Cs flies were obtained from the Bloomington Drosophila stock center. UAS-hiw^RNAi lines were obtained from the Vienna Drosophila RNAi center (VDRC; Dietzl et al., 2007). We obtained faf^EP3520, UAS-hiw^∆RING (UAS-hiw^DN)stocks from A. DiAntonio (Washington University in St.Louis, MO, USA), UAS-hiw^RNAi from the Vienna Drosophila Resource Center. GAL4 lines were selected based upon their expression throughout the brain or in neuronal populations known to be involved in sleep and plasticity. Drivers expressing broadly throughout the brain include elav-GAL4, MJ85B-GAL4, Cha-GAL4 and TH-GAL4. The Mushroom Bodies (MBs) play a role in sleep regulation and memory; drivers that express in the MBs include 247-Gal4, 30Y-GAL4, c309-GAL4, and 201y. The central complex drivers have been implicated in plasticity and include c232-GAL4 and c205-GAL4. The pars intercerebralis drivers include 50y-GAL4, c687-GAL4, Jan191-GAL4 and Feb194-Gal4. Finally, we obtained drivers expressing the circadian clock including cry39-GAL4, cry16-GAL4, cry-Gal80 stocks from M. Rosbash (Brandeis University), PDF-GAL4 flies from P. Taghert (Washington University in St.Louis, MO, USA). UAS-Rdl RNAi stocks and other GAL4 drivers were obtained from Bloomington stock center, P. Taghert, A. DiAntonio, S. Birman (Université de la Mediterannée), R. Greenspan (Neuroscience Institute, San Diego, CA, USA), A. Sehgal (University of Pennsylvania), and R. Allada (Northwestern University). For GAL4:UAS experiments, parental lines were outcrossed to Cs flies. Flies were cultured at 25°C, 50% to 60% humidity, in 12 h:12 h Light:Dark cycle, on a standard food containing yeast, dark corn syrup, molasses, dextrose and agar. Newly eclosed female adult flies were collected from culture vials daily under CO₂ anesthesia. Three day old flies were then individually placed into 65 mm glass tubes so that sleep parameters can be continuously evaluated using the Trikinetics activity monitoring system as previously described (Shaw et al., 2000)¹. Flies were sleep deprived using an automated SD apparatus that has been found to produce waking without nonspecifically activating stress responses (Shaw et al., 2002). Flies were sleep deprived using the SNAP from ZT 12 (beginning of the dark phase) to ZT 0 (beginning of the light phase). Unless otherwise stated, at least 16 flies were analyzed for each experimental condition. Differences in sleep time were assessed using either a Student’s t-test or analysis of variance (ANOVA) which were followed by planned comparisons using a Modified Bonferroni correction.

Flies were individually placed into 5 × 65 mm tubes with regular food and placed into constant conditions for 10 days. Locomotor activity was continuously recorded in 30 min bins using the Trikinetics system. Locomotor rhythms were analyzed for 6 or 9 days using MATLAB (Mathworks, Natick, MA, USA) based computational tools designed by Levine et al. (2002). At least 30 flies were analyzed for each genotype.

The APS performance test was performed as previously described (Le Bourg and Buecher, 2002). Flies are individually tested in a T-maze where they are allowed to choose between a dark and a lighted vial. Adult flies are phototaxic and choose the lighted alley in the absence of reinforcer. During the test a filter paper soaked with a quinine solution is placed in the lighted vial to provide an aversive association. In the course of 16 trials through the maze flies learn to make more frequent choices to the dark vial (photonegative choices). The number of photonegative choices is tabulated during four successive blocks of four trials and the performance score is the % of photonegative choices made in the last block of four trials. At least eight flies were evaluated for each condition. For each experiment, learning was evaluated by the same experimenter who was blind to genotype and condition. All flies were tested in the morning between ZT0 and ZT4. Flies were sleep deprived using the SNAP from ZT 12 (beginning of the dark phase) to ZT 0 (beginning of the light phase) and until each flies was tested for learning. Learning scores are normally distributed (Seugnet et al., 2009b). Differences between scores were assessed using either a Student’s t-test or ANOVA which were followed by planned comparisons using a Modified Bonferroni correction.

Photosensitivity was evaluated using the T-maze with no filter paper. The average proportion of choices to the lighted vial during 10 trials was calculated for each individual fly. The phototaxis index (PI) is the average of the scores obtained for at least 5 flies ± SEM.

Sensitivity to quinine/humidity was evaluated as Le Bourg and Buecher (2002) with the following modifications: each fly was individually placed in a 14 cm transparent cylindrical tube covered with filter paper, uniformly lighted and maintained horizontal. In one half of the apparatus the filter paper is soaked with quinine solution while the other half is kept dry. The quinine/humidity sensitivity index (referred to as Quinine Sensitivity Index or QSI) was determined by calculating the time in seconds that the fly spent on the dry side of the tube when the other side had been wetted with quinine, during a 5 min period.

Brains were removed from the head casing and fixed in 4% paraformaldehyde in phosphate buffer solution (PBS; 1.86 mM NaHPO, 8.41 mM NaHPO, and 175 mM NaCl) for 1 h and washed in PBS. Following a 2-h pre-incubation in 3% normal goat serum in PBS-TX (PBS containing 0.3% Triton X-100), brains were washed in PBS-TX. Primary antibodies were Rabbit anti-GFP (1:1000; Sigma) and mouse anti-PDF (1:50; DSHB, University of Iowa), washed in PBS-TX and incubated in the appropriate fluorescent secondary antibodies.

Confocal images with a 1 μm slice thickness were collected on an Olympus microscope provided by the Washington University Center for Cellular Imaging in St. Louis, MO, USA. Confocal stacks of PDF terminals were quantified as described previously (Donlea et al., 2009). Briefly, immuno-positive terminals were counted using the ImageJ binary thresholding algorithm. The number of synaptic terminals for the GAL4/+ parental control used to generate a mean. The mean of the GAL4/+ was used to normalize each individual UAS/+ and GAL4>UAS brain. The individual normalized values were then used to calculate the mean and standard error for the group. The normalized values for each group were then evaluated using a one-way ANOVA and a modified Bonferroni test.

All comparison were done using a Student’s T-test or, if appropriate, ANOVA and subsequent modified Bonferroni comparisons unless otherwise stated. All statistically different groups are defined as p < 0.05.

As mentioned, whole genome profiling revealed that faf was increased in ins-l flies compared to normal sleeping background controls identifying it as a potential contributor to insomnia phenotypes (Seugnet et al., 2009c). The increase of faf in ins-l flies was confirmed in an independent cohort using QPCR (data not shown). Given this expression profile, we hypothesized that over-expressing faf would recapitulate important features of the insomnia phenotype such as reduced sleep time and sleep fragmentation. The P-element line, faf^EP3520, contains a UAS that can be used to induce functional faf (DiAntonio et al., 2001). Thus, we conducted a mini-brain screen to express faf in circuits known to modulate sleep and waking in flies (Joiner et al., 2006; Pitman et al., 2006; Foltenyi et al., 2007; Sheeba et al., 2008; Donlea et al., 2011). Although previous reports demonstrate that knocking-down inc and Cul-3 with the pan-neuronal elav-GAL4 driver substantially reduced sleep time (Stavropoulos and Young, 2011; Pfeiffenberger and Allada, 2012), elav-GAL4>faf^EP3520 did not reduce night-time sleep compared to genetic controls (Figure 1A, left). However, we observed a ~20% reduction in total sleep time when we expressed faf^EP3520 using MJ85b-GAL4, a stronger driver that expresses broadly throughout the brain (Joiner and Griffith, 1997; Dubnau et al., 2001); no changes in sleep were observed when faf was expressed in cholinergic or dopaminergic neurons (Figure 1A). Surprisingly, expressing faf in other sleep/wake centers (e.g., Mushroom Bodies (MBs), Fan Shaped body (FB), Central Complex, Pars Intercerebralis (PI)) did not consistently alter sleep parameters (Figure 1A). In contrast, all five GAL4 drivers that express in clock neurons significantly reduced night-time sleep when compared to both parental controls. Indeed, the only drivers that reduced night-time sleep and also disrupted sleep consolidation when expressing faf^EP3520 were cry16-GAL4 and c929-GAL4 (Figures 1A,B). Since insomnia tends to become more prevalent with age in humans, we asked whether expressing faf in clock cells would disrupt sleep from the first day of life, similar to the results seen with inc and Cul-3, or whether the changes in sleep would develop over time in adults. In wild-type flies, sleep reaches a stable level by 3-days of age (Shaw et al., 2000). We did not observe differences in sleep in young c929-GAL4>faf^EP3520 and their parental controls during their first 2 days of adult life (data not shown). Similarly, c929-GAL4/+ (black), faf^EP3520/+ (gray) and c929-GAL4>faf^EP3520 (orange) flies slept similarly on day 3 (Figure 1C, left). However, while sleep remained stable over days in c929-GAL4/+ and faf^EP3520/+ controls, sleep progressively declined in mature c929-GAL4>faf^EP3520 flies and stabilized by 7–8 days of age (Figure 1C). Thus, disrupting the ubiquitin-proteasome system in clock cells (see below) by expressing a gain-of-function allele of faf disrupts sleep in mature adults, and recapitulates two key features seen in ins-l flies.

To further evaluate the role of ubiquitination pathways for sleep regulation, we obtained loss-of-function alleles for the E3 ubiquitin ligase highwire (hiw). As mentioned above, loss-of-function alleles for hiw phenocopy gain-of-function alleles for faf. As seen in Figures 2A,B knockdown of hiw in c929-GAL4 expressing neurons reduced sleep time using two independent RNAi lines that target a different portion of the gene. Reduced sleep was also observed when expressing a dominant negative form of highwire (hiw^DN; Figure 2C). As with gain-of-function alleles for faf, all three loss-of-function alleles for hiw also substantially disrupted sleep consolidation at night (Figure 2D). These phenotypes were not observed when expressing wild-type hiw (UAS-hiw^WT) with c929-GAL4 (data not shown).

Sleep loss impairs performance on many cognitive tasks including those that require short-term memory and response-inhibition (Frey et al., 2004; Chuah et al., 2006 Dissel and Shaw, 2013). Thus, to determine whether the reduction in sleep seen in c929-GAL4>hiw^DN flies is the result of a failed ability to generate adequate sleep or the consequence of a reduction in sleep need, we evaluated performance using Aversive Phototaxic Suppression (APS) assay. The APS is an established short-term memory assay that is extremely sensitive to sleep disruption (Seugnet et al., 2008, 2009a,c, 2011a,b; Thimgan et al., 2010, 2015; Dissel et al., 2015a,b,c, 2017). In the APS, flies are individually placed in a T-maze and allowed to choose between a lighted and darkened chamber over 16 trials. Flies that do not display phototaxis during the first block of four trials are excluded from further analysis (Le Bourg and Buecher, 2002; Seugnet et al., 2009a). During 16 trials, flies learn to avoid the lighted chamber that is paired with an aversive stimulus (quinine/humidity). The performance index is calculated as the percentage of times the fly chooses the dark vial during the last four trials of the 16 trial test and short term memory (STM) is defined as selecting the dark vial on two or more occasions during Block 4. Before being tested for STM, flies are first examined to ensure that they exhibit normal photosensitivity and quinine sensitivity (Le Bourg and Buecher, 2002; Seugnet et al., 2008, 2009b). This step is important since changes to sensory thresholds could confound the ability to detect true changes in associate learning (Kahsai and Zars, 2011; Dubnau and Chiang, 2013; Dissel et al., 2015a). As seen in Figure 2E, photosensitivity and quinine sensitivity for c929-GAL4>UAS-hiw^DN flies fall within the range seen in wild-type flies (Seugnet et al., 2008, 2009b) and are not statistically different from both parental controls. Importantly, c929-GAL4>UAS-hiw^DN flies exhibit performance deficits while c929-GAL4/+ and UAS-hiw^DN/+ parental controls display wild-type STM (Figure 2F). Thus, disrupting the ubiquitin-proteasome system in clock cells interferes with the ability of the flies to obtain needed sleep.

The pattern of expression for pdf-GAL4, cry16-GAL4 and c929-GAL4 all include the large ventral lateral clock neurons (lLNvs) indicating that the effects of faf and hiw are likely mediated through the large ventral lateral neurons (lLNvs; Grima et al., 2004; Stoleru et al., 2004). However, c929-GAL4 is expressed in other peptidergic neurons (Park et al., 2008). To determine whether faf over-expression in the lLNvs was responsible for the sleep reduction phenotype, we combined c929-GAL4 with cry-Gal80, which targets the GAL4 inhibitor GAL80 to all clock neurons (Stoleru et al., 2004). cry-GAL80 effectively removed GAL4 mediated induction in the lLNvs as assessed with a UAS-GFP reporter (Figure 3A bottom) and abolished the sleep reduction phenotype induced by c929-GAL4 with faf ^EP3520 (Figure 3B). Thus, expression of faf in the lLNvs regulates night-time sleep.

Neither inc, nor Cul3 alter free running circadian rhythms when disrupted in clock cells. To determine whether faf, and hiw would alter free running rhythms, we combined faf^EP3520, and UAS-Hiw^DN, with c929-GAL4 and PDF-GAL4. As seen in Figure 4, >40% of c929-GAL4 >faf^EP3520 and c929-GAL4 >UAS-hiw^DN where arrhythmic in free running conditions (Figure 4). Thus, in contrast to inc and Cul3, expression of faf in the lLNvs disrupts circadian locomotor rhythms.

To determine whether disrupting the ubiquitin-proteasome system would alter other aspects of sleep regulation, we evaluated sleep homeostasis following 12 h of SD. Previous studies have shown that the MB are important for sleep regulation (Joiner et al., 2006; Pitman et al., 2006). Moreover, performance in the APS involves MB neurons which are particularly sensitive to sleep loss (Seugnet et al., 2008, 2009b). As seen in Figures 5A,B, expressing faf^EP3520 or UAS-hiw^DN in the MBs using 247-GAL4 significantly attenuated sleep rebound compared to both parental controls. Previous reports suggest that sleep phenotypes are only observed when the ubiquitin-proteasome system is disrupted during development (Pfeiffenberger and Allada, 2012). Thus, we used the inducible GeneSwitch system (Osterwalder et al., 2001) to express faf^EP3520 or UAS-hiw^DN in adults. As seen in Figures 5C,D, adult GSW-elav>faf^EP3520 and GSW-elav>UAS-hiw^DN flies fed RU486 (RU+) exhibited a significantly reduced sleep rebound compared to genetically identical, vehicle fed siblings (RU−). It should be noted that we and others have consistently reported that RU does not influence a variety of phenotypes including lifespan, sleep, sleep homeostasis, short-term memory, short-term memory following SD, olfactory conditioning, phototaxis, geotaxis, locomotion, the escape response (Mao et al., 2004; Seugnet et al., 2008; Vanderheyden et al., 2013; Dissel et al., 2015c). Thus, disrupting the ubiquitin-proteasome system in adults attenuates sleep homeostasis.

The lack of a homeostatic response following SD may represent either an adaptation that allows animals to better withstand the negative effects of waking, or it may simply indicate that the flies have lost their capacity to respond normally to sleep loss (Seugnet et al., 2011a; Donlea et al., 2012; Dissel et al., 2015b). To distinguish between these two possibilities, we evaluated STM. If flies have reduced sensitivity to sleep loss, they should maintain their ability to learn after SD (Dissel et al., 2015b). However, if they are simply unable to generate the needed compensatory response they should be learning impaired (Dissel et al., 2015b; Dissel and Shaw, 2017). Unfortunately, learning could not be evaluated in faf over-expressing flies due to deficits in phototaxis. A previous study has found that expressing UAS-hiw^DN in MBs throughout development does not result in alterations in morphology (Huang et al., 2012). Thus we evaluated the effects of SD on learning in 247-GAL4/UAS-hiw^DN flies. As above, we first evaluated photosensitivity and quinine sensitivity to rule out the possibility that changes in behavior could be due to non-associative cues. As seen in Figure 5E, photosensitivity, quinine sensitivity and time to complete the task were similar between genotypes. Importantly, flies expressing hiw^DN in the MB maintain normal learning after SD while the outcrossed parental lines (247-GAL4/+ and UAS-hiw^DN/+) are significantly impaired (Figure 5F). The magnitude of the learning deficit observed in the parental lines following SD is similar to that previously reported for sleep deprived wild-type flies and flies lacking MBs (Seugnet et al., 2008). Moreover the deficits in learning following sleep loss in flies are within the range of effect sizes observed following sleep loss in humans and rodents across a number of cognitive domains (Frey et al., 2007; Fu et al., 2007; Pierard et al., 2007). Thus, reducing hiw function in the MB confers resistance to SD as measured by both sleep homeostasis and STM.

As mentioned above, we have hypothesized that Rdl may play a role in insomnia-phenotypes since Rdl is downregulated in ins-l flies (Seugnet et al., 2009c), is broadly expressed in the brain, including in neurons that impact sleep and waking (Liu et al., 2007; Parisky et al., 2008), and can enhance memory under some circumstances (Liu et al., 2007). While knocking down Rdl in lLNvs has been shown to increase waking, its role in other sleep wake-circuits and its impact on sleep homeostasis remains unclear (Agosto et al., 2008; Parisky et al., 2008; Chung et al., 2009). Thus, we conducted a mini-brain screen to express Rdl^RNAi in brain structures associated with sleep regulation. Consistent with previous reports, knocking down Rdl in clock neurons reduced night-time sleep and increased sleep fragmentation (Figures 6A,B).

To further characterize the effects of knocking down Rdl on sleep homeostasis and STM, we focused on c929-GAL4 and 247-GAL4. As seen in Figure 7A, knocking down Rdl in c929-GAL4 expressing cells significantly reduces total sleep time compared to c929-GAL4/+ and UAS-Rdl^RNAi/+ parental controls as previously described. Similar results were observed with two other RNAi line (Rdli4-5 and Rdl³¹²⁸⁶; data not shown). Although c929-GAL4>UAS-Rdl^RNAi flies are short sleepers, 12 h of additional sleep loss is not accompanied by a larger sleep rebound compared to parental controls (Figure 7B). Importantly, short-sleeping c929-GAL4>UAS-Rdl^RNAi flies exhibit intact STM which is not adversely affected by an additional 12 h of SD (Figure 7C). Note that while performance in c929-GAL4>UAS-Rdl^RNAi flies appears to be slightly lower than parental lines, the difference is not significant and well within the range observed for wild-type flies (Seugnet et al., 2008). Thus, in contrast to the learning impairments seen following disrupting faf and hiw in lLNvs, knocking down Rdl seems to protect flies from the deleterious effects of waking as measured by both a reduced sleep rebound and intact STM.

To determine if knocking down Rdl in the MBs would impact sleep homeostasis or STM, we expressed UAS-Rdl^RNAi using 247-GAL4; knocking down Rdl in the MBs throughout development does not result in obvious morphological defects (Liu et al., 2007). As seen in Figure 7D, knocking down Rdl in the MBs did not affect total sleep time. However, knocking down Rdl in the MBs altered sleep regulation as evidenced by a significantly reduced sleep rebound compared to 247-GAL4/+ and UAS-Rdl^RNAi/+ parental controls (Figure 7E). Although both 247-GAL4/+ and UAS-Rdl^RNAi/+ parental controls displayed impaired STM following SD 247-GAL4>UAS-Rdl^RNAi flies maintained normal STM both before and after sleep loss (Figure 7F). No changes were seen in photosensitivity of quinine sensitivity indicating that the observed results were not due to changes in sensory thresholds (data not shown). Together these data suggest that knocking down Rdl both in clock neurons (lLNvs) and in the MBs can protect flies from the deleterious effects of SD.

Previous studies have reported that SD increases synaptic markers (Gilestro et al., 2009; Bushey et al., 2011). Notably, the impact of sleep loss on synaptic plasticity can be readily examined by quantifying pigment dispersing factor (PDF) positive terminals in projections from the lLNvs (Vanderheyden et al., 2013). As noted above, behavioral plasticity is disrupted in short-sleeping c929/+>UAS-hiw^RNAi/+ flies but is preserved in short sleeping c929-GAL4/+>UAS-Rdl^RNAi/+ flies (Figures 2A,F, 7A,C). These data lead to the hypothesis that c929/+>UAS-hiw^RNAi/+ flies should exhibit an increase in the number of PDF-positive terminals typical of sleep deprived flies while c929-GAL4/+>UAS-Rdl^RNAi/+ should be unaffected. As seen in Figures 8A–C, short-sleeping c929/+>UAS-hiw^RNAi/+ flies display an increase in the number and size of PDF positive punctae compared to both of their normal sleeping parental controls (c929/+, UAS-hiw^RNAi/+). In contrast, the number, size and intensity of PDF-positive terminals is preserved in short-sleeping c929-GAL4/+>UAS-Rdl^RNAi/+ flies compared to their normal sleeping parental controls (c929-GAL4/+>UAS-Rdl^RNAi/+; Figure 8D). Thus, knocking down Rdl in lLNvs protects flies from the negative impact of waking as measured both by STM and by examining structural plasticity. In contrast, the increase in synaptic number and morphology induced by knocking down hiw in the lLNvs is deleterious to behavioral and structural plasticity.