The mitophagy receptor BNIP3L/Nix coordinates nuclear calcium signaling to modulate the muscle phenotype

ABSTRACT

Mitochondrial quality control is critical in muscle to ensure contractile and metabolic function. BNIP3L/Nix is a BCL2 member, a mitophagy receptor, and has been implicated in muscle atrophy. Human genome-wide association studies (GWAS) suggest altered BNIP3L expression could predispose to mitochondrial disease. To investigate BNIP3L function, we generated a muscle-specific knockout model. bnip3l knockout mice displayed a ragged-red fiber phenotype, along with accumulation of mitochondria and endo/sarcoplasmic reticulum with altered morphology. Intriguingly, bnip3l knockout mice were more insulin sensitive with a corresponding increase in glycogen-rich muscle fibers. Kinome and gene expression analyses revealed that bnip3l knockout impairs NFAT and MSTN (myostatin) signaling, with alterations in muscle fiber-type and evidence of regeneration. Mechanistic experiments demonstrated that BNIP3L modulates mitophagy, along with reticulophagy leading to altered nuclear calcium signaling. Collectively, these observations identify novel roles for BNIP3L coordinating selective autophagy, oxidative gene expression, and signaling pathways that maintain the muscle phenotype.

Introduction

Mitochondrial quality control involves the fission of dysfunctional mitochondrial fragments, the removal of these fragments by PINK1-PRKN/Parkin and receptor-mediated mitochondrial autophagy (i.e., mitophagy), combined with mitochondrial biogenesis and the fusion of nascent mitochondria with the existing mitochondrial network [1]. Aberrant mitochondrial quality control has been implicated in a number disease states, including muscle related pathologies, such as sarcopenia [2], cachexia [3], mitochondrial myopathies [4], muscular dystrophies [5], and the remodeling associated with obesity and insulin resistance [6].

BNIP3L/Nix is a BCL2 family member involved in the regulation of apoptosis, necrosis, macroautophagy, and mitophagy [7]. In addition, BNIP3L expression has recently been shown to be elevated during aging and has been implicated in starvation-induced muscle atrophy [8,9]. Human GWAS have identified non-coding single nucleotide polymorphisms within the BNIP3L gene that increase the risk of early-onset schizophrenia and cognitive decline, both of which involve mitochondrial pathology, suggesting that reduced BNIP3L expression may result in secondary mitochondrial myopathy [10,11].

BNIP3L is biologically active at both the mitochondria and the endoplasmic reticulum (ER) [7]. As a regulator of autophagy, BNIP3L activates macroautophagy at the ER by disrupting the interaction between BCL2 and BECN1, which has been proposed to take place at ITPR/IP3-receptor [7]. At the mitochondria, BNIP3L acts as a selective autophagy receptor recruiting Atg8-family proteins and the MTOR activator RHEB [12,13], where deletion of both Bnip3L and/or its homolog Bnip3 results in the age-related accumulation of large senescent mitochondria due to impaired mitochondrial pruning [14]. However, this model of BNIP3L function may be incomplete, as BNIP3 can also promote selective reticulophagy [15]. Moreover, during a mitophagy response, we previously demonstrated that BNIP3L promotes ER-dependent calcium signaling to activate the mitochondrial fission regulator DNM1L/DRP1, suggesting that both ER- and mitochondrial-localized BNIP3L contribute to mitophagy [6].

Given the diverse biological roles attributed to BNIP3L, we interrogated BNIP3L function in a novel muscle-specific knockout mouse with significant reduction in BNIP3L expression in both red and white muscle, but not in the heart. This approach allowed us to assess the role of BNIP3L in muscle biology without contaminant cardiovascular defects. Muscle-specific bnip3l knockout mice display a ragged-red fiber phenotype, along with accumulation of mitochondria and sarcoplasmic reticulum (SR) with altered morphology. Interestingly, bnip3l knockout also impaired NFAT and canonical MSTN/GDF8 (myostatin) signaling resulting in alterations in muscle fiber-type composition. In addition, we present evidence of myopathy and regeneration in the absence of BNIP3L. Mechanistic experiments in culture demonstrated that BNIP3L is both necessary and sufficient to regulate mitophagy, but also reticulophagy through a distinct mechanism leading to nuclear calcium signaling. Collectively, these observations indicate that BNIP3L has a biological role beyond autophagy/mitophagy, serving to modulate ER-SR homeostasis and signaling pathways that impact the muscle phenotype.

Results

Muscle-specific bnip3l knockout results in ragged red fibers, with the accumulation of mitochondria and sarcoplasmic reticulum with altered morphology

We used CRISPR-Cas9 technology to create a conditional allele with loxP sites flanking the second exon of the Bnip3l gene. Single-stranded donor DNAs were electroporated along with Cas9 and guide RNAs to insert loxP sites by homology directed repair (Figure 1A). Zygotes were implanted into pseudo-pregnant mice, and offspring were screened for insertion of loxP sites (Figure 1B). Mice containing both loxP sites were bred to homozygosity (Bnip3l^fl/fl) and crossed with human skeletal ACTA1/α-actin-Cre mice (ACTA1-Cre). Cre-positive mice (bnip3l-ACTA1-KO) displayed reduced BNIP3L expression in red, white, and mixed fiber-type muscles, but with no impact in the heart, while Cre expression alone did not influence BNIP3L expression (Figure 1C,D). The degree of knockout in whole gastrocnemius was similar in magnitude to other reports using the ACTA1 promoter [16]. Intriguingly, Gomori trichrome staining in bnip3l-ACTA1-KO mice revealed consistent evidence of ragged red muscle fibers in male mice by 10-weeks of age, which were notably absent in female mice, and mice of all other genotypes (Figure 1E; Figure S1A). As the pathology of ragged red fibers has been associated with mitochondrial myopathy and the accumulation of subsarcolemmal mitochondria, we performed transition electron microscopy, which revealed large mitochondria both beneath the sarcolemma and amongst the myofibrils with altered morphology and cristae structure (Figure 1F; Figure S1B, C). Furthermore, we did not observe a compensatory increase in other mitophagy genes Prkn, Bnip3, or Fundc1, but did observe a trend toward decreased expression (Figure S1D). In addition, we observed the accumulation of SR membranes in close association with mitochondria (Figure S1C). Intriguingly, removal of BNIP3L from muscle had no significant impact on the macroautophagy markers, MAP1LC3/LC3-II and SQSTM1/p62 (Figure S1E). To assess the impact of muscle-specific bnip3l knockout on metabolism, we performed metabolic caging, and observed a reduction in oxygen consumption, without changes in food consumption, body mass, or tibial length (Figure 1G; Figure S1F). In addition, we evaluated oxygen consumption rate (OCR) in soleus muscle explants, which confirmed that BNIP3L knockout reduced muscle oxygen consumption (Figure 1G). bnip3l-ACTA1-KO mice also displayed an increased resting respiratory exchange ratio (RER), suggesting increased reliance on carbohydrate metabolism, and a concurrent increase in blood lactate concentration (Figure 1H). Finally, we observed that bnip3l-ACTA1-KO mice had decreased running distance on a graded exercise treadmill test (Figure 1I). These observations implicate BNIP3L in the regulation of both mitochondria and SR structure, and muscle function.

Figure 1.

Signs of mitochondrial myopathy in muscle-specific bnip3l knockout mice. (A) Schematic illustrating the generation of the floxed BNIP3L gene using CRISPR-Cas9. (B) PCR detection of 5’ and 3’ loxP sites in heterozygous (het) and homozygous (Hom; Bnip3l^fl/fl) mice and presence of the cre transgene. (C) Real-time PCR detection of Bnip3l in Bnip3^fl/fl and knockout (bnip3l-ACTA1-ko) mice. Gastrocnemius/plantaris (gastroc), extensor digitorum longus (EDL), and soleus muscle (n = 4). (D) Representative immunoblots of BNIP3L and Cre by genotype. Wild-type (WT) and ACTA1-cre (Cre). (E) Gomori trichrome staining demonstrating ragged red fibers (red arrows) and quantification by genotypes (P-value determined by Chi-square test) in mice 10–12 weeks of age. (F) Transmission electron microscopy of gastrocnemius muscle fibres in longitudinal section showing morphology of subsarcolemmal (SS) and intermyofibrillar (IMF) mitochondria. Abnormal mitochondrial morphology (black arrows) and sarcoplasmic reticulum membranes (*). (G) Oxygen consumption and area under the curve quantification of Bnip3^fl/fl and bnip3l-ACTA1-ko male mice 10–12 weeks of age (n = 4; left). OCR of soleus muscle ex plants using a seahorse XFe24 analyzer (right). (H) RER calculated from resting metabolic cage data (left) and resting non-fasting blood lactate (right). (I) Maximal distance ran in graded exercise treadmill test (n = 6–10). Data shown as mean with error bars indicating standard error of the mean. *p < 0.05.

Kinome analysis of muscle-specific bnip3l knockout mice reveals increased signaling responses involved in nutrient storage and impaired calcium and TGFB signaling.

Previously, we demonstrated that BNIP3L orchestrates both ER-to-mitochondrial calcium transfer and MTOR signaling during a mitophagy response [6,17]. As ER calcium depletion, ER stress, and reticulophagy often occur concurrently, we hypothesized bnip3l-ACTA1-KO mice may present with defects in calcium signaling. To evaluate this hypothesis in vivo, we performed kinome microarray analysis using the workflow illustrated in Figure 2A, and described previously [18,19]. Pathway overrepresentation and Gene Ontology analyses were performed using InnateDB (Figure 2A; Figure S2A, B), which identified upregulation of pathways involved in lipid and glycogen biosynthesis, and downregulation of pathways involved with TGFB/TGF-β and calcium signaling. Inspection of individual signaling pathways, revealed altered phosphorylation of key proteins involved in the insulin signaling pathway and decreased phosphorylation of PHKA1 (phosphorylase kinase regulatory subunit alpha 1) involved in glycogen breakdown (Figure 2B). To evaluate the physiological significance of altered insulin signaling, we performed insulin tolerance tests in fasted Bnip3l^fl/fl and bnip3l-ACTA1-KO mice, which confirmed enhanced insulin sensitivity in the bnip3l-ACTA1-KO mice (Figure 2C; Figure S2C). In addition, periodic acid – Schiff (PAS) staining of gastrocnemius/plantaris muscle identified an increased number of glycogen-rich muscle fibers in bnip3l-ACTA1-KO mice (Figure 2D, Figure S2C). We also performed kinome microarray analysis in muscle tissue following administration of insulin. InnateDB and Gene Ontology revealed overrepresentation of insulin activated pathways, including RHO-RAC1 which are involved in SLC2A4/GLUT4 translocation, and several other pathways involved in cell growth and metabolism in the bnip3l-ACTA1-KO mice (Figure S2D).

Figure 2.

Deletion of BNIP3L alters multiple cell signaling cascades in muscle. (A) Kinomics analysis workflow and pathway analysis by InnateDB. (B) Volcano plot of individual phospho-residues showing fold change in bnip3l-ACTA1-ko mice over Bnip3l^fl/fl male mice at 10–12 weeks of age. P-values were adjusted using the false discover rate method (FDR)(n = 3). Table of phospho-residues involved in insulin, calcium, and TGFB-SMAD signaling as well as glycogen breakdown. (C) Insulin tolerance test in fasted BNIP3L^fl/fl and BNIP3L-ACTA1-ko male mice (n = 5). (D) Periodic acid Schiff (PAS) staining for glycogen content in gastrocnemius/plantaris muscle of male mice. E) Heat map generated from myogenesis and myopathy qPCR array. Green, red, and black colours indicate down-, up-regulation, or no change respectively (n = 4). (F) Gene expression of mb and mstn in gastrocnemius/plantaris muscle of male mice (n = 4). (G) Immunoblot of total and phospho-SMAD2 expression in extracts from gastrocnemius/plantaris muscle (n = 3). (H) Gene expression of Myh2, Myh4, Tnnt1, mb and mstn in soleus muscle extracts from male mice (n = 4). (I) Immunoblot of MB expression in soleus muscle extracts (n = 3). Data shown as mean with error bars indicating standard error of the mean. *p < 0.05.

Our kinome analysis also identified reduced phosphorylation of HDAC4 at Ser-632, and increased phosphorylation of NFATC3 at Ser186 (Figure 2B). As these residues have been previously implicated in calcium-dependent gene expression [20], we performed multiple PCR-based arrays targeting myogenesis and myopathy genes comparing male and female mice and red and white muscle groups in Bnip3l^fl/fl and bnip3l-ACTA1-KO mice (Figure 2E; Table S1). Interestingly, female knockout mice displayed a greater number of genes with reduced expression compared to male knockout mice, despite the absence of ragged red fibers. However, we observed that female mice also had reduced expression of PPARGC1A/PGC-1α and increased MYOG/myogenin expression, which may protect against this phenotype (Figure S2E). We also examined the expression of genes involved in the unfolded protein response/UPR and did not observe significant changes, although there was a trend toward reduced HSP90B1/GRP94 expression in bnip3l-ACTA1-KO mice (Figure S2F). Examination of two known NFAT-target genes, Mb (myoglobin) and Mstn (myostatin) [20,21], in gastrocnemius/plantaris and soleus muscle of male mice revealed that Mstn expression was reduced in gastrocnemius/plantaris, while Mb expression was reduced in soleus muscle (Figure 2F, -H). This suggests some degree of muscle-group or fiber-type specific effect of BNIP3L knockout, which is supported by the data in Table S1 comparing gene expression in different muscle groups. As MSTN is a member of the TGFB family that signals through SMAD2-SMAD3, we confirmed that bnip3l knockout reduced p-SMAD2 in gastrocnemius/plantaris (Figure 2G; Figure S3A), consistent with the kinome data (Figure 2B). Next, we examined the expression of other muscle oxidative genes in soleus muscle and observed reductions in Myh2 (myosin heavy chain 2; type IIa, fast oxidative) and Tnnt1 (troponin T1, slow skeletal type) (Figure 2H). Moreover, we observed a marked increase in Myh4 (myosin heavy chain 4; type IIb, fast glycolytic) expression. Finally, we confirmed the decreased expression of MB in the soleus of bnip3l-ACTA1-KO mice by western blot (Figure 2I; Figure S3B). Collectively, these observations suggest that BNIP3L knockout in muscle enhances insulin signaling to promote nutrient storage and impairs calcium-dependent gene expression and downstream myostatin activity.

BNIP3L is both necessary and sufficient to activate nuclear calcium signaling and gene expression

Using two independent C2C12 models of mitochondrial biogenesis, differentiation and electrical pacing, we observed that BNIP3L expression increased in parallel with PPARGC1A (Figure 3A, B). These observations suggest that quality control pathways are activated concurrently with mitochondrial biogenesis. To evaluate the role of BNIP3L in this hypothesis, we utilized the mitophagy biosensor mito-pHred [6], in combination with a validated lentiviral shRNA targeting Bnip3l (Lenti-shBnip3l; Figure S3C) [6] and electrical pacing. Shown in Figure 3C, electrical pacing (Stim) increased mito-pHred fluorescence, which was prevented by BNIP3L knockdown. As bnip3l-ACTA1-KO mice display evidence of ER/SR membrane accumulation, we speculated that BNIP3L may also be involved in selective removal of endoplasmic reticulum by reticulophagy. Thus, we transfected C2C12 cells with ER-targeted mCherry (ER-mCherry) and GFP-LC3 to monitor ATG8 recruitment to the ER in the presence of BNIP3L, which represents an early event in reticulophagy. We observed that BNIP3L significantly increases the colocalization of GFP-LC3 with ER-mCherry, but this colocalization was blocked by the ITPR antagonist 2APB (Figure 3D,E). In addition, we monitored ER calcium depletion in a parallel series of experiments using the ER-targeted calcium biosensor ER-GECO, which identified that ER calcium levels and reticulophagy have an inverse relationship (Figure 3D,E). BNIP3L also induced a similar degree of colocalization between ER-Emerald and LysoTracker, a late indicator of the autophagy process (Figure 3D). As an additional control, cells were treated with the lysosomal inhibitor bafilomycin A₁, which also increased LC3 colocalization with ER-mCherry (Figure S3E). As phosphorylation of BNIP3L at Ser35 has been previously shown to enhance LC3/Atg8 recruitment to mitochondria during mitophagy [12,22], we evaluated the effect of non-phosphorylable and phospho-mimetic mutations of BNIP3L at this residue (i.e., S35A and S35D, respectively). Intriguingly, wild-type BNIP3L, S35A, and S35D all had a similar impact on both GFP-LC3 recruitment to the ER and ER calcium depletion (Figure 3F,G); however, when we evaluated colocalization between GFP-LC3 and mito-mCherry, BNIP3L^S35D induced a greater degree of LC3 recruitment to the mitochondrial than BNIP3L^S35A (Figure S3F). Collectively, these data demonstrate that ER calcium depletion is a key mechanism by which BNIP3L induces reticulophagy, and that phosphorylation at Ser35 is more important in recruitment of LC3 to mitochondria.

Figure 3.

BNIP3L is both necessary and sufficient to activate nuclear calcium signaling and gene expression. (A-B) Immunoblots of BNIP3L and PPARGCA1 and quantification of protein levels in differentiating C2C12 myotubes (a) and in electrically paced (stim) C2C12 myotubes (b)(n = 3). (C) Representative images and quantification of the mitophagy biosensor mito-pHred in myotubes with or without electrical pacing and knockdown of Bnip3l with lentriviral RNA interference (lenti-shBnip3l). (D-E) C2C12 cells transfected with ER-mCherry, GFP-LC3, and BNIP3L, and treated with 2APB (2 µM), as indicated (above) or ER-GECO (below). Quantification performed by Pearson’s colocalization coefficient or mean fluorescence. (F-G) C2C12 cells were transfected as in (d) with BNIP3L, BNIP3L^S35A, or BNIP3L^S35D, and quantified as in (E). (H) Representative images and quantification of the nuclear calcium biosensor (NLS-GECO) in paced myotubes with and without lenti-shBnip3l. (I)=NLS-GECO in C2C12 cells transfected with empty vector or BNIP3L. (J) Representative images and quantification of sub-cellular distribution of NFAT-YFP in C2C12 cells expressing BNIP3L or empty vector. (K) Immunoblot and quantification of MB expression in C2C12 cells transfected with MYC-BNIP3L (n = 3). Endo-BNIP3L indicates endogenous BNIP3L. (L-N) C2C12 cells were transfected with an shRNA targeting Bnip3l (shBnip3l) or scrambled control (scramble) and allowed to differentiate for 6-days. RNA expression of Mb (l) Myh2 (m) and Myh4 (n) in C2C12 myotubes. (O) Immunofluorescence detection of specific muscle fiber-types in the central region of gastrocnemius/plantaris muscle of male mice 10–12 weeks of age. Type IIa (green; MYH2), type IIb (red; MYH4), and type I (blue; MYH7). **the percentage of type IIx fibers was estimated by non-staining regions. Floxed=Bnip3l^fl/fl; KO=bnip3l-ACTA1-ko. Data shown as mean with error bars indicating standard error of the mean. *p < 0.05.

Next, we evaluated the effect of BNIP3L on down-stream calcium signaling. We transfected C2C12 cells with the nuclear-targeted calcium biosensor, NLS-GECO [23], followed by differentiation and electrical pacing. Intriguingly, pacing increased nuclear calcium, but was attenuated by knockdown with Lenti-shBnip3l (Figure 3H). In addition, transfection of BNIP3L into C2C12 cells increased NLS-GECO fluorescence, which was inhibited by the ITPR antagonist 2ABP (Figure 3I; Figure S3G). Moreover, cell fractionation studies revealed that BNIP3L preferentially accumulates at the ER/SR following pacing (Figure S3H). These observations identify a dual role for BNIP3L, regulating selective autophagy and nuclear calcium signaling. To evaluate if increased nuclear calcium induced by BNIP3L translates into changes in gene expression, we first transfected NFAT-YFP with and without BNIP3L in C2C12 cells and observed that BNIP3L expression was sufficient to increase the nuclear accumulation of NFAT-YFP (Figure 3J). BNIP3L expression also increased the endogenous expression of the NFAT-target gene MB (Figure 3K). Next, we returned to the C2C12 differentiation model, but knocked-down BNIP3L using a plasmid-based shRNA prior to differentiation (Figure S3I) [6,24], and evaluated Mb expression, and the myosin heavy chain genes that were altered in vivo. Consistent, with the data in Figure 2H, shBnip3l decreased Mb and Myh2 expression, but increased Myh4 expression (Figure 3L–N). As calcium and NFAT signaling have been previously shown to alter the muscle fiber-type composition, promoting oxidative fiber-types [20], we evaluated fiber-type composition in Bnip3l^fl/fl and bnip3l-ACTA1-KO mice. Shown in Figure 3O, bnip3l-ACTA1-KO mice have a reduced number of oxidative type IIa fibers and an increase in glycolytic type IIb fibers in the gastrocnemius/plantaris muscle groups, without changes in type I or IIx fibers. Similar results were observed in soleus muscle, where bnip3l-ACTA1-KO mice have decreased numbers of oxidative fibers (Type I) and increased numbers of glycolytic fibers (Type IIb and IIx; Figure S3J).

Muscle-specific bnip3l knockout results in a myopathy with central nuclei and regeneration, without an increase in muscle fibrosis

Our Gene Ontology enrichment analysis in bnip3l-ACTA1-KO mice identified biological responses involved with proteolysis, cell growth, and phospholipid and steroid biosynthesis, all of which have been implicated in muscle repair and regeneration (Figure S2B). Consistent with this, muscle cross-sections from bnip3l-ACTA1-KO display central nuclei (Figure 4A), and we observed increased expression of embryonic myosin heavy chain (Myh3; Figure 2B). In addition, examination of gene expression identified increased Pax7 expression, with a corresponding increase in the satellite cell mitogen Tnf (Figure 4C) [25], which can operate in a reciprocal manner to MSTN during muscle regeneration. Interestingly, we also observed decreased expression of Dmd (dystrophin) and increased expression of Casp3 (caspase 3), known for its role in apoptosis but also as a regulator of PAX7 function (Figure 4C) [26]. The expression of NFE2L2/NRF2, a transcription factor involved in the regulation of antioxidant gene expression and modulation of satellite cell function, was also increased in bnip3l-ACTA1-KO mice (Figure 4D) [27]. We also examined signaling pathways identified through our kinome analysis and observed tyrosine phosphorylation of several receptors implicated in muscle regeneration, including INAR1, EPHB2, PDGFRA, and FGFR1 (Figure 4E). Our kinome analysis also identified phosphorylation of MAPK14/p38α at the inhibitory Thr123 residue, consistent with an established role of MAPK14/p38α in the repression of PAX7 expression [28,29]. Finally, we observed an increased number of PAX7-positive satellite cells within gastrocnemius/plantaris muscle in bnip3l-ACTA1-KO mice, without alterations in muscle fibrosis (Figure 4F,G). Collectively, these observations demonstrate that muscle-specific BNIP3L knockout results in a myopathy with compensatory regeneration.

Figure 4.

Muscle-specific BNIP3L knockout results in a myopathy with central nuclei and regeneration. (A) H&E staining of gastrocnemius/plantaris muscle from Bnip3l^fl/fl and bnip3l-ACTA1-ko male mice showing presence of centrally located nuclei (black arrows), and quantification in male mice 10–12 weeks of age. (B) Gene expression of embryonic Myh3 (myosin heavy chain 3), an indicator of regeneration (n = 4). (C) Gene expression of markers of muscle satellite cell activation in male mice. Tnf/TNFα, dmd (dystrophin), Casp3 (caspase 3) (n = 4). (D) Immunoblot and quantification of NFE2L2 protein (n = 3). (E) Kinome analysis of phospho-residues involved in muscle regeneration and/or satellite cell activation (n = 3). (F) Immunofluorescence and quantification of PAX7-positive cells (green with green arrow), LAM (laminin; red) in gastrocnemius/plantaris muscle (nuclei counterstained blue). Central nuclei identified by white arrows (n = 3). (G) Picrosirius red staining of gastrocnemius/plantaris muscle demonstrating no overt fibrosis in bnip3l-ACTA1-ko male mice 10–12 weeks of age (connective tissue stains red). Data shown as mean with error bars indicating standard error of the mean. *p < 0.05.

Discussion

Emerging GWAS evidence suggests that polymorphisms within the BNIP3L gene are associated with mitochondrial related pathology [10,11]. Although the precise impact of these polymorphisms is not currently known, BNIP3L has been experimentally implicated in muscle atrophy, aging, lipotoxicity, and cardiac remodeling [6,8,9,24,30], where mitochondrial dysfunction has been shown to contribute to the pathogenesis of these disorders. In this report, we interrogate BNIP3L function in muscle using a novel mouse knockout model, which present with evidence of compensated mitochondrial myopathy. Moreover, this mouse model has uncovered novel aspects of BNIP3L function, that could be of translational importance as skeletal muscle is often used as diagnostic proxy in the evaluation of neuropathology. The skeletal muscle phenotype has been described in terms of distinct muscle fiber-types that are commonly identified by the expression of a specific myosin heavy chain gene [20]. Our data identifies BNIP3L, a known regulator of mitophagy [6,14], as a modulator of muscle fiber-type composition by regulating several signaling pathways that target gene expression. Our observations suggest BNIP3L-induced mitophagy parallels mitochondrial biogenesis to mechanistically couple mitochondrial turn-over with oxidative gene expression, ultimately modifying the muscle phenotype.

Muscle-specific bnip3l knockout also results in the accumulation of ER/SR membranes, while mechanistic experiments demonstrate that BNIP3L modulates reticulophagy and calcium homeostasis. Interestingly, we observed that ER calcium depletion is an important mechanism triggering reticulophagy, which is distinct from BNIP3L-induced mitophagy which depends on direct Atg8/LC3 recruitment [12,22,31]. We also observed sex-specific differences in the present study, notably the absence of ragged red fibers in female mice. Interestingly, female mice displayed differential regulation of gene expression, including reduced PPARGC1A expression. Potentially, this prevents mitochondrial accumulation and averts the ragged red fiber phenotype. In addition, we observed increased MYOG expression in female mice, and increased IL1B/IL-1β in female knockout mice, which may contribute to increased muscle regeneration that prevented the appearance of ragged red fibers.

The present study also confirms and extends our previous work that identified BNIP3L as a modulator of MTORC1 in the regulation of insulin signaling [6,24]. Previously, we demonstrated that BNIP3L is responsive to diacylglyceride accumulation, known to promote insulin resistance [32], to activate mitophagy and inhibit insulin signaling through a mechanism contingent on IRS1 phosphorylation [6,24]. In the present study, we observed alterations in signaling pathways associated with glycogen and lipid metabolism in bnip3l-ACTA1-KO mice, which may be a direct effect of bnip3l deletion, or secondary to changes in the fiber-type. Future work should elucidate the role of BNIP3L in muscle during conditions of altered mitochondrial biogenesis, such as exercise training, denervation, and sarcopenia.

Perhaps the most intriguing observation in muscle-specific BNIP3L knockout mice is the alterations in TGFB-MSTN signaling, without altering expression of other secreted metabolic regulators, such as Fgf21 or Gdf15 (Table S1). Elevated MSTN secretion has been implicated in muscle atrophy, insulin resistance, satellite cell inhibition, activation of fibro-adipogenic progenitors/FAPs, and the repression of muscle genes, such as Myh4 [33–35]. Thus, the reduction in Mstn expression and impaired SMAD2 signaling in bnip3l-ACTA1-KO mice is consistent with the phenotypic alterations, including increased Myh4 expression, increased insulin sensitivity, and regeneration without overt fibrosis.

In summary, our results identify a new biological role for BNIP3L maintaining mitochondrial, ER and calcium homeostasis, ultimately modulating the oxidative muscle phenotype. Selective targeting of specific BNIP3L functions could alleviate many of the detrimental manifestations of muscle and metabolic disease.

Materials and methods

Generation of muscle-specific deletion of BNIP3L in mice

All procedures were approved by the Animal Care Committee of the University of Manitoba (#23–030), which adheres to the principles developed by the Canadian Council on Animal Care (CCAC). Bnip3l^fl/fl mice were designed and generated by the University of Manitoba Transgenic Facility using the IDT Alt-R^TM CRISPR-Cas9 system together with ssDNA donors in C57BL/6N zygotes to sequentially insert loxP sites flanking exon 2 of the Bnip3l gene. Guide RNAs, 5’-ggaactatttgagcgctttg-3’ (5’ side) and 5’-ttggttgacccgtttcatcc-3’ (3’ side) together with donors containing the loxP site and 60 bp arms matching the sequence upstream and downstream of the desired insertion site were purchased from Integrated DNA Technologies, Inc (USA). Removal of exon 2 creates a premature stop codon in the Bnip3l transcript that would be expected to result in nonsense mediated decay. This strain is available at The Jackson Laboratory Repository (039220). Hemizygous ACTA1-Cre mice were obtained from Jackson Labs (006149) and crossed with Bnip3l^fl/fl mice to conditionally ablate Bnip3 specifically in skeletal muscle (bnip3l-ACTA1-KO). Experimentation was carried out on 10- to 12-week-old mice.

Physiological assays

Insulin response was characterized by insulin tolerance test using intraperitoneal injection of bovine insulin (0.56 IU/g of body weight; Sigma, I0516). Exercise tolerance was assessed by treadmill running mice to exhaustion while measuring distance. Baseline metabolism measurements were performed by indirect calorimetry using metabolic cages (Columbus Instruments, CLAMS) over a period of 24 h after a familiarization period (24 h). Resting RER was calculated from individual data collected in a metabolic cage during the light cycle. Blood lactate was measured at rest from a left ventricle cardiac puncture using an EPOC Blood Analysis System (Siemens Healthcare Limited). OCR was evaluated using a Seahorse XFe24 Analyzer (Agilent) using intact soleus muscle ex plants [36].

Histology, immunofluorescence, and electron microscopy

Conventional histological stains (H&E, PAS, Picrosirius Red, Gomori Trichrome) were performed on formalin-fixed sections of muscle following standard protocols in the University of Manitoba Histology and Electron Microscopy core facility (https://umanitoba.ca/health-sciences/research/histology-services-imaging-facility-electron-microscopy-platform). Immunofluorescence experiments were performed on fresh-frozen sections of muscle using antibodies listed in Table 1. Transmission electron microscopy was performed in longitudinal sections of muscle fixed in glutaraldehyde and analyzed by an expert pathologist blinded to the experimental conditions [37].

Table 1.

Antibodies.

	Primary antibody	Dilution			Secondary	Dilution
Immunofluorescence
	MYH type IIa (DSHB, SC-71)	1:600			Alexa Fluor 488 IgG1	1:500
	MYH Type I (DSHB, BA-D5)	1:50			Alexa Fluor 555 IgM	1:500
	MYH Type IIb (DSHB, BF-F3)	1:100			Alexa Fluor 350 IgG2b	1:500
	PAX7 (DSHB, PAX7)	1:10			Alexa Fluor 488 IgG	1:500
	LAM/laminin (Abcam, ab11575)	1:500			Alexa Fluor 568 IgG	1:500
	Antibody	Dilution		Notes
Immunoblotting
	BNIP3L/Nix (Cell Signaling Technology [CST], 12396)	1:1000
	BNIP3 (CST, 3769)	1:1000
	Cre (CST 15,036)	1:1000
	TUBB/β-tubulin (CST 86,298)	1:10000
	SMAD2-SMAD3 (CST, 8685)	1:1000
	Phoshp-SMAD2 (CST 18,338)	1:1000
	MB/Myoglobin (CST 25,919)	1:10000		Reduced Protein (5 µg). Reduced secondary antibody (1:100,000)
	PPARGC1A/PGC1-α (Santa Cruz Biotechnology 13,067)	1:100
	NFE2L2/NRF2 (CST 12,721)	1:1000
	AIFM1/AIF1 (CST, 5318	1:1000
	ATP2A/SERCA CST, 9580)	1:1000
	SQSTM1/p62 (CST, 5114)	1:1000
	LC3A (CST, 4599)	1:1000
	Donkey anti-rabbit HRP (Jackson Immunoresearch, 711-035-152)	1:10000
	Donkey anti-mouse HRP (Jackson Immunoresearch, 715-035-150)	1:10000

DSHB, Developmental Studies Hybridoma Bank; all DSBH antibodies used are purchased as concentrated supernatant.

Cell culture, electrical pulse stimulation, and transfection

C2C12 myoblasts were cultured as described previously [6]. C2C12 cells were differentiated into myotubes for up to 10 days in DMEM (without pyruvate; HyClone, SH30081.01) with 0.5% FBS and supplemented with ITS (Gibco 41,400–045), as indicated. Ten-day differentiated myotubes were paced by electrical pulse stimulus (12 V, 1 hz, 1 hr; Ion Optix C-Pace EM). Cells were transfected with plasmids (Table 2) using jetPRIME transfection reagent (Polyplus 101,000,015). The BNIP3L plasmids are available through Addgene (Table 2) [17]. Ser35 mutations were constructed using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs, E0554S). Primers are listed in Table 3.

Table 2.

Plasmids.

Plasmid	Addgene number	Deposited by
Myc-Nix	100795	Joseph Gordon
Bnip3L RNAi pSuper (shBnip3l)	17469	Wafik El-Deiry
pLKO-shNix	100770	Joseph Gordon
GW1-Mito-pHRed	31474	Gary Yellen
CMV-NLS-R-GECO	32462	Robert Campbell
Myc-Nix-S35A	197562	Joseph Gordon
Myc-Nix-S35D	197563	Joseph Gordon
pEGFP-LC3	24920	Toren Finkel
mCherry-ER-3	55041	Michael Davidson
mCherry-mito-7	55102	Michael Davidson
mEmerald-ER-3	54082	Michael Davidson
CMV-ER-LAR-GECO1	61244	Robert Campbell
NFAT-YFP		A gift from T. Miyake and J. McDermott (York University, Toronto)

Table 3.

Primers.

Gene	Primer sequence	Notes
Bnip3l	Forward TGATGTTGAGATGCACACCAG
	Reverse GTGGGATGTTTTCGGGTCTA
Bnip3l 5’ LoxP site	Forward GGAGAGACACACATCTGTAGAATAG	Genotyping
	Reverse ATGACTTAGCAACATCATACAGTTC
Bnip3l 3’ LoxP site	Forward AGTCAGGGCTGTATAGTAAGGT	Genotyping
	Reverse GGTGTATATGTGGAAGCCAGAG
ACTA1-Cre	Forward GCGGTCTGGCAGTAAAAACTATC	Genotyping
	Reverse GTGAAACAGCATTGCTGTCACTT
Myh1	Forward CACTTACCAAACTGAGGAAGACC
	Reverse CCAGGTTGACGTTGGATTG
Myh2	Forward GAGTGAGCTGAAGTCGAAGGA
	Reverse CCCCTTGATAACTGAGAAACCAG
Myh3	Forward GCTAACACGGGAGAAGAAGG
	Reverse TTTTGTTCCTTTCTAGGTCCACA
Myh4	Forward TAGGAACACACAGGGAATGCT
	Reverse TAGCTCTTGCTCAGCCACTC
Myh7	Forward AAGAGCCGGGACATTGGT
	Reverse TTGGAGCTGGGTAGCACAAG
Fundc1	Forward CGGACCTATGGTAGAAAAATACTCA Reverse AGAAGGAAACCACCACCTACTG
Ddit4/Chop	Forward AGCCTGGTATGAGGATCTGC Reverse ACGCAGGGTCAAGAGTAGTG
HSP90B1/GRP94	Forward AGAATGAAGGAAAAACAGGACAA Reverse TCAGAAGTCTCTCAACAAATGGA
Xbp1	Forward CTGAGTCCGCAGCAGGTG Reverse AGAGTCCATGGGAAGATGTTCTG	For splice variant
Bnip3	Forward CCAGACACCACAAGATACCAAC Reverse GTCGACTTGACCAATCCCATATC
Fgf21	Forward CACAGATGACGACCAAGACAC Reverse GACACCCAGGATTTGAATGACC
Gdf15	Forward AGGACTCGAACTCAGAACCAA Reverse CTTCAGGGGCCTAGTGATGT
Prkn	Forward GCTCAAGGAAGTGGTTGCTA Reverse ATGACTTCTCCTCCGTGGTC
Mb	Forward CCAGCCTCTAGCCCAATCA Reverse CCCGGAATGTCTCTTCTTCAG
Il1B	Forward AGGAGGCGAAACAAATCCAC Reverse TATGAGCTTGGAGCGGTACTC
Actb	Forward CTGTGTGGATTGGTGGCTCTA Reverse AAAACGCAGCTCAGTAACAGTCC
Bnip3lS35A	Forward: CCTCAACAGTgccTGGGTGGAGCTACCCATGAACAG	Mutagenesis
	Reverse: CCGGCCGGCGGGGGCAGA
Bnip3lS35D	Forward: CCTCAACAGTgacTGGGTGGAGCTACCCATGAACAGCAGCAATGGC Reverse: CCGGCCGGCGGGGGCAGA	Mutagenesis

Real time qPCR, immunoblotting, and kinome analysis

RNA and protein were isolated from muscle and cells [6]. Array-based qPCR (Bio-Rad, myogenesis and myopathy, SAB gene list) used the built-in primers, while conventional qPCR was performed using primers listed in Table 3. Immunoblot analysis of proteins was performed by SDS-PAGE followed by immunoblotting with antibodies listed in Table 1. Kinome analysis was performed on protein lysates, as previously described [18,19].

Statistical analyses

Statistical analyses were performed using GraphPad Prism software. Unpaired 2-tailed t-test, Chi square, one-way ANOVA (post-hoc: Bonferroni), and two-way ANOVA (post-hoc: Tukey) were used to evaluate significance (α = 0.05).

Funding Statement

This work was support by the Natural Science and Engineering Research Council (NSERC) Canada through a Discovery Grant, and a Diabetes Canada-End Diabetes Grant to JWG. Seed funding was provided by the Children’s Hospital Research Institute of Manitoba, the DREAM research theme, and the Manitoba Centre for Nursing and Health Research. Transgenic and histological services were subsidized by the University of Manitoba Rady Faculty of Health Sciences. J.K. is supported by a Tier 2 Canada Research Chair provided by the Canadian Institutes of Health Research (950–231498; CRC-2021-00098). J.T.F. is supported by an Alexander Graham Bell studentship from NSERC Canada.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2025.2476872