Eukaryotic cells replicate and partition their organelles between the mother cell and the daughter cell at cytokinesis. Polarized cells, notably the budding yeast Saccharomyces cerevisiae, are well suited for the study of organelle inheritance, as they facilitate an experimental dissection of organelle transport and retention processes. Much progress has been made in defining the molecular players involved in organelle partitioning in yeast. Each organelle uses a distinct set of factors – motor, anchor and adaptor proteins – that ensures its inheritance by future generations of cells. We propose that all organelles, regardless of origin or copy number, are partitioned by the same fundamental mechanism involving division and segregation. Thus, the mother cell keeps, and the daughter cell receives, their fair and equitable share of organelles. This mechanism of partitioning moreover facilitates the segregation of organelle fragments that are not functionally equivalent. In this Commentary, we describe how this principle of organelle population control affects peroxisomes and other organelles, and outline its implications for yeast life span and rejuvenation.
One hallmark of eukaryotic cells is the presence of membrane-enclosed organelles that permit the coexistence of different chemical microenvironments, each precisely tailored for a defined set of biochemical reactions. Cells must maintain their organelles to retain the benefits of compartmentalization. As it is either impossible or energetically unfavorable to make organelles anew, cells have evolved elaborate mechanisms to control their organelle populations. With every round of cell division, organelles are duplicated and distributed between the two resultant cells. This process, termed organelle inheritance, ensures the faithful propagation of organelles to future generations of cells.
Cells of the yeast Saccharomyces cerevisiae undergo a repetitive pattern of growth and division termed budding. Cell division in S. cerevisiae is asymmetrical, with the formation of a bud that is initially much smaller than its mother. Thus, in contrast to cells that divide by median fission, yeast cells must actively deliver a portion of their organelles to the growing bud (Fig. 1). The remaining organelles are immobilized at defined structures in the mother cell. Organelle inheritance in budding yeast can therefore be divided into distinct processes of organelle transport and organelle retention. The ability to experimentally dissect a complex phenomenon into its individual components, along with ease of genetic manipulation, has made S. cerevisiae a powerful tool for the study of organelle inheritance.
The last decade has seen the discovery of a battery of factors involved in various aspects of organelle partitioning. Research now has shifted from the identification of individual components to an elucidation of how these factors work together in the cell to effect organelle partitioning. We are beginning to understand how inheritance factors coordinate their activities to maintain a stable population of a particular organelle, as well as how the cell orchestrates the partitioning of its organelle populations as a whole.
Here, we provide a synopsis of the recent advances in the field of organelle inheritance. Based on these findings, we propose that all organelles, regardless of origin or copy number, are partitioned between mother and daughter cells by the same fundamental mechanism involving the division of organelles and their subsequent segregation. We provide examples of how inheritance factors and the divisional machinery of an organelle cooperate to divvy up the organelle population and how the seemingly antagonistic processes of organelle transport and organelle retention act synergistically during organelle inheritance. Moreover, we outline how this mechanism of organelle population control affects other cellular processes, such as aging and rejuvenation.
Moving out – organelle transport processes
The bud-directed movement of organelles in S. cerevisiae is powered by molecular motors moving along cytoskeletal tracks. At the beginning of each cell cycle, formins assemble at the site of future bud emergence and catalyze the incorporation of actin monomers into filaments (Evangelista et al., 2002; Pruyne et al., 2002; Sagot et al., 2002). The actin cables thus formed converge at the bud neck region of the cell and radiate back into the mother cell. These cables form the tracks along which most organelles travel for their inheritance (Fig. 1).
The actin-based movement of organelles is catalyzed by the class V myosins, Myo2p and Myo4p. These molecular motors contain N-terminal domains that are required for binding to F-actin and generating force through the hydrolysis of ATP, while their C-termini form globular tails that attach to cargoes by binding to organelle-specific adaptor molecules (Sellers and Veigel, 2006). Myo4p and Myo2p divide the labor of organelle transport unevenly, as Myo4p moves only cortical endoplasmic reticulum (cER) (Estrada et al., 2003), whereas Myo2p powers the movement of secretory vesicles, Golgi, mitochondria, peroxisomes and vacuoles (Fagarasanu et al., 2010). Lipid droplets also segregate to buds in a Myo2p-dependent manner (Knoblach and Rachubinski, 2014).
The identities of many of the receptors connecting the different organelles to the cargo-binding domains of the myosin V motors have now been determined. She3p and She2p are required to attach cER elements and certain mRNAs to Myo4p, which results in the bud-specific expression of these transcripts (Estrada et al., 2003; Schmid et al., 2006; Shepard et al., 2003). Peroxisomes attach to Myo2p through inheritance of peroxisomes protein 2 (Inp2p), a peroxisomal integral membrane protein (Fagarasanu et al., 2006). Myo2p also moves a portion of the vacuole from mother cell to bud by interacting with the Vac17p–Vac8p complex (Ishikawa et al., 2003; Tang et al., 2003). The binding of Myo2p to Mmr1p and the Rab GTPase Ypt11p plays a role in the segregation of mitochondria (Frederick et al., 2008; Itoh et al., 2002; 2004). Both proteins bind Myo2p directly and have been implicated in mitochondrial inheritance, as small buds of mmr1Δ and ypt11Δ cells lack mitochondria (Frederick et al., 2008; Itoh et al., 2004). However, the localization of Ypt11p to the ER (Buvelot Frei et al., 2006) and late Golgi vesicles (Arai et al., 2008) raises the question as to whether Ypt11p attaches mitochondria directly to Myo2p or moves ER elements to which mitochondria adhere (see also Box 1). Adaptor proteins that attach Myo2p to secretory vesicles include the Rab GTPases Ypt31p, its paralog Ypt32p and Sec4p, as well as the exocyst component Sec15p, which mediates the polarized delivery of transport vesicles to the plasma membrane (Jin et al., 2011; Lipatova et al., 2008; Santiago-Tirado et al., 2011). Myo2p also carries the plus ends of cytoplasmic microtubules to the bud through interaction with the Kar9p–Bim1p complex (Beach et al., 2000; Yin et al., 2000) and thus orients the nucleus prior to mitosis. Of note, the nucleus itself, along with the perinuclear ER, is segregated to the bud by microtubules (Du et al., 2001; Huffaker et al., 1988; Jacobs et al., 1988) (Fig. 1).
In vivo time-lapse microscopy of organelle dynamics has revealed that multi-copy organelles such as peroxisomes are transferred from mother cell to bud in a highly ordered vectorial process (Fagarasanu et al., 2006; Hoepfner et al., 2001; Rossanese et al., 2001). Moreover, Myo2p moves different cargoes at distinct times in the cell cycle (Eves et al., 2012); for example, movement of peroxisomes always precedes movement of mitochondria (Knoblach et al., 2013). How does the cell coordinate organelle movement so as not to impair organelle function or impede cell cycle progression? Stringent control mechanisms have been identified at the level of both the myosin V motor and the organelle adapters.
Our understanding of how Myo2p associates with cargo was advanced significantly by the determination of the crystal structure of the Myo2p tail (Pashkova et al., 2006). Mutagenesis of conserved surface amino acids in the tail identified patches required for its binding specifically to vacuoles, mitochondria, secretory vesicles and peroxisomes (Altmann et al., 2008; Catlett et al., 2000; Catlett and Weisman, 1998; Fagarasanu et al., 2009; Pashkova et al., 2006). Confirmation that these patches represent bona fide binding sites for organelles was obtained by demonstrating that they are the sites at which the corresponding organelle-specific adaptors bind to Myo2p (Fagarasanu et al., 2009; Ishikawa et al., 2003; Lipatova et al., 2008). Later work showed that binding of eight of the nine known Myo2p cargo adaptors overlaps at one of two distinct Myo2p surface patches (Eves et al., 2012). This finding has challenged the concept that Myo2p acts as a scaffold that simultaneously exposes sites for attachment to its different cargoes; rather it suggests that different types of organelles compete for access to Myo2p by steric exclusion of other organelles and that the availability of Myo2p adaptors on the surface of organelles has to be tightly regulated to dictate the precise timing of Myo2p recruitment and organelle motility (Fagarasanu et al., 2009).
Inp2p and Vac17p are extensively studied Myo2p adaptors that form transport complexes with peroxisomes and vacuoles, respectively. Both proteins have structural and regulatory elements in common. Coiled-coil domains are in part overlapping with the Myo2p-binding domains of Vac17p and Inp2p, whereas conserved phosphorylation sites and a PEST motif identify the proteins as targets for post-translational modification and rapid proteolytic turnover (Fig. 2). Inp2p and Vac17p fluctuate in level with the cell cycle; they reach maximum levels during peroxisome and vacuole inheritance and are degraded after their organelles have been delivered to the bud (Fagarasanu et al., 2006; Peng and Weisman, 2008). Inp2p and Vac17p each contain several recognition sites for the cyclin-dependent kinase Cdk1p and are phosphorylated during the cell cycle (Fagarasanu et al., 2009; Peng and Weisman, 2008). Direct phosphorylation of Vac17p by Cdk1p has been implicated in the increased affinity of Vac17p for Myo2p (Peng and Weisman, 2008). The p21-activated kinases (PAKs) Cla4p and Ste20p have been implicated in terminating vacuole inheritance (Bartholomew and Hardy, 2009), but whether phosphorylation of Vac17p is required for its degradation and whether Vac17p is a direct target of PAKs remained unclear. A recent study has shown that different stages of vacuole inheritance are controlled by distinct Vac17p phosphorylation events. Phosphorylation of Vac17p at several Cdk1p sites leads to recruitment of Myo2p and the start of vacuole transport. By contrast, phosphorylation of Vac17p at a threonine residue located in its PEST motif leads to the recruitment of a ubiquitin ligase, proteasomal degradation of Vac17p, and termination of vacuole transport. It is noteworthy that the ubiquitin ligase is recruited to the transport complex only after the vacuole has entered the bud (Yau et al., 2014).
The signaling cascade described above exemplifies how the cell integrates spatiotemporal cues to finely tune organelle movement. Analysis of Myo2p mutants defective in binding Inp2p, and thus impaired in delivering peroxisomes to the bud, has provided another example of how class-V-myosin-driven organelle partitioning is regulated. Peroxisome inheritance is effectively uncoupled from the cell cycle in these mutants; the levels of Inp2p are increased, implying that the amount of Inp2p is not controlled by the cell cycle but rather by the distribution of peroxisomes in the cell, that is, by organelle-positioning-specific cues. The timing of Inp2p phosphorylation, by contrast, remains unaffected, and thus appears to be coupled to cell cycle progression. The increase in Inp2p abundance furthermore hints at the existence of a compensatory mechanism as the cell supplies more of the peroxisome-specific transport adaptor when peroxisome delivery to the bud is compromised (Fagarasanu et al., 2009). Indeed, organelle delivery between mother and daughter cells is apparently intricately choreographed.
In the next section we discuss mechanisms for organelle retention, which is needed for the equitable distribution of organelles between mother and daughter cells at cytokinesis.
Holding on – organelle retention processes
Budding S. cerevisiae cells retain a portion of their organelles in the mother cell, that is, they never lose the entire population of one organelle to the bud. Organelle retention is accomplished by immobilizing organelles at distinct cell structures. Some organelles, such as mitochondria and peroxisomes, are anchored at the cell periphery (Fagarasanu et al., 2005; Klecker et al., 2013; Kornmann et al., 2009; Lackner et al., 2013; Yaffe, 1999), whereas others, specifically lipid droplets, cluster around the nuclear envelope (Adeyo et al., 2011; Jacquier et al., 2011; Knoblach and Rachubinski, 2014; Wolinski et al., 2011). Organelle retention is not confined to the mother cell but also occurs in the bud to prevent the backtracking of newly inherited organelles to the mother cell (Boldogh et al., 2004; Fehrenbacher et al., 2004). Processes for organelle retention have long been proposed (Fagarasanu et al., 2005; Yang et al., 1999), yet only lately have their molecular mechanisms been elucidated. Organelles are retained in the cell by the formation of membrane contact sites, or tethers, with other organelles. Membrane contact sites not only guarantee that an organelle population is maintained over generations of cells but also define the architecture and spatial distribution of organelles and facilitate inter-organelle communication. Examples of tethers that might not affect organelle segregation directly include the mitochondrial contact site and cristae organizing system (MICOS) located in the inner mitochondrial membrane (for a review, see Pfanner et al., 2014), membrane contacts between the vacuole and mitochondria that serve to mediate lipid exchange (Elbaz-Alon et al., 2014; Hönscher et al., 2014), and membrane contacts between the ER and mitochondria involved in lipid synthesis, Ca2+ signaling and mitochondrial remodeling (for a review, see Rowland and Voeltz, 2012). We will discuss organelle tethers as they pertain to organelle dynamics and inheritance; their roles in signaling, lipid trafficking and metabolism have been reviewed elsewhere (Klecker et al., 2014; Prinz, 2014; Stefan et al., 2013).
Organelle tethers assemble at discrete foci where membranes from two organelles come into close contact but do not fuse. Specific proteins enriched at the membrane contact sites, either alone or in complex, form the tether by binding to both membranes. Tethering proteins are either integral to the membrane of one of the compartments or can be classified as membrane-associated ‘hinge’ proteins that link the integral membrane proteins to each other. Tethers that connect peroxisomes (Knoblach et al., 2013) and mitochondria (Kornmann et al., 2009) to the cER, as well as those that connect mitochondria (Klecker et al., 2013; Lackner et al., 2013) and the cER to the plasma membrane (Manford et al., 2012), have recently been resolved. As the molecular architectures of tethers have become better defined, features common to all tethers have begun to emerge. First, tethers are often made between the ER and a second organelle. Second, some organelles are maintained by more than one tether or are found to tether to more than one organelle. Third, tethers do not simply anchor organelles but actively regulate organelle dynamics.
We recently described a tether that connects peroxisomes to the cER in yeast. The tether is comprised of two proteins, Inp1p and the peroxisome biogenic protein Pex3p (Knoblach et al., 2013). Cells lacking Inp1p contain only mobile peroxisomes that are eventually driven to the bud by the Inp2p–Myo2p transport complex, whereas cells overproducing Inp1p maintain all of their peroxisomes in fixed cortical positions in the mother cell and do not transfer any peroxisomes to the daughter cell (Fagarasanu et al., 2005). Inp1p not only decorates peroxisomes but also has an affinity for a structure lining the cell cortex. Thus, Inp1p was concluded to be required for attaching peroxisomes to a then uncharacterized cortical anchor (Fagarasanu et al., 2005). The identification of mutants of Pex3p that are unable to recruit Inp1p to the peroxisomal membrane and exhibit the same phenotype as that of inp1Δ cells (Munck et al., 2009) indicated that there is a direct interaction between Pex3p and Inp1p. Pex3p is essential for peroxisome biogenesis and is integral to both the ER and the peroxisomal membranes (Hoepfner et al., 2005; Thoms et al., 2012). We showed that Inp1p contains at least two binding sites for Pex3p and acts as a molecular hinge by bridging ER-bound Pex3p and peroxisomal Pex3p into an ER–peroxisome tethering complex (Knoblach et al., 2013) (Fig. 3).
Mitochondrial distribution and partitioning are regulated by several pathways. The ER–mitochondria encounter structure (ERMES) is one of at least two mitochondrial tethers acting at the cortex in the mother cell. ERMES links mitochondria to the cER and is formed by a complex made up of both ER and mitochondrial proteins. Maintenance of mitochondrial morphology protein 1 (Mmm1p), formerly believed to be a bona fide mitochondrial protein, is actually an ER-resident integral membrane protein (Kornmann et al., 2009) that bridges mitochondria to the ER by binding to a protein complex in the mitochondrial outer membrane (Fig. 3). Members of the protein complex share a synaptotagmin-like mitochondrial lipid-binding protein domain that is important for their localization at the ER-mitochondrion junction (Toulmay and Prinz, 2012). Cells lacking ERMES contain enlarged spherical mitochondria, frequently lack mitochondria in the bud and exhibit growth deficits on non-fermentable carbon sources. This phenotype can be rescued by the expression of an artificial mitochondrion–ER tether (Kornmann et al., 2009).
Mitochondrion–plasma-membrane contacts are mediated by a complex containing nuclear migration protein 1 (Num1p), which binds to the plasma membrane through its C-terminal pleckstrin homology domain, and mitochondrial distribution and morphology protein 36 (Mdm36p), a peripheral membrane protein of mitochondria that is thought to connect Num1p to a yet unidentified receptor in the outer mitochondrial membrane (Klecker et al., 2013; Lackner et al., 2013) (Fig. 3). Num1p provides a cortical attachment for microtubules during positioning of the mitotic spindle (Heil-Chapdelaine et al., 2000), but surprisingly num1Δ mutants also show defects in mitochondrial distribution (Cerveny et al., 2007). Mitochondria are normally anchored to the cell cortex, but this peripheral localization is lost in cells lacking Num1p. The Num1p tether promotes mitochondrial retention in the mother cell, as deletion of the NUM1 gene leads to the formation of mother cells without mitochondria (Cerveny et al., 2007; Klecker et al., 2013; Lackner et al., 2013). The mitochondrial distribution and morphology defects observed in num1Δ cells can be corrected by expression of an artificial mitochondrion–plasma-membrane tether, which demonstrates the importance of Num1p in the cortical anchoring of mitochondria (Klecker et al., 2013; Lackner et al., 2013).
Anchorage of mitochondria in buds requires two factors, Mmr1p and Ypt11p (Frederick et al., 2008; Itoh et al., 2002; Itoh et al., 2004). Mmr1p, a member of the DSL1 family of tethering proteins, has been reported to link mitochondria with the cER (Swayne et al., 2011). Super-resolution microscopy has revealed that Mmr1p enriches at punctate structures on opposing surfaces of mitochondria and ER sheets underlying the bud tip. Deletion of Mmr1p does not affect the normal dynamics of mitochondria or the distribution of the cER but impairs anchoring of mitochondria at the bud tips (Swayne et al., 2011). Of note, although the cER and mitochondria are moved to the bud by two independent motors (Estrada et al., 2003), a defect in ER inheritance has also been shown to impair mitochondrial inheritance. Ypt11p is a Rab-like protein that localizes to the cER (Buvelot Frei et al., 2006) and also interacts with Myo2p (Arai et al., 2008; Itoh et al., 2004). Deletion of YPT11 not only prevents the accumulation of cER at the bud tip (Buvelot Frei et al., 2006) but also inhibits the anchoring of mitochondria at the bud tip (Itoh et al., 2002; Swayne et al., 2011).
Some proteins involved in anchoring and partitioning mitochondria have been ascribed multiple and sometimes contradictory functions (Westermann, 2014) (Box 1). Mmr1p and Num1p act antagonistically, as the distribution of mitochondria is shifted to the mother cell in mmr1Δ cells (Itoh et al., 2004; Swayne et al., 2011) but to the bud in num1Δ cells (Klecker et al., 2013; Lackner et al., 2013). Not surprisingly, deletion of both MMR1 and NUM1 alleviates the phenotype observed in cells deleted for one of the genes (Hoppins et al., 2011). In a num1Δ mmr1Δ double mutant, equidistribution of mitochondria between the mother cell and the bud is partially restored (Klecker et al., 2013) because mitochondria can move freely in both directions without the constraint of cortical anchors (Klecker et al., 2014). Mmr1p and Ypt11p, by contrast, have similar functions, as overexpression of YPT11 compensates for loss of MMR1 (Swayne et al., 2011). Inheritance of mitochondria by the bud is blocked in a mmr1Δ ypt11Δ double mutant, which additionally suggests that these proteins have redundant functions (Chernyakov et al., 2013). Notably, the defect in mitochondrial inheritance can be rescued by expression of the chimeric mitochondrion-specific motor Myo2p–Fis1p, which carries a mitochondrial outer membrane anchor in place of the cargo-binding domain of Myo2p (Förtsch et al., 2011).
The junction between the ER and the plasma membrane is comprised of six ER-resident proteins, which all need to be eliminated to reduce the normally extensive interactions between the ER and the plasma membrane (Manford et al., 2012). This suggests that these six proteins mediate ER–plasma-membrane tethering independently of one another. Three conserved protein families are involved in contact formation: the synaptotagmin-like tricalbins, VAP family members and Ist2p (Toulmay and Prinz, 2012; Wolf et al., 2012). Four of the six proteins are integral to the ER membrane and have conserved cytosolic lipid-binding domains (Fig. 3). ER–plasma-membrane tethering is thus likely mediated through a protein–lipid, rather than protein–protein, interaction (Stefan et al., 2013).
Loss of ER–plasma-membrane tethering results in a separation of the cER from the plasma membrane, repositioning of the cER to the cell interior, elevated levels of phosphoinositides at the plasma membrane and a constitutive activation of the unfolded protein response (Manford et al., 2012). How this collapse of the cER affects the intracellular distribution of other organelles, such as peroxisomes and mitochondria, that normally attach to it has not been investigated.
Organelle division – integrating transport and retention
Organelle transport and retention have long been considered to be mutually exclusive and antagonistic processes that need to play a ‘tug of war’ to achieve an equidistribution of organelles between mother cell and bud (Fagarasanu et al., 2007). This notion is derived in part from the observation that transport and retention factors have inverse concentration gradients across the cell division axis and thus appear to be preferentially associated with a subset of organelles. Inp2p can be detected only on bud-localized peroxisomes in wild-type cells (Fagarasanu et al., 2006), whereas Inp1p is confined to mother cell peroxisomes (Knoblach et al., 2013). These observations do not explain how individual peroxisomes are chosen to acquire one inheritance factor over the other, how the cell rebalances its peroxisome population after the Inp1p- and Inp2p-containing peroxisomes have met different segregation fates, and how the cell ‘counts’ its peroxisomes so as not to lose control of its peroxisome population as a whole (Knoblach and Rachubinski, 2013).
Tracking the movement of individual peroxisomes by photoconversion of a fluorescently tagged peroxisomal matrix protein has revealed that peroxisomes in the mother cell are not fully released from their cortical tethers, whereas peroxisomes that travel to the bud contain photoconverted reporter and must therefore be the product of a division event in the mother cell (Knoblach et al., 2013). Mutants defective in peroxisome division contain a single giant peroxisome (Hoepfner et al., 2001; Kuravi et al., 2006). Inp1p and Inp2p are present on this enlarged peroxisome but are polarized to opposite ends. Inp2p is enriched on the part of the peroxisome that protrudes into the bud, whereas Inp1p is found on the part that is retained in the mother cell (Fig. 4). The concentration gradient of Inp1p can be explained by the geometry of the ER–peroxisome tether. Inp1p attaches to the ER membrane before it connects with the peroxisomal surface; it is therefore confined to the part of the peroxisome that is in direct contact with the ER (Knoblach et al., 2013). The Inp2p concentration gradient is likely established by Myo2p, which attaches to the peroxisomal membrane through Inp2p and exerts an anterograde pulling force, thereby concentrating Inp2p at the tip of the peroxisomal tubule. Of note, the giant peroxisome is split apart late in the cell cycle as it is overextended by Inp2p–Myo2p-dependent pulling forces on one side and by Pex3p–Inp1p-dependent retention forces on the other side. Thus, the mother cell keeps a peroxisome fragment containing Inp1p, whereas the daughter cell receives a peroxisome fragment containing Inp2p (Fig. 4). A gradient of inheritance factors similar to the one observed in wild-type cells is thus established in the peroxisome divisional mutants. We contend that every peroxisome replicates and splits between mother cell and bud, thus ensuring that the mother cell keeps, and the daughter cell receives, their fair and equitable allotment of peroxisomes (Knoblach and Rachubinski, 2013).
If organelle division is obligatory for organelle inheritance, how are these processes coordinated? Peroxisomes and mitochondria are known to share components of their divisional machineries (Delille et al., 2009). New evidence suggests that peroxisomes and mitochondria also have a common mechanism of distributing themselves between mother cell and bud, which is controlled by their tethering to the ER (Fig. 4).
Inp1p, the hinge protein that anchors peroxisomes to the cER by forming a bridge between ER-bound Pex3p and peroxisomal Pex3p, has been ascribed a second function in controlling peroxisome size and number. Cells lacking Inp1p have fewer and larger peroxisomes than wild-type cells, and cells overexpressing Inp1p contain numerous small peroxisomes. These findings suggest that there is a role for Inp1p in peroxisome division (Fagarasanu et al., 2005). Inp1p physically interacts with Pex25p, Pex30p and Vps1p (Fagarasanu et al., 2005), proteins that modulate the elongation and scission steps of peroxisome division (Schrader et al., 2012). Pex25p is a member of the Pex11 family of proteins that cause tubulation of peroxisomes through membrane remodeling (Koch et al., 2010), and it has also been implicated in the de novo formation of peroxisomes at the ER (Agrawal and Subramani, 2013). Pex30p, a protein localized at the ER–peroxisome interface, regulates peroxisome size and number (David et al., 2013; Yan et al., 2008). The dynamin-related GTPase Vps1p carries out the final scission step of peroxisome division, as in its absence only large peroxisomes with a beads-on-a-string morphology are observed (Hoepfner et al., 2001). How Inp1p coordinates its dual activities in peroxisome immobilization and division is unknown, but whatever its mode of regulation, Inp1p ensures that peroxisomes are not only ‘holding on’ in the mother cell but also ‘shipping out’ to the bud.
Mitochondria form a highly dynamic tubular network, which is required for their faithful partitioning and is maintained by opposing fusion and fission events. Mutants lacking factors that are essential for mitochondrial fusion, such as Fzo1p, are unable to inherit mitochondrial DNA (Hermann et al., 1998). This defect can be rescued in part by deleting genes that are required for mitochondrial fission (Sesaki and Jensen, 1999). The dynamin-related GTPase Dnm1p is a core component of the mitochondrial divisional machinery (Bleazard et al., 1999; Mozdy et al., 2000). Dnm1p assembles into spirals on the outside of the mitochondrion and uses GTP hydrolysis to sever the mitochondrial membranes (reviewed by Friedman and Nunnari, 2014). In-vitro-reconstituted Dnm1p rings have a diameter of ∼100 nm, which is too small to encircle mitochondria, as they have a diameter of ∼300 nm (Ingerman et al., 2005). How mitochondrial fission sites are selected and whether external forces contribute to the constriction of mitochondria were unresolved questions; however, ER tubules have recently been shown to encircle mitochondria at the sites at which future division would occur (Friedman et al., 2011). The ER is connected to mitochondria at sites of fission by the ERMES. Remarkably, the ERMES subunit Mmm1p has been found in small, punctate structures close to a subset of matrix-localized mitochondrial DNA nucleoids (Hobbs et al., 2001), suggesting that at least some nucleoids are connected to the ER by a protein complex that spans both mitochondrial membranes. The nucleoids colocalizing with Mmm1p are actively replicating (Meeusen and Nunnari, 2003) and display oscillatory movements, which likely contributes to their effective segregation (Murley et al., 2013). Moreover, the ERMES remains associated with one of the two mitochondrial fragments after fission has occurred, thereby liberating the other fragment (Murley et al., 2013) (Fig. 4). It is notable that a fraction of the Dnm1p molecules forms complexes with Num1p, and num1Δ dnm1Δ double mutant cells often lack mitochondria in their mother cells and accumulate all mitochondria in their buds (Cerveny et al., 2007). Mitochondria thus seem to recapitulate a segregation strategy in which one portion of the organelle remains tethered at the cortex of the mother cell, while another portion is split off for delivery to the bud.
Recently, Myo2p has been shown to play a prominent role in organelle division in addition to its function in organelle transport. Mutants in which Myo2p can no longer attach to the peroxisomal surface do not exhibit bud-directed polarization of peroxisomes (Motley and Hettema, 2007), and peroxisomes remain associated with the same cell over many generations because they are not released from their cortical tethers (Knoblach et al., 2013). These peroxisomes then also accumulate Inp2p and cannot be severed by Myo2p and transported to the bud (Fagarasanu et al., 2009; Knoblach et al., 2013).
In the next section we will provide examples of how a replication-based system of organelle partitioning contains an intrinsic quality control mechanism that affords cells tight control over their organelle populations. The implications of this process on cell aging and rejuvenation are discussed.
Functional consequences – inequality in organelle segregation
Even when organelles divide in an apparently symmetrical manner, they can contribute to the asymmetric segregation of their components if the fragments kept by the mother cell and inherited by the daughter cell are not functionally equivalent (Fig. 5) (Ouellet and Barral, 2012). The nucleus, whose envelope remains intact during mitosis in yeast, is an extensively studied example of an asymmetrically segregated organelle.
Extra-chromosomal DNA (rDNA) circles, which pop out of the chromosome as a result of homologous recombination between adjacent copies of rDNA repeats, are thought to contribute to replicative aging by titrating cellular components as they accumulate (Sinclair et al., 1998). Consistent with this hypothesis are observations that the expression of high-copy number plasmids reduces replicative lifespan (Falcón and Aris, 2003), whereas lowering the frequency of recombination at the rDNA locus leads to lifespan extension (McCormick et al., 2014). rDNA circles are segregated unevenly between mother and daughter cells, with a retention frequency of >98% by the mother cell nucleus (Sinclair and Guarente, 1997). A diffusion barrier consisting of septins and sphingolipids in the envelope of anaphase nuclei controls the segregation of nuclear constituents (Boettcher et al., 2012; Clay et al., 2014; Shcheprova et al., 2008), but its mechanism is still controversial. One study suggests that the diffusion barrier prevents the passage of old nuclear pore complexes (NPCs) to the daughter cell nucleus and that the binding of rDNA circles to NPCs might be responsible for their retention (Shcheprova et al., 2008). Other studies report that existing NPCs can be inherited by the daughter cell nucleus (Khmelinskii et al., 2010), although NPCs that are damaged or lack essential components are retained in the mother cell (Colombi et al., 2013; Makio et al., 2013). It is noteworthy that mutations affecting the strength of the diffusion barrier in the nuclear envelope also affect the frequency of plasmid retention by the mother cell nucleus (Gehlen et al., 2011; Lindstrom et al., 2011; Shcheprova et al., 2008). Visualization of episomes confirms that they are retained in the mother cell nucleus but diffuse freely throughout the nucleoplasm (Gehlen et al., 2011). The geometry of the nucleus, as well as the short yeast division cycle, might thus contribute to the uneven segregation of rDNA circles (Ouellet and Barral, 2012).
The ER is an endomembrane system comprised of sheets and tubules that is essential for cell viability. Because of its complex geometry, it is difficult to track the inheritance of individual ER components. It has been observed that yeast cells respond to acute stress by completely inhibiting ER inheritance (Babour et al., 2010), thus producing daughter cells that are inviable. This extreme case of asymmetric organelle division ensures that only undamaged organelle material is segregated to the daughter cell, thereby avoiding imposing an immediate functional disadvantage on her. However, the ER fragments retained by the mother cell and inherited by the daughter cell are non-equivalent even under normal cell division. A diffusion barrier similar to the one found in the outer membrane of the nuclear envelope polarizes the ER into separate mother and bud domains (Chao et al., 2014; Clay et al., 2014; Luedeke et al., 2005). A septin ring at the bud neck enables the recruitment of polarization factors and sphingolipids to this site. The concerted action of these factors prevents the free diffusion of transmembrane, and potentially luminal, ER proteins between mother cell and bud (reviewed by Higuchi-Sanabria et al., 2014).
Peroxisomes and mitochondria are multi-copy organelles that regulate the size of their networks through proliferation and fission events (Rafelski et al., 2012; Schrader et al., 2011). The mitochondrial reticulum is maintained by continuous exchange of materials between individual mitochondria, and mutations in the mitochondrial fusion and fission machineries affect yeast lifespan (Scheckhuber et al., 2007; Scheckhuber et al., 2011). Most mitochondrial and peroxisomal proteins appear to be homogenously partitioned between mother and daughter cells (Menendez-Benito et al., 2013), but new findings indicate that some proteins preferentially associate with mother-cell- or bud-localized peroxisomes, respectively (Fig. 4).
Mitochondria play a central role in cellular aging and are considered to be both targets of the aging process and contributors to it (Hughes and Gottschling, 2012). Mitochondria are indirectly influenced by vacuoles and are non-equivalent in older and younger cells. Vacuoles in buds have lower pH than those in mother cells, which is due to mother cell retention of the plasma membrane ATPase Pma1p, which pumps protons from the cytosol, thus limiting the number of protons available to acidify the vacuole (Henderson et al., 2014). As vacuolar acidity declines in aging mother cells, neutral amino acids accumulate and become sequestered by mitochondria, causing mitochondrial fragmentation and loss of function (Hughes and Gottschling, 2012). Moreover, mitochondria are partitioned asymmetrically between mother and daughter cells. Only ‘healthy’ mitochondria, which contain less reactive oxygen species and are more reducing than mother cell mitochondria, are inherited by the bud (Higuchi et al., 2013; McFaline-Figueroa et al., 2011; Vevea et al., 2013). Actin dynamics and anchorage of mitochondria at the bud tip ensure that mitochondria are segregated according to their functional state (reviewed by Higuchi-Sanabria et al., 2014; Vevea et al., 2014). Actin cables undergo a continuous retrograde flow from bud to mother cell, which is caused by actin nucleation at the bud tip and pulling forces exerted by the type II myosin Myo1p at the bud neck (Huckaba et al., 2006). Consequently, mitochondria have to swim ‘upstream’ for their inheritance. Increasing the rate of retrograde actin cable flow (RACF) has been shown to increase mitochondrial fitness in buds, whereas decreasing RACF has the opposite effect (Higuchi et al., 2013). This suggests that RACF acts as a filter to prevent the inheritance of less-fit mitochondria. In addition, once mitochondria have entered the bud, they need to be arrested at the cER of the bud tip to maintain their asymmetric distribution. In mmr1Δ and ypt11Δ mutants, healthy mitochondria are not effectively segregated from less-healthy mitochondria, which leads to breakdown in the mother–daughter age asymmetry of cells (McFaline-Figueroa et al., 2011; Rafelski et al., 2012).
Replicative lifespan is defined as the number of times a cell can divide before it enters senescence. Conceptually, a bud is born ‘young’, that is, it has a full replicative lifespan even if it comes from an old mother cell. Mother–daughter age asymmetry is a consequence of asymmetric cell division. Protein aggregates containing damaged or unfolded proteins are aging determinants that are retained by the mother cell through their sequestration in quality control compartments (Kaganovich et al., 2008; Spokoini et al., 2012), but are also selectively removed from daughter cells (Hill et al., 2014; Liu et al., 2010; Song et al., 2014). Rejuvenation determinants, such as cytosolic catalase that protects cells from oxidative damage, have a higher activity in the bud than in the mother cell (Erjavec and Nyström, 2007). A replication-based means of organelle partitioning evidently contributes to, and might be a prerequisite for, the establishment of a mother–daughter age asymmetry in yeast.
Why should organelle inheritance require the multiple levels of control described above? After all, each daughter cell need only receive some material that can be used to seed the expansion of an organelle. However, we have come far from the view that organelle inheritance is a random event and that mother and daughter cells retain and acquire their organelle populations simply by chance. Recent discoveries have led not only to the identification of many factors involved in organelle inheritance but have also afforded us a much better understanding of how these factors cooperate to ensure that organelles are equitably, though not evenly, shared between mother cell and daughter cell. As our journey continues, we expect even more exciting and surprising discoveries about how organelle inheritance enables cells to maintain the advantages afforded by compartmentalization of function.
ERMES and Num1p provide attachment of mitochondria to the cER and the plasma membrane of mother cells (blue spheres). These tethers are non-equivalent, as only the loss of Num1p leads to defects in mitochondrial retention (Cerveny et al., 2007; Klecker et al., 2013; Lackner et al., 2013). Two models have been proposed for mitochondrion–plasma-membrane tethering. In the first model, Num1p links mitochondria directly to the plasma membrane (Klecker et al., 2013), whereas the second model suggests a participation of the cER in the formation of a mitochondrion–ER–plasma-membrane tether (Lackner et al., 2013). It is thus currently unknown how many membranes participate in Num1p-mediated mitochondrial tethering in the mother cell (‘?’ in figure). Active segregation of mitochondria and their tethering at the bud tip depend on Mmr1p and Ypt11p (Chernyakov et al., 2013). Both proteins have been ascribed multiple functions. Mmr1p has been identified as the mitochondrion-specific transport adaptor for Myo2p (Itoh et al., 2004) and as a tether that connects mitochondria to the cER at the bud tip (Swayne et al., 2011) (brown sphere). Bud-specific expression of the MMR1 transcript (Shepard et al., 2003) makes it questionable whether Mmr1p can act in directed movement of mitochondria from mother cell to bud. Conversely, if the primary function of Mmr1p is to act as a tether, it would be worthwhile to identify its binding partner(s) at the ER membrane (‘?’ in figure). Ypt11p has been identified as an adaptor protein that connects late Golgi cisternae to Myo2p through its interaction with the coatomer subunit Ret2p (Arai et al., 2008), as an adaptor for Myo2p on mitochondria (Itoh et al., 2002), as required for the retention of newly inherited mitochondria at the bud tip (Boldogh et al., 2004) and as a docking protein for cER at the bud tip (Buvelot Frei et al., 2006). Ypt11p is a Rab GTPase that is regulated by phosphorylation and degradation (Lewandowska et al., 2013).
This work was supported by the Canadian Institutes of Health Research [grant number 9208 to R.A.R.].
The authors declare no competing or financial interests.