Bioenergetic Analysis of Peroxisome Proliferator-activated Receptor γ Coactivators 1α and 1β (PGC-1α and PGC-1β) in Muscle Cells

Peroxisome proliferator-activated receptor γ coactivator (PGC)-1α is a coactivator of nuclear receptors and other transcription factors that regulates several components of energy metabolism, particularly certain aspects of adaptive thermogenesis in brown fat and skeletal muscle, hepatic gluconeogenesis, and fiber type switching in skeletal muscle. PGC-1α has been shown to induce mitochondrial biogenesis when expressed in muscle cells, and preliminary analysis has suggested that this molecule may specifically increase the fraction of uncoupled versus coupled respiration. In this paper, we have performed detailed bioenergetic analyses of the function of PGC-1α and its homolog PGC-1β in muscle cells by monitoring simultaneously oxygen consumption and membrane potential. Cells expressing PGC-1α or PGC-1β display higher proton leak rates at any given membrane potential than control cells. However, cells expressing PGC-1α have a higher proportion of their mitochondrial respiration linked to proton leak than cells expressing PGC-1β. Although these two proteins cause a similar increase in the expression of many mitochondrial genes, PGC-1β preferentially induces certain genes involved in the removal of reactive oxygen species, recently recognized as activators of uncoupling proteins. Together, these data indicate that PGC-1α and PGC-1β profoundly alter mitochondrial metabolism and suggest that these proteins are likely to play different physiological functions.

Mitochondria play a central role in metabolism by coupling cellular respiration to the production of ATP. However, this coupling is not perfectly tight. Indeed, it is estimated that approximately 20% of the standard metabolic rate in mammals is due to a leak of protons across the mitochondrial inner membrane in a manner that uncouples cellular respiration from ATP production, thereby generating heat (1). This cycle is called basal proton leak. In addition to this basal leak, there is an inducible leak of protons catalyzed by uncoupling protein 1 (UCP1) 1 in brown fat. Two close homologs of this protein have been discovered, UCP2 and UCP3 (2)(3)(4)(5)(6). Although the function of these homologs is not clear, recent work suggests that they might have an important role in the protection against reactive oxygen species (ROS) (7,8) and the modulation of cellular ATP levels, especially in insulin-secreting ␤ cells (9,10). Interestingly, none of the genetic studies using either knockout mice or mice overexpressing moderate levels of UCP2 and UCP3 show a significant effect of these proteins in determining standard metabolic rate by uncoupling cellular respiration (11,12).
Many changes in the cellular environment result in modulation of mitochondrial metabolism. Basal proton leak rate changes in response to hormonal status and metabolic depression (13)(14)(15)(16)(17)(18)(19). Also, small mammals with high standard metabolic rates have higher proton leak rates than large mammals with low standard metabolic rates (20,21). Furthermore, proton leak rates differ between phylogenetic groups; it is higher in endotherms than in ectotherms (22). The fact that mitochondrial functions can be altered in response to environmental stimuli is due to the execution of a coordinated program of genes expression. PPAR␥ coactivator-1␣ (PGC-1␣) has been shown to be a powerful regulator of multiple aspects of mitochondrial gene expression. Indeed, PGC-1␣ is cold-induced in brown fat and muscle where it plays a role in adaptive thermogenesis by regulating a complex program of increased mitochondrial biogenesis and uncoupled respiration via coactivation of peroxisome proliferator-activated receptors (PPARs), nuclear respiratory factor 1, and perhaps other transcription factors (23,24). Furthermore, transgenic mice expressing physiological levels of PGC-1␣ protein in skeletal muscle display an increased content of oxidative type I fibers compared with their wild-type counterparts (25).
To date, there are no bioenergetic mechanisms that might explain how PGC-1␣ increases uncoupled respiration, nor any studies examining a role for its closest homolog PGC-1␤ in mitochondrial metabolism. In this report, we measure the respiration and proton leak kinetics of C2C12 muscle cells expressing either PGC-1␣ or PGC-1␤. We report that PGC-1␤, like PGC-1␣, increases mitochondrial metabolism. Although both PGC-1␣ and PGC-1␤ increase proton leak, cells expressing PGC-1␣ have a higher proportion of their mitochondrial respiration linked to proton leak than those expressing PGC-1␤.

EXPERIMENTAL PROCEDURES
Cell Culture-C2C12 myoblasts were maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum at 37°C, 5% CO 2 . The differentiation of myoblasts into myotubes was induced by the standard protocol of allowing the cells to become confluent and changing the medium to Dulbecco's modified Eagle's medium with 2% horse serum. The differentiation into myotubes took 3-5 days. The cells were then infected with adenoviruses expressing GFP, PGC-1␣, or PGC-1␤. The GFP and PGC-1␣ viruses were made using the adeno-X expression system (Clontech). For PGC-1␤, a recombinant adenovirus containing the full-length cDNA encoding PGC-1␤ was prepared by subcloning of the cDNA into the pACCMV.pLpA vector and cotransfection of the recombinant plasmid with plasmid pJM17 into 293 cells using previously described methods (29). Two days after infection of myotubes, the cells were harvested to determine mitochondrial membrane potential, respiration rate, mitochondrial volume density, and cristae surface density.
The cells were isolated by washing with phosphate-buffered saline and trypsinizing for 5 min. Dulbecco's modified Eagle's medium with 2% horse serum was added to stop the reaction, and the cells were resuspended and spin twice at 1000 rpm for 5 min at room temperature. Finally, the cells were resuspended in Dulbecco's phosphate-buffered saline supplemented with 25 mM glucose, 1 mM pyruvate, and 2% bovine serum albumin. Cell viability was determined using trypan blue. In all experiments, viability was approximately 90% or higher.
Proton Leak Kinetics in C2C12 Myotubes-Proton leak kinetics analyses in cells were carried out as described in Ref. 30. The cells were incubated for 30 min with 0.2 Ci/ml [ 3 H]triphenylmethylphosphonium (TPMP), 0.2 M TPMP, and 1.5 M tetraphenylboron as carriers, as well as 2.5 g of oligomycin/10 6 cells. Controls were done to ensure that the oligomycin concentration was saturating for each group of cells. After this incubation period, the respiration rates and membrane potentials were determined. The first point of each proton leak curve represents the resting respiration rate and membrane potential of cells in the presence of oligomycin. Increasing amounts of myxothiazol (0.024, 0.048, and 0.072 M) were then added gradually to decrease membrane potential and respiration rate. Finally, 2.5 M myxothiazol and 2 M carbonylcyanide p-trifluoromethoxyphenylhydrazone were added to ascertain nonmitochondrial respiration. Nonmitochondrial respiration was subtracted from the total respiration to obtain mitochondrial respiration. To calculate membrane potential, we used the mitochondrial volume density values measured in the present study (see "Results"), cell volume of 4 l/10 6 cells (data not shown using the method described in Ref. 30), resting plasma membrane potential of Ϫ46 mV (31), and the following TPMP binding corrections: mitochondrial, 0.4; cytoplasmic, 0.2; and external medium, 0.7 (17,30).
Contribution of ATP Turnover and Proton Leak to Mitochondrial Respiration-The contribution of ATP turnover and proton leak to mitochondrial respiration was calculated from the results of the proton leak kinetics. ATP turnover represents the fraction of mitochondrial respiration sensitive to oligomycin. Proton leak represents the fraction of mitochondrial respiration that is insensitive to oligomycin.
Electron Microscopy-The samples were fixed for 1 h in a mixture of 2.5% glutaraldehyde, 1.25% paraformaldehyde, and 0.03% picric acid in 0.1 M sodium cacodylate buffer (pH 7.4), washed in 0.1 M cacodylate buffer, postfixed with 1% osmiumtetroxide/1.5% potassium ferrocyanide for 1 h, washed in water, and stained in 1% aqueous uranyl acetate for 30 min followed by dehydration in grades of alcohol (5 min at 70%, 5 min at 90%, and 2 ϫ 5 min at 100%). The samples were then infiltrated and embedded in TAAB Epon (Marivac Canada Inc., St. Laurent, Canada). Ultrathin sections (about 60 nm) were cut on a Reichert Ultracut-S microtome, picked up on to copper grids stained with uranylacetate and lead citrate, and examined in a JEOL 1200EX. The quantification of mitochondrial volume density and cristae surface density of cells expressing GFP, PGC-1␣, and PGC-1␤ was carried out as described in Ref. 32.
Isolation of Mitochondria from Transgenic Mice-Wild-type and transgenic mice expressing PGC-1␣ from the muscle creatine kinase promoter in the muscle tissues (transgenic line 31) were housed on a 12D:12N photoperiod cycle and fed ad libitum (25). The mice were sacrificed, and their whole leg muscle mass was excised and minced with razor blades. The minced tissue was diluted 1/10 (w/v) in isolation buffer containing 100 mM KCl, 50 mM Tris-HCl, 2 mM EGTA, 0.5% bovine serum albumin, pH 7.4, at 4°C and homogenized in a mortar with six passes using a medium tight pestle. The homogenate was centrifuged at 2000 ϫ g at 4°C for 5 min. The supernatant was collected and centrifuged at 10,000 ϫ g for 10 min. The supernatant was discarded, and the pellet was resuspended in isolation medium and centrifuged again at 10,000 ϫ g for 10 min. The supernatant was discarded, and the pellet was resuspended in 500 l of isolation medium. The protein concentration of the mitochondrial suspensions was determined using the bicinchoninic acid kit with bovine serum albumin as a standard.
Proton Leak Kinetics in Isolated Mitochondria-The proton leak kinetics in isolated mitochondria was determined as described in Ref. 30. Mitochondria were incubated (0.3 mg/ml) in assay medium containing 120 mM KCl, 5 mM KH 2 PO 4 , 3 mM Hepes, and 1 mM EGTA (pH 7.2 at room temperature) in a chamber thermostatted at 37°C using a recirculating water bath. The oxygen consumption rates, measured with a Clark-type electrode, and the membrane potential values, determined using a TPMP electrode, were recorded simultaneously. Rotenone (5 M), oligomycin (1 g/mg mitochondrial protein), and nigericin (80 ng/ml) were present at the beginning of each run. TPMP was added up to 1.3 M for calibration. Mitochondria were fed succinate (4 mM), and the inhibitor malonate was gradually added up to 1.3 mM to inhibit mitochondrial respiration and membrane potential. Finally, p-trifluoromethoxyphenylhydrazone (0.15 M) was added to determine the drift of the TPMP electrode, if any. The oxygen solubility of the medium was considered 406 nmol of oxygen/ml (33), and the TPMP binding correction was considered 0.4 (l/mg mitochondrial protein) Ϫ1 (30).
Northern Blot Analyses-Total RNA was isolated from C2C12 myotubes using Trizol (Invitrogen); 20 g of RNA was analyzed by electrophoresis and transferred on a membrane. The RNA blots were hybridized with specific cDNA probes.
Statistical Analyses-All of the statistical analyses were performed using Sigma Stat 2.0. Comparisons of the contribution of ATP turnover and proton leak to mitochondrial respiration between cells infected with GFP and PGC-1␣ or with GFP and PGC-1␤ were carried out with paired Student t test. Comparisons of mitochondrial volume density and cristae surface density values between GFP-, PGC-1␣-, and PGC-1␤-expressing cells were done with one-way analysis of variance and the a posteriori Tukey test.

Proton Leak Kinetics of Muscle Cells Expressing PGC-1␣ or
PGC-1␤-To study the functions of PGC-1␣ and PGC-1␤, these coactivators were expressed with adenoviral vectors in C2C12 muscle cells. These cells were used because skeletal muscle in vivo expresses both PGC-1s, but the cultured cells express much less of these proteins, making them an attractive system for studying PGC-1 function. As shown below, these infections achieved roughly equivalent levels of PGC-1␣ and PGC-1␤ mRNA. In addition, mRNAs for several well established target genes of PGC-1␣ (cytochrome c and UCP2) are activated equivalently in cells infected with PGC-1␣ or PGC-1␤.
We first measured the proton conductance in these cells, as well as cells expressing a control GFP. Proton conductance can be defined as the flow of protons (proton leak) across the mitochondrial membrane at a given membrane potential and represents the fraction of mitochondrial respiration that is not coupled to ATP production. Respiration, the consumption of oxygen, is used as a surrogate for proton leak (30). The highest point of a proton leak curve represents the respiration rate and membrane potential of cells/mitochondria in the presence of oligomycin, an inhibitor of the F 1 F 0 -ATP synthase (30). Then respiration and membrane potential are inhibited gradually by adding increasing amounts of myxothiazol, a specific inhibitor of the electron transport chain (30). As shown in Fig. 1 (A and  B), the respiration rate of PGC-1␣and PGC-1␤-expressing cells at any given membrane potential was substantially higher than GFP controls. For example, the respiration rate at 190 mV of PGC-1␣ cells and the paired GFP control was ϳ7 and 3 nmol of oxygen/min/10 6 cells, respectively, an increase of 2.5-fold (Fig. 1A). The respiration rate of PGC-1␤ cells and the paired GFP control at 172 mV was ϳ3 and 0.5 nmol of oxygen/min/10 6 cells, respectively, a 6-fold increase (Fig. 1B).
To compare the proton leak kinetics of the cells expressing PGC-1␣ and PGC-1␤, we expressed the respiration rates and membrane potentials of these cells as percentages of the initial respiration rate and membrane potential of paired GFP controls (Fig. 1C). The proton conductance of cells expressing PGC-1␤ was approximately three times higher than those expressing PGC-1␣ (Fig. 1C). Indeed, the respiration rate of PGC-1␤, when normalized to its paired GFP control, was at least twice that of PGC-1␣ cells at any given membrane potential (Fig. 1C).
Contribution of ATP Turnover and Proton Leak to PGC-1related Mitochondrial Respiration-Mitochondrial respiration is divided into two blocks: ATP turnover and proton leak. ATP turnover is the fraction of mitochondrial respiration coupled to ATP production and is sensitive to oligomycin. Proton leak is the fraction of mitochondrial respiration not coupled to ATP production and is insensitive to oligomycin. Metabolic efficiency represents the balance between the ATP turnover and proton leak blocks. Therefore, when total mitochondrial respiration is altered, metabolic efficiency can only be preserved if ATP turnover and proton leak are affected in a similar way. For example, if mitochondrial respiration doubles, metabolic efficiency will be preserved if both ATP turnover and proton leak double. We determined the proportion of mitochondrial respiration devoted to ATP turnover and proton leak in cells expressing PGC-1␣, PGC-1␤, and GFP. We used two different approaches to carry out this analysis; because each approach has its own limitation, together they provide a more reliable overview of the metabolic organization of these cells.
In the first calculation, we expressed the respiration rate of PGC-1␣, PGC-1␤, and GFP cells in the presence of oligomycin (from Fig. 1) as a fraction of their mitochondrial respiration (data not shown) without taking into account membrane potential values. The potential problem with this approach is actually that it does not take into account differing membrane potential values of the different cells. By this method, PGC-1␣expressing cells devote 37% of their mitochondrial respiration to proton leak, compared with 23% for paired GFP controls ( Fig. 2A). In other words, PGC-1␣ increases uncoupled respiration more than coupled respiration. This is in agreement with the initial study by Wu et al. (24) using C2C12 myotubes infected with retroviruses reporting that PGC-1␣ increases uncoupled respiration more than coupled respiration. In contrast, cells expressing PGC-1␤ display a fraction of mitochondrial respiration caused by proton leak similar to that of the GFP controls (Fig. 2B). Indeed, the ATP turnover and proton leak blocks increased coordinately in PGC-1␤-expressing cells so that they are almost as coupled as GFP control cells (Fig. 2B).
As a second approach, we took into account the membrane potential values and calculated the proportion of mitochondrial respiration devoted to proton leak at the resting membrane potential of PGC-1␣, PGC-1␤, and GFP cells. This is the membrane potential in the absence of oligomycin (data not shown). The membrane potential value of these cells was slightly higher in the resting state than in the presence of oligomycin, which is the highest point of the proton leak curves presented in Fig. 1. To obtain the predicted proton leak rate at the resting membrane potential of these cells, we fitted a linear curve through the proton leak points of PGC-1␣, PGC-1␤ cells, and paired GFP controls from Fig. 1 (A and B) (19). A caveat with The first point of each proton leak curve represents the resting mitochondrial respiration rate and membrane potential in the presence of oligomycin. The initial respiration rates and membrane potentials were decreased gradually by adding increasing amounts of myxothiazol (see "Experimental Procedures'' for details). A, comparison of the proton leak kinetics between C2C12 myotubes expressing GFP and PGC-1␣ (n ϭ 6). B, comparison of the proton leak kinetics between C2C12 myotubes expressing GFP and PGC-1␤ (n ϭ 5). C, comparison of the proton leak kinetics between C2C12 myotubes expressing GFP, PGC-1␣, and PGC-1␤. The results are from A and B. The respiration rates and membrane potentials for each panel were expressed as percentages of the resting respiration rate and membrane potential of the paired GFP control. this approach is that the proton leak rate value at the resting membrane potential of these cells must be predicted without being able to measure it experimentally. Using this method, we determined that cells expressing GFP, PGC-1␣, and PGC-1␤ have 37, 62, and 44%, respectively, of their mitochondrial respiration linked to proton leak. Although these values are higher than the ones determined above in absolute terms, the general conclusion remains the same: cells expressing PGC-1␣ are less efficient than GFP-or PGC-1␤-expressing cells. Whether PGC-1␤ significantly alters the fraction of respiration related to proton leak will require more detailed experiments.
Ultrastructural Analyses of Muscle Cells Expressing PGC-1␣ or PGC-1␤-Modification in mitochondrial volume density and changes in the intrinsic properties of mitochondria, such as membrane properties and cristae surface density, are key factors affecting proton conductance in cells (21). To determine whether the increased proton conductance of cells expressing PGC-1␣ or PGC-1␤ was associated with elevated mitochondrial volume density and/or cristae surface density, we carried out ultrastructural analyses of these cells and of GFP controls via transmission electron microscopy. To calculate mitochondrial volume density, a grid was laid on randomly selected micrographs, and the number of points falling onto mitochondria was expressed as a fraction of those falling onto the cell area (32). An analogous approach was used to determine the cristae surface density of individual mitochondrion (32). Muscle cells expressing PGC-1␣ and PGC-1␤ showed 25 and 75% increases in mitochondrial volume density, respectively, compared with GFP controls (Figs. 3A and 4A). Concerning the ultrastructure of individual mitochondrion, we observed that mitochondria from C2C12 myotubes expressing PGC-1␣ displayed a 10% increase in cristae surface density compared with GFP controls, which was statistically significant (Figs. 3B and 4B).

Proton Leak Kinetics of Mitochondria Isolated from Mice
Expressing PGC-1␣ Transgenically-Because cells expressing PGC-1␣ and PGC-1␤ displayed increased proton leak rates, we wanted to determine whether this effect was present in isolated mitochondria. However, performing this type of experiment has been very difficult with cultured cells because of the problems associated with large scale purification. Hence, we took advantage of the availability of mice expressing PGC-1␣ transgenically in skeletal muscle and compared the proton leak kinetics of mitochondria isolated from their leg muscle with that of wild-type mice. Importantly, these mice do not grossly overexpress this coactivator; they express PGC-1␣ in type II muscle fibers at the level ordinarily seen in type I fibers (25). Generally, the concentration of mitochondrial proteins from a given muscle mass was higher in transgenic mice than in wild-type animals (data not shown), suggesting that the ectopic expression of PGC-1␣ in vivo also leads to mitochondrial biogenesis. The highest point of a proton leak curve measured in isolated mitochondria represents the state 4 respiration rate and membrane potential. As shown in Fig. 5, mitochondria from transgenic mice displayed a slightly higher respiration rate than those from wild-type mice at any given membrane potential. In addition, the state 4 respiration rate was 50% higher in mitochondria from transgenic than wild-type mice, suggesting an increased substrate oxidation capacity (Fig. 5). Together, these results are consistent with the measurements performed in cells and demonstrate that PGC-1␣ can affect the intrinsic functional properties of mitochondria, notably increasing proton leak rate.
Mitochondrial Gene Expression in Cells Expressing PGC-1␣ or PGC-1␤-To gain possible insights into the molecular basis for the changes observed in respiration and proton leak, we examined the expression of several key mitochondrial compo- nents of the electron transport system (cytochrome c and ATPsynthase), the metabolism of ROS (manganese-superoxide dismutase and glutathione peroxidase-1), as well as uncoupling proteins (UCP2 and UCP3). The expression of cytosolic enzymes involved in the metabolism of ROS (glutathione peroxidase-1, ␥-glutamylcysteine synthetase light subunit, and ␥-glutamylcysteine synthetase heavy subunit) was also examined because the increased mitochondrial volume density and respiration of PGC-1␣ and PGC-1␤ cells could lead to elevated cellular levels of ROS. As shown in Fig. 6A, cells infected with these viruses expressed similar levels of PGC-1␣ or PGC-1␤ mRNA, and both coactivators induced mRNAs for the well studied PGC-1␣ target genes cytochrome c and UCP2 to the same extent. Specifically, the mRNA expression level of cytochrome c, ATP synthase, UCP2 and UCP3 (Fig. 6A), as well as manganese-superoxide dismutase and glutathione peroxidase-1 (Fig. 6B) were all higher in cells expressing PGC-1␣ and PGC-1␤ than in control cells (Fig. 6). The quantitative induction in the expression level of mRNA for certain markers, like ATP synthase and UCP3, in PGC-1␣ and PGC-1␤-expressing cells matched the 1.25-and 1.75-fold inductions in their mitochondrial volume density, respectively (Figs. 4 and 6). Others, like cytochrome c, UCP2, manganese-superoxide dismutase, and glutathione peroxidase-1, displayed higher inductions of at least 5-fold (Fig. 6). The fact that expression of these mRNAs displays a higher induction than the mitochondrial volume density suggests that each mitochondrion might have an elevated content of these proteins. Together, these results support the observation that myotubes expressing PGC-1␣ and PGC-1␤ ectopically have an increased mitochondrial volume density and also suggest that these cells have mitochondria with different intrinsic properties. Interestingly, PGC-1␤ cells displayed a higher expression of two cytosolic enzymes involved in the metabolism of ROS, ␥-glutamylcysteine synthetase light subunit, and ␥-glutamylcysteine synthetase heavy subunit, compared with PGC-1␣ and GFP cells (Fig. 6B). DISCUSSION The coactivator PGC-1␣ has been shown to induce mitochondrial biogenesis in cultured adipocytes as well as skeletal and cardiac myocytes (23,24,34). Initial studies using the inhibitor oligomycin suggested that PGC-1␣ might specifically induce an increased proton leak in skeletal muscle cells, but these studies had to be considered suggestive because a bona fide ascertainment of alterations in proton leak can only be determined by examining experimental and control cells at different mitochondrial membrane potentials. In this study, we have performed a much more complete and mechanistic analysis of the effects of PGC-1␣ and the related coactivator PGC-1␤ on the bioenergetics of mitochondria.
One basic question addressed here is whether PGC-1␤ has the capacity to induce mitochondrial biogenesis and increase respiration. As shown in Figs. 1B, 3A, and 4A, PGC-1␤ powerfully stimulates mitochondrial biogenesis and respiration. Indeed, it is more potent in this regard than PGC-1␣ at similar levels of mRNA, even though several genes are induced by both coactivators to a similar extent.
Another key issue addressed in this study is the fraction of mitochondrial respiration coupled or uncoupled in cells expressing PGC-1␣ or PGC-1␤. There are several possible explanations for the increased fraction of uncoupled respiration observed in the presence of PGC-1␣ versus PGC-1␤. First, it is necessary to examine how coupled and uncoupled respiration in particular can be regulated. A key factor that needs consideration with respect to respiration rate relates to the elevated mitochondrial volume density of these cells. We can account for the differences in mitochondrial volume density between cells expressing PGC-1␣ or PGC-1␤ and control cells simply by dividing the respiration rates of these cells by their respective mitochondrial volume density value. This calculation gives an indication of the respiration rate of mitochondria inside the cells. After correcting for differences in mitochondrial volume density, the total respiration rate of cells expressing PGC-1␣ or PGC-1␤ is similar to control cells. However, the proton leak rate (highest point of the proton leak curves in Fig. 1) of cells expressing PGC-1␣ remains higher than that of cells expressing PGC-1␤ or control cells, both of which have similar proton leak rates after correction. These data suggest the proton leak rate of individual mitochondria is higher in the presence of PGC-1␣ than PGC-1␤ and that mitochondrial volume density is not the only explanation for the higher fraction of leak-related respiration in cells expressing PGC-1␣.
Modifications in the intrinsic properties of mitochondria could explain higher proton leak rates in cells expressing PGC-1␣ versus PGC-1␤. Cells expressing PGC-1␣, but not PGC-1␤, displayed a statistically significant increase of 10% in cristae surface density (Figs. 3B and 4B), which would be expected to augment their proton leak by increasing the area of membrane across which protons can re-enter the matrix compartment. In support of the idea that elevated PGC-1␣ levels increase the proton leak of individual mitochondria, a separate experiment using mitochondria isolated from the skeletal muscle of mice expressing PGC-1␣ at the level found in type I muscle fibers revealed that these mitochondria have higher resting proton leak rate (highest point of proton the leak curves in Fig. 5) and proton conductance than those from wild-type mice. Therefore, modifications in the intrinsic properties of mitochondria play an important role in the difference between PGC-1␣ and PGC-1␤ for the proportion of mitochondrial respiration linked to proton leak.
Another possible difference between cells expressing PGC-1␣ and PGC-1␤ could relate to the function of the uncoupling proteins, that is the expression of UCP2 and/or UCP3, combined with altered levels of superoxide, activators of the UCPs. Cells expressing PGC-1␣ and PGC-1␤ both had similarly elevated expression levels of mRNA encoding UCP2 and UCP3 (Fig. 6). Intriguingly, only cells expressing PGC-1␤ had increased levels of two cytosolic genes, namely glutamylcysteine synthetase light subunit and glutamylcysteine synthetase heavy subunit, both involved in the removal/degradation of ROS (Fig. 6B). It is conceivable, or even likely, that both PGC-1␣ and PGC-1␤ action result in the generation of ROS via the activation of mitochondrial metabolism, thereby activating UCP2 and/or UCP3. However, it is also possible that PGC-1␤ is better at the removal of these ROS because mRNAs for two cytoplasmic enzymes playing a role in metabolism of ROS are induced by this coactivator but not by PGC-1␣. Whether regulation of ROS in the cytoplasm can influence superoxide on the matrix side of the inner mitochondrial membrane and regulate UCP2 or UCP3 is an open question (35). The relative ability of PGC-1␣ and PGC-1␤ to produce ROS and activate UCPs via this mechanism remains to be determined.
Another important aspect to consider in this analysis concerns ATP turnover, the fraction of mitochondrial respiration providing ATP for the ATP consumers. There are important differences regarding ATP turnover between cells expressing PGC-1␣ and PGC-1␤. Indeed, 75% of the increase in mitochondrial respiration in cells expressing PGC-1␤ was due to an increase in ATP turnover, and 25% was explained by an in- FIG. 5. Comparison of proton leak kinetics between skeletal muscle mitochondria isolated from wild-type mice and mice expressing PGC-1␣ transgenically in muscle fibers. The experiments were carried out in the presence of bovine serum albumin, which binds fatty acids, known activators of the UCPs. The UCPs are unlikely to contribute to leak in these experiments. The mitochondria were incubated in the presence of oligomycin, nigericin, rotenone, and TPMP. The mitochondria were fed succinate as substrate, and increasing amounts of malonate were added to gradually inhibit respiration rate and membrane potential. The experimental details are described under "Experimental Procedures." The first point of each proton leak curve represents the resting respiration rate and membrane potential of isolated mitochondria and is called the state 4 point. The closed squares represent mitochondria from wild-type mice; the open squares represent mitochondria from mice expressing PGC-1␣ transgenically in their muscle fibers (n ϭ 4).
FIG. 6. Northern blots of C2C12 myotubes expressing GFP, PGC-1␣, and PGC-1␤. The experiments were carried out in duplicate. mRNA isolation and Northern blots were performed as described under "Experimental Procedures." A, Northern blots of mRNA for UCP2, UCP3, and electron transport chain components. B, Northern blots of mRNA encoding enzymes involved in the metabolism of ROS. Mn-SOD, manganese-superoxide dismutase; Gpx1, glutathione peroxidase-1; Gamma-GCSl, ␥-glutamylcysteine synthetase light subunit; Gamma-GCSh, ␥-glutamylcysteine synthetase heavy subunit. crease in proton leak. For cells expressing PGC-1␣, the situation was completely reversed; only 25% of the increase in mitochondrial respiration in these cells was due to an increase in ATP turnover, and 75% was explained by an increase in proton leak. Together, these data indicate that the activity of the ATP consuming reactions was increased in cells expressing PGC-1␤, providing another potential influence related to the higher fraction of coupled respiration in these cells.
Overall, the data presented here indicate that the nature of the respiration induced by PGC-1␣ or PGC-1␤ is different. Indeed, cells expressing PGC-1␣ have a less efficient mitochondrial respiration than those expressing PGC-1␤ or GFP. This is particularly interesting in light of the fact that PGC-1␣ is induced in the cold in both brown fat and skeletal muscle, two key thermogenic tissues (23). In contrast, PGC-1␤, which does not lead to a particularly inefficient mitochondrial metabolism (present study), is not cold-inducible in these tissues (26). However, it is important to appreciate that PGC-1␣ is also induced in muscle by exercise and is expressed in type 1 muscle, muscle associated with resistance to fatigue. Also, PGC-1␣ can induce an increase in type 1 fibers in muscle composed mainly of type 2 fibers (25). Furthermore, PGC-1␣ is highly expressed in the heart, a tissue very dependent on ATP production. Certainly, mitochondrial uncoupling in these situations would be predicted to cause a reduced exercise tolerance in skeletal muscle and heart failure. These observations can be reconciled by an integrated view of the relative dependence of ATP turnover and proton leak on mitochondrial membrane potential (Fig. 7). In a state of physical rest, such as present in nonexercising muscle or cold-induced muscle or brown fat, a relatively low ATP demand will cause an increased mitochondrial membrane potential, which will in turn increase proton leak. In stark contrast, high ATP demand, expected upon the performance of muscular work, will slightly decrease membrane potentials resulting in a tighter coupling of metabolism because proton leak decreases rapidly as membrane potential is lowered. Said another way, PGC-1␣, by setting up a relatively uncoupled state at rest, probably provides a system where rapid ATP turnover, such as occurs during the performance of exercise, will effectively slow down excessive proton leak. PGC-1␤, in contrast, drives relatively less proton leak at rest, so it will provide more ATP when this is needed but cannot otherwise provide metabolic flexibility. Because both PGC-1␣ and PGC-1␤ are present in many tissues, the relative expression levels of these coactivators likely will play an important role in setting metabolic efficiency. FIG. 7. Model showing the dependence of proton leak on ATP demand. At low ATP demand, the membrane potential is high and an important fraction of protons re-enter the matrix side of the inner mitochondrial membrane by the proton leak pathway. At high ATP turnover, the membrane potential decreases, which reduces the entry of protons by the proton leak pathway in favor of the ATP synthase.