Balanced mitochondrial function at low temperature is linked to cold adaptation in Drosophila species Free

Most insects have limited capacity for physiological heat production and their body temperature is therefore primarily determined by the surrounding environmental conditions (Woods et al., 2015). As all biochemical and physiological processes are temperature dependent, it follows that environmental temperature will exert a strong influence on the physiological rates of insects, their performance, and ultimately their survival should temperatures become sufficiently extreme (Cossins and Bowler, 1987; Harrison et al., 2012; Hochachka and Somero, 2002). Given the overarching influence of temperature on physiological processes, it is not surprising that interspecific variation in thermal tolerance is a strong determinant of the biogeographical distribution of insects (Addo-Bediako et al., 2000; Bishop et al., 2017; Sunday et al., 2019). Limits for cold tolerance have been found to correlate particularly well with distribution in insects including the genus Drosophila, which represents more than 1400 chill-susceptible insect species distributed across tropical, temperate and subarctic climates (Kellermann et al., 2012; Kimura, 2004).

The cold biology of chill-susceptible insects is determined by their capacity to maintain physiological homeostasis when temperatures fluctuate seasonally, diurnally or during cold spells, and cold mortality in these species is therefore not related to ice formation (MacMillan and Sinclair, 2011a; Nedved, 2000; Overgaard and MacMillan, 2017). Cold tolerance of chill-susceptible insects varies with acclimation (Weaving et al., 2022) and adaptation (Kellermann et al., 2012; Kimura, 2004), and it is well documented that chill tolerance is closely linked to the insect's ability to maintain ion and water balance of their extracellular fluids (Koštál et al., 2004; Overgaard and MacMillan, 2017; Overgaard et al., 2021). Accordingly, chill-susceptible insects enter a state of neuromuscular paralysis (chill coma or critical thermal minimum, CT_min) at a temperature where they are unable to maintain extracellular K⁺ balance within the central nervous system (Andersen et al., 2018; Armstrong et al., 2012), and chronic cold injury occurs when low temperature impairs osmoregulatory capacity to a degree that causes hemolymph hyperkalemia (Koštál et al., 2006; MacMillan and Sinclair, 2011b; Overgaard et al., 2021).

Because chill tolerance relies on the ability of insects to balance active and passive ion transport, it follows that the ability to fuel active transport at low temperature could represent an important cold adaptation. CT_min is seemingly unaffected by ambient oxygen availability at low temperature (Boardman et al., 2016; Stevens et al., 2010), and cold exposure, even below CT_min, is usually not associated with a decrease in total ATP concentration in the whole animal (Colinet, 2011; Williams et al., 2018), muscle (MacMillan et al., 2012) or the fat body (Koštál et al., 2004). Nevertheless, it has been discussed whether mitochondrial capacity including the electron transport system (ETS) and associated oxidative phosphorylation is compromised by low temperature in insects (Colinet et al., 2017; Lubawy et al., 2022; Menail et al., 2022; Wood and Nordin, 1980).

All biochemical rates are affected by temperature as a result of thermodynamic molecular interactions, and the thermal relationship is often described using the thermal sensitivity quotient Q₁₀, which describes the change in rate with a temperature difference of 10°C (often Q₁₀≈2–3, i.e. the rate is 2–3 times lower with a 10°C decrease) (Arrhenius, 1915; Schmidt-Nielsen, 1997). In accordance, exposure of Drosophila melanogaster (whole body) mitochondria to 4°C acutely decreases oxygen consumption rate (OCR) relative to that measured at 25°C in both control and cold-acclimated flies, and chronic exposure to 4°C further decreases the capacity for mitochondrial oxygen consumption and ATP synthesis (Colinet et al., 2017). However, the mitochondrial ATP/O ratio (ATP produced per molecule of oxygen consumed) and respiratory control ratio (RCR – oxygen consumption associated with oxidative phosphorylation versus proton leak) were only moderately affected by acute and chronic cold; cold-acclimated flies tended to slightly increase both ratios over time spent at 4°C, while no significant changes were observed in the control flies. These findings suggest that the effects of cold stress on mitochondrial capacity were caused by mitochondrial breakdown rather than altered function of the isolated mitochondria (Colinet et al., 2017). Nevertheless, the same study found a tendency for cold-acclimated D. melanogaster to maintain ATP/O and RCR better during chronic cold. Similar tendencies were reported for cold-acclimated mayfly larvae (Drunella spp.), which retain relatively higher respiration capacity and RCR at low temperature compared with their warm-acclimated conspecifics (Havird et al., 2020). Cold-induced changes in insect mitochondrial capacity and coupling could also be associated with unbalanced temperature effects on different complexes in the electron transport system (Havird et al., 2020; Menail et al., 2022). For example, Menail et al. (2022) found that at 18°C, a non-stressful temperature, mitochondrial OCR is strongly stimulated by the addition of complex I substrates (pyruvate and malate; CI-OXPHOS) in both D. melanogaster and honeybees Apis mellifera carnica, but oxygen consumption does not increase significantly after the addition of alternative substrates [proline, succinate and glycerol 3-phosphate (G3P)]. At low temperatures (6 and 12°C), CI-OXPHOS was significantly reduced, but when alternative substrates were added they compensated for the decreased CI-OXPHOS and the total OCR was partially restored, suggesting that low temperature acutely alters the relative importance of ETS complexes (Menail et al., 2022). The findings discussed above suggest that insect mitochondrial function is modified or challenged by low temperature, but at present there is very limited knowledge of how cold adaptation is manifested at the functional level in mitochondria.

To investigate whether cold adaptation modifies mitochondrial capacity and function at low temperature, we examined the association between cold adaptation and mitochondrial respiration capacity in seven Drosophila species. The seven Drosophila species were chosen broadly from the phylogeny to represent independent examples of more cold-sensitive and more cold-adapted species as attested by their marked difference in chill tolerance (CT_min ranging from 7.02 to −1.94°C, measured in the present study). For each species, mitochondrial function was studied at five temperatures (19, 10, 7, 4, 1°C) using a substrate–uncoupler–inhibitor titration (SUIT) protocol to evaluate the contribution of individual mitochondrial complexes to the overall OCR. Given the temperature effect on biochemical processes and the requirement for sustained energy production to maintain homeostasis at low temperature, we hypothesized that (1) mitochondrial oxygen consumption will decrease with reduced temperature due to simple Q₁₀ effects, (2) the temperature effect on mitochondrial oxygen consumption and coupling will correlate with species cold tolerance, i.e. cold-adapted species will maintain higher OCRs and remain well coupled at low temperature compared with cold-sensitive species, and (3) that the relative contribution of ETS complexes will be stable in cold-adapted species but may change with temperature in cold-sensitive species.

Mitochondrial respiration was measured in seven Drosophila species that were selected broadly from the Drosophila genus to include both cold-sensitive and cold-adapted species from each of the two main subgenera (Sophophora and Drosophila). The species used are listed in Table 1 with decreasing level of cold tolerance (CT_min mean±s.e.m.; see methods for CT_min measurements below) and information on their cold tolerance group, which we here use to separate the species.

Flies were kept under common conditions (19°C, 22 h:2 h light:dark) in 250 ml plastic bottles containing 35 ml oat-based Leeds medium (Andersen et al., 2015). Parental flies were transferred to new bottles twice a week to prevent high larval density, and eclosed flies for experiments were transferred to new bottles every 2–3 days. Only female flies aged 3–12 days were used for measurements, because of their larger body size. All species had been kept in the lab for at least 20 generations before use in experiments.

Individual flies were placed in 5 ml glass vials (sample sizes in Table 1), mounted on a rack and submerged in a tank connected to a cooling bath with ethylene glycol set to 19°C (LAUDA-Brinkmann, Delran, NJ, USA). Temperature was gradually decreased by 0.1°C min⁻¹, and flies were checked for movement with increasing frequency as the flies started to show uncoordinated movement at low temperature. Once spontaneous movement stopped, vials were prodded, and flies were again checked for movement; the temperature where flies no longer responded was taken as the CT_min.

Mitochondrial OCRs were measured in permeabilized thoraces using the Oxygraph-O2K system (Oroboros Instruments, Innsbruck, Austria) and the associated software (DatLab v7.0.4.1, Oroboros Instruments). Permeabilized thoraces were chosen because of the high metabolic activity in insect flight muscle (Weis-Fogh, 1964) and the small number of flies required for measurements. Measurements were performed at five temperatures: 19°C (rearing temperature) and lower temperatures (10, 7, 4 and 1°C) that collectively cover critical temperatures for most of the tested species (CT_min ranging from 7.02 to −1.94°C). The protocol was similar to that used in Jørgensen et al. (2021), but is briefly outlined below.

Preparation of thoraces was performed over ice. Flies were placed on ice to incapacitate them and with a scalpel and a pair of tweezers, the thorax was separated from the head, abdomen, wings and legs and then placed in ice-cold biological preservation solution (BIOPS; 2.77 mmol l⁻¹ CaK₂EGTA, 7.23 mmol l⁻¹ K₂EGTA, 5.77 mmol l⁻¹ Na₂ATP, 6.56 mmol l⁻¹ MgCl₂, 20 mmol l⁻¹ taurine, 15 mmol l⁻¹ Na₂-phosphocreatine, 20 mmol l⁻¹ imidazole, 0.5 mmol l⁻¹ dithiothreitol and 50 mmol l⁻¹ K-MES, pH 7.1; Simard et al., 2018). Thoraces were gently punctured with a pair of fine-tipped forceps for initial mechanical permeabilization and then placed on a plate rotator (100 rpm) for 15 min in Eppendorf tubes with saponin-supplemented BIOPS (62.5 µg ml⁻¹, prepared daily) for chemical permeabilization. To stop the chemical permeabilization, thoraces were transferred to Eppendorf tubes with ice-cold respiration medium [RESPI; 120 mmol l⁻¹ KCl, 5 mmol l⁻¹ KH₂PO₄, 3 mmol l⁻¹ Hepes, 1 mmol l⁻¹ MgCl₂ and 1 mmol l⁻¹ EGTA, adjusted to pH 7.2 before adding 0.2% (w/v) fatty acid free BSA; Simard et al., 2018] and placed on the plate rotator (100 rpm) for 10 min. Thoraces were then gently blotted dry on a tissue and weighed [MSE6.6S-000-DM micro balance (0.001 mg), Sartorius, Göttingen, Germany] before being placed in a small droplet of RESPI on Parafilm over ice. The number of thoraces (1–3) for each chamber was chosen considering the species thorax size and the experimental temperature (balancing a clear oxygen consumption signal with the risk of depleting chamber oxygen during the protocol). Generally, more thoraces were used for the smaller species (D. melanogaster, D. teissieri and D. bunnanda) and, overall, the median mass used was 1.27 mg (first to third quartile: 0.98–1.80). The sample size for each combination of species and temperature is reported in Table 2.

To calibrate the oxygraphs, temperature was set and 2.5 ml RESPI was added to each chamber after which chambers were closed with stoppers and excess RESPI was aspirated to ensure a chamber volume of 2 ml. Then, stoppers were lifted with the associated spacer to equilibrate the oxygen concentration to that of air, and once the OCR was stable at ±1 pmol O₂ s⁻¹ ml⁻¹, the chambers were calibrated with respect to the pertaining barometric and water vapor pressure (DatLab v7.0.4.1, Oroboros Instruments).

OCRs were measured using a SUIT protocol (Jørgensen et al., 2021; Simard et al., 2018) (Fig. 1). To start the measurements, pyruvate (10 mmol l⁻¹), malate (2 mmol l⁻¹) and fly thoraces were added to each chamber to stimulate electron transport and proton pumping through complex I without coupling to oxidative phosphorylation (CI-LEAK; Fig. 1, step 1). As insect flight muscle can have very high rates of oxygen consumption, chambers were supplemented with oxygen to increase the oxygen concentration to ∼650 μmol l⁻¹ before closing. The SUIT protocol was then initiated, and each substance was injected once the OCR was stable. First, ADP (5 mmol l⁻¹) was added to couple proton pumping to the oxidative phosphorylation by ATP synthase (CI-OXPHOS; Fig. 1, step 2). Next, cytochrome c (10 µmol l⁻¹) was injected to examine the integrity of the outer mitochondrial membrane (Fig. 1, step 3), and in subsequent analysis, measurements with cytochrome c-induced increases in OCR exceeding 15% were discarded (Kuznetsov et al., 2008). Then, proline (5 mmol l⁻¹) was added to stimulate proline dehydrogenase (ProDH), a complex that transfers electrons to the Q-junction then complex III (CI+ProDH-OXPHOS; Fig. 1, step 4), followed by succinate (20 mmol l⁻¹) to stimulate electron transport through complex II (CI+ProDH+CII-OXPHOS; Fig. 1, step 5). Then, sn-glycerol 3-phosphate (15 mmol l⁻¹, G3P) was injected to stimulate electron transport through mitochondrial G3P dehydrogenase (mtG3PDH), another complex transferring electrons to the Q-junction (CI+ProDH+CII+mtG3PDH-OXPHOS; Fig. 1, step 6).

The uncoupler carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) was sequentially injected in doses of 1–2 μmol l⁻¹ to examine whether the ETS could increase electron transport capacity when the proton gradient was uncoupled from oxidative phosphorylation. Injections of FCCP were made until the OCR no longer increased (FCCP-ETS; Fig. 1, step 7). At this point, the main ETS complexes were sequentially inhibited; complex I by rotenone (0.5 μmol l⁻¹; Fig. 1, step 8), complex II by malonate (5 mmol l⁻¹, prepared daily; Fig. 1, step 9) and complex III by antimycin A (2.5 μmol l⁻¹; Fig. 1, step 9). The effect of the individual inhibitors on the OCR was not recorded for all samples, but the residual OCR remaining after all inhibitors were added originates from non-mitochondrial oxidative side reactions and was therefore subtracted from all measured OCRs before data processing. If oxygen concentration decreased below 180 μmol l⁻¹ after inhibition, the oxygen concentration was increased before the next steps (same procedure as above). Following stabilization of the OCR, ascorbate (2 mmol l⁻¹) and N,N,N′,N,-tetramethyl-p-phenylenediamine (TMPD, 0.5 mmol l⁻¹) were injected to stimulate electron transport through cytochrome c which delivers electrons to complex IV (CIV) (Fig. 1, step 11a). After the OCR had peaked, sodium azide (20 mmol l⁻¹; Fig. 1, step 11b) was injected to inhibit CIV, and the system was left for 15–20 min with the OCR gradually decreasing to subsequently account for auto-oxidation of TMPD when calculating the maximal OCR of CIV.

Unless otherwise stated, OCRs are reported as means±s.e.m. of mass-specific rates (pmol O₂ s⁻¹ mg⁻¹ permeabilized thorax).

Effects of species–temperature interactions on the OCR was examined for each step of the SUIT protocol. To aid comparison of temperature responses in OCRs between species that can have different basal levels of oxygen consumption, the OCRs were standardized by dividing each OCR by the species-specific mean OCR of CIV at 19°C for a relative value (unitless). This reference value was chosen as, for most of our measurements, the ETS exhibits the highest OCR when complexes I–III are inhibited and CIV is stimulated, and, moreover, the highest OCR on average was obtained at 19°C. Additionally, it is frequently observed that CIV has excess capacity compared with the ETS (Gnaiger et al., 1998; Pichaud et al., 2011).

Statistical data analyses were performed in R v4.2.1 (http://www.R-project.org/). Data were analyzed using two-way ANOVA with temperature and species as independent, categorical variables and included an interaction term. If a significant interaction was found (P<0.05), data were split to examine the simple main effect of species within temperature (for standardized OCRs and FCFs as dependent variables) or examine the simple main effect of temperature within species (for absolute OCRs, j_≈P and UCR as dependent variables) using one-way ANOVA and followed by a post hoc HSD Tukey's test if the one-way ANOVA was significant. When the interaction term was not significant, the interaction was removed and the main effects of temperature and species were examined using a post hoc HSD Tukey's test. F-statistics, degrees of freedom and P-values from the ANOVA are given in Tables S1–S3.

To investigate how low temperature affects mitochondrial function, mass-specific OCR was measured by successive stimulation of different parts of the ETS at five temperatures (1, 4, 7, 10 and 19°C) in seven Drosophila species.

Low temperature generally decreased OCR for all the examined steps of the ETS when compared within species. Two-way ANOVA for each specific step of the SUIT protocol indicated that there was a significant interaction between temperature and species in the OXPHOS state, but not for LEAK (CI-LEAK) and in the ETS state (FCCP-ETS and CIV-ETS) (Table S1). CI-LEAK generally decreased with decreasing temperature, but for some species (D. persimilis, D. melanogaster and D. sulfurigaster) it was not possible to detect this decline with decreasing temperature (Fig. 2; Fig. S2, Table S1). The UCR based on FCCP-ETS/CI+ProDH+CII+mtG3PDH-OXPHOS was always above 1 irrespective of species and temperature, indicating that the electron transport was limited by the phosphorylation system, i.e. there was an untapped capacity for electron transport in the system (Table 2). There was a significant interaction between temperature and species, and in five of the seven species, we found a slight tendency for UCR to increase with decreasing temperature, but this increase was only significant for D. persimilis and D. melanogaster. In D. teissieri, there was also an effect of temperature, but here UCR generally decreased with temperature (Table 2).

To identify putative interspecific patterns in the response to lowered temperature, the OCRs were standardized relative to the mean of the OCR of CIV at 19°C for each species (Fig. 3; Fig. S3). For CI-OXPHOS, when mitochondria were stimulated with pyruvate, malate and ADP, the relative OCRs revealed a pattern of species-specific temperature effects (supported by a significant interaction between temperature and species, P<0.001). Low temperature led to a larger reduction of CI-OXPHOS in the three cold-sensitive species (D. teissieri, D. bunnanda and D. sulfurigaster) relative to the more cold-adapted species (D. montana, D. persimilis and D. obscura), while the species with an intermediate level of cold tolerance (D. melanogaster) had a response in-between (Fig. 3A; one-way ANOVA followed by Tukey's HSD post hoc test, F-statistics given in Table S2). The species differences in temperature response were significant at the three lowest temperatures (1, 4 and 7°C) and the interspecific pattern was most obvious at 4°C where the level of CI-OXPHOS for D. melanogaster was intermediate to that of the three cold-sensitive and -adapted species, respectively.

The interspecific pattern of temperature sensitivity (and significant interaction between temperature and species, P<0.003) persisted after the addition of proline to stimulate CI+ProDH-OXPHOS (Fig. 3B) and after injection of succinate to fuel complex II (CI+ProDH+CII-OXPHOS; Fig. 3C). In both cases, the cold-adapted species had significantly higher relative OCRs than D. melanogaster and the cold-sensitive species at 1°C, while at 4°C, the relative OCRs of the cold-adapted species were significantly higher than those of the cold-sensitive species, with D. melanogaster intermediate to the sensitive and tolerant species. Furthermore, at 7°C, the cold-sensitive species had significantly lower relative OCR for CI+ProDH-OXPHOS and CI+ProDH+CII-OXPHOS than the four more tolerant species, while no interspecific differences were found for relative OCR at 10 and 19°C.

Addition of G3P to stimulate mtG3PDH (CI+ProDH+CII+mtG3PDH-OXPHOS) changed the interspecific pattern (Fig. 3D) such that differences in relative OCR were not significant in most species comparisons, although there was still an interaction between temperature and species (P=0.024). Accordingly, G3P stimulated OCR more in the cold-sensitive species (see below).

To examine the relationship between mitochondrial function at low temperature and species cold tolerance, CI-OXPHOS and CI+ProDH+CII+mtG3PDH-OXPHOS rates measured from 1 to 10°C were normalized relative to the respective mean OCR at 10°C. Normalization to 10°C was chosen as all examined species are able to move and exhibit normal neuromuscular function at this temperature (Andersen et al., 2015; MacLean et al., 2019). In this analysis, we estimated the temperature where the normalized OCR was reduced to 50% of the activity at 10°C, using interpolation of the normalized OCRs associated with the two bracketing temperatures. There was a tendency for the cold-sensitive species to experience a reduction in CI-OXPHOS OCR at a higher experimental temperature than the cold-adapted species (Fig. 4A). A less obvious pattern was found for CI+ProDH+CII+mtG3PDH-OXPHOS OCR, where the cold-adapted species generally displayed a 50% reduction at lower temperatures than the other species, but here the cold-sensitive D. teissieri also required a very low temperature to reduce the OCR (Fig. 4B). These interspecific patterns are also reflected in the thermal sensitivity (Q₁₀) of CI-OXPHOS and CI+ProDH+CII+mtG3PDH-OXPHOS for the interval 1–10°C. The Q₁₀ of CI-OXPHOS was higher for cold-sensitive species than for cold-adapted species, while thermal sensitivity was more alike between species for the maximal coupled OCR CI+ProDH+CII+mtG3PDH-OXPHOS (Fig. 4A,B).

To associate mitochondrial capacity with species cold tolerance, we regressed the index of mitochondrial function (the estimated temperature for 50% reduction in normalized OCR for each species) to species cold tolerance, CT_min. The resulting linear regression indicates that there was a strong correlation between the decrease in CI-OXPHOS and the level of cold tolerance (R²=0.73; F_1,5=17.55, P=0.009; Fig. 4C), while the correlation for the maximal coupled OCR CI+ProDH+CII+mtG3PDH-OXPHOS was not significant (R²=0.23; F_1,5=2.77, P=0.157; Fig. 4D).

Considering the significant association between CI-OXPHOS and cold tolerance, we examined the temperature effect of coupling efficiency at the level of CI (j_≈P), which describes the relationship between CI-LEAK and CI-OXPHOS. There was a significant interaction between temperature and species, but comparison of j_≈P between temperatures within species only revealed significant temperature effects in three of the seven species: D. persimilis, D. melanogaster and D. bunnanda (Table 3; Table S1). For D. persimilis and D. bunnanda, the OXPHOS coupling efficiency observed at 1°C was significantly lower than under control conditions at 19°C, and for D. melanogaster, j_≈P was significantly different between 4 and 19°C, but increased at 1°C.

The observation that the fully stimulated maximal OXPHOS (CI+ProDH+CII+mtG3PDH-OXPHOS) had a weaker association with species cold tolerance than CI-OXPHOS suggests that addition of ‘alternative’ substrates can partially compensate for the reduced CI-OXPHOS found particularly in the cold-sensitive species. To examine this explicitly, we calculated FCFs within temperatures for each species to investigate how much OCR increases with the addition of each added substrate. As the FCF reports the relative increase in OCR after injection of a substrate, it is sensitive to both the absolute increase in OCR and the prevailing condition before addition of the substrate (i.e. if OCR is low before substrate injection, then FCF will increase more with a fixed increase in OCR as it reports the relative change).

Our analysis of FCFs revealed significant temperature-dependent differences in mitochondrial substrate oxidation between species (Fig. 5; Table S3). Injection of proline generally had a low effect on the OCR, and there was no consistent pattern between species across temperatures (Fig. 5A; interaction between temperature and species in two-way ANOVA, P=0.072). When succinate was injected at low temperatures (1 and 4°C), there was a tendency for larger responses in OCR for the four most cold sensitive species compared with the three cold-adapted species, supported by the significant interaction between temperature and species (P<0.001), and this difference was significant at 1°C (Fig. 5B). Further, D. teissieri responded significantly more to the addition of succinate at 7°C than all the other species. A similar but even stronger increase in FCFs was found following addition of G3P (Fig. 5C; interaction between temperature and species in two-way ANOVA, P<0.001), where cold-sensitive species responded relatively more to this substrate at 1 and 4°C (also at 7°C for D. teissieri). At 10 and 19°C, FCFs were generally low across species for the three alternative substrates, highlighting that the FCF is sensitive to the pertaining OCR before injection of alternative substrates (at 10 and 19°C, CI-OXPHOS is not compromised and accordingly the starting point for the calculation is high).

This study examined the acute effects of low temperature on mitochondrial function in permeabilized flight muscle of seven species of Drosophila that represent different levels of cold tolerance. The study system included cold-adapted and cold-sensitive species from both subgenera of the Drosophila genus (Fig. S1), to ensure that a relationship between the level of cold tolerance and mitochondrial function at low temperature was not simply an artefact of species relatedness. Overall, our results show that mitochondrial OCR is strongly influenced by temperature; all seven species had reduced oxygen consumption at the lower temperatures (from 1 to 7°C) compared with the higher temperatures (from 10 to 19°C) (Fig. 2). Notably, cold-sensitive species (D. sulfurigaster, D. bunnanda and D. teissieri) were characterized by larger reductions in complex I-supported respiration at low temperature than cold-adapted species (D. obscura, D. persimilis and D. montana) and the intermediate D. melanogaster (Figs 3 and 4). As a result, these cold-sensitive species may rely more on oxidation of alternative metabolic substrates (Fig. 5).

In accordance with expectations of normal temperature effects on biochemical processes (Cossins and Bowler, 1987; Hochachka and Somero, 2002; Jørgensen et al., 2022), we found that lowering the temperature decreased OCRs of all complexes in the ETS. This temperature response was similar across all seven species when analyzed in the temperature interval between 10 and 19°C, a temperature range where all species are able to maintain neuromuscular function. Thus, mitochondrial OCR decreased with an average Q₁₀ of 2.1 (range: 1.6–3.1) and 1.8 (range: 1.3–2.2) for CI-OXPHOS and CI+ProDH+CII+mtG3PDH-OXPHOS, respectively, between 10 and 19°C. Although the capacity for mitochondrial oxygen consumption is not directly comparable to resting metabolic rate at the organismal level (Davison, 1971; Iverson et al., 2020; Tribe and Bowler, 1968), we note that the thermal sensitivity of mitochondrial OCR at these permissive temperatures is close to the average Q₁₀ of 2.7 found for standard metabolic rate between 10 and 20°C across more than 60 species of temperate and tropical Drosophila (Messamah et al., 2017).

The interspecific patterns of mitochondrial thermal sensitivity change markedly when OCRs are measured in the range of temperatures from 1 to 10°C. This range of temperatures will initially challenge the physiological performance of the three cold-sensitive species (D. sulfurigaster, D. bunnanda and D. teissieri) that enter cold coma at 6–7°C and, subsequently, cold will stress the ‘intermediate’ species (D. melanogaster) that enters cold coma at 3.5°C. In contrast, the three cold-adapted species (D. obscura, D. persimilis and D. montana) only enter chill coma once temperatures have decreased below 0°C. In the temperature interval between 10 and 1°C, we found clear differences in the dynamics and capacity of the mitochondrial respiration rate associated with species cold tolerance. This was particularly evident with respect to the reduction in complex I-supported respiration, which was much more pronounced in the cold-sensitive species (Figs 2 and 3). Here, Q₁₀ of CI-OXPHOS was 7.6 (range: 5.6–9.0) for the four most cold-sensitive species while the three cold-adapted species retained ‘normal’ thermal sensitivity (Q₁₀=2.9; range 2.4–3.1; Jørgensen et al., 2022; Schmidt-Nielsen, 1997) (Fig. 4A). A similar change in thermal sensitivity of complex I-driven respiration was recently reported for D. melanogaster with Q₁₀=8.8 at temperatures between 6 and 12°C and Q₁₀=1.8 between 12 and 18°C (Menail et al., 2022). The association between cold tolerance and preservation of CI-OXPHOS is highlighted by the strong correlation found between the temperature estimated to induce a 50% reduction of CI-OXPHOS relative to the rate at 10°C (an index for decreased CI-OXPHOS capacity) and our measure of cold tolerance, CT_min (Fig. 4A,C). A recent study examined the effect of cold acclimation in mayfly larvae and found a tendency for increased capacity for CI-supported respiration at low temperature compared with warm-acclimated conspecifics (Havird et al., 2020), but a comparative study directly examining a potential association between acclimation that increases cold tolerance and increased reliance on CI-OPXHOS has not yet been made. While our analysis suggests that preservation of CI-OXPHOS is important for cold tolerance, it is important to emphasize that there is no mechanistic link to associate a particular rate of CI-OXPHOS (i.e. a specific rate in pmol s⁻¹ mg⁻¹) directly to the onset of CT_min. Our findings simply suggest that thermal sensitivity of CI-OXPHOS is higher than that of organismal metabolic rate, which may cause unbalanced mitochondrial function as temperature decreases. CI-OXPHOS is a key contributor to the proton gradient across the inner mitochondrial membrane (Wikström and Hummer, 2012) and the observed correlation suggests that the activity of this complex in the ETS may be of importance to support the capacity for ATP production. The few studies that have investigated ATP/O ratios at low temperature in insects have found these ratios to be well maintained at low temperature (Colinet et al., 2017; Lubawy et al., 2022; Wood and Nordin, 1980), but considering the differences in CI-OXPHOS found between species in the present study, it would be relevant to further examine the association between CI-OXPHOS and ATP/O ratio.

Interestingly, we recently found a similar strong association between heat failure of CI-OXPHOS and heat tolerance in six species of Drosophila (Jørgensen et al., 2021), again suggesting that preservation of CI-OXPHOS is important for homeostasis and thermal tolerance in Drosophila. It is likely, however, that the failure of CI-OXPHOS at high and low temperature is caused by different underlying mechanisms. At stressfully high temperature, the significant decrease in CI-OXPHOS is associated with a dramatic hyperthermic breakdown (Jørgensen et al., 2021). Such an acute disruption is not found at low temperature, and we instead suggest that the gradual decline in CI-OXPHOS at low temperature simply indicates that species are endowed with different thermal sensitivities that disproportionally reduce CI-OXPHOS in cold-sensitive compared with cold-adapted species (Fig. 4A).

Associated with the cold-induced suppression of CI-OXPHOS, we found that low temperature changed the relative contribution of mitochondrial substrate oxidation in all of the examined Drosophila species. The substrates succinate and G3P were of higher importance and had higher FCFs at low temperatures (1–7°C) than at high temperatures (10–19°C) (Fig. 5B,C), while the effect of proline was generally low and uniform across temperatures (Fig. 5A). Importantly, the temperature effect on mitochondrial substrate oxidation differed significantly between species in relation to cold tolerance such that cold-sensitive species responded more to the addition of succinate and G3P at low temperatures. However, the FCFs are calculated from the preceding absolute OCR, and the low rates of CI+ProDH-OXPHOS and CI+ProDH+CII-OXPHOS prior to addition of succinate and G3P (as will be characteristic of cold-sensitive species) will accordingly inflate the calculated FCFs for the same absolute increase. Thus, when examining the absolute increases in OCR after addition of succinate and G3P at low temperatures (Fig. 2), we did not observe marked differences between species. Nevertheless, because the OCRs associated with alternative substrates are less sensitive to low temperature across all species, we observed that species achieve similar relative maximal OCR, i.e. generally offset the differences in OCR that stem from species effects on CI-OXPHOS, through the use of alternative substrates (Figs 3 and 4). A similar compensation with alternative substrates at low temperature has also been found with addition of succinate and G3P in D. melanogaster and honeybees (Menail et al., 2022), or with addition of succinate in warm-acclimated mayfly larvae (Havird et al., 2020) and a planarian (Scott et al., 2019). Thus, when considering Q₁₀ for maximal OCR coupled to phosphorylation (after G3P injection) between 10 and 1°C, we found more similar thermal sensitivities (Q₁₀=3.2 for the four most sensitive species and Q₁₀=2.6 for the three most tolerant species) (Fig. 4B), which is also in line with Menail et al. (2022), who found Q₁₀=2.3 between 6 and 12°C in D. melanogaster when all substrates were available. Compensation using alternative substrates was also found following failure of CI-OXPHOS respiration at high temperature in six Drosophila species where particularly mitochondrial G3P oxidation served to compensate for a decreased CI-OXPHOS at stressfully high temperatures (Jørgensen et al., 2021). At the genetic level, genes involved in the glycerophosphate shuttle (the mitochondrial shuttle that G3P participates in, which is essential for maintaining redox balance in insect flight muscle; Sacktor, 1975) have been found to be more expressed at higher latitudes in D. melanogaster (Lavington et al., 2014), further suggesting that G3P oxidation may be important at low environmental temperatures. Finally, a study of the arctic blowfly and the temperate housefly found that G3P-supported respiration (with rotenone inhibiting complex I) was maintained at temperatures down to 2°C (Wood and Nordin, 1980). Although the temperature sensitivity (activation enthalpy) increased markedly in both species below 11.5°C, the cold-tolerant arctic blowfly displayed a smaller increase in thermal sensitivity, again supporting the potential importance of G3P as an alternative substrate at low temperature (Wood and Nordin, 1980).

A switch in substrate oxidation associated with decreased function of complex I may alter the amount of reactive oxygen species (ROS) produced. mtG3PDH is a potent producer of superoxide, while complex I mainly contributes through reverse electron flow (Miwa et al., 2003). Reverse electron flow is possible if there is a high membrane potential and little substrate available for complex I, which is supported by the observation that mild uncoupling or inhibition of complex I by rotenone strongly reduces superoxide formation (Miwa et al., 2003). As complex I is a large contributor to the membrane potential, reducing its activity in the ETS may counter the adverse production of ROS; however, the increased reliance on G3P oxidation may increase the ROS formation from mtG3PDH, which additionally is less sensitive to uncoupling (Miwa et al., 2003). Future studies should therefore consider whether the reliance on G3P versus complex I substrates is associated with an increased ROS production, which could contribute to the effects of cold stress.

A high degree of OXPHOS coupling, which indicates the effect of coupling oxygen consumption to oxidative phosphorylation relative to that used to offset proton leak, has been associated with a high mitochondrial efficiency for ATP production, although the measure itself does not encompass ATP production efficiency directly (Colinet et al., 2017; Gnaiger, 2014; Salin et al., 2018). In a study on an arctic blowfly (Wood and Nordin, 1980), the ratio between mtG3PDH-OXPHOS and mtG3PDH-LEAK, corresponding to the OXPHOS coupling efficiency reported in the present study, decreased markedly below 10°C. Meanwhile, the efficiency of the phosphorylating system (ATP/O) was relatively unaffected by temperature; it only decreased by 10% when temperature was decreased from 22 to 2°C (Wood and Nordin, 1980). Unfortunately, the study by Wood and Nordin (1980) did not examine phosphorylation efficiency in the more cold-sensitive housefly nor did they measure complex I respiration, making direct comparison to the present study difficult. In the present study, we found that OXPHOS coupling efficiency was generally maintained at a high level, despite the decreased CI-OXPHOS at low temperature (Table 3). Although there was a tendency for a decreased coupling at lower temperature in some of the species, there was no correlation with cold tolerance. This suggests that proton leak does not pose a significant expense in terms of oxygen consumption at low temperature.

Uncoupling the ETS from phosphorylation increased the mitochondrial oxygen consumption (UCR>1) across all temperatures and in all species. This indicates that the phosphorylating system is not able to utilize the full capacity of the ETS and therefore limits/controls the flux through the ETS (Table 2). Thus, high UCRs indicate that ATP synthesis capacity (ATP synthase) or the capacity of the adenine nucleotide translocase (ANT), which switches ATP for ADP across the inner mitochondrial membrane, is limiting respiration rate (Pichaud et al., 2011). The UCRs found in the present study were generally well above 1, and slightly higher than the UCRs reported in other insect studies examining the effect of temperature on mitochondria (range: 0.85–1.5) (Chamberlin, 2004; Jørgensen et al., 2021; Menail et al., 2022). While there was a tendency for increased UCRs at lower temperatures, this was rarely supported statistically and seemingly not associated with species cold tolerance (Table 2).

This study examined the effect of low temperature on mitochondrial function, and how interspecific differences in mitochondrial function are related to cold tolerance of Drosophila species chosen broadly from the genus, to include cold-adapted and cold-sensitive species from both subgenera. As expected, mitochondrial respiration rates decreased with lowered temperature in all species, but the dynamics of this response differ considerably between cold-adapted and cold-sensitive species (Fig. 2). Most notably, the reduction of complex I-supported respiration was larger and occurred at a higher temperature in cold-sensitive species (Figs 3 and 4), and we therefore found a strong correlation between the maintenance of CI-OXPHOS at low temperature and the species cold tolerance. Cold-sensitive species were able to partially compensate their lowered complex I-supported respiration by increased reliance on alternative substrates, particularly G3P (Fig. 5). Together, our findings suggest that cold adaptation is manifested in a more stable and well-regulated reduction in mitochondrial metabolism, particularly complex I-supported respiration, while cold-sensitive species are forced to rely more on alternative metabolic substrates at low temperature, which may potentially disrupt optimal mitochondrial function. Future studies should examine whether the temperature-induced change in substrate oxidation affects the species ROS production and/or their ability to maintain sufficient ATP production to support physiological homeostasis at low temperature.

The authors would like to thank Kristoffer Neldeborg Jensen for help with some of the Oroboros measurements and Elin Ellebæk Petersen for help with reagent preparation.