Centrosome maturation requires phosphorylation-mediated sequential domain interactions of SPD-5

Centrosomes, the major microtubule organizing centers (MTOCs) in animal cells, consist of two centrioles and a surrounding amorphous mass of proteins called the pericentriolar material (PCM). The size of the PCM dramatically changes in a cell-cycle dependent manner. During interphase, a small amount of PCM is assembled adjacent to the centrioles. When cells enter mitosis, centrosomal proteins including γ-tubulin complexes are recruited to the PCM to expand its size and increase its microtubule nucleation activity. This process is called centrosome maturation (Palazzo et al., 2000).

During centrosome maturation, PCM scaffold proteins play pivotal roles to recruit other centrosomal proteins around centrioles. One of the well-characterized PCM scaffold proteins is Centrosomin (Cnn) in Drosophila melanogaster and its ortholog CDK5RAP2 in humans (Megraw et al., 1999; Fong et al., 2008). In Caenorhabditis elegans, SPD-5 is thought to be a functional homolog of Cnn/CDK5RAP2, even though sequence homology is undetectable (Hamill et al., 2002). These functional homologs contain multiple predicted coiled-coil regions, and scaffold assembly of these proteins is facilitated by the interaction with the conserved coiled-coil proteins SPD-2 (also known as Cep192 in humans) and phosphorylation by the polo-like kinase 1 (PLK-1; also known as PLK1 in humans, Polo in Drosophila) (Conduit et al., 2014a; Decker et al., 2011; Giansanti et al., 2008; Gomez-Ferreria et al., 2007; Haren et al., 2009; Kemp et al., 2004; Pelletier et al., 2004). These similarities suggest that a universal mechanism of PCM scaffold assembly might exist.

Through studies of Cnn, its distinct domains and their roles in scaffold assembly have been identified. Cnn has two motifs that are conserved in CDK5RAP2; the N-terminal Centrosomin motif 1 (CM1) domain that recruits γ-tubulin complexes to the centrosomes (Fong et al., 2008; Zhang and Megraw, 2007), and the C-terminal CM2 domain that is implicated in centrosomal targeting (Barr et al., 2010; Feng et al., 2017). During early mitosis in Drosophila, Cnn is recruited to the centrioles mainly by Spd-2, then phosphorylated by Polo at the central region called the phospho-regulated multimerization (PReM) domain (Alvarez-Rodrigo et al., 2019; Conduit et al., 2014a, b). The phosphorylation of the PReM domain enhances the interaction between this domain and the CM2 domain. This phosphorylation-dependent conformational change of Cnn leads to the assembly of the PCM scaffold (Feng et al., 2017).

In C. elegans, PCM scaffold protein SPD-5 is recruited to centrioles by centriolar/PCM protein SPD-2, and they depend on each other for localization to the expanded PCM (Hamill et al., 2002; Kemp et al., 2004; Pelletier et al., 2004). SPD-2 also functions as a PLK-1 recruiter, and PLK-1 phosphorylation activity is required for centrosome maturation (Decker et al., 2011; Erpf et al., 2019). The mutation of the PLK-1 phosphorylation sites in the SPD-5 central region results in the failure of scaffold assembly in vivo and in vitro (Woodruff et al., 2015; Wueseke et al., 2016). In vitro reconstitution studies have further demonstrated that purified SPD-5 assembles into spherical condensates which can recruit other downstream PCM proteins. This SPD-5 scaffold assembly is robustly promoted by the addition of SPD-2 and PLK-1 (Woodruff et al., 2017, 2015). Recently, the centriolar/PCM protein PCMD-1 was identified as a novel factor involved in PCM organization. Genetic analyses have indicated that PCMD-1 is required for the recruitment of SPD-5 around centrioles and the integrity of the PCM scaffold (Erpf et al., 2019). Although the involvement of SPD-2, PCMD-1 and PLK-1 is implicated, how SPD-5 interacts with each of these proteins and is regulated by phosphorylation in the PCM scaffold assembly is poorly understood.

In this study, we conducted in vivo domain analysis of SPD-5 and identified three functional domains. We propose a model for SPD-5 scaffold assembly during centrosome maturation through sequential interactions of these domains mediated by PLK-1-dependent phosphorylation.

To understand the mechanism(s) by which SPD-5 is localized to the centrosomes and participates in PCM scaffold formation, we conducted a domain analysis of SPD-5 in vivo (Fig. 1A). Several truncated versions of SPD-5 proteins tagged with green fluorescent protein (GFP) at their N-terminus were expressed in early C. elegans embryos from a transgene inserted into the specific locus on Chromosome II that is distinct from the spd-5 locus. The localization of transgenic proteins and embryonic phenotypes were investigated in the presence and the absence of endogenous SPD-5. The expression of each SPD-5 fragment was confirmed by measuring the GFP signal intensity of whole embryos and western blotting of worm extracts (Fig. 1B; Fig. S1A).

In both the presence and absence of endogenous SPD-5, the full-length SPD-5 [GFP::SPD-5(FL); amino acids (aa) 1−1198] localized to the centrosomes throughout the cell cycle in all observed early embryos, as previously reported (Woodruff et al., 2015) (n=15 and 21, respectively) (Fig. 1A). As GFP::SPD-5(FL) fully rescued the embryonic lethality of spd-5(RNAi) (n=21), we concluded that this fusion protein is functional in vivo. However, we noted that in the absence of the endogenous SPD-5, the shape of the centrosomal signal was slightly distorted, which might be caused by the lower amount of GFP::SPD-5(FL) relative to that of endogenous SPD-5 (Fig. S1A) and/or the effect of the GFP tag.

We first divided the SPD-5 protein into two regions, the N-terminal half (N-half, aa 1−522) and the C-terminal half (C-half, aa 523−1198) (Fig. 1A). In all observed embryos, the GFP::SPD-5(N-half) signal was barely detected at centrosomes in either the presence (n=10) or absence (n=5) of endogenous SPD-5, although in the latter case some condensates were detected in the cytosol at prometaphase to anaphase (Fig. S1B).

In contrast, GFP::SPD-5(C-half) localized to the centrosome in the presence of endogenous SPD-5 (100%, n=11) (Fig. 1A). In the absence of endogenous SPD-5, the centrosomal signal of GFP::SPD-5(C-half) became weaker and smaller, indicative of its centriolar localization (100%, n=15) (Fig. 1A,C). Our observation that GFP::SPD-5(C-half) is insufficient for the full expansion of the PCM implies the involvement of the N-half region in PCM scaffold assembly. However, GFP::SPD-5(C-half) can interact with endogenous SPD-5 to localize to the centrosomes, suggesting that the C-terminal half of SPD-5 is involved in its targeting to the centrioles and interacts with endogenous SPD-5.

We further dissected the SPD-5 C-terminal half for its ability to interact with SPD-5 itself and its function in centriolar targeting. When the last 272 residues of the C-terminal (C1 region, aa 927−1198) were deleted [GFP::SPD-5(ΔC1)], localization to the PCM and the centrioles was abolished in either the presence (100%, n=4) or absence (100%, n=5) of endogenous SPD-5 (Fig. 1A), suggesting that the C1 region is necessary for both self-interaction and centriolar localization.

In turn, the C1 region [GFP::SPD-5(C1)] localized to the centrosome in the presence of endogenous SPD-5 (100%, n=9; Fig. 1A), but the signal area was significantly smaller [27% of that of GFP::SPD-5(FL)] and was fragmented (Fig. 1D; Fig. S1C). GFP::SPD-5(C1) accumulated around the centrioles in the presence of endogenous SPD-5 (Fig. 1A; Fig. S1C). Immunostaining using an antibody against SPD-5 (anti-SPD-5) that recognizes both endogenous SPD-5 and GFP::SPD-5(C1) revealed that endogenous SPD-5 colocalized with the disordered and fragmented signal of GFP::SPD-5(C1) (100%, n=8; Fig. 2A). Notably, this strain showed ∼46% embryonic lethality (Fig. 2B). These results suggest that the C1 fragment disrupts the ability of endogenous SPD-5 to assemble the PCM scaffold in a dominant-negative manner.

In the absence of endogenous SPD-5, the GFP::SPD-5(C1) signal became even weaker and smaller (100%, n=11; Fig. 1A,C), implying that this region of SPD-5 localized only around the centrioles. These results suggest that the C1 region is involved in centriolar targeting and in intermolecular interactions with other SPD-5 molecules.

The C-terminal half of the C1 region contains a region that is highly conserved among the Caenorhabditis genus, which we refer to as the C2 region (aa 1061−1198). In the presence of endogenous SPD-5, GFP::SPD-5(C2) localized to the centrosome (100%, n=8), although the signal area was ∼15% smaller than that of SPD-5(FL) (Fig. 1A,D). A further deletion of 30 residues (C3, aa 1091−1198) abolished this centrosomal localization (100%, n=5; Fig. 1A). Unlike GFP::SPD-5(C1), GFP::SPD-5(C2) did not show a dominant-negative effect, and the localization of endogenous SPD-5 was unaffected (Fig. 2A,B). This centrosomal localization of GFP::SPD-5(C2) was lost in the absence of SPD-5 (92%, n=12; Fig. 1A). Thus, the C2 region interacts with endogenous SPD-5 in the central region of the PCM, but it lacks the ability to independently localize around the centrioles.

The fragment lacking the majority of the C2 region [GFP::SPD-5(ΔC3)] localized around the centrioles in the presence of endogenous SPD-5 but failed to expand to the PCM (100%, n=4; Fig. 1A). This suggests that the C2 region is dispensable for centriolar localization but is necessary for incorporation into the PCM.

The remaining N-terminal half of the C1 region is referred to as the C4 region (aa 927−1060). In both the presence and absence of endogenous SPD-5, GFP::SPD-5(C4) localization was represented by very small signals, plausibly around the centrioles (92%, n=13; 100%, n=12, respectively) (Fig. 1A,C; Fig. S1C,D), suggesting that the C4 region by itself can localize to the centrioles, independently of the interaction with endogenous SPD-5. GFP::SPD-5(C4) did not show a dominant-negative effect on PCM scaffold assembly of endogenous SPD-5 (Fig. 2A,B).

Taken together, these results identified two functional regions in the C-terminal half of SPD-5: C2 (aa 1061−1198), which is involved in the intermolecular interaction with endogenous SPD-5, and C4 (aa 927−1198), which is necessary and sufficient for centriolar targeting.

To identify proteins involved in the recruitment of the SPD-5(C4) domain to the centrioles, we conducted a yeast two-hybrid screen. Using SPD-5(C1) as the bait, the centriolar protein PCMD-1 was identified. The interaction between SPD-5 and PCMD-1 has been reported previously (Stenzel et al., 2021). PCMD-1 also interacted with SPD-5(C4), but not with SPD-5(C2) (Fig. 3A). SPD-5(C4) interacted with the N-terminal half region of PCMD-1 (aa 1−242), which contains predicted coiled-coil and disordered regions (Fig. 3B).

Colocalization of PCMD-1 and the SPD-5(C4) domain was confirmed by immunostaining using the anti-PCMD-1 and anti-GFP antibodies. In the absence of endogenous SPD-5, both GFP::SPD-5(C4) and GFP::SPD-5(C1) colocalized with PCMD-1 on the centriole (100%, n=18; 100%, n=18, respectively), whereas GFP::SPD-5(C2) did not (100%, n=16; Fig. 3C). To verify that the centriolar localization of SPD-5(C4) is dependent on PCMD-1, the localization of GFP::SPD-5(C4) was observed in the pcmd-1(RNAi) embryo. The centriolar localization of GFP::SPD-5(C4) significantly decreased in pcmd-1(RNAi) embryos (Fig. 3D,E) (peak values; P<0.0001, Welch's t-test). From these results, we concluded that the C4 region of SPD-5 is recruited to the centrioles by PCMD-1 through direct interaction. The C4 region is hereafter called the CL (centriole localization) domain.

Additionally, a yeast two-hybrid screen using PCMD-1 as the bait identified the centrosomal proteins TAC-1, PLK-1 and PLK-2 (Fig. S2). The interaction between PCMD-1 and PLK-1 has recently been reported (Stenzel et al., 2021).

Next, we examined which region of SPD-5 interacts with the C2 region by yeast two-hybrid assays. The C2 region (and the C1 region, which contains C2) interacted with the middle region of SPD-5 (aa 272−732) (Fig. 4A). The shorter fragments SPD-5(272−680), SPD-5(400−680) and SPD-5(523−732) did not interact with the C2 region (Fig. 4A), implying that a relatively long (∼460 aa) central region is required for the interaction with the C2 region.

As SPD-5(272−732) contains the PLK-1 phosphorylation sites (S530, S627, S653 and S658) required for the expansion of the SPD-5 scaffold (Woodruff et al., 2015; Wueseke et al., 2016), we next examined by pull-down assays whether their phosphorylation by PLK-1 affects the interaction with SPD-5(C2). Purified glutathione S-transferase (GST)-tagged SPD-5(272−732) proteins with or without PLK-1 pre-treatment were incubated with purified SPD-5(C2) proteins (all proteins used are shown in Fig. S3A,B). The interaction of SPD-5(C2) with wild-type SPD-5(272–732_WT) or phosphoresistant SPD-5(272–732_4A) (containing the S530A, S627A, S653A and S658A mutations) significantly increased by PLK-1 phosphorylation (Fig. 4B), but interaction with SPD-5(272–732_4A) was ∼25% lower than that with SPD-5(272–732_WT). These results imply that PLK-1-phosphorylation in SPD-5(272–732) at multiple residues, including the four serine residues, strongly enhances C2 binding.

To verify that the centrosomal localization of SPD-5(C2) is dependent on the interaction with phosphorylated SPD-5 by PLK-1, the localization of full-length or truncated SPD-5 fragments [GFP::SPD-5(xx)] was observed in the absence of PLK-1 activity. Embryos permeabilized by perm-1(RNAi) were treated with the PLK-1 inhibitor BI-2536 to inactivate PLK-1, or the solvent DMSO. SPD-5 fragments that contain the CL domain [GFP::SPD-5(FL), GFP::SPD-5(C1) and GFP::SPD-5(C4)] localized to the centrioles in the absence of PLK-1 activity (100%, n=28; 100%, n=25; 100%, n=27, respectively) (Fig. 4C,D). We noted that the amount of GFP::SPD-5(C1) localized to centrioles was higher than that of GFP::SPD-5(FL) or GFP::SPD-5(C4) (Fig. 4D). GFP::SPD-5(FL) localization around the centrioles also indicated that a small amount of endogenous SPD-5 remained around the centrioles even after PLK-1 inhibition. On the other hand, GFP::SPD-5(C2) that lacks the CL domain did not show any localization (77%, n=31), or showed weak centriolar localization (23%) (Fig. 4C,D). These results indicate that GFP::SPD-5(C2) localizes to the centrosomes through interaction with endogenous PLK-1-phosphorylated SPD-5.

These in vivo and in vitro domain analyses revealed two functional regions of SPD-5 (Fig. 5A). The SPD-5(272−732) region, which contains the key PLK-1 phosphorylation sites and was essential for SPD-5 scaffold expansion, is the PReM domain, named after the Centrosomin PReM domain in Drosophila (Conduit et al., 2014a). The C2 region, which was necessary and sufficient for the phosphorylation-dependent interaction with SPD-5(272–732) (PReM) is referred to as the PReM association (PA) domain.

Here, we propose a model for SPD-5 scaffold formation during centrosome maturation (Fig. 5B). (i) SPD-5 is recruited to the centrosomes through the interaction between its CL domain and PCMD-1, which is localized to the centrioles (Erpf et al., 2019). (ii) As PCMD-1 interacts with both SPD-5 and PLK-1 (this study and Stenzel et al., 2021), this interaction promotes phosphorylation of the PReM domain of SPD-5 by PLK-1 in the vicinity of the centrioles. The phosphorylated PReM domain interacts with the PA domain of the same and/or other SPD-5 molecules. (iii) This interaction between the phosphorylated PReM domain and the PA domain promotes SPD-5 multimerization to expand the PCM scaffold.

We speculate that both intermolecular and intramolecular interactions between the PA domain and the phosphorylated PReM domain are involved in PCM formation. The intermolecular interaction is indicated by the localization of GFP::SPD-5(PA) to the PCM in the presence of endogenous SPD-5. The requirement of intramolecular interaction is indicated by the result that GFP::SPD-5(ΔC3) that contains the PReM domain but not the PA domain did not expand to the PCM even though it localized to the centrioles. A previous study (Wueseke et al., 2016) showed that the phosphoresistant mutant of SPD-5 [SPD-5(4A)] could not form the mitotic PCM by itself even though it could localize to the PCM in the presence of endogenous SPD-5, which is consistent with our model that the intramolecular and intermolecular interaction between the PA domain and the phosphorylated PReM domain is crucial for PCM expansion.

We noted some discrepancies between the in vitro and in vivo data. Our in vitro assay showed the interaction between PLK-1-treated SPD-5(272–732_WT and 4A) (PReM) and SPD-5(C2) (PA) fragments (Fig. 4B). However, GFP::SPD-5(ΔC1) and GFP::SPD-5(ΔC3) that contain the PReM domain did not localize to the PCM in the presence of endogenous SPD-5 in early embryos (Fig. 1A). These discrepancies might be explained by the difference in protein concentrations between the in vitro and in vivo experiments. In our in vitro assay, high concentrations of purified SPD-5 fragments were incubated in the absence of other proteins, which might be preferable for the protein interactions than physiological conditions. In the in vivo experiment, the expression levels of GFP::SPD-5(ΔC1) and GFP::SPD-5(ΔC3) were several fold lower than the endogenous SPD-5 (Fig. S1A).

A dominant-negative effect of the CL+PA fragment (C1) also supports the importance of the intramolecular interactions between the PReM and the PA domains in the process of PCM expansion. The CL+PA fragment (C1) caused a dominant-negative effect on PCM expansion, whereas GFP::SPD-5(C-half), which contains a part of the PReM domain, in addition to the CL and the PA domains, did not. SPD-5(C1) interacted with both centriolar PCMD-1 and the PReM domain of SPD-5 and accumulated around the centrioles only in the presence of endogenous SPD-5, raising the possibility that over-accumulation of SPD-5(C1) with endogenous SPD-5 around the centrioles prevents the physical connection between the centrioles and the PCM (Fig. 5C). Consistent with this hypothesis, the shape of the PCM in the GFP::SPD-5(C1)-expressing embryo resembled that in a pcmd-1 mutant or the centriole-ablated condition that causes centriole–PCM connection defects in which the PCM is ripped apart by microtubule-mediated pulling forces (Erpf et al., 2019; Cabral et al., 2019). We speculate that a part of the PReM domain contained in GFP::SPD-5(C-half) is required for dissociation from centrioles, further implicating the requirement of intramolecular interactions between the PReM and the PA domains in the process of PCM expansion.

In this study, we demonstrated that SPD-5 is recruited to the centrioles through the interaction between its CL domain and centriolar PCMD-1. Previous studies proposed another model that SPD-5 binds throughout the pre-built PCM, then assembles into the PCM scaffold (Laos et al., 2015; Zwicker et al., 2014; Cabral et al., 2019). We speculate that these two routes of SPD-5 incorporation to the centrosome are not mutually exclusive and might reflect the two phases of mitotic PCM formation; at the beginning of mitotic PCM formation, SPD-5 is recruited to the centrioles in a PCMD-1-dependent manner, then once the PCM scaffold is assembled, SPD-5 can also be isotropically incorporated into the PCM.

Previous in vitro experiments suggested that SPD-5 plays a crucial role in PCM scaffold formation in C. elegans, which is dependent on phosphorylation-regulated interactions among coiled-coil proteins (Woodruff et al., 2017, 2015), but the involvement of specific domains has not been reported. Our results indicated that PLK-1 phosphorylation promotes specific interactions between the PReM and the PA domains of SPD-5 that are crucial for scaffold formation. Thus, PCM scaffold formation is not solely dependent on non-specific interactions between coiled-coil proteins, but rather, it requires well-organized, specific interactions.

SPD-5 has been thought to be a functional homolog of Drosophila Cnn and human CDK5RAP2, although sequence homology is undetectable (Fong et al., 2008; Hamill et al., 2002; Megraw et al., 1999).

Despite the lack of sequence similarity, Cnn and SPD-5 have some common features in their domain structures. Their N-termini have a binding domain for a γ-tubulin complex (Ohta et al., 2021; Zhang and Megraw, 2007). Both proteins have a domain for centrosomal targeting at their C terminus (CM2 for Cnn and CL for SPD-5) (Feng et al., 2017) and have a PReM domain in their central region that contains Polo/PLK-1 phosphorylation sites (Conduit et al., 2014a; Woodruff et al., 2015). Phosphorylation in the PReM domain by Polo/PLK-1 triggers the interaction between the PReM domain and a C-terminally located domain (CM2 for Cnn and PA for SPD-5), which leads to PCM scaffold assembly (Feng et al., 2017). Notably, the CM2 domain in Cnn carries out the combined roles of the CL and PA domains of SPD-5.

In conclusion, we have revealed the PCM scaffold assembly process mediated by SPD-5 through in vivo domain analyses. SPD-5 is recruited to the centrioles through direct interaction between its CL domain and PCMD-1. Phosphorylation by centriolar PLK-1 of the PReM domain of SPD-5 promotes its intramolecular and/or intermolecular interactions with the PA domain, which leads to mitotic PCM scaffold assembly. Our findings further suggest an evolutionarily conserved domain organization between Cnn and SPD-5, and their phosphorylation-dependent regulation. Further analysis of SPD-5 and other PCM proteins will lead to a better understanding of the universality and diversity of the PCM formation mechanisms.

All C. elegans strains were cultured as described (Brenner, 1974). The strains used in this article are listed in Table S1 and were maintained at 23°C.

Transgenes for GFP-tagged SPD-5 variants were inserted into the same genomic position with the MosSCI ttTi5605 allele on chromosome II (Frøkjaer-Jensen et al., 2008, 2012) by CRISPR/Cas9-mediated genome editing. We constructed a homologous repair template vector, pHK6, to insert SPD-5 variant transgenes into the ttTi5605 site. pHK6 contains arms for homologous recombination at the ttTi5605 site (Chr. II armL, 920 bp and armR, 967 bp), a GFP-encoding sequence under the control of the germline-specific pie-1 promoter and 3′ untranslated region (3′UTR), the self-excising cassette (SEC) (Dickinson et al., 2015) for hygromycin B selection and the ccdB sequence, which was substituted by each SPD-5 variant sequence (pHK6: Chr. II armL_pie-1p_coGFP_SEC_3×FLAG_ccdB_pie-1 3′UTR_Chr. II armR). The SEC_3×FLAG_ccdB fragment was from pDD282 (Addgene #66823), except that the let-858 3′UTR was substituted by the par-5 3′UTR to prevent gene silencing in the germline. To construct homologous repair templates for SPD-5 variants, pHK6 was digested with ApaI restriction enzyme, which recognizes sites at both ends of the ccdB sequence, and the coding sequence for individual SPD-5 variants was assembled using the Gibson assembly method (Gibson et al., 2009).

For single guide RNA (sgRNA) expression, we constructed the pTK73 vector, which was modified from pRB1017 (Fire et al., 1998) to incorporate the sgRNA^(F+E) sequence (Chen et al., 2013). Then, the target sequence for the sgRNA (5′-GATATCAGTCTGTTTCGTAA-3′) at the ttTi5605 site was inserted into pTK73 (pTK73_ ttTi5605).

For the transgene insertion by CRISPR/Cas9, the following injection mixture was injected into N2 worms: the homologous repair templates (pHK6_spd-5 fragments, 10 ng/μl), the sgRNA plasmid (pTK73_ ttTi5605, 50 ng/μl), the Cas9 expression plasmid pDD162 (Peft-3::Cas9, 50 ng/μl, Addgene #47549) (Dickinson et al., 2013) and the three injection markers pCFJ90 (Pmyo-2::mCherry, Addgene #19327), pCFJ104 (Pmyo-3::mCherry, Addgene #19328) and pGH8 (Prab-3::mCherry, Addgene #19359) (Frøkjaer-Jensen et al., 2008). Hygromycin B-resistant Roller worms that lacked mCherry fluorescence were selected as integrant candidates, and then were heat-shocked to remove the SEC. The integration and selection cassette excision were confirmed by sequencing.

Strain SA278 (mCherry::TBG-1; mCherry::Histone H2B) was constructed by microparticle bombardment (Toya et al., 2010). Dual-color strains [GFP::SPD-5(xx); mCherry::TBG-1; mCherry::Histone H2B] and [GFP::SPD-5(xx); mCherry::SAS-4]) were constructed by crossing.

For the depletion of endogenous SPD-5 by RNA interference (RNAi), we used the sequence corresponding to the spd-5 3′UTR or an N-terminal open reading frame (ORF) region (1−1566 bp). For pcmd-1 RNAi, we used the sequence corresponding to the ORF region (232−1079 bp). Template DNA fragments for synthesizing double-stranded RNAs (dsRNAs) were amplified by PCR. dsRNAs were synthesized in vitro with T7 RiboMAX Express Large Scale RNA Production System (Promega, Madison, USA) and then were purified by phenol–chloroform extraction. RNAi was carried out using the soaking method (Maeda et al., 2001). Briefly, 10 hermaphrodite early fourth larval stage (L4) worms were soaked in 2 mg/ml dsRNA solution and were incubated at 24.5°C for 24 h. Adult worms were then recovered from the dsRNA solution and incubated at 23°C on nematode growth medium (NGM) plates for 14−24 h before the worms were used for live imaging.

To permeabilize the eggshell for the in vivo PLK-1 inhibition assay, perm-1 RNAi was performed by feeding method. An E. coli strain HT115 that expresses perm-1 dsRNA was grown on NGM plates containing 1 mM isopropyl ß-D-1-thiogalactopyranoside (IPTG) for 24 h at 23°C, and L4 worms were fed for 19–21 h at 20°C.

Fluorescence images were obtained as described (Honda et al., 2017) with an Orca-R2 digital CCD camera (Hamamatsu Photonics, Shizuoka, Japan) mounted on an IX71 microscope (Olympus, Tokyo, Japan) using a CSU-X1 spinning disc confocal system (Yokogawa 1Electric Corporation, Tokyo, Japan). MetaMorph software (Molecular Devices, CA, USA) was used to control the microscopes. Live images of one-cell embryos were obtained using a UPlanSApo 60×/1.30 silicone oil objective lens (Olympus) with camera gain 255 and binning 2, with 40 s intervals, 15 z-slices at intervals of 1 μm and a 300 ms exposure for GFP and 1000 ms exposure for mCherry. Immunofluorescence images were obtained using a UPlanSApo 100×/1.40 oil objective lens (Olympus) with camera gain 0 and without binning; with 20 z-slices at intervals of 0.5 μm and 250 ms exposure for Alexa Fluor 488, 100 ms exposure for Alexa Fluor 568 and 1000 ms exposure for DAPI. Images of the pcmd-1(RNAi) embryos and the in vivo PLK-1 inhibition assay were obtained using an ORCA-Flash4.0 V3 digital CMOS camera (Hamamatsu Photonics) on the same microscope system as above with a UPlanSApo 60×/1.30 silicone oil objective lens (Olympus), 40 z-slices at intervals of 0.5 μm and a 300 ms exposure for GFP and 1000 ms exposure for mCherry without binning.

ImageJ/Fiji software (NIH; https://imagej.net/Fiji) was used for processing and analyzing the obtained images. The distribution and diameter of centrosomal GFP signals (Fig. 1C; Fig. 3E; Fig. 4D; Fig. S1C,D) were measured using the Plot Profile tool along the short axis of embryos, and the average cytosolic signal was subtracted as the background. The diameter was defined as the length that exceeded the common threshold value. The area of centrosomal GFP signal (Fig. 1D) was calculated using the threshold set by the Yen algorithm.

To generate antibodies against SPD-5 and PCMD-1, full-length recombinant SPD-5 or a PCMD-1 peptide (CDEGFDSSSLKNNPASLQRD; aa 6−24) was injected into rabbits as per standard protocols (MBL Life Science and Cosmo Bio, respectively).

For immunostaining, embryos were freeze cracked and fixed with −20°C methanol for 5 min and washed in 0.05% Tween 20 in phosphate-buffered saline (PBS-T). After being incubated for 30 min in PBS-T containing 0.1% bovine serum albumin (BSA) for blocking, the slides were incubated with rat anti-GFP (1:200, diluted with PBS-T containing 0.1% BSA; Nacalai Tesque, cat#04404-84, lot#M0P5692) and rabbit anti-SPD-5 (1:10,000; this study) or rabbit anti-PCMD-1 (1:10,000; this study) at 4°C overnight. Slides were washed with PBS-T and then were incubated with Alexa Fluor 488-conjugated donkey anti-rat IgG (1:500; Invitrogen, California, USA; RRID AB_2535794, lot# 2180272), Alexa Fluor 594-conjugated donkey anti-rabbit IgG (1:500; Invitrogen; RRID AB_141637, lot#52958A) and DAPI (1 mg/ml, 1:500; Dojinbo). After a 2 h incubation at room temperature, the slides were washed with PBS-T and mounted with ProLong Diamond (Thermo Fisher Scientific).

To score embryonic lethality, L4 worms were transferred to NGM plates with OP50 and incubated at 23°C for 24 h. Each 1-day-old adult was then transferred to a new plate and was allowed to lay eggs for 6 h before being removed. The plates with eggs were incubated at 23°C for 24 h, and the number of eggs and larvae was counted. The embryonic lethality was calculated as the number of eggs out of the total number of eggs and larvae.

PLK-1::6×His and 6×His::FLAG::SPD-5(C2) proteins used for the pull-down assay were obtained as follows: PLK-1::6×His was expressed in Sf9 insect cells and purified as described (Woodruff and Hyman, 2015). Briefly, PLK-1::6×His was expressed in Sf9 cells, and the cells were harvested 72 h after the infection of baculovirus generated by the Bac-to-Bac system (Invitrogen, cat#10359016). HisTALON Superflow Cartridge (Takara Bio, Shiga, Japan) was used for purification and the protein was eluted with 80 mM imidazole, and the buffer was then exchanged to imidazole-free buffer using PD-10 Columns (GE Healthcare). Proteins were concentrated using 30K Amicon Ultra Centrifugal Filters (Merck, Darmstadt, Germany).

6×His::FLAG::SPD-5(C2) was expressed in E. coli BL21-CodonPlus (DE3) (Agilent Technologies, CA, USA) using a cold-induced promoter (pCold, Takara Bio). Protein expression was induced by culturing for 18 h at 16°C with 0.5 mM IPTG. Flash-frozen bacterial pellets were dissolved in lysis buffer [20 mM HEPES, 1 mM DTT, 300 mM NaCl, 10% Glycerol, 1× cOmplete EDTA-free protease inhibitors (Roche, Basel, Switzerland) and PhosSTOP (Roche)] and sonicated (1 min, six times) before centrifugation (20,000 g, 30 min, 4°C). HisTrap HP (GE Healthcare) was used for purification and the protein was eluted with linear gradient from 25 to 500 mM imidazole, and the buffer was then exchanged to imidazole-free buffer and concentrated using 10K Amicon Ultra Centrifugal Filters (Merck).

GST, GST-tagged SPD-5(272−732_WT) or SPD-5(272−732_4A) bound to Glutathione Sepharose 4B beads were prepared as follows. The three proteins were expressed in E. coli in the same method as 6×His::FLAG::SPD-5(C2) described above. Bacterial pellets were dissolved in the binding buffer [PBS with 0.5 mM EDTA, 1 mM DTT, 1 mM PMSF, 26.3 μM MG-132 (Chemscene, New Jersey, USA), 2× cOmplete EDTA-free protease inhibitors (Roche), 1× PhosSTOP (Roche) and 1% TritonX-100], sonicated and centrifuged. The supernatant was filtered using a Millex-HV filter (0.45 μm; Merck) and incubated with Glutathione Sepharose 4B beads (GE Healthcare) for 2 h at 4°C. Beads were washed three times with the binding buffer.

Next, GST, GST-tagged SPD-5(272−732_WT) or SPD-5(272−732_4A) on Glutathione Sepharose 4B beads were treated by the purified PLK-1::6×His protein. The beads were incubated with or without 6.8 μM purified PLK-1::6×His in reaction buffer (25 mM HEPES, pH 7.4; 150 mM KCl; 0.5 mM DTT; 0.2 mM ATP; 10 mM MgCl₂; 0.025 mg/ml ovalbumin) (modified from Woodruff and Hyman, 2015). After a 60 min incubation at room temperature, the beads were washed three times with the C2-binding buffer (PBS with 150 mM NaCl, 0.5 mM EDTA, 1 mM DTT, 1 mM PMSF, 26.3 μM MG-132, 2× cOmplete EDTA-free protease inhibitors and 1% Triton X-100).

The beads with or without PLK-1 phosphorylation were incubated with 12.5 mM 6×His::FLAG::SPD-5(C2) at 4°C for 2 h. The beads were washed three times with the C2-binding buffer, then SDS-PAGE sample buffer (62.5 mM Tris-HCl, pH 6.8; 10% glycerol; 5% 2-mercaptoethanol; 2.5% SDS) was added. The samples were boiled at 98°C for 5 min and analyzed by western blotting.

C. elegans eggs were permeabilized by perm-1 RNAi, as described above. The worms were dissected in 43% egg buffer (25 mM HEPES pH 7.3, 118 mM NaCl, 48 mM KCl, 2 mM MgCl₂, 2 mM CaCl₂) (Toya et al., 2010) containing 1% DMSO or 10 μM PLK-1 inhibitor BI-2536 (diluted from 1 mM stock in DMSO; Selleckchem, TX, USA). Eggs at the stage of pronuclear appearance were observed 15 min after the mounting.

To check the expression level of transgenes, 15 one-day adult worms were collected in 10 μl of distilled water and flash-frozen in liquid nitrogen. The sample was then thawed and mixed with 3 μl of 4× SDS-PAGE sample buffer and 1.2 μl of 10% NP-40 and was boiled at 98°C for 5 min.

The samples were loaded onto a 5−20% gradient SDS-PAGE gel (Wako, Osaka, Japan) and transferred to an Immobilon-P Transfer Membrane (Merck). Antibodies were diluted with Signal Enhancer HIKARI for western blotting and enzyme-linked immunosorbent assays (ELISA; Nacalai Tesque). The primary antibodies rabbit anti-SPD-5 (1:2000) and rat anti-GFP (1:10,000) were generated against full-length recombinant proteins (MBL Life Science). Additional primary antibodies were rabbit anti-FLAG (1:2000; Sigma-Aldrich, MO, USA; cat#F7425, lot#078M4886V), mouse anti-GST B-14 (1:2000; Santa Cruz Biotechnology, TX, USA; cat#sc-138, lot#l1009) and mouse anti-α-tubulin DM1A (1:2000; Sigma-Aldrich; cat#T9026, lot#029M4880V). Secondary antibodies were horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (H+L) (1:50,000; Jackson ImmunoResearch Laboratories, Pennsylvania, USA; RRID AB_2338511, lot#93599), HRP-conjugated donkey anti-rabbit IgG (H+L) (1:50,000; Jackson ImmunoResearch Laboratories; RRID AB_10015282, lot#95427) and HRP-conjugated donkey anti-rat IgG (H+L) (1:50,000; Jackson ImmunoResearch Laboratories; RRID AB_2340638, lot#96323). Signals were detected using Chemi-Lumi One Ultra (Nacalai Tesque), and blot images were obtained with a ChemiDoc Touch MP (Bio-Rad, CA, USA). Images of full blots are shown in Fig. S4.

The ProQuest Two-Hybrid System (Invitrogen) was used for the yeast two-hybrid assays. Protein coding sequences were integrated into the plasmids pDEST22 and pDEST32 (Invitrogen, cat#PQ1000101), downstream of the GAL4 DNA-activation domain and DNA-binding domain, respectively. Yeast strain Mav203 (Invitrogen, cat#PQ1000101) was treated with Frozen-EZ Yeast Transformation II (Zymo Research, CA, USA) to generate competent cells and was then transformed with bait and prey plasmids. The transformants were selected by culturing on SC-Trp-Leu plates for 3 days at 30°C. Each yeast strain was cultured in 2 ml of liquid SC-Trp-Leu for 13 h at 30°C and was analyzed on selection plates containing 3-amino-1,2,4-triazole (3AT).

For the yeast two-hybrid screen, we first transformed yeast Mav203 with bait plasmids (the coding sequence of SPD-5 fragments or PCMD-1 subcloned into pDEST32) only and incubated the cells on SC-Trp plates. The competent yeast cells with a single bait vector were then used in a transformation with the prey vector (cDNA fragments cloned into a pPC86 plasmid, Invitrogen) for yeast two-hybrid screening. Transformation efficiency was measured based on the number of colonies on SC-Trp-Leu plates. For the screening, transformants were spread onto 20 SC-Trp-Leu+25 mM 3AT selection plates and incubated for 5−7 days at 30°C. Individual colonies were then picked and spread on new SC-Trp-Leu plates. As candidate interactors with SPD-5(C1), 21 colonies were obtained from 1.9×10⁵ cells on selection plates. As candidate interactors with PCMD-1, 12 colonies were obtained from 6.6×10⁴ cells on selection plates. The coding sequences of the candidate clones were PCR amplified and sequenced (Forward primer: 5′-TATAACGCGTTTGGAATCACT-3′, Reverse primer: 5′-AGCCGACAACCTTGATTGGAGAC-3′), and their identities were determined by a WormBase BLAST/BLAT search (https://wormbase.org/tools/blast_blat). For all positive prey plasmids, interactions were confirmed by isolation from the positive colonies and re-transformation to Mav203 with the bait vector.

The details for quantification, sample numbers and statistical analyses are described in the main text, the figures and the figure legends. The values given for n represent the number of embryos imaged to determine phenotypes (Fig. 1A–D, Fig. 2A, Fig. 3D,E; Fig. S1C,D), the number of embryos counted to measure embryonic lethality (Fig. 2B) or the number of centrosomes imaged to determine phenotypes (Figs 3C, Fig. 4C,D).

GraphPad Prism 9 software was used to analyze and represent the data. Results are expressed as the mean value±s.d. (Fig. 1B–D, Fig. 3E, Fig. 4D) or the mean value±s.e.m. (Fig. 2B). One-way ANOVA and Tukey's multiple comparisons test were performed to determine significant differences (Fig. 1D, Fig. 2B). Significance levels: not significant, n.s., P≥0.05; *P<0.05; **P<0.01; ***P<0.001.

We thank Yuji Kohara (National Institute of Genetics, Mishima, Japan) for providing cDNA clones and the members of the Sugimoto laboratory for discussions.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.259025.