Dystroglycan and dystroglycanopathies: Report of the 187th ENMC Workshop 11–13 November 2011, Naarden, The Netherlands

1. Introduction

The 187th ENMC Workshop on dystroglycan and dystroglycanopathies in Naarden from 11th to 13th November 2011, brought together 20 researchers from seven different countries (Germany, Italy, The Netherlands, Sweden, Switzerland, UK and USA) working on the clinical and basic aspects of the post-translational modification of dystroglycan. Topics included the pathophysiology in patients, animal models of dystroglycanopathies, and cellular approaches addressing the effects of post-translational modification on dystroglycan function in a variety of systems. Specific issues which were addressed included factors that might determine the development of central nervous system involvement in patients, if current cell and animal models of the dystroglycanopathies are appropriate for the study of the disease and their utility for therapeutic screening and whether dystroglycan signalling pathways in the cytoskeleton, cancer and polarity can provide new insight into the dystroglycanopathies.

Dystroglycan is a core and essential component of the dystrophin glycoprotein complex of muscle and brain. With one rare exception [1], no primary mutations are known in dystroglycan itself, presumed to be due to embryonic lethality. However mutations have been identified in a number of genes involved in the post-translational modification of dystroglycan giving rise to a large number of secondary dystroglycanopathies. The post translational modification of dystroglycan by either glycosylation, phosphorylation or proteolysis has profound effects on the functional capabilities of dystroglycan.

Several forms of neuromuscular disease are now known to be associated with mutations in genes including; POMT1, POMT2, POMGnt1, LARGE, fukutin, fukutin related protein (FKRP) and most recently DMP2 and DMP3. All encode for proteins that are either putative or determined glycosyltransferases or in the case of fukutin and FKRP proteins that are of unknown function but are nonetheless necessary for α-dystroglycan glycosylation lending support to the idea that the aberrant post-translational modification of proteins represents an important mechanism of pathogenesis in the muscular dystrophies [2–4]. One key characteristic of the muscle of patients with mutations in these genes is a marked hypoglycosylation of α-dystroglycan [5–7] which has led to the suggestion of dystroglycanopathies as a general term to describe these conditions.

α-Dystroglycan is a central component of the dystrophin associated complex and is expressed in a number of different tissues including the brain. Its abnormal glycosylation affects its interaction with members of the extracellular matrix, including laminin, perlecan, neurexin and agrin within the basement membrane. This reduction in ligand binding within the basement membrane is thought to underlie both the muscular dystrophy and the structural brain defects. Mutations in POMT1/2 [7,8], POMGnT1 [9], fukutin [10]), fukutin-related protein (FKRP) [11], LARGE [12,13] are responsible for a spectrum of clinical phenotypes that ranges from Walker–Warburg syndrome (WWS), Muscle–Eye–Brain disease (MEB), Fukuyama-type muscular dystrophy (FCMD), congenital muscular dystrophy (CMD) types MDC1C and MDC1D and the limb girdle type 2 variants with onset in childhood or adult life (LGMD2I, LGMD2L, and LGMD2N) [14–17]. At the severe end of the spectrum these diseases are associated with central nervous system involvement that present as cortical malformations (polymicrogyria and cobblestone lissencephaly) and ocular defects in addition to muscular dystrophy.

Interestingly the phenotype of patients belonging to this group of disorders depends not so much on the specific gene primarily affected (i.e. POMT1, POMT2, POMGnT1, LARGE, FKRP or fukutin) but rather the severity of the specific mutation and presumably its effect on the structure and thus the function of the gene product [16].

Whilst the hypoglycosylation of α-dystroglycan is the key pathogenic defect in the dystroglycanopathies, other posttranslational modifications such as phosphorylation and proteolysis have been recorded that also have profound consequences for dystroglycan function [18]. Indeed there is evidence to suggest that some of these latter modifications may also be consequent on the former. For example in many adenocarcinomas where α-dystroglycan is hypoglycosylated due to downregulation of LARGE [19], there are associated changes: phosphorylation and proteolysis in β-dystroglycan [20,21]. Although studied less extensively in muscle, it is extremely likely that secondary post-translational modification of β-dystroglycan is an important part of the aetiology of the dystroglycanopathies.

2. Clinical and pathological features of the dystroglycanopathies

Francesco Muntoni opened the workshop by presenting an overview of his experience of the phenotypic spectrum of patients affected by a secondary dystroglycanopathy.

He first described the pragmatic classification that his group proposed in 2007 [16] and re-elaborated in a recent review [22]. According to this proposed scheme, patients with mutations in the six classical genes involved in secondary dystroglycanopathies (POMT1; POMT2; FKRP; FKTN; POMGnT1; LARGE) are divided into seven different categories, ranging from severe WWS to adult onset LGMD without any central nervous system involvement, and with a number of intermediate categories that represent the spectrum observed in this group of conditions. These include CMD with structural brain defects (with and without associated with eye involvement) of the clinical severity seen in FCMD and MEB; CMD with isolated cerebellar cysts (CRB); CMD with mental retardation (MR) and normal brain structure; CMD with normal intelligence and normal brain structure; LGMD with normal brain structure but mental retardation; LGMD with normal brain structure and function (See Table II in [22]).

When analysing the data of the correlations between genotype and phenotype in the UK population, the following was observed:

• WWS: – Most common: POMT1 = POMGnT1 (however POMGnT1 consistently at the milder end of spectrum); – Less common: POMT2 = FKRP = FKTN; – Rare: LARGE.

• MEB/FCMD like: – most common. POMT2 – less common POMT1 and FKRP and POMGnT1; – Rare LARGE.

• CMD-CRB: – Common: FKRP = POMGnT1; – less common FKTN; – rare: POMT1 or 2/LARGE.

• CMD-MR: – common POMT1 and POMT2; – rare: FKTN.

• CMD no MR: – common FKRP – less common FKTN.

• LGMD-MR: – Only observed with POMT1 and POMT2.

• LGMD no MR: – common FKRP; Less common FKTN; – v.rare POMGnT1/POMT2.

Additional rare variants of the secondary dystroglycanopathies include patients with defects in DPM2 and DPM3 who show abnormalities in both N and O linked glycosylation and whose clinical phenotypes include CMD with severe CNS involvement (DPM2); and a form of LGMD-MR (DPM3).

Professor Muntoni also described two different families with a FCMD/MEB like CMD variants with mutations in two novel genes responsible for a form of secondary dystroglycanopathy. The approach taken for the gene identification was that of whole exome sequencing in collaboration with the UK10K Wellcome Trust Sanger Centre initiative in Cambridge. At the time of the workshop, no other family had been identified with mutations in these two novel genes. However work done also in collaboration with the Sanger Centre using morpholino knock-down of the respective transcripts in zebrafish resulted in affected morphants with a clear neuromuscular and central nervous system phenotype highly resembling the one observed in the affected families. Further families are being screened for mutations in these novel genes.

Hans van Bokhoven described the work of his group to identify known and novel mutations at the severe end of the dystroglycanopathy spectrum. For genetic studies, they have collected DNA and cell lines from patients diagnosed with Walker–Warburg syndrome (WWS) or the slightly milder Muscle–Eye–Brain (MEB) phenotype. The patients are being tested through a DNA-diagnostic protocol for the six genes known to be associated with this severe end phenotype; POMT1, POMT2, POMGnT1, FKTN, FKRP and LARGE. In consanguineous families the mutation analysis is preceded by a SNP microarray analysis (250K Affymetrix SNP array) to identify potential homozygous copy number variants (CNVs) and regions of homozygosity. Any of the known six genes that falls into a region of homozygosity will be tested for mutations by Sanger sequencing of the corresponding genes. In his labs experience (130 families), mutations in the six genes are found in approximately one third of all cases. Of the remaining families, a total of 61 are used for further genetic research aimed to identify new dystroglycanopathy genes. Of these, 30 have been used for SNP-array homozygosity mapping, which revealed no common locus for the remaining families. Rather, the array data suggest the existence of at least nine other genes that may underlie WWS/MEB.

To identify these novel genes, they have followed two parallel strategies. First, on the basis that the hypoglycosylation of α-dystroglycan is a common denominator in WWS/MEB, they have selected candidate genes encoding proteins that are predicted or hypothesised to be involved in post-translation modification of dystroglycan. Examples of such genes include, SDF2/SDF2L1, DPM3 (dolichyl-phosphate beta-D-mannosyltransferase), Galactosyl/sialyltransferases, B4GalTs/ST3Gals, DAG1 (dystroglycan), GMPPB (GDP-mannose pyrophosphorylase B), Furin (paired basic amino acid cleaving enzyme) and Fut9 (fucosyltransferase 9 (alpha (1,3) fucosyltransferase). A possible homozygous mutation was identified in one of the glycosyltransferase genes. The causative nature of the mutations was confirmed by functional analyses in zebrafish (in collaboration with Dr. Stemple, D and Dr. Lin YY, Wellcome Trust Sanger Institute) and by the impaired capacity to modify dystroglycan as evidenced by α-dystroglycan IIH6 staining.

The second strategy to identify novel WWS/MEB genes consists of exome sequencing. Currently, probands from 10 WWS families have been analysed by exome sequencing. In two families they found mutations in already known α-dystroglycanopathy genes: POMT2 and POMGnT1. Diagnostic sequencing and homozygosity mapping prior to whole exome sequencing did not reveal these mutations, prompting their diagnostic service to change the diagnostic screening procedure to one of standard sequencing of all known genes in each patient suspected with a WWS/MEB phenotype. In another patient a mutation in RAB3GAP1 was found. Mutations in this gene are associated with the Warburg-micro syndrome, which shares several symptoms with the Walker–Warburg syndrome.

In the other families they found several homozygous variants that create premature stop codons and missense mutations in possible candidate genes. This group is now testing the segregation of these genes.

In addition to the severe WWS/MEB syndromes, this group has identified families with abnormal dystroglycan O-mannosylation as the cause of isolated dilated cardiomyopathy due to mutations in the dolichol kinase (DOLK) gene. Mutations in this gene were previously identified in a congenital disorder of glycosylation (DK-CDG). Their results show that mutations in DOLK give rise to both N- and O-glycosylation defects, the latter seen by reduced IIH6 staining in cardiac muscle and impaired α-dystroglycan binding to laminin. In addition, the presented data showed that homozygous DOLK1 mutations can give rise to isolated heart defect [23].

Caroline Sewry summarised pathological aspects of the dystroglycanopathies and discussed difficulties in interpretation and the value of examining other tissues, such as skin. The dystroglycanopathies comprise approximately 15% of all congenital muscular dystrophy patients referred to the NCG service at Great Ormond Street Hospital for Children in London. They are thus more common than MDC1A and second only to Ullrich CMD. Although their summary of pathology in the dystroglycanopathies was published in 2009 [24] their overall conclusions have not changed significantly since then. The degree of pathology seen with routine histological stains is variable and does not correlate with phenotype; necrosis may not be obvious. Some samples have profound amounts of adipose and/or fibrotic tissue and may contain few muscle fibres which can influence interpretation, particularly of sarcolemmal-associated proteins. Sampling from a single site at only one time point has to be taken into account in interpretation, and there are limited pathological studies of muscles differentially involved in any neuromuscular disorders.

Inflammation is generally not a pronounced of feature of the dystroglycanopathies, even in cases that are steroid responsive, although some samples may contain some inflammatory cells.

Immunolabeling of the glycosylated epitope of α-dystroglycan with the IIH6 antibody can vary from none, to traces, to a mild reduction, to near normal amounts. There is an apparent correlation between the detected amount on sections and genotype/phenotype in POMT1, POMT2 and POMGNT1 cases but less so in those with FKRP and FKTN mutations. Thus mild cases with FKRP or FKTN mutations with a limb-girdle phenotype and no mental retardation can show little or no labelling with IIH6 whereas severe cases of MDC1C may show more. The possibility of epitope masking has not been excluded. Cases of limb-girdle muscular dystrophy 2I with mutations in FKRP may show a reduction in labelling with IIH6 but this is not consistent. Interpretation of labelling can be difficult if the reduction is mild. The antibody does not always give consistent results between and within sections, Fluorescent labelling is subject to fading and poor preservation of the sample has a marked adverse effect. Reduction of labelling of other proteins, including laminin α2, laminin β1, occasionally laminin γ1, C-terminal dystrophin, β-dystroglycan, and the core of α-dystroglycan, is also variable. Similarly, the number of fibres with foetal/neonatal myosin is variable and is not only a reflection of regeneration. The varying stages of maturity of fibres in a single sample has to be considered.

Assessment of other tissue components, such as blood vessels, show that β-dystroglycan and core α-dystroglycan are detected on large blood vessels but not capillaries, whereas IIH6 is not seen. The expression in human brain is not known although blood vessel labelling of laminin α2 is known to be different.

The secondary changes in immunolabelled proteins seen in muscle can also be observed in sections of skin, and may be more pronounced. Skin can therefore be a useful alternative if an adequate sample from muscle is not available, and skin can also be cultured to provide fibroblasts for in vitro study. Laminin α2 is absent from some skin samples from patients in whom no mutation in the CMD genes has been found. In FKRP patients laminin α2 in skin is also absent and reduced laminin β1 may also be seen. The heterotrimer containing laminin β2 may also have a role as it is absent from chorionic villi in MDC1A but normal in cases with FKRP mutations.

3. Dystroglycan intramolecular association structure and processing

Sabine Strahl described how yeast could be used to dissect biochemical and functional aspects of human O-mannosyltransferases. Protein O-mannosylation is an essential modification among fungi and mammals. It is initiated at the endoplasmic reticulum by a conserved family of dolichyl phosphate-mannose:protein O-mannosyltransferases (PMT/POMTs). Mutations in POMTs cause dystroglycanopathies covering a wide spectrum of clinical severities ranges from WWS to LGMD.

Formation of dimeric PMT complexes is crucial for PMT activity in yeast, mammals and humans, but the direct cause is not known to date. She has developed a photoaffinity probe based on an artificial mannosyl acceptor peptide. Photoaffinity labelling and site directed mutagenesis identified an Asp–Glu motif that is highly conserved among PMTs. Based on her data she proposes that the loop1 regions of dimeric complexes form part of the peptide binding and/or the catalytic site.

Professor Strahl’s group also studied the impact of mutations in POMT1 on the clinical phenotype. Characterisation of the POMT1 substrate dystroglycan and POMT in vitro mannosyltransferase activity showed that the severity of the clinical phenotype of the patients analysed is inversely correlated with POMT activity. Their results suggest that dermal fibroblasts can be applied to facilitate the diagnostic analysis of dystroglycanopathy patients.

Paul T. Martin described new work detailing changes, or lack of changes, in the monosaccharide composition between recombinant α-dystroglycan purified from CHO cells and from CHO cells overexpressing LARGE. These studies confirm the notion that LARGE overexpression can stimulate expression of the bioactive glycans required for laminin binding to α-dystroglycan, but suggest that LARGE does not dramatically alter the overall glycan content, on a molar basis, relative to protein. Other work described a cat muscular dystrophy model where dystroglycan expression, but not glycosylation, is altered, and changes in α-dystroglycan glycosylation and/or expression in paediatric cancers, including rhabdomyosarcoma and neuroblastoma. Here, α-dystroglycan glycosylation and laminin binding are altered in a manner that appears to be similar to changes found in the dystroglycanopathies.

Dr. Martin also discussed attempts by the lab, in collaboration with Professor Federica Montanaro, to immunoprecipitate dystroglycan in the context of the intact dystrophin-associated glycoprotein complex from mouse skeletal muscle. Studies were described that suggest that this can be accomplished such that dystrophin or utrophin, sarcoglycans, syntrophins, and dystrobrevins are all co-precipitated with dystroglycan. A comparative approach was taken to describe differences, and commonalities, between wild type and mdx skeletal muscles, as well to compare these to Galgt2 transgenic skeletal muscles. Galgt2 is an enzyme that when overexpressed in mdx muscles can inhibit the development of muscle pathology [25]. This approach may be developed to help identify new dystroglycan-associated proteins and to identify changes that occur to dystroglycan in various muscle diseases.

Derek Blake presented data on the proteolytic processing of the dystroglycan precursor and the effect of FKRP mutations on the trafficking and folding of the mutant protein. Dystroglycan is processed from a single precursor into α- and β-dystroglycan by proteolytic cleavage at the glycine–serine bond between amino acids 653 and 654 in the human precursor. This site is reminiscent of the auto-catalytic SEA (sea urchin, enterokinase, agrin)-module found in several O-glycosylated proteins including some mucins and agrin. Site-directed mutagenesis at key residues flanking the processing site suggested that N-linked glycosylation is required for efficient cleavage of the dystroglycan. Professor Blake also highlighted a conserved site for O-linked glycosylation on β-dystroglycan that was not required for SEA-module processing.

Professor Blake also presented novel findings on the potential affects of FKRP missense mutations that cause different forms of congenital or limb-girdle muscular dystrophy with or without brain involvement. In excess of 150 different mutations have been described in the human FKRP gene however, in the absence of any assay for the biological activity of FKRP, their effect on protein function is unknown. He also presented data modelling the folding and trafficking of a panel of FKRP mutants in a heterologous cell system. Findings from this study suggest that most FKRP mutations result in conspicuous misfolding that sometimes affects trafficking of the mutant protein between the endoplasmic reticulum and Golgi-apparatus. Dr. Blake suggested that therapeutic strategies aimed at improving the folding of FKRP might have some utility in treating CMD and LGMD caused by mutations in the FKRP gene.

Andrea Brancaccio described ongoing work to generate a novel transgenic mouse line, representing a multiple (tetra) knockin of key amino acids at the dystroglycan α–β interface. Heterozygous animals have been already obtained and appear healthy. Animals will require further more extensive characterisation and further genetic manipulations to remove a Neo cassette placed within the intron of DAG1. Future work will aim to investigate the presence of any possible phenotype emerging in skeletal muscle, heart and brain either in heterozygotes or in homozygotes if not lethal.

Dr. Brancaccio also presented data based on recent assessment by computational biochemistry of a second Ig-like domain within the C-terminal region of α-DG. Based on rational design they inserted a myc-tag at position K500 situated between the mucin-like domain of α-DG and the Ig-like domain. The novel construct does not interfere with DG maturation, processing or trafficking to the membrane as revealed by confocal microscopy. In addition, pull-down of myc-tagged α-DG peptides (both full-length and the truncated N-terminal domain) retrieved from transfected total cell extracts of Ebna293 cells has been achieved indicating appropriate sized bands as determined by SDS–PAGE and Western blotting. This novel construct is likely to be an extremely valuable tool and hopefully will pave the road to improved biochemical analysis and fluorescent imaging of the two DG subunits in a variety of cell based systems.

4. Animal models of the dystroglycanopathies

Kevin Campbell presented the first genetic evidence showing that a primary mutation of dystroglycan causes muscular dystrophy with cognitive impairment. Hypoglycosylation of α-dystroglycan and a consequent reduction of α-dystroglycan binding to ECM proteins are observed in patients with various forms of congenital and limb-girdle muscular dystrophies. Causative mutations for these disorders are found in known or putative glycosyltransferases that participate in the maturation of phosphorylated O-mannosyl glycan. However, despite recent advances in our understanding of the molecular mechanisms underlying secondary dystroglycanopathies, it remains unclear whether dystroglycan is the only target of these enzymes or other substrates contribute to the pathogenesis of these diseases. In a collaborative study between the University of Iowa and Hacettepe University a missense mutation (c. 575C > T, T192M) in the dystroglycan gene of a patient with limb-girdle muscular dystrophy with cognitive impairment was identified. Dr. Campbell showed in an in vitro study that this mutation does not affect dystroglycan expression, but instead causes α-dystroglycan hypoglycosylation and impaired binding between α-dystroglycan and the extracellular matrix protein laminin. A mouse model harbouring this mutation recapitulates immunohistochemical and neuromuscular abnormalities observed in the patient. Furthermore, the affected residue was found to selectively impair modification of dystroglycan post-phosphoryl chains, owing to disruption of the interaction between dystroglycan and like-acetylglucosaminyltransferase (LARGE). Thus, he proposed a novel pathogenic mechanism for muscular dystrophy: disruption of the enzyme-substrate complex that is required to initiate maturation of phosphorylated O-mannosyl glycans on dystroglycan. This work constitutes evidence for supporting the view that dystroglycan is the main – and probably only – protein that is subject to the glycosylation abnormalities that cause muscular dystrophy. Moreover, the knock-in mouse will be a powerful tool as a valid disease model to establish therapeutic strategies for glycosylation-dependent muscular dystrophy and cognitive impairment.

Sue Brown presented an overview of the muscle eye and brain phenotype of her FKRP knockdown (FKRP^KD) mice. This included observations of the inner limiting membrane (ILM) of the eye where defects in laminin α1 and core dystroglycan deposition at birth appeared to be central to breaches in the basement membrane. Interestingly the α-dystroglycan IIH6 epitope appears not be present at the ILM in the newborn mouse eye. Studies of the eye at E12.5 shows however, that IIH6 is highly expressed at the ILM. These observations suggest that the hypoglycosylation of α-dystroglycan may influence retinal development from an early stage. In the brain FKRP^KD mice show a marked disturbance in the deposition of laminin α1, α2, α4 and α5, with transcript levels of laminin α1 and γ1 chain being significantly up-regulated relative to controls. This was associated with a diffuse pattern of laminin deposition below the pial surface that correlated with an abrupt termination of many of the radial glial cells. Defects in the pial basement membrane and radial glial scaffold contributed to the abnormal positioning of both early and late born neurons. Muscle fibre number was reduced in some muscles but not others of the newborn FKRP^KD mice relative to controls. Since these mice die at birth this line has now been crossed to one expressing Cre recombinase in the developing central nervous system. This has resulted in a mouse with a normal lifespan but which develops a muscular dystrophy by 12 weeks of age. This new model is now being used to evaluate the therapeutic value of the up-regulation of LARGE.

Qi Lu discussed how mutations in the fukutin-related protein (FKRP) gene have been associated with a spectrum of muscular dystrophies with increasing severity from mild LGMD2I, congenital muscular dystrophy, to Walker–Warburg syndrome (WWS) and Muscle–Eye–Brain disease (MEB) with defects in eyes and central nervous system. The biochemical feature of the diseases is the hypoglycosylation of α-dystroglycan (α-DG). However, it is not understood how mutations in the single gene can create such a wide range of disease phenotypes and no effective therapy is available. Dr. Lu described how he and colleagues at the McColl-Lockwood Laboratory had developed several mouse models with FKRP mutations detected in patients. These mutations included the common L276I mutation, P448L, and a nonsense mutation. Similar to what is observed in clinics, these mutant mice exhibit wide variation in disease severity, from mild LGMD2I with later onset of muscular dystrophy to severe type of CMD with defects involving the central nerve system and eye development. Many factors, including levels of FKRP expression, genetic background and the specific mutations, were found to be implicated in the disease severity. Using these mouse models, it is possible to study disease mechanism(s) and to test experimental therapies. In collaboration with Dr. Xiao (UNC Chapel Hill, NC), Dr. Lu described how they have now started testing the feasibility of AAV-mediated gene therapy for treating FKRP mutation related diseases. Preliminary data indicate that effective expression of FKRP can be achieved to all skeletal and cardiac muscles with significant improvement in muscle pathology and functions. Expression of LARGE can also achieve partial restoration of functional glycosylation of α-DG in the FKRP mutant mice. These mouse models are essential for developing experimental therapies and for study of disease mechanism(s) of FKRP related muscular dystrophy.

Derek Stemple spoke about recently published zebrafish data, which indicate that fukutin or FKRP knockdown using antisense morpholino oligonucleotides (MOs) leads to a more severe phenotype than is seen with dystroglycan null mutants [26]. Specifically, there is loss of laminin 1 protein expression in FKRP MO injected embryos and activation of unfolded protein response in both fukutin and FKRP MO injected embryos. He then presented two Sanger Institute model organism mutation research projects. The Zebrafish Mutation Project (www.sanger.ac.uk/Projects/D_rerio/zmp/) aims to identify nonsense or disruptive splice-site mutations in every protein-coding gene in zebrafish over the next two years. Currently, disruptive mutations have been identified in 3525 genes, which is 13% of the protein-coding genes. The Mouse Genetics Project (www.sanger.ac.uk/mouseportal/) will produce 1000 lines of mutant mice over the next five years. Currently, 538 lines of mice have been produced of which 364 have undergone extensive primary phenotype analysis, which includes an analysis of plasma creatine kinase (CK) levels. Thus far no lines have shown increased levels of CK, however, Whrn and Ndufs3 mutants show significantly reduced CK levels. Both the zebrafish and mouse projects freely distribute resources and data. There was some discussion about forming a consortium with interest in neuromuscular disease to make best use of these zebrafish and mouse resources available through a coordinated secondary phenotyping effort.

Volker Straub summarised the recent findings from their work on zebrafish models for the dystroglycanopathies. Using a morpholino knockdown approach the Newcastle group focussed on vascular anomalies in zebrafish morphants, as in their cohort of LGMD2I patients there had been one incident of a haemorrhagic stroke and one incident of retinal bleeding. In addition there had also been reports in the literature about vascular anomalies in dystroglycanopathy patients [27]. For their experiments the Newcastle group used transgenic zebrafish expressing enhanced green fluorescent protein in blood vessels that were treated with morpholino antisense oligonucleotides that blocked the expression of fukutin, FKRP and dystroglycan. All morphant fish showed muscle damage and vascular abnormalities at 1 day post fertilisation. The intersegmental vessels of somites failed to reach the dorsal longitudinal anastomosis and in more severe phenotypes retracted further or were in some cases even completely missing. In contrast, the eye vasculature was only distorted in the fukutin and FKRP morphants, but not in dystroglycan morphants or control fish. These findings suggested that fukutin and FKRP have functions that affect ocular development in zebrafish independently of dystroglycan [28]. Overall the fukutin morphant fish had the most severe skeletal muscle and eye phenotype.

Steve Winder described how the phosphorylation of β-dystroglycan on tyrosine was a key regulatory step in determining dystroglycan function. Not only does it modulate the ability of β-dystroglycan to interact with different partners by acting as a molecular switch, but is also a possible part of the mechanism associated with dystroglycan loss from the sarcolemma in Duchenne muscular dystrophy (DMD). Professor Winder described a novel knockin mouse model with a point mutation in a key regulatory tyrosine in dystroglycan. The knockin mouse muscle is phenotypically normal, and when crossed onto mdx there was a significant improvement in histopathological and pathophysiological parameters of muscle function in the mdx:knockin cross as compared to the mdx alone. This suggested that targeting the pathways leading to the loss of dystroglycan from the sarcolemma could provide the basis for a new therapeutic strategy to treat DMD. Such a strategy would be aimed at more traditional drugable targets and utilise more conventional small molecule inhibitors, thus potentially providing a systemic and universal therapy for DMD. Professor Winder also outlined results from their recent work using dystrophic zebrafish as a screen to test for such therapeutic agents and presented as proof of principle their success with the proteasomal inhibitor MG132 and a novel method for analysis of zebrafish muscle integrity based on Fourier analysis [29].

5. Extracellular matrix, signalling, polarity, proteolysis and phosphorylation

Jonathan Jones compared the role of dystroglycan (DG) in muscle cells where it is thought to play a structural role in mediating the interaction between the cytoskeleton and the extracellular matrix, with its functions in epithelial cells, where its roles are less clear. There is emerging data however, that in certain epithelial cells, DG participates in the conversion of mechanical forces in the form of stretch into chemical signals in the cytoplasm [30,31]. Specifically, when cultured primary epithelial cells derived from the alveolar compartment of the lung are subject to stretch the extracellular signal-regulated kinases 1 and 2 (ERK1/2) and the adenosine 5′-monophosphate-activated protein kinase (AMPK) signalling cascades are activated in a DG-dependent fashion [31]. Moreover, AMPK activation in an intact mouse lung subject to non-injurious mechanical ventilation is also DG dependent [30]. Professor Jones described their efforts to characterise cytoplasmic molecules that interact with DG and mediate signalling events in the lung epithelium in vivo and in vitro. To this end, he demonstrated that DG co-localises with a cytoskeleton-crosslinker and signalling scaffold protein called plectin. In cultured lung epithelial cells the localisation was at the substratum-attached surface of the cells, in focal contact-like structures [31]. Moreover, DG co-precipitated with plectin, ERK1/2 and AMPK. Mechanoactivation of both the ERK1/2 and AMPK signalling pathways was significantly attenuated in lung cells deficient in plectin [31]. Thus, he concluded that a DG/plectin complex plays a central role in transmitting mechanical stress from the extracellular matrix (ECM) to the cytoplasm in lung cells. But what is the physiological relevance of DG/plectin mediated mechanosignaling? AMPK activation inhibits production of reactive oxygen species in mechanically stimulated lung cells in vitro and, hence, may protect the cells from the detrimental effects of stretch. On the other hand, activation of the ERK1/2 signalling pathway may be harmful to lung cells. Precise analyses of the consequences of mechanical ventilation in mice in which DG/plectin signalling is perturbed should shed light on this issue.

Alessandro Sgambato presented a comprehensive overview of dystroglycan function in cancers. Tumour development is characterised by disruption of the cell-extracellular matrix (ECM) interactions which allow maintenance of the normal architecture and adhesion. Dystroglycan (DG) is expressed in a wide variety of tissues at the interface between the basement membrane and the cell membrane and plays an important role in cell-ECM interaction by connecting the ECM network to the cytoskeleton and likely activating signalling pathways. Thus, DG was expected to play an important role in the function(s) of epithelial cells and, indeed, work from the Sgambato and other laboratories have demonstrated that it is implicated in several cell functions such as growth, adhesion, basement membrane formation, differentiation, and polarity. Those functions are all relevant in the process of tumour development and loss of a functional DG complex likely contributes to their disruption. Indeed, we and others have demonstrated that DG expression, and mainly α-DG, is reduced or lost in a variety of human cancer cell lines and primary tumours. Loss of a functional DG complex appears to be an early event in human tumourigenesis and in several types of tumours a reduction in the expression levels of DG is most pronounced in high-grade, less differentiated diseases and correlates with clinical outcome of patients. In vitro studies have demonstrated that expression of an exogenous DG cDNA inhibits proliferation and reduces the anchorage-independent growth and tumourigenicity of several types of tumour cells confirming an important role of this molecule in the maintenance of the transformed phenotype. Three major mechanisms have been proposed in cancer cells for disruption of DG function: (i) proteolytic cleavage of the DG complex; (ii) altered glycosylation of α-DG; (iii) overall reduction of DG expression. In conclusion Professor Sgambato concluded that the available studies support a role for DG as a tumour suppressor gene and suggested the addition of “cancer progression” to the list of dystroglycanopathies.

Mutations in genes encoding proteins of the extracellular matrix cause several distinct forms of congenital muscular dystrophies (CMDs). One of the most severe forms of CMD is caused by mutations in the laminin-α2 chain, giving raise to the major laminin isoform in skeletal muscle, called laminin 211 (formerly merosin). This form of CMD, (MDC1A), is characterised by floppy appearance of newborns and a frequent failure to achieve independent ambulation.

Whereas preclinical mouse models of other CMDs often do not reiterate the disease course of the patients, laminin 211-deficient mice show very much the same symptoms as the human disease. In his presentation, Markus Ruegg reported in their preclinical work to develop potential treatment option for MDCIA. Using dy^W/dy^W mice, which is the most often used mouse model for MDC1A, they have performed proof-of-principle studies that show that a miniaturized form of the extracellular matrix protein agrin, by virtue of its binding to the laminins and to α-dystroglycan, was able to ameliorate the phenotype in the dy^W/dy^W mice [32]. The ameliorations include the improvement of the histology and the force of muscle, and, most importantly, a prolongation of life span to approximately 40 weeks (the mice usually die with 8 weeks). Similar beneficial effects were found using an artificial fusion constructs of the N-terminal region of agrin and the α-dystroglycan-binding part of perlecan and the amelioration was still seen when the mini-agrin protein was expressed after onset of the disease phenotype [33]. These experiments thus provide evidence that such linker molecules can largely substitute laminin 211. As the translation of this approach to patients is still difficult because gene therapy is not yet developed sufficiently, pharmacological approaches might be much faster. Such an approach had been reported previously using genetic and pharmacological inhibition of apoptosis [34,35]. However, the efficacy of those treatments appears much lower than the treatment with mini-agrin. Thus, in another approach, the combination of apoptosis inhibition and mini-agrin was tested to eventually detect additive or synergistic effects. Indeed, some aspects of the disease including the loss of muscle force and the capability to regenerate muscle after injury were further ameliorated by the combination. However, no significant further prolongation of life span was observed [36]. These experiments are thus proof-of-concept studies that provide mechanistic insights and unequivocal preclinical evidence for the potential use of those therapies for MDC1A. As the dy^W/dy^W mice are a valid animal model for the human condition, these preclinical data may be of high predictive value and future efforts should be directed to translating those findings into the clinics.

Madeleine Durbeej-Hjalt presented recent work from her group on the molecular functions of laminin α2 and potential therapies for MDC1A. She showed that in previous work transgenically expressed laminin a1 chain significantly ameliorates muscular dystrophy and reduces peripheral neuropathy in the dy^3K/dy^3K mouse model of MDC1A. Dystroglycan and integrins are major laminin receptors. Unlike the laminin a2 chain, the a1 chain binds the receptors by separate domains; laminin globular (LG) domains 4–5 and LG1–3, respectively. To provide insights into the functions of laminin a1LG domains and their division of roles in MDC1A pathogenesis and rescue, a laminin a1 chain that lacks the dystroglycan binding LG4–5 domains was transgenically overexpressed in dy^3K/dy^3K mice. Resultant mice displayed prolonged lifespan and improved health. Diaphragm and heart muscles were spared, whereas limb muscles were dystrophic, indicating that different muscles may have different requirements for LG4–5 domains. Furthermore, the regenerative capacity of the skeletal muscle did not depend on laminin α1LG4–5. However, this domain was crucial for preventing apoptosis in limb muscles, essential for myelination in peripheral nerve and important for basement membrane assembly. These results show that laminin α1LG domains and consequently their receptors have disparate functions in the neuromuscular system. Finally, as muscle atrophy is a significant characteristic of MDC1A, they investigated if increased protein degradation is a feature of laminin α2 chain deficient muscles. Indeed, the ubiquitin–proteasome system, which plays a major role in muscle wasting, appeared overactive in MDC1A. Also, the autophagy-lysosome pathway, which is the other major system involved in degradation of proteins and organelles, was dysregulated in MDC1A. Moreover, separate treatments with a proteasome inhibitor (MG-132) and an autophagy inhibitor (3-methyladenine) significantly improved the dystrophic dy^3K/dy^3K phenotype. These findings indicate that increased activity of the ubiquitin–proteasome system and enhanced autophagic activity are pathogenic in MDC1A and that proteasome and autophagy inhibition, respectively, holds a promising therapeutic potential in the treatment of MDC1A [37].

6. Therapeutic strategies for the dystroglycanopathies

Francesco Muntoni presented ongoing collaborative work on therapeutic approaches in dystroglycanopathies using dystroglycan hyperglycosylation strategy or gene replacement genetic therapies.

Regarding dystroglycan hyperglycosylation, in collaboration with the pharmaceutical company Summitt, the Muntoni group at the Institute of Child Health screened a library of the iminosugar “Seglins” using myoblast H2K cells and identified six compounds which appear to consistently induce increased IIH6 expression in these cells. Experiments using patients cell lines are underway to confirm these findings; definitive positive hits will be characterised further in the FKRP and POMGnT1 deficient mice in collaboration with Sue Brown and Nic Wells at the RVC London.

Professor Muntoni also described ongoing approaches to utilise trans-splicing to correct defects in FKRP. In collaboration with Luis Garcia and Thomas Voit they hope to create an efficient trans-splicing vector that could be used to replace defective FKRP exon 4 which contains the whole coding region of FKRP. Work is in progress in patient cell lines at this point in time.

Jeffrey Chamberlain reported on his group’s efforts aimed at developing gene therapy for the muscular dystrophies, with a focus on DMD and LGMD2I. His lab has obtained efficient gene transfer to striated muscles of mice and dogs using delivery vectors derived from adeno-associated virus serotype 6. Intravenous delivery of the AAV6 vectors is able to achieve uniform gene expression in limb muscles, diaphragm, and heart. Two approaches are underway to extend the results to LGMD2I. First, his group is raising antisera against human FKRP using a peptide antibody and a longer fusion protein as antigens for monoclonal and polyclonal antibody production. They are also preparing a variety of FKRP expression vectors in AAV, and will begin testing them for expression levels, potential toxicity and localisation of the exogenous protein in mouse tissues. Many of these studies will be pursued collaboratively with other attendees of the workshop. Finally, Chamberlain reported on recent work restoring the dystrophin-glycoprotein complex in dystrophin:utrophin deficient mice by expressing a mechanically non-functional form of dystrophin. Those studies support an important role for dystroglycan in muscle tissue independent of the mechanical protection provided by the dystrophin/DGC axis.

Kevin Campbell presented his latest findings showing the like-acetylglucosaminyltransferase-dependent modification of dystroglycan occurs at Thr-317/319 and is required for laminin binding α-dystroglycan is a highly O-glycosylated extracellular matrix receptor that is required for anchoring of the basement membrane to the cell surface. The like-acetylglucosaminyltransferase (LARGE) is a key molecule that binds to the N-terminal domain of α-dystroglycan and attaches ligand-binding moieties to phosphorylated O-mannose on α-dystroglycan. LARGE is particularly interesting since upregulation in patient cells with various glycosyltransferase mutations can rescue functional glycosylation. However, it has been unclear which residues are specifically modified by LARGE, and how these residues determine the high ligand binding affinity. Dr. Campbell showed that the LARGE modification required for laminin- and virus-binding occurs on specific threonine (Thr) residues located at the very N-terminus of the mature form of α-dystroglycan. Deletion and mutation analyses demonstrate that the ligand binding activity of α-dystroglycan is conferred primarily by LARGE modification at Thr-317 and -319, within the highly conserved first 18 amino acids of the mucin-like domain. The importance of these paired residues is confirmed by laminin clustering and arenavirus cell entry assays with full-length dystroglycan. He further demonstrated that the five amino-acid sequence (Thr³¹⁷ProThr³¹⁹ProVal) contains phosphorylated O-glycosylation and when modified by LARGE is sufficient for the ligand binding. These results also suggest that the maximal ligand binding activity of α-dystroglycan is ensured by (1) the proximity of the Thr-317/319 to the ligand molecules including LG domains of ECM proteins and arenaviruses, and (2) the cluster of LARGE-dependent modification. These findings provide insights into the mechanisms that lead to a functional, glycosylated dystroglycan product, and thereby contribute to our understanding of α-dystroglycan-related pathogenesis of muscular dystrophy. Moreover, these findings provide a framework for establishing therapeutic strategies for treating a variety of neuromuscular diseases related to dystroglycan hypoglycosylation.

Volker Straub provided an update about the global FKRP registry (www.fkrp-registry.org/) that was initiated by TREAT-NMD (www.treat-nmd.eu/). The registry collects genetic and clinical data from patients either affected by LGMD2I, MDC1C or any other condition caused by a mutation in the FKRP gene. In the global FKRP patient registry this information is both provided by the patient and the professionals involved in the patient’s care after full consent by the patient. Patients can self-register online and can select a clinician in a drop down menu and authorise him to provide clinical data. Currently more than 150 patients from 17 different countries have signed up to the registry. The global FKRP registry has established close links with the Cure CMD international registry (www.cmdir.org/), which collects information about patients with congenital muscle diseases.

Professor Straub also presented data from a multicentre natural history study in LGMD2I patients with the homozygous founder mutation. The study was conducted between the neuromuscular centres in Newcastle Upon Tyne, London (Queens Square), Copenhagen and Paris (Institut de Myologie). The 12 months study focussed on physical assessments and quantitative muscle magnetic resonance imaging (MRI) using the three point Dixon technique with the main objective to contribute to the “trial readiness” of LGMD2I patients. Interestingly quantitative muscle MRI was much more sensitive to detect an increase in muscle pathology over a 12 months period than any of the tests that were measuring muscle strength or function, including the 6 min walking test. The study suggested that quantitative muscle MRI might be a reliable and objective outcome measure in patients with LGMD2I to monitor therapeutic efficacy in future clinical trials. One drug that could potentially be studied in patient with LGMD2I would be prednisone, as anecdotal reports have suggested a positive effect of glucocorticoids in FKRP-deficiency.

7. Conclusions

In addition to sharing her observations of variabilities in dystroglycanopathy phenotypes and her experiences of caring for her own daughter who has a POMT1 mutation giving rise to LGMD2K, Anne Rutkowski provided an eloquent and fitting summary of many of the topics that had been discussed throughout the workshop as well as highlighting the prospects and challenges for the future. Defining the exact role of α-dystroglycan and the effect of hypoglycosylation in skeletal, cardiac and brain pathogenesis presents a central challenge to scientists and clinicians alike. Traditionally, a disrupted dystroglycan-dystrophin complex has been implicated as the driver of dystrophic pathology through contraction induced sarcolemmal membrane rupture with resultant muscle cell death as evidenced in patients with mutations in dystrophin, dystroglycan and putative glycosyltransferases by elevated serum creatine kinase and muscle uptake of Evans blue dye. In zebrafish, primary dystroglycan mutations parallel findings in laminin α2 mutations of muscle fibre detachment with intact sarcolemmal membranes. Thus, it remains unclear if mutations in either dystroglycan itself or putative glycosyltransferases in patients with an α-dystroglycan related dystrophy (αDG-RD) lead to impaired signalling or structural instability with muscle fibre detachment and/or sarcolemmal membrane rupture.

Prior to developing rational roadmaps for translation in the αDG-RDs, careful dissection to understand the role of α-dystroglycan and its glycosylation relative to that of its binding partners, dystrophin and utrophin, in embryonic development, muscle extracellular matrix and intracellular signalling pathways, intracellular organelle function (endoplasmic reticulum) and satellite cell regeneration are needed. Using proof of concept studies developed in laminin α2 related dystrophy to query relative contribution of fibrosis, the autophagy-proteasome pathway and compensatory laminin and integrin signalling mechanisms can answer targeted queries, implicating overall disease pathology. While skeletal muscle effects are described, the αDG-RDs are a multi-systemic disease requiring a nascent understanding of involvement of neuro-cognitive function, cardiac muscle, the neuromuscular junction, the myotendinous junction, the smooth muscle of the intestinal tract, the endocrine axis and vascular integrity. Case reports highlight the potential role of steroid therapy in patients with an αDG-RD, making both a preclinical trial in the appropriate animal model and clinical trial an urgent priority, providing opportunities to investigate steroid targets and implicated mechanisms while defining treatment effect. Development of sound hypotheses to describe the clinical aspects of disease progression resulting in atrophy, dystrophy, weakness and fatigue will help inform relevant targets, biomarker discovery and timing of interventions.

Participants

Hans van Bokhoven, Nijmegen Medical Centre, The Netherlands.

Derek Blake, Cardiff University, UK.

Andrea Brancaccio, ICRM CNR Rome, Italy.

Sue Brown, Royal Veterinary College, UK.

Kevin Campbell, University of Iowa, USA.

Jeff Chamberlain, University of Seattle, USA.

Madeleine Durbeej-Hjalt, Lund University, Sweden.

Jonathan Jones, Northwestern University, USA.

Qi Lu, Carolinas Healthcare, USA.

Paul Martin, Ohio State University, USA.

Francesco Muntoni, Institute of Child Health, UK.

Markus Ruegg, Biozentrum Basel, Switzerland.

Anne Rutkowski, Cure CMD, USA.

Caroline Sewry, Imperial College, UK.

Alessandro Sgambato, Catholic University Rome, Italy.

Sabine Strahl, University of Heidelberg, Germany.

Derek Stemple, Sanger Center, UK.

Volker Straub, University of Newcastle, UK.

Steve Winder, University of Sheffield, UK.

Daniel Zollinger, ENMC.