mTORC1 signaling and primary cilia are required for brain ventricle morphogenesis

The vertebrate brain forms around a ventricular cavity filled with cerebrospinal fluid (CSF) that undergoes dynamic morphological changes during development. The neural tube wall constitutes a pseudostratified epithelium composed of elongated progenitor cells called radial glial cells (RGCs) that progressively generate, directly or indirectly, all neurons and glial cells. The brain ventricular system plays crucial roles in brain development, function and homeostasis, and leads to severe neurological disorders when defective (Lehtinen and Walsh, 2011; Lowery and Sive, 2009). Deciphering the mechanisms that regulate brain ventricle morphogenesis is thus essential to understand neocortical histogenesis and identify new therapeutic targets for these common pathologies.

The cerebral cortex originates from the dorsal telencephalon and is composed of six layers, each containing different classes of neurons. Cortical neurons and glia are formed through asymmetric divisions of RGCs, which are bipolar cells that traverse the width of the epithelium, with a basal domain at the pial surface and a ciliated apical contact at the ventricular surface. The primary cilium is a microtubule-based membrane protrusion, the axoneme of which is nucleated by the basal body, a modified mother centriole. Its dynamics are intimately linked to the progression of the cell cycle; it grows during the G1 phase and is resorbed just before mitosis (Nigg and Stearns, 2011). The primary cilium is an important signaling organelle at all developmental stages and has been shown to be crucial for morphogenetic processes such as brain patterning and tissue homeostasis (Eggenschwiler and Anderson, 2007; Willaredt et al., 2008). The primary cilium is chemosensitive and detects a variety of signaling molecules that are important for development (e.g. Shh, Wnt, Pdgf) (Guemez-Gamboa et al., 2014). In the kidney, it also acts as a mechanosensor that detects shear stress, leading to the downregulation of the mTOR pathway necessary for the proper control of cell size (Boehlke et al., 2010; Orhon et al., 2016). The cerebrospinal fluid secreted by the choroid plexus provides diffusible signals that are essential for the early development of RGC (Lehtinen et al., 2011; Chau et al., 2015; Higginbotham et al., 2013). The apical localization of the primary cilium is thus optimal for detecting chemo- or mechano-sensory signals from the CSF, and thus participates in brain morphogenesis by regulating RGC physiology (Lehtinen and Walsh, 2011).

Ciliopathies are a group of genetic diseases characterized by impaired ciliary function. Common brain pathologies encountered in ciliopathy patients are neurocognitive impairments, epilepsy and hydrocephalus. Hydrocephalus is a complex multifactorial brain disorder leading to reduced cortical thickness and enlarged ventricular cavities due to the accumulation of CSF in the ventricles at postnatal stages. It is one of the most common birth defects (Zhang et al., 2006) and has been correlated with a wide range of neurodevelopmental disorders, including schizophrenia (Shenton et al., 2001). Abnormal brain ventricle enlargement often develops at early postnatal stages as a result of defects in ependymal cilia (Ibanez-Tallon et al., 2003) or the choroid plexus (Banizs et al., 2005; Del Bigio, 2010). It was also shown that postnatal hydrocephalus could result from increased apoptosis and impaired proliferation of cilium-less NG2⁺ and PDGFR-α⁺ neural progenitors (Carter et al., 2012). In this study, ventriculomegaly and hydrocephalus are defined as brain ventricular enlargement at embryonic or postnatal stages, respectively. To date, the mechanisms leading to prenatal ventriculomegaly have remained obscure.

Using conditional mouse mutants, we show that depletion of the primary cilia of RGCs from E10.5 onwards results in progressive enlargement of the lateral ventricles (ventriculomegaly) without affecting brain patterning. This phenotype is associated with progressive enlargement of the surface area of RGC apical endfeet induced by abnormal upregulation of the mTORC1 pathway. In addition, we show that cilia depletion leads to misorientation of the RGC mitotic spindle, which is associated with an increased number of basal progenitors in the somatosensory cortex. The size of the apical endfeet, the increased number of basal progenitors and the ventricular enlargement are all rescued by rapamycin treatment. These results suggest that the mTORC1 pathway controlled by the primary cilium regulates ventricle morphogenesis and corticogenesis, and thus constitutes a new potential therapeutic target for the treatment of hydrocephalus.

During mouse brain development, neuroepithelial (NEP) cells and RGCs extend a primary cilium into the ventricles (Higginbotham et al., 2013; Fig. S1A). The dynamics are tightly correlated with the cell cycle and interkinetic nuclear migration (Kim et al., 2011; Paridaen et al., 2013). To study the role of the primary cilium in forebrain development, we generated conditional ciliary knockouts of floxed Ift88 or Kif3a genes (Haycraft et al., 2007; Marszalek et al., 1999) to ablate cilia from the apical surfaces of either NEP cells (using FoxG1-Cre mice: FoxG1-K3A^cKO) or RGCs (using Nestin-Cre mice: Nestin-K3A^cKO or Nestin-Ift88^cKO, Fig. S1B-E). In FoxG1-K3A^cKO mice, cilia depletion in NEP cells leads to patterning defects and altered GLI3 processing, as previously reported (Benadiba et al., 2012; Besse et al., 2011; Higginbotham et al., 2013; Willaredt et al., 2008; Laclef et al., 2015; Fig. S1F-H). In contrast, no polarity or patterning defects were observed in Nestin-Cre conditional ciliary mutants (Nestin-K3A^cKO or Nestin-Ift88^cKO), in which Cre was expressed in the forebrain from embryonic day (E)11 onwards and recombination leads to complete cilia ablation from E14.5 (Fig. S1B-E). Indeed, both the expression patterns of Ngn2 (Neurog2; a marker of the dorsal telencephalon) and Dbx1 (a marker of the pallium-subpallium boundary), and Gli3 processing were unaltered in Nestin conditional knockout (cKO) mutant forebrains at E12.5 (Fig. S1F-H). Furthermore, the apical positions of the centrosomes (Fig. S1C and Fig. S3A,B) and thus the polarity of RGCs were normal. As previously described (Tong et al., 2014), we observed that Nestin-K3A^cKO conditional ciliary mutants show severe hydrocephalus at postnatal stages (Fig. 1). However, we noticed that at earlier stages (from E18.5 onwards), Nestin-K3A^cKO mice displayed a progressive enlargement of the lateral ventricles (Fig. 1A,B), and reduced cortical thickness (Fig. 1A,C) associated with a moderate but significant decrease in the number of post-mitotic neurons labeled with a Ctip2 antibody (a marker of early-born neurons) (Fig. 1D,E). Similar phenotypes were observed in Nestin-Ift88^cKO mice (data not shown). The enlarged ventricles and the reduced brain tissue observed in the ciliary mutants before motile cilia develop both suggest that ablation of primary cilia at E11 leads to the development of prenatal ventriculomegaly, which might initiate postnatal hydrocephalus.

Primary cilia extend from the apical surface of RGCs in the brain ventricles. To further identify the cellular mechanisms leading to ventricular enlargement, we analyzed the apical domains of RGCs on whole-mount somatosensory cortical ventricular walls in Nestin-K3A^cKO ciliary mutant and control embryos (Fig. 2A) stained with an antibody against ZO-1 (Tjp1) that labels the tight junctions and delimits individual RGC apical domains (Fig. S2A,B,D). The size of the apical domains was quantified and color-coded with Packing Analyzer (Aigouy et al., 2010) and CellProfiler (Lamprecht et al., 2007) software (Fig. 2B-F). In control embryos, the mean size of the apical domains increased from E12.5 to E16.5; they were more heterogeneous at E16.5 than they were at E14.5 (Fig. 2C-F). Thus, the size of the RGC apical domain increases progressively during normal development, as previously shown (Nishizawa et al., 2007). Interestingly, the increase was significantly greater in ciliary mutants than in controls at all stages analyzed, and the difference increased with age: the relative difference in size between ciliary mutant and controls was 7% at E12.5, 35% at E14.5 and 39% at E16.5 (Fig. 2C-F). Similar results were obtained in Nestin-Ift88^cKO embryos compared with controls (Fig. S2B-C). Finally, as Nestin is also expressed in the lateral ganglionic eminence (LGE), we confirmed that apical domain enlargement was also present in Nestin-K3A^cKO lateral ganglionic eminence at E14.5 (Fig. S2D,E). These results suggest that cilia abrogation leads to a progressive increase in the size of RGC apical domains during forebrain development.

RGCs are highly polarized cells in which nuclear interkinetic migration takes place in the ventricular zone (VZ). Their nucleus is larger than the surface of their apical domain, suggesting that the size of the apical domain reflects the distance of the cell body from the ventricle and, thus, the phase of the cell cycle. To examine this possibility and determine whether the increase in the size of the apical domain in cilia mutants is correlated with a given phase of the cell cycle, whole-mount cortical ventricular walls from ciliary mutants and controls were triple-labeled with antibodies against ZO-1, 4A4 (phospho-vimentin; cytoplasmic mitotic marker) and γ-tubulin (Tubg1; marker of the pericentriolar material). Apical domains in three categories of cells were examined: interphase cells (γ-tubulin⁺/4A4⁻), mitotic cells (γ-tubulin⁻/4A4⁺) and other cells (γ-tubulin⁻/4A4⁻) (Fig. S3A,B, Fig. 3A-D). The size of the apical domains is significantly smaller during the interphase than during mitosis both in controls and Nestin-K3A^cKO mutants, suggesting that its size is precisely controlled as the cell cycle advances. The percentage of mitotic cells was similar in controls and mutants (Fig. 3A,B), in accordance with the observation that the number of mitotic cells in the ventricular zone was unaffected in cilium-less RGCs (see below, Fig. 3G-I). However, the size of the apical domains in both interphase and mitotic cells was greater in mutants than in controls, suggesting that cilia abrogation leads to a similar enlargement of both interphase and mitotic apical domains (Fig. S3A,B, Fig. 3C,D).

Cell shape and mechanical cues have been shown to guide division orientation in epithelia (Wyatt et al., 2015) and mitotic spindle orientation contributes to cell fate determination (Paridaen and Huttner, 2014). We therefore tested whether RGC apical domain enlargement in ciliary mutants could perturb mitotic spindle orientation. We measured anaphase spindle orientation in controls and Nestin-K3A^cKO mutants. Subtle defects in spindle orientation were observed: the median at E12.5 was 6.2° in controls but 8.7° in mutants (Fig. S3C). This defect worsened at E14.5: the median was 6.9° in controls and 10.2° in mutants (Fig. 3E,F). To test whether these defects affected other cells in the cortex, E14.5 coronal sections were immunostained with antibodies against PH3 (Phc3; mitotic cells), Tbr2 (Eomes; basal progenitors, BPs) and Ctip2 (Bcl11b; early born neurons). Although the abundance of mitotic or Pax6⁺ (RGC) cells was similar in the VZ of controls and Nestin-K3A^cKO and Nestin-Ift88^cKO mutants (Fig. 3G-I, Fig. S3D-F), the number of mitotic cells increased 2-fold in the subventricular zone (SVZ) of the ciliary mutants compared with controls and was associated with a significant increase in the number of basal progenitors (Fig. 3G-K). These results suggest that cilia abrogation leads to apical domain enlargement, impaired mitotic spindle orientation, increased cell proliferation and the generation of supernumerary progenitors in the SVZ at E14.5.

Other non-ciliary effects of Kif3a or Ift88 knockout have been reported to affect microtubule dynamics, leading to spindle misorientation (Delaval et al., 2011; Kodani et al., 2013; Teng et al., 2005); ninein (Nin)⁺ sub-distal appendages were absent from Kif3a embryonic knockout mouse fibroblasts (Kodani et al., 2013) and removal of ninein from the developing neocortex caused premature depletion of progenitors from the VZ (Wang et al., 2009) and enlargement of the endfoot area of apical progenitors (Shinohara et al., 2013). Here, no defects in subdistal appendages or ninein localization at the centrosome were detected in cilium-less RGCs at E14.5 (Fig. S4A,B). Similarly, no defects in polarity were observed in cortical ventricular walls labeled with adherens junction markers on en face views of the RGCs lacking a primary cilium, suggesting that cilia abrogation did not affect markers of apico-basal cell polarity (Fig. S4C). Thus, cilia abrogation in these ciliary mutants does not lead to the loss of sub-distal appendages. Mitotic spindle misorientation observed in Nestin-K3A^cKO ciliary mutant is probably due to increased cell size rather than impaired attachment of microtubules to the centrosome. These results, together with the similar phenotype observed in two different ciliary mutants (Kif3a and Ift88), suggest that mitotic spindle misorientation might be directly linked to cilia abrogation.

The number of Ctip2⁺ neurons at E14.5 was similar in controls and ciliary mutants (Fig. S3D-F) but it was decreased at postnatal day (P)2 (Fig. 1D,E). Interestingly, no defects in cell survival could be observed at either age in controls and Nestin-K3A^cKO mutant mice, suggesting that the decrease in Ctip2⁺ cells at P2 was not due to increased cell death. Indeed, the numbers of activated caspase-3⁺ cells (data not shown) and TUNEL⁺ cells were similar in Nestin-K3A^cKO neonatal mice (7±2 TUNEL⁺ cells per section) and controls (6±1 TUNEL⁺ cells per section). These results thus confirm previous findings that cilia abrogation in Nestin-K3A^cKO mice does not lead to increased cell death (Tong et al., 2014). To test whether thinning of the cortex could be due to tissue stretch, we quantified cell density on tissue sections at P2. However, no significant differences were observed (Fig. S4E). Altogether, these results suggest that the thinner cortex at P2 is due to premature differentiation rather than defects in cell survival or changes in cell density.

In the kidney, deletion of the primary cilium by knockout of Kif3a caused cell enlargement as a result of abnormal upregulation of the mTORC1 pathway (Boehlke et al., 2010; Orhon et al., 2016). To test whether cilia regulate the size of the RGC apical domain through the mTORC1 pathway, we analyzed dorsal telencephalic lysates from controls and ciliary mutants, at E14.5, for phosphorylation of mTOR and its targets p70S6K, S6RP (S6 ribosomal protein) and 4E-BP1, which are established markers of mTORC1 activity. The levels of phosphorylated mTOR and downstream targets increased significantly in ciliary Nestin-K3A^cKO mutants compared with controls (Fig. 4A,B, Fig. S5A). Immunostaining of the downstream mTORC1 effector with an antibody against phosphorylated ribosomal S6 protein (p-S6RP) revealed its localization at the mother centriole of the centrosome and at the apical surface of mitotic RGCs (Fig. 4C,D). Interestingly, ciliary mutant apical domains contained more p-S6RP than controls, as shown by the increased intensity of p-S6RP fluorescence at the mutant ventricular surface (Fig. 4E).

To further examine whether the mTOR pathway regulates the size of the RGC apical domain, we injected the mTORC1 inhibitor rapamycin (6 mg/kg) or vehicle intraperitoneally at E12.5. In control embryos, rapamycin injection at E12.5 had no significant effect on the size of RGC apical domain (Fig. 5A,B, Fig. S5B). In Nestin-K3A^cKO embryos, in which the size of RGC apical domains increased by 35% compared with controls (Fig. 2C), rapamycin injection at E12.5 led to an apical domain size comparable to that observed in control embryos at E14.5 (Fig. 5A-D). These results suggest that rapamycin treatment prevents excessive apical domain enlargement at E14.5 in Nestin-K3A^cKO, which might result from ineffective downregulation of mTORC1 due to the absence of primary cilia. Interestingly, the misorientation of the mitotic spindle and the number of PH3⁺ basal mitotic figures in the developing somatosensory cortex was also rescued in cilium-less RGCs of Nestin-K3A^cKO embryos at E14.5 after a single injection of rapamycin at E12.5 (Fig. 5E,F). These results thus show that cilia abrogation in developing RGCs leads to an abnormal increase in the size of apical domains as a result of upregulation of the mTORC1 pathway, which leads to spindle misorientation in apical mitotic cells and to an abnormal increase in the number of PH3⁺ basal mitotic figures.

Since the increase in size of the cells can be rescued by rapamycin administration, to further confirm that progressive enlargement of RGC apical domains leads to ventriculomegaly, we injected rapamycin at a low dose (1 mg/kg) once a day between E14.5 and E17.5 and analyzed the size of the lateral ventricles at E19. This low dose of rapamycin was used because we observed a high rate of mortality with consecutive treatments at higher doses. Interestingly, four successive injections of rapamycin rescued the size of the ventricle in ciliary mutants, which was not significantly different from that of control embryos injected with vehicle (Fig. 6). Furthermore, although a single injection of rapamycin had no effect on controls at E14.5 because the mean area of the apical domain was similar at E12.5 and E14.5 (4.7 µm² at E12.5 and 4.8 µm² at E14.5; Fig. 2C and Fig. 5A,B), we observed that consecutive rapamycin injections significantly decreased the size of the ventricles in control embryos (Fig. 6), suggesting that mTORC1 inhibition by rapamycin treatment acts as a brake on the enlargement of RGC apical domains but does not decrease the size of the apical domains. These results suggest that the progressive increase in RGC apical domain size during normal cortical development contributes to ventricle morphogenesis and that it is regulated by the mTOR pathway (Fig. 2), although rapamycin treatment might also affect other cell types surrounding the ventricles (Magri and Galli, 2013).

In the present study, we genetically ablated the primary cilium of RGCs using Ift88 or Kif3a conditional mutant mice and observed a progressive increase in the size of their ventricular apical domains through upregulation of the mTORC1 pathway. These defects are associated with a significant perturbation of their mitotic spindle orientation, an overproduction of basal mitotic progenitors at E14.5, a decreased number of cortical neurons at later stages and progressive enlargement of the lateral ventricles, defined as ventriculomegaly. Interestingly, all these defects were attributable to an abnormal increase in mTORC1 pathway activity as they were rescued by one or several injections of the mTORC1 inhibitor rapamycin. Altogether, this study shows for the first time that the primary cilium of RGCs regulates ventricular morphogenesis through control of the size of their ventricular contact by acting as a brake on the mTORC1 signaling pathway.

Here, we used conditional knockout of ciliary genes (Ift88 and Kif3a) in Nestin-Cre mice to deplete RGCs of their primary cilium, bypassing its role in brain patterning and the polarity of neuroepithelial cells (Benadiba et al., 2012; Besse et al., 2011; Higginbotham et al., 2013; Willaredt et al., 2008; Wilson et al., 2011). Although Nestin-Cre expression starts around E10.5 in the forebrain, cilia abrogation in the VZ was complete only at E14.5, as previously shown (Liang et al., 2012; Fig. S1A-E). The phenotypes observed at E12.5 might thus be partially due to inefficient recombination before E12.5. Moreover, Nestin is broadly expressed at later developmental stages, suggesting that some of the phenotypes observed in ciliary mutants are, in part, due to non-cell-autonomous effects.

During their development from RGCs (Spassky et al., 2005), multiciliated ependymal cells enlarge their apical domain (Mirzadeh et al., 2010). It is thus possible that the enrichment of ciliary mutants in large apical domains at a given developmental stage corresponds to premature ependymal differentiation. However, the numbers of ventricular CD24⁺/centrin-2 (Cetn2)-GFP⁺ cells at E18.5 were similar in controls and ciliary mutants, ruling out this hypothesis (Fig. S4D). Another possibility is that depletion of cilia leads to defects in the cytoskeleton and adherens junctions, resulting in increases in ventricular cell delamination and available space at the ventricular surface, and ultimately to enlargement of the apical domains. However, no defects in either adherens junctions or actin cytoskeleton were observed by immunostaining the cortical ventricular surface during corticogenesis in mutant RGC (Fig. S4C). Altogether these observations suggest that ventriculomegaly is due to apical domain enlargement of RGC before their differentiation into ependymal cells, ruling out a contribution of ependymal dysfunction during embryogenesis. The prenatal ventriculomegaly described here precedes and might contribute to the initiation of hydrocephalus observed at postnatal stages (this study and Tong et al., 2014). Postnatal hydrocephalus can thus be caused by a variety of factors including defects in ventricular morphogenesis (this study), disturbed CSF flow due to impaired ependymal cilia function, over-production of CSF by the choroid plexus (Del Bigio, 2010) or combinations of these factors.

In control embryos, RGC apical domains are highly diverse in size at a given developmental stage. The sizes increase throughout development, leading to greater size diversity at later stages (Nishizawa et al., 2007; this study). Enlargement is also observed in ciliary mutants and is greater than in control embryos. Further studies should determine the effects of apical domain morphology, mode of division and long-term potentials on the mechanical constraints exerted by RGCs on the morphogenesis of brain ventricles. We have shown here that the enlargement of the RGC apical domain is due to rapamycin-sensitive control of localized cell growth by the primary cilium. mTORC1 signaling, the targets of which are enriched at the apical domain of RGCs, is known to control cell growth by upregulating protein synthesis (Fig. 4A). This raises the question whether local protein synthesis is involved in apical cell enlargement. Although the size of RGC apical domains has not been studied, microcephaly and ventricular enlargement have been reported in mouse mutants in which the activity of the mTOR pathway was, respectively, defective or aberrantly elevated (Cloetta et al., 2013; Way et al., 2009; Magri et al., 2011). Mechanosensory proteins polycystic kidney disease 1 (Pkd1) and Pkd2 are expressed in primary cilia of RGCs, and their ablation leads to postnatal hydrocephalus and planar cell polarity defects in mouse ventricular epithelium (Wodarczyk et al., 2009; Ohata et al., 2015). Moreover, it was shown that Pkd1 can inhibit the mTOR pathway and regulate cell size in the kidney (Distefano et al., 2009). We thus tested whether Pkd1 contributed to the control of the mTOR pathway and cell size in RGCs by measuring the size of the apical domains in Nestin-Cre conditional knockouts of floxed Pkd1. However, no difference in the surface area of RGC apical domains was observed at E14.5 (Fig. S6), suggesting that primary cilia do not regulate the size of RGCs through the mechanosensory protein Pkd1 via the mTOR pathway. Interestingly, it was recently discovered that primary cilium-dependent autophagy regulates kidney epithelial cell size through the LKB1-AMPK-mTOR signaling pathway in response to fluid flow (Orhon et al., 2016). Whether the mTOR pathway and the mechanical stress pathway are coordinated in the control of RGC size in the primary cilium-dependent autophagy pathway is an attractive hypothesis that remains to be investigated. Further studies are thus required to identify the upstream signals of the mTOR pathway in these cells. Altogether, our results shed light on a new role of the primary cilium in the local control of cell growth and ventricular morphogenesis during brain development.

All animal care was in accordance with French and European legislation. All the mice used in this study were previously described: Kif3a^fl, Kif3a^ko/+ (Marszalek et al., 1999), IFT88^fl, IFT88^ko/+ (Haycraft et al., 2007), Nestin-Cre (Tronche et al., 1999), FoxG1-Cre (Hébert and McConnell, 2000), Rosa26^mTmG (Muzumdar et al., 2007), Centrin-2-GFP (Higginbotham et al., 2004). To produce conditional Kif3a and IFT88 mutants, we crossed Kif3a^fl/fl or IFT88^fl/fl mice with Kif3a^ko/+ Nestin-Cre^+/− or IFT88^ko/+ Nestin-Cre^+/− mice, respectively. To obtain conditional Kif3a FoxG1 mutants, we crossed Kif3a^fl/fl with Kif3a^ko/+ FoxG1-Cre^+/−. Nestin::Cre;Kif3a^ko/fl, Nestin::Cre;Ift88^ko/fl and Nestin::Cre;FoxG1^ko/fl were used as ciliary mutants, whereas Nestin::Cre;Kif3a^fl/+, Nestin::Cre;Ift88^fl/+ and Nestin::Cre;FoxG1^fl/+ mice were used as controls. We also crossed Kif3a^fl/fl Centrin-2^GFP/GFP mice with Kif3a^ko/+ Nestin-Cre^+/− mice to generate conditional mutants with GFP⁺ centrioles. The mice were maintained on a C57Bl6/J background. E0.5 was defined as noon on the day the vaginal plug was detected.

Mouse brains were fixed by immersion in 4% paraformaldehyde at 4°C for 4 h (embryonic brains) or overnight (postnatal brains) then washed several times in PBS. Coronal sections (80 or 100 µm) were cut on a Vibratome (S1000, Leica). For en face views, telencephalic hemispheres were first isolated in DMEM/F12 (Life Technologies) and fixed in 4% paraformaldehyde (PFA) in PBS for 1 h at 4°C. Tissues were then washed in PBS and immunostained as described below. Before mounting, the ventral telencephalon was removed with Pascheff–Wolff microscissors on a silicone rubber plate, and the dorsal telencephalon was flat-mounted by making a few cuts around the edge.

For immunostaining, samples were blocked for 30 min to 1 h in PBS containing 0.1% Triton X-100 and 10% fetal bovine serum, and then incubated overnight with the primary antibodies. The following antibodies were used: rabbit anti-activated caspase-3 (1:1000, Cell Signaling, 9662), rabbit anti-ZO-1 (1:100, Life Technologies, 402200), rat anti-ZO-1 (1:100, Developmental Studies Hybridoma Bank, AB2205518), rabbit anti-Ninein (gift from Michel Bornens; Mogensen et al., 2000), rabbit anti-Tbr2 (1:500, Abcam, ab23345), rabbit anti-Adenylate cyclase 3 (1:500, Santa Cruz, sc-588), mouse anti-Arl13b (1:500, NeuroMab, 75-287), mouse anti-N-cadherin (1:100, BD Biosciences, 610921), rabbit anti-phospho-Histone H3 (Ser10) (1:500, Millipore, 06-570), mouse anti-Vimentin (phospho S55) [4A4] (1:1000, Abcam, ab22651; Weissman et al., 2003), rabbit anti-Pax6 (1:1000, Millipore, AB5409), rabbit anti-Ctip2 [25B6] (1:1000, Millipore, ab18465), mouse anti-β-catenin (1:500, Millipore, 05-665), mouse anti-acetylated tubulin [6-11-B1] (1:500, Sigma-Aldrich, T6793), mouse anti-γ-Tubulin [GTU-88] (1:500, Sigma Aldrich, T6557), mouse anti-Actin (1:100, Millipore, mab1501), rabbit anti-phospho-S6 ribosomal protein (1:100, Cell Signaling, 2211), rat anti-CD24 (1:100, BD Pharmingen, 557436) and species-specific Alexa Fluor 488, 594 and 647 (1:1000, Life Technologies) secondary antibodies. Tissues were counterstained with DAPI (Sigma), mounted in Fluoromount (Southern Biotech) and examined with a confocal fluorescence microscope (SP5 Leica) or a fluorescence microscope (Zeiss Observer.Z1 with Apotome and Hamamatsu camera). For TUNEL staining, coronal cryostat or vibratome sections were labelled with In Situ Cell Death Detection Kit according to the manufacturer's instructions (Roche).

For quantification of cells expressing Ctip2, PH3, Tbr2 or Pax6, a single confocal fluorescent optical section was analyzed. A rectangle with a fixed width (220 µm) spanning the entire thickness of the cortex was placed at a 45° angle to the cortex to approximately mark the somatosensory cortex. Six regions per coronal section on a total of six sections at different rostro-caudal levels were analyzed for each condition. Quantification results are shown as indicated in graphs as a number per unit area (see Fig. 1E and Fig. 3G-J, Fig. S3E-F).

Anaphase spindle orientations were measured on reconstituted Z-projections from en face view acquisitions, using PH3 and γ-tubulin to identify anaphase and pericentriolar material, respectively. The anaphase angle was the angle between the vector connecting the two anaphase-centrosomes and the vector connecting apical surface centrosomes. For in situ hybridization, E12.5 mouse brains were fixed overnight in 4% PFA at 4°C. 100 µm vibratome sections were hybridized as described (López-Bendito et al., 2006) with digoxigenin-labeled Ngn2 and Dbx1 probes (Lu et al., 1994).

Semi-automated segmentation of ZO-1 was performed with Packing Analyzer software (Aigouy et al., 2010), which allows manual correction after automated detection. Segmented images were then vectorized in Illustrator CS6 (Adobe). Finally, images were analyzed with CellProfiler software (Lamprecht et al., 2007) to generate data and color-coded area maps. Edges of the picture were excluded from the analysis.

For scanning electron microscopy, brain embryos were dissected in PBS and fixed in 2% PFA/2.5% glutaraldehyde. Fixed samples were treated with 2% osmium and washed several times in ultrapure water, dehydrated in a graded series of ethanol concentrations and prepared for scanning electron microscopy using the critical point procedure (CPD7501, Polaron). Their surfaces were coated with a 20 nm gold layer using a gold spattering device (Scancoat Six, Edwards). Samples were observed under a Cambridge S260 scanning electron microscope at 15 keV.

For transmission electron microscopy, dorsal E12.5 telencephalons were dissected in PBS then fixed in a 2% PFA/1% glutaraldehyde in 0.1 M phosphate buffer for 3 h. After rinsing in PBS, the tissue was postfixed in 1% osmium tetroxide for 30 min on ice, protected from light, with shaking. The tissue blocks were then dehydrated in 50% and 70% ethanol baths for 7 min each then stained in 1% uranyl acetate in methanol. After the final dehydration, the samples were immersed for 40 min in a graded series of ethanol/Epon solutions (2:1, 1:1, 1:2 ratios), then in pure Epon. The samples were mounted in Epon blocks for 48 h at 60°C to ensure polymerization. Ultrathin sections (70 nm) were cut sagittally on an ultra-microtome (Ultracut E; Leica) and analyzed with a transmission electron microscope (Technai 12, Philips).

Dissected dorsal E12.5 or E14.5 telencephalons were frozen in liquid nitrogen; four or five of the same genotype were pooled. Tissue was homogenized in RIPA lysis buffer (50 mM Tris-HCl pH8, 150 mM NaCl, 1% Igepal CA-630, 0.5% sodium deoxycholate, 0.1% SDS) with Complete ULTRA protease inhibitor cocktail (Roche) and PhosSTOP phosphatase inhibitor cocktail (Roche). The homogenates were centrifuged at 13,500 g for 30 min at 4°C and the clear lysates stored at −80°C until used. Total protein content was quantified with the BCA protein assay (Thermo Fisher Scientific). Samples were then diluted in 2× Laemmli SDS Sample buffer (BioRad) and boiled for 5 min at 95°C. Equal amounts of protein (50 or 100 µg) were run on precast 7.5% Mini Protean TGX Gels (BioRad) and transferred overnight at 4°C onto nitrocellulose membranes (BioRad). The membranes were blocked in 5% non-fat dry milk in TBS-T (pH 7.4 with 0.1% Tween 20) for 1 h at room temperature. Primary antibodies were diluted in 3% BSA (Sigma Aldrich) and the membranes were incubated overnight at 4°C. We used the following antibodies (1:1000 dilution, from Cell Signaling unless otherwise indicated): rabbit anti-phospho-mTOR (Ser 2448) (2971), rabbit anti-mTOR (2972), rabbit anti-p70S6 kinase (9202), rabbit anti-phospho-p70S6 kinase (9205), rabbit anti-S6 ribosomal protein (2217), rabbit anti-phospho-S6 ribosomal protein (4857), mouse 6F5 anti-Gli3 (1:500, Genentech), mouse anti-GAPDH (1:10,000, Ambion). After removing the primary antibodies, the blots were washed several times in TBS-T then incubated with horseradish peroxidase-conjugated secondary antibodies (1:10,000, Jackson ImmunoResearch). Secondary antibodies were detected by SuperSignal West Dura or West Femto Maximum Chemiluminescent substrate (Thermo Scientific). Signals were analyzed with an ImageQuant LAS 400 device (GE Healthcare). Unsaturated bands were quantified using ImageJ (http://rsb.info.nih.gov/ij).

Rapamycin (Merck) was dissolved in 100% ethanol, stored at −20°C and diluted in vehicle containing 5% Tween 80 and 5% PEG400 (Sigma-Aldrich) just before injection. For in vivo administration, pregnant E12.5 dams received an intraperitoneal injection of rapamycin (1 or 6 mg/kg) or vehicle.

For normally distributed data (data shown as the mean±s.e.m.), a two-tailed Student's t-test was performed. For nonparametric distributions (data shown as the median), a two-tailed Mann–Whitney test was performed. *P<0.05, **P<0.01 and ***P<0.001.

Area of the lateral ventricle was measured by outlining the ventricle of three representative sections and the mean value plotted on the graph (n=3) (Figs 1 and 6).

We thank Amandine Correia and Christophe Auger (IBENS animal facility) and Benjamin Mathieu (IBENS Imaging facility) for excellent technical assistance. We wish to thank Benoit Aigouy for the Packing Analyzer software, Genentech for the anti-Gli3 antibody, Michel Bornens for the anti-Ninein antibody, Bradley Yoder, Larry Goldstein and Joseph Gleeson for their kind gifts of the IFT88-lox mutant mice, Kif3a-lox and knockout mouse lines and Centrin-2-GFP transgenic mice, respectively. We are very grateful to the members of the Morin and Spassky labs for insightful and stimulating discussions and I. Caillé for critical reading of the manuscript.