Cdc2-mediated Phosphorylation of CLIP-170 Is Essential for Its Inhibition of Centrosome Reduplication*

CLIP-170, the founding member of microtubule “plus ends tracking” proteins, is involved in many critical microtubule-related functions, including recruitment of dynactin to the microtubule plus ends and formation of kinetochore-microtubule attachments during metaphase. Although it has been reported that CLIP-170 is a phosphoprotein, neither have individual phosphorylation sites been identified nor have the associated kinases been extensively studied. Herein, we identify Cdc2 as a kinase that phosphorylates CLIP-170. We show that Cdc2 interacts with CLIP-170 mediating its phosphorylation on Thr287 in vivo. Significantly, expression of CLIP-170 with a threonine 287 to alanine substitution (T287A) results in its mislocalization, accumulation of Plk1 and cyclin B, and block of the G2/M transition. Finally, we found that depletion of CLIP-170 leads to centrosome reduplication and that Cdc2 phosphorylation of CLIP-170 is required for the process. These results demonstrate that Cdc2-mediated phosphorylation of CLIP-170 is essential for the normal function of this protein during cell cycle progression.

Microtubule dynamics consist of alternating phases of growth and shortening, a pattern of behavior known as dynamic instability (1). This process is tightly regulated by a group of proteins that bind specifically to the plus ends of the growing microtubules (2). Cytoplasmic linker protein (CLIP) 3 -170, the founding member of the microtubule plus end family (3), is composed of three separate regions: N terminus, central coiled-coil region, and C terminus. In addition to two conserved cytoskeleton-associated protein glycine-rich (CAP-Gly) domains, the N terminus has three serine-rich regions. The N-terminal domain plays an essential role in microtubule targeting (4), the long central coiled-coil domain is responsible for dimerization of the protein, and the C-terminal region, which contains two zinc-finger domains interferes with microtubule binding by interacting with the N terminus (5). Experiments in a variety of organisms have demonstrated that CLIP-170 plays an important role in microtubule dynamics (6,7). In addition to its positive role in regulating microtubule growth in both yeast and humans (8,9), CLIP-170 is involved in recruitment of dynactin to the microtubule plus ends and in linking microtubules to the cortex through Cdc42 and IQGAP (10,11). The role of CLIP-170 during mitosis was recently examined by loss-offunction approaches. It was shown that CLIP-170 localizes to unattached kinetochores in prometaphase and that such localization is essential for the formation of kinetochore-microtubule attachments (12,13).
It was previously reported that CLIP-170 is a phosphoprotein and that overall phosphorylation of CLIP-170 affects its microtubule binding ability (14). More recently, metabolic labeling experiments indicated that CLIP-170 is phosphorylated at multiple sites (15). However, individual phosphorylation sites have not been identified. Moreover, the FKBP12-rapamycin-associated protein (FRAP) is the only kinase identified to date for CLIP-170 (15). Therefore, to fully understand the regulation of CLIP-170, it is important to identify individual phosphorylation sites and the responsible kinases. In this communication, we describe a novel kinase/substrate partnership between Cdc2 and CLIP-170. We provide evidence that Cdc2 phosphorylates CLIP-170 at Thr 287 , and the Cdc2-mediated phosphorylation of CLIP-170 is essential for its localization at microtubule plus ends in the G2 phase and the G2/M transition.

EXPERIMENTAL PROCEDURES
Cell Culture, Synchronization, and DNA Transfections-HEK293T and U2OS cells were cultured in Dulbecco modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum, 100 units ml Ϫ1 penicillin, and 100 units ml Ϫ1 streptomycin at 37°C in 8% CO 2 . For synchronization, U2OS cells were treated with 2.5 mM thymidine for 16 h, released for 8 h, then blocked with 2.5 mM thymidine for another 16 h to arrest at the G1/S boundary. After release into fresh medium for different times, cells were stained with various antibodies. DNA transfection was performed using MegaTran 1.0 (ORI-GENE) with the protocol as described by the manufacturer.
Recombinant Protein Purification-Various domains of CLIP-170 were subcloned into pGEX-KG vector and expressed in Escherichia coli. Expression was induced by 0.5 mM isopropyl ␤-D-1-thiogalactopyranoside at 37°C for 3 h after the cell density had reached 0.6 at 600 nm. Recombinant GST fusion proteins were affinity-purified by incubation with glutathione-agarose beads (Sigma, catalogue number G4510), followed by extensive washes with STE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 150 mM NaCl) and elution with glutathione elution buffer (40 mM glutathione, 50 mM Tris-HCl, 10 mM dithiothreitol, 200 mM NaCl, pH 8.0).
Metabolic Labeling and Phosphopeptide Mapping-HEK293T cells were transfected with GFP-CLIP-170 (WT or T287A) and treated with 100 ng ml Ϫ1 of nocodazole for 10 h to arrest cells in mitosis. Cells were labeled for 3 h with [ 32 P]orthophosphate (ICN) at 1 mCi ml Ϫ1 in phosphate-free DMEM in the absence or presence of roscovitine, followed by incubation with 1.0 M okadaic acid for 30 min. Total cell lysates were prepared and incubated with an anti-GFP antibody overnight at 4°C, followed by incubation with protein A PLUS-agarose beads for an additional 3 h. The beads were washed at least five times with TBSN buffer containing 500 mM NaCl, and at least 10 times with TBSN buffer containing 150 mM NaCl. Samples were resolved by SDS-PAGE and stained with Coomassie Brilliant Blue, followed by autoradiography. For phosphopeptide mapping, phosphoproteins were immunoprecipitated with anti-GFP antibodies, resolved by SDS-PAGE, and transferred to nitrocellulose filters. Labeled GFP-CLIP-170 was detected by autoradiography, excised, exhaustively digested with trypsin, and separated by 40% alkaline acrylamide gel (16).

RESULTS AND DISCUSSION
Physical Interaction between Cdc2 and CLIP-170-Although CLIP-170 was reported to be a phosphoprotein almost 20 years ago (14), kinases responsible for the multiple phosphorylations are largely unknown (15). Recently, it was shown that knockdown of CLIP-170 with RNAi results in an almost complete block of cells in mitosis (13). Considering the essential role of Cdc2 in mitosis, we explored the possibility of a potential direct connection between CLIP-170 and Cdc2.
To test whether CLIP-170 is a binding partner of Cdc2, a co-immunoprecipitation experiment was performed. Briefly, HEK293T cells were transfected with GFP-CLIP-170, treated with or without nocodazole, harvested, and immunoprecipitated with a Cdc2 antibody, followed by anti-GFP Western blot. Nocodazole is a drug to block cells at mitosis through microtubule depolymerization (17). As shown in Fig. 1A, overexpressed GFP-CLIP-170 protein was specifically pulled down by the Cdc2 antibody, indicating an association between CLIP-170 and Cdc2. The increased level of GFP-CLIP-170 that co-immunoprecipitates with Cdc2 in nocodazole-treated cells suggests a substrate/kinase partnership in mitosis. Furthermore, we also examined the interaction between endogenous CLIP-170 and Cdc2. As shown in Fig. 1B, CLIP-170 co-immunoprecipitates with a Cdc2 antibody, confirming an interaction between these two proteins.
Cdc2 Phosphorylates CLIP-170 at Thr 287 -Considering the in vivo association between CLIP-170 and Cdc2, we asked whether Cdc2 directly phosphorylates CLIP-170. Commercial Cdc2/cyclin B was incubated with four purified GST-CLIP-170 fragments (amino acids 1-365, 355-734, 716 -1101, and 1089 -1392) in the presence of [␥-32 P]ATP. The reaction mixtures were resolved by SDS-PAGE, followed by autoradiography. As shown in Fig. 2A, only the fragment containing amino acids 1-365 was phosphorylated by Cdc2, indicating that this N-terminal region of CLIP-170 is a Cdc2 substrate in vitro. To map the potential phosphorylation site(s) for Cdc2, we generated a series of serine to alanine mutants in CLIP-170 spanning amino acids 1-365. Compared with the phosphorylation levels of wild-type CLIP-170 and other mutants, the phosphorylation level of the T287A mutant was completely abolished, indicating that Thr 287 is the major phosphorylation site for Cdc2 in vitro (Fig. 2B). The consensus phosphorylation motif for Cdc2 is (S/T)PX(R/K), in which the Pro at ϩ1 position is absolutely required, and a basic residue at the ϩ3 position is preferred but not essential (18). The sequence context of Thr 287 ( 287 TPAK) matches this consensus phosphorylation motif (Fig. 2F). To confirm specificity of the kinase assay, we repeated the experiment described above in the absence or presence of roscovitine, a Cdc2 inhibitor (19). Addition of roscovitine strongly inhibited the phosphorylation of CLIP-170-WT by Cdc2, supporting the specificity of the kinase assay (Fig. 2C).
To examine whether Cdc2-dependent phosphorylation of CLIP-170 at Thr 287 also occurs in vivo, metabolic labeling experiments were performed. In brief, HEK293T cells were transfected with GFP-CLIP-170-WT (wild type) or T287A. At 16-h post-transfection, roscovitine was added to inhibit Cdc2 activity. After an additional overnight incubation, cells were treated with nocodazole for 10 h, and metabolically labeled with [ 32 P]orthophosphate for 3.5 h. To inhibit possible dephosphorylation of CLIP-170, cells were further treated with okadaic acid for 0.5 h before harvesting. GFP-CLIP-170 protein was immunoprecipitated with anti-GFP antibody, resolved by SDS-PAGE, and subjected to autoradiography. As indicated in Fig. 2D, the total phosphorylation level of GFP-CLIP-170 was not significantly reduced in cells expressing the T287A mutant, suggesting that Thr 287 is only one of multiple sites phosphorylated in vivo. Moreover, treatment of cells with roscovitine did not dramatically reduce the phosphorylation level of CLIP-170-WT, indicating that Cdc2 is only one of multiple enzymes responsible for phosphorylation of CLIP-170 in vivo (Fig. 2D). These results are consistent with previous analyses that suggest that CLIP-170 is phosphorylated at multiple sites by multiple kinases  OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 28777 (15). To further analyze phosphorylation of CLIP-170-T287 in vivo, samples prepared as in Fig. 2D were excised from the filter, digested with trypsin, and resolved on a 40% alkaline gel. We observed disappearance of one specific band (indicated by the arrow, Fig. 2E) in the lanes loaded with the T287A mutant and with roscovitine-treated CLIP-170-WT, supporting the notion that Cdc2 does phosphorylate CLIP-170-Thr 287 in vivo. Quantification of the relative 32 P intensity of the peptide containing Thr 287 further confirmed the conclusion. Finally, in a separate study using mass spectrometry to analyze CLIP-170 phosphorylation, Thr 287 is also identified as one of multiple sites phosphorylated in vivo. A detailed description of additional phosphorylation sites and additional kinases using this approach will be reported later. 4 In sum, the data show that Cdc2 phosphorylates CLIP-170 at Thr 287 in vivo.

CLIP-170 Is Phosphorylated by Cdc2
Cdc2-mediated Phosphorylation of CLIP-170 at Thr 287 Regulates Its Binding Pattern with Tubulin-To explore the role of Cdc2-mediated phosphorylation of CLIP-170, U2OS cell lines were generated that stably express rat GFP-CLIP-170 (WT, T287A, or T287D). As shown in Fig. 3A, the expression levels of these GFP-CLIP-170 constructs are similar in the stably transfected cells.
Because CLIP-170 is a plus end microtubule-binding protein and the Cdc2 phosphorylation site is located at the microtubule binding domain (Fig. 2F), we asked whether the Cdc2-mediated phosphorylation of CLIP-170 regulates its microtubule binding ability. To address this question, U2OS cells stably expressing GFP-CLIP-170 (WT, T287A, or T287D) were analyzed by immunofluorescence microscopy using ␣-tubulin antibody. As indicated in Fig. 3, B and C, both GFP-CLIP-170-WT and the GFP-CLIP-T287D mutants localized to plus ends of microtubules. In striking contrast, 75% of cells expressing GFP-CLIP-170-T287A showed a diffuse pattern of GFP signal, rather than localization at microtubule plus ends (see the enlarged images of the far-right column in Fig. 3B). Because Cdc2 is critical for the later stages of the cell cycle, U2OS cells stably expressing GFP-CLIP-170-T287A were synchronized by the double thymidine block protocol in the G1/S boundary, released for different times, and stained with ␣-tubulin antibody (Fig. 3D). In late G1/early S phase (0 h of release) when Cdc2 activity is low, most GFP-CLIP-170-T287A was observed at plus ends of microtubules. In late S phase (6 h of release), G2 phase (8 h of release), and mitosis (10 h of release), most GFP-CLIP-T287A had lost its ability to localize at the plus ends of microtubules; instead, it localized throughout the cytoplasm. In the next G1 phase (12 h of release), a significant portion of GFP-CLIP-T287A relocalized back to plus ends of microtubules. All together, we conclude that Cdc2-associated phosphorylation of CLIP-170 is essential for its correct localization at the plus ends of microtubules at the later stages of the cell cycle. Neither the ␣-tubulin nor the ␥-tubulin staining pattern was affected in cells expressing different forms of GFP-CLIP-170 (WT, T287A, or T287D) (Fig. 3, B and E).
Cdc2-mediated Phosphorylation of CLIP-170 Is Essential for the G2/M Transition-To further analyze the functional significance of Cdc2-associated phosphorylation of CLIP-170, we compared cell cycle progression of U2OS cells, stably expressing different forms of GFP-CLIP-170 (WT, T287A, or T287D) in the absence of endogenous protein. Vectorbased RNAi was used to deplete CLIP-170. Specifically, U2OS cells were co-transfected with pBS/U6-CLIP-170 and pBabe-puro, a vector expressing a puromycin resistance gene. After puromycin selection, cells were harvested and analyzed by anti-CLIP-170 Western blots. As indicated in Fig. 4A, this approach can efficiently deplete CLIP-170. Next, U2OS cells stably expressing rat GFP-CLIP-170 (WT, T287A, or T287D) were depleted of endogenous CLIP-170 using vector-based RNAi as in Fig. 4A, treated with or without nocodazole, and harvested for Western blots using various antibodies as indicated (Fig. 4, B-D). Because of the high specificity of RNAi, rat GFP-CLIP-170 was not targeted by RNAi designed to deplete the human sequence.  4, 5, and 6, Fig. 4B). Addition of nocodazole to the three cell lines in the absence of endogenous protein resulted in identical levels of cyclin B. Interestingly, in this case the Plk1 level in cells expressing CLIP-T287A was clearly lower than that of cells expressing CLIP-170-WT and -T287D (compare lanes 7, 8, and 9 in Fig. 4, B-D). We interpret these results to indicate that CLIP-170-T287A expression causes a G2 block under this condition, as the level of Plk1 continues to increase in mitosis, whereas the level of cyclin B remains the same after G2 (20).
Next, control or CLIP-170-depleted U2OS cells were synchronized with the double thymidine block to arrest in the G1/S boundary, released for different times, and stained with phospho-H3 antibodies. We found that both control and CLIP-170depleted cells started to enter mitosis at 10 h after release, reached a peak at 12 h after release, and exited mitosis at 16 h after release. Although CLIP-170 depletion did not affect mitotic entry and exit, CLIP-170-depleted cells had significantly higher levels of phospho-H3-positive staining at 10, 12, and 14 h postrelease, indicating that CLIP-170 is involved in mitotic progression (data not shown). This is consistent with the published CLIP-170 loss-of-function phenotypic analysis (12). To assess the contribution of Cdc2 phosphorylation of CLIP-170 during mitosis, similar experiments were performed using U2OS cells stably expressing GFP-CLIP-170 (WT, T287A, or T287D) in the absence of endogenous CLIP-170 (Fig.  4E). Compared with cells expressing CLIP-170-WT and -T287D, a significant lower percentage of cells expressing CLIP-170-T287A could go through mitosis, indicating that  Cdc2 phosphorylation of CLIP-170 is essential for its function during cell cycle progression.
To further confirm the phenotypes we described above, U2OS cells stably expressing GFP-CLIP-170 (WT, T287A, or T287D) were grown on coverslips and stained with cyclin B or phosphohistone H3 antibodies. In agreement with the Western results in Fig. 4B, cells expressing CLIP-170-T287A show significantly higher levels of cyclin B (50%) than those of cells expressing CLIP-170-WT (19%) or -T287D (23%). In striking contrast, the levels of phosphohistone H3, a mitotic marker, of these cells are the same (Fig. 4F), arguing that cells expressing CLIP-170-T287A tend to arrest in G2, but not M phase.
Next, a series of FACS analysis was performed with U2OS cells stably expressing GFP-CLIP-170 (WT, T287A, or T287D). In randomly growing cultures, cells expressing CLIP-170-T287A show a dramatic increase in population with 4N DNA content even before the gating. After gating the GFP-positive cells, we detected only the population with 4N DNA content in cells expressing CLIP-T287A. This enrichment was not observed in cells expressing GFP-CLIP-170-WT and -T287D (Fig. 4G). We also tried to follow the cell cycle progression after synchronization of these cells. Toward that end, cells were blocked at the G2/M boundary with roscovitine and released for different times. Both cells expressing GFP-CLIP-170-WT and -T287D were able to go through mitosis, as indicated by a gradual decrease of 4N DNA population and an increase of 2N DNA population at the same time. As expected, no obvious cell cycle progression was detected in cells expressing GFP-CLIP-170-T287A even after 8 h of release from the G2 block (Fig. 4H).
During the G2/M transition, centrosome separation is one essential step to ensure subsequent bipolar spindle formation in metaphase. We thus monitored centrosome separation in synchronized U2OS cells by staining with anti-␥-tubulin, a centrosome marker (Fig. 4I). As expected, U2OS cells stably expressing both rat GFP-CLIP-170-WT and GFP-CLIP-170-T287D showed gradual increases of centrosome separation as cells went through the G2/M transition. In striking contrast, centrosome separation was strongly inhibited in cells stably expressing CLIP-170-T287A (Fig. 4J).
As CLIP-170 is a centrosomal protein (Fig. 3E), we continued to analyze its loss-of-function phenotypes using ␥-tubulin staining. Strikingly, depletion of CLIP-170 in U2OS cells leads to formation of supernumerary ␥-tubulin dots (compare Fig. 5,  A versus B). Similar experiments were performed in U2OS cells stably expressing GFP-CLIP-170 (WT, T287A, or T287D) to assess the contribution of Cdk-mediated phosphorylation of CLIP-170 in the process. While both cells expressing GFP-CLIP-170-WT and -T287D still have normal centrosomal numbers in the absence of endogenous protein, cells expressing  OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42 GFLP-CLIP-170-T287A failed to reverse the phenotype (Fig. 5,  C and D).

CLIP-170 Is Phosphorylated by Cdc2
It has been reported that CLIP-170 is phosphorylated at multiple sites, most likely by multiple kinases (15). To date, specific phosphorylation sites have not been determined, and information about the upstream kinases is very limited. In this communication, we provide evidence that Cdc2/cyclin B is responsible for phosphorylation of Thr 287 of CLIP-170, and that Cdc2-mediated phosphorylation of CLIP-170 is required for its plus end microtubule localization. More significantly, we show that the Cdc2-mediated phosphorylation of CLIP-170 is essential in the G2/M transition. Specifically, cells expressing the Cdc2 unphosphorylatable mutant of CLIP-170 accumulate Plk1 and cyclin B and fail to separate the centrosomes in late G2. As the critical role of CLIP-170 in mitosis was established using both dominant negative and RNAi approaches, we have shown that CLIP-170 function during the G2/M transition is also regulated by Cdc2/cyclin B, the master kinase that controls cell division. The mechanism by which the Cdc2-mediated phosphorylation of CLIP-170-Thr 287 regulates the G2/M transition is currently unknown. Different possibilities are currently under investigation.
Centrosomes duplicate once and only once per cell cycle, although how this process is controlled is still elusive. To our knowledge, nucleophosmin is the only protein that has been identified as yet to prevent centrosome duplication (21,22). It was shown that nucleophosmin dissociates from the centrosomes upon Cdk2/cyclin E-mediated phosphorylation in late G1, thus allowing centrosome duplication in the S phase (21). Based on its subcellular locations during different phases of the cell cycle and its inhibitory role of centrosome duplication, it was further proposed that nucleophosmin acts as a licensing factor to ensure that the centrosome duplicates once and only once per cell cycle (21,22). However, nucleophosmin does not bind to centrosomes during S and G2 phases, only re-associating with centrosomes in mitosis (21), suggesting that additional factors rather than nucleophosmin must act to inhibit centrosome reduplication during the G2 phase. CLIP-170 clearly localizes at the centrosomes throughout the entire cell cycle, including S and G2 phases (Fig. 3E), and its depletion leads to centrosome reduplication, indicating that CLP-170 might be a factor that prevents centrosome reduplication after the S phase. Consistent with the cell cycle regulation of the levels of cyclin B, the essential role of CLIP-170 to inhibit centrosome reduplication during G2 is Cdc2/cyclin B phosphorylationdependent. Our data provide a novel mechanism of the centrosome cycle.