The germinal matrices in the developing dentate gyrus are composed of neuronal progenitors at distinct differentiation stages



Background: Differentiation of granule cells (GCs) begins from late embryonic stage in the developing dentate gyrus (DG). Migration of the neurogenic stem cells and progenitors in the developing DG makes understanding of the DG morphogenesis difficult. The proliferative area in the developing DG was divided into the three germinal matrices (GMs). However, the stage of the progenitor cells in each GM along the GC differentiation process is not clear. Results: Here, we extensively compared expression of neurogenic transcription factors (TFs) of which sequential expression in the neocortical development and the adult DG neurogenesis was reported. The GC differentiation marked by Prox1 expression takes place from embryonic day 16.5 in the tertiary GM. Although neurogenesis in each GM basically proceeds along the radial axis of the forming GC layer, cells expressing stem cell markers were observed intermingling with NeuroD/Prox1 expressing differentiated cells in the tertiary GM at postnatal day 5, and gradually restricted in the subgranular zone as the development went on. Conclusions: We describe expression pattern of neurogenic TFs during DG development, which suggests conserved sequential expression of TFs in the GC lineage, and spatiotemporal relationships of GC differentiation and DG morphogenesis during embryonic and early postnatal periods. Developmental Dynamics 242:1442–1453, 2013. © 2013 Wiley Periodicals, Inc.


The hippocampal formation is one of the brain components of the limbic system playing important roles in learning and memory as well as regulation of emotional behaviors (Amaral and Witter, 1989; Deng et al., 2010; Goshen et al., 2011; Liu et al., 2012; Nakashiba et al., 2012). The hippocampal formation is composed of two major neuronal structures, the cornu ammonius (CA1, CA2, and CA3) and the dentate gyrus (DG). The DG mainly consists of granule cells (GCs), and it is the primary gateway of the tri-synaptic pathway for inputs into the hippocampal formation receiving from the entorhinal cortex (Witter, 2007). The neural stem cells (NSCs) in the subgranular zone (SGZ) of the adult DG produce new neurons throughout life, and incorporation of new neurons into the neuronal circuit probably plays important roles in normal brain function and pathogenesis of human psychiatric diseases (Altman and Das, 1965; Seki and Arai, 1993; Deng et al., 2010; Aimone et al., 2011).

The development of DG begins in embryonic stage and continues to postnatal stage (Angevine, 1965; Altman and Bayer, 1990a,b). The morphogenesis of DG is a very complex process, since unlike other cortical regions, proliferative progenitors migrate away from the neuroepithelium to form presumptive DG (Fig. 1). The progenitors of the GCs originate from the medial embryonic cortical area called the dentate neuroepithelium (DNE) (Fig. 1A). The DNE is morphologically marked by a ventricular indentation called the dentate notch (Fig. 1B; Altman and Bayer, 1990b). Altman and Bayer (1990a,b) marked proliferative cells in the developing rat DG by incorporation of tritium thymidine, and they defined three germinal matrices (GMs) in the developing DG. The primary (1ry) GM is in the DNE. The secondary (2ry) GM is the population of proliferative cells above the 1ry GM, and these cells subsequently form the dentate migratory stream (DMS) along the meninges (Fig. 1C). The tertiary (3ry) GM is formed at the distal end of the DMS (Fig. 1C,D). The 1ry and 2ry GMs gradually disappear, and the 3ry GM becomes prominent after birth (Fig. 1E,F). In 3 weeks postnatal DG, the 3ry GM is not observed, and cell proliferation is restricted in the SGZ (Fig. 1G; Altman and Bayer, 1990a).

Figure 1.

The embryonic and postnatal development of the dentate gyrus. The morphogenesis of the dentate gyrus and the histological terminology at E12.5 (A), E14.5 (B), E16.5 (C), E18.5 (D), P5 (E), P10 (F), and P20 (G) are schematically shown. The colored dots indicate progenitor cells in the 1ry germinal matrix (GM) (blue), 2ry GM (green), and 3ry GM (yellow), and the subgranular zone (red). The anatomical axes are indicated in A (D, V, M, and L indicate dorsal, ventral, medial, and lateral, respectively). CA, cornu ammonius; CH, cortical hem; DNE, dentate neurpepithelium; FDJ, fimbrio-dentate junction; GCL, granule cell layer; Hf, hippocampal fissure; HNE, hippocampal neuroepithelium; LBL, lower blade of the GCL; ML, molecular layer; SGZ, subgranular zone; UBL, upper blade of the GCL; 1ry, primary germinal matrix; 2ry, secondary germinal matrix;, 3ry, tertiary germinal matrix.

Sequential transcription factor (TF) cascade (Pax6 → Tbr2 → NeuroD) regulates neurogenesis to produce glutamatergic neurons in the developing neocortex (Englund et al., 2005; Hevner et al., 2006). The radial glial cells function as the NSCs expressing Pax6 and Sox2 in the VZ (Gotz et al., 1998; Wen et al., 2008). Radial glial cells generate not only neurons but also Tbr2-expressing intermediate progenitors (IP) that form the subventricular zone (SVZ) (Hevner et al., 2006; Arnold et al., 2008; Sessa et al., 2008). Differentiating IPs in the upper SVZ express NeuroD, and NeuroD expression persists in immature neurons (Lee et al., 2000; Hevner et al., 2006). In the adult DG, the GC differentiation process was divided into several stages, and the proliferative cells were designated to type1, type2, and type3 cells by their morphology and expressing molecules (Kempermann et al., 2004; von Bohlen and Halbach, 2011). The type1 cells are the radial glia-like stem cells that express NSC markers such as Pax6 and Sox2, the type2 cells are transit amplifying cells that intensively express IP marker Tbr2, and the type3 cells are neuroblasts that highly express NeuroD (Kempermann et al., 2004; Maekawa et al., 2005; von Bohlen und Halbach, 2011). The sequential TF cascade found in the developing neocortex is conserved in the adult SGZ (Hodge et al., 2008, 2012; von Bohlen und Halbach, 2011).

The neurogenic TFs, Sox2, Pax6, Tbr2, and NeuroD, are expressed during DG development (Li et al., 2009; Duan et al., 2012; Hodge et al., 2012). Prox1 is expressed in postmitotic GCs (Liu et al., 2000; Pleasure et al., 2000; Hodge et al., 2013). Gene knockout studies indicated that these TFs are important for differentiation of GCs. Sox2 is required for NSC maintenance in the postnatal DG, and deletion of Sox2 leads hypoplasia of the postnatal DG (Favaro et al., 2009). Loss of Tbr2 increased Sox2-expressing progenitors and decreased NeuroD-expressing cells in the DG development, suggesting that Tbr2 is essential for transition from Sox2-expressing proliferative stage to NeuroD-expressing immature neuronal stage (Hodge et al., 2012). NeuroD-deficient mouse (Miyata et al., 1999; Liu et al., 2000) and Prox1-deficient mouse (Lavado et al., 2010) lack the DG by cell death of the GC lineage. Because an induced Prox1 knockout in the postmitotic GC changes cellular characteristics from the GC to the CA3 pyramidal cell, Prox1 is required for maintenance of GC features (Iwano et al., 2012).

Although the importance of these TFs in the DG development was reported previously, detailed analyses linking spatiotemporal expression pattern of TFs and the DG morphogenesis have not been carried out. Here, we performed immunohistochemistry to compare expression of these TFs to obtain insights into morphogenetic processes of the developing DG.


Pax6 Expression Marks the Boundary Between the Dentate Neuroepitheilum and the Cortical Hem

We have been studying Pax6 function in the brain development (Osumi et al., 2008). Pax6 is expressed in the NSCs in the VZ of the dorsal forebrain, and it is well known that Pax6 expression forms gradient along the rostro-caudal and dorso-ventral axes of the dorsal forebrain (Stoykova et al., 1997; Bishop et al., 2000; Cocas et al., 2011). Here, we found an areal difference in intensity of Pax6 immunoreactivity in the medial telencephalon at embryonic day (E)12.5 (Fig. 2A–C). It was reported that EphB1 expression demarcates the DNE and the cortical hem (CH) (Tole and Grove, 2001). Simultaneous detection of Pax6 and EphB1 indicated that the dorsal side of the boundary exhibiting higher Pax6 expression is the DNE and the ventral side expressing Pax6 weakly is the CH (Fig. 2A–D). Sox2 is also expressed in the NSCs in the VZ (Fig. 2E) (Wen et al., 2008; Hodge et al., 2012). Interestingly, strong expressions of Pax6 and Sox2 showed a spatially complimentary pattern (Fig. 2E–G). Tbr2-expressing cells were observed in the meningeal layer in both DNE and CH, but few Tbr2-expressing cells were observed in the ventricular side at the DNE/CH boundary (Fig. 2H). Thus, the NSCs in the DNE and the CH are molecularly different at E12.5 in terms of the expression level of TFs, which are involved in stem cell function (Episkopou, 2005; Osumi et al., 2008).

Figure 2.

The boundary between the dentate neuroepithelium and the cortical hem is defined by Pax6 expression at embryonic day 12.5. A: Expressions of Pax6 (green) and EphB1 (red) in the medial telencephalon. B–D: Higher magnification view of a boxed area in A. E–G: Expressions of Pax6 (green) and Sox2 (red). H: Expressions of Tbr2 (green) and Sox2 (red). The dashed lines indicate the boundary between the dentate neuroepithelium and the cortical hem. DNE, dentate neuroepithelium; CH, cortical hem. Scale bars = 50 μm.

Figure 3.

Pax6-expressing cells are proliferative in the developing dentate gyrus. Expressions of Pax6 (green) and PCNA (red) in the developing dentate gyrus at E12.5 (A), E14.5 (D), E16.5 (G), E18.5 (L), P5 (Q), P10 (R), and P20 (S). B,C, E,F, H-K, M-P: Higher magnification views of boxed areas in A,B,C,D,G,L,Q, R, and S, respectively. White broken line shows the boundary between the DNE and CH (B and C). Asterisks (D, G, and L) point the dentate notch. a–c: Higher magnification views of the boxed areas in F showing the hippocampal neuroepithelium, the dentate neuroepithelium, and the cortical hem, respectively. The white broken line in J shows the outline of the 3ry GM. The abbreviations are the same as those in Figure 1. Scale bars = 50 μm (A, D, G, L, Q, R, and S) and 20 μm (B, C, E, F, H–K, and M–P).

Pax6 Expression Marks Granule Cell Progenitors of the Dentate Gyrus

Because strong Pax6 expression was observed in the VZ of the DNE at E12.5, we decided to follow Pax6 expression during DG development. Pax6 expression was observed in three GMs (Altman and Bayer, 1990a,b; Fig. 1C–F in the developing DG; Fig. 3). Proliferating cell nuclear antigen (PCNA) expression (Celis and Celis, 1985; Leonhardt et al., 2000) was used to determine proliferation of the Pax-expressing cells. PCNA expression was heterogeneous in the proliferative areas, because it increases in late G1 and, reaches the highest level in S-phase, and decreases in G2 and M-phases (Bolton et al., 1992; Linden et al., 1992). Apparent PCNA expression was observed in the meningeal cells, which outline the surface of the morphogenetic field of the DG from E12.5 up to postnatal day (P)10. These cells are not neurogenic, because they did not co-express Pax6 (Fig. 3A,D,G,L,Q,R), nor neurogenic TFs as described below. At E12.5, PCNA was expressed in the VZ throughout the embryonic cortex (data not shown), and there was no difference in its expression level between the CH and the DNE (Fig. 3A,C).

At E14.5, PCNA expression around the dentate notch (indicated by an asterisk in Fig. 3D) was weaker than those in the hippocampal neuroepithelium (HNE) and the CH (Fig. 3D, F, and a–c). The region dorsal to the dentate notch is presumably the HNE (Altman and Bayer, 1990b; Fig. 1B), which gives rise to the Cornu Ammonius (CA). Pax6 expression level in the DNE was similar to that in the HNE (Fig. 3D,E). Expression pattern of PCNA in the DNE and its vicinity area is consistent to localization of proliferative cells detected by the radiograms (Altman and Bayer, 1990b). The 2ry GM cells migrating out from the DNE expressed Pax6 and PCNA at E14.5 (Fig. 3D), suggesting reentry of these cells into the cell cycle.

At E16.5, Pax6 expression was strong in the 1ry GM (the neuroepithelium dorsal to the point indicated by an asterisk in Fig. 3G). Pax6-expressing cells in the 1ry and 2ry GMs coexpressed PCNA (Fig. 3H,I), whereas Pax6+/PCNA cells were observed in the surface layer of the forming 3ry GM (Fig. 3G,J,K). At E18.5, strong Pax6 expression in the 1ry GM persisted (Fig. 3L). At E18.5 Pax6+/PCNA+ cells were continuously observed in the 1ry and 2ry GMs (Fig. 3L,M,N). The dentate migratory stream (DMS) (Altman and Bayer, 1990b), which is a stretch of the cells between the 2ry and 3ry GMs along the meninges (Fig. 1C,D), contained Pax6+/PCNA+ cells (Fig. 3L,O,P).

At P5, Pax6+/PCNA+ cells were evenly distributed in the presumptive DG area (3ry GM) (Fig. 3Q). The 1ry and 2ry GMs marked by Pax6 and PCNA expression were no longer observed at P10. In the 3ry GM, cells expressing Pax6 and PCNA segregated into the future SGZ (Fig. 3R) At P20, the distribution of these cells became indistinguishable from that in the adult DG (Fig. 3S) (Maekawa et al., 2005).

Unlike in the VZ of the neocortical area, we observed a substantial number of Pax6+/PCNA cells in the GMs during DG development. The lack of PCNA expression in these Pax6-expressing cells suggests that the Pax6+ cell population in the developing DG contains postmitotic cells.

The Tertiary Germinal Matrix Contains Postmitotic Granule Cells

Because Prox1 is clearly expressed in the postmitotic GC (Liu et al., 2000; Pleasure et al., 2000; Hodge et al., 2013), and its genetic disruption abolishes DG (Lavado et al., 2010), we used Prox1 as a reliable postmitotic GC marker. Because GMs are marked by Pax6 expression as described above (Fig. 3), we expected to obtain information about GC production from the GMs by simultaneous detection of Pax6 and Prox1.

Very faint Prox1 immunoreactivity was detectable in the 1ry and 2ry GMs from E14.5 (Fig. 4A). Few sporadic cells expressed Prox1 moderately within this faintly stained region (Fig. 4A,G). Evident Prox1 expression was observed in the 3ry GM at E16.5, whereas weak Prox1 immunoreactivity was observed in very small number of cells in the 2ry GM and the DMS (Fig. 4B,H). Pax6 expression and apparent Prox1 expression were basically exclusive to each other until this developmental stage (Fig. 4G,H). In E18.5, Pax6+ cells and Prox1+ cells were intermingled in the 3ry GM (Fig. 4C,I). After birth, Prox1+ cells gradually segregated to form the GCL (Fig. 4D–F) and Pax6+ cells were gradually restricted in the SGZ as the development went on (Fig. 4J–L). In every stage we examined, some Pax6+/Prox1+ cells were observed (arrows in Fig. 4G–L; Pax6+•Prox1+/Pax6+ cells were 11.4% [E18.5], 20.4% [P5]), 27.6% [P10], 27.9% [P20]). These cells may be at a transit situation from DG progenitors to postmitotic GCs.

Figure 4.

Distribution of the granule cells and progenitor cells in the developing dentate gyrus. Postmitotic cells of the granule cell lineage was detected by Prox1 expression in the developing dentate gyrus at E14.5 (A), E16.5 (B), E18.5 (C), P5 (D), P10 (E), and P20 (F). White broken lines in A–C show the outline of the developing dentate gyrus. G, H, I, J,K, and L show high-magnification views of boxed areas in A–F, respectively, indicating expressions of Pax6 (green) and Prox1 (red). Small arrows indicate cells coexpressing Pax6 and Prox1. The abbreviations are the same as those in Figure 1. Scale bars = 50 μm (A–F) and 20 μm (G–L).

Expression of Neurogenic Transcription Factors in the Developing Dentate Gyrus

Because the GC differentiation indicated by strong Prox1 expression was observed exclusively in the 3ry GM, we examined differentiation stages of cells in the three GMs by comparing expression of Pax6, Sox2, Tbr2, and NeuroD. These TFs are known to exhibit a sequential expression in the developing neocortex and the SGZ of the adult DG (Hevner et al., 2006; von Bohlen und Halbach, 2011).

Pax6 and Sox2 were coexpressed in all cells in the 1ry GM at E14.5 (Fig. 5A). In the 2ry GM at E14.5, Pax6 expression was observed (Fig. 5A), but Sox2 expression was not observed (Fig. 5B). In the 2ry GM, 75.0% of Pax6+ cells coexpressed Tbr2, and 68.6% of Tbr2+ cells coexpressed Pax6. NeuroD expression, indicating a more differentiated stage (Hevner et al., 2006), was observed dorsally to Pax6+ and/or Tbr2+ cells in the 2ry GM (Fig. 5C,D). Pax6+/NeuroD+ and Tbr2+/NeuroD+ cells were observed above the 2ry GM (Fig. 5C,D).

Figure 5.

Expression of transcription factors in the embryonic dentate gyrus. A–D: The 1ry and 2ry GMs at E14.5. E–H: The 1ry and 2ry GMs at E16.5. I–L: The 3ry GM at E16.5. M–P: The 2ry GM at E18.5. Q–X: The 3ry GM at E18.5. U–X show higher magnification views of the boxed areas in Q–T. A, E, I, M, Q, and U: The green and red signals indicate expression of Pax6 and Sox2, respectively. B, F, J, N, R, and V: The green and red signals indicate expression of Tbr2 and Sox2, respectively. C, G, K, O, S, and W: The green and red signals indicate expression of Pax6 and NeuroD, respectively. D, H, L, P, T, and X: The green and red signals indicate expressions of Tbr2 and NeuroD, respectively. Asterisks (A–H) point the dentate notch. Scale bars = 50 μm

Figure 6.

Expression of transcription factors in the postnatal dentate gyrus. A–I: The forming dentate gyrus at P5 (A–C), at P10 (D–F), and at P20 (G–I). A, D, and G: Expression of Pax6 (green) and Sox2 (red) are indicated. The sections were counterstained by DAPI (blue). B, E, and H: Expression of Pax6 (green) and NeuroD (red). C, F, and I: Expression of Tbr2 (green) and NeuroD (red). a–l: Higher magnification views of the boxed areas in A–I. The abbreviations are the same as those in Figure 1. Scale bars = 50 μm

At E16.5, Pax6+/Sox2+ cells were observed in the 2ry and 3ry GMs (Fig. 5E,I). Coexpression of Sox2 and Tbr2 was scarcely observed (Fig. 5F,J). NeuroD+ cells increased in the dorsal side of the 2ry GM as development went on (Fig. 5C,D,G,H,O,P). NeuroD+ cells were abundant in the 3ry GM (Fig. 5K), and Tbr2+ cells were predominantly observed in the surface layer of the 3ry GM (Fig. 5L). Similar spatial expression pattern of these TFs was observed in the 2ry and 3ry GMs at E18.5 (Fig. 5M–X), increasing rates of NeuroD+ cells (Fig. 5S,T,W,X). At E18.5 in the developing DG area, 30.0% of Pax6+ cells were Tbr2+, and 65.0% of Tbr2+ cells were Pax6+.

In the postnatal 3ry GM, the C-shaped GCL gradually became apparent by expression of NeuroD (Fig. 6). At P5, Pax6+, Sox2+, and Tbr2+ cells localized predominantly in area surrounding the GCL forming area (Fig. 6B,C), while less cells expressing these TFs were also observed evenly throughout the forming GCL (Fig. 6A–C). Coexpression of Pax6 and Sox2 was observed in these cells (Fig. 6a,b). Pax6 expression was weaker in the cells inside of the forming GCL (Fig. 6A,a,b). A large fraction of Pax6+ cells in the zone surrounding GCL coexpressed Tbr2 (34.4% of Pax6+ cells were Tbr2+ at P5, 36.7% at P10, and 34.7% at P20. 63.9% of Tbr2+ cells were Pax6+ at P5, 70.1% at P10, and 84.6% at P20). NeuroD+ cells above the cell-dense area of the forming GCL strongly expressed Pax6 (Fig. 6c) or Tbr2 (Fig. 6e), whereas the majority of NeuroD cells in the cell-dense area did not express or weakly expressed Pax6 (Fig. 6d) and Tbr2 (Fig. 6f). It is likely that Pax6+/Sox2+/ NeuroD- cells in the forming GCL at P5 are the NSCs. Expression of Pax6, Sox2, and Tbr2 gradually became restricted into the SGZ (Fig. 6A–I, a–l). Until P20 Pax6+ cells were a larger population than Sox2+ cells (Fig. 6g,j), and some Pax6+ cells coexpressed NeuroD (Fig. 6h, and k), suggesting that Pax6 expression persisted longer than that of Sox2 in the GC cell lineage. In the 3ry GM, 38.4% (E18.5), 57.6% (P5), 46.8% (P10), and 43.4% (P20) of Pax6+ cells were NeuroD+ in the histological sections we observed.

Until P5, the proximal end of the DMS is localized above the presumptive fimbria (Figs. 1D, 3L) and this region is called the fimbrio-dentate junction (or juncture) (FDJ) (Rickmann et al., 1987; Li et al., 2009). Then, 2ry GM including the DMS detaches from the fimbrial region and becomes the lateral end of the lower blade of the GCL (Fig. 1D–F; Li et al., 2009; Hodge et al., 2012). At P5, Tbr2+ cells occupied the pial surface layer of the lateral end of the lower blade (Fig. 7C). Pax6+/NeuroD+ cells and Tbr2+/NeuroD+ cells were observed in this region (Fig. 7B,C). Pax6-, Sox2-, and Tbr2-expressing cells were gradually restricted into the SGZ (Fig. 7). These TFs expression profiles in this region were similar to the embryonic 3ry GM at E16.5 and E18.5 (Fig. 5I–L, Q–T). Thus, neurogenesis in the proximal end of the DMS remains active until later stages than that in the distal part of the DG.

Figure 7.

Expression of transcription factors in the lateral end of the lower blade. A–I: The lateral end of the lower blade of the dentate gyrus at P5 (A–C), at P10 (D–F), at P20 (G–I). A, D, and G: Expression of Pax6 (green) and Sox2 (red) indicated. The sections were counterstained by DAPI (blue). B, E, and H: Expression of Pax6 (green) and NeuroD (red). C, F, and I: Expression of Tbr2 (green) and NeuroD (red). The abbreviations are the same as those in Fig. 1. Scale bars = 50 μm.


Molecular Anatomy of the Germinal Matrices in the Developing Dentate Gyrus

The pioneering works by Altman and Bayer (1990a,b) showed localization of proliferative cells and defined GMs in the developing DG. However, their studies did not provide information about the GC differentiation process in each GM because of lacking markers to distinguish differentiating steps along the GC lineage. We found the appearance of Sox2-expressing cells in the 2ry GM and DMS at E16.5 (Fig. 5F). Because proliferative cells come from the DNE (Altman and Bayer, 1990a,b; Li et al., 2009), these cells exited from the DNE (or 1ry GM) later than E14.5 to join the 2ry GM and then DMS. The cell segregation at a distal region of the DMS primarily forms the 3ry GM. Prox1 expression was evident only in the 3ry GM, indicating that GC differentiation takes place in the 3ry GM from E16.5, and cells in the 1ry and 2ry GMs are all IPs and/or NSCs. These IPs and NSCs are subsequently incorporated into the DG.

Expression of TFs observed in this report is summarized in Figure 8. Using Chip-on-chip and the microarray analyses, Sansom et al. (2009) showed that Pax6 directly binds to regulatory genomic sequences to enhance expression of neurogenic factors, such as Tbr2 and NeuroD. Wen et al. (2008) demonstrated that Pax6 also modulates Sox2 expression by binding Sox2 promoter region. These previous reports are consistent with our observation that Pax6 expression is merged with that of Sox2, Tbr2, and NeuroD. Hodge et al. (2012) reported that Tbr2 binds to Sox2 promoter to suppress Sox2 expression. This molecular mechanism is consistent with our observation that expression of Sox2 and Tbr2 is exclusive of each other. Although Sox2+/Tbr2+ cells were observed, they are likely at a transitional situation from the Sox2+ to Tbr2+ stage. Pax6 and Prox1 also were expressed exclusive to each other, suggesting there is a similar molecular regulation between them. In the adult neurogenesis, type1, type2, and type3 cells predominantly express Pax6/Sox2, Tbr2, and NeuroD, respectively (Hodge et al., 2008; Fig. 8). According to TFs expression, this terminology can be applied to the cells in the GMs during the DG development. The 1ry GM at E14.5 consists of type1 cells. The 2ry GM at E14.5 consists of type1, type2, and type3 cells. In the 2ry GM and the DMS during embryogenesis, the type3 cells located 2 or 3 cells away from the meningeal surface (Fig. 5D, H, and P). In the DMS, the layer facing the meninges was predominantly occupied by type2 cells (Fig. 5B, D, F, H, and P). The type1 cells were observed evenly throughout the developing DG up to E18.5 (Fig. 5F, J, N, and R). Interestingly, color gradation was observed in our immunostaining images of TFs in the 2ry and 3ry GMs (Figs. 5G,H, K, L, O, P, W, X, 6B, C). This gradation likely indicates transition of GC differentiating stages, suggesting conservation of sequential expression of these neurogenic TFs. Based on the sequential TFs cascade (Englund et al., 2005; Hevner et al., 2006), the differentiation process of the GCs is suggested as described below. Evenly distributed type1 cells produce type2 cells, which predominantly migrate toward the meninges or hippocampal fissure. As type2 cells become type3 cells, they begin to form the GCL, and start GC differentiation as strong Prox1 expression begins. Direct observation of behaviors of living cells expressing Pax6, Tbr2, or NeuroD by time-lapse imaging will help confirming this presumption.

Figure 8.

Summary of expression pattern of the transcription factors in the dentate gyrus.

Different Developmental Process Between the Upper and Lower Blades of the Dentate Gyrus

The mature GCL is divided into two parts, the upper-blade (UBL) and the lower-blade (LBL), which locates dorsal and ventral to the hilus, respectively (Fig. 1E–G). At E16.5, the surface layer of the future UBL region contained Pax6+ and Tbr2+, but not Sox2+ cells (Fig. 5I,J). At E18.5, Pax6-, Tbr2-, and NeuroD-coexpressing cells were abundantly observed in the ventral side of the 3ry GM, which gives rise to the LBL after birth, whereas NeuroD single positive cells were dominant in the future UBL side (Fig. 5W,X). Disappearance of Pax6, Sox2, and Tbr2 expression in the outside of the UBL and LBL was observed at P10 and P20, respectively (Fig. 6D–I). These observations suggest that development of UBL precedes that of LBL, as previously suggested from a classical histological study (Angevine, 1965).

Origin of Adult Neural Stem Cells in the Dentate Gyrus

Sox2+/Pax6+ cells were evenly distributed throughout the 3ry GM at E18.5 (Fig. 5U) and P5 (Fig. 6A), and gradually restricted into the SGZ (Fig. 6D,G). Change of Sox2+/Pax6+ cells distribution possibly indicates migration of these cells along the radial axis of the developing DG. Alternatively, postnatal change of Sox2+/Pax6+ cells' distribution indicates maintenance of undifferentiated cells in the future SGZ and differentiation of other Sox2+/Pax6+ cells observed in earlier stages. Li et al. (2013) suggested migration of the adult NSC from the ventral part of the developing DG, but the migration pathway in the developing DG is not very clear yet. Establishment of adult NSC in the SGZ of DG requires SHH signaling, which is expressed in the ventral forebrain and differentiating DG (Li et al., 2013). There are Cajal-Retzius (CR) cells in the molecular layer surrounding the developing DG (Berger et al., 2007). Reelin protein, which is expressed by CR cells, has activity to promote neuronal differentiation (Kwon et al., 2009). In the neocortical development, the meninges is involved in regulation of neurogenesis by secretion of retinoic acid (Siegenthaler et al., 2009). These facts support latter developmental mechanism, suggesting that Pax6+/Sox2+ cells localizing far from resources of such neurogenesis promoting factors stay undifferentiated by an effect of SHH signal, and some of these cells are incorporated into the stem cell niche in the forming SGZ to become adult NSCs in the mature DG.



C57BL/6J mice were purchased from Japan Charles River Laboratories (Tokyo, Japan). The morning of the day when the virginal plug is observed was termed E 0.5; the day of birth was referred to as P0. Animals were housed under standard conditions (12 hr-light/dark cycle), and all experimental procedures were carried out in accordance with the National Institute of Health guidelines for the care and use of laboratory animals, and were approved by The Committee for Animal Experiments in Tohoku University School of Medicine.

Tissue Preparation

Embryonic stage

Embryos at embryonic day 12.5 (E12.5) were dissected from mothers that were anesthetized with a lethal dose of pentobarbital sodium (0.1 ml 200 mg/kg, i.p.) and immersion-fixed in 4% paraformaldehyde (PFA, Sigma, St. Louis, MO) in 0.01 M phosphate-buffered saline (PBS) at 4°C for 3 hr. The brains were rinsed in PBS, and cryoprotected in 20% sucrose in 0.01 M PBS at 4°C. Embryos from E14.5, E16.5, and E18.5 stages were chilled in cold PBS and then perfused transcardially with cold 4% PFA in 0.01 M PBS. The brains were postfixed in 4% PFA at 4°C for 3 hr, rinsed in PBS, and cryoprotected in 20% sucrose in 0.01 M PBS at 4°C.

Postnatal stage

Postnatal mice at P5, P10, and P20 were anesthetized with a lethal dose of pentobarbital sodium (0.1 ml 200 mg/kg, i.p.), followed by transcardial perfusion for 10 min with cold 4% PFA in 0.01 M PBS. Brains were dissected out of the skull, rinsed in PBS, and cryoprotected in 20% sucrose in 0.01 M PBS at 4°C.


Brains were sectioned at 14 μm thickness using a cryostat (CM3050, Leica, Exton, PA), and mounted onto slides (S9441, Matsunami). The slides were washed three times in 0.3% Triton X-100 in PBS (PBT), and nonspecific staining was blocked with 5% normal donkey serum (Sigma) in PBT at room temperature (RT) for 1 hr. The slides were incubated with primary antibodies overnight at 4°C. The slides were washed three times in PBT, and then incubated with secondary antibodies at RT for 1 hr. After three times washing in PBT, the slides were coverslipped with fluorescent mounting medium, Vectashield (H-1000, Vector Labs, Burlingame, CA).

Sections were immunostained for primary antibodies against PCNA (mouse monoclonal; MCL-PCNA, Novocastra, Bannockburn, IL; 1:100), Pax6 (rabbit polyclonal; 1:1,000) (Inoue et al., 2,000), Sox2 (goat polyclonal; AF2018, R&D Systems, Minneapolis, MN; 1:500), Tbr2 (rabbit polyclonal; ab23345, Abcam, Cambridge, MA; 1:1,000), NeuroD (goat polyclonal; sc-1084, Santa Cruz Biotechnology, Santa Cruz, CA; 1:200), Prox1 (goat polyclonal; AF2727, R&D Systems; 1:1,000), EphB1 (goat polyclonal; sc-926, Santa Cruz; 1:200). Secondary antibodies were donkey anti-rabbit Dylight 488 (Jackson Immunoresearch, West Grove, PA; 1:500), donkey anti-mouse Cy3 (Jackson Immunoresearch; 1:500), or donkey anti-goat Cy3 (Jackson Immunoresearch; 1:500). Nuclei were counterstained with DAPI (Sigma, 1:1,000). Fluorescent images were photographed under confocal laser scanning fluorescent microscope (Zeiss META510, Carl Zeiss, Thornwood, NY). Detection of PCNA using the mouse monoclonal antibody (Novocastra) gave us high background signals in mouse tissue. To decrease this background signal, we treated the tissue sections prior to primary antibody incubation by Donkey anti-mouse monovalent Fab fragment, then the slides were washed in PBT three times. To detect PCNA efficiently, we boiled the slides in 10 mM sodium citrate solution (pH 6.0) at 80°C for 10 min (Jiao et al., 1999). After heating, the slides were kept at RT for 3 hr to reduce the temperature of the solution, and washed in PBT. These slides were incubated with Fab Donkey anti-mouse IgG (1:10 diluted in PBT, Jackson Immunoresearch, 715-007-003) overnight at RT, washed three times in PBT, and incubated with anti-PCNA primary antibody.


We thank Ms. Emi Otsuki, and Sayaka Makino for their technical support, and members of the laboratory for helpful discussions. We also acknowledge the support of the Biomedical Research Core of Tohoku University Graduate School of Medicine. Y.K. was supported by KAKENHI of MEXT (Kiban C No.22570202). T.S. was supported by a Booster Grant from Tohoku University Graduate School of Medicine.