Diabetes in Friedreich Ataxia


Address correspondence and reprint requests to Miriam Cnop, Laboratory of Experimental Medicine, Université Libre de Bruxelles CP-618, Route de Lennik 808, B-1070 Brussels, Belgium. Email: mcnop@ulb.ac.be


Diabetes is a common metabolic disorder in patients with Friedreich ataxia. In this Supplement article, we review the clinical data on diabetes in Friedreich ataxia, and the experimental data from rodent and in vitro models of the disease. Increased body adiposity and insulin resistance are frequently present in Friedreich ataxia, but pancreatic β cell dysfunction and death are a conditio sine qua non for the loss of glucose tolerance and development of diabetes. The loss of frataxin function in mitochondria accounts for these pathogenic processes in Friedreich ataxia. Mitochondria are essential for the sensing of nutrients by the β cell and for the generation of signals that trigger and amplify insulin secretion, known as stimulus-secretion coupling. Moreover, in the intrinsic pathway of apoptosis, pro-apoptotic signals converge on mitochondria, resulting in mitochondrial Bax translocation, membrane permeabilization, cytochrome c release and caspase cleavage. How and at which level frataxin deficiency impacts on these processes in β cells is only partially understood. A better understanding of the molecular mechanisms mediating β cell demise in Friedreich ataxia will pave the way for new therapeutic approaches.

Abbreviations used

endoplasmic reticulum






reactive oxygen species

Diabetes in patients with Friedreich ataxia

It has been known for more than a century that individuals with Friedreich ataxia have an increased risk of developing diabetes. Reported incidence rates vary between 8 and 32% (Hewer 1968; Hewer and Robinson 1968; Andermann et al. 1976; Shapcott et al. 1976; Harding 1981; Finocchiaro et al. 1988; De Michele et al. 1996; Durr et al. 1996; Filla et al. 1996; Cossee et al. 1999; Delatycki et al. 1999). This variability is partially because of the choice of the diagnostic test. In one study clinically diagnosed diabetes was present in 8% of Friedreich ataxia patients, while an additional 10% was detected by oral glucose tolerance testing (Hewer and Robinson 1968). In a review of a series of fatal Friedreich ataxia cases, the prevalence of clinically diagnosed diabetes was 23% (Hewer 1968). An additional reason for the variable prevalence rates is that the diagnostic criteria for diabetes have substantially changed over time. In a recent study of Friedreich ataxia patients without a pre-existing clinical diagnosis of diabetes, 49% of participants had impaired fasting glucose and/or impaired glucose tolerance, and 12% had diabetes (Cnop et al. 2012b) (Fig. 1a), illustrating that this is a very common metabolic complication of the disease. Indeed, the prevalence of diabetes is several-fold higher in Friedreich ataxia patients than in an age-matched population. Contrary to cardiomyopathy, diabetes has never been reported as the presenting symptom of Friedreich ataxia. When present, hyperglycemia develops at an average time of 15 years after the onset of the neurological symptoms (Harding 1981; De Michele et al. 1996). Diabetes onset is often acute, with the patient requiring insulin at diagnosis (Ashby and Tweedy 1953). In a number of cases, the first presentation of diabetes in Friedreich ataxia was ketoacidosis, which may be fulminant (Hewer 1968; Bird et al. 1978).

Figure 1.

(a–b) Impaired glucose tolerance and decreased pancreatic β cell function in Friedreich ataxia. (a) Glucose tolerance was calculated as the area under the glycemic curve during the oral glucose tolerance test in 41 Friedreich ataxia patients and 53 healthy controls. (b) The disposition index, a measure of pancreatic β cell function adjusted for the individual's insulin sensitivity, was measured by intravenous glucose tolerance test. Data are modified from (Cnop et al. 2012b). §p < 0.05, §§p < 0.01, by unpaired t-test. (c) Establishment of an in vitro model to study the consequences of frataxin deficiency in β cells. Clonal β cells were transfected with a control siRNA (siCT) or an siRNA targeting frataxin (siFx), and knockdown was assessed by western blot for frataxin protein expression. (d) Mitochondrial membrane potential in control (siCT) and frataxin-deficient β cells (siFx) evaluated by JC-1 staining. The green color indicates loss of mitochondrial membrane potential. (e) Glucose-induced ATP/ADP ratio after frataxin silencing; data are expressed as% of control siRNA (siCT) at 1.6 mM glucose. (f) Glucose-induced insulin secretion expressed as fold of siCT at low glucose. (g) Decreased pancreatic β cell number in Friedreich ataxia. The presence of pancreatic β cells in islets of Langerhans was assessed by insulin staining of pancreas sections from controls (left) and Friedreich ataxia patients (right). Images are courtesy of Dr Ihsane Marhfour. (h) Impact of frataxin knockdown on β cell viability. The apoptotic index was calculated as (% apoptosis siFx–% apoptosis siCT)/(100%–% apoptosis siCT) × 100% for control (CT) or 24-h treatment with 0.5 mM oleate (OL), 25 μM cyclopiazonic acid (CPA) or 0.1 μg/mL brefeldin (BR). Data are modified from (Cnop et al. 2012b) and are shown as means ± SE of 4–16 independent experiments. *treated versus CT; #siFx versus siCT. One symbol p < 0.05, two p < 0.01, three p < 0.001, by paired t-test.

Diabetes is correlated with the size of the GAA trinucleotide expansion in the FRDA gene in some (Filla et al. 1996; Delatycki et al. 1999) but not all studies (Durr et al. 1996; Montermini et al. 1997; Mateo et al. 2004). Patients with diabetes have a longer duration of Friedreich ataxia and a younger age at onset (Delatycki et al. 1999).

In any given patient, diabetes may result from defects in insulin secretion by the pancreatic β cell, impaired insulin action, or both. In Friedreich ataxia, both insulin deficiency and insulin resistance have been reported. Several case reports described ketoacidosis, the hallmark of absolute insulin deficiency (Bird et al. 1978). This insulin-dependent ketosis-prone diabetes was suggested to be caused by non-autoimmune loss of insulin-producing β cells (Schoenle et al. 1989). In other studies, no perturbation of glucose-stimulated insulin secretion was found in Friedreich ataxia patients (Tolis et al. 1980; Finocchiaro et al. 1988); the patients displayed a preserved oscillatory pattern of insulin secretion (Meyer et al. 2006). Insulin resistance was demonstrated at the whole body and cellular level (Khan et al. 1986; Fantus et al. 1991, 1993; Coppola et al. 2009).

Most studies on the pathogenesis of diabetes in Friedreich ataxia were conducted before the genetic cause of the disease was elucidated (Campuzano et al. 1996) and before important concepts in glucose homeostasis were established (Bergman et al. 1987; Kahn et al. 1993; Kahn 2003). One of these key concepts is that insulin secretion is regulated by insulin sensitivity (Bergman et al. 1987; Kahn et al. 1993), similar to hormone production in other endocrine systems. Thus, pancreatic β cells secrete insulin as much as needed to maintain glucose tolerance normal. This implies that the insulin secretory response must be interpreted with regard to the prevailing insulin sensitivity (Kahn 2003; Cnop et al. 2007). This was accomplished in a recent study in 41 Friedreich ataxia patients. While being young and lean, the Friedreich ataxia patients displayed increased body fat content and were insulin resistant. Importantly, this was not compensated for by increased insulin secretion, indicating pancreatic β cell failure (Fig. 1b). In multiple regression analyses, the loss of glucose tolerance in Friedreich ataxia was shown to be essentially driven by β cell dysfunction (Cnop et al. 2012b). In parallel with these functional findings, a decrease in islet β cell mass was found in post-mortem pancreas tissue from nine patients (Cnop et al. 2012b) (Fig. 1g). These novel findings agree with previous anecdotal reports of decreased islet and β cell numbers in patients with Friedreich ataxia (Thoren 1962; Koeppen 2011). In summary, while insulin resistance is frequently present in Friedreich ataxia, β cell dysfunction and death are pre-requisites for development of diabetes.

Frataxin and mitochondrial metabolism

Frataxin is the gene product of the FRDA gene, which carries an intronic GAA trinucleotide expansion in the disease (Campuzano et al. 1996). It is widely accepted that this leads to reduced frataxin expression. Frataxin is a mitochondrial protein. Its role is still not fully clarified. When expression of frataxin is deficient, iron accumulates in mitochondria (Puccio et al. 2001). This has been observed in yeast as well as in several mammalian cellular systems. Based on these observations, it has been proposed that frataxin serves as a chaperone in mitochondria, assisting in assembly of iron-sulfur clusters (Muhlenhoff et al. 2002). These clusters are essential structural parts of a number of proteins that are critical for mitochondrial function. These include aconitase and several proteins in complex I and II of the respiratory chain. Thus, a deficiency of frataxin would impact the function of these proteins. Indeed, experimental ablation of frataxin as well as studies in patients has revealed that mitochondrial function depends on proper frataxin expression. It seems that cells are increasingly vulnerable to reactive oxygen species (ROS) when frataxin is compromised (Armstrong et al. 2010). Conversely, over-expression of frataxin activates mitochondrial energy conversion and enhances oxidative phosphorylation (Ristow et al. 2000). In summary, loss of frataxin function in mitochondria likely accounts for most of the pathogenic processes in Friedreich ataxia.

Role of mitochondria in β cell function and survival and insulin action

Given the essential role of frataxin for mitochondrial function, it is not surprising that Friedreich ataxia patients are at risk for developing diabetes. In fact, mitochondrial defects both at the level of the pancreatic β cell and in insulin target tissues may underlie such pathology.

Mitochondria play a critical role in β cells by controlling their function and number. Insulin secretion is regulated by a series of reactions termed stimulus-secretion coupling. The fundamental concept is that mitochondria in β cells are supply driven: this implies that their main function is to respond to cellular fuel levels (Maechler and Wollheim 2001; Mulder and Ling 2009; Wiederkehr and Wollheim 2012). In most other cellular systems, mitochondria are demand driven: here, their main role is to generate energy, metabolites, and reducing power for cellular function. Fuel supply in β cells is chiefly determined by glucose. The sugar is transported into the β cell in proportion to its extracellular concentration. When mitochondrial metabolism in β cells is activated by a rise in plasma glucose, ATP, reducing equivalents and other metabolites are produced. These factors, together, orchestrate the triggering and amplification of insulin secretion through plasma membrane depolarization, calcium influx and insulin granule exocytosis. This response to the postprandial rise in glucose serves to distribute the energy supply to the metabolically active tissues throughout the body. Insulin action in the target tissues accommodates this requirement, ensuring proper control of whole body metabolism. In diabetes, this control is lost. Consequently, plasma glucose levels remain elevated. Glucose metabolism in peripheral tissues and liver becomes deregulated. Ultimately, metabolism overall is perturbed, including lipid abnormalities that underlie many of the chronic complications of diabetes.

Not only the function of β cells but also their number determines secretion of insulin. The role of reduced β cell mass in the pathogenesis of type 2 diabetes has been debated for decades (Rahier et al. 2008). There are limitations to how this can be studied in humans. β Cell mass is regulated differently in rodents. This notwithstanding, the current view is that a reduction in β cell mass contributes to insulin deficiency in type 2 diabetes (Butler et al. 2007). The mechanisms for β cell loss in type 2 diabetes are not fully understood. Clearly, there is no autoimmune process in islets but mild inflammatory events are apparent (Igoillo-Esteve et al. 2010). ROS production by mitochondria could be a common denominator for many of the potential pathogenic processes culminating in β cell loss. Endoplasmic reticulum (ER) stress is another cellular response potentially leading to β cell dysfunction and apoptosis (Cnop et al. 2012a).

Mitochondria play a very important role in transmission and amplification of death signals. In eukaryotic cells apoptosis can be triggered through two different death pathways: the extrinsic and intrinsic pathway of apoptosis. The former involves the activation of cell surface ‘death receptors’, while the latter, also called the mitochondrial pathway of apoptosis, is activated in response to DNA damage, cytotoxic insults, ER stress, viral infection, growth factor deprivation, and mitochondrial dysfunction. The activation of the intrinsic pathway culminates in outer mitochondrial membrane permeabilization, cytochrome c release, caspase activation and cell death (Youle and Strasser 2008; Hotchkiss et al. 2009; Lee et al. 2009). The cellular outcome of the mitochondrial pathway of apoptosis depends on the balance and interactions between anti- and pro-apoptotic proteins of the Bcl-2 family (Gross et al. 1998; Kim et al. 2006, 2009; Brunelle and Letai 2009; Letai 2009). The Bcl-2 family members have been classified in three groups: pro-survival (Bcl-2, Bcl-XL, Bcl-W, Bcl-B, Mcl-1, and A1), pro-death (Bax, Bak, and Bok) and BH3-only proteins (DP5, Puma, Bid, Bim, Bad, Bmf, Bik, and Noxa). A hierarchical model has been proposed for their action (Kim et al. 2006, 2009; Brunelle and Letai 2009). A growing body of evidence suggests that mitochondrial apoptosis is central to β cell demise in diabetes. Pro-inflammatory cytokines, viral infections, saturated free fatty acids, and ER stress activate this death pathway in β cells, through stimulus-specific usage of BH3-only proteins (Gurzov et al. 2009, 2010; Barthson et al. 2011; Colli et al. 2011; Gurzov and Eizirik 2011; Thomas and Kay 2011; Cunha et al. 2012).

Whether mitochondrial dysfunction plays a primary role in the development of insulin resistance is the subject of ongoing debate. On one hand, aging and physical inactivity both are major risk factors for insulin resistance and type 2 diabetes as well as major determinants of mitochondrial function and mass (Petersen et al. 2004). On the other hand, mitochondrial deficiencies may arise as a consequence of the pathological metabolic milieu. Increased levels of glucose and lipids, that is, gluco- and lipotoxicity, are known to perturb mitochondrial function. In Friedreich ataxia patients, impaired mitochondrial function and deficient ATP production have been demonstrated in muscle using 31P magnetic resonance spectroscopy (Lodi et al. 1999; Vorgerd et al. 2000). This was proposed to be the result of primary frataxin deficiency, on the one hand, and secondary to muscle atrophy and motor handicap, on the other (Vorgerd et al. 2000).

Animal models to study diabetes in Friedreich ataxia

Modeling frataxin deficiency in vivo has been a challenge, since a complete knockout (KO) is embryonically lethal. To circumvent this problem, a pancreatic β cell-specific frataxin KO was created (Ristow et al. 2003). The frataxin-deficient mice become glucose intolerant and subsequently overtly diabetic at nine months of age. This is caused by reduced insulin secretion because of loss of β cell mass. Interestingly, insulin secretion from isolated islets, before loss of β cells is apparent, remains unchanged. This suggests that the anticipated effects on stimulus-secretion coupling in β cells, owing to perturbation of oxidative phosphorylation, were marginal. This notwithstanding, increased production of ROS is observed in the islets from frataxin-deficient mice. Also, there is an increased rate of apoptosis in islets. Taken together, the results suggest that diabetes in the frataxin KO mice is caused by ROS-induced apoptotic destruction of β cells leading to insufficient insulin secretion. Admittedly, the experiments could not establish a causal relationship between ROS production and apoptosis but the findings agree with observations of increased markers for ROS in Friedreich ataxia patients.

It should be kept in mind that the tissue-specific KO of frataxin shows a rapid degeneration of virtually all tissues tested (Puccio et al. 2001; Ristow et al. 2003; Simon et al. 2004; Thierbach et al. 2005). Animal models with reduced rather than absent frataxin may resemble human disease more closely. Knock in mice carrying a (GAA)230 repeat in the mouse frataxin gene crossed with frataxin+/− mice have generated KIKO (knockin/knockout) mice, which exhibit a 65% reduction in frataxin expression compared to wild type mice (Miranda et al. 2002). This is a moderate reduction compared to most patients, and results in a mild phenotype without massive cell loss (Coppola et al. 2006). This makes it a good model to study biochemical and molecular abnormalities. Microarray analysis of KIKO muscle and liver showed that gene networks promoting lipogenesis and insulin resistance are up-regulated (Coppola et al. 2009). The muscle and liver gene expression signatures pointed to down-regulation of the PPAR-γ/PGC-1α pathway, which plays a key role in mitochondrial biogenesis and function. In parallel, the transcription factor SREBP1, a master regulator of lipid biosynthesis, was up-regulated. These coordinated gene expression changes provide the molecular basis for the development of insulin resistance in Friedreich ataxia (Coppola et al. 2009). Studies of pancreatic β cell function and survival in the KIKO mouse are presently ongoing (Igoillo-Esteve and Cnop, unpublished observations).

In vitro models

Substantial effort has been put in the development of in vitro models for Friedreich ataxia. Initial in vitro studies involved the use of patient fibroblasts and lymphoblasts; these contributed to our understanding of the biological function of frataxin and the consequences of its deficiency on cell metabolism, function and survival (Wong et al. 1999; Jiralerspong et al. 2001; Paupe et al. 2009; Gakh et al. 2010). Recently, induced pluripotent stem cells have been derived from patient fibroblasts to generate neurons and cardiomyocytes, the two cell types primarily affected in the disease (Ku et al. 2010; Liu et al. 2011; Du et al. 2012). Since Friedreich ataxia is characterized by partial but not absolute frataxin deficiency, RNA interference technology has been an invaluable tool to silence frataxin expression in other cell types, including yeast, Drosophila, HeLa cells, neuronal and kidney cell lines, Schwann cells, and clonal and primary rodent and human islet cells (Santos et al. 2001; Stehling et al. 2004; Lu and Cortopassi 2007; Napoli et al. 2007; Kakhlon et al. 2008; Lu et al. 2009; Cnop et al. 2012b). We have recently shown that siRNA-mediated frataxin deficiency in β cells (Fig. 1c) reduces mitochondrial membrane potential, glucose-induced ATP/ADP ratio and insulin secretion (Fig. 1d–f). In addition to the functional defects, frataxin silencing also induced β cell apoptosis, under basal condition and following metabolic or ER stress (Fig. 1h). This increased sensitivity to pro-apoptotic stimuli may explain the loss of pancreatic β cell mass seen in patients with Friedreich ataxia (Fig. 1g). Different from other cell types, in which frataxin deficiency was directly linked to sensitization to oxidative stress (Wong et al. 1999; Jiralerspong et al. 2001; Anderson et al. 2008), no change in exogenous H2O2-induced apoptosis was detected in frataxin-depleted β cells (Cnop et al. 2012b). Because intracellular ROS levels were not measured, a potential role of oxidative stress in β cell demise cannot be ruled out. Taken together, the data from these in vitro models suggest that β cell dysfunction in Friedreich ataxia is the consequence of mitochondrial defects and an increased sensitivity to metabolic and ER stress-induced apoptosis.

Association of diabetes with other neurodegenerative diseases

Several neurodegenerative disorders are associated with an increased frequency of diabetes. It is beyond the scope of this Supplement article to discuss this at length and has been subject of an extensive review (Ristow 2004). For instance, in Huntington disease a diabetes prevalence of 10% has been reported. This is remarkable both given the comparably young age of the afflicted individuals and that they exhibit wasting rather than obesity and other metabolic traits characteristic of the common form of type 2 diabetes. It should be borne in mind that, as in Friedreich ataxia (Cnop et al. 2012b), the apparent leanness of these patients does not preclude them from having increased body adiposity.

The underlying mechanisms for the association of these two diseases seem to differ among the neurodegenerative disorders. While Huntington's disease is also caused by a trinucleotide repeat expansion, the CAG repeats in this disorder encode a polyglutamine stretch that appears to be pathogenic in susceptible cells. It has been studied in pancreatic β cells in R6/2 mice, which ubiquitously express a CAG-containing exon 1 of the human IT15 gene, encoding huntingtin. The β cells in R6/2 mice contain protein aggregates characteristic of populations of cortical neurons in the human disease (Bjorkqvist et al. 2005). R6/2 mice develop insulin-deficient diabetes reminiscent of what was observed in the β cell frataxin KO mice. However, in R6/2 mice, apoptosis is not apparent in β cells while replication is diminished, which leads to a reduction in β cell mass. Moreover, in vitro insulin secretion and exocytosis are perturbed in R6/2 mice but not in β cell frataxin KO mice. It was later shown that the mutant huntingtin protein interferes with β-tubulin, vesicular transport and hence the exocytotic process in clonal β cells (Smith et al. 2009). In pancreatic sections from Huntington's disease patients neither aggregates nor β cell loss could be observed in islets (Bacos et al. 2008).

Why these pathogenic processes are shared by neurons and pancreatic β cells is not known. Although neurons and endocrine cells have a distinct embryonic origin, they share many functional and structural features: exocytosis, peptide production, electric excitability, and a high level of metabolic activity. In addition, both cell types are very long lived (Spalding et al. 2005; Cnop et al. 2010; Perl et al. 2010). The neuronal traits of β cells include transcription factor and gene expression networks (Atouf et al. 1997), and these are required for normal β cell function (Abderrahmani et al. 2004). It may not be so surprising that the pathogenic processes of Friedreich ataxia affect these two cell populations similarly.


Diabetes in Friedreich ataxia patients develops as a result of insufficient insulin secretion by pancreatic β cells, in a context of increased insulin requirements because of insulin resistance (Fig. 2). The clinical studies in humans and data from in vivo and in vitro models point to a role for both β cell dysfunction and death. In both processes mitochondria play a crucial role. Mitochondria are central to stimulus-secretion coupling in β cells, sensing the nutrients and driving both the triggering and amplification steps of insulin secretion. Pro- and anti-apoptotic signals converge on mitochondria where the triggering of the intrinsic pathway of apoptosis takes place (Fig. 2). How and at which level frataxin deficiency affects these processes in β cells is only partially understood. A better understanding of the molecular mechanisms mediating β cell demise in Friedreich ataxia will open the way for new therapeutic approaches.

Figure 2.

Summary of the pathogenic mechanisms of diabetes in Friedreich ataxia. In Friedreich ataxia, mitochondrial dysfunction in liver, fat and muscle, either directly or indirectly, leads to the development of insulin resistance. This is a source of metabolic stress for the pancreatic β cells. The frataxin deficiency in β cells results in mitochondrial dysfunction and increased sensitivity to ER stress (shown in italic). Mitochondria are central to stimulus-secretion coupling in β cells, sensing the nutrients and driving both the triggering and amplification steps of insulin secretion through generation of ATP and other second messengers. Mitochondrial dysfunction in β cells leads to impaired insulin secretion. Pro- and anti-apoptotic signals, possibly generated as a result of ER stress, converge on mitochondria where the triggering of the intrinsic pathway of apoptosis takes place. Increased ROS production also promotes β cell apoptosis. Thus, both pancreatic β cell dysfunction and death contribute to the insulin deficiency that is central to the development of diabetes in Friedreich ataxia. Illustrations adapted from Servier Medical Art.

Outstanding questions related to

Pathogenic mechanisms contributing to diabetes in Friedreich ataxia

How does frataxin deficiency induce mitochondrial β cell dysfunction? Which are the dysfunctional iron-sulfur complex-containing proteins in the tricarboxylic acid cycle and respiratory chain? ATP formation is impaired; what other metabolic coupling signals are defective to attenuate the triggering and amplifying pathways in stimulus-secretion coupling? Are mitochondria lost later in disease development?

How does frataxin deficiency promote β cell death? Is ROS production increased, and do ROS play a role in β cell dysfunction and apoptosis? Is the intrinsic pathway of apoptosis activated by frataxin deficiency, and how? Which are the Bcl-2 family members at work?

Treatment of diabetes in Friedreich ataxia

Because Friedreich ataxia patients are both insulin deficient and insulin resistant, therapeutic interventions could be used that act at these two levels. Lifestyle modifications (to decrease adiposity and increase physical activity) and metformin or thiazolidinedione treatment improve insulin sensitivity. Because metformin and thiazolidinediones inhibit complex I of the mitochondrial respiratory chain, they should probably be used with caution in mitochondrial disease (Owen et al. 2000; Brunmair et al. 2004). In addition, thiazolidinediones are associated with congestive heart failure (Singh et al. 2007), which poses an additional risk in Friedreich ataxia. Thiazolidinediones also cause expansion of the adipose tissue, which would be undesirable in Friedreich ataxia patients. Sulfonylurea, glucagon-like peptide 1 analogs and dipeptidyl peptidase IV inhibitors stimulate insulin secretion. Exogenous insulin administration – the oldest and most effective treatment for diabetes – is often required in Friedreich ataxia patients. There are no controlled studies comparing these different diabetes therapies in Friedreich ataxia. The potential β cell and neuroprotective properties of glucagon-like peptide 1 (Holst et al. 2011; Cnop et al. 2012b) beg the question whether incretin analogs should be considered in Friedreich ataxia.


We thank Françoise Féry for thoughtful comments on the text, and Massimo Pandolfo, Dcio L Eizirik and members of the Laboratory of Experimental Medicine for stimulating discussions. Our work reviewed herein was supported by the European Union (Collaborative Projects CEED3 and BetaBat in Framework Program 7), a European Foundation for the Study of Diabetes EFSD/Lilly grant, and the National Ataxia Foundation USA. MC, HM, and MIE wrote the manuscript; all authors approved the final version.

Conflicts of interest

The authors have no conflict of interest.