Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. 2006 Sep 12; 103(37): 13843–13847.
Published online 2006 Aug 28. doi:  10.1073/pnas.0605903103
PMCID: PMC1564207
From the Cover

A Drosophila model for age-associated changes in sleep:wake cycles


One of the most consistent behavioral changes that occurs with age in humans is the loss of sleep consolidation. This can be quite disruptive and yet little is known about its underlying basis. To better understand the effects of aging on sleep:wake cycles, we sought to study this problem in Drosophila melanogaster, a powerful system for research on aging and behavior. By assaying flies of different ages as well as monitoring individual flies constantly over the course of their lifetime, we found that the strength of sleep:wake cycles decreased and that sleep became more fragmented with age in Drosophila. These changes in sleep:wake cycles became faster or slower with manipulations of ambient temperature that decreased or increased lifespan, respectively, demonstrating that they are a function of physiological rather than chronological age. The effect of temperature on lifespan was not mediated by changes in overall activity level or sleep amount. Flies treated with the oxidative stress-producing reagent paraquat showed a breakdown of sleep:wake cycles similar to that seen with aging, leading us to propose that the accumulation of oxidative damage with age contributes to the changes in rhythm and sleep. Together, these findings establish Drosophila as a valuable model for studying age-associated sleep fragmentation and breakdown of rhythm strength, and indicate that these changes in sleep:wake cycles are an integral part of the physiological aging process.

Keywords: aging, circadian rhythms, sleep fragmentation

Many physiological functions deteriorate with age. One of the most striking is the loss of sleep consolidation, namely increased daytime sleep and increased nighttime wakefulness in the elderly (1). In fact, loss of sleep consolidation has been used as one of several measures of frailty in elderly people (2). As a result, there is increasing awareness that treatment of their sleep problems could significantly improve quality of life for older individuals.

Drosophila is well established as a model for the study of circadian rhythms. Molecular mechanisms of the Drosophila circadian clock are conserved in mammals. In fact, a human circadian disorder, familial advanced sleep phase syndrome, has been linked to mutations in clock genes that were first discovered in flies (37). In recent years, Drosophila has also emerged as a useful system to study sleep (8, 9). Since sleep was first described in flies, multiple approaches have led to the identification of molecules involved in regulation of sleep (1014), some of which are associated with sleep phenotypes in mammals (15). In addition, because of its relatively short lifespan and its amenability to genetic approaches, Drosophila is becoming increasingly popular as a model for aging (16). As in the case of circadian rhythms and sleep, pathways that affect aging in flies have conserved functions in mammals (1719). Among other pathways, those involved in the response to oxidative stress have been implicated in aging in all organisms tested (20). Molecular interventions that down-regulate oxidation generally increase lifespan in diverse organisms including worms, flies, and mammals (2125). Metabolic regulators, such as insulin-like growth factor 1 (IGF1), that affect lifespan may also do so through effects on the redox activity in cells (22).

Here, we report that flies undergo an age-associated breakdown of sleep:wake cycles reminiscent of that seen in humans. Sleep becomes fragmented with age, and the rate at which this fragmentation occurs depends on ambient temperature and lifespan, suggesting that it is an integral part of the physiological aging process. Because an increase in oxidative stress disrupts sleep:wake cycles in a manner similar to that of aging, we propose that the accumulation of oxidative damage with age, at least in part, accounts for the breakdown of sleep:wake cycles.


The Strength of the Sleep:Wake Cycle Decreases with Age.

To identify changes that occur over the course of a lifetime, we monitored the sleep:wake behavior of individual flies in activity tubes continuously from a few hours after eclosion until their death. Under these monitoring conditions, wild-type flies lived for >2 months on average at 25°C. This is comparable with the lifespan reported in other longevity studies (12, 2428), although it is less than what we see when we maintain groups of flies in vials (see Fig. 5A, which is published as supporting information on the PNAS web site).

Importantly, the longitudinal approach provided invaluable information about changes in sleep:wake patterns of the same flies as they aged. We found that whereas flies had long, uninterrupted bouts of sleep when they were young, sleep became fragmented as they aged (Fig. 1A). Although there were some differences in sleep patterns between male and female flies, sleep became more evenly distributed over the 24-h period with age for both males and females (Fig. 1B). Young female flies slept mostly at night, and as they aged, daytime sleep increased and nighttime sleep decreased. Young male flies, on the other hand, had two peaks of activity at dawn and dusk, and slept almost as much in the middle of the day as at night. As male flies aged, they slept more at dawn and dusk, and slept less at midday and night.

Fig. 1.
Sleep:wake cycles become weaker and sleep becomes fragmented with age: longitudinal study. (A) Activity records of representative female and male flies when they were <10 days old (Young) and when they were >2 months old (Old). The records ...

Rhythm strength, as measured by relative fast Fourier transform (FFT) values, steadily decreased with age (Fig. 1C). In addition, a number of sleep parameters changed with age. The average duration of sleep bouts decreased relative to their peak values, and the number of sleep bouts increased, both during the day and at night (Figs. 1C and 5B). We also saw an increase in the number of brief awakenings (Fig. 1C). However, of the sleep parameters examined, daily sleep bout number showed the clearest change and was least variable. Changes in these sleep parameters indicate that sleep becomes fragemented in aging flies as it does in aging humans (1). In contrast, unlike in humans, total sleep amount did not decrease with age; there was a moderate increase in sleep amount and a corresponding decrease in activity counts in aging females, whereas these measures changed little after the first 3 weeks in males (Fig. 5C).

Age-associated decline in rhythm strength occurred in the presence of light:dark cycles, suggesting that it did not just result from loss of circadian clock function; clock mutants can be driven by light:dark cycles to show rhythmic behavior (29, 30). Indeed, the timeless (tim) clock mutant driven by a light:dark cycle showed a steady decline in rhythm strength with age (see Fig. 6, which is published as supporting information on the PNAS web site). tim flies had relatively poorly consolidated sleep compared with wild-type flies, namely they had more sleep bouts and a shorter sleep bout duration (compare Figs. 1C and 6) and did not show a substantial increase in sleep fragmentation with age. Although sleep bout numbers increased somewhat with age in tim flies, the average duration of sleep bouts did not decrease. These results suggest that the circadian clock promotes consolidation of sleep and that the age-associated decline in rhythm strength has multiple causes, including disruption of circadian clock function and of noncircadian mechanisms that control sleep.

To validate the data obtained through the longitudinal studies, we conducted conventional cross-sectional studies in which sleep:wake behavior was assayed in flies of different ages. Groups of flies that had been aged for different lengths of time were collected and monitored for sleep:wake cycles. Although the flies were somewhat longer-lived in cross-sectional assays (Fig. 5A), the data generally mirrored those obtained from longitudinal studies. Thus, rhythm strength and the average duration of sleep bouts declined over the lifetime of the flies, whereas the number of sleep bouts increased (see Fig. 7, which is published as supporting information on the PNAS web site).

Ambient Temperature Affects the Rate of Breakdown in Sleep:Wake Cycles.

To determine whether the breakdown of sleep:wake cycles reported above is linked to physiological or chronological age, we assayed sleep:wake cycles under environmental conditions that affect lifespan. An increase or decrease in ambient temperature is known to decrease or increase lifespan in Drosophila (31). We monitored the behavior of flies maintained at 29°C, 25°C, or 18°C. The temperature manipulation had a large effect on lifespan. Flies kept at 18°C lived ≈2.4 times longer than those kept at 25°C, which, in turn, lived ≈1.7 times longer than those kept at 29°C. The experiment was performed using two different stocks of wild-type Canton-S flies. These stocks have been maintained in different labs for many years, and the males differ considerably in average lifespan (e.g., median lifespan at 18°C was 167 and 115 days, respectively), most likely because of genetic drift. Despite the differences in lifespan in the two stocks, a consistent pattern of age-associated changes in sleep:wake cycles emerged. At lower ambient temperatures, flies lived longer and the strength of sleep:wake rhythms declined at a slower rate (Fig. 2A; and see Fig. 8, which is published as supporting information on the PNAS web site). Similarly, age-associated sleep fragmentation, as measured by the number of sleep bouts and sleep bout length, occurred at a slower rate at lower temperatures (Fig. 2 B and C and Fig. 8). This finding supports the idea that the strength of sleep:wake cycles and sleep consolidation change as a function of physiological, rather than chronological, age.

Fig. 2.
Decline of rhythm strength and fragmentation of sleep is slower at 18°C and faster at 29°C. Rhythm strength (A), number of sleep bouts (B), and average duration of sleep bouts (C) at 18°C (blue), 25°C (green), and 29°C ...

The Effect of Ambient Temperature on Lifespan Is Not Mediated by Changes in Overall Activity.

Mutants with increased activity or decreased sleep tend to have shortened lifespan (10, 32), although there are exceptions to this general trend (13). One explanation might be that a higher metabolic rate at a higher temperature leads to increased overall activity and reduced sleep, which may contribute to short lifespan. Contrary to this hypothesis, we did not find a simple monotonic relationship between longevity and activity level (Fig. 3A). Notably, flies maintained at 18°C lived approximately four times longer than those at 29°C, but there was little difference in activity levels. The effect of temperature on overall sleep amount followed a similar pattern (Fig. 3B). Thus, the effect of temperature on lifespan cannot be accounted for by changes in activity level or sleep amount. It is, however, closely linked to changes in the rate at which sleep fragmentation and decline in rhythm strength occur.

Fig. 3.
The effect of ambient temperature on lifespan is not mediated by changes in locomotor activity or amount of sleep. (A) Total activity counts were not monotonically related to lifespan. For males (Right), there was little difference in activity levels ...

Oxidative Stress Effects a Breakdown of Sleep:Wake Cycles.

Aging is associated with an accumulation of oxidative damage (20). In the course of other experiments in the laboratory, we found that young flies lacking the FOXO protein, which protects cells from oxidative stress in mammalian tissue culture (33), showed rhythm phenotypes. In particular, foxo mutant flies were unable to sustain sleep:wake rhythms in the presence of the oxidative stress-producing agent paraquat (X. Zheng and A.S., unpublished observations). These data suggested that an increase in oxidative damage can cause deterioration of sleep:wake cycles. To determine whether paraquat can produce similar effects in wild-type flies, we maintained flies on food containing 1 mM paraquat throughout life, while constantly monitoring their behavior. At this low dosage, wild-type flies could live on average for ≈1 month, which is approximately half of their normal lifespan. For the first 3 weeks of treatment, paraquat had little effect on sleep:wake cycles. At ≈3 weeks of age, however, compared with control flies, flies treated with paraquat showed a faster rate of decrease in rhythm strength and average duration of sleep bouts and a faster rate of increase in sleep bout numbers (Fig. 4; and see Fig. 9A, which is published as supporting information on the PNAS web site). Paraquat treatment also promoted faster changes in overall activity and total sleep amount in females (Fig. 9B). As mentioned above, these measures showed little change in males after ≈3 weeks and were relatively unaffected by paraquat treatment. These results resemble those found in flies whose aging process was accelerated with high ambient temperature (Fig. 2) and suggest that the effects of increased oxidative stress on sleep:wake cycles are similar to those that occur with aging.

Fig. 4.
Sleep:wake cycles become weaker and sleep becomes fragmented with paraquat treatment. Wild-type female (Left) and male (Right) flies were maintained with (black) or without (gray) 1 mM paraquat throughout life. Paraquat-treated flies had decreased lifespan ...


We report here that flies undergo age-associated changes in rhythm strength and sleep consolidation as do humans. The rate of these changes could be altered by manipulations of ambient temperature that alter lifespan, indicating that they are a function of physiological rather than chronological age. Lowered ambient temperature is thought to extend lifespan by lowering the rate of aging (18). Our results indicate that lowered temperature also lowers the rate of sleep fragmentation and breakdown of sleep:wake cycles, suggesting a close connection between the aging process and changes in sleep:wake behavior. At this point, we do not know the nature of the connection. It is tempting to speculate that the causal link is bidirectional. On the one hand, physiological changes that occur with age may cause cumulative damage to the mechanism regulating sleep:wake cycles. On the other hand, resulting sleep fragmentation may contribute to and accelerate the aging process. If so, interventions that increase sleep consolidation may slow down the aging process and increase lifespan.

Whereas the relationship between sleep fragmentation and lifespan has not been explored extensively, several studies show a relationship between sleep deprivation and lifespan. Rats die after a few weeks of sleep deprivation, supporting the idea that sleep is essential for life (34). Likewise, cycle (cyc) mutant flies, which have a defect in the homeostatic control of sleep, die after 12 h of sleep deprivation (14). Although wild-type flies are still alive under these conditions, further prolonged sleep deprivation is lethal for them as well. It has recently been shown that mutant alleles of the voltage-gated sodium channel gene Shaker have a short sleep phenotype and they are also short-lived (10), further strengthening the connection between sleep and lifespan. Given these results, it seemed reasonable to hypothesize that the effect of ambient temperature on lifespan is mediated by its effect on sleep amount and/or activity level. However, we found little evidence to support the hypothesis. We found that the rate of decline in sleep:wake cycles, not the total amount of sleep or activity, was monotonically related to lifespan. Although there may be a close relationship between sleep (or activity) and aging, our results implicate other factors that may modulate the relationship.

The effect of paraquat on sleep cycles suggests that an increase in oxidative stress can diminish the strength of the sleep:wake cycle. Because oxidative damage is known to build up with age, it is quite likely that it contributes to, if not accounts for, age-associated sleep fragmentation. Interestingly, both paraquat and age also affected the molecular cycling of clock gene expression in peripheral tissues, further suggesting that the mechanism involved may be the same (X. Zheng and A.S., unpublished observations). However, reversing age-related changes in sleep may not be a simple matter of treating with antioxidants. Even the biological enzymes known to have strong antioxidant activity have relatively minor effects on lifespan, indicating that they have limited potential for preventing age-associated damage (2125). Nevertheless, it is important to identify the mechanisms underlying the changes in physiology with age, with a view to developing treatments in the future.

Although we focused primarily on the strength of the sleep:wake cycle and upon sleep fragmentation, other aspects of sleep and circadian rhythms are likely also affected by age. In the suprachiasmatic nucleus, the master circadian clock in mammals, the amplitude of the firing rhythm in individual suprachiasmatic nucleus neurons decreases with age and the circadian waveform also becomes more variable (35). In addition, with age, the response of the clock to photic and nonphotic stimuli is reduced, suggesting that it is compromised in its ability to synchronize to the environment (36, 37). Clocks in other parts of the body (so-called peripheral clocks) are also affected with age, some more so than others, which may be indicative of a loss of synchrony between the suprachiasmatic nucleus and these peripheral oscillators (38). Future studies of such cellular and molecular effects in Drosophila will help in identifying the precise mechanisms that underlie the behavioral changes reported here.

Drosophila has long been used for studies of behavior and of aging. Our data here indicate that it is a powerful model to study age-related changes in behavior. We report robust changes in rhythm strength and sleep fragmentation similar to those that occur in humans. We have developed a method to monitor individual flies throughout life. This longitudinal assay is expected to facilitate many future approaches to the study of sleep and aging; for instance, experiments that involve molecular or pharmacological intervention may yield more precise results because baseline data and long-term effects can be determined for the same individual.

Materials and Methods

Behavioral Assay.

Two wild-type Canton-S fly stocks and tim0 mutants were maintained in vials containing molasses, cornmeal, and yeast media in a 12 h:12 h light:dark cycle at 25°C. For longitudinal studies, individual virgin flies were transferred to monitor tubes containing the same media on the day of eclosion. Their locomotor activity was monitored by using the Drosophila Activity Monitoring System (Trikinetics, Waltham, MA) at 18°C, 25°C, or 29°C, as indicated in the figure legends, continuously until at least half of them died. The flies were transferred to fresh tubes once a week. For cross-sectional studies, we maintained groups of 30 virgin male or female flies in vials, transferring them to fresh vials every 2 days. Flies that had been aged for different lengths of time were then transferred to monitor tubes and assayed for locomotor behavior.

Quantitative Analysis of Data.

Activity counts were collected in 5-min bins. As previously established (9, 39), a 5-min bin with no activity was considered to be sleep. A brief awakening was defined as a 5-min period of four or fewer counts, preceded and succeeded by at least 5 min of no activity. Circadian FFT values were computed for each day by using a moving window of a 6-day period. Because many flies, especially males, had strong morning and evening peaks of activity separated by 12 h, we used the sum of 12- and 24-h FFT values as a measure of circadian rhythm strength. To control for potentially confounding effects of overall activity level, circadian FFT values were normalized to noncircadian values. Each measure was computed separately for each fly and each day, and the mean and SEM for each group were calculated for each day. To examine the rate at which each measure changed over time, a linear regression line was fitted to individual data between days 10 and the median lifespan. For each group of flies, only flies that lived longer than the median lifespan were included in subsequent analyses. The number of flies in each condition included in the data analyses was as follows. For the longitudinal study, the numbers were 47 (25°C), 39 (29°C), and 32 (18°C) CS1 females; 46 (25°C), 45 (29°C), and 31 (18°C) CS1 males; 34 (25°C), 25 (29°C), and 34 (18°C) CS2 females; and 30 (25°C), 29 (29°C), and 30 (18°C) CS2 males. For the cross-sectional study, each data point represents data from at least 13 flies, except for the last data points for males, in which only 8–10 flies are represented because of high mortality. For comparison of paraquat-treated and control flies, the numbers were 33, 31, 32, and 29 for female control, female paraquat, male control, and male paraquat, respectively. The number of flies at the start of each experiment was approximately twice the number included in the analyses.

Supplementary Material

Supporting Figures:


We thank S. Helfand for Canton-S stock; S. Harbison, W. Joiner, M. Wu, and Q. Yuan for critical reading of the manuscript; and other members of the laboratory for helpful discussions. This work was supported in part by National Institutes of Health Grant AG017628 (to J.C.H. and A.S.) and a University of Pennsylvania Institute of Aging pilot grant (to A.S). A.S. is an Investigator of the Howard Hughes Medical Institute.


FFTfast Fourier transform.


Conflict of interest statement: No conflicts declared.


1. Pandi-Perumal SR, Seils LK, Kayumov L, Ralph MR, Lowe A, Moller H, Swaab DF. Ageing Res Rev. 2002;1:559–604. [PubMed]
2. Mitnitski AB, Graham JE, Mogilner AJ, Rockwood K. BMC Geriatr. 2002;2:1. [PMC free article] [PubMed]
3. Konopka RJ, Benzer S. Proc Natl Acad Sci USA. 1971;68:2112–2116. [PMC free article] [PubMed]
4. Price JL, Blau J, Rothenfluh A, Abodeely M, Kloss B, Young MW. Cell. 1998;94:83–95. [PubMed]
5. Wang GK, Sehgal A. Curr Opin Neurobiol. 2002;12:331–338. [PubMed]
6. Xu Y, Padiath QS, Shapiro RE, Jones CR, Wu SC, Saigoh N, Saigoh K, Ptacek LJ, Fu YH. Nature. 2005;434:640–644. [PubMed]
7. Toh KL, Jones CR, He Y, Eide EJ, Hinz WA, Virshup DM, Ptacek LJ, Fu YH. Science. 2001;291:1040–1043. [PubMed]
8. Hendricks JC, Finn SM, Panckeri KA, Chavkin J, Williams JA, Sehgal A, Pack AI. Neuron. 2000;25:129–138. [PubMed]
9. Shaw PJ, Cirelli C, Greenspan RJ, Tononi G. Science. 2000;287:1834–1837. [PubMed]
10. Cirelli C, Bushey D, Hill S, Huber R, Kreber R, Ganetzky B, Tononi G. Nature. 2005;434:1087–1092. [PubMed]
11. Hendricks JC, Williams JA, Panckeri K, Kirk D, Tello M, Yin JC, Sehgal A. Nat Neurosci. 2001;4:1108–1115. [PubMed]
12. Hendricks JC, Lu S, Kume K, Yin JC, Yang Z, Sehgal A. J Biol Rhythms. 2003;18:12–25. [PubMed]
13. Kume K, Kume S, Park SK, Hirsh J, Jackson FR. J Neurosci. 2005;25:7377–7384. [PubMed]
14. Shaw PJ, Tononi G, Greenspan RJ, Robinson DF. Nature. 2002;16:287–291. [PubMed]
15. Graves LA, Hellman K, Veasey S, Blendy JA, Pack AI, Abel T. J Neurophysiol. 2003;90:1152–1159. [PubMed]
16. Helfand SL, Rogina B. Annu Rev Genet. 2003;37:329–348. [PubMed]
17. Guarente L, Picard F. Cell. 2005;120:473–482. [PubMed]
18. Partridge L, Pletcher SD, Mair W. Mech Ageing Dev. 2005;126:35–41. [PubMed]
19. Tatar M, Bartke A, Antebi A. Science. 2003;299:1346–1351. [PubMed]
20. Merry BJ. Aging Cell. 2004;3:7–12. [PubMed]
21. Feng J, Bussiere F, Hekimi S. Dev Cell. 2001;1:633–644. [PubMed]
22. Holzenberger M, Dupont J, Ducos B, Leneuve P, Geloen A, Even PC, Cervera P, Le Bouc Y. Nature. 2003;421:125–126.
23. Lin YJ, Seroude L, Benzer S. Science. 1998;282:943–946. [PubMed]
24. Orr WC, Sohal RS. Science. 1994;263:1128–1130. [PubMed]
25. Parkes TL, Elia AJ, Dickinson D, Hilliker AJ, Phillips JP, Boulianne GL. Nat Genet. 1998;19:171–174. [PubMed]
26. Clancy DJ, Gems D, Harshman LG, Oldham S, Stocker H, Hafen E, Leevers SJ, Partridge L. Science. 2001;292:104–106. [PubMed]
27. Rogina B, Reenan RA, Nilsen SP, Helfand SL. Science. 2000;290:2137–2140. [PubMed]
28. Hwangbo DS, Gershman B, Tu MP, Palmer M, Tatar M. Nature. 2004;429:562–566. [PubMed]
29. Helfrich-Forster C. J Biol Rhythms. 2000;15:135–154. [PubMed]
30. Sehgal A, Price JL, Man B, Young MW. Science. 1994;263:1603–1606. [PubMed]
31. Vieira C, Pasyukova EG, Zeng ZB, Hackett JB, Lyman RF, Mackay TF. Genetics. 2000;154:213–227. [PMC free article] [PubMed]
32. Rogina B, Helfand SL. Genetics. 1996;143:1643–1651. [PMC free article] [PubMed]
33. Kops GJ, Dansen TB, Polderman PE, Saarloos I, Wirtz KW, Coffer PJ, Huang TT, Bos JL, Medema RH, Burgering BM. Nature. 2002;419:316–321. [PubMed]
34. Rechtschaffen A, Bergmann BM. Sleep. 2002;25:18–24. [PubMed]
35. Aujard F, Herzog ED, Block GD. Neuroscience. 2001;106:255–261. [PubMed]
36. Van Reeth O, Zhang Y, Reddy A, Zee P, Turek FW. Brain Res. 1993;607:286–292. [PubMed]
37. Zhang Y, Kornhauser JM, Zee PC, Mayo KE, Takahashi JS, Turek FW. Neuroscience. 1996;70:951–961. [PubMed]
38. Yamazaki S, Straume M, Tei H, Sakaki Y, Menaker M, Block GD. Proc Natl Acad Sci USA. 2002;99:10801–10806. [PMC free article] [PubMed]
39. Huber R, Hill SL, Holladay C, Biesiadecki M, Tononi G, Cirelli C. Sleep. 2004;27:628–639. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...