An auxiliary, membrane-based mechanism for nuclear migration in budding yeast.

How nuclear shape correlates with nuclear movements during the cell cycle is poorly understood. We investigated changes in nuclear morphology during nuclear migration in budding yeast. In preanaphase cells, nuclear protrusions (nucleopodia [NP]) extend into the bud, preceding insertion of chromosomes into the bud neck. Surprisingly, formation of nucleopodia did not depend on the established nuclear migration pathways. We show that generation and maintenance of NP requires nuclear membrane expansion, actin, and the exocyst complex. Exocyst mutations cause nuclear positioning defects and display genetic interactions with mutations that deactivate astral microtubule-dependent nuclear migration. Cells that cannot perform DNA replication also fail to form nucleopodia. We propose that nuclear membrane expansion, DNA replication, and exocyst-dependent anchoring of the nuclear envelope to the bud affect nuclear morphology and facilitate correct positioning of nucleus and chromosomes relative to the cleavage apparatus.

INTRODUCTION

Spatial coordination of cell cleavage with chromosome segregation ensures that the cleavage apparatus bisects the partitioned chromosomes and the nucleus after anaphase. To achieve this in Saccharomyces cerevisiae, the nucleus, the intranuclear mitotic spindle, and the chromosomes migrate toward the bud and insert into the bud neck, the future site of cytokinesis. Two conserved microtubule-dependent pathways facilitate nuclear/spindle migration in budding yeast (Miller ; Moore and Cooper, 2010). The Kar9 pathway mediates migration of the nucleus–spindle–chromosome system toward the bud neck and orients the mitotic spindle with the old spindle pole facing the bud (Pereira ; Liakopoulos ). Insertion of the nucleus and the spindle into the bud neck occurs with the help of dynein-mediated spindle oscillations, during which the entire spindle traverses the bud neck (Saunders ; Yeh ).

During spindle elongation in anaphase, the anaphase spindle stretches and partitions the nuclear envelope between mother and bud, resulting in generation of two nuclei at the end of mitosis. Nuclear duplication requires that both the nuclear membrane and the perinuclear endoplasmic reticulum (ER) increase in mass during the cell cycle (Jorgensen ). Growth of the nuclear membrane depends on the coordinated action of the diacylglycerol kinase Dgk1 and the mammalian lipin homologue, the phosphatidate phosphatase Pah1 (Siniossoglou, 2009). These enzymes regulate the production of phosphatidic acid, which acts as a positive signal for nuclear membrane growth (Loewen ). Nuclear membrane expansion is induced upon phosphorylation and deactivation of Pah1 by multiple kinases, including Cdk1/cyclin B, whereas dephosphorylation of the enzyme by the heterodimeric phosphatase Spo7-Nem1 activates Pah1 and inhibits growth of the nuclear membrane (Santos-Rosa ; O'Hara ; Karanasios ; Choi ; Su ).

Of interest, nuclear membrane growth and nuclear shape are linked (Zhang and Oliferenko, 2012). Nuclear membrane overexpansion leads to deformation of the nuclear envelope (Han ), and membrane/vesicle trafficking maintains nuclear shape during proliferation of the nuclear membrane (Webster ). The DNA in the nucleus seems to also participate in the regulation of nuclear shape. Expansion of the nuclear membrane takes place at the area of the nuclear envelope next to the nucleolus (Campbell ; Witkin ). In contrast, the nuclear envelope associated with the DNA mass seems to resist nuclear membrane expansion. Thus association of chromosomes with the nuclear envelope may contribute to preservation of spherical nuclear shape.

Nuclear migration into the bud is accompanied by dramatic changes in nuclear morphology. In preanaphase, the nuclear envelope loses its spherical shape and forms protrusions that extend into the bud and lack DNA (Fehrenbacher ). These poorly characterized structures exhibit sweeping motions and contact the cell cortex. Of importance, these nuclear protrusions form before anaphase and extend into the bud while the metaphase spindle is still in the mother cell. It is not clear what causes the nucleus to migrate into the bud ahead of the mitotic spindle and how the nuclear protrusions are formed and maintained.

RESULTS

Nuclear migration into the bud can occur independently of Kar9 and dynein

In preanaphase, the nucleus of budding yeast obtains a dumbbell morphology (Palmer ). This occurs because the mitotic spindle and the nucleus are pulled by astral microtubules into the bud neck. The bud neck constricts the nucleus, making it appear like a dumbbell (Figure 1A). Surprisingly, we and others observed that some preanaphase cells generated dynamic nuclear protrusions that grew into the bud ahead of the mitotic spindle, before dumbbell formation (Figure 1B and Supplemental Videos S1 and S2; Fehrenbacher ). We named these protrusions nucleopodia (NP). Nucleopodia have been suggested to contribute to spindle alignment and nuclear migration (Fehrenbacher ). We verified that NP contained little, if any, DNA (Figure 1B) and that they were indeed protrusions of the nucleus, since they contained nucleoporins (Nup1–green fluorescent protein [GFP]), nucleoplasm (visualized with tetR-NLS-GFP), and the nucleolar protein Nop1–cyan fluorescent protein [CFP] (Figures 1, B and D, and 2A, and Supplemental Videos S1 and S2). Nucleopodia formed in full or selective media at 30, 34, and 37°C.

FIGURE 1:

The nuclear envelope forms protrusions (NP) into the bud before anaphase. (A) Dumbbell nuclei in cells arrested in preanaphase by Cdc20 depletion (P cells in glucose). DAPI staining of DNA. (B) Images of living cells showing the relative positions of the nuclear envelope (marked with the nucleoporin fusion Nup1-GFP), the mitotic spindle, and the DNA (visualized with the histone H2B fusion HTB1-CFP). Top, a preanaphase-arrested P cell displaying NP (arrow) and two cells in the “dumbbell” stage. Bottom, two cells with nucleopodia (arrows). In all cases, the mass of the DNA is associated with the spindle. In dumbbell nuclei, spindle and DNA are inserted in the bud neck, whereas NP enter the bud while the spindle and the bulk of the DNA are still in the mother cell. (C) Quantification of the three different stages of nuclear migration of P cells for different time points after release from G1 (α-factor, black; nucleus in mother, magenta; NP, white; dumbbells). Cells were released into glucose-containing medium to arrest them in preanaphase (counted >100 cells/time point, N > 500/time course; three to five independent experiments were performed). Bars, SEM. (D) Image sequence showing generation of NP. P cells were released from G1 (α-factor) in medium containing glucose. Migration of the nuclear envelope and the mitotic spindle was monitored at different time points. Arrows indicate NP. The NP is visible until 13.5 min, when the spindle is pulled into the bud neck (at this stage a dumbbell nucleus is generated).

FIGURE 2:

Formation of nucleopodia does not depend on the dynein and Kar9 nuclear migration pathways. (A) Image sequence showing generation of NP during preanaphase arrest of a dyn1∆ P cell expressing the nucleoplasmic marker tetR-NLS-GFP. Arrows indicate NP. Note that the spindle does not migrate into the bud neck in dyn1∆ cells, and no dumbbell nucleus is generated; thus NP can be easily identified. (B) Formation of NP can occur in the absence of the KAR9 and DYN1 nuclear migration pathways. Quantification of nuclear migration stages during preanaphase arrest of dyn1∆ and kar9∆ cells. Bud growth was also monitored (for bud size measurements see Supplemental Figure S1B). Bars, SEM. (C) A NP has formed in a dyn1∆kar9∆ mutant cell. Note that anaphase has taken place entirely inside the mother cell (these cells arrest at anaphase due to the spindle position checkpoint).

We investigated NP formation during the cell cycle more closely. After release of Nup1-GFP, mCherry-Tub1–expressing cells from G1, 10% of the cells formed NP, and anaphase followed immediately after (Supplemental Figure S1A and Supplemental Video S3; Tub1 is the yeast α-tubulin). To conveniently observe NP formation, we prevented cells from undergoing anaphase after release from G1. For this we placed the chromosomal copy of the CDC20 gene under the control of the glucose-repressible GAL1-10 promoter (P cells). Addition of glucose in the medium results in G2/M arrest of the P cells due to depletion of Cdc20, an activator of the anaphase-promoting complex. When we released P cells expressing Nup1-GFP and mCherry-Tub1 from G1 (α-factor) in glucose-containing medium, nearly half of the cell population formed NP (Figure 1, C and D; for this and all following quantifications, N > 100 cells/time point were counted, N > 500 cells/time course, and three to five independent experiments were performed). Time-lapse microscopy showed that NP form after one pole of the mitotic spindle is pulled transiently through the bud neck (Figure 1D), in agreement with Fehrenbacher ). Subsequently the spindle pole is pulled back into the mother cell, leaving part of the nuclear membrane inside the bud. There the nuclear membrane forms dynamic protrusions (see also Supplemental Video S3). Eventually, the mitotic spindle and the associated DNA mass are pulled into the bud neck, forming a dumbbell nucleus.

Spindle insertion into the bud and oscillating spindle movements depend on the activity of cytoplasmic dynein, Dyn1 (Yeh ). Surprisingly, dyn1∆ P cells did not display significant defects in NP formation; Figure 2, A and B, and Supplemental Figure S1B), whereas insertion of the spindle into the bud (dumbbell nuclei) was severely reduced, as expected. Formation of NP was largely unaffected in kar9∆ cells (Figure 2B and Supplemental Figure S1B). We also occasionally observed NP in dyn1∆ kar9∆ cells that lack both pathways and display an extremely slow growth phenotype (Figure 2C; Miller ). Depolymerization of the microtubule cytoskeleton inhibited NP generation (Figure 3A). However, disruption of microtubules did not affect already formed NP, indicating that the microtubule cytoskeleton is not essential for maintenance of NP (Figure 3B). These results suggest that dynein-dependent nuclear oscillations may facilitate NP formation but are not essential for it. Thus the nucleus is able to protrude into the bud by a dynein- and Kar9-independent mechanism.

FIGURE 3:

Microtubules are required for formation but not for maintenance of nucleopodia. (A) NP nearly fail to form in absence of microtubules. Cells released from G1 (α-factor) into medium containing nocodazole (NZ) to disrupt microtubules. (B) Maintenance of NP does not require microtubules. P cells were released from α-factor into glucose-containing media for 90 min to form NP, followed by nocodazole or dimethyl sulfoxide (DMSO; mock) addition. Lengths of NP are shown. Bars, SEM.

Protrusion of the nucleus into the bud requires nuclear membrane growth

Nuclear membrane expansion relies on phospholipid biosynthesis and is up-regulated by mitotic Cdk1 complexes (Santos-Rosa ). We reasoned that NP might form after entry into the mitotic cell cycle by growth of the nuclear membrane into the bud. To investigate this possibility, we generated cells that had a large bud but had not entered mitosis. CDC34 encodes a ubiquitin-conjugating enzyme required for the progression from G1 to the S phase of the cell cycle. Shift of temperature-sensitive cdc34-2 cells to the restrictive temperature prevents cells from entering S phase but does not affect bud growth, which occurs normally in late G1. G1-arrested cdc34-2 cells did not form NP, even after growing very large buds (Figure 4A). Nuclear volume was reduced in these cells compared with preanaphase-arrested P cells at the same temperature (cdc34-2: 6 ± 1.6 vs. 11 ± 0.7 μm3), suggesting that the nuclear membrane expands only after cells enter mitosis.

FIGURE 4:

Formation of nucleopodia requires entry into the cell cycle and nuclear membrane growth. (A) G1-arrested cdc34-2 cells with large buds do not form NP. In these cells neither replication nor mitotic Cdk1 activation takes place, the nucleus does not grow (see the text), and NP do not form. (B) NP were not observed after addition of the fatty acid synthase inhibitor cerulenin or the acyl-CoA-carboxylase inhibitor soraphen A (ethanol was used as solvent for both drugs). See also Supplemental Figure S1, C, D, and E. Bars, SEM. (C) Inhibition of nuclear membrane growth by co-overexpression of the Nem1-Spo7 phosphatase subunits (from a plasmid, under the control of the GAL1-10 promoter) abrogates growth of NP in G2/M-arrested P cells (CDC20 was under the control of the methionine-repressible MET25 promoter in this experiment; cells were G2/M arrested after addition of methionine to the medium after release from α-factor). For spindle lengths see Supplemental Figure S1F.

We next investigated whether inhibition of nuclear membrane growth would abrogate NP formation. Growth of the nuclear membrane depends on fatty acid synthesis (Schneiter ) and requires the activity of fatty acid synthase. First, we used cerulenin, an inhibitor of fatty acid synthase that has been shown to inhibit growth of the nuclear membrane in Schizosaccharomyces pombe and Schizosaccharomyces japonicus (Yam ). Treatment of cells with 10 μM cerulenin abrogated NP formation (Figure 4B) without affecting cell cycle progression, as cerulenin-treated cells exhibited bud sizes and spindle lengths comparable to those of solvent-treated cells (Supplemental Figure S1, C and E). Moreover, the effect was not due to cell death because cell viability was not severely reduced during the course of the treatment (Supplemental Figure S1D). We also blocked lipid synthesis by inhibiting the yeast acetyl-CoA-carboxylase Acc1. For this we used the specific inhibitor of Acc1, soraphen A (Vahlensieck ). Similar to cerulenin, soraphen A inhibited formation of NP without severely affecting cell growth (Figure 4B and Supplemental Figure S1E). Finally, we used genetic means to inhibit nuclear membrane growth and overexpressed the Nem1-Spo7 phosphatase (Santos-Rosa ). Formation of nucleopodia was severely reduced in Nem1-Spo7–overexpressing cells (Figure 4C). Again, cell cycle progression to G2/M was not affected in Nem1-Spo7–overexpressing cells, since spindle length was similar to control cells (Supplemental Figure S1F). Taken together, these data indicate that protrusion of the nuclear envelope into the bud requires nuclear membrane growth in mitosis.

Nucleopodia formation and maintenance depend on the actin cytoskeleton and the exocyst complex

We next asked whether disruption of the actin cytoskeleton would affect NP formation. In this experiment we used P cells to avoid dumbbell formation and observe only nucleopodia. We depolymerized actin in cells that had not yet formed any NP (Figure 5A) by treating cells with latrunculin A (latA) 60 min after release from G1 in glucose medium. We did not observe any further increase in the percentage of cells with NP at later time points. When latA was added after cells had formed NP, the latter shrunk, became rounded, and lost their contact to the bud cortex (Figure 5, B and C). We concluded that both NP formation and maintenance require an intact actin cytoskeleton.

FIGURE 5:

Actin is required for nucleopodia formation and maintenance. (A) The actin cytoskeleton is required for NP formation. Quantification of NP formation after addition of latA. To avoid confusion between NP and dumbbells, experiments were performed in a dyn1∆ background. DMSO was used as a solvent for latA. (B) Maintenance of NP requires an intact actin cytoskeleton. P cells were released from α-factor into glucose-containing medium for 90 min to form NP, followed by latA or DMSO (mock) addition. Lengths of NP are shown. (C) Images showing NP before and after addition of latA. Bars, SEM; ***p < 10−3 in a Student's t test.

The cortical ER depends on the actin cytoskeleton for proper inheritance and cortical anchoring (Prinz ; Fehrenbacher ; Wiederkehr ). Therefore we reasoned that defects in generation of NP upon actin depolymerization could reflect defects caused to cortical ER anchoring, to which the nuclear envelope is connected (Preuss ; Fehrenbacher ; West ). To test this idea, we examined generation of NP in a series of P strains harboring mutations that impair ER inheritance (Du ). Neither myo4∆ cells nor swa2∆ cells (Du ; Estrada ) showed any severe defect in NP formation (Supplemental Figure S2A). In contrast, 105 min after release from G1, NP were almost absent when we shifted exocyst mutant sec3-2 cells to the restrictive temperature, whereas >50% of wild-type cells with comparable bud size had formed NP at this temperature (Figure 6, A and B, and Supplemental Figure S2B). Cells with a temperature-sensitive mutation in Sec4, a GTPase that mediates exocytosis through the exocyst complex (Guo ), also displayed a severe defect in NP formation (Figure 6C). Thus a functional exocyst is required for NP generation.

FIGURE 6:

The exocyst is required for nucleopodia formation and maintenance. (A) Generation of NP depends on a functional exocyst complex. Quantification of NP formation during preanaphase arrest of P mutants at permissive (28°C) and restrictive temperature (34°C). Control quantification at 34°C and bud sizes are shown in Supplemental Figure S2, A and B, respectively. (B) Images showing lack of NP in arrested P cells at the restrictive temperature. Arrows show NP in cells grown at 28°C. (C) Quantification of NP formation during preanaphase arrest of P cells. (D) The exocyst is required for maintenance of NP. The quantification shows reduction of the number of cells with NP after shift of arrested P cells to the restrictive temperature. (E) Shrinkage of NP (arrow) after shift of a P cell to the restrictive temperature. Note that the spindle–nucleus falls back into the mother cell after NP shrinkage. The gray dashed line is drawn for comparison of the spindle position. Bars, SEM; **p < 10−2 in a Student's t test.

The exocyst was also required for NP maintenance. When P with NP were shifted to the restrictive temperature, we observed gradual loss of NP (from 40% cells with NP to 20% in 30 min) and a parallel increase of cells without NP (Figure 6D). Time-lapse imaging showed that NP retracted after 20 min at the restrictive temperature (Figure 6E). Of importance, after retraction of the NP, the nucleus lost its position close to the bud neck and fell back into the mother cell. These data suggested that anchoring of NP could be important for maintaining the position of the nucleus and the spindle close to the bud.

Exocyst mutants display nuclear- and spindle-positioning defects

If NP facilitate nuclear positioning, mutations defective in NP formation or maintenance should exacerbate the nuclear- and spindle-positioning defects of kar9∆ and dyn1∆ cells and lead to increased numbers of binucleated cells. Indeed, all tested mutant alleles encoding different exocyst subunits (sec3-2, sec4-8, sec6-4, and sec10-1) displayed synthetic growth defects upon deletion of DYN1 or KAR9 (Figure 7A). Moreover, sec3-2 and sec4-8 cells displayed a high percentage of binucleated cells, which increased when the sec3-2 and sec4-8 mutations were combined with kar9∆ and dyn1∆ deletions (Figure 7B; 16.8% binucleated sec3-2 kar9∆ and 16.5% binucleated sec3-2 dyn1∆ cells vs. 10% for sec3-2, 3.7% for kar9∆, and 5.6% for dyn1∆ single mutants). In fact, sec4-8 kar9∆, and sec4-8 dyn1∆ cells had increased numbers of binucleated cells even at the permissive temperature. We did not observe any mislocalization of Kar9 or dynein in exocyst mutants (Supplemental Figure S2C). Finally, we found that the distance between the spindle and the bud neck increased in sec3-2 cells at the restrictive temperature, indicating that spindles are mispositioned in this mutant (Figure 7C). To directly examine whether NP facilitate spindle positioning, we measured the position of the spindle in cells with or without NP. To eliminate any contribution of dynein-dependent insertion of the nuclear envelope into the bud, we performed this measurement in dyn1∆ cells. We found that the spindle resided closer to the bud neck in cells with NP (Figure 7C). Therefore NP seem to stabilize the position of the nucleus close to the bud and facilitate spindle positioning.

FIGURE 7:

The exocyst and DNA replication facilitate nuclear migration. (A) Exocyst mutations display synthetic growth defects when combined with dyn1∆ or kar9∆ deletion. Shown are serial dilutions of cells. (B) Exocyst mutants display nuclear migration defects, resulting in a high percentage of binucleated cells. The defects are additive when combined with dyn1∆ or kar9∆ mutations, already at permissive temperatures. No semipermissive temperature could be identified for sec4-8 cells. (C) Left, exocyst mutants display spindle-positioning defects. Shown is the distance of the middle of the mitotic spindle to the middle of the bud neck. Right, the presence of NP correlates with improved position of the spindle. The spindle is closer to the bud neck in dyn1∆ cells that have NP compared with cells that lack NP. (D) Selected deconvolved image slices from 3D image stacks showing tubular ER connections (yellow arrow) between NP (white arrow), the cortical ER, and Sec3. Each image is a different z-slice; distance between slices is 0.2 μm. (E) Cells that do not perform DNA replication do not form NP. Deconvolved image of an arrested cdc7-1 cell (left) at the restrictive temperature (37°C) and quantification of NP formation during preanaphase arrest. Compared are two time points at which cells of the two strains have the same bud sizes. The nuclear DNA is associated with the spindle and surrounded by the nuclear membrane. DAPI staining outside the nucleus is due to mitochondrial DNA.

We sought evidence to determine whether the exocyst mediates a physical connection between NP and the cortical ER. In P cells expressing Sec61-GFP (the major subunit of the ER translocon), Sec3-CFP, and mCherry-Tub1, we detected ER tubules connecting NP to the cortical ER at sites occupied by the exocyst subunit Sec3 (Figure 7D and Supplemental Videos S4–S6). Therefore, consistent with the role of NP and the exocyst in nuclear migration, NP may contribute to stabilization of the nuclear/spindle position by anchoring the nucleus to the cortical ER in the bud.

DNA replication is required for nucleopodia formation

Previous studies showed that nuclear envelope proliferation does not take place around the bulk of the DNA, which seems to restrict membrane expansion (Campbell ; Witkin ). DNA replication was shown to disrupt anchoring of telomeres to the nuclear envelope (Taddei ; Ebrahimi and Donaldson, 2008). We thus wondered whether, besides nuclear membrane expansion, untethering of chromosomes from the nuclear envelope due to DNA replication would be required for generation of NP.

To test this, we examined cdc7-1 cells that are unable to replicate their DNA and arrest due to activation of the spindle assembly checkpoint (Biggins and Murray, 2001) with high mitotic Cdk1/Cdc28 activity, a mitotic spindle, and inactivated APCCdc20. In contrast to Cdc20-depleted cells, cdc7-1 cells were defective in NP formation (Figure 7E). Of importance, the inability of replication-defective cdc7-1 cells to form NP was not due to failure to expand their nuclear membrane. The nuclear volume of G2/M arrested cdc7-1 cells (10.8 ±1 μm3) did not differ significantly from the nuclear volume of arrested P cells (11.3 ± 0.7 μm3). We attempted to restore NP formation in cdc7-1 cells by deactivating factors shown to tether telomeric regions to the nuclear envelope. However, we did not observe any increase in NP formation after combining the cdc7-1 mutation with esc1∆, yku70∆ (Taddei ; Ferreira ), mps3∆pom152∆ (the pom152∆ mutation was used here to suppress the inviability of mps3∆; Friederichs ), or siz1∆siz2∆ (Taddei ; Ferreira ). Nevertheless, these data suggest that DNA replication is required for NP formation and efficient nuclear migration into the bud.

DISCUSSION

In this article we analyzed the changes in nuclear morphology accompanying positioning of the nucleus relative to the cleavage apparatus at the bud neck. We showed that the nuclear membrane can migrate ahead of the mitotic spindle into the bud by generating protrusions, which depend on nuclear membrane growth. Clearly, NP have been previously described and can be observed in any cycling yeast culture (Prinz ; Fehrenbacher ). However, to unambiguously distinguish NP from dumbbell or anaphase nuclei, the mitotic spindle (or the DNA) and the nuclear envelope must be simultaneously visualized. Nucleopodia are most easily observed whenever cells (also cycling cells) display a G2/M delay. We do not think that generation of NP is a checkpoint-specific process; rather, the G2/M cell cycle delay simply reveals the formed NP and its connection to the bud, which are obscured in fast-cycling cells.

Our data indicate that NP generation depends on phospholipid biosynthesis, actin, and the exocyst complex. Biosynthesis of phospholipids takes place in the ER, whereas actin is required for the delivery of cortical ER into the daughter cells (Estrada ). Nucleopodia could form when ER tubules connect the nuclear envelope with the cortical ER in the bud as the nucleus migrates close to the bud neck (Figure 8). Because the bud ER may be biosynthetically more active, incorporation of new lipids causes the nuclear envelope to proliferate in the direction of the bud. It was recently suggested that vesicle trafficking might regulate the availability of the nuclear membrane and the shape of the nuclear envelope (Webster ). We were not able to test mutants defective in vesicular trafficking (Novick ) for defects in NP formation because these mutants have a very sharp transition from permissive to restrictive temperature and immediately stopped growing. Nevertheless, we found that these mutations cause a high percentage of binucleated cells, which further increases after combination with dyn1∆ or kar9∆ mutations (Supplemental Figure S2D). These data support the finding that nuclear membrane proliferation is required for the shape changes that accompany migration of the nuclear envelope into the bud.

FIGURE 8:

Model showing events during nuclear migration/spindle positioning in budding yeast. Early in the cell cycle the KAR9 pathway moves the nucleus close to the bud neck. After entry into mitosis, replication starts and the nuclear membrane expands. Dynein-dependent spindle oscillations pull part of the nuclear envelope into the bud, which creates NP, as it continues to proliferate. This nuclear subdomain is connected to the cortical ER (dashed line) by the exocyst, stabilizing the position of the nucleus and the spindle close to the bud. Eventually, dynein pulls the spindle and the DNA mass into the bud neck, creating a dumbbell nucleus. Why DNA replication is required for NP formation is not understood; it may untether chromosomes from the nuclear envelope, enabling the nuclear membrane to move through the bud neck. Neither Kar9, dynein, nor NP formation is absolutely essential for nuclear migration. However, all three pathways likely cooperate to guarantee robust inheritance of nucleus and chromosomes.

To our surprise, we found that NP formation does not depend on dynein-mediated spindle oscillations. This raises the question of whether the opposite could be the case: nuclear membrane expansion may be required for dynein-dependent oscillations to occur. It would be interesting to study spindle oscillations when nuclear membrane growth is inhibited (i.e., upon Nem1-Spo7 overexpression).

The exocyst complex seems to be a docking/attachment site for ER tubules and growing NP. One could also speculate that the exocyst may play an active role in formation of NP, since exocyst homologues are involved in generation of membrane nanotubes (Hase ). In this respect, the Myo2 motor, which directly interacts with exocyst components Sec15 and Sec4, could be involved in generation of NP: indeed, myo2 mutants are defective in NP formation (Supplemental Figure S2E). However, myo2 mutants also affect actin polarity (Lillie and Brown, 1994), and therefore their defect in NP formation is not clear.

Why is DNA replication required for NP formation? Previous reports demonstrated that the DNA mass restricts nuclear membrane expansion, which instead takes place at the nuclear membrane adjacent to the nucleolus (Campbell ; Witkin ). Consistent with this, we find that NP are devoid (or contain only minor amounts) of DNA. Thus interaction of chromosome regions with the nuclear envelope in cells with unreplicated DNA may inhibit membrane expansion around the DNA mass and formation of NP. However, disruption of telomeric tethering to the nuclear envelope did not restore NP formation in cells with unreplicated DNA. Therefore chromosomal regions other than telomeres may interact with the nuclear envelope to restrain nuclear membrane expansion.

Formation of NP seems to act in parallel to the KAR9 and DYN1 pathways to facilitate nuclear/spindle migration. There are several examples for the role of cortical ER in spindle positioning. During ascidian asymmetric divisions, the cortical centrosome attracting body, a mass of ER containing asymmetrically segregated mRNAs and the cortical PAR-3/PAR-6/aPKC complex, keeps one centrosome in the proximity of the cortex by regulating microtubule–cortical interactions (Munro, 2007). Spindle positioning through anchoring of one centrosome to the spectrosome, an ER-rich structure of Drosophila germline stem cells, is another example (Lin ; Yamashita ). In a variation of this mechanism in budding yeast, NP formation and exocyst-dependent connections to the cortical ER may stabilize the position of the nucleus and the mitotic spindle, facilitating chromosome segregation in anaphase. It will be interesting to explore whether similar connections contribute to stabilization of the nuclear position in higher eukaryotic cells.

MATERIALS AND METHODS

Yeast strains, growth conditions, and drug concentrations

For strains see Supplemental Table S1. All strains were routinely grown in standard yeast extract/peptone/dextrose (YPD) or selective medium. Strains expressing CDC20 under the control of the GAL1-10 promoter were grown in YPSG (YP supplemented with 2% sucrose + 2% galactose). Endogenously tagged fusions were created by standard yeast techniques. For α-factor arrest, cells were incubated with 5 μg/ml α-factor for 3 h. Latrunculin A was used at 100 μM, nocodazole at 10 μg/ml, cerulenin at 10 μM, and soraphen A at 0.25 μg/ml final concentration. After release from α-factor arrest, cells were incubated for 30 min at 28°C to allow the formation of a small bud before addition of drugs. For 4’,6-diamidino-2-phenylindole (DAPI) staining of living cells, cells from logarithmically growing cultures were briefly incubated in 200 μl of phosphate-buffered saline (PBS), containing 2 μl of 1 μg/ml DAPI in PBS. Cells were washed twice to remove the DAPI and resuspended in 2.5 μl of nonfluorescent medium (Waddle ) for microscopy.

Fluorescence microscopy and image analysis

For GFP and mCherry visualization, cells were grown overnight in YPSG liquid cultures containing additional adenine, tryptophan, and uracil. Yeast cells expressing mCherry-Tub1 and tagged Nup1-GFP were analyzed by time-lapse fluorescence microscopy using an Olympus IX81 microscope (Tokyo, Japan) equipped with the CellR imaging system. Images were acquired with an APO 100× objective as z-series of five focal planes separated by 0.5 μm and projected into a single plane. Images were analyzed using ImageJ (National Institutes of Health, Bethesda, MD), and statistical analysis was performed with Excel (Microsoft, Redmond, WA).

For the three-dimensional (3D) reconstruction of yeast nuclei and deconvolution, a z-series of 41–81 z-stacks with 0.1-μm z distance were background subtracted and deconvolved using Huygens Essential 3.4 (Scientific Volume Imaging, Hilversum, Netherlands) with a point spread function experimentally obtained for the specific lens. Deconvolved 3D images were thresholded and nuclear volumes were calculated by Imaris 5.7.2 (Bitplane, Zurich, Switzerland). Thresholding was arbitrary, but only nuclei that generated a closed surface were chosen for quantification of nuclear volumes. Recalculation of the volume of the same nucleus resulted in a value that differed by <5% from the first calculated value.