Despite many studies performed in the last twenty years, the mechanism leading to the emergence of FSHD remains poorly understood. The 3C data reported here provide the first experimental evidence that, in this genetic disease, molecular events occur that involve chromosomal segments located at a very large linear distance on the partially deleted chromosome 4q. Specifically, we have observed that in FSHD myoblasts, the subtelomeric 4qA/B marker strongly interacts with the promoter of the FRG1 gene which is located dozens of kbp proximally on the chromosome, depending on the number of remaining D4Z4 repeats. Even more strikingly, we documented a direct interaction of 4qA/B with the promoter of the ANT1 gene which lies at a linear distance greater than 5 Mbp on the centromeric side. This interaction is FSHD-specific as, in control myoblast cells, the 4qA/B marker did not interact with the FRG1, or the ANT1 promoters.
4qA/B is a 10 kb-long polymorphic segment directly adjacent to the D4Z4 repeat array. It exists in two allelic forms, 4qA and 4qB, which are 92% identical and equally common in the general population. FSHD, however, has been reported to occur only in individuals with the 4qA allele 
. The main difference between the two alleles resides in a tract of β-satellite repeats present in 4qA but not 4qB 
. This difference may bear consequences either in the predisposition to deletions occurring within the D4Z4 repeat array or in the pathological consequences thereof.
Another surprising observation was that, in both normal and FSHD cells, the D4Z4 marker interacted only with its related sequence DUX4c among the various segments tested. No interactions were detected with the promoter regions of ANT1
. In accordance, the hypothesis of a transcriptional regulation through a direct contact of the D4Z4 array with the promoters of these three genes 
appears unlikely. DUX4 and DUX4c are two genes that have been shown to be transcribed within the D4Z4 repeats 
. Thus, our results suggest that the D4Z4 enhancer, within the D4Z4 repeat array, may directly regulate the transcription of the DUX4
We then found that DUX4c crosslinked with the FRG1 and FRG2 promoter regions in both normal and FSHD myoblasts (). We therefore postulate that DUX4c plays a key role in the three-dimensional organization of the locus. Sequence alignment analysis (data not shown) suggests that DUX4c contains a transcriptional enhancer. Moreover, DUX4 interacts with DUX4c which, in turn, makes contact with FRG1 and FRG2. This may provide a molecular basis for the transcriptional regulation of neighbor genes by DUX4/DUX4c.
We then detected a new enhancer element in the 4qA allele that may regulate the expression of the FRG1
genes specifically in FSHD cells through a direct interaction with the respective gene promoters. Indeed, both ANT1
are activated in FSHD patients 
. It is noteworthy that the (1.5 to 3 fold) up-regulation of these two genes seen in FSHD patients is consistent with the relatively weak effect of the 4qA enhancer in the luciferase assay.
Recently, we have reported that in FSHD myoblasts, the nuclear matrix attachment site FR-MAR was specifically delocalized from the nuclear matrix 
. In normal cells, this S/MAR may constrain the flexibility of the region by anchoring it to the nuclear matrix, thus restricting interactions of adjacent sequences in the three dimensional nuclear space. This could particularly affect the 4qA/B marker which is separated from neighbor genes by the S/MAR. In FSHD cells, the delocalization of FR-MAR would thus result in an increased flexibility of the corresponding chromosomal segment and additional possibilities of interaction for the 4qA/B marker. This may provide an explanation for the FSHD-specific, direct interaction of 4qA/B with the ANT1
gene promoters we observed in FSHD myoblasts. In the present 3C experiments, no interactions were detected that involved FR-MAR. This should not be surprising since previous 3C studies have already stressed that S/MARs appear to interact only with other SMARs 
The experimental approach used here provides new ways to systematically explore the higher-order chromatin structure of any chromosomal region. In this study, we have found that the binding of DUX4c to the FRG1
gene promoters appears to play a key role in structuring the 4q35 region in normal cells. Other interactions take place in FSHD cells and this is the likely result of a global reorganization of the locus in relation with the contraction of the number of D4Z4 tandem repeats. This reorganization is schematized in the three dimensional model shown in . In FSHD cells (), the deletion of D4Z4 repeats and the delocalization of the proximal S/MAR would result in the formation of a giant loop where the subtelomeric 4qA/B sequence is now brought in close proximity not only to DUX4C
but also to the proximal ANT1
gene promoter which lies 5 Mbp away on the centromeric side of the region. This major structural rearrangement, as compared to the normal situation (), would make gene promoters accessible to the DUX4c and 4qA enhancers specifically in FSHD myoblasts. One hypothesis to explain how such long-range changes in higher order chromatin structure can occur relates to differences in the methylation status of the corresponding regions 
. Further studies are clearly needed to explore this and other hypotheses.
A–B. Models schematizing the proximity between 4q subtelomeric fragments in control (A) and FSHD (B) nuclei.