TPX2 Inhibits Eg5 by Interactions with Both Motor and Microtubule*

Background: TPX2 is a mitotic microtubule-associated protein that regulates the kinesin, Eg5. Results: Full-length TPX2 is a more potent inhibitor of Eg5 velocity than truncated TPX2; differential regulation by TPX2 was not observed for monomeric Eg5. Conclusion: TPX2 inhibits Eg5 as a roadblock and by direct interaction with Eg5. Significance: TPX2 may contribute to the spatial and temporal regulation of the mitotic motor Eg5. The microtubule-associated protein, TPX2, regulates the activity of the mitotic kinesin, Eg5, but the mechanism of regulation is not established. Using total internal reflection fluorescence microscopy, we observed that Eg5, in extracts of mammalian cells expressing Eg5-EGFP, moved processively toward the microtubule plus-end at an average velocity of 14 nm/s. TPX2 bound to microtubules with an apparent dissociation constant of ∼200 nm, and microtubule binding was not dependent on the C-terminal tails of tubulin. Using single molecule assays, we found that full-length TPX2 dramatically reduced Eg5 velocity, whereas truncated TPX2, which lacks the domain that is required for the interaction with Eg5, was a less effective inhibitor at the same concentration. To determine the region(s) of Eg5 that is required for interaction with TPX2, we performed microtubule gliding assays. Dimeric, but not monomeric, Eg5 was differentially inhibited by full-length and truncated TPX2, demonstrating that dimerization or residues in the neck region are important for the interaction of TPX2 with Eg5. These results show that both microtubule binding and interaction with Eg5 contribute to motor inhibition by TPX2 and demonstrate the utility of mammalian cell extracts for biophysical assays.

Accurate chromosome segregation during cell division requires the assembly and function of the mitotic spindle. The spindle is composed of a bipolar array of dynamic microtubules that are required for chromosome alignment and segregation. Mitotic motor proteins play important roles in regulating microtubule organization and dynamics and in generating the forces required for spindle formation and chromosome motion. Despite the characterization of many mitotic motor proteins, how their activity is regulated both spatially and temporally in the spindle remains incompletely understood (1).
TPX2 is a conserved mitotic microtubule-associated protein (MAP) 3 that was originally identified as a protein required for the dynein-dependent targeting of the Xenopus kinesin Xklp2 to mitotic spindle poles (2). In mammalian cells, TPX2 localizes to the nucleus in interphase and to the spindle in mitosis with an enrichment near spindle poles (3,4). Depletion of TPX2 using siRNA results in short bipolar or multipolar spindles that fail to progress through mitosis (3,4). The N terminus of TPX2 binds and activates the mitotic kinase Aurora A and is required to localize the kinase to spindle microtubules (5)(6)(7). During spindle formation, TPX2 is required for microtubule formation near kinetochores, an activity that requires GTP-bound Ran, which relieves the inhibitory action of importin ␣/␤ on TPX2 (8). In addition, it has been demonstrated that the C terminus of TPX2 binds to the bipolar kinesin Eg5 and targets the motor to spindle microtubules (9,10). Expression of TPX2 lacking the C-terminal 35 amino acids, which contribute to Eg5 binding, results in defective spindles with greatly reduced Eg5 on spindle microtubules, unfocussed spindle poles, and bent and buckled microtubules (11).
Because Eg5 plays a critical and conserved function in establishing spindle bipolarity, it is important to understand how this motor is regulated in the spindle. Previous in vitro experiments have shown that purified TPX2 reduces the velocity of Eg5-dependent microtubule gliding and microtubule-microtubule-dependent sliding (11). Eg5 accumulation on microtubules is enhanced in the presence of full-length TPX2, but less in the presence of TPX2 lacking the Eg5 binding domain (11). These results directly demonstrate that TPX2 inhibits the ability of Eg5 to translocate microtubules, but the mechanism of inhibition is not established. Here we use in vitro assays and single molecule TIRF microscopy to characterize the interaction of TPX2 with microtubules and to examine the behavior of Eg5 in the presence of TPX2. Our results demonstrate that TPX2 blocks Eg5 motility both by a direct interaction with Eg5 and by binding to microtubules and acting as a roadblock. Using microtubule gliding assays, we further show that dimeric, but not monomeric, Eg5 is differentially inhibited by full-length and truncated TPX2. Our experiments provide new insight into the microtubule-associated protein TPX2 and its regulation of the mitotic kinesin Eg5.

Experimental Procedures
Materials-All chemicals, unless otherwise specified, were purchased from Sigma-Aldrich.
Cell Culture-LLC-Pk1 cells were cultured in a 1:1 mixture of Ham's F-10 medium and Opti-MEM (Life Technologies, Inc.) containing 7.5% fetal bovine serum and antibiotics at 1ϫ and 5% CO 2 . Cell extracts were made from LLC-Pk1 cells stably expressing LAP-tagged Eg5 from a bacterial artificial chromosome (12). To prepare the extract, a confluent 100-mm diameter cell culture dish was washed twice with 5 ml of room temperature PBS. Then 300 l of extraction buffer containing 40 mM HEPES/KOH, pH 7.6, 100 mM NaCl, 1 mM EDTA, 1 mM PMSF, 10 g/ml leupeptin, 1 mg/ml pepstatin, 0.5% Triton X-100, and 1 mM ATP (13) was added dropwise to the dish and incubated for ϳ2 min without disrupting the monolayer. The cell extract was transferred to a microcentrifuge tube and centrifuged at 14,500 rpm at 4°C for 10 min in a tabletop centrifuge. The supernatant was recovered and aliquoted into small tubes, flash-frozen, and stored in liquid nitrogen. Protein concentration was determined using the method of Lowry (14). For fluorescent intensity measurement experiments with Eg5-mEmerald, an siRNA-resistant Eg5-mEmerald construct was transiently co-transfected into LLC-Pk1 cells with siRNA directed against endogenous Eg5 (target sequence CUGAA-GACCUGAAGACAAU). The extract was made 48 h posttransfection as described. For the extracts used in size exclusion chromatography, cells were not treated with siRNA against endogenous Eg5.
Construction of Plasmids-For bacterial expression, desired nucleotide sequences of human TPX2 constructs (full-length or truncated at amino acid 710) were cloned into a pGEX vector following an N-terminal GST tag and a ULP1 protease cleavage site (15). At the C terminus of TPX2, the stop codon was removed, and the Halo tag sequence was introduced. Constructs were verified by sequencing. For expression in SF9 insect cells, nucleotides coding for full-length human TPX2 or its first 710 amino acids were cloned into the pFast Bac A vector after an N-terminal His 6 tag, and the constructs were verified by sequencing. The virus for infecting the cells was obtained following the Bac-to-Bac protocol (Invitrogen). The plasmid for monomeric Eg5-367 containing the first 367 amino acids of human Eg5 was a kind gift from the laboratory of Dr. Sarah Rice. The plasmid for the expression of dimeric Eg5-513 containing the first 513 amino acids of Eg5 was a kind gift from the laboratory of Dr. Susan Gilbert.
Full-length TPX2-Halo and TPX2-710-Halo were expressed and purified from Escherichia coli Rosetta DE3 pLysS cells. In short, 500 ml of culture was grown to an optical density of 0.5-0.8 and induced with 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside at 18°C for 16 h. The bacteria were harvested and washed with ice-cold distilled water. The cell pellet was resuspended in 2ϫ lysis buffer (60 mM HEPES, pH 7.4, 0.4 mM EGTA, 2 mM DTT, 1.4 g/ml pepstatin, 1.0 mM Pefabloc, 4 g/ml leupeptin, and 2 g/ml aprotinin), diluted to 1ϫ with cold distilled H 2 O, sonicated on ice (three times for 30 s each at maximum setting), and clarified at 15,000 ϫ g for 20 min at 4°C. The supernatant was incubated for 1 h at 4°C with glutathione-Sepharose beads that were pre-equilibrated in lysis buffer. The beads were then washed three times in wash buffer (10% glycerol, 300 mM KCl, 0.1% Triton X-100, 1 mM DTT, 0.7 g/ml pepstatin, and 0.5 mM Pefabloc) and twice in TEV buffer (5 mM Tris, pH 8.0, 150 mM KCl, 10% glycerol, 0.1% Triton X-100, 1 mM DTT, and 0.5 mM Pefabloc). The beads were resuspended in TEV buffer and incubated with 23 M Halo tag Alexa Fluor 660 (Promega, Madison, WI) for 15-20 min at room temperature and then washed to remove unbound ligand. The beads were then resuspended in TEV buffer containing Ulp1 protease and incubated at 16°C for 1 h to cleave protein off of the beads. The supernatant containing the protein was collected by centrifugation, and aliquots were flash-frozen and stored in liquid nitrogen.
Monomeric Eg5-367 was purified from E. coli as described (16). Briefly, 500 ml of bacteria was grown and induced at an OD of 0.5-0.8 with 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside and incubated at 16°C for 16 h. The bacteria were pelleted and washed with ice-cold water. The pellet was resuspended in lysis buffer (10 mM sodium phosphate buffer, pH 7.2, 20 mM NaCl, 2 mM MgCl 2 , 1 mM EGTA, 1 mM DTT, and 0.1 mM ATP with protease inhibitor tablet) and lysed by sonication. The lysate was clarified by centrifuging at 15,000 ϫ g for 30 min at 4°C. The supernatant was incubated with pre-equilibrated nickel-nitrilotriacetic acid-agarose beads for 90 min at 4°C. The beads were then washed in wash buffer (same as lysis buffer containing 20 mM imidazole) and eluted in elution buffer (same as lysis buffer with 300 mM imidazole). The eluate was then dialyzed against buffer containing 20 mM HEPES, pH 7.2, 5 mM magnesium acetate, 0.1 mM EDTA, 0.1 mM EGTA, 50 mM potassium acetate, 1 mM DTT, and 5% sucrose for 4 h at 4°C. Aliquots were flash-frozen in liquid nitrogen and stored at Ϫ80°C. Dimeric Eg5-513 was expressed and purified from E. coli exactly as described (11).

TPX2 Inhibition of Eg5
TPX2 Co-sedimentation with Microtubules-Unlabeled tubulin prepared from porcine brains (17) was polymerized and resuspended in PEM 100 buffer containing 50 M Taxol. 500 nM full-length TPX2 or TPX2-710 was incubated with indicated concentration of unlabeled polymerized microtubules at room temperature for 10 min. The mixture was centrifuged for 10 min at room temperature in a tabletop centrifuge at maximum speed. The supernatant and pellet fractions were carefully separated. Samples for SDS electrophoresis were prepared by boiling the samples with SDS protein sample buffer and run on an 8% polyacrylamide gel. The proteins were then transferred to a PVDF membrane and probed using antibodies against TPX2 (Novus Biologicals, Littleton, CO) and tubulin (DM1A, Sigma-Aldrich). The blots were developed by chemiluminescence and captured on a Bio-Rad imaging station. Analysis of band intensities were performed using ImageJ. Data were plotted using KaleidaGraph and fit with a quadratic equation (15).
TPX2-Halo Microtubule Binding Assays-For TPX2-Halo binding experiments, first 10 l of 10% rat YL 1 ⁄ 2 (0.1 mg/ml) anti-tubulin antibody (Accurate Chemical, Westbury, NY) was added to the flow chamber and incubated for 2 min. Second, 0.1 mg/ml rhodamine-labeled microtubules (untreated or treated with subtilisin A) were flowed in and incubated for 2 min. Subtilisin A-treated microtubules were prepared as described (15). Third, the surface was blocked by adding 5% Pluronic F-127 and incubated for 2 min. For assays done in epifluorescence, the chamber was incubated with the indicated concentration of TPX2-Halo for 2 min in PEM 100 (100 mM K-Pipes, pH 6.8, 2 mM MgSO 4 , and 2 mM EGTA) plus 0.5% Pluronic F-127, 50 M Taxol, 5 mM DTT, 15 mg/ml glucose, 1.23 mg/ml glucose oxidase, and 0.375 mg/ml catalase). Salt (KCl) from a 10ϫ stock of the working concentration was added directly to the buffer. Wide field images were acquired with a constant exposure time.
To measure dwell times of TPX2-Halo and TPX2-710-Halo, experiments were performed using TIRF microscopy.
Eg5 Single Molecule Experiments-The concentrations of Eg5 in the extracts were measured by quantitative Western blots. For the single molecule experiments, the perfusion chambers were made from glass slides, silanized coverslips, and double stick tape. 10 l of 10% rat YL 1 ⁄ 2 anti-tubulin antibody (Accurate Chemical) was flowed into the chamber and incubated for 3 min. Then the chamber was blocked by flowing in 5% Pluronic F-127 for 3 min. Diluted Cy5-labeled microtubules (composed of a mixture of Cy5 tubulin (Cytoskeleton, Inc., Denver, CO) and unlabeled brain tubulin) were flowed into the chamber and incubated for 3 min, followed by a second block of 5% Pluronic F-127. Eg5 was diluted to 1 or 1.5 nM in motility buffer containing PEM 50 (50 mM Pipes, pH 6.9, 2 mM EGTA, 2 mM MgSO 4 ), 0.5% F-127, 5 mM ATP, 1 mM DTT, 25 M Taxol supplemented with an oxygen-scavenging system (15 mg/ml glucose, 1.23 mg/ml glucose oxidase, and 0.375 mg/ml catalase) and flowed into the chamber and imaged. For preincubation experiments with TPX2, the indicated concentrations of TPX2 were added to the motility buffer along with Eg5 in extract and incubated on ice for 2 min before flowing into the chamber.
Kinesin-1 Single Molecule Experiments-Perfusion chambers were made as described above. 10 l of 10% rat YL 1 ⁄ 2 anti-tubulin antibody, 5% Pluronic F-127, and diluted Cy5-labeled microtubules were added sequentially and incubated for 5 min each. The chamber was washed with PEM 100 plus Taxol. Kinesin-1 was diluted in PEM 100 with 10 mM DTT. This was then added to the motility buffer (PEM 100, 25 M Taxol, 0.5% Pluronic F-127, 0.5 mg/ml BSA, oxygen-scavenging system, and 0.5 mM ATP) and flowed into the chamber and imaged. For experiments with TPX2 addition, TPX2-Halo was diluted into the motility buffer (without kinesin-1) and flowed into the chamber during image acquisition.
Microtubule-Microtubule Gliding Assays-Biotinylated, Cy5labeled microtubules were immobilized on silanized coverslips using anti-biotin antibody (Sigma-Aldrich). The chamber was blocked using 5% Pluronic F-127. Eg5-EGFP from extracts was preincubated with rhodamine-labeled microtubules for 3 min, and the mixture was flowed into the chamber. Finally, motility buffer was added, followed by acquisition on a TIRF microscope.
Size Exclusion Chromatography-Eg5-EGFP was purified from SF9 insect cells as per the manufacturer's instructions (Bac to Bac, Invitrogen). The extract from LLC-Pk1 cells was prepared as mentioned before. The Superose 6 10/300 GL column (GE Healthcare) was pre-equilibrated with 10 mM HEPES, pH 7.6, 0.05% Triton X-100, 100 mM NaCl, 1 mM ATP before use. 100 l of the purified protein was loaded onto the column and run at a constant flow rate of 0.2 ml/min. The elution profile of Eg5-EGFP was directly followed by measuring absorbance at 488 nm. For the size exclusion of LLC-Pk1 cell extracts, 175 l of cell extract was loaded on the column and run under identical conditions. The collected fractions were separated by SDS-PAGE, transferred to a PVDF membrane, and probed for the presence of Eg5 using Western blot.
Microtubule Surface Gliding Assays-Perfusion chambers of ϳ10 l in volume were made using glass slides and coverslips with a double stick tape spacer. For gliding assays with the Eg5-367 monomer, the chamber was incubated with anti-His antibody and 2 mg/ml BSA for 3 min, followed by two washes with motility buffer (80 mM PIPES, pH 6.8, 2 mM MgCl 2 , 1 mM EGTA, 0.2 mg/ml BSA, and 150 mM sucrose). Then the chamber was incubated with Eg5-367 for 3 min and washed again with motility buffer. Finally, the activation mix, consisting of motility buffer containing oxygen-scavenging system, ATP, Taxol, and diluted Cy5-labeled microtubules was added and imaged on a Nikon TiE microscope using epifluorescence. Surface gliding experiments with the dimeric Eg5-513 were performed exactly as described by Ma et al. (11). For the TPX2 addition experiments, the TPX2 constructs were added to the activation mix, incubated for 2 min on ice, and then flowed into the chamber.
Microscope Imaging and Analysis-TIRF microscopy was performed using a microscope (Ti-E; Nikon Instruments, Melville, NY) equipped with a ϫ60, 1.4 numerical aperture objective lens. The system was run by Elements software (Nikon Instruments). Images were acquired using a 512 ϫ 512-pixel camera (Cascade II; Photometrics, Tuscon, AZ). A ϫ4 image expansion telescope in front of the camera was used. The nm/pixel ratio was 68.5 nm/pixel. A blue diode laser (488 nm, 50 milliwatts) was used. Images were acquired every 2 or 3 s for 10 min. For two-color TIRF, a 488-nm argon laser and a 647-nm diode laser were used on a custom-built TIRF system on a Nikon TiE stand, run by Elements software. A ϫ60 objective lens was used; exposure times for both red and green illumination were 50 -100 ms. Wide field imaging for Eg5-513 gliding assays and for binding of TPX2-Halo to microtubules was performed using epifluorescence illumination.
Quantification of Gliding Velocity, Single Molecule Velocity, and MSD-The velocity of Eg5-513-and Eg5-367-dependent microtubule gliding movement was calculated using the MTrackJ plugin in ImageJ. To calculate the velocity of Eg5-EGFP single molecules from TIRF images, ImageJ was used to generate a kymograph of moving molecules. Velocities were calculated by manually tracking individual puncta. The data were ported to Excel, and a polynomial 2 trend line was added to the MSD versus time plot to determine the diffusion coefficient (D).

Results
TPX2 Binding to Microtubules-To examine the regulation of mammalian Eg5 by TPX2, we expressed and purified fulllength TPX2 and a truncated version lacking the C-terminal 35 amino acids (referred to as TPX2-710) that mediate the interaction with Eg5 (10, 11) (Fig. 1A). To characterize the microtubule binding of these proteins, we performed microtubule co-sedimentation experiments. Both full-length TPX2 and TPX2-710 co-sedimented with microtubules with apparent dissociation constants of 125 and 240 nM, respectively (Fig. 1, B and C). Both full-length and truncated TPX2 could be released from the microtubule lattice by adding KCl to the buffer, with negligible binding at 250 mM KCl. This demonstrates that, like other microtubule-associated proteins, TPX2 makes ionic interactions with the microtubule lattice ( Fig. 2A).
Microtubule-associated proteins are thought to make electrostatic interactions specifically with the negatively charged C-terminal E-hooks of tubulin, named for the abundance of glutamic acid residues (18). To determine whether the E-hooks are either a requirement for or facilitate TPX2 binding to microtubules, polymerized microtubules were digested with the enzyme subtilisin A to cleave off the E-hooks, and binding of TPX2 and TPX2-710 to control and digested microtubules was measured. The results show that binding of full-length or truncated TPX2 to microtubules was not different for untreated, compared with subtilisin-digested, microtubules (Fig. 2B).
To examine the interaction of individual molecules of TPX2 with the microtubule, we performed single molecule TIRF microscopy of Halo-tagged TPX2 full-length and TPX2-710. Individual molecules were stationary on the microtubule lattice, and at the concentration examined, no enrichment at either end of the microtubule was observed. The average dwell time of full-length TPX2-Halo, measured from image sequences acquired at 2-s intervals for 15 min, was 60.1 s (Fig. 2C).
The average dwell time of Halo-tagged TPX2-710 was 46.6 s and was not statistically different from that of the Halo-tagged full-length TPX2.
Together, these results demonstrate that TPX2 binds to the microtubule lattice with high affinity and that the C-terminal 35 amino acids do not contribute significantly to this interaction. Additionally, TPX2 does not require the tubulin E-hook for microtubule binding, suggesting that other tubulin residues are responsible for the interaction.

TPX2 Inhibition of Eg5
depending on the extract (Fig. 3A). The concentration of TPX2 in these cytoplasmic extracts was less than ϳ1 nM, consistent with the localization of TPX2 to the nucleus during interphase (data not shown).
To analyze Eg5-EGFP motors in cell extracts, the extract was diluted into motility buffer (see "Experimental Procedures") to achieve a final motor concentration of ϳ1 nM. Diluted extract was added to flow chambers containing rhodamine-labeled, Taxol-stabilized microtubules immobilized to the surface using anti-tubulin antibodies (see "Experimental Procedures"). Using TIRF microscopy, bright puncta were observed to bind to the microtubules in the absence of ATP. Upon the addition of ATP, robust motility of nearly all puncta was observed ( Fig. 3B and supplemental Movie 1). We observed an accumulation of motors at one end of most microtubules in the field of view ( Fig.  3B and supplemental Movie 1), indicating that motors remain associated with the microtubule plus-end after motion. This is consistent with the previously observed tethering of microtubules near the microtubule end in sliding assays using Xenopus Eg5 (see below) (21). At higher motor concentration, microtubules were uniformly coated with fluorescence, and individual puncta could not be resolved. The average velocity of individual puncta was 14.7 Ϯ 0.9 nm/s, (S.E.) (n ϭ 205) similar to the velocity of purified Xenopus and Drosophila Eg5 motors (22)(23)(24) (Fig. 3D). The average association time of Eg5 with the microtubules was not determined because motors rarely dissociated over the course of a 10-min movie, and longer movies resulted in photobleaching of individual puncta (supplemental Movie 2). Finally, motor behavior was not altered following storage in liquid nitrogen for several weeks, so a single extract could be used for multiple experiments, making this a robust and versatile method for studying motor behavior (see "Discussion").
To determine the directionality of motor motion, kinesin-1-EGFP, a plus-end-directed motor, was added to the chamber, and the direction of motion was observed. Next, the chamber was washed with 5 chamber volumes of ATP-containing motility buffer to remove the kinesin-1-EGFP, Eg5-EGFP was added to the same chamber, and motor behavior was followed in the same field of view. In all cases, both kinesin-1-EGFP and Eg5-EGFP moved to the same end of the microtubule (Fig. 3C), demonstrating that the motile puncta in the mammalian extract walk to the microtubule plus-end.
Next, we wished to determine whether the Eg5-EGFP motors in the extract were present as tetramers. Because the cells expressing Eg5-EGFP also express endogenous Eg5, the motile motors could be composed of between one and four EGFP molecules. In this cell line, the Eg5-EGFP is not resistant to the siRNA designed to deplete endogenous Eg5 (12). Therefore, to estimate the number of labeled Eg5 motors in the motile puncta, we depleted endogenous Eg5 from parental cells cotransfected with siRNA-resistant Eg5-mEmerald and prepared cell extracts. The average fluorescence intensity of Eg5-mEmerald puncta was measured and compared with the fluorescence intensity of bacterially expressed kinesin-1-EGFP dimers imaged under identical conditions (Fig. 3E). On average, the Eg5-mEmerald puncta were twice as bright as the kinesin-1-EGFP dimers, indicating that Eg5 was predominately tetrameric. The increase in quantum fluorescence yield of mEmerald alone is not sufficient to explain the nearly 2-fold increase in fluorescence intensity (25). Additionally, the siRNA may not deplete 100% of the endogenous Eg5, which could form tetramers with the Eg5-mEmerald, resulting in decreased fluorescence of some puncta.
To determine whether the Eg5-EGFP molecules function as tetramers, we examined the ability of Eg5-EGFP from extracts to cross-link two microtubules. To do this, we added Eg5-EGFP to immobilized microtubules in a flow chamber and then added additional microtubules. The added microtubules bound to the immobilized microtubules and were translocated upon the addition of ATP, demonstrating that Eg5-EGFP was capable of cross-linking and sliding microtubules (Fig. 3F). In addition, the moving microtubule remained associated with the tip of the track microtubule, consistent with previous observations (21). Together, these experiments demonstrate that Eg5-EGFP from extracts is tetrameric (Fig. 3, E and F).
Next, we performed size exclusion chromatography on the extracts from LLC-Pk1 cells. Due to the low abundance of Eg5-EGFP in our extracts, we used cells overexpressing Eg5-mEmerald to aid in the detection. The Western blots of the fractions obtained show that Eg5-mEmerald elutes around the same fractions as the Eg5-EGFP molecules, which are purified from SF9 insect cells, suggesting that the Eg5 molecules obtained from cell extracts are tetramers (Fig. 3G).
Eg5 has been shown to exhibit diffusive behavior on microtubules at physiological salt concentration (22,28). To determine whether mammalian Eg5 present in diluted cell extracts showed similar diffusive behavior, we added increasing concentrations of KCl to the motility buffer and examined motor behavior. At 20 mM KCl, the velocity of Eg5 was 12.1 nm/s, similar to that observed in 0 mM KCl, and the diffusion coefficient, D, obtained from plots of MSD over time, was 1588 nm 2 /s. At 50 mM KCl, motor velocity dropped to 3 nm/s, and the value of D was 4556 nm 2 /s (Fig. 3I). These results demonstrate that motor processivity is dependent on the ionic conditions, consistent with previous results using Xenopus Eg5 (23,28).
Together, our results show that Eg5-EGFP motors in mammalian cell extracts behave in a manner similar to purified Xenopus and Drosophila Eg5 tetramers. Specifically, the velocity, directionality, sensitivity to STLC and FCPT and diffusive behavior in higher ionic strength buffer are all consistent with previously reported properties of purified Xenopus and Drosophila kinesin-5 motors. Somewhat surprisingly, the behavior of mammalian Eg5 motors has not been examined previously. Our data show that these motors are similar to insect and other vertebrate Eg5 motors and distinct from kinesin-5 motors from yeast that show directional switching (29,30). Importantly, the similar properties of the mammalian motors strongly suggest that components that are present in the cell extract do not have a major effect on motor behavior.
Interactions of TPX2 with the Microtubule and with Eg5 Both Contribute to Inhibition of Motility-Previous work demonstrated that the gliding of microtubules by surface-attached Eg5 dimers is inhibited by full-length TPX2 and to a lesser extent by TPX2-710 (11). Full-length TPX2 also inhibits Eg5-mediated microtubule-microtubule sliding (11). In both of these assays, however, the behavior of populations of motors is examined, so how individual Eg5 molecules are regulated by TPX2 was not revealed. To gain a better understanding of the mechanism of inhibition of Eg5 by TPX2, we performed single molecule experiments.
To determine the effect of TPX2 on Eg5 behavior, the cytoplasmic extract containing Eg5-EGFP was diluted in motility buffer and added to chambers of immobilized microtubules, and motors were imaged. Next, TPX2 was added to the chamber during image acquisition (Fig. 4A). For these experiments, the velocity of motors following the addition of TPX2 is expressed as a percentage of the velocity prior to the addition of TPX2. The data show that full-length TPX2 is a potent inhibitor of the velocity of individual Eg5 motors; at 250 nM, TPX2 reduced Eg5 velocity by 83%, and at 50 nM, Eg5 velocity was reduced by 32% (Fig. 4B).
To understand how the interaction of TPX2 with Eg5 contributes to motor inhibition, we repeated the experiment using TPX2-710. The addition of TPX2-710 also substantially reduced the velocity of Eg5-EGFP, indicating that microtubule binding by TPX2 contributes to the reduction in motor velocity. At low concentrations (50 nM), both TPX2 and TPX2-710 showed similar inhibition of Eg5 (32 and 24%, respectively. However, at higher concentrations (250 nM), TPX2-710 was a less effective inhibitor of Eg5-EGFP than full-length TPX2 (inhibition of 53 and 83%, respectively) (Fig. 4B).
The results further show that TPX2 reduces the velocity of Eg5-EGFP motors without inducing dissociation of most motors from the microtubule (Fig. 4A), consistent with the established role of TPX2 in targeting Eg5 to spindle microtu-bules (11). In the presence of TPX2-710, more motors appeared to dissociate from the microtubule, although photobleaching precluded accurate quantification. In some cases, we saw that, following the addition of TPX2 to the motility chamber, motors from solution associated with the microtubule, and these motors also moved with reduced velocity (Fig. 4A).
To confirm the specificity of the Eg5-TPX2 interaction, we added full-length TPX2-Halo covalently tagged with an Alexa 660 ligand to kinesin-1-EGFP dimers in a single molecule assay (Fig. 4C). Consistent with prior results from microtubule gliding assays, TPX2 addition did not alter the motility of kinesin-1-EGFP (11).
To visualize the interaction between Eg5 and TPX2 in the single molecule experiments, we used TPX2-Halo covalently tagged with an Alexa 660 ligand (Fig. 4D). In this experiment, the addition of TPX2-Halo (at 20 nM) reduced the velocity of Eg5-EGFP. Analysis of kymographs showed that individual motors that encountered TPX2-Halo walked at reduced velocity. In some cases, a motor that has reduced velocity can resume motion when it encounters an area of the microtubule that is relatively free of TPX2 (Fig. 4D (right panels) and supplemental Movie 3). Note that in this experiment, a lower concentration of TPX2 was used because at higher concentration, fluorescent TPX2 coated the entire microtubule surface.
We also performed experiments in which 50 nM TPX2 or TPX2-710 was premixed with the motor in motility buffer before addition to the chamber. This method allows Eg5 and TPX2 to potentially interact, and both molecules are introduced to the chamber simultaneously (Fig. 4, E and F). This experiment also showed greater inhibition of Eg5-EGFP by the full-length compared with the truncated TPX2 (Fig. 4, E and F). Interestingly, premixing full-length TPX2 with Eg5-EGFP resulted in greater inhibition than when the same concentration of TPX2 was added to motors prebound to microtubules (60% versus 32% inhibition; Fig. 4, B and F). This result indicates that when the motor and TPX2 bind to the microtubule at the same time, stronger inhibition results. In contrast, when TPX2 is added to motors already bound to microtubules, TPX2 can bind to the microtubule at sites distant from the motors and thus not immediately impact motor velocity. Interestingly, in the case of TPX2-710, inhibition of Eg5-EGFP was similar regardless of whether the motors were premixed or added sequentially (Fig. 4, B and F).
Finally, to exclude the possibility that adding a Halo tag to TPX2 affected the TPX2-Eg5 interaction, we compared inhibition of Eg5 by untagged and Halo-tagged TPX2. As seen in Fig.  4F, inhibition of Eg5-EGFP by TPX2 was not changed by the presence of the Halo tag, demonstrating that the Halo tag was not detectably affecting TPX2-Eg5 interaction (Fig. 4F).
Together, the results of these experiments demonstrate that Eg5 in cytoplasmic extracts is inhibited by TPX2. Full-length TPX2, which can interact with Eg5 and with the microtubule, is a more potent inhibitor than TPX2-710, which lacks the Eg5 interaction domain. However, by binding to the microtubule lattice, TPX2-710 also substantially reduces the velocity of individual Eg5 puncta.

TPX2 Differentially Inhibits Microtubule Gliding by Eg5
Dimers but Not Monomers-To determine how Eg5-EGFP motors are inhibited by TPX2, we performed microtubule gliding assays using Eg5 dimers and Eg5 monomers (see "Experimental Procedures"). Dimers supported microtubule gliding at an average rate of ϳ20 nm/s. The velocity of gliding was reduced to ϳ6 nm/s by 250 nM full-length TPX2; the addition of the same concentration of TPX2-710 reduced the velocity of gliding to ϳ15 nm/s, demonstrating that TPX2-710 was a less effective inhibitor than the full-length protein (Fig. 5A). This result demonstrates that dimeric Eg5 retains the ability to interact with TPX2, consistent with previous in vitro binding assays  (11). In contrast, the velocity of microtubule gliding driven by monomeric Eg5 was inhibited to a similar extent by either fulllength or truncated TPX2 (Fig. 5B). It should be noted that the velocity of microtubule gliding driven by monomeric Eg5 is approximately half the rate of the dimeric construct, presumably due to the uncoordinated action of monomers. Further, our results also show that monomer-driven microtubule gliding is inhibited at lower concentrations of TPX2 or TPX2-710 (Fig. 5B). For example, the addition of 25 nM TPX2 or TPX2-710 nearly completely halted microtubule gliding by monomeric Eg5, whereas a 20-fold greater concentration of TPX2 is required to result in a similar reduction in the velocity of microtubule gliding by Eg5 dimers. The reason for this increased sensitivity is not known, but it may relate to the presence of a single motor head. These results suggest that the stalk region in the dimeric construct or the dimer conformation is required for differential inhibition by TPX2 and TPX2-710.

Discussion
TPX2 Binding to Microtubules-The results of our experiments provide insight into the interaction of TPX2 with the microtubule lattice. Our data show that TPX2 and TPX2-710 bind with relatively high apparent affinity to microtubules; these results are similar to the previously reported K D of 0.5 M for full-length Xenopus TPX2 (31). The similar binding of TPX2 and TPX2-710 suggests that the C-terminal region is not a major contributor to microtubule binding and instead interacts with the motor. The apparent affinity of TPX2 for microtubules is similar to that of other MAPs, including the dynein regulator, She1, MAP2, and Cep170 (15,32,33). The relatively strong interaction of TPX2 with the microtubule lattice is also reflected in the long dwell time measured for individual TPX2 puncta using microscopy (ϳ60 s). Our experiments did not reveal a diffusive component to TPX2 behavior under the conditions used. Under physiological ionic conditions, however, we expect that TPX2 would bind less strongly to the microtubule lattice and could exhibit one-dimensional diffusive behavior that is characteristic of many MAPs. Somewhat surprisingly, our results show that the C-terminal tails of tubulin, the E-hooks, do not contribute to TPX2 interaction with the microtubule. Other MAPs, including Tau, She1, and XMAP215, require the tubulin E-hooks for microtubule binding (15,34,35). Additionally, the processivity of kinesin-1 and dynein motors and the diffusive end targeting of MCAK are enhanced by the E-hooks (36 -38). The observation that the interaction of TPX2 with microtubules does not require the E-hooks indicates that TPX2 binds to tubulin residues that are located between protofilaments, as opposed to along the ridge where the E-hooks are located. This is consistent with the observation that TPX2 does not inhibit single molecule motion of kinesin-1-EGFP dimers on microtubules (Fig. 4C) and previous work demonstrating that TPX2 does not inhibit kinesin-1 in a gliding assay (11). In contrast, Tau, which binds along the outer ridge and requires the E-hook for lattice diffusion (35), induces the release of kinesin motors, including both kinesin-1 and Eg5, from the microtubule. This suggests that the location of MAP binding to the microtubule lattice results in differential effects on motor behavior (e.g. inhibition of motility versus release) (11).
The high affinity interaction of TPX2 with microtubules is consistent with its established role in promoting microtubule assembly near kinetochores (8) and branched microtubules in the spindle (39). However, the interaction of TPX2 with microtubules in cells is dynamic, as evidenced by its poleward motion in the spindle (9). One possibility that can account for these differences is that modifications, such as phosphorylation, or interactions with other binding partners, such as dynein, regulate TPX2 dynamic behavior in cells.
Functional Eg5 from Mammalian Cell Extracts-Previous work showed that the behavior of motors present in extracts of cultured mammalian or budding yeast cells is comparable with the behavior of the purified motor (13,30). We found that the rate of Eg5-EGFP stepping along the microtubule was ϳ14 nm/s in the plus-end direction, which is similar to the values obtained for purified Xenopus or Drosophila Eg5. To our knowledge, our results are the first report of single molecule data on the behavior of Eg5-EGFP tetramers from a mammalian source, although the behavior of human Eg5 dimers has been measured previously (40). Mammalian Eg5-EGFP showed exclusively plus-end-directed motion at low salt and diffusive behavior at higher salt concentrations. At the highest salt concentration that we examined, no processive minus-end directed motion was detected, indicating that mammalian Eg5 does not show directional switching like the yeast homolog, Cin8 (29,30).
There are several potential advantages to using motors present in mammalian extracts for biophysical experiments. For example, the contribution of specific domains or potential phosphorylation sites can be determined using extracts prepared from cells transfected with fluorescent constructs encoding mutant versions of the protein of interest. Similarly, to eliminate binding partners or accessory subunits implicated in motor regulation or function, cells can be treated with siRNA prior to preparation of the extract. In addition, cells can be arrested at particular stages of the cell cycle prior to preparation of the extract to determine how cell cycle-dependent modifications may impact motor function. Finally, the ease of preparation and robust motile behavior demonstrate that biophysical measurements of motors from cell extracts are a powerful tool for future experiments.
TPX2 Inhibits Eg5 by Interactions with Both the Motor and Microtubule-The single molecule data presented here demonstrate that TPX2 has two modes of inhibition for Eg5. Truncated TPX2 that cannot interact directly with Eg5 still bound to the microtubule lattice and substantially reduced the velocity of Eg5. Full-length TPX2, which binds to both the microtubule and the motor, was an even more potent inhibitor of Eg5. These results show that both binding to the microtubule and to Eg5 contribute to inhibition of the motor.
Our work is consistent with previous work demonstrating that MAPs can regulate motor behavior. For example, Tau results in differential regulation of kinesin-1 and dynein; upon encountering a Tau patch, kinesin-1 motors frequently detach from the microtubule, whereas dynein motors are likely to reverse direction or pause (41). However, a direct interaction of either kinesin-1 or dynein with Tau has not been reported. Other MAPs function to target motors to microtubules. For example, Cep170 is important for targeting the kinesin-13, Kif2b, to the spindle (33), and the yeast microtubule-associated protein She-1 prolongs the attachment of dynein to the microtubule in a stalled state in addition to inhibiting dynein motility (15). In contrast, the MAP ensconsin recruits and activates kinesin-1 (42) independent of microtubule binding by ensconsin (43).
Recent in vitro experiments show that TPX2 inhibits the stepping behavior of the kinesin-12, Kif15, the human homolog of Xklp2 (44). TPX2 enhances the binding of Kif15 to microtubules in pelleting assays and increases motor binding to the microtubule under load in optical trapping experiments (44). Purified Eg5 has been shown to detach from the microtubule before stalling (45), suggesting that TPX2 may function to increase Eg5 binding to the microtubule under load, as is the case for Kif15. In mitotic cells, Eg5 and Kif15 act redundantly to establish and maintain spindle bipolarity. Furthermore, minusend-directed forces generated by cytoplasmic dynein antagonize force generated by these plus-end-directed motors. Although TPX2 slows Eg5 and Kif15 motion on microtubules, by increasing the force-generating capacity of these motors, it