We investigated gene expression during postnatal OB neurogenesis by in vivo electroporation of neural stem cells in the lateral and dorsal aspects of the forebrain lateral ventricle at postnatal day 1 (P1), followed by the isolation of homotypic cohorts at different time points by microdissection and FACS. Microarray analyses provided detailed insight into gene expression changes between the two neurogenic lineages (‘in space’) and during the progression from stem cells to young neurons (‘in time’; Figure 1A; for detail see Tiveron et al., 2017).

These analyses showed that Vax1 was confined to the neurogenic lineage derived from the lateral ventricular wall (Figure 1B). Vax1 mRNA was induced at low levels at 1 day post-electroporation (dpe) when most GFP-positive cells were transit amplifying precursors [Boutin et al., 2008; Figure 1B]. Expression strongly increased at 2dpe and remained stably high at 4dpe, when most cells were migratory neuronal precursors, before steeply decreasing at 7dpe when cells arrived in the OB and emigrated from the RMS to invade the granule cell (GCL) and the glomerular (GL) layers (Tiveron et al., 2017). In comparison, isolates from dorsally electroporated brains showed no obvious Vax1 mRNA expression over all analyzed time points (Figure 1B).

Next, we investigated Vax1 co-expression with known markers of defined neuronal subsets or differentiation stages in the postnatal V-SVZ. In the absence of a Vax1 antibody that provided reliable signals on postnatal tissue sections, we combined in situ hybridization for Vax1 mRNA with immunohistochemistry for the proteins PAX6, ASCL1, KI67, and DLX2 (Figure 1C–H). Vax1 mRNA always showed a gradient-like distribution along the lateral ventricular wall, with highest expression in the ventral-most aspect and extending far into dorsal regions (Figure 1C,D,E,F). Fainter expression was observed along the septal wall. In the lateral wall, Vax1 mRNA was excluded from the VZ but expressed in the underlying SVZ (Figure 1C’,D’,E’,F’). Vax1 mRNA in the lateral wall was generally non-overlapping with PAX6, a marker for the dorsal stem cell pool and a determinant for the DA-N lineage (Figure C,C’). Vax1 mRNA-positive cells rarely expressed ASCL1 (3.3%±0.32), a marker for transit amplifying precursors (Figure 1D,D’), but the vast majority was labeled with a DLX2 antibody (90%±0.89) (Figure 1E,E’; arrowheads), confirming the preferential expression of Vax1 in migratory neuronal precursors (Doetsch et al., 2002; Figure 1I). Finally, about one third (32.6%±3.07) of the Vax1+ cells in the SVZ expressed the proliferation marker KI67 (Figure 1F,F’, arrowheads), likely representing mitotic neuroblasts. In the RMS, Vax1 mRNA was also present in a ventro-dorsal gradient, complementary to PAX6 immunostaining (Figure 1G), although this organization was less evident than in the SVZ, probably due to cell intermingling in this migratory compartment (Figure 1G’, G”). Like in the SVZ, Vax1 mRNA in the RMS co-localized strongly with DLX2 immunoreactivity (Figure 1H, close up H’, H’).

Altogether, the combination of microarray studies in defined neuronal lineages and histological approaches led to the conclusion that Vax1 is expressed in the lateral and ventral SVZ in a subset of proliferating precursors and in most neuroblasts, the latter maintaining expression during their migration in the RMS.

Previous work demonstrated that CB-N destined for the GL are generated from the ventral-most region of the anterior lateral ventricles (LV), and SHH signaling has been implicated in their specification (Merkle et al., 2007; Ihrie et al., 2011). As Vax1 is strongly expressed in this area, and has been shown to act as an intracellular mediator of SHH signaling (Take-uchi et al., 2003; Furimsky and Wallace, 2006; Zhao et al., 2010), we first asked if the TF is implicated in the generation of the CB-N subtype.

Vax1 conditionally mutant mice (Vax1cKO) (Hoffmann et al., 2016) were bred to R26tdTomato mice to monitor CRE-induced recombination and to follow the distribution and fate of mutant and control cells over time. We used postnatal in vivo brain electroporation to express CRE protein in the lateral wall (Figure 2A). Since targeting of the Vax1-positive ventral region of the ventricular wall with DNA-based expression constructs is inefficient, we used Cre mRNA, that is highly efficient for the transfection of stem cells along the entire wall, including the most ventral aspect (Bugeon et al., 2017). Animals were electroporated at P0 and analyzed 15 days later, when labeled neurons reached the OB and integrated into the GCL and GL (Figure 2B). Quantification of labeled neurons in the GCL and GL revealed no significant differences between control and mutant mice (Figure 2C; Figure 2—figure supplement 2). TH-positive PGC were also not significantly affected (Figure 4I). However, there was a significant loss in the small population of CB-positive neurons in the GL (Figure 2D,E).

Thus, Vax1 expression in the ventro-lateral-derived neurogenic lineage is necessary for the correct generation of CB-N in the GL.

In situ hybridization indicated that Vax1 mRNA was expressed in a ventro-dorsal gradient (Figure 1). To confirm the existence of such a gradient, we micro-dissected V-SVZ tissue from the dorsal, dorso-lateral, and ventro-lateral regions of the ventricular walls of postnatal and adult mice and subjected the isolates to RT-qPCR analyses for Vax1 mRNA. In agreement with the histological data, Vax1 mRNA showed a steep ventro-dorsal gradient, opposed to, and partially overlapping with, the well-described localization of Pax6 mRNA, that extends dorso-ventrally (Figure 3A, Figure 3—figure supplement 1; de Chevigny et al., 2012a). This observation appeared significant for two main reasons. First, several studies provided evidence that Vax1 can negatively regulate Pax6 expression during development (Bertuzzi et al., 1999; Hallonet et al., 1999; Mui et al., 2005). Second, repression of Pax6 translation along the lateral wall is necessary to confine Pax6 protein, and consequently DA-N phenotype, to the dorsal stem cell pool (de Chevigny et al., 2012a).

Based on this information, we hypothesized that Vax1 is implicated in Pax6 down-regulation in the postnatal SVZ. To test this, we overexpressed Vax1 in the dorsal and lateral neurogenic lineages by electroporation and investigated the impact on Pax6 expression. A Vax1 expression plasmid (pCAG-Vax1), or an empty control vector, was co-electroporated with pCX-GFP into either the dorsal or the lateral ventricular wall (Figure 3B). Two days later, animals were sacrificed and intensity of Pax6 immunostaining in GFP-positive cells was measured in the dorsal and dorso-lateral SVZ (Figure 3C,D,E). GFP-positive cells generated in both compartments showed a significant decrease in Pax6 expression levels at this early time point (Figure 3D,E).

Then we asked if Pax6 expression was affected at late time points, after the arrival of newborn neurons in the OB GL. Indeed, 15 and 25 days after pCAG-Vax1 electroporation into the dorsal wall the proportion of GFP-positive neurons that showed Pax6 expression was reduced by over 80% (Figure 3F,G).

We conclude that Vax1 has the capacity to act as a negative regulator of Pax6 expression during postnatal OB neurogenesis.

DA-N in the OB GL are derived from the dorsal and dorso-lateral aspects of the ventricle walls (Merkle et al., 2007; Fernández et al., 2011; de Chevigny et al., 2012a) and Pax6 expression is necessary and sufficient for the acquisition of this neurotransmitter phenotype (Hack et al., 2005; Kohwi et al., 2005). We asked if Vax1 overexpression in Pax6-positive cells was sufficient to inhibit the generation of DA-N in the OB.

We first targeted the dorsal compartment (Figure 4A–D), where the majority of DA-N are generated (Fernández et al., 2011; de Chevigny et al., 2012a). Ectopic expression of Vax1 led to a significant loss of TH/GFP-positive neurons 15 days later in the OB (Figure 4B). Loss of TH-positive cells was robust over time and could be observed at 25 and 60 dpe (Figure 4C). Two observations pointed toward the specific loss of dopaminergic neurons. First, the number of the second major identified neuron type that is generated in the dorsal ventricular wall, CR-N of the glomerular layer (Fernández et al., 2011; Tiveron et al., 2017), was unaffected by Vax1 expression (Figure 4B), arguing against a fate shift toward this neuron type (Tiveron et al., 2017). Second, the density of total GFP+ cells in the GL of Vax1-electroporated animals was significantly reduced (Figure 4D), whereas the number of GFP+ granule cells was not affected (Figure 4—figure supplement 1). As dopaminergic neurons represent a substantial population of all dorsal generated periglomerular cells such an overall loss is coherent with a loss of the DA-N subtype.

Next, we targeted the lateral ventricular wall (Figure 4E–G), where smaller but still significant numbers of DA-N are produced from a dorso-lateral stem cell pool (Figure 4F). Overexpression of Vax1 in the lateral wall induced a significant loss of TH-positive neurons in the GL 15 days later (Figure 4F). At the same time point, numbers of CB-N and CR-N were unchanged (Figure 4F), indicating again that no phenotypic switch toward these subtypes occurred. The density of GFP+ cells was also unaffected in both GL (Figure 4G) and GCL (Figure 4—figure supplement 1). Thus, Vax1 overexpression specifically inhibits DA-N phenotype of newborn neurons in the OB.

We also investigated whether Vax1 loss-of-function had a positive impact on DA-N phenotype in the OB. NSCs along the lateral wall of Vax1cKO animals were electroporated with pCX-CRE (Figure 4H) and the proportion of TH-positive neurons in the OB was analyzed 15 days later. These analyses failed to show a significant increase in DA-N at a confidence level of p≤0.05 (Figure 4I).

We conclude that Vax1, likely via regulation of Pax6, has the capacity to negatively control the generation of DA-N for the OB. Moreover, these data show that while Vax1 is necessary for the generation of CB-N, it is not sufficient.

Previous work demonstrated that mature microRNA miR-7 is expressed in a ventro-dorsal gradient along the lateral ventricular wall and post-transcriptionally inhibits Pax6 protein expression. This interaction confines the generation of DA-N to the very dorso-lateral aspect (de Chevigny et al., 2012a). As Vax1 and miR-7 are expressed in a similar gradient, we hypothesized that the repression of Pax6 protein expression by Vax1 is mediated by miR-7. To address this idea, we overexpressed Vax1 together with GFP in the lateral stem cell compartment and isolated GFP-positive cells 2 days later by microdissection, dissociation and flow cytometry cell sorting (Figure 5A). qRT-PCR analyses demonstrated that augmented Vax1 expression (Figure 5B) led to a strong increase in miR-7 levels (Figure 5C), suggesting that Vax1 regulates MiR-7 expression. In agreement, bioinformatical analyses of the proximal promoters of the three MiR-7 loci (MiR-7–1, MiR-7–2, MiR-7b) identified a significant match of the Vax1 DNA binding motif within 500 bp of the transcription start site of each of the three MiR-7 loci present in the mouse genome (Figure 5D). Altogether, these results suggest that the negative impact of Vax1 on Pax6 expression is, at least in part, mediated via the positive regulation of MiR-7.

All animal procedures were carried out in accordance to the European Communities Council Directie 2010/63/EU and approved by French ethical committees (Comité d'Ethique pour l'expérimentation animale no. 14; permission numbers: 00967.03; 2017112111116881 v2). Animals were held on a 12 h day/night cycle and had access to food and water ad libitum. Animals of both sexes were used for experiments. CD1 mice (Charles River, Lyon, France) were used for in vivo electroporation and expression pattern analyses. Vax1^flox (Vax1cKO) conditional mutants (Hoffman 2016) and Rosa26^tdTomato reporter mice (Ai14, Jackson Laboratories, USA, RRID:IMSR_JAX:007914) were bred on a mixed C57BL/6*CD1 genetic background. Vax1flox genotyping was performed with Vax1flox forward: 5’-GCCGGAACCGAAGTTCCTA; Vax1wt forward: 5’-CCAGTAAGAGCCCCTTTGGG, reverse 5’-CGGATAGACCCCTTGGCATC. Ai14 genotyping was performed with R26wt forward: 5’-AAGGGAGCTGCAGTGGAGTA, reverse 5’-CCGAAAATCTGTGGGAAGTC; R26tdTom forward: 5’- CTGTTCCTGTACGGCATGG and reverse: 5’- GGCATTAAAGCAGCGTATCC.

The full-length rat cDNA sequence of Vax1 was excised from pCMV2-Rn-Vax1-FLAG (a gift of Kapil Bharti) and subcloned into pCAGGS vector to produce pCAG-Vax1. Postnatal day 0 (P0) or day 1 (P1) pups were electroporated as previously described (Boutin et al., 2008; de Chevigny et al., 2012a) with plasmid DNA or RNA (Bugeon et al., 2017). CRE Recombinase mRNA (130-101-113, a generous gift from S. Wild and A. Bosio, Miltenyi Biotec, Bergisch Gladbach, Germany) and pCX-CRE (Morin et al., 2007) were used at a concentration of 0.5 µg/µl. Recombination efficiency was tested by qRT-PCR from tdTomato+ cells isolated by FACS (Figure 2—figure supplement 1). In CD1 pups, pCX-EGFP (Morin et al., 2007) was co-injected with pCAG-Vax1 or empty pCAGGS (as control) in a 1:2 molecular ratio to label the electroporated cells. Targeting of the dorsal or lateral wall of the lateral ventricle was directed by distinct orientation of the electrodes. Brains were collected at different time points after electroporation.

For all procedures, tissues were fixed by intracardiac perfusion with ice-cold 4% paraformaldehyde (wt/vol) in phosphate buffered saline (PBS) and cryoprotected in 30% sucrose solution in PBS. Brains were sliced coronally using either a microtome (Microm Microtech, France) or a cryostat (Leica Biosystems, France).

Mouse Vax1 cDNA clone (GeneBank: BC111818, Imagene) was used to produce the anti-sense RNA probe. Combined in situ hybridization (ISH) and immunohistochemistry (IHC) were performed as previously described (Tiveron et al., 1996) on 16 µm cryosections at postnatal day 3 (P3) using anti-digoxigenin antibody (Sheep polyclonal, 1:1000, Roche, 11093274910, RRID:AB_514497). Two days exposure to alkaline phosphatase (AP) substrate NBT/BCIP (Promega, S3771) was necessary to detect AP-DIG-labeled Vax1 mRNA at maximal level. Subsequently, antibodies directed against PAX6 (rabbit polyclonal, 1: 1000, Millipore, AB2237, RRID:AB_1587367), KI67 (mouse IgG1k clone B56, 1:200, BD Biosciences, 550609, RRID:AB_393778), ASCL1 (mouse, clone 24B72D11.1, 1:100, BD Biosciences, 556604, RRID:AB_396479), or DLX2 (rabbit polyclonal, 1: 2000 a gift from Prof. K. Yoshikawa) were applied to the ISH treated sections and were revealed with secondary antibodies coupled to horseradish peroxidase (Jackson ImmunoResearch Laboratories, UK), using 3, 3’-Diaminobenzidine (Invitrogen, 750118) as a substrate. Sections were finally mounted in fluoromount medium and analyzed by light microscopy using Axiovision Rel. 4.8 software (Zeiss, Germany).

For immunofluorescence, 50 µm floating sections were blocked in PBS supplemented with 0.5% Triton-X100, 10% foetal calf serum (FCS) and incubated overnight at 4°C in PBS, 0.1% Triton, 5% FCS with primary antibodies: anti-Calbindin D-28K (rabbit polyclonal, 1:1000, Millipore, AB1778, RRID:AB_2068336 or mouse IgG1, 1:3000, Swant, 300, RRID:AB_10000347), anti-Calretinin (mouse IgG1, clone 37C9, 1:1000, Synaptic systems, 214111, RRID:AB_2619906), anti-Tyrosine hydroxylase (chicken IgY, 1:1000, Avès labs, TYH, RRID:AB_10013440). After washing in PBS, Alexa Fluor-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories) were applied diluted at 1:500 in blocking solution for 2 hr at room temperature. After staining of cell nuclei with Hoechst 33258 (Invitrogen, H3569), sections were mounted with Mowiol (Sigma-Aldrich, 81381).

Animals (P1-P3) were decapitated and brains were cut into 300 μm thick sections using a Vibrating-Blade Microtome (Thermo Scientific, HM 650V). V-SVZ and RMS tissues were micro-dissected under a binocular microscope and kept in cold Hank’s balanced salt solution (HBSS, Gibco, 14170120). RNA was extracted using the miRNAeasy kit (Qiagen, 217004) or by Trizol reagent (life technologies, 15596026) according to manufacturer instructions, allowing the recovery of long and short RNAs. For Pax6 and Vax1 expression analysis, cDNA was prepared using superscript III reverse transcriptase (ThermoFisher Scientific, 12574–030) following manufacturer instructions and quantitative PCR was performed on a BioRad CFX system using SYBR GreenER qPCR SuperMix (ThermoFisher Scientific, 11762100) in technical triplicates. ß-Actin was used as reference gene. Primers used for mRNA detection are the following: Beta Actin forward 5’-CTAAGGCCAACCGTGAAAAG and reverse 5’-ACCAGAG GCATACAGGGACA; Pax6 forward 5’-TGAAGCGGAAGCTGCAAAGAAA and reverse 5’-TTTGGCCCTTCGATTAGAAAACC; Vax1 forward 5’- GCTTCGGAAGATTGTAACAAAAG and reverse 5’- GGATAGACCCCTTGGCATC.

For miRNA expression, cDNA was prepared using miScript II (Qiagen, 218160) and qPCR was performed using miScript SYBR Green PCR Kit (Qiagen, 218073) together with miRCURY LNA miRNA PCR Assay (Qiagen, 339306) on a StepOne Real-Time PCR System (Applied Biosystems). Let-7a was used as normalizer miRNA. Three independent technical replications for each condition were processed in triplicates.

To generate single-cell suspensions for FACS, dissected tissues were subjected to cell dissociation using a Papain solution and mechanical dissociation as described previously (Lugert et al., 2010). Cells were resuspended in HBSS/Mg/Ca supplemented with 10 mM HEPES (Gibco, 15630080), 40 μg/ml DNase I (Roche, 10104159001), 4.5 g/L Glucose (Gibco, A2494001) and 2 mM EDTA, filtered through a 30 μm pre-separation filter (Miltenyi Biotec, 130-041-407) and sorted on MoFlo Astrios EQ cytometer (Beckman- Coulter) gating on GFP- or tdTomato- positive population.

Significant Vax1 DNA binding motif (JASPAR Core position-weight matrix MA0722.1) occurrences in the three MiR-7 loci (MiR7-1, MiR7-2, MiR7b) proximal promoters (defined as 500 bp upstream of the transcription start site) from the Mus musculus mm10 genome were determined using FIMO from the MEME suite (Bailey et al., 2009) with a maximum p-value limit of 0.001. FIMO-derived p-values of motif matches in the promoters were 0.000173 (MiR7-1), 0.000333 (MiR7-2), and 0.000806 (MiR7b). Genome browser images were created using the UCSC Genome Browser.

All images were analyzed blind to the experimental condition. Optical images were acquired with an Axioplan2 ApoTome microscope (Zeiss, Germany) using ZEN software (Zeiss, RRID:SCR_013672), and processing was performed using Fiji software (RRID:SCR_002285, [Schindelin et al., 2012]) or ImageJ (NIH, https://imagej.nih. gov/ij/, RRID:SCR_003070). ISH/IHC double staining experiments were analyzed by counting labelled SVZ cells along the lateral ventricular wall on 40X magnification images (n = 3 SVZ for Ki67 and Dlx2, n = 8 SVZ for Ascl1). For cell counting in V-SVZ and OB tissues from electroporation experiments, a minimum of three sections per brain (n = animal) were analyzed and the totality of GFP+ or Tomato+ cells of each section were counted. To measure mean intensity of fluorescence after antibody staining, ROI were applied on individual electroporated cells.

Control and Vax1 electroporation assays were performed from the same litter and were reproduced in two or three independent experiments for lateral or dorsal electroporation, respectively. Histograms were drawn with GraphPad Prism version 8 (GraphPad Software, San Diego, CA). Data are presented as mean ± SD. Each sample (n = animal/brain) is represented on histograms by dot. Statistical analyses were performed using R software (RRID:SCR_001905) and R Commander Package (https://CRAN.R-project.org/package=Rcmdr). The non-parametric two-tailed Mann Whitney U test was performed for all in vivo experiments, on pooled experimental repetitions when appropriate. Differences were considered statistically significant when p≤0.05.