Introduction

Overall, spinocerebellar ataxias (SCAs) are considered to be rare disorders (prevalence ∼3:100 000). Machado–Joseph disease (MJD) (MIM 109150), also known as SCA type 3 (SCA3), is one of the most common SCAs worldwide,1 reaching its highest prevalence value in Azores Islands (Portugal) (∼1:3472).2 MJD is a clinically heterogeneous autosomal dominant neurodegenerative disorder of late onset (occurring at a mean age of 40.2 years), involving the cerebellar, ocular motor, pyramidal, extrapyramidal and peripheral motor systems.3

MJD is caused by the expansion of a CAG repeat at exon 10 of the ATXN3 gene (14q32.1), which encodes for ataxin-3.4, 5 Wild-type alleles present 12–44 CAG units, whereas expanded alleles contain between 53 and 87 CAG repeats.6, 7 Because in genomic DNA (gDNA) the (CAG)n number in the expanded alleles is only partially correlated with the age at onset, accounting for 50–80% of its variability,8, 9 precise predictions of the onset based on repeat size are unfeasible.

As in other polyglutamine disorders, both germinal8 and somatic10 instability have been described for expanded MJD alleles at the gDNA level. The existence of such molecular instability, which is related to DNA replication and repair processes, raises the hypothesis that changes in the CAG repeat number could also occur during transcription. Previous studies on yeast models have highlighted the possibility that transcription-related mechanisms could account for the generation of longer transcripts.11 However, this hypothesis is still much unexplored for polyglutamine disorders.

The main aim of this work was to determine whether there are discrepancies in the size of the (CAG)n tract at the ATXN3 gene between gDNA and mRNA from MJD patients and controls, and to establish whether there is a better association between age at onset and repeat size at the mRNA level.

Materials and methods

Subjects

After obtaining informed consent, blood samples were collected from 52 clinically confirmed MJD patients of Azorean ancestry, for whom age at onset was available (mean age 37.92 years). Twenty-eight apparently healthy individuals, of the same background and with no familial history for MJD, were used as controls.

gDNA and cDNA samples

DNA and total RNA were extracted from peripheral blood leukocytes, using a salting-out procedure and Trizol reagent (Invitrogen, Ltd, Paisley, UK), respectively. Total RNA (weight 2 μg) was subjected to direct cDNA synthesis using the ThermoScript reverse transcription (RT)-PCR System (Invitrogen) with Oligo(dT)20, at 60°C for 50 min.

Allele size determination

For gDNA, fragments containing the (CAG)n tract of the ATXN3 gene (198 bp+(CAG)n) were amplified using previously described conditions.12 For cDNA, the same methodology was applied, after substituting the forward primer by MJD814F (5′-TATTCAGCTAAGTATGCAAG-3′). The amplification products obtained also contained the (CAG)n tract plus 198 bp. Allele size determination was made according to Bettencourt et al.12

Statistical analysis

Wilcoxon's test was performed using SPSS v.15.013 to detect discrepancies between the sizes of the expanded ATXN3 alleles in gDNA and mRNA. An exact test was conducted in GENEPOP v.1.2 (http://genepop.curtin.edu.au/)14 to test for differences in the expanded allele size distribution between gDNA and mRNA. Correlation analysis was performed to determine the relationship between the CAG repeat number in the expanded alleles and the age at onset, using SPSS v.15.0.13

Results

With regard to wild-type alleles, no discrepancies were observed between the CAG repeat number in gDNA and in mRNA (Table 1). In the expanded alleles (Table 1), however, differences were observed in 48 cases (92.3%), with statistically significant (Wilcoxon's test, Z=−6.239, P<0.000) increases of +1 (alleles 63–78) or +2 CAG repeats (alleles 66–78) in mRNA (inferred from cDNA). Nevertheless, when comparing the expanded allele size distributions (gDNA versus mRNA), no significant differences were found (exact test, P=0.286) (Figure 1), indicating that, globally, the allelic distribution is similar in both types of samples.

Table 1 Summary statistics of ATXN3 allelic profiles in MJD patients and controls
Full size table
Figure 1
figure 1

Distribution of the expanded ATXN3 alleles at DNA and mRNA levels in samples of MJD patients (N=52).

Full size image

Negative correlations between (CAG)n size and age at onset were observed at both gDNA and mRNA levels (r=−0.825 and r=−0.819, respectively), with similar percentages of explanation of onset variance (68% gDNA and 67% mRNA).

Discussion

The main aim of this work was to determine whether there are discrepancies in the size of the (CAG)n tract at the ATXN3 gene between gDNA and mRNA in expanded and wild-type alleles. On analysing cDNA, we consistently detected differences from +1 to +2 repeats between gDNA and mRNA in expanded alleles. The possibility that this could result from a methodological artefact produced during in vitro RT could be raised. However, and to minimize the occurrence of in vitro errors, an enzyme with high thermal stability was used, enabling the increase of RT reaction temperature (in this case to 60°C), which reduces the probability of RNA secondary structure formation and improves priming specificity. On the other hand, the absence of discrepancies for wild-type alleles, and the fact that there was no relationship between increases of +1 or +2 repeats and the size of the expanded allele in gDNA, does not support a systematic error of the technique. Therefore, it is most likely that the size differences between gDNA and mRNA observed for the expanded alleles are in fact produced in vivo during transcription. It is known that triplet repeat sequences are able to form alternative structures, such as slipped strand structures.15 The presence of those structures along the gDNA template may cause transient pausing and backward slippage of the RNA polymerase complex. As it was proposed previously, from the study of yeast models with long CAG or CTG tracts,11 this could result in the resynthesis of the same RNA sequence, leading to the formation of longer transcripts. The present results support the hypothesis that transcriptional slippage occurs at the MJD locus in a size-dependent way, as it was observed only in expanded ATXN3 alleles.

To our knowledge, comparisons at the gDNA and mRNA levels with regard to the size of the expanded ATXN3 alleles were only made in a previous study,16 which analysed neuronal and non-neuronal post-mortem tissues of a single MJD patient. The referred study additionally included the analysis of the corresponding loci in one patient with dentatorubral-pallidoluysian atrophy (DRPLA) and another with spinal and bulbar muscular atrophy (SBMA). Their results indicated that the tissue-specific size variation of the expanded CAG repeats present in the gDNA was directly transcribed into the mRNAs of the three analysed loci (MJD, DRPLA and SBMA). In the present study, the number of MJD patients was largely increased; results showed that the CAG repeats in gDNA were not always directly transcribed into the mRNAs. The slight discrepancies observed, however, were insufficient to cause significant differences in the expanded alleles' distribution, and no improvement in onset variance explanation was obtained at the mRNA level. Other factors (genetic, environmental or both) should be influencing the onset variance and should be further analyzed.