Cell-cycle-regulated expression of STIL controls centriole number in human cells

Vertebrate centrioles are cylindrical structures that are made up of nine triplet microtubules and display an evolutionarily conserved ninefold rotational symmetry (Azimzadeh and Marshall, 2010). Two centrioles, embedded in a meshwork of proteins termed pericentriolar material (PCM), constitute the centrosome, the main microtubule-organizing center (MTOC) of animal cells (Bornens, 2002; Luders and Stearns, 2007). By nucleating and anchoring microtubules, centrosomes influence important cellular properties such as intracellular transport of organelles and vesicles, cell shape, polarity and motility (Doxsey et al., 2005). Centrosomes also promote the formation of a bipolar spindle at the onset of mitosis, thus contributing to the faithful segregation of chromosomes during cell division (Bettencourt-Dias and Glover, 2007; Nigg and Raff, 2009). The second major role of centrioles relates to their ability to function as basal bodies for the formation of cilia and flagella, which in turn play key roles in motility as well as in chemo- and mechanosensation (Goetz and Anderson, 2010). Mutations in genes coding for centrosomal proteins or centriole and basal body components have been linked to brain diseases and ciliopathies (Nigg and Raff, 2009; Bettencourt-Dias et al., 2011). Moreover, numerical and/or structural centrosome aberrations have long been implicated in carcinogenesis (Nigg, 2002; Zyss and Gergely, 2009). Biogenesis and propagation of appropriate numbers of centrioles is crucial for cell function and genome integrity. During the cell cycle, the two centrioles that make up the G1 centrosome need to be duplicated exactly once (Strnad and Gonczy, 2008). Centriole duplication begins at the onset of S phase, when one new centriole (termed procentriole or daughter centriole) forms at a perpendicular angle next to the proximal end of each pre-existing centriole (termed the parental or mother centriole). Procentriole elongation then continues through G2 phase so that before mitosis each centrosome contains one pair of centrioles. At the onset of mitosis, the duplicated centrosomes separate and contribute to the formation of the bipolar mitotic spindle. Association with the spindle ensures the equal distribution of centrioles to nascent daughter cells. Late in mitosis, the tight connection between mother and daughter centriole is severed in a process termed centriole disengagement, which requires the protease separase as well as the mitotic polo-like kinase PLK1 (Tsou and Stearns, 2006a; Tsou et al., 2009). Furthermore, daughter centrioles acquire the competence to organize PCM during passage through mitosis and this also requires the activity of PLK1 (Wang et al., 2011). Thus, traverse of mitosis sets the stage for a new round of centriole duplication in the ensuing cell cycle (Tsou and Stearns, 2006b).

Pioneering work in the nematode Caenorhabditis elegans led to the identification of five proteins, a kinase termed ZYG-1 and four coiled-coil proteins, named SPD-2, SAS-4, SAS-5 and SAS-6, whose recruitment to the preexisting centriole is essential for centriole duplication (Strnad and Gonczy, 2008). The first protein to be recruited to the centriole in C. elegans is SPD-2, which in turn is essential for the localization of ZYG-1. Subsequently, the assembly of a complex of SAS-5 and SAS-6 promotes the formation and elongation of a central tube and SAS-4 is thought to assist in the deposition of singlet microtubules onto this tube, resulting in the formation of a new daughter centriole (Pelletier et al., 2006). In other organisms, notably in Chlamydomonas and vertebrates, procentriole formation is characterized by the assembly of a so-called cartwheel structure, which presumably serves as an assembly platform for the outgrowth of the procentriole. Importantly, SAS-6 has recently been identified as a key component of the cartwheel (Nakazawa et al., 2007) and shown to confer ninefold symmetry to this structure (Kitagawa et al., 2011; van Breugel et al., 2011).

Of the centriole duplication factors described above, SPD-2, SAS-4 and SAS-6 show clear evolutionary conservation in other species. Curiously, no obvious homolog of ZYG-1 has been identified outside of nematodes, but polo-like kinase 4 (PLK4; also known as SAK) clearly plays a functionally analogous role in Drosophila and vertebrates. C. elegans SAS-5 also lacks obvious structural homologs outside of nematodes. Thus, it is of considerable interest that the Drosophila protein Ana2 has recently been identified as a potential functional ortholog of SAS-5 (Stevens et al., 2010). Furthermore, database searches led to the suggestion that STIL (SCL/TAL1 interrupting locus), a large cytosolic protein in human cells, could represent a SAS-5 and Ana2 ortholog in vertebrates. In support of this view, all three proteins share sequence similarity within a short, ∼90 aa C-terminal domain called the STAN (STIL/Ana2) motif (Stevens et al., 2010). The human STIL gene was first identified in the context of a genomic rearrangement that leads to T-cell acute lymphoblastic leukemia (Aplan et al., 1991). In mouse and zebrafish, STIL was shown to be essential for early vertebrate development (Izraeli et al., 1999; Golling et al., 2002; Pfaff et al., 2007). Moreover, STIL expression was reported to be high in lung cancer (Erez et al., 2004) and most interestingly, mutations in STIL were recently shown to cause primary microcephaly (Kumar et al., 2009). Additional studies point to a possible role for STIL in cell proliferation, mitotic regulation and centrosome integrity (Izraeli et al., 1997; Campaner et al., 2005; Erez et al., 2007; Castiel et al., 2011), but the precise function of this protein remains unknown.

Here, we have explored a possible role for STIL in centriole biogenesis in human cells. In view of the close relationship between SAS-6 and SAS-5 in C. elegans, we have focused particular attention on a possible functional interaction between STIL and human SAS-6. Our results unequivocally identify human STIL as a key centriole duplication factor that is essential for cell-cycle-regulated centriole formation. Although no stable complex between STIL and SAS-6 could be observed in human cells, we show that the two proteins strongly resemble each other with regard to their localization and cell-cycle-regulated expression. We further show that overexpression of STIL leads to the near-simultaneous formation of multiple daughter centrioles surrounding each mother, a phenotype previously observed for overexpression of both SAS-6 and the kinase PLK4 (Habedanck et al., 2005; Leidel et al., 2005; Strnad et al., 2007). This argues for a tight cooperation between STIL, SAS-6 and PLK4 in centriole biogenesis in human cells.

To test whether STIL is necessary for centriole duplication in human cells, we depleted the protein from asynchronously growing U2OS cells by RNA interference. Three different siRNA oligonucleotides efficiently depleted STIL after 72 hours of treatment, as confirmed by western blot analysis using a commercial antibody against STIL (Fig. 1A). Centrioles were counted after staining of cells with antibodies against Cep135 and CP110, which mark, respectively, the proximal and distal ends of centrioles (representative images are shown in Fig. 1B). Owing to the proximity of their proximal ends, mother and daughter centrioles are difficult to resolve by anti-Cep135 staining, so that anti-Cep135 antibodies generally produce a two-dot staining pattern throughout interphase, regardless of centriole duplication state. By contrast, the distal ends of mother and daughter centrioles are well separated so that anti-CP110 antibodies usually stain four dots shortly after daughter centrioles begin to form (Kleylein-Sohn et al., 2007). Cells with either two or four CP110-positive dots were regarded as normal, whereas cells harboring only a single centriole or lacking centrioles altogether were considered defective in centriole duplication. The majority of cells treated with control siRNA oligonucleotides (GL2) contained either two or four centrioles (CP110 dots). By stark contrast, about 60% of STIL-depleted cells possessed fewer than two centrioles (Fig. 1C). A similar reduction in centriole number was observed when PLK4, a key regulator of centriole duplication (Bettencourt-Dias et al., 2005; Habedanck et al., 2005), was depleted. Virtually identical results were also obtained after siRNA treatment of HeLa S3 cells (supplementary material Fig. S1A,B). Concomitant with a reduction of centriole numbers in approximately 90% of STIL-depleted HeLa cells, we also observed a significant increase in abnormal mitotic spindles, as reported previously (Pfaff et al., 2007). About 60% of STIL-depleted mitotic HeLa cells contained monopolar spindles, whereas nearly 30% showed bipolar spindles with one acentriolar pole (supplementary material Fig. S1C). Taken together, these data strongly support the notion that STIL is required for centriole duplication.

Because the antibody used above did not detect endogenous STIL by immunofluorescence microscopy, we generated a polyclonal antibody targeting the C-terminus of STIL (amino acids 938–1287, supplementary material Fig. S2A). In western blots, this antibody (termed ca66) readily recognized FLAG-tagged STIL ectopically expressed in HEK293T cells, but endogenous protein was undetectable (supplementary material Fig. S2B). By immunofluorescence microscopy, ca66 clearly stained the centrosome and this staining was lost upon siRNA-mediated depletion of STIL, confirming antibody specificity (supplementary material Fig. S2C). Ca66 was then used to determine the precise subcellular localization of STIL by immunofluorescence microscopy. When co-staining S, G2 or early mitotic cells with antibodies against STIL and CP110, STIL was consistently detected as a single dot at each centriole pair (marked by two CP110 dots) (Fig. 2A). To distinguish between mother and daughter centrioles within each pair, co-staining was performed using antibodies against Cep164, a distal appendage protein that marks mature mother centrioles (Graser et al., 2007). Invariably, STIL was found closer to the Cep164-negative CP110 dot, implying that it associates with the daughter centriole (Fig. 2B). This conclusion was further strengthened by the observation that STIL colocalized with SAS-6 (Fig. 2C), an evolutionarily conserved component of the cartwheel structure that assembles transiently at the proximal end of newly forming daughter centrioles (Leidel et al., 2005; Kleylein-Sohn et al., 2007; Strnad et al., 2007; Kitagawa et al., 2011; van Breugel et al., 2011). Finally, the association of STIL with newly forming daughter centrioles could also be demonstrated by immunoelectron microscopy (Fig. 2D). Taken together, these data indicate that STIL localizes specifically to the proximal part of daughter centrioles, where it colocalizes with the centriole duplication factor SAS-6.

To study the association of STIL with centrioles throughout the cell cycle, we stained asynchronously growing U2OS cells with antibodies against STIL, CP110 and Cep135. Cep135 served as a marker for centrioles and CP110 was used to time the emergence of newly forming centrioles during S phase (Chen et al., 2002; Kleylein-Sohn et al., 2007). In parallel, DNA was stained with DAPI to distinguish between interphase and mitotic cells and to identify different mitotic stages. Of the G1 cells, characterized by the presence of only two centrioles, some lacked detectable STIL, whereas others were clearly positive for STIL, even though no CP110-positive daughter centrioles could yet be seen (Fig. 3A). This strongly suggests that STIL associates with nascent daughter centrioles at a very early stage of centriole duplication, in excellent agreement with our immunoelectron microscopy data (Fig. 2D, right panel). To analyze the timing of STIL recruitment more precisely, nocodazole-arrested U2OS cells were released into G1 phase and both the abundance and localization of STIL were monitored at different time points (supplementary material Fig. S3). Western blot analysis revealed that STIL protein levels were low in early G1 phase (4 hours after release) and progressively augmented towards the G1–S transition (8–12 hours after release; marked by the appearance of the CDK2 activator cyclin E1) (supplementary material Fig. S3A). As shown by immunofluorescence microscopy, the vast majority of cells lacked STIL at their two (unduplicated) centrioles 4 hours after the nocodazole release; however, by 8 hours after release, STIL was present at centrioles in all cells showing duplicated centrioles (four CP110 dots) and also in half of those that showed only two CP110 dots (unduplicated centrioles) (supplementary material Fig. S3B). This indicates that STIL is recruited to daughter centrioles before these can be resolved by CP110 staining. Finally, by 12 hours after release, most cells had undergone centriole duplication and about 90% of all cells showed STIL-positive centrioles (supplementary material Fig. S3B,C). Thus, whenever a cell contained duplicated centrioles, STIL was invariably present, confirming that STIL is recruited very early during the formation of new centrioles.

After the G1–S transition, centriolar STIL staining gradually increased as cells approached mitosis, so that maximal staining was seen at the poles of pro- and prometaphase cells (Fig. 3A). However, at around the time of the metaphase-to-anaphase transition, STIL staining began to decline and was completely abolished in late anaphase and telophase cells, suggesting that STIL either gets displaced from centrosomes or degraded during exit from mitosis. To distinguish between these possibilities, we analyzed STIL protein levels in U2OS cells released from a nocodazole-induced prometaphase arrest. As revealed by western blotting, STIL levels diminished 1–2 hours after release from nocodazole, coincident with a drop in levels of cyclin B1, a prominent substrate of the anaphase promoting complex/cyclosome (APC/C) (Peters, 2006) (Fig. 3B). Identical results were also seen in HeLa S3 cells (supplementary material Fig. S4), confirming and extending an earlier study showing that levels of murine STIL protein, but not mRNA, are reduced in late mitosis (Izraeli et al., 1997). Addition of the proteasome inhibitor MG132 completely blocked STIL degradation, indicating a requirement of the proteasome (Fig. 3B). To rule out cell-cycle effects, nocodazole-arrested cells were also treated with the Cdk1 inhibitor RO-3306 to enforce exit from mitosis even in the presence of MG132 (Fig. 3C). Under these conditions, mitotic exit occurred, as demonstrated by use of an anti-phospho-histone H3 (Ser10) antibody, and STIL was rapidly degraded in the control sample, but stable in the absence of proteasomal activity. Interestingly, inactivation of Cdk1 by RO-3306 resulted in a noticeable increase in the electrophoretic mobility of STIL (Fig. 3C, right panel), suggesting that STIL is a Cdk1-dependent phosphorylation substrate (Campaner et al., 2005). The above results suggested that STIL is a likely substrate of APC/C^Cdc20 and/or APC/C^Cdh1. To explore this possibility, Cdc20 and/or Cdh1 were depleted by siRNA before synchronizing HeLa S3 cells in prometaphase and monitoring STIL levels upon release of cells in the presence of the Cdk1 inhibitor RO-3306. As shown in Fig. 3D, STIL was clearly stabilized in response to depletion of Cdc20, which is similar to results with cyclin B1. Depletion of Cdh1 alone did not produce major effects, but co-depletion of Cdh1 with Cdc20 clearly enhanced stabilization of STIL. The most likely interpretation of these results is that Cdc20 is the major APC/C adaptor responsible for STIL degradation and that Cdh1 contributes at later stages of mitosis.

To analyze the structural and functional relationship between STIL, Ana2 and SAS-5 in more detail, we first performed a BLAST (Basic Local Alignment Search Tool) analysis to search for homologs of human STIL in other organisms. To our surprise, we readily identified apparent STIL homologs not only in other vertebrate species, but also in Hemichordata (Saccoglossus kowalevskii; 24% identity, 38% similarity), Cnidaria (Nematostella vectensis; 20% identity, 39% similarity) and even Placozoa (Trichoplax adhaerens; 17% identity, 26% similarity) (Fig. 4A). However, although the STIL-related proteins of Chordata, Hemichordata and Cnidaria comprise up to 1300 amino acids, those of Drosophila species (Ana2) comprise only about 400 residues (Fig. 4A). As emphasized previously (Stevens et al., 2010), extensive sequence similarity between the various STIL-related proteins is seen in the STAN motif (Fig. 4B). In addition, we noticed a second region with high sequence similarity between the STIL and Ana2 families (Fig. 4B). We refer to this second motif as TIM (truncated in microcephaly), because it localizes to the extreme C-terminus of STIL that is truncated in microcephaly patients (Kumar et al., 2009). Notably, the nematode protein SAS-5 shows lower sequence conservation over the STAN and the TIM motif (Fig. 4B).

In C. elegans, SAS-5 has been shown to form a complex with SAS-6 (Leidel et al., 2005) and likewise, Drosophila Ana2 was found to bind DSAS-6 (Stevens et al., 2010). Thus, we asked whether human STIL might similarly interact with SAS-6. We co-expressed FLAG- or Myc-tagged versions of STIL and SAS-6 and performed co-immunoprecipitation experiments, followed by western blotting. Regardless of whether tags were placed at the N- or C-termini of the two proteins, we have so far been unable to detect any stable complex between STIL and SAS-6 (supplementary material Fig. S5A,B). Likewise, neither endogenous nor overexpressed STIL could be seen after immunoprecipitation of endogenous SAS-6 (supplementary material Fig. S5C). Although these results do not exclude that subpopulations of STIL and SAS-6 might interact in vivo, we have so far been unable to demonstrate such an interaction.

To determine whether STIL and SAS-6 exhibit mutual dependencies for centriole localization, we depleted the two proteins separately from asynchronously growing U2OS cells and stained prophase cells (selected by DAPI staining) with antibodies against STIL, SAS-6 and CP110. Upon depletion of STIL, SAS-6 was readily detected at centrioles, even though levels were reduced to about 35% of those seen in control siRNA-treated cells (Fig. 5A,B). Similar to STIL, SAS-6 is completely degraded during each passage through mitosis (Strnad et al., 2007). This argues against the possibility that the SAS-6 protein detected in the STIL-depleted prophase cells could represent residual SAS-6 that localized to centrioles before the 72 hour siRNA treatment. Rather, it seems likely that SAS-6 can localize to centrioles in the absence of STIL, although STIL apparently contributes to its efficient integration into newly forming daughter centrioles. Conversely, STIL could not be detected at centrioles when SAS-6 was depleted (Fig. 5A,B), suggesting that SAS-6 is essential for efficient recruitment and/or maintenance of STIL at centrioles. To rule out the possibility that these observations reflect changes in the respective protein levels, siRNA-treated U2OS cells were analyzed in parallel by western blotting. As shown in Fig. 5C, the efficient depletion of STIL did not reduce the total abundance of SAS-6 or vice versa, suggesting that the two proteins do not depend on each other for expression or stability.

Overexpression of PLK4 or SAS-6 results in centriole amplification (Leidel et al., 2005; Peel et al., 2007; Strnad et al., 2007; Cunha-Ferreira et al., 2009; Rogers et al., 2009). To ask whether overexpression of STIL might produce a similar phenotype, we overexpressed FLAG-tagged STIL in asynchronously growing U2OS cells and scored centriole numbers (Fig. 6). About 60% of all STIL-transfected cells showed extra copies of centrioles, which was similar to the extent of centriole amplification seen upon overexpression of PLK4 or SAS-6 (Fig. 6A). Strikingly, about 25% of the STIL-transfected cells showed centrioles arranged in a flower-like pattern (Fig. 6B), which was highly reminiscent of the arrangements seen upon overexpression of PLK4 and SAS-6 (Habedanck et al., 2005; Leidel et al., 2005; Duensing et al., 2007; Kleylein-Sohn et al., 2007; Strnad et al., 2007). Co-staining of STIL-transfected cells with antibodies against Cep164, a distal appendage marker present only on mature mother centrioles (Graser et al., 2007), confirmed the formation of multiple daughter centrioles around a single mother centriole (Fig. 6C). In agreement with STIL localization to the proximal ends of daughter centrioles (Fig. 2), STIL localized as a ring around the mother centriole and co-staining with antibodies against SAS-6 revealed extensive colocalization to these ring-like structures, indicating that the two proteins act in close proximity (Fig. 6D).

In conclusion, our data demonstrate that depletion of STIL interferes with daughter centriole formation, whereas overexpression of STIL causes the near-simultaneous formation of multiple daughter centrioles. This indicates that STIL plays a crucial and direct role in centriole duplication and further suggests that levels of this protein must be tightly controlled in human cells to prevent centriole overduplication.

Here we demonstrate that STIL participates directly in the formation of new centrioles in human cells. We further show that STIL expression is regulated during the cell cycle and that physiological levels of STIL are crucial for the control of centriole numbers. siRNA-mediated depletion of STIL from human cells severely suppresses centriole duplication. Conversely, overexpression of STIL triggers centriole amplification, characterized by the formation of multiple daughter centrioles around a single preexisting mother centriole, which is similar to the phenotypes seen upon overexpression of PLK4 and SAS-6. So far, we have not been able to detect stable complexes between STIL and SAS-6. However, the two proteins partly depend on each other for centriole association and they are subject to strikingly similar cell cycle regulation. Similar to SAS-6, STIL associates with newly forming procentrioles at an early stage of centriole formation during G1 phase. STIL then remains associated with daughter centrioles until the metaphase-to-anaphase transition of mitosis, when the protein undergoes APC/C^Cdc20–Cdh1-dependent proteasomal degradation. Collectively, these data identify STIL as a protein that is crucial for the formation of centrioles in correct numbers and they point to close cooperation between STIL, SAS-6 and PLK4 in centriole biogenesis.

Of the five gene products required for centriole duplication in C. elegans (Strnad and Gonczy, 2008), three (SPD-2, SAS-4 and SAS-6) have obvious structural homologs (CEP192, CPAP and SAS-6, respectively) in humans. Furthermore, PLK4/Sak almost certainly represents a functional homolog of nematode ZYG-1. By striking contrast, bioinformatics approaches failed to identify a clear ortholog (outside nematodes) for the coiled-coil protein SAS-5. Our current study suggests that proteins of the STIL/Ana2 family are present throughout Metazoa, but show only limited sequence similarity to nematode SAS-5. This notwithstanding, we provide direct evidence for a crucial role of STIL in centriole biogenesis and number control. By both immunofluorescence and immunoelectron microscopy we show that STIL localizes at the proximal end of newly forming daughter centrioles. At this site, it colocalizes with SAS-6, a highly conserved centriole duplication factor and component of the cartwheel structure (Leidel et al., 2005; Nakazawa et al., 2007; Strnad et al., 2007; van Breugel et al., 2011). Importantly, although the localization of STIL to the centriole depends on SAS-6, SAS-6 can associate with centrioles in the absence of STIL, albeit in reduced amounts. This observation illustrates a striking difference to C. elegans, where SAS-5 and SAS-6 are mutually dependent for their centriolar localization (Leidel et al., 2005). A further difference between STIL and nematode SAS-5 concerns the fact that we have not been able to recover a stable complex between STIL and human SAS-6. Clearly, these negative results do not exclude a transient interaction between the two proteins. However, we emphasize that STIL and SAS-5 also differ with regard to localization. Whereas STIL localization is clearly restricted to daughter centrioles (this study) and Ana2 also shows an asymmetrical localization (Stevens et al., 2010), SAS-5 shuttles between the cytoplasm and both mother and daughter centrioles throughout the cell cycle (Delattre et al., 2004). Thus, the question of whether or not STIL should be considered a genuine ortholog of C. elegans SAS-5 remains difficult to answer.

The properties of STIL in vertebrate organisms have been explored in previous studies. Although these have not directly addressed a role of STIL in centriole biogenesis, we propose that a centriole-related function of STIL can explain most, if not all, of the phenotypes that were previously observed upon inactivation of the gene encoding STIL in zebrafish and mouse. In particular, loss-of-function mutations of STIL (SIL) in zebrafish resulted in embryonic lethality, with embryos showing an increase in mitotic index and disorganized mitotic spindles (Pfaff et al., 2007). The same authors also reported mitotic spindle defects in HeLa cells, as confirmed in our present study. Similarly, STIL-deficient mouse embryonic fibroblasts were shown to display defects in cell cycle progression (Castiel et al., 2011). STIL-knockout (Stil^–/–) mice die at embryonic day 10.5, with prominent axial midline defects and randomized cardiac looping, consistent with a block in Sonic hedgehog (Shh) signaling (Izraeli et al., 1999). These phenotypes are in line with the notion that Shh signaling operates through the ciliary apparatus (Wong and Reiter, 2008; Goetz and Anderson, 2010). In humans, mutations in the STIL gene have been linked to primary autosomal microcephaly (Kumar et al., 2009), a neurodevelopmental disease characterized by abnormally small brain size. Interestingly, most of the genes identified to date that cause this disease code for centrosomal proteins (Thornton and Woods, 2009). Finally, it is intriguing that STIL expression is regulated by the transcription factor E2F (Erez et al., 2008), which has previously been shown to be important for the induction of centriole duplication in somatic cells (Meraldi et al., 1999). Thus, as a target of E2F, STIL might represent an important element in the coupling of centriole duplication to cell cycle cues.

STIL expression is tightly regulated during the cell cycle, with maximal protein levels seen during early mitosis. This might explain why a previous immunolocalization study (Pfaff et al., 2007) emphasized an association of STIL with mitotic spindle poles, but did not detect the association of STIL with interphase centrioles that we report in this study. Although STIL levels at daughter centrioles increase as cells approach mitosis, STIL disappears from centrioles at around the metaphase-to-anaphase transition, as a result of proteasome-mediated degradation. Regulation of STIL protein levels is crucial because overexpression of STIL induces centriole amplification. In particular, excess STIL causes the near-simultaneous formation of multiple daughter centrioles, which is highly reminiscent of the phenotype seen upon overexpression of PLK4 and SAS-6 (Habedanck et al., 2005; Leidel et al., 2005; Duensing et al., 2007; Kleylein-Sohn et al., 2007; Strnad et al., 2007). Thus, STIL cooperates with PLK4 and SAS-6 in the maintenance of constant centriole numbers during cell proliferation. The same conclusion has independently been drawn in a parallel study (Tang et al., 2011).

A full-length cDNA for STIL (clone IRCMp5012H1125D) was obtained from ImaGenes and amplified by PCR using the following oligonucleotides: 5′-CAAGCGGCCGCTTAAAATAATTTTGGTAACTGTC-3′ and 5′-CAAGCGGCCGCAATGGAGCCTATATATCCTTTTG-3′. A NotI digest of the PCR product was cloned into pcDNA3.1-NFLAG and pcDNA3.1/3xmyc-A (Invitrogen). The full-length cDNA of STIL was further amplified using oligonucleotides 5′-TTTTGGCCGGCCATCATGGAGCCTATATATCCTTTTG-3′ and 5′-TTTTCTCGAGATCATGGAGCCTATATATCCTTTTG-3′ and cloned into pcDNA3.1-CFLAG using restriction enzymes FseI and XhoI. A full-length SAS6 cDNA was amplified using oligonucleotides 5′-TTTTGGTACCATCATGAGCCAAGTGCTGTTC-3′ and 5′-TTTTGCGGCCGCACTGTTTGGTAACTG-3′. The PCR product was digested with NotI and KpnI and cloned into pcDNA3.1-Cmyc, using restriction enzymes NotI and KpnI. Full-length SAS6 cDNA was further cloned into pcDNA3.1-NFLAG using restriction enzymes AflII and XhoI. Cloning of PLK4 plasmids was described previously (Habedanck et al., 2005).

Rabbit polyclonal antibodies against His-STIL (aa 938–1287) were raised at Eurogentec (LIEGE Science Park, Seraing, Belgium) and immunoglobulins purified on a Protein A column according to standard protocols. Anti-CP110 (Schmidt et al., 2009), anti-Cep135 (Kleylein-Sohn et al., 2007), anti-PLK4 (Guderian et al., 2010), anti-SAS-6 (Kleylein-Sohn et al., 2007) and anti-Cep164 (Graser et al., 2007) antibodies were described previously. Anti-α-tubulin-FITC and anti-FLAG antibodies were purchased from Sigma. CP110 and SAS-6 antibodies were directly coupled to Alexa Fluor 555, Cep135 antibodies to Alexa Fluor 488 and Cep164, CP110 and Cep135 antibodies to Alexa Fluor 647, using labeling kits (Invitrogen). Alexa-Fluor-488-labeled secondary anti-mouse and anti-rabbit antibodies were purchased from Invitrogen. 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 a PlanApo 100×/1.4 NA oil-immersion objective. Serial optical sections were collected 0.2 μm apart along the z-axis and processed using a deconvolution algorithm and projected into one picture using Softworx (Applied Precision). For quantification of centrosomal STIL and SAS-6 protein levels in ImageJ, images from control and treated samples were acquired with the same exposure time and sum projected. Background signal intensity was subtracted from STIL and SAS-6 signal intensity.

For electron microscopy, cells were grown on coverslips and fixed with methanol for 5 minutes at –20°C. Blocking in PBS with 2% BSA was performed for 30 minutes. Primary antibody incubations were performed for 60 minutes, followed by incubation with goat-anti-rabbit IgG–Nanogold (1:50, Nanoprobes) for 45 minutes. Cells were then fixed with 2.5% glutaraldehyde for 1 hour. Nanogold was silver enhanced with HQ Silver (Nanoprobes). Cells were further processed as described previously (Fry et al., 1998).

U2OS, HeLa S3 or HEK293T cells were grown under standard conditions. Transient transfections were performed using TransIT-LT1 transfection reagent (Mirus Bio) according to the manufacturer’s protocol.

STIL, Cdc20 and Cdh1 were depleted using siRNA duplex oligonucleotides targeting the following sequences: STIL1, 5′-CTGTCACTCGATCGAACCAAA-3′; STIL2, 5′-AAGTAAAGAACCTTAAACCAA-3′; STIL3, 5′-AACTGAGGATTTGGAATTAAA-3′; Cdc20, 5′-AACATCAGAAAGCCTGGGCTT-3′; Cdh1, 5′-AATGAGAAGTCTCCCAGTCAG-3′. PLK4 and SAS-6 were depleted using the siRNA duplex oligonucleotides described previously (Habedanck et al., 2005; Leidel et al., 2005). Luciferase duplex GL2 was used as a control (Elbashir et al., 2001). siRNA duplex oligonucleotides were purchased from Qiagen. Transfections were performed using Oligofectamin (Invitrogen) according to the manufacturer’s protocol.

Cell lysates and western blot analysis were performed as described previously (Chan et al., 2009) using Tris lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% IGEPAL CA-630) containing protease and phosphatase inhibitors. For immunoprecipitation experiments, 20 μg of antibody crosslinked to Protein A beads (Affi-Prep protein A matrix, Bio-Rad Laboratories) were used. Beads were incubated with 2–5 mg of lysate for 2 hours at 4°C. Beads were washed twice with Tris lysis buffer followed by three washes in PBS. Bound proteins were resolved by SDS-PAGE. The following antibodies were used for western blotting: polyclonal anti-STIL antibodies (ab89314, Abcam), monoclonal anti-myc (9E10) (Evan et al., 1985), monoclonal anti-cyclin-B1 (GNS3, Millipore), monoclonal anti-cyclin-E (HE12, Abcam), monoclonal anti-SAS-6 (Kleylein-Sohn et al., 2007), monoclonal anti-α-tubulin (Sigma), polyclonal anti-phospho-Histone H3 (Millipore), polyclonal anti-Cdc20 (sc8358, Santa Cruz) and monoclonal anti-Cdh1 (DH01, Millipore).

To analyze endogenous STIL protein levels during mitosis, HeLa S3 cells were presynchronized in thymidine (2 mM, Sigma) before they were released and arrested at prometaphase by incubation for 14 hours with nocodazole (50 ng ml^–1, Sigma). U2OS cells were directly arrested at prometaphase by incubation with nocodazole (as for HeLa S3 cells). After mitotic shake off and release into fresh medium, the synchronized cells were collected at different time points for western blot analysis. To inhibit the 26S proteasome, nocodazole-arrested cells were released and treated with MG132 (10 μM, Calbiochem). To force mitotic exit of MG132-treated or Cdc20- and Cdh1-depleted cells, Cdk1 inhibitor RO-3306 (9 μM, Enzo Life Science) was provided in addition.

STIL homologs were identified by BLAST (Basic Local Alignment Search Tool) analysis. Protein sequences were aligned using the MUSCLE (multiple sequence alignment by log-expectation) sequence aligning algorithm (Edgar, 2004) in Jalview (Waterhouse et al., 2009) and were colored according to the BLOSUM62 coloring scheme (conservation color increment value was set to 30). Regions with high sequence conservation were subsequently determined by using the Blocks Multiple Alignment Processor (minimum block width was set to 5, maximum block width was set to 200) (Fred Hutchinson Cancer Research Center, Washington, DC). Phylogenetic trees were calculated on the phylogeny.fr platform (http://www.phylogeny.fr) (Dereeper et al., 2008). Protein sequences were first aligned with MUSCLE (Edgar, 2004), alignments were checked for accuracy with G-blocks (Castresana, 2000) and PhyML was used for tree building (maximum-likelihood method) (Guindon and Gascuel, 2003; Anisimova and Gascuel, 2006). Trees were rendered using Treedyn (Chevenet et al., 2006).

We thank Elena Nigg for excellent technical assistance and all members of our laboratory for helpful discussions. We also thank Conrad von Schubert for critical reading of our manuscript.