The Yeast Dynamin-like Protein, Mgm1p, Functions on the Mitochondrial Outer Membrane to Mediate Mitochondrial Inheritance

Yeast strains used in this study are listed in Table I. Culture and genetic analysis of yeast were performed by standard procedures (Rose et al., 1990). Semisynthetic lactate medium was prepared as described (Daum et al., 1982). Plasmid DNA was prepared from Escherichia coli strain DH5α.

The mdm17 mutant was isolated in a screen for novel alleles of another mitochondrial inheritance gene, mdm13. Strain MYY900, containing the mdm13 lesion, was crossed to the SL collection of temperature-sensitive strains which were derived from strain MYY290, and the resulting diploids were tested for growth at 37°C on yeast extract/peptone/glucose (YPD) medium. Diploids that failed to grow were sporulated and the meiotic progeny were analyzed by backcrossing and allelism tests. The recovered mdm17 spores were backcrossed three times to the wild-type parental strain, and the temperature-sensitive growth and mitochondrial distribution defects were shown to cosegregate in a 2:2 pattern. The backcrossed strain no longer displayed nonallelic noncomplementation with mdm13 and that property apparently depended on the presence of a third, unlinked mutation in the original mdm17 mutant strain. The MDM13 gene has yet to be isolated.

mdm17 cells were transformed with a yeast genomic DNA library in the centromere-based LEU2 vector p366 (M. Hoekstra, ICOS Inc.). 10,000 Leu⁺ transformants were screened for complementation of the temperature-sensitive growth defect by replica plating to YPD medium and to yeast extract/peptone/glycerol (YPG) medium at 37°C. Four different complementing clones were isolated, and restriction analysis demonstrated that they contained overlapping yeast DNA inserts. The identity of the inserts was determined by nucleotide sequence analysis of the ends of the inserts and comparison to sequences in the Saccharomyces Genome Database.

A full-length version of the MGM1 gene was synthesized using PCR. The high-fidelity polymerase, Pfu (Stratagene), was used with primers 5′-CTCTCTAGAGTTCTTCTGCTCGCTAATGGTAAATG-3′ and 5′-CTCCTCGAGGCAAGAAGATGAGTTGGATGAAGG-3′ to amplify the MGM1 open reading frame together with ∼500 bp of flanking DNA sequences. The PCR product was phosphorylated with T4 DNA kinase and ligated into vector pRS313 (Sikorski and Hieter, 1989) that had been digested with SmaI and EcoRV and treated with calf intestinal phosphatase. The resulting plasmid was designated pRS313-MGM1.

For integrative mapping, a 2.6-kb HindIII fragment adjacent to the MGM1 locus was subcloned into the YIp5 vector (Struhl et al., 1979). The plasmid was linearized by digestion with HpaI and the DNA was transformed into strain MYY290. Ura⁺ transformants were crossed to strain MYY971 containing the mdm17 mutation. The meiotic progeny of the resulting diploid consisted of 19 parental ditype and 1 tetratype, mapping the mdm17 lesion to within 2.5 cM of the MGM1 locus.

A gene replacement cassette was created by PCR as described by Baudin et al. (1993). The oligonucleotides 5′-ATGAGTAATTCTACTTCATTAAGGGCCATCCCAAGAGTGGATTGTACTGAGAGTGCACC-3′ and 5′-TCATAAATTTTTGGAGACGCCCTTGTAGCTTTTCTTGAAAGTGCGGTATTTCACACCGC-3′ were used as PCR primers, and the LEU2 gene in plasmid pRS305 (Sikorski and Hieter, 1989) served as the template. The resulting PCR product was used to transform the diploid strain MYY298. Leu⁺ transformants were screened by PCR to identify strains containing a replacement of one of two copies of MGM1 with LEU2. The heterozygous diploid strain was sporulated to obtain haploid segregants with the mgm1::LEU2 (mgm1-null) mutation.

For phenotypic analysis, cultures were first grown overnight in YPD-liquid medium at 23°C and diluted to 0.5 A₆₀₀/ml before incubation at 37°C for varying times. To evaluate respiration competence, cells were plated on YPD-agar medium, cultured at 23°C, and replica-plated to YPG medium. Colonies that failed to grow on YPG at 23°C were scored as having lost respiratory function. At least 700 colonies were scored for each time point.

To characterize mitochondrial distribution and morphology in living cells, mitochondria were stained with the membrane potential–sensitive dye 2-(4-dimethylaminostryl)-1-methylpyridium iodide (DASPMI), and cells were examined by fluorescence microscopy (Yaffe, 1995). Indirect immunofluorescence microscopy was performed on fixed yeast cells as described (McConnell et al., 1990).

Antibodies were raised against a peptide, CGGYKGVSKNL, of which the last eight residues correspond to the extreme COOH terminus of Mgm1p. Peptide synthesis, coupling of peptide to keyhole-limpet hemocyanin, and immunization of rabbits was carried out by Research Genetics, Inc. The antibodies were purified on an affinity column prepared by coupling the peptide antigen to Affigel 10 (Bio-Rad Laboratories).

Cells were grown in semisynthetic lactate medium (Daum et al., 1982), converted to spheroplasts, homogenized, and fractionated by differential centrifugation as previously described (Schauer et al., 1985; Yaffe, 1991). Mitochondrial subfractions were isolated as described (Daum et al., 1982). Mitochondria were subjected to trypsin treatment or extraction with sodium carbonate or sodium chloride as previously described (Sogo and Yaffe, 1994).

Wild-type cells were grown in minimal medium lacking methionine and cysteine to an A₆₀₀ = 0.5–0.8. Cells were harvested, resuspended in a solution of 40 mM KPi, pH 6.0, 1% glucose, and labeled for 5 min at 30°C by addition of 0.1 mCi [³⁵S]-Translabel (ICN) per ml. The chase was initiated by adding unlabeled methionine and cysteine to 20 μM and, in some cases, cycloheximide to 1 μg/ml. Cells were collected and processed for immunoprecipitation as previously described (Yaffe, 1991).

Gel filtration analysis of Mgm1p was similar to that described by Rapaport et al. (1998). Briefly, 5 mg of mitochondria was pelleted and resuspended in 1 ml buffer A (150 mM potassium acetate, 1% Triton X-100, 4 mM magnesium acetate, 0.5 mM EDTA, 0.5 mM PMSF, and 30 mM Tris-HCl, pH 7.4). The resuspended mitochondria were sonicated for 15 min in an ice-water bath, and the mixture was centrifuged for 15 min at 90,000 g. The supernatant was loaded onto a 100-ml Sephacryl-300 column (Pharmacia Biotech Inc.) that had been equilibrated with buffer A and fractionated with an FPLC system (Pharmacia Biotech Inc.) at a flow rate of 0.5 ml/min. 1-ml fractions were collected, proteins were precipitated with TCA, and analyzed by SDS-PAGE and immunoblotting. Protein bands on the immunoblots were scanned and quantified using NIH Image software.

The mdm17 gene was amplified by PCR from genomic DNA isolated from mutant cells using Taq polymerase. DNA products were cloned into the TA vector (Invitrogen Corp.) and sequenced by the dideoxy method (Sanger et al., 1977). Additionally, portions of mdm17 were amplified from genomic DNA using Pfu polymerase (Stratagene) and sequenced directly using the CircumVent Kit (New England Biolabs, Inc.).

Site-directed mutagenesis to create point mutations in MGM1 was carried out using the “PCR SOEing” technique (Horton, 1995). The PCR products containing the mutations were used to replace corresponding wild-type sequences in pRS313-MGM1 via standard cloning techniques. DNA sequence analysis confirmed the presence of the novel mutations and the absence of any PCR-generated errors. Plasmids containing the mutant genes were transformed into the heterozygous (MGM1/mgm1::LEU2) diploid strain MYY973. Transformants were sporulated, tetrads dissected, and mgm1-null spores harboring each of the mutant constructs were identified and analyzed.

The mdm17 mutant was isolated in a screen for novel alleles of mdm13, an uncharacterized gene that is required for normal mitochondrial inheritance and morphology. This screen involved crossing a haploid strain harboring the temperature-sensitive mdm13 lesion to a collection of temperature-sensitive strains and analyzing growth of the resulting diploids at the nonpermissive temperature (37°C). One diploid strain displayed temperature-sensitive growth, and analysis of its meiotic progeny indicated that one of two mutations conferring temperature-sensitive growth mapped to a genetic locus unlinked to mdm13. Therefore, rather than being an allele of mdm13, the new mutation defined a distinct gene. Analysis of cells harboring only the new mutation (after its genetic isolation from mdm13) by staining with the mitochondria-specific, vital dye DASPMI and fluorescence microscopy revealed defects in mitochondrial distribution and morphology after incubation at the nonpermissive temperature (Fig. 1). Complementation and allelism tests indicated that the new mutation was unlinked to previously characterized mdm mutations. Genetic analysis further demonstrated that the defects in growth at 37°C, mitochondrial distribution, and mitochondrial morphology were caused by a single nuclear mutation. This novel mutation was designated mdm17.

To quantify the effect of mdm17 on mitochondrial distribution and morphology, cells were stained with DASPMI and examined by fluorescence microscopy. In wild-type cells incubated at either 23°C or 37°C, mitochondria appeared as extended, tubular structures, distributed throughout the cytoplasm of both mother and bud portions of the cell (Fig. 1, A and B). Mitochondria in cells harboring the mdm17 mutation appeared very similar to wild-type at 23°C (Fig. 1 A) but displayed dramatic changes in mitochondrial distribution and morphology after incubation at 37°C (Fig. 1 B). In particular, mitochondrial aggregation and empty daughter buds were apparent in many cells. These alterations in mitochondrial distribution and morphology were confirmed by indirect immunofluorescence microscopy (data not shown).

To examine the development of mutant phenotypes at 37°C, populations of mutant cells were analyzed by indirect immunofluorescence microscopy at various times after shifting to the nonpermissive temperature (Fig. 2, A and B). After 1 h at 37°C, most cells (87%) contained aggregated mitochondria, and a significant fraction of cells (18%) possessed buds devoid of mitochondria. Some of the cells with empty buds still possessed mitochondria with normal tubular morphology (data not shown). By 2 h, 90% of cells displayed aggregated mitochondria and a majority (56%) possessed empty buds. These results demonstrate that the mdm17 mutation rapidly causes defects in mitochondrial distribution and morphology after shifting cells to the nonpermissive temperature.

The fraction of cells in the mutant population that showed mitochondrial staining with the potential-dependent dye DASPMI decreased during incubation at the nonpermissive temperature, suggesting that the cells were becoming respiration deficient. To analyze this phenotype, mdm17 cells were incubated at 37°C for various times, plated on glucose (YPD) medium, and cultured at 23°C. Colonies were tested for respiratory competence by replica plating onto glycerol (YPG) medium. At early times, almost all cells were respiration competent (Fig. 2 C). Significant numbers of respiration-deficient cells began appearing in the population only after 4 h at 37°C. After 24 h, >95% of the cells were respiration deficient (Fig. 2 C). When cells from the 16-h time point were stained with the DNA-binding dye 4,6-diamino-2-phenylindole and examined by fluorescence microscopy, mitochondrial DNA staining was no longer evident in a majority of cells (data not shown), indicating that the respiration deficiency reflected a loss of mitochondrial DNA.

To identify the molecular basis for the mdm17 phenotypes, the wild-type MDM17 gene was cloned by complementation of the temperature-sensitive growth defect of mdm17 mutant cells. From 10,000 transformants, four strains harboring complementing plasmids were isolated. Restriction mapping of the yeast genomic DNA inserts in these plasmids indicated that all contained overlapping clones. DNA sequence analysis and comparison with sequences in the Saccharomyces Genome Database indicated that the genomic inserts in the plasmids corresponded to a region of chromosome XV. Deletion analysis of the plasmids localized complementing activity to a single open reading frame corresponding to the previously identified MGM1 gene (Jones and Fangman, 1992). Transformation of mdm17 cells with a centromere-based plasmid containing only a single copy of MGM1 complemented all of the mutant phenotypes. Finally, the mdm17 mutation was localized to the MGM1 locus by integrative transformation and mapping (described in Materials and Methods). These results demonstrate that mdm17 is a mutant allele of MGM1.

Previous studies reported that mgm1 mutant cells readily lost mitochondrial DNA and contained aggregated mitochondria but showed normal mitochondrial distribution to buds (Guan et al., 1993). This latter property differs from the phenotype of the mdm17 mutant, where a large fraction of cells exhibited buds devoid of mitochondria. Accordingly, the phenotype of a null mutant in mgm1 generated in the same strain background as the mdm17 mutant was examined. Indirect immunofluorescence microscopy revealed many mgm1-null cells with buds devoid of mitochondria, even in cells cultured at 23°C (Fig. 3). As previously described (Guan et al., 1993), mitochondria in the mgm1-null strain were extensively aggregated.

Previously reported phenotypes of the mgm1 mutant, together with the resemblance of the protein's predicted NH₂ terminus to a mitochondrial targeting sequence, suggested that Mgm1p is a mitochondrial protein (Guan et al., 1993). However, direct evidence of the protein's subcellular location was lacking. To determine the functional location of Mgm1p, antibodies were generated that were specific for a peptide corresponding to the extreme COOH terminus. These antibodies were used to detect the protein in subcellular fractions. Immunoblotting of total cellular extracts indicated that the affinity-purified antibodies recognized two polypeptides of ∼100 and 90 kD (Fig. 4 A). Neither of these species was apparent in mgm1-null cells (Fig. 4 A), indicating that both polypeptides are products of the MGM1 gene. Furthermore, both species displayed substantially increased levels in cells harboring a multicopy (2 μ) plasmid encoding MGM1 (Fig. 4 A). These increased levels had no apparent effect on mitochondrial distribution and morphology (data not shown).

The exact relationship between the two Mgm1p species is unclear, but the 90-kD form appears to be derived from the 100-kD form by proteolysis. The relative amounts of the two species varied randomly in different experiments, and pulse–chase analysis in intact cells failed to reveal a biosynthetic precursor-product relationship (data not shown). However, prolonged incubation of subcellular fractions in vitro (even at 0°C in the presence of a cocktail of protease inhibitors) led to conversion of the larger species to the smaller form (data not shown). These results suggest that the 90-kD form is a product of the partial proteolytic degradation of the 100-kD species.

Earlier characterization of MGM1 failed to identify which of five NH₂-terminal methionine residues in the predicted open reading frame actually constituted the translational start of the protein (Jones and Fangman, 1992; Guan et al., 1993). To resolve this uncertainty, in vitro mutagenesis was used to generate versions of mgm1 in which individual methionine codons were converted to alanine codons. These mutant versions were tested for their complementation of the phenotypes of mgm1-null cells. No complementation was observed (and the protein was not detected) when methionine-2 was changed to alanine (data not shown). Mutation of the remaining four methionines did not affect complementation of the mutant defects, and both 100- and 90-kD forms of Mgm1p were detected in cells harboring these genes. These results indicate that the authentic translational start residue is methionine-2 (referred to henceforth as residue 1 of the coding region).

The subcellular distribution of Mgm1p was characterized by immunoblotting of proteins extracted from subcellular fractions isolated from wild-type cells. Mgm1p was localized predominantly to a subcellular fraction enriched in mitochondrial proteins (Fig. 4 B). Analysis of purified submitochondrial fractions revealed that Mgm1p was associated with the mitochondrial outer membrane (Fig. 4 C). Furthermore, Mgm1p was susceptible to protease treatment of intact mitochondria under conditions where internal proteins, such as the matrix-localized Mas2p, were protected (Fig. 4 D).

To examine the nature of Mgm1p association with the outer membrane, isolated mitochondria were extracted with sodium chloride and with sodium carbonate. The latter treatment strips peripheral proteins but leaves integral proteins embedded in the membrane (Fujiki et al., 1982). Mgm1p was resistant to extraction with 1 M NaCl (Fig. 5), indicating that the protein was tightly associated with the membrane. After carbonate treatment, the 90-kD form of Mgm1p was recovered in the soluble fraction, while the 100-kD form remained associated with the membrane (Fig. 5). This result indicates that the full-length Mgm1p is an integral membrane protein. The behavior of the 90-kD fragment suggests that Mgm1p is embedded in the membrane via the putative membrane-spanning domain in the NH₂ terminus, since both Mgm1p species are recognized by the antibody specific for the extreme COOH terminus.

The state of Mgm1p in the mitochondrial outer membrane was further characterized by gel filtration analysis of mitochondrial proteins solubilized in 1% Triton X-100. About 60% of Mgm1p eluted in a peak corresponding to a molecular mass of ∼400 kD, while the remainder of the protein eluted in a peak appropriate to the monomeric size (90–100 kD) (Fig. 6). The 90-kD (fragment) form of the protein eluted in a distinct peak with an apparent molecular mass of ∼120 kD (data not shown). The elution behavior of Mgm1p was distinct from that of two other integral proteins of the mitochondrial outer membrane, Tom70p and OM45 (Fig. 6). These results suggest that Mgm1p is part of a high molecular mass complex either in a homo-oligomeric state or complexed with other proteins on the mitochondrial surface.

To explore further the function of Mgm1p, the identity of the mdm17 mutation was determined. The mutant gene contained a single change: a transition of G to A at nucleotide 880, resulting in a change of Glu²⁹⁴ to Lys. This lesion mapped near the conserved, tripartite GTP-binding motif, suggesting a role for this site in Mgm1p function. To test the importance of the GTP-binding site, a mutant form of MGM1 incorporating a change in a conserved residue, S224N, was created (the numbering of this residue is based on the designation of the second methionine in MGM1 as the NH₂-terminal residue as described above). The equivalent mutation in the H-ras-GTPase (S17N) eliminates the protein's binding of GTP (Feig and Cooper, 1988), and a corresponding mutation in rat dynamin (S45N) blocks the protein's function in endocytosis (Herskovits et al., 1993). Immunoblot analysis of cellular extracts detected Mgm1p in mgm1-null cells harboring the mutant gene (Fig. 7 A). However, the mgm1-S224N gene failed to complement any of the mutant phenotypes of a cell containing an mgm1-null lesion (Fig. 7 B and data not shown). In addition, the mgm1-S224N gene did not complement the growth or mitochondrial distribution and morphology defects of the mdm17 mutant cells (Fig. 7 C). These results indicate that an intact GTP-binding domain is essential for Mgm1p function.

Overexpression of dynamin family members mutated in critical residues of the GTP-binding site was previously found to cause dominant-negative effects in wild-type cells (Warnock and Schmid, 1996). To test whether this is also true for MGM1, high-copy plasmids encoding either wild-type MGM1 or mgm1-S224N were transformed into wild-type cells. Only a small number of transformants harboring the mutant gene was recovered, and Western blot analysis indicated that the mutant protein was not overexpressed in these cells (perhaps reflecting a regulatory or suppression phenomenon). However, both wild-type and mutant proteins were overexpressed in mgm1-null cells harboring the respective plasmids (data not shown). These transformed cells were mated to wild-type cells, and the effect of overexpression in the resulting zygotes was analyzed by fluorescence microscopy. High levels of wild-type MGM1 had no apparent effect on mitochondrial distribution or morphology (Fig. 8 A), but zygotes with elevated levels of the mutant protein displayed defects very similar to those caused by the original mdm17 mutation (Fig. 8 A). In particular, 90% of the latter zygotes displayed aggregated mitochondria (Fig. 8 B), and 52% of the buds produced by these zygotes contained no mitochondria (Fig. 8 B). These results demonstrate that the mgm1-S224N mutant gene product confers dominant-negative effects on mitochondrial distribution and morphology.

We have identified a role for Mgm1p in mitochondrial inheritance through the analysis of cells harboring mdm17, a mutation causing temperature-sensitive growth and alterations in mitochondrial distribution and morphology. Genetic analysis revealed that mdm17 is a mutant allele of MGM1, a nuclear gene originally identified by its role in maintenance of the mitochondrial genome (Jones and Fangman, 1992). Two key observations indicate that the primary functions of Mgm1p are to facilitate the maintenance of mitochondrial morphology and to mediate mitochondrial inheritance and that mitochondrial genome maintenance is a secondary role of the protein. First, analysis of mdm17 cells revealed that alterations in mitochondrial morphology and distribution appeared very quickly after a shift to elevated temperature, whereas loss of respiratory competence and the mitochondrial genome occurred only much later. Second, the location of Mgm1p on the mitochondrial surface is consistent with a primary function of the protein in morphology and distribution, but not with a direct role in maintenance of the matrix-localized mitochondrial DNA. In addition, several other mutations that cause aberrant mitochondrial morphology also cause increased loss of mitochondrial DNA (Burgess et al., 1994; Sogo and Yaffe, 1994; Berger et al., 1997). This loss appears to be a secondary effect of the morphological changes.

Mgm1p is one of a small group of outer membrane proteins that have been shown to be required for mitochondrial inheritance. Mdm10p, Mdm12p, and Mmm1p are also integral proteins of the outer membrane whose loss leads to dramatic alterations in mitochondrial morphology and defects in transmission of mitochondria into developing daughter buds (Burgess et al., 1994; Sogo and Yaffe, 1994; Berger et al., 1997). Although the specific molecular activities of these other components are unknown, their location in the mitochondrial outer membrane and their exposed cytoplasmic domains suggest two potential models of function. One possibility is that these proteins serve as binding sites or anchor points for interaction with cytoskeletal structures or molecular motors. Such interactions might pull mitochondria into tubular structures and mediate mitochondrial movement into buds. A second possibility is that the outer membrane components regulate structural properties of the outer membrane, such as membrane fluidity, which are themselves essential for normal mitochondrial morphology and inheritance. Additionally, mitochondrial inheritance may depend on a tubular morphology that is maintained by these outer membrane proteins. Mgm1p may function in concert with the other outer membrane proteins, or it may fulfill a unique role. This latter possibility is suggested by the observation that phenotypes caused by mdm10, mdm12, and mmm1 are suppressed by the SOT1 mutation (Berger et al., 1997) but defects caused by mutations in mgm1 are not (data not shown).

Mgm1p is a member of the dynamin family of large GTP-binding proteins. The best characterized member of this family is dynamin itself, which plays an essential role in clathrin-mediated endocytosis in animal cells (Herskovits et al., 1993; van der Bleik et al., 1993). Although S. cerevisiae does not have a true dynamin orthologue, this yeast possesses two other dynamin-like proteins, Vps1p (Vater et al., 1992) and Dnm1p (Gammie et al., 1995). Vps1p is localized to the Golgi apparatus and participates in the vesicular transport of proteins to the vacuole (Vater et al., 1992). Dnm1p may function in the endosomal pathway (Gammie et al., 1995), but was recently found to play an important role in mitochondrial distribution and morphology (Otsuga et al., 1998). In particular, cells deleted for Dnm1p displayed mitochondrial tubules collapsed along one side of the cell and extending from the mother portion of the cell into the daughter bud. A related phenotype, the collapse of mitochondrial tubules into perinuclear aggregates, was observed in an independent study in which a mutant form of the human dynamin-like protein, Drp1, was expressed in cultured mammalian cells (Smirnova et al., 1998). These observations suggest that Dnm1p and its mammalian homologue Drp1 may mediate the lateral distribution or branching of mitochondrial tubules.

Many molecular details of dynamin's function in endocytosis remain to be clarified, but key features of the protein's activity include GTP hydrolysis and assembly of dynamin into oligomeric structures (Herskovits et al., 1993; Hinshaw and Schmid, 1995). Dynamin has been proposed to act as a mechanical constrictor for the neck of an invaginating coated pit (Warnock and Schmid, 1996), or as a regulator or recruiter of the constriction machinery (Roos and Kelly, 1997). Although it is not apparent that a constricting activity is necessary for the maintenance of mitochondrial morphology and distribution, our results suggest that Mgm1p shares two key functional features with dynamin. First, mutational analysis revealed that the GTP-binding site is essential for Mgm1p function, suggesting the importance of GTP hydrolysis. Second, the dominant-negative effect of the S224N mutation and the wild-type protein's gel filtration profile suggest that Mgm1p functions as part of an oligomeric complex. Mgm1p might play a structural role in locally influencing the properties of the outer membrane. For example, Mgm1p could assemble into a structure that promotes a tubular mitochondrial morphology. Alternatively, Mgm1p might fulfill a regulatory function in controlling the activity of other membrane components. More details of the protein's function should emerge from identification of interacting proteins.

This work was supported by grant GM44614 from the National Institutes of Health.

DASPMI

2-(4-dimethylaminostryl)- 1-methylpyridinium iodide

mdm

mitochondrial distribution and morphology mutant

YPD

yeast extract/peptone/glucose

YPG

yeast extract/ peptone/glycerol