Simply put, diabetes occurs as a result of an absolute or relative deficiency of insulin. The former occurs in autoimmune forms of diabetes, e.g., type 1 diabetes mellitus, or latent autoimmune diabetes in adults, in which progressive destruction of insulin-secreting β cells leads to an absolute deficiency of insulin. Relative insulin deficiency is far more pervasive and in its most common form, T2DM, is caused by insulin resistance (most often due to obesity) coupled with progressive failure of the β cell to secrete sufficient insulin to compensate for the increased insulin resistance (3). In addition to typical T2DM, we now know of several monogenic diabetes syndromes that lead to a relative deficiency of insulin (reviewed in ref. 2). These include autosomal dominant T2DM (also known as maturity onset diabetes of the young), autosomal recessive syndromes of extreme insulin resistance, maternally inherited diabetes and deafness (MIDD), and partial and complete lipoatrophic diabetes syndromes. Although relatively rare, these syndromes have provided important insights into the molecular and cellular basis of glucose homeostasis. For example, some forms of autosomal dominant T2DM are due to defects in transcription factors necessary for normal β cell growth and differentiation (e.g., hepatocyte nuclear factors [HNFs], β2/neuroD, insulin promoter factor-1 [IPF-1]) (4), while others are due to mutations in molecules involved in glucose-regulated insulin secretion (e.g., glucokinase or the regulatory subunit of the ATP-sensitive potassium channel) (4, 5). MIDD, caused by mutations in mitochondrial DNA, is associated with defective insulin secretion and also some element of insulin resistance (6). Although the specific mechanisms whereby mitochondrial DNA mutations lead to MIDD have not been fully elucidated, this rare syndrome points to mitochondrial function as a key factor in glucose homeostasis that may be relevant to the more common forms of T2DM (also see below).

In his 1987 Lilly Lecture, Defronzo elegantly described the triumvirate of β cell, muscle, and liver as a collusion responsible for T2DM (13). The idea that T2DM results from insulin resistance in muscle (causing decreased glucose uptake) and liver (causing increased gluconeogenesis), combined with declining β cell function is now widely accepted. As mentioned above, adipose tissue is now also recognized as an important player in glucose homeostasis. This conceptual framework has shaped how physicians treat patients with diabetes and has guided the identification of drug targets that reduce insulin resistance in muscle and liver and enhance and/or preserve β cell function.

For several decades, it was proposed that the biochemical basis of hyperglycemia in T2DM could be explained by the Randle cycle, by which, simply stated, increased fatty acid oxidation causes a commensurate decrease in glucose oxidation, leading to decreased glucose uptake and hyperglycemia. Recently, functional genomics and transgenic approaches have identified a key common pathway that may play an important role in the pathophysiology of T2DM. Two studies utilized cDNA microarrays to examine differences in gene expression profiles among muscle of humans with T2DM, impaired glucose tolerance (prediabetes), and normal glucose tolerance (20, 21). Both studies found subtle decreases in expression of a subset of genes involved in mitochondrial oxidative phosphorylation (OXPHOS) in T2DM and prediabetic subjects. The subset of downregulated OXPHOS genes is known to be coordinately regulated by the PPARγ coactivators–1α and –1β (PGC-1α and PGC-1β), which were also subtly decreased in muscle tissue from diabetic and prediabetic subjects. Indeed, overexpression of PGC-1α in a mouse skeletal muscle cell line resulted in upregulation of the same set of genes found to be downregulated in human T2DM muscle (20). Transgenic mice overexpressing PGC-1α in muscle had increased formation of mitochondria-rich oxidative type 1 myofibers and reduced formation of glycolytic type 2 myofibers (22). In mice, PGC-1α has been shown to be a major mediator of cold-induced mitochondrial biogenesis and adaptive thermogenesis in brown fat (23, 24), and in liver, PGC-1α is a pivotal regulator of gluconeogenesis (25). Finally, recent evidence suggests that PGC-1α may also influence β cell energy metabolism and insulin release (26). While the role of PGC-1α in energy homeostasis is reasonably well established, the role of PGC-1β is less clear. Transgenic mice overexpressing PGC-1β have increased systemic fat oxidation and are resistant to diet-induced obesity, which suggests coordinated transcriptional effects on mitochondrial genes involved in fat oxidation and, relative to PGC-1α, little effect on mitochondrial genes involved in glucose metabolism (27).

The results reported by Ling et al. (31) are, for the most part, consistent with and extend the findings of others regarding the importance of PGC-1–regulated mitochondrial pathways in glucose homeostasis and the pathophysiology of T2DM. However, there are several caveats that are worth mention and further consideration. For these physiological studies it was necessary to exclude twins with known diabetes. However, this might have preferentially excluded those individuals with a high genetic burden of T2DM susceptibility genes or individuals exposed to a diabetogenic environment from the older group compared with the younger group, since the older group would have had more time to express the diabetic (exclusionary) phenotype. This potential ascertainment bias might be expected to underestimate any differences in diabetes-related traits between young and old groups. Second, direct comparisons of young and old groups are potentially confounded by a cohort effect. For example, nutritional or other environmental factors that may have differed over the approximately 40-year span between the birth of the young and old cohorts could have important influences on the outcomes measured, particularly by virtue of birth weight and intrauterine effects on adult glucose metabolism (32). Third, although decreased PGC-1α and PGC-1β expression are likely to be causally related to T2DM pathogenesis, the correlative cross-sectional nature of this study does not prove causality. Fourth, the authors propose that association of the Gly482Ser PPARGC1 variant with decreases in both PGC-1α and PGC-1β mRNA levels may be due to decreased function of the Gly482Ser variant upon activation of transcription of its own gene and PGC-1β. It is also possible that the Gly482Ser variant is in linkage disequilibrium with a 5′ or 3′ polymorphism that affects transcription or mRNA stability. Finally, it is argued that monozygotic twins are genetically identical. Although they are clearly more genetically similar to each other than dizygotic twins, it should be kept in mind that even monozygotic twins have potentially important differences in their genetic makeup — most notably for this study, differences in mitochondrial DNA, as well as differences in genomic DNA methylation that may alter expression of individual genes. These variations could affect the expression of PGC-1α and PGC-1β and other OXPHOS genes and pathways, creating genetic differences even in monozygotic twins.