Fzo1p Is a Mitochondrial Outer Membrane Protein Essential for the Biogenesis of Functional Mitochondria in Saccharomyces cerevisiae*

Fzo1p is a novel component required for the biogenesis of functional mitochondria in the yeast Saccharomyces cerevisiae. The protein is homologous to DrosophilaFzo, the first known protein mediator of mitochondrial fusion. Deletion of the FZO1 gene results in a petite phenotype, loss of mitochondrial DNA, and a fragmented mitochondrial morphology. Fzo1p is an integral protein of the mitochondrial outer membrane exposing its major part to the cytosol. It is imported into the outer membrane in a receptor-dependent manner. Fzo1p is part of a larger protein complex of 800 kDa, and presumably is the first identified component of the yeast mitochondrial fusion machinery.

tioning of mitochondria to the daughter cell. None of these proteins appears to play a direct role in mitochondrial fusion.
Because the integrity of the three-dimensional structure of the cell depends on fusion of intracellular membranes, the identification and characterization of the molecular components responsible for this process is subject to intense investigation. The best characterized system of intracellular membrane fusion is that of the organelles of the secretory pathway which are interconnected by a complex network of transport vesicles (13). Both the transport vesicles and the target membranes carry a specific set of integral membrane receptors on their surface, the SNARE 1 proteins which determine the identity of organelles and control the specificity of docking between intracellular membranes (14,15). At the same time, SNAREs constitute the minimal machinery for membrane fusion (16). The ATPase NSF and SNAP proteins play an essential role in SNARE-dependent fusion processes in vivo and are thought to be a general machinery for the recycling of SNARE proteins after fusion (16,17). Mitochondrial fusion, however, appears to be independent of the action of NSF (4), and no SNARE-like proteins are known in mitochondria. Thus, it seems likely that mitochondria employ a different mechanism for fusion.
Recently, the first known protein mediator of mitochondrial fusion has been identified in Drosophila (18). The fuzzy onions (fzo) gene encodes a large predicted transmembrane GTPase that is expressed during spermatogenesis late in meiosis II. In male fzo mutants, mitochondria aggregate and are defective in postmeiotic fusion. They develop structures that look like "fuzzy onions." This deficient organellar development results in defective sperm production and male sterility. Similar proteins of unknown function exist in mammals, nematodes, and yeast (18).
Here, we report the characterization of the yeast homolog of Fzo, Fzo1p. Disruption of the FZO1 gene in yeast results in a petite phenotype and in the loss of mitochondrial DNA, indicating an important function of Fzo1p in mitochondrial biogenesis. Cells lacking Fzo1p show a fragmented mitochondrial morphology. Fzo1p is located in the mitochondrial outer membrane exposing the major part of the protein to the cytosol, and can be imported into isolated mitochondria in a receptor-dependent manner. Fzo1p is part of a high molecular weight complex, and presumably is the first identified component of a yeast mitochondrial fusion machinery.

EXPERIMENTAL PROCEDURES
Yeast Strains and Disruption of the FZO1 Gene-Standard genetic techniques were used for growth and manipulation of yeast strains (19). Transformation of yeast was carried out as described (20).
To obtain disruption mutants of FZO1, most of the open reading frame was first replaced with the Tn903 kanamycin resistance gene in the diploid strain YPH501 (21) by a PCR-based approach as described (22). The primers used to generate the disrupting DNA fragment were 179KA-5 (5Ј-GGT GAT GTA AAT ACT GGT GCT AGC GCT CTT TGC AAC TCT C) and 179KA-3 (5Ј-GCA AAG AGC GCT AGC ACC AGT ATT TAC ATC ACC). Haploid deletants were obtained after sporulation and tetrad dissection. The wild type strains D273-10B (ATCC 24657) and W303A were used for the preparation of subcellular and submitochondrial fractions.
Recombinant DNA Techniques and Plasmids Constructions-Standard methods were used for the manipulation of DNA (23). To obtain the construct for in vitro transcription of FZO1, the FZO1 open reading frame was amplified from genomic DNA by PCR using the primers Z36U (5Ј CCC GGA TCC ACC ATG TCT GAA GGA AAA CAA C) and Z36L (5Ј CCC GTT AAC GTC GAC CTA ATC GAT GTC TAA A) and cloned into the BamHI and SalI sites of the in vitro expression vector pGEM4 (Promega).
To obtain the construct for expression of the green fluorescent protein (GFP) in mitochondria, the presequence of the subunit 9 of the F 0 ATPase of Neurospora crassa was amplified by PCR using the primers SU9N (5Ј GGG AAG CTT ATG GCC TCC ACT CGT GTC C) and SU9C (5Ј GGG GGA TCC GGA AGA GTA GGC GCG CTT) and cloned into the HindIII and BamHI sites of the vector pGEM3 (Promega) yielding plasmid pGEM3-Su9 . The open reading frame coding for GFP from Aequorea victoria containing the S65T mutation (24) was amplified by PCR from plasmid pFP20 (kind gift of Dr. F. Parlati, New York) using the primers GFP-N (5Ј CGG GTA CCA GAT CTA TGA GTA AGG GTG AAG AAC TTT TC) and GFP-C (5Ј CGG AAT TCT TAT TTG TAT AGT TCA TCC) and cloned into the KpnI and EcoRI sites of pGEM3-Su9(1-69) yielding plasmid pGEM3-Su9-GFP2. The HindIII/EcoRI fragment of pGEM3-Su9-GFP2 was subcloned into the yeast expression vector pYES2.0 (Invitrogen) yielding plasmid pYES-GFP2.
Import of Precursor Proteins into Mitochondria-The in vitro import of Fzo1p was carried out essentially as described (25). Precursor protein was synthesized in the presence of [ 35 S]methionine in reticulocyte lysate (Promega). Import mixtures (100 l) usually contained 1-3% reticulocyte lysate (v/v) in 3% bovine serum albumin (w/v), 0.6 M sorbitol, 10 mM MOPS-KOH, 80 mM KCl, pH 7.2. Protease treatment was performed by adding proteinase K or trypsin at the indicated concentrations for 15 min at 0°C followed by addition of 1 mM PMSF.
Gel Filtration Analysis-Mitochondria (1 mg) were pelleted for 10 min at 10,000 ϫ g and resuspended at a concentration of 5 mg/ml in buffer A (1% Triton X-100, 150 mM K-acetate, 4 mM Mg-acetate, 0.5 mM EDTA, 0.5 mM PMSF, 30 mM Tris-HCl, pH 7.4). After incubation for 15 min at 4°C under agitation, mitochondrial extracts were centrifuged for 15 min at 90,000 ϫ g. The supernatant was loaded onto a Superose 6 gel filtration column (25-ml column volume; Amersham Pharmacia Biotech) and chromatographed in buffer A at a flow rate of 0.3 ml/min. Fractions (0.5 ml) were collected and analyzed by SDS-PAGE and immunostaining with antibodies against Fzo1p. Calibration standards used were as follows: cytochrome b 2 , 210 kDa; apoferritin, 440 kDa; Hsp60, 850 kDa.
Miscellaneous Methods-Antisera against the N and the C termini of Fzo1p were raised in rabbits by injecting the chemically synthesized peptides MSEGKQQFKDSNKC (N-terminal) or CKLMVEEINLDID (C-terminal) that had been coupled to activated KLH (Pierce).
Subfractionation of yeast cells and carbonate extraction (26), isolation of mitochondria (25), and preparation of outer membrane vesicles (27) were performed as described.
Protein determination and SDS-PAGE were performed according to published methods. The detection of proteins after blotting onto nitrocellulose was performed using the ECL detection system (Amersham Pharmacia Biotech).
Standard fluorescence and interference contrast microscopy was performed using an Axioplan microscope with a Plan-Neofluar 100x/1.30 Oil objective (Carl Zeiss Jena GmbH) using a 100 W mercury lamp and an excitation wavelength of 450 -490 nm for the visualization of mt-GFP.

FZO1 Is Essential for the Formation of Respiratory-competent
Mitochondria-The FZO1 gene in yeast (systematic name YBR179c) encodes a protein that is 20% identical to Drosophila Fzo (18). Both proteins possess a similar domain structure, namely an N-terminal predicted coiled-coil, followed by a highly conserved GTPase domain, a second coiled-coil region, a predicted transmembrane domain in the C-terminal third of the protein, and a third coiled-coil close to the C terminus (18). Based on these similarities, we reasoned that like Fzo in Drosophila, Fzo1p might play an important role for the biogenesis of mitochondria in yeast. To test whether the FZO1 gene is essential for the viability of yeast cells, we disrupted one of the two copies of FZO1 in diploid cells. The disruption was done by replacing almost the entire open reading frame of FZO1 by the kanamycin resistance gene (see "Experimental Procedures"). After sporulation and tetrad dissection, we found that all four spores in each tetrad were viable on glucose-containing medium, but that two spores in each tetrad showed a slow growth phenotype (Fig. 1A). Cells from the small colonies carried the disrupted gene (⌬fzo1), whereas cells from normal-sized colonies carried the wild type gene. The deletion mutant failed to grow on nonfermentable carbon sources (Fig. 1B), suggesting an important role of Fzo1p for mitochondrial function.
The cells containing the disruption in FZO1 had lost a functional mitochondrial genome. This was concluded from two independent observations. First, mating the ⌬fzo1 strain with a rho°tester strain lacking mitochondrial DNA, ⌬mdj1 (26), did Dissected tetrads were incubated for 3 days on YPD medium (2% glucose) at 30°C. B, replica-plate analysis of dissected tetrads from a FZO1/fzo1::Tn903 heterozygous diploid strain. Yeast cells were grown on a fresh YPD plate and were replica-plated onto a YPD plate, a YPG plate (3% glycerol), and a YPD plate containing 500 g/ml Geneticin (G418). Plates were incubated for 2 days at 30°C. The latter plate allows growth of only those cells that carry the ⌬fzo1 allele disrupted by the Tn903 kanamycin resistance marker. C, fluorescence microscopy of wild type cells and the ⌬fzo1 mutant expressing mt-GFP. Cells were grown in liquid YPGal medium (2% galactose) at 25°C, washed in water, and immobilized in 0.5% low melting point agarose. For each strain, one representative cell is shown. The upper part shows DIC images, and the lower part shows mitochondrial staining. not result in diploid cells able to grow on glycerol. Thus, the ⌬fzo1 strain did not contain a functional mitochondrial genome that would be able to restore growth on nonfermentable carbon sources in the presence of the wild type copy of the FZO1 gene in the diploid strain (not shown). Second, we performed an in organello translation assay for mitochondrial-encoded proteins (28). Mitochondria were incubated in the presence of 35 S-labeled methionine, which resulted in the labeling of mitochondrial-synthesized proteins in wild type mitochondria. In mitochondria isolated from the ⌬fzo1 strain, this labeling was completely absent (not shown). Moreover, COXII, a protein encoded by the mitochondrial DNA, was absent in immunoblots of ⌬fzo1 mitochondria. F 0 ATPase subunit e (Tim11p) and the Rieske iron-sulfur protein (Fe/S), two cytoplasmic-synthesized components of the respiratory chain, could not be detected, and the steady state levels of several other proteins involved in respiration were greatly reduced, namely cytochrome c 1 , F 1 ATPase subunit ␣, and Bcs1p (not shown). Mitochondria with reduced levels of components of the respiratory chain can often be found in petite or rho°strains (29,30). We conclude that the FZO1 gene is crucial for the maintenance of mitochondrial DNA.
We asked whether the deletion of FZO1 leads to an abnormal mitochondrial morphology. When green fluorescent protein fused to a mitochondrial presequence (mt-GFP) was expressed in wild type cells, the characteristic mitochondrial network below the cell cortex was seen in the fluorescence microscope (Fig. 1C). An identical staining was obtained with mitochondria-specific dyes such as Mitotracker or DiOC 6 (not shown). In the ⌬fzo1 strain expressing mt-GFP, the mitochondrial morphology was completely altered. Mitochondria were highly fragmented, and only in a few cells could some short tubular structures be seen (Fig. 1C). Because it is known that the morphology of yeast mitochondria is changed in the absence of an intact mitochondrial genome (31), we expressed mt-GFP in a cytoplasmic petite strain lacking mitochondrial DNA that is otherwise wild type (W303 rho°). In this strain, a mitochondrial morphology similar to that in the ⌬fzo1 mutant was observed. Most of the rho°cells harbored fragmented mitochondria, and only a few well developed tubular structures were seen (data not shown). We conclude that Fzo1p is required for a normal mitochondrial morphology. It is, however, unclear as to whether the fragmented mitochondrial morphology is a direct consequence of deletion of FZO1, or an indirect effect of the rho°state of the deletion mutant.
Fzo1p Is a Mitochondrial Protein Located at the Outer Membrane-To determine the location and topology of Fzo1p, antisera against an N-terminal and a C-terminal peptide were prepared (see "Experimental Procedures"). First, we investigated the subcellular location of Fzo1p. Cells were harvested from a liquid culture at mid-logarithmic growth phase, and subcellular fractions were prepared and used for immunoblotting. Fzo1p was detected in the mitochondrial fraction and cofractionated with Tom40, a mitochondrial protein. Fzo1p could not be detected in the cytosolic or nonmitochondrial membrane fractions (Fig. 2A). Thus, Fzo1p is a mitochondrial protein. Next, the submitochondrial location of Fzo1p was determined. Hydropathy analysis suggests that Fzo1p may contain one or two putative membrane-spanning segments (Fig. 2B). Indeed, Fzo1p could not be extracted from mitochondrial membranes following treatment with 0.1 M sodium carbonate, indicating that it is an integral membrane protein (Fig. 2C). To determine in which of the two mitochondrial membranes Fzo1p resides, we prepared outer membrane vesicles. Fzo1p is enriched in these vesicles together with other outer membrane proteins such as Tom70 (Fig. 2C). In contrast, inner membrane proteins such as COXII were hardly detectable in these vesicles. Furthermore, Fzo1p was accessible to externally added proteases in intact mitochondria under conditions where proteins exposed to the intermembrane space were protected (Fig.  2C) indicating that at least a part of the protein is exposed to the cytosol. We conclude that Fzo1p is an integral protein of the mitochondrial outer membrane.
To understand the role of Fzo1p in the biogenesis of normal mitochondria, it is important to know its topology in the outer membrane. We noticed that after treatment of mitochondria with very low amounts of protease, fragments were generated that were slightly smaller than the full-length protein. Because these fragments could be detected in a Western blot using the antibody against a C-terminal peptide of Fzo1p (Fig. 2D), the protease must have cleaved the N terminus of the protein, indicating that this part is exposed to the outside. Thus, the major part of Fzo1p including the predicted GTPase domain faces the cytosol.
In Vitro Translated Fzo1p Can Be Imported into Isolated Mitochondria-Next, we tested the in vitro import of Fzo1p into isolated mitochondria. This is an independent approach to show its mitochondrial localization, and at the same time, Fzo1p is a novel model protein to study protein import into the mitochondrial outer membrane. Radiolabeled protein was synthesized in vitro and incubated with isolated wild type mitochondria. After the import reaction, carbonate extraction was performed. Most of the imported protein was recovered in the pellet, indicating insertion of the protein into the membrane (Fig. 3A). As a control, carbonate extraction was performed on lysate in the absence of mitochondria. Here, the protein was found in the supernatant, excluding the possibility that it was aggregated under these conditions. The inserted protein like the endogenous one was sensitive to proteinase K (Fig. 3A). As expected from the lack of a typical cleavable presequence, no change in the molecular mass of Fzo1p was observed upon import. We investigated the kinetics of the import process and found that, within 5 min, most of the protein was inserted into the outer membrane (Fig. 3B). Similar to other outer membrane proteins, the import of Fzo1p was independent of a membrane potential ⌬⌿ across the inner membrane as the addition of the uncoupler CCCP did not change the import efficiency (Fig. 3C). Pretreatment of mitochondria with trypsin to cleave import receptors on the mitochondrial surface significantly reduced the amount of inserted protein, suggesting that Fzo1p uses protease-sensitive import receptors for its insertion into the outer membrane (Fig. 3C). The import and protease sensitivity of Fzo1p was not affected by the presence or absence of nucleotides such as AMP, ADP, ATP, or GTP ( Fig. 3C and data not shown). These observations are consistent with a localization of Fzo1p in the mitochondrial outer membrane. It appears that its import is receptor-dependent and does not require ATP-dependent cytosolic chaperones.
Fzo1p Is Part of a High Molecular Weight Complex-It is conceivable that Fzo1p interacts with other proteins to fulfill its role in mitochondrial biogenesis. In particular, the predicted coiled-coil domains may be responsible for formation of hetero and/or homo oligomers. Thus, we investigated whether Fzo1p is part of a high molecular weight complex. Mitochondria were solubilized with detergent, and the proteins were separated on a gel filtration column. Fzo1p was eluted from the column in a relatively sharp peak, corresponding to a molecular weight of about 800 kDa (Fig. 4). The protein was eluted in a fraction corresponding to a similar size when GTP was present during lysis and elution (not shown). These results suggest that Fzo1p is part of a complex of high molecular weight, which might represent the mitochondrial fusion machinery in yeast, or a part thereof. DISCUSSION Fzo in Drosophila is the first known protein mediator of mitochondrial fusion (18,32). Several lines of evidence suggest that Fzo1p might play a similar role in yeast. First, Fzo1p shares a significant sequence homology with the Drosophila protein and possesses a similar predicted domain structure (18). Second, Fzo1p in yeast cells and Fzo in sperm cells have the same subcellular location in mitochondria. Third, Fzo1p is an integral protein of the mitochondrial outer membrane with the major part facing the cytosol, a topology which would be expected for a fusion protein that has to interact with proteins on the opposite membrane or in the cytosol.
The assumption that Fzo1p plays a role in mitochondrial fusion is further strengthened by the observed phenotype of the deletion mutant. Deletion of FZO1 leads to the loss of mitochondrial DNA. The inheritance of mitochondrial DNA in yeast is an ordered event that is thought to depend on the integrity of ]methionine and incubated with isolated mitochondria for 10 min at 25°C. After import, mitochondria were isolated by centrifugation and resuspended in import buffer (Total). Samples were treated with 80 g/ml proteinase K (PK) where indicated. Another sample of the import reaction was resuspended in 0.1 M Na 2 CO 3 and separated by centrifugation into pellet (P) and supernatant (S) fractions (Carbonate/Import). As a control, carbonate extraction was performed with radiolabeled Fzo1p in the absence of mitochondria (Carbonate/Lysate). Proteins were analyzed by SDS-PAGE, blotting to nitrocellulose, and autoradiography. B, import kinetics. Radiolabeled Fzo1p was imported into isolated mitochondria at 25°C as described above. The import reaction was stopped after the indicated time periods by shifting the samples to 0°C, and mitochondria were washed twice with SHKCl buffer (250 mM sucrose, 20 mM HEPES/KOH, pH 7.2, 80 mM KCl). The imported protein was analyzed by SDS-PAGE, blotting to nitrocellulose, and autoradiography. The imported material was quantified by phosphoimaging (Fuji X BAS 1500). C, import characteristics of Fzo1p. Radiolabeled Fzo1p was imported for 10 min at 25°C as described above. One sample was treated with 40 g/ml trypsin to cleave import receptors before the addition of the radiolabeled protein (Trypsin). CCCP (30 M), EDTA (4 mM), or nucleotides (4 mM) were added to the import mixture where indicated. After the import reaction, mitochondria were washed once, and carbonate extraction was performed. Radiolabeled protein recovered in the pellet fraction was analyzed and quantified as described above. The amount recovered from mitochondria that did not get any treatment was taken as 100%.
the mitochondrial compartment (4). It is conceivable that deletion of a component important for mitochondrial structure results in a defect of segregation of mitochondrial DNA to the daughter cell. Thus, several proteins are known that are important for both mitochondrial morphology and maintenance of mitochondrial DNA. Deletion mutants of the mitochondrial outer membrane proteins, Mdm10p and Mdm12p, and the dynamin-like protein, Mgm1p, harbor one or few giant mitochondria per cell, implying a role of these proteins for mitochondrial morphology (6,9,11). At the same time, deletion of the genes leads to the loss of the mitochondrial genome (33,6,11). The ⌬fzo1 mutant is also rho°, it has, however, a fragmented mitochondrial morphology, a phenotype which is consistent with a possible role of Fzo1p in mitochondrial fusion.
Proteins involved in fusion events in the secretory pathway do not seem to play a role in mitochondrial fusion. Neither these proteins themselves or homologues of them have been found associated with mitochondria. And, vice versa, no Fzo1plike protein has been found to play a role in fusion events in the secretory pathway. Thus, mitochondria appear to employ a fusion mechanism that is fundamentally different. If SNAREdependent fusion machineries are mediating fusion of so many diverse membranes in the cell, such as endoplasmic reticulum, Golgi apparatus, vacuole, plasma membrane, endosomes etc., why is not a related machinery fusing the mitochondrial membranes? It is conceivable that at the time when the endosymbiotic ancestors of mitochondria entered the primitive eukaryotic cell, the organelles of the secretory pathway already carried a complex system of SNARE-like proteins on their surfaces. These SNAREs presumably had evolved from a single pair of primitive SNARE-like fusion proteins (16) long before mitochondria, or any other organelles stemming from endosymbiontic bacteria, were present in the eukaryotic cell. Furthermore, as double membrane-bounded organelles, mitochondria are faced with the problem of fusing four membranes. It is obvious that, for this challenge, a different fusion machinery had to be developed.
The mitochondrial fusion machinery must be able to meet the following criteria. First, it must provide a means for the specific recognition of the membranes that are to be fused. Second, it has to supply energy to overcome the energy barrier of membrane fusion. Third, it must coordinate fusion with fission events to maintain the continuity of the mitochondrial network during vegetative growth. Fourth, it has to intimately link the fusion of the mitochondrial outer membrane to the fusion of the inner membrane. It is unlikely that all these requirements can be fulfilled by a single protein. We have identified Fzo1p as a part of an 800-kDa protein complex. Up to now, we cannot exclude the possibility that this complex is a homo-oligomer of Fzo1p, however, preliminary data indicate that proteins of different molecular weights are present in the same complex. 2 Thus, Fzo1p is likely to be only the first identified component of a much more complex mitochondrial fusion machinery.
The precise role of Fzo1p in mitochondrial fusion is still unknown. Our data indicate that it is an integral protein of the mitochondrial outer membrane with the major part of the protein exposed to the cytosol. The exposed part includes the GTPase domain and two predicted coiled-coil regions. Such a topology seems to be ideally suited for a fusion protein that is expected to interact with other proteins on the opposite membrane and/or in the cytosol. It is still unclear whether the C-terminal end of the protein, which carries an additional putative coiled-coil region, may contribute to such interactions. It was not possible to detect a protected C-terminal fragment in Western blots of protease-treated mitochondria or outer membrane vesicles, which would be expected if the C terminus would be in the intermembrane space, or even in the matrix (not shown).
One possibility is that Fzo1p itself plays a central role in the recognition of the partner organelle and/or fusion of lipid bilayers. The predicted coiled-coil regions might be the domains mediating such interactions, similar to pairing of cognate SNARE proteins on opposite membranes via coiled-coils. An intact GTPase domain has been shown to be essential for the function of the protein in Drosophila (18). Similar to dynamin GTPases, GTP hydrolysis could provide biomechanical energy that could be used for membrane fusion. Alternatively, Fzo1p could be a key regulator for mitochondrial fusion, as many GTPases play regulatory roles in diverse biological processes. Rab GTPases, for example, are important regulators for fusion events in the secretory pathway (34). The challenge for the future is to identify the components interacting with Fzo1p, and to determine the precise role of Fzo1p and each of its yet unknown partner proteins in mitochondrial membrane fusion.