep disturbance [76,77].

In Drosophila and mammals, sleep is associated with reduced metabolic rate [8,9,78]. In Drosophila, metabolic rate is elevated across the circadian cycle in Nf1 mutants, and this is associated with reductions in energy stores and starvation resistance [40]. Here, we have applied indirect calorimetry to examine metabolic rate across individual sleep bouts and find that Nf1 is required for reductions in metabolic rate associated with prolonged time spent asleep. These findings raise the possibility that Nf1 signaling is a sleep output that specifically serves to regulate metabolic rate. Supporting this notion, Nf1 is proposed to be an output of the circadian clock because circadian gene expression is normal in Nf1 mutants, yet flies are arrhythmic [25], and Nf1 impacts the physiology of neurons downstream of clock circuits [75]. In Drosophila, numerous populations of neurons contribute to regulating sleep and wakefulness, presenting a challenge to localizing the integration of sleep and metabolic regulation [79]. The failure of Nf1 mutant flies to integrate sleep and metabolic rate may provide a pathway to identify output from sleep neurons that regulate metabolic state.

There is growing evidence that Drosophila, like mammals, possess light and deep sleep [32]. For example, readouts of both broad electrical activity and defined neural circuits suggest light sleep associates with periods early in a sleep bout, while deeper sleep associates with periods later in a sleep bout [15,21,80]. Furthermore, periods later in a sleep bout are associated with an elevated arousal threshold that indicates deeper sleep [21]. These findings are supported by functional evidence that consolidation of sleep bouts is required for critical brain functions, including waste clearance and memory, and that these processes are impaired when sleep is disrupted [10,34,56]. Here, we provide evidence that Nf1 flies fail to enter deep sleep, including sleep fragmentation, mathematical modeling of sleep pressure, reduced arousal threshold, and a loss of sleep-dependent reductions in metabolic rate. These findings suggest that Nf1 mutants fail to enter deep sleep, even during prolonged sleep bouts. While it is possible that the lack of metabolic downscaling during sleep bouts is due to a lack of deep sleep, it is also possible that metabolic downscaling itself is required for deep sleep. Examining local field potentials and neural activity within Nf1 mutants, as has been previously described [21,80], is likely to inform the neural basis for the loss of deep sleep.

Nf1 is broadly expressed and regulates numerous behaviors and brain functions. For many behaviors, Nf1 function has been localized to different subsets of neurons, suggesting localized changes in Nf1 regulate distinct behaviors. For example, Nf1 is broadly required within the pacemaker circuit to regulate 24-hour rhythms, while Nf1 in the mushroom bodies regulates clock-dependent wakefulness [74,75]. For a number of other processes, including grooming behavior and metabolic rate, the specific population of neurons where Nf1 functions has not been identified [29,40]. The broad and diverse effect of Nf1 raises the possibility that it functions widely in many circuits, and that it may be challenging to localize its function to defined cell types. Here, we find that the metabolic and sleep phenotypes of Nf1 mutant flies are phenocopied in flies with specific loss of Nf1 in GABA_A receptor/Rdl-expressing neurons. These findings raise the possibility that sleep dysregulation is due to altered GABA signaling. Supporting this notion, GABA signaling to the mushroom bodies, the Drosophila memory center, is dysregulated in Nf1 mutants [41]. It has also previously been reported that Nf1 knockdown in GABA_A receptor neurons leads to shortened sleep bouts and reduced sleep duration [73]. These findings support the notion that Nf1 modulates GABA signaling. Future studies defining the specific population(s) of neurons where Nf1 functions may reveal novel neural mechanisms regulating sleep-dependent regulation of metabolic rate.

Epidemiological data and individuals with chronic sleep loss reveal a link between shortened sleep duration and serious health problems [81,82]. Additionally, in several model organisms, sleep restriction can lead to premature death [48,83–86]. Lifespan is reduced in Nf1 mutant flies [87], but its relationship to reduced sleep or circadian dysregulation has been unclear. In Drosophila, lifespan is reduced in short-sleeping genetic mutants and by chronic sleep deprivation [47–49]. However, there are many examples of short sleeping genetic mutants, as well as in artificially selected short-sleeping flies and chronically sleep deprived wildtype flies where lifespan is not impacted [50–52,88]. Therefore, further studies are required to more fully elucidate the interactions between sleep regulation and longevity in Nf1 deficient flies, and it remains possible that sleep loss does not directly impact lifespan.

Sleep loss induced by acute manipulations in young flies, or during aging, results in increased sensitivity to ROS, suggesting the generation of ROS, or changes in clearing ROS, may be a critical function of sleep that is necessary for survival [89–91]. Further, evidence suggests that sleep deprivation leads to increased accumulation of ROS in the gut, resulting in gut permeability and death [49]. We find the ROS and permeability phenotypes in the guts of aged Nf1 flies phenocopy those of animals that have been mechanically sleep deprived. Interestingly, chronic sleep deprivation did not change intestinal permeability, suggesting the decreased lifespan associated with chronic sleep loss is distinct from accelerated aging [49]. Evidence suggests ROS accumulation contributes to aging-related pathologies [92,93], including intestinal barrier dysfunction [53,94]. Our observations of both intestinal barrier dysfunction and ROS accumulation in the gut in aged Nf1-deficient flies suggests this genetic model may more faithfully phenocopy an accelerated aging genetic model. Together, these findings suggest that sleep consolidation, or deep sleep, is important for maintaining gut homeostasis. However, given the pleiotropic effects of Nf1 on behavior and physiology, it remains a challenge to causally link this disruption of deep sleep with dysregulation of gut homeostasis. Future studies using other, less severe alleles of Nf1, in addition to other genetic models with reduced sleep quality will further elucidate this link. Ultimately, genetic models with reduced sleep quality may resemble human sleep disorders more closely than chronic sleep deprivation.

Dietary restriction (DR) extends lifespan from flies to humans [95–98], and has potent effects on gut health, including ROS and intestinal permeability [53,99–101]. In mammals, DR increases antioxidant defense during aging [102,103]. A previous study found that Nf1 mutants have elevated ROS production and that this could be rescued by feeding antioxidants [87], supporting the notion that DR may be restorative in this genetic model. Our findings that DR can restore longevity and gut homeostasis, but not sleep depth in Nf1-deficient flies suggests sleep depth and gut homeostasis are separable. It is possible that the effects of Nf1 on gut homeostasis are downstream of its effects on sleep, and that restoration of gut homeostasis with DR restores longevity without affecting sleep. Alternatively, Nf1 may exert its effects on sleep depth and gut homeostasis through two different pathways, and defining the specific population(s) of neurons where Nf1 functions may provide additional insight into this possibility. In mammals, DR reduces ROS production during aging, but is dependent on the duration of restriction [102]. Since our measurement of sleep depth was performed in young Nf1-deficient flies that have been maintained on a restricted diet for a short period of time, it is also possible that a longer duration of DR may ameliorate sleep depth in these flies. Future studies investigating how sleep depth changes during aging and whether it can also be rescued by DR will further our understanding of the interactions between diet and sleep regulation during aging.

Taken together, our findings reveal a novel and complex role for Nf1 in regulating sleep depth. Loss of Nf1 induces multiple phenotypes classically associated with a loss of deep sleep. Human mutations in NF1 have been introduced to Drosophila and phenocopy many aspects of the human disease [40,104–106]. Future studies examining the effects of human NF1 alleles, as well as additional Drosophila alleles that result in similar interactions between sleep depth, longevity, and gut homeostasis, will help to elucidate whether these phenotypes can be dissociated. In addition, the identification of these phenotypes in Nf1 mutants allow for future work assessing the contributions of antioxidants, additional sleep promoting drugs, and sleep-regulating environmental stimuli on longevity and gut function [49,107]. Therefore, these findings establish Nf1 mutants as a model to study the function of deep sleep and provide the ability to investigate the function of disease-causing mutations on sleep regulation.

Methods

Supporting information

Acknowledgments

We are thankful to members of the Keene laboratory for helpful discussions and technical support.