Regulation of asymmetric cell division in mammalian neural stem and cancer precursor cells.

Abstract

Stem and progenitor cells are characterized by their abilities to self-renew and produce differentiated progeny. The balance between self-renewal and differentiation is achieved through control of cell division mode, which can be either asymmetric or symmetric. Failure to properly control cell division mode may result in premature depletion of the stem/progenitor cell pool, or abnormal growth and impaired differentiation. In many tissues, including the brain, normal neural stem cells and oligodendrocyte precursor cells undergo asymmetric cell division through the establishment of cell polarity. Decrease or loss of asymmetric cell division can be associated with reduced differentiation during ageing or impaired remyelination as seen in demyelinating diseases. Glioma precursor cells show decreased asymmetric cell division rates and increased symmetric divisions which lead to the assumption that asymmetric cell division have tumor suppressive function in brain tumors. Cancer stem cells are the malignant counterpart of neural stem cells and undergo asymmetric cell division albeit at low levels. Proper control of cell division mode is therefore not only necessary to generate cellular diversity during development and to maintain adult tissue homeostasis, but may also prevent disease or determine disease progression. Proteins establishing cell polarity are critical regulators of asymmetric cell division. This chapter will review the current knowledge on molecular mechanisms that regulate asymmetric cell divisions in the neural and oligodendroglial lineage as well as the potential connection between polarity machinery and cancer. Such evidences will point to a better understanding of the causal relationship between asymmetric cell division loss and cancer initiation.

Developmental and temporal dynamics of cell division mode in neural stem and progenitor cells

From embryonic stages to adult, the development of the mammalian cortex relies on a temporal and tight control of Neural Stem and Progenitor Cell (NSPCs) mode of division. Stem and progenitor cells across species undergo either symmetric self-renewing divisions (SSD) to expand their pool, symmetric differentiating divisions (SDD) to generate two differentiated progeny, or asymmetric cell divisions (ACD) to generate one self-renewing and one differentiating progeny in a single division (Florio and Huttner, 2014) (Figure 1). The control of cell division mode affects development and size of the brain and sustains neuro- and oligodendroglio-genesis in the adult brain. The regulation of ACD has been intensively studied in the Drosophila neuroblasts (NBs), showing that deregulation of ACD leads to aberrant proliferation and genomic instability (Gomez-Lopez et al., 2014). Recent evidences show that adult neural stem cells (NSCs) and oligodendrocyte progenitor cells (OPCs) undergo ACD (Noctor et al., 2004; S Sugiarto et al., 2011). The study of the regulation of ACD is important for a better understanding of ageing and diseases, as decreased ACD has been evidenced in older brains, brain tumors (S Sugiarto et al., 2011) and other cancer types (Cicalese et al., 2009; Wu et al., 2007).

Figure 1: Evolution of cell division mode of neuronal and oligodendrocyte precursors throughout life in the rodent brain.

(A, B) RGs cells from the ventral hippocampus migrate from the temporal to septal poles and remain as NSCs at adult ages (purple) (Li et al., 2013) while NSCs in the adult SVZ emerge from the embryonic lateral ganglionic eminences (LGE; green) (Fuentealba et al., 2015). Adult OPCs from the CC are first formed in the developing cortex (blue) (Kessaris et al., 2006). (C) Neuroepithelial cells (NE), before the onset of neurogenesis (E11-E12), divide symmetrically to amplify the pool of NE cells. As the developing brain gets thicker, NE processes elongate and transforms into radial glial (RG) cells. Around E12.5, RG divide asymmetrically to generate oligodendrocytes through oligodendrocyte progenitor cells (OPCs) or neurons through intermediate progenitor cells (IPCs). Around birth, most RG cells convert into astrocytes while OPC production continues. (D) Between E13.5 and E15.5, RG cells enter quiescence and become the origin of quiescent NSCs in the adult SVZ. Quiescent NSCs have the ability to enter proliferation and transform into activated NSCs, which give rise to transit amplifying neuronal progenitors (TACs). After an amplification of neuronal progenitors by 3–4 symmetrical TACs divisions and 1–2 symmetrical immature neuroblasts divisions, the latter differentiate into migrating neuroblasts that will reach the olfactory bulbs (OB) to produce neurons. 10% of the NSCs have an oligodendrocyte fate (Ortega et al., 2013) and produce OPCs that will migrate to the corpus callosum (CC). Resident and newly coming OPCs from the CC have the ability to divide symmetrically or asymmetrically depending on intrinsic and extrinsic factors (E Boda et al., 2015; S Sugiarto et al., 2011). (E) In the aging brain, the number of NSCs and OPCs stays unchanged (Capilla-Gonzalez et al., 2013; Daynac et al., 2014). However, aNSCs lengthen their G1 phase and produce less TACs, reducing the number of neuroblasts reaching the olfactory bulbs (Daynac et al., 2016). OPCs also undergo a dramatic age-related cell cycle lengthening (Young et al., 2013). The division mode and rate of NSCs and progenitors from the aging brain is still unclear. LGE (Lateral ganglionic eminence); OPC (oligodendrocyte progenitor cell); IPC (intermediate progenitor cell); qNSC (quiescent neural stem cell); aNSC (activated neural stem cell); TAC (Transit-amplifying cell); Im. Nbs (immature neuroblasts); Mig. Nbs (Migrating Neuroblasts); SVZ (sub-ventricular zone); CC (corpus callosum); SGZ (sub-granular zone).

Division mode of NSPCs during embryonic brain development.

In the mammalian developing neocortex, neural stem cells (NSCs) are the source for the three functional cell types that will populate and shape the adult brain: neurons, astrocytes and oligodendrocytes. (Alvarez-Buylla et al., 2001; Doetsch et al., 1999a). Before the onset of neurogenesis (E≤11–12 in mice), NSCs are located in the monolayer epithelium that constitutes the neural plate and are therefore called neuroepithelial cells (NECs). This monolayer epithelium becomes rapidly pseudostratified, constituting the early neural tube (Florio and Huttner, 2014). To expand their pool, NECs will first undergo symmetric, self-renewing divisions, resulting in a fast thickening of the neuroepithelium (Rakic, 1995). The production of cortical mice neurons begins at E11–12 (Gao et al., 2014). At this point the NECs switch their division mode from symmetric to asymmetric, generating a NEC and a more differentiated cell expressing glial markers, consequently called radial glia cells (RGs) (Gotz and Huttner, 2005). Remaining NECs will then progressively transform into Radial Glial Cells (RG), which line the apical layer of the developing cortex, called the ventricular zone (VZ). RG undergo symmetric self-renewing cell divisions at first but progressively switch to ACDs, and the peak of neurogenesis (E13–18) is associated almost exclusively with ACD, producing a self-renewing RG and either a post-mitotic neuron or an intermediate progenitor (IP) cell (Noctor et al., 2008). The neurons issued from this divisions form the lower cortical layer while IPC will progressively divide in the sub-ventricular zone, once this structure is formed. Some IPC can undergo one or more rounds of symmetric self-renewing divisions to increase their pool and enlarge the SVZ and form the intermediate zone (IZ) but most of them will divide and differentiate into two neurons that will form the upper cortical layers (Hansen et al., 2010; Noctor et al., 2004). All these steps are schematically represented in Figure 1C.

Recent clonal analyses using the MADM (Mosaic Analysis with Double Markers) technique gives an unprecedented look into the remarkably tight and predictable control of the cell division mode of RG progenitors in the formation of the neocortex (Gao et al., 2014). In the developing mouse cortex, RG cells transit from symmetric, self-renewing (SSD) division to asymmetric neurogenic division around E11–E12, and produce a defined number of neurons at the onset of neurogenesis, under a tight control by the transcription factor OTX1 (Gao et al., 2014; Greig et al., 2013).

The number of neurons produced is amplified by a more recently identified type of progenitor, the outer radial glial (oRGs) cells situated in the outer SVZ (oSVZ) of mammals (Fietz et al., 2010; Hansen et al., 2010; X. Wang et al., 2011). In both human and rodents, oRG are produced by ACD of radial glial and will continue dividing asymmetrically during the peak of neurogenesis (E13–18) to produce new lineage of progenitor cells and a large number of neurons (Hansen et al., 2010; X. Wang et al., 2011).

ACD in the subgranular zone of the adult hippocampus

At completion of brain development, most RG cells will detach from the apical surface of the ventricular zone and migrate toward the cortical plates to produce astrocytes and ependymal cells (Kriegstein and Alvarez-Buylla, 2009; N Spassky et al., 2005; Voigt, 1989). However, new neurons continue to be produced in the neonatal brain (Luskin, 1993) and in the adult rodent brain in two main neurogenic niches: the SVZ along the lateral ventricles (Doetsch et al., 1999a; Lois and Alvarez-Buylla, 1993) and the subgranular zone (SGZ) of the dentate gyrus of the hippocampus (Gage et al., 1998; Kaplan and Hinds, 1977). Whether adult NSC similar to their embryonic counterparts divide asymmetric or symmetric is the topic of ongoing research. These investigations are frequently tackled in the context of the questions of whether NSC descend from embryonic RG cell divisions and whether adult NSC division patterns change with age. Several studies indeed point to an embryonic origin or adult NSC (G. Li et al., 2013; Furutachi et al., 2015; Fuentealba et al., 2015) In the developing brain, RGs cells from the ventral hippocampus migrate from the temporal to septal poles and are retained as NSCs in the adult SGZ (Figure 1A,B). NSC in the SGZ exhibit a polarized morphology with a long process extending from the SGZ towards the molecular layer (Mignone et al., 2004). Pulse labeling with thymidine analogue 5-bromo-2-deoxyuridine, BrdU, in a nestin-cyan fluorescent protein (CFP) reporter transgenic mouse, showed that NSC have a low level of proliferation and are therefore also called quiescent neural progenitors (QNP). In situ analyses of BrdU-labeled, Gfap+ immunostained reporter transgenic mice further revealed that QNP divide asymmetrically to generate a Gfap+ NSC and a transient amplifying cells, also called amplifying neural progenitors (ANPs). The long process and Gfap are inherited by the QNP and not the ANP, which makes it possible to identify the two different daughter cells by morphologic criteria. ANPs then go on to generate neuronal progenitor cells or neuroblasts, which will ultimately differentiate into granule neurons (Encinas et al., 2006). Genetic inducible fate mapping using either Nestin-CreER (Francesca Balordi and Fishell, 2007) or Gli-CreER (Encinas et al., 2011) lines confirmed the presence of QNP/ANP pairs and thereby corroborated that QNP undergo ACD. A subsequent report by Bonaguidi et al used a clonal labeling strategy to genetically mark single cells in Nestin-CreER^T2; Z/EG mice and to trace the fate of dividing cells. In addition, they applied mosaic-analysis with double markers (MADM) to label clones with two colors. Cell pairs were identified based on the close proximity of two cells and cell fate of daughter cells was determined by morphology and co-staining for Gfap, neuronal progenitor marker Tbr2, neuronal marker Prox1 and astrocyte marker S100β as well as proliferation marker MCM2. Both approaches gave similar results, which showed that QNP predominantly divide asymmetric to self-renew and generate a daughter cell of distinct fate and morphology. Symmetric self-renewing divisions yielding two QNP cells were found at low rates and of the progeny of these divisions, one either differentiated or continued to self-renew or, both cells differentiated. These data indicate that a fraction of QNP are also terminally differentiated following symmetric divisions. As expected from previous studies, the majority of divisions generated QNP/Neuronal lineage cell pairs, whereby the QNP re-entered quiescence and the non-QNP cell was either an ANP/intermediated progenitor or an immature neuron, which continued to proliferate. Surprisingly, this study also found QNP/mature astroglia pairs, whereby both, the QNP and the astroglia cells seize to proliferate. Previous studies which used BrdU labeling and virus injection to track cell pairs have missed these types of divisions probably due to technical limitations (Bonaguidi et al., 2011). Bonaguidi et al. further corroborated earlier data showing that the intermediate progenitors or ANPs divide symmetrically. Taken together, several conclusions can be drawn from Bonaguidi et al and earlier studies, including the fact that NSC are largely quiescent in the adult brain. NSC appear to re-enter quiescence following an ACD and are very heterogeneous in their output, using different cell division modes to self-renew, in the same temporal window. Lastly, NSCs exhibit the capacity for multi-lineage differentiation and long-term self-renewal through ACD. Despite these advances in identifying cell division mode in the dentate gyrus, and correlating it with cell fate, more sophisticated techniques advances such as live imaging of NSC divisions in situ will be necessary to improve our understanding of how the heterogeneity in cell division mode is achieved and regulated. Several environmental factors and stimuli have been identified that influence hippocampal neurogenesis. Studies in the 1990s have already shown that an enriched environment increases the number of hippocampal neurons in mice (Kempermann et al., 1997). On the other hand, glucocorticoid treatment mimics stress and exacerbates the loss of hippocampal granule cells due to Kainate-induced injury in the adult rat brain (Stein-Behrens et al., 1994). The molecular effects of stress and environmental changes on cell division mode are largely unknown. Interestingly, adult mice exposed to prolonged treatment with the antidepressant drug and serotonin reuptake inhibitor (SSRI) fluoxetine show increased cell proliferation and elevated numbers of new neurons in the dentate gyrus (Malberg et al., 2000). These changes are dues to increased symmetric divisions of ANP and not caused by increased rates of QNP ACD, suggesting that ANP are the fluoxetine target population in the brain (Encinas et al., 2006). These studies underline the importance of ACD for adult neuro- and gliogenesis. Further insights into the molecular mechanism that regulates cell division mode are needed to address the questions whether cell division mode is a cell intrinsic property of NSC and/or whether it is regulated by extrinsic signals from the local niche or distant sources.

Cell division mode in the adult subventricular zone

The adult SVZ encompasses the ventricle with cerebrospinal fluid, ependymal cells, NSC, NSC progeny and capillaries (Zaman Mirzadeh et al., 2008; Shen et al., 2008). In the SVZ, similar to the SGZ, adult NSCs are mostly quiescent and only a fraction of the NSCs (8.6% according to Ponti et al., 2013) from the SVZ are actively proliferating (Furutachi et al., 2013; Ponti et al., 2013).

Adult SVZ NSC, frequently referred to as type B cells, divide to generate transit amplifying progenitors (TACs; Doetsch et al. 1997; Doetsch, Garcia-Verdugo, et al. 1999; Figure 1). TACs in turn produce post-mitotic progeny, including neuroblasts, also called type A cells, (Doetsch et al., 1999b) and, to a lesser extent, oligodendrocytes (Menn et al., 2006). Until recently, it was proposed that RG cells retract their processes and differentiate into astrocytes and ependymal cells at embryonic stages (Nathalie Spassky et al., 2005; Voigt, 1989). It was therefore not clear where adult NSC originate and whether they would undergo ACD, similar to their embryonic counterparts, the RG cells. The origin of adult NSC was elusive until a recent lineage tracing study identified a slowly dividing population of embryonic NSCs as an origin of the majority of young adult SVZ NSCs (Furutachi et al., 2015). These embryonic quiescent NSC express high levels of the cyclin-dependent kinase inhibitor p57 that helps to maintain them in a slow-proliferative condition until adulthood. Another recent study has shed light on the origin of adult NSC and used a retroviral library carrying over 100 000 genetic tags to prove that the majority of the future adult NSCs are produced between E13.5 and E15.5 and enter quiescence until their eventual reactivation at adult ages (Fuentealba et al., 2015) (Figure 1C,D). Moreover, the ventral SVZ contains a specific population of Nkx2.1 NSCs derived from Nkx2.1 expressing cells present in the embryonic brain at early ages (E13.5–15.5), indicating that the regional specification of adult NSCs is triggered by ventral germinal zones at early embryonic ages, contributing to a unique positional heterogeneity of adult SVZ niches (Delgado and Lim, 2015).

While NSC in the SGZ have a long process, similar to embryonic RG cells in the ventricular zone, that was easy to detect in thin sections of the adult brain, this method was not suited to fully outline the morphologic polarity of NSC in the adult SVZ. The Alvarez-Buylla lab analyzed immunostainings of whole-mount preparations of the SVZ, which were isolated from adult mice injected into the lateral ventricle with adenovirus expressing green fluorescence protein from a Gfap promoter. Using these approaches, they showed that aNSC extend a very long basal process that contacts a blood vessel. In addition, these cells have a small ciliated domain that touches the lateral ventricle and is connected with the ependymal layer through adherens junctions. They call the polarized aNSC B1 cells, in juxtaposition to the B2 cells, which are the SVZ astrocytes with a bushy morphology (Z Mirzadeh et al., 2008). Due to the embryonic origin and some morphologic similarities, such as intrinsic cell polarity, it could be assumed that adult NSC undergo ACD in the adult brain. Given the complexity of their niche and their high level of quiescence and a clear morphologic mark of polarity, it has been challenging to assess the cell division mode of adult NSCs in situ.

Live imaging of SVZ derived cells ex vivo has provided first insights into the cell division modes utilized by NSCs to generate their progeny in the adult rodent brain (Figure 1C). Continuous live imaging of primary cultures of SVZ cells was conducted at low density and under adherent conditions without growth factors (Costa et al., 2011). Slow-dividing NSCs from the adult SVZ have an extended cell cycle (≥36 hours) and undergo one or two symmetric divisions to give rise to proliferating “activated” NSCs (aNSCs). These experiments revealed that the majority of cells divided symmetrically to give rise to Tuj1-positive neuronal progenitor cells. In addition, some cells showed a considerably slower cell cycle (> 52hrs) and those were large cells and they divided to generate morphologically identical cells in a single symmetric division. The progeny of this symmetric division went on to generate an asymmetric lineage tree, with one branch retaining astro-glial fate and eventually entering quiescence and the second branch expressing neuronal markers. The slow-dividing founder cells of these asymmetric clones were identified as Gfap+ cells with the help of a transgenic Gfap-RFP line and were considered NSCs. The model arising from these data is that NSCs exit quiescence, whereby they upregulate Gfap and divide symmetrically to give rise to two activated, fast-proliferating NSCs (aNSC), which in turn divide asymmetrically into aNSCs/TAC pairs. TACs down-regulate Gfap expression and divide typically two to three (up to five) times symmetrically to give rise to mostly neurons, whereas aNSCs divide up to two times, before re-entering quiescence. This suggests that a reservoir of quiescent NSC is maintained after a bout of divisions and that aNSC use ACD to theoretically undergo unlimited self-renewal and at the same time to generate TACs. However, due to the limitations of the experimental approach, it cannot be ruled out that aNSC terminally differentiate and become permanently post-mitotic in vivo, similar to what has been proposed in QNP or the dentate gyrus (Encinas et al., 2006). Subsequent in situ analyses using serial injections of thymidine analogs into mice, followed by detection of specific cell fate by immunofluorescence in the brain of injected mice for the most part corroborated ex vivo findings of the cell division rate of NSCs and their direct progeny (aNSC: one TACs: three amplifying divisions, Nbs: one or two divisions). However, these analyses did not unequivocally distinguish between asymmetric and symmetric cell division mode of activated NSCs (Ponti et al., 2013).

More recently, Calzolari et al. have used clonal analyses with a mouse line targeting NSCs (GLASTCreERT2) crossed with the ‘Confetti” multicolor reporter to show that a single activated NSC generates multiple subsequent subclones by supposedly undergoing two to three rounds of ACD (Calzolari et al. 2015). Subclone analyses are consistent with the earlier study by Ponti et al (Ponti et al., 2013), showing 3 to 4 rounds of symmetrical divisions TACs before giving rise to neuroblasts, which symmetrically divide 1–2 times before migrating to the OBs. Long-term clonal analyses carried out over the course of 4–6 months indicated that aNSC continue to generate new progeny once activated through multiple (2–3) rounds of divisions and then they terminally differentiate or die, rather than enter quiescence (Calzolari et al., 2015). This study together with an earlier report (Francesca Balordi and Fishell, 2007) indicate that the self-renewal potential of aNSC is limited. Further studies are needed to unequivocally determine whether the cell division mode of q/aNSC in the adult SVZ is asymmetric or symmetric.

One hallmark of a classic ACD is that cell fate markers and determinants distribute unequally during mitosis. Gfap would be such a marker expected to segregate asymmetric during aNSC divisions, which generates an aNSC/TAC pair, because TACs are negative for Gfap expression. Since Gfap is expressed in the processes of NSC and these can be quite removed from the cell body identified by nuclear staining and thymidine incorporation, it has been very difficult to assess whether Gfap is unequally inherited during aNSC mitosis. Other cell fate determinants, that are proposed to segregate asymmetrically during aNSC mitosis in the adult SVZ include epidermal growth factor receptor (EGFR; Sun et al. 2005), dual specificity kinase 1 (Dyrk1; Ferron et al. 2010) and Delta-like-ligand1 (Dll1; Kawaguchi et al. 2013).

Their potential function in ACD will be discussed in a separate paragraph below. Further proof that cell fate markers and determinants segregate asymmetric may come from time-lapse imaging of mitoses of aNSC, which express fluorescence-tagged cell fate determinants such as Dll1. This approach has so far been unattainable for adult brain although it is feasible for tissue sections isolated from embryonic neocortex (Kawaguchi et al., 2013; Noctor et al., 2004). It is likely that ACD of adult NSC are rather rare in the SVZ and yet it is deemed an important puzzle to solve. One reason for distinguishing between asymmetric and symmetric cell division mode is that they are regulated very differently. A better understanding of the regulatory mechanism behind asymmetric aNSC divisions, should they exist, will pave the way for therapeutic approaches utilizing aNSC for regenerative purposes. On the other hand, aNSC are cells-of-origin of adult gliomas in experimental mouse models, and understanding the mechanism by which they are activated and divide might lead to insights into neoplastic transformation and tumorigenesis.

The origin of oligodendrocytes in the central nervous system.

Oligodendrocytes are specialized, myelin-producing cells that form a myelin sheath around axons and thereby enable them to carry action potentials by salutatory conduction. Oligodendrocytes arise from multiple pools of oligodendrocyte precursor cells (OPCs) present in the developing brain and spinal cord. The production of OPC in mice starts in the spinal cord and occurs in three distinct phases: first, OPC are born at E12.5 at the ventral neural tube, then at E15.5 at the dorsal neural tube, and after birth at the central canal subependyma (Gallo and Deneen, 2014; Lu et al., 2000; Orentas et al., 1999; Rowitch and Kriegstein, 2010; Vallstedt et al., 2005). Similarly, genetic fate mapping using three different transgenic mice each expressing Cre recombinase from a different cell-type specific promoter, have shown that OPCs arise in three distinct phases in the mouse forebrain and these are characterized by the selective expression of transcription factors (Kessaris et al., 2006). The first wave occurs at E12.5 from the median ganglionic eminences under the control of Shh signaling and transcriptional regulator Nkx2.1 to colonize the entire forebrain (Kessaris et al., 2006; Klämbt, 2009; Tekki-Kessaris et al., 2001). At E15.5, transcriptional regulator Gsx2-positive OPCs are produced in the lateral ganglionic eminences of the medial forebrain and migrate to the neocortex (Chapman et al., 2013; Kessaris et al., 2006; Klämbt, 2009). The final phase of OPC generation occurs just before birth with OPCs, expressing the transcription factor Emx1, projecting radially from the dorsal SVZ to populate the newly forming adjacent corpus callosum (Kessaris et al., 2006). Upon reaching their final destination in the adult brain at perinatal stages, OPCs will proliferate locally and expand. OPC subsequently exit the cell cycle and produce the majority of myelinating oligodendrocytes within postnatal week three and four (Miller RH, 2002). Cre-lox fate mapping studies using NG2-Cre mice found that OPC can give rise to protoplasmic astrocytes (Zhu et al., 2008). In contrast, in vivo lineage tracing using a tamoxifen-inducible NG2-Cre driver (X Zhu et al., 2011) or a PDGFRα-Cre driver (Kang et al., 2010) did not observe any astrocyte production but only oligodendrocytes in grey and white matter, proofing that postnatal and adult NG2+ OPC become restricted to the oligodendrocyte lineage. Moreover, it has been debated as to whether NG2+ OPC are neuronal progenitors in the postnatal brain. Fate mapping results using PDGFRα-CreER^T2 or PLP-CreER^T mice (Guo et al., 2010; Rivers et al., 2008) indicated production of a small number of piriform neurons. Using different Cre-drivers, Zhu et al (X Zhu et al., 2011) and Kang et al (Kang et al., 2010) detected no differentiation of NG2+ OPCs into neurons consistent with results obtained with Olig2-CreERT2 transgenic mice (Dimou et al., 2008). Continued analyses of NG2+ OPC fate in multiple OPC-specific Cre drivers will be crucial in the following years to fully unravel the lineage specificity of OPC. Collectively, the results suggest that not all Cre driver lines are equally suited to study OPC cell fate and that OPC marked by co-expression of NG2 and PDGFRα are restricted to the oligodendrocyte lineage. In addition to their embryonic origin, OPC arise from NSCs in the dorsal/ventral SVZ in the adult brain (Gonzalez-Perez et al., 2009; Menn et al., 2006; Ortega et al., 2013), whereby around one in ten adult NSCs are capable to give rise to oligodendrogenic lineage cells (Figure 1C) (Ortega et al., 2013). Single cell tracking in vitro suggested that neuronal and oligodendroglial clones issued from adult NSCs are distinct, and after two to three rounds of symmetrical expansion, a sub-population of TACs are able to produce oligodendrocyte progenitor cells (OPCs) (Ortega et al., 2013). Whether OPC can arise from cells other than TACs and whether resident and SVZ-derived OPC are functionally distinct are open questions, which will be addressed by future research.

ACD of oligodendrocyte progenitor cells in the adult brain

OPC proliferation peaks in the mouse brain during early postnatal stages, however, OPC continue to proliferate (Dimou et al., 2008; Kang et al., 2010; Rivers et al., 2008). Despite a decrease in OPC proliferation and differentiation with age parenchymal OPC density is stable and OPC continue to self-renew and generate myelinating oligodendrocytes (Kang et al., 2010; K Psachoulia et al., 2009; X Zhu et al., 2011). These observations raised the questions how OPC maintain their pool and whether they use ACD to balance self-renewal with differentiation. Wren and Noble first documented the notion that OPC undergo long-term self-renewal through ACD by studying nerve progenitors, called O-2A, which they have isolated from the adult rat optic nerve. They discovered that clones derived from single adult O-2A cells and grown under adherent conditions contained both, differentiated post-mitotic oligodendrocytes and proliferating progenitor cells. These observations led them to speculate that O-2A cells undergo ACD to self-renew and generate mature cells. They also found albeit at lower frequency oligodendrocyte-only and progenitor-only clones, which would indicate that O-2A cells also undergo symmetric, self-renewing as well as symmetric, differentiating divisions (Wren et al., 1992). Nineteen years later, their predictions were directly tested and validated experimentally in a study of murine adult OPC, which were acutely isolated from the corpus callosum of P60-P90 mice (S Sugiarto et al., 2011). OPC were subjected to a pair cell assay, that has been initially designed to distinguish between symmetric and asymmetric cell divisions of embryonic neural progenitors (Shen et al. 2002; Y Sun et al. 2005). Detailed immunocytochemical analyses of cell pairs indicated that the NG2 proteoglycan itself was inherited unequally between daughter cells within cell pairs (Sugiarto et al., 2011). The rates for asymmetric versus symmetric, self-renewing cell divisions in vitro were balanced. In vitro studies further indicate that when left in culture for 48 hours the NG2-negative daughter cell acquires expression at a greater frequency while the NG2+ daughter has a greater frequency of proliferation (Sista Sugiarto et al., 2011). These findings indicated that mammalian adult OPCs undergo ACD to self-renew and generate mature oligodendrocytes at a one-to-one ratio. Immunohistochemistry analyses of dividing OPCs in the adult mouse brain have indeed confirmed the presence of symmetrically dividing OPCs producing two NG2⁺ OPCs, and asymmetrically dividing OPCs that generate one NG2⁺ OPC and one NG2-negative daughter.

Subsequent in vivo and ex vivo studies have confirmed that OPC divide asymmetrically as well as symmetrically in the adult brain. Live imaging of single cells on brain slices from early post-natal stages of NG2CreBAC transgenic mice, followed by immunostaining for differentiation marker CC1, has revealed that OPCs can self-renew and/or differentiate and that they use three cell division modes, ACD, symmetric self-renewing (SSD) or symmetric differentiating divisions (SDD), albeit at different rates (X Zhu et al., 2011). In vivo cell pair studies, using thymidine analogs in the adult brain to identify dividing cell pairs, further confirmed that OPCs undergo ACD. In these studies, co-immunofluorescence to detect the thymidine analogs and OPC-specific molecules such as NG2 and PDGFRα discovered that these markers inherit unequally between daughter cells at a rate of 35–45%. Conversely, previous studies using time lapse imaging in adult cortex have only evidenced symmetric OPC divisions (Hughes et al., 2013; Kukley et al., 2008). Neuro-anatomical as well as experimental differences between these studies might explain the distinct outcome of the analyses of cell division mode. Hughes et al. visualized OPC in the outer layers of the cortex while the other two studies focused on corpus callosum (Sista Sugiarto et al., 2011) as well as the entire cortex (Enrica Boda et al., 2015). Since numerous studies have indicated that OPC are heterogeneous in proliferation and differentiation rate depending on their neuro-anatomical differences (Hill et al., 2014; Kang et al., 2010; Young et al., 2013) and response to growth factor signaling (Hill et al., 2013), it is feasible that cell division mode show regional variations. The paper by Hughes et al. also did not investigate NG2 segregation during mitosis by time-lapse videomicroscopy but rather distribution of eGFP reporter expression under the control of the NG2 promoter. The stability of eGFP reporter after mitosis might have concealed the asymmetric distribution of NG2 in the sister cells (Hughes et al., 2013). As shown by two independent studies, OPC specific molecules like NG2 and PDGFRα are asymmetrically distributed in P20-P60 mice at about the same rate, with ~40% of asymmetric division, ~50% of symmetric, self-renewing divisions and the remaining ~10% of symmetric, differentiating divisions (Enrica Boda et al., 2015; Sista Sugiarto et al., 2011). Additional staining for phospho-histone-H3 further confirmed that cell pairs with asymmetric NG2 distribution result from a single OPC division (Sista Sugiarto et al., 2011). Cell fate analyses of cell pairs using thymidine analog injections and immunostaining for differentiation markers further revealed that single OPC produce daughter cells of distinct cell fate and that asymmetry in cell fate persists for months (E Boda et al., 2015). Taken together, the evidence for asymmetric OPC divisions and their role in balancing differentiation and self-renewal is strong and future studies will determine the mechanism and functional importance of this division type.

Aging and cell division mode

Neural stem cells and ageing

The production of neurons and oligodendrocytes occurs throughout adult lifespan in the mammalian brain. In the human brain, both the SGZ of the hippocampus (Knoth et al., 2010) and the SVZ of the lateral ventricle (Sanai et al., 2011; C. Wang et al., 2011a) contain proliferating neuronal precursors, with a peak rate in the perinatal period. Adult neurogenesis is important for learning, emotions and memory and abnormal adult neurogenesis has been linked to brain diseases, including Alzheimer’s disease, demyelinating disease, depression and brain tumorigenesis (Liu and Song, 2016). The human SVZ and rostral migratory stream (RMS), where neuroblasts form chain-like structures to migrate into the olfactory bulb, send a large number of doublecortin+ (Dcx+) neuroblasts to the olfactory bulb and the prefrontal cortex during the first 18 months of age and according to the report by Sanai et al, this activity subsides in older children. The RMS was not detected in the adult brain in one study (Sanai et al., 2011) and in a second study only sparse, migrating neuroblasts were found in what the authors call an “RMS-like pathway” (C. Wang et al., 2011b). A few mitotically active Dcx+ cells were found in the SVZ which frequently appeared as doublets, which suggests that neural progenitors divide symmetrically to generate neuroblasts in the adult brain (C. Wang et al., 2011b). Collectively, these studies suggest that the SVZ-derived neuron production is highest in young children but drops to a low frequency in adults. Adult neurogenesis appears to be more relevant in the hippocampus, were a breakthrough studies using ¹⁴C level analyses in genomic DNA have proved that nearly all dentate granule neurons turn over in the adult human hippocampus, with ≈700 new neurons per day in each hippocampus (Spalding et al., 2013).

In the rodent, the aging brain is subject to a progressive reduction in proliferating progenitor cells from the SGZ and SVZ, leading to a dramatic drop in the number of neurons produced during aging (Blackmore et al., 2009; Bouab et al., 2011; Enwere et al., 2004; Maslov et al., 2004; Tropepe et al., 1997). Accordingly, the number of TACs and Neuroblasts is strongly decreased with aging (F Balordi and Fishell, 2007; Blackmore et al., 2009; Daynac et al., 2014; Shook et al., 2012). Unexpectedly, the pools of quiescent and activated NSCs remain stable until middle-age in mice (12 months) (Daynac et al., 2014; Piccin et al., 2014), but activated NSCs lose their proliferative capacities (Ahlenius et al., 2009; Capilla-Gonzalez et al., 2014; Daynac et al., 2016, 2014) and progressively enter into quiescence (Bouab et al., 2011; Lugert et al., 2010). Clonal analyses with confetti-mice revealed that activated NSCs undergo self-renewing ACDs, but clones size is progressively reduced with age, suggesting a limited self-renewal capacity of activated NSCs (Calzolari et al., 2015). Further recent evidence suggests that activated NSCs cell cycle is altered as early as 6 months of age in mice, due to a G1 phase lengthening in response to an age-related vascular over-production of TGFβ1 (Figure 1E) (Daynac et al., 2016, 2014; Pineda et al., 2013). The activation process of quiescent NSCs is controlled by Shh signaling. Indeed, long-term activation of Shh signaling through Ptc inactivation in NSCs provokes quiescent NSCs to switch their division mode from ACD to SCD and accumulate at a quiescent, inactive state (Ferent et al., 2014). Further studies are required to address whether these long-lived quiescent NSCs can be activated by external stimuli and reproduce a functional niche, as observed with anti- TGFβ1 treatment in old and irradiated mice (Pineda et al., 2013).

OPC and ageing

The aged brain reportedly has decreased remyelination which slows recovery from demyelinating diseases such as multiple sclerosis (Sim et al., 2002). The underlying causes for this impairment are partially attributable to age-related changes in the oligodendrocyte lineage, which are not fully understood. OPCs are a reservoir for differentiating oligodendrocytes and continue to produce new myelinating oligodendrocytes throughout adulthood. Thus, decreases in OPC populations with age could potentially impact the rate of myelination and changes in cell division mode could be the factor in changing OPC frequencies. OPC numbers, however, were found to be stable during aging in mouse, suggesting that there is no exhaustion of the progenitor pool (Capilla-Gonzalez et al., 2013; Rivers et al., 2008; van Wijngaarden and Franklin, 2013). Between post-natal and young adult age notably, recent studies using cumulative labeling with EdU showed that all OPCs throughout the central nervous system are able to undergo at least one cell division and that a large fraction of newly-formed oligodendrocytes survive (Clarke et al., 2012; Young et al., 2013). Importantly, while OPC numbers are stable, their capacity to differentiate into myelinating oligodendrocytes significantly decreases with the transition from young adult (P2) to adult (P60) in the mouse brain, in one study (from 66% in P62 versus 39% in P120 mice; Xiaoqin Zhu et al., 2011). This age-related decrease in differentiation capacity is speculated to be an underlying cause for impaired remyelination and repair of demyelinated lesion in old versus young rat brains (Sim et al., 2002). Changes in cell division mode, such as a switch from ACD to SSD could potentially decrease rates of differentiation, since asymmetrical distribution of NG2, EGFR, GPR17, and PDGFRα is required for the long-term generation of cells distinct fate and control of proliferation versus differentiation (E Boda et al., 2015; S Sugiarto et al., 2011). To address whether the ability to generate daughter cells through ACD changes with age, Boda et al. quantified the asymmetric and symmetric distribution of PDGFRα in OPC cell pairs in mice at different ages and for up to one year. While the rate of ACD of OPC increases significantly when mice transition from early post-natal to adult stages (between P20 and P60) it subsequently decreases again in old mice (12 months of age). This age-related decrease is accompanied by an increase in symmetric, self-renewing divisions and a significant decline in newly generated OPC pairs (E Boda et al., 2015). A separate study addressed whether OPC change cell cycle dynamics with age and found a dramatic increase in cell cycle lengths of NG2+ OPC in post-natal versus middle-aged mice (Figure 1E) by around 2/3 of a day, every day after birth (cell cycle lenghts (CCL) <2 days at P6; CCL = 36 days at P60 and >70 days in P240 mice) (Konstantina Psachoulia et al., 2009; Simon et al., 2011; Young et al., 2013). Collectively, these data suggest that OPC overall lower their proliferation rate by becoming more quiescent and maintain the progenitor pool by increasing symmetric, self-renewing divisions. Decreased rates in ACD occur with age by a completely unknown mechanism and associate with decreased differentiation potential. Hypothetically, these switches in cell division mode could be the underlying cause for age-related decrease in differentiaion and myelin repair but much more research will be needed to proof or dissprove this notion and to fully elucidate the dynamics of cell division of OPCs and its functional implications during aging.

Regulation of ACD by cell polarity proteins

ACD are an evolutionary conserved mechanism utilized by prokaryotes as well as unicellular and multicellular eukaryotes to generate cell fate diversity within daughter cells. Drosophila melanogaster embryonic neuroblasts (NBs) are stem-like cells, which undergo several rounds of ACD, during which determinants of differentiating fate concentrate at the basal cell cortex before mitosis and segregate unequally during cytokinesis, to generate each time another NB and a more restricted progenitor called ganglion mother cell (GMC). We glean many hypotheses on ACD in mammalian stem and progenitor cells from the enormous amount of studies in Drosophila NBs. NBs break symmetry in a stepwise process, which exemplifies the classic, cell-intrinsic type of ACD. Prior to mitosis, NBs establish cell polarity and as a critical next step align the mitotic spindle along the axis of polarity and, simultaneously organize cell fate determinants in subcellular compartments near the mitotic spindle poles, such that those cell fate determinants with equal functions segregated together and only into one of the future daughter cells (Gomez-Lopez et al., 2014). Here, we will focus on the mechanisms by which cell polarity is established and how it links to the subsequent steps of ACD.

How is cell polarity established?

First-generation Drosophila NBs delaminate from the polarized neuro-ectoderm and from it inherit the partition-defective-3 (Par3) homolog Bazooka (Baz) as an apical polarity cue. Baz/Par3 assembles a polarity complex in NBs during late interphase/early prophase (Schaefer et al., 2000). Baz/Par3 binds to and activates Cdc42, a GTPase of the Rho family (Atwood et al., 2007), which in turn recruits atypical protein kinase C (aPKC) and the aPKC inhibitory subunit Par6 (Atwood et al., 2007; Peterson et al., 2004). The apical complex binds the adaptor protein Inscuteable (Insc) (Wodarz and Ramrath, 2000) and thereby initiates the assembly of partner of Insc (Pins) (Schaefer et al., 2000) and the heterotrimeric G protein coupled subunits Gαι and Gβγ. The WD40 protein lethal giant larvae (Lgl) binds to aPKC/Par6, which leads to a mutual antagonism of Lgl and aPKC, as demonstrated by loss-of-function experiments. When bound to the polarity complex Lgl restricts aPKC localization to the apical cortex and thereby helps to maintain cell polarity. Genetically enforced reduction of aPKC levels suppresses the NBs overproliferation phenotype of a Lgl1 hypomorphic mutation. Moreover, aPKC loss-of-function mutant larval NBs loose the polarized localization of Lgl and cell fate determinant Miranda, indicating that aPKC - despite its antagonism of Lgl function - is required for apical binding of Lgl and subsequent asymmetric distribution of cell fate determinants (Rolls et al., 2003). After delaminated, first-generation NBs have undergone cytokinesis, the NB daughter downregulates cell apical polarity components in interphase. Several studies indicated that to the contrary the centrosome remains stable and provides spatial memory for proper spindle axis formation in subsequent, divisions (Januschke et al., 2011; Rebollo et al., 2007; Siller and Doe, 2009). The centrosome also shows asymmetry, when it duplicates into a newer, “daughter” and an older, “mother” centriole and distributes between daughter cells in a non-random fashion, with the daughter centriole consistently remaining in the NB (Januschke et al., 2011; for review see Conduit, 2013). Centrosome asymmetry is regulated by Polo kinase phosphorylation of human CENTROBIN homolog CnB which promotes the formation of aster microtubules and anchors the centrosome (Singh et al., 2014; Januschke et al., 2013) and provides directionality for the mitotic spindle. More recent studies have shown that NBs mutant for centrosome asymmetry, undergo seemingly normal ACD (Singh et al., 2014), suggesting that centrosome asymmetry is not the dominant mechanism to establish proper spindle alignment. Noteworthy, mammalian radial glial cells non-randomly segregate their centrosomes during ACD and centrosomal components control cell fate (Wang et al., 2009), which suggest that centrosome asymmetry is a conserved feature of asymmetric dividing cells.

Cortical cell polarity as dominant mechanism to control mitotic spindle alignment

A dynamic crosstalk between the apical complexes (Par/aPKC and Pins/G-protein) with centrosomes and astral microtubules is crucial for proper spindle orientation in Drosophila NBs. In mitosis, the centriole duplicate and the daughter centrosome generates large aster microtubules, which bind and link the Drosophila NuMa homologue and coiled-coiled domain protein mushroom body-defective (Mud) to Pins/Gαi to form a large apical complex which tethers the mitotic spindle to the cortex and aligns it along the apico-basal axis (Bowman et al., 2006; Izumi et al., 2006; Siller et al., 2006). Further supporting the cortical-microtubule interactions, the MAGUK Disc large (Dlg) binds to kinesin Khc-73 at the microtubule plus ends (Siegrist and Doe, 2007, 2005) and this interaction feeds back to the cortex and maintains polarity. Spindle positioning is linked to global control of mitosis by Polo and AurA kinases (Lee et al., 2006a; Wang et al., 2007, 2006). In mammalian neurogenesis, RG cell divisions are biased towards aligning the mitotic spindles within 30° of the ventricular surface (Konno et al., 2008; Postiglione et al., 2011). LGN, an orthologue of Pins, is highly expressed in RG and localizes to the lateral membrane of dividing RG cells (Du et al., 2001). Genetic ablation (Konno et al., 2008) or siRNA-mediated knockdown (Shitamukai et al., 2011) of LGN alters the position of the mitotic spindle, which leads to asymmetric inheritance of the Par polarity complex and impaired differentiation (Shitamukai et al., 2011). The function of mammalian Insc is less clear and appears be opposite from LGN, which is different from Drosophila and does not match a conserved function of Insc as a cellular linker of the Par complex and LGN/Gαι. The phosphoprotein Treacle at the centrosome is critical not only for mitosis but also for correct spindle positioning and mitotic progression potentially through its interaction with Polo-kinase 1 (Plk1), a mammalian Polo homolog (Sakai et al., 2012). A Treacle/Plk1-controlled checkpoint that links cell polarity with mitotic progression has been proposed for RG. A potentially similar polarity checkpoint is relevant in cancer stem cells where it has been shown the Plk1 activity is critical for ACD and vice versa cell polarity is required for progression through mitosis (Lerner et al., 2015).

How does cell polarity link to cell fate control?

In metaphase, the mitotic kinase Aurora A (AurA) phosphorylates Par6 and thereby activates aPKC (Ogawa et al., 2009; Wirtz-Peitz et al., 2008; Wodarz and Ramrath, 2000). Active aPKC phosphorylates Lgl (Betschinger et al., 2003; Wirtz-Peitz et al., 2008) and Numb, thereby releasing them from the apical cortex (Smith et al., 2007; Wirtz-Peitz et al., 2008). Phosphorylated Numb then localizes to the basal side of the cell (Smith et al., 2007) by interacting with Polo kinase-activated Partner of Numb (PON) (Lu et al., 1998; Wang et al., 2007) and once inherited by the GMC prevents self-renewal, predominantly by antagonizing Notch signaling (Wang et al., 2006). Genetically enforced deletion of the two mammalian homologs of Drosophila Numb, Numb and Numblike (Numbl), in the postnatal SVZ, resulted in defects in lateral wall integrity and impaired neurogenesis. The results of the mouse double-knock out study indicate a conserved role in promoting neuronal differentiation. Since they did not show that Numb/Numbl regulate ACD in the adult NSC lineage nor that the defects are Notch-dependent it is unclear to what extent the function of Numb is conserved (Kuo et al., 2006). Interestingly, in vitro studies have shown that in the murine brain a fraction of adult SVZ-derived NSCs asymmetrically distributes activated Notch and EGFR between daughter cells. Activated Notch is unequally distributed during NSC division and it has been shown to directly regulate Egfr transcription (Andreu-Agulló et al., 2009). Moreover, Dyrk1a kinase is asymmetrically segregated to the EGFR^high daughter, where it prevents EGFR degradation (S. R. Ferron et al., 2010). Notch ligand delta-like 1 (Dll1) was found to be expressed in aNSC and TACs and Dll1 deletion promotes the activation of qNSC in the in the adult SVZ. Dll1 protein segregates asymmetrically in embryonic and adult NSC in vitro and associates with neuronal progenitor fate. Given that Notch is expressed in qNSC, the data suggest that Dll1 expression in a NSC and TACs provides a signal to maintain qNSC by activating Notch receptor (Kawaguchi et al., 2013).

Drosphila NBs have a second adaptor-cell fate determinant system consisting of Miranda (Mira) and its binding partners which localize asymmetrically and segregate to the basal GMC (Shen et al., 1998). APKC also phosphorylates Mira (Atwood and Prehoda, 2009; Wirtz-Peitz et al., 2008), but the functional significance of aPKC phosphorylation for Mira asymmetry is not clear (Lee et al., 2006a; Wang et al., 2006) and studies imply a range of alternative mechanisms of Mira localization acting upstream or in parallel to aPKC. These include proteasome-independent mechanism of ubiquitylation, kinase activation, the anaphase-promoting complex (Slack et al., 2007), MyoVI-directed vesicle transport (Petritsch et al., 2003) and passive diffusion (Erben et al., 2008). Loss of Miranda function leads to aberrant cell fate specification due to mislocalization of cell fate determinants, including the transcriptional regulator Prospero (Pros) (Ikeshima-Kataoka et al., 1997; Shen et al., 1997), the double stranded (ds) RNA-binding protein Staufen (Stau) (Matsuzaki et al., 1998; Shen et al., 1998) and the NHL-domain protein Brain tumor (Brat) (Betschinger et al., 2006; Lee et al., 2006b). Staufen2 (Stau2), the mouse homolog of Staufen, is apically localized in RG cells and preferentially segregates to the progenitor daughter (Kusek et al., 2012; Vessey et al., 2012). The Stau loss-of-function phenotypes cannot entirely be reconciled with each other in the two studies, making it difficult to pinpoint the exact role of Stau in the mammalian brain. Importantly, Ssimilar to Drosophila Stau, Stau2 is part of a ribonucleoprotein complex that cargos mRNAs for mammalian homologs of prospero and brat, Prospero-related homeobox 1 (Prox1) (Vessey et al., 2012) and Tripartite-motif containing 32 (Trim32) (Kusek et al., 2012), respectively. Prox1 is known to mediate cell-cycle exit and neurogenesis in the neural retina (Dyer et al., 2003) and promotes oligodendrocyte identity of adult NSC in the SVZ (Bunk et al., 2016) and inhibits proliferation and promotes differentiation of NG2+ OPC (Kato et al., 2015). Trim32 induces neurogenesis (Schwamborn et al., 2009). These data collectively suggest that, similar its Drosophila homolog, Stau2 functions by binding and localizing determinants of differentiation away from the self-renewing daughter. It will be exciting to see how future studies elucidate the exact function of mRNA localization in ACD and cell fate specification of mammalian NSC and OPC.

In summary, Drosophila NB cell polarity is established through two major mechanisms, the first is the dynamic physical association of scaffold proteins that activate GTPase and kinase activities. Moreover, the centrosome provides positional memory during cytokinesis that determines the orientation of subsequent cell divisions. The integrity of the polarity complex is essential for activating G protein signaling in a transmembrane receptor-independent manner (Willard et al., 2004) and in the absence of nucleotide exchange (Yu et al., 2005). As a result, it restricts proliferation and proper cell fate determination and centrosome positioning might be a back-up mechanism for cortical polarity to establish proper spindle alignment. Noteworthy, the activation of G protein signaling through Pins occurs cell intrinsically and not only stabilizes apical polarity, but also positions the nascent mitotic spindle along the apico-basal axis and determines its size asymmetry. As a result of cell polarization, the mitotic spindle aligns such that cell fate determinants with opposing functions are localized to opposite poles during mitotis and segregate to one daughter cell only. The unequal inheritance of cell fate determinants leads to one daughter cell to differentiate and seize proliferation, whereas the opposite daughter continues to proliferate and self-renewal. The specification of fates is irreversible and no cases of spontaneous de-differentiation have been observed in the NB lineage. Cell polarity proteins are guard-keepers to hierarchical lineage differentiation and maintain a strict balance of proliferation and differentiation. Many polarity proteins originally identified in Drosophila, are conserved in the mammalian genome, which raises the possibility that they may have similar functions in mammalian neural stem cells and oligodendrocyte progenitor cells. It is very likely, that cell type and temporal variations exist in the mechanisms establishing cell polarity.

The comparison of cell-fate determinants during ACD in drosophila Nbs and adult mouse NSC is schematized in figure 2.

Figure 2: Asymmetric distribution of cell-fate determinants in Drosophila neuroblasts and Mouse adult neural stem cells.

(A) During metaphase and telophase in drosophila neuroblasts, the mitotic spindle is anchored along to the apico-basal polarity and cell-fate determinants are asymmetrically distributed in the self-renewing neuroblast (NB) and the differentiating ganglion mother cells (GMC). (B) The asymmetric cell division of adult mouse NSCs is not well characterized; still there are evidences of asymmetrically segregated cell-fate determinants between quiescent NSCs and their proliferating progeny.

A first glimpse of the mechanism involved in asymmetric cell fate determination in OPC comes from the data that NG2 not only tracks self-renewing fate, but also instructs EGFR to co-segregate to the proliferative progeny. Thereby, each asymmetric OPC division generates one NG2⁺ OPC that activates EGFR and self-renews, and a NG2-negative cell that becomes a differentiated oligodendrocyte (Sista Sugiarto et al., 2011). The ratio between SCD and ACD in OPCs is highly dependent on physiological and pathological factors like stab-wound, induced demyelination or aging (E Boda et al., 2015). The molecular mechanisms that determine whether an OPC undergoes an asymmetric or symmetric division have yet to be identified.

Concluding remarks:

Founding studies in Drosophila have paved the way to the understanding of neural progenitor cell division mode in higher organisms. The elucidation of neural and glial progenitor cell division mode in mammals is challenging due to functional redundancy and heterogeneity. Progenitor populations change during developmental and ageing. Moreover, subpopulations may exist in different neuro-anatomical location, as suggested by transcriptome analyses of OPC, for example (Marques et al., 2016). It will be necessary to completely understand the underlying causes and consequences of this heterogeneity in order to fully assess the functional significance of asymmetric cell divisions of OPC. We expect that future studies will determine the existence and function of ACD in adult neural progenitor cells and whether deregulation of ACD in adult stem and progenitor cells causes brain tumorigenesis. Improving the understanding of the regulation of ACD such as polarity, cell fate and differentiation, holds the promise to develop new therapeutics targeting tumors and other degenerative diseases of the brain.