Multiple Pathways Influence Mitochondrial Inheritance in Budding Yeast
Rebecca L. Frederick, Koji Okamoto, Janet M. Shaw


Yeast mitochondria form a branched tubular network. Mitochondrial inheritance is tightly coupled with bud emergence, ensuring that daughter cells receive mitochondria from mother cells during division. Proteins reported to influence mitochondrial inheritance include the mitochondrial rho (Miro) GTPase Gem1p, Mmr1p, and Ypt11p. A synthetic genetic array (SGA) screen revealed interactions between gem1Δ and deletions of genes that affect mitochondrial function or inheritance, including mmr1Δ. Synthetic sickness of gem1Δ mmr1Δ double mutants correlated with defective mitochondrial inheritance by large buds. Additional studies demonstrated that GEM1, MMR1, and YPT11 each contribute to mitochondrial inheritance. Mitochondrial accumulation in buds caused by overexpression of either Mmr1p or Ypt11p did not depend on Gem1p, indicating these three proteins function independently. Physical linkage of mitochondria with the endoplasmic reticulum (ER) has led to speculation that distribution of these two organelles is coordinated. We show that yeast mitochondrial inheritance is not required for inheritance or spreading of cortical ER in the bud. Moreover, Ypt11p overexpression, but not Mmr1p overexpression, caused ER accumulation in the bud, revealing a potential role for Ypt11p in ER distribution. This study demonstrates that multiple pathways influence mitochondrial inheritance in yeast and that Miro GTPases have conserved roles in mitochondrial distribution.

MITOCHONDRIA contribute to many cellular processes, including calcium homeostasis, cell death, cellular respiration, and metabolism. Studies in yeast, flies, worms, and mammals have established that mitochondrial shape and distribution are important for organelle function and cell survival (recently reviewed by Karbowski and Youle 2003; Chan 2006; Frazier et al. 2006; Szabadkai et al. 2006). Moreover, mitochondrial function is particularly important in neurons, where synaptic mitochondria provide the necessary energy for neurotransmitter release and recycling (Hollenbeck 2005; Ly and Verstreken 2006; Rikhy et al. 2007). Studies in the budding yeast Saccharomyces cerevisiae have advanced understanding of processes that impinge on mitochondrial function, including regulation of mitochondrial shape and distribution (recently reviewed by Shaw and Nunnari 2002; Okamoto and Shaw 2005; Escobar-Henriques and Langer 2006; Griffin et al. 2006). Even in yeast, where mitochondrial respiration is dispensable, mitochondria are essential for cell survival (Altmann and Westermann 2005; Kispal et al. 2005). Mitochondria cannot be generated de novo and must arise from existing organelles in the mother cell. Therefore, transfer of mitochondria to the emerging daughter cell is an essential process (McConnell et al. 1990; Warren and Wickner 1996).

Inheritance of the tubular mitochondrial network is actively regulated in dividing yeast cells. Mitochondria are inherited by small buds soon after bud emergence and initially appear to be associated with the bud tip. In addition, anchoring in the mother cell is thought to ensure that the mother retains a subset of mitochondria. By the time the mother and daughter are separated by cytokinesis, approximately equal amounts of mitochondria are distributed into each cell (Simon et al. 1997; Boldogh et al. 2001).

Although many aspects of mitochondrial distribution remain unclear, efficient mitochondrial distribution and inheritance in yeast requires the actin cytoskeleton. Monomeric actin polymerizes to form actin filaments, which are either bundled to make actin cables or clustered into actin cortical patches (Young et al. 2004). Cables are oriented along the mother-bud axis to facilitate bud-directed movement of cellular materials to the growing bud. Cortical actin patches, which typically function as sites of endocytosis (Kaksonen et al. 2003; Huckaba et al. 2004), localize to the new bud tip. As the bud enlarges and shifts to isotropic growth, cortical patches are distributed in the bud cortex (Pruyne and Bretscher 2000a,b; Moseley and Goode 2006). Simultaneous live imaging of yeast mitochondria and actin cytoskeleton revealed that mitochondria move along actin cables (Fehrenbacher et al. 2004). In addition, mutations that affect the organization or stability of filamentous actin structures lead to defects in mitochondrial inheritance (Drubin et al. 1993; Singer et al. 2000; Altmann and Westermann 2005).

A common feature of mutants with defects in mitochondrial inheritance is that they often display a delay, rather than a block, in inheritance. Thus, in viable mutants, newly emerged small buds lack mitochondria, but upon further growth, larger buds contain mitochondria. This observation implies that mitochondrial inheritance is not restricted to small buds. Rather, mitochondria can be moved into larger buds as well.

Two potential mechanisms for mitochondrial movement have been suggested. First, the mitochore complex, which consists of Mdm10p, Mdm12p, and Mmm1p, has been implicated in mitochondrial-actin associations, perhaps by recruitment of the Arp2/3 complex (Boldogh et al. 1998, 2003). This interaction has been proposed to facilitate anterograde, bud-directed mitochondrial movement via actin polymerization by a mechanism similar to Listeria movement (Fehrenbacher et al. 2003a,b). However, Arp2/3 functions to nucleate actin and form cortical actin patches, which themselves move in a retrograde fashion along cables toward the mother pole (Kaksonen et al. 2003; Huckaba et al. 2004). In addition, Mdm10p and Mmm1p, components of the mitochore complex, have been directly implicated in mitochondrial import of β-barrel outer membrane proteins (Meisinger et al. 2004, 2007). Moreover, Mmm1p mutations affect mitochondrial DNA nucleoid organization (Hobbs et al. 2001; Hanekamp et al. 2002; Meeusen and Nunnari 2003). Thus, the primary function(s) of the mitochore complex and its role in mitochondrial movement remain unclear.

The second proposed mechanism for mitochondrial movement depends on association with a motor. In higher organisms, kinesins and dyneins facilitate microtubule-based mitochondrial movement (Hollenbeck 1996; Hollenbeck and Saxton 2005). In yeast, the myosin V isoforms, Myo2p and Myo4p, are involved in bud-directed transport of several organelle cargoes, including vacuoles, secretory vesicles, and Golgi membranes (Pruyne et al. 1998; Schott et al. 1999; Rossanese et al. 2001; Pashkova et al. 2005). Mutational analysis of the Myo2p tail has generated alleles that specifically affect binding of single adapter proteins and therefore disrupt inheritance of specific organelles (Catlett et al. 2000; Ishikawa et al. 2003; Pashkova et al. 2006). At least one allele, myo2-573, specifically impairs mitochondrial inheritance without affecting cell polarization or vacuolar inheritance, suggesting that mitochondrial movement into buds is motor based (Itoh et al. 2002; Altmann and Westermann 2005). Overexpression of Mmr1p, a peripheral outer mitochondrial membrane protein required for efficient mitochondrial inheritance, can specifically suppress myo2-573. Co-immunoprecipitation studies raised the possibility that Mmr1p functions as a mitochondrial adapter for Myo2p (Itoh et al. 2004). Ypt11p, a Rab GTPase, also affects mitochondrial inheritance in a Myo2p-dependent manner. Ypt11p overexpression drives mitochondria into the bud, similar to what has been observed for Mmr1p (Itoh et al. 2002, 2004). Genetic interactions suggest that Mmr1p and Ypt11p do not require each other to function (Itoh et al. 2004). Ypt11p has also been suggested to anchor mitochondria at the bud tip in a Myo2p-dependent manner (Boldogh et al. 2004). Thus, at least two proteins can function to promote mitochondrial inheritance via association with Myo2p.

The mitochondrial rho (Miro) family of GTPases is conserved between yeast and higher eukaryotes (Fransson et al. 2003). Miro proteins contain two GTPase domains that flank a pair of calcium-binding EF-hand motifs. The yeast Miro ortholog, Gem1p, is important for maintenance of tubular mitochondrial morphology. In the absence of GEM1, cells contain large, globular mitochondria and display an inheritance defect in small-budded but not large-budded cells (Frederick et al. 2004). In fly larvae lacking Miro, mitochondria accumulate in cell bodies and are depleted from axons and synaptic boutons (Guo et al. 2005). The fly and human Miro orthologs bind to an adapter protein, Milton, which itself forms a complex with kinesin heavy chain (Stowers et al. 2002; Gorska-Andrzejak et al. 2003; Fransson et al. 2006; Glater et al. 2006). Several studies suggest motor protein attachment to mitochondria is modulated by Miro and Milton to facilitate transport along microtubules (Cox and Spradling 2006; Glater et al. 2006; Rice and Gelfand 2006).

We sought to understand whether the link between Miro function and mitochondrial movement is conserved in yeast. In this study, we identified synthetic sickness interactions between gem1Δ and deletion alleles of genes annotated to function in mitochondrial or endoplasmic reticulum (ER) distribution and morphology. Further analysis of selected double mutants revealed that strain fitness correlated with the efficiency of mitochondrial transfer into large buds. Genetic interactions and overexpression studies demonstrated that Gem1p, Mmr1p, and Ypt11p independently contribute to mitochondrial inheritance. Finally, ER inheritance can occur independently from mitochondrial inheritance. However, Ypt11p may function to promote inheritance of both organelles.


Strains and plasmid construction:

Standard methods were used to manipulate yeast (Sherman et al. 1986; Guthrie and Fink 1991) and bacterial (Maniatis et al. 1982) strains. All mutations, disruptions, and constructs were confirmed by PCR and DNA sequencing as appropriate. Yeast strains used for the synthetic genetic array (SGA) screen were from the consortium knockout collection (Research Genetics, Birmingham, AL) and were of the BY4741 background. All other strains were newly generated in the W303 background by homologous recombination of a PCR-generated marker cassette. Proper integration was confirmed by PCR, and strains were backcrossed prior to use. Strains and plasmids depicted in this study are listed in Tables 1 and 2. p416-GAL1-MMR1 was generated by PCR amplification of the MMR1 open reading frame that was cloned into p416-pGAL1 using the restriction enzymes XmaI and XhoI. p416-MET25-YPT11 was generated by PCR amplification of the YPT11 open reading frame which was cloned into p416-MET25 using the restriction enzymes XmaI and XhoI. For visualization of ER, pRS406-HMG1-eGFP (Du et al. 2001) was linearized with StuI for integration at the URA3 locus. pRS303-SSH1-GFP (Du et al. 2006) was linearized with EcoRI for integration at the SSH1 locus, such that SSH1-GFP was the sole copy of SSH1. Integrants were confirmed by PCR and backcrossed prior to analysis.

View this table:

Yeast strains

View this table:


The gem1Δ mmr1Δ ypt11Δ triple mutant was generated by isolation of colonies directly from sporulation plates followed by low efficiency transformation with the mtGFP plasmid. Suppressors of the slow growth of this mutant often arose. During our manipulations and analysis, we eliminated cultures that grew faster than the initial doubling time of >12 hr. Multiple independently generated triple-mutant strains showed similar levels of mitochondrial inheritance.

Synthetic lethal screen and subsequent analysis:

A synthetic genetic array approach was performed essentially as described (Tong et al. 2001; Tong and Boone 2006). The gem1Δ haploid query strain (JSY8071) or ura3Δ (JSY8040) control strain was mated to the MATa haploid deletion collection (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0). Diploids were sporulated, and selections were used to generate a collection of colonies derived from spores that were MATa double mutants. Strain handling was achieved at 1536 colony spots per Nunc Omnitray using a Biomek 2000 robot and floating pin tool. Pins were sterilized by sonication for 20 sec in 10% bleach, rinsed for 10 sec in water and then 12 sec in 100% ethanol, and dried over a fan for 30 sec. Half of the colonies were mated to a control query strain (ura3Δ∷natMX4) and half to the gem1Δ query strain such that each Omnitray contained both the query strain and control strain in duplicate for each of 384 deletion collection spots. Diploids were selected and sporulated followed by selection of MATa haploids, then MATa haploids containing the deletion collection allele, and finally MATa double mutants exactly as described (Tong and Boone 2006). Small or absent gem1Δ double-mutant colonies that were present in all previous steps of the screen were scored as having synthetic sickness, provided no fitness defects were observed for control double mutants (ura3Δ).

Via repeated screens, each deletion in the knockout collection was scored four times for genetic interactions with gem1Δ. Genes with at least two hits were considered putative interactors. Genes that were linked on the left arm of chromosome 1 where GEM1 resides were eliminated from further consideration. Of 453 putative interactions, 231 were chosen for retesting by random spore analysis (Tong and Boone 2006). Thirteen diploids did not sporulate well enough for random spore analysis to be conclusive.

Dissection of tetrads was used to confirm several interactions. Sporulation of heterozygous diploids was achieved in liquid media (1% potassium acetate plus amino acids) for 7 days at room temperature. Tetrads were digested with β-glucoronidase (Sigma, St. Louis), dissected, and allowed to germinate for 3 days at 30°. Analysis of auxotrophic markers was used to determine spore genotypes. In several cases, knockout strains were newly generated in the W303 background and used to analyze fitness defects of double mutants. For W303, diploids were sporulated for 2–3 days at 30° and allowed to germinate for 2–3 days. Images of dissection plates were acquired by a flatbed scanner and processed using Adobe Photoshop CS and Adobe Illustrator CS.

Observation of organelle morphology and inheritance:

Mitochondria were labeled by transforming cells with pYX142-Su9 (aa 1–69)-GFP+ or p414-GPD-Su9 (aa 1–69)-RFPff (Frederick et al. 2004), henceforth referred to as mitochondrial-targeted GFP (mito-GFP) and RFP (mito-RFP). The expressed fusion proteins contain the targeting sequence of subunit 9 of the F0 ATPase and localize to the mitochondrial matrix. ER was visualized as described previously with Hmg1p-eGFP or Ssh1p-GFP (Du et al. 2001, 2006). Yeast strains were grown to log phase (OD600 0.3–1.2) in appropriate synthetic dextrose medium for observation of mitochondria and ER. Organelle inheritance was scored as the presence of fluorescent signal in the bud of small- (bud diameter approximately one-third of mother cell) or large- (diameter one-half to two-thirds of mother cell) budded cells. Mitochondrial inheritance was scored by direct observation of cultures. ER inheritance and mitochondrial inheritance were simultaneously quantified by analysis of epifluorescent (mitochondria) and deconvolved epifluorescent (ER) images of random fields of cells. Each deconvolved z-section was analyzed for the presence of cortical ER in the bud periphery to prevent buds above or below the plane of the mother nucleus from being scored incorrectly. Gain settings were adjusted during scoring as needed to visualize signal. Quantification of phenotypes represents the average of three independent experiments with total n ≥ 300 cells unless noted. Bars indicate standard deviation between experiments.

Microscopy and image analysis:

Cells were observed using a Zeiss Axioplan 2 microscope with 100× oil immersion objective (NA = 1.4). Images were acquired, deconvolved, analyzed, and assembled using Zeiss Axiovision version 4.1, Adobe Photoshop CS, and Adobe Illustrator CS. Brightness and contrast were adjusted using only linear adjustments applied to the entire image. For mitochondria, each z-stack slice of 0.275 μm was deconvolved with a regularized inverse filter algorithm and all slices were projected on the transparency setting. For mitochondria and ER images, separate stacks of mito-RFP and ER-GFP were obtained with z-stack slices 0.2 μm apart and deconvolved with a regularized inverse filter algorithm. Mitochondrial projections were performed as described above. The ER depicted was a single z-stack slice after deconvolution. For each cell, a peripheral section was chosen that displayed a cortical network but no nuclear outline (Figure 4A, third section). Center sections were those in which the perinuclear ER and mother cortical ER were visible in typical rim staining patterns (Figure 4A, fourth section).

Overexpression of Mmr1p and Ypt11p:

Strains containing either p416-GAL1-MMR1 or p416-GAL1 (as a control) were grown in synthetic (S) raffinose (2%), diluted to 0.2 OD600/ml in SGalactose (2%) medium for induction, and observed 1 hr later. For overexpression of Ypt11p, cells containing either p416-MET25-YPT11 or p416-MET25 were grown to log phase in SDextrose medium, washed and induced at 0.2 OD600/ml in SDextrose lacking methionine. Organelle distribution was observed 3 hr later.

Accumulation of mitochondria in the bud was scored by visually comparing the fluorescence intensity of mito-GFP in live cells. Buds were scored as having accumulated mitochondria if the bud (Figure 3B) or region near the bud neck (Figure 3C) had greater fluorescence intensity than the mother cell. Accumulation of ER in the bud of Ypt11p or Mmr1p-overexpressing cells was scored by analysis of average fluorescence intensity in deconvolved peripheral z-sections. Buds with greater intensity were scored as containing accumulated ER. Scoring of ER accumulation was spot checked by calculation of the average pixel value of the mother and bud using ImageJ (Rasband 2007). Accumulation of mitochondria in these dual-labeling experiments was scored as above from projections of deconvolved z-stacks.


Disruption of GEM1 and genes that encode mitochondrial proteins causes synthetic growth defects:

We screened for nonessential yeast genes that displayed synthetic sickness as double mutants with gem1Δ. This strategy (Tong et al. 2001) often reveals alternate pathways that contribute to a single essential process. A gem1Δ query strain or ura3Δ control strain was mated to each of >4000 strains lacking a single nonessential gene. Diploids were sporulated and subjected to a series of selections to generate a double-mutant collection, which was then evaluated for fitness defects. Of 453 putative interactions, 231 were chosen for retesting because they displayed strong synthetic sickness and/or encoded proteins with annotated mitochondrial or cytoskeletal localization or cellular signaling function. Retesting by random spore analysis (Tong et al. 2001; Tong and Boone 2006) revealed 46 reproducible interactions (Table 3). Several of these were also verified by dissection (our unpublished data) or by independent confirmation in another strain background (as indicated, Table 3). This screen was quite sensitive as some confirmed interactors exhibited moderate synthetic growth defects in double mutants (our unpublished data). Many of the genes that interacted with gem1Δ encode mitochondrial proteins (Table 3, underlined gene names), including four of the five strongest interactors. At least some of the genetic interactions identified in the screen were not strain-background specific as we observed synthetic sickness in the W303 strain background of gem1Δ with mmr1Δ (Figure 1A), tom70Δ, fmp13Δ, and ice2Δ (our unpublished data). gem1Δ mmr1Δ double mutants, in particular, were significantly sicker than either single mutant (Figure 1A), confirming a strong genetic interaction between these two loci.

Figure 1.—

Synthetic interactions among gem1Δ and genes that affect Myo2p-dependent mitochondrial inheritance (MMR1 and YPT11) revealed that multiple pathways contribute to strain viability. gem1Δ mmr1Δ (A) and mmr1Δ ypt11Δ (C) double mutants displayed growth defects. gem1Δ ypt11Δ (B) double mutants grew as well as either single mutant. gem1Δ mmr1Δ ypt11Δ (D) triple mutants germinated poorly and grew very slowly (doubling time >12 hr). Diploid W303 strains with the partial genotypes indicated at the top were dissected, and the four progeny from a single tetrad were arranged in vertical columns. Each spore genotype is indicated by a symbol in the bottom sections corresponding to its position in the top.

View this table:

Synthetic genetic interactors with gem1Δ

Six genes that produce synthetic growth defects in combination with gem1Δ had annotated defects in mitochondrial or ER distribution and/or morphology (Table 3, gene names in boldface type; fmp13Δ strains had mitochondrial morphology defects, our unpublished data). Because recent work in mammals and flies suggested that Miro interacts with a mitochondrial motor adapter complex required for mitochondrial distribution (Fransson et al. 2006; Glater et al. 2006), we focused on this class of genes. Within this class, gem1Δ mmr1Δ strains had the strongest synthetic growth defect, therefore this interaction was used to understand the contribution of Gem1p to mitochondrial inheritance. In addition, because Ypt11p, like Mmr1p, was previously implicated in Myo2p-dependent mitochondrial distribution and because YPT11 and MMR1 genetically interact, we included Ypt11p in our analysis.

Unlike gem1Δ mmr1Δ double mutants, gem1Δ ypt11Δ double mutants grew nearly as well as either single mutant (Figure 1B). This observation was consistent with results from the screen, in which ypt11Δ was represented but no synthetic interaction was found. Previous studies demonstrated that mmr1Δ ypt11Δ double mutants were synthetically lethal (Itoh et al. 2004). In our W303 strain background, mmr1Δ ypt11Δ double mutants displayed a severe growth defect, but germinated and could be propagated for further analysis (Figure 1C).

Synthetic growth defects in gem1Δ mmr1Δ and mmr1Δ ypt11Δ mutants correlate with defective mitochondrial inheritance:

To test the hypothesis that the mitochondrial inheritance defects previously observed in gem1Δ, mmr1Δ, and ypt11Δ single mutants (Itoh et al. 2002, 2004; Frederick et al. 2004) contribute to synthetic sickness in double mutants, we visualized mitochondria using mito-GFP. The presence of any detectable mitochondria in the bud was scored as inheritance (Figure 2, A, B, D, G, and H). In wild-type cells, mitochondria were efficiently inherited by both small (91%) and large buds (100%; Figure 2, A, D, and I). In the gem1Δ or mmr1Δ single mutants, mitochondrial inheritance was defective in some small-budded cells (61% inheritance for both), but occurred later in the cell cycle such that faithful inheritance was observed in large buds (98–99%; Figure 2I). The gem1Δ mmr1Δ double mutant displayed a severe inheritance defect in small-budded cells (28% inheritance; Figure 2C). In contrast to the single mutants, this defect did not merely represent an inheritance delay as only 51% of large-budded cells inherited mitochondria (Figure 2, E and F). Of the large buds that displayed mitochondrial inheritance, about half contained only one or two mitochondrial pieces (our unpublished data, Figure 2, G and H). In addition, and as observed in other severe mitochondrial inheritance mutants (McConnell et al. 1990), gem1Δ mmr1Δ cells were often multibudded, contained large vacuoles, and appeared generally unhealthy (our unpublished data; Figure 2H). By contrast, the gem1Δ ypt11Δ mutant, which grew well, had efficient inheritance in large-budded cells (95%) and slightly impaired inheritance (44%) in small-budded cells compared to the single gem1Δ mutant (Figure 2I). In agreement with previous observations (Itoh et al. 2004), deletion of both MMR1 and YPT11 conferred severe mitochondrial inheritance defects in both small- and large-budded cells (5 and 28% inheritance, respectively; Figure 2I). As in the gem1Δ mmr1Δ mutant, large-budded mmr1Δ ypt11Δ cells that inherited mitochondria often had only one or two small mitochondrial pieces (our unpublished data). Thus, defective mitochondrial inheritance by large buds correlated with strain sickness (Figure 1). Importantly, mutant cells with normal mitochondrial morphology can still display severe mitochondrial inheritance defects (Table 4), indicating that these defects do not correlate with abnormal mitochondrial morphology.

Figure 2.—

Mitochondrial inheritance defects correlated with reduced viability in synthetic double mutants. (A, B, D, and E) Examples of successful mitochondrial inheritance. (C and F) Examples of defective mitochondrial inheritance. (G and H) Examples of severely reduced mitochondrial inheritance (≤2 mitochondrial pieces per bud). Representative small- (A–C) and large- (D–H) budded cells of wild type (A and D) and gem1Δ mmr1Δ (B, C, and E–H) cultures. (I) Quantification of small- (shaded bars) and large- (solid bars) budded inheritance in strains of the indicated genotype. Presence of any mitochondria in the bud was scored as successful mitochondrial inheritance. Signal derived from mito-GFP was overlaid on the DIC image in this and other figures. N ≥ 150 for each genotype and bud size. Error bars indicate standard deviation between at least three independent experiments in this and other figures. Bar, 5 μm.

View this table:

Mitochondrial morphology does not correlate with severity of mitochondrial inheritance defects

Gem1p, Mmr1p, and Ypt11p each contribute to an essential process:

In cases where mitochondrial inheritance is absolutely blocked, buds fail to separate, resulting in lethality (McConnell et al. 1990). To determine whether the Gem1p-, Mmr1p-, and Ypt11p-dependent pathways make individual contributions to mitochondrial inheritance, we constructed the gem1Δ mmr1Δ ypt11Δ triple mutant. As shown in Figure 1D, cells lacking all three genes only formed microcolonies. Thus, Gem1p, Mmr1p, and Ypt11p all make independent contributions to cell viability.

To confirm that the severe synthetic sickness of the triple mutant does not simply reflect a “sick and sicker” interaction, we grew several triple mutants and analyzed mitochondrial inheritance. gem1Δ mmr1Δ ypt11Δ strains used for this analysis grew very slowly (doubling time >12 hr) and were often multibudded. Fewer gem1Δ mmr1Δ ypt11Δ triple-mutant large buds inherited mitochondria (13 ± 1%) compared to mmr1Δ ypt11Δ double-mutant large buds (28 ± 4%). The Student's t-test (P = 0.021) indicates this is a statistically significant difference. Those buds that inherited mitochondria almost always had only one or two small mitochondrial pieces. It was not possible to score small buds because they were rarely found in these cultures. This residual level of mitochondrial inheritance could indicate that additional molecules or pathways mediate inheritance in the triple mutant. Alternatively, it could reflect the minimum level of mitochondrial inheritance that can be observed in large-budded cells due to stochastic events that allow single mitochondrial tubules to cross the bud neck and enter the bud.

Mmr1p- and Ypt11p-mediated mitochondrial accumulation phenotypes do not depend on Gem1p:

To further test whether these proteins can function independently, we took advantage of previous observations that overexpression of MMR1 or YPT11 causes accumulation of mitochondria in buds with a concomitant decrease in the complement of mother cell mitochondria (Itoh et al. 2002, 2004). We constructed overexpression plasmids for Mmr1p and Ypt11p and characterized the mitochondrial distribution in wild-type and mutant strains.

During overexpression of MMR1, more mitochondria were observed in the bud than the mother cell, as reported previously (Figure 3B) (Itoh et al. 2004). In addition, mitochondria accumulated at the bud neck in large-budded cells (Figure 3C), a phenotype consistent with a Myo2p-dependent process (Myo2p participates in cytokinesis, Wagner et al. 2002). By contrast, control cells contained approximately equal amounts of well-distributed mitochondria in mother and daughter cells (vector only; Figure 3A). Mmr1p-mediated mitochondrial accumulation did not depend on Gem1p, as gem1Δ cells also exhibited accumulation in the bud during Mmr1p overexpression (Figure 3, E and F). Consistent with previous observations (Itoh et al. 2004), mitochondrial accumulation in YPT11 and ypt11Δ cells overexpressing Mmr1p was similar (Figure 3G), indicating that Mmr1p can function independently of Ypt11p. We further tested independence of these pathways by overexpressing Mmr1p in the absence of both Gem1p and Ypt11p. gem1Δ ypt11Δ buds exhibited mitochondrial accumulation during Mmr1p overexpression (Figure 3G). Thus, Mmr1p does not require either Ypt11p or Gem1p for its function.

Figure 3.—

Mmr1p or Ypt11p overexpression cause mitochondria to accumulate in buds independent of Gem1p function. (A–G) Mmr1p overexpression was achieved by galactose induction for 1 hr after preculture in raffinose. (A and D) GEM1 and gem1Δ strains containing empty vector. (B and E) Mitochondria accumulated in the bud of wild-type and gem1Δ strains. (C and F) Mitochondria accumulated at the bud neck of wild-type and gem1Δ strains. (G) Quantification of accumulation at the bud or bud neck region in control cells (shaded bar) and cells overexpressing Mmr1p (solid bar) of the indicated genotype. n ≥ 300 budded cells. (H–N) Ypt11p overexpression. (H and K) GEM1 and gem1Δ cells containing empty vector. (I, J, L, and M) Representative images of GEM1 and gem1Δ cells grown in medium lacking methionine for 3 hr to overexpress MET25-driven YPT11. (N) Quantification of mitochondrial accumulation in control experiments (shaded bars) or during Ypt11p overexpression (solid bars) in strains of the indicated genotypes. n ≥ 300 medium and large-budded cells. Bar, 5 μm.

Overexpression of Ypt11p in wild-type cells caused mitochondrial accumulation in 73% of wild-type buds compared to 3% of buds in vector-only control populations (Figure 3, H–J and N), consistent with previous observations (Itoh et al. 2002). In the absence of GEM1, mitochondria still accumulated in the daughter cell during Ypt11p overexpression (Figure 3, K–M and N). Ypt11p overexpression caused mitochondrial accumulation even in the absence of Mmr1p, in agreement with previous observations (Figure 3N) (Itoh et al. 2004). Moreover, deletion of both Mmr1p and Gem1p did not affect accumulation phenotypes. Therefore, Ypt11p requires neither Mmr1p nor Gem1p for its function. Interestingly, despite the gross mitochondrial morphology defects in the absence of Gem1p, mitochondria could still be efficiently moved into the daughter bud during this overexpression experiment (globular mitochondria, Figure 3M), demonstrating that globular mitochondria are still capable of interacting with at least some of the transport machinery.

Mitochondrial inheritance is not a prerequisite for ER inheritance:

Several observations raised the possibility that ER and mitochondrial inheritance are linked. First, physical interactions between mitochondria and peripheral ER in mammalian cells and yeast were discovered (Csordas et al. 2006; Perktold et al. 2007). Second, ypt11Δ, a mutant with established mitochondrial inheritance defects (Figure 2) (Boldogh et al. 2004; Itoh et al. 2004), was reported to reduce cortical ER abundance in small buds (Buvelot Frei et al. 2006). Third, several of the genes that interact genetically with gem1Δ (Table 3) promote cortical ER distribution, including ICE2 (Estrada de Martin et al. 2005) and PTC1 (Du et al. 2006). Despite these observations, careful analysis of ER distribution under conditions that impair mitochondrial inheritance had never been performed. Thus, we assessed mitochondrial inheritance and cortical ER distribution together in mutants that have mild or severe mitochondrial inheritance defects.

Cortical ER inheritance occurs when an ER tubule moves into the bud tip. Soon afterwards, ER is distributed around the cortex of the small bud (Fehrenbacher et al. 2002; Du et al. 2004). This cortical distribution was visualized as a rim stain around the edge of the bud in z-slices through the center of the bud (Figure 4A, fourth section). Perinuclear ER, which surrounds the nucleus, was also visible in the mother cell. Cortical ER was also often visible as networks in peripheral z-sections closer to the top or bottom of the cell (Figure 4A, third section). We observed faithful ER inheritance and cortical distribution in 88% of small buds with our imaging conditions (Figure 4D), while only 12% of small-budded wild-type cells displayed a single tubule (Figure 4C) or completely lacked ER staining in the bud (not depicted; <3% of cells in all strains). ER distribution in small buds occurs in ∼80% of gem1Δ, mmr1Δ, and ypt11Δ single mutants, in which mitochondrial inheritance is disrupted in up to 50% of small buds. Notably, even though mitochondrial inheritance was drastically reduced in gem1Δ mmr1Δ (Figure 4B) and mmr1Δ ypt11Δ double mutants, ER inheritance and distribution to the bud cortex occurred faithfully in at least 77% of small-budded cells (Figure 4D). The ptc1Δ mutation was previously shown to cause defects in both ER distribution (Du et al. 2006) and mitochondrial inheritance (Roeder et al. 1998; Du et al. 2006). ptc1Δ mutants contained a single tubule of ER in ∼45% of small buds but did not distribute ER to the cortex in those daughter cells and displayed significant defects in mitochondrial transfer to small buds (Figure 4). In Figure 4C, only dim ER signal rather than a bright cortical network was visible in the peripheral slice. In sections from the center of the cell, a lone ER tubule is visible (Figure 4C; fourth section), in stark contrast to the cortical ER distribution observed in wild-type cells (Figure 4A; fourth section). Slightly lower mitochondrial inheritance efficiency was observed in all strains during this experiment compared to efficiencies reported in Figure 2. This was due to reduced sensitivity when scoring from digital images rather than live cells.

Figure 4.—

Mitochondrial inheritance did not influence distribution of ER in the bud. Sections from left to right are DIC, mito-RFP deconvolved projection, Hmg1p-GFP-labeled ER deconvolved single z-sections at the periphery, and at the center of the cell. Inheritance and distribution of ER in wild type (A), and gem1Δ mmr1Δ (B) to the cortex of small buds is evident, regardless of mitochondrial inheritance status. (C) ptc1Δ mutant cells display inheritance of a single ER tubule but do not appropriately distribute ER to the cortex of small buds. Arrowheads mark the mother-bud neck. (D) Quantification reveals that mitochondrial inheritance is not a prerequisite for ER inheritance and cortical distribution. In the schematic, cortical ER is represented by shaded and mitochondria by solid lines. Bar, 5 μm.

Mitochondrial accumulation is not sufficient to cause alterations in ER distribution:

To further address whether mitochondrial and ER inheritance are linked, we tested whether ER accumulates in the bud when mitochondrial accumulation is caused by Mmr1p or Ypt11p overexpression. Consistent with its ER localization (Buvelot Frei et al. 2006), overexpression of Ypt11p was sufficient to cause a higher average fluorescence intensity of ER-localized Ssh1p-GFP in the bud than in the mother cell (Figure 5B, third section). This accumulation was evident in 75% of Ypt11p-overexpressing cells compared to 24% of control cells (Figure 5), and accumulation of both ER and mitochondria occurred more frequently than either alone (Figure 5C). By contrast, overexpression of Mmr1p, which has not previously been linked to ER inheritance or distribution, did not alter ER distribution (Figure 5C), even though mitochondria had accumulated in buds. Thus, mitochondrial accumulation in the bud was not sufficient to drive ectopic ER into the daughter cell. Ypt11p can therefore function in inheritance of both mitochondria and ER.

Figure 5.—

Overexpression of Ypt11p caused ER to accumulate in buds regardless of mitochondrial distribution. Sections from left to right are DIC, mito-RFP deconvolved projection, Ssh1p-GFP-labeled ER deconvolved single z-sections at the periphery, and at the center of the cell. (A) Wild-type cells expressing empty vector. (B) Overexpression of YPT11 as in Figure 3. (C) Quantification revealed that ER accumulates in buds during Ypt11p overexpression but not Mmr1p overexpression. Arrowheads mark the mother-bud neck. Bar, 5 μm.


Mitochondria cannot be generated de novo, so their distribution and inheritance in budding yeast are crucial for cell viability. We identified genetic interactions between Myo2p-dependent mitochondrial inheritance pathways and the Miro GTPase, Gem1p (Figure 1 and Table 3). Sick double-mutant strains displayed more severe mitochondrial inheritance defects than the single mutants, particularly in large-budded cells (Figure 2). Overexpression experiments indicated that each of the proteins can function independently (Figure 3). Additional studies revealed that mitochondrial inheritance status does not affect ER inheritance or cortical distribution (Figures 4 and 5).

Multiple mitochondrial inheritance pathways:

Several lines of evidence suggest that Gem1p, Mmr1p, and Ypt11p function in independent pathways to promote mitochondrial inheritance. First, the additive genetic interaction of the triple mutant compared to the single and double mutants reveals that each protein normally contributes to mitochondrial inheritance (Figure 1). Second, overexpression experiments indicate that none of the proteins are required for downstream effects of Mmr1p or Ypt11p. Because overexpression of Gem1p did not cause a substantial phenotype in wild-type cells (our unpublished data), we were unable to formally test whether Ypt11p and/or Mmr1p function downstream of Gem1p. Typically, however, additive effects such as those observed for mitochondrial inheritance and viability in the gem1Δ mmr1Δ and gem1Δ mmr1Δ ypt11Δ mutants indicate that parallel pathways have been compromised.

In other cases, additive phenotypes are due to disruption of the same step of the same pathway, often when members of the same complex are mutated (Youngman et al. 2004; Measday et al. 2005). This interpretation could apply to Mmr1p and Ypt11p because both form a complex with Myo2p. However, Mmr1p localizes to mitochondria (Itoh et al. 2004) while Ypt11p localizes to the ER (Buvelot Frei et al. 2006). Although both proteins are proposed to bind the Myo2p tail, allele specific interactions suggest they may form independent complexes with Myo2p (Itoh et al. 2002, 2004). In addition, Mmr1p and Ypt11p affect ER distribution differently when overexpressed (Figure 5), suggesting that they have distinct in vivo functions. Therefore, our data are most consistent with the interpretation that Gem1p, Mmr1p, and Ypt11p function independently to promote mitochondrial inheritance.

Our observations raise the question of why organisms would have multiple, independent mechanisms to coordinate organelle inheritance. For essential processes like mitochondrial inheritance, having a so-called “backup” system would allow organism survival during failure of one mitochondrial inheritance mechanism. It seems more likely, however, that these pathways differentially contribute to inheritance during the yeast life cycle (Suda et al. 2007) or under other physiological conditions not replicated in the laboratory.

The role of physical associations between organelles in inheritance:

Recently, physical associations between mitochondria and ER have been observed in mammalian cells and yeast (Csordas et al. 2006; Perktold et al. 2007). These physical contacts could, in principal, contribute to distribution of both organelles during division. However, our findings indicate that ER inheritance and distribution are not affected by mitochondrial inheritance nor can mitochondrial accumulation drive ER accumulation in the bud. Thus, in yeast, the physical association between mitochondria and ER is not essential for ER inheritance. Previous studies showed that the sec3Δ mutant, which displays ER inheritance defects, does not have mitochondrial inheritance defects (Wiederkehr et al. 2003). Together, these observations demonstrate that mitochondrial inheritance and ER inheritance are not strictly coupled processes. In addition, we found no evidence for correlation between mitochondrial inheritance and nuclear inheritance in cells with defective mitochondrial inheritance (our unpublished data).

Conserved functions for Miro GTPases:

We previously demonstrated that Gem1p functions to maintain mitochondrial morphology, retain mitochondrial DNA nucleoids, and promote mitochondrial inheritance in yeast (Frederick et al. 2004). More recent work has linked Miro GTPases to kinesin proteins required for microtubule-based mitochondrial movement in higher organisms (Guo et al. 2005; Cox and Spradling 2006; Fransson et al. 2006; Glater et al. 2006). We also observed both mitochondrial morphology and inheritance defects in miroΔ fission yeast (our unpublished data). The synthetic interactions we report here demonstrate that the contribution of Gem1p to mitochondrial inheritance in yeast is important for cell viability. Therefore, Gem1p/Miro has a conserved function in mitochondrial distribution. However, mitochondrial movement in yeast occurs on actin cables rather than microtubules, suggesting that the molecular details of Miro function must differ in different organisms.

Miro proteins have been implicated in kinesin-based mitochondrial movement in multicellular organisms, although a stable physical interaction between Miro and kinesin has not yet been captured. We considered the possibility that Gem1p coordinates with Myo2p to facilitate mitochondrial inheritance. However, to date we have been unable to find evidence for this hypothesis. Yeast two-hybrid and co-immunoprecipitation studies failed to detect interactions between Gem1p and Myo2p (our unpublished data). Identification of additional Miro protein-binding partners and characterization of hypomorphic alleles should help illuminate the molecular mechanisms by which Miro proteins contribute to mitochondrial distribution.


We thank members of the Shaw laboratory for helpful discussions. Charlie Boone (University of Toronto), Susan Ferro-Novick (Yale University, New Haven, CT), Yasushi Matsui (University of Tokyo), Mike Yaffe (University of California, San Diego), Lois Weisman (University of Michigan Life Sciences Institute), Brad Cairns (University of Utah), and Markus Babst (University of Utah) graciously provided strains or plasmids used in this work. Jim Metherall, Darren Warnick, and Casey Barnett assisted with robotics development and training. Support for this work was provided by the National Science Foundation Graduate Research Fellowship Program (to R.L.F.), the United Mitochondrial Disease Foundation (to K.O.), the American Heart Association (to K.O.), and the National Institutes of Health (to J.M.S., GM-53466). The University of Utah Health Sciences DNA/Peptide and Sequencing Facilities are supported by National Cancer Institute grant 5-P30CA42014.


  • 1 Present address: Department of Embryology, Carnegie Institution of Washington, Baltimore, MD 21218.

  • 2 Present address: Division of Molecular Cell Biology, National Institute for Basic Biology, Myodaiji, Okazaki 444-8585, Japan.

  • Communicating editor: F. Winston

  • Received October 6, 2007.
  • Accepted November 30, 2007.


View Abstract