Neurodegenerative diseases affect specific subsets of neurons and in some cases non-nervous system structures as well, but the reasons underlying such selective cellular vulnerability are largely unknown. Reduced frataxin levels in FRDA are likely to have some downstream consequences that are the same in all cell types, along with consequences and adaptive responses that are cell and tissue-specific. We took an unbiased, genome-wide approach to obtain clues on FRDA pathogenesis and selective vulnerability. We analyzed how global gene expression profiles are affected in the heart (affected in FRDA), skeletal muscle and liver (not clinically affected) from a mouse model expressing frataxin at levels (about 30–35% of normal) corresponding to mildly affected FRDA patients, and yet not showing obvious clinical or pathological abnormalities. We observed the same basic pattern of dysregulation in skeletal muscle and liver, consistent with their roles in energy metabolism. However heart muscle, which is significantly affected by myopathy in the human disease, showed changes suggesting a fiber-type switch and dysregulation of contractile proteins, possibly consistent with cardiomyopathy.
Dysregulation of the PPARγ/PGC1A pathway was observed in tissues from animal and cellular models of frataxin deficiency, as well as in cells from FRDA patients, suggesting that this is a general downstream effector of frataxin deficiency. Pgc1a is the most studied of the peroxisome proliferator activated receptor co-activators (PGCs), a family of transcriptional co-activators that regulate mitochondrial biogenesis, energy substrate and utilization, and oxidative metabolism (23
). In KIKO skeletal muscle and liver, our results indicate that Pgc1a
activity is downregulated, as shown by the increased expression in both tissues of a set of genes involved in lipogenesis that are normally repressed by Pgc1a, including Acas2
. These changes are also supported by the simultaneous upregulation of the transcription factor Srebp1
. In skeletal muscle, changes in expression of contractile proteins that would result in increased glycolytic fast-twitch fibers and decreased slow, oxidative fibers type I and IIa, provide further evidence of downregulated Pgc1a activity (37
). Pgc1a downregulation, and the related Srebp1 upregulation, are known to occur in insulin resistance and diabetes (38
), along with reduced mitochondrial oxidative phosphorylation and lower oxidative-to-glycolytic muscle fiber ratio (40
). The increased risk of diabetes in FRDA patients is therefore likely to be a consequence of the downregulation of Pgc1a
in key tissues for insulin response and fuel metabolism control. FRDA patients are insulin resistant before being diabetic (9
), and higher incidence of glucose intolerance and insulin resistance has been reported in family members of FRDA patients (42
). However, other reports suggest a primary beta-cell involvement in FRDA (44
), and a pathogenic mechanism primarily involving beta-cell failure has been proposed based on studies in other mitochondrial disorders (45
) and on the pancreatic conditional frataxin knock-out (6
). Here, we confirm that insulin resistance is present in all tested non-diabetic FRDA patients, strongly supporting the hypothesis suggested by our gene expression data that insulin target tissues are relevant in the pathogenesis of diabetes in FRDA. Furthermore, we show here for the first time that the degree of insulin resistance of FRDA patients correlates with GAA repeat size (hence to lower residual frataxin levels) and is independent of neurological impairment, in agreement with the hypothesis that it is a direct consequence of frataxin deficiency.
Though the direct mechanistic link between frataxin and PGC1A remains to be completely defined, it is very likely related to mitochondrial dysfunction caused by frataxin deficiency. Mitochondrial dysfunction also occurs in T2D (47
), as shown by 30% reduction of ATP production in skeletal muscle, associated with increased intramyocellular lipid (IML) content and decreased aerobic/anaerobic muscle fiber ratio (40
). IML content, a consequence of increased lipid synthesis, is strongly correlated with glucose intolerance and insulin resistance in diabetic patients and can trigger insulin resistance (26
). No data are available on IMLs in FRDA muscle, but the finding of increased cytosolic malic enzyme activity (18
), in perfect concordance with gene expression data from KIKO mice, at least suggests increased lipogenesis and a metabolic shift similar to T2D. We propose therefore that in FRDA, like in T2D, a state of insulin resistance and associated metabolic changes long precedes the onset of clinically overt diabetes, which is eventually due to pancreatic beta-cell failure. Beta-cell failure may in turn be accelerated by intrinsic mitochondrial dysfunction due to the gene defect (Fig. ).
Figure 4. Diabetes in FRDA and type-2 diabetes may have a final common pathway. In normal conditions (left panel), lipid breakdown and biosynthesis are tightly controlled by the two master regulators Pgc-1a and Srebp-1. In type-2 diabetes (center), increased dietary (more ...)
In contrast with skeletal muscle, we observed a downregulation of fast fibers in cardiac muscle, which also suggests an increased Pgc1a effect. We also observed a trend towards Pgc1a
upregulation in HL-1 cardiomyocytes treated with shRNAiFxn
. Though apparently contradictory, these data actually confirm Pgc1a as a primary target of frataxin deficiency while underlining tissue-specific differences in this response. Our findings are indeed consistent with the observation that Pgc1a
mRNA levels are elevated in the myocardium in a mouse model of metabolic syndrome (49
) and with data suggesting a role for increased Pgc1a levels in diabetic cardiomyopathy (50
). Furthermore, chronic overexpression of Pgc1a
in hearts of transgenic mice induces cardiomyopathy (51
), while only mild cardiac abnormalities are observed in the hearts of Pgc1a
knock-out mice (53
). The idea that Pgc1a can have opposite changes in different tissues is supported by the fact that its overexpression induces uncoupled respiration in adipose tissue and coupling in cardiac myocytes (51
). Pgc1a binding to tissue-specific transcription factors, enabling the activation of diverse metabolic programs in different tissues (23
), offers an explanation for these differences and the distinct metabolic effects observed in skeletal and cardiac muscle in this model of frataxin deficiency (Supplementary Material, Fig. S4
This model was supported by recent reports that (i) PGC1A overexpression in skeletal muscle cells (56
) and (ii) treatment of FRDA cells with the PPARγ agonist Azelaoyl-PAF (36
) are able to increase frataxin levels. PPARγ agonists such as rosiglitazone and pioglitazone are in clinical use as oral antidiabetics. The latter molecule may be of particular interest because of its capacity to cross the blood–brain barrier. Clearly, the opposite dysregulation of Pgc1a in the heart imposes caution for any attempt to clinically test such a drug in FRDA. Maybe not surprisingly, cardiac toxicity is a known side effect of this class of drugs.
In conclusion, we have identified distinct pattern of gene expression changes in metabolically distinct tissues in a model of frataxin deficiency. These data suggest Pgc1a as a key regulator of gene expression that is directly affected by frataxin deficiency and in turn affects frataxin expression. Further studies (such as frataxin overexpression) are needed to characterize the relationship between Fxn and Pgc1a. Downregulation of Pgc1a is the rule in several frataxin-deficient cell types, including skeletal muscle, liver, lymphoblasts, fibroblasts and possibly the pathologically relevant CNS cells, engaging a feedback loop that leads to even lower frataxin levels, reduced antioxidant defenses and mitochondrial function. Such a mechanism appears to be responsible in particular for the vulnerability to diabetes of FRDA patients. In these cell types, upregulation of Pgc1a using PPARγ agonists may be an appealing therapeutic approach.