foraging alters resilience/vulnerability to sleep disruption and starvation in Drosophila

Results

We hypothesized that polymorphisms in foraging would influence the response to SD as measured by both sleep homeostasis, the increase in sleep seen following sleep loss, and short-term memory. Because diet strongly modulates the behavior of the foraging alleles, we first asked whether for^R flies would sleep significantly longer than for^s2 mutants when tested under our laboratory conditions as described previously (7). As seen in Fig. S1, under our dietary conditions, for^R flies sleep significantly longer than for^s2 mutants.

Next, we exposed for^R, for^s, and for^s2 flies to 12 h of SD during their primary sleep period using the sleep-nullifying apparatus. As seen in Fig. 1A, for^R flies did not compensate for lost sleep during 48 h of recovery, whereas both for^s flies and for^s2 mutants displayed a sleep rebound similar to that previously seen in Canton-s (Cs) flies (19–21). The lack of a homeostatic response seen in for^R flies may represent either an adaptation that allows animals to better withstand the negative effects of waking, or it may indicate that foraging disrupts regulatory processes, thereby preventing flies from obtaining needed sleep. Because deficits in short-term memory are a robust consequence of sleep loss (13, 22, 23), we evaluated short-term memory (STM) using aversive phototaxic suppression (APS) in for^R, for^s, and for^s2 flies following 12 h of SD. In the APS, flies are individually placed in a T maze and allowed to choose between a lighted and darkened chamber (13, 24). During 16 trials, flies learn to avoid the lighted chamber, which is paired with an aversive stimulus (quinine/humidity). The performance index is calculated as the percentage of times the fly chooses the dark vial during the last 4 trials of the 16-trial test (13, 25). As seen in Fig. 1B, for^R flies maintain their ability to learn following SD, whereas for^s flies are significantly impaired; for^s2 mutants showed impaired performance in the APS both under baseline conditions and following SD. for^R, for^s, and for^s2 flies did not differ in sensory thresholds as measured by either the photosensitivity index (PI; percentage of photopositive choices in 10 trials in the absence of quinine) or the quinine sensitivity index (QSI; time in seconds flies reside on the nonquinine side of a chamber) (Table S1) (13, 25). Given that for^R flies maintain their ability to learn following 12 h of SD and do not appear to be sleepy, as indicated by the absence of a sleep rebound, for^R flies are considered to be resistant to SD. In contrast, both for^s and for^s2 flies remain vulnerable to the negative effects of extended waking as measured by learning deficits and an increased sleep rebound.

Interestingly, the deleterious effects of extended waking are absent when waking is induced by starvation (18). Given that foraging alters the response to food deprivation, we hypothesized that for^R, for^s, and for^s2 mutants would show different vulnerabilities to waking induced by starvation. As seen in Fig. 1C, when for^R flies are placed in recording tubes with agar and water (starvation), they exhibit an immediate and sustained increase in waking behavior and show no evidence of a sleep rebound when placed back on their standard diet 12 h later. Interestingly, whereas the wake-promoting effects of starvation are absent in for^s flies as previously described (17, 18), for^s2 mutants respond to starvation with a significant increase in sleep (Fig. 1C). If for^R flies are resistant to the negative effects of waking induced by starvation, they should maintain their ability to learn in the APS as they did following SD. However, in contrast to waking induced by SD, for^R flies are impaired following waking induced by starvation (Fig. 1D). Surprisingly, for^s2 mutants, which exhibit impaired short-term memory both under baseline conditions and after SD, recover their ability to learn when starved; previous studies have shown that neither SD nor starvation alters PI or QSI t (13, 18, 25, 26). Thus, for^s2 mutants sleep more and display normal cognitive behavior following starvation, whereas for^R flies display an unexpected vulnerability in short-term memory when waking is induced by the absence of food. Consistent with previous results (17, 18), SD and starvation likely invoke distinct physiological responses even though each manipulation produces an increase in waking.

Resilience to sleep loss is indicated by the ability to maintain optimal performance after sleep disruption. Thus, although baseline learning in for^s2 mutants is at the level observed in memory mutants (13, 25), neither for^s2 mutants nor for^R flies, which are in the same genetic background, show performance decrements following sleep loss. With that in mind, we conducted a complementation test to examine learning in for^s/for^s2 flies under baseline and in response to both SD and starvation. As seen in Fig. 1E, for^s/for^s2 flies do not learn under baseline or after SD, but display STM after starvation. These results, along with the gain- and loss-of-function data presented below, indicate that it is the levels of foraging, rather than genetic background, which most likely account for differences in learning following SD and starvation.

Although waking up to forage during starvation would enhance the opportunity to find food, it requires additional energy expenditure. In contrast, sleeping would minimize the ability to find food but would likely conserve energy. Thus, we asked whether the alternate behavioral strategies exhibited by for^R and for^s2 mutants would be associated with changes in survival during starvation. The average difference in the LD₅₀ in hours to starvation between for^R and for^s2 mutants was 10.25 ± 3.19 (t = 2.306, P = 0.01, one-sample t test, n = 4 replicates); a representative example of survival during starvation is shown in Fig. 1E. Thus, whereas for^R flies appear resistant to the behavioral consequences of SD, for^s2 flies appear more suited to withstand the challenge of overnight starvation.

Given that sleep plays a role in memory consolidation (27) and that the foraging polymorphism has been shown to independently alter both sleep and memory (7, 8), we examined the relationship between sleep and plasticity in for^R, for^s, and for^s2 flies. Previous studies have shown that enriched social environments induce synaptic elaboration in mammals and flies and that these changes are followed by several days of increased sleep (28–30). Thus, we evaluated sleep in for^R, for^s, and for^s2 flies after they had been exposed to either social enrichment, which consists of ∼60 flies maintained in a 50-mL vial, or social isolation, which consists of flies being housed individually in TriKinetics tubes, for 5 d (31, 32). Surprisingly, neither for^s nor for^s2 flies, which have long-term memory (LTM) using olfactory conditioning (8), display an increase in sleep following social enrichment (Fig. 2A). In contrast, for^R flies, which have impaired LTM using olfactory conditioning, maintain their ability to increase sleep following social enrichment (Fig. 2A). Together, these data extend results with STM and LTM by showing that foraging plays a role in an additional type of plasticity, and further suggest that foraging may play a unique role when plasticity is induced in a social context.

To further test this hypothesis, we evaluated LTM in male flies using a spaced training protocol in a courtship conditioning assay that results in decreased courtship behavior for at least 48 h after training (31, 32). As seen in Fig. 2B, for^R and for^s flies display a significant reduction in courtship 48 h following spaced training (T) compared with their naïve siblings (N), indicating that they developed LTM. In contrast, for^s2 males show no reduction in courtship, indicating that they have impaired memory consolidation (Fig. 2B, Right). Note that whereas naïve courtship was low in for^s males, it was not so low as to preclude the development of LTM. Moreover, naïve courtship was also low in for^s2 mutants, which are in the same genetic background as for^R. Thus, it is likely that the reduced level of naïve courtship is due to the foraging gene and not due to genetic background. The observation that for^R flies show both an increase in sleep in response to social enrichment and LTM following courtship conditioning suggests that foraging may be particularly relevant for plasticity induced in a social context.

We have previously shown that sleep is increased following courtship conditioning in Cs flies and that LTM is disrupted if flies are sleep-deprived immediately following training (31). As seen in Fig. 2C, male for^R flies sleep significantly more following spaced training than their naïve siblings. Similarly, for^s2 mutants, which did not develop LTM, did not increase their sleep following training, consistent with previous reports that courtship behavior in the absence of LTM formation does not alter sleep (31). Given that for^s flies developed an LTM, it is unclear why they did not show an increase in posttraining sleep (Fig. 2C). However, one explanation may be that the changes in the behavior of for^s flies shown in Fig. 2B were too small to effectively induce changes in sleep. We next asked whether posttraining SD would disrupt memory consolidation in for^R and for^s flies. Interestingly, 4 h of SD immediately following spaced training (T+SD) did not disrupt LTM in for^R flies, whereas LTM was disrupted in for^s flies. Thus, foraging appears to allow memory consolidation to proceed in the absence of sleep.

Given that the mushroom bodies (MBs) modulate both sleep and memory (13, 33, 34), we hypothesized that for signaling in the MBs would phenocopy for^R and confer resistance to SD. Sleep homeostasis and performance in the APS were evaluated following 12 h of SD in flies overexpressing for in the MBs of otherwise for^s homozygous background. As seen in Fig. 3 A and B, when for is overexpressed primarily in the α/β lobes of the MB using the c739 or 30y GAL4 drivers, sleep rebound is significantly attenuated. Thus, expressing for in the MBs recapitulates the sleep rebound phenotype observed in for^R flies. Interestingly, overexpression of for using the 201y GAL4 driver, which expresses predominantly in the λ lobes and only weakly in the α/β lobes, does not significantly alter sleep rebound (Fig. 3C). As mentioned above, a low sleep rebound could represent either an adaptation that allows animals to better withstand the negative effects of waking, or a disruption in regulatory processes that prevent flies from obtaining needed sleep. Consistent with the for^R phenotype described above, overexpressing for using c739 or 30y in an otherwise for^s background also prevented deficits in short-term memory following SD (Fig. 3 D and E). 201y/+; for^s control flies were altered in the APS under baseline conditions, such that the effect of MB γ-lobe overexpression using 201y on short-term memory after SD could not be assessed. No differences in sensory thresholds were observed between genotypes. To determine whether reducing foraging within the MBs would phenocopy for^s2 mutants, we expressed UAS-for^RNAi using UAS-dicer²;30y-GAL4. As seen in Fig. 3F, UAS-dicer²;30y-GAL4/+>UAS-for^RNAi flies, but not parental controls, showed disrupted STM during baseline and after SD, whereas STM was restored following starvation. These results are similar to those observed in for^s2 mutants. These data indicate that for activity in the mushroom bodies, particularly the MB α/β lobes, recapitulates the sleep phenotypes observed in for^R and for^s2 flies.

A previous report indicates that for^R flies and flies overexpressing for within the MBs using c739, 30y, and 201y GAL4 drivers have impaired LTM following olfactory conditioning (8). However, the data presented above indicate that for^R flies can generate LTM following courtship conditioning. To further define the role of foraging in LTM induced by courtship conditioning, we expressed for in the MBs of an otherwise for^s background. Consistent with the results reported for olfactory conditioning, expressing for in the MB α/β lobes using c739 or 30y significantly disrupted LTM (Fig. 3 G and H). Together, these data suggest that the GAL4 system most likely produces a higher level of for within the MBs than is seen in the for^R flies. In contrast to its effect on olfactory conditioning, expressing for using 201y did not disrupt LTM (Fig. 3I). Once again, these data indicate that whereas the expression of for may appear to provide resilience in one environmental context (sleep deprivation), it may confer an unexpected vulnerability in other situations (LTM).

Finally, we asked whether for overexpression in the MBs would phenocopy the for^R response to starvation. Indeed, expressing for using either c739 or 30y in an otherwise for^s background results in a for^R response to 15 h of starvation beginning 3 h before lights off (Fig. 4 A and B, triangles). However, when for is expressed primarily in the γ lobes using the 201y GAL4 driver, the change in sleep during starvation does not differ from parental controls (Fig. 4C). Thus, expressing for, primarily in the α/β lobes but not the γ lobes, recapitulates many of the for^R phenotypes. In for^R flies, the increased waking observed during starvation is associated with reduced survival. Thus, we asked whether the increased waking observed in starved c739 or 30y flies overexpressing for would alter survival. As seen in Fig. 4 D and E, survival was not altered during starvation when for was expressed using c739 or 30y. However, survival during starvation was increased when using 201y to express for in the γ lobes (Fig. 4F). Although determining precisely how driving expression of for within the γ lobes extends survival is beyond the scope of the current investigation, these data suggest that the MB γ lobes may play a role in controlling and/or responding to metabolic signals. In any event, these data show that the localized expression of for within the MBs can alter both short-term and long-term susceptibility to starvation.

Discussion

Our results not only show that the naturally occurring foraging polymorphism modulates sleep homeostasis but also demonstrate that the resistance to sleep loss conferred by higher levels of foraging has a tradeoff that is revealed as an increased vulnerability to starvation. In contrast, lower levels of foraging are associated with resistance to starvation and a corresponding tradeoff, as indicated by an increased vulnerability to SD. Importantly, the phenotypes seen in foraging alleles can be largely recapitulated by overexpressing or reducing foraging in the α/β lobes of the MBs.

Does the variability in resilience to sleep loss that is conferred by the for polymorphism have ecological relevance? Currently, it is not clear whether the ability to withstand sleep loss can confer an advantage in reproductive fitness and, thus, influence natural selection at the for locus. Furthermore, the for locus is notably pleiotropic and has been implicated in modulation of learning and memory (8) as well as metabolic plasticity (35), making it difficult to specify which phenotype might respond to a given selection pressure aimed at changing the allelic variation at the for locus. It has been established, however, that flies carrying a given for allele have a relative fitness advantage when that allele is more rare (4). This finding is consistent with the idea that flies might exploit the resiliencies conferred by their for genotype to increase their chances of reproduction. For example, if a rover fly lives in a population where the for^s allele is most frequent, it might increase its reproductive fitness by forgoing sleep to mate at night while its sitter neighbors must rest. This strategy may allow the rover to reduce the competition for a mate yet still maintain optimal functioning the following day. Conversely, a sitter fly might outcompete rover rivals by forgoing a feeding to mate. Under this hypothesis, both natural for alleles (along with their associated resiliencies) could be maintained within a given population of flies.

Ecological pressures have been shown to affect cavefish, which have moved from living near the surface of lakes to deeper inside caves (36). Shifting ecological pressures have independently led each of these populations to sleep less than their surface-dwelling ancestors and, importantly, all three have converged upon similar genetic adaptations to adapt to a decrease in sleep time (36). It is possible that the polymorphism in for evolved in response to such ecological conditions, such as a prolonged food shortage that might select for animals able to withstand starvation or to seasonal changes in the length of nights that might place constraints on sleep time.

Ultimately, the extent to which resiliency to sleep loss contributes to the frequency of for alleles in clinically varying natural populations of flies remains to be determined. It is important to note, however, that roles for PKG in sleep regulation have been identified in Caenorhabditis elegans (7) and in mice (6), indicating that the influence of PKG on sleep regulation is likely to be evolutionarily conserved.

Human studies indicate that individuals vary greatly in their vulnerability to sleep loss (1). With that in mind, several laboratories have begun to examine naturally occurring polymorphisms in humans to determine their role in this differential sensitivity (2, 37, 38). For example, a polymorphism in PERIOD3 (PER3) is associated with larger cognitive deficits following SD (2). Similarly, a functional polymorphism in adenosine deaminase results in increased sleep pressure and increased sensitivity to SD (37). Moreover, a functional polymorphism in brain-derived neurotrophic factor alters EEG slow-wave activity (0.75–4.5 Hz) during both baseline and recovery following SD (38). Pharmaceuticals are commonly used to offset the negative results of SD (e.g., caffeine, Modafinil, etc.). Not surprisingly, polymorphisms also influence the efficacy of drugs to improve performance during SD (39). In this context, the present study suggests that cGMP signaling and PKG are a candidate pathway for sleep resilience. Single-nucleotide polymorphism in the human for ortholog PRKG1 could be investigated for association with sleep loss.

Thus, human studies have begun to identify molecular pathways that alter not only sleep time but resilience to sleep loss. Unfortunately, determining whether a polymorphism in humans is also associated with unexpected tradeoffs is time-consuming and costly. However, such experiments are tractable in the fly. Indeed, a previous report has found natural genetic variants that contribute to baseline sleep time during the fly's primary waking period (40). Our data extend these findings to show that naturally occurring polymorphisms alter sleep homeostasis and, importantly, can confer resilience to sleep loss. In addition, our data suggest that the power of Drosophila genetics can be applied to these questions to determine the mechanism and extent to which a polymorphism has unexpected tradeoffs. Understanding how these tradeoffs confer resilience or vulnerability to specific environmental challenges is highly relevant for understanding both the importance of sleep during evolution and translational sleep research.

Materials and Methods

Supplementary Material