The mitochondrial protein MTP18 contributes to mitochondrial fission in mammalian cells

Mitochondria are dynamic organelles that change morphology by controlled fission and fusion events. Mitochondrial fission is regulated by a conserved protein complex assembled at the outer membrane. Human MTP18 is a novel nuclear-encoded mitochondrial membrane protein, implicated in controlling mitochondrial fission. Upon overexpression of MTP18, mitochondrial morphology was altered from filamentous to punctate structures suggesting excessive mitochondrial fission. Mitochondrial fragmentation was blocked in cells coexpressing either the mitochondrial fusion protein Mfn1 or Drp1K38A, a dominant negative version of the fission protein Drp1. Also, a loss-of function of endogenous MTP18 by RNA interference (RNAi) resulted in highly fused mitochondria. Moreover, MTP18 appears to be required for mitochondrial fission because it is blocked after overexpression of hFis1 in cells with RNAi-mediated MTP18 knockdown. In conclusion, we propose that MTP18 functions as an essential intramitochondrial component of the mitochondrial division apparatus, contributing to the maintenance of mitochondrial morphology.

GTPase dynamin 1-like protein (DNM1L, here referred to as Drp1) is recruited to the mitochondrial outer membrane to exert fission of mitochondria. In Caenorhabditis elegans, the assembly of Drp1 on mitochondria results in severing the outer membrane (Labrousse et al., 1999). Similarly, genetic and biochemical studies in yeast uncovered a multi-step pathway for mitochondrial fission, whereby preassembled Drp1 structures, in concert with other mitochondrial outer membrane components, are sufficient to cut apart both mitochondrial membranes (Mozdy et al., 2000;Sesaki and Jensen, 1999;Shaw and Nunnari, 2002;Tieu and Nunnari, 2000;Tieu et al., 2002). However, exactly how Drp1 is recruited to mitochondria and how the fission process is then executed is still not fully understood.
Like Drp1, the mitochondrial outer-membrane protein Fis1 appears to be a key player during mitochondrial fission. Human Fis1 [tetratricopeptide repeat protein 11 (TTC11), hereafter referred to as hFis1] is a highly conserved 17-kDa integral mitochondrial protein of the outer membrane. Recently, several reports have provided strong evidence that hFis1, like its yeast homologue Fis1p, is involved in mediating mitochondrial fission, presumably by setting up the fission complex on the cytoplasmic face of the outer membrane through (direct or indirect) recruitment of Drp1 (Frieden et al., 2004;James et al., 2003;Stojanovski et al., 2004;Suzuki et al., 2003;Yoon et al., 2003). In addition, hFis1 has been implicated in the control of Drp1-mediated fission during apoptosis. Mitochondrial fission and mitochondrial-membrane remodeling have been shown to be the initiating step of the apoptotic pathway (Bossy-Wetzel et al., 2003;Fantin et al., 2002;Frank et al., 2001; Mitochondria are dynamic organelles that change morphology by controlled fission and fusion events. Mitochondrial fission is regulated by a conserved protein complex assembled at the outer membrane. Human MTP18 is a novel nuclear-encoded mitochondrial membrane protein, implicated in controlling mitochondrial fission. Upon overexpression of MTP18, mitochondrial morphology was altered from filamentous to punctate structures suggesting excessive mitochondrial fission. Mitochondrial fragmentation was blocked in cells coexpressing either the mitochondrial fusion protein Mfn1 or Drp1 K38A , a dominant negative version of the fission protein Drp1. Also, a loss-of function of endogenous MTP18 by RNA interference (RNAi) resulted in highly fused mitochondria. Moreover, MTP18 appears to be required for mitochondrial fission because it is blocked after overexpression of hFis1 in cells with RNAi-mediated MTP18 knockdown. In conclusion, we propose that MTP18 functions as an essential intramitochondrial component of the mitochondrial division apparatus, contributing to the maintenance of mitochondrial morphology. Karbowski et al., 2002;Scorrano et al., 2002). Transient expression of hFis1 in mammalian cells leads to excessive mitochondrial fission. Consequently, cytochrome c is released from mitochondria and triggers the onset of programmed cell death, highlighting the direct link between mitochondrial function and apoptosis (James et al., 2003). In addition to the proposed interaction of Drp1 with Fis1 during recruitment of Drp1 to the mitochondrial membrane, another report discusses Drp1 as a target for protein modification processes through interaction with modifying proteins such as Ubc9 (ubiquitinconjugating enzyme 9) and Sumo1 (small ubiquitin-like modifier 1) (Harder et al., 2004).
Furthermore, a few viral proteins have been reported to be actively involved in fragmentation of mitochondria after their expression in mammalian cells, however, with different effects on the apoptotic behavior of mitochondria . We have recently discovered another novel human mitochondrial protein, MTP18, that is implicated in the control of mitochondrial morphology (Tondera et al., 2004).
In this study, we present further experimental evidence for MTP18's function in maintaining the balance of fission and fusion in mammalian cells. MTP18-induced mitochondrial fragmentation upon its overexpression can be inhibited by increasing counteracting mitochondrial fusion activity. Knockdown of MTP18 by RNA interference (RNAi) blocks hFis1-induced mitochondrial fission, therefore suggesting that MTP18 is required to facilitate the complete fission step.
Cell culture and transfections COS-7 and HeLa cells were cultured as described (Tondera et al., 2004). Transfection of MTP18-expression constructs using the Effectene transfection reagent (Qiagen) was carried out according to manufacturer's instructions. For transient coexpression of specific shRNAs and recombinant proteins, HeLa cells were first transfected with shRNA-expressing plasmids for 24 hours and then again with shRNA-expressing plasmids and overexpression plasmids (ratio 9:1) for another 24 hours.

Subcellular fractionation and proteinase K treatment
Isolation of mitochondria was done as described (Tondera et al., 2004). For membrane protein analyses, mitochondrial pellets were resuspended in 400 µl of 0.1 M Na 2 CO 3 or 0.5 M NaCl, vortexed and incubated on ice for 30 minutes. The insoluble membrane fractions were centrifuged at 100,000 g for 10 minutes, and the supernatant was precipitated with 10% (v/v) trichloroacetic acid (TCA). For proteinase K digestion, isolated mitochondria were suspended in isotonic mitochondrial buffer (MB; 250 mM sucrose, 10 mM HEPES, 10 mM KCl, 2 mM MgCl 2 , 1 mM EGTA pH 7.4) and incubated at room temperature with 5 µg/ml proteinase K for the times indicated. Digestion was terminated with 2 mM phenylmethylsulfonyl fluoride (final concentration). Mitochondrial proteins were separated by SDS-PAGE and proteins were detected by western blotting with anti-MTP18, anti-hFis1 and anti-Hsp60 antibodies.
Antibodies, dyes and immunoblotting Rabbit polyclonal antiserum against MTP18 has been previously described (Tondera et al., 2004). Monoclonal anti-cytochrome c, anti-Hsp60 and anti-Drp1/DLP1 antibodies were purchased from BD Transduction Laboratories. The rabbit polyclonal anti-Mfn1 antibody was kindly provided by Minx Fuller (Santel et al., 2003). Anti-hFis1 was obtained from Alexis Inc. (Portland, OR). Mouse-and rabbitspecific secondary antibodies coupled to Alexa fluor488, Alexa fluor596 or Alexa fluor633, as well as mitochondria-specific dye MitoTracker Red were purchased from Molecular Probes. Preparation of cell extracts and immunoblot analysis was carried out as described (Klippel et al., 1998).

Immunofluorescence microscopy
Before fixation, cells were treated for 1 hour with 5 µM nocodazole (Figs 5, 6, and 7) and for 15 minutes with 1 µM MitoTracker Red at 37°C. For immunofluorescence, cells were fixed for 15 minutes at room temperature in 4% formaldehyde-phosphate-buffered saline. All subsequent steps for indirect immunofluorescence were carried out as previously described (Santel and Fuller, 2001). An LSM 510 META confocal microscope (Zeiss) was used for microscopy. Multitrack scanning mode was used to record double-labelled cells.
the Drp1 fission protein (Tondera et al., 2004). More precisely, transient expression of MTP18-myc in COS-7 cells resulted in Drp1-mediated mitochondrial fission giving rise to punctated mitochondria (Tondera et al., 2004). In contrast to long filamentous mitochondria in untransfected cells, the mitochondria in transfected cells expressing MTP18-myc appeared to be bend and circular (Fig. 1A, three small images) and showed distinct Drp1 foci that assembled before fragmentation (Fig. 1A). MTP18 was not particularly localised to Drp1 fission sites. In transfected cells exhibiting fragmented mitochondria, Drp1 was detected at the end of punctated mitochondria (Fig. 1B). A close look at mitochondria with several Drp1 foci revealed ring-like Drp1 structures at constriction sites (Fig. 1B, three small images). Hence, mitochondria with high levels of MTP18-myc underwent Drp1mediated fission.
MTP18 is an internal mitochondrial membrane protein Sequence analyses suggests that MTP18 contains three α-helical stretches of high hydrophobicity ( Fig. 2A and Fig. 3A, TM1-3) that are probably transmembrane (TM) domains ( Fig. 2A). Interestingly, for the annotated Drosophila melanogaster MTP18homologue CG7772 only one TM domain (corresponding to the C-terminally located TM3) was predicted (data not shown). Other protein domain motifs, such as coiled-coil domains, were not predicted (not shown).
The association of MTP18 with membranes was analysed by treating mitochondrial cell fractions with NaCl or Na 2 CO 3 (Fig. 2B), which caused the release of peripheral membrane associated proteins. MTP18 and the outer-membrane protein hFis1 still precipitated with the membrane fraction after centrifugation, in contrast to Hsp60, a marker protein for the soluble fraction (Fig. 2B).
To analyse MTP18's submitochondrial localisation, isolated mitochondria were  (B) Western blot of mitochondrial extracts treated with Na 2 CO 3 and NaCl after separation by ultracentrifugation in soluble (S, peripheral membrane proteins) and pelleted proteins (P, integral membrane proteins). Blots were probed with anti-Hsp60, anti-hFis1 and anti-MTP18. (C) Protease digestion of mitochondria with proteinase K (PK) or trypsin (lanes 2 and 4, or 3, respectively). Extracts were subjected to SDS-PAGE and western blots were carried out with anti-MTP18, anti-Mfn1, anticytochrome c and anti-Hsp60. Mitochondria were also osmotically shocked to disrupt their outer membrane and treated with PK (lane 4). MTP18 becomes degraded like cytochrome c upon osmotic shock. (D) Confocal microscopy of mitochondria from MTP18-myc transfected COS-7 cells stained with anti-MTP18 (green) and anti-Hsp60 (red). MTP18-myc did not completely colocalise with the mitochondrial matrix protein Hsp60. Arrows indicate arcas of mitochondria where MTP18 does not colocalise with Hsp60; double arrow indicates mitochondria from an untransfected cell. treated with proteinase K and trypsin to digest the outer-membrane protein Mfn1 (Fig. 2C, lanes 2, 3 compared to untreated control lane 1). MTP18, however, the mitochondrial inter-membrane-space (IMS) protein cytochrome c and the mitochondrial matrix protein Hsp60, were not affected, suggesting that MTP18 is an internal mitochondrial protein.
Treating mitochondria with proteinase K after having had them subjected to osmotic shock (to selectively break the mitochondrial outer membrane) resulted in the degradation of MTP18 and cytochrome c, while Hsp60 was still protected (lanes 4, 5). This suggests that MTP18 is mainly exposed to the intermembrane space.
Furthermore, the submitochondrial localisation of MTP18 was investigated by confocal microscopy to support the biochemical data. Colocalisation experiments were performed in COS-7 cells transfected with MTP18-myc because polyclonal anti-MTP18 antibody only recognizes recombinant MTP18 but not the lower levels of endogenous MTP18. In transfected cells with very low levels of recombinant MTP18-myc (cells that still exhibit filamentous mitochondria morphology), MTP18-myc did not completely colocalise with the matrix protein Hsp60 (Fig. 2D) supporting the above biochemical data that identified the location of MTP18 in the IMS. In addition, in cells that had been co-transfected with GFP-Mfn1 and MTP18-myc, MTP18-myc was distinctly localised beyond the outer membrane (Fig. 4D). Taken together, these results indicate that MTP18-myc is not associated with the outer membrane facing the cytoplasm, but is exposed to the intermembrane space. We refer to MTP18 as an intramitochondrial protein because different models for its topology are conceivable (see Discussion).
The C-terminus of MTP18 is crucial for localisation and function of MTP18 The transient expression of a truncated version of MTP18 lacking the C-terminus, affects proper mitochondrial localisation of MTP18 and blocks mitochondrial fission. Microscopic studies were carried out with COS-7 cells transfected with expression constructs for myc-tagged full-length MTP18, or truncated variants lacking the N-terminal half (∆1-90-MTP18-myc) or the C-terminal half (∆91-166-MTP18-myc) (Fig. 3A). Transient expression of the full-length and the deletion constructs was confirmed with polyclonal anti-MTP18 antiserum in western blots (Fig. 3A). Transfected cells with high levels of recombinant MTP18-myc were discriminated from untransfected cells by immunofluorescence-staining with polyclonal anti-MTP18 antiserum. The MTP18 antibody detected recombinant MTP18-myc, but not low levels of endogenous MTP18 protein in mammalian cells. The full-length MTP18 protein and the N-terminally truncated variant localised to mitochondria as shown by counterstaining with an antibody against cytochrome c. Both variants caused fragmentation of mitochondria, demonstrating that the N-terminal half of the Journal of Cell Science 118 (14) Fig. 3. Effect of the MTP18 C-terminus on protein localisation and mitochondrial morphology. (A) (left) diagram of myc-tagged (C-terminal red boxes) MTP18 wild-type and mutants that lack the N-terminus (including the predicted TM1) or the C-terminus (including the predicted TM2 and TM3). Transient protein expression was confirmed by western blot (right panel) probed with anti-MTP18 (*, endogenous MTP18; arrow, full-length MTP18-myc; arrowhead, truncated protein variants. (B) Representative confocal immunofluorescence images of transiently transfected COS-7 cells expressing wild-type (upper row), N-terminally truncated (middle row) and C-terminally truncated MTP18 (lower row). Shown is distribution of MTP18 (green, anti-MTP18) and mitochondrial morphology (red, anti-Hsp60). Bars, 20 µm. (C) Immunofluorescence staining of ∆91-166-MTP18-myc in transfected COS-7 cells with anti-MTP18 (green) and anti-Hsp60 (red) to investigate its subcellular localisation. Arrowheads, untransfected cells; arrow, cell with low expression levels of ∆91-166-MTP18-myc; double-arrow, cell with high expression levels of ∆91-166-MTP18-myc.

Journal of Cell Science
protein is not required for proper targeting and fragmentation (Fig. 3B). By contrast, when high levels of an MTP18 variant that lacks the C-terminal half of the protein (∆91-166-MTP18myc) were transiently expressed, it did not colocalise with cytochrome c, but instead showed a conspicuous cytoplasmic pattern. Moreover, the mitochondria did not appear fragmented but clustered (Fig. 3B), even when ∆91-166-MTP18-myc is transiently expressed at very high levels. By contrast, a more detailed analysis of transfected cells revealed that low levels of ∆91-166-MTP18-myc associated with mitochondria ( Fig. 3C, arrow), but did not colocalise with Hsp60 (Fig. 3C, inset). Increasing amounts of this variant localised to the cytoplasm and nucleus in transfected cells as revealed by immunofluorescence staining with anti-MTP18 (Fig. 3C, double-arrow). Therefore, MTP18 might behave as an inner mitochondrial membrane protein containing more than one particular import signal (see Neupert, 1997). Notably, mitochondria remained intact and did not become fragmented in cells with low levels of recombinant ∆91-166-MTP18-myc. Mis-localisation of ∆91-166-MTP18-myc presumably due to impaired import might affect the mitochondrial translocation machinery resulting in commonly observed perinuclear clustering of mitochondria upon transient protein expression (Santel and Fuller, 2001). Taken together, the COOH-terminus of MTP18 bearing the predicted TM2 and TM3 regions contributes to the correct mitochondrial protein targeting and mitochondrial fragmentation activity.
Mitochondrial fission caused by transient overexpression of MTP18 is blocked by dominant negative Drp1 (Drp1 K38A ) Overexpression of MTP18 in cell culture induced fragmentation of mitochondria (Tondera et al., 2004), as shown by transient coexpression of MTP18-myc and the GFP-tagged mitochondrial marker protein ANT-GFP (adenosine nucleotide transporter (Fig. 4A) (Santel and Fuller, 2001). Mitochondrial morphology is maintained by controlled fusion and fission events. Both events are controlled by two large GTPases the mitofusins (Mfn1 and Mfn 2) and Drp1. Whereas fission is executed by the Drp1 protein, mitochondrial fusion is mediated by the mitochondrial outer membrane GTPase Mfn1. To study the effect of these proteins on MTP18-induced mitochondrial fragmentation, the following expression constructs were used: myc-tagged MTP18, HA-tagged Drp1, a dominant negative variant of Drp1 (HA-Drp1 K38A ), GFP-tagged Mfn1 and GFPtagged variant of the GTPase-defective mutant Mfn1 (GFP-Mfn1 K88T ) (Fig. 4B).
Transient coexpression of MTP18-myc with Drp1 K38A interfered with unopposed mitochondrial fission. Drp1 K38A carries a missense mutation in the predicted GTPase domain of Drp1 and acts as dominant negative by blocking mitochondrial fission and promoting fusion (Smirnova et al, 1998;Pitts et al., 1999). By contrast, transiently overexpressed wild-type Drp1 has no effect on altering mitochondrial morphology (data not shown) (Pitts et al., 1999;Smirnova et al., 1998). After transient expression of Drp1 K38A in COS-7 cells, 67% of the transfected cells showed highly interconnected mitochondria (Fig. 4C, upper row; Fig. 4E). Consequently, loss of fission activity resulted in uncontrolled mitochondrial fusion. When MTP18-myc was transiently coexpressed with Drp1 K38A , mitochondria in 30% of doubly transfected cells appeared as a network of long and highly interconnected filaments that partially clustered around the nucleus and scattered throughout the cytoplasm. By contrast, of MTP18-myc transfected cells less than 5% exhibited this mitochondrial morphology (Fig. 4E). Notably, 50% of doubly transfected cells exhibited fragmented mitochondria like seen in transfected cells that overexpressing MTP18-myc alone (>95% showing fragmented mitochondria) (Fig. 4E). The appearance of mitochondrial networks indicates that Drp1 K38A overexpression can partly suppress MTP18-mediated fission activity.

Knockdown of MTP18 results in mitochondrial fusion
The presented overexpression and coexpression studies demonstrated that MTP18 might participate in controlling mitochondrial morphology by modulating the balance of fission and fusion. We have previously shown that overexpression of MTP18 can mediate mitochondrial fragmentation (Tondera et al., 2004). To test whether mitochondrial fission depends on MTP18 action, we analysed the effect of MTP18 loss-of-function in mammalian cells by shRNA-mediated RNA interference (for a review, see Dorsett and Tuschl, 2004;Shi, 2003). For this purpose, we transiently transfected HeLa cells with an expression construct that gives rise to high levels of double-stranded RNA specifically designed to target MTP18 mRNA (shRNA-MTP18). In the same manner, we generated shRNA expression constructs to target Mfn1 (shRNA-Mfn1) and Drp1 (shRNA-Drp1) as controls to compare loss-of-function effects on mitochondrial morphology. An unrelated shRNA (PI 3-kinase subunit p110α) was used as negative control (Czauderna et al., 2003). The 2day-long transient expression of these shRNA-expressing constructs in HeLa cells resulted in a clear and target-specific Journal of Cell Science reduction of target protein levels as assessed by western blot; Fig. 5A shows two representative examples. At the same time protein levels of a non-target control protein, hFis1, were not affected (Fig. 5A).
For microscopic analysis, HeLa cells were transfected with the above mentioned shRNA expression constructs together with a GFP expression construct (to identify transfected cells) at the ratio 9:1. Transfected cell that showed reduced levels of Mfn1, Drp1 and MTP18 showed dramatic changes in mitochondrial morphology when compared with the nonrelated shRNA control sample (Fig. 5B). Reduction of the general mediator of mitochondrial fusion Mfn1 resulted in the fragmentation of mitochondria in approximately 80% of transfected -GFP positive -cells, whereas shRNA-Drp1expressing cells (~78%) displayed long interconnected filamentous mitochondrial network (Fig. 5B). For shRNA-Mfn1, the observed morphology resembled the phenotype described for mitochondria in Mfn1 -/mouse embryonic fibroblast (MEFs) (Chen et al., 2003) and cells transfected with synthetic siRNA molecules for RNAi (Eura et al., 2003), whereas shRNA-Drp1 displayed the anticipated opposite effect of mitochondrial morphology.
Transient expression of shRNA-MTP18 gave rise to long mitochondrial filaments in approximately 90% of transfected cells, similar to the described effect for RNAi against Drp1 (Fig. 5B). The same result was confirmed with another shRNA-MTP18 sequence (data not shown). Therefore, we propose that RNAi-mediated MTP18 protein reduction results in uncontrolled fusion due to suppressed MTP18 fission activity. This observation is in agreement with our initial knockdown analysis of MTP18 in COS-7 cells using antisense molecules (Tondera et al., 2004). These RNAi data along with the overexpression results suggest that MTP18 is involved in mitochondrial fission.

hFis1 controlled mitochondrial fission appears to depend on MTP18
The level of MTP18 is crucial for hFis1-mediated mitochondrial fission. To study the influence of MTP18 on hFis1-mediated mitochondrial fission, we studied the knockdown effects of MTP18/hFis1 in combination with overexpression on mitochondrial morphology in transiently transfected HeLa cells (Fig. 6A). The mitochondrial hFis1 protein controls mitochondrial fission. Consequently, knockdown of hFis1 by shRNA-induced RNAi (Fig. 6A) resulted in the formation of long filamentous fused mitochondria (Fig. 6B upper row). By contrast, overexpression of hFis1 induced fragmented mitochondria in more than 90% of transfected cells (Fig. 6B, lower row). The fragmented mitochondria appeared round and swollen with an average size of 1.4 µm (s.d. 0.4 µm; n=60 mitochondria from two cells). Fragmented mitochondria in hFis1 transfected cells differed in size and morphology from those in MTP18 transfected cells giving rise to mitochondria with an average size of 0.7 µm (s.d. 0.1 µm; n=60). Notably, while overexpression of hFis1 finally led to cytochrome c release and apoptosis (data not shown) (James et al., 2003), overexpression of MTP18 had no effect on viability or apoptosis in transfected cells (Tondera et al., 2004).
Transient overexpression of myc-tagged MTP18 in HeLa cells along with RNAi-mediated knockdown of hFis1 led to fragmentation in many of the transfected cells (~50%). This result suggests that low levels of hFis1 are sufficient to facilitate MTP18-induced mitochondrial fission (Fig. 6C,E, red bars). By contrast, reduction of MTP18 opposed the hFis1induced mitochondrial fission in cells that had been doubly transfected for the transient coexpression of shRNA-MTP18 and hFis1 protein (Fig. 6D,E, green bars). Compared with control cells and MTP18-shRNA-expressing HeLa cells that express endogenous hFis1 (Fig. 6C, arrows), elevated levels of hFis1 (Fig. 6C, double arrow; and Fig. 6A for western blot) had no effect on the highly filamentous morphology phenotype (~84% of hFis1-overexpressing cells with fused mitochondria and 4% with fragmented mitochondria in shRNA-MTP18/ hFis1 doubly transfected cells) because of the suppressed MTP18 expression (Fig. 6E, green bars). Thus, the effect of hFis1 overexpression on mitochondrial morphology was suppressed by knockdown of MTP18. Based on these data, we conclude that reduced MTP18 expression abolished the mitochondrial fragmentation. Therefore, MTP18 appears to be required for the mitochondrial fission process.

MTP18 as a novel intramitochondrial component for mitochondrial fission
Drp1 is responsible for mitochondrial fission. As shown above, dominant negative activity of Drp1 K38A can partially block MTP18 induced mitochondrial fission. Overexpression of MTP18 in HeLa cells with reduced Drp1 levels (as assessed by western blot see Fig. 7A) revealed fused mitochondrial networks. Owing to loss of Drp1 function, fusion of mitochondria occurred (Fig. 7B, double arrow) but could not be disrupted by additional levels of MTP18 fission activity Two examples of specific RNAi-mediated protein reduction is shown on western blots with extracts from HeLa cells transiently transfected with shRNA-expressing plasmids (as indicated) to target MTP18, Mfn1, Drp1 and p110α; hFis1 served as loading control. (B) Representative confocal images of HeLa cells (right panels, higher magnification) transiently transfected with indicated shRNAexpressing vectors labelled with anti-Hsp60 and treatment with nocodazole before staining to reveal mitochondrial morphology. Bars, 20 µm. Diagrams show the percentage of transiently transfected HeLa cells (corresponding to the cell shown on the left) with indicated mitochondrial morphology (Fu, highly fused mitochondria; Nw, filamentous mitochondrial network; Fi, highly fragmented mitochondria). The majority of cells that show highly fused mitochondria were found in cells expressing shRNA against MTP18 and Drp1. Fragmented mitochondria were found in cells expressing shRNA against Mfn1. No obvious mitochondrial phenotype was found in cells expressing p110α-shRNA. Journal of Cell Science (Fig. 7B, arrow indicates MTP18-myc expressing cells). This result emphasises the requirement of Drp1 for carrying out the fission steps. MTP18-induced fission relies on Drp1, but apparently not on hFis1 in the employed transient assay system. In summary, our biochemical as well as RNAi and overexpression studies suggest that MTP18 is a novel intramitochondrial component contributing to the mitochondrial fission process.

Discussion
Mitochondrial fission is a complex process for the coordinated division of two mitochondrial membranes. So far, only a few conserved protein components of the outer membrane, such as the integral protein hFis1 and the peripheral recruited Drp1 protein, are known to be key mediators of mitochondrial fission. Overexpression of Drp1 in mammalian cells has no effect on the mitochondrial fission rate (Smirnova et al., 1998;Smirnova et al., 2001) implying that Drp1 itself is not sufficient (although necessary) to drive the division of both membranes. Mammalian proteins involved in the fission of the mitochondrial inner membrane are currently not known, but the existence of separate protein complexes that are responsible for mediating fission of the outer and inner mitochondrial membrane can be hypothesised.
We propose that MTP18 is a novel mitochondrial protein component involved in the maintenance of mitochondrial morphology by contributing to the control of mitochondrial fission. Changes in the expression levels of MTP18 by either overexpression or RNAi-mediated knockdown in mammalian cells affect mitochondrial morphology. The changes most probably interfere with the balance of mitochondrial fission and fusion. The effects on mitochondrial morphology caused by different levels of MTP18 are reminiscent of mitochondrial fission proteins: loss of MTP18 results in pronounced mitochondrial fusion due to the block of fission and consequently unopposed mitochondrial fusion. By contrast, high levels of MTP18 stimulate frequent mitochondrial fission, which results in a high number of fragmented mitochondria by overriding endogenous fusion processes. Furthermore, an increase of mitochondrial fusion activity, by either overexpressing the main mitochondrial fusion protein Mfn1 or the dominant negative form of the fission protein Drp1, antagonises MTP18-induced mitochondrial fission. These results indicate a functional role of MTP18 in controlling the balance of fission and fusion processes. Our applied transient assays in mammalian cells proves the contribution of MTP18 to the maintenance of mitochondrial morphology and resembles the genetic system in yeast used to study novel regulatory components of the mitochondrial morphology machinery. In addition, by using RNAi (shRNA-MTP18) together with protein overexpression (hFis1), we were able to show that MTP18 might be required for the mitochondrial fission process (Fig. 5). Nevertheless, RNAi only allows to assess the effect of protein reduction (knockdown), in contrast to 'real' loss-of-function effects caused by null mutations in the endogenous gene. Therefore, further genetic analysis (e.g. gene knockout) is needed to confirm the hFis1-independent requirement of MTP18 in regulating mitochondrial fission.
Our biochemical and microscopic data imply that MTP18 is an integral protein residing inside mitochondria. MTP18 acts most probably as a component of the inner membrane (IM) facing the inner membrane space (IMS), but it is also conceivable that MTP18 is anchored to the mitochondrial outer membrane (OM) from the inner leaflet and exposed to the IMS. Since the number of predicted TM domains has not been fully determined, the exact localisation and TM topology of MTP18 remains to be clarified. Thus, it remains unclear, in which mitochondrial compartment the N-terminus and C-terminus reside. Since the antibody used in this study was raised against the entire protein, the exact localisation of the protein cannot be resolved in detail. A putative association of MTP18 as a bipartite TM protein of the outer membrane -with both protein termini facing the intermembrane space -is conceivable, if cytoplasmic portions of MTP18 were protected by peripheral outer membrane proteins from proteinase K digestion. Nevertheless, our results indicate that MTP18 does not act like Corresponding protein extracts were subjected to SDS-PAGE and the amount for endogenous and recombinant proteins was assessed by western blot. Mfn1 protein levels served as loading control. (B) RNAi for hFis1 gave rise to highly fused mitochondria in shRNA-expressing, transiently transfected HeLa cells as revealed by anti-cytochrome c staining (representative picture, upper left panel; higher magnification, right panel). Cells were treated with nocodazole before fixation and immunofluorescence staining. Overexpression of hFis1 resulted in fragmentation of mitochondria (second row of images) as shown by immunofluorescence staining with anti-hFis1 (green) -to identify HeLa cells expressing high levels of hFis1 -and anti-cytochrome c (red) to assess mitochondrial morphology (arrow indicates an untransfected cell). The third and fourth panel show merged images with hFis1 and cytochrome c colocalisation and a close-up view of fragmented mitochondria. (C) HeLa cells were doubly transfected with MTP18-myc expression construct and shRNA-Fis1 RNAi vector in order to target hFis1. Cells with high levels of MTP18 were identified by immunofluorescence using anti-MTP18 (green). Mitochondria were visualised by labelling with anti-cytochrome c (red). High levels of MTP18 resulted in fragmented mitochondria (double arrow). By contrast, cells that do not transiently express of MTP18-myc show fused mitochondria due to RNAi of hFis1 RNA (arrow). Overexpression of MTP18 led to fragmented mitochondria in cells with reduced hFis1 levels (merged higher magnification image on right). (D) Reduced levels of MTP18 abolished hFis1-mediated mitochondrial fragmentation after transient overexpression. HeLa cells doubly transfected with shRNA-MTP18 and hFis1 expression constructs are shown. hFis1 was detected by immunofluorescence with anti-hFis1 (green). Arrows indicates cells with endogenous levels of hFis1; double-arrows points to cells transiently expressing high levels of hFis1. Mitochondrial morphology was revealed by immunostaining with anti-cytochrome c. High levels of hFis1 had no effect on the fused mitochondrial morphology due to loss of MTP18 by RNAi (merged image and higher magnification, last two panels on right). Bars in the first three panels of the second row in B, and in C and D, 20 µm. (E) Percentage of doubly transfected HeLa cells (RNAi in combination with overexpression) that show the indicated mitochondrial morphology. Fu, mainly fused mitochondria; Nw, network of filamentous mitochondria; Fi, mitochondria with highly fragmented appearance. RNAi against hFis1 (green bars) together with overexpressed MTP18; RNAi for MTP18 together with overexpressed hFis1 (red bars). Mfn1 as a 'common' outer membrane protein in the proteinaseprotection experiment (Fig. 2C). Furthermore, we observed that GFP-tagged MTP18 fusion protein appeared to be processed upon transient overexpression in mammalian cell culture (D.T., unpublished observation). This preliminary result points to the possibility that MTP18 might become processed during import or maturation inside the mitochondrion, which raises the possibility for differential sublocalisation inside the IMS compartment -as reported for OPA1 (Satoh et al., 2003). The mitochondrial fusion protein OPA1/Mgm1 (Delettre et al., 2000;Griparic et al., 2004;Olichon et al., 2003;Olichon et al., 2002;Wong et al., 2000;Wong et al., 2003) has recently been discovered to be a target of the Rhomboid-like protease Rbd1/Pcp1 in yeast implying the occurrence of proteolytic cleavage mechanisms for the control of mitochondrial morphology (Herlan et al., 2003;McQuibban et al., 2003). An in-depth analysis with epitopespecific antibodies for MTP18 together with in-vitro import studies will clarify putative processing steps and reveal the exact topology of endogenous MTP18.
The intramitochondrial localisation of MTP18 suggests the following function during mitochondrial fission: it is conceivable that MTP18 is actively involved in setting up the fission complex inside the mitochondrion to allow division of both membranes. The RNAi experiments indicate that MTP18 is required to allow hFis1-induced fission, even if elevated levels of hFis1 are present. This result suggests that hFis1 function is not sufficient in setting up the division complex on the surface of the outer membrane, if MTP18 was lacking. In this scenario, a hypothesised mitochondrial division machinery that lacks the intramitochondrially localised MTP18containing part cannot execute the fission step for both membranes. In other words, it is conceivable that MTP18 completes the set-up of a fission complex inside the mitochondrion by bridging division complexes of the outer membrane with intramitochondrial division components. Conversely, lack of the inner mitochondrial fission machinery abrogates complete fission and might only result in partial division of the mitochondrial outer membrane. This idea is in agreement with the observation, that overexpression of hFis1, in contrast to MTP18, results in the formation of differently appearing fragmented mitochondria, which finally release cytochrome c. For this reason, we speculate that MTP18 action might be the limiting step during the fission process. When overexpressed, MTP18 possibly promotes the assembly of complete fission machineries that are responsible for faithful mitochondrial division. Consequently, release of cytochrome c does not occur. Interestingly, overexpression of Drp1 itself does not increase mitochondrial fission, presumably owing to the lack of sufficient mitochondrial division components such as hFis1 for the outer membrane and MTP18 for the inner mitochondrial membrane/intramitochondrial compartment. This proposed mode of MTP18 action is reminiscent of MDM33, the yeast mitochondrial inner membrane protein.
MDM33 was proposed to be involved in controlling the fission of the mitochondrial inner membrane by homotypic interaction, resulting in the constriction of the inner membrane (Messerschmitt et al., 2003). However, to date no homologue or orthologue of MDM33 is known in higher eukaryotes. The elucidation of the topology of MTP18 inside mitochondria and the identification of interacting proteins will help to clarify contribution of MTP18 to the fission process. Fig. 7. Knockdown of Drp1 affects MTP18-induced mitochondrial fission. (A) HeLa cells were co-transfected with an shRNA-Drp1 expression plasmid for RNAi and with an MTP18-myc expression vector. The resulting knockdown of Drp1 together with overexpression of MTP18mycis was shown by western blotting. No changes in protein levels were detected for hFis1 and Mfn1. Cells were treated with nocodazole and stained with anti-MTP18 (green) to show recombinant MTP18-myc overexpressing cells (arrows indicate representative examples) and anti-cytochrome c (red) to show mitochondrial morphology. Mitochondria show highly fused morphology (merged picture and high magnification, two panels on right) due to RNAiinduced reduction of Drp1 in cells with endogenous levels of MTP18 (double-arrow; not detectable with anti-MTP18) as well as in cells with high levels of MTP18-myc (arrow).