The methods for the generation of transgenic mice overexpressing PGC-1α-b in skeletal muscle were described previously ^[13]. The human α-skeletal actin promoter provided by Drs. E. C. Hardeman and K. L. Guven (Children's Medical Research Institute, Australia) was used to express PGC-1α-b in skeletal muscle. The transgenic mice (heterozygotes, BDF 1 background) and wild-type C57BL6 mice were crossed and the offspring (heterozygote and wild-type, born at the same period) were used for the experiments. Mice were cared for in accordance with the NIH Guide for the Care and Use of Laboratory Animals and our institutional guidelines. All animal experiments were conducted with the approval of the National Institute of Health and Nutrition Ethics Committee on Animal Research (approval ID: No 908, 1008, and 1111).

Mice were scanned with a Lunar PIXI mus2 densitometer (Lunar Corporation, Madison, WI), equipped for dual energy X-ray absorptiometry (DEXA) ^[17].

RNA preparation methods and quantitative real-time RT-PCR were performed as described previously ^[13]. The mouse-specific primer pairs used were as shown in Table 1.

Frozen skeletal muscle and heart was powdered under liquid nitrogen, and then homogenized in RIPA Lysis Buffer (Upstate Bitechnology, Lake Placid, NY) containing 1 mM sodium orthovanadate, 1 mM sodium fluoride, and the “Complete Mini, EDTA-free” protease inhibitor cocktail (1 tablet per 10 ml). After three cycles of freezing and thawing, the supernatant was separated by centrifugation at 20,400 g for 15 min at 4°C. Fifteen µg of protein from the supernatant was applied onto an SDS-PAGE gel. PGC-1α protein was identified by Western blotting with anti-PGC-1α (Rabbit polyclonal IgG) against the carboxyl terminus, 777–797 (Calbiochem, San Diego, CA).

The mitochondrial DNA copy number was normalized to the copy number of a gene contained in the nuclear genome. The mitochondrial DNA copy number in PGC-1α-b transgenic mice was expressed as a percentage of the copy number in control wild-type mice. Real-time PCR was used to estimate the copy number of specific genes in skeletal muscles as described previously ^[18]. CS activity was measured as described previously ^[19].

To obtain a pure and intact mitochondrial fraction, differential centrifugation was used ^[20]–^[22]. All procedures were performed at 0–4°C. Media used were as follows: medium I: 100 mM KCl, 5 mM MgSO₄·7H₂O, 5 mM EDTA, and 50 mM Tris·HCl, pH 7.4; medium II: medium I plus 1 mM ATP, pH 7.4; medium III: 220 mM sucrose, 70 mM mannitol, 10 mM Tris·HCl, and 1 mM EDTA, pH 7.4. Fresh skeletal muscle was immediately placed in ice-cold medium I and then blotted and weighed. Muscle was minced with scissors in 1 ml of medium II. Minced muscle was homogenized in 20 vol of ice-cold medium II using a Teflon pestle. The homogenate was treated with a protease (Sigma P5380, 0.025 ml/g tissue) for 5 min to digest the myofibrils. Addition of 10 ml of ice-cold medium II was used to diminish the action of the protease. Samples were spun at 5,000 g for 5 min, and the supernatant was removed. The pellet was resuspended in 10 vol of medium II and spun at 800 g for 10 min. The mitochondria found in the supernatant were spun at 12,000 g for 10 min. The pellet was washed in medium II and spun at 12,000 g for 10 min at 4°C. The pellet was resuspended in 1 µl medium III/mg tissue.

Mitochondrial oxygen consumption was measured polarographically using an Oxygraph oxygen electrode unit (Hansatech Instruments, Kings Lynn, UK) at 30°C. The respiration was measured in a medium containing 225 mM mannitol, 75 mM sucrose, 10 mM KCl, 10 mM Tris-HCl, and 5 mM KH₂PO₄, pH 7.2 ^[23]. For the measurement of pyruvate and malate oxidation, the following substrate combinations were used: 5 mM pyruvate+5 mM malate+1 mM ATP. Oxidative phosphorylation (state III respiration) was initiated by adding 125 nmol ADP in the presence or absence of 5 µg/ml oligomycin.

Samples of the tibialis anterior (TA) muscle at 10 weeks of age were frozen in liquid nitrogen-cooled isopentane. The identification of capillaries was performed using the anti-CD31 antibody (#553370, BD-Pharmingen, Franklin Lakes, NJ, USA), which recognizes platelet endothelial cell adhesion molecule-1(PECAM), a transmembranous glycoprotein strongly expressed by vascular endothelial cells ^[24], ^[25]. The acetone fixed slides were incubated for 1 h with anti-CD31 antibody and for 30 min with Simple Stain Mouse MAX PO (rat) (Nichirei, Tokyo, Japan). Visualization of antibody staining was achieved using 3,3′-diaminobenzidine tetrahydrochloride as the chromogen. Capillary density (number/mm²) and the capillary-to-fiber ratio were measured directly from images of micrographs projected on a screen.

To examine spontaneous physical activity, mice were housed in cages equipped with an infrared sensor, the activity was measured for 24 h, and the data were analyzed using a Supermex system (Muromachi Kikai, Tokyo, Japan).

Exercise capacity was determined by an exercise tolerance test during the peak oxygen uptake challenge as described previously ^[16] with modifications (no incline in our protocols). To determine peak VO₂, mice were placed in an air-tight treadmill (Muromachi Kikai) for 60 min at 0 m/min. The mice were then challenged at 10 m/min. The speed increased 2 m/min every 3 min until exhaustion. The mice ran until exhaustion, which is defined as remaining on the shocker plate for more than 10–15 s, as described previously ^[16]. Please make a note that the definition of exhaustion was not the same as used in other studies, in which volitional fatigue was taking into consideration ^[26], ^[27].

Open-circuit indirect calorimetry was performed with an O₂/CO₂ metabolism measuring system for small animals (MK-5000RQ; Muromachi Kikai). The system monitored VO₂ and VCO₂ at 3-min intervals and calculated the respiratory quotient (RQ) ratio (VCO₂/VO₂). Measurements were performed during the dark (from 19:00 to 7:00) or light (from 7:00 to 16:30) period under fasting conditions.

For measurement of VO₂ and VCO₂ during exercise, mice were allowed to acclimatize to the air-tight treadmill chamber (Muromachi Kikai) for 30 min, at which point VO₂ and VCO₂ were stable. Measurements were continued for another 30 min while mice were maintained in a sedentary state. Mice were then exercised as described above.

The substrate utilization rate and energy production rate were calculated using the formula used by Ferrannini where the rate of glucose oxidation (g/min) = 4.55VCO₂ (l/min)-3.21VO₂ (l/min)-2.87N(mg/min), the rate of lipid oxidation (g/min) = 1.67 (VO₂-VCO₂)-1.92N, and the rate of energy production (kcal/min) = 3.91 VO₂+1.10VCO₂-3.34N, and where N is the rate of urinary nitrogen excretion used to estimate protein oxidation ^[28]. However, considering that only a small portion of resting and exercise energy expenditure arises from protein oxidation, the contributions of protein oxidation were ignored ^[29].

Glycogen content in the gastrocnemius and liver was measured as glycosyl units after acid hydrolysis ^[30]. Blood glucose concentration was measured with a glucose analyzer (Glucometer DEX; Bayer Medical, Tokyo, Japan). Lactate levels were measured by Lactate Pro (Arkray, Kyoto, Japan). Blood samples were obtained by cutting the tail tip.

Data were analyzed by one-way ANOVA. Where differences were significant, each group was compared with the other by Student's t test (JMP 5.1.2; SAS, Campus Drive, Cary, NC, USA). In the exercise tolerance test, a Kaplan-Meier survival curve was obtained, and a comparison of groups was performed using the log-rank test (StatView 5.0). Statistical significance was defined as P<0.05. Values are shown as mean ± SE.

Skeletal muscle specific PGC-1α-b mice were made with a DNA construct containing the 5′-flanking skeletal muscle-specific regulatory region and the promoter of the human α-skeletal actin gene, and a cDNA encoding a PGC-1α-b ^[13]. Quantitative real-time RT-PCR showed that PGC-1α mRNA was expressed 29.2- and 26.8-fold higher in skeletal muscles of transgenic lines A and B, respectively, than in wild-type mice, but there was no difference in heart muscle (Fig. 1A).

PGC-1α protein was identified by Western blot analysis with an antibody against the carboxyl terminus of the PGC-1α-a protein, because the carboxyl terminus is the same in all PGC-1α isoforms (the amino terminus of PGC-1α-b protein differs from PGC-1α-a protein) ^[13]. In previous studies on mouse skeletal muscle and cultured cells, this antibody detected a 113 kDa protein that was considered the full length PGC-1α-a protein ^[31], ^[32]. In skeletal muscle taken from the transgenic mice in this study, increased labeling of the bands at 110 kDa (16.6-fold in line A and 24.8-fold in line B), 85 kDa (4.5-fold in line A and 5.5-fold in line B) and 45 kDa (5.8-fold in line A and 10.9-fold in line B) were detected with this antibody (Fig. 1B). In the transgenic mice, a decrease in the 40 kDa band was also observed. This might be due to the effects of alternate splicing of endogenous PGC-1α, as suggested in a previous study ^[32]. In heart, no significant change was observed between the genotypes, which confirmed that PGC-1α-b protein was not over-expressed in these transgenic mice. The increase in the reaction of several other proteins to this antibody might be due to post-translational processes of PGC-1α ^[33], its degradation products, or non-specific binding to unrelated proteins, although the precise nature of this is unknown.

In the transgenic mice, the expression of the PGC-1α target genes, COX2 and COX4, was also elevated in skeletal muscle but not in heart (Fig. 1A), confirming that expression of PGC-1α-b is specific to skeletal muscles.

Body weight, body composition and tissue weight were measured in male transgenic mice at 10 weeks of age. The body weight, lean body weight, fat weight, and fat% were not different between PGC-1α-b transgenic mice and wild-type littermates (Table 2). In PGC-1α-b transgenic mice, the weights of gastrocnemius, quadriceps, TA and extensor digitorum longus (EDL) were significantly lower than in wild-type littermates, however, this difference was not observed in the soleus.

The expression of mRNA in skeletal muscles of genes related to muscle fiber type and metabolism was determined by quantitative real-time RT-PCR (Fig. 2). Expression of myosin heavy chain (MHC) 1 (slow-twitch oxidative fibers, rich in mitochondria) and 2A (fast-twitch oxidative fibers) in quadriceps was increased only in PGC-1α-b transgenic mice (line B), but the increase in MHC1 was not significantly different. Compared to wild-type littermates, expression of MHC 2B (fast-twitch glycolytic fibers) was decreased to 37% in line A and 13% in line B, and the expression of MHC 2X (mouse fast-twitch oxidative fibers) was increased to 426% in line A and 462% in line B PGC-1α-b transgenic mice. These data suggested that expression of oxidative fibers was increased and glycolytic fibers was decreased in PGC-1α-b transgenic mice, similar to MCK-PGC-1α-a transgenic mice ^[34]. The expression of genes involved in glycogenolysis, such as phosphorylase kinase alpha (PHKA) 1 and muscle glycogen phsphorylase (PYGM), were significantly decreased to 20–30% of wild-type in both lines of transgenic mice. Glucose transporter (GLUT) 4 was decreased to 75% in both lines of transgenic mice. The key enzymes for glycolysis, such as muscle phosphofructokinase (PFKM), 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase (PFKFB) 3, and muscle pyruvate kinase (PKM) 2, were decreased significantly in the transgenic mice (the decrease of PFKM mRNA in line A transgenic mice was not significant), suggesting production of pyruvate was decreased in skeletal muscle that overexpressed PGC-1α-b. Pyruvate dehydrogenase kinase (PDK) 4 expression was increased only in line A transgenic mice. On the other hand, the expression of genes encoding proteins involved in fatty acid transport and fatty acid oxidation, such as lipoprotein lipase (LPL), CD36, fatty acid transport protein (FATP) 1, plasma membrane fatty acid binding protein (FABP-pm), fatty acid binding protein (FABP) 3, carnitine palmitoyltransferase (CPT) 1 and medium chain acyl-CoA dehydrogenase (MCAD), was higher in the transgenic mice. Myoglobin (Mb) and vascular endothelial growth factors (VEGFs) were also increased in the transgenic mice. Because VEGF-B might increase endothelial lipid uptake via stimulation of fatty acid transporter expression in endothelial cells ^[35], the increase in VEGF-B and fatty acid transporters in skeletal muscle overexpressing PGC-1α-b could enhance the rate of fatty acid transport from the blood stream into skeletal muscle. Neuronal nitric oxide synthase (nNOS) was decreased in the transgenic mice. Endothelial NOS (eNOS) was increased in the skeletal muscle overexpressing PGC-1α-b, suggesting that the numbers of endothelial cells derived from capillaries in the skeletal muscle tissue was increased. One of the NR4A orphan nuclear receptor proteins, Nur77, which regulates expression of genes linked to glucose metabolism ^[36], was significantly decreased in the transgenic mice.

To estimate the numbers of mitochondria in the skeletal muscle from PGC-1α-b transgenic mice, the mitochondrial encoded-DNA (COX2) copy number relative to nuclear encoded-DNA (COX4) copy number and the CS activity were measured. The ratio of COX2 copy number to COX4 copy number in PGC-1α-b transgenic mice was higher than in wild-type mice (TA, 2.1-fold in line A and 2.4-fold in line B; EDL, 2.3-fold in line A and B; soleus, 1.4-fold in line A and B) (Fig. 3A). CS activity in the gastrocnemius (mixture of red and white fibers) and EDL (white fibers) of PGC-1α-b transgenic mice was increased 3.7- and 3.8-fold in line A and 4.0- and 3.8-fold in line B, respectively, relative to wild-type mice. Because CS activity of the soleus (red fibers) was 1.7-fold higher than those of the gastrocnemius and EDL in the wild-type, the activity in the soleus of PGC-1α-b transgenic mice was increased only 1.7-fold in line A and 1.6-fold in line B relative to wild-type mice. These data suggested that over-expression of PGC-1α-b in skeletal muscle increased the biogenesis of mitochondria and that this increase was larger in white glycolytic muscle than red oxidative muscle.

To study the function of the skeletal muscle mitochondria themselves, we compared the respiration rates ( = oxygen consumption) in isolated mitochondria from gastrocnemius, quadriceps, and TA of PGC-1α-b transgenic mice with those in wild-type mice. Both lines of PGC-1α-b transgenic mice showed almost the same respiration rate as wild-type on a protein basis of a mitochondria-rich fraction under conditions where the mitochondrial substrate and ADP were not limiting (Fig. 3B), which mimicked state III respiration ^[37]. To measure uncoupled respiration, oligomycin, an inhibitor of the F₁F₀-ATP synthase, was added to inhibit oxidative phosphorylation prior to measurement of respiration rates. Oligomycin-insensitive respiration ( = uncoupling) constituted about 20% of the total respiration in wild-type and PGC-1α-b transgenic mice. CS activity in the isolated mitochondrial fraction was not different between wild-type and PGC-1α-b transgenic mice.

These data suggested that the number of mitochondria was increased in skeletal muscle from PGC-1α-b transgenic mice and that the mitochondrial oxidative phosphorylation was the same as in wild-type.

Exercise promotes angiogenesis ^[38]–^[42]. PGC-1α-b might be a candidate for inducing angiogenic processes because transfection of PGC-1α-b into differentiated primary skeletal myocytes isolated from mice increased VEGF and PDGF-B mRNA expression ^[12]. This causal relationship was also observed in vivo. Overexpression of PGC-1α-b in skeletal muscle increased VEGF-A mRNA expression (Fig. 2). Furthermore, immunohistochemical analysis of TA muscle was performed to investigate whether PGC-1α-b induces angiogenesis in vivo (Fig. 4). Each section was analyzed in superficial and deep regions, and 209–774 and 476–1670 of CD31 positive capillaries were identified in each photograph of the superficial and deep regions, respectively. The capillary-to-fiber ratio was increased 2.4–2.6 and 1.3–1.4 fold in the superficial and deep regions of PGC-1α-b transgenic mice, respectively. Because the superficial region in TA muscle from wild-type mice was rich in white glycolytic fibers and had a lower capillary density, it was suggested that PGC-1α-b-induced angiogenesis in vivo occurred especially in regions where white glycolytic fibers changed to red oxidative fibers.

Increases in mitochondria and capillary numbers in skeletal muscles in transgenic mice may alter substrate utilization. To examine this possibility, 8 week-old males of both sedentary PGC-1α-b transgenic mice and wild-type littermates were subjected to fasting and their oxygen consumption measured and RQ ratio calculated. The oxygen consumption and RQ ratio for PGC-1α-b transgenic and wild-type mice did not differ during the dark cycle (feeding period) or light cycle (sleeping period) (Table 3). Daily activity levels were not different between PGC-1α-b transgenic mice and wild-type littermates. Body weight was similarly reduced in the two groups after 24-hours of fasting (if energy expenditure was enhanced, fasting-induced weight reduction increased), suggesting that over-expression of PGC-1α-b in skeletal muscles did not affect fatty acid oxidation during fasting.

These data suggested that increases in mitochondria and capillaries in skeletal muscles did not affect substrate utilization in the sedentary state.

The ability of PGC-1α-b transgenic mice to tolerate a bout of exercise might be altered. To examine this possibility, mice were started off running on a treadmill at 10 m/min, and then the speed was increased by 2 m/min every 3 min until exhaustion (Fig. 5A). PGC-1α-b transgenic mice could run for a significantly longer duration with higher exercise-intensity than wild-type littermates (P = 0.002), although the weights of gastrocnemius, quadriceps, TA and EDL were significantly lower than in wild-type littermates (Table 2). Wild-type littermates ran for a duration of 47 min at a maximum speed of 40 m/min (total distance was 1.2 km), whereas PGC-1α-b transgenic mice ran for a duration of 64 and 66 min at a maximum speed of 54 and 56 m/min in lines A and B (total distances were 2.2 or 2.4 km), respectively (Fig. 5A).

To investigate the fuel utilization during exercise and the peak oxygen consumption, VO₂ and VCO₂ were monitored simultaneously until mice were exhausted. VO₂ and VCO₂ were increased gradually as speed increased until exhaustion in every group of mice (data not shown). In wild-type mice, the RQ ratio during exercise, which indicated fuel utilization, was increased as the treadmill speed increase, suggesting that glucose oxidation became predominant with increasing exercise intensity (Fig. 5B). However, in both lines of PGC-1α-b transgenic mice, an increase in the RQ ratio and glucose oxidation with increasing treadmill speed was not observed. The calculated level of lipid oxidation during exercise was always higher in PGC-1α-b transgenic mice than wild-type mice, suggesting that lipid oxidation was the predominant fuel source for exercise in the transgenic mice. Both lines of transgenic mice demonstrated a 20% increase in their peak oxygen uptake compared to wild-type mice (Fig. 5C).

Because of the difference in maximum running speed between wild-type and transgenic mice, it is possible that increased fatty acid oxidation in transgenic mice was due to their increased exercise capacity. However, when the data are expressed as the percentage of maximal speed, PGC-1α-b transgenic mice still showed a lower RQ ratio and higher levels of calculated fat oxidation (Fig. 5D).

These data indicated that PGC-1α-b transgenic mice show increased fat oxidation and decreased carbohydrate oxidation at either the absolute exercise intensity, or the same relative exercise intensities.

Lipid oxidation was the predominant fuel source for exercise in PGC-1α-b transgenic mice, whereas consumption of carbohydrate during exercise was reduced. Glycogen concentration in skeletal muscles and blood lactate concentration were measured at rest and 30 min after the exercise tolerance test, a time when the shift of fuel usage to glucose was manifested. At rest, the glycogen content in the gastrocnemius in line A and line B PGC-1α-b transgenic mice was 1.9- and 2.2-fold greater than in wild-type mice (Fig. 6). At the 30 min mark of the exercise tolerance test, skeletal muscle glycogen content was decreased to 52% in wild-type mice but no significant decrease was observed in transgenic mice. The liver glycogen content was not significantly changed between rest and 30 min of exercise in either wild-type or transgenic mice (data not shown). The blood lactate concentration at 30 min from the start of exercise was increased in wild-type mice, however this increase was not observed in transgenic mice (Fig. 6). Furthermore, an increase in blood lactate concentration at exhaustion was not observed in the transgenic mice (data not shown), also suggesting that glycogen in skeletal muscle was not used preferentially during exercise in transgenic mice. The blood glucose concentration did not differ significantly between the two groups of mice (data not shown).