Distal appendages (DAs) of the mother centriole are essential for the initial steps of ciliogenesis in G1/G0 phase of the cell cycle. DAs are released from centrosomes in mitosis by an undefined mechanism. Here, we show that specific DAs lose their centrosomal localization at the G2/M transition in a manner that relies upon Nek2 kinase activity to ensure low DA levels at mitotic centrosomes. Overexpression of active Nek2A, but not kinase-dead Nek2A, prematurely displaced DAs from the interphase centrosomes of immortalized retina pigment epithelial (RPE1) cells. This dramatic impact was also observed in mammary epithelial cells with constitutively high levels of Nek2. Conversely, Nek2 knockout led to incomplete dissociation of DAs and cilia in mitosis. As a consequence, we observed the presence of a cilia remnant that promoted the asymmetric inheritance of ciliary signaling components and supported cilium reassembly after cell division. Together, our data establish Nek2 as an important kinase that regulates DAs before mitosis.
The centrosome is the main microtubule-organizing center of most animal cells. Each centrosome comprises a pair of orthogonally arranged centrioles surrounded by a well-ordered matrix of pericentriolar material. Several proteins including microtubule nucleators, cell cycle regulators, and signaling molecules represent essential elements of pericentriolar material. Centrosomes perform important functions related to the control of chromosome segregation, cell cycle progression, and cell motility and polarization. In addition, centrioles are the seeding points for the formation of cilia. The primary, nonmotile cilium is a microtubule-based projection that serves as an “antenna” for detecting and responding to external signals (Nigg and Raff, 2009). Cilia can be assembled on almost all cell types in the human body (Olsen, 2005). Dysfunctions in cilia formation lead to impaired cell signaling and/or embryonic development and a wide range of diseases known as ciliopathies (Reiter and Leroux, 2017), whereas loss of cilia, or centrosomes, is associated with multiple types of cancer (Hassounah et al., 2013; Emoto et al., 2014; Nobutani et al., 2014).
Primary cilia formation occurs when cells enter the G0/G1 phase of the cell cycle (Seeley and Nachury, 2010). As the primary cilium is an important signaling hub, the control of cilia assembly and disassembly is critical. In most differentiated somatic cells, the primary cilium is fully disassembled before mitosis before being reassembled following division in G1 of the next cell cycle (Pugacheva et al., 2007; Ford et al., 2018). By contrast, certain neuronal stem cells only partially disassemble the cilium in mitosis. In those cells, a ciliary remnant is retained at one of the two centrosomes during mitosis, to be inherited by one of the two daughter cells after cell division (Paridaen et al., 2013). The molecular mechanisms accounting for synchronizing cilium assembly/disassembly with the cell cycle are only partially understood (Wang and Dynlacht, 2018).
Cilia formation requires the older mother centriole (M-centriole) that is generated one cell cycle earlier than the younger daughter centriole (D-centriole). Only the M-centriole associates with a subset of proteins that form macromolecular structures at the subdistal and distal ends of this centriole. These components appear in electron micrographs as spikelike protrusions that originate from the centriole wall that are referred to as subdistal appendages (SDAs) and distal appendages (DAs). Several components of SDAs and DAs have been identified. SDA formation requires the protein ODF2 that recruits additional SDA components, including Cep170, Cep128, CCDC120, CCDC68, centriolin, and ninein (Mazo et al., 2016). SDAs are involved in microtubule anchoring and centriole/cilia positioning (Vorobjev and Nadezhdina, 1987; Hung et al., 2016; Mazo et al., 2016). DAs comprise the proteins C2CD3 and Cep83 that are essential for the recruitment of additional DAs, such as Cep123/Cep89, SCLT1, LRRC45, Cep164, and FBF1 (Fig. 1 A; Tanos et al., 2013; Ye et al., 2014; Kurtulmus et al., 2018; Wang et al., 2018). DAs are essential at initial stages of cilia formation in order to establish the interaction between the basal body (a modified M-centriole) and the ciliary membrane compartment (Schmidt et al., 2012; Tanos et al., 2013).
The establishment of appendage proteins requires 1.5 generations after formation of the new centriole as the D-centriole is converted into an M-centriole that can function as a basal body during ciliation (Nigg and Stearns, 2011). Once assembled, at least some appendage proteins, such as ODF2, remain at the M-centriole throughout the cell cycle (Lange and Gull, 1995), which suggests that cilia disassembly or lack of cilia in mitosis is not regulated at the level of the presence or absence of centriolar appendage proteins. The general validity of this concept was, however, challenged by a number of reports showing that the level of some appendage components drastically decreases during mitosis, in a cell type–dependent manner (Schmidt et al., 2012; Sillibourne et al., 2013; Kong et al., 2014). The underlying molecular mechanism and functional significance of this mitotic regulation remain to be established.
In this study, we investigate the role of the mitotic kinase Nek2 in DA regulation. The first established function of Nek2 was the promotion of centrosome separation at the G2/M transition, when the levels and activity of this kinase peak (Fry et al., 1998b, 2012; Hames et al., 2001; Hames and Fry, 2002; Whitfield et al., 2002; Faragher and Fry, 2003). Nek2 was also shown to regulate cilia disassembly via phosphorylation of the microtubule-depolymerizing kinesin Kif24, to prevent the outgrowth of cilia in proliferating cells (Kim et al., 2015). Nek2 localizes to the proximal ends of both M- and D-centrioles (Fry et al., 1998a). In addition, Nek2 associates with the distal end of the M-centriole and the basal body. Its accumulation at the distal part of the M-centriole is cell cycle dependent and appears to peak in late G2 (Spalluto et al., 2012; Kim et al., 2015). Furthermore, Nek2 phosphorylates the M-centriole–specific component ninein-like protein (Nlp), a protein implicated in microtubule anchoring that is released from the centrosome upon mitotic entry (Casenghi et al., 2003; Rapley et al., 2005). These observations suggest that Nek2 could regulate additional appendage components at the M-centriole at the G2/M transition.
Here, we show that Nek2 is required for the release of a subset of DAs including Cep123, Cep164, and LRRC45 from the M-centriole at the onset of mitosis. In the absence of Nek2, ciliary disassembly is incomplete, leading to the persistence of a ciliary remnant at the older centrosome of mitotic cells. This facilitates cilia elongation shortly after division. Our data thus define Nek2 as an important kinase that controls DAs and timing of cilia formation.
A subset of DAs is displaced from centrosomes in mitosis
To investigate the behavior of DAs on the two centrosomes during mitosis, we quantified the levels of DAs at various cell cycle phases (Fig. 1 and Fig. S1, A and B). Because DAs were previously shown to behave differently depending on the cell type (Schmidt et al., 2012; Sillibourne et al., 2013; Kong et al., 2014), we performed our analysis using human immortalized retina pigment epithelial cells (RPE1) and primary human CD34+ hematopoietic stem and progenitor cells (HSPCs) isolated from healthy donors (Figs. 1 and S1 C). This allowed us to analyze the behavior of DAs throughout the cell cycle in differentiated versus immature/stem cells. We used HSPCs because they are one of the few primary multipotent cells that can readily be identified and isolated from the bone marrow of healthy human subjects.
In agreement with previous work, the levels of Cep164 and Cep123 at the M-centriole were drastically reduced before mitosis in RPE1 cells (Fig. S1 B; Schmidt et al., 2012; Sillibourne et al., 2013; Kong et al., 2014). Specifically, removal of Cep164 and Cep123 from the M-centriole occurred at the G2 phase, marked by centrosome separation and nuclear accumulation of the centromere protein F (CENPF; Fig. 1, B and C; Liao et al., 1995). Reduced levels were maintained from prophase to telophase (Fig. 1 D). Similar results were obtained for HSPCs (Fig. S1 C). LRRC45 signals emulated those of Cep164 in declining in intensity at one centrosome before mitosis in RPE1 cells and HSPCs (Fig. 1, C and D; and Fig. S1 C). Notably, FBF1 showed a different behavior in RPE1 cells compared with HSPCs. During mitosis, FBF1 levels decreased in HSPCs but maintained interphase levels at one centrosome in RPE1 cells (Figs. 1 D and S1 C) to define a clear cell type–specific regulation for FBF1 recruitment at DAs. Finally, the levels of Cep83 and SCLT1 did not decrease at the M-centriole during mitosis in RPE1 cells (Fig. 1, C and D) or HSPCs (Fig. S1 C) and resembled the behavior of the SDA ODF2 (Figs. 1 D and S1 C).
In conclusion, our data show that the mitotic behavior of appendage proteins reflects the hierarchical network of their assembly (Fig. 1 A). The assembly of appendage proteins, which were released from the centrosome in mitosis (Cep164, FBF1, Cep123, and LRRC45), has been shown to depend on the presence of components that stay on the M-centrosome in mitosis (Cep83 and SCLT1; Tanos et al., 2013; Kurtulmus et al., 2018). Together, our data illustrate the differential centrosome binding of some cohorts of DAs during the cell cycle and suggest a common mode of regulation of appendage dynamics with some cell type–specific components, such as FBF1.
The kinase Nek2 colocalizes with appendages
To understand the molecular mechanisms involved in the regulation of DAs during mitosis, we next asked whether the G2/M Nek2 kinase is involved. Concomitant with appendage release, Nek2 accumulates in cells and at centrosomes before mitosis (Fig. 2 A; Fry et al., 2012; Spalluto et al., 2012; Kim et al., 2015).
To define the localization overlap of Nek2 with DAs, we performed 3D structural illumination microscopy (3D-SIM). We found that in addition to its proximal pool, Nek2 fully colocalized with Cep164 in 49% of RPE1 cells (Fig. 2 B). A partial colocalization with the SDA ODF2 was observed in 60% of the cells (Fig. 2 B). The Nek2 signal was specific as it was absent in Nek2 knockout (KO) cells (Figs. 2 C and S2 A). As the intensity of the Nek2 signal was much stronger on the proximal versus distal pool (Fig. 2 B), we monitored Nek2 localization in RPE1 C-Nap1 KO cells, which lack the proximal Nek2 pool (Fig. 2 D; Spalluto et al., 2012; Hardy et al., 2014; Panic et al., 2015). The colocalization of Nek2 with Cep164 or ODF2 in interphase cells persisted (Fig. 2 D), indicating that Nek2 binds to the distal ends of the M-centriole independently of the proximal pool. When the plane of observation was aligned with the centriole cross-sectional axis, Nek2 appeared as dots or incomplete rings (Fig. 2 D), to suggest that Nek2 association with appendages might be transient. Together, the localization pattern of Nek2 supports the view of Nek2 being a potential kinase that regulates DAs.
Nek2 removes DAs from M-centrioles before mitosis
Next, we asked whether Nek2 is involved in the release of Cep164, Cep123, and LRRC45 before mitosis. For this, we first monitored the fluorescence levels of these proteins throughout the cell cycle in RPE1 Nek2 KO cells. The KO of Nek2 did not interfere with the cell cycle profile of the cells (Fig. S2 B). In contrast to WT cells, high levels of Cep164, Cep123, and LRRC45 remained at the M-centriole in all Nek2 KO cells during mitosis (Fig. 3, A and B). The mitotic attachment of Cep164, Cep123, and LRRC45 on the M-centrosome in Nek2 KO cells was reverted upon expression of low levels of Nek2A in the Nek2 KO background (Fig. S2, C and D), which only moderately decreased the interphase levels of appendages in comparison with control cells (20% reduction; Fig. S2 E). This indicates that the regulation arose from a specific property of Nek2. Our data establish a key role for Nek2 in the release of DAs before mitosis.
DAs are released from interphase centrosomes by higher levels of Nek2
As Nek2 levels peak at G2/M transition (Hames and Fry, 2002), we asked whether overexpression of active Nek2A prematurely displaces DAs from the M-centriole. To address this question, we made use of RPE1 cells stably expressing doxycycline (DOX)–inducible Nek2A or kinase-dead Nek2A (Nek2A-KD) with a K37R substitution (Fry et al., 1995) fused to mNeonGreen (Fig. 4, A and B; and Fig. S3 A). Expression of mNeonGreen-Nek2A and mNeonGreen-Nek2A-KD was observed in ∼75% of cells only in the presence of DOX (Figs. 4 A and S3 A). The overexpression of active, but not inactive, Nek2A caused a significant reduction in Cep164, Cep123, FBF1, and LRRC45 signals at the M-centriole in interphase (Fig. 4 B) in the majority of the Nek2A-overexpressing interphase cells (65–70%). For LRRC45, we analyzed the protein levels in RPE1 C-Nap1–depleted cells, thereby allowing us to exclusively detect the distal pool of LRRC45 (He et al., 2013; Kurtulmus et al., 2018). The analysis of DAs upon C-Nap1 knockdown also confirmed that Nek2A overexpression induced the release of Cep164, Cep123, and FBF1 from centrosomes in the absence of C-Nap1 (Fig. S3 B). This excludes the involvement of Nek2’s proximal pool in DA regulation. Notably, the overexpression of Nek2A-KD led to a significant increase of LRRC45, but no other DAs at the interphase centriole (Fig. 4 B), to reveal a dominant impact of Nek2A upon LRRC45. The removal of DAs by Nek2 did not arise from a pleiotropic impact on all DAs because the levels of Cep83 and SCLT1 remained unchanged upon Nek2A overproduction (Fig. 4, A and B). Also, this effect of Nek2 upon DAs was not due to a shift in cell cycle distribution, as no significant difference was observed in the FACS profile of Nek2-overexpressing and WT cells (Fig. S2 B).
We next considered the possibility that proteolysis of the phosphorylated proteins by the proteasome could account for the Nek2-driven dissociation of DA proteins from centrioles. However, the intensity of DA signals was still significantly decreased upon Nek2A overexpression when the proteasome was inhibited with MG132 (Fig. 4 C). As described previously (Hames et al., 2005), centrosomal intensity of endogenous and overexpressed Nek2A rose upon MG132 treatment (Fig. S3, C and D). This shows that, despite the lack of impact upon the DA components, MG132 treatment had been effective. These data indicate that protein degradation is unlikely to account for DA removal following Nek2A overproduction. Furthermore, DA protein levels did not decrease in mitotic extracts of RPE1 cells, arguing against a global protein degradation of DAs in mitosis (Fig. S3 E). It is therefore most likely that it is the physical dissociation of the proteins from the DA that accounts for their removal.
We next asked whether the elevation of Nek2 expression in cell lines other than RPE1 affected DAs. To this end, we focused our attention on mammary epithelial cell lines because of the recorded up-regulation of Nek2 in those cells (Hayward et al., 2004; Zhou et al., 2013; Kim et al., 2015). We analyzed Cep164 centrosomal protein levels in MCF10A (a near-diploid and nontransformed human mammary epithelial cell line; Soule et al., 1990), MCF10AT, and MCF10CA1 lines. MCF10AT and MCF10CA1 lines were derived from MCF10A cells that had been transformed with oncogenic Ras (V12G) and clonally selected after passage in mice. They represent a model for premalignant mammary cells (MCF10AT) and invasive, metastatic carcinoma (MCF10CA1; Dawson et al., 1996; Santner et al., 2001).
Compared with the intensity of signals in RPE1 cells (Fig. 5 A), there was a marked increase in the proportion of cells exhibiting lower levels of Cep164 at interphase centrosomes in MCF10A, MCF10AT, and MCF10CA1 cells. The decrease in Cep164 levels in MCF10-derived cell lines was a direct impact arising from Nek2 function because depletion of Nek2 using specific siRNA reverted the phenotype (Figs. 5 B and S3 F). This impact of Nek2 upon appendages became more apparent when Nek2-positive and -negative cells within the same line were scored separately (Fig. 5 C), as Nek2-positive cells, in which Nek2 accumulated at centrosomes (Fig. 5 D), had significantly diminished levels of Cep164. Thus, high Nek2 levels repress DA association with the M-centriole.
Previously, Nek2 and Plk1 were proposed to work synergistically to remove the centrosomal component Nlp from centrosomes before mitosis (Rapley et al., 2005). By contrast, we find no reliance upon Plk1 function for the impact of Nek2 upon DA loss from centrioles as the presence or absence of the Plk1-kinase inhibitor BI-2536 did not alter the outcome of Nek2 induction (Figs. 6 A and S3 G), yet it led to an accumulation of cells in prometaphase, as reported previously (Fig. 6 B; Sumara et al., 2004). Consistently, overexpression of hyperactive Plk1 had no impact upon the level of Cep164, Cep123, LRRC45, or FBF1 at interphase centrosomes (Fig. 6, C and D), indicating that Plk1 does not phenocopy Nek2 in respect to DA removal. Furthermore, Nek2A overexpression in G1-arrested cells was sufficient to displace DAs similarly to the effect observed in nonarrested interphase cells (Fig. 6, E and F). These data suggest that Nek2 exerts its function over DAs in the absence of most other mitotic kinases (Fig. 6, E and F).
Regulation of DAs by Nek2 does not require ODF2
ODF2 was proposed to be required for DA establishment in mouse embryonic stem cells (Ishikawa et al., 2005); however, depletion of ODF2 using siRNAs did not influence Cep164 centrosome association in RPE1 cells (Fig. S4 A; Kuhns et al., 2013). To clarify this discrepancy, we used CRISPR/Cas9 to produce an RPE1 ODF2 KO cell line (Fig. S4, B and C). Electron microscopy confirmed that the absence of ODF2 disrupted SDA formation (Fig. S4 D), while immunofluorescence showed that this loss of SDAs was accompanied by a drastic reduction in centriolin and Cep128 levels at centrosomes (Fig. 7, A and B; Ishikawa et al., 2005; Tateishi et al., 2013; Mazo et al., 2016). Furthermore, RT-PCR of RPE1 Nek2 KO cells indicated that all sequenced transcripts contained the mutated exon (data not shown), excluding the possibility of exon skipping (Uddin et al., 2015; Rodriguez-Rodriguez et al., 2018; Meraldi, 2019; Zhang et al., 2019). Interestingly, the levels of DAs at interphase centrosomes remained unchanged in ODF2 KO cells (Fig. 7 C), even after treatment of these cells with ODF2 siRNAs (Fig. S4 A), further excluding the possibility that partial ODF2 mutants would compensate for the lack of ODF2. Furthermore, Cep164, Cep123, and LRRC45 centrosomal levels decreased during mitosis in ODF2 KO cells to the same degree as in WT cells (Figs. 7 D and S4 E). The most logical conclusion that ODF2 is not required for DA regulation is supported by the decline in the levels of Cep164, Cep123, LRRC45, and FBF1 upon Nek2A overexpression in ODF2-depleted cells (Fig. 7 E; and Fig. S4, F–H). Likewise, Cep164 persisted at the M-centriole during mitosis upon Nek2 knockdown in ODF2 KO cells (Figs. 7 F and S4 I). Thus, we conclude that ODF2 is dispensable for DA formation and regulation by Nek2 in RPE1 cells.
Impairment of ciliogenesis upon Nek2 overexpression correlates with reduction of DAs
DAs are required for the initial steps of cilia formation (Graser et al., 2007; Schmidt et al., 2012; Tanos et al., 2013). We postulated that high levels of Nek2 might decrease DAs to levels that are unable to support ciliogenesis. To investigate the correlation between Nek2 overexpression, appendage loss, and ciliation, we overexpressed Nek2A in serum-containing medium and induced ciliogenesis after serum withdrawal. In agreement with previous reports (Spalluto et al., 2012; Kim et al., 2015; DeVaul et al., 2017), we observed a significant reduction in the frequency of ciliation when serum starvation was preceded by overexpression of Nek2A (Fig. 8 A). In line with Kim et al. (2015), ciliation was not decreased when Nek2A overexpression was induced after cells had been serum starved (unpublished data), reinforcing the conclusion that Nek2 is unlikely to act on assembled cilia (Kim et al., 2015; DeVaul et al., 2017). Upon Nek2A induction, Cep164 levels at DAs in nonciliated cells were significantly reduced below the levels seen in either ciliated or nonciliated cells in which Nek2A had not been induced (Fig. 8 B).
To understand why loss of DAs and cilia was only seen in such a small percentage of Nek2A-overproducing cells, we analyzed Nek2 protein levels at both the level of total cell lysates and at the single cell level. The overproduced Nek2A levels were drastically reduced in the absence of serum (Fig. 8 C), as previously noted for endogenous Nek2 (Kim et al., 2015). In agreement with low Nek2 levels upon serum starvation, we could not detect overexpressed Nek2A in the majority of cells (Fig. 8 D). Only ∼30% of the cells had Nek2 at the centrosomes at medium-to-high levels (Fig. 8 D). The reduction of Cep164 at DAs and loss of cilia were only pronounced in cells with higher Nek2 centrosomal levels (Fig. 8, E and F), explaining the low penetrance of this phenotype upon serum starvation.
Nek2 has been shown to regulate cilium disassembly via phosphorylation of Kif24 to prevent the outgrowth of cilia in proliferating cells (Kim et al., 2015). We next asked whether Nek2 could regulate DAs via Kif24. To address this hypothesis, we asked whether centriolar removal of Cep164 upon Nek2A overexpression required Kif24. We found that Cep164 declined to similar levels following Nek2 overexpression in interphase cells, irrespective of whether cells had been treated with control or Kif24 siRNA (Fig. 9, A and B). Furthermore, the mitotic decline in Cep164 centrosomal levels in RPE1 cells was independent of Kif24 depletion (Fig. 9 C). Finally, Kif24 overexpression had no impact upon Cep164 levels (Fig. 9, D and F), yet reduced ciliogenesis (Fig. 9 G), as reported previously (Kobayashi et al., 2011; Kim et al., 2015). We therefore concluded that Kif24 is dispensable for appendage regulation by Nek2.
A ciliary remnant in mitotic Nek2 KO cells leads to the asymmetric inheritance of ciliary signaling components and asynchronous cilium reassembly
As the primary cilium disassembles before mitosis and assembles again in G1 following division (Wang and Dynlacht, 2018), we asked whether the persistence of DAs in mitosis would influence this process. To this end, we analyzed the behavior of mitotic Nek2 KO cells that retain DAs at their M-centriole. While no cilium or cilia remnant was observed during mitosis in WT cells, ∼20% of mitotic Nek2 KO cells showed the cilia membrane marker Arl13B as a dot-like staining at the M-centriole decorated by Cep164 in cycling cells cultured in serum-containing media (data not shown). The number of mitotic cells with a ciliary remnant increased further after serum-starved ciliated cells were restimulated with serum to induce cilia loss. In this case, no cilia or cilia remnant was observed in mitotic WT cells, yet nearly half of the mitotic cells retained a cilia membrane dot (marked by Arl13B) on the M-centrosome (marked by Cep164) in Nek2 KO cells (Fig. 10 A). This indicates that an Arl13B-associated ciliary membrane (referred as ciliary remnant) remains associated with the M-centriole during mitosis, as previously reported for cells with reduced levels of Nek2 (Spalluto et al., 2012). The phenotype was specific to Nek2 loss, as it could be rescued by the inducible expression of Nek2A in the Nek2 KO background (Fig. S5 A). Furthermore, no mitotic remnant was detected after depletion of Cep164 (Fig. S5, B and C), confirming that Arl13B docked at the M-centriole in a Cep164-dependent manner. This phenotype was not a consequence of differences in ciliogenesis because RPE1 WT and Nek2 KO interphase cells displayed a similar capacity for ciliation (Fig. S5 D).
To understand how cilia behave in Nek2 KO cells, we performed live-cell imaging. The videos revealed that ciliated Nek2 KO cells shortened the cilium to a remnant during mitosis, from which a new cilium was formed after division (Fig. 10 B and Video 1). We determined that the daughter cell that retained the cilia remnant reformed a cilium in 2 ± 2.6 h after cell division (Fig. 10, C and D). We did not observe cilia reformation in the daughter cell that inherited the younger centrosome (not decorated by Cep164) within a time period up to 19.3 h. These data suggest that the presence of DAs during mitosis in Nek2 KO cells might contribute to cilia remnant maintenance during mitosis and faster cilia reformation after cell division. This mechanism reflects asymmetric cell division since it gives rise to two different daughter cells.
Finally, we asked whether the mitotic cilia remnant in RPE1 Nek2 KO cells was able to recruit signaling components. The membrane protein Smoothened (Smo), one of the hedgehog (Hh) pathway components, localizes to cilia when stimulated by Hh ligand or by pathway agonists such as N-methyl-N′-(3-pyridinylbenzyl)-N′-(3-chlorobenzo[b]thiophene-2-carbonyl)1,4-diamino-cyclo-hexane (SAG; Chen et al., 2002). In addition, overexpressed Smo enters the cilium in nonstimulated RPE1 cells and, similar to Arl13B, serves as a cilia membrane marker (Rohatgi et al., 2009). Like Arl13B, the mitotic cilia remnant was visible during live-cell imaging of Nek2 KO cells using overexpressed Smo-EGFP (Video 2). Immunofluorescence using an antibody that recognizes endogenous Smo indicated that Smo decorated the mitotic remnant upon SAG activation (Fig. 10 E). We thus conclude that the mitotic cilia remnant can recapitulate the cilia property of retaining signaling molecules.
DA behavior during mitosis
The importance of DAs for cilia biogenesis is well documented (Graser et al., 2007; Schmidt et al., 2012; Tanos et al., 2013), yet their behavior and function during mitosis remain poorly understood. The systematic analysis of DAs in RPE1 cells and HSPCs identified common modes of mitotic behavior for most of the DAs analyzed, including Cep83, SCLT1, FBF1, LRRC45, Cep123, and Cep164. Of these DA components, Cep83 and SCLT1 remained at the older centrosome throughout the cell cycle, to emulate the behavior of the SDA marker ODF2 (Fig. 1; Kong et al., 2014). In contrast, LRRC45, Cep123, and Cep164 were displaced from the M-centriole before mitosis. Interestingly, this behavior of DAs reflects the mode of their assembly (Tanos et al., 2013; Kurtulmus et al., 2018): a stable core composed of Cep83 and SCLT1 recruits the components decreasing in mitosis (Cep164, Cep123, and LRRC45). Similar results were recently reported in a study applying super-resolution microscopy (Bowler et al., 2019).
We propose that DAs can be subdivided into two categories: “core components” with scaffolding functions and “mobile components,” the latter of which are released from the older centrosome in every cell cycle. This suggests a dynamic cell cycle–dependent regulation of DA composition. Such a regulation would allow cells to differentially modulate the recruitment of DAs to mitotic centrosomes depending on the cell type or cellular environment. Cep164 is a good example for a DA shown to behave differently between two cell types (Graser et al., 2007; Schmidt et al., 2012; Kong et al., 2014). Another example is FBF1, which remains at the older centrosome in RPE1 cells (Fig. 1; Tanos et al., 2013), but not in HSPCs (Fig. S1 C). In this context, FBF1 is removed from mitotic M-centrioles as reported for HeLa cells (Fig. S1 C; Kong et al., 2014). The physiological significance of this fascinating cell type–specific regulation remains to be determined. It is puzzling that FBF1 stays at the M-centriole in mitosis in RPE1 cells; yet, the FBF1-targeting protein LRRC45 (Kurtulmus et al., 2018) is partly removed from this structure. This suggests that either residual levels of LRRC45 are sufficient to keep FBF1 at mitotic centrioles or FBF1 might be recruited to mitotic centrioles in an LRRC45-independent manner. Indeed, FBF1 was recently reported to localize at the proximal ends of the centriole in RPE1 and HeLa cells (Inoko et al., 2018). Super-resolution microscopy also indicated that FBF1 localization differs from that of other DAs as FBF1 forms bladelike structures that intercalate with DAs (Yang et al., 2018). Therefore, the hierarchical appendage formation model might not apply to FBF1, depending on the cell type or cell cycle stage.
DA regulation by Nek2
Several lines of evidence support a key role for Nek2 kinase in DA regulation. First, Nek2 increases in levels and activity concomitantly with the decrease of appendage levels in the G2/M transition (Fry et al., 1998b, 2012; Hames et al., 2001; Hames and Fry, 2002; Whitfield et al., 2002). Second, besides its localization at the proximal ends of both M- and D-centrioles (Fry et al., 1998a), Nek2 has also been mapped to colocalize with the distal ends of the centriole by super-resolution microscopy (Fig. 2). Furthermore, we found that overexpression of active Nek2A, but not Nek2A-KD, prematurely displaced Cep164, Cep123, and LRRC45 from M-centrioles in interphase. This displacement was independent of C-Nap1, indicating that Nek2 controls DAs independent of Nek2 regulation of the centrosome linker. Finally, in the absence of Nek2, DAs (Cep164, Cep123, and LRRC45) remained at the M-centriole throughout mitosis.
The mechanism by which Nek2 removes DAs remains to be unraveled. As proteasome inhibition did not prevent the release of DAs upon Nek2A overexpression, we reason that displacement rather than degradation underlies the regulation of DAs by Nek2. This assumption is further supported by the observation that DAs are not degraded in mitosis (Fig. S3 E; Schmidt et al., 2012).
Nek2 might, however, not be the only kinase regulating DAs. In the case of Nlp, Rapley et al. (2005) showed that Nek2 was joined by hyperactive Plk1 in displacing Nlp from the centrosome when overexpressed. In our analysis, Plk1 inhibition had no impact on DA removal by overproduced Nek2A. In addition, overexpression of hyperactive Plk1 did not change DA levels in interphase, suggesting that this process is primarily dependent on Nek2. However, a role of Plk1 in mitosis cannot be excluded. Indeed, a recent study reports that inhibition of Plk1 increased the levels of mitotic DAs to some extent (Bowler et al., 2019). These results point toward a regulation of DAs by multiple mitotic kinases.
Nek2 is a proto-oncogene that is overexpressed in several forms of cancer, including breast cancer (Hayward et al., 2004; Zhou et al., 2013; Cappello et al., 2014). By analyzing mammary epithelial MCF10-derived cell lines with increased levels of Nek2 but decreased primary cilia (Yuan et al., 2010; Kim et al., 2015), we detected a higher percentage of interphase cells with reduced Cep164 levels compared with RPE1 cells. Cep164 levels could be partly restored upon Nek2 depletion, implying that high Nek2 levels were indeed the cause of reduced centrosomal Cep164. It is thus tempting to speculate that DA mis-regulation might be one underlying mechanism that contributes to cilia loss in tumors overexpressing Nek2.
Influence of SDAs in DA regulation
ODF2 has been previously shown to be required for DA assembly in mouse embryonic F9 cells (Ishikawa et al., 2005; Tateishi et al., 2013); however, the analysis of other cell types did not confirm this observation after ODF2 depletion (Kuhns et al., 2013; Tanos et al., 2013), suggesting that residual ODF2 levels could be sufficient for DA assembly. Using RPE1 ODF2 KO cells, we now show that DA assembly and regulation by Nek2 are independent of SDAs in RPE1 cells. In ODF2 KO cells, the levels of DAs at the M-centriole during interphase were not decreased in comparison with those of WT cells. This implies that the dependency of ODF2 for DA formation is most likely cell type dependent.
Consequences of impaired DA regulation
Why should DAs be released from the M-centriole during mitosis? One possibility is that DAs perform functions outside the centrosome. For instance, Cep164 was proposed to be involved in DNA damage control, genomic stability, and cell cycle progression (Sivasubramaniam et al., 2008; Pan and Lee, 2009; Chaki et al., 2012; Slaats et al., 2014). However, centrosome-independent functions of Cep164 have been disputed in a recent publication using RPE1 Cep164 KO cells (Daly et al., 2016). Another reason could be related to their role in cilia biogenesis. Our data show that in RPE1 WT cells, the cilium is fully disassembled before mitosis, as no residual ciliary membrane marker could be observed at mitotic centrosomes. This scenario was different in Nek2 KO cells. In Nek2 KO cells, the percentage of mitotic cells in which the older centrosome was decorated with the cilia marker Arl13b increased considerably. This is in agreement with a previous report showing that siRNA-mediated depletion of Nek2 compromised the cells’ ability to resorb cilia in G2 and prophase (Spalluto et al., 2012). Kim et al. (2015) showed that Nek2 prevents outgrowth of cilia in proliferating cells via phosphorylation of the kinesin motor protein Kif24, thereby likely blocking cilia formation at the G2/M transition. However, in our study the impact of Nek2 upon DAs was not dependent on the activity of Kif24. These data suggest that the Nek2-dependent appendage release at the G2/M transition might constitute a fail-safe mechanism that prevents the persistence of ciliation throughout mitosis and that most likely operates in parallel to the Nek2–Kif24 pathway. We propose that Nek2 has a dual function in controlling ciliogenesis: it regulates Kif24 to inhibit cilia assembly (Kim et al., 2015), while also promoting the mitotic displacement of DAs to ensure the detachment of ciliary vesicles during mitosis (Fig. 10 F).
Our live-cell imaging analysis established that cilia growth from older centrosomes associated with a ciliary remnant seems to initiate faster than a cilium is nucleated from the “naked” older centrosomes that lack a remnant (Anderson and Stearns, 2009). We therefore suppose that, by regulating the extension of cilia disassembly, cells could modulate the speed of cilia regrowth in daughter cells just after cell division. The retention of DAs in mitosis could be part of such a mechanism. Interestingly, neural stem cells were shown to keep a ciliary remnant at the older centrosome in mitosis. The inheritance of the ciliary remnant seems to give cells a temporal advantage in reforming the cilium and becoming responsive to Hh ligands (Paridaen et al., 2013). The centrosomal association of the ciliary membrane in dividing neural stem cells decreased at the later stages of neurogenesis when cells differentiate (Paridaen et al., 2013). The molecular mechanisms underlying this change in behavior are unknown. It would be interesting to determine whether Nek2 controls the behavior of daughter cells by modulating the levels of DAs during mitosis in differentiated versus stem cells.
Materials and methods
Plasmids were L13-Arl13B-GFP (no. 40879; Addgene; Larkins et al., 2011); pRetroX-TRE3G (Takara Bio) with murine Smo-EGFP, monomeric mNeonGreen, and pQCXIZ-TUBG1-mRuby2 (a gift of Elmar Schiebel, University of Heidelberg, Heidelberg, Germany); and AcGFP1-Kif24 (Miyamoto et al., 2015). pRetroX-TRE3G-mNeonGreen-PLK1-T210D was generated by subcloning PLK1-T210D from pCMV-3Tag-1C-PLK1-T210D (Schmidt et al., 2012) into pRetroX-TRE3G-mNeonGreen. For expression of gRNAs and Cas9, the two oligos containing the forward and reverse guide sequences targeting exon 8 of ODF2 isoform 9 (also known as cenexin1, NM_002540.4) were 5′-CACCGTCCCCCCTTACATGTTCACG-3′ and 5′-AAACCGTGAACATGTAAGGGGGGAC-3′. Briefly, oligos were annealed by incubating 1 µl of each oligo (100 µM) with 0.5 µl of T4 Polynucleotide kinase (no. M0201S; New England Biolabs) and 1 µl of 10× ligation buffer in a total of 10 µl for 30 min at 37°C. The reaction was further incubated at 95°C for 5 min with subsequent cooling to 25°C at a cooling rate of 5°C/min. Annealed oligos (1 µl from 1:200 diluted annealing reaction) were ligated into the PX458 plasmid (no. 48138; Addgene) and digested with BbsI using the standard New England Biolabs T4 ligase protocol.
Cell lines used in this study were hTERT-immortalized RPE1 (ATCC CRL-4000), HEK293T cells (ATCC CRL-3216), and GP2-293 (Takara Bio). MCF10A, MCF10CA1 (CA1d), and MCF10AT1 (ATk.cl2; Santner et al., 2001) were a courtesy of Christian Conrad and Katharina Jechow (German Cancer Research Centre and BioQuant Center, Heidelberg University, Heidelberg, Germany).
RPE1 cells stably expressing Tet3G were a gift of Elmar Schiebel. RPE1 Nek2 KO cells (clones 18 and 27), RPE1 Nek2 KO clone 27 with Tet3G, RPE1 Nek2 KO clone 27 with Tet3G pRetrox-Tre3G-mNeonGreen-Nek2A, RPE1 Tet3G pRetrox-Tre3G-mNeonGreen-Nek2A, and RPE1 Tet3G pRetrox-Tre3G-mNeonGreen-Nek2A-KD cell lines are described in Hata et al. (2019). RPE1 C-Nap1 (CEP250) KO cell lines were described previously (Panic et al., 2015).
Stable cell lines expressing L13-Arl13B-GFP were generated using lentiviral transduction. For this, HEK293T cells were transiently transfected with L13-Arl13B-GFP, vesicular stomatitis virus glycoprotein envelope expressing plasmid pMD2.G (no. 12259; Addgene), and lentiviral packaging plasmid psPAX2 (no. 12260; Addgene). Cells were harvested 48 h after transfection. The supernatant was filtered, supplemented with Polybrene at a final concentration of 4 µg/ml, and used to transduce RPE1 and RPE1 Nek2 KO cells. After 2–4 d, GFP-positive cells were selected by FACS (FACSAria II SORP cell sorter; BD Biosciences) and collected as a pool.
Retroviral transduction was used to generate stable RPE1 or RPE1 Nek2 KO L13-Arl13B-GFP cell lines carrying pQCXIZ-TUBG1-mRuby2 (γ-tubulin-mRuby2) and RPE1 or RPE1 Nek2 KO Tet3G pRetroX-TRE3G cell lines carrying Smo-EGFP as well as mNeonGreen-PLK1-T210D. For this, the plasmid carrying the gene of interest was cotransfected with pMDG2 into the HEK293-based retroviral packaging cell line GP2-293. Media were changed after 24 and 48 h. The virus-containing media were harvested and filtered using a 0.45-µm filter (Millipore). Four parts of filtered medium were supplemented with two parts fresh media, one part FBS, and 4 µg/ml Polybrene (Sigma-Aldrich). RPE1 and RPE1 Nek2 KO cells were preseeded in a six-well plate and were infected by adding the virus-containing solution three times every 6 h. Cells were split 24 h after the first transduction. GFP- and mRuby-positive cells were selected by FACS and collected as a pool.
Construction and characterization of ODF2 KO cells
RPE1 ODF2 KO clones were generated by transfecting 106 cells with 15 µg of the gRNA-containing plasmid using the electroporation NEPA21 Transfection System (Nepa Gene) according to the manufacturer’s instructions. 48 h after transfection, GFP-positive cells were single sorted in 96-well plates containing 50% normal and 50% conditioned medium. After expansion, genomic DNA was extracted using the GeneJET Genomic DNA Purification Kit (Thermo Fisher Scientific) and sequenced.
ODF2 KO cells were further characterized by RT-PCR. Briefly, total RNAs were purified from the RPE1 WT and ODF2 KO cell line using the NucleoSpin RNA Plus Purification Kit (Macherey-Nagel) and reverse transcribed to first-strand cDNAs using ReverseAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific) and the oligo(dT)18 primer according to the manufacturer’s recommendations. The resulting products were then used as templates for subsequent PCR with a forward primer (5′-AAGGAAAAAAGCGGCCGCAATGAAGGACCGCTCTTCAAC-3′) and a reverse primer (5′-CTTCGCCTCCGTGATCTTCA-3′) that were designed to target the full-length cenexin1 (isoform 9). The obtained PCR fragments were cloned into pJET 2.1 (CloneJET Kit; Thermo Fisher Scientific) for sequence analyses with pJET1.2 forward and reverse primers. Reads were trimmed and aligned in SnapGene (GSL Biotech LCC). Among 23 independent clones sequenced per genotype, isoform 9/NM_002540.5 (also known as human cenexin1), isoform 13/NM_001351578.1 (human cenexin1 variant 1), isoform 8/NM_001242354.1, and isoform 18/NM_001351587.2 were found.
Isolation of human HSPCs
All experiments involving the use of human HSPCs were approved by the Ethics Committee of the Medical Faculty, University of Heidelberg (Heidelberg, Germany) and performed after obtaining informed consent from all voluntary donors in accordance with relevant guidelines and regulations. Human HSPCs were derived from umbilical cord blood or from healthy allogeneic stem cell donors. The latter had received a mobilization regimen with granulocyte–colony-stimulating factor (10 µg/kg body wt/d subcutaneously for 5 d). A sample of 60 ml of peripheral blood was taken for this study before leukapheresis. HSPCs were isolated as described previously (Wein et al., 2010; Ludwig et al., 2014). Briefly, mononuclear cells were isolated by density gradient centrifugation using the Ficoll-Hypaque technique (Merck KGaA). CD34+ cells from the mononuclear cell fraction were enriched by labeling with magnetic microbeads and sorted twice using an affinity column with the AutoMACS ystem (Miltenyi Biotec GmbH).
All cell lines were grown at 37°C under 5% CO2. RPE1 and derived cell lines were grown in DMEM/F12 (Sigma-Aldrich) supplemented with 10% FBS (Biochrom), 2 mM l-glutamine (Thermo Fisher Scientific), and 0.348% sodium bicarbonate (Sigma-Aldrich). HEK293T and GP2-293 cells were cultured in DMEM with high glucose supplemented with 10% FBS. MCF10A, MCF10A, and MCF10AT1 cell lines were maintained in DMEM/F12 supplemented with 0.1 µg/ml cholera toxin, 10 µg/ml insulin, 0.5 µg/ml hydrocortisone, 0.02 µg/ml EGF, and 5% horse serum. Primary CD34+ cells (HSPCs) were provided by the Department of Internal Medicine, University Hospital Heidelberg (Heidelberg, Germany). Primary HSPCs were cultured in Stemline II Hematopoietic Cell Expansion Medium (Sigma-Aldrich), supplemented with l-glutamine (2 mmol/liter; Thermo Fisher Scientific), thrombopoietin (100 ng/ml; R&D Systems), stem cell factor (100 ng/ml; R&D Systems), granulocyte–colony-stimulating factor (100 µg/ml; R&D Systems), and Flt-3 Ligand (500 ng/ml; R&D Systems). HSPCs were expanded for 4 d after isolation on retronectin-coated dishes (T100B Recombinant Human Fibronectin Fragment; Takara Bio). The coating was performed by incubating the dish for 2 h at 37°C with retronectin at a final concentration of 20 µg/ml in PBS, followed by washing in PBS.
Cell culture treatments and transfection
RPE1 cells were transiently transfected with plasmid DNA by either electroporation using the NEPA21 Transfection System (Nepa Gene) or FuGENE 6 (Promega) according to the manufacturer’s protocol and fixed 24–48 h after transfection. Transient transfection in HEK293T and GP2-293 cells was performed using polyethyleneimine (PEI 25000; Polysciences) as described previously (Longo et al., 2013).
RPE1 cells were seeded to subconfluence (25,000 cells/well in a 24-well plate) and incubated in serum-free medium for 48–72 h to induce cilia formation. To induce cilia disassembly, cells were serum starved for 24 h, followed by 24 h of serum restimulation. Expression of Tet-on-inducible constructs was induced by the addition of DOX (Sigma-Aldrich) at a concentration of 10 ng/ml for 24 and 72 h as indicated in the respective figure legends. In the rescue cell line Nek2 KO with Tet3G pRetrox-Tre3G-mNeonGreen-Nek2A, Nek2A expression was induced by the addition of DOX at a concentration of 1 ng/ml for 24 h. SAG (no. 4366; Tocris) was used at a final concentration of 0.4 µM for 24 h to activate Hh signaling. For visualization of tubulin during live-cell imaging, 20 nM sirTubulin (Cytoskeleton Kit 251SC006; tebu-bio) and 5 µM verapamil (Cytoskeleton Kit 251SC006; tebu-bio) were added 3 h before imaging. Cells were treated with MG132 (SMQ-SIH-537; Biozol) for 4 h at a final concentration of 20 µM to inhibit the proteasome. Cells were treated with BI-2536 (Boehringer Ingelheim) at a concentration of 200 nM for 1 or 6 h to inhibit Plk1. For G1 arrest, cells were treated with the Cdk4/6 inhibitor palbociclib (no. S1116; Selleckchem) at a concentration of 0.1 µM for 24 h and for an additional 24 h at a concentration of 1 µM as indicated in the respective figure legends. All drugs were dissolved in DMSO, and solvent only was used as a negative control. Upon Nek2 overexpression using 10 ng/µl DOX, only cells with medium or high Nek2 levels (detected by immunofluorescence) were considered for appendage quantification, if not stated otherwise.
Transfections of siRNA were performed using Lipofectamine RNAiMAX transfection reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. 30,000–40,000 cells were seeded per well in a 24-well plate and analyzed 72 h after the initial transfection. The following siRNAs (50 nM final concentration) were used: siLuciferase (siControl), 5′-UCGAAGUAUUCCGCGUACG-3′ (Dharmacon ON-TARGET plus/Ambion Silencer Select; Knodler et al., 2010); siNek2, 5′-GAUGCAAUUUGGUCAUUAAUU-3′ and 5′-GAAAGGCAAUACUUAGAUGUU-3′ (Dharmacon ON-TARGET plus; Kim et al., 2015); siCep164, 5′-GGUGACAUUUACUAUUUCA-3′ (Ambion Silencer Select; Graser et al., 2007); siHsODF2-1, 5′-AGACUAAUGGAGCAACAAG-3′ (Ambion Silencer Select; Soung et al., 2009); siODF2-2, 5′-GGAUCUUUAUGUCGCUGAATT-3′ (Ambion Silencer Select); siKif24, 5′-GGAACACCCTGGAGAATAGTT-3′ and 5′-GAGTTGAGCTCTCCTTTGGTT (Dharmacon ON-TARGET plus; Kobayashi et al., 2011); and siC-Nap1 (human CEP250), 5′-GAGCAGAGCUACAGCGAAU-3′ and 5′-AAGCUGACGUGGUGAAUAA-3′ (Dharmacon ON-TARGET plus; Panic et al., 2015).
Real-time quantitative PCR (qPCR)
Real-time qPCR was used to analyze the knockdown efficiency of Kif24 depletion. Briefly, RPE1 cells treated with control and Kif24 siRNA were detached by trypsin treatment and centrifuged to form a cell pellet. RNA was isolated using the RNeasy Mini Kit (no. 74104; Qiagen) according to manufacturer’s instructions. Total RNA was quantified on an ND-1000 spectrophotometer (Nanodrop). Real-time qPCR was performed with 100 ng of RNA using One-Step SYBR PrimeScript Real-Time PCR Kit II (Perfect Real Time no. RR086A; Takara Bio) with ROX reference dye on a 7300 Real-Time PCR System (Applied Biosystems) according to the manufacturer’s protocols. Forward and reverse primer pairs for Kif24 were 5′-GCATGAGAAGAAAGAAGCAGTTGACC-3′ and 5′-GGTGCACGCCTCACCAAAGA-3′ and for GAPDH were 5′-TGCACCACCAACTGCTTAGC-3′ and 5′-GGCATGGACTGTGGTCATGAG-3′. Primer pairs were reconstituted in nuclease-free water. Results were exported to Excel (Microsoft) for analysis. All the corresponding real-time qPCR data were analyzed using the ΔΔCT method and normalized against GAPDH (housekeeping gene; Milstein et al., 2013). The RQ value (fold change) was calculated by using the formula 2(−ΔΔCT).
The following primary antibodies were used: mouse anti-Nek2 (1:250, D-8, sc-55601; Santa Cruz Biotechnology), mouse anti-Nek2 (immunofluorescence 1:100, Western blot 1:500, no. 610593; BD Biosciences), rabbit anti-ARL13B (1:500, no. 17711-1-AP; Proteintech), mouse anti–γ-tubulin clone GTU-88 (1:500, no. T6557; Sigma-Aldrich), rabbit anti–γ-tubulin (1:500, no. T5192; Sigma-Aldrich), rabbit anti-Cep83 (CCDC41, 1:400, no. HPA038161; Sigma-Aldrich), rabbit anti-SCLT1 (1:250, no. HPA036561; Sigma-Aldrich), rabbit anti-FBF1 (1:500, no. HPA023677; Sigma-Aldrich), mouse anti-actin (1:1,000, MAB1501; EMD Millipore), rabbit anti-Cofilin (1:3,000, ab42824; Abcam), mouse anti-GFP (1:800, no. 11814460001; Roche), anti–nuclear pore complex (mAB414, ab24609; Abcam), goat anti-C-Nap1 (1:1,000; Panic et al., 2015), guinea pig anti-Cep164N (immunofluorescence 1:500) and rabbit anti-Cep164M (Western blot 1:400; Schmidt et al., 2012), guinea pig anti-ODF2 (1:500; Kuhns et al., 2013), guinea pig anti-LRRC45 (immunofluorescence 1:400, Western blot 1:500; Kurtulmus et al., 2018), guinea pig anti–Cep123-N (immunofluorescence 1:500, Western blot 1:1,000; Kurtulmus et al., 2018), rabbit anti-centriolin (1:100; a gift of Elmar Schiebel), rabbit anti-Cep128 (1:500, ab118797; Abcam), mouse anti–Plk1 F-8 (1:500, sc-17783; Santa Cruz Biotechnology), rabbit anti–Phospho-Histone3(Ser10) (1:1,000, no. 06-570; Millipore), mouse anti–Smo E-5 (1:100, sc-166685; Santa Cruz Biotechnology), and anti-CENPF (1:100, clone 11; BD Biosciences). Secondary antibodies for immunofluorescence analysis were coupled to Alexa Fluor 488, 546, 594, or 647 dyes (Thermo Fisher Scientific). Antibodies for Western blot analysis were coupled to HRP (Dianova).
Cells were grown on coverslips (no. 1.5; Thermo Fisher Scientific) and fixed in methanol at −20°C for 5 min or prefixed with 3% PFA for 3 min before methanol fixation. Cells were blocked with blocking solution containing 3% IgG-free BSA (Jackson ImmunoResearch) and 0.1% Triton X-100 (Sigma-Aldrich) in PBS for 30 min and incubated with primary antibodies dissolved in blocking solution for 1 h at room temperature. The cells were incubated with conjugated secondary antibodies diluted in blocking solution containing 0.1% DAPI (Sigma-Aldrich) for 1 h at room temperature. Coverslips were mounted with Mowiol (EMD Millipore).
For the staining of HSPCs, cells were fixed in suspension as described above. The washing steps were performed by centrifugation (at 650 g for 5 min at room temperature). After fixation, cells were resuspended in blocking solution and incubated for 1 h at room temperature. Approximately 20–30 µl of cell suspension was dropped per well on Teflon-coated slides (HTC supercured; Thermo Fisher Scientific) and allowed to dry in a laminar flow hood. Dried cells were washed with 30 µl of water and subsequently stained as described above for adherent cells.
Cell cycle stages were determined using 5-ethynyl-2′-deoxyuridine (EdU) and CENPF stainings. The EdU assay was performed using the Click-iT EdU Alexa Fluor 555 imaging kit (Life Technologies). Briefly, cells were treated with 10 µM EdU for 30 min before fixation with methanol at −20°C for 5 min. The Click-IT reaction was performed according to the manufacturer’s recommendations, and samples were subsequently subjected to indirect immunofluorescence analysis using antibodies against CENPF. CENPF and EdU double-negative cells were classified as G1 phase cells. EdU-positive and CENPF-negative cells were classified as S phase cells. CENPF-positive and EdU-negative cells were classified as G2–early prophase cells. Cells with condensed chromosomes and intact nuclei (assessed with DAPI staining) were classified as prophase cells. Mitotic cells were identified by immunofluorescence microscopy based on cell rounding, chromosome morphology (condensed chromosomes), nuclear envelope breakdown, or intercentrosomal distances.
For live-cell imaging, cells were seeded in CELLview culture dishes (Greiner Bio-One). Experiments were performed over a 24-h period at 37°C with 5% CO2. Seventeen Z-stacks of 1.25-µm spacing were taken every 10 min for 24 h. Images were acquired using an Axio Observer Z1 fluorescence motorized microscope (Zeiss) equipped with 63× NA 1.4 Plan-Apochromat oil immersion objective and AxioCam MRm charge-coupled device camera using Zeiss Efficient Navigation software. For 3D-SIM, cells were processed for immunofluorescence microscopy as described above, except that cells were mounted in Prolong Gold antifade reagent (Molecular Probes). Samples were analyzed with a 3D-SIM Ti inverted microscope (Nikon) equipped with three lasers (488, 561, and 647 nm), an Apo total internal reflection fluorescence 100 × 1.49 NA oil immersion objective (Nikon), and an iXon3 DU-897E single-photon detection electron multiplying charge-coupled device camera (Andor). After image capture, raw images were reconstructed in the NIS-Elements program of the microscope to obtain 3D-SIM images.
Image processing and analysis
Images were processed in Fiji (Schindelin et al., 2012) and Illustrator CS5 (Adobe). For figures, the maximum projection of representative images was generated using Fiji. Image brightness and contrast were adjusted equally in Fiji. Quantification of fluorescence intensity was performed using maximum projection of images. Centrosomes were segmented based on γ-tubulin staining. For signal intensity measurements, the mean fluorescence intensity of an area of 25 square pixels around the centrosome was calculated. The mean fluorescence intensity of an area adjacent to the centrosome was used for background correction. In all quantifications of mitotic cells, centrosome 1 was considered the centrosome carrying higher levels of appendages (M-centrosome, as defined by higher levels of ODF2). Intercentriolar distances were manually measured using Fiji. Intensity measurements from replicate experiments were normalized to the average of each repetition and combined. Statistical tests were performed in Excel (Microsoft), R (Synergy Software), and Prism 8 (GraphPad). The data were plotted as boxplots using R and as bar graphs using Excel. Co-localization of signals in 3D-SIM images was analyzed using the Plot Profile tool in Fiji, and graphs were normalized and plotted in Excel.
Cell cycle profile of control, Nek2 KO, and Nek2A-overexpressing RPE1 cells were determined by analyzing total DNA content stained with propidium iodide. Briefly, cells were fixed with ice-cold ethanol (70%) and incubated with staining solution (50 µg/ml propidium iodide, 0.08% Triton X-100, 0.2 mg/ml RNase A, and 1 mM EDTA in PBS) for 30 min. Cells were subjected to analysis on a BD FACSCanto II flow cytometer and BD FACSDiva 8 software (BD Biosciences).
Cells were grown on 12-mm coverslips, rinsed twice with PBS, and fixed at room temperature for 30 min in 2.5% glutaraldehyde in 50 mM cacodylate buffer containing 2% sucrose. Cells were subsequently washed with 50 mM cacodylate buffer and postfixed with 2% osmium tetroxide in 50 mM cacodylate buffer for 45 min on ice and in darkness. Cells were rinsed with water and stained overnight at 4°C with 0.5% uranyl acetate in water. Cells were rinsed with water and dehydrated with ethanol in a stepwise manner using 40, 60, 70, 80, 90, 95, and 100% water-free ethanol. Capsules filled with Spurr resin were immediately placed on top of each coverslip and polymerized for 24 to 48 h at 60°C. The embedded cells were sectioned using a Reichert Ultracut S microtome (Leica Instruments) to a thickness of 70 nm. Poststaining was performed with 3% uranyl acetate in water and lead citrate. Sections were imaged in a JE-1400 transmission electron microscope (JEOL Ltd.) at 80 kV, equipped with a 4k × 4k digital camera (F416; TVIPS). Micrographs were adjusted in brightness and contrast using ImageJ (Schneider et al., 2012).
Cell lysis and Western blot
Cells were scrapped out of the dishes, transferred into 1.5-ml tubes, and washed with PBS by centrifugation. Pellets were lysed in SDS sample buffer (63 mM Tris-HCl, pH 6.8, 10% [vol/vol] glycerol, 2% [wt/vol] SDS, 0.01% [wt/vol] bromphenol blue, and 100 mM DTT) containing 0.1% Benzonase solution (101654; Merck) at 37°C for 30 min. For the mitotic shake-off analysis (Fig. S3 E), cells were treated with 50 nM CENP-E inhibitor in DMSO or DMSO only (control) for 16 h. Mitotic cells were collected by shaking off the dishes (Jackman and O’Connor, 1998). Cell pellets were lysed in RIPA buffer (25 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS, and 0.1% Benzonase). Lysates were cleared by centrifugation (20,000 g for 5 min at room temperature), heated at 95°C for 5 min, and subjected to SDS-PAGE. For analysis of mNeonGreen-PLK1-T210D (Fig. 6 C), cells were lysed in 8 M urea containing 0.1% Benzonase for 1 h at room temperature, and SDS sample buffer was added before loading. After transfer to a nitrocellulose or polyvinylidene fluoride membrane, immunoreactive proteins were detected with the respective antibodies and visualized with enhanced chemiluminescence (Pierce no. 32106; Thermo Fisher Scientific).
Statistical analysis was applied for the mean values of each repetition. Two-tailed Student’s t tests were applied in case two conditions were compared (Figs. 5 C; 6 C; 7 C; 9 E; S2 E; S3 B; S4, F and H; and S5 D). When more than two conditions were compared, one-way ANOVA tests were applied followed by post hoc tests. Tukey’s post hoc test was applied when every condition was compared with every other condition (Figs. 1 C; 6, A and F; 7 E; 8, B, E, and F; 9 A; S2 B; S4 A; and S5 B); Dunnett’s post hoc test was applied if several conditions were compared with one control condition (Figs. 1,D; 3 B; S1 C; and S2 D). Holm–Šídák post hoc test was applied to compare selected experimental pairs of conditions (Figs. 4, B and C; 5 B; S3 F; and S5 C). Significance probability values are as follows: NS, P > 0.05; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001. The sampling sizes for quantifications are indicated in the respective figures or figure legends. Boxplots show the lower and upper quartiles; the whiskers show the minimum and maximal values excluding outliers. Outliers were calculated as values greater or lower than 1.5 times the interquartile range. Box/dot plots show outliers. The line inside the box indicates the median. Bar graphs show the mean ± SD of the independent experiments.
Online supplemental material
Fig. S1 illustrates a scheme for centrosome maturation, representative images for the cell cycle–dependent behavior of the DA Cep164, and the cell cycle behavior of appendage proteins in primary HSPCs. Fig. S2 shows the characterization of the Nek2 KO cell line and the phenotypical rescue of Nek2 KO cells. Fig. S3 demonstrates the analysis of Nek2A overexpression and DAs. Fig. S4 shows the characterization of ODF2 KO cells and the effect of ODF2 KO on DA regulation. Fig. S5 demonstrates that the ciliary remnant in mitotic Nek2 KO cells can be rescued upon siRNA of Cep164 and expression of Nek2 in the Nek2 KO background. Video 1 shows representative mitotic progression of Nek2 KO cells stably expressing Arl13B-GFP and γ-tubulin-mRuby2. Video 2 shows representative mitotic progression of Nek2 KO cells overexpressing Smo-EGFP.
We thank Marko Panic and Elmar Schiebel (University of Heidelberg, Heidelberg, Germany), Christian Conrad and Katharina Jechow (German Cancer Research Center, Heidelberg, Germany), Patrick Horn and Caroline Pabst (University of Heidelberg, Heidelberg, Germany) for sharing cell lines and reagents; Iain Hagan (Cancer Research UK Manchester Institute, Manchester, UK) for sharing protocols; Angela Lenze (University of Heidelberg, Heidelberg, Germany) for purification of CD34+ HSPCs; Astrid Hofmann (member of the Pereira lab) for excellent technical assistance; Volker Eckstein and Monika Langlotz (University of Heidelberg, Heidelberg, Germany) for FACS and help with cell cycle analysis; Damir Krunic and Felix Bestvater (German Cancer Research Center, Heidelberg, Germany) and Ulrike Engel (University of Heidelberg, Heidelberg, Germany) for providing access to microscopes, help with imaging, and image processing; and Elmar Schiebel, Iain Hagan, and members of the G. Pereira laboratory for critical comments on the paper.
This project was funded by the collaborative research grant of the Deutsche Forschungsgemeinschaft (SFB873 Project A14) to G. Pereira. Core funding for microscopy and FACS was provided by the Deutsche Forschungsgemeinschaft SFB873 projects Z03 and Z02, respectively. L. Viol is a member of the Heidelberg Biosciences International Graduate School and funded by the Deutsche Forschungsgemeinschaft SFB873. G. Pereira holds a Heisenberg Professorship from the Deutsche Forschungsgemeinschaft (PE1883/3).
The authors declare no competing financial interests.
Author contributions: L. Viol performed and analyzed all the experiments and assembled figures, except Fig. S4 D. A. Neuner performed and analyzed the electron microscopy experiment reported in Fig. S4 D. S. Hata and A. Pastor-Pediro constructed cell lines overexpressing Nek2 and Nek2 KO cell line. F. Murke, P. Wuchter, A.D. Ho, and B. Giebel provided tools and cell lines. S. Hata contributed to the conceptualization of the paper. G. Pereira conceptualized and supervised the project and wrote the paper, with contributions from all authors.
P. Wuchter’s present address is Institute of Transfusion Medicine and Immunology, Medical Faculty Mannheim, Heidelberg University, German Red Cross Blood Service Baden-Württemberg–Hessen, Mannheim, Germany.