Antagonistic Interplay between Necdin and Bmi1 Controls Proliferation of Neural Precursor Cells in the Embryonic Mouse Neocortex

Neural precursor cells (NPCs) in the neocortex exhibit a high proliferation capacity during early embryonic development and give rise to cortical projection neurons after maturation. Necdin, a mammal-specific MAGE (melanoma antigen) family protein that possesses anti-mitotic and pro-survival activities, is expressed abundantly in postmitotic neurons and moderately in tissue-specific stem cells or progenitors. Necdin interacts with E2F transcription factors and suppresses E2F1-dependent transcriptional activation of the cyclin-dependent kinase Cdk1 gene. Here we show that necdin serves as a suppressor of NPC proliferation in the embryonic neocortex. Necdin is moderately expressed in the ventricular zone of mouse embryonic neocortex, in which proliferative cell populations are significantly increased in necdin-null mice. In the neocortex of necdin-null embryos, expression of Cdk1 and Sox2, a stem cell marker, is significantly increased, whereas expression of p16, a cyclin-dependent kinase inhibitor, is markedly diminished. Cdk1 and p16 expression levels are also significantly increased and decreased, respectively, in primary NPCs prepared from necdin-null embryos. Intriguingly, necdin interacts directly with Bmi1, a Polycomb group protein that suppresses p16 expression and promotes NPC proliferation. In HEK293A cells transfected with luciferase reporter constructs, necdin relieves Bmi1-dependent repression of p16 promoter activity, whereas Bmi1 counteracts necdin-mediated repression of E2F1-dependent Cdk1 promoter activity. In lentivirus-infected primary NPCs, necdin overexpression increases p16 expression, suppresses Cdk1 expression, and inhibits NPC proliferation, whereas Bmi1 overexpression suppresses p16 expression, increases Cdk1 expression, and promotes NPC proliferation. Our data suggest that embryonic NPC proliferation in the neocortex is regulated by the antagonistic interplay between necdin and Bmi1.


Introduction
Higher brain functions of mammals are performed by a vast number of neurons in the cerebral neocortex. A large population of neocortical neurons arises from NPCs or neural stem cells residing in the neural tube during early developmental period. Early NPCs proliferate by dividing symmetrically for self-renewal (expansion phase) and then asymmetrically to produce young neurons (neurogenic phase) [1]. Nascent neurons migrate radially to form the cortical plate, which gives rise to a typical six-layered structure of the neocortex after maturation. During early neocortical neurogenesis, NPCs proliferate rapidly to expand their pool because the number of postmitotic neurons correlates closely with that of NPCs. Although it is speculated that expansion of neocortical NPCs is tightly regulated in each mammalian species, there is limited information about molecular mechanisms underlying the regulation of neocortical NPC proliferation.
Cell cycle regulators are expressed in the neocortex at early stages of mammalian development [2,3]. Regulation of the cell cycle is dependent on the control of cyclin-dependent kinases (Cdks), whose activities are positively regulated by cyclins and negatively by Cdk inhibitors such as p16 Ink4a (p16), p21 Cip1 (p21) and p27 Kip1 (p27). These inhibitors suppress Cdk activities and reduce phosphorylation of the retinoblastoma protein (Rb) family proteins such as Rb, p107, and p130. Hypophosphorylated Rb family proteins repress the activities of E2F family transcription factors that activate downstream genes involved in cell cycle progression. The Rb family proteins are differentially expressed in the embryonic brain during embryogenesis [2,4]. However, detailed mechanisms whereby these cell cycle-related proteins regulate the self-renewal and proliferation of embryonic NPCs remain elusive.
Necdin was originally identified as a hypothetical protein encoded by a neural differentiation-induced gene in murine embryonal carcinoma P19 cells [5]. Necdin is abundantly expressed in virtually all of postmitotic neurons and skeletal muscle cells at early stages of development [6,7]. Ectopic expression of necdin strongly suppresses the proliferation of tumor-derived cell lines [8][9][10][11]. Necdin, like Rb, binds to E2F1 and E2F4 [9,12], and interacts with E2F1 on the Cdk1 (Cdc2) promoter to repress the transcriptional activation of the Cdk1 gene [13]. Thus, necdin is likely to downregulate the expression of E2Fdependent cell cycle-related genes in proliferative cells and exerts its anti-mitotic activity during neurogenesis. However, there is little information about the molecular mechanism whereby necdin controls cell divisions of NPCs during embryonic neurogenesis.
In this study, we examined whether necdin regulates proliferation of neocortical NPCs in necdin-null mouse embryos. Analyses using neocortical NPCs prepared from necdin-null mice show that necdin suppresses NPC proliferation and induces significant changes in p16 and Cdk1 expression. We also demonstrate that necdin and Bmi1, a key transcription factor involved in NPC proliferation, interact to modulate their downstream cell-cycle regulatory systems. The present study provides insights into the regulatory mechanisms underlying NPC proliferation in the mammalian neocortex.

Necdin Deficiency Increases Proliferative Cell Populations in the Embryonic Neocortex in vivo
We examined the distribution pattern of necdin in the embryonic forebrain at E14.5 by immunohistochemistry (Fig. 1A). The necdin immunoreactivity was detected mainly in the cerebral neocortex and septum, consistent with our previous data [22]. In contrast, little necdin immunoreactivity was detected in the forebrain of necdin-null mice. In the neocortex, necdin was strongly expressed in the cortical plate (CP) where bIII-tubulinpositive (bIII-tubulin + ) neurons were located (Fig. 1B). Necdin was also expressed moderately in the proliferative zone including the ventricular zone (VZ) where the stem cell markers Sox2 and nestin were expressed.
Because necdin has an anti-mitotic activity, we tested whether proliferative cell populations are increased in the neocortex of necdin-null mice at E14.5. The population expressing the mitosis marker phospho-histone H3 (pH3) increased by 62% in necdinnull neocortex (Fig. 1C). The numbers of pH3 + cells were increased in the apical and basal regions of the proliferative zone by 61% and 67%, respectively, of control levels. To determine whether the increased proliferative cell populations in necdin-null neocortex are limited to specific developmental periods, we analyzed the pH3 + cell population in the necdin-null neocortex at E12.5 and E16.5 (Fig. S1). The numbers of pH3 + neocortical cells in necdin-null mice significantly increased at E12.5 and E16.5 by 65% and 27%, respectively, suggesting that necdin negatively regulates the proliferation of NPCs throughout the period of embryonic neurogenesis. When the number of DNA-synthesizing cells in the neocortex of E14.5 mice was analyzed by 59-bromo-29deoxyuridine (BrdU) incorporation assay, the BrdU + cell population significantly increased by 28% in necdin-null neocortex (Fig. 1D). These changes of proliferative cell populations were observed predominantly in the proliferative zone of necdin-null mice.
Although we found significant increases in proliferative cell populations in necdin-null neocortex in vivo, there was little or no difference in the cortical thickness at E14.5 (Fig. S2). We assumed that the balance between proliferation and death of NPCs is maintained, resulting in no apparent change of the neocortical thickness. We thus analyzed apoptosis in the neocortex of E14.5 mice by terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL). The number of TUNEL + cells significantly increased in the proliferative zone of necdin-null mice at E12.5, E14.5, and E16.5 (Fig. S3), suggesting that both proliferation and apoptosis of NPCs are enhanced in developing neocortex of necdin-null mice.

Necdin Deficiency Affects Expression Levels of Cell Cyclerelated Genes in the Embryonic Neocortex
To elucidate the molecular mechanism that increases NPC proliferation in necdin-null mice, we determined the mRNA levels of cell cycle regulators in E14.5 mouse neocortex in vivo by quantitative reverse-transcription PCR (qRT-PCR) ( Fig. 2A). The p16 mRNA level was significantly low in the necdin-null neocortex, whereas no significant differences in the p19, p21, p27, and p57 mRNA levels were noted between wild-type and necdin-null mice. In contrast, the expression levels of Cdk1 and Sox2 mRNAs were significantly high in the neocortex of necdinnull mice. To examine whether these mRNA levels correlate with the protein levels, we analyzed the protein levels of p16, Cdk1, and Sox2 by Western blotting (Fig. 2B, C). In the neocortex of necdinnull mice, the p16 protein level decreased to 45% of the control level, whereas the Cdk1 and Sox2 protein levels increased 2.0-and 2.6-fold, respectively. We then examined the distribution pattern of p16 + cells in the neocortex of E14.5 mice by immunohistochemistry (Fig. 2D, E). The p16 protein was mainly detected in the CP and VZ. Fluorescence microphotometry revealed that the p16 protein level in the necdin-null neocortex markedly decreased to 40% of the wild-type control level. These results suggest that necdin deficiency downregulates p16 expression and upregulates Cdk1 expression in the neocortex to increase the NPC population in vivo.

Necdin Deficiency Enhances Proliferation of Neocortical NPCs in vitro
To examine whether necdin is expressed in neocortical NPCs, we prepared NPCs from the neocortex at E14.5, cultured for 48 hrs in the presence of epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF), and carried out immunocytochemical analysis (Fig. 3A). Necdin was expressed in virtually all NPCs, most of which also expressed Sox2 and nestin (necdin + / Sox2 + , 9062%; necdin + /nestin + , 9163%; total 200 cells analyzed; n = 3). The necdin immunoreactivity was detected in the cytoplasm and nucleus of these NPCs. When neocortical NPCs were differentiated in vitro in the absence of EGF and bFGF, bIIItubulin + neurons and glial fibrillary acidic protein (GFAP) + astrocyte-like cells were differentiated, suggesting that primary NPCs are multipotent (Fig. 3B). To determine whether necdin regulates the proliferation of NPCs in a cell-intrinsic manner, we analyzed BrdU incorporation into nuclear DNA of primary NPCs. In necdin-null NPCs, the number of BrdU + cells was 1.69 times that in wild-type NPCs (Fig. 3C). In addition, the TUNEL + cell population in necdin-null NPCs was 4 times that in wild-type NPCs (Fig. 3D). These results suggest that necdin suppresses both proliferation and apoptosis of primary NPCs as seen in the neocortex in vivo. We also analyzed the expression levels of cell cycle-regulatory genes in NPCs by qRT-PCR (Fig. 3E). In necdin-null NPCs, the p16 mRNA expression level was 50% of that in wild-type NPCs, whereas the Cdk1 mRNA level was 1.6 times the control level. In contrast, the expression levels of p19 and p21 mRNAs were unchanged.
Because p16 expression was markedly diminished in necdin-null NPCs, we investigated whether necdin increases p16 expression by suppressing Bmi1, a Polycomb group protein that represses p16 expression to promote NPC proliferation [23]. We examined the subcellular localization of necdin and Bmi1 in primary NPCs by immunocytochemistry (Fig. 3F) and Western blotting (Fig. 3G). Necdin was mainly present in the cytoplasm but moderately in the nucleus of primary NPCs, whereas Bmi1 was restricted to the nucleus. We then analyzed the expression levels of necdin and Bmi1 mRNAs in NPCs and neurons by qRT-PCR. The necdin mRNA level in differentiated neurons was 2.5 times the NPC level (Fig. S5). In contrast, the Bmi1 mRNA level in NPCs was 3.1 times that in differentiated neurons. There was no difference in the Bmi1 expression level between wild-type and necdin-null mice in

Necdin Physically Interacts with Bmi1
To determine whether necdin binds to Bmi1 by the proteinprotein interaction, we constructed deletion mutants of Bmi1 encompassing known motifs such as the Ring-finger (RF), helixturn-helix (HTH), and proline-serine-rich domains (PS) [24,25] (Fig. 4A). When necdin and Myc-tagged Bmi1 deletion mutants (DCT, RF, HTH) were coexpressed in HEK293A cells, Bmi1 fulllength (FL) and HTH were co-precipitated with necdin, and necdin was conversely co-precipitated with FL and HTH (Fig. 4B). The positive control p53, but not p53DN lacking the N-terminal necdin-binding site [26], was co-precipitated with necdin. In this assay, the HTH expression level was low, and expression of the Cterminal proline-serine-rich (PS) mutant was undetectable (not shown), presumably owing to the instability of these mutant proteins [27]. We next examined the direct interactions between necdin and Bmi1 deletion mutants by in vitro binding assay using bacterially expressed GST-tagged Bmi1 mutants and virally expressed His-tagged necdin (Fig. 4C). Necdin bound efficiently to Bmi1 FL, C-terminal deletion mutant (DCT) and HTH, but failed to bind to the N-terminal RF or C-terminal PS. These results suggest that necdin directly binds to Bmi1 via the central HTH domain. To detect an endogenous complex of necdin and Bmi1, we carried out the co-immunoprecipitation assay using neocortical NPC lysates prepared from wild-type and necdin-null mice at E14.5 (Fig. 4D). Bmi1 was co-precipitated with necdin in the NPC extract prepared from wild-type mice, but not with proliferating cell nuclear antigen (PCNA) used as a negative control, suggesting that Bmi1 and necdin interact endogenously in primary NPCs.

Necdin and Bmi1 Counteract each other to Control Cdk1 and p16 Promoter Activities
We investigated whether the interaction between necdin and Bmi1 affects the proliferation rate of HEK293A by BrdU incorporation assay (Fig. 5A). Necdin reduced the BrdU + cell population by 59% (Fig. 5B). Bmi1 efficiently counteracted the proliferation-suppressive activity of necdin. We then performed the luciferase reporter assay for Cdk1 promoter activity (Fig. 5C). Necdin had no appreciable effect on basal Cdk1 promoter activity but efficiently suppressed E2F1-dependent Cdk1 promoter activity. Remarkably, Bmi1 almost completely relieved the necdin-induced suppression. Because Bmi1 directly binds to the p16 promoter and downregulates its transcriptional activity [28], we examined the effect of necdin on p16 promoter activity (Fig. 5D). Although necdin had little or no effect on basal p16 promoter activity, necdin efficiently restored the Bmi1-induced suppression to the control level.

Necdin Regulates NPC Proliferation and Expression of p16 and Cdk1 mRNAs
To examine whether the protein-protein interaction between necdin and Bmi1 regulates NPC proliferation, we used lentivirusmediated gene transfer system to express necdin, Bmi1, and Bmi1 small hairpin RNA (shRNA) in primary NPCs. NPCs infected with a necdin-expressing lentivirus expressed high levels of the necdin protein as analyzed by Western blotting (Fig. 6A). We examined the effect of lentivirus-mediated necdin overexpression on proliferation of wild-type and necdin-null NPCs using BrdU incorporation assay (Fig. 6B). In wild-type NPCs, necdin overexpression reduced the BrdU + cell population to 44% of the control level. The BrdU + cell population of necdin-null NPCs increased 2-fold, and necdin overexpression reduced it by 66%. We next examined the effects of necdin overexpression on the expression levels of p16 and Cdk1 mRNAs in primary NPCs (Fig. 6C). In wild-type NPCs, necdin overexpression induced a 2.2fold increase in the p16 mRNA level. In necdin-null NPCs, the p16 mRNA level decreased to 40% of the control level and was increased 3.5-fold by necdin overexpression. These results imply that necdin overexpression promotes p16 transcription via Bmi1 repression in NPCs. In contrast, necdin overexpression reduced the Cdk1 mRNA level by 40% in wild-type NPCs (Fig. 6D). In necdin-null NPCs, the Cdk1 mRNA level was 1.7 times that of wild-type control level, and necdin overexpression reduced it by 71%. These results suggest that necdin efficiently downregulates Cdk1 expression in neocortical NPCs.

Bmi1 Regulates NPC Proliferation and Expression of p16 and Cdk1 mRNAs
We determined the effects of Bmi1 overexpression and knockdown on the proliferation rates of neocortical NPCs. Western blot analysis revealed that the Bmi1 protein levels in primary NPCs infected with lentiviruses expressing Bmi1 and Bmi1 shRNA were elevated and diminished, respectively (Fig. 7A). Neither Bmi1 overexpression nor Bmi1 knockdown affected endogenous necdin levels in wild-type NPCs. In wild-type NPCs, Bmi1 overexpression induced a 2.2-fold increase in the BrdU + cell population, and Bmi1 knockdown decreased it by 58% (Fig. 7B). In necdin-null NPCs, Bmi1 overexpression had no significant effect on the BrdU + cell population, whereas Bmi1 knockdown reduced it by 42%. We then examined the effects of Bmi1 overexpression and knockdown on the p16 and Cdk1 mRNA levels in NPCs. Bmi1 overexpression markedly reduced the p16 mRNA level by 70%, whereas Bmi1 knockdown induced a 1.7-fold increase (Fig. 7C). The p16 mRNA level was significantly reduced by 50% in necdin-null NPCs, in which Bmi1 knockdown increased it by 80%, indicating that endogenous Bmi1 downregulates p16 expression in a necdin-independent manner. In contrast, Bmi1 overexpression induced a 1.6-fold increase in the Cdk1 mRNA level, whereas Bmi1 knockdown reduced it by 40% (Fig. 7D). In necdin-null NPCs, the Cdk1 level was 1.9-times the wild-type level, whereas neither Bmi1 overexpression nor Bmi1 knockdown exerted significant effects on the Cdk1 mRNA level, suggesting that the regulatory effects of Bmi1 on Cdk1 expression are mediated by endogenous necdin in NPCs.
Because Bmi1 downregulates the expression of p19 mRNA [29,30] and p21 mRNA [31,32] in NPCs, we quantified these mRNA levels in necdin-null NPCs (Fig. S6). There were no significant differences in the p19 and p21 mRNA levels between wild-type and necdin-null NPCs. Furthermore, necdin overexpression exerted no significant effects on these mRNA levels, suggesting that expression of p19 and p21 genes is regulated in a necdin-independent manner. In contrast, Bmi1 overexpression reduced the p21 and p19 mRNA levels by 40%, whereas Bmi1 knockdown induced 1.7-and 2.2-fold increases in the p19 and p21 mRNA levels, respectively, consistent with the previous reports [29][30][31][32] (Fig. S7). There were no significant differences in the degrees of these mRNA changes between wild-type and necdinnull NPCs.

Discussion
An accumulating body of evidence has suggested that Bmi1 is involved in NPC proliferation during embryonic and postnatal periods [23,[30][31][32][33][34][35]. The present study has shown that necdin and Bmi1 counteract each other to control proliferation of embryonic NPCs residing in the developing neocortex. On the basis of present data, we propose that the protein-protein interaction between necdin and Bmi1 controls NPC proliferation via Cdk1 and p16 pathways (Fig. 8). Under normal conditions, necdin, which suppresses Cdk1 transcription, and Bmi1, which suppresses p16 transcription, antagonize each other to regulate the proliferation of NPCs. In necdin-null NPCs (Necdin-KO), Cdk1 expression is upregulated, whereas p16 expression is downregulated via Bmi1 derepression, resulting in the enhancement of NPC proliferation. Bmi1 overexpression suppresses necdin and induces changes similar to those seen in necdin-null NPCs (Bmi1 OE). In necdin-overexpressing NPCs (Necdin OE), Cdk1 expression is downregulated, whereas p16 expression is upregulated via Bmi1 repression, resulting in the reduced NPC proliferation. Bmi1 knockdown (Bmi1 KD) suppresses Cdk1 expression via necdin derepression and induces changes similar to those seen in necdinoverexpressing NPCs. We assume that these events in primary NPCs in vitro also occur in NPCs in vivo in the embryonic neocortex.
The present findings suggest that necdin and Bmi1 use their downstream cell-cycle regulatory systems via Cdk1 and p16 pathways to control NPC proliferation: Necdin suppresses Cdk1 expression (independently of Bmi1), increases p16 expression by repressing Bmi1, and suppresses NPC proliferation, whereas Bmi1 suppresses p16 expression (independently of necdin), increases Cdk1 expression by repressing necdin, and increases NPC proliferation. Necdin binds directly to E2F1 on the Cdk1 promoter to suppress Cdk1 transcription [13]. Cdk1 is essential for controlling cell divisions during embryonic development and executes all the cell cycle-related events even in the absence of interphase Cdks such as Cdk2, Cdk4, and Cdk6 [36][37][38]. Accordingly, necdin may exert its anti-mitotic effect by suppressing Cdk1 expression at the transcription level in various cellular contexts. On the other hand, the proliferation capacity is reduced in Bmi1-deficient NPCs in which p16 expression is upregulated, and p16 deficiency reverses their proliferation defect [23], suggesting that endogenous Bmi1 upregulates NPC proliferation by repressing p16 expression. In the present study, p16 expression was significantly downregulated in necdin-null NPCs in vivo (Fig. 2). In necdin-null mice, the p16 immunoreactivity was markedly diminished in most neocortical cells but not in a specific cell population such as that undergoing apoptosis (Fig. 2D), suggesting that necdin upregulates p16 expression by repressing Bmi1 in a ubiquitous manner. Thus, necdin is likely to exert its anti-mitotic effect on embryonic NPCs by repressing Cdk1 expression and increasing p16 expression via Bmi1 repression.
Expression of the necdin gene (Ndn) is regulated by genomic imprinting, a mammal-specific epigenetic mechanism whereby certain genes are silenced in a parent-of-origin-specific manner [39,40]. Human NDN is located in chromosome 15q11-12, a region responsible for the pathogenesis of the human neurodevelopmental disorder Prader-Willi syndrome. The maternal NDN allele is silenced through hypermethylation of CpG-rich sequences, and necdin is expressed only from the paternal NDN allele, whose deletion causes a complete defect of necdin expression. Bmi1, a component of the Polycomb repressive complex 1 that regulates gene silencing by remodeling the chromatin structure, affects gene silencing through epigenetic mechanism [41,42]. However, the present study demonstrated that Bmi1 overexpression or knockdown exerted no significant effect on necdin expression in primary NPCs (Fig. 7A). This may be because acute lentivirus-mediated modulations of Bmi1 expression are insufficient to affect epigenetic controls of necdin expression. The previous study on the Bmi1-dependent transcriptional control of imprinted genes has shown that necdin expression is significantly derepressed in lung epithelial stem cells prepared from Bmi1mutated mice [42]. These findings suggest that Bmi1 deficiency during critical periods for genomic imprinting is indispensable for the increase in necdin expression.
The present study has shown that necdin deficiency significantly increases the apoptotic cell population in embryonic neocortex in vivo and primary NPCs, indicating that necdin suppresses both proliferation and apoptosis of neocortical NPCs. It is noteworthy that Drosophila MAGE, a necdin-homologous MAGE protein, is expressed in neural stem cells (neuroblasts) and their progeny (ganglion mother cells and postmitotic neurons) in postembryonic neurogenesis [43], and that downregulation of MAGE expression in the developing mushroom bodies increases both NPC proliferation and neuronal apoptosis [44], implying that these evolutionally distant MAGE proteins possess the conserved function required for neurogenesis. The present data also demonstrate that necdin-null mice have an increased population of neocortical cells in the postnatal period (Fig. S4), suggesting that the increased proliferation rate of necdin-null NPCs exceeds the increased apoptosis rate. Thus, we suggest that necdin is It has previously been demonstrated that necdin-null mice exhibit an enhanced number of proliferating hematopoietic stem cells during hematopoietic regeneration, suggesting that necdin prevents their excessive proliferation [18]. Similarly, necdin may exert its anti-mitotic effect on NPCs to prevent their excessive proliferation. It is also possible that necdin in NPCs is involved in terminal differentiation-associated withdrawal from the cell cycle. In this case, necdin deficiency may increase the number of differentiated postmitotic cells. We have previously found that necdin is moderately expressed in preadipocytes or mesenchymal stem cells residing in white adipose tissues, and that the number of terminally differentiated adipocytes is markedly increased in necdin-null mice fed a high-fat diet [20]. Similarly, the number of postmitotic neurons was slightly but significantly increased in the neocortex of necdin-null postnatal mice in the present study (Fig. S4). On the other hand, mitotically quiescent postmitotic neurons express higher levels of necdin than these proliferating stem cells or progenitors during mouse development [6]. We speculate that necdin prevents postmitotic neurons from aberrant re-entry into the cell cycle.
During development of necdin-null mice, apoptosis is enhanced in sensory neurons of the dorsal root ganglia [45,46], granule neurons and their progenitors in the internal and external granule layers, respectively, of the cerebellum [13], and motor neurons of the spinal cord [47]. In the present study, the number of apoptotic cells was significantly increased only in the proliferative zone of necdin-null mice (Fig. S3). The apoptotic NPC population in vivo is much smaller in the neocortex than that in the GEs [21]. These findings indicate that the apoptosis rates in necdin-null mice vary depending on cell types and developmental stages. This may be because necdin interacts with various pro-apoptotic and prosurvival proteins that are expressed in a cell type and stage-specific manner [13,21,[45][46][47]. Although the mechanism whereby necdin prevents apoptosis of neocortical NPCs remains to be elucidated, we speculate that necdin targets p53 which is highly expressed in neocortical NPCs in vivo [48]. Because necdin strongly suppresses p53-mediated neuronal apoptosis induced by DNA damage [49], we are currently investigating whether necdin suppresses p53mediated apoptosis of neocortical NPCs.
Necdin, like Rb, interacts with the viral oncoproteins SV40 large T antigen and adenovirus E1A [9], suggesting that these viral proteins interact with endogenous necdin and inhibit its anti-mitotic activity to promote malignant transformation of infected cells. However, the mechanisms that regulate the function of necdin under physiological conditions have been poorly understood until recently. We have most recently reported that the necdin protein in GE-derived NPCs is degraded under hypoxic conditions by hypoxia-inducible factor-2a via the ubiquitinproteasome pathway [21], implying that the function of necdin in NPCs is modulated in an oxygen tension-dependent manner. The present study has demonstrated another mechanism whereby Bmi1 suppresses necdin by the direct protein-protein interaction. These findings suggest that multiple pathways that control the activity of endogenous necdin are operative in embryonic NPCs to maintain their proper proliferation rates. Thus, we propose that necdin serves as a molecular rheostat to regulate the expansion of embryonic NPCs.

Animals
Ndn mutant mice (Ndn tm1Ky ) were generated and maintained as described in [45]. Heterozygous male mice (Ndn +/2 ) (.20 generations in ICR background) were crossed with wild-type female mice (Ndn +/+ ) to obtain wild-type (Ndn +m/+p ) and paternal Ndn-deficient (Ndn +m/2p ) littermates. All mice were housed in a 12hour light/dark cycle with room temperature at 2363uC. Pregnant female mice at gestation day 14.5 were sacrificed by cervical dislocation, and embryos were collected. Genotypes of all mice were analyzed by PCR for mutated Ndn locus. Experiments using gene-targeted mice were approved by the Recombinant DNA and Animal Experiment Committees of the Institute for Protein Research, Osaka University, and performed in accordance with institutional guidelines and regulations.

Cell Proliferation Assay
NPCs were plated on coverslips in 12-well plates (1610 5 cells per well), cultured for 24 hrs, incubated for another 4 hrs in the presence of BrdU (10 mM), and fixed with 70% ethanol containing 20 mM glycine-HCl (pH 2.0) for 30 min at 220uC. For lentivirusinfected NPCs, BrdU was added 48 hrs after viral infection, and NPCs were incubated for another 4 hrs and fixed for immunocytochemistry. BrdU was detected by fluorescence immunocytochemistry as described in [20,26]. For transfected cells, HEK293A cells were plated on coverslips in 35-mm dishes, cultured in DMEM supplemented with 10% FBS, and transfected with pEGFP-C2 (1 mg) (Clontech), pcDNA3.1+ carrying necdin cDNA (0.5 mg) and 66Myc-Bmi1 cDNA (5 mg). The empty vector pcDNA3.1+ was added to equalize the amounts of transfected DNA (6 mg/assay). BrdU (10 mM) was added to the medium 48 hrs after transfection, and the cells were fixed 4 hrs later. BrdU and GFP were detected by fluorescence immunocytochemistry. For cell proliferation assay in vivo, pregnant mice were injected i.p. with 59-ethynyl-29-deoxyuridine (EdU, Life Technologies) (50 mg/ kg body weight) at gestation day 14.5. Mice were sacrificed at P4 by decapitation, and forebrain tissues were fixed for EdU histochemistry. Incorporated EdU was detected using Click-iT EdU Alexa Fluor Imaging Kit (Life Technologies).

Apoptosis Assay
Nuclear DNA fragmentation was analyzed by terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TU-NEL) visualized with Texas Red as described in [53,54]. Primary NPCs were incubated for 48 hrs, dispersed, and fixed 24 hrs later. TUNEL analysis in brain tissues was performed as described in [54]. TUNEL + cells in each 200-mm-wide column of the neocortex were counted.
In vitro Binding Assay cDNAs encoding Bmi1 and its deletion mutants were subcloned into pCold-GST (Takara), and glutathione-S-transferase fusion proteins were synthesized in E. coli BL21 as described in [55]. The GST-Bmi1 fusions immobilized on glutathione Sepharose 4B (100 ml) (GE Healthcare) were incubated with His-tagged necdin (200 ng) at 4uC for 2 hrs in 1 ml of 50 mM Tris buffer (pH 8.0) containing 300 mM KCl and 2 mM 2-mercaptoethanol as described in [26]. After washing with 50 mM HEPES buffer (pH 8.0) containing 300 mM KCl and 1 mM DTT, bound proteins were eluted and detected by Coomassie Brilliant Blue staining. Bound His-tagged necdin was detected by Western blotting.
Transfection efficiency was normalized with the activity of coexpressed Renilla luciferase.

Lentivirus Vectors
Recombinant lentiviruses were produced in HEK293FT cells transfected with SIN vector plasmids and two helper plasmids as described in [20,56]. Necdin and Bmi1 cDNAs were subcloned into pENTR1A entry vector (Life Technologies) to construct the destination vectors CSII-EF1a-necdin-IRES-EmGFP and CSII-EF1a-Bmi1-IRES-EmGFP, in which Emerald Green Fluorescent Protein (EmGFP) (Life Technologies) was contained for an expression indicator. The lentivirus vector expressing Bmi1 shRNA CSII-EF1a-EmGFP-Bmi1 shRNA was constructed using BLOCK-iT Pol II miR RNAi Expression Vector Kits (Life Technologies) and double-stranded oligonucleotides described in [31]. The viral titer was measured by serial dilution on HEK293FT cells (Life Technologies) and determined as GFP + cell population by immunocytochemistry. For viral transduction, lentiviral vectors were added to primary NPC cultures at a multiplicity of infection (moi) of 10, and the infected cells were incubated for 48 hrs prior to analyses.

Statistics
Statistical significance was tested using an unpaired Student's t test or one-way ANOVA followed by Tukey's post hoc test. A significance of p,0.05 was required for rejection of the null hypothesis.  Figure S4 Distribution of E14.5-born neurons in the postnatal neocortex. Pregnant female mice were injected with EdU at gestational day 14.5, and the forebrains of wild-type (WT) and necdin-null (KO) mice were prepared at P4. EdU + neurons were detected by Alexa Fluor fluorescence (green) after staining DNA with Hoechst 33342 (blue) (A). HE, hematoxylin-eosin stain. Neocortical layers I-VI were divided into 10 equal bins, and the number of EdU + cells in each bin was counted within each 500mm-wide column (boxed in upper panels) after threshold setting (B). Scale bars; 500 mm (upper panel), 100 mm (lower panels) in (A). Arrows and arrowheads point to representative cells with high and low fluorescence intensities, respectively. Values represent the mean 6 SEM, n = 3. *p,0.05, **p,0.01. (TIF) Figure S5 Necdin and Bmi1 mRNA levels in primary NPCs and neurons. Primary NPCs and neurons were prepared from the neocortex of E14.5 wild-type (WT) and necdin-null (KO) mice. The necdin (A) and Bmi1 (B) mRNA levels were analyzed by qRT-PCR. Values represent the mean 6 SEM, n = 3. (TIF) Figure S6 Necdin exerts no modulatory effects on p19 and p21 expression in neocortical NPCs. Neocortical NPCs prepared from wild-type (WT) and necdin-null (KO) mice at E14.5 were infected with lentiviruses for GFP expression (Control) and necdin overexpression (Ndn OE). The p19 (A) and p21 (B) mRNA levels in lentivirus-infected NPCs were analyzed by qRT-PCR. Values represent the mean 6 SEM, n = 3. NS, not significant (p.0.05). (TIF) Figure S7 Bmi1 exerts modulatory effects on p19 and p21 expression in neocortical NPCs. Neocortical NPCs prepared from wild-type (WT) and necdin-null (KO) mice at E14.5 were infected with lentiviruses for GFP expression (Control), Bmi1 overexpression (Bmi1 OE), and Bmi1 shRNA (Bmi1 KD). The p19 (A) and p21 (B) mRNA levels in lentivirus-infected NPCs were analyzed by qRT-PCR. Values represent the mean 6 SEM, n = 3. *p,0.05, **p,0.01. (TIF)