Neurod1 is essential for the survival and maturation of adult-born neurons

Abstract

The transcriptional program that controls adult neurogenesis is unknown. We generated mice with an inducible stem cell–specific deletion of Neurod1, resulting in substantially fewer newborn neurons in the hippocampus and olfactory bulb. Thus, Neurod1 is cell-intrinsically required for the survival and maturation of adult-born neurons.

The ability to generate new neurons provides the brain with an important level of plasticity for maintaining cellular homeostasis and potentially underlies a response to injury. Studies of adult neurogenesis have revealed extrinsic signals controlling the fate of resident neural stem cells¹. However, the cell-intrinsic transcription factors required to generate and promote the survival of newborn neurons have not been fully elucidated. During development, the proneural basic helix-loop-helix (bHLH) transcription factors are essential for embryonic neurogenesis². To identify genes required for adult neurogenesis, we focused on the bHLH transcription factor Neurod1 (also known as NeuroD or Beta2)^3,4. Neurod1 is predominantly expressed in the nervous system late in development and is therefore more likely to be involved in terminal differentiation, neuronal maturation and survival^3,5,6. Because Neurod1-null mice die perinatally from neonatal diabetes as a result of the loss of Ins gene expression⁷, we used an inducible gene ablation approach and found that Neurod1 is necessary for newborn granule cells to survive and fully mature in the adult mammalian brain.

To define the cells in the hippocampus subgranular zone (SGZ) that express Neurod1, we stained for NeuroD1 and a variety of cell type–specific markers in nestin promoter–driven GFP transgenic reporter mice⁸ (Fig. 1 and Supplementary Fig. 1). NeuroD1 was not expressed in the type 1 (nestin-GFP and GFAP double positive) radial stem-like cell (Fig. 1a) and was first detected in ~20% of type 2a (nestin-GFP positive, GFAP negative and DCX negative) progenitors (Fig. 1a). NeuroD1 colocalized with Ki67, a marker of dividing cells, in a small subset of progenitors (Fig. 1a) and did not colocalize with the majority of Sox2-positive cells, although we detected rare cells expressing both NeuroD1 and Sox2 (Supplementary Fig. 2). NeuroD1 expression peaked in 490% of type 2b (nestin-GFP and DCX double positive) progenitors and in ~100% of type 3 (DCX positive only) neuroblasts (Fig. 1a). NeuroD1 also overlapped with the homeobox transcription factor Prox-1, which labels immature and mature granule neurons, and was downregulated in mature NeuN-expressing granule neurons (Fig. 1a). These data suggest that NeuroD1 has a transient and dynamic expression profile that is restricted to progenitors/neuroblasts (type 2b and 3 cells) as they transition to immature granule neurons.

Figure 1. Critical role of Neurod1 in adult hippocampal neurogenesis.

(a) Immunostaining of dentate gyrus of nestin-GFP mice. Scale bar represents 50 µm. (b) Immunostaining of YFP-positive cells (green) and Neurod1 (red) in dentate gyrus of wild-type (WT) and NeuroD1 cKO mice at 40 d post TAM treatment (top). Arrowheads correspond to insets of enlarged orthogonal images of YFP and NeuroD1 double-positive cells in wild-type mice and YFP-positive and NeuroD1-negative cells in NeuroD1 cKO mice, respectively. We immunostained the dentate gyrus of wild-type and NeuroD1 cKO mice for YFP (green) and DAPI (nuclear counterstain, blue) at 6 and 40 d post TAM treatment (bottom). Bottom insets show enlarged images of YFP-positive cells that are type 1 with radial stem-like morphology. Scale bar represents 100 µm. (c) Time course analysis of the percentage of YFP and Neurod1 double-positive cells in wild-type mice (* P < 0.01 versus 1 d, ANOVA with Bonferroni post hoc, data represent ± s.d. from n = 3 mice). (d) Quantification of YFP-positive cells in the dentate gyrus of wild-type and NeuroD1 cKO at 6 and 40 d post TAM treatment (* P < 0.001, t test, data represent ± s.d. from n = 8–10 per group). (e) Number of YFP-positive cells in dentate gyrus that are type 1 and 2/3 cells or granule neurons 40 d post TAM treatment of wild-type and NeuroD1 cKO mice (* P < 0.001, t test, data represent ± s.d., n = 6–8 per group).

To study the role of Neurod1 in adult neurogenesis, we ablated Neurod1 in the neurogenic compartments of the adult brain: the subventricular zone (SVZ) and the SGZ. We crossed mice harboring a tamoxifen (TAM)-inducible form of Cre recombinase (CreER^T2) under nestin transcriptional control⁹ with Neurod1^loxP/loxP mice¹⁰, combined with the Rosa26 (R26R)-yfp reporter line, to achieve deletion of Neurod1 in all nestin-expressing stem cells and their progeny (NeuroD1 cKO; Supplementary Fig. 3 and Supplementary Methods).

To determine the dynamics of Neurod1 expression in recombined YFP-positive cells, Nes-creER^T2; R26R-yfp mice were killed at different time points after TAM treatment (Fig. 1c). About 10% of the YFP-positive cells were positive for Neurod1 by 1 d after TAM treatment, which increased to ~40% at 8 d post TAM treatment, mostly in type 2 and 3 cells (Fig. 1c). The percentage of YFP and Neurod1 double-positive cells reached a plateau after 8 d, consistent with an earlier study⁹.

In light of the Neurod1 expression dynamics, we chose two time points, 6 and 40 d post TAM treatment, for further analysis. Wild-type littermates (controls) and NeuroD1 cKO mice were killed at the indicated time points post TAM treatment. At both time points, we found successful Cre-mediated recombination and deletion of Neurod1 in NeuroD1 cKO mice (wild type, 3,216 ± 1,551; cKO, 510 ± 218 YFP and Neurod1 double-positive cells; Fig. 1b, Supplementary Fig. 2 and data not shown). We quantified the number of YFP-positive cells in the SGZ and found no difference between NeuroD1 cKO and wild-type mice at 6 d post TAM treatment (Fig. 1b,d), suggesting that Neurod1 is not required in early stem/progenitor cells. In contrast, there was a substantial decrease in the number of YFP-positive cells between wild-type and NeuroD1 cKO mice at 40 d post TAM treatment (Fig. 1b,d). Morphological assessment of these YFP-positive cells showed selective loss of newborn granule neurons, without changing the stem/progenitor cell pool (Fig. 1e and Supplementary Fig. 4), suggesting that Neurod1 is required in late-stage progenitors as they differentiate into hippocampal granule neurons but is dispensable in early stem/progenitor cells.

Consistent with our morphological analyses, the number of Sox2, GFAP and YFP triple-positive (Type 1), Sox2 and YFP double-positive, or Ascl1-positive (a proneural bHLH factor upstream of Neurod1) cells did not change between wild-type and NeuroD1 cKO mice at 40 d post TAM treatment (Fig. 2 and Supplementary Fig. 5). In addition, proliferation of YFP-positive cells (Ki67- and BrdU-positive cells) (data not shown) remained unchanged between the two groups at 6 d post TAM treatment (Fig. 2b,d). In contrast, there was a marked decrease in the number of YFP and Prox-1 double-positive (Fig. 2c,d) and YFP and DCX double-positive cells (Supplementary Fig. 6) at 40 d post TAM treatment. The deletion of Neurod1 could decrease the number of YFP-positive cells by halting the progression of progenitors/immature neurons to mature granule cells. Indeed, wild-type mice had mature YFP, Prox-1 and NeuN triple-positive granule neurons at 40 d post TAM treatment, which showed a decreasing trend in NeuroD1 cKO mice (Fig. 2c,d). Consistent with this, we observed a reduction, albeit to a lesser extent compared with what we observed at 40 d, of YFP-positive cells in NeuroD1 cKO mice compared with wild-type mice at an intermediate time point of 18 d post TAM treatment (Supplementary Fig. 7). These results suggest that Neurod1 is important in the differentiation of progenitors to immature and mature neurons, similar to a recent report¹¹.

Figure 2. Neurod1 is required for the survival and maturation of adult-born granule cell neurons.

(a) Type 1 radial stem-like cells expressed GFAP, YFP and Sox2. Arrows indicate triple-labeled cells in NeuroD1 cKO mice. (b) Representative image of Ki67, YFP and Neurod1 triple-positive cells (arrows) in the dentate gyrus of NeuroD1 cKO mice. (c) An example of a YFP and Prox-1 double-positive immature granule neuron (arrowhead) and a YFP, Prox-1 and NeuN triple-positive mature granule neuron (arrow) in NeuroD1 cKO mice. Scale bar represents 25 µm. (d) Quantification of YFP-positive cells expressing various markers in the dentate gyrus of wild-type and NeuroD1 cKO mice at 40 d post TAM treatment (* P < 0.001, t test, data represent ± s.d., n = 4–6 per group). (e) A representative image of two AC3-positive cells. Arrowheads represent enlarged insets of dentate gyrus of NeuroD1 cKO mice (white, top panel; black, bottom panel). Scale bars represent 100 µm and 15 µm (insets). (f) Dendrites of YFP-positive adult-born neurons. Shown are sample projections of Z series confocal images of YFP-positive neurons. Scale bar represents 20 µm. DG, dentate gyrus; ML, molecular layer.

The decrease in YFP-positive granule neurons in NeuroD1 cKO mice could also be a result of defects in the survival and/or the maturation of newborn granule cells. Notably, we found more active caspase 3 (AC3) positive cells at 40 d, but not at 6 or 18 d, post TAM treatment in NeuroD1 cKO mice (227 ± 85 cells) compared with wild types (77 ± 24 cells) (Fig. 2e and data not shown). Closer examination revealed that AC3-positive cells in the dentate gyrus of NeuroD1 cKO and wild-type mice did not colocalize with DCX at 40 d post TAM treatment, although AC3 and DCX double-positive cells were readily observed at 18 d post TAM treatment (Supplementary Fig. 8), suggesting that the death of newborn neurons is a transient event and/or the increase of dead/dying cells occurs after the DCX-positive immature neuron stage. Furthermore, we observed a reduction in dendritic lengths of YFP-positive granule neurons in NeuroD1 cKO mice (363 ± 72 µm) compared with wild types (598 ± 116 µm), which is consistent with early studies describing dendritic defects in the absence of Neurod1 (Fig. 2f)^6,12. Together, these data suggest that Neurod1 functions cell intrinsically in progenitors/immature neurons to promote their survival and maturation into mature granule neurons.

To examine whether Neurod1 is involved in olfactory bulb neurogenesis, we looked for YFP and Neurod1 double-positive cells in the SVZ, rostral migratory stream (RMS) and olfactory bulb in wild-type mice (Supplementary Fig. 9). Notably, YFP and Neurod1 double-positive cells were only detected in the caudal RMS and were rarely seen in the SVZ and olfactory bulb, suggesting that Neurod1 is involved in the migration of progenitors as they differentiate into olfactory bulb neurons. Because of the density of cells in the RMS, we quantified YFP-positive cells in the olfactory bulb glomerular cell and granule cell layers. The number of YFP-positive cells decreased in the glomerular cell and granule cell layers of NeuroD1 cKO mice at 40 d post TAM treatment compared with wild types (Supplementary Fig. 10). These data suggest that Neurod1’s function is also important for promoting the survival and differentiation of adult-born olfactory bulb neurons.

To further delineate the role of Neurod1 in adult neurogenesis, we prepared hippocampus and SVZ neurospheres from wild-type and NeuroD1 cKO mice and infected cells with adenoviruses expressing GFP (control) or Cre-GFP. We observed a decrease in Tuj1-positive neurons in Cre-GFP–infected NeuroD1 cKO cells (28 ± 1.2%) after neuronal differentiation compared with GFP-infected cells (41 ± 7%) (Fig. 3a). In contrast, there was no change in GFAP-positive astrocytes between Cre-GFP–infected NeuroD1 cKO cells (62 ± 12%) and wild-type cells (65 ± 6%) (Fig. 3b). Moreover, wild-type cells over-expressing a full-length Neurod1 cDNA showed an increase in Tuj1-positive neurons (from 0 to 70 ± 12%) under proliferating conditions (Fig. 3c). Neurod1 and Tuj1 mRNA levels decreased in Cre-GFP–infected NeuroD1 cKO cells (Fig. 3d). In the absence of Neurod1, there was a decrease in the expression of neuronal/pro-survival genes, whereas the expression of Gfap was not dramatically altered (Fig. 3d). Thus, Neurod1 is both necessary and sufficient to promote neuronal differentiation and maturation in vitro, consistent with its essential role in vivo.

Figure 3. Neurod1 is required for neuronal, but not astrocyte, differentiation in neurosphere cultures.

(a) Immunostaining for the neuronal marker Tuj1 (red) and adenovirus expressing control GFP or Cre-GFP (green) in SGZ and SVZ neurospheres from NeuroD1 cKO mice. (b) Immunostaining for GFAP (red) and adenovirus expressing Cre-GFP (green) in SGZ and SVZ neurospheres from wild-type and NeuroD1 cKO mice. (c) Immunostaining for Tuj1 (red) and IRES-GFP or Neurod1–IRES-GFP (green) in SGZ and SVZ neurospheres from wild-type mice. DAPI is stained in blue. Scale bars represent 10 µm. (d) RT-PCR analysis of NeuroD1 cKO neurospheres under neuronal differentiation conditions after infection with adenovirus expressing GFP or Cre-GFP. Gapdh is used as a normalization control.

Using a genetic approach in mice, we found that Neurod1 is a cell-intrinsic factor required for the survival and differentiation of newborn neurons in the adult SGZ and SVZ. Recent studies have shown that a wide variety of stimuli, including seizures, running and enriched environment, can profoundly induce adult neurogenesis¹. An exciting possibility is that these stimuli may act, in part, via Neurod1 target genes to control the survival and maturation of newborn neurons¹³. Although the cause-and-effect relationship between cell death and failure to mature has not been fully established, our results suggest that Neurod1 is not only critical for the survival and differentiation of adult-born granule neurons but also for their maturation and integration into the neuronal circuitry. These studies suggest that Neurod1 is a critical regulator of adult neurogenesis and may serve as a potential therapeutic target in regenerative medicine and in the prevention of age-related cognitive decline.