Formation of a new centriole adjacent to a pre‐existing centriole occurs only once per cell cycle. Despite being crucial for genome integrity, the mechanisms controlling centriole biogenesis remain elusive. Here, we identify RBM14 as a novel suppressor of assembly of centriolar protein complexes. Depletion of RBM14 in human cells induces ectopic formation of centriolar protein complexes through function of the STIL/CPAP complex. Intriguingly, the formation of such structures seems not to require the cartwheel structure that normally acts as a scaffold for centriole formation, whereas they can retain pericentriolar material and microtubule nucleation activity. Moreover, we find that, upon RBM14 depletion, a part of the ectopic centriolar protein complexes in turn assemble into structures more akin to centrioles, presumably by incorporating HsSAS‐6, a cartwheel component, and cause multipolar spindle formation. We further demonstrate that such structures assemble in the cytoplasm even in the presence of pre‐existing centrioles. This study sheds light on the possibility that ectopic formation of aberrant structures related to centrioles may contribute to genome instability and tumorigenesis.
New centrioles form by a highly regulated and ordered assembly pathway, starting from a distinctive “cartwheel” structure. Identification and characterization of the novel centriole biogenesis regulator RBM14 uncovers a non‐canonical, cartwheel‐independent pathway that allows procentriole formation even in the presence of pre‐existing centrioles.
RBM14 is an interactor of the human centriole assembly factor STIL.
RBM14 interferes with complex formation of STIL and CPAP during centriole assembly.
RBM14 knock‐down leads to ectopic formation of functional centriolar protein complexes in cells.
De novo assembly of centriolar protein complexes occurs without a cartwheel structure.
RBM14 depletion affects spindle assembly and chromosome segregation during mitosis.
Centrosome duplication is tightly orchestrated with cell cycle progression to yield the correct number of centrosomes, which ensures robust formation of a mitotic bipolar spindle and faithful chromosome segregation (Nigg & Raff, 2009). The centrosome consists of a pair of centrioles surrounded by an electron‐dense pericentriolar material (PCM). Formation of a new centriole adjacent to each pre‐existing centriole is fundamental for centrosome duplication. Centriole formation starts with the assembly of a cartwheel structure onto which microtubules are then added (Brito et al, 2012; Gönczy, 2012). Besides this canonical pathway, the de novo formation of centrioles can be induced in the case of natural/physical loss of pre‐existing centrioles (Marshall et al, 2001; Khodjakov et al, 2002; La Terra et al, 2005), in unfertilized Drosophila eggs overexpressing core centriole duplication factors (Peel et al, 2007; Rodrigues‐Martins et al, 2007; Dzhindzhev et al, 2010; Stevens et al, 2010a), during early development in the mouse embryo (Courtois & Hiiragi, 2012) or in multicilia formation of terminally differentiated cells (Nigg & Raff, 2009). Although it is believed that the existence of pre‐existing centrioles normally suppresses the de novo assembly in proliferating cells, exactly how this suppression is achieved remains unknown.
The SAS‐6 family of proteins have been recently identified as crucial components of the cartwheel that is essential for centriole formation (Kilburn et al, 2007; Nakazawa et al, 2007; van Breugel et al, 2011; Kitagawa et al, 2011b). Moreover, human SAS‐6 (HsSAS‐6) overexpression induces formation of multiple procentrioles around a parental centriole, presumably through formation of multiple cartwheels (Strnad et al, 2007). Similar phenotypes have also been observed for overexpression of STIL (SCL/TAL1 interrupting locus) (Kitagawa et al, 2011a; Tang et al, 2011; Arquint et al, 2012; Vulprecht et al, 2012) and polo‐like kinase 4 (Plk4) (Kleylein‐Sohn et al, 2007), implying that there is a tight relationship between these critical components of the centriole assembly. In fact, in human cells, Plk4 acts upstream of HsSAS‐6, regulating the presence of centriolar HsSAS‐6 (Kleylein‐Sohn et al, 2007; Strnad et al, 2007). In line with the fact that both HsSAS‐6 and STIL localize to the proximal end of procentrioles (Kleylein‐Sohn et al, 2007; Strnad et al, 2007; Tang et al, 2011; Arquint et al, 2012; Vulprecht et al, 2012), HsSAS‐6 and STIL seem to be dependent on one another for their presence at procentrioles (Tang et al, 2011; Arquint et al, 2012; Vulprecht et al, 2012). HsSAS‐6 and STIL are needed in turn for CPAP loading (Kohlmaier et al, 2009; Tang et al, 2009, 2011), which presumably promotes subsequent addition of centriolar microtubules and the elongation of procentrioles (Kohlmaier et al, 2009; Schmidt et al, 2009; Tang et al, 2009). Consistently, the induction of procentriole amplification by Plk4 overexpression is mediated by HsSAS‐6 and STIL (Kleylein‐Sohn et al, 2007; Tang et al, 2011; Arquint et al, 2012; Vulprecht et al, 2012). Given that the function of their relatives in centriole formation is evolutionarily conserved (Dammermann et al, 2004; Delattre et al, 2004; Leidel et al, 2005; Kilburn et al, 2007; Nakazawa et al, 2007; Rodrigues‐Martins et al, 2007; Yabe et al, 2007; Culver et al, 2009; Stevens et al, 2010a) and that their relationship in other systems seems to be, at least in part, conserved (Delattre et al, 2006; Pelletier et al, 2006; Kitagawa et al, 2009; Stevens et al, 2010b), these three proteins are critical for the onset of centriole assembly. To conduct faithful centriole duplication, the function and expression levels of the critical components must be strictly regulated across the cell cycle.
In this study, we identify RNA‐Binding Motif protein 14 (RBM14) as a STIL‐interacting protein that interferes with a complex formation of STIL and CPAP. In addition, our data establish that RBM14 suppresses the formation of aberrant structures related to centrioles in the cytoplasm and thus preserves mitotic spindle integrity.
Depletion of RBM14 induces ectopic formation of centriolar protein complexes
We set out to investigate the mechanisms limiting the number of newly formed centrioles. We reasoned that such regulators could physically interact with and directly control critical centriole proteins such as HsSAS‐6 and STIL. Accordingly, we employed the mass spectrometry‐based proteomic analysis to identify candidate proteins interacting with HsSAS‐6 or STIL. Small‐scale RNAi screening was subsequently conducted to address their function in centriole formation in human culture cells. This screening led to the identification of RNA‐Binding Motif protein 14 (RBM14, also known as COAA, transcription co‐activator, or PSP2, paraspeckle protein 2) (Iwasaki et al, 2001; Fox et al, 2002), a potential tumor suppressor conserved within vertebrates (Kang et al, 2008), as a STIL‐interacting protein (Supplementary Table S1). We analyzed mitotic cells, which normally have four centrioles marked with centrin signals and assemble a bipolar spindle (Fig 1A). Intriguingly, we discovered that siRNA‐mediated depletion of RBM14 resulted in a significant increase of centrin foci in human U2OS (33 ± 2% compared with 7 ± 2% in control cells, P < 0.01, n = 90 in triplicate; Fig 1A and Supplementary Fig S1A), HeLa cells (19 ± 1% compared with 1 ± 1% in control cells, P < 0.01, n = 90 in triplicate; Supplementary Fig S1B) and RPE1 cells (17 ± 4% compared with 3 ± 2% in control cells, P < 0.05, n = 90 in triplicate; Supplementary Fig S1C). We also found a similar phenotype in interphase cells (Supplementary Fig S1D). The amplification of centrin foci was detectable within 24 h after siRNA transfection in U2OS cells without a significant cell cycle arrest or cytokinesis defect (Supplementary Fig S1E and F). It has been previously reported that RBM14 acts as a component of the nuclear paraspeckle and functions in RNA processing and transcription (Auboeuf et al, 2004). The nuclear paraspeckle is a small intranuclear structure that is known to contain RNA and several ribonucleoproteins. We therefore assessed the possibility that RBM14 depletion somehow affects centriole formation through transcriptional regulation. However, mis‐localization of RBM14 from the paraspeckle caused by depletion of p54nrb/NONO, a protein essential for paraspeckle organization (Sasaki et al, 2009), did not induce amplification of centriole foci (Supplementary Fig S1H). In addition, expression levels of STIL and other centriole proteins were unaffected by RBM14 depletion (Supplementary Figs S1A, S4A, S5A, and S9B). Furthermore, incubation with 0.5 mM cycloheximide, a translation inhibitor, did not suppress excess formation of centrin foci by RBM14 depletion (Supplementary Fig S1G). Based on these observations, we conclude that the amplification of centrin foci by RBM14 depletion likely stems from reduction of unknown function of RBM14 rather than its known function as a component of the nuclear paraspeckle.
Next, to distinguish whether the amplification of centrin foci is because of multiple rounds of centriole duplication or concurrent formation of multiple procentrioles and/or procentriole‐like premature structures, we monitored the number of C‐Nap1 or Cep164‐labeled parental centrioles (Fry et al, 1998; Graser et al, 2007) and Centrobin, a procentriole marker (Zou et al, 2005), foci in mitotic RBM14‐depleted cells. We found that 31 ± 4% of the cells harbored an increased number of Centrobin foci (P < 0.05, n = 90 in triplicate; Fig 1B), whereas the number of parental centrioles was not statistically affected compared with control cells (Fig 1A and Supplementary Fig S1I). These results indicate that the ectopic centrin foci in RBM14‐depleted cells likely represent amplification of procentrioles or procentriole‐like premature structures, or both, rather than multiple rounds of centriole duplication.
We also noted that most of these structures contained other centriolar proteins such as CPAP (Hung et al, 2000) (65 ± 3%, n = 150 in triplicate; Supplementary Fig S1J) and CP110 (Chen et al, 2002) (55 ± 3%, n = 150 in triplicate; Supplementary Fig S1J) that constitute the middle‐distal parts of centrioles. Consistently, we found significant increase in the number of CPAP and CP110 foci in the cells depleted of RBM14 (23 ± 2% compared with 0% in control cells, 30 ± 2% compared with 6 ± 2% in control cells, respectively; for both cases, P < 0.01, n = 90 in triplicate; Fig 1C and Supplementary Fig S1K). Furthermore, we observed ectopic foci of acetylated tubulin in the cells depleted of RBM14 (11 ± 1% compared with 1 ± 1% in control cells, P < 0.01, n = 90 in triplicate; Fig 1D) and also found that some of ectopic centrin‐containing structures appeared to incorporate acetylated tubulin (26 ± 2%, n = 150 in triplicate; Supplementary Fig S1J), suggesting that they contain the centriolar microtubules. We next tested whether the ectopic centrin‐containing structures are functional enough to assemble PCM. Upon depletion of endogenous RBM14, the ectopic centrin‐containing structures efficiently accumulated PCM proteins such as γ‐tubulin and Cep192 (Stearns et al, 1991; Zheng et al, 1991; Gomez‐Ferreria et al, 2007; Zhu et al, 2008) (39 ± 2% and 41 ± 2%, respectively, n = 150 in triplicate; Supplementary Fig S1J). In line with this, we found significant increase in the number of ectopic PCM formation (22 ± 3% compared with 4 ± 1% in control cells for γ‐tubulin, P < 0.01, n = 90 in triplicate; 22 ± 2% compared with 3 ± 2% in control cells for Cep192, P < 0.05, n = 90 in triplicate; Fig 1E and Supplementary Fig S1L). Based on these observations, we refer to the ectopic structures containing centriolar proteins and induced in RBM14‐depleted cells as “centriolar protein complexes” in this article.
Cytoplasmic RBM14 can inhibit centriole amplification in cells arrested in S phase
To confirm that the appearance of ectopic centriolar protein complexes specifically results from loss of RBM14 and also to identify the region of RBM14 required for its function, we tested whether full‐length (FL) and/or fragments of RBM14 can rescue the phenotype provoked by treatment with siRNA against the 3′ UTR targeting solely endogenous RBM14. We found that expression of the C‐terminal half of RBM14 (RBM14[C]) lacking the RNA‐recognition motif (RRM) as well as RBM14 FL markedly suppressed the amplification of centriolar protein complexes in the cells depleted of endogenous RBM14 (Fig 2A and B, and Supplementary Fig S2A).
Next, to further investigate the function of RBM14 in normal centriole formation or centriole re‐duplication in cells arrested in S phase, we ectopically expressed RBM14 FL and RBM14[C] in U2OS cells. Although overexpression of RBM14 had only a minor effect on normal centriole duplication and PCM assembly (Supplementary Fig S2B and C) or formation of multiple procentrioles induced by overexpression of Plk4, STIL or HsSAS‐6 (Supplementary Fig S2D), we found that it significantly inhibited centriole amplification in cells treated with hydroxyurea (HU) to prolong S phase, judged by the frequency of cells with more than 4 centrin foci (26 ± 3% in cells expressing RBM14 FL, 27 ± 2% in cells expressing RBM14[C] compared with 53 ± 4% in control cells, P < 0.05, n = 90 in triplicate; Fig 2C). We noted in addition that ectopically expressed RBM14 FL proteins mainly existed in the nucleus, but also localized in the cytoplasm as is the case for the endogenous proteins (Fig 2C and Supplementary Fig S2A, E and F). By contrast, we found that RBM14[C] could occasionally localize in the vicinity of centrioles during interphase (14.4 ± 2.9% compared with 2.2 ± 1.1% of RBM14 FL and 1.1 ± 1.1% of endogenous RBM14, n = 90; Fig 2C and Supplementary Fig S2E), implying that this region may bind to a centriolar protein. These observations prompted us to investigate where RBM14 could act on centriole biogenesis. To address this, we manipulated the localization of RBM14 by fusing NES (nuclear export signal) or PACT domain, a centrosomal targeting motif, to the C‐terminal end of RBM14 (Fig 2B). Intriguingly, we found that when overexpressing RBM14‐NES which localized to the cytoplasm, but seemingly not to centrioles (only 2.2 ± 1.1% of RBM14‐NES in the vicinity of centrioles during interphase, n = 90; Fig 2D), HU‐induced centriole amplification was efficiently suppressed as is the case for overexpression of the native full‐length protein (21 ± 1% in cells expressing RBM14‐NES compared with 28 ± 3% in cells expressing RBM14, P < 0.05, n = 90 in triplicate; Fig 2D). On the other hand, overexpression of RBM14‐PACT localizing to centrioles (33.3 ± 3.8%, n = 90; Fig 2D) and in the cytoplasm also suppressed the centriole amplification, but not more effectively (34 ± 4%, n = 90 in triplicate; Fig 2D), suggesting that the centriolar loading of RBM14 itself seems not to be essential for its ability to suppress centriole amplification. Furthermore, we established that expression of RBM14‐NES suppressed the phenotype provoked by depletion of endogenous RBM14 (Supplementary Fig S2G). Overall, these data lead us to propose that cytoplasmic localization of RBM14 is crucial for its function to suppress amplification of centrioles or ectopic centriolar protein complexes.
RBM14 interacts with STIL and prevents a complex formation of STIL and CPAP
We next confirmed by co‐immunoprecipitation experiments with the endogenous proteins that RBM14 is a bona fide STIL‐binding protein in vivo (Fig 3A and Supplementary Fig S3A). On the other hand, we could not detect interaction between endogenous STIL and CPAP proteins in these experiments. Moreover, yeast two‐hybrid, GST pull‐down and co‐immunoprecipitation assays using full‐length and fragments of STIL and RBM14 established that the N‐terminal region of STIL (STIL[N]) directly bound to the C‐terminal region of RBM14, which is crucial for the ability of RBM14 to suppress the formation of ectopic centriolar protein complexes (Fig 3B and C, and Supplementary Fig S3B–D). Furthermore, using GST pull‐down assays with several deletion mutants of RBM14[C], we determined that the TRBP (thyroid hormone receptor‐binding protein)/Ncoa6‐interacting domain (307–584 aa) (Iwasaki et al, 2001) is responsible for STIL–RBM14 binding (Fig 3C and Supplementary Fig S3E and F). This domain is composed of a serine/alanine/glycine/tyrosine‐rich region and thought to be a protein–protein binding domain of RBM14. Consistently, functional analysis revealed that expression of the RBM14 mutant proteins that lack the TRBP‐interacting domain did not efficiently rescue the phenotype provoked by depletion of endogenous RBM14 (Supplementary Fig S3G), suggesting that direct binding of RBM14 to STIL through the TRBP‐interacting domain is required for the function of RBM14 suppressing ectopic formation of centriolar protein complexes. These data prompted us to examine whether this interaction influences the expression levels or centriolar localization of STIL, or both. However, both immunofluorescence and Western blot analyses in RBM14‐depleted U2OS cells showed that fluctuating expression at centrioles and total expression levels of STIL across the cell cycle (Tang et al, 2011; Arquint et al, 2012; Vulprecht et al, 2012) seemed comparable to the control (Fig 3D and Supplementary Fig S4A). There was also no alteration in the localization of STIL at the proximal end of procentrioles (Tang et al, 2011; Arquint et al, 2012; Vulprecht et al, 2012) (G. Shiratsuchi, D. Kitagawa, unpublished observation). However, we noted that relatively small ectopic centrin foci, but not seemingly mature centrin foci, tended to contain STIL signals during mitosis (9 ± 3% for RBM14‐depleted cells, 1 ± 1% for control cells, P < 0.05, n = 90 in triplicate; Fig 3D), implying that STIL could localize to premature centriolar protein complexes only transiently. Based on these data, we postulated that the binding of RBM14 to STIL might affect the function of STIL in centriole formation. Given that STIL[N] contained the region responsible for CPAP‐binding (231–781 aa) (Tang et al, 2011), we assumed that RBM14 may disrupt the interaction between STIL and CPAP. To address this, we conducted in vitro pull‐down assay to test whether RBM14[C] and the STIL‐binding region of CPAP, CPAP[SBD], compete with each other for binding to STIL[N]. We found this to be the case, supporting the model in which RBM14 prevents the formation of STIL/CPAP complex (Fig 3E). Furthermore, we found that addition of RBM14 FL or RBM14[C], but not RBM14[N], efficiently dampened the complex formation of STIL and GFP‐CPAP in U2OS cells (Fig 3F). These findings are in line with the fact that the C‐terminal region of RBM14 is responsible for STIL binding (Fig 3B and C, and Supplementary Fig S3). Importantly, we revealed, using siRNA‐based double knockdown experiments, that the formation of ectopic centrin foci by RBM14 depletion depends on CPAP and STIL (Fig 3G). Moreover, to further confirm the biological relevance of the complex formation of STIL and CPAP in this process, we tested whether expression of STIL mutants, STIL[N] and STIL[CBD], that contain CPAP‐binding domain (CBD), but lack the conserved STAN motif, could act in a dominant‐negative manner to inhibit the formation of the ectopic centriolar protein complexes in RBM14‐depleted cells. Accordingly, we found that this was indeed the case (Supplementary Fig S4B). Overall, these findings lead us to propose that the interaction of RBM14 with STIL suppresses the inherent ability of the STIL/CPAP complex for the ectopic formation of centriolar protein complexes.
The ectopic formation of centriolar protein complexes occurs in the cytoplasm in a HsSAS‐6‐independent manner
Next, to test the dependency of a cartwheel structure on the formation of ectopic centriolar protein complexes, we conducted RNAi‐mediated reduction of HsSAS‐6 in the RBM14‐depleted cells. Strikingly, this analysis revealed that the formation of ectopic centrin foci was mostly independent of the existence of HsSAS‐6 (Fig 4A and Supplementary Fig S5A and B). Similarly, depletion of Plk4 did not affect the amplification of centriolar protein complexes (Supplementary Fig S5C). These results suggest that the amplification of centriolar protein complexes induced by RBM14 depletion occurs independently of HsSAS‐6 and possibly of a cartwheel structure. We also observed that assembly of PCM components was not critical for this process (Supplementary Fig S5D) even though we somehow detected the interaction of RBM14[C] with γ‐tubulin (Supplementary Fig S5E).
To investigate when and how the ectopic structures are formed, we employed long‐term fluorescence time‐lapse recording with HeLa cells stably expressing GFP‐centrin (Piel et al, 2000) and depleted of RBM14. Intriguingly, we found that the ectopic GFP‐centrin foci were formed in the cytoplasm (Fig 4B and Supplementary Movie S1). Our automated particle analysis demonstrated that the number of centrin foci whose signal intensity rose above the threshold was gradually increased and reached saturation (4.8 ± 0.7 in RBM14‐depleted cells compared with 1.6 ± 0.4 in control cells, respectively, n = 5; Fig 4B) within approximately 3 h from the appearance of the foci. Importantly, further detailed observation of the behavior of individual foci revealed that some small centrin foci in the cytoplasm fused with each other and grew to a size comparable to the centrin foci of pre‐existing centrioles (Supplementary Movies S2 and S3). On the other hand, the signal intensities and pattern of centrin foci at pre‐existing centrioles were not altered, suggesting that formation of ectopic centrin foci was not because of fragmentation of pre‐existing centrioles. We also noted that such ectopic formation of centrin foci was unlikely dependent on microtubules (Supplementary Fig S5F). Moreover, such assembly of ectopic centrin foci seemed to occur approximately from late G1 to S phase (Fig 4C and Supplementary Movie S4). Indeed, cell cycle arrest experiments and PCNA staining in RBM14‐depleted U2OS cells confirmed that ectopic formation of centriolar protein complexes occurred during S phase (Fig 4D and Supplementary Fig S5G and H). Lovastatin‐induced G1 phase arrest significantly inhibited amplification of centriolar protein complexes, whereas HU‐induced S phase arrest did not suppress the phenotype.
These data prompted us to analyze the morphological features of the ectopic centriolar protein complexes in RBM14‐depleted cells at the ultrastructural level using transmission electron microscopy (TEM). Repeated TEM analysis revealed that most of the structures corresponded to amorphous electron‐dense materials (white squares in Fig 4E and Supplementary Fig S5I). The average diameter of these structures (130 ± 11 nm, n = 16) is shorter than pre‐existing centrioles (225 ± 8 nm, n = 10). Some amorphous structures contained microtubule structures, although they were not cylindrically arranged (Fig 4E). CLEM (Correlative Light Electron Micro‐scopy) analysis further confirmed that the amorphous structures contained CPAP that is a critical component of ectopic centriolar protein complexes in RBM14‐depleted cells (Fig 4F and G). In addition, although it has been shown that the centriolar satellites can be morphologically detected as small electron‐dense granules that surround centrosomes, the ectopic centriolar protein complexes in RBM14‐depleted cells did not co‐localize with foci of PCM‐1, a major centriolar satellite protein (Balczon et al, 1994) (Supplementary Fig S5J). Overall, these results show that, upon RBM14 depletion, the amorphous centriolar protein complexes formed in the cytoplasm during S phase in a HsSAS‐6‐independent manner.
Ectopic centriolar protein complexes can be the sites of microtubule nucleation
Since the ectopic centriolar protein complexes induced by RBM14 depletion could recruit PCM components (Fig 1E), we next tested whether they are potent to nucleate microtubules. Given that formation of the centriolar protein complexes frequently resulted in pseudo‐bipolar spindle formation in which an abnormal number of centriolar protein foci clustered into spindle poles, we sought to establish an experimental setup in which it is possible to decluster the centriolar protein complexes and pre‐existing centrioles and evaluate their ability to nucleate microtubules in mitosis. To this end, we carried out a microtubule regrowth assay using mitotic U2OS cells (Fig 5A). As expected, in control mitotic cells, we detected efficient microtubule regrowth and subsequent spindle formation within 30 min after cold treatment (Fig 5B). Importantly, we found that, upon RBM14 depletion, the ectopic centriolar protein complexes with PCM cloud formed extra spindle poles, which often induced formation of multipolar spindles at 30 min after cold treatment (Fig 5C–E). 46 ± 5% of the centriolar protein complexes appeared to harbor microtubule nucleation activity in mitosis (n = 150 in triplicate). However, we observed that extra small spindle poles in RBM14‐depleted cells tended to be clustered into a pseudo‐bipolar spindle assay at 60 min after cold treatment (Fig 5C). Taken together, these results indicate that the ectopic centriolar protein complexes induced by RBM14 depletion are at least in part potent to nucleate microtubules.
Ectopic centriolar protein complexes can assemble into structures more akin to procentrioles
We next sought to analyze the molecular components of the ectopic centriolar protein complexes that can organize a mitotic spindle. In line with the observation in the microtubule regrowth assays (Fig 5), we first confirmed that there was an increase in the frequency of multipolar spindle formation in RBM14‐depleted cells (~9% for RBM14‐depleted HeLa cells and ~2% for the control cells, respectively, n ≥ 44; 7 ± 2% for RBM14‐depleted U2OS cells compared with 0% for the control cells, P < 0.05, n = 90 in triplicate; Supplementary Fig S6A and B). Furthermore, we found that most of the cells depleted of RBM14 and with multipolar spindles harbored extra HsSAS‐6 foci and acetylated‐tubulin foci (Fig 6A and Supplementary Fig S6C). Consistently, we noted that only a small but statistically significant portion of the RBM14‐depleted cells exhibited excess foci of HsSAS‐6 overlapping with the ectopic centrin foci (8 ± 1% compared with 2 ± 1% in control cells, P < 0.05, n = 90 in triplicate; Fig 6B and Supplementary Fig S6D). These ectopic centriolar protein complexes containing HsSAS‐6 appeared to co‐localize with Cep135/Bld10p and Cep152/Asterless (80 ± 2% and 78 ± 1%, respectively, n = 90 in triplicate; Supplementary Fig S6E) both of which are presumably implicated in the early stage of procentriole formation (Kleylein‐Sohn et al, 2007) (Blachon et al, 2008). Moreover, we found that HsSAS‐6 depletion substantially offsets the increase in the frequency of the cells with multipolar spindle formation in RBM14‐depleted cells (Fig 6C). In contrast, most pseudo‐bipolar spindles in which an abnormal number of centrin foci clustered into two spindle poles did not contain extra HsSAS‐6 or acetylated‐tubulin foci (n = 44; Fig 6A and Supplementary Fig S6C). Together, these results suggest that the ectopic centriolar protein complexes that have incorporated HsSAS‐6 become more potent for spindle assembly.
Next, we further investigated the molecular components of HsSAS‐6‐containing centriolar protein complexes by conducting triple‐labeling immunofluorescence analysis with HeLa cells expressing GFP‐centrin. Intriguingly, this experiment revealed that HsSAS‐6‐containing centriolar protein complexes presented almost appropriate placement of HsSAS‐6, centrin and CP110 that mark the centriole proximal and distal ends (Kleylein‐Sohn et al, 2007; Strnad et al, 2007) (Fig 6D). However, we found that the distance between the HsSAS‐6 signal at the proximal end and centrin/CP110 signals at the distal end of centriolar protein complexes was a bit shorter than that of the endogenous procentrioles (n = 7, respectively; Fig 6D). Together, these findings suggest that centriolar protein complexes in RBM14‐depleted cells could become the structures more related to procentrioles by incorporating HsSAS‐6 and other centriolar proteins although they are still abnormal in length.
To further address this possibility, we ectopically expressed HsSAS‐6‐GFP in RBM14‐depleted cells to see whether there is an increase in the number of centriolar protein complexes recruiting HsSAS‐6. Notably, we found that simultaneous HsSAS‐6‐GFP overexpression and RBM14 depletion synergistically enhanced the frequency of formation of the centriolar protein complexes containing HsSAS‐6 foci compared with HsSAS‐6‐GFP overexpression or RBM14 depletion alone (Supplementary Fig S7A). To strengthen this notion, using long‐term fluorescence time‐lapse recording, we provided direct evidence that ectopic HsSAS‐6‐GFP foci assembled in the cytoplasm in ~25% of RBM14‐depleted cells, whereas that was rarely the case in control cells (Supplementary Fig S7B and C, and Supplementary Movie S5). Furthermore, we also set up the experimental system in which we could observe the dynamics of both ectopic DsRed‐centrin‐2 and HsSAS‐6‐GFP foci in RBM14‐depleted cells, and established that incorporation of HsSAS‐6‐GFP into ectopic DsRed‐centrin‐2 foci indeed occurred (~16% of ectopic DsRed‐centrin‐2 foci in RBM14‐depleted cells compared with ~3% of those in control cells; Fig 6E and F, and Supplementary Movie S6). Overall, these results suggest that the centriolar protein complexes could serve as a platform to incorporate HsSAS‐6 and other centriolar proteins required for the initial stage of procentriole formation. In fact, when the formation of ectopic centriolar protein complexes upon RBM14 depletion was inhibited by concurrent CPAP depletion, the increase of HsSAS‐6 foci was significantly suppressed, suggesting that formation of ectopic HsSAS‐6 foci is dependent on the existence of ectopic centriolar protein complexes in RBM14‐depleted cells (Fig 6G and Supplementary Fig S7D).
Since an extra spindle pole in RBM14‐depleted cells likely incorporated ectopic HsSAS‐6 and other centriolar proteins aligned in a procentriole‐like manner (Fig 6A–G), we hypothesized that the ectopic centriolar protein complexes forming an extra spindle could be structures more akin to procentrioles. To address this possibility, we conducted Live CLEM analysis using HeLa cells expressing GFP‐centrin and depleted of RBM14 (Fig 6H, Supplementary Fig S8A–C, and Supplementary Movie S7). In the cells depleted of RBM14, we sometimes observed the pseudo‐bipolar spindles in which pre‐existing centriole pairs failed to separate each other and organize a single spindle pole while only ectopic GFP‐centrin foci assembled the other spindle pole (Fig 6H and Supplementary Figs S6A and S8A–C). Although the reason why pre‐existing centriole pairs did not separate in mitosis in such cases remains elusive, we speculate that RBM14 depletion might somewhat disturb the separation of centrosomes. We fixed such cells in which ectopic GFP‐centrin foci gathered to assemble a spindle pole (red arrowheads in Fig 6H, red circle in Supplementary Fig S8C). We then analyzed the structures in those spindle poles by using serial section EM analysis. Although we could not find a cartwheel structure with ninefold symmetrically arranged microtubules, at the spindle pole composed of only ectopic GFP‐centrin foci, probably due to technical reasons, we found a single centriole‐like structure in a cylindrical arrangement (arrowhead 2 in Fig 6H) as well as other electron‐dense structures (arrowheads 1 and 3 in Fig 6H). We also found such centriole‐like structures in the daughter cell containing only ectopic GFP‐centrin foci (Supplementary Fig S8C). Together, these results support the notion that RBM14 depletion induces ectopic formation of centriolar protein complexes, a part of which could become a procentriole‐like structure even in the presence of pre‐existing centrioles.
In conclusion, we have, for the first time, identified RBM14 as a suppressor of ectopic assembly of centriolar protein complexes that can function as microtubule organizing centers (MTOCs) (Fig 7). Our work indicates that when RBM14 is depleted, the STIL–CPAP complex can form aberrant centriolar protein complexes in the cytoplasm presumably without assembling a cartwheel structure. Given the heterogeneity of the ectopic structures, there might not be a well‐defined order by which centriolar components are incorporated into the structures. As the cell cycle progresses, a part of such centriolar protein complexes seem to assemble into procentriole‐like structures that incorporate a set of centriolar proteins required for the initial stage of procentriole formation. We also performed the functional analysis of RBM14 in vivo using early mouse embryos and found that RBM14 negatively regulates the assembly of centriolar protein complexes under physiological conditions (Supplementary Fig S9).
How can RBM14 spatiotemporally limit the function of the STIL–CPAP complex? It is possible that RBM14 proteins in the cytoplasm normally suppress STIL function by direct association during interphase. However, in the absence of RBM14, STIL can overly form a complex with CPAP in the cytoplasm, and this complex induces the ectopic formation of centriolar protein complexes during interphase. Considering the observation that overexpression of RBM14‐NES which forcibly localized to the cytoplasm, but seemingly not to centrioles, suppressed HU‐induced centriole amplification, it is likely that RBM14 acts on cytoplasmic STIL rather than on the STIL implicated in procentriole formation near pre‐existing centrioles. Furthermore, given the observation that the STIL/CPAP complex formation was enhanced in the S phase‐arrested cells upon treatment with HU (Supplementary Fig S4C), we speculate that RBM14 could inhibit centriole amplification in the S phase‐arrested cells by suppressing excess formation of the STIL/CPAP complex in the cytoplasm. By contrast, in the process of canonical centriole formation, formation of the STIL/CPAP complex could be strictly limited in the cytoplasm so as to happen only at the assembly site of procentrioles since RBM14 overexpression had only minor effect on this process in cycling cells (Supplementary Fig S2B). It has been shown that SAS‐4/CPAP provides a scaffold for cytoplasmic assembly of PCM components, which implies that PCM assembly could begin in the cytosol before recruitment of this complex to the centrosome (Gopalakrishnan et al, 2011). In a similar way, we propose that when RBM14 is depleted, the ectopic STIL/CPAP complex gives a scaffold for assembly of centriolar and PCM components in the cytoplasm and that such complexes including γ‐tubulin can be the site to nucleate microtubules during mitosis in RBM14‐depleted cells.
Recent studies have shown that depletion of CP110‐binding proteins such as Cep76 and Neurl4 induced accumulation of centriolar proteins near pre‐existing centrioles and resulted in the formation of the ectopic structures similar to the centriolar protein complexes that we demonstrated in this study (Tsang et al, 2009; Li et al, 2012). However, the ectopic structures in Cep76‐depleted cells contained HsSAS‐6 signals, and their formation was HsSAS‐6 dependent (Tsang et al, 2009). In addition, the ectopic structures disassembled and failed to form a functional microtubule array during mitosis. On the other hand, depletion of Neurl4 led to the formation of centrin‐containing ectopic structures without incorporating HsSAS‐6 (Li et al, 2012), implying that they might be more related to the centriolar protein complexes shown in this study. However, where this assembly takes place in the cells and whether the ectopic structures can finally assemble into the more completed centrioles remains unclear. More recently, it has been reported that depletion of LGALS3BP caused formation of ectopic structures composed of centriolar proteins such as CPAP, acetylated tubulin and centrin (Fogeron et al, 2013). However, the formation of those structures was dependent on Plk4, suggesting that the assembly mechanism is different from what is shown in this study.
Further study will be needed to describe the dynamics of the aberrant assembly of centriolar components in more detail and uncover the mechanisms by which they can assemble into substantial and functional structures to efficiently serve as MTOCs. Considering that the ectopic formation of centriolar protein complexes takes place during S phase in RBM14‐depleted cells, and also that de novo formation of premature centrioles caused by loss of pre‐existing centrioles is known to occur during S phase (Khodjakov et al, 2002; La Terra et al, 2005), there could be a close linkage between the two assembly pathways for centriolar proteins.
Given that RBM14 has been reported as a potential tumor suppressor in human cancer cells (Kang et al, 2008), the formation of these structures could contribute to genome instability by inducing chromosome mis‐segregation. We speculate that the accumulation of that kind of minor errors could be influential for tumor formation in the long term. Therefore, it will be important to investigate whether such aberrant assembly of centriolar proteins can be a cause of some types of cancer.
Materials and Methods
Cell culture and cell lines
Human U2OS, HEK293T, RPE1, HeLa and HeLa cells stably expressing GFP‐centrin 1 (Piel et al, 2000), and mouse NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) at 37°C in 5% CO2 incubator. To generate inducible cell lines, cells were transiently transfected with pEBTet‐EGFP‐CPAP (Kitagawa et al, 2011a) and pEBTet‐HsSAS‐6‐EGFP at 80–90% confluency. 24 h later, cells were exposed to selective medium containing 1 μg/ml puromycin, which led to substantial death of non‐transfected cells over 4–5 days. After amplification under selective conditions for 1–2 weeks, cells were frozen in 10% DMSO and stored at −80°C. Expression was induced using 1 μg/ml doxycycline for 24–48 h.
Cell cycle synchronization, cell cycle arrest and FACS analysis
For cell synchronization at G1/S transition, cells were treated with 2 mM thymidine for 16 h, washed twice with phosphate‐buffered saline (PBS), released for 8 h in fresh medium without thymidine and treated again with 2 mM thymidine for 16 h. For cell synchronization at G2/M transition, cells were treated with 200 ng/ml nocodazole for 16 h. For cell cycle arrest, cells were treated with 2 mM hydroxyurea (HU) for 16 h for S phase arrest or 25 μM lovastatin for 16 h for G1 phase arrest.
For FACS analysis, cells cultured on dishes were trypsinized, washed twice with PBS and fixed in 70% cold ethanol at −20°C at each time point. The fixed cells were incubated with 0.4 μg/ml RNase for 10 min at 37°C and stained with 100 μg/ml propidium iodide (PI). The DNA contents of the cells were then measured using JSAN desktop cell sorter (Bay Bioscience).
Molecular biology, RNA interference and transfection
The following siRNAs were used: Silencer Select siRNA (Life Technologies) against RBM14 (s20406), mRBM14 (s80246), STIL (s12863), SAS‐6 (s46487), CPAP (s31623), NONO (s9614), Cep192 (s30227) and Negative control #1 (4390843); Stealth siRNA (Life Technologies) against 3′ UTR of RBM14 (5′‐GCATTCTGGCCATAGAGCTCGTATT‐3′) used in Fig 2A and Negative control Low GC Duplex #2 (12935110); custom siRNA (Sigma Genosys) against 3′ UTR of Plk4 (5′‐CTCCTTTCAGACATATAAG‐3′).
pcDNA3‐FLAG‐RBM14 FL and pcDNA3‐FLAG‐RBM14[N] expression constructs were gifts from Dr. Toshiharu Iwasaki. The expression construct for FLAG‐RBM14[C] was created by insertion of the subcloned fragment into HindIII‐XhoI site‐digested pcDNA3 vector (Invitrogen). The pcDNA3‐FLAG‐RBM14‐NES construct was created by insertion of HIV‐Rev NES sequence (LQLPPLERLTLD) onto C‐terminus of RBM14. pEGFP‐RBM14‐PACT was created by insertion of the full‐length sequence of RBM14 into the pEGFP‐PACT expression vector (gift from Dr. Mikiko Takahashi). pcDNA3‐FLAG‐RBM14 T1, T2, T3, T4 and ΔTRBP‐ID were made from pcDNA3‐FLAG‐RBM14[C] using Prime STAR MAX mutagenesis Basal kit (TaKaRa). pHA‐STIL FL/[N]/[CBD]/[C] expression constructs were created by insertion of subcloned fragments into SpeI‐digested pHA vector. The expression construct for RFP‐histone H2B was a gift from Dr. Tetsuya Hori. pEBTet‐EGFP‐CPAP and pEBTet‐HsSAS‐6‐EGFP were created using Gateway System (Invitrogen). pcDNA3‐FLAG‐Plk4‐ΔPEST (Yamashita et al, 2001) was a gift from Dr. Hiroyuki Mano. pCMV5‐Myc‐HsSAS‐6 was created by insertion of the subcloned fragment into SalI site‐digested pCMV5 vector. The expression construct for DsRed‐centrin2 was purchased from Addgene (Plasmid # 29523).
Transfection of siRNA or DNA constructs into U2OS, HeLa, RPE1, HEK293T, and NIH3T3 cells was performed using Lipofectamine RNAiMAX (Life Technologies) or Lipofectamine 2000 (Life Technologies) according to the manufacturer's instruction. Unless otherwise noted, transfected cells were analyzed 72 h after transfection.
The following primary antibodies were used in this study: rabbit polyclonal antibodies against STIL (Abcam, ab89314, IF 1:1,000, WB 1:1,000), RBM14 (Abcam, ab12325, IF 1:300, WB 1:1,000), SAS‐6 (Santa Cruz Biotechnology, Inc., sc‐98506, IF 1:300, WB 1:1,000), Cep135 (Abcam, ab75005, IF 1:300), Cep192 (a gift from Laurence Pelletier, IF 1:1,000), Cep152 (Bethyl laboratories, A302‐480A, IF 1:1,000), CP110 (a gift from Brian David Dynlacht, IF 1:500), CPAP (a gift from Pierre Gönczy, IF 1:500, WB 1:1,000), CPAP/CENP‐J (Proteintech, 11517‐1‐AP, IF 1:500, WB 1:1,000), C‐Nap1 (a gift from Erich Nigg, IF 1:1,000), Cep164 (Novus, 45330002, IF 1:500), γ‐tubulin (Sigma‐Aldrich, T5192, IF 1:1,000), α‐tubulin (Abcam, ab18251, IF 1:1,000), centrin‐1 (Abcam, ab11257, IF 1:1,000), PCM‐1 (Sigma‐Aldrich, HPA023370, IF 1:300), GFP (MBL, IF 1:300), FLAG‐tag (Sigma‐Aldrich, F7425, IF 1:200) and HA‐tag (Abcam, ab9110, WB 1:1,000); mouse monoclonal antibodies against centrin‐2 (Millipore, 20H5, IF 1:1,000, WB 1:1,000), Centrobin/NIP2 (Abcam, ab70448, IF 1:500), α‐tubulin (Sigma‐Aldrich, T5168, IF 1:1,000), HsSAS‐6 (sc‐81431, Santa Cruz Biotechnology, Inc., WB 1:1,000), acetylated tubulin (Sigma‐Aldrich, T6793, 6‐11B‐1, IF 1:1,000), histone H4 (MBL, D214‐3, WB 1:1,000) and PCNA (Santa Cruz, sc‐56, IF 1:1,000); goat polyclonal antibodies against actin (Santa Cruz Biotechnology, Inc., ac‐1616, WB 1:1,000); and goat polyclonal antibody against GFP, FITC‐conjugated (Abcam, ab6662, IF 1:300).
The following secondary antibodies were used: Alexa Fluor 488 goat anti‐mouse IgG (H+L) (Molecular Probes, 1:500), Alexa Fluor 568 goat anti‐rabbit IgG (H+L) (Molecular Probes, 1:500) and goat polyclonal antibody against mouse IgG (H+L)‐CY5 (Abcam, ab6563, 1:500) for IF and goat polyclonal antibodies HRP against mouse IgG (Promega, W402B, 1:5,000), rabbit IgG (Promega, W401B, 1:5,000) and HRP‐labeled protein A (KPL, 14‐50‐00, WB: 1:2,500) for WB.
Indirect immunofluorescence and immunoblotting
For indirect immunofluorescence microscopy, the cells cultured on coverslips were fixed using −20°C methanol for 10 min. The cells were then permeabilized with PBS/0.05% Triton X‐100 (PBSX) for 5 min, washed with PBS three times and incubated for blocking in 1% BSA in PBSX for 30 min at 37°C. The cells were then incubated with primary antibodies for 1 h at 37°C, washed with PBSX three times and incubated with secondary antibodies for 1 h at 37°C. The cells were thereafter washed with PBSX twice, stained with 1 μg/ml Hoechst 33258 in PBS for 5 min at RT, washed again with PBSX and mounted onto glass slide. Counting the number of immunofluorescence signals was performed using an Axioplan2 fluorescence microscope (Carl Zeiss) with a 100×/1.4 NA plan‐APOCHROMAT objective. Confocal images were recorded using a Fluoview FV‐1000D confocal laser scanning microscope system (Olympus) with a 100×/1.4 NA UPlanSApo objective and analyzed using FV10‐ASM software (Olympus) and Adobe Photoshop.
For the preparation of human soluble cytosolic fractions for immunoblotting, cells were collected, washed in PBS and lysed on ice for 30 min in lysis buffer [20 mM Tris–HCl pH 7.5, 100 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.5% NP‐40 and 1/1,000 protease inhibitor cocktail (nacalai tesque)]. Lysates were cleared by centrifugation for 15 min at 16,000 g at 4°C and the supernatant collected. SDS–PAGE was performed using 8–12% polyacrylamide gels, followed by transfer on Immobilon‐P membrane (Millipore corporation). The membrane was probed with the primary antibodies, followed by incubation with their respective HRP‐conjugated secondary antibodies (Promega). Washes were performed in PBS containing 0.02% Tween (PBST). The signal was detected as chemiluminescence (GE healthcare, ECL Prime). Signal intensity of immuno‐reactive bands was measured using ImageJ software (http://rsb.info.nih.gov/ij/).
Electron microscopy/Correlative light and electron microscopy (CLEM)
For electron microscopy, the cells cultured on 6‐well dishes were washed with PBS and fixed with 2% glutaraldehyde and 1.5% Tannic Acid in PBS for 1 h. The cells were postfixed with 1% OsO4 for 1 h on ice and stained with 1% uranyl acetate for 30 min on ice. The samples were then dehydrated and embedded in Quetol 812 (Nisshin EM). Ultrathin sections were post‐stained with 7% aqueous uranyl acetate for 20 min and 0.8% lead citrate for 2 min and then observed by a transmission electron microscope JEM‐1010 (JEOL). Images were acquired with a VELETA CCD camera (SEIKA Corporation) and analyzed with iTEM software (SEIKA Corporation) and Adobe Photoshop.
For CLEM, U2OS cells expressing GFP‐CPAP were cultured on 35‐mm glass‐bottom dishes with grid (MatTek Co., P35G‐2‐14‐C‐GRID). The cells were fixed with 4% paraformaldehyde 24 h after transfection with RBM14 siRNA and induction of GFP‐CPAP. Fluorescence microscopy was performed on a Cell Voyager CV1000 Confocal Scanner Box (Yokogawa Electric Co.) with a 63× oil immersion objective. After fixation and dehydration as described above, the cells were embedded in Quetol 812 by inverting the Quetol 812‐filled beam capsules onto the glass‐bottom dishes. The regions including the cells of interest were found by a grid reference using stereoscopic microscope. Ultra‐thin sectioning and following staining were performed as described above.
Immunoprecipitation and mass spectrometry
For preparing soluble cytosolic fractions of HeLa, U2OS or HEK293T cells, cells were washed by PBS and lysed in ice‐cold lysis buffer. The lysates were vortexed for 15 min at 4°C, and insoluble material was removed after centrifugation for 15 min. For STIL immunoprecipitation, soluble cytosolic fractions of HeLa, HEK293T or U2OS cells were incubated with protein A agarose for 1 h at 4°C for preclear and then incubated for 2 h at 4°C with protein A agarose that had been incubated with anti‐STIL antibodies or normal rabbit IgG for mock immunoprecipitation. The beads were washed at least three times with lysis buffer and resuspended in SDS‐sample buffer before loading onto a SDS–PAGE gel. To identify STIL‐interacting proteins, the gel stained with SimplyBlue™ Safe (Invitrogen) was cut into five pieces. The peptides were digested in gel by trypsin and extracted from the gel. The peptides were subsequently desalted and analyzed by a HCT‐Ion trap mass spectrometer (Bruker Daltonik GmbH).
For co‐immunoprecipitation assays, the lysates of HEK293T cells transfected with the indicated plasmids (pcDNA3‐FLAG‐RBM14 FL/[N]/[C], pHA‐STIL FL/[N]/[C]) for 24 h were lysed in ice‐cold lysis buffer and incubated with anti‐FLAG M2 agarose beads (Sigma‐Aldrich, A2220) for 2 h at 4°C. The beads were then washed three times with ice‐cold lysis buffer and resuspended in SDS‐sample buffer.
GST pull‐down assay and in vitro competitive binding assay
DNA encoding fragments of human STIL or RBM14 were cloned in pGEX system vectors (GE healthcare) encoding for GST‐tags. DNA encoding fragments of CPAP[SBD] were cloned into pET30a vector encoding for His‐tags. The recombinant protein expression of the fragments was performed in E. coli strain BL21‐Gold (DE3) in LB medium. Protein expression was induced at 22°C by addition of 0.3 mM IPTG and allowed to proceed for 18 h. Cell pellets expressing GST‐STIL and RBM14 were lysed by lysozyme treatment and sonication and resuspended in lysis buffer containing 50 mM Tris–HCl (pH 7.5), 500 mM NaCl, 5 mM EDTA, 1 mM DTT, 1:1,000 protease inhibitor cocktail (Sigma) and 0.5% Triton X‐100. The lysates were incubated with Glutathione sepharose beads (GE healthcare). The beads were then washed ten times with lysis buffer supplemented with additional 500 mM NaCl. For preparing N‐terminal fragment of STIL, proteins were eluted from the beads by removal of the GST‐tags by prescission protease (GE healthcare) in a buffer containing 20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 0.5 mM EDTA and 1 mM DTT. Cell pellets expressing His‐CPAP[SBD] were lysed by lysozyme treatment and sonication and resuspended in lysis buffer containing 50 mM Tris–HCl (pH 7.5), 500 mM NaCl, 20 mM imidazole and 0.5% Triton X‐100. The lysates were incubated with Ni‐NTA agarose beads (QIAGEN). The beads were then washed ten times with lysis buffer. Proteins were eluted from the beads with a buffer containing 80 mM PIPES‐KOH (pH 6.8), 80 mM KCl, 2 mM MgCl2 and 400 mM imidazole.
For GST pull‐down assays, the Glutathione sepharose beads which retain bacterially purified GST‐RBM14 fragments were resuspended in ice‐cold lysis buffer and incubated with ~5 μg bacterially purified N‐terminal fragment of STIL for 2 h at 4°C. The resin was washed five times with ice‐cold lysis buffer and resuspended in SDS‐sample buffer. Proteins were separated by SDS–PAGE and stained with SimplyBlue™ Safe (Invitrogen). For in vitro competitive binding assays, the GST pull‐down experiment was performed in the same way with purified STIL[N] and GST‐RBM14[C] in the presence of the indicated amount of purified His‐CPAP[SBD] in Fig 3E.
Yeast two‐hybrid analysis
Yeast strain MAV103 was grown in YPD medium and transformed with a modified version of the vectors pPC86 and pPC97 (Clontech) that contained full‐length or fragments of RBM14 or STIL. Positive colonies were cultured in yeast dropout medium without leucine (L) and tryptophan (W) overnight. On the next day, cells were diluted to an OD600 = 1. 10 μl of this solution was spotted on yeast (−L/−W) plates supplemented with 50 mM 3‐amino‐triazol. Two serial dilutions to an OD of 0.1 and 0.01 were prepared, and 10 μl of these dilutions was spotted as well. Plates were placed at 30°C until clear colonies had formed on a positive control, which was SAS‐6 coiled‐coil self‐interaction in the initial experiments.
Time‐lapse fluorescence microscopy and image analysis
Time‐lapse fluorescence microscopy was performed on a Cell Voyager CV1000 Confocal Scanner Box with a 63× oil immersion objective in an equilibrated chamber in 5% CO2 at 37°C. HeLa cells stably expressing GFP‐centrin1 were cultured on 35‐mm glass‐bottom dishes (MatTek Co.), and they were transfected with RBM14 siRNA or control siRNA together with a plasmid DNA coding RFP‐histone H2B. U2OS cells expressing HsSAS‐6‐GFP were cultured in the same way and transfected with RBM14 siRNA or control siRNA together with or without a pDsRed‐N1‐Centrin2 vector. After 1 day, the U2OS cells were then treated with 1 μg/ml doxycycline and/or 6 μM aphidicolin. After 2 days from transfection, the cells were analyzed by fluorescence and bright field microscopy every 5 or 10 min over 24–48 h. After time‐lapse recording, we generated the movies and figure panels by using Z‐stack images from 25 to 30 Z‐planes covering approximately the whole height of the cells along the z‐axis to capture fluorescent foci that move extensively in all three dimensions. For quantification of the number of fluorescent foci and their intensity, the acquired sequential TIFF files were analyzed using ImageJ software (http://rsb.info.nih.gov/ij/).
Live correlative light and electron microscopy (Live CLEM)
For Live CLEM, the cells were cultured on 35‐mm glass‐bottom dishes with grid. We treated RBM14‐depleted cells with a moderate concentration of monastrol (50 μM) for 12 h to arrest the cells in mitosis and select the cells which harbor extra mitotic spindle poles containing only ectopic centrin foci although not all cells responded to that concentration of the drug in our experimental condition (Fig 6H, Supplementary Fig S8A and B, and Supplementary Movie S7, but no monastrol treatment for the experiment in Supplementary Fig S8C). The cells were thereafter observed by time‐lapse fluorescence microscopy as described above. We also acquired immunofluorescence images of paraformaldehyde‐fixed cells after the time‐lapse recording as individual GFP‐centrin foci correspond to the structures on EM images precisely in the same location. Then, we performed CLEM analysis as described in the method of CLEM.
RNAi experiments in mouse early embryos
Capped synthetic mRNAs were transcribed with the use of a mMessage mMachine kit (Ambion) using pCS2 tdTomato and pCS2 memTomato which consisted of tdTomato and the NH2‐terminal sequence of neuromodulin (nucleotides 679–738) from pECFP‐mem (BD Biosciences). For single‐cell injection, we used a Leica DMIRB microscope (Leica Microsystems), manipulators with Femto jet (Eppendorf) and piezo‐xpert (Eppendorf). The injection was performed in modified Whitten's medium (mWM). Fertilized eggs were obtained by mating CMV‐EGFP::centrin2 transgenic male (Higginbotham et al, 2004) with superovulated female mice. Single blastomeres of 8–16 cell embryos were injected with 16 μM Silencer Select siRNA (Life Technologies) against mRBM14 (s80246) or negative control #1 (4390843), and fluorescence mRNA. Injected embryos were cultured for 40 h in mWM solution in a 5% CO2 incubator and checked under M205FA fluorescence stereo microscope (Leica Microsystems). Only embryos that had Tomato and GFP expression were observed with FV1000 confocal microscope (Olympus).
For immunofluorescence analysis, embryos were recovered in PBS. They were carefully staged on the basis of their morphology. They were dehydrated with methanol series and were fixed overnight at 4°C in 100% methanol, washed twice with PBS and incubated for 1 h at 4°C with TSA blocking reagent (Perkin‐Elmer). Immunostaining was performed with rabbit antibodies against RBM14 (Abcam, ab12325, 1:50) and Alexa Fluor 647‐conjugated chick antibodies to rabbit immunoglobulin G (Invitrogen). Samples were washed with PBS between staining steps. Nuclei were stained by incubation for 30 min with 4′,6‐diamidino‐2‐phenylindole dihydrochloride (1/2,000 dilution in PBS, Wako). Images were obtained using FV1000 confocal microscope (Olympus).
GS and DK designed the study; GS, TA and DK performed experiments; KT and HH performed in vivo mouse experiments; GS and DK designed experiments and analyzed data; GS, KT and DK wrote the manuscript, which was commented on by all authors.
Conflict of interest
The authors declare that they have no conflict of interest.
Supplementary Figures and Table S1
Supplementary Movie S1
Supplementary Movie S2
Supplementary Movie S3
Supplementary Movie S4
Supplementary Movie S5
Supplementary Movie S6
Supplementary Movie S7
Legends for Supplementary Table, Figures and Movies
We are grateful to Diego Chiappe and Marc Moniatte (Proteomic core facility, School of Life Sciences, EPFL) for mass spectrometry, to Michel Bornens for HeLa cells stably expressing GFP‐centrin1, to Erich Nigg for C‐Nap1 antibodies, to Laurence Pelletier for Cep192 antibodies, to Brian David Dynlacht for CP110 antibodies, to Emiko Suzuki for supporting EM analysis, to Tetsuya Hori and Tatsuo Fukagawa for supporting live cell imaging and a construct coding RFP‐H2B, to Toshiharu Iwasaki and Eli Lilly and company for a construct coding RBM14, to Hiroyuki Mano for a construct coding Plk4, to Mikiko Takahashi for a construct coding PACT as well as to Midori Ota and Rieko Matsuura for critical reading of the manuscript. We thank Pierre Gönczy for CPAP antibodies, constructs and fruitful discussions. This work was supported by Grant‐in‐Aid for JSPS Fellows, Grant‐in‐Aid for Young Scientists (A) and for Scientific Research on Innovative Areas from the Ministry of Education, Science, Sports and Culture of Japan, by Improvement of Research environment for young researchers Grant from JST, by the Uehara Memorial Foundation, by Takeda Science Foundation, by the Mitsubishi foundation, by Mochida Memorial Foundation for Medical and Pharmaceutical Research, by Astellas Foundation for Research on Metabolic Disorders, by The Sagawa Foundation for Promoting Cancer Research, by Novartis Foundation for the Promotion of Science, by the Center for the Promotion of Integrated Sciences of Sokendai and by the Transdisciplinary Research Integration center.
This is an open access article under the terms of the Creative Commons Attribution 4.0 License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
- © 2014 The Authors. Published under the terms of the CC BY 4.0 license