• linkage disequilibrium
• SCA2
• trinucleotide repeat expansion
• founder effect

Autosomal dominant cerebellar ataxias (ADCA) are clinically and genetically heterogeneous neurodegenerative disorders. Patients typically present with progressive cerebellar ataxia and additional neurological signs.

Molecular genetic studies have permitted identification of seven distinct ADCA loci1-3: spinocerebellar ataxia 1 (SCA1) on chromosome 6p22-p23, SCA2 on 12q24.1, SCA3 or Machado-Joseph disease (SCA3/MJD) on 14q32.1, SCA4 on chromosome 16q22.1, SCA5 on chromosome 11q, SCA6 on chromosome 19p13, and SCA7 on 3p12-p21.1. Five of the seven genes have been identified, and all contain a trinucleotide (CAG)_n repeat expansion in the coding region. The same type of mutation is also responsible for Huntington’s disease, spinobulbar muscular atrophy, and dentatorubral-pallidoluysian atrophy (DRPLA).

The SCA2 mutation was found in 33 out of 220 families of different geographical origins with ADCA type I (15%), 32 of which were previously reported.4 Twenty six unrelated SCA2 families with at least two affected members were selected. Eight were from France, four from Germany, three from Austria, three from other European countries, five from North Africa, and three from the French West Indies (table 1).

Genotypes were determined for three micosatellite markers that span 550 kilobases (Kb) on chromosome 12p24.1 as follows: cen-D12S1332-350 Kb-D12S1672-200 Kb-D12S1333-tel.5D12S1332 and D12S1333 flank the SCA2 gene. D12S1672 is in the first intron, close to the CAG repeat.6 Microsatellite polymorphisms were analysed by PCR with pairs of primers, one of which was 5′ end labelled with either FAM, TET, or HEX (PE Applied Biosystems). PCR was carried out as previously described.7One μl of each PCR product was then mixed with 3 μl formamide, 0.5 μl loading buffer (dextran blue 50 mg/ml, containing 25 mmol/l of EDTA), and 0.5 μl of Genescan 500-TAMRA (Perkin Elmer), denatured at 94°C for two minutes, then cooled on ice and electrophoresed through a 5% acrylamide gel containing 6 mol/l of urea in 1 × TBE buffer for two hours. Product sizes were determined using GeneScan™ software version 3.0, and alleles were numbered according to the order of product sizes.

Families were separated into two groups for analyses of linkage disequilibrium (LD): northern Europe (Germany, Austria, France, Belgium, and Serbia, n=17) and the whole population (northern Europe, Morocco, Algeria, Tunisia, and Portugal, n=23). Owing to the lack of appropriate controls, the three families from the French West Indies were not used for LD analysis. For calculation of LD, allele frequencies at each locus were determined, for the first group, in spouses and unrelated controls from northern Europe (n=77). For the second group, populations corresponding to the 23 families tested were used (n=104). The difference in allele distribution between normal and carrier chromosomes was evaluated by χ² and two tailed Fisher’s exact tests. Yates’s correction was used when appropriate. The most frequent allele on disease bearing chromosomes was defined as a single allele, and the others were pooled to form a second allele. The presence of LD was tested byD = χ²/N. A difference of p<0.01 was considered to be statistically significant. The proportion of carrier chromosomes bearing the original associated allele was calculated with the equation δ=(P _D-P _N)/(1-P _N), where P _Dand P _N are the frequencies of carrier and normal chromosomes, respectively.8 Only the ancestral haplotype from each family was used, if it could be determined.

As shown in table 1, 15 disease linked haplotypes (D12S1332-D12S1672-D12S1333) were found. The three French West Indian families had the same haplotype for the two markers that flank the mutation (D12S1672-11, D12S1333-5), suggesting a common origin. In a previous study,9 using more distant flanking markers, only two French West Indian kindreds of the three shared the same haplotype. The four German and the Serbian families shared a common haplotype for the three markers (5-5-8). Another haplotype (6-7-2) was shared by the three Austrian families suggesting the existence of a common but different founder in both countries. In France, four different haplotypes were found in eight families, two of which were the 5-5-8 haplotype found in the German and Serbian families, and the 6-7-2 haplotypes shared with the Austrians. This might reflect common founders in countries that are geographically close.

No common haplotypes were found in North Africans families. The haplotypes in the Algerian, Moroccan, Belgian, Tunisian, and Portuguese kindreds probably resulted from different ancestral mutations. The D12S1672 alleles, segregating with the disease, were different in all but two of these kindreds. Since D12S1672 is within 20 kb of the SCA2 mutation,10 recombination events separating both loci are very unlikely, as well as mutations in D12S1672 that would generate at least five different alleles. This suggests an independent origin for the SCA2 mutation among these families. However, a very ancient founder effect cannot be excluded. The existence of multiple ancestral mutations is supported by haplotype analysis in smaller series of SCA2 families.10 11

We observed significant LD between the SCA2 mutation and the three microsatellite markers. Alleles 5 of D12S1332 and D12S1672 and allele 8 of D12S1333 were significantly more frequent in patients than in controls in the northern European group (p<0.0017), as well as in the group as a whole (p<0.0017) (table 2). In both groups, the most significant LD was obtained for D12S1672-5 and D12S1333-8, as shown by the D values (table 2). This could be explained by the fact that the SCA2 CAG repeat is located between D12S1672 and D12S1333. The δ value for D12S1672 reached 0.48 in the northern European population, whereas the largest δ value in the whole population was for D12S1333 (δ=0.48). The decrease in the δ value for D12S1672 in the whole population (0.34), compared with northern Europeans, suggests that recombination events occurred independently in northern Europeans, and in other populations, especially in North Africans.

Haplotype 6-7-2 which was found on 31% of the disease bearing chromosomes and 9% of controls was not in significant LD with the SCA2 mutation (table 2). Haplotype 5-5-8, absent in the controls (n=50), represented 35% of disease bearing chromosomes (p<0.001), confirming that these French, German, and Serbian alleles derived from a common founder (table 2).

Independent ancestral mutations, with founder effects in limited regions, have also been reported for other ADCA loci. At the SCA1 locus, haplotypes differed according to ethnic origin,12and LD with linked markers was reported in Japanese and Italian families.13 14 For SCA3/MJD, a founder effect or recurrent mutations were found in the French15 and in the Japanese populations.16 17 Probably, several founders accounted for the SCA6 mutations.18 Finally, LD exists between closely linked markers and the SCA1, SCA3, and SCA6 mutations in several countries, but several distinct ancestral mutations were found in different ethnic groups. SCA2 linked haplotypes might represent at risk chromosomes for CAG expansion, as suggested for SCA3,15 but the de novo cases, which are needed to confirm this hypothesis, have not yet been described.