Driven to drink: A study of reward

Natural rewards and abused drugs affect the function of the brain’s reward systems, and abnormal function of these brain regions is associated with addictive behavior (1–3). Some aspects of drug reward can be modeled in the genetically tractable fruit fly, Drosophila melanogaster. Flies exhibit complex addiction-like behaviors, including a lasting attraction for a cue that predicts ethanol intoxication (4) and a preference for consuming ethanol-containing food, even if made unpalatable (5). Here, we extend studies in the Drosophila model to incorporate the effect of social experiences, which can have long-lasting effects on behavior (6, 7).

We used two distinct sexual experiences to generate two cohorts of male flies. One cohort, rejected-isolated, was subjected to courtship conditioning (8); they experienced 1-hour sessions of sexual rejection by mated females, three times a day, for 4 days (Fig. 1A). Such conditioning suppresses future male courtship behavior, even toward receptive virgin females (9) (fig. S1). Flies in the mated-grouped cohort experienced 6-hour sessions of mating with multiple receptive virgin females (ratio 1:5) for 4 days. Flies from each cohort were then tested in a two-choice preference assay (10), in which they voluntarily choose to consume food with or without 15% ethanol supplementation (11). Results for the two cohorts differed markedly. The value of the ethanol preference index (which when positive signifies attraction) was consistently higher for the rejected-isolated cohort (Fig. 1B). The experiences did not alter food consumption when tested in the absence of ethanol (fig. S2).

 

Question

How does the sexual history of male flies impact their propensity to consume alcohol?

Experimental design

The timeline for the experiments is shown on the top part of panel A.

The green rectangles represent 6-hour mating sessions that the "mated-grouped" cohort experienced once a day for four consecutive days. The smaller blue rectangles below represent 1 hour periods of sexual rejection that the "rejected-isolated" cohort experienced three times a day for four consecutive days.

Once this was complete, the researchers allowed the flies to freely access two food sources for three days (represented by the large red rectangle titled "Ethanol consumption"). One source simply contained regular food. The other contained food that had been spiked with alcohol. The authors used a capillary feeder, or CAFE, for this part of the experiment because this device allows one to precisely measure how much food from each source is consumed. They then used these measurements to calculate a score for how much the flies "preferred" the alcohol. That is, how much more of the alcohol-spiked food they ate compared to the regular food.

The following video further explains this process: https://www.jove.com/video/54910/a-simple-way-to-measure-alterations-reward-seeking-behavior-using

Results

The graph in panel B shows the difference in preference for alcohol between the mated-grouped cohort (in green) and the rejected-isolated cohort (in blue). The X-axis represents days where food consumption measurements were carried out once per day for three consecutive days. The Y-axis represents the preference index score where numbers greater than zero indicate that the flies preferred the alcohol-spiked food, whereas numbers smaller than zero indicate that the flies preferred the regular food. Each point is an average value for all the flies in that group. As you can see the mated-group cohort either preferred the regular food or had no real preference between the two food options. The rejected-isolated group, on the other hand, consistently preferred the food with alcohol in it.

One of the differences between the mated-group cohort and the rejected-isolated cohort was in their housing conditions, as demonstrated in panel C. The mated-grouped flies were housed with fellow male flies, whereas the rejected-isolated flies were housed entirely by themselves. In order to see whether this social housing aspect could explain the observed preference for alcohol. the researchers repeated the same experiment, but this time they housed the sexually-rejected flies with other males (referring to them as the virgin-grouped cohort, shown in pink). The X and Y axes in panel C are the same as in panel B. The virgin-grouped flies also demonstrated a preference for alcohol, whereas the mated-grouped flies had no obvious preference (as was observed in the first experiment).

The authors also wanted to assess whether it was the social aspect of sexual rejection, rather than the lack of copulation, that led the virgin flies to consume more alcohol. To test this, they repeated the experiment, this time exposing the virgin male flies to dead, decapitated females rather than live ones, thus taking the social experience of rejection out of the picture. These flies also showed a preference for alcohol (shown in orange) relative to the mated-grouped flies (shown in green).

Finally, in panel E, the researchers wanted to see if flies who had been sexually deprived could be made to reverse their preference for alcohol if allowed to mate just prior to food exposure. Using the same initial conditions from earlier iterations, this time they took some of the rejected-isolated flies and allowed them to mate just before consuming food. These flies ended up having no preference for alcohol (shown in purple), whereas the flies who remained sexually deprived did (shown in blue).

Conclusions

Across various conditions, the researchers found that sexual deprivation enhanced flies' preference for alcohol, and that this preference could not be explained by other tested factors.

The rejected-isolated and mated-grouped cohorts differ in several respects in addition to sexual deprivation (lack of copulation) per se, including individual versus group housing, exposure to the social experience of rejection, and exposure to aversive chemosensory cues found on mated females. Several experiments were designed to determine which of these was the predominant contributor to the enhanced ethanol preference seen in rejected-isolated males (11). First, we compared males that differed in sexual experience but not in housing conditions—that is, mated and virgin males that were both group-housed. The virgin males showed higher ethanol preference, although in general not quite as high as rejected-isolated males. This argues that isolation is not the major explanation for the enhanced ethanol preference.

We next investigated ethanol preference in males that were sexually deprived (blocked from copulating) but not exposed to the social experience of rejection. For this purpose, males were exposed individually to decapitated virgin females on the same schedule as the rejected-isolated cohort, using a protocol that results in courtship suppression (9). These males, which experience neither rejection nor copulation, showed enhanced ethanol preference when compared to the mated-grouped cohort (Fig. 1C); the preference index was similar to that displayed by the rejected-isolated cohort. These results point to sexual deprivation per se, rather than rejection, as the major factor influencing ethanol preference.

Finally, we sought to establish whether there was a role for the repellant chemosensory cue cis-vaccenyl acetate (cVA), which is found on the cuticle of mated but not virgin females (12). We compared males trained with either mated females (rejected-isolated) or decapitated virgin females (11). Both groups endured sexual deprivation (lack of copulation), but only the former was exposed to cVA. There was no difference in ethanol preference between these two cohorts (fig. S3A). There was also no difference between males exposed individually to biologically relevant concentrations of cVA (13) and vehicle-exposed controls (fig. S3B) (11). Together, these experiments point to sexual deprivation per se, rather than other factors, as the major contributor to enhanced ethanol preference.

To further test the strength of this conclusion, we divided a cohort of rejected-isolated males into two subgroups, one of which was left undisturbed, and the other of which was allowed to mate with virgin females for 2.5 hours immediately before testing. Ethanol preference was markedly lower in the rejected, then mated subgroup (Fig. 1E) compared to the subgroup that had only experienced rejection. Thus, the effects of sexual deprivation can be reversed by copulation, which is consistent with sexual deprivation being the major contributor to ethanol preference.

We focused on Drosophila neuropeptide F (NPF) as a potential mediator of the effects of sexual experience. The mammalian NPF homolog, neuropeptide Y [NPY (14)], regulates ethanol consumption (15), the NPF–NPF receptor (NPFR) system regulates acute ethanol sensitivity in Drosophila (16), and the Caenorhabditis elegans NPY receptor homolog NPR-1 regulates ethanol behaviors (17). Intriguingly, stressful experiences regulate mammalian NPY levels. These include restraint stress and early maternal separation in rodents and post-traumatic stress disorder in humans (18–20). However, a direct connection between social experience, NPY, and ethanol-related behaviors has not been established.

To investigate whether NPF mediates ethanol preference in Drosophila, we first compared NPF transcript levels in heads of males subjected to different sexual experiences: rejected-isolated, virgin-grouped, and mated-grouped (11). Rejected-isolated males showed the lowest transcript levels, virgin-grouped males showed higher levels, and mated-grouped males showed the highest (Fig. 2A and fig. S4). Rejected-isolated males also showed markedly lower NPF protein levels than mated-grouped males by immunohistochemistry (Fig. 2, B to D).

 

Question

What is the biological/molecular basis for this observed behavior?

Experimental design

The authors tested whether the amount of neuropeptide F (NPF) in the flies' brains was correlated with sexual history. They did this by first measuring the amount of NPF transcript (which can be used as a proxy for the molecule itself) in the brains of rejected flies, regular virgin flies, and mated flies. They also then directly measured the amount of NPF protein levels in rejected and mated flies.

Results

Panel A depicts a bar graph showing the average levels of NPF transcript (on the Y-axis) across the three types of flies (on the X-axis). The rejected group has the least amount of NPF transcript, whereas the mated flies have the most. Unrejected, virgin flies fall in between the other two groups.

Panels C and D show an example of a fly's brain from the mated (panel C) and rejected (panel D) groups. The first image on the left of both panels shows all of the places in the brain where the NPF molecule was found (shown in green). The middle image shows all of the places in the brain where there are neurons (shown in red). The right image overlays the left and center columns on top of one another, thus showing all of the neurons in the brain where NPF was found (shown in yellow). Just by looking at these images (especially the right column), one can see that the amount of NPF in the brains of rejected flies (panel D) is far less than the amount in the brains of mated flies (panel C).

Panel B represents a quantification of the phenomenon observed in panels C and D. The researchers did this by simply counting the number of yellow pixels in the images (on the Y-axis) across the two conditions (on the X-axis). Clearly, the amount of NPF is higher in the brains of the mated flies as compared to rejected flies.

Conclusions

The amount of NPF in a male fly's brain is inversely correlated with its sexual history, suggesting that NPF might be involved in affecting how much alcohol a fly prefers.

To determine whether the inverse correlation between NPF levels and ethanol preference reflects a cause-and-effect relationship, we manipulated the NPF-NPFR system genetically. We first tested the effect of NPFR down-regulation by expressing an NPFR-specific short interfering RNA (UAS-NPFR^RNAi) pan-neuronally (using elav-GAL4). This manipulation significantly reduced ethanol preference in mated males, which have elevated NPF levels, but not in virgin males (Fig. 3, A and B). Second, we tested the effect of artificial activation of NPF neurons by expressing the heat-activated cation channel dTRPA1 (21) under NPF-GAL4 control (22). There was no effect on ethanol preference when virgin males were tested at 20°C, when the channel is inactive, but there was aversion to ethanol-supplemented food at 29°C, when the channel is active (Fig. 3, C and D). An intermittent dTRPA1 activation protocol that more closely mimics our conditioning protocol produced similar aversion (fig. S5). These data suggest a causal relationship between sexual experience, NPF levels, and ethanol preference.

 

Question

Does the inverse correlation between the amount of NPF and amount of alcohol consumed reflect a cause-and-effect relationship?

Experimental design

To address this question, the authors sought to artificially decrease the effect of NPF in the brains of mated flies to see if doing so would make them behave more like virgin flies (i.e. increase their preference for alcohol). They did this genetically by inserting a type of molecule (RNAi) that prevented neurons from synthesizing any receptors for NPF, thus rendering NPF useless. As a control, the researchers also carried out two other, unrelated genetic manipulations in a different set of flies, but this manipulation had no effect on NPF.

Next, the scientists manually over-activated any neurons containing NPF to see if doing so would cause virgin flies to behave more like mated flies (i.e. decrease their preference for alcohol). They did this by inserting a molecule known as dTRPA1 into all NPF-containing neurons because dTRPA1 can activate neurons at high temperatures. They also made two control groups of flies who underwent some genetic manipulation, but did not have dTRPA1. They then tested the experimental (dTRPA1-containing) and control flies as virgins on their preference for alcohol. They did this once at a low temperature (when NPF neurons would not be overactivated due to dTRPA1), and once at a high temperature (when NPF neurons would overactivate due to dTRPA1).

Results

The graph in panel A shows alcohol preference (on the Y-axis) over four days (on the X-axis) for virgin flies who had their NPF receptors blocked with RNAi (shown in red). The two genetic controls that did not affect NPF receptors are shown in blue and green. Clearly, the flies who had their NPF receptors reduced (thus making NPF ineffective) were no different from the controls. This is not surprising however, since NPF levels in male, virgin flies are low to begin with.

The graph in panel B shows alcohol preference again over the three conditions (one where NPF receptors were blocked with RNAi and two controls), this time with mated flies. Here we see that diminishing NPF receptors (shown in red) caused flies to prefer alcohol, which is very similar to what was observed with virgin flies. The flies in the control groups, on the other hand (shown in blue and green), were unaffected and did not have an alcohol preference.

Graphs in panels C and D also support this relationship where male, virgin flies who had dTRPA1 inserted into their NPF neurons (shown in red) also show preference for alcohol. There are also two control groups (shown in blue and green). In panel C at low temperatures when NPF neurons were not activated, the experimental cohort of virgin flies (red) continued to prefer alcohol, much like the controls. Notice the similarities between the graphs in panels A and C. However, when temperatures were raised and NPF neurons were activated (as shown in panel D), the experimental virgin flies no longer preferred alcohol, even though the controls did.

Conclusions

The relationship between NPF and alcohol preference is causal, since when the activity of NPF is reduced in mated flies, they behave like virgin flies, and when the activity of NPF is enhanced in virgin flies, they behave like mated flies.

We propose that the activity of the NPF-NPFR system may be a neural representation of the state of the Drosophila reward system. If so, experiences that change NPF-NPFR activity should promote behaviors that restore the system to its normal state. In this model, sexual deprivation would create an NPF deficit that increases reward-seeking behavior such as ethanol consumption. Conversely, successful copulation would create a NPF surfeit that reduces reward seeking. This model predicts that mating and ethanol consumption should be rewarding (4, 5), that activation of the NPF-NPFR pathway is rewarding per se, and that artificial activation of the NPF circuit will diminish ethanol reward-seeking behavior.

To test these predictions, we used a series of conditioning assays in which male flies were trained to associate the proposed rewarding experiences (mating, ethanol exposure, or NPF circuit activation) with one of two neutral odor cues. After 24 hours, flies were tested for their odor preference; development of a preference for the odor associated with these experiences would imply that flies found the events rewarding. To test if mating is rewarding, males were exposed sequentially for 30 min to two odorants [ethyl acetate (EA) or isoamyl alcohol (IAA)], one in the absence and the other in the presence of virgin females, and tested for odor preference 24 hours later in the absence of females. A conditioned odor preference index (CPI) for mating was calculated by averaging preference indices for reciprocally trained groups of flies. Positive CPI values indicate conditioned preference, negative values indicate aversion. Males displayed a strong preference for the mating-associated odor (Fig. 4A). We have separately shown that flies exhibit conditioned preference for an odor associated with ethanol intoxication in a similar assay (4). Together, these results indicate that both mating and ethanol intoxication, the latter of which is likely achieved in the two-choice consumption assay (5), are indeed rewarding experiences to male flies.

 

Question

Are mating and activation of NPF neurons inherently rewarding experiences? If so, can the rewarding experience of NPF neuron activity lead to a reduction in alcohol consumption?

Experimental design

The researchers used a conditioning assay, which is believed to reveal what an organism does or doesn't find rewarding (for more on this, see the "Author’s experiments" annotation for "conditioning assay" in the main text). In short, they trained flies to associate an odor with either the experience of mating or the experience of artificial activation of NPF neurons. Then, they provided flies with a choice between this paired-odor or a second, neutral odor to see whether the flies would gravitate towards the paired-odor, even in the absence of the reward it was initially paired with.

In order to activate NPF neurons, the researchers once again used dTRPA1 (similar to as in Figure 3C and D). This molecule can activate neurons at high temperatures. To achieve the odor pairing, they made the flies experience alternating high and low temperatures, each of which was associated with a distinct odor (shown with the red and blue rectangles in panel B). Thus, one odor was always experienced at high temperatures when NPF neurons would have been active thanks to dTRPA1, whereas the odor was always experienced when these same neurons would have been relatively inactive. The next day, they tested these flies as described above (represented by the green rectangle) to see which odor they would prefer. The timeline for this experiment is outlined in panel B.

Results

Panel A shows the flies' preference for an odor that was paired with mating (over the second odor which was not paired with any substance/experience). The Y-axis indicates the calculated "preference score," where positive values indicate a preference for the mating-paired odor. As you can see, male flies had a significantly higher preference for this odor over the neutral odor, indicating that mating is an inherently rewarding experience.

The bar graph in panel C shows flies' preference for an odor that was paired with NPF activation (right-most bar) relative to the second, unpaired odor. The Y-axis once again indicates the preference score. The left-most and center bars show the preference of two cohorts of flies in the control groups who underwent a partial genetic manipulation but did not experience NPF activation. From the bar graph you can see that the control animals did not particularly prefer either odor, whereas the flies that experienced NPF activation in conjunction with one odor significantly preferred that odor.

The graphs in panels D and E show the flies' preference for an ethanol-paired odor over an unpaired odor across two conditions. In one condition, the ethanol-odor pairing process was also accompanied with artificial overactivation of NPF neurons (this is shown in panel E where the temperature was high and the dTRPA1 inserted into the NPF neurons therefore activated them). In the other condition shown in panel D, the temperature was kept low, so the dTRPA1 had no effect and the NPF neurons behaved as normal. The orange bars correspond to genetic controls in both conditions that were unaffected by dTRPA1. The red bars correspond to the experimental group (in which dTRPA1 was functional). Notably, when the temperature was low and NPF neurons were allowed to behave as they normally would during the ethanol-odor pairing process, both the experimental and control flies demonstrated a preference for this paired odor when tested 24 hours later. However, when the temperature was high and NPF neurons were overactivated during the ethanol-odor pairing process, the preference for the ethanol paired odor was abolished.

Panel F shows the change in NPF transcript levels (which serve as a proxy for the amount of NPF present in the brain) in response to exposure to air (shown in blue) as well as ethanol vapor (shown in red). NPF transcript levels were tested an hour after the exposure, as well as 24 hours after the exposure. When exposed to ethanol, NPF transcripts rose to significantly greater levels compared to air in the hour immediately following the exposure. However, within 24 hours, this spike in NPF transcript in response to the ethanol exposure returned back to normal and was no different than the amount of NPF transcript found in response to air.

Conclusions

Ethanol, mating, and artificial activation of NPF neurons are inherently rewarding. Further, if NPF neuron activity is artificially made to compete with the experience of ethanol intoxication, it can abolish the rewarding effects of the ethanol.

To test whether activation of the NPF-NPFR pathway is rewarding per se, we trained virgin males to associate artificial activation of NPF neurons with either EA or IAA. Males expressing dTRPA1 in NPF neurons (NPF-GAL4 + UAS-dTRPA1) and the genetic controls each carrying only one of the two transgenes were trained for three 1-hour sessions at 29°C, with dTRPA1 active, interspersed with three 1-hour rest periods at 18°C, with dTRPA1 inactive (Fig. 4B). When tested 24 hours later, males in the experimental group demonstrated strong preference for the odor associated with NPF neuron activation. The genetic controls, which did not undergo NPF neuron activation, but were exposed to the same training protocol, developed no odor preference (Fig. 4C). Other controls, which underwent NPF neuron activation but were not exposed to the training protocol, similarly developed no odor preference (fig. S6C). Thus, activation of the NPF-NPFR system is in itself rewarding to flies.

We next tested whether artificial activation of the NPF-NPFR system diminishes ethanol reward seeking. Flies were trained to associate EA or IAA with a moderately intoxicating exposure of ethanol vapor (three 10-min training sessions spaced by 1 hour) as described before (4). Wild-type flies normally show conditioned aversion to ethanol (negative CPI) when tested 30 min after training, and conditioned preference 24 hours later (4). We used this assay to compare virgin male flies that expressed dTRPA1 in NPF neurons (NPF-GAL4 + UAS-dTRPA1) with genetic controls that did not. Artificial activation of NPF cells, which occurs at 30°C but not 22°C, had no effect on the initial aversion (fig. S6, A and B), but abolished conditioned preference for ethanol (Fig. 4, D and E). Thus, NPF neuron activation, which is in itself rewarding to flies, interferes with the ability of flies to form a positive association between ethanol intoxication and an odor cue, which is reflected in lower ethanol consumption.

If the NPF-NPFR system were to function generally to signal the state of the Drosophila reward system, NPF levels should be increased by rewarding experiences other than mating, such as exposure to intoxicating levels of ethanol. To test this hypothesis, we exposed virgin males to ethanol vapor using an exposure paradigm previously shown to be rewarding (three 10-min exposures spaced by 1 hour) (4). NPF transcript levels increased 1 hour after exposure and returned to basal levels 24 hours later (Fig. 4F). Because the ethanol-induced changes in NPF levels are transient, whereas the memory of ethanol reward lasts for several days, it is possible that ethanol-induced changes in NPF levels set in motion a process, likely involving dopaminergic systems (4, 22), that modifies the fly reward system. Indeed, the activity of the NPF circuit could remain altered long after the levels of NPF had returned to baseline. Regardless of the exact mechanism, these data suggest that activity of the NPF system is regulated by at least two rewarding experiences, mating and ethanol intoxication.

NPF has been shown to influence several complex behaviors in flies, including larval intake of noxious food (23), a switch in feeding behavior (24), and responses to physical stressors (25) and ethanol (16). In addition, NPF neurons modulate the effect of satiety on sugar reward memory (22). In our paradigm, NPF appears to play a different role: Its expression is regulated by sexual experience and ethanol intoxication, and activation of NPF neurons acts as a reward signal, thereby abolishing ethanol reward and the enhanced ethanol consumption observed after sexual deprivation. It is likely that the effects of NPF we describe here are mediated by a different set of NPFR-expressing neurons than those mediating NPF’s role in sugar reward memory.

Mammalian NPY has several distinct behavioral functions that are mediated by different brain regions, including roles in feeding (26, 27), anxiety, stress (28), sleep regulation (29), sexual motivation (30), and ethanol consumption (15, 31). Stressors also regulate NPY levels (18–20). In addition, injection of NPY into the nucleus accumbens of rats is rewarding (32), and NPY administration relieves the negative affective states of drug withdrawal and depression (33, 34).

Our findings are thus not only consistent with known functions of mammalian NPY and its mode of regulation, but also provide evidence for NPF functioning as a key molecular transducer between social experience and drug reward. Drosophila is a useful and accessible model system in which to decipher the mechanisms by which social experiences interact with reward systems.

Supporting Online Material

www.sciencemag.org/cgi/content/full/335/6074/1351/DC1

Materials and Methods

Figs. S1 to S6

References

† Present address: Howard Hughes Medical Institute, Janelia Farm Research Center, Ashburn, VA 20174, USA.

References and Notes

1. A. E. Kelley, K. C. Berridge, J. Neurosci. 22, 3306 (2002).
2. S. E. Hyman, Am. J. Bioeth. 7, 8 (2007).
3. G. F. Koob, Neuropharmacology 56 (suppl. 1), 18 (2009).
4. K. R. Kaun, R. Azanchi, Z. Maung, J. Hirsh, U. Heberlein, Nat. Neurosci. 14, 612 (2011).
5. A. V. Devineni, U. Heberlein, Curr. Biol. 19, 2126 (2009).
6. G. E. Robinson, R. D. Fernald, D. F. Clayton, Science 322, 896 (2008).
7. V. Krishnan, E. J. Nestler, Curr Top. Behav. Neurosci. 7, 121 (2011).
8. S. M. McBride et al., Neuron 24, 967 (1999).
9. J. E. Mehren, A. Ejima, L. C. Griffith, Curr. Opin. Neurobiol. 14, 745 (2004).
10. W. W. Ja et al., Proc. Natl. Acad. Sci. U.S.A. 104, 8253 (2007).
11. Materials and methods are available as supporting material on Science Online.
12. A. Ejima et al., Curr. Biol. 17, 599 (2007).
13. J. C. Billeter, J. Atallah, J. J. Krupp, J. G. Millar, J. D. Levine, Nature 461, 987 (2009).
14. R. S. Hewes, P. H. Taghert, Genome Res. 11, 1126 (2001).
15. T. E. Thiele, D. J. Marsh, L. Ste. Marie, I. L. Bernstein, R. D. Palmiter, Nature 396, 366 (1998).
16. T. Wen, C. A. Parrish, D. Xu, Q. Wu, P. Shen, Proc. Natl. Acad. Sci. U.S.A. 102, 2141 (2005).
17. A. G. Davies, J. C. Bettinger, T. R. Thiele, M. E. Judy, S. L. McIntire, Neuron 42, 731 (2004).
18. A. Thorsell, K. Carlsson, R. Ekman, M. Heilig, Neuroreport 10, 3003 (1999).
19. P. A. Jiménez-Vasquez, A. A. Mathé, J. D. Thomas, E. P. Riley, C. L. Ehlers, Brain Res. Dev. Brain Res. 131, 149 (2001).
20. R. Sah et al., Biol. Psychiatry 66, 705 (2009).
21. M. Rosenzweig et al., Genes Dev. 19, 419 (2005).
22. M. J. Krashes et al., Cell 139, 416 (2009).
23. Q. Wu, Z. Zhao, P. Shen, Nat. Neurosci. 8, 1350 (2005).
24. Q. Wu et al., Neuron 39, 147 (2003).
25. J. Xu, M. Li, P. Shen, J. Neurosci. 30, 2504 (2010).
26. A. Thorsell, R. Rimondini, M. Heilig, Neurosci. Lett. 332, 1 (2002).
27. T. M. Hahn, J. F. Breininger, D. G. Baskin, M. W. Schwartz, Nat. Neurosci. 1, 271 (1998).
28. M. Heilig, Neuropeptides 38, 213 (2004).
29. M. Dyzma, K. Z. Boudjeltia, B. Faraut, M. Kerkhofs, Sleep Med. Rev. 14, 161 (2010).
30. S. P. Kalra, J. T. Clark, A. Sahu, M. G. Dube, P. S. Kalra, Synapse 2, 254 (1988).
31. A. Thorsell, Peptides 28, 480 (2007).
32. S. A. Josselyn, R. J. Beninger, Pharmacol. Biochem. Behav. 46, 543 (1993).
33. J. P. Redrobe, Y. Dumont, R. Quirion, Life Sci. 71, 2921 (2002).
34. K. A. Stogner, P. V. Holmes, Eur. J. Pharmacol. 387, R9 (2000).

Acknowledgments: We thank F. Wolf, M. Schuldiner, O. Schuldiner, A. Devineni, K. McClure, R. Joseph, and N. Velazquez Ulloa for advice and comments on the manuscript; J. Levine for discussions; P. Shen for fly strains and antibodies; and the Bloomington Stock Center for fly strains. Funding was provided by the Sandler Research Fellowship (G.S.-O.), the Program for Breakthrough Biomedical Research at the University of California, San Francisco (U.H.), and NIH–National Institute on Alcohol Abuse and Alcoholism (U.H.). The authors declare that they have no competing interests. A detailed description of all materials and methods as well as supplementary figures are available as supporting online material.