Translational Research and Therapeutic Perspectives in Dysferlinopathies

Florian Barthélémy,1,2 Nicolas Wein,1,2 Martin Krahn,1,2,3 Nicolas Lévy,1,2,3 and Marc Bartoli1,2

1University of the Mediterranean, Marseille Medical School, Marseille, France; 2Inserm UMR_S 910 “Medical Genetics and Functional Genomics” Marseille, France; and 3Marseille Children's Hospital, Department of Medical Genetics, Marseille, France

Dysferlinopathies are autosomal recessive disorders caused by mutations in the (DYSF) , encoding the dysfer- lin . DYSF mutations lead to a wide range of muscular phenotypes, with the most prominent being Miyoshi myopathy (MM) and limb girdle muscular dystrophy type 2B (LGMD2B) and the second most common being LGMD. Symptoms generally appear at the end of childhood and, although disease progression is typically slow, walking impairments eventually result. Dys- ferlin is a modular type II transmembrane protein for which numerous binding partners have been identified. Although dysfer- lin function is only partially elucidated, this large protein contains seven calcium sensor C2 domains, shown to play a key role in muscle membrane repair. On the basis of this major function, along with detailed clinical observations, it has been possible to design various therapeutic approaches for dysferlin-deficient patients. Among them, exon-skipping and minigene transfer strategies have been evaluated at the preclinical level and, to date, represent promising approaches for clinical trials. This re- view aims to summarize the pathophysiology of dysferlinopathies and to evaluate the therapeutic potential for treatments cur- rently under development. © 2011 The Feinstein Institute for Medical Research, www.feinsteininstitute.org Online address: http://www.molmed.org doi: 10.2119/molmed.2011.00084

INTRODUCTION myopathy (DACM) (alternatively called phenotype that combines both proximal Dysferlinopathies are autosomal reces- distal myopathy with anterior tibial and distal myopathies of the lower sive muscular dystrophies caused by onset [DMAT]; OMIM number 606768) limbs. However, in every case, by late mutations in the gene encoding dysferlin and proximo-distal phenotypes (PD) stages of disease progression, all limb- (DYSF; Online Mendelian Inheritance in (4,5). LGMD2B is characterized by proxi- girdle muscles are affected. Man [OMIM] gene number 603009, 2p13, mal weakness, initially affecting the The dysferlin gene is located on 2p13 GenBank NM_003494.2) (1,2). Dysfer- muscles of the pelvic and shoulder gir- and encodes several transcripts. The linopathies are rare muscular dystro- dles, with age of onset in the late teens or most common transcript is composed phies, as the number of adult patients early adulthood. For MM, however, the of 55 exons, and alternative splicing whose symptoms meet the full diagnos- muscles initially affected are distal, in events (such as exon 1 for DYSF_v1 tic criteria is estimated between 1/ particular, the gastrocnemius. Age of [GenBank DQ267935], exon 5a [GenBank 100,000 to 1/200,000 (3). Disease-causing onset is at the end of childhood (as with DQ976379] and exon 40a [GenBank mutations in DYSF mainly cause limb LGMD2B), and progression is typically EF015906]) generate transcript diversity. girdle muscular dystrophy type 2B slow. Proximodistal forms of dysfer- Fourteen different splice isoforms, con- (LGMD2B; OMIM number 253601) (1) linopathies are interesting as they dis- taining exon 5a, 17 or 40a, have been de- and distal Miyoshi myopathy (MM; play a range of symptoms, with a hy- scribed for human DYSF (6). Seven of OMIM number 254130) (2). However, pothesized influence of both genetic and these isoforms are transcribed from the other phenotypes associated with muta- environmental factors contributing to the “canonical” DYSF promoter (GenBank tions in DYSF have been identified, in- phenotypic variation. At onset, nearly accession number AF075575), whereas cluding the distal anterior compartment 35% of patients present with a mixed the remaining seven are transcribed from the alternative DYSF_v1 promoter (Gen- Bank accession number DQ267935). Iso- Address correspondence and reprint requests to Marc Bartoli, Faculté de Médecine de form 8 (NM_003494.2) (which excludes Marseille, Université de la Méditerranée, Inserm UMR_S 910 “Génétique Médicale et 5a and 40a, but includes exon 17) is ex- Génomique Fonctionnelle,” Marseille, France. Phone: +33 491 324 906; Fax: +33 491 804 pressed at high levels in skeletal muscle 319; E-mail: [email protected]. and is commonly used to describe the Submitted March 9, 2011; Accepted for publication May 5, 2011; Epub (www.molmed.org) DYSF gene. In addition to these splice ahead of print May 6, 2011. forms, >400 different DYSF sequence

MOL MED 17(9-10)875-882, SEPTEMBER-OCTOBER 2011 | BARTHÉLÉMY ET AL. | 875 DYSFERLINOPATHIES THERAPEUTIC PROSPECTS

variations were reported in the Leiden cisely, it is important to determine ferlin complex (DGC), since in dysferlin null Muscular Dystrophy database (http:// C2 domain topology. Different three- mice, no alteration of the DGC complex www.dmd.nl), including deleterious dimensional shapes adopted by these was evident (29,30). These data clearly in- mutations, as well as nonpathogenic protein domains allow ferlin members to dicate that mechanisms involved in dys- polymorphisms. Most of the reported interact with binding partners; therefore, ferlinopathies must be different from the disease-causing variants are point muta- C2 structural data would be very inform- ones caused by DGC deficiencies. Inter- tions and small insertions/deletions (7), ative (18,19). Within the ferlin family, estingly, a role of dysferlin in the mainte- but large exonic deletions and duplica- dysferlin is particularly well conserved nance of muscle reparation capabilities tions have also been described (8). To among mammals (90% similarity be- has been described. In dysferlin-deficient date, no mutations have been identified tween human and mouse [20,21]) and mice, muscle recovery after injuries is in the alternatively spliced exon 1 of contains seven C2 domains. Multiple measurably slower than in wild-type DYSF_v1, 5a or 40 (9). copies of C2 domains are frequently mice, since recovery in the mutant tissue Translation of isoform 8 results in a found in this particular protein family involves myogenesis rather than mem- 237-kDa transmembrane protein, named (22). Recently, their topology was de- brane repair (31). In fact, dysferlin was dysferlin. This protein belongs to the fer- scribed (C2A, C2B and C2E are type 1 C2 directly implicated in sarcolemmal reseal- lin family, which includes otoferlin, myo- domains, while all the others are type 2), ing of muscle fiber damage resulting ferlin, fer-1-like 5 (FER1L5) and the origi- and it appears that C2 secondary struc- from mechanical stress, two-photon laser nally identified Caenorhabditis elegans tures are important for protein function lesions or muscle elongation in humans Fer-1. Fer-1 mediates calcium-dependent (18,19). The C2 domains of dysferlin are and mice (29,32–35). The involvement of membrane fusion during spermatogene- similar to the well-characterized C2 do- dysferlin in membrane repair is also sup- sis (10). Otoferlin is involved in vesicle main of synaptotagmin VII, a protein ported by the numerous abnormalities in exocytosis and is important for synaptic predominantly involved in vesicle exo - muscles of LGMD2B/MM patients, vesicle–plasma membrane fusion (medi- cytosis in neurons (23,24). Besides C2 do- mainly subsarcolemmal vesicle accumu- ated by Ca2+) at the inner hair-cell ribbon mains, two other motifs are found in this lation (36,37). It was shown in vitro that synapse (11). As such, mutations in protein family: dysferlin sequences (3 of Dysf _/_ cultures have delayed expression otoferlin were shown to cause nonsyn- 6 human ferlin family members contain of myogenin, resulting in defective mus- dromic deafness (deafness, autosomal re- this sequence) and ferlin sequences (car- cle differentiation (38). Finally, two addi- cessive 9 [DFNB9; OMIM number ried by all the members of the ferlin fam- tional functions for dysferlin were de- 601071]) (12). Myoferlin is involved in ily). Both domains are of unknown func- scribed recently: (i) dysferlin is required both myoblast–myoblast fusion (13) and tion (20). Moreover, dysferlin sequences for T-tubule homeostasis in humans (39) myoblast–myofiber fusion during muscle are conserved between orthologs. In ad- and (ii) dysferlin is associated with in- differentiation and maturation, respec- dition, dysferlin also has several nuclear flammation (40,41). tively (14). Myoferlin has also been im- localization sequences (although there is Because of its size and its numerous plicated in vesicle recycling during my- no evidence for its nuclear localization). protein domains, dysferlin likely interacts oblast fusion because of its interaction Finally, several proline, glutamic acid, with numerous partners, some of which with EH-containing domain 2 (EHD2) serine and threonine (PEST) sequences, have already been described. Identified (15). Data published recently suggest involved in protein–protein interaction, interacting are involved in di- that FER1L5 can interact, as previously or in protein ubiquitinylation and degra- verse physiological processes including described for myoferlin, with EHD1 and dation by the proteasome, have been membrane maintenance (annexin 1 and 2, EHD2 to engage myoblast fusion (16). described (25). affixin and caveolin-3), regu- However, FER1L5 has not been associ- The DYSF gene, although expressed lation (calpain-3, AHNAK [desmoyokin]) ated with any human pathology to date. in a variety of tissues including brain, and membrane repair (mitsugumin 53 All members of this gene family en- lung and placenta, is most highly en- [MG53]) (Figure 1). Indeed, it is well doc- code type II transmembrane proteins that riched in skeletal and heart muscles and umented that dysferlin interacts with contain several C2 domains (also called monocytes/ macrophages (1,2,26,27). Dys- caveolin-3, a protein concentrated at the calcium sensors). They are involved in ferlin is expressed in skeletal muscle as intersection between the T-tubule net- many biological processes including early as 5–6 weeks of human embryonic work and the sarcolemma (42) and whose phospholipid interactions (13,14) and development (28), where it is found mutation results in LGMD1C (OMIM Ca2+-dependent interactions. Other pro- mainly in the intracellular network of 607801) (43–45). Caveo lin-3 may serve as teins that contain the C2 domain are typ- muscle fiber (T-tubule and in the plasma a chaperone for dysferlin, directing it to ically involved in active membrane dy- membrane). Furthermore, it was demon- the plasma membrane. Annexins I and II namics and membrane repair (17). To strated that dysferlin is not an important (which bind phospholipids and are in- understand ferlin function more pre- regulator of the -glycoprotein volved in Ca2+ channel formation) (46,47)

876 | BARTHÉLÉMY ET AL. | MOL MED 17(9-10)875-882, SEPTEMBER-OCTOBER 2011 REVIEW ARTICLE

same complex. Although dysferlin and

Extracellular Matrix MG53 (60) are the only proteins known to be involved in muscle repair, their pre-

Laminin cise function in this context remains largely unknown. Ca2+ Dysferlin colocalizes with the dihydro - pyridine receptor in T-tubules, suggest- ing a role for dysferlin in T-tubulogenesis Affixin Caveolin 3 Annexin 1 Integrin (39). The T-tubule system is an extensive internal membrane invagination that MG53 Dysferlin passes transversally from the sarco lemma Calpain 3 AHNAK across a myofibril of striated muscle. T- Ca2+ ch tubules are involved in sarco mere stimu- Tubulin + Na ch lation by triggering the action potential deep in the fiber, allowing the release of calcium ions from the sarco plasmic retic- Sarcoplasmic Reticulum DHPR RyR ulum (61). Because of binding partners such as the annexins and caveolin-3, dys- ferlin could play a role in “invagina- tion/evagination” of the membrane. In- terestingly, T-tubule biogenesis requires both fusion and “invagination/ evagina- tion” of the membrane (62) (see Figure 1). Figure 1. Binding partners of dysferlin. In normal myofibers, dysferlin is anchored by its Recent work has shown that dysferlin’s C-terminal domain to the plasma membrane and membrane network to cytosolic vesi- plasma membrane localization is transi- cles. In these contexts, it associates with different partners according to its localization. tory; membrane anchorage never exceeds Schematized here are the binding partners of dysferlin and their localization within the 3 hours, and the protein’s half-life is only myofiber. ch, Channel. 5 hours. The syntaxin-4-associated endo- cytosis pathway plays a critical role in and affixin (or β-parvin, a focal adhesion recent publication clearly demonstrated regulating this process (63). protein) have each been shown to interact that AHNAK is also found in the costa - with dysferlin in human skeletal muscles meric network and is not present in the FROM SYMPTOMS TO DIAGNOSIS (48). Calpain-3 (mutated in LGMD2A) T-tubule system, leading to the hypothe- Dysferlinopathies, for which onset also interacts with dysferlin (49,50) sis that AHNAK is involved in fiber generally occurs early in the second dec- (OMIM 253600). Calpain-3 is a calcium- stiffness (59). ade of life, are characterized by atrophy dependent protease for which substrates Finally, dysferlin has binding partners and weakness of the gastrocnemius mus- are present in sarcolemme/sarcomeres that play direct and indirect roles in cle and/or the anterior tibial muscles. If and function as a regulator in muscle re- membrane repair, such as MG53 (also a patient presents with these symptoms, modeling (51). Finally, AHNAK has been called TRIM72). MG53 associates with the first step toward diagnosis is to mea- shown to interact with dysferlin. Interest- phosphatidylserine in cell membranes sure serum creatine kinase (which is ele- ingly, along with annexin I & II and af- and acts as a sensor for the entry of ex- vated in most cases of LGMD2B). In fixin, AHNAK has not been directly im- tracellular oxidative molecules. As a re- some cases, this test is supplemented plicated in a human disease (52–54). sult, MG53-containing vesicles are at- with a skeletal muscle computed tomog- AHNAK is a large protein found at the tracted to sites of injury (due to the raphy scan or muscle magnetic imagery enlargeosome (large-scale vesicle) surface presence of an oxidative environment), analysis. The second diagnostic step in- and is involved in cellular processes in- thereby recruiting dysferlin into this volves immunoblot analysis of a muscle cluding cell membrane differentiation, re- niche. However, the interaction between biopsy or blood monocytes. Reduction or pair and signal transduction through its dysferlin and MG53, along with caveo - absence of dysferlin supports the dysfer- protein-protein interactions (52,54–56). lin-3, has only been described in vitro by linopathy diagnosis. It should be men- According to Huang et al. (57), AHNAK coimmunoprecipitation experiments (60). tioned that there are secondary dysfer- and dysferlin colocalize at the sarcolem- There is still no direct evidence of an in linopathies, where low dysferlin levels mal membrane, where dysferlin was vivo interaction between these proteins, by Western blot result from mutations in shown to interact with α-tubulin (58). A but it is likely that they are part of the other , for instance, CAPN3

MOL MED 17(9-10)875-882, SEPTEMBER-OCTOBER 2011 | BARTHÉLÉMY ET AL. | 877 DYSFERLINOPATHIES THERAPEUTIC PROSPECTS

(LGMD2A). This step increases the com- tion (ERAD) system and autophagy (74). from A/J mice), which will be helpful plexity of dysferlinopathy diagnosis. ERAD is used to clear nonaggregated na- when assessing novel therapeutic strate- Histology of affected muscles shows tive or mutant dysferlin, whereas auto - gies. In addition to these cell culture sys- when extensive fiber degeneration- phagy removes aggregated versions of tems, several animal model systems also regeneration has taken place, revealed by the mutated protein. In situations where exist. In particular, two natural mice centronuclear fibers of varying sizes and unfolded-dysferlin degradation is mis- strains (SJL/J and A/J) were identified immune infiltration (predominantly regulated, the pathophysiology could be that have dystrophic phenotypes. The CD8+ cells and macrophages). No B cells explained by the formation of amyloid- SJL/J mouse strain has a splicing muta- are detected, probably because intracel- like aggregates (75). In silico analysis of tion in the dysferlin gene, making it a lular material is released after membrane dysferlin predicted a bi-partite nuclear natural model for LGMD2B and MM lesion (64). localization sequence (NLS) near its (21,78). Genomic analysis identified the N-terminus (www.genscript.com/psort/ mutation as a 141-bp genomic deletion WHAT ARE THE MUTATIONS INVOLVED psort2.html). Whereas endogenous dys- involving a tandem repeat and altering IN DYSFERLINOPATHIES? ferlin is typically not found in the nu- the 3′ splicing site of exon 45, therefore Because the discovery that DYSF dys- cleus, this motif could become relevant leading to the splicing of this 171-bp long function underlies both LGMD2B (1) and for truncated proteins that lack trans- exon. Another strain of mice carrying a MM (2), a wide spectrum of DYSF muta- membrane domains. Indeed, truncated deficient dysferlin was fortuitously dis- tions has been identified, including mis- dysferlins may be directed to inappropri- covered in The Jackson Laboratory. The sense, nonsense, frameshift deletions/ ate cellular compartments, in particular A/J strain presents mild progressive insertions, splicing mutations and large the nucleus, because of the NLS se- muscular dystrophy. Genetic analysis of deletions (1,2,7,8,46,65,66). Although no quence. Unfortunately, this result has this strain demonstrated that an early mutational hotspots were identified not yet been observed because of the transposon (ETn) retrotransposon in- (UMD-DYSF database, Leiden muscular C-terminal specificity of the Hamlet anti- serted itself in the intron 4 of the dysfer- dystrophy pages: http:// www. dmd.nl), body. Every DYSF mutation reported lin gene. However, research involving five founder mutations were reported: to date affects protein expression levels these strains is not easy because of a c.2372C>G (p.Pro791Arg) for a native in skeletal muscle, except for one mild phenotype and the absence of a Canadian population, c.4872_4876delin- (TG573/574AT; pVal67Asp), which af- matched control strain. Thus, genetically sCCCC (p.Glu1624 AspfsX9) for a Libyan fects dysferlin calcium-binding proper- engineered and backcrossed mice Jewish population, c.2875C>T ties (76). Before effective therapeutic ap- (dysf–/–, C57BL/10.SJL-Dysf, B6.129- (p.Arg959Trp) for an Italian population, proaches can be developed, it is critical Dysftm1Kcam/Mmmh, B6.A/J-Dysfprmd) c.5713C>T (p.Arg1905X) for a Spanish to fully characterize DYSF sequence vari- were created (29,78–80). Issues still re- population (region of Sueca) and ations that result in pathologies. main with these dysferlinopathy models, c.2779delG (p.Ala927LeufsX21) for a since none truly recapitulate the human Caucasian Jewish population (67–71). THERAPEUTIC APPROACHES FOR disease. Although not perfect, these Loss of dysferlin is frequently associ- DYSFERLINOPATHIES models will serve as valuable tools to ated with nonsense or frameshift muta- Today, no curative treatment for dys- develop and test novel therapeutic tions (7,66). However, except for rare ferlinopathies exists. In the following approaches. cases (for example, exon 32 deletion section, several validated approaches for causes mild dysferlinopathy [see below, dysferlinopathy therapy will be dis- TRANSFER OF FULL-LENGTH DYSFERLIN 72]), no measurable phenotype to geno- cussed, including synthesis of functional The ultimate goal is to alleviate the type correlation was established. In gen- protein after adeno-associated virus symptoms associated with dysferlin eral, the dysferlin gene has a high level (AAV) vector transfer, gene surgery deficiency. Currently, gene transfer ap- of polymorphisms, once again increasing (exon skipping, trans-splicing) and phar- proaches are the main strategy tested to the complexity for dysferlinopathy mo- macological or immunological preclinical treat recessive diseases such as dysfer- lecular diagnosis. treatments. As always, therapeutic ap- linopathy. Molkentin demonstrated that There are several DYSF mutations that proaches must first be evaluated in cells the restoration of dysferlin only in skeletal lead to mRNA instability and degrada- and/or animal models. Recently, fibro - muscle fibers would be sufficient to res- tion through the nonsense-mediated blasts derived from human patients have cue the dysferlin-deficient mice (81). The mRNA decay (NMD) process (73). At been successfully converted into myo- AAV vector is the most common viral the protein level, Fujita et al. (2007) have blasts (77). Several mouse cell culture vector used for muscle gene therapy (82). described two major pathways for degra- systems have been developed recently For dysferlinopathy, however, there are dation of misfolded dysferlin: the endo- (dysf–/– C2C12 cells, single-cell cloned major technical issues associated with plasmic reticulum–associated degrada- shRNA cell line and immortalized cells using AAV in therapeutic approaches. The

878 | BARTHÉLÉMY ET AL. | MOL MED 17(9-10)875-882, SEPTEMBER-OCTOBER 2011 REVIEW ARTICLE

ITR ITR ITR ITR DNA X-1 X X+1 minidysferlin into dysferlin-deficient Pro Dysf N-Ter Dysf C-Ter Transcripon mice and observed membrane insertion Concatemerizaon = AON ITR ITR pre-mRNA X-1 X X+1 of minidysferlin, which restored mem- ITR 3'OH 5' brane repair. Although this minidysferlin Pro Dysf N-Ter Dysf C-Ter Transcripon Reprogrammed X-1 X+1 mRNA prevented the appearance of severe dys- ATG STOP pre-mRNA Dysf N-Ter Dysf C-Ter ferlinopathies, it was not completely Splicing ATG STOP functional, since no histological improve- mRNA Full Length Dysferlin ments were seen in treated mice (88). AAV-Concatemerizaon Exon-skipping Therefore, our experiments demonstrate that the minidysferlin retains membrane N A B C D E F G C Mutated DYSF N A B C D E F G C repair function, although functions asso- ciated with missing C2 domains are lack- Mini-gene Transplicing ing. For example, the C2A domain inter- acts with phospholipids, AHNAK and α-tubulin (58). C2A (as well as C2B, C2C, ITR ITR DNA X-3 X-2 X-1 X C2D and C2E) is absent from the natural Pro Mini-Dysf Transcripon minidysferlin expressed in this patient. Transcripon pre-mRNA X-3 X-2 X-1 X Y pre-trans-splicing This result certainly explains why restor- ATG STOP molecule mRNA Mini-Dysf Trans-splicing ing minidysferlin is not sufficient to res- Reprogrammed PTC cue all aspects of the dysferlin pheno- mRNA X-3 X-2 X-1 Y type. Nevertheless, these results indicate Figure 2. Therapeutic strategies currently being tested. Dysferlinopathies are an ensemble that both the AAV injection and the level of pathologies, due to mutations in the DYSF gene. Wide spectrums of mutations are re- of expression of a functional minidysfer- sponsible for the pathologies, including missense, nonsense, frameshift, deletion, insertion lin are not toxic. This is important infor- and splicing mutations. Currently, two main types of strategies are envisaged: gene trans- mation for future clinical developments, fer (AAV concatemerization, or minigene) and mRNA surgery (exon-skipping and trans- since we have shown that the overall splicing strategies). Black arrows indicate strategies available for all gene mutations, function of dysferlin can be, at least par- whereas dashed arrows indicate mutation-specific strategies (mutations in exon 32 for tially, carried by a large deleted amino exon skipping, 3′ mutations for trans-splicing). Inverted terminal repeats (ITRs) are the acid sequence protein. This result also adeno-associated vector backbone. pinpoints the necessity to develop ge- encapsidation size is limited to approxi- and his colleagues have also tried to use nomic explorations in patients affected mately 4 kb, the equivalent of three C2 one AAV to encode full-length dysferlin by muscular dystrophies to identify new domains (83). To solve this problem, re- (using the AAV5 serotype) (86). They con- natural truncated mutated proteins. searchers used the concatemerization clude that this strategy allows expression property of AAV (84,85). Two independent of dysferlin at detectable levels; however, MESSENGER SURGERY BY EXON AAV vectors are used: one carrying 5′ dys- dysferlin expression could be in part the SKIPPING ferlin cDNA fused with an intronic se- result of a recombination event between In dysferlinopathies, correlations be- quence that carries a donor splice site and different parts of the transgene carried by tween genotype and phenotype are rare the other containing the 3′ cDNA–bearing AAVs (87). (as previously mentioned). However, the intronic sequence and a splice acceptor Sinnreich group identified a family in site. This approach allows expression, TRANSFER OF DYSFERLIN MINIGENES which two severely affected sisters were after concatemerization and splicing, of a Recently, a minidysferlin was discov- homozygous for a dysferlin null muta- full-length dysferlin transcript (a scheme ered by our team in a patient with a tion (72, Figures 2 and 3). Their mother representing this strategy, as seen in Fig- moderately severe muscular phenotype, was heterozygous for the same null ure 2). This process can be used to reverse and this discovery has opened up new mutation and had a lariat branch point membrane repair deficits associated with therapeutics prospects (88) (see Figure 2). mutation in intron 31 (leading to an in- dysferlin mutant myofibers. Functionally, This minidysferlin, which contains only frame skipping of exon 32) in trans. This this treatment results in improved loco- the last two C2 domains (out of seven) patient presented a mild phenotype, so it motor activity in injected mice, which is a and the transmembrane domain, is ex- is likely that the exon 32 skipped-allele promising result. However, the use of two pressed in muscle biopsies and correctly partially complemented the null muta- AAV vectors and the low levels of expres- targeted to the plasma membrane in this tion. In collaboration with the Garcia sion are issues that must be addressed be- patient. In collaboration with I Richard’s group, we have considered this discovery fore use as a viable therapy (80). Brown team (88), we injected AAV encoding as a “natural” proof of concept for an

MOL MED 17(9-10)875-882, SEPTEMBER-OCTOBER 2011 | BARTHÉLÉMY ET AL. | 879 DYSFERLINOPATHIES THERAPEUTIC PROSPECTS

A * for dysferlinopathy patients treated with Rituximab™ (a human/murine chimeric Allèle 1 : c.3443-33 A>G monoclonal antibody directed against Allèle 2 : c.4872-4876del_InsCCCC CD20-positive B cells) (98). The authors suggest a possible role for B cells in dys- B ferlin pathophysiology. After the tran- scriptomics studies performed in EP Hoff- WT N A B C D E F G C man’s laboratory, demonstrating the upregulation of proteins involved in rho Allele 1 N A B C D E F G C signaling, they started preclinical studies using an inhibitor of a rho kinase (Fa- Allele 2 N A B C D E F C sudil™) (99). This preclinical treatment demonstrated some benefits: notably, a C2 domain DYSF domain Deletion Exon 32 lower level of inflammatory cells in the Fer domain TM domain Premature Stop Codon muscles and an increase in body weight, but no impact in another physiological Figure 3. A natural proof of exon skipping. Depicted is a family studied by Sinnreich et al. process such as muscle fiber degenera- (72), in which the index case (identified by an asterisk) carries two mutant alleles. The fam- ily tree (A) and the nature of the two mutant alleles (B) are shown. tion/regeneration or in muscle force (100). Nevertheless, this study is a starting point for drug-testing trials in dysferlin- exon-skipping therapeutic approach for PHARMACOLOGICAL AND deficient mice, by a standardized mea- dysferlinopathies. The use of an exon- IMMUNOLOGICAL APPROACHES surement of dysferlin functions that will skipping strategy was first applied in Pharmacological or immunological ap- be useful for high-throughput screening. Duchenne muscular dystrophies (89), and proaches have also been envisaged to clinical trials using antisense oligonu- treat dysferlinopathies. Recently, PTC124® CONCLUSION cleotides (AONs) are underway (90). The (also known as Ataluren™) was shown to It is essential to characterize dysferlin results obtained in preclinical trials are read through nonsense mutations. Wang at the molecular level to functionally un- promising and pave the way for similar et al. (94) showed that treating myotubes derstand this sarcolemmal protein in approaches for dysferlinopathies (91). In from a patient with a premature stop both normal and pathological conditions. particular, using an exon- skipping strat- codon allowed the expression of a suffi- This knowledge is absolutely required if egy to treat DYSF mutations lying in cient functional dysferlin to rescue my- adapted therapeutic strategies are to be exon 32 is promising. According to the otube membrane blebbing. However, in developed for dysferlinopathies (and UMD database, approximately 4% of dys- dysferlinopathies, less than one-third of possibly other forms of muscular dystro- ferlin patients have disease-causing mu- mutations are nonsense mutations. phies), where enhancement of sarcolem- tations in this exon. AON directed In 2007, the use of Dantrolen™ as a po- mal resealing could be of benefit. Re- against splicing regulatory elements of tential treatment for dysferlinopathies searchers are still working to define exon 32 were tested on myo tubes transd- was described. CPK levels were reduced functional roles for each C2 domain and ifferentiated from patient fibro blasts pre- in response to Dantrolen (95); however, to accurately describe dysferlin activity viously infected by a lentivirus encoding no amelioration of the muscle phenotype in a variety of physiological processes, MyoD. The efficiency of exon skipping has been seen to date. In 2001, the pres- such as T-tubule homeostasis and mem- induced by these AONs was validated by ence of a complement deposit in non- brane repair. As molecular diagnostic reverse –polymerase chain necrotic muscle fibers deficient for dysfer- technologies improve, a better under- reaction. Indeed, a fragment correspon- lin (96) was observed, and it was standing of these diseases will emerge, ding to a sequence deleted of exon 32 hypothesized that the complement con- enabling the development of appropriate was observed in samples from AON- tributed to the deleterious effects of the therapeutic strategies on the basis of clin- treated cells (92,93). Even if this technol- dysferlin absence. A clinical trial was ical observations. ogy proves promising, it will be impor- launched to try to inhibit accumulation of tant to improve efficiency of AON this deposit by the administration of in- ACKNOWLEDGMENTS delivery and to determine the efficiency travascular immunoglobulin (IV-IG); We warmly acknowledge the constant of this “quasi-dysferlin” before switching however, the initially encouraging results support of the Association Française con- to clinical trials. The different approaches were not pursued (reviewed in 97). Re- tre les Myopathies (AFM) and the Jain described in this section have been cently, promising results were published Foundation. F Barthélémy and N Wein schematically represented in Figure 3. regarding the increase of muscle strength received PhD fellowship grants from

880 | BARTHÉLÉMY ET AL. | MOL MED 17(9-10)875-882, SEPTEMBER-OCTOBER 2011 REVIEW ARTICLE

AFM and the Fondation pour la 15. Doherty KR, et al. (2008) The endocytic recycling 34. Meldolesi J. (2003) Surface wound healing: a Recherche Médicale (FRM), respectively. protein EHD2 interacts with myoferlin to regu- new, general function of eukaryotic cells. J. Cell. late myoblast fusion. J. Biol. Chem. 283:20252–60. Mol. Med. 7:197–203. 16. Demonbreun AR, et al. (2010) Myoferlin is re- 35. von der Hagen M, et al. (2005) The differential DISCLOSURE quired for insulin-like growth factor response profiles of proximal and distal The authors declare that they have no and muscle growth. FASEB J. 24:1284–95. muscle groups are altered in pre-pathological competing interests as defined by Molecu- 17. Zhang D, Aravind L. (2010) Identification of dysferlin-deficient mice. Neuromuscul. Disord. lar Medicine, or other interests that might novel families and classification of the C2 do- 15:863–77. be perceived to influence the results and main superfamily elucidate the origin and evolu- 36. Cenacchi G, Fanin M, De Giorgi LB, Angelini C. tion of membrane targeting activities in eukary- (2005) Ultrastructural changes in dysferlinopathy discussion reported in this paper. otes. Gene. 469:18–30. support defective membrane repair mechanism. 18. Nalefski EA, Falke JJ. (1996) The C2 domain cal- J. Clin. Pathol. 58:190–5. REFERENCES cium-binding motif: structural and functional di- 37. Piccolo F, Moore SA, Ford GC, Campbell KP. 1. Bashir R, et al. (1998) A gene related to Caeno - versity. Protein Sci. 5:2375–90. (2000) Intracellular accumulation and reduced rhabditis elegans spermatogenesis factor fer-1 is 19. Therrien C, Di Fulvio S, Pickles S, Sinnreich M. sarcolemmal expression of dysferlin in limb: gir- mutated in limb-girdle muscular dystrophy type (2009) Characterization of lipid binding specifici- dle muscular dystrophies. Ann. Neurol. 48:902–12. 2B. Nat. Genet. 20:37–42. ties of dysferlin C2 domains reveals novel inter- 38. de Luna N, et al. (2006) Absence of dysferlin al- 2. Liu J, et al. (1998) Dysferlin, a novel skeletal mus- actions with phosphoinositides. Biochemistry. ters myogenin expression and delays human cle gene, is mutated in Miyoshi myopathy and 48:2377–84. muscle differentiation “in vitro.” J. Biol. Chem. limb girdle muscular dystrophy. Nat. Genet. 20. Lek A, Lek M, North KN, Cooper ST. (2010) Phy- 281:17092–8. 20:31–6. logenetic analysis of ferlin genes reveals ancient 39. Klinge L, et al. (2010) Dysferlin associates with 3. Moore SA, et al. (2006) Limb-girdle muscular eukaryotic origins. BMC Evol. Biol. 10:231. the developing T-tubule system in rodent and dystrophy in the United States. J. Neuropathol. 21. Vafiadaki E, et al. (2001) Cloning of the mouse human skeletal muscle. Muscle Nerve. 41:166–73. Exp. Neurol. 65:995–1003. dysferlin gene and genomic characterization of 40. Roche JA, et al. (2010) Extensive mononuclear in- 4. Nguyen K, et al. (2007) Phenotypic study in 40 the SJL-Dysf mutation. Neuroreport. 12:625–9. filtration and myogenesis characterize recovery patients with dysferlin gene mutations: high fre- 22. Cho W, Stahelin RV. (2006) Membrane binding of dysferlin-null skeletal muscle from contrac- quency of atypical phenotypes. Arch. Neurol. and subcellular targeting of C2 domains. Biochim. tion-induced injuries. Am. J. Physiol. Cell Physiol. 64:1176–82. Biophys. Acta. 1761:838–49. 298:C298–312. 5. Ueyama H, et al. (2002) Clinical heterogeneity in 23. Shao X, Davletov BA, Sutton RB, Sudhof TC, 41. De Luna N, et al. (2010) Role of thrombospondin dysferlinopathy. Intern. Med. 41:532–6. Rizo J. (1996) Bipartite Ca2+-binding motif in C2 1 in macrophage inflammation in dysferlin myo- 6. Pramono ZA, et al. (2009) Identification and char- domains of synaptotagmin and protein kinase C. pathy. J. Neuropathol. Exp. Neurol. 69:643–53. acterisation of human dysferlin transcript vari- Science. 273:248–51. 42. Hernandez-Deviez DJ, et al. (2006) Aberrant dys- ants: implications for dysferlin mutational 24. Sutton RB, Davletov BA, Berghuis AM, Sudhof ferlin trafficking in cells lacking caveolin or ex- screening and isoforms. Hum. Genet. 125:413–20. TC, Sprang SR. (1995) Structure of the first C2 pressing dystrophy mutants of caveolin-3. Hum. domain of synaptotagmin I: a novel Ca2+/ Mol. Genet. 15:129–42. 7. Krahn M, et al. (2009) Analysis of the DYSF mu- phospholipid-binding fold. Cell. 80:929–38. 43. Capanni C, et al. (2003) Dysferlin in a hyperCK- tational spectrum in a large cohort of patients. 25. Rechsteiner M, Rogers SW. (1996) PEST se- aemic patient with caveolin 3 mutation and in Hum. Mutat. 30:E345–75. quences and regulation by proteolysis. Trends C2C12 cells after p38 MAP kinase inhibition. Exp. 8. Krahn M, et al. (2009) Identification of different ge- Biochem. Sci. 21:267–71. Mol. Med. 35:538–44. nomic deletions and one duplication in the dysfer- 26. Ho M, et al. (2002) A novel, blood-based diagnos- 44. Matsuda C, et al. (2001) The sarcolemmal pro- lin gene using multiplex ligation-dependent probe tic assay for limb girdle muscular dystrophy 2B teins dysferlin and caveolin-3 interact in skeletal amplification and genomic quantitative PCR. and Miyoshi myopathy. Ann. Neurol. 51:129–33. muscle. Hum. Mol. Genet. 10:1761–6. Genet. Test. Mol. Biomarkers. 13:439–42. 27. Vandre DD, et al. (2007) Dysferlin is expressed in 45. Walter MC, et al. (2003) Variable reduction of 9. Krahn M, Labelle V, Borges A, Bartoli M, Levy N. human placenta but does not associate with caveolin-3 in patients with LGMD2B/MM. (2010) Exclusion of mutations in the dysferlin al- caveolin. Biol. Reprod. 77:533–42. J. Neurol. 250:1431–8. ternative exons 1 of DYSF-v1, 5a, and 40a in a co- 28. Anderson LV, et al. (1999) Dysferlin is a plasma 46. Cagliani R, et al. (2005) Mutation finding in pa- hort of 26 patients. Genet. Test. Mol. Biomarkers. and is expressed early in tients with dysferlin deficiency and role of the 14:153–4. human development. Hum. Mol. Genet. 8:855–61. dysferlin interacting proteins annexin A1 and A2 10. Washington NL, Ward S. (2006) FER-1 regulates 29. Bansal D, et al. (2003) Defective membrane repair in muscular dystrophies. Hum. Mutat. 26:283. Ca2+-mediated membrane fusion during C. ele- in dysferlin-deficient muscular dystrophy. Na- 47. Lennon NJ, et al. (2003) Dysferlin interacts with gans spermatogenesis. J. Cell Sci. 119:2552–62. ture. 423:168–72. annexins A1 and A2 and mediates sarcolemmal 11. Roux I, et al. (2006) Otoferlin, defective in a 30. Matsuda C, et al. (1999) Dysferlin is a surface wound-healing. J. Biol. Chem. 278:50466–73. human deafness form, is essential for exocytosis membrane-associated protein that is absent in 48. Matsuda C, et al. (2005) Dysferlin interacts with at the auditory ribbon synapse. Cell. 127:277–89. Miyoshi myopathy. Neurology. 53:1119–22. affixin (beta-parvin) at the sarcolemma. J. Neu- 12. Yasunaga S, et al. (1999) A mutation in OTOF, en- 31. Roche JA, Lovering RM, Bloch RJ. (2008) Im- ropathol. Exp. Neurol. 64:334–40. coding otoferlin, a FER-1-like protein, causes paired recovery of dysferlin-null skeletal muscle 49. Anderson LV, et al. (2000) Secondary reduction in DFNB9, a nonsyndromic form of deafness. Nat. after contraction-induced injury in vivo. Neuro - calpain 3 expression in patients with limb girdle Genet. 21:363–9. report. 19:1579–84. muscular dystrophy type 2B and Miyoshi myo- 13. Davis DB, Doherty KR, Delmonte AJ, McNally 32. Bansal D, Campbell KP. (2004) Dysferlin and the pathy (primary dysferlinopathies). Neuromuscul. EM. (2002) Calcium-sensitive phospholipid bind- plasma membrane repair in muscular dystrophy. Disord. 10:553–9. ing properties of normal and mutant ferlin C2 Trends Cell Biol. 14:206–13. 50. Huang Y, et al. (2008) Calpain 3 is a modulator of domains. J. Biol. Chem. 277:22883–8. 33. Doherty KR, McNally EM. (2003) Repairing the the dysferlin protein complex in skeletal muscle. 14. Doherty KR, et al. (2005) Normal myoblast fusion tears: dysferlin in muscle membrane repair. Hum. Mol. Genet. 17:1855–66. requires myoferlin. Development. 132:5565–75. Trends Mol. Med. 9:327–30. 51. Duguez S, Bartoli M, Richard I. (2006) Calpain 3:

MOL MED 17(9-10)875-882, SEPTEMBER-OCTOBER 2011 | BARTHÉLÉMY ET AL. | 881 DYSFERLINOPATHIES THERAPEUTIC PROSPECTS

a key regulator of the sarcomere? FEBS J. 273: dysferlin deficiency in Libyan Jews: clinical and Trans-splicing vectors expand the utility of 3427–36. genetic features. Brain. 123:1229–37. adeno-associated virus for gene therapy. Proc. 52. Borgonovo B, et al. (2002) Regulated exocytosis: a 69. Cagliani R, et al. (2003) Molecular analysis of Natl. Acad. Sci. U. S. A. 97:6716–21. novel, widely expressed system. Nat. Cell. Biol. LGMD-2B and MM patients: identification of 85. Xu Z, et al. (2004) Trans-splicing adeno- 4:955–62. novel DYSF mutations and possible founder ef- associated viral vector-mediated gene therapy 53. Haase H, et al. (1999) Signaling from beta- fect in the Italian population. Neuromuscul. Dis- is limited by the accumulation of spliced adrenoceptor to L-type calcium channel: identifi- ord. 13:788–95. mRNA but not by dual vector coinfection effi- cation of a novel cardiac protein kinase A target 70. Vilchez JJ, et al. (2005) Identification of a novel ciency. Hum. Gene Ther. 15:896–905. possessing similarities to AHNAK. FASEB J. founder mutation in the DYSF gene causing clin- 86. Allocca M, et al. (2008) Serotype-dependent 13:2161–72. ical variability in the Spanish population. Arch. packaging of large genes in adeno-associated 54. Hohaus A, et al. (2002) The carboxyl-terminal re- Neurol. 62:1256–9. viral vectors results in effective gene delivery in gion of ahnak provides a link between cardiac 71. Leshinsky-Silver E, et al. (2007) Dysferlinopathy in mice. J. Clin. Invest. 118:1955–64. L-type Ca2+ channels and the actin-based cyto - the Jews of the Caucasus: a frequent mutation in 87. Hirsch ML, Agbandje-McKenna M, Samulski skeleton. FASEB J. 16:1205–16. the dysferlin gene. Neuromuscul Disord 17:950–4. RJ. (2010) Little vector, big gene transduction: 55. Benaud C, et al. (2004) AHNAK interaction with 72. Sinnreich M, Therrien C, Karpati G. (2006) Lariat fragmented genome reassembly of adeno- the annexin 2/S100A10 complex regulates cell branch point mutation in the dysferlin gene with associated virus. Mol. Ther. 18:6–8. membrane cytoarchitecture. J. Cell Biol. 164:133–144. mild limb-girdle muscular dystrophy. Neurology. 88. Krahn M, et al. (2010) A naturally occurring 56. Gentil BJ, et al. (2001) The giant protein AHNAK 66:1114–6. human minidysferlin protein repairs sarcolem- is a specific target for the calcium- and zinc- 73. Wenzel K, et al. (2006) Novel sequence variants in mal lesions in a mouse model of dysferlinopa- binding S100B protein: potential implications for dysferlin-deficient muscular dystrophy leading thy. Sci. Transl. Med. 2:50ra69. Ca2+ homeostasis regulation by S100B. J. Biol. to mRNA decay and possible C2-domain mis- 89. Brun C, et al. (2003) U7 snRNAs induce correc- Chem. 276:23253–61. folding. Hum. Mutat. 27:599–600. tion of mutated dystrophin pre-mRNA by exon 57. Huang Y, et al. (2007) AHNAK, a novel compo- 74. Fujita E, et al. (2007) Two endoplasmic reticulum- skipping. Cell. Mol. Life Sci. 60:557–66. nent of the dysferlin protein complex, redistrib- associated degradation (ERAD) systems for the 90. van Deutekom JC, et al. (2007) Local dystrophin utes to the cytoplasm with dysferlin during novel variant of the mutant dysferlin: ubiquitin/ restoration with antisense oligonucleotide skeletal muscle regeneration. FASEB J. 21:732–42. proteasome ERAD(I) and autophagy/lysosome PRO051. N. Engl. J. Med. 357:2677–86. 58. Azakir BA, Di Fulvio S, Therrien C, Sinnreich M. ERAD(II). Hum. Mol. Genet. 16:618–29. 91. Kinali M, et al. (2009) Local restoration of dys- (2010) Dysferlin interacts with tubulin and mi- 75. Spuler S, et al. (2008) Dysferlin-deficient muscu- trophin expression with the morpholino oligomer crotubules in mouse skeletal muscle. PLoS One. lar dystrophy features amyloidosis. Ann. Neurol. AVI-4658 in Duchenne muscular dystrophy: a 5:e10122. 63:323–8. single-blind, placebo-controlled, dose-escalation, 59. Marg A, Haase H, Neumann T, Kouno M, 76. Therrien C, Dodig D, Karpati G, Sinnreich M. proof-of-concept study. Lancet Neurol. 8:918–28. Morano I. (2010) AHNAK1 and AHNAK2 are (2006) Mutation impact on dysferlin inferred 92. Levy N, et al. (2010) Therapeutic exon ‘switch- costameric proteins: AHNAK1 affects transverse from database analysis and computer-based ing’ for dysferlinopathies? Eur. J. Hum. Genet. skeletal muscle fiber stiffness. Biochem. Biophys. structural predictions. J. Neurol. Sci. 250:71–8. 18:969– 70; author reply 971. Res. Commun. 401:143–8. 77. Chaouch S, et al. (2009) Immortalized skin fibro - 93. Wein N, et al. (2010) Efficient bypass of muta- 60. Cai C, et al. (2009) Membrane repair defects in blasts expressing conditional MyoD as a renew- tions in dysferlin deficient patient cells by anti- muscular dystrophy are linked to altered interac- able and reliable source of converted human sense-induced exon skipping. Hum. Mutat. tion between MG53, caveolin-3, and dysferlin. muscle cells to assess therapeutic strategies for 31:136–42. J. Biol. Chem. 284:15894–902. muscular dystrophies: validation of an exon- 94. Wang B, et al. (2010) Membrane blebbing as an 61. Franzini-Armstrong C. (1991) Simultaneous mat- skipping approach to restore dystrophin in assessment of functional rescue of dysferlin- uration of transverse tubules and sarcoplasmic Duchenne muscular dystrophy cells. Hum. Gene deficient human myotubes via nonsense sup- reticulum during muscle differentiation in the Ther. 20:784–90. pression. J. Appl. Physiol. 109:901–5. mouse. Dev. Biol. 146:353–63. 78. Bittner RE, et al. (1999) Dysferlin deletion in SJL 95. Hattori H, et al. (2007) A novel compound het- 62. Klinge L, et al. (2007) From T-tubule to sarco - mice (SJL-Dysf) defines a natural model for limb erozygous dysferlin mutation in Miyoshi lemma: damage-induced dysferlin translocation girdle muscular dystrophy 2B. Nat. Genet. 23:141–2. myopathy siblings responding to dantrolene. in early myogenesis. FASEB J. 21:1768–76. 79. Ho M, et al. (2004) Disruption of muscle mem- Eur. J. Neurol. 14:1288–91. 63. Evesson FJ, et al. (2010) Reduced plasma mem- brane and phenotype divergence in two novel 96. Selcen D, Stilling G, Engel AG. (2001) The earli- brane expression of dysferlin mutants is attrib- mouse models of dysferlin deficiency. Hum. Mol. est pathologic alterations in dysferlinopathy. uted to accelerated endocytosis via a syntaxin-4- Genet. 13:1999–2010. Neurology. 56:1472–81. associated pathway. J Biol. Chem. 285:28529–39. 80. Lostal W, et al. (2010) Efficient recovery of dysfer- 97. Albrecht DE, et al. (2009) 3rd Annual Dysferlin 64. Gallardo E, et al. (2001) Inflammation in dysferlin lin deficiency by dual adeno-associated vector- Conference 2–5 June 2009, Boston, Massachu- myopathy: immunohistochemical characteriza- mediated gene transfer. Hum. Mol. Genet. setts, USA. Neuromuscul. Disord. 19:867–73. tion of 13 patients. Neurology. 57:2136–8. 19:1897–907. 98. Lerario A, et al. (2010) Effects of rituximab in 65. Aoki M, et al. (2001) Genomic organization of the 81. Millay DP, et al. (2009) Genetic manipulation of two patients with dysferlin-deficient muscular dysferlin gene and novel mutations in Miyoshi dysferlin expression in skeletal muscle: novel in- dystrophy. BMC Musculoskelet. Disord. 11:157. myopathy. Neurology. 57:271–8. sights into muscular dystrophy. Am. J. Pathol. 99. Kesari A, et al. (2008) Dysferlin deficiency 66. Nguyen K, et al. (2005) Dysferlin mutations in 175:1817–23. shows compensatory induction of LGMD2B, Miyoshi myopathy, and atypical dys- 82. Daniele N, Richard I, Bartoli M. (2007) Ins and Rab27A/Slp2a that may contribute to inflam- ferlinopathies. Hum. Mutat. 26:165. outs of therapy in limb girdle muscular dystro- matory onset. Am. J. Pathol. 173:1476–87. 67. Weiler T, et al. (1999) Identical mutation in pa- phies. Int. J. Biochem. Cell. Biol. 39:1608–24. 100. Rayavarapu S, et al. (2010) Characterization of tients with limb girdle muscular dystrophy type 83. Bueler H. (1999) Adeno-associated viral vectors dysferlin deficient SJL/J mice to assess preclini- 2B or Miyoshi myopathy suggests a role for for gene transfer and gene therapy. Biol. Chem. cal drug efficacy: fasudil exacerbates muscle modifier gene(s). Hum. Mol. Genet. 8:871–7. 380:613–22. disease phenotype. PLoS One. 5:e12981. 68. Argov Z, et al. (2000) Muscular dystrophy due to 84. Yan Z, Zhang Y, Duan D, Engelhardt JF. (2000)

882 | BARTHÉLÉMY ET AL. | MOL MED 17(9-10)875-882, SEPTEMBER-OCTOBER 2011