Mitofusin 2 displays fusion-independent roles in proteostasis surveillance

Abstract

Mitochondria are essential organelles and their functional state dictates cellular proteostasis. However, little is known about the molecular gatekeepers involved, especially in absence of external stress. Here we identify a role of MFN2 in quality control independent of its function in organellar shape remodeling. MFN2 ablation alters the cellular proteome, marked for example by decreased levels of the import machinery and accumulation of the kinase PINK1. Moreover, MFN2 interacts with the proteasome and cytosolic chaperones, thereby preventing aggregation of newly translated proteins. Similarly to MFN2-KO cells, patient fibroblasts with MFN2-disease variants recapitulate excessive protein aggregation defects. Restoring MFN2 levels re-establishes proteostasis in MFN2-KO cells and rescues fusion defects of MFN1-KO cells. In contrast, MFN1 loss or mitochondrial shape alterations do not alter protein aggregation, consistent with a fusion-independent role of MFN2 in cellular homeostasis. In sum, our findings open new possibilities for therapeutic strategies by modulation of MFN2 levels.

Fig. 1. MFN2 controls mitochondrial mass and composition.

a Confocal images after immunostaining with the inner and outer mitochondrial membrane proteins ATP5β (in magenta) and TOM20 (in cyan), respectively, of HEK WT or corresponding knockout cells for MFN1 (1KO) or MFN2 (2KO). Scale bar: 5 µm. Insets of white dotted boxes are shown on the right side of each image. b Principal Component Analysis (PCA) of proteomes of HEK WT, 1KO, 2KO and 2KO cells stably expressing MFN2^FLAG (2 + 2) (n = 4 biological replicates). Groupings are color-coded and the explained variance is indicated in brackets. c Boxplot analysis showing the log2 fold change distribution of 1KO, 2KO and 2 + 2 versus WT cells (n = 4 biological replicates) for MitoCarta 3.0 (in red) and non-MitoCarta 3.0 (in gray) protein groups. Boxes borders indicate the 25% and 75% quantiles and outliers are indicated by a + sign (greater distance than 1.5 * inter quantile range). The minimum and maximum values are shown by the whiskers excluding outliers. d Volcano plot of the log2 fold change between 2KO versus WT cells (n = 4 biological replicates). The significantly different (Two-sided unpaired t-test followed by a permutation-based FDR correction FDR < 0.05, s0 = 0.1, #permutations = 500) protein groups of different MitoCarta 3.0 pathways are highlighted by color. e Bar graph displaying the result of the 1D Enrichment from the MitoCarta 3.0 based normalization data showing the median log2 fold change (all protein groups annotated by the given pathway) of 2KO versus WT cells (n = 4 biological replicates). The number of proteins identified within the total number of proteins in each pathway are annotated next to respective bar. FDR < 0.02. Source data are provided as Source Data file.

Fig. 2. MFN2 loss impairs protein import, increasing PINK1 levels.

a Western blot analysis of total cell lysates from HEK WT, 2KO and 2 + 2 cells, immunoblotted with anti-TOM20. Staining of total protein with PoS was used as loading control. Bars represent the mean fold change relative to WT ± SD (n = 5 biological replicates). Individual values of each experiment are represented in white or black filled squares triangles or circles. RM one-way ANOVA was applied. P-value left to right (p L-R): **** <0.001, ** 0.0087. b Western blot analysis as in (a) from primary cultures of WT or Mfn2 depleted (2KO) mouse cortical neurons, immunoblotted with anti-TOM20 and anti-MFN2. (n = 3 biological replicates). Paired t-test was applied. P value: 0.0388 c Western blot as in (a) of total cell lysates from HEK WT, 2KO and 2 + 2 cells, immunoblotted with anti-PINK1 (FL: full-length; CL: cleaved) and anti-HSP60. (n = 6 biological replicates). RM one-way ANOVA was applied. P L-R: ** 0.0074, ** 0.0041. d Western blot analysis as in a) of total cell lysates from HEK WT and 2KO cells treated with epoxomicin (1 µM, 2 h), immunoblotted with anti-PINK1 (revealing its cleaved (CL) form). (n = 4 biological replicates). RM one-way ANOVA was applied. P L-R: ** 0.0044, * 0.015. e Western blot analysis as in (a) of total cell lysates from HEK WT and 2KO cells transiently transfected with a control vector (-) or with the artificial mito-ER tether (+), immunoblotted with anti-PINK1 (FL: full-length; CL: cleaved). (n = 3 biological replicates). Two-way ANOVA was applied. f MtKeima mitophagy measurement in HEK WT, 2KO, 2 + 2 and 2 + 2^K357N cells treated with DMSO (-) or with IU1 ( + , 100 µM). Bars represent mean fold change relative to DMSO-treated WT (n = 3 biological replicates) ± SD. Individual values of each experiment are discriminated in black filled triangles and white or black filled circles. Two-way ANOVA was applied. P L-R: * 0.0227, *** 0.0005, ** 0.0024. Source data are provided as Source Data file.

Fig. 3. MFN2 is constitutively ubiquitylated and binds the proteasome.

a Volcano plot of significantly enriched proteins (p = 0,05, 2-fold enrichment) identified by label-free quantification of immunoprecipitated MFN2^FLAG. Proteasomal subunits: light green. Quality control (QC)-related proteins: blue. E3 ligases: dark green. (n = 3 biological replicates). b Confocal images of proximity ligation assays of HEK WT cells, with antibodies against MFN2 and PSMD11 (left upper panel) or against MFN2 and 20S subunits (left lower panel) (in red) and DAPI staining (in blue). Scale bar: 10 µm. Quantification of number of puncta per cell (right panels) using exclusively each of the antibodies or both (n = 2 biological replicates). Ordinary one-way ANOVA was applied. P values: **** <0.0001. Boxes borders indicate the 25% and 75% quantiles. The minimum and maximum values are shown by the whiskers excluding outliers. c Western blot analysis of MFN2, immunoprecipitated from total cell lysates of 1KO cells, immunoblotted with anti-MFN2 and anti-ubiquitin. Arrowhead points to endogenous unmodified MFN2, arrows point to modified MFN2, corresponding to ubiquitylated forms. d Western blot analysis as in (c) of total cell lysates from HEK WT cells, untreated (-) or treated (+) with USP21 DUB (30 min, 5 µm, 37 °C), immunoblotted with anti-MFN2 and anti-ubiquitin. Arrowhead points to endogenous unmodified MFN2, arrow points to modified MFN2, corresponding to ubiquitylated forms. e Western blot analysis of MFN2 immunoprecipitated from total cell lysates of WT cells, untreated (-) or treated with MLN-7243 ( + ; 0.5 µM, 4 h), immunoblotted with anti-MFN2 and anti-ubiquitin. Arrowhead points to endogenous unmodified MFN2, arrow points to modified MFN2, corresponding to ubiquitylated forms. Staining of the input with PoS was used as loading control. f Proximity ligation assay of HEK WT cells untreated (-) or treated with MLN-7243 ( + ; 0.5 µM, 4 h) as in (b) (n = 3 biological replicates). Ordinary one-way ANOVA was applied. P value: **** <0.0001. Boxes borders indicate the 25% and 75% quantiles. The minimum and maximum values are shown by the whiskers excluding outliers. Source data are provided as Source Data file.

Fig. 4. MFN2 controls MFN1 turnover.

a Western blot analysis of total cell lysates from HEK WT cells untreated (0) or treated with cycloheximide (CHX, 100 µg/mL) for 4 or 8 hr, simultaneously treated with epoxomicin (1 µM), when indicated, and immunoblotted with anti-MFN1. Staining of total protein with PoS was used as loading control. Shown are the mean and SD of the individual timepoints, relative to the correspondent untreated control, labeled as “0” on the Western blot (n = 3 biological replicates). Two-way ANOVA was applied. P L-R: ** 0.0023, **** <0.0001. b Western blot analysis as in a) of total cell extracts from HeLa WT, 1KO and 2KO cells untreated (0) or treated for 2 or 6 h with ActD (1 µM), immunoblotted with anti-MFN1/2 and anti-PARP1. Shown are the mean and SD of the individual timepoints, relative to the correspondent untreated control, labeled as “0” on the Western blot (n = 4 biological replicates). Two-way ANOVA was applied. P L-R: * 0.0367, **** <0.0001. c Western blot analysis as in a) of total cell extracts from HeLa WT and 2KO cells untreated (0) or treated for 2 or 6 hr with ActD (1 µM), immunoblotted with anti-PINK1 (FL: full-length; CL: cleaved). Bars represent the mean fold change relative to WT ± SD (n = 3 biological replicates). Individual values of each experiment are discriminated in triangles and white or black filled circles. Two-way ANOVA was applied. P value: * 0.0076. d Western blot analysis as in a) of total cell extracts from HeLa WT and 2KO cells untreated (0) or treated for 2 or 6 h with ActD (1 µM), immunoblotted with anti-MCL1. Bars represent the mean fold change relative to WT ± SD (n = 3 biological replicates). Individual values of each experiment are discriminated in triangles and white or black filled circles. Two-way ANOVA was applied. P value: **** <0.0001. Source data are provided as Source Data file.

Fig. 5. Absence of MFN2 causes aggregation.

a Confocal images of proximity ligation assay of HEK WT cells, with antibodies against MFN2 and HSC70 (in red) and DAPI staining (in blue) (left panel). Scale bar: 10 µm. Quantification of number of puncta per cell (right panel) using exclusively each of the antibodies or both (n = 2 biological replicates). Ordinary one-way ANOVA was applied. P value: **** <0.0001. Boxes borders indicate the 25% and 75% quantiles. The minimum and maximum values are shown by the whiskers excluding outliers. b Proximity ligation assay of HEK WT cells untreated (-) or treated with MLN-7243 ( + ; 0.5 µM, 4 h), as in a) (n = 3 biological replicates). Ordinary one-way ANOVA was applied. Boxes borders indicate the 25% and 75% quantiles. The minimum and maximum values are shown by the whiskers excluding outliers. c Staining of protein aggregation with the PROTEOSTAT® Aggresome detection kit (in red) in HEK WT, 1KO, 2KO and 2 + 2 cells (left panel). Scale bar: 10 µm. Quantification of number of protein aggregates per cell (right panel). n = 3 biological replicates, color-coded in greyscale, with at least 100 cells in each were quantified. Ordinary one-way ANOVA was applied. P value: **** <0.0001. Boxes borders indicate the 25% and 75% quantiles. The minimum and maximum values are shown by the whiskers excluding outliers. d Confocal images of HEK WT and 2KO cells co-stained with PROTEOSTAT® Aggresome detection kit (in red), anti-TOM20 (in green) and DAPI (in blue). Scale bar: 10 µm. Insets of white dotted boxes are shown on the right side of each image. e–g Quantification of amount of protein aggregates per cell in HEK WT and 2KO cells, untreated (-) or treated (+) with CCCP (20 µM, 2 h) on e, with VER-155008 (50 µM, 2 h) on f, or with CHX (5 µM, 5 h) on g. n = 1 biological replicate with at least 100 cells was quantified. Ordinary one-way ANOVA was applied. P value: **** <0.0001 (on e), P L-R: **** <0.0001, * 0,0166 (on f), P value: **** <0.0001 (on g). Boxes borders indicate the 25% and 75% quantiles. The minimum and maximum values are shown by the whiskers excluding outliers. Source data are provided as Source Data file.

Fig. 6. Protein aggregation in 2KO cells depends on ubiquitylation.

a–c Right panels: Confocal images of HEK WT and 2KO cells, untreated or treated with Bafilomycin A (100 nM, 4 h) on a, with MG132 (10 µM, 5 h) on b, or with MLN-7243 (MLN; 0.5 µM, 4 h) on c, stained with PROTEOSTAT® Aggresome detection kit. Scale bar: 10 µm. Left panels: Quantification of number of protein aggregates per cell in HEK WT and 2KO cells for the conditions depicted in (a–c). n=1 biological replicate with at least 100 cells was quantified. Ordinary one-way ANOVA was applied. P value: **** <0.0001 (on a), P L-R: **** <0.0001, ** 0.003, **** <0.0001 (on b), P value: **** <0.0001. Boxes borders indicate the 25% and 75% quantiles. The min and maximum values are shown by the whiskers excluding outliers. Source data are provided as Source Data file.

Fig. 7. CMT2A patient fibroblasts have increased protein aggregation.

a–c Confocal images after immunostaining with anti-TOM20 (in green) and DAPI (in blue), of HEK WT and 2KO cells, untreated or treated with VER-155008 (50 µM, 2 h) on a, with CCCP (20 µM, 2 h) on b, or with CHX (5 µM, 5 h) on c. Scale bar: 10 µm. Insets of white dotted boxes are shown below each image. d Confocal images of two control and three CMT2A patients’ primary fibroblasts carrying the point mutations R94Q, R94W and R104W in MFN2, co-stained with the PROTEOSTAT® Aggresome detection kit (in red), anti-TOM20 (in green) and DAPI (in blue). Scale bar: 10 µm. Insets of white dotted boxes are shown below each image.

Fig. 8. Roles of MFN2 in proteostasis control.

MFN2 maintains healthy mitochondria, which present in a tubular form, and exhibit functional oxidative phosphorylation (OXPHOS) capacity. This provides a healthy cellular environment, with active mitochondrial import, surveyed by cytosolic chaperones that bind MFN2. Furthermore, MFN2 is constitutively modified with short ubiquitin chains, binds p97 and the proteasome, allowing the turnover of other outer membrane proteins, including its cognate MFN1 and PINK1. In contrast, cells depleted for MFN2 (MFN2KO) fail to locally maintain this quality control at the mitochondrial surface. Mitochondria are swollen, fragmented and dysfunctional, with extremely reduced OXPHOS capacity. TOM translocase components are also decreased, reducing import capacity into mitochondria. Consequently, PINK1 accumulates, causing the upregulation of mitophagy, and proteins abnormally aggregate outside mitochondria.