Myopathy associated LDB3 mutation causes Z-disc disassembly and protein aggregation through PKCα and TSC2-mTOR downregulation

Mechanical stress induced by contractions constantly threatens the integrity of muscle Z-disc, a crucial force-bearing structure in striated muscle. The PDZ-LIM proteins have been proposed to function as adaptors in transducing mechanical signals to preserve the Z-disc structure, however the underlying mechanisms remain poorly understood. Here, we show that LDB3, a well-characterized striated muscle PDZ-LIM protein, modulates mechanical stress signaling through interactions with the mechanosensing domain in filamin C, its chaperone HSPA8, and PKCα in the Z-disc of skeletal muscle. Studies of Ldb3Ala165Val/+ mice indicate that the myopathy-associated LDB3 p.Ala165Val mutation triggers early aggregation of filamin C and its chaperones at muscle Z-disc before aggregation of the mutant protein. The mutation causes protein aggregation and eventually Z-disc myofibrillar disruption by impairing PKCα and TSC2-mTOR, two important signaling pathways regulating protein stability and disposal of damaged cytoskeletal components at a major mechanosensor hub in the Z-disc of skeletal muscle.

H ighly ordered arrangement of actin and myosin filaments in sarcomeres produces physical force in striated muscle. To this end, the Z-disc is of prime importance, anchoring actin filaments from adjoining sarcomeres. Striated muscle fibers are constantly exposed to strong mechanical stress. The Z-disc protein assemblies together with the dystrophin-associated glycoprotein complex and the integrin complex in the sarcolemma play essential roles in withstanding the extreme mechanical force generated during muscle contraction 1 . Genetic defects in the components of each of these large protein complexes lead to degenerative muscle diseases, indicating the importance of these interconnecting systems for the muscle integrity and function 2,3 . Mutations in genes encoding many of the Z-disc proteins have been found to cause myofibrillar myopathies (MFMs), which are characterized by primary dissolution of myofibrils near the Z-disc and accumulation of degraded proteins in the sarcoplasm of muscle fibers 4,5 . The dominant p.Ala165Val mutation in exon 6 of LIM domain-binding 3 gene (LDB3; HGNC 15710; rs121908334; NM_001080114.2:c.494 C > T, NP_001073583.1:(p. Ala165Val)) has been reported in several unrelated families of European ancestry 6,7 . Whereas this mutation is interpreted as likely benign (https://varsome.com/variant/hg38/rs121908334), its penetrance in the long-studied Markesbery-Griggs pedigree is 100% by age 60 years and molecular studies of six unrelated families indicated a founder mutation with a common ancient ancestry 6 . Affected individuals present with adult-onset muscle weakness prominently affecting the calf muscles [6][7][8] . Cardiomyopathy is usually very late and only seen in a minority of patients, which may be due to extremely low expression of LDB3 isoforms containing the exon 6 in heart 9,10 .
LDB3 (ZASP; Cypher) is a highly conserved PDZ-LIM protein that plays an essential but as yet undefined role in maintaining the Z-disc integrity in contracting muscle fibers in flies, zebrafish, and mice [11][12][13] . The PDZ domain interacts with structural Z-disc proteins and the LIM domains bind to and are phosphorylated by protein kinase C (PKC) isoforms 14,15 . Alternative splicing of LDB3 is known to generate three major isoforms in skeletal muscle 9,10 . The shorter isoform lacks the C-terminal LIM domains. The longer isoforms with LIM domains either contain exon 10 or exclude it (LDB3-L and LDB3-LΔex10, respectively). LDB3-L is replaced by LDB3-LΔex10 in skeletal muscle during postnatal development 9 . Gene deletion studies in mice showed that the longer isoforms, but not LDB3-S, are important for Zdisc integrity in striated muscle 16 . Moreover, the p.Ala165Val mutation in LDB3-LΔex10, but not other isoforms, causes F-actin disruption in transfected muscle cells 17 . The actin-binding domain of LDB3 is mutated in MFM, but recombinant mutant proteins are correctly folded and show unaffected actin-binding affinity and kinetics 17,18 . Over-expression of the LDB3-LΔex10-p. Ala165Val via intramuscular injection leads to MFM-like pathology in mouse tibialis anterior muscle fibers 17 . However, short-term, variable, and heterogeneous expression of mutant protein in electroporated muscle fibers limited optimal investigation of disease mechanisms. Whereas knockout and cardiacspecific models have helped to characterize some LDB3 functions 11,16,19,20 , the murine models of LDB3- MFM have not yet been reported and the LDB3 interactions relevant to the MFM phenotype remain unknown.
In this study, a heterozygous knock-in of the p.Ala165Val mutation in mouse Ldb3 gene closely recapitulated the genetic mutation in patients and allowed physiological levels of LDB3 isoforms in mouse tissues. The Ldb3 Ala165Val/+ mice developed muscle weakness and classic MFM pathology. Our results indicate that LDB3 acts as a signaling adapter in a major mechanosensor assembly through interactions with filamin C, its chaperone HSPA8, and PKCα at skeletal muscle Z-disc. The LDB3 p.
Ala165Val mutation impairs PKCα and TSC2-mTOR mediated homeostasis in this large protein assembly leading to protein aggregation myopathy.

Results
Generation of Ldb3 Ala165Val/+ knock-in mice. We introduced the p.Ala165Val point mutation (chr14:34571772 C > T; GRCm38/ mm10; C57BL/6 N), responsible for MFM 6,7 , into exon 6 of the endogenous mouse Ldb3 gene to generate Ldb3 Ala165Val/+ mice by homologous recombination (Fig. 1a; Supplementary Fig. 1a). The mutated residue is conserved, and overall amino acid identity is >92% for LDB3 isoforms between human and mouse. Targeted gene sequencing and Southern blot analysis confirmed the accuracy of gene editing and the absence of other mutations in the recombined region ( Supplementary Fig. 1b). The presence of the NP_001034164.1:(p.Ala165Val) mutation was further validated by Sanger sequencing (Fig. 1b). The levels of Ldb3 mRNA and that of the major LDB3 protein isoforms in the vastus muscle of Ldb3 Ala165Val/+ mice were similar to Ldb3 +/+ littermates ( Fig. 1c; Supplementary Fig. 1c; Supplementary Data 1), suggesting that the point mutation is not affecting transcript or protein stability in muscle tissue. Mating of Ldb3 Ala165Val/+ mice resulted in 26% Ldb3 +/+ , 54% Ldb3 Ala165Val/+ , and 20% Ldb3 Ala165Val/Ala165Val mice (n = 84 mice; 10 litters), indicating that the mutation did not affect mouse viability. Ldb3 Ala165Val/+ mice had a normal lifespan and weight. Their cage behavior, feeding, and grooming activities were comparable to Ldb3 +/+ littermates.
LDB3 p.Ala165Val mutation causes a classic pathological MFM phenotype in mouse skeletal muscle. Skeletal muscle histology was assessed at 4, 6, and 8 months of age in the hindlimb muscles of Ldb3 Ala165Val/+ mice and their Ldb3 +/+ littermates. Muscle histology in mutant mice at 4 months of age was similar to that in wildtype mice (Fig. 2a). In contrast, Gomori trichrome staining of transverse sections of the soleus muscle of 8-month-old Ldb3 Ala165Val/+ mice showed dark blue to blue-red hyaline granular deposits and rimmed vacuoles in the sarcoplasm of muscle fibers (Fig. 2a). These abnormal fibers showed decrease in oxidative enzyme activity in the region of protein deposits and increased enzyme activity mostly in the periphery, suggesting abnormal mitochondrial function. In addition, non-specific myopathic changes such as increases in internal nuclei, hypertrophic fibers, and muscle fibers with rounded contour were observed in the vastus and tibialis anterior muscles of mutant mice (Fig. 2b). To quantitate the changes in myonuclear location, we performed morphometry with wheat germ agglutinin (WGA) to outline muscle membrane and distinguish muscle nuclei. Relative to wildtype littermates, Ldb3 Ala165Val/+ mice had a significantly higher percentage of fibers with internal nuclei in the tibialis anterior muscle (p < 0.01; Supplementary Fig. 2a). Muscle fiber type distribution was unchanged in the soleus and vastus muscles of Ldb3 Ala165Val/+ mice (Supplementary Fig. 2b-d). Muscle fiber size distribution for types I and IIA in Ldb3 Ala165Val/+ mice was shifted towards right compared to those from Ldb3 +/+ mice, whereas distribution of type IIB fiber size was unchanged ( Supplementary Fig. 2e). Mean minimal Feret's diameter for fiber types I and IIA was higher in Ldb3 Ala165Val/+ mice relative to wildtype mice, but the change was not statistically significant in the Bonferroni's multiple comparison test ( Supplementary  Fig. 2f). Accordingly, variance coefficient of the minimal muscle fiber diameter was unchanged for all fiber types in Ldb3 Ala165Val/+ mice ( Supplementary Fig. 2g).
Immunostaining studies showed sarcoplasmic accumulations of the MFM-associated Z-disc proteins including LDB3, myotilin, desmin, filamin C, and αB-crystallin, as well as ubiquitin in same fibers of 8-month-old Ldb3 Ala165Val/+ mice (Fig. 2c). Counts of a mean of 498 fibers in transverse soleus muscle section of five mutant mice each showed a mean of 64% and 44% muscle fibers contained filamin C and ubiquitin accumulations, respectively ( Supplementary Fig. 3a). These protein accumulations occurred in multiple or diffuse form in muscle sarcoplasm. The abnormal fibers were distributed focally surrounded by muscle fibers without protein aggregation. Such protein aggregates were not observed in Ldb3 +/+ mice. Electron microscopy studies showed the Z-disc myofibrillar disruption, dislocated enlarged mitochondria, and autophagic vacuoles in the soleus muscle fibers of 6-month-old Ldb3 Ala165Val/+ mice and normal Z-disc and sarcomere architecture in their Ldb3 +/+ littermates ( Fig. 2d-f). An in vivo "autophagy flux" assay 21 detected an increase in autophagy markers LC3-II and sequestosome 1 (p62/SQSTM1) following 3 days of treatment with colchicine in skeletal muscle of 4-month-old Ldb3 Ala165Val/+ mice compared to Ldb3 +/+ littermates ( Fig. 2g-i; Supplementary Data 5). These findings demonstrate that Ldb3 Ala165Val/+ mice develop typical MFM pathology characterized by the Z-disc streaming, sarcoplasmic protein aggregates, rimmed vacuoles, and increased lysosomal autophagy in skeletal muscle.
Cardiac ventricular muscle fibers of Ldb3 Ala165Val/+ mice displayed normal histology at 9 months of age ( Supplementary  Fig. 4a, b), which correlates with clinical observations of cardiomyopathy not being a regular feature in patients known to have the same mutation 22 . A knockout-validated custom-made rabbit polyclonal LDB3ex6ab antibody detected LDB3 in the skeletal muscle of mice but not in cardiac muscle ( Supplementary  Fig. 5a, b), supporting previous observations that the LDB3 isoforms containing exon 6 are predominantly expressed in skeletal muscle in mice 9 .
LDB3 Ala165Val protein triggers early aggregation of filamin C and its chaperones, eventually leading to aggregation of the mutant protein and Z-disc myofibrillar disruption. Patients usually present with advanced stages of myopathic disease at the time of biopsy, thereby limiting studies of early pathological features in the MFM. We examined the early cellular and molecular events that lead to the MFM pathology in Ldb3 Ala165Val/+ mice. We found filamin C aggregates in muscle fibers that had normal b Sanger sequence shows the heterozygous C > T mutation changing the codon GCT (Ala) to GTT (Val). c Immunoblotting analysis (blot and dot plot) of the LDB3 isoforms expression relative to β−actin in the vastus muscle of 8-month-old Ldb3 Ala165Val/+ mice (blue) and Ldb3 +/+ littermates (red). n = 4 each, represents triplicate assay. d Dot plot comparing maximum all paws grip force of five pulls per mouse normalized to body weight (g/g) between Ldb3 Ala165Val/+ mice (blue) and Ldb3 +/+ littermates (red) at 3, 6, and 9 months of age. Mean ± SEM: 9.2 ± 0.3 g/g versus 10.7 ± 0.3 g/g at 3 months, 6.9 ± 0.3 g/g versus 10.4 ± 0.3 g/g at 6 months, and 6.5 ± 0.2 g/g versus 7.9 ± 0.3 g/g at 9 months; n = 10-13 Ldb3 Ala165Val/+ mice and n = 8-10 Ldb3 +/+ mice per age group. The two-way ANOVA Bonferroni's multiple comparisons test significant p values between groups are shown. e Dot plot shows maximal isometric force (mN/mm 2 ) generated by the extensor digitorum longus muscle of 6-month-old male Ldb3 Ala165Val/+ mice (blue) versus Ldb3 +/+ littermates (red) against stimulation frequencies (Hz). Mean ± SEM: 148 ± 5 mN/mm 2 versus 187 ± 3 mN/mm 2 ; 160 Hz; n = 4 Ldb3 Ala165Val/+ mice and n = 3 Ldb3 +/+ mice. The two-way ANOVA Bonferroni's multiple comparisons test significant p values between groups are shown. f Dot plot shows the maximum holding impulse (N*s) in the four-limb wire grid holding test in 9-month-old male Ldb3 Ala165Val/+ mice and Ldb3 +/+ littermates. Mean ± SEM: 32 ± 4 N*s versus 81 ± 8 N*s; n = 9 Ldb3 Ala165Val/+ mice and n = 7 Ldb3 +/+ mice. Male mice were used in these tests as female mice showed intra-and inter-group variability. The two-tailed unpaired t-test comparison significance between groups is shown. *p < 0.05, **p < 0.01, ****p < 0.0001. The error bars in dot plots represent Mean (SEM).
LDB3 immunostaining in transverse vastus and soleus muscle sections of 6-month-old Ldb3 Ala165Val/+ mice (Fig. 3a). Counts of a mean of 623 fibers from vastus muscle transverse sections of five mutant mice each showed a mean of 15% fibers contained filamin C accumulations ( Supplementary Fig. 3b). Such protein aggregates were not seen in wildtype littermates. The filamin C accumulations were more prominent compared to LDB3 in the vastus lateralis muscle fibers of patients sharing the same mutation (Fig. 3a). Filamin C, an MFM gene product 23 , acts as a primary mechanosensitive crosslink for actin filaments at the Z-discs 24 . Mechanical strain on the actin network results in unfolding of the filamin C crosslinks, which are degraded through the conserved tension-induced chaperone-assisted selective autophagy (CASA) pathway 25 . Immunostaining of serial muscle transverse sections showed that filamin C aggregates were accompanied with the CASA chaperones BAG3, HSPA8, HSPB8, and ubiquitin in muscle fibers of Ldb3 Ala165Val/+ mice but not in their Ldb3 +/+ littermates (Fig. 3b, c). These sarcoplasmic protein aggregations were also found in the vastus muscle fibers of MFM patients with the LDB3 p.Ala165Val mutation (Fig. 3d).
Serial longitudinal tibialis anterior muscle sections of 4-and 8month-old Ldb3 +/+ mice showed normal Z-disc distribution of LDB3, filamin C, myotilin, BAG3, and HSPA8 proteins ( Fig. 4a and f). Immunostaining of the tibialis anterior muscle sections in 4-month-old Ldb3 Ala165Val/+ mice showed aggregates of filamin C, BAG3, and HSPA8 co-localizing at the Z-discs of same muscle fibers, whereas staining for LDB3 and myotilin was normal ( Fig. 4b-e). Co-aggregation of LDB3 with filamin C, myotilin, BAG3, and HSPA8 with or without Z-disc myofibrillar disruption occurred at a later disease stage in 8-month-old Ldb3 Ala165Val/+ mice ( Fig. 4g-j). Prominent accumulations of the CASA chaperones were found in muscle fibers harboring filamin C and other myofibrillar protein aggregates but not in adjoining fibers without protein aggregation (Figs. 3c, d and 4d, e, i, j). These results indicate that the LDB3 p.Ala165Val mutation causes aggregation of damaged filamin C before LDB3 and other MFM proteins, eventually leading to Z-disc myofibrillar disruption in skeletal muscle. Muscle fibers react to ubiquitinated protein aggregation with accumulations of the CASA chaperones, but the protein degradation systems are not efficient in clearing the damaged protein aggregates.
LDB3 WT and LDB3 Ala165Val proteins interact with mechanosensing domain in filamin C and HSPA8, a chaperone in tension-induced autophagy pathway. We found early and prominent filamin C aggregates in skeletal muscle fibers of Ldb3 Ala165Val/+ mice, indicating that the LDB3 p.Ala165Val mutation likely affects the stability or degradation of filamin C. Immunoblotting studies showed that whole muscle levels of filamin C and the CASA chaperones were not altered in skeletal muscle of Ldb3 Ala165Val/+ mice (Fig. 5a). The association of modular protein-interacting domains enables LDB3 to serve as an adapter protein that recruits multiple proteins to a localized site of action in a subcellular domain. To this end, we conducted a yeast two-hybrid (Y2H) screen of human skeletal muscle cDNA library using a human LDB3 peptide encoded by exons 8 and 11 as bait to identify novel interacting proteins. This bait region is exclusively present in the postnatal LDB3-LΔex10 isoform. The region is known to interact with skeletal actin, and its presence is crucial for the deleterious effects of the p. Aal165Val mutation on actin cytoskeleton in transfected muscle cells 17 . The Y2H screen identified 46 prey clones encoding six potential LDB3 interactors including two Z-disc associated proteins filamin C and HSPA8, with strong confidence in the interaction ( Fig. 5b; Supplementary Table 1).
Interestingly, the five filamin C clones encoded the rod domain 2 immunoglobulin-like (Ig) 17-21 repeats, a precisely tuned mechanosensor that detects and responds to the small cytoskeletal forces during muscle contraction 26,27 . In addition, three prey clones encoded an internal region of HSPA8 (aa 237-468; NM_006597.5) that is recruited by BAG3 to the Ig19-21 repeats of unfolded filamin C as a part of the CASA complex forming the molecular basis for sensing force-induced unfolding 25 . Pairwise Y2H assays showed that inclusion of the exon 10-encoded region in the bait corresponding to LDB3-L isoform abolished LDB3 binding to HSPA8 (Fig. 5c). These findings support our previous observations of isoform-specific LDB3 interactions with skeletal actin 17,18 . GST pulldown assays using tagged wildtype (WT) and mutant LDB3-LΔex10 proteins as bait showed that the LDB3 p. Ala165Val mutation did not impair filamin C and BAG3 interactions (Fig. 5d). Binding of WT and mutant LDB3-LΔex10 with filamin C Ig 17-21 domain and HSPA8 was validated by independent co-immunoprecipitation (IP) of tagged proteins from transfected COS-7 cell lysates (Fig. 5e). Endogenous interactions of LDB3 with filamin C and HSPA8 in skeletal muscle were confirmed by co-IP of these proteins from the tibialis anterior muscle myofibrillar fraction of 4-month-old Ldb3 Ala165Val/+ mice and Ldb3 +/+ littermates (Fig. 5f). These findings point to mechanosensing roles of LDB3 through interactions with filamin C and HSPA8 in the Z-disc. The p.Ala165Val mutation does not appear to affect these LDB3 interactions in skeletal muscle.
Ldb3 +/+ littermates. Whole muscle levels of PKCα in 4-and 8month-old Ldb3 Ala165Val/+ mice were decreased to about 50% of those in Ldb3 +/+ littermates, and levels of TSC2 in the mutant mice were reduced to almost two-thirds of those in Ldb3 +/+ mice ( Fig. 6c-f; Supplementary Data 8,9). Downregulation of PKCα and TSC2 indicates a likely mechanism for the observed aggregation of damaged filamin C in muscle fibers of Ldb3 Ala165Val/+ mice as PKCα stabilizes filamin C in muscle Z-disc through phosphorylation 34 , and TSC2 initiates CASA-mediated degradation of damaged filamin through local mTORC1 inhibition 35 .

Discussion
Here, we unveil a previously undescribed mechanism by which a disease-associated mutation in LDB3, a PDZ-LIM protein, causes protein aggregation and myofibrillar Z-disc disruption in skeletal muscle. Our data demonstrate that LDB3 directly binds to the mechanosensing domain of filamin C, and its chaperone HSPA8, a central player in protein folding and proteastasis control. Studies in knock-in mice show that the MFM-associated LDB3 p.Ala165Val mutation causes early aggregation of filamin C and its chaperones in the tension-induced CASA pathway before aggregation of the mutant LDB3 protein in muscle fibers. The LDB3 p.Ala165Val mutation destabilizes filamin C and stalls the removal of damaged filamin C through impaired PKCα and TSC2-mTOR signaling at arguably the most important mechanosensor hub in the Z-disc of skeletal muscle (summarized in Fig. 7). The presence of chaperones and ubiquitin in these protein aggregates likely represents impairment of protein degradation pathways. Further, pathology studies show that muscle fibers with protein aggregates develop the Z-disc myofibrillar disruption in skeletal muscle of Ldb3 Ala165Val/+ mice.
LDB3 is well-poised to play a role in mechanosensor function in striated muscle. Homozygous LDB3-null mice display fragmented Z-discs in the diaphragm muscle after birth, but not during embryonic development 11 . Since the diaphragm does not Fig. 4 Progressive protein aggregation and Z-disc myofibrillar disruption in skeletal muscle of Ldb3 Ala165Val/+ mice. Representative immunofluorescence staining on perfused tibialis anterior muscle consecutive longitudinal sections of 4-month-old (a-e) and 8-month-old (f-j) Ldb3 Ala165Val/+ mice and their Ldb3 +/+ littermates. a Muscle sections of 4-month-old Ldb3 +/+ mice (n = 4) show normal Z-disc staining for LDB3, filamin C, myotilin, BAG3, and HSPA8 proteins. b, c Muscle sections of 4-month-old Ldb3 Ala165Val/+ mice (n = 6) show same muscle fiber co-stained with LDB3 and filamin C antibodies (b), and LDB3 and myotilin antibodies (c). d, e Muscle sections of 4-month-old Ldb3 Ala165Val/+ mice (n = 6) show same muscle fiber co-stained with filamin C and HSPA8 antibodies (d), and BAG3 and HSPA8 antibodies (e). Filamin C, HSPA8, and BAG3 aggregates are seen at the Zdisc spanning multiple sarcomeres in same fiber. In contrast, the LDB3 and myotilin antibodies show normal Z-disc staining in the fiber with filamin C aggregates. f Muscle sections of 8-month-old Ldb3 +/+ mice (n = 5) show normal Z-disc staining for LDB3, filamin C, myotilin, BAG3, and HSPA8 proteins.
Top and bottom panel each shows same muscle fiber co-stained with LDB3 and filamin C antibodies (g), and LDB3 and myotilin antibodies (h) in muscle sections of 8-month-old Ldb3 Ala165Val/+ mice (n = 7). Muscle sections of 8-month-old Ldb3 Ala165Val/+ mice (n = 7) show same muscle fiber co-stained with LDB3 and BAG3 antibodies (i), and LDB3 and HSPA8 antibodies (j). LDB3 aggregates colocalize with filamin C, myotilin, BAG3, and HSPA8 at the Zdiscs in muscle fibers. The protein aggregates are seen in muscle fibers with normal periodicity of the Z-discs and with disorganized Z-disc periodicity indicating Z-disc myofibrillar disruption. A part of adjacent muscle fiber with normal staining pattern is shown at bottom in each image for comparison. Scale bar = 10 μm and applies to all images.  7,17 . Domain composition of the prey clones are shown. c Pairwise Y2H assays demonstrating interaction between LDB3 peptides and HSPA8. Positive interactions show yeast growth on the media deficient in HIS3 and ADE2. Yeast cells co-transformed with empty bait vector and the HSPA8 prey show no growth (labeled -). Transformation efficiency was uniform for all constructs. Sequential tenfold yeast dilutions are shown. d GST pulldown assay shows that GSTtagged wildtype (WT) and mutant LDB3-LΔEX10 (Ala165Val) but not GST alone pulled down filamin C and its interactor the CASA cochaperone BAG3 from the vastus muscle lysates of wildtype mice. Data represent n = 3 mice and triplicate assays. e Co-immunoprecipitation (co-IP) assays show that a FLAG antibody pulled down the FLAG-tagged WT and mutant (Ala165Val) LDB3-LΔex10 together with HA-tagged filamin C rod domain Ig17-21 and HSPA8 in Cos7 cells. Data represent triplicate assays. The proteins were detected with anti-FLAG and anti-HA antibodies. f Co-IP assays show that an LDB3 antibody pulled down LDB3 isoforms together with filamin C and HSPA8 from the tibialis anterior muscle lysates of Ldb3 Ala165Val/+ and Ldb3 +/+ mice. Data represent n = 3 mice per group and triplicate assays.
contract until after birth, these observations suggest that LDB3 is required for maintaining the sarcomere structure during muscle contraction. LDB3 interacts with skeletal actin filaments, the actin crosslinkers α-actinin and myotilin, as well as integrins that play a role in sensing external tension in striated muscle [13][14][15]17,18 . In this study, we identified an LDB3-binding site in filamin C rod domain 2 that is involved in multiple protein interactions including BAG3, myotilin, dystroglycans, and integrins ( Fig. 7a) 25,36,37 . Whereas the mechanical properties of filamin C remain unknown, previous studies of filamin A showed that the rod 2 domain undergoes unfolding upon mechanical strain that changes affinity of the binding proteins, thus providing a molecular basis for mechanosensing 24,26,27 . LDB3 also binds to HSPA8, which together with BAG3 degrades unfolded filamin C through the CASA pathway upon mechanical strain (Fig. 7b) 25 . These interactions were unaffected by the MFM-associated LDB3 p.Ala165Val mutation, supporting that toxicity is not through altered ligand binding properties of mutant protein. The filamin C-and HSPA8-binding domain in LDB3 also interacts with skeletal actin and is uniquely present in the postnatal longer LDB3 isoform 17 . Previously we have shown that the LDB3 p. Ala165Val mutation in the postnatal longer LDB3 isoform, but not other isoforms, leads to myofibrillar disruption 17 , suggesting that the LDB3 interactions through this domain likely play essential roles in the pathogenesis of MFM. This is supported by early and prominent aggregation of filamin C that seeds aggregation of other proteins in muscle fibers of Ldb3 Ala165Val/+ mice. Mutations in many proteins in this interactome, such as LDB3, filamin C, myotilin, BAG3, DNAJB6, and HSPB8 lead to MFM phenotype 5 , indicating that most of the disease pathways converge onto a final pathway centered on the large LDB3-filamin C-CASA chaperone scaffold for this diverse group of myopathies. These findings underscore critical roles for LDB3 in recruiting key proteins to strategic sites and facilitating mechanosensing in striated muscle.
LDB3 is known to anchor PKC isozymes at the Z-disc of striated muscle 14 . PKCα accounts for 97% of the conventional PKC function in skeletal muscle 38 . We found a significant decrease (~50%) in PKCα expression in skeletal muscle of Ldb3 Ala165Val/+ mice, before the MFM phenotype. Mechanisms by which the MFM-associated LDB3 p.Ala165Val mutation downregulates PKCα in skeletal muscle are yet unknown, but presumably the mutation affects ligand binding, phosphorylation, or autoinhibition of the kinase that effectively leads to its degradation 39,40 . The longer LDB3 isoforms have been shown to negatively regulate the PKCα expression in heart 16 . To this end, the p.Ala165Val mutation likely acts as a gain-of-function mutation leading to the MFM phenotype. Interestingly, PKCα has been viewed as an attractive therapeutic target in many diverse cancers and degenerative diseases of heart and brain [39][40][41] . Loss-of-function mutations in PKCα are prevalent in cancers, and studies in cancer cell lines showed that small changes in the PKC expression have large effects on cellular functions 40 . It is known that PKCα phosphorylates filamin C and prevents its calpainmediated proteolysis, thus preserving its dimerization function 34 . An MFM-associated nonsense mutation in filamin C, which disrupts its dimerization domain, causes self-aggregation of mutant protein and leads to aggregation of interacting  Table 2. Only PKCα and TSC2 are known to localize at skeletal muscle Z-disc. c, d Immunoblotting analysis (blot and dot plot) of PKCα and TSC2 protein levels relative to vinculin in the vastus muscle of 4-month-old Ldb3 Ala165Val/+ mice (n = 5 and 4, respectively) and Ldb3 +/+ littermates (n = 4 and 3, respectively). e, f Immunoblotting analysis (blot and dot plot) for protein levels of PKCα and TSC2 relative to vinculin in the vastus muscle of 8-month-old Ldb3 Ala165Val/+ mice (n = 4 and 5, respectively) and Ldb3 +/+ littermates (n = 3 and 4, respectively). The error bars in dot plots are Mean (SEM) and represent triplicate assays. The two-tailed unpaired t-test comparison significant p values are shown. **p < 0.01; ***p < 0.001; ****p < 0.0001. proteins 42,43 . Taken together, reduction in PKCα likely predisposes filamin C to calpain-mediated proteolysis leading to protein aggregation in skeletal muscle of Ldb3 Ala165Val/+ mice (Fig. 7c).
The aggregation of damaged filamin C aggravates the autophagy pathway in muscle fibers of Ldb3 Ala165Val/+ mice. This is evidenced by marked accumulations of the CASA chaperones in the muscle fibers containing protein aggregates and increased levels of the autophagy markers after colchicine blockage in skeletal muscle of 4-month-old mice. However, it appears that the strain on the cellular degradation system ultimately limits the capacity of the CASA chaperones to remove the damaged misfolded proteins and causes myopathy. This connection is supported by the hallmark MFM pathology changes in the Ldb3 Ala165Val/+ mice that resemble muscle pathology caused by mutations in filamin C and CASA components BAG3, DNAJB6, and HSPB8, which have been proposed to negatively affect CASA function 23,[44][45][46][47][48] . The CASA pathway constantly operates at the Zdisc and removes mechanically damaged proteins such as filamin C, thus it differs from the atrophy-driven autophagy pathways 49 . Recruitment of TSC1:TSC2 -mTORC1 assemblies by BAG3 to skeletal muscle Z-disc is required for the disposal of damaged filamin C through the CASA pathway 35 . Within the complex, TSC1 stabilizes TSC2 50,51 , whereas TSC2 acts as a GTPase activating protein and integrates signals from various kinases leading to mTORC1 regulation in the cell 33 . It is possible that the reduction of TSC2 in skeletal muscle of Ldb3 Ala165Val/+ mice perturbs CASA function, leading to filamin C aggregation in muscle fibers (Fig. 7c). In addition, chronic changes in PKCα and TSC2-mTORC1 levels likely affect the balance of lysosomal autophagy in skeletal muscle of Ldb3 Ala165Val/+ mice 52 .
In Ldb3 Ala165Val/+ muscle, the decrease in PKCα levels may be related to reduced TSC2 through mTORC2 inhibition 53,54 . PKCα is one of the best-characterized substrates of TSC2-mTORC2 that potentially regulates subcellular localization of mTORC1 and mTORC2 components, thereby permitting selective activation of specific targets 55,56 . It remains to be determined whether the MFM-associated mutation affects subcellular translocation of mTOR complexes to the Z-disc of skeletal muscle upon mechanical strain. The RPPA showed that the levels of TSC1 and Akt were unchanged in Ldb3 Ala165Val/+ mice, thus suggesting alternative mechanisms for TSC2 destabilization 50,51,57 . Furthermore, expression of total and phosphorylated S6K and 4E-BP1, two key regulators of translation was unaffected 57 , suggesting that the MFM mutation does not affect the anabolic process of protein synthesis in skeletal muscle of Ldb3 Ala165Val/+ mice.
Patients with LDB3 p.Ala165Val mutation typically develop a late-onset myopathy 6,7,22 . Similarly, Ldb3 Ala165Val/+ mice developed normally until 3 months of age, and filamin C aggregation is not visible until 4 months of age. The mutant mice did not develop respiratory failure, cardiomyopathy, or neuropathy, which reflects the clinical disease in patients known to have the same mutation and is consistent with the fact that only negligible expression of exon 6 containing LDB3 isoforms is found in mouse heart and nervous system 9 . In contrast, BAG3 mutations are associated with severe congenital MFM in children, suggesting that BAG3 mutations dramatically impact CASA and other protein degradation mechanisms, leading to a severe widespread disease 46,47 . Selective vulnerability of skeletal muscle to the LDB3 p.Ala165Val mutation may be related to high expression of the mutated actin-binding domain in muscle tissue 17 . Late-onset disease in Ldb3 Ala165Val/+ mice and patients with the same mutation suggests that the CASA failure is limited to select substrates and that alternate protein quality control mechanisms may compensate to some extent until a later age and disease stage 47 . Moreover, age-induced increased dependence on the BAG3-mediated autophagy for disposal of damaged proteins that cannot be degraded by the proteasome may contribute to lateonset MFM 58 .
In conclusion, the LDB3 p.Ala165Val mutation defies the prevailing thinking for pathogenesis of protein aggregation in neurodegenerative disases 59 . The mutant protein does not seed the aggregation process through misfolding or self-aggreation 18 , but triggers aggregation of filamin C and its chaperones at the Zdisc of skeletal muscle. Our findings indicate new roles for LDB3 as a modulator of mechanosensing through interactions with filamin C and its chaperone, as well as PKCα in the Z-disc of skeletal muscle. And, they demonstrate that the LDB3 p.Ala165Val mutation leads to Z-disc disassembly and protein aggregation by affecting the dynamic signaling pathways mediated by PKCα and TSC2-mTOR at a major mechanosensory hub in skeletal muscle. Ultimately, our results support signaling modulation by small molecules or allele silencing as potential therapeutic strategies for this debilitating degenerative disease with as of yet no approved treatment.

Methods
Mice. Wildtype mice (C57BL/6 N) were from Charles River laboratories (Wilmington, MA). The β-actin-cre mice were described previously 60  Antibodies. Primary antibodies are listed in Supplementary Table 3. Antibodies against LDB3 were generated by immunizing rabbits with a peptide corresponding to amino acid residues 116-130 encoded by exon 6 of human LDB3 (NP_001073585; LDB3ex6ab; Supplementary Fig. 5a). We validated the LDB3ex6ab antibody for immunoblotting and immunofluorescence using transfected COS-7 cells and tissues of Ldb3 +/+ and Ldb3 −/− mice ( Supplementary  Fig. 5b-d). Alexa Fluor-, IRDye-, and HRP-conjugated goat anti-mouse or antirabbit secondary antibodies were purchased from Invitrogen, LI-COR biosciences, and Jackson ImmunoResearch Laboratories, respectively. Generation of Ldb3 Ala165Val/+ mice. Strategy for generating the p.Ala165Val knock-in mutation in the mouse Ldb3 gene is shown in Supplementary Fig. 1a. The RP23-244C16 Bacterial Artificial Chromosome (BAC) clone containing the mouse Ldb3 gene (C57BL/6 N) was modified to reproduce the MFM causative p. Ala165Val mutation (chr14:34571772 C > T; GRCm38/mm10; C57BL/6 N), by a galK-based recombineering in SW102 cells, as described previously 62 . The targeting construct homologous to the region flanking the exon 6 mutation was generated by PCR using two 100-mer oligonucleotides with 20 bp overlap at the 3′ end. Next, the Ldb3 genomic region between introns 3 and 8 (chr14:34567235-34578097) containing the mutation was retrieved from the modified BAC into PL253 genetargeting vector by "gap repair", as described previously 63 . And finally, a neomycin selection marker cassette (PGK-neo), flanked by loxP sites, was inserted within intron 6 between chr14:34570741 and 34570742 base pairs in the direction of gene transcription by gap repair using PL452 vector containing~500 bp homology arms Fig. 7 Model of proposed mechanism for protein aggregation and Z-disc disassembly by LDB3 p.Ala165Val mutation. a, b Diagrams show LDB3filamin C -CASA complex interactions at skeletal muscle Z-disc relevant to MFM pathogenesis. LDB3 binds to the mechanosensor Ig 17-21 repeats (orange) in filamin C where the binding sites for the MFM-associated proteins BAG3 and myotilin, as well as a myopathy-associated integrin complex have been identified 25,36,37 . LDB3 binds to myotilin and PKCα, which regulates filamin C stability through phosphorylation (P) [14][15][16]34 . LDB3 interacts with HSPA8, a BAG3 partner that together with the MFM-associated HSPB8 chaperone degrades damaged filamin C through CASA pathway (adapted from 25 ). BAG3 cooperates with the HSPA8-associated ubiquitin ligase STUB1 and its partner UBE2D in the ubiquitination of chaperone-bound filamin C, which is recognized by sequestosome-1 leading to lysosomal disposal. These interactions are unaffected by the MFM-associated LDB3 p.Ala165Val mutation. c Model of LDB3 p.Ala165Val mutation effects on the integrity of LDB3 -filamin C -CASA complex interactome under mechanical strain. The activity of each protein and interaction is presented with color, blue for decrease and red for increase. An arrow indicates an activation, a bar at the end of an edge indicates an inhibitory interaction. A downward arrow within a box indicates decrease in the protein levels. Under normal conditions, LDB3 enables PKCαmediated phosphorylation of filamin C and other interacting proteins at the Z-disc and protects these proteins from proteolysis by calpains. The CASA pathway constantly operates at the Z-disc mediating degradation of large cytoskeleton components including filamin C damaged during mechanical strain 49 . The CASA activity depends on local mTORC1 inhibition through BAG3-recruited TSC2:TSC1 signaling 35 . PKCα is modulated by TSC2-mTORC2 signaling and the kinase may regulate mTOR assemblies including spatial localization (dashed lines) [53][54][55][56] . The LDB3 p.Ala165Val downregulates PKCα and TSC2-mTOR, two signaling proteins monitoring the integrity of the Z-disc assembly. Decreased PKCα promotes proteolysis by calpains leading to aggregation of damaged proteins. Reduced TSC2-mTOR together with increased strain on capacity of degradation pathway leads to impaired CASA function aggravating damaged protein aggregation, eventually leading to the Z-disc disassembly and myofibrillar disruption.
flanking the insertion site, as described previously 63 . The final modified PL253 gene-targeting vector was linearized with Not I and electroporated into embryonic stem (ES) cells generated from B6x129 hybrid mice. Targeted ES cells were identified by Southern blotting and confirmed by Sanger sequencing. Positive ES cell clones were injected into blastocysts to obtain chimeric mice as described previously 64 . The presence of the selection cassette disrupted allelic expression in Ldb3 flox-Ala165Val/+ mice. Functional knock-in lines were activated by crossing Ldb3 floxed-Ala165Val/+ mice with β-actin-cre transgenic mice (C57BL/6 N). Offsprings were bred with wildtype mice to remove Cre-recombinase. The resulting Ldb3 Ala165Val/+ mice were backcrossed for at least 5 generations on C57BL/6 N background.
Targeted gene sequencing of the Ldb3 recombined region. Mouse genomic DNA was PCR amplified using primer sets described in Supplementary Table 4. DNA library was prepared using the Nextera library preparation kit (Illumina). The concentrations of the indexed libraries were analyzed on the Agilent 2200 TapeStation using the D1000 Kit (Agilent Technologies). Equimolar amounts of the indexed libraries were pooled to obtain a 4 nM library mixture. After denaturing and further diluting, the final 12 pM library was loaded into an Illumina cartridge. Sequencing was performed using the Illumina MiSeq Reagent Kit v2 (500 Cycles) on the Illumina MiSeq instrument following the manufacturer's instructions.
Southern blotting. Tail DNA was extracted by digestion in buffer (1 M Tris-HCl pH 8.5, 0.5 M EDTA, 5 M NaCl, 10%SDS, 10 mg/ml proteinase K) overnight at 55°C with continuous shaking in low speed. Lysates were cleared by centrifugation at 14,000 rpm for 5 min, DNA was precipitated in equal volume of isopropanol, pelleted at 14,000 rpm for 5 min, washed three times in 70% ethanol, air dried, and rehydrated in water. Digoxigenin-labeled DNA probes hybridized to the 5′ (nucleotides 34579195-34579596; NC_000080.6) and 3′ (nucleotides 34564772-34565187) flanking sequence of the recombined Ldb3 genomic region were used. Mouse genomic DNA samples (10 μg) were digested with EcoRV (New England Biolabs) overnight at 37°C, separated on a 0.7% agarose gel over 20 h and blotted onto a positively charged nylon membrane (Roche Diagnostic) by capillary transfer for 48 h. Hybridization was done overnight at 45°C with agitation at 10 rpm. Membranes were washed twice with 2X SSC containing 0.1% SDS for 5 min at 65°C and then twice with 0.5X SSC containing 0.1% SDS for 15 min at 65°C. DIGlabeled probes after hybridization to target DNA were detected using DIG High Prime Detection kit II (Roche Diagnostic). Images were acquired on a ChemiDoc imager (BioRad) using Image Lab software (version 5.2).
Mouse genotyping. Mouse genotypes were identified by PCR using primers to detect the residual loxP site as well as a SNP genotyping assay on genomic DNA ( Supplementary Fig. 1d, e). The presence of mutation was validated at gene and transcript levels by Sanger sequencing. Primers and probes are listed in Supplementary Table 4. Genomic DNA was extracted from mouse tails or ear punches using Direct PCR reagent (Viagen Biotech, Los Angeles, CA). Genotyping PCR reactions were done using Taq DNA polymerase (GE Biosciences) and primers flanking the residual loxP site in intron 6 of the mutant allele for Ldb3 Ala165Val/+ mice. TaqMan SNP genotyping assays were run on QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems) using primers flanking the mutation and Taq-Man allele-specific probes (Life Technologies). For confirmation by Sanger sequencing, DNA fragments were amplified using Platinum Taq DNA polymerase High Fidelity (Thermo Fischer Scientific) and primers flanking the mutation in Ldb3 exon 6.
RNA Isolation from tissue. Mouse skeletal muscle tissue was homogenized using Ambion Trizol Reagent (Life Technologies) and RNAse-free 0.5 mm zirconium oxide beads in a Bullet Blender homogenizer (Next Advance, Troy, NY). Total cellular RNA was isolated by phase separation using 1-Bromo-3-Chloropropane followed by purification with RNAeasy Mini Kit (Qiagen). RNA integrity was assessed using Agilent 2100 Bioanalyzer. Samples with RNA integrity number >8.5 were used for all assays. RNA was stored at −80°C.
Reverse transcription (RT). RT was done in 20 μl volume using 1μg total cellular RNA and Superscript VILO master mix or High Capacity cDNA RT kit containing random hexamers according to manufacturer's recommendations (Thermo Fischer Scientific). The cDNA samples were stored at −20°C.
Real-time quantitative PCR (qPCR). qPCR assays were run on QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems, Life Technologies) using Taq-Man probe for Ldb3 spanning constitutively spliced exons 2 and 3 (NM_001039071.2; Mm01208763_m1), thus allowing assessment of total Ldb3 transcript levels. The ΔΔC T method was used for quantification of Ldb3 transcript expression relative to a reference gene (Supplementary Fig. 1c). Expression levels were normalized to Myom1 and Rer1 genes (Taqman; Mm00440394_m1 and Mm00471276_m1, respectively) because these genes showed the lowest intragroup and between group variability for wildtype and mutant mice in RNAseq and microarray data (n = 20 mice). All primers and probes were validated for optimal amplification efficiency and all samples were used within the dynamic range of the assays. Linearity was established over four orders of magnitude using cDNA titration series. Multiple biological and technical replicates were used in each assay (n ≥ 3). At least three independent RT-qPCR assays were performed for each sample.
Gross motor performance and muscle strength testing. The rotarod test was done as described previously (RotaRod Series 8, IITC Life Sciences, CA, USA) 65 . Four paw grip strength was assessed by BIOSEB grip strength meter (BIOSEB, Vitrolles, France) using a method adapted from Treat-NMD Standard Operating Procedure (DMD_M.2.2.001). Five consecutive measurements of peak pull force were performed per day at 2 min intervals. The maximum strength normalized to body weight was used for analysis. These test sessions were performed during the afternoon hours of the light cycle (11 AM to 2 PM) in the behavioral core facility. A two-way analysis of variance with post hoc Bonferroni correction for multiple comparisons was used to determine effects for age and genotype (GraphPad Prism 8). The four-limb hang test was adapted from Treat-NMD Standard Operating Procedure (DMD_M.2.1.005). Mice were allowed to grasp the wire grid by all four limbs before inverting at 40 cm above a layer of bedding (NIH Instrumentation Core Apparatus). The hang time (s) was measured from the time the grid was inverted to the time that the mouse fell off the wire grid using a stopwatch. The test was repeated in mice two more times with 5 m of rest period in between testing. The longest suspension time from the three trials was used for analysis. The mouse body weight (g) was measured immediately before testing. The holding impulse [N*s; weight (g) × 0.00980665 (N/g) × the maximum hang time (s)] was compared between the groups. These tests were done during the afternoon hours of the light cycle (3 PM to 5 PM) in the behavioral core facility. An unpaired t-test (two-tailed) was applied for group comparison (GraphPad Prism 8). We used alpha level of 0.05 for all statistical tests. Investigators were blinded to the genotype of the mice. See Supplementary Data 2 and 4 for source data.
In vitro muscle contractility measurements. Mice were anesthetized by intraperitoneal tribromoethanol (250 mg/kg). Tissue isolation and measurement were adapted from Treat-NMD Standard Operating Procedure (DMD_M.1.2.002) with some modifications. Stainless steel hooks were fastened onto the proximal and distal tendons of the extensor digitorum longus muscle using 4.0 silk suture before muscle removal. The muscle was quickly installed into the measurement chamber containing Ringer's solution (1.2 mM NaH 2 PO 4 , 1 mM MgSO 4 , 4.83 mM KCl,137 mM NaCl, 2 mM CaCl 2 , 10 mM glucose, 24 mM NaHCO 3 , pH 7.4) pre-warmed to 37°C and equilibrated with 95% O 2 /5% CO 2 . An Aurora 1200 A in vitro dual lever system (Aurora Scientific, Aurora, ON, Canada) was used for all recordings. For each muscle, optimal length (L o ) and supramaximal (110% of maximal) current were determined at the beginning of each session using a series of isometric stimulations (1 Hz, 0.2 ms pulse width) separated by 20 s. The measured Lo for isometric force measurements was similar between mutant and wildtype mice (mean ± SEM: 11.3 mm ± 0.2 mm versus 11.6 ± 0.4 mm, p = 0.28). A forcefrequency curve was established using a series of tetanic isometric stimulations (20-250 Hz, 0.2 ms pulse width, 500 ms train duration). The muscle cross-sectional area (CSA) was calculated using the formula: muscle mass (mg)/[L o (mm) × 0.44 × 1.06 mg/mm 3 ]. Data are presented as specific isometric force (mN/mm 2 ), calculated using the formula: Force (mN)/CSA (mm 2 ). A two-way analysis of variance with post hoc Bonferroni correction for multiple comparisons was used to determine effects for genotype (GraphPad Prism 8). Investigators were blinded to the genotype of the mice. See Supplementary Data 3 for source data.
Histology. Frozen 10 μm thick transverse muscle sections were stained with modified Gomori trichrome and nicotinamide adenine dinucleotide dehydrogenase-tetrazolium reductase (NADH-TR) enzymatic activity using standard procedures. Heart tissues obtained from mice were fixed in 4% paraformaldehyde (PFA) overnight at 4°C and then cryoprotected in 30% sucrose at 4°C. For histological examinations, hearts were embedded into wax blocks of paraffin and 5 μm thick sections were stained with hematoxylin and eosin.
Immunofluorescence. Frozen 10 μm thick transverse muscle sections were fixed in 4% PFA for 10 min, permeabilized and blocked with 2.5% goat serum, 0.2% Triton X-100/phosphate buffered saline (PBS) for 1 h. Subsequently, slides were incubated with appropriate primary antibody (Supplementary Table 3) diluted in 1.25% goat serum with 0.1% Triton X-100/PBS overnight at 4°C. Sections were incubated with fluorophore-conjugated secondary goat anti-mouse or anti-rabbit antibodies (Jackson ImmunoResearch Laboratories) for 1 h at room temperature before being mounted and sealed. Images were acquired on a Leica confocal microscope (TCS SP5 576 II) equipped with a 40×/NA 1.3 oil immersion Plan-Apochromat objective lens and processed with Adobe Photoshop Creative Cloud v2017. Identical settings were used on all images of a given experiment and comparisons were made for muscle sections stained on the same slide.
For longitudinal muscle sections, mice were transcardially perfused with 4% PFA in PBS and skeletal muscle tissues were post-fixed in same buffer overnight at 4°C. Muscle tissues were cryoprotected in 30% sucrose for 48 h at 4°C, embedded in Optimal Cutting Temperature compound, and stored at −80°C until use.
Longitudinal sections (8 μm) were collected onto slides and antigen retrieval was carried out by boiling slides in citrate buffer (pH 6.0) for 10 min before immunostaining. Images were acquired on a Leica confocal microscope (TCS SP5 576 II) using a 63×/NA 1.4 oil immersion Plan-Apochromat objective and processed as described for frozen sections.
For fiber-type staining, unfixed frozen transverse soleus and vastus muscle sections (10 and 8 µm, respectively) were blocked with 10% goat serum/PBS for 1 h and then incubated overnight at 4°C in antibodies to myosin heavy chain (MyHC)  types I, IIA, IIB, and IIX (BA-F8, SC-71, BF-F3, and 6H1, respectively,  Developmental Studies Hybridoma Bank; Supplementary Table 3). Fluorescenceconjugated secondary antibodies to different mouse immunoglobulin subtypes (IgG2b -Alexa Fluor 488 or 568, and IgM -Alexa Fluor 568, ThermoFisher Scientific; CY5-IgG1, Abcam) were applied for 1 h at room temperature to visualize MyHC expression. In addition, wheat germ agglutinin -Alexa Fluor 488 conjugate (WGA; ThermoFisher Scientific) was applied to stain muscle membrane for fiber type-size quantitation. Sections were mounted with Fluoromount -G with DAPI (Southern Biotech). Images were obtained using a Leica epifluorescent microscope (DMI6000 SD) equipped with a motorized stage. Images were acquired in Leica Application Suite (LAS) X software with a 20X/NA 0.4 objective. The LAS X Navigator tool was used to take images from adjacent fields and digitally stitch them (with 10% overlap) to form a single image of the entire soleus muscle crosssection for analysis. For vastus muscle, nonoverlapping areas were photographed for analysis. Myonuclei counting, muscle fibers with sarcoplasmic protein aggregate quantitation, muscle fiber typing (using the counting tool), and minimal Feret's diameter measurement for fiber size were performed for all fibers excluding those on the edges by two individuals using open-source image processing software Image J/Fiji (National Institutes of Health, Bethesda, MD).
Electron microscopy. Mice were transcardially perfused with 2% glutaraldehyde and 2% paraformaldehyde in PBS and skeletal muscles were post-fixed in same buffer in 0.1 M sodium cacodylate. The tissue embedding and staining were carried out in the NINDS electron microscopy facility using a standard approach. Images were acquired on an electron microscope (JEOL 200CX, Jeol, Inc.) and processed with Adobe Photoshop Creative Cloud v2017.
Immunoblotting. Protein was extracted from transverse muscle midbelly cryosections (n = 15, 10 μm thick) using SDS buffer (100 mM Tris-HCl; pH 6.8, 4% SDS, 20% glycerol, 2% β-ME, 25 mM EDTA pH 8) supplemented with protease inhibitor (Roche). Equal volumes were loaded for relative protein quantification by immunoblotting. For PKCα and TSC2 levels, protein was extracted from muscle tissue (30 mg) homogenized in 600 μl T-PER Reagent (Thermo-Fisher) supplemented with protease and phosphatase inhibitors (Roche) using polytron homogenizer (ThermoScientific). Supernatants were collected after centrifuging the samples at 10,000 × g for 5 min at 4°C. Proteins were quantified using Bradford assay. Proteins resolved on Tris-Glycine polyacrylamide gels were transferred to nitrocellulose membrane (Life Technologies). After blocking in 5% BSA, 0.2% Tween 20/PBS for 1 h at room temperature, the membranes were incubated with primary antibodies in 5% BSA, 0.1% Tween 20/PBS overnight at 4°C and then with secondary antibodies for 1 h at room temperature. Secondary antibodies were conjugated with IRDye (LI-COR Biosciences). Odyssey Imaging Systems and Image Studio Lite software were used for detection and quantification, respectively (LI-COR Biosciences). HRPconjugated secondary antobodies (Jackson ImmunoResearch Laboratories) and the ChemiDoc XRS + Molecular Imager (Bio-Rad, Hercules, CA) were used for detection in GST pulldown assay and for blots shown in Supplementary Fig. 5. At least three independent assays were performed for each sample. Multiple biological replicates were used in each assay (n ≥ 3). See Supplementary Fig. 6 for full-length blot images and Supplementary Data 1, 8, and 9 for source data.
Immunoprecipitation. Mouse tibialis anterior muscle lysates were homogenized on ice using polytron homogenizer in buffer (20 mM Tris-HCl pH 7.5, 100 mM KCl, 5 mM EGTA, 1% Triton X-100, and EDTA-free protease inhibitor). The homogenized lysates were incubated for 1 h at 4°C with gentle agitation and then centrifuged at 3000 g for 30 min at 4°C. The pellet containing myofibrillar fraction was homogenized one more time and washed twice in buffer (20 mM Tris-HCl pH 7.5, 100 mM KCl, 1 mM DDT, and EDTA-free protease inhibitor) before resuspending in storage buffer (20 mM Tris-HCl pH 7.5, 100 mM KCl, 1 mM DDT, 20% glycerol). The lysates containing myofibrillar fraction were precleared with protein A Dynabeads (Invitrogen) for 3 h at 4°C. The cleared lysates were incubated with either mouse anti-LDB3 antibody (Abnova) or mouse IGg2b (Cell signaling) for overnight at 4°C and subsequently with protein A Dynabeads for 3 h at 4°C. The beads retaining immune complexes were washed five times with wash buffer (50 mM Tris-HCl pH 8.0, 0.1% Triton X-100, 5 mM EDTA) and then heated in SDS sample loading buffer for 5 m at 95°C. The eluates and lysates were analyzed by immunoblotting. Immunoprecipitation of tagged proteins from transfected COS7 cell lysates was done as described previously 17 .
GST pulldown assay. The pulldown assays using recombinant GST-tagged LDB3 proteins were done as described previously 17 . At least three independent experiments were performed using the vastus muscle lysates of three wildtype mice.
Autophagosome flux. Measurement of autophagosome flux in mouse skeletal muscle tissue using lysosomal blockade by colchicine was done as described previously 21 . Mice were treated with intraperitoneal colchicine (0.4 mg/kg/day; Sigma) or vehicle alone (PBS pH 7.4) for 3 days. The day following the last injection, hindlimb muscles were collected for protein isolation. Levels of autophagosome markers LC3-II and sequestosome 1 (p62) were quantified by immunoblotting. GAPDH was used as a loading control. The samples were run on two gels (one mini gel could not accommodate all sample loadings) in the same electrophoresis tank. The protein transfer from both gels was done simultaneously on a single membrane which was then processed for immunoblotting. For densitometric analysis, all the band intensities were normalized to that of one PBS-treated Ldb3 +/+ sample in each membrane. See Supplementary Fig. 6b for full-length blot images. Autophagosome flux values were derived by subtracting corresponding protein values in vehicle treated from the colchicine treated conditions (Ldb3 +/+ COL minus Ldb3 +/+ VEH and Ldb3 Ala165Val/+ COL minus Ldb3 Ala165Val/+ VEH). Three independent assays were performed for each sample. See Supplementary Data 5 for source data.
Yeast two-hybrid screen. The yeast two-hybrid (Y2H) screen was performed by Hybrigenics Services (Paris, France) using human LDB3 exons 8 and 11 corresponding to LDB3-LΔex10 as a bait (amino acid 221 -320; NM_001080114; NP_001073583), and a human skeletal muscle cDNA library for interactors.
Yeast two-hybrid pairwise assay. Pairwise Y2H assays were done using the Matchmaker Gold system (Clontech), as described previously 17 .
RPPA analysis. The RPPA analysis on proteins isolated from frozen vastus muscle of 4-and 8-month-old Ldb3 +/+ and Ldb3 Ala165Val/+ mice (n = 5 per group, total 20 mice) was performed by the Functional Proteomics RPPA Core facility at MD Anderson Cancer Center (Houston, TX). Tissue lysate samples were serially diluted two-fold for 5 dilutions (undiluted, 1:2, 1:4, 1:8; 1:16) and arrayed on nitrocellulose-coated slides in an 11 × 11 format to produce sample spots. Sample spots were then probed with 248 antibodies (159 validated for RPPA) against total and phosphorylated proteins by a tyramide-based signal amplification approach and visualized by DAB colorimetric reaction to produce stained slides. Stained slides were scanned on a Huron TissueScope scanner (Huron Digital Pathology, Ontario, Canada) to produce 16-bit tiff images. Sample spots in tiff images were identified and their densities quantified by Array-Pro Analyzer 6.3. Relative protein levels for each sample were determined by interpolating each dilution curve produced from the densities of the 5-dilution sample spots using a "standard curve" (SuperCurve) for each slide (antibody). SuperCurve 1.5.0 (via Super-CurveGUI_2.1.1) was constructed by a script in R-package. All relative protein level data points were normalized for protein loading and transformed to linear values, which were designated "Normalized Linear" or "NormLinear" and were used for expression analysis. Only antibodies with quality control scores above 0.8 in the heatmaps were included in analysis. To identify differentially expressed proteins between age and gender-matched Ldb3 Ala165Val/+ mice and Ldb3 +/+ littermates, raw abundances detected per sample were log transformed (base 2), median centered, and variance-stabilized (median absolute deviation) using a Rpackage. Proteins having an absolute difference of means ≥ 1.5× and a corrected p ≤ 0.05 (Welch modified t-test under Benjamini-Hochberg false discovery rate multiple comparison correction; rounded to the nearest hundredth) were deemed to be differentially expressed between the groups. See Supplementary Data 6 and 7 for source data.
Statistics and reproducibility. The statistical analyses were performed using either GraphPad Prism 8 or R package. All individual data points are plotted as dot plot or bar-dot plot and presented as mean ± SEM or SD throughout the manuscript. Statistical testing method, sample size, replicates of experiments, and the p value are indicated in figure legends. A two-way ANOVA using the Bonferroni's multiple comparison test or an unpaired two-tailed t-test were done to determine differences between groups. Statistical significance was defined as p < 0.05. Welch modified t-test under Benjamini-Hochberg false discovery rate multiple comparison correction was applied to identify differentially expressed proteins between genotypes in the RPPA analysis using a R-package. Statistical significance was defined as absolute difference of means ≥ 1.5× and a corrected p ≤ 0.05.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Source data underlying plots shown in main figures are provided in Supplementary Data. Full blots are shown in Supplementary Fig. 6 in Supplementary Information.