We developed a Xenopus
egg extract system to characterize the action of Plk4 in centriole duplication. Xenopus
embryos can make thousands of centrioles in the absence of transcription or translation (Gard et al., 1990
), and egg extract can assemble centrosomes (Félix et al., 1994
; Stearns and Kirschner, 1994
) and duplicate centrioles (Hinchcliffe et al., 1999
; Lacey et al., 1999
). To study the function of Plk4 in centriole assembly, we made use of the ability of egg extracts to translate input mRNA (Murray and Kirschner, 1989
). mRNA encoding the Xenopus
Plk4 orthologue, Plx4, was added to translation-competent interphase Xenopus
egg extract. A polyclonal antibody raised against Plx4 recognized the translated protein, whereas endogenous Plx4 was only detectable when immunoprecipitated from a larger volume of egg extract ().
Figure 1. Plx4 overexpression drives centriole amplification in Xenopus egg extract. (A) Western blot of Xenopus egg extracts after addition of Plx4 mRNA or immunoprecipitation (IP) of endogenous Plx4 using anti-Plx4 or nonimmune IgG as a control. Immunoblotted (more ...)
We first tested whether translated Plx4 could stimulate the formation of multiple centrioles around sperm centrioles added to the extract. Addition of Plx4 mRNA and sperm centrioles together to egg extract resulted in the formation of centrosomes bearing multiple centrin foci surrounding a single bright centrin focus, presumably representing the sperm centriole, within 2 h at 16°C (). We next tested whether translated Plx4 could stimulate de novo centriole formation. Addition of Plx4 mRNA alone to egg extract resulted in the formation of many centriole-like structures within 2 h at 16°C at a mean density of 2 × 104
/µl extract (). We judged these structures to be centrioles based on three criteria: (1) they were able to organize centrosomes, as demonstrated by their ability to form microtubule asters in mitotic extract (Fig. S1 A
), (2) they were labeled by antibodies against centrosomal markers including γ-tubulin, Plx4, and acetylated α-tubulin (Fig. S1 B), and (3) they were able to serve as sites for assembly of new centrin foci after extended incubation in interphase extract (Fig. S1 C). Thus, overexpression of Plx4 in Xenopus
extracts promotes de novo centriole formation similar to that observed in Drosophila
eggs (Peel et al., 2007
; Rodrigues-Martins et al., 2007
Based on the ability of Plx4 to generate centrioles in Xenopus
egg extract, we sought to identify Plx4-interacting proteins from this source. For this purpose, we used a kinase-dead form of Plx4 (Plx4-D154A), as this protein might have a more sustained interaction with binding partners. Plx4-D154A was translated in egg extract and purified by affinity chromatography, and associated proteins were identified by mass spectrometry. Seven Xenopus
proteins were identified: Cep152, Brg1, Pbrm1, Wdr33, Cad, Chd1, and Atxn2 (Fig. S1 D). Cep152 was identified by three unique peptides and was chosen for further analysis because it was previously associated with centrosome structure and function (Andersen et al., 2003
; Varmark et al., 2007
; Blachon et al., 2008
; Dobbelaere et al., 2008
We examined the localization of endogenous Cep152 and a GFP-Cep152 fusion protein in human RPE-1 cells. Antibody against endogenous Cep152 revealed that it localized to the centrosome throughout the cell cycle (). A similar distribution was observed for transiently expressed GFP-Cep152 (unpublished data). In some cases, the apparent amount of Cep152 on the two centrosomes differed (); however, this was not correlated with centriole age or presence of a primary cilium, and likely represents experimental variation. Cep152 localization in G1 phase centrosomes of U2OS cells was examined by deconvolution microscopy and found to be distinct from that of centrin, a marker of the distal end of the centriole, and γ-tubulin, a marker of the pericentriolar material (). In duplicated centriole pairs, Cep152 signal partially overlapped with that of Sas6 (), which localizes to the proximal end of new centrioles (Kleylein-Sohn et al., 2007
; Strnad et al., 2007
). The observed pattern of localization is most consistent with Cep152 being concentrated at the proximal end of the mother centriole.
Figure 2. Cep152 localizes to centrioles throughout the cell cycle. (A) RPE-1 cells at the indicated cell cycle stages were fixed and stained for Cep152, centrin, or polyglutamylated (poly-glu) tubulin, which marks centrioles and primary cilia, and DNA. Insets (more ...)
If Plk4 and Cep152 function together in centriole duplication, we would expect them to have a similar evolutionary distribution. Several of the core centriole components, including Sas6, CPAP/Sas-4, and Cep135/Bld10, are widely conserved structurally and functionally in organisms with centrioles (Leidel and Gönczy, 2003
; Dammermann et al., 2004
; Matsuura et al., 2004
; Leidel et al., 2005
; Basto et al., 2006
; Kleylein-Sohn et al., 2007
; Carvalho-Santos et al., 2010
; Hodges et al., 2010
). In contrast, Plk4 has a more limited distribution, being present in chordates, insects, and a few other invertebrate groups, but absent from C. elegans
and most ciliated single-celled eukaryotes (Carvalho-Santos et al., 2010
; Hodges et al., 2010
). By searching for reciprocal BLAST hits, we found that Cep152 has the same distribution as Plk4, including absence from C. elegans
, which is similar to what was reported for the Drosophila
orthologue Asterless (Hodges et al., 2010
). We used multiple sequence alignment of vertebrate and invertebrate orthologues of Cep152 with Drosophila
Asterless to identify two conserved regions in Cep152, CR1 and CR2 (Fig. S2 A
). CR1 is found in Cep152 in vertebrates and invertebrates, whereas CR2 is only present in vertebrates. In addition to these two regions, Cep152 orthologues share homology in their two structural maintenance of chromosomes (SMC)–like coiled-coil domains (, SMC-A and SMC-B).
Figure 3. Cep152 and Plk4 interaction and localization. (A) Schematic of Cep152 full length (FL) and deletion constructs showing the ability to interact with Plk4 and Cep152 and to localize to the centrosome (Cent). Interactions were determined by coimmunoprecipitation (more ...)
To characterize the interaction of Cep152 with Plk4, we first tested whether the interaction could be observed in human cells as in frog egg extracts. Epitope-tagged human Plk4 and Cep152 expressed in human HEK293T cells were able to interact with the other in reciprocal immunoprecipitation experiments (Fig. S2 B). By expressing deletion constructs bearing the indicated parts of Cep152 and Plk4 as epitope-tagged proteins in HEK293T cells (), we found that the first 217 residues of Cep152, which includes CR1, are necessary and sufficient to bind Plk4 () and that the crypto Polo-box region of Plk4 is necessary and sufficient to bind Cep152 (). Using two differently tagged versions of Cep152, we found that Cep152 is able to interact with itself and that at least one of the SMC-like domains is necessary for interaction with full-length Cep152 (). Lastly, we found that a region including CR2 was necessary and sufficient for localization of a GFP fusion protein to the centrosome ( and S2 C). To determine whether the observed interaction of Plk4 and Cep152 is also likely to be occurring at the centrosome, we examined their localization in U2OS cells. Similar, overlapping patterns of localization were observed for both epitope-tagged versions of Plk4 and Cep152 and the endogenous proteins ().
We next tested whether Cep152, like Plk4, is required for centriole duplication in mammalian cells. U2OS cells were transfected with siRNAs against Cep152, and Cep152 protein depletion was apparent starting at 48 h after transfection, reaching maximum extent at 72 h (). Centriole number was determined by centrin labeling in mitotic cells that lacked Cep152 labeling. In contrast to cells transfected with control siRNAs, we observed a stepwise decrease in centriole number over time in Cep152-depleted cells, which is consistent with a defect in centriole duplication followed by segregation in mitosis (). In cells that had two or more centrioles 96 h after RNAi treatment, these centrioles were single centrioles rather than pairs, suggesting that they too resulted from failure of centriole duplication followed by a failure of segregation at mitosis (). Similar results were observed with Cep152 depletion in HeLa cells (unpublished data). Both bipolar and monopolar spindles were observed in cells with a single centriole after Cep152 depletion (unpublished data), which is similar to what is observed upon depletion of Plk4 (Habedanck et al., 2005
). The centriole duplication phenotype could be rescued by expressing RNAi-resistant GFP-Cep152, but not by GFP, indicating that the phenotype is specific to Cep152 depletion (Fig. S3
Figure 4. Cep152 is required for centriole duplication and amplification. (A) Western blotting of U2OS cell lysates 72 h after transfection with control (Ctl) siRNAs or Cep152 siRNAs immunoblotted for Cep152 or α-tubulin as a loading control. The indicated (more ...)
We next tested whether Cep152 is also required for Plk4-induced centriole amplification. Plk4 overexpression was induced in RPE-1 cells that had been depleted of Cep152, and centriole number was determined by centrin labeling. Cep152 depletion was less efficient in these cells compared with U2OS cells ( and ), and centriole duplication occurred normally, allowing us to specifically assay the requirement of Cep152 in centriole amplification. 18 h after induction of Plk4 overexpression, 99% of control mitotic cells had the expected amplified centriole “rosettes” at the spindle poles (Habedanck et al., 2005
), whereas 64% of Plk4-overexpressing, Cep152-depleted cells had only two centrioles at each pole (). These results indicate that Cep152 is required both for normal centriole duplication and for Plk4-induced centriole amplification.
Figure 5. Functional interaction of Cep152 and Plk4. (A) Plk4 is lost from the centrosome in cells expressing GFP-Cep152 (1–217). U2OS cells transfected with GFP or GFP-Cep152 (1–217) were fixed 24 h after transfection and labeled with antibodies (more ...)
Based on the aforementioned results, we envisioned several possibilities for the functional relationship between Cep152 and Plk4. The first is that one protein serves to localize the other to the centrosome and that localization is required for centriole duplication. Plk4 has been shown to localize to the centrosome via the crypto Polo-box domain (Habedanck et al., 2005
). Because Cep152 binds the Plk4 crypto Polo-box (), expression of a Cep152 mutant containing the Plk4-binding domain but lacking the centrosome localization domain (Cep152 [1–217]) might mislocalize Plk4. Consistent with this, we found that expression of Cep152 (1–217) resulted in loss of centrosomal Plk4 labeling () and subsequent progressive loss of centrioles, which is similar to that observed for Cep152 depletion (not depicted). Next, we directly tested whether Cep152 is required for Plk4 localization in cells depleted of Cep152. Plk4 localization to the centrosome was unchanged in U2OS cells depleted of Cep152; similarly, Cep152 localization to the centrosome was unchanged in cells depleted of Plk4 (). These results suggest that a Cep152 fragment that binds Plk4 is sufficient to outcompete one or more other factors that localize Plk4 to the centrosome but that neither protein is necessary for localization of the other.
Plk4 is required for the localization of Sas6 and other centriole assembly proteins during centriole assembly (Kleylein-Sohn et al., 2007
; Strnad et al., 2007
). To determine whether Cep152 is similarly required for early centriole assembly, we assayed Sas6 localization in cells depleted of Cep152 (). Control U2OS cells had one focus of Sas6 at each centriole doublet in S phase–arrested and mitotic cells (, S phase cell). In contrast, Cep152-depleted cells that had failed to duplicate centrioles lacked Sas6 on mitotic centrioles and had either no Sas6 (65%; n
= 150 cells) or reduced Sas6 (25%) on S phase centrioles (). We observed the same phenotype for Plk4 depletion (unpublished data), which is similar to the previous description of the relationship between Plk4 and Sas6 localization (Kleylein-Sohn et al., 2007
; Strnad et al., 2007
). These results suggest that both Cep152 and Plk4 act early in the centriole assembly pathway.
We next investigated whether Plk4 can phosphorylate Cep152. GFP-tagged Cep152 isolated from human cell lysates was incubated with purified wild-type GST-Plk4 (Plk4-WT) or the kinase-dead mutant GST-Plk4 D154A (Plk4-KD) and assayed for 32
P incorporation (). Plk4-WT but not Plk4-KD was able to phosphorylate itself, as previously reported (Holland et al., 2010
; Sillibourne et al., 2010
). Both full-length GFP-Cep152 (, left) and the GFP-Cep152 (1–217) fragment that binds to Plk4 (, middle) were phosphorylated when incubated with Plk4-WT but not with Plk4-KD. To test whether this reflected direct phosphorylation of Cep152 by Plk4, we purified Cep152 (1–217) fused to maltose-binding protein (MBP) and incubated it with GST-Plk4. MBP-Cep152 (1–217) was phosphorylated by Plk4-WT but not by Plk4-KD (, right). In all reactions, the phosphorylation was specific to the Cep152 portion of the fusion proteins, as neither GFP nor MBP alone were phosphorylated (unpublished data). Thus, Plk4 can interact directly with Cep152 and phosphorylate it in vitro. In addition, at least one phosphorylated region corresponds to the N-terminal Cep152 fragment that is required for interaction with Plk4.
We propose that Plk4 and Cep152 act together early in centriole duplication. We have shown that Cep152 is a centriolar protein that interacts with Plk4 and can be phosphorylated by it in vitro. Cep152 is required for centriole duplication, Plk4 overexpression–induced centriole amplification and, as for Plk4, for localization of Sas6 to centrioles, which is an early step in their assembly. The hypothesis that Plk4 and Cep152 act together in centriole duplication is also reinforced by their pattern of evolutionary conservation. The identification of Cep152 as a Plk4-associated factor provides a new tool for understanding the function of Plk4 in centriole duplication and how this process is spatially and temporally regulated.