Cep57 and Cep57l1 function redundantly to recruit the Cep63–Cep152 complex for centriole biogenesis

In mammalian cells, centrosomes function as spindle poles in mitosis and basal bodies for cilia formation. Typically, a centrosome is made of two cylindrical centrioles surrounded by pericentriolar material (PCM) (Nigg and Raff, 2009). In dividing animal cells, the duplication of centrioles must be tightly controlled with only a procentriole (herein referred to as daughter centriole) forming next to each of the two pre-existing centrioles (mother centrioles) per cell cycle to ensure the correct bipolar spindle formation, which is important for the integrity of the genome. Aberration in centriole duplication is commonly present in numerous tumors and other diseases (Bettencourt-Dias et al., 2011; Gönczy, 2015; Nigg and Holland, 2018). In recent years, great progress has been made in the understanding of the molecular mechanisms of centriole biogenesis. To initiate centriole duplication in mammals, Plk4 is firstly recruited by the Cep192-containing ring around the proximal mother centrioles at the early G1 phase and subsequently translocated to the larger Cep152-containing ring when Cep152 is recruited to the mother centriole (Kim et al., 2013; Sonnen et al., 2013). At the Cep152 ring, Plk4 recruits and phosphorylates Stil (also known as SAS-5 in Caenorhabditis elegans) to load Sas-6 for cartwheel formation (Arquint et al., 2015, 2012; Cizmecioglu et al., 2010; Dzhindzhev et al., 2010; Moyer et al., 2015; Ohta et al., 2014, 2018). Several proteins, including Cep135, Cpap (also known as CENPJ), Cp110 (CCP110) and Cep120, contribute to building the centriolar microtubule (MT) wall and mediate centriole elongation (Azimzadeh and Marshall, 2010; Brito et al., 2012; Carvalho-Santos et al., 2012; Comartin et al., 2013; Hung et al., 2004; Kohlmaier et al., 2009; Lin et al., 2013a,b; Loncarek and Bettencourt-Dias, 2018; Schmidt et al., 2009). It is known that Cep152 is recruited by Cep63 to act as the cradle for the mother centriole-dependent (MCD) centriole biogenesis (Brown et al., 2013; Kim et al., 2019; Sir et al., 2011; Zhao et al., 2013). However, it is still unclear how Cep63 or a Cep63–Cep152 complex is targeted to the proximal end of mother centrioles.

Cep57, an FGF-2 binding and trafficking protein, has been reported to be required for normal centriole duplication, proper chromosome segregation, PCM organization and centriole engagement (Bossard et al., 2003; Cuevas et al., 2013; Meunier et al., 2009; Snape et al., 2011; Watanabe et al., 2019; Wu et al., 2012; Zhou et al., 2016). Cep57, Cep63 and Cep152 form a stable complex at the proximal end of the centrioles, and Cep57 is a proximity interactor of Cep63 (Firat-Karalar et al., 2014; Lukinavičius et al., 2013; Sieben et al., 2018), suggesting that Cep57 may be involved in the centriolar targeting of the Cep63–Cep152 complex for centriole duplication.

In this study, we find that Cep57 and its paralog Cep57l1 function redundantly to recruit the Cep63–Cep152 complex for centriole duplication.

To investigate the spatial relationship of Cep57, Cep63 and Cep152, we co-stained Cep57 with either Cep63 or Cep152 in U2OS cells. Consistent with results from the previous study (Lukinavičius et al., 2013), 3D-structured illumination microscopy (3D-SIM) results showed that Cep57 formed a ring-like structure with Cep63 and Cep152 at the proximal end of both centrioles (Fig. 1A). Cep57 colocalized with Cep63 in rings of a similar diameter (Fig. 1A,B). Consistent with this, GFP–Cep57 was also enriched at the base of mother centrioles in U2OS cells. In addition to the centriolar localization, overexpressed GFP–Cep57 forms bundled filaments in the cytoplasm (Momotani et al., 2008; Wu et al., 2012; Zhou et al., 2016), and both Cep63 and Cep152 were recruited to these filaments (Fig. 1C,D), indicating that Cep57, Cep63 and Cep152 form a complex, and Cep57 could recruit Cep63 and Cep152 in vivo.

During multiciliation, mouse tracheal epithelial cells (mTECs) assemble hundreds of centrioles through both MCD and deuterosome-dependent (DD) pathways, respectively driven by Cep63 and its paralog Deup1 (Zhao et al., 2019, 2013). Cep152 localizes to both parental centrioles and deuterosomes but Cep63 only displays parental centriole localization (Zhao et al., 2013). Interestingly, Cep57 behaved similarly to Cep63 at the early stages of the centriole amplification. It colocalized with Cep152 at the proximal ends of parental centrioles from stage I but showed no deuterosome localization in stages II and III (Fig. 1E–G). In stage IV, Cep57 started to occupy the base of procentrioles formed around the parental centrioles and deuterosomes. In stages V and VI, Cep57 persistently localized to the proximal ends of nascent centrioles as a full ring structure (Fig. 1E–G), while Cep152 and Cep63 displayed as a dot (Zhao et al., 2013). Therefore, Cep57 only contributes to the MCD pathway and could recruit Cep152 to parental centrioles together with Cep63.

To explore the relationship among Cep57, Cep63 and Cep152, we performed co-immunoprecipitation and GST pulldown experiments (Fig. 2; Fig. S1). Cep57 localized to the centriole via its C-terminal MT-binding domain (Momotani et al., 2008) and interacted with Cep63 through its N-terminal coiled-coil domain (Fig. 2A; Fig. S1A,B). Cep63 bound to Cep57 via its N-terminal centriole-binding region, and its C-terminus interacted with the centrosomal-targeting region of Cep152 (152M) (Fig. 2B; Fig. S1C,D) (Hatch et al., 2010; Zhao et al., 2013). In the GST pulldown assay, GST–Cep57 directly interacted with His–Cep63, but could not pull down Cep152M in the absence of Cep63 (Fig. 2C), indicating that Cep57 forms the complex with Cep152 through Cep63 and these three proteins interact in a hierarchical manner (Fig. 2D). These results indicate that Cep57 localizes to the proximal mother centriole through its C-terminus and recruits the Cep63–Cep152 complex by directly binding to the centriolar targeting region of Cep63.

To test whether Cep57 is required for the centrosomal targeting of Cep63 and Cep152, we generated Cep57 and Cep63 knockout (KO) U2OS cell lines using the CRISPR/Cas9 method (Fig. 2E; Fig. S2A,B) (Cong et al., 2013). Consistent with previous studies (Brown et al., 2013; Sir et al., 2011), bipolar spindle formation and centriole duplication were mildly affected by the depletion of Cep63 (Fig. S2C–E). Immunofluorescence analysis of Cep63 KO or Cep57 KO cell lines revealed a complete loss of Cep63 or Cep57 ring-like signals at the proximal end of centrioles (Fig. 2F–I). As reported previously (Lukinavičius et al., 2013; Sir et al., 2011), the Cep152 centrosomal signal was reduced dramatically in Cep63 KO cells compared to that in control cells (Fig. 2F). The centrosomal localization of Cep57 was slightly affected in Cep63 KO cells (Fig. 2G). Surprisingly, both Cep63 and Cep152 were still able to localize at the centrosome in Cep57 KO cells (Fig. 2H–I). These results are in accordance with previous findings showing that Cep63 is essential for Cep152 centrosomal localization, whereas Cep57 seems to be dispensable for loading the Cep63–Cep152 complex to the centrosome (Lukinavičius et al., 2013; Sir et al., 2011).

The dispensable role of Cep57 in the centriolar targeting of the Cep63–Cep152 complex is contradictory to the fact that Cep57 forms a stable complex with Cep63 and Cep152 by directly binding to the centriolar-targeting region of Cep63. The paradox raises the possibility that other proteins may function redundantly with Cep57 to recruit Cep63 and Cep152.

Cep57l1, the paralog of Cep57, has been reported to function in the anterior and posterior cruciate ligaments of the knees (Liu et al., 2015). Cep57l1 has similar function domains and has an ∼43% identity to Cep57 (Fig. 3A; Fig. S3). We hypothesized that Cep57l1 might localize at the centriole and function redundantly with Cep57 to recruit the Cep63–Cep152 complex. We first tested whether Cep57l1 also interacts with Cep63 as Cep57. We found that the N-terminus of Cep57l1 was necessary and sufficient to bind Cep63 (Fig. 3B,C,G). Cep63 also interacted with Cep57l1 through its N-terminal centrosomal-targeting fragment (Fig. 3E–G).

Owing to the lack of an anti-Cep57l1 antibody, we examined the subcellular localization of Cep57l1 by expressing GFP–Cep57l1 in U2OS cells. GFP–Cep57l1 also formed a ring-like structure with Cep63 and Cep152 at the proximal end of centrioles during interphase with a diameter comparable to that of the Cep57 rings (227±33 nm, 44 centrioles from 24 cells) (Fig. 3D). GFP–Cep57l1, however, did not form bundled filaments (Fig. 3D,H). Although both the N- and C-termini of Cep57l1 localized to the centriole, their fluorescence intensities were weaker by at least 4-fold as compared to that of full-length Cep57l1 (Fig. 3D; Fig. S4A), indicating that both regions are required for the strong centriolar localization of Cep57l1.

In multiciliated mTECs, GFP–Cep57l1 displayed a similar subcellular distribution to that of Cep57 and Cep63 (Fig. 3I versus Fig. 1E–G) (Zhao et al., 2013). It also specifically associated with the parental centrioles but not with deuterosomes at the early stages of centriole amplification in mTECs (stages II and III) (Fig. 3I). In stage IV, GFP–Cep57l1 started to accumulate at the base of nascent centrioles to form a goblet-like structure (Fig. 3I). Like Cep57, GFP–Cep57l1 still localized at the proximal end of centrioles in a ring-like pattern in stages V and VI. Collectively, these findings imply that Cep57l1 interacts with Cep63 at the proximal end of centrioles.

To examine whether Cep57l1 is required for the localization of Cep63 and Cep152, we depleted Cep57l1 by RNAi in both wild-type and Cep57 KO U2OS cells (Fig. 4). Given the lack of an anti-Cep57l1 antibody, we selected Cep57l1 siRNA oligonucleotides by monitoring the GFP–Cep57l1 signal when co-expressed with various siRNA oligonucleotides, and two oligonucleotides, 57l1i-1 and 57l1i-2, were chosen based on knockdown efficiency (Fig. S4B).

In contrast to what is seen upon the depletion of Cep57 (Watanabe et al., 2019), the depletion of Cep57l1 did not affect the integrity of PCM (Fig. S4C). In addition, and similar to what is seen with Cep57 deficiency (Fig. 2H,I), the depletion of Cep57l1 alone did not impair the centriolar localization of Cep63 or Cep152 in control U2OS cells (Fig. 4A). The double depletion of Cep57 and Cep57l1, however, significantly reduced the centriolar localization of Cep63 and Cep152 (Fig. 4B). Furthermore, the centriolar localization of Cep63 and Cep152 was rescued by expressing either GFP–Cep57 or an siRNA-resistant GFP–Cep57l1 (GFP–Cep57l1R), but not GFP–centrin1 (GFP–Cetn1) (Fig. 4C,D). Therefore, Cep57 and Cep57l1 function redundantly to ensure the centriolar localization of the Cep63–Cep152 complex.

To investigate whether the co-depletion of Cep57 and Cep57l1 interferes with procentriole assembly, we pulse-labeled S-phase cells with EdU, and examined procentrioles marked by Sas-6 in the EdU-positive cells. We observed a marked reduction in procentriole biogenesis in the co-depleted cells (Fig. 4E), demonstrating a critical function of Cep57 and Cep57l1 in the initiation of centriole duplication.

In this study, we find that Cep57 and its paralog Cep57l1 localize to the proximal end of parental centrioles and function redundantly to recruit the Cep63–Cep152 complex. Co-depletion of Cep57 and Cep57l1 disrupts the centriolar localization of the Cep63–Cep152 complex and consequently inhibits normal procentriole biogenesis in S-phase cells. They appear to play similar roles in the parental centriole-mediated centriole amplification in multiciliated cells (Figs 1E–G and 3I). Interestingly, their strong basal body localization (Figs 1E–G and 3I) also implies that they have roles in multicilia formation and function, which should be clarified in the future.

Recently, Cep295 has been identified as the key protein for the conversion of daughter centrioles into functional mother centrioles, and the loading of Cep152 by Cep135 and Cep295 to daughter centrioles is critical for the conversion (Fu et al., 2016; Izquierdo et al., 2014; Tsuchiya et al., 2016). Interestingly, Cep295 is only required for targeting of Cep152 to the young mother centriole, but not to the old mother centriole, implying that other scaffold proteins might play a role in anchoring Cep152 to the old mother centriole. Our findings here raise a possibility that Cep63 recruited by Cep57 and/or Cep57l1 may function in detaining Cep152 at the mother centriole after its initial loading by Cep135–Cep295 during the centriole-to-centrosome conversion. In the future, it will be interesting to investigate the relationship among Cep57 and Cep57l1, Cep63 and Cep295 in the centriolar localization of Cep152.

mTECs were isolated from 4-week C57BL/6J mice and cultured as described previously (Zhao et al., 2013). Experiments involving mouse tissues were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of Institute of Biochemistry and Cell Biology. U2OS (HTB-96, ATCC) and HEK 293T (CRL-11268, ATCC) cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific) at 37°C and 5% CO₂. Cells were routinely tested for mycoplasma contamination by PCR. For U2OS cells, plasmids were transfected using Lipofectamine 2000 according to the manufacturer's instructions (ThermoFisher). For ectopic protein expression and lentiviral production in HEK 293T cells, plasmids were transfected by the calcium phosphate method. To observe the process of centriole amplification, mTECs were harvested at day 3 after the air-liquid interface (ALI) culture, and lentiviral infection of mTECs was performed as described previously (Zhao et al., 2013). Human Cep57l1 protein was depleted using two different siRNA duplex oligonucleotides that targeted the sequences 5′-CGTTGTACTCTTCTAGAGA-3′ (57l1i-1) and 5′-GCTTCAAACTGGACTTGAA-3′ (57l1i-2). A non-targeting siRNA duplex oligonucleotide was used as a control with the sequence 5′-TTCTCCGAACGTGTCACGTtt-3′ (Ctrli). U2OS cells were transfected with siRNA oligonucleotides using Lipofectamine RNAiMAX reagent (Thermo Fisher Scientific) for 48 h. To improve the efficiency of knockdown, a second round of siRNA transfection was conducted for another 48 h.

Full-length mouse cDNAs of Cep57 (NM_026665), Cep63 (NM_001301689), and Cep57l1 (NM_029132), centrin1 (NM_007593) and PCNT (1485-2896 aa, NM_001282992) were amplified from mTEC cDNA by PCR amplification. Full-length human cDNAs of Cep57 (NM_014679) and Cep57l1 (NM_173830) were amplified from HEK 293T cDNA and the fragments were subcloned into pEGFP-C1 (Clontech, 632470). For ectopic expression, fragments of the indicated regions of mouse Cep57, Cep63 and Cep57l1 were made by PCR amplification of and cloned into pEGFP-C1 or pcDNA3.1-NFLAG [a version of pcDNA3.1 (ThermoFisher, V79020), modified to encode the FLAG tag], respectively. To express RFP–centrin1, the full-length of centrin1 was cloned into pLV-RFP (generated by replacing the GFP cassette in pLV-GFP with an RFP cassette; kindly provided by Dr Q. Zhai, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences). To infect mTECs, Cep57 or Cep57l1 were cloned into pLV-GFP (kindly provided by Dr Q. Zhai). For rescue experiments, RNAi-resistant GFP–Cep57l1 (GFP–Cep57l1R) was made by site-directed mutagenesis of pEGFP-Cep57l1. For antibody production, the full-length of mouse Cep57 and Cep152 [amino acids (aa) 1–735] were cloned into pGEX-4T-1 (GE Healthcare, 27-1542-01) to express GST-tagged proteins, and the PCNT (aa 1485–2896) fragment was cloned into pET-32a (Novagen, 69015) to express His-tagged proteins. All constructs were verified by sequencing.

GST and GST–Cep57 were purified with glutathione–agarose beads (Sigma, G4510). His–Cep63 and His–Cep152M were purified with Ni-NTA agarose beads (Qiagen, 30230). In vitro GST pulldown assays were performed by incubating recombinant proteins in lysis buffer [500 mM NaCl, 50 mM HEPES-KOH pH 7.8, 5 mM EDTA, 3 mM DTT, 0.5 mM PMSF and complete protease inhibitors (Sigma, 539134)] for 4 h at 4°C. After three washes with lysis buffer, proteins were eluted with SDS-PAGE sample buffer and analyzed by SDS-PAGE.

The Cep57- and Cep63-knockout U2OS cell lines were generated by the CRISPR/Cas9-mediated gene editing system (Cong et al., 2013). The sequences of sgRNAs were Cep57 (5′-CACAAGCCCTAGCCATGCCG-3′), Cep63 (5′-AACATGCTGTAGAAGCTATA-3′) and Scramble (5′-GCACTACCAGAGCTAACTCA-3′). The synthesized primers were annealed and inserted into pLV-Cas9-puromycin vector.

U2OS cells were transfected with the plasmids for 48 h and were treated with 1.5 μg/ml Puromycin (Sigma, P8833) for 10 days. The cells were then serially diluted and single colonies were examined by immunofluorescence microscopy to confirm the loss of the indicated proteins. Genomic DNAs isolated from the selected cells were PCR amplified with the following primers: Cep57-p (5′-GTGGTGTAGAGAATATAACTTCAAT-3′), Cep57-r (5′-CTCGATCATTGC ACAAAGCTTTACT-3′), Cep63-p (5′-CTCATACATGAGGCCAGAATACAAA-3′) and Cep63-r (5′-GCCTAGTGCTTAATACCTAGTCATT-3′). The PCR products were cloned into a TA-cloning vector and for each KO cell line, six clones were picked for sequencing.

Co-immunoprecipitation was performed as described previously (Zhao et al., 2013). In brief, HEK293T cells were transiently transfected, cultured for 48 h, washed with PBS and lysed in lysis buffer [500 mM NaCl, 50 mM HEPES-KOH pH 7.8, 5 mM EDTA, 3 mM DTT, 0.5 mM PMSF and complete protease inhibitors (Sigma, 539134)]. Lysates were further cleared by centrifugation at 14,000 g and subjected to co-immunoprecipitations using anti-Flag agarose beads (Sigma, A2220) or GFP-Trap agarose beads (Chromotek, gat-20). Beads were washed with lysis buffer, eluted in SDS-loading buffer and run on SDS-PAGE gels.

The following primary antibodies were used for western blotting in this study: mouse monoclonal antibodies against Flag (Sigma, F3165, 1:5000), His (Sigma, H1029, 1:3000) and β-actin (Sigma, A5316, 1:5000); rabbit antibodies against Cep63 (Millipore, 06-1292, 1:2000), GFP (Proteintech, 50430-2-AP, 1:2000) and GST (Shanghai Wolwo biotech, Mab-GS01, 1:3000). Secondary antibodies were HRP-conjugated goat anti-mouse IgG (H+L) (ThermoFisher, G21040, 1:5000) and goat anti-rabbit IgG (H+L) (ThermoFisher, G21243, 1:5000).

Primary antibodies purchased for immunostaining were: chicken antibody against GFP (ThermoFisher, A10262, 1:200); mouse antibodies against centrin (Millipore, 04-1624, 1:200), acetylated tubulin (Sigma, T6793, 1:1000), α-tubulin (Sigma, T5168, 1:1000) and Sas-6 (Santa Cruz Biotechnology, sc-81431, 1:50); and rabbit antibody against CP110 (Proteintech, 12780-1-AP, 1:200) and Cep63 (Millipore, 06-1292, 1:200). Rabbit anti-PCNT (against 1485–2896 aa of mouse PCNT, 1:200), chicken anti-Cep152 (against 1–735 aa of mouse Cep152, 1:300) and rat anti-Cep57 (against full-length of mouse Cep57, 1:200) polyclonal antibodies were generated with purified GST- or His-tagged proteins (Abclonal). The following secondary antibodies purchased from ThermoFisher were used for immunofluorescence: goat anti-mouse IgG (H+L) Alexa Fluor 405 (A-31553, 1:200), goat anti-rabbit IgG (H+L) Alexa Fluor 405 (A-31556, 1:200), donkey anti-rabbit IgG (H+L) Alexa Fluor 488 (A-21206, 1:1000), goat anti-chicken IgY Alexa Fluor 488 (A-11039, 1:1000), goat anti-chicken IgY Alexa Fluor 546 (A-11040, 1:1000), goat anti-rat IgG (H+L) Alexa Fluor 546 (A-11081, 1:1000), and goat anti-rabbit IgG (H+L) Alexa Fluor 546 (A-11035, 1:1000).

For immunofluorescence experiments, mTECs were pre-extracted with 0.5% Triton X-100 in PBS for 3 min before fixation with 4% fresh paraformaldehyde in PBS for 15 min at room temperature. For U2OS cells grown on coverslips, cells were pre-extracted with 0.5% Triton X-100 in PBS for 40 s and then fixed with 4% paraformaldehyde in PBS for 15 min at room temperature. After fixation, cells were permeabilized with 0.5% Triton X-100 in PBS for 15 min and blocked with blocking buffer (4% BSA in PBS) for 1 h. The incubation with primary and secondary antibodies was performed at room temperature for 2 h and 1 h, respectively. 3D-SIM images were collected on the DeltaVision OMX V3 imaging system (GE Healthcare). The setting of the OMX V3 system was as described previously (Zhao et al., 2013). Confocal images were captured with the Leica TCS SP8 microsystem with a 100×/1.40 NA oil immersion objective at 400-nm intervals. Serial images were processed with maximum intensity projections.

Images of each grouped sample were captured under the same condition (gain, offset, exposure time, pixel size and laser settings). The fluorescence intensity and the diameters of the indicated proteins were measured with Photoshop (Adobe) and Image J (Fiji), respectively. Statistical results are presented as mean±s.d. Two-tailed Student's t-test (GraphPad Prism software) was used to calculate P-values between unpaired samples. Differences were considered significant when P<0.05.

The authors thank the Centre for Biological Imaging, Institute of Biophysics, CAS, for support on 3D-SIM imaging, and institutional core facilities for cell biology and molecular biology for instrumental and technical support.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.241836.reviewer-comments.pdf