Metabolic function in Drosophila melanogaster in response to hypoxia and pure oxygen

Understanding how organisms respond to, and tolerate low O₂tensions has become an area of intense scientific study. Much of this interest is due to the recent identification of a widely conserved set of genes, termed hypoxia inducible factors (HIF), that sense and respond to hypoxia(Kaelin, 2005; Semenza, 2004). In addition to providing basic insights into how organisms cope with reduced levels of O₂, understanding the hypoxic response has important medical ramifications in the treatment of ischemic injuries induced by events such as stroke or cardiac failure, and in the treatment of cancer(Dayan et al., 2006; Hochachka et al., 2002; Murphy and Steenbergen, 2008; Pouyssegur et al., 2006; Yun et al., 2002). Studies using model organisms such as the nematode Caenorhabditis elegans and Drosophila melanogaster, have provided valuable insights into the underlying molecular and physiological responses to hypoxia(Bishop et al., 2004; Centanin et al., 2005; Chen et al., 2004; Epstein et al., 2001).

Oxygen is required for all known metazoans to complete their life cycle(Fenchel and Finlay, 1995; Jensen, 1995). Under aerobic conditions, animals utilize O₂ in the mitochondrial electron transport chain for oxidative phosphorylation and the synthesis of adenosine triphosphate (ATP) (Saraste,1999). Metazoans have evolved a variety of mechanisms to either supply O₂ to cells, or to compensate for reduced O₂levels when faced with fluctuations in O₂ levels. Factors that expose organisms to low O₂ tensions include environmental events,such as decreased O₂ partial pressures with increased altitude, or chemical reactions that remove or displace O₂ in aquatic and terrestrial environments (Baumgartl et al.,1994; Drew, 1992; Hoback and Stanley, 2001). Hypoxia can also occur at the cellular level during heavy exercise, and during physiological insults, such as strokes or heart attack, which limit perfusion(Hochachka, 1998). Although the requirement for O₂ is absolute, animals do have the ability to tolerate varying degrees and periods of hypoxia(Hochachka, 1980).

Organisms respond to reduced O₂ levels with adaptations that include increasing rates of glycolytic flux, reducing metabolic rate and modulating the levels of hypoxia inducible factors (HIFs) that regulate hypoxia-sensitive genes (Baumgartl et al.,1994; Brooks and Storey,1997; Chandel and Schumacker,2000; Clegg, 1997; Johansson et al., 1995). Even with such modifications O₂ levels can still become limiting to cellular function, forcing the organism to withstand the harmful effects of hypoxia or anoxia (Hand,1998). At the cellular level hypoxia can rapidly induce a wide range of potentially detrimental physiological effects that include alterations in levels of nicotinamide adenine dinucleotide hydride (NADH),adenosine diphosphate (ADP), adenosine monophosphate (AMP), inorganic phosphate, ATP and pH (Foe and Alberts,1985).

The deleterious effects of hypoxia are thought to be mediated mainly by reducing the rate of ATP production. This results in the rapid failure of ATP-driven ion pumps, which allows the unopposed flow of ions into the cell causing rapid membrane depolarization. The resulting unregulated Ca²⁺ influx causes a wide array of cellular disruptions including spontaneous stimulation of phospholipases, proteases and nucleases(Choi, 2005; Lutz and Nilsson, 1997; Murphy and Steenbergen, 2008; Storey and Storey, 1990; Thompson et al., 2006). Toxic effects of hypoxia also include neuron excitotoxicity, free radical damage,inflammation and immune system over-activation(Bickler et al., 2004).

These hypoxia-induced changes at the cellular level result in alterations in physiological function of the organism. At the level of the whole organism two well-studied indicators of the effects of hypoxia are the O₂tension at which metabolic rate begins to be reduced, and how well an organism can survive a complete lack of O₂. The point at which a reduction in metabolic rate occurs is commonly referred to as the critical oxygen tension (P_c). This value can vary from normoxic O₂ tensions of 21 kPa in frog muscle to 1 kPa or less in some invertebrates (Boutilier and St Pierre,2002; Holter and Spangenberg,1997; Taylor and Moore,1995). Whereas exposures to low O₂ levels are eventually lethal to all metazoans, the length of time that organisms can tolerate anoxic and/or hypoxic conditions varies enormously. A few minutes without O₂ can be lethal to most vertebrates(Arthur et al., 1997; Hermes-Lima and Zenteno-Savin,2002), whereas copepod eggs buried in anoxic lake sediments remained viable for up to 332 years(Hairston et al., 1995; Marcus et al., 1994). Although some animals can survive long periods of anoxia(Hand, 1998; Hand and Hardewig, 1996; Jackson, 2000; Lutz and Nilsson, 1997), the great majority of free-living, eukaryotic organisms cannot withstand more than 24 h of anoxia (Clegg,1997).

Elevated O₂ levels can also be toxic to organisms. Hyperoxia toxicity is thought to be caused by an increased production of free radicals and hydrogen peroxide generation (Joenje,1989; Paget et al.,1987). Among other effects, hyperoxia can disrupt metabolic pathways and reduce respiration rates(Gille and Joenje, 1992; Jamieson, 1989; Schoonen et al., 1990). Humans exposed to pure O₂ for more than 24 h begin to show compromised lung function and exposure to several days of pure O₂ is lethal to many mammals and insects (Mockett et al.,2001; Smith and Gottlieb,1975).

Insects have long been used to study respiration physiology and the response of organisms to varied O₂ tensions(Chadwick and Gilmour, 1940; Ellenby, 1953; Fenn et al., 1967; Harrison et al., 2006; Hetz and Bradley, 2005; Hoback and Stanley, 2001; Jarecki et al., 1999; Lighton and Schilman, 2007). A major advantage in the use of insects to study physiological responses to varying O₂ levels is that the tissues in insects are typically directly exposed to the ambient O₂ tensions. This contrasts with many animals (e.g. mammals) that rely on respiratory pigments and a circulatory system to meet gas exchange requirements. This means the O₂ tensions that cells are exposed to can be quite different from ambient levels, particularly at increased O₂ tensions(Massabuau, 2003).

The well-characterized development and genetics of Drosophila make it particularly useful in studies of hypoxia tolerance(Douglas et al., 2003; Haddad, 2006). As in most insects, gas exchange in Drosophila occurs via pairs of spiracles on the segments of the thorax and abdomen from where branching tracheae penetrate the tissues terminating in blind-ended tracheoles that deliver O₂ directly from the atmosphere to the organs and tissues. Ambient gases enter into Drosophila through spiracles located laterally along the thorax and abdomen. Small muscles control valves that regulate the flow of air in response to factors such as humidity and metabolic demand (Heymann and Lehmann,2006; Lehmann et al.,2000; Wilson et al.,2005). This study examined the metabolic response of D. melanogaster to a wide range of O₂ concentrations. As well as better characterizing the response of Drosophila to hypoxia, such studies allow changes in the physiology of the whole organism to be compared with studies on the effects of varied O₂ tensions on gene expression patterns.

The metabolic response of a common laboratory wild-type strain of Drosophila melanogaster Meigen (strain w¹¹¹⁸) to varying O₂ tensions was assayed in groups of approximately 100 mixed-sex, 4- to 8-day-old post-emergent flies. Previous studies have shown that male and female flies appear to respond equally to hypoxia in physiological and behavioral assays(Krishnan et al., 1997). Fly stocks were maintained on a standard yeast-cornmeal-agar medium at 24°C in 50 ml glass vials. The day before the flow-through respirometry measurements were made the flies were sedated with light CO₂ anesthesia and their sex determined. Flies recovered from this CO₂ exposure for 24 h before the start of metabolic measurements.

On the day of the metabolic measurement flies were immobilized with a humidified stream of N₂ and placed in two 25 ml glass metabolic chambers at a density of around 100 flies per chamber. Exposing flies to this short period of anoxia does not significantly affect their subsequent metabolic rate (Van Voorhies et al.,2004). The chambers were then flushed with CO₂-free,H₂O-saturated air at 20 ml min^–1 (STPD). CO₂ was removed from the inflowing gas stream with an Ascarite gas scrubbing column. Gas flow into the chambers was regulated with a mass flow controller (Sierra Instruments, Monterrey, CA, USA) and a rotameter (Gilmont Instruments, Barrington, IL, USA) directly calibrated against a mass flow controller before each experiment. To prevent the flies from being stressed by low humidity conditions, the gas stream was rehydrated before entering the metabolic chamber by passing it through a series of glass syringes filled with sterile H₂O and cotton wool. After flowing through the metabolic chamber, water was removed from the air stream with magnesium perchlorate filters before entering the CO₂ and O₂ analyzers. The water vapor content of the air stream entering the metabolic chamber was at essentially 100% RH (18.7 mg H₂O l^–1) and contained 1–2 p.p.m. CO₂ as assayed with a Li-Cor 6262 CO₂/H₂O analyzer (Lincoln, NE, USA).

The small amount of CO₂ produced by outgassing from the water-filled syringes was subtracted from the final metabolic reading. Metabolic chambers were maintained at room temperature (23±1°C) as monitored by a Hobo data logger (Bourne, MA, USA). At the end of the collection of metabolic data flies were frozen at –80°C and weighed on a Sartorius M2P microbalance (Sartorius AG, Göettingen, Germany). The weight of flies stored at –80°C is stable for >1 year(Van Voorhies et al.,2004).

Oxygen and carbon dioxide concentrations were analyzed at a 2 s sampling interval beginning 10 min after the flies were sealed in the chambers. Data were recorded using Sable Systems DATACAN data acquisition hardware and software (Sable Systems, Las Vegas, NV, USA). A group of ∼100 flies produced around a 0.02% CO₂ enrichment of the air stream under normoxic conditions. To normalize for the slight differences in the initial metabolic rates between the different groups of flies, the pre-hypoxic metabolic rate of each group was given a relative value of 1.00. The metabolic rates of two experimental groups of flies were measured simultaneously using a Li-Cor 6251, and 6262 analyzer operated independently of each other. The gas stream then fed into a dual channel Oxzilla fuel-cell O₂ analyzer(Sable Systems) to measure O₂ concentration. Data from the CO₂ and O₂ analyzers were lag corrected to compensate for the time required for the gas sample to flow from the metabolic chambers to the analyzers. The O₂ concentration of the inflowing airstream would have been reduced by the metabolic consumption of O₂ by the flies. However, even at their maximum metabolic rate a group of flies depleted the inflowing air by ∼0.02 kPa so this would have a minor effect on the O₂ level. The CO₂ gas analysis system was set to zero daily against CO₂-free air, and calibrated regularly with a 51 p.p.m. certified gas standard (Air Products, Long Beach, CA, USA). The O₂ analyzer was calibrated daily with well-mixed atmospheric air scrubbed of H₂O with a column of magnesium perchlorate.

After allowing the flies to equilibrate in the chambers for ∼1 h, a stream of N₂ gas (>99.9% purity) was mixed into a 2 l glass mixing flask leading into the metabolic chamber. Over the course of ∼2 h the O₂ level in the air stream leading to the experimental group was reduced from 19 to 2 kPa and after 4 h reached <0.1 kPa. Flies were then exposed to 45 min of anoxia by passing pure N₂ directly into the metabolic chambers. At the end of this period atmospheric air was introduced into the mixing flask. Data were collected from six independent groups with two groups measured on three separate occasions(N=6).

Each experiment also included two to four control groups of 15–20 flies that were exposed to normoxic conditions. Because the gas analyzers were in continuous use in monitoring the hypoxic groups, the CO₂production of the control groups could only be assayed immediately after the control flies were placed in the metabolic chambers and after the completion of the hypoxia recovery measurements. Since these flies had only a short period to equilibrate in the chambers before the first metabolic measurement was taken the initial metabolic rate data for this group probably overestimates the actual initial metabolic rate(Van Voorhies et al.,2004).

After being exposed to approximately 5 h of graded hypoxic conditions flies were reintroduced to normoxic conditions by adding room air into the mixing flask. The flies were then directly exposed to an increase in O₂concentration. This increased the P_O2 of the airstream to 2 kPa within 1 min and to 16 kPa within 3 min. The metabolic rate(calculated as the amount of CO₂ produced) of the flies was then recorded for an additional 4 h period. These data were used to plot the metabolic recovery of flies exposed to hypoxia.

Respiratory quotients (RQ) were determined for flies while exposed to hypoxia, during recovery from hypoxia, and in hyperoxia, using a Sable Systems PA-1 paramagnetic O₂ analyzer and a Li-Cor 6262 gas analyzer. For the RQ experiments flies were sealed in chambers with a defined predetermined O₂ concentration. The chambers were filled and resampled from cylinders containing pressurized mixes of air of defined O₂concentrations and scrubbed of CO₂. At the end of the sampling interval the chamber was flushed at a known rate with the same gas mixture that was used to fill the chamber. The bolus of CO₂enriched/O₂ depleted air was flushed into the CO₂ and O₂ analyzers located downstream. Gas flow into the chamber was regulated with a mass flow controller located upstream of the chamber. The accuracy of the system was assessed through injection of known volumes of a calibration gas standard with defined amounts of CO₂ and N₂ and by calculating the RQ generated from the combustion of pure ethanol. Based on these methods the CO₂ and O₂ analyzers gave readings within 2% of the predicted values.

To determine if metabolic substrate utilization was altered in hypoxia,O₂ consumption and CO₂ production were measured in flies progressively exposed to O₂ tensions of 1.2, 0.8, 0.4, 0.2 kPa. O₂ consumption and CO₂ production were initially sampled at 1 h intervals. Because the metabolic rate of the hypoxic flies was greatly reduced, the sampling intervals were increased to 1.5 h during extreme hypoxia. For these measurements an average of approximately 75, 5- to 6-day-old post-emergent male and female flies were placed in 25 ml glass metabolic chambers. Five experimental groups were exposed to hypoxia, and three control groups exposed only to normoxic conditions. The amount of CO₂ produced and O₂ consumed was calculated using DATACAN software.

To determine if exposure to hypoxia induced a large O₂ debt which was repaid during hypoxic recovery, RQ values were determined for flies after a 4 h exposure to 0.2 kPa O₂. RQs were calculated on data from five groups of flies both during the hypoxic exposure and after the flies were in normoxic conditions for 0.5 and 1.5 h. The RQs of two groups of control flies that remained in normoxic conditions were also calculated.

RQ values of flies exposed to 100% O₂ for 24 h were calculated for six groups of flies each with equal numbers of males and females flies. RQs were calculated for these groups after 4, 7, 11 and 26 h.

To determine if survival varied between flies exposed to different hypoxic conditions groups of flies were subjected to O₂ levels of 0, 0.1,0.2, 1.0 and 19 kPa (normal atmospheric O₂ concentration at the site of the experiments) for periods ranging from 2–16 h. Five separate groups of ∼20 mixed-sex, flies were used for each time point and O₂ tension. The gas stream leading into the fly chambers were hydrated with glass syringes filled with water-saturated cotton wool that were preflushed with 99.999% N₂ for 48 h to purge O₂. Gas mixtures were made by mixing N₂ with room air in a 20 l pressurized gas cylinder and assaying the final O₂ contents with a PA-1 O₂ analyzer. Gas mixes of different O₂ tensions entered the chambers at 20 ml min^–1 (STPD).

Control flies had a small amount of Drosophila instant food placed in the chamber with them. The other groups of flies were essentially immobilized during exposures and were incapable of feeding. For this reason no food was placed in these chambers. Previous experiments with this same strain of flies have shown that no mortality occurs in flies starved for 24 h. Chambers were kept at room temperature, which varied between 23 and 25°C as monitored with a Hobo data logger. Survival was assayed after allowing the hypoxia-exposed flies to recover in normoxia for up to 24 h.

To determine if multiple, rapid exposure to hypoxia altered metabolic function, groups of flies were immobilized with a stream of humidified N₂ at a flow of 1 l min^–1 (STPD) until there was no detectable fly movement. This occurred within 15–20 s of exposure to the stream of N₂. The chamber was then flushed with normoxic air at 100% RH until >90% of the flies were actively walking in the vial(typically ∼2 min) and then immediately exposed to another anoxic event. Five groups of 12–15 flies were exposed to five such anoxic events while five groups of flies were exposed to 10 bouts of anoxia and reperfusion.

For each experiment there were three groups of ∼15 control flies that were exposed to a single anoxic exposure equal in duration to the total time the cyclic groups were exposed. The time that the control groups were exposed to pure N₂ was 90 s for the groups exposed to five anoxic bouts and 180 s for groups exposed to 10 bouts. Flies exposed to cyclic anoxia were in a flowing gas stream for 11.5 to 23 min. This time included both the interval of the flushing with N₂ (1.5 or 3 min) plus the time that the chambers were flushed with normoxic air during the recovery period (10 or 20 min). To control for the potential effect of exposing flies to a moving air stream the flies in the control group were exposed to a normoxic flow of air of 1 l min^–1 for a period equal to that of the experimental group. Thus, both the experiment and control flies had equal time exposures to a flow of N₂ or a flow of air. At the end of the set of anoxic exposures the anoxic and control groups were sealed in 50 ml glass metabolic chambers with a small amount of food and placed in incubators at 24°C. The flies were allowed to recover in the chamber for 1 h before collecting CO₂ and O₂ data for the next 22 h at a 1 h sampling interval for the first 6 h and every 2 h thereafter. After 22 h,CO₂ production from the food was measured and subtracted from the output.

Hyperoxia measurements were conducted under conditions similar to the hypoxic experiment except that flies were supplied with a small amount of sterile food. To avoid the potential confounding effect of the metabolic activity of eggs and larvae laid by female flies only male flies were used. For these measurements flies were immobilized with N₂ and placed in the metabolic chamber. After allowing the flies to equilibrate for 1 h in the chamber the metabolic rate of the flies in normoxia was recorded for 1.5 h to obtain a control metabolic rate. Gaseous 100% O₂ was mixed into a 2-l flask and then into 25 ml glass metabolic chambers containing an average of 20, 5-day-old male flies. The same flow rates, sampling intervals and humidification methods were used in the hyperoxia conditions as used in the hypoxia experiments.

The metabolic rate of the fly groups in 100% O₂ was recorded for a 24 h period. The flies were then exposed to normoxic conditions and their metabolic rate was recorded for another 2 h. The CO₂ output from the food was measured at the end of each experiment and subtracted from the final output to determine the final metabolic reading. The CO₂contribution from the food was always a minor portion (<2%) of the total CO₂ output. The CO₂ production of the control flies continually maintained under normoxic conditions was measured several times during each experiment. Data were collected from measurements done on five separate occasions. Fly mortality was assayed several times during each experiment.

To determine if the use of CO₂ to immobilize flies had a measurable effect on metabolic rate, groups of 20 flies were placed in 12 separate metabolic chambers divided between three groups: control flies, flies to be knocked out with CO₂, and flies to be knocked out with N₂. A baseline reading was taken after the flies were in the chambers for 20 min and then the chambers of the experimental groups were flushed with 100% humidified CO₂ or N₂ for 5 min at a flow of 200 ml min^–1 and then the chambers were sealed for 25 min. After being exposed to pure N₂ or CO₂ for ∼30 min the chambers were flushed with normoxic air for 5 min. Three additional metabolic rate measurements were then taken of each group at 30 min intervals. To determine if high CO₂ levels would immobilize flies in the presence of normoxic O₂ levels a group of mixed-sex flies was exposed to a mixture of 20 kPa O₂ and 80 kPa CO₂.

An ANOVA analysis (Statview 5.1) was used to determine at what O₂ level the metabolic rate was significantly reduced relative to normoxic levels. As O₂ tensions dropped below 0.7 kPa the rate of reduction in metabolic rate relative to the reduction in O₂ tension appeared to increase. To quantify this effect, the slopes were compared between the change in metabolic rate and O₂ tension, for O₂ tensions between 0.1–0.6 kPa and 0.7–3.1 kPa with a Student's t-test (Statview 5.1). Student's t-tests were also used to analyze for differences in RQ between flies in hypoxic or normoxic conditions and to compare the early metabolic recovery data from flies exposed to repeated hypoxic and reperfusion events. The data were analyzed for differences in survival with ANOVA. To determine if exposure to cyclic anoxia influenced the subsequent metabolic rate, data were analyzed for differences in metabolic rate using repeated-measures ANOVA (Statview 5.1).

The metabolic rate of flies exposed to graded hypoxia remained relatively unchanged until an O₂ tension of ∼3.1 kPa (P=0.015, F=2.6, d.f.=9), whereupon it decreased in a approximately linear manner until the O₂ partial pressure reached 0.6 kPa(Fig. 1A). The rate of metabolic depression relative to the decrease in O₂ tension was more pronounced as O₂ tensions were reduced from 0.6 kPa to anoxia(Fig. 1B). The average slope of the line between these O₂ tensions increased more than fourfold relative to the slope of the line between 0.7 and 3.1 kPa, a highly significant difference(P0.001, t=11.1, d.f.=10). The metabolic rates of the flies in extreme hypoxia are at least an order of magnitude lower than under normoxic conditions. The O₂ tension at which fly metabolic rate was reduced to 50% of normoxic levels was 1.0 kPa.

Flies exposed to hypoxia followed a characteristic set of behaviors. They initially rested quietly on the sides or ends of the chamber. At an O₂ tension of approximately 3 kPa the flies became more agitated and began actively walking around the chambers. As the O₂ tension dropped below around 2 kPa the flies moved to the bottom of the chamber where they stood with greatly reduced movement. Fly movement decreased even more as the O₂ tension dropped to around 1.4 kPa but the flies still maintained an upright position. At O₂ tensions below ∼0.8 kPa all of the flies were completely immobile and were lying on their sides. All flies survived exposure to several hours of hypoxic conditions and >99% of the flies were alive 15 h after the end of the exposure to hypoxia.

The metabolic rate of flies exposed to 5 h of graded hypoxia, including 45 min of anoxia, quickly returned to near pre-hypoxic levels after reintroduction of normoxic conditions (Fig. 1C). Within 5 min of being exposed to normoxia the metabolic rate of the flies increased almost 10-fold compared with their lowest hypoxic value. Despite the rapid recovery in metabolic rate, it typically took at least 30 min of normoxic flow before the first flies began to move and close to an hour for all the flies in a group to recover. During this recovery period the metabolic rate of the flies was around 50% higher than flies in 1.4 kPa O₂, an O₂ tension at which flies are still capable of activity.

The RQ of flies in normoxia was near 0.95, a value consistent with other measurements of Drosophila RQ(Van Voorhies et al., 2004). The RQ of flies exposed to O₂ tensions between 1.2 and 0.2 kPa increased to between 1.11 and 1.39 (Fig. 1D). Such increases in RQ are probably caused by a hypoxia-induced increase in anaerobic metabolism. The RQ of flies in 0.2 kPa O₂ was significantly higher than flies in normoxia (P=0.003, t=4.15, d.f.=8). Because the RQ values were higher in more extreme hypoxia, the use of CO₂ to measure metabolic rate is a conservative estimation of the metabolic depression that occurs in very hypoxic conditions.

The post-hypoxic RQ of flies during the first 30 min of the recovery period averaged 0.91 [±0.01 (s.e.m.) N=5]. The RQ increased slightly,but significantly (P=0.004, t=4.0, d.f.=8), during the 60–90 min recovery period to 0.98 (±0.02, N=5). The RQ of the two control groups was 0.87 and 0.90 for these two time periods. The RQ of flies in 100% O₂ for 24 h was constant [0.96±0.01 (mean± s.e.m.) N=24] over the four sample intervals. The maintenance of this high RQ value over this period indicates that the flies were feeding during the entire time they were exposed to 100% O₂. There was no difference in RQ between the male and female flies.

All flies were dead after a 16 h exposure to anoxia(Fig. 2). Flies survived slightly better at 1 kPa but still did far worse than normoxic control flies. There was generally a direct correlation between the O₂ tension and survival. Survival of flies remained high (>75%) in hypoxia, up to 4 h exposure, but dropped rapidly with longer exposures. Around 10% of the flies in 1 kPa O₂ group showed slight movement (e.g. twitching), whereas all the other groups of flies were immobile within a few minutes of exposure to the lowered O₂ tensions. Flies were much slower to recover from hypoxia of longer than 2 h and often were not moving after 1 h in normoxia.

Exposure to multiple hypoxic and anoxic events did not have a significant long-term effect on metabolic rate or respiratory quotients, although it did cause a short-term depression in metabolic rate(Fig. 3A,B). Metabolic rate varied significantly over the sampling interval for groups exposed to five and 10 anoxic bouts (P<0.001 for the 5× anoxic group and 10× hypoxic groups). With the exception of the first sample, there were no significant differences between the metabolic rates of the control groups and the cyclic anoxic groups for each time point (P=0.99 for the 5× anoxic group and P=0.13 for the 10× hypoxic groups). The initial metabolic rate of the group exposed to 10 cyclic anoxic events was reduced significantly compared with the control groups for the first reading(P=0.05, t=2.41, d.f.=6). There were no deaths in either the control or cyclic groups over 24 h, and feeding state of the groups recovering from the cyclic hypoxia treatment did not appear altered as the RQ remained near 1.00.

There were essentially no differences in metabolic rate between the normoxic group and groups exposed to 100% O₂(Fig. 4). All flies in both the hyperoxia and control groups were alive at the end of 8 h and after 24 h 97%of the flies in hyperoxia and control groups were alive. Two of the 100%O₂ groups had ∼15% mortality after 24 h. To avoid the potentially complicating problem of this high mortality rate these groups were excluded from the analysis.

At the end of the sample period the flies in all of the groups were alive. After 24 h, 99% of the control group was alive, and 90% of the N₂and CO₂ groups were alive. The metabolic rate of flies exposed to pure N₂ remained constant over the entire sample period relative to the starting metabolic rate. Flies exposed to pure CO₂ for 30 min did not show a subsequent reduction in metabolic rate over a 2 h period(Fig. 5), and if anything the metabolic rate of this group was elevated, but not significantly, compared with the control group (P=0.06, t=3.1, d.f.=3). Flies exposed to an atmosphere of 20% O₂ and 80% CO₂ were completely immobilized within 1 min of exposure to the gas mixture.

D. melanogaster maintains a near constant metabolic rate as O₂ tensions were reduced from normoxic levels to approximately 3 kPa – an O₂ tension equivalent to being at an altitude twice that of the summit of Mount Everest(Denny, 1993). Maintaining normal metabolic function at this low O₂ tension is remarkable by mammalian standards. However, several factors, including having low resting metabolic rates relative to maximal metabolic rates and the presence of highly efficient gas exchange systems, pre-adapt insects to low critical O₂ tensions (Greenlee and Harrison, 1998).

The critical O₂ tensions (P_c) in resting D. melanogaster would be expected to be low since the metabolic rate of flies at rest or walking is a small fraction of flight metabolic rate. Flight increases metabolic rate approximately tenfold in D. melanogaster (Lehmann et al.,2000) and can increase it by >50-fold in other insects(Harrison et al., 2001). As such, the high apparent safety margin for O₂ delivery in resting insects may be greatly reduced or nonexistent during flight(Harrison et al., 2001; Harrison et al., 2006; Heymann and Lehmann, 2006). Consistent with this prediction, D. repleta does not fly at O₂ tensions below 6 kPa and flight activity is reduced in D. melanogaster placed in an altitudinal equivalent of 7 kPa O₂(Chadwick and Gilmour, 1940; Dillon and Frazier, 2006).

The critical O₂ tension of an organism provides a useful estimate of the point at which its metabolic function may be limited by rates of gas exchange, and to predict the hypoxia tolerance of an organism(Herreid, 1980; Holter and Spangenberg, 1997; Loudon, 1989). P_c for insects typically range between 1–5 kPa. Higher P_c values are more common in larger insects and are probably a result of increased diffusional distances(Loudon, 1988). Previous studies have reported P_cs for Drosophila ranging from 1.6 to 2.8 kPa (Greenlee and Harrison, 1998; Loudon,1988). Many invertebrates respond to hypoxia by reducing metabolic rates, but in mammals a reduction in metabolic rate in response to hypoxia only occurs in smaller species such as rats(Korducki et al., 1994).

Although measures of an organism's P_c are valuable,this value does not have a precise set-point(Frazier et al., 2001). Factors such as experimental and growth temperature, activity level and developmental stage all affect P_c values(Herreid, 1980; Lighton, 2007; Morris and Taylor, 1985). It also is apparent that insects reared at reduced O₂ tensions well above levels that reduce metabolic rate can still show physiological alterations compared with those reared in normoxia. Placing D. melanogaster embryos at O₂ tensions of 5 kPa increases the expression level of hypoxia inducible factors(Teodoro and O'Farrell, 2003). Although the P_c for adult D. melanogaster is around 3 kPa, flies reared at 10 kPa had reduced growth rates, smaller adult body size, and higher mortality than normoxic counterparts(Frazier et al., 2001). Similarly, beetle larvae (Tebebrio molitor) reared at 11 kPa O₂, an O₂ tension above its P_c, grew more slowly, had increased mortality and an altered sex ratio than larvae reared at 15 or 21 kPa O₂(Loudon, 1988).

As O₂ tensions decreased below 3 kPa the metabolic rate of D. melanogaster showed an approximately linear rate of decrease until∼0.7 kPa. Below this point there was a sharp decrease in the metabolic rate relative to the reduction in O₂ tension. If the only factor limiting respiration in hypoxia was the rate of gas diffusion, the rate of decrease in metabolic rate should remain linear as O₂ levels decrease. This suggests that the more rapid decrease in metabolic rate observed in very hypoxic conditions may be due to flies actively reducing metabolic rate in response to hypoxia. Hypoxia causes numerous changes in mitochondrial function that include opening of mitochondrial ion channels, and a reduction in mitochondrial complex IV cytochrome oxidase activity. Factors such as these may be responsible for the increased depression of metabolic rate at very low O₂ levels(Lahiri et al., 2005). Also there are proteins in D. melanogaster that appear to reduce metabolism under conditions of environmental stress(Teleman et al., 2005), and it is possible that such proteins could become more active under extreme hypoxia.

It also appears that the hypoxic response in D. melanogasterdiffers from that caused by the loss of cellular energy reserves caused by metabolic inhibitors such as cyanide(Teodoro and O'Farrell, 2003). This implies that the physiological response to hypoxia is not a simple passive consequence of the collapse of cellular energy reserves due to insufficient levels of O₂ to carry out oxidative phosphorylation. Regardless of the reasons, such a downregulation in metabolic rate should be adaptive for surviving hypoxia since suppression of energy turnover provides the greatest protection against hypoxia, and is critical to long-term anoxic survival (Arthur et al., 1997; Brooks and Storey, 1997; Hand, 1998; Hochachka et al., 1996).

The RQ of flies exposed to O₂ tensions of ≤1.2 kPa increased to between 1.11 and 1.39, and is consistent with results from other animals exposed to hypoxia. For example, the RQ of rats (Rattus norvegicus)in hypoxia increased from 0.75 to 1.15(Frappell et al., 1995) and the RQ of the leafroller moth (Platynota stultana) increased from around 0.75 in normoxia, to 1.3 in 1–2 kPa O₂(Zhou et al., 2001). The most probable cause for this increase in RQ is an increased use of anaerobic metabolism in hypoxic conditions, but other factors, such as changes in acid–base balance, could also be responsible for these increased RQ values (Greenlee and Harrison,1998). Although the rate of anaerobic metabolism may increase in flies exposed to hypoxia, this presumed increase in the rate of anaerobic metabolism was insufficient for the flies to maintain movement.

In general, D. melanogaster flies tolerated hours of exposure to hypoxic and/or anoxic conditions with no apparent adverse effects. Unlike mammals, insects have the ability to spontaneously recover from extended periods of anoxia (Kolsch et al.,2002). The metabolic rate of hypoxia-exposed flies recovered to near pre-hypoxic levels within ∼1 h of being returned to normoxic conditions, even though exposure to hypoxic conditions reduced metabolic rates by more than an order of magnitude compared with normoxic levels. D. melanogaster can rapidly replenish ATP levels depleted by hypoxia. The ATP levels of D. melanogaster embryos exposed to 30 min of extreme hypoxia (<0.1 kPa O₂) declined to 38% of control levels, but recovered to 86% of pre-hypoxic treatment within 5 min of being placed in normoxia (DiGregorio et al.,2001). Other investigators have found that D. melanogaster recovers from anoxia as a non-linear function of exposure time, with >2 h of anoxia causing a disproportionate increase in recovery time (Krishnan et al.,1997).

There was no indication of either a large metabolic overshoot or a major alteration in RQ in flies recovering from hypoxia. The metabolic response of D. melanogaster during hypoxia recovery is consistent with that of many invertebrates that do not show an apparent O₂ debt during recovery from hypoxia. The lack of an alteration in the RQ of D. melanogaster is also consistent with the general observation that insects make minimal use of anaerobic metabolism when exposed to hypoxia(Kolsch et al., 2002; Wegener and Moratzky, 1995). The increase in O₂ consumption seen in some animals after exposure to hypoxia or anoxic is primarily due to increased energy demands for the removal of anaerobic end products and recharging of the phosphagen and ATP pools (Ellington, 1983).

The factors that allow D. melanogaster to survive several hours of exposure to hypoxic and/or anoxic conditions are not clear. Insects possess very efficient respiratory systems and appear to seldom function anaerobically(Wegener, 1996). An argument could be made that since insects are rarely exposed to hypoxic conditions at the cellular level, they would be poorly adapted to survive severe hypoxia. This pattern is seen in organisms such as birds and mammals that rely on aerobic metabolism for the bulk of their metabolic demands. These animals can suffer irreparable damage after more than a few minutes of anoxia(Chapman et al., 2002; Hermes-Lima and Zenteno-Savin,2002).

A general requirement for surviving hypoxia is having both a low initial metabolic rate and the ability to greatly reduce metabolic rate in hypoxia. Turtles, Chrysemys picta, are an example of this. Anoxic survival is temperature dependent with turtles surviving for 12 h at 20°C, but 90 days at 3°C. This difference in survival is thought to be mediated by a temperature-induced reduction in metabolic rate at lower temperature(Jackson, 2000). In contrast to the low metabolic rate of turtles, the mass-specific metabolic rate of non-flying D. melanogaster is well above the maximal metabolic rate recorded in human elite athletes (Coyle,2005), leading to the prediction that they would have a limited ability to survive hypoxia.

The response of Drosophila to anoxia is similar to mammals, in that both show a rapid loss of muscle control and are quickly immobilized. The difference is that Drosophila can survive in this state for hours whereas mammals suffer irreversible damage after a few minutes of anoxia(Gu and Haddad, 1999). Among the factors potentially responsible for the survival of Drosophila in hypoxia is the ability of Drosophila neurons (unlike mammals) to become hyperpolarized in hypoxia. This reduces neuron excitability and may save energy (Gu and Haddad,1999).

Although D. melanogaster is very tolerant of short-term hypoxia,exposure to hypoxic conditions for periods longer than a few hours rapidly increased mortality. There was typically a linear relationship between O₂ tension and survival. A similar result has been reported for D. melanogaster embryos, with embryos dying more quickly in 1 kPa than 2 kPa O₂ (DiGregorio et al., 2001). Anoxic conditions were the most stressful, and survival of this group was reduced compared with the hypoxic groups for all but the 2 h exposures. This contrasts with survival of hypoxic conditions in the nematode Caenorhabditis elegans in which embryos exposed to anoxia survive longer than embryos exposed to hypoxia(Nystul and Roth, 2004). The ability to withstand hypoxic and/or anoxic conditions also varies with the developmental state. Adult flies exposed to anoxia do not survive >12 h,but embryos can survive at least 36 h of anoxia with nearly 100% survival(DiGregorio et al., 2001; Foe and Alberts, 1985).

The survival of flies in hypoxia decreased rapidly after >4 h exposure,a finding reported by other investigators(Chen et al., 2002). This increase in mortality may be caused by large increases in protein aggregation which increased over 4-fold after 4 h of anoxia(Chen et al., 2002). The survival of D. melanogaster in hypoxia is comparable with that of other insects. Hypoxia and anoxia tolerance in insects varied widely from an LT₅₀ of 1.5 h for the yellow-fever mosquito (Adedes ageypti) to 36 h for a flour beetle (Tribolium confusum)(Knipling et al., 1961). Locusts (Locusta migratoria) survive up to 4 h of anoxia, and hawk moths (Manduca sexta) up to 24 h(Wegener and Moratzky,1995).

With the exception of the first metabolic reading, no significant differences were apparent between the recovery metabolic rate or RQ of flies exposed to multiple bouts of anoxia, and exposure to a single anoxic episode. This result is unexpected because it has been well demonstrated that fluctuating hypoxia can increase production of mitochondrial reactive oxygen species (ROS) by up to 10-fold (Chandel and Schumacker, 2000; Drew et al., 2002; Guzy et al.,2005; Mansfield et al.,2005; Michiels et al.,2002). These ROS bursts produce DNA damage that can lead to mitochondrial apoptosis (Wyllie,1997). Additionally in mammals, hypoxic-reperfusion events increase levels of neuronal nitric oxide synthase and generation of nitric oxide, which causes mitochondrial dysfunction(Vannucci and Hagberg, 2004). Because ROS formation can only occur in hypoxic, not anoxic conditions, once a cell is anoxic formation of ROS should be minimal, and cycling in and out of anoxia should be far more damaging than a single anoxic–reperfusion event.

Although there was no evidence from this study that anoxic–reperfusion events had a long-term effect on metabolic function,a study by Lighton and Schilman (Lighton and Schilman, 2007) found that O₂ reperfusion events did cause a significant reduction in recovery metabolic rates in D. melanogaster. The factor that appears responsible for the difference between these two studies is the amount of time that the flies were exposed to anoxia. In this study the flies were in anoxic conditions for between 1.5 and 3 min whereas in the study by Lighton and Schilman flies were exposed to anoxia for 7.5 to 120 min before reperfusion occurred. The shorter anoxic exposure time used this study was probably below the level that would induce mitochondrial damage and cause a reduction in metabolic rate. It is also worth noting that in this study the group of flies exposed to 10 cycles of anoxic reperfusion did have a significantly reduced metabolic rate compared with the control group for the first recovery metabolic rate.

Flies exposed to an environment of 100% O₂ for 24 h showed no apparent alteration in metabolic rate or RQ relative to flies in normoxic conditions. This result is unexpected. Studies in mammalian cells in culture have found that hyperoxia can inactivate many enzymes crucial to normal metabolic function (Gardner et al.,1994; Joenje,1989). Functional consequences of such changes include protein misfolding, catalytic inactivation, the loss of protein functions and a reduction in the function of glycolysis and the tricarboxylic (TCA) acid cycle(Das et al., 2001; Schoonen et al., 1990; Yan et al., 1997; Yan and Sohal, 1998). Hyperoxia can also increase rates of oxidation of pyridine nucleotides (NAD and NADH), significantly increasing ROS production from mitochondria, as well as damage to mitochondria (Arthur et al.,1997; Gille and Joenje,1992; Wispe et al.,1992). Many of these deleterious effects occur within a short time of exposure to hyperoxia (Das et al.,2001). Studies in insects have found that hyperoxia can cause a profound depression in metabolic rate. For example, the metabolic rate of army worm pupae (Prodenia eridania) exposed to hyperbaric hyperoxia was reduced 20- to 50-fold relative to normoxic controls(Clark and Cristofalo, 1960). Mammalian cell cultures exposed to hyperoxia also show large (approximately threefold) decreases in metabolic rate compared with normoxic cells(Schoonen et al., 1990).

It has generally been assumed that insects use tracheae and tracheoles to distribute gasses directly to cells. As a result, O₂ tensions at the cell surface should closely mirror ambient O₂ tensions(Miquel et al., 1975; Mockett et al., 1999). However, one method by which flies could withstand a hyperoxia insult would be to limit spiracle opening to reduce gas flow(Hetz and Bradley, 2005). Several lines of evidence argue that this is not occurring in adult Drosophila exposed to hyperoxia. Many studies have used both direct microscopic examination and biochemical assays to demonstrate damage to many different internal cells types in Drosophila exposed to elevated O₂ tensions (Miquel et al.,1975; Philpott et al.,1974; Yan et al.,1997; Yan and Sohal,1998). The deleterious effects of high O₂ levels can be seen after a few hours of exposure, and the mortality rate of Drosophila increases significantly after a 24 h exposure to hyperoxia(Das et al., 2001). Additionally, the high metabolic rate of D. melanogaster, even at rest, predicts that it should continually keep it spiracles at least partially opened to avoid a rapid build-up of CO₂(Heymann and Lehmann, 2006; Lighton and Schilman, 2007). These factors do not support the hypothesis that flies are able to maintain low internal O₂ levels even in the face of high external O₂ tensions.

The RQ of flies exposed to 100% O₂ remained near 0.95 over the 24 h period of the measurements, which is essentially the same as normoxic controls. Although there are examples of hyperoxia inhibiting carbohydrate metabolism (Haugaard, 1968),the maintenance of this high RQ indicates that the flies were actively feeding and using carbohydrates as their main metabolic substrate. It is not clear how D. melanogaster sustains the damage that is expected to be caused by hyperoxia and still maintains a normal metabolic rate. It is possible that reduction in metabolic rate is not seen because the standard metabolic rate of the flies is much less than flight metabolic rate.

Exposing D. melanogaster for 30 min to either pure N₂or CO₂ had relatively minor effects on the metabolic rate of the flies during recovery. The limited effect of N₂ exposure on the subsequent metabolic rate of flies is consistent with data showing that the metabolic rate of flies exposed to graded hypoxia quickly recovers to pre-exposure levels when placed in normoxic conditions(Van Voorhies et al., 2004)(and this study).

It is more surprising that a 30 min exposure to pure CO₂ did not cause a subsequent reduction in metabolic rate. Unlike N₂, the anesthetic effect of CO₂ is caused by factors other than CO₂ simply displacing O₂ and producing anoxic conditions(Badre et al., 2005). D. melanogaster exposed to a mixture of 20% O₂ and 80%CO₂ were quickly immobilized, even though O₂ tensions were far higher than required for normal movement. Additionally, exposing Drosophila to elevated CO₂ levels has many effects including influencing gene expression levels, reducing longevity and fecundity, and affecting mating behavior(Barron, 2000; Leenders and Beckers, 1972; Perron et al., 1972).

Studies in other insects have shown that even low levels of CO₂can also have large effects on insect respiration. Honeybees (Apis mellifera) exposed to CO₂ levels of a few percent showed a significant inhibition of succinic dehydrogenase, a critical enzyme of the Krebs TCA cycle (Seeley,1960). The metabolic rate of lepidopteran pupae (Platynota stultana) placed in an atmosphere containing >20% CO₂ was reduced up to 80% compared with CO₂-free air, even when 21%O₂ was present (Zhou et al.,2001). In view of the potent effects of CO₂ on the physiology of flies it could be reasonably predicted that exposure to pure CO₂ would have some effect on the metabolic rate of flies recovering from this exposure. However, at least in the short term, no such effect was detected.

A notable feature of this study is the quantification of the remarkable ability of D. melanogaster to maintain or recovery normal metabolic function in the face of numerous environmental insults such as exposure over many hours to anoxic, hypoxic, pure O₂ or CO₂environments and multiple episodes of hypoxia–normoxic reperfusion. Exposing most mammals, including humans, to such conditions would result in death or greatly compromise metabolic function.