A mouse strain with a targeted deletion of the NENF gene, provided by Merck-Serono under a material transfer agreement, was used, as previously reported (Novais et al., 2013). All experiments were performed in 2-month old male mice lacking the expression of NENF and the respective wild-type littermate controls in a C57BL/6J mouse background. Littermate controls (Nenf^+/+) and neudesin-null (Nenf^−/−) animals were obtained by crossing heterozygous animals. Animals, housed in groups of five, were maintained under 12 h light/dark cycles at 22 ± 1°C, 55% humidity and fed with regular rodent’s chow and tap water ad libitum. The Portuguese national authority for animal experimentation, Direção Geral de Veterinária, approved this study (permission ID: DGV9457). All experiments were performed following the guidelines for the care and handling of laboratory animals, as described in the Directive 2010/63/EU of the European Parliament and of the Council. A different set of animals was used for fear conditioning and immunohistochemistry experiments.

5-bromo-2′-deoxyuridine (BrdU) is a thymidine analog that incorporates in the DNA during the S phase of the mitotic cycle. To label proliferative cells in the SVZ and SGZ, mice were intraperitoneally (i.p.) injected with 50 mg/kg of BrdU (Sigma Aldrich, St. Louis, MO, USA) 24 h prior to sacrifice. To label newly born granular neurons at the DG, animals were i.p. injected with BrdU twice a day for five consecutive days and sacrificed 7 weeks later (Deng et al., 2013; Ferreira et al., 2017). During this period of chase, neuroblasts proliferate and mature, thus BrdU positive cells co-label with a mature neuronal marker of integrated newborn granular neurons in the DG circuitry, represent not only the morphologically mature cells but also functionally mature DG granule cells (van Praag et al., 2002; Zhao et al., 2006; Kee et al., 2007).

For immunohistochemistry, 2-month-old littermate control and neudesin-null mice were used (n = 7 in each group). Animals were transcardially perfused, under deep anesthesia [ketamine hydrochloride (150 mg/Kg) with medetomidine (0.3 mg/Kg)], with 4% of paraformaldehyde (PFA) in 0.01 M phosphate-buffered saline (PBS). Brains were removed, postfixed for 1 h in 4% PFA solution, cryoprotected in 30% sucrose solution for 12 h, and then embedded in optimal cutting temperature compound (ThermoFisher Scientific, Waltham, MA, USA), snap-frozen in liquid nitrogen and kept at −20°C until sectioning in a cryostat. Immunohistochemistry was performed on coronal 20 μm thick serial sections, pre-treatment of the brain tissue involved antigen retrieval in pre-heated 95–100°C citrate buffer (10 nM, pH6; Sigma) bath for 15 min; additionally for BrdU immunostaining, the sections received a pre-treatment for DNA denaturation with HCl for 30 min at room temperature (RT). Slides were incubated with an anti-BrdU primary antibody at a dilution of 1:50 (mouse anti-bromodeoxyuridine, Clone Bu20a, DAKO, Barcelona, Spain) and an anti-Ki67 primary antibody at a dilution of 1:300 (rabbit; Novocastra, Newcastle, UK). These were detected using a secondary antibody with the Ultravision Detection System (Lab Vision, Freemont, CA, USA), and the reaction developed with 3,3′-diaminobenzidine substrate (Sigma). The sections were subsequently counterstained with hematoxylin.

For immunofluorescence analysis, six littermate control and six neudesin-null mice were used. Immunofluorescence analysis was performed on fixed 20 μm coronal sections, pre-treatment using antigen retrieval and DNA denaturation (for BrdU) was also performed. The sections were further incubated with the following primary antibodies: doublecortin (DCX; rabbit polyclonal, Abcam, Cambridge, UK) at a dilution of 1:300, BrdU (rat anti-BrdU, BU1/75 clone, Abcam) at a dilution of 1:100, Calbindin (Calb; rabbit polyclonal, Abcam) at a dilution of 1:300 and neudesin at a dilution of 1:50 (NENF; rabbit polyclonal; Sigma). Fluorescent secondary antibodies (Invitrogen, Carlsbad, CA, USA), anti-rabbit (for DCX, Calb and NENF) and anti-rat (for BrdU), were used to detect the primary antibodies at a dilution of 1:500. To label the nucleus, incubation with 4’,6-diamidino-2-phenylindole (DAPI; Sigma) at a dilution of 1:200 was performed. Antibodies were diluted in PBS-0.5% Tween with 10% fetal bovine serum (FBS) and incubated overnight at RT, for the primary antibodies, and 2 h at RT, for the secondary antibodies.

The number of BrdU positive cells was assessed in the neurogenic niches using an Olympus BX51 microscope (Olympus, Hamburg, Germany) associated with the Visiopharm Integrator system (VIS) software (version 2.12.3.0). Coronal sections comprised coordinates between Bregma 0.86 mm and 0.14 mm for the SVZ (Falcão et al., 2013), −1.22 mm to −2.30 mm for the dorsal SGZ and −2.80 mm to −3.88 mm for the ventral SGZ. These boundaries were chosen as a representation of the septal and temporal poles of the hippocampus, known to be distinct in morphology and connectivity (Pinto et al., 2015). The number of BrdU positive cells and Ki67 positive cells were normalized for the respective area (mm²).

Hippocampi DG and SVZ from 3 days to 6 days old Nenf^+/+ and Nenf^−/− male mice were carefully dissected and meninges and choroid plexus removed. Tissue was triturated in DMEM F12 (Gibco, Rockville, MD, USA) with 10% FBS with the help of a P1000 micropipette; upon completion of tissue dissociation, the cell suspension was centrifuged at 900 rpm and the supernatant carefully discarded. Single cells obtained were plated in DMEM F12 media supplemented with GlutaMAX (Gibco), 100 U/ml penicillin, 100 μg/ml streptomycin (Gibco), 1% B27 (Gibco), 10 ng/ml epidermal growth factor (Gibco), and 10 ng/ml basic fibroblast growth factor (Gibco). Neurospheres were allowed to form and grow at 37°C in a 95% air-5% CO₂ humidified atmosphere for 5 days. Splitting of neurospheres was performed by mechanical dissociation with a P1000 micropipette using 0.05% of trypsin in DMEM F12, followed by incubation for 3 min at 37°C; after trypsin inactivation with 10% FBS, the cells were centrifuged at 1,000 rpm and the supernatant discarded. Single cells were counted using a Neubauer chamber and plated in the same media with growth factors in 48-wells plates at a 1 cell/μL density, six wells per each condition and neurospheres were counted every 24 h for 4 days. Two independent experiments were performed.

To determine the number of proliferating neuroblasts (DCX⁺/BrdU⁺) and the number of newly-born granular neurons (Calb⁺/BrdU⁺) at the SGZ, sections from the dorsal hippocampus were used (Franklin and Paxinos, 2013). Six animals per genotype (Nenf^+/+ and Nenf^−/−) were assessed and four sections from each animal were analyzed. BrdU positive cells were identified using a confocal microscope (FV1000; Olympus, Hamburg, Germany) and the total number of double DCX⁺/BrdU⁺ and Calb⁺/BrdU⁺ cells was counted and normalized for the total number of BrdU positive cells and presented as a percentage.

For the behavior assessment, groups had the following composition: eight male Nenf^+/+ and 10 male Nenf^−/−. We performed a non-cued paradigm of the CFC where animals placed in an operant chamber (Med Associates) explored it for 3 min, after which three mild foot shocks (0.5 mA) were given randomly for the following 15 min. This conditioning phase was performed for two consecutive days. On the 3rd day, animals were placed for 3 min in the conditioned context (context A) followed by 3 min in a different context (context B; Fanselow and LeDoux, 1999; Anagnostaras et al., 2001; Mateus-Pinheiro et al., 2016; Ferreira et al., 2017). In this new context, a vanilla scent was added to the chambers and the chamber fan turned on; the walls were covered with patterned posters and the grid floor covered with white plastic. Each animal was scored for the percentage of time freezing in each chamber of the discrimination session day. Freezing behavior was scored manually and consisted of the total absence of movement besides regular breathing for more than 2 s. Parameters analyzed included the percentage of time freezing during the discrimination session, the total percentage of time freezing in contexts (A,B), and the index of discrimination between contexts as the ratio of percentage of time freezing (contexts A − B)/percentage of time freezing (contexts A+B).

For the gene expression study, the following groups of naïve mice were used: 10 male Nenf^+/+ and 10 male Nenf^−/−. Total RNA was isolated from DG samples using RNeasy Plus Micro Kit (Qiagen, Hamburg, Germany), following the manufacturer’s instructions. RNA quantification and quality were confirmed using the NanoDrop ND-1000 (NanoDrop, Thermo Scientific, Wilmington, DE, USA), and 500 ng of RNA from each sample was reverse transcribed into cDNA using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA, USA) following the manufacturer’s instructions. Quantitative real-time PCR analysis (qPCR) was used to measure the expression levels of hypoxanthine guanine phosphoribosyl transferase (Hprt), 5-alpha-reductase 2 and 3 (Srd5a2 and Srd5a3) and gamma-aminobutyric acid A receptor, delta (Gabrd).

Primers were designed using PrimerBlast software tool from National Center for Biotechnology Information (Bethesda, MD, USA), based in the respective GenBank sequences accession numbers; primers sequence are provided upon request. The qPCR reactions were conducted using equal amounts of cDNA from each sample and were performed in a Bio-Rad CFX96TM using SsoFast™ EvaGreen^® Supermix (Bio-Rad) according to the manufacturer’s instructions. The Hprt gene was used as an internal reference gene; its expression levels did not vary between experimental groups.

All quantitative data are expressed as Mean ± standard error mean (SEM). Data followed a Gaussian distribution, revealed by the normality test (using the Shapiro-Wilk test). Therefore, statistically significant differences between groups were determined using unpaired two-tailed Student’s t-test for two comparisons, and Cohen’s d for effect size is also presented. For the neurospheres clonal analysis, a two-way repeated measures ANOVA with a Bonferroni’s multiple comparisons post-test was used. P-values lower than 0.05 were considered statistically significant. All statistical procedures were performed using GraphPad Prism, version 7 (GraphPad Software Inc., San Diego, CA, USA).

To evaluate the effect of NENF ablation in the adult neurogenic niches, we first injected the DNA thymidine-analog BrdU 24 h before sacrifice, to assess dividing cells. We found that the absence of NENF (Nenf^−/− mice) reduces cell proliferation at the SGZ in ~50% when compared to controls (Nenf^−/− mice) [(t₍₁₁₎ = 2.356; d = 0.29; p < 0.05); Figure 1A]. In the SVZ proliferation did not differ between genotypes (Figure 1B).

The hippocampal formation is known to have topographical specificities (Pinto et al., 2015); thus we evaluated SGZ proliferation in dorsal and ventral poles. We observed lower levels (45% decrease when compared to wild-type mice) of proliferation in the dorsal SGZ of Nenf^−/− mice [(t₍₁₀₎ = 2.405; d = 1.52; p < 0.05); Figure 1C]; we did not observe significant differences in proliferation between genotypes in the ventral SGZ, despite a trend for less BrdU⁺ cells in the absence of NENF (Figure 1D). Using Ki67, a standard endogenous marker for proliferation index, we observed a significant decrease in Ki67⁺ cells in the dorsal SGZ of Nenf^−/− mice when compared to littermate controls [(t₍₆₎ = 3.14; d = 2.22; p < 0.05); Figure 1E]. No significant differences were observed in the ventral SGZ, although, similarly to BrDU, a trend for a decrease in Ki67⁺ cells is observed (Figure 1F).

In vitro neurosphere cultures were performed to confirm the in vivo proliferation deficits observed in Nenf^−/− mice. Postnatal derived stem cells were isolated from the hippocampal DG and the SVZ of Nenf^+/+ and Nenf^−/− mice, and their self-renewal capacity evaluated in a clonal neurosphere assay. We observed a significant 3-fold decrease in the total number of primary neurospheres formed from Nenf^−/− when compared to Nenf^+/+ cells in days 3 and 4. Analysis of the neurosphere counts along the days showed a significant interaction between variables [(F_(3,21) = 14.68 η² = 0.15 p < 0.0001); Figure 2A], with a major effect of genotype (F_(1,7) = 48.99 η² = 0.26 p < 0.01), and of time (F_(3,21) = 48.25 η² = 52.16 p < 0.01); Bonferroni post hoc tests showed significant differences between genotypes at days 3 and 4 (p < 0.0001). Day 3 revealed the most significant difference between genotypes SGZ [(t₍₇₎ = 9.19; d = 0.92; p < 0.00001; Figure 2A. Differences at day 3 in the SVZ were also present, although Nenf^−/− mice showed significantly more neurospheres then controls [(t₍₅₎ = 4.89; d = 3.66; p < 0.001; Figure 2B]. On the other hand, we did not observe differences between genotypes in SVZ-derived neurospheres counts along the days (Figure 2B). This finding confirms the specificity (SGZ but not SVZ) of NENF’s neurotrophic modulation observed in vivo in the adult brain.

The diminished proliferation observed in the dorsal SGZ of Nenf^−/− mice prompted us to further characterize neurogenesis specifically in the dorsal pole of the hippocampus. Proliferative neuroblasts were labeled with both DCX and BrdU (24 h; BrdU⁺/DCX⁺; Figure 3A). The proliferative deficits observed in the absence of NENF seem to result from an alteration in the neuroblasts population, since Nenf^−/− mice SGZ displayed 24% less DCX⁺/BrdU⁺ cells when compared to control littermates [(t₍₈₎ = 2.57; d = 2.71; p < 0.05); Figure 3C]. Of notice, NENF protein expression was not detected in 24 h BrdU⁺ cells but its expression in BrdU⁺ is detected later in differentiation, after 7 weeks of BrdU labeling (Figure 3B). Thus, to assess differentiation we chose a late neuronal maturation marker (Calb) that begins its expression after 4 weeks of differentiation, instead of a global neuronal marker such as NeuN, to evaluate BrdU⁺ cells and assess late stages of differentiation. Herein, we estimated newborn neurons in the DG by performing BrdU labeling for five consecutive days and analyzed the DG after 7 weeks (Figure 3A). In the absence of NENF, there is a significant 1.8-fold decrease in the number of newborn neurons (Calb⁺/BrdU⁺) in the DG when compared to control mice [(t₍₆₎ = 5.04; d = 0.35; p < 0.01); Figure 3D].

Adult hippocampal neurogenesis was implicated in dorsal hippocampal related behaviors, such as pattern separation (Saxe et al., 2006). Animal models with negative modulation of adult hippocampal neurogenesis display impairments in CFC paradigms involving the discriminations of similar contexts (Saxe et al., 2006; Sahay et al., 2011). We used a non-cued CFC paradigm to test dorsal hippocampal function based on lesion studies that show significantly less involvement of the amygdala compared to cued CFC paradigms (Anagnostaras et al., 2001). A two-session conditioning phase was chosen to ensure that both genotypes discriminate the shock-context association (Figure 4A). The animals were positively conditioned since all froze significantly more in the aversive context (context A) than in the new context (context B). Nenf^−/− mice froze significantly more in both contexts [context A: (t₍₁₀₎ = 2.235; d = 0.453; p < 0.05), Figure 4B; and context B: (t₍₁₁₎ = 3.484; d = 0.28; p < 0.01), Figure 4C] indicating a generalized exacerbation of the fear response in the absence of NENF. Ultimately, discrimination between contexts revealed that in the absence of NENF, animals show a significant inferior index of discrimination, [(t₍₁₀₎ = 2.62; d = 0.84; p < 0.05); p < 0.01); Figure 4D]. These results suggest that there is a moderate decrease in contextual discrimination revealing a cognitive impairment in the absence of NENF.

Progesterone metabolization through 5α-reductase results in neurosteroids synthesis, which by allosteric binding to delta⁺ GABA A receptors sustains the excitability/inhibition balance in the DG, modulating tonic inhibition (Figure 5A). On the other hand, NENF was suggested to play a role in the progesterone stabilization complex to potentiate the activation of membrane progesterone receptors (Kimura et al., 2013). Given the potential interaction between NENF and progesterone-related signaling, we evaluated if genes related to progesterone metabolism could be impaired in the DG of Nenf^−/− mice (Figure 5A). We observed that the expression of the isoform 2 of 5α-reductase (Srd5a2) gene is significantly increased (50% higher) in the absence of NENF [(t₍₁₈₎ = 3.53; d = 0.73; p < 0.01; Figure 5C]. We observed a similar increase in the expression of the isoform 3 of 5α-reductase (Srd5a3) [(t₍₇₎ = 2.95; d = 1.58; p < 0.05); Figure 5D], revealing a sharp difference between genotypes in the expression of the progesterone catabolism enzyme. Furthermore, we tested the gene expression of delta subunit of the GABA A receptor because it is modulated by the action of progesterone metabolites (via 5α-reductase) in the DG. We found a significant 1.8-fold increase in the expression of the Gabrd gene [(t₍₁₆₎ = 3.28; d = 1.55; p < 0.01); Figure 5E], indicative of a potential alteration in the DG tonic inhibition control. Together these results suggest that the proposed mechanism for NENF neurotrophic support in the DG can be through progesterone availability for metabolization into neurosteroids (Figure 5B).