GCP6 is a substrate of Plk4 and required for centriole duplication

Centrioles are microtubule-based structures that organize the centrosome and nucleate cilia in most eukaryotes (Nigg and Raff, 2009). They are surrounded by the pericentriolar material (PCM), a protein matrix that contains protein complexes required for centrosome-associated functions such as microtubule nucleation. Excessive numbers of centrioles lead to tumorigenesis in flies (Basto et al., 2008) and have been linked to chromosomal instability in human cells (Ganem et al., 2009). Thus, controlling centriole numbers ensures that cells have the proper number of centrosomes and cilia. Centriole duplication must be coordinated with the cell cycle to ensure that the number of centrioles in the cell doubles precisely during each cell cycle. In mammalian cells, a single procentriole starts forming perpendicular to the wall of each parental centriole around the G1–S transition. Once the assembly of the two new procentrioles has been initiated, further centriole duplication is inhibited until the cells have passed through mitosis (Wong and Stearns, 2003).

Several human protein kinases were shown to be crucial for centriole duplication and assembly. Cyclin-dependent kinase 2 (Cdk2) is necessary for the initiation of centrosome duplication (Hinchcliffe and Sluder, 2002). A few candidate substrates including CP110, NPM/B23 and Mps1 have been identified as being necessary for the function of Cdk2 in centrosome duplication (Chen et al., 2002; Fisk et al., 2003; Okuda et al., 2000). Polo-like kinase 2 (Plk2) is also involved in the regulation of centriole reproduction (Warnke et al., 2004). It is known that Plk2 phosphorylates NPM/B23 to trigger centriole duplication and CPAP to control its biological activity for centriole elongation (Chang et al., 2010; Krause and Hoffmann, 2010). A central role in the control of centriole biogenesis and duplication has been attributed to Plk4 (Sak in Drosophila) (Bettencourt-Dias et al., 2005; Habedanck et al., 2005). Following activation of Plk4 and its recruitment to the centrosome by its interacting protein Cep152 (Cizmecioglu et al., 2010; Dzhindzhev et al., 2010; Hatch et al., 2010), a sequential assembly of several crucial proteins including human Sas6, Cep135, CPAP, γ-tubulin and CP110 is induced (Kleylein-Sohn et al., 2007). The F-box protein Slimb in Drosophila or its human homologue β-TRCP mediate proteolytic degradation of autophosphorylated Plk4 (Cunha-Ferreira et al., 2009; Guderian et al., 2010; Rogers et al., 2009).

A key component of the centrosome is the γ-tubulin ring complex, γ-TuRC, whose major function is to nucleate microtubule polymerization. In Drosophila and budding yeast, two molecules of γ-tubulin and one copy of each the corresponding orthologues of γ-tubulin complex protein 2 (GCP2) and GCP3 form the γ-tubulin small complex (γ-TuSC) (Kollman et al., 2008; Oegema et al., 1999). In metazoans, multiple γ-TuSCs assemble with several additional proteins, including GCP4, GCP5, GCP6 and Nedd1 (GCP-WD), into γ-TuRCs (Fava et al., 1999; Moritz et al., 2000; Murphy et al., 2001). Recently, additional components of the γ-TuRC denoted MOZART1, MOZART2 (GCP8), and NME7 have been identified (Choi et al., 2010; Hutchins et al., 2010; Teixido-Travesa et al., 2010). It is assumed that the γ-TuRC-specific proteins enhance the microtubule nucleating activity of the γ-TuSC (Kollman et al., 2008; Moritz et al., 2000; Oegema et al., 1999). However, in a number of organisms only the γ-TuSC components were shown to be essential. Depletion of any of theγ-TuSC proteins is lethal and results in defective monopolar mitotic spindles (reviewed by Raynaud-Messina and Merdes, 2007), whereas the deletion of all γ-TuRC-specific genes in fission yeast or Aspergillus nidulans produced viable mutants (Anders et al., 2006; Xiong and Oakley, 2009). Moreover, after depletion of any of the four γ-TuRC-specific proteins in Drosophila melanogaster γ-TuSC subunits were still recruited to the centrosome. Less severe defects in mitotic spindle formation compared with γ-TuSC protein depletion were observed (Verollet et al., 2006). In contrast to this, the role of the γ-TuRC-specific protein Nedd1 is more crucial in human cells. Similar to its Drosophila orthologue Dgp71WD, Nedd1 is not required for γ-TuRC assembly (Haren et al., 2006; Luders et al., 2006). However, unlike Dgp71WD, which is dispensable for γ-TuSC localization to the centrosome in Drosophila (Verollet et al., 2006), human Nedd1 functions as an attachment factor for the γ-TuRC. The role of the other γ-TuRC-specific proteins GCP4, GCP5 and GCP6 in human cells and their contribution to spindle formation is still unclear. In addition, little is known about upstream regulators in the control of γ-TuRC function.

In this report, we show that the human γ-TuRC-specific proteins, GCP4, GCP5 and GCP6 are required for bipolar mitotic spindle formation. Further characterization of GCP6 revealed that it is indispensable for assembly of the γ-TuRC and for microtubule nucleation. We found that GCP6 is located both at the centrioles and the PCM and that it is required for centriole formation as well as centriole reduplication. In addition, Plk4-induced centriole overduplication is dependent on GCP6. GCP6 interacts with and is phosphorylated by Plk4. Furthermore, we show that Plk4-dependent phosphorylation of GCP6 regulates centriole duplication.

To investigate the function of the human γ-TuRC proteins we first analyzed the effect of GCP4, GCP5 and GCP6 on mitotic spindle formation. GCP4, GCP5 and GCP6 were downregulated in HeLa cells using specific short interfering RNAs (siRNAs) (Fig. 1A). Whereas around 25–30% of the cells treated with siRNAs against GCP4 and GCP5 formed monopolar spindles, the percentage of cells showing this phenotype was much more pronounced and was around 55% when GCP6 was downregulated (Fig. 1B,C). These findings point to a crucial role of GCP6 in the regulation of mitotic spindle formation. We therefore decided to study the functions of GCP6 in more detail.

We examined whether GCP6 depletion would affect γ-TuRC formation in human cells. HeLa cell lysates were treated with siRNA against GL2 (firefly luciferase) as control or against GCP6 and then subjected to sucrose gradient sedimentation followed by western blot analysis to detect different γ-TuRC components. In the control sample, GCP6 sedimented at higher molecular weight fractions together with GCP2 and γ-tubulin (Fig. 2A, right). These fractions probably represent the γ-TuRC, which is known to sediment at ~32S (Stearns and Kirschner, 1994). The γ-TuSC-specific components, GCP2 and γ-tubulin, were also identified in the lower molecular weight fractions, which might represent the γ-TuSC or breakdown products of the γ-TuRC. Upon siRNA-mediated downregulation of GCP6, however, the formation of the γ-TuRC was impaired, whereas the lower molecular weight complexes remained intact (Fig. 2A, left), suggesting that GCP6 is a crucial component for the formation of the γ-TuRC. Next, we checked whether absence of GCP6 function had an effect on the centrosomal localization of γ-tubulin. We found that in mitotic monopolar spindles, γ-tubulin localization to the centrosome was markedly reduced in the absence of GCP6 (Fig. 2B), whereas centrosomal localization of another PCM protein, pericentrin, was not decreased (Fig. 2C). These results suggest that in human cells γ-tubulin is recruited to the centrosomes by the γ-TuRC.

The main function of the γ-TuRC is to facilitate microtubule nucleation. We performed microtubule-regrowth experiments to detect whether centrosomal microtubule nucleation depends on GCP6. In HeLa cells treated with GCP6-specific or control siRNAs, microtubules were depolymerized upon cold treatment, and microtubule regrowth was induced by short incubation of the cells at 37°C. As shown in supplementary material Fig. S1, treatment with control siRNA led to microtubule regrowth in ~75% of the cells, whereas only ~15% of cells depleted of GCP6 regrew centrosomal microtubule asters.

Taken together, these results show that in contrast to its Drosophila orthologue, human GCP6 is required for the recruitment of γ-tubulin to the centrosome. The results suggest that the γ-TuRC plays a pivotal role in microtubule nucleation and mitotic spindle formation in human cells.

The γ-TuRC is a well-known component of the PCM. For γ-tubulin, however, an additional localization was described in the lumen of the centriole (Fuller et al., 1995; Moudjou et al., 1996). To analyze the centrosomal localization of GCP6 we made use of HeLa cells that expressed localization and affinity purification (LAP)-tagged GCP6 at the same levels as the endogenous protein (Hutchins et al., 2010). We first performed colocalization experiments of GCP6 with centrin-2, a centriolar marker, localized to the centriolar distal lumen. In G1 phase cells, two centrioles were detectable that showed centrin-2 staining. GCP6 localized predominantly between the two centrioles and showed a partial colocalization with one of the centrin dots (Fig. 3A). Immunoelectron microscopy was used to obtain more definitive insights into the localization of GCP6. We found that GCP6 was localized both to the PCM surrounding the proximal end of the centriole and in the distal part of the centriolar lumen (Fig. 3B). Taken together, these results show that GCP6, similarly to γ-tubulin, is present in the PCM but is also localized to the centriole.

To determine whether centriole duplication required the presence of GCP6, the protein was downregulated in HeLa cells by specific siRNAs and cells were subsequently stained with antibodies against centrin. As shown in Fig. 4A,C, the monopolar spindles observed after GCP6 depletion showed reduced centriole numbers, with mainly two instead of four centrioles per mitotic spindle. To rule out RNAi off-target effects, we treated HeLa cells expressing GCP6–LAP with siRNA against GCP6 (Fig. 4B). These cells stably express mouse LAP-tagged GCP6 under the control of the endogenous GCP6 promoter. This system was used previously to perform RNAi rescue experiments, as mouse genes are frequently resistant to siRNAs targeting human genes (Hutchins et al., 2010). Both the effects on mitotic spindle formation as well as the reduction in centriole numbers could be reverted by the expression of mouse GCP6–LAP (Fig. 4B,C). An efficient downregulation of the endogenous GCP6 but not the mouse GCP6–LAP was confirmed by western blot analysis (supplementary material Fig. S2). Together, these results imply that GCP6 plays a role in the unperturbed centrosome duplication cycle in HeLa cells.

Next, we set out to investigate whether centriole reduplication in S-phase-arrested U2OS cells was also dependent on GCP6. In some cell lines, e.g. U2OS, centrosome duplication can be experimentally uncoupled from DNA replication in the presence of the DNA synthesis inhibitors hydroxyurea or aphidicolin. Upon treatment of U2OS cells with aphidicolin they continue to duplicate their centrosomes (Fig. 4D, upper left panel). To test whether ablation of GCP6 could disturb centriole reduplication, the protein was downregulated by two siRNAs (A: nucleotide 154–172 and B: nucleotide 3801–3819 of GCP6 coding sequence) targeting GCP6 (Oriolo et al., 2007). Because Cdk2 activity is required for centriole reduplication (Lacey et al., 1999; Matsumoto et al., 1999; Meraldi et al., 1999), we used siRNA targeting Cdk2 as a positive control. As summarized in Fig. 4D, similar to the absence of Cdk2, ablation of GCP6 function interfered with centriole reduplication. Next, we analyzed whether GCP6 is required for Plk4-induced centriole overduplication. Therefore, we made use of a hemagglutinin (HA)-tagged Plk4 overexpressing HeLa cell line under the control of the tetracycline-inducible promoter (Cizmecioglu et al., 2010). HA–Plk4 was induced through addition of doxycycline and this resulted in a multiplication of centrin-positive dots. Upon simultaneous downregulation of GCP6, the centrosome overduplication phenotype was abrogated. Whereas around 70% of the control siRNA (GL2)-treated cells contained more than four centrioles, downregulation of GCP6 resulted in only 40–45% of cells with more than four centrioles (Fig. 4E). Taken together, these results suggest that GCP6 is involved in centriole duplication as well as in Plk4-induced centriole biogenesis.

In order to resolve whether other γ-TuRC specific GCPs also play a role in centriole duplication, centrin-positive dots were counted in GCP4- or GCP5-depleted cells. supplementary material Fig. S3 shows that the monopolar spindles observed after GCP4 or GCP5 depletion also contained reduced centriole numbers, which indicates that a complete γ-TuRC is required for centriole duplication.

Downregulation of GCP6 led to a reduction in the amounts of γ-tubulin localized to the centrosome. Because γ-tubulin is also required for centriole duplication, the effects of GCP6 knockdown on centriole duplication could be simply explained by a decrease in γ-tubulin levels. However, as γ-tubulin is involved in Plk4-induced centrosome overduplication (Kleylein-Sohn et al., 2007) it is possible that GCP6, as a crucial subunit of the γ-TuRC, is regulated by Plk4 to induce centriole duplication. To analyze this in more detail, we first looked at whether complexes between endogenous Plk4 and GCP6 could be detected in vivo. As seen in Fig. 5A, endogenous GCP6 was present in Plk4 immunoprecipitates. In support of this notion, interactions between ectopically expressed Myc-tagged Plk4 and FLAG-tagged GCP6 could also be detected in reciprocal co-immunoprecipitations (Fig. 5B). These results demonstrate that Plk4 and GCP6 associate in vivo. To determine whether the two proteins directly interact, we performed pulldown assays using GCP6 that had been translated in vitro in the presence of [³⁵S]-methionine and then precipitated with maltose-binding protein (MBP)-tagged Plk4 bound to amylose beads. A clear recovery of GCP6 was observed, suggesting that the proteins probably bind directly to each other (Fig. 5C).

To explore the possibility that GCP6 is directly phosphorylated by Plk4, we carried out in vitro kinase assays. We used a Zz-tagged (Zz tag consists of two IgG-binding domains from protein A at the N-terminus and a C-terminal His tag) kinase-active form of Plk4 and immunoprecipitated FLAG-tagged GCP6 and found that GCP6 was phosphorylated by Plk4 (Fig. 6A). Together, these results show that GCP6 is a novel Plk4-interacting protein and a substrate of Plk4.

To identify Plk4-specific phosphorylation sites, we subjected in vitro phosphorylated GCP6 to mass spectrometry and identified ten Plk4-specific phosphorylation sites on serine residues. Remarkably, the majority of the Plk4 phosphorylation sites were found within the tandem repeat segment of GCP6 (nine tandem repeats of a 27 amino acid sequence are present in GCP6). Furthermore, three additional phosphorylation sites were identified. One of these (Ser392) was located within the N-terminal grip1 domain (Fig. 6B), which describes a region conserved among the different GCPs (Guillet, 2011). The two other phosphorylation sites (Ser1437 and Ser1465) were located in close proximity to the C-terminal grip2 domain (Fig. 6B). Fig. 6C (left) shows that a GST–GCP6 fragment (amino acids 1027–1286), including the nine tandem repeat region of GCP6 (denoted hereafter as GST–GCP6 RR) was phosphorylated efficiently by immunoprecipitated FLAG-tagged wild-type Plk4 (denoted FLAG–Plk4 WT) but not by a kinase-inactive mutant in which Lys41 was exchanged for arginine (denoted Plk4 K41R) in vitro. No incorporation of [³²P] was detected upon incubation of the GST tag alone with FLAG–Plk4 WT (Fig. 6C, right). To confirm that the phosphorylation sites within the repeat region of GCP6 identified by mass spectrometry were the major sites phosphorylated by Plk4 in this region, we substituted them with alanine. The tandem repeats exhibited a very high sequence identity, therefore it is likely that the identified serine residues (Ser7 and Ser15 of a repeat) are phosphorylated in every repeat. We therefore mutated these 17 serine residues (Fig. 6B, highlighted in gray) into nonphosphorylatable alanine residues (GCP6 17×A). As shown in Fig. 6D, the incorporation of [³²P] was almost completely abolished in the GST–GCP6 RR 17×A mutant (Fig. 6D). Also, the Zz–Plk4 dependent phosphorylation of the full-length FLAG–GCP6 17×A was markedly reduced (Fig. 6E), although not completely abolished. When we additionally mutated the three serine residues into alanine that were found to be phosphorylated outside the GCP6 repeat region (FLAG–GCP6 20×A), the phosphorylation decreased only slightly more (Fig. 6E). Taken together, these results suggest that Plk4 phosphorylates GCP6 mainly in the repeat region but that there are additional phosphorylation sites located outside this region.

To explore the impact of the phosphorylation sites on centriole duplication, we first analyzed whether the GCP6 20×A mutant was still functional with respect to its localization on the centrosome and in complex formation with other γ-TuRC components. We found that, similar to wild-type GCP6 (GCP6 WT), the GCP6 20×A mutant showed a centrosomal localization (Fig. 7A). In addition, the expression of an siRNA-resistant GCP6 20×A mutant rescued the reduced centrosomal γ-tubulin levels observed after GCP6 depletion as efficiently as the expression of siRNA-resistant wild-type GCP6 (Fig. 7A). Moreover, in 293T cells depleted of endogenous GCP6, the GCP6 20×A mutant was still able to interact with other γ-TuRC proteins to a similar extent as wild-type GCP6 (supplementary material Fig. S4A). These results suggest that the GCP6 20×A mutant is functional and that phosphorylation by Plk4 is not required for these functions. However, we observed that in GCP6-depleted cells, expression of a siRNA-resistant GCP6 17×A or GCP6 20×A mutant inhibited centriole duplication compared with cells expressing wild-type GCP6 (supplementary material Fig. S4B). This points to a function of the PLK4-dependent phosphorylation of GCP6 in centriole biogenesis.

To assess whether expression of the GCP6 17×A or GCP6 20×A mutant would interfere with Plk4-mediated centriole formation, FLAG–GCP6 WT, FLAG–GCP6 17×A or FLAG–GCP6 20×A were expressed in HeLa Tet-on cells and Plk4 expression was induced. Whereas expression of wild-type GCP6 did not impair centriole overduplication, expression of GCP6 17×A and GCP6 20×A led to a 7 and 11% reduction of the centriole overduplication phenotype, respectively (Fig. 7B, left). Endogenous GCP6 was still present in this experiment and it might counteract the effect of the GCP6 alanine mutants. Therefore, endogenous GCP6 was downregulated by RNAi, which led to an inhibition of Plk4-induced centriole overduplication by ~25%. Expression of an RNAi-resistant wild-type GCP6 protein clearly rescued the phenotype caused by downregulation of GCP6 (Fig. 7B, right). By contrast, the RNAi-resistant nonphosphorylatable 20×A mutant was not able to rescue ablation of GCP6 function. Furthermore, expression of the siRNA-resistant GCP6 17×A mutant did not rescue the defects in centriole overduplication as efficiently as expression of wild-type GCP6. This suggests that the phosphorylation of GCP6 is essential for Plk4-induced centriole duplication.

Taken together, these results suggest that GCP6 is regulated by Plk4 and is therefore a crucial component of Plk4-induced centriole biogenesis.

We report the functional analysis and regulation of human GCP6, a core component of the γ-TuRC. Our results demonstrate that GCP6 is involved in mitotic spindle formation, γ-TuRC assembly and the localization of γ-tubulin to the centrosome. Furthermore, we show that GCP6 is an integral component of the centriole and required for centriole duplication. Moreover, we find that GCP6 interacts in vitro and in vivo with Plk4. We show that phosphorylation of GCP6 by Plk4 is required for Plk4-induced centriole overduplication.

Although components of the γ-TuSC are crucial for normal mitotic spindle assembly and function (reviewed by Raynaud-Messina and Merdes, 2007), the functions of GCP4, GCP5 and GCP6 are less clear. For GCP6 it was suggested that the protein has a specialized function in organizing microtubules in epithelial cells (Oriolo et al., 2007). In Drosophila melanogaster, only γ-tubulin and the orthologues of GCP2 and GCP3 are essential. Interestingly, although the absence of GCP4, GCP5 and GCP6 leads to mitotic defects in Drosophila, γ-tubulin is still recruited to centrosomes and contributes to microtubule nucleation (Verollet et al., 2006). Similar observations have been made in the fungus Aspergillus nidulans, where deletion of GCP4, GCP5 and GCP6 does not affect localization of γ-TuSC components (Xiong and Oakley, 2009). On the contrary, deletion of the GCP6 homolog Xgrip210 in Xenopus laevis blocked the recruitment of γ-tubulin to the centrosome (Zhang et al., 2000). Importantly, our own data show that in human cells both γ-TuRC formation and the localization of γ-tubulin to the centrosome is impaired when GCP6 is absent (Fig. 2A,B). These findings therefore suggest that in vertebrates cells the formation of a complete γ-TuRC is required for the recruitment of γ-tubulin to the centrosome.

Additionally, we observed a high frequency of monopolar mitotic spindle formation when GCP6 was absent (Fig. 1 and Fig. 4A,C). This might reflect the failure of γ-TuRC recruitment to the centrosome (Fig. 2B) and could result from a decrease in the centrosomal nucleation capacity that is required to push the two spindle poles apart. Moreover, the reduction in centriole numbers observed after GCP6 depletion (Fig. 4) might contribute to the defects in mitotic spindle formation.

A role of γ-tubulin in centriole duplication was reported earlier (Haren et al., 2006; Kleylein-Sohn et al., 2007; Luders et al., 2006; Ruiz et al., 1999; Shang et al., 2002). In Paramecium and Tetrahymena thermophila, depletion of γ-tubulin led to a block in basal body duplication (Ruiz et al., 1999; Shang et al., 2002). Similarly, when γ-tubulin was depleted in Caenorhabditis elegans embryos, centriole assembly was impaired (Dammermann et al., 2004). Interfering with γ-tubulin expression in Drosophila resulted in shortened centrioles, indicating a role of γ-tubulin in centriole morphogenesis (Raynaud-Messina et al., 2004). Most importantly, in human cells it was reported that γ-tubulin is required for Plk4-induced centriole overduplication (Kleylein-Sohn et al., 2007).

In this study, we demonstrate that GCP4, GCP5 and GCP6 also function in centriole duplication (Fig. 4 and supplementary material Fig. S3), suggesting that the whole γ-TuRC is involved in this process. It is likely that the γ-TuRC facilitates the nucleation of centriolar microtubules. This is supported by cryo-electron tomography studies of human procentrioles, which revealed that the proximal end of the first of the centriolar triplet microtubule is capped by a conical structure supposed to be the γ-TuRC (Guichard et al., 2010). Our electron microscopy data suggest that GCP6 is localized to the PCM but also to the distal end of the centriolar lumen (Fig. 3B). The fraction of GCP6 localizing to the PCM in close proximity to the proximal end of the centriole might be involved in the initiation of centriolar microtubule nucleation, whereas the localization within the centriolar lumen could point to a role in centriole elongation or maturation. A similar centriolar localization has been found for human POC5, a protein that is indeed required for centriole elongation (Azimzadeh et al., 2009). Interestingly, Plk4 was found around the proximal end of centrioles (Kleylein-Sohn, 2007), but in addition was observed to localize to the distal end of the centriole, where it might also be involved in centriole maturation (Sillibourne et al., 2010).

Despite the important function of the γ-TuRC for microtubule nucleation events, little is known about the upstream regulators in the control of these processes. Nedd1, a key member of the γ-TuRC components, contributes to γ-TuRC attachment to the centrosome (Haren et al., 2006). Nedd1 is sequentially phosphorylated by Cdk1 and Plk1. This promotes its interaction with γ-tubulin for targeting the γ-TuRC to the centrosome (Zhang et al., 2009). Recently, it has also been shown that γ-tubulin is regulated by the SADB protein kinase by phosphorylation on Ser131, which triggers centriole duplication (Alvarado-Kristensson et al., 2009). We find that GCP6 is directly phosphorylated by Plk4. Mutation of the Plk4-specific serine sites in GCP6 into nonphosphorylatable alanine residues inhibits Plk4-induced centriole overduplication (Fig. 7B). Therefore, an attractive hypothesis is that GCP6 is the subunit of the γ-TuRC that is regulated by phosphorylation to trigger centriole duplication.

A striking feature of the human GCP6 sequence is that it contains a tandem repeat region, which is located between the two conserved grip domains and which is not found in other GCPs (Murphy et al., 2001). Interestingly, this region is conserved down to the Xenopus laevis GCP6 orthologue Xgrip210 (Zhang et al., 2000). Whereas the grip domains were reported to be required for the interaction between GCPs and γ-tubulin during complex formation, the interface between the two conserved domains is assumed to form a flexible hinge in GCP3 (Guillet, 2011; Kollman et al., 2008). Conformational changes in this flexible region, which might be regulated by post-translational modifications, were suggested to be required for triggering the microtubule-nucleating activity of the γ-TuRC (Guillet, 2011; Kollman et al., 2008). Because the majority of Plk4 phosphorylation sites were identified in a corresponding region between the grip1 and grip2 domains of GCP6, including the repeat region, it is tempting to speculate that a phosphorylation of GCP6 by Plk4 activates centriolar microtubule nucleation and thereby procentriole formation.

In conclusion, our work identified GCP6, a crucial subunit of the γ-TuRC, as a novel substrate for Plk4 in centriole duplication.

Human GCP6 cDNA inserted into a pOTB7 vector was kindly provided by Pedro Salas (University of Miami, Miami, FL) and cloned between BamHI and XhoI sites of pCMV-3Tag-1 (Agilent Technologies) and pCDNA3.1 (+) (Invitrogen). To generate GCP6 mutants in which Ser7 and Ser15 of every repeat are mutated to alanine (17×A) the region of GCP6 located between and including the endogenous restrictions enzyme sites SbFI and PflFI was synthesised by Geneart (Invitrogen). Thereby, the corresponding point mutations were introduced as well as seven silent mutations between base pairs 3801 and 3819 to ensure siRNA resistance. The synthesized DNA supplied in a pMA cloning vector was subcloned into the GCP6 internal SbFI and PflFI restriction enzyme sites of pCMV-3Tag-1-GCP6. Three additional mutations were introduced by PCR-based site-directed mutagenesis (Stratagene) using pCMV-3Tag-1-GCP6 17×A as a template. For generation of GST–GCP6 RR, the region between nucleotides 3079 and 3859 of GCP6 was amplified by PCR using pCMV-3Tag-1-GCP6 or pCMV-3Tag-1-GCP6 17×A as a template and cloned into the BamHI and XhoI sites of pGEX4T1 (GE Healthcare). Construction of pCMV-3Tag-2-Plk4, pQE80zz-Plk4 and pMAL-c2-Plk4 was as described (Cizmecioglu, 2010).

Anti-GCP6 rabbit polyclonal antibody was raised against human GST–GCP6 RR (amino acids 1027–1286) (Innovagen). Rabbit and mouse anti-Plk4 antibodies were as described (Cizmecioglu et al., 2010). Rabbit anti-GCP2, rabbit anti-GCP4 and mouse anti-pericentrin antibodies were a kind gift from Andreas Merdes (CNRS/Université de Toulouse, Toulouse, France). Rabbit anti-CP110 antibody was kindly provided by Kunsoo Rhee (Seoul National University, Seoul, Korea). Rabbit anti-GCP5 (H-300), rabbit anti-centrin-2 (N-17-R) and mouse anti-Myc (9E10) antibodies were purchased from Santa Cruz Biotechnology. Mouse anti-α-tubulin (T5168), mouse anti-γ-tubulin (T6557), rabbit anti-γ-tubulin (T3559) and mouse anti-FLAGM2 (F3165) antibodies were obtained from Sigma-Aldrich. Rabbit anti-GFP (NB600-303) antibody, used for western blots and immunofluorescence, was from Novus Biologicals. Rabbit anti-GFP used for immuno-gold electron microscopy was from Iain Mattaj (EMBL, Heidelberg, Germany). Secondary antibodies for western blot were peroxidase-conjugated donkey anti-rabbit and goat anti-mouse (Jackson ImmunoResearch Laboratories). Secondary antibodies for immunofluorescence were goat anti-mouse IgG coupled to Alexa Fluor 488 or Alexa Fluor 594 and goat anti-rabbit IgG coupled to Alexa Fluor 488 or Alexa Fluor 594 (Molecular Probes).

All cells were grown in DMEM containing 1 g/l glucose (Sigma-Aldrich) and supplemented with 10% fetal calf serum (TAA) and 2 mM glutamine at 37°C with 5% CO₂. U2OS cells stably expressing GFP–centrin1 were provided by Stefan Duensing (University of Pittsburgh, Pittsburgh, PA). HeLa TDS cells expressing murine GCP6–LAP were provided by Anthony Hyman (MPI, Dresden, Germany) and were grown in medium supplemented with 400 μg/ml G418 (Invitrogen). HeLa Tet-on cells stably expressing HA–Plk4 under a doxycycline-inducible promoter were generated as described (Cizmecioglu et al., 2010). HA–Plk4 expression was induced by supplementation of media with 2 μg/ml doxycycline for 24 hours. For the centriole reduplication assays, the media of U2OS GFP–centrin1 cells was supplemented with 1.9 μg/ml aphidicolin for 65 hours.

Plasmid DNA transfections of 293T cells were preformed with calcium phosphate according to standard protocols. HeLa, HeLa GCP6–LAP, HeLa Tet-on and U2OS GFP–centrin1 cells were transfected with plasmid DNA or siRNA (Applied Biosystems) using Lipofectamine 2000 (Invitrogen). For the combined transfection of 293T cells with plasmid DNA and siRNA, Lipofectamine 2000 was used. The following siRNA sequences were used: GCP6 A, 5′-GAUGAG-ACUCAACAGCUGCtt-3′; GCP6 B, 5′-CACCCAUGUACCCAUCCCUtt-3′; GCP4, 5′-CGGAAAGGAGCACAAAGAUtt-3′; GCP5, 5′-GGAACAUCAUG-UGGUCCAUCAtt-3′; Cdk2, 5′-GAUGGACGGAGCUUGUUAUtt-3′; firefly luciferase (GL2), 5′-CGUACGCGGAAUACUUCGAtt-3′. Cells transfected with siRNA were analyzed 3 or 6 days after transfection. In the case of 6-day siRNA transfection, cells were transfected a second time with siRNA after 3 days.

For immunofluorescence, cells grown on coverslips were fixed with −20°C methanol for 10 minutes at −20°C. Afterwards, cells were washed once with PBS and blocked for 30 minutes with 5% BSA in PBS. Cells were incubated with primary antibody for 1 hour. Alexa-Fluor-488- or Alexa-Fluor-594-conjugated secondary antibodies (Invitrogen) were incubated for 30 minutes. In the case of staining two different proteins on one coverslip, primary as well as secondary antibodies were incubated consecutively. Following each step of antibody incubation, cells were washed three times with PBS. DNA was stained with Hoechst 33258 (Molecular Probes) for 10 minutes.

Depolymerization of microtubules was induced by incubating cells grown on coverslips in pre-cooled medium for 1 hour on ice. To stimulate microtubule regrowth, the cold medium was replaced by medium pre-warmed to 37°C and cells incubated for 2 minutes at 37°C. Regrowth was stopped by methanol fixation.

Images were taken with a PerkinElmer Ultra-View spinning disc confocal on a Nikon Ti inverted microscope equipped with a 100× NA 1.0 oil immersion objective and an electron multiplying charge-coupled device camera (Hamamatsu Photonics). Exposure times were constant for all samples being compared. Images were processed with NIH ImageJ and Photoshop (Adobe Systems). For quantification of immunofluorescence intensities, Photoshop was used to measure the mean fluorescence intensity of a defined area around the centrosome and near the centrosome (background). Background intensities were subtracted from each measurement. For all samples to be compared, staining was performed in parallel and images were taken on the same day under equal exposure times.

Hela TDS GCP6–LAP cells were grown on coverslips, fixed with 2% formaldehyde for 10 minutes, and permeabilized with 0.05% saponin in PBS for 5 minutes. Cells were incubated with rabbit anti-GFP antibody (Iain Mattaj, EMBL, Heidelberg, Germany) for 3 hours and anti-rabbit IgG Nanogold antibody overnight. Cells were further fixed with 2.5% glutaraldehyde in 50 μM cacodylatebuffer for 20 minutes at 4°C, and Nanogold was silver-enhanced with HQ silver (Nanoprobes). Cells were dehydrated and embedded in epoxy resin.

Cell lysates for western blotting and immunoprecipitation were obtained by incubation of cells with NP40 buffer (40 mM Tris pH 7.5, 150 mM NaCl, 0.5% NP40, 5 mM EDTA, 10 mM β-glycerophosphate, 5 mM NaF, 1 mM DTT, 0.1 mM Na₃VO₄ and protease inhibitors) for 30 minutes on ice and centrifuged for 15 minutes at 13,200 r.p.m. (16,000 g) at 4°C to remove cellular debris. Western blotting was performed according to a standard protocol (Hassepass et al., 2003).

For immunoprecipitations of overexpressed proteins, 293T cells were transfected with the corresponding constructs. Cells were harvested 1.5 days after transfection and lysed with NP40 buffer. The supernatant was incubated with anti-FLAG or anti-Myc antibodies for 2 hours (in the case of FLAG–Plk4 immunoprecipitation for in vitro kinase assay) or overnight at 4°C followed by addition of 10 μl protein G-Sepharose (GE Healthcare) and incubation for 1 hour at 4°C. Beads were washed three to four times with NP40 buffer, boiled in sample buffer and analyzed by western blotting. Endogenous Plk4 was immunoprecipitated form HeLa cell lysates with mouse anti-Plk4 antibody using normal mouse IgGs (Santa Cruz Biotechnology) as control.

For the in vitro pulldown assay, GCP6 was in vitro translated (IVT) in the presence of [³⁵S]-labeled methionine (PerkinElmer) using the TNT-Coupled Reticulocyte Lysate System (Promega), followed by an incubation of 20 μl of the IVT reaction with 10 μg MBP or MBP–Plk4 immobilized on 10 μl amylose beads (NEB) in NP40 buffer for 2 hours at 4°C. Beads were washed three times using NP40 buffer, boiled in sample buffer, and analyzed by SDS-PAGE followed by autoradiography.

For sucrose gradient sedimentation, cells were lysed for 10 minutes on ice in HEPES buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM MgCl₂, 1 mM EGTA) containing protease inhibitors and 0.5% Triton X-100. After centrifugation for 10 minutes at 13,200 r.p.m. (16,000 g) at 4°C supernatants (500 μg of protein) were loaded onto a 2.2 ml 5–40% sucrose gradient prepared as described (Zhang et al., 2000) and centrifuged in a TLS-55 rotor (Beckman Coulter) for 3 hours and 10 minutes at 49,000 r.p.m. at 4°C. Fractions were collected and analyzed by western blotting. Thyroglobulin (660 kDa, 19S) (Calbiochem) was used as molecular mass marker and analyzed in parallel.

Expression of Zz–Plk4 and MBP–Plk4 was described previously (Cizmecioglu et al., 2010; Jaäkel and Görlich, 1998). GST–GCP6 RR WT or GST–GCP6 RR 17×A were expressed in BL21-Rosetta and affinity-purified using glutathione-Sepharose (GE Healthcare) according to the manufacturer's protocol. The GST-tagged proteins were dialyzed against PBS overnight.

For in vitro kinase assays with recombinant Zz–Plk4, FLAG–GCP6 WT, FLAG–GCP6 17×A or FLAG–GCP6 20×A was expressed in 293T cells and immunoprecipitated using anti-FLAG antibodies as described above. FLAG–GCP6 bound to Sepharose beads was washed three times in NP40 buffer and once in kinase assay buffer (50 mM Tris pH 7.5, 10 mM MgCl₂, 10 μM MnCl₂, 1 mM DTT) followed by an incubation with 2.5–5 μg Zz–Plk4 in the presence of 5 μCi [γ-³²P]-ATP (PerkinElmer) in kinase assay buffer supplemented with 33 μM ATP for 5–30 minutes at 30°C. Reactions were stopped by adding sample buffer and heating at 95°C. Samples were analyzed by SDS-PAGE followed by Coomassie Blue staining and autoradiography. For quantification of kinase assays, Photoshop was used to measure the mean phosphorylation intensity, which was normalized to the loading.

For in vitro kinase assays with FLAG–Plk4 expressed in 293T cells, the immunoprecipitated kinase was incubated with 200 ng GST–GCP6 RR WT or GST–GCP6 RR 17×A in the presence of 5 μCi [γ-³²P]-ATP in kinase assay buffer supplemented with 33 μM ATP for 20–30 minutes at 30°C.

SDS-PAGE purified proteins were excised, destained, and in-gel digested with trypsin or chymotrypsin as described (Seidler et al., 2009). Mass spectrometric analyses were performed using a nanoAcquity UPLC (Waters) coupled to a LTQ-Orbitrap XL (Thermo). Gradients and solvents were exactly as described (Seidler et al., 2010). Data was acquired using a Top3 Data dependent acquisition: tandem mass spectrometry (MSMS) spectra of the three most intense precursors with charge ≥2 were recorded. Data were analyzed using Xcalibur 2.0.6 and MASCOT 2.2.2. All phosphorylation sites identified were controlled manually.

We thank Stefan Duensing, Anthony Hyman, Iain Mattaj, Andreas Merdes, Kunsoo Rhee and Pedro Salas for reagents. We acknowledge Ulrike Engel and Christian Ackermann from the Nikon Imaging Center at the University of Heidelberg for equipment and assistance in implementation of experiments. Andreas Merdes, Onur Cizmecioglu and Mei Zhu are thanked for critically reading the manuscript and Lena Ehret for expert technical assistance.