Kinesins klp5⁺ and klp6⁺ are required for normal chromosome movement in mitosis

Chromosome segregation in mitosis is achieved through movements that depend on dynamic interactions between spindle microtubules and kinetochores(reviewed by Skibbens and Heiter,1998; Mitchison and Salmon,2001). A few spindle microtubules attach to each kinetochore by their plus ends (i.e. the fast growing ends), while corresponding microtubule minus ends lie proximal to the spindle poles. Kinetochore attachments to plus microtubule ends are dynamic because changes in chromosome position require changes in microtubule length, both during prometaphase congression to the spindle equator and during anaphase A. Studies with labeled tubulin have shown that much of this length change occurs by tubulin subunit gain and loss at the kinetochores, so this chromosomal specialization must be able to retain microtubule binding while it allows tubulin subunit exchange. Several studies have implicated microtubule-dependent motor enzymes as components of this labile attachment mechanism (reviewed byManey et al., 2000;Mitchison and Salmon, 2001). Kinetochore associated motors also appear to be instrumental in setting up the spindle assembly checkpoint, the sensing mechanism that assures proper chromosome attachment and regulates the initiation of sister chromatid separation (Skibbens and Heiter,1998; Pidoux and Allshire,2000; Wassmann and Benezra,2001). Therefore, understanding the details of kinetochore motor composition is likely to be essential for elucidating the mechanisms of mitosis.

Two kinesin-like protein families, CENP-E(Yen et al., 1992) and KinI(MCAK and XKCM1) (Wordeman and Mitchison,1995; Walczak et al.,1996) have been localized to mitotic kinetochores in metazoans(reviewed by Goldstein and Philip,1999). Functional studies implicate CENP-E in prometaphase congression to the spindle equator and the maintenance of metaphase(Wood et al., 1997;Schaar et al., 1997;Yao et al., 2000;Yucel et al., 2000;McEwen et al., 2001), and MCAK/XKCM1 in anaphase A chromosome movement towards the spindle poles(Walczak et al., 1996;Maney et al., 1998).

The fission yeast, Schizosaccharomyces pombe, is an excellent model system whose suitabilities for genetic and cytological studies make it particularly valuable for the study of kinetochore function. For example, S. pombe contains just three chromosomes that segregate on a highly ordered mitotic spindle, so the cytological analysis of chromosome movement and kinetochore function is comparatively easy. Through the use of kinetochore markers (Goshima et al., 1999)sister chromatid separation can be monitored for each replicated chromosome simultaneously, thus allowing distinctions between pre-anaphase motions and the segregation of sister chromatids at anaphase. In addition, the S. pombe kinetochores resemble those of metazoans in several ways: they are built upon tens of kilobases of DNA, which is organized as a central core flanked by large repeating units; they attach 2-4 microtubules; and are visible as a region with special staining characteristics(Ding et al., 1993).

Here we describe the mitotic role(s) of two kinesin-like motors in fission yeast, klp5⁺ and klp6⁺. Previous work has shown that these two proteins have similar primary structures but that neither of them is essential, either alone or together. Null alleles do,however, have unusually stable microtubules; they are highly resistant to the microtubule drug, thiabendazole, and they contain unusually long microtubules in interphase (West et al.,2001). Both Klp5p and Klp6p localize to cytoplasmic microtubules in interphase, and their deletion leads to morphological defects in especially long cells. These results led to the hypothesis that Klp5p and Klp6p foster microtubule disassembly in vivo. In this paper, we present data indicating that both of these kinesins are also necessary for proper sister chromatid separation in mitosis. We propose that the defects in chromosome movement are the result of changes in spindle dynamics and/or the interaction between the spindle and the kinetochores.

Cell culture and genetic manipulations were performed using standard techniques (Moreno et al.,1991). The construction of the null and (green fluorescent protein) GFP-tagged alleles of klp5⁺ and klp6⁺, and other mutants with klpΔbackgrounds is described elsewhere (West et al., 2001). Spore viability tests were carried out as described previously (West et al.,2001). Cell transformations were carried out using the lithium acetate/sorbitol (Moreno et al.,1991) or polyethylene glycol(Elble, 1992) protocols. The klp5Δ, klp6Δ and klp5Δ klp6Δ strains are collectively referred to as`klpΔ' when either the same experiment was done with all three strains or the same general conclusion is being drawn from experiments done with all three strains.

Cell cycle arrest and release experiments with cdc25-22 were done as follows. Cultures were grown in YES medium to early log phase(OD₅₉₅ 0.1-0.2) at permissive temperature (25°C), shifted to restrictive temperature (36°C) for 3 hours, then returned to permissive temperature. Samples were collected at 10 minute intervals, fixed with methanol as described below, and stained for DNA with DAPI (4′,6-diamino-2-phenylindole) (Sigma, St Louis, MO). Mitotic cells were identified by having >1 mass of DNA, but no septum. Abnormal mitotic cells were defined as those containing two or more asymmetric masses of DNA.

For microscopy, cells were grown to early-mid log phase(1-3×10⁵cells/ml). Observations on strains expressing GFP tagged genes were done at 25°C unless otherwise noted. DNA staining was performed on cells fixed by adding one-tenth of the volume of a cell culture to methanol at -20°C and incubating at -20°C for 2-15 minutes. Cells were collected by centrifugation, resuspended in phosphate-buffered saline (pH 7.4) containing 1 μg/ml DAPI and mounted on glass slides.

Live cells were analyzed by visualizing microtubules with GFP-α-tubulin (Ding et al.,1998), Klp5p and Klp6p with their corresponding GFP fusions(West et al., 2001),kinetochores with Mis 12p-GFP (Nabeshima et al., 1998), and DNA with Hoechst 33342 (Sigma). The GFP-α-tubulin plasmid pDQ 105 (Ding et al., 1998) was transformed into cells, and the resultant strains were grown in defined medium(Moreno et al., 1991)containing 5 μg/ml thiamine (Sigma) to limit expression levels(Maundrell, 1990) of the GFP-α-tubulin. Cells were collected by centrifugation (5-10 ml) at 2056 g for 2 minutes in a Beckman CS-6 centrifuge, and resuspended in 5 ml YES, pH 7.5, containing 2 μg/ml Hoechst 33342 for ∼20 minutes. Cells were then mounted on glass coverslips, and images collected with a Cooke SensiCam slow-scan CCD camera on a Zeiss AxioplanII fluorescence microscope,using the SlideBook software package (Intelligent Imaging Innovations, Denver,CO) as previously described (West et al.,2001).

Wild-type and klpΔ cells were followed through mitosis with GFP-α-tubulin and Hoechst 33342 staining by deconvolution microscopy as described above. Image stacks of eight focal planes each were collected for every time point and processed into a 2D image as described above. The collection of time points were compiled in Corel Photo-Paint and exported as a QuickTime movie. The wild-type series begins towards the end of anaphase A. The klp5Δ and klp6Δ series each start with a single mass of DNA asymmetrically placed on a spindle. The DNA subsequently undergoes three asymmetric divisions to produce even segregation of the DNA. The klp5Δ spindle elongates and breaks (yellow arrow) at the end of the time course. These movies are at available at http://jcs.biologists.org/supplemental; selected images are represented inFig. 2C.

Given that kinesin-like proteins are required for several steps in mitosis,we have sought evidence for roles of klp5⁺ and klp6⁺ by characterizing mitosis in both fixed and live cells deleted for one or both of these genes. The phenotypes described below were found to be the same for klp5Δ, klp6Δ and the klp5Δ klp6Δ double mutants, so these mutants are subsequently referred to as `klpΔ'. In mitotic fission yeast, the chromosomes appear at metaphase as a single mass of DNA at the equator of the spindle; this mass then divides into two equal masses that lie at either end of a short spindle (∼2.5 μm; anaphase A)(Fig. 1A;Fig. 2A). Subsequently, the DNA masses move to the ends of the cell as mitosis is completed (anaphase B)(Fig. 2A)(Hagan, 1998;Nabeshima et al., 1998).

The klpΔ cells likewise contain a single mass of DNA as they enter mitosis, but fixed cells also frequently displayed more than one mass of DNA distributed over an elongated spindle; the most common abnormality observed was three unequal masses of DNA separated over 4-6 μm(Fig. 1B;Table 1). Other complex and asymmetric arrangements of DNA were occasionally seen in mutant cells(Fig. 1C), but on longer spindles the DNA was usually present as two equal masses, as observed for wildtype (Fig. 1D;Table 1). These data suggest that klpΔ cells do not progress normally from prophase chromosome condensation to a symmetric segregation of sister chromatids in anaphase A, but that by late anaphase B mitosis is often like wildtype.

The abnormal pattern of chromosome segregation was further examined by characterizing live klpΔ cells as they progressed from early mitosis to the formation of daughter cells (see movies;http://jcs.biologists.org/supplemental/). Wild-type fission yeast mitosis has been described previously as occurring in three distinct stages(Nabeshima et al., 1998;Ding et al., 1993), and this was confirmed by our methods (Fig. 2A). Phase 1 included spindle formation in the nucleus and the equatorial alignment of DNA on a short spindle (∼2.5 μm), indicating the establishment of metaphase (Fig. 2A, 0 minutes). During Phase 2, the spindle elongates very slowly(∼0.1 μm/minute), but towards the end of this phase the chromosomes segregate rapidly (∼1 μm/minute) to the poles, marking the completion of anaphase A on a spindle that is still <3.0 μm. This is apparent as an even division of the chromatin mass (Fig. 2A, 3.1 minutes). Phase 3 is initiated by the abrupt onset of spindle elongation from ∼2.5 μm to 12-15 μm(Fig. 2A, 3.1-9.4 minutes).

Most klpΔ cells showed Phase 1 spindles (<2.5 μm spindle length) with chromosomes abnormally placed at or near one pole(Fig. 2B, 0 minutes)(Table 2, early). These cells progressed to show several deviations from wild-type Phase 2 spindles. First,the DNA sometimes began to separate, but then collapsed back into a single mass before re-initiating separation (Fig. 2B) (20-30% of cells). Second, most Phase 2 spindles with more than one mass of DNA were longer than a wild-type spindle during anaphase A,but anaphase B did not begin until the spindle was unusually long(Table 3). Third, DNA segregation did not usually occur by a single, uniform separation into two equal masses, but rather by a series of unequal divisions(Fig. 1B,C;Fig. 2C;Table 2, mid). Despite this complex pattern, proper separation was usually achieved before mitosis was completed (Fig. 1D;Fig. 2C, 25.5 minutes)(Table 2, late). Phase 3 spindle elongation appeared largely normal in the klpΔ cells,as no significant differences from wildtype were observed in the rate of spindle elongation (Table 3). However, Phase 3 frequently began before the completion of Phase 2, a phenomenon not observed in wild-type cells(Fig. 2C, 12.5-25.5 minutes).

These observations showed that replicated chromosomes move back and forth along the klpΔ spindle, and suggested that sister chromatids would subsequently separate at any position along the spindle. Thus, when anaphase A started, the pairs of sister chromatids could segregate independently of each other to reach their respective poles, at times passing other chromatids that were moving in the opposite direction along the spindle(Fig. 2C).

The abnormal chromosome movements observed in klpΔ cells were surprising, given that these mutants have virtually wild-type viability(West et al., 2001). A series of experiments was undertaken to determine the mechanism by which klpΔ cells with abnormal DNA arrangements achieved successful chromosome segregation. Given that the most common defect in DNA segregation was the appearance of three unequal masses of DNA (Tables2,3), and that fission yeast contains three chromosomes, we inferred that each mass might be an individual chromosome. To examine this possibility, the six kinetochores were followed through mitosis using a tagged kinetochore protein, Mis 12p-GFP(Fig. 3)(Goshima et al., 1999).

Wild-type cells carrying this marker showed kinetochores clustered at the spindle pole bodies (SPBs) early in mitosis, as others have previously described (Funabiki et al.,1993; Goshima et al.,1999). The kinetochores appeared briefly in early mitosis as six dots distributed over 1.5-2 μm across the nucleus(Fig. 3A,B), but then clustered into two groups near the SPBs, at opposite ends of the dividing nucleus(Fig. 3C,D).

By contrast, the kinetochores in klpΔ cells spread out along the spindle as it slowly elongated to ∼6 μm(Fig. 3E). Each DNA mass was associated with two Mis12p-GFP dots that moved together along the spindle,consistent with each being a replicated but unsegregated chromosome, not an individual chromatid. However, the distance between kinetochore pairs was greater than that observed in wildtype (0.5-0.8 μm versus 0.2-0.3 μm),suggesting that the klpΔ sister kinetochores are under tension as a result of forces acting toward the spindle poles (compareFig. 3A,B to 3E). These results demonstrated that each of the three DNA masses scattered along the klpΔ spindles represents an individual chromosome. Therefore,the abnormal chromosome movements in the klpΔ cells precede anaphase A onset.

The observations of individual live klpΔ cells suggested that klpΔ cells are slow to complete mitosis. To test this possibility on populations of cells, we followed synchronous cultures of klpΔ and control strains through mitosis. Cells were synchronized at G₂/M by constructing klpΔ, cdc25-22double mutants and placing these cells at restrictive temperature for this cell cycle regulating phosphatase. After cells had accumulated at the G₂/M boundary, they were released into mitosis by returning the cultures to permissive temperature. Previous work showed no temperature sensitivity in the klpΔ mutants alone(West et al., 2001). Cells with two or more masses of DNA in the absence of a septum were scored as`mitotic', and these were further distinguished as `normal' (two equal masses of DNA) or `abnormal' (two or more unequal DNA masses). The frequency of abnormal DNA distributions in the cdc25-22 cells after block/release was less than 1%, indicating that the cdc25-22 mutation does not, in itself, disorganize chromosome segregation (data not shown). The klpΔ, cdc25-22 mitotic cells showed a high frequency of an abnormal DNA distribution as the culture entered mitosis, but abnormalities decreased rapidly as the culture completed mitosis(Fig. 4A). These data indicate that abnormal chromosome movements occur early in mitosis but are corrected later, a conclusion that is consistent with the observations made on individual cells (Figs 1,3;Table 3).

These experiments also revealed a delay in the progression through mitosis in klpΔ strains. Progress of a synchronized cell culture through M phase is revealed by the fraction of mitotic cells as a function of time after release from cell cycle arest(Fig. 4B). The klpΔ,cdc25-22 mutants showed a significant increase in the total time spent in mitosis, compared with cdc25-22 alone. This was caused by a delay in the completion of mitosis, but not in its initiation, as indicated by the positions of the ascending and descending lines of the two mitotic distributions (Fig. 4B). The klpΔ, cdc25-22 cultures were mitotic 25-50% longer than cdc25-22 alone.

The delay in mitotic progression, together with abnormal spindle elongation prior to anaphase onset, suggested that klpΔ cells might be activating their spindle assembly checkpoint, thus allowing time to rectify their mitotic abnormalities. This checkpoint is a regulatory pathway that monitors the attachment of kinetochores to the spindle and/or the tension on these kinetochores, delaying sister chromatid separation until all the chromosomes are properly attached to the spindle (reviewed inAmon, 1999;Burke, 2000). Disruption of the checkpoint should then lead to a loss of viability in a klpΔbackground. To test this prediction, double and triple mutants of the klpΔ alleles and components of the spindle assembly checkpoint were constructed and assayed for viability. A significant loss of viability was observed in the klpΔ, bub1Δ mutants(Bernard et al., 1998) but not in klpΔ strains with backgrounds in mph1Δ[MPS1 homologue (He et al.,1998)], mad2Δ(He et al., 1997), or cdc16-116^ts [BUB2 homologue(Minet et al., 1979)]. The klpΔ bub1Δ cells produced small colonies at 25°C which then failed to growth when replica plated to 36°C (44 tetrads analyzed) (data not shown).

One of the regulatory targets of the spindle assembly checkpoint is the cyclosome, or anaphase-promoting complex (APC), which catalyzes the loss of sister chromatid cohesion at the onset of anaphase A (reviewed byPage and Hieter, 1999;Nasmyth et al., 2000). Double and triple mutants with the klpΔ strains and mutants of the APC were constructed and assayed for viability. Double and triple mutants of klpΔ and nuc2-663^ts(Hirano et al., 1988;Yamada et al., 1997) were not isolated by tetrad analysis (36 tetrads) or random spores (∼35,000). Double and triple mutants with klpΔ and cut4-533^ts (Yamashita et al., 1996) were not found from either tetrads (12 tetrads) or random spores (5300) indicating that these mutant combinations were synthetically lethal. Mutants with klpΔ and cut9-665^ts (Samejima et al., 1993) produced small colonies, but failed to grow upon re-streaking to new plates (6 tetrads/3800 random spores) data not shown).

The aberrant chromosome behavior described here in the klpΔmutants, together with the effects of these mutants on cytoplasmic microtubules and cell morphology described previously(West et al., 2001), suggested that klp5⁺ and klp6⁺ affect both microtubule and kinetochore functions. Similar functions have been described for Dis1p, a member of the TOG1/XMAP215 family of microtubule-associated proteins; Dis1p binds to both microtubules and centromeric DNA, and mutants display abnormalities in both microtubule organization and chromosome segregation (Nabeshima et al.,1995; Nakaseko et al.,1996; Nakaseko et al.,2001). We therefore looked for genetic interactions between the klpΔ mutants and the cold-sensitive mutant dis1-1^cs (Ohkura et al., 1988), by constructing double and triple mutants using tetrad analysis. The dis1-1^cs strain failed to grow at 20°C,as previously reported (Ohkura et al.,1988), whereas both klpΔ and dis1-1^cs parental strains grew virtually at wild-type levels at 32°C. By contrast, the klpΔ, dis1-1^cs double and triple mutants failed to grow at 32°C, although the viability of the sibling spores was normal in these crosses (20 recombinant tetrads examined for each klpΔ dis1-1^cs cross). Spore formation in these crosses resembled wildtype, suggesting normal karyogamy and meiosis, but the klpΔ dis1-1^cs spores often failed to germinate. When colonies formed, they never progressed past two or three divisions. These results indicate a strong interaction between klp5⁺ and klp6⁺ and the dis1⁺ gene.

Previous work showed that Klp5p and Klp6p localize to microtubules in both interphase and mitotic cells (West et al.,2001). Here, we examined the mitotic localization with finer temporal resolution as cells progressed through mitosis. As the cells entered mitosis, each Klp-GFP localized as two or three spots within the nucleus(Fig. 5A, t=0 minutes). These spots briefly separated into 4-6 resolvable spots(mean=4.8±0.5, n=18) lying along a ∼2 μm line(mean=1.8±0.4 μm, n=18)(Fig. 5B, and t=3.7-13.7 minutes). This pattern resembles that of wild-type kinetochores (Fig. 3B)(Saitoh et al., 1997;Goshima et al., 1999;Wigge and Kilmartin, 2001). As the DNA mass became oblong, indicating the completion of anaphase A(Fig. 3C), the Klp-GFP signal re-localized to the entire spindle (mean length of spindle staining=3.1±0.4 μm, n=10), rather than to the pole-associated dots characteristic of a kinetochore marker(Fig. 5D; 16 minutes). The Klp-GFP remained in this arrangement as the spindle elongated, thus becoming restricted to a decreasing fraction of the spindle midzone(Fig. 5D; 27.4 minutes). The Klp-GFP re-appears on interphase cytoplasmic microtubules coincident with the formation of the post-anaphase array, which arises from the point of septation as the cells enter G₁ (West et al., 2001).

To better discern the localization of Klp-GFP proteins during pre-anaphase,we attempted to take advantage of the clarity of kinetochore positions seen in klpΔ cells, as revealed by expressing Mis12-GFP(Fig. 3). Double mutant strains were constructed in which Klp5p-GFP was expressed in the klp6Δbackground, and vice versa. In these cells, there was no concentration of GFP signal on kinetochores when the chromosomes were distributed across a long spindle (4-6 μm); these cells usually showed spindle staining (data not shown). These results imply a co-dependence for kinetochore localization between Klp5p and Klp6p at this stage, or that each Klp requires the other to be restricted to the kinetochores before anaphase A. In any case, these results suggest that proper localization of both Klp5p and Klp6p in pre-anaphase A cells requires the expression of both of these kinesins.

Our analyses of klp5⁺ and klp6⁺suggest that these genes encode kinetochore motors important for normal chromosome movement prior to sister chromatid separation. In wild-type cells,the kinetochores remain clustered near the SPBs through most of the cell cycle. They separate from the SPBs only briefly when the spindle is ∼2μm long, but they quickly re-associate as two pools of dots on opposite sides of the dividing nucleus. By contrast, the kinetochores in klpΔ cells spread out along a 4-6 μm spindle. They remain associated as three pairs of dots, and each pair moves independently along the spindle for several minutes before separating. As these movements precede sister chromatid separation, we interpret this stage in klpΔcells as prometaphase (i.e. after chromosomes have attached to the spindle,but before anaphase A). It remains uncertain whether these cells fail to establish metaphase in the first place, or subsequently fail to maintain it until chromatid separation begins. However, the high frequency of asymmetrically placed DNA masses on short spindles suggests a lack of proper chromosome congression to the metaphase plate. The absence of a proper metaphase in klpΔ cells may be due to changes in the dynamics of kinetochore microtubule interactions, as both Klp5p and Klp6p have been implicated in microtubule disassembly(West et al., 2001). It is also a formal possibility that anaphase A has been activated normally in these cells, but that the loss of either klp5⁺ or klp6⁺ changes kinetochore microtubule dynamics in a way that prevents proper sister chromatid migration to the poles. However, the genetic interactions between klpΔ strains and APC mutants,together with the delay in anaphase B onset until after kinetochore segregation is initiated, suggest that the activities of regulatory proteins are also at play in the klpΔ phenotype. Furthermore, the rate of anaphase B, once started, is normal in klpΔ cells, whereas anaphase B is retarded in separin and APC mutants, which fail to separate sister kinetochores (Samejima et al.,1993; Nabeshima et al.,1998; Hirano et al.,1988).

The localization of Klp5p and Klp6p to the kinetochores from prophase to the onset of anaphase A suggests that these motors play a role in chromosome alignment prior to chromatid separation. The strong genetic interactions between klpΔ and one component of the spindle assembly checkpoint, bub1⁺, as well as three components of the APC,are consistent with a function for these kinesins before or at anaphase A onset. However, once chromosome separation has occurred in klpΔcells, the chromatids migrate to the poles normally (i.e. there is no apparent`lagging chromosome' phenotype, as has been described for other mitotic mutants) (Pidoux et al.,2000).

Our results suggest that anaphase A can be delayed when proper metaphase chromatid alignment is not achieved, but that chromatid separation can eventually be initiated from any point on the spindle. A strikingly similar phenotype has been observed in other S. pombe mutants, including a dominant allele of the cdc2⁺ kinase(Labib et al., 1995) and a disruption of dis1⁺ or its related protein mtc1⁺/alp14⁺, proteins that bind to both microtubules and kinetochores(Nabeshima et al., 1998;Garcia et al., 2001;Nakaseko et al., 2001). In each of these mutants, three masses of DNA are observed on an unusually long spindle prior to sister chromatid separation, indicating a block to sister chromatid separation without inhibition of spindle elongation. The dis1⁺ disruption mutants are, however, distinct from the klpΔ cells with respect to spindle length in that they skip mitotic Phase 2 (metaphase/anaphase A) and appear to proceed directly into Phase 3 (anaphase B); klpΔ cells have a longer Phase 2 with a distinct transition to anaphase B (Table 3). In this way, the klpΔ phenotype more closely resembles mutants of two kinetochore components, mis6⁺ and mis12⁺ (Goshima et al., 1999).

The extensive, albeit unusual, chromosome movements observed in klpΔ mutants demonstrate that the spindle in S. pombecontains mechanisms other than Klp5/6p that are capable of moving chromosomes. Several candidate motors have been identified, including two KAR3homologues (Pidoux et al.,1996; Troxell et al.,2001), dynein (Yamamoto et al., 1999), and a potential chromokinesin (pombe genome project). A role in chromosome segregation for one of the KAR3 homologues, pkl1⁺, is implied from the chromosome missegregation phenotype of a cold-sensitive allele of γ-tubulin, which is synthetically lethal with a pkl1Δ mutation(Paluh et al., 2000). However,the quadruple pkl1Δ, klp2Δ, klp5Δ, klp6Δ mutant is viable, indicating that fission yeast possesses some other mechanism to segregate its chromosomes(West et al., 2001). Alternatively, microtubule dynamics, in the absence of motor activity per se,have been shown to move chromosomes in vitro(Lombillo et al., 1995). Sorting out the roles of each of these different sources of motive force will required a systematic analysis of all the genes involved.

Our observations of both fixed and live cells may be summarized by the statement that the deletion of either klp5⁺ or klp6⁺ disrupts the balance of forces exerted on the kinetochores, which in turn leads to the misalignment of chromosomes at metaphase and a delay in the onset of anaphase. This interpretation motivated our search for interactions between the klpΔ mutations and spindle assembly checkpoint genes, but only bub1Δ showed an effect. This implies that either bub1⁺ functions in a spindle checkpoint independently of the other checkpoint genes, or that bub1⁺ has functions outside the checkpoint. At present, we cannot distinguish between these possibilities and, indeed, they are not mutually exclusive. Recent work in HeLa cells supports a model in which a BUB1-dependent checkpoint detects tension on the kinetochores, while a MAD2-dependent checkpoint detects kinetochore attachment to the spindle(Skoufias, et al., 2001). In vitro assays have also demonstrated that BubR1 can interact with Cdc20 independently of Mad2 to inhibit the APC complex(Tang et al., 2001). By analogy, the synthetic effects of klpΔ with bub1Δ, but not mad2Δ or mph1Δ,can be interpreted as a requirement only for Bub1p to keep klpΔcells from a self destructive anaphase onset. Since the kinetochores in klpΔ cells appear to be under greater tension than in wild-type cells, this interpretation leads to a disagreement with Skoufias et al.(Skoufias et al., 2001),concerning the tension versus microtubule attachment aspects of the spindle assembly checkpoint. Alternatively, it may be that the delay in anaphase onset is not required to maintain viability in klpΔ cells, but that the synthetic lethality of klpΔ with bub1Δarises from a bub1⁺ function distinct from checkpoints,such as a role in the integrity of the kinetochore.

The specificity for bub1⁺ among the checkpoint components for the viability of klpΔ strains was surprising and suggests that klp5⁺ and klp6⁺ interact more directly with bub1⁺ than with other checkpoint genes. Previous work indicated that BUB1 and MPS1 genes function at the top of a regulatory hierarchy, with MAD2 downstream of both of these genes (reviewed by Amon,1999; Abrieu et al.,2001). However, in fission yeast the interactions among the klpΔ and checkpoint mutants indicate that bub1⁺ has functions independent of mad2⁺. The fission yeast mad2⁺ and mph1⁺ genes do share some features of checkpoint regulation with the budding yeast homologues, including a co-dependence for metaphase arrest (He et al.,1997; He et al.,1998). However, differences between these yeast genes in the regulation of SPB duplication have been previously described(He et al., 1998).

A distinction between BUB1/BUBR1 and MAD2 has also been described recently(Skoufias et al., 2001;Tang et al., 2001). In mammals, MAD2 is thought to respond to the attachment of kinetochores to the spindle microtubules, while BUB1/BUBR1 respond to tension on kinetochores(Skoufias et al., 2001;Li and Nicklas, 1995;Waters, et al., 1998). If an analogous system functions in S. pombe, the interaction between the klpΔ mutants and bub1Δ would imply that the chromosomes are attached but not under tension in klpΔ cells. The lack of obviously unattached chromosomes, together with active movement of the chromosomes in klpΔ cells are consistent with the chromosomes remaining attached to the spindle. However, increased distance between sister kinetochores implies that the chromosomes are under increased tension (Fig. 3).

The synthetic interactions between the klpΔ mutants and the APC components are also consistent with a kinetochore function for these kinesins, but the relation between these results and the checkpoint genes remains to be determined. Mad2p is thought to be the principle regulator of the APC through its interactions with CDC20 (reviewed byAmon, 1999), but BubR1 also interacts with CDC20 independently of Mad2p(Tang et al., 2001). Although our results show no interactions between the klpΔ and mad2Δ mutants, both of these mutants show synthetic lethal interactions with the same APC components (cut4⁺, cut9⁺ and nuc2⁺)(He et al., 1997). By contrast, in mammalian cells, BUBR1 co-immunoprecipitates with Cdc16p(cut9⁺ homologue) and Cdc27p (nuc2⁺homologue) (Chan et al., 1999). The fission yeast bub1⁺ gene has checkpoint functions(Bernard et al., 1998) and synthetic interactions with the klpΔ mutants, but it remains to be determined whether it also interacts with APC components. Clearly,additional work will be required to sort out the complex roles of both motor proteins and checkpoint regulators at the kinetochore.

Prometaphase congression in metazoans may be dependent on the kinesin-like protein CENP-E (Wood et al.,1997; Schaar et al.,1997; Yucel et al.,2000). Although the sequences of Klp5p and Klp6p are not closely related to those of known CENP-E proteins, there are striking similarities in mutant phenotypes. Disruption of CENP-E function, either by depletion with antibodies (Yen et al., 1991;Wood et al., 1997;Schaar et al., 1997) or mutation (Yucel et al., 2000)causes a failure in metaphase chromosome alignment and a delay in the onset of anaphase. Microinjection of CENP-E antibodies into CF-PAC cells did not show the same degree of chromosome misalignment, but a decrease in the number of kinetochore microtubules and a loss of tension at the kinetochores(McEwen et al., 2001).

Further, the genetic interactions between klp5Δ, klp6Δ and bub1Δ in fission yeast are reminiscent of the physical association reported between CENP-E and the kinetochore-associated spindle assembly checkpoint kinases BUB1 and BUBR1 in metazoans (Taylor and McKeon,1997; Chan et al.,1998; Yao et al.,2000). A functional role for this interaction is indicated by the BUBR1-dependent mitotic arrest in systems deficient in CENP-E(Chan et al., 1999;Yao et al., 2000), and the requirement for CENP-E in the activation of the checkpoint upon spindle damage(Abrieu et al., 2000). The synthetic lethal interactions reported here between klpΔ and bub1Δ strains indicate a similar functional relation in fission yeast.

In vertebrates, BUBR1 is required for the localization of CENP-E to the kinetochores (Chan et al.,1999; Sharp-Baker and Chen,2001). Bub1p in fission yeast also localizes to kinetochores, but it is not known whether this is dependent on other proteins(Bernard et al., 1998).

CENP-E in vertebrates localizes to kinetochores from prophase to the end of anaphase A, when it re-localizes to the spindle midzone(Brown et al., 1996). Klp5p and Klp6p also localize to kinetochores prior to anaphase A, and then they relocalize to the spindle midzone during anaphase B.

Previous work has shown that Klp5p and Klp6p may foster microtubule disassembly (West et al.,2001), reminiscent of the KinI (MCAK, XKCM1) subfamily of kinesins, which also localize to kinetochores(Walczak et al., 1996;Maney et al., 1998). Furthermore, sequence similarity has been described among the KIP3 subfamily members and the metazoan KinI subfamily(Severin et al., 2001). The activity of the KinI motors may be balanced by the activity of microtubule-associated proteins of the TOG/XMAP215 family (reviewed byAndersen, 2000). For example,in budding yeast, deletion of the KIP3 gene partially rescues anaphase B defects in a mutant of the TOG/XMAP215 homologue, stu2-10^ts (Severin et al., 2001). Here, we show that the klpΔ mutants are synthetically lethal with a cold sensitive allele of the fission yeast TOG/XMAP215 homologue, dis1⁺. Given that the klpΔ, dis1-1^cs mutants failed to grow beyond one or two divisions, we cannot assess the nature of the defects in these mutants. However, we have not detected any abnormalities in anaphase B in our klpΔ strains (Table 3). It remains to be determined whether klp5⁺and klp6⁺ also interact with the second TOG/XMAP215 homologue, mtc1⁺/alp14⁺(Garcia et al., 2001;Nakaseko et al., 2001).

With the completion of the genome sequencing projects for both yeasts, it is evident that they both lack obvious sequence homologues to CENP-E, and the kinesins they do express have only limited similarities to members of the KinI family (Severi, et al., 2001). We suggest that the KIP3 subfamily, represented by klp5⁺ and klp6⁺ in S. pombe and KIP3 in S. cerevisiae(West et al., 2001), is the functional analogue of one or both of these metazoan kinesin subfamilies. Severin et al. have previously suggested that Kip3p is the orthologue of the KinI subfamily (Severin et al.,2001). We emphasize functional similarities among klp5⁺, klp6⁺ and both CENP-E and KinI based on similarities in loss-of-function phenotypes. We conclude that mitotic functions can be subsumed by a variety of molecular entities and that a broad view of mitosis will require an emphasis upon mitotic functions, rather than on the identity of specific molecular players.

We thank Heidi Browning, Katya Grishchuk, Thomas Fiedler and Michele Jones for helpful suggestions and critical reading of the manuscript. We thank Lei Ding for help with the klpΔ, dis1-1^cscrosses. We thank Da-Qiao Ding for the pDQ105 (GFP-α-tubulin) plasmid,Shelley Sazer for the mad2Δ and mph1Δ strains,J.-P. Javerzat for the bub1Δ strain, and M. Yanagida for the Mis12p-GFP and Mis12-HA strains. This work was supported by National Institutes of Health (NIH) grant GM-36663 to J.R.M., who is a Research Professor of the American Cancer Society.