PDK4 Augments ER-Mitochondria Contact to Dampen Skeletal Muscle Insulin Signaling During Obesity

Abstract

Mitochondria-associated endoplasmic reticulum membrane (MAM) is a structural link between mitochondria and endoplasmic reticulum (ER). MAM regulates Ca²⁺ transport from the ER to mitochondria via an IP3R1-GRP75-VDAC1 complex-dependent mechanism. Excessive MAM formation may cause mitochondrial Ca²⁺ overload and mitochondrial dysfunction. However, the exact implication of MAM formation in metabolic syndromes remains debatable. Here, we demonstrate that PDK4 interacts with and stabilizes the IP3R1-GRP75-VDAC1 complex at the MAM interface. Obesity-induced increase in PDK4 activity augments MAM formation and suppresses insulin signaling. Conversely, PDK4 inhibition dampens MAM formation and improves insulin signaling by preventing MAM-induced mitochondrial Ca²⁺ accumulation, mitochondrial dysfunction, and ER stress. Furthermore, Pdk4^-/- mice exhibit reduced MAM formation and are protected against diet-induced skeletal muscle insulin resistance. Finally, forced formation and stabilization of MAMs with synthetic ER-mitochondria linker prevented the beneficial effects of PDK4 deficiency on insulin signaling. Overall, our findings demonstrate a critical mediatory role of PDK4 in the development of skeletal muscle insulin resistance via enhancement of MAM formation.

Figure 1

PDK4 resides and interacts with the IP3R1-GRP75-VDAC1 complex at the MAM interface. A: Immunoblot analysis of subcellular fractions isolated from gastrocnemius muscle in mice. Markers for MAM and SR/ER (protein disulfide isomerase [PDI]), mitochondria (cytochrome c oxidase [COX IV]), nucleus (proliferating cell nuclear antigen [PCNA]), and cytosol (β-tubulin). B: In situ PLA in VXY control and PDK4-flag C2C12 myoblasts (scale bars, 10 μm). C: Coimmunoprecipitation (IP) analysis in gastrocnemius muscle homogenate in mice. D: IP analysis in VXY- (control) and PDK4-flag–expressing C2C12 myotubes. E: Schematic illustration depicting detection of protein–protein interaction by in situ PLA at the MAM interface using ER-targeted Sec61B-GFP and mitochondria-targeted mito-BFP. F: Confocal microscope imaging of in situ PLA blobs at the MAM interface (indicated by white arrows) in PDK4-flag–expressing C2C12 myoblasts (scale bars, 5 μm; inset, yellow scale bars, 1 μm). -ve Con, negative control.

Figure 2

Inhibition of PDK4 kinase activity suppresses MAM formation. A and B: In situ PLA in VXY- and PDK4-flag–expressing C2C12 myoblasts ± 4 mmol/L DCA for 6 h (respective panels are quantified below). Scale bars, 20 μm; mean ± SD of n = 3 of >30 cells/experiment (*P < 0.05; **P < 0.01; one-way ANOVA). C: Coimmunoprecipitation (IP) analysis in VXY- and PDK4-flag–expressing C2C12 myotubes ± 4 mmol/L DCA for 6 h. D: Input controls of C. E: Quantification of C (mean ± SD of n = 3) (*P < 0.05, **P < 0.01, Student t test). F: In situ PLA in C2C12 myoblasts that were transfected with either pcDNA or ΔPDK4-flag (respective panels are quantified below). Scale bars, 20 μm; mean ± SD of n = 3 of >30 cells/experiment. ns, not significant.

Figure 3

Obesity enhances MAM formation in skeletal muscle. A: Immunoblot analysis of gastrocnemius muscle isolated from CD- and HFD-fed mice (left) and quantification (mean ± SD of n = 6/group; **P < 0.01, Student t test). B: Immunoblot analysis of gastrocnemius muscle isolated from WT and ob/ob mice (left) and quantification (mean ± SD of n = 3/group; **P < 0.01, Student t test) (right). C: Representative immunoblot analysis of subcellular fractions isolated from gastrocnemius muscle of CD- or HFD-fed mice (left) and quantification of MAM fractions (mean ± SD of n = 3/group; *P < 0.05; **P < 0.01, Student t test) (right). D: In situ PLA in isolated myofibers from CD- and HFD-fed mice (respective panels are quantified below). Scale bars, 100 μm; mean ± SD of ∼10–15 myofibers/sample (*P < 0.05, **P < 0.01, Student t test). E: Immunoblot analysis of subcellular fractions isolated from gastrocnemius muscle of WT and ob/ob mice (top) and quantification of MAM fractions (representative of gastrocnemius skeletal muscle pooled from n = 6 mice/group; mean ± SD of n = 3) (*P < 0.05, **P < 0.01, Student t test) (bottom). F: In situ PLA in isolated myofibers from WT and ob/ob mice (respective panels are quantified below). Scale bars, 100 μm; mean ± SD of ∼10–20 myofibers/sample (**P < 0.01, Student t test). G: Coimmunoprecipitation (IP) analysis in PDK4-flag–overexpressing C2C12 myotubes treated with 0.4 mmol/L palmitate (Pal)/BSA for 16 h. H: Input control of G. I: Quantification of G (mean ± SD of n = 3; *P < 0.05, Student t test). -ve, negative control.

Figure 4

Genetic ablation of PDK4 reduced MAM formation in skeletal muscle of mice with diet-induced obesity. A: Immunoblot analysis of subcellular fractions isolated from CD- or HFD-fed WT and Pdk4^−/− mice. B: Quantification of proteins in MAM fraction of A (gastrocnemius muscle pooled from n = 6 mice/group, and two independent experiments were performed) (*P < 0.05, one-way ANOVA). C: In situ PLA in isolated myofibers from CD- or HFD-fed WT and Pdk4^−/− mice (scale bars, 100 μm) (left) and quantification of in situ PLA blobs (mean ± SD of ∼10–20 myofibers/sample) (**P < 0.01, one-way ANOVA) (right). D: Representative TEM images showing the association of SR/ER and IMF in gastrocnemius muscle isolated from CD- or HFD-fed WT and Pdk4^−/− mice (red arrows indicate the SR–mitochondria contact sites) (scale bars, 200 nm). Inset: yellow dotted line indicating the surface area of MAM (M, mitochondria; TT, T-tubule) (inset scale bars, 100 nm). E: Percentage of MAM surface area per mitochondrion perimeter in each microscopic field (n = 20/group; mean ± SD; ***P < 0.001, one-way ANOVA). F: Percentage of tight MAM (<15 nm) and loose MAM (15–30 nm) in total MAM surface area (n = 28–41/group; mean ± SD; *P < 0.05, **P < 0.01, #P < 0.05 control vs. WT HFD; &P < 0.05 WT HFD vs. Pdk4^−/− HFD, one-way ANOVA).

Figure 5

PDK4 inhibition attenuates MAM-mediated Ca²⁺ transfer from ER to mitochondria, mitochondrial dysfunction, and ER stress. A: Mitochondrial Ca²⁺ flux was measured using 4mitD3 probe (Ai). Basal [Ca²⁺]_mito (Aii) and [Ca²⁺]_mito after stimulation with 500 nmol/L ATP (Aiii) in C2C12 myoblast treated with 50 μmol/L palmitate (Pal)/BSA and ± 4 mmol/L DCA for 16 h (n = 9–12/group; mean ± SD; *P < 0.05, **P < 0.01, one-way ANOVA). B: ER lumen Ca²⁺ was measured using D1ER (Bi). Basal [Ca²⁺]_ER (Bii) and [Ca²⁺]_ER after stimulation with 1 µmol/L CPA (Biii) in C2C12 myoblast treated with 50 μmol/L Pal/BSA and ± 4 mmol/L DCA for 16 h (n = 9–12/group; mean ± SD). C: Cytosolic Ca²⁺ measurement using Fura-2 dye (Ci). Basal [Ca²⁺]_cyto (Cii) and [Ca²⁺]_cyto after stimulation with 1 µmol/L CPA (Ciii) in C2C12 myoblast treated with 50 μmol/L Pal/BSA and ± 4 mmol/L DCA for 16 h (n = 9–12/group; mean ± SD; **P < 0.01, one-way ANOVA). D: Evaluation of mitochondrial ROS generation after 16-h incubation with 0.4 mmol/L Pal/BSA in control siRNA (siCon)/siPDK4-transfected C2C12 myotubes using CM-H2XRos normalized by MitoTracker green FM (n = 3; mean ± SD; **P < 0.01, one-way ANOVA). E: MMP in siCon/siPDK4-transfected C2C12 myotubes using JC-1 dye by FACS. Graph indicates the JC-1 Red and Green ratio (n = 3; mean ± SD; *P < 0.05, **P < 0.01, one-way ANOVA). F: Cellular ATP level was measured in siCon/siPDK4-transfected C2C12 myotubes by ATPlite luminescence assay (n = 3; mean ± SD; *P < 0.05, **P < 0.01, one-way ANOVA). G: Quantitative RT-PCR measurement of mtDNA (mitochondrial ND1) normalized by gDNA (nuclear Pecam1) in gastrocnemius muscle of CD- or HFD-fed WT and Pdk4^−/− mice (n = 5/group; mean ± SD; **P < 0.01, one-way ANOVA). H: Representative immunoblot analysis of ER stress markers in gastrocnemius muscle tissue of CD- or HFD-fed WT and Pdk4^−/− mice. I: Quantification of H (mean ± SD of n = 4 to 5/group; *P < 0.05, **P < 0.01, one-way ANOVA). J: Representative immunoblot analysis of ER stress markers in siCon/siPDK4-transfected C2C12 myotubes. K: Quantification of J (mean ± SD of n = 4; *P < 0.05, **P < 0.01, ***P < 0.001, one-way ANOVA). p-, phosphorylated; Sec, seconds.

Figure 6

PDK4 deficiency attenuates palmitate (Pal)–induced skeletal muscle IR via downregulation of MAM. A: Immunoblot analysis in control siRNA (siCon)/siPDK4-transfected C2C12 myotubes (top) and quantification (mean ± SD of n = 3; *P < 0.05, **P < 0.01, one-way ANOVA) (bottom). B: Immunoblot analysis in siCon/siPDK4-transfected C2C12 myotubes following 100 nmol/L insulin treatment for 10 min (top) and quantification (mean ± SD of n = 3; *P < 0.05, one-way ANOVA) (bottom). C: Immunoblot analysis in gastrocnemius muscle homogenate isolated from CD- or HFD-fed WT or Pdk4^−/− mice (left) and quantification (mean ± SD of n = 3/group; *P < 0.05, **P < 0.01, one-way ANOVA) (right). D: Immunoblot analysis of insulin action in CD- or HFD-fed WT or Pdk4^−/− mice (top) and quantification (mean ± SD of n = 3 to 4/group; *P < 0.05, one-way ANOVA) (bottom). E: Graphical representation of synthetic ER–mitochondria linker: linker-RFP HA (Linker-RH)–mediated MAM induction. F: Targeting of Linker-RH at MAM interface was examined in C2C12 myoblast using mito-BFP (mitochondria) and Sec61B (ER) by confocal microscopy (scale bars, 20 μm). G: Pearson coefficient for colocalization between mito-BFP (mitochondria) and Sec61B (ER) in pcDNA- or Linker-RH–transfected C2C12 myoblast (mean ± SD of n = 8; **P < 0.01, Student t test). H: Basal and stimulated (with 500 nmol/L ATP) mitochondrial Ca²⁺ flux was measured in mock or Linker-RH–transduced C2C12 myoblast using 4mitD3 probe (n = 9–12/group; mean ± SD; *P < 0.05, **P < 0.01, Student t test). I: Immunoblot analysis after mock or Linker-RH transduction in WT or Pdk4^−/− primary myotubes (top) and quantification (mean ± SD of n = 3/group; *P < 0.05, **P < 0.01, Student t test) (bottom). J: AKT activation, following 10 min of 100 nmol/L insulin stimulation, was evaluated in mock or Linker-RH–transduced WT or Pdk4^−/− primary myotubes (top). Bottom: quantification (mean ± SD of n = 3/group; *P < 0.05, Student t test). K: JNK phosphorylation was evaluated after mock or Linker-RH transduction in DMSO or 20 μmol/L SP600125 treated for 6 h in WT primary myotubes (top). Bottom: quantification (mean ± SD of n = 3; *P < 0.05, **P < 0.01, Student t test). L: Insulin-induced AKT activation was evaluated in mock or Linker-RH–transduced WT primary myotubes treated with DMSO or 20 μmol/L SP600125 for 6 h (top). Bottom: quantification (mean ± SD of n = 3; *P < 0.05, Student t test). Sec, seconds.

Figure 7

A graphical representation demonstrating the role of PDK4 in obesity-induced MAM formation and skeletal muscle IR. In skeletal muscle, PDK4 interacts with and stabilizes the IP3R1-GRP75-VDAC1 complex at the MAM interface. In physiological conditions, PDK4 is expressed at low levels, and moderate Ca²⁺ transfer from ER to mitochondria via the IP3R1-GRP75-VDAC1 complex formation stimulates mitochondrial ATP production. However, obesity- or free fatty acid (FFA)–induced overexpression of PDK4 via Foxo1 activation augments MAM formation and promotes mitochondrial Ca²⁺ accumulation, mitochondrial dysfunction, and ER stress. Subsequently, ER stress-mediated JNK activation leads to inhibition of the insulin signaling pathway.