Although several proteins involved in mediating mitochondrial division have been reported in mammals, the mechanism of the fission machinery remains to be elucidated. Here, we identified a human nuclear gene (named MTGM) that encodes a novel, small, integral mitochondrial inner-membrane protein and shows high expression in both human brain tumor cell lines and tumor tissues. The gene is evolutionarily highly conserved, and its orthologs are 100% identical at the amino acid level in all analyzed mammalian species. The gene product is characterized by an unusual tetrad of the GxxxG motif in the transmembrane segment. Overexpression of MTGM (mitochondrial targeting GxxxG motif) protein results in mitochondrial fragmentation and release of mitochondrial Smac/Diablo to the cytosol with no effect on apoptosis. MTGM-induced mitochondrial fission can be blocked by a dominant negative Drp1 mutant (Drp1-K38A). Overexpression of MTGM also results in inhibition of cell proliferation, stalling of cells in S phase and nuclear accumulation of γ-H2AX. Knockdown of MTGM by RNA interference induces mitochondrial elongation, an increase of cell proliferation and inhibition of cell death induced by apoptotic stimuli. In conclusion, we suggest that MTGM is an integral mitochondrial inner-membrane protein that coordinately regulates mitochondrial morphology and cell proliferation.
Mitochondria play a crucial role in cellular energy metabolism, producing ATP through oxidative phosphorylation. They are also involved in many cellular processes including β-oxidation of fatty acids, the urea cycle and programmed cell death. Mitochondrial dysfunction has been associated with a variety of pathological conditions, such as Alzheimer's and Parkinson's diseases (Liu and Ames, 2005), diabetes (Maassen et al., 2005), cardiovascular disease (Ballinger, 2005), osteoarthritis (Terkeltaub et al., 2002) and cancer (Grandemange et al., 2009).
Mitochondria are dynamic and ubiquitous organelles that, according to the cell's metabolic demands or pathological condition, frequently change their morphology via fission and fusion events (Kiefel et al., 2006; Okamoto and Shaw, 2005). These processes not only control the shape of mitochondria but also affect many cellular functions (Chan, 2006; Chen et al., 2005; Grandemange et al., 2009; Knott et al., 2008; Suen et al., 2008). Several recent studies reported that morphological changes of mitochondria occur during cell division (Arakaki et al., 2006; Lee et al., 2007). Mitochondria present as interconnected tubular structures in interphase, undergo a drastic reorganization and fragmentation at an early mitotic stage, and finally reform tubular structures in the two daughter cells. Phosphorylation of Drp1 (dynamin-related protein 1) is involved in these cell-cycle-dependent morphological changes of mitochondria (Taguchi et al., 2007).
In mammalian cells, the mitochondrial fusion machinery includes OPA1 (optic atrophy 1) and mitofusins (Mfn1 and Mfn2); the mitochondrial fission machinery consists mainly of Drp1 and Fis1 (fission 1). Additionally, MTP18 (mitochondrial protein 18 kDa), endophilin B1 and GDAP1 (ganglioside-induced differentiation-associated protein 1) have been found to be required for mitochondrial fission (Chan, 2006). The maintenance of mitochondrial shape depends on a balance between these two protein machineries. Although several key components of the protein machineries responsible for mitochondrial fission and fusion have been discovered in mammalian cells, the molecular mechanisms that regulate mitochondrial dynamics remain to be clarified and additional, so far unknown cofactors are likely to be required for these processes. Mitochondrial inner membrane proteins involved in the process of mitochondrial division remain poorly understood (Suen et al., 2008).
In the initial study, we identified a human nuclear gene showing high expression in brain tumor cell lines and tumor tissues. This gene encodes a small integral mitochondrial inner membrane protein, evolutionarily highly conserved throughout the eukaryotic kingdom. All orthologs of this protein contain a putative transmembrane (TM) segment characterized by a highly conserved tetrad of the GxxxG motif (Russ and Engelman, 2000), and it is therefore designated here as MTGM (mitochondrial targeting GxxxG motif) protein. During our study on the biological property of MTGM, Yoo and coworkers (Chung et al., 2006) reported their study on Romo1 (reactive oxygen species modulator 1), the same protein as MTGM. They showed that Romo1 (MTGM) increased the level of ROS (reactive oxygen species) in the cells, and that the enforced expression of Romo1 triggered premature senescence and also contributed toward induction of DNA damage (Chung et al., 2008). In this work, we demonstrated that overexpression of MTGM leads to mitochondrial fragmentation in a Drp1-dependent manner, inhibition of cell proliferation and release of mitochondrial Smac to the cytosol, but that overexpression has no effect on apoptosis and autophagy. Depletion of MTGM results in mitochondrial elongation, increased cell growth and resistance to apoptosis induced by staurosporine. These intriguing findings suggest that MTGM plays a key role in the regulation of mitochondrial morphology and in the coordination of mitochondrial dynamics with cell proliferation and mitotic events.
The MTGM gene is highly expressed in human brain tumors
By comparison of gene expression patterns in human brain tumors and the corresponding normal tissues using SAGE libraries, we have identified a gene named MTGM, which corresponds to the proteins C20orf52 and Romo1 (Chung et al., 2006). MTGM protein is upregulated in several different brain tumors (supplementary material Table S1). RT-PCR analysis showed that it is highly expressed in the two glioma cell lines U-343 MG and U-343 MGa Cl2:6 (Cl2:6) and the neuroblastoma cell line SH-SY5Y, relative to normal adult brain (supplementary material Fig. S1A). MTGM is also upregulated in several patient samples of medulloblastoma and astrocytoma (supplementary material Fig. S1B). Thus, we conclude that deregulation of MTGM expression occurs in human brain tumors. We also showed that MTGM was ubiquitously expressed in a panel of eight normal human adult tissues by RT-PCR (supplementary material Fig. S1C).
Molecular structural features of the MTGM gene and the putative MTGM protein
We analyzed the organization of the MTGM gene and potential alternative transcripts by RACE-PCR. This revealed that the gene consists of three exons, and that alternative splicing resulted in at least three transcript variants in human brain (α, β and γ) with alternative 5′ untranslated regions, but encoding the same protein (supplementary material Fig. S2, Table S2).
Bioinformatics analysis showed that the MTGM gene encodes a small protein of 79 amino acids with a theoretical molecular mass of 8.2 kDa. The protein contains a putative TM segment at the amino acid positions 22-44 (supplementary material Fig. S3), indicating that MTGM encodes a membrane protein.
Computational analysis further revealed the existence of MTGM homologs throughout the eukaryotic kingdom from yeast and plant to human (supplementary material Fig. S3), and all homologs in analyzed species contain a tetrad of the GxxxG motif (Russ and Engelman, 2000) in the TM segment. Moreover, its orthologs are 100% identical at the amino acid level in all analyzed mammalian species (supplementary material Fig. S3). Its high conservation in evolution suggests that MTGM plays an important role in cells.
MTGM encodes a mitochondrial targeting protein
To begin dissecting the function of MTGM, 293T cells were transfected with the pcDNA3.1-MTGM/V5-His expression plasmid (MTGM-V5). This resulted in a MTGM/V5-His fusion protein ∼13 kDa in size (Fig. 1A). Immunofluorescence revealed a punctate distribution of MTGM-V5 in the cytoplasm of transfected cells (Fig. 1B), co-localized with the mitochondrial marker MitoTracker (Fig. 1C), indicating that MTGM is a mitochondrial targeting protein. We generated an anti-MTGM specific antibody in rabbits, which specifically recognized both endogenous MTGM (∼8.5 kDa) and recombinant MTGM-V5 (Fig. 1D). Like exogenous MTGM-V5, endogenous MTGM was detected in mitochondria of 293T cells (Fig. 1E). Endogenous MTGM was also found in several indicated cell lines (Fig. 1F). Additionally, western blotting of isolated cytosolic and mitochondrial fractions confirmed that MTGM-V5 was present in the mitochondrial fraction (Fig. 1G). We furthermore constructed an N-terminally HA-tagged MTGM, which also localized to mitochondria in transfected 293T cells (data not shown), confirming that the expression of MTGM as a fusion protein did not affect its subcellular distribution. Thus, we conclude that MTGM encodes a mitochondrial targeting protein. Although MTGM is a mitochondrial protein, no putative N-terminal pre-sequence for mitochondrial targeting was predicted by computational analysis. However, many mitochondrial proteins of the inner and outer membrane have been reported to lack the N-terminal targeting sequence, and instead have internal targeting signal(s) (Habib et al., 2007; Pfanner and Geissler, 2001). We therefore speculate that MTGM might contain internal signal(s) for its targeting to mitochondria. For example, the TM segment and its flanking regions contain a few positively charged amino acid residues, which are probably involved as signals for mitochondrial targeting and membrane integration. To address this issue, further experiments will be required in the future.
MTGM is an integral mitochondrial inner-membrane protein
To demonstrate that MTGM is an integral membrane protein (see supplementary material Fig. S3), the mitochondrial fraction was isolated from 293T cells expressing MTGM-V5 and resuspended in mitochondrial buffer (MB) containing 0.1 M Na2CO3 (pH 11.5) or in MB buffer alone as a control. After centrifugation, supernatants and membrane pellets were analyzed by western blot. MTGM and the integral membrane protein Tim23 were retained in the mitochondrial membrane pellet after treatment with Na2CO3 (which results in release of soluble and peripheral membrane-associated proteins), whereas cytochrome c (cyt c), located in the intermembrane space, was released to the supernatant (Fig. 1H). These experiments indicated that MTGM is an integral mitochondrial membrane protein.
We next determined whether MTGM is an inner- or outer-membrane protein in mitochondria. Isolated mitochondria were incubated with or without proteinase K, which digests the cytosolic part of the outer-membrane proteins. Immunoblotting (Fig. 1I) showed that proteinase K did not affect MTGM and the inner-membrane protein Tim23, but it destroyed the outer-membrane protein Tom20 (lane 2). As a negative control, all these proteins were retained in the mitochondrial pellets when incubated with MB buffer alone (lane 1), and disappeared after incubation with MB buffer containing 1% Triton X-100 and proteinase K (lane 3). We conclude that MTGM is an integral mitochondrial inner-membrane protein.
Overexpression of MTGM induces mitochondrial fragmentation
We noted that mitochondria became extensively fragmented in 293T cells expressing MTGM-V5 (see Fig. 1) and also in HeLa cells expressing MTGM-V5 (data not shown). Thus, we examined the effect of MTGM on mitochondrial morphology by fluorescence and confocal microscopy. In most non-transfected 293T cells, the morphology of mitochondria was a mixed reticulum with tubular and round forms (95.2±0.8%) (mean ± s.d.), whereas only 4.6±0.7% of cells displayed tubular mitochondria (Fig. 2A,C). Cells with dramatically fragmented mitochondria were rarely observed in this population (0.2±0.3%). Similarly, in 293T cells transfected with a control GFP (green fluorescent protein) expression vector, most cells (94.9±2.5%) had a mixed reticulum of tubular and round mitochondria, and only 4.2±1.9% and 0.9±0.6% of cells had elongated tubular or fragmented mitochondria, respectively (Fig. 2C). By contrast, 89.5±1.9% of cells expressing MTGM-V5 (Fig. 2B, left; Fig. 2C) exhibited fragmented mitochondria, which was a significant increase compared with non-transfected cells or GFP-expressing cells (P<0.0001), whereas only 8.8±2.2% of the MTGM-expressing cells (Fig. 2B, right; Fig. 2C) had a mixed reticulum of tubular and round mitochondria, displaying a significant reduction relative to control cells (P<0.0001). A significant decrease was also seen in cells containing only tubular mitochondria in MTGM-expressing cell populations (1.76±0.7%) compared with non-transfected (4.6±0.7%, P<0.0005) or GFP-expressing cell populations (4.2±1.9%, P<0.05) (Fig. 2C). These data revealed that MTGM overexpression induces mitochondrial fragmentation. We also observed the ultrastructure of mitochondria by electron microscopy. It seemed that no marked internal structural changes were detectable in the MTGM-induced fragmented mitochondria except that they were seen as numerous small sphere-like structures. The inner mitochondrial membrane and the cristae were still present in these mitochondria (Fig. 2D).
Overexpression of MTGM results in inhibition of DNA synthesis, reduction of mitotic cells and stalling of cells in S phase
We subsequently examined the effect of MTGM overexpression on cell proliferation. We first used a BrdU incorporation assay to determine the fraction of cells undergoing DNA synthesis. 293T cells were transfected with MTGM-V5 or GFP expression vector as a control and analyzed by immunofluorescence. Results (Fig. 3A,B) showed that the number of BrdU-positive cells was dramatically lower (P<0.0001) in the MTGM-expressing cell population (4.3±2.0%) than in non-transfected (37.7±4.0%) or GFP-expressing cell populations (31.6±5.7%). This demonstrated that overexpression of MTGM significantly inhibits DNA synthesis.
To further evaluate the inhibitory effect of MTGM on cell proliferation, we assessed mitotic index of MTGM- and GFP-expressing cell populations using double analyses with anti-Ki67 antibody (a cell proliferation marker) and DAPI (for staining DNA). 2.3±0.9% of the GFP-expressing control cells were found to be in different stages of mitosis. By contrast, among MTGM-expressing cells examined from three independent transfections, none was found to undergo mitosis (Fig. 3C,D), suggesting that overexpression of MTGM totally excludes cells from mitosis.
We further studied the effect of MTGM overexpression on cell-cycle progression by a flow cytometry (FACS) analysis of DNA content. As shown in Fig. 3E, when the whole cell population was analyzed, overexpression of MTGM induced an increase of cells in the S phase (27.7±3.1%, P=0.012) relative to the distribution in control cells (empty vector, 19.3±1.1%) and a corresponding decrease of cells in the G0-G1 and G2-M phases of the cell cycle. However, no evident difference in the sub-G1 population (debris) was seen between MTGM-V5 transfected (7.53±3.0%) and empty vector transfected cells (10.1±2.4%, P=0.32), implying that expression of MTGM does not affect cell viability.
We also carried out the cell-cycle analysis separately on MTGM-V5-positive and V5-negative subpopulations of 293T cultures transfected with MTGM-V5 plasmid using immunolabeling with anti-V5 antibody, followed by FACS analysis of DNA content. Results revealed an apparent increase of S-phase cells (47.0%) and a corresponding decrease of G0-G1 cells (36.5%) in the MTGM-positive subpopulation compared to 34.5% S-phase cells and 53.2% G0-G1 cells in the MTGM-negative subpopulation (Fig. 3F). Taking all data together, we conclude that overexpression of MTGM halts cells in S phase of the cell cycle.
We subsequently examined whether MTGM overexpression affected the levels of cell-cycle regulators. Immunoblotting showed that the level of cyclin B1 was slightly increased in 293T cells transfected with MTGM-V5 for 24 hours, whereas cyclins D1, E and A, and Cdk inhibitors p21 and p27 were unaffected (supplementary material Fig. S4). A similar result was seen also at 48 hours post-transfection (data not shown). Additionally, we found that MTGM overexpression resulted in an increase of γ-H2AX-positive cells relative to control (Fig. 3G,H; P<0.0001), implying that MTGM overexpression and subsequent stalling of cells in S phase might be associated with DNA damage.
Overexpression of MTGM leads to a release of Smac/Diablo to the cytosol
To explore the potential effect of MTGM-induced mitochondrial fragmentation on release of proapoptotic factors from mitochondria, we first determined whether MTGM can lead to cyt c release. We showed that although the enforced expression of MTGM triggered mitochondrial fragmentation, cyt c was still retained in the fragmented mitochondria for 24 hours post-transfection (Fig. 4A) and for 48 hours post-transfection (data not shown). As a positive control, we showed that hFis1 triggered mitochondrial cyt c release in Myc-hFis1 overexpressing 293T cells (Fig. 4A). We subsequently examined the subcellular localization of Smac/Diablo (Smac) and found that 56.8±8.6% of cells expressing MTGM-V5 had lost Smac immunostaining in mitochondria (Fig. 4A,B). By contrast, mitochondrial Smac staining was observed in most non-transfected cells, and only 1.9±0.8% of these cells lacked Smac staining (Fig. 4B; P<0.0001). Immunoblotting further confirmed that the level of Smac decreased in mitochondria, whereas it increased in the cytosol of MTGM-V5 transfected cells. By contrast, the majority of cyt c was still retained in the mitochondrial fraction and there was only a minor increase of cyt c in the cytosol (Fig. 4C). Similarly, the majority of OPA1, Omi/HtrA2 and AK2 proteins were also retained in the mitochondrial fraction of cells expressing MTGM-V5 (Fig. 4C). Additionally, the death factor AIF was in mitochondria (Fig. 4A), and we could not demonstrate any translocation of Bax to mitochondria in cells expressing MTGM-V5 (data not shown).
Next, we examined whether MTGM overexpression affected OPA1 isoforms (Delettre et al., 2001), because MTGM is localized in the mitochondrial inner membrane like OPA1, which is involved in regulation of mitochondrial cristae morphology and release of cyt c during apoptosis (Estaquier and Arnoult, 2007; Frezza et al., 2006; Olichon et al., 2003). Immunoblotting using anti-OPA1 antibody revealed that MTGM expression did not affect the levels of several OPA1 isoforms in 293T cells (Fig. 4D) nor in HeLa cells (data not shown).
Overexpression of MTGM does not promote apoptotic and autophagic activities
We examined apoptotic activity in cells expressing MTGM-V5 using western blot. Active caspase 3 and cleaved PARP (polyADP-ribose polymerase) were assessed in eight cell lines. As shown in Fig. 5A, overexpression of MTGM did not increase the levels of either active caspase 3 (19-17 kDa) or cleaved PARP (∼89 kDa) in any of the analyzed cell lines transfected with MTGM-V5, except that a small increase of cleaved PARP was seen in 293T cells compared to control cells transfected with empty vector. Likewise, by using immunofluorescence with M30 (CytoDEATH) antibody (a marker for early apoptotic event), we did not detect any increase of M30-positive cells in cells expressing MTGM-V5 compared with non-transfected and empty-vector-transfected cells (data not shown). Neither did we detect any increase of autophagic activity in cells expressing MTGM-V5 using immunoblotting with anti-LC3B antibody (a marker of autophagy). MTGM did not induce a conversion of LC3B-I to LC3B-II (Fig. 5B).
Taking these data together with the results from FACS analysis, we suggest that overexpression of MTGM does not induce apoptotic and autophagic events, although it leads to mitochondrial fragmentation and a release of mitochondrial Smac to the cytosol. Moreover, inhibition of caspase activity by Z-VAD-FMK had no effect on mitochondrial fragmentation induced by MTGM-V5 (supplementary material Fig. S5).
MTGM-mediated mitochondrial fragmentation occurs before the release of Smac
To analyze the correlation between mitochondrial fragmentation and release of Smac induced by MTGM, we examined these effects at early time points post-transfection. We found that overexpression of MTGM triggered a rapid mitochondrial fragmentation at 8 hours post-transfection, at which time point 65.6±3.7% of MTGM-expressing cells exhibited fragmented mitochondria and only 3.9±1.0% of these cells showed release of mitochondrial Smac. Release of Smac was observed first at 24 hours (Fig. 5C). This marked difference in kinetics indicated that mitochondrial fragmentation occurred before Smac release.
MTGM-induced mitochondrial fission and Smac release require intact Drp1 function
Drp1 is a key component of the mitochondrial fission machinery (Chan, 2006). To examine whether the MTGM-induced fragmentation is coupled to the Drp1-mediated mitochondrial fission pathway, we coexpressed MTGM with HA-Drp1 and the dominant negative mutant HA-Drp1-K38A (Smirnova et al., 1998) in 293T cells. The transfected cells were co-immunostained with anti-V5 (for MTGM) and anti-HA (for Drp1 and its mutant) antibodies. Overexpression of Drp1-K38A alone induced mitochondrial fusion in 91.4% of transfected cells, whereas co-expression of wild-type Drp1 with MTGM-V5 had no effect on MTGM-induced fragmentation compared with MTGM-V5 alone (Fig. 6A,C). This is consistent with a previous report that overexpression of Drp1 does not affect mitochondrial fission due to an already abundant pool of endogenous Drp1 in the cytosol (Smirnova et al., 1998). By contrast, co-expression of MTGM with Drp1-K38A led to mitochondrial elongation and formation of the interconnected tubular cluster in most transfected cells (82.6%). Only 8.1% of MTGM/Drp1-K38A co-transfected cells exhibited fragmented mitochondria, indicating that Drp1-K38A highly blocked MTGM-induced mitochondrial fragmentation (Fig. 6B, upper panel, and Fig. 6C; P<0.0001). Further, Drp1-K38A significantly blocks MTGM-induced Smac release in most cotransfected cells (Fig. 6B, lower panel and Fig. 6D) compared with cells transfected with MTGM-V5 alone (also see Fig. 4B). The proportion of cotransfected cells with release of Smac was only 5.7% compared with 56.3% after transfection with MTGM-V5 alone (Fig. 6D; P<0.0001). This indicated that MTGM-induced mitochondrial fission and Smac release both require intact Drp1 function.
Depletion of endogenous MTGM induces mitochondrial elongation
To further study MTGM function, knockdown of endogenous MTGM was performed by RNA interference (RNAi). Four sets of siRNAs specific to MTGM and a scrambled control siRNA were separately transfected into 293T cells. siMTGM-10 and siMTGM-7 significantly decreased endogenous MTGM protein to less than 10%, as analyzed by immunoblotting with anti-MTGM antibody (Fig. 7A). siMTGM-8 led only to a slight reduction of endogenous MTGM, and siMTGM-6 did not affect expression of MTGM (data not shown). Using confocal microscopy, we showed that reduction of endogenous MTGM by siMTGM-10 significantly increased the number of cells with elongated mitochondria (58.8±7.7%, P<0.0001) compared with control cells (16.8±3.6%) (Fig. 7B,C). Similarly, siMTGM-7 resulted in a marked increase in number of cells with elongated mitochondria (40.7±3.6%, P<0.0001) (Fig. 7C; supplementary material Fig. S6A). We also showed that depletion of MTGM by siMTGM-10 and siMTGM-7 led to mitochondrial elongation in HeLa cells (supplementary material Fig. S6B). All these results indicate that endogenous MTGM is required for mitochondrial fission.
Knockdown of endogenous MTGM results in increased cell growth and in resistance to staurosporine-induced apoptosis
We noticed that cell growth was evidently better in cells treated with siMTGM-10 and siMTGM-7 than in cells treated with control siRNA, whereas cells treated with siMTGM-8 and siMTGM-6 were similar to cells treated with control siRNA (data not shown). Consistent with this, we found that the number of cells incorporating BrdU was significantly higher in MTGM-depleted cells (49.8±1.5%) than in control-siRNA-treated cells (28.8±2.9%, P<0.0001) (Fig. 8A,B). These data indicate that depletion of MTGM is favorable for cell growth. Further, we showed that the levels of cyclin D1 and the cdk inhibitor p21 were lower in MTGM-depleted cells, whereas the level of cyclinB1 was slightly increased (Fig. 8C), which is consistent with an increased cell-cycle progression.
It has been reported that inactivation of the components involved in mitochondrial fusion, such as Mfn1 and Mfn2 (Chen et al., 2005; Sugioka et al., 2004) and OPA1 (Lee et al., 2004), induces mitochondrial fragmentation and increases the sensitivity of cells to apoptosis induced by stimuli. In contrast to this, depletion of hFis1 and Drp1 by RNAi induces mitochondrial fusion and inhibits apoptosis (Lee et al., 2004). Therefore, we tested how depletion of MTGM affected the response of cells to apoptosis stimuli. To this end, staurosporine (STS) (Lee et al., 2004) was added at indicated time points to 293T cells transfected with MTGM siRNA or control siRNA, and cleavage of PARP was used to assess apoptosis by immunoblotting. We found that the induced activity of caspase 3 (indicated by PARP cleavage product) was much lower in MTGM RNAi cells than in control RNAi cells (Fig. 8D), indicating that downregulation of endogenous MTGM made the cells apoptosis-resistant. Likewise, depletion of MTGM led to a decrease of the sub-G1 cell fraction representing apoptotic cells induced by STS treatment (FACS analysis of DNA content; data not shown). Finally, STS-induced apoptotic cells were evaluated using M30 CytoDeath antibody labeling and FACS analysis. Consistently, the number of M30-positive cells was evidently lower in the cells treated with MTGM siRNA (7.6%) than in cells treated with control siRNA (17.5%) (Fig. 8E). All these data indicate that depletion of MTGM delays apoptosis induced by staurosporine. This is consistent with the idea that mitochondrial fusion increases resistance of cells to apoptosis (Chen and Chan, 2005).
In this report, we demonstrate that the highly conserved MTGM protein is located in mitochondria and is an integral mitochondrial inner-membrane protein. Overexpression of MTGM resulted in mitochondrial fragmentation and inhibition of cell proliferation, whereas depletion of endogenous MTGM led to mitochondrial elongation, increased cell proliferation and reduced STS-induced apoptosis. These new findings illustrate that MTGM plays an important role in the regulation of mitochondrial fission and cell proliferation. Although MTGM overexpression led to a release of Smac to the cytosol, this did not promote obvious apoptotic and autophagic activities. However, the fact that depletion of endogenous MTGM prevented STS-induced apoptosis indicates that endogenous MTGM still has a role in the apoptotic process.
Mitochondria are dynamic organelles, and their morphology is controlled by a balance of fission and fusion events. Several proteins have been described to be required for mitochondrial fission in mammals, including Drp1, Fis1, MTP18, endophilin B1 and GDAP1 (for a review, see Chan, 2006). Drp1 and endophilin B1 are mainly present in the cytosol, but they cycle dynamically between the cytosol and the outer mitochondrial membrane (Karbowski et al., 2004; Pitts et al., 1999; Smirnova et al., 1998). Fis1 and GDAP1 are integral mitochondrial outer-membrane proteins (James et al., 2003; Niemann et al., 2005). Fis1 plays a crucial role in controlling fission at the outer mitochondrial membrane by limiting recruitment of Drp1 to mitochondria (Yoon et al., 2003). MTP18 is located in the intermembrane space and is thought to regulate mitochondrial fission. Overexpression of MTP18 leads to mitochondrial fragmentation, and downregulation leads to mitochondrial elongation (Tondera et al., 2005; Tondera et al., 2004). However, mitochondrial inner-membrane proteins involved in the fission process remain poorly understood (Suen et al., 2008).
In this work, we demonstrate that MTGM is an integral mitochondrial inner-membrane protein and, to our knowledge, it is the first integral mitochondrial inner-membrane protein described as being involved in regulation of mitochondrial fission. MTGM is an evolutionarily highly conserved protein with orthologs throughout the entire eukaryotic kingdom. Its orthologs are 100% identical at the amino acid sequence level in all analyzed mammalian species. To our knowledge, such a shared protein in the mammalian kingdom is very rare in evolution. Its highly conserved features suggest that this protein has a crucial role in mitochondrial function. Moreover, the localization of MTGM in the inner membrane implies that the protein is involved in the fission process, possibly by affecting the mitochondrial inner membrane. It has been suggested that coordination of the outer and inner membranes in the fission process is likely to be required for mitochondrial division (Heath-Engel and Shore, 2006). Consistent with this hypothesis, we demonstrate that inhibition of the outer-membrane fission activity by coexpressing MTGM with a dominant-negative Drp1 mutant (Drp1-K38A) blocks MTGM-mediated mitochondrial fragmentation, indicating that intact Drp1 function is required for the MTGM-mediated effect. Therefore, there is a potential functional interaction between these two proteins, not necessarily involving direct protein-protein interaction. We also show that either disruption of Drp1 activity (involving the outer membrane) by overexpression of Drp1-K38A, or depletion of MTGM (involving the inner membrane) by siRNA alone (inhibiting the respective fission process) both result in elongated mitochondria. However, the mechanism(s) by which mitochondrial outer and inner membranes can coordinately perform these processes remains to be elucidated. The GxxxG motif has been found in plasma membrane proteins and is known to be important for mediating protein-protein interactions (Russ and Engelman, 2000). The presence of such a highly conserved tetrad of the GxxxG motif in all orthologs of MTGM suggests that the motif might play a key role in the function of MTGM. However, additional experimentation will be required to elucidate the potential role of this motif in MTGM.
Smac is normally present in mitochondria, and mature Smac protein is released along with cyt c during most situations of apoptosis and promotes cyt-c-dependent caspase activation by neutralizing inhibitor of apoptosis proteins (IAPs) (Du et al., 2000; Munoz-Pinedo et al., 2006; Rehm et al., 2003). In this study, we report that MTGM-mediated mitochondrial fragmentation induces a release of mitochondrial Smac to the cytosol, but does not induce activation of apoptosis and autophagy, suggesting that release of mitochondrial Smac alone is not sufficient to induce apoptosis. This is consistent with a previous report that expression of a mature form of Smac in the cytosol of mammalian cells does not promote apoptosis (Hao et al., 2004). The release of Smac might be taken as an indication that MTGM overexpression affects mitochondrial outer-membrane permeabilization to some degree. Additionally, OPA1, a protein essential for the inner mitochondrial membrane fusion (Meeusen et al., 2006), has been shown to be involved in cristae remodeling and required for complete release of cyt c during apoptosis (Frezza et al., 2006). We show that MTGM overexpression has no effect on the levels of several OPA1 isoforms, implying that MTGM probably does not disrupt cyt c within mitochondrial cristae controlled by OPA1. Taken together, our findings provide new evidence that excessive mitochondrial fragmentation can occur independently, with no effect on apoptosis (Alirol et al., 2006), although mitochondrial fission seems to be usually associated with apoptosis.
It is becoming increasingly clear that the proteins involved in mitochondrial dynamics are associated with a broad range of cellular functions, impacting on cell growth and survival, cell death and senescence, and also on energy production and calcium metabolism (Berman et al., 2008; de Brito and Scorrano, 2008; Grandemange et al., 2009; Parone et al., 2008; Suen et al., 2008). Several recent reports have shown that changes in mitochondrial morphology occur during cell division (Arakaki et al., 2006; Lee et al., 2007) and that Drp1 is phosphorylated by Cdk1-cyclin B in this process (Taguchi et al., 2007). We show that overexpression of MTGM results in mitochondrial fragmentation, stalling of cells in S-phase-associated DNA damage indicated by γ-H2AX staining, and exclusion of cells from mitosis, whereas depletion of MTGM results in mitochondrial elongation and cell-cycle progression. All these findings indicate a coupling between mitochondrial dynamics and cell-cycle progression. However, it remains unclear how MTGM affects cell-cycle progression. We speculate that the forced mitochondrial fragmentation resulting from MTGM overexpression might prevent the normal cell-cycle-dependent mitochondrial changes and affect energy production, thereby inhibiting cell-cycle progression. Biogenesis of mitochondria in S-phase cells and in cells preparing for mitosis normally involves extensive tubular networks (Lee et al., 2007).
In systemic genome-wide functional analysis of genes using RNA interference, it was found that loss of F15D4.3, a Caenorhabditis elegans ortholog of MTGM (with 50.6% identity), is non-lethal (Kamath et al., 2003; Sonnichsen et al., 2005). In yeast, the YPL098C deletion mutant (a putative Saccharomyces cerevisiae ortholog exhibiting 43.6% identity with MTGM) was shown to be viable (Giaever et al., 2002), but to be deficient in sporulation (Deutschbauer et al., 2002). Mitochondrial dynamics have been reported to be important in yeast sporulation. Yeast cells lacking fission proteins alone or both fission and fusion proteins produced more non-viable spores than did wild-type cells (Gorsich and Shaw, 2004).
Little is known about the possible role of the mitochondrial fission machinery in human diseases. Only GDAP1, known currently as a component of the fission machinery (Niemann et al., 2005), is associated with human Charcot-Marie-Tooth (CMT) disease because mutations in GDAP1 lead to severe forms of the peripheral neuropathy. In this work, we report that MTGM is a novel component of the mitochondrial inner membrane and is involved in regulating fission, and that it is upregulated in human brain tumors. Consistent with our findings, Yoo and coworkers (Chung et al., 2006) showed in a recent study that Romo1 (MTGM) is upregulated in several cancer cell lines. However, the role of MTGM (Romo1) in mitochondria of tumor cells is completely unknown. In two glioma cell lines, U-343 MG and U-343 MGa Cl2:6, mitochondrial morphology is normal with a semi-tubular network (data not shown) in spite of the fact that MTGM is upregulated in the cell lines. It should be stressed that the morphology of mitochondria is controlled by multiple components of the fission-fusion machineries. Therefore, we speculate that a re-balance of the different components of these two machineries might occur in the tumor cells to maintain a mitochondrial shape that is suitable for growth of tumor cells. Consistent with this hypothesis, analysis of a large number of microarray gene expression data from the Gene Expression Omnibus at NCBI (http://www.ncbi.nlm.nih.gov/geo/) (data not shown) showed that multiple components of the mitochondrial fission-fusion machineries (such as Drp1, GDAP1 and OPA1) were deregulated in human brain tumor tissues. The potential role of mitochondrial dynamics and the components of the mitochondrial fission-fusion machineries in brain tumors remain to be elucidated.
Materials and Methods
Serial Analysis of Gene Expression (SAGE) libraries (www.cgap.nci.nih.gov) were used for analysis of gene expression profiles. LOCtree (http://cubic.bioc.columbia.edu) and Lokero (http://www-bs.informatik.uni-tuebingen.de) were used for prediction of subcellular localization. The TMHMM was used for prediction of transmembrane regions. Multiple alignments were generated with the CLUSTALW (http://npsa-pbil.ibcp.fr/). SignalP3.0 (www.cbs.dtu.dk/services/SignalP/); TargetP 1.1 (www.cbs.dtu.dk/services/TargetP/) were used for prediction of a mitochondrial targeting sequence.
Mouse monoclonal antibodies (mAb) were used for: V5-tag (Invitrogen), cyt c (clone 7H8.2C12), cyt c (clone 6H2.B4), Smac/Diablo, PARP, Tom20, Tim23, OPA1 and p21 (BD Biosciences); BrdU (Amersham); GAPDH (Abcam); p27 and cyclin B1 (Santa Cruz); c-Myc-tag (Sigma); HA-tag (Nordic Biosite, Sweden); and Ki67 (Novocastra). Rabbit polyclonal antibodies (pAbs) were used for: Omi/HtrA2, caspase 3 and LC3B (Cell Signaling); AK2 and V5-tag (Abcam); and cyclin A, cyclin D1 and cyclin E (Santa Cruz). Secondary antibodies included the Rhodamine- and FITC-conjugated anti-mouse and anti-rabbit IgG (Abcam) for immunofluorescence and the peroxidase-conjugated anti-mouse and anti-rabbit IgG (Amersham). Rabbit polyclonal antibodies specific to MTGM were raised against the peptide sequence (N)-CIGMRGRELMGGIGKT-(C) (corresponding to aa 45-59 of MTGM) and affinity-purified against the immunogenic peptide by Innovagen (Lund, Sweden).
Isolation of human MTGM cDNA and sequence analysis
To obtain the full-length coding region of MTGM cDNA, a pair of primers (ex2b-F1a and ex3-R1) were designed (supplementary material Table S3) according to NM_080748. RT-PCR was performed by using the Human Fetal Brain BD Marathon-Ready cDNA (Clontech) as template. The PCR product was cloned into a pGEM-T easy vector (Promega). The generated recombinant plasmid (pGEM-MTGM) carrying the whole coding region of MTGM cDNA was verified by sequencing.
Rapid amplification of cDNA ends (RACE)
RACE-PCR for derivation of the 5′-ends of MTGM cDNAs was done using the Human Fetal Brain BD Marathon-Ready cDNA (Clontech). The outer primers ex3-R1 and Adaptor1 were used in the first round of RACE-PCR. Subsequently, the primary RACE-PCR product was used as template in a nested RACE-PCR with the inner primers ex3-end-R and Adaptor2. The purified RACE-PCR fragments were cloned into a pGEM-T easy vector (Promega) and sequenced.
RT-PCR analysis of MTGM gene expression in human tissues and tumor cell lines
Total RNA from U-343 MGa, Cl2:6 and SH-SY5Y cells, as well as from nine brain tumor samples obtained from children with medulloblastoma (six patients) and astrocytoma grade I-II (three patients) was prepared using the RNeasy Miniprep Kit (Qiagen). The ThemoScript RT-PCR System (Invitrogen) was used to perform the first strand cDNA synthesis. For analysis of expression profiles of MTGM, these cDNA templates and the Human Fetal Brain as well as the Human Brain (adult) Whole BD Marathon-Ready cDNAs (Clontech) were used in PCR reactions. For normal tissue distribution of MTGM expression, the Human Adult MTC Panel containing cDNAs from multiple tissues (Clontech) and the primers ex2-F1 and ex3-R1 (supplementary material Table S3) were used for PCR, resulting in a 224 bp band. All MTGM results were normalized against a β-actin control using a gel optical density analyzer. The use of human samples was approved by the local ethics committee (Ups d nr 02-566).
The cDNA with the full-length ORF of MTGM was amplified by PCR from the pGEM-MTGM plasmid using the primers (ex2b-F1a and ex3-end-R) and subcloned into the pcDNA3.1/V5-His-TOPO vector (Invitrogen) to produce a C-terminally V5-His-tagged MTGM. The generated pcDNA3.1-MTGM/V5-His (MTGM-V5) plasmid was verified by sequencing. We also generated an N-terminally HA-tagged MTGM plasmid in pcDNA3.1 (HA-MTGM). Expression constructs for HA-tagged human Drp1 and mutant Drp1-K38A, and Myc-tagged human Fis1 (hFis1) were kindly provided by Alexander M. van der Bliek (Smirnova et al., 1998) and Yisang Yoon (Yu et al., 2005), respectively.
Cell cultures and transfection
293T, HeLa, Cos-7 and MCF-7 cells were grown in Dulbecco's modified Eagle's medium with 10% FCS. The two clonal human malignant glioma cell lines, U-343 MG and U-343 MGa Cl2:6 (Cl2:6), and the neuroblastoma cell line SH-SY5Y were grown in Eagle's minimum essential medium with 10% FCS. U2OS osteosarcoma cells, HCT116 p53+/+ and HCT116 p53–/– colon carcinoma cells were grown in Iscove's modified Dulbecco's medium with 10% FCS. Transient transfections were performed using Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen). Expression plasmids for MTGM-V5, HA-MTGM, pcDNA3.1-GFP/His, HA-Drp1, HA-Drp1-K38A, Myc-hFis1 and pcDNA3.1-V5/His empty vector were used.
Protein extracts were separated by electrophoresis and transferred to PVDF membranes (Millipore). After blocking with 10% nonfat dry milk in PBS, membranes were incubated with primary antibodies and the peroxidase-conjugated secondary antibody (Amersham), and immunocomplexes were detected with an enhanced chemiluminescence (ECL) kit (Amersham).
Cells on coverslips were fixed with 4% paraformaldehyde in PBS, permeabilized with 0.5% Triton X-100 in PBS, blocked in PBS with 2% BSA and incubated with the primary antibodies and Rhodamine- or FITC-conjugated secondary antibodies. The stained cells were mounted with the Mounting Medium containing DAPI (Vector Laboratories). For mitochondrial staining, MitoTracker Red (Molecular Probes) was added to cultures for 15 minutes before fixation. Treatment with pan caspase inhibitor Z-VAD-FMK (BD, Biosciences) was started at 2.5 hours post-transfection with MTGM-V5 vector, at which time the cells were washed and incubated in fresh medium containing 100 μM Z-VAD-FMK for 24 hours. Specimens were examined by fluorescence microscopy (Zeiss) or confocal microscopy (Leica).
Transmission electron microscopy
Transfected 293T cells were cultured for 24 hours, fixed with 2% paraformaldehyde or 2% glutaraldehyde in cacodylate buffer pH 7.4 at 4°C, and post-fixed with 1% OsO4 at 4°C. The specimens were en-block stained with 1% uranyl acetate in 70% alcohol, dehydrated in a series of alcohols and embedded in Durcupan ACM (Fluka) resin. Series of sections were counterstained with uranyl acetate and lead citrate and analyzed in a Tecnai 12 (FEI) electron microscope.
The mitochondrial fraction was isolated as described (Parone et al., 2006). 293T cells were collected in ice-cold mitochondrial buffer (MB) (210 mM mannitol, 70 mM sucrose, 10 mM HEPES, 1 mM EDTA, pH 7.5) with proteinase inhibitors (Roche). Cells were broken by 15 passages through a 25-gauge needle, and the suspension was centrifuged for 5 minutes at 1500 × g at 4°C. The resulting supernatant (whole cell lysate) was further centrifuged at 16,000 × g for 30 minutes. The resulting supernatant (cytosolic fraction) was separated from the pellet (mitochondrial fraction). For proteinase-K digestion, mitochondria were suspended in MB buffer with 50 μg/ml proteinase K and incubated for 30 minutes at room temperature. Digestion was terminated with 5 mM phenylmethylsulfonyl fluoride. For analysis of membrane protein, mitochondrial pellets were resuspended in 100 μl of MB buffer, or MB buffer containing 0.1 M Na2CO3 (pH 11.5) and incubated on ice for 30 minutes. The insoluble membrane fractions were centrifuged at 16,000 × g for 15 minutes, and the supernatants precipitated with 10% (v/v) trichloroacetic acid.
293T cells were incubated with BrdU (20 μM/ml) in medium for 4 hours before harvest, and fixed in 1% formalin in PBS for 8 minutes, followed by 1% formalin, 0.3% Triton X-100 in PBS for 18 hours at 4°C. BrdU immunolabeling was performed with anti-BrdU antibody and detected with secondary antibodies conjugated with Rhodamine. For analysis of BrdU incorporation in cells expressing MTGM-V5, BrdU-labeled cells were subsequently immunostained with anti-V5 pAb.
Cell cycle analysis by flow cytometry (FACS)
The cell-cycle distribution was determined by FACS analysis of DNA content. Briefly, 293T cells were harvested at indicated times post-transfection, fixed in 70% ethanol and stained with DAPI. DNA content was analyzed using a BD LSR II flow cytometer.
Immunostaining of MTGM-V5-positive cells for FACS analysis of DNA content was performed as described (Berger and Haupt, 2003). Briefly, cells were harvested, fixed with cold methanol (–20°C), incubated with anti-V5 mAb followed by FITC-conjugated secondary antibody. DNA was stained with DAPI, and the cell-cycle distribution of MTGM-V5-positive and MTGM-V5-negative cells analyzed by FACS. For negative control, transfected cells were stained only with FITC-conjugated secondary antibody, and anti-V5 mAb was omitted.
RNA interference silencing of MTGM expression and test of effects on apoptosis
Four sets of siRNA duplexes specific for human MTGM mRNA: siMTGM-6 (sense 5′-GAGAGCUGAUGGGCGGCAUUGGGAA-3′), siMTGM-7 (sense 5′-GAGAGCUGAUGGGCGGCAUUGGGAA-3′), siMTGM-8 (sense 5′-CAGAGUGGCGGCACCUUUGGCACAU-3′), siMTGM-10 (sense 5′-CCAUUGGGAUGGGCAUCCGAUGCUA-3′) and a scrambled siRNA duplex (12935-400) with similar GC content recommended by the manufacturer were purchased from Invitrogen. RNA interference was carried out with Lipofectamine 2000 (Invitrogen). Cells were retransfected with siRNA duplex 36 hours after initial transfection and incubated for another 36 hours. In some experiments, cells were incubated with BrdU (20 μM/ml) in medium for 4 hours or treated with staurosporine (1.5 μM; Sigma) for indicated times (see Fig. 8D) before harvest.
Apoptosis and autophagy detection
Immunoblotting was performed with caspase 3, PARP and LC3-B1 antibodies, and active caspase 3, cleaved PARP product, and a conversion of LC3B-I to LC3B-II (a marker of autophagy) were used to assess activities of apoptosis and autophagy, respectively. Sub-G1 cell fraction representing non-viable cells was assessed by FACS analysis of DNA content. Apoptotic cells were analyzed by immunostaining with M30 antibody using a M30 CytoDEATH detection kit (Roche) followed by fluorescence microscopy and FACS analysis. As negative controls, 293T cells were labeled only with FITC-conjugated secondary antibody.
The unpaired Student's t-test was applied to evaluate differences between experimental groups. P values less than or equal to 0.05 were considered as statistically significant.
This work was supported by grants from the Swedish Cancer Society, the Swedish Childhood Cancer Foundation, the Cancer Society in Stockholm, Karolinska Institutet and the Swedish Research Council (VR-M). We would like to thank Alexander M. van der Bliek (Department of Biological Chemistry, UCLA School of Medicine, CA) for providing HA-Drp1 and HA-Drp1-K38A expression vectors, and Yisang Yoon (Department of Anesthesiology and of Pharmacology and Physiology, University of Rochester School of Medicine and Dentistry, NY) for Myc-hFis1 expression vector. We thank Mikael Lindström for valuable discussions. We also thank Dan Grandér for advice on apoptosis detection. The cDNA sequences described in this work have been deposited in GenBank under the accession numbers MTGM-α: AM397244, -β: AM397245 and -γ: AM397246