Mitochondrial DNA and TLR9 drive muscle inflammation upon Opa1 deficiency

Abstract

Opa1 participates in inner mitochondrial membrane fusion and cristae morphogenesis. Here, we show that muscle‐specific Opa1 ablation causes reduced muscle fiber size, dysfunctional mitochondria, enhanced Fgf21, and muscle inflammation characterized by NF‐κB activation, and enhanced expression of pro‐inflammatory genes. Chronic sodium salicylate treatment ameliorated muscle alterations and reduced the muscle expression of Fgf21. Muscle inflammation was an early event during the progression of the disease and occurred before macrophage infiltration, indicating that it is a primary response to Opa1 deficiency. Moreover, Opa1 repression in muscle cells also resulted in NF‐κB activation and inflammation in the absence of necrosis and/or apoptosis, thereby revealing that the activation is a cell‐autonomous process and independent of cell death. The effects of Opa1 deficiency on the expression NF‐κB target genes and inflammation were absent upon mitochondrial DNA depletion. Under Opa1 deficiency, blockage or repression of TLR9 prevented NF‐κB activation and inflammation. Taken together, our results reveal that Opa1 deficiency in muscle causes initial mitochondrial alterations that lead to TLR9 activation, and inflammation, which contributes to enhanced Fgf21 expression and to growth impairment.

Introduction

Mitochondrial metabolism, quality control, response to apoptotic stimuli, and hormone activity are partly controlled through the balance between mitochondrial fusion and fission (Twig et al, 2008; Liesa et al, 2009; Sebastian et al, 2012; Liesa & Shirihai, 2013; Munoz et al, 2013). Mitochondrial fusion in mammalian cells is regulated by the proteins mitofusin 1 and mitofusin 2 (Mfn1 and Mfn2) and optic atrophy 1 (Opa1). Various Opa1 isoforms are located in the mitochondrial intermembrane space or inserted within the inner mitochondrial membrane (Olichon et al, 2003; Cipolat et al, 2004; Griparic et al, 2004; Ishihara et al, 2006). The existence of multiple Opa1 isoforms and cleavage mechanisms may explain the role of this protein beyond mitochondrial inner membrane fusion, such as in cristae remodeling, and supercomplex formation (Frezza et al, 2006; Cogliati et al, 2013).

In humans, OPA1 mutations have been reported in patients affected by autosomal dominant optic atrophy or ADOA (Alexander et al, 2000; Delettre et al, 2000). Overall, data support the view that the pathogenesis of ADOA occurs as a result of haploinsufficiency (Pesch et al, 2001). Human fibroblasts carrying OPA1 mutations show impaired oxidative phosphorylation and mitochondrial fusion (Amati‐Bonneau et al, 2008). Some of these OPA1 missense mutations are associated with altered mitophagy and parkinsonism (Carelli et al, 2015). In addition to dominant optic atrophy, OPA1 mutations also cause a multi‐systemic disorder called ADOA plus syndrome, which results in severe myopathy (Amati‐Bonneau et al, 2008; Hudson et al, 2008; Zeviani, 2008). Interestingly, these patients show multiple mitochondrial DNA (mtDNA) deletions in muscles, which suggests a role of OPA1 on mtDNA stability.

The appropriate homeostasis of mitochondria is essential in the maintenance of cellular health. Mitochondria are a source of damage‐associated molecular patterns (DAMPs), and among others, mtDNA has been shown to induce a pro‐inflammatory state (Zhang et al, 2010; Oka et al, 2012; Wenceslau et al, 2014). MtDNA has been reported to activate immunity through two distinct pathways, namely Toll‐like receptor 9 (TLR9; Zhang et al, 2010; Oka et al, 2012; Liu et al, 2015), and cGAS activation (White et al, 2014; West et al, 2015). Under basal conditions, TLR9 is located in the endoplasmic reticulum (ER), and upon stimulation by inducers, TLR9 translocates to the membrane of endosomes or to lysosomes, bind to ligands, and initiate cellular inflammation (Latz et al, 2004; Zhang et al, 2010; Wei et al, 2015; De Leo et al, 2016). Interaction of TLR9 with mtDNA activates the nuclear factor kappa B (NF‐κB) signaling and increases the expression of other pro‐inflammatory cytokines such as tumor necrosis factor‐α (TNF‐α), interleukin (IL)‐6, IL‐1β (Julian et al, 2013; Yu & Bennett, 2014; Zhang et al, 2014).

Recently, cytosolic mtDNA has been reported to engage cytosolic antiviral signaling and to enhance the expression of interferon‐stimulated genes (Rongvaux et al, 2014; White et al, 2014; West et al, 2015). Thus, cytosolic mtDNA activates the DNA sensor cGAS and promotes STING‐IRF3‐dependent signaling (Rongvaux et al, 2014; White et al, 2014; West et al, 2015). Furthermore, neutrophils extrude interferogenic mtDNA by a process dependent in part on lysosomal activity and that occurs in the presence of a constitutive defect in mitophagy (Caielli et al, 2016),

Based on the observations that OPA1 mutations cause mtDNA instability (Kim et al, 2005; Amati‐Bonneau et al, 2008; Hudson et al, 2008; Yu‐Wai‐Man et al, 2010), we reasoned that OPA1 deficiency in a non‐immune cells should not only alter mitochondrial morphology but also mitochondrial stability, and in consequence trigger immune responses. Based on this, we analyzed the impact of Opa1 loss‐of‐function in skeletal muscle. Our data indicate that Opa1 deletion in skeletal muscle causes mitochondrial inflammatory myopathy characterized by impaired mitochondrial function, pro‐inflammatory cytokine production, altered myofiber morphology, and muscle dysfunction. In addition, we found that the activation of NF‐κB in Opa1‐deficient muscle cells is mediated by TLR9 and by mtDNA.

Results

Skeletal muscle‐specific Opa1 ablation causes reduced body growth and premature death

In preliminary studies, we monitored skeletal muscle regeneration in a cardiotoxin (CTX) injury‐induced model in gastrocnemius muscles in wild‐type animals at a range of times (Fig EV1A). Two days post‐injury (dpi), when satellite cells become active, muscles were treated with adenoviruses encoding for engineered miRNAs against Opa1 (miR Opa1) or with control adenoviruses encoding for miRNA with no homology in the mouse genome (miR Ctrl; Fig EV1B). Opa1 deficiency (miR Opa1) not only caused impaired muscle regeneration but also increased the presence of immune cells and reduced muscle fiber size in regenerating myofibers (Figs 1A and B, and EV1C). Immunostaining of the developmental form of MHC (dMHC) revealed a reduction in controls between 9 and 12 days post‐CTX treatment, whereas Opa1‐deficient muscles showed sustained high expression of dMHC (Figs 1C and EV1D), indicating impaired muscle fiber maturation.

Figure EV1. Skeletal muscle‐specific Opa1 ablation causes reduced body growth and premature death.

• A
  Scheme of the experimental design of CTX‐induced injury.
• B
  Opa1 protein levels in control C2C12 myoblasts, which were transduced with control miRNA (miR Ctrl) or in Opa1 loss‐of‐function C2C12 myoblasts, which were transduced with adenoviruses encoding for miRNA against Opa1 (miR Opa1) (n = 5).
• C
  Distribution of myofiber size in mice treated with miR Ctrl or miR Opa1 adenoviruses at dpi 12 (n = 5).
• D
  Representative image of dMHC immunohistochemistry from gastrocnemius muscle treated with miR Ctrl or miR Opa1 adenoviruses (12 dpi). Scale bars, 100 μm.
• E
  Partial genomic structure of the Opa1 gene showing the scission of exon 5, thus deleting all Opa1 protein isoforms.
• F
  Opa1 mRNA levels in the gastrocnemius muscle of loxP (non‐expressing Cre Opa1^loxP/loxP mice) and skeletal muscle‐specific KO mice (KO) (n = 10).
• G
  Opa1 protein levels in tissue homogenates from control (loxP) and skeletal muscle‐specific KO mice (KO).
• H, I
  Relative weight of muscles (gastrocnemius (Gast) and quadriceps (Quad)) (panel H) (n = 18) and organs (liver (L), heart (H), kidney (K), spleen (S), T thymus (T), subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and brown adipose tissue (BAT)) (panel I) (n = 32) of 9‐week‐old loxP and KO mice. These data are expressed as g of tissue/g of body weight (BW).
• J
  Distribution of myofiber size of 150 myofibers in quadriceps muscle.
• K, L
  Mean cross‐sectional area (CSA) (K) and distribution of myofiber size (L) of 150 myofibers in diaphragm muscle.
• M
  Representative images of hematoxylin and eosin‐stained lung sections from loxP and KO mice. Thick arrows indicate congestion and atelectasis. Thin arrows indicate normal parenchyma. Asterisks show bronchioles. Scale bars, 100 μm.

Data information: Data represent mean ± SEM. *P < 0.001 vs. control loxP mice. Data were analyzed using Student's t‐test.

Figure 1. Skeletal muscle‐specific Opa1 ablation causes reduced body growth and premature death.

• A
  Transversal sections of gastrocnemius muscles of 3‐month‐old mice injected with CTX as an injury‐induced model and, 2 days later, with adenoviruses encoding for non‐targeting miRNA (miR Ctrl) or miRNA targeting Opa1 (miR Opa1). Samples were taken on various days after the injury (dpi, n = 5). Scale bars, 100 μm.
• B
  Mean cross‐sectional area (CSA) of 150 myofibers per gastrocnemius muscle at dpi 12.
• C
  Quantification of positive MHC myofibers vs. total regenerating myofibers of gastrocnemius muscle treated with miR Ctrl or miR Opa1 adenoviruses at dpi 9 and 12 (n = 5).
• D
  Opa1 protein levels in tissue homogenates of control (loxP) and skeletal muscle‐specific KO mice (KO). The skeletal muscle used was gastrocnemius muscle (n = 6).
• E
  Survival curves for loxP and Opa1 KO mice (n = 25).
• F
  Picture of loxP and KO mice at 9 weeks of age.
• G
  Body weight of loxP and KO male and female mice (n = 25).
• H
  Grip strength in loxP and KO mice. loxP (n = 7) and KO (n = 4).
• I
  Mean cross‐sectional area (CSA) of 150 myofibers in quadriceps muscle.
• J–K
  Plasma concentration of growth hormone (GH) (J) and Igf1 (K) of loxP (n = 8) and KO mice (n = 10).

Data information: Data represent mean ± SEM. *P < 0.001 vs. control loxP mice (with the exception of panel B in which P < 0.01). Data were analyzed using Student's t‐test (B, C, I, J, and K) or analysis of variance test (G and H).

Based on these data, we analyzed the impact of Opa1 depletion on muscle homeostasis by the generation of skeletal muscle‐specific knockout mice. This was performed by crossing homozygous Opa1‐loxP/loxP mice with a mouse strain expressing Cre recombinase under the control of the myogenin promoter (Li et al, 2005; Figs 1D and EV1E–G). Opa1 ablation in this specific tissue induced a dramatic reduction in life span (Fig 1E) and impaired normal growth (Fig 1F and G). At 9 weeks of age, Opa1 KO mice showed a reduction in relative weight of skeletal muscle (Fig EV1H) and no change in various organs and tissues (Fig EV1I). Muscle force was also lower in Opa1 KO mice (Fig 1H). In keeping with this, histological analyses of KO mice indicated a marked decrease in fiber size and cross‐sectional area in quadriceps and diaphragm (Figs 1I and EV1J, K and L). Consistent with altered diaphragm morphology, we observed extensive congestion and atelectasis in the lungs of these mice (Fig EV1M). In keeping with the reduced growth, plasma levels of GH were high (Fig 1J) and IGF1 were low (Fig 1K). Characterization of the hepatic profile of Opa1 KO mice showed reduced phosphorylation of STAT5 (Fig 2A), and decreased expression of its target genes Igf1 and Fos (Fig 2B). Also, it revealed reduced expression of growth hormone receptor (Ghr) and increased expression of the STAT5 inhibitor Socs3 (Fig 2B).

Figure 2. Muscle‐specific Opa1 ablation induces growth hormone resistance.

• A
  Stat5 phosphorylation in livers of loxP and KO mice (n = 10).
• B
  mRNA levels of Igf1, Fos, growth hormone receptor (Ghr), and Socs3 in liver of loxP and KO mice (n = 5).
• C
  Fgf21 gene expression in muscle of loxP and KO mice (n = 8).
• D
  FGF21 levels in plasma of loxP (n = 8) and KO mice (n = 10).
• E–G
  Opa1 mRNA levels in muscle (E), and Fgf21 gene expression in muscle (F) and liver (G) of loxP, Opa1 KO, FGF21 KO, and double KO (Opa1 + FGF21 KO, DKO) (n = 8).
• H
  Hepatic expression of genes encoding Pgc1α, respiratory complex IV subunit Cox7a1, fatty acid oxidation components (Cpta1, Mcad, and Vlcad), and gluconeogenic enzymes (Pepck and G6p) in loxP, Opa1 KO, DKO, and Fgf21 KO mice (n = 7).
• I, J
  Relative weight of gastrocnemius and quadriceps muscles, (I) and organs (J) of loxP, Opa1 KO, DKO, and Fgf21 KO mice (n = 10). These data are expressed as g of tissue/g of body weight (BW).
• K
  Igf1, Fos, and Ghr mRNA levels in the livers of loxP, Opa1 KO, DKO, and Fgf21 KO mice (n = 10).

Data information: Data represent mean ± SEM. *P < 0.05 vs. control loxP mice. ^# P < 0.05 vs. DKO mice. Data were analyzed using Student's t‐test (B–D) or analysis of variance with Tukey's post hoc test (E–K).

Gene set enrichment analysis of genomic profiling in muscles of control and Opa1 KO mice revealed the up‐regulation of ATF4 target genes in muscles of Opa1 KO mice (Dataset EV1; Kim et al, 2013) among the large number of genes with deregulated expression (Dataset EV2; 486 genes up‐regulated and 175 repressed). Fgf21, an ATF4 target gene, was greatly increased in the muscles of Opa1 KO mice (Fig 2C). Circulating Fgf21 levels were fourfold greater in these animals (Fig 2D). Given the observation that Fgf21 causes resistance to GH (Inagaki et al, 2008), we explored the possibility of a rescue by generating a double knockout (DKO) mouse for Fgf21 and Opa1 (Fig 2E–G). DKO mice normalized hepatic PGC‐1α expression consistent with the regulatory role of Fgf21 (Potthoff et al, 2009; Fig 2H). In keeping with these observations, Opa1 KO mice showed enhanced expression of hepatic genes relevant in gluconeogenesis, lipid metabolism, or mitochondrial respiration, and DKO mice showed a normalized expression (Fig 2H). In contrast, DKO mice showed similar muscle and tissue weights to those of Opa1 KO mice (Fig 2I and J). Furthermore, DKO failed to enhance the hepatic expression of Igf1 or Fos genes (Fig 2K). In all, these data indicate that Fgf21 mediates the metabolic alterations occurring in liver from Opa1 KO mice, but it is not involved in the reduced growth of Opa1 KO mice.

Opa1 ablation causes inflammatory myopathy

Histological inspection of Opa1‐deficient muscle sections revealed a substantial number of necrotic myofibers, as well as regenerating fibers (centrally located nuclei; Fig 3A). Given the preliminary data obtained from the CTX‐treated muscles (Fig 1A), we questioned whether Opa1‐deficiency promotes muscle inflammation, and whether that could be the reason for the growth impairment. Non‐specific esterase (NSE) and major histocompatibility complex (MHC) class I stainings revealed severe inflammation in muscles from Opa1 KO mice (Fig 3A). In addition, the expression of NF‐κB target genes was also induced (Fig 3B), as well as the plasma levels of pro‐inflammatory cytokines IL‐1β and IL‐6 (Fig 3C).

Figure 3. Opa1 ablation causes mitochondrial inflammatory myopathy.

1. Representative images of hematoxylin and eosin (H/E) staining, non‐specific esterase (NSE) staining and MHC class I expression in gastrocnemius muscle sections of 6‐week‐old loxP and skeletal muscle‐specific Opa1 KO mice. Inset panels are 4× magnification of the outlined areas. Asterisks in H/E indicate necrotic fibers. Asterisks in NSE indicate the presence of macrophages. Scale bars, 100 μm.
2. Expression of NF‐κB target genes in muscles of loxP and KO mice (n = 7).
3. Circulating IL‐1β and IL‐6 measured in plasma of 8‐ to 10‐week‐old loxP and KO mice (n = 9).
4. Quantification of IL‐1β and CD68 expression in gastrocnemius, and CK activity in plasma of loxP and Opa1 KO animals at 3, 7, 10, and 13 weeks of age. Values of Opa1 KO mice are represented relative to the loxP group. The staining quantification was performed using Trainable Weka Segmentation plugin from ImageJ (5 images per group).
5. Expression of NF‐κB target genes in muscles of control and inducible skeletal muscle‐specific Opa1‐KO mice at 30, 50, and 90 days after tamoxifen treatment (n = 6).
6. Transcriptional activity of NF‐κB in myoblast expressing scramble shRNA (Scr) or shRNA against Opa1 (KD) (n = 5).
7. Expression of NF‐κB target genes in Scr and KD myoblasts (n = 6).
8. Levels of IL‐1β in culture media from C2C12 myoblasts (n = 6).

Data information: Data represent mean ± SEM. *P < 0.05 vs. control groups. Data were analyzed using Student's t‐test.

In order to determine whether inflammation is a primary consequence of Opa1 depletion, we followed the presence of IL‐1β and CD68 in muscle and the activity of creatine kinase (CK) in plasma in 3‐, 7‐, 10‐, and 13‐week‐old mice. Highly positive labeling for IL‐1β was already detected at 3 weeks in Opa1‐deficient muscles and was maintained with age (Fig 3D and Appendix Fig S1A). Plasma CK activity was increased from week 7 (Fig 3D). However, pro‐inflammatory macrophages (stained for CD68) were not detected before 13 weeks of age (Fig 3D and Appendix Fig S1B) in these mice, thereby indicating that the inflammation in skeletal muscle caused by Opa1 deficiency precedes macrophage infiltration. No evidence of apoptosis was detected in muscles of control or KO mice (Fig EV2A).

Figure EV2. Opa1 ablation causes mitochondrial inflammatory myopathy.

• A
  Caspase 3 expression in loxP and skeletal muscle‐specific Opa1 KO mice.
• B
  CD68 expression in muscles of control and inducible muscle‐specific Opa1 KO mice at 30, 50, and 90 days after tamoxifen treatment (n = 4).
• C
  Opa1 protein levels in C2C12 control myoblasts (Scr) and in stably depleted Opa1 myoblasts (KD).
• D
  LDH activity in cultured media of Scr and Opa1 KD myoblasts (n = 8).
• E, F
  Caspase 3 (E) and Parp (F) expression in Scr and Opa1 KD myoblasts. Apoptosis‐positive control was obtained by treating C2C12 myoblasts with 10 μm staurosporin for 4 h.

Data information: Data represent mean ± SEM. *P < 0.05 vs. control loxP mice. Data were analyzed using Student's t‐test.

To further confirm inflammation being a primary event, inducible skeletal muscle‐specific Opa1 KO mice were treated with tamoxifen, in order to specifically ablate Opa1 in adult muscles, and the expression of NF‐κB target genes was measured at different times. Thirty days after the onset of tamoxifen treatment, enhanced NLRP3 and ASC expression was detected under conditions in which muscle mass is normal (Fig 3E). At 50 days of tamoxifen treatment, the expression of genes S100A9, TNFα, and RAGE was also enhanced in Opa1‐deficient muscles under conditions in which muscle loss was already apparent (Fig 3E). These changes in the expression of NF‐κB target genes were detectable before alterations in CD68, a marker of pro‐inflammatory macrophages (Fig EV2B). These results indicate that Opa1 deficiency triggers muscle inflammation even when depletion occurs in mature muscles from adult mice.

To determine whether cultured muscle cells show a cell‐autonomous response, we analyzed whether Opa1‐deficient C2C12 myoblasts exhibit an inflammatory state. Phenocopying the muscle profile, Opa1 KD myoblasts (Fig EV3C) showed a marked increase of NF‐κB promoter activity (Fig 3F) and the expression of NF‐κB target genes (Fig 3G). IL‐1β was also increased in culture media from Opa1‐deficient cells (Fig 3H). Opa1 deficiency did not cause necrosis (Fig EV2D) or apoptosis (Fig EV2E and F).

Figure EV3. Salicylate treatment reduces inflammation and improves growth in Opa1 KO mice.

• A
  Expression of NF‐κB target genes in muscles of loxP and KO mice treated with sodium salicylate (Sal) or with PBS (PBS).
• B, C
  Circulating IL‐6 (B) and IL‐1β (C) measured in plasma of loxP and KO mice treated or not with sodium salicylate.
• D
  Body weight increase in loxP and KO mice treated or not with salicylate for 20 days.
• E
  Distribution of myofiber size in quadriceps muscles from loxP and KO mice treated or not with salicylate.
• F, G
  Igf1 (F) and Fos (G) mRNA levels in liver from salicylate‐treated and non‐treated mice.

Data information: Data represent mean ± SEM. *P < 0.05 vs. control loxP mice. ^# P < 0.05 vs. KO mice. Data were obtained from 8‐ to 10‐week‐old loxP and KO mice (n = 6). Data were analyzed using analysis of variance with Tukey's post hoc test.

In all, our data indicate that Opa1‐deficient muscle undergoes a primary inflammatory process characterized by NF‐κB activation, followed by necrosis and then macrophage infiltration. In addition, they reveal that Opa1 deficiency‐induced inflammation is a cells.

Salicylate treatment reduces inflammation and improves growth in Opa1 KO mice

Based on the previous observations, we examined whether the pro‐inflammatory state was responsible for the reduced growth of Opa1 KO mice. To this end, mice were treated for 30 days with sodium salicylate, a known NF‐κB inhibitor (Kopp & Ghosh, 1994; Yin et al, 1998) or with PBS. Anti‐inflammatory efficiency was documented by low expression of NF‐κB target genes and by low circulating levels of IL‐6 and IL‐1β (Fig EV3A–C). Salicylate‐treated Opa1 KO mice showed a notable increase in body weight and relative muscle weight compared to PBS‐treated mice (Figs 4A and EV3D). Moreover, fiber size and distribution were markedly enhanced in KO mice treated with salicylate (Figs 4B and C, and EV3E), and also CK activity was normalized (Fig 4D), consistent with reduced muscle necrosis. To further confirm that growth reduction is due to inflammation, we analyzed circulating and hepatic markers. Salicylate treatment markedly reduced plasma GH (Fig 4E), increased the hepatic expression of Ghr, Igf1, and Fos (Figs 4F and EV3F and G), and reduced SOCS3 expression (Fig 4G). Furthermore, the muscle expression and circulating levels of Fgf21 were normalized in Opa1 KO mice treated with salicylate (Fig 4H and I). In all, our data indicate that inflammation is a key pathological event leading to many metabolic alterations detected under Opa1 deficiency.

Figure 4. Salicylate treatment reduces inflammation and improves growth in Opa1 KO mice.

• A
  Relative weight of gastrocnemius and quadriceps muscles of loxP and KO mice treated with sodium salicylate (Sal) or PBS (PBS) for 30 days.
• B
  Mean CSA of 150 myofibers in quadriceps muscles.
• C
  Representative H/E staining of quadriceps muscles of LoxP and KO mice treated with salicylate or PBS. Scale bars, 50 μm.
• D
  Creatine kinase (CK) activity measured in plasma of salicylate‐treated and PBS‐treated mice.
• E
  Circulating GH levels from salicylate‐treated and PBS‐treated animals.
• F, G
  Growth hormone receptor (GHR) (F) and SOCS3 (G) mRNA levels in livers of treated and PBS‐treated animals.
• H
  Fgf21 mRNA levels measured in tibialis muscle of salicylate‐treated and PBS‐treated mice.
• I
  Concentration of Fgf21 in plasma of loxP and treated and PBS‐treated KO mice.

Data information: Data represent mean ± SEM. *P < 0.05 vs. control loxP mice. ^# P < 0.05 vs. KO mice. Data were obtained from 8‐ to 10‐week‐old loxP and KO mice (n = 6). Data were analyzed using analysis of variance with Tukey's post hoc test.

Opa1 depletion causes mitochondrial dysfunction and mitochondrial DNA stress

Next, we studied whether Opa1 ablation causes mitochondrial dysfunction in skeletal muscle. Analysis of mitochondrial morphology showed a markedly altered organization of the mitochondrial network and an enhanced mitochondrial density (Fig 5A). Opa1‐deficient muscles showed increased levels of various mitochondrial proteins, but neither of OXPHOS subunits nor of Pgc‐1α expression (Fig 5B and C and Appendix Fig S2A). Moreover, analysis of mitochondrial respiration in permeabilized muscles revealed a marked reduction in state 3 activity (Fig 5D). In parallel to muscles, MtDsRed labeling of Opa1 KD myoblasts revealed a complete fragmentation of the mitochondrial network (Fig 5E). In vivo cell imaging indicated that Opa1‐deficient cells showed movement of fragmented mitochondria and frequent mitochondria encounters but no fusion events (Movies EV1 and EV2). Mitochondrial fusion analysis by expression of a photo‐activatable mitochondrial GFP protein (PA‐GFPmt) revealed a time‐dependent dilution and diffusion of the fluorescence through the mitochondrial network in control cells, whereas Opa1‐deficient cells generated a static signal with time (Fig 5F and Appendix Fig S2B), compatible with lack of mitochondrial fusion. There were no changes in mitochondrial mass, as the expression of mitochondrial proteins PORIN, TIM44, SDHA, or TOM20 was not reduced in Opa1‐deficient cells (Fig 5G). Under these conditions, Opa1 deficiency also caused decreased mitochondrial respiration, which was characterized by enhanced respiratory leak and reduced maximal respiratory capacity (Fig 5H), increased glycolysis (Fig 5I), and unaltered ATP levels (Fig 5J).

Figure 5. Opa1 depletion causes mitochondrial dysfunction and mitochondrial DNA stress.

• A
  Representative images of mitochondrial network of tibialis muscles of loxP and KO mice electroporated with cDNA encoding DsRed2‐Mito vector (n = 3). Scale bars, 5 μm.
• B
  Relative mitochondrial protein expression of gastrocnemius muscles of loxP and KO mice. Values were corrected by α‐tubulin (10 observations per group).
• C
  PGC‐1α expression in muscles of loxP and KO mice (n = 7).
• D
  Mitochondrial respiration assayed in isolated myofibers of tibialis muscles (2–4 mg) from loxP or KO mice (n = 6).
• E
  Mitochondrial morphology of control (Scr) and Opa1 KD (KD) C2C12 myoblasts. Scale bars, 10 μm.
• F
  Estimation of mitochondrial fusion as PA‐GFP dilution in control (black line, n = 27) and Opa1‐silenced (purple line, n = 17) C2C12 myoblasts.
• G
  Expression of mitochondrial proteins in Scr and Opa1 KD myoblasts (n = 4).
• H
  Mitochondrial oxygen consumption rates (OCR) measured in intact Scr or Opa1‐silenced C2C12 myoblasts (n = 5). Five mM glucose was used as a substrate. Proton leak (respiration independent of ATP synthesis) was induced by 1.25 μM oligomycin (complex V inhibitor). Respiratory electron transfer system (ETS) capacity was analyzed using 1 μM FCCP. Non‐mitochondrial electron transport OCR was determined by the addition of the complex III inhibitor antimycin A (0.1 μM) and subtracted from the total OCR in order to obtain mitochondrial OCR.
• I
  Glycolytic flux assessed as extracellular acidification rates were measured in intact Scr or KD C2C12 myoblasts (n = 5). Five mM glucose was used as a substrate.
• J
  ATP content in Scr and Opa1 KD C2C12 myoblasts (n = 15).
• K
  mtDNA levels relative to nuclear DNA (Tert) (n = 4).
• L
  TFAM protein levels (n = 4).
• M, N
  Nucleoid composition in Scr and Opa1 KD myoblasts. Number of mtDNA nucleoids per cell (M) and area of each nucleoid (N). Data are from a representative experiment with 20 cells randomly quantified per condition.

Data information: Data represent mean ± SEM. *P < 0.05 vs. control groups. Data were analyzed using Student's t‐test.

Because some OPA1 mutations are linked to mtDNA instability (Amati‐Bonneau et al, 2008; Hudson et al, 2008), we analyzed the effect of Opa1 depletion on mtDNA. Indeed, Opa1 deficiency in muscle cells led to a near 50% reduction in mtDNA copy number measured by the amplification of seven different regions of the mtDNA (Fig 5K). Accordingly, TFAM protein levels were also reduced to a half (Fig 5L). Nucleoid composition analysis by confocal microscopy revealed a reduction in the number of nucleoids per cell (Fig 5M) and enhanced nucleoid size (Fig 5N). This pattern of changes suggests the existence of mitochondrial DNA stress similar to what detected upon TFAM repression (West et al, 2015).

Opa1 deficiency‐driven inflammation requires mtDNA and is independent of cGAS

Based on the development of mtDNA stress in Opa1‐depleted cells and because mtDNA triggers inflammation (Arnoult et al, 2011), we analyzed whether Opa1‐deficiency causes inflammation through mtDNA. We depleted mtDNA in C2C12 muscle cells by chronic treatment with ethidium bromide (EtBr). EtBr‐treated cells showed a 95% reduction of mtDNA in Opa1‐deficient cells (Fig 6A and Appendix Fig S3A). EtBr‐induced mtDNA depletion abolished the expression of NF‐κB target genes Il‐6, Rage, Hmgb1, or Tlr2 in Opa1‐deficient cells (Fig 6B). These results indicate the necessity of mtDNA to trigger the inflammatory response.

Figure 6. Opa1 deficiency‐driven inflammation requires mtDNA and is independent of cGAS .

• A, B
  mtDNA levels relative to nuclear DNA (D‐loop1 to Tert) and expression of NF‐κB target genes of Opa1 KD (KD) treated or not with EtBr. Relative values to untreated control myoblasts (n = 3).
• C
  Representative experiment of cytosolic extraction by digitonin protocol. Total homogenate (TH), pellet fraction after cytosolic extraction (P), and cytosolic extract (C).
• D
  mtDNA levels relative to nuclear DNA (D‐loop1 to Tert) in total homogenate, pellet and cytosolic fraction of Scr and Opa1 KD (KD) cells (n = 3).
• E
  Scr and KD mRNA expression of interferon‐stimulated genes (ISG) and Ifnβ (n = 6).
• F
  Expression of NF‐κB target genes in control (Scr), cGas depleted (cGas KD), Opa1 depleted (Opa1 KD), and double cGas‐Opa1 ablated (DKD) cells (n = 6).

Data information: Data represent mean ± SEM. *P < 0.05 vs. Scr control group or untreated group. Data were analyzed using Student's t‐test or analysis of variance with Tukey's post hoc test (F).

MtDNA stress has been reported to cause inflammation when it is found in the cytosol, through its recognition by cGAS resulting preferentially in a type I interferon (IFN) response (Rongvaux et al, 2014; White et al, 2014). Thus, we next analyzed the presence of mtDNA in the cytosol by using permeabilization with digitonin as reported (West et al, 2015). The method permits to isolate cytosol from the rest of the cellular material (Fig 6C). Under these conditions, mtDNA was retained in organelles and it was not detected in the cytosolic fraction either in Opa1‐depleted or in control cells (Fig 6D and Appendix Fig S3B). In this connection, we did not detect induction of interferon‐stimulated genes neither Ifnβ in Opa1‐deficient cells (Fig 6E), which lie downstream of cGAS. Furthermore, cGAS loss‐of‐function (Appendix Fig S3C and D) did not rescue NF‐κB activation in Opa1‐deficient cells (Fig 6F).

Opa1 deficiency disrupts mitophagy completion in muscle cells

Because Opa1 depletion does not lead to the release of mtDNA to the cytosol, we examined whether mtDNA is maintained to mitochondria. Analysis of the distribution of mtDNA and TOM20 (an outer mitochondrial membrane protein) revealed a complete co‐localization of both markers in control muscle cells (Appendix Fig S4A and B). The two markers also showed close apposition in Opa1‐deficient muscle cells (Appendix Fig S4A and B) suggesting that mtDNA and TOM20 were in the same compartment and that mtDNA is unlikely to be located in the cytosol.

Based on data indicating that autophagy malfunction caused mtDNA‐mediated inflammation (Nakahira et al, 2011; Oka et al, 2012), we next assessed the functionality of the mitochondrial quality control system. Mitochondrial autophagy was analyzed upon treatment of control of Opa1‐deficient muscle cells with the depolarizing agent CCCP. A rapid recruitment of LC3‐II to mitochondrial enriched fractions was detected upon incubation with CCCP for 30 min in both control and Opa1‐deficient cells, suggesting a normal mitophagy induction in Opa1‐deficient cells (Fig 7A and Appendix Fig S4C). Alterations in mitophagy resolution were assessed by inducing mitochondrial damage for longer times and analyzing the rate of mitochondrial nucleoid disappearance (Lazarou et al, 2015). While control cells responded with a marked reduction in mitochondrial nucleoids upon 9‐h incubation with CCCP, Opa1‐deficient cells showed no mitophagic response (Fig 7B and C). Mitochondrial autophagy was analyzed in skeletal muscle by measuring the abundance of LC3‐II in mitochondrial enriched fractions after treatment for 5 days with the lysosomal inhibitor chloroquine. Control muscles responded to chloroquine by enhanced LC3‐II abundance (Fig 7D, and Appendix Fig S4D) in keeping with prior data (Sebastian et al, 2016). In contrast, Opa1 ablated muscles only showed a weak non‐significant increase in LC3‐II (Fig 7D and Appendix Fig S4D). In keeping with these data, ultrastructural analysis of Opa1‐ablated myofibers of 3‐week‐old mice showed a dramatic accumulation of abnormal mitochondria (Fig 7E).

Figure 7. Opa1 deficiency affects mitophagy.

1. LC3‐II recruitment to mitochondrial enriched fractions in Scr and KD myoblasts after 30 min of CCCP treatment (n = 4).
2. Study of mitophagy completion by mtDNA nucleoid clearance upon long CCCP treatment in Scr and KD cells. Data represent mean ± SEM of three independent experiment with 20 cells randomly quantified per condition. *P < 0.05 vs. Scr untreated, ^# P < 0.05 vs. 9 h of treatment, and ^& P < 0.05 KD vs. Scr.
3. Representative images of the nucleoid clearance upon CCCP treatment. Scale bars 20 μm.
4. Relative LC3‐II protein levels in mitochondrial enriched fractions of quadriceps muscle after 5 days of intraperitoneal treatment with chloroquine (CQ) (10 observations per group).
5. Electron micrographs of gastrocnemius muscles from 3‐week‐old loxP and KO mice. Scale bars, 1 μm in left panels, and 500 nm in right panels.
6. Time‐course study of LC3‐II protein levels in Scr and KD C2C12 myoblasts under CQ treatment (n = 6).
7. Relative LC3‐II protein levels in total homogenates of gastrocnemius muscle after 5 days of intraperitoneal treatment with CQ (six observations per group).
8. Quantification of p62 expression and Gomori's Trichrome staining in gastrocnemius muscle of Opa1 KO animals at 3, 7, 10, 13, and 20 weeks of age. Values of Opa1 KO mice are represented relative to the loxP group. This quantification was performed using Trainable Weka Segmentation plugin from ImageJ (five images per group).
9. Relative expression of 3‐week‐old mice of p62 (seven and four observations in control and KO mice, respectively).

Data information: Data represent mean ± SEM. *P < 0.05 vs. control group. Data were analyzed using Student's t‐test (A, D, G, and I) or two‐way analysis of variance with Tukey's post hoc test (B, F, and H).

Studies in Opa1‐deficient muscle cells also revealed a defective autophagy assessed by the reduced buildup of LC3‐II upon incubation in the presence of chloroquine (Fig 7F and Appendix Fig S4E). Opa1‐depleted muscle also showed an impaired autophagy flux and treatment for 5 days with chloroquine caused a marked buildup of the autophagic protein LC3‐II in muscles from control mice, whereas only a weak induction was detectable in Opa1 KO mice (Fig 7G and Appendix Fig S4F). Time‐course studies revealed significantly increased levels of p62 and presence of ragged red fibers in the muscles of 3‐week‐old mice, which further accumulated with age (Fig 7H, Appendix Fig S4H and I). In addition, 3‐week‐old Opa1 KO mice showed enhanced p62 abundance compared to controls by Western blot (Fig 7I and Appendix Fig S4G). In all, our data are coherent with the view that Opa1 deficiency results in a defective mitophagy characterized by normal induction followed by defective resolution. However, measurements of mitophagy are difficult in skeletal muscle and therefore further studies are necessary to document the impact of Opa1 ablation on muscle mitophagy.

Localization of mitochondrial DNA in Opa1‐deficient muscle cells

Recent data have established a relationship between autophagy malfunction and presence of undegraded mtDNA, which is recognized by TLR9 and results in inflammation (Nakahira et al, 2011; Oka et al, 2012). Based on the observation indicating that Opa1 deficiency causes an impairment in the resolution of mitophagy in muscle cells, next we explored the localization of mitochondrial DNA, in order to provide preliminary evidence on a re‐routing of mitochondria‐containing autophagosomes to the endosomal pathway, which would enable the recognition of mtDNA by TLR9. To this end, we analyzed the cellular localization of mtDNA regarding the different endo‐lysosomal compartments. In some studies, we carried out the co‐staining of mtDNA and VPS4 (a subunit of ESCRT‐III complex, involved in the formation of multivesicular bodies). We detected the presence of aggregation of VPS4 in Opa1 KD cells compared to a diffuse cytosolic, barely noticeable in control cells (Appendix Fig S5A). In addition, we detected a close co‐apposition of the signals corresponding to VPS4 and mtDNA in Opa1 KD cells (60% of mtDNA was close to VPS4; Appendix Fig S5A and B). There was no co‐localization between LAMP1 and VPS4; thus, we suggest the structures labeled by these markers are different (Appendix Fig S5C). The lack of triple staining with a mitochondrial protein or with a late endosomal marker does not permit us to conclude at this stage whether mitochondrial DNA or mitochondria are located in late endosomes.

Analysis of the co‐staining of mtDNA and early endosomes (EEA1 protein marker) showed a complete exclusion of the markers under control or Opa1‐deficient conditions (Appendix Fig S6A). Given the reported presence of TLR9 in the lysosomes (De Leo et al, 2016), we examined whether this could be a relevant compartment. To this end, we performed double and triple staining using LAMP1 as a lysosomal marker. These studies showed no co‐localization between LAMP1 and TOM20 (containing 100% of mtDNA nucleoids; Appendix Fig S6B). Triple labeling of mtDNA, LAMP1, and VPS4 in control and Opa1‐deficient cells showed exclusion localization between mtDNA and LAMP1, and apposition of mtDNA and VPS4 signals (Appendix Fig S6C). Based on the close apposition of the mtDNA and VPS4 signals but lack of triple staining, a co‐localization in the same structure may be still possible, but is not sufficiently supported by data at this stage.

TLR9 mediates inflammation upon Opa1 deficiency

We next analyzed whether mtDNA co‐localizes with TLR9 in Opa1‐deficient muscle cells. In control cells, TLR9 was barely detectable, which was markedly enhanced in Opa1‐deficient cells (Appendix Fig S7A). In addition, Opa1 KD cells showed co‐localization of the mtDNA in TLR9 signals (Appendix Fig S7A). More than 50% of mtDNA signal co‐localized with TLR9 (Appendix Fig S7B). In addition, TLR9 did not co‐localize with lysosomes, when using LAMP1 as a marker, in control or in Opa1‐deficient cells (Appendix Fig S7C). Furthermore, TLR9 expression was increased in Opa1‐deficient muscles at various postnatal ages compared to controls (Fig 8A and Appendix Fig S7D). These data support the concept that TLR9 is activated upon Opa1 deficiency. Based on this, cells were treated with the TLR9 antagonist ligand ODN2088 or with the corresponding negative control. Incubation of Opa1 KD muscle cells with ODN2088 attenuated the expression of NF‐κB target genes, Myd88, S100a8, Hmgb1, or Rage (Fig 8B–E). In addition, the enhanced release of IL‐6 of Opa1‐deficient cells was inhibited by incubation with ODN2088 compared to the negative control (Fig 8F). Lastly, we analyzed whether similar effects were detected upon genetic depletion of TLR9. To this end, TLR9 was depleted in scramble or Opa1‐depleted muscle cells (Fig 8G and H). Depletion of TLR9 markedly reduced the induction of NF‐κB genes in Opa1‐deficient cells (Fig 8I), in agreement with the data obtained by the treatment with TLR9 antagonists. In all, our data are coherent with a TLR9 activation mediated by Opa1 deficiency.

Figure 8. TLR9 mediates inflammation upon Opa1 deficiency.

• A
  Quantification of TLR9 expression immunostaining in gastrocnemius muscle of Opa1 KO animals at 3, 7, 10, 13, and 20 weeks of age. This quantification was performed using Trainable Weka Segmentation plugin from ImageJ (five images per group). Represented values are relative to loxP group.
• B–E
  MyD88, S100A8, HMGB1, and RAGE expression in Scr and KD myoblasts treated with ODN2088 (ODN) or ODN2088 negative control (C) 1 μM (n = 9).
• F
  IL‐6 levels detected in cultured medium of myoblasts (Scr and KD) treated with ODN2088 (ODN) or ODN2088 negative control (C) (n = 9).
• G, H
  Expression of genes of TLR9 (G) and Opa1 (H) upon single or double knock‐down. n = 8 (G); n = 3 (H).
• I
  Expression of NF‐κB target genes in control (Scr), Tlr9 depleted (Tlr9 KD), Opa1 depleted (Opa1 KD), and double Tlr9‐Opa1 ablated (DKD) cells (n = 3).

Data information: Data represent mean ± SEM. *P < 0.05 when comparing the effect of Opa1 deficiency. ^# P < 0.05 when comparing the effects of ODN2088 (in panels B–F) or comparing to Opa1 KD cells (in panels G–I). Data were analyzed using analysis of variance with Tukey's post hoc test.

Discussion

In this study, we show that specific ablation of the mitochondrial fusion protein Opa1 in skeletal muscle results in whole body growth impairment, systemic inflammation, and premature death. The characterization of the Opa1 KO mice growth defects revealed low levels of circulating Igf1, high plasma growth hormone, and high Fgf21 expression in muscle as well as in circulation. We also found hepatic alterations in Opa1 KO mice such as enhanced Socs3 expression, reduced STAT5 phosphorylation, reduced gene expression of growth hormone receptor, and lower expression of growth hormone target genes. These data are consistent with the view that Opa1 deficiency in muscle causes global growth defects in conjunction with an altered Igf1 pathway (Wong et al, 2016). Given that high plasma concentrations of FGF21 have been observed to reduce growth hormone signaling, resulting in a marked growth defect (Inagaki et al, 2008), we examined whether the phenotype could be rescued by knocking out FGF21 expression globally. Analysis of double knockout mice proved that, although the hepatic expression of metabolic genes was restored, the growth defects due to Opa1 deficiency were independent of FGF21. Our observations also support the notion that muscle Opa1 exerts non‐cell‐autonomous actions on growth.

Inflammation has been reported to regulate growth (O'Connor et al, 2008). Growth inhibition is commonly observed in children with chronic inflammatory diseases such as juvenile idiopathic arthritis, chronic kidney disease, or chronic inflammatory bowel disease (Wong et al, 2016). In these patients, circulating concentrations of pro‐inflammatory cytokines such as IL‐1β and IL‐6 are elevated (Kutukculer et al, 1998; Ji et al, 2002). The use of mouse models has documented that overexpression of IL‐6 causes reduced circulating levels of Igf1 and reduced growth which is rescued by blocking anti‐IL‐6 antibodies (De Benedetti et al, 1997). In addition, TNF‐α, IL‐1β, and IL‐6 induce hepatic growth hormone resistance in mice (Zhao et al, 2014). In agreement with these data, here we report that Opa1 KO mice show systemic inflammation characterized by increased levels of IL‐1β and IL‐6. Furthermore, chronic treatment with salicylate resulted in substantial amelioration of muscle mass, body weight, muscle necrosis, and hepatic growth hormone resistance. Interestingly, anti‐inflammatory treatment also normalized FGF21 expression and its circulation levels, suggesting that the enhancement of the mitokine is downstream of inflammation. For now, we support the view that reduced growth in Opa1 KO mice is, at least in part, secondary to systemic inflammation.

In this study, we document the existence of a connection between Opa1 deficiency and muscle inflammation. Thus, we have demonstrated that skeletal muscle‐specific Opa1 ablation causes an inflammatory myopathy in mice, which is characterized by enhanced expression of NF‐κB target genes, and increased circulating levels of pro‐inflammatory cytokines IL‐1β and IL‐6. Inflammation was initially detected in 3‐week‐old Opa1 KO mice by enhanced IL‐1β abundance and normal CD68 labeling (a marker of macrophage infiltration), whereas at greater ages (13 weeks of age), CD68 labeling was enhanced, thereby indicating macrophage infiltration. In parallel to these data, tamoxifen‐induced Opa1 ablation revealed that the expression of some NF‐κB target genes such as NLRP3 and ASC was induced early under conditions in which muscle mass was normal. On the basis of these data, we conclude that Opa1 deficiency causes muscle inflammation which is independent of muscle mass loss, that occurs before macrophage infiltration, and that it is mediated via activation of NF‐κB. Cell culture studies in myoblasts also support the view that Opa1 deficiency causes a primary autonomous cellular response leading to inflammation. Opa1 deficiency in C2C12 muscle cells caused increased NF‐κB transcriptional activity, expression of NF‐κB target genes, and production of IL‐1β. This pattern of changes was not associated with cell death activation.

During the revision of this manuscript, two studies were published reporting the phenotype of muscle‐specific Opa1 depleted mice (Pereira et al, 2017; Tezze et al, 2017). In agreement with our data, they reported that ablation of Opa1 in skeletal muscle reduces muscle mass, and premature death. In addition, they also detected an enhanced muscle expression of Fgf21 in Opa1‐deficient muscles (Pereira et al, 2017; Tezze et al, 2017). We propose that the muscle inflammation driven by Opa1 deficiency is a primary event leading to the many alterations detected in this condition.

Dysregulation of mitochondrial dynamics has a major impact on mitochondrial function leading to a number of diseases (Liesa et al, 2009). OPA1 mutations causes an autosomal dominant optic atrophy (ADOA), and some OPA1 mutations lead to a multi‐systemic disorder called ADOA plus syndrome, which results in severe myopathy (Alexander et al, 2000; Delettre et al, 2000; Amati‐Bonneau et al, 2008; Hudson et al, 2008). Here, we report that Opa1 deficiency causes mitochondrial dysfunction both in skeletal muscle and in cultured muscle cells. This is characterized by mitochondrial fragmentation, reduced mitochondrial respiration, normal or enhanced expression of mitochondrial proteins, and unaltered expression of the mitochondrial biogenesis factor PGC1α.

Some OPA1 mutations are linked to mtDNA deletions, which has permit to propose a role of OPA1 on mtDNA stability (Amati‐Bonneau et al, 2008; Hudson et al, 2008). Here, we document that Opa1 deficiency was characterized by reduced mtDNA copy number, parallel to diminished TFAM protein abundance, reduced number of nucleoids, and increased nucleoid size. Opa1 deficiency has been reported to induce low mtDNA copy number and nucleoids abundance in human fibroblasts, which suggests that those alterations are not just restricted to muscle cells (Liao et al, 2017). Our data support the existence of mtDNA stress induced by Opa1 deficiency, which includes mtDNA loss and reduced nucleoids number.

Damaged mitochondria have been implicated in the induction of inflammation through the production of reactive oxygen species and the release of damage‐associated molecular patterns (DAMPs), namely formyl peptides and mtDNA (Zhang et al, 2010; Oka et al, 2012; Wenceslau et al, 2014). In this connection, we document that the inflammation detected in Opa1‐deficient muscle cells requires the presence of mtDNA so ethidium bromide‐treated Opa1‐deficient cells with more than 95% depletion of mtDNA, no longer show induction of NF‐κB target genes.

MtDNA stress triggered by TFAM deficiency has been reported to release mtDNA to the cytosol causing the activation of the DNA sensor cGAS, and to promote STING‐IRF3‐dependent signaling to induce interferon‐stimulated genes (West et al, 2015). However, here we report that Opa1 depletion causes mtDNA stress, but this occurs in the absence of leakage of mtDNA to the cytosol, and with no induction of interferon‐stimulated genes, and instead it causes induction of NF‐κB dependent on TLR9 activity.

A potential explanation for these different cellular responses under condition of mtDNA instability may lie on mitochondrial morphology. West et al (2015) showed a hyperfused phenotype of the mitochondrial network, and it has been reported the need for mitochondria to be fragmented prior to undergo mitophagy (Twig et al, 2008). Therefore, TFAM‐deficient mitochondria may not be included into autophagosomes, thus enabling mtDNA leakage to the cytosol. On the contrary, Opa1‐deficient mitochondria are completely fragmented which favors their entry into the mitophagic pathway, thus restraining the potential leakage to the inside of the vesicular system. Further studies regarding when and how mtDNA is released from mitochondria should be performed not only to understand whether there is an active form of transport of this mtDNA, but also to understand whether mtDNA has any control on the destination of the mito‐autophagosome. In this regard, it remains unclear the nature of the mechanisms by which Opa1 deficiency causes alterations in late stages of mitophagy, as well as in autophagy in muscle cells and whether this alteration also occurs in skeletal muscle. Studies need to be performed to understand whether this blockage of mitophagy completion is a consequence of general autophagy alterations, meaning that there are common players directly affected by Opa1 deficiency, or whether there are distinguishable defects for each pathway.

Based on the pattern of alterations detected, we speculate that in Opa1‐deficient cells, mitochondria‐containing autophagosomes are mainly diverted into the formation of amphisomes or amphisome‐like organelles, which is linked to the activation of TLR9. In this connection, further studies are required to determine whether Opa1 deficiency is linked to re‐routing of the mito‐autophagosome to the endosomal pathway, and if so, to address whether such a re‐routing may be an active programmed cause‐response mechanism or rather a consequence of the saturation of the lysosomal capacity due to the large amount of material to be degraded in an Opa1‐deficient context. The existence of communication mechanisms linking mitochondria and lysosomes may be responsible for the potential alterations in Opa1‐deficient cells. Thus, mitochondrial dysfunction has been reported to promote enhanced lysosomal biogenesis in some conditions, and reduced lysosomal function (Baixauli et al, 2015; Demers‐Lamarche et al, 2016; Fernandez‐Mosquera et al, 2017). Alternatively, Opa1 deficiency may regulate mTOR, which has been reported to regulate mitochondrial activity through translational control (Morita et al, 2013).

The existence of TLR9 activation is mainly suggested by the normalization of NF‐κB target genes expression upon TLR9 antagonist treatment or genetic depletion. The precise mechanism by how mtDNA may interact with TLR9 remains elusive, and our preliminary data indicate that mtDNA and TLR9 may be present in an intracellular compartment. However, a robust demonstration is lacking. We cannot exclude the possibility that Opa1 deficiency causes the activation of TLR9 by mtDNA through the formation of mitochondria‐derived vesicles, in spite of being reported under conditions of mitochondrial elongation (Neuspiel et al, 2008; McLelland et al, 2014, 2016).

In all, we document that Opa1 is required for mtDNA stability and mitochondrial quality control. As a result, Opa1 deficiency leads to the engagement of TLR9 resulting in the activation of NF‐κB inflammatory program. This is the mechanism by which Opa1 deficiency triggers muscle inflammation, which eventually becomes systemic, and leads to altered growth hormone/Igf1 axis, enhanced Fgf21 expression, and reduced growth. We propose that a deficient Opa1 activity may participate in the development of human inflammatory myopathies.

Materials and Methods

Refer to Appendix Supplementary Methods for a detailed description of all methods used.

Animal care

All animal experiments were done in compliance with the guidelines established by the Committee on Animal Care of the University of Barcelona. Mice were kept under a 12‐h dark‐light period and provided with a standard chow diet and water ad libitum. For detailed information of mouse strain generation, see Appendix Supplementary Methods.

Mice treatment

Sodium salicylate (200 mg/Kg, Sigma‐Aldrich) was administered daily by intraperitoneal injection for 30 days. Chloroquine (Sigma) was administered via i.p twice a day during 5 days at a concentration of 50 mg/kg into 3‐week‐old mice.

In vivo gene transfer and fluorescence microscopy of muscle sections

Left tibialis anterior was injected with the empty vector, while pDsRed2‐Mito (Clontech) was injected in the right muscle, and then muscles were electroporated. After 8 days, muscles were removed and fixed. 10‐μm cryosections were analyzed using a Leica TCS SP2 AOBS Systems confocal scanning microscope.

Histological analysis

For light microscopy, muscles were removed and embedded in OCT solution (TissueTek). Cryosections of 10 μm were stained with hematoxylin and eosin, Gomori's modified trichrome stain, or non‐specific esterase (Sigma) following standard protocols. Immunohistochemistry was performed with the indicated antibodies and labeling with the Vectastain ABC kit (Vector Laboratories), following the manufacturer's instructions.

Muscle regeneration studies

Regeneration of skeletal muscle was induced by intramuscular injection of 75 μl of 1 μM Cardiotoxin (CTX, Latoxan) in the gastrocnemius muscle (Kherif et al, 1999). Two days later, adenoviruses (microRNA Ctrl or microRNA Opa1) were injected in the region of the previous injury. Muscles were removed at 9, 12, and 22 days post‐injury. CSA was analyzed as described.

Respiration measurements in muscle

The respiration of permeabilized muscle was measured at 37°C by high‐resolution respirometry with an Oxygraph‐2k (Oroboros Instruments), as described (Kuznetsov et al, 2008).

Transmission electron microscopy

Gastrocnemius muscles were dissected, fixed, embedded, and then semi‐ and ultrathin cut and mounted, as described in the Appendix Supplementary Methods. JEM‐1010 electron microscope (Jeol, Japan) equipped with a CCD camera SIS Megaview III and the AnalySIS software were used for image acquisition.

Cell culture and stable knock‐down generation

Cells were grown in DMEM (Invitrogen) with 10% FBS and 100 U/ml of penicillin/streptomycin (Invitrogen) at 37°C in a humidified atmosphere of 5% CO₂/95% O₂. Lentiviral infection was used to deliver shRNA to C2C12 myoblasts using a procedure described in (Munoz & Zorzano, 2015).

Oxygen consumption measurements in C2C12 myoblasts

C2C12 myoblasts (Scr or Opa1 KD) were plated on SeaHorse Bioscience XF24 plates. After 48 h, oxygen consumption was measured using a Seahorse Bioscience XF24 extracellular flux analyzer. Extracellular acidification rate of C2C12 cells (ECAR) was measured in parallel. For detailed methods, see Appendix Supplementary Methods.

ATP content

ATP cellular content was measured using the ATP Determination Kit (Molecular Probes), following the manufacturer's instructions.

Immunofluorescence

Cells were fixed, permeabilized, and immunostained for the indicated components. For a detailed description of the protocol followed, see Appendix Supplementary Methods.

Mitochondrial fusion assays

One micro gram of mtPA‐GFP was transfected into C2C12 myoblasts (Scr or Opa1 KD) using Lipofectamine^® 2000 (Lovy et al, 2012). Experiments were performed 48 h after transfection, and the protocol was followed as described previously.

Mitophagy assessment

To analyze mitophagy initiation, Scr and Opa1 KD C2C12 cells were treated with 30 μM CCCP (Sigma) for 30 min. Mitochondrial fractions were examined for recruited LC3‐II protein. To analyze mitophagy resolution, cells were treated with 30 μM CCCP for 9 and 16 h, and then, together with control untreated cells, fixed and immunostained for dsDNA. MtDNA nucleoid number was quantified using ImageJ software.

Protein extraction and Western blotting

Mitochondrial enriched fractions, cytosolic fractions, total cellular homogenates, and tissue homogenates were used for Western blotting. For detailed methods and antibodies used, see Appendix Supplementary Methods.

Plasma and cell culture media measurements

Plasma concentrations of creatine kinase (Thermo Scientific), lactate dehydrogenase (LDH) (BioVision), IGF‐1, IL‐6, and IL‐1β (Abnova), and GH and FGF21 (Millipore) were measured following the manufacturer's instructions. Cell culture media was concentrated, and cytokine concentration was assessed in a similar way.

Statistics

The data presented here were analyzed using the Student's t‐test or analysis of variance (ANOVA) with an appropriate post hoc test. Data are presented as mean ± SEM unless stated otherwise. Significance was established at P < 0.05.

Author contributions

AR‐N and AD‐R performed cellular and animal experiments and wrote the manuscript, and contributed equally to this work; EN, FD‐S, XD, JPM, MR, NP, DS, JS, and JV performed some cellular or animal experiments; CT, VR, and MS performed some animal studies; ML and OS participated in some cell studies; EP performed statistical analyses; JG and MB performed histological analysis; FR, SRD, and FV contributed with materials and analysis tools; MP analyzed the experimental data; AZ directed the research, revised the experimental data, and wrote the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Dataset EV1

Dataset EV2

Movie EV1

Movie EV2

Review Process File

The EMBO Journal (2018) 37: e96553 29632021