Oligomer-dependent and oligomer-independent pathogenesis of muscular dystrophy-associated mutations within the penta-EF-hand domain of calpain-3[image]

TL;DR

This study explores how mutations in the PEF domain of calpain-3 cause muscular dystrophy through two distinct mechanisms.

Contribution

The study identifies both oligomer-dependent and oligomer-independent pathways in LGMDR1 pathogenesis.

Findings

Oligomer formation of CAPN3 through the PEF domain is crucial for efficient autolytic processing.

LGMDR1 mutants fail to localize at sarcomeric M-bands due to reduced titin binding, independent of oligomerization.

Mutations in the PEF domain disrupt CAPN3's physiological function via two distinct mechanisms.

Abstract

Limb-girdle muscular dystrophy R1 (LGMDR1) is an autosomal recessive disorder caused by dysfunction of calpain-3 (CAPN3; also known as p94), a muscle-specific, Ca2+-dependent cysteine protease. LGMDR1 mutations are distributed throughout the Capn3 gene. Nevertheless, our knowledge of the biochemical and biological properties of individual LGMDR1 mutants is limited, hindering a full understanding of LGMDR1 pathogenesis. Here, we comprehensively examined the functional properties of LGMDR1 mutants within the penta-EF-hand (PEF) domain at the COOH terminus of CAPN3, focusing on their autolytic processing, oligomerization, titin binding, and subcellular localization within sarcomeres of mouse skeletal muscle. We found that oligomer formation of CAPN3 through the PEF domain contributes to efficient NH2-terminal and IS1-region processing, which were impaired by specific LGMDR1 mutations within the PEF domain. Furthermore, while WT CAPN3 predominantly localized at the sarcomeric M-bands of tibialis anterior muscles in vivo, several LGMDR1 mutants were absent from the M-bands due to decreased binding to titin, a giant cytoskeletal protein, irrespective of their oligomerization status. These findings indicate that LGMDR1 mutations within the PEF domain disrupt the physiological function of CAPN3 through both oligomer-dependent and oligomer-independent mechanisms, highlighting two distinct pathways contributing to LGMDR1 pathogenesis.

Figures

Full text

Limb-girdle muscular dystrophy R1 (LGMDR1) is an autosomal recessive disease characterized by progressive atrophy and weakness of the proximal limb muscles, including the shoulder girdle muscles. LGMDR1 is the most common form of limb-girdle muscular dystrophy and is caused by inactivation of calpain-3 (CAPN3; also known as p94), a Ca^2+^-dependent cysteine protease predominantly expressed in skeletal muscle (1, 2).

Among the 15 mammalian calpain family members, CAPN3 is classified as a classical calpain based on its primary structure (3). Similar to the classical calpains CAPN1 and CAPN2, CAPN3 contains three conserved domains: a calpain-type cysteine protease conserved domain, composed of protease core 1 and 2; a calpain-type β-sandwich (CBSW) domain; and a penta-EF-hand (PEF) domain.

However, CAPN3 contains three additional specific insertional sequences: an N-terminal sequence (NS) at the NH_2_-terminus, insertion sequence 1 (IS1) within the protease core 2 domain, and IS2 between the CBSW and PEF domains. These specific insertion sequences confer distinct features to CAPN3, such as rapid autolysis (2, 4, 5, 6, 7, 8) and intramolecular complementation between the NH_2_- and COOH-terminal (CT) autolytic fragments (9), which together form a catalytic center for substrate hydrolysis. Notably, CAPN1 and CAPN2 exist as heterodimers with a common 28 kDa regulatory small subunit, CAPNS1 (10), whereas CAPN3 forms homodimers (11, 12, 13) and oligomers (14, 15) through the PEF domain.

To date, more than 500 LGMDR1 mutations have been reported (https://www.ncbi.nlm.nih.gov/clinvar/?term=CAPN3%5Bgene%5D&redir=gene). These mutations are distributed not only within the catalytic cysteine protease conserved domain but also throughout the CAPN3 protein. This complicates a precise understanding of how CAPN3 dysfunction contributes to LGMDR1 pathogenesis. Although many studies have examined the relationship between LGMDR1 mutations and their effects on the biochemical properties of CAPN3, particularly on its autolytic activities in vitro (3, 5, 16, 17, 18, 19, 20, 21, 22), our knowledge of the functional properties of CAPN3 in LGMDR1 remains limited (16, 22, 23).

In particular, despite the interesting observation that CAPN3 forms homo-oligomers through the PEF domain (11, 12, 13, 14), the biochemical and physiological roles of these complexes are largely unexplored. Moreover, the question of whether and how LGMDR1 mutations in the PEF domain affect CAPN3, including its activation, protein binding, and subcellular localization in skeletal muscle, remains unanswered.

To address these gaps, we investigated the biological and biochemical roles of CAPN3 homo-oligomers in autolytic processing and subcellular localization within sarcomeres. Specifically, we analyzed the functional properties of various LGMDR1 mutants within the PEF domain.

Results

Abnormal NH2-terminal processing of LGMDR1 variants in the PEF domain of CAPN3

To elucidate the functional properties of LGMDR1 mutants within the PEF domain, we exogenously expressed various enhanced green fluorescent protein (EGFP)-tagged LGMDR1 mutants in HEK293T cells. We then examined their autolytic processing (a hallmark of CAPN3 activation) patterns using a commercially available anti-CAPN3 antibody (Proteintech, 28476-1-AP) and our recently developed anti-AIS1 (autolysis within IS1) antibody, which specifically detects the IS1 autolytic cleavage site of CAPN3 (24). The structure of CAPN3 and the antigen regions of the two antibodies are illustrated in Figure 1A. The AIS1 antibody detected two bands (“fragment a” and “fragment b” in Fig. 1B, lower panel) in lysates from cells expressing WT CAPN3. Notably, the band intensity of “fragment b” was much stronger than that of “fragment a.”Figure 1**Abnormal NH_2_-terminal processing of the LGMDR1 mutants of CAPN3 within the PEF domain.**A, schematic illustration of CAPN3 and its LGMDR1 mutants within the PEF domain. Stars indicate the cleavage sites recognized by the anti-AIS1 antibody. The bar indicates the antigen region (488–666 amino acids encoded by BC146672) of a commercial polyclonal anti-CAPN3 antibody (Proteintech, 28476-1-AP). . C129, H334, and N358 are catalytic centers. Arrows indicate autolytic cleavage sites. B, Western blotting of the cell lysates transfected with WT, protease-inactive form (C129S), and LGMDR1 mutants of EGFP-CAPN3 with anti-CAPN3 (28476-1-AP, upper panel) and anti-AIS1 (lower panel) antibodies. The cell lysates of CAPN3:CS lack the two AIS1-recognizing fragments. C, band intensity ratio of “fragment b” to “fragment a” in the LGMDR1 mutants within PEF domains. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, one-way ANOVA with Dunnett’s multiple comparison test. AIS1, autolysis within IS1; NS, N-terminal sequence; IS, internal sequence; PC, protease core; CBSW, calpain-type β-sandwich; PEF, penta-EF motif; LGMDR1, limb-girdle muscular dystrophy R1; CAPN, calpain.

In contrast to WT CAPN3, we noticed that most LGMDR1 mutants in the PEF domain showed decreased band intensity of “fragment b” relative to that of “fragment a.” This suggests inefficient processing of “fragment a.” In particular, LGMDR1 variants with PEF domain deletions (H690RfsTer9, Q728Ter, N760KfsTer5, and R788SfsTer14) as well as F731S, a missense LGMDR1 mutant, completely abolished “fragment b” (Fig. 1B, lower panel). These LGMDR1 mutants also showed decreased processing activity within the IS1 region, and the amount of pre-autolytic full-length product (migrating at ∼130 kDa) was greater than that of the lower autolytic products (∼55 kDa) in HEK293T cells (Fig. 1B, upper panel). The ratio of the band intensity of “fragment b” to that of “fragment a” for individual LGMDR1 mutants is summarized in Figure 1C.

Oligomer formation of LGMDR1 mutants within the PEF domain

Since these LGMDR1 mutants, except F731S, lack the EF5 domain, which is critical for homodimer formation (13), we speculated that oligomerization of CAPN3 may affect the processing of “fragment a” into “fragment b.” Therefore, we examined the oligomerization capacity of LGMDR1 mutants within the PEF domain.

To accurately evaluate the intrinsic oligomerization ability of the mutants independent of their innate protease activity, we used catalytically inactive CAPN3 (CAPN3:C129S; (CS)) harboring the respective PEF mutations. To detect oligomeric states, we expressed EGFP-tagged LGMDR1 mutants in HEK293T cells, lysed the cells, and cross-linked the proteins with glutaraldehyde prior to SDS–PAGE and Western blotting (14). As shown in Figure 2A, EGFP-CAPN3:CS predominantly existed as oligomers migrating over 250 kDa, while monomers (∼130 kDa) were minimally detected. In contrast, some LGMDR1 mutants negatively affected oligomer formation, predominantly migrating as monomers. The ratio of oligomers to monomers for each LGMDR1 mutant is summarized in Figure 2B.Figure 2**Oligomerization of LGMDR1 mutants of CAPN3 within the PEF domain.**A, Western blotting of LGMDR1 mutant forms of EGFP-CAPN3 with (right panel) or without (left panel) cross-linking by 0.1% glutaraldehyde. B, ratio of the band intensity of oligomer (>250 kDa) to monomer (∼130 kDa) forms of the LGMDR1 mutants within the PEF domain. ∗∗p < 0.01, ∗∗∗p < 0.001, one-way ANOVA with Dunnett’s multiple comparison test. C, 3D structural model of the CAPN3 dimer and position of the I807 amino acid residue (red) within the EF5 loop (Protein Data Bank ID: 4OKH) (13). D, oligomerization of CAPN3:CS and CAPN3:CS:I807T expressed in HEK293T cells. E, relative ratio of oligomer to monomer. ∗p = 0.0121, Mann–Whitney U test. F, Western blotting of the cell lysates expressing WT and the I807T mutant form of EGFP-CAPN3 with anti-CAPN3 (upper panel) and anti-AIS1 (lower panel) antibodies. G, band intensity ratio of “fragment b” to “fragment a” of EGFP-CAPN3:I807T. ∗∗∗p = 0.00084, Mann–Whitney U test. H, scatter plot of the oligomer/monomer ratio against the NH_2_-terminal processing ratio (b/a) of various LGMDR1 mutants within the PEF domain. Mutants in red lack “fragment b” when expressed in HEK293T cells. I807T in light blue is an artificial mutant that lacks “fragment b” in the cell lysates. PEF, penta-EF motif; LGMDR1, limb-girdle muscular dystrophy R1; AIS1, autolysis within internal sequence1; CAPN, calpain; EGFP, enhanced green fluorescent protein.

In accordance with previous studies showing the contribution of the EF5 domain in CAPN3 oligomerization (13, 14), frameshift LGMDR1 mutants lacking this domain showed a significant reduction in the oligomer/monomer ratio. Moreover, the F731S missense mutant also markedly reduced this ratio (Fig. 2B). Remarkably, the pattern of the change in the oligomer/monomer ratio of LGMDR1 mutants (Fig. 2B) was very similar to that of the b/a band intensity ratio in Figure 1C, suggesting a close interaction between CAPN3 oligomerization and NH_2_-terminal processing.

Oligomerization abolishment and abnormal NH2-terminal processing by missense mutant I807T

The EF5 region of CAPN3 has been shown to not only contribute to dimerization but also associate with Ca^2+^ ion binding. This coordination occurs through the side chain carboxylates of D800, D802, and D804, the backbone carbonyl oxygen of I806, and two water molecules (13). To confirm the significance of CAPN3 oligomerization in NH_2_-terminal autolytic processing, we generated a point mutant in which isoleucine 807, predicted to localize at the hydrophobic interface of CAPN3 dimers, was replaced with threonine, a hydrophilic amino acid residue (I807T) (Fig. 2C).

As expected, CAPN3:CS:I807T exhibited decreased oligomer formation compared to CAPN3:CS (Fig. 2, D and E). Furthermore, although EGFP-CAPN3 was predominantly detected as a ∼ 55 kDa band because of extensive autolysis within the IS1 region, EGFP-CAPN3:I807T showed reduced autolysis and primarily appeared as a full-length EGFP-CAPN3 protein (∼130 kDa) in the transfected cell lysates (Fig. 2F, upper panel). When probed with the anti-AIS1 antibody, “fragment b” was absent in EGFP-CAPN3:I807T (Fig. 2F, lower panel), and the band intensity ratio of “fragment b” to “fragment a” was significantly decreased (Fig. 2G). A scatter plot of the oligomer/monomer ratio to the b/a band intensity ratio clearly indicated an apparent positive correlation between the two in LGMDR1 mutants (R = 0.77) (Fig. 2H).

We compared the location of individual PEF mutations in the dimer structure of the PEF domain retrieved from the Protein Data Bank (PDB ID: 4OKH); however, there was no specific region where the missense mutations severely affecting oligomer formation (L712F, F731S, Y763C, and R769Q) are concentrated (Fig. S1A). We also analyzed the change in stability of the LGMDR1-associated mutant homodimers using Molecular Operating Environment (MOE) software (https://www.chemcomp.com/en/index.htm). The effect of each mutant, expressed in two indices representing the shift in affinity between the two EF-hand domains (dAffinity) or stability of dimer complex (dStability), indicates that the mutations impair dimer formation to various degrees (Fig. S1B). However, the obtained values were not quantitatively correlated with our experimental data shown in Figure 2, B and H, which may be a limitation of the prediction analysis using the structure of the PEF domain but not of the full-length CAPN3.

On the other hand, the rationally designed mutant I807T was given high values in both dAffinity and dStability. The precise location of I807 at the monomer–monomer interface could not be determined in the recently solved hexamer structure (15). However, based on the consistency of the data, impaired oligomerization detected by cross-linking assay, and reduction in dimer stability calculated by the MOE software, a critical role for I807 in the assembly of full-length CAPN3 and the isolated PEF domain is suggested.

Taken together, these results indicate that oligomer formation through the PEF domain influences the autolytic processing of the NH_2_-terminal region of CAPN3.

Preferential binding of CAPN3 to the titin CT region over the N2A region in heterologous cells

We previously reported that CAPN3 binds to titin, a giant sarcomeric protein, using yeast two-hybrid screening and protein co-overexpression in heterologous cells. Based on this finding, we proposed that CAPN3 localizes to the N2A and M-band regions of sarcomeres in skeletal muscle through its interaction with titin (25, 26, 27). However, the significance of CAPN3 binding to these two titin regions in determining its subcellular localization remains unclear. Therefore, we examined the binding preference of CAPN3 and assessed how the LGMDR1 mutation affects its subcellular localization in skeletal muscle. Since WT CAPN3 is unstable owing to rapid autolysis in heterologous cells, CAPN3:CS, a mutant that is catalytically inactive and otherwise recapitulates WT properties, was considered a reasonable choice for analysis.

We coexpressed EGFP-CAPN3:CS with mCherry-fusion proteins containing the N2A-containing region (I80-PEVK; amino acids 8469–9141; NCBI Reference Sequence: NP_596869.4) or the CT region of human titin (titin CT; IgC2-9-is7-IgC2-10; amino acids 33,132–33,423) in HEK293T cells (Fig. 3A) and performed a coimmunoprecipitation assay. As shown in Figure 3B, the amount of CAPN3:CS coimmunoprecipitated with mCherry-titin CT was much larger than that with mCherry-titin I80-PEVK. The band intensity ratio of the coimmunoprecipitated CAPN3:CS to total CAPN3:CS in the lysate was approximately 3 to 4 times higher with mCherry-titin CT than that with mCherry-titin I80-PEVK (Fig. 3C).Figure 3**Analysis of CAPN3 binding to the two titin regions in cells with heterologous overexpression.**A, schematic illustration of the mCherry-fusion proteins with the two regions of human titin (NCBI reference sequence: NP_596869.4). The N2A region corresponds to the mouse mCN48 fragment identified as a p94-binding fragment via yeast two-hybrid screening in our previous study (25). B, EGFP-CAPN3:CS is coimmunoprecipitated with mCherry-titin CT more efficiently than with mCherry-titin I80-PEVK in HEK293T cells. Because of their weak interaction, the cells were treated with a dithiobis(succinimidyl propionate) cross-linker before cell lysis and immunoprecipitation. C, band intensity ratio of the coimmunoprecipitated CAPN3 to the CAPN3 in the lysate. ∗p = 0.0211, Mann–Whitney U test. D, immunostaining of EGFP-CAPN3:CS (green) coexpressed with mCherry fusion proteins of the two titin regions (magenta) in HeLa cells. Arrows (upper panel) and arrowheads (bottom panel) show filopodia with no GFP staining. The scale bar represents 20 μm. E, translocation of EGFP-CAPN3:CS to the membrane fractions by mCherry-titin CT-kRas, but not mCherry-titin I80-PEVK-kRas, in HeLa cells. HeLa cells were fractionated into cytosol and membrane fractions. The upper and the lower panels show the expression of EGFP-CAPN3:CS (Blot: anti-CAPN3 antibody), and mCherry-titin-kRas (Blot: anti-RFP antibody), respectively, and the protein bands were labeled with arrows. F, ratio of band intensity of EGFP-CAPN3:CS in the membrane fraction to that in the cytosol fraction. ∗∗p = 0.0073, one-way ANOVA with Dunnett’s multiple comparison test. a.a., amino acid number; LGMDR1, limb-girdle muscular dystrophy R1; CAPN, calpain; EGFP, enhanced green fluorescent protein; CT, COOH terminal; RFP, red fluorescent protein.

To further evaluate CAPN3–titin interactions, we used a membrane relocalization assay (28). mCherry-titin constructs were fused to the kRas membrane-targeting signal (29, 30) and the impact of their expression on the subcellular localization of EGFP-CAPN3:CS was examined in HeLa cells. We chose HeLa cells because they express neither CAPN3 nor titin, and their relatively flat morphology is suitable for analyzing the plasma membrane structure, including filopodia. The resultant mCherry-titin-kRas and control mCherry-kRas localized predominantly at the plasma membrane, including the filopodia (Figs. 3D and S2). Coexpression with mCherry-kRas revealed no significant localization of EGFP-CAPN3:CS at the cell membrane, whereas mCherry-kRas signals were observed at the tip of the filopodia (Fig. 3D, upper panel, arrows). CAPN3 signals gradually declined from the nucleus, possibly the endoplasmic reticulum, toward the plasma membrane of mCherry-kRas–expressing cells. In contrast, coexpression of mCherry-titin and CT-kRas markedly altered CAPN3 localization, with signals congregating at the tips of filopodia and becoming clearly visible at the plasma membrane (Fig. 3D, middle panels).

Interestingly, the coexpression of mCherry-titin I80-PEVK-kRas did not consistently result in strong CAPN3 signals at the tips of filopodia or at the plasma membrane (Fig. 3D, bottom panel, arrowheads). Correlation of the signal intensities between EGFP-CAPN3:CS and mCherry-titin CT-kRas was prominent in the line plot profile (Fig. S3, second panels from the top). In contrast, no such correlation between CAPN3 and mCherry signals was observed in the line plots of cells coexpressing EGFP-CAPN3:CS with either mCherry-kRas or mCherry-titin I80-PEVK-kRas (Fig. S3, top and third panels, respectively). Additionally, no signal correlation was observed when we coexpressed EGFP-CAPN3:CS with mCherry-titin N2A (I81–I83, 8619–8920)-kRas (Fig. S3, bottom panel).

The region covered by titin N2A corresponds to I81 to I83 (mCN48) of mouse titin (31), the particular titin fragment originally identified via yeast two-hybrid screening using CAPN3:CS as bait (Fig. S3, bottom panel). Therefore, our results indicate that the M-band region serves as a more prominent or constitutive localization site for CAPN3.

To further validate the physical association of EGFP-CAPN3 with mCherry-titin fragment-kRas at the plasma membrane, we performed subcellular fractionation of HeLa cells transfected with expression constructs. The cytosolic and membrane fractions were separately analyzed. As shown in Figure 3, E and F, the membrane-to-cytosol ratio of EGFP-CAPN3:CS was significantly increased in cells expressing mCherry-titin CT-kRas compared to those expressing mCherry-kRas. However, such an increase was not observed in cells coexpressing mCherry-titin I80-PEVK-kRas.

In our previous study, we demonstrated that endogenous CAPN3 immunosignals are predominantly detected at the M-band of cultured skeletal myotubes and sarcomeres of resting extensor digitorum longus skeletal muscle, as assessed using a KO-validated anti-CAPN3 antibody (24). Based on this, we focused on the effect of LGMDR1 mutations on CAPN3 binding to titin CT, which corresponds to the M-band region in skeletal muscle sarcomeres.

Decreased CAPN3 binding to the titin CT by LGMDR1 mutations within the PEF domain, irrespective of oligomerization

We coexpressed mCherry-titin CT-kRas with EGFP-CAPN3:CS variants harboring LGMDR1 mutations within the PEF domains in HeLa cells and analyzed their colocalization (Fig. 4). Coexpression of mCherry-titin CT-kRas with EGFP-CAPN3:CS greatly increased the Pearson’s correlation coefficient (r) of their immunosignals compared to mCherry-kRas coexpression (mCherry-kRas: r = ∼0.4; mCherry-titin CT-kRas: r = ∼0.8) (Fig. 4B). However, no significant increase in r values was observed when mCherry-titin CT-kRas was coexpressed with several LGMDR1 deletion mutants in the PEF domain (Q728Ter, H690RfsTer9, N760KfsTer5, and R788SfdTer14) (Fig. 4B). In addition, two missense mutants, EGFP-CAPN3:CS:A702V and EGFP-CAPN3:CS:D705H, did not show a similar increase in r value upon coexpression with mCherry-titin CT-kRas.Figure 4**Localization of LGMDR1 mutants of EGFP-CAPN3 and mCherry-titin CT in HeLa cells.**A, immunostaining of EGFP-CAPN3:CS (green) and its LGMDR1 mutants coexpressed with mCherry-titin CT (magenta) in HeLa cells. The scale bar represents 10 μm. B, quantification of the colocalization of EGFP-CAPN3:CS and its LGMDR1 mutants with mCherry-titin CT. Pearson correlation coefficient was calculated using Coloc2 software. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, one-way ANOVA with Dunnett’s multiple comparison test. LGMDR1, limb-girdle muscular dystrophy R1; CAPN, calpain; EGFP, enhanced green fluorescent protein; CT, COOH terminal.

Because EGFP-CAPN3:CS:A702V and EGFP-CAPN3:CS:D705H are at least capable of homo-oligomerization (Fig. 2), the reduced r values observed cannot necessarily be attributed to defective oligomer formation. This notion was supported by the finding that EGFP-CAPN3:CS:F731S, which has a defect in its oligomerization ability, showed increased r values when coexpressed with mCherry-titin CT-kRas, similar to EGFP-CAPN3:CS (Fig. 4B). These results indicate that certain LGMDR1 variants exhibit decreased binding affinity to titin CT, irrespective of their oligomerization ability.

To complement these cell-based assay results, we also performed a pull-down assay using a maltose-binding protein (MBP) fusion protein with titin CT (MBP-titin CT) and the HEK293T cell lysates transiently expressing EGFP-CAPN3:CS and its derivatives (Fig. S4). However, all LGMDR1 mutants examined showed a lower binding ability to MBP-titin CT than EGFP-CAPN3:CS without a distinctive feature in the extent (Fig. S4B). Thus, it seemed that the relocalization assay is superior to the pull-down assay for simulating the native CAPN3-titin CT interaction in living cells and evaluating the differential effect of mutations therein.

Subcellular localization of EGFP-CAPN3 in the tibialis anterior muscles of mice in vivo

We have previously reported that transiently expressed EGFP-CAPN3 predominantly localizes at the sarcomeric M-band of resting cultured myotubes (32). To further investigate the subcellular localization of CAPN3 in vivo, we injected adeno-associated virus (AAV) encoding EGFP or EGFP-CAPN3 in the tibialis anterior (TA) muscles of CAPN3-deficient (Capn3^−/−^) mice and analyzed localization after 4 weeks of infection.

While EGFP was diffusely localized within the sarcomere (Fig. 5, top panels), we observed strong EGFP-CAPN3:CS immunosignals at the M-band, situated at the center between actinin immunosignals at the Z-bands (Fig. 5, second-row panels). This specific M-band localization of CAPN3 was independent of its protease activity, since WT CAPN3 (EGFP-CAPN3) presented similar signals at the M-band (Fig. 5, third row panels). Furthermore, an EGFP-CAPN3 mutant lacking the IS2 region, which is believed to mediate binding to the N2A region of titin (EGFP-CAPN3:CS:ΔIS2) in vitro, was also detected at the sarcomeric M-bands (Fig. 5, bottom panels).Figure 5**Localization of EGFP-CAPN3 in TA muscles of Capn3^−/−^ mice following AAV-mediated expression.**A, immunohistochemistry of AAV-infected EGFP, EGFP-CAPN3, EGFP-CAPN3:CS, EGFP-CAPN3:CS:DIS2 (GFP, green), and actinin (magenta) in Capn3^−/−^ mice (left panels). Right panels show intensity plots of EGFP and actinin immunoreactivity along the line in each merged image. The scale bar represents 10 μm. B, immunohistochemistry of endogenous CAPN3 and actinin in the gastrocnemius muscle of 12-week-old Capn3^+/+^ and Capn3^−/−^ mice. Tissues were stained with anti-CAPN3 (green, Proteintech 28476-1-AP) and anti-actinin (magenta) antibodies (left panels). Right panels indicate the line plot of CAPN3 and actinin immunosignals within the merged images. The scale bar represents 10 μm. CAPN, calpain; EGFP, enhanced green fluorescent protein; AAV, adeno-associated virus; CT, COOH terminal; TA, tibialis anterior.

No noticeable differences were detected in subcellular localization between EGFP-CAPN3:CS and EGFP-CAPN3:CS:ΔIS2. The M-band localization of EGFP-CAPN3 and its derivatives resembled endogenous CAPN3 localization in the gastrocnemius (Fig. 5B) and extensor digitorum longus muscles, as detected by the KO-validated CAPN3 antibody (24). Similar M-band localization of EGFP-CAPN3 and EGFP-CAPN3:CS was observed in the TA muscles of infected Capn3^+/+^ mice (Fig. S5A).

Collectively, these results suggest that AAV-infected CAPN3 predominantly localizes at the M-band of sarcomeres in resting skeletal muscles, mirroring the subcellular localization of endogenous CAPN3.

Subcellular localization of LGMDR1 mutants within the PEF domain in sarcomeres of the Capn3−/− mouse skeletal muscles in vivo

We next expressed the EGFP-tagged LGMDR1 mutants within the PEF domain in the TA muscles of Capn3^−/−^ mice using AAV-mediated delivery. Capn3^−/−^ mice were used to exclude the potential influence of endogenous CAPN3 on the subcellular localization of the exogenously expressed LGMDR1 variants. To prevent autolytic degradation of the LGMDR1 mutants, we also expressed an EGFP-fusion protein with inactive CAPN3:CS, harboring either missense mutants (CS:A702V, CS:D705H, CS:D707G, CS:L712F, CS:F731S, CS:744, CS:R748Q, CS:Y763C, CS:D780H, CS:R769Q, CS:A798E, and CS:I807T) or frameshift mutants (CS:H690RfsTer9, CS:Q728Ter, CS:N760KfsTer5, and CS:R788fsTer14) of LGMDR1.

We observed two predominant types of subcellular localization for these LGMDR1 mutants: at the M-band (Fig. 6, A and B) and around the N2A region (Fig. 6C) of the sarcomere. Most missense LGMDR1 mutants, except for CAPN3:CS:A702V and CS:D705H, predominantly localized at the M-band, similar to CS (Fig. 6A). Notably, CAPN3:CS:R748Q, CS:D780H, and CS:R769Q exhibited broader immunosignal distributions around the M-bands than CAPN3:CS (Fig. 6B). This was confirmed by the line profiles of CAPN3 and actinin signals (Fig. 6B). In contrast, frameshift mutants lacking the PEF domain (CS:H690RfsTer9, CS:Q728Ter, CS:N760KfsTer5, and CS:R788fsTer14), as well as two missense mutants (CS:A702V and CS:D705H), were predominantly localized around the N2A regions of the sarcomeres.Figure 6**Subcellular localization of LGMDR1 mutants of EGFP-CAPN3:CS in Capn3^−/−^ mice TA following AAV-mediated expression.**A–C, immunohistochemistry of the AAV-infected LGMDR1 mutant form of EGFP-CAPN3:CS (anti-GFP antibody, green) and anti-actinin (magenta) antibody (left panels) in Capn3^−/−^ mice. Right panels show the intensity plot of EGFP and actinin immunoreactivity along the line in each merged image. Three patterns of CAPN3 immunosignals were observed in the LGMDR1 mutants within the PEF domain: fine M-band signals (A), broadened M-band signals (B), and signals around the N2A region of sarcomeres (C). The scale bar represents 10 μm. LGMDR1, limb-girdle muscular dystrophy R1; CAPN, calpain; EGFP, enhanced green fluorescent protein; AAV, adeno-associated virus; TA, tibialis anterior; PEF, penta-EF hand.

This abnormal localization appeared to be independent of oligomer formation, as CS:A702V and CS:D705H could at least form oligomers (Fig. 2). Conversely, EGFP-CAPN3:CS:S731F and CS:I807T, which are defective in oligomer formation, localized predominantly at the M-band (Figs. 6A and S5B), similar to EGFP-CAPN3:CS (Fig. 5A).

Finally, we examined the intracellular localization of EGFP-CAPN3:CS:A702V and CS:D705H in the sarcomere of TA muscles of Capn3^+/+^ mice. We speculated that endogenous CAPN3 forms hetero-oligomers with EGFP-CAPN3:CS:A702V and CS:D705H and rescues their abnormal localization in sarcomeres of Capn3^+/+^ mice. When the AAV encoding EGFP-CAPN3:CS:A702V was injected into TA muscles of Capn3^+/+^ mice, localization of EGFP-CAPN3:CS:A702V at the M-bands of the sarcomeres was detected. The width of signals was much broader than that of EGFP-CAPN3:CS, indicating the partial rescue of EGFP-CAPN3:CS:A702V abnormal localization (Fig. 7A, upper panels). However, such a rescue of mislocalization was not observed when we expressed EGFP-CAPN3:CS:D705H in TA muscles of Capn3^+/+^ mice (Fig. 7A, lower panels).Figure 7**Partial rescues of abnormal intracellular localization of EGFP-CAPN3:CS:A702V but not of EGFP-CAPN3:CS:D705H in Capn3^+/+^ mice.**A, immunostaining of actinin (magenta) and EGFP-CAPN3:CS:A702V and EGFP-CAPN3:CS:D705H (green) in the TA muscles of Capn3^+/+^ mice. Right panel shows the line plot of the merged image. The scale bar represents 10 μm. B, coexpression of CAPN3:CS ameliorated the localization of EGFP-CAPN3:CS:A702V but not EGFP-CAPN3:CS:D705H at the mCherry-titin CT (magenta) enriched plasma membrane in HeLa cells. Right panels show the line plot of mCherry-titin CT (magenta) and EGFP-CAPN3:CS, CAPN3:CS:A702V, and CAPN3:CS:D705H (green) within the left merged images. C, quantification of the colocalization of EGFP-CAPN3:CS and its LGMDR1 mutants with mCherry-titin CT upon coexpression of CAPN3 in HeLa cells. Pearson correlation coefficient was calculated using Coloc2 software. ∗∗∗p = 0.0019, one-way ANOVA with Bonferroni multiple comparison test. LGMDR1, limb-girdle muscular dystrophy R1; CT, COOH terminal; CAPN, calpain; EGFP, enhanced green fluorescent protein; TA, tibialis anterior.

To determine if WT CAPN3 could similarly rescue the mislocalization of these mutants in a heterologous system, we performed the membrane relocalization assay in HeLa cells. When we coexpressed EGFP-CAPN3:CS:A702V with CAPN3:CS, EGFP-CAPN3:CS:A702V localized at the membrane where mCherry-titin CT-kRas was enriched (Fig. 7B, middle panels). On the contrary, the localization of EGFP-CAPN3:CS:D705H was not significantly affected by coexpression of CAPN3:CS (Fig. 7B, bottom panels). These differences in the change of localization by coexpression of CAPN3:CS were clearly reflected in the r value: the r value of immunosignals of EGFP-CAPN3:CS:A702V and mCherry-titin CT-kRas, but not EGFP-CAPN3:CS:D705H and mCherry-titin CT-kRas, became comparable to that of EGFP-CAPN3:CS and mCherry-titin CT-kRas (Fig. 7C). At present, we do not have a reasonable explanation for the difference between EGFP-CAPN3:CS:A702V and CAPN3:CS:D705H in these recovery assays. Since the D705H mutant has slightly lower homo-oligomer–forming ability than the A702V mutant (Fig. 2B), D705H might not efficiently restore the titin binding upon heteromerization with endogenous CAPN3, resulting in poor rescue of localization measured in the above experiments. Alternatively, D705H might dominantly interfere with the CAPN3 binding to titin CT and negate the binding activity of the heteromerized endogenous CAPN3.

Altogether, these results suggest that CAPN3 localizes at the M-bands in resting skeletal muscle. Moreover, specific LGMDR1 variants within the PEF domain fail to localize correctly at the M-band, irrespective of their ability to form oligomers.

Discussion

Recent biochemical and structural studies have shown that CAPN3 is a unique Ca^2+^-dependent cysteine protease that forms homodimers and oligomers through the PEF domain in a tail-to-tail orientation, with the catalytic domains placed at opposite ends (12, 13, 14, 15, 33). However, the biological and physiological roles of the CAPN3 oligomers remain unknown.

In this study, we comprehensively analyzed the impact of LGMDR1 mutations within the PEF domain on CAPN3 oligomerization, autolytic activity, titin binding, and subcellular localization in mouse TA muscles. We found a close association between defective oligomerization and abnormal NH_2_-terminal processing of CAPN3. In addition, we demonstrated that certain LGMDR1 variants within the PEF domain decreased their binding to titin CT in HeLa cells, irrespective of oligomerization. This reduced binding was critically correlated with the absence of these variants from the M-band region of sarcomeres in vivo. These findings suggest that oligomerization through the PEF domain is critical for CAPN3 processing and activation but not for subcellular localization at the M-band of skeletal muscle.

CAPN3 is an apoenzyme that requires autolytic processing for limited substrate proteolysis (5). Autolysis depends on specific internal sequences within the NS and IS1 regions (4, 11). A previous study using a recombinant protease core domain containing NS and IS1 showed that autolysis within these regions is an intramolecular event (4). In the present study, we found a linear correlation between CAPN3 oligomerization and NS-domain processing, from “fragment a” to “fragment b,” in HEK293T cells (Fig. 2H). Therefore, oligomer formation is necessary for NS cleavage in living cells. In conventional calpains, for example, calpain-1 and calpain-2 (heterodimers CAPN1/CAPNS1 and CAPN2/CAPNS1, respectively), limited autolysis of the N-terminal region of the large subunit lowers the Ca^2+^ requirement for activation (34, 35, 36). Similarly, NS region processing in CAPN3 may modulate its proteinase activity. Notably, oligomer formation also appeared to influence IS1 region processing in heterologous cells. Oligomer-deficient LGMDR1 variants (F731S, H690RfsTer9, Q728Ter, N760KfsTer5, and R788fsTer14) showed increased levels of full-length CAPN3 (100–130 kDa) compared to those of the cleaved protein (50–58 kDa). In contrast, WT CAPN3 showed stronger accumulation of the cleaved fragment (Fig. 1B). These results suggest that the oligomerization of CAPN3 through the PEF domain significantly enhances its proteolytic activity, and further studies are warranted to clarify the physiological significance of NS region cleavage in CAPN3 activity.

Using yeast two-hybrid screening and coimmunoprecipitation of proteins expressed in heterologous cells, we previously reported that CAPN3 binds to two regions of the giant protein titin: the N2A and CT regions (25, 26, 27). However, a comparative analysis of the binding ability of CAPN3 to these two regions, specifically in terms of quantity and physiological relevance, has not yet been conducted. One exception is a previous in vitro study employing a solid-phase ELISA with purified CAPN3 and titin fragments, which showed that the C-terminal fragment of titin is a more competent binding partner than the N2A region (37). Since endogenous CAPN3 is predominantly detected at the M-band, and not in the vicinity of N2A in resting skeletal muscle (24), we analyzed its relative binding affinity to these two titin regions in living cells. We found that CAPN3:CS was coimmunoprecipitated more efficiently with mCherry-titin CT than with mCherry-titin I80-PEVK in HEK293T cells (Fig. 3, B and C).

As an alternative measure to evaluate the efficacy of these interactions, we established a cellular assay system in which the titin fragment of interest was expressed at the plasma membrane by introducing a membrane localization signal sequence of kRas at its C terminus (28). Compared to the yeast two-hybrid assay, this system reduces the likelihood of false negatives due to cellular toxicity caused by interactions among the proteins of interest. Coexpression of mCherry-titin CT-kRas led to a marked relocalization of CAPN3:CS from the cytoplasm to the plasma membrane and filopodia, as seen by immunostaining (Fig. 3D), which consolidated the results of subcellular fractionation (Fig. 3, E and F). However, regardless of the coimmunoprecipitation of CAPN3 with mCherry-titin I80-PEVK, CAPN3:CS did not show colocalized immunosignals with mCherry-titin I80-PEVK-kRas or enrichment in the membrane fraction (Fig. 3, D–F). This suggests that CAPN3 binding to titin CT may be significantly stronger than that to titin I80-PEVK in living cells. Our data are consistent with a previous study showing that CAPN3:CS binds more strongly to the COOH fragment M6–M10 (NM_133378; 101,254–103,252 bp) than to the titin N2A region (NM_133378; 28,316–29,860 bp) in a solid-phase ELISA (37). Collectively, these findings support the view that endogenous CAPN3 is strongly localized at the M-band in resting skeletal muscle through binding with titin CT (24).

A previous study established transgenic mice on a Capn3^−/−^ background to investigate the pathogenesis of two missense LGMDR1 mutations, D705G and R448H, both of which showed reduced titin-binding ability in vitro (22, 37). The R448H variant showed rapid degradation in muscle extracts of the transgenic mice compared with the WT CAPN3 protein, probably owing to the decreased myofibrillar binding of the R448H protein and its degradation by other cellular proteases. However, because of the low expression levels, the precise subcellular localization of CAPN3:R448H signals in vivo could not be confirmed. Furthermore, the D705G variant lacked sufficient stability for generating D705G transgenic mice.

In the present study, we expressed the protease-inactive form of LGMDR1 mutants within the PEF domain in skeletal muscle using AAV. This allowed us to specifically focus on their subcellular localization in skeletal muscle, independent of their intrinsic protease activity. Using this approach, we found a strong correlation between the reduced colocalization of certain LGMDR1 variants and mCherry-titin CT-kRas in HeLa cells and the absence of CAPN3 signals from M-bands in skeletal muscle in vivo. The relocalization assay in HeLa cells, but not the pull-down assay, also demonstrated the difference in titin CT binding to LGMDR1 mutants within the PEF domain (Figs. 4 and S4). Although the deletion of the PEF domain significantly perturbed the oligomeric state and subcellular localization of the mutants, impaired oligomerization did not necessarily trigger the detachment of CAPN3. The LGMDR1 missense variant F731S and the artificially produced missense mutant I807T, which abolished oligomerization, did in fact localize to the M-bands of sarcomeres (Figs. 6A and S5B). These data provide strong evidence that CAPN3 binding to titin CT mediates the subcellular localization of CAPN3 at the M-bands in resting skeletal muscle. Moreover, detachment of certain LGMDR1 variants from the M-bands could be a key mechanism in LGMDR1 pathogenesis.

Detection of EGFP-CAPN3:CS at the M-bands of sarcomeres in TA muscles after AAV transduction (Fig. 5A) contrasts with the findings of a previous study, which reported EGFP-CAPN3:CS localization at both the N2A (a major band) and M-bands (a minor band) in the flexor digitorum brevis muscles 3 days after electroporation of EGFP-CAPN3 variants (22). In that study, the electroporation of EGFP-tagged WT CAPN3, D705G, and R448H caused myofibrillar structural destruction in the muscles, preventing accurate assessment of their subcellular localization. Transgenic mice expressing WT CAPN3 and infected with AAV encoding CAPN3 show neither toxicity nor cytoskeletal disorganization in the skeletal muscle (38, 39, 40, 41, 42). This suggests that electroporation may lead to excessive CAPN3 overexpression beyond the titin-binding capacity in skeletal muscle. Alternatively, electroporation may cause slight muscle damage, including myofibrils, and a 3-day recovery period may not be sufficient for proper localization of CAPN3 in the muscles.

The subcellular localization of CAPN3 expressed by AAV has not yet been precisely analyzed, and only sarcomeric patterns of CAPN3 immunosignals have been reported at the macroscopic level (42). Our findings indicate that AAV-expressed EGFP-tagged CAPN3 localizes to the M-bands of skeletal muscle, similar to endogenous CAPN3 in resting skeletal muscle. This, coupled with the observation that CAPN3:CS binds more efficiently to the M-band fragment than to the N2A fragment of titin in vitro (37), suggests that the M-band is a major site for CAPN3 localization in resting skeletal muscle.

Importantly, we detected at least three immunosignal patterns for LGMDR1 variants in skeletal muscle: i) normal sharp M-band signals (D707G, L712F, F731S, S744G, Y763C, and A798E), ii) slightly broadened M-band signals (D780H, R769Q, and R748Q), and iii) signals localized around the N2A region (A702V, D705H, H690RfsTer9, Q728Ter, N760KfsTer5, and R788SfsTer14) (Fig. 6). Although the cause of the broadened immunosignals (D780H, R769Q, and R748Q) is unclear, they may result from impaired oligomerization of these LGMDR1 mutants. Alternatively, since Capn3^−/−^ mice exhibit M-band abnormalities at the electron microscopic level, the broadened signals of these LGMDR1 variants might reflect incomplete recovery of the M-band. Therefore, further studies are required to clarify this.

Additionally, we detected LGMDR1 variants around the N2A region in vivo (Fig. 6C). These variants lacked titin CT binding in HeLa cells and showed consistent absence of immunosignals at the M-bands, suggesting a shift in localization. Inactive Capn3 knock-in mice are known to exhibit greatly improved sarcomere ultrastructure compared with Capn3^−/−^ mice (32). Thus, the absence of CAPN3 at the M-bands plays a key pathogenic role, irrespective of the protease activity of these mutants. The abnormal localization of LGMDR1 variants around the N2A region may also contribute to LGMDR1 pathogenesis. However, similar to the previously reported instability of the D705G variant in transgenic mice (22), these variants may be degraded before accumulating around the N2A region because of residual intrinsic proteinase activities or targeting by other proteases.

The lack of colocalization between EGFP-CAPN3:CS and mCherry-titin I80-PEVK-kRas or mCherry-titin N2A-kRas in HeLa cells was surprising. One potential explanation is that the colocalization assays in heterologous cells have inherent limitations in recapitulating native protein–protein interactions, particularly those between CAPN3 and the titin N2A region. The titin N2A region has a unique mode of elasticity and possesses multiple binding sites for various proteins (43). In line with this, CAPN3 deficiency was initially predicted to be involved in the pathogenesis of muscular dystrophy with myositis (mdm), which harbors an in-frame deletion in the titin N2A region that overlaps with the CAPN3 binding site. However, introducing this mutation in Capn3^−/−^ mice showed no change in disease progression or severity, which emphasized the complexity of protein–protein interactions at the N2A region (44). Considering the undetectability of endogenous CAPN3 immunosignals around N2A in resting muscles (24) and the absence of EGFP-CAPN3 at the N2A region of skeletal muscle sarcomeres, the physiological relevance of CAPN3–N2A interactions remains uncertain and warrants further examination.

In summary, we demonstrated that CAPN3 oligomerization through the PEF domain plays a critical role in its processing and that certain LGMDR1 mutations within the PEF domain compromise both. Although we focused on PEF domain mutations in this study, missense mutations outside this domain can also disrupt CAPN3 processing by altering its oligomeric state, particularly through domains such as CBSW (14). Impairment of oligomerization by LGMDR1 mutants, despite being located far from the catalytic center, may partially explain the widespread distribution of LGMDR1 mutations across the Capn3 gene. Since our oligomer assay can be performed using mouse tissue lysates (14), it may also be applicable for human samples to predict LGMDR1 pathogenesis. We also confirmed the dominant subcellular localization of exogenously expressed EGFP-CAPN3 at the sarcomeric M-bands via AAV-mediated expression. This aligned with the subcellular localization of endogenous CAPN3 in resting skeletal muscle. In addition, certain LGMDR1 variants did not localize at the M-bands in resting skeletal muscle, which strongly correlated with decreased CAPN3 binding to titin CT, irrespective of their oligomerization ability. Capn3^−/−^ mice are known to exhibit more severe phenotypes, including abnormal sarcomere organization, compared with CAPN3:CS knock-in mice (32, 37, 45). This suggests that disrupted binding to titin CT, instead of (or in addition to) N2A, could underlie the apparent loss of CAPN3 protein in patients with corresponding mutations. These insights may help refine genotype–phenotype relationships to better explain the variable disease severity observed in patients with LGMDR1.

It is worth noting that WT CAPN3 cleaves titin CT upon coexpression in heterologous cells (46). Therefore, our present CAPN3-titin CT binding assay does not fully replicate endogenous CAPN3–titin interactions in mature muscles and may require additional factors, including accessory proteins and appropriate cytosolic calcium levels around the M-band in the muscle. Furthermore, while the AAV experiments in mouse skeletal muscle support the hypothesis that CAPN3 localization is dependent on interaction with titin CT, additional in vivo functional validation studies—evaluating sarcomere structure via electron microscopy, assessing muscle strength, or measuring CAPN3’s protease activity—would strengthen our conclusions regarding the relationship among LGMDR1 mutations, CAPN3–titin CT interaction, and the LGMDR1 pathogenesis.

Altogether, our findings demonstrate the oligomer-dependent and oligomer-independent effects of PEF domain mutations on the physiological functions of CAPN3, advancing our understanding of LGMDR1 pathogenesis. Further studies on the mechanisms underlying CAPN3 oligomerization and maintenance of sarcomere organization by CAPN3 at the M-band would help explore the new pharmacological and genetic strategies for treating LGMDR1.

Experimental procedures

Construction of expression vectors

Primers used in this study are listed in Table S1. The pAAV-CMV-EGFP-CAPN3 plasmid was constructed as follows. A fragment encoding the CMV-EGFP-polyA sequence was amplified via PCR using KOD Plus DNA polymerase (Toyobo), primers NotI-CMV-F and NotI-SV40polyA-R, and pEGFP-C1 vector (Takara Bio Inc) as a template. The amplified fragment and pAAV-Syn-GFP (Addgene plasmid #58867; generously provided by Dr Edward Boyden) were digested with Not I and then ligated to obtain pAAV-CMV-EGFP-polyA.

To construct the pAAV-CMV-EGFP-CAPN3 vector, pEGFP-CAPN3 and pEGFP-CAPN3:CS (47) were digested with XhoI and BamHI, and the XhoI-BamHI fragment encoding CAPN3 or CAPN3:CS was inserted into the BamHI-XhoI site of the pAAV-CMV-EGFP-polyA vector. Mutations in pAAV-CMV-EGFP-CAPN3 were introduced via site-directed mutagenesis and in-fusion cloning using the In-Fusion Snap Assembly Master Mix (Takara Bio Inc, 638947). The EcoRI-BamHI fragment containing each LGMDR1 mutation was amplified using specific primers: fragment 1 FW Primer (EcoRI), fragment 2 RV Primer (BamHI), and LGMDR1 mutant primers. pAAV-CMV-EGFP-CAPN3 was used as a template. The amplified product was inserted into the EcoRI-BamHI site of pAAV-CMV-EGFP-CAPN3 through in-fusion cloning.

For mCherry-kRas, annealed primers of mCherry-kRas-S and mCherry-kRas-AS, which encode the CT amino acid sequence of kRas4B (MSKDGKKKKKKSKTKCVIM∗) (29, 30), were ligated into the XhoI*-Eco*RI site of mCherry-C1.

To construct pmCherry-titin CT, the minimal CAPN3-binding region of human titin (IgC2-9-is7-IgC2-10) (26) was amplified using two primers (titin IgC2-9-S and titin IgC2-10-AS) with the pACT2-CN5 vector (25) as template. The amplified fragment was cloned into the XhoI*-Eco*RI site of mCherry-C1 via in-fusion cloning.

For mCherry-titin CT-kRas, the same region (IgC2-9-is7-IgC2-10) (26) was first amplified using appropriate primers, titin IgC2-9-S and titin IgC2-10-kRas-AS1. Then, the amplified fragment, IgC2-9-is7-IgC2-10-kRas, was used as a template and further amplified using primers titin IgC2-9-S and kRas-AS2. The resulting fragment was inserted into the XhoI-EcoRI site of pmCherry-C1 (Takara Bio Inc.) through in-fusion cloning.

To construct pmCherry-titin I80-PEVK, a fragment spanning amino acids 8469T–9141E of human titin was amplified using primers human titin I80-S and human titin PEVK-AS, with a human skeletal muscle complementary DNA library (25) as template. The fragment was cloned into the XhoI*-Eco*RI site of mCherry-C1 through in-fusion cloning.

For constructing mCherry-titin N2A-kRas, the human titin N2A region was amplified using primers mCherry-titin N2A-S and kRas-titin N2A-AS. The product, an mCherry-titin N2A-kRas fragment, was reamplified using mCherry-titin N2A-S and kRas-AS2 and then cloned into the XhoI*-Eco*RI site of mCherry-C1 through in-fusion cloning.

To construct a MBP fusion protein with titin CT (MBP-titin CT), the titin CT fragment amplified with c6T-titinCT-S and c6T-titinCT-AS primers was inserted into the NotI-SalI site of pMal-c6T vector (New England Biolabs Japan Inc).

The sequences were verified using the SupreDye v3.1 Cycle Sequencing Kit (M&S TechnoSystems). The amino acid residue numbers are based on the NCBI Reference Sequence: NP_596869.4.

Antibody

The antibodies used in this study included rabbit anti-CAPN3 polyclonal antibody (Proteintech, Tokyo, Japan, 28476-1-AP), anti-α-actinin (Sigma-Aldrich, EA-53), anti-GFP polyclonal antibody (MBL, Tokyo, Japan, Code No. 598), and anti-red fluorescent protein monoclonal antibody cocktail (MBL, Code No. M208–3), and anti-mCherry antibody (Proteintech, 26765-1-AP). The AIS1 antibody has been described in our previous study (24).

Transfection and immunoprecipitation

HeLa and HEK 293T cells were cultured in Dulbecco’s modified Eagle medium (Nacalai Tesque, 08458-16) supplemented with 10% fetal bovine serum, 100 units/ml of penicillin, and 100 μg/ml of streptomycin (Nacalai Tesque, 09367-34). Approximately 5 × 10^5^ cells were transfected with the appropriate plasmids in a 6 cm dish using PEI-MAX 40k (Polysciences Inc). After 36 h, the cells were lysed with a lysis buffer containing 10 mM Tris–HCl (pH = 7.0), 150 mM KCl, 1 mM EGTA-KOH (pH = 8.0), and 1.0% Triton, supplemented with cOmplete Protease Inhibitor Cocktail (Sigma-Aldrich, 04693116001) for 15 min on ice. The lysates were centrifuged at 20,630g for 10 min, and the resulting supernatants were mixed with 4×SDS–PAGE sample buffer and boiled at 95 °C for 5 min. For protein cross-linking, the supernatants were treated with 0.1% glutaraldehyde for 10 min at room temperature (RT) before the addition of 4×SDS–PAGE sample buffer, as previously described (14).

For immunoprecipitation, transfected cells were treated with PBS containing 1 mM dithiobis(succinimidyl propionate) (Thermo Fisher Scientific K.K, 22585) for 30 min at RT. The reaction was quenched by adding 100 mM Tris–HCl (pH 7.5) for 15 min at RT. After washing with PBS, the cells were lysed with 0.6 μl of the lysis buffer for 10 min at 4 °C and centrifuged at 20,630g for 10 min at 4 °C. Next, 2 μg of anti-mCherry antibody and 35 μl of 50% protein G Sepharose were added to the lysates, and the mixture was rotated at 4 °C for 2 h. The samples were centrifuged at 4490g for 1 min, and the precipitates were washed four times with lysis buffer. Finally, 35 μl of 2×SDS–PAGE sample buffer was added to the precipitates, and the mixture was boiled at 95 °C for 5 min.

MBP-titin CT production and pull-down experiment

The pMBP-Titin CT vector was transfected into Rosetta-gami B(DE3) pLysS Competent Cells (Sigma-Aldrich). Induction of MBP-titin CT fusion proteins was conducted by the addition of IPTG (0.3 mM) at 20 °C. The cells were harvested after 16 h, and the cell pellets were lysed with BugBuster Protein Extraction Reagent (Millipore) following the manufacturer’s instructions. The MBP-titin CT was purified with Amylose Resin (New England Biolabs) and washed with a lysis buffer (20 mM Tris–HCl (pH 7.0), 150 mM KCl, 1 mM EGTA-KOH (pH 8.0), and 1.0% Triton X-100).

HEK293T cells in a 12-well plate were transfected with 1.0 μg of pAAV-EGFP-CAPN3:CS and its derivatives using 3 μl of PEI-MAX 40k. After 36 h, the cells were lysed with 0.5 ml of lysis buffer and centrifuged at 15,000 rpm for 10 min. MBP-titin CT fusion proteins (5 μg) bound to the Amylose Resin were added to the resultant cell lysates, and the mixture was rotated for 3 h at 4 °C. After 3 h, the resin was washed three times with the lysis buffer. Twenty-five microliters of 2×SDS–PAGE sample buffer was added to the precipitates, and the mixtures were boiled at 95 °C for 5 min.

Western blotting

Protein samples were separated using Extra PAGE One Precast Gel 7.5 to 15% (Nacalai Tesque, 13065-54) and 8% SDS–PAGE, then transferred to a polyvinylidene difluoride membrane. The membrane was blocked with PBS containing 0.05% Tween 20 (PBST) and 5% skim milk for 1 h, followed by probing with the appropriate primary antibodies (1.0 μg/ml) for 1 h at RT. After washing with PBST thrice (5 min each), the membranes were probed with the appropriate horseradish peroxidase–conjugated secondary antibodies (Agilent, P0399, P0447) for 1 h. Following another three washes with PBST (5 min each), the protein bands were detected using Immobilon Western Chemiluminescent Horseradish peroxidase Substrate (Millipore) and visualized with LAS3000 (Fujifilm). Band intensities were analyzed using Fiji open-source software (https://imagej.net/software/fiji/) (48).

Subcellular fractionation

The transfected cells were homogenized in 0.7 ml of homogenization buffer containing 0.32 M sucrose, 10 mM Tris–HCl (pH 7.5), and 5 mM EDTA-KOH (pH 8.0), and then centrifuged at 600g for 10 min. Next, 40 μl of the resulting supernatants were mixed with 15 μl of 4×SDS–PAGE buffer and used as the total fraction. The remaining supernatants were further centrifuged at 20,630g for 10 min to obtain cytosol and pellet (membrane) fractions. Next, 40 μl of the supernatants were mixed with 15 μl of 4×SDS–PAGE buffer and used as the cytosol fraction. The pellets were dissolved in 30 μl of 2×SDS–PAGE buffer and boiled at 95 °C for 5 min.

Immunocytochemistry

For immunostaining, cells were cultured on coverslips placed at the bottom of 6 cm dishes, transfected with appropriate plasmids [1.0 μg pAAV-CMV-EGFP-CAPN3:CS and its derivatives and 1.0 μg pmCherry-titin-kRas (Fig. 4); 1.0 μg EGFP-CAPN3:CS:A702V or D705H, 2.0 μg pSRD-CAPN3:CS, and 1.0 μg mCherry-titin-kRas (Fig. 7)], and fixed with 4.0% paraformaldehyde in PBS for 10 min at RT after 36 h posttransfection. After washing with PBS, the cells were permeabilized with 0.2% Triton in PBS for 5 min at RT and blocked with 5% goat serum in PBS for 30 min. The cells were probed with anti-red fluorescent protein mAb Cocktail [1 μg/ml, Medical & Biological Laboratories Co., Ltd. (MBL), Code No. M208-3] and anti-GFP polyclonal antibody (1 μg/ml, MBL, Code No. 598) for 1 h at RT. After washing with PBS for 15 min in total, the cells were probed with goat anti-rabbit IgG (H+L) cross-adsorbed secondary antibody Alexa Fluor 488 (Thermo Fisher, A-11008, 1:1000) and goat anti-mouse IgG (H+L) highly cross-adsorbed secondary antibody Alexa Fluor 594 (Thermo Fisher Scientific, A-11032, 1:1000) for 30 min at RT. After three additional PBS washes (15 min in total), the cells were mounted with Vibrance Antifade Mounting Medium containing 4′,6-diamidino-2-phenylindole (Vector Laboratories, Inc, H-1800) and imaged using a confocal fluorescence microscope (FV3000; Evident Corporation). Images were analyzed using the open-source Fiji software (48). Colocalization analysis was performed with Fiji’s Coloc2 plugin, using individual transfected cells as regions of interest.

Production of AAV

We mixed 40 μg of three vectors—pHelper vector (AAVpro Helper Free System, Takara), pAAV2/9n (Addgene, #112865), and pAAV-CMV-EGFP-CAPN3 or its derivatives—with 480 μl of PEI-MAX 40k in 7 ml of OPTI-MEM in a 50 ml tube. After vortexing, the mixture was incubated for 15 min at RT. The mixture was then added to the culture medium of 80% confluent HEK293T cells seeded in a Collagen I–coated 225 cm^2^ flask (IWAKI 4143-010, AGC Techno Glass Co Ltd). After 4 h, the media was replaced with 35 ml of Dulbecco’s modified Eagle medium without fetal bovine serum, and the cells were cultured for an additional 5 days in a CO_2_ incubator at 37 °C, without changing the medium.

AAV was purified from the culture medium as previously described (49). Briefly, the culture medium containing AAV was centrifuged at 830g for 15 min at RT. The resulting supernatant was treated with 8.5 ml of AAVanced Concentration Reagent (AAV100A-1, System Biosciences) or 40% polyethylene glycol in 2.5 M NaCl. After 3 days of incubation at 4 °C, the solution was centrifuged at 830g for 30 min. The pellet was washed with 1 ml of cold PBS, centrifuged again at 830g for 1 min, and resuspended in 350 μl of cold PBS. The AAV solution was divided into aliquots and stored at −80 °C until use.

For in vivo delivery, AAV (5 × 10^10^ genome copies per muscle) was injected into the TA muscles of anesthetized mice (2–5 months old) with isoflurane using 27G insulin syringes (Terumo). After 4 weeks, the injected muscles were harvested, dissected, and processed for immunohistochemistry.

Immunohistochemistry

The experimental animals were handled according to the guidelines of the Experimental Animal Care and Use Committee of the Tokyo Metropolitan Institute of Medical Science (approval number: 24034). The mice were sacrificed by cervical dislocation, and the TA skeletal muscles were carefully dissected using forceps. The expression of GFP fusion proteins was confirmed using a fluorescent stereomicroscope (Leica Microsystems, M165 FC). The tissues were immersed in 4.0% paraformaldehyde in PBS for 3 h at 4 °C. The tissues were immersed overnight in PBS containing 30% sucrose at 4 °C, embedded in Optimal Cutting Temperature compound (Sakura Finetek Japan Co., Ltd), and frozen using isopentane prechilled in liquid nitrogen. Cryosections (10 μm thick) were prepared using a CM1950 cryostat (Leica, Biosystems) and mounted on MAS-coated glass slides (MAS-01, Matsunami Glass IND., Ltd).

After washing thrice with PBS (10 min each), the sections were permeabilized with 0.2% Triton in PBS for 5 min and subjected to antigen retrieval in boiled 10 mM Tris–HCl (pH 9.0) for 10 min. After an additional PBS washing step, the sections were blocked for 1 h with PBS containing 1.0% goat serum and 1.0% skim milk or 5% bovine serum albumin.

Next, the sections were probed overnight at 4 °C with appropriate primary antibodies (1.0 μg/ml) in blocking buffer. After washing thrice with PBS (10 min each), the sections were probed for 30 min at RT with goat Alexa Fluor 488–conjugated anti-rabbit IgG or goat Alexa Fluor 594–conjugated anti-mouse IgG antibodies (Thermo Fisher Scientific K.K., A-11034 and A11032, 1:1000). Finally, after a 30-min wash with PBS, the sections were mounted with Vibrance Antifade Mounting Medium containing 4′,6-diamidino-2-phenylindole and visualized using the FV3000 confocal fluorescence microscope.

Structural analysis using MOE software

The dimer structure of the PEF domain was obtained from the Protein Data Bank (PDB ID: 4OKH). Analysis of the 3D structure was performed using the MOE software (MOE, Ver. 2024.0601, Chemical Computing Group Inc., Montreal, Quebec, and Molsis Inc). After protonation and energy minimization of the structure, the changes in binding energy of the LGMDR1-associated mutant dimers compared to the WT dimer were calculated using the MOE program’s “residue scan” function. The larger the dAffinity and dStability, the greater the impairment in the affinity of the two domains and the stability of the dimer, respectively, caused by the mutations.

Statistical analysis

Statistical analyses were performed using EZR software (https://www.jichi.ac.jp/usr/hema/EZR/statmedEN.html) (50). All data are expressed as mean ± SD, and individual data are indicated as scatter plots in the Figures.

Data availability

All data are contained within the article.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.