Loss of atrogin-1 results in contraction-dependent fiber failure. To determine the role of atrogin-1 in skeletal muscle, we used CRISPR/Cas9 genome editing to generate an atrogin-1 mutant (atrogin-1pc43/pc43, referred throughout to as atrogin-1–/–) with a 34–base pair insertion in exon 1, which resulted in a frameshift and incorporation of a premature stop codon (Figure 1A). Examination of atrogin-1 mRNA levels revealed a significant reduction in atrogin-1–/– mutants compared with homozygous wild-type (atrogin-1+/+) sibling embryos (Figure 1B), suggestive of nonsense mediated decay of the mRNA. We next examined skeletal muscle integrity in the atrogin-1 mutants by staining the muscle with phalloidin, a high-affinity filamentous actin probe. While under normal conditions the muscle structure in atrogin-1 heterozygous (atrogin-1+/–) and atrogin-1–/– mutants 3 days after fertilization (3 dpf) was indistinguishable from that of atrogin-1+/+ embryos (Figure 1, C–F), atrogin-1+/– and atrogin-1–/– larvae displayed muscle fiber detachment following incubation in methyl cellulose, a viscous solution that increases the load on the muscle (Figure 1, G–J). Additionally, at 6 dpf, both atrogin-1+/– and atrogin-1–/– larvae displayed sporadic muscle fiber detachment (Figure 1, K–N), which was exacerbated following incubation in methyl cellulose (Figure 1, N–R). To validate that the observed phenotype was indeed due to a deficiency in atrogin-1, we compared the pathology evident in a second atrogin-1 mutant allele (atrogin-1pc44/pc44) previously generated using zinc finger nuclease technology. Examination of the muscle in 6 dpf atrogin-1pc44/pc44 mutants revealed presence of detached fibers identical to the phenotype seen in the atrogin-1–/– mutants (Supplemental Figure 1, A–D; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.167578DS1). As further evidence that atrogin-1 deficiency is responsible for the fiber detachment phenotype seen in the mutant, we performed a rescue experiment whereby we overexpressed fluorescently tagged actin (Lifeact-GFP) or fluorescently tagged atrogin-1 (atrogin-1-GFP) in the muscle of atrogin-1–/– mutants and examined their ability to prevent muscle fiber disintegration. As presented in Supplemental Figure 1, E–G, while 38% (23 of 60 fibers in 6 larvae) of Lifeact-GFP–expressing muscle fibers underwent detachment, atrogin-1-GFP expression (22 fibers in 6 larvae) was sufficient to prevent fiber disintegration. Taken together, a reduction in or a loss of atrogin-1 resulted in load-dependent muscle detachment.

Atrogin-1 deficiency results in contraction-dependent muscle fiber detachment. Schematic of wild-type atrogin-1 (atrogin-1+/+) and mutant atrogin-1 (atrogin-1–/–) protein structure and mRNA sequence, with the mutant predicted to incorporate a premature stop in exon 1. The mutant was generated using CRISPR/Cas9 genome editing, which resulted in a 34 bp insertion (red). Numbers in the protein box are amino acids, and numbers in the mRNA box are base pairs. (B) qRT-PCR analysis showing significant reduction in atrogin-1 levels in atrogin-1–/– mutants compared with atrogin-1+/+ wild-type larvae. Error bars represent mean ± SEM for 3 replicate experiments, with each experiment comprising a pooled sample of at least 5 fish. *P < 0.05 determined using a 1-way ANOVA with Tukey’s multiple correction post hoc test. Muscle fibers span the entire length of the somite in the 3 dpf atrogin-1+/+ (C), atrogin-1 heterozygous (atrogin-1+/–) (D), and atrogin-1–/– mutant (E) larvae, as seen by F-Actin labeling. (F) Quantification of the muscle phenotype, with atrogin-1+/+, atrogin-1+/–, and atrogin-1–/– displaying indistinguishable muscle structure, as determined using a χ2 test. Incubation of 3 dpf atrogin-1+/– (H) and atrogin-1–/– (I) in methyl cellulose results in muscle fiber detachment, which is not evident in atrogin-1+/+ larvae (G). (J) Percentage of affected atrogin-1+/+, atrogin-1+/–, and atrogin-1–/– larvae, with the latter 2 genotypes having a significant increase in the proportion of fish displaying the muscle fiber detachment, as determined using a χ2 test. At 6 dpf, atrogin-1+/– (L) and atrogin-1–/– (M) display sporadic muscle fiber detachment but not in atrogin-1+/+ larvae (K). (N) Percentage of affected atrogin-1+/+, atrogin-1+/–, and atrogin-1–/– larvae, with the latter 2 genotypes having a significant increase in the proportion of fish displaying the muscle fiber detachment, as determined using a χ2 test. Methyl cellulose incubation of 6 dpf atrogin-1+/– (P) and atrogin-1–/– (Q) results in muscle fiber detachment, which is not evident in atrogin-1+/+ larvae (O). (R) Percentage of affected atrogin-1+/+, atrogin-1+/–, and atrogin-1–/– larvae, with the latter 2 genotypes having a significant increase in the proportion of fish displaying the muscle fiber detachment, as determined using a χ2 test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All experiments were performed in triplicate, with the total number of fish examined in each replicate documented in Supplemental Table 2.

Atrogin-1 deficiency results in impaired membrane integrity and apoptosis. We next examined the biological basis of the atrogin-1 mutant muscle detachment phenotype by crossing it to a double transgenic line (Tg(actc1b:Lifeact-GFP);Tg(actc1b:CAAX-mCherry). In this line actin filaments within the muscle fibers are labeled with GFP, and membrane and t-tubules with mCherry. In methyl cellulose–treated 6 dpf wild-type sibling larvae, muscle cells span the entire length of the somite, with the sarcolemma fully surrounding each cell (Figure 2, A–C). In stark contrast, in atrogin-1–/– mutants, detached muscle fibers were surrounded with irregular sections of sarcolemma, with abnormal sarcolemma vacuoles also evident (Figure 2, D–F), indicative of a deficit in sarcolemma integrity. To test sarcolemmal permeability, we performed precardiac sinus injections of Evans blue dye, a small-molecular-weight dye that while impermeable in cells with normal sarcolemma, selectively accumulates in cells in which the sarcolemma lacks integrity. Using this technique, we revealed that while muscle fibers in atrogin-1+/+ larvae had no Evans blue dye uptake (Figure 2, G–I), consistent with the presence of intact sarcolemma, muscle cells in atrogin-1–/– mutants displayed an accumulation of dye (Figure 2, J–L), confirming an impairment in membrane integrity. Finally, to determine if the retracted muscle cells seen in the atrogin-1–/– mutant undergo apoptosis, we performed a TUNEL assay in methyl cellulose–treated, 6 dpf atrogin-1+/+ and atrogin-1–/– mutants. In line with the normal muscle structure observed in atrogin-1+/+ wild-type larvae, we observed no apparent TUNEL labeling (Figure 2, M–O). In contrast, atrogin-1–/– mutants displayed increased numbers of TUNEL+ nuclei, which coincided with areas of muscle detachment (Figure 2, P–R). Collectively, these results highlight that loss of atrogin-1 results in a loss in membrane integrity and increased apoptosis, consistent with the phenotypes presented in atrogin-1–deficient cardiomyocytes.

Atrogin-1 deficiency results in impaired membrane integrity and apoptosis. Live images of methyl cellulose–treated 6 dpf atrogin-1+/+ (A–C) and atrogin-1–/– (D–F) on the (Tg(actc1b:Lifeact-GFP);Tg(actc1b:CAAX-mCherry) background, whereby the actin filaments within the muscle fibers are labeled with GFP and membrane and t-tubules with mCherry. While muscle cells in the atrogin-1+/+ wild-type larvae span the entire length of the somite, with the sarcolemma fully surrounding each cell, atrogin-1–/– mutants display detached muscle fibers that are surrounded with irregular sections of sarcolemma, along with presence of abnormal sarcolemma vacuoles. Live images of methyl cellulose–treated 6 dpf atrogin-1+/+ (G–I) and atrogin-1–/– (J–L) larvae injected with Evans blue dye. atrogin-1–/– mutants displayed an accumulation of the dye, which is absent in atrogin-1+/+ larvae. TUNEL staining on 3 dpf methyl cellulose–treated atrogin-1+/+ (M–O) and atrogin-1–/– (P–R) larvae, with the latter displaying TUNEL in areas where the muscle cell had detached. Scale bar: 100 μm (A–F); 300 μm (G–R).

Untargeted proteomics identified a role of atrogin-1 in regulating BiP levels. Having characterized the atrogin-1 mutant phenotype, we wanted to examine the mechanisms by which loss of atrogin-1 results in fiber detachment. Given the striking similarity in phenotypes between the zebrafish atrogin-1–KO animals and the atrogin-1–KO cardiomyocytes, we hypothesized that the same protein(s) may be dysregulated in both systems following the absence of atrogin-1, resulting in the pathology observed. To identify proteins differentially regulated in the zebrafish mutant, we performed mass spectrometry on protein lysates obtained from 6 dpf atrogin-1+/+ and atrogin-1–/– larvae. Using this strategy, we identified a total of 4,242 distinct proteins across the 6 samples (Supporting Data Values file). Of these, 162 proteins were differentially expressed in the atrogin-1–/– mutant larvae compared with atrogin-1+/+ larvae (Figure 3A), of which, 69 were upregulated and 93 downregulated (Supporting Data Values file). Given that a loss in atrogin-1 is expected to result in an accumulation of its targets — as they are no longer targeted for degradation — we focused on proteins that were upregulated in the mutant. A comparison of the 69 proteins upregulated in the atrogin-1–/– mutant with those shown to also have increased expression (56 proteins) or reduced turnover (137 proteins) in the atrogin-1–KO cardiomyocytes (12) revealed an overlap of 7 proteins (Table 1). Of these 7 proteins, the most upregulated was BiP (also known as GRP-78 or heat shock 70 kDa protein 5 [HSPA5]), a member of the HSP70 family of proteins localized primarily to the ER, where it regulates multiple processes, including activation of the unfolded protein response (UPR) following accumulation of unfolded or misfolded proteins, protein transport, cell survival and apoptosis, calcium homeostasis, and ER-mitochondrial calcium crosstalk, which subsequently regulates mitochondrial function (16, 17) (reviewed in ref. 18). Importantly, prolonged ER stress and chronic upregulation of BiP results in apoptosis, which is a characteristic feature of the atrogin-1 mutant, and atrogin-1–deficient cardiomyocytes (12). BiP accumulation is therefore a prime candidate that could explain the manifestation of atrogin-1 mutant phenotype, and as such all subsequent analyses were focused on defining BiP’s role in maintaining muscle homeostasis.

Atrogin-1 mutants display increased levels of BiP, which is sufficient to cause muscle fiber detachment. (A) Volcano plot highlighting differentially regulated proteins in atrogin-1–/– larvae compared with atrogin-1+/+ wild-type larvae – identified from untargeted proteomics. Proteins significantly (q < 0.05) upregulated and downregulated are shown in red and blue, respectively, as determined using an unpaired t test. (B) Representative Western blot images for BiP, and total protein direct blue stain, on whole cell protein lysates obtained from 3 independent biological replicates, each containing multiple atrogin-1+/+ or atrogin-1–/– larvae. (C) Quantification of BiP levels normalized to total protein with atrogin-1–/– larvae displaying a significant reduction compared with atrogin-1+/+, as determined using an unpaired t test. Data are shown as mean ± SD. (D–F) 6 dpf tunicamycin- (Tm-) or thapsigargin-treated (Tg-treated) larvae display muscle fiber detachment following incubation in methyl cellulose. (G) The percentage of affected larvae, with Tm or Tg treatment resulting in a significant increase in the proportion of fish displaying the muscle fiber detachment, as determined using a χ2 test. (H and I) Confocal images of F-actin–stained, methyl cellulose–treated, 6 dpf atrogin-1–/– mutants on the Tg(actc1b:KalTA4;cryaa:GFPpc54Tg) only [labeled as Control (KaltA4)] or Tg(actc1b:KalTA4;cryaa:GFPpc54Tg) and Tg(4XUAS:NLSCas9;cmlc2:RFP gl37Tg) (labeled as BiP KO) background. While control atrogin-1–/– mutants display fiber detachment, atrogin-1–/– mutants with BiP deficiency specifically in the muscle show normal muscle structure. (J) The percentage of affected atrogin-1–/– control larvae and BiP-KO larvae, with the latter having a significant decrease in the proportion of fish displaying the muscle fiber detachment, as determined using Fisher’s exact test. *P < 0.05, **P < 0.01. All experiments performed in triplicate with the total number of fish examined in each replicate being documented in Supplemental Table 2. Scale bar: 200 μm.

Proteins upregulated in atrogin-1–deficient zebrafish that have increased expression/reduced turnover in the atrogin-1–KO cardiomyocytes

To confirm that BiP is indeed upregulated in the atrogin-1 mutant, we performed Western blotting for BiP on whole cell protein lysate. In line with our untargeted proteomics data, atrogin-1–/– larvae displayed a significant increase in BiP compared with atrogin-1+/+ larvae (Figure 3, B and C). One possible explanation for the increased levels of BiP is increased transcription. To test this possibility, we performed qRT-PCR for the UPR genes bip, chop, atf6, and atf4. Our results indicate no difference in the expression of these genes between atrogin-1+/+ wild-type and atrogin-1–/– larvae (Supplemental Figure 2A). Therefore, the increased levels of BiP seen in the mutant were not due to increased transcription. We next wished to determine if BiP is a direct target of atrogin-1, which could explain the increased levels of BiP in the atrogin-1 mutant. To this end, we conducted coimmunoprecipitation experiments in HEK293T cells by cotransfecting with plasmids encoding GFP (control), Myc-tagged zebrafish atrogin-1, and/or HA-tagged zebrafish BiP and, subsequently, performing a pull-down assay using anti-Myc–coated beads. While Myc-atrogin-1 was enriched in the Myc-atrogin-1 and in the Myc-atrogin-1– and BiP-HA–transfected cells, indicating successful pull down, no HA-tagged BiP was detected in any of the immunoprecipitated lysates (Supplemental Figure 2B). These results indicate that BiP is not a direct target of atrogin-1. Our results suggest that BiP levels are indirectly regulated by atrogin-1 and the increased abundance in the atrogin-1 mutant is likely a secondary consequence of atrogin-1 deficiency.

Accumulation of BiP results in muscle detachment following the loss of atrogin-1. Having confirmed that BiP is upregulated in atrogin-1–deficient larvae, we next wanted to determine if its increased level was responsible for the loss in muscle integrity observed in the atrogin-1–/– mutant. To address this, we treated 3 dpf larvae with the well-characterized ER stress inducers tunicamycin (Tm) or thapsigargin (Tg) for 3 days, which not only induced bip expression, but also expression of the UPR genes chop, atf6, and atf4 (Supplemental Figure 2C). Examination of muscle structure in Tm- or Tg-treated fish revealed a significant increase in the proportion of muscle fibers displaying muscle fiber detachment following incubation in methyl cellulose (Figure 3, D–G), consistent with the morphology seen in the atrogin-1–/– mutants. Therefore, chronic ER activation of ER stress can explain the characteristic phenotype evident in the atrogin-1–/– mutant.

To more explicitly implicate BiP accumulation as the mechanism responsible for the muscle fiber detachment seen in the atrogin-1–/– mutant, we made use of the compound HM03, which has been shown to selectively inhibit BiP activity by binding to its substrate binding domain (19). We treated 3 dpf atrogin-1–/– mutants with the BiP inhibitor HM03 or DMSO for 3 days, changing the chemical each day thereafter, and at 6 dpf we examined muscle integrity. Although the rescue was not complete, pharmacological inhibition of BiP resulted in a reduction in the number of atrogin-1–/– mutants displaying fiber disintegration (Supplemental Figure 2, D–F), supporting the role of BiP accumulation in driving the myopathic phenotype seen following atrogin-1 deficiency.

As an alternative approach, we attempted to generate a BiP mutant line using CRISPR/Cas9 genome editing. However, despite using low levels of BiP targeting guide RNAs, BiP crispant larvae displayed striking phenotypes, including edema in the brain and heart (Supplemental Figure 3, A and B) that were lethal and thus limited our ability to recover BiP germline mutants. Indeed, mice that are completely deficient in BiP display peri-implantation lethality (20). To overcome this issue, we developed a muscle-specific BiP-KO strategy to examine its ability to rescue the atrogin-1–/– mutant phenotype. Briefly, this involves the use of two transgenic lines on the atrogin-1–/– mutant background: a muscle-specific KalTA4 expressing line (Tg(actc1b:KalTA4;cryaa:GFPpc54Tg) crossed to a UAS-driven Cas9 line (Tg(4XUAS:NLSCas9;cmlc2:RFP gl37Tg) (Supplemental Figure 3C). Into this line, we injected two BiP-targeting guide RNAs at the 1-cell stage, which is predicted to result in muscle-specific mutagenesis of BiP (BiP KO) and subsequent loss in expression. To confirm the loss of BiP expression, we stained 3 dpf BiP-KO fish with antibody against myosin and BiP. While 89% of control fish, which were injected with the dual BiP gRNAs but lack KalTA4, displayed striated ER-like BiP localization, only 27% of BiP-KO fish showed clear striations, with the remaining 73% lacking this staining pattern (Supplemental Figure 3, D–F), confirming the loss of BiP expression. To further support this, examination of BiP levels using Western blot on whole cell lysates demonstrated a significant reduction of BiP in BiP-KO fish compared with control fish (Supplemental Figure 3, G and H). Collectively, these results confirm that our tissue-specific approach results in a reduction in BiP expression in the muscle.

Having confirmed the efficiency of our tissue-specific KO approach, we examined the muscle morphology in 6 dpf BiP-KO, atrogin-1–/– mutant larvae. Remarkably, muscle-specific KO of BiP resulted in a striking rescue of the fiber integrity defects of atrogin-1–/– mutant larvae (Figure 3, H–J). While 56% of atrogin-1–/– mutants displayed detached muscle fibers, this was significantly reduced to 21% following muscle-specific loss of BiP in atrogin-1–/– mutants (Figure 3J). A potential explanation for this rescue is that muscle-specific loss of BiP results in reduced muscle contraction, thus preventing fiber detachment. To exclude this possibility, we examined locomotor function, using the Zebrabox assay — which examines the average distance, time, and, thus, speed a fish moves over a 10-minute period — of 6 dpf, BiP-KO larvae. As shown in Supplemental Figure 3I, the average speed traveled by BiP-KO larvae is indistinguishable from that of control larvae, highlighting that muscle-specific loss of BiP does not affect motor performance. Taken together, these results highlight a role of atrogin-1–mediated BiP regulation in the maintenance of muscle homeostasis.

Systems proteomics reveals impaired mitochondrial dynamics as the mechanism of muscle fiber detachment in atrogin-1–deficient fish. While our results support a model in which BiP accumulation results in fiber detachment in atrogin-1 mutants, we wished to determine how this was regulated at a cellular level. To this end, we reexamined our atrogin-1 mutant proteomics data set to identify any potential pathways that may be dysregulated in the mutant. Enrichment analyses on all differentially regulated proteins in atrogin-1–/– larvae revealed a significant overrepresentation of proteins of the oxidative phosphorylation (OXPHOS) pathway (Figure 4A), which is responsible for the production of ATP in the mitochondria. Further examination of proteins within this Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway revealed that except for atp6v1ab and atp5f1b, all other proteins (ndufb6, ndufa10, ndufs3, cox4i1, cox5aa, atp5pb, atp6v1e1b, and atp5fa1) were downregulated in atrogin-1–/– larvae compared with atrogin-1+/+ larvae (Figure 4B). One possible explanation for this is the reduced transcription of each of these complexes. However, our qRT-PCR analyses revealed small, nonsignificant increases in expression of each of these genes, highlighting that reduced transcription is not responsible for the reduction in OXPHOS abundance observed in atrogin-1–deficient larvae (Supplemental Figure 4A). An alternative explanation for the changes in OXPHOS levels is a change in mitochondrial fission and fusion rates and the subsequent reduction in mitochondria number. To assess mitochondrial fusion and fission, we examined the expression of mitochondrial fission genes drp1 and fis1 and fusion genes mfn1, mfn2, and opa1. With the exception of opa1, all genes examined were significantly downregulated in the atrogin-1 mutant, suggestive of altered mitochondrial dynamics (Supplemental Figure 4, B and C). We also examined total mitochondrial content, using Western blot for VDAC1 on whole cell protein lysate, and, in line with our hypothesis, atrogin-1–/– larvae displayed a significant reduction in VDAC1 levels compared with atrogin-1+/+ larvae, indicating a reduction in mitochondrial content (Figure 4, C and D).

Atrogin-1 deficiency also results in altered mitochondrial dynamics. (A) Overrepresentation analyses on all differentially regulated proteins in atrogin-1–/– larvae revealed a significant enrichment of multiple Kyoto Encyclopedia of Genes and Genomes pathway terms, with oxidative phosphorylation (OXPHOS) being the most significant. (B) Heatmap of the relative abundance of OXPHOS proteins ndufb6, ndufa10, ndufs3, cox4i1, cox5aa, atp5pb, atp6v1e1b, atp6v1ab, atp5fa1, and atp5f1b in atrogin-1+/+ and atrogin-1–/– larvae. (C) Representative Western blot images for VDAC1, and total protein direct blue stain, on whole cell protein lysates obtained from 3 independent biological replicates, each containing multiple atrogin-1+/+ or atrogin-1–/– larvae. (D) Quantification of VDAC1 levels normalized to total protein, with atrogin-1–/p4 larvae displaying a significant reduction compared with atrogin-1+/+ larvae, as determined using an unpaired t test. Data are shown as mean ± SD. (E–H) Live images of 6 dpf methyl cellulose–treated atrogin-1+/+ (E and F) and atrogin-1–/– (G and H) larvae showing mosaic expression of actc1b:mitoGFP labeling the mitochondria in green. While atrogin-1+/+ larvae display small mitochondria some of which form an intricate network, mitochondria in atrogin-1–/– larvae are large and rounded. Scale bar: 100 μm (left); 20 μm (right). F and H are zoomed in views of E and G, respectively. (I) The proportion of muscle fibers displaying altered mitochondrial morphology in methyl cellulose–treated atrogin-1+/+ or atrogin-1–/– larvae, as per Fisher’s exact test. ****P < 0.0001. (J–M) Electron micrographs of the muscle in 6 dpf methyl cellulose–treated atrogin-1+/+ and atrogin-1–/– larvae. While atrogin-1+/+ larvae display normal sarcomeric and mitochondrial structure (J and K), atrogin-1–/– mutants (L and M) display fiber disintegration, evident by the disorganized arrangement of sarcomeres (arrow), and abnormal mitochondria with large and swollen matrices (arrowheads). K and M are zoomed in views of J and L, respectively. (N and O) 3 dpf methyl cellulose–treated atrogin-1–/– larvae show a significant reduction in basal (N) and maximum respiration (O) compared with atrogin-1+/+ larvae, as determined using an unpaired t test. Data are shown as mean ± SD.*P < 0.05; ****P< 0.0001. All experiments performed in triplicate, with the total number of fish examined in each replicate being documented in Supplemental Table 2.

Having identified altered mitochondrial biology in the atrogin-1 mutant, we wished to further characterize their mitochondrial structure and function. Mitochondrial morphology was examined by expressing a mito-GFP construct (generated by fusing the mitochondrial targeting sequence of Cox8a to GFP) specifically in the muscle of methyl cellulose–treated atrogin-1+/+ wild-type and atrogin-1–/– mutant larvae. Live imaging at 6 dpf revealed that while the majority of muscle fibers in methyl cellulose–treated atrogin-1+/+ fish display small mitochondria, some of which form an intricate network (Figure 4, E, F, and I), a significant proportion of muscle fibers in methyl cellulose–treated atrogin-1–/– mutant larvae displayed large and rounded mitochondria (Figure 4, G–I). Consistent with this, electron microscopy revealed that 6 dpf methyl cellulose–treated atrogin-1–/– mutants displayed fiber disintegration, evident by the disorganized arrangement of sarcomeres and abnormal mitochondria with large and swollen matrices (Figure 4, L, M, and O), with methyl cellulose–treated atrogin-1+/+ larvae displaying normal sarcomeric and mitochondrial structure (Figure 4, J, K, and N). Finally, oxygen consumption rates, a readout of mitochondrial function, were also examined in the atrogin-1–deficient larvae. We report a significant reduction in both basal (Figure 4N) and maximum respiration (Figure 4O) in 3 dpf methyl cellulose–treated atrogin-1–/– mutant larvae, indicating an alteration in mitochondrial function. Collectively, these results highlight that loss of atrogin-1 results in a reduction in mitochondria number and an impairment in mitochondrial structure and function.

We next wished to determine if the mitochondrial alterations observed could explain the muscle fiber detachment phenotype seen in atrogin-1–/– mutant larvae. To this end, we treated 3 dpf wild-type larvae with rotenone, a complex 1 inhibitor, for 3 days and examined the muscle using an antibody against F-actin. Remarkably, while DMSO treatment had no effect on muscle integrity (Supplemental Figure 4D), chronic inhibition of mitochondrial function resulted in muscle fiber detachment (Supplemental Figure 4, E and F), identical to the phenotype seen in atrogin-1–/– mutants. These results highlight that impaired mitochondrial dynamics is sufficient to cause muscle fiber detachment.

BiP accumulation is responsible for impaired mitochondrial biology. To determine if the mitochondrial phenotypes seen in the atrogin-1 mutant are caused by BiP accumulation, we treated 3 dpf larvae expressing the mito-GFP transgene with Tm or Tg for 3 days and examined mitochondrial morphology. Our analyses revealed that chronic treatment with Tm or Tg resulted in a significant increase in the proportion of muscle fibers displaying large and rounded mitochondria compared with those in F-actin DMSO-treated animals (Figure 5, A–G), consistent with the morphology seen in the atrogin-1–/– mutants. To more explicitly implicate BiP accumulation as the mechanism responsible for the mitochondrial phenotypes seen in the atrogin-1 mutant, we generated a construct to enable the muscle-specific overexpression of fluorescently tagged, full-length mouse BiP. To confirm that the fluorescently tagged form of BiP localized correctly to the ER, we stained BiP-mCherry–expressing fish with an anti-mCherry antibody and with an antibody against Ryr1, which is known to localize within the t-tubule. Using super resolution imaging, BiP-mCherry was found to localize to the terminal cristae of the sarcoplasmic reticulum (SR), a structure directly adjacent to the T-tubules, and more generally within the SR network (Supplemental Figure 5, A–C). Having confirmed that fluorescent tagging of BiP does not affect its localization, we coinjected the BiP-mCherry construct (or mCherry alone) along with the mitochondria labeling GFP plasmid to examine the effect of BiP overexpression on mitochondrial structure. Remarkably, while mCherry-expressing muscle cells displayed small, intricate mitochondrial networks (Figure 4I and Figure 5, H–K), BiP-mCherry–expressing fibers had predominantly large and rounded mitochondria that phenocopied the atrogin1 loss-of-function phenotype (Figure 5, L–O and P). This demonstrates that BiP upregulation alone is sufficient to cause the abnormal mitochondrial structure observed in the atrogin-1–deficient fish.

BiP accumulation is also responsible for the impaired mitochondrial dynamics. Live images of 6 dpf DMSO- (A and B), tunicamycin- (Tm-) (C and D), or thapsigargin-treated (Tg-treated) (E and F) larvae showing mosaic expression of actc1b:mitoGFP labeling the mitochondria in green. While DMSO-treated larvae display small mitochondria, some of which form an intricate network, mitochondria in Tm- and Tg-treated larvae were large and rounded, as determined using a χ2 test. B, D, and F (scale bar: 15 μm) are zoomed in views of A, C, and E (scale bar: 100 μm), respectively. (G) The proportion of muscle fibers displaying altered mitochondrial morphology following DMSO, Tm, or Tg treatment, determined using a χ2 test. (H–O) Live images of 6 dpf larvae coexpressing Mito-GFP with mCherry (H–K) or BiP-mCherry (L–O). K and O are zoomed in views of J and N, respectively. Scale bar: 150 μm (first, second, and third columns); 15 μm (last column). (P) The proportion of muscle fibers displaying altered mitochondrial morphology comparing mCherry overexpression with BiP overexpression, as per Fisher’s exact test. (Q and R) Live images of 6 dpf atrogin-1–/– mutant larvae on the Tg(actc1b:KalTA4;cryaa:GFPpc54Tg) only (labeled as Control (KaltA4)) or Tg(actc1b:KalTA4;cryaa:GFPpc54Tg) and Tg(4XUAS:NLSCas9;cmlc2:RFP gl37Tg) (labeled as BiP KO) background, showing mosaic expression of actc1b:mitoGFP labeling the mitochondria in green. While control atrogin-1–/– mutants display large and rounded mitochondria, atrogin-1–/– mutants with BiP deficiency specifically in the muscle have small mitochondria that form an intricate network, as determined using a χ2 test. Scale bar: 100 μm. (S) The proportion of muscle fibers displaying altered mitochondrial morphology in control and BiP-KO atrogin-1–/– mutants, as per Fisher’s exact test. *P < 0.05, ***P < 0.001. All experiments performed in triplicate with the total number of fish examined in each replicate being documented in Supplemental Table 2.

As a final approach, we used our muscle-specific BiP-KO system to examine if loss of BiP is sufficient to rescue the mitochondrial phenotype seen in the atrogin-1 mutant. Indeed, while 56% of muscle fibers in atrogin-1–/– mutant larvae contained large and rounded mitochondria, this was significantly reduced to 22% in BiP muscle–specific KO, atrogin-1–/– mutants, highlighting a rescue in the mitochondrial phenotypes (Figure 5, Q–S).

We next wished to determine if BiP overexpression altered mitochondrial dynamics as seen in the atrogin-1–/– mutant. To this end, wild-type embryos were injected with mCherry or BiP-mCherry RNA, and at 2 dpf qRT-PCR for UPR and mitochondrial fission and fusion genes was performed. Consistent with the injection of RNA, BiP-mCherry–injected fish displayed increased levels of BiP (Supplemental Figure 5D). Interestingly, we also observed a significant increase in the expression of atf4, with chop showing a small nonsignificant increase, suggesting that BiP overexpression may have triggered ER stress (Supplemental Figure 5D). Furthermore, in line with the reduced expression of mitochondrial fission and fusion genes observed in atrogin-1–/– mutants, BiP-mCherry RNA–injected larvae displayed a significant reduction in drp1, fis1, mfn2, and opa1, with mfn1 showing small but nonsignificant reduction (Supplemental Figure 5, E and F), highlighting a role of BiP in regulation mitochondrial dynamics.

Taken together, our results demonstrate that the loss of atrogin-1 results in the accumulation of BiP, which results in mitochondrial dysfunction and a subsequent detachment and apoptosis of muscle cells.

Atrogin-1 is a modifier in, and contributes to, the pathogenesis of DMD. The muscle fiber detachment observed in the atrogin-1–/– mutant is strikingly similar to the phenotype seen in zebrafish models of DMD, caused by a mutation in dystrophin (21). Given that our findings have implicated BiP accumulation in the presentation of the atrogin-1–/– mutant phenotype, we hypothesized that a similar mechanism may be contributing to the pathogenesis of DMD. Indeed, BiP upregulation has been reported in several mammalian models of DMD (22, 23), but whether a similar response occurs in zebrafish is not known. To determine this, we performed Western blotting for BiP on whole cells lysates of 2 dpf and 4 dpf dmd+/+ wild-type and dmd–/– mutant larvae. Consistent with the mammalian models, we observed a significant increase in BiP expression in 4 dpf dmd–/– mutant compared with the dmd+/+ wild-type larvae, although no change was observed at 2 dpf (Figure 6, A and B). To determine if loss of dystrophin results in increased ER stress and activation of the UPR, we performed qRT-PCR for the UPR genes bip, chop, atf6, and atf4. We observed a significant increase in the expression of bip and atf6, with chop and atf4 showing small but nonsignificant increases (Supplemental Figure 6A). Therefore, as shown in the mdx mouse model, and in skeletal muscle from patients with DMD (22–24), the loss of dystrophin in zebrafish also results in increased abundance of BiP and activation of the UPR.

Atrogin-1 is a modifier of and contributes to the pathogenesis of Duchenne muscular dystrophy. (A) Representative Western blot image for BiP, and total protein direct blue stain, on whole cell protein lysates obtained from 2 dpf and 4 dpf dmd+/+ or dmd–/– larvae. (B) Quantification of BiP levels normalized to total protein with 4 dpf dmd–/– larvae displaying a significant increase compared with dmd+/+ larvae, as determined using a 2-way ANOVA with Šidák’s multiple correction post hoc test. Data are shown as mean ± SD. (C–F) Representative birefringence images of 4 dpf dmd+/+; atrogin-1+/+ wild-type larvae and dmd–/– and atrogin-1–/– single and double mutants. Scale bar: 500 μm. (G) Quantification of normalized birefringence intensity, which is the mean birefringence intensity relative to area, in 4 dpf dmd+/+; atrogin-1+/+ wild-type larvae and dmd–/– and atrogin-1–/– single and double mutants, analyzed using a 1-way ANOVA with Tukey’s multiple correction post hoc test. Data are shown as mean ± SD. (H) Average speed, in mm/s, of dmd+/+; atrogin-1+/+ wild-type larvae and dmd–/– and atrogin-1–/– single and double mutants, as analyzed using as a 1-way ANOVA with Tukey’s multiple correction post hoc test. Data are shown as mean ± SEM for 3–4 biological replicates. (I and J) Live images of 4 dpf larvae expressing GFP RNA or atrogin-1-IRES-GFP RNA. Scale bar: 200 μm. (K–N) Birefringence images of 4 dpf dmd+/+ and dmd–/– larvae injected with GFP RNA or atrogin-1-IRES-GFP RNA. Scale bar: 500 μm. (O) Quantification of normalized birefringence intensity, which is the mean birefringence intensity relative to area, in 4 dpf dmd+/+ or dmd–/– larvae injected with GFP RNA or atrogin-1-IRES-GFP RNA, as analyzed using a 2-way ANOVA with Šidák’s multiple correction post hoc test. Data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All experiments performed in triplicate with the total number of fish examined in each replicate being documented in Supplemental Table 2.

Having confirmed that BiP is upregulated in zebrafish models of DMD, we wished to determine if the atrogin1-BiP axis we have identified contributes to DMD pathology and could be manipulated for potential therapeutic gain. As such, we crossed the dmd–/– mutant with the atrogin-1–/– mutant and examined muscle structure, using birefringence assays and locomotor function in the double mutants. As previously shown, dmd–/– mutants displayed a significant reduction in mean birefringence intensities compared with wild-type larvae (Figure 6, C, D, and G), highlighting a reduction in muscle fiber integrity. The birefringence intensities in atrogin-1–/– mutants on the other hand were indistinguishable from those of wild-type larvae (Figure 6, C, E, and G), consistent with the mild, sporadic phenotypes seen in these mutants. Simultaneous loss of both dystrophin and atrogin-1 resulted in a dramatic additive reduction in birefringence intensity within the myotomes of double mutant larvae, compared not only with wild-type and atrogin-1–/– mutants, but also dmd–/– mutants (Figure 6, C, F, and G). This highlights a potential role of atrogin-1 in modifying the muscle fiber detachment in DMD. We also examined if loss of atrogin-1 affects muscle function in dmd–/– mutants, specifically examining the average speed of larvae over a 10-minute period in a standard zebrabox locomotion assay (25). Similar to the birefringence assays, while dmd–/– mutants have a significant reduction in average speed, it is further reduced following the loss of atrogin-1 (Figure 6H). The exacerbation of the muscle detachment phenotype and reduction in muscle function in dmd–/– mutants following the loss of atrogin-1 demonstrates a role of the latter in DMD pathogenesis.

As further validation of atrogin-1’s role in DMD, we injected atrogin-1-IRES-GFP (or GFP control) RNA in dmd–/– mutants and examined muscle fiber integrity in 4-day-old animals. Expression of RNA was confirmed by the fluorescence of GFP protein in the myotome of injected larvae (Figure 6, I and J). atrogin-1-IRES-GFP mRNA injection significantly ameliorated the reduction in birefringence intensity evident in dmd–/– larvae, although the rescue was not complete — that is, atrogin-1-IRES-GFP–injected dmd–/– mutants still has a significant reduction in birefringence compared with dmd+/+ wild-type larvae injected with same RNA (Figure 6, K–O). This is a surprising finding, because it suggests that in DMD additional dystrophin-independent mechanisms regulated by atrogin-1 may be contributing to disease pathogenesis. Importantly, atrogin-1 overexpression did not have any detrimental effect on muscle integrity, as evident by indistinguishable birefringence intensities between GFP-injected and atrogin-1-IRES-GFP injected wild-type larvae (Figure 6, K–O). These results combined with the data on the dmd–/–; atrogin-1–/– double mutants implicates atrogin-1 in the presentation of DMD pathologies.

BiP inhibition rescues muscle function in DMD. While our results suggest that atrogin-1 may be manipulated for therapeutic gain in DMD, we wanted to examine whether manipulating levels of BiP, which is regulated by atrogin-1, could provide a possible alternative therapeutic strategy to combat DMD. To test this, we treated 3 dpf dmd+/+ wild-type larvae and dmd–/– mutants with the BiP inhibitor HM03, or DMSO control for 3 days, changing the chemical each day thereafter, and at 6 dpf we performed muscle integrity birefringence assays and zebrabox assays. Contrary to our hypothesis, HM03 treatment had no effect on muscle integrity, evident from the indistinguishable birefringence intensities between DMSO-treated and HM03-treated dmd–/– mutants (Figure 7, A–D). We confirmed this result by treating dmd–/– mutants on the Tg(actc1b:Lifeact-GFP);Tg(actc1b:CAAX-mCherry) background, whereby the actin filaments within the muscle fibers were labeled with GFP and membrane and t-tubules with mCherry (Supplemental Figure 6, B–D). Similar to the birefringence assays, HM03-treated dmd–/– mutants displayed similar severities of fiber detachment to DMSO-treated dmd–/– mutants. While these results are surprising, they suggest that rescue of muscle fiber integrity in the dmd–/– following atrogin-1 overexpression likely results from an atrogin-1 target independent of BiP.

BiP inhibition rescues muscle function in DMD. (A–C) Representative birefringence images of 6 dpf DMSO-treated dmd+/+ and dmd–/– larvae and HM03-treated dmd–/– larvae. (D) Quantification of normalized birefringence intensity, which is the mean birefringence intensity relative to area, in 6 dpf DMSO- or HM03-treated dmd+/+ and dmd–/– larvae, as analyzed using a 2-way ANOVA with Šidák’s multiple correction post hoc test. Data are shown as mean ± SD. (E) Average speed, in mm/s, of 6 dpf DMSO- or HM03-treated dmd+/+ and dmd–/– larvae. Data are shown as mean ± SEM for 3–4 biological replicates and were analyzed using a 2-way ANOVA with Šidák’s multiple correction post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All experiments performed in triplicate with the total number of fish examined in each replicate being documented in Supplemental Table 2.

We also examined the effect of HM03 treatment and subsequent BiP inhibition on the muscle function of dmd–/– mutants. Unlike the results for muscle integrity, HM03 treatment significantly improved the average speed of dmd–/– mutants compared with DMSO-treated dmd–/– mutants (Figure 7E). Remarkably, the mean speed of HM03-treated dmd–/– mutants was comparable to that of DMSO-treated dmd+/+ wild-type larvae, indicating that HM03 completely restored muscle function in the mutant fish (Figure 7E). Importantly, treatment of dmd+/+ wild-type larvae with HM03 had no effect on their speed, demonstrating that the improvement in muscle function seen in dmd–/– mutants was specific and not a generalized response.

Together, these results demonstrated that HM03 and the subsequent inhibition of BiP specifically improves muscle function in performance in dmd–/– mutants. Therefore, while the atrogin-1–mediated BiP may not be involved in the loss in fiber integrity seen in DMD, it does contribute to the reduction in muscle function, making this disease axis therapeutically relevant for potentially improving muscle performance in boys with DMD.