Previous studies using gain-of-function strategies have shown that the coactivator PGC-1α is capable of coactivating an array of transcription factors involved in energy metabolic processes including fatty acid oxidation, electron transport, and oxidative phosphorylation [6
]. Forced expression of PGC-1α triggers mitochondrial biogenesis by activating a complex circuitry of factors including NRF-1, NRF-2, and the orphan nuclear receptor estrogen-related receptor α [23
]. However, gain-of-function strategies cannot determine whether PGC-1α is essential for critical energy metabolic processes including mitochondrial biogenesis and function. Using targeted gene deletion in mice, we show here that PGC-1α is not essential for normal embryologic development or the fundamental events of mitochondrial biogenesis. However, several lines of evidence support the conclusion that PGC-1α is necessary for the programs that regulate postnatal mitochondrial function and cellular energy metabolism, processes that equip the organism for the energy metabolic rigors of the postnatal environment. First, mitochondrial volume density is diminished in slow-twitch skeletal muscle of PGC-1α−/−
mice. Second, mitochondrial respiratory capacity is modestly but significantly altered in skeletal muscle and liver of PGC-1α−/−
mice. Third, the growth of heart and soleus muscle, tissues with high reliance on mitochondrial energy production, is blunted. Fourth, control of body fat mass is abnormal in the PGC-1α−/−
mice. Finally, PGC-1α−/−
mice do not respond normally to a variety of physiologic and dietary stresses known to increase oxidative energy demands. Taken together, these results strongly suggest that PGC-1α is necessary for the terminal stages of mitochondrial maturation necessary to meet the energy demands of the postnatal environment.
Extensive phenotypic analyses demonstrated that mice lacking PGC-1α are unable to cope with physiologic stressors relevant to postnatal survival. For example, a skeletal muscle phenotype was unveiled in PGC-1α−/−
mice under conditions in which energy supply becomes limiting. This was most clearly demonstrated by the profound abnormalities exhibited by PGC-1α−/−
mice with exercise-to-exhaustion and repetitive muscle stimulation studies. Similarly, cardiac performance of PGC-1α−/−
mice was compromised following severe exertion. This effect was largely due to an abnormal heart rate response. The basis for the observed abnormalities of cardiac heart rate, including a blunted response to β-adrenergic stimulation, is unknown, but could be related to the effects of late-stage growth arrest and corresponding derangements in energy metabolism on sinus node function. PGC-1α was first identified as a coactivator in BAT [2
]. Indeed, we found that exposure of the PGC-1α−/−
mice to cold, another relevant physiologic stress, resulted in an untoward drop in core body temperature consistent with an abnormality in thermogenesis despite normal cold induction of UCP-1 mRNA in BAT. Studies with a β3
-adrenergic agonist confirmed that the peak oxygen consumption rate in thermogenic tissue is diminished in PGC-1α−/−
mice. We propose that the thermogenic phenotype is related to reduced capacity for mitochondrial respiration in BAT. Interestingly, this phenotype was only evident during a rather narrow window of postnatal life. Animals at an older age did not exhibit cold intolerance, possibly due to the insulating properties of increased body mass. Collectively, these results demonstrate the importance of PGC-1α as a key transducer of physiologic stimuli to the control of energy metabolism.
The observation of fasting-induced hepatic steatosis is another example of the inability of PGC-1α−/−
mice to respond to postnatal environmental metabolic demands. Following short-term starvation, we found that the PGC-1α−/−
mice developed marked hepatocyte triglyceride accumulation. Further analysis revealed that palmitate oxidation rates were reduced in hepatocytes isolated from the PGC-1α−/−
mice, which would predispose to lipid accumulation. Surprisingly, the reduction in fatty acid oxidation rates in PGC-1α null hepatocytes was not due to altered expression of PGC-1α/PPAR target genes involved in mitochondrial fatty acid oxidation. However, mitochondrial respiratory rates were diminished. In addition, we found that triglyceride synthesis was abnormally activated, and the expression of genes encoding SREBP-1c and SCD-1, key proteins in the hepatic lipogenic pathway, failed to be appropriately down-regulated in fasted PGC-1α−/−
mice. The mechanism involved in this latter finding is unknown. Indeed, the relative contribution of increased triglyceride synthesis rates to the steatotic phenotype cannot be fully discerned from our data, given that this response could reflect the direct effects of PGC-1α deficiency on target genes or a secondary compensatory response to hepatocyte fatty acid accumulation. Consistent with the former possibility, recent evidence indicates that PGC-1α coactivates the nuclear receptor FXR, a negative regulator of SREBP-1c expression and triglyceride synthesis [33
]. We conclude that reduced hepatocyte mitochondrial respiratory capacity, and possibly activation of lipogenic programs, result in hepatocyte triglyceride accumulation in the context of increased hepatic delivery of fatty acids such as occurs with fasting.
We found that after 18 wk of age, female PGC-1α−/−
mice exhibit a mild but significantly abnormal weight increase associated with increased fat stores. Lean mass was unchanged at the time points examined. With further aging, a modest but significant increase in body fat was also noted in male PGC-1α−/−
mice (unpublished data). The basis for the observed abnormalities in weight control is unknown. We did not find differences in food intake or activity levels in female PGC-1α−/−
mice. It is possible that a reduction in systemic energy utilization, related to the mitochondrial dysfunction, leads to increased fat mass and weight gain in the PGC-1α−/−
mice. Interestingly, an association between PGC-1α
gene polymorphisms and obesity in humans has been recently reported [26
]. Clearly, future studies of male and female PGC-1α−/−
mice in pure-strain backgrounds over a range of ages will be necessary to fully investigate the observed abnormalities in weight control and fat distribution.
We did not find evidence for glucose intolerance or insulin resistance in the PGC-1α−/−
animals on standard chow. Moreover, female PGC-1α−/−
mice were more glucose-tolerant and insulin-sensitive than PGC-1α+/+
controls when consuming a high-fat diet. These findings are surprising, given the results of several recent studies demonstrating reduced expression of PGC-1α in human diabetic skeletal muscle [24
]. It is certainly possible that compensatory metabolic regulatory mechanisms have been activated in the PGC-1α-deficient mice, accounting for this observation. Alternatively, PGC-1α could serve as a coactivator of factors that mediate diet-induced insulin resistance. Consistent with this notion, we and others have shown that mice lacking the PGC-1α target PPAR exhibit resistance to diet-induced glucose intolerance [21
Histologic surveys of the PGC-1α−/− mice revealed ultrastructural abnormalities in the central nervous system. Inspection of sections prepared from the brains of PGC-1α−/− mice revealed patchy areas of microvacuolation in the pyramidal neurons of the cerebral cortex, accompanied by a mild increase in the number of astrocytes in the basal ganglia. The basis for this interesting but relatively nonspecific finding is unknown. It is possible that PGC-1α plays an important role in lipid metabolism related to membrane synthesis. Alternatively, the normal process of cellular debris turnover could be altered due to a defect in the energetics of the microglial component of the central nervous system. Although overt neurologic dysfunction was not apparent in PGC-1α−/− mice during the first 6 mo of life (no group differences were found on five of six sensorimotor tests), the PGC-1α−/− mice showed clear deficits on the inverted screen test. These deficits are likely due to impaired muscle strength in the PGC-1α−/− mice, but contributions by peripheral or central nervous system determinants (or both) could be contributory. Moreover, evidence of altered emotionality from the 1-h locomotor activity test also suggests the possibility of altered brain function in PGC-1α−/− mice. It will be of interest to determine whether the neurologic abnormalities contribute to the systemic metabolic abnormalities of the PGC-1α null mice.
During the preparation of this manuscript, Lin et al. reported an independent mouse line in which the PGC-1α gene was targeted [37
]. Phenotypic comparison of the our PGC-1α-deficient line with that of Lin et al. reveals a number of similarities and several interesting differences. Both PGC-1α-deficient lines exhibit cold intolerance, reduced hepatocyte respiration rates, and neurologic lesions. However, a number of interesting differences are notable. First, in contrast to Lin et al., the PGC-1α−/−
mice described here do not exhibit any postnatal mortality. Second, we did not find evidence for a defect in gluconeogenesis based on fasting blood glucose levels (unpublished data). In addition, whereas Lin et al. found an abnormal expression profile for
CCAAT-enhancer-binding protein β and δ and the gluconeogenic genes encoding phosphoenolpyruvate carboxykinase and glucose-6-phosphatase at baseline and with fasting in the PGC-1α−/−
mice, we did not (unpublished data). Third, we found evidence for an age-related increase in body fat in PGC-1α−/−
mice (females earlier than males), whereas Lin et al. identified a male-specific resistance to diet-induced obesity and insulin resistance. We have also found that male PGC-1α-deficient mice are somewhat protected against diet-induced obesity (Figure S4
). However, we observed that the insulin-sensitive phenotype of the female PGC-1α−/−
mice occurred in the context of normal weight gain with high-fat diet. These latter results indicate that the insulin-sensitive phenotype of PGC-1α−/−
mice cannot be fully explained by a lean phenotype. Of interest, mice lacking the nuclear receptor estrogen-related receptor α, a known target of PGC-1α, exhibit resistance to diet-induced obesity similar to that of male PGC-1α null mice [38
]. Fourth, the PGC-1α−/−
mice described here exhibit a dramatic fasting-induced hepatic steatotic phenotype, whereas the Lin et al. mouse does not. Fifth, Lin et al. found a neurologic phenotype in males characterized by hyperactivity, whereas the PGC-1α−/−
mice described here show reduced locomotor activity. However, it should be noted that we did not study activity levels over an extended period of time in males as did Lin et al., so it is possible that our findings reflect an emotional disturbance that manifests only when the animals are placed in a new environment. Finally, we report significant skeletal muscle and cardiac functional abnormalities (although the report by Lin et al. did not address these phenotypes, so this may not represent a true difference).
The reasons for the interesting differences between the two PGC-1α-deficient mouse lines are not clear. It is possible that distinct genetic backgrounds related to hybrid strains confer different degrees of secondary compensatory responses. In addition, the incompletely penetrant postnatal mortality noted in the PGC-1α−/− mice reported by Lin et al. could have resulted in a selection bias toward greater levels of compensatory responses in liver and other tissues in the surviving group. It is also possible that the method of gene targeting led to different phenotypes. Lin et al. generated PGC-1α−/− mice by Cre recombinase-mediated excision of exons 3–5 in oocytes. The PGC-1α−/− mice described here were generated by a targeting event that involved a 3′ homologous recombination leading to an insertion of the targeting vector including an extra exon 3 between exons 5 and 6. The exon 3 insertion, which was confirmed by RT-PCR, results in a mutant transcript that encodes a truncated protein. We were unable to detect normal transcript containing an exon 5–6 border, indicating that the targeting was accurate and complete. In addition, we could not detect full-length or smaller PGC-1α proteins by Western blotting. However, we cannot exclude the possibility that the sensitivity of the immunoblotting was not high enough to pick up a small amount of mutant (truncated) PGC-1α protein that could have some activity, given that it would contain nuclear receptor-interacting domains and the amino-terminal activation domain. If small amounts of PGC-1α activity are present in the mice reported here, it could explain some of the observed differences between the models. However, the bulk of data presented here support the conclusion that the PGC-1α−/− mice described are completely deficient in PGC-1α. Future direct comparison of the two mouse lines in pure background strains will be of interest.
In summary, this body of work provides evidence that PGC-lα is critical for the adaptive responses necessary to meet postnatal energy demands. Our results also suggest a broader role for inducible transcriptional coactivators such as PGC-1α in transducing cellular signals triggered by physiologic and developmental cues to the transcriptional control of energy metabolism and other dynamic cellular processes. In this regard, the inducible coactivator PGC-1α serves as a transcriptional “booster” to augment the capacity of downstream metabolic pathways critical for metabolic maturation and postnatal growth. Indeed, although PGC-1α null mice survive in the protected environment of the laboratory, our results indicate that in the rigors of a typical external environment, PGC-1α would be necessary for survival. Lastly, we propose that the PGC-1α−/− mice should serve as a useful murine model to investigate the role of altered energy metabolism in obesity, diabetes, hepatic steatosis, and diseases of the heart, skeletal muscle, and central nervous system.