|Home | About | Journals | Submit | Contact Us | Français|
Centrosome functions are important in multiple brain developmental processes. Proper functioning of the centrosome relies on assembly of protein components into the pericentriolar material. This dynamic assembly is mediated by the trafficking of pericentriolar satellites, which are comprised of centrosomal proteins. Here we demonstrate that trafficking of pericentriolar satellites requires the interaction between Hook3 and Pericentriolar Material 1 (PCM1). Hook3, previously shown to link the centrosome and the nucleus in C. elegans, is recruited to pericentriolar satellites through interaction with PCM1, a protein associated with schizophrenia. Disruption of the Hook3-PCM1 interaction in vivo impairs interkinetic nuclear migration, a featured behavior of embryonic neural progenitors. This in turn leads to overproduction of neurons and premature depletion of the neural progenitor pool in the developing neocortex. These results underscore the importance of centrosomal assembly in neurogenesis, and provide potential insights into the etiology of brain developmental diseases related to the centrosome dysfunction.
Centrosome functions are important for multiple processes during embryonic brain development, including neuronal migration, neural polarity establishment, and neurogenesis (Badano et al., 2005; Higginbotham and Gleeson, 2007). Therefore, understanding the formation and function of the centrosome in embryonic brains will provide helpful insights into the regulation of these brain developmental processes. Electron microscopy studies revealed that the centrosome consists of a pair of centrioles and the pericentriolar material. The latter is a meshwork of various proteins intertlaced into a well-organized matrix (Bettencourt-Dias and Glover, 2007; Doxsey, 2001b). Most centrosomal functions are thought to be carried out in the pericentriolar material, including microtubule nucleation, microtubule anchorage, and mitotic spindle organization (Luders and Stearns, 2007; Rieder et al., 2001). The pericentriolar material is structurally very dynamic. Individual constituent proteins, such as centrin, pericentrin and ninein, undergo vigorous exchange between the centrosome-bound pool and a larger cytoplasmic pool (Doxsey, 2001a; Young et al., 2000). Consistent with this observation, many pericentriolar material components are not confined to centrioles, but are distributed in the cytoplasm and the nucleus (Baron et al., 1994; Paoletti et al., 1996). Furthermore, Fluorescence Recovery After Photobleaching (FRAP) experiments revealed rapid turnover of multiple centrosomal proteins (Khodjakov and Rieder, 1999; Valente et al., 2006). Maintenance of the structural dynamics depends on pericentriolar satellites, non-membranous granules with a diameter of 70-100nm, where were first observed as electron-dense vesicles by electron microscopy (Berns et al., 1977). Pericentriolar satellites shuttle between the centrosome and the cytoplasm, and may mediate the transport of proteins destined for the pericentriolar material (Kubo et al., 1999; Zimmerman and Doxsey, 2000).
The first identified molecular component of pericentriolar satellites was Pericentriolar Material 1 (PCM1) (Balczon et al., 1994; Baron and Salisbury, 1988). PCM1 serves as a scaffold protein and recruits various centrosomal proteins to pericentriolar satellites (Hames et al., 2005), which are then linked to dynein motors via the Bardet–Biedl syndrome 4 protein (Kim et al., 2004). Dynein motor complexes then transport pericentriolar satellites along microtubules towards the centrosome (Kubo et al., 1999). Compromising PCM1 function impairs the proper functioning of pericentriolar satellites, resulting in a reduction of protein integration into the centrosome (Dammermann and Merdes, 2002). Recently, PCM1 was discovered to be highly associated with schizophrenia in human genetic studies (Datta et al., 2008; Gurling et al., 2006; Kamiya et al., 2008). Furthermore, it interacts with and functions at the centrosome in cooperation with DISC1, another risk gene of schizophrenia (Kamiya et al., 2008). Although accumulating data suggest that brain developmental disorders, in particular, defects in cortical neural progenitor proliferation, contribute to the onset of this psychiatric disorder (Mao et al., 2009; Ross et al., 2006), whether PCM1 is involved in embryonic brain development is largely unknown.
In the developing neocortex, neural progenitor cells (radial glia and neuroepithelial cells) proliferate to establish a pool from which all neurons are generated. These neural progenitors reside in the ventricular zone (VZ), a pseudostratified columnar epithelium that lines the lateral ventricle. The neural progenitor pool replenishes itself during development, and this self-replenishment must be balanced with neurogenesis in order to ensure that a precise number of neurons are generated. Disrupting this balance results in severe neurological and neuropsychiatric disorders, ranging from microcephaly to bipolar disorder and schizophrenia (Badano et al., 2005; Mao et al., 2009). Interestingly, nuclei of neural progenitors oscillate within the VZ in correlation with cell cycle progression in a process termed interkinetic nuclear migration (INM) (Sauer, 1935). During INM, neural progenitors undergo mitosis along the ventricular surface. After mitosis, nuclei ascend to the basal end of the VZ, where cells duplicate their DNA during S phase. Subsequently, nuclei descend towards the apical side of the VZ and cells undergo mitosis (Sauer, 1935). INM has been suggested to regulate the balance between neurogenesis and maintenance of the neural progenitor pool by controlling the exposure of progenitor cells to neurogenic versus proliferative signals (Del Bene et al., 2008; Murciano et al., 2002). INM is regulated by microtubules, actin, and microtubule-associated proteins (Messier and Auclair, 1974; Tsai et al., 2005; Webster and Langman, 1978), and the centrosomal proteins Cep120 and TACC regulate INM by controlling the length of microtubules that attach to the centrosome (Xie et al., 2007). This underscores an essential function of the centrosome in the regulation of neural progenitor proliferation.
In embryonic neural progenitors, the centrosome resides at the ventricular surface when the nucleus undergoes ascending and descending movements during INM. PCM1-containing pericentriolar satellites are found at the apical surface of various epithelial cells, including neuroepithelia in the developing neocortex (Cohen et al., 1988; Kubo et al., 1999). However, it remains unknown whether PCM1-mediated centrosomal assembly is involved in cortical neurogenesis. Here, we demonstrate that pericentriolar satellite-mediated protein trafficking is essential for maintenance of proper centrosome function in neural progenitors. We found that trafficking of pericentriolar satellites relies on the interaction between PCM1 and Hook3. Hook3 is the mammalian homolog of C. elegans ZYG-12 that links the centrosome and the nucleus (Malone et al., 2003). In mammals, Hook3 is recruited to pericentriolar satellites through an interaction with PCM1.
Disrupting this interaction impairs trafficking of pericentriolar satellites and subsequently reduces protein assembly at the centrosome. This in turn compromises centrosome functions, including its ability to organize the microtubule cytoskeleton. Finally, silencing Hook3 or PCM1, or blocking the Hook3-PCM1 interaction in vivo, impairs INM, resulting in an over-production of neurons at the expense of neural progenitor pool. Our findings have broad implications in the understanding of centrosomal protein assembly during embryonic neurogenesis. Furthermore, it provides potential insight into the development of centrosome-related neural developmental disorders.
To evaluate centrosomal function during cortical development, we initiated our studies with Hook3, which has been reported to associate with the centrosome in C. elegans. Consistent with a role in neurogenesis, Hook3 is expressed in the developing mouse neocortex concurrent with neurogenesis (E11-E18, Figure 1A). Immunostaining of E14 cortical slices showed that Hook3 expression was relatively high in the VZ, where it overlapped with the neural progenitor marker nestin (Figure 1B). In particular, Hook3 fluorescence highlighted the apical edge of the VZ, where centrosomes of neural progenitors reside. Upon closer examination of the VZ apical surface, we found that Hook3 fluorescence covered the centrosome per se and was also distributed to the centrosome periphery (Figure 1C, D, en face view). To further elucidate the subcellular localization of Hook3, we examined Hook3 localization in a neuroblastoma cell line (N2A). Hook3 fluorescence appeared as small granules concentrated at the centrosome, but was also distributed throughout the cytoplasm (Figure 1E). This subcellular localization was also observed in primary cultured neural progenitors (Figure S1A). The nature of these granular structures will be addressed later.
To study the function of Hook3, we developed two RNAi constructs, Hook3RNAi-1 and -2, to silence Hook3 expression. Both constructs efficiently knocked down endogenous Hook3 in N2A cells (Figure 2A, B) and in the mouse cerebral cortex (Figure 2C). We then introduced Hook3 RNAi into E11.5 mouse embryonic brains together with Venus (a variant of EYFP) by in utero electroporation, a technique that allows analysis of acute effects of silencing genes of interest, and examined the brains at E14.5.
As Hook3 is highly expressed in neural progenitors, we first examined the division of progenitors cells in the VZ by staining with phospho-histone H3 (PH3) antibody (Figure 2D). While most cells divide on the ventricular surface in control brain sections (Figure 2D, arrows), a significantly higher percentage of cells divide away from the ventricular surface after Hook3 knockdown (Figure 2D, E, arrowheads). We define apical division as mitoses that occur within 20μm of the ventricular surface (Miyama et al., 2001). Ectopic mitoses were rescued by co-electroporation with RNAi-resistant full-length Hook3 (Figure 2D). The observed ectopic mitoses suggest that INM is uncoupled from cell cycle progression, as reported by previous studies (Tsai et al., 2005; Xie et al., 2007).
We next sought to identify the responsible factor that elicited the uncoupling. We first examined cell cycle progression by FACS analysis, but did not find any abnormalities (Figure S2). Another possibility is that the phenotype is caused by an impairment in INM. To address this possibility, we first examined the morphology of transfected nuclei. As Sauer noticed in his original study, nuclei assume an elongated morphology along the apical-basal axis during INM. When INM halts or is impaired, nuclei exhibit rounded shapes (Chenn and McConnell, 1995; Sauer, 1935). To evaluate this, we measured the length and width of nuclei of transfected cells, and found that the length to width ratio was significantly smaller after Hook3 knockdown, suggesting that nuclei assume a less elongated morphology. This phenotype was rescued by RNAi-resistant Hook3 (Figure 2F, G). These results indicate that INM might be impaired after Hook3 knockdown.
To directly evaluate Hook3 function in INM, we performed time-lapse imaging on acute cortical slices two days after in utero electroporation (Figure 2I). In the VZ, newborn neurons migrate upward (away from the ventricular surface), with their cell bodies amidst cell bodies of neural progenitors undergoing INM. Therefore, we focused on the descending movement (toward the ventricular surface) of transfected cells in order to exclude migrating neuron from our analysis. During the 8 hour imaging period, most cell bodies of control cells traversed long distances (Figure 2H,J; movie S1), and some cells divided at the ventricular surface (Figure 2H, arrow). In contrast, cell bodies of most Hook3 RNAi transfected cells oscillated at their original positions (Figure 2H,J; movie S2) and divided away from the ventricular surface (Figure 2H, arrowhead). The reduced nuclear motility of Hook3 knockdown cells was restored by co-expressing RNAi-resistant Hook3 (Figure 2H,J; movie S3). Taken together, we conclude that Hook3 knockdown impairs INM.
We next examined effects of Hook3 RNAi-mediated INM disruption on neurogenesis. At E14, neural progenitors predominantly undergo N-P divisions, in which one daughter cell differentiates into a neuron and the other remains a neural progenitor. In this way, neurons are continuously generated while the neural progenitor pool is maintained. To evaluate cell fate after ectopic divisions, we performed the cell cycle exit assay. Embryos were electroporated at E11.5, pulse-labeled with BrdU at E13.5, and analyzed at E14.5 by immunostaining for BrdU and Ki67 (labels cells in late G1, S, G2 and M-phase). We found that cells expressing Hook3 RNAi exhibited a significantly higher cell-cycle exit index compared to control cells (Figure 3A, B). In addition, transfected cells in Hook3 knockdown sections showed a significantly lower BrdU labeling index, which suggests a diminished size of the neural progenitor pool (Figure 3C).
In line with the increased cell cycle exit, neurogenesis significantly increased after Hook3 knockdown. This is first evidenced by the higher percentage of cells migrating to the CP/IZ, and lower percentage remaining in the VZ/SVZ (Figure 3D, E). Second, immunostaining of brain sections with neuron-specific markers Tuj1 and NeuN revealed that neurogenesis increased after Hook3 knockdown (Figure 3D, F; Figure S3D). Taken together, these data suggest that as a result of Hook3 knockdown-mediated INM disruption, cells differentiate into neurons after ectopic cell division, leading to premature depletion of the neural progenitor pool and overproduction of neurons.
To elucidate the molecular mechanism by which Hook3 is involved in INM, we performed a yeast-two-hybrid screen for Hook3 interacting proteins. One potentially interesting interactor was PCM1 (Figure 4A), given that its subcellular localization resembles that of Hook3 in previous studies (Balczon et al., 1994; Baron and Salisbury, 1988; Kubo and Tsukita, 2003). Co-immunoprecipitation from E14 mouse brain lysates confirmed the interaction between PCM1 and Hook3 in vivo (Figure 4B). We then examined the subcellular localization of both proteins by over-expressing EGFP-Hook3 in N2A cells and staining with PCM1 antibody (Figure 4C,D). As previously reported (Kubo et al., 1999; Kubo and Tsukita, 2003), PCM1 immunofluorescence appears as granules concentrated at the centrosome, with a small proportion scattered in the cytoplasm. These granules overlapped with EGFP-Hook3, particularly in the vicinity of the centrosome (Figure 4D). Furthermore, we confirmed the colocalization of the two proteins by immuno-staining using Biotin-conjugated Hook3 antibody, followed by PCM1 staining (Figure S1C-C’). In the developing cortex, PCM1 immunofluorescence concentrates at the ventricular surface, where it overlaps with and surrounds the centrosome (Figure 4E-F). These results suggest that PCM1 might be a component of the molecular machinery containing Hook3 that regulates neurogenesis.
As PCM1 is a component of pericentriolar satellites (Baron and Salisbury, 1988), its co-localization with Hook3 suggests that Hook3 might also be associated with pericentriolar satellites. To test this possibility, we examined the ultrastructural localization of Hook3 by immunogold electron microscopy. We used PCM1 antibody in parallel with Hook3 antibody, and found that both antibodies target gold particles to electron-dense granules (approximately 100nm in diameter), whose size and morphology resemble pericentriolar satellites (Figure 4G, arrows). Taken together, these data suggest that Hook3 interacts with PCM1 and both proteins are components of pericentriolar satellites.
To identify the interaction domain of Hook3 with PCM1, we used different Hook3 fragments (Table in Figure 4A) to perform yeast-two-hybrid interaction assays. We found that the C-terminus of Hook3 (Hook3-C) mediated the binding of the two proteins. To confirm this interaction, we over-expressed flag-tagged Hook3 fragments and GFP-PCM1 in N2A cells, and found that only Hook3-C co-immunoprecipitated with PCM1 (Figure 5A). Furthermore, the amount of PCM1 that co-immunoprecipitated with Hook3 decreased as the expression of Hook3-C increased (Figure 5B). These results indicate that Hook3-C inhibits the interaction between PCM1 and Hook3 in a dominant negative manner.
As PCM1 may function as a scaffold for pericentriolar satellites (Dammermann and Merdes, 2002; Hames et al., 2005; Kubo and Tsukita, 2003), we investigated the significance of the Hook3-PCM1 interaction on trafficking of pericentriolar satellites. To this end, we transfected N2A cells with control RNAi, Hook3 RNAi, or Hook3-C, and stained cells with PCM1 antibody. PCM1 is used here as a marker for pericentriolar satellites. In Hook3 RNAi and Hook3-C expressing cells, PCM1-containing granules were dispersed from the centrosomal periphery (Figure 5D, E). We confirmed that total PCM1 levels were unchanged in Hook3 RNAi and Hook3-C expressing cells to exclude the possibility that the loss of centrosomal localization of PCM1 might be due to PCM1 degradation (Figure 5G). Furthermore, PCM1 RNAi dispersed Hook3-containing granules from the centrosomal periphery (Figure 5F). The characterization of PCM1 RNAi is shown in supplemental Figure 3A-C. These results suggest that the Hook3-PCM1 interaction is required for trafficking of pericentriolar satellites towards the centrosome.
Further evidence was obtained by electron microscopy. Transfected cells were sorted by flow cytometry prior to fixation and processed for ultra-thin sectioning. We then collected all sections containing the centrosome and quantified the number of pericentriolar satellites within a square of 16μm2 surrounding the centrosome (with the centrosome in the center; arrows in Figure 5H;). The high magnification of pericentriolar satellites around the centrosome is shown in Figure 5I. Significantly less pericentriolar satellites were found surrounding the centrosome in Hook3 knockdown cells. A similar phenotype was observed with PCM1 RNAi and Hook3-C over-expression (Figure 5H, J). In summary, these data suggest that disrupting the Hook3-PCM1 interaction impairs the proper functioning of pericentriolar satellites, either by disrupting their assembly or blocking their transport to the centrosome.
As mentioned above, pericentriolar satellites mediate the assembly of centrosomal proteins (Dammermann and Merdes, 2002). Hence, we tested whether disrupting the Hook3-PCM1 interaction would affect centrosomal protein assembly. We chose to examine the assembly of pericentrin, CDK5Rap2, and ninein, proteins with well-characterized centrosomal functions (Bouckson-Castaing et al., 1996; Doxsey, 2001a). Compared to the control, N2A cells transfected with Hook3 RNAi, PCM1 RNAi, or Hook3-C displayed reduced centrosomal localization of these three proteins (Figure 6A, B).
To further establish the dynamic assembly of these centrosomal proteins, we performed FRAP analysis using CDK5Rap2 as a representative centrosomal protein (Hames et al., 2005; Thompson et al., 2004). In the control, centrosomal venus-CDK5Rap2 recovered to 73% of its original level within 3 minutes. In contrast, Hook3 and PCM1 knockdown significantly reduced the extent of recovery (Figure 6C, D). The minor recovery observed after Hook3 or PCM1 knockdown is likely due to diffusion of proteins from the neighboring cytoplasmic area. These results strongly indicate that Hook3 and PCM1 are essential for the dynamic assembly of centrosomal proteins.
Disruption of the dynamic assembly of pericentriolar material may result in abnormalities in various centrosomal functions. Here, we examined the most well-established functions of the centrosome: microtubule nucleation and anchorage. We chose NIH3T3 given their higher tolerance to nocodazole treatment compared to N2A cells. After transfection with Hook3 RNAi, PCM1 RNAi, or Hook3-C, NIH3T3 cells were treated with nocodazole and allowed to recover after drug washout. Nocodazole depolymerized microtubules and disrupted the trafficking of pericentriolar satellites, resulting in a dispersion of Hook3 and PCM1 from the centrosome (Figure S4A, B). After a 5 min recovery period, most transfected cells (white stars) in both the control and RNAi treated group showed a clear microtubule aster, indicating that initial microtubule nucleation was not affected (Figure 6E, G, Figure S4C). However, after 20 min of recovery, control cells contained clear microtubule asters organized radially from the centrosome (arrowheads), whereas microtubule asters in most Hook3 and PCM1 RNAi and Hook3-C transfected cells were lost (Figure 6F,G, Figure S4D). Taken together, these data suggest that after compromising the function of pericentriolar satellites, microtubule anchorage at the centrosome is impaired, likely due to insufficient targeting of microtubule anchorage proteins to pericentriolar material.
We next investigated the physiological role of the Hook3-PCM1 interaction in cortical development. We employed two RNAi constructs to silence endogenous PCM1 expression (Figure S3A-C). These RNAi constructs and the dominant negative Hook3-C construct were then introduced into E11.5 embryonic brains by in utero electroporation, and brains were examined at E14.5. PH3 immunostaining revealed a significantly higher percentage of ectopic mitoses compared to the control (Figure 7A-B). In addition, the ratio of length to width of transfected nuclei was significantly lower (Figure S5A, B). These results suggest that similar to Hook3 knockdown, PCM1 knockdown or disrupting the Hook3-PCM1 interaction impairs INM.
To directly assess involvement of the Hook3-PCM1 interaction in INM, we performed time-lapse imaging of E13.5 acute brain slices. We found that the long distance movement of cell bodies was significantly reduced by PCM1 knockdown or Hook3-C over-expression (Figure S5C, D; movie S5, S6). Taken together, these data suggest that the Hook3-PCM1 interaction is required for neural progenitors to carry out INM.
We next examined involvement of the Hook3-PCM1 interaction in embryonic neurogenesis. The cell cycle exit index was significantly higher and BrdU incorporation lower in brains electroporated with PCM1 RNAi or Hook3-C compared to the control (Figure 7C, D, E). Furthermore, the percentage of cells remaining in the SVZ and VZ was significantly lower; correspondingly, the percentage of cells that migrated to the IZ and CP was significantly higher (Figure S5E, F). In addition, the percentage of Tuj1-posivite or NeuN-positive cells increased, indicating a higher percentage of neurons in the cortex compared to the control (Figure S3, S5G). Taken together these data suggest that following disruption of the PCM1-Hook3 interaction, INM is impaired, leading to a temporary overproduction of neurons and premature depletion of the neural progenitor pool.
Neural progenitors in the VZ give rise to intermediate neural progenitors (INPs) that leave the VZ and undergo terminal divisions in the SVZ or IZ, producing two neurons (Noctor et al., 2004). One possibility consistent with our data is that over-proliferation of INPs due to Hook3 or PCM1 knockdown results in the temporary over-production of neurons. We tested this possibility by staining brain sections with Tbr2, a marker for INPs. Compared to the control, the percentage of Tbr2-positive cells was not significantly different after Hook3 or PCM1 knockdown, or after disrupting the interaction between these two proteins (Figure S6). Hence, the overproduction of neurons is not due to the over-proliferation of INPs.
Neural progenitors in the VZ are highly polarized. They anchor to each other at the apical end-feet on the ventricular surface through adherens junctions, which is important for maintenance of the neurogenic niche. We carefully examined the polarity and adhesion of neural progenitors by immunostaining brain sections with the adherens junction marker N-cadherin and F-actin (Figure S7). We did not find any obvious abnormalities after Hook3 or PCM1 knockdown, or after disrupting the interaction between the two proteins.
Various centrosomal proteins have been established to be important factors in normal embryonic brain development. Mutations in genes encoding these proteins cause severe developmental disorders that lead to CNS diseases, such as lissencephaly (Olson and Walsh, 2002), microcephaly (Badano et al., 2005), and schizophrenia (Mackie et al., 2007). Most of these proteins localize to the pericentriolar material, which underscores the importance of the pericentriolar material assembly in normal centrosome function. In this study, we examined the significance of pericentriolar material assembly in embryonic brain development and found that Hook3 and PCM1 are required for trafficking of pericentriolar satellites. Blocking the Hook3-PCM1 interaction abolished the dynamic assembly of centrosomal proteins, resulting in compromised centrosomal functions. This in turn impaired INM, leading to an overproduction of neurons at the expense of the neural progenitor pool. Our studies underscore the essential function of pericentiolar satellites in embryonic neurogenesis, and provide novel insight into the etiology of centrosome-related neurodevelopmental diseases.
Pericentriolar satellite-mediated protein transport is essential for the dynamic assembly of centrosomal proteins (Dammermann and Merdes, 2002; Hames et al., 2005; Kubo et al., 1999). In this study, we showed that Hook3 localizes to pericentriolar satellites through an interaction with PCM1 (Figure 4). Abolishing the Hook3-PCM1 interaction compromised the dynamic assembly of centrosomal components (Figure 5), which led to detachment of microtubules from the centrosome (Figure 6) and subsequent INM impairment. Based on these data, we propose a model for the dynamic assembly of pericentriolar material during neurogenesis (Figure 7F). In this model, PCM1 functions as a scaffold that recruits multiple centrosome proteins, including Hook3, to pericentriolar satellites. Pericentriolar satellites are then transported to the centrosome along microtubules in a dynein-dependent manner. When the Hook3-PCM1 interaction is disrupted, either the assembly or transport of pericentriolar satellites is impaired, leading to decreased protein assembly into the centrosome. Consequently, microtubules detach from the centrosome due to the lack of microtubule anchorage proteins, resulting in the disruption of nuclear movement during INM. In support of this model, we found that abolishing the Hook3-PCM1 interaction impaired INM (Figure (Figure7,7, S4).
Our immunostaining and EM data showed decreased localization of pericentriolar satellites around the centrosome after Hook3 knockdown (Fig 5D, 5H). This phenomenon may be attributed to defects in the assembly of pericentriolar satellites at the cell periphery or in their subsequent transport towards the centrosome. Although in the current study we can not unambiguously define whether Hook3 interacts directly with PCM1, we attempted to elucidate the interaction of Hook3 and PCM1 by epistatic analysis. We overexpressed one protein in the context of silencing the other, and examined the distribution of pericentriolar satellites in N2a cells, as well as INM and neuronal production in developing cortices. We did not find any significant difference between single gene knockdown and overexpressing one protein in the context of silencing the other (Data not shown). These data suggest that overexpressing either Hook3 or PCM1 in the context of silencing the other gene does not rescue the defects in the trafficking of pericentriolar satellites. We speculate that while PCM1 functions as a scaffold for pericentriolar satellites, Hook3 may mainly function in the transport step, but its recruitment to the pericentriolar satellite relies on PCM1.
Disruption of INM by in vivo knockdown of Hook3 or PCM1 resulted in mitoses away from the ventricular surface. After this ectopic division, both daughter cells differentiate into neurons. This results in premature depletion of the neural progenitor pool, which is normally used up at later developmental stages. Similar phenotypes have been reported in other studies when INM is impaired by Lis1 loss-of-function, or by compromising centrosome function through knockdown of centrosomal proteins (Gambello et al., 2003; Xie et al., 2007). A mathematical simulation based on a gradient of Notch signaling in the VZ predicts similar results (Murciano et al., 2002).
Why does ectopic cell division produce two neurons? During apical division, cells are exposed to neural inhibition signals such as Notch, which displays an apical-basal gradient in developing epithelia (Del Bene et al., 2008; Murciano et al., 2002). Ectopic divisions free cells from this inhibition, resulting in neuronal differentiation. Thus, INM may be a strategy used by the developing cortex to maintain a sufficient neural progenitor pool size during early developmental stages to ensure that the correct number of neurons is generated with temporal precision.
We performed several control experiments to ensure that other properties of neural progenitors were not affected by Hook3 or PCM1 knockdown. First, FACS analysis revealed that Hook3 knockdown did not affect cell cycle progression (Figure S2). Second, the polarity of neural progenitors is not affected, as evidenced by the intact adhesion and polarity of progenitors (Figure S6). We also found that apical processes attaching neural progenitors to the ventricular surface were not affected in knockdown cells (Figure (Figure2F,2F, S4A). Finally, we exclude the possibility that overproduction of neurons may be caused by over-proliferation of Tbr2-positive INPs (Figure S5). These results are consistent with our finding that the neural progenitor pool was depleted.
Multiple centrosomal proteins have been implicated in human brain development diseases, such as Lis1 (Lissencephaly), ASPM and CDK5Rap2 (microcephaly), Cep290 (Joubert Syndrome). A fraction of DISC1, the protein strongly associated with schizophrenia, also localize to the centrosome (Kamiya et al., 2005; Morris et al., 2003). Both Cep290 and DISC1 interact with two components of the pericentriolar satellite, BBS4 and PCM1 (Kamiya et al., 2005; Kim et al., 2008), suggesting that these two proteins might be the cargo of pericentriolar satellites. In addition, our study established the dependence of the centrosomal assembly of CDK5Rap2 on pericentriolar satellites. Furthermore, recent human genetic studies revealed that the PCM1 gene locus on chromosome 8p22 is strongly associated with schizophrenia (Datta et al., 2008; Gurling et al., 2006; Kamiya et al., 2008). Both DISC1- and PCM1- associated schizophrenic patients show reduced gray matter volume as revealed by magnetic resonance imaging (MRI) studies (Gurling et al., 2006; van Haren et al., 2004), suggesting a possible contribution of impaired neurogenesis to the onset of schizophrenia. In line with this, our previous study showed that DISC1 loss of function results in premature depletion of the neural progenitor pool in early embryonic developmental stages (Mao et al., 2009). In the current study, we showed that compromising PCM1 function leads to similar neurogenesis defects. Taken together, these findings suggest that pericentriolar satellite-mediated centrosomal protein assembly is a general mechanism for the proper functioning of the centrosome in embryonic brain development. Compromising this protein assembly may result in disturbance of multiple events during embryonic brain development, which eventually leads to CNS diseases. For the first time, our study revealed the significance of centrosome protein assembly in embryonic brain development, providing novel insight into the understanding of brain development and diseases.
RNAi plasmids were generated by inserting complementary hairpin oligonucleotides into pSilencer 2.0-U6 (Ambion). Hook3 RNAi-2, and PCM1 RNAI-1, and -2 were obtained from the Mission shRNA library (Sigma). A random sequence without homology to any known mRNA was used as the control RNAi. The Hook3 rescue construct was cloned using the human full-length Hook3 coding sequence; it is resistant to Hook3 RNAi-1 because the latter targets to the 3′-UTR of mouse Hook3. Human Hook3 constructs were provided by H. Kramer (Univ. Texas Southwestern medical center, Dallas, Texas). The EGFP-PCM1 plasmid was provided by A. Merdes (Institute of Science and Technology, Toulouse, France).
Venus plasmid (final concentration 1μg/μl) was co-injected with RNAi plasmid (final concentration 3μg/μl). For triple-electroporations, a mixture of venus, Hook3 RNAi, and wild-type Hook3 was prepared at a ratio of 1:3:3. After pregnant females (Swiss Webster, TACONIC) were anesthetized, embryos were exposed within the uterus, approximately 1μl DNA solution was injected into the lateral ventricle, and electroporation was performed (35 V for 50 ms, with 950ms intervals; 5 pulses). Embryos were then placed back into the abdominal cavity and the abdominal wall was sutured. For BrdU labeling experiments, in utero electroporation was performed at E11.5 and BrdU was injected once at 50 mg/g body weight intraperitoneally at E13.5. Embryos were harvested 3 days after electroporation. Brains were removed and fixed in 4% paraformaldehyde overnight, followed by cryoprotection in 30% sucrose in PBS overnight. Afterwards, brains were embedded in OCT and frozen in liquid nitrogen, and sliced into 12-20μm coronal sections.
Mouse embryos were electroporated at E11.5 and sacrificed at E14.5. Embryonic brains were excised and embedded in 3% low melt agarose (SIGMA, Type VII) and sectioned (coronal, 200μm) using a Vibrotome (Leica VT1000S). Slices were laid on a Millicell culture insert (Millipore) and cultured in Neurobasal medium (supplemented with 1% penicillin-streptomycin, 1% glutamine, N2, B27, and 5% horse serum) at 37°C for 1 hr. Thereafter, brain sections were viewed through a 20X objective (NA 0.75) of a Nikon inverted microscope linked to a DeltaVision deconvolution-imaging system (Applied Precision). Images were collected every 10 min for 6-8 hr.
For cryosections, brain slices were incubated in blocking solution (2% goat serum, 0.2% Triton X-100 in PBS) for 1 hr at room temperature. For BrdU staining, brain sections were incubated in 4N HCl solution for 2 hr at room temperature to unmask the antigen, followed by three washes in PBS. Subsequently, brain sections were blocked and incubated in primary antibodies diluted in blocking solution for 2 hr at room temperature, or 24 hr at 4°C. Primary antibodies used were anti-Hook3 (kindly provided by H. Kramer), anti-nestin (Rat401, 1:500; BD Pharmingen), anti-Tuj1 (monoclonal, 1:1000, Covance), anti-pericentrin (monoclonal, 1:100, BD Biosciences), anti-phospho histone H3 (polyclonal, 1:500, Upstate), anti-BrdU (monoclonal, 1:500, Dakocytomation), anti-Ki67 (poluclonal, 1:500, Neomarkers), anti-α-tubulin (1:1000, Sigma T5168), anti-PCM1 (polyclonal, 1:500, kindly provided by A. Merdes), anti-GFP (Chicken polyclonal, 1:1000, Aves Labs). Sections were then washed three times with PBS and incubated with secondary antibody against goat IgG conjugated to cy2, or cy3, or cy5 (1:500, Jackson ImmunoResearch), followed by staining with Hoechst 33258 (Sigma). Sections were mounted with Antifade reagent (Invitrogen) for fluorescence microscopy.
N2A cells were cotransfected with Venus and the indicated plasmids, and sorted by FACS to enrich for GFP positive cells the following day. Cells were fixed after a 2 day recovery period. For conventional EM, cells were fixed in 2.5% paraformaldehyde and 2.5% glutaraldehyde in PBS PH 7.4 for 10 min at room temperature, followed by three washes in PBS. Samples were then postfixed with 1% osmiumtetroxide/1.5% potassium ferrocyanide (in H2O) for 1 hour at room temperature in the dark, followed by three washes in water. Next, samples were incubated in 1% uranyl acetate (in H2O) for 30 min, and then dehydrated through an ethanol series, infiltrated in a 1:1 solution of propylene oxide and epon, and finally embedded in pure epon overnight at 60°C. Sections were prepared using an ultramicrotome (Reichert Ultracut –S), collected on Formvar-coated copper grids, post-stained at room temperature with 2% aqueous uranyl acetate, and treated with 0.2% lead citrate for 15 min.
For immuno-gold staining, cells were fixed in 4% paraformaldehyde in PBS for 10 min at room temperature, and permeabilized in 0.2% Triton X-100 in PBS for 10 min at room temperature. Samples were then blocked in 1% BSA in PBS for 30 min, and incubated in primary antibody (Rabbit anti-Hook3, 1:50; Rabbit anti-PCM1, 1:50) at 4°C overnight. After three washes in PBS, samples were incubated in 5nm colloidal gold-conjugated Protein A for 1-2 hr at room temperature. After 5 washes in PBS, samples were post-fixed in 2.5% glutaraldehyde in 0.1M cacodylate buffer pH7.4 for 0.5-1 hr. Samples were then processed as described above. Sections were observed and photographed on an electron microscope (Tecnai™ G2 Spirit BioTWIN).
NIH3T3 cells were transfected with venus-CDK5rap2 together with corresponding RNAis, and plated on glass-bottom dishes. FRAP was performed 3 days later with a Zeiss LSM 510 confocal microscope using a 63X water lens. A small region of interest of about 2μm in diameter centered on the centrosome was bleached with 10 iterations and 100% laser power (488-nm argon laser). Two images were captured before bleaching. After bleaching, images were taken every 2 seconds (488-nm argon laser at 2% power) over a 200 second period. At each time point, the fluorescence intensity of the photobleached area (P1) and an unbleached area of the same size in the cytoplasm were measured with LSM 510 software. Fluorescence intensity was normalized as B1/B2. Background fluorescence was defined as the normalized intensity of the first frame after photobleaching, and this value was subtracted from all frames to obtain the final fluorescence intensity of the ROI (Proi =B1/B2 - BI). Fluorescence recovery of a given time point was calculated as the final fluorescence intensity (Proi) divided by the final fluorescence intensity of the frame immediately prior to photobleaching.
We thank Dr. H. Kramer for providing the Hook3 antibody, Dr. A. Merdes for the PCM1 antibody and GFP-PCM1 construct, and Dr. M. Mogensen for the ninein antibody. We acknowledge Dr. Z. Xie, J. Buchman, K. Singh, T. Shu, and K Sanada for technical support, helpful discussion, and critical reading of the manuscript; M. Ericsson and L. Trakimas at the EM facility of Harvard Medical School for technical assistance. This work was supported by the NIH RO1 grant NS37007 to L.-H.T. L.-H. T. is an investigator of the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.