Disruption of mitochondrial dynamics triggers muscle inflammation through interorganellar contacts and mitochondrial DNA mislocation

Abstract

Some forms of mitochondrial dysfunction induce sterile inflammation through mitochondrial DNA recognition by intracellular DNA sensors. However, the involvement of mitochondrial dynamics in mitigating such processes and their impact on muscle fitness remain unaddressed. Here we report that opposite mitochondrial morphologies induce distinct inflammatory signatures, caused by differential activation of DNA sensors TLR9 or cGAS. In the context of mitochondrial fragmentation, we demonstrate that mitochondria-endosome contacts mediated by the endosomal protein Rab5C are required in TLR9 activation in cells. Skeletal muscle mitochondrial fragmentation promotes TLR9-dependent inflammation, muscle atrophy, reduced physical performance and enhanced IL6 response to exercise, which improved upon chronic anti-inflammatory treatment. Taken together, our data demonstrate that mitochondrial dynamics is key in preventing sterile inflammatory responses, which precede the development of muscle atrophy and impaired physical performance. Thus, we propose the targeting of mitochondrial dynamics as an approach to treating disorders characterized by chronic inflammation and mitochondrial dysfunction.

Fig. 1. Opposite mitochondrial morphologies result in mtDNA-dependent sterile inflammation associated with distinct mtDNA intracellular location.

a NFκB target and type I IFN response gene expression in myoblasts with fragmented mitochondria (n = 7). b IL1β and IFNβ levels in cultured media from myoblasts with fragmented mitochondria (n = 6). c NFκB target and type I IFN response gene expression in myoblasts with elongated mitochondria (n = 7). d IL1β and IFNβ levels in cultured media from myoblasts with elongated mitochondria (n = 6). NFκB target gene expression upon mtDNA depletion in myoblasts with e fragmented and f elongated mitochondria (n = 6). g Representative immunoblot of LAMP1, TIMM23 and Tubulin in subcellular fractionations (n = 3). Expression of mtDNA-encoded genes relative to nuclear-encoded gene (bActin) in cytosolic fractions of myoblast with h fragmented (n = 7) or i elongated (n = 6) mitochondria. a–d, i, h Two-sided Students’ t test per gene. e, f Two-way ANOVA test and post hoc t tests. Data are expressed as the mean of n independent experiments ± SEM.*p vs. Scr <0.05. ^#p vs cognate KD - ddC <0.05. a, c Created with BioRender.com. a–i Source data is provided in the Source Data File.

Fig. 2. Increased co-distribution between mtDNA and DNA sensors is associated with inflammation upon altered mitochondrial dynamics.

a Representative immunostainings of dsDNA (green) with subtraction of the nuclear signal (mtDNA), and TLR9 (red) in Scr, Mfn1KD, Mfn2KD, Fis1KD, and Drp1KD myoblasts. Scale bar 10 µm in MERGE and 3 µm in INSET. b Quantification of the Pearson’s correlation between mtDNA and TLR9 (n = 20 images per condition). c Representative immunostainings of dsDNA (green) with subtraction of the nuclear signal (mtDNA), and cGAS (red) in Scr, Mfn1KD, Mfn2KD, Fis1KD, and Drp1KD muscle cells. Scale bar 10 µm in MERGE and 3 µm in INSET. d Quantification of Pearson’s correlation between mtDNA and cGAS. (n = 20 images per condition). a, c White arrows point positive co-distribution. b, d Two-way ANOVA test and post-hoc t tests. Data are expressed as mean ± SEM. *p vs. Scr <0.05. b, d Created with BioRender.com. b, d Source data is provided in the Source Data File.

Fig. 3. Blockade of DNA sensors rescues the inflammatory profile upon altered mitochondrial dynamics.

a Inflammatory profile in Mfn1KD myoblasts upon ODN2088 or Ru.521 treatment (1 µM, 24 h) (n = 4). b Inflammatory profile in Mfn2KD myoblasts upon ODN2088 or Ru.521 treatment (1 µM, 24 h) (n = 4). c Inflammatory profile in Fis1KD myoblasts upon ODN2088 or Ru.521 treatment (1 µM, 24 h) (n = 8). d Inflammatory profile in Drp1KD myoblasts upon ODN2088 or Ru.521 treatment (1 µM, 24 h) (n = 7). e Inflammatory profile upon Vdac1, Bax, Ppid, or DNaseIIa acute downregulation in Scr and Mfn1KD myoblasts (n = 4). a–e Two-way ANOVA test and post-hoc t tests. Data are expressed as mean of n independent experiments ± SEM.*p vs. Scr + vehicle <0.05 and ^#p vs. cognate KD + vehicle <0.05 in a–d; *p vs. Scr + siCtrl <0.05 and ^#p vs. cognate KD + siCtrl <0.05 in e. a–e Source data is provided in the Source Data File.

Fig. 4. Mfn1 deficiency causes preferential location of mtDNA and mitochondria in early endosomes and promotes close contacts between Rab5⁺ early endosomes and mitochondria.

a Quantification of Pearson’s correlation between mtDNA and endosomal markers (n = 20 images per condition, except for mtDNA-LAMP1, where n = 39). Representative 3D reconstructions of immunostainings using dsDNA with nuclear subtraction (mtDNA, orange), mitochondria with Mitotracker Deep Red (turquoise) and b Rab5 (white) or c TLR9 (white) (Scale bar,10 µm). d Representative images of immunogold staining of Rab5 (gold particle 18 nm) and SdhA (gold particle 12 nm) in Scr- and Mfn1KD myoblasts (Scale bar, 200 nm). e Quantification of the distance between marked early endosomes and mitochondria in Scr (n = 38 contacts) and Mfn1KD myoblasts (n = 31 contacts). f Percentage of measured contacts <30 nm in Scr- and Mfn1-deficient muscle cells (n = 3, each point represents the mean of quantifications obtained in three independent experiments). b, c Arrows point to positive co-distribution. d Arrows point gold particles. a, e, f Two-sided Students’ t test. Data are expressed as the mean of n independent experiments ± SEM.*p vs. Scr <0.05. a, e, f Source data is provided in the Source Data File.

Fig. 5. Interaction between Mfn2 and Rab5C promotes mitochondria-early endosomal contacts upon Mfn1 deficiency, driving mtDNA and TLR9-dependent NFκB-mediated inflammation.

a Workflow of the generation of the MFN1-HA HeLa cell line, HA immunoprecipitation, and mass spectrometry analysis. b Graphical representation of MFN1 interacting candidates organized according to their fold change (FC) and Bayesian False Discovery Rate (BFDR). Identified in purple are the endosomal interactors of MFN1, and in green a known MFN1 interactor. c MFN1 and MFN2 representative immunoblots of immunoprecipitation (IP) in WT HeLa cells overexpressing FLAG-RAB5C (n = 6). d Mfn2 and Rab5C representative immunoblots of input, flowthrough (FT), and eluate (IP) fractions of FLAG immunoprecipitation in Scr and Mfn1KD muscle cells overexpressing FLAG-RAB5C (n = 5). e Quantification of the Mfn2/Rab5C ratio in the IP fraction (n = 5). (Representative immunostainings of dsDNA (green) with subtraction of the nuclear signal (mtDNA), and TLR9 (red) (Scale bar 10 µm in MERGE and 3 µm in INSET) upon siCtrl, siRab5C or siMfn2 transfection in Scr and Mfn1KD cells. White arrows point positive co-distribution. g Quantification of Pearson’s correlation between mtDNA and TLR9 or EEA1 upon siCtrl, siRab5C, or siMfn2 transfection in Scr and Mfn1KD cells (n = 20 images per condition). h NFκB target gene expression upon siCtrl, siRab5C, or siMfn2 transfection in Scr and Mfn1KD cells (n = 5). e Two-sided Students’ T test, g, h Two-way ANOVA test and post-hoc t tests. Data are expressed as mean of n independent experiments ± SEM.*p vs. Scr <0.05 in e; *p vs. Scr + siCtrl <0.05 and ^#p vs. Mfn1KD + siCtrl <0.05 in g, h. a Created with BioRender.com. c–e, g, h Source data is provided in the Source Data File.

Fig. 6. Specific ablation of Mfn1 in skeletal muscle promotes TLR9-dependent NFκB-dependent inflammation, muscle atrophy, reduced physical performance, and enhanced IL6 response to exercise.

a Mfn1 and Tubulin representative immunoblot in total homogenates of quadriceps muscles of LoxP (n = 4 mice) and SkM-Mfn1KO (n = 4 mice) male and female mice. b TEM images from cross-sectional sections of quadriceps muscles of LoxP and SkM-Mfn1KO male mice (n = 4 mice) and quantification of parameters representing mitochondrial morphology (Scale bar, 500 nm). c NFκB target and type I IFN response gene expression in quadriceps muscles of LoxP (n = 8 mice) and SkM-Mfn1KO mice (n = 14 mice). d NFκB target gene expression in quadriceps muscles of endotoxin-free water- or ODN2088-treated (100 µg/mice, necropsy 48 h after injection) LoxP (n = 4 mice) or SkM-Mfn1KO male mice (n = 6 mice). e Representative images of hematoxylin/eosin staining of cross-sectional sections of gastrocnemius muscles (Scale bar, 100 µm) and quantification of the CSA (n = 3 mice; 4 areas per mice; 30 fibers per area). f CK activity in plasma samples (n = 16 mice). g Fgf21 mRNA levels in quadriceps muscles of LoxP (n = 6 mice) and SkM-Mfn1KO male mice (n = 5 mice). h Plasma FGF21 levels in LoxP (n = 7 mice) and SkM-Mfn1KO male mice (n = 6 mice). i Distance run on the treadmill test and the difference in the distance run between day 2 and day 1 in LoxP (n = 12 mice) and SkM-Mfn1KO male mice (n = 11 mice). j Il6 expression levels in quadriceps muscles and k plasma IL6 levels of LoxP and SkM-Mfn1KO male mice at resting conditions or at different time-points after the treadmill test (n = 14 mice). b, e–k Two-sided Students’ T test, c Two-sided Students’ T-test per gene, and d two-way ANOVA test and post hoc t tests. Data are expressed as mean ± SEM. *p vs. LoxP <0.05. a–k Source data is provided in the Source Data File.

Fig. 7. Chronic anti-inflammatory treatment normalizes muscle inflammation, rescues muscle atrophy, ameliorates physical performance, and improves exercise-induced IL6 response.

a NFκB target gene expression in quadriceps muscles of PBS- or salicylate-treated (200 mg/kg, 28 days) LoxP (n = 4 mice) and SkM-Mfn1KO male mice (n = 5 mice). b Representative images of hematoxylin/eosin staining of cross-sectional sections of gastrocnemius muscles (Scale bar, 100 µm) and quantification of the CSA from PBS- or salicylate-treated mice (n = 3 mice; 4 areas per mice; 30 fibers per area). c CK activity in plasma samples in LoxP (n = 12 mice) and SkM-Mfn1KO PBS- or salicylate-treated male mice (n = 12 mice). d Fgf21 mRNA levels in quadriceps muscles of LoxP (n = 5 mice) and SkM-Mfn1KO PBS- or salicylate-treated male mice (n = 7 mice). e Plasma FGF21 levels in LoxP (n = 7 mice) and SkM-Mfn1KO PBS- or salicylate-treated male mice (n = 9 mice). f Distance run on the treadmill test and the difference in the distance run between day 2 and day 1 in LoxP (n = 7 mice) and SkM-Mfn1KO PBS- or salicylate-treated male mice (n = 7 mice). g Plasma IL6 levels of LoxP and SkM-Mfn1KO PBS- or salicylate-treated male mice at resting conditions or at different time-points after the treadmill test (n = 8 mice). a–g Two-way ANOVA test and post hoc t tests. Data are expressed as mean ± SEM. *p vs. LoxP + PBS < 0.05, ^#p vs. SkM-Mfn1KO + PBS < 0.05. a–g Source data is provided in the Source Data File.