Calpain 3 Is Activated through Autolysis within the Active Site and Lyses Sarcomeric and Sarcolemmal Components

Abstract

Calpain 3 (Capn3) is known as the skeletal muscle-specific member of the calpains, a family of intracellular nonlysosomal cysteine proteases. This enigmatic protease has many unique features among the calpain family and, importantly, mutations in Capn3 have been shown to be responsible for limb girdle muscular dystrophy type 2A. Here we demonstrate that the Capn3 activation mechanism is similar to the universal activation of caspases and corresponds to an autolysis within the active site of the protease. We undertook a search for substrates in immature muscle cells, as several lines of evidence suggest that Capn3 is mostly in an inactive state in muscle and needs a signal to be activated. In this model, Capn3 proteolytic activity leads to disruption of the actin cytoskeleton and disorganization of focal adhesions through cleavage of several endogenous proteins. In addition, we show that titin, a previously identified Capn3 partner, and filamin C are further substrates of Capn3. Finally, we report that Capn3 colocalizes in vivo with its substrates at various sites along cytoskeletal structures. We propose that Capn3-mediated cleavage produces an adaptive response of muscle cells to external and/or internal stimuli, establishing Capn3 as a muscle cytoskeleton regulator.

Calpains are a large family of intracellular cysteine proteases (31). This family to date comprises 14 different members, which are expressed in a ubiquitous or tissue-specific manner (29). Despite numerous studies, the precise physiological functions of calpains are still elusive. The ubiquitous calpains have been involved in a variety of signaling pathways, inducing irreversible modifications through limited proteolysis of specific targets (10, 36). The best-characterized function of ubiquitous calpains is their implication in the regulation of the cytoskeleton and focal adhesions. They have been shown to modulate these structures during cell migration (11), cell transformation (6), platelet activation (9), and wound healing (27).

Among the calpain family members, calpain 3 (Capn3) is a particularly interesting protein. This unique calpain carries two insertion sequences, IS1 and IS2, which are involved in the regulation of its function and activity (12). IS1 includes three autolytic sites, S1, S2, and S3 (14). IS2 carries a nuclear translocation signal and a binding site to the giant sarcomeric protein titin (29a). It is likely that the mechanisms that regulate Capn3 activity are different from those for the ubiquitous calpains, considering the very low calcium requirement for its function and the difference in interacting proteins. In addition, Capn3 has tissue-specific expression; it is mainly located in skeletal muscle (28). The importance of Capn3 in muscle homeostasis has been pointed out by the observation that its deficiency leads to limb girdle muscular dystrophy type 2A and is associated with a perturbation of the antiapoptotic pathway of NF-κB/IκBα (2, 7, 23). However, the precise function of Capn3 in muscle biology, as well as the identity of its substrates and its activation mechanism, remained largely unknown.

We were interested in investigating the biological role of this protease. Identification of a phenotype of cytoskeleton disruption associated with ectopic expression of Capn3 enabled us to demonstrate that autolysis constitutes its mechanism of activation. As Capn3 is seen mainly in an unprocessed form when extracted from muscle as well as from cultured myotubes (14), we can conclude that Capn3 is mostly in an inactive state. This particularity precluded the identification of substrates from these tissues and forced us to undertake the search in immature muscle cells, which are known to be devoid of full-length Capn3 (12). We showed that Capn3 activity is directed against several cytoskeletal components of the muscle sarcomere and costamere, with which it colocalizes. We propose that Capn3-mediated cleavages in these structures modify the properties of the muscle, which enables it to display efficient physiological response to external and/or internal stimuli.

MATERIALS AND METHODS

Expression constructs and site-directed mutagenesis.

pTOM plasmid containing the enhanced yellow and cyan fluorescent proteins (eYFP and eCFP), respectively, 5′ and 3′ of the multicloning site was obtained after digestion of peYFP and peCFP (Clontech) by XhoI and HpaI and religation of the eCFP fragment in peYFP. Capn3 cDNAs were synthesized by PCRs, with pSRD-Capn3, pSRD-Capn3^C129S, and pSRD-Capn3^ΔExon6 as templates (12). PCR products were subcloned into pTOM after digestion by BamHI and HindIII. Capn3^34-274 and Capn3^323-821 fragments were obtained by PCR on pSRD-Capn3 and were subcloned into pDONR.201 and then into the pDEST40 and pDEST47 plasmids by using Gateway Technology (Invitrogen). The TTN-741/948 fragment was obtained by PCR on human skeletal muscle cDNA and was subcloned into peGFP-CT (Invitrogen). TTN-952/1540, TTN-1607/2167, and Mex5 fragments were obtained by PCR on human skeletal muscle cDNA, and a putative cleavage product of filamin C (FLNC) was obtained by PCR on the IMAGE clone 5169921 (HGMP-MRC Service, Cambridge, United Kingdom). These PCR products were subcloned in the pcDNA3.1/V5-His-TOPO plasmid (Invitrogen). All sequences obtained by PCR were confirmed by automated sequencing.

Site-directed mutagenesis was performed by use of the QuikChange site-directed mutagenesis kit (Stratagene). pSRD-Capn3 was converted to pSRD-Capn3^Y274A by replacing the tyrosine at position 274 with an alanine. pSRD-Capn3 was converted to pSRD-Capn3^Y322A by replacing the tyrosine at position 322 with an alanine. Automated sequencing was carried out to confirm the mutations. Absence of autolysis in Capn3^Y274A was confirmed by Western blot analysis (see Fig. 2C).

FIG. 2.

Capn3 is activated through autolysis in the active site which involves intra- and intermolecular events. (A) Mouse myoblasts were transiently transfected with pSRD-Capn3^ΔExon6 or pSRD-Capn3^Y274A or cotransfected with plasmids coding for the 34-kDa autolysis fragment (pDEST40-Capn3^34-274) and the 55-kDa autolysis fragment (pDEST47-Capn3^323-821). Images were visualized by confocal microscopy. Capn3 was detected with anti-Capn3 RP1. Capn3^34-274 was detected with a monoclonal anti-V5 tag and Capn3^323-821 was detected with eGFP. Actin was labeled with phalloidin and focal adhesions were stained with a monoclonal anti-alpha actinin antibody or a monoclonal anti-vinculin antibody. Bar = 10 μm. (B) Schematic representation of Capn3 proteolytic domain with the C129S, Y274A, and Y322A mutations indicated in green, red, and blue, respectively. (C) 911 cells were transfected with pSRD-wt-Capn3, pSRD-Capn3^C129S, pSRD-Capn3^Y274A, and pSRD-Capn3^Y322A alone or in combination. Twenty-four hours after transfection, cells were harvested and lysed for protein extraction. Western blot analysis of Capn3 was performed with anti-Capn3 RP1. The full-length Capn3 94-kDa form and 58-kDa partially autolyzed and 55-kDa fully autolyzed fragments are indicated by black arrows. The asterisk indicates a nonspecific band detected with a particular batch of the RP1 antibody.

Cell culture, transfection, and cell sorting.

The 911 and NIH 3T3 cell lines were obtained from the American Type Culture Collection (Rockville, Md.). Primary mouse myoblasts were obtained from legs and gluteus muscles of 10- to 15-day-old 129Sv mice according to a previously described protocol (19). Myogenic differentiation of cells was initiated by replacing the growth medium, consisting of Dulbecco's modified Eagle's medium-Ham F-12 (1:1) supplemented with 20% fetal calf serum (FCS), 10 μg of gentamicin per ml, and 2% Ultroser G (BioSepra), with Dulbecco's modified Eagle's medium containing 5% FCS and by maintaining the cells in this medium for 8 days.

For plasmid transfections, cells were harvested, plated at 50% subconfluence (200,000 cells per well in 6-well microtiter plates), and allowed to grow for 24 h. Transfections were performed with 2 μg of plasmid and 6 μl of FuGENE 6 transfection reagent (Roche Applied Science) or 7 μl of ExGen 500 in vitro transfection reagent (Fermentas). In cases of cotransfections, plasmids were mixed at equimolar concentrations. Two days after transfection, cells were digested with trypsin and subjected to cell sorting on a DakoCytomation MoFlo apparatus (Dako) with respect to eYFP fluorescence.

Antibodies.

Mouse monoclonal antibodies against talin (8d4), vinculin (hVIN-1), and alpha actinin (BM-75.2) were purchased from Sigma Chemical Co. Mouse monoclonal antibodies against paxillin and focal adhesion kinase (FAK) were from BD Transduction Laboratories. Mouse monoclonal antibody against filamin A (Ab-1) was from NeoMarkers. Goat polyclonal antibodies against calpain 1 (C-20) and calpain 2 (C-19) were from Santa Cruz Biotechnology. Polyclonal rabbit antibodies RP1 and RP4 raised against Capn3 were purchased from Triple Point Biologics. Mouse monoclonal antibody against titin (NCL-TTN) that was determined to stain in the vicinity of N2A was from Novocastra. Rabbit polyclonal antibody against green fluorescent protein (GFP) was from Medical and Biological Laboratories and mouse monoclonal antibody against the V5 epitope was from Invitrogen. Antibodies against vinexin (rabbit polyclonal [16]), ezrin (rabbit polyclonal [1]), and palladin (mouse monoclonal [22]) were kindly provided by Noriyuki Kioka (Kyoto University, Kyoto, Japan), Paul Mangeat (University of Montpellier, Montpellier, France) and Carol Otey (University of North Carolina), respectively.

Preparation of protein samples and immunoblotting.

Cell cultures were rinsed with 1× phosphate-buffered saline and lysed in protein extraction buffer containing 50 mM HEPES, 1 mM dithiothreitol, 0.1 mM EDTA, 0.1% CHAPS (pH 7.4), and protease inhibitors (complete mini protease inhibitor cocktail; Roche Biomedicals). Muscles from wild-type and Capn3^−/− 6- to 8-week-old mice were quick-frozen in liquid nitrogen, pulverized to a fine powder, and then rapidly solubilized in LDS NuPage buffer (Invitrogen) containing protease inhibitors (complete mini protease inhibitor cocktail; Roche Biomedicals). After sonication and centrifugation at 12,000 × g for 10 min at room temperature, the supernatants were recovered for Western blot analysis and mixed with 1 mM dithiothreitol. Protein concentrations of samples were determined by the Amido-Schwartz methodology (25).

Protein samples were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis in precast 4 to 12% acrylamide gradient gels (NuPage system; Invitrogen) and electrotransferred onto nitrocellulose membranes. Membranes were stained with Ponceau Red for evaluation of protein transfer and probed with primary antibodies against talin (dilution, 1:200), filamin (dilution, 1:100), vinexin (dilution, 1:1,000), ezrin (dilution, 1:1,000), V5 epitope (dilution, 1:5,000), GFP (dilution, 1:2,000), and Capn3 (RP1; dilution, 1:1,000). Detection was performed with sheep anti-mouse or donkey anti-rabbit secondary antibody (dilution, 1:5,000; Amersham Pharmacia) coupled to horseradish peroxidase. Revelation was performed with the Super Signal Pico West kit.

Actin staining and immunohistochemistry.

Twenty-four hours after transfection, cells were fixed with 3.7% formalin for 20 min, permeabilized with 0.2% Triton X-100 for 20 min, and blocked with 20% FCS for 1 h. Cells were incubated with Alexa Fluor 546-labeled phalloidin (Molecular Probes). Primary anti-talin (1:200), anti-ezrin (1:200), anti-alpha actinin (1:200), anti-titin (1:200), anti-vinculin (1:200), and anti-Capn3 (RP1, 1:200, and RP4, 1:100) antibodies were incubated for 2 or 24 h in case of longitudinal section staining. Secondary antibodies conjugated to either Alexa Fluor 488, 546, or 633 (Molecular Probes) were incubated for 1 h at a dilution of 1:1,000. Images were collected with an Axiovert ×100 confocal microscope (Carl Zeiss).

In vivo plasmid delivery.

Endotoxin-free pTOM-Mex5 plasmid was prepared with the EndoFree Megaprep kit (Qiagen). Eight- to 10-week-old mice were injected in the tibialis anterior and the posterior compartment of the limb with 100 and 300 μg of plasmid, respectively. Immediately after injection, transcutaneous electric pulses were applied through two stainless steel plate electrodes placed on either side of the hind limb. Eight square-wave electric pulses were generated by an ECM-830 electropulsator (BTX) with an output voltage of 200 V/cm, a pulse length of 20 ms, and a frequency of pulse delivery of 2 Hz. The tibialis anterior and soleus were dissected and quick-frozen in liquid nitrogen 6 days after injection. Plasmid transduction efficiency was assessed by microscopic observation of eYFP fluorescence prior to freezing. Experimental protocols complied with the European guidelines for the humane care and use of experimental animals.

RESULTS

Capn3 ectopic expression induces cell rounding and leads to disruption of the actin cytoskeleton and disorganization of focal adhesions.

To assess Capn3 function, we used a model of ectopic expression to bypass the putative regulations that were supposed to be present in mature muscle, as suggested by the inactive state of Capn3 in this tissue. Several cell lines, including NIH 3T3 cells, 911 cells, and mouse myoblasts, were transfected with a construct containing full-length Capn3 cDNA tagged with GFP variants and driven by the cytomegalovirus promoter. Control cells were transfected with the same plasmid without Capn3 cDNA.

Most Capn3-expressing cells showed profound morphology alterations consisting of cell rounding and nuclear condensation (Fig. 1A). This phenotype was observed in all cell types tested. We also observed reduced adhesion of transfected cells compared to the control, as assessed by a higher motility in time-lapse video microscopy imaging (data not shown). These cells were negative for terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling as well as for caspase 3 activation, demonstrating that the phenotype was unrelated to apoptosis.

FIG. 1.

Capn3 ectopic expression leads to cell rounding and nuclear condensation associated with disruption of the actin stress fibers and disorganization of focal adhesions. (A) NIH 3T3 mouse fibroblasts were transiently transfected with pTOM-Capn3, encoding Capn3 tagged with two eGFP variants. Images were visualized by confocal microscopy. Transfected cells appear in green. Nuclei were stained with Topro 3 (red). Insert a shows examples of nuclear condensation in transfected cells and the shape of the nucleus in a nontransfected cell. White arrows in the phase image point to round cells. Control cells were transfected with pTOM. Bar = 10 μm. (B) Mouse myoblasts were transiently transfected with pSRD-Capn3 or pSRD-Capn3^C129S as a control. Transfected cells were stained with anti-Capn3 RP1. F-actin was labeled with phalloidin. Images were visualized by confocal microscopy. Two examples of pSRD-Capn3 transfected cells stained for actin are shown. Insert a shows actin aggregates in the cytoplasm of transfected cells. Insert b shows actin fibers in a nontransfected cell. Bar = 10 μm. (C) Mouse myoblasts were transiently transfected with pSRD-Capn3 or pSRD-Capn3^C129S as a control. Transfected cells were stained with anti-Capn3 RP1. Focal adhesions were stained with a monoclonal anti-alpha actinin antibody. Images were visualized by confocal microscopy. Insert a shows loss of the focal adhesion rod shape, and insert b shows normal rod-shaped focal adhesions. Bar = 10 μm.

To evaluate the cytoskeleton organization of Capn3-expressing cells, we examined actin stress fibers, focal adhesions, and the microtubule network in transfected primary mouse myoblasts. For these experiments, plasmids coding for wild-type (wt) Capn3 or for enzymatically inactive Capn3^C129S were used. Labeling of actin with phalloidin conjugates showed a complete absence of stress fibers in round cells, with actin being detectable as small aggregates in the cytoplasm and as a submembranous network at the cell periphery (Fig. 1B, example 1). Interestingly, some Capn3-expressing cells conserved a normal shape but also lacked organized actin fibers (Fig. 1B, example 2). Subsequently, we labeled focal adhesions with an antibody against alpha actinin. Staining showed a reduced number of focal adhesions in transfected cells associated with a loss of the classical rod shape (Fig. 1C). In some cells, colocalization of Capn3 with focal adhesions was observed. Finally, immunostaining of the microtubule network with a specific antibody against alpha tubulin showed no obvious modification (data not shown). Taken together, these results indicate that Capn3 proteolytic activity induces disorganization of the actin cytoskeleton and of focal adhesions.

Autolysis in the active site IS1 domain is required for Capn3 function.

In addition to its proteolytic activity against putative substrates, Capn3 has the unique property among calpains to undergo autolysis in its own catalytic core. We investigated the autolysis contribution to the cytoskeleton disorganization phenotype. For this purpose, we tested the ability of a Capn3 isoform with exon 6 spliced out (Capn3^ΔExon6) and of a Capn3 isoform that is unable to be processed (Capn3^Y274A) to recapitulate this phenotype. Exon 6 encodes two of the autolytic sites (S1 and S2), and Capn3^Y274A was obtained by replacement of tyrosine 274 in S1 with an alanine. Mouse myoblasts expressing Capn3^ΔExon6 or Capn3^Y274A had organized stress fibers and focal adhesions, as seen by staining of actin, alpha actinin, and vinculin (Fig. 2A, ledt and central panels). These results indicate that autolysis within the catalytic site is required for Capn3 function.

It was shown that Capn3 autolysis generates a small N-terminal fragment of 34 kDa and a large C-terminal fragment whose size ranges from 55 to 60 kDa during self-processing (14). When the 34-kDa (Capn3^34-274) and 55-kDa (Capn3^323-821) fragments were coexpressed in mouse myoblast cells, normal cell spreading and intact stress fibers were seen (Fig. 2A, right panel). The same result was obtained when the 34-, 60-, and 55-kDa fragments were expressed separately or in combination (data not shown). Furthermore, we performed colocalization and fluorescence resonance energy transfer analyses and showed that the two fragments remain associated in vivo (data not shown). Taken together, these results strongly suggest that autolysis does not result in the generation of individual fragments with specialized functions.

We were able to further dissect the proteolytic events leading to Capn3 activation by expression of Capn3 forms mutated either in the active site (Capn3^C129S), in S1 (Capn3^Y274A), or in S3 (Capn3^Y322A), followed by Western blotting (Fig. 2B). When the form Capn3^Y274A, which cannot be processed, was expressed alone, a single 94-kDa band corresponding to the full-length protein was observed (Fig. 2C, left panel), indicating that a single mutation in S1 completely abolished autolysis in S2 and S3. Conversely, the Capn3^Y322A form showed a band at 58 kDa (Fig. 2C, left panel), suggesting that a mutation in S3 did not abolish cleavage in S1 and S2. These results indicate that autolysis is sequential and that the cleavage in S1 is the first step of this process. Subsequently, we coexpressed Capn3^Y274A with Capn3^C129S. A single band at 94 kDa was observed, indicating that Capn3^Y274A was not able to lyse Capn3^C129S in S1 (Fig. 2C, center panel). This result suggests that the first autolysis in S1 strictly occurs intramolecularly. Finally, coexpression of wt Capn3 with Capn3^C129S (Fig. 2C, right panel) resulted in the disappearance of the full-length form, indicating that wt Capn3 cleaves Capn3^C129S and that autolysis can occur intermolecularly.

Capn3 activity promotes cleavage of talin, filamin A, vinexin, and ezrin.

With respect to the observed phenotype, we performed Western blot analysis of proteins known to be actin binding proteins or components of focal adhesions. Proteins were selected based on the above criteria and antibody availability. The wt Capn3, Capn3^ΔExon6, and Capn3^Y274A forms were transfected into mouse myoblasts to evaluate their proteolysis abilities. Control cells were transfected with the inactive Capn3^C129S form. To gain sensitivity, purification of the transfected population was performed by cell sorting, using cotransfection with eYFP or taking advantage of Capn3 fused to GFP variants.

Western blot profiles of vinculin, alpha actinin, FAK, calpain 1, calpain 2, and palladin were identical for wt Capn3, Capn3^ΔExon6, and Capn3^Y274A transfected cells compared to those for Capn3^C129S control cells (data not shown). In contrast, Western blot profiles of talin, filamin A, vinexin, and ezrin showed proteolytic cleavage products (Fig. 3A ). A talin 190-kDa fragment identical in size to a previously identified ubiquitous calpain-mediated cleavage product was found in wt Capn3 transfected cells as well as in Capn3^ΔExon6 and Capn3^Y274A forms but not in control Capn3^C129S transfected cells (Fig. 3A, top left panel). Filamin A, which is cleaved as a 220-kDa fragment by ubiquitous calpains, showed a band at the same molecular mass for wt Capn3, Capn3^ΔExon6, and Capn3^Y274A transfected cells but not for control Capn3^C129S cells (Fig. 3A, top right panel). The vinexin profile of wt Capn3 transfected cells showed an additional band at 60 kDa. A vinexin proteolysis product was found in Capn3^ΔExon6 but was not found in Capn3^Y274A or control Capn3^C129S transfected cells (Fig. 3A, bottom left panel). Finally, we detected a 50-kDa cleavage fragment of ezrin in wt Capn3 and Capn3^ΔExon6 transfected cells. Like vinexin, ezrin was not cleaved by Capn3^Y274A and Capn3^C129S (Fig. 3A, bottom right panel).

FIG. 3.

Capn3 cleaves talin, filamins A and C, vinexin, ezrin, and titin Z-disk and M-line domains. (A) Western blot analysis was done on extracts from mouse myoblasts transiently transfected with pTOM-Capn3 (WT), pTOM-Capn3^ΔExon6 (Δexon6), pSRD-Capn3^Y274A/pEYFP (Y274A), and pSRD-Capn3^C129S/pEYFP (C129S). Talin was revealed with a monoclonal antibody. The unprocessed talin form at 230 kDa and a 190-kDa proteolytic fragment are indicated. Filamin A was detected with a monoclonal antibody. The unprocessed filamin A form and a 220-kDa proteolytic fragment are indicated. Vinexin was revealed with a polyclonal antibody. The unprocessed alpha isoform of vinexin at 86 kDa and a proteolytic fragment at 50 kDa are indicated. Ezrin was stained with a polyclonal antibody. The unprocessed ezrin form at 80 kDa and a proteolysis product at 50 kDa are indicated. (B) Organization of titin Z disk, N2A, and M line and filamin C. Gray, black, and white boxes represent repetitive immunoglobulin-like, unique, and Z-repeat regions of titin, respectively. Known relationships between Capn3 and titin are indicated below the corresponding scheme. Cloned fragments are represented by lines with double arrows. The titin domain corresponding to residues 741 to 948 was fused to eGFP. Titin domains corresponding to residues 952 to 1540 and 1607 to 2167 and the Mex5 domain were tagged with the V5 epitope. The 53-kDa COOH terminus of filamin C was tagged with the V5 epitope. (C) Western blot analysis was done on extracts from mouse myoblasts cotransfected with the corresponding plasmids and pSRD-wt-Capn3 (WT), pSRD-Capn3^ΔExon6 (Δexon6), pSRD-Capn3^Y274A (Y274A), or pSRD-Capn3^C129S (C129S) as a negative control. Titin and filamin C domains were revealed with a monoclonal anti-V5 antibody and a polyclonal anti-GFP antibody. The unprocessed form of TTN-741/948 is shown at 65 kDa. A proteolytic 40-kDa fragment was observed for wt Capn3 and Capn3^ΔExon6, but not for Capn3^Y274A and Capn3^C129S transfected cells. TTN-952/1540 shows a single band at 70 kDa with no apparent proteolytic cleavage. Full-length TTN-1607/2167 and a proteolytic product are indicated at 70 and 60 kDa, respectively. The full-length Mex5 domain and a proteolytic fragment are indicated at 38 and 18 kDa, respectively. The full-length filamin C domain and a proteolytic fragment are indicated at 53 and 15 kDa, respectively.

The muscle cytoskeleton components titin and filamin C are also substrates of Capn3.

So far, the only identified Capn3 partner is the giant muscle-specific titin protein, which is a major organizer of the muscle cytoskeleton. Interactions between Capn3 and titin take place in the titin N2A and Mex5 domains (15, 29a). Furthermore, Capn3 has been detected at the Z disk in isolated muscle fibers, suggesting additional interaction with titin in this structure. It was shown previously that the N2A domain is not cleaved by Capn3 (14), but the Mex5 and titin Z-disk regions have not been investigated. To address this issue, we constructed fusions between Mex5, three Z-disk titin domains (TTN-741/948, -952/1540, and -1607/2167) (Fig. 3B), and either the V5 tag or eGFP and expressed them in myoblasts together with Capn3. Western blot analysis using anti-V5 or anti-GFP antibodies was performed to detect proteolytic cleavage products (Fig. 3C). As described previously, proteolysis by Capn3^ΔExon6 and Capn3^Y274A was also investigated. The results showed proteolytic fragments for TTN-741/948, TTN-1607/2167, and Mex5 which were also found with Capn3^ΔExon6 and Capn3^Y274A but not with Capn3^C129S. Interestingly, as for vinexin and ezrin, TTN-741/948 was not cleaved by Capn3^Y274A. No cleavage was detected for TTN-952/1540.

We also tested a putative cleavage product of filamin C (FLNC) by Capn3. Filamin A, which we found to be cleaved by Capn3, is down-regulated during muscle differentiation, while the expression of its paralog, FLNC, increases. In addition, FLNC was proposed to be a Capn3 substrate (35). Accordingly, the hinge region of FLNC was cloned as a fusion protein with the V5 tag (Fig. 3B) and was coexpressed with the various Capn3 forms. The results showed cleavage of FLNC by all three Capn3 forms but not by control Capn3^C129S transfected cells (Fig. 3C).

Capn3 colocalizes with its substrates at the Z disk, M line, costameres, and myotendinous junctions (MTJ) in mouse muscle.

Cleavage events involving talin, vinexin, and ezrin were assessed on endogenous proteins by Western blot analysis in myotubes and in muscle extracts. The same analyses were done on transfected proteins for titin domains and FLNC, except that only Mex5 was tested in vivo on mouse muscles. Cleavage products could not be observed at Western blot detection level under these conditions (data not shown).

Evidence for these proteins to be Capn3 substrates would be supported by in situ colocalization. For this purpose, we performed an extensive study of Capn3 subcellular localization on longitudinal sections of tibialis anterior mouse muscles. Sarcomeric structures including the Z disk, N2A, and M line were localized with antibodies against desmin and titin. MTJ were visualized with respect to their enrichment in vinculin. Four different Capn3 localization areas were observed. (i) Confocal imaging showed localization of Capn3 at the Z disk (Fig. 4A ). (ii) In some fibers, Capn3 staining was detected at the M line, although at a lower intensity and frequency than the Z-disk pattern (Fig. 4B). (iii) Capn3 was localized and enriched in most of the MTJ (Fig. 4C, arrowheads). Interestingly, staining of titin using the N2A antibody showed colocalization with vinculin but not enrichment (data not shown), suggesting an interaction(s) of Capn3 with another partner(s) besides titin at the MTJ. (iv) Localization of Capn3 at the N2A line was also observed (Fig. 4D). In parallel with these experiments, we also investigated a direct colocalization of Capn3 with talin. Both were found to be colocalized at the costameres (Fig. 4E). Finally, since ezrin is expressed in muscles (S. Baghdiguian, personal communication), we performed immunodetection of ezrin and found colocalization with desmin at the Z disk (Fig. 4F). Taken together, these in vivo Capn3 detection results reveal a complex pattern of Capn3 localization that coincides with its substrate localization at titin Z-disk domains, FLNC and ezrin in the Z disk, the Mex5 domain at the M line, and talin in the costameres and MTJ.

FIG. 4.

Capn3 colocalizes with its substrates in vivo at the Z disk, M line, costameres, and MTJ. Longitudinal sections of mouse tibialis anterior muscles were stained with anti-Capn3 RP1 or RP4 antibody.Images were visualized by confocal microscopy. White and green scale bars, 2 and 10 μm, respectively. (A) The Z disk was detected with a monoclonal anti-desmin antibody. Red arrows indicate colocalization (yellow) of Capn3 with desmin at the Z disk. (B) The N2A line was detected with the monoclonal anti-titin NCL-TTN antibody. White arrowheads show Capn3 staining at the M line. (C) MTJ were localized with respect to their enrichment in vinculin and to the vicinity of tendons (red arrow). White arrows indicate colocalization (yellow) of Capn3 with vinculin at the MTJ. (D) The N2A line was detected with the monoclonal anti-titin NCL-TTN antibody. (E) Talin was revealed with a monoclonal antibody. White arrowheads show colocalization (yellow) of Capn3 with talin. (F) Ezrin was localized with a specific rabbit polyclonal antibody. White arrowheads indicate colocalization (yellow) of ezrin with desmin at the Z disk.

DISCUSSION

Proteolysis, which leads to irreversible modifications, is a widely used and powerful means to regulate cellular functions. Since uncontrolled proteolysis may be deleterious for the cell, multiple strategies exist to control such events. Accordingly, Capn3 proteolysis seems to have several levels of regulation. Members of our laboratory previously reported complex transcriptional regulation of Capn3 (12). In the present study, we describe a mechanism of activation whereby Capn3, synthesized as a zymogen, is activated through auto-proteolysis in its active site. We and others have shown that when expressed in ectopic systems, Capn3 is essentially found in an autolyzed state (32). The significance of this auto-processing has remained enigmatic, since it theoretically disrupts the active site. As a consequence, it was interpreted as a degradation process. Our results indicate that Capn3 autolysis should be alternatively considered a mechanism of proteolytic activation. Our evidence is based on the observation of a cytoskeletal rearrangement induced by full-length Capn3. This phenotype was not found when Capn3 was devoid of IS1 or was mutated in the autolytic site S1, indicating the necessity of IS1 domain autolysis in Capn3 function. According to our results, the aim of this autolysis would not be the generation of separated fragments but rather the removal of an internal pro-segment. Subsequently, we showed that intramolecular autolysis in S1, the first event in Capn3 activation, is followed by intermolecular autolysis, as wt Capn3 is able to autolyze Capn3^C129S. This transition from intra- to intermolecular autolysis putatively initiates a signal amplification in which activated Capn3 molecules intermolecularly process inactive zymogens (Fig. 5A).

FIG. 5.

(A) Model for Capn3 activation. Capn3 is mainly present in a full-length inactive state in skeletal muscle, presumably through its binding to titin. Upon receiving an activation signal, a subset of Capn3 molecules undergoes intramolecular autolysis in S1. This first event allows the complete autoprocessing of these molecules, consisting of cleavage of S2 and S3. These fully activated Capn3 molecules can thereafter intermolecularly autolyze other Capn3 molecules which have not received the activation signal. This ultimate step generates an amplification cascade leading to global activation of the Capn3 pool. (B) Model for Capn3 cytoskeleton remodeling in skeletal muscle. The main components of costameres are depicted: integrins alpha and beta, talin, the dystrophin glycoprotein complex (DGC), and dystrophin. Examples of components of the Z disk (alpha actinin) and of the M line (M protein and myomesin) are also shown. Inactive Capn3 receives an activation signal, potentially of mechanical origin and occurring at the N2A line. Upon activation, Capn3 is able to cleave titin (Z-disk and Mex5 domains), filamin C, talin, vinexin, ezrin, and possibly other proteins at the Z disk, M line, and costameres. Regarding our knowledge of the function and structure of these proteins, the cleavage products could produce gains of function, inducing an adaptive response of muscle cells to the initial activation signal.

Interestingly, this is the second example, together with the caspase family, of activation by proteolytic cleavage in a linker region situated within the active site (26). However, Capn3 distinguishes itself from the caspase family, as this event occurs intramolecularly, whereas caspase activation arises intermolecularly after oligomerization (18). For caspases, cleavage of the linker situated within the active site empties the proteolytic cavity, enabling the substrate alignment segment to shift and adopt its active conformation (24). By similarity, we propose that autolysis in IS1 generates a structural transition of the active site, releasing constraints within the molecule and sterically freeing the proteolytic cleft. The activation mechanism for Capn3 may be unique among the calpain family members, as it is the only isoform that possesses an insertion segment within the active site. It is interesting that ubiquitous calpains also use a conformational shift for activation, but in their case it is induced by calcium binding (20). Our data illustrate the convergence of activation mechanisms between members of two different protease families (Capn3 and caspases) and divergence between members of the same family (Capn3 and ubiquitous calpains).

When extracted from muscle, Capn3 is seen mainly in unprocessed form (14) and therefore, according to our results, is presumably mostly in an inactive state in this tissue. Furthermore, when Capn3 was overexpressed in mature muscle, either in transgenic mice (33) or by gene transfer (unpublished data), no obvious phenotype was seen, suggesting that a robust regulator inhibits Capn3 proteolytic activity in this tissue, preventing extended proteolysis. A good candidate for this Capn3 inhibition is its giant partner, titin, possibly through the N2A domain, which binds Capn3 but is not cleaved by it (14). Moreover, it should be noted that titin contains, in the vicinity of the N2A domain, regions of calpastatin homology, which have been proposed to be potential inhibitors of Capn3 (30). Using ectopic systems, we were able to bypass this inhibition and reveal a number of Capn3 substrates, all of which are components of the muscle cytoskeleton. Although Capn3-mediated cleavage products could not be detected by Western blotting in vivo, as a consequence of its mostly inactive state, we can infer that they are potential substrates in mature muscles with respect to their colocalization.

The fact that the limb girdle muscular dystrophy type 2A phenotype is caused by the loss of Capn3 activity (21) reinforces the idea that Capn3-mediated cleavage products are necessary for the homeostasis of skeletal muscle. These substrate cleavage products do not correspond to a degradation process but rather to an irreversible regulatory event that would modulate the function of the substrates through the redistribution of functional domains or by relieving constraints within molecules. This is well illustrated in the case of talin, a cytoplasmic protein that links integrins to the actin cytoskeleton. Proteolysis of talin by ubiquitous calpains induces a 16-fold increased affinity for integrin beta3 (13, 37). As both Capn3 and ubiquitous calpains generate talin cleavage fragments that are identical in size, we can postulate that Capn3 has the same effect on talin-integrin interaction and therefore possibly on clustering and activation of integrin (4). The consequences at the muscle level would be a reinforcement of the linkage between the intracellular cytoskeleton and the extracellular matrix, as it was shown that such activation increased the ligand-binding affinity of integrin (5). In the same line of thinking, it is interesting that Capn3 activity is directed against a unique category of proteins, all components of the cytoskeleton, and that Capn3 is located in costameres and MTJ, which are sites of force transmission. This suggests that Capn3 could participate in an integrated regulation of the muscular cytoskeleton during processes such as force generation, adaptive response to exercise, or stretching or protection during contraction. Another line of evidence in favor of this hypothesis is the observation that deficiency in Capn3 seems to induce a decrease in fiber contraction ability (8).

In parallel to the modification of the mechanic properties of the muscle, Capn3 activity against these substrates could lead to a modulation of signal transduction. For muscle, studies have reported a regulation of signaling going through the cytoskeleton by means of phosphorylation-dephosphorylation, but our report opens up a new field showing that it could be positively regulated by proteolytic cleavages as well. Signals transduce through a complex web of interconnected pathways, especially in a muscle cell with such an organized and complex cytoskeleton. Proteolysis of components of these pathways would cause absolute closure of several possibilities, forcing the signal to go towards a specific direction. Specifically, both integrin clustering and vinexin participate in the regulation of the mitogen-activated protein kinase pathway (3, 34). In addition, it should be noted that recently, Kumar and Boriek identified mechanical stress as an activation signal for NF-κB (17). Considering previous results showing that Capn3 is a key regulator of the NF-κB survival pathway in this tissue (2), this suggests that Capn3 could be a key player in connecting mechanical stress to NF-κB activation. In summary, in response to a particular event that may include an internal mechanical aspect, Capn3 cleavage within important cytoskeletal structures leads to skeletal muscle adaptation by means of modification of its mechanical characteristics and modulation of gene expression.