PCMD-1 bridges the centrioles and the pericentriolar material scaffold in C. elegans

Centrosomes are dynamic, non-membranous organelles that serve as the major microtubule-organizing centers in animal cells and are thus essential for biological processes ranging from polarity establishment to the orchestration of cell division. Centrosomes comprise a centriole pair and the surrounding pericentriolar material (PCM). The size and material properties of the PCM change dynamically during the cell cycle (Woodruff et al., 2015, 2017; Mittasch et al., 2020).

PCM expansion during mitosis facilitates bipolar spindle assembly. At the root of PCM expansion is a proteinaceous matrix that serves as a scaffold for the recruitment of regulatory proteins, including mitotic kinases and microtubule nucleators. In Caenorhabditis elegans, this scaffolding function is fulfilled by self-assembly of the coiled-coil protein SPD-5 (functional homolog of human CDK5RAP2), which is controlled by phosphorylation through Polo-like kinase PLK-1 (homolog of human PLK1) and interaction with the coiled-coil protein SPD-2 (homolog of human CEP192) (Hamill et al., 2002; Decker et al., 2011; Woodruff et al., 2015, 2017; Cabral et al., 2019). Our previous findings have revealed that PCMD-1, a protein with a predicted coiled-coil domain, regulates the spatial integrity of the PCM scaffold, and, together with SPD-2, is required for the recruitment of SPD-5 (Erpf et al., 2019). Centrioles serve as condensation centers for PCM proteins. During PCM expansion in mitosis, centrioles contribute to the growth and structural integrity of the PCM scaffold (Cabral et al., 2019). A limited set of centriolar core proteins has been described in C. elegans (O'Connell et al., 2001; Kirkham et al., 2003; Leidel and Gönczy, 2003; Dammermann et al., 2004; Delattre et al., 2004, 2006; Kemp et al., 2004; Pelletier et al., 2004, 2006; Leidel et al., 2005; Dammermann et al., 2008; Sugioka et al., 2017). From these proteins, SPD-2, SAS-4 (CPAP homolog) and SAS-7 have been proposed to functionally bridge the PCM and the centrioles (Varadarajan and Rusan, 2018). SAS-4, which localizes to the centrioles and the PCM, plays a crucial role in microtubule assembly around the central tube of a forming centriole (Kirkham et al., 2003; Leidel and Gönczy, 2003; Dammermann et al., 2008; Delattre et al., 2006; Pelletier et al., 2006). SAS-7 facilitates the formation of the paddlewheel structure on centriolar microtubules and recruits SPD-2, which in turn is needed for centriole duplication and mitotic PCM scaffold expansion (Sugioka et al., 2017).

PCMD-1 is predominantly a centriolar protein, yet its depletion affects SPD-5 recruitment and the structural integrity of the PCM (Erpf et al., 2019), raising the possibility that it connects the PCM scaffold functionally to the centrioles. However, the precise mechanisms of PCMD-1 centriolar targeting and how PCMD-1 recruits PCM components is still to be elucidated. Here, we investigate the function of PCMD-1 in functionally bridging centrioles and the PCM in C. elegans embryos.

PCMD-1 localizes weakly to the PCM and strongly to centrioles. PCMD-1 does not require the PCM scaffold protein SPD-5 for its localization in one-cell embryos (Erpf et al., 2019). This raises the questions of whether PCMD-1 is recruited to the centrosome via outer centriolar proteins and whether PCMD-1, in turn, has a role in maintaining these proteins at centrioles. One candidate for such interaction is SAS-7, which is needed for the formation of paddlewheels, the outermost centriolar structures known in C. elegans (Sugioka et al., 2017). We investigated the spatial relationship between PCMD-1 and SAS-7 by analyzing embryos expressing endogenously tagged GFP::PCMD-1 and RFP::SAS-7 using lattice structured illumination microscopy (SIM) (Fig. 1A). We found that PCMD-1 and SAS-7 signals largely overlapped on both centrioles at spindle poles of mitotic blastomeres in early embryos (Fig. 1A). This observation prompted us to test whether SAS-7 and PCMD-1 localization are interdependent. Live-cell imaging of GFP::SAS-7 in pcmd-1(t3421) mutant embryos revealed that GFP::SAS-7 levels at centrioles are comparable to that of control embryos (Fig. 1B,C). We concluded that PCMD-1 is not involved in the recruitment of SAS-7 to the centrioles. To test inversely whether SAS-7 is required for PCMD-1 recruitment to the centrosome, we crossed an in-locus-tagged gfp::pcmd-1 with sas-7(or452) mutant animals. Whereas GFP::PCMD-1 at the centrosome was apparent in all control embryos, the centrosomal GFP::PCMD-1 signal was significantly reduced in sas-7(or452) one-cell embryos (Fig. 1D,E). Interestingly, a GFP::PCMD-1 signal was consistently detected at sperm-derived sas-7(or452) centrioles during pronuclear migration (Fig. S1A). Shortly thereafter, this signal decreased close to the detection limit (Fig. 1E, Fig. S1A). As an alternative means to address the SAS-7-dependent localization of PCMD-1, we performed immunostaining using antibodies against GFP and SAS-4 to mark the centrioles. In one-cell control embryos, all centrosomal SAS-4 foci colocalized with a clear GFP::PCMD-1 signal (Fig. 1F). In contrast, only a very weak GFP::PCMD-1 signal was observed at SAS-4 foci in sas-7(or452) embryos (Fig. 1F), probably reflecting the hypomorphic nature of the sas-7(or452) allele (Sugioka et al., 2017). Note that centrosomal GFP::PCMD-1 signal was detectable in sas-7(or452) multicellular embryos (Fig. S1B). In these late embryos, PCMD-1 could be recruited through a SAS-7-independent mechanism, as is the case at the ciliary base (Garbrecht et al., 2021; Magescas et al., 2021). Thus, SAS-7 is necessary to maintain PCMD-1 at the centrosome during the first cell division.

The genetic dependency of PCMD-1 centrosomal localization on SAS-7 raises the possibility that PCMD-1 is recruited to the centriole by direct interaction with SAS-7. To determine whether PCMD-1 and SAS-7 interact with each other and with other centrosomal proteins such as SAS-4, SPD-2, SPD-5 and PLK-1, we performed a candidate-based yeast two-hybrid screen (Fields and Song, 1989). We generated bait plasmids containing the cDNAs of SAS-7 and PCMD-1 and prey plasmids for the candidate proteins SAS-4, SAS-7, PCMD-1, SPD-2, SPD-5 and PLK-1. The readouts of positive interactions were growth and the expression of a GFP reporter. We categorized proteins showing both readouts on day 3 as strong interactors and on day 5 as weak interactors.

Previous yeast two-hybrid screens identified SAS-7 as a binding partner of SPD-2 (Sugioka et al., 2017; Boxem et al., 2008; Li et al., 2004) and SAS-4 (Boxem et al., 2008). We used these interactions to validate our yeast two-hybrid assay. As previously reported, we found that SAS-7 interacts with SPD-2 and SAS-4 (Fig. 2A,B). However, we could not detect an interaction between SAS-7 and PCMD-1 using either protein as bait or prey. Therefore, we tested whether PCMD-1 could interact with the centriolar proteins SAS-4 or SPD-2. We observed strong interaction of PCMD-1 with SAS-4 but not with SPD-2. Next, we tested whether PCMD-1 could bind to the PCM scaffold protein SPD-5, the kinase PLK-1 or itself. We detected strong interaction of PCMD-1 with itself and a weaker interaction with SPD-5 and PLK-1. In turn, SAS-7 did not interact with SPD-5 or with PLK-1 (Fig. 2A,B).

To confirm the interactions between the PCMD-1 bait and SAS-4, PLK-1 and SPD-5 preys, we reversed bait and prey. Again, we observed that PLK-1 and SAS-4 interacted with PCMD-1 (Fig. 2C). Because SPD-5 bait autoactivated, we turned to an alternative assay to confirm this interaction. We expressed C. elegans PCMD-1 tagged with EGFP (EGFP::PCMD-1) and SPD-5 tagged with mCherry and 6xHis (mCherry::SPD-5::6xHis) in human HEK293T cells and performed a co-immunoprecipitation. We successfully co-immunoprecipitated EGFP::PCMD-1 with SPD-5, confirming our yeast two-hybrid observation (Fig. 2D).

The yeast two-hybrid assay revealed no interaction of PCMD-1 with SAS-7 or SPD-2. However, both PCMD-1 and SAS-7 interact with SAS-4. Thus, rather than direct recruitment of PCMD-1 by SPD-2 or SAS-7, SAS-4 could theoretically act as a linker between PCMD-1 and SAS-7. However, in accordance with previous studies we found that in sas-7(or452) mutant embryos, where no PCMD-1 is found at the centrosome, SAS-4 foci are still present (Fig. 1F) (Sugioka et al., 2017). Therefore, it is highly unlikely that SAS-4 is accountable for the loss of PCMD-1 centrosomal localization in sas-7(or452) embryos.

In summary, PCMD-1 interacts with the centriolar protein SAS-4, the PCM protein SPD-5, the mitotic kinase PLK-1, and with itself. We propose that PCMD-1 acts as a functional bridge between the centrioles and the PCM scaffold.

PCMD-1 is required to recruit SPD-5 to the centrosome (Erpf et al., 2019). This finding is strongly supported by our yeast-two hybrid interaction data, which indicates that PCMD-1 anchors the PCM scaffold to the centriole. Therefore, we investigated whether PCMD-1 is also able to recruit SPD-5 to an ectopic location. To address this, we established a ‘translocation assay’ by targeting PCMD-1 to an ectopic location in the cell and testing whether PCMD-1 is capable of recruiting SPD-5 to this cellular location. To tether PCMD-1 to the plasma membrane, we fused the mkate2::PCMD-1 reporter to the plcδ1PH-domain and expressed it under a heat-shock promoter. Upon heat shock, PH::mkate2::PCMD-1 was expressed and reliably translocated to the plasma membrane (Fig. 3A,B).

We tested whether membrane-bound PH::mkate2::PCMD-1 could recruit GFP::SPD-5 in the pcmd-1(t3421) mutant, which has a premature stop codon, and in the presence of wild-type PCMD-1. In pcmd-1(t3421) animals, recruitment of SPD-5 to the centrosome is compromised owing to the lack of endogenous PCMD-1 (Erpf et al., 2019). Therefore, more GFP::SPD-5 is expected in the cytoplasm. In control embryos without heat shock, GFP::SPD-5 was never detected at the plasma membrane. Induction of PH::mkate2::PCMD-1 by heat shock resulted in GFP::SPD-5 localization to the plasma membrane in 95.2% of the pcmd-1(t3421) and 68.4% of the wild-type embryos (Fig. 3B,C, Fig. S2A,B).

Because PLK-1 is also a PCM component and interacts with PCMD-1 in the yeast two-hybrid assay, we tested whether the in-locus-tagged PLK-1::sGFP translocates to the plasma membrane in a PCMD-1-dependent manner. Similar to GFP::SPD-5, PLK-1::sGFP localized to the plasma membrane in 41.4% of pcmd-1(t3421) and 26.3% of wild-type embryos after heat shock but did not localize to the membrane without heat shock (Fig. 3B,C, Fig. S2A,B). Note that the PLK-1::sGFP signal at the membrane was much weaker compared with the GFP::SPD-5 signal.

SPD-5 phosphorylation by PLK-1 is essential for PCM maturation (Woodruff et al., 2015). In the absence of PLK-1 phosphorylation at four specific residues, SPD-5 only forms a centrosome core but fails to assemble the mitotic scaffold. Therefore, we tested whether the translocation of SPD-5 to the membrane requires phosphorylation by PLK-1 at these residues. For this, we used a strain carrying RNAi-resistant GFP::SPD-5(4A), in which the four PLK-1 phosphorylation sites were substituted by alanine residues (Woodruff et al., 2015). To eliminate the endogenous SPD-5, embryos were treated with RNAi against spd-5. We found that GFP::SPD-5(4A) was still able to translocate efficiently to the membrane (96.8%; Fig. 3D,E), indicating that the membrane-bound GFP::SPD-5 pool does not resemble the mitotic PCM scaffold of SPD-5. Membrane translocation of PLK-1 could be mediated through SPD-5. To test this possibility, we performed the experiment in the spd-5(RNAi) background. PLK-1::sGFP recruitment still took place (45.9%; Fig. 3D,E). Thus, PCMD-1 can recruit PLK-1 independently of SPD-5.

Next, we tested whether PCMD-1 can also induce translocation of centriolar proteins. Interestingly, PH::mkate2::PCMD-1 was unable to translocate GFP::SAS-4, GFP::SAS-7 or GFP::PCMD-1 in any of the analyzed pcmd-1(t3421) embryos (Fig. 3B,C). This was unexpected, especially for SAS-4, because the interaction in the yeast two-hybrid assay was very strong. One explanation could be that a centriole tethering of SAS-4 prevents membrane translocation by PCMD-1. To test this hypothesis, we used a GFP::SAS-4 construct that lacks the conserved T complex protein 10 (TCP) domain, needed for its binding to SAS-5 (Cottee et al., 2013). When endogenous SAS-4 is eliminated by RNAi, the RNAi resistant GFP::SAS-4(ΔTCP) protein is not tethered to the centrioles but still localizes to the PCM (Cottee et al., 2013) (Fig. S2C). In the translocation assay, PCMD-1 was unable to recruit GFP::SAS-4(ΔTCP) to the membrane (Fig. S2C,D), raising the possibility that the interaction between SAS-4 and PCMD-1 is not through the PCM pool and may require the local environment at the centrioles. Therefore, the ability of PCMD-1 to ectopically anchor proteins to the plasma membrane is specific to the PCM scaffold proteins SPD-5 and PLK-1. Alternatively, the deletion of the TCP domain could compromise the interaction between SAS-4 and PCMD-1.

Our findings suggest that PCMD-1 could recruit the initial PCM core to the PCM-less sperm centrioles in the C. elegans one-cell embryo. Therefore, we investigated whether PCMD-1 is loaded onto the sperm centrosomes before SPD-5. In the C. elegans zygote, SPD-5 is recruited to the sperm-derived centrioles after the completion of meiosis II of the female pronucleus and concomitant with the ability of the centrosome to nucleate microtubules (McNally et al., 2012). This paradigm allows us to investigate whether PCMD-1 is loaded to the centrioles prior to SPD-5 recruitment using marked mating experiments in which only the sperm or the oocyte expresses a fluorescent marker. First, we tested whether paternal GFP::PCMD-1 could be detected at the centrosome after fertilization. For this, we mated fog-2(q71) females lacking sperms with GFP::PCMD-1-expressing males, thus labeling sperm centrioles (Fig. S3A) (Erpf et al., 2019). We could not detect GFP::PCMD-1 at centrioles in any of the analyzed embryos after the completion of meiosis II. Second, we mated GFP::PCMD-1 females, treated with fem-1(RNAi) to block sperm production, with control fog-2(n71) males with unlabeled sperm centrioles. We found that the GFP::PCMD-1 signal was detected at the centrosomes in all analyzed embryos during the first mitotic division (Fig. S3C). These results suggest that sperm-derived PCMD-1 is not maintained and that the maternal PCMD-1 is recruited to the centrioles after fertilization.

To determine exactly when maternal PCMD-1 is recruited to the centriole after fertilization and to temporally map its loading with respect to SPD-5, we immunostained embryos from GFP::PCMD-1 females mated with males with unlabeled sperm centrioles (Fig. S3C,D), using antibodies against SPD-5 and GFP. In meiosis I embryos, neither GFP::PCMD-1 nor SPD-5 foci were present at the centrioles (Fig. S3E), indicating that maternal GFP::PCMD-1 was not yet incorporated in the centrioles. During meiosis II, we found that a GFP::PCMD-1 focus, without any detectable SPD-5, was visible at 90.5% of sperm centrioles (Fig. S3D,E). In the remaining 9.5% of embryos, categorized as early meiosis II, neither GFP::PCMD-1 nor SPD-5 foci were present (Fig. S3D). Therefore, we conclude that maternal GFP::PCMD-1 is recruited to the sperm centrioles at meiosis II. After meiosis II, when sperm pronuclei are decondensed, SPD-5 colocalized with GFP::PCMD-1 at the centrosomes in 87% of the embryos (Fig. S3D,E). We never observed embryos with centrosomes only labeled by SPD-5.

In summary, our results are consistent with a model in which maternal GFP::PCMD-1 is recruited to the sperm-derived centrioles shortly after fertilization and subsequently recruits SPD-5 and PLK-1 to form the centrosome core. These findings strengthen the hypothesis that PCMD-1 bridges centriolar and PCM proteins.

To determine how PCMD-1 is anchored to the centrosome, we next examined which part of the protein is necessary for its centrosomal targeting. PCMD-1 is predicted to have a single coiled-coil domain and six intrinsically disordered regions (IDRs 1-6), which partially overlap with low complexity regions (Fig. 4A) (UniProt Consortium, 2019; Schultz et al., 2000; Letunic and Bork, 2018). Coiled-coil domains often mediate protein-protein interactions, including oligomerization, and these interactions can have regulatory functions for centrosomal proteins (Leidel et al., 2005; Kitagawa et al., 2011; Qiao et al., 2012; Hilbert et al., 2013; Lettman et al., 2013; Rogala et al., 2015). Therefore, we set out to investigate the function of the coiled-coil domain in PCMD-1.

To examine the function of the coiled-coil domain, we deleted the sequence from E86 up to and including F118, predicted as the coiled-coil domain by the COILS program (Lupas et al., 1991; Lupas, 1996), using CRISPR/Cas9 in the in-locus-tagged GFP::PCMD-1 protein. We refer to this deletion as gfp::pcmd-1(ΔCC) (Fig. 4A). In a lethality test, 97.5% of the gfp::pcmd-1 embryos and 87.6% of the gfp::pcmd-1(ΔCC) embryos survived at 25°C (Fig. 4B). Thus, the deletion of the predicted coiled-coil domain compromised viability.

Next, we investigated whether the GFP::PCMD-1(ΔCC) protein could localize to the centrosome. We performed live-cell imaging on worms expressing GFP::PCMD-1 and GFP::PCMD-1(ΔCC). Whereas GFP::PCMD-1 efficiently localized to the centrosomes in all analyzed embryos, the GFP::PCMD-1(ΔCC) signal on average appeared much weaker (Fig. 4C, Movies 1 and 2). Measuring the mean centrosomal GFP signal intensity confirmed that PCMD-1 without the coiled-coil domain was significantly reduced at the centrosomes in comparison with wild-type PCMD-1, whereas cytoplasmic levels remained unchanged (Fig. 4D,E). Thus, the coiled-coil domain is necessary for efficient centrosomal loading of PCMD-1 but is not essential for the viability of the embryos. Given that coiled-coil domains are often implicated in the oligomerization of centrosomal proteins, we asked whether PCMD-1 self-interaction was compromised in the absence of the coiled-coil domain. To this end, we expressed PCMD-1(ΔCC) as a bait plasmid and probed its interaction with the PCMD-1(ΔCC) prey (Fig. 4F). In the absence of the coiled-coil domain, PCMD-1 self-interaction was lost (Fig. 4F).

This raises the question of how embryos with reduced centrosomal PCMD-1 levels can divide. To investigate whether these animals could still recruit the PCM scaffold, we immunostained GFP::PCMD-1(ΔCC) embryos using antibodies against GFP and SPD-5 and performed live-cell imaging of RFP::SPD-5. We found that SPD-5 was still recruited to the centrosome in all analyzed embryos, even in embryos in which GFP::PCMD-1(ΔCC) was almost undetectable (Fig. 4G, Fig. S4A). Although overall centrosomal RFP::SPD-5 levels remained similar to the control embryos in the absence of the coiled-coil domain (Fig. S4A,B), the SPD-5 centrosome matrix appeared to be much more dispersed and disorganized (Fig. 4G). To quantify the degree of disorganization, we measured centrosome circularity using immunofluorescence staining and found that mean centrosome circularity values were significantly lower in gfp::pcmd-1(ΔCC) embryos (Fig. 4H).

We propose that the coiled-coil domain facilitates PCMD-1 self-interaction and thereby promotes efficient PCMD-1 accumulation at the centrosome and the maintenance of PCM scaffold integrity.

The fact that GFP::PCMD-1(ΔCC) could still be recruited to the centrosome indicates that protein regions other than the coiled-coil domain might play a role in centrosomal anchoring. To map which part of the protein is involved, we used a previously established single-copy replacement system (Erpf et al., 2019). pcmd-1(t3421) mutant animals carry a stop codon before the coiled-coil domain and are lethal at 25°C (Fig. S5A). Reconstituting a single copy of the PCMD-1 cDNA under the regulatory elements of the mai-2 gene rescues survival rates to 96.2% (Fig. S5A). We used this assay to test the functionality and localization of different truncations of the PCMD-1 protein. GFP::PCMD-1(N) spans the region E2-N117, including the first two IDRs and the coiled-coil domain. GFP::PCMD-1(C) comprises amino acids F118 to the stop codon, spanning the remaining IDRs (Fig. 5A). In the survival assay, GFP::PCMD-1(C) could rescue the lethality of pcmd-1(t3421) to 90.8% survival, whereas GFP::PCMD-1(N) was not sufficient to rescue viability (0.3%) (Fig. S5A). Interestingly, GFP::PCMD-1(N) even had a dominant-negative effect on the viability of pcmd-1(t3421) at the permissive temperature of 15°C, reducing it from 41% to 28% (Fig. S5B).

Next, we assessed the ability of these constructs to localize to the centrosome by live-cell imaging. Centrosomal GFP signal was detected in all GFP::PCMD-1(C) embryos, albeit with slightly reduced GFP signal intensities compared with control animals (Fig. 5B,C). In contrast, we could not detect any GFP signal at the centrosome in embryos expressing the GFP::PCMD-1(N) constructs, even though the cytoplasmic levels were much higher (Fig. 5B, Fig. S5C). Therefore, the C-terminal part of PCMD-1, excluding the coiled-coil domain and IDR1 and IDR2, is sufficient for PCMD-1 anchoring to the centrosome.

To further map the part of PCMD-1 that targets the protein to the centrosome, we subdivided the C-terminal part into two fragments spanning F118-D342 (C1) and G343-stop codon (C2) (Fig. 5A). In the survival assay, neither GFP::PCMD-1(C1) nor GFP::PCMD-1(C2) rescued the lethality of pcmd-1(t3421) (Fig. S5A). However, GFP::PCMD-1(C2) still localized to the centrosome, whereas GFP::PCMD-1(C1) did not (Fig. 5B,C). This raises the possibility that the C2 part of the PCMD-1 could be interacting with SAS-4. Therefore, we expressed the PCMD-1(C2) as a bait together with SAS-4 as a prey in the yeast two-hybrid assay (Fig. 5D). The PCMD-1(C2) bait strongly interacted with the SAS-4 prey. A similarly strong interaction was detected between the SAS-4 bait and the PCMD-1(C2) prey (Fig. 5D).

Similar to the centrosomes in the embryo, only GFP::PCMD-1, GFP::PCMD-1(C) and GFP::PCMD-1(C2) localized to the ciliary base of adult animals (Fig. S6). In particular, signals from GFP::PCMD-1(C) and GFP::PCMD-1(C2) appeared stronger than the control, extending along the cilia (Fig. S6).

In pcmd-1(t3421) mutant animals, SPD-5 recruitment to the centrosome is severely compromised (Erpf et al., 2019). Therefore, we tested whether the truncated parts of PCMD-1 that localize to the centrosome can restore SPD-5 recruitment. In animals expressing GFP::PCMD-1(C), levels of RFP::SPD-5 at metaphase were similar to those in animals carrying the full-length GFP::PCMD-1. However, in GFP::PCMD-1(C2)-expressing animals, RFP::SPD-5 levels were largely reduced (Fig. S5C,D). Thus, the C2 part of PCMD-1, including IDR6, is sufficient to target PCMD-1 to the centrosome but is insufficient to recruit the PCM scaffold composed of SPD-5.

The fact that PCMD-1(C2) strongly localizes to the centrosome prompted us to test a construct that lacks the C2 part (GFP::PCMD-1(ΔC2) E2-D342). Surprisingly, GFP::PCMD-1(ΔC2) could still localize to the centrosome, although the centrosomal levels were reduced by 60% in comparison with the full-length construct (Fig. 5B,C). However, GFP::PCMD-1(ΔC2) fully rescued the viability of pcmd-1(t3421) and recruited RFP::SPD-5 to the centrosome (Fig. S5A,D,E).

Intriguingly, we could not detect GFP::PCMD-1(ΔC2) at the cilia, indicating that the C2 part is absolutely essential for ciliary base targeting (Fig. S6). Given that the C2 part contains IDR6, we generated a worm strain that expresses GFP::PCMD-1(ΔC2.2) E2-G514, a construct in which IDR6 is included. Centrosomal GFP::PCMD-1(ΔC2.2) levels were comparable to GFP::PCMD-1(ΔC2); however, the ciliary targeting was restored (Fig. S6).

In summary, multiple parts of PCMD-1 contribute to the centrosomal localization of PCMD-1. The PCMD-1(C2) region can bind SAS-4 and is sufficient for anchoring to the centrosome. The N-terminal construct lacking C2 can still localize to the centrosome, suggesting the presence of a second centrosome targeting region. The centrosomal PCMD-1(C2) is insufficient to recruit SPD-5 and for embryonic development. By contrast, the N-terminal construct lacking the C2 localizes to the centrosome at largely reduced levels and can restore centrosomal SPD-5 levels and function. The C2 part, and specifically the amino acids between D342 and G514 encompassing IDR6, are necessary for cilia localization.

In this study, we examine the mechanism by which PCMD-1 anchors the PCM scaffold to the centriole. We demonstrate that PCMD-1 interacts with the PCM scaffold protein SPD-5, the mitotic kinase PKL-1 and the centriolar protein SAS-4. Furthermore, tethering PCMD-1 at an ectopic cellular location is sufficient to recruit SPD-5 and PKL-1. In turn, the centrosomal localization of PCMD-1 depends on the outer centriolar protein SAS-7. Together with previous findings that PCMD-1 is required for SPD-5 recruitment to the PCM core in the one-cell embryo, these findings place the function of PCMD-1 between the centrioles and the PCM.

Our analysis revealed that PCMD-1 interacts with SPD-5. A similar protein-protein interaction has been established for pericentrin/Plp, the putative homolog of PCMD-1, and CDK5RAP2/Cnn the functional homolog of SPD-5 in vertebrates and Drosophila, respectively (Buchman et al., 2010; Lerit et al., 2015; Galletta et al., 2016). This interaction is especially important in C. elegans for the initial formation of the PCM core in the one-cell embryo. Paternally contributed centrioles in the one-cell embryo are deficient of the PCM core. SPD-5, which renders the centrioles’ microtubule-organizing activity, is recruited from the maternal pool after female meiosis (McNally et al., 2012). Our findings that PCMD-1 is associated with centrioles prior to SPD-5 recruitment and that tethering of PCMD-1 at the plasma membrane is sufficient to recruit SPD-5 suggest that PCMD-1 is needed for SPD-5 accumulation and the formation of the PCM core after fertilization. In the translocation assay, PCMD-1 was consistently more effective at recruiting SPD-5 and PLK-1 to the plasma membrane in the absence of an endogenous PCMD-1. In the pcmd-1(t3421) mutant background, SPD-5 and PLK-1 are not efficiently recruited to the centrosome and are expected to be more abundant in the cytoplasm. This finding suggests a ‘tug-of-war’ between the centrosomal and membrane-bound PCMD-1 pools for the recruitment of SPD-5 and PLK-1. Membrane-bound PCMD-1 can also efficiently recruit a phospho-deficient version of SPD-5 in which the four residues that are phosphorylated by PLK-1 and that play a key role in the expansion and maturation of the mitotic PCM scaffold are mutated (Woodruff et al., 2015). Therefore, we speculate that PLK-1 phosphorylation of these residues is not required for PCMD-1 to recruit SPD-5 and that the membrane-targeted SPD-5 is analogous to the PCM core. The fact that PLK-1 can be translocated to the membrane in the absence of SPD-5 confirms our previous findings that PCMD-1 contributes to the formation of the PLK-1 pool at the centrosome independently of SPD-5 (Erpf et al., 2019). By bringing together PLK-1 and SPD-5, PCMD-1 facilitates the initiation of the centrosome maturation process.

In contrast to the PCM proteins, PCMD-1 was unable to relocate itself and the centriolar proteins SAS-4 and SAS-7 to the plasma membrane. This is consistent with observations that the centriolar localization of both SAS-4 and SAS-7 is independent of PCMD-1 (this study; Erpf et al., 2019). At the same time, this is puzzling especially in the case of PCMD-1 and SAS-4, for which we found a strong protein-protein interaction in the yeast two-hybrid system. Assays such as the yeast two-hybrid system do not indicate exactly where and when these interactions occur. In the case of PCMD-1, this could be due to a stronger binding of PCMD-1 to other centrosomal proteins, which prevent delocalization.

The fact that PCMD-1 could not recruit SAS-4(ΔTCP) raises the possibility that the PCMD-1/SAS-4 interaction may need a specific local environment or modification, which is provided at the centrioles but is absent at the cytoplasmic pool of the proteins. As the TCP domain of SAS-4 associates with SAS-5 and the centrioles, SAS-4(ΔTCP) only represents the PCM fraction of SAS-4 (Cottee et al., 2013). Even though we cannot exclude the formal possibility that the deletion of the TCP domain additionally disrupts the interaction with PCMD-1, we consider it highly unlikely. We favor the possibility that either another protein bridges SAS-4 and PCMD-1 or that the proteins are post-translationally modified in the yeast two-hybrid system and at the centrosome, but not on the plasma membrane.

The outer centriolar protein SAS-7 is genetically upstream and required for PCMD-1 centrosomal recruitment in the one-cell embryo. This interaction could be direct or mediated through the known SAS-7 binding partners SPD-2 and SAS-4 (Sugioka et al., 2017). It has been shown previously that SAS-7 recruits SPD-2 to the centrosome, and that SAS-7 and PCMD-1 do not rely on a functional SPD-2 for their centrosomal localization (Sugioka et al., 2017; Erpf et al., 2019). In the yeast two-hybrid assay, PCMD-1 did not interact with SAS-7 or SPD-2. Thus, SPD-2 is an unlikely mediator of this interaction. As both PCMD-1 and SAS-7 interact with SAS-4 in the yeast two-hybrid assay, one could assume that the interaction is mediated through SAS-4. The absence of a fully formed daughter centriole in sas-4(RNAi) embryos does not allow us to make a conclusion about the requirement of SAS-4 for PCMD-1 localization. However, SAS-4 foci are still present in sas-7(or452) mutant embryos (Sugioka et al., 2017), in which little or no PCMD-1 is found at the centrosome. Although we cannot exclude the formal possibility that in sas-7(or452) mutant embryos SAS-4 changes its conformation or lacks a post-translational modification, which would hinder the interaction with PCMD-1, we do not favor a model in which SAS-4 is the only anchor for PCMD-1 on the centriole. We suggest that the interaction between SAS-7 and PCMD-1 is either indirect, through an unknown protein or an additional co-factor, or that a modification is required for interaction in the yeast two-hybrid and translocation assay. Structurally intact paddlewheels, which are altered in sas-7(or452) mutant animals, might be a prerequisite for PCMD-1 anchoring (Sugioka et al., 2017).

The presence of multiple IDRs in PCMD-1 suggests high conformational plasticity of the protein (Uversky, 2019). Our structure-function analysis revealed that multiple regions of PCMD-1 contribute to centrosomal targeting. The PCMD-1(C2) fragment, including IDR6, is sufficient for the centrosomal localization of PCMD-1. In the yeast two-hybrid assay, the same fragment strongly interacts with SAS-4, suggesting an interaction site with the centrioles (Fig. 6A). At the same time, centrosomal SPD-5 levels in the PCMD-1(C2) background are highly reduced and are comparable to the pcmd-1(t3421) mutants (Erpf et al., 2019). Thus, the PCMD-1(C2) fragment alone is insufficient to recruit SPD-5 and as a result does not support embryonic development. Therefore, we suggest that N-terminal parts of the protein are needed for SPD-5 recruitment and function (Fig. 6A). In accordance with this, constructs lacking the C-terminal region [PCMD-1(ΔC2) and PCMD-1(ΔC2.2)], which localize to the centrosome at much-reduced levels, are sufficient to accumulate SPD-5 and sustain viability.

The coiled-coil domain and IDR1 and IDR2 alone are insufficient for centrosomal targeting. Combination with the adjacent IDRs 3-5 restores localization and function. Therefore, the PCMD-1(C1) fragment reconstitutes major functional units. The dominant-negative effect on the viability of the coiled-coil domain and IDR1 and IDR2 could be explained by binding or sequestration of PCMD-1 interaction partners. In the context of the full-length protein, the coiled-coil domain significantly contributes to the accumulation of PCMD-1 at the centrosome and is required to form an organized PCM. The fact that the deletion of the coiled-coil domain only partially compromises viability is not surprising given that even the complete absence of an expanded SPD-5 scaffold at mitosis does not affect the viability (Woodruff et al., 2015). In animals with the deleted coiled-coil domain, SPD-5 levels are not altered, but the PCM appears more dispersed. This could be a direct effect of the coiled-coil domain on SPD-5 or an indirect effect as a result of the reduced PCMD-1 levels at the centrosome. In the yeast two-hybrid assay, we identified a strong self-interaction of PCMD-1, pointing to a tendency towards dimerization or formation of higher-order structures. Self-association is frequently found in centrosomal proteins. This has also been demonstrated for Plp in flies (Galletta et al., 2016). The self-interaction is abolished in the absence of the coiled-coil domains in both bait and prey but not when the coiled-coil domain is only deleted in the bait. This might indicate that PCMD-1/PCMD-1 self-interaction is mediated through binding of the coiled-coiled domain to a different part of the protein, rather than a dimerization of the coiled-coil domain. We speculate that the self-interaction of PCMD-1 plays a significant role in the PCM scaffold integrity. It would be interesting to investigate how exactly the physical properties of the mitotic PCM change in the absence of the coiled-coil domain of PCMD-1.

Recently it was shown that PCMD-1 and SPD-5 play an important role in the function of sensory cilia (Garbrecht et al., 2021; Magescas et al., 2021). After initial recruitment, PCMD-1 and SPD-5 maintain each other at the ciliary base through a positive-feedback loop (Garbrecht et al., 2021). Through our structure-function analysis, we could separate the cilia localization of PCMD-1 from the centrosome localization. A part of the C2 region, missing in PCMD-1(ΔC2) but present in PCMD-1(ΔC2.2) restores PCMD-1 at the ciliary base (Fig. S6). Interestingly, this region comprises the most C-terminal IDR (IDR6), suggesting a potential role in targeting the C2 construct to the cilia. Our observation that PCMD-1 localizes to astral and spindle microtubules (Fig. 1A), could indicate binding to microtubules or a microtubule-binding protein. The enhanced localization of the GFP::PCMD-1(C2) along the length of the cilium may reflect increased microtubule binding, which in the cilia is limited by the N-terminal region in the full-length PCMD-1.

Together we propose a model (Fig. 6B,C) in which PCMD-1 is recruited to the centriole via two anchor points: SAS-7, either directly or indirectly, and SAS-4. The C-terminal part, especially the C2 part of PCMD-1, is sufficient for its centrosomal recruitment and interacts with SAS-4. We speculate that SAS-4 acts in parallel to SAS-7 and that this interaction needs the local environment at the centrioles. Centrosomal PCMD-1 recruits SPD-5 through its N-terminal parts, including IDR3-5. By bringing together SPD-5 and PLK-1, PCMD-1 facilitates PCM core formation and PCM maturation. The coiled-coil domain enhances PCMD-1 centrosomal accumulation through its self-interaction and contributes to the stabilization of the mitotic PCM scaffold. Given that proteins other than PCMD-1, such as SPD-2, play a role in SPD-5 scaffold formation during mitosis, this model might be cell-cycle specific or needed for the initial recruitment of SPD-5 to the PCM core. In summary, we propose that PCMD-1 is one of the proteins that anchors the PCM to the centrioles and functionally bridges the two centrosomal components.

Worms were maintained on NGM plates seeded with the OP50 Escherichia coli strain under standard conditions at 15°C (Brenner, 1974). All worm strains used in this work are listed in Table S1. Unless indicated otherwise, for experimental use, progeny of worms shifted to 25°C at the L4 stage for 16-20 h, were analyzed. gfp::pcmd-1; sas-7(or452)/hT2 worms were allowed to lay eggs for 3 h at 25°C. The laid eggs developed into adults at 25°C for 68 h. Progeny of gfp::pcmd-1;sas-7(or452) worms were used for further analysis.

Worms carrying single-copy transgene insertions were generated by the Universal MosSCI system, according to previously published protocols (Frøkjær-Jensen et al., 2008). Transgenes with the pCFJ350 backbone were injected into EG6699 or EG8081 and the progeny were selected using selection markers. Insertions were verified by PCR. Multiple independent insertion lines were screened for expression of the transgenes.

The pcmd-1(syb1285 syb486[gfp::pcmd-1(ΔCC)])I allele was generated by SunyBiotech by deleting 33 amino acids from E86 including F118, spanning the coiled-coil domain ranging from amino acids E86 to N117, predicted by the COILS program (see below). The deletion was verified by PCR amplification using the oligos GCGCTCCGTTGAGAATCTCGTA and CACAAACGAGCCCGCACGGA and sequencing.

IDRs were annotated based on information provided by the UniProt Consortium (UniProt Consortium, 2019). The coiled-coil domain was defined via the COILS program with a 28-residue window comparing both MTK and MTIDK matrices (weighted and unweighted) (Lupas et al., 1991; Lupas, 1996). The domain structures were illustrated using DOG2.0 (Ren et al., 2009).

Growth and genetic manipulation of the Saccharomyces cerevisiae strain EGY 48/194 (MATα, trp 1, his 3, ura 3, leu2::2/4 LECAop-LEU2) were performed using standard genetic techniques. The yeast strain was transformed with plasmids using lithium acetate (1 M). The selection of the different plasmids was conducted with complete minimal medium lacking histidine/uracil/tryptophane/leucine. Yeast two-hybrid assays were performed using the Grow'N'Glow GFP Yeast Two-Hybrid System (Mobitech) according to the manufacturer's protocol. Full-length cDNA of C. elegans SAS-7, PCMD-1, PCMD-1(ΔCC), PCMD-1(C2), SPD-5, PLK-1 and SAS-4 were inserted into the bait vector pEG202 containing the DNA-binding domain LexA. pEG202_SPD-5 could not be used owing to autoactivation. The cDNA of the different candidates PCMD-1, PCMD-1(ΔCC), PCMD-1(C2), SPD-2, SPD-5, PLK-1, SAS-7 and SAS-4 were cloned into the prey vector pJG45 comprising the B42 transcription activation domain. The third plasmid transformed into the yeast was pGNG1-GFP containing the reporter gene gfp. pEG202-p53 with pJG45-LTA was used as a positive control, whereas pJG45 without an insert was used as a negative control. The presence of the plasmids in yeast was verified by plasmid extraction, followed by PCR amplification of the inserts and sequencing. Expression of the prey proteins was verified by immunoblotting against an HA-tag.

Human embryonic kidney (HEK) 293T cells were cultured at 37°C (5% CO₂) in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Biochrom), 100 units/ml penicillin and 100 µg/ml streptomycin (Gibco).

HEK293T cells were transfected with 4 µg DNA of pTMD143 and/or pTMD144. After 24 h, cells were chilled to 4°C. To prepare cell extracts, cells were washed once with 1×PBS and pelleted by centrifugation at 1000 g for 5 min. Cell pellets were lysed using 100 µl lysis buffer [10 mM Tris-HCl (pH8.0), 50 mM NaF, 0.3 mM Na₃VO₄, 10 mM sodium pyrophosphate, 5 mM β-glycerolphosphate, 120 mM NaCl, 0.25% NP40] containing protease and phosphatase inhibitors (PhosSTOP, Pefabloc, cOmplete Mini, Pepstatin A, Leupeptin, Sigma-Aldrich; Aprotinin, Carl Roth) by incubating them for 30 min at 4°C. Crude extracts were clarified by centrifugation at 12,000 g for 12 min at 4°C and incubated with RFP-trap agarose beads (ChromoTek). Beads were pre-blocked in 3% bovine serum albumin and washed three times in 500 µl lysis buffer before use. Beads were then collected by centrifugation at 2500 g for 5 min at 4°C, washed 1× in lysis buffer and resuspended in SDS gel loading buffer. The double volume was loaded for the CoIP compared with input (I) and flow-through (FT). Immunoblotting was performed by using primary antibodies mouse anti-GFP (1:1000; 11814460001, Sigma-Aldrich) and mouse anti-His (1:1000; MA1-135, Thermo Scientific) and HRP-linked anti-mouse secondary antibody [1:7500; Goat Anti-Mouse IgG (H+L)-HRP Conjugate, 170-6516, Bio-Rad)] and detected using ECL Prime Western Blotting Detection Reagent (Amersham).

L4 stage worms of different strains carrying the construct with the heat-shock promoter (TMD151, TMD157, TMD158, TMD159, TMD162, TMD165, TMD168, TMD167, TMD183 and TMD184) were shifted to 25°C for 15 h. Subsequently, worms laid eggs for 2 h at 25°C. The embryos were mounted on a 2% agarose pad and heat-shocked at 30°C for 1 h (Thermocycler, Bio-Rad). After 2 h recovery at 20°C, embryos were imaged using a SP5 Leica confocal microscope (see ‘Microscopy’ section). Control embryos were incubated at 20°C without heat shock. Feeding RNAi against spd-5 was performed for 20 h at 25°C by using I-2G08 for TMD168 and the pTMD118 feeding clone constructed against the re-encoded region (Woodruff et al., 2015, 2017; Mittasch et al., 2020) for TMD151 and TMD162. Soaking RNAi against sas-4 for TMD183 and TMD184 was performed against the re-encoded region (Cottee et al., 2013). dsRNA was made using the MEGAscript T7 Kit (Invitrogen) and primers TAATACGACTCACTATAGATGGCTTCCGATGAAAATATCGGTGC and TAATACGACTCACTATAGGCAGCGTGCTGTCCACTGTGGC. Worms were incubated in 1 µg/µl dsRNA for 48 h at 20°C and allowed to recover for 24 h at 25°C. Both RNAi treatments were validated on N2 worms by lethality tests and differential interference contrast (DIC) imaging. Embryos with a minimum of two cells in the case of a weak GFP signal or one cell with the GFP signal surrounding the whole circumference of this cell were considered as positive.

To mark the sperm centrioles in marked mating experiments fog-2(q71), females were mated with TMD119 males at 20°C and the progeny imaged by 4D microscopy. For the converse experiment, TMD119 L4 worms were fed fem-1(RNAi) overnight. The hatched progeny were raised on fem-1(RNAi) to block sperm production. Feminized animals were mated with fog-2(q71) males at 20°C. Progeny of the crosses were either imaged by 4D microscopy or used for indirect immunofluorescence. Meiotic stages of embryos in fixed samples were staged by the condensation state of the female DNA and the presence of the polar bodies.

Indirect immunofluorescence was performed using a protocol by Delattre et al. (2004). Hermaphrodite worms were cut in M9 buffer, covered with a coverslip, and placed on ice blocks. After freeze-cracking, slides were fixed in methanol, followed by incubation with primary antibodies anti-SAS-4 (1:500; sc-98949, Santa Cruz Biotechnology), anti-SPD-5 (1:1000, a generous gift from B. Bowerman; Hamill et al., 2002) and anti-GFP (1:500; 11814460001, Roche) overnight at 4°C and with Alexa 488-conjugated (1:500; A32723, Invitrogen Molecular Probes) and Alexa 568-conjugated (1:500; A-11011, Invitrogen Molecular Probes) secondary antibodies and Hoechst 33258 (1:1000, Sigma-Aldrich) at room temperature for 1 h.

Embryos treated for the translocation assay and indirect immunofluorescence samples were imaged at a resolution of 1024×1024 pixels with an HCX PL APO Lambda Blue 63×1.4 oil objective and a step size of 0.7 µm on a SP5 Leica confocal microscope using LAS software. For live-cell imaging, young adult worms were either dissected in 6 µl M9 and mounted on 2% agar pads or dissected in Polybead Microspheres 20.00 µm (diluted 1:10 in M9). Live-cell imaging was performed on an inverted Nikon Eclipse Ti spinning disc confocal microscope using an Andor DU-888 X-11056 camera (1024×1024 pixels), a 100×1.45-NA Plan-Apochromat oil immersion objective and controlled by NIS Elements 4.51 software. z-stacks were taken every 30 s with a step size of 0.7 µm and with 2×2 binning. Embryos for marked mating experiments were imaged using a Zeiss Axio Imager.M2 equipped with epifluorescence and ‘Time to Live’ software from Caenotec. Twenty-five DIC z-stacks were taken throughout the volume of the embryo every 35 s, and fluorescence scans were taken at required time points.

For SIM of live C. elegans embryos, we used the ZEISS Elyra 7 system in the Lattice SIM mode equipped with a Plan Apochromat 63×/1.40 oil immersion objective. Images were acquired with two pco.edge sCMOS 4.2 cameras simultaneously, using the ZEISS DuoLink adapter. Acquired z-stacks with a voxel size of 63×63×110 nm³ and field of view size of 1024×1024 pixels were processed using the SIM processing algorithm of ZEN Black 3.0 software. SIM processed z-stacks have a voxel size of 31.5×31.5×110 nm³. Yeast colonies were acquired using a Leica Stereomicroscope M205 FA, controlled by the Leica Application Suite software (3.2.0.9652) and equipped with a 1×2.11 NA Plan Apo lens and a Leica DigitalDFC340x FX camera.

GFP and RFP intensities were measured on raw images by analyzing z-stacks with ManualTrackMate in Fiji (Tinevez et al., 2017). The time point was either defined through the DNA condensation visualized by the mCherry::H2B marker (for EU3000, TMD107, TMD165, TMD166, TMD175, TMD177, TMD181, TMD202, TMD210) or through corresponding DIC recordings (for TMD119, TMD123, PHX1285, TMD178, TMD179, TMD203, TMD214, TMD216, TMD217). When there was no clear GFP signal recognizable, the centrosomal position was determined by DIC. A fixed radius was applied to measure all fluorescent signals (GFP::PCMD-1: 0.762 µm for Fig. 1E and 0.788 µm for all other figures, GFP::SAS-7: 0.828 µm, RFP::SPD-5: 4.062 µm), background signal, and cytoplasmic background signal outside the embryo, in 3D. Intensities were calculated for each centrosome as: intensity=(C-B)−(CS-B). The total intensity of the background (B) was subtracted from the total intensity of the centrosome (C) and from the total intensity of the cytoplasmic signal (CS). The cytoplasmic signal without background was then subtracted from the centrosomal signal without background. For GFP::SAS-7, the sum of the centrosomal fluorescence intensities per embryo was used to calculate the mean centrosomal fluorescence intensities because of separation defects in some pcmd-1(t3421) embryos. For cytoplasmic GFP values, the background (B) was subtracted from the cytoplasmic signal (CS).

Statistical analysis was performed by using R Studio version 1.2.5003 (R Core Team, 2014). Shapiro–Wilk test was used to test for normality. Levene's test was performed to compare variances. Depending on the normality, variance, and number of groups in the data sets, different comparison tests were performed (see figure legends). Mean values with the standard error of mean were plotted in Prism v6.

Centrosomal circularities were evaluated in one-cell embryos ranging from NEB to metaphase that were immunostained with an antibody against SPD-5. The cell stage was defined by DNA condensation, visualized with Hoechst staining. Image analysis was performed in Fiji (Schindelin et al., 2012). Maximum z-projections were created and the PCM shapes were converted into black/white outlines using the ‘Huang’ threshold. Statistical analysis was performed by using R Studio version 1.2.5003 (R Core Team, 2014). Shapiro–Wilk test was used to test for normality. Levene's test was performed to compare variances. Mann–Whitney U-Test was used for the comparison of circularity values. Mean values with the standard error of mean were plotted with Prism v6.

L4 worms were singled and maintained at the indicated temperature for 16-20 h; laid eggs and hatched adult worms were counted. Statistical analysis was performed in R Studio version 1.2.5003 (R Core Team, 2014). Shapiro–Wilk test was used to test for normality. Levene's test was used to test for homogeneity of variances. Depending on the normality, variance, and number of groups in the data sets, different comparison tests were performed (see figure legends).

We thank the laboratories of C. Osman and N. Wagener for help with the yeast two-hybrid experiments; the laboratory of A. Böttger for help with the cell culture; M. Antoniolli, N. Sharma, A. Trinca and M. Plotnikova for generating tools; A. Bezler for critical comments on the manuscript; the imaging facility CALM, especially H. Harz and J. Ryan for help with imaging, the whole Zeiss-team and especially M. Gorelashvili for giving us the opportunity to image using the Lattice SIM; N. Lebedeva and M. Nöcker for excellent technical assistance; P. Gönczy, K. Oegema and J. Feldman for providing worm strains; B. Bowerman for the anti-SPD-5 antibody; the LSM graduate school. Some strains used in this study were provided by the Caenorhabditis Genetic Center (CGC), which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440).

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.198416