Integrins: Redundant or Important Players in Skeletal Muscle?*

Migration of myogenic cells from the somites and their subse- quent fusion to myotubes are key steps during skeletal muscle development. Continuous interaction of the cells with the neigh- borhood most likely triggers all events necessary to induce the genetic programs for differentiation, migration, and fusion into multinucleated myotubes. Likewise, it is well recognized that the function and the maintenance of tissue integrity are dependent on specific interactions of cells with the surrounding extracellular matrix. Transmembrane receptors are involved in polymerization and assembly of the matrix (1, 2) and in addition provide both a mechanical link to the cytoskeleton and a means of transducing signals from the extracellular matrix to the nucleus (3, 4). In skeletal muscle two major types of extracellular-cytoskeletal link- ages exist at the cell plasma membrane. Numerous studies over the past decade have addressed the function of the dystrophin-glyco- protein complex (DGC). 1 Insights into the biological significance of integrin receptors for skeletal muscle development and function have been gained more recently, largely by gene targeting approaches and analyses of human diseases caused by integrin mu- tations. This review will therefore focus mainly on recent advances in understanding of the function of integrins in skeletal muscle.

Migration of myogenic cells from the somites and their subsequent fusion to myotubes are key steps during skeletal muscle development. Continuous interaction of the cells with the neighborhood most likely triggers all events necessary to induce the genetic programs for differentiation, migration, and fusion into multinucleated myotubes. Likewise, it is well recognized that the function and the maintenance of tissue integrity are dependent on specific interactions of cells with the surrounding extracellular matrix. Transmembrane receptors are involved in polymerization and assembly of the matrix (1,2) and in addition provide both a mechanical link to the cytoskeleton and a means of transducing signals from the extracellular matrix to the nucleus (3,4). In skeletal muscle two major types of extracellular-cytoskeletal linkages exist at the cell plasma membrane. Numerous studies over the past decade have addressed the function of the dystrophin-glycoprotein complex (DGC). 1 Insights into the biological significance of integrin receptors for skeletal muscle development and function have been gained more recently, largely by gene targeting approaches and analyses of human diseases caused by integrin mutations. This review will therefore focus mainly on recent advances in understanding of the function of integrins in skeletal muscle.

Integrins in Skeletal Muscle
Integrins form the major family of cell surface adhesion receptors, mediating both cell-cell and cell-matrix interactions. They are heterodimeric, transmembrane glycoproteins consisting of an ␣ and a ␤ chain that are non-covalently associated (5). To date, 18 ␣ and 8 ␤ chains have been identified, and these combine in a restricted manner to form at least 24 different dimers (6). Integrin diversity is increased still further through the expression of intraand extracellular splice variants for several subchains. Among these, the ␤ 1 integrin family forms the largest group of receptors for extracellular matrix proteins.
So far, most conclusions about function and expression of integrins in vertebrate skeletal muscle have been drawn from combined studies in human, rat, mouse, chicken, and quail, mainly utilizing in vitro approaches. It is difficult from these studies to evaluate the precise function of integrins at the tissue level, as some integrins have only been detected in one species but not in others (7). Of the 12 current members of the ␤ 1 integrin family, a subset has been shown convincingly to be expressed in mammalian at focal contacts, costameres, neuromuscular (NMJ) and myoten-dinous junctions (MTJ), or the sarcolemmal membrane during either muscle development or in the adult (Fig. 1).
Integrins in In Vitro Muscle Differentiation-Studies performed in avian have indicated that ␤ 1 integrins are involved in cell migration from the somite (8) and terminal differentiation of myoblasts into myotubes (9). In vitro studies in avian and rodent species imply that the ␣ 4 integrins, containing either the ␤ 1 or ␤ 7 subunit, and ␣ v , ␣ 5 ␤ 1 , ␣ 6 ␤ 1 , and ␣ 7 ␤ 1 integrins are the major players in muscle differentiation. These integrin chains are readily detected in myoblasts. The ␣ 4 integrins were thought to be of particular importance providing the major cell-cell contact for myotube formation during secondary myogenesis through a heterophilic interaction with the counter-receptor VCAM-1 (10). However, chimeric mice with a high percentage of ␣ 4 -deficient embryonic stem (ES) cells formed normal muscle (11). These data, together with the demonstration that, in vitro, ␣ 4 -deficient myoblasts can form myotubes strongly suggests that ␣ 4 -containing integrins are not required to establish the cell-cell contacts necessary for myoblast fusion (11).
Whereas ␣ 5 ␤ 1 is the classical fibronectin receptor, both ␣ 6 ␤ 1 and ␣ 7 ␤ 1 are exclusive laminin receptors. ␣ 5 ␤ 1 and ␣ 6 ␤ 1 are widely expressed and down-regulated after myotube formation (12)(13)(14), whereas ␣ 7 ␤ 1 is mainly restricted to skeletal and cardiac muscle and strongly up-regulated upon myoblast fusion (15,16). The role of ␣ 5 ␤ 1 and ␣ 6 ␤ 1 in muscle development and the reason they coexist at the myoblast stage as ligand-opposing receptors is not yet well defined. Elegant studies by Sastry et al. (17), however, suggested distinct functions for both integrins. Overexpression of the ␣ 5 subunit in primary quail myoblasts maintained them in a proliferative phase, whereas ectopically expressed ␣ 6 ␤ 1 induced myoblast differentiation (17). Given the switch in the muscle cell environment from a fibronectin-rich matrix into a laminin-containing basement membrane with the onset of terminal differentiation (18), these data suggested a fine-tuned regulation of differentiation and matrix assembly via these two integrin receptors. However, no obvious defects in muscle development have been reported either in mice with a targeted deletion of the ␣ 6 subunit (19) or in myoblasts devoid of ␣ 5 ␤ 1 that efficiently differentiated into myotubes (20).
Various results have implicated ␣ 7 ␤ 1 as the crucial receptor for myoblast migration (21)(22)(23), and its strong up-regulation in terminally differentiated myotubes further suggested a functional role in this process. Yet, as with other integrin ␣ chain-null mice, skeletal muscle develops normally in the absence of ␣ 7 ␤ 1 (24).
The apparently normal myogenesis in integrin ␣ chain-null mice could be explained by redundancy or overlap in function (25) because integrin ␤ 1 -inhibiting antibodies, which disrupt the function of all integrins concurrently, perturbed myotube formation in vitro (9). However, a critical role for integrins in muscle development became questionable, when it was demonstrated that skeletal muscle in chimeric mice derived from ␤ 1 -null ES cells formed normally in vivo, and ␤ 1 homozygous mutant myoblasts were shown to be fusion-competent, although differentiation of ␤ 1 -null ES cells into myotubes was delayed (26,27). These data supported the view that early muscle development is regulated mainly by transcription and growth factors (28), although integrins could be one of the downstream targets. The availability of conditional skeletal musclespecific ␤ 1 -integrin knock-out mice, however, has now shed new light on this field of research. Mice lacking ␤ 1 integrin specifically in muscle die immediately after birth with only poorly developed muscle fibers. 2 At first glance, these data are at odds with earlier results, suggesting that ␤ 1 integrins were not required for myogenesis (26,27). However, upon closer examination it is apparent that both models differ significantly. In the conditional ␤ 1 integrin-null mice, all myoblasts lack ␤ 1 integrins, whereas in the chimeric mice, ␤ 1 -deficient cells are interspersed in a mosaic pattern with wild-type cells. The tight contact between wild-type and mutant cells may commit ␤ 1 -deficient myoblasts to differentiation, or alternatively, wild-type cells may secrete soluble factors in a ␤ 1 integrindependent manner. This could be sufficient to induce the genetic program in the neighboring ␤ 1 knock-out cells. Obviously, the conclusions drawn about individual ␣ subunits based on the analysis in chimeric animals have to be reconsidered in view of the impact caused by the environment. Further work will undoubtedly yield important insights into the role of ␤ 1 integrins for skeletal muscle differentiation, and the race is on again to determine whether an individual or a combination of integrins underlies the phenotype observed in ␤ 1 -deficient skeletal muscle.
Accordingly, laminin receptors are thought to play pivotal roles for skeletal muscle function and integrity. All laminin isoforms detected in skeletal muscle are recognized by ␣ 3 ␤ 1 , ␣ 6 ␤ 1 , and ␣ 7 ␤ 1 (35)(36)(37). Yet, only two laminin receptor systems in the mature skeletal muscle have been shown to be present ubiquitously, namely the DGC and ␣ 7 ␤ 1 integrin. Both are thought to connect the extracellular matrix and the cytoskeleton and thus provide the necessary muscle stability during contraction. Mutations in the dystrophin gene cause Duchenne or Becker muscular dystrophies (38), and genes encoding components of the DGC have been shown to be mutated in various forms of muscular dystrophies (39,40). ␣ 7 ␤ 1 integrin is the major if not the exclusive integrin receptor found in adult skeletal muscle. It is highly enriched at MTJs (41) and is localized at the NMJ together with ␣ 3 -and ␣ v -containing integrins (42). There have been some arguments that the ␣ 10 ␤ 1 and ␣ 11 ␤ 1 integrins localize at the MTJ in adult muscle. Further analysis is still required to clarify whether these integrins are expressed by muscle or alternatively by tendon as their ligand specificity would suggest (43).
Structural and Functional Features of ␣ 7 ␤ 1 -The ␣ 7 ␤ 1 subunit was originally identified from myoblasts and melanoma cells as laminin-1-binding (␣1␤1␥1) integrin (44 ,45). The diversity of the ␣ 7 ␤ 1 integrin is further increased due to the occurrence of alternative mRNA splice variants for both the extracellular and cytoplasmic domains. The major variants in the extracellular domain are the mutually exclusive X1 and X2 domains located between domains III and IV of the conserved repeat motifs, common to all integrins (46,47), and the A and B variants for the cytoplasmic domain (46,48). In addition, some minor extracellular variants have been identified (49 -51), and in rat a third cytoplasmic C variant was described (15), which, however, is not found in the human and murine ITGA7 genes (50).
The spatiotemporal expression of the extracellular and the cytoplasmic variants are developmentally regulated. ␣ 7B is expressed in proliferating myoblasts, whereas the A variant is induced only upon terminal differentiation (46,48). Both forms have been immunolocalized to the NMJ and MTJ in the adult muscle, whereas at the sarcolemma the current data are not conclusive. The ratspecific cytoplasmic C variant was postulated to be the exclusive splice form found at the sarcolemma (15). In mouse and human, however, ␣ 7B was detected along the muscle fiber membrane (52)(53)(54) and in accordance with our data seems to be the predominant splice form found at the sarcolemmal membrane. 3 In contrast to the intracellular domains, only the extracellular X2 variant is found in adult skeletal muscle, whereas X1 and X2 are simultaneously expressed throughout muscle development. Initial data suggested that the ␣ 7 subunit is restricted to skeletal and cardiac muscle (55). The ␣ 7B subunit, however, is ubiquitously expressed as it has been detected in smooth muscle, neuronal, and trophoblast cells and a variety of tissues (48, 56 -58). Among the possible splice variants only those containing the cytoplasmic A variant appear therefore to be skeletal muscle-specific.
There is still much to learn about the specificities of the ␣ 7 integrin splice variants. Although integrin cytoplasmic domains are generally accepted to be involved in signaling events, so far no conclusive evidence has been provided that functionally distinguishes the A and the B variants of the ␣ 7 subunit in terms of cell migration, proliferation, or matrix assembly (16,59). Interestingly, however, cell adhesion and ligand-binding studies of transfected cells and recombinant soluble ␣ 7 ␤ 1 suggest that the extracellular X1/X2 variants are not identical in function. Ziober et al. (60) showed that whereas ␣ 7X2 is constitutively active in binding to laminin-1, the X1 variant only becomes adhesive after treatment with an activating ␤ 1 antibody. In line with these data, soluble ␣ 7X2 ␤ 1 bound to laminin-1, whereas in the presence of X1, ␣ 7 ␤ 1 preferentially interacted with laminin-8 (␣4␤1␥1) (37). Both extracellular splice variants, however, showed the same binding activity to laminin-2. As the laminin ␣4 chain is only present during muscle development and up-regulated in muscle regeneration (31,61), it is tempting to speculate that X1-or X2-containing ␣ 7 ␤ 1 might have subtly distinct functions in these processes.
Cytoplasmic variants have also been described for the ␤ 1 sub-3 M. Willem and U. Mayer, unpublished results.

FIG. 1. Schematic presentation of integrin expression in mouse skeletal muscle.
In migrating myoblasts, the ␣ 4 -␣ 7 , ␣ v , and ␤ 1A subunits are present. During secondary myogenesis at late embryonic and early postnatal stages the indicated subunits are present. Note that the ␤ 1A variant is replaced by ␤ 1D . In the adult, only ␣ 7 ␤ 1D is found at the sarcolemma and at MTJs, whereas at NMJs the ␣ 7 chain along with ␣ 3 and ␣ v subunits is present.

Minireview: Role of Integrins in Skeletal Muscle 14588
unit. The major forms represent ␤ 1A and ␤ 1D , whereas ␤ 1B and ␤ 1C are minor forms and only exist in human. ␤ 1A is ubiquitously expressed, with the exception of skeletal and cardiac muscle, where it is replaced by the homologous ␤ 1D variant. In skeletal muscle the ␤ 1D subunit exclusively associates with the ␣ 7 chain. Upon expression in integrin ␤ 1 -deficient cells, ␤ 1D caused multiple changes in the cellular morphology and strengthened the link between the cytoskeleton and the extracellular matrix as compared with ␤ 1A (62), a finding that is in accordance with the strong force transmission occurring in skeletal muscle. Further analysis suggested ␤ 1D to be involved in the inhibition of myoblast proliferation (63). Mice with a replacement of ␤ 1D by ␤ 1A , however, showed normal muscle function and histology, and only a mild disturbance of cardiac function was found (64). The significance of ␣ 7 in association with ␤ 1D remains therefore open at present.

Integrins in Skeletal Muscle Diseases
Muscular dystrophies are a group of heterogeneous inherited muscle-wasting diseases. Linked genes identified to date are highly diverse and comprise extracellular matrix proteins, transmembrane and associated proteins, proteases, enzymes, and cytoplasmic and nuclear membrane proteins (40,65,66). All interfere at certain stages with normal muscle structure and function, finally leading to similar pathological changes. The currently best accepted model underlying the etiology of the disease predicts that any disturbance of the molecular link between the extracellular matrix and the cytoskeleton results in impaired muscle stability. This theory arose when the first loss-of-function mutations in dystrophin were identified as the cause of DMD (66). Dystrophin binds to F-actin and with its C-terminal domain to ␤-dystroglycan, which in turn interacts with the laminin-binding ␣-dystroglycan (67). Furthermore, mutations within the laminin ␣2 chain result in severe forms of congenital muscular dystrophy (CMD) both in human and the corresponding mouse model dy/dy (68,69).
Muscular Dystrophies Caused by Integrins-Integrins provide a similar link between the cytoskeleton and the extracellular matrix as the DGC and were therefore strong candidate genes for unclassified forms of muscular dystrophies. Indeed, mice carrying an inactivated ␣ 7 integrin gene were shown to develop a mild but progressive muscular dystrophy soon after birth (24). The histological features characteristic for muscular dystrophies were restricted to the deeply located soleus muscle and the diaphragm, whereas the muscle fibers in all other muscles were unaffected. In the meantime, human patients have been identified with a primary integrin ␣ 7 deficiency, resulting from a 21-bp insertion due to a splice site mutation or deletions leading to frameshifts (53). The disease was apparent from birth with delayed motor milestones in the following years of age and was accordingly classified as congenital myopathy. Interestingly, no mutations in the coding region could be identified in one patient with a lack of ␣ 7 integrin staining and low levels of transcript (53). This implies a mutation in the promoter region or in a regulatory element of the gene, and its identification may yield valuable insight into the transcriptional regulation of the ITGA7 gene. The phenotypes observed in mouse and human are highly similar. Neither myofiber necrosis nor muscle fiber regeneration was predominant, and human muscle biopsies revealed only a mild fiber size variation. The staining pattern for the laminin ␣2 chain and components of the DGC were unaltered (24,53). The mild histopathology is in contrast to the severity of the disease in human species. The integrin ␣ 7 -deficient mice, however, might in part explain the etiology of the disease, as muscles are readily accessible. MTJs, the primary site of force transmission between the muscle and the tendon, were severely destructed in all muscles in these mice. The majority had lost their digit-like extensions, and the sarcomer was retracted from the muscle membrane, suggesting an impairment of function of the MTJ (24,70). A similar role of integrins for muscle integrity has been reported for two Drosophila PS integrin mutations, myospheroid (␤PS) and inflated (␣PS2). In both cases the muscles appeared to develop normally with correctly formed attachment sites, but on contraction the sarcomers detached from the cell membrane (71,72). Based on these animal models, it is highly likely that the same consequences occur in human patients and might explain the muscle weakness despite the rather mild myofiber damage.
So far no other human skeletal muscle diseases have been linked to mutations in integrin genes. Yet, another candidate gene for an unclassified CMD is the integrin ␣ 5 subunit. The analysis of integrin ␣ 5 Ϫ/Ϫ chimeric mice with a high contribution of ␣ 5 homozygous mutant cells in skeletal muscle revealed the characteristics for muscular dystrophies (20). Intriguingly, these changes were already evident at late embryonic stages and suggest that ␣ 5 ␤ 1 is more important than ␣ 7 ␤ 1 during muscle maturation. It is also worthwhile to note that ␣ 5 ␤ 1 is a classical fibronectin receptor favoring the idea that a fine balance between fibronectin and laminin-rich matrices and their receptors is critical for muscle integrity at late embryonic and early postnatal stages. As integrin ␣ 5 deficiency leads to embryonic lethality at midgestation due to mesodermal defects in mice (73), it will be interesting whether mutations in the ITGA5 gene, which interfere with ligand binding or result in a hypomorph due to promoter/enhancer mutations, can be identified. Secondary Reduction of ␣ 7 ␤ 1 in Muscular Dystrophies-The demonstration that mutations affecting ␣ 7 ␤ 1 caused muscular dystrophy led to a broad investigation about its expression pattern in other myopathies of known and unknown origin. At most, only minor changes in expression have been noted with the exception of severe forms of CMD linked to mutations in the laminin ␣2 chain. Concomitant with the loss or defective expression of the laminin-2 and -4 isoforms in the muscle basement membrane, several independent investigations of diseased human and murine muscle demonstrated a reduction for the integrin ␣ 7 and ␤ 1D subunits as a secondary consequence, whereas components of the DGC appeared unaffected (54,74,75). The same was also observed in Fukuyama CMD and muscle-eye-brain disease, for which a secondary reduction of the laminin ␣2 chain was reported (75). These findings underline that laminin ␣2-containing isoforms are indeed the major ligands for ␣ 7 ␤ 1 integrin at the muscle membrane.
Relationship of ␣ 7 ␤ 1 Integrin and the DGC-The critical presence of ␣ 7 ␤ 1 integrin and the DGC to maintain muscle integrity raised the question as to whether both can compensate for each other. A higher staining intensity for ␣ 7 ␤ 1 in sections of DMD patients and the corresponding mouse model, mdx, has been observed (54,74,75), although it is still controversial whether this is due to increased transcription (53,74). In contrast, components of the DGC were unchanged in ␣ 7 deficiency (24,53,54). Further, both complexes differ in that ␣ 7 ␤ 1 integrin and the DGC are essential for MTJ stability and the lateral integrity of the muscle fiber, respectively (70,76). This supports the notion that both are independently controlled receptor systems and that the underlying mechanisms leading to the disease are different. The presence of either complex, however, is essential, as double-mutant mice lacking dystrophin and ␣ 7 ␤ 1 develop a severe dystrophy and die within 4 weeks after birth. 3 ␣ 7 ␤ 1 as a Therapeutic Tool for DMD-DMD is a severe X-linked disease and affects about 1 in 3000 males. Over the past years a major effort has been put into therapeutic strategies to ameliorate FIG. 2. Integrins involved in muscular dystrophies. The diagram shows ␣ 5 ␤ 1 and ␣ 7 ␤ 1 and their respective ligands. Mutations in both integrins lead to muscular dystrophies in mice and human. How they are connected to the actin cytoskeleton and which signaling cascades they induce in skeletal muscle is still unknown.

Minireview: Role of Integrins in Skeletal
Muscle 14589 the disease. Various approaches tried to restore dystrophin in skeletal muscle by myoblast or virally mediated gene transfer (77). The most promising approach, however, dealt with the constitutive up-regulation of utrophin, a dystrophin homologue, which had been shown to compensate in mdx mice for the loss of dystrophin (66). As ␣ 7 ␤ 1 showed increased staining intensities in DMD patients and mdx mice, Burkin et al. (78) tested the idea that overexpression of the ␣ 7 subunit in dystrophin/utrophin double-mutant mice might restore muscle integrity. Although degeneration, as expressed by the percentage of muscle fibers with centrally located nuclei, was similar in the double mutants and those overexpressing the ␣ 7 subunit, life span increased 3-fold in conjunction with higher mobility upon overexpression of the ␣ 7 subunit (78). The high degree of de-and regeneration supports the view that ␣ 7 ␤ 1 and the DGC complex are distinctive receptors but that increased amounts of ␣ 7 ␤ 1 are capable of stabilizing muscle function and ameliorating the dystrophin-deficient phenotype. More work is needed to investigate this promising approach for DMD therapy.

Conclusions and Perspectives
The discoveries made in recent years emphasize that integrins are indispensable both for muscle development and muscle function in the adult and not redundant to the DGC. Although much progress has been made by identifying muscular dystrophies caused by loss-of-function mutations in ␣ 5 ␤ 1 and ␣ 7 ␤ 1 integrins, our understanding about the underlying mechanisms is still very incomplete, and several key questions remain open (Fig. 2). For example, through which linker proteins are both integrins connected to the actin cytoskeleton? Which signaling cascades are associated with the receptors in skeletal muscle? The answers to these questions will not only elucidate the pathogenesis of the disease but may also uncover new strategies to improve the concept for therapeutic approaches of DMD patients by integrins.