Author contributions: S.D.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript; J.M.: collection and/or assembly of data, data analysis and interpretation; F.H.: data analysis and interpretation, manuscript writing; P.K.D.: administrative support, data analysis and interpretation, manuscript writing; R.G.R. and T.M.: conception and design, financial support, administrative support, data analysis and interpretation, manuscript writing, final approval of manuscript. R.G.R. and T.M. contributed equally to this work.
First published online in STEM CELLSExpress March 5, 2009
Development of the mammalian brain relies on the coordinated expansion of neural cells in a relatively short time, spanning for a period of only a few days in mice. The molecular networks regulating neural cell birth and expansion, termed neurogenesis, are still unresolved, although many studies using genetically modified mice have revealed a growing number of genes that are involved in regulating these processes. The cAMP response element binding protein (CREB) lies at the hub of a diverse array of intracellular signaling pathways and is a major transcriptional regulator of numerous functions in adult neural cells, including learning and memory and neuronal survival. Recent studies have shown that activated CREB is highly expressed in immature dividing cells in adult mouse and zebrafish brains and that CREB regulates neural stem/progenitor cells (NSPCs) proliferation in embryonic zebrafish brain. Using genetically modified mice, we show that deletion of CREB, without the concomitant loss of the related compensating factor cAMP response element modifier, leads to defects in neural progenitor cell expansion and survival. Cultured primary CREB−/− NSPCs exhibited decreased expression of several target genes important for neuronal survival and growth, including brain-derived neurotrophic factor and neural growth factor and showed that the survival and growth defect can be rescued by the addition of wild-type NSPC-conditioned medium. This is the first study showing a specific role for CREB in mammalian embryonic neurogenesis. This role appears to be mediated via the expression of factors important for NSPC survival and growth and suggests that CREB is an important signaling regulator within the developing neurogenic niche. STEM CELLS 2009;27:1347–1357
In mouse, two major germinal waves of neural proliferation have been identified. The first of these arises directly from cells of the neural plate and forms the ventricular zone (VZ) and subsequently the layers of the cortex [1–3]. A second germinal wave forms the subventricular zone (SVZ), comprising the lateral ganglionic, medial ganglionic, and caudal ganglionic eminences (LGE, MGE, and CGE, respectively) as well as the fetal neocortical SVZ. These regions contain neural stem/progenitor cells (NSPCs), a small proportion of which persists as an active pool of stem and progenitor cells into adulthood [2, 4]. Therefore, these cells are the very earliest precursors of adult SVZ-NSPCs.
Among the many factors contributing to NSPC proliferation, survival, and differentiation, transcription factors orchestrate the first dynamic steps of biological information flow. Through constant temporal control of the expression of the cell's hard-wired genetic information, these cells can respond to signals and exhibit their specialized stem and progenitor cell characteristics.
CREB is a transcription factor with diverse roles in neural cell function and is regulated through the action of diverse intracellular signaling cascades. One of the major mechanisms by which CREB is activated is by phosphorylation at a key conserved serine residue, Ser133, allowing interaction with the transcriptional coactivators, CREB-binding protein and p300 . Recent findings show that CREB is constitutively activated in dividing immature neural cells which are present in neurogenic regions of both embryonic and adult vertebrate brains [6, 7]. This is in contrast to brain regions outside the neurogenic zones, where CREB is only transiently activated under very specific conditions, treatments, or stimuli. CREB is an important factor regulating neural plasticity, a phenomenon not only critical for higher order brain functions such as learning and memory but also postulated to be intimately associated with the neurogenic process .
Our understanding of the role and mechanisms by which CREB modulates NSPC function is still at an early stage. Evidence that CREB influences neurogenesis comes from studies in adult mouse showing that constitutively activated phospho-CREB (pCREB) colocalizes with the cell adhesion marker Poly-Sialated Neural Cell Adhesion Molecule (PSA-NCAM) and labeled bromodeoxyuridine neurons in neurogenic cells, and that drugs which activate CREB also increase neurogenic activity [7, 8, 12]. To our knowledge, there are only two other studies specifically investigating CREB and the proliferative stages of early neurogenesis. The first presents only indirect evidence where PI3K/AKT stimulation activates both CREB and neurogenesis . The second study uses conditional CREB mutant mice to show a disturbance in the differentiation and survival of immature migrating olfactory neurons . Using zebrafish, our recent studies have made some progress toward deciphering the role of CREB in NSPCs where we reported that neurogenesis was regulated by CREB activity . This was done by injecting either a constitutively active mutant CREB or dominant negative mutant CREB (dnCREB) mRNA into fertilized zebrafish embryos. Overexpression of activated CREB led to increased neural cell proliferation, while overexpression of dnCREB led to decreased neural cell proliferation.
Interestingly, and with the emerging view of the existence of cancer-initiating “cancer stem cells,” there is also an accumulating body of evidence showing that CREB possesses oncogenic properties. Recent studies show that CREB overexpression or activation is a feature of human myeloid , pancreatic , and lung tumors , and that transgenic overexpression of constitutively active mutant CREB in each respective cell type promotes tumor development and growth. A further study also implicates CREB in liver cancer in a transgenic mouse model . CREB overexpression has also been implicated in other hyperproliferative neurological disorders, including hemimegalencephaly .
Mice with a deletion of the Creb1 gene, resulting in complete loss of CREB protein, die as soon as they are born, due to respiratory distress . Examination of the brains of these CREB deficient mice showed a severe reduction in corpus callosum thickness and anterior cortical thickness, accompanied by increased lateral ventricle space. Conditional brain-specific CREB mutant mice survived to adulthood and showed deficits in hypothalamic and associated pituitary development . Because of functional compensation of CREB by the related transcription factor cAMP response element modifier (CREM), a compound brain-specific CREB–CREM mouse was also generated, resulting in severely reduced neuronal survival during late development and postnatally depending on the time when CREB was lost .
In this study, we investigated the role of CREB in embryonic mouse brain development by using a CREB-deficient mouse line to examine early neurogenesis, at day14.5 postcoitum (pc), a phase when neurogenic activity reaches its peak . We show that CREB loss leads to impairment of neural expansion and brain development. By growing CREB-deficient NSPCs in vitro, we show that CREB is required for efficient cell-autonomous NSPC expansion and that the expression of a number of factors critical for neural growth, including neural growth factor (NGF) and brain-derived neurotrophic factor (BDNF) are significantly reduced. Taken together, these data provide evidence supporting a role for CREB in mammalian neurogenesis during early development.
Materials and Methods
Generation of CREB−/− Mice
Creb1loxP/loxP mice on a C57Bl/6 genetic background (backcrossed >10 generations) were generated by gene-targeting as described previously  and used to generate Creb1+/− mice, where the mutated allele contained a Cre-mediated deletion of exons 10 of the mouse Creb1 gene. Timed matings of Creb1+/− were performed and fetal mice were collected by caesarian-section, sacrificed at E14.5 pc. All animal experimentation was approved by the Animal Ethics Committee at the Peter MacCallum Cancer Centre.
Embryos were fixed in cold 4% paraformaldehyde for 16 hours prior to embedding in paraffin. Paraffin sections were sectioned on a microtome at a thickness of 7 μm. pCREB (Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com) and Proliferating Cell Nuclear Antigen (PCNA) antibodies were used at dilutions of between 1:1,000 and 1:2,000 and detected using the Vectastain ABC kit (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). For immunofluorescence staining, anti-nestin 1:100 (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com), anti-β3-tubulin 1:200 (Millipore, Billerica, MA, http://www.millipore.com), anti-glial fibrillary acidic protein (GFAP) 1:200 (Abcam, Cambridge, U.K., http://www.abcam.com), and Alexa Fluor-488 and -633 secondary antibodies 1:500 (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) were used. 4′,6-diamidino-2-phenylindole (DAPI; Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com) was used at 10 μg/ml for 5 minutes.
Isolation and Culture of NSPCs
Mouse embryos were analyzed at day 14.5 (E14.5) following detection of vaginal plugs. Following dissection of the brain, the external cortex was removed to allow isolation of the ganglionic eminences. These were placed into 5 ml fresh Hank's balanced saline solution (HBSS) in a 15 ml Falcon tube and disaggregated by gentle trituration. Following centrifugation at 1,200g for 5 minutes, the supernatant was removed and the NSPCs were resuspended in 5 ml neurosphere basal medium (1:1 Dulbecco's modified Eagle's medium [DMEM] and F12, 4 μg/ml heparin, 100 μg/ml penicillin/streptomycin; all Invitrogen) supplemented with 10 ng/ml basic Fibroblast Growth Factor (bFGF; Chemicon, Temecula, CA, http://www.chemicon.com), 20 ng/ml epidermal growth factor (EGF; Chemicon), and 1:50 B27 without vitamin A (retinoic acid; media supplemented with growth factors will be referred to as “NSBM”). The cells were transferred to a 25 cm2 filter-capped flask (Thermo-Fisher Scientific, Waltham, MA, http://www.thermo.com), where they formed clonal aggregates (neurospheres). Media was changed every 2 days and replaced with fresh NSBM, containing 1:50 B27 supplements, 20 ng/ml bFGF, and 40 ng/ml EGF. Neurospheres were passaged by centrifugation at 700g for 7 minutes, resuspension in 500 μl dissociation solution (50 ml (HBSS), 0.01 g EDTA, 0.0125 g trypsin [0.25 mg/ml], 0.119 g HEPES, pH 8) for 3 minutes at room temperature (RT), followed by neutralization in 500 μl neutralization solution (50 ml HEPES-buffered DMEM, 0.007 g soybean trypsin inhibitor [0.14 mg/ml; Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com]) for a further 3 minutes at RT. Dissociated cells were then washed with 9 ml DMEM/F12, and centrifuged at 1,400g for 5 minutes. The cells were resuspended in NSBM and plated into fresh 25 cm2 flasks at a concentration of 1–5 × 105 cells per ml. All NSPC culture took place at 37°C and 5% CO2. Neurospheres were cultured at “high” (>1 × 106 cells per ml), “medium” (1 × 105–3 × 105 cells per ml), “low” (2.5 × 104 cells per ml), or “single-cell” densities (1–3 cells per Terasaki well), as indicated in the text. For supplemented neurosphere medium, pituitary adenylate cyclase-activating polypeptide 27 (PACAP-27) (100 nm), BDNF (40 ng/ml) or NGF (20 ng/ml) were added individually or all the three combined (all three factors were from Sigma-Aldrich).
Whole neurospheres (20–30 spheres/well) or single cells (1 × 105 cells/well) were plated in Falcon 8-chamber slides (Krackeler-Scientific, Albany, NY), coated with 0.5 mg/ml poly-dl-ornithine in 0.15 M boric acid (Sigma) and 20 μg/ml laminin (Sigma). Neurospheres were plated in 500 μl NSBM either supplemented with 1% Fetal Calf Serum (FCS; JRH Biosciences, Lenexa, KS) or no serum or growth factors for the 3-day experiment. For the 7-day experiment, every 2 days the media was removed and replaced with fresh NSBM + 1% FCS.
See supporting information for more detailed protocols.
pCREB Is Expressed Throughout the Embryonic brain, Including in Actively Proliferating Cells
To determine the extent of CREB activity in neurogenic areas of the developing mouse brain, coronal sections of E14.5 wild-type mouse brain were analyzed for expression of pCREB and the proliferation marker PCNA. Within the midbrain, abundant pCREB (Fig. 1A, 1C) and PCNA (Fig. 1B, 1D) expression was seen in the ventricular neurogenic zone. Conversely, regions containing few proliferating PCNA positive cells (Fig. 1D) exhibited only scattered pCREB expression (Fig. 1C), indicating an association between the expression profiles of PCNA and pCREB in NSPCs within the ventricular neurogenic zone. Within the developing neocortex, the region where progenitor cells migrate toward the outer cortical layers from the VZs and SVZs, showed almost no pCREB+ cells, while pCREB expression in both the lower layers of the developing neocortex and the outermost cortical layer within the cortical plate (CP) was strong and abundant (Fig. 1C). As formation of the most superficial cortical layers involves the migration of progenitors from the VZ past previously formed neuronal cortical layers, which lie closest to the VZ, this outermost layer represents the most recently migrated progenitors. These data suggest that although pCREB is expressed in both proliferating regions and immature differentiating progenitors, it is not expressed in differentiated cells of the deepest cortical layers.
Within the neurogenic region surrounding the third ventricle there were three subpopulations of cells with differing pCREB and PCNA expression profiles. PCNA-positive-only cells were present around the dorsal and lateral regions of the third ventricle, with cells at the most ventral extremities showing no PCNA expression (Fig. 1H); this PCNA-negative cell population exhibited pCREB expression (Fig. 1G). A third subpopulation of cells in the dorsal region of the third ventricle coexpressed pCREB and PCNA, demarcating a double-positive immature progenitor population.
pCREB Is Expressed in the Developing Retina
Because of the defect in CREB−/− eye development and the presence of neural progenitor cells in the retina, pCREB and PCNA expression was analyzed. Both pCREB and PCNA were expressed in the primitive ciliary marginal zone (CMZ; Fig. 1E, 1F), a region harboring both stem and proliferating progenitor cells . Both pCREB and PCNA were expressed within the outer layer of the developing lens (Fig. 1E, 1F). Within the retina proper, PCNA+ and pCREB+ cells were clearly localized to discrete, mutually exclusive regions. PCNA+ cells were restricted to the outermost nuclear layers of the retina (Fig. 1F), whereas pCREB expression was restricted to cells of the inner nuclear layer (but not the developing ganglionic cell layer; Fig. 1E), indicating that although the most primitive proliferating cells express both PCNA and pCREB, pCREB expression is reduced as cells differentiate and migrate.
Generation of CREB−/−Mice
To examine the role of CREB in mammalian embryonic brain development and neural cell proliferation, a mouse with germline deletion of CREB was employed. For this study mice were generated by germline deletion in CREBlox/lox mice, where exon 10 of the Creb1 gene is deleted and due to a frame-shift between exons 9 and 11 results in the loss of the C-terminal dimerization and DNA binding domains and no stable truncated CREB protein expression . Some germline mutants were identified during the breeding of CREBlox/lox × NestinCre mice, where recombination of the “floxed” Creb1 gene was mediated by the expression of Cre-recombinase under the control of a nestin promoter and intron one enhancer. The CREB−/− mice in this study harbor no exogenous gene, only the presence of a single loxP site at the point of deletion, while the CREB−/− mice of Rudolph et al.  contain a neomycin resistance gene replacement of Creb1 exons 10 and 11. Immunohistochemical analysis using anti-CREB antibodies (data not shown) and reverse transcription polymerase chain reaction (RT-PCR) analysis of NSPCs (Fig. 2A) confirmed the loss of CREB.
CREB-Deficient Embryos Are Smaller and Exhibit Defects in Brain and Retina
To assess the effects of CREB loss on early development, CREB+/− mice were plug-mated and embryos harvested at E14.5, a stage when neurogenesis peaks in the developing mouse forebrain . A total of 105 embryos were generated throughout the course of the study, from 19 separate females, with a median of five embryos per litter. Of the embryos analyzed at E14.5, 26 (25%) were wild-type, with respect to CREB expression (CREB+/+), 63 (60%) were heterozygous (CREB+/−), and 16 (15%) were homozygous null for CREB (CREB−/−). Chi-square test analysis (Fig. 2B) showed a significant decrease (p < .05) in the number of CREB−/− embryos (16 actual, 26 expected) to the expected “Mendelian” ratios. Furthermore, surviving CREB−/− embryos were approximately 65% of the weight of both CREB+/− and CREB+/+ embryos (Fig. 2C, 2D), due to reduced body size and mass.
Of the 16 CREB−/− embryos generated, four showed severe neural patterning defects, including hemisphere fusion (holoprosencephaly) and severe reduction of olfactory bulb size (Fig. 2E). Most (12/16) CREB−/− embryos also had severe eye defects, indicated by retinal malformation (Fig. 2C and 3C, 3D). Gross examination of CREB−/− embryos showed no defects in formation of tail, limbs, toes, liver, and external head and facial structures, indicating that the patterning of these tissues was not affected.
Examination of the brains of CREB−/− mice highlighted several neuroanatomical defects. CREB−/− embryos displayed reduced neuroepithelium thickness in both the diencephalic (65% ± 12% of wild-type, n = 3 embryos) and telencephalic ventricular walls (67% ± 18% of wild-type, n = 3 embryos; Fig. 3A, 3B; center panels). At the level of the mid and anterior hindbrain (Fig. 3A, 3B; bottom panels), the size and location of both the posterior horns of the lateral ventricles and the size of the third ventricle did not appear disrupted in CREB−/− embryos, although the fourth ventricle was enlarged to more than double the size compared to wild-type (2.6 ± 1 times larger when compared with wild-type, ranging from 1.6 to 3.8 in three different embryos). Coronal sections through the forebrain of the most severely affected CREB−/− mice (Fig. 3A, 3B; top panels) confirmed hemisphere fusion due to the absence of the intraventricular foramen and enlarged lateral ventricles. All other brain structures examined appeared normal.
Immunofluorescence staining for PCNA was employed to examine the number of cycling cells in the neurogenic zones of CREB+/+ and CREB−/− embryos. CREB−/− brains showed a reduction of PCNA labeling in the VZs (Fig. 3C, 3D). As cycling PCNA+ cells populate the growing brain, reduced neurogenesis may explain the reduced neuroepithelium thickness observed in CREB−/− brains.
CREB Loss Leads to Abnormal Nestin Immunoreactive Cell Morphology
To assess the effects of CREB loss in the neurogenic zone, we examined one of the major subpopulations of neurogenic cells, namely the nestin immunoreactive cells in the SVZs. Although the number of nestin positive cells was not significantly different in CREB−/− embryo brains compared with wild-type controls, there was a difference in the morphology of the nestin+ cells in CREB deficient mice (Fig. 4). Although wild-type mouse brains exhibited nestin+ cells with highly branched and long dendrite-like nestin-immunoreactive projections (Fig. 4A–4C), CREB−/− embryo brains exhibited shorter nestin-immunoreactive projections and less complex branching (Fig. 4D–4F).
CREB Has a Role in Cultured Neural stem/Progenitor Cell Growth and Survival
To determine whether the increased ventricular space observed in some CREB−/− embryos was due to an increase in cellular apoptosis, Terminal deoxynucleotidyl Transferase Biotin-dUTP Nick End Labeling assays were performed on CREB+/+ and CREB−/− brains (Fig. 5A). Automated morphometric quantitation (MetaMorph, Molecular Devices) of coronal sections through anterior, mid, and posterior regions of the brains of both mutant and control groups showed no significant difference in the number of apoptotic cells (as a proportion of all cells in the section); (Fig. 5B). These data indicate that apoptosis does not significantly contribute to reduced brain size, lack of olfactory bulbs, fused ventricles (holoprosencephaly) and enlarged ventricles in CREB−/− embryo brains.
To determine whether NSPCs isolated from CREB−/− embryos showed niche-independent defects, primary NSPCs were isolated from the LGE, MGE, and CGE of CREB−/− and CREB+/+ E14.5 embryos and cultured in vitro. Significantly fewer (p < .05) cells were generated from NSPCs derived from CREB−/− mice between 2 and 4 weeks culture (Fig. 5D), although maximum neurosphere size did not vary between CREB−/− and CREB+/+ cultures (data not shown).
To investigate NSPC survival, apoptosis was examined by labeling with the apoptotic membrane marker dye Annexin V (Fig. 5C). The number of apoptotic cells (Annexin V+) was significantly higher in CREB−/− (8.3% ± 0.7%) when compared with CREB+/+ cells (4.9% ± 0.5%), indicating that increased cell death is a contributing factor to the reduced cumulative cell number observed in vitro in NSPCs from CREB−/− animals. Furthermore, when cells from CREB−/− and CREB+/+ mice were cultured at high density, CREB−/− neurospheres adhered to the base of the well and began to sprout processes (Fig. 5E). This was not seen in cells from CREB+/+ mice (Fig. 5E). As sprouting neurons are a feature of differentiated cells, this observation suggests that CREB has a role in maintaining immature cell status. Overall, the data shows that CREB has a role in NSPC survival and growth, although the survival defect is only seen when NSPCs are removed from their in vivo niche, suggesting that CREB loss is compensated for by other factors in the neurogenic niche in vivo.
CREB Does Not Have a Role in NSPC Cell Cycle Progression In Vitro
As a knockout mouse model of cyclin D2 exhibited neurogenic phenotypes similar to those observed within the present study, particularly a defect in olfactory bulb formation  and cyclin D2 is a known direct target of CREB , we analyzed the expression of cyclin D2 mRNA in CREB−/− neurospheres cultured for 24 hours at low density and saw no change in cyclin D2 expression (supporting information Fig. 2). To test the hypothesis that a cell-cycle defect may contribute to the reduced cellularity observed in culture of NSPCs from CREB−/− mice, propidium-iodide (PI) staining was conducted to determine the cell-cycle profiles of cultured cells from CREB+/+ and CREB−/− mice. The resulting data were then analyzed by an automated cell-cycle analysis program, Modfit, to eliminate operator bias in interpreting relative numbers of cells in G1, S, or G2-M phases of the cell cycle. No difference was seen in the proportions of cells in each stage of the cell cycle in cells derived from CREB+/+ and CREB−/− mice (supporting information Fig. 2).
NSPC Self-Renewal and Survival in the Absence of CREB
To assess the self-renewal potential of NSPCs without CREB, we used the neurosphere assay to measure the number of neurospheres formed over 5 days following single-cell seeding of primary (freshly dissected) cells in Terasaki plates. After 5 days, significantly fewer neurospheres formed when cells from CREB−/− embryos were plated compared with wild-type littermate-derived cells (supporting information Fig. 3C). This suggested that CREB may contribute to neural stem cell self-renewal. However, since decreased cell-survival within the assay period could also have led to the observed reduction in neurosphere formation, cell viability was assessed 24 hours after single-cell plating by visual inspection and counting of phase-bright (light refractive; viable) cells per well. CREB−/− cells showed a marked reduction in cell survival 24 hours after plating, compared with 2 hours and corresponded with the reduction in neurosphere formation after 5 days. Moreover, cells viable after 24 hours were able to divide and generate neurospheres, as depicted in supporting information Figure 3D. Thus, although NSPCs from CREB−/− cells were able to form neurospheres and be serially passaged, the culturing of NSPCs at medium to high densities allows for cell–cell contact, paracrine signaling, thereby potentially masking cell-intrinsic effects through generation of a surrogate “in vitro niche.” Moreover, neurosphere formation under these conditions is not necessarily clonal, as neurospheres can form by cell aggregation rather than clonal cell division. To examine potential defects in cells from CREB−/− mice independently of this “in vitro niche,” cells were cultured at low density in six-well plates. To confirm the identity of cells used for low-density plating experiments, neurospheres were routinely genotyped to ensure that potential cross-contamination of CREB−/− cells with NSPCs from CREB+/+ culture had not occurred (data not shown). When primary neurospheres were dissociated and equal numbers of CREB+/+ and CREB−/− cells were plated at low densities, the number of viable secondary spheres formed following 7 days culture was significantly reduced in cells lacking CREB (Fig. 6A, 6B). Furthermore, neurospheres formed at low density from cells lacking CREB were smaller, with fewer cells than those formed from CREB+/+ mice (Fig. 6A). The incidence of neurosphere-forming cells (NFCs) from CREB+/+ mice was 14.3 ± 1.9 NFCs per 1,000 cells; in CREB−/− mice this was reduced to only 2.3 ± 1.1 NFCs per 1,000 cells (Fig. 6B). These data indicate severe defects in neurosphere-forming capacity in cells lacking CREB when plated at low density.
To analyze whether decreased neurosphere-forming capability of CREB−/− cells at low density was also due to increased apoptosis, neurospheres at 7 days were dissociated, and assessed for incorporation of Annexin V dye by fluorescence-activated cell sorting (Fig. 6C). Although the numbers of apoptotic cells in culture from CREB+/+ mice at low density were increased when compared with medium-high density culture (18.4 ± 2.0 at low density, compared with 4.9% ± 0.5% at medium-high density), the difference in the number of apoptotic cells was far greater in cells from CREB−/− mice (41.5% ± 1.2% at low density, 8.3% ± 0.7% at medium-high density; p < .01). Consistent with this, quantitative RT-PCR (qRT-PCR) analysis of mRNA from CREB−/− cells cultured for 24 hours at low density, showed decreased expression of the prosurvival CREB target gene, Bcl-2 (Fig. 7D).
Survival and Neurosphere Forming Ability of CREB−/− NSPCs is Rescued Following Culture in CREB+/+ NSPC-Conditioned Medium
To examine the hypothesis that loss of soluble paracrine signaling factors in low density CREB−/− NSPC culture contributed to decreased neurosphere formation and cellular survival, CREB−/− NSPCs were cultured in conditioned medium harvested from wild-type (CREB+/+) NSPCs. CREB−/− NSPCs were grown at clonal density in neurosphere basal medium (NSBM) supplemented with 0%, 25%, or 50% conditioned medium (Fig. 7A). Significantly higher numbers of viable neurospheres were observed following 7 days culture in NSBM supplemented with either 25% or 50% wild-type conditioned medium compared with growth in NSBM alone, suggesting that many of the NSPCs which undergo apoptosis in the absence of conditioned medium are able to survive and form neurospheres (Fig. 7B). Interestingly, the proportion of “large” neurospheres (>50 μm diameter) was significantly decreased in cultures supplemented with wild-type conditioned medium (Fig. 7C), suggesting that either the surviving NSPCs had insufficient access to mitogens in the media, or perhaps surviving NSPCs had only limited proliferative potential. Taken together, these data indicate that in vitro paracrine signaling independent of the mitogens EGF and bFGF, is an important mechanism for NSPC survival in vitro. Of the NGFs which can influence survival and are known CREB target genes, we chose to examine the expression of BDNF [27, 28], NGF [29, 30], Bcl-2 , and PACAP . qRT-PCR showed that the expression of mRNA of BDNF, NGF, PACAP, and Bcl-2 was significantly downregulated in CREB−/− cells cultured for 24 hours at low-density (Fig. 7D). To further investigate the survival defects of CREB−/− NSPCs in neurosphere culture conditions, we cultured both embryonic CREB-deficient cells and wild-type cells in medium containing twice the standard bFGF and EGF concentration. To test whether the observed decrease in the CREB-dependent factors, BDNF, PACAP, and NGF were the cause of the survival defect, standard neurosphere medium supplemented with BDNF (40 ng/ml) or PACAP-27 (100 nM) or NGF (20 ng/ml) or all three factors combined. No difference was observed in neurosphere survival, growth, or morphology between CREB-deficient and wild-type neurospheres in any tested condition (data not shown).
CREM and ATF-1 Cannot Compensate for the Absence of CREB in NSPCs
In the absence of CREB, functional compensation by other closely related CREB family members, CREM and Activating Transcription Factor 1 (ATF-1), has been demonstrated previously [22, 33]. Moreover, CREM has been shown to specifically compensate for CREB loss in late embryonic (>E16.5) brain to support neuronal survival . Compensatory upregulation of ATF-1 and CREM was not seen in CREB−/− E14.5 brains (data not shown) and in NSPCs grown in vitro (Fig. 7D). Instead, in NSPCs, CREM mRNA (all isoforms) expression was severely reduced in the absence of CREB, contrary to what was expected. This observation suggests that the Crem gene may be differentially regulated by CREB in early embryonic cells compared with late embryonic and adult cells. This is consistent with previous data where CREM is not expressed in early (up to E3.5, blastocyst stage) wild-type and CREB−/− embryos  and there is no compensation by CREM in neuronal survival prior to E16.5 . We speculate that CREB may positively regulate Crem expression prior to E16.5 but negatively regulate expression after this time. Atf-1 expression was not influenced by the absence of CREB in NSPCs, consistent with previous data in CREB mutant pluripotent embryonic cells  and adult somatic cells .
CREB Does Not Influence Glial–Neuronal Fate
To determine whether loss of CREB, previously implicated as a crucial factor for NSPC differentiation, and to determine whether a specific progenitor lineage was preferentially undergoing apoptosis in NSPC culture from CREB−/− mice, CREB+/+ and CREB−/− neurospheres were induced to differentiate by treatment with 5% FCS for 3 or 7 days (supporting information Fig. 1). Differentiated cells were stained for markers of differentiated neurons (β3-tubulin) and astrocytes (GFAP). Nuclei were visualized by staining with DAPI. To quantitate the relative percentages of differentiated neural cell types generated, a minimum of six different fields were selected for counting per culture (supporting information Fig. 1B). The population of β3-tubulin/GFAP double-negative cells (containing both oligodendrocytes and undifferentiated neuronal progenitors) was also quantified. The percentage of each cell type formed from CREB−/− NSPCs did not vary significantly from CREB+/+ cells, indicating that in vitro, NSPCs, once formed, can differentiate normally in the absence of CREB. Furthermore, CREB loss did not cause a specific committed progenitor population to preferentially differentiate or undergo apoptosis. As for the 7-day differentiation protocol, culturing in differentiation-promoting conditions without serum or growth factors for 3 days did not reveal any effect of CREB loss.
Data presented within this study show that activated CREB, in the form of pCREB is constitutively expressed in neurogenic cells of the embryonic mouse brain. Germ-line deletion of CREB in the mouse leads to developmental defects during embryogenesis, specifically in patterning of the brain and retina at E14.5. Furthermore, there is reduced cellularity and reduction in PCNA+ cells in CREB−/− embryo brains without an observed increase in apoptosis in vivo. NSPCs derived from these mice display severe defects in survival and neurosphere-forming ability in vitro, particularly when cultured at low density. Analysis of several CREB target genes showed a specificity of expression in NSPCs lacking CREB. Taken together, these data indicate that loss of CREB results in cell-intrinsic NSPC defects.
The expression of pCREB in the ventricular and SVZs of the E14.5 embryonic brain was consistent with both the distribution of PCNA+ proliferating cells and the localization of Nestin-positive neural stem and progenitor cells . Within the retina, pCREB is coexpressed with PCNA within the CMZ—the region containing the most immature stem and progenitor cells. Within the more mature retinal layers, PCNA and pCREB are clearly expressed in mutually exclusive populations of cells. This suggests specific spatiotemporal CREB activation and function during neurogenesis and differentiation. Specifically, CREB is activated in immature stem/progenitor cells, deactivated in mature progenitors, and once again phosphorylated when these progenitors begin to differentiate and integrate.
Previous analysis of E18.5 embryos lacking CREB  showed a similar distribution of genotype numbers and reduced embryo size to that seen in the present study. Fifteen percent of embryos at E18.5 were CREB−/− and the embryos were 70% the size of wild-type littermates ; in the present study 15% of embryos were CREB−/− at E14.5, and the weight of these embryos was 65% that of wild-type littermates. Although the previous study reported relatively subtle abnormalities in the brain morphology of CREB−/− embryos at E18.5 with reduced corpus callosum and enlarged lateral ventricles, embryos analyzed in the present study exhibited more severe morphological defects, including holoprosencephaly, failure of olfactory bulb development, and retinal crenation. These differences may be attributed to the different genetic backgrounds of the two respective model systems used (129-C57Bl/6 mix in the previous study compared with relatively pure C57Bl/6 in this study), or may reflect the more severely affected embryos not previously analyzed prior to E17.5 .
Immunochemical characterization of CREB mutant embryo brains revealed that one of the major cell types associated with neurogenesis, the nestin immunoreactive cells, exhibited abnormal morphology compared with wild-type control brains (Fig. 4). In wild-type brains, nestin+ cells showed highly branched and long dendrite-like nestin-immunoreactive projections, whereas CREB-deficient brains exhibited shorter nestin-immunoreactive projections and less complex branching. At this stage, the reason for the observed differences is not clear. However, in a recent report, similar morphological changes in nestin-immunoreactive cells has been attributed to ageing in rats , possibly due to the metabolic senescence of these actively dividing cells and neurogenic niche decline. This explanation could also apply to the CREB-deficient nestin+ cells in mice, given that CREB regulates many genes important for cell function and survival.
To assess “stem cell” properties which may be dependent on CREB function, measurement of self-renewal potential using the neurosphere assay was undertaken. Initial results showed that CREB−/− NSPCs plated at single-cell densities were unable to form neurospheres as readily as wild-type cells after five days. Further investigation of the survival of plated cells showed that CREB deficient cells exhibited severely reduced survival rates at 24 hours, accounting for the reduction in neurosphere formation after 5 days. Moreover, equal numbers of mutant and wild-type cells surviving beyond 24 hours were able to divide and generate neurospheres, implying that although CREB does not directly affect self-renewal, it may indirectly contribute to self-renewal by altering the threshold of survival of newly generated neural stem cells, in line with the well-characterized role of CREB and CREM in neuronal survival during development and in adult brain . Furthermore, a role for CREB and the related factor ATF-1 in promoting stem cell survival has been previously reported in mouse embryonic stem (ES) cells . Taken together with our data, CREB likely has a general role in regulating important survival genes in stem cells.
To examine potential mechanisms important for mediating NSPC survival, the expression of candidate CREB-target genes important for neural growth and survival were analyzed in CREB+/+ and CREB−/− NSPCs. Following quantitative real-time PCR analysis, the mRNA expression of the survival (Bcl-2) and soluble growth factors (BDNF, NGF, PACAP) was significantly decreased, indicative of a dual role for CREB in promoting survival via cell-intrinsic production of survival factors and via autocrine or paracrine production of signaling factors within the “niche”. As the expression of three key regulators of the cell-cycle, cyclins A2, D1, and D2 was not significantly altered in CREB−/− NSPCs, and cell cycle analysis showed no alteration in the cell-cycle profiles of CREB−/− NSPCs compared with NSPCs from CREB+/+ embryos (supporting information Fig. 2), a cell-cycling defect is not indicated in CREB−/− NSPCs.
Of the CREB family members examined, CREM mRNA expression was downregulated in CREB−/− NSPCs, whereas the expression of Atf-1 was unchanged (Fig. 7D). Although this observation is at odds with data reported previously, where CREM is upregulated in neuronal cells lacking CREB , this may be due to unique transcriptional properties of immature proliferating cells compared to more mature cells. CREM has been shown to specifically compensate for CREB loss in late embryonic (>E16.5) brain to support neuronal survival . Taken together it appears that CREB positively regulates Crem gene expression prior to E16.5 but negatively regulates expression after this time. This is in line with previous data where CREM is not detectable in very early (<E3.5) wild-type and CREB−/− embryos , nor is there evidence of CREM neuronal survival compensation prior to E16.5 . As mentioned, Atf-1 expression does not appear to be influenced by the absence of CREB in NSPCs, consistent with previous data in adult CREB mouse mutant tissue .
Overall, the data presented in this study point to the importance of CREB in the regulation of target genes involved in NSPC growth and survival. This work extends previous knowledge of CREB function in neurons with the identification of novel defects in neural development, and NSPC survival and proliferation. Data presented in this study show for the first time that pCREB is expressed in PCNA+ proliferating cells in several neurogenic regions in mouse. Furthermore, this study shows that CREB contributes to NSPC survival ex vivo, possibly through transcriptional regulation of several factors. Although the limited candidate gene analysis approach taken here has identified several CREB-target genes regulated by CREB which are likely to be involved in specific pathways mediating cellular survival and proliferation, a larger-scale target gene screen is necessary to reveal the NSPC CREB transcriptome. Future work will focus on determining the complete spectrum of differentially regulated factors, both soluble and endogenous, in NSPCs and whether upregulation of CREB in NSPCs results in dysregulated growth.
We thank members of the Peter MacCallum Cancer Centre Microscopy facility for technical assistance and Paul Neeson, Leigh Ellis, Ralph Rossi, Warren Raye, Sab Ventura, and Paul White for their assistance and advice. This work was supported, in part, by an National Health and Medical Research Council of Australia grant number 20895 and by a Monash University Faculty of Pharmacy and Pharmaceutical Sciences Small Grant. S. D. is currently affiliated with Rotary Bone Marrow Research Laboratories, Parkville, Victoria, Australia.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.