Impaired climbing and flight behaviour in Drosophila melanogaster following carbon dioxide anaesthesia

Laboratories that study Drosophila melanogaster or other insects commonly use carbon dioxide (CO2) anaesthesia for sorting or other work. Unfortunately, the use of CO2 has potential unwanted physiological effects, including altered respiratory and muscle physiology, which impact motor function behaviours. The effects of CO2 at different levels and exposure times were examined on the subsequent recovery of motor function as assessed by climbing and flight assays. With as little as a five minute exposure to 100% CO2, D. melanogaster exhibited climbing deficits up to 24 hours after exposure. Any exposure length over five minutes produced climbing deficits that lasted for days. Flight behaviour was also impaired following CO2 exposure. Overall, there was a positive correlation between CO2 exposure length and recovery time for both behaviours. Furthermore, exposure to as little as 65% CO2 affected the motor capability of D. melanogaster. These negative effects are due to both a CO2-specific mechanism and an anoxic effect. These results indicate a heretofore unconsidered impact of CO2 anaesthesia on subsequent behavioural tests revealing the importance of monitoring and accounting for CO2 exposure when performing physiological or behavioural studies in insects.

isoflurane anaesthesia 8 . Woodring et al. showed that CO 2 and anoxia affect aspects of growth, feeding, development, reproduction and behaviour in the cricket, Acheta domesticus 9 . Exposure to CO 2 disturbs normal development and affects movement in the German cockroach, Blattella germanica 10,11 , blocks glutamate receptors at neuromuscular junctions of D. melanogaster 12 and has widespread effects on the nervous system of the crayfish, Procambarus clarkii 13 . Carbon dioxide anaesthesia also increases D. melanogaster haemolymph acidity and causes a reduction in heart rate 12 . However, there have only been a limited number of behavioural studies of the effects of CO 2 anaesthesia in D. melanogaster, and much remains to be explored.
Use of 100% CO 2 to anaesthetize insects not only exposes them to high levels of CO 2 , but simultaneously exposes these organisms to a completely anoxic environment. Thus, exposure to pure CO 2 impairs or eliminates oxygen (O 2 ) delivery to the tissues, likely compromising important energy production pathways. Insects in general are very tolerant to hypoxia and/or anoxia for prolonged periods of time, but their responses vary significantly 14,15 . Drosophila melanogaster, for example, can tolerate four hours of pure nitrogen (N 2 ) exposure and survive 16 . In contrast, A. domesticus may experience significant physiological disruption after only a few minutes of anoxia 9 . Prolonged anoxic conditions conjoined with the overall harsh chemical nature of CO 2 may result in long lasting or permanent changes that may have not been previously accounted for. Due to the large amount of behavioural and physiological studies performed using insects as model organisms, it is of extreme importance to study the effects of CO 2 and anoxia on insects as this may confound data obtained in these studies. For appropriate selection of an anaesthetic method to use in insect behavioural and physiological studies, it is important to determine if anoxia and CO 2 anaesthesia have separate effects and how persistent these effects are. This study determined the effects of CO 2 , hypoxia and anoxia on D. melanogaster climbing and flight behaviours, which are routinely used to assay motor function and performance in flies.

Results
Drosophila melanogaster climbing and flight is inhibited by CO 2 exposure. Exposure to 100% CO 2 for 10 minutes reduced D. melanogaster climbing ability by 81% after being allowed to recover for one hour (Fig. 1a, P < 0.001). This reduced climbing ability persisted for at least 7 days (37% reduction, P < 0.001). The reduction in climbing exhibits a dose-dependent response, with longer CO 2 exposure times worsening the climbing deficits ( Fig. 2a, P < 0.001). Even a short, five minute exposure led to a 57% reduction in climbing after a one hour recovery (Fig. 2a, P < 0.001), which lasted for 24 hours. Air 100% CO 2 Gas Exposure a b Figure 1. Drosophila melanogaster climbing and flight abilities are reduced following a 10 minute exposure to 100% CO 2 . Flies were exposed to 100% CO 2 for 10 minutes and allowed to recover for varying times and assayed for the ability to (a) climb and (b) fly. CO 2 inhibited climbing at all recovery time points assayed while flight was only reduced through 8 hours. Data points represent the mean ± SEM of the percentage of flies able to climb or fly. Statistical analysis was performed by Student's t-tests. a = P < 0.001, b = P < 0.01. This experiment was performed in the Oregon-R wild-type strain. To determine if this was a strain-specific occurrence, other wild-type strains were investigated. A 10-minute exposure to 100% CO 2 induced climbing deficits after a 1 hour recovery in the Samarkand (20.4% reduction, P < 0.05, n = 7), Swedish-C (19.9% reduction, P < 0.01, n = 16) and Lausanne (24.5% reduction, P < 0.001, n = 19) strains. This indicates that this is not a strain-specific phenomenon. Drosophila melanogaster flight was negatively affected by a 10 minute exposure to 100% CO 2 ; showing a 32% reduction following a one hour recovery period (Fig. 1b, P < 0.001). These effects persist for at least 8 hours. Similar to the effects on climbing behaviour, a dose-dependent effect of CO 2 exposure is observed on the recovery of flight ability (Fig. 2b). While the negative effect of a five minute CO 2 exposure is absent by 8 hours, flies that were exposed for 15 minutes do not recover by 24 hours (Fig. 2b, P < 0.01). The flight deficit caused by a 15 minute CO 2 exposure at 24 hours was reduced by an additional 14% when the exposure time was increased to 30 minutes ( Fig. 2b, P < 0.05).
Anoxia reduces D. melanogaster flight but not climbing. To test whether the effect of CO 2 was due to anoxia, flies were exposed to 100% N 2 and their climbing and flight were assayed following a one hour recovery period. Following a 10 minute exposure to 100% N 2 , flight ability was reduced by 18% ( Fig. 3b, P < 0.001), while climbing was unaffected (Fig. 3a). However, there was no detrimental effect on flight or climbing with a 10 minute exposure to a 99% N 2 /1% O 2 mix (Fig. 3c,d).
Both severe and moderate CO 2 levels reduce D. melanogaster climbing. Different levels of CO 2 balanced with O 2 and N 2 were administered to D. melanogaster. Again, exposure to 100% CO 2 led Figure 2. Drosophila melanogaster climbing and flight abilities are reduced by 100% CO 2 in a dosedependent manner. Flies were exposed to 100% CO 2 for the times (minutes) noted in the legend and allowed to recover for varying times and assayed for the ability to climb and fly. (a) CO 2 exposure inhibited climbing at all recovery time points assayed. Compared to the flies that were exposed for 10 minutes, the five minute exposure flies climbed better at all recovery time points, while the 30 minute exposed flies performed worse at the one and two hour recovery time points. (b) Flight was reduced in flies exposed to 100% CO 2 for five minutes for up to four hours, and for 8 hours in flies exposed for 10 minutes. Flies that were exposed for 15 or 30 minutes had reduced flight at all recovery time points assayed. Compared to the flies that were exposed for 10 minutes, the 30 minute exposure flies performed worse at all recovery time points, while the 15 minute exposed flies flew less only at the 24 hour recovery time point. The Air-exposed points represent the average of all controls performed for that time point; however, the statistical analysis (Student's t-test) was performed with the individual control flies for each exposure length. The data points represent the mean ± SEM of the percentage of flies able to climb or fly. a = P < 0.01 vs. Air control flies, b = P < 0.05 vs. 10 minute CO 2 flies.
to a marked reduction in climbing ability of D. melanogaster (Fig. 4). A reduction in climbing ability when compared to their air-exposed controls was observed at both 99% CO 2 /1% O 2 and 98% CO 2 /2% O 2 mixes (P < 0.001), but was absent at 95% CO 2 /5% O 2 . Further reduction in CO 2 levels to 85% had no detrimental effect on climbing; however, when CO 2 levels were reduced to 80-65% [80% CO 2 /20% O 2 , 75% CO 2 /20% O 2 /5% N 2 , 70% CO 2 /20% O 2 /10% N 2 , 65% CO 2 /20% O 2 /15% N 2 ], climbing was reduced again (P < 0.01). Below the 65% CO 2 level, gas treatment no longer had any effect on climbing. Fly behaviour varied significantly over the range of CO 2 mixes used. At the 50% CO 2 level, the gas does not anaesthetize the flies; however, their movement slows considerably for as long as they are being exposed (Supplementary Movie 1). At higher levels of CO 2 (100-85%), the flies are anaesthetized completely, with   . Two different ranges of CO 2 levels adversely affect D. melanogaster climbing ability. Flies were exposed to varying levels of CO 2 for 10 minutes prior to a one hour recovery. The CO 2 mixes were initially offset by O 2 up to 20% and then with N 2 . Bars represent the mean ± SEM of the percentage of flies able to climb. The Air-exposed control bar represents the average of all controls performed for this experiment, while the statistical analysis (Student's t-test) was performed between the specific CO 2 level and their individual air-exposed flies. a = P < 0.01. CO 2 levels on anaesthetizing pads can affect D. melanogaster climbing. To better mimic laboratory conditions, a CO 2 pad was used to expose flies to different flow rates, in contrast to the precisely controlled exposure apparatus used above. Using this method, the minimal flow necessary to anaesthetize flies was determined empirically to be 2.4 L/minute. When flies were exposed to CO 2 for 10 minutes at a flow rate of 2.4 L/minute, and allowed to recover for one hour, there was no detrimental effect on climbing ability (Fig. 5). However, a minor increase in flow rate to 3.0 L/minute caused a 27% reduction in climbing ability (P < 0.001). This worsened to a 62% reduction in climbing when the CO 2 flow rate was increased to 5.0 L/minute (P < 0.001).
Flies recover from a 10 minute exposure to 99% CO 2 /1% O 2 significantly faster than to 100% CO 2 . When flies were exposed to a 99% CO 2 /1% O 2 mix in the exposure apparatus and allowed to recover for one hour, their climbing ability was reduced by 17% ( Fig. 6, P < 0.001). Unlike 100% CO 2 , the negative effect of 99% CO 2 /1% O 2 on climbing is absent by 16 hours. By increasing the 99% CO 2 /1% O 2 exposure time to 30 minutes, there is a greater reduction in climbing (39% at one hour, P < 0.001) that takes 24 hours to recover from.

Discussion
This study represents the first comprehensive analysis of the effects of CO 2 exposure and anoxia on D. melanogaster climbing and flight behaviours using precise gas mixtures. The results of the study reveal: 1) there is a chronic negative effect of a short exposure to 100% CO 2 on climbing (Fig 1a), which is potentially indefinite in length; 2) there is a long-lasting effect of 100% CO 2 exposure on flight for CO 2 exposures of 15 minutes or longer (Fig. 2); and 3) the mechanism behind these detrimental effects is a combination of a CO 2 -specific mechanism and an anoxic effect (Figs 1 and 3). Although there a number of studies that have shown negative effects of CO 2 on D. melanogaster, most notably on mating and reproduction 2,7 , negative long-term effects of CO 2 and anoxia on D. melanogaster climbing and flight have not been noted or observed before, to the best of our knowledge 6,16 . Indeed, it is thought that there . CO 2 flow rate through a CO 2 pad reduces D. melanogaster climbing ability in a flow ratedependent manner. Flies were exposed to varying levels of CO 2 on a CO 2 pad with varying flow rates and allowed to recover for one hour prior to being assayed for climbing. 2.4 L/minute is the minimum CO 2 flow rate found to anaesthetize flies and has no effect on subsequent climbing ability. However, increasing the flow to 3.0 and 5.0 L/minute decreased the ability to climb in a flow rate-dependent manner. Bars represent the mean ± SEM of the percentage of flies able to climb. The Air-exposed control bar represents the average of all controls performed for this experiment. Statistical analysis was performed by Student's ttests between the individual Air control flies and each CO 2 flow rate. a = P < 0.001 vs. the Air controls and b = P < 0.001 vs. the 3.0 L/minute CO 2 flies.
Scientific RepoRts | 5:15298 | DOi: 10.1038/srep15298 are no lasting effects from up to four hours of anoxia on D. melanogaster [17][18][19] , although some researchers have specifically avoided the use of CO 2 anaesthesia prior to performing assays for these behaviours 20 . Unfortunately, the majority of methods described for analysing climbing or flight either mention the use of CO 2 anaesthesia prior to performing the assays or make no mention of it at all [21][22][23][24][25][26][27][28][29] . When potential adverse effects of CO 2 anaesthesia are discussed in reference to these and other behaviours, a 24 hour waiting period has been suggested to allow for sufficient recovery 30,31 . The results from this study definitively show that flies do not recover normal climbing ability in 24 hours if they experience an exposure to 100% CO 2 for as little as five minutes, nor do they recover normal flight in 24 hours if exposed for 15 minutes (Fig. 2). In fact, a single 10 minute exposure of 100% CO 2 can affect climbing ability for up to 7 days post-exposure (Fig. 1). These findings have wide ranging impacts for the D. melanogaster research community when analysing not just climbing and flight behaviours, but any assay that requires coordinated movement, such as mating. In a broader scope, this study provides sufficient detailed analysis of significant lasting effects in D. melanogaster to warrant concern for using either CO 2 or anoxia prior to performing behavioural assays in insects. Therefore, the length of exposure to CO 2 should be minimized and avoided if possible. Additionally, for any study involving insects, it is essential to understand the impacts of CO 2 anaesthesia on the specific insect system being analysed before carrying out behavioural and physiological assays.
The mechanism behind the anaesthetic effect of CO 2 and its general physiological impacts is poorly understood. However, it has been shown that CO 2 directly inhibits glutamate receptors 12,32 , which function like the mammalian nicotinic receptors on motor end plates. This would be a direct mechanism to inhibit muscle activity in D. melanogaster. A less likely possibility could be the activation of the antennal CO 2 receptor, which is composed of the Gr21a and Gr63a chemosensory receptors 33 . However, these receptors are likely saturated at CO 2 levels much lower than those required for anaesthesia 34 . Neither of these mechanisms sufficiently explains the disparity between the effect of CO 2 on climbing and flight behaviour. However, the fact that the deficits last up to one week after exposure argues that long-term changes have occurred or damage has been done that is potentially irreversible.
While it appears that the negative effects on climbing caused by anoxia and CO 2 are through separate mechanisms, it is clear that anoxia is responsible for the long-lasting consequences observed with 100% CO 2 exposure, since flies exposed to 99% CO 2 /1% O 2 recover by 16 hours (Fig. 6). Climbing is unaffected by anoxia following a one hour recovery period (Fig. 3); however, this time point displays the strongest deficit with 100% CO 2 exposure (Fig. 1). This indicates that acute behavioural effects are likely due to a CO 2 -specific mechanism, while the anoxic effect takes longer to manifest. It appears that the anoxic effect cannot be produced by severe hypoxia, as 1% O 2 abolishes the detrimental effect on flight (Fig. 3). This is interesting because this level of hypoxia is extreme. While flies can survive short exposures to anoxia, flies cannot survive in 4% O 2 without conditioning 35,36 . Anoxia leads to the spiracles remaining open, which would increase tissue exposure to high CO 2 [37][38][39] . However, low (O 2 ) concentrations allow flies to continue to ventilate and close their spiracles, which would then mitigate exposure to CO 2 at the tissue level, thus allowing them to recover faster (Fig. 6). Anoxia contributes to tissue damage by increasing the mitochondrial production of reactive oxygen species [40][41][42] and there is strong evidence that low (O 2 ) leads to increased oxidative damage in insects 43,44 . These mechanisms could be driving the long-term effects and lack of recovery.  Figure 6. Exposure to 99% CO 2 reduces D. melanogaster climbing ability for an extended period of time in a dose-dependent manner. Flies that were exposed to 99% CO 2 /1% O 2 in the exposure apparatus had reduced climbing abilities at various recovery time points. Flies exposed for 10 minutes had reduced climbing through 8 hours, while the climbing of flies that were exposed for 30 minutes was reduced through 16 hours. The 30 minute exposure inhibited D. melanogaster climbing more than the 10 minute exposure. The Air-exposed points represent the average of all controls performed for that time point; however, the statistical analysis (Student's t-test) was performed with the individual control flies for each exposure length. The data points represent the mean ± SEM of the percentage of flies able to climb. a = P < 0.01 vs. Air control flies, b = P < 0.01 vs. 10 minute CO 2 flies.
The use of the exposure apparatus to treat multiple groups of flies evenly and precisely to gases is difficult to translate to a typical D. melanogaster laboratory. Flies are frequently anaesthetized by CO 2 pads that are composed of flat, porous surfaces that allow for diffusion of CO 2 . It is difficult to ascertain what CO 2 level flies are actually being exposed to in the boundary layer between the diffusing CO 2 and the environmental air on the CO 2 pad. However, it has been previously shown that the use of a CO 2 pad can alter response to anaesthetics in flies 8 . In order to relate the results from the precise exposure apparatus to actual practice in research laboratories, a CO 2 pad was incorporated into the study. 100% CO 2 delivered at a flow rate of 2.4 L/minute was the minimum level needed to anaesthetize flies on the pad. Typically, when the CO 2 flow rate through a CO 2 pad is insufficient to anaesthetize flies (i.e., < 2.4 L/minute), most users increase the flow enough to attain a more thorough and rapid anaesthetization. This likely leads to flow rates similar to 5.0 L/minute, or higher, which is closer to the 100% CO 2 data from the precisely controlled experiments (compare Figs 4 and 5). However, some users might only turn up the flow rate slightly (e.g., 3.0 L/minute), which is more approximate to the precise 99% CO 2 level. Given these similarities, minimal changes in flow rates will yield dramatically different impacts on fly behaviour. In most laboratories CO 2 flow rates are set by "ear, " that is, the flow is turned up until the CO 2 is heard to be flowing through the CO 2 pad. It is difficult to ascertain what flow rate this equates to. An 11 L/minute flow rate (the maximum the mass flow controllers allowed) was undetectable to our hearing, which is likely to be very close to a 100% CO 2 exposure. In respect to practices in D. melanogaster research laboratories, we suggest that modification of behavioural assays to account for this CO 2 effect is completely possible without any additional equipment for monitoring flow or mixing gases. Since the minimal flow necessary to anaesthetize D. melanogaster (2.4 L/minute) had no obvious negative effects on climbing, it is possible that if caution is exhibited, negative consequences can be avoided.
This study clearly demonstrates a vital need for D. melanogaster researchers to use CO 2 anaesthesia with greater caution. This has already been the standard of practice for research on mating behaviour, which has been shown to be affected by CO 2 2,45 . Our results encourage D. melanogaster researchers to take the same careful approach with any D. melanogaster behavioural assay. This care needs to be exercised in each individual laboratory, as the setups are unique and the CO 2 delivery to the fly might be very different. We suggest that each laboratory carefully take into account the amount of CO 2 that can be delivered to the fly without negative consequences on their behavioural assay. Or, conversely, determine the minimum time needed for the flies to recover from the effects of CO 2 exposure. Additionally, since there were marked differences in the negative effects of CO 2 exposure in flies exposed for five minutes versus 10 minutes (Fig. 2), it is also very important that control flies be exposed to the same amount of CO 2 as experimental flies. To work around this potential impact of CO 2 , many members of the fly community already use cold immobilisation as a method to avoid CO 2 exposure. However, this method has also been shown to affect some behaviours and physiology of D. melanogaster 2,46,47 . Therefore, we caution against the untested use of any form of anaesthesia or immobilisation in general.
This study shows that 100% CO 2 exposure has profound and long-lasting impacts on D. melanogaster climbing and flight. Furthermore, the mechanism behind this effect is a CO 2 -specific mechanism separate from an additional anoxic effect. In so doing, it conclusively demonstrates that both CO 2 and anoxia can markedly affect behavioural, and likely physiological, studies in D. melanogaster. We therefore advocate a more careful and considered use of any anaesthetic or immobilisation technique in respect to insect behavioural and physiological research. By monitoring the duration of exposure time, flow rate, or concentration of CO 2 the insects are exposed to, many of these impacts may be able to be minimized. Not only is this an issue for D. melanogaster, but it has the potential to be important to a wide variety of insect systems. Therefore, it would be worthwhile to test the effects of CO 2 anaesthesia more broadly across a wide range of species. Exposure to gas mixtures. Flies were exposed to a variety of gas mixtures using a custom-built exposure apparatus ( Supplementary Fig. 1), which distributes various gas mixtures rapidly and evenly to multiple vials of flies at set flow rates. The flies were kept in a polypropylene fly vial (Genesee Scientific) that had the bottom replaced with a stainless steel wire mesh (W. W. Grainger ® Inc., Lake Forest, IL; wire diameter of 0.012 inches and a mesh opening of 0.0213 inches) that was fused to the vial by heat. Flies were collected in groups of approximately 50 by weight (0.03-0.06 g). Four of these vials containing flies were connected in a series via silicone tubing (3/4 inch inner diameter, 1/8 inch wall thickness; United States Plastic Corp., Lima, OH). The series of vials were connected to a gas source regulated by mass flow controllers (Brooks ® Instruments models 5850i and 5841A, Tylan General ® model FC-2900v and Sable Systems International ® version 1.1 mass flow control electronics unit). For all experiments where flow was constant, flow was maintained at 3.0 L/minute. 100% CO 2 administered at this flow rate was found to cause rapid anaesthesia of the flies in the vials (Supplementary Movie 2). The time delay for anaesthesia to occur in the first vial to the last vial was < 2 seconds. The test gas mixture was administered to the flies for the length of time indicated and immediately flushed out of the exposure apparatus with compressed air at 3.0 L/minute for 10 minutes. The flies were then transferred to fresh food vials and maintained at 25 °C for the remainder of the recovery time. Control flies were treated in the same manner, except they were exposed to compressed air instead of the experimental gas mixes. All gas exposures were performed with an inline humidifier to prevent desiccation ( Supplementary Fig. 1).

Methods
Another set of experiments was carried out on a fly CO 2 anaesthetizing pad (FlyStuff Flypad with a flowbed frame, Genesee Scientific). The CO 2 pad replaced the vials in the exposure apparatus setup, so that flow, humidification and other variables could be controlled. The flies were counted on the pad and sorted into separate vials following the air flush. The compressed air flush was performed until the flies were visibly moving (typically < 10 minutes) on the pad and quickly transferred into their recovery food vial. Control flies were held against the CO 2 pad by an empty vial, since compressed air does not anaesthetize the flies.
Climbing assay. Flies were subjected to the experimental gases and allowed to recover for the various times and then subjected to a climbing assay previously described by Chambers et al. 48 . Groups of ten flies were placed in an empty climbing vial and then tapped down to the bottom. They were allowed 18 seconds to climb past a dotted line marked 5 cm from the bottom of the vial. The number of flies above the 5 cm mark at 18 seconds was recorded as a percentage of flies able to climb/vial. In all assays, flies were transferred to a new food vial the day before the climbing assay was performed to help reduce wet food from inhibiting their climbing ability. A minimum of 8 separate trials were run per condition.
Flight assay. The flight assay was performed as described in Chambers et al. 48 . Individual flies from a vial of approximately 10 flies were dropped into a clear acrylic flight box (28 cm × 28 cm × 28 cm) through a 3 cm entry hole in the center of the top. A fly was determined to be capable of flight if it maintained a steady elevation and flew in a controlled manner after being dropped. Data were collected as the percentage of flies able to fly/vial with at least 10 separate vials run per condition.
Statistical analysis. All statistical analysis was performed on the arcsin of the square root of the ratio of flies able to climb or fly. For the majority of analyses, Student's t-tests were performed between the compressed air control flies specific to the experimental group. In graphs where a single control group is plotted versus different experimental conditions, the plotted control group value represents the average of all control groups run at that time point/concentration/etc. Prior to this averaging, a one-way analysis of variance was performed on the transformed data from all the combined control groups to ensure that there were no differences between the control group values. However, the presented statistical analysis is always comparing the experimental group to its specific control. All statistics were run using Graphpad Software Prism ® Version 8 or Microsoft ® Excel 2010.