158th ENMC international workshop on congenital muscular dystrophy (Xth international CMD workshop) 8th–10th February 2008 Naarden, The Netherlands

1.1.1. Spectrum of dystroglycanopathies and genotype–phenotype correlations 

A series of presentations were made on conditions related to mutations in the known genes responsible for dystroglycanopathies (POMT1, POMT2, POMGnT1, FKTN, FKRP and LARGE). Muntoni discussed the experience of the London group, and updated the results of a recently published manuscript [1]. A total of 180 dystroglycanopathy patients have been studied for mutations in the known genes as part of a large international collaborative effort. Forty-four patients had clinical or structural evidence of brain involvement while the remaining cases had only muscle weakness with clinical severity ranging from severe congenital variants to relatively mild limb girdle variants. Mutations in the FKRP genes were the most frequent – 77 cases, of which 59 had LGMD2I; 14 had CMD with [8] or without [6] cerebellar cysts; 2 had a condition of equivalent severity to Muscle Eye Brain disease (MEB) and 2 to Walker-Warburg syndrome (WWS). The second most frequently mutated gene was POMT1, with 12 cases of which 6 had WWS, 1 CMD with microcephaly and 5 a LGMD variant also associated with microcephaly. Mutations in POMT2 were identified in 11 patients, 2 with WWS, 7 in patients with MEB-like, 1 CMD and 1 LGMD with microcephaly. Mutations in POMGnT1 were identified in 9 patients, 6 affected by WWS, 1 by MEB, and 2 affected by LGMD. Mutations in FKTN were identified in 8 families, 2 affected by WWS, 3 CMD with no structural brain involvement and 3 families affected by LGMD (and no brain involvement). Regarding LARGE, mutations were identified in 3 patients, 1 affected by WWS, 1 by MEB and 1 which was the previously described MDC1D case. Mutations in these 6 genes accounted for ∼64% of cases. Guicheney discussed the experience of the Paris group specifically on patients affected by CMD, with or without mental retardation and no mutation in FKRP. The strategy followed for mutation detection was that of performing linkage analysis first in the informative families, followed by mutation detection of the relevant gene. Twenty-seven patients were genotyped, belonging to 21 families. The most commonly mutated gene was POMT2, with 10 patients, followed by POMT1 (9 cases); FKTN (4 cases), POMGnT1 (3 cases) and 1 case with LARGE mutations. The linkage analysis allowed the identification of a founder mutation in POMT2, present in patients from Europe and also Morocco [2].

Topaloglu reported the Ankara experience (a few of these patients have been molecularly characterised in London and were, therefore, also mentioned in the presentation by Muntoni): mutations in POMGnT1 were identified in 7 patients with MEB, including cases with profound autistic behaviour but only mild muscle weakness; mutations in POMT1 in 8 cases with “LGMD2K” [3], and 1 with WWS; mutations in FKTN in 3 cases (of which 1 was of WWS severity and the others milder CMD variants).

Mercuri reported the experience of the Italian CMD network which collected 70 dystroglycanopathy CMD cases. Although the mutation analysis was incomplete at the time of the workshop as screening for LARGE and FKTN had not yet been performed in all cases, although no patient with mutations in these 2 genes has been identified so far, mutations in the remaining genes were identified in 37 cases, of which 14 carried POMT1 changes, 11 POMGNT1, 7 FKRP and 4 POMT2 [4] The previously reported cases of “Italian MEB” [5] were found to have mutations in either POMT1 or POMT2 [6], [7].

Van Bokhoven illustrated the cohort of patients studied in Nijmengen. This series of patients is at the severe end of the clinical spectrum, mostly affected by WWS or severe MEB-like conditions. Mutations in POMGNT1 were found in 11 families; FKTN mutations in 4 families, with 2 more families carrying possible mutations; FKRP was involved in 3 families, with 2 additional families carrying possible FKRP missense mutations. Mutations in POMT1 and POMT2 were found, respectively, in 21 and 7 families, while mutations in LARGE were found in 1 family [6], [8]. There was no clear genotype–phenotype correlation for most genes, although none of the patients with POMGNT1 mutations had changes similar to those found at the severe end of the WWS spectrum.

Seta presented the result of a large collaborative French study, in which a population of affected fetuses in whom complete neuropathological evaluation was available. Forty-seven fetal samples were collected from 41 families in the period 1990–2006. Mutations were identified in 23 of these 41 families (56%): 14 of these carried POMT1 mutations; 6 POMGnT1 mutations, 3 POMT2 mutations, 1 FKRP and 1 LARGE mutation. More recently an additional 18 fetuses were identified and a similar frequency of mutations identified. In terms of genotype–phenotype correlation, the most severe end of the spectrum included other organ involvement (abnormal genitalia, lung lobulation, renourinary system, cleft lips and palate, heart malformations). These features were found in fetuses carrying POMT1 or POMT2 mutations and in the single case with FKRP mutations, while POMGnT1 mutations were associated with milder phenotypes [9].

Vissing discussed the mild end of the dystroglycanopathy spectrum. Molecular analyses of 99 Danish LGMD2 patients revealed that 38 had LGMD2I [10]. All 38 patients carried the common 826C>A mutation, in either homozygous or compound heterozygous form. This makes LGMD2I the commonest cause of LGMD in Denmark (∼40%), with a heterozygote frequency of the common mutation of 1:200. A third of the LGMD2I patients, homozygous for the common mutation, presented with myoglobinuria, mimicking a metabolic condition.

Moore presented a large series of North American patients with dystroglycanopathy (both CMD and LGMDs) studied in Iowa City. One hundred and sixty-eight dystroglycanopathy cases were ascertained. The mutation distribution was as follows: 50 FKRP cases (26 LGMD and 24 CMD); 7 FKTN cases (1 LGMD and the remaining CMDs, including 3 WWS); 3 POMT1 cases (all severe CMD/WWS); 5 POMT2 cases (all severe CMD/WWS), and 2 POMGnT1 cases (both MEB). (Some of the FKRP-LGMD cases were reported in Moore et al. [11].

1.1.2. Diagnostic issues 

Sewry reviewed the muscle pathology of 17 patients studied in London carrying mutations in the different genes. The antibodies used were IIH6 and VIA4-1, which are both directed against glycosylated epitopes, and 2 antibodies raised against the core protein of alpha-dystroglycan (GT20 polyclonal raised by Campbell and 1 peptide polyclonal antibody raised by Kroger), laminin α2, laminin α5 and β dystroglycan. Immunolabelling of ADG ranged from absent or traces, to mild or minimal levels. The main outcome of this study was that there appeared to be a better correlation between the degree of hypoglycosylation and the phenotype in POMT1, POMT2 and POMGnT1 and LARGE but less so in FKTN, and, to some extent, in FKRP. In addition in individual patients GT20 labelling appeared reduced, while labelling with the peptide antibody raised by Dr. Kroger was more frequently severely reduced. Laminin α2 was frequently reduced especially in the patients at the severe end of the spectrum, and laminin α5 upregulated (although some of this related to immature fibres) [12].

Barresi reviewed the Newcastle experience with Western blots in 18 patients with a dystroglycanopathy, of whom 10 were molecularly confirmed. She stressed that the secondary reduction or absence of laminin α2 80kD band in LGMD2I, was possibly present with other gene defects although this has been less well investigated. She also reported that in all FKRP patients calpain 3 was variably down-regulated on Western blot; conversely, in a few patients with LGMD2A ADG appeared abnormal on section.

1.1.3. Glycobiology of dystroglycanopathies 

Guicheney discussed the role of POMGnT1 and POMT enzymatic activity in immortalised lymphocytes as an aid to the diagnosis. In collaboration with Endo she characterised patients with mutations in FKRP, POMT1, POMT2 and POMGnT1. The POMGnT1 and POMT enzymatic activities are consistently normal in patients with FKRP mutations but invariably reduced in patients with the relevant mutations in POMT1, POMT2 and POMGnT1, respectively. There was no strict correlation between clinical severity and degree of enzymatic down-regulation. This analysis, therefore, appears helpful both to direct the molecular diagnosis and to assign pathogenicity in patients carrying previously unreported missense mutations [13].

Schachter reminded the participants that, as in other disorders of glycosylation, proteins other than glycosyltransferases can participate in the process and hence be candidate proteins for uncharacterised CMD variants. In particular lectins, nucleotide sugar transporters, convertases, glycosidases, conserved oligomeric Golgi proteins (COGS), all participate to the process of glycosylation. He also mentioned that while O-mannosylation plays a crucial role in ADG function, and accounts for ∼70% of all ADG glycans, other glycosylation processes are also involved and could potentially be responsible for dystroglycanopathy cases.

Campbell suggested that in addition to the Man-O-Ser/Thr and GalNAc-O-Ser/Thr cores, ADG most likely contains a third, unknown structure mediated by LARGE. His group previously demonstrated that the N-terminal domain of ADG interacts directly with LARGE and this is a requirement for normal glycosylation [14]. The interaction between LARGE and dystroglycan occurs despite the N-terminal domain of ADG being cleaved in the mature protein, [14]; so far the precise portion of the N-terminus that is binding LARGE has not been determined.

Kanagawa presented studies aimed at addressing fukutin function. The transmembrane domain of fukutin is responsible for its localisation in the Golgi apparatus; as POMGnT1 is also localised in the Golgi, recent co-localisation and co-immunoprecipitation studies have shown that these 2 proteins physically interact and that the interaction is mediated by the fukutin transmembrane domain. He went on to show that the Y371C FKTN mutation induces a mis-localisation of fukutin in the ER and a secondary mis-localisation of POMGnT1. Interestingly fukutin deficient mice have reduced enzymatic POMGnT activity [15]. Similarly, the brain of FCMD patients has been previously shown to have reduced POMGnT activity (Noguchi et al., 2006 World Muscle Society meeting abstract book, Neuromuscular Disorders 16, S54, 2006).

1.1.4. Search for novel genes 

Several groups are investigating the molecular basis of the dystroglycanopathies for which no involvement of any of the 6 known genes is found. Regarding the candidate gene approach, Seta has studied the genes B4GALT1 and B4GALT2; GnTIX (GnTVB) and DAG1 in the population of lissencephalic fetuses after excluding mutations in the known genes, but failed to identify any mutation.

The group of van Bokhoven is following both a candidate gene approach and a genome wide linkage strategy. They have collected a large population of consanguineous WWS families (>100) of which only 30% carry mutations in 1 of the known genes. Regarding the candidate gene strategy approach, genes that have so far been studied unsuccessfully are DAG1, FUT9, SDF2, SDF2L1, B4GALT1-6, ST3GAL3-6, DPM1-3 and iGnT. Regarding the genome wide studies, informative families are being studied and a number of candidate gene loci are being investigated further.

Cirak and Brown illustrated the strategy being followed in London, utilising both candidate gene and linkage analysis approaches. Candidate genes studied so far included LARGE2/GYLTL1B, GnTIX, iGNT, GYG1 and DAG1, while refinement of previously identified loci is being pursued in collaboration with Voit both for MDC1B, and for a locus for a CMD/LGMD variant without brain involvement, which has been mapped to chromosome 7. Clarke described the consanguineous families that are being studied in Paris using genome wide linkage mapping techniques.

1.1.5. General discussion 

The pick-up rate of mutations of the 6 genes responsible for dystroglycanopathies accounted for ∼30–60% of cases, probably as a result of the different populations and inclusion criteria of the various studies presented. This suggests that more genes are involved in dystroglycanopathy cases. However, it is also possible that rare mutations in the 6 genes are missed by the commonly used genomic mutation screening approaches: Guicheney for example reported the identification of 2 deletions in POMT2 and an insertion in LARGE, which were found using a cDNA approach. It is, therefore, likely that the prevalence of mutations studied at the genomic level and reported so far might be an underestimate of the actual figure. On the other hand heterozygosity for previously reported missense or nonsense mutations in patients with a dystroglycanopathy are not common, and this indirectly suggests that deletions or duplications have probably not a very frequent occurrence. The absence of a clear digenic inheritance in any of the patients identified so far was also highlighted.

The issue of how to attribute pathogenicity to previously unreported missense changes was discussed, and the role of the enzymatic assays for POMT and POMGnT1 stressed. However, even these assays need to be interpreted with caution, as illustrated by the case of 1 of the LGMD patients with mutations in POMGnT1 reported by Muntoni. This woman carried a homozygous point mutation c.1666G>A leading to the substitution of a conserved aspartic acid at position 556 to asparagine (p.Asp556Asn) in the POMGnT1 protein [16]. This change was in linkage with the condition as it was not found at the homozygous state in her 4 healthy siblings, was absent in more than 100 controls and the patient with this homozygous change had altered POMGnT1 enzymatic kinetics in fibroblast cultures. However, Seta reported the identification of 1 unaffected individual carrying the same homozygous change, which she also identified at the heterozygous state in 7 individuals. This would indicate that this sequence variant is likely to be a polymorphism; further studies aimed at understanding the significance of the altered POMGnT1 kinetics in the patient are planned. From a diagnostic perspective, the difficulty in recognising partial merosin deficiency from a dystroglycanopathy was emphasised. It was recommended to study carefully laminin α2 expression using antibodies which recognise both the 80 and 300kDa fragment in all cases in whom a dystroglycanopathy is suspected. While partial reduction of laminin α2 is observed frequently in dystroglycanopathy patients, so far no patient was reported to have complete absence for laminin α2 using the aforementioned antibodies; a complete absence of staining with at least 1 the laminin α2 antibodies is a strong indicator for MDC1A.

1.2.1. Animal models of dystroglycanopathies 

1.3.1. Brain involvement in dystroglycanopathies 

Muntoni reported the spectrum of brain involvement detected on MRI in a population of 26 patients with proven mutations in POMT1 [4], POMT2 [9], POMGnT1 [7], FKTN [4] or LARGE [2]. A wide spectrum of structural defects were documented, ranging from complete lissencephaly in patients with Walker-Warburg syndrome-like disorders to isolated cerebellar involvement. Cerebellar cysts and/or cerebellar dysplasia and hypoplasia were the predominant features in 4 patients. Polymicrogyria (n=12) appeared more severe in the fronto-parietal regions in 7 cases, with an occipital-frontal gradient in 2 cases whilst involvement was more generalised in 3 cases. The signal intensity of the corticospinal tracts was abnormal in 4 patients, all with a Muscle–Eye–Brain–like phenotype. No pattern of abnormality unequivocally allowed the identification of the primary gene defect; however, prominent cerebellar cysts were seen in all patients with POMGnT1 mutations, while they were only rarely observed in cases with POMT1 and POMT2 mutations. On the whole the gradient of the supratentorial changes appeared more variable than has been previously described in FKRP gene mutations.

Mercuri summarised the results of a collaborative Italian study; this included 11 patients with POMGnT1 mutations; 14 with POMT1, 5 with POMT2 and 7 with FKRP mutations. The main outcomes of this study were that the predominant phenotype in POMGnT1 deficient patients tended to be that of MEB and that severe cerebellar hypoplasia was more commonly seen in patients with POMT1 or POMT2 deficiencies. The wide spectrum of involvement seen in patients with involvement of the same gene was highlighted once more.

Voit reported data from patients carrying a severe FKRP mutation, Tyr307Asn, associated with a MEB phenotype. This FKRP mutation, originally reported at the homozygous state in 1 MEB patient, was found also at the homozygous state in 3 more MEB cases. These patients had severe cerebellar dysplasia with cerebellar cysts; although neuropathology of the cystic lesions in these patients was not available, studies on WWS fetuses suggest that the cysts originate prenatally and are the result of entrapment of islands of meningeal tissue between adjacent folia.

Interestingly, Vissing reported that the Tyr307Asn mutation was found in the heterozygous state in 6 Danish patients, suggesting a possible founder Scandinavian mutation. These patients all carried, in the other allele, the common Leu276Ile FKRP mutation and followed a relatively severe LGMD2I phenotype, resembling DMD, in keeping with the interpretation that the Tyr307Asn is a relatively severe mutation.

Clarke reported 1 patient with LARGE mutations who, at variance with the originally described MDC1D patient, had cerebellar dysplasic changes and cerebellar cystic lesions.

Brown discussed in detail the brain phenotype of the FKRP deficient mice carrying the homozygous “MEB” mutation. The brain of these mice showed a clear disruption of the neuronal layering of the cerebral cortex and partial fusion across the interhemispheric fissure. The radial glial cell population in these mice was highly disorganised; there was also a marked alteration in the density and distribution of nuclei at different levels throughout the cortex, and a reduction, relative to control littermates in the total number of nuclei near the pial basement membrane. Immunocytochemistry indicated a defective basement membrane that resulted in a disruption of both radial glia and neurons.

Moore discussed in detail both the role of dystroglycan in determining the CNS phenotype observed in dystroglycanopathies and also which cell types may be responsible for the CNS involvement, neurons or glial cells. These issues are being addressed by using a variety of conditional KO mouse models in which the DG gene is selectively disrupted in different subpopulations of cells.

Regarding the first question, nearly complete deletion of DG in brain using the GFAP-CRE/DG null mice suggests that ablation of DG is sufficient to cause the CNS malformations observed in dystroglycanopathies [22]. These mice have a selective loss of DG in brain, spinal cord and peripheral nerve, and have severely disrupted glia limitans, evidence of cerebral cortical dysgenesis with fusion of hemispheres and glial-neuronal heterotopia, fusion of adjacent lobules of the cerebellum, and malformations of the hippocampal dentate gyrus. More complete deletion of DG earlier in development using Mox-2-Cre causes all the developmental pathology seen in the GFAP-Cre mice and additionally results in hydrocephalus and ocular pathology that closely mimics WWS [17]. Hydrocephalus does not appear to be the result of aqueductal stenosis, but possibly is caused by occlusion of the subarachnoid space by heterotopic glial-neuronal tissue. This, therefore, suggests that deletion of DG is sufficient to create the broad spectrum of CNS abnormalities detected in patients with WWS.

Regarding the second question, Moore discussed unpublished data from lines of mice in which the CRE deletion of DG is driven by additional promoters. These data suggest that dystroglycan expression in glial cells is required for appropriate CNS formation.

1.4.1. Therapeutic perspectives 

The discussion focused on current or planned human studies, and on preclinical developments. Several participants (Muntoni; Vissing, Voit) reported anecdotal evidence of improvement in muscle strength and function in dystroglycanopathy patients treated with corticosteroids. In these patients the corticosteroid therapy had been initiated following an erroneous diagnosis of polymyositis. Some of these cases have been described in the literature and appeared to document a sustained benefit following the initiation of the therapy. It was discussed that the role of corticosteroids in dystroglycanopathy patients ought to be explored in more detail, and that LGMD2I would be an ideal population of patients to study. It was decided to form a steering committee to discuss the trial design, outcome measures and sponsorship. It was decided that TREAT-NMD could take the lead in the organisation of this collaborative trial.

Vissing discussed the beneficial effect of aerobic exercise training in LGMD2I. He studied a cohort of patients who received a 12-week aerobic training exercise program, consisting of a total of 50 sessions of 30min each [23]. In several of the patients studied, a muscle biopsy was performed before and after the 12-week sessions, to assess if the training had resulted in increased muscle damage. No sign of additional muscle damage was detected; from a functional perspective there was a statistically significant improvement of oxidative capacity and endurance, and all patients reported improvements in a questionnaire on activity of daily living. Regarding the preclinical studies, Barresi summarised work performed in collaboration with Campbell which showed that the overexpression of LARGE affects the post-translational modification of ADG by inducing the synthesis of species that are enriched in glycans recognised by the IIH6 antibodies [24], [25], [26]. This hyperglycosylation of ADG is followed by increased laminin binding. A closely related homologue of LARGE, glycosyltransferase-like 1B (GYLTL1B) or LARGE2 [26] also induces ADG hyperglycosylation [25]. Transfection of patients’ fibroblasts and myoblasts with recombinant LARGE adenovirus is capable of inducing the synthesis of the IIH6 antigen even in patients in whom the O-mannosylation pathway is disrupted, such as for example in POMT1 and POMGnT1 [24]. This indirectly suggests that LARGE may exert its action via the modification of N-glycans and/or mucin O-glycans rather than on the O-mannosyl residues [27]. This alternative structure mimics the O-mannose glycan in its ability to bind the extracellular matrix proteins associated with ADG, and therefore, seems capable of bypassing defects in patients with a dystroglycanopathy [28]. The upregulation of LARGE could, therefore, represent an effective therapeutic strategy for patients with dystroglycanopathies. Muntoni presented the recently generated transgenic lines overexpressing LARGE under the control of a ubiquitous promoter. Preliminary data on 2 transgenic lines suggests that the long-term overexpression of LARGE was not associated with any detrimental effect (Brockington et al., in preparation). Tinsley suggested that upregulation of LARGE could be achieved either by the identification of compounds active on its promoter (and/or LARGE2); or with compounds which would affect the recycling of LARGE (and/or LARGE2) from the Golgi. Different possibilities are represented by the identification of pharmacological agents able to affect the transcription of LARGE; or drugs aimed at modulating the recycling of LARGE, so that the concentrations of LARGE at the Golgi are increased. The pharmacological upregulation of LARGE (or LARGE2) could be particularly valuable for patients affected by the common LGMD2I variant as they have only a relatively modest decrease in ADG glycosylation.