Limb-girdle muscular dystrophies (LGMDs) represent a clinically heterogeneous group of genetic diseases characterised by progressive weakness of the pelvic and shoulder girdle muscles. An autosomal dominant form ( LGMD1A ) has been mapped at 5q22.3–31.3, while five genes responsible for the autosomal recessive forms were mapped respectively at: 15q15.1 ( LGMD2A ), 2p12-p16 ( LGMD2B ), 13q12 ( LGMD2C ), 17q12–q21.33 ( LGMD2D ) and 4q12 ( LGMD2E ). Among 17 autosomal recessive (AR) LGMD Brazilian families with at least three affected sibs, we were able to exclude four families (one mild and three severe) from all these five known loci as well as from the dystroglycan and syntrophin genes. Therefore, we have performed a genome-wide search in two of the severely affected families, which are α-sarcoglycan negative. We demonstrate linkage of these two Duchenne muscular dystrophy-like families to 5q33–34, and propose to classify them as LGMD2F. In addition, linkage analysis in the other two genealogies that are α-sarcoglycan positive suggests that there is at least one other gene which causes AR LGMD.
Limb-girdle muscular dystrophies (LGMDs) represent a clinically heterogeneous group of genetic diseases characterised by progressive weakness of the pelvic and shoulder girdle muscles. An autosomal dominant form ( LGMD1A ) has been mapped at 5q22.3–31.3 ( 1 ), but it has been estimated that at least 90% display autosomal recessive inheritance ( 2 ). The dominant disease shows a benign clinical course ( 1 , 3 ), while the autosomal recessive (AR) forms are associated with a wide spectrum of clinical severity and age of onset ( 4–7 ). This last group of diseases (AR LGMD) has been classified mainly as severe (which resemble clinically the Xp21 Duchenne muscular dystrophy, the DMD-like subtype) or mild, depending on the rate of progression and the age of wheelchair confinement. Five genes responsible for the autosomal recessive forms were mapped respectively at: 15q15.1 [LGMD2A ( 8 )], 2p12–p16 [LGMD2B ( 9 )], 13q12 [LGMD2C ( 10 )], 17q12–q21.33 [LGMD2D ( 11 )] and 4q12 [LGMD2E ( 12 , 13 )]. Except for LGMD2B , all other genes have been cloned and their products identified as follows: muscle-specific calpain 3 or CANP3 [LGMD2A ( 14 )], α-sarcoglycan [LGMD2D ( 11 , 15 , 16 )], β-sarcoglycan [ LGMD2E ( 12 , 13 )] and γ-sarcoglycan [ LGMD2C17 )]. CANP3 is a protease while the other three genes encode proteins of the dystrophin-glycoprotein complex (DGC).
The DGC is a large oligomeric complex of sarcolemmal proteins and glycoproteins that acts as a structural link between the cytoskeleton and the extracellular matrix through laminin ( 18–20 ). It is believed to confer stability to the sarcolemma and to protect muscle cells from contraction-induced damage and necrosis ( 18 ). Besides dystrophin (a large F-actin binding intracellular protein), the DGC consists of three main complexes: the dystroglycan complex [α-and β-dystroglycan which are translated from a single mRNA from a gene at 3p21( 21 , 22 )]; the sarcoglycan complex [α-sarcoglycan, β-sarcoglycan, γ-sarcoglycan and possibly the 25 DAG ( 18 , 19 )] and the syntrophin complex [α-syntrophin at 20q12 ( 23 ); β1-syntrophin at 8q23–24 ( 24 ); β2-syntrophin at 16q23 ( 25 , 26 ); AO ( 19 )].
All genes within this complex have been considered strong candidate loci for LGMD due to their close association with dystrophin. Indeed, in addition to the three sarcoglycan genes, it has been shown that mutations in merosin (the α2-subunit of laminin) cause one form of congenital muscular dystrophy ( 27 ). Except for 25 DAG and AO, all others have already been mapped.
Linkage analysis depends on families with multiple affected children, mainly in conditions where there is genetic heterogeneity, which are very rare, in particular in the case of severe diseases. Among 17 Brazilian AR LGMD families with at least three affected sibs (12 with mild and five with severe phenotype), we were able to exclude four families (one mild and three severe) from the LGMD2A, 2B, 2C and 2D loci ( 28 ) and also from therecently cloned LGMD2E gene (unpublished results). Two of these four families (one mild and one with severe phenotype) are positive for α-sarcoglycan, while the two remaining ones with a DMD-like phenotype are negative for α-sarcoglycan, suggesting that they might result from mutations in the same gene.
In the present report, we demonstrate linkage of these two DMD-like families to 5q33–34. In addition, linkage analysis of the other two genealogies suggests that there is at least one other gene which causes AR LGMD.
Four genealogies (LG61, LG6, LG26 and LG43) excluded from the five known mapped/cloned loci for AR LGMD were included in the present linkage analysis. Genealogy LG61, is a caucasian non-consanguineous family, where six of eight sibs, aged 25–41 years old, are affected. All the patients are still ambulant and show a mild clinical course, with onset of the disease in the second decade. Serum creatine kinase (CK) was only slightly increased in all of them (average 1-to 2-times above normal). The muscle biopsy from one of the patients showed a typical myopathic pattern with a positive reaction for dystrophin and α-sarcoglycan. Among the three DMD-like families, all negroids, only one is non-consanguineous (LG6). The clinical data of eight patients from these genealogies who were personally examined showed that age at onset ranged from 2 to 7 years of age, and confinement to a wheelchair occurred between 11 and 16 years old. Among three patients already deceased, the age of death varied from 9 to 19 years old. Serum CK was grossly increased in all of them (10–50 times above normal). Muscle biopsy from at least one patient of each family revealed a typical myopathic pattern and a positive immunofluorescence (IF) labelling for dystrophin in all three families. In relation to the other DGC proteins, one is positive for α-sarcoglycan (LG6) while the other two are totally negative (LG26 and LG43). Based on the dystrophin and α-sarcoglycan pattern and the similarity in clinical course among the affecteds, these two latter families were assembled for linkage analysis, while the first two (the mild LG61 and the severe LG6) were analysed separately.
Linkage analysis was started with markers from four candidate regions (3p21, 8q23–24, 16q23 and 20q12), where the dystroglycan and syntrophin genes have been mapped ( 21–26 ). No evidence of linkage was observed in any of the four families ( Table 1 ). Therefore, we performed a genome-wide search in the two α-sarcoglycan-negative families (LG43 and LG26).
A total of 310 markers spaced at 10–20 cM were tested in these two families. Evidence of linkage was observed with the marker D5S210 for family LG43 (Z max = 1.03; θ = 0.10), although the expected homozygosity due to the consanguinity was not observed with this marker ( Fig. 1 ). The following additional markers were tested in both families: D5S210/D5S209-D5S413/D5S434-3 cM- D5S470/D5S640/D5S636-3 cM- D5S673-1 cM- D5S410-5 cM- D5S820-2 cM- D5S422-5 cM- D5S621-2 cM- D5S415 ( 29 , 30 ). Two-point linkage analysis between these markers (which span >21 cM of the 5q33–34 region) and each of the families are described in Table 2 . Confirmation of linkage was obtained independently in family LG26 (lod score >3) while findings in family LG43 are entirely compatible with linkage to the same markers.
In family LG26, the observed recombinations between the disease gene and markers D5S422/D5S621 in three patients (IV–15, IV–16 and IV–18) positioned the disease gene proximal to D5S422 . In addition, one affected and one unaffected sib (IV–13 and IV–15) are homozygous for D5S210 , suggesting, therefore, that the gene must lie in the interval between D5S210 and D5S422 .
Homozygosity for a 12 cM region spanned by the markers D5S209–D5S820 was observed in all the affecteds from the two families. A common haplotype, flanked by the markers D5S470-D5S820 , that are 9 cM distant apart, was observed in the two families. The alleles for each one of these marker loci segregating with the disease gene are not rare ( Fig. 1 ) and therefore, these results do not support evidence of a strong linkage disequilibrium between the disease gene and any of these particular maikers.
The subsequent analysis of the two α-sarcoglycan-positive families (LG6, LG61) showed negative lod scores, suggesting that the gene mapped at 5q33–34 is not that responsible for the disease in these genealogies (LG6: Z =−0.81 at θ = 0.01; Lg61: Z =−2.74 at θ = 0.01 with the marker D5S636 ).
In the present report, we were able to demonstrate linkage of another subtype of LGMD to 5q33–34, which we propose to be classified as LGMD2F. Considering the recombinants and the region of homozygosity flanked by the markers D5S470 and D5S820 , we suggest that this gene might lie in this 9 cM interval. Interestingly, LGMD1A has been mapped to 5q22.3–31.3, with the markers IL9 and D5S210 ( 1 ), a region near to LGMD2F . However, the marker D5S210 in the two present DMD-like families is out of the critical region for the LGMD2F gene. Therefore, these results exclude the possibility that the same gene is responsible for both conditions.
All patients with LGMD2A and LGMD2B reported so far have a mild phenotype. In both conditions, the DGC complex is normal (Vainzof, unpublished observations), which suggests that CANP3 (the product of LGMD2A ) and the as yet unknown product of the LGMD2B gene do not seem to interfere with this complex. In relation to the LGMD2C and LGMD2E forms, it has been reported that the phenotype can be extremely variable, ranging from DMD-like forms to very mild ones ( 12 , 13 ). Although the 13q region has been associated with a severe phenotype in Tunisian families ( 10 ), the analysis of our LGMD2C families also showed that 13q-linked patients may have a relatively mild clinical course ( 28 ).
Therefore, the fact that the two families with absence of α-sarcoglycan reported here have a severe phenotype does not exclude the possibility that this gene may also lead to a milder clinical course as shown for the other three genes of the sarcoglycan complex. Particularly interesting was our observation of unrelated patients with a variable clinical course, ranging from severe to mild, although carrying the same α-sarcoglycan R77C missense mutation ( 31 ; unpublished results).
Among the proteins from DGC, the transmembrane 25 DAP and AO have not yet been mapped. It has been predicted that mutations in any of the sarcoglycan glycoproteins may lead to deficiency (or absence) in all proteins of this particular complex, as has been already demonstrated in patients with mutations in the β-and α-sarcoglycan genes ( 12 , 13 , 17 ). Therefore, we suggest that mutations in the 25 DAP gene may cause LGMD2F , since the 5q-linked patients reported here are also α-sarcoglycan negative. The cloning of this gene will allow confirmation of this hypothesis.
The two families excluded for all the mapped/cloned genes from the DGC as well as for the present 5q region, are dystrophin and α-sarcoglycan positive. These findings suggest that mutations in at least one other gene are responsible for another form of AR LGMD, which probably does not encode or interfere with proteins of the DGC.
The identification of the genes causing AR LGMD confirms the prediction that this group of diseases is very complex. In addition, it contributes to the elucidation of the process of muscle degeneration and shows that not only structural proteins are involved in this form of muscular dystrophy. The identification of all these genes has been extremely important for enhancing our comprehension of the mechanisms of genetic heterogeneity, that is how mutations in different genes may lead to a similar phenotype.
Materials and Methods
Four families (LG6, LG26, LG43 and LG61) with autosomal recessive limb-girdle patients previously reported ( 28 ) were included in the present analysis. All of them were excluded for the loci 2p, 13q, 15q, 15q and 17q ( 28 ). Negative lod scores with the markers D4S392 and D4S1577 were observed in these families, suggesting that the γ-sarcoglycan gene is not responsible for the phenotype in these genealogies (unpublished data). Adhalin and dystrophin immunofluorescence analysis were performed as previously reported ( 31 , 32 ).
The following markers were used for linkage analysis in LG6, LG26, LG43 and LG61 families (15 affecteds and 32 normals) to test the candidate regions of the dystroglycan and syntrophin genes: D3S1766 for dystroglycan; D8S199 and/or D8S200 for β1-syntro-phin; D16S266 and/or D16S422 for β2-syntrophin; D20S119 and/or D20S110 for α-syntrophin. For the genomic-wide search, DNA samples from family members of LG26 and LG43 (six affecteds and 27 normals) were analysed by polymerase chain reaction (PCR) amplification using microsatellite markers purchased from Research Genetics and Isogen. For the refinement of the candidate area, the markers were selected from Gyapay et al . ( 29 ). Standard protocols were used for PCR reactions; the products were visualised on 6.5% denaturing gels, which were dried and exposed to X-ray films for 2–24 h as previously reported ( 31 ).
Two-point linkage analysis was performed using the computer program MLINK ( 33 ). A gene frequency of 0.001 for the disease gene and equal recombination for both sexes were considered. An equal frequency of the alleles for most of the markers were considered; however, for the markers mapped at the region where linkage was detected, allele frequencies were estimated for 76 chromosomes from normal controls of our population, including caucasians and negroids.
We would like to express our gratitute to Débora F. Rodrigues for her valuable technical help and Mrs Constância Urbani for secretarial assistance. Also, we would like to acknowledge the following persons: Andréa Sertié, Simone Campiotto, Antonia Cerqueira, Dr Rita de Cassia, Dr Ivo Pavanello, Denilce Sumita and Paula Iughetti. We are also extremely grateful to Professor Kay Davies for the basic training in molecular biology techniques (to MRPB) some years ago that was extremely important for the achievement of the present results and to Dr E. Bakker for his constant support and for offering his lab facilities for updating new molecular biology techniques. This research was supported by Fundação de Amparo a Pesquisa do Estado de Sào Paulo (FAPESP), PADCT, CNPq, FINEP and Associação Brasileira de Distrofia Muscular (ABDIM).