Conserved molecular signatures of neurogenesis in the hippocampal subgranular zone of rodents and primates

Abstract

The neurogenic potential of the subgranular zone (SGZ) of the hippocampal dentate gyrus is likely to be regulated by molecular cues arising from its complex heterogeneous cellular environment. Through transcriptome analysis using laser microdissection coupled with DNA microarrays, in combination with analysis of genome-wide in situ hybridization data, we identified 363 genes selectively enriched in adult mouse SGZ. These genes reflect expression in the different constituent cell types, including progenitor and dividing cells, immature granule cells, astrocytes, oligodendrocytes and GABAergic interneurons. Similar transcriptional profiling in the rhesus monkey dentate gyrus across postnatal development identified a highly overlapping set of SGZ-enriched genes, which can be divided based on temporal profiles to reflect maturation of glia versus granule neurons. Furthermore, we identified a neurogenesis-related gene network with decreasing postnatal expression that is highly correlated with the declining number of proliferating cells in dentate gyrus over postnatal development. Many of the genes in this network showed similar postnatal downregulation in mouse, suggesting a conservation of molecular mechanisms underlying developmental and adult neurogenesis in rodents and primates. Conditional deletion of Sox4 and Sox11, encoding two neurogenesis-related transcription factors central in this network, produces a mouse with no hippocampus, confirming the crucial role for these genes in regulating hippocampal neurogenesis.

Keywords: Development; Hippocampus; Neurogenesis; Rhesus monkey; Subgranular zone; Transcriptome.

Fig. 1.

Identification of neurogenic niche genes in the adult mouse SGZ. (A) Thionin-stained cryosection of mouse DG showing the subgranular zone (SGZ), located at the base of the granule cell layer (GCL) and adjacent to but separate from the hilus. (B) Schematic representation of the many different known cell types in the SGZ. Also shown are excitatory mossy cells in the hilus and mature granule cells. Corresponding labels share the same color as the cell type. The schematic was derived from labeled cells shown in previously published studies (Amaral, 1978; Claiborne et al., 1986; Jessberger et al., 2008; Kosaka and Hama, 1986; Lawrence and McBain, 2003; Palmer et al., 2000; Seki and Arai, 1999). (C-E) SGZ enrichment for Dcx and Sox11 corroborated by ISH (C,D) and microarrays (E). Dcx and Sox11 ISH also show enrichment in the wall of the lateral ventricle (LV) and rostral migratory stream (RMS). Insets in C,D show high magnification views of the boxed areas. In E, box and whiskers represent 25/75% and 5/95% quantiles, respectively. (F) Highly significant agreement was seen between SGZ-enriched genes found using ISH and microarrays (P<10^-35). y-axis is the density distribution (scaled average count) across paired t-test statistics for regional differential expression between SGZ and GCL (x-axis). Genes confirmed by ISH (red) show a significant skew towards SGZ enrichment relative to those not found in the SGZ using ISH (black). Vertical red and black lines show the mean of each distribution. Scale bars: 100 μm (A); 500 μm (C,D).

Fig. 2.

Genes labeling interneurons, oligodendrocytes and vascular cells in the adult SGZ. (A-C) ISH demonstrating enriched cellular expression in the SGZ compared with the overlying GCL selectively in GABAergic interneurons (A), oligodendrocytes (B) and vascular cells (C) in coronal (A,C) or sagittal (B) sections through the P56 mouse hippocampus. Inset panels show high magnification views of the GCL and SGZ corresponding to boxed regions in low magnification images. High magnification images are oriented with SGZ positioned below the GCL regardless of which blade of the dentate gyrus is represented. Scale bar: 500 μm.

Fig. 3.

Known and predicted gene markers for neurogenesis-related cell types found in the adult SGZ. (A-C) ISH demonstrating enriched cellular expression in the SGZ compared with the overlying GCL in coronal or sagittal sections through the P56 mouse hippocampus. Genes enriched in astrocytes (A) and dividing cells (B) are shown, along with other genes with diverse hippocampal expression patterns and striking SGZ enrichment (C). Inset panels show high magnification views of the GCL and SGZ corresponding to boxed regions in low magnification images. High magnification images are oriented with SGZ positioned below the GCL regardless of which blade of the dentate gyrus is represented. Scale bar: 500 μm.

Fig. 4.

Genome-wide analysis of macaque SGZ-enriched genes across postnatal development. (A) Nissl-stained coronal sections of rhesus monkey DG at 0, 3, 12 and 48 months. (B) High magnification images of the DG corresponding to boxed regions in A, showing developmental changes in the appearance and cellular makeup of the DG. MO, molecular layer; GCL, granule cell layer; SGZ, subgranular zone; PO, polymorphic layer. (C) Multidimensional scaling (MDS) plot of the top 388 ANOVA genes (P<10^-8). Dots are samples, and text is mean value in group (green G, GCL; red S, SGZ; blue P, polymorphic layer). Percentages indicate the variance explained by the first (x-axis; cell layer) and second (y-axis; age) principal components. (D) Genes marking proliferating cells (MKI67), immature neurons (DCX), astrocytes (GFAP) and interneurons (SLC32A1) show expected spatiotemporal expression patterns across time. Each dot represents the mean expression level of that gene at the labeled age point. Scale bars: 500 μm.

Fig. 5.

Good agreement between mouse and macaque markers for SGZ. (A) Highly significant agreement between rhesus monkey microarray data and the SGZ-enriched genes in mouse. Genes confirmed as SGZ-enriched by ISH in mouse (red) show a significant skew towards SGZ enrichment relative to those not found in the SGZ using ISH (black). Labeling as in Fig. 1F. (B) Many top macaque markers (t>6 in A; P<0.0001) also show significant enrichment in mouse microarray (P<0.01) and ISH. Fold changes (FC) of mean SGZ versus mean GCL expression for 25 select genes in macaque (light gray) and mouse (dark gray) microarrays are shown. Genes with red asterisks are shown by ISH in mouse in Figs 1, 2, 3, and the remaining genes are shown in supplementary material Fig. S3.

Fig. 6.

Network analysis distinguishes subclasses of SGZ-enriched genes based on developmental expression profiles. (A) WGCNA cluster dendrogram groups genes into distinct modules. The y-axis corresponds to distance determined by the extent of topological overlap (1-TO). Dynamic tree cutting identified highly parsimonious modules, generally dividing them at significant branch points in the dendrogram. (B) Top color band: Genes in the 18 modules are color-coded. Genes not assigned to a module are labeled gray. Second color band: Genes in mouse SGZ based on ISH are labeled black, and are enriched in the cyan, green-yellow, midnight-blue and tan modules (^*P<10^-11). Third color band: Plotted t-values of SGZ versus GCL samples, collapsed across all ages. Red corresponds to SGZ-enriched and green to GCL-enriched. Fourth and fifth color bands: Plotted t-values of 0-month versus 3-month, and 3-month versus 48-month samples in SGZ, respectively. Red corresponds to increased and green to decreased expression with age. (C) Module eigengene (ME) expression for the four modules showing SGZ enrichment. y-axes show ME expression (in arbitrary units). Bars represent individual samples (ordered ascending by age). Modules are labeled using selected category enrichments (see supplementary material Table S5 for complete enrichments).

Fig. 7.

Gene network experimentally confirmed as neurogenesis-related. (A) Tan module expression in SGZ correlates strongly with the number of proliferating cells across postnatal development (R=0.99, P=7.7×10^-10). Each point represents an SGZ sample for which age is labeled in months (0, 3, 12 and 48). y-axis shows module eigengene expression (in arbitrary units). x-axis shows the average number of proliferating cells in the DG of age-matched macaque monkeys from Jabès et al. (Jabès et al., 2010). The 48-month count was estimated using reported counts for earlier and later ages. (B) Network depiction of this module allows visualization of intramodular connections and hub (central) genes. Gray nodes indicate genes that are included in the mouse SGZ list. Large nodes are hub genes with 15+ of the 250 total connections. Node locations were chosen to highlight network structure and allow visibility of all gene names. (C) Nissl stain of sagittal section through control mouse at P0 showing normal hippocampal development. (D) Nissl stain of plane-matched section in a Sox4;Sox11 cdKO mouse shows severe anatomical malformations, including an undeveloped hippocampus. (E,F) Adjacent sections of control (E) and cdKO (F) mice stained for BCL11B (CTIP2) IHC. BCL11B expression is disrupted in cortex and missing in the hippocampal region of cdKO mice. Box indicates location of hippocampus. Scale bar: 1 mm.

Fig. 8.

Consistent spatiotemporal expression patterns in mouse and macaque SGZ. (A) ISH for DCX and SEMA3C in coronal sections of rhesus monkey show SGZ enrichment and decreasing expression level with increasing age, consistent with microarray data. (B) Tan module genes show consistent spatiotemporal patterning in mouse SGZ. Top row: Nissl sections of sagittally sectioned mouse DG at E18.5, P4, P14, P28 and P56. Remaining rows: Expression patterns of Dcx, Sema3c, Sox11, Sox4, Cd24a and Tnc. (C) Schematic of DG development in mouse and rhesus monkey, with time points aligned on the basis of gene expression patterns in A and B. Scale bars: 500 μm.