Genotype Influences Day-to-Day Variability in Sleep in Drosophila melanogaster

Genotype Influences Day-to-Day Variability in Sleep in Drosophila melanogaster Abstract Patterns of sleep often vary among individuals. But sleep and activity may also vary within an individual, fluctuating in pattern across time. One possibility is that these daily fluctuations in sleep are caused by the underlying genotype of the individual. However, differences attributable to genetic causes are difficult to distinguish from environmental factors in outbred populations such as humans. We therefore employed Drosophila as a model of intra-individual variability in sleep using previously collected sleep and activity data from the Drosophila Genetic Reference Panel, a collection of wild-derived inbred lines. Individual flies had significant daily fluctuations in their sleep patterns, and these fluctuations were heritable. Using the standard deviation of sleep parameters as a metric, we conducted a genome-wide association study. We found 663 polymorphisms in 104 genes associated with daily fluctuations in sleep. We confirmed the effects of 12 candidate genes on the standard deviation of sleep parameters. Our results suggest that daily fluctuations in sleep patterns are due in part to gene activity. intra-individual variability, sleep standard deviation, Drosophila melanogaster, sleep Statement of Significance Previous studies in humans have observed intra-individual variability in sleep phenotypes, and there may be a connection between this variability and both sleep and mental health disorders. This study, conducted in Drosophila, reveals that day-to-day fluctuations in sleep are due, in part, to the underlying genotype of the individual. We identified 104 candidate genes for intra-individual variability, some of which have predicted homology with human genes. We confirmed the effects of 12 of these genes on intra-individual variability. Daily fluctuations in sleep in flies are thus regulated in part by genes, and these results may extend to human sleep as well. Introduction Although the need to sleep is persistent within a species, the manner in which that need is fulfilled often varies considerably among individuals. Modern-day human populations living in pre-industrial environments have characteristic sleep patterns consistent with ecological needs [1]. Sleep in some groups features a single bout of sleep at night with rare instances of napping [2], whereas others have a combination of night sleep bouts and daily napping [1, 3]. Historical evidence suggests that humans typically had two sleep bouts during the night [4], and this differs from the modern concept of a single night sleep bout. Although different among individuals, some measures of sleep and sleep-related characteristics are remarkably stable within individuals and are characterized by low within-individual variance [5–8]. However, some characteristics of sleep are highly variable within an individual. Often measured as the standard deviation or coefficient of variation [9–14], total sleep time or time in bed [13, 15–17], sleep onset latency [13, 15, 16], sleep quality or efficiency [15, 16], daytime sleepiness [17], and wake after sleep onset [13, 15, 16] can be more variable within a single individual than among individuals. This intra-individual variability has been observed using sleep diaries, actigraphy, and polysomnography in both children [13, 16, 17] and adults [13, 15]. Although the causes and consequences of daily fluctuations in sleep are not understood, one intriguing possibility is that high intra-individual variability in sleep is a hallmark of sleep disorder and mental illness. Greater variability in night-to-night sleep characteristics has been observed in those with chronic insomnia than in controls [18, 19], and patients with post-traumatic stress disorder (PTSD) exhibit exceptionally fragmented sleep patterns [20]. Daily fluctuations in obstructive sleep apnea (OSA) severity class and periodic leg movements have been observed as well [21, 22]. Variability in sleep duration has been observed in those with bipolar disorder [23, 24], as has interdaily stability, the similarity of activity patterns across days [24]. Intra-individual variability in activity patterns, including the amount, predictability, and intricacy, is also associated with bipolar disorder (reviewed in Ref. [25]). Intra-individual variability in sleep onset time and duration in children could be used to classify those with attention deficit hyperactivity disorder (ADHD), a distinction that could not be made using the averages of sleep parameters [14]. Correlates between physiological endophenotypes and high intra-individual variability in sleep characteristics have also been described, including flattening cortisol responses [26], increasing blood concentrations of inflammatory markers [27], and abdominal obesity [28]. Stress, work demands, iron deficiency, and moderate exercise can all mediate intra-individual variability in sleep [10–12, 29, 30]. Thus, intra-individual variability in sleep can be said to be plastic, i.e., it can be altered in response to changes in the environment [31]. The degree to which an individual can adapt to changing environmental conditions is ultimately due to the underlying genotype, however. Few studies have attempted to map genes influencing intra-individual variability in sleep thus far. Linkage mapping in humans identified two candidate genes, JARID1a and CACNA1C, associated with interdaily stability [24]. Spada and colleagues attempted to map sleep onset irregularity, measured as the standard deviation of the timing of sleep onset, but the associations did not reach their genome-wide threshold of significance, though some nominal associations were found [32]. These difficulties occur in part because of the difficulties in estimating trait variability in an outbred population accurately. Sources of heritable genetic variance and environmental variance are confounded in outbred populations, making it challenging to distinguish between the two [33]. Model organism populations have the potential to address this problem. Model organisms can be inbred so that many simultaneous measurements can be made on a single genotype, and environmental conditions can be controlled to a greater extent. Previously, model organism populations have been used to study the genetic basis of inter-individual variability in morphological and fitness traits in Arabidopsis and maize [34–36]; gene expression in isogenic strains of yeast [37, 38] and inbred populations of flies [39]; and wing shape [40], bristle number [41], food consumption [42], locomotor handedness [43], chill coma recovery, startle response, starvation resistance [44], and sleep [45] in flies. Measured as the coefficient of environmental variation (CVE), the differences in sleep parameters among individual flies were found to have a heritable, mappable basis [45]. Thus, model organism populations offer the opportunity to investigate among-animal variability and its genetic underpinnings. We wanted to determine whether this strategy could also be applied to study intra-individual variability in sleep, and if so, whether this variability had an underlying genetic cause. We used previously collected data from a genome-wide association study of the Drosophila Genetic Reference Panel (DGRP) to investigate this possibility [45]. We examined daily fluctuations in the sleep patterns of 10239 flies. Daily fluctuations in sleep were more variable in some genotypes than in others, revealing that the variation is heritable. Expressing the variability in sleep traits as a standard deviation enabled us to conduct a genome-wide association study of intra-individual variability. We mapped 104 candidate genes to seven sleep parameters varying over days in flies. These genes included plausible candidates from previous studies in flies and mapped to 51 human homologs. Day-to-day fluctuations in sleep characteristics may therefore depend in part upon underlying genotype. Methods Assessment of Individual Differences in Day-to-Day Sleep We reanalyzed previously collected [45] data to determine whether sleep fluctuated across days in individual flies and whether that fluctuation had a heritable component. In that experiment, 168 lines of flies from the DGRP [46, 47] were randomly divided into four equal blocks with four replicates. Each replicate had eight flies per line per sex, for a total of 64 flies per line. Flies were assayed for rest and activity behavior under standard environmental conditions (cornmeal–molasses–agar medium, 25°C, 60%–75% relative humidity, and 12-hr light:dark cycle). Virgin males and females were assayed in order to avoid any effect of mating status on sleep [48], and 30 same-sex flies were maintained per vial for 4 days to standardize social exposure effects [49]. Sleep and activity recordings were made for each fly using the Drosophila Activity Monitoring System (Trikinetics, Waltham, MA). Seven days of recordings were made for each fly; the first day was discarded to mitigate any potential effect of CO2 anesthetization recovery on sleep and activity. The flies were visually inspected after the experiment and data from dead flies were eliminated from the data set. The raw sleep and activity data were used to calculate sleep parameters for each day using an in-house C# program. Sleep duration, numbers of sleep bouts, and average sleep bout length during the day and night were calculated. In addition, waking activity, the number of activity counts per minute spent awake, was determined. We assessed the change in sleep parameters across days using the four-way random ANOVA model, Y=μ+D+B+L(B)+R(B)+D×L(B)+D×R(B)+L×R(B)+D×L×R(B)+ε where D represents the random effect of day, B represents the random effect of block, L represents the random effect of line, R represents the random replicate effect, and ε represents the error variance. Analyses were performed separately for males and females. A significant D term indicates that the sleep phenotype is different across days, whereas a significant D × L(B) term indicates that the fluctuation in a sleep parameter across days has a heritable component. In addition, we evaluated males and females of each line separately for day effects using the reduced model, Y = µ + D + R + D×R + ε. We used SAS 9.3 (SAS Institute, Cary, NC) to evaluate these models. Quantitative Genetic Analysis We found a significant D × L(B) term for all the sleep traits, indicating that sleep was variable across days and that a portion of this variability was under genetic control (Supplementary Table S1). We used the within-fly standard deviation in sleep across days, σ, to represent the variability in each sleep trait per fly [9–13]. We noted that average day bout length σ and waking activity σ were not normally distributed traits. We normalized these two traits by taking the natural log of σ for each fly. We analyzed σ for each trait using the ANOVA model,  Y=μ+B+ L(B)+S+R(B)+S×L(B)+L×R(B)+S×R(B)+S×L×R(B)+ ε where B, L, and R are as defined above, and S is sex. We analyzed the data for each sex separately using the reduced model, Y=μ+B+ L(B)+R(B)+L×R(B)+ εThe variance components for each model were estimated using the restricted maximum likelihood (REML) method. The broad sense heritability for each trait was estimated as  H2=(σL2+σSL2)/(σL2+σSL2+σε2) for both sexes combined, where σL2 is the line variance component, σSL2 is the line by sex variance component, and σε2 is the sum of the other variance components [50]. The broad sense heritability was estimated as H2=(σL2)/(σL2+σε2) for each sex separately. We calculated the cross-sex genetic correlation as rmf=(σL2)/(σLM2+σLF2) where σL2 represents the line variance component for both sexes combined, σLM2 represents the line variance component for males, and σLF2 represents the line variance component for females [50]. In addition, we computed phenotypic correlations between sleep σ traits and other traits measured in the DGRP having line means available, including mean sleep and sleep CVE [45]; morphological traits (eye area and size, pigmentation level, wing centroid size, and eye interocular distance) [51–53]; lifespan and metabolic traits [54, 55]; behavior (reproductive or courtship traits, food consumption, chemosensation, startle response, geotaxis, and aggression) [42, 44, 56–61]; environmental responses (oxidative stress, methylmercury tolerance, radiation stress, chill coma, and starvation) [44, 61–64]; and genome size [65]. Phenotypic correlations were calculated as Pearson product moment correlation coefficients. We also estimated the genetic correlations rG between mean sleep and sleep σ traits as cov12/σL12×σL22 [50]. SAS 9.3 and JMP 13.0.0 (SAS Institute, Cary, NC) were used to evaluate these models. Genotype–Phenotype Associations Whole-genome sequence data are available for the DGRP, and a website tool (http://dgrp2.gnets.ncsu.edu/, accessed 21 December 2017) is available to calculate genome-wide associations among variants in this population and any trait of interest [46, 47]. Using the average σ per line for each sleep trait, we computed genome-wide associations with 1920276 polymorphisms segregating in the DGRP. Only polymorphic variants with minor allele frequencies of 5% or more were used in the associations. Some lines of the DGRP are known to have Wolbachia pipientis infection and large chromosomal inversions. We checked for associations between sleep σ parameters and Wolbachia and inversions. No significant associations with Wolbachia infection were present, but there were significant associations with chromosomal inversions for average day bout duration σ (associated with the In(2R)NS and In(3R)Mo inversions, p = .0003 and p < .0001, respectively), average night bout duration σ (associated with In(2L)t, p = .0144), night bout number σ (associated with In(2R)NS, p = .0051), and waking activity σ (associated with In(3R)Mo, p = .0008). We applied the linear mixed model, y=Xb+Zu+ε, where y are the adjusted phenotypes, X is the fixed effect of polymorphisms, and Z represents the covariance matrix accounting for population structure [46]. Associations with a p-value of 1 × 10–5 or less were called significant, a threshold applied in many DGRP studies [42, 52, 57, 66–70]. This threshold was supported by Q–Q plots (Supplementary Figure S1), which generally deviated little from expected values. In addition, false discovery rates (FDRs) were calculated for each polymorphic association using the method of Benjamini and Hochberg [71]. We determined the linkage disequilibrium (LD) (r2) among significant polymorphisms using PLINK [72]. Finally, human homologs of Drosophila genes were identified using the DIOPT website, which compiles homology data from multiple genomic databases [73]. Sleep Assays and Phenotypes For all sleep assays, flies were maintained in a constant temperature (25°C), constant humidity (60%) incubator and assayed on standard food (http://flystocks.bio.indiana.edu/Fly_Work/media-recipes/bloomfood.htm, accessed 21 December 2017) in standard light:dark cycle (12:12 hr) conditions. Male and female virgins were collected and maintained in single-sex vials of 20 flies for 4 days before sleep measurements to mitigate potential effects of mating status [74] and social exposure [49]. We loaded individual flies into Drosophila Activity Monitors (DAM2, Trikinetics, Waltham, MA) to measure sleep and activity parameters. Numbers of activity counts for each fly were recorded per minute; we defined sleep as 5 min without an activity count [75]. We discarded the first day’s data as the flies would be recovering from CO2 anesthesia and/or adjusting to the monitor tubes during the first day. We examined each fly after the experiment; the data from flies not surviving the experiment were discarded. We used an in-house C# program to calculate sleep parameters for each day: night, day, and 24-hr sleep duration in minutes; night and day sleep bout number; night and day average sleep bout length; sleep latency, the number of minutes before the fly’s first sleep bout after lights are turned off; and waking activity, the number of activity counts per minute spent awake. Using these parameters, we calculated the standard deviation of each sleep trait over the 6-day monitoring period for each fly. Sleep Deprivation Protocol The most variable DGRP lines were deprived of sleep using the following procedure. We placed individual flies into the DAM2 Trikinetics monitors using the protocol above. The flies acclimated to the monitors for a single day, and then we recorded 2 baseline days of rest and activity. We used a vortexer (Troemner, Thorofare, NJ) fitted with a custom mounting plate (Trikinetics, Waltham, MA) to apply a mechanical stimulus on the night of the third day. Each minute, the vortexer gently shook the flies for 5 s every minute of the 12-hr night sleep period (shaking speed = 2.0). The flies’ recovery was monitored for 2 additional days. An identical set of flies were loaded into Trikinetics monitors on a different shelf of the same incubator to serve as nonsleep-deprived controls. We used this protocol to deprive the following DGRP lines: DGRP_41, DGRP_73, DGRP_153, DGRP_335, DGRP_338, DGRP_409, DGRP_646, and DGRP_892. We sleep-deprived 16 flies of each sex per line. We used an in-house Python algorithm to determine the average minutes of sleep per 30 min. We evaluated differences in 24-hr sleep for each DGRP line separately using the ANOVA model, Y = µ + T + D + S + T×D + T×S + D×S + T×D×S + ε, where T is the control or sleep-deprived treatment, D is day, and S is sex. We analyzed the data for each population or line separately using the reduced ANOVA model, Y = µ + D + S + D×S + ε, where D and S are as defined above. We applied a post hoc Tukey analysis to identify significant loss and gain of sleep. Verification of Candidate Genes We tested 17 Minos element (Mi{ET1} and Mi{y[+mDint2] = MIC}) insertion lines in 14 candidate genes identified by the GWAS: 5-hydroxytryptamine (serotonin) receptor 1A (5-HT1A), CG42260, CG5888, CG7985, fruitless (fru), histone deacetylase 4 (HDAC4), rhomboid (rho), scab (scb), sallimus (sls), solwind (sowi), super sex combs (sxc), Transmembrane channel-like (Tmc), turtle (tutl), and ZnT35C (Supplementary Table S2). The inserts for sls and sowi were not viable as homozygotes and were tested as heterozygotes. Lines with Mi{ET1} insertions were created in an isogenic w1118 background (w1118[5905]); hence, we used this background as a control [76]. Lines with Mi{y[+mDint2] = MIC} insertions were crossed to a common background when created: y[1] w[*]; nub[2] b[1] sna[Sco] pr[1] cn[1]/CyO [3628] [77]. We used this background as a control for lines with the Mi{y[+mDint2]=MIC} insertion. All stocks were obtained from the Bloomington, IN stock center. Sleep phenotypes were measured as described above for 16 flies per sex per genotype in the mutants and their respective controls. Mutant sleep σ phenotypes were compared with the control using the ANOVA model, Y = µ + G + S + G×S + ε, where G is genotype, S is sex, and ε is the error term. Males and females were analyzed separately using the reduced ANOVA model, Y = µ + G + ε. Results Quantitative Genetics of Day-to-Day Variability in Sleep Daily fluctuations occurred in all sleep traits among individual flies. Figure 1 contrasts night sleep traits among individuals from the least variable DGRP line with those of the most variable line, illustrating how extreme the differences can be (for day sleep traits, see Supplementary Figure S2). Night sleep duration in DGRP_41, the least variable line, was high and changed little across time (Figure 1A). In contrast, sleep in DGRP_153, the most variable line, fluctuated through the full range of possible night sleep values (Figure 1A). Numbers of night bouts were relatively low in DGRP_531, but fluctuated greatly in DGRP_882 (Figure 1B). Stable and fluctuating patterns were also observed for average night bout length (Figure 1C). These daily fluctuations among individuals translated into changes in overall patterns of sleep for some DGRP lines. For example, male flies of DGRP_181 had increasing night nap length over the course of 6 days; simultaneously, the average number of night bouts decreased, whereas overall night sleep duration stayed constant (Figure 2A). Thus, these males had less fragmented sleep patterns over the 6-day observation period. Females of DGRP_761, on the other hand, had increased numbers of nightly naps and decreased average nap times, suggesting an increasingly fragmented pattern (Figure 2B). In addition, some flies, such as males of DGRP_73, had stable sleep characteristics over time (Figure 2C). We analyzed the entire population for significant changes across day. With the exception of day sleep duration in females, all sleep traits varied significantly across time (Supplementary Table S1). We observed a highly significant interaction of day and line for all sleep traits (Supplementary Table S1; all pLine×Day(Block) < .0001), suggesting that this variability was heritable. We also used a reduced model ANOVA to determine the numbers of DGRP lines with significant day-to-day fluctuations in sleep (Table 1). Waking activity was variable for nearly every line, with 87.5% and 91% of the lines having significant daily fluctuations in males and females, respectively. Average day bout length was the least variable trait, with only 26.2% of the lines varying across day. These results show that daily fluctuations in sleep parameters are common in flies, with some genotypes exhibiting stability while others are characteristically variable. Table 1. Numbers of Lines With Significant Daily Fluctuations in Sleep Parameters Sleep Trait  Males  Females  No. of variable lines  Percentage  No. of variable lines  Percentage  Day sleep (min)  98  58.3  84  50.0  Night sleep (min)  89  53.0  85  50.6  Day bout number  69  41.1  67  39.9  Night bout number  69  41.1  59  35.1  Avg. day bout length (min)  49  29.2  41  24.4  Avg. night bout length (min)  54  32.1  50  29.8  Waking activity (cts/min)  147  87.5  153  91.1  Sleep Trait  Males  Females  No. of variable lines  Percentage  No. of variable lines  Percentage  Day sleep (min)  98  58.3  84  50.0  Night sleep (min)  89  53.0  85  50.6  Day bout number  69  41.1  67  39.9  Night bout number  69  41.1  59  35.1  Avg. day bout length (min)  49  29.2  41  24.4  Avg. night bout length (min)  54  32.1  50  29.8  Waking activity (cts/min)  147  87.5  153  91.1  The table lists the number and percentage of DGRP lines with significant daily variability in the sleep parameter indicated (pDay < .05). The total number of DGRP lines with sleep phenotypes is 168. cts = counts. View Large Figure 1. View largeDownload slide Example of variability in night sleep among individual flies. The plots contrast sleep changes over days for the least variable DGRP line with the most variable DGRP line for that trait. (A) Night sleep duration. The least variable line is DGRP_41 (n = 64); the most variable line is DGRP_153 (n = 61). (B) Night bout number. The least variable line is DGRP_531 (n = 62); the most variable line is DGRP_882 (n = 54). (C) Average night bout length. The least variable line is DGRP_42 (n = 64); the most variable line is DGRP_338 (n = 64). Figure 1. View largeDownload slide Example of variability in night sleep among individual flies. The plots contrast sleep changes over days for the least variable DGRP line with the most variable DGRP line for that trait. (A) Night sleep duration. The least variable line is DGRP_41 (n = 64); the most variable line is DGRP_153 (n = 61). (B) Night bout number. The least variable line is DGRP_531 (n = 62); the most variable line is DGRP_882 (n = 54). (C) Average night bout length. The least variable line is DGRP_42 (n = 64); the most variable line is DGRP_338 (n = 64). Figure 2. View largeDownload slide Examples of changes in night sleep parameters over days. Box plots of night sleep duration, night bout number, and average night bout length are plotted for each day. (A) Males of DGRP_181 (n = 32). (B) Females of DGRP_761 (n = 32). (C) Males of DGRP_73 (n = 32). Figure 2. View largeDownload slide Examples of changes in night sleep parameters over days. Box plots of night sleep duration, night bout number, and average night bout length are plotted for each day. (A) Males of DGRP_181 (n = 32). (B) Females of DGRP_761 (n = 32). (C) Males of DGRP_73 (n = 32). Lines with characteristically variable sleep could also potentially have altered responses to disruption of the sleep homeostat. We examined the response of several DGRP lines to a single night of mechanical sleep deprivation: DGRP_338, DGRP_892, DGRP_41, and DGRP_646, which had the most variable average night bout length; DGRP_335, DGRP_409, DGRP_73 (and DGRP_646), which had the most variable waking activity; and DGRP_153, which had the most variable night sleep. Flies from all eight lines were significantly deprived of sleep compared with their baseline sleep times (Figure 3). However, the sleep loss in flies of line DGRP_335 was quite low; only 104.7 min of sleep were lost on average (Figure 3). All of the lines had a statistically significant increase in their 24-hr sleep on the day after sleep deprivation compared with their baseline sleep, with the exception of DGRP_335 and DGRP_73. DGRP_335 did not lose much sleep, so one possibility is that the homeostat was not perturbed sufficiently in these flies to elicit a response. However, flies from line DGRP_73 lost 540.5 min of sleep during the night, and only gained 99.8 min during the next 24-hr period. Thus, although the recovery sleep in most of the lines was similar to that of other wild-derived and laboratory strains [78, 79], the rebound sleep of DGRP_335 and DGRP_73 suggests that genotypes with variable sleep may also have an altered homeostatic response. Figure 3. View largeDownload slide Sleep loss and recovery in DGRP lines with the greatest daily fluctuations in sleep. *p < .05. For DGRP_41, n = 32; DGRP_73, n = 29; DGRP_153, n = 31; DGRP_335, n = 30; DGRP_338, n = 30; DGRP_409, n = 28; DGRP_646, n = 30; and DGRP_892, n = 32. Figure 3. View largeDownload slide Sleep loss and recovery in DGRP lines with the greatest daily fluctuations in sleep. *p < .05. For DGRP_41, n = 32; DGRP_73, n = 29; DGRP_153, n = 31; DGRP_335, n = 30; DGRP_338, n = 30; DGRP_409, n = 28; DGRP_646, n = 30; and DGRP_892, n = 32. We modeled daily fluctuations in sleep traits as the standard deviation (σ) per fly (Supplementary Table S3) in a manner analogous to studies of intra-individual variability in human sleep parameters [9–13]. We observed highly significant differences in σ among lines (all pLine(Block) < .0001), and the variability was sexually dimorphic as well (all pLine×Sex(Block) < .0001) (Supplementary Table S4). Histograms of σ for night sleep traits (Figure 4) and day sleep traits (Figure 5) illustrate this variability among lines of the DGRP. The significant differences among lines suggested that sleep σ was heritable; thus, we calculated broad-sense heritabilities to quantify the genetic contribution to each trait. Heritabilities for day and night sleep duration σ were 0.49 and 0.26, respectively; 0.07 and 0.13 for day and night bout number σ, respectively; 0.30 for both day and night average bout length σ; and 0.21 for waking activity σ. With the exception of the low heritability for day bout number σ, heritabilities were moderate, suggesting that daily fluctuations in sleep are partially driven by genotype. Figure 4. View largeDownload slide Histograms of night sleep σ traits. The histogram shows male line means in blue, and female line means in pink. (A) Night sleep duration σ. (B) Night bout number σ. (C) Avg. night bout length σ. Figure 4. View largeDownload slide Histograms of night sleep σ traits. The histogram shows male line means in blue, and female line means in pink. (A) Night sleep duration σ. (B) Night bout number σ. (C) Avg. night bout length σ. Figure 5. View largeDownload slide Histograms of day sleep σ traits. The histogram shows male line means in blue, and female line means in pink. (A) Day sleep duration σ. (B) Day bout number σ. (C) Normalized avg. day bout length σ. (D) Normalized waking activity σ. Figure 5. View largeDownload slide Histograms of day sleep σ traits. The histogram shows male line means in blue, and female line means in pink. (A) Day sleep duration σ. (B) Day bout number σ. (C) Normalized avg. day bout length σ. (D) Normalized waking activity σ. We calculated the phenotypic and genetic correlations among sleep σ phenotypes (Supplementary Table S5). Many of these correlations were significantly different from zero; however, the correlations were not necessarily high (i.e., the correlations were not close to 1). For example, night sleep duration σ had significant genetic correlations with all other sleep σ traits, but genetic correlations with other night-related traits were higher (0.573 and −0.459 with night bout number σ and average night bout length σ, respectively) than day and activity traits (0.256, −0.222, and −0.152 with day bout number σ, average day bout length σ, and waking activity σ, respectively). These correlations suggest a shared genetic architecture underlying daily fluctuations in sleep, but the overlap is far from complete. We also calculated the phenotypic and genetic correlations with the previously published mean and CVE sleep phenotypes for these lines. Sleep σ traits tended to be correlated with their respective mean and CVE, and, in general, these correlations were moderate to high (Supplementary Table S6). Night sleep σ and mean night sleep had a negative genetic correlation of rG = −0.80, whereas day sleep σ and mean day sleep had an rG of −0.44; that is, lower mean sleep duration was more variable. The trend for night sleep σ may also be true of single-gene mutations; we examined data from a previous p-element mutagenesis screen [80] and estimated the mutational genetic correlation as rG = −1.07 (p < .0001) between night sleep σ and night sleep. Day sleep σ and day sleep had a positive mutational genetic correlation of rG = 0.61 (p < .0001), however, suggesting the opposite trend. Day bout number σ and waking activity σ were the least correlated among sleep traits, whereas average day and night bout length σ was highly correlated with their mean. Thus, intra-individual variability in sleep has, in part, some distinctive genetic features. We wondered if sleep σ traits might be correlated with other types of traits in flies. The DGRP has been used extensively to study many different complex traits; thus, it was possible to assess the phenotypic correlations with these traits and sleep σ. We calculated the phenotypic correlations between sleep σ traits and morphological traits; lifespan and metabolic traits; behavior; responses to environmental stresses; and genome size [42, 44, 51–65]. Many of the correlations with other traits were significantly different from zero, but their magnitudes were quite low (Supplementary Table S7). Only six correlations were higher than 0.25. Night sleep duration σ was negatively correlated with mean lifespan (r = −0.27) [55] and inter-individual variability in starvation resistance (r = −0.26) [44]. Average night bout length σ was positively correlated with variability in intraocular distance (r = 0.29), a morphological measure related to reproductive fitness in flies [53]. Night bout number σ was correlated with eclosion rate (r = 0.30) [62]. Waking activity σ was negatively correlated with startle response on control medium (r = −0.28) and menadione bisulfite-supplemented medium (r = −0.30), an inducer of oxidative stress [61]. Thus, correlations between sleep σ traits and other types of traits were generally quite low, suggesting a distinct genetic basis for sleep σ. Genotype–Phenotype Associations Using the average σ of each sleep trait per line, we associated each sleep σ phenotype with all polymorphisms known to be segregating in the DGRP which had a minor allele frequency of 0.05 or more: 1920276 variants in total [46]. Associations were evaluated for sexes combined, each sex separately, and the difference between sexes (male avg. − female avg.) (see Supplementary Table S8 for details). Table 2 shows the numbers of polymorphisms that were associated with each sleep σ trait at a threshold p-value of 1 × 10–5 or less. Numbers of significant associations were modest, with 14%–26% of the associated polymorphisms located within the coding region of a gene, and the remaining polymorphisms mapping between genes (Supplementary Figure S3). A preponderance of significant associations had low minor allele frequencies as the median allele frequency was 0.12, a typical result in studies using the DGRP [45, 47, 61, 62, 64, 68]. Thus, variants with larger effect sizes tended to be relatively rare. However, significant associations were also observed at more common alleles as roughly 10% of the associations had minor allele frequencies above 0.40. Only 20 of the 663 associations had FDRs of 0.05 or less, and only 34 had FDRs of 0.10 or less. Despite the modest statistical support, effect sizes for some sleep σ traits were very high. We discuss these results in detail below. Table 2. Numbers of Polymorphisms Significantly Associated With Sleep σ Traits Sleep trait  Number of polymorphisms  Percent within genes  Day sleep σ (min)  145  25.5  Night sleep σ (min)  103  16.5  Day bout number σ  43  14.0  Night bout number σ  76  19.7  Day avg. bout length σ [ln(min)]  86  26.7  Night avg. bout length σ (min)  124  13.7  Waking activity σ [ln(cts/min)]  87  18.4  Sleep trait  Number of polymorphisms  Percent within genes  Day sleep σ (min)  145  25.5  Night sleep σ (min)  103  16.5  Day bout number σ  43  14.0  Night bout number σ  76  19.7  Day avg. bout length σ [ln(min)]  86  26.7  Night avg. bout length σ (min)  124  13.7  Waking activity σ [ln(cts/min)]  87  18.4  View Large Effect sizes for associations with average night bout length σ were very high in magnitude and ranged from −39.2 to 17.7 min. Unlike mammalian genomes, LD decays very rapidly in the DGRP, on the order of 10–30 bp on average [46, 47]. We therefore assumed that a significant variant located within a gene implicated that gene, unless (1) LD was high (r2 ≥ 0.8) among variants, or (2) the trait was associated with a known chromosomal inversion (Methods). LD was generally very low among polymorphisms for average night bout length σ (Figure 6A) except for two regions: one region was on chromosome 2L between 5056207 and 5594455 bp; the other was on chromosome 2R between 5191057 and 5472092 bp. The first region of LD was also within the In(2L)t inversion, which was itself significantly associated with average night bout length σ (Methods). Variants within candidate genes in these two regions, CG31918 and sxc, were therefore not distinguishable from the other polymorphisms in LD around them (8 and 21 variants, respectively). Most of the remaining 93 polymorphisms associated with average night bout length σ were intergenic, but 15 additional polymorphisms were in 12 candidate genes. These candidate genes included 5-HT1A, flower, fru, and HDAC4 (Supplementary Table S8). Figure 6. View largeDownload slide Genome-wide association results for sleep σ traits. Significant polymorphisms (p < 1 × 10−5) are plotted for each trait. The points are color-coded to indicate the most significant association (sexes combined, male, female, or sex difference) as indicated in the legend. The top panel shows the minor allele frequency (MAF) for each polymorphism plotted across chromosomal position. The middle panel plots the effect size a normalized by the genetic standard deviation σG. The lower panel shows the p-values for each polymorphism plotted as the -log of the p-value. The bottom triangle is the LD among polymorphisms calculated as r2. Black lines denote chromosome boundaries. (A) Avg. night bout length σ. (B) Waking activity σ. (C) Night sleep duration σ. (D) Day sleep duration σ. Figure 6. View largeDownload slide Genome-wide association results for sleep σ traits. Significant polymorphisms (p < 1 × 10−5) are plotted for each trait. The points are color-coded to indicate the most significant association (sexes combined, male, female, or sex difference) as indicated in the legend. The top panel shows the minor allele frequency (MAF) for each polymorphism plotted across chromosomal position. The middle panel plots the effect size a normalized by the genetic standard deviation σG. The lower panel shows the p-values for each polymorphism plotted as the -log of the p-value. The bottom triangle is the LD among polymorphisms calculated as r2. Black lines denote chromosome boundaries. (A) Avg. night bout length σ. (B) Waking activity σ. (C) Night sleep duration σ. (D) Day sleep duration σ. We also observed large effect sizes for associations with waking activity σ; these ranged from 0.76 to 1.25 counts per minute. For waking activity σ, 87 significant polymorphisms were found, and virtually no LD among these polymorphisms was observed. As with average night bout length σ, most of these variants were intergenic, but 16 polymorphisms were in 13 genes (Figure 6B; Supplementary Table S8), including rho, roughoid (ru), and Heat shock protein 26. Seventeen polymorphisms had very low FDRs (<0.0001), including one in roughoid and two in Heat shock protein 26. Variants associated with night and day sleep duration σ had smaller effect sizes. Night sleep σ effects ranged from −17.6 to 12.1 min and day sleep σ effects ranged from −14.3 to 7.5 min. 103 polymorphisms were significantly associated with night sleep σ. A large region on Chromosome 3L was in LD (23531038 to 24006114 bp) among 10 intergenic polymorphisms significantly associated with night sleep σ (Figure 6C). Eighteen variants implicated 14 candidate genes, including sls, Secretory 8, and unpaired 1 (Supplementary Table S8). One hundred forty-five polymorphisms were associated with day sleep σ. There were 29 genes identified for day sleep σ, including skittles, minibrain, and kirre (Supplementary Table S8). Little LD was evident for day sleep σ (Figure 6D). One possibility is that the genetic architecture for sleep σ overlaps with mean/CVE sleep as the genetic correlations among these traits were generally significant. We therefore compared our results with the previous genome-wide association study of sleep in Drosophila [45]. We found 22 unique polymorphisms that overlapped with sleep σ traits, which was 3.3% of the total identified in this study (Supplementary Table S9). Most of the overlap was of intergenic variants, but polymorphisms in the genes CG13699, CR45711, flower, and rho overlapped among traits. Many polymorphisms identified in this study are therefore unique to sleep σ traits. Finally, we searched for human homologs of candidate sleep σ genes using the DIOPT tool. DIOPT searches for homologous genes across 12 databases; DIOPT scores indicate the number of databases that predicted a given homologous gene [73]. A total of 438 human genes had predicted homology with 78 Drosophila genes, though in many cases the homology was not high. We found 51 human genes that were predicted to have the highest homology with 42 Drosophila genes (Supplementary Table S10). Twenty-seven Drosophila genes had no known human homolog. Thus, a portion of the genes we identified for sleep σ traits have plausible human homologs. Verification of Candidate Genes We tested 17 Minos element constructs in 14 candidate genes for their effects on sleep σ traits (Supplementary Table S2). We chose sleep σ traits with the largest effect sizes to confirm: night average bout length σ, night sleep σ, and waking activity σ (Supplementary Table S8). We tested all genes with available Minos constructs that were identified for these three traits. We found that 12 of the candidate genes confirmed the GWAS prediction (Table 3). 5-HT1A, CG42260, and fru had two Minos insertions available. Both 5-HT1A insertions had significant differences in night average bout length σ from the control; likewise, both fru insertions were significantly different for night average bout length σ. Only one of the CG42260 insertions had significant differences in night sleep σ from the control line, however. All the remaining genes had one Minos construct available for test, and the following genes were confirmed: CG7985, HDAC4, and sxc for night average bout length σ; heterozygous sls for night sleep σ; and tutl, Tmc, rho, heterozygous sowi, and scb for waking activity σ. In addition, some of these constructs had pleiotropic effects on other sleep traits (Supplementary Table S11), consistent with previous observations of gene-level pleiotropy in Drosophila genome-wide studies [45, 81]. These mutant tests validate these genes as candidates affecting daily fluctuations in sleep. Table 3. Verification of Candidate Genes Gene  Allele tested  Sleep trait  Diff. from control  No.  5HT1A  MB09812  Night avg. bout length σ (min)  14.32***  32  5HT1A  MB09978  Night avg. bout length σ (min)  10.61**  31  CG7985  MI00275  Night avg. bout length σ (min)  −113.93***  31  fru  MB02472  Night avg. bout length σ (min)  28.29*  32  fru  MB01996  Night avg. bout length σ (min)  7.09*  30  HDAC4  MI05513  Night avg. bout length σ (min)  −103.05***  32  sxc  MI02286  Night avg. bout length σ (min)  −125.68****  32  CG42260  MB00428  Night sleep σ (min)  −27.87****  31  sls  MI011177 Het.  Night sleep σ (min)  30.78****  32  rho  MI08786  Waking activity σ (cts/min)  −0.85****  32  scb  MI010788  Waking activity σ (cts/min)  −0.82****  32  sowi  MI03342 Het.  Waking activity σ (cts/min)  −0.97****  32  Tmc  MI02041  Waking activity σ (cts/min)  −0.78****  32  tutl  MI08144  Waking activity σ (cts/min)  −0.90****  31  Gene  Allele tested  Sleep trait  Diff. from control  No.  5HT1A  MB09812  Night avg. bout length σ (min)  14.32***  32  5HT1A  MB09978  Night avg. bout length σ (min)  10.61**  31  CG7985  MI00275  Night avg. bout length σ (min)  −113.93***  31  fru  MB02472  Night avg. bout length σ (min)  28.29*  32  fru  MB01996  Night avg. bout length σ (min)  7.09*  30  HDAC4  MI05513  Night avg. bout length σ (min)  −103.05***  32  sxc  MI02286  Night avg. bout length σ (min)  −125.68****  32  CG42260  MB00428  Night sleep σ (min)  −27.87****  31  sls  MI011177 Het.  Night sleep σ (min)  30.78****  32  rho  MI08786  Waking activity σ (cts/min)  −0.85****  32  scb  MI010788  Waking activity σ (cts/min)  −0.82****  32  sowi  MI03342 Het.  Waking activity σ (cts/min)  −0.97****  32  Tmc  MI02041  Waking activity σ (cts/min)  −0.78****  32  tutl  MI08144  Waking activity σ (cts/min)  −0.90****  31  Mutations were evaluated for their effect on the sleep σ traits predicted by the GWAS. Mean differences from the respective control line are given. The number of w1118[5905] control flies was 32; the number of y[1] w[*]; nub[2] b[1] sna[Sco] pr[1] cn[1]/CyO [3628] control flies was 27. Waking activity σ differences reflect un-transformed values; statistical tests were ln-transformed. Asterisks reflect FDRs to correct for multiple tests. *0.01 < FDR ≤ 0.05; **0.001 < FDR ≤ 0.01; ***0.0001 < FDR ≤ 0.001; ****FDR < 0.0001. FDRs = false discovery rates; Het. = heterozygote. View Large Discussion Here, we demonstrated that individual flies experience daily fluctuations in sleep characteristics. Some flies altered their sleep patterns dramatically from day to day, whereas others were remarkably stable. This intra-individual variability, which we quantified as the standard deviation (σ) of each sleep trait over time, differed among genotypes and was heritable. Importantly, sleep σ traits were partially but not completely genetically correlated with summary sleep traits (mean sleep and sleep CVE). Likewise, sleep σ traits were relatively uncorrelated with other complex traits studied in the DGRP. Daily fluctuations in sleep therefore have a distinctive genetic basis, with a partial overlap with summary sleep traits. Fourteen polymorphisms overlapped between sleep σ traits and their respective summary traits, demonstrating this partial overlap. Although further experimentation is required to understand the mechanistic basis of the predicted associations here, some of the candidate genes were previously reported to affect summary sleep phenotypes. 5-HT1A was identified as a gene regulating sleep duration and bout length in flies via signaling in the mushroom bodies [82]; here, it was implicated in average night bout length σ, and tests of flies with two different Minos insertions in 5-HT1A verified these effects. VAChT was associated with day bout number σ; mutations in this gene were previously reported to affect mean day bout number [68]. Two other genes known to affect summary sleep phenotypes, rho [83] and kirre [68], were also associated with waking activity σ and day sleep σ, respectively. We confirmed the effects of rho on waking activity σ. The associations with sleep σ traits differ somewhat from the sleep summary traits previously reported, indicating that these genes may be pleiotropic. The activity of these genes may therefore vary from day to day to produce the summary effects observed previously. Other genes we identified are plausible candidates for daily fluctuations in sleep. Interestingly, several of these genes have known effects on circadian rhythms. super sex combs was associated with average night bout length σ; this gene interacts with the GSK3β gene sgg to affect circadian rhythms [84]. We confirmed the effect of sxc on average night bout length σ. HDAC4 was also identified in the GWAS and verified for average night bout length σ. It is an intriguing candidate as it is involved in epigenetic modification and transcription [85], which might be anticipated characteristics of a gene affecting daily fluctuations in sleep. HDAC4 also functions in long-term memory [86], and mutations of this gene affect circadian rhythmicity in flies [87]. RNAi-mediated knockdown of unpaired1, which was associated with night sleep σ in this study, results in arrhythmicity [88]. minibrain was identified for day sleep σ; double-mutant flies minibrain/small optic lobes increased instability in rest/activity rhythms [89]. In addition, roughoid, like rhomboid, is a serine protease functioning in the EGFR pathway [90]; this gene was associated with waking activity σ. About 80% of the polymorphic variants that we identified were in intergenic regions. This observation is consistent with other GWA studies [91], but reveals the difficulty in fully understanding the genetic basis of complex traits such as sleep. Intergenic variants might alter gene function via changes in enhancer or DNA binding sites, which in turn may alter the transcription or translation of a gene or genes. Enhancer or promoter effects can occur far upstream or downstream of the gene they affect, however. As a first step, these variants could be studied as potential cis-regulatory genes by using genome editing to perturb each position and by examining the impact of these perturbations on transcript abundance in neighboring genes. Polymorphisms within the coding region of candidate genes are more tractable, as they can be tested directly with allelic replacement via genome editing. Overall, we identified 663 polymorphisms and 104 candidate genes for intra-individual variability in sleep. Our study has several limitations. First, the statistical support for polymorphisms associated with intra-individual variability in sleep was relatively low. This is partly due to the relatively small size of the DGRP and relatively low heritabilities, both of which make the detection of significant associations more difficult. Thus, the associations reported here probably represent a subset of the true associations, and for some of the traits such as day and night bout number σ, FDRs were high, decreasing the confidence in those predictions. However, quantile–quantile plots supported the 1 × 10−5 threshold p-value for night sleep σ, day sleep σ, average night bout length σ, and waking activity σ, indicating that there was adequate statistical power for these measures (Supplementary Figure S1). This makes sense as the effect sizes for these four traits were the largest we observed. Accordingly, our follow-up mutational tests focused on the genes with the largest predicted effect sizes. This was an effective strategy for confirming candidate genes as 12 of 14 genes had significant effects on the predicted sleep σ trait. Second, the variability in sleep patterns that we observed is confounded with age. The original study evaluated sleep in flies that were 4 to 14 days old, so it is not known how sleep patterns in these lines might change in either very young or very old flies. We observed both increasing and decreasing patterns of fragmentation in these relatively young flies. A previous longitudinal study of sleep in flies documented a monotonic trend for increased fragmentation across the entire lifespan, with some fluctuation in sleep patterns over short periods [92]. Varying trends have been observed in human sleep as well over time, from inverted U-shaped distributions [93] to declines in intra-individual variability with age [13]. It would therefore be interesting to examine daily fluctuations in sleep across the entire lifespan in this population. Finally, our study focuses on behavioral correlates for sleep and does not address any potential fluctuation in sleep intensity, which vary with genotype in flies [94] and are critical to understanding sleep need [95]. Conclusion We have demonstrated that Drosophila sleep patterns fluctuate from day to day. We defined the standard deviation as a measure of the daily fluctuations in sleep for each fly and found it to be heritable. Using all available variants in the DGRP, we mapped 104 candidate genes, some of which have known roles in sleep and circadian rhythms. We also confirmed the role of 12 genes on sleep intra-individual variability. Many of these genes have predicted homology with human genes, suggesting that the underlying genetic architecture of daily fluctuations in sleep may be conserved. Supplementary Material Supplementary material is available at SLEEP online. Acknowledgments The authors thank Y. Lin for helpful discussions. Funding This research was supported by the Intramural Research Program at the National Institutes of Health, the National Heart, Lung, and Blood Institute. Disclosure Statement None declared. References 1. Samson DR, Crittenden AN, Mabulla IA, Mabulla AZ, Nunn CL. Hadza sleep biology: evidence for flexible sleep-wake patterns in hunter-gatherers. Am J Phys Anthropol . 2017; 162( 3): 573– 582. Google Scholar CrossRef Search ADS PubMed  2. Yetish G, Kaplan H, Gurven Met al.   Natural sleep and its seasonal variations in three pre-industrial societies. Curr Biol . 2015; 25( 21): 2862– 2868. Google Scholar CrossRef Search ADS PubMed  3. Samson DR, Manus MB, Krystal AD, Fakir E, Yu JJ, Nunn CL. Segmented sleep in a nonelectric, small-scale agricultural society in Madagascar. Am J Hum Biol . 2017; 29( 4): e22979. Google Scholar CrossRef Search ADS   4. Ekirch AR. Segmented sleep in preindustrial societies. Sleep . 2016; 39( 3): 715– 716. Google Scholar CrossRef Search ADS PubMed  5. Van Dongen HP, Baynard MD, Maislin G, Dinges DF. Systematic interindividual differences in neurobehavioral impairment from sleep loss: evidence of trait-like differential vulnerability. Sleep . 2004; 27( 3): 423– 433. Google Scholar PubMed  6. Bliese PD, Wesensten NJ, Balkin TJ. Age and individual variability in performance during sleep restriction. J Sleep Res . 2006; 15( 4): 376– 385. Google Scholar CrossRef Search ADS PubMed  7. Lim J, Choo WC, Chee MW. Reproducibility of changes in behaviour and fMRI activation associated with sleep deprivation in a working memory task. Sleep . 2007; 30( 1): 61– 70. Google Scholar CrossRef Search ADS PubMed  8. Geiger A, Huber R, Kurth S, Ringli M, Jenni OG, Achermann P. The sleep EEG as a marker of intellectual ability in school age children. Sleep . 2011; 34( 2): 181– 189. Google Scholar CrossRef Search ADS PubMed  9. Knutson KL, Rathouz PJ, Yan LL, Liu K, Lauderdale DS. Intra-individual daily and yearly variability in actigraphically recorded sleep measures: the CARDIA study. Sleep . 2007; 30( 6): 793– 796. Google Scholar CrossRef Search ADS PubMed  10. Mezick EJ, Matthews KA, Hall Met al.   Intra-individual variability in sleep duration and fragmentation: associations with stress. Psychoneuroendocrinology . 2009; 34( 9): 1346– 1354. Google Scholar CrossRef Search ADS PubMed  11. Buman MP, Hekler EB, Bliwise DL, King AC. Exercise effects on night-to-night fluctuations in self-rated sleep among older adults with sleep complaints. J Sleep Res . 2011; 20( 1 Pt 1): 28– 37. Google Scholar CrossRef Search ADS PubMed  12. Angulo-Barroso RM, Peirano P, Algarin C, Kaciroti N, Lozoff B. Motor activity and intra-individual variability according to sleep-wake states in preschool-aged children with iron-deficiency anemia in infancy. Early Hum Dev . 2013; 89( 12): 1025– 1031. Google Scholar CrossRef Search ADS PubMed  13. Dillon HR, Lichstein KL, Dautovich ND, Taylor DJ, Riedel BW, Bush AJ. Variability in self-reported normal sleep across the adult age span. J Gerontol B Psychol Sci Soc Sci . 2015; 70( 1): 46– 56. Google Scholar CrossRef Search ADS PubMed  14. Gruber R, Sadeh A, Raviv A. Instability of sleep patterns in children with attention-deficit/hyperactivity disorder. J Am Acad Child Adolesc Psychiatry . 2000; 39( 4): 495– 501. Google Scholar CrossRef Search ADS PubMed  15. Shoji KD, Tighe CA, Dautovich ND, McCrae CS. Beyond mean values: Quantifying intraindividual variability in pre-sleep arousal and sleep in younger and older community-dwelling adults. Sleep Sci . 2015; 8( 1): 24– 30. Google Scholar CrossRef Search ADS PubMed  16. Bei B, Allen NB, Nicholas CL, Dudgeon P, Murray G, Trinder J. Actigraphy-assessed sleep during school and vacation periods: a naturalistic study of restricted and extended sleep opportunities in adolescents. J Sleep Res . 2014; 23( 1): 107– 117. Google Scholar CrossRef Search ADS PubMed  17. Könen T, Dirk J, Schmiedek F. Cognitive benefits of last night’s sleep: daily variations in children’s sleep behavior are related to working memory fluctuations. J Child Psychol Psychiatry . 2015; 56( 2): 171– 182. Google Scholar CrossRef Search ADS PubMed  18. Buysse DJ, Cheng Y, Germain Aet al.   Night-to-night sleep variability in older adults with and without chronic insomnia. Sleep Med . 2010; 11( 1): 56– 64. Google Scholar CrossRef Search ADS PubMed  19. Cheek RE, Shaver JL, Lentz MJ. Lifestyle practices and nocturnal sleep in midlife women with and without insomnia. Biol Res Nurs . 2004; 6( 1): 46– 58. Google Scholar CrossRef Search ADS PubMed  20. Straus LD, Drummond SP, Nappi CM, Jenkins MM, Norman SB. Sleep variability in military-related PTSD: a comparison to primary insomnia and healthy controls. J Trauma Stress . 2015; 28( 1): 8– 16. Google Scholar CrossRef Search ADS PubMed  21. Prasad B, Usmani S, Steffen ADet al.   Short-term variability in apnea-hypopnea index during extended home portable monitoring. J Clin Sleep Med . 2016; 12( 6): 855– 863. Google Scholar CrossRef Search ADS PubMed  22. Sforza E, Haba-Rubio J. Night-to-night variability in periodic leg movements in patients with restless legs syndrome. Sleep Med . 2005; 6( 3): 259– 267. Google Scholar CrossRef Search ADS PubMed  23. Bei B, Wiley JF, Trinder J, Manber R. Beyond the mean: A systematic review on the correlates of daily intraindividual variability of sleep/wake patterns. Sleep Med Rev . 2016; 28: 108– 124. Google Scholar CrossRef Search ADS PubMed  24. Pagani L, St Clair PA, Teshiba TMet al.   Genetic contributions to circadian activity rhythm and sleep pattern phenotypes in pedigrees segregating for severe bipolar disorder. Proc Natl Acad Sci U S A . 2016; 113( 6): E754– E761. Google Scholar CrossRef Search ADS PubMed  25. Scott J, Murray G, Henry Cet al.   Activation in bipolar disorders: a systematic review. JAMA Psychiatry . 2017; 74( 2): 189– 196. Google Scholar CrossRef Search ADS PubMed  26. Bei B, Seeman TE, Carroll JE, Wiley JF. Sleep and physiological dysregulation: a closer look at sleep intraindividual variability. Sleep . 2017; 40( 9): zsx109. doi: 10.1093/sleep/zsx109. Google Scholar CrossRef Search ADS   27. Okun ML, Reynolds CF3rd, Buysse DJet al.   Sleep variability, health-related practices, and inflammatory markers in a community dwelling sample of older adults. Psychosom Med . 2011; 73( 2): 142– 150. Google Scholar CrossRef Search ADS PubMed  28. He F, Bixler EO, Liao Jet al.   Habitual sleep variability, mediated by nutrition intake, is associated with abdominal obesity in adolescents. Sleep Med . 2015; 16( 12): 1489– 1494. Google Scholar CrossRef Search ADS PubMed  29. Gander P, van den Berg M, Signal L. Sleep and sleepiness of fishermen on rotating schedules. Chronobiol Int . 2008; 25( 2): 389– 398. Google Scholar CrossRef Search ADS PubMed  30. Stalder T, Evans P, Hucklebridge F, Clow A. State associations with the cortisol awakening response in healthy females. Psychoneuroendocrinology . 2010; 35( 8): 1245– 1252. Google Scholar CrossRef Search ADS PubMed  31. Callahan HS, Pigliucci M, Schlichting CD. Developmental phenotypic plasticity: where ecology and evolution meet molecular biology. Bioessays . 1997; 19( 6): 519– 525. Google Scholar CrossRef Search ADS PubMed  32. Spada J, Scholz M, Kirsten Het al.   Genome-wide association analysis of actigraphic sleep phenotypes in the LIFE Adult Study. J Sleep Res . 2016; 25( 6): 690– 701. Google Scholar CrossRef Search ADS PubMed  33. Hill WG, Mulder HA. Genetic analysis of environmental variation. Genet Res (Camb) . 2010; 92( 5-6): 381– 395. Google Scholar CrossRef Search ADS PubMed  34. Hall MC, Dworkin I, Ungerer MC, Purugganan M. Genetics of microenvironmental canalization in Arabidopsis thaliana. Proc Natl Acad Sci U S A . 2007; 104( 34): 13717– 13722. Google Scholar CrossRef Search ADS PubMed  35. Sangster TA, Salathia N, Undurraga Set al.   HSP90 affects the expression of genetic variation and developmental stability in quantitative traits. Proc Natl Acad Sci U S A . 2008; 105( 8): 2963– 2968. Google Scholar CrossRef Search ADS PubMed  36. Ordas B, Malvar RA, Hill WG. Genetic variation and quantitative trait loci associated with developmental stability and the environmental correlation between traits in maize. Genet Res (Camb) . 2008; 90( 5): 385– 395. Google Scholar CrossRef Search ADS PubMed  37. Ansel J, Bottin H, Rodriguez-Beltran Cet al.   Cell-to-cell stochastic variation in gene expression is a complex genetic trait. PLoS Genet . 2008; 4( 4): e1000049. Google Scholar CrossRef Search ADS PubMed  38. Blake WJ, Balázsi G, Kohanski MAet al.   Phenotypic consequences of promoter-mediated transcriptional noise. Mol Cell . 2006; 24( 6): 853– 865. Google Scholar CrossRef Search ADS PubMed  39. Lin Y, Chen ZX, Oliver B, Harbison ST. Microenvironmental gene expression plasticity among individual Drosophila melanogaster. G3 (Bethesda) . 2016; 6( 12): 4197– 4210. Google Scholar PubMed  40. Whitlock MC, Fowler K. The changes in genetic and environmental variance with inbreeding in Drosophila melanogaster. Genetics . 1999; 152( 1): 345– 353. Google Scholar PubMed  41. Mackay TF, Lyman RF. Drosophila bristles and the nature of quantitative genetic variation. Philos Trans R Soc Lond B Biol Sci . 2005; 360( 1459): 1513– 1527. Google Scholar CrossRef Search ADS PubMed  42. Garlapow ME, Huang W, Yarboro MT, Peterson KR, Mackay TF. Quantitative genetics of food intake in Drosophila melanogaster. PLoS One . 2015; 10( 9): e0138129. Google Scholar CrossRef Search ADS PubMed  43. Ayroles JF, Buchanan SM, O’Leary Cet al.   Behavioral idiosyncrasy reveals genetic control of phenotypic variability. Proc Natl Acad Sci U S A . 2015; 112( 21): 6706– 6711. Google Scholar CrossRef Search ADS PubMed  44. Morgante F, Sørensen P, Sorensen DA, Maltecca C, Mackay TF. Genetic architecture of micro-environmental plasticity in Drosophila melanogaster. Sci Rep . 2015; 5: 9785 doi:10.1038/srep09785. Google Scholar CrossRef Search ADS PubMed  45. Harbison ST, McCoy LJ, Mackay TF. Genome-wide association study of sleep in Drosophila melanogaster. BMC Genomics . 2013; 14: 281 doi:10.1186/ 1471-2164-281. Google Scholar CrossRef Search ADS PubMed  46. Huang W, Massouras A, Inoue Yet al.   Natural variation in genome architecture among 205 Drosophila melanogaster Genetic Reference Panel lines. Genome Res . 2014; 24( 7): 1193– 1208. Google Scholar CrossRef Search ADS PubMed  47. Mackay TF, Richards S, Stone EAet al.   The Drosophila melanogaster Genetic Reference Panel. Nature . 2012; 482( 7384): 173– 178. Google Scholar CrossRef Search ADS PubMed  48. Isaac RE, Li C, Leedale AE, Shirras AD. Drosophila male sex peptide inhibits siesta sleep and promotes locomotor activity in the post-mated female. Proc Biol Sci . 2010; 277( 1678): 65– 70. Google Scholar CrossRef Search ADS PubMed  49. Ganguly-Fitzgerald I, Donlea J, Shaw PJ. Waking experience affects sleep need in Drosophila. Science . 2006; 313( 5794): 1775– 1781. Google Scholar CrossRef Search ADS PubMed  50. Falconer DS, Mackay TF. Introduction to Quantitative Genetics . 4th ed. Edinburgh Gate, Harlow: Addison Wesley Longman Limited; 1996. 51. Chow CY, Kelsey KJ, Wolfner MF, Clark AG. Candidate genetic modifiers of retinitis pigmentosa identified by exploiting natural variation in Drosophila. Hum Mol Genet . 2016; 25( 4): 651– 659. Google Scholar CrossRef Search ADS PubMed  52. Dembeck LM, Huang W, Magwire MM, Lawrence F, Lyman RF, Mackay TF. Genetic architecture of abdominal pigmentation in Drosophila melanogaster. PLoS Genet . 2015; 11( 5): e1005163. Google Scholar CrossRef Search ADS PubMed  53. Vonesch SC, Lamparter D, Mackay TF, Bergmann S, Hafen E. Genome-wide analysis reveals novel regulators of growth in Drosophila melanogaster. PLoS Genet . 2016; 12( 1): e1005616. Google Scholar CrossRef Search ADS PubMed  54. Unckless RL, Rottschaefer SM, Lazzaro BP. A genome-wide association study for nutritional indices in Drosophila. G3 (Bethesda) . 2015; 5( 3): 417– 425. Google Scholar CrossRef Search ADS PubMed  55. Ivanov DK, Escott-Price V, Ziehm Met al.   Longevity GWAS using the Drosophila Genetic Reference Panel. J Gerontol A Biol Sci Med Sci . 2015; 70( 12): 1470– 1478. Google Scholar CrossRef Search ADS PubMed  56. Morozova TV, Huang W, Pray VA, Whitham T, Anholt RR, Mackay TF. Polymorphisms in early neurodevelopmental genes affect natural variation in alcohol sensitivity in adult drosophila. BMC Genomics . 2015; 16: 865. Google Scholar CrossRef Search ADS PubMed  57. Arya GH, Magwire MM, Huang W, Serrano-Negron YL, Mackay TF, Anholt RR. The genetic basis for variation in olfactory behavior in Drosophila melanogaster. Chem Senses . 2015; 40( 4): 233– 243. Google Scholar CrossRef Search ADS PubMed  58. He X, Zhou S, St Armour GE, Mackay TF, Anholt RR. Epistatic partners of neurogenic genes modulate Drosophila olfactory behavior. Genes Brain Behav . 2016; 15( 2): 280– 290. Google Scholar CrossRef Search ADS PubMed  59. Gaertner BE, Ruedi EA, McCoy LJ, Moore JM, Wolfner MF, Mackay TF. Heritable variation in courtship patterns in Drosophila melanogaster. G3 (Bethesda) . 2015; 5( 4): 531– 539. Google Scholar CrossRef Search ADS PubMed  60. Chow CY, Wolfner MF, Clark AG. Large neurological component to genetic differences underlying biased sperm use in Drosophila. Genetics . 2013; 193( 1): 177– 185. Google Scholar CrossRef Search ADS PubMed  61. Jordan KW, Craver KL, Magwire MM, Cubilla CE, Mackay TF, Anholt RR. Genome-wide association for sensitivity to chronic oxidative stress in Drosophila melanogaster. PLoS One . 2012; 7( 6): e38722. Google Scholar CrossRef Search ADS PubMed  62. Montgomery SL, Vorojeikina D, Huang W, Mackay TF, Anholt RR, Rand MD. Genome-wide association analysis of tolerance to methylmercury toxicity in Drosophila implicates myogenic and neuromuscular developmental pathways. PLoS One . 2014; 9( 10): e110375. Google Scholar CrossRef Search ADS PubMed  63. Vaisnav M, Xing C, Ku HCet al.   Genome-wide association analysis of radiation resistance in Drosophila melanogaster. PLoS One . 2014; 9( 8): e104858. Google Scholar CrossRef Search ADS PubMed  64. Weber AL, Khan GF, Magwire MM, Tabor CL, Mackay TF, Anholt RR. Genome-wide association analysis of oxidative stress resistance in Drosophila melanogaster. PLoS One . 2012; 7( 4): e34745. Google Scholar CrossRef Search ADS PubMed  65. Ellis LL, Huang W, Quinn AMet al.   Intrapopulation genome size variation in D. melanogaster reflects life history variation and plasticity. PLoS Genet . 2014; 10( 7): e1004522. Google Scholar CrossRef Search ADS PubMed  66. Dembeck LM, Boroczky K, Huang W, Schal C, Anholt RR, Mackay TF. Genetic architecture of natural variation in cuticular hydrocarbon composition in Drosophila melanogaster. Elife . 2015; 4: e09861. Google Scholar CrossRef Search ADS PubMed  67. Hunter CM, Huang W, Mackay TF, Singh ND. The genetic architecture of natural variation in recombination rate in Drosophila melanogaster. PLoS Genet . 2016; 12( 4): e1005951. Google Scholar CrossRef Search ADS PubMed  68. Lobell AS, Kaspari RR, Serrano Negron YL, Harbison ST. The genetic architecture of ovariole number in Drosophila melanogaster: genes with major, quantitative, and pleiotropic effects. G3 (Bethesda) . 2017; 7( 7): 2391– 2403. Google Scholar CrossRef Search ADS PubMed  69. Shorter J, Couch C, Huang Wet al.   Genetic architecture of natural variation in Drosophila melanogaster aggressive behavior. Proc Natl Acad Sci U S A . 2015; 112( 27): E3555– E3563. Google Scholar CrossRef Search ADS PubMed  70. Zwarts L, Vanden Broeck L, Cappuyns Eet al.   The genetic basis of natural variation in mushroom body size in Drosophila melanogaster. Nat Commun . 2015; 6: 10115 doi:10.1038/ncomms10115. Google Scholar CrossRef Search ADS PubMed  71. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Statist Soc B . 1995; 57( 1): 289– 300. 72. Purcell S, Neale B, Todd-Brown Ket al.   PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet . 2007; 81( 3): 559– 575. Google Scholar CrossRef Search ADS PubMed  73. Hu Y, Flockhart I, Vinayagam Aet al.   An integrative approach to ortholog prediction for disease-focused and other functional studies. BMC Bioinformatics . 2011; 12: 357 doi: 10.1186/1471-2105-12-357. Google Scholar CrossRef Search ADS PubMed  74. Isaac RE, Li C, Leedale AE, Shirras AD. Drosophila male sex peptide inhibits siesta sleep and promotes locomotor activity in the post-mated female. Proc Biol Sci . 2010; 277( 1678): 65– 70. Google Scholar CrossRef Search ADS PubMed  75. Huber R, Hill SL, Holladay C, Biesiadecki M, Tononi G, Cirelli C. Sleep homeostasis in Drosophila melanogaster. Sleep . 2004; 27( 4): 628– 639. Google Scholar CrossRef Search ADS PubMed  76. Bellen HJ, Levis RW, He Yet al.   The Drosophila gene disruption project: progress using transposons with distinctive site specificities. Genetics . 2011; 188( 3): 731– 743. Google Scholar CrossRef Search ADS PubMed  77. Venken KJ, Schulze KL, Haelterman NAet al.   MiMIC: a highly versatile transposon insertion resource for engineering Drosophila melanogaster genes. Nat Methods . 2011; 8( 9): 737– 743. Google Scholar CrossRef Search ADS PubMed  78. Cirelli C, Bushey D, Hill Set al.   Reduced sleep in Drosophila Shaker mutants. Nature . 2005; 434( 7037): 1087– 1092. Google Scholar CrossRef Search ADS PubMed  79. Vienne J, Spann R, Guo F, Rosbash M. Age-related reduction of recovery sleep and arousal threshold in Drosophila. Sleep . 2016; 39( 8): 1613– 1624. Google Scholar CrossRef Search ADS PubMed  80. Harbison ST, Sehgal A. Quantitative genetic analysis of sleep in Drosophila melanogaster. Genetics . 2008; 178( 4): 2341– 2360. Google Scholar CrossRef Search ADS PubMed  81. Mackay TFC, Huang W. Charting the genotype-phenotype map: lessons from the Drosophila melanogaster Genetic Reference Panel. Wiley Interdiscip Rev Dev Biol . 2017 doi:10.1002/wdev289. 82. Yuan Q, Joiner WJ, Sehgal A. A sleep-promoting role for the Drosophila serotonin receptor 1A. Curr Biol . 2006; 16( 11): 1051– 1062. Google Scholar CrossRef Search ADS PubMed  83. Foltenyi K, Greenspan RJ, Newport JW. Activation of EGFR and ERK by rhomboid signaling regulates the consolidation and maintenance of sleep in Drosophila. Nat Neurosci . 2007; 10( 9): 1160– 1167. Google Scholar CrossRef Search ADS PubMed  84. Kaasik K, Kivimäe S, Allen JJet al.   Glucose sensor O-GlcNAcylation coordinates with phosphorylation to regulate circadian clock. Cell Metab . 2013; 17( 2): 291– 302. Google Scholar CrossRef Search ADS PubMed  85. Cho Y, Griswold A, Campbell C, Min KT. Individual histone deacetylases in Drosophila modulate transcription of distinct genes. Genomics . 2005; 86( 5): 606– 617. Google Scholar CrossRef Search ADS PubMed  86. Fitzsimons HL, Schwartz S, Given FM, Scott MJ. The histone deacetylase HDAC4 regulates long-term memory in Drosophila. PLoS One . 2013; 8( 12): e83903. Google Scholar CrossRef Search ADS PubMed  87. Fogg PC, O’Neill JS, Dobrzycki Tet al.   Class IIa histone deacetylases are conserved regulators of circadian function. J Biol Chem . 2014; 289( 49): 34341– 34348. Google Scholar CrossRef Search ADS PubMed  88. Luo W, Sehgal A. Regulation of circadian behavioral output via a MicroRNA-JAK/STAT circuit. Cell . 2012; 148( 4): 765– 779. Google Scholar CrossRef Search ADS PubMed  89. Helfrich C. Role of the optic lobes in the regulation of the locomotor activity rhythm of Drosophila melanogaster: behavioral analysis of neural mutants. J Neurogenet . 1986; 3( 6): 321– 343. Google Scholar CrossRef Search ADS PubMed  90. Shilo BZ. Regulating the dynamics of EGF receptor signaling in space and time. Development . 2005; 132( 18): 4017– 4027. Google Scholar CrossRef Search ADS PubMed  91. Hindorff LA, Sethupathy P, Junkins HAet al.   Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc Natl Acad Sci U S A . 2009; 106( 23): 9362– 9367. Google Scholar CrossRef Search ADS PubMed  92. Koh K, Evans JM, Hendricks JC, Sehgal A. A Drosophila model for age-associated changes in sleep:wake cycles. Proc Natl Acad Sci U S A . 2006; 103( 37): 13843– 13847. Google Scholar CrossRef Search ADS PubMed  93. Becker SP, Sidol CA, Van Dyk TR, Epstein JN, Beebe DW. Intraindividual variability of sleep/wake patterns in relation to child and adolescent functioning: A systematic review. Sleep Med Rev . 2017; 34: 94– 121. Google Scholar CrossRef Search ADS PubMed  94. van Alphen B, Yap MH, Kirszenblat L, Kottler B, van Swinderen B. A dynamic deep sleep stage in Drosophila. J Neurosci . 2013; 33( 16): 6917– 6927. Google Scholar CrossRef Search ADS PubMed  95. Aulsebrook AE, Jones TM, Rattenborg NC, Roth TC2nd, Lesku JA. Sleep ecophysiology: integrating neuroscience and ecology. Trends Ecol Evol . 2016; 31( 8): 590– 599. Google Scholar CrossRef Search ADS PubMed  Published by Oxford University Press on behalf of Sleep Research Society (SRS) 2017. This work is written by (a) US Government employee(s) and is in the public domain in the US. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png SLEEP Oxford University Press

Genotype Influences Day-to-Day Variability in Sleep in Drosophila melanogaster

Loading next page...
 
/lp/ou_press/genotype-influences-day-to-day-variability-in-sleep-in-drosophila-Z4KBtWDAqL
Publisher
Oxford University Press
Copyright
Published by Oxford University Press on behalf of Sleep Research Society (SRS) 2017. This work is written by (a) US Government employee(s) and is in the public domain in the US.
ISSN
0161-8105
eISSN
1550-9109
D.O.I.
10.1093/sleep/zsx205
Publisher site
See Article on Publisher Site

Abstract

Abstract Patterns of sleep often vary among individuals. But sleep and activity may also vary within an individual, fluctuating in pattern across time. One possibility is that these daily fluctuations in sleep are caused by the underlying genotype of the individual. However, differences attributable to genetic causes are difficult to distinguish from environmental factors in outbred populations such as humans. We therefore employed Drosophila as a model of intra-individual variability in sleep using previously collected sleep and activity data from the Drosophila Genetic Reference Panel, a collection of wild-derived inbred lines. Individual flies had significant daily fluctuations in their sleep patterns, and these fluctuations were heritable. Using the standard deviation of sleep parameters as a metric, we conducted a genome-wide association study. We found 663 polymorphisms in 104 genes associated with daily fluctuations in sleep. We confirmed the effects of 12 candidate genes on the standard deviation of sleep parameters. Our results suggest that daily fluctuations in sleep patterns are due in part to gene activity. intra-individual variability, sleep standard deviation, Drosophila melanogaster, sleep Statement of Significance Previous studies in humans have observed intra-individual variability in sleep phenotypes, and there may be a connection between this variability and both sleep and mental health disorders. This study, conducted in Drosophila, reveals that day-to-day fluctuations in sleep are due, in part, to the underlying genotype of the individual. We identified 104 candidate genes for intra-individual variability, some of which have predicted homology with human genes. We confirmed the effects of 12 of these genes on intra-individual variability. Daily fluctuations in sleep in flies are thus regulated in part by genes, and these results may extend to human sleep as well. Introduction Although the need to sleep is persistent within a species, the manner in which that need is fulfilled often varies considerably among individuals. Modern-day human populations living in pre-industrial environments have characteristic sleep patterns consistent with ecological needs [1]. Sleep in some groups features a single bout of sleep at night with rare instances of napping [2], whereas others have a combination of night sleep bouts and daily napping [1, 3]. Historical evidence suggests that humans typically had two sleep bouts during the night [4], and this differs from the modern concept of a single night sleep bout. Although different among individuals, some measures of sleep and sleep-related characteristics are remarkably stable within individuals and are characterized by low within-individual variance [5–8]. However, some characteristics of sleep are highly variable within an individual. Often measured as the standard deviation or coefficient of variation [9–14], total sleep time or time in bed [13, 15–17], sleep onset latency [13, 15, 16], sleep quality or efficiency [15, 16], daytime sleepiness [17], and wake after sleep onset [13, 15, 16] can be more variable within a single individual than among individuals. This intra-individual variability has been observed using sleep diaries, actigraphy, and polysomnography in both children [13, 16, 17] and adults [13, 15]. Although the causes and consequences of daily fluctuations in sleep are not understood, one intriguing possibility is that high intra-individual variability in sleep is a hallmark of sleep disorder and mental illness. Greater variability in night-to-night sleep characteristics has been observed in those with chronic insomnia than in controls [18, 19], and patients with post-traumatic stress disorder (PTSD) exhibit exceptionally fragmented sleep patterns [20]. Daily fluctuations in obstructive sleep apnea (OSA) severity class and periodic leg movements have been observed as well [21, 22]. Variability in sleep duration has been observed in those with bipolar disorder [23, 24], as has interdaily stability, the similarity of activity patterns across days [24]. Intra-individual variability in activity patterns, including the amount, predictability, and intricacy, is also associated with bipolar disorder (reviewed in Ref. [25]). Intra-individual variability in sleep onset time and duration in children could be used to classify those with attention deficit hyperactivity disorder (ADHD), a distinction that could not be made using the averages of sleep parameters [14]. Correlates between physiological endophenotypes and high intra-individual variability in sleep characteristics have also been described, including flattening cortisol responses [26], increasing blood concentrations of inflammatory markers [27], and abdominal obesity [28]. Stress, work demands, iron deficiency, and moderate exercise can all mediate intra-individual variability in sleep [10–12, 29, 30]. Thus, intra-individual variability in sleep can be said to be plastic, i.e., it can be altered in response to changes in the environment [31]. The degree to which an individual can adapt to changing environmental conditions is ultimately due to the underlying genotype, however. Few studies have attempted to map genes influencing intra-individual variability in sleep thus far. Linkage mapping in humans identified two candidate genes, JARID1a and CACNA1C, associated with interdaily stability [24]. Spada and colleagues attempted to map sleep onset irregularity, measured as the standard deviation of the timing of sleep onset, but the associations did not reach their genome-wide threshold of significance, though some nominal associations were found [32]. These difficulties occur in part because of the difficulties in estimating trait variability in an outbred population accurately. Sources of heritable genetic variance and environmental variance are confounded in outbred populations, making it challenging to distinguish between the two [33]. Model organism populations have the potential to address this problem. Model organisms can be inbred so that many simultaneous measurements can be made on a single genotype, and environmental conditions can be controlled to a greater extent. Previously, model organism populations have been used to study the genetic basis of inter-individual variability in morphological and fitness traits in Arabidopsis and maize [34–36]; gene expression in isogenic strains of yeast [37, 38] and inbred populations of flies [39]; and wing shape [40], bristle number [41], food consumption [42], locomotor handedness [43], chill coma recovery, startle response, starvation resistance [44], and sleep [45] in flies. Measured as the coefficient of environmental variation (CVE), the differences in sleep parameters among individual flies were found to have a heritable, mappable basis [45]. Thus, model organism populations offer the opportunity to investigate among-animal variability and its genetic underpinnings. We wanted to determine whether this strategy could also be applied to study intra-individual variability in sleep, and if so, whether this variability had an underlying genetic cause. We used previously collected data from a genome-wide association study of the Drosophila Genetic Reference Panel (DGRP) to investigate this possibility [45]. We examined daily fluctuations in the sleep patterns of 10239 flies. Daily fluctuations in sleep were more variable in some genotypes than in others, revealing that the variation is heritable. Expressing the variability in sleep traits as a standard deviation enabled us to conduct a genome-wide association study of intra-individual variability. We mapped 104 candidate genes to seven sleep parameters varying over days in flies. These genes included plausible candidates from previous studies in flies and mapped to 51 human homologs. Day-to-day fluctuations in sleep characteristics may therefore depend in part upon underlying genotype. Methods Assessment of Individual Differences in Day-to-Day Sleep We reanalyzed previously collected [45] data to determine whether sleep fluctuated across days in individual flies and whether that fluctuation had a heritable component. In that experiment, 168 lines of flies from the DGRP [46, 47] were randomly divided into four equal blocks with four replicates. Each replicate had eight flies per line per sex, for a total of 64 flies per line. Flies were assayed for rest and activity behavior under standard environmental conditions (cornmeal–molasses–agar medium, 25°C, 60%–75% relative humidity, and 12-hr light:dark cycle). Virgin males and females were assayed in order to avoid any effect of mating status on sleep [48], and 30 same-sex flies were maintained per vial for 4 days to standardize social exposure effects [49]. Sleep and activity recordings were made for each fly using the Drosophila Activity Monitoring System (Trikinetics, Waltham, MA). Seven days of recordings were made for each fly; the first day was discarded to mitigate any potential effect of CO2 anesthetization recovery on sleep and activity. The flies were visually inspected after the experiment and data from dead flies were eliminated from the data set. The raw sleep and activity data were used to calculate sleep parameters for each day using an in-house C# program. Sleep duration, numbers of sleep bouts, and average sleep bout length during the day and night were calculated. In addition, waking activity, the number of activity counts per minute spent awake, was determined. We assessed the change in sleep parameters across days using the four-way random ANOVA model, Y=μ+D+B+L(B)+R(B)+D×L(B)+D×R(B)+L×R(B)+D×L×R(B)+ε where D represents the random effect of day, B represents the random effect of block, L represents the random effect of line, R represents the random replicate effect, and ε represents the error variance. Analyses were performed separately for males and females. A significant D term indicates that the sleep phenotype is different across days, whereas a significant D × L(B) term indicates that the fluctuation in a sleep parameter across days has a heritable component. In addition, we evaluated males and females of each line separately for day effects using the reduced model, Y = µ + D + R + D×R + ε. We used SAS 9.3 (SAS Institute, Cary, NC) to evaluate these models. Quantitative Genetic Analysis We found a significant D × L(B) term for all the sleep traits, indicating that sleep was variable across days and that a portion of this variability was under genetic control (Supplementary Table S1). We used the within-fly standard deviation in sleep across days, σ, to represent the variability in each sleep trait per fly [9–13]. We noted that average day bout length σ and waking activity σ were not normally distributed traits. We normalized these two traits by taking the natural log of σ for each fly. We analyzed σ for each trait using the ANOVA model,  Y=μ+B+ L(B)+S+R(B)+S×L(B)+L×R(B)+S×R(B)+S×L×R(B)+ ε where B, L, and R are as defined above, and S is sex. We analyzed the data for each sex separately using the reduced model, Y=μ+B+ L(B)+R(B)+L×R(B)+ εThe variance components for each model were estimated using the restricted maximum likelihood (REML) method. The broad sense heritability for each trait was estimated as  H2=(σL2+σSL2)/(σL2+σSL2+σε2) for both sexes combined, where σL2 is the line variance component, σSL2 is the line by sex variance component, and σε2 is the sum of the other variance components [50]. The broad sense heritability was estimated as H2=(σL2)/(σL2+σε2) for each sex separately. We calculated the cross-sex genetic correlation as rmf=(σL2)/(σLM2+σLF2) where σL2 represents the line variance component for both sexes combined, σLM2 represents the line variance component for males, and σLF2 represents the line variance component for females [50]. In addition, we computed phenotypic correlations between sleep σ traits and other traits measured in the DGRP having line means available, including mean sleep and sleep CVE [45]; morphological traits (eye area and size, pigmentation level, wing centroid size, and eye interocular distance) [51–53]; lifespan and metabolic traits [54, 55]; behavior (reproductive or courtship traits, food consumption, chemosensation, startle response, geotaxis, and aggression) [42, 44, 56–61]; environmental responses (oxidative stress, methylmercury tolerance, radiation stress, chill coma, and starvation) [44, 61–64]; and genome size [65]. Phenotypic correlations were calculated as Pearson product moment correlation coefficients. We also estimated the genetic correlations rG between mean sleep and sleep σ traits as cov12/σL12×σL22 [50]. SAS 9.3 and JMP 13.0.0 (SAS Institute, Cary, NC) were used to evaluate these models. Genotype–Phenotype Associations Whole-genome sequence data are available for the DGRP, and a website tool (http://dgrp2.gnets.ncsu.edu/, accessed 21 December 2017) is available to calculate genome-wide associations among variants in this population and any trait of interest [46, 47]. Using the average σ per line for each sleep trait, we computed genome-wide associations with 1920276 polymorphisms segregating in the DGRP. Only polymorphic variants with minor allele frequencies of 5% or more were used in the associations. Some lines of the DGRP are known to have Wolbachia pipientis infection and large chromosomal inversions. We checked for associations between sleep σ parameters and Wolbachia and inversions. No significant associations with Wolbachia infection were present, but there were significant associations with chromosomal inversions for average day bout duration σ (associated with the In(2R)NS and In(3R)Mo inversions, p = .0003 and p < .0001, respectively), average night bout duration σ (associated with In(2L)t, p = .0144), night bout number σ (associated with In(2R)NS, p = .0051), and waking activity σ (associated with In(3R)Mo, p = .0008). We applied the linear mixed model, y=Xb+Zu+ε, where y are the adjusted phenotypes, X is the fixed effect of polymorphisms, and Z represents the covariance matrix accounting for population structure [46]. Associations with a p-value of 1 × 10–5 or less were called significant, a threshold applied in many DGRP studies [42, 52, 57, 66–70]. This threshold was supported by Q–Q plots (Supplementary Figure S1), which generally deviated little from expected values. In addition, false discovery rates (FDRs) were calculated for each polymorphic association using the method of Benjamini and Hochberg [71]. We determined the linkage disequilibrium (LD) (r2) among significant polymorphisms using PLINK [72]. Finally, human homologs of Drosophila genes were identified using the DIOPT website, which compiles homology data from multiple genomic databases [73]. Sleep Assays and Phenotypes For all sleep assays, flies were maintained in a constant temperature (25°C), constant humidity (60%) incubator and assayed on standard food (http://flystocks.bio.indiana.edu/Fly_Work/media-recipes/bloomfood.htm, accessed 21 December 2017) in standard light:dark cycle (12:12 hr) conditions. Male and female virgins were collected and maintained in single-sex vials of 20 flies for 4 days before sleep measurements to mitigate potential effects of mating status [74] and social exposure [49]. We loaded individual flies into Drosophila Activity Monitors (DAM2, Trikinetics, Waltham, MA) to measure sleep and activity parameters. Numbers of activity counts for each fly were recorded per minute; we defined sleep as 5 min without an activity count [75]. We discarded the first day’s data as the flies would be recovering from CO2 anesthesia and/or adjusting to the monitor tubes during the first day. We examined each fly after the experiment; the data from flies not surviving the experiment were discarded. We used an in-house C# program to calculate sleep parameters for each day: night, day, and 24-hr sleep duration in minutes; night and day sleep bout number; night and day average sleep bout length; sleep latency, the number of minutes before the fly’s first sleep bout after lights are turned off; and waking activity, the number of activity counts per minute spent awake. Using these parameters, we calculated the standard deviation of each sleep trait over the 6-day monitoring period for each fly. Sleep Deprivation Protocol The most variable DGRP lines were deprived of sleep using the following procedure. We placed individual flies into the DAM2 Trikinetics monitors using the protocol above. The flies acclimated to the monitors for a single day, and then we recorded 2 baseline days of rest and activity. We used a vortexer (Troemner, Thorofare, NJ) fitted with a custom mounting plate (Trikinetics, Waltham, MA) to apply a mechanical stimulus on the night of the third day. Each minute, the vortexer gently shook the flies for 5 s every minute of the 12-hr night sleep period (shaking speed = 2.0). The flies’ recovery was monitored for 2 additional days. An identical set of flies were loaded into Trikinetics monitors on a different shelf of the same incubator to serve as nonsleep-deprived controls. We used this protocol to deprive the following DGRP lines: DGRP_41, DGRP_73, DGRP_153, DGRP_335, DGRP_338, DGRP_409, DGRP_646, and DGRP_892. We sleep-deprived 16 flies of each sex per line. We used an in-house Python algorithm to determine the average minutes of sleep per 30 min. We evaluated differences in 24-hr sleep for each DGRP line separately using the ANOVA model, Y = µ + T + D + S + T×D + T×S + D×S + T×D×S + ε, where T is the control or sleep-deprived treatment, D is day, and S is sex. We analyzed the data for each population or line separately using the reduced ANOVA model, Y = µ + D + S + D×S + ε, where D and S are as defined above. We applied a post hoc Tukey analysis to identify significant loss and gain of sleep. Verification of Candidate Genes We tested 17 Minos element (Mi{ET1} and Mi{y[+mDint2] = MIC}) insertion lines in 14 candidate genes identified by the GWAS: 5-hydroxytryptamine (serotonin) receptor 1A (5-HT1A), CG42260, CG5888, CG7985, fruitless (fru), histone deacetylase 4 (HDAC4), rhomboid (rho), scab (scb), sallimus (sls), solwind (sowi), super sex combs (sxc), Transmembrane channel-like (Tmc), turtle (tutl), and ZnT35C (Supplementary Table S2). The inserts for sls and sowi were not viable as homozygotes and were tested as heterozygotes. Lines with Mi{ET1} insertions were created in an isogenic w1118 background (w1118[5905]); hence, we used this background as a control [76]. Lines with Mi{y[+mDint2] = MIC} insertions were crossed to a common background when created: y[1] w[*]; nub[2] b[1] sna[Sco] pr[1] cn[1]/CyO [3628] [77]. We used this background as a control for lines with the Mi{y[+mDint2]=MIC} insertion. All stocks were obtained from the Bloomington, IN stock center. Sleep phenotypes were measured as described above for 16 flies per sex per genotype in the mutants and their respective controls. Mutant sleep σ phenotypes were compared with the control using the ANOVA model, Y = µ + G + S + G×S + ε, where G is genotype, S is sex, and ε is the error term. Males and females were analyzed separately using the reduced ANOVA model, Y = µ + G + ε. Results Quantitative Genetics of Day-to-Day Variability in Sleep Daily fluctuations occurred in all sleep traits among individual flies. Figure 1 contrasts night sleep traits among individuals from the least variable DGRP line with those of the most variable line, illustrating how extreme the differences can be (for day sleep traits, see Supplementary Figure S2). Night sleep duration in DGRP_41, the least variable line, was high and changed little across time (Figure 1A). In contrast, sleep in DGRP_153, the most variable line, fluctuated through the full range of possible night sleep values (Figure 1A). Numbers of night bouts were relatively low in DGRP_531, but fluctuated greatly in DGRP_882 (Figure 1B). Stable and fluctuating patterns were also observed for average night bout length (Figure 1C). These daily fluctuations among individuals translated into changes in overall patterns of sleep for some DGRP lines. For example, male flies of DGRP_181 had increasing night nap length over the course of 6 days; simultaneously, the average number of night bouts decreased, whereas overall night sleep duration stayed constant (Figure 2A). Thus, these males had less fragmented sleep patterns over the 6-day observation period. Females of DGRP_761, on the other hand, had increased numbers of nightly naps and decreased average nap times, suggesting an increasingly fragmented pattern (Figure 2B). In addition, some flies, such as males of DGRP_73, had stable sleep characteristics over time (Figure 2C). We analyzed the entire population for significant changes across day. With the exception of day sleep duration in females, all sleep traits varied significantly across time (Supplementary Table S1). We observed a highly significant interaction of day and line for all sleep traits (Supplementary Table S1; all pLine×Day(Block) < .0001), suggesting that this variability was heritable. We also used a reduced model ANOVA to determine the numbers of DGRP lines with significant day-to-day fluctuations in sleep (Table 1). Waking activity was variable for nearly every line, with 87.5% and 91% of the lines having significant daily fluctuations in males and females, respectively. Average day bout length was the least variable trait, with only 26.2% of the lines varying across day. These results show that daily fluctuations in sleep parameters are common in flies, with some genotypes exhibiting stability while others are characteristically variable. Table 1. Numbers of Lines With Significant Daily Fluctuations in Sleep Parameters Sleep Trait  Males  Females  No. of variable lines  Percentage  No. of variable lines  Percentage  Day sleep (min)  98  58.3  84  50.0  Night sleep (min)  89  53.0  85  50.6  Day bout number  69  41.1  67  39.9  Night bout number  69  41.1  59  35.1  Avg. day bout length (min)  49  29.2  41  24.4  Avg. night bout length (min)  54  32.1  50  29.8  Waking activity (cts/min)  147  87.5  153  91.1  Sleep Trait  Males  Females  No. of variable lines  Percentage  No. of variable lines  Percentage  Day sleep (min)  98  58.3  84  50.0  Night sleep (min)  89  53.0  85  50.6  Day bout number  69  41.1  67  39.9  Night bout number  69  41.1  59  35.1  Avg. day bout length (min)  49  29.2  41  24.4  Avg. night bout length (min)  54  32.1  50  29.8  Waking activity (cts/min)  147  87.5  153  91.1  The table lists the number and percentage of DGRP lines with significant daily variability in the sleep parameter indicated (pDay < .05). The total number of DGRP lines with sleep phenotypes is 168. cts = counts. View Large Figure 1. View largeDownload slide Example of variability in night sleep among individual flies. The plots contrast sleep changes over days for the least variable DGRP line with the most variable DGRP line for that trait. (A) Night sleep duration. The least variable line is DGRP_41 (n = 64); the most variable line is DGRP_153 (n = 61). (B) Night bout number. The least variable line is DGRP_531 (n = 62); the most variable line is DGRP_882 (n = 54). (C) Average night bout length. The least variable line is DGRP_42 (n = 64); the most variable line is DGRP_338 (n = 64). Figure 1. View largeDownload slide Example of variability in night sleep among individual flies. The plots contrast sleep changes over days for the least variable DGRP line with the most variable DGRP line for that trait. (A) Night sleep duration. The least variable line is DGRP_41 (n = 64); the most variable line is DGRP_153 (n = 61). (B) Night bout number. The least variable line is DGRP_531 (n = 62); the most variable line is DGRP_882 (n = 54). (C) Average night bout length. The least variable line is DGRP_42 (n = 64); the most variable line is DGRP_338 (n = 64). Figure 2. View largeDownload slide Examples of changes in night sleep parameters over days. Box plots of night sleep duration, night bout number, and average night bout length are plotted for each day. (A) Males of DGRP_181 (n = 32). (B) Females of DGRP_761 (n = 32). (C) Males of DGRP_73 (n = 32). Figure 2. View largeDownload slide Examples of changes in night sleep parameters over days. Box plots of night sleep duration, night bout number, and average night bout length are plotted for each day. (A) Males of DGRP_181 (n = 32). (B) Females of DGRP_761 (n = 32). (C) Males of DGRP_73 (n = 32). Lines with characteristically variable sleep could also potentially have altered responses to disruption of the sleep homeostat. We examined the response of several DGRP lines to a single night of mechanical sleep deprivation: DGRP_338, DGRP_892, DGRP_41, and DGRP_646, which had the most variable average night bout length; DGRP_335, DGRP_409, DGRP_73 (and DGRP_646), which had the most variable waking activity; and DGRP_153, which had the most variable night sleep. Flies from all eight lines were significantly deprived of sleep compared with their baseline sleep times (Figure 3). However, the sleep loss in flies of line DGRP_335 was quite low; only 104.7 min of sleep were lost on average (Figure 3). All of the lines had a statistically significant increase in their 24-hr sleep on the day after sleep deprivation compared with their baseline sleep, with the exception of DGRP_335 and DGRP_73. DGRP_335 did not lose much sleep, so one possibility is that the homeostat was not perturbed sufficiently in these flies to elicit a response. However, flies from line DGRP_73 lost 540.5 min of sleep during the night, and only gained 99.8 min during the next 24-hr period. Thus, although the recovery sleep in most of the lines was similar to that of other wild-derived and laboratory strains [78, 79], the rebound sleep of DGRP_335 and DGRP_73 suggests that genotypes with variable sleep may also have an altered homeostatic response. Figure 3. View largeDownload slide Sleep loss and recovery in DGRP lines with the greatest daily fluctuations in sleep. *p < .05. For DGRP_41, n = 32; DGRP_73, n = 29; DGRP_153, n = 31; DGRP_335, n = 30; DGRP_338, n = 30; DGRP_409, n = 28; DGRP_646, n = 30; and DGRP_892, n = 32. Figure 3. View largeDownload slide Sleep loss and recovery in DGRP lines with the greatest daily fluctuations in sleep. *p < .05. For DGRP_41, n = 32; DGRP_73, n = 29; DGRP_153, n = 31; DGRP_335, n = 30; DGRP_338, n = 30; DGRP_409, n = 28; DGRP_646, n = 30; and DGRP_892, n = 32. We modeled daily fluctuations in sleep traits as the standard deviation (σ) per fly (Supplementary Table S3) in a manner analogous to studies of intra-individual variability in human sleep parameters [9–13]. We observed highly significant differences in σ among lines (all pLine(Block) < .0001), and the variability was sexually dimorphic as well (all pLine×Sex(Block) < .0001) (Supplementary Table S4). Histograms of σ for night sleep traits (Figure 4) and day sleep traits (Figure 5) illustrate this variability among lines of the DGRP. The significant differences among lines suggested that sleep σ was heritable; thus, we calculated broad-sense heritabilities to quantify the genetic contribution to each trait. Heritabilities for day and night sleep duration σ were 0.49 and 0.26, respectively; 0.07 and 0.13 for day and night bout number σ, respectively; 0.30 for both day and night average bout length σ; and 0.21 for waking activity σ. With the exception of the low heritability for day bout number σ, heritabilities were moderate, suggesting that daily fluctuations in sleep are partially driven by genotype. Figure 4. View largeDownload slide Histograms of night sleep σ traits. The histogram shows male line means in blue, and female line means in pink. (A) Night sleep duration σ. (B) Night bout number σ. (C) Avg. night bout length σ. Figure 4. View largeDownload slide Histograms of night sleep σ traits. The histogram shows male line means in blue, and female line means in pink. (A) Night sleep duration σ. (B) Night bout number σ. (C) Avg. night bout length σ. Figure 5. View largeDownload slide Histograms of day sleep σ traits. The histogram shows male line means in blue, and female line means in pink. (A) Day sleep duration σ. (B) Day bout number σ. (C) Normalized avg. day bout length σ. (D) Normalized waking activity σ. Figure 5. View largeDownload slide Histograms of day sleep σ traits. The histogram shows male line means in blue, and female line means in pink. (A) Day sleep duration σ. (B) Day bout number σ. (C) Normalized avg. day bout length σ. (D) Normalized waking activity σ. We calculated the phenotypic and genetic correlations among sleep σ phenotypes (Supplementary Table S5). Many of these correlations were significantly different from zero; however, the correlations were not necessarily high (i.e., the correlations were not close to 1). For example, night sleep duration σ had significant genetic correlations with all other sleep σ traits, but genetic correlations with other night-related traits were higher (0.573 and −0.459 with night bout number σ and average night bout length σ, respectively) than day and activity traits (0.256, −0.222, and −0.152 with day bout number σ, average day bout length σ, and waking activity σ, respectively). These correlations suggest a shared genetic architecture underlying daily fluctuations in sleep, but the overlap is far from complete. We also calculated the phenotypic and genetic correlations with the previously published mean and CVE sleep phenotypes for these lines. Sleep σ traits tended to be correlated with their respective mean and CVE, and, in general, these correlations were moderate to high (Supplementary Table S6). Night sleep σ and mean night sleep had a negative genetic correlation of rG = −0.80, whereas day sleep σ and mean day sleep had an rG of −0.44; that is, lower mean sleep duration was more variable. The trend for night sleep σ may also be true of single-gene mutations; we examined data from a previous p-element mutagenesis screen [80] and estimated the mutational genetic correlation as rG = −1.07 (p < .0001) between night sleep σ and night sleep. Day sleep σ and day sleep had a positive mutational genetic correlation of rG = 0.61 (p < .0001), however, suggesting the opposite trend. Day bout number σ and waking activity σ were the least correlated among sleep traits, whereas average day and night bout length σ was highly correlated with their mean. Thus, intra-individual variability in sleep has, in part, some distinctive genetic features. We wondered if sleep σ traits might be correlated with other types of traits in flies. The DGRP has been used extensively to study many different complex traits; thus, it was possible to assess the phenotypic correlations with these traits and sleep σ. We calculated the phenotypic correlations between sleep σ traits and morphological traits; lifespan and metabolic traits; behavior; responses to environmental stresses; and genome size [42, 44, 51–65]. Many of the correlations with other traits were significantly different from zero, but their magnitudes were quite low (Supplementary Table S7). Only six correlations were higher than 0.25. Night sleep duration σ was negatively correlated with mean lifespan (r = −0.27) [55] and inter-individual variability in starvation resistance (r = −0.26) [44]. Average night bout length σ was positively correlated with variability in intraocular distance (r = 0.29), a morphological measure related to reproductive fitness in flies [53]. Night bout number σ was correlated with eclosion rate (r = 0.30) [62]. Waking activity σ was negatively correlated with startle response on control medium (r = −0.28) and menadione bisulfite-supplemented medium (r = −0.30), an inducer of oxidative stress [61]. Thus, correlations between sleep σ traits and other types of traits were generally quite low, suggesting a distinct genetic basis for sleep σ. Genotype–Phenotype Associations Using the average σ of each sleep trait per line, we associated each sleep σ phenotype with all polymorphisms known to be segregating in the DGRP which had a minor allele frequency of 0.05 or more: 1920276 variants in total [46]. Associations were evaluated for sexes combined, each sex separately, and the difference between sexes (male avg. − female avg.) (see Supplementary Table S8 for details). Table 2 shows the numbers of polymorphisms that were associated with each sleep σ trait at a threshold p-value of 1 × 10–5 or less. Numbers of significant associations were modest, with 14%–26% of the associated polymorphisms located within the coding region of a gene, and the remaining polymorphisms mapping between genes (Supplementary Figure S3). A preponderance of significant associations had low minor allele frequencies as the median allele frequency was 0.12, a typical result in studies using the DGRP [45, 47, 61, 62, 64, 68]. Thus, variants with larger effect sizes tended to be relatively rare. However, significant associations were also observed at more common alleles as roughly 10% of the associations had minor allele frequencies above 0.40. Only 20 of the 663 associations had FDRs of 0.05 or less, and only 34 had FDRs of 0.10 or less. Despite the modest statistical support, effect sizes for some sleep σ traits were very high. We discuss these results in detail below. Table 2. Numbers of Polymorphisms Significantly Associated With Sleep σ Traits Sleep trait  Number of polymorphisms  Percent within genes  Day sleep σ (min)  145  25.5  Night sleep σ (min)  103  16.5  Day bout number σ  43  14.0  Night bout number σ  76  19.7  Day avg. bout length σ [ln(min)]  86  26.7  Night avg. bout length σ (min)  124  13.7  Waking activity σ [ln(cts/min)]  87  18.4  Sleep trait  Number of polymorphisms  Percent within genes  Day sleep σ (min)  145  25.5  Night sleep σ (min)  103  16.5  Day bout number σ  43  14.0  Night bout number σ  76  19.7  Day avg. bout length σ [ln(min)]  86  26.7  Night avg. bout length σ (min)  124  13.7  Waking activity σ [ln(cts/min)]  87  18.4  View Large Effect sizes for associations with average night bout length σ were very high in magnitude and ranged from −39.2 to 17.7 min. Unlike mammalian genomes, LD decays very rapidly in the DGRP, on the order of 10–30 bp on average [46, 47]. We therefore assumed that a significant variant located within a gene implicated that gene, unless (1) LD was high (r2 ≥ 0.8) among variants, or (2) the trait was associated with a known chromosomal inversion (Methods). LD was generally very low among polymorphisms for average night bout length σ (Figure 6A) except for two regions: one region was on chromosome 2L between 5056207 and 5594455 bp; the other was on chromosome 2R between 5191057 and 5472092 bp. The first region of LD was also within the In(2L)t inversion, which was itself significantly associated with average night bout length σ (Methods). Variants within candidate genes in these two regions, CG31918 and sxc, were therefore not distinguishable from the other polymorphisms in LD around them (8 and 21 variants, respectively). Most of the remaining 93 polymorphisms associated with average night bout length σ were intergenic, but 15 additional polymorphisms were in 12 candidate genes. These candidate genes included 5-HT1A, flower, fru, and HDAC4 (Supplementary Table S8). Figure 6. View largeDownload slide Genome-wide association results for sleep σ traits. Significant polymorphisms (p < 1 × 10−5) are plotted for each trait. The points are color-coded to indicate the most significant association (sexes combined, male, female, or sex difference) as indicated in the legend. The top panel shows the minor allele frequency (MAF) for each polymorphism plotted across chromosomal position. The middle panel plots the effect size a normalized by the genetic standard deviation σG. The lower panel shows the p-values for each polymorphism plotted as the -log of the p-value. The bottom triangle is the LD among polymorphisms calculated as r2. Black lines denote chromosome boundaries. (A) Avg. night bout length σ. (B) Waking activity σ. (C) Night sleep duration σ. (D) Day sleep duration σ. Figure 6. View largeDownload slide Genome-wide association results for sleep σ traits. Significant polymorphisms (p < 1 × 10−5) are plotted for each trait. The points are color-coded to indicate the most significant association (sexes combined, male, female, or sex difference) as indicated in the legend. The top panel shows the minor allele frequency (MAF) for each polymorphism plotted across chromosomal position. The middle panel plots the effect size a normalized by the genetic standard deviation σG. The lower panel shows the p-values for each polymorphism plotted as the -log of the p-value. The bottom triangle is the LD among polymorphisms calculated as r2. Black lines denote chromosome boundaries. (A) Avg. night bout length σ. (B) Waking activity σ. (C) Night sleep duration σ. (D) Day sleep duration σ. We also observed large effect sizes for associations with waking activity σ; these ranged from 0.76 to 1.25 counts per minute. For waking activity σ, 87 significant polymorphisms were found, and virtually no LD among these polymorphisms was observed. As with average night bout length σ, most of these variants were intergenic, but 16 polymorphisms were in 13 genes (Figure 6B; Supplementary Table S8), including rho, roughoid (ru), and Heat shock protein 26. Seventeen polymorphisms had very low FDRs (<0.0001), including one in roughoid and two in Heat shock protein 26. Variants associated with night and day sleep duration σ had smaller effect sizes. Night sleep σ effects ranged from −17.6 to 12.1 min and day sleep σ effects ranged from −14.3 to 7.5 min. 103 polymorphisms were significantly associated with night sleep σ. A large region on Chromosome 3L was in LD (23531038 to 24006114 bp) among 10 intergenic polymorphisms significantly associated with night sleep σ (Figure 6C). Eighteen variants implicated 14 candidate genes, including sls, Secretory 8, and unpaired 1 (Supplementary Table S8). One hundred forty-five polymorphisms were associated with day sleep σ. There were 29 genes identified for day sleep σ, including skittles, minibrain, and kirre (Supplementary Table S8). Little LD was evident for day sleep σ (Figure 6D). One possibility is that the genetic architecture for sleep σ overlaps with mean/CVE sleep as the genetic correlations among these traits were generally significant. We therefore compared our results with the previous genome-wide association study of sleep in Drosophila [45]. We found 22 unique polymorphisms that overlapped with sleep σ traits, which was 3.3% of the total identified in this study (Supplementary Table S9). Most of the overlap was of intergenic variants, but polymorphisms in the genes CG13699, CR45711, flower, and rho overlapped among traits. Many polymorphisms identified in this study are therefore unique to sleep σ traits. Finally, we searched for human homologs of candidate sleep σ genes using the DIOPT tool. DIOPT searches for homologous genes across 12 databases; DIOPT scores indicate the number of databases that predicted a given homologous gene [73]. A total of 438 human genes had predicted homology with 78 Drosophila genes, though in many cases the homology was not high. We found 51 human genes that were predicted to have the highest homology with 42 Drosophila genes (Supplementary Table S10). Twenty-seven Drosophila genes had no known human homolog. Thus, a portion of the genes we identified for sleep σ traits have plausible human homologs. Verification of Candidate Genes We tested 17 Minos element constructs in 14 candidate genes for their effects on sleep σ traits (Supplementary Table S2). We chose sleep σ traits with the largest effect sizes to confirm: night average bout length σ, night sleep σ, and waking activity σ (Supplementary Table S8). We tested all genes with available Minos constructs that were identified for these three traits. We found that 12 of the candidate genes confirmed the GWAS prediction (Table 3). 5-HT1A, CG42260, and fru had two Minos insertions available. Both 5-HT1A insertions had significant differences in night average bout length σ from the control; likewise, both fru insertions were significantly different for night average bout length σ. Only one of the CG42260 insertions had significant differences in night sleep σ from the control line, however. All the remaining genes had one Minos construct available for test, and the following genes were confirmed: CG7985, HDAC4, and sxc for night average bout length σ; heterozygous sls for night sleep σ; and tutl, Tmc, rho, heterozygous sowi, and scb for waking activity σ. In addition, some of these constructs had pleiotropic effects on other sleep traits (Supplementary Table S11), consistent with previous observations of gene-level pleiotropy in Drosophila genome-wide studies [45, 81]. These mutant tests validate these genes as candidates affecting daily fluctuations in sleep. Table 3. Verification of Candidate Genes Gene  Allele tested  Sleep trait  Diff. from control  No.  5HT1A  MB09812  Night avg. bout length σ (min)  14.32***  32  5HT1A  MB09978  Night avg. bout length σ (min)  10.61**  31  CG7985  MI00275  Night avg. bout length σ (min)  −113.93***  31  fru  MB02472  Night avg. bout length σ (min)  28.29*  32  fru  MB01996  Night avg. bout length σ (min)  7.09*  30  HDAC4  MI05513  Night avg. bout length σ (min)  −103.05***  32  sxc  MI02286  Night avg. bout length σ (min)  −125.68****  32  CG42260  MB00428  Night sleep σ (min)  −27.87****  31  sls  MI011177 Het.  Night sleep σ (min)  30.78****  32  rho  MI08786  Waking activity σ (cts/min)  −0.85****  32  scb  MI010788  Waking activity σ (cts/min)  −0.82****  32  sowi  MI03342 Het.  Waking activity σ (cts/min)  −0.97****  32  Tmc  MI02041  Waking activity σ (cts/min)  −0.78****  32  tutl  MI08144  Waking activity σ (cts/min)  −0.90****  31  Gene  Allele tested  Sleep trait  Diff. from control  No.  5HT1A  MB09812  Night avg. bout length σ (min)  14.32***  32  5HT1A  MB09978  Night avg. bout length σ (min)  10.61**  31  CG7985  MI00275  Night avg. bout length σ (min)  −113.93***  31  fru  MB02472  Night avg. bout length σ (min)  28.29*  32  fru  MB01996  Night avg. bout length σ (min)  7.09*  30  HDAC4  MI05513  Night avg. bout length σ (min)  −103.05***  32  sxc  MI02286  Night avg. bout length σ (min)  −125.68****  32  CG42260  MB00428  Night sleep σ (min)  −27.87****  31  sls  MI011177 Het.  Night sleep σ (min)  30.78****  32  rho  MI08786  Waking activity σ (cts/min)  −0.85****  32  scb  MI010788  Waking activity σ (cts/min)  −0.82****  32  sowi  MI03342 Het.  Waking activity σ (cts/min)  −0.97****  32  Tmc  MI02041  Waking activity σ (cts/min)  −0.78****  32  tutl  MI08144  Waking activity σ (cts/min)  −0.90****  31  Mutations were evaluated for their effect on the sleep σ traits predicted by the GWAS. Mean differences from the respective control line are given. The number of w1118[5905] control flies was 32; the number of y[1] w[*]; nub[2] b[1] sna[Sco] pr[1] cn[1]/CyO [3628] control flies was 27. Waking activity σ differences reflect un-transformed values; statistical tests were ln-transformed. Asterisks reflect FDRs to correct for multiple tests. *0.01 < FDR ≤ 0.05; **0.001 < FDR ≤ 0.01; ***0.0001 < FDR ≤ 0.001; ****FDR < 0.0001. FDRs = false discovery rates; Het. = heterozygote. View Large Discussion Here, we demonstrated that individual flies experience daily fluctuations in sleep characteristics. Some flies altered their sleep patterns dramatically from day to day, whereas others were remarkably stable. This intra-individual variability, which we quantified as the standard deviation (σ) of each sleep trait over time, differed among genotypes and was heritable. Importantly, sleep σ traits were partially but not completely genetically correlated with summary sleep traits (mean sleep and sleep CVE). Likewise, sleep σ traits were relatively uncorrelated with other complex traits studied in the DGRP. Daily fluctuations in sleep therefore have a distinctive genetic basis, with a partial overlap with summary sleep traits. Fourteen polymorphisms overlapped between sleep σ traits and their respective summary traits, demonstrating this partial overlap. Although further experimentation is required to understand the mechanistic basis of the predicted associations here, some of the candidate genes were previously reported to affect summary sleep phenotypes. 5-HT1A was identified as a gene regulating sleep duration and bout length in flies via signaling in the mushroom bodies [82]; here, it was implicated in average night bout length σ, and tests of flies with two different Minos insertions in 5-HT1A verified these effects. VAChT was associated with day bout number σ; mutations in this gene were previously reported to affect mean day bout number [68]. Two other genes known to affect summary sleep phenotypes, rho [83] and kirre [68], were also associated with waking activity σ and day sleep σ, respectively. We confirmed the effects of rho on waking activity σ. The associations with sleep σ traits differ somewhat from the sleep summary traits previously reported, indicating that these genes may be pleiotropic. The activity of these genes may therefore vary from day to day to produce the summary effects observed previously. Other genes we identified are plausible candidates for daily fluctuations in sleep. Interestingly, several of these genes have known effects on circadian rhythms. super sex combs was associated with average night bout length σ; this gene interacts with the GSK3β gene sgg to affect circadian rhythms [84]. We confirmed the effect of sxc on average night bout length σ. HDAC4 was also identified in the GWAS and verified for average night bout length σ. It is an intriguing candidate as it is involved in epigenetic modification and transcription [85], which might be anticipated characteristics of a gene affecting daily fluctuations in sleep. HDAC4 also functions in long-term memory [86], and mutations of this gene affect circadian rhythmicity in flies [87]. RNAi-mediated knockdown of unpaired1, which was associated with night sleep σ in this study, results in arrhythmicity [88]. minibrain was identified for day sleep σ; double-mutant flies minibrain/small optic lobes increased instability in rest/activity rhythms [89]. In addition, roughoid, like rhomboid, is a serine protease functioning in the EGFR pathway [90]; this gene was associated with waking activity σ. About 80% of the polymorphic variants that we identified were in intergenic regions. This observation is consistent with other GWA studies [91], but reveals the difficulty in fully understanding the genetic basis of complex traits such as sleep. Intergenic variants might alter gene function via changes in enhancer or DNA binding sites, which in turn may alter the transcription or translation of a gene or genes. Enhancer or promoter effects can occur far upstream or downstream of the gene they affect, however. As a first step, these variants could be studied as potential cis-regulatory genes by using genome editing to perturb each position and by examining the impact of these perturbations on transcript abundance in neighboring genes. Polymorphisms within the coding region of candidate genes are more tractable, as they can be tested directly with allelic replacement via genome editing. Overall, we identified 663 polymorphisms and 104 candidate genes for intra-individual variability in sleep. Our study has several limitations. First, the statistical support for polymorphisms associated with intra-individual variability in sleep was relatively low. This is partly due to the relatively small size of the DGRP and relatively low heritabilities, both of which make the detection of significant associations more difficult. Thus, the associations reported here probably represent a subset of the true associations, and for some of the traits such as day and night bout number σ, FDRs were high, decreasing the confidence in those predictions. However, quantile–quantile plots supported the 1 × 10−5 threshold p-value for night sleep σ, day sleep σ, average night bout length σ, and waking activity σ, indicating that there was adequate statistical power for these measures (Supplementary Figure S1). This makes sense as the effect sizes for these four traits were the largest we observed. Accordingly, our follow-up mutational tests focused on the genes with the largest predicted effect sizes. This was an effective strategy for confirming candidate genes as 12 of 14 genes had significant effects on the predicted sleep σ trait. Second, the variability in sleep patterns that we observed is confounded with age. The original study evaluated sleep in flies that were 4 to 14 days old, so it is not known how sleep patterns in these lines might change in either very young or very old flies. We observed both increasing and decreasing patterns of fragmentation in these relatively young flies. A previous longitudinal study of sleep in flies documented a monotonic trend for increased fragmentation across the entire lifespan, with some fluctuation in sleep patterns over short periods [92]. Varying trends have been observed in human sleep as well over time, from inverted U-shaped distributions [93] to declines in intra-individual variability with age [13]. It would therefore be interesting to examine daily fluctuations in sleep across the entire lifespan in this population. Finally, our study focuses on behavioral correlates for sleep and does not address any potential fluctuation in sleep intensity, which vary with genotype in flies [94] and are critical to understanding sleep need [95]. Conclusion We have demonstrated that Drosophila sleep patterns fluctuate from day to day. We defined the standard deviation as a measure of the daily fluctuations in sleep for each fly and found it to be heritable. Using all available variants in the DGRP, we mapped 104 candidate genes, some of which have known roles in sleep and circadian rhythms. We also confirmed the role of 12 genes on sleep intra-individual variability. Many of these genes have predicted homology with human genes, suggesting that the underlying genetic architecture of daily fluctuations in sleep may be conserved. Supplementary Material Supplementary material is available at SLEEP online. Acknowledgments The authors thank Y. Lin for helpful discussions. Funding This research was supported by the Intramural Research Program at the National Institutes of Health, the National Heart, Lung, and Blood Institute. Disclosure Statement None declared. References 1. Samson DR, Crittenden AN, Mabulla IA, Mabulla AZ, Nunn CL. Hadza sleep biology: evidence for flexible sleep-wake patterns in hunter-gatherers. Am J Phys Anthropol . 2017; 162( 3): 573– 582. Google Scholar CrossRef Search ADS PubMed  2. Yetish G, Kaplan H, Gurven Met al.   Natural sleep and its seasonal variations in three pre-industrial societies. Curr Biol . 2015; 25( 21): 2862– 2868. Google Scholar CrossRef Search ADS PubMed  3. Samson DR, Manus MB, Krystal AD, Fakir E, Yu JJ, Nunn CL. Segmented sleep in a nonelectric, small-scale agricultural society in Madagascar. Am J Hum Biol . 2017; 29( 4): e22979. Google Scholar CrossRef Search ADS   4. Ekirch AR. Segmented sleep in preindustrial societies. Sleep . 2016; 39( 3): 715– 716. Google Scholar CrossRef Search ADS PubMed  5. Van Dongen HP, Baynard MD, Maislin G, Dinges DF. Systematic interindividual differences in neurobehavioral impairment from sleep loss: evidence of trait-like differential vulnerability. Sleep . 2004; 27( 3): 423– 433. Google Scholar PubMed  6. Bliese PD, Wesensten NJ, Balkin TJ. Age and individual variability in performance during sleep restriction. J Sleep Res . 2006; 15( 4): 376– 385. Google Scholar CrossRef Search ADS PubMed  7. Lim J, Choo WC, Chee MW. Reproducibility of changes in behaviour and fMRI activation associated with sleep deprivation in a working memory task. Sleep . 2007; 30( 1): 61– 70. Google Scholar CrossRef Search ADS PubMed  8. Geiger A, Huber R, Kurth S, Ringli M, Jenni OG, Achermann P. The sleep EEG as a marker of intellectual ability in school age children. Sleep . 2011; 34( 2): 181– 189. Google Scholar CrossRef Search ADS PubMed  9. Knutson KL, Rathouz PJ, Yan LL, Liu K, Lauderdale DS. Intra-individual daily and yearly variability in actigraphically recorded sleep measures: the CARDIA study. Sleep . 2007; 30( 6): 793– 796. Google Scholar CrossRef Search ADS PubMed  10. Mezick EJ, Matthews KA, Hall Met al.   Intra-individual variability in sleep duration and fragmentation: associations with stress. Psychoneuroendocrinology . 2009; 34( 9): 1346– 1354. Google Scholar CrossRef Search ADS PubMed  11. Buman MP, Hekler EB, Bliwise DL, King AC. Exercise effects on night-to-night fluctuations in self-rated sleep among older adults with sleep complaints. J Sleep Res . 2011; 20( 1 Pt 1): 28– 37. Google Scholar CrossRef Search ADS PubMed  12. Angulo-Barroso RM, Peirano P, Algarin C, Kaciroti N, Lozoff B. Motor activity and intra-individual variability according to sleep-wake states in preschool-aged children with iron-deficiency anemia in infancy. Early Hum Dev . 2013; 89( 12): 1025– 1031. Google Scholar CrossRef Search ADS PubMed  13. Dillon HR, Lichstein KL, Dautovich ND, Taylor DJ, Riedel BW, Bush AJ. Variability in self-reported normal sleep across the adult age span. J Gerontol B Psychol Sci Soc Sci . 2015; 70( 1): 46– 56. Google Scholar CrossRef Search ADS PubMed  14. Gruber R, Sadeh A, Raviv A. Instability of sleep patterns in children with attention-deficit/hyperactivity disorder. J Am Acad Child Adolesc Psychiatry . 2000; 39( 4): 495– 501. Google Scholar CrossRef Search ADS PubMed  15. Shoji KD, Tighe CA, Dautovich ND, McCrae CS. Beyond mean values: Quantifying intraindividual variability in pre-sleep arousal and sleep in younger and older community-dwelling adults. Sleep Sci . 2015; 8( 1): 24– 30. Google Scholar CrossRef Search ADS PubMed  16. Bei B, Allen NB, Nicholas CL, Dudgeon P, Murray G, Trinder J. Actigraphy-assessed sleep during school and vacation periods: a naturalistic study of restricted and extended sleep opportunities in adolescents. J Sleep Res . 2014; 23( 1): 107– 117. Google Scholar CrossRef Search ADS PubMed  17. Könen T, Dirk J, Schmiedek F. Cognitive benefits of last night’s sleep: daily variations in children’s sleep behavior are related to working memory fluctuations. J Child Psychol Psychiatry . 2015; 56( 2): 171– 182. Google Scholar CrossRef Search ADS PubMed  18. Buysse DJ, Cheng Y, Germain Aet al.   Night-to-night sleep variability in older adults with and without chronic insomnia. Sleep Med . 2010; 11( 1): 56– 64. Google Scholar CrossRef Search ADS PubMed  19. Cheek RE, Shaver JL, Lentz MJ. Lifestyle practices and nocturnal sleep in midlife women with and without insomnia. Biol Res Nurs . 2004; 6( 1): 46– 58. Google Scholar CrossRef Search ADS PubMed  20. Straus LD, Drummond SP, Nappi CM, Jenkins MM, Norman SB. Sleep variability in military-related PTSD: a comparison to primary insomnia and healthy controls. J Trauma Stress . 2015; 28( 1): 8– 16. Google Scholar CrossRef Search ADS PubMed  21. Prasad B, Usmani S, Steffen ADet al.   Short-term variability in apnea-hypopnea index during extended home portable monitoring. J Clin Sleep Med . 2016; 12( 6): 855– 863. Google Scholar CrossRef Search ADS PubMed  22. Sforza E, Haba-Rubio J. Night-to-night variability in periodic leg movements in patients with restless legs syndrome. Sleep Med . 2005; 6( 3): 259– 267. Google Scholar CrossRef Search ADS PubMed  23. Bei B, Wiley JF, Trinder J, Manber R. Beyond the mean: A systematic review on the correlates of daily intraindividual variability of sleep/wake patterns. Sleep Med Rev . 2016; 28: 108– 124. Google Scholar CrossRef Search ADS PubMed  24. Pagani L, St Clair PA, Teshiba TMet al.   Genetic contributions to circadian activity rhythm and sleep pattern phenotypes in pedigrees segregating for severe bipolar disorder. Proc Natl Acad Sci U S A . 2016; 113( 6): E754– E761. Google Scholar CrossRef Search ADS PubMed  25. Scott J, Murray G, Henry Cet al.   Activation in bipolar disorders: a systematic review. JAMA Psychiatry . 2017; 74( 2): 189– 196. Google Scholar CrossRef Search ADS PubMed  26. Bei B, Seeman TE, Carroll JE, Wiley JF. Sleep and physiological dysregulation: a closer look at sleep intraindividual variability. Sleep . 2017; 40( 9): zsx109. doi: 10.1093/sleep/zsx109. Google Scholar CrossRef Search ADS   27. Okun ML, Reynolds CF3rd, Buysse DJet al.   Sleep variability, health-related practices, and inflammatory markers in a community dwelling sample of older adults. Psychosom Med . 2011; 73( 2): 142– 150. Google Scholar CrossRef Search ADS PubMed  28. He F, Bixler EO, Liao Jet al.   Habitual sleep variability, mediated by nutrition intake, is associated with abdominal obesity in adolescents. Sleep Med . 2015; 16( 12): 1489– 1494. Google Scholar CrossRef Search ADS PubMed  29. Gander P, van den Berg M, Signal L. Sleep and sleepiness of fishermen on rotating schedules. Chronobiol Int . 2008; 25( 2): 389– 398. Google Scholar CrossRef Search ADS PubMed  30. Stalder T, Evans P, Hucklebridge F, Clow A. State associations with the cortisol awakening response in healthy females. Psychoneuroendocrinology . 2010; 35( 8): 1245– 1252. Google Scholar CrossRef Search ADS PubMed  31. Callahan HS, Pigliucci M, Schlichting CD. Developmental phenotypic plasticity: where ecology and evolution meet molecular biology. Bioessays . 1997; 19( 6): 519– 525. Google Scholar CrossRef Search ADS PubMed  32. Spada J, Scholz M, Kirsten Het al.   Genome-wide association analysis of actigraphic sleep phenotypes in the LIFE Adult Study. J Sleep Res . 2016; 25( 6): 690– 701. Google Scholar CrossRef Search ADS PubMed  33. Hill WG, Mulder HA. Genetic analysis of environmental variation. Genet Res (Camb) . 2010; 92( 5-6): 381– 395. Google Scholar CrossRef Search ADS PubMed  34. Hall MC, Dworkin I, Ungerer MC, Purugganan M. Genetics of microenvironmental canalization in Arabidopsis thaliana. Proc Natl Acad Sci U S A . 2007; 104( 34): 13717– 13722. Google Scholar CrossRef Search ADS PubMed  35. Sangster TA, Salathia N, Undurraga Set al.   HSP90 affects the expression of genetic variation and developmental stability in quantitative traits. Proc Natl Acad Sci U S A . 2008; 105( 8): 2963– 2968. Google Scholar CrossRef Search ADS PubMed  36. Ordas B, Malvar RA, Hill WG. Genetic variation and quantitative trait loci associated with developmental stability and the environmental correlation between traits in maize. Genet Res (Camb) . 2008; 90( 5): 385– 395. Google Scholar CrossRef Search ADS PubMed  37. Ansel J, Bottin H, Rodriguez-Beltran Cet al.   Cell-to-cell stochastic variation in gene expression is a complex genetic trait. PLoS Genet . 2008; 4( 4): e1000049. Google Scholar CrossRef Search ADS PubMed  38. Blake WJ, Balázsi G, Kohanski MAet al.   Phenotypic consequences of promoter-mediated transcriptional noise. Mol Cell . 2006; 24( 6): 853– 865. Google Scholar CrossRef Search ADS PubMed  39. Lin Y, Chen ZX, Oliver B, Harbison ST. Microenvironmental gene expression plasticity among individual Drosophila melanogaster. G3 (Bethesda) . 2016; 6( 12): 4197– 4210. Google Scholar PubMed  40. Whitlock MC, Fowler K. The changes in genetic and environmental variance with inbreeding in Drosophila melanogaster. Genetics . 1999; 152( 1): 345– 353. Google Scholar PubMed  41. Mackay TF, Lyman RF. Drosophila bristles and the nature of quantitative genetic variation. Philos Trans R Soc Lond B Biol Sci . 2005; 360( 1459): 1513– 1527. Google Scholar CrossRef Search ADS PubMed  42. Garlapow ME, Huang W, Yarboro MT, Peterson KR, Mackay TF. Quantitative genetics of food intake in Drosophila melanogaster. PLoS One . 2015; 10( 9): e0138129. Google Scholar CrossRef Search ADS PubMed  43. Ayroles JF, Buchanan SM, O’Leary Cet al.   Behavioral idiosyncrasy reveals genetic control of phenotypic variability. Proc Natl Acad Sci U S A . 2015; 112( 21): 6706– 6711. Google Scholar CrossRef Search ADS PubMed  44. Morgante F, Sørensen P, Sorensen DA, Maltecca C, Mackay TF. Genetic architecture of micro-environmental plasticity in Drosophila melanogaster. Sci Rep . 2015; 5: 9785 doi:10.1038/srep09785. Google Scholar CrossRef Search ADS PubMed  45. Harbison ST, McCoy LJ, Mackay TF. Genome-wide association study of sleep in Drosophila melanogaster. BMC Genomics . 2013; 14: 281 doi:10.1186/ 1471-2164-281. Google Scholar CrossRef Search ADS PubMed  46. Huang W, Massouras A, Inoue Yet al.   Natural variation in genome architecture among 205 Drosophila melanogaster Genetic Reference Panel lines. Genome Res . 2014; 24( 7): 1193– 1208. Google Scholar CrossRef Search ADS PubMed  47. Mackay TF, Richards S, Stone EAet al.   The Drosophila melanogaster Genetic Reference Panel. Nature . 2012; 482( 7384): 173– 178. Google Scholar CrossRef Search ADS PubMed  48. Isaac RE, Li C, Leedale AE, Shirras AD. Drosophila male sex peptide inhibits siesta sleep and promotes locomotor activity in the post-mated female. Proc Biol Sci . 2010; 277( 1678): 65– 70. Google Scholar CrossRef Search ADS PubMed  49. Ganguly-Fitzgerald I, Donlea J, Shaw PJ. Waking experience affects sleep need in Drosophila. Science . 2006; 313( 5794): 1775– 1781. Google Scholar CrossRef Search ADS PubMed  50. Falconer DS, Mackay TF. Introduction to Quantitative Genetics . 4th ed. Edinburgh Gate, Harlow: Addison Wesley Longman Limited; 1996. 51. Chow CY, Kelsey KJ, Wolfner MF, Clark AG. Candidate genetic modifiers of retinitis pigmentosa identified by exploiting natural variation in Drosophila. Hum Mol Genet . 2016; 25( 4): 651– 659. Google Scholar CrossRef Search ADS PubMed  52. Dembeck LM, Huang W, Magwire MM, Lawrence F, Lyman RF, Mackay TF. Genetic architecture of abdominal pigmentation in Drosophila melanogaster. PLoS Genet . 2015; 11( 5): e1005163. Google Scholar CrossRef Search ADS PubMed  53. Vonesch SC, Lamparter D, Mackay TF, Bergmann S, Hafen E. Genome-wide analysis reveals novel regulators of growth in Drosophila melanogaster. PLoS Genet . 2016; 12( 1): e1005616. Google Scholar CrossRef Search ADS PubMed  54. Unckless RL, Rottschaefer SM, Lazzaro BP. A genome-wide association study for nutritional indices in Drosophila. G3 (Bethesda) . 2015; 5( 3): 417– 425. Google Scholar CrossRef Search ADS PubMed  55. Ivanov DK, Escott-Price V, Ziehm Met al.   Longevity GWAS using the Drosophila Genetic Reference Panel. J Gerontol A Biol Sci Med Sci . 2015; 70( 12): 1470– 1478. Google Scholar CrossRef Search ADS PubMed  56. Morozova TV, Huang W, Pray VA, Whitham T, Anholt RR, Mackay TF. Polymorphisms in early neurodevelopmental genes affect natural variation in alcohol sensitivity in adult drosophila. BMC Genomics . 2015; 16: 865. Google Scholar CrossRef Search ADS PubMed  57. Arya GH, Magwire MM, Huang W, Serrano-Negron YL, Mackay TF, Anholt RR. The genetic basis for variation in olfactory behavior in Drosophila melanogaster. Chem Senses . 2015; 40( 4): 233– 243. Google Scholar CrossRef Search ADS PubMed  58. He X, Zhou S, St Armour GE, Mackay TF, Anholt RR. Epistatic partners of neurogenic genes modulate Drosophila olfactory behavior. Genes Brain Behav . 2016; 15( 2): 280– 290. Google Scholar CrossRef Search ADS PubMed  59. Gaertner BE, Ruedi EA, McCoy LJ, Moore JM, Wolfner MF, Mackay TF. Heritable variation in courtship patterns in Drosophila melanogaster. G3 (Bethesda) . 2015; 5( 4): 531– 539. Google Scholar CrossRef Search ADS PubMed  60. Chow CY, Wolfner MF, Clark AG. Large neurological component to genetic differences underlying biased sperm use in Drosophila. Genetics . 2013; 193( 1): 177– 185. Google Scholar CrossRef Search ADS PubMed  61. Jordan KW, Craver KL, Magwire MM, Cubilla CE, Mackay TF, Anholt RR. Genome-wide association for sensitivity to chronic oxidative stress in Drosophila melanogaster. PLoS One . 2012; 7( 6): e38722. Google Scholar CrossRef Search ADS PubMed  62. Montgomery SL, Vorojeikina D, Huang W, Mackay TF, Anholt RR, Rand MD. Genome-wide association analysis of tolerance to methylmercury toxicity in Drosophila implicates myogenic and neuromuscular developmental pathways. PLoS One . 2014; 9( 10): e110375. Google Scholar CrossRef Search ADS PubMed  63. Vaisnav M, Xing C, Ku HCet al.   Genome-wide association analysis of radiation resistance in Drosophila melanogaster. PLoS One . 2014; 9( 8): e104858. Google Scholar CrossRef Search ADS PubMed  64. Weber AL, Khan GF, Magwire MM, Tabor CL, Mackay TF, Anholt RR. Genome-wide association analysis of oxidative stress resistance in Drosophila melanogaster. PLoS One . 2012; 7( 4): e34745. Google Scholar CrossRef Search ADS PubMed  65. Ellis LL, Huang W, Quinn AMet al.   Intrapopulation genome size variation in D. melanogaster reflects life history variation and plasticity. PLoS Genet . 2014; 10( 7): e1004522. Google Scholar CrossRef Search ADS PubMed  66. Dembeck LM, Boroczky K, Huang W, Schal C, Anholt RR, Mackay TF. Genetic architecture of natural variation in cuticular hydrocarbon composition in Drosophila melanogaster. Elife . 2015; 4: e09861. Google Scholar CrossRef Search ADS PubMed  67. Hunter CM, Huang W, Mackay TF, Singh ND. The genetic architecture of natural variation in recombination rate in Drosophila melanogaster. PLoS Genet . 2016; 12( 4): e1005951. Google Scholar CrossRef Search ADS PubMed  68. Lobell AS, Kaspari RR, Serrano Negron YL, Harbison ST. The genetic architecture of ovariole number in Drosophila melanogaster: genes with major, quantitative, and pleiotropic effects. G3 (Bethesda) . 2017; 7( 7): 2391– 2403. Google Scholar CrossRef Search ADS PubMed  69. Shorter J, Couch C, Huang Wet al.   Genetic architecture of natural variation in Drosophila melanogaster aggressive behavior. Proc Natl Acad Sci U S A . 2015; 112( 27): E3555– E3563. Google Scholar CrossRef Search ADS PubMed  70. Zwarts L, Vanden Broeck L, Cappuyns Eet al.   The genetic basis of natural variation in mushroom body size in Drosophila melanogaster. Nat Commun . 2015; 6: 10115 doi:10.1038/ncomms10115. Google Scholar CrossRef Search ADS PubMed  71. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Statist Soc B . 1995; 57( 1): 289– 300. 72. Purcell S, Neale B, Todd-Brown Ket al.   PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet . 2007; 81( 3): 559– 575. Google Scholar CrossRef Search ADS PubMed  73. Hu Y, Flockhart I, Vinayagam Aet al.   An integrative approach to ortholog prediction for disease-focused and other functional studies. BMC Bioinformatics . 2011; 12: 357 doi: 10.1186/1471-2105-12-357. Google Scholar CrossRef Search ADS PubMed  74. Isaac RE, Li C, Leedale AE, Shirras AD. Drosophila male sex peptide inhibits siesta sleep and promotes locomotor activity in the post-mated female. Proc Biol Sci . 2010; 277( 1678): 65– 70. Google Scholar CrossRef Search ADS PubMed  75. Huber R, Hill SL, Holladay C, Biesiadecki M, Tononi G, Cirelli C. Sleep homeostasis in Drosophila melanogaster. Sleep . 2004; 27( 4): 628– 639. Google Scholar CrossRef Search ADS PubMed  76. Bellen HJ, Levis RW, He Yet al.   The Drosophila gene disruption project: progress using transposons with distinctive site specificities. Genetics . 2011; 188( 3): 731– 743. Google Scholar CrossRef Search ADS PubMed  77. Venken KJ, Schulze KL, Haelterman NAet al.   MiMIC: a highly versatile transposon insertion resource for engineering Drosophila melanogaster genes. Nat Methods . 2011; 8( 9): 737– 743. Google Scholar CrossRef Search ADS PubMed  78. Cirelli C, Bushey D, Hill Set al.   Reduced sleep in Drosophila Shaker mutants. Nature . 2005; 434( 7037): 1087– 1092. Google Scholar CrossRef Search ADS PubMed  79. Vienne J, Spann R, Guo F, Rosbash M. Age-related reduction of recovery sleep and arousal threshold in Drosophila. Sleep . 2016; 39( 8): 1613– 1624. Google Scholar CrossRef Search ADS PubMed  80. Harbison ST, Sehgal A. Quantitative genetic analysis of sleep in Drosophila melanogaster. Genetics . 2008; 178( 4): 2341– 2360. Google Scholar CrossRef Search ADS PubMed  81. Mackay TFC, Huang W. Charting the genotype-phenotype map: lessons from the Drosophila melanogaster Genetic Reference Panel. Wiley Interdiscip Rev Dev Biol . 2017 doi:10.1002/wdev289. 82. Yuan Q, Joiner WJ, Sehgal A. A sleep-promoting role for the Drosophila serotonin receptor 1A. Curr Biol . 2006; 16( 11): 1051– 1062. Google Scholar CrossRef Search ADS PubMed  83. Foltenyi K, Greenspan RJ, Newport JW. Activation of EGFR and ERK by rhomboid signaling regulates the consolidation and maintenance of sleep in Drosophila. Nat Neurosci . 2007; 10( 9): 1160– 1167. Google Scholar CrossRef Search ADS PubMed  84. Kaasik K, Kivimäe S, Allen JJet al.   Glucose sensor O-GlcNAcylation coordinates with phosphorylation to regulate circadian clock. Cell Metab . 2013; 17( 2): 291– 302. Google Scholar CrossRef Search ADS PubMed  85. Cho Y, Griswold A, Campbell C, Min KT. Individual histone deacetylases in Drosophila modulate transcription of distinct genes. Genomics . 2005; 86( 5): 606– 617. Google Scholar CrossRef Search ADS PubMed  86. Fitzsimons HL, Schwartz S, Given FM, Scott MJ. The histone deacetylase HDAC4 regulates long-term memory in Drosophila. PLoS One . 2013; 8( 12): e83903. Google Scholar CrossRef Search ADS PubMed  87. Fogg PC, O’Neill JS, Dobrzycki Tet al.   Class IIa histone deacetylases are conserved regulators of circadian function. J Biol Chem . 2014; 289( 49): 34341– 34348. Google Scholar CrossRef Search ADS PubMed  88. Luo W, Sehgal A. Regulation of circadian behavioral output via a MicroRNA-JAK/STAT circuit. Cell . 2012; 148( 4): 765– 779. Google Scholar CrossRef Search ADS PubMed  89. Helfrich C. Role of the optic lobes in the regulation of the locomotor activity rhythm of Drosophila melanogaster: behavioral analysis of neural mutants. J Neurogenet . 1986; 3( 6): 321– 343. Google Scholar CrossRef Search ADS PubMed  90. Shilo BZ. Regulating the dynamics of EGF receptor signaling in space and time. Development . 2005; 132( 18): 4017– 4027. Google Scholar CrossRef Search ADS PubMed  91. Hindorff LA, Sethupathy P, Junkins HAet al.   Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc Natl Acad Sci U S A . 2009; 106( 23): 9362– 9367. Google Scholar CrossRef Search ADS PubMed  92. Koh K, Evans JM, Hendricks JC, Sehgal A. A Drosophila model for age-associated changes in sleep:wake cycles. Proc Natl Acad Sci U S A . 2006; 103( 37): 13843– 13847. Google Scholar CrossRef Search ADS PubMed  93. Becker SP, Sidol CA, Van Dyk TR, Epstein JN, Beebe DW. Intraindividual variability of sleep/wake patterns in relation to child and adolescent functioning: A systematic review. Sleep Med Rev . 2017; 34: 94– 121. Google Scholar CrossRef Search ADS PubMed  94. van Alphen B, Yap MH, Kirszenblat L, Kottler B, van Swinderen B. A dynamic deep sleep stage in Drosophila. J Neurosci . 2013; 33( 16): 6917– 6927. Google Scholar CrossRef Search ADS PubMed  95. Aulsebrook AE, Jones TM, Rattenborg NC, Roth TC2nd, Lesku JA. Sleep ecophysiology: integrating neuroscience and ecology. Trends Ecol Evol . 2016; 31( 8): 590– 599. Google Scholar CrossRef Search ADS PubMed  Published by Oxford University Press on behalf of Sleep Research Society (SRS) 2017. This work is written by (a) US Government employee(s) and is in the public domain in the US.

Journal

SLEEPOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off