The mitochondrial protein Dmel_CG41128 is homologous to MIC10 from yeast and humans

The genome of D. melanogaster encodes three different proteins with noticeable sequence similarity to the MIC10 proteins from yeast (ScMIC10) and humans (HsMIC10): MINOS1a/Dmel_CG12479; MINOS1b/Dmel_CG41128; and MINOS1c/Dmel_CG13564. Like MIC10 from yeast and humans, these three proteins contain two putative transmembrane domains (TMDs) with conserved glycine-rich motifs (Fig 1A). Each N-terminal TM segment contains a GxxxG motif (Engelman motif), which has been reported to mediate oligomerization of several membrane proteins (Russ & Engelman, 2000).

Each C-terminal TM domain contains a related, highly conserved GxGxGxG motif, which has been shown to be crucial for the oligomerization of ScMIC10 (Barbot et al, 2015; Bohnert et al, 2015). Of the three orthologues, only Dmel_CG41128 is ubiquitously expressed throughout all analyzed developmental stages, whereas Dmel_CG12479 and Dmel_CG13564 are testes-specific proteins (FlyBase FB2022_06). Therefore, we decided to focus on the investigation of Dmel_CG41128.

The expression of Dmel_CG41128-FLAG in S2 cells and subsequent immunolabeling and fluorescence microscopy demonstrated that Dmel_CG41128 indeed localized to mitochondria (Fig 1B). We next used AlphaFold2 (Jumper et al, 2021) to predict the overall structure of Dmel_CG41128, ScMIC10, and HsMIC10. The AlphaFold2 algorithm predicted a hairpin topology reminiscent of some ER-resident reticulons (Yang & Strittmatter, 2007) for all three proteins, with the conserved GxxxG and GxGxGxG motifs being oriented to each other in a similar way (Fig 1C). The predicted structures are fully in line with previous experimental studies on ScMIC10, which demonstrated that two TM domains of ScMIC10 are linked by a short loop that points toward the matrix, whereas the termini of the protein point toward the IMS (Barbot et al, 2015; Bohnert et al, 2015).

We conclude that Dmel_CG41128 is a mitochondrial protein that shows sequence homology with known MIC10 proteins and presumably features a comparable hairpin-like shape, making it a promising candidate for the MIC10 subunit from D. melanogaster.

Loss of Dmel_CG41128 reduces the life span and the fertility of flies

In humans, the loss of subunits of the MIC10 subcomplex is associated with severe diseases such as mitochondrial encephalopathy, myopathy, or cognitive impairment (Guarani et al, 2016; Zeharia et al, 2016; Benincá et al, 2021). To investigate the influence of Dmel_CG41128 on mitochondria and the life span of flies, we generated flies deficient for Dmel_CG41128 using CRISPR/Cas9 genome editing (Figs 2A and B and S1). The loss of Dmel_CG41128 reduced the average life span of flies by around 40% (Fig S2A). Moreover, the loss of Dmel_CG41128 reduced the fly’s overall fertility, indicated by a significantly decreased seminal vesicle area (Fig S2B), which is in line with the finding that Dmel_CG41128 was present in both testes and ovaries from D. melanogaster (Fig S3A–H).

Dmel_CG41128 is the major MIC10 orthologue in D. melanogaster

In both humans and yeast, MIC10 controls the stability of the MIC10 subcomplex. Depletion of MIC10 leads to the degradation of MIC13 and strongly disturbs cristae architecture (Harner et al, 2011; Hoppins et al, 2011; von der Malsburg et al, 2011; Alkhaja et al, 2012; Callegari et al, 2019; Kondadi et al, 2020; Stephan et al, 2020). To investigate the consequences of Dmel_CG41128 depletion on mitochondrial architecture in D. melanogaster, we isolated mitochondria from WT and Dmel_CG41128-deficient flies and analyzed them by SDS–PAGE and immunoblotting. Similar to the situation in yeast and human cells (von der Malsburg et al, 2011; Guarani et al, 2015; Kondadi et al, 2020), loss of Dmel_CG41128 caused the loss of DmMIC13 when analyzing steady-state levels (Fig 2A). Next, we performed co-immunoprecipitation experiments on the isolated WT mitochondria using Dmel_CG41128 as a bait (Fig 2B). We found that Dmel_CG41128 interacted with both DmMIC13 and DmMIC60, central subunits of both MICOS subcomplexes (Guarani et al, 2015). Likewise, when DmMIC60 was used as a bait, Dmel_CG41128 and DmMIC13 were pulled down (Fig 2B), indicating that Dmel_CG41128 is a part of the Drosophila MICOS complex.

To test whether Dmel_CG41128 also interacts with subunits of mammalian MICOS complexes, we expressed Dmel_CG41128-FLAG and FLAG-tagged human MIC10 (HsMIC10-FLAG) in human HeLa cells and COS-7 cells from the green African monkey Cercopithecus aethiops. Approximately, the same amounts of MICOS subunits were pulled down with these proteins as baits in COS-7 cells. Also, in HeLa cells Dmel_CG41128-FLAG pulled down subunits of MICOS, although HsMIC10-FLAG was a more efficient bait (Fig 2C). We conclude that Dmel_CG41128-FLAG interacts also with the mammalian MICOS complex.

MIC10-deficient mitochondria from yeast and human cells exhibit highly disturbed mitochondrial ultrastructure (Harner et al, 2011; Hoppins et al, 2011; von der Malsburg et al, 2011; Guarani et al, 2015). Transmission electron microscopy of brain tissue mitochondria of flies deficient for Dmel_CG41128 revealed similar mitochondrial phenotypes with most of the mitochondria exhibiting aberrant crista morphologies including tube-like and onion-shaped cristae (Fig 2D and E).

Altogether, the ubiquitously expressed Dmel_CG41128 shares sequence homology with other MIC10 proteins, presumably features a hairpin-like structure, and binds to the MICOS complexes of D. melanogaster, H. sapiens, and C. aethiops. Dmel_CG41128 further regulates the levels of DmMIC13 in flies, and its depletion strongly affects the cristae architecture. These findings support the view that Dmel_CG41128 is the ubiquitously expressed MIC10 orthologue in D. melanogaster. Hence, in accordance with the uniform nomenclature for MICOS (Pfanner et al, 2014), we will refer to it as DmMIC10b from here on.

DmMIC10b has a propensity to polymerize into filaments

When analyzing the subcellular localization of overexpressed DmMic10b in S2 fly cells using confocal fluorescence microscopy, it became apparent that at higher expression levels, DmMic10b-FLAG influenced the overall shape of the mitochondrial network (Fig 3A). We next performed super-resolution microscopy to investigate the distribution of DmMIC10b in more detail. At moderate expression levels, STED nanoscopy revealed the formation of distinct DmMIC10b clusters, comparable to the situation in yeast and human cells (Jans et al, 2013). Intriguingly, long DmMIC10b-containing filaments, apparently forming bundles of filaments, were observable at higher expression levels of DmMIC10b-FLAG (Fig 3A and B). These DmMIC10b-containing filaments seemed to be located inside of mitochondria and pervaded the mitochondrial network (Fig 3A–C). Even when scrutinized by STED microscopy, the filaments appeared to be contiguously labeled, suggesting that they might be formed by DmMIC10b only, rather than being a mosaic out of DmMIC10b and other proteins (Fig 3B). To explore this idea, we first expressed DmMIC10b-FLAG in two heterologous cellular systems, namely, HeLa and COS-7 cells. Similar to the overexpression in S2 cells, the expression of DmMIC10b in these cells resulted in the formation of DmMIC10b-containing filaments, which formed exclusively inside mitochondria (Figs 4A and S4A and B).

As DmMIC10b interacted with mammalian MIC60 (Fig 2C), we analyzed the fine distribution of MIC60 in these DmMIC10b-overexpressing cells. Dual-color 2D STED recordings of COS-7 cells indeed suggested occasional spatial connections between the filaments and some MIC60 clusters (Fig 4A). MIC60 seemed to be localized in clusters along the IBM as described before (Harner et al, 2011; Jans et al, 2013; Stoldt et al, 2019; Pape et al, 2020), whereas DmMIC10b-containing filaments often seemed to be situated more toward the center of the mitochondrial tubules (Fig 4A, Inset 1). To localize DmMIC10b in 3D, we next performed 4Pi-STORM of COS-7 cells expressing DmMIC10b-FLAG (Bates et al, 2022). At low expression levels, DmMIC10b was found in close proximity to MIC60 clusters, suggesting that it was mainly located at CJs or in the IBM (Fig S5A–C). However, the 3D recordings confirmed that at higher expression levels, a large fraction of DmMIC10b was present in filamentous structures (Video 1). Most of these long filaments were located along the center of mitochondrial tubules with no obvious connection to the MIC60 clusters indicating the IBM (Figs 4B and S5A and Video 2).

Filament formation does not require other MICOS proteins and is independent of the C-terminal tag

The nanoscopy data supported the idea that DmMIC10b-FLAG filaments may form independent from other MICOS proteins. To test this further, we next expressed DmMIC10b-FLAG in human MIC10-KO and MIC60-KO cells, as these cells are devoid of the MIC10 subcomplex and the entire MICOS complex, respectively (Stephan et al, 2020). STED recordings revealed that the formation of DmMIC10b filaments occurred in the absence of the mammalian MIC10 subcomplex or mammalian holo-MICOS complex (Figs 4C and S6A and B). We conclude that additional MICOS subunits are not required for the polymerization of DmMIC10b into filaments.

As all DmMIC10b filaments shown thus far relied on the overexpression of a C-terminally FLAG-tagged version of DmMIC10b, we next expressed non-tagged DmMIC10b in mammalian HeLa, U-2 OS, and COS-7 cells and labeled them with antibodies against DmMIC10b to test whether the formation of the filamentous structures was induced by the FLAG epitope. Also, untagged DmMIC10b formed filaments at high expression levels, demonstrating that its propensity to form a filamentous structure is independent of the tag (Fig S6C–E).

We next investigated whether purified DmMIC10b is able to polymerize into filamentous structures in vitro. To this end, His₆-DmMIC10b was expressed in Escherichia coli and purified to homogeneity from inclusion bodies as described previously (Barbot et al, 2015). After the removal of detergent by dialysis, purified His₆-DmMIC10b exhibited a distinct ladder pattern on SDS–PAGE (Fig 4D, right panel), as previously reported for ScMIC10 (Barbot et al, 2015; Bohnert et al, 2015). This suggests that the purified DmMIC10b can associate into oligomers. In line with this observation, negative stain electron microscopy demonstrated the formation of filamentous structures upon the removal of the detergent (Fig 4E).

Taken together, we conclude that DmMIC10b can polymerize into homo-oligomeric filaments both in mitochondria and in vitro.

Filament formation seems to be specific for DmMIC10b

Although the formation of filamentous MICOS structures had been suggested by early studies conducted on yeast (Hoppins et al, 2011), several studies using super-resolution microscopy have not substantiated the existence of extended MICOS-only filaments in yeast or mammalian cells (Jakobs et al, 2020). Still, elongated MIC60 assemblies in mitochondria of COS-7 cells (Bates et al, 2022) or in mitochondria of HeLa cells depleted of the dynamin-like GTPase optic atrophy 1 (Stephan et al, 2020) have been reported. However, these ring- or arc-like structures only wrapped around the mitochondrial tubules and were by orders of magnitude shorter than the DmMIC10b filaments reported here. Previous studies overexpressing HsMIC10 did not report on the formation of filaments at various expression levels (Stephan et al, 2020) either. To further explore whether MIC10 from yeast or humans has a tendency to polymerize when expressed in a heterologous system, we expressed ScMIC10 in COS-7 cells and S2 cells and HsMIC10 in S2 cells. No filaments were observed, suggesting that the propensity to form filaments is a specific characteristic of DmMIC10b (Fig S7A–C).

DmMIC10b filaments remodel the crista membranes

As MIC10 is a membrane-shaping protein (Barbot et al, 2015; Bohnert et al, 2015), it appeared possible that the DmMIC10b filaments influence the overall IM architecture. To test this, we expressed the cristae marker COX8A fused to a SNAP-tag (Stephan et al, 2019) together with DmMIC10b in HeLa and COS-7 cells and visualized the fusion protein inside mitochondria using live-cell STED nanoscopy. Both mitochondria of WT HeLa and COS-7 cells showed lamellar cristae as previously reported (Stephan et al, 2019; Liu et al, 2022). The overexpression of DmMIC10b strongly altered the cristae architecture, with the IM often apparently collapsed along the DmMIC10b filaments (Fig 5A).

To investigate the influence of DmMIC10b polymerization on the cristae architecture in more detail, we next recorded electron tomograms of chemically fixed COS-7 (Figs 5B and S8A) and S2 (Fig S8B) cells expressing DmMIC10b. Electron microscopy revealed filament bundles oriented along the mitochondria. As these filaments were absent in WT cells, we assume that these are DmMIC10b filaments. Whereas some of these filaments were in close contact with the IM, other filaments seemed to form between OM and IBM and inside the crista lumen, thereby widening the IMS or causing the formation of aberrant, tubular cristae (Figs 5B and S8). As STED nanoscopy recordings of human MICOS knockout cells expressing DmMIC10b had indicated a similar arrangement of the DmMIC10b filaments (Fig S6A and B), we analyzed those cells using electron microscopy as well (Fig S9). In MIC60-KO cells, we observed most of the filaments in bundles between OM and IBM (Fig S9A). Despite its ability to interact with human MICOS, the expression of DmMIC10b could not rescue the aberrant cristae morphology in human MIC10-KO cells (Fig S9A–C). Instead, we observed the formation of bundles of filaments that remodeled the crista membranes in a similar way as observed in WT cells (Fig S9D).

Together, our data suggest that upon overexpression, DmMIC10b polymerizes into extended filaments that associate with bundles within the IMS and strongly influence the mitochondrial IM architecture.

Filament formation of MIC10b is regulated by conserved amino acids

DmMICOS subunits can suppress filament formation of DmMIC10b

The formation of extended DmMIC10b filaments in fly cells was only observed upon the overexpression of DmMIC10b, which is a condition that results in a shifted balance between DmMIC10b and the other MICOS proteins. We speculated that at physiological conditions, filament formation of DmMIC10b is suppressed by interacting MICOS proteins. To explore this, DmMIC10b-FLAG was co-expressed with several Spot-tagged versions of MICOS proteins of the MIC10 and the MIC60 subcomplexes, and filament formation was determined by STED nanoscopy (Fig 6D). To ensure the balanced expression of the respective two overexpressed proteins, we expressed both proteins as a translational fusion, separated by a self-cleaving 2A peptide.

The co-expression of DmMIC10b with the MIC10 subcomplex subunits DmMIC13 or DmMIC26 suppressed DmMIC10b filament formation in most (>80%) of the cells, whereas DmMIC19, part of the MIC60 subcomplex, reduced filament formation less efficiently (∼50%). Upon the co-expression of the human DmMIC13 orthologue HsMIC13, most of the analyzed cells still formed filaments, suggesting that the fly homologs have co-evolved to efficiently suppress the ability of DmMIC10b to form long filaments at physiological conditions (Fig 6D and E). The suppression of filaments by DmMIC13 was reduced when its only cysteine residue (C90) was replaced by a serine residue (Fig 6D and E), further supporting the notion that in D. melanogaster, cysteine residues are key for the regulation of the oligomerization status of MIC10.

Taken together, DmMIC10b, the major MIC10 from D. melanogaster, has the propensity to form extended filaments in vitro and in cells. Upon overexpression, the filaments reside in the IMS and deform the IM. The formation of filaments requires glycine-rich motifs within the TMDs, as well as a conserved cysteine residue (C19) in close proximity to the N-terminal TMD. The formation of filaments requires excess of DmMIC10b and is effectively suppressed by the co-overexpression of other constituents of the MIC10 subcomplex.

Plasmids

Expression plasmids for S2 cells

Expression plasmids for mammalian cells

Co-expression of DmMIC10b-FLAG and Spot-tagged MICOS proteins

Expression plasmids for bacteria

Purification of DmMIC10b from E. coli and preparation for EM

The purification of His-tagged DmMIC10b was performed as described previously (Barbot et al, 2015). In brief, E. coli BL21 (DE3) cells were collected by centrifugation after the expression of His₆-DmMIC10b (1 mM isopropyl-β-D-thiogalactopyranoside [IPTG], 3 h, 37°C) and stored at −20°C until purification. After thawing, cells were lysed, and inclusion bodies were isolated and subsequently dissolved in resuspension buffer containing 8 M urea, 150 mM NaCl, 20 mM Tris–HCl, 40 mM imidazole, and 0.5 mM DTT, pH 8.0. The mixture was applied to a HisTrap column (5 ml) and eluted with resuspension buffer supplemented with 500 mM imidazole. Isolated MIC10 was further subjected to a HiLoad 16/600 Superdex 200 size-exclusion column (GE Healthcare). Separated fractions were analyzed by SDS–PAGE and Coomassie brilliant blue staining.

Expression of DmMIC10b mutants in E. coli and analysis by Western blotting

DmMIC10b mutants were expressed in E. coli BL21 (DE3). Cells were collected by centrifugation after expression (1 mM isopropyl-β-D-thiogalactopyranoside [IPTG], 3 h, 37°C) and washed in salt buffer containing 150 ml NaCl and 10 mM Hepes, pH 7.4. The pellet was resuspended in lysis buffer containing 50 mM Tris (pH 8.0), 150 mM NaCl, 0.1 mg/ml lysozyme, 1 mM MgCl₂, and cOmplete protease inhibitor (Merck), and the cells were homogenized by sonication. The homogenate was supplemented with Benzonase (Sigma-Aldrich) and stirred for 30 min at 4°C. After centrifugation, the pellet was washed with Triton X-100 wash buffer containing 50 mM Tris, 150 mM NaCl, 1 mM EDTA, and 2% (wt/vol) Triton X-100 (pH 8.0) followed by washing with a wash buffer containing 50 mM Tris, 100 mM NaCl, 10 mM DTT, and 1 mM EDTA (pH 8.0). The pellet was dissolved with 8% (wt/vol) sarcosyl and 1 M urea (in 50 mM Tris, pH 8.0), and insoluble material was removed by centrifugation. The solution was diluted with solubilization buffer (20 mM Tris and 150 mM NaCl, pH 8.0) to a final concentration of 1.5% (wt/vol) sarcosyl, supplemented with 0.1% (wt/vol) DDM, and dialyzed against solubilization buffer overnight. Samples were analyzed by SDS–PAGE followed by Western blotting. His-tagged DmMIC10b was detected using an anti-His antibody (34660; QIAGEN N.V.).

Cultivation of flies and life span assay

Flies were maintained on standard fly food with cornmeal, yeast, and agar at 25°C on a 12/12-h light/dark cycle. Fly stocks used were DmMIC10b[KO] (this study), w[1118] (#6326; Bloomington Drosophila Stock Center [BDSC]), and OR-R (#5; BDSC). WT control flies used were the heterozygous progeny of w[1118] males and OR-R females.

The life span assay was performed using four biological replicates per genotype. 15 male and 15 female young flies were placed in each vial. Every 2–4 d, flies were transferred to new food vials, and both living and dead flies were counted for consistency. Counting was continued until all flies had died.

Generation of knockout flies

Dmel_CG41128[KO] flies were generated by a commercial transformation service (BestGene Inc). The injection stock was RRID:BDSC_55821 (Bloomington Drosophila Stock Center).

Cell culture

Drosophila S2 cells (Cat. No. R69007, Lot No. 2082623; Thermo Fisher Scientific/Gibco) were cultivated in Schneider’s Drosophila medium (Cat. No. 21720024; Merck/Sigma-Aldrich/Gibco) supplemented with 1mM sodium pyruvate (Cat. No. S8636; Merck/Sigma-Aldrich) and 10% (vol/vol) fetal bovine serum (Cat. No. FBS. S 0615, FBS Superior stabil; Bio & Sell GmbH). Cells were cultivated at 28°C and ambient CO₂ levels. Kidney fibroblast-like cells (COS-7) from the green African monkey C. aethiops (Cat. No. 87021302, Lot No. 05G008; Merck/Sigma-Aldrich) and human cervical cancer cells (HeLa) were cultivated in DMEM, containing 4.5 g/liter glucose and GlutaMAX additive (Cat. No. 10566016; Thermo Fisher Scientific) supplemented with 1 mM sodium pyruvate (Cat. No. S8636; Merck/Sigma-Aldrich) and 10% (vol/vol) FBS at 37°C and 5% CO₂. Human osteosarcoma cells (U-2 OS) (Cat. No. 92022711, ECACC) were cultivated at 37°C and 5% CO₂ and grown in McCoy’s medium (Thermo Fisher Scientific) supplemented with 10% (vol/vol) FBS and 1% (vol/vol) sodium pyruvate.

Transfection of HeLa, COS-7, and S2 cells

S2 cells were split the day before transfection. On the day of transfection, 1.5 × 10⁶ cells were seeded per well of a six-well dish. Transient transfection was achieved using Effectene transfection reagent (Cat. No. 301425; QIAGEN) and 2 μg of plasmid DNA per well. After 23 h, the cells were detached and reseeded on glass coverslips coated with concanavalin A (Cat. No. C5275; Merck/Sigma-Aldrich) for 1 h.

Mammalian cells were seeded on coverslips (for light microscopy), on Aclar film (for electron microscopy), and in cell culture dishes (for biochemistry) 1 d before transfection. Transient transfection was achieved by jetPRIME transfection reagent (Cat. No. 114-15; Polyplus-transfection) following the manufacturer’s protocol.

Generation of antibodies against DmMICOS proteins

Drosophila anti-MICOS polyclonal antibodies were produced by injecting purified His-tagged DmMIC10b or synthetic peptides into rabbits (Prof. Hermann Ammer, Munich, Germany). For anti-MIC10 antibody generation, recombinantly expressed and affinity-purified MIC10 protein, and for anti-MIC60 and anti-MIC13 antibody generation, AAKPKDNPLPRDVVELEKA and GDSDQTDKLYNDIKSELRPH synthetic peptides were injected into rabbits, respectively. All antibodies were affinity-purified.

Mitochondrial isolation

Mitochondria were isolated as described previously (Panov & Orynbayeva, 2013) with slight modifications. Flies were taken up in cold TH buffer, containing 300 mM trehalose, 10 mM KCl, 10 mM Hepes (pH 7.4), 2 mM PMSF, 0.1 mg BSA/ml, and protease inhibitor mix (04693116001; Roche). The flies were then two times homogenized with a Dounce homogenizer (800 rpm/min), and the large remaining fragments were pelleted at 400g for 10 min at 4°C. Afterward, the remaining pieces were removed by centrifugation (800g, 8 min, 4°C) and the mitochondria-containing supernatant was saved in a new vial. To collect mitochondria, a further spin at 11,000g, 10 min, 4°C was performed, and pellets from separate reaction tubes were pooled and washed with BSA-free TH buffer. The mitochondrion concentration was determined using the Bradford assay.

Affinity purification from fly mitochondria

Isolated mitochondria (1 mg) were transferred to lysis buffer (150 mM NaCl, 10% glycerol [vol/vol], 20 mM MgCl₂, 2 mM PMSF, 50 mM Tris–HCl, pH 7.4, 1% digitonin [vol/wt], and protease inhibitor [04693116001; Roche]) and agitated for 30 min at 4°C. Debris was removed by centrifugation (15 min, 16,000g, 4°C), and the cleared supernatant was transferred on protein A-Sepharose conjugated with anti-DmMIC60 or control antisera. After 1 h binding at 4°C, beads were washed 10 times with washing buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 10% glycerol [vol/vol], 20 mM MgCl₂, 1 mM PMSF, and 0.3% digitonin [wt/vol]). Proteins were eluted by the addition of 0.1 M glycine, pH 2.8, at RT for 5 min. Eluates were neutralized and analyzed by SDS–PAGE and immunoblotting using specific antibodies against DmMIC60 (this study), DmMIC10b (this study), DmMIC13 (this study), NDUFA9, and RIESKE (Dennerlein et al, 2021).

Affinity purification from cell lysates

COS-7 and HeLa cells were seeded in 15-cm cell culture dishes and cultured overnight. Cells were transfected with AAVS1-TRE3G-MIC10-FLAG-T2A-EGFP (Stephan et al, 2020, EMBOJ), AAVS1-TRE3G-DmMIC10-FLAG-T2A-EGFP, or AAVS1-TRE3G-EGFP (Qian et al, 2014) (52343; Addgene). The medium was exchanged 4 h after transfection, and expression was induced by adding 1 μg/ml doxycycline hyclate for 24 h. Cells were harvested by trypsinization and centrifugation. All the following steps were performed at 4°C. The pellet was washed with PBS and resuspended in 1.5 ml lysis buffer containing 20 mM Tris (pH 7.0), 100 mM NaCl, 1 mM EDTA, 10% (vol/vol) glycerol, 1% (wt/vol) digitonin (Carl Roth GmbH), and cOmplete protease inhibitor (Merck). Samples were rotated on a wheel for 1 h. After centrifugation at 13,000g for 10 min, the supernatant was transferred onto equilibrated magnetic anti-FLAG M2 magnetic beads (Merck). Samples were rotated on a wheel for 105 min. The supernatant was removed, and beads were washed 10 times with wash buffer containing 20 mM Tris, 100 mM NaCl, 1 mM EDTA, 5% (vol/vol) glycerol, 0.25% (wt/vol) digitonin, and cOmplete protease inhibitor (pH 7.0). Elution was performed by adding 0.1 M glycine (pH 2.8) and shaking the samples at 1,200 rpm for 20 min. Eluates were neutralized and analyzed by SDS–PAGE and immunoblotting using specific antibodies against FLAG (F3165; Sigma-Aldrich), MIC13 (SAB1102836; Sigma-Aldrich), MIC19 (HPA042935; Atlas Antibodies), COX3 (55082-1-AP; Proteintech), MIC60 (10179-1-AP; Proteintech), and MFN1 (ab57602; Abcam).

Preparation of fly tissues for fluorescence microscopy

Adult testes were dissected in ice-cold PBS, then fixed at RT using 4% formaldehyde in PBT (PBS + 0.2% [vol/vol] Triton X-100) for 20 min. After fixation, testes were rinsed in PBT (multiple rinses for a minimum of 30 total minutes), then blocked using PBTB (PBT + 0.2% [wt/vol] BSA + 5% [vol/vol] normal goat serum) for 30 min. After blocking, samples were incubated in primary antibody diluted in PBTB overnight at 4°C. Then, the primary antibody was removed, and samples were rinsed in PBT for 30 min, then for 30 min in PBTB. Secondary antibodies diluted 1:500 in PBTB were added to the samples overnight at 4°C. Samples were then rinsed again with PBT for 30 min, treated with DAPI for 10 min to stain DNA, and finally rinsed again with PBT. Samples were stored in Vectashield (Vector Laboratories) until being mounted on slides and imaged. Primary antibodies used for staining testes were rabbit anti-DmMIC10b (1:3,000, this work) and mouse anti-ATP5A (1:300, ab14748; Abcam).

Sample preparation of fixed S2 cells for STED nanoscopy

Fixation and labeling of S2 cells were done as described previously (Wurm et al, 2010). In brief, cells were fixed using an 8% (wt/vol) formaldehyde solution, permeabilized by incubation with a 0.25% (vol/vol) Triton X-100 solution, and blocked with a 5% (wt/vol) BSA solution. Proteins of interest were labeled with antisera against the FLAG-tag (Thermo Fisher Scientific) and ATP5A (Abcam). Detection was achieved via secondary antibodies custom-labeled with Alexa Fluor 594 or STAR RED.

Sample preparation of fixed HeLa and COS-7 cells for STED nanoscopy and 4Pi-STORM

Cells were transfected with pcDNA3.1-DmMIC10b-FLAG or with AAVS1-TRE3G-DmMIC10b-FLAG-T2A-EGFP. In case of the latter construct, cells were induced with 1 μg/ml doxycycline. Fixation and labeling were performed 24 h after transfection or induction. In brief, cells were fixed using 4% (wt/vol) or 8% (wt/vol) formaldehyde solution, permeabilized by incubation with a 0.25% (vol/vol) Triton X-100 solution, and blocked with a 5% (wt/vol) BSA solution (Wurm et al, 2010). Proteins of interest were labeled with antibodies against DmMIC10b (this study), FLAG-tag (Merck/Sigma-Aldrich), Spot-tag (ChromoTek), TOMM22 (Merck/Sigma-Aldrich), and MIC60 (Abcam). For STED nanoscopy, detection was achieved via secondary antibodies custom-labeled with Alexa Fluor 594 or STAR RED. Cells were mounted using Mowiol containing 1,4-diazabicyclo[2.2.2]octane (DABCO).

For 4Pi-STORM, samples were prepared on 18-mm-diameter glass coverslips, which were coated over one quarter of their surface with a reflective aluminum layer, which was used for initial alignment of the sample within the 4Pi-STORM microscope. Cells were plated on the coated side of the coverslip and fluorescently labeled as described before (Bates et al, 2022). For two-color imaging, cells were fixed and stained with antibodies against MIC60 (Abcam) and the FLAG epitope (Sigma-Aldrich). Primary antibodies were detected with Fab fragments coupled to Alexa Fluor 647 (Thermo Fisher Scientific) or antibodies labeled with Cy5.5 (Jackson ImmunoResearch). Before imaging, the sample was washed with PBS and mounted in STORM imaging buffer in a sandwich configuration with a second glass coverslip. The two coverslips were sealed together around their edge with fast-curing epoxy glue (UHU GmbH).

Sample preparation for live-cell STED nanoscopy

HeLa and COS-7 cells were seeded in glass bottom dishes (ibidi GmbH) and grown at 37°C and 5% CO₂ overnight. Cells were co-transfected with AAVS1-TRE3G-DmMIC10b-FLAG-T2A-EGFP and AAVS1-Blasticidin-CAG-COX8A-SNAP (Stephan et al, 2019). 4 h after transfection, the culture medium was exchanged and the expression of DmMIC10b-FLAG-T2A-EGFP was induced by adding 1 μg/ml doxycycline hyclate (Sigma-Aldrich) to the growth medium for 24 h. Before STED imaging, cells were stained with 1 μM SNAP-cell SiR-647 (NEB) for 30 min at 37°C and 5% CO₂. Transfected cells were identified based on the cytosolic EGFP reporter and recorded by live-cell STED nanoscopy at RT.

Sample preparation of cultivated cells for electron tomography

Aclar 33C disks were punched with 18 mm diameter using a 0.198-mm-thick Aclar film (Plano GmbH) and sterilized with 70% ethanol before usage. On these, COS-7 cells or MICOS-depleted HeLa cells (Stephan et al, 2020) were grown to ∼70% confluency and transfected with pcDNA3.1-DmMIC10b-FLAG. 24 h after transfection, samples were immobilized and fixed with 2.5% (vol/vol) glutaraldehyde in 0.1 M cacodylate buffer at pH 7.4, first for 1 h at RT, then transferred to 4°C until the next day. Alternatively, S2 cells were seeded on coated glass coverslips and transfected the next day as described above. They were fixed by immersion with 2.5% (vol/vol) glutaraldehyde in 0.1M phosphate buffer (pH 7.4).

Before post-fixation, the cells were in addition stained with 1% osmium tetroxide and 1.5% (wt/vol) K₄[Fe(CN)₆] in 0.1 M cacodylate buffer at pH 7.4 for 1 h at RT. After post-fixation in 1% osmium tetroxide for 1 h at RT and pre-embedding en bloc staining with 1% (wt/vol) uranyl acetate for 30 min at RT, samples were dehydrated and embedded in Agar 100 resin (Plano GmbH).

Sample preparation of fly tissues for electron microscopy

After extraction, fly brains were fixed in bulk by immersion with 2% (vol/vol) glutaraldehyde in 0.1 M cacodylate buffer at pH 7.4. Fixation was completed overnight at 4°C. After three washing steps with 0.1 M cacodylate buffer, the brains were stained with 1% (wt/vol) osmium tetroxide for 1 h at RT followed by en bloc staining with 1% (wt/vol) uranyl acetate for 30 min at RT. Subsequently, the brains were dehydrated in a graded series of ethanol and finally flat-embedded in a thin layer of Agar 100 resin in between two Aclar sheets. Individual brains were manually cut out of the resin with a saw, and 80-nm-thick sections were collected on Formvar-coated copper grids. 10 × 10 tiles at 8,600x original magnification were recorded at randomly selected areas of the fly brains.

Preparation of recombinant DmMIC10b for negative stain EM

DmMIC10b purified in the presence of urea was dialyzed (against 20 mM Tris, 100 mM NaCl, and 2 mM DTT, pH 8.0) to precipitate DmMIC10b. The precipitate was solubilized by adding 10% (wt/vol) sarcosyl and 5% (wt/vol) digitonin followed by thorough pipetting. The obtained solution was afterward diluted with a suspension buffer containing 20 mM Tris, 100 mM NaCl, 3% (wt/vol) DDM, and 2mM DTT, pH 8.0, up to a final concentration of 0.2% sarcosyl and 0.05% digitonin. The solution was supplemented with 10 mM imidazole and loaded onto magnetic His-Beads. Samples were rotated on a wheel for 1.5 h at 4°C. The supernatant was removed, and the beads were washed five times with a wash buffer containing 20 mM Tris, 100 mM NaCl, 10 mM imidazole, 1 mM DTT, and 0.05% (wt/vol) DDM, pH 8.0. Afterward, the beads were washed five times with a wash buffer containing 20 mM Tris, 100 mM NaCl, 1 mM imidazole, 1 mM DTT, and 0.01% (wt/vol) DDM, pH 8.0. Elution was performed with 500 mM imidazole. Samples were dialyzed (against 20 mM Tris, 100 mM NaCl, 1 mM DTT, and 0.01% [wt/vol] DDM, pH 8.0) using a 5 kD cutoff membrane. The samples were analyzed by SDS–PAGE and Coomassie brilliant blue staining. For negative staining EM, the obtained protein samples were bound to a glow-discharged carbon foil–covered 400-mesh copper grid. After successive washing with water, samples were stained with 1% (wt/vol) uranyl acetate aq. and evaluated at RT using a Talos L120C transmission electron microscope (Thermo Fisher Scientific).

Fluorescence microscopy

Imaging of testes was performed on a Zeiss LSM 700 confocal laser-scanning microscope (Carl Zeiss AG). Confocal light microscopy images of cultured cells were acquired using a Leica TCS SP8 confocal laser-scanning microscope (Leica Microsystems) equipped with an HC PL APO 63x/1.40 Oil objective (Leica Microsystems).

STED nanoscopy was performed using an Expert Line quad scanning STED microscope (Abberior Instruments) equipped with a UPlanSApo 100x/1.40 Oil objective (Olympus). Alexa Fluor 594 was excited at a 561-nm wavelength, and STAR RED was excited at a 640-nm wavelength. STED was performed at a 775-nm wavelength. SiR-647 was excited at a 640-nm wavelength, and STED was performed at a 775-nm wavelength as described previously (Stephan et al, 2019, SciRep).

For 4Pi-STORM, cells were imaged on the 4Pi-STORM microscope by illuminating with 642-nm laser light and recording the on–off blinking events of individual fluorescent molecules on a CCD camera, for ∼100,000 camera frames at a rate of 100 Hz. To achieve a larger imaging depth, the sample was periodically shifted along the optical axis (z-coordinate) in steps of 500 nm during the recording. Afterward, in a post-processing step, each blinking event was analyzed to determine a 3D molecular coordinate, and the full set of coordinates plotted together in 3D space forms the 3D 4Pi-STORM image. Multicolor imaging was achieved by distinguishing the Alexa Fluor 647 and Cy5.5 according to the ratio of photons for each fluorophore detected in the four image channels of the microscope. The spatial resolution of the processed 4Pi-STORM image is ∼10 nm in all dimensions. Full details of the microscope, image procedure, and data analysis are described in Bates et al (2022).

Electron microscopy

For 2D analysis, images of ultrathin sections of ∼70 nm thickness were recorded on a Philips CM120 BioTwin transmission electron microscope (Philips Inc.). Usually, 2D images of at least 20 different cells were randomly recorded for each sample at 8,600× original magnification using a TemCam 224A slow-scan CCD camera (TVIPS).

For electron tomography, tilt-series of thin sections of ∼230 nm that were in addition decorated with 10-nm gold beads on both sides were recorded on a Talos L120C transmission electron microscope (Thermo Fisher Scientific/FEI Company) at 17,500× original magnification using a Ceta 4k × 4k CMOS camera in an unbinning mode. Series were recorded from −65.0° to 65.0° with a 2° dose-symmetric angular increase. The series were calculated using Etomo (David Mastronade; http://bio3d.colorado.edu/).

Imaging data processing

4Pi-STORM data were analyzed using the custom MATLAB code. Single X,Y,Z coordinates were extracted from raw HDF5 files generated by STORM acquisition software, and split into two-channel datasets based on logical values embedded in the data structure. Nearest neighbor distances (minimum pairwise distances) were then calculated between the two channels, and complete pairwise distances were calculated within each channel, using inbuilt MATLAB functions. For the complete pairwise self-distances for each channel, a global threshold distance of 200 nm was used to filter the data, in order to probe the differences at distances comparable to mitochondrial diameters. The data were then plotted as empirical cumulative distribution functions and displayed on plots comparing the different conditions in the context of MIC10 expression levels. For volume renderings of 4Pi-STORM data, we relied on Imaris (Bitplane).

TEM recordings of thin sections and ET data were processed in Fiji using the median filter. For the analysis of cristae morphology, the images were manually categorized in a blinded approach based on the ultrastructure of individual mitochondrial cross sections. For automated segmentation of mitochondrial membranes, we employed MemBrain-Seg (https://github.com/teamtomo/membrain-seg) using the current provided model V10 and a sliding window size of 128. The membrane model was further refined manually in Amira 6.0.1 (Thermo Fisher Scientific). Filaments were manually segmented in IMOD 4.11 and imported into Amira for visualization. If not stated otherwise, STED nanoscopy data were smoothed with a low-pass filter using Imspector software (Abberior Instruments). For deconvolution, we used the Richardson Lucy algorithm in Imspector software. Confocal images of the fly tissue were processed using ImageJ and Adobe Illustrator. The seminal vesicle areas were calculated using the “Measure” feature in ImageJ.