SUMOylation of MFF coordinates fission complexes to promote stress-induced mitochondrial fragmentation

Abstract

Mitochondria undergo fragmentation in response to bioenergetic stress, mediated by dynamin-related protein 1 (DRP1) recruitment to the mitochondria. The major pro-fission DRP1 receptor is mitochondrial fission factor (MFF), and mitochondrial dynamics proteins of 49 and 51 kilodaltons (MiD49/51), which can sequester inactive DRP1. Together, they form a trimeric DRP1-MiD-MFF complex. Adenosine monophosphate-activated protein kinase (AMPK)-mediated phosphorylation of MFF is necessary for mitochondrial fragmentation, but the molecular mechanisms are unclear. Here, we identify MFF as a target of small ubiquitin-like modifier (SUMO) at Lys¹⁵¹, MFF SUMOylation is enhanced following AMPK-mediated phosphorylation and that MFF SUMOylation regulates the level of MiD binding to MFF. The mitochondrial stressor carbonyl cyanide 3-chlorophenylhydrazone (CCCP) promotes MFF SUMOylation and mitochondrial fragmentation. However, CCCP-induced fragmentation is impaired in MFF-knockout mouse embryonic fibroblasts expressing non-SUMOylatable MFF K151R. These data suggest that the AMPK-MFF SUMOylation axis dynamically controls stress-induced mitochondrial fragmentation by regulating the levels of MiD in trimeric fission complexes.

Fig. 1.. MFF is poly-SUMOylated at Lys¹⁵¹.

(A) HEK293T cells were co-transfected with GST-tagged DRP1 receptors [Fis1 (mouse), MFF (human, isoform 1), and MiD49 or MiD51 (mouse)] and either FLAG-SUMO1 or FLAG-SUMO2. GST immunoprecipitates and lysate were immunoblotted for FLAG and GST. (B) Wild-type (WT) or K151R cyan fluorescent protein (CFP)–MFF transfected cells were lysed in buffer ± 2% SDS and probed for CFP. (C) Blot of endogenously SUMOylated MFF. CFP-MFF (WT or K151R) immunoprecipitates from HEK293T cells were probed for endogenous SUMO1 or SUMO2/3. Note that, in experiments, when FLAG-SUMO is expressed, the mono-SUMOylated species of MFF resolves at ~130 kDa, whereas, when probing for endogenous SUMO, this corresponds to the ~115-kDa band, due to the lack of tag on SUMO. (D) Analysis of MFF SUMOylation-deficient mutants. GST pulldowns of the indicated mutants were blotted for FLAG and GST. (E) Quantification of SUMO-deficient MFF mutants. n = 3, ****P < 0.001, one-sample t test. (F) SUMO2/3 immunoprecipitation from HEK293T cell lysate, probed for MFF. Lanes 1 and 2 are control lanes using protein G beads. Lanes 3 and 4 are SUMO2/3-enriched samples using anti–SUMO2/3-conjugated beads. HEK293T cells were lysed in buffer ± 4% SDS and 20 mM N-ethylmaleimide (NEM) to preserve or inhibit SENP activity in the lysate. Four percent SDS was then diluted to 0.1% in lysis buffer before incubation with beads. Enrichment of SUMO2/3-conjugated proteins in lane 3 was confirmed by SUMO2/3 reprobe (bottom left blot), and deconjugation of SUMO2/3 was confirmed in the lysate blot. Arrowhead indicates endogenous SUMOylated MFF, and asterisk indicates the nonspecific antibody bands.

Fig. 2.. Mitochondrial morphology analysis and DRP1 recruitment in MFF-KO MEF cells expressing MFF-WT or MFF-K151R.

(A) Confocal images of WT and MFF–knockout (KO) MEF cells stained for DRP1 and mitochondria using MitoTracker. Enlargements show zoomed section of the highlighted area. Scale bars, 10 μm. (B) MEF cell lysate probed for MFF. MFF-KO cells lack all detectable isoforms of MFF. (C) Manders’ colocalization quantification of DRP1 and MitoTracker, n = 3, 84 to 98 cells were imaged. (D to F) Mitochondrial morphology analysis of MEF WT and MFF-KO cells. (D) number of branches per network, (E) mean mitochondrial length, and (F) free-end index. Mann-Whitney test was used to determine significance, 70 to 74 cells were imaged from three independent experiments, ****P < 0.0001. (G) Confocal images of MFF-KO MEF cells expressing either GFP, GFP-MFF-WT, or MFF-K151R. Enlargements show zoomed section of the highlighted area. Scale bars, 10 μm. (H) Viral titers of GFP, GFP-MFF-WT, or K151R infection of WT MEF cells were used to determine appropriate viral amount to infect cells with. The volume in lanes 8 and 12 were used for subsequent experiments. (I) Manders’ colocalization of DRP1 and MitoTracker [GFP, n = 2, 52 cells; and GFP-MFF (WT and K151R), n = 3, 85 to 91 cells]. (J to L) Mitochondrial morphology analysis of MFF-KO cells expressing GFP, WT-MFF, or K151R-MFF. (J) Network branching, (K) mitochondrial length, and (L) free-end index. n = 3, 73 to 95 cells, *P < 0.05 and ***P < 0.0005, Kruskal-Wallis test followed by Dunn’s multiple comparisons test. n.s., not significant.

Fig. 3.. MFF SUMOylation is promoted by AMPK-mediated phosphorylation.

(A and B) SUMOylation of MFF phosphorylation mutants. HEK293T cells were co-transfected with CFP-tagged 2SA/D MFF mutants and either (A) FLAG-SUMO1 or (B) FLAG-SUMO2. Immunoprecipitates and lysates were blotted for FLAG and CFP. Blot was cropped from larger blot (fig. S2E). Arrowhead indicates mono-SUMOylated MFF band, and asterisk shows higher–molecular weight bands. (C and D) Quantification of SUMOylation of MFF phosphorylation mutants. One-sample t test was performed to determine significance between mutants and WT, and unpaired t test was performed to determine significance between mutants, n = 4 or 5; *P < 0.05, **P < 0.01, and ***P < 0.005. (E) Ser¹⁵⁵ phosphorylation of MFF-K151R. HEK293T cells were transfected with the indicated CFP-MFF mutants, and immunoprecipitates were blotted for Ser¹⁵⁵ phosphorylation using an AMPK substrate motif antibody. S155A mutant was used to confirm specificity of the antibody. (F) Quantification of phosphorylation state of MFF-K151R, n = 3, one-sample t test. (G) Ser¹⁵⁵ phosphorylation of MFF-WT and K151R. HEK293T cells were transfected with CFP-MFF (WT or K151R), and immunoprecipitates were blotted for Ser¹⁵⁵ phosphorylation. Arrowhead corresponds to the band similar to the size of the mono-SUMOylated MFF species, and asterisk represents higher–molecular weight (MW) species.

Fig. 4.. MFF posttranslational modifications regulate the MiD/DRP1 ratio in the fission complex.

(A) Representative blot of endogenous DRP1 binding to CFP-MFF mutants. HEK293T cells were transfected with the indicated CFP-MFF mutants, and immunoprecipitates were probed for DRP1. (B) Quantification of DRP1 binding, n = 5 or 6. Representative blot of (C) endogenous MiD49 and (E) MiD51-HA binding to MFF mutants. Immunoprecipitates from HEK293T were co-transfected with MiD-HA, and the indicated CFP-MFF mutants were probed for HA. Uncropped blot of E in fig. S3D. (D and F) Quantification of (D) MiD49 and (F) MiD51-HA binding to MFF mutants, n = 3 or 4, *P < 0.05, ***P < 0.005, and ****P < 0.001, one-sample t test used to determine significance from WT; one-way analysis of variance (ANOVA) was used to determine significance between groups. (G) The table shows the relative amounts of DRP1 and MiD within the different MFF complexes, compared to WT, obtained from the quantifications [(B), (D), and (F)]. The MiD-to-DRP1 ratio is highlighted in yellow, calculated from the values in blue. (H) Schematic of DRP1-MiD-MFF rearrangement in response to MFF phosphorylation and SUMOylation. Created with BioRender.com.

Fig. 5.. AMPK activation enhances MFF SUMOylation and MiD51 displacement in response to CCCP.

(A) Confocal images of HEK293T cells transfected with mito-DsRed following 1 hour of treatment with 10 μM CCCP. Scale bars, 10 μm. (B) HEK293T cells were transfected with CFP-MFF (WT or K151R) and treated with 10 μM CCCP for 1 hour before lysis, either alone or in combination with the AMPK inhibitor compound C (10 μM). CFP-MFF immunoprecipitates were blotted for SUMO2/3, Ser¹⁵⁵ phosphorylation (low- and higher-exposure blots are shown), and CFP. (C) Quantification of SUMOylation of WT-MFF following 1 hour of CCCP in the presence or absence of compound C. Representative of three independent experiments. One-sample t test for conditions versus vehicle control (Ctl), and two-sample t test for CCCP versus CCCP + CC conditions. **P < 0.01. (D and E) HEK293T cells were transfected with CFP-MFF (WT) and treated with rotenone (250 ng/ml) or 1 mM AICAR for 1 hour before lysis. CFP immunoprecipitates were blotted for SUMO2/3 and CFP. Quantification presented in (E), n = 5 (rotenone) and n = 4 (AICAR). Uncropped blot is shown in fig. S4C. (F) HEK293T cells expressing GFP-MFF (WT or K151R) and MiD51-HA were treated with CCCP (10 μM, 1 hour). Co-immunoprecipitates were immunoblotted for HA and GFP. (G) Quantification of MiD51-HA binding to MFF following CCCP treatment, n = 4 or 5. One-sample t test for CCCP conditions versus vehicle controls, and two-sample t test for WT versus K151R CCCP conditions. *P < 0.05 and **P < 0.01.

Fig. 6.. MFF SUMOylation is not necessary for DRP1 recruitment under CCCP treatment but is required for promoting mitochondrial fragmentation.

(A) Confocal imaging of CCCP-induced mitochondrial fragmentation in MFF-KO MEF cells virally expressing GFP alone, or WT or K151R GFP-MFF. Cells were treated with CCCP (10 μM, 1 hour). Mitochondria were stained using MitoTracker Deep Red, endogenous DRP1 stain is shown in green, and GFP channel is shown in cyan. Processed images of mitochondrial stain with enlargements of highlighted area. Scale bar 10 μm. (B) Manders’ colocalization analysis of DRP1 with MitoTracker. Kruskal-Wallis test, 57 to 119 cells were imaged from three independent experiments, **P < 0.01 and ****P < 0.0001. (C) Quantification of the free-end index, data were generated from three independent experiments, expressed as percentage of DMSO control, 72 to 110 cells were imaged (for GFP-MFF–expressing cells); two independent experiments, 50 to 53 cells were imaged for GFP-expressing cells. Unpaired t test, **P < 0.01.

Fig. 7.. Working model of MFF SUMOylation–dependent stress-induced fission.

MFF, DRP1, and MiD49/51 proteins exist in a trimeric complex. Multiple types of DRP1 oligomers exist in the trimeric complex. Upon AMPK activation, MFF is phosphorylated at Ser¹⁵⁵ and Ser¹⁷², leading to MFF SUMOylation at Lys¹⁵¹ (for simplicity, only one phosphorylation site is shown). In the WT condition, this results in reduced MiD association and displacement from the complex, leading to the formation of MFF-DRP1 fission–competent complexes. How MiD proteins transfer DRP1 to MFF remains to be determined. When MFF cannot be SUMOylated, DRP1 is still recruited, and MFF is still phosphorylated, but MiD proteins remain associated in the trimeric complex. MFF-DRP1 fission complexes are not efficiently formed, leading to impaired fragmentation in response to stress. Created with BioRender.com.