Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Dev Biol. Author manuscript; available in PMC 2010 December 3.
Published in final edited form as:
PMCID: PMC2996717

Bmi-1 over-expression in neural stem/progenitor cells increases proliferation and neurogenesis in culture but has little effect on these functions in vivo


The Polycomb gene Bmi-1 is required for the self-renewal of stem cells from diverse tissues, including the central nervous system (CNS). Bmi-1 expression is elevated in most human gliomas, irrespective of grade, raising the question of whether Bmi-1 over-expression is sufficient to promote self-renewal or tumorigenesis by CNS stem/progenitor cells. To test this we generated Nestin-Bmi-1-GFP transgenic mice. Analysis of two independent lines with expression in the fetal and adult CNS demonstrated that transgenic neural stem cells formed larger colonies, more self-renewing divisions, and more neurons in culture. However, in vivo, Bmi-1 over-expression had little effect on CNS stem cell frequency, subventricular zone proliferation, olfactory bulb neurogenesis, or neurogenesis/gliogenesis during development. Bmi-1 transgenic mice were born with enlarged lateral ventricles and a minority developed idiopathic hydrocephalus as adults, but none of the transgenic mice formed detectable CNS tumors, even when aged. The more pronounced effects of Bmi-1 over-expression in culture were largely attributable to the attenuated induction of p16Ink4a and p19Arf in culture, proteins that are generally not expressed by neural stem/progenitor cells in young mice in vivo. Bmi-1 over-expression therefore has more pronounced effects in culture and does not appear to be sufficient to induce tumorigenesis in vivo.

Keywords: Stem cell, Bmi-1, transgenic mouse, over-expression, central nervous system (CNS), glioma, glioblastoma, tumorigenesis


Bmi-1 is a Polycomb group protein that functions as one component of the repressor complex 1 to modify histones, regulating chromatin structure and repressing transcription (Valk-Lingbeek et al., 2004). Bmi-1 is required for the postnatal self-renewal of stem cells from diverse tissues including the hematopoietic system (Lessard and Sauvageau, 2003; Park et al., 2003) and the CNS (Bruggeman et al., 2005; Molofsky et al., 2005; Molofsky et al., 2003). Bmi-1 deficient mice exhibit a progressive postnatal depletion of stem cells from these tissues, leading to hematopoietic failure, defects in cerebellum development, neurological abnormalities, and death by early adulthood (Leung et al., 2004; van der Lugt et al., 1994). Bmi-1 promotes the self-renewal of neural stem cells and other stem cells largely, but not exclusively, by repressing the expression of Ink4a and Arf (Bruggeman et al., 2005; Jacobs et al., 1999a; Molofsky et al., 2005; Molofsky et al., 2003). Ink4a and Arf encode the p16Ink4a and p19Arf tumor suppressor proteins that inhibit cell cycle progression and induce cellular senescence (Lowe and Sherr, 2003). Bmi-1 thus promotes the maintenance of CNS stem cells throughout adult life by repressing tumor suppressors associated with cellular senescence.

Bmi-1 was originally identified as a transgene that could co-operate with myc to induce hematopoietic malignancies (Haupt et al., 1993; Jacobs et al., 1999; van Lohuizen et al., 1991). Elevated Bmi-1 expression has been observed in hematopoietic malignancies (Bea et al., 2001), medulloblastomas and gliomas (Bruggeman et al., 2007; Leung et al., 2004), and cancers from other tissues (Song et al., 2006; Tateishi et al., 2006; Vonlanthen et al., 2001; Wang et al., 2007). Indeed, Bmi-1 is necessary for the maintenance of cancer stem cells from acute myeloid leukemias (Lessard and Sauvageau, 2003) as well as gliomas (Bruggeman et al., 2007). Bmi-1 is thus necessary for the progression of certain types of cancer and increased Bmi-1 expression may contribute to tumorigenesis in certain contexts.

These observations raise the question of whether over-expression of Bmi-1 in neural stem/progenitor cells is sufficient to enhance the self-renewal of these cells or to render them tumorigenic. Retroviral over-expression of Bmi-1 can increase hematopoietic stem cell self-renewal in culture and enhance the ability of these cells to reconstitute irradiated mice (Iwama et al., 2004). However, it is important to distinguish between the effects of Bmi-1 in culture and in vivo because the Bmi-1 target gene Ink4a-Arf is regulated differently in culture: Ink4a-Arf is rarely expressed by normal cells in fetal or young adult mice in vivo but is induced in cultured cells as a stress response to unphysiological conditions (Lowe and Sherr, 2003; Sherr and DePinho, 2000). Consistent with this, Bmi-1 can be necessary for maintaining stem cell self-renewal in culture in a way that is not observed in vivo in certain contexts: fetal neural stem cells from Bmi-1 deficient mice exhibit a self-renewal defect in culture but no detectable defect in Bmi-1 deficient mice in vivo (Molofsky et al., 2003). Thus, as a result of increased Ink4a-Arf expression in culture, Bmi-1 can sometimes exhibit functions in culture that are not observed in vivo. Nonetheless, Bmi-1 also promotes stem cell self-renewal and cancer cell proliferation through Ink4a-Arf-independent mechanisms (Bruggeman et al., 2007; Chagraoui et al., 2006). Resolving the question of whether Bmi-1 over-expression is sufficient for tumorigenesis and increased stem cell self-renewal in vivo will be critical to determine whether elevated Bmi-1 expression can be an initiating event in the etiology of brain tumors.

We examined Bmi-1 expression in both high-grade and low-grade human gliomas as well as in normal CNS tissue. Consistent with a recent report (Bruggeman et al., 2007), we observed elevated Bmi-1 expression in most high grade and low grade gliomas. To test whether Bmi-1 over-expression is sufficient to promote self-renewal and tumorigenesis by stem/progenitor cells in the CNS, we generated Nestin-Bmi-1-GFP transgenic mice that expressed Bmi-1 throughout CNS stem and progenitor cells. While Bmi-1 over-expression was sufficient to increase self-renewal, overall proliferation, and neuronal differentiation by CNS stem and progenitor cells in culture, little change in neural stem/progenitor cell function, neurogenesis or gliogenesis was observed in vivo. A minority of the transgenic mice developed idiopathic hydrocephalus and died early in adulthood, but none of the transgenic mice developed CNS tumors at any developmental stage up to and including two years of age. These results indicate that the consequences of Bmi-1 over-expression in stem/progenitor cells are more dramatic in culture and that Bmi-1 over-expression is not sufficient to drive tumorigenesis by CNS stem/progenitor cells in vivo.


Generation of Nestin-Bmi-1-HA-GFP transgenic mice

Full-length mouse Bmi-1 cDNA was PCR amplified from total mouse embryo cDNA and HA-tagged, then inserted into the pMIG vector immediately upstream of the IRES site. The Bmi-1-IRES-GFP fragment was then isolated by NotI digestion and inserted into NotI digested nes1852tk/lacZ plasmid (kindly provided by Dr. Urban Lendahl). This nes1852tk/lacZ plasmid carried the Nestin second intronic enhancer that has previously been shown to drive transgene expression in CNS stem and progenitor cells (Kawaguchi et al., 2001) as well as the thymidine kinase minimal promoter (Lothian and Lendahl, 1997). The entire sequence (~7600 bp) was then linearized with ScaI and microinjected into mouse zygotes (Hogan et al., 1994). Founder males were bred to wild-type C57Bl/6 females. Offspring were screened by GFP phenotyping of the brain in neonatal pups, and by genotyping using PCR at later ages using the primer set: 5′-TGGACACAAAGACCTCTGTG-3′ and 5′-GGTTGTTCGATGCATTTCTGC-3′.

Isolation of CNS progenitors and flow cytometry analysis

Adult lateral ventricle subventricular zone (SVZ) cells from 2-4 month old wild type or transgenic mice were obtained by micro-dissecting the lateral walls of the lateral ventricles, then dissociating for 20 min at 37 °C in 0.025% trypsin/0.5 mM EDTA (Calbiochem; San Diego, CA) plus 0.001% DNase1 (Roche; Indiannapolis, IN). After quenching the enzymatic dissociation with staining medium (L15 medium (Invitrogen; Carlsbad, CA) supplemented with 1 mg/ml BSA (Sigma; St. Louis, MO), 10 mM HEPES (pH 7.4) and 1% penicillin/streptomycin (Invitrogen)) containing 0.014% soybean trypsin inhibitor (Sigma) and 0.001% DNase1, the cells were washed and resuspended in staining medium, triturated, filtered through nylon screen (45 μm, Sefar America; Depew, NY), counted by haemocytometer, and plated.

For flow-cytometric analysis of GFP expression in transgenic mouse brain tissue, freshly dissociated single cell suspensions from neonatal mouse cerebral cortex and cerebellum, or adult SVZ were analyzed by a FACS VantageSE flow-cytometer (Becton-Dickinson; San Jose, CA). Dead cells were excluded from analysis by DAPI staining.

Neural stem cell culture, differentiation, cell proliferation and self-renewal assays

For neurosphere culture, dissociated SVZ cells were plated at clonal density (1.3 cells/μl of culture medium) onto ultra-low binding plates (Corning, Lowell, MA). The culture medium (self-renewal medium) was based on a 5:3 mixture of DMEM-low glucose: neurobasal medium (Invitrogen) supplemented with 20 ng/ml human bFGF (R&D Systems, Minneapolis, MN), 20 ng/ml EGF (R&D Systems), 1% N2 supplement (Invitrogen), 2% B27 supplement (Invitrogen), 50 μM 2-mercaptoethanol (Sigma), 1% pen/strep (Invitrogen) and 10% chick embryo extract (prepared as described (Stemple and Anderson, 1992)). All cultures were maintained at 37°C in 6% CO2/balance air. Neurosphere numbers and diameters were determined on the tenth day of culture.

For neurosphere culture from single cells, P5 wild type and transgenic mouse SVZ cells were dissociated as described above. To enrich neural stem cells, the dissociated SVZ cells were stained with anti-CD24 (BD Biosciences, San Jose, CA) and anti-CD15 (clone MMA) antibodies, and single CD24CD15hi cells were sorted into each well of 96 well ultra-low binding plates. Single cell sorting was confirmed in control plates stained with DAPI. Cultures were maintained at 37°C in 6% CO2/balance air. Neurospheres were measured and subcloned for self-renewal quantification on the tenth day of culture.

To assess the differentiation of primary neurospheres, individual neurospheres were replated onto Poly-D-Lysine (PDL, Biomedical technology, Stoughton, MN) and fibronectin (Biomedical technology) coated 48-well plates and cultured adherently in differentiation medium for 7 days before being fixed and stained for markers of neurons, oligodendrocytes, and astrocytes as previously described (Molofsky et al., 2003). Differentiation medium was the same as self-renewal medium except that it contained no EGF, and reduced concentrations of chick embryo extract (1%) and FGF (10 ng/ml).

For cell proliferation assays, dissociated adult SVZ cells were plated onto PDL and laminin (Invitrogen) coated tissue culture plates at clonal density (0.7 cells/μl) and cultured for 6 days, then pulse-labeled with 10 μM BrdU for 1 hour at 37°C before being fixed and stained with an antibody against BrdU as described (Molofsky et al., 2003). Only stem cell colonies (identified by large compact colony morphology) were included in this analysis.

To assess the self-renewal potential of cultured neural stem cells, individual primary neurospheres were mechanically dissociated by trituration and then replated at clonal density (1.3 cells/μl) into non-adherent cultures. Secondary neurospheres were counted 10 days later and differentiated in the same manner as described above. Self-renewal is reported as the number of multipotent secondary neurospheres that arose per subcloned primary neurosphere.

Generation of Bmi-1-GFP-MSCV virus and infection of neural stem cells

Bmi-1 bearing retrovirus was made using a mouse stem cell virus (MSCV) vector (pMIG) which contains an internal ribosome entry site (IRES) followed by GFP (Van Parijs et al., 1999). Full length mouse Bmi-1 cDNA was PCR amplified from perinatal mouse lateral ventricle SVZ cDNA. Hpa1 and Bgl2 sites were added to either end and the sequence was inserted into the MSCV vector 5' to the IRES GFP. Preparation of high-titer virus was carried out by calcium phosphate transfection of BOSC packaging cells (Pear et al., 1993) with the Bmi-MSCV vector. Vesicular stomatitis virus G protein (VSV-G) plasmid was added during transfection to increase the infectivity of the packaged virus (Lee et al., 2001). Cells were allowed to produce virus for 24-48 hours, then cell supernatants were collected, 0.2 micron filtered, and stored at −80°C or used fresh to infect cultured neural stem/progenitor cells.

Neonatal lateral ventricular zone (VZ) tissue from wild type or Ink4a-Arf deficient mice was dissected, dissociated, and cultured as described above. Cells were plated adherently at high density (20,000 cells per well of a six-well plate). After 24-48 hours of culture, viral supernatant was added for 12-24 hours. Then the culture medium was replaced for 24-48 hours, and cells were briefly trypsinized and replated at clonal density (0.7 cells/μl of media). Viral titers allowed infection of 25 to 90% of colonies.

Immunohistochemistry of tissue sections

Immunohistochemical staining of Bmi-1 on normal human brain tissue and brain tumor samples was performed on the DAKO Autostainer (DAKO, Carpinteria, CA) using DAKO Envision+ and diaminobenzadine (DAB) as the chromogen. De-paraffinized sections of formalin fixed tissue at 5μm thickness were labeled with mouse monoclonal anti-Bmi-1 (1:400, Clone F6, Upstate Chemicals, Pickens, SC) for 60 minutes, after microwave citric acid epitope retrieval. Appropriate negative (no primary antibody) and positive controls (high grade brain tumor) were stained in parallel with each set of tumors studied. The immunoreactivity was scored by a three-tier (negative, low (+) and high (++) positive) grading scheme.

For immunofluorescence staining of mouse brain sections, whole embryos or postnatal brains were fixed in 4% paraformaldehyde overnight, then cryoprotected in 30% sucrose, embedded in OCT (Sakura Finetek, Torrance, CA) and flash frozen. Twelve micron sections were cut on a Leica cryostat and air dried over night. Tissue sections were first blocked in PG/HN (5% horse serum, 5% goat serum, 1% BSA, 0.05% sodium azide and 0.05% Triton X-100 in 1x PBS) for 1 hour at room temperature (RT). Primary antibodies were diluted in PG/HN and incubated with the sections for 1 hour at RT, followed by washing then secondary antibody incubation for 1 hour at RT. Antibodies included mouse monoclonal anti-GFAP (1:500, Clone GA-5, Sigma); mouse monoclonal anti-b-III-Tubulin (1:500, Clone TUJ1, Covance); rat monoclonal anti-myelin basic protein (MBP, 1:100, Millipore, Temecula, CA), rabbit polyclonal anti-BFABP (1:2000, gift from T. Muller, Max-Delbruck-Center, Berlin, Germany), mouse monoclonal anti-Reelin (1:1000, Clone G10, Abcam), rabbit polyclonal anti-FoxP2 (1:500, Abcam) and Alexa 488 conjugated rabbit polyclonal anti-GFP (1:400, Invitrogen). Alexa Fluor conjugated secondary antibodies were purchased from Invitrogen and used at 1:500 dilution. Nuclei were visualized using DAPI (2μg/ml, Sigma) and slides were mounted with Prolong antifade solution (Invitrogen). Images were collected on a Leica TCS SP5 laser scanning confocal microscope, and multi-channel overlays were assembled in Adobe Photoshop.

For immunohistochemical staining of Bmi-1 on E15 telecephalon, 12 μm frozen sections of E15 wild type and transgenic mouse telencephalon were first treated with 3% H2O2/0.1% NaN3 for 1h at RT to quench endogenous peroxidase activity, then blocked with PG/HN before being labeled with mouse monoclonal anti-Bmi-1 (1:400, Clone F6, Upstate Chemicals) for 60 minutes at room temperature. The staining were then visualized using DAKO Envision+ and diaminobenzadine (DAB) as the chromogen.

To quantify adult SVZ proliferation, mice were injected with 100 mg/kg of 5-bromo-2-deoxyuridine (BrdU, Sigma), and sacrificed 2 hour later. Brains were fixed and embedded as described above. Twelve micron thick coronal sections were cut on a Leica cryostat. For detection of BrdU, DNA was first denatured in 2N HCl for 45 min at RT and neutralized with 0.1M Sodium Borate for 10 minutes. Sections were then pre-blocked in PG/HN for 1 hour at RT and stained with primary rat anti-BrdU antibody (1:500, Accurate Chemical, Westbury, NY) diluted in PG/HN for 1 hour at room temperature, followed by biotinylated goat-anti-rat IgG (Jackson Immunoresearch) for 30 minutes at RT. BrdU staining were then visualized with Alexa594-conjugated streptavidin (Invitrogen) and nuclei were counter-stained with DAPI. Slides were mounted using ProLong antifade solution (Invitrogen) and imaged on a Leica TCS SP5 laser scanning confocal microscope using a 63x oil immersion objective (NA=1.4).

To assess proliferation in E14.5 telencephalon, timed pregnant females were injected with a single dose of 100mg/ml BrdU 15 minutes prior to euthanization. Tissue processing and BrdU staining were then performed in a similar manner as described for adult tissue.

To quantify olfactory bulb neurogenesis, 1-2 month old mice were given a single injection of 100mg/kg of BrdU and kept on BrdU-containing drinking water (1mg/ml) for the next 7 days. They were switched to regular water for an additional 4 weeks before being sacrificed. Brains were fixed and embedded as described. Ten micron coronal sections of the olfactory bulb were cut on a Leica cryostat. Tissue sections were blocked with PG/HN, stained with anti-NeuN (1:100, Clone A60, Millipore), re-fixed with 4% PFA then stained for BrdU as described above. NeuN staining was visualized with Alexa555-conjugated goat-anti-mouse IgG1 secondary antibody (Invitrogen). BrdU staining was visualized with biotinylated goat-anti-rat IgG secondary antibody and Alexa594 conjugated streptavidin. Nuclei were counter-stained with DAPI. Slides were mounted in Prolong antifade (Molecular Probes), and imaged on a Leica TCS SP5 laser scanning confocal microscope using a 63x oil immersion objective (NA=1.4). Twenty five random fields of view (each containing 300-500 cells) spanning the full thickness of the olfactory bulb were counted in ImageJ software.

Neuron birth dates were determined by injecting timed pregnant females with a single dose of 100mg/ml 5-chloro-2-deoxyuridine (CldU, Sigma) at E17. Pups were sacrificed at P20, brains were fixed and embedded as described, and 40μm-thick floating sections were cut on a Leica cryostat. CldU staining was performed in a similar manner as described for BrdU, using CldU-specific (1:500, Clone BU1/75, Accurate Chemical) antibody (Kiel et al., 2007); Cux-1 was stained using a rabbit polyclonal anti-Cux1 antibody (1:50, Santa Cruz Biotechnology; Santa Cruz, CA). CldU staining was visualized by biotinylated goat-anti-rat IgG secondary antibody and Alexa594 conjugated streptavidin, and Cux-1 staining was visualized using Alexa555 conjugated goat-anti-rabbit IgG secondary antibody.

MRI analysis

All mice analyzed were anesthetized with 2% isoflurane/air mixture throughout MRI. Mice lay prone, head first in a 7.0T Varian MR scanner (183-mm horizontal bore, Varian, Palo Alto, CA), with the body temperature maintained at 37°C using circulated heated air. A double-tuned volume radio frequency coil was used to scan the head region of the mice. Axial T2-weighted images were acquired using a fast spin-echo sequence with the following parameters: repetition time (TR)/effective echo time (TE), 4000/47.456 ms; field of view (FOV), 30×30 mm; matrix, 256×128; slice thickness, 0.5 mm; slice spacing, 0 mm; number of slices, 25; and number of scans, 1 (total scan time ~1 min).


Bmi-1 expression is elevated in a subset of human CNS tumors

We examined the expression of Bmi-1 in normal and diseased human parenchymal brain tissue by performing immunohistochemistry on fixed tissue sections. We did not detect Bmi-1 expression in any of 3 samples of normal brain tissue (Suppl. Fig. 1A). We did observe strong Bmi-1 staining in the vast majority of cells within a white matter biopsy from a patient after an epileptic seizure (Suppl. Fig. 1B). Bmi-1 staining was also observed in most specimens from patients that exhibited gliosis due to damage (mesial temporal lobe sclerosis) or abscess (4 of 5 specimens; Suppl. Fig. 1C). These results demonstrate that Bmi-1 can be upregulated, particularly in glia, after brain injuries.

Similar to what has been reported recently (Bruggeman et al., 2007), we also observed elevated Bmi-1 expression in a subset of human gliomas (Suppl. fig. 1D-F). Strong Bmi-1 expression was observed in most cells from 10 of 30 glioblastoma specimens (Suppl. Fig. 1F), while 7 of 30 specimens exhibited detectable, but lower levels of Bmi-1 expression in fewer cells. Variable levels of Bmi-1 expression were detected in some cells in low grade gliomas (1 of 2 specimens, data not shown), pilocytic astrocytomas (3 of 4 specimens; Suppl. Fig. 1D), and oligodendrogliomas (4 of 5 specimens; Suppl. Fig. 1E). These data indicate that Bmi-1 expression is elevated in most human CNS tumors, though no strong correlation between Bmi-1 expression level and the type or grade of the tumors was observed (Suppl. Fig. 1H). These data raise the question of whether elevated Bmi-1 expression is sufficient to drive tumorigenesis in CNS stem/progenitor cells.

Generation of Nestin-Bmi-1-GFP transgenic mice

To examine the effects of Bmi-1 over-expression on neural stem/progenitor cell function and CNS tumorigenesis in vivo, we generated transgenic mice in which the human Nestin second intron enhancer was used to drive the expression of HA-tagged mouse Bmi-1 protein, as well as a GFP reporter in CNS stem and progenitor cells (Fig. 1A). Nestin is expressed in CNS stem and progenitor cells (Dahlstrand et al., 1992b; Kawaguchi et al., 2001), as well as in various CNS tumors (Almqvist et al., 2002; Dahlstrand et al., 1992a; Rutka et al., 1999). The human Nestin second intron enhancer has been shown to drive transgene expression in mouse CNS stem and progenitor cells during embryonic development and adulthood (Lothian and Lendahl, 1997).

Figure 1
Transgene expression in Nestin-Bmi-1-GFP transgenic mice

Ten of 77 (13%) pups born after zygote injection harbored the Nestin-Bmi-1-GFP construct based on PCR genotyping of tail DNA as well as by observation of GFP expression in the CNS (analyzed by transillumination of the head). Although one transgenic pup died shortly after birth of unknown causes, most of the transgenic offspring appeared healthy and developed normally. Southern blotting was performed on the surviving founders to confirm transgene integration and to determine copy number (Suppl. fig. 2A). Five lines that gave germ line transmission of the transgene were successfully established. One line (line B*) exhibited transgene segregation among F1 progeny in a way that suggested the original founder had multiple transgene integration sites (Suppl. fig. 2B). As a result, two sub-lines from line B* were established: line B that had up to five copies of the transgene in a single integration site and line K that had more than 25 copies of the transgene.

To select for lines that exhibited transgene expression in CNS stem/progenitor cells throughout development and adulthood, we analyzed GFP expression in neonatal and adult F1 transgenic mouse brains by direct observation of GFP fluorescence and by analysis of dissociated cells from various CNS regions using flow-cytometry. Surprisingly, only the two transgenic lines (lines B and C) with the lowest transgene copy number (approximately 5 copies of the transgene inserted into a single site in the genome) maintained GFP expression in both the newborn and adult brain (Suppl. fig. 3A). All other lines either showed no transgene expression (lines D and K) at either stage, or showed only weak GFP expression at birth that was no longer detectable in the adult brain (lines E and G) (Suppl. fig. 3A). Therefore, for our analysis of the consequences of Bmi-1 over-expression in vivo we studied lines B and C. The transgene was inherited with expected Mendelian frequencies in these lines (Suppl. fig. 4A).

Characterization of transgene expression in vivo

GFP expression in the transgenic mouse CNS could be detected at all developmental stages in patterns that resembled endogenous Nestin expression (Lothian and Lendahl, 1997) in both lines B and C. At embryonic day (E)14.5, GFP was ubiquitously expressed throughout the developing CNS (Fig. 1B). In contrast, we did not observe GFP fluorescence in tissues outside of the nervous system or in littermate controls (Fig. 1B). At birth (P0), GFP was still widely expressed in all brain regions of transgenic mice but not littermate controls (Fig. 1C). The highest levels of expression appeared to be in the ventricular zones throughout the brain as well as in nerve fiber bundles within the corpus callosum and anterior commissure (Fig. 1C). This widespread transgene expression was also confirmed by analyzing freshly dissociated cells from various neonatal brain regions by flow-cytometry: most cells from the P0 transgenic lateral ventricle germinal zone (VZ; 90±8%), cerebral cortex (88±9%), and cerebellum (66±11%) expressed GFP (Fig. 1G).

In contrast, GFP expression in the adult transgenic brain became more spatially restricted. Whole brain GFP fluorescence was greatly reduced as compared with E14.5 or P0 brain (Fig. 1D); and most of the GFP expressing cells were localized to brain regions that contain neural stem/progenitor cells, such as the SVZ of the lateral ventricle (Fig. 1E) and the subgranular cell layer in the dentate gyrus of the hippocampus (Fig. 1F). Brain layers that contained differentiated cells showed no detectable GFP expression (Fig. 1E,F). GFP expression was also not observed in tissue lacking appreciable Nestin expression, such as in adult bone marrow (Suppl. fig. 3B). The enrichment of GFP expressing cells in adult SVZ was further confirmed by flow-cytometry, which showed that 76±9% of all dissociated SVZ cells were GFP+ (Fig. 1H). GFP expression thus correlated with the expected pattern of Nestin expression in stem/progenitor cells in vivo, throughout development and into adulthood.

To directly confirm that the transgene was indeed expressed by neural stem and progenitor cells, GFP+ and GFP cells from freshly dissociated neonatal cerebral cortex, VZ, and cerebellum were sorted into culture. The vast majority (>94%) of neurospheres arose from the GFP+ fraction of cells while GFP cells formed few or no neurospheres (Fig. 1I). Upon differentiation, some of the GFP+ neurospheres formed multilineage colonies (containing neurons, astrocytes and oligodendrocytes) while other GFP+ neurospheres formed colonies that contained only astrocytes, or neurons and astrocytes. GFP neurospheres usually did not form multilineage colonies. This indicates that virtually all multipotent cells (>99%) expressed the transgene but that GFP expressing cells were functionally heterogeneous and contained both neural stem and progenitor cells.

The transgene was also expressed by neural stem/progenitor cells from the brains of adult transgenic mice. SVZ cells were dissociated from the brains of two to three month old transgenic mice and GFP+ or GFP cells were sorted into culture. Over 97% of neurosphere-forming cells were GFP+ and most of these neurospheres underwent multilineage differentiation (Fig. 1J,K). Since GFP expression was observed throughout the adult SVZ, and most adult SVZ cells are restricted progenitors that fail to form multipotent colonies in culture, the data suggest that many restricted progenitors in the adult brain also expressed the transgene. These data suggest that newborn and adult mice exhibited transgene expression in neural stem cells as well as in restricted neural progenitors.

The transgene increases Bmi-1 expression in neural stem cells

To assess the level of Bmi-1 over-expression as a result of transgene expression, we performed quantitative (real-time) reverse transcription polymerase chain reaction (qRT-PCR) and western blotting using freshly isolated SVZ cells and cultured neurospheres. In cultured neurospheres from adult SVZ, Bmi-1 transcript levels were 15-fold and 9-fold higher in line B and Line C transgenic neurospheres, respectively, as compared with neurospheres from littermate controls by qRT-PCR (Fig. 2A). Using uncultured tissue, Bmi-1 transcript levels were 8.5 to 11.1-fold higher in P0 ventricular zone and in the adult SVZ of line B and Line C transgenic mice as compared to littermate controls (Fig. 2A). At the protein level, using anti-HA antibody, we were able to detect the expression of the HA-tagged Bmi-1 in both cultured neurospheres from adult SVZ and in freshly dissected E14.5 telencephalon tissue and P0 brain from lines B and C (Fig. 2B-D). Using anti-Bmi-1 antibody, this translated to a significant overall increase in Bmi-1 expression by cultured neurospheres and uncultured telencephalon/P0 brain cells from transgenic mice (Fig. 2B-D). Furthermore, immunohistochemical staining of Bmi-1 in tissue sections from E15 transgenic and control mouse telencephalon showed significantly increased Bmi-1 expression in most VZ cells in the transgenic mice (Fig. 2E). Together, these data demonstrate that transgene expression led to Bmi-1 over-expression in neural stem and progenitor cells in vitro and in vivo throughout CNS development. Moreover, the level of Bmi-1 expression within transgenic mice appeared similar or higher than the level observed within human glioma specimens by both immunohistochemistry (compare Fig. 2E to Suppl. Fig. 1A-F) and western blot (Suppl. Fig. 1G).

Figure 2
Bmi-1 expression was elevated in transgenic neural stem/progenitor cells

To test whether the increase in Bmi-1 expression in transgenic mice correlated with increased Bmi-1 function, we assessed the expression of the Bmi-1 target genes Ink4a and Arf. The Bmi-1-containing polycomb repressor complex 1 represses Ink4a and Arf expression (Bracken et al., 2007; Bruggeman et al., 2007; Bruggeman et al., 2005; Molofsky et al., 2005). Cultured neurospheres from line B and line C exhibited lower levels of p16Ink4a and p19Arf expression as compared to neurospheres cultured from littermate controls (Fig. 2B). This demonstrated that transgenic Bmi-1 was functionally active and that Bmi-1 over-expression attenuated the induction of p16Ink4a and p19Arf that occurs in cultured cells. Consistent with earlier studies that failed to detect p16Ink4a or p19Arf expression in fetal and young adult mouse brain (Bruggeman et al., 2007; Bruggeman et al., 2005; Molofsky et al., 2005; Molofsky et al., 2003; Nishino et al., 2008) we were also unable to detect these proteins in transgenic mice or littermate controls at E14.5 (Fig. 2C) or P0 (Fig. 2D). The observation that over-expressed Bmi-1 was able to reduce p16Ink4a and p19Arf expression in culture, but that p16Ink4a and p19Arf expression could not be detected in vivo raised the possibility that Bmi-1 over-expression would promote self-renewal/proliferation to a greater degree in culture.

Transgenic neural stem cells show increased self-renewal and neurogenesis in culture

To assess the consequences of Bmi-1 over-expression we compared the frequency of adult SVZ cells from transgenic and control mice that formed multipotent neurospheres, the diameter of these neurospheres, as well as their self-renewal potential. A modestly (1.7-fold) but significantly (p<0.05) higher percentage of dissociated adult SVZ cells from line B and line C transgenic mice formed multipotent neurospheres in culture (Fig. 3A). However, when multiplied by the total number of dissociated SVZ cells obtained from each mouse, this did not translate into a significant increase in the total number of multipotent neurospheres per SVZ in the transgenic mice (Fig. 3B). The slight increase in the frequency of multipotent neurospheres could either reflect a slightly increased frequency of stem cells in the SVZ of transgenic mice in vivo or slightly increased survival of these cells in culture. The neurospheres formed by transgenic cells were significantly larger than the neurospheres from littermate controls (Fig. 3C), demonstrating that elevated Bmi-1 expression increased the proliferation of stem/progenitor cells in culture. This is further supported by the significantly increased proliferation rate observed in adherently cultured transgenic neural stem cell colonies based on BrdU incorporation (Fig. 3G). To focus on the self-renewal potential of stem cells rather than the overall proliferation of the heterogeneous progenitors that compose stem cell colonies, we dissociated and subcloned individual primary neurospheres into secondary cultures and counted the number of multipotent secondary neurospheres that arose per primary neurosphere. Stem cells from line B and line C transgenic mice exhibited dramatically increased self-renewal in culture as compared to stem cells from littermate controls (Fig. 3D).

Figure 3
Bmi-1 over-expression increased the self-renewal and neuronal differentiation of neural stem/progenitor cells in culture

At the extraordinarily low cell densities used in our experiments (approximately 1 cell/μl) control experiments indicated that most neurospheres were clonally derived (Suppl. fig. 5). Nonetheless, to confirm that Bmi-1 increased the self-renewal of clonal neural stem cell colonies we sorted individual CD15+CD24 SVZ cells from P5 wild type and transgenic mice into different wells of 96 well plates. CD15+CD24 SVZ cells are enriched for neural stem cells (Capela and Temple, 2002). Neurospheres arose in only 10-17% of wells seeded with single CD15+CD24cells. We found that transgenic neurospheres were significantly larger and had significantly more self-renewal potential than wild-type neurospheres (Fig. 3E, F). These data demonstrate that elevated Bmi-1 expression increases the proliferation and self-renewal of CNS stem cells in clonal culture.

Loss of Bmi-1 reduces neuronal differentiation from CNS stem cells in culture (Bruggeman et al., 2007; Zencak et al., 2005). To test whether over-expression of Bmi-1 affects stem cell differentiation, we differentiated primary neurospheres from both transgenic lines to assess their ability to generate Tuj-1+ neurons, GFAP+ astrocytes and O4+ oligodendrocytes in culture. The frequency of oligodendrocytes was comparable in transgenic and wild type multipotent stem cell colonies (Fig 3I; 0.5±0.6% of cells in wild type, 1.0±1.1% of cells in line B transgenic and 1.1±1.0% in line C transgenic respectively). However, transgenic stem cell colonies consistently contained an increased proportion and absolute number of neurons as compared to control colonies (3.9±4.9% in wild type, 23±18% in line B transgenic and 31±17% in line C transgenic respectively) (Fig. 3H,I;). Transgenic stem cell colonies also contained a modestly reduced frequency of astrocytes as compared to wild-type colonies (95±4% in wild type, 76±18% in line B transgenic and 67±17% in line C transgenic respectively). Since multipotent transgenic colonies contained many more cells than multipotent wild type colonies (3,700±2,700 cells/ wild type colony versus 21,000±17,000 cells/ line B colony and 10,000±8,000 cells/ line C colony after 10 days in culture), the reduced frequency of astrocytes in transgenic colonies reflected the dramatic increase in the number of neurons in these colonies, rather than a decrease in the absolute number of astrocytes per colony. Bmi-1 over-expression thus increased the neurogenic capacity of CNS stem cells in culture.

Bmi-1 over-expression has little effect on neural stem/progenitor cell proliferation in vivo

To evaluate the effects of Bmi-1 over-expression in vivo, we compared various measures of stem/progenitor cell function between control and Bmi-1 transgenic mice. We administered a pulse of BrdU to 2-4 month old transgenic and littermate control mice two hours before they were sacrificed to test whether Bmi-1 over-expression increased the overall proliferation of neural stem and progenitor cells in the SVZ. In contrast to the significantly enhanced proliferation exhibited by cultured transgenic neural stem/progenitor cells (Fig. 3G), neither line B nor line C transgenic mice showed a significant difference in the frequency of BrdU+ cells in the SVZ when compared with littermate controls (Fig. 4A).

Figure 4
Bmi-1 over-expression had little effect on fetal or adult neural stem/progenitor cell proliferation in vivo

To test whether Bmi-1 over-expression could affect neural stem/progenitor cell proliferation during fetal development, we administered BrdU to timed pregnant females to mark dividing cells in the developing telencephalon of wild type and transgenic embryos. After a 15 minute BrdU pulse, many cells in the VZ of both wild type and transgenic embryos became BrdU+, indicating rapid proliferation. However, we did not observe any difference in the distribution or frequency of BrdU+ cells between transgenic and control embryos (Fig. 4B). In addition, the overall thickness of the cerebral cortex and the VZ did not significantly differ between transgenic and wild-type telencephalon (Fig. 4C). These data suggest that Bmi-1 over-expression does not cause a notable increase in proliferation within the telencephalon during fetal development.

Bmi-1 over-expression has little effect on fetal or adult neurogenesis in vivo

To assess whether Bmi-1 over-expression affected the rate of olfactory bulb neurogenesis in vivo, we administered BrdU to 1-2 month old transgenic and littermate control mice for one week to mark dividing SVZ progenitors followed by a one month chase without BrdU to allow these cells to migrate into the olfactory bulb and to differentiate into neurons. We then quantified the percentage of NeuN+ neurons that were also BrdU+ by confocal microscopy. Twenty-five random fields of view spanning the whole thickness of the olfactory bulb were imaged and counted. Transgenic mice did not exhibit an increased rate of neurogenesis as compared to littermate controls (Fig. 5A). Indeed, line B transgenic mice exhibited a significantly lower rate of neurogenesis than littermate controls because a subset (approximately 25%) of these mice were found to be in the process of developing hydrocephalus when sacrificed, and the mice with hydrocephalus tended to have reduced SVZ proliferation and olfactory bulb neurogenesis (data not shown). Thus we could find no evidence of increased SVZ proliferation or neurogenesis in adult transgenic mice in vivo, in contrast to what we had observed in culture.

Figure 5
Bmi-1 over-expression had little effect on fetal or adult neurogenesis

To test whether Bmi-1 over-expression could affect neurogenesis during fetal CNS development, we examined the onset of cortical neurogenesis as well as the organization of the cerebral cortex in the forebrain of transgenic and control mice. Most cortical neurogenesis in mice occurs from E10 to E17, when ventricular zone progenitors give rise to waves of neuronal precursors that migrate toward the pial surface, forming six distinct cortical layers. To examine whether Bmi-1 over-expression affected the onset of cortical neurogenesis, we stained E12 telencephalon with antibody against Tuj1 to label the early born Cajal-Retzius neurons within the marginal zone (Hatten, 1999). We did not observe any difference between transgenic or control mice in terms of the density or localization of these Tuj-1+ cells (Fig. 5B). This suggests that Bmi-1 over-expression did not significantly delay the onset of cortical neurogenesis.

Next we asked whether the overall organization of the cerebral cortex in transgenic mice was altered by Bmi-1 over-expression. We cut sections through the cerebral cortex of P0 or adult wild type and transgenic mice and stained with the layer specific neuronal markers Reelin (staining Cajal-Retzius neurons in layer I; (Soriano and Del Rio, 2005)), Cux-1 (staining layer 2-4 neurons; (Nieto et al., 2004)), and FoxP2 (staining layer 6 neurons in the adult brain; (Ferland et al., 2003)). In no case did we detect any difference between transgenic mice and littermate controls in the frequency or localization of Reelin+ neurons at P0 (Fig. 5C), FoxP2+ neurons in adults (Fig. 5D), or Cux-1+ neurons at P20 (Fig. 5F). This suggests that cortical neuronal progenitors acquire normal identities and a normal laminar fate despite Bmi-1 over-expression.

To assess whether neurons were born according to a normal schedule we pulse labeled wild type and transgenic embryos with the nucleotide analog 5-chloro-2-deoxyuridine (CldU) at E17 (a stage when upper layer neurons are generated (Polleux et al., 1997)), and stained sections cut at P20 with antibodies against CldU and Cux-1. The CldU staining was concentrated in upper cortical layers as expected (Fig. 5E,F) and there was a similar degree of CldU/Cux-1 co-localization in both wild-type and transgenic mice (Fig. 5G-I). Therefore, we did not detect any significant change in the timing or pattern of cortical neurogenesis in the transgenic mouse brain, suggesting that Bmi-1 over-expression did not significantly alter neurogenesis in vivo, though we cannot rule out the possibility of subtle changes.

Bmi-1 over-expression has little effect on glial differentiation in vivo

While we did not observe significant effects of Bmi-1 over-expression on SVZ proliferation or neurogenesis in embryonic or adult mice, we wondered whether Bmi-1 over-expression affected gliogenesis since a loss of Bmi-1 leads to an increase in astrocytes in vivo (Zencak et al., 2005). To test this we first examined the expression of the astrocyte marker GFAP and the oligodendrocyte marker MBP in adult mice. We did not observe any difference between line B or line C transgenic mice and littermate controls in the numbers or positions of GFAP+ cells or MBP+ cells in the corpus callosum or striatum (Fig 6A,B). While this suggested that transgenic mice did not have gross changes in gliogenesis, we also examined P4-P5 mice to assess whether there were any early changes. We did not observe any difference between line B or line C transgenic mice and littermate controls in the numbers or positions of BFABP+ glia at P5 (Fig. 6E). We also did not observe any differences among line C transgenic mice and littermate controls in terms of the numbers or positions of GFAP+ astrocytes in the corpus callosum or MBP+ oligodendrocytes in the corpus callosum and striatum at P5 (Fig. 6C,D). However, we did observe a delay in oligodendrogliogenesis in line B transgenic mice, which had fewer MBP+ oligodendrocytes as compared to littermate controls in the P4 corpus callosum and striatum while the frequency of GFAP+ astrocytes was relatively normal (Fig. 6 F,G). The fact that this delay was not observed in line C mice suggests that it may not be attributable to increased Bmi-1 expression; however, we cannot rule out the possibility that increased Bmi-1 expression delays glial differentiation in some mice under certain circumstances. Overall, our data suggest that elevated Bmi-1 expression has little effect on gliogenesis in vivo.

Figure 6
Bmi-1 over-expression had little effect on gliogenesis in vivo

Bmi-1 over-expression does not further enhance the self-renewal or neurogenic potential of cultured Ink4a-Arf−/− neural stem cells

The increased overall proliferation, stem cell self-renewal, and neurogenesis that occurred upon Bmi-1 over-expression in vitro (Fig. 3) but not in vivo (Fig. 4--6)6) correlated with the ability of over-expressed Bmi-1 to reduce p16Ink4a and p19Arf expression in vitro but not in vivo (Fig. 2). To test whether the effects of Bmi-1 over-expression on neural stem cell function in culture were mediated by its ability to oppose the induction of p16Ink4a and p19Arf expression in culture, we over-expressed Bmi-1 in neural stem cells cultured from newborn (P2-P6) wild-type and Ink4a-Arf deficient mice. This was done using a bicistronic retrovirus (MSCV-Bmi1) expressing both mouse Bmi-1 and GFP. Adherently plated newborn lateral ventricle germinal zone cells from wild type or Ink4a-Arf−/− mice were infected with this virus or a control retrovirus bearing only GFP (MSCV-GFP) then replated into non-adherent cultures to generate neurospheres. Infected cells could thus be distinguished from non-infected cells based on GFP expression. qRT-PCR indicated that Bmi-1 expression was increased by MSCV-Bmi1 infection as compared to neurospheres that were infected with the control virus (Fig. 7A).

Figure 7
Bmi-1 over-expression increased proliferation, self-renewal, and neurogenesis by wild-type but not Ink4a-Arf deficient neural stem/progenitor cells in culture

To assess overall proliferation and stem cell self-renewal we measured the diameter of neurospheres and the capacity of multipotent neurospheres to form secondary multipotent neurospheres upon subcloning, respectively. Wild type neurospheres infected with MSCV-Bmi1 virus were significantly larger than neurospheres infected with MSCV-GFP control virus or uninfected neurospheres (Fig. 7B), which is consistent with what we observed in Bmi-1 transgenic cells (Fig 3C). In contrast, Ink4a-Arf−/− neurospheres were significantly larger than wild type neurospheres but over-expression of Bmi-1 in Ink4a-Arf−/− neurospheres did not further increase their size (Fig. 7B). Similar results were observed in the self-renewal assay. While MSCV-Bmi1 infected wild type neurospheres generated significantly more multipotent secondary neurospheres as compared to uninfected neurospheres or neurospheres infected with control virus (Fig. 7C), over-expression of Bmi-1 in Ink4a-Arf−/− neurospheres did not further enhance their self-renewal potential (Fig. 7C). Finally, Bmi-1 over-expression or Ink4a-Arf deficiency each increased the amount of neuronal differentiation observed in neural stem cell colonies; however, over-expression of Bmi-1 in Ink4a-Arf deficient neural stem cells did not significantly further increase neuronal differentiation beyond what was observed with Ink4a-Arf deficiency alone (Fig. 7D).

A similar experiment was performed using freshly isolated SVZ cells from adult Bmi-1 transgenic mice crossed onto an Ink4a-Arf deficient background. Consistent with the data from retroviral Bmi-1 over-expression, transgenic Bmi-1 over-expression in Ink4a-Arf deficient mice no longer increased the frequency of multipotent neurospheres, neurosphere diameter, or self-renewal potential (Fig. 7G-I). Our data suggest that the increased proliferation, self-renewal, and neuronal differentiation of cultured neural stem cells upon Bmi-1 over-expression are largely attributable to the repressive effects of Bmi-1 on p16Ink4a and p19Arf expression.

A subset of Bmi-1 transgenic mice developed hydrocephalus but not CNS tumors

Bmi-1 transgenic mice from lines B and C were born at Mendelian frequencies (Suppl. fig. 4A). The mice had a normal appearance at birth and no behavioral abnormalities were noted; however, the transgenic mice consistently exhibited enlarged lateral ventricles (Fig. 8A). We cut sections throughout lateral ventricles to look for blockages that could prevent the flow of cerebrospinal fluid out of the lateral ventricles but never detected any blockage. Thus it is unclear why Bmi-1 transgenic mice were born with enlarged lateral ventricles.

Figure 8
A minority of Bmi-1 transgenic mice developed idiopathic hydrocephalus but showed no sign of brain tumors

Between 4 and 8 weeks of age, a minority of transgenic mice from both transgenic lines (7 of 28 (25%) mice from line B and 14 of 61 (23%) mice from line C) developed hydrocephalus. The mice with hydrocephalus could be distinguished based on the dome shaped appearance of their heads, severely enlarged lateral ventricles, and behavioral changes that included weight loss, lethargy, and ruffled fur. Hydrocephalus was confirmed in these mice by magnetic resonance imaging (MRI) (Fig. 8B). MRI images showed that lateral ventricles were most affected in these mice; the third ventricle, sylvius aqueduct, and fourth ventricle appeared relatively normal (Fig. 8B). However, we could not find any obstruction between the lateral and third ventricles nor any blockage within the third ventricle or the aqueduct that would explain the hydrocephalus phenotype (Fig. 8C). Nonetheless, it remains possible that a physical blockage exists in the ventricular system in vivo that is not readily detected by histology or MRI.

We also assessed other common causes for mouse hydrocephalus such as over-proliferation of choroid plexus, or defects in the ependyma lining the lateral ventricle. No abnormal mass of choroid plexus was observed in the transgenic mice (data not shown). Some transgenic mice with hydrocephalus did exhibit an abnormally folded ependymal cell lining of the lateral ventricle (Fig. 8D), a similar phenotype was observed in transgenic mice that never developed hydrocephalus. Thus it was unclear whether the ependymal layer abnormality contributed to the development of hydrocephalus. The ependymal cells in young adult mice did not incorporate a two-hour pulse of BrdU (data not shown), so they were not proliferating rapidly as would be expected if these cells were neoplastic. Finally, we also tested whether hydrocephalus was associated with cortical atrophy due to increased apoptosis in the cerebral cortex. However, sections from hydrocephalic transgenic mouse brains did not exhibit increased activated caspase-3 staining relative to littermate controls and little apoptosis was observed in either the control or transgenic brains (data not shown). Overall, we detected no evidence of lateral ventricle malformations or blockages that could explain why hydrocephalus developed and we never detected any tumors in the brains of these mice by either histology or MRI.

The 75% of transgenic mice that were not affected by hydrocephalus remained healthy and we never detected any abnormalities in appearance or behavior. Line C mice had normal body mass and brain mass while line B mice had slightly but significantly increased brain mass and slightly but significantly decreased body mass by young adulthood (Suppl. fig. 4B,C). We aged these mice for up to 25 months and detected no evidence of premature death or illness as compared to control littermates. To look for evidence of CNS tumors in the transgenic mice, we sacrificed newborn (3 line B and 4 line C), 1-4 month old (5 line B and 16 line C) and 15-24 month old (2 line B and 3 line C) transgenic mice and analyzed serial sections of the whole brain by histology (Fig. 8E). We also grossly examined the brains from all of the transgenic mice used in earlier experiments (more than 50 mice from each line). We detected no evidence of brain tumors in any mouse at any age (Fig. 8E). Some transgenic and wild type mice spontaneously developed non-neural neoplasms (mostly hematopoietic) after being aged for more than 20 months; however, none of these neoplasms contained GFP-expressing cells, suggesting that transgene expression did not contribute.

Immunohistochemical staining of Bmi-1 in sections (compare Fig 2E to Suppl. Fig. 1A-F) and Bmi-1 western blot (Suppl. Fig. 1G) suggest that the level of Bmi-1 expression in the transgenic mouse brain is comparable to the level of expression observed in brain tumors, at least in the fetal telencephalon. However, we have been unable to examine levels of Bmi-1 protein expression in the SVZ of adult transgenic mice because the adult SVZ yielded insufficient material for western blot, and high background staining of the adult SVZ by immunohistochemistry (the anti-Bmi-1 antibody was made in mouse) made this difficult to interpret. Nonetheless, Bmi-1 transcript levels were elevated in the transgenic adult SVZ by 8.5 to 11-fold relative to control SVZ (Fig. 2A). As a result, available data suggest that Bmi-1 expression levels in CNS stem/progenitor cells from transgenic mice were comparable to those observed within brain tumors, but we cannot rule out the possibility that further increases in expression within adult CNS cells might render these cells tumorigenic. Nonetheless, it is not clear whether it is possible to achieve further increases in expression as all transgenic mice with higher transgene copy numbers silenced the transgene. Since prior studies have suggested that Bmi-1 is degraded when unable to bind other polycomb repressor complex 1 components (Ben-Saadon et al., 2006; Hernandez-Munoz et al., 2005), the stoichiometry of other complex components may limit the degree to which Bmi-1 expression and function can be increased.


In this study, we generated Nestin-Bmi-1-GFP transgenic mice to study the effect of Bmi-1 over-expression in neural stem/progenitor cells on CNS development and tumorigenesis. Bmi-1 over-expression in CNS stem/progenitor cells in vitro was sufficient to significantly increase stem cell self-renewal, overall proliferation, and neuronal differentiation (Fig. 3). In contrast, Bmi-1 over-expression in vivo had little effect on SVZ proliferation, adult olfactory bulb neurogenesis, or neurogenesis/gliogenesis during development (Fig. 4--6).6). Bmi-1 transgenic mice showed only a mild increase in neural stem cell frequency (<2-fold; Fig. 3A) with no increase in the total number of neural stem cells per mouse (Fig. 3B). They also exhibited no evidence of tumorigenesis, even when aged for more than 20 months (Fig. 8E). These data demonstrate that Bmi-1 over-expression has much more profound effects on CNS stem cell function in culture than in vivo.

The increased effect of Bmi-1 over-expression in culture correlated with its ability to reduce p16Ink4a and p19Arf expression in culture. Although p16Ink4a and p19Arf are not detectably expressed by neural stem/progenitor cells in developing or young adult mice (Bruggeman et al., 2007; Bruggeman et al., 2005; Molofsky et al., 2005; Molofsky et al., 2003; Nishino et al., 2008) (Fig. 2C, D), p16Ink4a and p19Arf are induced in culture in neural stem cells (Molofsky et al., 2003) in response to the stress associated with adaptation to the unphysiological culture environment (Lowe and Sherr, 2003; Sherr and DePinho, 2000). Since p16Ink4a and p19Arf were not detectably expressed by neural stem/progenitors cells in vivo, Bmi-1 over-expression had no effect on p16Ink4a and p19Arf expression in vivo. In contrast, Bmi-1 over-expression did reduce p16Ink4a and p19Arf expression in culture (Fig. 2B). This suggested that the increased effects of Bmi-1 over-expression in culture were caused by its ability to reduce p16Ink4a and p19Arf expression in culture. Consistent with this, Bmi-1 over-expression no longer enhanced self-renewal, proliferation, or neurogenesis in Ink4a-Arf deficient neural stem cells in culture (Fig. 7). These data demonstrate that caution must be exercised when inferring in vivo functions for Bmi-1 based on observations collected in culture because the increased expression of p16Ink4a and p19Arf in culture increases the reliance of cells upon Bmi-1 and makes cells respond to Bmi-1 over-expression in a way that is not observed in vivo.

In Bmi-1 deficient mice in vivo, neural stem cells require Bmi-1 postnatally, but not during fetal development, for self-renewal. Fetal Bmi-1 deficient mice do not exhibit detectable p16Ink4a and p19Arf expression in neural tissue in vivo and are born with normal numbers of neural stem cells (Molofsky et al., 2003). In contrast, postnatal Bmi-1 deficient mice exhibit elevated p16Ink4a and p19Arf expression in neural tissues in vivo and neural stem cells become increasingly depleted with time (Molofsky et al., 2003). This depletion of neural stem cells can be largely, but not completely, rescued by Ink4a and/or Arf deficiency (Bruggeman et al., 2005; Molofsky et al., 2005; Molofsky et al., 2003). These data suggest that the primary physiological role for Bmi-1 in the nervous system is to repress Ink4a-Arf in postnatal stem cells and to promote the maintenance of these cells throughout adulthood (Bruggeman et al., 2005; Molofsky et al., 2005; Molofsky et al., 2003). Nonetheless, a recent study found that shRNA knockdown of Bmi-1, primarily in cultured fetal neural stem cells, resulted in a self-renewal defect (Fasano et al., 2007). These authors speculated that acute knockdown of Bmi-1 revealed a physiological role for Bmi-1 in fetal neural stem cells that was not evident in germline knockout mice. One possibility is that unknown mechanisms in germline Bmi-1 knockout mice compensate for the loss of Bmi-1 during fetal development. Another possibility is that shRNA knockdown of Bmi-1 in cultured fetal neural stem cells reflected a function for Bmi-1 that is not required in vivo because these cells express p16Ink4a and p19Arf in culture in a way that does not occur in vivo. To distinguish between these possibilities it will be necessary to study Bmi-1floxed mice to determine whether conditional deletion of Bmi-1 in vivo leads to neurodevelopmental defects during fetal development.

Consistent with other recent studies (Bruggeman et al., 2007; Leung et al., 2004), we observed elevated levels of Bmi-1 expression in many brain tumors as compared to normal brain tissue, irrespective of whether the tumors were low or high grade (Suppl. fig. 1). Indeed, Bmi-1 is required for the growth of gliomas and for glioma stem cell function, through Ink4a-Arf dependent and independent mechanisms (Bruggeman et al., 2007). However, our data demonstrate that the degree of Bmi-1 over-expression that can be achieved in transgenic mice does not promote tumorigenesis in the CNS. Thus, increased Bmi-1 expression is not likely to be an initiating event in the formation of brain tumors. Rather, Bmi-1 may be required to cooperate with other mutations, such as to reduce the extent to which oncogenic stimuli induce p16Ink4a and p19Arf expression. In medulloblastomas, elevated Bmi-1 expression may reflect mutations that activate sonic hedgehog signaling (Leung et al., 2004). Elevated Bmi-1 expression in brain tumors may be the consequence of other mutations, or may confer a selective advantage in the context of mutations, but is not likely to reflect a primary role for increased Bmi-1 expression in the initiation of tumorigenesis.

Supplementary Material

Supplementary figure 1

Bmi-1 expression was elevated in human brain tumors:

A-F) Immunohistochemistry for Bmi-1 in representative sections from normal human brain tissue and various types of brain tumors. Bmi-1 (brown, arrowheads) was not detected in uninjured normal human brain tissue (A), but was detected in white matter after an epileptic seizure (B), in gliosis (C), pilocytic astrocytoma (D), oligodendroglioma (E) and glioblastoma (F). Sections are counterstained with hematoxylin. G) Western blotting showed Bmi-1 was faintly detected in normal human brain and cerebral cortex (brain, cortex) but it was elevated to various degrees in human glioma specimens obtained directly from patients (AA1, GBM3) as well as from mouse xenografts (PA1, GBM1,GBM2). In comparison, telencephalon cells from Bmi-1 transgenic mice showed the highest level of Bmi-1 expression. AA1: anaplastic astrocytoma (grade 3); PA1: pilocytic astrocytoma (grade 1); GBM1, GBM2, GBM3: glioblastoma multiforme (grade 4). H) Comparison of Bmi-1 expression between normal human brain tissues and various brain tumors. A three-tier grading system was used to evaluate Bmi-1 expression levels: strong Bmi-1 expression in many cells (red, ++); variable Bmi-1 expression in some cells (orange, +), no Bmi-1 expression above background (blue, −). Numbers in parentheses indicate the number of specimens examined for each category.

Supplementary figure 2

Southern blot analysis of transgene copy number and insertion site in transgenic founders:

A). Southern blots of genomic DNA from the ten transgenic founders showed variable transgene copy numbers. ScaI linearized transgene was used for copy number standards. Five lines (B*, C, D, E, G) gave germ line transmission. B). Genomic DNA from lines B, K, and C were digested with ScaI or EcoRV (which each cut at single sites within the transgene) and analyzed by Southern blot using an internal probe against the transgene. Line B and Line K mice were both derived from the original line B* founder; however, they had different transgene copy numbers and different transgene integration sites. Note that the presence of only 1 or 2 bands indicates a single chromosome integration site. Line C samples were included for transgene copy number comparison. All experiments were performed on line B and C mice, which had approximately 5 copies of the transgene inserted at a single integration site.

Supplementary figure 3

Characterization of transgene expression in lines that gave germline transmission:

A). GFP expression in transgenic mice was examined at P0 and P30 by analyzing dissociated cells by flow-cytometry. Only two transgenic lines with low copy numbers of the transgene (lines B and C) showed GFP expression in the CNS at both time points examined, while the other lines showed no appreciable expression at either developmental stage. GFP expression was not detected in tissues outside of the nervous system such as in bone marrow hematopoietic cells. B). Representative flow-cytometry histograms show the absence of GFP expression in transgenic mouse bone marrow cells (black open histograms show littermate control cells; green solid histograms show transgenic cells). Line B adult SVZ cells were included as a positive control for GFP expression.

Supplementary figure 4

Bmi-1 transgenic mice were born at expected Mendelian frequency and showed no substantial growth defects:

A) Bmi-1 transgenic mice from both lines were born at Mendelian frequency and showed no increased mortality relative to control littermates within the first month after birth. Numbers of surviving wild type and transgenic mice were counted at P30. B, C) Bmi-1 transgenic mice from both lines exhibited similar brain and body masses as control littermates at P60, though a subset (~25%) of transgenic mice that developed hydrocephalus in early adulthood tended to exhibit increased brain mass and decreased body mass. Since line B mice tended to develop hydrocephalus earlier than line C mice, these effects led to line B mice exhibiting slightly but significantly lower body masses and slightly but significantly higher brain masses than littermate controls (n=3 independent experiments, with 6 mice per genotype; *, p<0.05, error bars represent s.d.).

Supplementary figure 5

Most neurospheres grown under our culture conditions were clonal:

A 1:1 mixture of freshly isolated adult SVZ cells from UBC-GFP and ROSA-26 mice were plated at the indicated total cell density in non-adherent cultures, and neurospheres were allowed to form. After 10 days in culture spheres were scored for GFP and LacZ expression. While most neurospheres appeared to be clonally derived (containing GFP+ cells or LacZ+ cells but not both), a small fraction of neurospheres contained both GFP+ and LacZ+ cells suggestive of sphere-fusion (red bar) (Singec et al., 2006). Data represent mean±s.d. from 3 experiments.


This work was supported by the McDonnell Foundation and the Howard Hughes Medical Institute. Thanks to David Adams, Martin White, and Ann Marie Deslaurier of the University of Michigan (UM) Flow-Cytometry Core Facility. Flow-cytometry was supported in part by the UM Comprehensive Cancer Center NIH CA46592, and the UM Multipurpose Arthritis Center NIH AR20557. Thanks to Amanda Welton and the UM Center for Molecular Imaging for help with MRI analysis. Thanks to Galina Gavrilina and Thom Saunders of the UM Transgenic Animal Core Facility for generating Nestin-Bmi-1-HA-GFP transgenic mice with partial financial support from the UM Comprehensive Cancer Center NIH CA46592, the UM Center for Organogenesis, and the Michigan Economic Development Corporation (grant 085P1000815). Thanks to Daisuke Nakada for technical assistance with Southern and western blotting.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Almqvist PM, Mah R, Lendahl U, Jacobsson B, Hendson G. Immunohistochemical detection of nestin in pediatric brain tumors. J Histochem Cytochem. 2002;50:147–58. [PubMed]
  • Bea S, Tort F, Pinyol M, Puig X, Hernandez L, Hernandez S, Fernandez PL, van Lohuizen M, Colomer D, Campo E. BMI-1 gene amplification and overexpression in hematological malignancies occur mainly in mantle cell lymphomas. Cancer Res. 2001;61:2409–12. [PubMed]
  • Ben-Saadon R, Zaaroor D, Ziv T, Ciechanover A. The polycomb protein Ring1B generates self atypical mixed ubiquitin chains required for its in vitro histone H2A ligase activity. Mol Cell. 2006;24:701–11. [PubMed]
  • Bracken AP, Kleine-Kohlbrecher D, Dietrich N, Pasini D, Gargiulo G, Beekman C, Theilgaard-Monch K, Minucci S, Porse BT, Marine JC, Hansen KH, Helin K. The Polycomb group proteins bind throughout the INK4A-ARF locus and are disassociated in senescent cells. Genes Dev. 2007;21:525–30. [PubMed]
  • Bruggeman SW, Hulsman D, Tanger E, Buckle T, Blom M, Zevenhoven J, van Tellingen O, van Lohuizen M. Bmi1 controls tumor development in an Ink4a/Arf-independent manner in a mouse model for glioma. Cancer Cell. 2007;12:328–41. [PubMed]
  • Bruggeman SW, Valk-Lingbeek ME, van der Stoop PP, Jacobs JJ, Kieboom K, Tanger E, Hulsman D, Leung C, Arsenijevic Y, Marino S, van Lohuizen M. Ink4a and Arf differentially affect cell proliferation and neural stem cell self-renewal in Bmi1-deficient mice. Genes Dev. 2005;19:1438–43. [PubMed]
  • Capela A, Temple S. LeX/ssea-1 is expressed by adult mouse CNS stem cells, identifying them as nonependymal. Neuron. 2002;35:865–875. [PubMed]
  • Chagraoui J, Niessen SL, Lessard J, Girard S, Coulombe P, Sauvageau M, Meloche S, Sauvageau G. E4F1: a novel candidate factor for mediating BMI1 function in primitive hematopoietic cells. Genes Dev. 2006;20:2110–20. [PubMed]
  • Dahlstrand J, Collins VP, Lendahl U. Expression of the class VI intermediate filament nestin in human central nervous system tumors. Cancer Res. 1992a;52:5334–41. [PubMed]
  • Dahlstrand J, Zimmerman LB, McKay RD, Lendahl U. Characterization of the human nestin gene reveals a close evolutionary relationship to neurofilaments. J Cell Sci. 1992b;103(Pt 2):589–97. [PubMed]
  • Fasano CA, Dimos JT, Ivanova NB, Lowry N, Lemischka IR, Temple S. shRNA knockdown of Bmi-1 reveals a critical role for p21-Rb pathway in NSC self-renewal during development. Cell Stem Cell. 2007;1:87–99. [PubMed]
  • Ferland RJ, Cherry TJ, Preware PO, Morrisey EE, Walsh CA. Characterization of Foxp2 and Foxp1 mRNA and protein in the developing and mature brain. J Comp Neurol. 2003;460:266–79. [PubMed]
  • Hatten ME. Central nervous system neuronal migration. Annu Rev Neurosci. 1999;22:511–39. [PubMed]
  • Haupt Y, Bath ML, Harris AW, Adams JM. BMI-1 transgene induces lymphomas and collaborates with Myc in tumorigenesis. Oncogene. 1993;8:3161–3164. [PubMed]
  • Hernandez-Munoz I, Lund AH, van der Stoop P, Boutsma E, Muijrers I, Verhoeven E, Nusinow DA, Panning B, Marahrens Y, van Lohuizen M. Stable X chromosome inactivation involves the PRC1 Polycomb complex and requires histone MACROH2A1 and the CULLIN3/SPOP ubiquitin E3 ligase. Proc Natl Acad Sci U S A. 2005;102:7635–40. [PubMed]
  • Hogan B, Beddington R, Costantini F, Lacy E. Production of transgenic mice. Manipulating the Mouse Embryo: A Laboratory Manual. Cold Spring Harbor Press; Plainview, New York: 1994. pp. 218–252.
  • Iwama A, Oguro H, Negishi M, Kato Y, Morita Y, Tsukui H, Ema H, Kamijo T, Katoh-Fukui Y, Koseki H, van Lohuizen M, Nakauchi H. Enhanced self-renewal of hematopoietic stem cells mediated by the polycomb gene product Bmi-1. Immunity. 2004;21:843–51. [PubMed]
  • Jacobs JJL, Scheijen B, Voncken J-W, Kieboom K, Berns A, von Lohuizen M. Bmi-1 collaborates with c-Myc in tumorigenesis by inhibiting c-Myc-induced apoptosis via INK4a/ARF. Genes and Development. 1999;13:2678–2690. [PubMed]
  • Kawaguchi A, Miyata T, Sawamoto K, Takashita N, Murayama A, Akamatsu W, Ogawa M, Okabe M, Tano Y, Goldman SA, Okano H. Nestin-EGFP transgenic mice: visualization of the self-renewal and multipotency of CNS stem cells. Molecular and Cellular Neuroscience. 2001;17:259–273. [PubMed]
  • Kiel MJ, He S, Ashkenazi R, Gentry SN, Teta M, Kushner JA, Jackson TL, Morrison SJ. Haematopoietic stem cells do not asymmetrically segregate chromosomes or retain BrdU. Nature. 2007;449:238–42. [PMC free article] [PubMed]
  • Lee H, Song JJ, Kim E, Yun CO, Choi J, Lee B, Kim J, Chang JW, Kim JH. Efficient gene transfer of VSV-G pseudotyped retroviral vector to human brain tumor. Gene Ther. 2001;8:268–73. [PubMed]
  • Lessard J, Sauvageau G. Bmi-1 determines the proliferative capacity of normal and leukaemic stem cells. Nature. 2003;423:255–60. [PubMed]
  • Leung C, Lingbeek M, Shakhova O, Liu J, Tanger E, Saremaslani P, Van Lohuizen M, Marino S. Bmi1 is essential for cerebellar development and is overexpressed in human medulloblastomas. Nature. 2004;428:337–41. [PubMed]
  • Lothian C, Lendahl U. An evolutionarily conserved region in the second intron of the human nestin gene directs gene expression to CNS progenitor cells and to early neural crest cells. Eur J Neurosci. 1997;9:452–62. [PubMed]
  • Lowe SW, Sherr CJ. Tumor suppression by Ink4a-Arf: progress and puzzles. Current Opinion in Genetics & Development. 2003;13:77–83. [PubMed]
  • Molofsky AV, He S, Bydon M, Morrison SJ, Pardal R. Bmi-1 promotes neural stem cell self-renewal and neural development but not mouse growth and survival by repressing the p16Ink4a and p19Arf senescence pathways. Genes Dev. 2005;19:1432–7. [PubMed]
  • Molofsky AV, Pardal R, Iwashita T, Park IK, Clarke MF, Morrison SJ. Bmi-1 dependence distinguishes neural stem cell self-renewal from progenitor proliferation. Nature. 2003;425:962–7. [PMC free article] [PubMed]
  • Nieto M, Monuki ES, Tang H, Imitola J, Haubst N, Khoury SJ, Cunningham J, Gotz M, Walsh CA. Expression of Cux-1 and Cux-2 in the subventricular zone and upper layers II-IV of the cerebral cortex. J Comp Neurol. 2004;479:168–80. [PubMed]
  • Nishino J, Kim I, Chada K, Morrison SJ. Hmga2 promotes neural stem cell self-renewal in young but not old mice by reducing p16Ink4a and p19Arf Expression. Cell. 2008;135:227–39. [PMC free article] [PubMed]
  • Park IK, Qian D, Kiel M, Becker MW, Pihalja M, Weissman IL, Morrison SJ, Clarke MF. Bmi-1 is required for maintenance of adult self-renewing haematopoietic stem cells. Nature. 2003;423:302–5. [PubMed]
  • Pear WS, Nolan GP, Scott ML, Baltimore D. Production of high-titer helper-free retroviruses by transient transfection. Proc. Natl. Acad. Sci. U.S.A. 1993;90:8392–8396. [PubMed]
  • Polleux F, Dehay C, Kennedy H. The timetable of laminar neurogenesis contributes to the specification of cortical areas in mouse isocortex. J Comp Neurol. 1997;385:95–116. [PubMed]
  • Rutka JT, Ivanchuk S, Mondal S, Taylor M, Sakai K, Dirks P, Jun P, Jung S, Becker LE, Ackerley C. Co-expression of nestin and vimentin intermediate filaments in invasive human astrocytoma cells. Int J Dev Neurosci. 1999;17:503–15. [PubMed]
  • Sherr CJ, DePinho RA. Cellular senescence: mitotic clock or culture shock? Cell. 2000;102:407–10. [PubMed]
  • Singec I, Knoth R, Meyer RP, Maciaczyk J, Volk B, Nikkhah G, Frotscher M, Snyder EY. Defining the actual sensitivity and specificity of the neurosphere assay in stem cell biology. Nat Methods. 2006;3:801–6. [PubMed]
  • Song LB, Zeng MS, Liao WT, Zhang L, Mo HY, Liu WL, Shao JY, Wu QL, Li MZ, Xia YF, Fu LW, Huang WL, Dimri GP, Band V, Zeng YX. Bmi-1 is a novel molecular marker of nasopharyngeal carcinoma progression and immortalizes primary human nasopharyngeal epithelial cells. Cancer Res. 2006;66:6225–32. [PubMed]
  • Soriano E, Del Rio JA. The cells of cajal-retzius: still a mystery one century after. Neuron. 2005;46:389–94. [PubMed]
  • Stemple DL, Anderson DJ. Isolation of a stem cell for neurons and glia from the mammalian neural crest. Cell. 1992;71:973–85. [PubMed]
  • Tateishi K, Ohta M, Kanai F, Guleng B, Tanaka Y, Asaoka Y, Tada M, Seto M, Jazag A, Lianjie L, Okamoto M, Isayama H, Yoshida H, Kawabe T, Omata M. Dysregulated expression of stem cell factor Bmi1 in precancerous lesions of the gastrointestinal tract. Clin Cancer Res. 2006;12:6960–6. [PubMed]
  • Valk-Lingbeek ME, Bruggeman SW, van Lohuizen M. Stem cells and cancer; the polycomb connection. Cell. 2004;118:409–18. [PubMed]
  • van der Lugt NM, Domen J, Linders K, van Roon M, Robanus-Maandag E, te Riele H, van der Valk M, Deschamps J, Sofroniew M, van Lohuizen M, et al. Posterior transformation, neurological abnormalities, and severe hematopoietic defects in mice with a targeted deletion of the bmi-1 proto-oncogene. Genes Dev. 1994;8:757–69. [PubMed]
  • van Lohuizen M, Verbeek S, Scheijen B, Wientjens E, van der Gulden H, Berns A. Identification of cooperating oncogenes in E mu-myc transgenic mice by provirus tagging. Cell. 1991;65:737–52. [PubMed]
  • Van Parijs L, Refaeli Y, Lord JD, Nelson BH, Abbas AK, Baltimore D. Uncoupling IL-2 signals that regulate T cell proliferation, survival, and Fas-mediated activation-induced cell death. Immunity. 1999;11:281–8. [PubMed]
  • Vonlanthen S, Heighway J, Altermatt HJ, Gugger M, Kappeler A, Borner MM, van Lohuizen M, Betticher DC. The bmi-1 oncoprotein is differentially expressed in non-small cell lung cancer and correlates with INK4A-ARF locus expression. Br J Cancer. 2001;84:1372–6. [PMC free article] [PubMed]
  • Wang H, Pan K, Zhang HK, Weng DS, Zhou J, Li JJ, Huang W, Song HF, Chen MS, Xia JC. Increased polycomb-group oncogene Bmi-1 expression correlates with poor prognosis in hepatocellular carcinoma. J Cancer Res Clin Oncol. 2007 [PubMed]
  • Zencak D, Lingbeek M, Kostic C, Tekaya M, Tanger E, Hornfeld D, Jaquet M, Munier FL, Schorderet DF, van Lohuizen M, Arsenijevic Y. Bmi1 loss produces an increase in astroglial cells and a decrease in neural stem cell population and proliferation. J Neurosci. 2005;25:5774–83. [PubMed]