Evolutionary diversity of muscle OXPHOS efficiency in vitro across ectothermic vertebrates

Abstract

Oxidative metabolism meets the majority of vertebrate energy demands through the coupling of mitochondrial respiration to ATP production (OXPHOS). In endotherms, variations in OXPHOS coupling efficiency influence metabolic thermogenesis, locomotor economy and reactive oxygen species (ROS) generation. However, the extent of these variations and their functional implications in ectotherms are less clear. We measured mitochondrial oxygen consumption, ATP production and ROS production in permeabilized skeletal muscle fibres from salamanders, frogs and lizards representing ectotherm clades with low, medium and high standard metabolic rates (SMRs), respectively. Consistent with predicted associations with SMR, lizards had the highest capacities for muscle mitochondrial ATP production, while salamanders had the lowest. Unexpectedly, corresponding rates of oxygen consumption followed an opposite trend, reflecting 8.5-fold variations in OXPHOS coupling efficiency between salamanders (the lowest) and lizards (the highest). Intrinsic proton permeability of the inner mitochondrial membrane was the primary source of OXPHOS coupling variation across species, being highest in salamanders and lowest in lizards. Basal proton leak mediated by uncoupling proteins and the adenine nucleotide translocase was only seen in lizards, where it limits mitochondrial ROS production. We infer that diverse evolutionary selection pressures drive unexpectedly wide variations in muscle OXPHOS efficiency with different functional implications across ectotherm clades.

1. Introduction

The transformation and utilization of chemical energy is a fundamental property of life that varies widely across animals. Across tetrapod vertebrates, much of this variation comes from reliance on metabolically costly endothermic versus ectothermic thermoregulation [1]. Within ectothermic tetrapods, body temperature and activity patterns drive variation in energy utilization [2,3]. However, variation also emerges from physiological differences at the biomechanical, cellular and molecular levels. In motion, species differ in the minimum cost of transport (COT_min) [4]. Under resting conditions, species differ in mass-specific standard metabolic rate (SMR), the amount of energy required to maintain homeostasis for an organism.

Species in the different ectothermic tetrapod clades differ in their SMRs; lizards and snakes have higher SMRs than frogs, which in turn have higher SMRs than salamanders, although ranges overlap [5–7]. The causes and consequences of these differences are not fully understood. Membrane lipid unsaturation levels have been posited to underlie the difference in metabolic rates between endotherms and ectotherms, and may contribute to SMR variation among ectotherms as well, although this remains understudied [8–11]. Cell size—and, by association, cell surface area to volume ratio—may also contribute to SMR variation across ectotherms [12]. For example, the salamander clade has the largest cells among tetrapods and the lowest SMRs [2,13,14], consistent with lower membrane maintenance costs of large cells translating into lower SMR [15]. However, within ectothermic clades, including the salamanders themselves, the relationship between SMR and cell size is weak or inconsistent [5,6,16–18]. Overall, we lack an understanding of the biochemical and cellular differences that give rise to the broad-scale differences in SMR across ectothermic tetrapod clades.

SMR is typically measured as mass-specific organismal oxygen consumption per unit of time, which estimates the rate of oxidative metabolism occurring in vivo. However, this rate may not necessarily equate to systemic rates of energy production or the capacity of body tissues to meet their energy demands. Cellular energy demand is met largely by adenosine triphosphate (ATP) produced through oxidative phosphorylation (OXPHOS) in mitochondria. ‘OX’—the transfer of electrons from oxidized metabolic intermediates through respiratory enzyme complexes to oxygen (cellular respiration)—is coupled to ‘PHOS’ (adenosine diphosphate (ADP) phosphorylation) by formation of a proton gradient (Δp) across the inner mitochondrial membrane. Dissipation of the Δp through the F_oF₁ ATP synthase provides the kinetic energy to phosphorylate ADP to ATP. However, OXPHOS coupling is imperfect, as dissipation of the inner membrane Δp also occurs by several processes independent of ADP phosphorylation. Taken together, these mechanisms facilitate proton ‘leak’ respiration (LEAK), whereby oxygen consumption enables substrate oxidation in the absence of ATP synthesis or demand. This uncoupled LEAK respiration accounts for up to 30% of basal metabolic rates in endotherms, where it contributes to metabolic thermogenesis and protection against the formation of harmful reactive oxygen species (ROS) [19]. In this manner, LEAK respiration has the potential to positively impact lifespan and resilience [20], but may also reduce the bioenergetic efficiency of locomotion [21]. Much less is known about the extent of variation in OXPHOS coupling, its physiological impacts or its relationship to SMR in ectothermic vertebrates. A better understanding of these relationships may shed light on the evolutionary basis for variation in SMR and bioenergetic efficiency across diverse biological lineages.

To this end, we quantified maximal rates of OXPHOS-linked ATP production, OXPHOS coupling efficiency (ATP produced per oxygen consumed) and mitochondrial respiratory control (regulation of substrate oxidation by ADP) in skeletal muscle from six exemplar ectothermic vertebrates representing clades ranging from low to high SMR: two salamanders, two frogs and two lizards. We also investigated mechanisms of LEAK respiration and their impacts on mitochondrial ROS production across species. We used the data from these experiments to test the hypothesis that differences in SMR across vertebrates reflect underlying differences in OXPHOS capacity and efficiency. We predicted that variation in muscle OXPHOS capacity would reflect the broad-scale differences in functional demands imposed by SMR across clades. Interestingly, our results are consistent with this hypothesis for maximal rates of ATP synthesis (PHOS), but not for corresponding rates of oxygen consumption (OX), revealing unexpectedly wide variation in OXPHOS coupling efficiency across ectothermic vertebrates. Based on these results, we posit that OXPHOS efficiency in ectotherms has been shaped by a variable balance of evolutionary forces across lineages, including relaxed functional demands in some cases and increased functional demands in others. More generally, our results demonstrate that measurements of respiration rate alone may not allow reliable comparisons of ATP production across ectotherms because of variation in OXPHOS coupling efficiency across species.

2. Methods

(a). Taxon sampling

We chose six species that exemplify ectothermic vertebrate clades that differ in SMR, and our exemplars also encompass variation in physiological ecology. We sampled 4−7 adult individuals from each of two salamander species (Ambystoma macrodactylum (long-toed salamander) N = 5 and Plethodon cinereus (red-backed salamander) N = 5; lunged and lungless, respectively), two frog species (Hymenochirus curtipes (western dwarf-clawed frog) N = 7 and Dryophytes (Hyla) cinereus (American green tree frog) N = 5; aquatic and terrestrial, respectively) and two lizard species (Anolis sagrei (brown anole) N = 4 and Hemidactylus frenatus (common house gecko) N = 4; diurnal and nocturnal, respectively). The A. macrodactylum were collected from the wild in May 2018 under the wildlife collection permit no. 180226 issued by the Idaho Department of Fish and Game and the P. cinereus were collected in July 2018 under the New York State Department of Environmental Conservation scientific collection permit no. 2303. Frogs and lizards were obtained commercially from the pet trade. Frogs and lizards were housed at 24°C and fed crickets weekly, and salamanders were kept at 15°C and fed fruit flies weekly. Animals were fasted for 3 days prior to euthanasia. Amphibians were euthanized by submersion in a 1% solution of MS-222, and lizards were euthanized by intracoelomic injection of MS-222 (0.75 ml 1% followed by 0.25 ml 50%) [22] under Colorado State University IACUC Protocol 17-7189A.

(b). Sample preparation

Following euthanasia, skeletal muscle tissue was immediately dissected from the thigh and transferred into ice-cold biopsy preservation (BIOPS) solution containing (in mM): 10 Ca-EGTA (0.1 μM free calcium), 20 imidazole, 20 taurine, 50 K-MES, 0.5 dithiothreitol, 6.56 MgCl₂, 5.77 adenosine triphosphate and 15 phosphocreatine, pH 7.1. Dissected muscle fibre bundles were immediately trimmed free of connective tissue and gently teased with needle-tip forceps in ice-cold BIOPS, followed by incubation with 25 μg ml⁻¹ saponin in BIOPS on ice for 20 min with gentle rocking to permeabilize cell membranes while leaving mitochondrial membranes intact [23,24]. Permeabilized fibre bundles were then transferred to mitochondrial respiration medium (MiR05) containing (in mM) 0.5 EGTA, 3 MgCl₂ hexahydrate, 60 lactobionic acid, 20 taurine, 10 KH₂PO₄, 20 HEPES, 110 sucrose and 0.1% BSA, pH 7.1 with KOH, and rinsed by rocking for 2 × 15 min on ice. Fibre bundles were gently blotted dry for 10−15 s on Whatman paper and weighed immediately before adding 3−4 mg to the oxygraph chamber for fluo-respirometry experiments.

(c). Mitochondrial fluo-respirometry

Mitochondrial respiration and ATP synthesis rates were measured in permeabilized muscle fibre bundles in MiR05 buffer using the Oxygraph-2k high-resolution fluo-respirometry system (Oroboros Instruments, Innsbruck, Austria). Assays were performed in hyperoxic conditions (250–400 μM O₂) to prevent well established O₂ diffusion limitations on mitochondrial respiration and ROS production rates in permeabilized muscle fibres [23,24]. While ectotherm species have different preferred body temperatures, a consistent assay temperature of 25°C was used for all experiments because it was high enough to support detectable rates of respiration and ATP synthesis from tissue quantities available from all individuals while still being within the ecologically relevant body temperature range for all focal taxa.

Maximal OXPHOS rates were induced in permeabilized fibre bundles by the addition of saturating concentrations of tricarboxylic acid (TCA) cycle substrates, including (in mM) 1 malate, 5 pyruvate, 10 glutamate and 10 succinate, in the presence of 2.5 mM ADP. Mass-corrected oxygen consumption rates (pmol O₂ mg⁻¹ wet weight s⁻¹; JO₂) were monitored in real time by resolving changes in the negative time derivative of the chamber O₂ concentration. Mitochondrial ATP synthesis rates (JATP) were measured simultaneously using a custom-fitted Fluo LED2-Fluorescence Module (Oroboros Instruments) to monitor the decline of chamber Mg²⁺ content with the magnesium green fluorophore (MgG; 5.5 μM, 503_Ex/530_Em nm). This assay is based on the higher binding affinity for buffer Mg²⁺ of ATP than ADP and the 1 : 1 exchange of ATP for ADP across the mitochondrial inner membrane by the adenine nucleotide translocase (ANT) [25,26]. To prevent confounding impacts of ATP hydrolysis by myosin ATPase and the sarcolemmal Na⁺/K⁺-ATPase on JATP measurements, blebbistatin (60 μM) and ouabain (25 μM) were included in the assay buffer to inhibit activities of these enzymes, respectively [27,28]. Importantly, neither compound had any impact on JO₂ of fibres in any of the selected species as determined in preliminary experiments. Calibration of the chamber MgG fluorescence signal in terms of free [Mg²⁺] and the different dissociation constants (K_d) for ADP and ATP were performed under experimental conditions (25°C at approx. 350 μM oxygen with all substrates and inhibitors at assay concentration) as described in detail in the electronic supplementary material. These data were used to calculate rates of ATP production (i.e. phosphorylation of chamber ADP to ATP) from changes in chamber MgG fluorescence as described by Oroboros Instruments (https://wiki.oroboros.at/index.php/MgG_Fluo-Manual) based on the method established by Chinopoulos et al. [29]. OXPHOS coupling efficiency was then calculated for each individual sample as the ratio of ATP synthesized per oxygen consumed (JATP/O) using JO₂ values obtained during the same 3−5 min period of stable flux. Notably, calibrations were performed in the presence and absence of permeabilized fibre bundles from each of the ectotherm clades and found to produce essentially identical MgCl₂ and nucleotide K_d titration curves (electronic supplementary material, figure S1), indicating no significant interference or biological impacts of the fibre bundles on assay performance.

Given the wide range of OXPHOS efficiencies revealed across ectotherms using the protocol above, we performed a second series of experiments on a subset of taxa to more closely investigate rates of LEAK respiration, the extent of mitochondrial respiratory control by ADP, and their impacts on mitochondrial ROS production. Our representative ectotherms for these experiments were the salamander P. cinereus (N = 4), the frog D. cinereus (N = 5) and the lizards A. sagrei (N = 5) and H. frenatus (N = 5). Maximal LEAK respiration was measured first by the addition of the same saturating concentrations of substrates used in the first experiment, but in the absence of ADP (LEAK_N JO₂). In the presence of reducing substrates, but no ADP to power the ATP synthase, Δp increases to its maximum and exerts negative feedback on electron flow through the respiratory chain, slowing the rate of O₂ consumption. LEAK_N JO₂ proceeds under these conditions at a rate that reflects the extent to which protons ‘leak’ back across the inner mitochondrial membrane into the matrix through a combination of (i) non-specific proton permeability of the membrane itself and (ii) membrane proteins capable of dissipating the Δp—primarily uncoupling proteins (UCPs) and the ANT [30,31]. Following stable LEAK_N JO₂, maximal OXPHOS-linked JO₂ was induced by adding 2.5 mM ADP to the respirometry chamber, fully activating the ATP synthase and markedly increasing respiration rate. Finally, to isolate and quantify LEAK JO₂ facilitated by non-specific proton permeability of the inner mitochondrial membrane, we inhibited the ANT with 5 μM carboxyatractyloside (CAT) to prevent ADP/ATP exchange (and, thus, F_oF₁ ATP synthase activity) along with basal proton leak through the ANT (LEAK_CAT) [32]. UCP activities are also inactive during the LEAK_CAT state owing to inhibitory effects of CAT and 2.5 mM quantities of adenylate nucleotides (ADP and ATP) present from the previous titration [33,34]. A conceptual schematic of OXPHOS coupling and LEAK respiration is provided in figure 1.

Figure 1.

Conceptual schematic of OXPHOS coupling and LEAK respiration. Oxidation of substrates derived from nutrient catabolism (OX) supplies electrons (e⁻) to the electron transfer system (ETS) in the inner mitochondrial membrane (IMM), which shuttles them to oxygen to form H₂O during cellular respiration (red arrows). This process is obligatorily coupled to the translocation of protons (H⁺) by ETS proteins from the mitochondrial matrix to the inner membrane space (IMS), which generates a proton gradient (Δp) that powers phosphorylation of ADP by the F_oF₁ ATP synthase (PHOS). In this manner, ADP phosphorylation and ADP/ATP exchange by the adenine nucleotide translocase (ANT) regulate the rate of oxygen consumption determined by cellular energy demands. However, non-productive dissipation of the Δp also occurs via translocation of protons back into the matrix by uncoupling proteins (UCPs), the ANT and non-specific permeability of the IMM (dashed arrows). This IMM proton leak facilitates additional flow of electrons to oxygen, referred to as LEAK respiration, which was induced herein by provision of substrates in the absence of ADP (LEAK_N; all dashed arrows) and following inhibition of the ANT and UCPs with carboxyatractyloside in the presence of substrates and adenylate nucleotides (LEAK_CAT; blue dashed arrow).

The extent to which the OXPHOS-linked JO₂ exceeds LEAK JO₂ is referred to as mitochondrial respiratory control (or ‘coupling control’), which reflects the capacity of the ATP synthase and ADP/ATP exchange by the ANT to increase the rate of substrate oxidation (JO₂) during OXPHOS relative to an index LEAK JO₂. Accordingly, mitochondrial respiratory control is commonly expressed as a ratio of these two respiration states (i.e. the respiratory control ratio, RCR; OXPHOS/LEAK JO₂ [35]) or as a coupling control factor (CCF; 1–(LEAK/OXPHOS JO₂)) that ranges from 0 (no respiratory control by ADP) to 1 (100% respiratory control by ADP) [24]. CCFs are preferred for use in high-resolution respirometry titration protocols, and so were calculated for all samples using the LEAK_N and OXPHOS-linked JO₂ from the same protocol as 1–(LEAK_N/OXPHOS JO₂). To exclude the impact of proton leak facilitated by UCPs and the ANT present during LEAK_N on calculations of mitochondrial respiratory control, CCFs were also calculated using the LEAK_CAT JO₂ as 1–(LEAK_CAT/OXPHOS JO₂). The contribution of UCPs and the ANT to LEAK was further examined by calculating the LEAK_N/LEAK_CAT JO₂ ratio, where values >1 reflect a greater contribution of these proteins than of non-specific inner membrane proton permeability to LEAK JO₂.

(d). Mitochondrial reactive oxygen species release

In addition to reducing oxygen at complex IV of the ETS to form H₂O, electrons transferred from oxidized substrates can singly reduce molecular oxygen to form superoxide and other ROS during mitochondrial respiration. Hydrogen peroxide (H₂O₂) is the primary form of ROS released from mitochondria into the cytosolic compartment, where it can participate in cell signalling and increase cellular oxidative stress if not quenched by endogenous antioxidant systems. Rates of mitochondrial ROS production increase as inner membrane Δp increases owing to a slowing of electron flow through ETS protein complexes, favouring their ‘slip’ to molecular oxygen at multiple sites [36]. Consequently, processes that dissipate the inner membrane Δp or membrane potential tend to reduce mitochondrial ROS formation, including the activities of UCPs [37] and the ANT [38]. Given these relationships between inner membrane Δp, respiration and mitochondrial ROS production, we measured the net rates of H₂O₂ release (JH₂O₂) from permeabilized fibre bundles in our second cohort of ectotherms during the LEAK_N, OXPHOS and LEAK_CAT respiratory states described above. This was accomplished using the O2k-Fluo LED2-module (Oroboros Instruments, Innsbruck, AT) by monitoring the chamber accumulation of resorufin fluorescence during respiration experiments, which forms from 1 : 1 oxidation of Amplex UltraRed (AmR; 5 μM; 563_Ex/587_Em nm) by H₂O₂ generated by the sample in the presence of horseradish peroxidase (HRP; 1 U ml⁻¹) [23]. Sample JH₂O₂ values were corrected for chemical background rates of resorufin fluorescence development assessed prior to starting each experiment at 25°C and approximately 350–400 μM O₂, and calibrated by averaging fluorescence responses to four successive 0.1 μM H₂O₂ titrations without sample in the presence of substrates under the same experimental conditions. JH₂O₂ is expressed per milligram of tissue wet weight and as a percentage of sample JO₂ recorded simultaneously.

(e). Statistical analyses

Data from all experiments are presented as means ± s.e.m., with individual values plotted along with bar graphs. Mean comparisons across the six-species cohort were performed using a 3 × 2 ANOVA to evaluate differences between and within the three ectotherm clades, with pairwise comparisons of individual species performed using post hoc Tukey HSD tests. Mean comparisons across the four-species cohort were performed using one-way ANOVA with post hoc Tukey HSD tests, as well as by grouped-mean comparisons of lizards versus amphibians by independent-sample t-tests. A p-value ≤0.05 was considered significant for all statistical analyses.

3. Results

(a). Discordant trends in maximal JO₂ and JATP underlie wide variation in OXPHOS coupling efficiency across ectotherms

Maximal OXPHOS-linked respiration rates (OXPHOS JO₂) varied 3.2-fold across the six species studied, ranging from 20.26 ± 1.74 pmol O₂ mg⁻¹ s⁻¹ in the lizard H. frenatus to 64.75 ± 6.41 pmol O₂ mg⁻¹ s⁻¹ in the salamander A. macrodactylum (p < 0.0001 by ANOVA; figure 2a). Between-clade differences were all statistically significant (p < 0.01–0.0001, ANOVA main effect), with no significant within-clade differences. All across-clade pairwise comparisons of salamanders versus frog and lizard species were significantly different (p < 0.02 by Tukey post hoc tests), but no other pairwise comparisons across species were significantly different.

Figure 2.

Maximal OXPHOS-linked respiration and ATP synthesis rates of skeletal muscle across ectotherm species. Bars show mean ± s.e.m. with individual data points for (a) the OXPHOS-linked respiration rate (OXPHOS JO₂; supported by respiratory substrates and ADP), (b) the corresponding rate of ATP synthesis (JATP), and (c) the calculated JATP per oxygen consumed (JATP/O) of permeabilized muscle fibres from the lizards (white bars) Anolis sagrei (A_sa) and Hemidactylus frenatus (H_fr), frogs (light grey bars) Dryophytes cinereus (D_ci) and Hymenochirus curtipes (H_cu), and salamanders (dark grey bars) Plethodon cinereus (P_ci) and Ambystoma macrodactylum (A_ma). Different capital letters over horizontal bars indicate significant clade-level differences by ANOVA (p < 0.01), while lowercase letters indicate significant differences between individual species by Tukey HSD post hoc tests (p < 0.02). Illustrations in (c) reflect the significant differences in OXPHOS coupling efficiency between the three focal ectotherm clades.

In contrast to maximal OXPHOS-linked JO₂, corresponding rates of ATP synthesis (JATP) were lowest in the salamander A. macrodactylum (35.75 ± 3.84 pmol mg⁻¹ s⁻¹) and highest in the lizard A. sagrei (121.17 ± 4.21 pmol mg⁻¹ s⁻¹), varying 3.4-fold across the six species examined (p < 0.0001 by ANOVA; figure 2b). Between-clade differences were all statistically significant (p = 0.02–0.0001, ANOVA main effect), and both salamander species were significantly lower than all other species by pairwise species comparison (p < 0.03–0.0001), with the exception of P. cinereus versus H. curtipes, which differed significantly from the other frog D. cinereus (p = 0.01) and lizard A. sagrei (p < 0.0001) by Tukey post hoc testing.

Muscle OXPHOS efficiencies expressed as maximal JATP/O varied 8.5-fold across the six ectotherm species, ranging from 0.30 ± 0.03 in the salamander A. macrodactylum to 2.56 ± 0.10 in the lizard A. sagrei (p < 0.0001 by ANOVA; figure 2c). Between-clade differences were all statistically significant (p < 0.001 ANOVA main effect), with the highest efficiencies in lizards > frogs > salamanders. Pairwise comparisons between all individual species were statistically significant (p ≤ 0.05 by Tukey HSD) except for the two salamander species.

In summary, OXPHOS-linked rates of ATP synthesis are highest in species representing the clade with the highest SMR (lizards), whereas OXPHOS-linked respiratory capacity of skeletal muscle tissue (JO₂) is highest in species representing the clade with the lowest SMR (salamanders). Taken together, these results suggest that the clade with the lowest SMR (salamanders) has the lowest capacity to produce ATP, and does so in the least bioenergetically efficient way. In contrast, the clade with the highest SMR (lizards) has the highest capacity to produce ATP, and does so in the most bioenergetically efficient way.

(b). Maximal LEAK respiration rates are highest in salamanders and lowest in lizards

Consistent with salamanders having the lowest OXPHOS efficiency of the three ectotherm clades examined, maximal LEAK_N respiration rate is highest in the salamander P. cinereus (11.77 ± 2.51 pmol O₂ mg⁻¹ s⁻¹) and lowest in the lizard A. sagrei (4.65 ± 0.61 pmol O₂ mg⁻¹ s⁻¹), varying approximately 2.5-fold among the four species examined in our follow-up studies (p = 0.0067 by ANOVA; figure 3a). However, while all pairwise comparisons of LEAK_N between P. cinereus and the three other ectotherm species are significantly different (p < 0.03 by Tukey HSD), no pairwise comparisons between the frog and lizard species are statistically significant.

Figure 3.

Proton leak-mediated respiration rates across ectotherm species. Bars show mean ± s.e.m. with individual data points for (a) rates of LEAK respiration of permeabilized muscle fibres induced by saturating substrate concentrations in the absence of ADP (LEAK_N JO₂), (b) LEAK respiration rates induced by carboxyactratyloside in the presence of substrates and ADP/ATP (LEAK_CAT JO₂) and (c) the calculated LEAK_N /LEAK_CAT JO₂ ratio for the lizards Anolis sagrei (A_sa) and Hemidactylus frenatus (H_fr), the frog Dryophytes cinereus (D_ci) and the salamander Plethodon cinereus (P_ci). Different lowercase letters indicate significant differences between individual species by Tukey HSD post hoc tests (p < 0.05).

In contrast to LEAK_N rates, maximal LEAK_CAT rates vary approximately ninefold across species (p = 0.0012 by ANOVA; figure 3b), but were also highest in P. cinereus (11.89 ± 3.02 pmol O₂ mg⁻¹ s⁻¹) and lowest in A. sagrei (1.38 ± 0.09 pmol O₂ mg⁻¹ s⁻¹). Similar to LEAK_N rates, pairwise comparisons of LEAK_CAT between P. cinereus and the two lizard species are significantly different (p < 0.006 by Tukey HSD), but not between frog and lizard species.

The ratio of LEAK_N to LEAK_CAT was calculated to assess the extent to which the mitochondrial Δp can be dissipated by protein-mediated proton flux relative to the non-specific proton permeability of the inner mitochondrial membrane. This ratio varies 3.6-fold across species (p = 0.001 by ANOVA, figure 3c), averaging above 1.0 in the lizards A. sagrei (3.46 ± 0.53) and H. frenatus (1.85 ± 0.35), but near 1.0 in the frog D. cinereus (0.97 ± 0.05) and salamander P. cinereus (1.07 ± 0.15). However, only pairwise differences between A. sagrei and the other three species are statistically significant (p < 0.04 by Tukey HSD).

Taken together, these results indicate that while maximal leak respiration rates are highest in salamanders across the ectotherm clades examined, variation across species is greater when UCPs and the ANT are inactive (during LEAK_CAT), which reflects a greater impact of these proteins on LEAK_N rates in lizards compared with frogs and salamanders.

(c). Calculations of mitochondrial respiratory control using LEAK_CAT, but not LEAK_N, parallel OXPHOS coupling efficiency

Mitochondrial respiratory control by ADP is a commonly used proxy for OXPHOS efficiency in permeabilized cells and isolated mitochondria, which is calculated from LEAK and OXPHOS-linked JO₂ supported by identical substrates during the same respirometry protocol [39]. However, indices of respiratory control do not always parallel direct measures of OXPHOS efficiency [40], which we posit may result from differences in the LEAK JO₂ state being compared with the OXPHOS-linked JO₂. Therefore, we calculated mitochondrial respiratory control as the OXPHOS coupling control factor (CCF) 1–(LEAK/OXPHOS JO₂) using both LEAK_N and LEAK_CATJO₂, and related these to the corresponding mean JATP/O from the same species.

Mean 1–(LEAK_N/OXPHOS) CCFs vary by 13%, from 0.73 ± 0.04 in the lizard A. sagrei to 0.83 ± 0.02 in the frog D. cinereus, with no statistically significant pairwise differences across groups (ANOVA p = n.s.; figure 4a). In contrast, the 1–(LEAK_CAT/OXPHOS) CCF varied by 25% across species, ranging from 0.74 ± 0.07 in the salamander P. cinereus to 0.92 ± 0.01 in the lizard A. sagrei (p < 0.05 by ANOVA; figure 4b), with significant pairwise differences between these two species (p < 0.01 by Tukey HSD), but no others. Interestingly, while trends for mean LEAK_N CCFs were largely unassociated with mean JATP/O across species (figure 4c), trends for mean LEAK_CAT CCF closely paralleled mean JATP/O across species (figure 4d), suggesting that non-specific dissipation of the inner mitochondrial membrane Δp is a primary determinant of OXPHOS efficiency in the four species examined under saturating substrate and ADP concentrations. Importantly, outer mitochondrial membrane integrity was found to be high across all samples and species, indicated by a negligible effect of a 10 μM cytochrome c titration of sample JO₂ following respirometry protocols (data not shown), confirming a uniformly high quality of sample preparations across species.

Figure 4.

Respiratory coupling control factors across ectotherm species. Bars show mean ± s.e.m. with individual data points of OXPHOS coupling control factors (CCFs) calculated as 1–(LEAK/OXPHOS JO₂) using (a) LEAK_N (substrates alone) and (b) LEAK_CAT (respiratory substrates + ADP and carboxyatractyloside to inhibit the adenine nucleotide translocase) with the corresponding OXPHOS JO₂ (respiratory substrates+ADP) obtained from the same respirometry protocol on permeabilized muscle fibres from the lizards Anolis sagrei (A_sa) and Hemidactylus frenatus (H_fr), the frog Dryophytes cinereus (D_ci) and the salamander Plethodon cinereus (P_ci). Relationships between mean JATP/O and CCFs calculated using LEAK_N (c) and LEAK_CAT (d) from the same species. Different lowercase letters indicate significant differences between species by Tukey HSD post hoc tests following a significant ANOVA (p < 0.05).

(d). Reactive oxygen species release rates are greatest in the salamander Plethodon cinereus during all respiratory states

In addition to influencing OXPHOS efficiency, the inner membrane Δp impacts production of ROS by electron transfer system proteins influencing the redox state of electron carrier molecules [41]. Consistent with this effect, raw rates of muscle mitochondrial JH₂O₂ (fmol H₂O₂ mg⁻¹ s⁻¹) were lowest across all taxa during OXPHOS when protons are dissipated through the ATP synthase, higher during LEAK_N when the ATP synthase is inactive, and highest during LEAK_CAT when the ATP synthase, ANT and UCPs are inhibited (p < 0.0001 for respiratory state main effect by ANOVA; table 1). During all three respiratory states examined, JH₂O₂ was greater in the salamander P. cinereus than all other species (p < 0.003 by Tukey HSD). Hemidactylus frenatus was also lower than A. sagrei and D. cinereus in LEAK_CAT (p < 0.05 by Tukey HSD), and averaged lower than D. cinereus across all respiratory states (p < 0.05 by Tukey HSD), but there were no other significant pairwise comparisons across species.

Table 1.

Rates of muscle H₂O₂ release across respiratory states and ectotherm species. Values are mean JH₂O₂ ± s.e.m. in fmol mg⁻¹ s⁻¹ for each species during the three respiratory states examined. Group comparisons by a 3 (respiratory state) × 4 (species) ANOVA revealed significant main effects of both respiratory state and species on JH₂O₂ (p < 0.0001), with no interaction between factors. Significant differences were also found for mean JH₂O₂ values across species (all states) and respiratory states (all species) by ANOVA (p < 0.01). Capital superscripts indicate significant differences between respiratory states across species, and lowercase superscripts indicate significant differences between species for a given respiratory state by Tukey HSD post hoc tests (p < 0.05).

respiratory state	A. sagrei (N = 5)	H. frenatus (N = 5)	D. cinereus (N = 5)	P. cinereus (N = 4)	state mean
OXPHOS	A28 ± 1a	A30 ± 1a	A65 ± 4a	A149 ± 25b	A64 ± 12
LN	A47 ± 5a	A43 ± 4a	A86 ± 4a	AB192 ± 37b	A87 ± 15
LCAT	B131 ± 7a	A75 ± 4c	B139 ± 7a	B234 ± 29b	B140 ± 14
species mean	69 ± 3ab	49 ± 2a	97 ± 4b	192 ± 30c	97 ± 14

(e). JH₂O₂ normalized to JO₂ is similar across species in LEAK_N, highest in salamanders during OXPHOS and highest in lizards during LEAK_CAT

Across all four species, muscle JH₂O₂ expressed as a percentage of JO₂ (JH₂O₂/O₂) during respirometry experiments averaged approximately 10-fold lower during OXPHOS than during LEAK_N, and was highest during LEAK_CAT (figure 5). JH₂O₂/O₂ during OXPHOS varied 2.2-fold across species, ranging from 0.14 ± 0.01% in the lizard H. frenatus to 0.31 ± 0.03% in the salamander P. cinereus (p = 0.001 by ANOVA; figure 5a). Pairwise comparisons revealed significant differences between P. cinereus and the other three species by Tukey HSD post hoc tests (p < 0.018), but no other differences across species. Similarly, JH₂O₂/O₂ during LEAK_N varied twofold across species, ranging from 0.87 ± 0.12% in H. frenatus to 1.77 ± 0.41% in P. cinereus, but there were no significant differences across species (ANOVA p = 0.11) (figure 5b).

Figure 5.

Mitochondrial reactive oxygen species release during OXPHOS and LEAK respiration states. Bars show mean ± s.e.m. with individual data points for rates of H₂O₂ release (JH₂O₂) from permeabilized muscle fibres expressed as a percentage of oxygen consumed during (a) OXPHOS (respiratory substrates + ADP), (b) LEAK_N (substrates alone), and (c) LEAK_CAT (respiratory substrates + ADP and carboxyatractyloside to inhibit the adenine nucleotide translocase), and (d) the calculated LEAK_CAT/LEAK_N JH₂O₂/O₂ ratio for the lizards Anolis sagrei (A_sa) and Hemidactylus frenatus (H_fr), the frog Dryophytes cinereus (D_ci) and the salamander Plethodon cinereus (P_ci). Different lowercase letters indicate significant differences between species by Tukey HSD post hoc tests following a significant ANOVA (p < 0.05).

In contrast to OXPHOS and LEAK_N states, mean JH₂O₂/O₂ during LEAK_CAT varied 4.3-fold across species, ranging from 2.28 ± 0.18% in the frog D. cinereus to 9.83 ± 1.02% in the lizard A. sagrei (p < 0.0001 by ANOVA; figure 5c). Nearly all of this variability was accounted for by fourfold higher rates in A. sagrei compared with all other species (p < 0.0001 by Tukey HSD), with no other significant pairwise comparisons.

(f). Lizards rely more heavily on protein-mediated proton leak to mitigatemitochondrial reactive oxygen species release than do amphibians

JH₂O₂/O₂ during LEAK_CAT to LEAK_N represents the extent to which activities of the ANT and UCPs (present during LEAK_N, but not LEAK_CAT) mitigate mitochondrial ROS release from a maximally reduced respiratory chain induced by saturating substrate concentrations and an inactive ATP synthase. The mean LEAK_CAT/LEAK_N ratio varies 7.5-fold across species, ranging from 1.4 ± 0.3 in the salamander P. cinereus to 10.5 ± 2.3 in the lizard A. sagrei (p = 0.002 by ANOVA; figure 5d). This reflects a much greater increase in JH₂O₂ in A. sagrei compared with the three other species when the ANT and UCPs are inhibited during LEAK_CAT (p < 0.02 by Tukey HSD). JH₂O₂ averaged 3.6 ± 1.0-fold higher in LEAK_CAT versus LEAK_N in the lizard H. frenatus, while this ratio in the frog D. cinereus (1.6 ± 0.1) was similar to that in the salamander P. cinereus. Taken together, these results suggest a greater reliance on the ANT and/or UCP activity to limit mitochondrial ROS release in the two reptile versus amphibian species (p < 0.001 by independent samples t‐test of combined group LEAK_CAT/LEAK_N).

4. Discussion

(a). Wide variation in muscle OXPHOS capacity and efficiency across ectotherms

The present study is the first to investigate the evolutionary diversity of muscle OXPHOS efficiency and its relationship to mitochondrial ROS production across ectotherm clades with widely varying SMRs. We hypothesized that muscle OXPHOS capacity or efficiency would be predicted by the functional demands imposed by SMR. Despite only two exemplar species per clade, our OXPHOS results largely show the same ordinal pattern as SMR. Mass-corrected rates of ATP synthesis (JATP) in permeabilized muscle fibres were highest in lizards, followed by frogs and salamanders. However, corresponding rates of tissue oxygen consumption (JO₂) during OXPHOS followed the opposite trend, producing unexpectedly wide variation in OXPHOS coupling efficiency across species. The primary source of this variation was intrinsic proton permeability of the inner mitochondrial membrane (LEAK_CAT JO₂), which was highest in salamanders and lowest in lizards. In contrast, contributions to proton leak mediated by UCPs and/or the ANT (active during LEAK_N JO₂) do not parallel the trends in OXPHOS efficiency across species, but may serve to mitigate ROS production in lizards. Taken together, these results shed new light on the evolutionary basis for variations of OXPHOS efficiency in ectotherms, with broad implications for the study of ectotherm metabolism and comparative bioenergetics.

(b). Low OXPHOS efficiency of salamanders may reflect relaxed selection on metabolic output

Given the low SMR of salamanders compared with other tetrapod vertebrates, the result that salamander muscle mitochondria have the highest maximal respiratory capacity of our three focal ectotherm clades is surprising. This result is, in principle, consistent with two possible evolutionary explanations. The first is that the low metabolic demands of salamanders have led to relaxed functional constraints on the ETS machinery, permitting the accumulation of slightly deleterious mutations that have diminished OXPHOS efficiency, either by increasing basal proton leak or by impairing proton translocation during electron transfers. This scenario is consistent with the much lower OXPHOS efficiencies in salamanders compared with frogs and lizards that parallel higher non-specific proton leak across the inner mitochondrial membrane (LEAK_CAT), with no appreciable contributions of UCPs or the ANT (LEAK_N/LEAK_CAT ≈ 1). Importantly, maximal respiration capacities induced by saturating concentrations of substrates may not be representative of system behaviour under natural conditions. These experimental conditions may, in fact, amplify the consequences of ‘sloppy’ or inefficient OXPHOS machinery that has nonetheless remained more than adequate to meet the low functional demands of salamanders in their natural environments over tens to hundreds of millions of years.

A second potential evolutionary explanation for the higher rate of LEAK respiration in salamanders is that it provides functional benefits that made inefficient OXPHOS machinery the target of natural selection. A high capacity for oxidative metabolism despite low ATP demand enables greater cellular redox potential, which could support anabolic challenges imposed by large genome and cell size, such as increased nucleotide synthesis [42]. Additionally, higher LEAK respiration reflects a greater capacity to dissipate the inner membrane Δp, which could decrease mitochondrial ROS production by limiting the ‘backup’ of electron flow through the ETS that favours superoxide formation [36,41]. However, we consider a scenario in which high LEAK evolved because it provides functional benefits to be unlikely for multiple reasons. First, patterns of molecular evolution in salamander OXPHOS protein-coding genes are consistent with relaxed purifying selection relative to that seen in frogs, suggesting weak functional constraints on metabolic output in salamanders [43]. Second, patterns of anatomical evolution in salamander metabolic organs (heart and liver) are consistent with organ structure following the constraints imposed by large cell size on developmental systems, again suggesting weak functional constraints on metabolic output in salamanders [44]. Third, despite having the highest LEAK JO₂, rates of ROS release from P. cinereus mitochondria were higher than from frogs and lizards across all respiratory states (table 1), including during OXPHOS, expressed as a proportion of JO₂. Considering these lines of evidence, we posit that weak functional constraints on metabolic output in salamanders have allowed the evolutionary decline of OXPHOS efficiency, but without strongly detrimental effects on salamander fitness.

(c). High OXPHOS efficiency and reliance on protein-mediated LEAK to limit mitochondrial reactive oxygen species production may reflect increased functional demands of higher standard metabolic rates in lizards

Consistent with having the highest SMRs of the ectotherm clades studied, lizards had the highest capacity for muscle mitochondrial ATP synthesis and the highest OXPHOS efficiency measured as JATP/O. Lizards also had the lowest LEAK_CAT JO₂ and the highest accompanying coupling control (CCF), reflecting a greater capacity to increase rates of oxidative metabolism under states of high metabolic demand (elevated ADP/ATP levels) than frogs and salamanders. Relative to amphibians, lizards routinely experience higher body temperatures and higher activity levels, which translate into higher overall energy expenditure. Additionally, the biomechanical conflict between breathing and running experienced by lizards translates into intermittent locomotion and apnoea, producing wide variation in metabolic flux during routine activity [45,46]. Therefore, lizards may have evolved higher muscle OXPHOS capacity, efficiency and control to meet the increased functional demands associated with higher overall energy expenditure and wider variations in metabolic flux.

Lizard muscle mitochondria also exhibited a greater capacity to dissipate their Δp through the activity of UCPs and/or the ANT than frogs and salamanders (higher LEAK_N/LEAK_CAT JO₂), although this was only statistically significant for A. sagrei (figure 3c). UCPs and the ANT are primary mediators of basal and inducible inner membrane proton leak in skeletal muscle mitochondria [30,47], which serves to limit ROS formation under conditions favouring high Δp [48,49]. The wide fluctuations in metabolic activity routinely experienced by lizards generally favour greater ROS production through rapid transient changes in Δp and mitochondrial redox status [41]. Therefore, lizards may have evolved a greater reliance on UCPs and/or the ANT to mitigate muscle mitochondrial ROS production under these conditions, as evidenced by much greater JH₂O₂ during LEAK_CAT versus LEAK_N in A. sagrei compared with amphibians. Future studies that expand taxon sampling within lizards and integrate additional datasets (e.g. western blots to quantify protein levels) are necessary to further test this hypothesis, as well as to investigate whether differences between our exemplar lizard species reflect nocturnality versus diurnality of these two lineages [50] or factors unrelated to occupying different activity period (and therefore thermal) niches. More generally, our study design compares two exemplars from each of three clades, allowing us to reveal patterns of variable mitochondrial physiology, but precluding us from drawing firm conclusions about adaptation. These patterns lay the foundation for future work that more explicitly tests for correlation between OXPHOS capacity and efficiency, metabolic demand and fitness based on broader taxon sampling and comparative methods that control for phylogeny [51,52].

(d). Effects of temperature on the mitochondrial economy

Because temperature has a profound effect on the rates of chemical reactions in ectotherms, we measured all species at the same temperature to enable inter-species comparison. We chose 25°C to ensure that we would have detectable rates of ATP synthesis and JO₂ under all respiratory states from all species. Because species of amphibians and reptiles differ in their preferred body temperatures, this temperature represents a different departure from the physiological preference for each species. For the salamanders, 25°C is on the higher side of their ecologically relevant temperature range [53]. However, critical thermal maxima (CT_max) for salamanders, irrespective of acclimation temperatures, are at or above 32°C [54]. For P. cinereus, preferred temperature in the laboratory ranges from 16 to 21°C, but the CT_max is 32°C for animals acclimated at 15°C [55]. Similarly, A. opacum, a congener of A. macrodactylum used here, has a CT_max of 37°C when acclimated at 15°C [54]. For the frogs, 25°C is closer to the centre of their ecologically relevant temperature range and below the critical thermal maxima of 37°C (H. cinerea) [56–58]. For the lizards, 25°C is on the lower side of their ecologically relevant temperature range, but well above the critical thermal minima of 9°C (A. sagrei) and 12°C (H. frenatus) [59–61]. Therefore, the assay temperature used in our studies is well below the temperatures at which salamanders experience physiological failure and well above the temperatures at which lizards experience physiological failure.

Even in the absence of physiological failure, deviations in body temperature from the thermal optimum may still alter the mitochondrial economy. Previous studies of other ectotherms demonstrated impacts of elevated temperatures on mitochondrial metabolism, including (i) increased LEAK respiration, which shows greater temperature sensitivity than OXPHOS-linked respiration; (ii) increased ROS release rates; and (iii) decreased ATP synthesis rates [62]. Because these effects are consistent with our observed differences between salamanders and other ectotherms, it is important to consider whether these across-clade differences are an artefact of higher-than-preferred assay temperatures for salamander tissue. We consider this unlikely for several reasons. First, the Arrhenius breakpoint temperature (ABT)—the temperature above which substantial changes in the mitochondrial economy are found because of discontinuities in the relationships between temperature and reaction rates—is typically higher than species’ CT_max, suggesting that our data represent normal mitochondrial physiology [62]. Second, complex I is typically the first to be functionally impaired by elevated temperatures [62], but OXPHOS in our studies was supported by millimolar concentrations of substrates that supply electrons to the ETS through complex II (succinate) in addition to complex I (malate, pyruvate and glutamate). Third, previous studies of ectotherm mitochondrial flux as a function of temperature suggest that the magnitude of change in mitochondrial economy is far less than we reveal across species. For example, a 10° difference in assay temperature (from 20 to 30°C) is associated with an approximately 33% change in ATP/O in toad liver mitochondria [63], whereas our focal species differ by approximately 850%. Similarly, H₂O₂ production by toad liver mitochondria during OXPHOS increased only 30% from 20 to 30°C, whereas variation of JH₂O₂ across our focal species is approximately 200%. Nonetheless, future studies comparing mitochondrial bioenergetics of these species across a range of temperatures will strengthen and perhaps expand upon the conclusions drawn here.

5. Conclusions

Results of the present study demonstrate marked across-clade variations in mitochondrial economy, despite wide variation in physiological ecology encompassed by our focal taxa. Overall, we interpret the lizards’ greater capacity for skeletal muscle ATP synthesis, higher OXPHOS efficiency and greater ability to reduce ROS production through facilitated proton LEAK to reflect selection to meet the functional demands of higher SMR and energy expenditure. In contrast, we interpret the salamanders’ lower capacity for muscle ATP synthesis, lower OXPHOS efficiency and greater intrinsic rates of LEAK and ROS production to reflect relaxed selection due to lower functional demands of lower SMR and energy expenditure. More generally, we demonstrate that muscle oxidative capacity (max JO₂), respiratory control factors (different CCFs) and OXPHOS efficiency (ATP/O) provide different information about mitochondrial bioenergetics that may differentially align with SMR. We advocate for further studies that expand taxon sampling within each clade, while controlling for phylogenetic relatedness, to (i) further corroborate the clade-level differences suggested here, and (ii) characterize any differences in mitochondrial bioenergetics that may underlie specific variation in physiological ecology within each clade.

Ethics

We sampled 4−7 adult individuals from each of two salamander species (Ambystoma macrodactylum (long-toed salamander) N = 5 and Plethodon cinereus (red-backed salamander) N = 5; lunged and lungless, respectively), two frog species (Hymenochirus curtipes (western dwarf-clawed frog) N = 7 and Dryophytes (Hyla) cinereus (American green tree frog) N = 5; aquatic and terrestrial, respectively) and two lizard species (Anolis sagrei (brown anole) N = 4 and Hemidactylus frenatus (common house gecko) N = 4; diurnal and nocturnal, respectively). The A. macrodactylum were collected from the wild in May 2018 under the wildlife collection permit no. 180226 issued by the Idaho Department of Fish and Game, and the P. cinereus were collected in July 2018 under the New York State Department of Environmental Conservation scientific collection permit no. 2303. Frogs and lizards were obtained commercially from the pet trade. Frogs and lizards were housed at 24°C and fed crickets weekly, and salamanders were kept at 15°C and fed fruit flies weekly. Animals were fasted for 3 days prior to euthanasia and were humanely euthanized with MS-222 under Colorado State University IACUC Protocol 17-7189A.

Data accessibility

Data supporting this study are openly available from Zenodo [64].

Supplementary material is available online [65].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

R.L.M.: conceptualization, funding acquisition, project administration, resources, supervision, writing—original draft, writing—review and editing; L.C.L.P.: data curation, formal analysis, investigation; M.W.I.: investigation, methodology; A.J.C.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed herein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

These studies were funded in part by the US National Science Foundation.