The centrosome, which consists of centrioles and pericentriolar material (PCM), becomes mature and assembles mitotic spindles by increasing the number of microtubules (MTs) emanating from the PCM. Among the molecules involved in centrosome maturation, Cep192 and Aurora A (AurA, also known as AURKA) are primarily responsible for recruitment of γ-tubulin and MT nucleators, whereas pericentrin (PCNT) is required for PCM organization. However, the role of Cep215 (also known as CDK5RAP2) in centrosome maturation remains elusive. Cep215 possesses binding domains for γ-tubulin, PCNT and MT motors that transport acentrosomal MTs towards the centrosome. We identify a mitosis-specific centrosome-targeting domain of Cep215 (215N) that interacts with Cep192 and phosphorylated AurA (pAurA). Cep192 is essential for targeting 215N to centrosomes, and centrosomal localization of 215N and pAurA is mutually dependent. Cep215 has a relatively minor role in γ-tubulin recruitment to the mitotic centrosome. However, it has been shown previously that this protein is important for connecting mitotic centrosomes to spindle poles. Based on the results of rescue experiments using versions of Cep215 with different domain deletions, we conclude that Cep215 plays a role in maintaining the structural integrity of the spindle pole by providing a platform for the molecules involved in centrosome maturation.
The centrosome is the major microtubule (MT)-organizing center in animal cells and is composed of a pair of centrioles and surrounding pericentriolar material (PCM). During M phase, the centrosomes become localized at each spindle pole to assemble the mitotic spindle by undergoing a process called maturation, which is characterized by an increased number of MTs emanating from the centrosome. Some MTs are directly polymerized onto MT nucleators (γ-tubulin and γ-TuRCs) embedded in the PCM (Kuriyama and Borisy, 1981). Other MTs are formed around the chromatin and at the walls of pre-existing MTs, and are then transported toward the centrosome by MT motor proteins (reviewed by Meunier and Vernos, 2016; Prosser and Pelletier, 2017). Stabilization of MT minus ends at the centrosome is also an important factor to assure the association of a large number of MTs with the centrosome. To accommodate the variety of molecules and protein complexes required for MT nucleation, anchorage and stabilization, the PCM expands enormously during maturation (Lawo et al., 2012; Sonnen et al., 2012). This expansion of the PCM is one of the hallmarks of centrosome maturation.
Molecules involved in centrosome maturation include a subset of core centrosomal proteins – Cep192 (Gomez-Ferreria et al., 2007; Zhu et al., 2008), Cep215 (also known as CDK5RAP2; Fong et al., 2008), pericentrin (PCNT; Zimmerman et al., 2004) and γ-TuRC components (Lüders et al., 2006) – and two mitotic kinases, Aurora A (AurA, also known as AURKA) and Plk1 (Lane and Nigg, 1996; Hannak et al., 2001; Berdnik and Knoblich, 2002; Haren et al., 2009). High-resolution analysis has revealed that PCNT, an elongated scaffolding protein, extends toward the centrosomal periphery to provide the structural frame of the expanding mitotic PCM (Lawo et al., 2012). Cep192 is essential for MT polymerization by recruiting γ-TuRC to the centrosome (Gomez-Ferreria et al., 2007; Zhu et al., 2008). During M phase, more binding sites for γ-TuRC are created on Cep192 as a result of protein phosphorylation controlled by AurA and Plk1 (Joukov et al., 2014). In contrast to these molecules, little is known of the role of Cep215 in centrosome maturation.
Cep215 is a mammalian homolog of Drosophila Centrosomin (Cnn) (Zhang and Megraw, 2007), and mutations in the gene encoding human Cep215 induce primary microcephaly, a neurodevelopmental disorder (Woods et al., 2005). Members of the Cnn family are present in a wide range of species, from yeast to humans (Verde et al., 2001; Sawin et al., 2004; Pitzen et al., 2018). Although their sizes and sequences are diverse, all members of the family contain the highly conserved Cnn motif 1 (CM1) and motif 2 (CM2) (Zhang and Megraw, 2007). Because CM1 is capable of binding γ-tubulin and γ-TuRC to initiate MT nucleation (Terada et al., 2003; Choi et al., 2010), Cep215 has been predicted to be important for centrosome maturation by inducing MT polymerization onto the centrosome (Fong et al., 2008). However, this notion has been challenged (Kim and Rhee, 2014). CM2 is responsible for the centrosome targeting of Cep215 (Barr et al., 2010; Wang et al., 2010) and for the binding of PCNT (Buchman et al., 2010; Wang et al., 2010). Therefore, Cep215 is likely important for the construction of the mitotic PCM scaffold in collaboration with PCTN (Lawo et al., 2012; Kim and Rhee, 2014). Indeed, Cnn is widely acknowledged to play a major role in the assembly of mitotic centrosomes in fly embryos (Conduit et al., 2010). Besides CM1 and CM2, Cep215 includes the binding sites of the MT motors dynein–dynactin (Jia et al., 2013) and HSET (also known as KIFC1; Chavali et al., 2016), both of which are important for transporting and focusing acentrosomal MTs to spindle poles. In fact, abnormal poles disconnected from mitotic centrosomes have been reported in cells lacking not only Cnn/Cep215 (Lucas and Raff, 2007; Barr et al., 2010; Barrera et al., 2010; Lee and Rhee, 2010; Chavali et al., 2016), but also the Cnn/Cep215 binding partners Ncd/HSET and dynein–dynactin (Morales-Mulia and Scholey, 2005; Chavali et al., 2016).
To understand how Cep215 is involved in centrosome maturation, we analyzed the mouse Cep215 sequence to identify two previously unknown centrosome-targeting domains. One (215M) includes the dynein–dynactin-binding sequence (Jia et al., 2013) and the other (215N) targets Cep215 to the centrosome during M phase. The 215N domain binds to Cep192 and AurA phosphorylated at T288 (pAurA), both of which are essential for 215N recruitment to the centrosome. Deletion of Cep215 does not significantly interfere with γ-tubulin recruitment to mitotic centrosomes. However, as reported previously (Barr et al., 2010), the most prominent phenotype of the Cep215 knockout is centrosome separation from the spindle poles. We show that this phenotype is rescued by the collaborative efforts of multiple binding domains of Cep215 that are specific to γ-tubulin, Cep192, pAurA, MT motor proteins and PCNT, suggesting that Cep215 serves as a platform for molecules important for organization of mitotic spindle poles during centrosome maturation.
Domain analysis of Cep215
We first analyzed deletion constructs of the mouse Cep215 sequence by expressing a series of c-Myc (myc)-tagged truncated polypeptides. We employed Chinese hamster ovary (CHO) cells, which divide faster than the majority of human cell lines (∼13 h versus ∼24 h doubling time) and are highly amenable to mitotic synchronization (∼75% mitotic index) and transfection (>50% efficiency) (Fig. S1). As reported previously (Barr et al., 2010; Wang et al., 2010), the truncated polypeptide CM2, consisting of a C-terminal segment of Cep215 (Fig. 2A), functions as a centrosomal-targeting domain throughout the cell cycle (Fig. 1A, panels 1,2). By labeling Cep215 with antibodies recognizing the sequence outside CM2, we found that CM2 colocalizes well with endogenous Cep215 in interphase centrosomes (Fig. 1A, panels 3a,b). CM2 includes the PCNT-binding sequence (Buchman et al., 2010; Wang et al., 2010), thus PCNT shows an almost identical distribution to CM2 (Fig. 1A, panels 3a,c) and endogenous Cep215 (Fig. 1A, panels 3b,c). CM1, another conserved domain of Cep215, is capable of γ-tubulin and γ-TuRC binding (Fong et al., 2008). Although exogenous CM1 can initiate MT nucleation (Terada et al., 2003; Choi et al., 2010), it failed to localize to both interphase and mitotic centrosomes, as reported previously (Fong et al., 2008; Fig. 1B).
The central region of Cep215 between CM1 and CM2 is relatively uncharacterized. Nonetheless, a unique sequence termed PReM (Conduit et al., 2014) and domains specific to several MT-interacting proteins, including dynein light chain 8 (also known as DYNLL1; Jia et al., 2013), EB1 (also known as MAPRE1; Fong et al., 2009) and a minus-end-directed kinesin-like motor (HSET; Chavali et al., 2016) have been identified. By cutting in the middle of this long stretch of ∼1300 amino acids, we generated near N-terminal (215N, amino acids 166–696) and mid-portion (215M, amino acids 697–1472) polypeptides (Fig. 2A,B). 215M, which includes the sequences capable of interacting with dynein and EB1, was found at the interphase and mitotic centrosomes (Fig. 1C). However, its fluorescence signal was found to be much weaker than that of CM2. Furthermore, interphase centrosomes labeled with 215M appeared differently in different cells. Some centrosomes appeared as a discrete dot, while others were diffuse and irregular in shape. Centrosomal localization of 215M was observed in HeLa and U2OS cells (Fig. S2), but we detected only a very faint signal, if any, in RPE1 cells (data not shown).
The 215N polypeptide, which consists of an ∼500-amino-acid sequence downstream of CM1 (Fig. 2A), was observed to evenly disperse in the entire cytoplasm of interphase cells (arrowheads in Fig. 1D). As the cell cycle proceeded, the myc signal became concentrated at the centrosomes in late G2 cells (arrow in Fig. 1D, panel 1′) prior to histone H3 phosphorylation at Ser10 (pHH3, arrow in Fig. 1D, panel 2). The 215N polypeptide persisted at the spindle poles during mitosis, and then gradually faded away by the end of cell division (Fig. 1D, panels 4–6). Cells depleted of endogenous Cep215 by RNAi were also capable of recruiting 215N to the centrosome at each spindle pole (Fig. 1D, panel 7). The 215N fragment was targeted to the mitotic centrosome in all cell types tested thus far, and was better detected in cells transfected with significantly lower amounts of plasmid DNA than usual (see Materials and Methods). With higher amounts of DNA, a high background of cytoplasmic fluorescence interfered with clear visualization of 215N signal at mitotic centrosomes.
The 215N and CM2 regions are major domains targeting Cep215 to the mitotic centrosome
The presence of newly identified centrosome-targeting domains indicates that Cep215 is recruited to the mitotic centrosome via different pathways. To assess the contribution of individual domains, we quantified the fluorescence intensity of myc signal at the spindle poles (Fig. 2C). The level of 215N fluorescence corresponded ∼30% of that of the full-length protein [mean±s.e.m. of 3.11±0.15 arbitrary units (a.u.) versus 10.50±0.62 a.u., respectively; P<0.001] (Fig. 2C and Fig. 2D, panels 2,3). An 8–9% higher level of fluorescence intensity of myc–215N was detected in cells where endogenous Cep215 was knocked out by introducing siRNA specific to the sequence outside the 215N region (215N*). CM2 (∼21%) and 215M (∼11%) produced lower intensities of centrosomal fluorescence than that of 215N. The sum of these three intensity values (∼62%) is less than 100%, possibly because adjacent targeting domains have been separated. When the coding sequence of 215N was removed from Cep215 (Δ215N), the spindle pole fluorescence signal fell to ∼32% of the control level (Fig. 2D, panel 4), which is a greater decrease than that seen in cells expressing the CM2-lacking sequence (ΔCM2, ∼40% of control level) (Fig. 2C). Because the difference in fluorescence intensity between 215N and CM2, and between Δ215N and ΔCM2 was ∼8–9%, it is reasonable to conclude that 215N is more potent than CM2 for targeting Cep215 to the mitotic centrosome in CHO cells. The contribution of 215M was significantly lower than that of 215N and CM2 (Fig. 2C). Statistical analysis found this difference to be highly significant (P<0.001), thus we conclude that recruitment of Cep215 to the mitotic centrosome is primarily achieved by two domains, 215N and CM2. Indeed, a chimeric fusion protein of 215N and CM2 (215N/CM2; Fig. 2A and Fig. 2D, panel 5) retained nearly 80% of the targeting activity, whereas only ∼4% of centrosomal fluorescence remained in cells expressing Cep215 lacking both 215N and CM2 (Δ215N/ΔCM2; Fig. 2A and Fig. 2D, panel 6). We obtained similar results of fluorescence intensity of Cep215 domains in human cells. Bipolar spindles formed in cells expressing such an extremely low level of Cep215 fluorescence at the poles were thin and composed of significantly reduced amounts of spindle MTs (Fig. 2D, panel 7), which is further analyzed in Fig. 8.
The 215N domain is different from the HSET-binding domain
Almost half of the C-terminal 215N region partially overlaps with the sequence previously identified as an HSET-binding domain (amino acids 500–700 in human Cep215; Chavali et al., 2016; Fig. S3A). To determine the relationship between 215N and the HSET-binding domain, we cut 215N in the middle to generate 215N-N2 (amino acids 166–435) and 215N-C (amino acids 436–696) polypeptides (Fig. S3A). Despite a significant reduction in myc fluorescence at the centrosomes, myc-tagged 215N-N2 was still able to localize at mitotic centrosomes (fluorescence levels ∼22% of 215N; Fig. S3B, panel 4). In contrast, 215N-C totally lost the ability to associate with the centrosome (Fig. S3B, panel 3 and Fig. S3C, panel 3′). Truncation of 35 amino acids from the N terminus of 215N-N2 reduced the targeting activity of 215N-N3 (amino acids 201–435) to ∼16% of 215N. This difference (∼6%) between 215N-N2 and 215N-N3 is more than that between 215N-N2 (∼22%) and 215N-N1 (∼24%). Because the addition of ∼70 amino acids at the C-terminal end of 215N-N2 resulted in only ∼2% increase of the centrosomal target, it is reasonable that the N-terminal sequence of 215N is more crucial than the C terminus for centrosome-targeting activity. Immunostaining of 215N and HSET in mitotic cells showed that the vast majority of HSET, both endogenous and exogenous, localizes along spindle fibers (Fig. S3C, panels 1–3; Kuriyama et al., 1995). In contrast, 215N is found exclusively at the centrosome (Fig. S3C panels 1′,2′), suggesting that 215N is different from the HSET-binding domain.
The 215N domain binds pAurA, and their centrosomal localizations are mutually dependent
Because 215N is one of the major centrosome-targeting domains, we investigated this sequence further. We have previously reported the close relationship between Cnn and AurA (Terada et al., 2003). Furthermore, the timing of 215N emergence at the centrosome resembles that of AurA phosphorylated at T288 (pAurA). Thus we first compared the distributions of 215N and pAurA in mitotic spindles by employing two AurA antibodies: an anti-AurA monoclonal antibody (mAb) labeling non-phosphorylated AurA, and an anti-pAurA polyclonal antibody (pAb) specific to the phospho-epitope at T288. The specificity of the pAb to pAurA was demonstrated by loss of immunoreactivity of mitotic centrosomes after dephosphorylation by phosphatase treatment (Fig. S4, panels 1,2). As has been previously well-documented, non-phosphorylated AurA was primarily seen at spindle poles and fibers, whereas phosphorylated AurA appeared as a discrete dot at each spindle pole (Fig. 3A, panels 1,2; Fig. S4, panel 3). The association of pAurA and AurA with spindle fibers and spindle poles was retained in isolated spindles (Fig. S4, panels 4–6). It is clear that the 215N distribution is more similar to that of pAurA than that of AurA. Fig. 3A, panel 3, shows GFP-tagged centrin-1 (GFP–CETN)-expressing cells immunostained for pAurA and Cep135 (a marker for the proximal end of the mother centriole). pAurA localized at a position more distal than Cep135, but more proximal than CETN, which is known to localize in the middle to distal region of the centriole. Like pAurA, 215N displayed preferential localization at and around the proximal end of the mother centriole. Although similar, the distribution of the two molecules did not overlap entirely (arrows and arrowheads in Fig. 3A, panels 4a,b).
The close proximity of 215N and pAurA suggests the possibility of physical interaction between the two molecules, which we analyzed by immunoprecipitation (Fig. 3B–D). GFP–AurA and myc–215N were specifically pulled down by anti-myc and anti-GFP antibodies, respectively (Fig. 3B). The presence of phosphorylated AurA, both endogenous (pAurA) and exogenous (GFP–pAurA), was also confirmed in the precipitated fractions (Fig. 3C). The amount of pAurA co-precipitating with 215N was elevated when exogenous AurA was co-expressed with 215N (lanes 3 and 4 in Fig. 3C). This suggests that AurA phosphorylation is stimulated by interaction with 215N. Fig. 3D shows immunoprecipitation of myc–215N with mutant AurA, in which T288 is replaced with non-phosphorylatable alanine (T288A). GFP–AurA (T288A) was not recognized by anti-pAurA antibodies and no longer co-immunoprecipitated with 215N. These results indicate that 215N specifically binds to phosphorylated AurA, but does not bind to non-phosphorylated AurA.
To examine how Cep215/215N and pAurA relate to each other at the centrosome, we prepared RNAi cells (Fig. 4A) to analyze the presence or absence of the molecules at the mitotic centrosome (see summary in Fig. 4B). In AurA-depleted cells, Cep215 was successfully recruited to the centrosomes (arrow in Fig. 4C, panel 1′). In the reverse situation, AurA was still found at the spindle fibers and spindle poles in Cep215-depleted cells (arrows in Fig. 4E, panels 1,2). AurA-depleted cells expressing CM2 and 215N are shown in Fig. 4C, panels 2 and 3, where both Cep215 domains are seen to be targeted to the mitotic centrosomes (arrows). CM2 also localized to the centrosome in cells lacking pAurA (arrow in Fig. 4D, panel 1′). The independence of CM2 and pAurA is supported by the distinctive distributions of these molecules (Fig. 4D, panel 2). Unlike pAurA, which is confined to a small area of the centrosome, CM2 disperses widely to the PCM, occasionally along the flare-like PCM particles similar to those previously reported by Megraw et al. (2002) in fly embryos. In contrast to CM2, 215N failed to localize to the mitotic centrosomes after depletion of detectable pAurA in AurA-knockdown cells (arrow in Fig. 4D, panel 3′). These results indicate that pAurA is essential for centrosomal recruitment of 215N. We noted that pAurA was sometimes more resistant than AurA to exclusion from the centrosome following RNAi of AurA. It is thus possible that some of the AurA-negative cells shown in Fig. 4C, panel 3, still had pAurA, allowing 215N to localize to the centrosome. Whereas Cep215 was dispensable for the centrosomal localization of AurA (arrows in Fig. 4E, panels 1,2), pAurA was no longer detected at the mitotic centrosome in Cep215-depleted cells (Fig. 4E, panel 2″). Panels 3 and 4 of Fig. 4E show results of rescue experiments. The 215N domain, but not myc–Cep215 lacking the 215N sequence (Δ215N), was able to restore pAurA to the centrosome in Cep215-depleted cells. From these results, we conclude that 215N and pAurA are mutually dependent for localization to the mitotic centrosome.
Centrosomal targeting of 215N is inhibited by both AurA and Plk1 inhibitors
To confirm the interdependency between 215N and pAurA, we examined the effect of MLN8237, a small molecule inhibitor of AurA. Although entry into mitosis was slightly delayed, cells treated with MLN8237 entered M phase to initiate spindle assembly. Due to the lack of pole separation, a well-known phenotype of AurA inhibition, the vast majority of cells formed monopolar spindles associated with only a few spindle MTs (Fig. 5A, panel 2). In those cells, the fluorescence intensity of AurA at the poles was reduced to less than half of that of mock-treated cells [mean±s.e.m. of 2.16±0.09 a.u. versus 5.11±0.24, respectively; P<0.001]. The majority of pAurA was also removed from the spindle pole (∼32% of control levels remained), but a trace amount of the protein signal persisted in cells treated with up to 1 µM MLN8237 (Fig. 5A, panel 5′). Similar to pAurA fluorescence, the fluorescence intensity of 215N diminished to ∼20% of the control level (Fig. 5A, panel 8), but centrosomal targeting of CM2 was not affected by drug treatment (Fig. 5A, panel 11). Because Cep215 was found to be recruited to the mitotic centrosomes via two independent pathways (an MLN8237-sensitive 215N-mediated pathway and an MLN8237-insensitive CM2-mediated pathway), it is reasonable that we detected residual Cep215 fluorescence (∼44% of the control level) at the mitotic centrosome in MLN8237-treated cells (Fig. 5A, panel 2′, Fig. 5B).
Mitotic cells were also treated with a Plk1 inhibitor (BI2536), which interferes with the formation of functional bipolar spindles, as does MLN8237. However, unlike AurA inhibition, BI2536 did not significantly affect astral MTs (Fig. 5A, panels 3,6), which is in good agreement with previous reports (De Luca et al., 2006; Hanisch et al., 2006). Plk1 inhibition efficiently removed both AurA and pAurA from centrosomes (∼25% of AurA and 10% of pAurA remained) (Fig. 5A, panels 6,6′ and Fig. 5B). It also profoundly affected the centrosomal targeting of 215N and removed nearly 95% of fluorescence from the centrosome (Fig. 5A, panel 9, Fig. 5B). This is in striking contrast to the effect of Plk1 inhibition on CM2, which remained entirely intact, as in control cells (Fig. 5A, panel 12, Fig. 5B). Consequently, the fluorescence intensity of Cep215 became almost halved at mitotic centrosomes in Plk1-inhibited cells (Fig. 5A, panel 3′, Fig. 5B). The inhibitory effect of BI2536 on Cep215 has previously been reported; Haren et al. (2009) calculated that less than 20% of Cep215 fluorescence remains at mitotic centrosomes in BI2536-treated HeLa cells. This is much less than the 46–47% that remained in CHO cells (Fig. 5B). We repeated these experiments using RPE1 cells and found that Plk1 inhibition reduced centrosomal fluorescence of Cep215 to 32% of the control level (mean±s.e.m. of 4.17±0.36 a.u. versus 13.02±0.81, respectively; n=54; P<0.001). This number is lower than that seen in CHO cells but still higher than that reported by Haren et al. (2009). The difference in the degree of inhibition may be due to, at least partly, the different cell types, species or reagents and protocols used to quantify fluorescence intensities. From these results, we conclude that the 215N-dependent, but not the CM2-dependent, centrosome-targeting pathway is under the control of Plk1.
Cep192 interacts with 215N and is required for centrosomal localization of 215N
We next examined the relationship of 215N with Cep192, a key molecule in centrosome maturation. Like 215N, Cep192 preferentially localizes to the proximal end of the centriole (Zhu et al., 2008; Lawo et al., 2012), and the two proteins indeed colocalized quite well at each centrosome (Fig. 6A). Cells co-expressing myc–215N and Flag–Cep192 also revealed identical distribution of the two proteins (Fig. 6B, panel 1). At increasing levels of protein expression, Flag-tagged proteins induced cytoplasmic aggregates of various sizes and numbers throughout the cell cycle (Fig. 6B, panels 1–4). Importantly, myc–215N became associated with each aggregate in a perfectly overlapping manner. This is strikingly different from cells expressing myc–215N alone, where no such protein aggregates were induced, no matter how strongly myc–215N was expressed (arrow in Fig. 1D, panel 3; Fig. S3, panel C1′). In contrast to 215N, other domains of Cep215 (CM1, 215M and CM2) did not appear to colocalize well with Flag–Cep192 (Fig. 6B, panels 5–7). To confirm a physical interaction between 215N and Cep192, we performed immunoprecipitation experiments (Fig. 6C). Flag–Cep192 and myc–215N co-expressed in cells were successfully precipitated with antibodies against the tag attached to the counterpart protein. Conduit et al. (2010) have reported co-immunoprecipitation of Cnn and DSpd-2 (homologs of Cep215 and Cep192, respectively) in fly embryos. Although the band intensity was weak, we confirmed co-precipitation of endogenous Cep215 with Flag–Cep192 in mammalian cells (lower panel of Fig. 6C). Panels 8 and 8′ of Fig. 6B show colocalization of Flag–Cep192 aggregates with endogenous AurA in interphase cells, which is consistent with the previous report of Joukov et al. (2014) showing interaction of Cep192 with AurA.
To determine whether Cep192 and 215N are required for each other to localize at mitotic centrosomes, we prepared Cep192-deleted cells using RNAi (Fig. 6D). Without Cep192, no 215N was found at the mitotic centrosome (Fig. 6E, panel 2), but recruitment of CM2 to the centrosomes was not affected by the loss of Cep192 in both mitotic and interphase cells (Fig. 6E, panels 3,4). This indicates that the 215N-dependent, but not the CM2-dependent, centrosome targeting of Cep215 is controlled by Cep192. We next analyzed centrosomal localization of Cep192 in Cep215-depleted cells. Among three mitotic cells included in the same frame shown in Fig. 6F, one cell expressed almost no Cep215 (arrow), but residual amounts of Cep215 were still detected in two other cells (arrowheads). Regardless of the presence or absence of remaining Cep215, considerable amounts of Cep192 were still detected at the centrosome in all cells (Fig. 6F, panel 1″). We thus conclude that Cep192 is recruited to the mitotic centrosome independently of Cep215. However, because no 215N-binding domain of Cep192 has as yet been identified, we cannot rule out the possibility that Cep192 and Cep215 are partially interdependent in localizing at the mitotic centrosome.
Cep215 does not significantly contribute to γ-tubulin recruitment to mitotic centrosomes
The MT-nucleating activity of mitotic centrosomes is primarily controlled by Cep192 and pAurA, to which Cep215 binds via 215N. To determine whether Cep215 is also required for promoting MT assembly at the centrosome, we measured the ability of Cep215-depleted cells to recruit γ-tubulin to mitotic centrosomes (Fig. 7A, panel 1). Because the RNAi treatment was not completely efficient, a trace amount of Cep215 [1.8±0.3% of control levels (mean±s.e.m.), n=68, P<0.001] was still detected at the centrosome in cells where a large amount of γ-tubulin (65.1±4.8% of control levels) is still retained (Fig. 7B). Two different siRNAs, specific to the N- and C-terminal sequences of Cep215, yielded almost identical results. Depletion of Cep192 and AurA by siRNA treatment was less efficient than Cep215 depletion; the amount of centrosomal fluorescence quantified (as described in Materials and Methods) revealed ∼7% of Cep192 and ∼19% of pAurA fluorescence still present at mitotic centrosomes in siRNA-treated cells (Fig. 7A, panels 2,3 and Fig. 7B). Nonetheless, much lesser amounts of γ-tubulin remained in those cells (∼12% for Cep192 and ∼24% for pAurA) compared with the level remaining in Cep215-depleted cells. Because statistical analysis showed these differences to be highly significant (P<0.001), we conclude that Cep215 plays a minor role in centrosomal recruitment of γ-tubulin during mitosis.
The role of Cep215 in spindle pole formation
If Cep215 is not actively involved in γ-tubulin recruitment to the centrosome, what is the role of Cep215 in centrosome maturation? To address this question, we analyzed the mitotic phenotypes of CHO cells in which Cep215 was deleted (Fig. 8A). Previous studies have shown that the loss of Cep215 in HeLa cells induces abnormal spindles, including monopolar spindles, bipolar spindles with less distance between the two poles, and a lack of prominent astral MTs emanating from the poles (Fong et al., 2008; Lee and Rhee, 2010). We found a wide range of spindle abnormalities (Fig. 8A, panels 2–6), including monopolar or short spindles (∼17%; Fig. 8A, panel 2 and Fig. 8B), which is consistent with previous reports. The proportion of cells with bipolar spindles consisting of sparse MTs (Fig. 8A, panel 3) and thin spindles (Fig. 8A, panel 4) was also increased after Cep215 depletion (∼22%; Fig. 8B). This type of spindle has not previously been documented, possibly because of the shape of mitotic spindles. It would be difficult to identify thin spindles in cells where the distance between two poles is reduced. It is likely that depletion of Cep215 induces more monopolar and short bipolar spindles in HeLa cells (Lee and Rhee, 2010) than in CHO cells.
The most notable feature of spindles formed in Cep215 RNAi cells was the presence of ‘knots’, which were observed inside or protruding outside the spindle depending on the direction from which the spindle was viewed (arrows in Fig. 8A, panels 7–10). Among antibodies tested thus far, we found that NuMA (nuclear mitotic apparatus protein) was the best for probing these structures. As an MT-binding protein interacting with dynein–dynactin, NuMA is important to focus transported spindle MTs on the pole (Merdes et al., 1996). Therefore, the knots labeled with NuMA are likely to correspond to the dislocated centrosomes and spindle poles, which were first reported by Barr et al. (2010) in Cep215 knockout chicken DT40 cells. Split NuMA was detected in cells with multipolar spindles (Fig. 8A, panels 5,11). Some spindles were composed of totally disorganized MT arrays and randomly dispersed NuMA (Fig. 8A, panels 6,12). We observed that ∼38% of Cep215-depleted cells produced extra NuMA-binding sites (Fig. 8B). Even cells with normal-looking bipolar spindles revealed the presence of extra NuMA sites, and the center of many monopolar spindles was occupied by a cluster of NuMA dots of various sizes and shapes (arrows in Fig. 8A, panels 7′,8′). Additional locations where extra NuMA dots were commonly seen are the shoulder of mitotic spindles (Fig. 8A, panel 13), the tip of MT bundles shooting out from the lateral side of the spindle (Fig. 8A, panel 14) and the end of MT bundles outspreading to form the unfocused pole at one side of the spindle (Fig. 8A, panel 15).
Similar effects on split NuMA were also seen in cells from which Cep192 (Fig. 8C) or AurA (Fig. 8D) were depleted. These observations are in good agreement with those of De Luca et al. (2006), who reported that small extra poles are induced in AurA-depleted cells. Overall, the aberrant spindle shapes and random distribution of NuMA were strikingly similar among the three types of knockout cells. Furthermore, split NuMA occured at similar frequencies in those cells (∼32 to ∼37%) (Fig. 8F). When the three proteins were simultaneously depleted (triple RNAi), a significantly increased number of cells had mitotic spindles consisting of highly randomized MT arrays and NuMA distribution (>70%; Fig. 8E,F). The additive effect of Cep215, Cep192 and AurA–pAurA on NuMA dispersion implies that these three molecules function as a protein complex to coalesce NuMA at each spindle pole.
We next asked which domains of Cep215 are responsible for NuMA coalescence. A series of rescue constructs lacking one of four domains of Cep215 (ΔCM1, Δ215N, Δ215M, ΔCM2) were prepared, and they were introduced into cells from which endogenous Cep215 was depleted by RNAi. Each deleted domain is capable of binding to specific molecules: CM1 can bind to γ-tubulin and γ-TuRC; 215N can specifically bind to Cep192 and pAurA, and HSET-binding sequences overlap with 215N; 215M can interact with dynein–dynactin; and CM2 can associate with PCNT (Fig. 8H). We found that ∼5% of mock-treated and ∼50% of RNAi-treated cells induced abnormal spindles with split NuMA. Introduction of myc–Cep215 into Cep215 RNAi cells reduced the number of cells with abnormal spindles to ∼13% (Fig. 8G). Expression of myc–Cep215 lacking 215N rescued the phenotype. However, centrosome–pole separation was still seen in almost 33% of mitotic cells, which is statistically significant (P<0.05). Besides Δ215N, other deletion constructs (ΔCM1, Δ215M and ΔCM2) also partially restored NuMA coalescence at strikingly similar efficiencies (30–35%) to that of Δ215N. Because each domain binds the specific partner required for tethering the ends of MTs to the PCM (Fig. 8H), the role of Cep215 in centrosome–spindle pole connection is likely achieved by the collaboration of multiple binding sites spanning the entire length of the Cep215 sequence.
The 215N domain is a newly identified mitosis-specific centrosome-targeting domain of Cep215 that interacts with Cep192 and pAurA. Cep215 appears to be recruited to the mitotic centrosome via multiple pathways. CM2 has previously been considered the only domain required for targeting Cep215 to interphase and mitotic centrosomes (Barr et al., 2010; Wang et al., 2010). CM2 interacts with PCNT (Buchman et al., 2010; Wang et al., 2010), and Cep215 and PCNT require each other to localize to the centrosome (Kim and Rhee, 2014). Here, we show that centrosome targeting of Cep215 via 215N is dependent on Cep192 (Fig. 6E). Because no Cep215-binding domain(s) of Cep192 has as yet been identified, it is not known whether centrosome targeting of Cep192 is independent or partially dependent on Cep215. In addition, Cep192 and PCNT have been shown to be mutually dependent (Gomez-Ferreria et al., 2007; Zhu et al., 2008). This suggests that the mechanism of centrosome localization of these three core proteins is under a complex three-way control. The contribution of individual pathways for targeting each molecule may vary among cell types, species, or even among the same cells under different physiological conditions.
It has recently been shown that Cnn oligomerizes through the interaction of CM2 with the leucine zipper-containing sequence located in the middle of Cnn (Feng et al., 2017; Citron et al., 2018). Apparently, this central sequence of Cnn, termed PReM (phosphoregulated multimerization domain; Conduit et al., 2014), is well-aligned to 215N of Cep215, in which the leucine zipper motif is conserved (Feng et al., 2017). However, the probability that 215N is targeted to the mitotic centrosome via CM2 interaction is low, because 215N is still able to locate to mitotic centrosomes in cells from which CM2-containing endogenous Cep215 has been deleted (Fig. 1D, panel 7). Therefore, the centrosomal target of 215N is, at least partially, independent of CM2. Is 215N, or a domain functionally equivalent to 215N, conserved among Cnn family members? Although fluorescence signals at the centrosome are weak, Cep215 or Cnn devoid of CM2 can localize to the spindle pole in human (Kim and Rhee, 2014), chicken (Barr et al., 2010) and Drosophila (Feng et al., 2017) cells. These results indirectly support the presence of a centrosome-targeting domain other than CM2. More directly, Barrera et al. (2010) and Sukumaran et al. (2017) have reported that the N-terminal sequences of Cep215 in mouse (amino acids 1–435) and human (amino acids 1–580) can be at the interphase centrosome in MEFs derived from a Cep215 knockout mouse and HeLa cells, respectively. As summarized in Fig. S3A, the mouse clone constructed by Barrera et al. corresponds to the sequence of the CM1 and 215N-N2 regions. The human clone reported by Sukumaran et al. encodes CM1 (amino acids 1–165) plus an additional ∼410 amino acids, a region shorter than 215N (amino acids 166–696) but longer than 215N-N1 (amino acids 166–508) (Fig. S3A). We confirmed that polypeptides from three constructs, consisting of CM1 plus 215N, CM1 plus 215N-N1, or CM1 plus 215N-N2, are targeted to the centrosome. In contrast to the previous two reports, however, they are found there only in mitotic cells, and not in interphase cells. Although the question of cell cycle dependency is still open, these results indicate the presence of multiple centrosomal targeting domains of Cep215 in mammalian cells other than those of mice.
PCNT plays a major role in construction of the mitotic PCM lattice in mammalian cells (Lawo et al., 2012). In contrast, a role for the Drosophila Pericentrin-like protein (D-PLP) appears to be more limited than PCNT, because its localization is restricted to the outermost region of centrosomes (Fu and Glover, 2012; Mennella et al., 2012). The impact of its loss-of-function mutation is also limited to weakening the outer structure of the PCM (Richens et al., 2015). Assembly of mature centrosomes in Drosophila embryos is predominantly controlled by Cnn (Conduit et al., 2010), which forms PCM scaffold-like micron-scale structures (Feng et al., 2017; Citron et al., 2018). Cnn interacts with D-PLP at two distinctive sites (Lerit et al., 2015). In contrast, CM2 is the only PCNT-binding site identified so far in Cep215 (Buchman et al., 2010; Wang et al., 2010). This difference between Cnn and Cep215 may be attributed to their distinct functions in PCM assembly.
AurA and Plk1 orchestrate the process of nuclear and cytoplasmic divisions by controlling a series of mitotic events, including centrosome maturation. Both Plk1 and AurA have long been postulated to be essential for increasing the MT number associated with the mitotic centrosome (Lane and Nigg, 1996; Hannak et al., 2001; Berdnik and Knoblich, 2002; Haren et al., 2009; Joukov et al., 2014). We showed previously that Cnn interacts with AurA (Terada et al., 2003). Here, we specify that phosphorylated AurA interacts with Cep215 via 215N (Figs 3,4). The centrosomal targeting of 215N depends on pAurA (Fig. 4), and MLN8237 interferes with pAurA localization to the centrosome (Fig. 5A, panel 5′). It is thus reasonable that 215N is blocked from localizing to mitotic centrosomes in MLN8237-treated cells due to the absence of pAurA at the centrosome (Fig. 5A, panel 8). The 215N domain is also efficiently removed from the centrosome by treatment with BI2536 (Fig. 5A, panel 9). This is likely due to a loss of centrosomal pAurA caused by Plk1 inhibition (Fig. 5A, panel 6′; De Luca et al., 2006; Hanisch et al., 2006). Plk1 is known to phosphorylate PCNT, which is essential for recruitment of several PCM proteins, including Cep192, to the centrosome (Lee and Rhee, 2011). Because 215N requires Cep192 to localize to the centrosome, Plk1 inactivation inhibits the centrosome targeting of 215N due to the absence of Cep192. It is noteworthy that Cep215 also relies on PCNT to localize to the centrosome. However, this does not require Plk1 phosphorylation of PCNT (Lee and Rhee, 2011). This is consistent with our observation that the CM2-dependent centrosome targeting of Cep215 was not affected by BI2536 (Fig. 5A, panel 12 and Fig. 5B). Another possibility is a direct effect of Plk1 on 215N. It has recently been shown that Plk1 phosphorylates PReM, the central region of Cnn, which is essential for organization of the Cnn scaffold of fly centrosomes (Conduit et al., 2014). Because this central domain corresponds to 215N of Cep215 (Feng et al., 2017), BI2536 may inhibit 215N recruitment to the mitotic centrosome by interfering with oligomerization of Cep215 and assembly of PCM scaffolds.
In mitotic cells, 215N bound to pAurA and Cep192 (Fig. 3C, Fig. 6C), and Cep192 is known to interact with both AurA and pAurA (Joukov et al., 2014). It is thus highly probable that Cep215 associates with AurA indirectly via Cep192. However, Joukov et al. reported that Cep192, but not Cep215, is pulled down in frog extracts with beads coated with anti-AurA antibodies. This raises the possibility that co-precipitation of Cep215 with anti-AurA-antibody-coated beads was more limited than that of Cep192 with anti-AurA beads, although it is unknown what percentage of beads were covered with phosphorylated AurA. It is possible that the amount of Cep215 pulled down by beads may have been below the detection level. Identification of the Cep192/pAurA-binding domain of Cep215 implies that Cep215 is a core component of the signaling cascade organized by Cep192 to induce centrosome maturation described by Joukov et al. (2014). AurA and Plk1 are bound to Cep192 and transported to the centrosome in a PCNT-dependent manner, where AurA becomes activated by autophosphorylation at T288. Active AurA (pAurA) phosphorylates Plk1, which in turn causes phosphorylation of Cep192 to generate multiple attachment sites for γ-TuRCs. Delivering the Cep192–kinases complex to centrosomes is achieved by Plk1-phosphorylated PCNT (Lee and Rhee, 2011). To the best of our knowledge, however, no evidence of direct binding of Cep192 and PCNT has been presented. Rather negative/almost negative immunoprecipitation was reported between the Cep192 homolog DSpd-2 and the PCNT homolog D-PLP in fly embryos (Conduit et al., 2010). It is thus possible that Cep215 plays a role in carrying Cep192 to the centrosome, either independently or in collaboration with PCNT, via its 215N domain for binding Cep192 and its CM2 domain for binding PCNT. Upon delivery to the mitotic centrosome, both Cep192 and pAurA become firmly attached to the PCM lattice through Cep215/CM2–PCNT interactions. Furthermore, assembly of Cep215 scaffolds promoted by Plk1 may facilitate recruitment of the Cep192–kinase complex to the expanding PCM. Subsequent accumulation of active AurA creates the γ-tubulin binding sites on Cep192. The 215N domain and pAurA are co-dependent on their localization at the centrosome. This suggests that interaction of AurA with 215N may initiate and/or promote AurA phosphorylation at the centrosome. Alternatively, 215N ensures preferential localization of phosphorylated AurA over non-phosphorylated AurA at the mitotic centrosome.
We propose that the function of Cep215 in connecting the mitotic centrosome and spindle poles is achieved by collaborative efforts of individual subdomains of Cep215 through their ability to interact with different molecules important for assembly of functional spindles and spindle poles (Fig. 8H). MTs become massively associated with mitotic centrosomes as a result of increased MT-nucleating activity of the centrosome. This activity is primarily achieved by Cep192 and pAurA (Joukov et al., 2014), to which Cep215 binds via 215N. Acentrosomal MTs are transported by motor proteins, and their minus ends are coalesced to focus on the centrosome at each spindle pole. Cep215 is able to interact with dynein–dynactin and HSET, and thus is likely to anchor acentrosomal MT ends to the centrosome via its binding to the motor proteins. Bringing the two types of MTs close together, Cep215 next connects all MT ends tightly to the PCM lattice by interaction with PCNT via CM2. CM1 binds to γ-tubulin, which allows Cep215 to associate with γ-tubulin and γ-TuRCs located at the minus ends of both centrosomal and acentrosomal MTs (Meunier and Vernos, 2016; Prosser and Pelletier, 2017). It is thus plausible that CM1 participates in anchoring the MT ends at the centrosome along with other domains. It is likely that Cep215 acts to connect the mitotic centrosome to each spindle pole by serving as a platform for the molecules involved in centrosome maturation and bipolar spindle formation. Because individual domains are important for the centrosome–spindle pole connection, it is reasonable to detect the same phenotype of centrosome–spindle pole separation in cells depleted of each Cep215 domain (Fig. 8G; Barr et al., 2010; Chavali et al., 2016). A lack of structural integrity of the mitotic centrosome and spindle poles may disrupt centrosomal cohesion, split centrioles and displace centrioles from the PCM (Lucas and Raff, 2007; Barrera et al., 2010), all of which have been identified as unique phenotypes of Cep215 and Cnn knockout cells.
We probed the minus end of spindle MTs with NuMA, and Cep215 interacts with NuMA indirectly through MTs–γ-tubulin and dynein–dynactin. Haren and Merdes (2002) reported that ectopic NuMA expression induces multiple spindle poles. Overexpression of NuMA may create a situation similar to Cep215 depletion in cells where excess NuMA fails to tether into the centrosome due to the paucity of centrosomal Cep215. It would be interesting to know whether there are specific Cep215 domains that directly interact with NuMA to connect the MT minus end to the mitotic centrosome.
MATERIALS AND METHODS
Cell culture, synchronization, and drug treatment
CHO and human (RPE1, U2OS, HeLa and HEK293T) cells were cultured in Ham's F-10 and DMEM-GlutaMAX (Gibco) medium containing 10% FBS and antibiotics. GFP tagged CETN-expressing CHO, RPE1 and HeLa cells were described previously (Ohta et al., 2002; Steere et al., 2012). For synchronization of mitotic cells, CHO cells were treated with RO3306 (Alexis Biochemicals) for 5 h at a final concentration of 10 µM, followed by incubation with fresh F-10 medium for 20–40 min to prepare cells at different mitotic stages. To inhibit AurA and Plk1, MLN8237 (0.5–1.0 µM; Selleckchem) and BI2563 (0.1–1.0 µM; Axon Medchem) were added to cells at 3.5 h after incubation with RO3306 and further cultured for an additional 1.5 h. After washing out RO3306, cells were cultured for 30–60 min in the presence of kinase inhibitors before fixation.
Plasmid preparation, RNAi and transfection
Mouse cDNA encoding full-length Cep215 (GenBank accession: AK129411) and human Flag-tagged Cep192 (NP_115518) were obtained from Timothy Megraw (Florida State University, Tallahassee, FL) and Laurence Pellitier (University of Toronto, Canada), respectively. Myc-tagged full-coding and deletion constructs of Cep215 were generated by PCR amplification and subcloned into an N-terminally 3×myc-tagged pCS2 vector obtained from Jeffrey R. Miller, Estrella Mountain Community College, Avondale, AZ. For generation of individual constructs, we used the following internal restriction sites and primers specific to each nucleotide position: full length (1–5,466), CM1 (1–495: XbaI), 215N (495: XbaI-2089: NheI), 215M (2089: NheI-4417: XhoI), CM2 (4417: XhoI-5466), 215N-N1 (495: XbaI-1524: XhoI), 215N-N2 (495: XbaI-1305), 215N-N3 (603–1305), 215N-C (1305–2089: NheI), CM1+215N (1–2089: NheI), CM1+215N-N1 (1–1524: XhoI), CM1+215N-N2 (1–1305), ΔCM1 (495: XbaI-5466), Δ215N (1–495: XbaI fused with 2089: NheI-5466), Δ215M (1–2089: NheI fused with 4417: XhoI-5466), ΔCM2 (1–4417: NheI), 215N/CM2 (495: XbaI-2089: NheI fused with 4417: NheI-5466), Δ215N/ΔCM2 (1–495: XbaI fused with 2089: NheI-4417: XhoI). cDNA encoding human AurA (GenBank accession: NM_001323305) was cloned as previously described (Terada et al., 2003) and ligated into a GFP-tagged pCS2 vector (Jeffrey R. Miller). The non-phosphorylatable AurA mutant (T288A) was generated using the protocol of DpnI mediated site-directed mutagenesis (Steere et al., 2012). For RNAi, we applied the following species-specific siRNAs: Cep215 (C-terminus specific), nucleotide positions 3417–3441, 5′-GGACCAUAUUGAUGAAGAAGAGAGG-3′; Cep215 (N-terminus specific), nucleotide positions 609–627, 5′-GGACAGACUGAUUGAGGAG-3′; Cep192, nucleotide positions 7160–7178, 5′-CUAAAGAGCCUCACAUGAA-3′; AurA, nucleotide positions 164–182, 5′-AGAAAGCUGUCUCAGGUCA-3′. For mock transfection, control siRNA (sc-3707, Santa Cruz Biotechnology) was used.
For transfection, cells were seeded on coverslips in a 12-well plate 1 d before and treated with plasmid DNA and/or siRNA using Lipofectamine for CHO and HEK293T cells, and Lipofectamine 2000 for other types of human cells, as previously described (Ohta et al., 2002; Steere et al., 2012). For expression of truncated Cep215 polypeptides, plasmid DNAs were reduced to 20–50% (0.2–0.25 µg of 215N DNA and 0.4 µg of other DNA clones per well of a 12-well plate) of the normal amount (0.83 µg per well).
Cells cultured on coverslips were fixed with methanol at −20°C. After rehydration in phosphate-buffered saline (PBS) containing 0.05% Tween-20, cells were incubated for 30 min at 37°C with the following primary antibodies: two anti-Cep215 pAbs raised against the mouse Cep215 sequence (amino acids 24–278; a gift from Timothy Megraw) and the human Cep215 sequence at amino acids 1307–1382 (HPA035820, Sigma-Aldrich), anti-Cep192 pAb (a gift from Laurence Pelletier), anti-AurA mAb (35C1, Invitrogen), anti-pAurA pAb at T288 (ab83968, Abcam Biotechnology; #3091, Cell Signaling Technology), anti-NuMA (pAb, Maekawa and Kuriyama, 1993; human auto-antibodies, Maekawa et al., 1991), anti-γ-tubulin mAb (Sigma-Aldrich, clone GTU-88), anti-PCNT (pAb, PRB-432C, Covance; human auto-antibodies, a gift from Stephen Doxsey, University of Massachusetts Medical School, Worcester, MA), anti-Cep135 (mouse pAb and rabbit pAb, Ohta et al., 2002), anti-HSET mAb (Kuriyama et al., 1995), anti-α-tubulin mAb (Sigma-Aldrich, clone B-5-1-2), anti-phospho-histone H3 at Ser10 (pAb, #06-570, Upstate Biotechnology; mAb, #9706, Cell Signaling Technology), anti-c-Myc (mAb, 9E10, Santa Cruz Biotechnology; pAb, CM-100, Gramsch Laboratories), anti-Flag mAb (M2, Sigma-Aldrich), anti-HA mAb (12CA5, Roche Diagnostics) and anti-GFP pAbs (sc-8334, Santa Cruz Biotechnology; #8363, Clontech Laboratories, a gift from Ken-Ichi Takemaru, Stony Brook University, Stony Brook, NY). Primary antibodies were used at a dilution of 1:300 to 1:5000. Fluorescein-conjugated anti-mouse IgG plus IgM and Texas Red-conjugated anti-rabbit IgG antibodies were used as the secondary antibodies (Jackson ImmunoResearch). For triple staining with GFP, we used Cy3- and Cy5-conjugated anti-mouse and anti-rabbit secondary antibodies (Molecular Probes).
Immunoprecipitation and western blotting
Immunoprecipitation was carried out as reported previously (Steere et al., 2012). Briefly, transfected HEK293T cells with Flag–Cep192, myc–215N, GFP–AurA, and GFP–mutant AurA (T288A) were washed three times with PBS and collected into ice-cold lysis buffer (20 mM Tris-HCl, pH 7.2, 135 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, 1% Triton X-100 and 10% glycerol) containing 1 mM Na3VO4 and a mixture of protease inhibitors (20 μM leupeptin, 0.4 μM chymostatin, 1–2.5 mM PMSF and 15 μM aprotinin; Sigma-Aldrich). To enrich mitotic populations, cells were treated with 0.25 µg/ml nocodazole for 6–10.5 h. After centrifugation at 13,000 g for 10 min at 4°C, the supernatants were co-immunoprecipitated with anti-myc and anti-Flag mAbs. For immunoprecipitation with GFP antibodies, magnetic beads conjugated with rat anti-GFP mAb (Medical & Biological Laboratories) were used. After separation on 7–7.5% SDS-PAGE gels, bands were visualized by blotting with primary antibodies as indicated for each figure, followed by incubation with alkaline phosphatase-conjugated secondary antibodies (Hyclone Laboratories) and BCIP/NBT chromogen (Ohta et al., 2002). Primary antibodies used were: anti-Cep215 pAb (a gift from Timothy Megraw; 1:10,000), anti-Cep192 pAb (a gift from Laurence Pelletier; 1:500), anti-AurA mAb (35C1, Invitrogen; 1:3000), anti-pAurA pAb (ab83968, Abcam; 1:1000), anti-GFP pAb (sc-8334, Santa Cruz Biotechnology; 1:1000) and anti-c-Myc mAb (9E10, Santa Cruz Biotechnology; 1:1000).
Microscopy and statistical analysis of fluorescence intensity at mitotic centrosomes
Microscopic observations were carried out using a Nikon Eclipse microscope with a 100× oil immersion objective (NA 1.4) and a Photometrics CoolSNAP camera. Image slices (0.2 µm) were merged using the SlideBook 4.1 program. Some images were processed through deconvolution using the program included in the SlideBook software. All images were exported to Adobe Photoshop (CS5) to acquire 8-bit files.
To measure the fluorescence intensity of the centrosomal proteins/truncated Cep215 at spindle poles (Fig. 2C, Fig. 5B and Fig. 7B), immunostained cells were kept in the dark at −20°C for several days to ensure stabilization and equilibration of the mounting medium. Images of individual spindle poles were captured under identical exposure conditions. During measurements of γ-tubulin shown in Fig. 7B, we observed that depletion of Cep192 and pAurA by RNAi was less efficient than depletion of Cep215. Cells that were not obviously depleted of Cep192 were eliminated from quantification of fluorescence intensity. After subtraction of background fluorescence, fluorescence intensity was quantified in arbitrary units (a.u.) using the ImageJ/Fiji image processing program, and individual data points were plotted using the GraphPad Prism software. The mean and s.e.m. were calculated and are indicated as long and short bars in each frequency distribution. The signal intensity of individual Cep215 domains was lower than that of the full-coding sequence, suggesting that no inhibitory sequences are included in Cep215 (Fig. 2C). In addition, treatment with kinase inhibitors and individual depletion of Cep215, Cep192 and AurA were reported to reduce the fluorescence intensity of γ-tubulin at spindle poles (Figs 5B and 7B). Because a one-tailed z-test has more statistical power than a two-tailed z-test against the alternative hypothesis that treatments decrease signal intensity, we performed one-tailed z-tests for assessment of statistical significance by comparing the mean of each treated group with the control mean.
We thank Drs T. Megraw and L. Pelletier for sharing Cep215 and Cep192 reagents.
Conceptualization: R.K.; Methodology: R.K., C.R.F.; Software: R.K., C.R.F.; Validation: R.K., C.R.F.; Formal analysis: R.K., C.R.F.; Investigation: R.K., C.R.F.; Resources: R.K., C.R.F.; Data curation: R.K., C.R.F.; Writing - original draft: R.K.; Writing - review & editing: R.K., C.R.F.; Visualization: R.K.; Supervision: R.K.; Project administration: R.K.; Funding acquisition: R.K.
This work was supported by a grant to R.K. from the National Science Foundation (MCB1140033).
Peer review history
The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.240267.reviewer-comments.pdf
The authors declare no competing or financial interests.