Primary microcephaly (MCPH) is a neurodevelopmental disorder characterized by small brain size with mental retardation. CPAP (also known as CENPJ), a known microcephaly-associated gene, plays a key role in centriole biogenesis. Here, we generated a previously unreported conditional knockout allele in the mouse Cpap gene. Our results showed that conditional Cpap deletion in the central nervous system preferentially induces formation of monopolar spindles in radial glia progenitors (RGPs) at around embryonic day 14.5 and causes robust apoptosis that severely disrupts embryonic brains. Interestingly, microcephalic brains with reduced apoptosis are detected in conditional Cpap gene-deleted mice that lose only one allele of p53 (also known as Trp53), while simultaneous removal of p53 and Cpap rescues RGP death. Furthermore, Cpap deletion leads to cilia loss, RGP mislocalization, junctional integrity disruption, massive heterotopia and severe cerebellar hypoplasia. Together, these findings indicate that complete CPAP loss leads to severe and complex phenotypes in developing mouse brain, and provide new insights into the causes of MCPH.
The centrosome, which is composed of a pair of centrioles surrounded by a matrix of pericentriolar material (PCM), serves as the microtubule organization center (MTOC) in cells. The paired centrioles include a mother centriole with subdistal appendages (SDAs) and distal appendages (DAs) that regulate microtubule anchoring and cilia formation, and a daughter centriole that lacks any DAs or SDAs; this creates structural and functional asymmetry (Nigg and Stearns, 2011). Recent studies of the centrosomal core proteins have improved our understanding of the roles of centrioles/centrosomes in modulating cell polarity, functioning as MTOCs during cell divisions, and/or as basal bodies for flagellogenesis and ciliogenesis (Nigg and Holland, 2018; Gönczy, 2015; Conduit et al., 2015; Bettencourt-Dias et al., 2011). Previous elegant works identified at least four centrosomal core proteins (ZYG–1/Sak, SAS-4, SAS-5/Ana2 and SAS-6) that are essential for centriole duplication in Caenorhabditis elegans and Drosophila melanogaster (Basto et al., 2006; Delattre et al., 2004; Kirkham et al., 2003; Leidel et al., 2005; Leidel and Gönczy, 2003; O'Connell et al., 2001; Dammermann et al., 2004; Pelletier et al., 2006). The identified human functional homologs of ZYG-1, SAS-4, SAS-5, and SAS-6 are PLK4 (Habedanck et al., 2005; Bettencourt-Dias et al., 2005), CPAP (also known as CENPJ) (Hung et al., 2000; Bond et al., 2005), STIL (Stevens et al., 2010; Tang et al., 2011; Aplan et al., 1991; Kumar et al., 2009), and SAS-6 (also known as SSAS6) (Leidel et al., 2005), respectively. In human cells, centriole duplication begins with PLK4–STIL activation (Arquint et al., 2015; Ohta et al., 2014; Dzhindzhev et al., 2014; Klebba et al., 2015; Kratz et al., 2015; Moyer et al., 2015), which triggers the assembly of a SAS-6-containing cartwheel at the proximal end of a mother centriole. CPAP cooperates with other proteins, including CEP120 and SPICE1, to promote the assembly and elongation of a newborn (daughter) centriole during S/G2 phase (Lin et al., 2013c; Comartin et al., 2013). CEP120 then interacts with Talpid3 (Wu et al., 2014; Tsai et al., 2019) and C2CD3 (Tsai et al., 2019) to assemble centriole appendages, which are required for subsequent ciliogenesis.
Primary microcephaly (MCPH) is an autosomal recessive neurodevelopmental disorder characterized by smaller brain size with mild to severe mental retardation. Currently, at least 19 MCPH-associated genes, including RTTN, have been identified (Jayaraman et al., 2018; Shamseldin et al., 2015). Interestingly, mutations in the genes that encode the four centrosomal core proteins described above have been reported to cause MCPH in humans (Jayaraman et al., 2018), implying that centriole biogenesis plays a central role in MCPH (Chen et al., 2017; Tang et al., 2011; Lin et al., 2013a) via a yet-unidentified mechanism. Among the centrosomal core proteins, human CPAP was initially described as a centrosomal protein carrying tubulin dimer-binding and microtubule-interacting domains (Hsu et al., 2008; Hung et al., 2004). These domains are known to regulate centriole elongation and cilium assembly (Wu and Tang, 2012; Tang et al., 2009; Kohlmaier et al., 2009; Schmidt et al., 2009; Zheng et al., 2016). Studies in C. elegans have also found that SAS-4 is required for centriole duplication (Leidel and Gönczy, 2003), and that deficiency of SAS-4 in Drosophila prevents centriole duplication and disrupts flagellogenesis/ciliogenesis (Basto et al., 2006). SAS-4/CPAP also reportedly contributes to mitosis. RNAi knockdown experiments initially revealed that human CPAP plays a structural role in maintaining centrosome integrity and normal spindle morphology (Cho et al., 2006). Later studies further demonstrated that Aurora-A-phosphorylated CPAP functions to cohere the pericentriolar material (PCM) and maintains centrosomal integrity (Chou et al., 2016), while Cdk1-phosphorylated SAS-4/CPAP creates a Polo/PLK1-docking site that helps recruit Polo/PLK1 to daughter centrioles and is required to recruit PCM proteins to centrosomes (Novak et al., 2016; Ramani et al., 2018).
In the past decade, a Cpap/Cenpj/Sas-4 hypomorphic mouse line [Cenpjtm1a(EUCOMM)Wtsi] generated by inserting a gene-trap cassette between exons 4 and 5 of the Cenpj locus resulting in splicing over the cassette and cryptic splicing, which produces low levels of various abnormally spliced Cenpj mRNAs (McIntyre et al., 2012), has been established. This Cenpj hypomorphic mouse line was reported to recapitulate many of the clinical features of microcephalic primordial dwarfism disorder, which phenocopies human Seckel syndrome (McIntyre et al., 2012). Subsequently, a Cenpj-knockout allele (CenpjΔE5*, also designated as Sas-4−/−) was generated by Cre-mediated deletion of the floxed exon 5 in Cenpjtm1a(EUCOMM)Wtsi allele; mouse embryos carrying these homozygous Sas-4−/− mutations died at mid-gestation around embryonic day (E)8.5 and exhibited activation of p53 (also known as Trp53)-dependent apoptosis (Bazzi and Anderson, 2014). Further studies of Nestin-Cre-mediated conditional deletion (deletion in the central nervous system) of exon 5 in Cenpj mutant mice (hereinafter referred to as Sas-4 cKO) revealed that these mice recapitulated many human microcephaly phenotypes (Insolera et al., 2014), implying that cortical neurogenesis is sensitive to centrosomal dysfunction.
To better understand the correlation between CPAP loss and microcephaly, we herein generated a different mutant Cenpj mouse line (Lin et al., 2013b) that carries a previously undescribed Cpap conditional knockout allele (hereinafter referred to as Cpap cKO), with targeted deletion of exons 6 and 7 and introducing a frameshift stop codon in the CPAP protein (Fig. S1A,B). Our analyses revealed that loss of CPAP in the central nervous system inhibits centriole duplication, impairs cilia formation, produces severe mitotic abnormality in radial glial progenitors (RGPs) and induces p53-dependent neuronal cell death. We further observed that simultaneous removal of p53 can rescue this neuronal cell death, and that Cpap deletion leads to severe and complex phenotypes in developing brains. The functional implication of CPAP loss in the RGPs of developing mouse brains and their correlation with human primary microcephaly is discussed.
Loss of CPAP inhibits centriole duplication, impairs cilium formation, and induces p53-dependent cell death in the developing mouse brain
Our previous studies showed that the N-terminal region of CPAP carries a tubulin dimer-binding (PN2-3) and a microtubule-binding domain (A5N), and that these domains play crucial roles in regulating microtubule assembly (Hsu et al., 2008; Hung et al., 2004) and centriole elongation (Tang et al., 2009). Taking this advantage, we generated a new gene-targeted Cpap mouse line carrying a conditional knockout allele, Cpapflox (Fig. S1A), for Cre-mediated deletion of the loxP-flanked exons 6 and 7. This targeted deletion removes 1730 nucleotides from the 4035-nt ORF and creates a frameshift in the recombined CpapΔE6E7* allele (Fig. S1B).
Consistent with previous reports regarding homozygous Sas-4−/− mice with deletion of exon 5 (Bazzi and Anderson, 2014), our homozygous Cpap-knockout mice (CpapΔE6E7*/ΔE6E7*, hereinafter Cpap KO) showed reduction of fetal growth and growth arrest by E9.5, prior to embryonic turning (Lin et al., 2013b). Immunostaining of CPAP and the centrosomal protein, pericentrin (PCNT), revealed that mouse embryonic fibroblasts (MEF) obtained from Cpap KO embryos exhibited loss of CPAP-labeled centrioles (Fig. S1C, upper panel) and dispersion of centrosomal PCNT (Fig. S1C, lower panel).
To explore the roles of CPAP during cortical neurogenesis, we used Nestin-Cre (Tronche et al., 1999) to drive the deletion of Cpap in neural progenitors of the central nervous system (CNS) beginning at E10.5. Centrioles were clearly lost, as evidenced by severe reductions of CPAP (Fig. 1A) and the daughter centriole-enriched CEP120 (Mahjoub et al., 2010; Lin et al., 2013c) (Fig. 1B) in RGPs located at the ventricular zone of the cerebral cortex of Nestin-Cre-mediated Cpap conditional knockout embryos [hereafter Cpap cKO (Cpapflox/−;Nestin-Cre+)]. We also observed a dramatic reduction of ARL13B (a ciliary membrane marker) in the RGPs of Cpap cKO embryos (Fig. 1C), indicating that cilia assembly is impaired. By contrast, both centrioles (Fig. 1A) and cilia (Fig. 1C) were detected at the RGPs of wild-type control (CTL) and p53-knockout mice (p53 KO), implying that p53 loss does not affect centriole and cilium formation. Since previous studies have shown that centrosome loss could activate p53-dependent apoptosis (Bazzi and Anderson, 2014; Insolera et al., 2014; Lin et al., 2013b), we next examined whether loss of CPAP could induce neuronal cell death in Cpap cKO embryonic cortices. Similar to the previous reports (Insolera et al., 2014), we observed widespread apoptotic neuronal cell death (as evidenced by the signal for cleaved caspase 3) with upregulation of p53 in E14.5 Cpap cKO cortices, and this apoptosis could be effectively rescued by the absence of p53 in the p53; Cpap double knockout [hereafter dKO (p53−/−;Cpapflox/−;Nestin-Cre+)] cortex (Fig. 1D). However, in contrast to the normal bipolar spindles previously reported in Sas-4−/−;p53−/− conditional knockout embryonic cortices (Insolera et al., 2014), we frequently observed a high percentage of mitotic monopolar spindles (40–60%) in the RGPs of Cpap cKO and dKO mouse cortices (Fig. 1E).
Furthermore, unlike the microcephalic phenotype observed in Sas-4 cKO mice (Insolera et al., 2014), all of our Cpap cKO pups were born with cerebral hemorrhage and died shortly after birth. Histological analysis showed that most of the cerebral cortex was absent from these mice, as were other mitotically active regions in the midbrain and the hindbrain; this was due to extensive neuronal cell death, and only post-mitotic neurons remained in the ventral CNS by E18.5 (Fig. S2A). Interestingly, when only one of the two p53 alleles was removed from our Cpap cKO embryos [hereafter hcKO (p53+/−;Cpapflox/−;Nestin-Cre+)], the hcKO cortex showed fewer apoptotic cells compared to the cKO cortex (Fig. S2B; 25.5±9.5% in hcKO vs 42.1±8.9% in cKO; mean±s.d.). This resulted in microcephalic brains with reduced brain sizes observed in hcKO mice (Fig. S2A), similar to that reported for Sas-4 cKO mice (Insolera et al., 2014). Together, these results show that our Cpap conditional knockout allele produced more severe phenotypes than the Sas-4 conditional knockout allele used in the previous studies (Bazzi and Anderson, 2014; Insolera et al., 2014), and that p53 heterozygosity could reduce this severe p53-dependent apoptosis to yield microcephalic brains in hcKO (p53+/−;Cpapflox/−;Nestin-Cre+) mice.
Loss of CPAP in cultured neural progenitor cells
Cpap conditional knockout by developmentally regulated Nestin-Cre usually caused incomplete loss of centrioles/centrosomes and cilia at the apical end-feet of RGPs at E16.5 (Fig. 1A–C), possibly due to the gradual cell-cycle-dependent degradation of the CPAP protein (Tang et al., 2009). To better examine the effects of centriole dysfunction in the complete absence of CPAP protein, we isolated neural progenitor cells from embryonic cortexes (E12.5) derived from control (CTL, CTLCre), p53-KO, and dKO mice; we excluded Cpap cKO mice from this analysis because the p53+/+;Cpap−/− neural progenitors proved non-viable after serial passage. To avoid artificial transformation after long-term passage, we characterized the phenotypic features of isolated primary neural progenitors within 10 passages. The loss of full-length CPAP protein was confirmed by western blot analysis of dKO lysates using an anti-mouse (m)CPAP antibody (Fig. 2A) raised against the N-terminal 300 amino acids of mouse CPAP protein. Further analysis showed that no apparently truncated CPAP protein products were detected in the dKO lysates after prolonged exposure of the blot membranes (Fig. S1D), implying that the truncated CPAP proteins may be more susceptible to proteolytic degradation.
Immunofluorescence analysis showed that most of the dKO neural progenitors exhibited complete loss of CPAP-labeled centrioles (green, Fig. 2Biv,v), γ-tubulin-labeled centrosomes (red, Fig. 2Bv) and pericentrin-labeled centrosomes (red, Fig. 2Cv). However, a portion of the dKO neural progenitors (∼25%) exhibited dispersion of centrosomal proteins, including γ-tubulin (red, Fig. 2Biv) and pericentrin (red, Fig. 2Civ), and thus resembled cells treated with siRNA against CPAP showing dispersion of the pericentriolar material (PCM) (Chou et al., 2016). Further studies demonstrated that ARL13B-labeled cilia were completely lost from dKO neural progenitors (Fig. 2Civ,v) but not from neural progenitors of control (CTL, CTLCre) or p53-KO mice (Fig. 2Ci–iii), indicating that dKO neural progenitors failed to assemble cilia.
In contrast to the previous report of normal bipolar spindles in Sas-4 cKO mice (Insolera et al., 2014), monopolar spindles were commonly detected in the RGPs of our dKO embryonic cortices (∼55%; Fig. 1E). Consistent with this finding, neural progenitors isolated from E12.5 dKO embryonic cortex also exhibited multiple mitotic abnormalities with predominantly monopolar spindles (Fig. 3A). Together, our findings indicate that mitotic spindle defects are commonly observed in in vitro cultured dKO neural progenitors (Fig. 3A) and in vivo cortex of dKO embryonic brains (Fig. 1E).
We next examined whether treatment with centrinone (CN), which is a PLK4 inhibitor that was previously reported to prevent centriole duplication (Wong et al., 2015), could induce p53-dependent neural cell death similar to that found in the Cpap cKO embryonic cortex. Both immunostaining for cleaved caspase 3 (Fig. 3B) and a TUNEL assay (Fig. 3C) showed that CN-treated control (CTL) neural progenitors exhibited a time-enhanced neuronal cell death (Fig. 3B,C) that was correlated with increased p53 expression (Fig. 3D, left) and decreased centriole numbers (Fig. 3D, right). In contrast, p53-KO (Fig. 3B,C) or dKO (data not shown) neural progenitors exhibited few or no apoptotic cells under CN treatment. Together, our findings indicate that the survival of neural progenitors is sensitive to centriole loss, and that p53 can rescue the centriole-loss-induced apoptosis.
Cpap deletion leads to mislocalized RGPs, disrupted junction integrity and massive heterotopia in dKO embryonic cortex
Because the neuronal cell death of Cpap cKO mice could be effectively rescued by the concomitant deletion of p53 in dKO embryonic brains (Insolera et al., 2014; Fig. 1D, this report) and the removal of p53 does not seem to impair cortical development (Insolera et al., 2014), we next examined the cell fates of the rescued neural progenitors and the progeny neurons. During cortical neurogenesis, PAX6+ or SOX2+ RGP cells are found at the ventricular zone (VZ). Since cKO mice exhibit extensive neuronal cell death resulting in no visible cortex at E18.5 (Fig. S2A), we then examined the effects of CPAP loss in cKO and dKO embryonic cortices at E16.5. Consistent with a previous report (Insolera et al., 2014), Cpap deletion led to a decrease of RGPs in cKO mice (Fig. 4A) and delocalization of PAX6+/SOX2+ RGPs from the VZ (Fig. 4A) in dKO mice. A significant portion of mislocalized PAX6+ (Fig. 4A, upper panel; Fig. 4B) and SOX2+ (Fig. 4A, lower panel) RGPs were detected outside the VZ (extra VZ), suggesting that some (but not all) RGPs exhibiting various degrees of CPAP loss possibly due to the progressive action of Nestin-Cre lose their apical processes prematurely in the dKO embryonic cortex.
We next examined the identity of neurons of the extra VZ below the cortical plate of the E18.5 cortex. Compared with control mice, neurons in the extra VZ of dKO mice were predominantly composed of neurons positive for CUX1 (an upper-layer neuron marker) and CTIP2 (also known as BCL11B; a lower-layer neuron marker) (Fig. 4C). There were significantly more CUX1+ neurons in the dKO cortex, and these neurons formed massive bands of bilateral neuronal heterotopia in the E18.5 dKO cortex (Fig. 4C; Fig. S3A,B). Interestingly, the massive heterotopic bands of CUX1+ neurons were surrounded by residual mislocalized PAX6+ RGPs (Fig. S3B), suggesting that the mislocalized PAX6+ cells seen at E16.5 may contribute to the formation of the heterotopic cortex (HC) (Fig. 4B). Together, our results suggest that loss of CPAP in RGPs causes abnormal RGP (PAX6+ or SOX2+) localization, and that an excessive production of CUX1-expressing neurons, which predominantly occupy the HC, in the dKO embryonic cortex.
Recently, conditional disruption of the crucial apical polarity gene, Par3 (also known as Pard3; partitioning-defective 3), was shown to cause massive heterotopia in the developing mouse brain (Liu et al., 2018). Furthermore, disruption of N-cadherin, which plays a role in cell–cell contact at the apical junctions of RGPs, reportedly caused premature retraction of apical processes in apical progenitors (Imai et al., 2006; Das and Storey, 2014; Rousso et al., 2012). We therefore examined whether the loss of CPAP could affect the localizations of these apical proteins. As shown in Fig. 5, loss of CPAP in Cpap cKO or dKO embryonic brains significantly decreased the intensity and integrity of the lattice-like signals of PARD3 (Fig. 5A,C) and N-cadherin (Fig. 5B) at apical junctions. Collectively, our findings suggest that loss of CPAP in apical RGPs perturbs the localization of PARD3 and N-cadherin to the apical membrane surface (apical end-foot) and impairs junctional integrity at the ventricular surface.
An in vitro pair-cell assay shows increased neurogenesis among neural progenitors isolated from dKO embryonic brain
To better understand the effects of CPAP loss on the cell fate determination of neural progenitors, we used an in vitro pair-cell assay (Jensen and Parmar, 2006) to investigate the effect of CPAP loss on the intrinsic cell fate determination in neurosphere-derived progenitors. The cell pairs were immunostained for PAX6 (a neural progenitor marker) and TUJ1 (an early neuron marker) to identify the cell fates of daughter cells that were newly divided from neural progenitors (Fig. 6A). The percentages of symmetric versus asymmetric divisions in progenitor–progenitor pairs (P-P; symmetric division), progenitor–neuron (P-N pairs; asymmetric division) and neuron–neuron pairs (N-N; symmetric division) were quantified (Fig. 6B). We excluded Cpap cKO mice from this pair-cell assay because p53+/+;Cpap−/− neural progenitors were non-viable after serial passage. There was significant difference in the P-P, P-N, or N-N divisions between the p53 KO group and dKO group in this in vitro pair-cell assay (Fig. 6B). Over 80% of the cells in the p53 KO group underwent P-P division, while less than 10% cells showed P-N or N-N divisions. By contrast, the dKO progenitor cells showed significantly more P-N (∼16.5%) and N-N (∼17.2%) divisions (Fig. 6B). Together, our in vitro pair-cell assay results suggest that loss of CPAP in RGPs promotes premature neuronal differentiation.
p53;Cpap dKO mice show reduced postnatal viability, hydrocephalus and severe cerebellar hypoplasia
We found that p53 deletion could rescue neural progenitors from CPAP loss-induced apoptosis, but that most dKO pups failed to survive postnatally past weeks 3 or 4. Interestingly, the dKO pups suffered from ataxia-related complications, including loss of motor coordination and growth retardation associated with feeding difficulty. Histological examination for neurological defects in the CNS showed that dKO pups displayed severe cerebellar hypoplasia with no foliation (Fig. 7A) and deteriorating hydrocephalus (Fig. 7B). Further immunocytochemical analysis showed that cilia were absent from the cerebellum of dKO mice but not p53 KO mice (Fig. 8A). We detected only very few scattered PAX6+ granule cells in the underdeveloped cerebellum of dKO mice (Fig. 8B), suggesting that an almost complete loss of granule cells is responsible for the later absence of the internal granule cell layer (GCL) and the external granular layer (EGL). The lack of Ki67+ proliferating cells (Fig. 8B) and CUX1+ cells (Fig. 8C) further confirmed the loss of granule cell formation in the underdeveloped dKO cerebellum. Together, our results indicate that dKO mouse brains exhibit severe cerebellar hypoplasia and hydrocephalus.
A Cpap/Cenpj/Sas-4 hypomorphic mouse line was previously generated by inserting a gene-trap cassette between exon 4 and exon 5 of the Cpap locus (McIntyre et al., 2012), and Sas-4 cKO mice carrying a Nestin-Cre that targeted the 112-nt exon 5 were found to recapitulate many microcephaly phenotypes (Insolera et al., 2014). In the present study, we produced a new conditional Cpap knockout mouse line (Cpap cKO) with targeted deletion of exons 6 and 7, which resulted in producing out-of-frame products that are susceptible to proteolytic degradation (Fig. S1B,D). Compared to the Sas-4 cKO mice with targeting of exon 5 (Insolera et al., 2014), our Cpap cKO mice exhibited more severe effects in the developing brain, including robust neuronal apoptosis (Fig. 1D), loss of the whole cerebral cortex (Fig. S2A), increased mitotic abnormalities (e.g. monopolar mitotic spindles, Fig. 1E), and impaired junctional integrity of RGPs (Fig. 5). The last two phenotypes have not been reported in Sas-4 cKO mice. Furthermore, our dKO mice revealed poor development of the cerebellum (Fig. 7A) and severe hydrocephalus (Fig. 7B). The less severe phenotypes observed in the Sas-4 conditional exon 5 knockout (Sas-4 cKO) mice may reflect the generation of trace amounts of mutant CPAP proteins from ORFs that were restored by cryptic splicing (McIntyre et al., 2012). Further characterization of the phenotypic differences between various Cpap knockout alleles may help us understand the many aspects of centrosome dysfunction that are involved in the disease spectrum of centrosome-related microcephaly (O'Neill et al., 2018).
In contrast to our homozygous Cpap KO mice (Cpap−/−), which exhibited embryonic lethality, there are human patients with reported homozygous CPAP mutations; these individuals usually exhibit microcephalic brain (Bond et al., 2005; Gul et al., 2006) and primordial dwarfism (Seckel syndrome) (Al-Dosari et al., 2010). In considering why human MCPH patients can survive into adulthood, we hypothesize that partially functional CPAP proteins (hypomorphic, mutation-induced cryptic-spliced mutants or missense mutants) are present and alleviate clinical symptoms in these surviving patients. Our hypothesis is supported by a previous report that identified a homozygous CPAP/CENPJ splicing mutation (c.3302-1G>C) in a Saudi Arabian microcephalic family and showed that this mutation abolishes the consensus splice acceptor site, leading to the production of three different transcripts (Al-Dosari et al., 2010). The hypothesized retention of partial protein function may also apply to other microcephaly gene-encoded proteins (PLK4, STIL, RTTN, etc.) reported to critically participate in the key pathway of centriole biogenesis. Although Cpap deletion also induces p53-dependent neuronal cell death (Insolera et al., 2014; this report), we found that removal of only one allele of the p53 gene could ameliorate this neuronal cell death (hcKO, Fig. S2B) and produce microcephalic cortex (hcKO, Fig. S2A). Thus, a reduced dosage of p53 protein can attenuate p53-mediated apoptosis (Attardi and Jacks, 1999; Lemon et al., 2014) and decrease the neuronal cell death induced by CPAP loss (this report). We speculate that the severe clinical symptoms in human patients with reported MCPH-associated gene mutations could be attenuated by reduced p53 levels or a defective p53 gene. Future examination of both the p53 status and the mutant forms of various MCPH proteins in microcephalic patients will be needed to elucidate the correlation between clinical phenotypes and the functional roles of MCPH proteins.
We previously proposed that interfering with the production of intact and functional centrioles in neuronal progenitor cells is one of the major causes of microcephalic cortex in human MCPH patients (Tang et al., 2011; Lin et al., 2013a; Chen et al., 2017). Recent studies showed that several microcephaly proteins form separate complexes, including CEP152–CPAP (Cizmecioglu et al., 2010), CEP152–CEP63 (Sir et al., 2011), WDR62–CEP63–ASPM (Jayaraman et al., 2016), PLK4–STIL (Ohta et al., 2014), CPAP–STIL (Tang et al., 2011), CPAP–CEP135–SAS6 (Lin et al., 2013a) and RTTN–STIL (Chen et al., 2017), all converge into a common centriole biogenesis pathway to build an intact centriole. Intriguingly, we herein show that Cpap-removal-induced centriole loss not only impairs cilia formation (Fig. 1C), but also induces mitotic abnormalities, including monopolar spindles, multipolar spindles and abnormal bipolar spindles in RGPs (Fig. 1E). This suggests that RGP survival is very sensitive to defective centrioles.
Centrioles are microtubule-based organelles that recruit PCM to form the centrosome, which is the main MTOC in animal cells. The centrosome is essential for accurate chromosome segregation during mitosis, which is critical for the maintenance of genomic stability. Mitotic catastrophe was initially delineated as a type of cell death resulting from aberrant mitosis (Kroemer et al., 2009) and later functionally redefined as an apical mechanism that senses mitotic failure and drives the cell to apoptosis, necrosis, or senescence (Vitale et al., 2011). At least three different modes of mitotic catastrophe have been described to date: immediate mitotic death (cells experiencing aberrant mitosis die without exiting mitosis), subsequent G1 cell death (cells reach the G1 phase of the subsequent cell cycle and undergo cell death) and senescence (cells exit mitosis and undergo senescence) (Vitale et al., 2011). Our present results showed that complete loss of CPAP induces mitotic abnormalities and subsequent p53-dependent massive neuronal cell death (Fig. 1D). Intriguingly, the neuronal cell death induced by Cpap deletion seems to be mainly restricted to cells outside the VZ (Fig. 1D). We speculate that centriole loss due to Cpap removal may induce a mitotic catastrophe that drives neuronal cell death via a G1 cell death mode. In the future, it will be interesting to investigate the associated linkage between mitotic catastrophe executors [e.g. spindle assembly checkpoint (SAC) components and p53] and Cpap-deletion-induced neuronal cell death.
Previous studies have identified several key pathways that regulate the asymmetric/symmetric cell divisions of neural progenitor cells (NPCs) in worm and fly (Kemphues, 2000; Jan and Jan, 2001; Wodarz and Huttner, 2003; Knoblich, 2008; Fuerstenberg et al., 1998). Among the components of these pathways, the partitioning-defective (PARD) protein complex (PARD3, aPKC, etc.), which is abundantly expressed in the developing brain (Manabe et al., 2002), seems to be positioned at the top of the genetic regulatory hierarchy (Johnson and Wodarz, 2003). Recently, Liu and colleagues reported that PARD3 dysfunction in conjunction with dynamic HIPPO (and NOTCH) signaling in RGPs leads to massive heterotopia and excessive production of CUX1+ upper-layer neurons that predominantly occupy the heterotopia (Liu et al., 2018). Similar to PARD3 dysfunction, loss of CPAP perturbs the localization of PARD3 (Fig. 5A) and leads to massive heterotopia (Fig. 4C) containing CUX1+ neurons (Fig. 4C, Fig. S2A). Our findings suggest that loss of centrioles due to CPAP loss in RGPs could perturb PARD complex localization, induce premature neuronal differentiation and cause heterotopia formation. Interestingly, our in vitro pair-cell assay also showed that premature differentiation was increased in Cpap-deleted NPCs (Fig. 6A,B). Consistent with this finding, results with other MCPH gene-deleted mouse models also support this hypothesis (O'Neill et al., 2018). In the future, it will be interesting to investigate whether centriole loss or centriole defects in NPCs can alter the HIPPO signaling pathway in Cpap cKO and dKO mice.
Finally, one of the major functions of centrioles is to produce cilia in non-proliferating cells. It was reported that primary cilia are required for cerebellar development and Shh-dependent expansion of cerebellar granule cell precursors (GCPs) (Spassky et al., 2008). Interestingly, our immunocytochemical analysis showed that cilia were absent from the cerebellum of p53;Cpap dKO mice but not p53 KO mice (Fig. 8A), and that dKO mice exhibit severe cerebellar hypoplasia without foliation (Figs 7A and 8B) and deteriorating hydrocephalus (Fig. 7B). Because the cerebellar foliation in mice involves rapid and robust expansion of granule cells, which occurs mostly in the first 2 weeks after birth upon receipt of sonic hedgehog (SHH) signals secreted by Purkinje cells (Wechsler-Reya and Scott, 1999), it is likely that the severe cerebellar hypoplasia and the lack of foliation in dKO mice (Figs 7A and 8B) could be caused by the loss of cilia and cilia-mediated signaling (Fig. 8A). Furthermore, dysfunctional cilia were reported to lead to altered choroid plexus function, resulting in the formation of hydrocephalus (Banizs et al., 2005). The lack of cilia in NPCs/ependymal cells of dKO cerebral cortex due to Cpap deletion (Fig. 1B) could be a major factor contributing to the severe hydrocephalus seen in dKO mice.
MATERIALS AND METHODS
To generate Cpap conditional knockout allele (Cpapflox), the recombineering system (Liu et al., 2003) was used to insert a 5′-loxP-FRT-HindIII site at 241 bps upstream of the 103-bp exon 6 and a 3′-EcoRV-loxP-HindIII site at 175 bps downstream of the 1627-bp exon 7 by homologous recombination in C57BL/6 ES cells. After the Pgk-Neo cassette was removed by mating the generated mice with Flp-expressing mice, the resulting targeted Cpapflox allele was maintained on the C57BL/6 background. The Cre-recombined CpapΔE6E7* (or Cpap−/−) allele was generated by mating the above-described mice with Cre deleter mice to remove a 1730-bp genomic region of exon 6 and 7 that encodes the essential tubulin-dimer-binding and microtubule-binding domains of CPAP, yielding an exon 5-to-8 spliced transcript with a frameshift that creates nonsense stop codons (p.P314IfsX8 for the 1344-aa CPAP ORF). Nestin-Cre mice (#003771, Jackson Laboratories, Tronche et al., 1999) were used to drive recombination of loxP sites for Cpap conditional knockout in the central nervous system. To study the p53-dependent apoptotic pathway in Cpap knockout mice, we used complete p53 knockout mice (Trp53tm1Brd, Donehower et al., 1992) to constitutively inhibit p53 expression. Animals were genotyped and assigned to groups, and both male and female animals were used in the experiments. For comparative analysis with knockout animals, age-matched control littermates were used. All animal studies were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of Academia Sinica, Taiwan.
Primary embryonic neural progenitor culture
Primary embryonic neural progenitors were isolated from E12.5 mouse embryo neurospheres as previously described (Reynolds and Weiss, 1992). Briefly, after mice were euthanized, individual embryonic tail tips were collected for genotyping and embryonic cerebral cortices were dissected out for primary neurosphere culture. The embryonic cortex was digested enzymatically in Accutase (AT-104, Innovative Cell Technologies) at room temperature for 5 min. HBSS (Invitrogen) was added, the mixture was centrifuged at 160 g for 5 min, the cell pellets were resuspended, and the dissociated cells were seeded to a 24-well plate in a total 2 ml/well of complete medium (DMEM/F12; Invitrogen) with B27 (Invitrogen), 20 ng/ml bFGF (Preprotech), and 20 ng/ml EGF (Preprotech). Culture was performed in a humidified incubator with a 5% CO2 atmosphere at 37°C. Small floating spheroids (neurospheres) formed after ∼3 days, depending on the rate of cell growth. The medium was refreshed every 2 days and neurospheres were passaged every 7 days. Only primary cells with less than 10 passages were used for experiments, and neurospheres were split before reaching confluence. For centrinone treatment, progenitor cells were plated on poly-D-lysine-coated coverslips in complete medium, 125 nM centrinone or DMSO (control) was added to the medium, and the cultures were incubated for the indicated days.
Immunocytochemistry and confocal imaging
For immunohistochemistry, mouse brains were fixed overnight at 4°C with 4% paraformaldehyde in PBS. Cryosectioned samples were post-fixed in 4% paraformaldehyde in PBS for 20 min. Antigen retrieval was performed for both paraffin-embedded sections and cryosections using intermittent microwave boiling in 10 mM sodium citrate (pH 6.5) containing 0.05% Tween 20 for 30–40 min. Blocking was performed with 10% BSA in PBS blocking buffer containing 0.01% Triton X-100, and immunostaining was performed with antibodies in blocking buffer. Mouse neurosphere-derived progenitors and MEF cells grown on coverslips were fixed in methanol and incubated with the indicated primary antibodies. After washing, the cells were incubated with secondary antibodies conjugated with Alexa Fluor 488, Alexa Fluor 568, or Alexa Fluor 647 (1:500 dilution, Invitrogen), and DNA was counterstained with 4′,6-diamidino-2-phenylindole (DAPI). In TUNEL experiments, cells were incubated with in situ cell death detection reagents (Roche) before incubation with secondary antibodies. The samples were mounted in Vectashield mounting medium (Vector Laboratories) and visualized using a LSM700 system or a LSM880 Airyscan system (Carl Zeiss).
To generate the rabbit anti-mCPAP antibody (1:250 dilution), recombinant His-tagged mouse CPAP protein (residues 1–300) was used as the immunogen, and the antibody was affinity purified using recombinant CPAP protein immobilized on PVDF membranes. The antibodies against CEP120 (1:500) and NuMA (1:250 dilution) were generated as described previously (Lin et al., 2013c; Tang et al., 2004). The commercially available antibodies used in this study were: mouse anti-pericentrin (611814, BD Bioscience, 1:500), mouse anti-γ-Tubulin (T6793, Sigma-Aldrich, 1:500), rabbit anti-ARL13B (17711-1-AP, Proteintech, 1:250), rabbit anti-PAX6 (PRB-278P, Covance, 1:250), mouse anti-TUJ1 (MMS-435P, Covance, 1:1000), rabbit anti-Ki67 (ab16667, Abcam, 1:250), rabbit anti-cleaved Caspase3-Asp175 (5A1E, 9664, Cell Signaling, 1:250), rabbit anti-p53 (CM5, NCL-p53-CM5p, Leica Biosystems, 1:100), rat anti-CTIP2 (ab18465, Abcam, 1:250), anti-CUX1 (sc-13024, Santa Cruz, 1:100), and mouse anti-Centrin3 (H0000170-M01, Abnova, 1:250). Brain samples were fixed overnight at 4°C in 4% paraformaldehyde/PBS for subsequent processing.
To examine the levels of endogenous CPAP protein, mouse neurospheres were lysed in cell lysis reagent (C2978, Sigma) with protease inhibitor cocktails (P1860, Sigma), 100 mM NaF, 2.5 mM Na3VO4, 1 µg/µl leupeptin, 1 µg/µl pepstatin and 1 µg/µl aprotinin for 30 min on ice. The cell lysates were centrifuged at 14,000 g at 4°C for 15 min, and the supernatants were analyzed by SDS-PAGE.
In vitro pair-cell assay
The in vitro pair-cell assay was performed as previously described (Shaker et al., 2015; Shen et al., 2002). Briefly, cells were plated on poly-D-lysine-coated coverslips at clonal density in serum-free medium [DMEM (Invitrogen) containing L-glutamate (Sigma), sodium pyruvate (Sigma), 1 mM N-acetyl-L-cysteine (Sigma), B27 (Invitrogen), N2 (Invitrogen) and 10 ng/ml bFGF (Preprotech)]. Plated cells were cultured for 24 h, rinsed twice with PBS, fixed in 4% paraformaldehyde, and immunostained with antibodies against protein markers of neural progenitors and neurons.
For quantification analysis, cells positively stained by the corresponding antibodies were counted individually or in a columnar area of width described in the figure legends. At least three animals (n≥3) or at least three batches of cultured progenitor cells in each group with the actual numbers described in the figures and figure legends were analyzed. Statistical analyses were performed using GraphPad Prism 8, and the data are presented as the mean±standard derivation (s.d.) from at least three independent experiments. Statistical differences between two data sets were analyzed using a two-tailed unpaired Student's t-test (*P<0.05, **P<0.01, ***P<0.001, n.s., not significant).
We thank Dr Yu-Ting Yan (Institute of Biomedical Sciences, Academia Sinica), Dr Shen-Ju Chou (Institute of Cellular and Organismic Biology, Academia Sinica), and Dr Li-Huei Tsai (Massachusetts Institute of Technology) for advice on experimental design, and Dr Hsin-Fang Yang-Yen (Institute of Molecular Biology, Academia Sinica) for p53 knockout mice. We thank lab members Dr Yuan-Chang Hsu and Dr Kuo-Tai Yang for testing experimental conditions not included in this study. We also thank the sequencing core facility (IBMS, AS-CFII-108-115), the confocal imaging core facilities (IBMS, NPAS, ABRC, AS-CFII-108-116), the Transgenic Core Facility (Institute of Molecular Biology, Academia Sinica) for the generation of targeted Cpap mutant mice, and the Animal Facility (Institute of Biomedical Sciences, Academia Sinica) for mouse husbandry support.
Conceptualization: Y.-N.L., Y.-S.L., T.K.T.; Methodology: Y.-N.L., Y.-S.L., S.-K.L.; Validation: Y.-N.L., Y.-S.L., S.-K.L., T.K.T.; Formal analysis: Y.-N.L., Y.-S.L., S.-K.L., T.K.T.; Investigation: Y.-N.L., Y.-S.L., S.-K.L., T.K.T.; Data curation: Y.-N.L., Y.-S.L., S.-K.L., T.K.T.; Writing - original draft: Y.-N.L., T.K.T.; Writing - review & editing: T.K.T.; Visualization: T.K.T.; Supervision: T.K.T.; Funding acquisition: T.K.T.
This work was supported by grants from Academia Sinica (AS-IA-104-L01, AS-IA-109-L04, and AS-TP-108-L08) and the Ministry of Science and Technology, Taiwan (MOST 108-2321-B001-026).
Peer review history
The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.243592.reviewer-comments.pdf
The authors declare no competing or financial interests.