DNA repeats can be the source for an extraordinary amount of information about biological processes. As passive markers they provide clues for evolutionary events and processes of mutation and selection, as active agents repeats have remodeled the genome by causing ectopic rearrangements, by creating entirely new genes and by modifying and reshuffling existing genes. Specifically, segmental duplications involve the transfer of 1–200 kb blocks of genomic sequences to one or more locations in the genome and may contain various types of repeats [1]. These transfers have been shown to occur between homologous and non-homologous chromosomes [2, 3]; their preference for pericentromeric and subtelomeric sites can be explained by a damage-control mechanism that facilitates insertion of chromosomal breakage products into poor-gene regions [4, 5]. A particular case of interchromosomal duplication is that observed at the subtelomeric 4q35 and 10q26 loci [6]. The sequence homology between these subtelomeric regions is very high (nearly 99%) and is not restricted to a polymorphic array of 3.3 kb KpnI repeat units, designated D4Z4, but continues distally for at least 6 kb and proximally for nearly 42 kb [7, 8]. The reduction in the number of 3.3 kb D4Z4 repeat to an array of 1–10 units at the 4q35 locus is tightly linked to FSHD, an autosomal-dominant myopathy with an estimated prevalence of 1/20.000 [9]. Similar contractions involving 10q26 D4Z4 repeats are not pathogenic. In normal subjects the single copy probe p13E-11 reveals polymorphic EcoRI fragments in the range 40 to 300 kb including the repeat array, whereas in sporadic and familial cases of FSHD the disease cosegregates with a EcoRI shorter fragment ranging between 10 and 40 kb [10–12]. The identification of a BlnI restriction site within each 10q repeat unit that is absent in the homologous 4q repeat [13] and the use of pulsed field gel electrophoresis (PFGE) allows to distinguish between the p13E-11 fragments of 4q and 10q origin, to study their number and size and to detect sequence exchanges between the two loci [14, 15]. A high number of variant alleles has been detected in human population and is probably consistent with the occurrence of rearrangements between 4q and 10q homologous subtelomeric regions, even if the "de novo" translocation events were never detected. Sequence variations consist in the presence of BlnI-sensitive repeats (10q-type) on chromosome 4 or vice versa of BlnI-resistant repeats (4q-type) on chromosome 10. These exchanges were found in 20–30% of normal subjects and caused either a reduction of the number of BlnI-resistant fragments or an increase of the number of BlnI-resistant fragments in the genome [15], interfering with the correct assignment of DNA fragments to a specific chromosome pair and with an accurate genetic diagnosis of the FSHD [16]. Moreover in healthy individuals a biallelic variation was identified at the 4q region distal to D4Z4, named qA type and qB type. In FSHD patients the contracted D4Z4 fragments implicated in the disease are exclusively located on 4q alleles of qA type, while the 10q alleles always have a qA type telomere [17, 18].

In this paper we performed a detailed analysis of 4q and 10q subtelomeric regions by characterizing the D4Z4 repeat assortment and the distal telomeric markers in Italian healthy subjects and FSHD patients, thus extending previous studies in other populations. We determined: i) the size distribution of polymorphic 4q alleles as compared with 10q alleles; ii) the frequency of variant 4q and 10q alleles; iii) the D4Z4 repeat assortment of 4q and 10q variant alleles; and finally; iv) the qA and qB telomeric sequences both in standard and variant alleles.

Patients and controls Diagnostic criteria for FSHD followed the guidelines proposed by the European Expert Group on FMD [19]. Neurological examinations of 70 unrelated FSHD families for a total of 230 individuals (116 affected and 114 unaffected) were performed at Institute of Neurology of Catholic University of Rome and at the Centre for Neuromuscular diseases, U.I.L.D.M., Rome. For each FSHD family we chose only the proband, both for the study of size distribution of 4q–10q alleles and the types of variant arrays. Healthy subjects were chosen among hospital and laboratory staff as well among the spouses of FSHD families (genetically unrelated to FSHD patients) for a total of 106 individuals. DNAs were obtained from healthy subjects and patients after informed consent in agreement with the guidelines of the Ethical Committee of our Institution.

qA and qB telomeric markers

To perform the allele-typing for qA and qB, the DNA of healthy and affected subjects was digested with HindIII restriction enzyme (Roche Mannheim, Germany) (Fig. 1). The fragments were separated with PFGE, transferred to a Nylon membrane and hybridized with probes qA and qB, supplied by S. M. van der Maarel [17]. The hybridizations were performed overnight at 60°C in a mixture containing 0.125 M Na₂HPO₄ (pH 7.2), 0.25 M NaCl, 1 mM EDTA and 7% SDS. The membranes were washed in 2 × SSC at 42°C and then in 0.1 × SSC and 0.1% SDS at 65°C for 15 minutes.

The high level of sequence homology between the 4q and 10q loci would imply for both polymorphic regions a similar size distribution and recombination products. This is not the case: 10q alleles displayed a different size distribution as compared with 4q alleles with a high percentage of alleles shorter than 40 kb (18%). The differences are more evident when we analysed the structure of variant alleles. 4q variant alleles were heterogeneous and always contained a mixture of BlnI-sensitive (10q-type) and BlnI-resistant (4q-type) repeats, on the contrary 10q variant alleles usually displayed a homogeneous set of BlnI-resistant (4q-type) repeats. The variant alleles revealed that such interactions probably occurred in the past, but the different structures of 4q and 10q variant alleles support the idea that the exchanges between the two loci are rare and these subtelomeric regions appear to evolve independently under different evolutionary pressure. Since reduction in the number of D4Z4 on 4q locus results in FSHD, it is not surprising that 4q chromosomes have incurred a strong selection pressure during human evolution. Gabellini et al. reported that a repressive protein complex binds a specific sequence within each D4Z4 repeat unit on chromosome 4. The reduction in the number of repeats in FSHD patients abolishes this binding and results in the loosening of the transcriptional repression at 4q region and in the overexpression of upstream genes (FRG2, FRG1, ANT1) [28, 29]. Accordingly we observed that 4q type repeats are preserved on 4q chromosome as they are probably essential for maintaining the structural and functional properties of this subtelomeric region. In particular the distal repeats, close to qA and qB sequences, were always maintained on chromosome 4.

Moreover we described qB telomeres on 10q variant alleles carrying homogeneous 4q-type repeat arrays. This suggests that qB sequences are not exclusively associated with 4q chromosomes, but they are always linked to BlnI-resistant repeats.

It was proposed that an ancestral duplication event generated 4q and 4p telomeric regions, then 4q sequences diverged and gave origin to a 4qA type telomere that was duplicated onto chromosome 10. A more recent duplication of telomeric sequences from 4p onto 4q gave origin to 4qB-type telomere [7]. Therefore the presence of qB on chromosome 10 demonstrates that a telomeric sequence exchange has occurred between the 4qB and 10q chromosomes after the duplication events.

We reported a detailed molecular analysis of the 10q and 4q subtelomeric regions in FSHD patients and controls with the aim of characterizing the D4Z4 repeat array assortment and the distal telomeric region (qA and qB) of 4q and 10q alleles. We obtained new interesting findings about the structure of these subtelomeric regions that should be taken into account for a correct molecular diagnosis and genetic counseling of FSHD. In some cases the chromosome origin of the p13E-11 fragments could be misinterpreted: i) we found several 10q variant alleles with a BlnI-resistant array and a size smaller than 40 kb (Fig. 5) mimicking a FSHD-linked alleles; ii) in one patient we observed a 4q FSHD-linked allele carrying BlnI-sensitive repeats upstream the BlnI-resistant repeats, mimicking a 10q allele (Fig. 4).

Furthermore we studied in a different population the 4qter polymorphism distal to D4Z4 described in a previous Dutch study [17] and confirmed the FSHD association only with qA type sequence, but we also detected the presence of borderline alleles in the range of 30–40 kb of qA type, not associated with the disease. This suggests that other concomitant factors, either genetic or environmental, are implicated in the pathogenesis of the FSHD. It appears that the 4qA allele-typing is necessary but not sufficient to assign the pathogenic role to a 4q short allele.