Cell cycle-dependent translocation of PRC1 on the spindle by Kif4 is essential for midzone formation and cytokinesis

Abstract

The spindle midzone, a conspicuous network of antiparallel interdigitating nonkinetochore microtubules between separating chromosomes, plays a crucial role in regulating the initiation and completion of cytokinesis. In this study, we report the use of time-lapse microscopy and a human kinesin endoribonucleases RNase III-prepared short interfering RNA (esiRNA) library to identify Kif4 as a motor protein that translocates PRC1, a spindle midzone-associated cyclin-dependent kinase substrate protein, to the plus ends of interdigitating spindle microtubules during the metaphase-to-anaphase transition. We show that Kif4 binds to PRC1 through its “stalk plus tail” domains and Kif4 and PRC1 colocalize on the spindle midzone/midbody during anaphase and cytokinesis. Suppression of Kif4 expression by Kif4 esiRNA results in the inhibition of PRC1 translocation, a block of the midzone formation, and a failure of cytokinesis. PRC1 translocation and midzone formation can be restored, and the cytokinetic defects can be rescued in Kif4 esiRNA-treated cells by coexpression of Kif4 but not its motor dead mutant Kif4md. Furthermore, we show that cyclin-dependent kinase phosphorylation of PRC1 controls the timing of PRC1 translocation by Kif4. These results, in light of the crucial role of PRC1 in midzone formation, indicate that cell cycle-dependent translocation of PRC1 by Kif4 is essential for midzone formation and cytokinesis.

The spindle midzone, a conspicuous network of antiparallel interdigitating nonkinetochore microtubules (MTs) between separating chromosomes, has been shown to play an essential role in regulating the initiation and completion of cytokinesis in animal cells (1). Midzone assembly occurs during the metaphase-to-anaphase transition, at the time of cleavage furrow initiation. Recent studies from cultured mammalian cells, Caenorhabditis elegans and Drosophila, have begun to reveal factors that are involved in the midzone assembly process. These factors include the kinesin-like motors and the associated proteins, chromosomal passenger proteins, kinases, phosphatase, and the spindle midzone bundling protein PRC1 (2–7).

The kinesins are a family of MT-based motor proteins that generate directional movement along MTs and are involved in many crucial cellular processes including cell division (8). Gene disruption experiments in Saccharomyces cerevisiae indicated that five of six kinesins play crucial roles in regulating formation of the spindle and segregation of chromosomes in mitosis (9). In higher eukaryotes, several studies have shown that a number of kinesins are crucial for spindle assembly and function, chromosome segregation, mitotic checkpoint control, and cytokinesis (10–12). For instance, the midzone-associated kinesin MKLP-1 was reported to play an essential role in regulating midzone/midbody formation and cytokinesis in various organisms (5, 7). Klp3A, a Drosophila homologue of human kinesin-4 member Kif4, was shown to be essential for midzone formation and/or maintenance during cytokinesis (13–15). However, the precise role for Klp3A (or Kif4) in regulation of midzone formation is not clear.

Previously, we identified the human mitotic spindle midzone-associated cyclin-dependent kinase (Cdk) substrate protein PRC1 and showed that PRC1 plays an essential role in regulating midzone formation and cytokinesis (4, 16). PRC1 has MT binding and bundling activities, and Cdk phosphorylation of PRC1 appears to be important for negatively regulating PRC1 function in early mitosis (4). In this study, we report the identification of Kif4 as a motor protein that translocates PRC1 to the plus ends of interdigitating MTs on the spindle during the metaphase-to-anaphase transition. We also show that Cdk phosphorylation of PRC1 controls the timing of PRC1 translocation by Kif4. Our results indicate that the cell cycle-dependent translocation of PRC1 by Kif4 plays an essential role in midzone formation and cytokinesis.

Materials and Methods

Plasmids, a Human Kinesin/Dynein Endoribonucleases RNase III-Prepared Short Interfering RNA (esiRNA) Library, and Antibodies. The cDNAs of full-length coding region of Kif4 and its point mutation or deletion mutant were generated by PCR and subcloned into BglII–SalI sites of the mammalian expression vector pEGFP-C1. Fig. 1 summarizes these Kif4 constructs: (i) EGFP-Kif4; (ii) EGFP-Kif4md (motor dead), in which the Kif4 ATP-binding Walker A site GKT (amino acid residues 93–95) were changed to AAA; (iii) EGFP-motor plus stalk domains of Kif4; (iv) EGFP-stalk plus tail domains of Kif4; (v) EGFP-motor domain of Kif4; (vi) EGFP-stalk domain of Kif4; and (vii) EGFP-tail of Kif4. Full-length human PRC1 cDNA (16) was subcloned into EcoRI/SalI sites of the mammalian expression vector pEYFP-C1 or pFLAG-1. pECFP-H2B plasmids were generated as described (17). All constructs were fully sequenced. A human kinesin/dynein esiRNA (18) library will be described elsewhere (unpublished work).

Fig. 1.

Abnormal subcellular localization of PRC1 and failure of cytokinesis in HeLa cells treated with Kif4 esiRNA. HeLa cells grown on glass-bottom microwell dishes were transfected with pEYFP-PRC1 and pECFP-H2B together with 100 nM control (luciferase) (A) or Kif4 (B) esiRNA. Thirty-six hours posttransfection, cells were placed on a 37°C heated stage, and time-lapse images were collected every 2 min for 10 h with a Zeiss Axiovert 100M inverted fluorescence microscope and an automatic digital charge-coupled device camera. The movies were edited with slidebook 3.0 software. Representative images of the movies are shown (EYFP-PRC1, pseudocolored red and ECFP-H2B, pseudocolored green). Blue arrows indicate midzone/midbody localization of PRC1 in control esiRNA-treated cells, and black arrows indicate abnormal spindle localization of PRC1 in Kif4 esiRNA-treated cells. The numbers indicate time periods in minutes. For details, see Movies 1 and 2.

Rabbit anti-Kif4 antibodies (481/482) were generated against His-tagged Kif4-C-term fusion protein (amino acids 1000–1232). Alexa Fluor 594-conjugated anti-PRC1 antibodies were made following the instruction for Amine-Reactive Probes (Molecular Probes). Anti-Cdk-phosphorylated PRC1 (T481) antibodies were purchased from Santa Cruz Biotechnology. All secondary antibodies were purchased from Jackson Immunoresearch.

Cell Culture, Transfection, Immunoprecipitation, Immunoblotting, and Immunofluorescence Analyses. HeLa cells were cultured in 6- or 24-well plates in DMEM supplemented with 10% FBS and transfected with 100 nM esiRNA and/or 0.2–1 μg of plasmid(s) by using Oligofectamine (Invitrogen). Two to 3 days after transfection, cells were harvested or fixed for immunoprecipitation, immunoblotting, or immunofluorescence analysis as described (16).

Time-Lapse Microscopy. HeLa cells grown on 35-mm glass-bottom microwell dishes (MatTek, Ashland, MA) were transfected with 0.5 μg of pEYFP-PRC1 and pECFP-H2B plasmids together with 100 nM of Kif esiRNAs by using Oligofectamine (Invitrogen). Twenty-four hours after transfection, cells were cultured with CO₂-independent medium (GIBCO) containing 10% FBS overnight. Then, dishes were covered with mineral oil (Sigma) and transferred to a heated stage (37°C) on a Zeiss Axiovert 100M microscope. Phase-contrast and fluorescence images of live cells were collected at 2-min intervals for 9–10 h and processed by using slidebook 3.0 software (Intelligent Imaging Innovations, Santa Monica, CA).

Cell Cycle Synchronization and Pull-Down Assay. For cell cycle synchronization, HeLa cells were first treated with 2 mM thymidine for 16 h, released into fresh medium for 6 h, and then blocked with medium containing 40 ng/ml nocodazole for 12 h as described (16). Mitotic cells were collected by shake-off, washed with PBS, transferred into fresh medium, and harvested every 20 min.

For pull-down assay, 1 μg of bacterially expressed glutathione-agarose beads with bound GST or GST-PRC1 were incubated with or without 0.2 μg of baculovirus-expressed purified Cdc2/cyclin B1 in 20 μl of Cdk kinase buffer containing 100 μM ATP at 30°C for 2 h as described (16). After washing with cell lysis buffer, the beads were incubated with 150 μg of early mitotic HeLa cell lysates at 4°C for 2 h followed by washing three times with 1 ml of cell lysis buffer. Beads-bound Kif4 protein was analyzed by immunoblotting with anti-Kif4 antibodies.

Results and Discussion

Kif4 Is Required for the Spindle Midzone/Midbody Localization of PRC1 and Cytokinesis in Human Cells. PRC1 associates with the mitotic spindle poles and the mitotic spindle in early mitosis and then translocates to the spindle midzone/midbody during anaphase and telophase (ref. 16 and Fig. 5, which is published as supporting information on the PNAS web site). Because PRC1 is not a MT-based motor protein, we hypothesized that the movement of PRC1 along the spindle during mitosis and cytokinesis would be regulated by motor proteins. To identify such motor protein(s), we screened a human kinesin/dynein esiRNA library that we recently generated (see Materials and Methods)by time-lapse microscopy using HeLa cells expressing enhanced yellow fluorescent protein (EYFP)-PRC1 and enhanced cyan fluorescent protein (ECFP)-histone H2B fusion proteins. HeLa cells were simultaneously transfected with EYFP-PRC1 and ECFP-H2B plasmids together with individual human kinesin/dynein esiRNAs, and live cell images of transfected cells were obtained by time-lapse microscopy. Fig. 1 A and Movie 1, which is published as supporting information on the PNAS web site, show time-lapse fluorescent and phase-contrast images of control (luciferase) esiRNA-transfected HeLa cells expressing EYFP-PRC1 and ECFP-H2B. Consistent with the previous immunofluorescence analysis (4), ectopically expressed EYFP-PRC1 localized in cytoplasmic cytoskeleton arrays in interphase and then localized on the mitotic spindle during mitosis whereas ECFP-H2B was in the nuclei in interphase and associated with chromosomes during mitosis. EYFP-PRC1 associated with the mitotic spindle poles and the mitotic spindle in early mitosis, translocated to the spindle midzone during anaphase, and then concentrated to the midbody during cytokinesis. Like untransfected cells, cells expressing EYFP-PRC1 and ECFP-H2B progressed through mitosis and cytokinesis normally.

Cotransfection of the majority of kinesin esiRNAs with EYFP-PRC1 and ECFP-H2B in HeLa cells did not perturb the dynamic subcellular localization of EYFP-PRC1 during mitosis and cytokinesis. However, cells transfected with esiRNA against one kinesin, Kif4, displayed striking abnormalities of subcellular localization of EYFP-PRC1 during mitosis and cytokinesis. As shown in Fig. 1B and Movie 2, which is published as supporting information on the PNAS web site, although EYFP-PRC1 localized on the mitotic spindle in Kif4 esiRNA-treated cells in early mitosis, it could not translocate to the spindle midzone/midbody during anaphase and telophase. Instead, EYFP-PRC1 was found along the entire mitotic spindle. Kif4 esiRNA-treated cells formed very elongated anaphase spindles with separating chromosomes that moved extremely close to the spindle poles; midzone formation was not evident. As the cleavage furrow formed and ingressed in Kif4 esiRNA-treated cells, the EYFP-PRC1-associated mitotic spindle twisted around. Ultimately, these cells failed to complete cytokinesis and became binucleated cells. Perturbation of translocation of endogenous PRC1 to the midzone/midbody in Kif4 esiRNA-treated cells also was observed by immunofluorescence analysis (Fig. 5). Because PRC1 is crucial for spindle midzone formation and cytokinesis (4, 16), the results suggested that Kif4 might be the motor protein that translocates PRC1 along the mitotic spindle to the plus ends of interdigitating MTs, which is necessary for organizing midzone formation and cytokinesis.

Kif4 and PRC1 Associate with Each Other and Colocalize on the Mitotic Spindle During Mitosis and Cytokinesis. If Kif4 is the motor protein that translocates PRC1 along the mitotic spindle during mitosis, it should interact with PRC1. To test this, human 293 cells were cotransfected with mammalian expression plasmids expressing FLAG-tagged PRC1 together with EGFP (control), EGFP-tagged Kif4 (WT), or EGFP-tagged Kif4md (a motor dead mutant in which the ATP binding Walker A consensus site was mutated from GKT to AAA) (Fig. 2A). Two days after transfection, cells were lysed, and cell lysates were immunoprecipitated with anti-FLAG antibody. Immunoblotting analysis with anti-GFP antibody showed that EGFP-Kif4 and EGFP-Kif4md coimmunoprecipitated specifically with FLAG-PRC1, indicating that FLAG-PRC1 and EGFP-Kif4 interacted with each other in vivo regardless of the motor activity of Kif4 (Fig. 2B). We also observed coimmunprecipitation of endogenous PRC1 and Kif4 in late mitosis of HeLa cells by using affinity-purified anti-PRC1 and anti-Kif4-specific antibodies (see Fig. 4). Together, the results indicated that Kif4 and PRC1 specifically associated with each other in human cells.

Fig. 2.

Kif4 and PRC1 interact with each other in vivo and colocalize at the mitotic spindle and spindle midzone/midbody during mitosis and cytokinesis. (A) A schematic illustration shows EGFP-tagged Kif4, its motor dead version, and a series of deletion mutant constructs used in coimmunoprecipitation assays (lanes a–g; for details see Materials and Methods). The asterisk indicates triple point mutations in the ATP-binding Walker A consensus site of the Kif4 motor domain (GQTGSGKTYSMG to GQTGSAAAYSMG). (B) Coimmunoprecipitation analysis of EGFP-tagged Kif4 or EGFP-tagged Kif4md (motor dead mutant) with FLAG-tagged PRC1. EGFP-tagged Kif4 (lanes a) or EGFP-tagged Kif4md (lanes b) were coexpressed with FLAG-tagged PRC1 in 293T cells. Cell lysates were made and immunoprecipitated with anti-FLAG antibody. Immunoprecipitates (IP) and 1/10 of whole-cell lysates (WCL) used in immunoprecipitations were subjected to SDS/PAGE, transferred to poly(vinylidene difluoride) membrane, and immunoblotted with anti-GFP antibody (Upper) or anti-FLAG antibody (Lower). (C) Coimmunoprecipitation analysis of EGFP-tagged Kif4 or its mutant proteins with FLAG-tagged PRC1. EGFP-tagged Kif4 or its mutant proteins (lanes a and c–g) were coexpressed with FLAG-tagged PRC1 in 293T cells. Cell lysates were made and immunoprecipitated with anti-FLAG antibody. Immunoprecipitates (IP) (Left) and 1/10 of whole-cell lysates (WCL) used in immunoprecipitations (Right) were subjected to SDS/PAGE, transferred to poly(vinylidene difluoride) membrane, and immunoblotted with anti-GFP antibody (Upper) or anti-FLAG antibody (Lower). (D) Colocalization of Kif4 and PRC1 on the mitotic spindle and spindle midzone/midbody during mitosis and cytokinesis is shown. Asynchronous HeLa cells grown on coverslips were fixed with 3% formaldehyde and then stained with rabbit anti-Kif4 antibodies (green), Alexa Fluor 594-conjugated anti-PRC1 antibodies (red), mouse anti-α-tubulin antibody (blue), and DAPI (DNA, white). Kif4 and PRC1 colocalized on mitotic spindle in early mitosis and spindle midzone and midbody in late mitosis. (Scale bar: 5 μm.)

Fig. 4.

Kif4 specifically binds to unphosphorylated PRC1. (A) HeLa cells were synchronized at the G₂/M boundary by a thymidine/nocodazole treatment. Mitotic cells were collected by shake-off, released into fresh medium, and harvested at 10, 30, 50, 70, or 90 min after release. Percentages of cells at different stages of mitosis were determined by counting under a microscope. Whole-cell lysates (WCL) were made, subjected to SDS/PAGE, and immunoblotted with corresponding antibodies as indicated. (B) Lysates as in A were immunoprecipitated by anti-PRC1 or anti-Kif4 antibodies. The immunoprecipitates (IP) were subjected to SDS/PAGE and immunoblotted with anti-Kif4, anti-phospho-PRC1, or anti-PRC1 antibodies. (C) One microgram of bacterially expressed glutathione agarose-bound GST or GST-PRC1 was incubated with or without 0.5 μg of purified baculovirus-expressed Cdc2/cyclin B1 (K1/B1) in the presence of 100 μM ATP for 2 h at 30°C. (Middle) After washing, 1/10th of beads were subjected to SDS/PAGE and immunoblotted with anti-phospho-PRC1 antibodies. The rest of the beads were then incubated with 200 μg of early mitotic HeLa cell lysates (10 min release from a thymidine/nocodazole block) for 2 h at 4°C. (Top) After washing, bead-bound proteins were subjected to SDS/PAGE and immunoblotted with anti-Kif4 antibodies. (Bottom) One microgram of GST, GST-PRC1, and Cdk-phosphorylated GST-PRC1 was used in the pull-down assay on SDS/PAGE by Coommasie blue staining. (D) The proposed model for cell cycle-dependent translocation of PRC1 to the plus ends of interdigitating MTs by Kif4 during the metaphase-to-anaphase transition (for details, see text). APC, anaphase promoting complex.

The coimmunoprecipitation assays were used to determine the Kif4 interaction domain(s) needed for association with PRC1. FLAG-PRC1 and a series of EGFP-tagged deletion mutants of Kif4 were coexpressed in 293 cells, and cell lysates were immunoprecipitated with anti-FLAG antibody (Fig. 2 A). Immunolotting analysis of these immunoprecipitates with anti-GFP antibody indicated that the stalk plus tail domains of Kif4 specifically coimmunoprecipitated with FLAG-PRC1 (Fig. 2B). These results indicated that Kif4 requires its stalk and tail domains to interact with PRC1 in vivo.

We then examined the subcellular localization of Kif4 and PRC1 during mitosis and cytokinesis in HeLa cells by using purified anti-Kif4 and anti-PRC1 antibodies. Immunofluorescence analysis showed that both endogenous Kif4 and PRC1 were nuclear proteins in interphase cells (data not shown). Kif4 and PRC1 colocalized to the mitotic spindle in early mitosis (Fig. 2D), and in anaphase translocated to the spindle midzone where they colocalized as a series of narrow MT-bundle bars at the midzone. During cytokinesis, Kif4 and PRC1 were coincidentally concentrated on the midbody. Thus, the association of Kif4 with PRC1 and colocalization of Kif4 with PRC1 on the mitotic spindle further support the hypothesis that Kif4 is the motor protein that translocates PRC1 along the mitotic spindle during mitosis.

Previous studies reported that Kif4 also colocalized with chromosomes during mitosis (19). We found that detection of endogenous Kif4 on chromosomes was inefficient under the conditions of formaldehyde fixation. However, under the conditions of methanol fixation or ectopic expression of EGFP-tagged Kif4, we observed chromosomal association of Kif4 (Fig. 3 and Fig. 6, which is published as supporting information on the PNAS web site). The role of Kif4 in chromosome segregation was reported recently (20) although we did not observe any significant defects in chromosome congression, alignment, or segregation in cells treated with Kif4 esiRNA (Figs. 1 and 5).

Fig. 3.

Restoration of the localization of PRC1 and rescue of cytokinetic defects in Kif4 esiRNA-treated cells by ectopic expression of GFP-Kif4. (A) HeLa cells grown in a 6-well plate were transfected with 400 ng of pEGFP, pEGFP-Kif4, or pEGFP-Kif4md. Twenty-four hours later, cells were transfected with 100 nM of control (luciferase) esiRNA or Kif4A esiRNA. Two days after esiRNA transfection, cells were lysed and lysates were subjected to SDS/PAGE and immunoblotted with anti-Kif4 antibody or anti-β-actin antibody. (B) HeLa cells grown on coverslips were transfected with plasmids and esiRNAs as in A. After transfection, cells were fixed and stained with anti-PRC1 antibodies (red), anti-α-tubulin antibody (blue), and DAPI (DNA, white). The expression of GFP proteins is shown in green. (Scale bar: 5 μm.) (C) HeLa cells grown on coverslips were transfected with plasmids and/or esiRNAs, fixed, and stained with antibodies and DAPI as in B. The percentages of polyploidy in GFP-negative or GFP-positive cells transfected with the indicated esiRNAs and/or plasmids were scored under a fluorescence microscope. More than 300 cells were counted in each experiment.

Ectopic Expression of Kif4, but Not the Kif4md Mutant, Relocalizes PRC1, Reestablishes the Midzone, and Rescues the Cytokinetic Defect in Kif4 esiRNA-Treated Cells. If Kif4 is required for translocating PRC1 to the ends of interdigitating MTs and allows PRC1 to bundle them and form the midzone, one might expect that expression of exogenous Kif4, but not the Kif4md mutant, would restore the midzone localization of PRC1, reestablish the midzone, and rescue the cytokinetic defects we observed in Kif4 esiRNA-treated cells. To this end, we performed rescue experiments in Kif4 esiRNA-treated cells. HeLa cells were cotransfected with Kif4 esiRNA, which targets the 3′ UTR of Kif4 mRNA, together with mammalian expression vectors expressing EGFP (control), EGFP-Kif4, or EGFP-Kif4md by using Kif4 cDNA lacking the native 3′ UTR. As shown in Fig. 3A, immunoblotting analysis indicated that cotransfection of Kif4A esiRNA with EGFP, EGFP-Kif4, or EGFP-Kif4md plasmid in HeLa cells effectively ablated the expression of endogenous Kif4. However, cotransfection of Kif4 esiRNA with EGFP-Kif4 or EGFP-Kif4md plasmid did not affect the expression of the exogenous EGFP-Kif4 or EGFP-Kif4md in these cells. These results indicated that Kif4 esiRNAs specifically and effectively suppress the expression of endogenous Kif4 but not the expression of exogenous EGFP-Kif4 or EGFP-Kif4md.

Expression of EGFP-Kif4 protein, but not EGFP-Kif4md or EGFP, restored the midzone localization of PRC1, reestablished the mitotic spindle midzone, and rescued the cytokinetic defects in endogenous Kif4-depleted cells. As shown in Fig. 3 B and C, when compared with cells transfected with Kif4 esiRNA alone (GFP-cells) or cells cotransfected with Kif4 esiRNA and control vector EGFP, the majority of cells cotransfected with Kif4 esiRNA and EGFP-Kif4 plasmid showed normal spindle morphology during mitosis and cytokinesis, similar to that of untransfected cells (Figs. 3B and 5). In cells cotransfected with Kif4 esiRNA and EGFP-Kif4 plasmid, the midzone was reestablished, EGFP-Kif4 and PRC1 colocalized with interdigitating MT bundles in anaphase, and the number of binucleated/multinucleated cells was dramatically reduced (Fig. 3 B and C and Fig. 7, which is published as supporting information on the PNAS web site). By contrast, in cells cotransfected with Kif4 esiRNA and EGFP-Kif4md plasmid, the midzone localization of PRC1 was not restored and the midzone was not reestablished even though PRC1 and EGFP-Kif4md colocalized on the abnormal mitotic spindle (Fig. 3B). The percentage of binucleated/multinucleated cells was dramatically increased in cells cotransfected with Kif4 esiRNA and EGFP-Kif4md plasmid when compared with cells transfected with Kif4 esiRNA alone, suggesting that EGFP-Kif4md might have dominant-negative effects (Fig. 3C). Taken together, these results indicated that the expression of EGFP-Kif4, but not its motor dead mutant EGFP-Kif4md, in Kif4 esiRNA-treated cells compensated for the depletion of endogenous Kif4, which was necessary for relocalizing PRC1 to the plus end of interdigitating MTs of the spindle, reestablishing the midzone, and completing cytokinesis.

Cdk Phosphorylation of PRC1 Controls the Timing of Kif4-Mediated PRC1 Translocation to the Plus Ends of Interdigitating MTs at the Metaphase-to-Anaphase Transition. PRC originally was identified as a Cdk substrate in an in vitro phosphorylation screen (16). It has been suggested that Cdk phosphorylation of PRC1 negatively regulates its function during mitosis (4, 16). One attractive possibility is that Cdk phosphorylation of PRC1 negatively regulates its interaction with Kif4. In this scenario, PRC1 is phosphorylated by Cdks (mainly Cdc2/cyclin B) in early mitosis, and Cdk phosphorylation of PRC1 would prevent PRC1 interaction with Kif4 although both PRC1 and Kif4 associate with mitotic spindle. During metaphase-to-anaphase transition, when all cellular Cdk activity is eliminated because of the degradation of cyclin B by the anaphase-promoting complex, PRC1 would be dephosphorylated by an active mitotic phosphatase (e.g., Cdc14A or PP1γ). Dephosphorylation of PRC1 would promote interaction of Kif4 with PRC1, allowing Kif4 to translocate unphosphorylated PRC1 to the plus ends of interdigitating MTs where PRC1 could bundle the interdigitating MTs to organize formation of the midzone, which is necessary for cytokinesis. To test the model, we examined the interactions of Kif4 and PRC1, especially the interactions of Kif4 with phosphorylated PRC1, during different stages of mitosis.

HeLa cells were synchronized at the G₂/M boundary by a thymidine/nocodazole block and then released into mitosis. Cells at different stages of mitosis were lysed, and whole-cell lysates were subjected to immunoblotting analysis with anti-Kif4, anti-PRC1, anti-Cdk-phosphorylated PRC1 (T481), or anti-cyclin B1-specific antibodies. Expression levels of Kif4 and PRC1 remained constant during mitosis. However, the levels of Cdk-phosphorylated PRC1 fluctuated during mitosis. The levels of Cdk-phosphorylated PRC1 were high in early mitosis but decreased dramatically during the metaphase-to-anaphase transition, coincident with the decrease of the expression levels of cyclin B1 (Fig. 4A).

Consistently, similar amounts of PRC1 from different stages of mitosis were immunoprecipitated by anti-PRC1 antibodies (Fig. 4B). High levels of phosphorylated PRC1 were detected in the early mitosis anti-PRC1 immunoprecipitates. As cells entered anaphase, the levels of phosphorylated PRC1 in anti-PRC1 immunoprecipitates were dramatically decreased (Fig. 4B). In contrast, very low levels of Kif4 were detected in the anti-PRC1 immunoprecipitates from early mitosis, but high levels of Kif4 proteins were detected in the anti-PRC1 immunoprecipitates from late mitosis (Fig. 4B). Reciprocal immunoprecipitation experiments also were performed with anti-Kif4 antibodies. As shown in Fig. 4B, similar amounts of Kif4 from different stages of mitosis were immunoprecipitated by anti-Kif4 antibodies. Very low levels of PRC1 were detected in the anti-Kif4 immunoprecipitates from early mitosis but high levels of PRC1 proteins were detected in the anti-Kif4 immunoprecipitates from late mitosis. No Cdk-phosphorylated PRC1 was detected in any anti-Kif4 immunoprecipitates. These results suggest that Kif4 preferentially interacts with unphosphorylated PRC1.

To confirm that Kif4 specifically interacts with unphosphorylated PRC1, but not phosphorylated PRC1, we performed in vitro pull-down analyses. Bacterially expressed GST-PRC1 or GST alone was bound to glutathione-agarose beads and incubated with or without baculovirus-expressed purified Cdc2/cyclin B1 in the presence of ATP. As shown in Fig. 4C, Cdc2/cyclin B1 could efficiently phosphorylate GST-PRC1 in vitro as determined by immunoblotting analysis using antiphosphorylated PRC1(T481)-specific antibodies. GST-phosphorylated PRC1, GST-PRC1, or GST controls then were incubated with early mitotic HeLa cell lysates. GST-PRC1, but not Cdc2/cyclin B1-phosphorylated GST-PRC1 or GST controls, was able to pull down endogenous Kif4 (Fig. 4C). These results indicated that unphosphorylated PRC1, but not phosphorylated PRC1, specifically interacts with Kif4. Taken together, our results indicate that PRC1 is phosphorylated in early mitosis and phosphorylated PRC1 does not associate with Kif4. During the metaphase-to-anaphase transition, PRC1 is dephosphorylated and interacts with Kif4.

We conclude that Kif4 is the motor protein that translocates PRC1 along mitotic spindle to the plus ends of interdigitating MTs during the metaphase-to-anaphase transition and that Cdk phosphorylation of PRC1 controls the timing of the translocation of PRC1 by Kif4. Several lines of evidence support our conclusion. First, previous studies showed that PRC1 is a MT binding and bundling Cdk substrate protein that plays an essential role in regulating the spindle midzone formation and cytokinesis in human cells (4, 16). PRC1 interacts with MTs directly and bundles antiparallel MTs into MT bundles in vitro. Because PRC1 translocates from spindles to the spindle midzone/midbody, it had been proposed that PRC1 bundles the spindle plus ends of interdigitating MTs to form the midzone that is necessary for cytokinesis. Our immunofluorescence and time-lapse video studies, in which we monitored the movement of PRC1 on the mitotic spindle in live cells, demonstrated that depletion of Kif4 expression by Kif4 esiRNA impaired the translocation of PRC1 on the spindle during the metaphase-to-anaphase transition, inhibited midzone formation and caused cytokinetic defects. Furthermore, we have shown that ectopic expression of Kif4, but not its motor dead Kif4md mutant, restores the translocation of PRC1 along the spindle, reestablishes the midzone, and rescues the cytokinetic defects in Kif4 esiRNA-treated cells. These results strongly indicate that Kif4 is the motor protein required for PRC1 translocation on the spindle during mitosis and that Kif4-dependent translocation allows PRC1 to bundle interdigitating MTs and form the midzone that is necessary for cytokinesis.

Second, based on the position of the motor domain, Kif4 is categorized as a N-class kinesin (21). The motor domains of N-class kinesins use ATP to fuel the movement along MTs anterogradely toward the plus ends. Besides the motor domains, N-class kinesins have different stalk and tail domains at the C terminus that mediate oligomerization, regulation of motor activity, and interactions with their specific cargos (8). We show that Kif4 interacts with PRC1 in vivo and that the interaction occurs through the Kif4 C-terminal stalk plus tail domain. As the translocation of PRC1 on the spindle is to the plus ends of the interdigitating MTs and dependent on Kif4, our results indicate that PRC1 is a critical downstream cargo of Kif4 on the spindle during mitosis. While this article was in preparation, Kurasawa et al. (22) reported the interaction between Kif4 and PRC1 although the study did not show that Kif4 is required for translocation of PRC1 on the spindle during mitosis.

Third, cytokinesis is a carefully regulated process that must coordinate precisely with other events of the cell cycle. The best-known cell cycle regulators are Cdks. Previous studies showed that inactivation of Cdk activity through destruction of mitotic cyclin B is a critical event for the metaphase-to-anaphase transition. For example, cells expressing a stable nondegradable cyclin B mutant arrest in anaphase and fail to undergo cytokinesis (2). Thus, coupling cytokinesis to Cdk inactivation is a key mechanism for cells to ensure that division is not initiated until chromosomes have been separated. We find that the interaction of Kif4 and PRC1 depends on Cdk phosphorylation of PRC1. We show that PRC1 is phosphorylated in early mitosis and phosphorylated PRC1 does not associate with Kif4. During the metaphase-to-anaphase transition, PRC1 is dephosphorylated and interacts with Kif4. These results, together with the fact that Kif4 is required for the translocation of PRC1 to the plus ends of interdigitating MTs during the metaphase-to-anaphase transition, indicate that Cdk phosphorylation of PRC1 controls the timing of the translocation of PRC1 by Kif4.

Taken together, we present the following model for the cell cycle-dependent translocation of PRC1 to the plus ends of interdigitating MTs by Kif4 during the metaphase-to-anaphase transition (Fig. 4D). In early mitosis, PRC1 is phosphorylated by Cdks (mainly Cdc2/cyclin B1), and Cdk phosphorylation of PRC1 prevents PRC1 interaction with Kif4 although both PRC1 and Kif4 associate with the mitotic spindle. During the metaphase-to-anaphase transition, the anaphase promoting complex triggers proteolysis of cyclin B, resulting in the elimination of all cellular Cdk activity. In the absence of Cdk activity, PRC1 is dephosphorylated by an active mitotic phosphatase (e.g., Cdc14A or PP1γ). Dephosphorylation of PRC1 promotes interaction of Kif4 with PRC1. Kif4 translocates PRC1 to the plus ends of interdigitating MTs, allowing PRC1 to bundle the interdigitating MTs and organize formation of the midzone during anaphase that is necessary for cytokinesis.