Membrane fusion plays an important role in controlling the shape, number, and distribution of mitochondria. In the yeast Saccharomyces cerevisiae, the outer membrane protein Fzo1p has been shown to mediate mitochondrial fusion. Using a novel genetic screen, we have isolated new mutants defective in the fusion of their mitochondria. One of these mutants, ugo1, shows several similarities to fzo1 mutants. ugo1 cells contain numerous mitochondrial fragments instead of the few long, tubular organelles seen in wild-type cells. ugo1 mutants lose mitochondrial DNA (mtDNA). In zygotes formed by mating two ugo1 cells, mitochondria do not fuse and mix their matrix contents. Fragmentation of mitochondria and loss of mtDNA in ugo1 mutants are rescued by disrupting DNM1, a gene required for mitochondrial division. We find that UGO1 encodes a 58-kD protein located in the mitochondrial outer membrane. Ugo1p appears to contain a single transmembrane segment, with its NH2 terminus facing the cytosol and its COOH terminus in the intermembrane space. Our results suggest that Ugo1p is a new outer membrane component of the mitochondrial fusion machinery.
Mitochondrial fusion is a fundamental process required for establishing and maintaining the specialized shapes and numbers of mitochondria in many cell types (Tyler 1992; Bereiter-Hahn and Voth 1994). In the yeast Saccharomyces cerevisiae, mitochondrial fusion and its opposite activity, division, are highly regulated during growth, mating, and sporulation (Hermann and Shaw 1998; Yaffe 1999; Jensen et al. 2000). During vegetative growth, mitochondria constitutively fuse, divide (Nunnari et al. 1997), and change their number depending on growth conditions (Stevens 1977). For instance, exponentially growing cells contain several branched, tubular mitochondria. When cells enter stationary phase, the mitochondrial tubules fragment into numerous small organelles (Hoffman and Avers 1973; Stevens 1977). When yeast cells mate, mitochondria fuse immediately after cell fusion, mixing their contents, including mitochondrial DNA (mtDNA) and proteins (Thomas and Wilkie 1968; Dujon 1981; Nunnari et al. 1997; Okamoto et al. 1998). When diploids go through meiosis and sporulation, mitochondria undergo several fusion and division events, eventually encircling each of the four haploid nuclei (Miyakawa et al. 1984).
Mitochondrial fusion in yeast requires the Fzo1 protein (Hermann et al. 1998; Rapaport et al. 1998). Fzo1p was identified as a homologue to the Drosophila fuzzy onions protein, which is required for mitochondrial fusion during fly spermatogenesis (Hales and Fuller 1997). Fzo1p is a mitochondrial outer membrane protein with a cytosolic GTPase domain at its NH2 terminus (Hermann et al. 1998; Rapaport et al. 1998). In FZO1 disruption mutants, cells contain many small mitochondrial fragments instead of the few tubular mitochondria seen in wild-type cells (Hermann et al. 1998; Rapaport et al. 1998), and a defect in mitochondrial fusion in fzo1 mutants has been directly demonstrated using a mating assay (Hermann et al. 1998). In addition to fusion, Fzo1p is also important for maintenance of mtDNA. fzo1 mutants lack mtDNA, but the mechanism by which mtDNA is lost in fzo1 cells is not understood (Hermann et al. 1998; Rapaport et al. 1998).
The fragmentation of mitochondria in fzo1 mutants depends on mitochondrial division. Dnm1p is a dynamin-related GTPase (Gammie et al. 1995; Otsuga et al. 1998) and dnm1 mutants are defective in mitochondrial division (Bleazard et al. 1999; Sesaki and Jensen 1999). Cells disrupted for DNM1 contain a single mitochondrion consisting of a network of interconnected tubules. In dnm1 fzo1 double mutants, normal tubular-shaped mitochondria are seen, suggesting that mitochondrial shape and number is normally controlled, at least in part, by a balance between division and fusion mediated by Dnm1p and Fzo1p, respectively (Sesaki and Jensen 1999). Like fragmentation, the loss of mtDNA in fzo1 mutants requires Dnm1p and fzo1 dnm1 double mutants to maintain mtDNA (Bleazard et al. 1999; Sesaki and Jensen 1999; Jensen et al. 2000).
In this report, we have used a novel genetic screen to isolate new yeast mutants defective in mitochondrial fusion. We show that one of these mutants, ugo1, identifies a new mitochondrial outer membrane protein required for the fusion of mitochondria.
Materials and Methods
Strains, Media, and Genetic Methods
Yeast strains used in this study are listed in Table. Yeast media, including YEPD (yeast-extract peptone [YEP] medium containing 2% glucose), YEPGE (YEP medium containing 2% glycerol and 2% ethanol), YEPGal (YEP medium containing 2% galactose), SD (synthetic medium containing 2% glucose), SRaf (synthetic medium containing 2% raffinose), SGal (synthetic medium containing 2% galactose), SGalSuc (synthetic medium containing 2% galactose and 2% sucrose), 5FOAD (synthetic medium containing 0.1% 5-fluoro-orotic acid [5FOA] and 2% glucose), and 5FOAGE (synthetic medium containing 0.1% 5FOA, 2% glycerol, and 2% ethanol) are as described (Boeke et al. 1984). Standard molecular genetic techniques were used (Adams et al. 1997).
pHS29, a CEN-LEU2 plasmid containing DNM1-111 (Jensen et al. 2000) tagged with the triple influenza hemagglutinin (HA) epitope (Field et al. 1988) at the COOH terminus (Dnm1p-111-HA), was constructed as follows. The DNM1-111 gene, which contains two mutations in the GTPase domain of Dnm1p (Jensen et al. 2000), was PCR amplified from yeast genomic DNA (Hoffman and Winston 1987) prepared from DNM1-111 strain, YHS15 (Jensen et al. 2000), using oligos 268 (5′-CCGCTCGAGGAATACGATACAGAGGAAG-3′) and 269 (5′-ATAAGAATGCGGCCGCCCAGAATATTACTAATAAG-3′). PCR product was digested with XhoI and NotI, and subcloned into XhoI-NotI–digested pAA3 (Sesaki and Jensen 1999).
pHS50, a 2μ-URA3 plasmid expressing Dnm1p-111–HA, was constructed by cotransforming a PvuII fragment from pHS29 with PvuII- and XhoI-HindIII–digested pRS426 (Sikorski and Hieter 1989) into yeast cells. Homologous recombination between the DNM1-111–HA fragment and the linearized plasmid (Oldenburg et al. 1997) created plasmid pHS50.
pHS58, a CEN-LEU2 plasmid containing the UGO1 gene, was constructed by PCR amplifying UGO1 from yeast genomic DNA using oligos 446 (5′-GGGCTCGAGTGATTTCTTTGAGCGAC-3′) and 448 (5′-GAAGCGGCCGCTAGACAAGTGGTGGAGG-3′). The PCR product was subcloned into XhoI-NotI–digested pRS314 (Sikorski and Hieter 1989).
pHS51, a CEN-URA3 plasmid expressing red fluorescent protein (RFP) fused to the presequence of Cox4p under the control of the GAL1 promoter was constructed as follows. RFP was PCR amplified from plasmid ST10 (a gift from B. Glick, University of Chicago, Chicago, IL) using oligos 461 (5′-GGGTCTAGACCACCGGTCGCCACCATG-3′) and 462 (5′-ATCTAGAGTCGCGGCCG-3′). The PCR fragment was digested with XbaI and NotI and subcloned into XbaI-NotI–digested pHS12 (Sesaki and Jensen 1999), forming pHS12-RFP. COX4-RFP was PCR amplified from pHS12-RFP using oligos 473 (5′-GGGCTCGAGATGCTTTCACTACGTCAATC-3′) and 474 (5′-GGGGGATCCCTACAGGAACAGGTGGTG-3′). The PCR fragment was digested with XhoI and BamHI, and subcloned downstream of the GAL1 promoter in XhoI-BamHI–digested pRS316GU (Nigro et al. 1992).
pHS52, a CEN-URA3 plasmid expressing cyan fluorescent protein (CFP) fused to the presequence of Cox4 from the GAL1 promoter, was constructed by PCR amplifying CFP from pECFP (CLONTECH Laboratories, Inc.) using oligos 355 (5′-TGCTCTAGAATGGTGAGCAAGGGC-3′) and 498 (5′-CCGGATCCTTACTTGTACAGCTCGTC-3′). The PCR fragment was digested with XbaI and BamHI and subcloned into XbaI-BamHI–digested pHS51.
pHS57, a CEN-TRP1 plasmid expressing Ugo1p with the triple myc epitope (Munro and Pelham 1986) at its NH2 terminus, was constructed by amplifying the promoter region of UGO1 from yeast genomic DNA (Hoffman and Winston 1987) using oligos 446 and 517 (5′-GGGATCGATTGGGGGGTTGAGTTAAAC-3′). The PCR product was digested with XhoI and ClaI and subcloned into XhoI-ClaI–digested pRS314 (Sikorski and Hieter 1989), forming pHS57-1. The UGO1 open reading frame was PCR amplified using oligos 518 (5′-GGGATCGATATGAACAACAATAATGTTAC-3′) and 448 (5′-GAAGCGGCCGCTAGACAAGTGGTGGAGG-3′), digested with ClaI and NotI, and subcloned into ClaI-NotI–digested pHS57-1, forming pHS57-2. The triple myc epitope was PCR amplified from KB241 (a gift from D. Kornitzer, S. Kron, and G. Fink, Whitehead Institute, Cambridge, MA) using oligos 519 (5′-GGGATCGATGAGCTCGGTACCCGGGG-3′) and 520 (5′-GGGATCGATGCATGCCTGCAGGTCGAC-3′), digested with ClaI, and subcloned into ClaI-digested pHS57-2, forming pHS57.
pHS55, a CEN-LEU2 plasmid containing Ugo1p with the triple HA epitope at its COOH terminus, was constructed by amplifying UGO1 as from yeast DNA using oligos 446 and 447 (5′-GAAGCGGCCGCCGAACTTCTCTTGTTCCATG-3′). The PCR product was digested with XhoI and NotI, and subcloned into XhoI-NotI–digested pAA3 (Sesaki and Jensen 1999).
Strains YHS60 and YHS61 were constructed by transforming MATa and MATα W303 strains (Thomas and Rothstein 1989) with pHS50, a 2μ-URA3 plasmid which expresses Dnm1p-111–HA. These parental strains were mutagenized with 3% ethane methylsulfonate (Sigma-Aldrich) in 50 mM potassium phosphate, pH 7.0, for 1.5 h at 30°C to ∼30% survival (Lawrence 1991). Cells were plated onto SD medium lacking uracil at 30°C and formed 16,000 colonies after 7 d. Cells were then replica-plated onto 5FOAD medium. We isolated 11 mutants that gave rise to red colonies on a synthetic medium containing 2% glucose (SD-Ura), but formed white colonies on 5FOAD medium. These mutants were also replica-plated from SD-Ura medium onto YEPGE and 5FOAGE media. All 11 mutants were viable on YEPGE, but failed to grow on 5FOAGE. Of those, eight mutants were found to be defective in a single gene in crosses to wild-type strains YHS1 and YHS2 (Sesaki and Jensen 1999) and were further analyzed. For complementation analysis, mutants were crossed to fzo1Δ cells (YHS21 and YHS22) and to mgm1Δ cells (strains 2467 and 12467).
Identification of the UGO1 Gene
ugo1-1 leu2 strain YHS62 was transformed with a library of yeast genomic DNA carried in the CEN-LEU2 plasmid p366 (a gift from F. Spencer, Johns Hopkins University). 5,000 Leu+ transformants were selected and tested for growth on 5FOAGE medium. Eight colonies that grew on 5FOAGE medium were isolated and plasmid DNA was obtained from each (Hoffman and Winston 1987). Plasmids were partially sequenced and were found to contain overlapping fragments of chromosome IV. To localize the region of the complementing activity, we digested one of the complementing plasmids, pa3-1/1A, with XbaI and religated, forming pa3-1/1A/XbaI. pa3-1/1A/XbaI complemented the ugo1-1 growth defect and carried two open reading frames, YDR469w and YDR470c. To test which open reading frame contains UGO1, pa3-1/1A/XbaI was digested using NcoI and religated to remove YDR470c. The resulting plasmid failed to complement the ugo1-1 growth defect. To confirm that YDR470c carries the complementing activity, we constructed pHS58, which carries only YDR470c, and showed that this plasmid rescued the growth defect on 5FOAGE of ugo1-1. Furthermore, we constructed cells disrupted for YDR470c (YHS73; see below) and crossed them to ugo1-1 cells. The resulting diploids failed to grow on 5FOAGE.
Complete disruptions of the UGO1, FZO1, and DNM1 genes were constructed by PCR-mediated gene replacement as described (Lawrence 1991) into diploid strain FY833/844 (Winston et al. 1995). For ugo1::HIS3, we used the HIS3 gene from plasmid pRS303 (Sikorski and Hieter 1989) and for fzo1::KAN and dnm1::KAN, we used kanMX4 plasmid pRS400 (Brachmann et al. 1998). Heterozygous diploids were sporulated and dissected to obtain MATa ugo1Δ strain YHS72, MATα ugo1Δ strain YHS73, MATa fzo1Δ strain YHS74, MATα fzo1Δ strain YHS75, MATa dnm1Δ strain YHS83, and MATα dnm1Δ strain YHS84. MATa ugo1Δ dnm1Δ strain YHS85 and MATα ugo1Δ dnm1Δ strain YHS86 were constructed by crossing MATa ugo1Δ strain YHS72 and MATα dnm1Δ strain YHS84.
Mitochondrial Fusion Assay
Mitochondrial fusion during mating was observed as described (Nunnari et al. 1997; Okamoto et al. 1998), but with the following modifications. MATa strains that carry pGAL1-COX4–RFP (pHS51) were grown to log phase in SRaf medium overnight, pelleted by centrifugation, and resuspended in S medium with 2% galactose and 2% sucrose (SgalSuc) to an OD600 of 0.2 for 3–5 h to induce COX4-RFP expression. MATα strains carrying pGAL-COX4-CFP were grown to log phase in SGalSuc medium overnight. Cells were collected, washed, and resuspended in 2 ml of YEPD medium at an OD600 of 0.2. Two strains were mixed and collected by centrifugation. Cells were resuspended in 5 μl of YEPD medium and placed on a nitrocellulose membrane and excess solution was removed by placing the membrane on filter papers. The nitrocellulose membrane was then incubated on YEPD medium at 30°C for 3.5 h. Zygotes were examined by fluorescence microscopy.
Cells were fixed in 4% paraformaldehyde, converted to spheroplasts, attached to poly-l-lysine coated coverslips, and permeabilized as described (Harlow and Lane 1988). Samples were incubated with a 1:100 dilution of antibodies to the myc epitope (9E10; Covance) in PBS containing 1% BSA and 0.05% Tween 20), and with 1:100 dilution of antiserum to the β subunit of the F1-ATPase (a gift from M. Yaffe, University of California, San Diego, CA) for 1 h, washed three times in PBS containing 0.05% Tween 20, and then stained a 1:200 dilution of FITC-conjugated goat anti–mouse IgG (Boehringer) and a 1:500 dilution of rhodamine-conjugated goat anti–rabbit IgG (Boehringer) for 1 h. Samples were washed and mounted in 95% glycerol containing 0.1% p-phenylenediamine and observed.
Subcellular and Submitochondrial Fractionation
Yeast cells were grown to an OD600 of ∼2 in SGal medium. Cells were converted to spheroplasts, homogenized, and separated into a mitochondrial pellet and a postmitochondrial supernatant by centrifugation at 9,600 g for 10 min as described (Daum et al. 1982). Separation of outer membrane and inner membrane vesicles on sucrose gradients was performed as described (Ryan et al. 1994). For protease digestion, mitochondria were resuspended at 1 mg/ml in 250 mM sucrose, 20 mM Hepes-HCl, pH 7.5, and treated with 200 μg/ml trypsin (Sigma-Aldrich) for 20 min on ice, followed by the addition of 2 mg/ml soybean trypsin inhibitor (Sigma-Aldrich). To disrupt the mitochondrial outer membrane, mitochondria were resuspended at 1 mg/ml in 20 mM Hepes-HCl, pH 7.5, and incubated on ice for 30 min.
For analysis, proteins were separated on SDS-PAGE (Laemmli 1970) and transferred to Immobilon filters (Millipore; Haid and Suissa 1983). Filters were probed with antibodies to the myc epitope (9E10), the HA epitope (12CA5; Niman et al. 1983), F1β ATPase, Tim23p (Emtage and Jensen 1993), and OM45p (Yaffe et al. 1989), all at 1:10,000 dilution, or hexokinase (Kerscher et al. 2000) at 1:20,000 dilution. Immune complexes were visualized using 1:10,000 dilution of HRP-conjugated secondary antibodies (Amersham Pharmacia Biotech) followed by chemiluminescence (SuperSignal; Pierce Chemical Co.).
Cells were observed using a Axioskop microscope (ZEISS) with a 100× Plan-Neofluar objective. Fluorescence and differential interference contrast (DIC) images were captured with a MicroMax CCD camera (Princeton Instruments) using IP Lab software v3.2.0 (Signal Analytics Co.).
Isolation of Mutants That Lose Mitochondrial DNA in a Dnm1p-dependent Manner
fzo1 mutants are defective in mitochondrial fusion and also lose mtDNA. The loss of mtDNA in fzo1 cells can be suppressed by inactivation of Dnm1p function (Bleazard et al. 1999; Sesaki and Jensen 1999; Jensen et al. 2000). To identify new components required for fusion, we screened for mutants that maintain mtDNA when Dnm1p activity is absent, but lose mtDNA in the presence of functional Dnm1p. We controlled Dnm1p activity using the URA3 plasmid pHS50, which carries a dominant negative version of Dnm1p, Dnm1p-111 (Fig. 1 A; Sesaki and Jensen 1999; Jensen et al. 2000). Cells that contain pHS50 lack Dnm1p function, which is restored upon loss of the plasmid. To monitor the presence of mtDNA, we took advantage of the ade2 mutation. ade2 cells that contain mtDNA are competent for respiration, producing a red pigment and forming red colonies (Reaume and Tatum 1949). Cells that lose mtDNA and the ability to respire form white colonies.
Strains YHS60 and YHS61, which contain the plasmid pHS50, were mutagenized and plated on SD-Ura. Colonies were then replica-plated to 5FOAD to select for cells that had lost the URA3-DNM1-111 plasmid (Boeke et al. 1984). Out of 16,000 colonies screened, 11 formed red colonies on SD-Ura and then formed white colonies when replica-plated onto 5FOAD medium (Fig. 1 B), suggesting that the 11 mutants contained mtDNA when the DNM1-111–containing plasmid was present, but lost mtDNA when pHS50 was absent. Since mtDNA is required for cell growth on nonfermentable carbon sources, we tested the 11 mutants for their ability to grow on glycerol- and ethanol-containing medium. When cells contained pHS50 (YPGE medium), all 11 mutants were able to grow, but in the absence of pHS50 (5FOADE medium), all 11 mutants failed to grow (Fig. 1 B). We crossed the 11 mutants to wild-type cells and found that all were recessive. After meiotic analysis, we found that 8 of the 11 mutants were defective in single genes and these 8 were further analyzed.
Our genetic screen identified four different genes. In crosses to fzo1 cells, we found that one of our mutants carried an fzo1 allele (Fig. 1 C). Since our genetic screen was based on the behavior of fzo1 mutants, this result was expected. Furthermore, since mgm1 mutants lose mtDNA in a Dnm1p-dependent manner (Fekkes et al. 2000), we anticipated that we would find mgm1 mutants. Five of our mutants carried mgm1 alleles. The two remaining mutants formed two new complementation groups, which we have called ugo1 and ugo2 (ugo is Japanese for fusion). We examined mitochondrial shape in our ugo1 and ugo2 mutants (Fig. 1 D). Wild-type, ugo1-1, and ugo2-1 cells were transformed with a plasmid expressing green fluorescent protein (GFP) fused to the mitochondrial outer membrane protein, OM45 (Cerveny et al. 2001). In wild-type cells, mitochondria were seen as long, tubular structures with occasional branches. In contrast, in both ugo1-1 and ugo2-1 mutants, fragmented mitochondria were found, similar to those seen in fzo1 cells (Hermann et al. 1998; Rapaport et al. 1998; see below). In this report, we focus on the characterization of the ugo1 mutant and the description of ugo2 is the subject of another study.
UGO1 Encodes a Novel 58-kD Protein
Since ugo1 cells lose mtDNA in the presence of functional Dnm1p, we isolated UGO1 by screening a genomic library for clones that allow ugo1-1 to maintain mtDNA. ugo1-1 cells containing the URA3-DNM1-111 plasmid pHS50 were transformed with a yeast genomic DNA library. We then selected for loss of pHS50 and asked if cells could retain their mtDNA by replica-plating transformants to 5FOA medium with glycerol and ethanol as the sole carbon source. Eight plasmids with overlapping inserts were identified that allowed ugo1-1 cells to maintain mtDNA in the absence of pHS50. Subcloning studies localized the UGO1-complementing activity to open reading frame YDR470c, a previously uncharacterized protein.
UGO1 encodes a 503–amino acid protein (∼58 kD), and hydropathy analysis (Kyte and Doolittle 1982) predicts that Ugo1p contains a single transmembrane domain in the middle of the protein, between residues 295 and 311. Interestingly, we found that Ugo1p carries two mitochondrial energy transfer protein signatures (Nelson et al. 1998) at residues 132–144 and 310–319. This motif consists of 10 loosely conserved amino acids and is found in many mitochondrial carrier proteins such as AAC2, CTP1, and DIC1 (Nelson et al. 1998). We also found that Ugo1p is ∼22% identical to a Schizosaccharomyces pombe protein encoded by the SPAC1B2.02c gene.
Ugo1p Is Essential for Growth on Nonfermentable Carbon Sources and for Maintenance of mtDNA
To further investigate the function of Ugo1p, we created a null allele by replacing the UGO1 open reading frame with the yeast HIS3 gene (ugo1Δ). UGO1/ugo1Δ diploid cells were sporulated and the haploid segregants were allowed to grow on YEPD. We found that all four spores in each tetrad were viable, but that two spores formed small colonies (Fig. 2 A). The slower growing cells were shown to carry the ugo1Δ gene, whereas cells from larger-sized colonies contained UGO1. Ugo1p appears to be required for normal cell growth. As shown in Fig. 2 B, cells carrying the ugo1Δ disruption failed to grow on YEPGE. Since mtDNA is essential for growth on glycerol and ethanol, we asked if ugo1Δ cells lack mtDNA by staining them with the DNA-specific dye, DAPI. DAPI staining showed that mtDNA nucleoids were present in wild-type cells (Fig. 3). In contrast, ugo1Δ cells contained little or no mtDNA and only faint staining of nuclear DNA was seen, similar to wild-type cells lacking mtDNA (rho0 WT). Our results indicate that UGO1 is required for growth on nonfermentable carbon sources and to maintain mtDNA.
ugo1Δ Cells Contain Fragmented Mitochondria
To further probe the function of UGO1, we examined mitochondria in wild-type, fzo1Δ, and ugo1Δ cells. Mitochondria were visualized using an outer membrane–targeted GFP fusion protein, OM45-GFP (Fig. 3). Although wild-type cells contained a few elongated mitochondrial tubules with occasional branches, ugo1Δ and fzo1Δ cells showed many small mitochondrial fragments. In most ugo1Δ cells (64%, n = 102) mitochondria were distributed uniformly at the cell periphery (Fig. 3, left panel of ugo1Δ image). In the remaining ugo1Δ cells, mitochondria were somewhat aggregated (Fig. 3, right panel of ugo1Δ image). Although the morphology of mitochondria in ugo1Δ cells was dramatically altered, mitochondrial transmission was not altered. Mitochondrial fragments were always seen in both mother and daughter cells. The altered mitochondrial shape in ugo1Δ cells was not due to lack of mtDNA. Wild-type cells which lack mtDNA showed normal tubular-shaped mitochondria (Fig. 3, rho0WT).
Although the mitochondrial fragments seen in ugo1Δ cells are similar to those seen in fzo1Δ cells, we found noticeable differences between ugo1Δ and fzo1Δ mutants (Fig. 3). For example, ugo1Δ cells contained slightly larger fragments (0.47 μm in average diameter, n = 50) than fzo1Δ cells (0.40 μm, n = 50). Mitochondria in ugo1Δ cells showed markedly different sizes (ranging from 0.21 to 1.06 μm), whereas the organelles in fzo1Δ cells displayed relatively uniform sizes (ranging from 0.28 to 0.69 μm). We also noticed that mitochondria in ugo1Δ cells were aggregated more often than mitochondria in fzo1Δ cells. 36% (n = 102) of ugo1Δ cells contained aggregated mitochondria, whereas only 10% (n = 110) of fzo1Δ mitochondria were clumped.
The morphology of intracellular structures other than mitochondria was not defective in ugo1Δ cells. When we stained vacuoles using FM4-64 (Vida and Emr 1995), their morphology in ugo1Δ cells was indistinguishable from vacuoles seen in wild-type cells (Fig. 4 A). The endoplasmic reticulum, visualized with Sec63p-GFP (Prinz et al. 2000), also displayed normal shape in ugo1Δ cells (Fig. 4 B). Since the actin cytoskeleton is important for the shape of mitochondria in budding yeast (Drubin et al. 1993; Boldogh et al. 1998), we examined the organization of the actin in wild-type and ugo1Δ cells using Alexa 594–phalloidin. We found that the distribution of actin cables and patches in ugo1Δ cells was the same as that in wild-type cells (Fig. 4 C). Thus, our results suggest that the defect in ugo1Δ cells is limited to mitochondria.
Disruption of DNM1 Rescues Fragmentation of Mitochondria and Loss of mtDNA in ugo1Δ Cells
Mitochondrial fusion and division are normally balanced in cells, leading to the few tubular-shaped mitochondria seen in wild-type cells. The fragmentation of mitochondria in fzo1 results from continued division in the absence of fusion, and disruption of a gene required for division, DNM1, in fzo1 cells restores normal mitochondrial shape and number (Sesaki and Jensen 1999). To test if ugo1 shows a similar interplay with dnm1 as that seen with fzo1 and dnm1, we compared mitochondrial shape in either ugo1Δ mutants or dnm1Δ mutants to those in ugo1Δ dnm1Δ double mutants. In dnm1Δ mutants, a single mitochondria consisting of a network of interconnected tubules is seen, resulting from ongoing fusion in the absence of mitochondrial division (Fig. 5 A; Bleazard et al. 1999; Sesaki and Jensen 1999). As noted previously, ugo1Δ cells contain many small mitochondrial fragments (Fig. 3 and Fig. 5). In contrast, in the majority of ugo1Δ dnm1Δ cells (90%, n = 100), mitochondria appeared as elongated tubules, similar to those in wild-type cells (Fig. 5 A) or in fzo1 dnm1 double mutants (Sesaki and Jensen 1999). In ∼56% of ugo1Δ dnm1Δ cells mitochondrial tubules were often collapsed to one side of the cell and appeared to be bundled (Fig. 5 A, left panel of ugo1Δ dnm1Δ images). In 34% of ugo1Δ dnm1Δ cells individual mitochondrial tubules were clearly separated from other tubules (Fig. 5 A, right panel of ugo1Δ dnm1Δ images). Only a small fraction of ugo1Δ dnm1Δ cells (∼10%) showed mitochondria that appeared to be fragmented and aggregated. Thus, our results demonstrate that the fragmentation of mitochondria in ugo1Δ cells can be suppressed by dnm1 disruption. Ugo1p, like Fzo1p, appears to function in mitochondrial fusion, an activity antagonistic to the Dnm1p-mediated division of mitochondria.
Disruption of DNM1 also rescued the loss of mtDNA in ugo1Δ cells. When wild-type cells, ugo1Δ mutants, dnm1Δ mutants, or ugo1Δ dnm1Δ double mutants were stained with DAPI, we found that mtDNA was absent from ugo1Δ cells (Fig. 5 A). However, similar amounts of mtDNA nucleoids were found in wild-type cells, dnm1Δ mutants, and ugo1Δ dnm1Δ mutants (Fig. 5 A). Furthermore, in contrast to ugo1Δ mutants, we found that ugo1Δ dnm1Δ cells were able to grow on a glycerol and ethanol-containing medium, indicating that the double mutant contained mtDNA (Fig. 5 B). We note that ugo1Δ dnm1Δ cells grew more slowly than wild-type and dnm1Δ cells on both glucose and glycerol/ethanol media. This growth appears to result from lack of Ugo1p, since ugo1Δ and ugo1Δ dnm1Δ cells grew more slowly than wild-type or dnm1Δ on glucose-containing medium (Fig. 2 A and 5 B).
ugo1Δ and ugo1Δ dnm1Δ Cells Are Defective in Mitochondrial Fusion
To ask if Ugo1p plays a direct role in fusion, we examined the ability of ugo1 cells to fuse their mitochondria after yeast cell mating (Nunnari et al. 1997; Okamoto et al. 1998). The mitochondria in MATa cells were labeled using a matrix-targeted RFP (pGAL1-COX4–RFP) expressed from pHS51. In MATα cells, mitochondria were visualized with a matrix-targeted CFP (pGAL1-COX4–CFP) carried on pHS52. Both plasmids express the fusion protein under control of the inducible GAL1 promoter. MATa and MATα cells were pregrown in galactose-containing medium to induce the expression of the fusion proteins and transferred to glucose medium to inhibit their further synthesis. Cells were mixed and allowed to mate on glucose-containing medium. If mitochondrial fusion occurred in the resulting zygotes, RFP and CFP fluorescence should completely overlap due to the diffusion of the matrix COX4-RFP and COX4-CFP proteins. If no fusion occurs, RFP and CFP should be seen in separate organelles.
We found that ugo1Δ and ugo1Δ dnm1Δ mutants are defective in fusion. In Fig. 6, representative examples of zygotes containing a medial diploid bud from each mating mixture are shown, but >50 zygotes for each mating mixture were actually examined. When two wild-type cells were mated, mitochondria in the zygote efficiently fused and a complete overlap of the RFP and CFP fluorescence was seen. Consistent with previous studies (Bleazard et al. 1999; Sesaki and Jensen 1999), mitochondrial fusion also occurred in dnm1Δ/dnm1Δ zygotes. Interestingly, dnm1Δ/dnm1Δ zygotes often contained a single tubule emerging from the mitochondrial network of each parent. Fusion appeared to occur at a discrete point near the middle of the zygote. In contrast to wild-type and dnm1Δ cells, fusion was defective in ugo1Δ mutants. ugo1Δ/ugo1Δ zygotes contained many mitochondrial fragments, but these organelles contained only RFP or CFP fluorescence. No organelles containing both fluorophores were seen. Although disruption of DNM1 suppresses the fragmentation of mitochondria and the loss of mtDNA of ugo1Δ mutants, ugo1Δ dnm1Δ double mutants still failed to fuse their mitochondria. Although ugo1Δ dnm1Δ cells displayed tubular mitochondrial shape, each mitochondrial tubule contained only RFP or CFP. We found that mitochondria in ugo1Δ/ugo1Δ or ugo1Δ dnm1Δ/ugo1Δ dnm1Δ zygotes were often closely positioned to each other near the middle of zygotes, but nonetheless did not fuse. Our results thus indicate that ugo1Δ and ugo1Δ dnm1Δ cells are defective in mitochondrial fusion and argue that Ugo1p plays a direct role in the fusion pathway.
Ugo1p Is a Mitochondrial Outer Membrane Protein, with its NH2 Terminus Facing the Cytosol and COOH Terminus in the Intermembrane Space
To localize Ugo1p in yeast cells, we constructed two epitope-tagged versions of Ugo1p, myc-Ugo1p, and Ugo1p-HA. myc-Ugo1p carries the myc epitope (Munro and Pelham 1986) fused to the NH2 terminus of the Ugo1 protein, and Ugo1p-HA contains the influenza HA epitope (Field et al. 1988) at its COOH terminus. Cells that expressed either myc-Ugo1p (Fig. 7 A) or Ugo1p-HA (data not shown) contained a single protein of ∼65 kD. We found that both fusion proteins were functional and ugo1Δ cells expressing either myc-Ugo1p (Fig. 7 B) or Ugo1p-HA (data not shown) maintained mtDNA and normal mitochondrial shape.
Immunofluorescence studies showed that Ugo1p is a mitochondrial protein (Fig. 7 B). ugo1Δ cells expressing the myc-Ugo1p fusion protein were fixed, permeabilized, and then incubated with antibodies to the myc-epitope and the mitochondrial ATPase β subunit (F1β) protein. When immune complexes were visualized using fluorescence microscopy, we found that myc-Ugo1 protein colocalized with the mitochondrial F1β protein. Cell fractionation experiments also confirmed the mitochondrial localization of Ugo1p (Fig. 7 C). Cells expressing Ugo1p-HA were homogenized and separated into a mitochondrial fraction and a postmitochondrial supernatant. We found that Ugo1p cofractionated with the mitochondrial Tim23 protein, whereas little or no Ugo1p was found in the supernatant along with cytosolic hexokinase protein.
Ugo1p is an integral membrane protein located in the outer membrane. When mitochondria isolated from cells expressing Ugo1p-HA were treated with 1.5 M sodium chloride or 0.1 M sodium carbonate (Fig. 8 A), Ugo1p was not extracted from the mitochondria like the peripheral membrane protein, the β subunit of the F1-ATPase (F1β). Instead, Ugo1p remained in the membrane pellet with the integral membrane protein, Tim23p. To determine which mitochondrial membrane contains Ugo1p, we prepared membrane vesicles from myc-Ugo1p mitochondria and separated them into outer membrane and inner membrane fractions on sucrose gradients. As shown in Fig. 8 B, myc-Ugo1 cofractionated with the outer membrane vesicle fraction, along with OM45. The F1β protein, a marker for the inner membrane, was found in more dense fractions, separate from myc-Ugo1p and OM45.
The COOH terminus of Ugo1p faces the intermembrane space (IMS). When Ugo1p-HA mitochondria were treated with trypsin, the outer membrane protein, OM45, was completely digested, whereas the inner membrane proteins, Tim23p and F1β, remained intact (Fig. 8 C). Trypsin digested the 65-kD Ugo1-HA protein, producing an ∼25 kD fragment that reacted with our HA antibodies (Fig. 8 C, asterisk). The size of this fragment is consistent with the length of the COOH terminus of Ugo1p including its transmembrane segment. When the mitochondrial outer membrane was disrupted by osmotic shock to form mitoplasts, protease treatment now digested both Tim23p and Ugo1p-HA. Since our antiserum recognizes the NH2 terminus of Tim23p, which faces the IMS, we conclude that the COOH-terminal HA epitope of Ugo1p-HA similarly faces the IMS. Trypsin treatment of mitoplasts did not digest the matrix-facing F1β protein. We also directly showed that NH2 terminus of Ugo1p faces the cytosol. Mitochondria were isolated from cells expressing myc-Ugo1p and treated with trypsin. myc-Ugo1p and OM45p, but not Tim23p and F1β protein, were digested in intact mitochondria (Fig. 8 D). No protected fragments of myc-Ugo1p were seen. Therefore, we conclude that Ugo1p is an outer membrane protein, with its NH2 terminus facing the cytosol and its COOH terminus in the IMS.
We have identified a new outer membrane protein, Ugo1p, required for mitochondrial fusion. Ugo1p was identified using a genetic screen for mutants that maintain mtDNA in the absence of mitochondrial division, but lose mtDNA when division is active. Our screen was based on previous studies with fzo1 mutants, which are defective in an outer membrane protein that mediates mitochondrial fusion. fzo1 cells lose mtDNA, but the loss of mtDNA can be suppressed by disruption of DNM1, a gene required for mitochondrial division (Bleazard et al. 1999; Sesaki and Jensen 1999; Jensen et al. 2000). Like fzo1 mutants, the loss of mtDNA in ugo1 mutants is suppressed by inactivation of Dnm1p function. ugo1 cells that carry the dominant negative DNM1-111 mutant or ugo1Δ dnm1Δ double mutants maintain mtDNA. Similar to fzo1 mutants, the mitochondrial morphology defect in ugo1Δ cells is suppressed by DNM1 disruption. Both fzo1Δ and ugo1Δ mutants contain many small mitochondrial fragments, instead of the few long tubular-shaped mitochondria found in wild-type cells. ugo1Δ dnm1Δ cells contain mitochondrial tubules similar to those seen in wild-type cells and in fzo1 dnm1 double mutants (Sesaki and Jensen 1999). Our results suggest that a balance of fusion and division regulates mitochondrial shape and number. In the absence of mitochondrial fusion, mediated by Fzo1p and Ugo1p, ongoing division produces numerous small organelles. When division is defective, continuous fusion leads to the single interconnected mitochondrial network seen in dnm1 cells (Bleazard et al. 1999; Sesaki and Jensen 1999).
We have found that ugo1Δ cells, like fzo1 mutants (Hermann et al. 1998), are defective in mitochondrial fusion. Although ugo1Δ dnm1Δ double mutants contain tubular-shaped mitochondria, they do not fuse their mitochondria. Similarly, fzo1Δ dnm1Δ cells remain blocked for fusion (Bleazard et al. 1999; Sesaki and Jensen 1999). Therefore, the normal-looking mitochondria found in ugo1 dnm1 and fzo1 dnm1 mutants (Sesaki and Jensen 1999) suggest that some aspects of mitochondrial shape result from mechanisms independent of fusion and division. For example, tubular-shaped mitochondria may arise by directed growth of preexisting organelles along cytoskeletal filaments. Alternatively, internal mitochondrial proteins may provide a scaffold for tubulation of mitochondria. Regardless of how tubules are formed, we suggest that a balance between mitochondrial fusion and division regulates the length, number, and connection of mitochondrial tubules. For example, in ugo1 and fzo1 mutants where fusion is blocked, mitochondria form numerous tubular structures, but the length of each tubule is very short. In dnm1 mutants (Bleazard et al. 1999; Sesaki and Jensen 1999), mdv1/gag3/net2 mutants (Fekkes et al. 2000; Tieu and Nunnari 2000; Cerveny et al. 2001), or fis1 mutants (Mozdy et al. 2000) where division is defective, mitochondria form a single organelle consisting of interconnected tubules.
Our genetic screen identified five mgm1 mutants. mgm1 mutants lose mtDNA (Jones and Fangman 1992; Guan et al. 1993) and mtDNA loss can be suppressed by inactivation of Dnm1p (Fekkes et al. 2000). Mgm1p is a mitochondrially associated, dynamin-related GTPase, although its exact location in mitochondria is unclear (Shepard and Yaffe 1999; Wong et al. 2000). In mgm1 mutants, mitochondria are fragmented like those in fzo1 and ugo1 cells, suggesting a role in mitochondrial fusion. Recently, mgm1 mutants have been shown to be defective in mitochondrial fusion (Wong et al. 2000). However, in contrast to fzo1 dnm1 (Bleazard et al. 1999; Sesaki and Jensen 1999) and ugo1 dnm1 double mutants, mgm1 dnm1 double mutants are able to fuse their mitochondria, suggesting that Mgm1p plays an indirect role in mitochondrial fusion (Wong et al. 2000).
Ugo1p is embedded in the mitochondrial outer membrane with its COOH terminus of nearly 200 amino acids facing the IMS. Most other proteins involved in membrane fusion, such as SNAREs (Rothman and Warren 1994; Pelham 1999), the influenza HA protein (White et al. 1996), and Fzo1p (Hermann et al. 1998; Rapaport et al. 1998), contain few if any residues on the opposite side of the membrane to where fusion takes place. Mitochondria, in contrast to most other organelles, have two membranes. The mitochondrial inner membrane appears to fuse immediately after outer membrane fusion (Okamoto et al. 1998), suggesting a coupling between both fusion events. We speculate that the COOH terminus of Ugo1p may interact with inner membrane fusion machinery. We note that the COOH terminus of Ugo1p contains a mitochondrial energy transfer protein motif (Nelson et al. 1998) which is found in many inner membrane proteins. Studies to determine the role Ugo1p plays in mitochondrial inner and outer membrane fusion are in progress.
Although mitochondrial fusion occurs predominately at the tips of mitochondrial tubules (Nunnari et al. 1997), our studies show that Ugo1p is present throughout the mitochondrial outer membrane. Similarly, Fzo1p shows a uniform distribution along the mitochondrial tubule (Hermann et al. 1998). It is possible that mitochondrial fusion is activated only at sites of fusion. For example, the Fzo1p GTPase may act as a molecular switch that regulates mitochondrial fusion by activating the fusion machinery at the appropriate time. Alternatively, the fusion machinery may be transiently concentrated at fusion sites. It is also possible that mitochondria are competent to fuse anywhere along the tubule, but fusion is directed by controlling contact between organelles. For example, the cytoskeleton may play a crucial role in positioning mitochondria during their fusion.
Is Ugo1p part of a fusion machine? Gel filtration studies of detergent-solubilized mitochondria show that Fzo1p is found in an ∼800-kD complex (Rapaport et al. 1998). Since both Fzo1p and Ugo1p are located in the mitochondrial outer membrane and play essential roles in mitochondrial fusion, it is possible that both proteins are part of the same complex. Both proteins appear to be defective in a late step in the fusion pathway. In matings between ugo1Δ mutants and mitochondria the two parent cells are closely paired in the neck of the zygote, but do not fuse. Similar connections between unfused mitochondria were seen in matings between fzo1Δ cells (Hermann et al. 1998). However, preliminary studies have shown that Fzo1p and Ugo1p do not coimmunoprecipitate (Sesaki, H., unpublished observations). Therefore, it is possible that Fzo1p and Ugo1p mediate distinct steps in mitochondrial fusion and do not physically interact. We note that whereas fzo1 and ugo1 mutants both contain fragmented mitochondria, the organelles tend to aggregate in ugo1 mutants, but mitochondria remain dispersed in fzo1 cells. Further studies are clearly needed to determine the role of Ugo1p and Fzo1p in mitochondrial fusion.
We thank K. Cerveny for pKC2, F. Spencer for a yeast genomic DNA library, B. Glick for pST10, P. Silver for pPS1530, R. Rothstein for W303, M. Yaffe for anti-F1b antibody, A. Aiken Hobbs for pAA3, D. Kornitzer, S. Kron, and G. Fink for KB241, and A. Aiken Hobbs and O. Kerscher for help with submitochondrial fractionation. We also thank C. Machamer, K., Wilson, A. Aiken Hobbs, K. Cerveny, M. Youngman, C. Dunn, J. Holder, and T. Kai for valuable comments on the manuscript.
This work was supported by US Public Health Service grant RO1-GM46803 to R.E. Jensen and a postdoctoral fellowship from the Japan Society for the Promotion of Science to H. Sesaki.
Abbreviations used in this paper: CFP, cyan fluorescent protein; DIC, differential interference contrast; 5FOA, 5-fluoro-orotic acid; GFP, green fluorescent protein; HA, hemagglutinin; IMS, intermembrane space; mtDNA, mitochondrial DNA; RFP, red fluorescent protein.