The Forkhead-associated Domain of Ki-67 Antigen Interacts with the Novel Kinesin-like Protein Hklp2*

The Ki-67 antigen (pKi-67) is widely used as a cell proliferation marker protein. Its actual role in the cell cycle progression, however, is presently unclear. Using a two-hybrid screening in yeast, a novel protein, termed Hklp2 (human kinesin-like protein 2), was identified and shown to interact with the forkhead-associated (FHA) domain of pKi-67. Hklp2 has 1388 amino acids and shows a striking similarity (a 53% identity in amino acids) to Xklp2, a plus-end directed kinesin-like motor found in Xenopus . The interaction domain of Hklp2 was mapped to the portion that comprised residues 1017– 1237 and that was phosphorylated in vitro by incubating with mitotic but not interphasic HeLa cell extracts. That the interaction was striking in the mitotic extract was also verified. In addition, immunofluorescence using specific antibodies revealed an association between pKi-67 and Hklp2 at the periphery of mitotic chromosomes, largely in close proximity to the centromeres. These findings suggest that pKi-67 is involved in the progression of mitosis via its interaction with Hklp2. The Ki-67 antigen (pKi-67) was originally identified as an antigen for a monoclonal antibody raised against the nuclear extract from a Hodgkin’s lymphoma derived cell line and is characterized as a class

The Ki-67 antigen (pKi-67) was originally identified as an antigen for a monoclonal antibody raised against the nuclear extract from a Hodgkin's lymphoma derived cell line and is characterized as a class of proteins that are localized around mitotic chromosomes (1). Because of this, it is assumed that pKi-67 is involved in mitotic chromosome organization. Moreover, pKi-67 is a convenient cell proliferation marker, because its expression is restricted to growing cells (2). Although the recent identification and characterization of a marsupial counterpart of pKi-67, which is referred to as chmadrin, suggests a role of pKi-67 in the organization of higher order chromatin structure (3), the actual function of the molecule in the cell cycle progression is presently unclear.
To better understand the role of pKi-67 in the cell cycle, a two-hybrid screening from a HeLa cDNA library was carried out using the N-terminal portion of pKi-67 as the bait. This portion is well conserved between human pKi-67 and chmadrin (the putative ortholog found in marsupial cells) and contains the FHA 1 domain. It was originally reported that the FHA domain constituted a conserved region that is found in a subset of forkhead-type transcription factors (4). The sequence profile was also found in variety of proteins with diverse functions (transcription, DNA repair, cell cycle progression, and so on) (4). In several instances, the FHA domain was shown to preferentially recognize partner proteins when they are phosphorylated (5,6), and, thus, it is currently thought to be a general phosphopeptide recognition motif (7). A search for the interaction partner(s) of the FHA domain of pKi-67, which might be phosphorylated, is an intriguing issue because the interaction would be a significant component of the regulation of the cell cycle progression.
Here, we report the interaction between the FHA domain of pKi-67 and a novel kinesin-like protein (herein referred to as Hklp2), which is a human homolog of Xklp2 (Xenopus kinesinlike protein 2) (8). Xklp2 is a plus-end directed kinesin-like motor that has been reported to be required in centrosome separation and the maintenance of spindle bipolarity during mitosis in Xenopus egg extracts (8). Because pKi-67 is localized around mitotic chromosomes (9), Hklp2 might also function at a point near mitotic chromosomes, via its interaction with pKi-67. We also characterized the mode of interaction between the FHA domain of pKi-67 and Hklp2. As described previously for the FHA domains of other proteins (5-7), the FHA domain of pKi-67 preferentially was likely to recognize the phosphorylated form of the partner protein, Hklp2. Our findings provide a possible molecular link between pKi-67 and the cell cycle and further provide a new clue to understand dynamism of mitotic chromosomes. This study will also be of use for understanding the mode of interaction between the FHA domain and specific partner phosphoproteins.

EXPERIMENTAL PROCEDURES
Two-hybrid Interaction Screening-A yeast two-hybrid screen was performed essentially as described previously (10). Briefly, yeast strain Y190, which expresses the fusion protein of the GAL4 DNA-binding domain and Ki-FHA (residues 1-99 of pKi-67) was transformed with a HeLa cDNA library constructed in a plasmid pACT (purchased from CLONTECH) and plated on selective medium supplemented with 25 mM 3-aminotriazole. Bait loss was carried out by successive inoculation with a medium devoid of leucine.
Molecular Cloning of Hklp2-One of the clones obtained by the two-hybrid screen (clone 3␣) showed a striking similarity to Xklp2 (8). The clone spanned the region that corresponds to approximately onethird from the C-terminal end of the coding region of Hklp2. Using the sequence of Xklp2 as a query, the human expressed sequence tag data base was searched with the result that clone AA332437 was likely to encode a portion of the more upstream region of Hklp2. The 300-base * This work was supported by Grant-in-Aid for Scientific Research on Priority Areas (B) 11237202 and Grant-in-Aid for COE Research 07CE2006 from the Japanese Ministry of Education, Science, Sports and Culture and by funds from the Mitsubishi Foundation and the Human Frontiers Science Program. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB035898.
Immunofluorescence-Immunofluorescence was performed essentially as described previously (3), except that cold (Ϫ20°C) methanol was used for the fixative. Anti-Hklp2 antibodies were used at 2 g/ml. The anti-␣-tubulin monoclonal antibody (Sigma) was used at 1:200. MIB-1 (Immunotech), a monoclonal antibody against pKi-67, was used at 1:100. Human autoimmune antibodies against centromere proteins (12) were generous gifts from Dr. Hiroshi Masumoto (Nagoya University) and were used at 1:160. In the experiments presented in Fig. 3C, nocodazole was added to a final concentration of 0.2 g/ml for 3 h before fixation. Deconvolution microscopy was carried out with a Delta Vision System (Applied Precision, Inc.), and the acquired digital images were processed by using Adobe Photoshop software.
In Vitro Kinase Assay-HeLa cells, at a 70% confluence, were incubated with 50 ng/ml of nocodazole for 14 h, and mitotic cells were collected by mechanical shake-off. To prepare cell extracts, asynchronous or mitotic HeLa cells were washed once with phosphate-buffered saline (137 mM NaCl, 3 mM KCl, 8 mM Na 2 HPO 4 , and 1 mM KH 2 PO 4 ) containing 0.2 mM phenylmethylsulfonyl fluoride before resuspending in kinase buffer (50 mM HEPES-KOH, pH 7.5, 1% Nonidet P-40, 100 mM NaCl, 25 mM NaF, 25 mM ␤-glycerophosphate, and 10 g/ml of leupeptin, aprotinin, and pepstatin) to give 2 ϫ 10 4 cells/l. Cell suspensions were left on ice for 20 min and passed 10 times through a 27-G needle, followed by centrifugation for 10 min at 10,000 ϫ g at 4°C. Supernatants were stored at Ϫ80°C until used for kinase assay. To examine the activity of the cell extracts with respect to the phosphorylation of Hklp2, purified His-Hklp2-tail (0.5 mg/ml) was incubated with the extracts in the presence of 10 M ATP and 4 Ci of [␥-32 P]ATP for 15 min at 30°C. As control substrates, histone H1 and GST (0.5 mg/ml) were used. Reactions were stopped by the addition of sample buffer containing SDS. Samples were then heated for 5 min at 60°C before analysis by SDS-polyacrylamide gel electrophoresis and autoradiography.
Preparation of Cell Extracts and Pull-down Assay-HeLa cells (asynchronous or mitotic-arrested) were washed once with phosphate-buffered saline containing 0.2 mM phenylmethylsulfonyl fluoride and resuspended in EB150 (50 mM HEPES-KOH, pH 7.5, 150 mM NaCl, 2 mM MgCl 2 , 5 mM EGTA, 1 mM dithiothreitol, 0.5% Nonidet P-40, 10% glycerol, 0.2 mM phenylmethylsulfonyl fluoride, 25 mM NaF, 25 mM ␤-glycerophosphate, and 10 g/ml of leupeptin, aprotinin, and pepstatin) to give 2 ϫ 10 4 cells/l. The cell suspensions were left on ice for 20 min, passed through a 27-G needle 10 times, and centrifuged at A, summary of the cDNA cloning of Hklp2. Clone 3␣ was obtained via twohybrid screening of a HeLa cDNA library using the FHA domain of pKi-67 as the bait. The expressed sequence tag clone AA332437 was found by a data base search using the amino acid sequence of Xklp2 as a query. Clone 18-1 was obtained via the screening of a HeLa cDNA library using the probes depicted (probes 1 and 2). The most 5Ј region was obtained by 5Ј rapid amplification of cDNA ends. B, schematic alignment of Hklp2 and Xklp2. The kinesin motor domains are represented by black boxes. Arrows indicate the regions corresponding to the leucine zipper motif. The tail region, which was used for immunization, is shaded. Hklp2 and Xklp2 show 53.2 and 87.5% identity in their full-length and kinesin motor domains, respectively. C, predicted amino acid sequence of Hklp2. Underlining, kinesin motor domain; bold type, leucine zipper motif.
14,000 ϫ g for 10 min at 4°C. The supernatants were used as EB150 extracts. EB150 extracts were incubated with GST or GST-Ki-FHA coupled to glutathione-Sepharose beads (GS-4B; Amersham Pharmacia Biotech) for 1 h at 4°C, after which the beads were washed extensively with EB150. Associated proteins were boiled out in sample buffer and separated in 2-15% gradient gel (Multigel 2/15, Daiichi Pure Chemicals). Hklp2 was detected by Western blotting using anti-Hklp2 antibodies at 0.5 g/ml.

Identification of Hklp2 as a Molecule That
Interacts with the FHA Domain of pKi-67-To identify cellular proteins that might interact with the FHA domain of pKi-67, a yeast twohybrid screen of a HeLa cDNA library was carried out using residues 1-99 of human pKi-67 as bait. From the ϳ5 ϫ 10 6 cDNA clones screened, 10 positive clones were obtained. Sequence analyses revealed that these clones could be placed in one of two possible classes. One class (containing five clones), which corresponded to a novel RNA-binding protein, which will be described elsewhere. 2 Here, we focus on the other class (containing two clones). A full-length sequence was first obtained using the strategy summarized in Fig. 1A. The predicted protein showed striking similarity to Xklp2 (53% identity in amino acids over their entire length) and was therefore termed Hklp2 (human kinesin-like protein 2). Hklp2 and Xklp2 share some structural features; kinesin motor domain at the N-terminal portion, coiled-coil structure at the rest of the motor domain, and leucine zipper motif at the C-terminal portion (Fig. 1B).
Identification of the Binding Domain of Hklp2 with pKi-67-To map the interaction domain of Hklp2, a deletion analysis in the yeast two-hybrid system was carried out. For this analysis, deletion derivatives of Hklp2 (depicted in Fig. 2A) were fused to the GAL4 activation domain and transformed into yeast strain Y190 together with a GAL4 DNA-binding domain fusion of the residues 1-99 of pKi-67 (Ki-FHA). This experiment indicated that residues 1017-1238 of Hklp2 were sufficient for interaction with Ki-FHA. We were not able to narrow the interaction domain further, which suggests that the entire sequence was essential for the interaction. When secondary structural prediction methods (13) were applied to this region, seven ␣-helices, inserted between turns, were predicted (Fig. 2B). In the first, second, and seventh helices, leucine or isoleucine residues were predicted to be clustered and to form a hydrophobic surface on each helix (Fig. 2C). A protein conformation properly mediated by these leucine zipper-like structures might be significant, in terms of its interaction with Ki-FHA.
To confirm that the interaction observed above is not merely specific to yeast cells, we performed an in vitro binding assay using purified recombinant proteins. GST fused to Ki-FHA (GST-Ki-FHA) was expressed in Escherichia coli, immobilized on glutathione-Sepharose beads, and incubated with purified histidine-tagged Hklp2 (1017-1238). The matrix-attached Hklp2 (1017-1238) was detected by Western blotting. Hklp2 (1017-1238) was retained on beads coupled to GST-Ki-FHA but not on control beads (Fig. 2D). These data are consistent with Ki-FHA interacting with Hklp2, both in yeast and in vitro. Subcellular Distribution of Hklp2 in Relation with Mitotic Spindles, Chromosomes, and pKi-67-As an approach to understanding the biological significance of the interaction between the FHA domain of pKi-67 and Hklp2, the subcellular localization of Hklp2 was examined using polyclonal antibodies raised against its C-terminal portion (residues 1132-1338). The antibodies specifically recognized a 160-kDa protein from the total HeLa extract by immunoblotting (Fig. 3A). In the interphase, Hklp2 was detected on centrosomes as described for Xklp2 (data not shown). From prometaphase to metaphase, Hklp2 was detected primarily on the mitotic spindles emanating from the spindle pole regions (Fig. 3B). Hklp2 was not uniformly located over the entire length of the spindles. A certain population of Hklp2 in metaphase was localized in the equatorial region, distinct from the microtubules (Fig. 3B, panels d and e). It is likely that the population is integrated in, or closely associated with, the chromatin structure (Fig. 3B, panels d-f). When the mitotic spindles were disrupted by nocodazole before fixing the cells, the chromatin-integrating population of Hklp2 was detected in the form of dots on mitotic chromosomes (Fig. 3C). These dots were not co-localized with but were close to the centromeres (Fig. 3C). In anaphase, Hklp2 was also detected at the equatorial region of the spindle in a punctate pattern (Fig. 3B, panels g and h), raising the possibility that Hklp2 may be involved in the sliding of antiparallel microtubules in anaphase B. Considering the fact that pKi-67 is localized around the mitotic chromosomes (9), pKi-67 and Hklp2 could, in theory, be co-localized at the surface of the chromosomes where mitotic spindles are pointed to. In fact, it is likely that a certain population of Hklp2 is colocalized with pKi-67 on mitotic chromosomes (Fig. 3D, yellow).
Hklp2 Interacts with pKi-67 Preferentially in Mitosis-A previously published report (7) concluded that the FHA domain interacts preferentially with a phosphorylated peptide. We examined the issue of whether this was also the case for the FHA domain of pKi-67. We first asked whether the partner protein, Hklp2, was modified differently through the cell cycle. When the electrophoretic mobilities of Hklp2 derived from differently synchronized HeLa cells were compared, the mitotic Hklp2 showed a slower mobility than the interphasic Hklp2 (Fig. 4A). Whether this was due to mitosis-specific phosphorylation was not determined in this study.
We next examined the issue of whether Hklp2 is phosphorylated within the interaction domain to pKi-67 (residues 1017-1238). When histidine-tagged Hklp2 (1017-1238) was purified and incubated with differently synchronized HeLa extracts in the presence of [␥-32 P]ATP, the protein was labeled only when the mitotic HeLa extract was used, suggesting that Hklp2 is phosphorylated within this region in the mitotic phase. The possible presence of other phosphorylation sites in the Hklp2 sequence, however, cannot be excluded. To test the possibility that the interaction between pKi-67 and Hklp2 is regulated through the cell cycle, cell extracts were prepared from asynchronized or mitotic HeLa cells. These extracts were then incubated with a GST fusion of Ki-FHA immobilized on Sepharose beads. After incubation, the bound proteins were analyzed by Western blotting, and Hklp2 was detected with anti-Hklp2 antibodies. The slowly migrating form of Hklp2 contained in the mitotic extract was pulled down more efficiently than the quickly migrating form of Hklp2. DISCUSSION A novel kinesin-like protein, referred to as Hklp2, has been identified and shown to interact with the FHA domain of pKi-67 (Ki-FHA). The interaction occurred efficiently in the mitotic phase, during which Hklp2 was likely to be phosphorylated. These findings suggest that the Ki-67 antigen might play an crucial role in the progression of mitosis, thus providing a clue to our understanding of the strict correlation between the expression of pKi-67 and cell proliferation. Further analysis of the mode of interaction between Ki-FHA and Hklp2, including an investigation of whether phosphorylation itself is essential, will obviously be required for a more complete understanding of the biological significance of the interaction. Because the phosphoproteins that are recognized by the FHA domain have, thus far, been specified only in limited cases (5,6), these data will be also helpful in exploring the general mechanism of phosphorylation-dependent protein recognition mediated by the FHA domain.
Given the mitotic localization of pKi-67, it is possible that it co-operates with Hklp2 on the surface of mitotic chromosomes to allow mitosis to proceed. Given the fact that Hklp2 is a plus end-directed motor by analogy to XKlp2 (8), Hklp2 could generate "polar ejection force" (14) by acting along chromosome arms via an interaction with pKi-67. However, this notion is apparently inconsistent, given the observed localization of Hklp2 (Fig. 3C), in which a certain population of this molecule is associated with chromosomes in a punctate pattern but are not distributed throughout the entire chromosome arms. It is possible that we overlooked detecting the localization of Hklp2 on the chromosome arms because of its low levels in that area. Alternatively, our protocol (nocodazole treatment and methanol fixation) might disrupt the actual localization. In any case, it would be intriguing to examine the issue of whether the working point of Hklp2 exists on the mitotic chromosome arms.
In an early study, we identified a molecule, termed chmadrin, from a marsupial cell line (3). Chmadrin shows certain similarities to pKi-67 in its primary structure and cellular localization and thus is considered to be a marsupial counterpart of pKi-67. The C-terminal portion of chmadrin has a novel domain that is characterized by repetitive occurrences of LR (leucine and arginine) pairs with irregular spacing. The domain, called the LR domain, gave rise to severe heterochromatin formation when overexpressed (3) and, as a result, can be implicated in the organization of higher order chromatin structure. The C-terminal portion of human pKi-67 also possesses the LR domain and, in fact, has the activity to drive heterochromatin formation in a manner similar to that of chmadrin. 3 Thus, pKi-67 appears to play a role in chromatin condensation in its C-terminal region (LR domain). If pKi-67 serves as an adapter between chromosomes and the mitotic apparatus in its N-terminal region (FHA domain), this protein may be a multifunctional protein that acts throughout the cell cycle.