Purification and Characterization of Native Conventional Kinesin, HSET, and CENP-E from Mitotic HeLa Cells*

We have developed a strategy for the purification of native microtubule motor proteins from mitotic HeLa cells and describe here the purification and characterization of human conventional kinesin and two human kinesin-related proteins, HSET and CENP-E. We found that the 120-kDa HeLa cell conventional kinesin is an active motor that induces microtubule gliding at ;30 mm/min at room temperature. This active form of HeLa cell kinesin does not contain light chains, although light chains were detected in other fractions. HSET, a member of the C-terminal kinesin subfamily, was also purified in native form for the first time, and the protein migrates as a single band at ;75 kDa. The purified HSET is an active motor that induces microtubule gliding at a rate of ;5 mm/min, and microtubules glide for an average of 3 mm before ceasing movement. Finally, we purified native CENP-E, a kinesin-related protein that has been implicated in chromosome congression during mitosis, and we found that this form of CENP-E does not induce microtubule gliding but is able to bind to microtubules.

Conventional kinesin and kinesin-related proteins (KRPs) 1 constitute a rapidly expanding superfamily of microtubuleassociated motor proteins that perform a variety of cellular functions (1,2). Conventional kinesin was originally identified as a fast moving (ϳ30 m/min) microtubule-based motor in squid giant axons (3), and it is believed to function as a heterotetramer consisting of two heavy chains and two light chains (reviewed in Ref. 4). It has since been identified in virtually all cell types and has multiple roles in vesicle trafficking (5)(6)(7)(8)(9); however, it is not yet clear if conventional kinesin has a role in animal cell mitosis. Microinjection of antibodies specific to kinesin heavy chain had no effect on mitotic progression of early sea urchin embryos (10), and no mitotic defects in kinesin heavy chain mutants have been detected in Drosophila melanogaster (11). However, severe defects in early mitoses were found in C. elegans kinesin heavy chain mutants (12). In contrast to conventional kinesin, many KRPs are known to have mitotic functions, including roles in spindle formation, spindle maintenance, and chromosome movement. These mitotic KRPs include members of the BimC subfamily, the Cterminal subfamily, the MKLP1 subfamily, chromokinesins, and others (reviewed in Refs. 13 and 14).
More than 200 kinesins and KRPs have been identified and catalogued. 2 Whereas most of the biochemical characterization performed on these motors has been on recombinantly expressed protein or protein fragments, the characterizations of the biochemical properties of kinesins and KRPs in their native forms from natural host cells have been limited. In the case of conventional kinesin, which has been the most extensively studied kinesin in native form, the cell types from which it has been isolated and thoroughly studied are restricted for the most part to brain tissue and early embryonic cells (reviewed in Ref. 4). Furthermore, only two KRPs have been isolated from natural host cells. These are KRP85/95 and KRP130, both of which were purified from embryonic cells and found to exist in multimeric complexes (16,17).
In order to study human kinesins and KRPs in their native forms, we have developed a method for the purification of microtubule motor proteins from HeLa cells. We were especially interested in mitotic kinesins; thus, we used cells that were blocked in the mitotic stage of the cell cycle. In this study, we describe a method for the large scale purification of native kinesin and KRPs and focus on the purification of three motors: conventional kinesin, HSET (human spleen, embryo, testes; Ref. 18), and CENP-E (centromere protein-E). We found that the active form of purified HeLa cell kinesin lacks light chains and moves microtubules at ϳ30 m/min. In addition, we purified the KRP HSET, a 75-kDa protein that moves microtubules at a rate of ϳ5 m/min. Finally, we purified the native form of the mitotic KRP, CENP-E, and found that although it binds to microtubules, it apparently lacks microtubule gliding activity.

EXPERIMENTAL PROCEDURES
Large Scale Culturing and Harvesting of HeLa Cells-HeLa S3 cells were grown in suspension in 20-liter carboys maintained on large spinner plates at 37°C in half Ham's F-12 medium and half Dulbecco's modified Eagles's medium supplemented with modified Eagle's medium-nonessential amino acids, penicillin (5000 units/ml), streptomycin (5 mg/ml), 10 mM Hepes, 1.5 g/liter glucose, and 2% iron-rich calf serum. Asynchronously dividing cells were collected from exponentially growing cultures after reaching a density of ϳ10 6 cells/ml (ϳ4% of the cells were in mitosis, and 96% were in interphase). A highly enriched mitotic * This work was supported in part by United States Public Health Service Grants NS13560 and CA57291 and by the Materials Research Laboratory Program of the National Science Foundation under Awards DMR96-32716 and DMR00-80034. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Purification of Motors-Approximately 70 liters of cells from several carboys were harvested using a continuous flow rotor adapter system (Sorvall model KSB-R, Kendro Laboratories, Newtown, CT) in a Sorvall RC5B centrifuge. The cell suspension was pumped into a modified SS-34 rotor with a peristaltic pump and centrifuged continuously at 8000 rpm. Cell pellets were washed once by resuspension in 50 ml of 50 mM Pipes, 1 mM EGTA, 1 mM MgSO 4 , and 0.05% sodium azide at pH 6.9 (PEM 50) and sedimented to a pellet in a tabletop clinical centrifuge. The pellet was resuspended in 150 ml of PEM50 buffer containing 1 mM dithiothreitol, 0.1 M microcystin, and a protease inhibitor mixture (2.5 M 4-(2-aminoethyl)benzenesulfonyl fluoride, 1 g/ml pepstatin A, 1 g/ml leupeptin, 1 mM tosyl arginine methyl ester, and 10 g/ml aprotinin) (PEM50-P). The resuspended cells were lysed by pulse sonication at low energy on ice at 0°C. Cell lysates were incubated for 30 min on ice to ensure complete microtubule disassembly in order to maximize the proportion of soluble motor proteins. Lysates were centrifuged at 100,000 ϫ g for 1 h (4°C) to obtain a clarified high speed supernatant (HSS). The HSS, which contained between 1 and 2 g of protein in 50 -100 ml, was batch-adsorbed to 50 -100 ml of hydrated DEAE-cellulose (DE52, Whatman, Maidstone, England) (equal volumes of HSS to hydrated DE52) by incubation on a rocking platform (4°C for 30 min). The slurry was then poured into a 2.5 ϫ 50-cm column and allowed to pack. The flow-through (DE-FT), which contained the motors characterized in this study, was collected. The column was then washed with two column volumes of PEM50-P, which was added to the DE-FT. The remaining adsorbed proteins were eluted with ϳ1 1 ⁄2 column volumes of 150 mM NaCl in PEM50-P, and the eluate, which contained additional motors, was collected but not further pursued here. The proteins present in the DE-FT fraction were precipitated by the gradual addition of ammonium sulfate to 60% saturation (w/v) and incubation (24 h, 4°C). Precipitated proteins were sedimented by centrifugation in a Sorvall centrifuge (SS-34 rotor, 20,000 rpm, 20 min, 4°C) and resuspended in 4 -8 ml of PEM100-P (identical to PEM50-P but containing 100 mM Pipes). Resuspended samples were clarified by centrifugation for 20 min at 20,000 rpm (SS-34 rotor, 4°C), and depending upon the motor of interest, the resulting supernatant (ϳ8 -10 ml) was loaded onto a 2.5 ϫ 120-cm column containing either 500 ml of a BioGel-A 1.5-m or 500 ml of a BioGel-A 15-m gel filtration matrix. Proteins were filtered through the columns with PEM100-P, and 7-ml fractions were collected. HSET and CENP-E were identified in specific column fractions by SDS-polyacrylamide gel electrophoresis and immunoblotting with specific antibodies, and conventional kinesin was identified by immunoblotting and analysis of motor activity (see "Results"). Microtubule motors in the gel filtration fractions were further purified by a microtubule affinity procedure. First, microtubules were polymerized by the stepwise addition of taxol (final taxol concentration, 20 M) to a solution of 150 M purified bovine brain tubulin in PEM100-P buffer at room temperature. The motors were then adsorbed onto the taxol-stabilized microtubules by the addition of the column fractions and 2 mM AMPPNP to the microtubule suspension and incubation on a rocking platform at 4°C for 1 h. The microtubules and adsorbed motors were sedimented by centrifugation in a Beckman L5-50 high speed centrifuge (60 Ti rotor, 4°C, 100,000 ϫ g, 1 h). The resulting pellets were washed two times in 5 ml of PEM100-P containing 20 M taxol, 0.1 mM AMPPNP, and 300 mM NaCl. The washed microtubules were then sedimented as described above and resuspended in 500 l of a release buffer consisting of PEM100-P, 20 M taxol, 10 mM MgATP, and 300 mM NaCl. The motors were released from the microtubules by incubation for 1 h at 0°C on ice with occasional gentle mixing on a vortex mixer or occasional homogenization in a Dounce homogenizer. Samples were centrifuged as described above, yielding released proteins in the supernatants and microtubules and bound proteins in the pellets. Final purification of motors was accomplished by sucrose density centrifugation. The supernatants were applied to 4.6-ml 5-50% sucrose gradients in PEM100-P buffer, and the gradients were centrifuged in a Beckman L5-50 high speed centrifuge using a SW50 swinging bucket rotor for 14 h (200,000 ϫ g, 4°C). Forty-five two-drop fractions were collected by hand after puncturing the bottom of the tube with a 22-gauge needle.
Microtubule Motor Assays-Fluorescent microtubule gliding assays were used to characterize the motors at various steps in the purification procedures. Short stable microtubules were prepared by polymerizing purified bovine brain tubulin in the presence of 1 mM GTP, 1 mM MgCl 2 , and 10% glycerol at 37°C and shearing through a 25-gauge needle five times. Rhodamine-labeled microtubule segments were assembled onto the short microtubules by incubation with a mixture of rhodaminetubulin and purified bovine brain tubulin (8:1 ratio) plus 1 mM GTP and 1 mM MgCl 2 (1 h, 37°C). The fluorescent microtubules were stabilized by the addition of taxol (20 M), sedimented by centrifugation through a taxol-containing 60% glycerol cushion in PEM100 (200,000 ϫ g, 1 h, 4°C), and resuspended in 40 l of PEM100 containing 20 M taxol and antifade reagents (1 mg/ml catalase, 1 mg/ml glucose oxidase, 1% ␤-mercaptoethanol, 5% glycerol, and 0.1 M glucose). The microtubules were diluted 1:20 in PEM100 buffer containing 20 M taxol and the same antifade reagents as above in the gliding assays. Three aliquots of a single protein sample were sequentially perfused into a flow cell chamber (chambers of ϳ3-l volume were made using a glass microscope slide, two pieces of double stick tape, and a glass coverslip) and maintained in a humid environment produced by inverting a glass Petri dish over the chambers surrounded by wet paper towels. Each sample was allowed to remain for 3 min (room temperature) and then removed by wicking with Whatman No. 1 filter paper. Chambers were then washed with 10 l of PEM100-P plus 20 M taxol. Three l of rhodamine-labeled taxol-stabilized microtubules, diluted 1:20 in PEM100-P plus 20 M taxol, were then perfused into the chambers and allowed to incubate for 2 min at room temperature. The chambers were then washed with PEM100-P plus 20 M taxol and 10 mM ATP, and both sides of the chambers were sealed with VALAP (vasoline, lanolin, paraffin; 1:1:1). Images were collected every 3-15 s using a Nikon Eclipse E800 fluorescence microscope (A. G. Heinze, Lake Forest, CA), Hamamatsu Orca II camera, Uniblitz shutter, and controller driven by a Metamorph Imaging System (Universal Imaging, Westchester, PA). For determination of kinesin directionality, we constructed microtubules composed of brightly fluorescent and dimly fluorescent segments (modified from Ref. 19). Brightly labeled microtubule seeds were prepared by incubating 1.5 l of 10 mg/ml rhodamine-labeled bovine brain tubulin, 0.75 l of glycerol, and 0.25 l of PEM100 buffer containing 10 mM GTP for 1 h at 37°C. Seeds were sheared five times with a 25-gauge needle that had the beveled tip removed with scissors. Sheared seeds (0.5 l) were incubated with 2.5 l of 10 mg/ml rhodamine labeled tubulin, 6 l of PEM100, and 1 l of 10 mM GTP (1 h, 37°C). To cross-link the resulting brightly labeled microtubules, 1 l of 15 mM ethylene glycol bis(succinamidyl succinate) was added, the solution was incubated (15 min, 37°C), transferred into 50 l quenching buffer (10 mM sodium glutamate, 0.5 mM ␤-mercaptoethanol, and 50% sucrose), mixed by vortexing, and incubated (room temperature, 1 h). The solution was then applied to a 200-l 60% sucrose cushion in PEM100 and centrifuged for 1 h at 4°C at 200,000 ϫ g. The microtubule pellet was resuspended in 25 l of PEM100 and sheared by passage 10 times through a 25-gauge needle to produce short brightly labeled seeds. Dimly fluorescent microtubules were then polymerized off the plusends of brightly labeled seeds by incubation of 20 l of seeds, 2 l of 10 mg/ml rhodamine-labeled tubulin, 10 l of 10 mg/ml unlabeled purified bovine brain tubulin, 12 l of 10 mg/ml NEM-modified bovine brain tubulin (19), 18 l of PEM100, and 6.7 l of 10 mM GTP for 45 min at 37°C. Microtubule constructs were stabilized by the addition of 3.3 l of 0.4 mM taxol in methanol and gentle vortexing. The stabilized microtubule constructs were applied to two 200-l 60% sucrose cushions in PEM100 containing 20 M taxol and centrifuged for 45 min at 200,000 ϫ g. Final microtubule pellets were resuspended in a total of 32 l of PEM100 containing 20 M taxol and antifade components described above. The fluorescent microtubule constructs were used at a 1:20 dilution in motility assays.
Determination of Sedimentation Coefficients-Proteins with known S values were centrifuged through 5-50% sucrose gradients in PEM100 buffer for 14 h at 200,000 ϫ g. The protein standards used were catalase (11.3 S), ␤-amylase (8.98 S), and bovine serum albumin (4.4 S). Twodrop fractions were collected from each gradient, protein peaks were identified by the Bradford protein assay (20), and the fraction volume was plotted against the known S values to generate a standard curve.
Antibodies-A monoclonal kinesin heavy chain antibody (SUK-4) was used for immunoblots at a dilution of 1:1000 (Covance, Richmond, CA). HSET antibodies were generously provided by Dr. Duane Compton (Dartmouth Medical School, Hanover, NH) and used for immunoblots at a dilution of 1:1000. Antibodies to CENP-E were generously provided by Dr. Tim Yen (Fox Chase Cancer Center, Philadelphia, PA) and used for immunoblots at a dilution of 1:6000.

RESULTS
Our primary goal was to characterize motors from mitotic cells; thus, a highly enriched mitotic cell population was obtained by incubation of a 10 6 cells/ml suspension with 12 nM vinblastine sulfate for one cell cycle (see "Experimental Procedures"). This concentration was chosen to produce a cell popu-lation with a high mitotic index (90%) and intact spindles (21,22). We determined that we could isolate motor proteins in sufficient quantities for characterization from about 70 ml of packed cells, obtained from ϳ70 liters of suspension culture. A flow chart outlining the basic purification scheme for motor proteins present in the mitotic cells is shown in Fig. 1. For the present study, we focused on the purification and characterization of three of the motors: conventional kinesin, HSET, and CENP-E. These motors were identified and followed during the purification procedures using specific antibodies and fluorescent microtubule gliding assays (see "Experimental Procedures").
Purification of HeLa Cell Conventional Kinesin-We found that the initial HSS (see "Experimental Procedures") contained a robust motor activity that moved microtubules at a rapid rate of ϳ30 m/min at 25°C (Fig. 2). The individual microtubules glided for several hundred m without dissociating from the glass surface; thus, the motor appeared to be processive. Based on these properties, we suspected that the fast motor might be conventional kinesin, and we designed a purification scheme for its isolation. The HSS was batch-absorbed to DEAE-cellulose, packed into a column, and eluted with PEM50-P (see "Experimental Procedures"). The protein flow-through fraction (DE-FT; Fig. 1), which contained the fast motor as well as other motors (data not shown), was collected, concentrated, and fractionated on a BioGel-A 1.5-m gel filtration column (Fig. 3). The fractions containing the fast motor (fractions 34 -42) were pooled (Fig. 4, lane 1) and subjected to a microtubule affinity purification step in the presence of AMPPNP, a procedure routinely used for affinity purification of kinesin and KRPs (4). A large number of proteins present in the pooled fractions did not adhere to the microtubules (Fig. 4, lane 2), while the fast motor bound tightly to taxol-stabilized microtubules in the presence of 2 mM AMPPNP (data not shown). The microtubules with the bound fast motor and other proteins were sedimented by centrifugation at 100,000 ϫ g (see "Experimental Procedures") and resuspended in PEM100-P containing 300 mM NaCl. Proteins that dissociated from the microtubules in the presence of NaCl are shown in Fig. 4, lane 3. The microtubules and all proteins that remained bound in the presence of NaCl were again sedimented and resuspended in PEM100P plus 10 mM MgATP and 300 mM NaCl to release the proteins that were bound in an ATP-sensitive manner (Fig. 4, lane 4). The released proteins were then fractionated on a 5-50% sucrose gradient, and two-drop fractions were collected (see "Experimental Procedures"). The fast motor fractionated as a single peak in fractions 26 -29 of the gradient (data not shown). The pooled fractions contained a single polypeptide that migrated at ϳ120 kDa (Fig. 4, lane 5, upper arrow) and a trace amount of residual bovine brain tubulin (Fig. 4, lane 5, lower arrow). No other proteins could be detected on heavily overloaded Coomassie Blue-stained gels or on silver-stained gels (data not shown); thus, we conclude that the motor had been purified to homogeneity. The purified protein had a sedimentation coefficient of ϳ7 S as determined by comparison with proteins of known S values (data not shown).
The 120-kDa protein was identified as conventional kinesin heavy chain by mass spectrometry. The band was excised from a Coomassie Blue-stained SDS gel, and the protein was sequenced by microcapillary reverse-phase high pressure liquid chromatography nanoelectrospray tandem mass spectrometry (Harvard Microchemistry Facility). The sequences of multiple individual tryptic fragments of the protein corresponded exactly with sequences present in human ubiquitous kinesin heavy chain (data not shown). In addition, the kinesin heavy chain antibody, SUK-4, recognized the 120-kDa polypeptide on immunoblots (Fig. 4, lane 5Ј), further indicating that the protein was conventional kinesin.
The active purified kinesin motor did not appear to contain light chains, since no additional protein bands could be detected either by Coomassie Blue stain (Fig. 4, lane 5) or by silver stain (data not shown). To determine if kinesin light chains were present in the active kinesin motor fractions, we performed Western blots using antibodies against the conserved region of kinesin light chain. No cross-reactivity to the antibodies was detected in the active motor fractions. The antibodies did, however, recognize light chains in other gradient fractions 3 and so clearly were able to detect light chains. Therefore, the active human 7 S conventional kinesin that we isolated is composed solely of the heavy chain(s).
Motility Characteristics of Human Kinesin-The purified 7 S kinesin moved microtubules at a rate of 32.2 Ϯ 1.5 m/min at 25°C (n ϭ 240), a rate similar to that of the activity detected in the clarified lysate. When assayed at physiological temperature (37°C), the gliding rate increased ϳ2-fold to 61.5 Ϯ 2.0 m/min (n ϭ 50). The directionality of the kinesin was tested using polarity-marked fluorescent microtubules prepared with NEMlabeled tubulin (19). The microtubules glided with their brightly labeled minus-ends leading (Fig. 5), indicating that the kinesin is a plus-end-directed motor. However, when directionality was tested using polarity-marked microtubules in the absence of NEM, the microtubules glided in both directions (data not shown).
To determine whether the HeLa cell kinesin was similar to other conventional kinesins, we examined the nucleotide specificity of the gliding activity and inhibition by known kinesin inhibitors. As shown in Table I, HeLa cell kinesin could utilize GTP to produce microtubule gliding but at only 10% of the efficiency of ATP. Neither 1 mM ATP␥S, 1,N 6 -etheno-ATP, nor 8-bromo-ATP supported microtubule gliding. Microtubule gliding was supported by 1 mM 2Ј-deoxy-ATP, 3Ј-deoxy-ATP, and 2Ј,3Ј-dideoxy-ATP, but the rate was reduced to 84, 63, and 18%, respectively, of the gliding rate in the presence of 1 mM ATP. As reported in Table II, the HeLa cell kinesin gliding was inhibited with 1 mM AMPPNP, whereas gliding was not affected by 3 DeLuca and Wilson, unpublished observation. HeLa cells blocked in mitosis by vinblastine were used as the source material for purified kinesin, since we were primarily interested in isolating mitotic motor proteins. However, we also purified kinesin from asynchronous HeLa cells, which consisted primarily of cells in interphase (ϳ96%). The biochemical properties of kinesin purified from the interphase cells were not distinguishable from those of the kinesin purified from mitotic cells. Specifically, the active form of the motor consisted only of heavy chains and induced plus-end-directed microtubule gliding at a rate of ϳ30 m/min (data not shown).
Purification of HSET-We found, using specific antibodies, that the DE-FT fraction also contained HSET, the human homolog of Drosophila Ncd (non-claret disjunction). HSET belongs to the C-terminal kinesin motor subfamily, since its motor domain resides in the C terminus of the polypeptide. The basic purification scheme (Fig. 1) used to purify kinesin from mitotic HeLa cells was modified to purify HSET. The proteins in the DE-FT were concentrated and applied to a BioGel-A 1.5-m gel filtration column, from which HSET eluted in the void volume (data not shown). The void volume fractions were pooled and subjected to a microtubule affinity purification step similar to that described previously for kinesin. In Fig. 6, lanes 1-5 show the supernatant and pellet fractions from the microtubule-binding and release steps. Proteins released from the microtubules with ATP (Fig. 6, lane 5) were fractionated on a 5-50% sucrose gradient. SDS-polyacrylamide gel electrophoresis followed by immunoblot analysis revealed that HSET migrated as a single peak in fractions 31-33 of the gradient (data not shown). The peak obtained from the pooled sucrose fractions is shown in Fig. 6, lane 6, as a Coomassie Blue-stained gel and as an immunoblot in lane 6Ј. The protein migrated with a sedimentation coefficient of ϳ4.5 S (data not shown). HSET (Fig. 6, upper arrow) was the only protein detected in the peak other than a minor amount of bovine brain tubulin remaining from the microtubule affinity purification step (Fig. 6, lower  arrow). Thus, the HSET motor contained no associated proteins.
Purified native HSET produced microtubule gliding at a mean rate of 5.8 Ϯ 3.0 m/min (n ϭ 28) at 25°C (Fig. 7). The microtubules glided for an average distance of 3.6 Ϯ 1.0 m (n ϭ 28) before either ceasing to move or detaching from the glass.
Purification of Human CENP-E-One of our initial aims of this work was to purify CENP-E, a mitotic kinesin-related protein (23). Thus, the basic purification scheme for kinesin was modified for purification of CENP-E. Antibodies to the stalk region of CENP-E were used to identify the protein throughout the purification procedure. The majority of CENP-E was detected in the DE-FT, although some CENP-E was detected in the 150 mM eluate (data not shown). We found that a critical step for the purification of CENP-E was the use of a BioGel-A 15-m gel filtration column rather than the Bio-Gel-A 1.5-m column used for kinesin and HSET. Thus, the proteins in the DE-FT were concentrated as described previously for kinesin and applied to a BioGel-A 15-m gel filtration column and eluted with PEM100-P buffer. The fractions containing CENP-E were pooled and subjected to a microtubule affinity purification step as described for kinesin (Fig. 8, lanes  1-4). Proteins released from microtubules with ATP (Fig. 8, lane 5) were fractionated on a 5-50% sucrose gradient. The CENP-E migrated as a discrete peak in fractions 26 -28 with a sedimentation coefficient of ϳ8.5 S. The pooled sucrose-gradient fractions containing the CENP-E are shown in Fig. 8, lane  6, as a Coomassie Blue-stained gel, and the corresponding immunoblot is shown in Fig. 8, lane 6Ј. In the majority of the preparations, no other protein bands were present other than a trace of contaminating bovine brain tubulin remaining from the microtubule affinity step. In a small number of the CENP-E preparations, however, a minor protein band at ϳ125 kDa was apparent on Coomassie Blue-stained gels. By immunoblotting, we determined that this protein was Eg5, a microtubule motor that belongs to the BimC subfamily of KRPs. Eg5 did not appear to co-migrate with CENP-E by BioGel-A 15-m gel filtration, but the Eg5 peak did slightly overlap with the CENP-E. Thus, when the CENP-E fractions were collected from the 15-m column in some of the preparations, there may have been contaminating Eg5, which then co-purified along with CENP-E throughout the microtubule affinity purification step and the sucrose gradient fractionation step.
Initial attempts to demonstrate microtubule binding or gliding by purified CENP-E using the protocol described previously for kinesin (see "Experimental Procedures") were negative. It is conceivable that the CENP-E did not bind to the glass slide under those conditions. To facilitate CENP-E binding to the slide, antibodies to the C-terminal tail domain of CENP-E were

TABLE I
Nucleotide specificity of HeLa cell kinesin a Control gliding is the gliding rate of purified kinesin at 25°C in the presence of 1 mM ATP (32.7 Ϯ 1.9 m/min, n ϭ 92). All nucleotides and analogs were used in the assays at 1 mM, and gliding was tested at 25°C. The results are an average of 3 experiments per nucleotide or analog, and in each experiment, an average of 30 microtubules was scored.  first adsorbed to the slide by adding a solution of antibody (100 g/ml) onto the slide. After a 10-min incubation, unbound antibodies were washed off, and the purified CENP-E was adsorbed to the antibody-coated glass surface. When assayed under these conditions, CENP-E did support microtubule binding (Fig. 9). Binding was specific for CENP-E; glass slides coated with antibody alone or sucrose gradient fractions not containing CENP-E did not support microtubule binding.
Although microtubules did bind to CENP-E, they did not glide upon the addition of ATP at concentrations as high as 10 mM. Many assay conditions were tested, but none supported gliding. Specifically, the pH was varied between pH 5.5 and 8.3, and the NaCl concentration was varied between 0 and 200 mM. In addition, we tried several buffers known to support gliding, including Pipes buffers (10 -100 mM), Tris buffers (10 -50 mM), and Hepes buffers (10 -50 mM). No gliding occurred under any of these conditions. While the CENP-E did not exhibit microtubule motor activity, it did support a pivoting behavior, since microtubules bound to CENP-E-coated slides pivoted about fixed points on the microtubules. Interestingly, pivoting was not dependent on added ATP, since microtubules bound to the CENP-E on slides still pivoted after extensive washing with buffer containing no ATP. No other motor or protein fraction that we tested produced the pivoting movement; thus, this behavior was specific to CENP-E. DISCUSSION A major goal of this work was to develop a strategy for large scale purification of kinesin and KRPs in their native forms from mitotic human cells. While there are advantages to studying recombinantly expressed motors and motor fragments, including large yields and relative ease of purification (24), there are also important advantages to studying native motors.
Namely, native motors may retain modifications acquired in vivo, they may retain their natural folded state, and they may remain complexed with important associated proteins.
We were interested primarily in purifying motors from mitotic cells. Therefore, HeLa cells were an ideal source, since they exhibit a high mitotic index (ϳ90%) when treated with low concentrations of vinblastine (22). Under these conditions, the cells block in late prometaphase with an intact, relatively normal mitotic spindle (21,22). This was important, because we had hypothesized that many mitotic motors might be bound to the spindle microtubules. These mitotically blocked cells differ slightly from naturally occurring mitotic cells in that the blocked cells are incapable of exiting mitosis in a normal fashion. The cells are in a "checkpoint-active" state, in which the cell cycle checkpoint machinery that monitors the progression into anaphase is "on," preventing further cell cycle advancement.
Mitotic HeLa cells proved to be valuable for purification of native mitotic human microtubule motor proteins. In this report, we describe the purification and partial characterization of three motors: conventional kinesin, the C-terminal mitotic KRP HSET, and the mitotic KRP CENP-E. In addition, using this method, we identified and partially purified additional native microtubule motor proteins including the KRPs MKLP1, KLP2, and Eg5 and also a large cytoplasmic dynein complex (data not shown). The purification scheme described in this study for HeLa cell kinesin and KRPs is similar to previously described purification schemes in that a critical enrichment step is the binding of motors to microtubules in the presence of AMPPNP and the release of motors with ATP and NaCl (3,(25)(26)(27)(28). The major difference in our purification is that we developed a strategy for large scale HeLa cell culture growth and harvesting; thus, we were able to purify human kinesins in their native form in sufficient quantities for biochemical study. The large majority of previous studies on purified native kinesin or KRPs has involved the use of either neuronal tissue (3,28) or embryonic cells (16,17,25,27), since these systems can also yield ample protein for biochemical characterization.
Human Conventional Kinesin-We found that mitotic HeLa cell extracts contained an active form of conventional kinesin. The microtubule gliding rate (ϳ30 m/min at 25°C) did not change throughout purification, suggesting that the motor was not altered during the purification procedure. The gliding rate of the HeLa cell kinesin is similar to most, but not all, conventional kinesins previously studied (3,25,26). For example, conventional kinesins purified from sea urchin embryos and from bovine brain tissue glide at ϳ30 m/min (27,28), whereas Drosophila kinesin glides along microtubules at ϳ54 m/min (25), and kinesin from Neurospora glides at ϳ177 m/min (26).
The nucleotide requirements and inhibition patterns of HeLa cell kinesin were similar to those of other previously purified native kinesins (26, 27, 29 -31). Conventional kinesins promote microtubule gliding in the presence of the deoxy-ATP analogs in the order 2Ј Ͼ 3Ј Ͼ 2Ј3Ј and do not promote efficient gliding in the presence of ATP␥S (26,27,31), consistent with the results we obtained (Table I). However, the activities we observed with GTP and 1,N 6 -etheno-ATP do differ from those reported in the literature. In the case of GTP, the HeLa cell kinesin produced gliding at only 10% of the rate it produced with ATP, while both Neurospora and sea urchin egg kinesins use GTP more efficiently, at 50 -100% of control gliding in ATP (26,27). These results may suggest that the nucleotide requirement for HeLa kinesin is more specific than that of other kinesins. Alternatively, the differences may simply reflect contamination of earlier kinesin preparations with a nucleotide diphosphokinase. Also, in contrast with bovine brain and Neurospora (26,31), HeLa cell kinesin did not produce gliding in the presence of 1,N 6 -etheno-ATP. While this may suggest that the kinesins vary in their abilities to hydrolyze 1,N 6 -etheno-ATP, it could be that the various motors have different affinities for nucleotide analogs.
The inhibition pattern of HeLa cell kinesin (Table II) was found to be similar to those patterns of previously characterized conventional kinesins purified from bovine brain tissue, sea urchin eggs, and Neurospora (26,27,30). The gliding activity of the HeLa cell kinesin was insensitive to low concentrations of vanadate, but at concentrations above ϳ50 M, gliding was inhibited, consistent with previous data (26,27,30). In addition, AMPPNP has been shown to reduce or inhibit the gliding activity of conventional kinesins at Ն1 mM (26,27), while the alkylating agent NEM at 1-2 mM has no effect on gliding activity (26,27,30). These findings are consistent with our data regarding the HeLa cell kinesin (Table II).
We examined the directionality of the purified human kinesin on microtubules in which tubulin was grown onto stable seeds in the presence of NEM-modified tubulin ( Fig. 6; see "Experimental Procedures"). This procedure results in microtubule constructs with microtubules grown only at the plusends of the seeds, making designation of direction unambiguous. The microtubules glided with the plus-ends trailing the seeds, indicating that the kinesin produced plus-end-directed gliding (Fig. 5). These results are consistent with previous studies on the motility of conventional kinesin (3,25,27). It is interesting, however, that when we tested the direction of movement with polarity-marked microtubules polymerized without NEM-modified tubulin (32), the HeLa cell kinesin produced both plus-and minus-end-directed microtubule gliding (data not shown). We are currently investigating further the directionality of the purified HeLa kinesin.
A major difference between the active HeLa cell kinesin and previously characterized conventional kinesins is the lack of association of the HeLa cell kinesin with light chains. The HeLa cell kinesin had a sedimentation coefficient of ϳ7 S and did not appear to be associated with any other proteins. Also, light chain antibodies did not react with any proteins in the active purified kinesin preparation, although the antibodies did recognize light chains in other fractions (not shown).
Although the majority of conventional kinesins purified to date are heterotetrameric (reviewed in Ref. 4), containing both heavy and light chains, this is not the first example of an active native kinesin composed solely of heavy chains. Hackney et al. (33) purified two active bovine brain kinesins, a 7 S form consisting only of heavy chains and a 9 S form containing both heavy and light chains. In addition, purified conventional kinesin from Neurospora lacks light chains (26). Finally, we found that HeLa cells do contain a form of kinesin that contains both heavy and light chains, but this form lacks microtubule gliding activity. 3 HSET-HSET belongs to the C-terminal family of KRPs; members of this family have their motor domains in the C terminus as opposed to conventional kinesin, whose motor domain resides in the N terminus. The recombinant forms of the family members Kar3, CHO2, and Ncd exhibit minus-enddirected gliding activity at a rate of 1-8 m/min (34 -36). These motors have been proposed to provide "inward" directed forces on the mitotic spindle to oppose the "outward" directed forces provided by the BimC subfamily (reviewed in Ref. 37). In vitro experiments have suggested that the C-terminal family motors may cross-link spindle microtubules and exert force by motoring toward the MT minus-ends (38). Interestingly, the Xenopus C-terminal family member, XCTK2, does not exhibit microtubule gliding activity in vitro and cannot cross-link microtubules (39). Therefore, the known members of this family may not be functionally equivalent.
We identified HSET in mitotic HeLa cell lysates using an HSET antibody and purified it by modifying the basic purification scheme (Fig. 1) for kinesin. HSET migrated as a single band at ϳ75 kDa on SDS gels and did not appear to purify with any associated proteins. The native HSET is an active microtubule motor that moves microtubules at ϳ5 m/min (Fig. 7). This is consistent with previous results demonstrating microtubule motor activity for several other C-terminal family homologs including Ncd, CHO2, and Kar3 (34 -36). However, this is the first demonstration that a native C-terminal KRP supports microtubule gliding activity and, moreover, that human HSET possesses gliding activity. Many additional experiments probing the biochemical nature of active native HSET are now feasible, including detailed characterization of its motor activity and determination of its microtubule cross-linking ability and ultrastructure.
CENP-E-CENP-E is a cell cycle-regulated, 312-kDa KRP that associates transiently with the kinetochores of mitotic chromosomes (23,40). Disruption of CENP-E function by microinjection of antibodies to CENP-E or by antisense oligonucleotide transfection results in a failure of complete chromosome alignment at metaphase (41,42), suggesting that CENP-E is involved in mitotic chromosome movements. This idea is supported by experiments of Lombillo et al. (43) in which antibodies to CENP-E blocked microtubule depolymerizationdependent minus-end-directed chromosome movements in vitro.
The nature of CENP-E's motor activity and thus its function in mitosis remains poorly understood. In initial attempts to purify CENP-E from HeLa cells, it was found to be tightly associated with a fast, minus-end-directed microtubule motor activity (44). The form of CENP-E purified here does not appear to possess such activity. One simple explanation is that in the previous work, the partially purified CENP-E was contaminated with another motor. Another possibility is that the robust motor activity was associated with a different form of CENP-E. More recently, the recombinant motor domain of the Xenopus homolog, XCENP-E, exhibited slow, plus-end-directed microtubule gliding activity (45). Thus, it is clear that at least certain CENP-E segments can glide along microtubules.
The form of CENP-E we purified from the mitotic HeLa cells does not possess intrinsic microtubule gliding activity but rather may function to tether microtubules, perhaps to the kinetochores. Using a wide range of motility assay conditions, we were unable to detect any microtubule gliding, only microtubule binding. Many of the CENP-E-tethered microtubules pivoted around a fixed point, a behavior that occurred in similar fashion in the presence or absence of ATP. Because pivoting occurred in the absence of ATP, we suspect that the pivoting motion may be driven by diffusion rather than by a genuine motor activity.
It is possible that other forms of CENP-E might function as motors. For example, we found that a portion of the HeLa cell CENP-E remained bound to the DE-52 column and did not emerge in the DE-FT. This CENP-E form, which could be eluted from the column in 150 mM NaCl, has not yet been characterized, but it could have motor activity. In support of this idea, the CENP-E-containing 150 mM NaCl eluate did contain a slow microtubule gliding activity that moved microtubules when assayed in PEM50-P buffer but not in PEM100-P buffer. A second possibility is that the native CENP-E we purified may be modified (e.g. by phosphorylation) so that its gliding activity is inhibited. Consistent with this possibility, phosphorylation of CENP-E during mitosis has been implicated in its microtubule binding ability (46).
Another possibility is that the quantity of native CENP-E in our purified preparation may be too low for detectable gliding. In support of this possibility, it has been previously shown that when very low concentrations of conventional kinesin were examined in motility assays, the kinesin supported a similar behavior to that which we observed with CENP-E (microtubule binding and pivoting about a fixed point) (47,48). However, in those experiments, some microtubules did exhibit gliding, whereas we never observed gliding with the HeLa cell CENP-E.
An interesting possibility is that the role in mitosis of the form of CENP-E we purified involves tethering of spindle microtubules to the kinetochores. In support of this idea, Lombillo et al. (43) found that antibodies to CENP-E caused chromosomes to lose their tethering to microtubules in vitro. Active microtubule gliding activity was not required in their assay, since ATP was not needed for the chromosome movement. This would not be the first example of an essential mitotic KRP that is incapable of producing microtubule gliding, since XKCM1, a KIN I kinesin that is essential for mitosis, does not support microtubule gliding but does induce microtubule disassembly in an ATP-dependent manner (15).
In summary, our experiments with CENP-E are consistent with a model in which the CENP-E binds to mitotic kinetochores and serves to tether the kinetochore microtubules. As suggested by Lombillo et al. (43), CENP-E may be essential for proper poleward chromosome movements by tethering them to actively depolymerizing microtubules. With such a mechanism, gliding activity would not be required for CENP-E's function. Further characterization of the properties of native CENP-E will undoubtedly shed further light on its function.