The E3 ubiquitin ligase Mib1 regulates Plk4 and centriole biogenesis

Centrioles play important functions in the organization of microtubule arrays. A pair of centrioles, surrounded by pericentriolar material (PCM), constitutes the centrosome, the main microtubule‐organizing center of animal cells (Bornens, 2012; Gönczy, 2012; Nigg and Stearns, 2011). In addition, centrioles function as basal bodies, or templates, for the axonemes of cilia and flagella (Ishikawa and Marshall, 2011; Santos and Reiter, 2008). In proliferating cells, centrioles duplicate exactly once per cell cycle through the formation of one new centriole close to the proximal end of each pre‐existing centriole (Azimzadeh and Bornens, 2007; Firat‐Karalar and Stearns, 2014; Nigg and Stearns, 2011). Proper control of centriole biogenesis is crucial for embryogenesis, tissue homeostasis and genome stability (Bettencourt‐Dias et al., 2011; Godinho and Pellman, 2014; Chavali et al., 2014; Nigg et al., 2014). Dysfunction of the centriole‐ciliary apparatus and/or centrosomes is implicated in the etiology of ciliopathies, brain diseases, dwarfism and cancer (Bettencourt‐Dias et al., 2011; Godinho and Pellman, 2014; Hildebrandt et al., 2011; Nigg and Raff, 2009; Thornton and Woods, 2009).

Centriole biogenesis in human cells and Drosophila melanogaster is governed by Polo‐like kinase 4 (Plk4, also known as Sak), a distant member of the Polo kinase family (Bettencourt‐Dias et al., 2005; Habedanck et al., 2005); in Caenorhabditis elegans an analogous function is attributed to the kinase ZYG‐1 (O'Connell et al., 2001). Tight control of Plk4 levels is crucial, as its depletion leads to a gradual loss of centrioles and its overexpression to centriole amplification (Bettencourt‐Dias et al., 2005; Habedanck et al., 2005; Kleylein‐Sohn et al., 2007; Rodrigues‐Martins et al., 2007). One important mechanism for controlling Plk4 abundance is based on the ability of Plk4 to dimerize and trans‐autophosphorylate, which results in the generation of a phosphodegron for recognition by Skp, Cullin, F‐box containing complex coanatining β‐TrCP (SCF^β‐TrCP), an E3 ligase, and subsequent degradation of Plk4 by the proteasome (Cunha‐Ferreira et al., 2013; Cunha‐Ferreira et al., 2009; Guderian et al., 2010; Holland et al., 2012; Holland et al., 2010; Klebba et al., 2013; Sillibourne et al., 2010). Other regulatory mechanisms are likely to exist, and a full understanding of Plk4 regulation is important, particularly in view of the frequent deregulation of this kinase in human cancers (Chng et al., 2008; Macmillan et al., 2001; Mason et al., 2014; van de Vijver et al., 2002) and the impact of Plk4 mutations on brain development and body growth (Martin et al., 2014).

Here, we report on a new interaction between Plk4 and the E3 ubiquitin ligase Mind bomb 1 (Mib1). This comes as a surprise, as Mib1 has previously been studied mainly in the context of Notch and nuclear factor (NF)‐κB signaling (Barsi et al., 2005; Daskalaki et al., 2011; Itoh et al., 2003; Li et al., 2011; Liu et al., 2012). However, recent reports have described associations of Mib1 with centriolar satellites, small multiprotein clusters in the vicinity of centrosomes (Akimov et al., 2011; Jakobsen et al., 2011; Tollenaere et al., 2014; Villumsen et al., 2013). Here, we confirm that Mib1 localizes to centriolar satellites but additionally show that Mib1 relocates to centrioles in response to stimulation of centriole biogenesis. Moreover, we demonstrate that Mib1 interacts with Plk4 and promotes its ubiquitylation. This in turn affects Plk4 abundance and localization, with important consequences for centriole homeostasis.

Recent proteomics and yeast two‐hybrid studies have hinted at a centrosome‐associated function for the E3 ubiquitin ligase Mib1 (Akimov et al., 2011; Jakobsen et al., 2011; Tseng et al., 2014). In line with these results, we found that Mib1 colocalizes with PCM‐1 (Fig. 1A, top left), a marker for centriolar satellites (Dammermann and Merdes, 2002). Furthermore, we could detect Mib1 in the vicinity of centrosomes in interphase but not mitotic cells (supplementary material Fig. S1), in agreement with reports of PCM‐1 localization (Balczon et al., 1994). Centriolar satellite localization of both Mib1 and PCM‐1 was abolished by taxol treatment, suggesting a dependency on intact microtubules (Fig. 1A, top right). Moreover, siRNA‐mediated depletion of PCM‐1 resulted in a redistribution of Mib1 to centrioles (Fig. 1A, bottom), reminiscent of the behavior of the centriolar satellite proteins Cep72 and Cep290 (Lopes et al., 2011; Stowe et al., 2012). Furthermore, overexpression of Myc–Plk4 in either U2OS or RPE‐1 cells also triggered re‐localization of Mib1, but not PCM‐1, to centrioles (Fig. 1B; supplementary material Fig. S2A,B). The same phenotype was seen when Plk4 levels were raised through depletion of β‐TrCP (supplementary material Fig. S2C) or when centriole amplification was induced through transient transfection of Plk4 or the centriole duplication factors STIL or Sas6 (also known as SASS6) (supplementary material Fig. S2D). Moreover, MCF‐7 breast cancer cells, which are prone to centriole over‐duplication upon loss of p53 (D'Assoro et al., 2004), showed increased levels of Plk4 and, concomitantly, more pronounced Mib1 signals at centrioles, when compared to RPE‐1 cells (supplementary material Fig. S2E–G). Thus, we conclude that Mib1 localizes to centrioles in response to Plk4 elevation and centriole over‐duplication, raising the prospect of a functional link between Mib1 activity and the regulation of centriole duplication.

Next, we examined a possible interaction between Mib1 and Plk4. Only trace amounts of wild‐type Flag–Mib1 could be co‐immunoprecipitated with Myc–Plk4 but, remarkably, strong binding was seen between Plk4 and a Flag–Mib1ΔRING mutant (Fig. 1C,D), which is catalytically inactive (Berndt et al., 2011; Itoh et al., 2003). Myc–Cep135, analyzed as a control, was unable to bind either wild‐type or mutant Mib1 (Fig. 1C). A Mib1 mutant carrying a point mutation (C997S) that disrupts one of the three RING fingers (Berndt et al., 2011) also showed poor binding to Plk4 (Fig. 1C,D), in line with observations that this mutant is at least partially active when overexpressed in cells (Berndt et al., 2011; and data not shown). Using Mib1 deletion mutants (Fig. 1E), binding of Myc–Plk4 could be mapped to the N‐terminal but not C‐terminal part of Mib1 (Fig. 1C,D). Complex formation could also be detected between Myc–Plk4 and endogenous Mib1 (supplementary material Fig. S2H) and between Plk4 and Mib1 proteins produced by in vitro translation (supplementary material Fig. S2I), supporting a direct interaction.

The striking difference in the ability of catalytically active and inactive Mib1 to co‐immunoprecipitate with Plk4 raised the possibility that Mib1 might act as an ubiquitin ligase for Plk4. Indeed, transient overexpression of wild‐type Flag–Mib1 resulted in the appearance of higher migrating forms of Myc–Plk4 (Fig. 2A), but not Myc–Cep135 (Fig. 2B). Importantly, this Plk4 modification was induced only by Mib1 full length protein, but not by Mib1 deletion mutants (Fig. 2A), even though the latter showed much stronger binding to Plk4 (Fig. 1C,D). To determine the nature of this Mib1‐dependent modifications, we co‐expressed Myc–Plk4 with His‐tagged ubiquitin and either wild‐type Flag–Mib1 or Mib1ΔRING, respectively, and then isolated poly‐ubiquitylated protein species. Remarkably, Flag–Mib1 wild‐type, but not catalytically inactive Mib1ΔRING, efficiently promoted poly‐ubiquitylation of Myc–Plk4 (Fig. 2C), but not Myc–Cep135 (supplementary material Fig. S3A).

Given that steady‐state levels of Plk4 are regulated by the E3 ligase SCF^β‐TrCP (Cunha‐Ferreira et al., 2009; Guderian et al., 2010; Rogers et al., 2009), we tested the possible involvement of this ligase in Mib1‐mediated ubiquitylation of Plk4. As expected, depletion of β‐TrCP caused an increase in active β‐catenin (ABC; Fig. 2D), a direct target of SCF^β‐TrCP (Hart et al., 1999; van Noort et al., 2002), as well as increased levels of Plk4 at centrosomes (supplementary material Fig. S3B). In striking contrast, β‐TrCP depletion did not affect Mib1‐induced ubiquitylation of Myc–Plk4 (Fig. 2D), arguing that SCF^β‐TrCP does not mediate Mib1 action on Plk4. We have not been able to obtain active full‐length Mib1 to demonstrate in vitro ubiquitylation of purified Plk4 and thus cannot formally exclude an indirect mechanism. However, the most parsimonious interpretation of our results is that Mib1 acts directly on Plk4.

Next, we used mass spectrometry to explore the nature of the Mib1‐induced ubiquitylation of Plk4. We identified a total of 15 ubiquitylated lysine residues within Plk4, many of which clustered within the kinase domain, the linker region, and polo box domains (Fig. 2E). When different moieties of Plk4 were examined for their response to Mib1, most fragments showed a mobility shift after 18 h of co‐expression and downregulation after 36 h of co‐expression (supplementary material Fig. S3C). In contrast, although mass spectrometry had revealed ubiquitylated lysine residues within the PB1 or PB2 region of Plk4, ubiquitylation of a corresponding fragment (residues 608–970) did not lead to either a mobility upshift or downregulation (supplementary material Fig. S3C). Taken together, these experiments indicate that Mib1 modifies Plk4 on multiple sites, likely involving more than one type of ubiquitylation. Depending on the types of linkages, ubiquitin modifications might either trigger protein degradation or serve to regulate protein localization, activity or protein–protein interactions (Al‐Hakim et al., 2008; Ciechanover, 1998; Chen and Sun, 2009; Komander, 2009; Pickart and Eddins, 2004; Xu et al., 2009). To obtain information about the types of Mib1‐mediated ubiquitin chains formed on Plk4 we performed ubiquitylation assays using His–ubiquitin mutants that are able to contribute to only one particular type of linkage. Mib1 mediated extensive ubiquitylation of Plk4 in presence of wild‐type ubiquitin or mutants conferring K11‐, K29‐ and K48‐based linkages, but K63‐based ubiquitylation was much less efficient (Fig. 2F). These results suggest that Mib1 is able to induce formation of ubiquitin chains on Plk4 that are typically associated with fast proteasomal degradation (K11 and K48 linkages). In addition, Mib1 clearly induces formation on Plk4 of at least one additional type of ubiquitin chain that is commonly associated with regulatory events (K29 linkages).

Our identification of K11‐ and K48‐based ubiquitin linkages suggested that Mib1 might control Plk4 abundance. To examine this possibility, we transfected Myc–Plk4‐inducible U2OS (denoted USOS U2OS:Myc–Plk4) cells with control or Mib1 small interfering RNA (siRNA) oligonucleotides, induced Myc–Plk4 expression by tetracycline and 2 h later added cycloheximide to inhibit protein translation (Fig. 3A). As shown by western blotting (Fig. 3B) and quantified in Fig. 3C, Mib1 depletion led to a significant increase in the levels of exogenous Myc–Plk4. Overexpression of Plk4 in U2OS:Myc–Plk4 cells has been previously shown to trigger centriole over‐duplication (Habedanck et al., 2005; Kleylein‐Sohn et al., 2007) as reflected in the formation of multiple procentrioles around each pre‐existing centriole (Fig. 3D). Interestingly, we found that centriole over‐duplication was enhanced upon Mib1 depletion from tetracycline‐treated U2OS:Myc–Plk4 cells, particularly when these were synchronized in S phase by aphidicolin, indicating that Mib1 counteracts excess Plk4 activity (Fig. 3E). In contrast, depletion of Mib1 from parental U2OS cells caused neither a significant increase in centrosomal levels of Plk4 (Fig. 3F) nor centriole amplification (Fig. 3G), although both these phenotypes could readily be observed upon depletion of β‐TrCP (Fig. 3F,G). These results suggest that the ubiquitin ligase activity of Mib1 becomes important primarily in response to excess Plk4, whereas normal cell cycle regulation is exerted primarily by SCF^β‐TrCP.

We also examined the impact of Mib1 overexpression on Plk4 levels. We found that overexpression of Flag–Mib1, but not Flag–Mib1ΔRING, caused a marked decrease of Myc–Plk4 levels at centrioles (Fig. 4A,B), and similar effects were seen also upon endogenous Plk4 (Fig. 4C,D). Similarly, Mib1 caused downregulation when Plk4 levels were elevated by β‐TrCP depletion (supplementary material Fig. S4A,B), and expression of Flag–Mib1 in the U2OS:Myc‐Plk4 cell line greatly impaired the ability of exogenous Myc–Plk4 to trigger centriole over‐duplication (Fig. 4E,F). Taken together, these results demonstrate that Mib1 is able to reduce centriolar levels of Plk4 and thereby counteract Plk4‐induced centriole over‐duplication. Flag–Mib1 overexpression also caused a reduction in centriolar levels of Sas6 (supplementary material Fig. S4C,D), in line with recent studies showing that Plk4 is required for the recruitment of Sas6 to centrioles (Dzhindzhev et al., 2014; Kratz et al., 2015; Ohta et al., 2014). Interestingly, Mib1 also caused a mobility upshift in Sas6, but not in STIL (supplementary material Fig. S4E). The significance of this upshift will require further study. One possibility is that Sas6 is an additional target of Mib1, in which case Sas6 ubiquitylation might also contribute to the effects of Mib1 on centriole duplication.

In addition to K48‐linked ubiquitin chains, we observed ubiquitin chains on Plk4 that are not generally related to fast proteasomal protein degradation (Al‐Hakim et al., 2008; Komander, 2009; Xu et al., 2009). This most likely explains why Mib1‐induced ubiquitylated species of Plk4 could readily be detected without use of proteasome inhibitors (Fig. 3A–C). Considering that we were able to map ubiquitylated lysine residues to regions that are essential for Plk4 functionality, notably the kinase domain, the linker region and the polo box domains (Fig. 2E), we asked whether ubiquitylation might affect the interaction of Plk4 with its binding partners Cep192 and Cep152 (Cizmecioglu et al., 2010; Dzhindzhev et al., 2010; Hatch et al., 2010; Kim et al., 2013; Sonnen et al., 2013). We found that affinity purified LAP–Plk4, ubiquitylated in a Mib1‐dependent manner, indeed showed reduced in vitro binding to both the GST‐tagged N‐terminal of Cep192 and Cep152 (Fig. 4G–I). Taken together, these results suggest that ubiquitylation by Mib1 negatively affects the ability of Plk4 to bind its centriole‐recruitment factors Cep152 and Cep192 (Kim et al., 2013; Sonnen et al., 2013).

In conclusion, we show that Mib1 is important for regulating Plk4 levels, particularly under conditions of aberrant Plk4 expression. Such conditions are known to occur in breast cancer as well as other tumor types (Chng et al., 2008; Macmillan et al., 2001; Mason et al., 2014; van de Vijver et al., 2002). Specifically, we propose that SCF^β‐TrCP is the major E3 ligase controlling steady‐state Plk4 levels under physiological conditions (Cunha‐Ferreira et al., 2013; Cunha‐Ferreira et al., 2009; Holland et al., 2012; Klebba et al., 2013) and that Mib1 might become important primarily in response to excess Plk4 (this study). As illustrated in Fig. 4K, our data indicate that Mib1‐induced ubiquitylation results in at least two functional consequences for Plk4 and hence centriole numbers. First, Mib1 contributes to control turnover of Plk4 through ubiquitin‐dependent proteolytic degradation. Second, Mib1‐dependent ubiquitylation of Plk4 also impairs the ability of Plk4 to interact with Cep152 and Cep192, hence reducing its association with centrioles. Combined, these Mib1‐induced effects on Plk4 abundance and localization contribute to control centriole numbers. In future it will be interesting to explore the role of Mib1 in the control of Plk4 under pathophysiological conditions, notably in human tumors with high levels of centriole amplification. As one example, we have shown here that Mib1 levels are enhanced at the centrioles of breast cancer cells, MCF‐7, which express higher levels of Plk4 than RPE‐1 cells, and our model would indicate that this reflects a response to deregulation of Plk4. However, it is also attractive to speculate that tumors might exist in which a loss of Mib1 (or Mib2) activity causes Plk4 overexpression and concomitant centriole duplication.

U2OS, U2OS:Myc‐Plk4 (Kleylein‐Sohn et al., 2007), MCF‐7 and HEK293T cells were propagated in Dulbecco's modified Eagle's medium (DMEM), and RPE‐1 cells were grown in DMEM with F12. Media were supplemented with 10% fetal bovine serum (FBS) and penicillin‐streptomycin (all from Life Technologies, Carlsbad, CA; FBS for U2OS:Myc‐Plk4 from PAA). Transfection was performed with TransIT‐LT1 (Mirus Bio, Madison, WI). siRNA experiments were performed using Oligofectamine (U2OS cell) or Lipofectamine RNAiMAX (HEK293T cells), both from Life Technologies. Control (GL2) and Mib1 (targeting sequence, TTCTCATCCACAATCCATGGT), PCM‐1 (targeting sequence, AATCAGCTTCGTGATTCTCAG), or β‐TrCP1 and β‐TrCP2 (targeting sequence, GTGGAATTTGTGGAACATC) specific oligonucleotides were used at concentrations of 50–100 nM for 36–48 h. In some experiments, cells were transfected with plasmids 32–36 h after siRNA transfection, as described above. Expression of Myc‐Plk4 in the U2OS:Myc‐Plk4 cell line was induced by tetracycline (1 µg/ml) for 2–12 h. Where indicated, cells were treated with cycloheximide (50 µg/ml) or aphidicolin (1.6 µg/ml; 24 h).

Insert‐containing entry vectors for use in the GATEWAY system (Life Technologies) were generated by PCR, using Pfu Ultra II Fusion DNA polymerase (Agilent, Santa Clara, CA). Inserts were verified by sequencing and subsequently cloned into pDEST‐Myc, pDEST‐Flag, pDEST‐GST (Invitrogen), or pg‐LAP1 (Addgene plasmid 19702) GATEWAY destination vectors. All human cDNAs used here have been described previously: Mib1 (Berndt et al., 2011), Plk4 (Habedanck et al., 2005), Cep152 and Cep192 (Sonnen et al., 2013), Cep135 (Kleylein‐Sohn et al., 2007), Ubiquitin wt and chain‐specific mutants (Chastagner et al., 2006).

Methanol fixation of cells, blocking, incubation with primary and secondary antibodies, and washing were performed as described (Sonnen et al., 2012). The following antibodies were used: rabbit anti‐PCM‐1 was raised by Charles River Laboratories against His‐PCM‐1 (amino acids 1022–1202), rabbit anti‐Mib1 (M5948; Sigma‐Aldrich, St.Louis, MI), rabbit anti‐Cep135, rabbit‐CP110, rabbit anti‐Plk4 (all described in Kleylein‐Sohn et al., 2007), mouse anti‐Plk4 (Guderian et al., 2010), mouse anti‐Myc (clone 9E10), mouse anti‐Flag (clone M2; Sigma‐Aldrich), Alexa‐Fluor‐488‐conjugated donkey anti‐mouse, Alexa‐Fluor‐488‐conjugated donkey anti‐rabbit, Alexa‐Fluor‐488‐conjugated donkey anti‐mouse, and Alexa‐Fluor‐488‐conjugated donkey anti‐rabbit (all from Life Technologies). Direct labeling of primary antibodies was performed using an Alexa‐antibody labeling kit (Life Technologies). Imaging was performed in Glycergel mounting medium (Agilent) at room temperature, using a DeltaVision Core system (Applied Precision, Issaquah, WA) with a 60×/1.2 or 100×/1.4 Apo plan oil immersion objective, Photometrics CoolSNAP CCD camera and solid state illumination. Image stacks were taken with a z distance of 0.2 µm, deconvolved (conservative ratio, three to five cycles), and projected as maximal intensity images by using SoftWoRX (Applied Precision). For cell counts, 50–100 cells per condition and experiment were analyzed. Densitometry analysis was performed in 16‐bit TIFF images with ImageJ within selected regions of interest (ROIs); 15–20 cells per experiment and condition were analyzed. Data are presented as relative staining intensity (staining intensity of a protein of interest normalized to the intensity of centriolar marker Cep135). Relative Mib1 levels at the centrosome (supplementary material Fig. S2G) were measured as total intensities in ROIs using Cell Profiler. The area of Cep135 signal, enlarged by 2ppx, was used to create a mask to measure centriole‐associated Mib1 staining. Contrast and/or brightness adjustment and cropping of final images were done using Photoshop CS5 (Adobe Systems, San Jose, CA).

At 18 h post‐transfection, cells were washed in PBS and lysed [62.5 mM Tris‐HCl, pH 6.8, 1% SDS, 10% glycerol (all from Sigma‐Aldrich)]. Samples were processed as described previously (Bryja et al., 2005) and analyzed by western blotting. The following antibodies were used, in addition to those listed above: rabbit anti‐Myc (sc‐789; Santa Cruz Biotechnology, Dallas, TX), rabbit anti‐Flag (F7425; Sigma‐Alrich), goat anti‐GST (27457701V; GE Healthcare, Issaquah, WA), rabbit anti‐GFP (Ab290; Roche, Basel, Switzerland), mouse anti‐α‐tubulin (T9026; Sigma‐Aldrich), mouse anti‐active‐β‐catenin (clone 8E7; Millipore, Billerica, MA), horseradish peroxidase (HRP)‐conjugated goat anti‐mouse (170‐6516) and goat anti‐rabbit (170‐6515, both from BioRad, Hercules, CA), and HRP‐conjugated donkey anti‐goat (sc‐2020; Santa Cruz Biotechnology). Signal was revealed using the SuperSignal Femto detection kit (Thermo Scientific, Rockford, IL) and detected using LAS3000 (GE Healthcare), ChemiDoc (BioRad) or film (Fig. 2F, Agfa). For immunoprecipitation–pulldown experiments, cells were lysed in IP Lysis buffer [20 mM Tris‐HCl pH 7.4, 150 mM NaCl, 25 mM β‐glycerol phosphate, 0.5% Igepal CA630, 0.5% Triton X‐100 (all from Sigma‐Aldrich), and 1× Complete proteasome inhibitors (Roche)]. Following centrifugation (15,000 g for 10 min at +4°C), cleared extracts were incubated (6 h at +4°C in an orbital shaker) with anti‐Myc (9E10), anti‐Flag (M2) G‐protein–sepharose (GE Healthcare) or S‐protein agarose (Novagen, Billerica, MA). Each batch of beads with LAP–Plk4 was tested for the presence of a Mib1‐induced upshift and stored as a 25% slurry (IP Lysis buffer plus 10% glycerol and 100 µM DTT; −80°C) until used in binding assays. His pulldowns were performed in denaturing conditions [50 mM Tris‐HCl, pH 7.4, 300 mM NaCl, 0.5% Triton‐X‐100, 0.05% SDS, 10 mM Imidazol and 6 M Urea (all from Sigma‐Aldrich)] using Ni‐NTA agarose (Qiagen, Hilden, Germany). Complexes were pelleted, washed, and subsequently analyzed by western blotting.

Myc–Plk4, Flag–Mib1 and Flag–Cep152 were translated in vitro (TNT Quick system; Promega, Madison, WI), mixed with modified IP Lysis buffer (as above; 0.05% Igepal CA630, 100 µM DTT) and subjected to immunoprecipitation and washing. GST‐tagged proteins were purified as described previously (Čajánek and Nigg, 2014). Binding of Mib1‐modified LAP–Plk4 with the GST‐tagged Cep192 N‐terminus (recombinant, purified from bacteria) or Cep152 N‐terminus (translated in vitro) was performed in pulldown buffer (20 mM Tris‐HCl, pH 7.4, 150 mM NaCl, 25 mM β‐glycerol phosphate, 0.5% Igepal CA630, 100 µM DTT) for 45 min (+4°C; orbital shaker). Where indicated, densitometry analysis of blots was performed with ImageJ or ImageLab (Fig. 2D; BioRad).

Following 4 h of treatment with proteasome inhibitor MG132 (10 µM), LAP–Plk4 was isolated from HEK293T cell extracts prepared 18 h after transfection, using S‐protein agarose. Proteins were eluted, reduced, alkylated, digested with trypsin overnight, and then purified with C‐18 Microspin columns (Harvard Apparatus, Holliston, MA). Peptides were separated by using an Easy‐Nano‐LC system, and liquid‐chromatography–tandem‐MS analysis was performed on a hybrid LTQ‐Orbitrap mass spectrometer (both from Thermo Scientific). Obtained protein spectra were searched against the human proteome database (UniProt). The detection of ligated ubiquitin was based on ubiquitin‐derived di‐glycine mass increments of ubiquitylated peptides (114.1 Da per modified lysine residue) (Kirkpatrick et al., 2005).

Statistical analyses (Student's t‐test or ANOVA with Tukey's multiple comparison test) were performed using Prism 6 (GraphPad Software, La Jolla, CA). *P<0.05, **P<0.01 and ***P<0.001 were considered statistically significant differences. Results are presented as mean±s.e.m.

We thank Jason Berndt (University of Washington), Christel Brou (Institut Pasteur), Peter Jackson (Genentech), Matthias Peter (ETH Zurich), Xiumin Yan (IBCB Shanghai), Christian Arquint, and Patrick Redli (both Biozentrum) for sharing reagents, Gernot Guderian for communicating initial observations on a possible Mib1‐Plk4 interaction, and Elena Nigg, the imaging core facility and the proteomics core facility for assistance.