Plk4 trans-autophosphorylation regulates centriole number by controlling βTrCP-mediated degradation

Centrioles are the main constituents of the mammalian centrosome and act as basal bodies for ciliogenesis. Centrosomes organize the cytoplasmic microtubule network during interphase and the mitotic spindle during mitosis, and aberrations in centrosome number have been implicated in chromosomal instability and tumor formation. The centriolar protein Polo-like kinase 4 (Plk4) is a key regulator of centriole biogenesis and is crucial for maintaining constant centriole number, but the mechanisms regulating its activity and expression are only beginning to emerge. Here, we show that human Plk4 is subject to βTrCP-dependent proteasomal degradation, indicating that this pathway is conserved from Drosophila to human. Unexpectedly, we found that stable overexpression of kinase-dead Plk4 leads to centriole overduplication. This phenotype depends on the presence of endogenous wild-type Plk4. Our data indicate that centriole overduplication results from disruption of Plk4 trans-autophosphorylation by kinase-dead Plk4, which then shields endogenous Plk4 from recognition by βTrCP. We conclude that active Plk4 promotes its own degradation by catalyzing βTrCP binding through trans-autophosphorylation (phosphorylation by the other kinase in the dimer) within homodimers.

While this work was in progress, Drosophila Plk4 was shown to be degraded in an SCF Slimb/TrCP -dependent manner (Cunha-Ferreira et al., 2009;Rogers et al., 2009). Consequently, depletion of the SCF ubiquitin ligase Slimb (mammalian TrCP) led to stabilization of Plk4 and to centriole overduplication (Cunha-Ferreira et al., 2009;Rogers et al., 2009). As expected for a proteasome-dependent degradation mechanism, human Plk4 protein levels also increased upon proteasome inhibition with 1 M MG132 for 16 hours (supplementary material Fig. S4) and similar results were independently reported by others (Holland et al., 2010;Sillibourne et al., 2010). This prompted us to speculate that Plk4-KD might cause centriole overduplication by interfering with the TrCP-mediated degradation of endogenous Plk4. To explore this notion, we first investigated whether human Plk4 protein levels are also controlled by TrCP. Asynchronously growing U2OS cells were depleted of TrCP by siRNA transfection and centriole numbers monitored by immunofluorescence microscopy. Upon depletion of TrCP, Plk4 protein levels at the centrosome increased about sevenfold compared with those of control cells ( Fig. 2A,B). Moreover, TrCP-depleted cells exhibited centriole overduplication, partially in a rosette-like arrangement of procentrioles, reminiscent of Plk4 overexpression in human cells (Kleylein-Sohn et al., 2007) and earlier work in Drosophila (Cunha-Ferreira et al., 2009;Rogers et al., 2009). To directly demonstrate a role of Plk4 in the observed phenotype, we analyzed the effects of TrCP depletion in the absence of Plk4. Whereas 48% of TrCP-depleted control cells exhibited overduplicated centrioles, virtually no centriole overduplication was observed after co-depletion of TrCP and Plk4, similar to results observed after depletion of Plk4 alone ( Fig.  2A,C). Instead, these latter treatments increased the proportion of :myc-Plk4-KD cells were arrested with aphidicolin for 24 hours before expression of myc-Plk4-WT or myc-Plk4-KD was induced for 16 hours. No tetracycline was added to controls. Cells were fixed and stained with antibodies for the myc epitope (green), CP110 (red) and Cep135 (blue). (B)U2OS:myc-Plk4-WT or U2OS:myc-Plk4-KD cells were transfected for 24 hours with siRNA oligonucleotides targeting GL2, the 3Ј-UTR of Plk4 or human Sas-6 prior to induction of Plk4 expression (myc-Plk4-WT or myc-Plk4-KD) for 16 hours. Cells were stained for the myc epitope (green), CP110 (red) and Cep135 (blue). (C)Percentage of cells, treated as described in B, that exhibited centriole overduplication. Data from three independent experiments (n100) are shown. Error bars denote s.e.m. Scale bars: 1m. cells with fewer than two centrioles to 67% and 73%, respectively (Fig. 2C). Hence, the centriole-overduplication phenotype produced by depletion of TrCP clearly requires Plk4. To demonstrate that TrCP modulates levels of Plk4 protein, we depleted TrCP for 72 hours before inducing expression of myc-Plk4-WT for the last 24 hours of siRNA treatment. Compared with cells treated with control siRNA duplexes (siGL2), depletion of TrCP led to a 1.5-fold increase in Plk4-WT protein (Fig. 2D). Also, Plk4 siRNA treatment (carried out for control) abolished Plk4 expression, as expected (Fig. 2D). Conversely, coexpression of TrCP and Plk4-WT in 293T cells led to a decrease in Plk4 protein (Fig. 2E). Together, the above data demonstrate that TrCP modulates Plk4 protein levels in human cells and thus contributes to the maintenance of correct centrosome number. This confirms and extends earlier work in Drosophila (Cunha-Ferreira et al., 2009;Rogers et al., 2009) and shows that the TrCP-Plk4 pathway is conserved in Drosophila and mammals (see also Guardavaccaro et al., 2003;Holland et al., 2010;Sillibourne et al., 2010). Yet another recent study also demonstrates centriole overduplication in U2OS cells upon depletion of the SCF component Cul1, although a role for TrCP was not emphasized (Korzeniewski et al., 2009).
To further explore our proposition that Plk4-KD might cause centriole overduplication by interfering with the degradation of endogenous (active) Plk4, we next investigated whether Plk4-KD is able to bind to TrCP. Usually, TrCP binds its substrates via a DSGxx[S/T] motif (DSG motif) in the substrate protein and this interaction is thought to be regulated by phosphorylation of two phospho-acceptor sites (S/T) within this so-called phosphodegron (Nakayama and Nakayama, 2006). Human Plk4 carries an evolutionarily conserved DSG motif spanning residues 284 to 289 (DSGHAT). Indeed, an interaction between human Plk4-WT and TrCP could readily be demonstrated by co-immunoprecipitation and, as predicted, this interaction required an intact DSG motif (  in this context. Importantly, under the exact same experimental conditions, Plk4-KD did not interact with TrCP, strongly suggesting that Plk4 activity is required for this interaction. Confirming this conclusion, the Plk4-TrCP complex could be disrupted by -phosphatase (PPase) treatment (Fig. 3B). Furthermore, Plk4-KD was ubiquitylated less efficiently than Plk4-WT and in this regard resembled Plk4-WT-DSG AA (Fig. 3C). Consistent results were obtained in vivo (Fig. 3C) and in vitro (supplementary material Fig. S6), arguing against co-precipitation of other ubiquitylated proteins with Plk4-WT. One would expect that lack of ubiquitylation should stabilize Plk4 by protecting it from degradation via the 26S proteasome. Indeed, whereas Plk4-WT was degraded in cells treated with cycloheximide, Plk4-KD was stabilized to a similar extent as was Plk4-WT-DSG AA (Fig.  3D). Together, these data suggest that Plk4 kinase activity is necessary for its interaction with TrCP and, consequently, its polyubiquitylation and subsequent degradation. Overall, these results are in good agreement with two recent studies that independently demonstrate a role of Plk4 autophosphorylation in controlling Plk4 stability (Holland et al., 2010;Sillibourne et al., 2010). Interestingly, Holland and co-workers observed only partial stabilization of a DSG-phosphodegron mutant, whereas more extensive stabilization was observed for a version of Plk4 that lacked a stretch of 24 residues comprising this motif (Holland et al., 2010). This observation prompted the authors to propose that a second, TrCP-independent, pathway might also contribute to the regulation of Plk4 stability (Holland et al., 2010).
The finding that Plk4-KD cannot interact with TrCP argues against the possibility that excess Plk4-KD causes centriole overduplication through sequestration of TrCP. This led us to explore an alternative model involving dimerization of Plk4. As shown previously, Plk4 dimerizes via its C-terminal coiled-coil region (Leung et al., 2002;Habedanck et al., 2005). This dimerization is independent of kinase activity, as confirmed here by co-immunoprecipitation experiments (supplementary material Fig. S7). Several Plk4 fragments differing in their ability to dimerize were overexpressed in U2OS cells and assayed for their ability to trigger centriole overduplication. As shown in supplementary material Figs S8-S10, Plk4 1-608 is clearly active as a kinase and interacts with TrCP but does not dimerize owing to truncation of its C terminus, whereas Plk4 609-970 is able to dimerize with Plk4-WT but does not interact with TrCP owing to truncation of the kinase domain. Remarkably, Plk4 609-970 caused strong centriole overduplication, occasionally resulting in a rosette-like arrangement of procentrioles, whereas Plk4 1-608 failed to do so (Fig. 4A). This reinforces the view that excess Plk4-KD is able to cause centriole overduplication, provided that its ability to dimerize with endogenous Plk4 is preserved.
The above data led us to conclude that excess Plk4-KD triggers centriole overduplication by virtue of its ability to (hetero-)dimerize with endogenous, active Plk4. If this is the case, the Plk4-KD polypeptide could potentially be phosphorylated in trans by the Plk4-WT polypeptide (but not vice versa), and phosphorylated Plk4-KD could then sequester SCF TrCP by acting as a decoy. A corollary of this model is that autophosphorylation in trans should convert Plk4-KD to a TrCP-binding species. To test this prediction, we expressed various combinations of myc-or FLAG-tagged Plk4 proteins differing in their activity status (WT or KD) and/or ability to be recognized by TrCP (DSG-WT or DSG AA ). In these experiments, the myc-tagged constructs served as bait for TrCP binding, whereas the FLAG-tagged constructs, competent to dimerize with Plk4 but incompetent to bind TrCP, provided kinase activity. The ability of the immunoprecipitated complexes to bind to TrCP was then analyzed via an in vitro binding assay. Coexpression of FLAG-Plk4-KD-DSG AA with myc-Plk4-KD failed to restore TrCP binding, as expected, considering the Anti-myc immunoprecipitations were performed and immunoprecipitates treated with -phosphatase (PPase) where indicated. The co-immunoprecipitated proteins were detected by immunoblotting. (C)Myc-Plk4-WT, myc-Plk4-KD or myc-Plk4-DSG AA were coexpressed for 24 hours with HA vector or HA-ubiquitin. Cell extracts were subjected to anti-myc immunoprecipitations and probed by immunoblotting for the indicated proteins. (D)FLAG-Plk4-WT, FLAG-Plk4-KD and FLAG-WT-DSG AA were expressed in 293T cells before protein synthesis was blocked by cycloheximide. Cells were harvested at the indicated time points (hours) and protein levels analyzed by immunoblotting.
absence of trans-autophosphorylation. By stark contrast, coexpression of FLAG-Plk4-WT-DSG AA with myc-Plk4-KD fully restored the binding of myc-Plk4-KD to TrCP (Fig. 4B). This demonstrates that autophosphorylation in trans is required to confer TrCP-binding properties to Plk4. In excellent agreement with this conclusion, Plk4-WT was independently shown to promote destruction of Plk4-KD through intermolecular phosphorylation (Holland et al., 2010).
Whether autophosphorylation is not just required, but is sufficient for Plk4-TrCP binding is not presently known. A priori, it is possible that Plk4 trans-autophosphorylation directly activates the phosphodegron for TrCP binding (Fig. 4C, model I). Alternatively, Plk4 might autophosphorylate in trans on sites distinct from the phosphodegron that then serve to recruit a different kinase X, which in turn phosphorylates Plk4 on the phosphodegron or in close proximity to this motif (Fig. 4C, model II). In support of this latter possibility, we emphasize that degradation of several TrCP targets, e.g. -catenin (Liu et al., 2002), Wee1 (Watanabe et al., 2004) and Erp1 (Liu and Maller, 2005;Rauh et al., 2005;Hansen et al., 2006), involves recruitment of phosphodegron-directed kinases through phosphorylation-dependent docking sites.
In conclusion, our study shows that autophosphorylation controls TrCP-mediated degradation of Plk4. In line with observations on the activation-dependent degradation of other protein kinases (Kang et al., 2000;Lu and Hunter, 2009), we propose that active Plk4 catalyzes its own degradation and that this provides a tight coupling between activity status and protein abundance. We further show that Plk4 degradation involves autophosphorylation in trans, and this provides a rational for the observation that excess Plk4-KD can trigger centriole overduplication through a mechanism requiring endogenous, active Plk4. Our data provide not only important mechanistic insight into the regulation of Plk4, but also raise interesting new questions. Most importantly, future research should aim at exploring the timing of Plk4 degradation during the cell cycle and the identification of a putative kinase X that is proposed here to contribute to control Plk4 stability.

Plasmids and antibodies
Cloning of Plk4 and TrCP1 cDNA has been described previously (Habedanck et al., 2005;Chan et al., 2008). Sequence mutations in Plk4 were inserted by using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions using the following primers: Plk4 S285A/T289A 5Ј-GAAGACTCAA -2167 Plk4 autophosphorylation and stability myc-Plk4 1-608 or myc-Plk4 609-970 for 48 hours. Cells were stained for the myc epitope (green), CP110 (red) and Cep135 (blue). Scale bar: 1m. (B)293T cells were transfected with the indicated plasmids. Anti-myc immunoprecipitates were incubated with in-vitrotranslated [ 35 S]-methionine-labeled HA-TrCP in an in vitro binding assay. The co-immunoprecipitated proteins were analyzed by immunoblotting and autoradiography. (C)Two schematic models. According to model I, Plk4 autophosphorylation directly phosphorylates the DSG motif in trans and this is sufficient for TrCP binding. Alternatively (model II), Plk4 autophosphorylation in trans creates a docking site for an unknown kinase X, which then phosphorylates the DSG motif. In both cases, phosphorylation of the DSG motif is proposed to initiate the degradation of Plk4. Ubi, ubiquitin. TTGATGCTGGGCATGCCGCAATTTCTACTGC-3Ј; Plk4 S285D/T289D 5Ј-GAA-GACTCAATTGATGACGGGCATGCCGACATTTCTACTGC-3Ј. HA-ubiquitin was generously provided by Stefan Müller (Max-Planck Institute of Biochemistry, Martinsried, Germany).

Cell culture and transfections
Transient transfections of 293T cells were performed using TransIT-LT1 transfection reagent (Mirus Bio) according to the manufacturer's protocol.
To assay protein-degradation kinetics, translation was inhibited by the addition of 25 g/ml cycloheximide for the indicated time.
For immunoprecipitations, the extracts were incubated with protein-G beads (GE Healthcare) and 10 g of the appropriate antibodies for 1.5 hours at 4°C. Immunocomplexes bound to beads were washed three times with wash buffer (lysis buffer with 300 mM NaCl). Bound proteins were eluted by boiling in 2ϫ SDS sample buffer, resolved by SDS-PAGE and analyzed by immunoblotting.
For in vitro binding assays, the washed immunocomplexes were suspended in lysis buffer and incubated for 1.5 hours at 4°C with HA-TrCP, which had been in vitro translated using the TNT-T7 quick coupled transcription/translation system (Promega) with [ 35 S]-methionine according to the manufacturer's protocol. After washing three times with wash buffer, the bound proteins were eluted by boiling in 2ϫ SDS sample buffer, resolved by SDS-PAGE, and analyzed by immunoblotting and autoradiography.
In vitro ubiquitylation of in-vitro-translated [ 35 S]-methionine-labeled Plk4 was carried out using a HeLa-lysate-based ubiquitin-conjugation kit (Enzo Life Sciences) according to the manufacturer's protocol. Conjugation was visualized by immunoblotting and autoradiography.
In vitro kinase assays using immunoprecipitated Plk4 were carried out at 30°C in kinase buffer (50 mM HEPES, pH 7.0, 100 mM NaCl, 10 mM MgCl 2 , 5% glycerol, 1 mM DTT). Reactions were stopped after 30 minutes by addition of sample buffer. Samples were then analyzed by immunoblotting and autoradiography.

Microscopic techniques
Cells were fixed in methanol for 5 minutes at -20°C. Antibody incubations and washings were performed as described previously (Meraldi et al., 1999). Stainings were analyzed using a DeltaVision microscope on a Nikon TE200 base (Applied Precision), equipped with an APOPLAN 100ϫ/1.4 N.A. oil-immersion objective. Serial optical sections obtained 0.2-m apart along the z-axis were processed using a deconvolution algorithm and projected into one picture using Softworx. For quantitation of Plk4 levels at the centrosome with ImageJ, z-stacks from control and treated samples were acquired with the same exposure and maximum-intensity projections were carried out. Background signal intensity was subtracted from Plk4 signal intensity.