Membrane and organelle rearrangement during ascospore formation in budding yeast

SUMMARY

In ascomycete fungi, sexual spores, termed ascospores, are formed after meiosis. Ascospore formation is an unusual cell division in which daughter cells are created within the cytoplasm of the mother cell by de novo generation of membranes that encapsulate each of the haploid chromosome sets created by meiosis. This review describes the molecular events underlying the creation, expansion, and closure of these membranes in the budding yeast, Saccharomyces cerevisiae. Recent advances in our understanding of the regulation of gene expression and the dynamic behavior of different membrane-bound organelles during this process are detailed. While less is known about ascospore formation in other systems, comparison to the distantly related fission yeast suggests that the molecular events will be broadly similar throughout the ascomycetes.

INTRODUCTION

In fungi, a common response to changing environmental conditions is the formation of spores: specialized, stress-resistant cells that are capable of dispersal in the environment (1). Despite these common properties, there are many different types of spores. Often, spores are derived directly from mitotic cells, such as conidiospores in hyphal fungi (2). Alternatively, sexual spores, such as basidiospores or ascospores, are formed after mating and meiosis (1). These sexual spores are the fungal equivalent of gametes in metazoans.

The formation of ascospores occurs by the process of sporulation and is the defining feature of ascomycete fungi. Ascospore formation is linked to the meiotic nuclear divisions and occurs through a form of cell division in which a single cell gives rise to multiple daughter cells by envelopment of multiple individual newly made nuclei within newly formed plasma membranes, called prospore membranes, in budding yeast (3). Prospore membranes form de novo and are not connected to the existing plasma membrane of the cell. Thus, the mother cell remains intact during sporulation, eventually forming the ascus that surrounds the spores.

Most ascomycetes produce asci that contain either four spores (tetrads) or eight spores (octads). In the case of tetrads, the nuclei packaged into spores are the four daughter nuclei produced by meiosis. In octads, the meiotic products undergo one round of DNA replication and an additional mitotic division before being packaged into spores (4). Despite these changes in cell cycle regulation, cytological studies have revealed that sporulation is broadly similar in all ascomycetes with new intracellular double membranes appearing and ultimately engulfing nuclei (5–9).

Ascospore formation has been described at the ultrastructural level in many ascomycetes, but detailed molecular analyses have only been performed in Saccharomyces cerevisiae and Schizosaccharomyces pombe. Ascospore formation in these two yeasts requires extensive reorganization of the cytoskeleton, endomembrane system, and regulatory networks found in vegetative cells (3, 10–12). In this review, I focus on recent developments in our understanding of the timing of gene expression and membrane rearrangements required for ascospore formation in Saccharomyces cerevisiae, with a focus on how prospore membrane growth is promoted and the regulation of organellar inheritance into newly formed spores.

A BRIEF OVERVIEW OF SPORULATION IN S. CEREVISIAE

In S. cerevisiae diploids that are heterozygous for mating type, nitrogen starvation in the presence of a non-fermentable carbon source triggers exit from the mitotic cell cycle and entry into meiosis (13). This developmental decision is driven by the transcription factor Ime1 (14–16) (Fig. 1A). IME1 transcription is regulated by a variety of environmental and genetic factors. The Ime1 protein activates transcription of a set of meiosis-specific “early genes” required for pre-meiotic S phase as well as interhomolog recombination, chromosome pairing, and synapsis during prophase I (17–19). In addition to these functions, Ime1 triggers the expression of a second transcription factor, Ndt80, which is the central regulator driving prophase I exit, the meiotic divisions, and spore formation (20, 21) (Fig. 1A).

Fig 1.

Overview of ascospore formation. (A) A MATa/MATα diploid cell starved of nitrogen in the presence of a non-fermentable carbon source arrests in G1 of the cell cycle and then enters meiosis, driven by the sequential activity of the transcription factors Ime1 and Ndt80. The nuclear envelope is shown as a black circle, and the plasma membrane is a blue circle. (B) During meiosis II, MOP complexes (orange) are assembled on the cytoplasmic faces of the spindle pole bodies (curved lines). These MOPs are the initiation sites for the generation of prospore membranes (green). (C) The prospore membranes expand during meiosis II to engulf the adjacent nuclear lobes. (D) After exit from meiosis II, each prospore membrane encloses a daughter nucleus, creating four daughter cells surrounded by double lipid bilayers. (E) Spore walls are formed around each daughter cell within the lumen of the double bilayer, maturing the cells into ascospores. (F) After spore wall maturation, the remnant of the mother cell collapses, creating the ascus. This review focuses on events occurring as cells progress from panels B to D. ER, endoplasmic reticulum; MOP, meiosis II outer plaque.

Progression through meiosis results in four discrete haploid genomes in the cytoplasm of the mother cell. The packaging of each of these genomes into daughter cells begins during the second meiotic division (22, 23). One prospore membrane is formed at each of the four meiosis II spindle poles (22, 24, 25) (Fig. 1B). To generate prospore membranes, the cytoplasmic face of each spindle pole body (SPB) is converted from a site of microtubule nucleation to a membrane initiation site (24). This conversion requires the assembly of a complex on each SPB termed the meiosis II outer plaque (MOP) (24, 26). Post-Golgi secretory vesicles dock onto the surface of the MOPs and subsequently fuse to create prospore membranes (25, 27).

As meiosis II continues, haploid chromosome sets segregate into nuclear lobes, with an SPB at the apex of each lobe. Each prospore membrane expands and engulfs the nuclear lobe to which it is linked (Fig. 1C). At exit from meiosis II, the spindles disassemble, and each nuclear lobe pinches off to create four discrete haploid nuclei. At the same time, prospore membranes extend completely around the haploid nuclei. The ends of each membrane then fuse, enclosing each nucleus within a prospore (Fig. 1D). Prospore membrane closure is a cytokinetic event that separates the cytoplasm of the spore from the cytoplasm remaining in the mother cell (now referred to as the ascus) (28). Closure also separates the double membrane of the prospore membrane into two distinct bilayers. The bilayer closest to the nucleus becomes the spore plasma membrane, while the outer membrane disappears during the subsequent formation of the spore wall (29).

After closure, extended prospore membranes rapidly round up (30). The prospores created by membrane closure then mature into ascospores by assembly of spore walls (29, 31) (Fig. 1E). Spore wall materials are initially deposited into the luminal space between the spore plasma membrane and the outer membrane (31). The two inner layers of the spore wall, composed primarily of mannoproteins and β-1,3 linked glucan chains, are formed in this luminal space (29, 32, 33). The outer membrane then disappears, bringing the forming spore wall into direct contact with the ascal cytoplasm, and the outer layers of the spore wall are formed (29). These outer layers contain the β-1,4 linked glucosamine polymer chitosan, a polyaromatic composed of the di-amino acid dityrosine, and triglycerides (34–38). It is these outer layers that provide spores with increased resistance to a wide variety of external stresses (39–41).

Finally, after visible spore walls have formed, the original mother cell collapses around the spores to form the mature ascus (Fig. 1F). This process, termed sporoptosis, involves release of degradative enzymes from vacuoles that remain in the ascal cytoplasm after prospore membrane closure (42, 43). These enzymes degrade cellular components that are not segregated into the spores. This process occurs only after spore wall development so that spores are not affected by the release of vacuolar hydrolases, though the specific triggers of sporoptosis are unknown (44).

COORDINATING SPORE FORMATION

Control of gene expression in the NDT80 regulon

The central regulator of sporulation is the transcription factor Ndt80 (20, 21, 45). Activation of Ndt80 at the end of meiotic prophase I leads to induction of genes driving three interrelated functions: (i) cell cycle genes that encode, for example, cyclin proteins and the polo-like kinase Cdc5 that allow exit from prophase I and progression through the meiotic divisions; (ii) genes involved in formation and expansion of the prospore membrane; and (iii) genes involved in the post-meiotic events of spore wall assembly and maturation.

NDT80 transcription is itself induced early in the meiotic prophase, but the activity of the transcription factor is inhibited until the creation of crossover precursors by meiotic recombination and homolog synapsis is complete (46–50). Upon activation, Ndt80 induces expression of both itself and ~300 “middle” sporulation genes (20, 45). This auto-activating loop leads to rapid amplification of the signal and induction of the full suite of Ndt80 targets. However, this rapid induction actually poses a difficulty for the cell. Though transcription of the entire Ndt80 regulon is induced with similar kinetics, the protein products of these genes are required at different times during sporulation. Factors driving meiosis I are required immediately, those involved in meiosis II and prospore membrane growth not until meiosis I is completed, about 1.5 hours after Ndt80 activation, and those involved in spore wall assembly not until after meiosis II is completed, about 3 hours after Ndt80 activation.

The first indication that Ndt80-induced transcripts may be expressed as proteins at different times was provided by studies of cyclin genes during the meiotic divisions (51). Highly synchronous progression through meiosis can be achieved by placing the NDT80 gene under the control of an estradiol-inducible promoter (17). Using this system to clearly distinguish meiosis I from meiosis II cells, studies showed that the expression of different B-type cyclins differs during the meiotic divisions (51). Specifically, although CLB1 and CLB3 are transcriptionally induced at the same time by Ndt80, Clb1 protein is immediately produced. By contrast, translation of CLB3 is delayed so that the protein does not appear until meiosis II (51).

Ribosome profiling of mRNAs from sporulating cells revealed that translational delay of transcripts in the Ndt80 regulon is a widespread phenomenon (52). Examination of the association of individual mRNAs with ribosomes across highly synchronized meiotic time courses provides a measure of when individual transcripts are translated during meiosis and sporulation. This study demonstrated that, although transcripts of the Ndt80 regulon are all induced at the same time, their peak association with ribosomes (i.e., maximal translation) is divided broadly into three classes: class 1, mRNAs that are translated immediately upon induction, like CLB1; class 2, mRNAs whose translation is delayed until the meiosis I/meiosis II transition like CLB3; and class 3, mRNAs that are delayed until the end of meiosis II, such as SPS4 (52) (Fig. 2).

Fig 2.

Translation controls the timing of gene expression in the Ndt80 regulon. Upon exit from meiotic prophase, Ndt80 induces the expression of several hundred genes. The majority of these transcripts are translated immediately, such as CLB1 (class 1), while a subset is bound by the Rim4 protein and sequestered into phase-separated condensates. Late in meiosis I, increased activity of the Cdc5-Ime2-Hrr25 cascade leads to phosphorylation and degradation of Rim4, releasing these transcripts. Some of these transcripts, such as CLB3 (class 2), are then immediately translated in early meiosis II, while others associate with additional RNA-binding factors such as Pes4 and Mip6, leading to a further delay in their translation to the end of meiosis II, such as SPS4 (class 3).

While the majority of Ndt80-induced genes are class 1, a significant fraction fall into classes 2 and 3 (52). Though not a perfect correlation, in many cases, the delay in timing is consistent with the known function of the protein product. For instance, class 2 genes are involved in meiosis II-specific processes such as prospore membrane growth, while many of the class 3 genes function in the post-meiotic process of spore wall development (52, 53). Thus, translational regulation is a critical regulatory feature of gene expression during meiosis in yeast.

Ime2/Rim4 regulates translational timing during meiosis

Translational timing in meiosis is controlled by a pathway involving the meiosis-specific protein kinase Ime2 and the RNA-binding protein Rim4 (Fig. 2). IME2 was originally identified for its role early in meiosis. While IME2 itself is an early gene required for pre-meiotic S phase, it is also important for full Ime1 activity that drives the transcription of many early genes (15, 19). IME2 also indirectly plays a role in triggering transcription of the middle genes by enabling Ime1-mediated expression of NDT80 (54–56). The low level of Ndt80 protein is then inhibited by the meiotic recombination checkpoint until chromosomes are fully synapsed (46). Once the meiotic recombination checkpoint is satisfied, Ndt80 is activated and induces transcription of itself, as well as the genes in its regulon (Fig. 2). After Ndt80 induction, Ime2 kinase activity drops then rises again as cells progress through the meiotic divisions (17).

Carboxy-terminal truncations of Ime2 stabilize the protein and create a form in which kinase activity remains high throughout meiosis (57). In strains carrying this constitutively active Ime2, CLB3 mRNAs are now translated immediately (53) (Fig. 2). Constitutive activation of Ime2 relieves the translational delay of both CLB3 and other class 2 transcripts as well as class 3 transcripts (53, 58). Thus, reduced Ime2 activity is essential to establish translational repression in meiosis, and increased Ime2 activity promotes the translation of these mRNAs.

RIM4 encodes a protein with multiple RNA recognition motif (RRM) domains that is required for both transcription of IME1 and IME2 at the onset of sporulation and the translational delay of class 2 and 3 transcripts (53, 59) (Fig. 2). Because rim4∆ diploids fail to enter meiosis, Rim4’s later role in translational regulation was revealed using strains carrying rim4 alleles with point mutations in individual RRM domains. In these mutants, IME1 and IME2 are transcribed and cells enter meiosis, but CLB3 is translated prematurely, indicating that Rim4 RNA binding is required to delay the translation of CLB3 (53, 59). In fact, Rim4 can be co-precipitated with CLB3 mRNA, as well as several other class 2 and class 3 transcripts, suggesting that Rim4 directly inhibits the translation of all these transcripts (53).

Rim4 binds to different transcripts through its RRM domains (53). In addition, Rim4 can coalesce to form phase-separated condensates through low-complexity sequences in its C-terminal region (60) (Fig. 2). These condensates sequester mRNAs from the translation machinery and are important for Rim4 binding to transcripts (60–62). The increase in Ime2 activity as cells progress through meiosis triggers dissolution of the condensates and Rim4 degradation (53, 60). Though Ime2 can phosphorylate Rim4, the action of Ime2 on Rim4 is indirect. Ime2 acts as part of a kinase cascade (63). SPO13 encodes a sporulation-specific regulatory subunit of the polo-like kinase Cdc5 (64). While free Cdc5 activates Ime2, Spo13-Cdc5 inhibits Ime2 activity (63). Degradation of Spo13 at metaphase I frees Cdc5 to activate Ime2. In turn, Ime2 activates the casein kinase Hrr25, and active Hrr25 promotes Rim4 degradation (63). As a result of Rim4 turnover, class 2 transcripts are now translated (Fig. 2). In contrast, class 3 mRNA translation remains delayed by the action of different mRNA-binding proteins (Fig. 2).

Interestingly, Rim4 degradation is mediated largely by an autophagy pathway rather than the ubiquitin/proteosomal pathway (65). Therefore, Rim4 must release transcripts prior to its degradation, presumably when the condensate disassembles. Otherwise, the mRNAs would be engulfed and degraded, along with the Rim4 protein. The specific receptor that recruits Rim4 into autophagosomes has not been identified. Cdc14 phosphatase, which is activated at the end of meiosis I, broadly upregulates autophagy during meiosis (66). This increased autophagic activity may help reinforce the rapid destruction of Rim4.

Ime2-mediated inactivation of Rim4 explains how translation of class 2 messages is delayed until the meiosis I/meiosis II transition. However, class 3 Ndt80 targets have their translation delayed until the end of meiosis II (52) (Fig. 2). In addition to delayed translation, many class 3 transcripts share another feature, termed “protection”. Yeast cells pass a commitment point during meiosis I after which they will complete sporulation when returned to rich medium (67, 68). Nonetheless, reintroduction of nutrients to committed cells has a profound effect on the transcriptome, with the transcript levels of most Ndt80 target genes dropping immediately (69). Within this regulon, a group of ~20 protected transcripts maintain their levels even after reintroduction of nutrients (69). These protected transcripts are all class 3. Mutation of an RNA-binding domain in Rim4 or hyperactivation of Ime2 leads to both loss of protection and early translation of one of these transcripts, SPS4, demonstrating that delayed translation and protection from degradation are linked (58, 70).

PES4 and MIP6 encode related RNA-binding proteins that are class 1 Ndt80-induced transcripts (52). In a pes4∆ mip6∆ mutant, over half of the protected transcripts become sensitive to nutrients (i.e., are no longer protected) (71). Moreover, the translational delay of the SPS4 gene is lost, although the delay in CLB3 translation is not affected (71). Thus, Pes4 and Mip6 function specifically in control of class 3 mRNA translation. Meiotic commitment is defective in pes4∆ mutants, suggesting that protection of these transcripts is important for blocking the return to meiotic growth (72).

These results suggest the model that, after Rim4 degradation, class 3 transcripts associate with additional RNA-binding proteins, which further delays translation of these mRNAs (Fig. 2). For SPS4, association with Pes4/Mip6 is responsible for this delay. Pes4 and Mip6 regulate only a subset of the class 3 transcripts, indicating that there are additional regulators of late translation still to be identified (71). A global analysis of protein abundance during meiosis revealed that both Pes4 and Mip6 are degraded at the end of meiosis II (73), suggesting that turnover of Pes4/Mip6 is the basis for triggering translation of their target mRNAs. The general model that emerges from these studies is that Ime2/Rim4 is the essential backbone of translational regulation in meiosis II, while association with additional RNA-binding factors can modify the precise timing of translation for specific mRNAs (71).

HOW TO MAKE A NEW CELL

Initiation of a new membrane

During mitosis and meiosis I, the cytoplasmic face of each SPB contains the Spc72 gamma tubulin receptor associated with the gamma tubulin complex (74, 75). Spc72 and the gamma tubulin complex must be replaced by the MOP on the meiosis II SPBs for prospore membranes to form (Fig. 3A) (26, 76). The meiosis-specific proteins that make up the MOP (Mpc54, Spo74, Spo21/Mpc70, and Ady4) are derived from Ndt80-dependent class 1 transcripts that are translated without delay (26, 52, 77, 78). Although these proteins are present during meiosis I, their recruitment to meiosis I SPBs is prevented by Spc72 (76). Prior to meiosis II, each meiosis I SPB duplicates to form a “mother-daughter” pair (Fig. 3A). The newly formed daughter SPBs are not inhibited from MOP assembly because they lack Spc72 (Fig. 3B). However, for prospore membrane formation to occur at all four SPBs, Spc72 and the gamma-tubulin complex must first be removed from the mother SPBs by proteolysis (76).

Fig 3.

Initial steps of spore formation at SPBs. (A) Conversion of the Meiosis I SPB outer plaques to MOPs. In vegetative or meiosis I cells, the outer plaque of the SPB contains Spc72, which acts as a receptor for the gamma-tublin complex, allowing the formation of cytoplasmic microtubules. At the end of meiosis I, SPBs duplicate, creating two pairs of “mother” and “daughter” SPBs. (B) Assembly of the MOP at SPBs. Spo13/Cdc5 inhibits Spc72 proteolysis until after SPB duplication, at which time Spc72 is degraded from mother SPBs and cytoplasmic microtubules are lost. Hrr25 and cyclin-Cdk1 then promote recruitment and assembly of the MOP components to both types of SPB. (C) Prospore membrane initiation on the MOP. The N-terminal domain of Spo21 binds to membranes, tethering vesicles to MOPs. These vesicles then undergo MPC54-dependent docking onto the MOP surface, resulting in the vesicle fusion that forms prospore membranes. MOP, meiosis II outer plaque; NE, nuclear envelope; SPB, spindle pole body.

The Spo13-Cdc5 complex, which stabilizes Rim4, also stabilizes Spc72 at SPBs (63, 76). Anaphase-promoting complex/cyclosome (APC), an E3 ubiquitin ligase that regulates the stability of a large number of cell cycle-associated proteins (79–81), is activated at the end of meiosis I, causing inactivation of the cyclin-Cdk1 kinase as well as degradation of Spo13 (63). Loss of Spo13 in turn leads to degradation of Spc72 (63, 76) (Fig. 3B). Spc72 removal alone is not sufficient for MOP assembly, however. Cyclin-Cdk1 and Hrr25 are also necessary for MOP formation on both mother and daughter SPBs at meiosis II (63, 82) (Fig. 3B). Since cyclin-Cdk1 activity drops at the end of meiosis I and then rises again at metaphase of meiosis II, this requirement delays MOP assembly, and therefore prospore membrane formation, until meiosis II is under way (76). How the activity of Spo13-Cdc5 protects Spc72 against degradation and how phosphorylation by cyclin-Cdk1 and Hrr25 promotes MOP assembly are unknown.

MOP assembly is also under nutritional control. Cells reduce the number of spores formed in response to carbon limitation (83). This control is exercised at the level of MOP assembly, since only SPBs with MOPs can generate the prospore membranes necessary to encapsulate chromosomes within a spore (84, 85). Chromosomes that segregate to meiosis II SPBs lacking MOPs remain behind in the ascus, where they are degraded by sporoptosis during ascal maturation (42). When carbon is limiting, MOPs are preferentially formed on daughter SPBs at meiosis II (84, 85). How the number of MOPs is reduced and how daughter SPBs are specifically selected are unknown. Removal of Spc72 is required for MOP assembly. However, Spc72 is removed from the mother SPBs that lack MOPs during carbon limitation, and mutations in SPC72 have at most modest effects on SPB choice during sporulation in limiting carbon (84, 86, 87). Thus, this process is likely distinct from the Spo13-Cdc5 activity described above (84).

The vesicle fusion that initiates prospore membrane formation involves many of the same components required for vesicle fusion at plasma membranes: for example, the exocyst tethering complex, the Rab protein Sec4, and the Sso1 and Snc1/Snc2 soluble NEM-sensitive factor attachment protein receptors (SNAREs) (25, 88, 89). In addition to this core membrane fusion machinery, there are requirements specific to prospore membranes. For example, MOPs are meiosis-specific platforms on SPBs where vesicles fuse to initiate prospore membrane formation (26, 77). The phospholipase D Spo14 is present in both vegetative and sporulating cells but is only necessary for vesicle fusion at prospore membranes, not plasma membranes (27, 90).

Vesicle fusion at MOPs is proposed to occur in two steps: (i) tethering, where vesicles are loosely associated at a distance from a MOP, and (ii) docking, where vesicles are tightly connected to a MOP (91) (Fig. 3C). The N-terminal region of the Spo21 protein contains a prospore membrane-binding amphipathic helix (92). Alphafold predicts that this helix sits in an otherwise unstructured region of the protein, suggesting that this helix can extend into the cytoplasm to interact with vesicle membranes at a distance from a MOP (93, 94) (Fig. 3C). Point mutants in this helix show reduced sporulation and prospore membrane defects consistent with poor attachment of prospore membranes to MOPs (92). That vesicles do tether to MOPs prior to docking was revealed by electron microscopy studies using separation of function point mutants in MPC54. In cells carrying these mpc54 alleles, vesicles localize around intact MOPs but are not tightly juxtaposed to them (91). These tethered vesicles do not fuse, indicating that tethering is not sufficient for vesicle fusion.

SSO1 encodes a SNARE that directly catalyzes membrane fusion (95, 96). Deletion of SSO1 leads to accumulation of vesicles on MOPs (27). Moreover, in electron microscopy tomography reconstructions of wild-type cells, unfused vesicles are docked on to the MOP surface prior to prospore membrane formation (3). Together these results indicate that docking of vesicles to MOPs occurs prior to fusion and is a pre-requisite for fusion.

How might docking to a MOP allow the subsequent fusion of the vesicles? In the fusion process, Sec4 functions to recruit the exocyst, which in turn helps drive SNARE-mediated fusion (97, 98). Sec4 is a GTPase that binds to membranes only in its GTP-bound state, and conversion to the GTP-bound state is controlled by partner proteins that act as guanine nucleotide exchange factors (GEFs) (99–101). In mpc54 mutants that display only tethered vesicles, Sec4 is not present at MOPs; however, it is present in sso1∆ mutants that accumulate docked vesicles (91). Thus, docking to the MOP surface may promote fusion by triggering recruitment of Sec4. Interestingly, a MOP protein in the fission yeast Schizosaccharomyces pombe, Spo13 (unrelated to S. cerevisiae Spo13), acts as a GEF (102), though none of the S. cerevisiae MOP proteins has yet been found to have such an activity.

The lipid composition of prospore membrane precursor vesicles is important for docking and fusion. For example, ADY4 encodes a non-structural MOP component that shuttles on and off the SPB (103) (Fig. 3B). Ady4 binds to the phosphatidylinositol-4-phosphate 5-kinase Mss4 and recruits it to SPBs during meiosis (92). This recruitment is expected to generate a pool of phosphatidylinositol-4,5 bis-phosphate (PI4,5P2) in membranes near SPBs. ady4∆ mutants form prospore membranes, but these membranes frequently detach from MOPs, suggesting that PI4,5P2 helps maintain the connection of prospore membranes to MOPs, perhaps through binding to the amphipathic helix of Spo21 (92, 103). Sso1 also binds to PI4,5P2, indicating this lipid might play a role in promoting fusion of vesicles at MOPs as well (104).

A critical lipid involved in prospore membrane initiation is phosphatidic acid that is generated by Spo14 (105). Spo14 localizes to prospore membranes where it catalyzes hydrolysis of phosphatidylcholine to phosphatidic acid (90, 106). The catalytic activity of Spo14 is required to form spores, indicating that production of phosphatidic acid is essential for making prospore membranes (106). Furthermore, spo14∆ diploids show a very similar phenotype to sso1∆ with unfused vesicles decorating the MOP surface (27). Therefore, phosphatidic acid produced by SPO14 is required for vesicle fusion at MOPs after vesicle docking.

The need for phosphatidic acid is tied, at least in part, to the function of the meiosis-specific SNARE protein Spo20 that replaces the plasma membrane SNARE Sec9 as a partner for Sso1 (25, 107). Spo20 lacks a transmembrane helix but localizes to prospore membranes via an amphipathic helix in its N-terminal region (108). In isolation, this helix preferentially binds to phosphatidic acid-rich membranes in vivo (108). Additionally, fusion mediated by Sso1-Snc2-Spo20 SNARE complexes in vitro is stimulated by addition of phosphatidic acid to liposomes (109). Thus, Spo14 activity promotes Spo20-mediated fusion in vivo by both recruitment of Spo20 to prospore membranes and enhancement of Spo20’s activity. Phosphatidic acid generated by Spo14 must play additional roles, however, as spo14∆ mutants lack prospore membranes entirely, while spo20∆ strains display abnormally small prospore membranes (25, 90). The additional functions of phosphatidic acid in prospore membrane formation remain to be identified.

Expansion of the prospore membrane

Fusion events that initiate prospore membrane formation occur on the MOP surface. Once a membrane “cap” has formed, it rapidly extends beyond the MOP (Fig. 1B and 4). Prospore membrane expansion utilizes the fusion of vesicles delivered through the secretory pathway, as well as the transfer of lipids directly into the membrane from various sources. The requirement for vesicle transport is shown by the prospore membrane phenotypes of mutants defective in post-Golgi vesicle fusion or earlier steps in the secretory pathway (25, 90, 110). GIP1 encodes a sporulation-specific regulatory subunit of protein phosphatase 1 (33). Loss of GIP1 results in a block to sporulation with small prospore membranes (33, 111). As described in detail below, gip1∆ mutants are altered in vesicle formation at the endoplasmic reticulum (ER), providing further evidence for the central role of vesicular delivery in prospore membrane growth (112).

Fig 4.

Expansion of the prospore membrane. The shape of the prospore membrane as it grows is governed by two cytoskeletal elements, the leading-edge complex at the lip of the membrane (blue dots) and septins, which form sheets lining the interior surface of the membrane (red bars). Expansion is driven by the delivery of lipids by (i) fusion of secretory vesicles (green circles) with the prospore membrane and (ii) direct delivery of lipids from the ER or lipid droplet via Vps13 or Sfh3, respectively. It is unknown whether lipid delivery via either of these routes is biased to one side of the prospore membrane or is to both sides, as shown. Throughout expansion, the prospore membrane remains anchored to the nuclear envelope via a MOP.

However, vesicular transport is insufficient to supply all of the lipids required for prospore membrane expansion. Lipids are also delivered to prospore membranes from the ER via the highly conserved Vps13 protein (113). VPS13 is required for prospore membrane expansion and thus spore formation (114). Vps13 is a member of a protein superfamily termed “bridge-like transfer proteins” that act as channels for the bulk transfer of lipids between membranes of different intracellular organelles (115). The VPS13 gene is constitutively expressed, and the protein is involved in lipid transport at a variety of different membrane contact sites in vegetative cells (116–118). Localization of Vps13 is regulated by interaction with different organelle-specific adaptor proteins (119). These proteins are unrelated but compete for binding to Vps13 through a shared sequence motif, termed a proline-x-proline (PxP) motif (119).

Vps13 functions in prospore membrane growth with two partner proteins, Spo71 and Spo73. SPO71 encodes a prospore membrane adaptor for Vps13 (119, 120). Spo71 localizes to prospore membranes and is required for Vps13 localization at prospore membranes (120). Spo71 carries a PxP motif, and this small region of Spo71 is capable of driving Vps13 localization when expressed in vegetative cells (119). Deletion of SPO71 results in sporulation defects very similar to those of vps13∆ (114, 120). SPO73 encodes a second sporulation-specific protein that interacts with both Vps13 and Spo71 (121, 122). Like mutants in VPS13 or SPO71, spo73∆ cells display abnormally small prospore membranes, indicating a defect in membrane expansion (110, 114, 120–122). However, unlike spo71∆, Vps13 localization to the prospore membrane is normal in spo73∆ cells, suggesting that SPO73 plays a regulatory role in Vps13 activity.

Spo71 binds the C-terminal half of Vps13, while the extreme N-terminus of Vps13 is sufficient to bind to the ER (113, 119) (Fig. 4). This configuration suggests that Vps13 mediates ER-prospore membrane contacts. Indeed, three proteins found at ER-plasma membrane junctions in vegetative cells, Ist2, Tcb3, and Scs2, localize in patches adjacent to prospore membranes in meiotic cells (113). All three of these integral membrane proteins localize to the ER, so the meiotic localization pattern is consistent with the existence of ER-prospore membrane contact sites. Moreover, association of Ist2 with prospore membranes is altered in cells lacking SPO71 or VPS13 (113). Together, these observations suggest that Vps13 connects prospore membranes with the ER and that the Vps13-Spo71-Spo73 complex is necessary for stable ER-prospore membrane junctions. At these junctions, Vps13 could deliver lipids directly from their site of synthesis in the ER to prospore membranes (Fig. 4). Expansion of the prospore membrane in this respect is analogous to expansion of the phagophore membrane during autophagy, in which lipids are delivered from the ER to the growing phagophore via the Vps13-related Atg2 protein (123).

The lipid composition of the prospore membrane plays an important role in expansion as well as initiation. During expansion, the prospore membrane is rich in both phosphatidyl-inositol-4-phosphate (PI4P) and phosphatidic acid, but lacking in PI4,5P2 (108, 113, 124). The level of all three lipids is reduced in the smaller prospore membranes formed in vps13∆ mutants, perhaps due to reduced delivery of phosphatidylinositol (114). Raising phosphatidic acid levels by overexpression of Spo14 can restore prospore membrane size, though not sporulation, in a vps13∆ strain (114). Moreover, reducing PI4P levels via recruitment of the catalytic domain of the PI4P phosphatase Sac1 to the prospore membrane can partially suppress the sporulation defect of spo71 or spo73 mutants (113). While the mechanism of suppression by manipulation of lipids is unclear, together these results suggest that Vps13 activity may be modulated by the lipid composition of the prospore membrane.

Finally, lipid droplets also provide lipids to growing prospore membranes. Electron micrographs show extensive contacts between the outer bilayer of prospore membranes and lipid droplets (31, 125). Lipids are moved from the droplets to the membrane by Sfh3, a member of the Sec14-family of lipid transfer proteins that localizes specifically to lipid droplets (125). Mutant strains lacking lipid droplets exhibit prospore membrane defects including the appearance of intraluminal vesicles within the prospore membrane, a phenotype also seen in vps13∆ and spo71∆ mutants (114, 120, 126). However, the sporulation defects in cells lacking lipid droplets are not as severe as vps13∆, indicating that the ER is the major lipid source for prospore membrane expansion.

Guiding membrane growth

As prospore membranes expand, their growth is guided by two membrane-associated cytoskeletal structures (3). The first is the leading-edge complex that is located at the lip of the prospore membrane (Fig. 4) (127). The proteins that make up this complex, Ssp1, Ady3, Irc10, and Don1, are all Ndt80 targets and so are sporulation-specific in their expression (45, 127–129). Interdependent interactions between the leading-edge proteins are necessary to form the complex. For example, Ssp1 localization to the leading edge is similar to wild type in ADY3, IRC10, or DON1 single mutants, but is abnormal in ady3∆ irc10∆ diploids, and sporulation is blocked (127, 129). In addition, in an ssp1∆ diploid, all the other leading-edge complex proteins are absent and no spores are made (127, 129). This sporulation defect is due to aberrant prospore membrane growth; in particular, the membranes collapse onto the nuclear envelope and close prematurely (124, 127).

The second membrane-associated cytoskeletal structure is composed of septins (Fig. 4). In budding cells, septins form filaments at bud necks important for cytokinesis and for creating a diffusion barrier between mother cells and their buds (130, 131). The basic unit of a septin filament at the bud neck is an octamer containing two copies each of the proteins Cdc3, Cdc10, Cdc11, and Cdc12 (132). In sporulating cells, CDC3 and CDC10 are transcriptionally induced by Ndt80, whereas CDC11 and CDC12 are not (45). Instead, two sporulation-specific septins, Spr3 and Spr28, form an octamer with Cdc3 and Cdc10 and assemble into sheets or bars that line the interior of the growing prospore membrane (33, 133–135) (Fig. 4). In vitro studies demonstrate that the assembly properties of the sporulation-specific septin complex are distinct from the vegetative complex, accounting for the different structures formed in vivo (133). The precise role of these septin sheets in membrane growth is unclear. In some strain backgrounds, removal of septins has no strong sporulation phenotype, whereas in others, there are significant spore formation defects (134–136). In the latter strains, prospore membrane morphology resembles that seen in ssp1∆ mutants, suggesting that the leading-edge complex and the septins collaborate to control prospore membrane growth (136). Consistent with this possibility, in these strains, mutation of the septins disrupts the localization of the Ady3 protein to the leading edge (136). In fact, the leading-edge and septin complexes may be linked by the action of the Ysw1 protein, which interacts with Ady3 in the two-hybrid system but in sporulating cells co-localizes with septins (137). Cells lacking YSW1 show a modest sporulation defect with an increased appearance of morphologically abnormal prospore membranes (137). These latter experiments were done in a strain background where septin mutants do not cause significant sporulation defects. It would be of interest to examine the ysw1∆ phenotype in a strain more sensitive to septin loss.

Prospore membrane closure

Prospore membrane expansion coincides with the segregation of haploid chromosomes to opposite poles of each meiosis II spindle so that by telophase of meiosis II, the chromosomes are contained within the compartments created by the prospore membranes. At this time, the meiosis II spindle breaks down, and the nuclear envelope pinches off from the base of the each lobe to make four haploid nuclei (22) (Fig. 5). To complete cytokinesis, the ends of each prospore membrane then fuse, thereby enclosing each nucleus within a double membrane to create prospores that lack spore walls (28) (Fig. 5). This fusion is topologically similar to abscission of the plasma membrane at the end of a mitotic division in that the cytoplasmic face of the membrane must fuse with itself to make two separate cytoplasmic compartments. However, little is known about how prospore membrane closure is achieved or regulated. Unlike the cytokinesis that separates the bud from its mother, the actin cytoskeleton plays no role in prospore membrane closure (138).

Fig 5.

Prospore membrane closure. After nuclear lobes pinch off to create haploid nuclei, prospore membranes close to complete cytokinesis. Closure is coordinated with meiotic exit through the actions of both the Cdc15-Sps1 pathway and the APC^Ama1 complex. These factors act in parallel to remove Ssp1 and the leading-edge complex from the lips of the membrane, thus allowing the fusion necessary for closure. The molecular steps leading to membrane fusion after Ssp1 removal are unknown.

Prior to closure, the leading-edge complex is removed by proteolysis mediated by the APC (30, 124). This version of the APC contains a sporulation-specific activating subunit, Ama1 (80) (Fig. 5). APC^AMA1 activity is tightly regulated. First, AMA1 transcription is induced upon entry into meiosis by Ime1 (139). Second, the AMA1 RNA undergoes meiosis-specific splicing followed by Rim4-mediated translational repression (53, 140). Third, the activity of the APC^AMA1 complex is inhibited by both the Mnd2 subunit of the APC/C and the Spo13-Cdc5 kinase (63, 141, 142). Collectively, this regulation keeps APC^AMA1 activity low until the end of meiosis II. A low level of activity of APC^AMA1 is required in meiotic prophase I to trigger turnover of mitotic cell cycle factors; however, high activity at this time results in meiotic defects (141–143). At the end of meiosis II, increased activity of APC^AMA1 drives the turnover of several meiotic factors, including cyclins, Cdc5, and Ndt80, leading to exit from meiosis (63, 82).

Cells lacking AMA1 progress through the meiotic divisions; however, no spores are formed, and about 50% of prospore membranes fail to close (30, 80). This failure results in continued expansion and morphologically abnormal membranes (124, 144). The Ssp1 protein, which is normally degraded at the end of meiosis II, is stabilized in ama1∆ mutant cells, and inactivation of a conditional allele of SSP1 can partially rescue the ama1 sporulation defect (30). Thus, Ama1-mediated degradation of Ssp1 is important for prospore membrane closure (Fig. 5).

Several other genes besides AMA1 are required for prospore membrane closure (114, 120, 144). In particular, mutations in SPS1, encoding a sporulation-specific kinase; SPO77, originally identified as a high copy suppressor of sps1; or CDC15, encoding Hippo kinase, all exhibit closure defects similar to ama1∆ with closure of ~50% of the prospore membranes (30, 144) (Fig. 5). Strains lacking SPS1 or SPO77 also show persistent Ssp1 localization to the leading edge, similar to ama1∆ (144). An ama1∆ sps1∆ double mutant shows more severe prospore membrane closure and Ssp1 stabilization defects than either of the single mutants, suggesting overlapping functions that act in parallel to promote prospore membrane closure in S. cerevisiae (144) (Fig. 5).

Cdc15 is required for phosphorylation of Sps1 but not vice versa, suggesting that Cdc15 functions upstream of Sps1 (145). In vegetative cells, Cdc15 is stimulated by association with the Tem1 GTPase and activates the downstream Dbf2/Dbf20 kinases as part of the mitotic exit network, the yeast analog of the mammalian Hippo signaling pathway (146–149). However, TEM1 and DBF2/DBF20 are dispensable for sporulation, while CDC15 is essential (87, 150–152). Thus, Cdc15-Sps1 represents an alternative, sporulation-specific Hippo pathway (145). How Cdc15 activity is triggered in meiotic cells remains to be determined.

CDC15/SPS1 and AMA1 are important for other post-meiotic events, in addition to prospore membrane closure. For example, AMA1, CDC15, and SPS1 are all required for disassembling meiosis II spindles (145, 150, 152). It was therefore possible that the cytokinesis defects exhibited in mutants of these genes were indirectly due to persistence of meiosis II spindles creating a steric block to closure. This idea was ruled out by the observation that prospore membranes still failed to close in sps1∆ or ama1∆ cells when meiosis II spindles were artificially disassembled by addition of microtubule depolymerizing agents (153). Comparison of the degradation of different spindle components in wild-type, ama1∆, and sps1∆ cells demonstrated that Ama1 and Cdc15/Sps1 regulate distinct targets to promote disassembly (153).

As cells exit meiosis, there is a dramatic increase in chromatin compaction such that chromatin in spore nuclei is more highly condensed than in nuclei from vegetative cells (154). This increased chromatin condensation requires phosphorylation of serine 1 on histone H4 (154). SPS1, SPO77, and CDC15, but not AMA1, are required for this increase in H4 serine 1 phosphorylation (145, 154). Thus, the Cdc15/Sps1 pathway drives nuclear compaction at the end of meiosis II.

AMA1 is required for activation of the sporulation-specific protein kinase Smk1, a key regulator of spore wall development (155–157). The transcription of SMK1 and its inhibitor, ISC10, is activated by Ndt80, and both mRNAs are class 1 transcripts. Although the proteins are present prior to meiosis II, Smk1 activity is inhibited by Isc10 binding (158). APC^AMA1 mediates degradation of Isc10 at the end of meiosis II, concomitant with translation of the Smk1 activator Ssp2 (encoded by a class 3 transcript), thereby triggering Smk1 activity and promoting spore wall synthesis (158–160). Together, the Cdc15/Sps1 Hippo and APC^Ama1 pathways drive the transition from meiotic exit to spore development.

ORGANELLAR INHERITANCE AND CELLULAR REJUVENATION

Altered ER organization

A viable spore not only requires the presence of a nucleus but also must contain membrane-bound organelles such as mitochondria, vacuoles, Golgi apparatus, and ER. Organellar inheritance during mitotic growth in budding yeast is largely driven and regulated by the polarized nature of budding cells (161). Polarized microtubules and actin fibers running from the mother into the bud provide a basis for vectoral transport of organelles into the daughter cell (Fig. 6). For example, different adaptors connect mitochondria, peroxisomes, Golgi elements, and small vacuoles to the myosin motor Myo2, which then delivers these organelles along actin filaments into the bud (162–165). By contrast, the actin cytoskeleton is not polarized during sporulation (138), and multiple daughter cells (spores) are formed simultaneously. Therefore, different mechanisms are required to ensure proper organellar inheritance.

Fig 6.

Segregation of the nucleus in vegetative and sporulating cells. During budding, the nucleus is split evenly between the mother and the bud. During sporulation, the nucleus is divided into five parts. Each of the haploid chromosome sets is segregated into prospores, but much of the nuclear envelope and nucleoplasm remain behind in the ascus in a compartment termed the GUNC, or gametogenesis uninherited nuclear compartment.

In vegetative cells, segregation of nuclear envelope into a bud occurs as part of the mitotic division of the spindle, which is arranged with one SPB in the mother cell and one in the bud (Fig. 6). Division of the nucleus results in two identical nuclei, one in each cell. By contrast, at the end of meiosis II, the nucleus is divided into not four, but five different compartments: four sets of haploid chromosomes and a fifth, centrally located, nuclear compartment that remains behind in the ascus (166, 167) (Fig. 6). This remnant body is bounded by nuclear envelope and contains the nucleolar material, most nuclear pores, and concentrations of “senescence factors” such as extrachromosomal rDNA circles and Hsp40-containing protein aggregates (166, 167). This chromosome-free nuclear remnant has been dubbed the gametogenesis uninherited nuclear compartment (GUNC) (168). Thus, the nuclear envelope that is enclosed by the prospore membrane is distinct from that of the GUNC as the GUNC is enriched in nuclear pore complexes (NPCs), while the chromosome containing regions are relatively devoid of NPCs but carry SPBs (Fig. 6 and 7).

Fig 7.

Mitochondrial and ER rearrangements during sporulation. (A) In vegetative or pre-meiotic cells, the cytoplasm is organized with cortical ER, and mitochondria are distributed around the cell cortex through ER-plasma membrane or mitochondria-plasma membrane tethers. ER exit sites are scattered around the cortical ER, and nuclear pores are found uniformly throughout the nuclear envelope. (B) Upon entry into the meiotic divisions, the mitochondria-plasma membrane tethers are degraded and mitochondria associate with the nuclear envelope. The cortical ER is fragmented, leaving remnants associated with ER-plasma membrane tethers, and the remaining membranes are either degraded through autophagy or are associated with the nuclear envelope. As the haploid chromosome sets segregate into nuclear lobes at meiosis II, ER exit sites reappear within the growing prospore membranes. The leading edge of the prospore membrane acts as a “strainer” to limit the entry of nuclear pores and mitochondria into the forming prospore compartment.

The separation of the nuclear pore complexes into the GUNC involves not just segregation of these structures but also their transient reorganization. In particular, the “nuclear basket,” a subcompartment of the nuclear pore that projects into the nucleoplasm, dissociates from the rest of the nuclear pore structure during the meiotic divisions (169). This dissociation/reassociation cycle occurs during each meiotic division (169). The first dissociation during meiosis I is mediated by phosphorylation of the nucleoporin Nup60 by Cdc5, and the second dissociation at meiosis II allows the basket to escape from the GUNC and end up incorporated into nuclear pores within the spore nucleus (169). Interestingly, the same dissociation of basket subunits is seen during the meiotic divisions in S. pombe (169). The permeability barrier created by nuclear pores breaks down transiently during meiosis in S. pombe (170, 171). Though it is not known if a similar membrane permeabilization occurs during S. cerevisiae meiosis, it is tempting to speculate that reorganization of nuclear pore complexes is tied to this change in permeability.

The rearrangement of the nucleus and segregation of nuclear and nuclear envelope components into the GUNC requires not only the meiotic spindle to pull the chromosomes apart but also formation of the prospore membrane on the cytoplasmic side of the nucleus (167). In the absence of prospore membranes, nuclear pore complexes and other materials are not concentrated in the GUNC but rather are found throughout the nucleus (167). Further, the leading-edge complex at the lip of the prospore membrane is also required (167). These results suggest that the leading-edge complex acts like a “strainer” to exclude nuclear pore complexes from the portion of the ER that enters the prospore membrane. A similar role in regulating diffusion of ER components has been proposed for the septin ring at the bud neck in vegetative cells (131).

In its role in driving segregation of nuclear components, the leading-edge complex may act through the nuclear envelope proteins Src1 and Heh2 (172). These proteins of the inner nuclear membrane recruit the ESCRT-III complex to the nuclear envelope during meiosis to drive division of the nuclear envelope. The recruitment of ESCRT-III is also required for concentration of nuclear pores into the GUNC (172). How the position of the leading edge is communicated to the interior of the nucleus is not known.

The ER in yeast consists of both the nuclear envelope and cortical ER (Fig. 7A). The latter consists of tubules and sheets of ER membranes that extend through the cytoplasm and underlie the plasma membrane (173). In vegetative cells, cortical ER elements are inherited separately from the nuclear envelope, delivered into the bud along actin filaments (173, 174). During meiosis, however, the cortical ER is dramatically reorganized and is inherited in association with the nuclear envelope (175, 176). As cells enter meiosis II, cortical ER-specific proteins, such as Rtn2, relocalize from the cell periphery to the cell center close to the nuclear envelope (176) (Fig. 7B). The collapse of the cortical ER is driven by NDT80, which causes the separation of cortical ER into membrane fragments with different fates. Cortical ER at ER-plasma membrane contact sites, where ER is anchored to the plasma membrane through tethering proteins, remains as patches of cortical ER associated with the plasma membrane of the mother cell (175) (Fig. 7). These patches of ER remain in the ascus and are not inherited by spores (175).

The remainder of the cortical ER is released from the cell periphery (175). These released membranes can either be degraded by an ATG40-mediated ER-phagy pathway or associate closely with the nuclear envelope (175). A fraction of the nuclear envelope-associated membranes are then captured within prospore membranes as the nuclear lobes are engulfed (Fig. 7). Thus, cortical ER inheritance is linked directly to delivery of the nuclear envelope to the prospore. Moreover, in sporulating cells, unlike vegetative cells, the bulk of the cortical ER is not inherited by daughter cells. It instead either remains connected to the ascal plasma membrane, is degraded via ER-phagy, or simply remains behind in the ascus, associated with the nuclear envelope of the GUNC.

How NDT80 expression triggers cortical ER collapse is not entirely understood. Release of cortical ER from the periphery involves the function of several proteins, Rtn1, Rtn2, Lnp1, and Yop1, known to be involved in regulating the fission and fusion of ER tubular networks (177–179). Induction of NDT80 likely shifts the balance toward fission, resulting in fragmentation of the cortical ER (175). However, none of these genes are Ndt80 targets. Therefore, expression of NDT80 may stimulate their activity indirectly. Turnover of ER via ATG40-mediated ER-phagy is also increased after NDT80 induction (175). However, in this case, ATG40 expression is increased in meiosis II cells, providing a clear connection between NDT80 and increased ER phagy (52, 175).

Another significant change in the ER during meiosis II is the relocalization of ER exit sites (Fig. 7). ER exit sites are specialized regions from which COPII coated vesicles bud off the ER carrying cargo into the secretory pathway (180). In vegetative yeast cells, after budding from the ER, these vesicles fuse with Golgi-derived vesicles to create cis-Golgi elements that then mature into medial- and trans-Golgi elements before vesiculating (181, 182). Consistently, as shown in Pichia pastoris, a steady supply of COPII vesicles is required to maintain the Golgi apparatus in in yeast (183). This is likely to be true during sporulation as well.

ER exit sites are marked by concentrations of the Sec16 protein, a peripheral membrane protein required for COPII vesicle formation (184, 185). In mitotic or pre-meiotic cells, Sec16 foci are present throughout the cytoplasm (112, 185). As cells enter meiosis, the number of Sec16 foci drops and Golgi elements disappear (112). After NDT80 induction, Sec16 foci reappear, but these foci are concentrated in ER elements that are within growing prospore membranes (112). The Gip1/Glc7 phosphatase complex, which is essential for sporulation, is required for the reorganization of Sec16 foci (33, 112). Loss of GIP1 results in abnormally small prospore membranes (111), suggesting that initiating vesicular transport from within prospore membranes is important for proper membrane growth. Consistent with this idea, loss of the ER exit site component Sed4 results in defective ER exit site localization during sporulation and variable prospore membrane formation, with those prospore membranes that encompass normal numbers of ER exit sites showing normal expansion and those with reduced numbers displaying defects (112). The rearrangement of ER exit sites is tied to the collapse of the cortical ER. In the absence of RTN1 or YOP1, which is required for cortical ER relocalization, rearrangement of ER exit sites does not occur (112). Thus, the dynamic collapse of the ER during sporulation is necessary to rearrange ER exit sites and properly form prospore membranes. The growing prospore membranes then “feed back” to help sculpt the proper segregation of the nuclear envelope and its contents.

Mitochondrial inheritance

Mitochondrial behavior during sporulation mirrors that of cortical ER in that the mitochondria release from the cell periphery and collect around the nuclear envelope during meiosis II (186, 187) (Fig. 7). Also like the ER, mitochondria are incompletely segregated to daughter cells with around 50% of the mitochondrial mass excluded from spores (186, 188, 189) (Fig. 7). The increase in Ime2 activity after Ndt80 induction leads to relocalization of mitochondria (190). Mitochondria are anchored at the cell periphery by a membrane contact site complex, MECA, consisting of Mdm36 and Num1 (191) (Fig. 7). Ime2 phosphorylates both proteins in vitro, and at least Num1 in vivo, and both proteins are degraded in meiosis II in an NDT80 and IME2-dependent fashion (190). Thus, entry into the meiotic divisions leads to mitochondrial relocalization by degradation of the cortical mitochondrial anchor.

The released mitochondria associate tightly with the nuclear envelope, probably via mitochondrial-ER contacts (Fig. 7) (187, 190). Mitochondria enter into the prospore while associated with the nuclear envelope, and this association is proposed to be the mechanism by which segregation of mitochondria into spores is achieved (176, 190). Therefore, in contrast to mitotic cells where the inheritance of both mitochondria and ER is driven by actin-based transport, the inheritance of both organelles during sporulation is mediated by spindle-driven segregation of the nuclear envelope into the developing spore. In both cases, this results in distribution of a fraction of the organellar mass into spores but leaves a significant amount behind in the ascus. Why mitochondrial segregation is incomplete and whether there is a difference between those mitochondria that are inherited by spores and those left behind in the ascus are not known. In vegetative cells, a mitochondrial quality control system assures that less-fit mitochondria remain in the mother cell, and higher-functioning mitochondria are delivered to the bud (192, 193). Whether this same system functions in sporulation is not known, but it is tempting to speculate that, similarly, the mass left in the ascus might be a mitochondrial equivalent of the GUNC.

Other organelles

While some of the pre-existing ER and mitochondria are distributed to prospores during sporulation, other membrane-bound organelles are left behind entirely in the ascus and must be constructed de novo within the spores (42, 188). One example is the vacuole (176, 194) (Fig. 8). Fluorescent markers of both the vacuolar lumen and vacuolar membrane are absent from newly formed spores, though fluorescence appears within the spores after a couple of hours (176, 194). By contrast, Golgi elements are visible within the prospore membranes even before membrane closure (176). However, these appear to be generated from the ER exit sites that are localized within the cytoplasm enveloped by the growing prospore membrane (112). Presumably, the Golgi elements formed from these ER exit sites then mature to generate different Golgi cisternae as shown in vegetative cells (181, 183).

Fig 8.

Organellar inheritance in spore formation. At the end of meiosis, the newly formed spores have inherited some nuclear envelope/ER and some mitochondria from the mother cell, though in both cases, the bulk of the pre-existing organellar mass is left behind in the ascus. Also left behind in the ascus is the vacuole (which will lyse during sporoptosis) and prospore membrane-associated lipid droplets marked by Sps4. New Golgi elements are formed from ER exit sites within the spore. Pet10-marked lipid droplets may also be formed de novo within spores.

Lipid droplets also display dynamic behavior during sporulation. Lipid droplets are formed by the budding of neutral lipids out of the ER. In vegetative cells, most droplets remain closely associated with the ER (195, 196). Early in sporulation, many droplets are associated with vacuolar membranes rather than the ER (126). During meiosis II, this distribution changes again, and a subset of droplets is associated with the ascal side of prospore membranes (38, 125). These droplets remain in the ascus after closure where they play a role in spore wall development [38, 125: (126) #970, 197] (Fig. 8). Different classes of lipid droplets can be distinguished in sporulating cells by their associated proteins. The Sps4, Nus1, Srt1, and Pdr16 proteins are found only on prospore membrane-associated droplets and at least one protein, Pet10, localizes only to lipid droplets within prospores (38, 125, 129, 197) (Fig. 8). While it is not known if Pet10-carrying droplets inherited by spores are newly formed from the ER/nuclear envelope that is enveloped by the prospore membrane, that would be consistent with the behavior of the Golgi apparatus. In any event, it is clear that a large fraction of lipid droplets is left behind in the ascus and not inherited by spores, as for other organelles.

If there is a common theme to organellar inheritance during sporulation, it is that most of the contents of the mother cell are left behind to be destroyed in the ascus and new components are regenerated within the spore. This behavior is not only true of membrane-bound organelles but also extends to individual proteins. For example, activation of Ndt80 leads to a transient downregulation of expression of the cytoplasmic enzyme superoxide dismutase, Sod1, as cells enter the meiotic divisions (198). The existing pool of Sod1 is then induced to form aggregates that are degraded. Transcription of SOD1 resumes within prospores so that only newly synthesized Sod1 is present in the spores (198).

There are multiple possible explanations why cells discard existing cellular components during sporulation, only to resynthesize them within the spores. The presence of new constituents may make the spores more robust. This seems to be the case for Sod1, as failure to discard the old Sod1 pool results in spores that are more sensitive to oxidative stress (198). Minimizing cytoplasmic and nucleoplasmic inheritance could also minimize transmission of RNA viruses (199). Ascospores are more resistant to digestion by insects than vegetative cells, suggesting that this cell type may be specialized for dispersal by insect vectors (40). It is possible, though not proven, that degradation of the ascal contents releases volatile compounds that attract insects, analogous to the way bright colors of fruit attract birds for dispersal of seeds (40). Finally, discarding old and damaged cellular components may be important for resetting of the aging clock (200, 201).

Individual yeast cells have a limited replicative lifespan, defined as the number of times an individual cell can produce a bud before the cell senesces (202). When a cell sporulates, the resulting four spores all have the same replicative potential (all are “young” cells), whether the original cell that sporulated was a young cell (never budded) or an aged cell (has already budded multiple times) (203). Thus, sporulation can reset the aging clock in yeast. This “reset” may be tied to the removal of old and damaged material, for example, the protein aggregates and extrachromosomal rDNA circles that are left behind in the GUNC. Indeed, spores from src1∆ mutants that cannot concentrate material in the GUNC showed a reduced replicative lifespan (172). In vegetative cells, segregation of higher-functioning mitochondria into the bud is important for maintaining the replicative lifespan of the daughter cell, and the restricted segregation of mitochondria into spores could play a similar role (193).

Consistent with the idea that restricted segregation of organelles into spores might be related to aging, induction of NDT80 and formation of a spore are required for reset (203). However, induction of NDT80 in aged vegetative cells is sufficient by itself to extend lifespan, suggesting that resetting of the clock in sporulation may also involve NDT80 targets independent of the replacement of older cell materials (203).

ARE THE MECHANISMS OF ASCOSPORE FORMATION CONSERVED?

The cytology of ascospore formation is broadly similar in all ascomycetes. Double membranes engulf haploid chromosomes, giving rise to spores within the interior of the mother cell (5–9). Therefore, the molecular mechanisms of spore formation might be expected to be conserved across ascomycetes as well. However, whether the details worked out in S. cerevisiae are generally true in ascomycetes or are unique to Saccharomyces remains to be determined. The other system where ascospore formation has been extensively studied is the fission yeast Schizosaccharomyces pombe (12). The S. pombe lineage branched very early within the ascomycete tree (204). Thus, if a process is mechanistically similar in both S. pombe and S. cerevisiae, it is likely to be conserved in most ascomycetes.

Comparison of the requirements for sporulation in budding and fission yeasts suggests that the overall process is conserved, though many of the specific proteins involved are not (Table 1). For example, NDT80 is the central regulator of sporulation in S. cerevisiae; however, NDT80-related transcription factors are only found in other budding yeasts, and even then, they are not always involved in sporulation (20, 205, 206). So how useful is the S. cerevisiae regulatory system as a model for sporulation in other ascomycetes? S. pombe uses the same regulatory logic described in S. cerevisiae but with a non-orthologous transcriptional activator called Mei4 (207). Mei4 activates transcription of genes required for the meiotic divisions and for ascospore assembly (207–210). Transcripts of the Mei4 regulon are induced together but translated at different times in meiosis (211). Included in the Mei4 regulon is an RRM-containing protein, Meu5, and, analogous to Pes4/Mip6, mutants of meu5 exhibit earlier translation of a subset of these delayed Mei4-induced transcripts (209, 211, 212). While the regulatory details in S. pombe have yet to be elucidated and will certainly differ from S. cerevisiae in terms of the specific proteins involved, these observations suggest that the basic regulatory mechanisms may be the same in both yeasts.

TABLE 1.

Analogous sporulation functions in S. cerevisiae and S. pombe

Function	S. cerevisiae protein (references)	S. pombe protein (references)	Orthologs (yes/no)
Meiotic transcription regulator	Ndt80 (20, 21)	Mei4 (207)	No
Translational regulation	Pes4/Mip6 (71)	Meu5 (212)	No
Meiotic APC regulator	Ama1 (80)	Fzr1 (213, 214)	No
Meiotic outer plaque proteins	Spo74/Mpc54/Spo21 (26, 77, 78)	Spo13/Spo2 (215)	No
SNAREs for prospore/forespore membrane fusion	Sso1/Snc1,2/Spo20 (25, 88, 89)	Psy1/Syb1/Sec9 (216–218)	Yes
Rab protein for prospore/forespore membrane fusion	Sec4 (25)	Ypt2 (219)	Yes
Sporulation-specific septins	Spr3/Spr28 (134, 135)	Spn5/Spn6 (220)	Yes
Leading-edge component	Ssp1 (127)	Meu14 (221)	No

Forespore membrane formation in S. pombe

A similar pattern of “same but different” applies to initiation of forespore membranes (the name for prospore membranes in S. pombe). Forespore membranes initiate in meiosis II on a modified cytoplasmic surface of the S. pombe SPB (12). Also, as in S. cerevisiae, there are several sporulation-specific MOP components required for proper forespore membrane initiation (215, 222, 223). However, the S. pombe MOP proteins are not orthologous to the S. cerevisiae proteins. Therefore, although membrane initiation begins in the same way and at the same intracellular location in the two yeasts, the molecular details have diverged.

A mixed pattern of conservation can be seen at many steps in ascospore formation. Fusion of vesicles with the forespore membrane in S. pombe is mediated by a SNARE complex including the syntaxin Psy1, the synaptobrevin Syb1, and Sec9 (216–218). The first two proteins are the S. pombe orthologs of S. cerevisiae Sso1 and Snc1/Snc2 involved in fusion at prospore membranes (88, 89). However, S. pombe lacks the sporulation-specific form of Sec9, Spo20, which completes this trimer in S. cerevisiae (25, 88, 89). Rather, S. pombe Sec9 mediates fusion at both the plasma and forespore membranes (217).

The fusion machinery in both budding and fission yeasts includes the Rab protein Sec4. Sec4 (named Ypt2 in S. pombe) is required for vesicle fusion in both systems, but the GEF that activates Sec4 has diverged (25, 102, 219). S. pombe contains two Sec4 GEFs, Sec2 and Spo13. Sec2 is likely involved in vesicle fusion at the plasma membrane, while Spo13 is a sporulation-specific MOP component necessary for forespore membrane formation (102). By contrast, S. cerevisiae carries only Sec2, and it promotes prospore membrane formation without being itself an SPB component (224). Many hyphal ascomycetes encode both Sec2 and Spo13 homologs, while budding yeasts do not, suggesting that the S. pombe arrangement is ancestral and that Spo13 was subsequently lost in the lineage leading to S. cerevisiae (102).

Septin sheets underlie the expanding prospore/forespore membranes in both S. cerevisiae and S. pombe (134, 135). Moreover, S. pombe expresses a pair of sporulation-specific septins, orthologs of S. cerevisiae Spr3 and Spr28, that are important for spore formation (220). Therefore, the unique biochemical properties described for the S. cerevisiae sporulation-specific septin octamer may be generally important for ascospore formation (133). Also similar to S. cerevisiae, there is a ring structure found at the leading edge of the forespore membrane that is essential for proper forespore membrane growth, although the key protein of this structure, Meu14, is not orthologous to the budding yeast protein, Ssp1 (221). Finally, S. pombe encodes a sporulation-specific APC activator, fzr1⁺ (also known as mfr1⁺), that is essential for sporulation (213, 214). Though Fzr1 is not orthologous to Ama1, it has been implicated in exit from meiosis II, providing another example of conservation of regulatory architecture, if not of specific proteins (225, 226).

While non-orthologous proteins may have similar functions in budding and fission yeast sporulation, the complementary situation is also observed; that is, S. pombe contains orthologs of proteins which are required for spore formation in S. cerevisiae but have no sporulation defects when the genes are mutated. For example, S. pombe contains two paralogs of VPS13 and orthologs of SPO71 and SPO73 that are transcriptionally induced in sporulation (209). However, mutations in any of these genes display, at most, mild sporulation defects (227, 228). The function of these genes in S. pombe remains to be determined.

The great evolutionary distance between budding and fission yeast, and having only the two systems to compare, makes it difficult to establish what the ancestral machinery for ascospore formation looked like. What is greatly needed is a detailed genetic and molecular analysis of ascospore formation in a filamentous ascomycete, for example, Aspergillus or Neurospora, that lies between these two yeasts in the fungal tree (204). Insights from such an analysis could greatly clarify what the conserved, central features of ascospore formation are.

Biography

Aaron M. Neiman is a Professor in the Department of Biochemistry and Cell Biology at Stony Brook University. He performed his doctoral work with Ira Herskowitz at UCSF and his postdoctoral training with Rolf Sternglanz at Stony Brook University. He joined the faculty at Stony Brook in 1999. A postdoctoral project intended to identify genes important for nuclear organization instead identified multiple genes involved in ER export. This led to an interest in vesicle trafficking and, in turn, the realization that spore formation provided an opportunity to study how intracellular transport is altered during cellular differentiation. The Neiman lab has spent the last 25 years exploring how yeast cells rearrange their architecture to package the meiotic products into stress resistant spores.