We have demonstrated that the MDM20 gene encodes a novel protein required for proper mitochondrial segregation in yeast. Cells lacking Mdm20p are temperature sensitive for growth and fail to transport mitochondria into newly formed buds. Our results indicate that the defect in the mdm20 mutant specifically disrupts mitochondrial transport but not mitochondrial morphology in these cells. In addition, the mdm20 mutation does not severely interrupt the division and segregation of the nucleus, and other cytoplasmic organelles, such as vacuoles, are always detected in mdm20 daughter cells that fail to inherit the mitochondrial compartment. Together, our findings establish an essential role for MDM20 in mitochondrial partitioning during mitotic division.
Two different cytoskeletal elements in yeast, actin (Drubin et al., 1993
; Lazzarino et al., 1994
) and an intermediate filament–like protein (McConnell and Yaffe, 1992
), have been proposed to play a central role in mitochondrial inheritance. In this study we present two lines of in vivo evidence that mitochondrial inheritance in yeast is an actin-mediated process. First, two major phenotypic consequences of disrupting MDM20
appear to be the cessation of mitochondrial movement into buds during division and the loss of observable actin cables in cells. Second, overexpression of two well-characterized actin binding proteins, Tpm1p and Tpm2p, suppresses temperature-sensitive growth and mitochondrial inheritance defects and partially restores actin cables in the mdm20
mutant. Although it is possible that Mdm20p fulfills distinct functions in mitochondrial inheritance and actin cable organization, the observation that overexpression of Tpm1p and Tpm2p can partially rescue defects in both processes in mdm20
mutants suggests that these two phenotypes cannot be separated from one another. Moreover, a direct link between mitochondrial segregation and the actin cytoskeleton is consistent with the recent observations that yeast mitochondria colocalize with actin cables in vivo (Drubin et al., 1993
), can bind to phalloidin-stabilized actin filaments in vitro (Lazzarino et al., 1994
), and contain an actin-dependent myosin-like motor activity on their surface (Simon et al., 1995
). Thus, our data strongly support a model in which mitochondria are transported into yeast buds along polarized actin filaments or cables.
The organization and stabilization of the actin cytoskeleton in yeast depends on the activities of a large number of actin binding proteins and regulatory factors. Mdm20p is a new member of this important group of proteins required for the integrity of actin cables in cells. Previous studies indicate that defects in actin organization cause a variety of phenotypes including temperature-sensitive lethality, osmotic sensitivity, and defects in cell morphology, mitochondrial dynamics, secretion, endocytosis, and nuclear segregation (for review see Bretscher et al., 1994
; Welch et al., 1994
). In addition to the phenotypes we have reported here for the mdm20
mutant, our preliminary studies suggest that mdm20
cells are rounder than wild type and heterogeneous in size, exhibit defects in chitin localization, and accumulate small vesicles (unpublished observations). These data provide further support that MDM20
is required for normal functions of the actin cytoskeleton. Most importantly, MDM20
defines a new class of genes that control both actin organization and mitochondrial transport during division.
Previous studies of the actin–myosin complex identified a myosin “footprint” on the actin monomer (Rayment et al., 1993
; Schroder et al., 1993
). Charge-to-alanine substitutions located under the equivalent myosin footprint in the yeast actin monomer have been shown to cause defects in mitochondrial morphology, specifically mitochondrial aggregation (Drubin et al., 1993
). The mutants with aberrant mitochondrial morphology exhibit severe actin organization defects that are most pronounced at 37°C, including loss of actin cables and/or patches, delocalized patches, and reduced or faint cables. In contrast, we find that mdm20
mutants lacking actin cables exhibit disrupted mitochondrial transport but not mitochondrial morphology. Similarly, we do not observe defects in mitochondrial morphology in isogenic tpm1
mutant cells. If the actin cytoskeleton plays a role in regulating mitochondrial morphology, then the structures required must still be present in mdm20
mutant cells. Alternatively, other cytoplasmic structures or cytoskeletal proteins may be responsible for controlling yeast mitochondrial morphology.
Our observation that mitochondrial morphology and inheritance is normal in the tpm1Δ
mutant is interesting in light of previous reports that tpm1Δ
strains generate petites (lose mitochondrial genome function) at a higher frequency than wild type (Liu and Bretscher, 1992
). We also find that mdm20Δ
cells generate petites at a high frequency when maintained on dextrose. One possible explanation for these results is that tpm1
daughter cells frequently receive DNA-free mitochondrial compartments. In our experiments, mitochondrial morphology and distribution were measured by labeling the mitochondrial compartment with the potential dependent dye, DiOC6
, and labeling mtDNA nucleoids with DAPI. We observed that tpm1Δ
cells segregated both the mitochondrial compartment (Table ) and mitochondrial nucleoids (data not shown) into daughter cells. Furthermore, we did not detect a large number of tpm1
cells lacking mt nucleoids. Thus, the increased petite formation in tpm1
cells is not simply the result of a block in mtDNA transfer to buds. Further experiments are required to determine the mechanism by which these strains become petite.
Some of our studies suggest that Tpm1p and Mdm20p might perform overlapping functions required for the integrity of the actin cytoskeleton. The disruption of both genes causes a similar loss of actin cables in cells and combinations of mutations in MDM20
result in synthetic lethality. In addition, like tpm1
exhibits synthetic growth defects in combination with mutations in either BEM2
but not SAC6
(Fig. ). Furthermore, we observe that overexpression of Tpm1p restores actin cables and rescues mitochondrial inheritance defects in mdm20.
However, this suppression is not reciprocal; overexpression of Mdm20p does not rescue mutant phenotypes in tpm1
cells, suggesting that their roles in organizing actin differ. Mdm20p and Tpm1p do not appear to be structurally homologous since their amino acid sequences and sizes are quite different. If Mdm20p and Tpm1p do not have redundant functions, then an alternative explanation for the rescue of mdm20
mutant phenotypes by overexpressed Tpm1p is required. One possibility is that the overexpression of Tpm1p leads to the global stabilization of all actin filaments and cables in mdm20
cells, including those necessary for mitochondrial transport and inheritance. In support of this model, overexpression of Tpm1p is reported to cause more prominent cables and actin networks in wild-type cells and can restore actin cables in act1-2
mutant strains (Liu and Bretscher, 1989
). Although overexpression of Tpm2p also suppresses mutant phenotypes in mdm20
, we did not observe synthetic lethality in mdm20 tpm2
double mutants. Previous studies showed that the level of Tpm2p expression in wild-type cells is much lower than that of Tpm1p, and that cells lacking TPM2
do not exhibit defects in actin organization or growth (Drees et al., 1995
). Thus, mdm20Δ tpm2Δ
cells may contain enough Tpm1p to compensate for the loss of Tpm2p and prevent the cell death observed in mdm20Δ tpm1Δ
Interestingly, although mutations in MDM20 and TPM1 cause similar defects in actin organization, the mutant phenotypes in these two strains are distinct. mdm20Δ mutant cells exhibit severe defects in mitochondrial inheritance (58%) and only slight defects in nuclear segregation (5%). Conversely, tpm1Δ strains exhibit weak defects in mitochondrial inheritance (5%) and stronger defects in nuclear segregation (16%). As discussed above, while overexpression of Tpm1p can rescue mutant phenotypes in mdm20, the converse is not true. These observations suggest that Mdm20p and Tpm1p perform different roles in vivo and support the hypothesis that there are functionally distinct classes of actin-containing structures in cells. It is important to note that mdm20 and tpm1 mutant cells lacking actin cables may contain individual actin filaments that are not detected by rhodamine-phalloidin staining. Thus, actin filaments that play a role in mitochondrial inheritance may be present in the tpm1 mutant but absent in the mdm20 mutant. By the same reasoning, actin-containing structures required for nuclear segregation may be more severely disrupted in the tpm1 mutant than in the mdm20 mutant. In this way, MDM20 and TPM1 could be acting to establish molecular and functional heterogeneity in the actin cytoskeleton.
As illustrated in Fig. , our synthetic lethal studies suggest that MDM20
functions in a cellular process that also requires TPM1
, and MYO2.
Assembly and/or stabilization of the actin cytoskeleton is an obvious candidate for this process since defects in actin organization have been reported for mutations in all of these genes (Liu and Bretscher, 1989
; Johnston et al., 1991
; Kim et al., 1994
; Wang and Bretscher, 1995
). The synthetic growth defects we observe between mdm20
, and myo2
may simply reflect the fact that partial defects in actin organization can be tolerated at 25°C, but combinations of defects cannot. However, this interpretation is not supported by the findings that double mutants lacking both MDM20
(Table and Fig. ) or SAC6
(Adams et al., 1993
) are viable. An alternative possibility is that synthetic lethality between mutations in these genes results from a failure to coordinate bud site assembly with the actin polarization required for directed membrane transport and bud growth. Lillie and Brown (1994)
have suggested that Myo2p localized at the incipient bud site might function to anchor actin filaments that are required for the polarized delivery of vesicles to the bud. In support of this model, both myo2
mutants accumulate small vesicles, and we have observed similar vesicles in a small number of mdm20
cells (data not shown). Our studies of the mdm20
mutant indicate that polarized actin cables are also required for the delivery of mitochondria to buds. Physical and genetic interactions reported for BEM2
suggest that the Bem2p protein may also localize to the bud site where it is postulated to regulate Tpm1p-containing actin filaments (Wang and Bretscher, 1995
). Based on the synthetic lethal interactions observed to date, we speculate that viable mdm20
cells lacking actin cables still contain some undetected actin filaments that are properly polarized to the bud site. However, when mdm20
are combined with mutations in genes that control actin filament polarization to the bud site, synthetic lethality results.
Although MDM20 encodes a novel protein required for the organization of the actin cytoskeleton, analysis of the predicted Mdm20 amino acid sequence does not reveal any known actin binding domains. Mdm20p does contain two potential heptad repeats, however, which could modulate its homo- or heterodimerization through the formation of α-helical coiled-coils. While the specific biochemical activity of Mdm20p is unknown, it is likely to participate in the assembly or function of actin-containing structures necessary for mitochondrial segregation. In this regard, Mdm20p could bind directly to filamentous actin or might act in a signaling pathway that directs the assembly or stabilization of actin filaments or cables that transport mitochondria during cell division. We think it is unlikely that Mdm20p is a major structural component of actin filaments or cables for two reasons. First, our studies with the HA-tagged Mdm20p suggest that relatively low levels of the protein are able to completely rescue mutant phenotypes in the mdm20 strain (see Fig. ). Second, we have been unable to localize the low abundance of HA-tagged Mdm20p in wild-type or mdm20 cells (data not shown). In cells overexpressing HA-tagged Mdm20p, the protein is found dispersed throughout the cytoplasm and does not appear to be concentrated along actin cables or at actin patches (data not shown). Based on these results, we suggest that Mdm20p serves a regulatory function in cells. We are currently studying the function of this interesting protein in organizing actin and in directing mitochondrial transport and inheritance.