I. Richard, Généthon, CNRS UMR8115, 1 rue de l'Internationale, 91000 Evry, France Fax: +33 1 60 77 86 98 Tel: +33 1 69 47 29 38 E-mail: email@example.com
Calpain 3 is a 94-kDa calcium-dependent cysteine protease mainly expressed in skeletal muscle. In this tissue, it localizes at several regions of the sarcomere through binding to the giant protein, titin. Loss-of-function mutations in the calpain 3 gene have been associated with limb-girdle muscular dystrophy type 2A (LGMD2A), a common form of muscular dystrophy found world wide. Recently, significant progress has been made in understanding the mode of regulation and the possible function of calpain 3 in muscle. It is now well accepted that it has an unusual zymogenic activation and that cytoskeletal proteins are one class of its substrates. Through the absence of cleavage of these substrates, calpain 3 deficiency leads to abnormal sarcomeres, impairment of muscle contractile capacity, and death of the muscle fibers. These data indicate a role for calpain 3 as a chef d'orchestre in sarcomere remodeling and suggest a new category of LGMD2 pathological mechanisms.
Calpains (EC 220.127.116.11) are nonlysosomal cysteine proteases with activity that is calcium dependent (for a detailed review of calpains, see ). The most well-known ones are the ubiquitous heterodimeric calpains (µ and m), which have been known since the eighties, and calpain 3, loss-of-function mutations of which lead to limb-girdle muscular dystrophy type 2A (LGMD2A, MIM No. 253600 ).
LGMD2A is one of the most common LGMDs, accounting for about 35% of cases, with a prevalence estimated to be 1 : 15 000–1 : 150 000 depending on the area (for a complete list of publications reporting mutations, see Table 1 and supplementary Doc S1; [3,4]). To date, 286 distinct pathogenic calpain 3 mutations (13 nonsense, 74 deletion/insertion, 38 splice site, and 161 missense) have been characterized in the literature and the Leiden database (http://www.dmd.nl/capn3_home.html). They are distributed along the entire length of the gene (Fig. 1). Patients with LGMD2A, like other LGMD2 patients, classically present with progressive muscle weakness and atrophy of the shoulder and pelvic girdle musculature, an elevated serum creatine kinase activity and a degeneration/regeneration pattern in muscular biopsy samples . Interestingly, patients homozygous for null mutations usually have no protein and a more severe phenotype, suggesting that a correlation between clinical phenotype and genotype may exist.
Table 1. . Publications reporting calpain 3 mutations in chronological order (for full reference details see supplementary Doc S1). The geographic origin of patients reported in each publication is indicated in the second column. A website reporting calpain 3 mutations is indicated at the end of the table.
Richard et al. 1995 PMID: 7720071
Brazil, France, Reunion Island
Dincer et al. 1997 PMID: 9266733
Richard et al. 1997 PMID: 9150160
France, Israel, Italy, Turkey, USA
Haffner et al. 1998 PMID: 9452114
Penisson-Besnier et al. 1998 PMID: 9655129
Brazil, France, Reunion Island
Kawai et al. 1998 PMID: 9771675
Urtasun et al. 1998 PMID: 9762961
Chou et al. 1999 PMID: 10102422
Italy, Mexico, Poland, USA
Passos-Bueno et al. 1999 PMID: 10069710
Minami et al. 1999 PMID: 10567047
Richard et al. 1999 PMID: 10330340
Bulgaria, Canada, France, Germany, Greece, USA, Italy, Japan, Lebanon, the Netherlands, Poland, Russia, Spain, Switzerland, Turkey, UK, USA, Vietnam
Calpain 3 is the only calpain known to cause a monogenic disease, and its implication in LGMD2A underscores its crucial role in muscle homeostasis. Recently, significant progress has been made in the comprehension of its mode of regulation and its possible function in muscle. This review summarizes the current knowledge about the calpain 3 gene and protein, as well as the disease pathogenesis.
The calpain 3 gene and its expression
The human calpain 3 gene is located on chromosome 15q15.1-q21.1 and covers a genomic region of ≈138 kb (Ensembl gene ID ENSG00000092529; Fig. 1). The predominant product of this gene is encoded by 24 exons corresponding to a 3316-bp mRNA and is principally expressed in adult skeletal muscle in fast-twitch and slow-twitch fibers [2,6]. Accordingly, the phenotype of LGMD2A affects both types of fiber .
In addition to the main product, multiple alternative transcripts have been detected in human, mouse, rat and rabbit tissues, but usually with an expression level 100- to 1000-fold lower (for listing of isoforms, see [8,9]). Some of these transcripts are expressed from an additional alternative ubiquitous promoter known to be present in human and mouse genomes or from a lens-specific promoter detected in mouse, rat and rabbit genomes but absent from the human genome. As the phenotype observed in patients with LGMD2A is muscle-restricted, we will not discuss the role of calpain 3 outside skeletal muscle.
Structure of the calpain 3 protein
Translation of the main calpain 3 gene product leads to a 94-kDa protein of 821 amino acids consisting of a short N-terminal region (domain I), a papain-type proteolytic domain (domains IIa and IIb), a C2-like domain (domain III) and a calcium-binding domain composed of five EF-hands (domain IV) [6,10] (Fig. 1). In addition, calpain 3 possesses three unique sequences not found in any other calpains, NS (N-terminal sequence), IS1 and IS2 (inserted sequences 1 and 2). NS is a 20–30 amino-acid N-terminal domain rich in proline encoded by exon 1. This region is in domain I which corresponds to a regulatory propeptide found in various cysteine proteinases . IS1 is a polypeptide of about 50 amino acids encoded by exon 6 and embedded in the proteolytic domain. It contains three autolytic sites: Y274, N292 and Y322. As a consequence, calpain 3 in which IS1 is deleted no longer autolyzes, although it is still proteolytically competent . Calpain 3 autolysis occurs rapidly in heterologous cells or inadequately extracted muscle samples and presumably after physiological activation in living muscle [12,13]. It generates a small fragment of 30 kDa and a large C-terminal fragment, the size of which ranges from 60 to 55 kDa depending on the extent of autolysis. IS2 is a peptide of about 80 amino acids encoded by exons 15–16 and located between domain II and domain III. A basic PVKKKKNKP sequence encoded by exon 15 seems to act as a nuclear translocation signal at least in human and COS-7 cells [12,14]. IS2 has been demonstrated to be important in the control of the activity of calpain 3, as exon 15 deletion leads to a Ca2+ independence of autolytic activity, and exon 16 deletion leads to loss of substrate proteolysis [8,15].
Because of the rapid autolysis of calpain 3, it has so far been impossible to obtain crystals of the full molecule. However, Jia and colleagues established a 3D model of calpain 3 based on the known structure of m-calpain . The model shows that the proteolytic domain can be subdivided into two globular subdomains (domain IIa and IIb), forming a catalytic cleft at their interface. As in ubiquitous calpains, domain III of calpain 3 fits a C2 motif. In this model, IS1 and IS2 have been structured as loops protruding out of the globular core structure. However, Diaz and colleagues have shown that, instead of protruding, IS1 is composed of an α-helix flanked by loops that close the catalytic cleft, blocking its access to substrates and inhibitors .
Recently, other structural analyses have revealed that calpain 3 could homodimerize through its penta EF-hand domain . The dimer would be in a tail to tail orientation, placing the catalytic domains at both ends. This homodimerization is reminiscent of the heterodimeric structure of the ubiquitous calpains, the large subunits of which associate with a small subunit of 30 kDa . It has been suggested that the small subunit may act as a chaperone or that dissociation from the catalytic subunit is part of the activation process of the ubiquitous calpains [19,20]. Therefore, the interesting observation of calpain 3 dimerization raises the question of whether and how it can intervene in the regulation of calpain 3 activation or binding to partners.
Subcellular localization of calpain 3
Insights into the subcellular localization of calpain 3 have come from a yeast two-hybrid screening in which calpain 3 was found to bind to the I-band and M-line regions of titin [21,22]. This extremely large molecule spans half the sarcomere from the Z-disc to the M-line and participates in the construction and overall elasticity of the myofibrils . The binding of calpain 3 to the I-band was restricted to the Ig83 immunoglobulin-like domain of titin which is located in the N2A region, close to the extensible PEVK domain (Fig. 2). In the M-line, it was mapped to a unique region flanked by two immunoglobulin C2 motifs encoded by Mex5, the next to last exon of titin (Fig. 2). The minimal region for the binding to N2A of calpain 3 involves the IS2 region, comprising residues 570–639 . Concerning the binding to the M-line, no minimal domain has been found . However, the replacement of exon 1 by the lens-specific first exon abolished the binding to both sites, whereas the absence of IS1 and IS2 seems to increase the binding . In conclusion, the mechanism of calpain 3 binding to the two titin regions is apparently different, suggesting distinct physiological functions.
Subsequent immunolocalization studies carried out in humans and mice confirmed that calpain 3 is localized in several regions of the sarcomere. In addition to the N2A and M-line location, calpain 3 also seems to be localized in the Z-disc [13,22,24]. Beside these localizations, calpain 3 has also been found at the costameres and myotendinous junctions in mouse muscle and in the nucleus in human muscle [13,14].
Regulation of the proteolytic activity of calpain 3
As a cytoplasmic protease, the activity of calpain 3 must be tightly regulated temporally and spatially to be effective and to avoid unwanted damage. Besides control at the transcriptional level and regulation via its compartmentalization, another interesting regulatory mechanism has been identified at the protein level [13,16]. When extracted from fresh muscle, calpain 3 is seen mainly in an unprocessed form that has been shown to correspond to an inactive protein [13,16,25]. Once activated, it first undergoes intramolecular proteolysis at one autolytic site in IS1. Then, calpain 3 can intermolecularly proteolyze the other sites, removing the IS1 loop and leaving the two other parts of the molecule associated through noncovalent bounds. Thus, IS1 acts as an inhibitory peptide of calpain 3 activity.
In contrast with the situation in muscle, calpain 3 is fully active when expressed in nonmuscle cells . Therefore, it can be postulated that the muscle inhibition arises through interaction with a muscle-specific protein, a good candidate for which is its partner, titin. Interestingly, when calpain 3 was overexpressed in muscle, either in transgenic mice or by gene transfer, no obvious phenotype was seen, indicating a high buffering capacity of the muscle [26,27]. On the other hand, overexpression of calpain 3 in muscular dystrophy with myositis (mdm) mice, a strain carrying a deletion of several amino acids in the Ig83 domain of titin, aggravates the phenotype, suggesting the necessity for N2A binding to control calpain 3 activity [28,29]. However, coexpression with the N2A region in COS-7 cells does not by itself impair the proteolytic activity of calpain 3 . However, downstream of N2A, titin has a domain presenting some homology with calpastatin, an inhibitor of the ubiquitous calpains . In addition, in tibialis muscular dystrophy, a muscle disease caused by specific mutations in the M-line titin, calpain 3 was reduced or absent as an unprocessed protein [32,33]. Taken together, these observations pinpoint titin as a reservoir of inactive calpain 3 molecules and suggest that dissociation from titin corresponds to calpain 3 activation.
The question remains what is the signal leading to calpain 3 activation? The main activation signal of the ubiquitous calpains is Ca2+. After much debate, the Ca2+ dependency of calpain 3 activity is now well established thanks to research with mutants and in vitro analyses of the catalytic subdomains [15,34–36]. However, Ca2+ cannot be considered a signal for calpain 3 because a trace amount is sufficient for activation . Another signal that has been excluded is exercise, as calpain 3 is down-regulated after eccentric exercise and does not autolyze with exhaustive or endurance exercises in humans [37,38]. Therefore, further studies are still needed to obtain a clear picture of the calpain 3 activation signal.
Calpain 3 substrates
Coexpresssion experiments and in vitro studies have led to the identification of numerous proteins that can be cleaved by calpain 3, including titin, filamin C, vinexin, ezrin and talin [13,39,40]. Although in vivo confirmation is awaited, the fact that these proteins are located in the vicinity of calpain 3 renders them likely physiological substrates. The absence of a consensus sequence at the cleavage sites indicates that there is no specificity and suggests that calpain 3 cleaves destructured regions as the ubiquitous calpains do. The pattern of cleaved products suggests limited proteolysis as a means to irreversibly modulate the function of substrates (Fig. 3). For example, the cleavage of filamin C at the extreme C-terminus abolishes the interaction with γ-sarcoglycans and δ-sarcoglycans and dissociates the dimerization domain from the rest of the molecule . Another example comes from talin. The ubiquitous calpains cleave talin at almost the same position as calpain 3 . This cleavage induces a 16-fold increased affinity for integrin β-3, showing that calpains may induce an increase in the function of their substrates . These data pinpoint a putative function for calpain 3 in the adjustment of cytoskeleton/membrane links.
Lack of proteolysis of substrates as the origin of LGMD2A pathogenesis
Fine correlation analyses of LGMD2A mutations with perturbations on calpain 3 features may be the first step to understanding the pathogenesis of the disease. The numerous patients who present two null mutations leading to the absence of the protein together with the recessive pattern of inheritance clearly indicates that LGMD2A is due to a deficiency in the function of calpain 3. This piece of evidence has further been validated by the reproduction of the phenotype when the gene has been knocked-out in mice. However, analysis of patient biopsy specimens showed that a proportion of them have normal calpain 3 expression on western blot (in particular for the patients homozygous for the mutations T184M, G222R, G496R, S606L, R490W, R490Q, R489Q and R461C). However, further analysis showed that some of these mutations could be associated with impairment of autolytic activity, and therefore indicative of perturbation of calpain 3 function .
To add more complexity, it has been shown that S606L, a mutation located in IS2, leads to a normal calpain 3 level, with autolytic activity as well as correct subcellular localization . Interestingly, in vitro analysis of other LGMD2A missense mutations (S744G and R769Q) indicated that they retain autolytic activity as well . Even though they have the ability to cleave calpain 3 intermolecularly, they are no longer able to cleave the endogenous fodrin in transfected COS-7 cells. These data indicate that, in these mutants, the intramolecular and intermolecular proteolysis is not affected and suggests a problem in substrate recognition. Others mutations were shown to impair titin binding, including the fully active R448H, D705G mutants [10,40]. In those cases, it is possible that the resulting abnormal compartmentalization may have the consequence of preventing the cleavage of substrates. Taken together, these observations are consistent with the hypothesis that LGMD2A pathogenesis is related to the loss of proteolytic activity against substrates. A preliminary requirement to confirm that this is true in vivo would require the identification of a condition in which a physiological cleavage of substrates could be observed.
Calpain 3 activity is needed in fully mature myofibers
Calpain 3 is not essential for building functional muscles, as indicated by the fact that muscles of patients develop normally and that the mean age of onset of the disease is in the second decade. Along with this fact, expression of the full-length calpain 3 during both human and mouse skeletal muscle development is a relatively late event, subsequent to muscle innervation and therefore to myoblast proliferation and fusion . It was also consistently observed that its expression is concomitant with the appearance of neoformed myotubes and reinnervation in the regeneration process occurring after experimental degeneration [44,45] and with myoblast differentiation during in vitro myogenesis in C2C12 cells [8,45,46]. It can be concluded that calpain 3 is not required for myoblast proliferation and fusion, in contrast with the ubiquitous calpains, nor for the regulation of muscle regeneration and reinnervation. The function of calpain 3, which manifests itself as proteolysis of substrates, seems to be important during the life of fully differentiated fibers; its absence leads to degeneration and death of the fibers.
Calpain 3 in the life and death of myofibers
The ubiquitous calpains have been shown to participate in the initial proteolytic events that accompany muscle wasting, whereas calpain 3 can be excluded from this process for several reasons. First, calpain 3 deficiency results in muscle atrophy in LGMD2A. Secondly, in two models of cachexia (transgenic mice overexpressing interleukin 6 and Yoshida AH-130 rat ascites hepatoma), calpain 3 mRNA has been shown to be down-regulated [47,48]. Thirdly, calpain 3 is also down-regulated during the atrophic phase seen after nerve section . In all cases, calpain 3 activity correlates negatively with muscle degradation, again in contrast with the ubiquitous calpains which show a positive correlation .
We can also state that the myofiber degeneration observed in patients with LGMD2A is not related to membrane disruption, in contrast with other muscular dystrophies caused by mutations in proteins of the dystrophin–glycoprotein complex (for a relatively recent review, see ). In fact, there is a normal amount and correct localization of sarcolemmal proteins such as dystrophin, sarcoglycans and merosin in LGMD2A [51–54]. Even if some Evans blue-positive cells can occasionally be seen in calpain-3-deficient muscles, they probably reflect dying fibers rather than membrane permeability. Furthermore, no increased numbers of Evans blue-positive cells after exercise and no deficiency in membrane resistance of stretched isolated muscles have been observed .
Beside membrane fragility, a second pathogenic mechanism leading to LGMD2 has been identified in the form of deficiency in membrane repair . This defect was observed in LGMD2B due to mutations in dysferlin, a member of the newly described ferlin family . It is interesting to note that there is a secondary reduction in dysferlin in calpain 3-deficient muscle  and that an interaction has been identified between calpain 3 and dysferlin . However, the lack of Evans blue-positive cells argues against the participation of calpain 3 in the repair process.
Another mechanism is needed to explain LGMD2A pathology. Indeed, interesting observations can be put together to link calpain 3 deficiency with abnormal sarcomere organization. First, biopsy samples from LGMD2A patients and calpain 3-deficient mice present aspecific ultrastructural changes such as the presence of lobulated fibers, fragmentation and disorganization of myofibers [34,40,59]. Secondly, calpain 3-deficient primary myotubes from knock-out mice lacked well-organized sarcomeres and presented a misincorporation of adult myosin heavy chain . Finally, antisense oligonucleotides against calpain 3 led to immature Z discs and diffuse distribution of α-actinin in myotubes . Altogether, these data suggest a role for calpain 3 in sarcomere maintenance in mature muscle cells.
During adult life, skeletal muscles must constantly adapt to respond to metabolic, mechanical or hormonal conditions. These adaptations involve altered patterns of both protein synthesis and protein degradation and promote changes in contractile and metabolic proteins to optimize muscle function . Considering the highly organized structure of the muscles, the exchange of myofibrillar proteins during these processes, known as sarcomere remodeling, necessitates the intervention of proteolytic systems. Indeed, numerous studies have shown that the ubiquitous calpains intervene in the initial phase of myofibril disassembly, and the ubiquitin/proteasome system is in charge of the degradation of proteins that are no longer needed [49,62,63]. Interestingly, the recovery phase subsequent to unloading is associated with an increase in calpain 3 expression, whereas calpain 3-deficient muscles failed to regain their full weight under this condition . In addition, there is an increase in ubiquitination of proteins in reloading that is not seen in the absence of calpain 3. It is noteworthy that a reduction in the expression of several ubiquitin/proteasome system components was observed in calpain 3-deficient mice . In conclusion, it is possible that calpain 3 deficiency impairs the remodeling response consequent to perturbation of the ubiquitin/proteasome system.
These abnormal sarcomeres seem to have a twofold effect: (a) a decrease in the force-generating capacity of the fibers related to impaired contractility of the muscle fibers ; (b) an increase in cellular stress as indicated by the up-regulation of heat-shock proteins in the muscles of knock-out mice and the presence of apoptotic myonuclei in patients and mice [14,54,64]. However, it is not possible to know whether the perturbation of the apoptosis-controlling pathway of NF-κB observed in patients with LGMD2A is subsequent to the stress response, to the adjustment of the nuclei number to the volume of the atrophying fibers, or is a direct consequence of the lack of calpain 3 activity on the NF-κB/IκBα pathway.
Ten years ago, the gene responsible for LGMD2A was identified as coding for the enigmatic protease, calpain 3. This finding was the starting point for molecular diagnosis for patients and had the consequence of designating LGMD2A as a common form of muscular dystrophy. The recognition of LGMD2A is still a challenge at the protein level as some mutations maintain the protein but in an inactive form and secondary reductions are observed in a number of muscular dystrophies. From a therapeutic point of view, treatment of this recessive disease by gene transfer can be proposed and tested based on information about the gene involved. Indeed, we recently demonstrated the safety and efficacy of adeno-associated virus (AAV)-mediated calpain 3 cDNA transfer in a mouse model of LGMD2A . However, gene therapy still has some obstacles that remain to be worked out before it can become a therapeutic solution in human beings. Hopefully, identification of the role of calpain 3 will eventually lead to an understanding of the pathogenesis of the disease and proposals of original pharmacological treatment for this disorder. In fact, we are on the verge of understanding the full extent of calpain 3 regulation and physiological function. In addition to its regulation of transcription, alternative splicing and subcellular compartmentalization, calpain 3 has an interesting and unusual internal zymogenic mechanism of activation that is unique in the protease world. In mature innervated fibers, calpain 3 seems to play a role in sarcomere remodeling by cleaving cytoskeletal proteins during muscular adaptation. This role is in agreement with the cytoskeletal nature of the known in vitro substrates of calpain 3. Identifying the signal that triggers its activity is the next step for the validation of its physiological substrates and determination of the consequences on muscle regulation. Placing calpain 3 in the context of the biological pathway in which it acts will then make it possible to envisage how and when to intervene therapeutically to bypass this pathway or compensate for its perturbation.
Overall, calpain 3 can be envisaged as a ‘chef-d’ orchestre' in the homeostasis of the muscle sarcomere. From this proposed role, it can be postulated that deregulation of sarcomere remodeling would constitute the origin of LGMD2A pathogenesis. It suggests the existence of a new pathogenic mechanism besides membrane fragility and membrane repair which may also be applied to other muscular dystrophies caused by mutations in sarcomeric proteins.
We would like to acknowledge Dr Nathalie Daniele, Dr Susan Cure and Dr Oliver Danos for critical reading of the manuscript. This work was supported by the Association Française contre les Myopathies.