Skeletal muscle fuel selection occurs at the mitochondrial level

Post-absorptive mammals rest at a whole-body respiratory quotient (RQ) near 0.80, with fat providing ~2/3 of the energy substrates (Brooks and Donovan, 1983; Willis et al., 2005). Skeletal muscle arterio-venous RQ is yet lower, showing an even stronger bias toward fat oxidation in resting muscle (Dagenais et al., 1976). Fasted birds generally rest at a lower whole-body RQ. For example, 2.8 g hummingbirds rest at a RQ of 0.72 (Suarez et al., 1990), 23 g sparrows at 0.71 (Walsberg and Wolf, 1995) and 480 g ravens at 0.80 (Hudson and Bernstein, 1983). These observations taken together support the concept that, in the post-absorptive state, resting mammalian and avian skeletal muscle cells preferentially select lipid as their oxidative substrate.

Most mammalian and avian skeletal muscle cells rest at very low fractions of their aerobic capacity, probably less than 1%. Studies in several mammalian species have clearly demonstrated the enormous aerobic scope of mammalian skeletal muscle. For example, human muscle, which has been shown to rest at about 2 ml O₂ min⁻¹ kg⁻¹ (Dagenais et al., 1976; Zurlo et al., 1990), may achieve maximum O₂ consumption rates in excess of 300 ml min⁻¹ kg⁻¹ (Andersen and Saltin, 1985; Richardson et al., 1993). The metabolic control signals underlying this expansive dynamic range, while not definitively understood, have been well characterized in many mammalian species using several methodologies (Chance et al., 1985; Connett, 1987; Jeneson et al., 1995). Contractile activity elevates the rate of ATP hydrolysis and results in the cytosolic accumulation of the ATP hydrolysis products ADP and inorganic phosphate (P_i), and thus decreased (less negative) ATP free energy (ΔG_ATP). A nearly linear relationship between O₂ consumption rate () and ΔG_ATP, consistent with simple feedback control, has been repeatedly demonstrated both in vivo (Kushmerick et al., 1992; Jeneson et al., 1995) and in isolated mitochondria in vitro (Rottenberg, 1973; Davis and Davis-van Thienen, 1989; Willis and Jackman, 1994; Messer et al., 2004; Glancy et al., 2008; Glancy et al., 2013). Because of the near-equilibrium maintained by the creatine kinase and adenylate kinase reactions, in vivo P_i rises nearly linearly with muscle , while ADP, and particularly AMP, rise more abruptly as high fractions of the aerobic capacity are approached (Funk et al., 1989; Foley et al., 1991). All three of these energy phosphates exert strong and independent positive modulation on important rate-controlling enzymes of the glycogenolytic pathway (Stanley and Connett, 1991; Lambeth and Kushmerick, 2002). As both ADP and AMP rise exponentially as whole-body maximum O₂ consumption rate () is approached, this energy phosphate control of mitochondrial oxidative phosphorylation appears consistent with the observation that mammalian muscle exponentially increases the recruitment of carbohydrate into the fuel supply as exercise intensity rises toward , (Brooks and Mercier, 1994; Roberts et al., 1996). At and above moderate exercise intensity, mammalian muscle remains critically dependent on carbohydrate even in the absence of high blood lactate or glucose concentrations (O'Brien et al., 1993), and actively suppresses fat oxidation at high ATP turnover rates (Romijn et al., 1995; Sidossis and Wolfe, 1996). Indeed, glycogen depletion precipitates the failure of mammalian muscle contractile output (Wahren et al., 1971; Baldwin et al., 1973; Fitts et al., 1975), despite the presence of a vast lipid energy reserve.

In profound contrast, birds fly at high percentages of , combusting fat at nearly the complete exclusion of carbohydrate (Rothe et al., 1987; Suarez et al., 1990), which indicates that the fuel selection of highly active avian muscle categorically differs from that of mammals. Moreover, avian skeletal muscle must achieve extremely high aerobic energy turnover to support flight and is equipped with a mitochondrial density 2–5 times that of mammalian skeletal muscle to do so. Pigeon pectoralis muscle provides an informative example: 300 g pigeons fly at a whole-body of roughly 200 ml min⁻¹ kg⁻¹ body mass (Rothe et al., 1987), and the pectoralis, which is about 10% of body mass (30 g), must account for well over half of this. This conservative estimation gives a muscle tissue of at least 3000 ml min⁻¹ kg⁻¹ muscle, almost all

of which goes toward the combustion of lipid, because whole-body RQ is ~0.72 (Rothe et al., 1987). Thus, while mammalian muscle transitions from a lipid-based fuel metabolism at rest to near-complete reliance on carbohydrate as an aerobic maximum perhaps 100-fold higher is approached, avian muscle rising far higher continues to select lipid as its oxidative fuel.

The mechanisms underlying this dramatically different fuel metabolism remain poorly understood. The state 3 (maximum) rate of mitochondrial oxygen consumption, J_O, of pigeon pectoralis mitochondria oxidizing pyruvate is roughly equal to that of palmitoyl-l-carnitine (Rasmussen et al., 2004), and the pioneering work of Suarez et al. showed similar trends in hummingbird pectoralis (Suarez et al., 1986). Data from pectoralis mitochondria of the house sparrow further confirm this avian pattern, as recently reported by our laboratory (Kuzmiak et al., 2012). As pyruvate (carbohydrate) and β-oxidation (fatty acid) pathways elicit equal rates of oxygen consumption, the selection of fatty acid fuel during avian flight cannot be simply explained by a higher catalytic potential for fat oxidation. Moreover, in vivo studies on pigeon and hummingbird clearly show that avian muscle exhibits substantial metabolic flexibility under certain conditions. From the complementary pigeon studies of Rothe et al. (Rothe et al., 1987) and Butler et al. (Butler et al., 1977) emerges the concept that a large glycogenolysis initiated by the onset of flight mobilizes carbohydrate carbon as lactate into the blood, as whole-body respiratory exchange ratio (RER) rises toward 1.00 (Butler et al., 1977; Rothe et al., 1987). Continued steady-rate flight over the next 30 min or so results in a progressively falling RER toward a steady-state value near 0.7 (Rothe et al., 1987). Another example of avian metabolic flexibility is the hovering hummingbird, in which whole-body RER promptly rises from near 0.7 to 1.0 when nectar feeding is allowed (Suarez et al., 1990). Mammalian skeletal muscle also exhibits metabolic flexibility as it adjusts fuel selection toward the oxidation of lactate (Mazzeo et al., 1986; Stanley et al., 1986) and glucose (Horowitz et al., 1999; Koutsari and Sidossis, 2003) when these carbohydrate fuels are made available.

The present study focused on the metabolism of mitochondria isolated from the locomotory muscle of a mammal [mixed hindlimb muscle of rat, Rattus norvegicus (Berkenhout 1769)] and a bird [pectoralis of house sparrow, Passer domesticus (Linnaeus 1758)]. Three fundamental characteristics were studied. (1) In avian muscle, very little is known about the sensitivity of respiratory control (e.g. the relationship between ATP free energy and or the K_m,ADP for respiration) or the extent to which the fuel supply available to mitochondria might influence this control sensitivity. We therefore determined this relationship between energy phosphate levels and mitochondrial O₂ consumption rate (J_O) in mitochondria provided with saturating levels of three oxidative substrates: pyruvate, glutamate and palmitoyl-l-carnitine, as well as combinations of these fuels. (2) With these same fuels, we assessed the extent to which fuels competed for oxidation (fuel competition) or simply added together to a higher flux (fuel additivity) at three experimentally fixed levels of ATP free energy (ΔG_ATP), −14.0, −13.5 and −13.0 kcal mol⁻¹. (3) Finally, it has been shown that a reduced (high NADH/NAD⁺ ratio) mitochondrial matrix strongly inhibits the β-oxidation of fatty acids (Lumeng et al., 1976). Our previous work has shown NADH cytochrome c reductase activity is 80% higher in sparrow versus rat skeletal muscle, indicating a greater maximal rate of electron transport chain (ETC) flux in birds than in mammals (Kuzmiak et al., 2012). Because an ETC capable of achieving a given at a low matrix NADH/NAD⁺ should provide a competitive advantage for fatty acid to contribute to the oxidative fuel supply, we predicted higher conductance for electron flow down the ETC in avian than in mammalian mitochondria. We therefore evaluated the kinetics of NADH oxidation by the ETC in sonicated mitochondria.

The results show that, with every fuel and fuel combination tested, avian muscle mitochondria control respiration across a wider range of ATP free energy than mammals or, stated in familiar kinetic terms, the avian K_m,ADP for respiration was significantly higher. Further, it is clearly shown in both mammalian and avian mitochondria, but especially in the bird, that muscle mitochondria provided with fatty acid suppress pyruvate oxidation more strongly than can be explained by the relative catalytic potentials of the two competing pathways. The findings also demonstrate that glutamate conversion to aspartate (the mitochondrial steps of the malate–aspartate shuttle) suppresses the oxidation of fatty acid in both types of mitochondria, but does so to a much greater extent in mammalian muscle. Finally, as predicted, the electron conductance of avian ETC exceeded that of the mammal, by over 50%.

Isolated mitochondria were of good functional integrity as indicated by high maximal respiration rates and energy coupling. With pyruvate (P) + malate (M) as substrates, state 3 J_O was 753±81 nmol mg⁻¹ min⁻¹ in rat and 548±9 nmol mg⁻¹ min⁻¹ in sparrow mitochondria (supplementary material Table S1). The corresponding ADP/O ratios were 2.83±0.15 and 3.00±0.04, and respiratory control ratios were 5.5±0.8 and 5.4±1.0 (supplementary material Table S1).

Typically, the relationship between ΔG_ATP and J_O is nearly linear. The slope of this relationship is one index of respiratory control sensitivity and can be described as the elasticity of mitochondrial J_O to ΔG_ATP. Near-linear relationships were observed in mitochondria oxidizing all fuel combinations in the present study (Fig. 1), and the slopes, or elasticities, often differed depending on species and substrate (Table 1). For example, rat mitochondria exhibited similar slopes with all substrates with the notable exceptions of palmitoyl-l-carnitine (PC) + M and glutamate (G) + M + arsenite (A) (Table 1). Specifically, elasticity with P+M was over 7 and 5 times greater than that with PC+M and G+M+A, respectively (Table 1). Sparrow mitochondria exhibited similar force:flow slopes for all substrates with two exceptions: the addition of glutamate to P+M significantly increased the slope by 72% and the addition of arsenite to G+M significantly reduced the slope by about 64% (Fig. 1, Table 1). Indeed, the addition of the 2-oxoglutarate (2-OG) dehydrogenase (OGDH) inhibitor arsenite to mitochondria oxidizing G+M decreased respiration in both species by ~54% at the calculated V_max J_O (Table 2). Sparrow mitochondria, however, were slightly less affected by this OGDH inhibition; at ΔG_ATP of −13.0 kcal mol⁻¹ the G+M+A/G+M ratio was significantly higher in sparrow than in rat (0.66±0.03 and 0.53±0.03, respectively).

Using the data set above, the K_m,ADP and V_max for respiration for each fuel combination were determined from Eadie–Hofstee analysis; these are reported in Table 2. In rat mitochondria, P+M V_max was 2.5 times higher than PC+M V_max and 2.3 times higher than G+M+A V_max (Table 2). In contrast, in sparrow P+M V_max was similar to PC+M V_max, and also to all other substrate combinations with one exception. Specifically, the addition of glutamate to P+M almost doubled V_max and similarly doubled the K_m,ADP (Table 2). In contrast, this addition of glutamate to P+M did not significantly increase V_max over P+M alone in rat mitochondria, with the average increase a modest 13% compared to the 96% increase in sparrow mitochondria (Table 2). Importantly, the V_max values reported in Table 2, determined using Eadie–Hofstee analysis of steady state conditions established with the creatine kinase energy clamp, are in excellent agreement with state 3 rates determined conventionally with saturating addition of ADP as seen in supplementary material Fig. S1, and also as recently reported by our laboratory (Kuzmiak et al., 2012).

In sparrow mitochondria, no fuel combination K_m,ADP was significantly different from the P+M value, with the exception of the much higher K_m,ADP observed with P+G+M noted above (Table 2). For all fuel combinations except P+PC+M, sparrow K_m,ADP values were significantly higher than rat (Table 2). In rat mitochondria, very low K_m,ADP values were observed for fat oxidation (PC+M) and for G+M+A, the mitochondrial steps of the malate–aspartate shuttle (Table 2).

Fuel J_O values across all energy states are reported as a percentage of V_max in Table 3. These values represent the mean for all fuel combinations at a given ΔG_ATP. At the three clamped energy states (ΔG_ATP=−14.0, −13.5 and −13.0 kcal mol⁻¹), average rat J_O were 60, 75 and 86% V_max, respectively, while corresponding sparrow values were 36, 53 and 69% V_max. J_O/V_max values for each fuel combination can be found in supplementary material Table S2.

In sparrow mitochondria under any energetic condition, pyruvate metabolism was strongly suppressed by the addition of any other fuel (Fig. 2). The addition of glutamate, fatty acid or their combination markedly decreased pyruvate utilization (Fig. 2B) and this suppression was even more pronounced when expressed relative to the oxygen consumption rate (Fig. 2D). In marked contrast, in the rat, suppression of pyruvate utilization was only observed at the highest energy state (ΔG_ATP=−14.0 kcal mol⁻¹), which is the energetic condition most similar to the low flux of rest, or when both G and PC were added (Fig. 2C).

The percentage decrease in pyruvate utilization resulting from the addition of palmitoyl-l-carnitine and/or glutamate is shown in Fig. 3A,B. The figure clearly shows that sparrow mitochondria reduce the pyruvate utilized, by roughly 70–80%, when either β-oxidation or the malate–aspartate shuttle, or a combination of the two, is fueled, regardless of energy state. Rat mitochondria also spare pyruvate from oxidation when other fuels are available, but, unlike the sparrow, the extent of pyruvate sparing is highly dependent upon both the energy state (Fig. 3B) and what alternative fuels are available (Fig. 3A). For example, at an energy state closer to that of resting muscle (ΔG_ATP=−14.0 kcal mol⁻¹), pyruvate utilization decreases 67±4% when both palmitoyl-l-carnitine and glutamate are also available. But at an energy state associated with moderate intensity locomotion (ΔG_ATP=−13.0 kcal mol⁻¹), the same additional fuel supply decreased pyruvate utilization 50±3%.

The average relative contribution of each fuel to oxygen consumption is presented in Figs. 4 and 5. In sparrow mitochondria, fatty acid consistently provided the majority of the fuel regardless of energy state, while pyruvate contributed very little (Fig. 5B–D). Glutamate competed slightly more successfully with fatty acids than pyruvate, but nevertheless failed to make more than a minor contribution under any energetic condition (Fig. 5B–D). In dramatic contrast, rat mitochondrial fuel metabolism was dominated by pyruvate and glutamate (Fig. 5A–D). While, as noted above, fatty acid out-competed pyruvate under high ATP energy (low flux) conditions, it is much less successful against glutamate, especially as ATP free energy was relaxed and flux rose (Fig. 5C). The delivery of ‘glycolytic pathway products’ (the combination of pyruvate and glutamate) to rat muscle mitochondria nearly abolished fatty acid metabolism under all energetic conditions (Fig. 5D).

A linear thermodynamic force:metabolic flow (ΔE_h:J_O) relationship was observed for NADH oxidation in sonicated mitochondria from both rat and sparrow (Fig. 6A). A slight loss of linearity, suggesting kinetic saturation, was seen at the highest ΔE_h and J_O values. Eliminating these data improved the fit of the regression and did not affect the differences between the species. The slope of the ΔE_h:J_O relationship was 1.7-fold higher in sparrow than in rat mitochondria (14,589±1811 versus 8797±1322, respectively); these slopes converted to proportional conductance units gave values of 94±12 versus 57±9 mS mg⁻¹, respectively (Fig. 6A,C). Additionally, the V_max of NADH oxidation was over 2 times greater in sparrow mitochondria (1074±118 versus 516±42 nmol O₂ min⁻¹ mg⁻¹, respectively) (Fig. 6B).

Three fundamental functional differences between avian and mammalian skeletal muscle mitochondria were revealed by these studies. (1) Mammalian mitochondria control respiration over a more negative range of ATP free energy. In more familiar kinetic terms, the K_m,ADP for respiration is lower in rat than in sparrow muscle mitochondria. This is particularly the case when fatty acid was the respiratory substrate. (2) Avian mitochondria primarily oxidize fatty acids and strongly spare pyruvate from oxidation across all energy states. The ability of a fuel to outcompete another for oxidation was not the simple consequence of a higher V_max. (3) Avian mitochondria possess much greater conductance for electron flow down the ETC, suggesting that a given can be achieved with a more oxidized matrix, i.e. higher NAD⁺/NADH ratio [at any given level of downstream backpressure exerted by the protonmotive force (Δp) and ATP free energy].

For almost all fuels and combinations, the K_m,ADP was higher in sparrow than in rat mitochondria (Table 2), indicating the sparrow controls respiration over a greater range of ΔG_ATP. At the three values of ΔG_ATP investigated in the present study, sparrow mitochondria were operating at a lower percentage of V_max (Table 3). It is therefore likely that ΔG_ATP can fall lower in sparrow than in rat mitochondria and still maintain useful contractile function. It is also likely that sparrow mitochondria energy production is activated by both [ADP] and [P_i]. The experiments in the present study employ a [P_i] of 10 mmol l⁻¹. In unpublished experiments from our lab (S.K.-G. and D. L. Gardner), we found that NAD-linked isocitrate dehydrogenase (IDH), which is activated by ADP in mammalian mitochondria, is dependent on [P_i] in sparrow mitochondria over the range 0–25 mmol l⁻¹. In fact, in the absence of P_i, IDH activity was undetectable, corroborating very low rates reported for whole muscle (Alp et al., 1976). This suggests an activation of NADH production by rising [P_i] before ΔG_ATP falls significantly.

It is also important to point out that compared with mammalian limb muscle that supports locomotion, avian pectoralis muscle has a much higher mass-specific ATP turnover rate. The maximal (state 3) mitochondrial oxygen consumption rate expressed per mg of mitochondrial protein is not very different between these species (Table 2; supplementary material Fig. S1) (Kuzmiak et al., 2012). As such, the energy requirements of flight can only be met because of the higher mitochondrial volume and protein density of avian myocytes. As pioneered by Weibel and colleagues and examined for avian species by our lab, a linear relationship exists between mitochondrial protein density and , demonstrating mammals and birds exhibit similar respiratory capacities per mitochondrial unit (Weibel et al., 1991; Weibel et al., 2004; Kuzmiak et al., 2012).

The mass-specific rate of energy demand required to support avian flight is enormous (Schmidt-Nielsen, 1984) and the energy density of carbohydrate in its stored, hydrated form is a mere ~1/10 that of lipid (Flatt, 1995; Jenni and Jenni-Eiermann, 1998). Many avian species are able to fly long distances without dietary energy replacement as a result of an almost exclusive reliance on fat oxidation (Rothe et al., 1987). Sparrow mitochondria provided with fatty acid strongly suppressed pyruvate utilization, by 69–78% (Fig. 2B,D), across the entire range of energy states tested. In rat mitochondria, pyruvate sparing was much more flux/energy state dependent: At low ATP turnover (ΔG_ATP=−14.0 kcal mol⁻¹), fatty acids inhibited pyruvate oxidation by 60%, consistent with the dominance of fatty acid oxidation in mammalian muscle at rest and during low intensity exercise (Fig. 2A,C) (Romijn et al., 1993; Roberts et al., 1996; Kelley and Mandarino, 2000; Willis et al., 2005). But as energy state was relaxed and the fuel oxidation rate rose, fatty acid addition spared only 30% of the pyruvate oxidized in its absence (Fig. 2A,C). Thus, in the mammalian mitochondria, as the ATP turnover rate increased, pyruvate combustion markedly rose in both absolute and relative terms. Viewed in the other direction, PC accounted for 16% of oxygen consumption at ΔG_ATP=−13.0 kcal mol⁻¹, while it accounted for 59% at −14.0 kcal mol⁻¹. These findings are similar to the pyruvate dehydrogenase (PDH) flux results of Ashour and Hansford, who used an ATP-utilizing reaction (hexokinase + glucose) to experimentally establish the in vitro rate of ATP turnover (Ashour and Hansford, 1983).

In mammals, fat oxidation rate rises with exercise intensity, as long as the intensity remains within the mild to moderate range (Romijn et al., 1993; Willis et al., 2005), i.e. at intensities below the lactate threshold. Across this range, myocyte NAD⁺/NADH also rises (Sahlin et al., 1987). At intensities exceeding roughly 60–70% , i.e. the lactate threshold, however, myocyte NAD⁺/NADH falls (Sahlin et al., 1987), as does the fat oxidation rate (Jones et al., 1980; Romijn et al., 1993). There is evidence that mitochondrial fatty acid oxidation is actively suppressed at these high intensities (Romijn et al., 1995), but the mechanism proposed, that of malonyl-CoA inhibition at carnitine palmitoyl transferase I (CPT-I), is not consistent with the observation that malonyl-CoA levels fail to change in the predicted direction (Winder et al., 1989; Odland et al., 1996).

An alternative mechanism is, however, supported by available evidence in a compelling fashion. The falling myocyte NAD⁺/NADH levels, associated with over tenfold increases in muscle lactate:pyruvate ratios (Sahlin et al., 1987), reflect a fundamental requirement for aerobic glycolysis: the shuttling of cytosolic NADH electrons into mitochondria. Almost 40 years ago, Lumeng et al. demonstrated in rat liver mitochondria incubated with saturating levels of palmitoyl-l-carnitine that the reconstruction of the malate–aspartate electron shuttle abolished fatty acid oxidation (Lumeng et al., 1976). Fat oxidation was also completely inhibited when the glycerol 3-phosphate (G3P) electron shuttle was fueled in liver mitochondria isolated from rats pre-treated with thyroid hormone, which upregulates hepatic mitochondrial G3P dehydrogenase. In both cases, β-oxidation pathway flux decreased as the matrix NADH/NAD ratio increased. Sahlin's laboratory has also suggested the possibility of control of β-oxidation flux by NADH/NAD⁺ (Mogensen and Sahlin, 2005). In the present study, as in the study by Lumeng et al. (Lumeng et al., 1976), fatty acid was provided to mitochondria as PC, which bypasses CPT-I entirely. Corroborating Lumeng et al., glutamate addition to fuel the mitochondrial steps of the malate–aspartate shuttle (Bookelman et al., 1979) suppressed fatty acid utilization concomitant with markedly increased J_O flux, which is consistent with an elevated matrix redox pressure. This pattern was dramatically the case in mammalian mitochondria, while it occurred to a minor extent in avian mitochondria. The findings above strongly suggest that fatty acid oxidation may be a more competitive substrate in sparrow mitochondria because of their very high ETC electron conductance.

In both species, electron flow down the ETC is a linear function of the driving force (ΔE_h) (Fig. 6), providing a compelling example of a pathway flux conforming to a linear thermodynamic force:metabolic flow relationship (Rottenberg, 1973; Van der Meer et al., 1980; Kushmerick et al., 1992; Glancy et al., 2013). V_max J_O in disrupted mitochondria oxidizing NADH was higher in sparrow than in rat, in agreement with previous avian mammalian comparisons (Rasmussen et al., 2004; Kuzmiak et al., 2012). This greater catalytic potential in avian ETC might facilitate at least two functional outcomes. First, during routine flight, sparrow pectoralis mitochondria could generate a high ETC flux at a low matrix redox potential, which would minimize inhibition of β-oxidation, allowing fatty acids to produce acetyl-CoA and inhibit pyruvate oxidation. Second, under conditions of rising lactate availability due, for example, to a rapid glycogenolysis proceeding in the same or another cell (Rothe et al., 1987), glycolytic products (pyruvate and the malate–aspartate shuttle) would additively fuel a yet higher rate of oxidative phosphorylation. Our data show that bird mitochondria oxidizing a single substrate increase V_max J_O by 50% when multiple substrates are added, in comparison with an 11% increase in rat mitochondria (Table 2).

Mitochondrial fuel oxidation and selection mirrored that of the whole body: in rat mitochondria the reliance on carbohydrate increased and the relative contribution of fat decreased as the rate of oxygen consumption increased, whereas fat dominated under all conditions in the sparrow. This indicates that fuel selection, at least in part, can be modulated at the level of the mitochondrial matrix when multiple substrates are present at saturating levels. As an increase in matrix redox potential has been linked to a suppression of palmitoyl-l-carnitine (Lumeng et al., 1976), we suggest a high ETC conductance relative to dehydrogenase activity in avian compared with rat mitochondria observed by our lab and others (Rothe et al., 1987) may result in a selective oxidation of fatty acids for energy in avian compared with mammalian mitochondria.

All procedures are in accordance with the guiding principles in the care and use of animals at Arizona State University. House sparrows (P. domesticus) were captured at a Livestock Auction (Phoenix East Valley, AZ, USA) by mist net the morning of each experiment. The sparrows, weighing 21–28 g, were fasted for 2–3 h during transport to the laboratory. Sprague–Dawley rats (R. norvegicus, weighing 300–400 g) and sparrows were killed with an overdose of CO₂. Pectoralis muscle was extracted and removed from the keel of a sparrow and the quadriceps femoris and triceps surae groups were extracted from the rats. These species were chosen as (1) a characterization of the oxygen consumption and reactive oxygen species production in these species was previously conducted (Kuzmiak et al., 2012), (2) rat hindlimb muscle provided adequate tissue for mitochondrial isolation and functional assessment, (3) the fuel mixture oxidized by the rat is representative of mammalian patterns, (4) the glucose and fatty acid uptake by sparrow muscle has been previously examined (Sweazea and Braun, 2005; Sweazea and Braun, 2006) and (5) there is no apparent correlation between body mass and mitochondrial oxygen consumption rate (Rasmussen et al., 2004). Extracted muscles were immediately placed in an ice-cold solution of (in mmol l⁻¹) 100 KCl, 40 TrisHCl, 10 Tris base, 5 MgCl₂, 1 EDTA and 1 ATP, pH 7.4 (solution I) (Makinen and Lee, 1968) and mitochondria were isolated as described previously (Kuzmiak et al., 2012). Protein concentrations of the final mitochondrial suspensions were 15.0±1.3 and 8.4±0.9 mg mitochondrial protein ml⁻¹ for sparrows and rats, respectively.

Respiration studies were performed in ~2.0 ml of a respiration medium (RM) containing (in mmol l⁻¹) 100 KCl, 50 MOPS, 20 glucose, 10 KPO₄, 10 MgCl₂ and 1 EDTA, with 0.2% BSA, pH 7.0. A Clark-type O₂ electrode was used to measure mitochondrial O₂ consumption (J_O) at 37°C (Rank Brothers, Cambridge, UK) as described previously (Messer et al., 2004). Other assay conditions and substrate additions are as noted in the legends of figures and tables.

Substrates provided to respiring mitochondria, either alone or in combination, included 0.5 mmol l⁻¹ P, 5.0 mmol l⁻¹ G and 10 μmol l⁻¹ PC. Importantly, PC is translocated directly into the mitochondrial matrix in exchange for carnitine, i.e. fat metabolism in these studies did not involve CPT-I. In all incubations, the citrate cycle was primed with 0.5 mmol l⁻¹ M. After malate priming, fuels were provided alone as P+M, G+M or PC+M, as well as in combination as P+G+M, P+PC+M, G+PC+M or P+G+PC+M. Additionally, G+M was added with 2.0 mmol l⁻¹ A, which specifically reconstructs the mitochondrial steps of the malate–aspartate shuttle. Arsenite inhibition of OGDH ensures that the MDH reaction Mal + NAD⁺ → OAA + NADH is the sole site generating reducing power, as glutamate transaminates with OAA to form 2-OG and aspartate and the formed 2-OG and aspartate then exchange with extra-mitochondrial malate and glutamate, respectively. The G3P shuttle was not assessed as previous work indicates it does not operate in sparrow mitochondria (Kuzmiak et al., 2012).

Timed incubations were carried out to determine fuel oxidation rates. The energy clamp components and substrates were added to the incubation medium and, after a few minutes equilibration period, the incubation was initiated by the addition of ~200 μg mitochondrial protein. A ‘pre-incubation’ sample (500 μl) was immediately withdrawn and quenched in 150 μl 25% HClO₄ and J_O was followed continuously until a second ‘post-incubation’ aliquot was similarly sampled and quenched. Incubations were ~10 min in duration; all durations were precisely measured and recorded times were used to calculate substrate fluxes. Acidified samples were promptly centrifuged for 1.0 min at 14,000 g and supernatants were neutralized with 2 mol l⁻¹ KOH + 0.5 mol l⁻¹ MOPS. An aliquot of this acid/neutralized extract was stored on ice to fluorometrically assay for pyruvate on the same day; the remainder was stored at −80°C for subsequent metabolite analyses.

For each fuel combination, J_O was plotted against ΔG_ATP to obtain the thermodynamic force:metabolic flow relationship, a measure of mitochondrial sensitivity to a respiratory signal. Using the same data set, K_m,ADP and V_max J_O were determined using Eadie–Hofstee analysis, plotting J_O against J_O/[ADP].

All metabolite assays were carried out by enzymatically linking substrate metabolism to the oxidation/reduction of NAD⁺/NADH by fluorescence changes (excitation λ 360 nm, emission λ 465 nm) in a 96-well microplate reader (TECAN GENios, Durham, NC, USA). Assays were run in duplicate using 10–40 μl of acidified/neutralized extract in a final assay volume of 200 μl. Standard curves were run daily. Pyruvate was assayed by a method adapted from a previous publication (Bergmeyer and Bernt, 1974). Malate, 2-OG and aspartate assays were adapted from Williamson and Corkey (Williamson and Corkey, 1969). Preliminary experimental incubations were assayed for OAA, citrate, isocitrate, lactate and alanine, and these metabolites did not show measurable accumulation with any substrate or substrate combination used (data not shown). Net metabolite disappearance and appearance rates were expressed per mg of mitochondrial protein in the incubation and can be found in supplementary material Table S3.

The net contributions of pyruvate, glutamate and palmitoyl-l-carnitine to the mitochondrial fuel supply were based on the principle of mass balance and the measured rates of O₂ consumption, pyruvate utilization, malate utilization and aspartate formation (see Fig. 7A). For these calculations, O₂ consumption rates and substrate utilization and formation rates were averaged for each condition so the data represent the mean for all experiments. In these fuel-utilization studies, 100–200 μg of mitochondrial protein was incubated in a 2 ml incubation system sealed from the environment. This quantity of mitochondria carries with it essentially no stored oxidizable substrates (Messer et al., 2004). All reducing equivalents available to the mitochondria were therefore experimentally controlled as fuel added in the form of malate, pyruvate, glutamate and/or palmitoyl-l-carnitine. In this highly controlled environment, the total O₂ consumed provides the total number of electron pairs extracted from added fuels. As seen in Fig. 7A, the pyruvate and palmitoyl-l-carnitine pathways converge at acetyl-CoA; thus, both require OAA to advance to citrate via the citrate synthase reaction. In turn, OAA is derived from malate, which was added to prime the citrate cycle at 0.50 mmol l⁻¹. Glutamate metabolism also utilizes malate priming carbon when it transaminates with OAA to form aspartate and 2-OG. Thus, the metabolism of all three substrates involves 2-OG formation, which can either be oxidized at OGDH or exchanged with extra-mitochondrial malate via the dicarboxylate carrier into a vast volume of respiration medium where it would accumulate. Net 2-OG accumulation was measured to assess the extent to which carbon either exited the mitochondrial matrix or advanced through OGDH into the second span of the citrate cycle. Please contact corresponding author for details on fuel utilization calculations. Briefly, the approach and assumptions on which these calculations are based can be fairly described as a process that rigorously assesses the contributions of pyruvate and glutamate oxidation to fuel metabolism, and thus uses O₂ consumption and mass balance to estimate the reducing equivalents generated by the β-oxidation of fatty acids.

NADH oxidation by sonicated mitochondria assessed the electron conductance of the ETC from complex I to complex IV. Frozen mitochondria were diluted 50% with 10 mmol l⁻¹ KPO₄, pH 7.0, and sonicated in three, 10 s bouts with a Branson Sonifier at 40% power; the mitochondria were rested on ice for 5 min between each bout. The assay began with 2.0 ml RM to which was added ~100 μg of sonicated mitochondrial protein followed by a 750 nmol bolus of NADH. The O₂ content of the RM was monitored continuously as it fell exponentially toward a zero slope plateau, indicating the complete oxidation of added NADH. Because of the 1:1 stoichiometry between NADH oxidation to NAD⁺ and atomic oxygen reduction to water, the O₂ electrode signal could be used to calculate J_O (hence J_e−, the electron current), as well as the ratios of NAD⁺/NADH and ½O₂/H₂O at any point along the progress curve. The total oxygen consumed was determined and the top and bottom 10% of the oxidation curve were excluded. The remaining field was divided into five equally spaced regions and in each both flux (J_O or J_e−) and the driving force (the voltage drop) down the ETC were analyzed.

Differences in the ΔG_ATP:J_O slope relationship, V_max J_O, K_m,ADP, and pyruvate utilization within a species were determined using a one-way ANOVA with a Tukey post hoc test with P<0.05. Independent t-tests were used to determine differences in V_max, K_m,ADP, percentage decreases in pyruvate utilization, as well as ETC conductance between the species with P<0.05.

The authors would like to thank Doree Lynn Gardner for her work on phosphate activation of isocitrate dehydrogenase activity.