Noncanonical PDK4 action alters mitochondrial dynamics to affect the cellular respiratory status

Abstract

Dynamic regulation of mitochondrial morphology provides cells with the flexibility required to adapt and respond to electron transport chain (ETC) toxins and mitochondrial DNA-linked disease mutations, yet the mechanisms underpinning the regulation of mitochondrial dynamics machinery by these stimuli is poorly understood. Here, we show that pyruvate dehydrogenase kinase 4 (PDK4) is genetically required for cells to undergo rapid mitochondrial fragmentation when challenged with ETC toxins. Moreover, PDK4 overexpression was sufficient to promote mitochondrial fission even in the absence of mitochondrial stress. Importantly, we observed that the PDK4-mediated regulation of mitochondrial fission was independent of its canonical function, i.e., inhibitory phosphorylation of the pyruvate dehydrogenase complex (PDC). Phosphoproteomic screen for PDK4 substrates, followed by nonphosphorylatable and phosphomimetic mutations of the PDK4 site revealed cytoplasmic GTPase, Septin 2 (SEPT2), as the key effector molecule that acts as a receptor for DRP1 in the outer mitochondrial membrane to promote mitochondrial fission. Conversely, inhibition of the PDK4-SEPT2 axis could restore the balance in mitochondrial dynamics and reinvigorates cellular respiration in mitochondrial fusion factor, mitofusin 2-deficient cells. Furthermore, PDK4-mediated mitochondrial reshaping limits mitochondrial bioenergetics and supports cancer cell growth. Our results identify the PDK4-SEPT2-DRP1 axis as a regulator of mitochondrial function at the interface between cellular bioenergetics and mitochondrial dynamics.

Keywords: OCR; dynamin-related protein 1; mitochondrial fission; pyruvate dehydrogenase kinase 4; septin 2.

Fig. 1.

PDK4 mediates energy stress-induced mitochondrial fission. (A) Representative confocal images of the mitochondrial morphology of wild-type (Pdk4^+/+) or Pdk4 KO (Pdk4^−/−) MEFs, transduced with an adenovirus overexpressing PDK4 or mock control, and treated for 1 h with vehicle (DMSO), 1 μM of rotenone (Rot), or 10 μM antimycin A (AMA). Mitochondria and nucleus were visualized using an antibody to TOM20 and DAPI, respectively. (Scale bars, 15 μm.) (B) Quantification of the mitochondrial length of the cells shown in (A), >200 cells counted for each condition. (C) Time-lapse confocal images of Pdk4^+/+ or Pdk4^−/− MEFs stained with MitoTracker Green and NucBlue. The indicated treatment was started at 0 min. A magnification of a portion of the mitochondrial network (dotted square) is included for each image. (Scale bars, 20 μm.) (D) OCR of Pdk4^+/+, Pdk4^−/−, and Pdk4^−/− MEFs transduced with an adenovirus overexpressing PDK4 or mock control. (E) Representative confocal images of the mitochondrial morphology of wild-type (sgCtrl) or Pdha1 KO (sgPdha1) MEFs treated with DMSO or 1 μM Rot for 1 h. Mitochondria were stained with MitoTracker deep red. (Scale bars, 20 μm.) (F) Quantification of the mitochondrial length of the cells shown in (E), >100 cells counted for each condition. (G) Immunoblot analysis of indicated proteins in the subcellular fractions (Cyto: cytosolic fraction; Mito: mitochondrial fraction) of Pdk4^+/+ or Pdk4^−/− MEFs treated with DMSO or 1 μM Rot for 1 h. Data are shown as the mean ± SEM of at least three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001, ^##P < 0.01; ^####P < 0.0001 (DMSO treated Pdk4^+/+ vs. Rot/AMA treated samples) by two-way ANOVA using Tukey’s multiple comparison test.

Fig. 2.

PDK4 phosphorylates SEPT2 and enhances SEPT2-DRP1 interaction. (A) Schematic depiction of SILAC-LC-MS/MS analysis of C2C12 myoblasts transiently transduced with adenovirus (Ad)-mock or PDK4 as indicated. (B) Venn diagram depicting down-, unchanged- and up-regulated proteins in the mitochondrial fraction of the cells in (A). (C) Heat map showing the abundance of phosphorylated proteins in the mitochondrial fraction of the cells in (A). (D) Schematic illustration of SEPT2 domain structure and location of the PDK4 substrate motif, serine 218 (S218) relative to functionally annotated GTP-binding domain at amino terminus. (E) ClustalW alignment of SEPT2 amino acid sequences showing that S218 site is conserved among the mammalian species. (F) Coimmunoprecipitation of SEPT2 with PDK4 in Pdk4^+/+ MEFs. PDK4 immunoprecipitates and lysate was immunoblotted with the indicated antibodies. (G) Coimmunoprecipitation of SEPT2 with PDK4 in Pdk4^+/+ and, Pdk4^−/− MEFs transduced with adenovirus-mock or PDK4. Phospho-serine/threonine (pS/T) immunoprecipitates and lysates were immunoblotted with the SEPT2 antibody. (H) Schematic of SEPT2 cDNA illustrating the generation of SEPT2 S218A (phospho-mutant) and S218D (phospho-mimetic) mutant construct at the location of serine 218. (I) AD-293 cells were cotransfected with HA-tagged WT or SEPT2 S218A mutant and empty vector (−) or PDK4-FLAG (+). HA/pS/T immunoprecipitations and total cell lysates were immunoblotted as indicated. (J) Immunoblot analysis of in vitro kinase assay of recombinant PDK4 protein and SEPT2 WT or S128A Immunoprecipitates. (K) AD-293 cells cotransfected with HA-tagged WT or SEPT2 S218A mutant and empty vector (−) or PDK4-FLAG (+). HA immunoprecipitates and lysates were immunoblotted with the indicated antibodies and phosphomotif antibody that recognizes phosphorylated SEPT2 at Ser218. (L) Immunoblot analysis of indicated proteins in Pdk4^+/+ and, Pdk4^−/− MEFs transduced with adenovirus-mock or PDK4. Phosphorylation of endogenous SEPT2 was detected by the antibody specific to SEPT2 Ser218. (M) Protein-protein interactions between SEPT2 and PDK4/DRP1 were visualized by Proximity Ligation Assay (PLA) in AD-293 cells transiently transfected with PDK4-FLAG (Left) or endogenous DRP1 (Right) using anti-SEPT2 and anti-FLAG or anti-DRP1 antibodies. Predominantly, PLA blobs were observed on the mitochondrial fission sites (arrows). Magnified portion of the mitochondrial network (dotted square) is included for each image. Manders’ coefficient (M) indicates fraction of PLA blobs overlapping mito-GFP. (Scale bars, 10 μm.).

Fig. 3.

PDK4-dependent SEPT2 phosphorylation is critical for DRP1 recruitment to the mitochondria. (A) Immunoblot analysis of indicated proteins in WT or QKO (Mff^−/−Fis1^−/−Mid49/51^−/−) MEFs. (B) Representative confocal images of the mitochondrial morphology of WT and QKO MEFs, transduced with an adenovirus overexpressing PDK4 or mock control, after transient knockdown of Septin2 using siRNA (siSept2). Mitochondria and nucleus were visualized using an antibody to TOM20 and DAPI, respectively. (Scale bars, 20 μm.) (C) Quantification of the mitochondrial length of the cells shown in (B). (D) Immunoblot analysis of indicated proteins in the subcellular fractions (WCL: whole cell lysate; Mito: mitochondrial fraction) of WT or QKO MEFs, transduced with adenovirus-mock (−) or PDK4 (+), after transient knockdown of Septin2 using siRNA (siSept2) (+) or siCtrl (−). (E) DRP1 and exogenous SEPT2 interaction were visualized by PLA using anti-HA and anti-DRP1 antibodies in AD-293 cells ectopically expressing HA-tagged WT, SEPT2 S218A, or SEPT2 S218D mutants and PDK4-FLAG together with a plasmid encoding for mitochondrial-targeted GFP (mito-GFP). (Scale bars, 10 μm.) (F) Quantification of the PLA blobs of the cells shown in (E). (G) Quantification of the mitochondrial length of the cells shown in (E). (H) DRP1 recruitment to the mitochondria were visualized by immunofluorescence (IF) using anti-DRP1 antibody in AD-293 cells ectopically expressing HA-tagged WT, SEPT2 S218A, or SEPT2 S218D mutants and PDK4-FLAG together with a plasmid encoding for mitochondrial-targeted GFP (mito-GFP). (Scale bars, 15 μm.) (I) Quantification of the colocalization between mito-GFP and DRP1 of the cells shown in (H). (J) Immunoblot analysis of indicated proteins in the subcellular fractions (WCL: whole cell lysate; Mito: mitochondrial fraction) of AD-293 cells transiently transfected with HA-tagged WT, SEPT2 S218A, or SEPT2 S218D mutants and PDK4-FLAG as indicated. Data are shown as the mean ± SEM of at least three independent experiments with >200 cells counted for each condition. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by ordinary one-way or two-way ANOVA using Tukey’s multiple comparison test.

Fig. 4.

Inactivation of PDK4-SEPT2 axis corrects mitochondrial dynamics and function in Mfn2 deficient cells. (A) Immunoblot analysis of indicated proteins in wild-type (Mfn2^+/+) or Mfn2 KO (Mfn2^−/−) MEFs. (B) Representative confocal images of the mitochondrial morphology of Mfn2^+/+ or Mfn2^−/−MEFs. Mitochondria were visualized using an antibody to TOM20. (Scale bars, 20 μm.) (C) Quantification of the mitochondrial length of the cells shown in (B). (D) OCR of the cells shown in (B). (E) Representative confocal images of the mitochondrial morphology of Mfn2^−/− MEFs transiently transfected with control (Ctrl), Pdk4 or Sept2 siRNAs for 48 h. Mitochondria were visualized using an antibody to TOM20. (Scale bars, 20 μm.) (F) Quantification of the mitochondrial length of the cells shown in (E). (G) OCR of the cells shown in (E). Data are shown as the mean ± SEM of at least three independent experiments with >200 cells counted for each condition. **P < 0.01; **P < 0.01; ***P < 0.001; ****P < 0.0001 by two-way ANOVA using Tukey’s/Sidak’s multiple comparison test.

Fig. 5.

Inhibition of PDK4-SEPT2 pathway limits cancer cell growth. (A) Immunoblot analysis of indicated proteins in tumor (T) and corresponding adjacent nontumor (N) tissue lysates from patients with lung adenocarcinoma. (Left) Immunoblot quantification of PDK4 and pS218 were normalized with ACTB and SEPT2, respectively. *P < 0.05; **P < 0.01 by paired t test. (B) Representative transmission electron microscopic (TEM) images and analysis of mitochondrial morphology of tissue samples (patient 1 and 2) shown in (A). (Scale bars, 1 μm.) (C) Analysis of mitochondrial morphology of the TEM images shown in (B). (D) Immunoblot analysis of indicated proteins in H460 and Calu-3 cells transiently transfected with Pdk4 siRNA for 48 h. (E) PLA of indicated proteins in H460 and Calu-3 cells transiently transfected with Pdk4 siRNA for 48 h using anti-SEPT2 and anti-DRP1antibodies. (Scale bars, 15 μm.) (F) Quantification of the PLA blobs of the cells shown in (E). (G) Representative confocal images of the mitochondrial morphology of H460 and Calu-3 cells transiently transfected with Pdk4 siRNA for 48 h. Mitochondria were visualized using an antibody to TOM20. (Scale bars, 15 μm.) (H) Quantification of the mitochondrial length of the cells shown in (G). (I and J) Extracellular acidification rate (ECAR) measurement of H460 (I) and Calu-3 (J) cells transiently transfected with Pdk4 siRNA for 48 h. (K) Representative images of spheroid cultures of H460 and Calu-3 cells transiently transfected with Pdk4 siRNA after 72 h. (Scale bars, 500 μm.) (L) Quantification of the spheroid surface area (in %) of the cells shown in (K). (M) Molecular model depicting the role of PDK4-SEPT2-DRP1 axis in regulating mitochondrial dynamics in lung tumor. Data are shown as the mean ± SEM of at least three independent experiments with >200 cells counted for each condition. Data are shown as the mean ± SEM of at least three independent experiments with >200 cells counted for each condition. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by unpaired t test or two-way ANOVA using Sidak’s multiple comparison test.