Motile Properties of the Kinesin-related Cin8p Spindle Motor Extracted from Saccharomyces cerevisiae Cells*

We have developed microtubule binding and motility assays for Cin8p, a kinesin-related mitotic spindle motor protein from Saccharomyces cerevisiae. The methods examine Cin8p rapidly purified from crude yeast cell extracts. We created a recombinant form of CIN8 that fused the biotin carrying polypeptide from yeast pyruvate carboxylase to the carboxyl terminus of Cin8p. This form was biotinated in yeast cells and provided Cin8p activity in vivo. Avidin-coated glass surfaces were used to specifically bind biotinated Cin8p from crude extracts. Microtubules bound to the Cin8p-coated surfaces and moved at 3.4 ± 0.5 μm/min in the presence of ATP. Force production by Cin8p was directed toward the plus ends of microtubules. A mutation affecting the microtubule-binding site within the motor domain (cin8-F467A) decreased Cin8p’s ability to bind microtubules to the glass surface by >10-fold, but reduced gliding velocity by only 35%. The cin8–3 mutant form, affecting the α2 helix of the motor domain, caused a moderate defect in microtubule binding, but motility was severely affected.cin8-F467A cells, but not cin8–3 cells, were greatly impaired in bipolar spindle forming ability. We conclude that microtubule binding by Cin8p is more important than motility for proper spindle formation.

We have developed microtubule binding and motility assays for Cin8p, a kinesin-related mitotic spindle motor protein from Saccharomyces cerevisiae. The methods examine Cin8p rapidly purified from crude yeast cell extracts. We created a recombinant form of CIN8 that fused the biotin carrying polypeptide from yeast pyruvate carboxylase to the carboxyl terminus of Cin8p. This form was biotinated in yeast cells and provided Cin8p activity in vivo. Avidin-coated glass surfaces were used to specifically bind biotinated Cin8p from crude extracts. Microtubules bound to the Cin8p-coated surfaces and moved at 3.4 ؎ 0.5 m/min in the presence of ATP. Force production by Cin8p was directed toward the plus ends of microtubules. A mutation affecting the microtubule-binding site within the motor domain (cin8-F467A) decreased Cin8p's ability to bind microtubules to the glass surface by >10-fold, but reduced gliding velocity by only 35%. The cin8 -3 mutant form, affecting the ␣2 helix of the motor domain, caused a moderate defect in microtubule binding, but motility was severely affected. cin8-F467A cells, but not cin8 -3 cells, were greatly impaired in bipolar spindle forming ability. We conclude that microtubule binding by Cin8p is more important than motility for proper spindle formation.
Eukaryotic chromosome segregation is mediated by the mitotic spindle, a microtubule-based motile structure that undergoes a distinct program of morphological changes. It is now clear that many spindle movements are accomplished by microtubule-based motor proteins. Perhaps the best characterized type of spindle motor is that of the BimC subfamily of kinesinrelated proteins. Members of the BimC family, first discovered in Aspergillus nidulans (1), have been found in numerous eukaryotic species (2)(3)(4)(5)(6)(7). These proteins are conserved in amino acid sequence of the motor (force-producing) domain and apparently perform similar roles in many different cell types (8 -10). BimC motors are required for bipolar spindle assembly; elimination of their function blocked this essential early mitotic step in fungal, insect, and mammalian cells. The yeast Saccharomyces cerevisiae expresses two BimC-related motors that overlap in function, Cin8p and Kip1p. Although neither is individually essential, one of the pair is required for viability (3,4,11). Loss of KIP1 causes less severe phenotypes than loss of CIN8, suggesting that Cin8p is more important for successful yeast spindle function (3). Besides essential roles in spindle assembly, Cin8p and Kip1p are also required for the maintenance of spindle bipolarity following assembly and are responsible for producing most of the spindle-elongating force during anaphase (11,12).
In addition to genetic experiments, several lines of evidence suggest that BimC motors act by cross-linking and sliding antiparallel microtubules found in the spindle midzone. BimC motors have been localized exclusively to the microtubules that lie between the spindle poles (3,4,(13)(14)(15). In vitro, BimC motors from Drosophila melanogaster and Xenopus laevis have been found to move exclusively toward microtubule plus ends, albeit at a rate much slower than that achieved by kinesin (ϳ2 m/min versus 20 -40 m/min for kinesin) (16). In particular, the "bipolar" molecular structure determined for the Drosophila BimC motor Klp61F suggested a mechanism by which these motors may contribute to spindle structure and elongation. Klp61F is an elongated bipolar homotetramer, with two motor domains positioned at each end of a rodlike structure (17). Such a bipolar molecule would have the ability to cross-link and slide antiparallel microtubules (10). Despite these suggestive observations, the actual molecular role of BimC motors during spindle dynamics has yet to be established.
In this study, we developed in vitro assays for the analysis of the S. cerevisiae Cin8p motor. Our purpose was 2-fold. First, the in vitro properties of Cin8p have not been described. Mere analogies to other BimC members are clearly indirect and unsatisfactory. Second, the ease of genetic and cell cycle manipulations in yeast demands a similarly direct method to examine and dissect Cin8p functions in vitro. Although exogenous expression of motors allows initial reconstituted studies, only endogenous expression in yeast allows direct study of Cin8p cell cycle regulation and different mitotic roles. With these goals in mind, we developed a rapid assay for the activity of Cin8p expressed in yeast cells. For a one-step purification and adsorption to avidin-coated surface, the Cin8 protein was fused to a peptide (biotin carrier peptide or BCP) that is endogenously biotinated in yeast. Subsequent analysis could distinguish microtubule binding from translocation activities, and photoactivation of caged-ATP further increased the sensitivity of our assays. Using these assays, the study of two Cin8p mutants revealed the relative importance of binding and motility for Cin8p in vivo functions.

MATERIALS AND METHODS
Yeast Strains and Media-The S. cerevisiae strains used in these experiments are derivatives of S288C and are listed in Table I. The cin8::HIS3 and kip1::HIS3 alleles were described previously (12). Rich (YPD) and minimal (SD) media were as described by Sherman et al. (18). To derepress galactose-inducible genes, cells were grown in 2% raffinose minimal media at 26°C for 24 h prior to induction by the addition of galactose to 2%. Cycloheximide (Sigma) was added to a final concentration of 5 g/ml. All media for expression of biotinated Cin8p were supplemented with 24 mg/liter biotin (Sigma) (19). Cells were synchronized in G 1 phase by the addition of 4 -6 g/ml of ␣-factor (Bachem) to liquid media (pH 4), for about 4 h until Ͼ85% were unbudded.
DNA Manipulations-To create a biotinated version of Cin8p, we used the polymerase chain reaction to amplify the sequence of yeast PYC1, encoding the last 90 amino acids of pyruvate carboxylase (20,21), and add NotI sites at each end. The 5Ј side primer used was AATAA-AAGCGGCCGCACTGTTACTAAATCCAAAGCA, and the 3Ј primer was AATAAATGCGGCCGCTGCCTTAGTTTCAACAGG. The BCP fragment was inserted into a NotI site that had previously been created at the 3Ј-end of CIN8 1 in the CEN vector pRS317, (22), creating pLG3. Standard DNA manipulations were used to place CIN8-BCP under the control of the galactose-inducible promoter on a CEN vector (pLG8). The cin8 -3 (cin8-R196K) mutation was introduced into CIN8-BCP constructs using an MscI to PstI fragment containing the mutant allele. The presence of cin8 -3 was confirmed by sequencing. The cin8-F467A mutation was created by the unique site elimination method (23). The primer for mutagenesis was GTCAATTTCGATTCACGGGCAGGTAT-ATGGCCGCTTTTATC, which also introduced a BspMI site with the amino acid change. The presence of the cin8-F467A mutation was confirmed by digestion with BspMI.
Yeast Extracts and Protein Analysis-Yeast protein extracts for SDS gel electrophoresis were prepared by vortexing cells with glass beads (diameter of 425-600 m) in 50 mM Tris-HCl (pH 7.4), 250 mM NaCl, 50 mM NaF, 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride. Cells were vortexed eight times at the maximum speed for 15 s, with 15 s on ice in between. Lysates was separated from cell debris and glass beads via centrifugation for 5 min in the cold. An aliquot was reserved for protein concentration determination using the BCA assay (Pierce) to ensure equal protein loading in gel lanes. Samples were run on 7% acrylamide gels, and the proteins were transferred to polyvinylidene difluoride Immobilon-P (Millipore Corp.) membranes using standard techniques (24). Reagents for immunoblot analysis, including streptavidin-conjugated alkaline phosphatase, were obtained from Tropix. Band intensities were quantified by scanning film images of blots and analyzing with commercial software, IDL (Research Systems).
For the in vitro assays, 20 -50-ml cultures were grown to midlog phase. Cells were pelleted, washed, and ground with a mortar and pestle under liquid nitrogen in 1 ml of motility buffer 1 (MB1): 30 mM Tris, 35 mM PIPES, final pH 7.2, 175 mM NaCl, 2 mM EDTA, 1 mM EGTA, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol. Extracts were clarified by centrifugation at 13,000 rpm in a microcentrifuge at 4°C. Protein concentration in these extracts was typically 1-3 mg/ml.
In Vitro Microtubule Binding and Motility Assays-To assay Cin8p-BCP-induced microtubule gliding and binding on avidin-or streptavidin-coated surfaces, we modified the protocol of Berliner et al. (25). Acid-treated number 1 borosilicate glass coverslips were prepared by overnight incubation in Chromerge (Thomas Scientific), followed by washing under running filtered deionized water for 20 -30 min and spin-dried on a custom coverslip spinner. Flow chambers (ϳ10 l) were formed between microscope slide and the coverslip mounted with double stick tape. For side-by-side comparison of different conditions, up to four chambers separated by silicone grease lines were prepared on a single slide and coverslip. All incubations below were for 10 min at room temperature. Biotinamidocaproyl bovine albumin (Sigma), 10 l of a 1.7 mg/ml solution in H 2 O, was introduced to the chambers, incubated, and washed twice with H 2 O. The chambers were treated with 10 l of 5 mg/ml UltraAvidin TM (deglycosylated and charge-neutralized avidin; from Leinco Technologies) in 10 mM Tris, pH 8, 1 mM EDTA, incubated, and rinsed twice with H 2 O, incubated with casein (up to 5 mg/ml) in H 2 O, rinsed twice with H 2 O, and then rinsed twice with MB1. Crude yeast extracts, 10 l of 0.2-1 mg/ml total protein, were introduced, incubated for 4 -6 min, and washed twice with MB1. In experiments in which the ability of biotin to block binding was assessed, the H 2 O rinse after UltraAvidin incubation was replaced with biotin-saturated H 2 O, and the consecutive rinses and incubations contained a 1% dilution of biotin-saturated H 2 O. The final rinse in MB1 and motility and binding assays were performed without the addition of biotin. For motility, taxol (40 M)-stabilized microtubules (0.05 mg/ml) in MB1 were added to the flow chamber along with 1-5 mM ATP. Average microtubule length was 5 m. When tested, other nucleotides were added to a final concentration of 1 mM. For motility assays, MB1 was supplemented with 5 mM MgSO 4 . To visualize microtubules, video-enhanced DIC microscopy was used as described previously (26). Briefly, an inverted Zeiss microscope (Axiovert 100TV, 100ϫ Plan Neofluar 1.3 NA objective) was modified by installing high transmission polarizers to project images onto a Newvicon camera (Dage VE1000), and images were enhanced using an image processor (Hamamatsu Argus 100). Video sequences were recorded on sVHS tape (Panasonic AG7650 recorder), and microtubule gliding velocities were measured using custom software (Imaging Technologies AFG image processor) from these videos.
ATP uncaging experiments were done in the presence of 1 mM of 1 T. Kingsbury and M. A. Hoyt, unpublished results. MATa ade2 lys2 his3 leu2 ura3 cin8::HIS3 kip1:: NPE-ATP 2 (Molecular Probes, Inc., Eugene, OR), a "caged" nucleotide from which ATP is liberated by exposure to UV irradiation. The assay was performed similarly to experiments with myosin (27,28). Photoactivation of caged ATP required a custom dichroic (400DCLPL from Chroma Technology Corp.) to reflect light from a 75-watt xenon arc lamp. To prevent uncaging during video-enhanced differential interference contrast microscopy, the mercury arc illumination passed through an OG515 Schott Glass optical filter as well as through a 546-nm interference filter. UV illumination of the sample was between 1 and 2 min, followed by observation of the sample for an additional minute. Under these conditions, no effect on the length of microtubules was observed.
Polarity-marked microtubules were prepared according to Refs. 29 and 30 by using a 1:1 ratio of NEM-treated to NEM-untreated tubulin with a total tubulin concentration of 2 g/ml. Sea urchin (Strongylocentrotus purpuratus) axonemes (a gift from Herb Miller and Jennifer Gregory) were used as seeds and were prepared by the method of Bell et al. (31) as modified by Walker et al. (32). The correct orientation of microtubules labeling was confirmed by a motility assay using bovine brain kinesin (Cytoskeleton).
The microtubule bundling assay was performed using flow chambers constructed as described above. Yeast cell extracts (4 l of 2 mg/ml) were mixed with taxol-stabilized microtubules in MB1 (16 l of 3.3 g/ml). Bundling was observed in solution and on the glass surface by video-enhanced differential interference contrast optics.

A Biotinated Form of Cin8p Provides Activity in Vivo-For
this study, we developed an in vitro assay to examine the microtubule binding and motile properties of the Cin8p kinesin-related motor from S. cerevisiae. We constructed a version of CIN8 that encodes a fusion between Cin8p and a peptide from S. cerevisiae pyruvate carboxylase (product of PYC1; Refs. 20 and 21) attached to the Cin8p carboxyl terminus. This region of pyruvate carboxylase, derived from its carboxyl terminus, acts as a covalent acceptor of biotin from a specific ligase (20,25). We refer to this region of pyruvate carboxylase as BCP for biotin-carrying polypeptide. Fusion proteins that contained the homologous region of E. coli pyruvate carboxylase become endogenously biotinated both in E. coli and S. cerevisiae (19). The CIN8-BCP fusion gene was cloned into low copy centromere-containing (CEN) and high copy (2-m origin) vectors and was expressed from either the endogenous CIN8 promoter or a high level galactose-inducible promoter. Relative to the level of expression provided by the endogenous promoter/ CEN plasmid construct, the endogenous promoter/2-m version and the galactose promoter/CEN version (induced for 4 h) produced approximately 35-and 500-fold more Cin8p-BCP, respectively.
On account of overlapping functions (3,4), the ability of CIN8-BCP to complement a CIN8 deletion allele was tested in a strain that was deleted for both CIN8 and KIP1. Since cin8⌬ kip1⌬ double mutants are not viable, the viability of the tester strain was maintained by a CIN8 plasmid that also carried CYH2. Despite the presence of a recessive cyh2 cycloheximide resistance allele in the genome, this strain is sensitive to cycloheximide due to the dominant CYH2 allele carried on the plasmid (Fig. 1A). Transformation with a second plasmid carrying CIN8-BCP, however, allowed cells to live without the CIN8-CYH2 plasmid and therefore to grow on cycloheximidecontaining media. This indicated that CIN8-BCP provides CIN8 activity. The cin8⌬ kip1⌬ (CIN8-BCP plasmid) cells were indistinguishable from cin8⌬ kip1⌬ (CIN8 plasmid) cells for growth rate and temperature resistance. In addition, we found that Cin8p-BCP localized to spindle microtubules in a manner indistinguishable from Cin8p (data not shown). We conclude that Cin8p-BCP acts similarly to Cin8p in vivo.
The Cin8p-BCP fusion protein was endogenously biotinated in S. cerevisiae (Fig. 1B). Biotination was detected by gel electrophoresis of total cell protein extract followed by blotting and probing with streptavidin-conjugated alkaline phosphatase. In cells expressing CIN8-BCP from the galactose-inducible promoter (on a CEN plasmid), the addition of galactose resulted in the appearance of an ϳ128-kDa protein, consistent with the predicted molecular weight of the fusion protein (Fig. 1B). This protein band increased in intensity with the time of galactose induction and was absent from cells that are deleted for CIN8 (Fig. 1B). In cells expressing CIN8-BCP from its native promoter, either on low copy (CEN) or high copy (2-m) plasmids, the amount of Cin8p-BCP produced was correspondingly lower (Fig. 1B). The 2-m expression level was intermediate between that of CEN and P GAL expression levels (Fig. 1B). The noninducible biotinated protein band of ϳ120 kDa is probably S. cerevisiae pyruvate carboxylase, one of the five endogenously biotinated proteins in this organism (33).
Bundling of Microtubules in Extracts Containing High Levels of Cin8p-We observed that extracts from yeast cells overproducing Cin8p exhibited a microtubule bundling activity ( Fig. 2 and Table II). This activity may be related to the ability of BimC motors to cross-link spindle microtubules (10). When extracts from cells overproducing Cin8p or Cin8p-BCP under the control of the galactose promoter were mixed with taxolstabilized microtubule, bundles were observed. In the presence of 1 mM AMP-PNP, a nonhydrolyzable ATP analogue, 90% of the observable microtubule structures were bundles. Bundles were found in 70 -80% of scored microscope fields (n ϭ 100 fields of 400 m (2). Without the addition of AMP-PNP, the degree of bundling was slightly reduced. At lower overexpression levels (Cin8p-BCP from its normal promoter but on a high copy 2-m plasmid) bundling was observed, but at a reduced level (Table II). No bundling was observed using extracts from yeast producing wild-type levels of Cin8p or deleted for CIN8 (n Ͼ 100 fields for each). We found that the addition of avidin, which interacts with the biotin moiety of BCP, did not alter the extent of Cin8p-BCP-specific bundling. This indicates that the ability of Cin8p-BCP to interact with microtubules is independent of its interaction with avidin. When 1 mM ATP was added to bundles formed in the absence of added nucleotides, dissolution of bundles occurred within 5-10 min. In some cases, we were able to observe the sliding apart of bundles following the addition of ATP. The relative polarity of the sliding microtubules (parallel versus antiparallel) was not determined, however.
The bundling in crude cell extracts was specific for overproduced Cin8p. Under the conditions described above, extracts from cells that overproduced either a Cin8p motor domain mutant (Cin8p-R394A, H396A) or kinesin-related Kip2p or Kip3p never bundled microtubules. Interestingly, Cin8p-871, a tail deletion mutant that forms a dimer instead of the wild-type tetramer, 3 was unable to form bundles. This suggests that bundling reflects the ability of BimC motors to cross-link microtubules via their bipolar arrangement of motor domains.
Cin8p-BCP Specifically Binds Microtubules to Avidin-coated Surfaces-Avidin-coated glass coverslips were used to specifically bind Cin8p-BCP from crude protein extracts. Two criteria were used to establish that any observed microtubule binding to these surfaces was specific for Cin8p: 1) extracts that lack biotinated Cin8p should exhibit no microtubule binding, and 2) added biotin should block microtubule binding. Fig. 3 summarizes six experiments in which microtubule binding was assessed in the absence of added nucleotide. The microtubule binding activity was high from extracts in which Cin8p-BCP was overproduced but was greatly reduced by the addition of biotin or from extracts of cin8⌬ cells or cells overexpressing Cin8p (without the BCP tag). The main factors that influenced the specificity of this assay were ionic strength and the concentration of casein, which was used to block nonspecific interactions. We found that different extracts from the same yeast strain sometimes required different casein concentrations to block nonspecific microtubule binding. Therefore, for every extract preparation we determined the casein concentration that, in the absence of added nucleotide, would allow high microtubule binding in the absence of a biotin block and very low binding in the presence of a biotin block. Reducing NaCl concentrations below 150 mM also increased the nonspecific binding of microtubules. However, NaCl concentration above 200 mM interfered with binding of microtubules to Cin8p. The optimal conditions were achieved with 175 mM NaCl in the buffers for both extract preparation and assays. The number of specifically bound microtubules was higher on modified avidin (UltraAvidin TM ; see "Materials and Methods") surfaces than on other avidin-and streptavidin-coated surfaces.
In Vitro Motile Activity of Cin8p-The spindle pole-separating activity of Cin8p is most easily explained by a plus enddirected motile activity (3,12,34). To observe Cin8p motility in vitro, 1 mM ATP was added to microtubules preadsorbed to Cin8p-BCP on avidin-coated surfaces. The Cin8p-BCP source for these initial experiments was an extract from cells in which it had been overproduced from the galactose promoter. Under these conditions, efficient microtubule gliding was observed (Fig. 4A). Of the 3-10 microtubules visible per 400-m (2) field, 70 -80% moved in the presence of ATP. The nonmotile microtubules divided approximately equally between two categories: immobile microtubules and those that detached. For about 10 -20% of the moving microtubules, pivoting around a single point on the surface was observed during movement. Such movement probably reflects the activity of a single surface motor (35). For about 10% of the moving microtubules, the movement was episodic, where changes in velocity or stops and starts were observed. Buckling of microtubules, which would indicate the presence of binding but nonmotile motors on the surface (36), was not observed.
To determine the directionality of Cin8p-induced motility, we used minus end-marked microtubules nucleated from sea urchin sperm axonemes (see "Materials and Methods"). The direction of all gliding microtubules was minus end leading (n ϭ 16), indicating that Cin8p exhibits plus end-directed activity (Fig. 4B).
The velocity distribution of Cin8p-induced microtubule gliding is shown in AMP-PNP ϳ0 a The indicated protein was expressed from its endogenous promoter on a low copy centromere plasmid (CEN), its endogenous promoter on a high copy 2-m plasmid, or the galactose promoter (P GAL ) on a low copy centromere plasmid. The strains utilized are MAY2063, -4816, -6013, -6015, -6016, -6022, -6023, and -6024.
b The percentage of observed microtubule structures that were clearly bundled (as opposed to single microtubules).
ing extracts. The average velocity of Cin8p-BCP induced gliding was 3.4 Ϯ 0.5 m/min (S.E. shown in parentheses). A small but significant number of microtubules moved at rates as much as 3 times this average, causing the velocity distribution to be quite broad. The distribution of the overall distance moved by microtubules (running length) is shown in Fig. 5B. About 75% of microtubules moved relatively short distances, smaller than 5 m, before the microtubule either paused or detached from the glass surface. High microtubule densities did not change the episodic nature of movement. The short running length may suggest that Cin8p is less processive than kinesin (37,38), which was also suggested in a kinetic study on the vertebrate BimC motor Eg5 (39).
The effects of added nucleotide on microtubule binding and gliding are shown in Table III. The number of microtubules found on the surface in the presence of AMP-PNP was more than 10-fold larger than that with ATP. However, unlike experiments that lacked added nucleotide (Table III and Fig. 3), a high fraction of the AMP-PNP-bound microtubules could not be blocked by preincubation with biotin. Nonetheless, this residual microtubule binding with AMP-PNP appeared specific to Cin8p; extracts from cin8⌬ cells did not show microtubule surface binding in the presence of AMP-PNP. As expected, AMP-PNP did not support any microtubule gliding. With ADP there were fewer bound microtubule than with ATP or AMP-PNP or in the absence of added nucleotide (Table III). This observation is consistent with what is known for conventional kinesin for which affinity to microtubules is strong in the presence of AMP-PNP or in the absence of added nucleotide and is reduced in the ADP-bound state. We also found that 1 mM GTP was unable to support Cin8p-induced motility. GTP supports motility by kinesin at a reduced level relative to ATP (40,41), but tests on BimC motors have not previously been reported.
The above microtubule binding and motility experiments were performed using extracts in which Cin8p-BCP had been overexpressed. We also used this assay to examine Cin8p activity at closer to normal expression levels. While we have not been able to detect microtubule binding to the glass surface using Cin8p-BCP from cells producing it from its native promoter on a low copy (CEN) plasmid, we have been able to assay Cin8p-BCP activity when produced from its native promoter on a high copy (2-m) plasmid. Under these conditions, Cin8p levels are ϳ35-fold higher than wild type but lower than the ϳ500-fold amplification produced by P GAL . P GAL overexpressed CIN8 is toxic to cells, but the 2-m plasmid is tolerated and complements the growth defect of cin8⌬ kip1⌬ cells. In addition, Cin8p expression from the 2-m plasmid resembles the cell cycle-specific fluctuations of Cin8p levels found in the wildtype situation. 4 Although Cin8p-BCP from 2-m extracts could immobilize microtubules in the presence of AMP-PNP, introducing ATP released all the bound microtubules before motility could be analyzed (Table III). To overcome this problem, we used NPE-ATP, a "caged" nucleotide from which ATP is liberated by exposure to UV irradiation. Prior to uncaging with UV, microtubules were bound to the surface, similar to the absence-ofnucleotide rigor state. Upon activation with UV light, the pulse of released ATP caused gliding of the preadsorbed microtubules. The motility characteristics in caged ATP protocol were slightly different from the simple ATP addition protocol, described above. In direct comparison with the P GAL extracts, fewer microtubules moved, and running lengths were shorter in the caged ATP experiment. The average running length was 3.3 Ϯ 0.3 and 1.1 Ϯ 0.1 m for steady ATP and caged ATP, respectively. However, microtubule gliding velocities were similar for P GAL caged ATP, P GAL steady ATP, and 2-m caged ATP assays (Table III), validating the use of caged ATP to examine Cin8p activity in situations where increased sensitivity is required.
Analysis of Cin8p Motor Domain Mutants in Vivo and in Vitro-An advantage of our experimental system is that it allows us to analyze both the phenotypic and biochemical consequences of motor gene mutations. To demonstrate this potential, we examined two Cin8p forms mutant within their motor domains. One mutant is encoded by cin8 -3, a form that has been subjected to significant in vivo analyses (3,11,12). cin8 -3 causes a change from arginine to lysine at amino acid 196, located in the region corresponding to the ␣2 helix of human kinesin (42). The second mutant form is encoded by cin8-F467A, the change occurring within a region corresponding to the kinesin L12 loop, important for microtubule interaction (43). Both mutations cause temperature-sensitive growth at 33°in the absence of the functionally overlapping KIP1 (data not shown), but both are clearly compromised at room temperature as well (see below). We could detect no difference in the expression level of these two mutant forms compared with wild-type Cin8p when produced at low copy from the CIN8 promoter (data not shown). Therefore, the phenotypic effects of these mutations can be attributed to the differences in their activities. In the experiments described below, we examined wild-type and mutant forms expressed from P GAL in cells that were deleted for CIN8. All analyses were performed at room temperature (ϳ25°C). Fig. 6 compares the ability of the Cin8p mutants to bind microtubule in vitro. Extracts producing the Cin8p-BCP forms were passed over avidin-coated glass surfaces to adsorb the motor. Microtubules were added in the presence of AMP-PNP, and the number bound per 400-m (2) field was determined by microscopy. We also determined, by immunoblotting, that all three forms were expressed at comparable levels (Fig. 6B) and were bound to the glass surface equally well (data not shown). Cin8p-F467A displayed severely reduced microtubule binding. Even higher expression levels from longer induction times (Fig.  FIG. 3. Microtubule binding to Cin8p-BCP on glass surfaces. Extracts were prepared from cells that overexpressed (by P GAL ) either biotinated or untagged Cin8p or were deleted for CIN8. The bars represent the mean Ϯ S.E. of microtubules bound per 400-m 2 field. For each category, 120 fields were counted (six experiments, 20 fields/ category/experiment). For each experiment, one coverslip was divided into four chambers, which each received one of the four indicated extracts (bottom of graph). Microtubule attachment to the glass surface was measured in the absence of added nucleotides. During adsorption, the total extract protein concentration in each chamber was 1 mg/ml. For blocking with biotin, the chamber surface was saturated with biotin after the adsorption of avidin but prior to the addition of cell extract (see "Materials and Methods"). Strains utilized were MAY2063, -6015, and -6016. 6; 4.5 h of induction) did not rescue the reduction in microtubule binding. Cin8p-3 also exhibited a reduction in microtubule binding, but this was less pronounced (typically ϳ50% reduced). By optically focusing on the glass surface, we could estimate the rates of microtubule capture and release. For wild-type Cin8p surfaces, almost all microtubules that landed adhered for at least 5 min (the length of the observation time). For Cin8p-F467A and Cin8p-3 surfaces, 30 -70% of the landing microtubules rapidly detached (within 1 s). However, of the microtubules that were captured to the mutant surfaces, most remained attached for at least 5 min. This suggests that the reduced microtubule binding affinity of the mutant proteins reflects a diminished capture rate rather than an increased detachment rate.
Because of their reduced ability to capture microtubules, experiments for in vitro motility of mutants required the more sensitive caged ATP method (see above). In this set of experiments (Table IV), the average gliding velocity of wild-type Cin8p-BCP measured after uncaging ATP was 2.8 Ϯ 0.5 m/ min, within the margin of error of that measured using ATP (3.4 Ϯ 0.5 m/min). Although Cin8p-F467A was greatly reduced in its ability to bind microtubules, the gliding velocity achieved was reduced by only 35% (Table IV). In addition, the fraction of surface-bound microtubules that responded to ATP uncaging by either moving or detaching was similar to that of the wild type. This finding, combined with the results in Fig. 6, leads us to conclude that the effect of this mutation is mainly on the ability of Cin8p to bind microtubules rather than on motility. In contrast, for Cin8p-3, ATP uncaging resulted in a lower fraction of microtubules moving or detaching from the surface. However, the few motile microtubules moved at a velocity close to the wild type.
To summarize mutant reconstitution studies, the main defect of Cin8p-F467A mutant was its ability to bind microtubules in vitro. In comparison, Cin8p-3 mutant was less im-paired for microtubule binding but was able to move fewer microtubules after ATP uncaging.
A major mitotic role for BimC motors is the assembly of the bipolar spindle. By monitoring the extent of spindle bipolarity and pole-to-pole distances, we were able to detect differential effects of the two Cin8p motor mutants on the spindle assembly process (Fig. 7). After release from ␣-factor block at G 1 , synchronized wild-type cells convert monopolar astral arrays of microtubules into short bipolar spindles, which subsequently elongate when cells enter anaphase. In these in vivo experiments, we used cells deleted for KIP1 encoding the functionally overlapping BimC motor. The CIN8 allelic forms were integrated into the genome and expressed from the native promoter (see "Materials and Methods"). Relative to CIN8, the cin8-F467A cells were impaired in their ability to form bipolar spindles; monopolar structures took longer to disappear, and bipolar structures took longer to appear (Fig. 7, A and B). The bipolar spindles formed by cin8-F467A also were shorter than CIN8 (0.5 Ϯ 0.1 versus 1.1 Ϯ 0.1 m). cin8-F467A cells were also delayed in their entry into anaphase as evidenced by the delayed appearance of cells with elongated spindles (Ͼ2 m; Fig. 7C). In contrast, we could not detect any effect of cin8 -3 on spindle assembly at 25°C, although this mutant may exhibit a slight anaphase entry delay (Fig. 7, B and C). Therefore, cin8-F467A, but not cin8 -3, significantly reduces the ability of Cin8p to form bipolar spindles. Further attempts to observe effects on anaphase proficiency were inconclusive, possibly due to the difficulty in separating the contribution of the dynein motor, which also contributes to anaphase pole separation.

DISCUSSION
The phenotypic analyses of S. cerevisiae cells mutant for BimC-related Cin8p have revealed many aspects of the in vivo function of this family of mitotic spindle motors (8 -10). The observed requirement for Cin8p during spindle assembly and elongation and its localization to microtubules between the spindle poles have led to the suggestion that Cin8p acts to cross-link and slide spindle midzone microtubules. The in vitro analysis presented here demonstrates that Cin8p indeed exhibits the properties expected for a kinesin-related microtubule-based motor protein. Cin8p can bind and bundle microtubules in the absence of nucleotide or in the presence of AMP-PNP. In the presence of ATP, Cin8p induced microtubule gliding in a plus end-directed fashion. If Cin8p acts to cross-link antiparallel midzone microtubules, plus end-directed sliding would lead to spindle elongation.
The assay we developed examines Cin8p isolated from crude extracts of S. cerevisiae cells. An active biotinated form was expressed in S. cerevisiae and rapidly isolated by adsorption to an avidin-coated glass surface. In principle, this assay has several advantages over studies of motors produced in exogenous host cells or assays in which extensive purification steps are required. Since Cin8p acts in a cell cycle-specific fashion, it seems likely that its activity will be subject to cell cycle regulation. We have determined that Cin8p levels fluctuate during the cell cycle, peaking in mitosis and rapidly decreasing during the G 1 stage. 5 It is possible that other forms of post-translational regulation are also imposed upon Cin8p. The rapid purification used in this assay, requiring minimal manipulation of the protein extracts, may permit (hypothetical) regulatory proteins to maintain their association with Cin8p, allowing their influence to be monitored. On account of the low intracellular abundance of Cin8p, we were unable to detect Cin8p activity in extracts of cells in which it was produced at normal levels. We were successful, however, with assays using cells harboring the gene (expressed from its normal promoter) in extra copies on a 2 m-based plasmid. Our crude extract assay is currently being used to examine differences in microtubule binding and translocation proficiency of Cin8p during cell cycle progression.
With our biotin-based assay, we characterized the microtubule binding and motility properties of Cin8p. The nucleotide    Table III, AMP-PNP rows for examples). B, Western blots of extracts used in A, probed with streptavidin-conjugated alkaline phosphatase. The induction time indicates the hours elapsed following the addition of galactose. The strains utilized were MAY6019, -6020, and -6021. specificity for microtubule binding was similar to that reported for other kinesins and exhibits strong binding in the presence of AMP-PNP and weak binding in the presence of ADP (Table  III). Conventional kinesin can move microtubules in the presence of GTP (40,41), but other kinesin-related proteins, such as Ncd, cannot utilize GTP (44,45). In our experimental conditions, GTP was unable to support motility by Cin8p.
In our assay, Cin8p produced microtubule gliding rates of ϳ3.4 m/min. This rate is approximately 10-fold slower than that produced by kinesin in vitro (16) but similar to the 1.0 -2.4 m/min reported for two other BimC motors (15,46,47). The distribution of microtubule velocities determined for Cin8p was broad (Fig. 6A), ranging between 1 and 10 m/min (although most fell in the 1-4-m/min range). A much narrower velocity distribution was reported for the Xenopus BimC homologue, Eg5, but this study utilized a fragment of the motor produced in bacteria (47). The specificity of movement in our assay to biotinated Cin8p makes it highly unlikely that other motors were contributing to this broad range of velocities. It is possible, therefore, that Cin8p motors with different motile properties exist in the extracts examined. It is tempting to speculate that the broad velocity distribution reflects cell cycle regulation of the motile properties of Cin8p.
Comparison of Cin8p Mutants in Vivo and in Vitro-Coupled with the in vivo assays of spindle motors in S. cerevisiae, the rapid in vitro assay described here allows a powerful analysis of the properties of mutant motor forms. In this study, we exam-ined the properties of two Cin8p mutants altered within their motor domains. The Cin8p-3 form is altered at an arginine (changed to lysine) that is conserved among kinesin family members. The corresponding residue in human kinesin lies in the motor domain ␣2, the P-loop helix (42). The role of this helix is not known. In the conditions of our assay, Cin8p-3 was moderately reduced in its ability to bind microtubules and greatly reduced in its ability to move microtubules following ATP uncaging. The ability of Cin8p-3 to form bipolar spindles in vivo (at room temperature) was not affected, suggesting that its motility defect does not impair its ability to assemble spindles.
The second mutant, Cin8p-F467A was specifically created in loop L12, which contains residues highly conserved among kinesin-related proteins (42). Mutant analysis of human kinesin revealed that L12 is the major domain for the kinesinmicrotubule interaction (43). Changes of two hydrophobic residues in the same region of human kinesin to alanine reduced the microtubule-binding affinity (43). In our assay, the F467A change reduced the number of microtubules bound to the Cin8p surface 10-fold (Fig. 6). Despite the pronounced effect on microtubule binding, the gliding velocity produced by Cin8p-F467A was reduced by only 35% (Table IV). A similar effect was also found for the human kinesin L12 mutants for which the effect on velocity was very mild as compared with the effect on the affinity to microtubule (43). Our findings indicate that the L12 region of Cin8p, like kinesin, is important for microtubule binding. In vivo, cin8-F467A was found to be defective for bipolar spindle assembly compared with either wild-type or cin8 -3 (at room temperature). Therefore, changes in different motor domain regions can cause different phenotypic effects. In mitosis, Cin8p acts prior to anaphase to assemble spindles and maintain their structural integrity. During anaphase, Cin8p is almost certainly the most important pole-separating motor in S. cerevisiae (12). Combined with the mutant study here, we suggest that microtubule binding ability is important for spindle assembly, while force production may be important for anaphase spindle elongation.
In summary, we have developed methods that allow the comparative studies of motor proteins in vivo and in vitro. In developing these techniques, we anticipate their applicability to the study of other motor proteins. As was the case for Cin8p, tagging with BCP need not interfere with the normal activity of the motor in vivo. This allows the examination of functional motor forms rapidly extracted from their endogenous host cell.