Article Text

Download PDFPDF

DNA methylation in intron 1 of the frataxin gene is related to GAA repeat length and age of onset in Friedreich ataxia patients
  1. I Castaldo1,
  2. M Pinelli1,
  3. A Monticelli1,2,
  4. F Acquaviva1,
  5. M Giacchetti1,
  6. A Filla3,
  7. S Sacchetti4,
  8. S Keller1,4,
  9. V E Avvedimento1,
  10. L Chiariotti1,4,
  11. S Cocozza1
  1. 1
    Department of Cellular and Molecular Biology and Pathology, University of Naples “Federico II” Naples, Italy
  2. 2
    IEOS CNR, Naples, Italy
  3. 3
    Department of Neurological Science, University of Naples “Federico II”, Naples, Italy
  4. 4
    Naples Oncogenomic Center NOGEC, CEINGE Biotecnologie Avanzate, Naples, Italy
  1. Dr I Castaldo, Department of Cellular and Molecular Biology and Pathology, University of Naples “Federico II”, via Pansini 5, 80131, Naples, Italy; icastald{at}


Background: The most frequent mutation of Friedreich ataxia (FRDA) is the abnormal expansion of a GAA repeat located within the first intron of FXN gene. It is known that the length of GAA is directly correlated with disease severity. The effect of mutation is a severe reduction of mRNA. Recently, a link among aberrant CpG methylation, chromatin organisation and GAA repeat was proposed.

Methods: In this study, using pyrosequencing technology, we have performed a quantitative analysis of the methylation status of five CpG sites located within the region upstream of GAA repeat, in 67 FRDA patients.

Results: We confirm previous observation about differences in the methylation degree between FRDA individuals and controls. We showed a direct correlation between CpG methylation and triplet expansion size. Significant differences were found for each CpG tested (ANOVA p<0.001). These differences were largest for CpG1 and CpG2: 84.45% and 76.80%, respectively, in FRDA patients compared to 19.65% and 23.34% in the controls. Most importantly, we found a strong inverse correlation between CpG2 methylation degree and age of onset (Spearman’s ρ  =  −0.550, p<0.001).

Conclusion: Because epigenetic changes may cause or contribute to gene silencing, our data may have relevance for the therapeutic approach to FRDA. Since the analysis can be performed in peripheral blood leucocytes (PBL), evaluation of the methylation status of specific CpG sites in FRDA patients could be a convenient biomarker.

View Full Text

Statistics from

Friedreich ataxia (FRDA; MIM 229300) is an autosomal recessive neurodegenerative disease that causes physical disability at an early age.1 The disorder is the most frequent hereditary ataxia in western countries with a prevalence of one in 80 000 in the European population.2 FRDA is absent in Japan and very rare among black Africans. The frequency of heterozygous carriers is about 1%.2 FRDA is clinically characterised by progressive gait and limb ataxia, cerebellar dysarthria, loss of lower limb tendon reflexes, and severely impaired position and vibration sense and extensor plantar responses.2 The onset of disease usually occurs before the age of 25 years and typically around puberty. The median survival from onset is 36 years.3 Non-neurological signs include cardiomyopathy, skeletal deformities (scoliosis and pes cavus) and increased incidence of diabetes mellitus. Cardiac involvement is the most common cause of premature death.1 2 Neuropathology in FRDA typically shows early degeneration of large sensory neurons in the dorsal root ganglia, associated with degeneration of the posterior columns and atrophy of the large sensory fibres in peripheral nerves.1 2 4

In most cases, FRDA is caused by expansion of GAA repeat affecting the first intron of the FXN gene (MIM 606829) on chromosome 9q13 that encodes a highly conserved protein called frataxin. The GAA mutation reduces the expression of frataxin. Normal chromosomes have FXN alleles carrying between 5–30 triplet repeats. The size of the expanded repeat is typically between 600–1200 triplets, but alleles ranging from 44–1700 triplets have been reported.5 A smaller proportion of patients (<5%) are compound heterozygous, having one expansion bearing chromosome and a deleterious point mutation on the other allele.2 6 The phenotypic variability among patients is heavily influenced by the size of the GAA expansion. In particular, the number of GAA repeats in the smaller allele correlates negatively with the age of onset.2

Studies in yeast, mice and patients have provided further insights into FRDA pathogenesis. The lack of frataxin is associated with mitochondrial dysfunction and intramitochondrial iron accumulation.7 8 Levels of mitochondrial iron are elevated in fibroblasts from FRDA patients.9 In cardiac muscle samples from FRDA patients, reduced activities of respiratory chain complexes I, II, and III and of the Krebs cycle enzyme aconitase have been found.10 11 All these enzymes contain iron–sulfur (Fe–S) clusters, and, therefore, frataxin could play a role in mitochondrial iron metabolism and in the genesis of Fe–S centres. Alternatively, these enzymes may be targeted because of their particular sensitivity to damage by oxygen free radicals.11 Evidence from several sources suggests that repeat expansion reduces mRNA frataxin expression by inhibition of transcription. The mechanism by which GAA expansion inhibits FXN transcription is not definitively elucidated. The formation of unusual DNA structures, such as intramolecular triplex model12 or altered chromatin packaging mediated by repetitions, have been proposed.13 Recently, an aberrant CpG methylation has been identified in the first intron of FXN.14 In particular, the authors found hypermethylation of the region in four FRDA patients compared with four healthy subjects.

In this study we examined the DNA methylation status at five CpG sites located in the island region immediately upstream to GAA repeat in a larger cohort of FRDA patients. We have performed a quantitative DNA methylation analysis (pyrosequencing) of genomic DNA from peripheral nucleated blood cells of 67 FRDA patients, all homozygote for GAA expansion. We found a strong correlation between CpG methylation degree of specific CpG sites with both the size of the expansion and the age of onset.



Sixty-seven FRDA patients from the Department of Neurology of the “Federico II” University in Naples were enrolled in the study. The clinical diagnosis of FRDA was determined by examination based on established diagnostic criteria. The age of onset varied from 2–29 years (mean (SE) 12.76 (0.72) years). The age of onset was defined as the date at which the patients or relatives noticed the first appearance of ataxic symptoms. Skeletal deformities were not considered to be the onset symptoms because it is difficult to establish their exact time of presentation. These patients were unrelated and the age range was 5–63 years (mean (SE) 38 (1.75) years). As control subjects 21 healthy volunteers were recruited in the laboratory of Genetica Medica of University “Federico II” Italy. The age of the unaffected controls ranged from 25–30 years (mean (SE) 28.42 (0.61) years). They were also genotyped for the GAA repeat to confirm the absence of any potential carrier heterozygotes. The local ethics committee approved the study and all patients and controls were informed of the aim of the study and gave their consent.

DNA isolation and measure of repeat size

From each individual a 5 ml peripheral blood sample was taken in 50 mM EDTA. Genomic DNAs were purified from peripheral blood leucocytes (PBL) using phenol–chlorophorm protocol.15 GAA repeats expansion in the FXN gene were analysed by polymerase chain reaction (PCR) as previously described.2 In the 67 patients, all homozygous for the GAA expansion, the number of repeats ranged from 117–1280. Most of these patients had two expanded alleles of different sizes. The mean (SE) number of GAA repeats on the smaller and the larger alleles were 708.5 (28.28) and 874.1 (26.86), respectively.

DNA methylation assay

Pyrosequencing technology was used for DNA methylation quantitative analysis. DNA samples (2 μg) were treated with sodium bisulfite with EZ DNA Methylation Kit (Zymo Research, Orange, California, USA), according to manufacturer’s instructions, and then eluted in 50 μl elution buffer and stored at −20°C until use. Treatment of DNA with sodium bisulfite converts the epigenetic difference between methylated and unmethylated cytosine into a single base variation of C/T type (5-methylcytosine remains unaltered, whereas cytosine is selectively deaminated to uracil and amplified as thymidine). Five CpG sites located upstream of GAA repeat were analysed (CpG1, nucleotide position −394 with respect to GAA repeat; CpG2, −382; CpG3, −359; CpG4, −329; CpG5, −326) (fig 1). Amplification was carried out on 10 ng of bisulfite treated DNA using HotStarTaq DNA polymerase (Qiagen, GmbH, Germany) under the following conditions: 15 min at 95°C, followed by 50 cycles of 30 s at 95°C, 40 s at 57.5°C, and 1 min at 72°C. Primers were designed using Pyrosequencing Assay Design Program. Primers used for amplification were (positions are indicated with respect of GAA repeat): FRAT-F3, 5′-GAGGGTTTTGAAGATGTTAAGGA-3′ (from −436 to −414, biotinylated); FRAT-R3, 5′-AACACTAACAACCAATCCCAAAAT-3′ (from −200 to −177). The amplicon size was 260 bp. Sequencing primers were: FRAT-S3, 5′-TTCATCTCCCCTAATACATA-3′ (from −324 to −305, for CpG3, CpG4 and CpG5) and FRAT-S1, 5′-TCCTAACCTTTACCCAA-3′ (from −379 to −363, for CpG1 and CpG2). Pyrosequencing reactions were performed in a PSQ 96MA System (Biotage AB, Uppsala, Sweden) according to the manufacturer’s protocols (representative pyrosequencing is shown in fig 2B). Raw data were analysed using the provided single nucleotide polymorphism (SNP) analysis software.

Figure 1 Schematic view of FXN gene region analysed in this study. The five CpG sites located upstream from the GAA repetition in the first intron of FXN gene are indicated by vertical arrows.
Figure 2 Quantitative DNA methylation analysis. (A) Methylation degree of five CpGs analysed in 67 patients. For any CpG site the mean value of the methylation level is shown both for the control group and for the Friedreich ataxia (FRDA) patients. The methylation levels in FRDA patients range from 76.80–98.08% while controls range from 19.65–66.42%. (B) Representative pyrograms quantifying methylation levels at CpG sites in the first intron of FXN gene. CpG5, CpG4 and CpG3 were detected with sequencing primer FRAT-S3; CpG2 and CpG1 were detected in the same amplicon with sequencing primer FRAT-S1. E and S indicate enzyme and substrate control in the reaction, respectively. Peaks may not correspond to the given percentage values when an A/G position is followed by other adenines in the original sequence (see CpG2 and CpG5). As the reverse strand was read, G peaks indicate methylated cytosine while A indicates unmethylated cytosine. The control, non-CpG cytosine residue indicates the complete conversion of cytosine to uracil by bisulfite treatment. An extra A highlighted in the control derives from an additional internal control chosen by the software and does not affect the quality of the control. The order of nucleotide dispensation in A/G positions is chosen by the software on the basis of the sequence context.

Statistical methods

Methylation differences between patients and controls for each CpG site were evaluated by one way analysis of variance (ANOVA). To test the association among the CpG methylation, the GAA size and the age of onset, a non-parametric correlation (Spearman’s ρ) was used. All statistical analyses were performed by SPSS for Windows (version 15, Chicago, Illinois, USA). The tests were considered significant at values of p<0.05.


Sixty-seven FRDA patients were examined for the GAA trinucleotide repeat expansions and methylation status of a part of the CpG island in the first FXN intron upstream to the GAA repeat. In all patients the long range PCR analysis of intron 1 of the FXN gene showed two large expanded alleles diagnostic of FRDA. The age at onset of the disease were known for 49 affected subjects. The mean (SE) age at onset was 12.76 (0.72) years (range 2–29 years).

By pyrosequencing we investigated the methylation degree of 5 CpG sites (5′-CpG1, CpG2, CpG3, CpG4, CpG5-3′) in the first intron of the FXN gene (fig 1). A representative pyrogram is shown in fig 2B. We found differences in the methylation degree between FRDA individuals and controls (fig 2A). Statistically significant differences were found for each CpG tested (ANOVA p<0.001). These differences were largest for CpG1 and CpG2: 84.45% and 76.80%, respectively, in FRDA patients compared to 19.65% and 23.34% in the controls. We also found a different methylation status among the FRDA individuals (data not shown).

The two FXN expanded alleles are generally called GAA1 (the smaller) and GAA2 (the larger), and previous studies demonstrated that several features of the disease are mainly related to GAA1 size.

To test the hypothesis of a possible relationship between the GAA repeat size and the methylation status of this region, we performed correlation studies. We found a significant correlation between CpG1 and CpG2 methylation levels and the GAA1 size (CpG1: ρ = 0.386, p = 0.001; CpG2: ρ = 0.561, p<0.001). The most significant correlation was found between GAA1 size and CpG2 methylation status (fig 3A). The other CpGs tested did not reveal a statistically significant correlation with the GAA length. These data indicate that alleles from patients with greater GAA1 expansions are more methylated, at specific CpG sites, than those with smaller pathological expansions.

Figure 3 Statistical analysis. Scatter plot (with regression line) between CpG2 methylation status (%) and GAA1 size (number of the GAA repeats) (A) and age of onset (years) in Friedreich ataxia (FRDA) subjects (B). The values of non-parametric correlation analysis (Spearman’s ρ) are reported.

Furthermore, we tested for a possible correlation between the CpG1 and CpG2 methylation with age of onset. A statistically significant inverse correlation was found between both CpG1 and CpG2 and the age of onset of the disease. The best correlation was found for CpG2 as shown in fig 3B (Spearman’s ρ  =  −0.550; p<0.001).

These results suggest that patients with higher methylation degree at specific CpG sites of expanded FXN gene developed the disease earlier.


Hypermethylation associated with silencing of genes is one of the most important and most extensively analysed epigenetic mechanisms in human diseases.16 An alteration of this mechanism has been invoked in the pathogenesis of cancer17 and several inherited diseases.18 In particular, hypermethylation has been found in myotonic dystrophy type 119 and fragile X syndrome.20 In these disorders, the presence of the hypermethylation seems to be responsible for the disease, lowering the amount of the mRNA of the involved gene. In the case of fragile X syndrome, these findings are well established, and diagnosis can be confirmed by the identification of specific aberrant methylation patterns.21

Recently, repeat mediated epigenetic changes have been invoked in the pathogenesis of FRDA.14 22 In particular, Greene and colleagues, studying by bisulfite sequence analysis the methylation of 15 CpG sites in the 5′ region of GAA repeat, demonstrated that three sites were more methylated in four FRDA patients compared with four healthy controls.14 In this work the authors stressed also the role of the specific CpG site 13 as a robust marker of epigenetic modification in FRDA. On the other hand, the same authors stated that methylation of this site per se was unlikely to be the primary cause of frataxin deficiency.

In the present study we investigated five of these 15 CpGs, selected in the middle of the region analysed by Greene and colleagues (CpG sites here numbered from 1 to 5 correspond to 6 to 10 in the Greene’s paper). The use of pyrosequencing allowed us to quantify precisely the methylation levels of these CpG sites in a larger cohort of 67 FRDA patients and 21 healthy controls.

The comparison of the overall methylation pattern revealed statistically significant differences between patients and controls. In the paper by Green et al,14 the CpG site number 6, corresponding to CpG1 in this report, is not methylated in the controls. In our study we analysed the peripheral blood of 21 controls, with an average methylation level of 19.65% at CpG1 residue. This apparent discrepancy could be attributed to the larger cohort analysed in this study or to the different sensitivity of the methods utilised. Furthermore it will be of interest to investigate the methylation state of the other CpG sites, for example, site 13, reported in the precedent study, by quantitative methodologies in the future.

Interestingly, in another study it was reported that epigenetic changes are also present in the brain, cerebellum and heart tissue of FRDA patients.23 Al-Mahdawi et al have identified three specific CpG residues within the FXN promoter and one CpG site within exon 1 that have a different methylation degree by bisulfite sequences. They have also shown, by chromatin immunoprecipitation assay, that there is overall decreased histone H3K9 acetylation together with increased H3K9 methylation of FRDA brain tissue.

How could the expanded GAA tract in FRDA cells influence the methylation status of the region? In the past, several models were proposed to explain the repeat-mediated transcriptional silencing observed in FRDA. In vitro experiments demonstrated that expanded GAA repeats interfere with in vitro transcription in a length and orientation dependent manner.12 More recently, it was supposed that expanded repeats confer variegation of expression by a position effect variegation (PEV)-like mechanism24 with the final effect of gene silencing. By this way the expansion of triplet repeats can promote the heterochromatin formation.24 Expanded FXN alleles were found associated with the epigenetic mark for heterochromatin in the region immediately near the expanded GAA repeats, and it is possible that the GAA expansion is responsible for a particular DNA structure inducing heterochromatin formation.22

We found a direct correlation between CpG methylation and triplet expansion size. Large repeat expansions were found associated with hypermethylation in myotonic dystrophy and fragile X,19 20 but no linear direct correlation has been described yet. Our data support a role for DNA methylation in this process, suggesting that these epigenetic changes occur in a gradual and expansion size dependent manner. Furthermore, we found an inverse correlation between CpG methylation and age at onset of the disease. In our sample, the R2 value obtained between age of onset and CpG2 methylation (R2  =  0.32) were slightly higher than that obtained with GAA1 length (R2  =  0.24). The methylation status of this region seems to account only partially for the age of onset variance. Despite the correlation between CpG2 methylation levels and the age of onset, the dependence of the latter on the former parameter is in the order of 30%. This may be due to several factors such as the use of PBL DNA, uncertainty in the determination of the age of onset, different genetic background and environmental factors.

The inverse correlation between increasing expansion size and severity of the disease is well known in FRDA. Therefore the correlations that we found were not unexpected. In previous reports we demonstrated the existence of a relationship between GAA1 expansion size and age of onset2 and between the size of the expansion and the levels of the frataxin mRNA.25 The results of the present study suggest that the CpG1 and CpG2 methylation status represent the possible link between GAA repeat size and reduced frataxin mRNA levels. The repeat-mediated transcriptional silencing observed in FRDA could be mediated by changes in the methylation status of the FXN gene.

Because epigenetic changes may cause or contribute to gene silencing, our data and those of others may have relevance for the therapeutic approach to FRDA. Recently, the possible use of histone deacetylases (HDAC) inhibitors has been proposed for the treatment of FRDA.26 Our findings could have an impact on the management of FRDA disease; it is likely that monitoring the methylation status may be a useful biomarker for tracking therapeutic benefit following administration of epigenetic drugs.


We would like to thank Professor Carmela Maria Laudando for her help. This work is dedicated to the memory of S Varrone.


View Abstract


  • Funding: This work was supported in part by grants from Friedreich’s Ataxia Research Alliance (FARA) and it was carried out as partial fulfillment of requirements of the PhD degree in “Patologia e Fisiopatologia Molecolare”, University of Naples “Federico II”.

  • Competing interests: None declared.

  • Patient consent: Obtained.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.