Cycling or not cycling: cell cycle regulatory molecules and adult neurogenesis

Abstract

The adult brain most probably reaches its highest degree of plasticity with the lifelong generation and integration of new neurons in the hippocampus and olfactory system. Neural precursor cells (NPCs) residing both in the subgranular zone of the dentate gyrus and in the subventricular zone of the lateral ventricles continuously generate neurons that populate the dentate gyrus and the olfactory bulb, respectively. The regulation of NPC proliferation in the adult brain has been widely investigated in the past few years. Yet, the intrinsic cell cycle machinery underlying NPC proliferation remains largely unexplored. In this review, we discuss the cell cycle components that are involved in the regulation of NPC proliferation in both neurogenic areas of the adult brain.

Introduction

Several decades of substantial advances in the field of neurosciences have highlighted the outstanding plasticity of the adult mammalian brain. One intriguing facet of brain plasticity is its ability to generate new neurons in two discrete areas [1]. Indeed, neural precursor cells (NPCs) of both the dentate gyrus (DG) of the hippocampus and the subventricular zone (SVZ) retain the ability to proliferate and undergo neuronal differentiation throughout adulthood [2]. Consequently, understanding the mechanisms of adult neurogenesis will contribute to the development of novel stem cell-based strategies to replace neuronal loss following neurodegenerative disorders such as Alzheimer’s disease, Parkinson’s disease or stroke [3, 4].

In mammals, the cell cycle is driven by the concerted action of cyclin-dependent kinases (Cdks) and their activating partners, the cyclins (Fig. 1) [5]. During the G1 phase, Cdk4/6-cyclin D complexes progressively phosphorylate the retinoblastoma protein (pRb), causing E2f transcription factors to promote the transcription of the genes required for cell cycle progression, including cyclin E [6, 7]. Cdk2-cyclin E complexes further phosphorylate pRb, leading to its complete inactivation and to a wave of transcriptional activity essential for DNA replication phase (i.e., S phase) entry [6, 7]. Cdk2 then partners with cyclin A to drive the cell through S phase, and at the end of this cell cycle phase, cyclin A associates with Cdk1 (Fig. 1). The resulting complexes facilitate the completion of the G2 phase. Finally, Cdk1-cyclin B complexes contribute to G2-M transition [8], and direct the structural and regulatory events during mitosis. Importantly, Cdk activity is negatively regulated by the members of the Ink4 and Cip/Kip family, the so-called Cdk-inhibitors (CKIs) (Fig. 1) [9]. Altogether, these elements orchestrate the progression through the different phases of the cell cycle that ultimately leads to the mitotic segregation of the two daughter cells [10, 11].

Fig. 1.

Cell cycle regulation in the adult SGZ and SVZ compared to the classical model of the cell cycle. The dashed line on Cdk2 in the SGZ cell cycle indicates its dispensability to the process. Variations of cell cycle and cell cycle phase durations may represent a key aspect of the regulation of NPC proliferation in the adult brain, as it does during embryonic brain development

A number of studies over the past few years have challenged the importance of the canonical cell cycle pathway to organismal development. Notably, the use of mutant mouse models has revealed that a considerable number of cell cycle regulators are not essential for the survival of the organism [12–14]. Indeed, the requirement for particular cell cycle regulators is often cell-type specific, and functional redundancy occurs frequently within families of cell cycle regulators. Altogether, these data highlight the need for a more integrative view of the requirement for specific cell cycle regulatory molecules in defined tissues [15], including the adult brain.

Although several studies have highlighted the importance of intrinsic cell cycle components in regulating NPC proliferation during development [16–24], their putative role in the regulation of NPC proliferation in the adult neurogenic niches has received only minor consideration. However, we believe that an improved knowledge of the intrinsic cell cycle regulation of adult NPCs will be necessary to develop new cell-based regenerative therapies. In this context, we summarise the research that has already been done on the cell cycle constituents as key regulators of adult neurogenesis.

Cdks

Five Cdks directly control the progression through the mammalian cell cycle: Cdk1, Cdk2, Cdk3, Cdk4 and Cdk6, Cdk2/3/4/6 being considered as “interphase Cdks” [25]. However, although Cdk3 is expressed in human cells [26], most laboratory mouse strains lack Cdk3 owing to a naturally occurring mutation [27]. Beyond these cell cycle-related Cdks, an atypical Cdk, Cdk5, has been characterised. Unlike Cdk1 and interphase Cdks, Cdk5 is regulated by its own activators p35 and p39 rather than by cyclins [28–30], plays no apparent role in cell cycle regulation and is predominantly expressed in postmitotic neurons [31].

The case of Cdk1 is special as mice lacking Cdk1 fail to reach the morula stage (i.e., E2.5) [32], precluding any study of its requirement for the proliferation of a specific cell type at a given developmental time point or during adulthood using Cdk1 knockout mice. This issue will be discussed below (see “Conclusions and future directions”). The remaining cell cycle-related Cdks were studied in various tissues using mutant mouse models. Surprisingly, several studies reported that knockout mice mutant for a single interphase Cdk (i.e., Cdk2, Cdk4, Cdk6) were viable and developed normally until adulthood [12, 14, 33, 34].

In addition, even if combined deletion of Cdk4/6 [12] or Cdk2/4/6 [32] causes late embryonic lethality because of a major failure in haematopoiesis, most organogenesis and tissue development appear unaffected [12, 32]. From these observations emerged the idea that Cdks are endowed with interspersed compensatory functions and that specific Cdks are required in defined proliferative niches.

Cdk2

Albeit Cdk2 was thought to be essential to promote G1/S transition [26], knockout mice for this protein are viable and develop normally with a minor body weight reduction [14, 35]. Unexpectedly, Cdk2 is only essential for meiosis, and mice lacking Cdk2 are sterile because of defects in germ cell production [14]. In the adult brain, Cdk2 has been shown to be dispensable for DG neurogenesis [36] (Table 1). Indeed, Cdk2 deletion neither impairs proliferation of adult hippocampal NPCs in basal conditions nor following seizures [36]. In addition, while Cdk2 has been involved in the regulation of neuronal survival in vitro [37–39], no defect in newborn neurons survival/apoptosis was found in the DG of Cdk2-deficient mice [36]. Accordingly, Cdk2 might be absent in DG cells or functional redundancies among Cdks likely occur in this neurogenic niche. In contrast, in vivo and vitro gain- and loss-of-function approaches established a requirement for Cdk2 in cell proliferation and self-renewal capacities of adult SVZ-derived precursors [40]. Interestingly, no defects were observed in Cdk2-deficient SVZ precursors during early postnatal life thanks to a transient functional compensation by Cdk4 [40]. Therefore, in the postnatal/adult brain, the Cdk2 requirement is highly age and cell type dependent.

Table 1.

Consequences of cell cycle regulatory molecule deficiency on neurogenesis in postnatal, young and old adults in vivo and in vitro

		Postnatal (P1 < P10)		Adult (P42 < P120)		Old adult (1 year < 2 years)
		In vitro	In vivo	In vitro	In vivo	In vitro	In vivo
Cdk2−/−	DG	N.D.	N.D.	N.D.	No change in NPC proliferation	N.D.	N.D.
	SVZ	No change in neurosphere number	No change in NPC proliferation	Reduced neurosphere proliferation and self-renewal	Reduced NG2 + cell proliferation	N.D.	N.D.
Cdk4−/−	DG	N.D.	No change in NPC proliferation	N.D.	No change in NPC proliferation	N.D.	N.D.
	SVZ	N.D.	No change in NPC proliferation	N.D.	No change in NPC proliferation	N.D.	N.D.
Cdk6−/−	DG	N.D.	Reduced NPC proliferation	N.D.	Reduced type -2b and -3 NPC proliferation	N.D.	N.D.
	SVZ	N.D.	Reduced NPC proliferation	N.D.	Reduced type -2b and -3 NPC proliferation	N.D.	N.D.
Cyclin D2−/−	DG	N.D.	No change in NPC proliferation	Some neurospheres are formed	Absent NPC proliferation	N.D.	N.D.
	SVZ	N.D.	N.D.	N.D.	Absent NPC proliferation	N.D.	N.D.
pl6Ink4a−/−	DG	N.D.	N.D.	N.D.	No change in NPC proliferation	N.D.	No change in NPC proliferation
	SVZ	N.D.	N.D.	No change in neurosphere number nor self-renewal	No change in NPC proliferation	Increased neurosphere number and self-renewal	Increased type-B NPC proliferation
p21Cip1−/−	DG	N.D.	N.D.	N.D.	Increased NPC proliferation	N.D.	N.D.
	SVZ	No change in neurosphere number	N.D.	Increased neurosphere number and self-renewal	Increased type-B NPC proliferation	Reduced neurosphere number and self-renewal	Reduced type-B NPC proliferation
p27Kipl−/−	DG	N.D.	N.D.	Increased neurosphere number	Increased NPC proliferation	N.D.	N.D.
	SVZ	N.D.	N.D.	Increased neurosphere number	Increased type-C NPC proliferation	N.D.	N.D.
pl07−/−	DG	N.D.	N.D.	N.D.	N.D.	N.D.	N.D.
	SVZ	N.D.	N.D.	Increased neurosphere number and self-renewal	Increased type-B NPC proliferation	N.D.	N.D.
E2 fl−/−	DG	N.D.	N.D.	N.D.	Reduced NPC proliferation	N.D.	N.D.
	SVZ	N.D.	N.D.	N.D.	Reduced NPC proliferation	N.D.	N.D.

N.D. Non-determined

Cdk4/6

Since Cdk4 and Cdk6 amino acid sequences share 71% homology, it was thus suggested that these two kinases play similar activities. However, Cdk4-deficient mice exhibited pancreatic hypoplasia because of the reduced numbers of β cells [33, 34], and Cdk6-deficient mice displayed erythropenia [12]. Our laboratory recently described the expression pattern of Cdk4 and Cdk6, and showed exclusive expression in dividing NPCs of the adult SVZ and DG. Surprisingly, Cdk6 was the only one to drive NPC proliferation in vivo [41]. Specifically, the absence of Cdk6 induces a lengthening of G1 phase in neuronally committed NPCs [41]. According to the “cell cycle length hypothesis”, which predicts that G1 length influences differentiation[42], the lengthening G1 phase observed in the absence of Cdk6 may cause neuronally committed NPCs to prematurely withdraw from the cell cycle, thereby impeding their expansion capacity and thus the production of neurons in the adult brain.

Altogether, these studies demonstrate that, contrary to what occurs during embryonic development and in most adult proliferating tissues, proliferation of adult NPCs is highly dependent on the presence of specific interphase Cdks. A yet unsolved question is: why are these enzymes functionally distinct in the context of adult NPCs proliferation? A plausible explanation is that the required Cdks phosphorylate substrates that are unique to adult NPCs. It is also possible that only these Cdks can generate the necessary levels of kinase activity to drive adult NPCs division. Finally it would be interesting to determine whether or not Cdk1, the only Cdk required for embryonic development [32], is crucial for NPCs proliferation in the adult brain.

Cdk5

Cdk5 regulates multiple cellular processes of both the developing and mature CNS. Particularly, this enzyme mediates cytoskeletal changes involved in neuronal migration during embryonic development [43, 44]. Using in vitro and in vivo conditional knockout experiments, Hirota and colleagues demonstrated that Cdk5 is also a key molecule controlling migration of neuroblasts in the postnatal/adult SVZ [45]. For instance, Cdk5 deficiency impairs the chain formation, speed, directionality, and leading process extension of migrating cells in a cell-autonomous manner. Similarly, using retroviruses expressing either a dominant negative version of Cdk5 (DNCdk5) or a shRNA targeting Cdk5 mRNA, it was recently found that Cdk5 is critical for migration, dendritic extension and pathfinding of adult newborn dentate granule cells [46]. Interestingly, whereas viral expression of DNCdk5 only moderately reduced the survival of adult-born hippocampal neurons, genetic deletion of Cdk5 in adult hippocampal precursors using Nestin-CreERT2 transgenic mice resulted in dramatic reduction of neuronal survival [47], suggesting that low levels of Cdk5 activity are compatible with neuronal survival, but inadequate to promote neuritic pathfinding. Of note, ablation of Cdk5 expression only in mature dentate neurons decreases the number of neuroblasts without affecting cell proliferation, suggesting a non-cell autonomous role for Cdk5 in the survival of newborn neurons in the adult DG [47]. Overall, these studies highlighted crucial and non-redundant roles for Cdk5 during adult neurogenesis, but the molecular mechanisms underlying Cdk5 functions in that context remain to be elucidated.

D cyclins

Cyclin D family members are the regulatory subunits controlling Cdk4/6 activity [5]. Three D-type cyclins, D1, D2 and D3, have been described in mammals. Most cells express multiple D cyclins [48, 49], but studies in mutant mice revealed key roles for these proteins in specific cell types. For instance, mice lacking cyclin D1 display abnormalities in the retina and mammary glands [48–50], while mice lacking cyclin D2 have defects in the ovaries and testes [51, 52]. Finally, cyclin D3-deficient mice are less susceptible to T cell malignancies triggered by specific oncogenic pathways [53].

In the adult brain, analysis of bromodeoxyuridine (BrdU) incorporation revealed that mice deficient for cyclin D2, but not cyclin D1, showed reduced cell proliferation in the DG and SVZ [54, 55]. Of note, the few cells that are generated within the adult DG of cyclin D2 KO mice belong to the astroglial lineage [55], suggesting that loss of cyclin D2 mostly impedes neuronally committed NPC proliferation. Moreover, an enriched environment [56–58] failed to increase neurogenesis in mice lacking cyclin D2 [55]. Altogether, these experiments revealed how essential cyclin D2 is to adult neurogenesis [55]. This is consistent with D-type cyclins being the primary cell cycle core constituents that sense growth factor stimulation to initiate cell cycle entry [59, 60].

Noteworthy, mRNA analysis suggested that cyclin D2 is the only D-type cyclin to be expressed in the adult mouse brain [55]. Using immunolabellings, another study reported that cyclin D1 is expressed in the adult DG [61]. However, the finding that adult neurogenesis is virtually absent in cyclin D2 KO mice indicates that cyclin D1 is not able to compensate for cyclin D2 during adult neurogenesis in the DG. In contrast, at early stages of life (P5), WT and cyclin D2 KO mice display strikingly similar rates of proliferation and neurogenesis [55]. Interestingly, all D-type cyclins are present at P5, and they can compensate for each other to ensure NPC proliferation and neurogenesis [55]. It has been shown that early postnatal neurogenesis is responsible for the generation of most of the granule cell neurons present in the adult DG [62, 63] and that adult neurogenesis poorly contributes to the production of granule neurons [64]. Consistently, GCL volume is only slightly reduced in the cyclin D2-deficient mice [55]. Together, these data indicate that the requirement for the different D-type cyclins varies along developmental stages, with cyclin D2 being exclusively required for adult neurogenesis. More globally, it suggests that adult neurogenesis is a specifically regulated process.

Endogenous cdk inhibitors

p16^Ink4a

The four members of the Ink4 family (i.e., p16^Ink4a, p15^Ink4b, p18^Ink4c and p19^Ink4d) are specific inhibitors of Cdk4/6 activity. Among these, p16^Ink4a expression increases with age in a variety of tissues [65, 66], including the brain [67]. Age is a well-described negative regulator of NPC proliferation and neurogenesis [68–70], and loss of p16^Ink4a partially rescued the age-related decline in NPC proliferation in the SVZ [67]. Interestingly, in vitro and in vivo experiments showed that lack of p16^Ink4a expression increased the proliferation of early/uncommitted NPCs in the SVZ (i.e., type-B stem cells) to a larger extent than lineage-determined transit-amplifying type-C cells and type-A neuroblasts [67]. This was consistent with previous findings showing that the deletion of p16^Ink4a impedes embryonic neural stem but not lineage-restricted progenitor cell proliferation [71–73]. This raises the hypothesis that lineage determination modifies the cell cycle regulation of neural precursors in addition to restricting their developmental potential, causing the proliferation of lineage-determined cells to become p16^Ink4a-independent [72]. In addition, neurogenesis, but not gliogenesis, in the OB was also increased in p16^Ink4a-deficient old mice [67]. Altogether, these data indicate that increased expression of p16^Ink4a may account for an age-related decline of type-B cell proliferation and neurogenesis in the SVZ and OB, respectively. Conversely, the absence of p16^Ink4a did not detectably affect proliferation or neurogenesis in the aged DG [67], suggesting that, as exemplified by the deletion of Cdk2, the intrinsic cell cycle machinery driving NPC proliferation in the SVZ and DG is not always equivalent.

Cip/Kip family

The three members of the Cip/Kip family (p21^Cip1, p27^Kip1 and p57^Kip2) have a broader range of Cdk inhibitory activity compared to the Ink4 inhibitors [9, 74–76]. They form stable complexes with the Cdk enzyme before cyclin binding, thereby preventing their association [9, 77, 78].

p21^Cip1

Analysis of p21^Cip1-deficient mice has provided new insights into the comprehension of quiescence and stem cell longevity (i.e., long-term maintenance of self-renewal ability) within the adult brain. In vitro, p21^Cip1 deficiency increases the proliferation rate of SVZ-derived neurospheres from young adult (P60) mice, while the opposite situation occurs in old age (P480) [79]. However, some discrepancies were reported for in vivo analyses [79, 80], as no overall proliferation defect was reported in young adult p21^Cip1 knockout mice [79, 80]. In contrast, there was a marked reduction of the number of Ki67-positive NPCs in old mice lacking p21^Cip1 [79]. The reason for such a difference may be explained, as for p16^Ink4a knockout mice [67], by a loss of p21^Cip1 primarily affecting the proliferation of the more quiescent type-B NPCs [79], which account for a minority of actively dividing cells in the adult SVZ [81, 82]. It is possible that some rate variations of type-B precursor proliferation may not be detected when diluted within the overall cycling cell population from the young adult SVZ in vivo. However these results suggest that type-B NPCs that lack p21^Cip1 expression expand themselves to a higher rate during early adult stages, but are more rapidly exhausted during ageing [79].

Altogether, these data suggest that p21^Cip1 constrains SVZ NPC proliferation (i.e., the so-called relative quiescence), allowing their maintenance throughout the lifetime of the organism. This hypothesis is supported by the findings that haematopoietic stem cell quiescence is also maintained by p21^Cip1 [83].

On the other hand, data regarding the importance of p21^Cip1 in DG precursors are quite conflicting. For instance, although Pechnick et al. [84] observed an increase in cell proliferation in the DG in the absence of p21^Cip1, another study did not report any significant modification of the mitotic activity of p21^Cip-deficient DG precursors, except following ischemia. Since the genetic background and age of the animals used by these studies are similar, the observed discrepancies come probably from their different ways of evaluating cell proliferation.

p27^Kip1

The absence of p27^Kip1 in the adult SVZ leads to an increase in precursor proliferation [85], establishing p27^Kip1 as a negative regulator of proliferation in the adult brain. Interestingly, p27^Kip1 deletion induces a selective increase in transit-amplifying precursors (type-C) proliferation in the SVZ, while more quiescent precursors (type-B) are unchanged and neuroblasts (type-A cells) are decreased [85].

p27^Kip1 was previously shown to induce cell cycle withdrawal in oligodendroglial precursors [86–88]. It is thus tempting to speculate that the absence of p27^Kip1 prevents type-C cells from exiting the cell cycle, allowing them to perform extra rounds of divisions at the expense of lineage progression towards type-A neuroblasts [85]. This proposal correlates with findings in the haematopoietic system where p27^Kip1 deletion markedly affects the proliferation of actively dividing progenitors [83, 89].

In the DG, p27^Kip1 is particularly expressed in the SGZ [90]. A more detailed characterisation of its expression pattern revealed its localisation in the nucleus of neuronally commited NPCs [90], which reflects its cell cycle-active form where it can effectively inhibit Cdks [91, 92]. Upon deletion of the p27^Kip1 gene, the proliferation is increased in the DG [90], just as it is in the SVZ [85]. The possibility that the lack of p27^Kip1 selectively enhances the proliferation of neuronally committed NPCs requires further investigation. However, it was found that the number of newly born neurons in the DG is increased in the absence of p27^Kip1 [90].

Although a growing body of literature has substantiated our knowledge of the function of CKIs in adult neurogenesis, many questions remain unanswered. For instance, the reason why p16^Ink4a deficiency in old animals specifically rescues the rate of neurogenesis in the SVZ-OB niche but not in the DG remains to be fully elucidated. It is well known that there is an age-related decline of neurogenesis, but this decrease is much faster in the DG as compared to the SVZ-OB system [68]. Consequently, one reasonable hypothesis is that Molofsky et al. [67] analysed DG neurogenesis in excessively old mice, when the relative paucity of dividing precursors may mask effects of p16^Ink4a deletion. Hence, it will be interesting to analyse DG neurogenesis in younger p16^Ink4a-deficient mice to determine if the latter may transiently increases neuronal production. In addition, it is reasonable to assume that, in the SVZ, p21^Cip1 and p16^Ink4a maintain precursor quiescence in the young and old animals, respectively.

Considering the Cdk-cyclin specificity of the CKIs, it appears that type-B stem cells in the old SVZ are more likely to depend on D-type cyclin activation by growth factors [59, 60] to ensure their entry into the cell cycle. This last hypothesis is strongly supported by the fact that neurogenesis in the ageing brain can be stimulated by increasing the level of FGF-2[69, 93]. On the other hand, p27^Kip1 may belong to a molecular timer that defines the rate of expansion of type-C transit amplifying cells, indicating that the cell cycle machinery is clearly precursor type-specific. The expression and function of p27^Kip1 in neuronally determined NPCs is likely to be the preamble of its additional role(s) in the migration and differentiation of the newly generated neurons [94, 95]. Alternatively, the functions of both p16^Ink4a and p21^Cip1 in type-B stem cell proliferation and p27^Kip1 in type-C actively dividing NPCs might reflect the need of stem cells for an utter level of regulation of their proliferation compared to actively dividing cells. Finally one must emphasise that all studies dealing with CKI regulation of adult neurogenesis have been using germline KO mouse models. This suggests that the self-renewal phenotypes observed in these models could be at least partially non-cell autonomous and/or did not give acute information on the role of CKI in adult NPCs, specifically. Using inducible conditional gene inactivation or gene silencing is thus clearly required for deciphering the precise role of these molecules during adult neurogenesis.

Cdk-cyclin substrates (pRb/E2f)

Phosphorylation of the closely related Rb family members (i.e., pRb/p105, p107, p130) by Cdk4/6-cyclin D complexes leads to their partial inactivation and to the release of the E2f transcription factors. Among the three Rb family members, pRb function is by far the better studied and characterised [96]. However, both germline [97–99] and nervous system [100, 101] specific deletions of pRb result in embryonic/perinatal lethality, preventing any analysis of the role of pRb during adult neurogenesis. On the other hand, mice deficient in either p107 or p130 develop normally on a C57Bl/6 genetic background [102, 103]. In the adult brain, p107 is highly expressed in the SVZ, and p107-deficient mice show a twofold increase in the number of type-B stem cells compared to their WT littermates in vivo and in vitro [104]. It was later suggested that p107 regulates the neural precursor population and differentiation by the repression of Hes1, a downstream mediator of the Notch signalling pathway [105, 106]. p130 is thought to be upregulated and to maintain neurons in a differentiated state, but no specific analysis of the putative role of p130 in adult neurogenesis has been done yet.

The E2f family contains eight members most recognised for their ability to regulate cell cycle progression [107]. According to whether E2fs act positively or negatively on gene transcription, they are grouped into transcriptional activators (E2f1 to E2f3a) or suppressors (E2f3b, E2f4, E2f5, E2f7 and E2f8), E2f7 and 8 being considered as atypical E2fs [108]. As for cyclins, analysis of mutant mice revealed broad compensatory mechanisms among the E2f family members during development [109, 110]. E2f1 is the most prominently expressed in the embryonic nervous system [111, 112]. Mice with a targeted deletion of E2f1 depict testis atrophy while brain development proceeds normally [109, 110]. However, in adulthood, mice lacking E2f1 show half of the number of dividing cells within both the DG and SVZ [20]. The deficit of NPC expansion further impedes production of new neurons in the DG and OB [20]. The fact that embryonic brain development is not affected in these mutant mice [20] further suggests that cell cycle progression in adult neurogenesis is a specifically regulated process. However, the persistence of dividing cells in adult E2f1-deficient neurogenic niches indicates that other members of the E2f family are capable of some compensatory function. Supporting this is recent evidence that E2f3a or b proteins, the two products of the E2f3 gene, also sustains NPC proliferation in the adult SVZ, as mice deficient for E2f3 show a one-third reduction in NPC numbers [107].

Maintenance of adult NPCs niches: self- renewal and quiescence

A hallmark feature of adult stem cells is their capacity to self-renew over the lifespan with a very slow rate of cell division as compared to their foetal counterpart. It is widely accepted that this is linked to a quiescent state, a functionally important characteristic of adult stem cells [113]. Quiescence can be defined as a non-dividing state outside of the cell cycle (i.e., G0) where a cell can remain in stasis until activated by appropriate proliferative signals. This state is necessary in the adult brain (1) to limit the accumulation of mutations in NPCs leading to cancer and (2) to prevent the exhaustion of the NPCs pool [114]. Characterisation of the molecular regulation of NPCs-specific CDK inhibitors may provide insights into the signalling pathways underlying quiescence of adult NPCs.

Recent results showed that Bmi-1 promotes the self-renewal of NSCs by repressing the expression of p16^Ink4a and p19^Arf [71, 115]. However, despite ongoing Bmi-1 expression, p16^Ink4a expression increases with age, potentially reducing stem cell frequency and function [67]. p53, the well-known tumour suppressor gene, is also a critical regulator of cdk inhibitors. Loss of p53 results in increased proliferation of adult NPCs by a decrease of p21^cip1 expression, allowing the escape of NPCs from quiescence and promoting their entrance into the cell cycle [116, 117]. Adult NPCs lacking expression of phosphatase and tensin homolog (Pten), or forkhead box Os (FoxOs) show a similar phenotype to that of p21^cip1 in terms of cell cycling and exhaustion [118–120]. Whether these effects are a consequence of the regulation of cell cycle inhibitors remains to be demonstrated.

Conclusions and future directions

The ability of the brain to undergo neurogenesis in two restricted regions has challenged our view of brain plasticity. Given that new neurons are added or replace old ones in the DG and OB, respectively [64], one can easily understand that an accurate regulation of NPC proliferation is needed to produce the correct amount of newly generated mature neurons, particularly in a perspective of cell replacement following neurodegenerative disorders [3]. In this context, cell cycle regulatory molecules represent an attractive basis of investigation. Evolution has endowed higher mammals with a wide panel of cell cycle regulatory molecules capable of redundancy in order to prevent the catastrophic consequences upon the loss of a single cell cycle component. However, recent results show that redundancy is not complete, and a careful study of knockout mouse models reveals unique functions of many of these proteins. This specificity appears at the tissue level, and even more importantly at the cell scale. Therefore, in the brain, one should determine the cell cycle machinery in proliferating cells in the DG and the SVZ, and even in each neurogenic area in different subtypes of proliferating cells, e.g., uncommitted versus neuronally determined precursors.

Besides, Cdk1 deficiency is lethal at E2.5 in mice [32], precluding any analysis of its putative roles in NPC proliferation in the adult brain using a constitutive Cdk1-mutant mouse model. This particular point highlights a significant drawback to the studies that have so far addressed the contribution of cell cycle regulatory molecules to adult neurogenesis. To our knowledge, there are currently no reports examining the function of core cell cycle components in adult neurogenesis using conditional gene manipulation. Such experiments will be necessary to fully determine the cell cycle machinery governing cell proliferation in the two neurogenic regions of the adult brain, and constitute a prerequisite for manipulating NPC proliferation and neurogenesis in the frame of stem cell-based therapies.

In conclusion, much has been uncovered in the regulation of NPC proliferation in the adult brain, but a lot remains to be determined regarding the contribution of a basic biological process that is the cell cycle to adult neurogenesis. In our view, an in depth understanding of this issue is of utter importance to foresee the use of endogenous adult NPCs as a source for regenerative therapies.