|Home | About | Journals | Submit | Contact Us | Français|
The dentate gyrus (DG) is one of two areas in the mature brain where stem cells reside to continuously produce new neurons throughout adulthood. While much research is focused on the DG for its roles in adult neurogenesis, little is known regarding how this key region of the brain initially develops to form its distinct architecture. We show here that the murine EphB2 receptor tyrosine kinase is critical for embryonic/postnatal development of a specific region of the DG known as the lateral suprapyramidal blade (LSB). Intracellular truncation and point mutants demonstrate that EphB2 catalytic activity is essential for LSB formation. This is consistent with expression of EphB2 in nestin-positive neural progenitor cells that migrate medially from the lateral ventricle ‘dentate notch’ neuroepithelium to populate the tertiary matrix and form the DG near the midline of the brain. Animals lacking ephrin-B1 recapitulate loss of the receptor and show that this molecule acts as the ligand to stimulate EphB2 forward signaling and direct migration of the neural progenitors into the dorsal compartment of the tertiary matrix and form the LSB. Immunoreactivity against the extracellular matrix protein Reelin in a region directly above the developing LSB is dramatically reduced when EphB2 forward signaling is disrupted. Together, these results indicate ephrin-B1 interacting with EphB2 controls the migration of dentate progenitor cells into the dorsal half of the developing DG, perhaps in part by affecting Reelin expression in a key compartment directly above the LSB.
The subgranular zone (SGZ) of the dentate gyrus (DG), located in the hippocampus, produces neural progenitors in the adult that migrate a short distance and differentiate into granule cell neurons (Abrous et al., 2005; Zhao et al., 2008). The dentate stem/progenitor cells in the SGZ reside immediately under the mature granule cell layer and can be distinguished by the expression of markers such as nestin and GFAP (Alvarez-Buylla et al., 2002; Doetsch, 2003). These cells divide to self-renew and to produce rapidly-amplifying progenitor cells, the latter of which become marked by the expression of Doublecortin (DCX). The DCX-positive cells further differentiate and migrate into the granule layer where they begin to express markers associated with mature neurons and extend mossy fiber axons and dendrites to integrate and become a functional part of the hippocampal circuitry (van Praag et al., 2002; Zhao et al., 2006; Kee et al., 2007). Given the importance of the DG for learning and memory (Li and Pleasure, 2010; Deng et al., 2010; Abrous et al., 2005; Zhao et al., 2008), it is vital that we understand the molecules that contribute to adult neurogenesis as well as those that lead to the initial formation of this stem cell niche which occurs during late embryonic and early postnatal development.
One group of molecules that participate in stem cell biology are the Eph receptor tyrosine kinases and their membrane-anchored ephrin ligands. In the adult DG, EphB1 and EphB2 are expressed on SGZ stem/progenitor cells and analysis of EphB1−/− mutants revealed ~50% reduction in the number of early-staged nestin-positive cells (Chumley et al., 2007). Loss of EphB1 expression was also found to disrupt the organization of the SGZ neurogenic niche as stem/progenitor cells in the mutant brains showed abnormal polarity and their cell bodies were often inappropriately located in the granule cell and molecular layers. However, loss of EphB1 alone did not actually alter the number of mature granule cell neurons as the volume of the DG was relatively normal in these mice. In contrast, when expression of both the EphB1 and EphB2 receptors was eliminated in compound mutants, the adult DG now exhibited a drastic decrease in volume indicating a greatly reduced number of mature granule neurons.
While the above study implicates EphB1 and EphB2 in adult neurogenesis, these receptors are also expressed in the earliest SGZ progenitor cells as they migrate during development and form the DG. We show that EphB2 tyrosine kinase-dependent forward signaling is vital for the early migration of the SGZ progenitor cells that form a specific subregion of the DG known as the lateral suprapyramidal blade (LSB). We further show ephrin-B1 acting as the ligand controls the early migration of these SGZ progenitor cells to form the LSB, and find that ephrin-B1 stimulated EphB2 forward signaling is important for the normal expression of Reelin in this subcompartment of the DG. The data suggests complex interplay between EphB forward signaling and Reelin signaling in the migration of neuronal precursors in the developing brain.
Mice used in this study include EphB2− protein-null, EphB2lacZ C-terminal truncation (Henkemeyer et al, 1996), EphB2K661R, EphB2ΔVEV, EphB2KVEV point mutants (Genander et al, 2009), nestin-eGFP (Yu et al., 2005), ephrinB1loxP (Davy et al, 2004), GFAP-cre (Zhuo et al, 2001), and Synapsin-cre (Zhu et al, 2001). These mice were maintained on a CD1 background and housed in the Animal Resource Center at the University of Texas Southwestern Medical Center at Dallas. All experiments conducted on animals were carried out according to protocols approved by the Institutional Animal Care and Use Committee.
Brain sections were obtained from both adult mice (8–10 weeks old) and embryonic mice. Adult mice were anesthetized by ketamine/xylazine, and then transcardially perfused with 4% paraformaldehyde (PFA) in PBS. Brains were removed and fixed overnight in 4% PFA. 40μm coronal sections were collected on a vibratome from agarose-embedded brains and placed in a series of 9 wells in PBS with 0.1% sodium azide. IF was performed on adult sections by incubating free-floating sections with primary antibodies overnight at 4°C. Donkey secondary antibodies conjugated with Cy2, Cy3 or Cy5 were incubated with the free-floating sections for 8 hours at RT.
For embryonic sections, pregnant female mice were sacrificed using CO2, embryos removed and anesthetized by lowering body temperature on ice. Embryos were transcardially perfused with 4% PFA in PBS. The heads were removed and fixed in 4% PFA for 4 hours, then sunk in 30% sucrose in PBS. 14μm coronal sections were prepared on a cryostat and dried overnight on slides. For IF, primary antibodies were placed on slides overnight at 4°C. Secondary antibodies were placed on slides for 3 hours at RT.
BrdU was injected intraperitoneally into pregnant females at stage E18 at a concentration of 200mg/kg. 2 hours post-injection, the female was sacrificed using CO2 administration, and the embryos removed and transcardially perfused with 4% PFA. Cryostat sections were dried on slides, post-fixed in 1% PFA in PBS for 10 min. at RT, placed in 2N HCl in PBS for 30 min. at 37°C, then placed in 0.1M Borax pH 8.5 for 10 min. at RT. The sections were then exposed to primary and secondary antibodies.
Primary antibodies used in this study were rabbit anti-green fluorescent protein (GFP, 1:500, Invitrogen), guinea-pig anti-glial fibrillary acidic protein (GFAP, 1:500, Advanced Immunochemical Inc.), goat anti-Doublecortin (DCX, 1:500, Santa Cruz Biotechnology), mouse anti-NeuN (1:1000, Chemicon), rat anti-5-bromo-2-deoxyuridine (BrdU, 1:500, Abcam), goat anti-EphB2 (1:500, R&D systems), goat anti-ephrin-B1 (1:500, R&D systems), goat anti-Sox2 (1:1000, Santa Cruz), mouse G10 anti-Reelin (1/500, de Bergeyck et al., 1998). Other staining techniques used included Nissl stains (Poly Scientific), and fluorescent neurotrace (1:1000; Invitrogen).
Stereologic volume estimates were acquired by analyzing every ninth section spanning the hippocampus from bregma −0.82 to −4.04mm, spanning the length of the DG. Sections were stained with Nissl. Measurements were performed using Stereo Investigator 9 software (MBF Bioscience) and an Olympus BX51 microscope. Volumes were estimated according to the Cavalieri principle. To define the lateral and medial suprapyramidal blade bins for each section, the total length of the suprapyramidal blade from tip to the back point of the crest was measured. This length was bisected to produce two bins of equivalent length. The same method was used to define the lateral and medial infrapyramidal blade bins.
GFP-positive cells were quantified using the optical fractionator method, where every ninth section throughout the hippocampus is examined. The DG was subdivided into 4 bins using the same method as the volume estimation protocol. GFP-positive cells in the subgranular zone were labeled with DAB, and counted using an Olympus BX51 microscope. As every ninth section was examined, the total number of GFP-positive cells counted were multiplied by nine to give an estimate of total number of cells per bin.
BrdU-labeled cells at E18 were quantified by imaging a 20x field of view in equivalent coronal sections. The embryonic DG was divided into dorsal and ventral halves, and BrdU-positive cells quantified in each half.
Data illustrated in graphs represent the mean of repeated observations. Error bars represent the SEM of those observations. All statistical analysis was performed with Prism 4 (GraphPad Software, Inc.).
Previous analysis of EphB1−/−; EphB2−/− compound null adult mice revealed a large reduction in volume of the entire DG as determined by measurement of the mature granule cell layer, and that this is associated with a diminished number of nestin-positive neural progenitors throughout the SGZ (Chumley et al., 2007). However, while the EphB1−/− single mutant did exhibit a reduction in nestin-positive cells throughout the SGZ, an obvious reduction in DG volume was not observed. This suggests that EphB1 and EphB2 receptors may share common overlapping functions in the DG, or perhaps have independent but complementary activities. To investigate potential roles for EphB2 alone, we examined the DG of adult EphB2−/− single mutant adult mice (8–10 weeks old) in Nissl stains of coronal vibratome sections and observed a fully-penetrant reduction compared to wild-type (WT) littermates in the volume of the lateral portion of the upper granule layer of the DG, an area known as the LSB (Fig. 1A, asterisk). This reduction in the LSB is specific for the loss of EphB2 expression and was not observed in other Eph single null mutants analyzed, including EphB1−/−, EphB3−/−, EphB6−/−, or EphA4−/− mice (data not shown).
We also analyzed the EphB2lacZ C-terminal truncation mutation to determine if the reduced LSB is due to loss of forward signaling. The EphB2lacZ allele produces a protein in which the majority of the intracellular segment of EphB2 is replaced by an in-frame fusion to β-galactosidase (βgal), allowing it to be expressed on the cell surface and able to bind ephrins on adjacent cells to stimulate reverse signaling (Henkemeyer et al., 1996). However, as the EphB2-βgal fusion protein lacks the tyrosine kinase catalytic domain and PDZ binding motif, it is unable to transduce forward signals that require these intracellular sequences. Examination of the DG in adult EphB2lacZ/lacZ mice showed a similar fully-penetrant reduction in numbers of granule cell neurons specifically in the LSB (Fig. 1A, asterisk). This demonstrates that the EphB2 intracellular domain is vital for normal appearance of mature granule cell neurons in the LSB.
To quantify this defect, we first measured the total volume of the DG in the EphB2 mutant mice using stereological analysis. While there was a trend towards a slightly smaller DG in the EphB2−/− and EphB2lacZ/lacZ mutants, there was no significant difference in total DG volume compared to WT (Fig. 1B). As loss of EphB2 appeared to specifically affect the LSB while leaving the remainder of the DG relatively intact, we stereologically measured the volume of 4 separate areas within the DG, the LSB (bin 1), the medial suprapyramidal blade (MSB, bin 2), the medial infrapyramidal blade (MIB, bin 3), and the lateral infrapyramidal blade (LIB, bin 4), as illustrated in Fig 1C. Stereologic volume determination revealed that the LSB in the EphB2−/− and EphB2lacZ/lacZ mutants demonstrated a 40% reduction in volume compared to WT (Fig. 1D). The other areas of the DG showed no significant change in volume compared to WT in both EphB2−/− and EphB2lacZ/lacZ mutants, demonstrating that loss of EphB2 expression only perturbs the volume of the LSB. As another way to represent this volumetric data, the proportion of total DG volume that the LSB occupies was calculated with the value for WT mice being 25%. Using this method, the proportion of LSB volume in EphB2−/− and EphB2lacZ/lacZ mutants were both determined to be less than 20% of the total DG volume (Fig. 1E).
To examine for deficits in the stem/progenitor niche of the DG, the EphB2 mutants were crossed to a nestin-eGFP transgenic mouse (Yu et al., 2005) which labels early-stage stem/progenitor cells in the SGZ. The late-stage rapidly-amplifying progenitors derived from the nestin-positive cells can be identified by DCX expression. Using immunofluorescence (IF) to detect both markers, the entire SGZ was well populated in WT mice with both GFP-positive early progenitors and DCX-positive late-stage progenitors, while the corresponding cells in the EphB2−/− and EphB2lacZ/lacZ mice appeared obviously reduced (Fig. 1F). Quantification of GFP-positive progenitors revealed a 35% reduction in the total numbers of early-stage progenitor cells in the EphB2−/− and EphB2lacZ/lacZ mutants (Fig. 1G). Quantification of total GFP-positive cells in each of the four defined bins showed that a statistically significant loss of progenitors occurred in each of the four areas in EphB2 mutants (Fig. 1H). This loss was especially pronounced in the LSB as the EphB2−/− and EphB2lacZ/lacZ mutants had only 36% and 19%, respectively, of the GFP-positive cells in bin 1 compared to the WT mice. By calculating the proportion of total GFP-positive cells that are located in the LSB, it was determined that the EphB2−/− and EphB2lacZ/lacZ mutants contained only 14% and 8%, respectively, of the total complement of early-stage neural precursors. This data demonstrates that EphB2 forward signaling is important for normal numbers of progenitor cells throughout the dentate niche, and that it is most critical for the presence of progenitors and mature granule cell neurons in the LSB.
Genetic deletion of EphB1 also resulted in an overall decrease in progenitor cell numbers throughout the DG, similar to the data presented for loss of EphB2, although without selective preference for the LSB (Chumley et al., 2007). To determine if EphB2 shared other roles with EphB1, we examine the EphB2 mutant DGs for another phenotype observed in the EphB1−/− mouse. Targeted deletion of EphB1 led to disruption of the organization of the SGZ niche, as ectopic DCX-positive cells were observed in the granule cell and molecular layers of the DG (Chumley et al., 2007). The polarity and branching of the processes of the DCX-positive cells were also often abnormal. Analysis of DCX-positive progenitors in EphB2 mutants showed that the cells exhibited normal polarity and were confined to the SGZ, as in WT mice (Fig. 1J). EphB1 and EphB2 thus appear to have distinct roles in the organization of the dentate SGZ.
The intracellular segment of EphB2 contains a number of domains that may participate in forward signaling, most notably a tyrosine kinase catalytic domain of approximately 250 amino acids and a PDZ domain binding motif at the extreme C-terminal tail. To determine if either of these domains contributes to the formation of the LSB, the DG was examined in a number of EphB2 point mutant mice where catalytic activity and PDZ binding were selectively targeted (Fig. 2A) (Genander et al., 2009). EphB2ΔVEV deletes the last three amino acids of EphB2 eliminating the C-terminal PDZ-binding motif, EphB2K661R replaces the conserved lysine at position 661 in the tyrosine kinase domain with an arginine and renders the protein catalytically inactive, and EphB2KVEV targets both the catalytic domain and PDZ binding motif of EphB2. Nissl stains of adult brains from the various mutants showed that while EphB2ΔVEV/ΔVEV mutants appeared fairly normal, both EphB2K661R/K661R and EphB2KVEV/KVEV mutants showed an obvious reduction in the DG which, like that observed for the protein-null and C-terminal truncation, was particularly apparent in the LSB (Fig. 2B, asterisks). Stereological measurements confirmed a significant reduction in total DG volume in the kinase-defective mutants, while the DG in the EphB2ΔVEV/ΔVEV mutant line did not demonstrate a significant reduction in volume compared to WT (Fig. 2C). The total volumes of the DG in the kinase-defective mutants were comparable to the EphB2−/− and EphB2lacZ/lacZ mutant lines. Focusing on the LSB, stereological analysis confirmed a significant reduction of total dentate volume in this region of the DG in the EphB2K661R/K661R and EphB2KVEV/KVEV mutants, but not in the EphB2ΔVEV/ΔVEV mutants, which were similar to WT (Fig. 2D). Together these results indicate that the tyrosine kinase catalytic activity of EphB2 is essential for normal appearance of the LSB, while the ability of this receptor to couple to PDZ domain-containing proteins is dispensable.
To determine if EphB2 kinase activity and/or PDZ interactions are important for the neural progenitor population of the LSB, the nestin-eGFP transgene was crossed with the EphB2 signaling mutants. IF showed that the LSB in the EphB2ΔVEV/ΔVEV mutant was populated with both early- and late-stage progenitors. However EphB2K661R/K661R and EphB2 KVEV/KVEV mutants showed a drastic reduction in GFP-positive and DCX-positive cells in the LSB, similar to the EphB2 protein-null and C-terminal truncated mutants (Fig. 2E). Quantification of GFP-positive cells in the EphB2 point mutants revealed a significant loss on total GFP-positive cell numbers in both the PDZ-binding deficient and tyrosine kinase catalytically inactive EphB2 mutant lines (Fig. 2F). Interestingly, when the proportion of GFP-positive progenitors found in the LSB was quantified, the EphB2ΔVEV/ΔVEV mice showed no significant change in ratios, while the tyrosine kinase catalytically inert lines demonstrated a highly significant 50% reduction compared to WT (Fig. 2G). This data demonstrates that the tyrosine kinase activity of EphB2 is essential for a normal neurogenic cell population in the regionalized zone of the LSB, while the PDZ-binding activity of EphB2 is important for the maintenance of the progenitor cell population throughout the DG.
Formation of the DG starts during late embryogenesis as dentate precursors migrate from the neuroepithelium associated with a medial compartment of the lateral ventricular zone (the dentate notch) to the tertiary matrix, the site of DG development near the midline of the brain (Altman and Bayer, 1990a, b). The data above shows that loss of EphB2 tyrosine kinase signaling has a large impact on adult DG morphology, particularly affecting the LSB. This led us to investigate whether EphB2 signaling has a role in early development of the DG. To analyze the effects of EphB2 signaling on proliferating cells within the tertiary matrix, pregnant female mice carrying late-gestation embryos (E18) were injected with BrdU, and then following a 2 hour labeling period the developing DG were examined. In WT embryos, BrdU-positive cells were observed within both the upper (dorsal) and lower (ventral) halves of the tertiary matrix, the sites of the suprapyramidal and infrapyramidal blades of the DG, respectively. In contrast, BrdU-positive cells in EphB2−/− and EphB2lacZ/lacZ mutants were observed in the lower half of the tertiary matrix, but were essentially absent from the upper half (Fig. 3A, asterisks). Quantification of the BrdU-positive cells showed a significant reduction in the number of proliferating cells in the upper half of the tertiary matrix in EphB2−/− and EphB2lacZ/lacZ mutants when compared to WT, while the lower half of the tertiary matrix was unaffected (Fig. 3B). This indicates that EphB2 forward signaling is necessary for proliferating cells to populate the developing suprapyramidal blade of the dentate tertiary matrix.
We next determined if the lack of proliferating cells in the developing suprapyramidal blade coincides with a lack of neural precursors by labeling for the nestin-eGFP transgene and another marker of neural progenitors, Sox2. IF of embryos collected at E18 showed a large population of GFP and/or Sox2 positive cells in both halves of the tertiary matrix, including the site of the developing LSB in WT animals (Fig. 3C). In contrast, EphB2−/− and EphB2lacZ/lacZ embryos had very few if any GFP or Sox2-positive cells in the upper tertiary matrix where the developing suprapyramidal blade forms (Fig. 3C, asterisks), while the lower tertiary matrix was well populated with GFP/Sox2 labeled progenitors. Interesting, while a recent study suggested that stimulation of EphB2 forward signaling in Schwann cell cultures led to increased expression/stability of Sox2 (Parrinello et al., 2010), we observed no obvious change in Sox2 IF intensity in the neural progenitors of the EphB2−/− and EphB2lacZ/lacZ mutant brains analyzed here in the tertiary matrix or elsewhere. Nevertheless, our data shows that EphB2 forward signaling is necessary for neural progenitors to populate the upper compartment of the tertiary matrix in the embryonic brain.
Previously we demonstrated that EphB2 is expressed on postnatal and adult nestin-eGFP-positive and DCX-positive progenitor cells within the SGZ of the dentate gyrus (Chumley et al., 2007). We examined the late-stage embryonic hippocampus to determine if EphB2 was also expressed on the earliest progenitor cells that originally populate the DG. As shown in Fig. 4A, IF at E18 determined that EphB2 is expressed on the progenitor cells migrating from the dentate notch near the lateral ventricle (asterisk) to the tertiary matrix (DG) near the midline of the brain. There is a high level of expression of EphB2 on cells labeled with the nestin-eGFP transgene as indicated by the overlap of IF signal (yellow), as well as on cells forming the CA3 pyramidal layer and to a lesser extent the CA1 area (red). Interestingly, EphB2 is not expressed on GFP-positive stem/progenitor cells lining the ventricle that give rise to cortical neurons (green). This indicates EphB2 is specifically expressed only on the progenitor cells that leave the dentate notch and migrate in a medial direction towards the site of DG development. To confirm the specificity of the EphB2 antibody, tissue samples of EphB2−/− mutants were similarly analyzed and showed no binding in the GFP-positive progenitors (Fig. 4B).
Expression of the EphB2 ligand, ephrin-B1, was also characterized in the developing DG during late embryogenesis. IF for ephrin-B1 at E18 revealed expression of the protein in a number of areas in the developing hippocampus, including a GFAP-positive ridge that lies immediately below the migrating nestin-GFP/EphB2 positive dentate progenitors, the dentate notch (asterisk), and in the forming CA1 and CA3 areas (Fig. 4C). Interestingly, while expression of ephrin-B1 was not noted to overlap with nestin-GFP/EphB2 positive cells streaming towards the tertiary matrix, there was obvious co-expression of ephrin-B1 and nestin-GFP (but not EphB2) in migrating cortical neurons originating from the lateral ventricles. Confirming the specificity of the antibody, IF for ephrin-B1 showed no signal in the GFAP-positive ridge in ephrin-B1−/Y mutant brain tissue and greatly reduced signal elsewhere, likely owing to some cross-reactivity of the antibody used with ephrin-B2 and/or ephrin-B3 (Fig. 4D).
Closer examination of expression within the developing hippocampus indicate EphB2 and ephrin-B1 may interact to influence the development of the DG. EphB2 and ephrin-B1 are both expressed in fairly complex patterns in the dentate notch area (Fig. 4E). While the lateral ventricle epithelium of the dentate notch only expresses nestin-GFP (asterisk), the delaminating nestin-positive cells located immediately adjacent to the dentate epithelium were observed to co-express ephrin-B1 (bracket for ephrin-B1 panels). As the dentate progenitor cells migrate closer towards the developing hippocampus, EphB2 appears to become co-expressed with ephrin-B1 in the nestin-GFP cells. Then, as the nestin-GFP/EphB2/ephrin-B1 labeled progenitors migrate closer to the tertiary matrix, the co-expression of ephrin-B1 is lost and the GFP-positive cells now only express EphB2 (bracket for EphB2 panels). Within the tertiary matrix (Fig. 4F), EphB2 continues to be expressed at high levels in the nestin-GFP stem/progenitors that now populate the developing DG, while expression of ephrin-B1 is most strongly observed in the molecular layer above the developing suprapyramidal blade (Δ). There also appears to be a focal area of overlap of EphB2 and ephrin-B1 expression at the precise point where the LSB should form, and these cells do not appear to express nestin-GFP (arrows). Unfortunately, as both the EphB2 and ephrin-B1 antibodies are made in goats, we have not been able to co-label sections to better visualize this overlap of expression of these two proteins.
To determine if ephrin-B1 is involved in development of the DG, we first analyzed an ephrin-B1 knockout mouse obtained by crossing a floxed conditional allele (Davy et al., 2004) with a transgene that expresses Cre recombinase in the germline to delete the loxP-flanked sequences and generate a protein-null allele. Since this gene is X-linked, the ephrin-B1+/− females were then crossed to WT males to generate ephrin-B1−/Y and ephrin-B1+/Y hemizygous males for analysis. Nissl stains of resulting adults revealed a greatly reduced LSB in the mutant (Fig. 5A, asterisk). Stereological measurements of serial sections indicated a small, but statistically significant reduction in total volume (WT = 1.144 ±0.031 mm3; ephrin-B1−/Y = 0.890 ±0.072 mm3; unpaired t-test, P < 0.05, n = 4 per group analyzed). There was, however, a highly significant reduction in the LSB as it occupied only 17.54 ±0.37 % of the total DG volume in the ephrin-B1−/Y mice, a much smaller percentage than the 26.5 ±0.42 % measured in the WT mice (unpaired t-test, P < 0.001, n = 4 per group analyzed). To examine the effects of loss of ephrin-B1 on the stem/progenitor cell population of the DG, ephrin-B1−/Y mice containing the nestin-eGFP transgene were generated and analyzed for both GFP and DCX expressing cells. Like the EphB2 mutant mice, there is almost a complete loss of both early-stage and late-stage progenitors in the LSB of ephrin-B1−/Y mice, while the remainder of the DG contained an obvious complement of neural progenitors (Fig. 5B). Quantification of the total number of GFP-positive cells in ephrin-B1−/Y DG revealed a trend towards fewer early-stage progenitors compared to WT (WT = 18,350 ±2,179; ephrin-B1−/Y = 11,170 ±2,037; unpaired t-test, n = 4 per group analyzed). There was, however, a highly significant reduction in the percentage of GFP-positive cells within the LSB compared to the total progenitor pool (WT = 24.41 ±0.96 %; ephrin-B1−/Y = 8.27 ±1.27 %; unpaired t-test, P < 0.001, n = 4 per group analyzed).
To investigate the developmental effect of loss of ephrin-B1 expression, a 2 hour pulse of BrdU was conducted in utero at E18. The data showed that while the upper tertiary matrix of WT embryos contained a number of BrdU-positive cells, very few proliferating cells were observed in this compartment in the ephrin-B1−/Y embryos (Fig. 5C, asterisk). Quantification of BrdU-positive cells showed a significant decrease in proliferating cells in the dorsal half of the tertiary matrix in ephrin-B1−/Y mutants, while the numbers of proliferating cells in the ventral half were not significantly affected (Fig. 5D). Likewise, analysis of nestin-eGFP and Sox2 positive cells at E18 showed reduced numbers of neural progenitors in the developing suprapyramidal blade of ephrin-B1−/Y embryos (Fig. 5E and F, asterisks). As with the EphB2 mutants, no obvious change in Sox2 IF intensity was noted in the remaining neural progenitors of ephrin-B1−/Y mutant brains. Interestingly, GFAP expression surrounding the tertiary matrix appeared to be disrupted in the ephrin-B1−/Y mutants, particularly in the dorsal half of the developing DG. The phenotype observed in the DG of mice lacking ephrin-B1 is very similar to mice lacking EphB2 kinase-dependant forward-signaling, demonstrating that an ephrin-B1:EphB2 signaling partnership is vital for the correct formation of the DG.
Our analysis of the expression pattern of ephrin-B1 as shown in Fig. 4 suggests that ephrin-B1 is expressed in GFAP-positive cells, including a GFAP-positive ridge directly underneath the EphB2/nestin-GFP-positive progenitor cells that are migrating towards the tertiary matrix. To investigate the activity of ephrin-B1 in GFAP-positive cells, we crossed conditional ephrin-B1loxP/loxP female mice with GFAP-cre (Zhuo et al., 2001) transgenic males to eliminate ephrin-B1 expression in GFAP-positive cells while leaving it intact in other populations in male offspring. As a control, ephrin-B1loxP/loxP females were also crossed to Synapsin-cre (Zhu et al., 2001) transgenic males to delete ephrin-B1 in mature neurons. As shown in Fig. 6A, a reduced LSB was observed in the GFAP-cre;ephrin-B1loxP/Y combination (asterisk), while the Synapsin-cre;ephrin-B1loxP/Y brains appeared similar to the ephrin-B1loxP/Y controls that did not receive a Cre driver. Stereological analysis of total volume of the dentate in these mice revealed a significant loss in the GFAP-cre;ephrin-B1loxP/Y DG compared to controls (Fig. 6B). Furthermore, the LSB was particularly affected in the GFAP-cre;ephrin-B1loxP/Y combination, demonstrating an equivalent reduction in proportion of total dentate volume to the ephrin-B1−/Y protein-null mutants (Fig. 6C).
To determine if the LSB reduction observed in GFAP-cre;ephrin-B1loxP/Y mice was due to a developmental defect, we examined the developing DG at E18 for expression of Sox2 positive progenitors (Fig. 6D). The developing suprapyramidal blade in the GFAP-cre;ephrin-B1loxP/Y hippocampus showed a noticeable deficit of Sox2 positive progenitor cells (Fig. 6D, asterisk), demonstrating that the GFAP-cre;ephrin-B1loxP/Y and ephrin-B1−/Y DG share a similar developmental defect. In both the Synapsin-cre;ephrin-B1loxP/Y and ephrin-B1loxP/Y brains, this area was well-populated with Sox2 progenitor cells.
To examine in closer detail the activity of the cre drivers, we crossed a Rosa-YFP transgene into the cre backgrounds, where YFP is preceded by a stop codon flanked with loxP sites. We then examined the cre lines for YFP and ephrin-B1 expression at E18. Extensive expression of YFP was observed throughout the developing hippocampus and cortex of the GFAP-cre;ephrin-B1loxP/Ymice, indicating a number of cell populations contained active cre (Fig. 7). Expression of YFP in the Synapsin-cre;ephrin-B1loxP/Ywas much more restricted in the embryonic brain. Ephrin-B1 expression (red) in the Synapsin-cre;ephrin-B1loxP/Y hippocampus appeared very similar to the ephrin-B1loxP/Y controls. However, loss of ephrin-B1 expression was observed in the GFAP-cre;ephrin-B1loxP/Y brains throughout the ventricular proliferative zone, including the dentate notch (asterisk), and in the hippocampus, including the molecular layer above the tertiary matrix (Δ). Of particular interest, loss of ephrin-B1 expression was not observed in the GFAP-positive ridge in the GFAP-cre;ephrin-B1loxP/Y mice, indicating that ephrin-B1 expression here does not contribute to formation of the LSB.
The activity of the GFAP-cre driver illustrated in Fig. 7 indicates that there are two possible areas of ephrin-B1 expression that may influence the development of the LSB, namely the molecular layer above the LSB and the ventricular proliferative zone near the dentate notch. The molecular layer has been previously shown to influence DG development. This area, also known as the marginal zone, contains Cajal-Retzius cells that secrete Reelin, an extracellular matrix protein important for neuronal migration. As loss of Reelin has significant effects on DG morphology (Forster et al., 2002; Frotscher et al., 2003), we investigated its expression within the developing hippocampus and compared to EphB2 and ephrin-B1 expression patterns. We found that Reelin is expressed above the developing suprapyramidal blade in WT mice at E18 and is closely associated with both EphB2 and ephrin-B1, particularly where the LSB will form (Fig. 8A). EphB2 is intensely expressed below the band of Reelin expression, with some co-expression of EphB2 and Reelin observed, while ephrin-B1 is expressed above the band of Reelin expression.
We next examined Reelin in EphB2 and ephrin-B1 mutant brains collected at E18 and noted an obvious reduced expression in cells directly above where the LSB should form (Fig. 8B). This decrease in Reelin expression corresponds with the decrease in Sox2 and nestin-GFP positive progenitors occupying this area in the mutants. The observation of reduced Reelin expression in the EphB2lacZ/lacZ brain demonstrates that EphB2 forward signaling is necessary for robust expression of Reelin in the area of the hippocampus that gives rise to the LSB.
The formation of the DG and the molecules involved in this developmental process are beginning to be illuminated. Here we show that Eph-ephrin signaling contributes to the early development of this structure, as EphB2 tyrosine kinase-dependant forward signaling is required for the migration of neural progenitors into the dorsal region of the tertiary matrix to form the LSB. We demonstrate that ephrin-B1 acting as the ligand is also necessary for correct formation of the LSB, and that these molecules likely regulate the formation of the DG at least in part by influencing expression of Reelin in the marginal zone area directly above where the LSB will form.
Reelin is an extracellular matrix molecule expressed by Cajal-Retzius cells located in the marginal zone of the hippocampus (Del Rio et al., 1997). Loss of Reelin has drastic effects on the structure of the DG as observed in Reeler mutant mice, at least in part by controlling the formation of the radial glial scaffold of the developing DG (Forster et al., 2002; Frotscher et al., 2003). Reelin has also been shown to determine aspects of the migration of DG progenitors from the ventricle to the developing DG (Li et al., 2009). Interpreting the phenotypes observed in the EphB2 and ephrin-B1 mutants, Reelin is likely acting as an attractant to neural progenitors in the tertiary matrix, and loss of Reelin above the dorsal half of the developing DG leads to a failure of neural progenitors to migrate into this area. Interestingly, the effects on DG structure observed in EphB2/ephrin-B1 mutants compared to Reeler mice are strikingly different. Granule neurons in Reeler mice fail to coalesce into densely packed layers and are loosely distributed throughout the developing DG (Drakew et al., 2002), whereas granule neurons in EphB2 forward signaling mutants are selectively absent in the LSB and remain tightly packed throughout the remainder of the DG. This discrepancy may be explained by the observation that Reelin expression is specifically lost above the LSB in EphB2/ephrin-B1 mutant mice, and is still expressed surrounding the other regions of the developing DG.
Although our data strongly suggests that Eph-ephrin signaling in DG development is of primary importance immediately surrounding the developing LSB, we cannot rule out a role for these signaling events at the ventricular proliferative zone. Ephrin-B1 is expressed at a high concentration in the ventricular zone where both cortical and DG neuron precursors arise (Fig 4). In an early study involving thymidine labeling during embryonic development of the hippocampus, it was demonstrated that proliferating DG neuron progenitors arise in an area of the lateral ventricles termed the primary dentate neuroepithelium, otherwise known as the dentate notch (Altman and Bayer, 1990a). This area transitions into a secondary matrix, which produces proliferative cells that migrate medially from this ventricular zone subregion to the site of DG formation, termed the tertiary matrix. Previously, it was demonstrated that ephrin-B1 expressed in progenitor cells in the cortical ventricular zone is involved with the maintenance of the progenitor pool in cortical neurogenesis (Qiu et al., 2008). This function was linked to ephrin-B1 reverse signaling, which does not play a role in the DG phenotype described here, as we have analyzed ephrin-B1 reverse signaling mutants (Bush and Soriano, 2009) and did not observe any abnormalities in LSB morphology (data not shown). As the DG neuron progenitors leave the secondary matrix and migrate in a medial direction towards the tertiary matrix, ephrin-B1 expression is lost and EphB2 expression is observed. These EphB2 positive cells then migrate to the tertiary matrix. DG granule layer development occurs in a gradient beginning at the crest and expanding in a lateral direction over time (Altman and Bayer, 1990b). Our analysis of EphB2 and ephrin-B1 mutant mice demonstrated that disruption of EphB2 forward signaling did not lead to a reduction in progenitor numbers in the ventral half of the tertiary matrix, indicating a role in the migration of progenitor cells from the ventral to the dorsal half of the tertiary matrix, and not in the migration from the ventricular zone to the tertiary matrix.
This study has revealed a particularly interesting aspect of Eph-ephrin signaling in the hippocampus, as EphB1 and EphB2 appear to have both overlapping and separate roles in the DG. Both EphB1 and EphB2 control progenitor number throughout the SGZ in the adult brain, and we show that this activity is linked to the PDZ-binding ability of EphB2. However, only loss of the tyrosine kinase activity of EphB2 affected the number of mature granule cell neurons, a phenotype not observed in the EphB1−/− mutant mice (Chumley et al., 2007). EphB2 plays a major role in development of the DG as the formation of the LSB is disrupted in the embryonic brain. EphB1 does not appear important in the early development of the DG, as genetic deletion of EphB1 does not affect the DG volume. Interestingly, both EphB1 (Chumley et al., 2007) and EphB2 (this study) are expressed on migrating neural progenitors as they migrate to the site of DG development. While we have shown that EphB1 and EphB2 are expressed in the embryonic DG neural progenitors, we have not established levels of expression at the cell membrane. EphB2 could be expressed at a much higher level than EphB1, making it more influential in the early development of the DG than EphB1. Alternatively, EphB1 and EphB2 may transduce somewhat different forward signals to bring about their distinct effects on the migration and proliferation of dentate precursors. This latter possibility has foundations in other studies that show the intracellular signaling domains of EphB1 and EphB2 have very different abilities to mediate the ipsilateral routing of retinal ganglion cell axons despite a high degree of sequence identity between the two receptors (Petros et al., 2009).
The fact that such a specific part of the DG is affected while the remainder of the structure remains relatively normal could have specific effects on hippocampal function. A number of studies have shown a separation of function between the dorsal and the ventral DG. The dorsal DG has been linked to spatial learning and memory (Hunsaker and Kesner, 2008), while the ventral DG has been linked to behavior correlated with anxiety (Eadie et al., 2009). The structure of the adult DG in EphB2 and ephrin-B1 mutants suggests that these mice should have spatial information processing deficits, and yet anxiety behavior should remain relatively unaffected.
As well as the demonstrated effects on Reelin expression, Eph-ephrin signaling may affect DG morphogenesis via other signaling pathways that have been shown to control the development of the DG (Li and Pleasure, 2007). Elimination of the transcription factors Sox2 or neurogenin 2 in neural progenitors greatly decreases the size of the DG (Galichet et al., 2008; Favaro et al., 2009). Loss of Draxin, a protein linked to repulsive axon guidance functions, was also shown to cause a decrease in size of the entire DG (Zhang et al., 2010). Mice lacking the chemokine receptor CXCR4 and the ligands SDF-1 and Cxcl12 were shown to greatly decrease the number of cells that populate the DG (Lu et al., 2002; Li et al., 2009). Likewise, sonic hedgehog (Shh) signaling is also required for formation of the dentate stem cell niche (Machold et al., 2003). Other signaling molecules have been shown to specifically affect the radial glial scaffold, such as Wnts (Galceran et al., 2000; Zhou et al., 2004). Our analysis of ephrin-B1 knockout mutant DGs at E18 shows that the GFAP expression around the developing LSB is disrupted. However, unlike Wnt signaling mutants, the majority of the DG is still able to form correctly in the EphB2 and ephrin-B1 mutants with only the LSB affected, indicating that Eph-ephrin and Wnt signaling have separate roles in DG morphogenesis. Interestingly, these two signaling pathways have been linked in regulating the stem cell niche in the intestinal epithelium (Batlle et al., 2002; Holmberg et al., 2006), suggesting the possibility that Wnt and Eph-ephrin signals are common features that dictate stem cell migration and proliferation throughout the body.
It remains to be determined exactly how EphB2 kinase-dependant forward signaling affects Reelin expression in the MZ. We observed some co-expression of Reelin and EphB2 in Cajal-Retzius cells, suggesting that EpB2 forward signaling may play a cell-autonomous role in the expression and secretion of Reelin once activated by the ephrin-B1-positive cells surrounding the MZ. Alternatively, EphB2 activity may determine the migration of Cajal-Retzius cells into the MZ, and loss of forward signaling would result in fewer numbers of Cajal-Retzius cells above the LSB. Interestingly, we show that EphB2 and ephrin-B1 are expressed in opposing gradients surrounding Reelin, with EphB2 highly expressed below and ephrin-B1 highly above the observed Reelin expression in the MZ. Presumably, Eph-ephrin signaling is strongest at the point where these gradients overlap, immediately next to the cluster of Reelin above the developing LSB. This data suggests that Reelin expression may require a gradient of Eph-ephrin signaling strength.
We thank Steve Kernie for the nestin-eGFP transgenic mice, Alice Davy and Phil Soriano for the ephrin-B1loxP conditional mutant mice, Albee Messing for GFAP-cre transgenic mice, Jamey Marth for Synapsin-cre transgenic mice, Joachim Herz for the Reelin antibody, Franny Prince for genotyping, and Stacey Mohammadie for technical assistance. This research was supported by the NIH (R01 MH66332).
Conflict of interest: The authors declare no conflicts of interest or financial interests.