Kinesin-related KIP3 of Saccharomyces cerevisiae Is Required for a Distinct Step in Nuclear Migration

Genotypes and sources of the strains and plasmids used in this study are listed in Table I. Strains were constructed by standard genetic methods (Rose et al., 1990) and as described in the text. The yeast media used were SD minimal medium and sporulation medium supplemented with adenine, uracil, and appropriate amino acids; SC complete medium containing 5-fluoroorotic acid (5-FOA)¹ (1 mg/ml); and YPD rich medium (Rose et al., 1990). The frequency of chromosome loss was assayed by scoring kip3/kip3 and wild-type diploid strains for loss of heterozygosity at the mating type locus of chromosome III. The diploid strains were mated to MATa and MATα tester strains on YPD plates and replica printed to score prototrophic triploid colonies (Spencer et al., 1990). Sensitivity to benomyl was tested on YPD medium containing 1% DMSO and the desired concentration of benomyl (E.I. du Pont de Nemours, Wilmington, DE). Strains expressing the Nuf2p–green fluorescent protein (GFP) fusion protein for localization of spindle poles (Kahana et al., 1995) were constructed by integrating the NUF2–GFP gene fusion at the NUF2 locus using plasmid pJK67 (a gift of J. Kahana and P. Silver, Dana Farber Cancer Institute, Boston, MA). The NUF2–GFP gene fusion is expressed from the NUF2 promoter, and the GFP gene contains S65T and V163A mutations.

A precise deletion of the KIP3 open reading frame (kip3Δ) was constructed by PCR (Baudin et al., 1993). The 3′ ends of primers 1 and 2 contained sequences for PCR amplification of the HIS3 or TRP1 genes present in pRS403 or pRS404 (Sikorski and Hieter, 1989), and the 5′ ends of the primers contained sequences immediately flanking the KIP3 gene. Primer 1 was of sequence 5′-ACTTGAGTTTTCTTTCCAGCTGTATACTATTGACACTAACGAATTCGATTGTACTGAGAGTGCACC- 3′ , and primer 2 was of sequence 5′-GAAAGAAGTTATATTCGAT-AGTTTACGTAGGATATGTATGGAATTCCTGTGCGGTAT T TC A- CACC-3′. (Sequences flanking KIP3 are underlined, and EcoRI restriction sites are in bold type.) The KIP3 locus in a diploid yeast strain was replaced by this PCR product using the one-step gene replacement procedure (Rothstein, 1983). Correct recombinants were identified by PCR amplification and restriction digest analysis of genomic DNA. The primers used were primer 3, 5′-CCGGGATCCGACTCTTCTAATTGGTCTCT-3′ and primer 4, 5′-GGCGTCGACGTCTCCTGAGAAACGTTTT-3′. (BamHI and SalI restriction sites are in bold type.)

To recover the KIP3 gene, we screened a library of S. cerevisiae genomic DNA fragments cloned into the vector YCp50 (URA3 CEN ARS) by colony hybridization using a radiolabeled probe. The probe was a 556-bp sequence adjacent to the 5′ end of KIP3 that was prepared from a plasmid carrying the kip3Δ1::HIS3 allele using the EcoRI and BamHI sites introduced by primers 1 and 3. We identified a single plasmid that carried the full-length KIP3 gene on a 10–15-kb genomic DNA insert (pDR605) out of ∼4,100 colonies screened.

For epitope tagging of KIP3, the myc tag coding sequence (Evan et al., 1985) was introduced into the KIP3 coding sequence as follows. A 3.8-kb fragment containing the entire KIP3 open reading frame, but lacking the termination codon, was generated by PCR using the Expand High Fidelity PCR system (Boehringer Mannheim Corp., Indianapolis, IN). This fragment was cloned into a vector containing six tandem copies of the myc tag (pB896 [Li, 1997]). The resulting construct contained 1,419 bases of 5′ noncoding sequence, the entire KIP3 open reading frame, and the sequence encoding the myc epitopes just 5′ of the termination codon. To eliminate the possibility of PCR-generated errors, a 2.25-kb fragment of wild-type KIP3 was used to replace most of the PCR-generated clone. The remaining PCR-generated KIP3 sequences were verified using DNA sequencing. KIP3-6MYC (pB956) appears to be functional, as it fully complements the lethality of kip3Δ kar3Δ and kip3Δ dyn1Δ strains using the plasmid shuffle test described below.

To test whether the kip3Δ mutation was lethal in combination with other mutations, we generally constructed double mutants in the presence of a plasmid-based wild-type copy of one of the genes and then determined whether the strain remained viable after plasmid loss. The presence of the wild-type gene during strain construction prevented the potential isolation of aneuploid strains because of genomic instability. Synthetic lethality of kip3 and dyn1 mutations was tested by replacing the wild-type DYN1 gene with the dyn1Δ::HIS3 deletion allele (from plasmid p2-1H, a gift of E. Yeh and K. Bloom, University of North Carolina, Chapel Hill, NC) in the kip3Δ strain DS667, which carries the KIP3 URA3 plasmid pDR605, using one-step gene replacement. Correct deletion of DYN1 was confirmed by restriction analysis and Southern blotting of genomic DNA. Five independently constructed double mutants were inviable when plated on 5-FOA medium, which selects against cells carrying the URA3 plasmid (Boeke et al., 1987), indicating that kip3 is synthetically lethal with dyn1.

Synthetic lethality of kip3 with kar3 was tested by crossing a kip3Δ:: HIS3 strain with strain DS276 containing kar3Δ102::LEU2 and a KAR3 plasmid. The resultant diploid strain was sporulated, and the tetrads were dissected. The spores were 83% viable, and all 11 viable kip3 kar3 spores carried the KAR3 plasmid. An additional 16 inviable spores were inferred to carry both the kip3 and kar3 mutations based on analysis of markers in the remaining spores of each tetrad; these spores may have been inviable because they did not inherit the KAR3 plasmid. The viable kip3 kar3 strains were sensitive to 5-FOA at 30°C, indicating that the KAR3 plasmid was essential and that kip3 is synthetically lethal with kar3. In a second test of synthetic lethality, a diploid strain homozygous for kip3 and heterozygous for kar3 was sporulated, and tetrads were dissected. Two spores in each tetrad were kip3 KAR3 and viable, and two spores inferred to be kip3 kar3 germinated and divided two to six times before ceasing growth, confirming lethality of the double mutant.

Synthetic lethality of kip3 with kip1 and cin8 was tested by crossing a kip3Δ::HIS3 strain with strain DS118 containing kip1Δ1 and a KIP1 plasmid, and strain DS379 containing cin8Δ112 and a CIN8 plasmid. The resultant diploid strains were sporulated, and 24 tetrads from each strain were dissected. The spores were >90% viable, and double kip3 kip1 and kip3 cin8 mutants were recovered at a frequency of ∼25%. Some of the double mutants did not contain the complementing plasmid, and those that did were resistant to 5-FOA, indicating the plasmid was not essential. The kip3 kip1 and kip3 cin8 double mutants were capable of vegetative growth at 16, 23, 30, and 37°C, indicating that kip3 is not synthetically lethal with kip1 or cin8.

Synthetic lethality of kip3 with kip2 and smy1 was tested by crossing kip3Δ strains with kip2Δ strain MS2309 or smy1Δ strain SLY57. The resultant diploid strains were sporulated, and 24 tetrads were dissected. The spores were >90% viable, the kip3 kip2 and kip3 smy1 double mutant progeny were recovered at a frequency of ∼25%, and the double mutants were capable of vegetative growth at 16, 23, 30, and 37°C, indicating that kip3 is not synthetically lethal with kip2 or smy1.

Synthetic lethality of dyn1 with kar3 was tested by crossing a kar3Δ strain bearing a KAR3 plasmid (strain DS276) with the dyn1Δ strain DS730. The resultant diploid was sporulated, and tetrads were dissected. Spore viability was 65%, and only 5 out of 24 tetrads yielded 4 viable spores that exhibited 2:2 segregation of mating type and the three auxotrophic markers present. Three dyn1 kar3 double mutants were recovered from these valid tetrads, and each double mutant carried the KAR3 plasmid. The double mutants were unable to grow on medium containing 5-FOA at 30°C, indicating that the plasmid was essential and that dyn1 and kar3 are synthetically lethal.

A 4,822-bp EcoRI/SpeI fragment from plasmid pDR605 was subcloned into the EcoRI and SpeI sites of pRS314 (TRP1 CEN ARS) (Sikorski and Hieter, 1989) to yield the plasmid pB893 and further subcloned into pRS416 (URA3 CEN ARS) to yield plasmid pDR622. These plasmids carry the entire KIP3 open reading frame with 1,391 bp 5′ and 966 bp 3′ flanking DNA. Hydroxylamine mutagenesis of pB893 was performed by the method of Rose and Fink (1987), and temperature-sensitive (ts) alleles of KIP3 were obtained after transformation of strain DS738 or DS716 using the plasmid shuffle technique (Boeke et al., 1987). Transformation of the kip3-20(ts) dyn1Δ strain DS765 with the KIP3 plasmid pDR622 complemented the temperature-sensitive growth, but transformation with the vector (pRS416) did not. This confirmed that the mutation conferring temperature-sensitive growth was in KIP3 and that the mutation is recessive.

Cells were fixed for immunofluorescence microscopy by adding formaldehyde to 3.7% directly to the culture medium and incubating for 2 h at 23°C, or overnight at 4°C. Cells were prepared for immunofluorescence microscopy as described (Rose et al., 1990). For the Kip3p localization experiment, Kip3p-6myc was visualized with mAb 9E10, which recognizes the myc epitope (Santa Cruz Biotechnology, Santa Cruz, CA), and microtubules were visualized with the rat antitubulin antibody YOL 1/34 (Accurate Chemical and Scientific Corp., Westbury, NY). Fluorochrome-conjugated secondary antibodies were from Jackson Immunoresearch Labs, Inc. (West Grove, PA). For double labeling of Kip3p and microtubules, controls demonstrated that cross-reactivity and light channel spill-over did not contribute to the final images when species-specific secondary antibodies were used. For all other immunofluorescence experiments, tubulin was stained with the antitubulin monoclonal antibody BIBE2, a gift of F. Solomon (Massachusetts Institute of Technology, Cambridge, MA). The secondary antibody was FITC-conjugated goat anti–mouse antibody (Accurate Chemical and Scientific Corp.), which was absorbed against fixed yeast cells for 3 h at 4°C before use to minimize background signal. DNA was stained using 4,6-diamidino-2-phenylindole (DAPI) (Boehringer Mannheim Corp.). To observe Nuf2p–GFP localization in fixed cells, cells in culture medium were fixed for 1 h with 3.7% formaldehyde, washed, and applied to polylysine-treated Teflon^®-masked slides for viewing.

For live cell microscopy of spindle pole body location, cells were applied to a microscope slide with a thin pad of 1% agarose containing SC complete medium, and a coverslip was applied and sealed with a thin bead of a mixture of equal parts Vaseline, lanolin, and paraffin at the edges. Cells were imaged at 23–25°C using a microscope (model DMRBE; Leica, Inc., Deerfield, IL) equipped with a 100×/1.4 NA objective, a monochrome CCD camera (Cohu Inc., San Diego, CA), a 100-W mercury vapor lamp, and a fluorescein filter set. Images were captured using LG3 frame grabber hardware (Scion Corp., Frederick, MD) and NIH Image software (written by W. Rasband and available at http://rsb.info.nih.gov/nih-image/) with a custom set of macro programs to control the camera exposure settings, fluorescence illumination shutter, and motorized microscope stage. At each time point, both differential interference contrast (DIC) and fluorescence images were photographed at three focal planes spaced at 1-μm intervals to increase the likelihood of detecting both spindle pole bodies. Illumination during fluorescence photography was controlled by a shutter and was 200 ms or less for each image.

Micrographs were spatially calibrated using NIH image software and a stage micrometer. Distance and angle measurements were made using Object Image software, a modified version of NIH Image available at the above internet site. Measurements to determine nuclear migration indices were made on digital images of cells by the method of Jacobs et al. (1988), and the statistical significance of the difference between the variances of nuclear migration indices was tested using the two-tailed variance ratio test at a 5% significance level.

Upon completion of the S. cerevisiae genome sequencing projects, we searched the public databases using the BLAST program to identify additional kinesin-related genes. In addition to known genes KIP1, KIP2, KAR3, CIN8, and SMY1, this search detected an open reading frame on chromosome VII (YGL216w) that is predicted to encode an 805–amino acid protein of molecular mass 91 kD. Residues 86–464 showed 39% identity to the motor domain of human kinesin and include putative ATP-binding and microtubule-binding motifs (Fig. 1,A). Using the nomenclature employed for the kinesin-related genes KIP1 and KIP2, we have designated this gene KIP3. Subfamilies of kinesin-related proteins have been defined by exceptionally high sequence conservation in the motor domain (Moore and Endow, 1996), but sequence comparisons with Kip3p did not reveal any obvious subfamily relationship. The nonmotor, COOH-terminal 340 amino acids did not exhibit obvious sequence similarity to the nonmotor regions of other kinesin-related proteins, nor to any other protein in the current DNA, protein, and EST databases. The COOH-terminal domain contains several regions predicted to form α-helical coiled coils that potentially mediate protein multimerization (Fig. 1 B).

No known mutations map to KIP3. Therefore, to determine the function of KIP3 a deletion mutant was made by replacing the KIP3 coding sequence with the HIS3 gene in a diploid strain via one-step gene replacement (see Materials and Methods). Tetrad analysis demonstrated that KIP3 is not essential for mitotic growth. The efficiency of karyogamy was tested by crossing a kip3Δ mutant to wild-type and kip3Δ strains in a qualitative limited mating assay (Conde and Fink, 1976) and was found to occur at the wild-type efficiency. Meiosis was tested by sporulating the homozygous kip3Δ diploid strain and dissecting the tetrads. The efficiency of sporulation was normal and spore viability was >90%. Chromosome instability was tested using homozygous kip3Δ diploid strains in a qualitative assay for chromosome III loss and was found to occur at the wild-type frequency.

To investigate potential functions of KIP3 that are not essential for viability, we examined mitotic cultures of the kip3Δ mutant in more detail. The growth rate of the kip3Δ mutant at 30 and 37°C in liquid YPD medium was the same as an isogenic wild-type strain, and flow cytometry showed that the mutant cultures at 30 and 37°C had the same proportion of G1 and G2 cells as wild-type cultures. However, a subtle cell cycle delay that was not evident from flow cytometry analysis was evident from the analysis of cell morphology shown in Table II, where there was an increased frequency of preanaphase cells whose buds had grown beyond the size at which nuclear division normally occurs. Cells defined as delayed in metaphase had a single nucleus and a bud >50% the diameter of the mother cell. In the wild-type strain, cells delayed in metaphase comprised 3–4% of the culture, while in the kip3Δ mutant these cells comprised 5–14% of the culture, depending on incubation temperature.

Mutation of genes that regulate microtubule function often results in altered sensitivity to drugs that destabilize microtubules (Stearns et al., 1990). Sensitivity of the kip3Δ mutant to the antimicrotubule drug benomyl was tested by spotting known numbers of cells on YPD plates containing different concentrations of benomyl (Fig. 2). Both the wild-type and mutant strains exhibited substantial growth at 30°C on 0–20 μg/ml benomyl, while only the kip3Δ mutant grew at 30 μg/ml benomyl, indicating that kip3 mutants are more resistant to benomyl than wild-type. A similar degree of increased resistance to benomyl occurred in an isogenic strain containing a deletion mutation of the kinesin-related gene KAR3. The same relative pattern of increased resistance of the kip3Δ mutant was observed when the plates were incubated at 23 or 37°C. Consistent with increased microtubule stability in the kip3Δ mutant, immunofluorescent localization of tubulin in the kip3Δ mutant showed brighter staining of astral microtubules in the kip3Δ mutant than in wild-type (Fig. 3). In budded kip3Δ and wild-type cells, the astral microtubules extended from the nucleus and passed through the neck into the bud.

In addition to nuclear division, meiosis, and karyogamy, microtubule function is required for migration of undivided nuclei to a site near the neck of the newly emerging daughter bud, and for the subsequent insertion of nuclei into the neck as spindle elongation begins (Huffaker et al., 1988; Jacobs et al., 1988; Snyder et al., 1991; Yeh et al., 1995). Anaphase spindle insertion into the bud neck requires dynein function, but no microtubule-based motor protein has yet been shown to be required for migration of the nucleus to the neck of the emerging bud. We examined nuclear migration in the kip3Δ mutant using DIC and fluorescence microscopy of fixed cells and found that the kip3Δ mutant was defective in positioning nuclei within the cell, both before and during anaphase.

To assess nuclear migration before anaphase, we measured the nucleus-to-neck distance in budded cells with undivided nuclei. The nucleus-to-neck distances were measured on digital micrographs and were corrected for differences in the sizes of the mother cells using the nuclear migration index, which is a ratio of these two values (Jacobs et al., 1988). Fig. 3 shows representative cells, and Fig. 4 shows histograms of the number of cells observed with a given nuclear migration index. Wild-type cells from exponentially growing populations exhibited the expected nonrandom placement of nuclei near the bud neck; when the culture was grown at 30°C, the mean distance between the proximal edge of the nuclear DNA mass and the bud neck was 1.1 ± 0.66 μm (mean ± SD) and the corresponding nuclear migration index was 0.17 ± 0.093 (Fig. 4). In contrast to wild-type, the kip3Δ mean distance between the nucleus and bud was 1.8 ± 1.3 μm, and the corresponding nuclear migration index was 0.23 ± 0.17. The larger standard deviation observed with the kip3Δ mutant is due to the fact that undivided nuclei tended to be positioned more randomly in the mother cell, with larger populations of cells with the nucleus located both more closely and more distantly to the neck (Fig. 4). A variance ratio test demonstrated that the differences were significant for cultures grown at 11, 30, and 37°C.

To examine potential defects of the kip3Δ mutant in positioning the nuclei during anaphase spindle elongation, we used DIC and fluorescence microscopy to score the frequency of mother cells that contained two nuclei. In wild-type cells that contain two nuclei, one nucleus is found within the mother and another is found in the daughter. In contrast, 15% of the kip3Δ cells grown at 11°C contained both nuclei within the mother cell (Table II). The binucleate cells comprised 55% of the anaphase stage cells in the asynchronously growing culture. Microtubules were observed using tubulin immunofluorescence microscopy, and we found that all binucleate mother cells with an empty bud contained an intact mitotic spindle connecting the two nuclear DNA masses (Fig. 3). Frequently, the spindle was not aligned along the mother–bud axis as in wild-type cells undergoing anaphase.

Because the kip3Δ mutant was defective in nuclear positioning, we tested whether loss of KIP3 function also caused a defect in the potentially related process of spindle orientation. A specific orientation of the nucleus and its associated spindle pole body can be detected in wild-type cells when the bud begins to emerge and the spindle pole body faces the neck. Upon assembly of a bipolar mitotic spindle, one spindle pole is typically oriented toward the neck so that the spindle is approximately aligned along the mother–daughter axis. To investigate spindle orientation, we fixed cells from exponentially growing populations and measured the orientation of preanaphase spindles with respect to the mother–daughter axis. Spindle orientation was observed by fluorescence microscopy, using strains expressing a fusion of the spindle pole body protein Nuf2p to green fluorescent protein developed by Kahana et al. (1995). The Nuf2p–GFP fusion protein complements a nuf2 null mutation, and strains containing the fusion grow normally and exhibit anaphase spindle elongation kinetics similar to spindle movements observed independently by DIC microscopy of live cells (Kahana et al., 1995; Yeh et al., 1995). The measurements in Fig. 5 show that the majority of preanaphase spindles in the wild-type strain were oriented within 30 degrees of the mother–daughter axis, while the preanaphase spindles in the kip3Δ mutant were nearly randomly oriented. Thus, KIP3 function is required for normal preanaphase spindle orientation.

The pathway of nuclear migration begins at the end of anaphase when the spindle pole bodies of the newly divided nuclei are positioned distal to the neck. After mitotic spindle disassembly, the nuclei migrate to a position ∼1 μm from the neck of the new bud. In haploid cells, the site of bud emergence is usually adjacent to the previous division site (axial budding pattern), necessitating that the nucleus traverse the diameter of the cell. The timing of this event in the cell cycle and the path and rate of movement is not apparent by examining populations of fixed cells because there is not yet a bud to provide a temporal and spatial reference point. However, by observing movement of the nucleus in a live cell after anaphase, the timing, path, and rate of nuclear movement can be investigated. To observe mitotic spindle pole movements in live cells, we used time-lapse fluorescence microscopy of strains expressing the Nuf2p–GFP fusion (Kahana et al., 1995). We collected both fluorescence and DIC images at three focal planes at each time point and overlaid the fluorescence images of spindle poles on the corresponding DIC images of cells to accurately position the spindle poles within the cell.

To examine nuclear migration in haploid G1 phase cells, we monitored the position of the spindle pole body relative to the prebud site after anaphase. In wild-type cells after maximal anaphase spindle pole separation, the spindle pole body typically moved 1–2 μm toward the cell center in 4 to 8 min (Fig. 6 A). This immediate and rapid movement preceded cytokinesis and probably corresponds to a similar movement described previously in diploid cells (Yeh et al., 1995). After the rapid movement, the spindle pole body in wild-type cells typically traversed an additional 2–3 μm to the site near the neck within 10 to 30 min, and then its position became relatively stable until anaphase, exhibiting movements of 0.5 μm or less.

In the kip3Δ mutant, the initial rapid phase of spindle pole body movement after maximal anaphase spindle elongation appeared similar to that of wild-type strains (Fig. 6). However, subsequent to the initial movement, the spindle poles exhibited large oscillations in position, with the spindle pole alternately moving closer and farther from the neck. A single spindle pole body movement toward or away from the neck was typically 1.5–2.5 μm and occurred in 10–15 min (Fig. 6, C and D). The spindle pole also exhibited lateral movements that sometimes occurred without apparent large changes in the pole to neck distance (Fig. 6 C). The highly variable position of the spindle pole relative to the neck revealed by real-time analysis is consistent with the defective nuclear positioning shown by analysis of nuclear position in fixed cells. Furthermore, the variable nuclear positioning occasionally results in the presence of the nucleus near the neck, constituting an alternate pathway for nuclear migration that could account for the viability of kip3 mutants.

Microscopy of individual live cells has revealed several discrete phases of spindle elongation (Palmer et al., 1989; Kahana et al., 1995; Yeh et al., 1995; Yang et al., 1997). When the initial fast phase of anaphase commences, spindle elongation moves one pole of the mitotic spindle to the neck, and the nucleus passes through the narrow opening. The fast phase of elongation continues until the spindle is ∼4 μm long, and then elongation continues at a slower rate, sometimes with a pause between the two phases. To determine whether loss of KIP3 function affects the kinetics of spindle elongation, we used live cell microscopy of the Nuf2p–GFP fusion to observe these movements in a kip3Δ mutant. To focus specifically on anaphase, cells with a short bipolar spindle located near the neck were selected for time-lapse photography. The average rates of fast and slow spindle elongation were similar in wild-type and kip3 strains (Fig. 7 A). Fast elongation in the kip3 mutant was 0.69 ± 0.25 μm/min (n = 10) over the first 6–8 min of elongation, and in the wild-type strain it was 0.60 ± 0.19 μm/ min (n = 6). Slow elongation in the kip3 mutant was 0.14 ± 0.06 μm/min (n = 6), and in the wild type strain it was 0.14 ± 0.08 μm/min (n = 4). Thus loss of KIP3 function does not dramatically affect anaphase B kinetics.

During the early period of anaphase spindle elongation, when the dividing nucleus is in the neck, the nucleus and spindle can exhibit forward and reverse motion along the mother–daughter axis. These movements range in distance from 0.5–1 μm in wild-type diploid cells, and they are absent in dyn1 mutants (Yeh et al., 1995). We observed few forward and reverse motions of the nucleus relative to the neck in the dyn1 mutant, but we did observe such movements in wild-type and kip3 strains (Fig 7, B–I), suggesting that KIP3 is not essential for these movements.

The early anaphase spindle undergoes proportionally greater elongation before insertion of the nucleus into the neck in dyn1 mutants compared to wild-type (Yeh et al., 1995). To identify possible defects in this process in the kip3 mutant, we measured the pole-to-pole length of the spindle at the time that the daughter-bound pole passed through the neck (Fig. 7). Some elongation of the spindle occurred before insertion into the neck in all strains examined. Spindle length upon neck contact in the kip3 mutant was 2.9 ± 0.8 μm (n = 10), and in the wild-type control it was 2.3 ± 0.5 μm (n = 5). This difference in spindle lengths was not statistically significant (unpaired t test, P = 0.12), indicating that loss of KIP3 function has little effect on insertion of the spindle through the neck. In contrast, the spindle length of an isogenic dyn1 mutant at the time of insertion was 4.1 ± 0.7 μm (n = 8), which was significantly different from wild-type (P = 0.0005). Translocation was complete within 5 min in the wild-type, kip3, and dyn1 cells observed, indicating that transient elongated nuclei at the beginning of successful anaphase events make only a small contribution to the populations of binucleate cells reported in Table II.

Mutants defective in the gene for the dynein heavy chain (dyn1 or dhc1) have been reported to be defective in nuclear orientation, insertion of the dividing nucleus into the neck, and oscillatory movement of the anaphase-stage nucleus within the neck (Eshel et al., 1993; Li et al., 1993; Yeh et al., 1995). To compare the effect of loss of KIP3 function to loss of DYN1 function in isogenic strains, we constructed a dyn1Δ mutant and scored the frequency of binucleate cells and the position of undivided nuclei within mother cells. Consistent with the previously reported defects in anaphase-stage cells, the dyn1Δ mutant cultures contained binucleate mother cells at a frequency of 11– 31%, depending on incubation temperature (Table II). Proficiency in migration of undivided nuclei was assessed by determining the mean nuclear migration index and standard deviation. The histograms in Fig. 4 show that the dyn1Δ mutant grown at 11, 30, and 37°C showed a distribution of nuclear migration indices that was similar to the wild-type strain. The ratio of variances between wild-type and dyn1Δ strains at each temperature was below the 95% significance point, indicating no significant difference was present. Furthermore, spindle pole movement observed by microscopy of live dyn1 cells containing the Nuf2p– GFP fusion showed that the path and rate of movement was similar in dyn1 and wild-type cells (Fig. 6, E and F), although we did observe that the time between anaphase spindle disassembly and bud emergence was prolonged in the dyn1 mutant. Thus, in contrast to the kip3Δ mutant, the dyn1Δ mutant is not significantly impaired in migration of undivided nuclei to the site near the bud neck. This is consistent with previous measurements of nuclear distribution in hydroxyurea-arrested dyn1 cells, where nuclear translocation into the bud decreased, but accumulation of nuclei at the bud neck was largely unimpaired (Yeh et al., 1995).

To determine whether orientation of the preanaphase mitotic spindle is affected by loss of dynein function, we used the Nuf2p–GFP fusion protein employed above for the kip3 mutant to measure spindle orientation in fixed dyn1Δ mutant cells. Preanaphase spindles in the dyn1Δ mutant were oriented within 30 degrees of the mother– daughter axis in 68% of the cells, compared to 72% in wild-type cells. There was a small increase in the number of cells with a spindle misoriented by 60–90 degrees (13%, compared to 8% in wild-type). Thus, loss of dynein heavy chain function has little effect on preanaphase spindle orientation.

The phenotypes presented above of mutants singly defective in KIP3 or DYN1 indicate that KIP3 is required for nuclear migration to the bud neck and preanaphase spindle orientation, but that DYN1 is not. However, Kip3p and Dyn1p may perform overlapping or dependent functions that are not evident in the single mutants. For instance, the binucleate mother cell phenotype manifested by both kip3 and dyn1 mutants may reflect a common function in insertion of the anaphase nucleus through the bud neck. Alternatively, the cumulative effect caused by loss of the sequential migration and insertion steps of nuclear migration may be greater than that caused by loss of migration or insertion individually. Similarly, KIP3 could overlap with any of the other kinesin-related genes to perform an essential function, and this overlap would only result in a phenotype upon loss of both activities.

We tested for synthetic phenotypes by constructing double deletion mutants in the presence of a plasmid-based copy of one wild-type gene and then determined whether the strain remained viable after plasmid segregation (see Materials and Methods). The kip3 dyn1, kip3 kar3, and kar3 dyn1 strains all required a complementing plasmid for viability (Fig. 8 A), indicating that these combinations were synthetically lethal. The kip3 kip1, kip3 cin8, kip3 smy1, and kip3 kip2 double mutants were all viable and formed colonies equal in size to the single mutants at 16, 23, 30, and 37°C (at 37°C the cin8 and kip3 cin8 strains showed equally poor growth), indicating that these pairs of genes are not solely responsible for an essential function.

To distinguish whether kip3 and dyn1 mutations are synthetically lethal because of a functional overlap or due to a cumulative effect caused by partial loss of sequential functions, we examined a conditional double mutant. Temperature-sensitive kip3(ts) dyn1Δ strains were constructed by the plasmid shuffle technique (Fig. 8 B). A culture of the kip3-20(ts) dyn1Δ strain grown at 23°C contained 48% inviable cells, while kip3Δ and dyn1Δ single mutant cultures contained 95–100% viable cells. Incubation of the double mutant at 37°C for 4 h resulted in a doubling of cell number and a reduction of cell viability to 25%. Continued incubation at 37°C resulted in a second doubling of cell number by 16 h, at which time cell number ceased to increase and cell viability remained at 25%. This suggests that the synthetic lethality is not solely due to induction of a checkpoint that arrests cell cycle progression and prevents inviability.

Preanaphase nuclear migration was examined by microscopy of the double mutants after incubation at the nonpermissive temperature. The position of undivided nuclei was significantly more random in the kip3(ts) dyn1Δ double mutant than in the dyn1 single mutant and wild-type strains (Fig. 4). However, the variance of the kip3(ts) dyn1Δ double mutant was not significantly different from the kip3Δ single mutant. This indicates that combination of dyn1Δ and kip3(ts) mutations causes no greater defect in nuclear migration to the neck than that caused by the kip3Δ mutation alone.

Both kip3 and dyn1 mutations individually cause a small delay in the nuclear division cycle before anaphase (see above and Yeh et al., 1995). The combination of kip3(ts) dyn1Δ mutations also caused a delay in the nuclear division cycle, which we observed as an increased frequency of preanaphase cells whose buds had grown beyond the size at which nuclear division normally occurs. Double mutant cells incubated at 37°C for 4 h showed the delay at a frequency 2.5-fold higher than the kip3 and dyn1 single mutants and 10-fold higher than the wild-type strain (Table II). The kip3(ts) dyn1 cells incubated at the nonpermissive temperature exhibited an abnormal morphology that was not observed in either single mutant. Mother cells and mature buds were enlarged and elongated, but small buds had a normal ellipsoidal shape (Fig. 9). Abnormal cell elongation can be caused by mutations affecting a variety of functions (Blacketer et al., 1995), including defects in the switch from apical to isotropic growth that occurs at the G2/M transition (Lew and Reed, 1993).

To determine whether KIP3 and DYN1 overlap in function for insertion of the nucleus into the neck during anaphase spindle elongation, we measured the frequency of mother cells containing two discrete nuclear DNA masses. The kip3(ts) dyn1 double mutant culture incubated at 37°C for 4 h contained 10% binucleate mother cells (Table II), compared to 6% binucleate cells in kip3 and dyn1 single mutants. An independent kip3(ts) allele in a dyn1Δ strain had a similar cell type distribution. Since the double mutant culture incubated for 4 h at 37°C still contained some anaphase cells, we examined a culture incubated for 8 h at 37°C and found that the binucleate mother cells increased to 23%. Thus, simultaneous loss of KIP3 and DYN1 function causes defective insertion of anaphase spindles through the bud neck, in addition to causing defective preanaphase nuclear migration and a metaphase delay. This suggests that the synthetic lethality of kip3 dyn1 mutants is a cumulative effect caused by loss of partially overlapping and sequential functions.

Since one phenotype of the kip3Δ single mutant and the kip3(ts) dyn1Δ double mutant is increased microtubule abundance (Figs. 3 and 9), we tested whether the microtubule-depolymerizing drug benomyl could suppress the lethality of the kip3(ts) dyn1Δ double mutant. The double mutant showed little growth at 37°C in the absence of benomyl but exhibited significant growth on media containing 5 or 10 μg/ml benomyl (Fig. 8 C). This suggests that altered microtubule polymerization in the double mutant is a contributing factor to the synthetic lethality.

The kinesin-related Kar3p is a candidate to provide some of the same functions as Kip3p because loss of either KIP3 or KAR3 causes increased astral microtubule abundance (above and Saunders et al., 1997), and Kar3p is known to mediate one nuclear movement, nuclear congression, during karyogamy (Meluh and Rose, 1990). To determine the nature of the defects caused by simultaneous loss of KIP3 and KAR3 function, the plasmid shuffle technique was used to isolate a temperature-conditional double mutant (Fig. 8,B). Like the kip3(ts) dyn1 double mutant, lethality of the kip3-30(ts) kar3Δ double mutant was partially suppressed on medium containing a low concentration of benomyl (Fig. 8,C). To examine morphological defects, the kip3(ts) kar3Δ double mutant was incubated at the nonpermissive temperature, and the cells were compared to kar3Δ and kip3Δ single mutants by microscopy. The kar3Δ single mutant showed a high proportion of cells before anaphase (Table II), with many of the nuclei in very close proximity to the bud neck (Fig. 4), as previously reported (Meluh and Rose, 1990). When the kip3 kar3 double mutant was incubated at 37°C for 4 h, fewer nuclei were in very close proximity to the neck, and more were at the distal edge of the mother cell (Fig 4). This suggests that the migration defect caused by kip3 mutation is epistatic to the accumulation of nuclei in very close proximity to the bud caused by kar3 mutation. No other aberrant morphology unique to the double mutant was detected (Fig. 9). The data in Table II show that the frequency of cells in metaphase in the kip3 kar3 double mutant culture was similar to that in kar3 single mutant, and the frequency of binucleate cells was similar to that in the kip3 single mutant. Thus, the combination of kip3 and kar3 mutations caused the aberrant morphologies characteristic of the individual mutations to be superimposed and did not result in a detectable synthetic morphological phenotype.

Given the possibility of functional overlap suggested by the synthetic lethality of kip3 and kar3, we examined whether Kip3p could perform a previously described function of Kar3p in contributing to the balance of inwardly and outwardly acting forces that maintain mitotic spindle pole separation during metaphase. Mutations in kinesin-related genes KIP1 and CIN8 can cause collapse of the metaphase spindle, suggesting that Kip1p and Cin8p act to maintain spindle pole separation, and mutations in KAR3 can suppress this collapse phenotype and restore growth, suggesting that Kar3p generates an inward force on the spindle poles (Saunders and Hoyt, 1992; Hoyt et al., 1993). Mutations in KAR3 may suppress the kip1 cin8 collapse phenotype by stabilizing microtubules, and a mutation in the β-tubulin gene that stabilizes microtubules has an effect similar to kar3 mutation (Saunders et al., 1997). To determine whether Kip3p can perform a role similar to Kar3p in countering the forces generated by Kip1p and Cin8p, we tested whether a kip3Δ mutation can suppress the kip1 cin8 growth defect. Meiotic crosses were used to construct kip3 kip1 cin8 and kar3 kip1 cin8 triple mutants, and the sensitivity of growth to temperature was examined (Fig. 8 D). The kar3Δ mutation partially suppressed the temperature-sensitive growth of the kip1 cin8 mutant as expected, but the kip3Δ mutation did not. Thus, KIP3 fails to show genetic interactions with KIP1 and CIN8, indicating that Kip3 does not overlap with the suggested role of Kar3p in generating an inwardly directed spindle force, and further suggests that Kip3p and Kar3p perform separate functions.

The phenotypes of the kip3Δ mutant suggested roles for Kip3p primarily in astral microtubule function. To determine if Kip3p is associated with microtubules in vivo, we determined the intracellular localization of a functional epitope-tagged protein. A single-copy plasmid expressing myc-tagged Kip3p (KIP3-6MYC, see Materials and Methods) from its own promoter was introduced into a kip3Δ strain. Western blotting of an extract from this strain identified a single band at the predicted molecular mass of 100 kD (data not shown). In G1 cells, Kip3p colocalized with the astral microtubules (Fig. 10, A–C). In late S phase or early mitotic cells, Kip3p colocalized with short bipolar spindles (Fig. 10, D–F). In late anaphase or telophase cells, Kip3p colocalized with spindle microtubules and was frequently concentrated near the spindle midzone, the zone of overlap between microtubules from opposite poles (65% of late mitotic cells where Kip3p spindle localization was observed, n = 250; Fig.10, G–I). In the remaining late mitotic cells, Kip3p staining was evenly distributed along the length of the elongated spindle (Fig. 10, J–L). In addition to microtubule staining, expression of KIP3-6MYC resulted in bright punctate staining of the nucleus and cytoplasm of all cells observed. This punctate staining was not observed over the vacuole and was specific to Kip3p-6myc because it was not observed in an untagged KIP3 control strain (Fig. 10, M–O). We were able to detect microtubule localization of Kip3p in 11% of the cells examined (n = 300). It is possible that microtubule staining in many cells was obscured by the generalized staining.

A search of the complete yeast genome sequence revealed KIP3 as the single previously uncharacterized gene encoding a protein with strong sequence conservation with the force-generating domain of kinesin. Thus, budding yeast contains six genes encoding kinesin-related proteins (KIP1, KIP2, KIP3, KAR3, CIN8, and SMY1), and it is the first organism whose complete set of kinesin-related proteins has been identified. Three of the kinesin-related proteins are members of known kinesin subfamilies defined by sequence conservation; KIP1 and CIN8 belong to the BimC subfamily and KAR3 belongs to the COOH-terminal subfamily (Moore and Endow, 1996). Members of these two subfamilies show functional conservation as well as sequence conservation, with members of the BimC subfamily playing roles in mitotic spindle assembly and members of the COOH-terminal subfamily playing roles in spindle pole function. The kinesin-related proteins encoded by KIP2, KIP3, and SMY1 lack any strong subfamily homology.

The relatively small number of microtubule-mediated movements that occur in yeast can now be ascribed to the force generated by the six kinesins, dynein, and potentially to forces generated by microtubule polymerization and depolymerization. No single motor protein is essential for viability, perhaps because of cooperation between different force-generating mechanisms, such as the overlap between KIP1 and CIN8 for mitotic spindle assembly (Hoyt et al., 1992; Roof et al., 1992). Given the complete set of motor proteins, it is now possible to systematically dissect the contribution of each motor protein to a particular microtubule-based movement.

We used two experimental approaches to characterize the effect of loss of KIP3 function on nuclear migration. In the first approach, we examined populations of fixed cells to assess the position of the nucleus and orientation of the mitotic spindle relative to the bud. Undivided nuclei in budded wild-type cells had a strong tendency to be located ∼1 μm from the neck, and the associated mitotic spindle had a strong tendency to be oriented within 30 degrees of the mother–daughter axis, consistent with previous studies of S. cerevisiae morphology (Byers and Goetsch, 1975; Jacobs et al., 1988). Both position and orientation of the nucleus became essentially random upon loss of KIP3 function. Since these experiments relied on use of the bud as a reference point to quantitate the position and orientation of undivided nuclei, they define an essential role for KIP3 for directed nuclear migration and orientation during the cell cycle interval from Start to just before anaphase. We were unable to use fixed cells to quantitate nuclear movements in G1 phase cells because these cells lack a bud to use as a reference point.

In the second experimental approach to characterize nuclear migration, we used live cell microscopy to follow movement of the spindle pole body in individual cells. In addition to defining the path and kinetics of movement during nuclear migration, this approach enabled us to examine nuclear positioning during the G1 phase of the cell cycle. By observing nuclear migration in a live cell through the point of bud emergence, we could infer the site of bud emergence for use as a reference point in G1 phase cells. Bud emergence in haploid cells follows the axial pattern of bud site selection, in which the new bud emerges adjacent to the previous division site (Chant and Pringle, 1995). A consequence of axial budding is that anaphase spindle elongation places the nucleus at the edge of the cell opposite to the site of bud emergence.

The first spindle-independent movement of the nucleus occurs immediately after spindle disassembly, when the nucleus rapidly moves toward the cell center (Fig. 6 and Yeh et. al., 1995). We observed this movement in wild-type, kip3, and dyn1 strains. In the subsequent period of movement, the nucleus continued toward the prebud site, with steady forward movement and no large reversals in direction. The second period of movement sometimes occurred immediately after the initial movement, so that migration was complete within 15 min of spindle disassembly, before a bud could be seen. In other examples, there was a delay of about 20 min before the second period of movement occurred. The new bud can be detected only at the end of G1, but actin, Spa2p, and neck filament proteins localize to the prebud site before bud emergence (Kim et al., 1991; Snyder et al., 1991; Lew and Reed, 1993). Although there are differences in the reported timing of assembly of proteins at the prebud site, the intersection of astral microtubules with the prebud site provide a potential mechanism for directed nuclear migration during G1. When the nucleus reached a site ∼1 μm from the bud neck, its position became relatively stable. We observed similar patterns of preanaphase movement in wild-type and dyn1 strains.

Loss of KIP3 function caused a dramatically different course of phase 1 nuclear migration. After the kip3 mutant completed the rapid inward movement after spindle disassembly, nuclei exhibited large forward and reverse oscillations in position relative to the bud neck (see Fig. 6, C and D). The nucleus sometimes came within ∼1 μm of the neck, only to reverse direction and move 1–2 μm in the opposite direction. Since the nuclei were located far from the neck for long time intervals, the oscillations are consistent with the random position of nuclei observed in populations of fixed kip3 cells. Furthermore, the oscillations may account for the viability of mutants lacking KIP3 function because nuclei are occasionally at the “correct” location ∼1 μm from the neck.

Nuclear migration is a pathway comprised of a series of distinct movements that occur during interphase and mitosis. We found that mutations in KIP3 predominantly affected the first phase of nuclear migration as the undivided nucleus moved to a site ∼1 μm from the neck, and the metaphase spindle oriented along the mother–daughter axis (Fig. 11, Phase 1). In contrast, the dynΔ mutation predominantly affected the second phase of nuclear migration as the spindle began to elongate during anaphase, and the nucleus inserted into the neck, consistent with the previous detailed analysis of nuclear movement by Yeh et al. (1995). We detected only a small effect of kip3 mutation on the second phase of nuclear migration, and little effect of dyn1 mutation on the first phase of nuclear migration. Thus, the phenotypes of mutants singly defective in KIP3 and DYN1 fell into two classes that define distinct steps in the nuclear migration pathway.

The nuclear migration defects of dyn1 mutants during anaphase support the hypothesis that dynein is part of the force-generating machinery that squeezes the nucleus through the narrow neck opening. However, we found very little effect of loss of dynein function on nuclear position or spindle orientation from the time of cytokinesis to anaphase initiation. This is in contrast to Aspergillus nidulans and Neurospora crassa, where dynein is required after nuclear division for migration of nuclei into the mycelium (Plamann et al., 1994; Xiang et al., 1994). Our results do not contradict most previous reports of incorrect spindle orientation in S. cerevisiae mutants lacking dynein or dynactin function because these studies examined the orientation of anaphase spindles in binucleate mother cells and did not address nuclear position or spindle orientation in preanaphase cells (Eshel et al., 1993; Li et al., 1993; McMillan and Tatchell, 1994; Muhua et al., 1994). However, Yeh et al. (1995) suggest that orientation of the spindle as the nucleus contacts the neck during early anaphase is defective in dyn1 mutants. We have not examined spindle orientation as anaphase initiates in an unbiased set of kip3 or dyn1 cells and do not rule out a possible role for dynein in spindle orientation as the elongating anaphase nucleus docks onto the neck. The absence of a detectable effect of dynein mutation on interphase nuclear migration may be explained by functional overlap between dynein and other force-generating mechanisms.

Mutants defective in KIP3 or DYN1 have one phenotype in common: the generation of mother cells that contain two nuclei connected by a spindle that is frequently not aligned along the mother–daughter axis. Our analysis of dyn1 mutants suggests that binucleate cells in this mutant may arise not because of spindle misorientation, but instead because of failed translocation of the anaphase-stage nucleus through the neck. However, formation of binucleate mother cells upon loss of KIP3 function may be caused by a defect in an independent mechanism. The defects of kip3 mutants in phase 1 nuclear migration suggest that binucleate cells could be formed when anaphase is initiated while the nucleus is improperly located or oriented. The mechanism by which binucleate mother cells arise in kip3 and dyn1 mutants could be addressed more directly by observing nuclear position and spindle orientation during anaphase initiation when a binucleate mother cell is formed, but we have been unable to record such an event in either kip3 or dyn1 strains by microscopy of individual live cells. Mutations in many other S. cerevisiae genes cause binucleate mother cells to accumulate (for review see Morris et al., 1995). It would be interesting to use the approaches described here to test whether these mutations affect the first or second phase of the nuclear migration pathway.

A model for how Kip3p controls nuclear movement and orientation should take into account the observations that the kip3 mutant exhibits large oscillations of nuclear position and has higher resistance to the microtubule-depolymerizing drug benomyl than to wild-type strains. Both of these phenotypes could be the result of altered microtubule dynamics. One explanation for the benomyl resistance is that microtubules are more stable because loss of KIP3 function eliminates a microtubule depolymerizing activity. Destabilization of microtubules by kinesin-related proteins has been described for Kar3p at the minus ends of microtubules and for Xenopus XKCM1 protein at the plus ends, and altered astral microtubule dynamics occur in S. cerevisiae mutants lacking dynein (Endow et al., 1994; Walczak et al., 1996; Carminati and Stearns, 1997). Although Kip3p and Kar3p may both affect microtubule dynamics, these proteins differ in that Kar3p is concentrated at the spindle poles (Saunders et al., 1997), and Kip3p is located throughout the cytoplasm and nucleus and is present on both astral and spindle microtubules. Furthermore, loss of KAR3 function has a different effect on the balance of forces in the mitotic spindle perturbed by kip1 and cin8 mutation than does loss of KIP3.

One model for Kip3p function is that Kip3p interacts with the plus ends of microtubules, enabling it to control astral microtubule dynamics and thereby influence the forces applied to the nucleus. In the absence of Kip3p, the positional oscillations of preanaphase nuclei could be powered by unregulated polymerization and depolymerization of the astral microtubules that extend into the bud and terminate near the cortex. Other motor proteins could also power the positional oscillations, but the morphology of kip3 dyn1 and kip3 kar3 double mutants suggests that Dyn1p or Kar3p are not individually responsible for powering the nuclear oscillations (see below). In addition to regulating microtubule dynamics, Kip3p may directly generate the force applied to the nucleus for nuclear migration. The wide distribution of Kip3p in the cell suggests that astral microtubules interact with Kip3p either transiently, or with a specific subpopulation of Kip3p. One possibility is that a subpopulation of Kip3p is attached to an asymmetrically distributed cortical protein such as Num1p or proteins located at the prebud site (Snyder et al., 1991; Farkasovsky and Kuntzel, 1995; Pringle et al., 1995), and that these Kip3p molecules push or pull the nucleus toward the bud neck.

While mutants individually defective in KIP3 or DYN1 defined distinct steps of nuclear migration, these proteins could perform additional roles that are not evident in the single mutants because of functional overlap among different force generating mechanisms. We explored this possibility by testing for synthetic lethality between the kip3Δ mutation and mutations in the DYN1 gene and each of the kinesin-related genes, and we found that kip3 is synthetically lethal with dyn1 and kar3. However, in the W303-1a strain background, we could recover very slow-growing but viable kip3Δ dyn1Δ double mutants (data not shown). To determine whether the synthetic lethality was due to the loss of a single essential movement, we constructed conditional double mutants and examined cell and microtubule morphology. In both the kip3 dyn1 and kip3 kar3 conditional double mutants, there was no accumulation of a structural intermediate in nuclear migration or mitosis that suggested that a specific movement was abolished. The kip3 dyn1 conditional double mutant culture contained cells delayed in metaphase at a frequency 2.5 times that observed in the single mutants, and there was a higher frequency of cells with two separate DNA masses in the mother cell than in either single mutant (a fourfold increase). However, the binucleate mother cells may arise as a consequence of separate defects. Preanaphase nuclear positioning was essentially random in both the kip3 single mutant and in the kip3 dyn1 double mutant. Since no single detectable defect in the kip3 dyn1 double mutant appears to account for the synthetic lethality, we suspect that the synthetic lethality is caused by the cumulative effect of partial defects in sequential movements. Synthetic lethality due to partial loss of sequentially acting functions has been described (Simon et al., 1991). A similar conclusion is suggested by the morphology of the kip3 kar3 conditional double mutant, which essentially exhibits a superimposition of the defects observed in the single mutants. However, we cannot rule out the possibility that Kip3p and Dyn1p or Kip3p and Kar3p overlap to perform a movement or function whose loss does not generate a detectable morphological defect. Furthermore, the localization of Kip3p to central spindle microtubules suggests that our analysis has not revealed the full functional repertoire of Kip3p.

Nuclear positioning and spindle orientation are key regulated steps of cell growth and development in many organisms. Nuclear migration plays an important role for fertilization in higher cells and in yeast for the analogous process of karyogamy (Meluh and Rose, 1990; Rouviere et al., 1994). Nuclear migration is also central to development. In Caenorhabditis elegans, asymmetric positioning of the nucleus during the first cell division of the zygote results in the production of daughter cells that differ in size and developmental fate (Hyman and White, 1987). During the syncytial stage of development in Drosophila, nuclear migration determines the position of both the germ cells and the somatic cells (Baker et al., 1993). Also in Drosophila, patterned changes in spindle orientation mediate developmental fate in the parts of the nervous system (Kraut et al., 1996).

Although it is thought that both the tubulin and actin cytoskeletons are important for nuclear migration, few of the force-generating molecules required for nonmitotic nuclear movement have been identified. The S. cerevisiae kinesin–related protein Kar3p is essential for karyogamy (Meluh and Rose, 1990), and A. nidulans and N. crassa cytoplasmic dynein is required for movement of nuclei in the germ tube (Plamann et al., 1994; Xiang et al., 1994). In this study, we found that another yeast kinesin–related protein, Kip3p, plays a critical role in spindle orientation and nuclear position before anaphase. Kinesin-related proteins may therefore have a general role in nuclear migration. Furthermore, because kinesin-related proteins are found in all eukaryotes, these motors might also control developmentally regulated nuclear movement in multicellular organisms.

The authors would like to thank J. Kahana and P. Silver for the NUF2– GFP plasmid, E. Yeh and K. Bloom for the dynΔ plasmid, J. Carminati and T. Stearns for sharing unpublished results, and D. Gordon, B. Bucher, and T. Yen for comments on the manuscript.

This work was supported by grants from the National Institutes of Health (R55-GM50884) and the University of Pennsylvania Research Foundation to D. Roof. D. Pellman is supported by a Damon Runyon Scholar Award, and T. DeZwaan was supported by a training grant from the National Institutes of Health (GM07229).

5-FOA

5-fluoroorotic acid

DAPI

4,6-diamidino-2-phenylindole

DIC

differential interference contrast

GFP

green fluorescent protein

ts

temperature-sensitive