Phase separation of Polo-like kinase 4 by autoactivation and clustering drives centriole biogenesis

Abstract

Tight control of centriole duplication is critical for normal chromosome segregation and the maintenance of genomic stability. Polo-like kinase 4 (Plk4) is a key regulator of centriole biogenesis. How Plk4 dynamically promotes its symmetry-breaking relocalization and achieves its procentriole-assembly state remains unknown. Here we show that Plk4 is a unique kinase that utilizes its autophosphorylated noncatalytic cryptic polo-box (CPB) to phase separate and generate a nanoscale spherical condensate. Analyses of the crystal structure of a phospho-mimicking, condensation-proficient CPB mutant reveal that a disordered loop at the CPB PB2-tip region is critically required for Plk4 to generate condensates and induce procentriole assembly. CPB phosphorylation also promotes Plk4's dissociation from the Cep152 tether while binding to downstream STIL, thus allowing Plk4 condensate to serve as an assembling body for centriole biogenesis. This study uncovers the mechanism underlying Plk4 activation and may offer strategies for anti-Plk4 intervention against genomic instability and cancer.

Fig. 1

Plk4 triggers its symmetry-breaking ring-state-to-dot-state relocalization by autophosphorylating its CPB. a Schematic diagram showing the secondary structure of the Plk4 CTD. Numbers indicate amino acid residues. The positions of PC1 to PC8 are marked. b Multiple sequence alignment for the region containing PC3 was performed using the Clustal Omega software. The S698, S700, T704, and T707 residues phosphorylated in vivo are indicated. c 3D-SIM analysis of immunostained U2OS cells stably expressing the indicated EGFP-Plk4 constructs and silenced for endogenous Plk4 (siPlk4). The schematic diagram (right) illustrates multiple Cep152 ring (red), Plk4 dot (green), and Sas6 (black) signals. Bar, 0.5 μm. Quantification of images is shown in mean ± s.d. (n = 3 independent experiments). Note that like the catalytically inactive Plk4 KM mutant, the pc3 mutant remained at a ring state. **** P < 0.0001 (unpaired two-tailed t test). d Confocal analysis of immunostained U2OS cells stably expressing various EGFP-Plk4 constructs under Plk4 RNAi conditions (siPlk4). Bar, 20 μm. Dotted boxes, areas of image enlargement (bar, 3 μm). Quantified data are shown in mean ± s.d. (n = 3 independent experiments). Immunoblotting shows the expression of each construct. ***P < 0.001; ****P < 0.0001 (unpaired two-tailed t test). Asterisk denotes a cross-reacting protein. CBB, Coomassie Brilliant Blue-stained membrane. Source data are provided as a Source Data file for c, d

Fig. 2

A condensation-proficient Plk4 CP mutant drives procentriole formation by switching its interaction from Cep152 to STIL. a, b 3D-SIM analysis of immunostained U2OS cells stably expressing the indicated EGFP-Plk4 constructs under Plk4 RNAi (siPlk4) conditions. Bar, 1 μm. Quantification of images is shown in mean ± s.d. (n = 3 independent experiments). Arrowheads in a, dot-state Plk4 signals; schematic diagram in a shows the sites of the KM and the quadruple CP mutations. Elongated procentrioles in both centrosome (cent.) and cytosol (cyto.; as judged by the absence of the centrosomal Cep152 signal) are shown in b. ***P < 0.001; ****P < 0.0001 (unpaired two-tailed t test). c 3D-SIM analysis of immunostained U2OS cells stably expressing the EGFP-Plk4 CP mutant under various RNAi conditions. Bar, 1 μm. Quantified data are shown in mean ± s.d. (n = 3 independent experiments). ****P < 0.0001 (unpaired two-tailed t test). d–f Immunoprecipitation (IP) and immunoblotting analyses using HEK293T cells cotransfected with the indicated constructs. IP samples were then treated with λ phosphatase (PPase), where indicated, to convert phosphorylated, slow-migrating forms into a fast-migrating form, and then separated by 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) for immunoblotting. Numbers denote relative signal intensities. Asterisk denotes remnants of STIL signals from a prior anti-GFP blotting. Source data are provided as a Source Data file for  a–c

Fig. 3

A condensation-proficient Plk4 CP mutant generates nanoscale, electron-dense bodies and localizes along an elongated procentriole. a Thin-section transmission electron microscopy (TEM) showing U2OS cells stably expressing EGFP-Plk4 CP stained with an anti-GFP and a 6-nm nanogold-conjugated secondary antibody. Both centrosomal and cytosolic Plk4 condensates decorated with nanogold signals are shown. Bar, 0.2 μm. b Correlative light and electron microscopy (CLEM) analysis of U2OS cells stably expressing EGFP-Plk4 CP. Electron micrographs overlaid with GFP fluorescent 3D-SIM images are shown at the bottom. Bar, 2 μm. Dotted boxes, areas of enlargement. Bar, 0.5 μm. The dimension of an elongated electron-dense body induced by Plk4 CP is shown. c Two 3D-SIM images showing tubular EGFP-Plk4 signals observed in U2OS cells. Plk4 and Sas6 signals were detected along the length of elongated procentrioles. The images were subjected to 3D surface rendering (see Supplemental Movie S1). Bar, 1 μm. Dotted lines traversing the elongated procentrioles mark the region where the relative fluorescence intensities for Plk4 and Sas6 were measured (middle). Bars in the graph (right), an average peak-to-peak distance across the tubular EGFP-Plk4 signals (n = 13) with s.d. Source data are provided as a Source Data file for c

Fig. 4

Plk4 containing the phospho-PC3 motif is stable and colocalizes with STIL and Sas6 along the length of an elongated procentriole. a Immunoblotting analyses of U2OS cells stably expressing the indicated constructs treated with siRNA (siGL or siβTrCP) or cycloheximide (CHX). Total lysates were treated with λ phosphatase (PPase) to convert all phosphorylated, slow-migrating Plk4 forms into a fast-migrating form (arrows). Numbers, relative signal intensities. Quantification of relative signal intensities is shown in mean ± s.d. (n = 3 independent experiments). b 3D-SIM analysis and quantification of immunostained HEK293T cells stably expressing the indicated EGFP-Plk4 constructs and transfected with Flag-βTrCP. Bar, 1 μm. Enlarged images (bar, 0.2 μm) for dotted boxes are shown. The fraction of cells with or without colocalized Flag-βTrCP signals was quantified (right) (mean ± s.d., n = 3 independent experiments). ****P < 0.0001 (unpaired two-tailed t test). Note that elongated (inside dotted boxes) and dot-state (arrowheads) Plk4 signals are colocalized with STIL, but not with βTrCP, while the ring-state Plk4 (yellow arrow) is colocalized with Cep152 and βTrCP. c 3D-SIM analysis of U2OS cells immunostained with anti-Cep152, anti-Plk4, and Alexa Fluor 594-conjugated anti-Plk4 pSSTT antibodies. Images representing different stages from G1 to S/G2 are shown. Bar, 0.5 μm. See the confocal images in Supplementary Fig. 4e that demonstrate the specificity of the Alexa Fluor 594-conjugated Plk4 pSSTT antibody. Note that the Plk4 pSSTT epitope preferentially associates with the dot state (arrows) and elongated (barbed arrow) Plk4 signals rarely observed at native centrosomes. d Quantification of the relative intensities of Plk4 pSSTT, STIL, STIL pS1108, and Sas6 signals was carried out using confocal images from immunostained U2OS cells. Relative fluorescence intensities (mean ± s.d.) are shown. n = 90 (WT), 90 (KM), and 90 (CP) for Plk4 pSSTT/Plk4, n = 90 (WT), 91 (KM), and 97 (CP) for STIL/Plk4, n = 132 (WT), 103 (KM), and 142 (CP) for STIL pS1108/Plk4, and n = 90 (WT), 91 (KM), and 97 (CP) for Sas6/Plk4 with each sample pooled from three independent experiments. Numbers denote mean relative intensities. Representative images are shown in Supplementary Fig. 4f–h. ****P < 0.0001 (unpaired two-tailed t test). Numbers denote mean values. e 3D-SIM analysis of immunostained U2OS cells stably expressing EGFP-Plk4 CP. Bars, 5 μm. Enlarged images (green bars, 1 μm) for dotted boxes are shown. Quantification of images is shown in mean ± s.d. (n = 3 independent experiments). Source data are provided as a Source Data file for a, b, d, e

Fig. 5

Altered surface hydrophobicity and conformational nature of condensation-proficient Plk4 CP and CP_v1. a Fluorescence spectra of ANS in a buffer containing the indicated concentrations of CPB, CPB CP, or CPB CP_v1 at 4 or 20 °C. b Overall structures of the CPB CP_v1 and apo-CPB (4N9J) overlaid for comparison. Each subunit of a homodimeric CPB consists of PB1 and PB2 folds. Side views show the magnitude of β-sheet twisting along the PB2–PB2 axis (below). c Detailed structural comparison showing that the PB2-tip region of CPB CP_v1 is angularly distorted (left) and appears flexible with several disordered hydrophobic residues (right). Note that Y750, L752, V758, and L761 residues are either absent or devoid of their side chains. d Differences in the mode of interactions for PC3 WT and CP mutant residues. Important interactions (H-bond and/or salt bridge) are depicted as dashed lines (see text for details). e A hypothetical model illustrating that Cep152 60-mer residues (cyan) engaged in interactions with CPB residues (gray) fail to establish analogous interactions with CPB CP (brown), thus explaining why the CP mutations disrupted the Cep152-Plk4 interaction in Fig. 2d. Dotted lines indicate important interactions detected between Cep152 60-mer and CPB WT (4N7V). Source data are provided as a Source Data file for a

Fig. 6

Autophosphorylated CPB is sufficient to generate spherical condensates in a concentration-dependent manner. a Condensation profiles of CPB CP at various concentrations as measured by OD_{350 nm} at 20 °C. Different concentrations of soluble CPB CP were prepared at 4 °C. Upon shifting the temperature to 20 °C, optical density was measured every 3 s. Each point represents the average of two replicates ± s.d. b 3D-SIM analysis showing the formation of spherical CPB condensates at various concentrations in vitro. Condensates formed at 20 °C were decorated with FITC for imaging (see Methods). Black bar, 2 μm. Dotted boxes, areas of enlargement (bar, 1 μm). A 3D surface-rendered spherical condensate (bar, 1 μm) is shown at two different angles. The diameter and volume of spherical condensates were quantified (mean ± s.d., n = 200/each concentration from three independent experiments). Numbers denote mean values. c Condensation profiles of the indicated proteins (40 µM) analyzed concurrently with a at 20 °C. Each point represents the average of two replicates ± s.d. Photographs showing the turbidity of each sample were taken after incubating 20 min or 3 days. d 3D-SIM images for the indicated proteins incubated at 4 °C (top) or 20 °C (bottom) for 1 h and decorated with FITC. Bar, 5 μm. Dotted boxes denote areas of enlargement (bar, 1 μm). DIC, Differential interference contrast. e DIC and 3D-SIM analyses for CPB CP condensates formed at 20 °C for 1 h in the presence of 5% 1,6-hexanediol and decorated with FITC. Bar, 20 μm. f Live cell imaging of U2OS cells stably expressing EGFP-Plk4 CP before (0 s) or after 6% 1,6-hexanediol treatment for 10 s. Bar, 20 μm. Quantification of relative fluorescence intensities is shown in mean ± s.d. (n = 101 for each timepoint from three independent experiments). g FRAP analysis for EGFP-Plk4 CP condensates in U2OS cells left untreated (left) or treated with 6% 1,6-hexanediol for 5 min (right). Images are from Supplementary Video 3. Bar, 2 μm. Relative signal intensities were quantified from 16 independent condensates for each group. Bars, s.d. h FRAP was carried out after photobleaching the bottom half (dotted box) of an FITC-decorated CPB CP condensate formed in vitro. A representative time-lapse series is shown (left). Bar, 0.5 μm. Relative signal intensities from 10 photobleached experimental hemispheres and their respective control condensates were quantified (right). Bars, s.d. Source data are provided as a Source Data file for a–c and f–h

Fig. 7

Phosphorylation of CPB by Plk4 induces clustering in vitro. a 3D-SIM images showing CPB condensates generated by GST-Plk4∆PB3-dependent phosphorylation in vitro. Bar, 10 μm. Dotted boxes, areas of enlargement (bar, 2 μm). Yellow bar in the dotted box, 0.5 μm. Kinase reactions were carried out in two steps—Plk4∆PB3 activation step (1st rx) and CPB phosphorylation step (2nd rx). Reaction products were decorated with FITC to visualize CPB condensates and imaged (left). The same reaction products were separated by 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) for silver staining (right). Note that FITC-decorated condensates contributed by GST-Plk4∆PB3 (5th panel) are negligible. b Immunoblotting analysis showing Plk4∆PB3-phosphorylated CPB generates pelletable condensates in vitro. Following a kinase reaction, a sample was fractionated into supernatant (S15) and pellet (P15), and then separated by 8% SDS-PAGE. Note that Plk4∆PB3-reacted PC3-phosphorylated CPB (α-pSSTT panel) is present only in the P15 fraction (arrow)

Fig. 8

Mutations at the PB2 tip disrupt Plk4’s ability for ring-to-dot conversion, STIL and Sas6 recruitment, and procentriole formation. a Quantification of relative intensities of STIL, STIL pS1108, Sas6, and Plk4 fluorescence signals was carried out using confocal images acquired from immunostained U2OS cells stably expressing the indicated constructs under Plk4 RNAi conditions. The expression levels of Plk4 WT and mutants and their immunostained images are provided in Supplementary Fig. 7a, b. Relative fluorescence intensities (mean ± s.d.) are shown. n = 75 (WT), 75 (WT PB2-tip mt), 97 (CP), 81 (CP PB2-tip mt), and 78 (KM) for STIL/Plk4, n = 80 (WT), 75 (WT PB2-tip mt), 83 (CP), 77 (CP PB2-tip mt), and 75 (KM) for STIL pS1108/Plk4, and n = 75 (WT), 75 (WT PB2-tip mt), 97 (CP), 81 (CP PB2-tip mt), and 78 (KM) for Sas6/Plk4 with each sample pooled from three independent experiments. Numbers denote mean relative intensities. ***P < 0.001; ****P < 0.0001 (unpaired two-tailed t test). b 3D-SIM analysis of immunostained U2OS cells stably expressing the indicated constructs under Plk4 RNAi conditions. Arrows denote the ring-state Plk4 CP PB2-tip mutant colocalized with Cep152. Bar, 1 μm. c, d Quantification of the samples in b is shown in mean ± s.d. (n = 3 independent experiments). ***P < 0.001; ****P < 0.0001 (unpaired two-tailed t test). e Model illustrating the mechanism of how Plk4 breaks the symmetrical ring-state localization pattern and generates a dot-state assembling body for centriole biogenesis. In G1, as Cep152 is recruited to a region around a centriole, Plk4 interacts with Cep152 and localizes at the outskirts of the Cep152 scaffold (ring state). As predicted from extensive simulation analyses, a small fluctuation in the local concentration of Plk4 may allow its N-terminal catalytic activity to cross a critical point of phosphorylating the CPB PC3 motif, thus causing the phospho-CPB to cluster through the aggregative PB2-tip region (for simplicity, Plk4 is illustrated as a monomer) and generate a nanoscale condensate (dot state). The PC3 phosphorylation also promotes the Plk4–STIL interaction (Fig. 2). Since STIL activates Plk4^,, this interaction is expected to reinforce Plk4’s ring-to-dot conversion in a positive-feedback fashion and enable Plk4 to rapidly coalesce into a centriole duplication-competent assembling body. The CP mutations at the PC3 motif enable Plk4 to bypass both Cep152-dependent pericentriolar localization and N-terminal KD-dependent CPB phosphorylation, and, therefore, to efficiently induce multiple dot-like or elongated procentrioles. Source data are provided as a Source Data file for a, c, d