Does DNA methylation in the promoter region of the ATXN3 gene modify age at onset in MJD (SCA3) patients?

To the Editor : Machado–Joseph disease (MJD/SCA3) is a neurodegenerative disease caused by an unstable CAG repeat expansion in ATXN3 gene, leading to an expanded polyglutamine tract in ataxin-3, the corresponding protein (1). The distribution of age at onset (AO) is inversely correlated with CAG expansion size; however, repeat length is responsible only for 45–60% of AO variation in MJD (2, 3), indicating other still unidentified factors (genetic, environmental or others). The degree of neurodegeneration induced by the polyQ protein is correlated with protein storage levels (4). We therefore hypothesized that DNA methylation, specifically targeting the mutant allele and leading to transcriptional deregulation, might influence the levels of mutant ataxin-3 in affected cells and thus AO in MJD patients. Our aim was to assess the methylation degree at six CpG dinucleotides at the ATXN3 promoter and explore their role as potential modifiers for AO in MJD. One hundred and twenty-three Brazilian patients from 62 families with MJD were ascertained in Rio Grande do Sul, Brazil. The inclusion criterion was molecular confirmation in a symptomatic patient. AO was defined as the age at which the patient, or a close person, noticed the first symptoms (usually gait unbalance); 35 healthy individuals were also neurologically examined to be used as controls. This study was approved by the Hospital Ethics Committee. DNA was isolated from leukocytes, as described (5). Evaluation of the (CAG)n tract was performed by fluorescently labeled polymerase chain reaction (PCR) and capillary electrophoresis. Methylationspecific multiplex ligation-dependent probe amplification (MS-MLPA) analysis was carried out according to manufacturer’s instructions (MRC Holland, Amsterdam) (6). We designed four specific probes for the ATXN3 promoter, which were able to detect the methylation degree at six CpG dinucleotides, through restriction enzyme recognition. We have also designed three control probes, lacking a restriction enzyme recognition site, in genes located outside this region. Data analysis was performed by exporting peak areas to an Excel-based analysis program. Methylation differences between patients and controls, for each probe, were evaluated using the Student’s unpaired t-test. Generalized linear models (GLMs) by generalized estimating equations (GEEs) were used to test the effect of the methylation degree in AO variation. Controlled variables were the family and (CAG)n length (normal and expanded). Statistical analyses were performed using spss for Windows v.16. Length distribution of (CAG)n with AO is shown in Table 1. A significant inverse correlation was found between AO and repeat length (r2 = 0.57, p < 0.001). Probe 1 (containing one CpG) was predominantly methylated (average 94.8%). Probe 2, also containing one CpG dinucleotide, had an average methylation degree of 40.1%. Probes 3 and 4, both containing two CpGs, were predominantly non-methylated (averages of 6.7% and 3.6%) (Fig. 1). No statistically significant differences were found when comparing methylation status between the whole patients’ cohort and controls (p > 0.05). A stepwise analysis of each probe was performed. Given the factor effect under study being the family, a trend toward a direct relation between methylation degree for probe 1 and AO was suggested (p = 0.055), when GLM analyzed methylation status against AO only (Fig. 1a). The regression coefficient relating probe 1 and AO was 24.0; i.e. each 10% decrease in probe 1 methylation status was related to a 2.4-year reduction in AO. The effect of the methylation status was small; however, when compared to the effect of the expanded repeat: when (CAG)n