Moe1, a conserved protein in Schizosaccharomyces pombe, interacts with a Ras effector, Scd1, to affect proper spindle formation

Abstract

In fission yeast, Scd1/Ral1 is a putative guanine nucleotide exchange factor for Cdc42sp and also acts as a Ras1 effector necessary for the regulation of cytoskeleton organization. In this study, we have characterized a protein, Moe1, that binds directly to Scd1. A moe1 null (Δ) mutant exhibits numerous phenotypes indicative of abnormal microtubule functioning, including an abnormality in the spindle. moe1Δ mutants are resistant to microtubule destabilizing agents; moreover, moe1Δ rescued the growth defects of tubulin mutants containing unstable microtubules. These results suggest that Moe1 induces instability in microtubules. Biochemical and subcellular localization studies suggest that Moe1 and Scd1 colocalize in the nucleus. Furthermore, loss of function in Scd1 or Ras1 also induced abnormality in the spindle and is synthetically lethal with moe1Δ producing cells that lack a detectable spindle. These data demonstrate that Moe1 is a component of the Ras1 pathway necessary for proper spindle formation in the nucleus. Human and nematode Moe1 both can substitute for yeast Moe1, indicating that the function of Moe1 in spindle formation has been conserved substantially during evolution.

The ras protooncogenes encode GTP/GDP binding proteins that are essential elements in signal transduction pathways necessary for the regulation of cell growth and organization of the cytoskeleton (1). Ras proteins can cycle between the GDP and GTP-bound forms, but only the GTP-bound Ras is active. To activate Ras, extracellular signals act through guanine nucleotide exchange factors, which catalyze the exchange of the bound GDP for GTP. Activated Ras in mammalian cells has been shown to interact with the Raf-mitogen-activated protein kinase cascade to regulate cell proliferation (2, 3). The molecular pathways leading to the remodeling of the cytoskeleton, however, have not been firmly established, although they seem to involve the Rho-like GTPases (4).

The fission yeast Schizosaccharomyces pombe has one known Ras homolog, Ras1, which participates in at least two distinct signaling events. Ras1 can interact with the Byr2 protein kinase, a member of a mitogen-activated protein kinase cascade, to mediate mating pheromone signaling that is necessary for mating and sporulation (5, 6). In addition, we have identified a second Ras1 effector, Scd1 (shape and conjugation deficiency, also called Ral1; ref. 7), which is a putative guanine nucleotide exchange factor for Cdc42sp, a Rho-like GTPase (6). Our data suggest that Scd1 is necessary for maintaining an elongated cell shape but not for pheromone signaling. In the same study, we isolated Scd2/Ral3 (6, 7), which interacts directly with Scd1. Our data support a model in which Ras1 and Scd2 act cooperatively to regulate the interaction between Scd1 and Cdc42sp. We will refer to the protein complex of Ras1, Scd1, Scd2, and Cdc42sp as SCD. A putative effector for Cdc42sp has been identified, Shk1/Pak1/Orb2, which is a homolog of the mammalian Pak and budding yeast STE20 protein kinase (8–10).

It has been shown that in ras1Δ cells, both the actin and microtubule (MT) cytoskeleton appear disorganized (11, 12). However, neither the physiological significance of these observations nor the mechanisms by which Ras1 affects the organization of the cytoskeleton have been fully analyzed. In a separate study (Y-c.L., C-r.C., and E.C.C., unpublished work), we present evidence demonstrating that the Ras1/Scd1 pathway can affect the functioning of the spindle. To further understand the mechanism by which Scd1 affects spindle functioning, here we describe the study of a conserved protein, Moe1. Our data indicate that one of the principal functions of Moe1 is to reduce stability of MTs and that Moe1 and Scd1 can interact in the nucleus to affect spindle formation. Furthermore, we demonstrate that the function of Moe1 in spindle formation has been conserved substantially during evolution as human and nematode Moe1 both can fully substitute for yeast Moe1.

MATERIALS AND METHODS

Parental Strains and Microbial Manipulation.

The generic wild-type strain in our laboratory is SP870 (h⁹⁰, ade6.210, leu1.32, ura4-D18). Mutant strains containing the nda2-KM52 and nda3-KM311 mutations (13) and their parental wild-type strain (h⁻, leu1.32) were from Paul Nurse (Imperial Cancer Research Fund, London). For clarification we named the wild-type strain from the Nurse lab PN1. The rich and minimal medium were yeast extract medium supplemented with adenine and uracile (YEAU) and minimal medium (14). Hydroxyurea (HU; 11 mM) treatment was essentially as described (15) except that our strains were incubated in HU for 6 hr. Cell viability was determined by plating on YEAU at 30°C.

Yeast Two-Hybrid Screening.

The protocols for the screening and the β-galactosidase assay are as described (6). A BamHI fragment of scd1ΔB (encoding residues 1–815; ref. 6) was cloned into pVJL11 to create pLBD-SCD1ΔB. Approximately one million cDNA clones were screened. A total of six SCD1ΔB-specific clones were identified, one of which encodes Moe1ΔN. The interaction between the LBD-Scd1ΔB and GAD-Moe1ΔN detected by the β-galactosidase assay was performed for approximately 6 hr at 30°C.

Cloning of moe1 and Plasmid Constructions.

A 4.1-kb BamHI/ApaI fragment containing moe1 was obtained from cosmid no. 637 (16) and cloned into Bluescript to create pBSMOE1. A 1.8-kb XcmI and BglII fragment in pBSMOE1 was swapped with ura4 to create pMOE1U; similarly, pMOE1L was constructed by replacing ura4 with LEU2. The KpnI site in the polycloning site of pBSMOE1 was converted to a PstI site by a linker. A 4.1-kb SacI/PstI fragment was excised from the resulting plasmid and cloned into pAL (5) to generate pALMOE1. A BglII fragment of moe1 was created by PCR and cloned into pARTCM (6) and pTrcHisB (Invitrogen) to create pARCMOE1 and pHTMOE1, which express c-MYC-Moe1 in yeast and HT-Moe1 in bacteria, respectively. To express glutathione S-transferase (GST)-Scd1 in Sf9 cells, a BamHI/NheI fragment of scd1 was obtained from pGAD-SCD1 (6) and cloned into a modified pAcG1 (PharMingen) in which the SmaI site had been changed to NheI. pALLT17N was created by swapping the ura4 in pALUT17N (6) with LEU2. The BglII-moe1 was cloned into pALG (Y-c.L., C-r.C., and E.C.C., unpublished work), which expresses green fluorescence protein (GFP)-Moe1 (GFP:GFP[S65T]; ref. 17) under the control of the adh1 promoter. pNMTSCD1 (Y-c.L., C-r.C., and E.C.C., unpublished work) allows expression of GFP-Scd1 under the control of a weak nmt1 promoter (pREP41; ref. 18). The cDNA for CeMoe1 was amplified from the Caenorhabditis elegans cDNAs (provided by Michael Hengartner, Cold Spring Harbor Lab, Cold Spring Harbor, NY) to create unique PstI and KpnI sites and cloned into pART1 (19) to generate pCEMOE1. Human moe1 was amplified from the cDNAs of HeLa cells (from Greg Hannon, Cold Spring Harbor Lab, Cold Spring Harbor, NY) to contain unique BamHI sites and cloned into pARTCM to create pHSMOE1.

Strain Constructions.

To delete moe1, strain SP870 was transformed with an ApaI/SacI fragment released from pMOE1U. Uracile-prototrophic transformants were selected, and the deletion of moe1 was confirmed by PCR. One of these was named MOE1U and chosen for detailed studies. A moe1Δ strain, PNMOE1L, also was created in strain PN1 by using pMOE1L. Diploid cells homozygous for the deletion of scd1, scd2, and ras1 were constructed by protoplast fusion of haploid null cells, SPSCD1U, SPSCD2L, and SPRU, respectively (6). The ras1Δ moe1Δ, scd2Δ moe1Δ, and scd1Δ moe1Δ strains used in this study are named R1UME1L, SD2LME1U, and SD1UME1L.

MT Assembly and Stability in Cells.

Cells were pregrown at 30°C to early log phase and then transferred to 20°C to grow for an additional 12 hr. For cold shock, cells were transferred to an ice bath, and time points were taken for fixing and immunofluorescence microscopy as described below. To monitor MT polymerization, samples were subjected to cold shock as above for 30 min and then transferred to 20°C.

Immunoprecipitation and Far-Western Analysis.

pARCMOE1 and pALUSCD1 (6) were used to express the c-MYC-tagged Moe1 and the HA1-tagged Scd1, respectively, in a scd1 mutant, SP870M3 (6). Freshly transformed cells were used for the preparation of lysates. The lysis buffer was 100 mM Pipes, 1 mM MgSO₄, 1 mM EGTA, pH 6.9 (20) plus 1% Triton X-100. mAbs 12CA5 and 9E10 and the control mouse IgG were covalently coupled to protein-A beads (6). For the Far-Western analysis, purified proteins were separated by SDS/PAGE and transferred to a nitrocellulose membrane without using methanol in the transfer buffer. The membrane was incubated first in 3% nonfat dry milk in tris-buffered saline (TBS) for 1 hr; after washing, it was incubated with purified His-tagged HT-Moe1 (1 μg/ml in TBS; 0.05% Tween-20; 3% BSA) overnight at 4°C. An anti-T7 antibody from Novagen was used (1:10,000) to detect the presence of HT-Moe1.

Subcellular Localization.

The general procedures for immunostaining were as described (14). To visualize MTs, TAT1 (1:5, overnight) was used (21). F-actin was visualized by Rhodamine-phalloidin (Molecular Probes; 20 units/ml, 1 hr). To examine GFP-Moe1 and GFP-Scd1, cells were transformed with pALGMOE1 or pNMTSCD1 and seeded on plates containing 20 μM thiamine. Overexpression of Scd1 induced lethality in moe1Δ cells. To achieve the optimal level of expression of GFP-Scd1, cells were inoculated from the thiamine-rich to the thiamine-free medium and allowed to grow at 30°C for 2 days before being fixed (3% formaldehyde, 5 min). Under these conditions, the expression level of GFP-Scd1 was quite low and was detectable in only 30% of the cells. In addition to the nucleus, GFP-Scd1 also appeared at the spindle and cell equator. In a separate study (Y-c.L., C-r.C., and E.C.C., unpublished work), we have examined the localization of GFP-Scd1 in scd1Δ cells in which GFP-Scd1 can be expressed to a higher level. In that case, in addition to the described patterns of localization, GFP-Scd1 also could be detected at the cell tips. Therefore, we speculate that to detect GFP-Scd1 in the cell ends, GFP-Scd1 needs to be expressed at a high level.

RESULTS

Isolation and Sequence Analysis of moe1.

To identify genes encoding proteins that form complexes with Scd1, we carried out a yeast two-hybrid screen using as bait a truncated, but functional, Scd1 (Scd1ΔB; ref. 6). One of the isolated cDNA clones encodes the C-terminal 271 aa of a protein that we named Moe1 for “microtubules overextended,” based on the phenotype of the null strains as described below. The protein encoded by the truncated cDNA (Moe1ΔN) interacts specifically with Scd1ΔB but not with Scd2, Ras1, or Cdc42sp. The full-length moe1 gene was isolated from a cosmid library (16) and contains an uninterrupted ORF capable of encoding a protein of 568 aa. The truncated cDNA, moe1ΔN, encodes residues 298–568.

Databank searches revealed two Moe1 homologs in GenBank, one from humans and one from C. elegans. The amino acid sequences of these two Moe1 homologs share more than 50% identity with yeast Moe1 throughout the entire coding region. (A sequence alignment created with clustal is published as supplemental data to this article on the PNAS web site, www.pnas.org.) The C. elegans Moe1 homolog contains 574 residues (Cosmid clone R08D7_3; ref. 22); however, the function of this gene has not been characterized. The human Moe1 homolog, eIF3-p66, contains 548 aa and was identified as a protein copurified with the translation initiation protein complex (23). Whether eIF3-p66 actually is involved in translation remains unknown. Surprisingly, no Moe1 homolog is present in Saccharomyces cerevisiae although it contains homologs of Ras1, Scd2, Scd1, Cdc42sp (6), and Shk1 (8, 9).

Loss of Function in moe1 Induces Abnormalities in MTs and Spindle.

To study the function of Moe1, moe1Δ strains were created in which the entire moe1 ORF was removed. The resulting cells are viable, but grow less efficiently than wild-type cells as the growth temperature decreases (Fig. 1A). At 30°C, the optimal growth temperature for fission yeast, the generation times of wild-type and moe1Δ cells were 2.6 ± 0.3 and 3.4 ± 0.2 hr (mean ± SEM, n = 3 experiments), respectively. At 20°C, the growth rate of moe1Δ cells reduced to about half of wild type (generation time: 11.7 ± 0.2 vs. 6.7 ± 0.5 hr). The morphology of moe1Δ cells is also abnormal. Unlike wild-type cells, which are straight and elongated, moe1Δ cells are frequently bent and longer and wider (Fig. 1B). Unlike scd1Δ cells, which are sterile, moe1Δ cells are fertile.

Figure 1.

Growth curve, cell morphology, and MTs of various strains. (A) Cells were grown to ≈4 × 10⁶ cells/ml at 30°C and then diluted to 1 × 10⁶ cells/ml at t = 0. After 3 hr (marked by an arrow), cells either were transferred to 20°C or remained at 30°C, and the cell density was monitored over time. (B) Cells were pregrown in solid rich medium at 30°C for 24 hr before being examined. The frequency of bent cells was 20%. (C and D) Cells were grown at 20°C to mid-log phase and subjected to fluorescence microscopy to visualize MTs, F-actin, and DNA. scd1Δ and ras1Δ cells showed the same abnormal phenotype in MTs at 30°C. I, interphase; M, mitosis. Strains used: SP870 (wild type, WT), MOE1U (moe1Δ), SPSCD1U (scd1Δ), and SPRN (ras1Δ).

Because the cold-dependent growth defects and the presence of bent cells in moe1Δ strains resemble phenotypes of mutants defective in MTs (13, 24–26), we performed immunostaining to examine whether MTs are abnormal in moe1Δ cells. As shown in Fig. 1C, wild-type cells exhibit a typical pattern of interphase MTs with two or three dominant MT bundles extending between the two ends of the cell (27). In moe1Δ cells, however, each cell has more MT bundles (Fig. 1C). The number of MT bundles per cell was estimated by counting each in various focal planes. moe1Δ cells have on average at least 4–6 MT bundles per cell, compared with 2–3 in wild type. moe1Δ cells also show abnormalities in the spindle. The two dividing nuclei of a wild-type cell are separated by a single straight spindle (Fig. 1C). In moe1Δ cells, the spindle in 25% of cells in anaphase appears to be abnormally long and curves within the cell in a distinct “S”-shape (Fig. 1C). At a lower frequency (1% of the entire population), we observed what appears to be a fraying V-shaped spindle, which suggests that the MTs in the spindle are not properly interdigitated (Fig. 1C; ref. 28). At 30°C, by contrast, these abnormalities in MTs/spindle were rarely detectable.

We also examined the organization of F-actin in moe1Δ cells and found no abnormality at either 30°C or 20°C. That is, F-actin is concentrated at the growing ends and the cell equator as observed in wild-type cells (Fig. 1D; ref. 27). Hence, we conclude that moe1Δ cells have cold-dependent abnormalities in MTs and the spindle, whereas the organization of F-actin does not seem to be visibly altered.

Loss and Gain of Function of Moe1 Affect MT Stability and/or Assembly.

We attempted to decipher the influence of Moe1 on the functionality of MTs. As described above, both the number of MT bundles and the length of the spindle increase at 20°C as a result of moe1Δ. By using Western blot analysis, we determined that the levels of α tubulins in moe1Δ cells was the same as in wild type; thus, it is unlikely that the observed anomaly in MTs and spindle is caused by an overproduction of the α tubulins. Therefore, as an alternative hypothesis, we postulated that Moe1 negatively affects the assembly and/or stability of MTs.

To analyze the stability of MTs in moe1Δ cells, we exposed cells to treatments known to promote depolymerization of MT, including cold shock in ice (29) and thiabendazole (TBZ). To quantitate depolymerization of MTs under cold shock, we measured the percentage of cells that have long MTs during the time of the cold shock (Fig. 2A). We reasoned that the more stable the MTs are in a cell, the higher the percentage of cells that have long MT bundles and spindle at any given time after the cold shock. Our data show a marked decline in the percentage of cells containing long MT bundles and spindle in wild-type versus moe1Δ cells (Fig. 2A). In addition, when the cold-shocked cells were allowed to recover at 20°C, we found that MTs also regenerated more readily in moe1Δ cells than in wild type (Fig. 2B). These data support our model that moe1Δ renders MTs more stable and allows them to assemble more readily.

Figure 2.

The effects of loss of function of Moe1 on MT assembly and stability. (A) The percentages of cells with either a long MT (one that spans at least half the cell length) or an intact spindle (one that connects the two nuclei) are plotted against time after the cold shock. (B) Cells first were cold-shocked in ice for 30 min. The MTs then were examined at various time points after recovery at 20°C. (C) α1 tubulin mutant (nda2^cs) and the α1-tubulin Moe1Δ double mutant cells (nda2^cs moe1Δ) pregrown at 30°C to log phase were shifted to 20°C; the MTs were visualized after 30 min. (D) Serial dilutions of cells (1:5) were spotted on yeast extract medium supplemented with adenine and uracile (YEAU) plates (Materials and Methods) containing either no or 12.5 μg/ml of thiabendazole. The plates were incubated at 30°C for 3 days. Note that the plating efficiency of moe1Δ cells at 30/25°C is only 50% of wild type. We therefore routinely use twice as many moe1Δ as wild-type cells for this type of analysis. The wild-type and moe1Δ strains in A and D were SP870 and MOE1U, respectively, and in B were PN1 and PNMOE1L, respectively.

Consistent with the hypothesis that moe1Δ leads to a global stabilization of MTs, we found that moe1Δ rescued the growth defects of the cold-sensitive α1 (nda2-KM52) and β (nda3-KM311) tubulin mutants (13) at 20°C. The MTs/spindles in the moe1Δ nda2-KM52 double mutant also appeared more stable than those in the nda2-KM52 mutant cells at 20°C (Fig. 2C). Furthermore, we show that deletion of moe1 rendered cells more resistant to TBZ (Fig. 2D), whereas overexpression of moe1 rendered cells weakly more sensitive to TBZ (data not shown). Based on these data, we propose that Moe1 negatively affects the assembly and/or stability of MTs.

Moe1 and Scd1 Interact Directly in Vitro.

The two-hybrid data suggest that Moe1ΔN can form a protein complex with Scd1. We therefore proceeded to investigate whether full-length Moe1 and Scd1 can interact directly by using biochemical methods. Far-Western analysis revealed that poly-histidine tagged Moe1 (HT-Moe1) can interact specifically with immobilized GST-Scd1, but not with the control GST (Fig. 3A), demonstrating that Scd1 and Moe1 interact directly in vitro. Furthermore, immunoprecipitation experiments revealed that c-Myc epitope-tagged Moe1 coprecipitated with HA1-tagged Scd1 from fission yeast lysates, and vice versa (data not shown). These data indicate that Scd1 directly interacts with Moe1 in vitro and suggest a direct interaction between them in vivo.

Figure 3.

Physical interaction between Scd1 and Moe1 in vitro and in vivo. (A) Purified GST (3 μg) and GST-Scd1 (1 μg) were separated by SDS/PAGE and transferred to a nitrocellulose membrane. The blot (Right) was incubated with purified HT-Moe1 and then analyzed by using an anti-T7 antibody that detected the presence of HT-Moe1 that binds GST-Scd1. The same samples also were stained with Coomassie blue for comparison (Left). The protein band corresponding to the GST-Scd1 HT-Moe1 complex is marked by an arrowhead. (B) The localization of GFP-Moe1 and GFP-Scd1 in scd1Δ (SPSCD1U) and moe1Δ (MOE1U) cells was examined after short fixation. Cell samples were counterstained with 4′,6-diamidino-2-phenylindole to view DNA. GFP-Moe1 appeared in the nucleus in >85% of moe1Δ cells; in scd1Δ cells, however, GFP-Moe1 appeared rather diffused and only <25% of scd1Δ cells contained GFP-Moe1 in the nucleus. In contrast, in every moe1Δ cell that displayed detectable levels of GFP-Scd1 GFP-Scd1 was always in the nucleus.

Scd1 Is Required for Proper Nuclear Localization of Moe1.

Because Moe1 and Scd1 interact directly in vitro, we investigated whether Moe1 and Scd1 colocalize in the cell by using GFP fusion proteins. GFP-Moe1 or GFP-Scd1 both were able to fully complement the phenotypes of moe1Δ or scd1Δ cells, respectively. As shown in Fig. 3B, whereas the GFP control protein was dispersed over the entire cell, GFP-Moe1 appeared predominantly associated with the nucleus throughout the cell cycle. In S. pombe, the nuclear envelope does not break down during mitosis so the dividing nuclei remain connected by the nuclear membrane. Moe1 was not present between the two dividing nuclei (data not shown), which suggests that Moe1 is most likely in the nucleus, and not in the nuclear membrane. A functional GFP-Scd1 also can be detected in the nucleus throughout the cell cycle (Y-c.L., C-r.C., and E.C.C., unpublished work) (Fig. 3B). More important, the nuclear localization of GFP-Moe1 was dramatically reduced in scd1Δ cells (Fig. 3B). Together with the biochemical data, we hypothesize that Moe1 and Scd1 colocalize in the nucleus and that Scd1 is essential for proper nuclear localization of Moe1.

Deletion of Both moe1 and scd1 Blocks Spindle Formation.

We next constructed moe1Δ scd1Δ double mutants to define the physiological significance of the physical interaction between Scd1 and Moe1. The moe1 gene was deleted and replaced with a selectable marker, LEU2, in a diploid homozygous for an scd1 deletion. A total of 52 tetrads from the resulting diploid cells were examined; all produced only two viable spores of four. None of the viable spores were prototrophic for leucine, which suggests that moe1 and scd1 double deletions induce synthetic lethality.

To investigate the cause for the synthetic lethality, the diploid cells that are homozygous for scd1Δ (scd1∷ura4) and heterozygous for moe1Δ (moe1∷LEU2) were induced to sporulate, and the spores were allowed to germinate in minimal medium that would support the “growth” of scd1Δ moe1Δ cells, but not scd1Δ cells. We found that scd1Δ moe1Δ cells were able to germinate but unable to divide after 24 hr. The majority of these cells (75%) appeared to be in early mitosis with condensed chromosomes but without a spindle (Fig. 4A, Right). As described above, the spindle in moe1Δ cells appears abnormal. In a separate study (Y-c.L., C-r.C., and E.C.C., unpublished work), we found that the spindle in SCD-deficient cells is also abnormal (see also Fig. 1C). Therefore, we speculate that the lethality in scd1Δ moe1Δ cells is the result of the lack of a spindle.

Figure 4.

Loss of function in both Moe1 and SCD affects proper spindle formation. (A) Spores of diploid (scd1∷ura4/scd1∷ura4 +/moe1∷LEU2) were allowed to germinate in minimal medium that lacks leucine and uracile at 30°C. Cells were immunostained after 24 hr. The majority of cells (>75%) appear to accumulate in early M phase (Right), and the rest were still in interphase (Left for comparison). (B) Separation of chromosomes in cells that contain a spindle of 6 μm. (Scale bar = 3 μm.) (C) Aberrant morphology of synchronized scd2Δ moe1Δ cells (I–IV) after two generations at 20°C. Cells were fixed in methanol and stained with 4′,6-diamidino-2-phenylindole. (I) Cells with no or only traces of DNA signal. (II) Cells with unequal number of nuclei in the two cell compartments separated by a septum. (III) Binucleate cells without a spindle (MT staining is not shown). (IV) Cells with at least one of the nuclei “cut” by the septum. The position of the septum is indicated by an arrowhead. The frequency at which the abnormal cells occurred is indicated. The strains used in this study were: SP870 (wild type, WT), SD1UME1L (scd1Δ moe1Δ), and SD2LME1U (scd2Δ moe1Δ).

Deletion of moe1 in Other SCD Loss-of-Function Mutants Also Blocks Proper Spindle Formation.

Scd1 is part of a protein complex that contains at least three other proteins, Ras1, Scd2, and Cdc42sp. Hence, one would expect that components interacting with Scd1 also would interact with Ras1, Scd2, or Cdc42. In support of this hypothesis, we found that deletion of ras1 induced synthetic lethality in moe1Δ cells, and the terminal phenotype of these cells is indistinguishable from that of scd1Δ moe1Δ cells (data not shown).

Deletion of scd2 also created a synthetic lethality in moe1Δ cells, but, unlike scd1Δ moe1Δ cells, which are not viable at 30°C, scd2Δ moe1Δ cells grow but very slowly. At 20°C, however, the scd2Δ moe1Δ strain cannot form colonies. To investigate whether these scd2Δ moe1Δ cells are also defective in spindle formation, we first examined the percentage of synchronized cells that have a detectable spindle.

Cells were synchronized at 30°C by hydroxyurea, and then released from the blockage and allowed to grow at 20°C. We counted over time the percentage of cells at 20°C that had entered anaphase (the anaphase index), as judged by the presence of two dividing nuclei lying along the cell body by using 4′,6-diamidino-2-phenylindole staining. As the anaphase index peaked, approximately 56% of wild-type and scd2Δ cells and 47% of moe1Δ cells had a spindle; by contrast, in scd2Δ moe1Δ cells, only 24% had a spindle. Because the chromosomes in the majority of scd2Δ moe1Δ cells (60%) were condensed, the lack of a detectable spindle is unlikely to result from a mitotic block caused by the lack of chromosome condensation. These data support the hypothesis that the ability to generate a spindle is substantially inhibited in scd2Δ moe1Δ cells. Furthermore, for those scd2Δ moe1Δ cells that did make a long spindle of 6 μm, more than 80% could not efficiently separate the chromosomes so there was only a single DNA mass (Fig. 4B). In contrast, in the majority of wild-type, scd2Δ, or moe1Δ cells (>80%), the two sister-chromatids can fully separate (Fig. 4B). These data suggest that if a spindle is generated in scd2Δ moe1Δ cells, it cannot separate chromosomes efficiently. Thus, we conclude that scd2Δ moe1Δ cells are defective in the generation of a functional spindle.

To examine the terminal phenotype of scd2Δ moe1Δ cells, the cultures were kept at 20°C for one more generation. We observed no appreciable loss of viability in wild-type, scd2Δ, and moe1Δ cells after two generations (>80% viable). In contrast, more than 70% of scd2Δ moe1Δ cells were inviable with deformities that can be grouped into four major categories (Fig. 4C). We observed deformities indistinguishable from those shown in Fig. 4C in an asynchronous culture of scd2Δ moe1Δ cells at 20°C, indicating that these abnormalities are not artifacts caused by hydroxyurea. The accumulation of these abnormal cells occurs frequently in mutant strains defective in chromosome disjunction (26). One explanation is that septation/cytokinesis in S. pombe still can take place even though the chromosomes are not properly segregated, which ultimately results in uneven distribution of chromosomes in the two daughter cells. We conclude that the lethality in scd2Δ moe1Δ cells is caused by the absence of a functional spindle, which results in chromosome loss.

Deletion of cdc42sp alone induces lethality (30). To test whether a defective Cdc42sp can impart lethality to moe1Δ cells, we overexpressed a dominant negative Cdc42sp (Cdc42sp[T17N]; refs. 6, 8, and 9), and found that it, too, induced lethality in moe1Δ cells, producing abnormal cells indistinguishable from those in Fig. 4C.

On the basis of these double mutant studies, we conclude that SCD and Moe1 both are required for the construction of a functional spindle. In combination with moe1Δ, deletion of scd1 or ras1 appears to lead to a complete loss of a detectable spindle, whereas deletion of scd2 or overexpression of Cdc42sp[T17N] severely inhibits spindle formation and proper chromosome segregation.

Moe1 Function Is Conserved in Evolution.

To test whether the function of Moe1 is conserved in evolution, we investigated the effects of overexpressing human Moe1 on the phenotypes of S. pombe moe1Δ cells. Human Moe1 was able to rescue all the described phenotypes of moe1Δ cells: namely, the abnormal cell morphology, the presence of V- and S-shaped spindle, and the cold-dependent growth defect (Fig. 5). These results suggest that human Moe1 can affect the functioning of MTs in a manner very similar to that of yeast Moe1. Furthermore, human Moe1 also rescued the lethality of scd2Δ moe1Δ cells (data not shown), suggesting that human Moe1 also can interact with the Ras1/Scd1 pathway. The C. elegans Moe1 also rescued the phenotypes of moe1Δ cells (data not shown). These data demonstrate that the function of Moe1 is highly conserved in evolution. We suggest that the C. elegans and human Moe1 should be named CeMoe1 and HsMoe1, respectively.

Figure 5.

Moe1 function is conserved in evolution. Shown here are cell patches generated by replica plating containing various plasmids growing at 30°C for 2 days or at 20°C for 4 days. The plasmids tested were: pARTCM (C, empty vector), pALMOE1 (G-moe1, expression controlled by the genomic moe1 promoter), pARCMOE1 (adh-moe1, expression controlled by the adh1-promoter), and pHSMOE1 (Hs-moe1, expression of human moe1 controlled by the adh1-promoter). Strains used were SP870 (wild type, WT) and MOE1U.

DISCUSSION

In this report, we describe the characterization of a highly conserved gene from fission yeast, moe1. Biochemical and subcellular localization studies suggest that Moe1 and a Ras1 effector, Scd1, interact directly in the nucleus. We conclude from the genetic analysis of Moe1 SCD double mutants that Moe1 is a key component of the Ras1/Scd1 signaling pathway necessary for proper spindle formation.

Judging from the phenotypes of moe1Δ cells, we conclude that the deletion of moe1 primarily affects the function of MTs, but not the organization of F-actin. Furthermore, we determined that one of the principal functions of Moe1 is to destabilize MTs. The exact molecular mechanism by which this process occurs, however, is unresolved in this study. Increasing evidence suggests that motor proteins in the family of kinesin and dynein can play a key role in reducing the stability of MTs (31). In budding yeast, for example, a loss of function in a dynein-related motor protein, Dyn1 (32), induces cold-dependent spindle defects, and the dyn1 mutant cells are resistant to a MT-destabilizing drug, benomyl. These phenotypes are reminiscent of the phenotypes of moe1Δ cells. Hence, we speculate that Moe1 can affect MT stability through the action of motor proteins. Cold conceivably can retard the mobility of these motor proteins to reduce their effectiveness, which could explain why the phenotypes of moe1Δ cells are manifested predominantly in the cold. We note that our data are insufficient to judge whether the lack of Moe1 function promotes polymerization of MTs directly or it influences the organization and/or integrity of MTs, which then leads to an apparent increase in MT stability.

It is possible that persistent MTs block MT remodeling necessary for the generation of a functional spindle. Consistent with this hypothesis, it has been shown that the spindle of S. pombe is composed of several types of MTs, the length and number of which change dynamically during mitosis (33). It is conceivable that such MT morphogenesis would require cycles of MT assembly and disassembly in which Moe1 participates.

Subcellular localization studies illustrate that Moe1 and Scd1 can concentrate in the nucleus. The nucleus of S. pombe has been suggested to play an essential role in the regulation of MT functioning throughout the cell cycle. Interphase MTs that are experimentally disassembled regenerate from near the nucleus, suggesting the presence of microtubule organization centers (MTOCs) in or by the nucleus (ref. 29; wild-type cells in Fig. 2B). The generation of the spindle requires the spindle pole bodies, the yeast MTOCs responsible specifically for spindle formation, to integrate into the nuclear envelope. The importance of the nucleus in the regulation of MT assembly and stability is further underscored by the finding that essential constituents for the MTOCs, such as γ-tubulin (34), can be found in the nucleus. It is possible that the Moe1–Scd1 complex may exert at least part of its effects on MT assembly/stability by interacting with components of the MTOCs that are associated with the nucleus.

Our data show that the nuclear localization of Moe1 is markedly reduced in scd1Δ cells, whereas the nuclear localization of GFP-Scd1 was not markedly altered in moe1Δ cells. This result suggests that part of the function of Scd1 is to maintain Moe1 in the nucleus. In this scenario, Scd1 acts in a manner similar to other guanine nucleotide exchange factors, such as Sos (35), to assemble elements into a single protein complex that apparently can maximize the efficiency of signal transduction. However, the function of Scd1 in the regulation of spindle formation is unlikely to be limited to controlling the subcellular localization of Moe1. In a separate study, we show that Scd1 also can associate with the spindle (Y-c.L., C-r.C., and E.C.C., unpublished work). As shown here, however, Moe1 does not seem to associate with the spindle. Furthermore, GFP-Scd1 remains associated with the spindle in moe1Δ cells (unpublished results). These observations suggest that Scd1 can interact with the spindle and that the association between Scd1 and the spindle is independent of Moe1. Thus, in addition to recruiting Moe1 to the nucleus, Scd1 may anchor to the spindle to recruit other molecules necessary for spindle formation. Furthermore, because deletion of both moe1 and scd1 create a phenotype that is more severe than either single deletion, it is possible that Moe1 or Scd1 interacts with additional molecules to affect spindle formation.

A databank search has identified full-length sequences in C. elegans and humans that are more than 50% identical to Moe1. It is enticing to speculate that these Moe1 homologs also interact with the Ras pathway and MTs. In support of this hypothesis, we have shown that both nematode and human Moe1 can substitute for the function of yeast Moe1. Moreover, human Cdc42 (Cdc42Hs) has been shown to affect the functioning of the MTOCs. For example, helper T cells reorient MTOCs toward the antigen-presenting cells, which ultimately leads to a contact between the two cell types. Stowers and coworkers (36) have shown that this process can be blocked by the dominant negative Cdc42Hs. Finally, we note that the elevated levels of MT stability/assembly in moe1Δ cells resemble the centrosome hypertrophy observed in tumors in which MTs assemble robustly because of an increase in the number and size of the centrosomes (37). One can envision that centrosome hypertrophy also can lead to improper chromosome segregation, similar to that observed in yeast cells deficient in Moe1 and SCD, which would accelerate malignant transformation.

ABBREVIATIONS

MT

microtubule

GFP

green fluorescence protein

SCD

shape and conjugation deficiency

GST

glutathione S-transferase

HT

polyhistidine tagged

MTOC

microtubule organization center, TBZ, thiabendazole