Cullin-3–RING ubiquitin ligase activity is required for striated muscle function in mice

Control of protein homeostasis is an essential cellular process that, when perturbed, can result in the deregulation or toxic accumulation of proteins. Owing to constant mechanical stress, striated muscle proteins are particularly prone to wear and tear and require several protein quality–control mechanisms to coordinate protein turnover and removal of damaged proteins. Kelch-like proteins, substrate adapters for the Cullin-3 (Cul3)-RING ligase (CRL3) complex, are emerging as critical regulators of striated muscle development and function, highlighting the importance of Cul3-mediated proteostasis in muscle function. To explore the role of Cul3-mediated proteostasis in striated muscle, here we deleted Cul3 specifically in either skeletal muscle (SkM-Cul3 KO) or cardiomyocytes (CM-Cul3 KO) of mice. The loss of Cul3 caused neonatal lethality and dramatic alterations in the proteome, which were unique to each striated muscle type. Many of the proteins whose expression was significantly changed in the SkM-Cul3 KO were components of the extracellular matrix and sarcomere, whereas proteins altered in the CM-Cul3 KO were involved in metabolism. These findings highlight the requirement for striated muscle–specific CRL3 activity and indicate how the CRL3 complex can control different nodes of protein interaction networks in different types of striated muscle. Further identification of Cul3 substrates, and how these substrates are targeted, may reveal therapeutic targets and treatment regimens for striated muscle diseases.

Human cells express up to 10,000 proteins at a given time (1), many of which require folding and assembly into higher-order, macromolecular structures (2). Therefore, maintaining protein homeostasis (proteostasis) is critical for cellular integrity. To ensure proteostasis, cells have evolved an elaborate network of protein quality-control factors involved in protein synthesis, folding, and degradation (3). Damaged proteins must be recognized and degraded by the proteostasis network to maintain cellular function and to avoid aggregation and proteotoxic stress. Striated muscle, in particular, is especially vulnerable to protein damage, due to the persistent wear and tear of sarcomeric proteins during force-generating contraction. Several myopathies in both skeletal and cardiac muscle are characterized by protein aggregates, which result from mutations in sarcomeric proteins, such as desmin (4), nebulin (5), myotilin (6), and filamin C (7).
Several intracellular proteolytic systems exist in striated muscle that serve to remove damaged proteins and coordinate protein turnover: the calpain system (8), the autophagylysosome system (9), and the ubiquitin-proteasome system (UPS) 3 (10). The UPS is the primary degradation system responsible for clearance of soluble proteins that are misfolded, damaged, mutated, or oxidized (11,12). Perturbation of the UPS results in cellular dysfunction (13), and this is evidenced in muscle by a growing list of BTB-Back-Kelch (BBK) family members that give rise to muscular disease when mutated (14). BBK family members encode proteins containing an N-terminal bric a brac, tramtrack, and broad-complex (BTB) domain, a BACK domain, and a C-terminal Kelch repeat domain (15) and function as substrate-specific adapters for the Cullin-3 (Cul3) E3 ligase complex, facilitating the ubiquitination and, in many cases, the subsequent proteosomal degradation of their specific substrate (16 -18). Mutations in the BTB domain of Kelch-like protein 9 (Klhl9) result in an early onset autosomal dominant distal myopathy (19). As BBK proteins interact with Cul3 through their BTB domain (20), the mutation in Klhl9 most likely disrupts its interaction with Cul3, preventing proper targeting of substrate turnover (19). Moreover, mutations in Kbtbd13, Klhl40, and Klhl41 give rise to nemaline myopathy in mice and humans (21)(22)(23)(24)(25), and we recently demonstrated that mice lacking Klhl31 develop myopathy with centralized nuclei, central cores, Z-disc streaming, and myofibril degeneration (26). These findings indicate that elements of the Cul3-RING ligase (CRL3) complex are critical regulators of muscle development and function.
Cul3 is a member of the Cullin family of E3 ligases, which consists of seven members: Cul1, -2, -3, -4A, -4B, -5, and -7 (27). Whereas many substrate adapters use an intermediate protein to bind to Cullin family members, such as the SCF (Skp1-Cul1-F-box) complex, BBK proteins bind directly to Cul3 through their BTB domain (27). As Cul3 is ubiquitously expressed and interacts with multiple substrate adapters, it is not surprising that it regulates many processes, including mitosis (28,29), cellular stress (30 -32), electrolyte homeostasis (33), and cell death (34). Recently, the Cullin family was also demonstrated to be critical for the differentiation and maturation of myoblasts (35). Inhibiting Cullin E3 ligase activity in myoblasts with the neddylation inhibitor MLN4929 showed that Cullin activity is required for myoblast differentiation, fusion, and maturation (35).
Given the increasing evidence that Cul3 and its BBK substrate adapters play critical roles in muscle disease, we investigated the function of the CRL3 complex in striated muscle with the intention of identifying CRL3-regulated substrates that would illuminate critical targets and regulatory pathways in muscle biology and disease. To approach this problem, we generated mice with either skeletal muscle-specific (SkM-Cul3) or cardiomyocyte-specific (CM-Cul3) loss of Cul3. Our findings indicate that the CRL3 complex is essential for both skeletal and cardiac muscle function; SkM-Cul3 knockout (KO) and CM-Cul3 KO mice are neonatal lethal due to nonfunctional skeletal muscle and severe cardiomyopathy, respectively. Proteomic analyses indicate changes in protein groups involved in muscle contraction, the extracellular matrix, and metabolism and provide new insights into how the CRL3 complex regulates the muscle proteome during development and differentiation.

Skeletal muscle-specific deletion of Cul3 causes neonatal lethality
Mice with global loss of Cul3 are early embryonic lethal resulting from mitotic defects (28). To explore the effect of skeletal muscle-specific Cul3 loss of function and to avoid a general mitotic defect, we crossed mice carrying a conditional allele of Cul3 (36) with transgenic mice expressing Cre recombinase under control of the Myogenin promoter and the skeletal muscle-specific enhancer of the Mef2c gene (37). This transgene expresses in differentiating myocytes following myoblast proliferation at E8.5.
To assess the efficiency of Cul3 deletion by Myogenin:Cre, we immunoblotted for Cul3 using whole-hind-limb lysates from WT and SkM-Cul3 KO mice at postnatal day 0 (P0). Cul3 expression was decreased in SkM-Cul3 KO mice, as detected by Western blotting and quantitative RT-PCR analyses. Residual levels were observed in SkM-Cul3 KO mice, which were most likely derived from other tissue types in the hind limb ( Fig. 1, A-C). SkM-Cul3 KO embryos at E18.5 had a moon-shaped appearance, wrist drop, abnormal spinal curvature, and reduced body weight (Fig. 1, D and E), physical features that are characteristic of defective myogenesis or excitationcontraction (EC) coupling (38 -40). SkM-Cul3 KO mice were born in normal Mendelian ratios but became cyanotic shortly after birth and died, while maintaining a reduction in body weight (Fig. 2, A and B). Histological analysis of the SkM-Cul3 KO diaphragm at P0 revealed a substantial reduction in size (Fig. 2C), whereas SkM-Cul3 KO lungs contained smaller alve-oli compared with heterozygous mice (Fig. 2D), suggesting that the respiratory muscles are not capable of inflating the lungs following birth. As the hearts of SkM-Cul3 KO mice were beating at birth, these results indicate these mice expire perinatally due to the inability to breathe.

Musculature defects in SkM-Cul3 KO mice
Histological analysis of SkM-Cul3 KO tongue musculature revealed disorganized myofiber architecture compared with heterozygous controls (Fig. 2E). Myoblast fusion was unaffected, as evidenced by multiple nuclei in individual fibers; however, myofibers in SkM-Cul3 KO mice appeared smaller and had a more fibrous appearance (Fig. 3A). The decrease in fiber size resulted in an overall reduction in size of muscle groups in fore limbs of SkM-Cul3 KO (Fig. 2B). Indeed, crosssectional analysis demonstrated that SkM-Cul3 KO myofibers were significantly smaller (Fig. 3, C and D). EM of SkM-Cul3 KO quadriceps showed disorganized sarcomeres, characterized by Z-disc streaming and nearly complete loss of the sarcomeric M-line. (Fig. 3E). As SkM-Cul3 KO mice contain multinucleated myofibers but are born with paralysis, these findings suggest that the absence of Cul3 does not affect myoblast differentiation or fusion but may impair aspects of myofiber maturation, such as innervation, or EC coupling.

Cul3 is required for muscle function Altered proteome in SkM-Cul3 KO mice
To identify proteins regulated by Cul3, we harvested gastrocnemius/plantaris/soleus (GPS) muscles from WT and SkM-Cul3 KO embryos at E18.5 for proteomic analysis. There were 52 proteins down-regulated and 53 proteins up-regulated in SkM-Cul3 KO mice compared with WT (Table S1). Gene ontogeny analysis of the affected proteins showed that the most significantly changed proteins participate in muscle contraction (Fig. 4A). Other changed proteins were involved in collagen fibril organization and regulation of muscle adaptation. When we analyzed the pathways in which the affected proteins are associated, we found that many are components of extracellular matrix (ECM)-receptor interactions, muscle contraction, and regulators of cardiomyopathy (Fig. 4B). Moreover, most of the top 10 up-regulated proteins are associated with the ECM (Fig. 4C), whereas most of the top 10 down-regulated proteins are contractile proteins or are associated with the sarcomere (Fig. 4D). To confirm the changes observed in the proteomics analysis, we chose to immunoblot against one of the top down-regulated proteins found in the SkM-Cul3 KO mice. Western blot analysis for myosin light chain 2 (Myl2) confirmed the down-regulation in SkM-Cul3 KO mice observed in proteomics (Fig. 4E). These findings suggest that Cul3-depen-  Cul3 is required for muscle function dent proteostasis broadly controls elements of ECM deposition and sarcomere maturation/function.

Cardiomyocyte-specific deletion of Cul3 causes neonatal lethality
We next explored the role of Cul3 in cardiomyocytes by breeding Cul3 fl/ϩ mice with ␣MHC-Cre transgenic mice (41).
CM-Cul3 KO mice were born in normal Mendelian ratios and were indistinguishable from WT or ␣MHC:Cre;Cul3 fl/ϩ controls at birth. Western blot analysis from neonatal heart confirmed the down-regulation of Cul3 (Fig. 5, A and B). Although CM-Cul3 KO mice seemed unaffected at birth, they began to exhibit a failure-to-thrive phenotype and were much smaller compared with controls by P5 (Fig. 5C). CM-Cul3 KO mice Cul3 is required for muscle function died around P6 with severe cardiomyopathy, consistent with dysmorphic myocardial architecture, dilated right ventricle and atrium, and atrial thrombi (Fig. 5D). Closer inspection of myocardial histology of CM-Cul3 KO mice revealed widespread cardiomyocyte vacuolization and protein aggregate deposition (Fig. 5E). Heterozygous mice had a few sporadic vacuoles; however, CM-Cul3 KO hearts contained rampant vacuolization, consisting of vacuoles significantly larger than controls (Fig.  5F). To determine whether sarcomeric organization was disrupted, we stained heart sections for cardiac troponin T (cTnT). CM-Cul3 KO mice showed mislocalization and aggregation of cTnT compared with controls (Fig. 6). Whereas ␣MHC:Cre is transiently expressed at low levels around E8 -10, it is re-expressed at more robust levels at P2 (42). Given that CM-Cul3 KO mice were unaffected at birth but expired by P6, these findings demonstrate the importance of cardiac Cul3mediated proteostasis and illustrate how quickly the integrity of the cardiomyocyte is compromised when the proteome is perturbed.

Altered metabolic profile in CM-Cul3 KO hearts
To determine which proteins are regulated by Cul3, we harvested WT and CM-Cul3 KO hearts at P4 for proteomic analysis. A total of 969 proteins were significantly changed in CM-Cul3 KO mice (Table S2). The majority of affected proteins (582 proteins) were down-regulated, whereas 387 proteins were up-regulated. Gene ontogeny analysis of the affected proteins in CM-Cul3 KO mice revealed that the most significantly changed proteins participate in oxidation-reduction and other metabolic processes (Fig. 7A). Other changed proteins were involved in sarcomere organization and cardiac muscle contraction. When we analyzed the pathways in which the affected proteins are associated, we found that many are components of oxidative phosphorylation and metabolic pathways (Fig. 7B) and are found among the most up-regulated and down-regulated proteins (Fig. 7, C and D). Interestingly, protein aggregate diseases, such as Parkinson's, Huntington's, and Alzheimer's disease, were top pathways associated with the affected proteins in CM-Cul3 KO hearts. In addition to Cul3 being significantly down-regulated (Fig. 7D), one of the most up-regulated proteins was NAD(P)H dehydrogenase 1 (Nqo1), a known target of Nrf2, which is regulated in a Cul3-Keap1-dependent manner (43) (Fig. 7C). These findings suggest that Cul3-dependent proteostasis is a critical regulator of cardiac anti-oxidative and metabolic processes.

Discussion
For cells to maintain functionality, it is paramount that misfolded, unnecessary, and damaged proteins be recognized and degraded. Uncontrolled proteostasis leads to accumulated

Cul3 is required for muscle function
depositions of protein aggregates and inclusion bodies and, ultimately, cellular dysfunction. Proteostasis allows differentiated cells to adapt their proteome to intrinsic and environmental changes to prevent disease onset, which is crucial for postmitotic cell types, such as myofibers and cardiomyocytes. Mitotic cells have the ability to clear aggregation by sequestering protein aggregates and asymmetrically dividing, leaving one healthy daughter cell and one aggregate-containing daughter cell, which will proceed through apoptosis (44,45). In addition to being post-mitotic, striated muscle must overcome two additional hurdles to achieve proteostasis. First, muscle contains a multitude of contractile and specialized proteins involved in EC coupling that require constant maintenance due to the inherent wear and tear of contraction. Second, energy production in muscle (especially cardiomyocytes) primarily depends on oxidative phosphorylation, and, consequently, high levels of ROS are generated, which can oxidize and damage proteins (46,47).
Recent findings have underscored the importance of CRL3 components in muscle disease (14,26). In this study, we sought to gain a broader perspective on the role of CRL3 in striated muscle to identify CRL3-regulated targets and pathways in muscle biology. To achieve this, we generated mice lacking Cul3 in developing skeletal muscle or cardiomyocytes. SkM-Cul3 KO mice were born alive but had many features in common with mouse models of severe defects in myogenesis (38), neuromuscular junction (NMJ) function (39), and EC coupling (40). However, defects in myogenesis and myocyte fusion can be ruled out, as SkM-Cul3 KO mice have multinucleated myofibers. Following birth, SkM-Cul3 KO mice quickly became cyanotic and died. The loss of Cul3 in myocytes did not affect myoblast differentiation or fusion, as SkM-Cul3 KO mice had multinucleated myotubes, which suggests that there may be defects in NMJ function or EC coupling. Interestingly, a previous study reported that Klhl8 degrades the NMJ protein Rapsyn, thereby maintaining proper levels of Rapsyn and acetylcholine receptor clustering (48); however, Rapsyn was not changed in skeletal muscle lacking Cul3. Other than reduced muscle mass and structural aberrations, SkM-Cul3 KO muscle appears to differentiate and fuse normally, indicating that Cul3 is essential for muscle maturation.
Proteomic analysis identified numerous changes in protein levels, and surprisingly, many of the significantly changed proteins were down-regulated, which is in contrast to the current dogma of Cullin E3 ligase activity. Three possible scenarios could explain this. First, Cul3 could use most of its BTB-containing substrate adapters (i.e. Kelch-like proteins) to stabilize its targeted proteins in skeletal muscle, while only utilizing very few Kelch-like proteins to target substrates for degradation. Ubiquitination targets proteins not only for degradation, but also for receptor internalization, assembly of protein complexes, protein localization, DNA repair, and signaling (49). Second, Cul3 could function in a temporal manner. In other words, during skeletal muscle development and differentiation, the CRL3 complex could target substrates to maintain overall protein synthesis and growth during a process where multitudes of proteins are being synthesized and assembled into the highly complex sarcomere. Then, following birth and postnatal muscle growth and maturation, the CRL3 complex would function primarily to clear unnecessary and damaged proteins. Finally, and perhaps most likely, the loss of Cul3 in nascent myocytes results in the deregulation of several proteins that accumulate and cause proteotoxic stress, and in an attempt to compensate for this stress, the synthesis of sarcomeric proteins could be reduced.
Similar to SkM-Cul3 KO mice, CM-Cul3 KO mice also showed a vast number of proteins that were down-regulated. Perhaps the most surprising findings of the current study were the differences in phenotypes and groups of proteins that were changed between SkM-Cul3 and CM-Cul3 KO mice. CM-Cul3 KO mice had widespread vacuolization and protein aggregates, and gene ontogeny analysis indicated that most of the significantly changed proteins participate in metabolism. In contrast, SkM-Cul3 KO fibers lacked the severe vacuolization and protein aggregates, and gene ontogeny analysis indicated that most of the altered proteins in skeletal muscle are associated with the ECM and sarcomere. Additionally, there were few significantly changed proteins in common between SkM-Cul3 and CM-Cul3 KO mice. Interestingly, of the changed proteins that were shared between SkM-Cul3 and CM-Cul3 KO muscle, some were regulated in a contrasting manner, such as PDZ and LIM domain 5 (Pdlim5), myosin heavy chain 7 (Myh7), Ras homolog family member b (RhoB), and Nudix hydrolase 4 (Nudt4) ( Table 1). There were only three proteins that were changed in the same direction between SkM-Cul3 and CM-Cul3 KO mice: Cul3, transcription elongation factor A3 (Tcea3), and plastin 3 (Pls3). Little is known about Tcea3, and although it is highly enriched in skeletal muscle (http://www.biogps.org/), 4 there is only one report on a possible function in muscle, suggesting that it promotes differentiation of isolated bovine satellite cells (50). Moreover, we previously found Tcea3 to be regulated by Mef2 in a mouse model of skeletal muscle regeneration (51). Pls3 is an actin-bundling protein that has been suggested to partially rescue the effects of spinal muscular atrophy, in part by improving NMJ activity (52,53). In both SkM-Cul3 and CM-Cul3 KO mice, we observed changes in the cytoskeletal network that included microtubule-associated proteins and both sarcomeric and nonsarcomeric actins, suggesting that Cul3 plays a role in regulating the actin-cytoskeletal network (Tables S1 and S2).

Cul3 is required for muscle function
Keap1 is a well-characterized substrate adapter for Cul3 that targets cytoplasmic Nrf2 for degradation, prohibiting its nuclear entry where it activates antioxidant genes in response to oxidative stress (43). In the absence of Cul3, we observed significant increases in Nrf2 target genes in CM-Cul3 KO mice, such as NAD(P)H dehydrogenase (Noq1), pirin (Pir), peroxiredoxin (Prdx1), thioredoxin (Txn), thioredoxin reductase 1 (Txnrd1), NADP-dependent malic enzyme (Me1), and the regulatory and catalytic domains of glutamate cysteine ligase (Gcl) (54). Interestingly, none of these proteins were up-regulated in SkM-Cul3 KO mice. These findings are consistent with the very modest phenotype of skeletal muscle-specific Keap1 KO mice (55) and demonstrate how Cul3 differentially regulates the proteome in skeletal muscle and cardiomyocytes.
The results of the current study highlight the complexity of Cul3-dependent proteostasis and illustrate how Cul3 can regulate different aspects of cellular biology in two similar cell types. Muscle has a capacity to adapt to external stimuli and physiological demand, and one way in which this adaptation is accomplished is by altering the proteome. The identification of Cul3 substrates and determination of the ways in which these substrates are targeted at the molecular level during conditions such as growth and maintenance, cardiac remodeling, atrophy, and disease will enhance our understanding of the muscle proteostasis network and may uncover novel therapeutic approaches to muscle disease.

Mice
Cul3 conditional mice were purchased from the Jackson Laboratory (Cul3tm1Jdsr/J 028349) (36). SkM-Cul3 KO mice were generated by crossing Cul3 fl/fl mice with Myogenin:Cre mice (37), whereas CM-Cul3 KO mice were generated by crossing Cul3 fl/fl mice with ␣MHC:Cre mice (41). All mice were maintained on a C57BL/6 background. Animal work described in this paper was approved and conducted under the oversight of the University of Texas Southwestern institutional animal care and use committee.

Western blot analysis
Lysates were prepared by pulverizing flash-frozen tissue. Tissue powder was homogenized in radioimmune precipitation buffer (Sigma) with the addition of protease inhibitors (cOmplete ULTRA mini tablet) on ice in a glass Dounce homogenizer. Protein concentrations were determined using a BCA protein assay kit (Pierce). Samples were separated on Any kDa TM Tris-glycine-buffered polyacrylamide gels (Bio-Rad) and transferred onto Immobilon P membranes (Millipore). Membranes were blocked for 1 h at room temperature with 5% nonfat dry milk in TBST, and primary antibody hybridization was carried out overnight at 4°C using the following antibodies: Cullin-3 (1:250; Sigma) and glyceraldehyde-3-phosphate dehydrogenase (1:10,000; Millipore).

Histology
Tissues were fixed with 4% paraformaldehyde in PBS for 48 h at room temperature. Paraffin-embedded sections were then subjected to standard hematoxylin and eosin (H&E) staining protocols. Images were taken and processed with a Keyence BZ-X700 scope. For immunofluorescent staining, skeletal muscle tissues were embedded in a mixture of OCT (Fisher) and gum tragacanth (Sigma-Aldrich) and flash-frozen in a 2-methylbutane reservoir submerged in liquid nitrogen, followed by cryostat sectioning at 10 M. Sections were air-dried for 15 min and fixed with 1% paraformaldehyde for 2 min. Sections were then permeabilized for 15 min with PBST (0.3% Tween 20), followed by blocking with 5% goat serum (Sigma-Aldrich) in PBST for 30 min. Primary antibodies were diluted in 2% goat serum in PBST and added overnight at 4°C in a humidified chamber using the following antibodies: cTnT (Sigma; 1:1000), My32 (Sigma; 1:1000), wheat germ agglutinin (Thermo Fisher Scientific; 1:20), Alexa Fluor goat anti-mouse 555 (Thermo Fisher Scientific; 1:250), and Alexa Fluor goat anti-mouse 488 (Thermo Fisher Scientific; 1:250). Confocal images were taken with a Zeiss LSM-800 microscope. Electron microscopic images were obtained using E18.5 quadriceps as described previously (56).

Proteomics
CM-Cul3 KO in vivo TMT-labeling and 10-fraction LC/LC-MS/MS was performed by the Proteomics and Metabolomics Shared Resource at Duke University School of Medicine, as described previously (26). SkM-Cul3 KO in vivo TMT-labeling and 10-fraction LC/LC-MS/MS was performed by the Proteomics Core Laboratories at Washington University School of Medicine.

Bioinformatic analyses
Gene ontogeny analysis was performed using DAVID with ILLUMINA_ID identifiers and Mus musculus as the background set. Biological process level 1 was used for gene ontogeny term enrichment. The Kyoto Encyclopedia of Genes and Genomes (KEGG) was used for pathway analyses.

Statistics
Values are given as mean Ϯ S.D. Differences between two groups were assessed using unpaired two-tailed Student's t