ENRICHMENT AND FLUOXETINE NEUROGENIC STIMULI INCREASE ß3-TUBULIN IMMATURE NEURONS IN MENINGES THROUGH TRKB-MEDIATED SIGNALING

Neural precursors (NPs) present in the hippocampus can be modulated by several neurogenic stimuli including environmental enrichment (EE) and antidepressant treatment acting through BDNF-TrkB signaling. We have recently identified NPs in meninges, however menigneal niche response to pro-neurogenic stimuli has never been investigated. To this aim, we analyzed the effects of 4 weeks fluoxetine administration or 1 week EE treatment on NP distribution in mouse brain meninges. Following neurogenic stimuli, although we did not detect modification of meningeal cell number and proliferation, we observed, in meninges, an increased number of β3-Tubulin+ immature neuronal cells. Lineage-tracing experiment confirmed that EE-induced β3-Tubulin+ immature neuronal cells present in meninges originated from GLAST+ radial glia cells. To investigate the molecular mechanism responsible for this response, we studied the BDNF-TrkB interaction. Treatment with ANA-12, a TrkB non-competitive inhibitor, abolished the EE-induced increase of β3-Tubulin+ immature neuronal cells in meninges. Overall these data showed, for the first time, that the meningeal niche responded to neurogenic stimuli by increasing the immature neuronal population through TrkBmediated signaling. A better understanding of the neurogenic stimuli effects on NPs in meninges may be useful to improve the effectiveness of depression and mood disorders treatments.


INTRODUCTION
Mammal adult neurogenesis can be modulated by several stimuli including antidepressant treatment and exposure to enriched environment (EE) [1]. Fluoxetine, a Selective-Serotonin Reuptake Inhibitor (SSRI) antidepressant [2,3], as well as other antidepressants, has been shown to induce hippocampal neurogenesis [3][4][5]. Similarly, EE also increases hippocampal neurogenesis and its therapeutic effect has been used to ameliorate brain injury outcomes [6]. The molecular mechanisms responsible for the hippocampal neurogenic modulation by EE and fluoxetine have not been fully elucidated; however, a major role of the BDNF-TrkB signaling pathway has been described by several groups [7][8][9][10]. Besides the fluoxetine and EE effects on the hippocampal neurogenesis, little is known about their effects on other less studied brain NSC niches. [11]. Recently, a subset of meningeal-residing neural progenitors has been identified both during development and adulthood in different mammals, including humans [11][12][13][14][15][16]. Single-cell RNA sequencing (scRNAseq) analysis of meningeal cells identified a small fraction of cells with a signature corresponding to radial glia-like cells, expressing GLAST and a population with a neuroblast signature expressing β3-Tubulin [17]. These neural precursors can migrate from the meninges to the brain parenchyma and differentiate into functional cortical neurons or oligodendrocytes [17,18]. Meningeal stem cell niche is composed by specific trophic extracellular matrix components called fractones and it is able to sense and respond to neurotrophic stimuli including FGF2, NGF, and EGF [19][20][21]. Noteworthly, NPs in meninges have been shown to react to CNS diseases, including spinal cord and brain injuries, amygdala lesion, stroke and progressive ataxia [13,22]. Meninges may therefore represent a functional niche for neural progenitors, however, the response of NPs in meninges to neurogenic stimuli such as drugs (i.e., fluoxetine) [23] or EE tasks [24] has never been assessed. In this work we investigated the effects and the molecular mechanisms of menigneal niche response to pro-neurogenic stimuli including antidepressant treatment and EE exposure. We found that fluoxetine and EE increased the number of neural progenitors in meninges. Lineage tracing experiment further confirmed that EE-induced NPs in meninges derived from radial glia GLAST + cells. Pharmacological inhibition of TrkB receptor following EE administration abolished both the hippocampal and meningeal increase of neural progenitors supporting the role of TrkB-BDNF signaling in hippocampal and meningeal response to neurogenic stimuli.

Neural precursor cells and immature neurons increase in mouse brain meninges following Fluoxetine administration
Hippocampal adult neurogenesis can be enhanced by external stimuli such as antidepressant treatment [4,25,26]. Fluoxetine is a SSRI antidepressant which has been shown to ameliorate anxiety/depression-related behavior and hippocampal neurogenesis [4] . To investigate if pro-neurogenic stimuli were able to induce changes in meningeal niche, we exposed CD1 mice (4 weeks old) to chronic (4 weeks) administration of fluoxetine as described by others (Fig. 1A) [3,4,27]. We first confirmed the effectiveness of the fluoxetine treatment on animal behavior using an anxiety/Obsessive Compulsive Disorder (OCD)-evaluating behavioral test (the Marble Burying Test, MBT) [28,29]. As shown in Figure 1B-C, following 4 weeks of drug administration, the percentage of marbles buried by treated animals is significantly lower when compared with controls (percentage of marbles buried over the total marble number, CTRL: 79,33% ± 5,573, n=10; FLUOX: 31,37% ± 7,854, n=17; p=0,0002). As expected, the antidepressant was able to reduce OCD-like behavior and, thus, it suggested that fluoxetine exerted its pharmacological effect. We further confirmed fluoxetine efficacy on hippocampal neurogenesis by analyzing the expression of the proliferation marker Ki67 and of the immature neuron marker doublecortin (DCX) in the DG (Fig. 1D-G). In line with previous observations, we found that the number of Ki67 + cells and the extension of the DCX + expression area of the dentate gyrus (DG) were significantly higher in fluoxetine-treated mice compared to control mice (number of Ki67 + cells per mm 2 of DG area, CTRL 0,007801 ± 0,004454, n=3; FLUOX: 0,02541 ± 0,003082, n=3; p=0,0313) (percentage of DCX + area over the total DG area, CTRL: 1,381% ± 0,1229, n=3; FLUOX: 2,403% ± 0,1472, n=3; p=0,0060). These data confirmed fluoxetine treatment effectiveness, supporting its known effect on both OCD-like behavior and hippocampal neuronal differentiation [3,4,27]. A similar approach was used to characterize the possible antidepressant-induced effects on meninges. We first analyzed mouse brain meningeal cell number and proliferation following 4 weeks of fluoxetine treatment. The meningeal nuclei number as well as the number of cells expressing the proliferation marker Ki67 didn't change following the treatment ( Fig. 1H-K). NPs in meninges are characterized by features of radial glia-like cells expressing GLAST and of immature neurons expressing β3-Tubulin [13]. We found that both GLAST + neural precursors and β3-Tubulin + cells increased in the treated group compared to the control (Fig.1L-O) (number of GLAST + cells per 2 mm of cross-sectioned meninges: CTRL: 9,167 ± 0,9458, n=6; FLUOX: 13,17 ± 1,352, n=6; p=0,0358; number of β3-Tubulin + cells per 2 mm of cross-sectioned meninges: CTRL: 14,6 ± 1,72, n=5; FLUOX: 32,6 ± 3,736, n=5; p=0,0024). These data suggested that meningeal niche responded to fluoxetine increasing NPs number. We further investigated the impact of the antidepressant treatment on the meningeal niche by analyzing the extracellular matrix components and the innate immune cell macrophages [5,30]. The meningeal niche is endowed of fractones, small laminin-based structures able to capture growth factors and exerting a trophic role for neural stem cells niche [31]. We therefore, analyzed the distribution of the fractones by analysing laminin + puncta [12]. Importantly, the number of fractones was significantly higher in the fluoxetine-treated group when compared to controls (Fig.1P-Q) (number of fractones per 2 mm of cross-sectioned meninges: CTRL: 59 ± 0,5774, n=3; FLUOX: 76,33 ± 2,186, n=3; p= 0,0016), suggesting that a trophic activation of the meningeal niche was induced in response to neurogenic stimuli. Finally, as immune cells were proven to have an effect on neurogenesis [32], we assessed the number of macrophages in meningeal niche following fluoxetine treatment. We observed a trend suggesting a possible increase of CD68 + cells in fluoxetine treated mice, albeit the difference was not of statistical significance (Fig.1R-S). Altogether these data indicated that meningeal niche reacted to fluoxetine treatment by increasing neural precursors, immature neurons and the trophic ECM fractones suggesting sensitivity of meningeal NPs to neurogenic stimuli. acting on the hippocampal niche. Neural precursor cells and immature neurons change their representation in mouse brain meninges during EE EE is a pro-neurogenic stimulus known to have an antidepressant-like effect on hippocampal neurogenesis. In order to assess the effectiveness of the EE treatment on the meningeal niche, we exposed to EE CD1 mice for 7 days and then analyzed mouse brain by immunofluorescence and confocal analysis ( Fig.2A). To validate EE treatment, we first assessed its impact on hippocampal neurogenesis by analyzing the number of cells expressing of the proliferation marker Ki67 and the extension of the DCX-labelled area in the DG (Fig.2B-E). The number of Ki67 + cells and the DCX + area in DG of treated mice was significantly higher compared to control mice (number of Ki67 + cells per mm 2 of DG area: CTRL: 0,02625 ± 0,005611, n=3; EE: 0,09301 ± 0,01854, n=3; p=0,0261; percentage of DCX + area over DG total area: CTRL:1,664% ± 0,3338, n=3; EE: 3,061% ± 0,2613, n=3; p=0,0300). In line with the literature data, these results confirmed that EE increased hippocampal neurogenesis [7,33,34]. To investigate if EE affected meningeal niche, we analyzed the number of cell nuclei and the expression of cell proliferation marker Ki67 in 2 mm of cross-sectioned brain meninges ( Fig.2F-G). We didn't observe differences in both total cell number and proliferation index in meninges of control and EE treated group. Then, we analyzed in meninges the distribution of radial glia-like cells expressing GLAST and of immature neurons expressing β3-Tubulin ( Fig.2H-K). We found a significant increase of GLAST + and β3-Tubulin + cells in treated mice as compared to the control group (number of GLAST + cells per 2 mm of cross-sectioned meninges: CTRL: 8 ± 0,5774, n=3; EE: 12,17 ± 0,6009, n=3; p=0,0075; number of β3-Tubulin + cells per 2 mm of cross-sectioned meninges: CTRL: 11 ± 1,528, n=3; EE: 20,67 ± 1,667, n=3; p=0,0129). In line with the results obtained by fluoxetine administration, these data indicated that meningeal niche responded to the EE neurogenic stimulus by increasing NP cell number. We then focused our analysis on the evaluation of the meningeal trophic and immune state in response to this pro-neurogenic stimulus via quantification of the ECM component fractones and macrophages (Fig.2L-O). Similarly to fluoxetine experiments, we found a statistically significant increase in the meningeal fractones and in CD68 + cells after exposure to EE (number of fractones per 2 mm of cross-sectioned meninges: CTRL: 47 ± 11,58, n=4; EE: 94,25 ± 10,17, n=4; p=0,0220; number of CD68 + cells per 2 mm of cross-sectioned meninges: CTRL: 4,7 ± 0,05774, n=3; EE: 6,687 ± 0,5203, n=3; p= 0,0192). These data indicated that neural precursor cells, immature neurons, fractones and macrophages significantly increased in brain meninges after EE exposure, further supporting the responsiveness of meningeal niche to pro-neurogenic stimuli.

EE-induced NPs in meninges derive from GLAST + radial glia progenitors
To identify the lineage of the immature neurons which increased in meninges after the exposure to a pro-neurogenic stimulus, we took advantage of an inducible transgenic mouse model for radial-glial cells. GLAST-Cre ERT2 mice [35] intercrossed with the CAG-CAT-EGFP reporter line [36] (GLAST-GFP) allowed to label by GFP all the GLAST + cells and their progeny following tamoxifen administration. After three days of tamoxifen induction via oral gavage, we subjected the mice to two weeks of EE (Fig.3A) and then analysed the brain and meninges of control and treated mice [35]. At first, we confirmed the effectiveness of the EE protocol in GLAST-GFP mice by assessing the increase of DCX expression in the DG of treated mice (Fig.3B, D) (percentage of DCX + area over the total DG area, CTRL: 1,021 ± 0,1381, n=4; EE: 1,833 ± 0,319, n=3; p= 0,0485). In line with the results obtained in CD1 mice, we observed an increase of immature neurons in meninges as a response to EE exposure (Fig.3 C, E) (number of β3-Tubulin + cells per 2 mm of cross-sectioned meninges: CTRL: 11,33 ± 0,3333, n=3; EE: 18,3 ± 2,197, n=4; p= 0,0447). We then analyzed the distribution and fate of GFP + GLAST-derived progenitors. As expected, we found GFP + cells in SVZ, DG and meninges both in EE-treated and control mice (Fig.3F) [35]. In order to assess the fate of the GFP + cells in meninges following EE exposure, we analyzed the expression of the β3-Tubulin in these cells ( Fig.3G-H). We found the presence of a double positive GFP + / β3-Tubulin + cell population in EE treated mice meninges; on the contrary, double positive cells were absent in control mice, suggesting that immature neurons differentiated from GLAST derived cells (number of GFP + / β3-Tubulin + cells per 2 mm of cross-sectioned meninges: CTRL: 0 ± 0, n=3; EE: 1,478 ± 0,2629, n=8; p= 0,0088). Altogether, these data indicate that increased β3-Tubulin + immature neurons in meninges following EE exposure originated from radial glia GLAST + cells.

DISCUSSION
In this study, we described, for the first time, the meningeal niche response to neurogenic stimuli. Following fluoxetine and EE administration, meninges increased the number of immature neurons. In addition, administration of the TrkB blocker ANA-12 [38] inhibited this response, suggesting that the effects of neurogenic stimuli on the meningeal niche is at least in part mediated by BDNF receptor signaling. Hippocampal neurogenesis has been extensively investigated with pro-neurogenic paradigms like EE and antidepressant treatments. However, the effect of neurogenic stimuli on novel less-known NSC niches, like the meningeal one, is still unexplored [4,7,33]. We first verified the effectiveness of the pro-neurogenic stimuli active on hippocampal neurogenesis, then we investigated meningeal niche responses. Strikingly, in both fluoxetine and EE administration, we observed an increase in GLAST + NPs and in β3-Tubulin + immature neurons, without cellular proliferation. The meningeal response to neurogenic stimuli partially differs from the hippocampal one [4,26,33,39,40], as there's no apparent increase in cellular proliferation. However, the overall increase of immature neurons was observed in both meninges and hippocampus.
Meningeal niche was already shown to be able to sense and respond to different type of stimuli both in physiological and pathological conditions. Administration of FGF-2 and NGF in meninges induced hyperplastic changes within the meninges of the rat and monkey [20,21]. Injuries, including spinal cord injury (SCI) [22], progressive ataxia [41] and brain stroke [42] were able to increase the number of meningeal-derived doublecortin (DCX) positive immature neurons. Aside from effects on immature neurons and neural progenitors, the pro-neurogenic stimuli used in our study were also able to remodel in a significant way the extracellular matrix. In both neurogenic paradigms, we observed an increase in the fractones, specialized ECM components of the neurogenic niche able to retain trophic factors [43] suggesting an overall remodeling of the meningeal niche. As the presence/increase of fractones in a NSC niche has been associated with neurogenesis, this further corroborates the hypothesis that the responsiveness of meninges to EE and fluoxetine treatment is a form of neurogenesis, albeit different from the hippocampal one [31]. While the mechanisms underlying the pro-neurogenic effects of antidepressants and environmental enrichment have been not entirely clarified, the pivotal role played by BDNF in these contexts has been shown [4,7]. We examined BDNF role using ANA-12, a small molecule acting as a TrkB non-competitive inhibitor [38]. We observed that the effects on β3-Tubulin and TrkB expression induced by EE were partially reverted by ANA-12 administration. The number of CD68 + cells was not altered by the TrkB inhibitor, while a reduction trend in the number of meningeal fractones was observed suggesting that macrophage activity to generate ECM was partially reduced [43]. To identify the origin of the immature neurons emerging in meninges after the exposure to the neurogenic stimuli, we took advantage of a transgenic model used for radial-glia (RG) cells tracing [35,36]. We observed that GFP + radial glia derived neural precursors co-expressing TrkB and β3-Tubulin increased in meninges following EE administration suggesting the RG origin of the immature neurons. The RG origin of neurons present in the meningeal niche was already characterized in the healthy newborn mouse [17]; here we showed for the first time their presence in the adult mouse brain. The increase in the number of immature neurons observed in meninges after the pro-neurogenic stimulus may be due to differentiation of RG endogenous cells already present in meninges or to neural precursors migrating from brain neurogenic niche to the meninges. While the findings regarding the generation of immature neurons in brain meninges after exposure to neurogenic stimuli are completely novel, one question, out of many, remains to be answered: what is the function of these cells? In recent years, the presence of "stand-by" neuroblasts expressing markers of neural precursors (like nestin or GLAST) or of migrating cells (like DCX) while being in a quiescent state (no expression of Ki67, no BrdU incorporation) was described [11,[44][45][46]. Those quiescent cells were found in classical NSC niches, like the subgranular [47,48] and the subventricular zone [49], but also in newly described niches like the cortex, the striatum and the meninges themselves [11,17,47,[50][51][52]. The role covered by those stand-by neuroblasts is still object of speculation and hypothesis, as little is known about their function in the adult brain. Our study showed that, at least in meninges, quiescent precursors are sensitive to proneurogenic stimuli. This opens the stage to further investigations that may help clarifying the function of immature neurons in the regeneration and/or repair processes occurring in adult brain.

Animals
Animal housing and all experimental procedures were approved by the Istituto Superiore della Sanita` (I.S.S., National Institute of Health), Italy and the Animal Ethics Committee (C.I.R.S.A.L., Centro Interdipartimentale di Servizio alla Ricerca Sperimentale) of the University of Verona, Italy (authorization number: 237/2016-PR; date of approval: 3 rd March 2016; protocol number: 56DC9.13). Wild-type (WT) CD1 mice and GLAST -GFP transgenic mice [35,36] were used for all the experiments.

CD1 mice Fluoxetine administration
Four weeks old CD1 male and female mice (n=25) were treated via oral administration of fluoxetine (Fluoxetine Hydrochloride, LRAA9180, Sigma-Aldrich, St. Louis) for 4 consecutive weeks. The drug was dissolved into the water (0.16 mg/ml concentration) contained in the dispenser normally present in mice cages, and mice were able to access freely the water containing the drug. As fluoxetine is light-sensitive, a tinfoil sheet was used to cover the water dispenser, in order to avoid any kind of light-induced change on the drug. The drug-containing water was changed by the operator two times per week and the dispenser was weighted to evaluate the average water consumption for every animal. On the basis of this evaluation, on average, animals took 29 mg/kg/die of fluoxetine via parenteral administration. The control group consisted in equally aged CD1 male and female mice (n=15) which were administrated normal water.

CD1 mice exposure to Enriched Environment
Seven weeks old CD1 male mice (n=6) were housed together in a single rat cage for 1 week of time. A running wheel and nesting material were always present in the cage, while other toys (stairs, cardboard rolls and marbles) were added alternatively to the cage, in order to preserve novelty, according to what is shown in Table 1. Animals were sacrificed after 7 days of EE exposure. Control animals were equally aged CD1 male mice (n=5) housed singularly for 1 week of time. The cage was a normal mouse cage and no toys were introduced in the cages for the whole experiment. Animals were sacrificed 7 days after the start of the experiment. TABLE 1 Scheme of the toys differentially used to perform the environmental enrichment on 7 weeks CD1 male mice. The X in the square states that the toy (raw) was present in the cage at that specific day (column). If a cell is empty, it means that the toy (row) wasn't present at that specific day (column).

CD1 mice exposure to EE and TrkB inhibitor ANA-12
Seven weeks old CD1 male mice (n=6) were injected intraperitoneally with 10 µl/g of TrkB inhibitor ANA-12 [38] dissolved in sunflower seed oil + 1% DMSO at a 0,1 mg/ml concentration three consecutive days before the start of the experiment and at the third day of the experiment. This administration protocol was adapted from Moy et al 2019 [37]. After the injections, mice were either subjected to EE (n=3) as described above (CD1 mice exposure to Enriched Environment). Control vehicle animals were injected with vehicles only at the same time points as the ANA-12 treated ones, and then they were subjected to EE treatment for the same period of time.

GLAST-GFP exposure to EE
GLAST-Cre ERT2 mice [35] intercrossed with the CAG-CAT-EGFP reporter line [36] (GLAST-GFP) that allow to label by GFP all the GLAST + cells and their progeny following tamoxifen administration, creating the GLAST-GFP strain. Seven to ten weeks old GLAST-GFP mice were induced using 3 daily Tamoxifen (T5648-1G, Sigma-Aldrich, St. Louis) gavage before the start of the EE protocol. Tamoxifen was dissolved into sunflower seed oil at a 30mg/ml concentration and the mice received 3.5mg of Tamoxifen for 35g of bodyweight. Animals were then left 2 days alone to recover from the handling and then were either subjected to EE (n=3) or control treatment (n=4) as previously described. Due to the mouse strain the treatment was prolonged to 2 weeks.

Marble Test Administration
Behavioral marble test was administrated at the end of the 3 rd week of fluoxetine treatment. Animals were individually placed into a new cage containing 15 equally spaced marbles placed over approximately 5 cm of saw dust, and their behaviour was videorecorded for 30 minutes. The number of buried marbles (criterium of at least ¾ of its surface under the saw dust) was blindly assessed.

Tissue Preparation and Immunofluorescence
All animals were anesthetized by intraperitoneal injection of Zoletil (50 mg/kg) and Xylazine (7 mg/kg). Once the pedal reflex was lost, animals were sacrificed by intracardial perfusion of PBS with 4% paraformaldehyde (PFA)/4% sucrose (pH 7.4) solution. Brains were extracted, fixed in 4% PFA solution and transferred into 10% and subsequently 30% sucrose solution. By cryostat cutting, 35 µm thick medio lateral sagittal brain sections were obtained and processed by immunofluorescence as previously described [53]. Immunostaining on cryosections was performed after 30 minutes incubation in blocking solution (PBS 1X with 0.25% Triton X-100, 2% BSA). If required by the specific antibody combination, mouse serum (1:100) was added during incubation in blocking solution. Sections were then incubated with primary antibodies in blocking solution overnight at 4°C. After rinsing 6 times for 5 minutes in blocking solution, appropriate secondary antibodies were applied for 4 hours at room temperature. After final washing steps in blocking solution and then in PBS, nuclear staining with 4',6-Diamidino-2-Phenylindole Dihydrochloride (DAPI, Molecular Probes-Thermo Fisher Scientific) or TO-PRO™-3 Iodide (TO-PRO-3, Molecular Probes-Thermo Fisher Scientific) was performed and slides were mounted using 1,4-Diazabicyclo[2.2.2]octane (DABCO,Sigma-Aldrich). Staining for the nuclear marker of proliferation Ki67 required a different blocking solution (PBS 1X with 0.5% Triton X-100, 2% BSA).
For immunofluorescence using two rabbit antibodies, one conjugated and one nonconjugated, the following protocol was developed. Cryosectioned sagittal sections were obtained as previously described. Immunostaining was performed after a 30 minutes incubation in blocking solution (PBS 1X with 0.25% Triton X-100, 2% BSA). Sections were then incubated with the non-conjugated rabbit primary antibody in blocking solution overnight at 4°C. After rinsing 6 times for 5 minutes in blocking solution, the appropriate secondary antibody was applied for 4 hours at room temperature. Following additional rinsing in blocking solution 6 times for 5 minutes, the GFP-conjugated antibody was added to the sections, which were incubated overnight at 4°C. Final washing steps, nuclear staining and mounting procedure were performed as previously described.

Quantitative Analysis
Quantification of different marker and nuclei was done by counting positive cells above the basal lamina (identified by laminin reactivity) in at least 18 sections for each experimental group (n ≥ 3 animals analyzed). For analysis on post hoc tissues, imaging of the immunostained slices was performed using an Eclipse Ti Nikon microscope (Nikon, Tokyo, Japan) and a Zeiss L710 confocal microscope (Carl Zeiss, Munich, Germany). Acquisition parameter settings (pinhole, gain, offset, laser intensity) were kept fixed for each channel in different sessions of observation at the fluorescence and confocal microscope.

Statistics
Data are expressed as mean ± SEM. Statistical differences were calculated by Student's t-test using GraphPadPrism (GraphPad Inc., La Jolla, CA). p ≤ 0.05 was considered statistically significant.

ACKNOWLEDGMENTS
We acknowledge the CIRSAL (Centro Interdipartimentale di Servizi per la Ricerca che utilizza Animali da Laboratorio) for the help in the management of the animals needed for the experimental procedures. We thank Marzia Di Chio for her helpful technical assistance, Alberto Poli, Giulia Lucianer and Nicola Piazza for their help with immunofluorescence analysis. We acknowledge Prof. M. Goetz for kindly providing Glast-Cre ERT2 mice.

AUTHOR CONTRIBUTIONS
SZ performed experiments, analyzed and interpreted data, wrote the manuscript, final approval of manuscript; AC performed experiments, analyzed and interpreted data, wrote the manuscript, final approval of manuscript; AP collaborated for conceptualization and experiments on fluoxetine treatment; AA and FB contributed in confocal acquisition; GFF, CC, FB and ID reviewed the manuscript, final approval of manuscript and provided funding; FB and ID conceptualized the study; ID supervised the study.