Limb girdle muscular dystrophy type 2D (LGMD2D, OMIM600119) is a genetic progressive myopathy that is caused by mutations in the human α-sarcoglycan gene (SGCA). Here, we have introduced in mice the most prevalent LGMD2D mutation, R77C. It should be noted that the natural murine residue at this position is a histidine. The model is, therefore, referred as SgcaH77C/H77C. Unexpectedly, we observed an absence of LGMD2D-like phenotype at histological or physiological level. Using a heterologous cellular model of the sarcoglycan complex formation, we showed that the R77C allele encodes a protein that fails to be delivered to its proper cellular localization in the plasma membrane, and consequently to the disappearance of a positively charged residue. Subsequently, we transferred an AAV vector coding for the human R77C protein in the muscle of Sgca-null mice and were able to pharmacologically rescue the R77C protein from endoplasmic reticulum-retention using proteasome or mannosidase I inhibitors. This suggests a therapeutic approach for LGMD2D patients carrying mutations that impair α-sarcoglycan trafficking.
α-Sarcoglycan is a 50 kDa sarcolemmal protein that forms, together with β, δ and γ sarcoglycans, a subcomplex participating in the dystrophin–glycoprotein complex (DGC), an important link between the actin cytoskeleton, the sarcolemma and the extracellular matrix (1). Genetic defects in any of the sarcoglycans lead to a recessive limb-girdle muscular dystrophy (LGMD2), respectively, LGMD2D, E, C and F for α, β, γ and δ-sarcoglycans (2). Immunohistochemistry studies indicated that dysfunction of one of the sarcoglycan destabilizes the whole sarcoglycan complex, leading to a partial or complete disappearance of the other sarcoglycans at the membrane (3–5). This particularity was shown to impair the structural integrity of the muscle fiber during contraction, explaining, at least partly, the pathophysiological mechanism by which muscle cells degenerate (6).
Patients with α-sarcoglycan deficiency (LGMD2D, OMIM600119) present a clinical phenotype with proximal muscle involvement and a highly elevated serum creatine kinase level (7,8). At the histological level, muscles show marked regeneration and degeneration pattern, inflammatory infiltrates, fibrosis and variation in fiber size. Age of onset is variable, although, in most cases, first symptoms appear around the 20s. The progression of muscle weakness is also variable and is often correlated to the level of residual α-sarcoglycan expression (9,10). No treatment is currently available for this devastating disease. The most frequently reported mutation in the α-sarcoglycan gene is a C to T transition at position 229 in exon 3, resulting in the substitution of an arginine in position 77 by a cysteine (R77C). This mutation accounts for up to one-third of all mutations in European populations except in Finland where it is found in every LGMD2D patients often in both alleles (10,11).
Here, we have introduced this mutation in the murine genome. Unexpectedly, no apparent LGMD2D-like phenotype was induced. As a first step to explain this difference in outcome between the human and murine species, we showed that the R77C protein fails to be delivered to its proper sarcolemmal localization, and consequently to the disappearance of a positively charged residue at the 77 position. Most importantly, following expression of the human R77C protein in Sgca-null mice, we were able to pharmacologically rescue the R77C protein from retention in the secretory pathway by blocking the protein quality control using proteasome or mannosidase I inhibitors in living mice without evidence of toxicity of the treatment.
A cysteine at position 77 in Sgca induces no phenotype in mice
To explore the pathophysiological mechanism of the R77C mutation, we introduced in the murine genome a ‘knock-in’ mutation that encodes a cysteine residue at the 77 position of the α-sarcoglycan protein (Fig. 1; Supplementary Material, Fig. S1). It should be noted that the natural murine residue at this position is a histidine in place of an arginine in the human protein.
Obtaining homozygous mice, denoted as SgcaH77C/H77C mice, was confirmed by PCR and sequencing (Supplementary Material, Fig. S1). RT–qPCR and western blot analyses performed on skeletal muscle did not detect any difference between WT and SgcaH77C/H77C regarding the relative amount of α-sarcoglycan mRNA or protein (Fig. 1). To investigate the subcellular localization of α-sarcoglycan in SgcaH77C/H77C mice, we performed immunohistochemistry analysis using an α-sarcoglycan-specific antibody. Sections from wild-type (WT) mice and from an Sgca-null model previously described by Duclos et al. (6) were used as positive and negative controls, respectively. Staining of α-sarcoglycan was clearly observed at the fiber membrane in SgcaH77C/H77C and WT mice, whereas it was totally absent in Sgca-null mice (Fig. 2A). To examine the localization of DGC components, immunohistochemistry analyses were performed on sections with β-, γ-sarcoglycan and dystrophin antibodies. Noticeably, staining for sarcoglycan proteins was equivalent to WT in SgcaH77C/H77C muscles, whereas it was greatly reduced in Sgca-null skeletal muscles (Fig. 2A). Thus, the mutated α-sarcoglycan was expressed, localized correctly at the membrane and did not perturb the DGC assemblage in mouse muscle, in contrast to the human situation where α-sarcoglycan was not detected at the membrane in biopsies from patients carrying the R77C mutation (data not shown).
Muscle histology on deltoid, psoas, gastrocnemius, gluteus, extensor digitorum longus and quadriceps removed from SgcaH77C/H77C mice at 3 and 9 months of age showed no dystrophic features (Fig. 2B and data not shown). We submitted Sgca77C/77C mice of the same two ages to an eccentric contraction exercise protocol followed by an intraperitoneal injection of Evans blue dye (EBD). No noticeable uptake of the dye was detected in any of the tested muscles, indicating the absence of plasma membrane disruption (Fig. 2C, right panel). In contrast, EBD uptake was always observed as clusters of skeletal muscles from Sgca-null mice with 13% of fibers being positive (Fig. 2C, left panel). Thus, inconsistent with the human clinical features but in accordance with the presence of the sarcoglycan complex at the membrane, SgcaH77C/H77C mice showed no histological dystrophic signs, not even after exercise. As Sgca-null mice present a severe dystrophic phenotype, we know that α-sarcoglycan is as essential in mice as in humans and that no compensatory mechanism takes place in mice (12,13).
The R77C mutant is retained in the secretory pathway in human, and consequently to the disappearance of a positively charged residue
To explore the underlying cause of the difference in outcome between the two species, we established a heterologous cellular model. A human cell line (HER911) was cotransfected with pcDNA3 vectors expressing β, γ and δ-sarcoglycans from human together with WT or mutated α-sarcoglycan. Whereas α-sarcoglycan WT co-transfected with other members allowed the formation of the complex (Fig. 3A, left panel), transfection with the R77C mutant did not show any labeling when non-permeabilized cells were used, indicating an absence of the complex at the membrane (Fig. 3A, right panel). Western blot analyses confirm that the absence of labeling was not due to the absence of expression of the mutated proteins (Fig. 3, insets). Thus, these observations indicate that this model recapitulates features observed in human patients. It would have been also interested to analyze the fate of H77C in cells as this protein was able to reach the surface in mouse. However, the available antibody directed against the extracellular domain of α-sarcoglycan does not recognize the murine protein by immunohistochemistry. In addition, using the galactosyltransferase–GFP fusion protein as reporter of endoplasmic reticulum (ER), we observed, in permeabilized cells labeled with anti-α-sarcoglycan, accumulation of the R77C protein in the secretory pathway (Fig. 3B).
The presence of the cysteine residue in R77C may have been responsible for an illegitimate interaction, possibly by disulfide bond formation, resulting in retention in the secretory pathway. Yet, no membrane associated-labeling was detected when a non-conservative mutant, R77G was co-expressed with the three other human sarcoglycans, demonstrating that the presence of cysteine was not the cause of the pathology (Fig. 3C). In contrast, when a mutant that conserves a positively charged residue at the 77 position was expressed (R77K), a clear labeling at the plasma membrane was observed (Fig. 3D). In conclusion, it appeared that the phenotype observed was the consequence of the disappearance of a positively charged residue instead of the presence of a cysteine.
In vitro rescue of the R77C mutant protein
In an attempt to genetically rescue the R77C protein, we expressed it in a context where its immediate partners were of a murine origin by co-transfecting the human mutant with all other murine sarcoglycans. No membrane labeling was observed (Fig. 4A). As a control for the complex formation between species, we co-transfected the human WT α-sarcoglycan with all other murine sarcoglycans. A membrane-associated fluorescence was observed (Fig. 4B). These data demonstrated that the murine sarcoglycans could not rescue the mutated human α-sarcoglycan.
In a second step, considering the observation presented in Figure 3B, we hypothesized that it could be possible to act at the level of ER retention. The fate of newly synthesized glycoproteins is determined by the protein quality control machinery that includes ER chaperones and glycosylation/deglycosylation enzymes (14,15). Folding attempts are eventually interrupted upon substrate demannosylation by mannosidase I (16,17). Terminally misfolded polypeptides are retrotranslocated into the cytosol and degraded by the ubiquitin–proteasome pathway, a mechanism known as ER-associated degradation (ERAD). To test whether the R77C mutant trapped in the secretory pathway was recognized by this system, we treated our model cells with MG 132, a proteasome inhibitor. The use of this non-specific inhibitor of the ERAD system increased the amount of α-sarcoglycan that could reach the membrane. A secondary restoration of sarcoglycan complex at the membrane was observed (Fig. 4C). To confirm that R77C is a substrate of the ubiquitin/proteasome system, we performed an immunoprecipitation experiment. Extracts of HER911 cells co-transfected with the β-, γ- and δ-sarcoglycans together with a plasmid coding for R77C in the absence or presence of MG132 were immunoprecipitated with an α-sarcoglycan antibody. We subsequently analyzed the ubiquitination of the proteins by western blot using an anti-ubiquitin antibody. Poly-ubiquitinated proteins were detected in immunoprecipitates of cells expressing the R77C mutant treated with MG132 (Fig. 4D).
Mannosidase I enzymatic activity leads to the irreversible exclusion of the defective polypeptide from the chaperone folding attempts and therefore consists of a late control quality check point upstream of the retrotranslocation and proteasomal degradation. To investigate the participation of this enzyme in the abnormal trafficking of the R77C mutant, we treated transfected cells with kifunensine or deoxymannojirimycine, two inhibitors of the mannosidase I activity. A restoration of α-sarcoglycan at the membrane was clearly seen (Fig. 4E and F). As a control, we also treated cells transfected with the WT α-sarcoglycan (Fig. 4G and H). Altogether, results obtained in our cellular model seem to indicate that R77C mutant was retained in the secretory pathway and that inhibition of ERAD allowed the R77C mutant to reach the plasma membrane.
In vivo rescue of the R77C mutant protein
Our next step was to investigate whether mannosidase inhibition was able to readdress the mutant to the membrane in vivo in the perspective of using this type of intervention as a treatment in humans. We constructed an adeno-associated virus (AAV) vector containing a sequence coding for the human mutant protein placed under the control of a muscle-specific promoter (Supplementary Material, Fig. S2) and injected 1010 viral genome (vg) in the tibialis anterior (TA) muscles of 4-week-old Sgca-null mice. A group of Sgca-null mice injected with the AAV-R77C vector were treated for 2 weeks with kifunensine starting 20 days after AAV injection. To evaluate the subcellular distribution of α-sarcoglycan with and without the kifunensine treatment, we labeled muscle sections with α-sarcoglycan and calreticulin, a marker of the ER. Muscle section from mice non-treated with kifunensine showed accumulation of α-sarcoglycan around the nucleus and co-localization with calreticulin, indicative of aggregation of the transgenic protein in the ER (Fig. 5A, upper panels). Muscle from mice treated with kifunensine showed a decrease in the number of fibers containing such accumulation (Fig. 5A, lower panels). This reduction in ER aggregation is associated with an increase in the intensity of labeling of α-sarcoglycan at the plasma membrane (mean fluorescence level associated at the plasma membrane: 2.32 U.A. for the non-treated muscles versus 13.26 U.A. for the kifunensine treated muscle. This 575% increase indicates that kifunensine allowed the R77C mutant to by-pass the ER associated quality control and reach the plasma membrane.
Kifunensine-mediated localization of the mutated α-sarcoglycan at the plasma membrane allowed the formation of the sarcoglycan complex as seen by the β-sarcoglycan staining at the surface and its perfect co-localization with α-sarcoglycan (Fig. 5B). We also analyzed the effect of kifunensine on Sgca-null mouse muscles that have not been transduced with AAV. As no labeling was observed with β-sarcoglycan, we excluded the possibility that kifunensine promote the trafficking of the residual sarcoglycan complex in the absence of α-sarcoglycan (Fig. 5C). The presence of the complex in turn ameliorated the histological phenotype of transduced Sgca-null muscle (Fig. 5D). Morphological analyses showed a decrease number of centrally located nuclei and a reduced dispersion in fiber cross-section area in kifunensine-treated muscle compared with the non-treated muscles [dispersion factor (DF): 1.39 µm compared with 1.71 µm] (Fig. 5E). This improvement in phenotype was also associated with a reduction in the membrane instability evaluated by EBD staining (data not shown). All these data indicated that the treatment with kifunensine has the capacity to redirect the human mutant protein to the plasma membrane, to ensure the stability of the sarcoglycan complex in this murine context and to ameliorate the phenotype. It is worth to note in regards to potential safety of this approach that muscles from WT mice injected with kifunensine did not display any histological abnormalities (data not shown).
A large proportion of patients suffering LGMD2D have been reported to carry the R77C missense mutation in the α-sarcoglycan gene. In this study, we report the identification of a new pharmacological strategy that could potentially treat such patients. Our starting point was the observation of a lack of phenotype in a knock-in murine model carrying an equivalent mutation. This unexpected finding was confirmed by another study in which a different construct introducing the same mutation was used (18). In this other model, the mutant α-sarcoglycan was also correctly localized at the membrane and failed to give any dystrophic signs over the lifetime of the H77C/H77C mouse.
As a means to identify the possible therapeutic options, we thought it would be of interest to identify the reason for the difference in the outcome between the human and murine species. We provide evidences that the disappearance of a positively charged residue at the 77 position of the human α-sarcoglycan protein seems to prevent the correct sarcolemmal targeting of the protein in human. Our observations are further supported by a previous study showing that the R77C mutation leads to a fundamental defect in α-sarcoglycan protein synthesis which blocks its exit from the ER (19).
Kobuke et al. (18) also performed adenovirus-mediated transfer of the human R77C protein in Sgca-null pups and observed restoration of the sarcoglycan complex. This observation is to some extent different from our own gene transfer experiments in which, even if α-sarcoglycan was observed at the membrane, its level was reduced. It is possible that this difference may be related to the age of the mice used (neonates versus 3–4 weeks of age), especially as the disease is already widespread in muscle in the older animals, the nature of the vector (adenovirus versus AAV) or the strength of the promoter (Rous sarcoma virus promoter versus the muscle-specific C5-12 promoter). Altogether, both studies are in favor of a difference in the processing of the mutant protein between human and murine species. The weaker α-sarcoglycan staining after AAV-mediated transfer of R77C suggests that the human version of the mutated protein is still not quantitatively fully considered as a normal protein in a murine context. Even though, considering the absence of a dystrophic phenotype of both models, the murine mutant seems to be processed normally, it is possible that the shorter lifespan of the mouse could prevent the apparition of deleterious effect resulting from a slightly imperfect processing. In addition, variability in the efficacy of the processing pathways may provide an explanation for the variability of phenotype in human patients and components of this pathway are, hence, potential disease-modifying factors.
Following this demonstration that the central defect in the R77C patients is a failure in the correct delivery of the protein at the sarcolemma, we hypothesized that a pharmacological intervention could be beneficial for these patients. The first obvious target for such intervention is the proteasome. Indeed, we showed that inhibition of the ubiquitine/proteasome system by the use of MG132 increased the sarcolemmal targeting of the R77C protein in transfected cells. However, because of the potential toxicity of such treatment in humans, we decided to intervene upstream in the ERAD system by inhibiting mannosidase I, an ER enzyme that ensures that terminally misfolded glycoproteins are preferentially targeted for degradation. The inhibitor that we used, kifunensine, is routinely used in vitro to manipulate the ERAD system but has not been used in living animals. Most importantly, we showed that the in vivo use of kifunensine could block the ERAD upstream of the proteasome and allowed the rerouting of mutated protein to the plasma membrane. Noticeably, this treatment did not present any overt pathological effect. The animals tolerated well the injection and we did not notice any pathological signs in normal injected muscles. Importantly, the mutated R77C human protein is correctly targeted at the membrane following transfer in a murine context and treatment with kifunensine and once at the membrane is able to stabilize the sarcoglycan complex, demonstrating therapeutic efficiency. Increasing the amount of a mutant protein to the membrane by pharmacological agents has been previously used in several diseases, exemplified in the case of the most common cystic fibrosis (CF) mutation, deltaF508. In this disease, amelioration of the folding, and hence the presence at the membrane, by chemical chaperones (e.g. TMAO) has been successful in vitro and in vivo (20). However, the dose used to obtain such efficiency is very close to the 50% lethality dose. Increased trafficking of mutant forms of the CF transmembrane conductance regulator protein has been achieved in vitro by reducing its degradation by the proteasome (21). Although one proteasome inhibitor has reached the clinic, this type of drug is highly toxic in vivo. Compared with those compounds, mannosidase inhibitors have the potential to have a similar efficiency at a lower toxicity level in vivo.
The data presented may open a new therapeutic approach of LGMD2D. It has been reported that other mutations in α-sarcoglycan or in other sarcoglycan proteins could also lead to an abnormal trafficking and absence of the whole complex at the membrane (22). Therefore, these mutants, as well as mutations in other pathologies linked to defect in trafficking of secreted or membrane proteins, would undoubtedly be candidate for mannosidase inhibition.
In conclusion, our results, in addition of being a concrete example of the interest to investigate the underlying reason of difference in phenotype between species, define a new therapeutic strategy that could be of significance for other disease apart from LGMD2D.
MATERIALS AND METHODS
Cell culture, transient transfection and cell treatment
Human embryonic retinoblast HER911 cells were grown in Dulbecco's modified Eagle's medium supplemented with glutamine, antibiotic (Gentamycin), 10% fetal calf serum and MEM vitamin solution (100×, Sigma M6395). Cells were transfected using 6 µl of Fugene (ROCHE) for 1 µg of plasmid. Forty-three hours later, cells were treated for 5 h with a proteasome inhibitor (5 µm of MG132 dissolved in DMSO, Sigma C2211) or a mannosidase inhibitor (5 µm of kifunensine, VWR calb 422500-1). In six-well plates, 0.5 µg of each plasmid (α-sarcoglycan, α-sarcoglycan-R77C, α-sarcoglycan-R77G or α-sarcoglycan-R77K together with β-sarcoglycan, γ-sarcoglycan and δ1-sarcoglycan) was used. The Golgi compartment was labeled using pAcGFP1-Golgi vector which expressed the galactosyltransferase fused to EGFP (Clontech).
In vivo analysis
All experiments were performed in accordance with the European guidelines and regulations. Mice were submitted to a 30 min running session on a treadmill set to a 15° down slope at a speed of 10 m/min. At the end of the 3-day training period, the mice were injected intraperitoneally with EBD (1 mg/g of body weight, Sigma). The next day, mice were sacrificed and muscles were removed and quickly frozen in liquid nitrogen-cooled isopentane.
Female Sgca-deficient mice of 3 or 4 weeks of age were used throughout this study. Each rAAV2/1 viral preparation was injected (1010 vg in 30 µl in total) into the left TA muscle. At days 20, 21, 22, 25, 27, 29, 32 and 34, kifunensine (VWR, calb 422500-1) was injected directly into the same muscle. Initially, two quantities have been used (30 or 90 pmoles) with similar results. Subsequent experiments were performed only with the lowest dose. This mode of administration was chosen as no data exist about the bioavailability for oral or IV administration. Left and right TA muscles from untreated and treated Sgca-null mice were collected and quickly frozen in liquid nitrogen-cooled isopentane for further analysis.
The fiber diameter was analyzed by measuring the minimum cross-section area of more than 2000 individual fibers. The width of the confidence interval gives the dispersion in the size fibers. The equation used to calculate the DF was the following: DF=mean+(Z* (SD/square root (n)) − mean+(Z * (SD/square root (n)). The value for Z was defined as 1.96 for a confidence of 95%.
Muscle RNAs were extracted by the TRIZOL method after pulverization using a Fast-prep apparatus (Bio101). Synthesis of cDNA was performed from 1 µg of total RNA using the SuperScript II first-strand synthesis system for RT–PCR kit (Invitrogen) and random primers. PCR amplification of the α-sarcoglycan cDNA was done on this cDNA with the forward primer (mSgca.s: 5′-ACTGGCTAGGCCCCGGGCAA-3′) located just upstream of the ATG and the reverse primer (mSgca.as: 5′-GGCTGGGCTGTCAGTGCTGG-3′) located across the stop codon under the following conditions: 94°C for 3 min, then 30 cycles of 94°C for 30 s, 61°C for 40 s and 72°C for 1 min, then 3 min at 72°C. A 1195 bp product was obtained.
Western blot, immunohistology and immunoprecipitation
Western blot and immunohistology were performed as previously described (23) and briefly summarized in Supplementary Material and Materials and Methods.
Transfected HER911 cells were scraped from cell culture dishes. After centrifugation and washing with ice-cold PBS, the cell pellet was incubated 20 min with 200 µl of ice-cold lysis buffer [10 mm Tris (PH 7.4), 5 mm EDTA, 50 mm NaCl, 50 mm NaF, 1% (v/v) Triton X-100, 500 µm sodium orthovanadate, 1× Complete Mini protease inhibitor cocktail (Roche Biomedicals)]. After centrifugation, the lysate was incubated 2 h at 4°C with a monoclonal mouse antibody against α-sarcoglycan (1/100, Novocastra NVL-a-SARC) on rotating wheel. Lysates were mixed with protein A-sepharose (100 mg/ml) for 45 min at 4°C on rotating wheel. The beads were washed three times with lysis buffer. Immunoprecipitated proteins were eluted from the beads by incubating at 70°C in SDS–PAGE sample buffer for 10 min and processed for western blot analysis. Revelation was performed using anti-ubiquitin antibody (1/100, Santa Cruz Biotechnology Ub-P4D1).
Quantitation of confocal microscopy images of muscle sections from mice treated or not with kifunensine was performed using the Plot Profile tool of NIH ImageJ software. Appropriate linear slices were chosen to cross several fibers. The relative magnitude of fluorescence distribution along linear slices after background subtraction of the fiber was quantified. The average pixel intensity in 10–30 lines from n = 3 mice for each condition was determined. Histogram of the different conditions is presented and it reveals quantitative differences in plasma membrane localization. Values are presented as fluorescence arbitrary unit (AU) ± standard deviation (SD).
Supplementary Material is available at HMG Online.
This work was funded by the Association Française contre les Myopathies.
We acknowledge the excellent expertise of the In vivo Department of Généthon. We thank the Howard Hughes Medical Institute for providing us with Sgca-null mice (Iowa City, USA). We also thank Dr Serge Bouaziz for helpful discussion and Dr S. Cure for critical reading of the manuscript. The SgcaH77C/H77C mutant mice were developed with the support of the platform RIO-MNG with the technical help of Anne-Marie Mura and Mireille Richelme.
Conflict of Interest statement. A patent covering these results has been filled with IRi as an inventor.