Abstract
Carbon dioxide is a volatile and broad signal of many organic processes, and serves as a convenient cue for insects in search of blood hosts1–6, flowers7, decaying matter8–11, communal nests12, fruit13, and wildfires14. Curiously, although Drosophila melanogaster feed on yeast that produce CO2 and ethanol during fermentation, laboratory experiments suggest that flies actively avoid CO215–25. Here, we resolve this paradox by showing that both flying and walking fruit flies do actually find CO2 attractive, but only when they are in an active state associated with foraging. Aversion at low activity levels may be an adaptation to avoid CO2-seeking-parasites, or succumbing to respiratory acidosis in the presence of high concentrations of CO2 that are occasionally found in nature26,27. In contrast to CO2, flies are attracted to ethanol in all behavioral states, and invest twice as much time searching near ethanol compared to CO2. These behavioral differences reflect the fact that whereas CO2 is a generated by many natural processes, ethanol is a unique signature of yeast fermentation. Using genetic tools, we determined that the evolutionarily ancient ionotropic co-receptor IR25a is required for both CO2 and ethanol attraction, and that the receptors previously identified for CO2 avoidance are not involved. Our study lays the foundation for future research to determine the neural circuits underlying both state- and odorant-dependent decision making in Drosophila.
The life history of the fruit fly, Drosophila melanogaster, revolves around fermenting fruits, where they feed, mate, and deposit eggs. Their lifecycle from egg to adult takes approximately 10-14 days, roughly the same amount of time that most ripe fruit takes to decay. Thus, upon emerging from their puparia, adult flies need to locate a fresh ferment. The primary compounds produced by yeast fermentation are ethanol and CO2. Because of its high volatility, CO2 emission is greatest near the start of fermentation, whereas the ethanol emission increases more slowly (Fig. 1a). Other odors associated with fermentation, such as acetic acid and ethyl acetate, form later when bacteria begin to break down the ethanol.
To find an active rot, flies should therefore search near sources of both CO2 and ethanol. Variable air currents make it difficult to estimate the exact concentration of CO2 emitted from ferments in the wild; however, we measured the CO2 concentration in 500 mL bottles used to rear flies in many laboratories (see Methods and Fig. S1). Such bottles contain 0.5-1% CO2 depending on the amount of yeast and flies present (Fig. 1b), and serve as effective traps if left without a lid. In trap assays (Fig. 1c), Drosophila showed a preference for 2-day-old apple juice ferments compared to older solutions in which the yeast had flocculated and were no longer producing CO2 (Fig. 1d).
This casual evidence that CO2 attracts Drosophila contradicts many prior studies that concluded flies actively avoid CO2 in small chambers and T-mazes15–25. To study how flies respond to different odors under more ethologically relevant conditions, we recorded the flight trajectories28,29 of flies in a wind tunnel containing a fruit-sized landing platform, which we programmed to periodically release plumes of CO2 or ethanol (Fig. 2a-b). In the presence of either odor, flies were far more likely to approach and land on the platform. They also approached a dark spot on the floor of the wind tunnel (Fig. 2c-d), consistent with prior experiments with flies and mosquitoes2,29. Flies were more likely to approach the platform or the dark spot in the presence of ethanol compared to CO2, but were equally likely to land in the presence of either odor (Fig. 2e).
To quantify the behavior of flies after they land, we designed a new platform suitable for automated tracking (Fig. 3a-b). For a flow rate of 60 sccm CO2, the CO2 concentration near the surface of the platform was approximately 3% (Fig. 3c, S2). After landing near a source of CO2, ethanol, or apple cider vinegar, flies exhibited local search behavior (Fig. 3d), which we summarized using four descriptive statistics (Fig. 3e, S3). Flies spent approximately twice as much time exploring the platform in the presence of ethanol compared to CO2 or any other odor. Vinegar elicited smaller local searches than either CO2 or ethanol. While searching on the platform, flies approached the odor source most frequently for ethanol and CO2. Vinegar elicited slightly fewer approaches compared to CO2, consistent with the hypothesis that vinegar might indicate a less favorable, late-stage ferment. Flies spent significantly less time standing still on the platform in the presence of CO2 compared to any other odor, exhibiting an overall mean walking speed greater than 2mm s−1. When combined in a single odor stream, CO2 and ethanol together elicited a stronger search behavior than that exhibited to either odor alone.
One prior study using a tethered flight assay showed that Drosophila are attracted to CO2 while flying, a result that was attributed to the influence of the elevated levels of octopamine during flight30. Our results confirm this observation in freely-flying flies; however, we also found that flies continue to be attracted to CO2 after they land. One possible explanation for this discrepancy is that the elevated levels of octopamine during flight might influence the flies’ reactions to CO2 for a short time after landing. To test this hypothesis, we built an enclosed walking arena in which flies were unable to fly (Fig. 4a, S4-6), and presented them with pulses of 5% CO2 (close to the 3% concentration that elicited attraction in the wind tunnel assay). Starved flies presented with CO2 after acclimating to the arena for 10 min exhibited aversion, as has been previously reported in such chambers (Fig. 4b). However, if allowed to acclimate for two hours and then given a pulse of CO2, the animals exhibited attraction (Fig. 4c).
To study their responses to CO2 in more detail, we recorded the behavior of flies for 20 continuous hours in darkness, while offering 10 min long presentations of CO2 from alternating sides of the arena every 40 minutes (Fig. 4d). Throughout the experiments, both sides of the arena received 20 sccm of air saturated with water vapor. The flies exhibited a clear circadian rhythm in their activity within the chamber, as indicated by their mean walking speed. At times of peak activity — near their entrained dusk and dawn — flies showed a strong initial attraction to CO2, which decayed stereotypically during the 10 min presentation. At times of low activity — at mid-day and during the night — the flies exhibited a mild aversion to CO2. Starving flies for 24 hours prior to placing them in the chamber (instead of just 3 hours) changed their activity profile, resulting in a slightly elevated attraction during their subjective night. Ethanol, in contrast, elicited sustained attraction regardless of baseline activity or time of day (Fig. 4d).
Our experiments thus far suggest a possible correlation between activity and attraction to CO2. To test this hypothesis, we made several other environmental manipulations that are known to alter activity: increased temperature and wind speed (Fig. 4e). When we increased our bulk flow rate to 100 sccm, flies exhibited a peak walking speed (at dusk) of about 1.5 mm s−1, nearly half the speed we measured when using a flow rate of 20 sccm. This result is consistent with observations that flies stop moving in the presence of wind31. Instead of showing attraction, these flies exhibited aversion to 5% CO2 when it was presented at this higher flow rate; however, they still exhibited attraction to ethanol (Fig. 4e). This result helps to explain why previous studies that used high bulk airflow rates of 100-1000 sccm to present CO216,24 observed aversion. To further explore the effect of wind speed on behavior, we clipped the flies’ aristae. This manipulation destroys their primary means of detecting airflow but does not interfere with the detection of odors32. The aristae-less flies exhibited the same walking speed and attraction to CO2 at the high flow rate as exhibited by normal flies at the low flow rate. We also warmed flies with intact aristae to 32° C, which increased their baseline activity. These flies also exhibited attraction to CO2 at the higher flow rate. Pooling data across all our experimental conditions, we found that flies were attracted to CO2 when they had a baseline walking speed above ~2.4 mm s−1 (Fig. 4f). This result is similar to the mean walking speed value we observed in our wind tunnel assay, which was higher for CO2 than the other odors we tested. This suggests that there may be some underlying physiological connection between circuits regulating locomotor activity and those regulating CO2 attraction. To confirm that activity dependent attraction to CO2 is not a function of social interactions, we also performed experiments on 30 single flies, which on average behaved exactly as the cohorts of 10 (Fig. S7). We also tested three concentrations of CO2 (1.7%, 5%, 15%) and found that 5% elicited the strongest response, consistent with our wind tunnel experiments (Fig. S8).
Although flies’ responses to ethanol and CO2 were similar during the first minute of the stimulus, the attraction to ethanol was more sustained. The time course of behavior was remarkably similar in the walking arena and wind tunnel (Fig. 4g-j), suggesting that the behavioral dynamics of olfactory attraction are robust to the stimulus environment, and may represent an adaptation for utilizing information that ecologically broad (CO2) and more specific (ethanol) odorants provide.
Despite the ethological importance of both ethanol and CO2 as food cues for Drosophila, the olfactory receptors used to detect these odors during foraging are not known. To determine if CO2 attraction is mediated by either an olfactory (OR) or ionotropic (IR) receptor, we used our apparatus to test an IR8a; IR25a; Orco, Gr63a quadruple mutant, which lack the OR and IR co-receptors as well as a CO2-sensitive gustatory receptor (Fig. 5a-b). These near-anosmic mutants exhibited no detectable behavioral response to CO2. Flies in which we surgically removed the 3rd antennal segment also showed no response to CO2, despite otherwise normal levels of activity. Together with our arista ablations (Fig. 4e), these experiments show that CO2 attraction is mediated by the olfactory system in the 3rd antennal segment.
Prior research has shown that flies’ aversion to CO2 is mediated by a pair of olfactory receptors, Gr63a and Gr21a15,18,33, with high concentrations of CO2 also being detected by the acid-sensitive ionotropic receptor IR64a20 (which operates together with the co-receptor IR8a). Mutant flies lacking the IR64a receptor showed no significant change in their behavior compared to wild type (Fig. 5c). Mutants lacking the Gr63a receptor exhibited no aversion to CO2 (Fig. 5c), consistent with the prior literature; however, the same animals were still attracted to CO2 when more active. Homozygous Gr63a/IR64a double mutants behaved similarly to the Gr63a mutants. It is noteworthy that the characteristic decaying time course of attraction was unaffected in Gr63a mutants, even though these flies showed no aversion. This suggests that the decay in attraction to CO2 is not caused by an increase in aversion over time.
Collectively, our results suggest that none of the canonical CO2 receptors are responsible for attraction. The OR class of receptors is unlikely to mediate the response; indeed, Orco mutants exhibited a sustained attraction to CO2. IR8a mutants also exhibit normal attraction to CO2. Ir25a mutants, however, exhibited only aversion to CO2 at all activity levels, whereas rescuing Ir25a with a bacterial artificial chromosome34 rescued their attraction. Mutant flies lacking Orco, Ir8a, and Gr63a exhibit wild type attraction to CO2, indicating that none of these proteins are necessary for CO2 attraction. We further verified our results by testing a different mutant allele of the IR25a, which behaved in the same manner (Fig. S10).
Although ethanol is the most ethologically relevant cue related to fermentation and elicits the strongest attraction, the receptors for ethanol are still not known. We used the same set of co-receptor mutants to determine that IR25a, but not Orco or IR8a, is required for attraction to ethanol (Fig. 5d). IR25a is, however, not required for attraction to other odors associated with foraging, like apple cider vinegar (Fig. S9), confirming that these mutants are still capable of exhibiting attractive behaviors.
Prior studies reporting aversion to CO2 have suggested that it serves as a pheromonal cue (Drosophila Stress Odor, DSO) by which stressed flies signal others to flee a local enviroment15. Our result that active flies are attracted to CO2 is not consistent with this hypothesis. An alternative explanation for the prior findings is that stressed insects release CO2 simply because they have it stored in their tracheal system as part of the normal process of discontinuous respiration35,36. Indeed, we found that even mosquitoes (which are strongly attracted to CO2) release CO2 when shaken (Fig. S12). We suggest that the DSO hypothesis is a by-product of two unrelated behaviors: the release of tracheal CO2 by agitated flies and the avoidance of CO2 while in a behavioral state related to either low activity levels or being recently introduced to a new chamber (and thus likely to be in a behavioral state more associated with exploring a new environment than foraging). This aversive behavior may be an adaptation that helps sleeping flies either minimize encounters with parasites that are themselves attracted to CO2 as a means of finding hosts (parasitic wasps of Drosophila are attracted to yeast products37 and thus likely CO2; other hematophagous parasites are often attracted to CO21,3,4,6), or avoid succumbing to respiratory acidosis in the presence of high concentrations of CO2. Examples of insects being fatally attracted to high levels of CO2 have been reported in the literature27, and we have replicated this behavior in the lab (Fig. S13).
Our study adds Drosophila to the long list of insects that are attracted to CO238. This implies an ancient evolutionary role of CO2 in insect behavior, as well as a highly conserved means for detecting it. These hypotheses are supported by our finding that CO2 attraction in Drosophila requires the ionotropic co-receptor IR25a, the most highly conserved olfactory receptor among insects39 (over 550-850 million years old40). Curiously, attraction to CO2 in mosquitoes (as well as members of Coleoptera (Tribolium castaneum) and Lepidoptera (Bombyx mori)) is mediated—at least in part—by a system homologous to the Gr63a/Gr21a gustatory receptors that mediate aversion in Drosophila41. Other insect species that respond to CO2, including members of Hymenoptera (honeybees42 and ants12), Hemiptera (bed bugs3 and kissing bugs43), Blattodea (termites11), and Ixodida (ticks4), however, lack this receptor41. It is possible that these insects also use the same evolutionarily ancient IR25a dependent CO2 pathway that is responsible for attraction in Drosophila.
The different time course in attraction to CO2 and ethanol, as well as the state-dependent decision to move towards or away from CO2, make this system ripe for exploring ecologically relevant decision making. Unfortunately, the GAL4 driver for the IR25a promoter is only expressed in about half of the endogenous IR25a-expressing neurons44, making imaging, silencing, and activation experiments difficult to interpret at this time. By narrowing the possible pathways of CO2 and ethanol attraction to IR25a, we hope to motivate future efforts to develop new genetic reagents that will make it possible to study this system in greater detail.
Acknowledgements
We to thank Andrew Straw for providing the 3D tracking software for our flight experiments. Richard Benton offered helpful feedback on an early draft of the manuscript and also provided the IR8a; IR25a; Orco, Gr63a quadruple mutant. Ralf Stanewsky provided an IR25a mutant and the IR25a + BAC rescue line. Greg Suh provided an IR8a mutant. Elizabeth Hong and Jeff Riffell contributed helpful comments on the manuscript.