Heterogeneity of Z-band structure within a single muscle sarcomere: implications for sarcomere assembly

The vertebrate striated muscle Z-band connects actin filaments of opposite polarity from adjacent sarcomeres and allows tension to be transmitted along a myofibril during contraction. Z-bands in different muscles have a modular structure formed by layers of alpha-actinin molecules cross-linking actin filaments. Successive layers occur at 19 nm intervals and have 90 degrees rotations between them. 3D reconstruction from electron micrographs show a two-layer "simple" Z-band in fish body fast muscle, a three-layer Z-band in fish fin fast muscle, and a six-layer Z-band in mammalian slow muscle. Related to the number of these layers, longitudinal sections of the Z-band show a number of zigzag connections between the oppositely oriented actin filaments. The number of layers also determines the axial width of the Z-band, which is a useful indicator of fibre type; fast fibres have narrow (approximately 30-50 nm) Z-bands; slow and cardiac fibres have wide (approximately 100-140 nm) Z-bands. Here, we report the first observation of two different Z-band widths within a single sarcomere. By comparison with previous studies, the narrower Z-band comprises three layers. Since the increase in width of the wider Z-band is about 19 nm, we conclude that it comprises four layers. This finding is consistent with a Z-band assembly model involving molecular control mechanisms that can add additional layers of 19 nm periodicity. These multiple Z-band structures suggest that different isoforms of nebulin and titin with a variable number of Z-repeats could be present within a single sarcomere.     

In vitro cleavage of eIF4GI but not eIF4GII by HIV-1 protease and its effects on translation in the rabbit reticulocyte lysate system

The vertebrate muscle Z-band organizes and tethers antiparallel actin filaments in adjacent sarcomeres and hence propagates the tension generated by the actomyosin interaction during muscular contraction. The axial width of the Z-band varies with fibre and muscle type: fast twitch muscles have narrow (approximately 30-50 nm) Z-bands, while slow-twitch and cardiac muscles have wide (approximately 100-140 nm) Z-bands. In electron micrographs of longitudinal sections of fast fibres like those found in fish body white muscle, the Z-band appears as a characteristic zigzag layer of density connecting the mutually offset actin filament arrays in adjacent sarcomeres. Wide Z-bands in slow fibres such as the one studied here (bovine neck muscle) show a stack of three or four zigzag layers. The variable Z-band width incorporating variable numbers of zigzag layers presumably relates to the different mechanical properties of the respective muscles. Three-dimensional reconstructions of Z-bands reveal that individual zigzag layers are often composed of more than one set of protein bridges, called Z-links, probably alpha-actinin, between oppositely oriented actin filaments. Fast muscle Z-bands comprise two or three layers of Z-links. Here we have applied Fourier reconstruction methods to obtain clear three-dimensional density maps of the Z-bands in beef muscle. The bovine slow muscle investigated here reveals a Z-band comprising six sets of Z-links, which, due to their shape and the way their projected densities overlap, appear in longitudinal sections as either three or four zigzag layers, depending on the lattice view. There has been great interest recently in the suggestion that Z-band variability with fibre type may be due to differences in the repetitive region (tandem Z-repeats) in the Z-band part of titin (also called connectin). We discuss this in the context of our results and present a systematic classification of Z-band types according to the numbers of Z-links and titin Z-repeats.     

Three-dimensional structure of a vertebrate muscle Z-band: implications for titin and alpha-actinin binding

The Z-band in vertebrate striated muscles, mainly comprising actin filaments, alpha-actinin, and titin, serves to organise the antiparallel actin filament arrays in adjacent sarcomeres and to transmit tension between sarcomeres during activation. Different Z-band thicknesses, formed from different numbers of zigzag crosslinking layers and found in different fibre types, are thought to be associated with the number of repetitive N-terminal sequence domains of titin. In order to understand myofibril formation it is necessary to correlate the ultrastructures and sequences of the actin filaments, titin, and alpha-actinin in characteristic Z-bands. Here electron micrographs of the intermediate width, basketweave Z-band of plaice fin muscle have been subject to a novel 3D reconstruction process. The reconstruction shows that antiparallel actin filaments overlap in the Z-band by about 22-25 nm. There are three levels of Z-links (probably alpha-actinin) in which at each level two nearly diametrically opposed links join an actin filament to two of its antiparallel neighbours. One set of links is centrally located in the Z-band and there are flanking levels orthogonal to this. A 3D model of the observed structure shows how Z-bands of different widths may be formed and it provides insights into the structural arrangements of titin and alpha-actinin in the Z-band. The model shows that the two observed symmetries in different Z-bands, c2 and p12(1), may be attributed respectively to whether the number of Z-link levels is odd or even.     

The three-dimensional structure of the nemaline rod Z-band

In nemaline myopathy and some cardiac muscles, the Z-band becomes greatly enlarged and contains multiple layers of a zigzag structure similar to that seen in normal muscle. Because of the additional periodicity in the direction of the filament axis, these structures are particularly favorable for three-dimensional analysis since it becomes possible to average the data in all three dimensions and thus improve the reliability of the reconstruction. Individual views of the structure corresponding to tilted longitudinal and transverse sections were combined by matching the phases of common reflections. Examination of the tilted views strongly suggested that to the available resolution, the structure possesses fourfold screw symmetry along the actin filament axes. This symmetry could be used both in establishing the correct alignment for the combination of individual tilted views and to generate additional views not readily accessible in a single tilt series. The reconstruction shows actin filaments from one sarcomere surrounded by an array of four actin filaments with opposite polarity from the adjacent sacormere. The actin filaments show a right- handed twist and are connected by a structure that links adjacent filaments with the same polarity at the same axial level, then runs parallel to the filaments, and finally forms a link between two actin filaments whose polarity is opposite to that of the first pair. The connecting structure is probably composed of alpha-actinin which is located in Z-bands and cross-links actin filaments. The connecting structure may consist of two alpha-actinin molecules linking actin filaments of opposite polarity.     

A structural characterization of the interactions between titin Z-repeats and the alpha-actinin C-terminal domain

Titin and alpha-actinin, two modular muscle proteins, are with actin the major components of the Z-band in vertebrate striated muscles where they serve to organize the antiparallel actin filament arrays in adjacent sarcomeres and to transmit tension between sarcomeres during activation. Interactions between titin and alpha-actinin have been mainly localized in a 45-amino acid multiple motif (Z-repeat) in the N-terminal region of titin and the C-terminal region of alpha-actinin. In this study, we provide the first quantitative characterization of alpha-actinin-Z-repeat recognition and dissect the interaction to its minimal units. Different complementary techniques, such as circular dichroism, calorimetry, and nuclear magnetic spectroscopy, were used. Two overlapping alpha-actinin constructs (Act-EF34 and Act-EF1234) containing two and four EF-hand motifs, respectively, were produced, and their folding properties were examined. Complex formation of Act-EF34 and Act-EF1234 with single- and double-Z-repeat constructs was studied. Act-EF34 was shown quantitatively to be necessary and sufficient for binding to Z-repeats, excluding the presence of additional high-affinity binding sites in the remaining part of the domain. The binding affinities of the different Z-repeats for Act-EF34 range from micromolar to millimolar values. The strongest of these interactions are comparable to those observed in troponin C-troponin I complexes. The binding affinities for Act-EF34 are maximal for Zr1 and Zr7, the two highly homologous sequences present in all muscle isoforms. No cooperative or additional contributions to the interaction were observed for Z-repeat double constructs. These findings have direct relevance for evaluating current models of Z-disk assembly.     

Dynamics of Z-band based proteins in developing skeletal muscle cells

During myofibril formation, Z-bodies, small complexes of alpha-actinin and associated proteins, grow in size, fuse and align to produce Z-bands. To determine if there were changes in protein dynamics during the assembly process, Fluorescence Recovery after Photobleaching was used to measure the exchange of Z-body and Z-band proteins with cytoplasmic pools in cultures of quail myotubes. Myotubes were transfected with plasmids encoding Yellow, Green, or Cyan Fluorescent Protein linked to the Z-band proteins: actin, alpha-actinin, cypher, FATZ, myotilin, and telethonin. Each Z-band protein showed a characteristic recovery rate and mobility. All except telethonin were localized in both Z-bodies and Z-bands. Proteins that were present both early in development in Z-bodies and later in Z-bands had faster exchange rates in Z-bodies. These results suggest that during myofibrillogenesis, molecular interactions develop between the Z-band proteins that decrease their mobility and increase the stability of the Z-bands. A truncated construct of alpha-actinin, which localized in Z-bands in myotubes and exhibited a very low rate of exchange, led to disruption of myofibrils, suggesting the importance of dynamic, intact alpha-actinin molecules for the formation and maintenance of Z-bands. Our experiments reveal the Z-band to be a much more dynamic structure than its appearance in electron micrographs of cross-striated muscle cells might suggest.     

The interaction of titin and alpha-actinin is controlled by a phospholipid-regulated intramolecular pseudoligand mechanism

The assembly of stable cytoskeletal structures from dynamically recycled molecules requires developmental and spatial regulation of protein interactions. In muscle, titin acts as a molecular ruler organizing the actin cytoskeleton via interactions with many sarcomeric proteins, including the crosslinking protein alpha-actinin. An interaction between the C-terminal domain of alpha-actinin and titin Z-repeat motifs targets alpha-actinin to the Z-disk. Here we investigate the cellular regulation of this interaction. alpha-actinin is a rod shaped head-to-tail homodimer. In contrast to C-terminal fragments, full-length alpha-actinin does not bind Z-repeats. We identify a 30-residue Z-repeat homologous sequence between the actin-binding and rod regions of alpha-actinin that binds the C-terminal domain with nanomolar affinity. Thus, Z-repeat binding is prevented by this 'pseudoligand' interaction between the subunits of the alpha-actinin dimer. This autoinhibition is relieved upon binding of the Z-disk lipid phosphatidylinositol-bisphosphate to the actin-binding domain. We suggest that this novel mechanism is relevant to control the site-specific interactions of alpha-actinin during sarcomere assembly and turnover. The intramolecular contacts defined here also constrain a structural model for intrasterical regulation of all alpha-actinin isoforms.     

Binding of alpha-actinin to titin: implications for Z-disk assembly

Titin is an exceptionally large protein (M.Wt. approximately 3 MDa) that spans half the sarcomere in muscle, from the Z-disk to the M-line. In the Z-disk, it interacts with alpha-actinin homodimers that are a principal component of the Z-filaments linking actin filaments. The interaction between titin and alpha-actinin involves repeating approximately 45 amino acid sequences (Z-repeats) near the N-terminus of titin and the C-lobe of the C-terminal calmodulin-like domain of alpha-actinin. The conformation of Z-repeat 7 (ZR7) of titin when complexed with the 73-amino acid C-terminal portion of alpha-actinin (EF34) was studied by heteronuclear NMR spectroscopy using (15)N-labeling of ZR7 and found to be helical over a stretch of 18 residues. Complex formation resulted in the protection of one site of preferential cleavage of EF34 at Phe14-Leu17, as determined by limited proteolysis experiments coupled to mass spectrometry measurements. Intermolecular NOEs show Val16 of ZR7 to be positioned close in space to the backbone of EF34 around Phe14. These observations suggest that the mode of binding of ZR7 to EF34 is similar to that of troponin I to troponin C and of peptide C20W to calmodulin. These complexes would appear to represent a general alternative binding mode of calmodulin-like domains to target peptides.     

Tracking changes in Z-band organization during myofibrillogenesis with FRET imaging

There are a large number of proteins associated with Z-bands in myofibrils, but the precise arrangements of most of these proteins in Z-bands are largely unknown. Even less is known about how these arrangements change during myofibrillogenesis. We have begun to address this issue using Sensitized Emission Fluorescence Resonance Energy Transfer (SE-FRET) microscopy. Cultured skeletal muscle cells from quail embryos were transfected to express fusions of alpha-actinin, FATZ, myotilin, or telethonin with cyan and yellow fluorescent proteins in various pair wise combinations. FATZ and myotilin were selected because previous biochemical studies have suggested that they bind to alpha-actinin, the major protein of the Z-band. Telethonin was selected for its reported ability to bind FATZ. Statistical analysis of data from FRET imaging studies yield results that are in agreement with published biochemical data suggesting that FATZ and myotilin bind to alpha-actinin near its C-terminus as well as to each other and that a region near the amino-terminus of FATZ is responsible for its interaction with telethonin. In addition, our analysis has revealed changes in the arrangement of alpha-actinin and FATZ that take place during the transition as the z-bodies of premyofibrils fuse to form the Z-bands of mature myofibrils. There was no evidence for a change in the arrangement of myotilin as z-bodies transformed into Z-bands. Myotilin is one Z-band protein that does not exhibit decreased dynamics as z-bodies fuse to form Z-bands. These FRET results from living cells support a stepwise model for the assembly of myofibrils.     

Three-dimensional reconstruction of a simple Z-band in fish muscle

The three-dimensional structure of the Z-band in fish white muscle has been investigated by electron microscopy. This Z-band is described as simple, since in longitudinal sections it has the appearance of a single zigzag pattern connecting the ends of actin filaments of opposite polarity from adjacent sarcomeres. The reconstruction shows two pairs of links, the Z-links, between one actin filament and the facing four actin filaments in the adjacent sarcomere. The members of each pair have nearly diametrically opposed origins. In relation to one actin filament, one pair of links appears to bind along the final 10 nm of the actin filament (proximal site) and the other pair binds along a region extending from 5 to 20 nm from the filament end (distal site). Between one pair and the other, there is a rotation of approximately 80 degrees round the filament axis. A Z-link with a proximal site at the end of one actin filament attaches at a distal site on the oppositely oriented actin filaments of the facing sarcomere and vice versa. The length of each Z-link is consistent with the length of an alpha-actinin molecule. An additional set of links located 10-15 nm from the center of the Z-band occurs between actin filaments of the same polarity. These polar links connect the actin filaments along the same direction on each side of the Z-band. The three-dimensional structure appears to have twofold screw symmetry about the central plane of the Z-band. Only approximate twofold rotational symmetry is observed in directions parallel to the actin filaments. Previous models of the Z-band in which four identical and rotationally symmetrical links emanate from the end of one actin filament and span across to the ends of four actin filaments in the adjacent sarcomere are therefore incorrect.     

The Z-band lattice in skeletal muscle in rigor

Previous work with tetanized and relaxed muscle has shown a correlation between active tension and the structure of the Z-band. This suggests that there is a correlation between the cross-bridge binding in the A-band and the structure of the Z-band. Using electron microscopy and optical diffraction we have examined this correlation in glycerinated muscle in rigor and in unstimulated intact muscle. We have found that the Z-bands of muscles in rigor always show the basketweave form, while those of the unstimulated muscles always show the small square form. The basketweave form found in rigor muscles is similar in form and dimension to that found in tetanized muscle. Thus it appears that the small square form of the Z-band is found in physiological states with little cross-bridge binding and the basketweave form is found in states with a high degree of cross-bridge binding.     

Electron microscopy and x-ray diffraction evidence for two Z-band structural states

In vertebrate muscles, Z-bands connect adjacent sarcomeres, incorporate several cell signaling proteins, and may act as strain sensors. Previous electron microscopy (EM) showed Z-bands reversibly switch between a relaxed, “small-square” structure, and an active, “basketweave” structure, but the mechanism of this transition is unknown. Here, we found the ratio of small-square to basketweave in relaxed rabbit psoas muscle varied with temperature, osmotic pressure, or ionic strength, independent of activation. By EM, the A-band and both Z-band lattice spacings varied with temperature and pressure, not ionic strength; however, the basketweave spacing was consistently 10% larger than small-square. We next sought evidence for the two Z-band structures in unfixed muscles using x-ray diffraction, which indicated two Z-reflections whose intensity ratios and spacings correspond closely to the EM measurements for small-square and basketweave if the EM spacings are adjusted for 20% shrinkage due to EM processing. We conclude that the two Z-reflections arise from the small-square and basketweave forms of the Z-band as seen by EM. Regarding the mechanism of transition during activation, the effects of Ca²⁺ in the presence of force inhibitors suggested that the interconversion of Z-band forms was correlated with tropomyosin movement on actin.     

Characterization of components of Z-bands in the fibrillar flight muscle of Drosophila melanogaster

Twelve monoclonal antibodies have been raised against proteins in preparations of Z-disks isolated from Drosophila melanogaster flight muscle. The monoclonal antibodies that recognized Z-band components were identified by immunofluorescence microscopy of flight muscle myofibrils. These antibodies have identified three Z-disk antigens on immunoblots of myofibrillar proteins. Monoclonal antibodies alpha:1-4 recognize a 90-100-kD protein which we identify as alpha-actinin on the basis of cross-reactivity with antibodies raised against honeybee and vertebrate alpha-actinins. Monoclonal antibodies P:1-4 bind to the high molecular mass protein, projectin, a component of connecting filaments that link the ends of thick filaments to the Z-band in insect asynchronous flight muscles. The anti-projectin antibodies also stain synchronous muscle, but, surprisingly, the epitopes here are within the A-bands, not between the A- and Z-bands, as in flight muscle. Monoclonal antibodies Z(210):1-4 recognize a 210-kD protein that has not been previously shown to be a Z-band structural component. A fourth antigen, resolved as a doublet (approximately 400/600 kD) on immunoblots of Drosophila fibrillar proteins, is detected by a cross reacting antibody, Z(400):2, raised against a protein in isolated honeybee Z-disks. On Lowicryl sections of asynchronous flight muscle, indirect immunogold staining has localized alpha-actinin and the 210-kD protein throughout the matrix of the Z-band, projectin between the Z- and A-bands, and the 400/600-kD components at the I-band/Z-band junction. Drosophila alpha-actinin, projectin, and the 400/600-kD components share some antigenic determinants with corresponding honeybee proteins, but no honeybee protein interacts with any of the Z(210) antibodies.     

Analysis of myofibrillar structure and assembly using fluorescently labeled contractile proteins

To study how contractile proteins become organized into sarcomeric units in striated muscle, we have exposed glycerinated myofibrils to fluorescently labeled actin, alpha-actinin, and tropomyosin. In this in vitro system, alpha-actinin bound to the Z-bands and the binding could not be saturated by prior addition of excess unlabeled alpha-actinin. Conditions known to prevent self-association of alpha-actinin, however, blocked the binding of fluorescently labeled alpha-actinin to Z-bands. When tropomyosin was removed from the myofibrils, alpha-actinin then added to the thin filaments as well as the Z-bands. Actin bound in a doublet pattern to the regions of the myosin filaments where there were free cross-bridges i.e., in that part of the A-band free of interdigitating native thin filaments but not in the center of the A- band which lacks cross-bridges. In the presence of 0.1-0.2 mM ATP, no actin binding occurred. When unlabeled alpha-actinin was added first to myofibrils and then labeled actin was added fluorescence occurred not in a doublet pattern but along the entire length of the myofibril. Tropomyosin did not bind to myofibrils unless the existing tropomyosin was first removed, in which case it added to the thin filaments in the l-band. Tropomyosin did bind, however, to the exogenously added tropomyosin-free actin that localizes as a doublet in the A-band. These results indicate that the alpha-actinin present in Z-bands of myofibrils is fully complexed with actin, but can bind exogenous alpha- actinin and, if actin is added subsequently, the exogenous alpha- actinin in the Z-band will bind the newly formed fluorescent actin filaments. Myofibrillar actin filaments did not increase in length when G-actin was present under polymerizing conditions, nor did they bind any added tropomyosin. These observations are discussed in terms of the structure and in vivo assembly of myofibrils.     

Molecular structure of the sarcomeric Z-disk: two types of titin interactions lead to an asymmetrical sorting of alpha-actinin.

The sarcomeric Z-disk, the anchoring plane of thin (actin) filaments, links titin (also called connectin) and actin filaments from opposing sarcomere halves in a lattice connected by alpha-actinin. We demonstrate by protein interaction analysis that two types of titin interactions are involved in the assembly of alpha-actinin into the Z-disk. Titin interacts via a single binding site with the two central spectrin-like repeats of the outermost pair of alpha-actinin molecules. In the central Z-disk, titin can interact with multiple alpha-actinin molecules via their C-terminal domains. These interactions allow the assembly of a ternary complex of titin, actin and alpha-actinin in vitro, and are expected to constrain the path of titin in the Z-disk. In thick skeletal muscle Z-disks, titin filaments cross over the Z-disk centre by approximately 30 nm, suggesting that their alpha-actinin-binding sites overlap in an antiparallel fashion. The combination of our biochemical and ultrastructural data now allows a molecular model of the sarcomeric Z-disk, where overlapping titin filaments and their interactions with the alpha-actinin rod and C-terminal domain can account for the essential ultrastructural features.     

Amorphin is phosphorylase; phosphorylase is an alpha-actinin-binding protein

In a study of myofibrillar proteins, Chowrashi and Pepe [1982: J. Cell Biol. 94:565-573] reported the isolation of a new, 85-kD Z-band protein that they named amorphin. We report that partial sequences of purified amorphin protein indicate that amorphin is identical to phosphorylase, an enzyme important in the metabolism of glycogen. Anti-amorphin antibodies also reacted with purified chicken and rabbit phosphorylase. To explore the basis for phosphorylase's (amorphin's) localization in the Z-bands of skeletal muscles, we reacted biotinylated alpha-actinin with purified amorphin and with purified phosphorylase and found that alpha-actinin bound to each. Radioimmune assays also indicated that phosphorylase (amorphin) bound to alpha-actinin, and, with lower affinity, to F-actin. Negative staining of actin filaments demonstrated that alpha-actinin mediates the binding of phosphorylase to actin filaments. There are several glycolytic enzymes that bind actin (e.g., aldolase, phosphofructokinase, and pyruvate kinase), but phosphorylase is the first one demonstrated to bind alpha-actinin. Localization of phosphorylase in live cells was assessed by transfecting cultures of quail embryonic myotubes with plasmids expressing phosphorylase fused to Green Fluorescent Protein (GFP). This resulted in targeting of the fusion protein to Z-bands accompanied by a diffuse pattern in the cytoplasm.     

Ca2+-independent binding of an EF-hand domain to a novel motif in the alpha-actinin-titin complex

The interaction between alpha-actinin and titin, two modular muscle proteins, is essential for sarcomere assembly. We have solved the solution structure of a complex between the calcium-insensitive C-terminal EF-hand domain of alpha-actinin-2 and the seventh Z-repeat of titin. The structure of the complex is in a semi-open conformation and closely resembles that of myosin light chains in their complexes with heavy chain IQ motifs. However, no IQ motif is present in the Z-repeat, suggesting that the semi-open conformation is a general structural solution for calcium-independent recognition of EF-hand domains.     

Molecular identification and localization of cellular titin, a novel titin isoform in the fibroblast stress fiber

We previously discovered a large titin-like protein-c-titin-in chicken epithelial brush border and human blood platelet extracts that binds alpha-actinin and organizes arrays of myosin II bipolar filaments in vitro. RT-PCR analysis of total RNA from human megakaryoblastic (CHRF-288-11) and mouse fibroblast (3T3) nonmuscle cells reveal sequences identical to known titin gene exon sequences that encode parts of the Z-line, I-band, PEVK domain, A-band, and M-line regions of striated muscle titins. In the nonmuscle cells, these sequences are differentially spliced in patterns not reported for any striated muscle titin isoform. Rabbit polyclonal antibodies raised against expressed protein fragments encoded by the Z-repeat and kinase domain regions react with the c-titin band in Western blot analysis of platelet extracts and immunoprecipitate c-titin in whole platelet extracts. Immunofluorescent localization demonstrates that the majority of the c-titin colocalizes with alpha-actinin and actin in 3T3 and Indian Muntjac deer skin fibroblast stress fibers. Our results suggest that differential expression of titin gene exons in nonmuscle cells yields multiple novel isoforms of the protein c-titin that are associated with the actin stress fiber structures.     

Myogenic stage, sarcomere length, and protease activity modulate localization of muscle-specific calpain

p94/calpain 3 is a Ca(2+)-binding intracellular protease predominantly expressed in skeletal muscles. p94 binds to the N2A and M-line regions of connectin/titin and localizes in the Z-bands. Genetic evidence showing that compromised p94 proteolytic activity leads to muscular dystrophy (limb-girdle muscular dystrophy type 2A) indicates the importance of p94 function in myofibrils. Here we show that a series of p94 splice variants is expressed immediately after muscle differentiation and differentially change localization during myofibrillogenesis. We found that the endogenous N-terminal (but not C-terminal) domain of p94 was not only localized in the Z-bands but also directly bound to sarcomeric alpha-actinin. These data suggest the incorporation of proteolytic N-terminal fragments of p94 into the Z-bands. In myofibrils localization of exogenously expressed p94 shifted from the M-line to N2A as the sarcomere lengthens beyond approximately 2.6 and 2.8 microm for wild-type and proteaseinactive p94, respectively. These data demonstrate for the first time that p94 proteolytic activity is involved in responses to muscle conditions, which may explain why p94 inactivation causes limb-girdle muscular dystrophy.