First International “Institute of Myology Workshop” on Facioscapulohumeral Muscular Dystrophy, Paris, May 22, 2007

1. Introduction 

This one-day workshop was organized by Thomas Voit and assembled 13 researchers from five countries (Belgium, The Netherlands, France, Italy, USA) in order to create a collaborative interface between several international groups who are actively investigating the molecular background of facioscapulohumeral muscular dystrophy (FSHD) as well as to give an update on therapeutic projects on FSHD at the preclinical and clinical level.

Such a collective effort seemed particularly timely because there is still no unifying and generally accepted patho-physiological concept that could explain the muscle degeneration in FSHD or the less frequently associated symptoms of cochlear hearing loss, Coats’s syndrome or, in rare infantile cases, mental retardation. Furthermore, opinions diverge as to which genes are possibly directly or indirectly implicated, and no animal model exists that resumes all the particular aspects of FSHD.

2. Genetics and patho-physiology of FSHD 

Luis Garcia presented a synthesis of the main facts and hypotheses related to the molecular biology of FSHD: the genetic defect is a deletion that shortens the array of D4Z4 repeated elements in 4q35. No evidence for transcription of the DUX4 gene which is located within each D4Z4 element [1] has been provided to date, which has shed doubt on its causal implication in FSHD. The D4Z4 region however, carries repressor elements in the normal state, the loss of which leads to upregulation of other genes as had been shown by Rosella Tuppler. He also pointed out that there is evidence from the group of Yegor Vassetzky of an altered chromatin structure linked to a deformation in the DNA structure on the mutated allele: in control cells a nuclear matrix attachment site is found just centromeric of the D4Z4 repeat array but it is lost in the presence of a D4Z4 contraction.

This presentation elicited a debate as to what can be regarded as a pathogenic deletion in the diagnostic sense or why is a shortened allele not necessarily linked to muscle symptoms. Rosella Tupler recalled that in some subjects, the D4Z4 array is missing and they do not show symptoms. This provoked an animated discussion on the “grey zone”, when the number of D4Z4 elements is comprised between 8 an 10 (normal subjects have 11–150 repeats), and that this reduction is specifically linked to the 4qA allele and not to 4qB. Some subjects with 8–10 4qA alleles are not ill, others have atypical symptoms. Such observations had in the past been linked to ideas of a second chromosomal locus which however had never been confirmed. On the other hand, observations from Japan and China indicate that the allele number linked to disease may be variable on a differing ethnic background because these observations apply to the Caucasian population whereas in the Japanese and Chinese population lower allele frequencies of 6–10 D4Z4 elements were not associated to disease [2]. Nicolas Lévy pointed out that the clinical diagnosis of FSHD, notably in less severely affected patients, can be mimicked by other muscle disorders. In the cohort referred to his molecular genetic laboratory for a genetic diagnosis of FSHD he has detected several patients with D4Z4 repeats in this “grey zone” and some of these patients were subsequently found to harbour other mutations in different genes, particularly in calpain 3 and lamin A/C. The question was raised whether a mutation of lamin A/C might negatively interact with the binding of a shortened D4Z4 array to the nuclear membrane.

Rosella Tupler then gave some precisions concerning the calculation of the D4Z4 repetition number. She recalls that the D4Z4 array found in the genomic fragment cut by the EcoRI enzyme is preceded by 6019 base pairs (bp) on its centromeric side and followed by 2400bp on its telomeric side. As a matter of fact, (6019−2400=3619) is about the length of one D4Z4 element (3300bp). So, as a rule of thumb, one must remove the length of one D4Z4 element from the EcoRI fragment size measured on the gel electrophoresis. She also commented that she is following several families where some subjects related to the “grey zone” show symptoms whereas others do not. She therefore raised the following questions: (1) How many patients exist in the “grey zone”? (2) What is the proportion of normal subjects belonging to this zone? A work published in March 2007 by N. Thomas and M. Upadhyaya [3] concerning 146 subjects found no subjects with clinical symptoms having an EcoRI fragment larger than 38kb.

François Leterrier commented that this “grey zone” has given rise to discussions for more than ten years! He therefore proposed to decide for a common protocol on the measurement of the D4Z4 array length, and indeed different methods are used in different laboratories. Such a project should also take into account the recent results of Richard Lemmers (vide infra).

Kevin Flanigan showed the genealogic tree of an extended FSHD family from Utah spanning six generations. A short D4Z4 allele (20kb) remained stable during 12 meiotic events, a finding that strongly argued against anticipation which had previously been discussed for FSHD [4]. He recalls the comparative transcriptome study which compared muscles from FSHD subjects, normal subjects and from patients suffering from other neuromuscular diseases (particularly DMD) [5]. The main conclusion was that genes showing the most significant variations are those concerning the myogenic differentiation and that no gene from chromosome 4 in the vicinity of the D4Z4 region was found to be significantly altered. François Leterrier commented that these modifications of the myogenic differentiation in FSHD had independently been confirmed by two other groups [5], [6]. K. Flanigan then presented his very recent results concerning μ-crystallin over-expression in FSHD muscle observed by proteomics [7].

The discussion has followed on the difficulties encountered for the standardisation of transcriptome studies. Rosella Tupler indicates that there is less FRG1 expression in males than in females. Alexandra Belayew mentions the PITX1 gene implicated in the asymmetrical development of limbs as well as in inflammatory processes. This gene, present on chromosome 5, was shown by Yi Wen Chen [8] to be specifically activated in FSHD muscles as compared to 11 other neuromuscular disorders. This gene is activated in trans by DUX4 (see below). The PITX1 protein is a transcription and differentiation factor.

Cecilia Gelfi showed proteomic results obtained in her laboratory [9]. The principal finding was the increased expression of type I myosin. The most prominent protein variations were the following (observations on FSHD patients, aged more than 50 years):

The discussion on this topic commented on the high methodological difficulties encountered in proteomic studies. Considerable variations are observed from one laboratory to the other. To analyse muscle tissue, the most powerful method available today is 2D-Dige, which allows separation and quantification of a number of contractile, metabolic and signal proteins providing an overview of what is altered at the molecular level at different stages of disease. This methodology is not suitable for the high molecular weight contractile proteins (such as titin, nebulin). For this specific group of proteins a chromatographic separation followed by mass spectrometry might be more suitable. Unfortunately the latter methodology on muscle proteome studies is still in his infancy and no references in literature are available regarding muscle diseases characterization.

Yegor Vassetzky presented his results on the « geometric » structure of the D4Z4 array [10], [11]. A shortening of the D4Z4 region provokes a modification of its binding to the nuclear matrix. The normal D4Z4 array has two loops: the pathologic, shortened one has only one loop. The experimental demonstration of this finding is very well documented and it is very convincing since an accurate cellular imagery allows us to visualize these loops by FISH. Yegor Vassetzky indicated that this mode of binding to the matrix concerns the 4qA and not the 4qB allele.

This result was discussed as a very important piece of the “FSHD puzzle”. The question was raised: how does it correlate to the modification of gene transcription? An hypothesis to be explored further might be the following: are these loops directly implicated in the binding of chromosome 4 to the nuclear matrix? It has been suggested that a possible binding might occur mediated through lamin A or to a protein interacting with lamin A [12].

Alexandra BELAYEW has been studying the DUX4 gene for many years (first DUX4 mention in [1]). She mentioned the presence of a DUX4c gene near FRG2. DUX4c is located 42kb centromeric to the D4Z4 array in the vicinity of the nuclear matrix attachment site (MAR) that is lost in FSHD myoblasts [10], so that DUX4c, FRG2 and FRG1 are brought into a single chromatin loop with the D4Z4 repeat array. With a specific antiserum, she observed increased DUX4c protein expression in FSHD muscle biopsies with decreasing D4Z4 copy number (collaboration with Dalila Laoudj-Chenivesse) These data suggested that DUX4c could be a “sensor” of the chromatin structure. Her group has developed a monoclonal antibody against DUX4 and DUX4c with an affinity in the femtomolar order: the proteins are distinguished on a Western blot by their different sizes. This antibody allowed to detect the DUX4 protein in FSHD but not control myoblast extracts [9]. In collaboration with Alberto Rosa’s team in Argentina, A. Belayew’s group demonstrated that DUX4 forced expression was toxic in different cell lines [13]. It disturbed the nuclear distribution of emerin, induced caspase 3/7 activity and caused cell death. On the other hand, DUX4 activated the PITX1 gene promoter and induced the expression of the corresponding endogenous protein in C2C12 cells [8]. In consequence, A. Belayew proposed the following hypothesis: the shortening of the D4Z4 array activates expression of the DUX4 genes in the residual D4Z4 repeats, the DUX4 protein activates expression of the PITX1 gene, and both DUX4 and PITX1 proteins which are transcription factors disturb the gene expression pattern leading eventually to muscle pathology and cell death. In the discussion the methodological problem was raised: the transfection level of the DUX4 gene that results in cell death might be a consequence of using the very potent pCINeoDUX4 expression vector that results in too strong over-expression.

Rosella Tupler recalled the central hypothesis of her work: when the D4Z4 array is shortened, the complex “YY1, HMGB2, nucleolin” does not bind on it any more, and that induces the over-expression of FRG2, FRG1 and ANT1 [14], [15]. The over-expression of the human FRG1 gene in a transgenic mouse model induces a muscular dystrophy, the severity of which is proportional to the level of FRG1 protein expression. Diaphragm and masseter muscles are not affected in this model. Trapezius and vastus lateralis are the more affected. FRG1 protein is found in the nucleolus, in the Cajal bodies and in the nuclear “speckle”. It co-purifies with the spliceosome. FRG1 interacts with the polyA-binding protein. Splicing modifications are observed in these transgenic mice, they concern the pre-mRNA Tnnt3 and Mtmr1. This anomaly was also observed in FSH patients. R. Tupler proposed a pathologic model based on a modification of the alternative splicing, and possible analogies to myotonic dystrophy were raised [16].

Richard Lemmers presented his recent results on the specification of the FSHD locus. The D4Z4 proximal zone, where the FRG2 and DUX4c genes are located, seems not essential. Previously, two subtelomeric 4q variants have been identified, 4qA and 4qB. Only D4Z4 contractions on 4qA are pathogenic, the reason for this observation is unknown. By extensive genotyping, Lemmers showed that the subtelomeric region of chromosome 4q can be subdivided in nine different haplotypes. Interestingly, three different 4qA haplotypes have been identified, but only one has been shown to be pathogenic in 86 FSHD patients (4qA161). D4Z4 contractions on another 4qA haplotype (4qA166) and on the 4qB163 haplotype do not cause FSHD. It has been proposed that haplotype-specific sequence variations are essential for the 4qA161 specificity of FSHD [17], [18].

3. Therapeutic approaches to FSHD 

Julie Dumonceaux presented her experimental work aiming to develop a gene therapy approach for FSHD. She has induced a strong expression of the mouse FRG1 gene by its transfection in the mouse tibialis anterior muscle. The gene was packed up in an AAV and injected directly into the muscle. Mice were sacrificed at 1, 3 and 6 months. The FRG1mRNA was strongly expressed. At month 1, an important inflammation and changes compatible with a dystrophic process were observed and the muscle weight decreased. At month 3, there was no more inflammation; the necrosis-regeneration process was still observed, the muscle weight was not further modified. After month 6, the results were not known at the date of the meeting. Using RNA interference, it was possible to decrease the FRG1 mRNA over-expression and to reverse its dystrophic effect on skeletal muscle. This work seemed to corroborate an important role of FRG1 in a molecular mechanism leading to muscular dystrophy as had previously been suggested by the work of Rosella Tupler. The additional question was raised, what level of over-expression of FRG1 would be required to initiate such a process, and if the levels induced by a viral approach were not incompatible with those that had been observed in some studies on human tissue and not been confirmed by others [14], [18].

Claude Desnuelle recalled that cell therapy could be a useful means to slow down or stop the muscle destruction and might even favour regeneration. In the case of FSHD, the clinical trial coordinated in Nice is based on injecting cells obtained from a clinically non-affected muscle from a FSHD patient into an affected muscle of the same patient. Between the explant and the re-injection there is an important step of cell-expansion under GCP conditions performed by the company MYOSIX. This auto-transplantation will avoid immunological difficulties hampering cell therapy approaches. This clinical trial had been approved by the French regulation authority concerned by human therapeutics assays (AFSSAPS). Three patients have been treated so far. Six months after the injection, no major adverse effects were observed, but no functional amelioration has been documented so far. A Belayew raised the concern that the injection of apparently normal cells into a diseased muscle may “wake up” the pathological process in this tissue. C. Desnuelle answers that it is impossible to know the response without taking this risk, since no animal model exists. Another unsolved problem raised was that it is unknown if myogenic progenitors expanded from a given muscle such as the quadriceps femoris would reproduce a ‘quadriceps-specific’ expression pattern when expressed into a host muscle such as the deltoid. However, the trial had not been designed to answer this latter question as this would have necessitated additional muscle specimens for a transcriptional study.

Dalila Laoudj described the patho-physiological bases for antioxidant supplementation in FSHD. Experimental obtained data (in publication) suggest that mitochondrial dysfunction and oxidative stress play a role in FSHD pathogenesis. The objectives of a clinical study which is planned are to demonstrate the improvement of mitochondrial dysfunction and oxidative stress by antioxidant treatment and demonstrate their effect on peripheral skeletal muscle function and exercise tolerance. The study will include a total of 10 FSHD patients (between 21 and 54 years old), walking and climbing stairs without assistance, and 10 age and sex-matched healthy human volunteers with similar physical activity.

Rosella Tupler presented her project of pharmacological screening aiming to find molecules able to inhibit the FRG1 gene expression by interaction with its promotor. A construction associating the FRGI and luciferase gene, and the FRG1 and SV40 promotor is introduced into C2C12 cells. The readout implies that an active molecule must decrease the luminescence. Among 7000 tested molecules, 309 had such an effect. With another test, 28 hits were obtained. The molecules active in these test have complex structures associating many 6-atom-cycles, some of them are highly substituted N-heterocycles. Compounds thereby obtained will be further studied regarding their toxicity and effect in vitro and possibly in vivo.

In conclusion, new pieces complementing the puzzle of FSHD pathogenesis have been obtained but they do still not comfortably integrate into a unified theory of FSHD pathogenesis. The progress which is immediately useful for the patients concerns the molecular diagnostics and consequently the family counselling. Regarding the therapeutic options, none of the approaches discussed can claim a definite positive effect on muscle function in the short term, but clearly the strategies explored deserve to be followed in more detail. But it was very satisfying to hear the vivid and open exchanges between the scientists present in this meeting. We can hope that efficient cooperation will accelerate the unveiling of the FSHD mystery.

Participants