Prox1 Is Required for Granule Cell Maturation and Intermediate Progenitor Maintenance During Brain Neurogenesis

The transcription factor Prox1 plays a crucial role in intermediate progenitor maintenance and hippocampal neuron differentiation during adult neurogenesis in the dentate gyrus region of the hippocampus.


Introduction
In the brain, the dentate gyrus (DG) is the primary afferent pathway into the hippocampus. The DG has a crucial role in learning and memory [1,2,3]. In mammals, neurogenesis occurs in the subgranular zone (SGZ) of the DG throughout adulthood [4,5,6,7]; this activity is thought to be the basis for the acquisition of new memories [3,8,9].
The formation of the DG is a complex process that involves cell migration and neuronal differentiation [10,11]. Factors that regulate DG development are thought to have a similar function during adult neurogenesis. In the SGZ, astrocyte-like adult neural stem cells (NSCs) give rise to a series of intermediate progenitors that eventually differentiate into neurons [12]. Several signaling molecules, including Wnt, Noggin/BMP, Shh, and Notch, regulate adult NSC self-maintenance, proliferation, and progenitor differentiation [13,14]. However, little is known about how the generation of the proper number of descendants is controlled. It has been proposed that once generated, NSC descendants can trigger some type of feedback mechanism to stop stem cell differentiation [15]. In this context, Notch signaling has been considered a candidate to regulate such a feedback mechanism during adult neurogenesis [13].
The homeobox gene Prox1 is expressed in several brain regions (i.e., cortex, DG, thalamus, hypothalamus, cerebellum) during prenatal and postnatal stages of development [16,17,18]. Interestingly, Prox1 is expressed throughout all stages of DG development and in adult granule cells; therefore, Prox1 is commonly used as a specific marker for these cells [15,19]. However, no data are yet available on the functional role(s) of Prox1 during brain development.
We have now determined that functional inactivation of Prox1 during DG development results in defective granule cell maturation and the loss of this cell population. We also report that conditional inactivation of Prox1 in the SGZ during adult neurogenesis leads to the lack of intermediate progenitors, and as a consequence, the disruption of the mechanism involved in NSC self-maintenance. Therefore, we have identified a previously unknown non-cell autonomous regulatory feedback mechanism that links adult NSC self-maintenance with the generation of the proper number of descendants in the SGZ. Finally, we show that ectopic expression of Prox1 in NSCs promotes premature differentiation during DG development and adult neurogenesis in the SGZ. [22]. Adult Nestin-Cre;Prox1 F/F mice were viable but had only a few scattered Prox1 + /NeuN + wild-type granule cells in their hippocampi ( Figure 1B,D). During embryonic development, Prox1 expression is detected in both the dentate neuroepithelium (DNE) and the DG [17,23]. Therefore, we performed a detailed characterization of the development of the DG in Nestin-Cre;Prox1 F/F conditional-mutant embryos.
At E14.5, the DG of Nestin-Cre;Prox1 F/F embryos showed normal Ammon's horn formation (Figure 2A-L). At E16.5, and as shown by an anti-C-Prox1 antibody that recognizes only the wild-type form of Prox1 (see Figure S1A,B for more details), only a few cells escaped Cre-mediated deletion in the DG of Nestin-Cre;Prox1 F/F embryos ( Figure 1F). However, as indicated by an anti-N-Prox1 antibody that recognizes both wild-type and conditional-mutant forms of Prox1 ( Figure S1A,B), the number of N-Prox1 + cells was reduced in the DG of Nestin-Cre;Prox1 F/F embryos at this stage ( Figure 1H; Figure S1C,D). Reduced numbers of Notch1 + , Ngn2 + , and NeuroD1 + cells were also observed by ISH in the mutant DG ( Figure 1J,L,N); however, we found no obvious alterations in the expression of Wnt3a in the FNE ( Figure S2A,B) [24], or lef1 in the migratory stream (MS) ( Figure S2C,D) [25]. We also did not find obvious anomalies in radial glia scaffolding [26] in the mutant DG at E16.5 ( Figure S2E-H). These results indicate that in the Nestin-Cre;Prox1 F/F embryos, the N-Prox1 + cells are capable of migrating out of the DNE and reach the DG.

Prox1 Controls Cell Proliferation in the DNE
Prox1 absence leads to cell cycle alterations in the developing neuroretina [27]. To evaluate possible alterations in cell proliferation during DG development, we compared the number of cycling cells in the DNE of wild-type and Nestin-Cre;Prox1 F/F embryos by using a 1-h BrdU pulse starting at E14.5. Significantly fewer BrdU + cells were observed in the Nestin-Cre;Prox1 F/F DNE at E16.5 ( Figure 3D,G) and E18.5 ( Figure 3F,G). Similar results were obtained with Ki67 ( Figure 3E,H). Results using BrdU/Ki67 double immunostaining ( Figure 3I) determined that cells in the DNE were cycling more slowly in Nestin-Cre;Prox1 F/F embryos at E16.5 and E18.5. We also found that CyclinE expression was reduced in the Nestin-Cre;Prox1 F/F DNE at E16.5 ( Figure 3K). To confirm that the reduced proliferation of cells in the DNE was caused by the lack of Prox1 in that region at around E16.5, we used a tamoxifen (TM)-inducible Nestin-CreER T2 strain [28] to induce Prox1 deletion later during development. Following TM administration at E16.5 (see Figure S3A for details), most C-Prox1 + cells were missing in the DNE and MS of E18.5 Nestin-CreER T2 ;Prox1 F/F embryos ( Figure 3M). Fewer Ki67 + cells were also observed in the DNE of these mutant embryos ( Figure 3N). Thus, Prox1 is required to regulate cell proliferation in the DNE during early stages of DG development.

Granule Cell Maturation Is Arrested in the Absence of Prox1
During DG formation, Prox1 expression is upregulated in intermediate progenitors and immature granule cells, and throughout adulthood, its expression is maintained in mature granule cells [17,23]. Therefore, we analyzed whether Prox1 is necessary for the acquisition of granule cell identity by monitoring the D-Prox1 cells. Using this approach, we observed an increase in the number of D-Prox1 cells (cells recognized by the anti-N-Prox1 antibody but not by the anti-C-Prox1 antibody; see Figure S1A,B for details) in the DG region of Nestin-Cre;Prox1 F/F embryos until E18.5 ( Figure S1D,F and Figure 4E); however, at later stages, the number of D-Prox1 cells in the DG decreased. This reduction was particularly apparent at postnatal stages; by postnatal day (P) 15, most Prox1 + cells in the mutant DG corresponded to those that escaped deletion ( Figure 4B,D,E). As revealed by TUNEL assay, starting at E16.5 the number of TUNEL + cells increased in the DG of Nestin-Cre;Prox1 F/F embryos ( Figure 4F), a finding suggesting that Prox1 is necessary for the survival of intermediate progenitors and immature granule cells.
In the DG, terminal differentiation of granule cells occurs during early postnatal stages. Therefore, we analyzed whether the observed increase in the number of TUNEL + cells was due to defective differentiation of D-Prox1 granule cells. Previous studies have shown that the bHLH protein NeuroD1 is required for the maturation of granule cells [15,29,30]. We found that at P10, N-Prox1 + cells in the mutant DG were also NeuroD1 + ( Figure 4H); however, they did not co-express Dcx + [31] or Calretinin + [32] ( Figure 4J,L), and only the few C-Prox1 + granule cells that escaped Cre deletion co-expressed NeuN ( Figure 4N) [33]. These results suggest that lack of Prox1 activity arrested granule cell differentiation.
To determine whether Prox1 is necessary for Dcx expression, we cotransfected Nestin-Cre;Prox1 F/F neurospheres with GFP-expressing and full-length Prox1 cDNA-expressing plasmids. We found that Nestin-Cre;Prox1 F/F neurospheres transfected with Prox1 cDNA produced Dcx + cells (20 of 76 GFP + neurospheres), but those transfected only with the GFP-expressing plasmid did not (0 of 73

Author Summary
In the brain, the hippocampus has a crucial role in learning and memory. In mammals, neurogenesis (the birth of new neurons) occurs in the dentate gyrus region of the hippocampus throughout adulthood, and this activity is thought to be the basis for the acquisition of new memories. In this study we describe for the first time the functional roles of the transcription factor Prox1 during brain development and adult neurogenesis. We demonstrate that in mammals, Prox1 is required for the differentiation of granule cells during dentate gyrus development. We also show that conditional inactivation of Prox1 results in the absence of specific intermediate progenitors in the subgranular zone of the dentate gyrus, which prevents adult neurogenesis from occurring. This is the first report showing blockade of adult neurogenesis at the level of progenitor cells. Next, we demonstrate that in the absence of Prox1-expressing intermediate progenitors, the stem cell population of the subgranular zone becomes depleted. Further, we show that Prox1-expressing intermediate progenitors are required for adult neural stem cell self-maintenance in the subgranular zone. Finally, we demonstrate that Prox1 ectopic expression induces premature granule cell differentiation in the subgranular zone. Therefore, our results identify a previously unknown non-cell autonomous feedback mechanism that links adult stem cell self-maintenance with neuronal differentiation in the dentate gyrus and could have important implications for neurogenesis in other brain regions. GFP + neurospheres) ( Figure S4G,H). Thus, as seen in vivo, Prox1 is not necessary to induce neuronal differentiation in vitro but is required for the expression of later neuronal markers such as Dcx.

Prox1 Is Required for Granule Cell Maturation at Postnatal Stages
To confirm that the conditional inactivation of Prox1 at postnatal stages is directly responsible for the defective granule cell maturation phenotype observed in Nestin-Cre;Prox1 F/F mice, we next deleted Prox1 during postnatal stages by using the Nestin-CreER T2 transgenic mouse strain [28]. TM was administered to Nestin-CreER T2 ;Prox1 F/F pups at early postnatal stages (see Figure  S3B for details); at these stages, there are no Prox1 + NSCs in the hilus ( Figure S5A-E). At 2 mo of age, TM-treated Nestin-CreER T2 ;Prox1 F/F mice had smaller DGs than their control littermates ( Figure 5B). The number of Tbr2 + intermediate progenitors [34] was reduced in both P5 ( Figure S5R,S) and P10 ( Figure 5P,Q) pups. They also exhibited fewer Dcx + cells at P5 ( Figure S5T,U) and P10 ( Figure 5R,S). Moreover, the ratio of  Figure 4T-V) and active Caspase-3 assays ( Figure S6A,B), an increase in cell death was observed in the mutant DG at these stages. On the other hand, cell proliferation was reduced at P5 but not at P10 in the DG of Nestin-CreER T2 ;Prox1 F/F pups ( Figure S5P; Figure 5N). The number of Nestin + , Sox2 + , or Id1 + NSCs was normal at P5 and P10 ( Figure S5J,M; Figure 5C-K). These data suggest that the lack of Prox1 during postnatal stages is directly responsible for the reduction in the number of intermediate progenitors and the defect in granule cell differentiation that secondarily promotes an increase in cell death and ultimately a reduction in the size of the DG.

Prox1 Is Required for the Maintenance of Intermediate Progenitors in the SGZ During Adult Neurogenesis
In the SGZ, Prox1 expression is initially detected in Tbr2 + ( Figure 6E) [35] and Dcx + ( Figure 6F) Type-IIb intermediate progenitors [36] and is absent from adult NSCs ( Figure 6A,B,C) [37,38] and Ascl1 + intermediate progenitors ( Figure 6D) [36]. As neurogenesis progresses, Prox1 is detected in Calretinin + cells ( Figure 6H). Analysis of the SGZ of Nestin-CreER T2 ;Prox1 F/F pups treated with TM daily from P0 to P15 (see Figure S3B for the details of the TM treatment) identified a reduced number of Tbr2 + ( Figure 6K) and Dcx + ( Figure  6N) cells at P20; these cells were nearly absent at 4 and 8 mo of age ( Figure 6J,K,M,N). As indicated by C-Prox1 immunostaining, the few remaining Dcx + cells were those that escaped deletion ( Figure S7). Fewer Calretinin + immature neuron cells were also observed in the mutant brains ( Figure 6P,Q). To determine whether the lack of intermediate progenitors in adult stages is a direct or indirect result of previously identified alterations in postnatal DG development, we treated 8-wk-old (adult) Nestin-CreER T2 ;Prox1 F/F mice with TM 3 d/week (see Figure  S3C for the details of the TM treatment). Four weeks after the initial TM induction, the number of Tbr2 + ( Figure 7A-C) and Dcx + ( Figure 7D-F) cells was reduced in the SGZ of 12-wk-old Nestin-CreER T2 ;Prox1 F/F mice. Accordingly, the number of Calretinin + cells was also reduced ( Figure 7G-I). Similar results were observed 8 wk after induction (in 16-wk-old animals) ( Figure 7A,D,G).
To evaluate whether the lack of Prox1 activity leads to an increase in cell death, as it does during developmental stages, we performed double TUNEL immunohistochemistry in 12-and 16wk-old mice after TM induction. At both time points, we observed an increased number of TUNEL + Tbr2 + ( Figure 7J,K,M) and TUNEL + Dcx + ( Figure 7J-L) cells in the Nestin-CreER T2 ;Prox1 F/F SGZ, a result indicating that Prox1 is required for the survival of Tbr2 + and Dcx + cells.
To evaluate whether the lack of Tbr2 + intermediate progenitors and Dcx + cells affects adult neurogenesis in the SGZ, we performed BrdU labeling 15 d prior to collection of the brains for analysis. Fewer BrdU + newborn cells were observed in the SGZ of 12-wk-old ( Figure 7N-P) and 16-wk-old ( Figure 7Q-S) Nestin-CreER T2 ;Prox1 F/F mice. This result suggests that the lack of Prox1 in the SGZ affects adult neurogenesis.
Next, to determine whether the defective neurogenesis in the SGZ is solely because of a reduction in the number of intermediate progenitors or it is also because of an alteration in immature neurons, we analyzed the ratio of Tbr2 + :Calretinin + cells in the SGZ of 12-wk-old and 16-wk-old Nestin-CreER T2 ;Prox1 F/F mice. A consistent increase in the ratio of Tbr2 + :Calretinin + cells was observed in 12-wk-old (N = 3) (2.3960.58:1) and 16-wk-old (N = 3) (4.3961.92:1) TM-treated Nestin-CreER T2 ;Prox1 F/F mice (12-wkold control: 1.7760.34:1; 16-wk-old control: 1.5160.17:1; p value for both cases was p,0.1). This result suggests that granule cell maturation is also affected by the lack of Prox1 in the SGZ during adult neurogenesis.
We next examined whether the induction of neurogenesis rescues the reduction in the number of progenitor cells observed in the SGZ of adult Nestin-CreER T2 ;Prox1 F/F mice. At 12 wk of age, TM-induced and control mice were treated with kainic acid (KA), a compound that induces adult hippocampal neurogenesis (see Figure S3D for the details of the TM/KA treatment) [39,40]. Eight days after KA administration, the brains of control mice treated with TM and KA or only with KA showed an increased

Lack of Intermediate Progenitors Results in Defective Self-Maintenance of Adult Neural Stem Cells
It has been proposed that upon their differentiation, intermediate progenitors trigger a feedback mechanism necessary to stop NSC differentiation and support NSC maintenance [13,15]. Therefore, the lack of intermediate progenitors observed in the SGZ of Nestin-CreER T2 ;Prox1 F/F mice may impinge on this feedback mechanism and ultimately affect the number of adult NSCs produced in this region.
To assess whether the reduced number of intermediate progenitors in the SGZ of Nestin-CreER T2 ;Prox1 F/F mice treated postnatally with TM affected that of NSCs, we compared the number of Type-I NSCs in wild-type and Prox1 conditionalmutant brains. We detected no difference in the number of Nestin + , Sox2 + , or Id1 + adult NSCs at P20 (Figure 8A,D,G). However, we unexpectedly observed fewer Nestin + , Sox2 + , or Id1 + NSCs in the SGZ of Nestin-CreER T2 ;Prox1 F/F mice older than 2 mo ( Figure 8A-I). We also observed a similar reduction in the number of Nestin + Gfap + Blbp + NSCs in the SGZ of 4-mo-old Nestin- Figure S9). This result argued that the absence of intermediate progenitors observed in the SGZ of Nestin-CreER T2 ;Prox1 F/F brains leads to a non-cell autonomous reduction in the number of adult NSCs.
To determine whether defective maintenance of adult NSCs is a direct result of the lack of intermediate progenitors or an indirect result because of alterations in DG development, we treated 8-wkold control and Nestin-CreER T2 ;Prox1 F/F mice with TM (see Figure  S3C for the details of the TM treatment) and counted the number of adult NSCs. Four weeks after the beginning of the treatment, the number of Nestin + , Sox2 + , or Id1 + adult NSCs was higher in the SGZ of Nestin-CreER T2 ;Prox1 F/F mice than in wild-type littermates ( Figure 8J,M,P); however, 8 wk after TM treatment, we detected fewer Nestin + , Sox2 + , or Id1 + adult NSCs in the SGZ of Nestin-CreER T2 ;Prox1 F/F mice ( Figure 8J-R).
To examine whether Prox1 has a direct role in this defective maintenance of adult NSCs, we generated Nestin-CreER T2 ;Prox1 F/ F ;R26R mice and treated them with TM (3 d/wk) starting at 8 wk of age (see Figure S3C for the details of the TM treatment). After 4 wk of treatment, the ratios of Nestin + b-gal + cells and Sox2 + bgal + cells were the same in wild-type and Prox1 conditional-mutant mice ( Figure 8S). This result indicates that the absence of Prox1 did not directly affect the adult NSC population. Similar results were seen in mice analyzed 8 wk after the beginning of TM administration ( Figure 8T). Only the Tbr2 + b-gal + and the Dcx + bgal + cell populations were significantly reduced in the SGZ of Nestin-CreER T2 ,Prox1 F/F ;R26R mice at 4 and 8 wk after the beginning of TM treatment ( Figure 8S,T). These results, together with the absence of increased apoptosis in adult NSCs  ( Figure 7J,K), suggest that the defective maintenance of adult NSCs in the mutant SGZ is indirectly caused by the absence of intermediate progenitors.
We next investigated the possible mechanisms of this intermediate progenitor-dependent control of adult neurogenesis in the SGZ. In the developing telencephalon [41] and in the SGZ [42], Notch signaling is necessary for NSC maintenance, and the expression of Jagged1, a Notch receptor ligand, is restricted to intermediate progenitors in the SGZ [42]. We found a reduced number of Jagged1 + cells in the SGZ of Nestin-CreER T2 ;Prox1 F/F mice treated with TM starting at 8 wk of age and analyzed 4 wk later ( Figure 9C,D). To examine the relationship between the number of Nestin + Type-I cells in the SGZ and active Notch signaling, we performed Hes1 immunohistochemistry. In the SGZ of control mice, only 17% of the Type-I cells were Nestin + Hes1 -( Figure 9F); however, in the Nestin-CreER T2 ;Prox1 F/F mice, this number increased to 55% ( Figure 9F). The number was also higher in the SGZ of Nestin-CreER T2 ;Prox1 F/F mice (59%) when compared with wild-type controls (25%) at 16 wk of age ( Figure 9G). A reduced number of Hes5-expressing cells was also observed in the SGZ of Nestin-CreER T2 ;Prox1 F/F mice at 16 wk of age ( Figure S10). These results show that in the absence of intermediate progenitors, the proportion of adult NSCs with active Notch signaling at 12 and 16 wk is reduced. This mechanism may explain why the self-maintenance of adult NSCs is eventually overruled at 16 wk of age.

Ectopic Expression of Prox1 in NSCs Induces Their Premature Differentiation
Our results showed that Prox1 is necessary for granule cell differentiation and intermediate progenitor maintenance. Therefore, we next determined whether Prox1 activity is sufficient to induce granule cell differentiation. To do this, we generated a new mouse transgenic line in which Prox1 expression was under the control of the ubiquitous CMV promoter (CMV-CAG-loxP-eGFP-Stop-loxP-Prox1-Ires-b Gal; JoJo-Prox1 for brevity) [43]. Crossing this line with any available Cre strain will lead to the release of the Stop-signal cassette and the transcription of Prox1 in a tissuespecific manner. We used this novel strain to generate adult Nestin-Cre;JoJo-Prox1 mice that ectopically expressed Prox1 in several brain regions ( Figure S11). Importantly, we found that the number of granule cells was reduced in the DG of these mice ( Figure 10A,B,V).
Prox1 was ectopically expressed in the ventricular, subventricular, and mantle zones of several brain regions of Nestin-Cre;JoJo-Prox1 embryos ( Figure S11), including the hippocampal neuroepithelium and the hippocampal field ( Figure 10D). However, we observed no change in the number of Prox1 + (Figure 10E), Sox2 + ( Figure 10F), or Ki67 + cells ( Figure 10G) in the DNE or DG. These results suggest that ectopic expression of Prox1 in embryonic Nestin-expressing neuroepithelium is not sufficient to affect embryonic NSC differentiation.
To perform a similar analysis during postnatal stages, we administered TM to Nestin-CreER T2 ;JoJo-Prox1 pups daily starting at P0 (see Figure S3B for the details of the TM treatment). At P8, the DG of Nestin-CreER T2 ;JoJo-Prox1 pups exhibited an increased number of Dcx + ( Figure 10I) and NeuN + (Figure 10L) cells and a reduced number of Nestin + cells ( Figure 10N). At this stage, proliferation was already reduced in the DG of Nestin-CreER T2 ;-JoJo-Prox1 mice, as shown by Ki67 staining ( Figure S12G).
To determine the fate of cells that ectopically expressed Prox1, we performed double immunohistochemistry using antibodies against b-gal and Nestin, Dcx, NeuN, Gfap, or NG2 (Figure S12I-M). We determined that 2.11%60.38% of the b-gal + cells in the DG of Nestin-CreER T2 ;JoJo-Prox1 mice were Nestin + ; 64.57%612.99% were Dcx + ; 34.52%66.32% were NeuN + ; and less than 0.1% were Gfap + or NG2 + ( Figure 10O). No increase in the number of TUNEL + cells was observed in the DG of these mice ( Figure S12H). These results suggest that there is premature neuronal differentiation in the DG of Nestin-CreER T2 ;JoJo-Prox1 mice during postnatal stages of brain development.
Finally, we addressed the consequences of ectopic expression of Prox1 in the SGZ during adult neurogenesis. Analysis of the brains of P30 Nestin-CreER T2 ;JoJo-Prox1 mice revealed the presence of Nestin + ;Prox1 + adult NSCs ( Figure S12N). As a consequence of Prox1 ectopic expression, the numbers of Nestin + (Figure 10Q,T) and Sox2 + ( Figure S12P) cells were reduced at this stage. Accordingly, the numbers of Dcx + (Figure 10S,U) and Calretinin + ( Figure S12R) intermediate progenitors were also reduced. These results indicate that the ectopic expression of Prox1 depleted the adult NSC population.

Discussion
In this article, we report for the first time the functional role for Prox1 in mammalian brain development. We determined that in the mouse, Prox1 is required for the maturation of granule cells during DG development. Prox1 is expressed through all the stages of DG formation; therefore, it is possible that the defective granule cell maturation observed in Nestin-Cre;Prox1 F/F mice is an indirect consequence of the earlier absence of Prox1 at the intermediate progenitor level. However, it is also possible that Prox1 plays additional functional roles during granule cell formation. Moreover, the granule cells in the DG are one of the few types of brain cells that express Prox1 throughout adulthood. This suggests that Prox1 might be necessary not only for the maturation of granule cells but also for the regulation of other aspects of granule cell function. NeuroD1 is also required for granule cell maturation [15,29,30]. During embryogenesis and in the absence of Prox1 or NeuroD1, mutant granule cells express some neuronal markers but fail to fully differentiate and undergo apoptosis [15]. As a consequence, Nestin-Cre;Prox1 F/F and NeuroD1 -/mice have very few granule cells ( Figure 1) [15,30]. Few Prox1 + granule cells are present in the DG of NeuroD1 -/mice [15], and NeuroD1 is expressed in the DG of Prox1-conditional mutants. These results suggest that although NeuroD1 and Prox1 might control similar or parallel pathways of granule differentiation, they might not necessarily induce each other's expression.
We also showed that during adult neurogenesis, Prox1 is necessary for the survival of Tbr2 + intermediate progenitors in the SGZ. In this case and similarly to DG development, our data suggest that the lack of Prox1 results in defects in granule cell maturation that could also be an indirect consequence of the earlier lack of Prox1 at the progenitor level. Nevertheless, the lack of Prox1 leads to an increase in apoptosis in Tbr + and Dcx + cells and the absence of adult neurogenesis. Also in this case, similar results have been reported for NeuroD1-mutant mice during adult neurogenesis [44,45].
Our results provide the strongest evidence so far about a feedback mechanism involved in the regulation of adult neurogenesis and progenitor cell numbers in the adult SGZ. We also provide evidence supporting the proposal that Tbr2 + intermediate progenitors and Dcx + cells are required to maintain the adult NSC population in the SGZ niche. We showed that in the absence of Tbr2 + and Dcx + cells and as this feedback regulation becomes defective, adult NSCs continue generating new progeny such that the adult NSC population in the SGZ of Prox1-conditional mutants transiently expands but is ultimately depleted. We do not yet know whether this depletion of NSCs is due to their premature differentiation or their exhaustion resulting from increased proliferation [46].
How is this feedback mechanism regulated? We showed that in the SGZ of Prox1 conditional-mutant mice the lack of intermediate progenitors leads to a reduction in Jagged1 expression. Moreover, we determined that in these conditional-mutant mice the lack of intermediate progenitors leads to the absence of active Notch signaling in adult NSCs; active Notch signaling is necessary for the maintenance of adult stem cells in the brain, bone marrow, and gut [42,[47][48][49][50][51]. Previous work has shown that during embryonic development, the activity of mind bomb homolog 1 (Mib1) is required for Jagged and Delta-like-mediated Notch signaling. In the absence of Mib1, Notch signaling is not activated in radial glia cells, which results in their premature differentiation [41,52]. In the SGZ, the absence of Notch1 is sufficient to induce neuronal differentiation and reduce the proportion of adult NSCs and intermediate progenitors, which suggests that this receptor is involved in NSC self-renewal in the SGZ [42]. Therefore, the role of other Notch receptors during adult neurogenesis could be either redundant or restricted to a subpopulation of cells in the SGZ [42,53].
Together, these and others results [42] suggest that in the SGZ, the Notch1/Jagged1 pathway may regulate NSC self-maintenance. However, we cannot exclude the possibility that Notch activation in adult NSCs is also mediated by other Notch ligands expressed by other cell types (e.g., vascular endothelial cells) [54]. Also, we cannot rule out the possibility that other signaling molecules or trophic factors produced by Prox1-expressing intermediate progenitors may be involved in the feedback mechanism that controls adult neurogenesis in the SGZ.
In summary, we have provided evidence demonstrating that Prox1 plays a key role as a neurogenic factor during granule cell formation. In the absence of Prox1, granule cell maturation is affected at later stages of differentiation; ectopic expression of Prox1 in the brain is sufficient to overrule the mechanisms that control NSC self-maintenance and induce premature differentiation in the SGZ during postnatal stages of brain development and adult neurogenesis.

Mice
Prox1 F/F [21] mice, Nestin-Cre [22] mice, and Nestin-CreER T2 [28] mice were previously described. The JoJo-Prox1 construct was generated by the introduction of a 2.2 kb Prox1 cDNA into a CMV-CAG-loxP-eGFP-Stop-loxP-IRES-bGal expression vector [43]. Mice were kept in the NMRI background. TM (Sigma) was dissolved in safflower oil at 20 mg/ml. For prenatal induction, time-mated females were treated with TM by gavage at E16.5 and embryos were harvested at E18.5. To induce Cre recombination postnatally, pups were fed TM (4 mg/20 g body weight) daily from P0 until the collection day. As a consequence of postnatal TM administration, body weight was reduced in TM-treated versus non-TM-treated pups. This reduction was temporary; 2- mo-old postnatally TM-treated Nestin-CreER T2 ;Prox1 F/+ and Nestin-CreER T2 ;Prox1 F/F mice had a similar weight than non-TM-treated animals. In adults (8-wk-old), TM (4 mg/20 g body weight) was administered by gavage 3 times/wk for 4 wk. No obvious weight differences were observed in the TM-treated animals at the time of sample collection. Genotypes were determined by PCR analysis. Nestin-Cre;Prox1 F/+ , Nestin-CreER T2 ;Prox1 F/+ , and JoJo-Prox1 mice were used as controls.

Immunohistochemistry
For Hes1 staining during adult stages, brains were perfused with 4% PFA and embedded in paraffin. Antigen retrieval was performed for 7 min at 105uC in a decloaking chamber with Antigen Retrieval Solution (Dako). The ABC Kit (Vector Laboratories) and TSA Fluorescein System (Perkin Elmer) were used for signal amplification. Jagged1 staining at adult stages was performed on brains perfused with 4% PFA and cryoprotected in 30% sucrose. Signal amplification was obtained with the ABC Kit and the TSA Fluorescein System. Immunohistochemistry for the rest of the antibodies was performed as described [17]. The following antibodies and dilutions were used: rabbit anti-C-Prox1

TUNEL and Proliferation Assays
TUNEL assay was performed on tissue sections as previously described [56]. For proliferation assays at embryonic stages, timemated female mice were injected with BrdU (100 mg/g body weight, intraperitoneally), and embryos were harvested 1 h later. Embryos were fixed o/n in 4% PFA and cryoprotected in 30% sucrose. For proliferation assays at early postnatal stages, P5 and P10 pups were injected with BrdU (100 mg/g body weight, intraperitoneally) 1 h before harvest. Brains were perfused with 4%PFA and cryoprotected in 30% sucrose. For proliferation assays at adult stages, animals were injected with BrdU (100 mg/g body weight, intraperitoneally) 15 d before harvest. Brains were perfused in 4% PFA and cryoprotected in 30% sucrose. BrdU incorporation was exposed after 20-min treatment in 2N HCl. Mouse anti-BrdU (1:10; BD Biosciences) antibody was used. Sections were counterstained with DAPI.

Neurosphere Cultures and Immunocytochemistry
Neurosphere cultures were established as described, with modifications [57]. Briefly, E16.5 hippocampi were dissected, disaggregated in trypsin, and maintained in culture in neurosphere culture medium (Neurobasal medium with GlutaMAX, Pen/ Strep, B27, and N2) (Gibco) supplemented with 20 ng/ml EGF (Upstate) and 20 ng/ml FGF (Millipore). After the fourth passage, neurospheres were differentiated for 4 d in neurosphere culture medium with 10% fetal calf serum (FCS) without supplements, in Lab-Tek II CC 2 chamber slides (Nunc). Cells were fixed in 2% PFA for 15 min at room temperature (RT). Cells were blocked in 10% FCS for 30 min at RT and incubated with the appropriate primary and secondary antibodies (see above) in 2% FCS for 2 h. For the transfection of the neurospheres, a Mouse NSC NucleofectorH Kit (Amaxa) was used according to the manufacturer's specifications. For Prox1 expression, a Prox1 cDNA was cloned in an expression plasmid under the PGK promoter.

Kainic Acid Assay
Adult neurogenesis was induced by KA as described [39] with minor modifications. Briefly, TM (4 mg/20 g body weight) was administered by gavage 3 times/wk for 4 wk to 8-wk-old Nestin-CreER;Prox1 F/F and Nestin-CreER;Prox1 F/+ mice. At the end of TM treatment, animals were injected intraperitoneally with KA (30 mg/kg body weight; Sigma) dissolved in PBS. Approximately 40 min after KA injection, all mice displayed status epilepticus for 2 to 3 h. Approximately 15% of control and conditional mutant mice died 12 h after KA administration. Eight days after KA injection, the surviving animals were euthanized and their brains perfused with 4% PFA and cryoprotected in 30% sucrose.