|Home | About | Journals | Submit | Contact Us | Français|
During yeast sporulation, a forespore membrane (FSM) initiates at each spindle-pole body and extends to form the spore envelope. We used Schizosaccharomyces pombe to investigate the role of septins during this process. During the prior conjugation of haploid cells, the four vegetatively expressed septins (Spn1, Spn2, Spn3, and Spn4) coassemble at the fusion site and are necessary for its normal morphogenesis. Sporulation involves a different set of four septins (Spn2, Spn5, Spn6, and the atypical Spn7) that does not include the core subunits of the vegetative septin complex. The four sporulation septins form a complex in vitro and colocalize interdependently to a ring-shaped structure along each FSM, and septin mutations result in disoriented FSM extension. The septins and the leading-edge proteins appear to function in parallel to orient FSM extension. Spn2 and Spn7 bind to phosphatidylinositol 4-phosphate [PtdIns(4)P] in vitro, and PtdIns(4)P is enriched in the FSMs, suggesting that septins bind to the FSMs via this lipid. Cells expressing a mutant Spn2 protein unable to bind PtdIns(4)P still form extended septin structures, but these structures fail to associate with the FSMs, which are frequently disoriented. Thus, septins appear to form a scaffold that helps to guide the oriented extension of the FSM.
Yeast sporulation is a developmental process that involves multiple, sequential events that need to be tightly coordinated (59, 68). In the fission yeast Schizosaccharomyces pombe, when cells of opposite mating type (h+ and h−) are mixed and shifted to conditions of nitrogen starvation, cell fusion and karyogamy occur to form a diploid zygote, which then undergoes premeiotic DNA replication, the two meiotic divisions, formation of the spore envelopes (comprising the plasma membrane and a specialized cell wall), and maturation of the spores (74, 81). At the onset of meiosis II, precursors of the spore envelopes, the forespore membranes (FSMs), are formed by the fusion of vesicles at the cytoplasmic surface of each spindle-pole body (SPB) and then extend to engulf the four nuclear lobes (the nuclear envelope does not break down during meiosis), thus capturing the haploid nuclei, along with associated cytoplasm and organelles, to form the nascent spores (55, 68, 81). How the FSMs recognize and interact with the nuclear envelope, extend in a properly oriented manner, and close to form uniformly sized spherical spores is not understood, and study of this model system should also help to elucidate the more general question of how membranes obtain their shapes in vivo.
It has been shown that both the SPB and the vesicle trafficking system play important roles in the formation and development of the FSM and of its counterpart in the budding yeast Saccharomyces cerevisiae, the prospore membrane (PSM). In S. pombe, the SPB changes its shape from a compact dot to a crescent at metaphase of meiosis II (26, 29), and its outer plaque acquires meiosis-specific components such as Spo2, Spo13, and Spo15 (30, 57, 68). This modified outer plaque is required for the initiation of FSM assembly. In S. cerevisiae, it is well established that various secretory (SEC) gene products are required for PSM formation (58, 59). Similarly, proteins presumably involved in the docking and/or fusion of post-Golgi vesicles and organelles in S. pombe, such as the syntaxin-1A Psy1, the SNAP-25 homologue Sec9, and the Rab7 GTPase homologue Ypt7, are also required for proper FSM extension (34, 53, 54). Consistent with this hypothesis, Psy1 disappears from the plasma membrane upon exit from meiosis I and reappears in the nascent FSM.
Phosphoinositide-mediated membrane trafficking also contributes to the development of the FSM. Pik3/Vps34 is a phosphatidylinositol 3-kinase whose product is phosphatidylinositol 3-phosphate [PtdIns(3)P] (35, 72). S. pombe cells lacking this protein exhibit defects in various steps of FSM formation, such as aberrant starting positions for extension, disoriented extension and/or failure of closure, and the formation of spore-like bodies near, rather than surrounding, the nuclei, suggesting that Pik3 plays multiple roles during sporulation (61). The targets of PtdIns(3)P during sporulation appear to include two sorting nexins, Vps5 and Vps17, and the FYVE domain-containing protein Sst4/Vps27. vps5Δ and vps17Δ mutant cells share some of the phenotypes of pik3Δ cells (38). sst4Δ cells also share some of the phenotypes of pik3Δ cells but are distinct from vps5Δ and vps17Δ cells, consistent with the hypothesis that Pik3 has multiple roles during sporulation (62).
Membrane trafficking processes alone do not seem sufficient to explain how the FSMs and PSMs extend around and engulf the nuclei, suggesting that some other mechanism(s) must regulate and orient FSM/PSM extension. The observation that the FSM is attached to the SPB until formation of the immature spore is complete (68) suggests that the SPB may regulate FSM extension. In addition, the leading edge of the S. cerevisiae PSM is coated with a complex of proteins (the LEPs) that appear to be involved in PSM extension (51, 59). S. pombe Meu14 also localizes to the leading edge of the FSM, and deletion of meu14 causes aberrant FSM formation in addition to a failure in SPB modification (60). However, it has remained unclear whether the SPB- and LEP-based mechanisms are sufficient to account for the formation of closed FSMs and PSMs of proper size and position (relative to the nuclear envelope), and evidence from S. cerevisiae has suggested that the septin proteins may also be involved.
The septins are a conserved family of GTP-binding proteins that were first identified in S. cerevisiae by analysis of the cytokinesis-defective cdc3, cdc10, cdc11, and cdc12 mutants (41). Cdc3, Cdc10, Cdc11, and Cdc12 are related to each other in sequence and form an oligomeric complex that localizes to a ring in close apposition to the plasma membrane at the mother-bud neck in vegetative cells (12, 20, 25, 41, 47, 77). The septin ring appears to be filamentous in vivo (12), and indeed, the septins from both yeast (11, 20) and metazoans (31, 36, 69) can form filaments in vitro. The yeast septin ring appears to form a scaffold for the localization and organization of a wide variety of other proteins (8, 22), and it forms a diffusion barrier that constrains movement of membrane proteins through the neck region (7, 8, 73). In metazoan cells, the septins are involved in cytokinesis but are also implicated in a variety of other cellular processes, such as vesicular transport, organization of the actin and microtubule cytoskeletons, and oncogenesis (27, 70).
In S. cerevisiae, a fifth septin (Shs1) is also expressed in vegetative cells, but the remaining two septin genes, SPR3 and SPR28, are expressed at detectable levels only during sporulation (15, 17). In addition, at least some of the vegetatively expressed septins are also present in sporulating cells (17, 48), and one of them (Cdc10) is expressed at much higher levels there than in vegetative cells (32). The septins present during sporulation are associated with the PSM (15, 17, 48, 51), and their normal organization there depends on the Gip1-Glc7 protein phosphatase complex (71). However, it has been difficult to gain insight into the precise roles of the septins during sporulation in S. cerevisiae (59), because some septins are essential for viability during vegetative growth, and the viable mutants have only mild phenotypes during sporulation (15, 17), possibly because of functional redundancy among the multiple septins.
S. pombe seemed likely to provide a better opportunity for investigating the role of septins during spore formation. There are seven septin genes (spn1+ to spn7+) in this organism (23, 41, 63). Four of these genes (spn1+ to spn4+) are expressed in vegetative cells, and their products form a hetero-oligomeric complex that assembles during cytokinesis into a ring at the division site (2, 3, 10, 76, 79). The septin ring is important for proper targeting of endoglucanases to the division site (44), and septin mutants show a corresponding delay in cell separation (10, 41, 44, 76). However, even the spn1Δ spn2Δ spn3Δ spn4Δ quadruple mutant is viable and grows nearly as rapidly as the wild type (our unpublished results), a circumstance that greatly facilitates studies of the septins' role during sporulation.
spn5+, spn6+, and spn7+ are expressed at detectable levels only during sporulation (1, 45, 78; our unpublished results), and spn2+, like its orthologue CDC10 (see above), is strongly induced (45), but the roles of the S. pombe septins in sporulation have not previously been investigated. In this study, we show that the septins are important for the orientation of FSM extension, suggesting that the septins may have a more general role in dynamic membrane organization and shape determination.
S. pombe strains used in this study are listed in Table Table1;1; strain constructions are described below or in the table. PCR primers are listed in Table Table2.2. Yeast cells were grown in YE (rich) or EMM (synthetic) medium (50), and SSA (16) or MM-N (EMM without NH4Cl) medium was used to induce mating and/or sporulation. Except where noted, cells to be examined microscopically were grown for 12 to 18 h on an EMM plate at 30°C, transferred to an SSA plate at 30°C, and incubated for 8 to 14 h before examination. When appropriate, thiamine was added to media at 15 μM to repress expression of genes controlled by the nmt1 promoter.
Strains containing mutant alleles of spn1 to spn7 and pik1 were constructed as follows. The genes were amplified from genomic DNA by PCR and cloned by TA cloning into vector pT7Blue (Novagen). Plasmids containing partially deleted genes were then constructed as shown in Fig. Fig.1,1, using ura4+ and LEU2 marker cassettes obtained as HindIII fragments from plasmids KS-ura4 (6) and KS-LEU2. KS-LEU2 was constructed by cloning a HindIII fragment containing LEU2 from pART1 (46) into pBluescript II KS(+) (Stratagene). Fragments were released from the plasmids by digestion with the indicated restriction enzymes (Fig. (Fig.1)1) and were transformed into strain THP18, with selection for Ura+ or Leu+. The success of the constructions was checked by PCR and also (for spn6 and spn7) by showing that ura4+ and LEU2 were tightly linked to each other when the spnΔ::ura4+ and spnΔ::LEU2 strains were crossed. Double mutants were constructed by crosses among the single mutants.
In addition, the complete spn2 coding region was deleted or tagged at its C terminus with the GFP or mRFP1 (13) sequence by a PCR method (6) using pFA6a-kanMX6, pFA6a-GFP(S65T)-kanMX6, or pFA6a-mRFP-kanMX6 (62) as template and THP17, THP18, SG14, or MO822 as the recipient strain, yielding strains MO684, MO701, MO815, and MO832, respectively. Proper integration at the spn2 locus was confirmed by PCR (6).
A strain expressing green fluorescent protein (GFP)-tagged Psy1 was constructed as follows. First, plasmid pTN381 was constructed by inserting a SacI linker into the BamHI site of pBR(leu1) (56). Second, GFPS65T sequences were amplified by PCR from plasmid pEG5 (a gift from Y. Hiraoka, National Institute of Information and Communications Technology, Kobe, Japan), digested with XhoI and BamHI (sites in the primers), and cloned into XhoI/BamHI-digested pBluescript II, yielding pBS(GFP). The psy1+ open reading frame (ORF) was then amplified by PCR from genomic DNA, digested with SalI and SacI (sites in the primers), and ligated into SalI/SacI-digested pBS(GFP), yielding pBS(GFP-psy1). The psy1+ promoter region was then amplified from genomic DNA by PCR, digested with XhoI (sites in the primers), and ligated into XhoI-digested pBS(GFP-psy1), yielding pBS(psy1pro-GFP-psy1). The orientation of the promoter fragment was checked by restriction enzyme digestion. The ApaI-SacI fragment of pBS(psy1pro-GFP-psy1) was then ligated into ApaI/SacI-digested pTN381, yielding pBR(GFP-Psy1). Finally, pBR(GFP-Psy1) was linearized by digestion at the SnaBI site in the leu1+ 3′-noncoding region and transformed into strain SG14, yielding strain MO817, which was checked by confirming that Leu+ and Leu− progeny segregated 2:2 in crosses and that the expected GFP labeling of the FSM was seen.
Plasmids for the expression of septins tagged by GFP or maltose-binding protein (MBP) were constructed as follows. For GFP tagging, the genomic spn loci, including promoters, were amplified by PCR and inserted as BamHI-NotI or BglII-NotI fragments (sites in the primers) into BamHI/NotI-digested pAL(GFP) (62), yielding pAL(spn2-GFP), pAL(spn5-GFP), pAL(spn6-GFP), and pAL(spn7-GFP). The tagged proteins appeared to be functional based on their ability to rescue the sporulation defects of the corresponding mutants (data not shown). For MBP tagging, plasmid pMALBB was constructed by inserting a BglII linker into the BamHI site of expression vector pMAL-cRI (New England Biolabs). cDNAs of the spn2+, spn5+, spn6+, and spn7+ coding regions were amplified by reverse transcription-PCR (RT-PCR), using RNA extracted (see below) from strain THP18, and were cloned by TA cloning into vector pT7Blue. The N-terminal portions of the cDNAs were then isolated using a BglII site (included in the primer) and an appropriate site within each gene, as follows (numbers indicate base pairs of the cDNAs): spn2+, AccI (1-408); spn5+, AccII (1-601); spn6+, BamHI (1-585); and spn7+, BamHI (1-594). These fragments were cloned into BglII/HincII- or BglII-cut pMALBB, yielding pMAL(spn2N), pMAL(spn5N), pMAL(spn6N), and pMAL(spn7N), respectively. Site-directed mutagenesis of spn2+ was done in plasmid pAL(spn2-GFP) and in the pT7Blue spn2+ cDNA plasmid as described previously (67). The mutagenized pAL(spn2-GFP) plasmid was then cut with BamHI and NotI, and the spn2 fragment was recloned into BamHI/NotI-cut pAL(GFP), whereas the mutagenized spn2 cDNA from pT7Blue(spn2+) was cloned into pMALBB as described above.
Other plasmids were as follows. pAL(spo3-GFP), pAL(spo3-HA), and pREP81(GFP-psy1) were described previously (53), and pAUSK+ is like pALSK+ (75) except that it contains S. cerevisiae URA3 instead of LEU2. pAU(Spo3-HA) was created by inserting a BamHI-SacI fragment from pAL(spo3-HA) into pAUSK+. pREP41(GFP-spn2) and pREP41(GFP-spn5) were constructed by ligating the BglII-XhoI fragments from the pT7Blue(spn2+) and pT7Blue(spn5+) cDNA clones (see above), respectively, into BglII/XhoI-digested pGFT41 (a gift from Y. Watanabe) (constructed as described for pGFT81 , but using pREP41  as the parent plasmid). These plasmids express full-length GFP-Spn2 and GFP-Spn5 under the control of the medium-strength nmt41 promoter; the tagged proteins appeared to be functional based on their ability to rescue the sporulation defects of the corresponding mutants in media lacking thiamine (data not shown). To construct pREP41(GFP-PHOsh2-dimer), a SmaI-SacI fragment containing GFP-PHOsh2-PHOsh2-GFP was excised from pTL511 (66) and ligated into SmaI/SacI-digested pGADT7 (Clontech). An NdeI-XhoI (both sites from pGADT7) fragment from the resulting plasmid was then subcloned into NdeI/XhoI-digested pREP41 (9).
Total RNA was extracted using a simplified hot-phenol method based on that described previously (42). Briefly, the frozen cell pellet (~100 μl) was thawed on ice, suspended in 500 μl TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.5), mixed with 500 μl of phenol-chloroform by vortexing, and incubated at 65°C for 1 h with periodic vortexing. After centrifugation, the aqueous phase was removed and extracted as before, using phenol-chloroform. Finally, RNA was precipitated with ethanol and dissolved in diethyl pyrocarbonate-treated water.
Cells to be examined by phase-contrast microscopy were suspended in phosphate-buffered saline (PBS) without fixation. Except where noted, cells to be examined by fluorescence microscopy were growing exponentially in EMM liquid culture or were collected from a sporulation plate and suspended in MM-N medium. In either case, cells were usually fixed with formaldehyde, washed, stained for DNA with 4′,6-diamidino-2-phenylindole (DAPI), and (where appropriate) immunostained for Spo3-hemagglutinin (Spo3-HA) essentially as described previously (62). In one case, DAPI staining was accompanied by staining of F-actin using tetramethyl rhodamine isocyanate (TRITC)-phalloidin (Sigma) at 2 μM. In another case, cells were stained with DAPI after fixation with 30% ethanol for ≥15 min at room temperature. In another case, cells were fixed with methanol (8 min at −20°C) and acetone (1 min at 4°C) and washed twice in PBS; DNA was then stained with bisbenzimide H33258 (Sigma) at 50 μg/ml in PBS. Fluorescence of the GFP-PHOsh2 dimer was observed using unfixed cells because fixation destroyed the GFP signal, and some observations of GFP-red fluorescent protein (RFP) double labeling were made using cells that were suspended in PBS without fixation or DAPI staining. Cells were observed using a Zeiss Axiovert 100 fluorescence microscope with a Plan Apochromat 100×/1.40-numerical-aperture (NA) objective (most images), a Nikon Microphot SA microscope with a Plan Apo 60×/1.40-NA objective (see Fig. Fig.4A),4A), an Olympus IX71 inverted microscope with a UPlan Apo 100×/1.35-NA objective (see Fig. 6C and D and 10C), or a Nikon Eclipse 600-FN microscope with a Plan Apo 100×/1.40-NA objective (see Fig. 10B and 11D). To obtain three-dimensional (3D) reconstructed images, IPLab 3.6 software (Scanalytics) was used. The maximum projections of deconvoluted images (see Fig. 11D) were created from z series of eight 0.5-μm steps, using MetaMorph software 7.0 (MDS Analytical Technologies). Time-lapse imaging was performed using MM-N medium and the Axiovert 100 microscope essentially as described previously (62); the glass-bottomed dish was made by gluing a coverslip to the bottom of a Falcon 35-mm plastic dish through which a hole had been drilled.
Cells were grown on an EMM plate for 12 h at 30°C, transferred to an MM-N plate, incubated for 6 h at 30°C, scraped from the plate, mounted directly on a copper grid to form a thin layer, and plunged into liquid propane cooled with liquid nitrogen (−196°C) in a Leica EM CPC cryoworkstation (Leica Microsystems). The frozen cells were transferred to anhydrous acetone containing 2% OsO4 at −80°C in a Leica EM AFS automatic freeze-substitution apparatus, held at −80°C for 78 h, warmed gradually to 0°C over a period of 11.4 h, held at 0°C for 1.5 h, warmed gradually to 23°C (room temperature) over a period of 3.9 h, and held at 23°C for 2 h. After being washed three times with anhydrous acetone, samples were infiltrated with increasing concentrations of Spurr's resin in anhydrous acetone and, finally, with 100% Spurr's resin. After polymerization (5 h at 50°C plus 60 h at 60°C) in capsules, ultrathin sections were cut on a Leica Ultracut UCT microtome and stained with uranyl acetate and lead citrate. The sections were then viewed on a Hitachi H-7600 electron microscope at 100 kV.
Escherichia coli strain BL21 (Novagen) harboring plasmids encoding MBP fusion proteins was cultured at 15°C for 2 days, and the fusion proteins were purified by standard methods (43). A total of 2.5 μl of lipid solution containing 100 picomoles of a phosphoinositide (Wako Chemicals) dissolved in 1:1 chloroform-methanol was spotted onto a Hybond-C-Extra membrane (GE Healthcare) and dried at room temperature. The membrane was incubated with 3% fatty-acid-free bovine serum albumin (BSA) in TBS-T (10 mM Tris-HCl, pH 8.0, containing 150 mM NaCl and 0.1% Tween 20) at ~23°C for 1 h and then with TBS-T containing 0.5 μg/ml of the indicated protein overnight at 4°C. The membrane was then washed three times (10 min per wash) with TBS-T at ~23°C. Proteins interacting with the phosphoinositides were detected with an anti-MBP monoclonal antiserum (New England Biolabs) and a horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG antibody (BioSource International).
cDNAs of the full-length coding regions of spn2+ and spn5+ were cloned into the multicloning sites of the tandem-expression vector pET-Duet1 (Novagen) to achieve expression of untagged Spn2 and His6-tagged Spn5. Similarly, spn6+ and spn7+ sequences were cloned into pCOLA-Duet1 (Novagen) to create two plasmids; one of these expressed untagged Spn6 and untagged Spn7, whereas the other expressed His6-tagged Spn6 and untagged Spn7. To analyze complexes lacking one septin, similar plasmids were constructed that expressed a single septin. Pairs of plasmids were cotransformed into E. coli strain BL21(DE3) (Novagen). The transformants were grown in LB plus ampicillin and kanamycin at 37°C to an optical density at 600 nm (OD600) of 1.0, and expression was induced by adding isopropyl-β-d-thiogalactopyranoside for 16 h at 16°C. Preparation of high-speed supernatants and purification of the septin complexes were done exactly as described previously (36), in the presence of 0.5 M KCl. Polymerization of septin complexes was induced by diluting the solution in the same buffer without KCl.
To investigate septin function during mating and sporulation in S. pombe, we constructed septin mutants by deleting part of the septin ORF in a homothallic strain background (Fig. (Fig.1).1). As expected (see the introduction), all mutants were viable. When mating and sporulation were induced, a wild-type control strain formed mating tubes that fused to form zygotes; the initially narrow conjugation bridges then expanded to the full cell diameter, and four spores formed in most asci (Fig. 2A and B). Some apparent asci contained no visible spores, but very few asci (<5%) contained 1 to 3 spores (Fig. (Fig.2B).2B). The spn1Δ, spn3Δ, and spn4Δ mutants also formed mostly four-spored asci (Fig. 2A and B); taken together with the protein localization data (see below), these results suggest that these septins are not involved in spore formation. However, these mutants and the spn2Δ mutant showed a defect in the reshaping of the conjugation bridge, which remained narrow even after the spores had formed (Fig. (Fig.2A).2A). Consistent with the relative severities of the vegetative-cell defects in mutants lacking these septins (3; our unpublished data), the conjugation bridge defect was most severe in the spn1Δ and spn4Δ mutants and less so in the spn2Δ and spn3Δ mutants (Fig. (Fig.2A).2A). The failure to reshape the conjugation bridge was associated with a defect in actin organization at that site. In wild-type cells, F-actin localizes to cortical patches (presumably representing sites of endocytosis ) in the projection tip and conjugation bridge during cell fusion (64) (Fig. (Fig.3);3); these patches then disperse after karyogamy. In contrast, the F-actin patches in the spn1Δ, spn2Δ, spn3Δ, and spn4Δ mutants were dispersed before karyogamy (Fig. (Fig.3).3). Taken together with the protein localization data (see below), these mutant phenotypes suggest that the vegetative complex of Spn1, Spn2, Spn3, and Spn4 (3) continues to function during conjugation, perhaps to direct and/or anchor the exocytic and endocytic machinery to the conjugation site.
In contrast, the spn5Δ, spn6Δ, and spn7Δ mutants displayed no obvious defects in conjugation bridge enlargement (Fig. (Fig.2A)2A) or in actin patch localization to that site (data not shown), consistent with the evidence that these genes are not expressed until after conjugation bridge enlargement has occurred. However, these mutants, and to a lesser extent the spn2Δ mutant, showed a strong defect in spore formation in which the majority of asci contained fewer than four spores (Fig. 2A and B). Double mutants and even an spn5Δ spn6Δ spn7Δ triple mutant did not show a more severe defect (Fig. (Fig.2B;2B; data not shown), suggesting that Spn2, Spn5, Spn6, and Spn7 all function in the same process during sporulation. Homozygous spn2Δ, spn5Δ, spn6Δ, and spn7Δ diploids also formed many asci with fewer than four visible spores (data not shown) but had no obvious abnormality in meiotic nuclear division (Fig. (Fig.2C),2C), suggesting that the defect is indeed in the process of spore formation.
In wild-type cells, differential interference contrast (DIC) microscopy clearly showed the uniform, thick walls of the mature spores (Fig. (Fig.4A).4A). In contrast, in septin mutants, some presumptive spores did not appear to have mature walls (Fig. (Fig.4A,4A, arrows); these structures always lacked nuclei (Fig. (Fig.4A,4A, upper right panel), suggesting that engulfment of a nucleus is a prerequisite for spore maturation. The use of Spo3-GFP as a marker for the FSM revealed uniform cup-shaped or spherical FSMs, each of which engulfed a nucleus, in wild-type cells (53) (Fig. (Fig.4B,4B, top panels). In contrast, in septin mutants, the FSMs were frequently misshapen and failed to engulf a nucleus (Fig. (Fig.4B).4B). Time-lapse analysis using GFP-Psy1, another FSM marker (53, 55), revealed the defect in more detail (Fig. (Fig.4C;4C; see Movie S1 in the supplemental material). At the onset of FSM formation, although some FSMs appeared to behave normally, others extended in inappropriate directions and formed small spheres at an early stage (Fig. (Fig.4C,4C, arrows). These FSMs then continued to grow to seemingly normal size, but with aberrant shapes (Fig. (Fig.4C,4C, 20′ to 50′). In the spn5Δ strain, 15 of 36 FSMs observed (in nine independent asci) extended in such a disoriented manner. Similar time-lapse observations were made with spn6Δ (three asci) and spn5Δ spn6Δ (two asci) strains (data not shown).
Similar results were obtained by electron microscopy. In wild-type cells, as expected, the nuclei were always engulfed by the FSMs (Fig. (Fig.5B).5B). In contrast, the spn2Δ, spn5Δ, and spn6Δ mutants frequently left some nuclei bare (Fig. 5D, F, and G). Although the FSMs appeared to begin extension close to the SPBs in both wild-type and mutant cells (Fig. 5A, C, E, and F, arrowheads), in the mutant cells they frequently curved away from the nuclei, extended asymmetrically from the SPBs, or both (Fig. 5C, E, and F, arrows), and they eventually formed spore-like structures without nuclei (Fig. 5D and G, arrows).
Taken together, the results suggest that the septins are important for the proper orientation of FSM extension and that the formation of aberrant numbers of spores in septin mutants results from the reduced fidelity of this process.
To investigate how the septins regulate the direction of FSM extension, we examined septin localization during sporulation, using septins tagged at their C termini and expressed from their normal promoters, septins tagged at their N termini and expressed from the nmt41 promoter, or both. As expected (see the introduction), Spn1, Spn2, Spn3, and Spn4 all localized to a ring at the division site in vegetative cells (Fig. (Fig.6B,6B, top row; data not shown). In addition, these four proteins also localized to an apparent ring at the site of cell fusion during conjugation (Fig. (Fig.6B,6B, second row; data not shown); these rings disappeared by the time that expansion of the conjugation bridges was complete (data not shown). These localization data are consistent with the phenotypic data suggesting that the vegetative septin complex is involved in conjugation bridge expansion (see above).
In sporulating cells, we observed no localization of Spn1, Spn3, and Spn4 (Fig. (Fig.6A;6A; data not shown), consistent with the genetic evidence that these proteins are not involved in spore formation (see above). In contrast, Spn2, Spn5, Spn6, and Spn7 colocalized to partial or complete ring-like structures that curled around each haploid nucleus (Fig. (Fig.6B,6B, last four rows, and C). Localization of the septins to these structures was interdependent (Fig. (Fig.6D),6D), suggesting that in sporulating cells, as in vegetative cells, a septin complex (albeit one composed of a different set of septins) is the functional unit.
To test this hypothesis biochemically, we first tried to purify the putative septin complex from extracts of sporulating cells by using the tandem affinity purification (TAP) system. Surprisingly, however, we were unable to purify any of the TAP-tagged septin proteins, and indeed, these proteins did not appear to be soluble even in the presence of 0.5 M KCl (data not shown). These results differ from those obtained previously with many other septin complexes (3, 19, 20, 36). Indeed, the same Spn2-TAP protein could be purified successfully from vegetative cells (our unpublished results; also see reference 3), suggesting that the septin complexes in vegetative and sporulating cells have different properties.
To overcome this problem, we coexpressed the four septins in bacteria as described previously (11). When His6-Spn5 was coexpressed with Spn2, Spn6, and Spn7, complexes were purified that contained the four septins in nearly 1:1:1:1 stoichiometry (Fig. 7A and D, lane 1), indicating that these septins are indeed capable of forming a complex. These complexes appeared to polymerize when the KCl concentration was decreased (Fig. 7B and C), suggesting that septin function during sporulation involves the assembly of higher-order structures and that Spn2, Spn5, Spn6, and Spn7 are sufficient to form a functional complex that is capable of such assembly.
When Spn2 was omitted from the expression system, Spn5, Spn6, and Spn7 still formed a stoichiometric complex (Fig. (Fig.7D,7D, lane 2), and when Spn5 was omitted, Spn6 and Spn7 still formed a stoichiometric complex (Fig. (Fig.7D,7D, lane 3). These data suggest that the sporulation septins may form a linear core tetramer of Spn2-Spn5-(Spn6, Spn7), similar to the linear complexes proposed to form from the septins expressed in vegetative cells (3, 11, 47). The similarity extends even to the presence of Spn2 at an end of the tetramer, where its homotypic association could form the octamers that would in turn be the subunits for assembly of longer filaments (11, 47). However, a problem for this model is that when either Spn6 or Spn7 was omitted from the expression system, the other was also lost from the complexes, which contained only Spn2 associated with His6-Spn5 (Fig. (Fig.7D,7D, lanes 4 and 5). A possible explanation could be that an interaction between Spn6 and Spn7 is necessary for one (or conceivably both) of them to bind to Spn5.
Time-lapse observations revealed that the septin structures, like the FSMs, extended from the vicinity of the SPBs to surround the nuclei (Fig. (Fig.8A;8A; see Movie S2 in the supplemental material), and double labeling of a septin and a bona fide FSM marker confirmed the close association of the septin structure with the FSM throughout FSM formation (Fig. (Fig.8B).8B). Surprisingly, however, the double labeling also revealed that the septin structure was not coterminous with the FSM at any stage (Fig. (Fig.8B,8B, panels a to d). Examination of serial z sections (Fig. 9A and B; see Movies S3 to S5 in the supplemental material) and of rotated images (Fig. (Fig.9C;9C; see Movies S6 and S7 in the supplemental material) after three-dimensional deconvolution indicated that whereas the FSM itself is a cup-shaped structure that eventually forms a sphere enclosing the haploid nucleus, the septins appear to form a horseshoe-shaped structure that eventually becomes a ring (compare the images in Fig. Fig.99 to those in Fig. 8A and B, panels c and d; see Fig. 13A). In some cases, there appeared to be an additional projection of labeled septin material from the vicinity of the SPB (Fig. (Fig.9C,9C, arrows), suggesting that the nucleation of septin assemblies in this region might sometimes be more complex. However, these structures were short and observed around <5% of the SPBs, and we were unable to characterize them further.
The abnormalities in FSM formation seen in septin mutants (see above) are similar to those seen in mutants lacking the LEP Meu14 (60) (Fig. 10A). This suggested that the septins might function through the LEPs or vice versa. However, the localization of the septins and LEPs appeared to be independent of each other: a labeled septin was consistently associated with the FSMs in a meu14Δ strain, even when the FSMs were abnormal (Fig. 10A; see Fig. 13C), and the localization of Meu14 in septin mutants seemed as normal as could be expected given the abnormalities of FSM structure in these mutants (Fig. 10B). Moreover, the spore formation defect of each spnΔ meu14Δ double mutant was significantly more severe than that of either single-mutant parent (Table (Table3).3). Taken together, these data suggest that the septins and the LEPs contribute to FSM morphogenesis in parallel pathways rather than a linear pathway.
Mammalian SEPT4 and S. cerevisiae Cdc3, Cdc10, Cdc11, and Cdc12 have all been reported to bind directly to phosphoinositides in vitro via the polybasic regions near their N termini (14, 83). To ask if the S. pombe septins might behave similarly, we first examined their sequences. Unambiguous polybasic regions were found near the N termini of five of the seven proteins; in Spn1, Spn2, Spn4, and perhaps Spn3, the sequences were similar to those in the S. cerevisiae orthologues (Fig. 11A). We then used purified MBP-septin fusion proteins and a protein-lipid overlay assay to investigate phosphoinositide binding by the septins that are involved in sporulation. Spn2 and Spn7 bound to PtdIns(4)P and PtdIns(5)P but did not bind convincingly to the other phosphoinositides examined, whereas the less-basic septins Spn5 and Spn6 showed no detectable binding to any of the compounds tested (Fig. 11B). If the four sporulation septins indeed form a complex in vivo, then binding of two septins to phosphoinositides may be sufficient for the entire complex to associate with a membrane.
Because PtdIns(5)P has not been detected in S. pombe (49), we postulated that Spn2 and/or Spn7 might bind to PtdIns(4)P in the FSM. To visualize PtdIns(4)P, we used the GFP-PHOsh2 dimer (see Materials and Methods) as a fluorescent probe; in S. cerevisiae, this probe reveals both the Golgi and plasma membrane pools of PtdIns(4)P (66). In vegetative S. pombe cells, this probe gave a weak signal at the plasma membrane and stronger signals at newly formed septa and at intracellular foci that presumably represent the Golgi apparatus (Fig. 11C, panel a). As expected, these signals were not affected by deletion of the sole S. pombe PtdIns 3-kinase gene, pik3+/vps34+ (data not shown), whereas the presumed Golgi signal disappeared in a mutant defective in the putative Golgi-targeted PtdIns 4-kinase Pik1 (Fig. 11C, panel d; also see below), suggesting that the localization of the GFP-PHOsh2 dimer indeed reflects the distribution of PtdIns(4)P. During sporulation in wild-type cells, the GFP-PHOsh2 dimer was enriched at FSMs, while the signals at the plasma membrane and Golgi apparatus declined or disappeared (Fig. 11C, panels b and c), suggesting that PtdIns(4)P is indeed enriched in the FSMs during sporulation. Because the GFP-PHOsh2 dimer and Spn2-RFP signals were not coterminous (Fig. 11D), PtdIns(4)P does not appear to be enriched specifically in the zone of contact between the FSM and the septin structure but is instead distributed throughout the FSM; thus, the distribution of PtdIns(4)P cannot be solely responsible for determining the architecture of the septin structure (or vice versa).
To ask if PtdIns(4)P in the FSM is important for septin binding, we wished to mutate the relevant PtdIns kinase(s). S. cerevisiae has two type III PtdIns 4-kinases, Pik1 and Stt4, which synthesize PtdIns(4)P in the Golgi apparatus and plasma membrane, respectively (4, 5, 28), and the S. pombe genome encodes apparent orthologues of both enzymes. Although pik1 null mutations are lethal in S. cerevisiae (4), the S. pombe pik1 mutation that we constructed (Fig. (Fig.1)1) was nonlethal, although vegetative cells of the mutant contained only ~60 to 70% as much total PtdIns(4)P as wild-type cells, with apparently a much more severe reduction in the level of PtdIns(4)P in the Golgi apparatus (Fig. 11C, panel d; our unpublished data). Given the apparent lack of PtdIns(4)P in the Golgi apparatus and the construction of the FSM from Golgi complex-derived vesicles (see the introduction), we were surprised that staining of sporulating pik1 mutant cells with the GFP-PHOsh2 dimer revealed FSM-like structures containing PtdIns(4)P (Fig. 11C, panels e and f). However, these structures were clearly smaller than normal FSMs and were usually abnormal in shape (Fig. 11C, panels e and f), and essentially no recognizable spores were formed (none seen in >200 asci examined [our unpublished data]). Although Spo3 failed to localize to these abnormal FSM-like structures (and was apparently retained in the Golgi apparatus [data not shown]), Psy1 did appear to localize to them (Fig. 11F), and examination of this marker revealed that the FSM-like structures frequently failed to enclose the adjacent nuclei. This phenotype resembles that of the septin mutants, except for the smaller size of the FSM-like structures in the pik1 mutant (which presumably reflects the absence of Spo3 and/or inefficient vesicle transport from the Golgi apparatus to the FSM). Remarkably, the septins did associate with the abnormal FSM-like structures, beginning in what appeared to be a normal location near the poles of the meiosis II spindles (Fig. 11E, panel a, and F, panel a), and thus also formed structures that failed to enclose the adjacent nuclei (Fig. 11E, panels b and c, and F, panels b and c; also see Fig. 13C). These results suggest that the PtdIns(4)P content of the FSM-like membranes is sufficient for septin association but that septin association alone is not sufficient for the normal oriented growth of the FSMs. Attempts to generate strains in which the PtdIns(4)P content of the FSM is more severely reduced have thus far been blocked by the lethality of stt4 knockout strains (data not shown) and the unavailability as yet of a conditional stt4 mutant.
As another approach to assessing the importance of septin-phosphoinositide binding for FSM organization, we substituted glutamines for the four conserved basic amino acids in Spn2. The resulting protein, Spn24Q, showed only very weak PtdIns(4)P binding in the protein-lipid overlay assay (Fig. 12A). When expressed in spn2Δ cells, Spn24Q-GFP was found in structures that began to assemble near the spindle poles (Fig. 12B, panel a) and continued to form at least nascent rings (Fig. 12B, panels b and c); thus, it appears that the mutant protein can function as a subunit of the septin complex and that binding to phosphoinositides is dispensable for septin assembly. However, spn24Q did not complement the sporulation defect of spn2Δ cells (Fig. 12C), suggesting that a weakened interaction between the septin assemblage and the FSM resulted in a failure to encapsulate nuclei. Indeed, when we examined the spatial relationship between Spn24Q and the FSM (marked by Spo3) in spn2Δ cells, we found that the FSMs and septin assemblages often extended independently of each other and that neither structure regularly encapsulated the nuclei (Fig. 12D and E and 13D). These results suggest (i) that the septin assemblage can form an extended higher-order structure even when not associated with the FSM, (ii) that phosphoinositide binding by Spn2 is important for septin-FSM association, and (iii) that this association is important for both the septins and the FSMs to extend in a properly oriented manner.
During formation of the yeast spore envelope, the initially small FSM/PSM sac extends in an oriented fashion so as to enclose a nucleus, other organelles, and cytoplasm in a double sphere of membrane; the spore wall then forms between the inner and outer membranes. Earlier work has shown that FSM/PSM extension involves the modified cytoplasmic face of the SPB; the proteins involved in the formation, docking, and fusion of post-Golgi vesicles; phosphoinositide-mediated membrane trafficking; and the LEP complex at the leading edge of the extending membrane sac (see the introduction). However, it has remained unclear whether these mechanisms are sufficient to explain the precise orientation of FSM/PSM extension and its eventual closure to form nascent spores of uniform size. Indeed, studies of S. cerevisiae have indicated that the septins are also involved, although their precise role has been difficult to elucidate because of the inviability of some septin mutants in this species and the relatively mild sporulation phenotypes of the viable mutants. In this study, we have gained deeper insight into the role of septins in spore formation through studies of S. pombe, in which all septin mutants are viable and some have relatively strong sporulation phenotypes. The results also extend our understanding of how septins assemble and associate with membranes and provide insight into the mechanisms by which intracellular membrane systems can acquire specific shapes and three-dimensional organizations.
In vegetative S. pombe cells, Spn1, Spn2, Spn3, and Spn4 form a complex at the division site that seems orthologous to the S. cerevisiae complex of Cdc3, Cdc10, Cdc11, and Cdc12. spn1Δ and spn4Δ mutants have the strongest cell division phenotypes, suggesting that Spn1 and Spn4 are central to complex formation (3, 47). This complex apparently continues to function during conjugation and zygote formation. We found that Spn1, Spn2, Spn3, and Spn4 all localized to an apparent ring at the site of cell fusion and that in the absence of any of these septins, the narrow conjugation bridge failed to expand as it does in wild-type cells. As in vegetative cells, the defects in spn1Δ and spn4Δ mutants were significantly more severe than those in spn2Δ and spn3Δ mutants. Since all four septins are presumably involved in the formation of octamers and higher-order structures (3, 47), it is unclear why some mutants would have weaker phenotypes than others, although it may reflect an ability of septin complexes to form partially functional filaments even when lacking certain subunits (11, 47). Based on the premature dispersal of actin patches from the conjugation bridge in the septin mutants, it seems likely that the septin complex serves to anchor components that are involved in the reshaping of the cell membrane and/or cell wall at that site. Since the actin-nucleating formin Fus1 also localizes to the conjugation bridge (39, 65), it would be interesting to examine its possible interactions with the septins.
Remarkably, in sporulating cells, Spn1, Spn3, and Spn4 are expressed at low levels (45), fail to localize, or both, and spn1Δ, spn3Δ, and spn4Δ mutations have no detectable effect on sporulation efficiency. Instead, efficient spore formation depends on an alternative set of septins, Spn2, Spn5, Spn6, and Spn7, which can form a complex in vitro and assemble interdependently in vivo into a structure associated with the FSM. In the absence of the septins, the FSM appears to begin assembly and extension normally, but it fails, in ~50% of cases, to extend in a properly oriented fashion and to envelop the associated nucleus (Fig. 13B). Interestingly, although we could not detect localization of the other septins in the spn2Δ mutant, its spore formation defect was less severe than those of the spn5Δ, spn6Δ, and spn7Δ mutants, suggesting that like the case in vegetative and conjugating cells, Spn2 might be less critical than other septins to the formation of a functional septin complex. Indeed, Spn5, Spn6, and Spn7 still formed a complex in vitro in the absence of Spn2. Another interesting feature of the sporulation septin complex is that Spn7 appears to be an integral component despite its lack of features that characterize most septins, namely, a well-defined GTP-binding site (which in other septins appears to be critical for septin-septin interaction [47, 52, 69]) and a C-terminal coiled-coil domain.
Also remarkable is the overall form of the septin structure. Rather than a cup-shaped structure (like the FSM/PSM) or a ring associated with the lip of the FSM/PSM cup (like the LEP complex), the septins appear to form a curved, two-dimensional bar (a “horseshoe”) whose midpoint is at or near the SPB (Fig. 13A). As the FSM extends and closes to form the spherical prespores, the septin structure appears to extend and close to form an equatorial ring. It seems possible that an intrinsic propensity of septin assemblies to form curved bars that can become rings or spirals (18, 21, 36, 47, 80) has been exploited in this system to help ensure appropriate curvature of the extending FSM. Although it is not yet clear whether the septin structure lies on the inner or outer face of the FSM, the former possibility seems more likely because, in S. cerevisiae, tagged molecules of Cdc10 that were found in the PSM-associated structures could later be found at the bud necks in germinating spores (48), implying that they were present in the spore cytoplasm after PSM closure. This model predicts that the septin filaments are nucleated at or near the SPB and therefore are likely to constitute two (or occasionally more) (Fig. (Fig.9C)9C) distinct bars rather than a continuous horseshoe; in this case, septin organization along the FSM in S. pombe might be more similar to that thought to exist along the PSM in S. cerevisiae (59) than is apparent from the images available to date. However, we were unable to detect a gap in the septin staining near the SPB by fluorescence microscopy, and analysis at a higher resolution (e.g., by electron microscopy) will presumably be needed to resolve this issue.
Further analysis of the sporulation septins should contribute to a general understanding of the organization and assembly of septin complexes, particularly if the sporulation complexes are distinct from those in vegetative cells, as suggested both by Neiman (59) and by our biochemical data (insolubility of the complexes in sporulating cells and interdependent binding of Spn6 and Spn7 in vitro). It should also be instructive to explore further the apparent differences between the septin structures observed in S. pombe and those seen in S. cerevisiae (17, 37, 51, 59, 71).
Another interesting question is why properly oriented FSM extension and nuclear envelopment fail only about half of the time in the septin mutants. Similarly, it is remarkable that at least some of the FSMs that do not have contact with a septin structure and do not surround a nuclear envelope nonetheless appear to close to form quasispherical structures. These observations suggest that another system, such as the LEP complex, is also capable of guiding FSM extension and closure but is not robust in function in the absence of the septin structure. Since the available data suggest that the septins and the LEPs function in parallel (59, 71; see Results), it is possible that a differential ability of the LEPs to function in the absence of septins may account for the differential severities of septin mutant sporulation phenotypes in S. cerevisiae (15, 17, 59) and S. pombe (this study).
Although many of the biological roles of the septins involve a close association with membranes (12, 27, 41), the mechanisms of septin-membrane association have remained poorly understood. It has been reported that both mammalian and S. cerevisiae septins can bind specific phosphoinositides via clusters of basic amino acids near their N termini (14, 83), suggesting that septins might associate with membranes by direct interaction with the membrane phospholipids (47). Our results support and significantly extend this model. First, we found that Spn2 and Spn7, which have well-defined basic clusters, bound phosphoinositides in vitro, whereas Spn5 and Spn6, which lack such clusters, did not (at least with the assay used here). Second, phosphoinositide binding by Spn2 was essentially eliminated by replacing its cluster of basic amino acids by uncharged residues (Spn24Q). Third, we detected binding specifically to PtdIns(4)P and PtdIns(5)P but not to other phosphoinositides and showed that PtdIns(4)P was highly enriched in the FSM and thus could potentially account for the association of the septins with that structure. Fourth, although the pik1Δ117-198 mutation reduced the total cellular PtdIns(4)P level and produced gross alterations of FSM development, it did not greatly reduce the PtdIns(4) content of the FSM, and the septins still associated with the aberrant FSMs (Fig. 13C). Finally, although spn2Δ cells expressing Spn24Q could at least assemble nascent septin horseshoe/ring structures, these structures did not consistently associate with the FSMs, and neither the septin structures nor the FSMs efficiently enclosed the adjacent nuclei (Fig. 13D). Thus, the data suggest (i) that the basic amino acid cluster is critical for septin-phosphoinositide binding in vivo as well as in vitro, (ii) that such binding is critical for septin-membrane association, and (iii) that such association is critical for both the septin structure and the FSM to extend with a normal orientation (although each structure is capable of extension in the absence of association with the other).
Three additional points should be noted. First, we cannot entirely rule out the possibility that the Spn24Q protein has a subtly altered structure that in turn alters the structure of the septin complex and in this way blocks its association with the membrane. However, this seems unlikely given that Spn24Q appears fully competent to assemble into extended septin structures and that the septin-FSM association otherwise seems quite robust (as evidenced by its persistence even when the FSM is abnormal in meu14Δ and pik1 mutants). Second, although all of our data, like those of Casamayor and Snyder (14), are consistent with PtdIns(4)P being the phosphoinositide to which the septins bind in yeast membranes, it is not clear that the in vitro phospholipid-binding assays faithfully reflect binding in vivo. Thus, it remains possible, as McMurray and Thorner (47) have suggested, that binding in vivo is actually to a different phospholipid, such as PtdIns(4,5)P2. Indeed, localization data obtained using a different fluorescent marker (GFP-PHPLCδ) suggest that PtdIns(4,5)P2 is also enriched in the FSM (our unpublished observations). Finding ways to selectively reduce PtdIns(4)P or PtdIns(4,5)P2 in the FSM would allow us to address this question. Third, although the pik1Δ117-198 cells displayed substantial amounts of PtdIns(4)P in the FSM and septin association with that structure, their defect in FSM development and spore formation was actually more severe than that of the septin deletion and spn24Q mutants. Thus, Pik1-produced PtdIns(4)P must be important in some other way for FSM extension. Because the presumed Golgi PtdIns(4)P signal was lost in the pik1Δ117-198 mutant (see Results) and the aberrant FSM structures were consistently smaller in this mutant than in the septin mutants (our unpublished results), it seems likely that at least part of the role of Pik1-produced PtdIns(4)P is in the delivery of membrane from the Golgi apparatus to the developing FSM. However, further studies will clearly be necessary to clarify the role of Pik1.
We thank Y. Watanabe (University of Tokyo), H. Nojima (Osaka University), and T. Levine (University College London) for plasmids and strains and Sadayuki Yamada of the Y. Fukui laboratory for help with strain construction. Some reagents were provided via Japan's National BioResource Project—Yeast.
This work was supported by a postdoctoral fellowship from the Uehara Memorial Foundation (to M.O.), by grants-in-aid from the Ministry of Education, Science, Sports, and Culture of Japan (to Y.F.), and by National Institutes of Health grant GM-31006 (to J.R.P.).
Published ahead of print on 1 February 2010.
†Supplemental material for this article may be found at http://mcb.asm.org/.