Genetic compensation for sarcoglycan loss by integrin α7β1 in muscle

Disruption of the sarcoglycan complex leads to muscle membrane instability and muscular dystrophy in humans and mice. Through the dystrophin glycoprotein complex, sarcoglycan participates in connecting the internal cytoskeleton to the membrane and the extracellular matrix. Integrin α7β1 is also a transmembrane protein of skeletal and cardiac muscle that similarly links the cytoskeleton to the extracellular matrix. Mice lacking integrin α7 develop mild muscle degeneration, while sarcoglycan mutant mice display overt muscle degeneration and muscular dystrophy. In sarcoglycan-deficient muscle, integrin α7 protein was upregulated at the plasma membrane. To ascertain whether integrin α7 upregulation compensates for the loss of the transmembrane sarcoglycan linkage in sarcoglycan-deficient muscle, we generated mice lacking both integrin α7 and γ-sarcoglycan (gxi). These double-mutant gxi mice exhibit profound, rapid muscle degeneration leading to death before one month of age consistent with a weakened cellular attachment to the extracellular matrix. The regenerative capacity of gxi muscle was intact with increased embryonic myosin heavy chain expression, myofiber central nucleation and normal in vivo myoblast differentiation. Therefore, upregulation of integrin α7β1 compensates as a transmembrane muscle cell attachment for sarcoglycan consistent with overlapping roles for sarcoglycan and integrins in mediating cytoskeletal-membrane-extracellular matrix interaction.

Dystrophin mutations lead to Duchenne Muscular Dystrophy (DMD) and secondarily destabilize the sarcoglycan complex from the skeletal muscle plasma membrane (Rafael and Brown, 2000). In humans, mutations in sarcoglycan genes lead to a Duchenne-like muscular dystrophy (Bonnemann, 1996;Hack et al., 2000a). Gene targeting of murine sarcoglycan genes recapitulates the human muscular dystrophy phenotype where muscle degeneration is accompanied by muscle regeneration Heydemann et al., 2001;Ozawa et al., 2001). Both δand γ-sarcoglycan bind the cytoplasmic actin binding protein, filamin C (Thompson et al., 2000). Moreover, mutations that disrupt sarcoglycan cause redistribution of filamin C. Sarcoglycan also stabilizes the interaction of dystroglycan subunits (Durbeej et al., 2000;Straub et al., 1998). Dystroglycan is a broadly expressed transmembrane protein with its α subunit binding directly to the G domains of laminin-α2 in the ECM and its β subunit binding dystrophin in the cytoplasm (Henry and Campbell, 1999). Biochemical preparations of the DGC from sarcoglycan mutant muscle reveal a less tightly adherent α-dystroglycan subunit suggesting abnormal interaction between αand βdystroglycan in the absence of sarcoglycan (Durbeej et al., 2000;Straub et al., 1998). Additionally, the absence of sarcoglycan alters membrane integrity in that the muscle membrane becomes abnormally permeable to small molecular mass tracers such as Evans blue dye (EBD) (Hack et al., 1998; 3821 Disruption of the sarcoglycan complex leads to muscle membrane instability and muscular dystrophy in humans and mice. Through the dystrophin glycoprotein complex, sarcoglycan participates in connecting the internal cytoskeleton to the membrane and the extracellular matrix. Integrin α7β1 is also a transmembrane protein of skeletal and cardiac muscle that similarly links the cytoskeleton to the extracellular matrix. Mice lacking integrin α7 develop mild muscle degeneration, while sarcoglycan mutant mice display overt muscle degeneration and muscular dystrophy. In sarcoglycan-deficient muscle, integrin α7 protein was upregulated at the plasma membrane. To ascertain whether integrin α7 upregulation compensates for the loss of the transmembrane sarcoglycan linkage in sarcoglycan-deficient muscle, we generated mice lacking both integrin α7 and γ-sarcoglycan (gxi). These double-mutant gxi mice exhibit profound, rapid muscle degeneration leading to death before one month of age consistent with a weakened cellular attachment to the extracellular matrix. The regenerative capacity of gxi muscle was intact with increased embryonic myosin heavy chain expression, myofiber central nucleation and normal in vivo myoblast differentiation. Therefore, upregulation of integrin α7β1 compensates as a transmembrane muscle cell attachment for sarcoglycan consistent with overlapping roles for sarcoglycan and integrins in mediating cytoskeletal-membrane-extracellular matrix interaction. Matsuda et al., 1995;Straub et al., 1997). Thus, the sarcoglycan-dystroglycan complex mediates at least two links to the cytoskeleton, to dystrophin and filamin C, and coordinates interaction with laminin α2 in the ECM. Like the integrin complex, the sarcoglycan-dystroglycan complex mediates interactions between the ECM and membrane cytoskeleton.
Upregulation of integrin α7β1 has been observed in DMD muscle biopsies . A transgenic mouse model of integrin α7BX2 overexpression was bred with mice lacking dystrophin and utrophin where it ameliorated the severe phenotype seen in these mice (Burkin et al., 2001). The presence of the integrin α7BX2 transgene on the dystrophin/ utrophin mutant background reduced phagocytic cell infiltration and embryonic myosin heavy chain expression and improved lifespan (Burkin et al., 2001). Upregulation of integrin α7 produced from the transgene was relatively modest and the mechanism by which this increase reduced muscular dystrophy was not known.
Because both integrin α7β1 and the sarcoglycans are transmembrane proteins that mediate laminin interactions in muscle, upregulation of integrin may compensate for the loss of sarcoglycan. A conditional allele that targets the integrin β1 gene in skeletal muscle results in late embryonic lethality (Schwander et al., 2003). Integrin β1 muscle displays delayed myoblast fusion and disorganized sarcomere structure. Dystroglycan, dystrophin and laminin α2 appear to be normally localized in integrin β1 null muscle highlighting the importance of integrin complexes in muscle development. We found that integrin α7 was upregulated in mice lacking sarcoglycans. To determine whether integrin α7 upregulation is compensatory, we bred mice mutant for γ-sarcoglycan (gsg -/-) with mice mutant for integrin α7 (Itgα7 -/-) to create mice lacking both proteins (gxi). Double-mutant gxi mice die within one month of birth and examination of these mice showed pervasive muscle degeneration. gxi muscle showed more EBD uptake than gsg -/muscle indicating greater muscle degeneration. As loss of integrins may impair muscle regeneration, we determined that muscle regeneration was intact in gxi mice as gxi myoblasts showed normal in vitro and in vivo differentiation. These results argue that integrin and sarcoglycan have overlapping roles in maintaining muscle membrane stability and that integrin upregulation compensates for sarcoglycan loss.

Microsome preparation and immunoblotting
Heavy microsomes were purified as described (Ohlendieck and Campbell, 1991) with modifications Hack et al., 2000b). Microsomes were prepared from a minimum of three animals, using six distinct muscle groups from each animal (quadriceps, gastrocnemius, soleus, biceps, triceps and pectoralis). Microsomal protein content was determined for each sample using the BioRad (Hercules, CA) protein assay. Protein was subjected to denaturing and reducing conditions, resolved by SDS-PAGE using either 4-12% or 4-20% linear gradient gels (Novex, San Diego, CA) and transferred to Immobilon P membranes (Millipore, Bedford, MA). Equal loading was confirmed by Coomassie blue staining. Immunoblotting was performed as described previously (Hack et al., 1998) with antibodies (listed below). Detection was performed with ECL-Plus (Amersham-Pharmacia, Piscataway, NJ) and visualized on film or using a Storm 860 (Molecular Dynamics, Sunnyvale, CA) and quantified using ImageQuant software (Molecular Dynamics, Sunnyvale, CA).
For laminin α2 immunoblotting, laminin was extracted from muscle tissue essentially as described (Xu et al., 1994). Briefly, 0.1 g (wet weight) of frozen quadriceps muscle was ground in liquid nitrogen and transferred into 1 ml extraction buffer without EDTA [150 mM NaCl, 50 mM Tris-HCl, pH 7.4 and protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany)]. This mixture was homogenized briefly and then centrifuged at 16,000 g for 15 minutes at 4°C. The supernatant was kept and frozen while the pellet was resuspended in 0.3 ml of extraction buffer with 10 mM EDTA and incubated on ice for 1 hour with periodic mixing. The mixture was centrifuged as before, and the supernatant was kept while the pellet was discarded. The protein content of the EDTA extract was quantified using the BioRad protein assay. 20 µg of protein was separated on 4-10% SDS/PAGE gradient gels under non-reducing conditions at 160V for approximately 20 hours and then transferred to Immobilon P membrane. The membrane was blocked in 10% dry milk in phosphate buffered saline with 0.1% Tween-20 and then incubated with polyclonal anti-laminin α2 at 1:500 (Ab 1301) (Kuang et al., 1998) in fresh blocking buffer. Immunoreactive protein bands were visualized as described above.

Immunocytochemistry
Mice from representative genotypes were sacrificed, and skeletal muscle was dissected from gsg -/-, wild-type (WT), Itgα7 -/and gxi animals and frozen in liquid nitrogen-cooled isopentane. 7 µm sections were prepared using a cryostat at -20°C, fixed in ice-cold methanol for 2 minutes and blocked in a solution of phosphate buffered saline (PBS) with 5% fetal bovine serum (FBS) for one hour at room temperature (RT). Primary antibodies (see above) were diluted in blocking solution and incubated overnight at 4°C. Cy-3 and FITC-conjugated secondary antibodies (Jackson ImmunoResearch) were diluted in blocking solution at 1:2500 at RT for 2 hours. Sections were mounted with Vectashield containing DAPI (Vector Laboratories, Burlingame, CA) and photographed using a Zeiss Axiophot microscope equipped with an Axiocam (Carl Zeiss, Germany). Where double staining with a polyclonal and monoclonal antibody was necessary, serial sections were taken. In order to minimize the background caused by the anti-mouse secondary antibody, the mouse on mouse (MOM) immunodetection kit was used in conjunction with the avidin/biotin blocking kit (Vector Laboratories, Burlingame, CA). The experiment was carried out according to manufacturer's protocol.

Central nuclei quantification
Hematoxylin-and eosin-stained quadriceps sections of each genotype were analyzed for number of centralized nuclei. Sections from 6-12 animals were used for each genotype and two sections from each slide were counted. Eight random microscopic fields were analyzed per section, per slide. This resulted in 5000-7000 myofibers being analyzed per genotype.

Primary myoblast cultures
Primary myoblasts were isolated as described (Rando and Blau, 1997). Briefly, muscle was dissected from 1-3 day old mice and placed in PBS where it was minced using a razor blade. Approximately 2 ml collagenase/dispase/CaCl2 per gram of tissue was added and the mixture was incubated at 37°C for 30-45 minutes. This slurry was passed through an 80 µm nylon mesh filter and centrifuged for 5 minutes at 350 g. The pellet was resuspended in 8 ml F10-based primary myoblast growth medium and plated in a collagen-coated dish. The cells underwent a variable number of rounds of preplating to reduce fibroblast contamination and for enrichment of myoblasts. Differentiation was induced with Dulbecco's modified Eagle medium (DMEM) supplemented with 5% horse serum. Fusion indices were calculated as described (van der Putten et al., 2002). Briefly, myoblast cultures were allowed to differentiate for 6 days prior to fusion index calculation. A minimum of six low-power microscopic fields was counted for each genotype. Fusion index was calculated as the number of nuclei in myotubes divided by the total number of nuclei. A myotube was defined as having three or more nuclei.
Phenotypic analysis, histology, terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) assay and Evans blue staining Animals were observed daily after birth for phenotypic abnormalities and identification of the gxi genotype. Tissues for histology were frozen in liquid nitrogen-cooled isopentane and cryosectioned, then placed directly in 10% neutral buffered formalin overnight. Tissues were then stained with hematoxylin and eosin (H&E) and Masson trichrome. TUNEL assays were performed on fixed, cryosectioned tissues using the Apoptag TM fluorescein kit (Intergen, Purchase, NY). Evans blue dye (EBD) (Sigma) staining of muscle in vivo was performed by injecting 5 µl EBD (10 mg/ml in PBS)/gram body weight intraperitoneally. Mice were sacrificed and skinned 10-12 hours after EBD injection. Approximately 1000-20,000 individual myofibers from quadriceps muscle were counted for each genotype for either presence or absence of EBD. All muscles were age-matched to 18-22 days.
Quantitative RT-PCR Total RNA was prepared using 100 mg quadriceps tissue from each genotype. Frozen, powdered muscle was added directly to 1.5 ml Trizol reagent (Invitrogen, Carlsbad, CA). The mixture was homogenized through successively smaller needles (16, 18, 20 and 21 gauge). The remainder of the protocol followed the manufacturer's recommendations. 4 µg total RNA was added as a template for the Superscript II Reverse Transcription kit (Invitrogen, Carlsbad, CA). Following reverse transcription, 2 µl of a 1:4 dilution of each cDNA was used as a template for quantitative PCR using primers to integrin α7 (Itga7F GCTGATACCGCTGCTCTGTTC and Itga7R TCATGTTGGTGCTTCAGCCAC) and glyceraldehyde-3 phosphate dehydrogenase (G3PDH) G3PDHF ACCACAGTCCATGCCATCAC and G3PDHR TCCACCACCCTGTTGCTGTA. PCR was carried out using the DyNAmo SYBR Green qPCR kit using the manufacturer's recommended parameters (Finnzymes, Espoo, Finland) and an MJ Research Opticon Monitor. Cycle thresholds were determined for integrin α7 and GAPDH mRNA levels in wild-type and gsg -/muscle. The experiment was performed in duplicate.

Statistical methods
Comparisons between genotypic groups for EBD uptake, central nucleation and fusion indices were analyzed for statistical significance using GraphPad Instat 3.0 software. A nonparametric ANOVA Kruskal-Wallis test was followed by Dunn's multiple comparisons post-test to determine P values and significance.

Integrin is upregulated in sarcoglycan-deficient mice
We assessed the effect of sarcoglycan loss on the major sarcolemmal integrin complex, α7β1, using antibodies to the integrin α7 subunits. Integrin α7A and α7B derive from alternative splicing that alters the cytoplasmic domain of integrin α7; integrin α7B is the major form of integrin α7 expressed in skeletal muscle (Crawley et al., 1997). We studied microsomal membrane fractions from mice with null alleles in δ-sarcoglycan (dsg -/-) or γ-sarcoglycan (gsg -/-) (Hack et al., 2000b;Hack et al., 1998). Microsomal fractions enriched for membrane-associated proteins and removed contamination by sarcomeric proteins. We also studied mdx mice that lack full length dystrophin (Sicinski et al., 1989) and as a secondary consequence, show a decrease of the sarcoglycan subunits at the plasma membrane (Ohlendieck and Campbell, 1991). In common to each of these three mutants, dsg -/-, gsg -/and mdx is sarcoglycan complex loss from the plasma membrane. Each of these three mutants showed an increase in integrin α7 protein levels (Fig. 1). We evaluated whether integrin α7 mRNA was increased to account for the upregulation seen on immunoblotting. Using quantitative RT-PCR in two separate experiments, we found that integrin α7 mRNA was increased around threefold (average 3.1) in gsg -/muscle compared to wild-type muscle. Interestingly, the degree of integrin protein upregulation was more marked in δsarcoglycan mutant mice. Although the phenotype is identical between gsg -/and dsg -/mice, dsg -/mice have a more complete loss of the sarcoglycan complex (Hack et al., 2000b) potentially accounting for the greater increase of integrin α7 in dsg -/compared to gsg -/and mdx.

Double-mutant (gxi) mice display early lethality
To determine if integrin α7 upregulation compensates for the absence of sarcoglycan, gsg -/-Itgα7 -/double-mutant mice (gxi) were generated by crossing gsg +/-Itgα7 +/mice. We chose γ-sarcoglycan mice for this analysis as the defect in these mice derives wholly from a striated muscle-intrinsic defect, as opposed to a vascular smooth muscle defect as has been hypothesized for δ-sarcoglycan null mice (Coral-Vazquez et al., 1999). It was difficult to discern a phenotype in very young animals, as activity levels are very limited in all genotypes at this young age. By 10-14 days of age most gxi animals were smaller ( Fig. 2A) than their littermates and began to display an abnormal 'hopping' gait and evidence of kyphoscoliosis (see movie, http://jcs.biologists.org/supplemental/). In addition, gxi mice demonstrated little to no normal grooming behavior. The cervical-thoracic angle was more marked in gxi mice than in control mice (Fig. 2B). By 18-20 days after birth, all gxi animals were smaller and weighed about half as much as their littermates. The average lifespan of gxi mice was 21 days (Fig.  2C). gxi mice had reduced mobility in that their ability to walk on smooth surfaces was markedly compromised reflecting muscle weakness. Food and water were placed within reach of Journal of Cell Science 117 (17) Fig. 1. Upregulation of integrin α7 where sarcoglycan level is reduced. Immunoblotting of microsomal membrane fractions purified from skeletal muscle from mice lacking sarcoglycan subunits or dystrophin. The antibodies used are specific to the integrin α7 splice forms (Itgα7A and Itgα7B) representing splice variants that alter the cytoplasmic domains of integrin α7 (Mayer et al., 1997). A Coomassie blue-stained loading control is shown in the lower panel of each blot. (A) Upregulation of integrin α7 in muscle null for δsarcoglycan (dsg -/-). (B) Upregulation of integrin α7 in muscle null for γ-sarcoglycan (gsg -/-) and in dystrophin-null mdx muscle. mdx mice have a secondary reduction of sarcoglycan at the plasma membrane (Ohlendieck and Campbell, 1991). (C) Graphical representation of integrin α7 expression from blots shown in A and B. Upregulation of integrin α7 was seen in all three mutant models but was greatest in muscle lacking δ-sarcoglycan (dsg -/-) where sarcoglycan loss is greatest (Hack et al., 2000b). gxi mice have markedly reduced survival rates (n=10 for gsg -/and gxi mice, n=3 for control and Itgα7 -/mice). Integrin α7 mutants do not have reduced lethality less than 6 months of age (Mayer et al., 1997).
gxi mice, but despite this nearly all gxi mice died by 25 days of age.
Severe muscle degeneration results from loss of both integrin α7 and γ-sarcoglycan Histological examination of quadriceps skeletal muscle demonstrated a severe degenerative process in all muscle groups of gxi mice (Fig. 3). Muscular dystrophy is characterized by ongoing necrosis accompanied by regeneration giving rise to variation in myofiber size. An increase in the number of myofibers with centrally placed nuclei is consistent with regeneration. The degenerative process exceeds the regenerative potential so that replacement by connective and adipose tissue ensues. In dystrophin and sarcoglycan mutant muscle, the degenerative process is focal in nature. That is, normalappearing areas of muscle are found neighboring foci of degeneration (Fig. 3, pale blue area, lower left panel). In contrast, gxi muscle showed widespread degeneration and virtually no normal-appearing areas of muscle in any of the muscle groups examined. Fig. 3 is representative of the typical appearance of gxi muscle with interspersed fibrosis seen throughout the myofibers. Age-matched normal muscle showed homogeneity in myofiber appearance, as did age-matched Itgα7 -/muscle, consistent with the very mild degenerative process associated with deletion of the Itgα7 locus (Fig. 3, upper panels).

Enhanced muscle degeneration in gxi mice
The severe muscular dystrophy in gxi muscle may arise from increased degeneration from lack of muscle membrane integrity or decreased regeneration, or a combination thereof. To distinguish these possibilities, we studied gxi mice using the vital tracer EBD. Normal muscle is impermeable to EBD whereas muscle that lacks sarcoglycan or dystrophin becomes abnormally permeable to this tracer (Matsuda et al., 1995). Since gxi mice do not survive beyond 21-25 days, we studied 21-day-old gxi mice and gsg -/mice and found significant uptake of EBD in gxi muscles and comparatively little EBD uptake in gsg -/muscle (Fig. 4). Normal mice and 3-week-old Itgα7 -/mice show no EBD uptake grossly (data not shown) or microscopically (Fig. 4A, upper panels and Fig. 4B). The loss of both sarcoglycan and integrin α7 results in enhanced membrane permeability and degeneration.
To determine whether the major extracellular matrix attachment for integrin was present, we evaluated expression of laminin α2 (Fig. 5). Laminin α2 upregulation is seen in Masson trichrome staining shows that muscle from integrin α7 mutants (Itgα7 -/-) appears indistinguishable from the wild type in young mice (aged 3 weeks). Muscle from γ-sarcoglycan null mice (gsg -/-) displays focal degeneration, seen as blue areas. This regional fibrosis is often immediately adjacent to normal appearing regions of muscle (surrounding red). In contrast, double-mutant mouse muscle (gxi) has widespread degeneration affecting all parts of the muscle. At this low magnification view, fibrosis is seen interspersed throughout gxi muscle (note blue areas in lower right panel). Bar, 200 µm. Fig. 4. Evans blue dye (EBD) uptake is increased in gxi doublemutant mice. (A) EBD is a vital tracer and mutations in the dystrophin and sarcoglycan genes lead to enhanced muscle uptake of EBD. Following injection with EBD, muscles from wild-type (WT), integrin α7 mutant (Itgα7 -/-), γ-sarcoglycan mutant (gsg -/-) and double mutant (gxi) mice were examined grossly and microscopically where EBD staining appears red. Counterstaining with an anti-dystrophin antibody outlines myofibers (green). Bar, 100 µm. (B) Percentage of EBD-positive fibers in wild-type and mutant mice. EBD uptake is increased in both gxi and gsg -/muscle compared to control muscle (P<0.001 and P<0.01, respectively). EBD uptake is increased gxi and gsg -/muscle compared to Itgα7 -/muscle (P<0.001 and P<0.05, respectively). EBD uptake is also increased in gxi muscle compared to gsg -/muscle (P<0.05).
gsg -/muscle, and this increase may be functional as it can couple to the increased integrin α7 complex. As both integrin and sarcoglycan are absent in gxi mice, the upregulation of laminin α2 may be unable to function. In gxi muscle, scattered myofibers with decreased laminin α2 can be seen (asterisk, Fig.  5B) and are likely to represent those fibers undergoing degeneration. In both gxi and Itgα7 -/muscle, β-dystroglycan was upregulated. Since muscle degeneration is mild in Itgα7 -/muscle, upregulation of β-dystroglycan may be functional. Thus, in the setting of an intact sarcoglycan complex, as is present in Itgα7 -/muscle, β-dystroglycan upregulation may partially compensate for the loss of the integrin transmembrane linkage.

Regenerative properties of gxi muscle are intact
To evaluate whether doubly deficient gxi mice have impaired muscle regeneration, we quantified the number of central nuclei in the myofibers of each genotype since centrally placed nuclei indicate myofibers that have undergone regeneration. We found that gsg -/and gxi muscle had a statistically significant increase in centrally nucleated myofibers compared to either wild-type or Itgα7 -/muscle (Fig. 6). We also studied the expression of embryonic myosin heavy chain (eMyHC) as expression of this myosin isoform reflects myofiber regeneration. In gsg -/muscle, expression of eMyHC is noted surrounding regions of degeneration (Fig. 7A, lower left panel). In contrast, eMyHC expression was present diffusely throughout gxi muscle reflecting regeneration in concert with diffuse degeneration (Fig. 7A, lower right panel). This pattern parallels the widespread degeneration seen with histology and reflects the close coupling of degeneration and regeneration. Normal muscle displays no eMyHC expression and Itgα7 -/muscle taken from young mice also does not express eMyHC (Fig. 7A, upper panels). We evaluated the contribution of programmed cell death to the gxi phenotype with TUNEL labeling. An increase in TUNEL-positive nuclei characterizes dystrophin and sarcoglycan mutant muscle compared to normal muscle (Hack et al., 1998;Matsuda et al., 1995). The number of TUNELpositive nuclei appears insufficient to account for the degree of degeneration seen in sarcoglycan or dystrophin mutants, but the increase is nonetheless a consistent feature that occurs near regions of muscle degeneration. In gxi muscle, TUNELpositive nuclei were seen dispersed throughout the entire muscle consistent with an enhanced and widespread degenerative process (Fig. 7B).

In vitro fusion is normal in gxi myoblasts
To evaluate myofiber regeneration in gxi muscle further, we Journal of Cell Science 117 (17) Fig. 5. Upregulation of laminin α2 and β-dystroglycan in gxi mouse muscle. (A) Immunoblot with antibodies specific to laminin α2, βdystroglycan or sarcomeric actin in single (Itgα7 -/and gsg -/-) and double-mutant (gxi) mice. Upregulation of laminin-α2 may be effective through the integrin α7β1 complex that is upregulated in gsg -/muscle. An increase of β-dystroglycan is seen in Itgα7 -/mutant muscle and this upregulation may be significant as an intact sarcoglycan-dystroglycan complex is present. In contrast, upregulation of β-dystroglycan is ineffective in gxi muscle where the major laminin binding complexes, integrin and sarcoglycandystroglycan, are disrupted. (B,C) Photomicrographs showing muscle membrane staining for laminin α2 (B) and β-dystroglycan (C) in each genotype. Scattered fibers with reduced laminin α2 staining can be seen in gxi muscle (B, lower right panel*). Bars in B and C, 50 µm. Fig. 6. Regeneration is present in gxi muscle. The percentage of centrally placed nuclei is increased in gxi and gsg -/muscle reflecting enhanced regeneration. Centrally placed nuclei develop after myoblasts fuse to regenerate muscle. Quadriceps muscle was studied. There is no significant increase in number of centrally nucleated myofibers in integrin α7 mutant (Itgα7 -/-) muscle (P>0.05). γ-sarcoglycan mutant (gsg -/-) and double-mutant gxi muscle both have significantly increased centrally nucleated myofibers compared to the wild type and Itgα7 -/-(P<0.001 for all comparisons). No difference was detected between the number of centrally nucleated fibers between gsg -/and gxi mice (P>0.05). cultured primary myoblasts from normal, gsg -/-, Itgα7 -/and gxi muscle. In each case myoblasts were cultured from neonatal mice and upon serum starvation showed normal properties of differentiation and fusion to myotubes. These differentiated cultures were immunostained using antibodies directed against both eMyHC (red) and dystrophin (green) as expression of these proteins reflects normal differentiation of myotubes in culture (Fig. 8A). Normal and mutant cultures from each of the genotypes were able to express eMyHC and dystrophin in developing myotubes, indicating that gxi mice retained the ability to regenerate damaged muscle. We determined the fusion index for each genotype to measure the timeframe of myotube development. We found no statistically significant difference between any of the genotypes examined consistent with intact regenerative potential of gxi myotubes (Fig. 8B).

Upregulation of integrin α5
Integrin α5 is highly expressed in developing myotubes and serves as the major fibronectin linkage for myofibers (Muschler and Horwitz, 1991). We found that integrin α5 was increased in both dystrophin and sarcoglycan mutant mice (Fig. 9A). We examined the expression of integrin α5 in gxi, γ-sarcoglycan and integrin α7 mutant muscle and found that integrin α5 was upregulated more in gxi compared to gsg -/muscle. Immunostaining of gsg -/and gxi muscle revealed that integrin α5 upregulation was seen in fibers positive for embryonic myosin heavy chain expression (Fig. 9D). In gxi muscle, the increase in integrin α5 expression was also seen in areas with increased connective tissue deposition as can be noted by the cluster of DAPI positive nuclei in the right hand side of the image taken from gxi muscle (Fig. 9D, lower right panel). As integrin α5 upregulation is generally not associated with mature myotubes, it is less likely to contribute to myofiber stability or instability in this model.

Discussion
Muscle is a dynamic tissue that undergoes marked cell shape changes associated with significant stress and strain during muscle contraction. Mutations in cytoskeletal proteins such as dystrophin implicate the cortical cytoskeleton as important for maintenance of cell stiffness as disruption of this network leads to changes in muscle membrane properties that render muscle more susceptible to damage during muscle contraction (Harper et al., 2002;Petrof et al., 1993). The role of the remainder of the DGC, particularly the sarcoglycan subcomplex, is less well understood. Dystroglycan is a major cell surface receptor for laminin-2, and it is thought that the sarcoglycan complex stabilizes the α and β subunits of dystroglycan . Biochemical dissociation studies of sarcoglycan null muscle suggest that α-dystroglycan is less tightly adherent to the remainder of the complex resulting in a weakened connection from the dystrophin cytoskeleton to the ECM (Durbeej et al., 2000;Straub et al., 1998). Therefore, a complete sarcoglycan-dystroglycan complex is required for normal membrane stability.
The severe phenotype seen in gxi double-mutant mice provides genetic evidence that integrin and sarcoglycan are Fig. 7. Upregulation of embryonic myosin heavy chain expression in gxi mouse muscle. Embryonic myosin heavy chain is expressed in response to regeneration, as it is a developmentally expressed isoform of myosin. (A) Embryonic myosin heavy chain (eMyHC) expression is shown in green, with Evans Blue Dye showing in red. DAPI staining alone is shown (in blue) for both wild-type (WT) and integrin α7 mutant (Itgα7 -/-) sections as no embryonic myosin heavy chain staining was present. The pattern of embryonic myosin heavy chain expression differs between γ-sarcoglycan mutant (gsg -/-) and gxi double-mutant muscle. In gsg -/muscle, embryonic myosin is found in discrete regions consistent with the pattern of focal degeneration and regeneration (A, lower left panel). In contrast, gxi muscle shows widespread expression of eMyHC reflecting widespread degeneration and responsive regeneration. (B) TUNELpositive nuclei are shown in green and are increased near regions of degeneration in gsg -/and are seen scattered throughout gxi muscle. This is consistent with the widespread degenerative pattern and relatively small increase in TUNEL-positive nuclei compared to wild-type muscle. TUNEL-positive cells were absent in sections from the wild type and integrin α7 mutants, and only DAPI staining is shown (blue). Bars in A and B, 100 µm.
parallel pathways for the maintenance of muscle membrane integrity. This point is underscored by recent work showing the phenotype of integrin β1-null muscle. Integrin β1-null muscle has aberrant sarcomere patterning and is defective in myofiber development despite normal expression and localization of dystroglycan, dystrophin and laminin-2 (Schwander et al., 2003). Therefore, it may be that sarcoglycan and integrin α7β1 are compensatory proteins with regard to mature muscle membrane integrity but that integrin has a unique additional role in muscle development. Additional support for this assertion is the observed upregulation of the DGC protein γsarcoglycan seen in integrin α7-deficient muscle as assayed by immunoblot (data not shown). Taken together, the available data seem to point to a functional redundancy between sarcoglycan and integrin for maintenance of muscle membrane integrity but not for actual fusion of myoblasts and development into myotubes. In the absence of both transmembrane linkages, widespread degeneration occurs. In dystrophin or sarcoglycan-deficient muscle, there is focal necrosis with normal-appearing muscle immediately adjacent to areas of muscle damage. In contrast, muscle damage is seen spread throughout gxi muscle with little evidence of spared areas of muscle. gxi double-mutant muscle showed marked Journal of Cell Science 117 (17) Fig. 8. In vitro myoblast differentiation and fusion to myotubes is normal in gxi mutant mouse. (A) To assess the regenerative capacity of myoblasts, we cultured myoblasts from wild-type (WT), integrin α7 mutant (Itgα7 -/-), γ-sarcoglycan mutant (gsg -/-) and double mutant (gxi) neonatal mice. Each culture was differentiated to myotubes, and stained for dystrophin (green) and embryonic myosin heavy chain expression (red). Bar, 50 µm. (B) The fusion index was determined as described in Materials and Methods for each genotype and no significant differences were noted in the timing or degree of myoblast fusion to myotubes (P>0.05 for all comparisons). Fig. 9. Upregulation of integrin α5 is found in regenerating fibers. (A) Immunoblot and graphical representation of integrin α5 expression in wild-type (normal), δ-sarcoglycan mutant (dsg -/-), γsarcoglycan mutant (gsg -/-) and dystrophin mutant (mdx) muscle protein. Integrin α5 is a fibronectin receptor in muscle and is upregulated in DGC mutant muscle. β-actin labelling is shown in the lower panels as a loading control. (B) Immunoblot showing that integrin α5 is upregulated in double-mutant gxi and gsg -/mutant muscle compared to muscle from integrin α7 mutant (Itgα7 -/-). βactin labelling is again shown in the lower panel as a loading control. (C) Immunostaining with an antibody to integrin α5 shows little to no increase in Itgα7 -/muscle compared to wild-type (WT) muscle. (D) Immunostaining shows that integrin α5 is found in regions of regeneration as the same fibers that are positive for integrin α5 also stain with embryonic myosin heavy chain (green). The increased fibrosis in gxi muscle also reacts with the integrin α5 antibodies consistent with its expression outside of muscle. Bar, 50 µm (C,D). EBD uptake consistent with enhanced muscle damage. The early lethality in gxi mice appears to be the result of muscle weakness itself and probably involves a decline in the function of respiratory muscles. Examination of the hearts from gxi mice did not reveal evidence of focal necrosis like that seen in older gsg -/mice (Heydemann et al., 2001) (data not shown). In the case of gsg -/mice, cardiomyopathy typically develops by 6 months of age and often is associated with sudden unpredictable death probably arising from cardiac arrhythmias. In contrast, gxi mice appear unwell with reduced ambulation and an increased respiratory rate. The absence of overt cardiomyopathy may be explained because the mice do not survive long enough for focal degenerative cardiomyopathy to develop.
Within a myofiber, there may be regional roles for sarcoglycan and integrin in the maintenance of muscle membranes. Shear-type injury of muscle resulted in increased integrin expression at the ends of myofibers where integrins participate in forming myotendinous junctions. Shear stress that disrupted the long axis of the myofiber membrane produced upregulation of integrin and sarcoglycan, suggesting that these transmembrane linkages may both be important to lateral interactions of myofibers with the extracellular matrix and neighboring myofibers (Kaariainen et al., 2000;Kaariainen et al., 2001). The severe phenotype in gxi mice may arise from a fully defective ECM attachment along the lateral aspects of myofibers, although based on the known functions of integrins and the hypothesized roles of sarcoglycan, downstream signaling is likely to be impaired. Normal signals, transmitted perhaps through the nucleus and gene expression, are unable to respond and repair in the face of the fully defective ECM connection.
In muscle development, the fusion of myoblasts to myotubes is associated with an increase in integrin expression, notably integrin α5β1 (McDonald et al., 1995). In mature muscle, integrin α7β1 predominates (Crawley et al., 1997), although integrin α5 may also be important for a fibronectin interaction that stabilizes muscle membrane integrity (Taverna et al., 1998). We noted an increase in the fibronectin receptor, integrin α5, in sarcoglycan-deficient muscle. This receptor may also be capable of compensating for the loss of a functional sarcoglycan-dystrogycan unit, or alternatively, the increase in integrin α5 could be pathologic to muscle. Immunolocalization data demonstrated that integrin α5 expression is predominantly coincident with embryonic myosin heavy chain expression and therefore reflects regeneration. While it could be expected that gxi muscle derives its phenotype in part from ineffective regeneration, gxi muscle development does not appear significantly impaired since in vivo features of regeneration are normally present. Furthermore, in vitro myoblast fusion of gxi myoblast cultures was indistinguishable from normal muscle making ineffective regeneration unlikely to account for the severe muscle findings in gxi mice.
The compensatory role of integrin was suggested by recent work showing that overexpression of rat integrin α7β1 can ameliorate the muscular dystrophy phenotype as well as increase the life span of mdx/utr -/-DKO mice (Burkin et al., 2001). Interestingly, only a modest upregulation of integrin subunits was necessary to produce this improvement in phenotype (2.3-fold upregulation of integrin α7B and 1.5-fold upregulation of β1D integrin). Integrin α7β1 RNA levels are upregulated in both mdx mice and human Duchenne and Becker muscular dystrophy patients . More recently, integrin α7B was shown to be upregulated in muscle fibers deficient in dystroglycan, despite dystrophin being present in some of the fibers (Cote et al., 2002). The sarcoglycan complex is secondarily destabilized in dystrophin deficiency. Thus, it is likely that upregulation of integrin is compensating for the secondary loss of sarcoglycan in each of these cases.
The role of integrins as mechanosignaling molecules for cell adhesion has been established. Signals from extracellular matrix proteins such as laminin and fibronectin are transmitted through integrin complexes to intracellular proteins such as focal adhesion kinase (FAK) and phosphatidylinositol (PtdIns) 3-kinase (Howe et al., 2002;Schwartz, 2001). Like the integrins, the sarcoglycan complex may participate in both mechanical and signaling roles at the plasma membrane. Supporting this, the cytoplasmic tails of γand δ-sarcoglycan have conserved tyrosine residues that can be phosphorylated. Furthermore, the cytoplasmic domains of γand δ-sarcoglycan bind filamin C (Thompson et al., 2000). In other tissues, filamins have been shown to interact with integrin subunits, specifically to the β1 subunit of integrin. Thus, filamin may be one of the signaling mechanisms onto which both integrins and sarcoglycans converge. The synthetic lethal phenotype produced by the combined sarcoglycan and integrin loss from the surface of myofibers highlights not only the downstream signaling mechanisms that may function in parallel between these two cellular attachments, but also emphasizes the extracellular component, laminin, with which each of these complexes interacts. These findings help elucidate the role of the sarcoglycan complex, drawing parallels to the integrin complex, in its role in cytoskeletal-membrane-matrix interactions.