E4 ubiquitin ligase promotes mitofusin turnover and mitochondrial stress response

Abstract

Ubiquitin-dependent control of mitochondrial dynamics is important for protein quality and neuronal integrity. Mitofusins, mitochondrial fusion factors, can integrate cellular stress through their ubiquitylation, which is carried out by multiple E3 enzymes in response to many different stimuli. However, the molecular mechanisms that enable coordinated responses are largely unknown. Here we show that yeast Ufd2, a conserved ubiquitin chain-elongating E4 enzyme, is required for mitochondrial shape adjustments. Under various stresses, Ufd2 translocates to mitochondria and triggers mitofusin ubiquitylation. This elongates ubiquitin chains on mitofusin and promotes its proteasomal degradation, leading to mitochondrial fragmentation. Ufd2 and its human homologue UBE4B also target mitofusin mutants associated with Charcot-Marie-Tooth disease, a hereditary sensory and motor neuropathy characterized by progressive loss of the peripheral nerves. This underscores the pathophysiological importance of E4-mediated ubiquitylation in neurodegeneration. In summary, we identify E4-dependent mitochondrial stress adaptation by linking various metabolic processes to mitochondrial fusion and fission dynamics.

Keywords: CMT2A; Cdc48/p97; E4; Fzo1; MFN2; UBE4B; Ufd2; fusion; mitochondria; mitofusin; stress; ubiquitin.

Graphical abstract

Figure 1

Ufd2 controls mitofusin and mitochondrial fusion (A) Destabilization of Fzo1 upon CCCP treatment. Total cell extracts of WT cells after treatment with 10μM CCCP after 0, 10, 30, and 60 min were analyzed by sodium dodecyl sulfate- polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot using an Fzo1-specific antibody. (B) Stress-induced low Fzo1 levels depend on Ufd2. Protein steady state levels of WT or Δufd2 cells expressing hemagglutinin (HA)-Fzo1 upon treatment with 50μM CCCP, 1M NaCl, 5mM DTT, pH2.4, 0.2% oleate for 1h or 42°C for 20 min. Upper: total cell extracts, analyzed as in (A) but using an HA-specific antibody. Lower: quantification of Fzo1 levels from four independent experiments, depicted as mean (bars) and individual values (circles, squares, triangles, and inverted triangles), relative to the corresponding untreated samples. The Ponceau S (PoS) staining was used as a loading control. (C) Ufd2 facilitates mitochondrial fragmentation upon stress. Mitochondrial morphology of WT or Δufd2 cells upon treatment with 5 μM CCCP. Cells expressing mitochondrial-targeted mCherry were analyzed for mitochondrial (mCherry) and cellular (Nomarski) morphology by fluorescence microscopy (left). Quantification of three independent experiments with at least 200 cells each, showing the mean percentage of tubular mitochondria (bars) and individual values (circles, squares, and triangles) (upper middle). Automated quantification of the average mitochondrial size (upper right, mitochondrial aspect ratio (lower middle), and number of mitochondria (lower right) of at least 50 randomly selected cells, shown as Tukey boxplot. Scale bar: 5 μm. (D) Stress-induced fragmentation depends on the proteasome. Mitochondrial morphology of Δpdr5 cells upon treatment with 5 μM CCCP after pre-treatment with the proteasomal inhibitor MG132 for 1h, analyzed as in (C). Scale bar: 5 μm. All statistical significances were determined by ANOVA test. ns > 0.05, ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001.

Figure 2

Ufd2 ubiquitylates Fzo1 and targets it for proteasomal degradation (A) Higher molecular weight forms of Fzo1 in the cdc48-2 mutant background depend on Ufd2. Total cell extracts from the indicated strains expressing HA-Fzo1 were analyzed by SDS-PAGE and Western blot using an HA-specific antibody. Unmodified Fzo1, E3-dependent constitutive ubiquitin forms and E4-dependent degradative ubiquitin forms on Fzo1 are marked with an arrow, “E3-Ub” or “E4-Ub”, respectively. (B) Ufd2 specific ubiquitylation of HA-Fzo1 depends on SCFMdm30. cdc48-2Δpdr5 or cdc48-2Δpdr5Δmdm30 cells expressing genomically integrated HA-Fzo1 were treated with MG132 for 1h. Total cell extracts were analyzed as in (A). (C) Higher molecular weight forms of Fzo1 accumulate by proteasomal inhibition. The indicated strains expressing HA-Fzo1 were treated with 50μM MG132 or an equivalent amount of DMSO for 1h. Total cell extracts were analyzed as in (A) using HA- and ubiquitin (P4D1)-specific antibodies. (D) Higher molecular weight forms on Fzo1 represent ubiquitylation. Crude mitochondrial extracts were prepared from cdc48-2Δpdr5 cells expressing genomically integrated HA-Fzo1, previously treated with 50μM MG132 for 1h. Extracts were treated with the indicated concentrations of the purified catalytic domain of USP2 for 15 min at 37°C. Proteins were subsequently precipitated with trichloroacetic acid (TCA) and analyzed as in (A). (E) HA-Fzo1 levels in cdc48-2 depend on the catalytic activity of Ufd2. Total cell extracts from cdc48-2Δpdr5 or cdc48-2Δpdr5Δufd2 cells expressing genomically integrated HA-Fzo1 and the indicated variants of Ufd2-Myc were analyzed as in (A). (F) Ufd2P923A physically interacts with HA-Fzo1. Crude mitochondrial extracts from Δfzo1Δufd2 cells expressing HA-Fzo1, Ufd2P923A-Myc or the corresponding empty vectors, as indicated, were treated or not with DSP and subsequently solubilized, subjected to immunoprecipitation, and analyzed by SDS-PAGE and Western blot using HA- and Ufd2 specific antibodies.

Figure 3

Turnover of unstable Fzo1 variants requires Ufd2 (A) Ufd2 confers instability to Fzo1C805S. HA-tagged Fzo1 (black lines) or Fzo1C805S (red lines) were expressed in Δfzo1 (continuous lines) or Δfzo1Δufd2 (dotted lines) cells and the protein levels analyzed after treatment with the translation inhibitor cycloheximide (CHX). Total cell extracts were prepared 0, 1, or 3 h after synthesis shut off and analyzed by SDS-PAGE and Western blot using HA- and Ubc6-specific antibodies. Quantification of Fzo1 levels from three independent experiments is presented relative to corresponding untreated control labeled as “0”. Shown are mean and SEM of the individual time points. The PoS staining was used as a loading control. (B and C) Suppression of Fzo1C805S steady-state levels enables mitochondrial tubulation. Protein levels of Δfzo1 or Δfzo1Δufd2 cells expressing an empty vector, HA-Fzo1, or HA-Fzo1C805S were analyzed as in Figure 1B using an HA-specific antibody (in B). Mitochondrial morphology of the same cells expressing mitochondrial targeted GFP was analyzed as in Figure 1C, with at least 200 cells from three independent experiments (in C). (D) Ubiquitylation of HA-Fzo1C805S depends on Mdm30. The indicated cells were analyzed as in Figure 2A with at least 200 cells from three independent experiments. (E) Suppression of mitochondrial morphology of cells expressing HA-Fzo1C805S in absence of Ufd2 depends on SCFMdm30. Mitochondrial morphology of the indicated cells was analyzed as in (B), with at least 200 cells from three independent experiments. Scale bar: 5 μm. All statistical significances were determined by ANOVA test. ns > 0.05, ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001.

Figure 4

Non-functional Fzo1 and Ufd2 trap each other at mitochondria (A) Destabilization of Fzo1C805S depends on the catalytic activity of Ufd2 and its U-box functionality. HA-Fzo1C805S protein levels of WT or Δufd2 cells expressing the empty vector, Myc-tagged Ufd2, or Myc-tagged catalytic inactive (P923A) Ufd2 were analyzed as in Figure 1B using HA- and Myc-specific antibodies. (B) Mitochondrial fragmentation in cells expressing Fzo1C805S occurs independently of the catalytic activity of Ufd2. Mitochondrial morphology of Δfzo1 or Δfzo1Δufd2 cells expressing the indicated variants of HA-Fzo1 and Ufd2-Myc was analyzed as in Figure 1C with at 200 cells from three independent experiments. (C) Foci formation of Fzo1C805S-GFP in presence of catalytic-inactive Ufd2. Δfzo1Δdnm1Δufd2 cells expressing Fzo1-GFP or Fzo1C805S-GFP, Ufd2-Myc or Ufd2P923A-Myc and mitochondrial-targeted mCherry were analyzed using fluorescence microscopy. Shown are cellular morphology (Merge), mitochondrial morphology (Su9-mCherry), and Fzo1 localization (Fzo1-GFP). At least 100 cells were quantified per experiment. The number of Fzo1 foci, quantified in four independent experiments, is depicted as mean percentage (bars) and individual values (circles, squares, triangles and inverted triangles). Significant changes per “# of foci” category, in comparison to the control condition (WT Fzo1-GFP + WT Ufd2) of the same category, are represented. Scale bar: 5 μm. (D) Ufd2 P923A is recruited to Fzo1C805S foci. Δfzo1Δdnm1Δufd2 cells expression GFP-Fzo1 or GFP-Fzo1C805S, Ufd2P923A-mCherry and mitochondrial-targeted BFP were analyzed using fluorescence microscopy. Shown are the localization of Ufd2 P923A (Ufd2 P923A-mCherry), Fzo1 (GFP-Fzo1) or Fzo1C805S (GFP-Fzo1C805S), mitochondrial morphology (mtBFP), and cellular morphology (Nomarski). At least 200 cells were quantified per experiment. Colocalization of Ufd2 P923A with either the mitochondrial marker or with Fzo1 was quantified in three independent experiments using CellProfiler 4 77, and is depicted as mean (bars) and individual values (circles, squares, and triangles). Scale bar: 5 μm. All statistical significances were determined by ANOVA test. ns > 0.05, ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001.

Figure 5

Ufd2 controls the stability of CMT2A mutants (A) Ufd2 facilitates low levels of CMT2A mutant variants on Fzo1. Total cell extracts of WT or Δufd2 cells expressing the indicated variants of HA-Fzo1 were analyzed as in Figure 1B using an HA-specific antibody. Quantifications are shown as a relative to HA-Fzo1 in WT cells from three independent experiments and depicted as mean (bars) and individual values (circles, squares, and triangles). The PoS staining was used as a loading control. (B) Mitochondrial fragmentation of cells expressing CMT2A mutant variants depends on Ufd2. Mitochondrial morphology of Δfzo1 or Δfzo1Δufd2 cells expressing the indicated variants of HA-Fzo1 was analyzed as in Figure 1C with at 200 cells from three independent experiments. (C) Ufd2 dependent instability of Fzo1V183P requires the presence of WT Fzo1. Steady state levels of WT, Δufd2, Δfzo1, or Δfzo1Δufd2 cells expressing the indicated variants of HA-Fzo1 were analyzed as in (A) (upper). (D) Deletion of UFD2 increases respiratory growth of CMT2A mutant variants. Δfzo1 or Δfzo1Δufd2 cells expressing the indicated HA-Fzo1 variants after backcrossing with WT cells were spotted on fermentable (glucose) or non-fermentable (glycerol) carbon sources to evaluate their growth capacity. Quantification from four independent experiments of the growth on non-fermentable media normalized to growth on fermentable media, as a relative to the growth of Δfzo1 cells expressing the WT HA-Fzo1, depicted as mean (bars) and individual values (circles, squares, triangles, and inverted triangles). (E) Proteasomal inhibition prevents the degradation of Fzo1L302P and mitochondrial fragmentation. Δfzo1Δpdr5 cells expressing the indicated variants of HA-Fzo1 were treated with 50μM MG132 or Bortezomib (BTZ). Mitochondrial morphology (top) and protein steady-state levels (bottom) were analyzed as in Figures 1C and 1B, respectively. (F) Overview of Ufd2 sensitivity of CMT2A mutant variants of Fzo1. A total of 25 mutations were analyzed for protein stability in WT and Δufd2 cells and for mitochondrial morphology in Δfzo1 and Δfzo1Δufd2. Out of 12 mutant variants in the GTPase domain, 3 were Ufd2 sensitive for both protein levels and mitochondrial morphology. Out of 5 mutant variants in HB1, 1 was Ufd2-sensitive for protein levels, but not for mitochondrial morphology. Out of 6 mutant variants in HB2, 1 was Ufd2-sensitive for both protein levels and mitochondrial morphology. Out of 2 mutant variants in the TM, none were sensitive to Ufd2. All statistical significances were determined by ANOVA test. ns > 0.05, ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001.

Figure 6

MFN2 CMT2A variants are degraded dependent on UBE4B and the UPS (A) Stability of MFN2 WT and CMT2A variants. HEK293 cells transiently transfected with MFN2wtFlag or the indicated variants were left untreated (−) or treated for 3 h with CHX before cell harvesting. Total cellular extracts were analyzed by T/G-SDS-PAGE and immunoblotting using a Flag-specific antibody. The quantification of MFN2-variant levels below represents the mean from four independent experiments and the individual values of each experiment (circles, squares, triangles, and inverted triangles), normalized to the corresponding transiently transfected but untreated cells. PoS staining was used as a loading control. Statistical significance was determined by ANOVA test. (B) Proteasome sensitivity of MFN2 CMT2A variants. HEK293 cells transiently transfected with MFN2L92P-Flag, MFN2R104w-Flag or MFN2T706P-Flag, as indicated, were left untreated (−) or treated for 2 h with Epoxomicin before cell harvesting. Total cellular extracts were analyzed by T/G-SDS-PAGE and immunoblotting using an MFN2-specific antibody. The quantification of MFN2-variant levels below represents the mean from three independent experiments and the individual values of each experiment (circles, squares, and triangles), normalized to each transiently transfected untreated sample. PoS staining was used as a loading control. Statistical significance was determined by Student’s t-test. (C) Regulation of MFN2 CMT2A protein levels by UBE4B. HEK293 cells were treated for a total of 72h with scrambled (scr) or UBE4B siRNA (siUBE4B), as indicated. 48h after downregulation, cells were transiently transfected with MFN2L92P-Flag, MFN2R104w-Flag, or MFN2T706P-Flag and finally collected after 24h. Total cellular extracts were analyzed by T/G-SDS-PAGE and immunoblotting using a Flag-specific antibody. The quantification of MFN2-variant levels below represents the mean from at least four independent experiments and the individual values of each experiment (circles, squares, triangles, and inverted triangles), normalized to each transiently transfected scr-treated sample. PoS staining was used as a loading control. Statistical significance was determined by ANOVA test. (D) Mutation of lysine 532 blocks the degradation of MFN2T706P. HEK293 cells transiently transfected with MFN2T706P-3xFlag and MFN2K532R; T706P-3xFlag were analyzed and treated as in (A). The quantification of MFN2-variant levels was relative to the corresponding transiently transfected untreated cells. Four independent experiments were performed and the mean-MFN2 level (in bars) and individual values of each experiment (circles, squares, triangles, and inverted triangles) thereof is depicted. Statistical significance was determined by ANOVA test. ns > 0.05, ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001.

Figure 7

Model of mitofusin regulation by E4 ubiquitin ligases External stresses, including mitochondrial depolarization and CMT2A related mutations, trigger proteasomal turnover of mitofusins. This stress response requires the E4 ubiquitin ligase Ufd2/UBE4B, which extends ubiquitin chains previously added by E3 ubiquitin ligases. Degradation of mitofusins prevents their fusion activity and results in mitochondrial fragmentation by ongoing fission events. This impairs mitochondrial functions and may be at the base of MFN2-linked diseases. Inhibition of the E4 UBE4B or of the proteasome might rescue mitochondrial performance and provide a study target for therapeutic intervention.