Potential m-Calpain Substrates during Myoblast Fusion

Many studies have demonstrated that m-calpain was implicated in cell membrane reorganization-related phenomena during fusion via a regulation by calpastatin, the specific Ca²⁺-dependent proteolytic inhibitor. However, the real biological role of this protease is unclear because many targeted proteins are still unknown. Using different digestion experiments we have demonstrated that desmin, vimentin, talin, and fibronectin represent very good substrates for this proteinase capable of cleaving them in fragments which are immediately degraded by other enzymatic systems. Concerning intermediate filaments, we showed that during the phenomenon of fusion, the amount of desmin was significantly reduced while the concentration of vimentin presented a steady level. On the other hand, we have conducted biological assays on cultured myoblasts supplemented by exogenous factors such as calpain inhibitors or antisense oligonucleotides capable of stimulating or inhibiting m-calpain activity. The effect of such factors on fusion and concomitantly on the targeted substrates was analyzed and quantified. When m-calpain activity and myoblast fusion were prevented by addition of calpain inhibitors entering the cells, the amounts of desmin, talin, and fibronectin were increased, whereas the amount of vimentin was unchanged. Using antisense strategy, similar results were obtained. In addition, when the phenomenon of fusion was enhanced by preventing calpastatin synthesis, the amounts of desmin, talin, and fibronectin were significantly reduced. Taken together, these results support the hypothesis that m-calpain is involved in myoblast fusion by cleaving certain proteins identified here. This cleavage could modify membrane and cytoskeleton organization for the myoblasts to fuse.     

M-calpain levels increase during fusion of myoblasts in the mutant muscular dysgenesis (mdg) mouse

Previous studies have led to the hypothesis of a possible role for the calcium-dependent neutral protease m-calpain in myoblast fusion in culture. To evaluate this hypothesis, we chose as our model, the "muscular dysgenesis" mouse (mdg), which presents in vivo and in vitro characteristics of an elevated process of fusion (Yao and Essien, 1975; Dussartre, 1993; Ashby et al., 1993, Joffroy et al., 1999). The aim of this study was to demonstrate using myoblast cell lines and muscle biopsies from this mdg mutant, that the amount of m-calpain increases significantly as multinucleated myotubes are formed. Using immunoblot analysis, it was shown that the m-calpain concentration in a dysgenic cell line (GLT) increased 3-fold compared to what it was upon the introduction of the differentiation medium. On the other hand, in a normal cell line (NLT), the concentration of m-calpain did not vary significantly. Thus, when the transition from myoblasts to myotubes was slow, and the absolute level of fusion was reduced, as in the NLT cell line, the level of m-calpain was stable. In contrast, when the process of fusion was precocious and fast, and the level of fusion was elevated, such as in the GLT cell line, the concentration of m-calpain increased during fusion. Moreover, when myoblast fusion was prevented by the addition of calpain inhibitor II, the process was reduced by approximately 93%. Taking into account these observations, it is clear from our data that the muscular dysgenesis mouse provides a relevant model to study myoblast fusion and that m-calpain is involved in this process.     

Role of Ca2+ and Ca2+-activated protease in myoblast fusion

In this report, we have examined the effects of a calcium chelator, EGTA, and a calcium ionophore, A23187, on fusion of a cloned muscle cell line, L6. Our results confirm that EGTA essentially blocks all myoblast fusion because the lateral alignment of presumptive myoblasts cannot occur in the absence of extracellular calcium. A23187, however, promotes the precocious fusion of myoblasts, apparently by facilitating Ca2+ transport into myoblasts. We have also demonstrated that a Ca2+-activated protease, CAF (mM), appears to relocate in response to the Ca2+ flux, changing from a random, dispersed distribution in proliferative myoblasts to a predominantly peripheral distribution in prefusion myoblasts. Coincident with the mM CAF relocation is an altered distribution of a surface glycoprotein, fibronectin. Extracellular fibronectin is seen in abundance in proliferating myoblasts, but is essentially absent from the surface of fusing myoblasts. We suggest that mM CAF when activated by Ca2+ influx may act to promote the release of fibronectin from the myoblast cell surface, thus providing a mechanism by which the membrane of the fusing myoblast may be rearranged to accommodate fusion.     

Calpain and myogenesis: development of a convenient cell culture model

Previous studies have led us to hypothesize that m-calpain plays a pivotal role in myoblast fusion through its involvement in cell membrane and cytoskeleton component reorganization. To support this hypothesis, a convenient and simple myoblast culture model using frozen embryonic myoblasts was developed, which resolved a number of problems inherent to cell primary culture. Biological assays on cultured myoblasts using different media to define the characteristics of the fusion process were first conducted. Proteinase was detectable before the initiation of the fusion process and was closely correlated to the phenomenon of fusion under each culture condition studied. In addition, the study of calpastatin showed that the initiation of fusion does not require a decrease in the level of this endogenous inhibitor of calpains and also confirmed that calpastatin may be implicated in the determination of the end of fusion. On the other hand, analysis of the evolution of myogenic factors revealed that myogenins, MyoD and Myf5, increase very significantly during the formation of multinucleated myotubes. Moreover, the antisense technique against myogenin is capable of preventing the process of fusion by 50%, confirming the pivotal role of this factor in the early stages of differentiation. The possible role of myogenic regulator factors on m-calpain gene expression is discussed.     

The glycoprotein-processing inhibitors bromoconduritol and N-methyl-1-deoxynojirimycin alter the adhesion phenotype of skeletal myoblasts

Treatment of chick myoblasts with the glucosidase inhibitors bromoconduritol (BCD) or N-methyl-1-deoxynojirimycin (MDJN), but not the mannosidase I inhibitor 1-deoxymannojirimycin (ManDJN), decreased their rate of adhesion to fibronectin and laminin and increased their rate of adhesion to collagen types I and IV. The adhesion of chick myoblasts to fibronectin, collagen type IV, and laminin was predominantly mediated by beta 1-type integrin(s) as judged by inhibition of adhesion with the beta 1-specific monoclonal antibody JG22. Collagen binding in inhibitor-treated cells remained JG22-sensitive suggesting the inhibitors promote increased activity of a beta 1-type collagen-selective integrin. The effects of BCD, MDJN, and ManDJN on myoblast beta 1-integrin detectable at the myoblast cell surface with JG22 antibody correlated well with their effects on adhesion to fibronectin and laminin, and paralleled the previously reported effects of these agents on myogenesis. Interaction of integrin with the extracellular matrix appears to be required for myoblast terminal differentiation. We found that Mn2+ ions increased the adhesion of myoblasts to extracellular matrix proteins and antagonized the effect of BCD and MDJN on myoblast differentiation, supporting a role for cell-matrix interactions in myogenesis. Inhibition of myogenesis by BCD or MDJN was not reversed by growth under low serum conditions, suggesting these agents do not act by maintaining myoblast in a proliferative state.     

Interactions between Skeletal Muscle Myoblasts and their Extracellular Matrix Revealed by a Serum Free Culture System

Decellularisation of skeletal muscle provides a system to study the interactions of myoblasts with muscle extracellular matrix (ECM). This study describes the efficient decellularisation of quadriceps muscle with the retention of matrix components and the use of this matrix for myoblast proliferation and differentiation under serum free culture conditions. Three decellularisation approaches were examined; the most effective was phospholipase A2 treatment, which removed cellular material while maximizing the retention of ECM components. Decellularised muscle matrices were then solubilized and used as substrates for C2C12 mouse myoblast serum free cultures. The muscle matrix supported myoblast proliferation and differentiation equally as well as collagen and fibronectin. Immunofluorescence analyses revealed that myoblasts seeded on muscle matrix and fibronectin differentiated to form long, well-aligned myotubes, while myoblasts seeded on collagen were less organized. qPCR analyses showed a time dependent increase in genes involved in skeletal muscle differentiation and suggested that muscle-derived matrix may stimulate an increased rate of differentiation compared to collagen and fibronectin. Decellularized whole muscle three-dimensional scaffolds also supported cell adhesion and spreading, with myoblasts aligning along specific tracts of matrix proteins within the scaffolds. Thus, under serum free conditions, intact acellular muscle matrices provided cues to direct myoblast adhesion and migration. In addition, myoblasts were shown to rapidly secrete and organise their own matrix glycoproteins to create a localized ECM microenvironment. This serum free culture system has revealed that the correct muscle ECM facilitates more rapid cell organisation and differentiation than single matrix glycoprotein substrates.     

Role of Ca and Ca-activated protease in myoblast fusion

In this report, we have examined the effects of a calcium chelator, EGTA, and a calcium ionophore, A23187, on fusion of a cloned muscle cell line, L₆. Our results confirm that EGTA essentially blocks all myoblast fusion because the lateral alignment of presumptive myoblasts cannot occur in the absence of extracellular calcium. A23187, however, promotes the precocious fusion of myoblasts, apparently by facilitating Ca²⁺ transport into myoblasts. We have also demonstrated that a Ca²⁺-activated protease, CAP (mM), appears to relocate in response to the Ca²⁺ flux, changing from a random, dispersed distribution in proliferative myoblasts to a predominantly peripheral distribution in prefusion myoblasts. Coincident with the mM CAF relocation is an altered distribution of a surface glycoprotein, fibronectin. Extracellular fibronectin is seen in abundance in proliferating myoblasts, but is essentially absent from the surface of fusing myoblasts. We suggest that mM CAF when activated by Ca²⁺ influx may act to promote the release of fibronectin from the myoblast cell surface, thus providing a mechanism by which the membrane of the fusing myoblast may be rearranged to accommodate fusion.     

Microinjection of calpastatin inhibits fusion in myoblasts

Rat satellite cells (RSC) were microinjected with purified calpastatin or m-calpain, and myoblasts from a C2C12 mouse line were microinjected with purified calpastatin. Microinjection with calpastatin completely prevented fusion of myoblasts from both sources, whereas microinjection with m-calpain significantly increased the rate of fusion of cultured RSC; 44% of the nuclei of RSC cultures were in multinucleated myotubes within 48 h after microinjection with m-calpain plus labeled dextran, whereas only 15% of the nuclei were in multinucleated myotubes after microinjection with dextran alone. Western analyses indicated that neither RSC nor C2C12 myoblasts contained detectable amounts of mu-calpain before fusion. The levels of calpastatin in C2C12 myoblasts increased as cells passed from the proliferative stage to the onset of fusion, and these levels increased substantially in both the C2C12 and the RSC cells as they progressed to the late or postfusion stage. Both RSC and C2C12 myoblasts contained an 80-kDa polypeptide that was labeled with an anti-m-calpain antibody in Western blots. The results are consistent with a role of the calpain system (m-calpain in these myoblast lines) in remodeling of the cytoskeletal/plasma membrane interactions during cell fusion.     

The control of chick myoblast fusion by ion channels operated by prostaglandins and acetylcholine

Chick myoblast fusion in culture was investigated using prostanoid synthesis inhibitors to delay spontaneous fusion. During this delay myoblast fusion could be induced by prostaglandin E1 (PGE1), by raising extracellular potassium and by addition of carbachol. Carbachol-induced fusion, but not PGE-induced fusion, was prevented by the acetylcholine receptor blocker alpha-bungarotoxin. Fusion induced by any of these agents was prevented by the Ca channel blockers lanthanum and D600. The threshold for potassium-induced fusion was 7-8 mM; maximal fusion occurred at 16-20 mM. Low extracellular potassium inhibited spontaneous fusion. Intracellular potassium in fusion competent myoblasts was 101 m-moles/l cell. Calcium flux measurements demonstrated that high potassium increased calcium permeability in fusion-competent myoblasts. A 30-s exposure to high potassium or PGE1 was sufficient to initiate myoblast fusion. Anion-exchange inhibitors (SITS and DIDS) delayed spontaneous myoblast fusion and blocked fusion induced by PGE1 but not carbachol. Blocking the acetylcholine receptor shifted the dose-response relation for PGE-induced fusion to higher concentrations. PGE1-induced fusion required chloride ions; carbachol-induced fusion required sodium ions. Provided calcium channels were available, potassium always induced fusion. We conclude that myoblasts possess at least three, independent pathways, each of which can initiate myoblast fusion and that the PGE-activated pathway and the acetylcholine receptor-activated pathway act synergistically. We suggest that fusion competent myoblasts have a high resting membrane potential and that fusion is controlled by depolarization initiated directly (potassium), by an increase in permeability to chloride ions (PGE), or by activation of the acetylcholine receptor (carbachol); depolarization triggers a rise in calcium permeability. The consequent increase in intracellular calcium initiates myoblast fusion.     

Spontaneous myoblast fusion is mediated by cell surface Ca-dependent protein kinase(s)

Mononucleated myoblasts divide in vitro until they attain confluency and fuse, forming multinucleated myotubes. Fusion is an extracellular Ca2+-dependent process. We used for our studies an established line of skeletal myoblasts (L6) as well as a non-fusing Myo- alpha-amanitin-resistant mutant of this line (Ama102). Our results show that extracellular calcium at concentrations which elicit myoblast fusion activates the phosphorylation of a protein species of 48 kD, present at the surface of mononucleated myoblasts of the fusing wild type (L6). At fusion, as the cells become independent of the extracellular calcium concentration for their further differentiation, this activation can no longer be observed. In fusion inhibition experiments, where we used lowered calcium levels, the phosphorylation of the 48 kD protein band is clearly decreased. When the myoblasts are fed with standard medium, they fuse rapidly and the phosphorylation of the 48 kD species is markedly increased. The above-described phenomenon takes place at the cell surface and is completed in a short time. The use of Myo- mutant showed that it is developmentally regulated. In view of our results, it is reasonable to postulate that Ca2+-activated phosphorylation of the cell surface could be on the basis of spontaneous myoblast fusion.     

Specific knockdown of m-calpain blocks myogenesis with cDNA deduced from the corresponding RNAi

Fusion of mononuclear myoblast to multinucleated myotubes is crucial for myogenesis. Both µ- and m-calpain are ubiquitously expressed in most cells and are particularly abundant in muscle cells. Knockout of calpain-1 (catalytic subunit of µ-calpain) induced moderate platelet dysaggregation, preserving the normal development and growth, although knockout of calpain-2 (m-calpain) is lethal in mice. Therefore, there should be muscle-specific function of m-calpain per se. Previous methods lack direct evidence for the involvement of m-calpain, because the specific inhibitor to m-calpain has not been developed yet and the inhibition was less potent. Here, we show that screened RNA interference (RNAi) specifically blocked the m-calpain expression by 95% at both the protein and the activity levels. After transfection of adenovirus vector-mediated cDNA corresponding to the RNAi-induced short hairpin RNA, m-calpain in C₂C₁₂ myoblasts was knocked down with no compensatory overexpression of µ-calpain or calpain-3. The specific knockdown strongly inhibited the fusion to multinucleated myotubes. In addition, the knockdown modestly blocked ubiquitous effects, including cell migration, cell spreading, and alignment of central stress fiberlike structures. These results may indicate that m-calpain requiring millimolar Ca²⁺ level for the full activation plays specific roles in myogenesis, independent of µ-calpain, and leave us challenging problems in the future. RNA interference; muscle cell development; fusion; adenovirus vector     

Antibodies to 100- and 60-kDa surface proteins inhibit substratum attachment and differentiation of rodent skeletal myoblasts

A rabbit polyclonal antiserum was raised against membrane vesicles shed from the surface of fusing L6 rat myoblasts. In immunoblots the antiserum recognized fibronectin, a protein of approximately 100,000 Da (100-kDa), and a protein of approximately 60,000 Da (60 kDa). If added prior to cellular alignment, immunoglobulins from this serum inhibited fusion of both rat (L6) and mouse (C2) myoblasts in a dose-dependent fashion. To determine which component of this serum was responsible for fusion inhibition, antibodies against fibronectin, the 100- and 60-kDa proteins were microaffinity purified and tested, individually, for their effects on myoblast fusion. Antibodies against fibronectin had no effect on fusion. Antibodies against the 100-kDa protein released most cells from the substratum. Antibodies against the 60-kDa protein completely inhibited fusion. Fusion inhibition was accompanied by a corresponding inhibition of expression of two differentiation markers, creatine phosphokinase and the acetylcholine receptor. The 60-kDa protein was found, by immunoblot analysis, in smooth muscle-like cells (BC3H1 cells) and in variant L6 cells that do not differentiate and do not fuse. However, in the differentiation incompetent cells, the 60-kDa antigen appeared to be present in reduced amount. Indirect immunofluorescence of unpermeabilized L6 cells revealed alterations in the distribution of all three antigens during development. Fibronectin first appeared in long fibrillar arrays above the surface of cells that were beginning to align and fuse; fibronectin was not present on myotubes. The 100-kDa protein was seen initially in prominent fibrillar projections at the tips of prefusion myoblasts. During fusion the antigen was observed at sites of cell-cell contact and on extracellular vesicles. The 100-kDa protein appeared to be less abundant on myotubes. The 60-kDa protein first appeared in regions of cell-cell contact on cells that were beginning to align and fuse. As. fusion progressed, the 60-kDa protein was also found in extracellular vesicles. The 60-kDa protein was not observed on myotubes. As a result of this study we have identified two previously undescribed cell surface proteins involved in rodent skeletal myogenesis. The first is an approximately 100-kDa protein involved in early interactions of skeletal myoblasts with their substratum. The second is an approximately 60-kDa protein involved in myoblast differentiation. Both proteins are shed from the myoblast surface during myotube formation.     

Decorin modulates collagen I-stimulated, but not fibronectin-stimulated, migration of C2C12 myoblasts

Extracellular matrix factors, specifically fibronectin and collagen I, are essential for structural support during muscle regeneration. Decorin has been identified as an anti-fibrotic agent with binding sites located on both fibronectin and collagen I. Upon injury, activated myoblasts are required to migrate through the extracellular matrix factors deposited by the myofibroblasts to facilitate skeletal muscle regeneration. In this study we looked at the effects decorin on fibronectin- and collagen I-stimulated myoblast migration. Dose response studies demonstrated 10 μg/ml, 5 μg/ml and 25 μg/ml as the optimal stimulatory concentrations of decorin (1.2 fold increase), fibronectin (3.5 fold increase) and collagen I (2.4 fold increase), when compared with control respectively. A synergistic effect was identified when decorin and collagen I were added in combination; this effect was not evident when decorin was added with fibronectin. The effects of these factors on the ROCK signalling pathway were also analyzed. ROCK-2 was identified as the key Rho-activated kinase isoform involved in migration, due to its higher expression levels and localisation to focal points within migrating C2C12 myoblasts. Decorin and collagen I in combination stimulated an increase in the number of ROCK-2 localized focal points when compared with control, decorin and collagen I added separately. Fibronectin did not show any increase in ROCK-2 focal points when compared with control. These results show for the first time that decorin can modify collagen I-stimulated, but not fibronectin-stimulated myoblast migration in vitro. Furthermore, the synergistic, rather than additive, effect observed suggests a direct modification of collagen I signalling by decorin mediated, at least in part, by ROCK-2 rather than ROCK-1.     

Fibronectin expression during myogenesis

The biosynthesis and localization of fibronectin during chick muscle differentiation are described. This study employed two monoclonal antibodies, one that selectively killed mononucleated cells and one specific for avian fibronectin. These antibodies allowed precise analyses of fibronectin expression in well-defined cultures of myoblasts or myotubes and avoided the complications of exogenous fibronectin and contamination by fibroblasts or unfused myoblasts. Fibronectin synthesis, as a fraction of total protein synthesis, remains constant at 0.3-0.4% before and after myoblast fusion, suggesting that the absolute rate of fibronectin synthesis may increase somewhat when myotubes synthesize and accumulate myofibrillar proteins. The pattern of fibronectin arrangement does change during myogenesis. In myotube cultures, the appearance of pulse-labeled fibronectin at the cell surface and its secretion into the medium begin after a 2-3-h lag period, in contrast to the 30-min lag period observed in fibroblast cultures. This lag between polypeptide biosynthesis and the exteriorization of the new protein is thus a characteristic of each cell type rather than the protein. All of the major secretory proteins of myogenic cells, including fibronectin and collagenous components, share this 2-3-h intracellular transit time.     

m-Calpain implication in cell cycle during muscle precursor cell activation

Milli-calpain, a member of the ubiquitous cysteine protease family, is known to control late events of cell-cell fusion in skeletal muscle tissue through its involvement in cell membrane and cytoskeleton component reorganization. In this report, we describe the characterization of m-calpain compartmentalization and activation during the initial steps of muscle precursor cell recruitment and differentiation. By immunofluorescence analysis, we show that m-calpain is present throughout the cell cycle in the nucleus of proliferating myoblast C2 cells. However, when myoblasts enter a quiescent/G0 stage, m-calpain staining is detected only in the cytoplasm. Moreover, comparison of healthy and injured muscle shows distinct m-calpain localization in satellite stem cells. Indeed, m-calpain is not found in quiescent satellite cells, but following muscle injury, when satellite cells start to proliferate, m-calpain appears in the nucleus. To determine the implication of m-calpain during the cell cycle progression, quiescent myoblasts were forced to re-enter the cell cycle in the presence or not of the specific calpain inhibitor MDL 28170. We demonstrate that this calpain inhibitor blocks the cell cycle, prevents accumulation of MyoD in the G1 phase and enhances Myf5 expression. These data support an important new role for m-calpain in the control of muscle precursor cell activation and thus suggest its possible implication during the initial events of muscle regeneration.     

Involvement of calpains in growth factor-mediated migration

Previous research in our laboratory has already shown the importance of the role played by ubiquitous calpains during myoblast migration. The aim of this study was to investigate calpain expression during myoblast migration and, to enhance this phenomenon via calpain stimulation. Ubiquitous calpains are members of a large family of calcium-dependent cysteine proteases. They play an important role in numerous biological and pathological phenomena, such as signal transduction, apoptosis, cell-cycle regulation, cell spreading, adhesion, invasion, myogenesis, and motility. Myoblast migration is a crucial step in myogenesis, as it is necessary for myoblast alignment and fusion to form myotubes. This study started by examining changes in calpain expression during migration, then investigated the possibility of activating myoblast migration via the stimulation of calpain expression and/or activity. The migration rate of myoblasts overexpressing mu- or milli-calpain was quantified. The results showed that calpain overexpression dramatically inhibited myoblast migration. Growth-factor treatments were then used to enhance myoblast migration. The results showed that treatment with IGF-1, TGF-β1, or insulin induced a major increase in migration and caused a significant increase in m-calpain expression and activity. The increase in migration was totally inhibited by adding calpeptin, a calpain-specific inhibitor. These findings suggest that milli-calpain is involved in growth factor-mediated migration.     

Proteoglycans of L6J1 myoblast extracellular matrix: Properties and effect on myoblast adhesion

Proteoglycans were isolated from the extracellular matrix (ECM) of L6J1 rat myoblasts; their influence on myoblast adhesion has been studied. Proteoglycan digestion with chondroitinase AC and heparinase III, which degrade polysaccharide moieties, has revealed that chondroitin sulfate proteoglycans are a major class of myoblast extracellular matrix proteoglycans. Electrophoresis of enzymatically processed proteoglycans was used to examine their core proteins. Myoblast adhesion was suppressed by proteoglycans or a mixture of proteoglycans and a fibronectin-extracellular matrix. Myoblast adhesion to a substrate composed of fibronectin and proteoglycans is restored after the substrate was treated with chondroitinase AC. In conclusion, proteoglycans of L6J1 rat myoblast ECMs were isolated and purified. Chondroitin sulfate proteoglycans are a major class of proteoglycans. Isolated proteoglycans suppressed myoblast adhesion; the effect was mediated by polysaccharide moieties of proteoglycans.     

Thrombin, a survival factor for cultured myoblasts

Three members of the family of protease-activated receptors (PARs), PARs-1, -3 and -4, have been identified as thrombin receptors. PAR-1 is expressed by primary myoblast cultures, and expression is repressed once myoblasts fuse to form myotubes. The current study was undertaken to investigate the hypothesis that thrombin inhibits myoblast fusion. Primary rodent myoblast cultures were deprived of serum to promote myoblast fusion and then cultured in the presence or absence of thrombin. Thrombin inhibited myoblast fusion, but another notable effect was observed; 50% of control cells were apoptotic within 24 h of serum deprivation, whereas less than 15% of thrombin-treated cells showed signs of apoptosis. Proteolysis was required for the effect of thrombin, but no other serine protease tested mimicked the action of thrombin. Neither a PAR-1- nor a PAR-4-activating peptide inhibited apoptosis or fusion, and myoblast cultures were negative for PAR-3 expression. Myoblasts exposed to thrombin for 1 h and then changed to medium without thrombin accumulated apoptosis inhibitory activity in their medium over the subsequent 20 h. Thus the protective action of thrombin appears to be effected through cleavage of an unidentified thrombin receptor, leading to secretion of a downstream apoptosis inhibitory factor. These results demonstrate that thrombin functions as a survival factor for myoblasts and is likely to play an important role in muscle development and repair.