Bimodal Interaction of Mammalian Polo-Like Kinase 1 and a Centrosomal Scaffold, Cep192, in the Regulation of Bipolar Spindle Formation

Abstract

Serving as microtubule-organizing centers, centrosomes play a key role in forming bipolar spindles. The mechanism of how centrosomes promote bipolar spindle assembly in various organisms remains largely unknown. A recent study with Xenopus laevis egg extracts suggested that the Plk1 ortholog Plx1 interacts with the phospho-T46 (p-T46) motif of Xenopus Cep192 (xCep192) to form an xCep192-mediated xAurA-Plx1 cascade that is critical for bipolar spindle formation. Here, we demonstrated that in cultured human cells, Cep192 recruits AurA and Plk1 in a cooperative manner, and this event is important for the reciprocal activation of AurA and Plk1. Strikingly, Plk1 interacted with Cep192 through either the p-T44 (analogous to Xenopus p-T46) or the newly identified p-S995 motif via its C-terminal noncatalytic polo-box domain. The interaction between Plk1 and the p-T44 motif was prevalent in the presence of Cep192-bound AurA, whereas the interaction of Plk1 with the p-T995 motif was preferred in the absence of AurA binding. Notably, the loss of p-T44- and p-S995-dependent Cep192-Plk1 interactions induced an additive defect in recruiting Plk1 and γ-tubulin to centrosomes, which ultimately led to a failure in proper bipolar spindle formation and mitotic progression. Thus, we propose that Plk1 promotes centrosome-based bipolar spindle formation by forming two functionally nonredundant complexes with Cep192.

INTRODUCTION

As the major microtubule-organizing center in somatic animal cells, centrosomes play a critical role in establishing bipolar spindles. Centrosomes consist of a pair of centrioles surrounded by electron-dense pericentriolar material (PCM), which is thought to serve as a scaffold for recruiting various proteins that are critical for microtubule (MT) assembly. Prior to entering mitosis, centrosome size increases dramatically by recruitment of the γ-tubulin ring complex (γ-TuRC) and other PCM proteins, and this process, called centrosome maturation, confers to centrosomes a greater ability to nucleate MTs.

Centrosome maturation occurs through the actions of various PCM scaffolding proteins and regulatory kinases. One of the PCM scaffolds important for this process is a conserved centrosomal protein called Cep192. Early in the cell cycle, Cep192 is detected as an inner PCM ring structure with a diameter of ∼300 to 400 nm (1–3). As cells enter mitosis, the level of Cep192 increases severalfold, and it accumulates on mitotic PCM (4). Interestingly, depletion of Cep192 results in the almost complete loss of centrosome-associated γ-tubulin, whereas overexpression of Cep192 leads to the formation of ectopic puncta in the cytoplasm. These ectopic puncta are capable of recruiting γ-tubulin and other key components that are important for γ-tubulin recruitment. However, how Cep192 functions as a scaffold to support centrosomal maturation and how its function is integrated into the cell cycle have remained elusive.

Besides centrosomal scaffolds that serve as a platform for centrosome maturation, phosphorylation by kinases appears to play an important regulatory role in promoting this event. Data obtained from various studies show that two mitotic Ser/Thr kinases, Plk1 and AurA (and their orthologs in various organisms), play a key role in recruiting γ-tubulin and promoting bipolar spindle formation (5–9). Interestingly, recent studies with Xenopus laevis egg extracts revealed that xCep192 binds to and activates xAurA, and this event is important for the interaction with Plx1 (Xenopus Plk1 ortholog) and for promoting γ-tubulin recruitment to centrosomes (9, 10). In addition, studies with human HeLa cells suggested that these Cep192-mediated processes are largely conserved (9). Since AurA has been shown to function as an upstream kinase of Plk1 at the time of mitotic entry (11, 12), the formation of the xCep192-xAurA complex appears to be a key step in promoting Plx1-dependent centrosome maturation. Notably, however, the Xenopus xCep192(T46A) mutant, lacking T46-dependent Plx1 binding, or the xCep192Δ(543–747) mutant, lacking AurA binding, still maintained a large fraction (∼70%) of its MT-nucleating activity (9). These observations suggest the presence of an alternative pathway(s) that regulates the function of xCep192 in the Xenopus system.

In this study, we investigated the underlying mechanism of how human Cep192 functions together with Plk1 to promote γ-tubulin recruitment and bipolar spindle formation at mitotic centrosomes. Our results showed that in human cells, Plk1 is recruited to centrosomes through an interaction with either the p-T44 or the p-S995 motif of Cep192 and that the loss of both the T44- and S995-dependent interactions results in an additive defect in γ-tubulin recruitment and bipolar spindle formation. Remarkably, in the presence of Cep192-bound AurA, Plk1 preferentially interacted with the T44 motif by self-phosphorylating this site, whereas in the absence of Cep192-bound AurA, Plk1 favored the S995-dependent interaction. Based on these observations, we propose that Plk1 interacts with Cep192 in a bimodal fashion to localize to centrosomes and promote γ-tubulin recruitment and bipolar spindle formation.

MATERIALS AND METHODS

Plasmid construction.

To generate the pCI-neo-FLAG₃-Plk1 (pKM3789), pCI-neo-FLAG₃-Plk1(H538A K540M) (pKM3790), or pCI-neo-FLAG₃-Plk1(K82M) (pKM3791) construct, the respective KpnI (end-filled)-XhoI fragment of Plk1 prepared from either pUC19-Plk1 (pKM3464), pUC19-Plk1(H538A K540M) (pKM3465), or pUC19-Plk1(K82M) (pKM3611) was subcloned into a pCI-neo-FLAG₃ vector digested by PmeI and XhoI. The pCI-neo-HA-AurA (pKM4459) and pCI-neo-HA-AurA(T288A) (pKM4460) constructs were generated by inserting the respective PCR products into a modified pCI-neo-HA vector (2) digested by PmeI and NotI. The green fluorescent protein (GFP)-Bora construct was kindly provided by Dong Zhang (New York Institute of Technology, Old Westbury, NY).

The pEGFP-C2-Cep192 (pKM3105) construct encoding residues 1 to 2538 of Cep192 was generated by inserting a reverse transcription-PCR (RT-PCR)-generated N-terminal fragment of Cep192 into the pCR3.1-EGFP-Cep192(507–2538) (numbers indicate amino acid residues) construct (a gift from David Sharp, Albert Einstein College of Medicine, Bronx, NY). To generate the pEGFP-C1-Cep192 (pKM3552) and pCI-neo-HA-Cep192 (pKM3360) constructs, a SalI fragment containing full-length Cep192 was subcloned into a pEGFP-C1 vector (Clontech) digested by SalI and a pCI-neo-HA vector digested by XhoI, respectively. A pEGFP-C1-Cep192 construct containing the S43A (pK4623), T44A (pK4229), R47A (pK4624), D48A (pK4625), R49A (pK4626), or S995A (pKM3553) mutation or the T44A S995A double mutation (pKM4230) was generated by inserting a PCR-generated SalI fragment containing each respective mutation into the pEGFP-C1 vector digested by the same enzyme. The pCR3.1-EGFP-Cep192(507–1065)(S995A) construct (pKM3119) was generated by replacing the wild-type sequence with an HpaI-BamHI fragment containing the S995A mutation. A bacterial construct containing either Plk1 (pKM5548) or Cep192(1–1093) (pKM5552) was generated by inserting each respective XhoI-NotI fragment of the PCR product into the pET28b vector (Novagen) digested by SalI and NotI. The glutathione S-transferase (GST)-fused form of Cep192(1–647) (pKM3156) was cloned by inserting a SalI-digested fragment into the pGEX-4T3 vector (GE Healthcare) digested by the same enzyme. The pET32b-Aurora A construct was a kind gift from Takeshi Urano (Shimane University School of Medicine, Izumo, Japan).

For the construction of Cep192 deletion mutants, a SalI fragment containing Cep192 P1(1–647) (pKM3290), Cep192 P3(507–1400) (pKM3287), or Cep192 P6(1700–2538) (pKM3289) or a PmeI-SmaI fragment containing Cep192 P5(901–1800) (pKM3288) was cloned into the pEGFP-C1 vector digested by either SalI or SmaI, respectively.

A lentiviral Cep192-sil construct containing either the untagged wild type (WT) (pKM3235), the T44A (pKM4560) or S995A (pKM3236) mutation, or the T44A S995A double mutation (pKM4558) was generated by inserting each respective SalI fragment into a modified pHR′.J-CMV-SV-puro vector (pKM2994) (2) digested by the corresponding enzyme. To generate constructs for tandem affinity purification, an AscI-EcoRV fragment containing Cep192 P1(1–647) (pKM3291), Cep192 P3(507–1400) (pKM3292), Cep192 P5(901–1800) (pKM3293), or Cep192 P6(1700–2538) (pKM3294) was cloned into the pHR′.J-CMV-FLAG₃-TEV-ZZ vector (pKM3045) digested by AscI and PmeI. The ZZ tag is a derivative of Staphylococcus protein A, which has a high affinity for IgG. The TEV sequence provides a cleavage site for a highly site-specific protease found in tobacco etch virus (TEV).

Cell culture and transfection.

HeLa, U2OS, and HEK293T cells were cultured as recommended by the American Type Culture Collection. Transfections into these cells were carried out by using either Lipofectamine 2000 (Invitrogen) for protein overexpression or Lipofectamine RNAiMAX (Invitrogen) for small interfering RNA (siRNA)-based knockdowns. For the production of lentiviruses, transfection was performed by using a calcium phosphate coprecipitation method (13). To effectively deplete Cep192 proteins, U2OS cells were transfected twice with an siRNA against Cep192 (siCep192) (nucleotide positions 2407 to 2427) (14) and cultured for a total period of 96 h. To trap the majority of cells in either the S or M phase, cells were treated with 2.5 mM thymidine (Sigma-Aldrich) or 666 nM nocodazole (Sigma-Aldrich) for 20 h, respectively.

Lentivirus generation and infection.

Lentiviruses were generated by cotransfecting HEK293T cells with pHR′-CMVΔR8.2Δvpr-, pHR′-CMV-VSV-G (protein G of vesicular stomatitis virus)-, and pHR′-CMV-SV-puro-based constructs containing the target gene. Stable U2OS cell lines were generated by infecting the cells with lentiviruses expressing the gene of interest and were selected with 2 μg/ml of puromycin (Sigma-Aldrich). The selected cells were then transfected with siRNAs to deplete RNA interference (RNAi)-sensitive endogenous proteins.

Tandem affinity purification of Cep192-binding proteins.

HEK293T cells stably expressing various Cep192-Flag₃-TEV-ZZ constructs, namely, P1(1–647), P3(507–1400), P5(901–1800), and P6(1700–2538), were generated by using the appropriate lentiviruses and then treated with 666 nM nocodazole to enrich mitotic targets. The resulting cells were lysed in TBSN buffer (20 mM Tris-Cl [pH 8.0], 150 mM NaCl, 0.5% NP-40, 5 mM EGTA, 1.5 mM EDTA, 0.5 mM Na₃VO₄, 20 mM p-nitrophenyl phosphate, and a protease inhibitor cocktail [Roche]) (15) and subjected to tandem affinity purification as previously described (14).

GST-PBD and peptide pulldown assays.

To carry out Plk1 polo-box domain (PBD) pulldown assays, bacterially expressed, bead-bound GST, the GST-PBD WT, or GST-PBD(H538A K540M) (gifts from Michael B. Yaffe, Massachusetts Institute of Technology, Cambridge, MA) was incubated with an equal amount of mitotic HeLa cell lysates in TBSN buffer for 2 h and then precipitated. The resulting precipitates were washed four times with TBSN buffer and analyzed by immunoblotting with the indicated antibodies.

Peptide-based pulldown assays were carried out as previously described (16).

Antibody production.

Rabbit polyclonal antibodies were generated against the synthesized phospho-T44 (p-T44) (residues 34 to 53; SNLGLPVAVSpTLARDRSSTD [p-T44 is indicated in boldface type]) or phospho-S995 (p-S995) (residues 986 to 1004; PSTSPLSHSpSPSEISGTSS [p-S995 is indicated in boldface type]) peptide at Young In Frontier Co., Ltd. (Seoul, South Korea). Affinity purification of these antibodies was carried by out using C-(CH₂)₆-conjugated phosphopeptides and nonphosphopeptides that were immobilized separately on SulfoLink coupling resin (Pierce).

Rabbit anti-Cep192 was previously described (14). Other antibodies used in this study were purchased from commercial sources: rabbit anti-GFP, mouse anti-Plk1, and goat anti-γ-tubulin antibodies were obtained from Santa Cruz Biotechnologies; mouse anti-Flag antibody was obtained from Sigma-Aldrich; rat antihemagglutinin (anti-HA) antibody was obtained from Roche; and both rabbit anti-AurA and rabbit anti-AurA p-T288 antibodies were obtained from Cell Signaling.

In vitro kinase assay and GST pulldown.

For the experiments shown in Fig. 6E, kinase assays were carried out with kinase reaction buffer (50 mM Tris-HCl [pH 7.5], 10 mM MgCl₂, 2 mM dithiothreitol [DTT], 2 mM EGTA, and 20 mM p-nitrophenyl phosphate) in the presence or absence of 1 mM ATP. Bacterially expressed, purified His-Cep192(1–1097) and His-AurA were first incubated in kinase reaction buffer for 30 min at 30°C. Purified His-Plk1 was then added to the reaction mixture and incubated for an additional hour at 30°C. Reactions were terminated by mixing with SDS sample buffer and analyzed.

FIG 6.

Cell cycle-dependent regulation of the p-T44 and p-S995 epitopes and phosphorylation of the T44 site by AurA-activated Plk1. (A) Immunostaining was carried out by using HeLa cells harvested at the indicated time points (hours) after double-thymidine (DT) release (top), and centrosome-localized p-T44 and p-S995 signals were quantified (n ≥ 20 per sample at each time point) (bottom). Bar, 5 μm. Error bars indicate standard deviations. (B) Quantification of centrosome-localized p-T44 or p-S995 signals after treatment of HeLa cells with 0.5 μM MLN8237, 0.5 μM VX680, 0.1 μM BI2536, or 0.2 μM BMI1026 (n ≥ 20 per sample). To enrich the mitotic population, cells released from a DT block for 9 h were used. Error bars indicate standard deviations. (C) Immunoblot analysis was carried out with transfected HEK293T cells treated with nocodazole for 20 h. 82M, K82M mutant; 44A, T44A mutant; 995A, S995A mutant; AA, T44A S995A double mutant; 288A, T288A mutant. (D) Immunoblot analysis was performed by using transfected HEK293T cells treated with nocodazole for 20 h. (E) In vitro kinase assays were carried out by using purified protein, as described in Materials and Methods. The resulting samples were immunoblotted. Numbers indicate relative signal intensities.

For the GST pulldown experiments shown in Fig. 2D, kinase reactions were performed as described above except that bead-bound GST-Cep192(1–647) or the GST control was used. Reacted samples were diluted in TBSN buffer (20 mM Tris-Cl [pH 8.0], 150 mM NaCl, 0.5% NP-40, 5 mM EGTA, 1.5 mM EDTA, 0.5 mM Na₃VO₄, and 20 mM p-nitrophenyl phosphate) and then subjected to GST pulldown. After mixing of the glutathione (GSH)-agarose precipitates with SDS sample buffer, samples were separated by 10% SDS-PAGE and analyzed by immunoblotting.

FIG 2.

AurA and Plk1 form a ternary complex with Cep192 in a cooperative manner. (A) Immunoblot analysis was carried out by using lysates fractionated on a Superdex 200 gel filtration column. Arrowheads indicate fractions 21 to 23, which contain cofractionated Cep192, AurA, and Plk1. Numbers indicate relative signal intensities. (B and C) Immunoprecipitation analysis was carried out by using transfected HEK293T cells. Cells in panel B were treated with thymidine or nocodazole for 20 h. Numbers in panel C indicate relative signal intensities. (D) In vitro kinase assays were carried out, and the resulting samples were subjected to GST pulldown. To detect their cooperative binding to the GST-Cep192(1–647) ligand, limited amounts of AurA and Plk1 were added to the reaction mixture to ensure that ligands are in excess, and coprecipitated AurA and Plk1 were detected by immunoblotting. Numbers indicate relative signal intensities. (E) Immunoprecipitation analysis was carried out with cotransfected HEK293T cells as described above for panels B and C. AM, H538A K540M double mutant. Numbers indicate relative signal intensities.

Coimmunoprecipitation and immunoblotting.

Cells were lysed in TBSN buffer, and the resulting lysates were incubated with the primary antibody for 1 h at 4°C, followed by an additional 2 h of incubation with protein G-Sepharose beads (Santa Cruz). Where indicated, 20 μM the Plk1 PBD-binding PLHSpT phosphopeptide (17) was provided to competitively inhibit the interaction between the Plk1 PBD and its target. Immunoprecipitates were washed five times with TBSN buffer, separated by 10% SDS-PAGE (7% to detect Cep192), transferred onto a polyvinylidene difluoride (PVDF) membrane, and then detected by immunoblotting with the indicated antibodies using an enhanced chemiluminescence detection system (Pierce). The signal intensities of the protein bands of interest were quantified by using ImageJ software.

Size exclusion chromatography.

Size exclusion chromatography was performed with the Akta Explorer instrument (GE Healthcare), which was equipped with a HiLoad 16/60 Superdex 200 prep-grade column (GE Healthcare). The column was preequilibrated with loading buffer (20 mM Tris-HCl [pH 7.5], 1 mM MgCl₂, 1 mM EGTA, and 150 mM NaCl) and then applied with 1 ml (2 mg/ml) of precleared cell lysates. Gel filtration was carried out with loading buffer at 1 ml/min, and fractions were collected every 2 min (2 ml/fraction). The resulting samples were then subjected to immunoblot analysis.

Immunofluorescence microscopy and quantification.

Immunostaining analysis was carried out essentially as previously described (16), using the indicated primary antibodies and appropriate secondary antibodies (Alexa Fluor 488 or 594) from Invitrogen. Confocal images were obtained by using the Zeiss LSM 780 system mounted on a Zeiss Observer Z1 microscope. To quantify the fluorescence signal intensities, images of unsaturated fluorescence signals were acquired with the same laser intensity at 512 by 512 pixels and with a 12-bit resolution. The intensities of the fluorescence signals were quantified by using Zeiss ZEN confocal software and then plotted by using the GraphPad Prism 6 program.

RESULTS

Mammalian Plk1 PBD interacts with Cep192 in a phospho-dependent manner.

In our effort to better understand the function of Cep192 during the M phase of the cell cycle, we isolated Cep192-binding proteins by performing tandem affinity purification with HEK293T cells expressing various ligands and treated with nocodazole to enrich mitotic targets (Fig. 1A). Mass spectrometry analyses revealed that both the Cep192 P3(507–1400) and P5(901–1800) fragments specifically coprecipitated Plk1, whereas the Cep192 P1(1–647) fragment efficiently interacted with AurA (Fig. 1B). In good agreement with data from a previous report (9), a systematic analysis revealed that human Cep192(401–647) was necessary for AurA binding (data not shown). These data suggest that AurA and Plk1 bind to distinct regions on Cep192. A relatively low efficiency of Plk1 binding to Cep192, compared to that of AurA binding to Cep192, could be attributed to the fact that Plk1 binding requires the prior generation of a phosphoepitope on its target (18), whereas AurA binding does not (19).

FIG 1.

The Plk1 PBD interacts with Cep192 in a phospho-dependent manner. (A) Cep192 constructs used for tandem affinity purification. (B) Numbers of AurA and Plk1 peptides retrieved by mass spectrometry analysis. (C) Immunostaining was carried out with asynchronously growing U2OS cells. Bar, 5 μm. (D) Immunoprecipitation (IP) analysis was carried out with HeLa cells treated with thymidine or nocodazole for 20 h. Samples were then immunoblotted to detect endogenous proteins. Asterisks indicate cross-reacting proteins. (E) Immunoprecipitation analysis was carried out with nocodazole-treated HeLa cell lysates preincubated with buffer (−) or 20 μM either a Plk1 PBD-binding phosphopeptide (p), PLHSpT (17), or its respective nonphosphopeptide (non-p). Numbers indicate relative signal intensities, and asterisks indicate cross-reacting proteins. (F) Immunoprecipitation analysis was carried out with HEK293T cells transfected with the indicated constructs. AM, H538A K540M double mutant. (G) A GST pulldown assay was carried out by using nocodazole-treated HeLa lysates. The asterisk indicates a cross-reacting protein. CBB, Coomassie brilliant blue. The dotted line in the Coomassie brilliant blue gel indicates merged gels. (H) GST pulldown was carried out by using HEK293T cells transfected with the constructs described above for panel A and treated with nocodazole for 20 h. The resulting precipitates were immunoblotted with the indicated antibodies. p-Cdc25C precipitated with the Plk1 PBD served as a positive control. (I) A pulldown was carried out by using the indicated bead-immobilized phosphopeptides (top) and nocodazole-treated HeLa lysates, and the samples were then subjected to immunoblot analysis to detect coprecipitated endogenous Plk1. Phosphorylated residues are indicated in boldface type.

Previous studies showed that Cep192 localizes to centrosomes throughout the cell cycle and becomes most abundant at these sites during the M phase of the cell cycle (2, 4). Centrosome-localized Plk1 is detected as early as the late S phase and sharply peaks at mitotic centrosomes (15, 20). Consistent with these findings, the localization of Cep192 to centrosomes preceded that of Plk1 to centrosomes (Fig. 1C), suggesting that Cep192 could serve as a scaffold that recruits Plk1 to these structures. Since both Cep192 and Plk1 have been shown to play important roles in mitotic spindle assembly (4, 9, 10, 21), we therefore chose to investigate the physiological significance of the Cep192-Plk1 interaction during the cell cycle.

To examine whether Cep192 interacts with Plk1 at its endogenous levels, we performed coimmunoprecipitation analysis using HeLa cells treated with either thymidine (S-phase arrest) or nocodazole (M-phase arrest). Immunoprecipitation of Cep192 coprecipitated Plk1 from M-phase but not S-phase cells, whereas it coprecipitated AurA from both S- and M-phase cells (Fig. 1D). These observations suggest that under physiological conditions, Cep192 interacts with AurA prior to interacting with Plk1. Notably, Cep192 showed markedly retarded electrophoretic mobility in M phase (Fig. 1D, third panel), likely due to AurA- and Plk1-dependent phosphorylation (see below). Since the Plk1 PBD binds to a phosphorylated epitope (18), Plk1 may interact with phosphorylated Cep192 during M phase. Consistent with this view, the provision of a Plk1 PBD-binding phosphopeptide PLHSpT (17) but not its respective nonphosphopeptide, greatly diminished the level of the Cep192-Plk1 interaction, whereas it failed to alter the level of the Cep192-AurA interaction within the same sample (Fig. 1E).

In a second experiment, the H538A K540M (AM) mutation, which cripples the ability of the Plk1 PBD to recognize its phospho-target (18), completely annihilated the interaction between cotransfected Plk1 and Cep192 (Fig. 1F) or between the recombinant GST-PBD and endogenous Cep192 (Fig. 1G). Furthermore, the GST-fused Plk1 PBD, but not its respective PBD(AM) mutant, interacted with the Cep192 P5(901–1800) fragment (Fig. 1H). Analyses of 11 potential Plk1 PBD-binding phosphopeptides visually identified from the Cep192 primary sequence revealed that the p-S995-containing peptide was capable of efficiently interacting with endogenous Plk1 (Fig. 1I), hinting that the S995 residue is phosphorylated in vivo, and this event is likely important for the Cep192-Plk1 interaction.

Synergism in the formation of the Cep192-AurA-Plk1 ternary complex.

Studies with Xenopus egg extracts have shown that xAurA binds to xCep192, and the resulting xCep192-xAurA complex activates Plx1 to induce bipolar spindle assembly (9, 10). However, how Cep192, AurA, and Plk1 interact with one another to elicit this event remains uninvestigated. To examine the molecular nature of their interactions, we performed gel filtration chromatography using cotransfected HEK293T cells treated with nocodazole for 20 h. Because the level of endogenous Cep192 is low and because both AurA and Plk1 bind to many targets under physiological conditions, the use of cotransfected cell lysates was necessary. In the presence of GFP-fused Cep192 (i.e., GFP-Cep192), the majority of AurA coeluted with Cep192 in high-molecular-weight fractions (fractions 21 to 23) (Fig. 2A, arrowheads) immediately following the void volume (i.e., fraction 19) (Fig. 2A), suggesting that Cep192 and AurA form a high-molecular-weight complex. The ability of Cep192 and AurA to form a high-molecular-weight complex may stem from the capacity of Cep192 to establish homomeric interactions under nocodazole- and thymidine-treated conditions (Fig. 2B). As expected if Plk1 formed a complex with Cep192 and AurA, a significant fraction of coexpressed Plk1 was enriched in fractions 21 to 23 (Fig. 2A, arrowheads), while the majority of Plk1 was detected as a monomeric 68-kDa protein (Fig. 2A). The relatively low level of Plk1 coeluted with the high-molecular-weight Cep192-AurA complex (fractions 21 to 23) could be attributed in part to the fact that the generation of a phosphorylated epitope is a prerequisite step for the complex's interaction with the Plk1 PBD (18). In control GFP vector-transfected cells, both AurA and Plk1 failed to form a high-molecular-weight complex (Fig. 2A, bottom three panels), suggesting that AurA and Plk1 associate with each other through Cep192.

If Cep192, AurA, and Plk1 indeed form a thermodynamically stable ternary complex, the presence of either AurA or Plk1 may promote the binding of the other to Cep192. In support of this notion, expression of AurA greatly increased the Cep192-Plk1 interaction, while expression of Plk1 also enhanced the Cep192-AurA interaction but at a somewhat reduced level (Fig. 2C, right). Notably, the expression of either AurA or Plk1 moderately decreased the electrophoretic mobility of GFP-Cep192, whereas the expression of both AurA and Plk1 greatly decreased it (Fig. 2C, input). In a second experiment, we were able to reconstitute the findings shown in Fig. 2C by performing in vitro kinase reactions followed by GST pulldown (Fig. 2D) and demonstrated the formation of a ternary Cep192-AurA-Plk1 complex using bacterially expressed purified proteins (Fig. 2D, lane 6). Judging from the Ponceau-stained gel, it was apparent that AurA and Plk1 cumulatively phosphorylate Cep192 (Fig. 2D, bottom). These findings suggest that AurA and Plk1 directly phosphorylate Cep192 in an additive manner and that they cooperatively promote the formation of a Cep192-based ternary complex. In addition, consistent with the PBD-dependent Plk1-Cep192 interaction (Fig. 1F to H), WT Plk1, but not its respective Plk1(AM) mutant, coimmunoprecipitated Cep192 or the Cep192-AurA complex, and the coexpression of AurA substantially increased the Cep192-Plk1 interaction (Fig. 2E, compare lanes 2 and 4).

Cep192 mediates the interaction between and the reciprocal activation of AurA and Plk1.

The synergistic formation of a ternary Cep192-AurA-Plk1 complex raises the question of how Cep192 serves as a platform to regulate AurA and Plk1. To directly determine whether Cep192 serves as a scaffold that mediates the interaction between AurA and Plk1, we carried out coimmunoprecipitation analyses using cotransfected HEK293T cells that express AurA and Plk1 in the presence or absence of GFP-Cep192. In the total lysates, coexpression of Cep192 resulted in the generation of p-T288 epitope-containing activated AurA (22) (Fig. 3A, input, third panel) and p-T210 epitope-containing activated Plk1 (23) (Fig. 3A, input, bottom). Coimmunoprecipitation analyses under these conditions revealed that AurA and Plk1 reciprocally coprecipitated each other only in the presence of Cep192 (Fig. 3A, right panels). These findings demonstrated that Cep192 mediates the AurA-Plk1 interaction and that the formation of the Cep192-AurA-Plk1 complex could be important for activating AurA and Plk1. Notably, treatment of cells with either thymidine or nocodazole did not significantly alter the levels of activated AurA and Plk1 and the degree of the Cep192-dependent AurA-Plk1 interaction (Fig. 3A). These results hint that the formation of an active ternary Cep192-AurA-Plk1 complex can occur without the involvement of a third element, which may influence AurA or Plk1 function, in S- or M-phase-arrested cell lysates. Therefore, except where indicated, we used asynchronously cultured cells for further analysis.

FIG 3.

Cep192-dependent reciprocal activation of AurA and Plk1 is required for the proper Cep192-Plk1 interaction. (A) Immunoprecipitation analysis was carried out by using transfected HEK293T cells treated with thymidine or nocodazole for 20 h. (B and C) Immunoblot analysis was carried out by using HEK293T cells transfected with the indicated constructs. (D) Immunoprecipitation analysis was carried out by using transfected HEK293T cells. Numbers in panels B to D indicate relative signal intensities. 995A, S995A mutant; 288A, T288A mutant; 82M, K82M mutant. (E and F) HeLa cells released from a DT block for 9 h were treated with control dimethyl sulfoxide (DMSO), 0.5 μM MLN8237, or 0.1 μM BI2536 for 1 h and then harvested for immunostaining analysis. Cells treated with Cep192 siRNA (siCep192) were included for comparison. Costaining with either anti-AurA (E) or anti-Plk1 (F) antibody was carried out to assess the total level of centrosome-localized AurA or Plk1 signals. Representative images are shown. Bars, 5 μm. DAPI, 4′,6-diamidino-2-phenylindole. (G) Quantification of centrosome-localized p-T210 or p-T288 signals was carried out with metaphase HeLa cells, prepared as described above for panels E and F, from three independent experiments (n ≥ 20 per sample/experiment). Error bars indicate standard deviations. (H) Immunoblot analysis was carried out by using HEK293T cells transfected with the indicated constructs. Numbers indicate relative signal intensities.

To determine whether AurA and Plk1 can reciprocally activate each other in the presence of Cep192, we examined the level of activated AurA or Plk1 in total lysates coexpressing either Plk1 or AurA, respectively (Fig. 3B). Consistent with the results shown in Fig. 3A, coexpression of AurA markedly increased the level of p-T210 epitope-containing activated Plk1. In addition, as expected if Plk1 promoted the Cep192-AurA interaction shown in Fig. 2C, Plk1 coexpression increased the level of p-T288 epitope-containing activated AurA in the total lysates (Fig. 3B, compare the third panel with the total amount of AurA in the second panel). Although the degree of Plk1-dependent AurA activation was not as prominent as that of AurA-dependent Plk1 activation, these findings hint that AurA and Plk1 reciprocally activated each other in the presence of Cep192.

We then investigated whether the reciprocal activation of AurA and Plk1 required the kinase activity of each enzyme. To this end, HEK293T cells were cotransfected with either the WT or its respective kinase-inactive form of AurA or Plk1 and subjected to immunoblot analysis. The Cep192(S995A) mutant deprived of the p-S995 motif-dependent interaction with Plk1 (Fig. 1I) was included for comparison. The results showed that the expression of WT AurA, but not its kinase-inactive T288A mutant (22), efficiently induced the p-T210 epitope on Plk1 but only in the presence of Cep192 (Fig. 3C). In addition, WT Plk1, but not its kinase-inactive K82M mutant (15), was required to achieve the full activation of AurA, again in the presence of Cep192 (Fig. 3C). Notably, the Cep192(S995A) mutant, which is expected to abrogate the p-S995-dependent Cep192-Plk1 interaction, still possessed a significant capacity to promote AurA-dependent Plk1 p-T210 generation (Fig. 3C, compare lane 3 to lanes 1 and 2), suggesting that the S995 motif may not be the sole source of the interaction between Cep192 and Plk1 (see below).

To further investigate whether the kinase activities of AurA and Plk1 are important for their interactions with Cep192, we performed coimmunoprecipitation analysis using samples prepared as for Fig. 3C. We observed that cells expressing either the kinase-inactive AurA(T288A) or Plk1(K82M) mutant showed a markedly diminished level of interaction between Cep192 and Plk1, compared to cells expressing the respective WT AurA or WT Plk1 (Fig. 3D). In addition, Cep192(S995A)-expressing cells showed a much higher level of the Cep192-Plk1 interaction than did AurA(T288A)- or Plk1(K82M)-expressing cells (Fig. 3D), further suggesting the existence of an S995-independent Cep192-Plk1 interaction.

We next investigated whether AurA and Plk1 reciprocally activate each other at their endogenous levels and whether this event requires Cep192. To do this, HeLa cells enriched in M phase were treated with pharmacological inhibitors against AurA (MLN8237) or Plk1 (BI2536) for 1 h and then immunostained. To examine the effect of Cep192 depletion on cross-activation between AurA and Plk1, cells treated with Cep192 siRNA (siCep192) were also included. Quantification of the level of the centrosome-localized AurA p-T288 or Plk1 p-T210 epitope was performed by using cells with metaphase chromosome morphology to indirectly assess the overall AurA or Plk1 kinase activity at this location. The results showed that treatment of cells with MLN8237, which greatly diminished the level of the p-T288 epitope, also significantly reduced the level of the p-T210 epitope (Fig. 3G). In addition, treatment with BI2536 led to similar degrees of cross-reductions for p-T210 and p-T288 signals (Fig. 3G), thus confirming the results obtained with transfected lysates (Fig. 3C). Depletion of Cep192 drastically diminished the levels of centrosome-localized p-T288 and p-T210 signals (Fig. 3E to G), indicating that Cep192 is required for the cross-activation of AurA and Plk1 at their endogenous sites. In cotransfected HEK293T cells, GFP-fused Bora, a nuclear activator of AurA (24), was capable of activating coexpressed AurA and Plk1 but only marginally (judging from the levels of the activational p-T288 and p-T210 epitopes, respectively) (Fig. 3H, compare lanes 1 and 3). Under conditions where GFP-Cep192 and GFP-Bora were expressed at similar levels, GFP-Cep192 activated coexpressed AurA and, in turn, Plk1 more efficiently than did GFP-Bora (Fig. 3H, compared lanes 2 and 3).

Plk1 interacts with Cep192 in both AurA-dependent and -independent manners.

While we were searching for additional Plk1-binding motifs, Joukov et al. demonstrated that in Xenopus egg extracts, Plx1 interacts with the T46 motif of xCep192 and promotes microtubule recruitment (9). Notably, the analogous human T44 motif does not closely resemble the previously proposed Plk1 PBD-binding motif (Φ/P-Φ-T/Q/H/M-S-pS/pT-P/X, where X is any amino acid residue and Φ is a hydrophobic residue) (18) and therefore was not included in our phosphopeptide screens for Plk1 PBD binding (Fig. 1I). However, we found that a p-T44 peptide immobilized on beads precipitated endogenous Plk1 from mitotic HeLa cell lysates as efficiently as the p-S995 peptide, whereas their respective non-phospho-S995 and -T44 peptides did not detectably precipitate endogenous Plk1 from the same lysates (Fig. 4B). Furthermore, mutational analyses of the purportedly PBD-binding T44 motif and its neighboring residues confirmed that both the T44 and, somewhat less importantly, S43 residues were critical for Plk1 binding, while their downstream residues were not (Fig. 4A). The importance of a Ser residue at the −1 position and the nonessential role of residues downstream of the SpT dipeptide have been previously demonstrated (17, 18). These data strongly suggest that in addition to the S995 motif (Fig. 1I), the T44 motif is required for the Cep192-Plk1 interaction in humans. Highlighting the importance of both T44- and S995-dependent Plk1 PBD binding, these two residues are conserved among higher eukaryotic organisms (Fig. 4C).

FIG 4.

Two distinct modes of the Cep192-Plk1 interaction in the absence or presence of AurA. (A) Immunoprecipitation analysis was carried out by using cotransfected HEK293T cells lysates. The numbers indicate the relative levels of Plk1 coprecipitated with Cep192 and AurA. Because the p-T44-dependent Cep192-Plk1 interaction requires a prior Cep192-AurA interaction, AurA was cotransfected. (B) Peptide pulldown assays were carried out by using bead-immobilized nonphosphopeptides or their respective phospho-T44 (p-T44) or phospho-S995 (p-S995) peptides. Sequences of synthesized T44 or S995 peptides are shown at the left. The p-T44 and p-S995 residues are indicated in boldface type. (C) Sequence alignment of the T44 and S995 regions of Cep192 orthologs in Homo sapiens (H.s.), Mus musculus (M.m.), Gallus gallus (G.g.), Xenopus laevis (X.l.), and Danio rerio (D.r.). Identical residues are marked by boldface type. (D to F) Immunoprecipitation analysis was carried out by using HEK293T cells transfected with the indicated constructs. Numbers indicate relative signal intensities. 288A, T288A mutant; 995A, S995A mutant; 44A, T44A mutant; AA, T44A S995A double mutant.

To investigate the functional relationship between the T44- and the S995-dependent Cep192-Plk1 interactions, we closely examined the significance of these interactions under different conditions. When only Plk1 and Cep192 were coexpressed, we found that the S995A mutation greatly diminished the Cep192-Plk1 interaction, whereas the T44A mutation only mildly crippled the interaction (Fig. 4D). In contrast, under conditions where Cep192, AurA, and Plk1 were all coexpressed, the T44A mutation caused a much more pronounced defect in the Cep192-Plk1 interaction than did the S995A mutation (Fig. 4E). The T44A S995A (AA) double mutations appeared to cripple the Cep192-Plk1 interaction in an additive manner (Fig. 4D and E). These results suggest that Plk1 binds to two distinct phospho-motifs (i.e., the p-T44 and p-S995 motifs) on Cep192 and that the binding is contingent upon the presence or absence of AurA.

Since both AurA and Plk1 activities were required for the efficient formation of the Cep192-AurA-Plk1 ternary complex (Fig. 3D), we next investigated the relationship between AurA activity and T44- or S995-dependent Cep192-Plk1 interactions. Using cotransfected HEK293T cells expressing either WT AurA or the kinase-inactive AurA(T288A) mutant, we confirmed that AurA activity was required for the Cep192-Plk1 interaction (Fig. 4F, compare lanes 1 and 2). Under these conditions, the Cep192(S995A) mutant, which still bears the T44 motif, coprecipitated Plk1 in an AurA activity-dependent manner (Fig. 4F, lanes 3 and 4). In contrast, the Cep192(T44A) mutant, which contains the S995 motif, bound to Plk1 regardless of the presence or absence of AurA activity, although the level of the Cep192(T44A)-Plk1 interaction in AurA(T288A)-expressing cells was somewhat diminished (Fig. 4F, compare lanes 5 and 6). This finding suggests that unlike the T44-dependent Cep192-Plk1 interaction, the S995-dependent interaction is largely independent of AurA kinase activity. As expected, the Cep192(T44A S995A) double mutant failed to coprecipitate Plk1 because of the lack of both AurA-dependent (i.e., T44-dependent) and AurA-independent (i.e., S995-dependent) interactions between Cep192 and Plk1. Whether the AurA-dependent interaction between the T44 motif of Cep192 and Plk1 is regulated by either AurA activity itself or AurA-activated Plk1 activity is addressed below.

Cell cycle-dependent regulation of the p-T44 and p-S995 motifs and generation of the “self-primed” p-T44 epitope by AurA-activated Plk1.

The two distinct modes of Plk1 binding to either the T44 or S995 motif of Cep192 (Fig. 4) suggest that phosphorylation of the T44 and S995 residues could be differentially regulated. To investigate this possibility, we generated phospho-T44 and -S995 antibodies (Fig. 5) and carried out immunostaining analyses using cells released synchronously from a double-thymidine (DT) block. These results showed that the levels of both the p-T44 and p-S995 signals were hardly detectable or very low at interphase centrosomes (Fig. 6A). However, the levels of these signals increased sharply during M phase, peaking at metaphase centrosomes before precipitously decreasing in late mitosis (Fig. 6A). Interestingly, unlike the p-S995 epitope, which was detectable mostly at metaphase centrosomes, a significant level of the p-T44 epitope was detectable even at prophase centrosomes (Fig. 6A). This observation suggests that under physiological conditions, the p-T44-dependent formation of the Cep192-AurA-Plk1 complex could occur at this stage.

FIG 5.

Characterization of Cep192 p-T44 and p-S995 phospho-antibodies. (A and B) HEK293T cells transfected with the indicated constructs were treated with nocodazole for 20 h and then immunoblotted. Unlike the p-S995 epitope, coexpression of Plk1 was necessary to readily detect the p-T44 epitope. 82M, K82M mutant; 44A, T44A mutant; 995A, S995A mutant. Asterisks indicate cross-reacting proteins. (C to H) Immunostaining with Cep192 p-T44 or p-S995 antibodies was carried out by using U2OS cells silenced for either the control luciferase (siGL) or Cep192 (siCep192). The positions of centrosomes are marked by AurA (C), Cep192 (D), Cep152 (E and F), or γ-tubulin (G and H) signals. Bars, 5 μm.

Since AurA and Plk1 cooperated to form the Cep192-AurA-Plk1 ternary complex, and both AurA and Plk1 appeared to additively phosphorylate Cep192 (Fig. 3), we investigated whether their kinase activities are required to generate the p-T44 and p-S995 epitopes. Because the S995 residue is followed by a Pro residue at the +1 position, we also examined whether a Pro-directed mitotic kinase, Cdc2 (25), contributes to the production of the p-S995 epitope. To this end, HeLa cells enriched in mitosis from DT release were additionally treated with pharmacological inhibitors against AurA (MLN8237 and VX680) (26, 27), Plk1 (BI2536) (28), or Cdc2 (BMI1026) (29). Interestingly, although the treatment of cells with an AurA-selective inhibitor, MLN8237, or a pan-Aur inhibitor, VX680, only modestly reduced the level of the p-T44 signal, both inhibitors greatly diminished the level of the p-S995 signal (Fig. 6B). On the other hand, the treatment of cells with a Plk1 inhibitor, BI2536, selectively abolished the p-T44 epitope but not the p-S995 signal. Cells treated with a Cdc2 inhibitor, BMI1026, exhibited somewhat reduced levels of both the p-T44 and p-S995 signals (Fig. 6B). A short (20-min) treatment was required to minimize the degree of unscheduled mitotic exit by Cdc2 inhibition (29).

To further investigate whether AurA or Plk1 contributes to the production of the p-T44 or p-S995 epitope, HEK293T cells expressing Cep192 were coexpressed with either Plk1 or AurA and harvested for immunoblot analysis (we were not able to detect endogenous p-T44 and p-S995 epitopes with our antibodies). The results showed that coexpression of WT Plk1 but not its respective kinase-inactive K82M mutant efficiently induced the p-T44 epitope in Cep192 (Fig. 6C, lanes 1 and 2). Interestingly, the level of the p-T44 epitope remained unaltered by the Cep192(S995A) mutation (Fig. 6C, compare lanes 1 and 5), suggesting that Plk1-dependent production of the p-T44 epitope does not require prior phosphorylation at S995. AurA alone failed to detectably induce the p-T44 epitope (Fig. 6C, lanes 9 to 16). Under these conditions, we found that the level of the p-S995 epitope was unaltered by the presence or absence of Plk1 or AurA kinase activity or the T44A mutation (Fig. 6C, lanes 1 to 4 and 9 to 12), suggesting that the production of the p-S995 epitope is independent of AurA, Plk1, and the T44A mutation.

Although AurA activity was required for the T44-dependent Cep192-Plk1 interaction (Fig. 4F, compare lanes 3 and 4), AurA failed to directly induce the p-T44 epitope (Fig. 6C). Thus, we investigated whether AurA contributes to the T44-dependent interaction via Plk1. We found that cell lysates coexpressing Cep192 and Plk1 induced the p-T44 epitope at a modest level, whereas cell lysates coexpressing Cep192, AurA, and Plk1 induced the p-T44 epitope at a substantially (∼8.5-fold) increased level (Fig. 6D, compare lanes 2 and 4). Coexpression of the kinase-inactive AurA(T288A) mutant failed to enhance the level of the p-T44 epitope (Fig. 6D, compare lanes 4 and 5), suggesting that the activation of Plk1 by AurA is critical for the phosphorylation and generation of the p-T44 epitope. Judging from the levels of the AurA p-T288 and Plk1 p-T210 epitopes, the coexpression of AurA and Plk1 significantly cross-activated each other (Fig. 6D, fourth and sixth panels), confirming the findings shown in Fig. 3. In contrast to these findings, the level of the p-S995 epitope remained unchanged by the expression of Plk1, AurA, or both (Fig. 6C and D), implying that the markedly decreased p-S995 signal caused by MLN8237 or VX680 treatment (Fig. 6B) is likely an indirect consequence of AurA inhibition. Consistent with data shown in Fig. 6C and D, in vitro kinase assays with purified proteins showed that Plk1 directly phosphorylated and generated the p-T44 epitope, whereas AurA did not (Fig. 6E, lanes 4 and 5, respectively). The addition of AurA greatly enhanced (9.3-fold) the ability of Plk1 to generate the p-T44 epitope, closely reflecting the severalfold increased level of the p-T210 epitope on Plk1 (Fig. 6E, compare lanes 4 and 7). The effect of Plk1 on AurA activation was less apparent under these conditions, because bacterially expressed, purified AurA was already highly phosphorylated at its activational T288 residue (Fig. 6E, third panel, lane 2).

Both p-T44- and p-S995-dependent Cep192-Plk1 interactions are required for proper spindle formation and mitotic progression.

To determine the physiological significance of T44- and/or S995-dependent interactions between Cep192 and Plk1, U2OS cells expressing a lentivirus-based control vector or various RNAi-insensitive Cep192 (Cep192-sil) constructs were silenced for either the control luciferase (siGL) or endogenous Cep192 (siCep192) by siRNA transfection. Immunoblot analysis revealed that the silencing of Cep192 efficiently depleted endogenous Cep192 but not exogenously expressed Cep192-sil and that exogenous Cep192-sil forms were expressed at a level comparable to that of endogenous Cep192 (Fig. 7A). Immunostaining analysis showed that all Cep192-sil forms localized to centrosomes efficiently (Fig. 7B and C), suggesting that both the T44A and S995A mutations do not alter the subcellular localization of Cep192. Under these conditions, cells expressing the Cep192(T44A S995A) mutant exhibited significantly diminished γ-tubulin recruitment to centrosomes, while cells expressing either the Cep192(T44A) or the Cep192(S995A) single mutant displayed a somewhat modest defect in this process (Fig. 7B and C). These findings suggest that the γ-tubulin recruitment defect associated with the T44A or S995A mutation is additive. In line with these results, Cep192(AA)-expressing cells also appeared to exhibit an additive defect in the amount of the total Plk1 population and the active Plk1 p-T210 form recruited to centrosomes (Fig. 7B and C). Interestingly, although the level of centrosome-localized Cep192 remained unchanged by the T44A and S995A mutations, the levels of both centrosomally localized AurA and its active AurA p-T288 form were modestly diminished in cells expressing the Cep192(AA) mutant (Fig. 7B and C), likely because of the loss of the Plk1-promoted Cep192-AurA interaction shown in Fig. 2C. As a consequence of the defects in the proper recruitment of Plk1 and γ-tubulin, Cep192(AA) mutant cells exhibited a significantly increased mitotic block, with monopolar spindle morphology, whereas either Cep192(T44A) or Cep192(S995A) cells displayed these defects somewhat weakly (Fig. 7D to F).

FIG 7.

Additive defect of the T44A and S995A mutations in the recruitment of γ-tubulin and the establishment of bipolar spindles. (A) Immunoblotting was carried out for U2OS cells stably expressing RNAi-insensitive Cep192 (Cep192-sil) constructs and depleted of endogenous Cep192. CBB, Coomassie brilliant blue. (B to F) The resulting cells from panel A were immunostained (B and E) with the appropriate antibodies, and the fluorescent signal intensities (C) (n ≥ 20 per sample), numbers of mitotic cells (D) (n ≥ 4,000 per each sample), and numbers of cells with monopolar spindles (E and F) (n ≥ 45 per each sample) were quantified. Bars, 5 μm (B and E). Error bars in panels C, D, and F indicate standard deviations. 44A, T44A mutant; 995A, S995A mutant; AA, T44A S995A double mutant.

DISCUSSION

AurA and Plk1 cooperation in the formation of the Cep192-AurA-Plk1 ternary complex.

While characterizing the physiological function of human Cep192, Gomez-Ferreria et al. proposed that Cep192 serves as a scaffold that is critical for γ-TuRC recruitment to centrosomes and bipolar spindle formation at these structures (4). Later studies with Xenopus egg extracts suggested that xCep192 binds to and activates xAurA (10), which in turn activates Plx1 to promote centrosome maturation and bipolar spindle assembly (9). These observations highlight the importance of an xCep192-organized xAurA-Plx1 cascade in regulating γ-TuRC recruitment and MT assembly at centrosomes. However, the molecular mechanism of how Cep192, AurA, and Plk1 orchestrate their functions to elicit these processes in human cells has remained largely uninvestigated. In this study, we demonstrated that AurA and Plk1 are recruited to distinct regions of Cep192 and form a ternary complex in a cooperative manner (Fig. 2). Furthermore, Cep192-bound AurA and Plk1 reciprocally activate each other, although the degree of Plk1-dependent AurA activation was somewhat less substantial than that of AurA-dependent Plk1 activation (Fig. 3B and C). The underlying mechanism of how Plk1 activates Cep192-bound AurA remains to be further investigated. Since AurA functions as an upstream kinase of Plk1 (11, 12), the feedback activation of AurA by Plk1 constitutes a positive amplification loop that ultimately leads to the activation of Plk1 itself. Thus, the formation of a ternary Cep192-AurA-Plk1 complex is likely crucial for the rapid amplification of AurA-dependent Plk1 activation prior to mitotic entry.

The mechanism underlying the cooperative formation of the Cep192-AurA-Plk1 complex remains elusive. Here, we demonstrated that the initial formation of the Cep192-AurA complex does not require AurA or Plk1 kinase activity, but the presence of Plk1 enhances the formation of the Cep192-AurA complex (Fig. 2C and 3D). On the other hand, Plk1 recruitment to the Cep192-AurA complex required both AurA and Plk1 activities (Fig. 3D). Thus, once AurA binds to Cep192, AurA and Plk1 cooperate to generate the Cep192-AurA-Plk1 complex. These observations suggest that AurA- and Plk1-dependent Cep192 phosphorylations likely induce a conformational change in the Cep192 scaffold that allows both AurA and Plk1 to bind more efficiently to Cep192. Since Plx1-dependent xCep192 phosphorylation appears to promote γ-tubulin recruitment and bipolar spindle formation in Xenopus egg extracts (9), it will be interesting to further investigate whether AurA- and/or Plk1-dependent phosphorylations on Cep192 cooperatively promote γ-tubulin recruitment to the centrosome-associated Cep192 scaffold in human cells. Identification of Cep192 phosphorylation sites and subsequent investigation of the physiological significance of this event will be necessary for a better understanding of this process.

It should be noted that cells overexpressing Cep192, AurA, and Plk1 form a high-molecular-weight complex, which is significantly larger than the calculated size of the heterotrimeric Cep192-AurA-Plk1 complex (calculated molecular mass of ∼393 kDa). This observation suggests that Cep192, AurA, and Plk1 may form a multitrimeric complex under physiological conditions. Unlike in the Xenopus system, where xAurA is proposed to form a dimer/multimer (9), we found that Cep192 homomerizes under physiological conditions, whereas AurA and Plk1 do not detectably do so (Fig. 2A and B). Thus, we propose that in human cells, Cep192 homomerization is likely the driving force for the formation of a Cep192-AurA-Plk1 multitrimeric complex.

Two distinct modes of interaction between Cep192 and Plk1.

Studies with Xenopus egg extracts showed that the xCep192(T46A) mutant lacking T46-dependent Plx1 binding or the xCep192Δ(543–747) mutant lacking xAurA binding still maintained ∼70% of its MT-nucleating activity, suggesting the presence of a T46- and xAurA-independent xCep192 function (9). Here, we demonstrated that Plk1 binds to Cep192 through interactions between its C-terminal PBD and one of the two phosphorylated motifs (p-T44 and p-S995) on Cep192. Remarkably, the Plk1 interaction with the p-T44 motif was prevalent when Cep192 was in a complex with AurA, whereas the Plk1 interaction with the p-S995 motif was preferred when Cep192 was in an AurA-unbound homomeric state (Fig. 4). Since AurA phosphorylates Cep192 and retards its gel mobility (Fig. 2 and 3), AurA may have the capacity to induce a conformational change in Cep192 that favors the T44 motif-dependent Plk1 interaction. Interestingly, the defect associated with the impairment of T44- and S995-dependent functions appeared to be additive (Fig. 7; also see below). Thus, we propose that the bimodal interactions between Cep192 and Plk1 ultimately lead to the formation of two distinct Cep192-Plk1 complexes that function in parallel pathways to recruit downstream components, such as the γ-TuRC (Fig. 8). In this view, the bimodal formation of the Cep192-Plk1 complex may serve as a crucial step for providing versatility in Plk1-dependent downstream events. As with the importance of the biomodal nature of the Cep192-Plk1 interaction, both the T44 and S995 residues are highly conserved among vertebrates (Fig. 4C).

FIG 8.

Model illustrating the formation of the AurA-dependent Cep192 p-T44-Plk1 complex and the AurA-independent Cep192 p-S995-Plk1 complex and the regulation of MT nucleation by these complexes. During interphase, AurA binds to the region of Cep192 spanning residues 401 to 647 and becomes activated at centrosomes. In the presence of AurA binding, Plk1 preferentially phosphorylates and interacts with the T44 motif of Cep192 through the “self-priming and binding” mechanism (31, 32). The formation of the Cep192-AurA-Plk1 ternary complex allows efficient AurA-dependent Plk1 activation. Plk1 also reciprocally activates AurA, presumably by phosphorylating Cep192 and enhancing the Cep192-AurA interaction, thus creating a positive feedback loop to further potentiate Cep192-AurA-Plk1-mediated γ-tubulin recruitment and bipolar spindle formation. In the absence of AurA binding to Cep192 (considering a strong interaction between Cep192 and AurA, a minor fraction of Cep192 could be free of AurA at centrosomes), Plk1 binds to the p-S995 motif generated by an unknown kinase and promotes γ-tubulin recruitment and bipolar spindle formation. The resulting Cep192-Plk1 complex may function as a trigger of downstream events (thin arrows) or serve as an intermediate subcomplex for the formation of the Cep192-AurA-Plk1 ternary complex with available AurA. The ternary complex may also revert back to the Cep192-Plk1 binary complex upon AurA dissociation and p-T44 dephosphorylation. Notably, the AurA-Plk1-dependent production of the p-T44 epitope is independent of S995 phosphorylation, while the production of the p-S995 epitope is independent of AurA, Plk1, and T44 phosphorylation. In line with these observations, the loss of both T44- and S995-dependent Cep192-Plk1 interactions induces an additive defect in γ-tubulin recruitment and bipolar spindle formation.

Then how are the p-T44 and p-S995 epitopes generated on Cep192? We observed that Plk1 expression efficiently induced the p-T44 epitope in the presence of AurA (Fig. 6D), whereas Plk1 inhibition almost completely abolished it (Fig. 6B). These findings suggest that Cep192-bound AurA, which presumably induces a conformational change in Cep192 through phosphorylation, allows Plk1 to “self-prime” the T44 residue and bind to the resulting p-T44 epitope, thus generating a Cep192-AurA-Plk1 ternary complex (Fig. 8). Contrary to these findings, the mechanism of how the p-S995 epitope is generated remains puzzling at present. Our results showed that although AurA inhibition greatly diminished the level of the p-S995 epitope, AurA expression failed to increase it at all (Fig. 6). Consistent with these results, we found that the primary sequence of the S995 motif does not conform to the canonical AurA consensus phosphorylation site [(K/R)XX(S/T), where X is any residue] (30). Identification of the kinase responsible for generating the p-S995 epitope will be very important for a better understanding of the mechanism underlying the formation of an AurA-independent Cep192-Plk1 complex.

Physiological significance of T44- and S995-dependent Cep192-Plk1 interactions.

Prior to mitotic entry, centrosomes recruit various components of the pericentriolar material, including MT-nucleating factors such as the γ-TuRC. This process, called centrosome maturation, is critical for the proper establishment of bipolar spindles during mitosis. Data obtained from studies with various organisms suggest that Plk1 and AurA, and their orthologs, play an important role in this process (5–9). Recently, Cep192 has been proposed to function as an important centrosomal scaffold that promotes centrosome maturation and bipolar spindle formation (4). Subsequent studies with Xenopus egg extracts have shown that an xCep192-organized xAurA-Plx1 cascade promotes γ-TuRC recruitment and MT assembly (9).

The molecular mechanisms of how human Cep192 assembles functional complexes with AurA and Plk1 and how the formation of these complexes is regulated are beginning to emerge. Our data showed that the impairment of T44- and S995-dependent Cep192-Plk1 interactions induced additive defects in Plk1 recruitment and activation, γ-tubulin recruitment, and bipolar spindle formation (Fig. 7). These observations suggest that T44- and S995-dependent Cep192-Plk1 interactions function in parallel to promote Plk1-dependent biochemical and cellular processes that are important for bipolar spindle establishment (Fig. 8). Given that AurA efficiently binds to Cep192, as shown in Fig. 2A, the Cep192-Plk1 complex, which preferably forms in the absence of AurA, could serve as an intermediate subcomplex that facilitates the rapid reassembly of the Cep192-AurA-Plk1 ternary complex. However, we cannot exclude the possibility that the Cep192-AurA-Plk1 complex disassembles via AurA dissociation and p-T44 dephosphorylation to regenerate the Cep192-Plk1 binary complex (Fig. 8). In this regard, perhaps the intricate regulation of the two Plk1-bound states (i.e., Cep192-AurA-bound and Cep192-bound states) could be important in helping to maintain the proper levels of Cep192-associated Plk1 and, consequently, Plk1-mediated MT nucleation at centrosomes. A deeper understanding of how the Cep192-Plk1 interaction is regulated during the cell cycle and how this interaction promotes the recruitment of γ-tubulin to centrosomes will be critical for providing new insights into the mechanism of centrosome maturation and bipolar spindle formation at the time of mitotic entry.