Oxidative phosphorylation efficiency, proton conductance and reactive oxygen species production of liver mitochondria correlates with body mass in frogs

Physiological traits including longevity, fecundity, and endurance to starvation and desiccation have previously been shown to be closely related to body size, placing it as one of the most important parameters in an animal's life history (Schmidt-Nielsen, 1984). As energy is essential for all biological activity, the link between body size and energy metabolism has previously been extensively studied (Schmidt-Nielsen, 1984; Darveau et al., 2002; Glazier, 2005). Most notably, basal metabolic rate in mammals and birds has been correlated with many energy-consuming processes at the level of tissues, cells and mitochondria (Kunkel and Campbell, 1952; Hulbert and Else, 2000; Else et al., 2004), providing support for the ‘multiple-causes model’ of allometry (Darveau et al., 2002). Research has focused on the relationship between body mass (M_b) and mitochondrial function (Darveau et al., 2002; Brand et al., 2003; Porter and Brand, 1993). Understanding the link between body size and mitochondrial bioenergetics is of fundamental importance as mitochondria are essential organelles of eukaryotic cells responsible for the biosynthesis of many cellular metabolites and the generation of chemical energy in the form of ATP (Brand, 2005).

Allometric relationships of proton leak and oxidative activity with animal body size have been observed in the liver mitochondria of endotherms (Brand et al., 2003; Porter and Brand, 1993; Porter et al., 1996; Polymeropoulos et al., 2012). Proton leak refers to the mechanism where motive force is dissipated through the mitochondrial inner membrane proton conductance pathways independently of ATP synthase. In turn, proton leak diverts energy away from ATP production, resulting in the inefficiency of mitochondria. Therefore, previous studies suggest that small endotherms have a lower mitochondrial efficiency than larger species (Brand et al., 2003; Porter and Brand, 1993; Porter et al., 1996). These allometric relationships echo several experimental studies on (i) organisms treated with chemical uncouplers such as 2,4-dinitrophenol (Toyomizu et al., 1992; Caldeira da Silva et al., 2008; Salin et al., 2012a) or (ii) a genetic line of poultry selected for low feed efficiency (Bottje et al., 1999), showing that mitochondrial efficiency could constrain growth rate and so underpin the variability of M_b. However, the question of why mitochondrial inefficiency evolved in an allometric manner among taxa remains an open debate. One possible explanation is that mitochondrial efficiency is constrained by the generation of reactive oxygen species (ROS). ROS are the by-product of normal mitochondrial activity and play an important role in the rate of ageing and lifespan. According to the ‘uncoupling to survive’ theory, stimulating proton leak would lower mitochondrial efficiency in addition to the generation of ROS and this would result in an advantage to long-lived species (Brand, 2000).

However, there are not as many studies on ectotherms. In terrestrial ectotherms, respiratory metabolic rate correlates with M_b (Nagy, 2005), but this is less clear at the cellular and mitochondrial levels (Brookes et al., 1998; Else and Hulbert, 1985; Hulbert et al., 2002). However, studies investigating the relationship between mitochondrial activity and M_b in ectotherms are scarce. Studies available showing no relationship may simply reflect the disparate nature of the ectotherms studied or experimental procedures used as discussed in Hulbert et al. (2002). The aim of the present study was to investigate whether parameters of mitochondrial bioenergetics, including ATP synthesis efficiency, proton conductance and ROS generation, show an allometric relationship with M_b in frog species from the same family [Ranidae: Rana temporaria L., Pelophylax ridibundus (Pallas 1771) and Lithobates catesbeiana (Shaw 1802)]. Our results show a strong negative correlation between proton conductance and M_b, and a positive correlation between mitochondrial ATP synthesis efficiency and M_b. Interestingly, the larger frog species, which display high mitochondrial ATP synthesis efficiency and low inner membrane proton conductance, also exhibit a low level of mitochondrial ROS production.

M_b and liver mass of each of the frog species are listed in Table 1. The liver represented a smaller proportion of total M_b as the size of the frog species increased (Table 1). This trend resulted in an allometric relationship of liver mass to M_b with an exponent of 0.95 (Fig. 1).

Table 1 shows the respiratory parameters of liver mitochondria isolated from each of the frog species. The mitochondrial oxygen consumption measured at the maximum rate of ATP synthesis (state 3) was not significantly different between the three species, or significantly affected by M_b. In contrast, the rate of basal non-phosphorylating oxygen consumption, measured in the presence of oligomycin (state 4_oligo), and the rate of uncoupled oxygen consumption, measured in the presence of FCCP (state 3_FCCP), were both significantly correlated to M_b. The respiratory control ratios (RCRs) were significantly different between species and were significantly dependent on M_b (Table 1). Importantly, mean RCR values obtained (3.2±0.1) in this study fall within the range of values previously published for ectotherm liver mitochondria respiring on succinate (1.4 to 5.8; Akhmerov, 1986; Brand et al., 1991; Hulbert et al., 2002; Savina et al., 2006; Salin et al., 2012b). Calculated fractional mitochondrial aerobic scope (fAS_m) was significantly higher in larger individuals (Table 1), with a significant dependence on M_b (fAS_m=2.4×M_b^0.14, where M_b is in g; R²=0.28, P<0.001). These data are in line with previously published data in birds with an exponent of 0.09 (Brand et al., 2003) or in mammals with an exponent of 0.12 (Porter and Brand, 1993; Porter et al., 1996) and suggest that larger species have higher effective mitochondrial electron transport capacity than smaller species.

Fig. 2A shows the linear relationship between the rates of ATP synthesis and oxygen consumption measured in the presence of an ATP-regenerating system (hexokinase plus glucose) in liver mitochondria isolated from each of the frog species. As for the respiratory parameters reported above, the mean maximal rates of ATP synthesis and oxygen consumption (highest points to the right of the linear relationships) were not significantly different between species. The slopes of the linear relationships were positively mass dependent (ATP/O=0.4×M_b^0.23, R²=0.23, P=0.017), indicating that the larger frogs had a higher mitochondrial coupling efficiency than those with a smaller M_b (Fig. 2B). This is further highlighted by the fact that, when calculated at the highest common oxygen consumption rate (7.3 nmol O min⁻¹ mg⁻¹), liver mitochondrial ATP production increased as a function of M_b [ATP synthesis (nmol ATP min⁻¹ mg⁻¹)=1.2×M_b^0.33, R²=0.29, P=0.0006; mass effect: F=4.4, P=0.04; species effect: F=2.6, P=0.08; mass×species: F=2.5, P=0.1], with mitochondria from the largest frog (L. catesbeiana) producing on average 70% more ATP than mitochondria from the smallest frog (R. temporaria).

The values of proton-motive force (Δp) and its chemical (ΔpH) and electrical (mitochondrial membrane potential, Δψ) components were determined in frog mitochondria respiring on succinate under basal non-phosphorylating steady-state conditions (state 4_oligo). Table 2 shows that the values of Δp, ΔpH and Δψ in L. catesbeiana were significantly higher than values in R. temporaria. Fig. 3 reports the response of substrate oxidation and proton leak systems to mitochondrial membrane potential. The response of substrate oxidation rate was determined by manipulating the mitochondrial membrane potential with an increasing dose of FCCP, a known uncoupler (Brand and Nicholls, 2011). Fig. 3A shows that the liver substrate oxidation system of R. temporaria mitochondria was less active than that of mitochondria from the other two species, as it consumed oxygen more slowly at any given value of mitochondrial membrane potential. Mitochondrial proton leak at 150 mV, the highest common value of membrane potential (Fig. 3A), was significantly different between species (F=9.7, P=0.001; Fig. 3B), with mitochondria from the larger frog (L. catesbeiana) having only 50% of the proton conductance of liver mitochondria from the smaller frog (R. temporaria) on average. The proton leak rate at 150 mV was also negatively dependent upon mass (F=13.4, P=0.001) as shown in Fig. 4A.

ROS production showed a negative correlation with M_b (mass effect: F=9.3, P=0.005), but not between species (F=1.6, P=0.22). Interestingly, the interaction between these two factors was significant (mass×species: F=3.8, P=0.03) suggesting that the effect of M_b on ROS production is greater in R. temporaria than in P. ridibundus and L. catesbeiana. Therefore, liver mitochondria from the larger individuals produced significantly lower oxygen radical (per mg protein) than those from the smaller animals (Fig. 4B). When ROS production is estimated from the efflux of hydrogen peroxide, as here, it is assumed that the activity of superoxide dismutase is not limiting and not different between species. The activity of liver superoxide dismutase was not significantly different between frogs (21.0±0.9 U mg⁻¹ protein in R temporaria; 21.4±4.5 U mg⁻¹ protein in P. ridibundus; 20.7±2.3 U mg⁻¹ protein in L. catesbeiana). Therefore, the differences in hydrogen peroxide efflux rate between frog species would not be explained by differences in superoxide dismutase activity.

The mitochondrial coupling ATP/O ratio is of physiological and ecological importance as it shows how much oxygen and nutrients are needed to produce ATP, the main form of energy usable by cells. The results in this study indicate that mitochondria from larger frogs produce more ATP per amount of oxygen consumed and therefore per amount of nutrients oxidized than smaller frogs. This suggests that larger frogs have a lower cost of cellular functioning, as their mitochondria are set to minimize the cost of ATP production and therefore optimize energy output for cellular processes. Such a difference in mitochondrial coupling efficiency has previously been shown to correlate with intraspecific variation in M_b in natural populations of common frogs (Salin et al., 2012b). It has been suggested that a change in mitochondrial membrane proton leak is one of the underlying biochemical mechanisms involved in the variability of mitochondrial efficiency in common frogs (Salin et al., 2012b). To support this, we have found that there is a strong negative correlation between proton conductance and M_b, similar to previous studies on avian and mammalian liver mitochondria (Porter and Brand, 1993; Brand et al., 2003). However, this result contradicts previous studies that have shown no allometric relationship between mitochondrial proton leak and M_b in ectotherms (Hulbert et al., 2002). The differences in results between studies may simply reflect the disparate nature of the ectotherm species used in each study (fishes, toads, lizards, crocodiles, snails) in addition to the temperature and tissue studied (kidney, liver, skeletal muscle). Although the data were collected from a small number of frog species, they suggest that the negative relationship between proton conductance and M_b that is seen in endotherms (Porter and Brand, 1993; Porter et al., 1996; Polymeropoulos et al., 2012) may also occur in frogs. However, this may not necessarily be the case for all groups of ectotherms (Hulbert et al., 2002), as suggested in reptiles, where proton conductance of liver mitochondria from crocodiles appears greater than that of smaller reptiles (Brookes et al., 1998; Hulbert et al., 2002). The results in this study show that under resting conditions, liver mitochondria from larger frogs have lower costs of energy maintenance, consuming less oxygen to balance proton leak across the inner membrane.

The lower mitochondrial proton permeability found in the larger species might be explained by at least three mechanisms: decreased surface area per mass of protein, changes in mitochondrial phospholipid fatty acyl composition and/or changes in some specific membrane proteins. As an example, the surface area of liver mitochondria decreases as M_b increases in mammals (Else and Hulbert, 1985; Porter et al., 1996), accounting for almost 70% of the differences in proton leak (Porter et al., 1996). However, this does not seem to be the case in liver mitochondria from reptiles (Else and Hulbert, 1985) or birds (Else et al., 2004), showing that other characteristics of the inner membrane are required to explain the relationship between M_b and proton leak in these vertebrates. Another potential mechanism involves the fatty acid composition of the mitochondrial membrane phospholipids. Previous research has shown that the proton conductance of the native mitochondrial membrane positively correlates with the polyunsaturated fatty acid content of phospholipids (Porter et al., 1996; Brookes et al., 1998; Else et al., 2004). However, the loss of the phylogenetic and allometric differences in proton conductance found with liposomes made from the mitochondrial membrane (Brookes et al., 1997) further highlights that the presence of mitochondrial membrane proteins plays an important role in determining proton permeability (Stuart et al., 2001). For example, this has been suggested for the content of adenine nucleotide translocase (Talbot et al., 2004; Brand et al., 2005; Shabalina et al., 2006). In amphibians, some studies have shown that the mitochondrial inner membrane surface area might be greater in the mitochondria of liver and muscle from smaller species (Brookes et al., 1998; Hulbert et al., 2006; Berner et al., 2009). Therefore, the relationship between M_b and proton conductance in frogs might be due to variation in the mitochondrial inner membrane. This could be caused by changes in the inner membrane fatty acyl composition and/or in the inner membrane protein content (such as adenine nucleotide translocase isoforms); however, this remains to be clarified and requires future investigation.

Several studies have linked ROS production with proton-motive force (Korshumov et al., 1997; Miwa et al., 2003; Kikusato and Toyomizu, 2013). In this context, mitochondria with lower basal proton conductance would show a high membrane potential and greater ROS production. To investigate this further, we measured succinate-driven ROS production in the liver mitochondria, a process that occurs during reverse electron transport and is therefore critically sensitive to proton-motive force and mild uncoupling (Korshumov et al., 1997; Miwa et al., 2003; Mookerjee et al., 2010). We found that liver mitochondria from the larger individuals produced significantly lower levels of ROS (per mg protein) than those from the smaller animals. A similar negative relationship between M_b and ROS production can be found in the literature for liver, kidney, heart and fibroblasts in mammals (Sohal et al., 1989, 1990; Ku et al., 1993; Csiszar et al., 2012). However, such interspecific correlation between mammalian M_b and mitochondrial ROS generation is not systematically found, being mostly revealed when isolated mitochondria oxidize succinate in the absence of rotenone (Lambert et al., 2007) or when cells are metabolically stressed (Csiszar et al., 2012). However, it is apparent that the negative relationship reported between M_b and proton leak (Porter and Brand, 1993; Porter et al., 1996; Polymeropoulos et al., 2012) cannot be associated with a positive allometric variation in ROS generation (Sohal et al., 1989, 1990; Ku et al., 1993; Lambert et al., 2007; Csiszar et al., 2012). A similar result has been found in birds, with M_b shown to be negatively correlated with both proton leak (Brand et al., 2003) and ROS generation in liver mitochondria respiring on succinate (Lambert et al., 2007; Montgomery et al., 2012). Collectively, the data suggest (Ramsey et al., 2005; Hagopian et al., 2005), that basal ROS production does not have a negative relationship with basal proton leak. It has previously been reported that ROS generation in the presence of succinate is more sensitive to the pH gradient than to the proton-motive force or its electrical component (Lambert and Brand, 2004; Lambert et al., 2010). In this study, the low rates of ROS production from the larger frogs were, however, not explained by lower values of ΔpH (Table 2).

Another explanation for the difference in ROS production is the difference in the redox state of respiratory chain components. Interestingly, mitochondria from the smallest frogs (R. temporaria) exhibited a lower activity of the substrate oxidation system (Fig. 3A) while having similar non-phosphorylating and phosphorylating oxygen consumption rates to mitochondria from the other frogs. This indicates that mitochondria from the common frog have a low oxidative reserve as illustrated by the lower fAS_m. In turn, this implies that common frog mitochondria operated at a higher proportion of maximal respiration under the phosphorylating state than mitochondria from the other species (i.e. state 3_ADP of R. temporaria, P. ridibundus and L. catesbeiana operating at 80%, 75% and 65% of FCCP-induced maximal respiration rates, respectively; Table 2). Although we did not directly measure redox potentials of the electron chain carriers, the above result suggests that mitochondria from the small individuals had a more reduced respiration chain while functioning, which would explain the higher ROS production compared with mitochondria from the heavier individuals. Alternatively, because succinate-induced ROS generation is essentially located at complex I (Lambert et al., 2007, 2010), it has also been suggested that the differences in mitochondrial ROS production rates between species can be explained by differences in the number of superoxide production sites within complex I (Lambert et al., 2010). Whether low mitochondrial complex I content and/or low redox potential of the electron carriers may explain differences in mitochondrial ROS production between frog species remains unknown and requires further investigation.

In conclusion, liver mitochondria from frogs show a positive relationship between mitochondrial oxidative phosphorylation efficiency and M_b. This is due to a decrease in the mitochondrial proton conductance in the heaviest individuals. Interestingly, the negative relationship found between mitochondrial proton leak and M_b in frogs is similar to that in endothermic groups such as mammals (Porter and Brand, 1993), including marsupials (Polymeropoulos et al., 2012), and birds (Brand et al., 2003). Finally, and in contrast to our prediction, we found that liver mitochondria from the larger individuals produced significantly lower levels of ROS than those from the smaller animals. The molecular mechanism responsible for this negative allometric relationship between mitochondrial ROS production and M_b in frogs remains to be determined. Overall, the heaviest frogs (L. catesbeiana) exhibited a high mitochondrial coupling efficiency and a low level of mitochondrial ROS production, which would drive a high growth rate and promote longevity in this species. The results in this study argue that M_b is dependent on mitochondrial bioenergetics (mitochondrial efficiency, proton leak, ROS production) in frogs, raising the question of whether this would also be the case in other ectothermic groups. Future research should address this using more species before a conclusion can be drawn.

Three frog species from the same family (Ranidae) but with differing M_b were used. Common frogs (R. temporaria) were caught on the spawning sites located in Jurassian Bresse between Lons-le-Saunier and Dôle (Jura, France). Marsh frogs (P. ridibundus) were caught near Lyon (Rhône-Alpes, France) and bullfrogs (L. catesbeiana) were caught near Bordeaux (Gironde, France). Animals from the three frog species were maintained individually at 20±2°C, with a 12 h:12 h light:dark cycle and fed on crickets once a week for 2–3 weeks prior to killing by stunning and decapitation. All experiments were carried out according to the ethical principles of the French Ministry of Agriculture, and the French Department of Veterinary Services (DVS no. 69266347).

Individual mitochondrial preparations were from two to five pooled common frog livers, two marsh frog livers, or one bullfrog liver. Liver mitochondria was isolated using differential centrifugation in ice-cold isolation buffer containing 250 mmol l⁻¹ sucrose, 1 mmol l⁻¹ EGTA and 20 mmol l⁻¹ Tris-HCl brought to pH 7.4 at 4°C (Salin et al., 2012b). The protein content of the mitochondrial preparation was assayed at 540 nm using the biuret method with bovine serum albumin used as a standard. The mitochondrial preparation from frog livers contains a dark pigment which absorbs at 540 nm; therefore, the absorbance of the same volume of mitochondria in isolation buffer containing 0.6% potassium sodium l(+)-tartrate and 3% NaOH was subtracted.

Oxygen consumption was measured with a Clark oxygen electrode (Rank Brothers Ltd, France), in a closed and stirred glass cell of 0.5 ml volume and expressed per mg of mitochondrial protein at 25°C. Liver mitochondria were incubated in a respiratory buffer containing 120 mmol l⁻¹ KCl, 5 mmol l⁻¹ KH₂PO₄, 1 mmol l⁻¹ EGTA, 2 mmol l⁻¹ MgCl₂, 0.3% bovine serum albumin (w/v) and 3 mmol l⁻¹ Hepes at pH 7.4. Substrate concentrations used were 5 mmol l⁻¹ succinate plus 5 µmol l⁻¹ rotenone. The active state of respiration (state 3_ADP) was initiated by the addition of 500 µmol l⁻¹ ADP. The basal non-phosphorylating respiration rate (state 4_oligo) was measured in the presence of 5 µg ml⁻¹ oligomycin. The maximal uncoupled state of respiration (state 3_FCCP) was initiated by the addition of 2 µmol l⁻¹ carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (FCCP). The RCR refers to the ratio of ADP-stimulated oxygen consumption (state 3) to that consumed in the presence of oligomycin (state 4_oligo). We defined the fractional mitochondrial aerobic scope (fAS_m) as the ratio of FCCP-induced maximal uncoupling respiration rate (state 3_FCCP) to oligomycin-induced basal non-phosphorylating respiration rate (state 4_oligo). These metrics (RCR=state 3_ADP/state 4_oligo or fAS_m=state 3_FCCP/state 4_oligo) provide information on the function of oxidative phosphorylation, as state 3_ADP is controlled by the activity of ADP phosphorylation and substrate oxidation, state 3_FCCP is controlled exclusively by substrate oxidation and state 4_oligo is controlled predominantly by the rate at which protons leak across the mitochondrial inner-membrane (Brand and Nicholls, 2011).

Mitochondrial oxidative phosphorylation efficiency was determined by measuring ATP synthesis concurrently with oxygen consumption in respiratory buffer supplemented with 20 mmol l⁻¹ glucose and 1.5 U ml⁻¹ hexokinase. Different steady states of phosphorylation were obtained by adding ADP from 10 µmol l⁻¹ to 100 µmol l⁻¹. Mitochondrial ATP synthesis was determined from the glucose 6-phosphate content of the samples, which was measured enzymatically by spectrophotometry as previously described (Salin et al., 2012b). Briefly, after recording the respiration rate, four aliquots (60 µl each) of mitochondrial suspension were withdrawn from the respiratory chamber every 2 min and immediately added to 40 µl of perchloric acid solution (10% HClO₄, 25 mmol l⁻¹ EDTA) to stop the reaction. After centrifugation of the denatured protein (15,000 g for 5 min) and neutralization of the resulting supernatant, the glucose 6-phosphate content of the samples was measured by spectrophotometry at 340 nm in an assay medium consisting of 50 mmol l⁻¹ tri-ethanolamine-HCl, 7.5 mmol l⁻¹ MgCl₂, 3.75 mmol l⁻¹ EDTA, 0.5 mmol l⁻¹ NAD and 0.5 U glucose 6-phosphate dehydrogenase, pH 7.4. The rate of mitochondrial ATP production was calculated from the slope of the linear accumulation of glucose 6-phosphate over the sampling time interval (6 min). The linearity of glucose 6-phosphate accumulation was checked to ensure that the system was in a steady state. We also ensured that the ATP production measured was specific to mitochondrial ATP synthase activity by determining oxygen consumption and ATP synthesis rates in the presence of oligomycin (5 µg ml⁻¹). Over the range of ADP concentrations used, no oligomycin-insensitive ATP synthesis activity was measurable.

Respiration rate and membrane potential were measured simultaneously using electrodes sensitive to oxygen and to the potential-dependent probe triphenylmethylphosphonium (TPMP⁺) (Brand, 1995; Brookes et al., 1998; Salin et al., 2012b). Mitochondria were incubated in respiratory buffer supplemented with 5 µg ml⁻¹ oligomycin. The TPMP⁺ electrode was calibrated with four sequential 0.5 µmol l⁻¹ additions up to 2 µmol l⁻¹ TPMP⁺, and then 5 mmol l⁻¹ succinate was added to start the reaction and membrane potential was measured upon reaching the steady state. The chemical component of proton-motive force, ΔpH, was then measured as the change in membrane potential after ΔpH was converted to the electrical component of proton-motive force (Δψ) by addition of 60 ng ml⁻¹ nigericin (Brand, 1995). On the basis that proton-motive force=Δψ+ΔpH, ΔpH is simply calculated as ΔpH=proton-motive force (presence of nigericin)–Δψ (absence of nigericin) (Lambert and Brand, 2004). The kinetic response of proton leak and substrate oxidation to proton-motive force was determined by titration of non-phosphorylating respiration with malonate (up to 5 mmol l⁻¹) or FCCP (up to 800 nmol l⁻¹), respectively (Brand and Nicholls, 2011). After each run, 2 µmol l⁻¹ FCCP was added to dissipate the membrane potential and release all TPMP⁺ back into the medium for baseline correction. Membrane potential was calculated as described by Brand (1995), assuming a TPMP binding correction of 0.69 mg protein µl⁻¹ for liver mitochondria (Brand et al., 1991; Brookes et al., 1998). For P. ridibundus and L. catesbeiana, mitochondrial oxygen consumption and membrane potential were assayed at 25°C. For R. temporaria, mitochondrial oxygen consumption and membrane potential were assayed at 20°C (values are from Salin et al., 2012b) and corresponding values were recalculated to 25°C assuming a Q₁₀ of 2.3 for oxygen consumption and of 1.05 for membrane potential (Berner, 1999; Chamberlin, 2004; Rogers et al., 2007; Trzcionka et al., 2008; Guderley and Seebacher, 2011). For all three species, the rates of mitochondrial proton leak were calculated at the highest common membrane potential value of 150 mV from the oxygen consumption rate by assuming a constant stoichiometry of six protons pumped (and leaked) per atom of oxygen for succinate. It is important to note that one previous study has reported a proton leak activity of around 30 nmol H⁺ min⁻¹ mg⁻¹ protein in skeletal muscle mitochondria of common frog at 25°C (St-Pierre et al., 2000). Given that basal non-phosphorylating respiration rate and basal proton leak activity are on average 3.4- or 2.5-fold greater in skeletal muscle mitochondria than in liver mitochondria, respectively (St-Pierre et al., 2000; Hulbert et al., 2006; Rogers et al., 2007; Trzcionka et al., 2008; Guderley and Seebacher, 2011), the proton leak activity value of liver mitochondria from common frog should have ranged from 14 to 8.7 nmol H⁺ min⁻¹ mg⁻¹ protein (at 150 mV and 25°C). Hence, in the present study, the calculated mean value of 9.7±1.2 nmol H⁺ min⁻¹ mg⁻¹ protein (Fig. 3B) is in the range of expected values, showing that the above Q₁₀ values are reasonable.

The rate of H₂O₂ released from mitochondria was measured using a fluorescence spectrophotometer (Xenius, SAFAS Monaco) at excitation and emission wavelengths of 563 and 587 nm, respectively. The respiratory buffer was supplemented with 5 U ml⁻¹ horseradish peroxidase and 1 µmol l⁻¹ Amplex red fluorescent dye. Mitochondria and succinate were added to start the reaction. The fluorescent signal was calibrated using a standard curve obtained after successive addition of H₂O₂ (20–400 nmol l⁻¹). Electron leak was calculated as the fraction (%) of total electron flow that reduces O₂ into oxygen free radicals at the respiratory chain instead of reaching cytochrome c oxidase to reduce O₂ into H₂O. The total activity of liver superoxide dismutase was determined as previously described (Rey et al., 2010).

Results are presented as means±s.e.m. An ANCOVA was performed to test, for every parameter, both species and individual mass effects together with their interactions. Data were log transformed to homogenize variances when homoscedasticity was not observed. Statistical analyses were performed using JMP 7 (SAS Institute Inc., Cary, NC, USA). A 5% (P=0.05) level of significance was used in all tests.

We are grateful to Amy Sinclair for help with the English and to the reviewers for their helpful comments.