Separation-of-function MCPH-associated mutations in CPAP affect centriole number and length

Centrosomes are major microtubule-organizing centers required for cellular processes like division, motility, and signaling (Vertii et al., 2016). In non-dividing cells, they transition to basal bodies, which generate axoneme of cilia and flagella. A centrosome is a non-membranous cell organelle, mainly consisting of a pair of microtubule-based cylindrical structures at the core called the centrioles, embedded in a proteinaceous matrix referred to as pericentriolar material (PCM). The mature mammalian centriole is ∼250 nm in diameter and 450 nm in length. It exhibits a characteristic nine-fold radially symmetric arrangement of microtubule triplets (Anderson and Brenner, 1971; Gupta and Kitagawa, 2018; Vorobjev and Chentsov, 1982).

Several evolutionary conserved proteins first identified in Caenorhabditis elegans and Drosophila melanogaster tightly regulate centriole duplication during the S-phase of the cell cycle (Delattre et al., 2004; Kemp et al., 2004; Leidel and Gönczy, 2003; Leidel et al., 2005; Pelletier et al., 2006). In human cells, the new (procentriole) centriole generation involves the proximally localized parent centriole proteins, CEP152 (the functional ortholog of Asterless in flies) and CEP192 (the functional ortholog of SPD-2 in worms) (Sonnen et al., 2013). They initiate the recruitment of the centrosome-specific kinase PLK4 (the functional ortholog of ZYG1 in worms and SAK in flies) (Delattre et al., 2006; Habedanck et al., 2005; O'Connell et al., 2001). Once recruited, PLK4 brings STIL (SCL/TAL1 interrupting locus; SIL) protein to the centriole; the functional ortholog of STIL is SAS-5 in worms and ANA-2 in flies. PLK4-mediated phosphorylation of STIL promotes the recruitment of central cartwheel protein SAS-6 (Dzhindzhev et al., 2014). The oligomerization of SAS-6 (also known as SASS6) homodimers dictates the nine-fold symmetry of microtubules at the centriole (Kitagawa et al., 2011a; Nakazawa et al., 2007). The N-terminal of STIL protein contains a proline-rich conserved region (CR2) that interacts with the CPAP (centrosomal P4.1-associated protein; also known as CENPJ) protein (Cottee et al., 2013) and recruits CPAP to the outer region of cartwheel for procentriole formation (Tang et al., 2011). The C-terminal of SAS-6 interacts with the cartwheel pinhead protein CEP135, which directly interacts with CPAP and microtubules (Lin et al., 2013a).

Human CPAP was first identified in a yeast two-hybrid screen as an interacting partner of γ-tubulin complex (Hung et al., 2000). It contains three major conserved domains, which include the N-terminal microtubule-destabilizing PN2-3 domain (residues 311–422) (Cormier et al., 2009; Hung et al., 2004), followed by the microtubule-stabilizing A5N domain (residues 423–607) (Hsu et al., 2008) and the C-terminal located T-complex protein 10 (TCP) or glycine-rich G-box domain (residues 1150–1338) (Fig. 1A) (Zheng et al., 2014). Structural work has revealed that CPAP microtubule destabilizing/stabilizing domains act as a molecular cap for defining organelle size by engaging the PN2–3 domain with the β-tubulin site in longitudinal tubulin–tubulin interactions. The A5N domain engagement ensures microtubule stability. Accordingly, CPAP ensures slow and progressive growth of centriolar microtubules (Sharma et al., 2016). The C-terminal of CPAP (CP3, residues 895–1338), which includes the G-box domain, is involved in direct interaction with the STIL protein (Tang et al., 2011). CPAP protein also exhibits five short coiled-coil (CC) regions, and the CC5 region close to the G-box is responsible for the protein homo-dimerization (Zhao et al., 2010). SAS-4, the worm ortholog of CPAP, has been shown to regulate centrosome size by regulating the levels of PCM organized around the centrioles (Kirkham et al., 2003). The overexpression of CPAP increases centriole length in cycling human cell lines, which translates to abnormal spindle organization and cell division (Kohlmaier et al., 2009; Schmidt et al., 2009; Tang et al., 2009).

Abnormalities in centrosomes have been associated with human diseases like cancers, ciliopathies, and neurodevelopmental disorders. Autosomal recessive primary microcephaly (MCPH) is a neurodevelopmental disorder causing small head size and intellectual disability (Thornton and Woods, 2009). Variants in the core centriole proteins, CEP152, PLK4, STIL, SAS-6 and CPAP have been associated with MCPH, suggesting the importance of proper centrosome organization in brain development (Jaiswal and Singh, 2021). The missense variant, E1235V, in the G box domain of CPAP was identified in primary microcephaly (MCPH6) patients (Bond et al., 2005). This variant decreases the direct interaction of CPAP with the cartwheel protein STIL (Tang et al., 2009). CPAP E1235V overexpression causes excessively long centrioles, which could result in defective spindle organization and positioning (Kitagawa et al., 2011b). Structural work for the CPAP G-box domain from the Danio rerio shows its fibrillary structure and that it has a periodicity of ∼8 nm, a length comparable to the size of tubulin heterodimer and half the size of the successive cartwheel spoke in Tryconympha basal body (Guichard et al., 2012; Hatzopoulos et al., 2013). This suggests that CPAP is not only required for cartwheel assembly to regulate centriole number, but possibly also to make vertical connections with the cartwheel stacks to regulate the centriole size. Specific mutations separating these two functions of CPAP would be a powerful tool to investigate mechanisms dictating this crosstalk between the centriole size and length. Recently, another microcephaly-associated variant, D1196N has been mapped in the G-box of CPAP (Makhdoom et al., 2021). How this leads to MCPH remains largely unknown.

Here, we have utilized E1235V and D1196N to investigate the role of CPAP in regulating centriole length and number. Despite affecting the same G-box domain of CPAP, they caused distinct centriole phenotypes. The D1196N mutation mainly caused an increase in centriole number, and the E1235V mutation increased centriole length. Interestingly, CPAP D1196N does not interact with the cartwheel protein STIL but localizes to the centrioles, revealing an alternative recruitment route, which we show depends on CEP152. Computational analysis of zebrafish equivalent CPAP G-box crystal structures predicts destabilizing effects for E1235V and stabilizing effects for D1196N mutants on the protein structure. Finally, we propose that such defects in centriole organization translate to multipolar spindles and decrease cell viability. Given that proper spindle organization sets the division axis in neural stem cells of the neonatal brain, our work provides mechanistic insights on how MCPH-associated CPAP variants lead to disease.

CPAP is a core centriole protein linking the proteinaceous central cartwheel to the outer centriole microtubule triplet (Hatzopoulos et al., 2013). To investigate the effect of CPAP on centriole organization, we knocked down endogenous CPAP using small interfering RNA (siRNA) (Fig. S1A,B) and counted centriole numbers using centrin-1 as a marker (Paoletti et al., 1996). In control HeLa cells expressing only EGFP, the majority (84%) of interphase cells had two centrioles, and only a minor population had less than (10%) or more than (6%) two centrioles (Fig. 1B,C). Upon CPAP siRNA treatment, a significant increase (49%) in the population of interphase cells with less than two centrioles was observed. Importantly, we could show that the observed phenotype was due to CPAP, as expression of ectopic siRNA-resistant EGFP-tagged CPAP (EGFP–CPAP) in CPAP siRNA background, rescued the phenotype. However, overexpression of EGFP–CPAP did not affect the centriole number. These data suggest that CPAP is required for cartwheel assembly, thus affecting centriole numbers.

Although the G-box domain does not interact directly with tubulin, the previously reported primary microcephaly (MCPH6) missense variant E1235V in the G-box is known to increase centriole length (Bond et al., 2005; Kitagawa et al., 2011b) via an unknown mechanism. Recently, another MCPH variant (D1196N) in the same G-box domain has been mapped (Makhdoom et al., 2021); however, its effects on CPAP functioning remain to be tested. We compared the amino acid sequences encompassing these two residues from organisms like humans, mice, flies and zebrafish (Fig. 1A), which showed their conserved nature during evolution.

CPAP overexpression affects centriole length with minimal effects on their numbers (Kohlmaier et al., 2009). We investigated the impact of CPAP MCPH mutations on centriole number and length. HeLa cells treated with CPAP siRNA and transfected with the siRNA-resistant wild-type EGFP–CPAP, showed two centrioles in the majority (82.5%) of interphase cells (Fig. 2A,B) as indicated by centrin-1 immunostaining. Expressing EGFP–CPAP E1235V had non-significant effects on the number of centrioles, whereas EGFP–CPAP D1196N expression caused an ∼4-fold increase in the population of interphase cells with more than two centrioles both in the asynchronous (Fig. 2A,B) and S-phase synchronized cells (Fig. S3E). We also observed a significant reduction of CPAP centriole localization in the case of the E1235V mutant (Fig. 2C; Fig. S4A).

Although centrin-1 marked the entire elongated centrioles in the U2OS cells, it only marked the centriole foci in the HeLa cells (Figs S1C,D). Given that the E1235V mutation is known to affect centriole length (Kitagawa et al., 2011b), we investigated centriole length under the same conditions using an acetylated tubulin marker. We observed that 40% of the cell population had an increase in their centriole length on E1235V expression. Surprisingly, only a minor (7%) proportion of cells expressing D1196N displayed a very small effect on the centriole length (Fig. 2D–F). The expression of the C-terminal region, i.e. residues 895–1338 (CPAP-CP3) instead of the full-length protein also confirmed the same result (Fig. S1E–H). This suggests that the two separate MCPH mutations are affecting distinct functions of CPAP in regulating organelle number and length.

The E1235V mutation is known to disrupt the direct interaction of CPAP with the upstream STIL protein (Kitagawa et al., 2011b; Tang et al., 2009). However, the effect of novel D1196N MCPH mutation on CPAP–STIL interaction is unknown. The CPAP-CP3 region is sufficient for direct interaction with the STIL-CR2 region (residues 375–510) (Tang et al., 2011). We used recombinant His–MBP-tagged CPAP-CP3 (residues 895–1338) and GST-tagged STIL-CR2 (residues 375–510) proteins expressed in Escherichia coli to test their direct interaction using a GST pulldown assay. We found that just like the E1235V mutation of CPAP, the D1196N mutation also significantly reduced CPAP-CP3 interaction with the STIL-CR2 region (Fig. 3A).

Next, we investigated the centriole localization of CPAP MCPH mutants using fluorescence microscopy. Although the expression of EGFP–CPAP wild-type, E1235V and D1196N is comparable in transfected HeLa cells (Fig. 3B), we observed significantly reduced levels of E1235V at centrioles when compared to the wild-type protein. However, D1196N levels remained comparable to the levels of the wild-type CPAP in the transfected asynchronous (Figs 2C, 3C,D; Fig. S4B) and S-phase synchronized HeLa cells (Fig. S3A–D). Similar effects were observed in HeLa cells transfected with the CPAP-CP3 region (wild-type or mutants) (Fig. 3E,F). This motivated us to investigate the mechanism involved in CPAP recruitment at the centrioles.

STIL protein is required for CPAP localization at the procentriole (Tang et al., 2011). So, mutations disrupting this interaction should affect the localization of CPAP, which is visible for the E1235V mutation (Figs 2C and 3C,D). However, the novel D1196N MCPH mutation in CPAP, which does not interact with STIL (Fig. 3A), can localize normally to the centrioles, challenging the previous notion. To test this carefully, we depleted endogenous STIL using 90 nM siRNA (Fig. S2A,B) and observed centriole localization of transfected wild-type and mutant CPAP by fluorescence microscopy. STIL depletion caused reduced centriole levels of transfected wild-type EGFP–CPAP (Fig. 4A,C) Similar results were obtained for the EGFP–CPAP-CP3 region (Figs S2E,G). This data suggests the involvement of some alternative route to STIL responsible for CPAP localization at the centrioles.

The parent centriole protein CEP152 is an upstream regulator that interacts with the N-terminus of CPAP and is required for its recruitment (Cizmecioglu et al., 2010). Next, we depleted CEP152 using siRNA (Fig. S2C,D) and observed a drastic reduction in the levels of transfected CPAP at the centrioles (Fig. 4B,D). Similar effects were observed when CEP152 was depleted in HeLa cells expressing CPAP-CP3 (wild type and mutants) (Fig. S2F,H). Using immunoprecipitation, we observed that the CPAP MCPH mutants could interact with the endogenous CEP152 (Fig. S2I). The data suggest that STIL is not working alone to recruit CPAP at centrioles, and there must be an alternative route that is CEP152 dependent. In the case of the D1196N mutant, which loses interaction with STIL protein but could normally localize to centrioles, this is possibly because this mutation might enhance recruitment through this alternative route. Simultaneously, we observed that SAS6 localization remains unaffected in cells rescued with wild-type or mutant CPAP proteins (Fig. S3A,B). The localization of CEP135 was not affected by CPAP siRNA. However, the expression of the E1235V mutant in the complementation experiment causes a significant reduction in CEP135 levels at the centrosome compared to the control cells. The D1196N mutant did not affect CEP135 centriole levels (Fig. S3C,D). This suggests a possible dominant-negative effect of the E1235V mutation.

To understand the effect of MCPH mutation on the protein dynamics, we performed molecular dynamics simulation using the available crystal structure of wild-type (PDB ID: 4BXP) and E1021V (PDB ID: 4BXQ) (equivalent to the human E1235V residue mutation, Fig. 4E) of the TCP10/G-box domain of Danio rerio. The D982N structure (equivalent to the human D1196N residue mutation, Fig. 4E) was generated by mutating the respective residue in the wild-type TCP10 structure in UCSF chimera (https://www.cgl.ucsf.edu/chimera/). The normal mode-based geometric simulation was performed using the NMsim webserver (http://www.nmsim.de) for 5000 cycles with 0.5 step size. The root mean square deviation (RMSD) determines the deviation of the protein relative to its native conformation during the course of its simulation. The RMSD profile indicates that the wild-type structure reaches equilibration after ∼3500 conformations, which is similar to what was found for D982N. However, E1021V did not attain equilibration even at 5000 conformations (Fig. 4F). To analyze the effect of mutation on the residue-wise mobility of the proteins, we performed root-mean-square fluctuations (RMSF) analysis. The RMSF graph indicated higher fluctuations for E1021V as compared to the wild-type protein structure, whereas the D982N mutant protein had the lowest fluctuations. In the wild-type TCP10, major fluctuations were observed near the site at residues 60–120 that harbors D982 (residue 65 in the graph) and E1021 (residue no.104 in the graph), indicating conformational flexibility of the region (Fig. 4G). Alternatively, we also analyzed the above structures using the mutation Cutoff Scanning Matrix (mCSM), which utilizes the graph-based atom distance pattern to predict change in Gibbs free energy (ΔΔG), an indicator of protein stability and/or impact on protein interactions. The E1021V has ΔΔG<0, whereas D982N has ΔΔG>0, which indicates the destabilizing and stabilizing effect of these mutations on protein stability or interactions, respectively (Fig. S3F). The distance between the β1 (arginine 1) and β17 (alanine 169) strands of TCP10 wild-type, EV and DN mutant structures over the simulation cycles was significantly increased for the EV mutant (Fig. 4H,I; Movie 1). This indicates a high structural flexibility, which could affect its molecular interactions at centrioles.

Proper centriole organization is essential for bipolar spindle organization and division of neural progenitor cells in developing brains. We investigated the effect of CPAP microcephaly mutants on spindle formation. We observed that the depletion of endogenous CPAP by siRNA results in a 7-fold increase in the mitotic cell population with monopolar spindles (stained with α-tubulin), which supports the fact that CPAP depletion causes a decrease in centriole number (Fig. 5A,B). This phenotype is rescued by expressing exogenous siRNA-resistant EGFP–CPAP. However, the EGFP–CPAP E1235V mutation, which causes an increase in centriole length, and D1196N mutation, which causes an increase in centriole number, showed a 3-fold increase in the population of mitotic cells with a multipolar spindle. This suggests that although the two MCPH mutations in the G-box of CPAP control two aspects of its functioning, they ultimately cause similar spindle organization problems. Given that over-elongation of centrioles could also result in centriole amplification (Marteil et al., 2018), we scored for a fragmented (>2 foci) pericentrin signal (PCM marker) in control (EGFP), EGFP–CPAP- and EGFP–CPAP-E1235V-expressing cells in S-phase and M-phase synchronized cells. The S-phase cells showed no differences, but at M-phase, we observed a 3.5-fold increase in the cell population with fragmented PCM for the E1235V mutant compared to those cells with control or EGFP–CPAP (Fig. 5C,D). This explains the multipolar spindle phenotype observed for the E1235V mutant (Fig. 5A,B). Our finding also corroborated recent work that demonstrated that there are short G2 centrioles in CPAP-E1235V-expressing human induced pluripotent stem cell (hiPSC)-derived brain organoids (An et al., 2022). The CPAP MCPH-associated variants would be expected to affect the pool of neural stem cells, in turn causing a reduction in brain size and head circumference in individuals with this condition. In order to understand the effect of CPAP MCPH mutation on cell viability, we performed MTT assay. We show that CPAP MCPH mutations cause a decrease in cell viability (Fig. 5E), which could explain their contribution to the etiology of primary MCPH by affecting neural stem cell division and their viability in the neonatal brain.

The core structure of the centrosome has microtubule-based structures called centrioles. The centriole microtubules are more stable and slower growing than the cytoplasmic microtubules. However, it remains puzzling how these centrioles achieve definite size and stability. CPAP has emerged as an essential protein linking central cartwheel proteins to the growing outer microtubules and possibly stacking these cartwheels along the length of the centriole lumen. The CPAP microtubule destabilizing (PN2-3) and stabilizing (A5N) domains are known to interact directly with centriole microtubules. The C-terminal G-box domain of CPAP is engaged in direct interaction with the core centriole cartwheel protein STIL, which is involved in CPAP localization and procentriole generation.

Here, we simultaneously characterized the two reported MCPH mutations in the G-box of CPAP (E1235V and D1196N), which revealed that, despite these affecting the same domain of the proteins, they have different effects on CPAP functions. According to previous literature, the CPAP E1235V (MCPH6) caused overly long centrioles without affecting the centriole number. Interestingly, CPAP D1196N had minimal effect on centriole length. However, it caused a significant increase in their number. Using an in vitro pulldown assay, we found that both CPAP MCPH mutations affect direct interaction with the STIL protein. Moreover, we observed using fluorescence microcopy that E1235V reduces CPAP centriole localization, whereas D1196N does not affect its localization.

Further work revealed that STIL is only partially required for the centriole localization of CPAP, and we show an alternative CEP152-dependent pathway that is simultaneously involved in this process (Fig. 5F). The molecular dynamics simulations of the wild-type, E1201V and predicted D982N zebrafish G-box crystal structure suggest that these MCPH mutations affect protein dynamics differently. The E1235V increases protein flexibility, which could affect molecular interactions at centrioles. The overly long centrioles of CPAP E1235V, which eventually get fragmented by the M-phase, and the amplified centrioles in the case of D1196N, both resulted in multipolar spindles. We also show that the abnormal spindle organization in CPAP MCPH affects the cell viability.

This work highlights the importance of the CPAP G-box domain in regulating two distinct functions of CPAP, i.e. centriole number and centriole size and/or length. Given that important functional mutations map to the G-box domain of CPAP, further work is required to understand the involvement of this region in the CPAP activity. It would be interesting to look at the structural effects of E1235V and D1196N on the G-box organization. In addition, both investigated MCPH mutations replace negatively charged amino acids for uncharged amino acids, which points towards the involvement of protein interactions. Several centriolar proteins found in complexes with CPAP, like CEP135 (Lin et al., 2013a), centrobin (Gudi et al., 2011) and CEP120 (Lin et al., 2013b), are also involved in regulating centriole length. These would be attractive candidates for gaining mechanistic insights and their possible involvement in the CPAP centriole localization and functioning.

HeLa Kyoto cells and U2OS were cultured in DMEM (Himedia, #AL111) supplemented with 10% fetal bovine serum (Himedia, #RM10432), 2 mM antibiotic (penicillin and streptomycin) solution (Merck, #P4333), and 2 mM L-glutamine (Merck, #G7513) in a humidified incubator with 5% CO₂ at 37°C.

Full-length and C-terminal coding sequences (residues 895–1338; CP3) were PCR amplified from the siRNA-resistant CPAP and were fused in the same reading frame as EGFP in pCDNA5/FRT/TO (gift from Prof. Andrea Musacchio, MPI Dortmund, Germany) or to 6xHis-MBP tag in pET Duet vector (gift from Prof. Andrea Musacchio). Full-length and mutants (E1235V and D1196N) were generated by site-directed mutagenesis using the Phusion High-Fidelity DNA Polymerase (New England Biolabs, #M0530S), followed by DpnI digestion (Takara, #1235A). The cDNA of STIL (Addgene #80266) and the CR2 fragment (residues 375–510) sequences were PCR amplified and cloned in-frame to the GST tag in pGEX-6P1 vector (gift from Prof. Andrea Musacchio). The siRNA-resistant CPAP cDNA containing plasmid was obtained from Addgene (#46390).

E. coli (BL21) cells were transformed with the recombinant plasmid pETDuet_His-MBP_CPAP-CP3 (wild-type and mutants) and pGEX6P1_GST_STIL-CR2. The bacterial cultures were grown at 37°C until an optical density at 600 nm (OD₆₀₀) of 0.4–0.8 was reached. Protein expression was induced by adding either 0.4 mM IPTG to the cell culture with pET_Duet_His-MBP_CPAP-CP3 (wild-type and mutants) or 1 mM IPTG, 0.1 mM ZnCl₂, 2% glucose solution and 1.5% ethanol solution for pGEX6P1_GST_STIL-CR2. The cells were grown at 18°C for 18 h. For affinity purification of the His-tagged proteins, the bacterial cells were lysed using the lysis buffer (50 mM Tris-HCl pH 8.0, 300 mM NaCl and 5 mM imidazole). The cell lysate was sonicated at 35% amplitude on ice three times for 10 s with a 10 s break. The cell debris was centrifuged at 14,500 g, 10 min 4°C, and the supernatant was incubated with pre-equilibrated Ni-NTA agarose beads (Qiagen, #1018244) with shaking on ice for 3 h. The beads were spun down at 300 g for 2 min to remove the unbound fraction. This was followed by three washing steps with the washing buffer (50 mM Tris-HCl pH 8.0, 300 mM NaCl, 20 mM Imidazole). Finally, the protein was eluted from the beads using the elution buffer (50 mM Tris-HCl pH 8.0, 300 mM NaCl and 300 mM imidazole). GST-tagged proteins were purified by the same protocol using GST lysis buffer (50 mM Tris-HCl pH 8.0, 100 mM NaCl, 1% Triton-X, 1 mM DTT and 1 mM PMSF), glutathione superflow resin (Takara, #635607) and GST wash buffer (1× PBST, 1% NP-40 and 2 mM DTT).

For plasmid transfections, HeLa Kyoto cells at 80–90% confluency were transfected using Lipofectamine 3000 as per the manufacturer's protocol and analyzed after 48 h. To visualize metaphase cells, they were synchronized for the G₂/M phase using 5 μM RO-3306 (Merck, #SML0569) for 18 h followed by 45 min of release. Cells were synchronized at the S-phase using a double-thymidine block (2 mM) to detect endogenous STIL, CEP152, CEP135 and SAS6. Cells were incubated with thymidine for 18 h, followed by 9 h of release, and the second block was introduced for 15 h followed by cell fixation for immunofluorescence. Double-stranded CPAP siRNA oligonucleotides were synthesized with 3′-UU overhangs with the sequence 5′-AGAAUUAGCUCGAAUAGAA-3′ (Dharmacon). STIL and CEP152 siRNA duplexes were synthesized with the sequence 5′-GUUUAAGGGAAAAGUUAUU-rU-rU-3′ and 5′-GGAGGACCAUGUCAUUAGACU-rU-rU-3′, respectively (Merck).

Cells grown on coverslips were washed with 1× PBS to remove media. They were permeabilized using 0.1% Triton X-100 prepared in 1× PBS (1× PBST) and fixed with 4% (v/v) paraformaldehyde solution prepared in 1× PBS for 10 min at room temperature. The cells were washed three times with 1× PBST and blocked using 1% BSA in 1XPBST for 1 h. The cells were washed three times with 1× PBST and left in primary antibodies overnight at 4°C. After primary antibody incubation, cells were washed three times and incubated in secondary antibody for 1 h at room temperature. Cells were stained with DAPI (0.5 µg/ml) prepared in 1× PBS for 15 min and mounted on glass slides with the help of DABCO-Mowiol mounting solution. The fluorescent images were acquired using the 100× oil, 1.35 NA objective lens of the epi-fluorescence microscope (Olympus IX83).

Primary antibodies used for immunofluorescence were mouse anti-centrin-1 (Merck, #04-1624, 1:500), rabbit anti-centrin-1 (Abcam, #Ab156858, 1:500), mouse anti-pericentrin (Abcam, #Ab28144, 1:100), rabbit anti-STIL (Abcam, #Ab89314, 1:100), mouse anti-α tubulin (Merck, #T6074, 1:500), mouse anti-tubulin acetylated (Merck, #T6793), rabbit anti-CENPJ (Invitrogen, #PA597577, 1:300), SAS6 (Abcam, # Ab85993, 1:200), rabbit anti-CEP152 (Abcam, #Ab183911, 1:250) and CEP135 (Proteintech, #24428-1-AP, 1:100). Secondary antibodies used were anti-rabbit IgG (Fc) TRITC (Merck, # SAB3700846, 1:500), anti-rabbit IgG (Fc) FITC (Sigma, #F7512, 1:500), anti-mouse IgG (Fc) FITC (Sigma, #F5387, 1:500) anti-mouse IgG (Fc) TRITC (Merck, #SAB3701020, 1:500) and anti-rabbit IgG Alexa Fluor 405 (Invitrogen, #A-31556, 1:250). Images are represented as maximum projections of Z-stacks and analyzed using ImageJ. The bar graphs were plotted using Microsoft Excel software, and the figures were arranged on Microsoft PowerPoint.

GST pulldown assay was performed to test the interaction between the 6×His-MBP-tagged CPAP CP3 wild-type or mutant proteins and the GST–STIL-CR2 region. Bacterial cell lysates expressing His-MBP-tagged CPAP-CP3 constructs were incubated with immobilized GST–STIL-CR2 fusion protein at 4°C for 3 h in binding buffer (50 mM Tris-HCl pH 8.0, 100 mM NaCl, 1% Triton-X, 1 mM DTT and 1 mM PMSF). After incubation, the beads were washed three times with GST wash buffer. The proteins were eluted from the beads by incubating with SDS Laemmli buffer and then boiling at 95°C for 10 min. The proteins were resolved on 10% SDS-PAGE gels. The protein bands were transferred onto nitrocellulose membrane and analyzed by western blotting (Sambrook, 2001). Primary antibodies used for immunoblotting were rabbit anti-CENPJ (Invitrogen, #PA597577, 1:1000), mouse anti-His (Invitrogen, #MA1-21315, 1:1000), rabbit anti-GST (Merck, #ZRB1223,1:1000) and mouse anti-β-actin (Santa Cruz Biotechnology, #SC-47778, 1:1000). Secondary antibodies used were anti-mouse IgG HRP linked (Cell Signaling Technology, #CST 7076S,1:10,000) and anti-Rabbit IgG HRP linked (Cell Signaling Technology, #CST 7074S, 1:10,000). The original blot images for figures in the manuscript are provided in Fig. S5.

Normal mode-based geometric simulation method was used for multi-scale modelling of protein conformational changes. The X-ray diffraction structures of the wild-type (PDB ID: 4BXP) and E1021V (PDB ID: 4BXQ) (equivalent to the human CPAP E1235V mutant residue position) of CPAP TCP10/G-box domain of Danio rerio were used for the simulation. The D982N structure (equivalent to human CPAP D1196N mutant residue position) was generated by mutating the 982 residue of wild-type TCP10 in UCSF chimera (https://www.cgl.ucsf.edu/chimera/). The protein structures were submitted to the NMSim webserver (http://www.nmsim.de) for rigid cluster normal-mode analysis and simulated for 5000 cycles with 0.5 step size. Other parameters were kept at default during submission. The RMSD and RMSF graphs were plotted using Origin software. Additionally, the simulation results were validated using mCSM (https://biosig.lab.uq.edu.au/mcsm/stability), which uses the concept of graph-based structural signatures to study and predict the impact of single point mutations on protein stability.

Approximately 5000 cells were seeded per well in a 96-well tissue culture plate (Tarsons, #980040) and allowed to settle overnight. The next day, cells were transfected using Lipofectamine 3000 per the manufacturer's protocol. Each transfection was performed in duplicate. The cells were treated for 48 h at 37°C in a 5% CO₂ atmosphere. Then, 10 μl of 3-(4,5- dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT; Invitrogen, #M6494) stock solution prepared in 1× PBS at 5 mg/ml was added to each well in 100 μl serum-free medium. The plate was incubated at 37°C in a humidified incubator chamber for 4 h. 100 μl acidic isopropanol (prepared by adding 132 μl of 0.04 M HCl in 15 ml isopropanol) was added to each well and mixed thoroughly in a shaker to dissolve the purple formazan crystals. The absorbance was measured using a microplate reader (Biotek Synergy H1) at a wavelength of 570 nm. The results were analyzed and plotted in Microsoft Excel.

Centrin-1 was used as a centriole marker to draw a region of interest (ROI) in ImageJ for intensity measurements at centrioles. The ROI size depended on the centrin size (reference channel). For a cell, the same ROI was duplicated into the EGFP channel at the same location. An ROI with the same size was placed around the centrosome to measure the background. The absolute fluoresce intensity of EGFP was reported after subtracting the average background signal in the same channel (Fig. S2J).

All the bar graphs were generated in Microsoft Excel, which plots the mean±s.d. for each condition. Origin software was used to plot RMSF and RMSD graphs. Experiments were repeated at least twice, and the number of cells (n) analyzed was >50 for each experiment. A two-tailed unpaired Student's t-test was applied to compare two conditions to generate the P-values. The P-values for centriole and spindle pole counting experiments were generated using the chi-square test.

The HeLa Kyoto, U2OS, pCDNA5/FRT/TO, pETDuet1, and pGEX vectors were kind gifts from the laboratory of Prof. Andrea Musacchio, Max Planck Institute of Molecular Physiology, Dortmund, Germany. We thank Prof. Andrea Musacchio and Dr Clemens Cabernard for critically reading the manuscript. We are thankful to the Centre for Research & Development of Scientific Instruments at IIT Jodhpur for microscope facilities.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.261297.reviewer-comments.pdf