Patients F49-396 (the proband) and F49-398 were sibs born to non-consanguineous parents (Fig. 1), who presented with abnormal gait and speech in their second decade. The disease started in patient F49-396, at age 24, with gait ataxia, muscle cramps and spasticity, followed by dysarthria, by age 26 years. One year later, he could walk only with support, and became wheel-chair bound by age 29. He also complained of urinary urgency and mild dysphagia. On clinical examination (age 33 years), he had severe spasticity and hyperreflexia in upper and lower limbs, and bilateral Babinski sign. He also had mild muscle wasting in proximal upper limbs and moderate wasting in his lower limbs. His power was grade five in the upper and lower limbs. Also significant were mild upper-limbs ataxia and slow saccades. He could not perform heel-to-shin, due to the severe spasticity. Disease evolved until the age of death, at 38 years old.

In his sib, F49-398, onset was at age 20, with gait ataxia. One year later, he still walked independently, but had lost the ability to run, developed dysarthria and started having muscle cramps. Examined at age 24 years, this patient had hyperreflexia without spasticity in upper limbs, and severe spasticity and hyperreflexia in lower limbs and bilateral Babinski sign. His lower-limbs’ muscles were mildly wasted, but had a grade five power in upper and lower limbs. He showed moderate upper-limbs and mild lower-limbs ataxia. He had slow saccades, but apparently no limitation of eye movements. The time from the first symptoms to death was 7 years. Neither of the patients had bulging eyes, nor sensory or extrapyramidal involvement, including dystonic posturing.

An expanded ATXN3_(CAG)n allele was observed in each patient: F49-396 (29/74) and F49-398 (29/75) (Fig. 1). By segregation analysis, we were able to identify the allelic phase of the polymorphic markers with expanded alleles and construct haplotypes. We observed a “pure” Machado lineage associated with the MJD allele, covering a 15 kb region (Table 1, Fig. 1).

Previously, we studied SNP-based MJD haplotypes in more than 20 populations (Martins et al. 2007, 2012; Martins and Sequeiros 2018; Sharony et al. 2019); for this study, we analysed our current cohort of 393 MJD families, from 33 populations and identified the Machado lineage in 90 MJD families worldwide: 86 from Portugal, 2 from Spain and 2 from the USA (with no Portuguese ancestry). In addition, 3 Portuguese-Azorean families shared 29 of those 30 SNPs (differing only by allele C_rs12895357), but their flanking STR haplotype (H7) was common in other Machado families (which led us to hypothesize a recent recurrent back mutation G > C at rs12895357 for that sublineage). Therefore, a single origin should be shared by all 93 families and this Sudanese family, a total of 94 families for further analysis with the 7 flanking STRs.

We were, thus, able to reconstruct MJD-associated haplotypes in the Sudanese patients (H11:16-19-9-(CAG)exp-13-7-19-26), as well as in 79 other families of the Machado lineage (Table 2). The two Spanish families, although not reaching the threshold of 0.6 for the probability inferred (0.551 and 0.541) by the PHASE software (Stephens et al. 2001), shared the same 10-20-10-(CAG)exp-18–7-15–24 haplotype, the reason why we included them in this analysis. The 12 remaining families were excluded, as their disease-associated STR-haplotype could not be reliably inferred.

We observed a low gene diversity (0.204, SD = 0.139) in the Machado lineage, with only 12 STR-haplotypes being found among 82 families (Fig. 2). A founder haplotype (H1:10–20-10-(CAG)exp-18-7-19-24) was shared by 47 families, all of Portuguese origin. The two Spanish families showed a haplotype phylogenetically close to H1, probably resulted from a recombination in the founder haplotype. A recombination event (instead of stepwise mutations in STRs) is also the most parsimonious option to explain the origin of H4 and H5 from H1, since their downstream haplotype (4 STRs) is shared by all of them, with phylogenetically distant haplotypes observed only upstream.

Currently, in control populations, alleles 10_TAT223, 10_ATA194, 7_AAAC123, 19_GT190 and 24_AC190 are among the most frequent in all Europeans, Asians and Africans; it is noticeable, however, the very low frequency of allele 20_GT199 (5.7% in Europe (n = 405), 3% in Asia (n = 122), and 0% in Africa (n = 96)) and allele 18_AC21 (0.7% in Europe (n = 420); 3% in Asia (n = 119), and 0% in Africa (n = 97)).

We have currently analysed a total of 393 MJD families (from 33 populations; including unpublished data), and could draw a broad picture of MJD lineages (and sublineages) worldwide. This Sudanese family with MJD shares the ancestral (“pure”) Machado lineage, present in 24% of all families studied. Thus, we estimated the time ensued from their likely common ancestor. As the most ancient STR haplotypic background, we assumed H1 to be the founder core haplotype of the network (Fig. 2). From there, we estimated the number of stepwise STR mutations and recombinations that would be implied to originate the remaining STR haplotypes (Table 2).

To capture an accurate picture on the effect of recombination on age estimation, we calculated the recombination fraction for these STRs, based on (1) physical distance between the two farthest STRs typed (θ = 0.0058 cM); and (2) meiotic events (2 recombinations observed among 268 meioses; θ = 0.0075 cM). Thus, assuming a generation time of 25 years, and based on the most accurate recombination estimates (from family data), we estimated the Machado lineage to be 3,211 ± 693 years.