The term apical stem and progenitor cells (APs) refer to cells undergoing mitosis at the apical surface of the ventricular zone (VZ; Figure 1). This category is comprised of cells which have a wide range of mitotic capacity and proliferation/differentiation potential. In the subsequent sections, we will describe the different AP subtypes in order from the most multipotent down to unipotent cells.

Before the onset of neurogenesis, the developing brain is formed mostly by neuroepithelial cells (NECs). They are highly polarized epithelial cells exhibiting apico-basal polarity (Figure 1). Their apical plasma membrane is integrated into the adherens junctional (AJ) belt and lines the lumen of the neural tube, which is filled with lipoprotein- and membrane particle-rich embryonic cerebrospinal fluid (Lehtinen et al., 2011). The AJs are cell junctions surrounding the cell, they are linked to the actin cytoskeleton and they separate the apical and the basal domain of NECs (for an historical perspective on AJs refer to Franke, 2009; and references therein; see also Farquhar and Palade, 1963; Takeichi, 1977; Stocker and Chenn, 2015). The basal plasma membrane of NECs spans the neuroepithelium and reaches the basal lamina, a rich source of extracellular molecules (Vaccarino et al., 1999a,b; Raballo et al., 2000; Götz and Huttner, 2005; Fietz et al., 2010). This highly dynamic and rich micro-environment provides a “stem cell niche” to the NECs during development (Lehtinen et al., 2011) that is crucial for the regulation of neurogenesis. NECs undergo interkinetic nuclear migration (INM), that is, they move their nuclei in the VZ in concert with the cell cycle: after completing mitosis at the ventricular surface, their nuclei undergo apical-to-basal migration during G1. After exiting from S-phase at the basal part of the VZ the nuclei undergo basal-to-apical migration, so that their successive mitosis will occur again at the ventricular surface (see Taverna and Huttner, 2010; Lee and Norden, 2013 and references therein). The NECs mitosis is confined to the ventricular surface, as the apical plasma membrane harbors the primary cilium that is nucleated by the centrosomes that also builds the mitotic spindle (see Taverna and Huttner, 2010; and references therein). NECs undergo proliferative divisions to expand the NEC pool. Ultimately, they develop into radial glial cells. Although it is not the main topics of this review article, it is important to mention that a proportion of NECs are embryonic neural stem cells from which adult neural stem cells originate (Furutachi et al., 2015). Using a strategy to follow the cell cycle progression of NECs, it has been shown that a subpopulation of NECs at early developmental stage gives rise to adult neural stem cells. This NECs subpopulation can be therefore considered as an embryonic neural stem cell (Furutachi et al., 2015). Embryonic neural stem cells slow down their cell-cycle speed allowing cells remain in a quiescence state. It still remains unclear how this embryonic neural stem cell population is determined during development and if adult neural stem cells can have also a radial glial origin.

With the onset of neurogenesis, NECs differentiate into apical radial glial cells (aRGCs; Malatesta et al., 2003; Götz and Huttner, 2005). aRGCs are even more elongated than NECs. Their basolateral plasma membrane is divided in two sub-compartments: the apical and the basal process (Figures 1, 2). The apical process is the portion of the basolateral plasma membrane residing in the VZ and it accommodates the nucleus during the different phases of INM. The basal process is the portion of the cell that traverses the sub-ventricular zone (SVZ) and the forming neuronal layers and reaches the basal lamina. Of note, as neurogenesis proceeds and neuronal layers are formed, the width of the cortical wall increases radially: therefore, the basal process elongates. The basal process functions as a guide for radial neuronal migration, allowing newborn excitatory cortical neurons to translocate from the place of birth to their final destination. In addition to providing a migratory scaffold for neurons in their journey to the cortical plate (CP), the basal process is a subcellular compartment involved in signaling and fate specification (Stenzel et al., 2014). Furthermore, live imaging has revealed that the basal process is a very dynamic entity, with the basal end changing from highly branched to club-like during cortical development (Yokota et al., 2010; Figure 2).

aRGCs have been extensively studied in the last decades and it is now clear that they contribute to neurogenesis mainly via the generation of a second type of neural progenitor cells: the BPs (Pontious et al., 2008; Kowalczyk et al., 2009). Interestingly, one of the main difference between apical and BPs is the absence of apical polarity cues in the latter.

Short neural precursors (SNPs, also known as apical intermediate progenitors) were first described in the mouse developing neocortex, where they exhibit several features in common with aRGCs, such as the bipolar morphology and the integration into the AJ belt (Gal et al., 2006; Tyler and Haydar, 2013). Unlike aRGCs, SNPs feature a basal process that does not traverse the neuronal layer, but it is confined to the VZ (Figure 1). A potentially similar cell type was reported to be present in the mouse ventral telencephalon, where aRGCs give rise to interneurons (Tan et al., 2016). As development proceed, the aRGCs in the ventral telencephalon grow radial glial fibers that no longer reach the basal lamina, but rather contact periventricular vessels (Tan et al., 2016). The vessel-anchored aRGCs undergo INM, divide apically (as SNPs do) and maintain the radial fiber throughout mitosis (unlike SNPs). Furthermore, a recent paper (Nowakowski et al., 2016) shows that in the human developing neocortex, during the mid-neurogenesis stage, aRGCs transform into “truncated” aRGCs, with a basal process that no longer reaches the basal lamina, but terminates in the depth of the cortical wall. Several interesting questions remain: are SNPs, vessel-anchored aRGCs in the mouse ventral telencephalon and truncated aRGCs in human related? Is the lack of basal attachment affecting the radial migration of the daughter cell after division? Is the daughter cell migrating for shorter distances?

Subapical progenitors (SAPs) were identified in the mouse ventral telencephalon and in the dorsal telencephalon of gyrencephalic species (Pilz et al., 2013). SAPs are anchored to the ventricle with an apical process. However, they undergo mitosis at a subapical location (Pilz et al., 2013; Figure 1). Most likely, in SAPs the centrosome is not docked at the apical plasma membrane and is therefore free to move basally to nucleate the mitotic spindle, as opposed to what happens in NECs where the centrosome is restricted to the apical side (see above).

The term BPs indicates cells which undergo mitosis in the SVZ, the secondary germinal zone located basally compared to the VZ (Figure 1). BPs are generated by divisions of aRGCs and move basally via a process known as delamination. BPs are further divided into two classes: the intermediate progenitors (IPCs) and the basal radial glial cells (bRGCs).

IPCs represent the main class of BPs in rodents, as originally described independently by three different groups (Haubensak et al., 2004; Noctor et al., 2004; Miyata et al., 2004). IPCs are non-polar cells, as they lack both apical and basal polarity cues (Figure 1). The process of delamination of an IPC from the apical junctional belt very much resembles an epithelial-to-mesenchimal transition (EMT; Wilsch-Bräuninger et al., 2016), in which an epithelial cell gives rise to an unpolarized, highly motile cell (reviewed in Acloque et al., 2009; Itoh et al., 2013b). EMT is a process by which a polarized epithelial cell transforms into an unpolarized and highly motile mesenchymal cell. EMT occurs extensively during embryogenesis and it is crucial for gastrulation and neural crest formation. In pathological conditions, EMT is associated with in the initiation of metastasis and with cancer progression (Acloque et al., 2009; Itoh et al., 2013b; Wilsch-Bräuninger et al., 2016).

Basal radial glial cells were first described in the developing neocortex of gyrencephalic species, namely in humans and ferrets, and were subsequently found, albeit at a much lower abundance, also in the lissencephalic developing rodent brain (Fietz et al., 2010; Hansen et al., 2010; Reillo et al., 2011; Wang et al., 2011; Reillo and Borrell, 2012; Sauerland et al., 2016). From an evolutionary perspective, the pool of bRGCs greatly expanded in humans and other gyrencephalic species, leading to the generation of two separate basal germinal zones: the inner and outer sub VZ (ISVZ and OSVZ, respectively; Smart et al., 2002). Recently, bRGCs have attracted great interest, as their abundance seems to correlate with the extent of brain expansion across species and with gyrification (Reillo et al., 2011; Kelava and Huttner, 2013). Several lines of evidence, including wide occurrence of bRGs in the marsupials and wallaby, suggest that bRGs might have been present in the ancestor of all mammals (Kelava et al., 2012; Sauerland et al., 2016). From a cell biological point of view, bRGCs are monopolar cells lacking an apical attachment (Figure 1). Interestingly, bRGCs still maintain an attachment to the basal lamina via a basal process (Hansen et al., 2010; Florio et al., 2015; Nowakowski et al., 2016). Functional manipulation has shown that the basal process is crucial for the maintenance of the proliferative capacity of bRGCs (Fietz et al., 2010). bRGCs appear to be a rather heterogeneous cell population, as shown by high resolution live imaging of the developing macaque neocortex (Betizeau et al., 2013). The heterogeneity was found to be both morphological and transcriptional. In particular, macaque bRGCs differ in term of presence vs. absence of apical-directed and basal-directed processes (Betizeau et al., 2013).

In summary, APs and BPs show striking differences in cell polarity. Evidences are accumulating that polarity influences the behavior of neural stem and progenitor cells during brain development.

The organization of aRGCs along their apico-basal axis somehow mirrors the histological organization of the cortical wall. In that context, the basal and apical extensions of aRGCs could allow the aRGCs to sense, integrate and respond to different signals generated in different niches, either at the apical or at the basal pole. The apical plasma membrane delimits the ventricle, which is filled with cerebrospinal fluid, and contains different signaling molecules including morphogens (Lehtinen et al., 2011, 2013). At the basal pole, the basal end feet are physically in contact with meninges, extracellular matrix and Cajal-Retzius cells which are sources for morphogens promoting proliferation and/or differentiation (Siegenthaler et al., 2009; Griveau et al., 2010). The basal end foot is therefore in a privileged position to sense and respond to basal extracellular signals. These signals could regulate local biological processes such as mRNA translation (Tsunekawa et al., 2012; Pilaz et al., 2016). It is interesting to note that due to their cellular organization, different neural stem cell subtypes have different level of access to signaling niches localized along the apico-basal axis of the cell and of the tissue. Since during development and evolution there is a progressive shift from apical to basal mitosis, it is tempting to speculate that this shift also represents a shift in signaling, with stem cells being regulated by basal and apical polarity cues during early development, and then being regulated mainly by basal polarity cues (Stenzel et al., 2014). This could also represent a way to restrict the expansion of the ventricular surface and favor the expansion of the basal part of the developing neocortex.

The cell polarity of APs is the structural basis for symmetric vs. asymmetric division as it allows the equal vs. unequal distribution of cellular components to the daughter cells (Huttner and Kosodo, 2005). Several findings support the idea that the apical pole and subcellular structure therein are asymmetrically partitioned and influence cell fate. The apical plasma membrane constitutes a minor proportion of the total plasma membrane and can be either bisected or bypassed by the cleavage plane, resulting in only one daughter cell inheriting a portion of the apical plasma membrane (Kosodo et al., 2004). The cell inheriting the apical plasma membrane was reported to maintain proliferative potential (Kosodo et al., 2004). Not all the apical plasma membrane is partitioned based on the cleavage plane orientation: it was shown that the ciliary membrane, a specialized domain of the apical plasma membrane, is endocytosed at the onset of mitosis (Paridaen et al., 2013). The ciliary membrane preferentially associates with the mother centriole, is conserved throughout mitosis in association with one spindle pole and is asymmetrically inherited by one of the two daughter cells (Paridaen et al., 2013). The cell inheriting the ciliary membrane re-establishes the cilium faster and tend to maintain stem cell-like characteristics. These findings strongly suggest a role for the apical pole and subcellular structures therein in maintaining and influencing the choice between proliferation and differentiation.

It has also been shown that the basal process can be asymmetrically inherited by one of the two daughter cells, with the cell inheriting the basal process tending to maintain proliferative capacities (Konno et al., 2008; Shitamukai et al., 2011). How is the inheritance of the basal process linked to the choice between proliferation and differentiation? The inherited basal process could maintain aRGCs in a proliferative state thanks to the inheritance of the basal process-localized mRNAs and/or receptors for growth factors. Furthermore, data showed that both aRGCs and bRGCs are able to re-grow their basal process (Hansen et al., 2010; Shitamukai and Matsuzaki, 2012; Betizeau et al., 2013; Subramanian et al., 2017). What are the differences between an inherited vs. a regrown basal process? Is the re-growth of the basal process a mechanism to bypass the limitation imposed by the asymmetric inheritance of the basal process, so that both daughter cells are equally able to respond to extracellular signals? Is the re-grown basal process featuring different receptors compared to the inherited basal process? One might speculate that the newly delivered receptors in a re-grown basal process features different post-translational modifications and/or has a different desensitization status, allowing the two daughter cells to respond differentially to the same extracellular stimuli. Further research will be required to obtain a coherent picture on the interplay between polarity, asymmetric division and cell fate specification. It would also be extremely interesting to extend the pioneering work performed on aRGCs to other polarized progenitor cells, such as bRGCs, in order to understand to which extent similar cell biological principles are used in different cell types and how they act in generating neural stem cell diversity.

Down syndrome is the most common genetic cause of mental retardation. Patients diagnosed with Down syndrome showed an overall reduction of cerebrum gray matter volume (Weitzdoerfer et al., 2002) as well as a disorganized cortical lamination (Pinter et al., 2001). The reduction of Arp2/3 complex was reported in fetal Down syndrome brain and the conditional ablation of Arp2/3 complex in mice showed a reduction in neuronal number and highly disorganized cortical lamination (Wang et al., 2016). It would be important to understand to which extent the effects of Arp2/3 on the overall brain functions are due to the early effects of Arp2/3 on neural stem and progenitor cells, in particular on aRGCs (Tyler and Haydar, 2013).

Fragile-X syndrome (FXS) is the most common form of inherited intellectual disability and it is caused by mutations in Fragile X Mental Retardation 1 (FMR1) gene. Loss-of-functions of FMR protein (FMRP) showed: (i) a switch from AP to BP fate, indicating the depletion of aRGCs pool (Saffary and Xie, 2011); (ii) defects in neuronal positioning due to the misregulation of N-cadherin levels (La Fata et al., 2014); and (iii) early postnatal circuitry impairments possibly linked to abnormalities in the projection fibers (La Fata et al., 2014). Of interest, FMR1 is also the most common single genetic cause of autism spectrum disorders (ASD; Hagerman et al., 2011; Bagni et al., 2012). In line with that, clinical crosstalk has been reported between FXS and ASD (Hagerman et al., 2011; Bagni et al., 2012). Considering the involvement of FMRP in mRNA transport in aRGCs (Pilaz et al., 2016; Pilaz and Silver, 2017), it would be interesting to understand if and how impaired mRNA transport in aRGCs contribute to the etiology of FXS and ASD (for an extensive discussion on the link between FXS, ASD, neural progenitors and cortical neurogenesis refer to Callan and Zarnescu, 2011; Packer, 2016; Marchetto et al., 2017 and references therein).

Ciliopathies are genetic disorders of ciliary structure or function. Joubert syndrome (JS) and related disorders are ciliopathies clinically characterized by ataxia, psychomotor delay and cognitive impairment (Cantagrel et al., 2008). The classical form of JS is caused by mutations in Arl13b, a cilia-specific small GTPase. Arl13b mutations lead to an inverted apico-basal polarity of aRGCs and impair the ability of primary cilia to convey extracellular signals such as insulin-like growth factor (Igf; Higginbotham et al., 2013). Igf is highly enriched in the CSF. It would be interesting to understand if the relocation of the primary cilium from the apical to the basolateral plasma membrane upon APs to BPs fate transition (Wilsch-Bräuninger et al., 2012) changes the degree of exposure to extracellular signals generated from the CSF.

Schizophrenia shares several common characteristics with ASD in term of behavioral, social and cognitive disturbances and also in term of genes implicated in the disease etiology (Carroll and Owen, 2009). Disrupted In Schizophrenia (DISC1) is a common susceptibility gene for those disorders and it is also associated with bipolar and mood disorders (Khanzada et al., 2017). Mutations in DISC1 gene lead to schizophrenic or depressive behavior (Clapcote et al., 2007; Mao et al., 2009; Dachtler et al., 2016). DISC1 is highly expressed in aRGCs during development and a knock-down of DISC1 showed a decreased proliferation of aRGCs and premature neurogenesis (Mao et al., 2009; De Rienzo et al., 2011; Ishizuka et al., 2011; Ye et al., 2017). The neuronal phenotypes observed are reminiscent of Cdc42 loss-of-functions (Yokota et al., 2010; Ishizuka et al., 2011). Consistent with that, DISC1 regulates aRGC proliferation through GSK3β, a downstream effector of Cdc42 (Clapcote et al., 2007; Ishizuka et al., 2011; Dachtler et al., 2016).

Taken together, these data suggest that alteration of aRGCs polarity can trigger neurodevelopmental and psychiatric disorders.