Research Article

Identification of SLEEPLESS, a Sleep-Promoting Factor

See allHide authors and affiliations

Science  18 Jul 2008:
Vol. 321, Issue 5887, pp. 372-376
DOI: 10.1126/science.1155942

Abstract

Sleep is an essential process conserved from flies to humans. The importance of sleep is underscored by its tight homeostatic control. Through a forward genetic screen, we identified a gene, sleepless, required for sleep in Drosophila. The sleepless gene encodes a brain-enriched, glycosylphosphatidylinositol-anchored protein. Loss of SLEEPLESS protein caused an extreme (>80%) reduction in sleep; a moderate reduction in SLEEPLESS had minimal effects on baseline sleep but markedly reduced the amount of recovery sleep after sleep deprivation. Genetic and molecular analyses revealed that quiver, a mutation that impairs Shaker-dependent potassium current, is an allele of sleepless. Consistent with this finding, Shaker protein levels were reduced in sleepless mutants. We propose that SLEEPLESS is a signaling molecule that connects sleep drive to lowered membrane excitability.

Insufficient and poor-quality sleep is an increasing problem in industrialized nations. Chronic sleep problems diminish quality of life, reduce workplace productivity, and contribute to fatal accidents (1). Although the biological needs fulfilled by sleep are unclear (2), they are likely to be important, because sleep is conserved from flies to humans (37) and prolonged sleep deprivation can lead to lethality (810). Identifying mechanisms that control sleep may lead to novel approaches for improving sleep quality.

Sleep is regulated by two main processes: circadian and homeostatic (11, 12). The circadian clock regulates the timing of sleep, whereas the homeostatic mechanism controls the need for sleep. Homeostatic pressure to sleep increases with time spent awake and decreases with time spent asleep. Homeostatic control is thought to influence sleep under normal (baseline) conditions as well as recovery (rebound) sleep after deprivation. However, the molecular mechanisms underlying homeostatic regulation of sleep have remained unclear.

A powerful approach to unraveling a poorly understood biological process is to conduct unbiased genetic screens to identify molecules required for that process. The Drosophila model for sleep is well suited for such an approach, which proved invaluable for elucidation of the molecular basis of the circadian clock. Although several Drosophila genes have been implicated in sleep regulation [for example, (1315)], only one of these, the gene encoding the Shaker (Sh) K+ channel, was isolated as a result of a genetic screen (16). A mutation in this gene causes one of the shortest-sleeping phenotypes known, validating the use of screens and suggesting that control of membrane excitability is a critical requirement for sleep.

Using a large-scale, unbiased genetic screen, we identified a gene, sleepless (sss), which is required in Drosophila for both normal baseline sleep and rebound sleep after deprivation. We find that sss encodes a brain-enriched, glycosylphosphatidylinositol (GPI)–anchored membrane protein. We also show that quiver (qvr), a mutation causing impaired Sh-dependent K+ current (17, 18), is an allele of sss, and that Sh protein levels are reduced in sss mutant flies. We propose that the SSS protein signals homeostatic sleep drive by enhancing K+ channel activity and thus reducing neuronal excitability.

Identification of sss. To identify genes involved in sleep regulation, we carried out a forward genetic screen for Drosophila mutants with reduced daily sleep. We screened ∼3500 mutant lines bearing transposon insertions. A histogram summarizing the daily sleep of these lines is shown in Fig. 1A. We selected for further study the mutant line with the lowest amount of daily sleep, which we named sleepless (sss). To homogenize the genetic background, we outcrossed this strain five times into an isogenic wild-type strain, iso31, a line generated specifically for use in behavioral experiments (19). Both daytime and nighttime sleep were severely reduced in both male and female sss mutants relative to background controls (Fig. 1B). Indeed, a small percentage of sss flies (∼9% for both males and females) in our assay did not sleep at all—a phenotype never seen in control flies. To our knowledge, sss mutants exhibit the most extreme reduction in daily sleep (>85% for males and >80% for females; Fig. 1C) attributable to a single gene mutation.

Fig. 1.

Sleep phenotype of sss mutants. (A) Histogram showing the distribution of daily sleep for ∼3500 mutant lines (∼8 female flies per line). For each line, daily sleep is shown as the difference from the mean of a group of about 100 to 250 lines tested simultaneously. The arrow indicates the sss mutant line. (B) Sleep profile in 30-min intervals for sss flies (open diamonds) versus background controls (ctrl, solid diamonds). Data for male (M) and female (F) flies are shown. The bar below the x axis indicates 12-hour light and 12-hour dark periods. (C) Daily sleep amount for control (162 males and 148 females), control/sss (111 males and 113 females), and sss flies (146 males and 148 females). Data from the same flies are shown in (C) to (F). (D to F) Activity counts per minute awake (D), sleep bout duration (E), and daily number of sleep bouts (F) for male and female control, control/sss, and sss flies. In this and subsequent figures, error bars represent SEM. *P <0.05, **P < 0.0001. For (C), (E), and (F), significance level is shown for sss mutants compared to both control and control/sss flies. For (D), significance level is shown for pairwise comparisons as indicated by lines. In (E), sleep bout duration, which is not normally distributed, is presented as simplified box plots. The line inside each box indicates the median; the top and bottom represent 75th and 25th percentiles, respectively. About 9% of flies exhibiting zero sleep were excluded from calculation of sleep bout duration.

Despite this extreme reduction in daily sleep, waking activity (defined as activity counts per minute awake) was not significantly elevated in this mutant (Fig. 1D), which suggests that the mutant is not hyperactive when awake (20). The marked decrease in sleep amount was largely due to a sharp reduction in the duration of sleep bouts (Fig. 1E). However, decreased sleep in the sss mutant was also attributable in part to a significant reduction in the number of daily sleep bouts (Fig. 1F). These phenotypes are recessive in mutant animals, because flies bearing one copy of the sss mutation behaved similarly to background controls (Fig. 1, C to F).

SSS is a brain-enriched, GPI-anchored protein. sss mutants bear a P-element insertion (EY04063, which we refer to as P1) in the open reading frame of a gene designated CG33472 by the Drosophila Genome Project. The genomic structure of this gene consists of two noncoding exons and five coding exons, the last of which also contains a 3′ untranslated region (3′UTR) predicted to be ∼3.9 kb (Fig. 2A). In addition to the original P1 insertion line, there is a second line, which we call P2, bearing a transposon insertion (f01257) in the 3′UTR. The SSS protein is predicted to contain a signal peptide, an N-type glycosylation site, and a potential GPI attachment site (Fig. 2, B and C). SSS is well conserved in other insect species, and there is a potential Caenorhabditis elegans homolog (F31F6.8 in Wormbase, 46% similarity for amino acids 51 to 133) but no obvious vertebrate homologs. Nonetheless, there may be functional vertebrate homologs with conserved downstream pathways.

Fig. 2.

sss encodes a brain-enriched, GPI-anchored protein. (A) Schematic of the genomic structure of the sss locus. Noncoding regions of the cDNA are shaded; coding regions are shown in white. (B) Schematic of structural features of the SSS protein. The primary sequence contains a predicted signal peptide, an N-type glycosylation site (ψ), and a potential GPI attachment site (*). (C) Amino acid sequence of SSS (GenBank accession number EU816195). Amino acids 1 to 32 constitute the predicted signal peptide (boxed); the predicted N-type glycosylation site is underlined. Asterisk denotes the predicted GPI attachment site. Abbreviations for amino acids: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; Y, Tyr. (D) Glycosylation of the SSS protein. Western blot analysis with SSS antibody revealed two bands detected in head extracts from wild-type (ctrl) but not sss flies. Deglycosylation of head extracts by treatment with PNGase F resulted in detection of a single band. Because our antibody to SSS does not recognize glycosylated SSS well, Western blots were treated with peptide N-glycosidase F (PNGase F) before being probed with antibody to SSS. In this and subsequent Western blots, antibody to mitogen-activated protein kinase (MAPK) was used to control for loading. (E) Surface expression of SSS in cultured Drosophila cells. S2R+ cells were transfected with a pIZ-sss construct and stained with or without permeabilization to assay for total or surface expression, respectively. Transfection with the pIZ vector alone shows specificity of our SSS antibody. (F) Reduced surface expression of SSS after PI-PLC treatment. S2R+ cells transfected with a pIZ-sss construct were stained without permeabilization after PI-PLC (+) or mock (–) treatment. (G) Release of SSS into the culture medium by PI-PLC. Western blot analysis of S2R+ cells transfected with pIZ-sss was performed after PI-PLC (+) or mock (–) treatment. (H) Enrichment of SSS expression in brain and head versus body. An equal amount of total protein (∼40 μg) was loaded per lane. The experiments in (D) through (H) were performed three or four times with similar results.

To characterize the SSS protein, we used a peptide antigen to generate an antibody (21). This antibody recognized two bands on Western blots of wild-type head extracts that were not detectable in sssP1 mutant extracts (Fig. 2D), which suggests that sssP1 is a severe hypomorph or null allele. Because SSS contains a consensus site for N-type glycosylation, we deglycosylated proteins from head extracts and examined SSS mobility by Western blotting. Under these conditions, only a single band of a lower apparent molecular weight than the two untreated bands was detectable (Fig. 2D), indicating that SSS is glycosylated in vivo.

Because sss also contains a potential GPI attachment site, we next examined subcellular localization of SSS. Transfection of Drosophila S2R+ cells with a wild-type sss construct and staining with the SSS antibody under nonpermeabilizing conditions revealed a subset of the SSS protein expressed on the cell surface (Fig. 2E). Treatment of the cells with phosphatidylinositol-specific phospholipase C (PI-PLC) resulted in severe reduction of surface expression (Fig. 2F) and release of the SSS protein into the culture medium (Fig. 2G). These results show that the SSS protein is attached to the extracellular surface of the plasma membrane with a GPI anchor and can be released by cleavage with PLC.

Using our SSS antibody, we found that SSS protein levels are enriched in fly brain and head relative to body (Fig. 2H). Consistent with these findings, sss mRNA expression is enriched by a factor of 23 to 42 in brain relative to whole fly [Adult Drosophila Gene Expression Atlas (22)]. SSS protein levels did not cycle in a circadian fashion, nor did they change after sleep deprivation (fig. S1, A and B) (see below).

Genetic analysis of sleepless. To determine whether the sleep phenotype maps to the sss locus, we crossed sssP1 to two deficiencies that remove the locus. As predicted, both deficiencies failed to complement the short-sleeping phenotype of sssP1 (fig. S2, A and B). To confirm that the sleep phenotype in sssP1 mutants is caused by disruption of the sss gene, we mobilized the P-element to generate precise and imprecise excision lines. Precise excision of the P-element restored daily sleep amount in sss mutants to wild-type levels (Fig. 3A and fig. S2C). We also obtained an imprecise excision allele (Δ40) that removes part of the sss coding region and is likely to be a null allele (fig. S2D). Consistent with this interpretation, sssΔ40 mutants produced an undetectable level of the SSS protein (Fig. 3B). Sleep in this mutant is reduced as severely as in the P1 mutant; the phenotype maps to the sss gene, because the Δ40 allele failed to complement the P1 allele (Fig. 3A and fig. S2C).

Fig. 3.

Genetic analysis of sss. (A) Daily sleep amount for precise excision (Pr, n = 26), sssΔ40 imprecise excision (Im, n = 15), precise/sssP1 (Pr/sssP1, n = 24), and imprecise/sssP1 (Im/sssP1, n = 35) female flies. (B) Western blot analysis of SSS protein levels. Similar levels of SSS protein are seen in head extracts from background control (ctrl) and precise excision (Pr) flies. SSS protein is undetectable in sssP1 and sssΔ40 imprecise excision (Im) flies. Similar results were obtained in two additional experiments. (C) Daily sleep amount for female sssP1 mutant flies with (TG1, n = 15; TG2, n = 8; TG3, n = 16) or without (n = 16) agenomic sss transgene. TG1, 2, and 3 refer to three independent transgene insertions, and one or two copies of the transgene were present in the flies tested. (D) Daily sleep amount for sssP2 (n = 110) versus background control (ctrl, n = 80) as well as control/sssP1 (n = 80) versus sssP2/sssP1 (n = 112) female flies. (E) Reduced levels of SSS protein in sssP2 and trans-heterozygous sssP2/sssP1 flies. Similar results were obtained in three additional experiments. Data from male flies of the genotypes shown in (A), (C), and (D) are shown in fig. S2. *P < 0.05, **P < 0.0001.

We next tested whether expression of wild-type SSS from a transgene could rescue the sleep phenotype of sssP1 mutants. Daily sleep amount was fully rescued to wild-type levels in sssP1 mutants carrying a genomic sss transgene (Fig. 3C and fig. S2E). Together with the results of the deficiency and excision experiments, the rescue data provide strong evidence that disruption of the sss gene is responsible for the marked reduction in sleep in sssP1 mutants.

As described above, sssP2 mutants harbor an independent transposon insertion in the 3′UTR of the sss gene. Homozygous sssP2 mutant females had amounts of daily sleep similar to those of controls, whereas mutant males had slightly lower amounts of sleep than controls (Fig. 3D and fig. S2F). In contrast, sssP2/sssP1 trans-heterozygous mutants had a ∼30% reduction in daily sleep relative to control/sssP1 flies. These data suggest that the P2 insertion is a weaker allele than the original P1 insertion. To examine the biochemical basis of this possibility, we performed Western analysis on head lysates from mutant and control flies. As noted above, the P1 insertion severely reduced baseline sleep and rendered SSS undetectable (Figs. 2D and 3B). In contrast, the P2 insertion, which had a minimal effect on baseline sleep, caused a moderate reduction in the level of SSS protein relative to control flies (Fig. 3E). Finally, trans-heterozygous sssP1/sssP2 flies, which exhibit a ∼30% reduction in sleep, had a greatly reduced but still detectable level of SSS protein. These data suggest that the amount of daily sleep is correlated with the level of SSS protein and that large reductions of SSS protein are necessary to cause a substantial change in daily sleep.

Reduced homeostatic response in sss mutants. We next sought to determine whether sss mutants have defects in their homeostatic response to sleep deprivation. We did not observe rebound sleep in sssP1 flies, but sssP1 flies do not have much sleep to deprive. Thus, we tested sssP1/sssP2 trans-heterozygous flies, which still have moderate amounts of sleep, as well as sssP2 homozygotes, which have essentially normal amounts of sleep.

Mechanical stimulation resulted in equivalent sleep loss in sssP2 homozygous flies and controls; sleep loss was moderately reduced in sssP1/sssP2 flies relative to controls (Fig. 4A and fig. S3A). Whereas control flies showed substantial rebound sleep after deprivation, sssP1/sssP2 flies had little or none (Fig. 4B and fig. S3B). Unexpectedly, we observed a similar lack of rebound sleep in sssP2 homozygous flies. In addition, when lights were turned on, control animals went to sleep faster after deprivation, but this effect was significantly less pronounced or nonexistent in sssP2 and sssP1/sssP2 mutants (Fig. 4C and fig. S3C).

Fig. 4.

Reduced homeostatic response to sleep deprivation in female sss mutants. (A) Amount of sleep lost during 6 or 12 hours of deprivation by the end of the dark period for background control (ctrl), sssP2, control/sssP1, and sssP2/sssP1 flies. Data from 13 to 56 female flies are shown. (B) Amount of sleep gained during 6 hours of recovery after deprivation as in (A). (C) Change in sleep latency after deprivation relative to undisturbed controls as in (A). Sleep latency is defined as the time between the end of deprivation (which coincided with light onset) and the start of a sleep bout. Data from male files are shown in fig. S3. *P <0.05, **P < 0.001.

Although other genes have been suggested to play a role in homeostatic regulation of sleep, assessment of rebound sleep in animals bearing mutations in these genes is often confounded by concomitant reductions in baseline sleep (13, 16, 23, 24). The amount of rebound sleep generally increases with sleep lost (25, 26). Thus, when comparing the effects of sleep deprivation in animals with different amounts of baseline sleep (which leads to loss of different amounts of sleep), it is unclear whether rebound sleep should be compared in absolute terms or relative to amount of sleep lost. We have circumvented this problem by using the sssP2 mutant to study the contribution of SSS to sleep homeostasis. The finding that sssP2 animals exhibit markedly reduced rebound sleep, but minimally affected baseline sleep, provides strong evidence that sleep homeostasis is impaired in these mutants.

Effect of sss on other behaviors and longevity. To further characterize sss mutants, we examined several other behavioral phenotypes. Because mutations in certain central clock genes cause baseline and rebound sleep phenotypes (9, 2730), we analyzed the circadian rhythm phenotypes of sss mutants. Whereas sssP1 mutants exhibited weak rhythms, almost all sssP1/sssP2 trans-heterozygous mutants, which displayed a ∼30% reduction in daily sleep time, were rhythmic (Fig. 5, A and B, and table S1). Furthermore, daily oscillations in the level of PERIOD (PER) protein in the ventral lateral neurons (clock cells) remained intact in sssP1 mutants (Fig. 5C), which suggests that the reduced behavioral rhythmicity seen in these mutants is not due to a defect in the central clock.

Fig. 5.

Circadian rhythm and longevity phenotypes of sss mutants. (A) Average activity records for background control (ctrl, n = 64) and sssP1 male flies (n = 81) assayed in constant darkness (DD). The activity records are double-plotted so that each horizontal line represents data for 2 days. The gray and black bars above each activity record indicate subjective day and night, respectively. (B) Activity records showing average activity in DD for control/sssP1 and control/sssP2 (n = 76) versus sssP2/sssP1 (n = 65) male flies. Circadian data for control/sssP1 and control/sssP2 flies were statistically similar and thus were pooled. (C) Cycling of PER protein in large ventral lateral neurons in control and sssP1 mutants. Ventral lateral neurons for control and sssP1 flies were stained for PER at indicated Zeitgeber times (ZT). PER protein levels are elevated at ZT2 and ZT20 and are low at ZT8 and ZT14. (D) Survivorship curves of background control (solid diamonds) and sssP1 (open diamonds) flies. Female sss flies (n = 187) show a significantly shorter life span (P < 0.0001) than controls (n = 198). Data from male flies are shown in fig. S5.

Several other behaviors that we tested also appear normal. We found that the phototactic responses of sssP1 mutants are similar to those of controls (fig. S4A) and that sssP1 mutants perform as well as controls in a taste discrimination assay (fig. S4B). sssP1 flies (n = 43) did not exhibit a bang-sensitive paralytic phenotype, whereas 89% (n = 56) of easily shocked (eas1) flies used as a positive control did exhibit this phenotype. On the other hand, the sssP1 mutants appeared somewhat uncoordinated, and fewer mutants were able to climb a specific distance in given amounts of time relative to controls (fig. S4C). However, despite their apparent difficulties with coordination, sssP1 mutants spent more time walking than controls and were capable of flying and mating. Consistent with the widely held view that sleep serves essential biological functions, sssP1 mutants also exhibited a shortened life span relative to background controls (Fig. 5D and fig. S5).

sss is allelic to qvr and affects Sh expression. Because two short-sleeping mutants, Sh and Hyperkinetic (Hk), exhibit ether-induced leg shaking, we assayed sss mutants for this phenotype (16, 23). We found that both sssP1 and sssP2 mutants show ether-induced leg shaking. Notably, qvr, a mutant for which the underlying molecular defect is unknown, also has a leg-shaking phenotype, and this phenotype has been mapped close to sss (17). Because qvr mutants exhibit impaired Sh-dependent K+ current (18), identification of qvr as an allele of sss would implicate Sh as an effector of SSS function.

Genetic and molecular analyses confirmed that qvr is indeed an allele of sss. The qvr mutation failed to complement sssP1 for the leg-shaking phenotype. Similarly, after being outcrossed five times, qvr mutants showed a significant decrease in sleep relative to wild-type controls, and sssP1/qvr trans-heterozygotes showed a further reduction in sleep (Fig. 6A and fig. S6).

Fig. 6.

sss is allelic to qvr and affects Sh expression. (A) Daily sleep amount for qvr (n = 31), versus background control (n = 32) as well as control/sssP1 (n = 30) versus qvr/sssP1 (n = 32) female flies. **P < 0.0001. (B) Altered sss transcripts in qvr mutants. RT-PCR products were obtained with qvr and background control (ctrl) RNA and water was used as a negative control (neg). (C) Schematic representation of sss transcripts in qvr mutants. qvr 1, 2, and 3 correspond to the top, middle, and bottom bands, respectively. In background control transcripts, 163 nucleotides of intron 6 are spliced out. In contrast, the entire intron is present in qvr 1 transcripts. In qvr 2 and 3 transcripts, splice donor sites differ from the one used in wild-type control transcripts, as indicated by the nucleotide numbers for splice sites. (D) Sequence change in qvr genomic DNA in intron 6 of sss. The fifth nucleotide in intron 6 has a G → A transition. (E) Altered expression of SSS in qvr mutants. Fly head extracts from background control, qvr, and sssP1 flies were analyzed by Western blotting with SSS antibody. (F) Reduced expression of Sh in sss mutants. Western blot analysis of head extracts with Sh antibody reveals a Sh-specific band that is substantially reduced in sssP1 mutants relative to background control flies. Sh14 flies were used to identify a Sh-specific band, and Hk1 flies were used as an additional control. Nonspecific bands (*) may have obscured additional Sh bands. The experiments in (E) and (F) were performed three times with similar results.

We next investigated the molecular basis of the qvr mutation. Reverse transcription polymerase chain reaction (RT-PCR) of sss transcripts in qvr mutants produced three bands, whereas that of wild-type sss transcripts produced a single band (Fig. 6B), indicating splicing defects in qvr mutants. None of the three qvr bands showed the same electrophoretic mobility as the wild-type control band. Sequencing of the RT-PCR products revealed altered splicing of the last intron (intron 6) of sss in the qvr mutant (Fig. 6C). A single base change found in the intron is likely to be responsible for the defective splicing (Fig. 6D). Only one of the three qvr transcripts (qvr 2) is predicted to be in frame (resulting in an insertion of 21 amino acids) and thus has the potential to produce functional SSS protein. Western analysis of qvr mutants revealed a small amount of SSS with a slightly higher apparent molecular weight than wild-type SSS protein, which may correspond to the product of the inframe qvr 2 transcript (Fig. 6E).

Because qvr mutants were shown to have severely reduced Sh-dependent K+ current (18), we examined whether Sh protein levels are affected in sss mutants. We found that one form of Sh protein is expressed at a substantially reduced level in sssP1 mutants relative to wild-type flies (Fig. 6F), which suggests that SSS affects Sh at least in part through its protein expression. These results establish SSS as an important regulator of the Sh K+ channel.

Discussion. We have identified a Drosophila gene required for homeostatic regulation of sleep under normal conditions and after sleep deprivation. Although genes have been identified that regulate sleep-wake stability and baseline sleep amount, few have been shown to be important for sleep rebound (13, 15, 3135). Thus, further analysis of SSS function may provide a rare opportunity to gain mechanistic insight into the homeostatic regulation of sleep.

It is worth noting that sssP2 animals show a moderate reduction in SSS protein and a minimal reduction in baseline sleep, but have severely reduced sleep rebound. The differential requirement for SSS protein in normal versus rebound sleep may be explained in the context of the two-process model of sleep regulation, where sleep is postulated to be controlled by the opposing influences of circadian waking drive and homeostatic sleep drive (11, 36). In this context, for early-morning rebound sleep to occur, a strong homeostatic signal promoting sleep would be required to counteract a strong circadian input keeping the flies awake. At night, when circadian waking drive is weaker or absent, a relatively low level of homeostatic input may suffice to allow flies to sleep. The moderate level of SSS protein in sssP2 mutants may be within the range where sleep is possible when a wake-promoting circadian signal is low (at night), but not when it is high (in the early morning). In contrast, sssP1 and sssΔ40 mutants, which have undetectable levels of SSS expression, display severe reductions in both baseline and rebound sleep. In these mutants, the sleep-promoting signal may be too low to allow flies to sleep even when the circadian waking drive is weak at night.

Clues to the role of SSS at the cellular level come from our biochemical characterization of this molecule. The SSS protein is a GPI-anchored membrane protein enriched in the brain. GPI-anchored proteins can function as ligands or coreceptors and can also act as diffusible signals after cleavage of the GPI anchor (37, 38). Although we were unable to detect circadian or homeostatic regulation of the total levels of SSS protein, such regulation may occur at the level of cleavage of the GPI anchor. Regulation of release is known to be controlled by time of day for other proteins that do not cycle in overall levels, such as pigment-dispersing factor, a molecular output of clock neurons (39). Alternatively, SSS may be regulated in a subset of cells that express it, which would be undetectable on our Western blots.

A potential mechanism by which SSS regulates sleep is suggested by our finding that qvr is an allele of sss and that Sh protein levels are reduced in sss mutants. Furthermore, qvr mutants exhibit markedly impaired Sh-dependent K+ current at the larval neuromuscular junction (18). Thus, we propose that SSS lowers membrane excitability by modulating K+ channel expression and activity. It is striking that among thousands of mutants screened in Drosophila, two with the strongest sleep phenotypes affect the Sh K+ channel (16) and its putative regulator, sss. Reduced membrane excitability may thus be a central feature of sleep. Collectively, our data suggest that SSS is a signaling molecule that links homeostatic sleep drive to neuronal excitability.

Supporting Online Material

www.sciencemag.org/cgi/content/full/321/5887/372/DC1

Materials and Methods

Figs. S1 to S6

Table S1

References

References and Notes

View Abstract

Navigate This Article