OUP user menu

Prenatal pharmacotherapy rescues brain development in a Down’s syndrome mouse model

Sandra Guidi, Fiorenza Stagni, Patrizia Bianchi, Elisabetta Ciani, Andrea Giacomini, Marianna De Franceschi, Randal Moldrich, Nyoman Kurniawan, Karine Mardon, Alessandro Giuliani, Laura Calzà, Renata Bartesaghi
DOI: http://dx.doi.org/10.1093/brain/awt340 380-401 First published online: 12 December 2013

Summary

Intellectual impairment is a strongly disabling feature of Down’s syndrome, a genetic disorder of high prevalence (1 in 700–1000 live births) caused by trisomy of chromosome 21. Accumulating evidence shows that widespread neurogenesis impairment is a major determinant of abnormal brain development and, hence, of intellectual disability in Down’s syndrome. This defect is worsened by dendritic hypotrophy and connectivity alterations. Most of the pharmacotherapies designed to improve cognitive performance in Down’s syndrome have been attempted in Down’s syndrome mouse models during adult life stages. Yet, as neurogenesis is mainly a prenatal event, treatments aimed at correcting neurogenesis failure in Down’s syndrome should be administered during pregnancy. Correction of neurogenesis during the very first stages of brain formation may, in turn, rescue improper brain wiring. The aim of our study was to establish whether it is possible to rescue the neurodevelopmental alterations that characterize the trisomic brain with a prenatal pharmacotherapy with fluoxetine, a drug that is able to restore post-natal hippocampal neurogenesis in the Ts65Dn mouse model of Down’s syndrome. Pregnant Ts65Dn females were treated with fluoxetine from embryonic Day 10 until delivery. On post-natal Day 2 the pups received an injection of 5-bromo-2-deoxyuridine and were sacrificed after either 2 h or after 43 days (at the age of 45 days). Untreated 2-day-old Ts65Dn mice exhibited a severe neurogenesis reduction and hypocellularity throughout the forebrain (subventricular zone, subgranular zone, neocortex, striatum, thalamus and hypothalamus), midbrain (mesencephalon) and hindbrain (cerebellum and pons). In embryonically treated 2-day-old Ts65Dn mice, precursor proliferation and cellularity were fully restored throughout all brain regions. The recovery of proliferation potency and cellularity was still present in treated Ts65Dn 45-day-old mice. Moreover, embryonic treatment restored dendritic development, cortical and hippocampal synapse development and brain volume. Importantly, these effects were accompanied by recovery of behavioural performance. The cognitive deficits caused by Down’s syndrome have long been considered irreversible. The current study provides novel evidence that a pharmacotherapy with fluoxetine during embryonic development is able to fully rescue the abnormal brain development and behavioural deficits that are typical of Down’s syndrome. If the positive effects of fluoxetine on the brain of a mouse model are replicated in foetuses with Down’s syndrome, fluoxetine, a drug usable in humans, may represent a breakthrough for the therapy of intellectual disability in Down’s syndrome.

  • Down’s syndrome
  • brain development
  • intellectual disability
  • neurogenesis impairment
  • preventive pharmacotherapy

Introduction

Down’s syndrome is a genetic disorder of high prevalence (1 in 700–1000 live births) caused by trisomy of chromosome 21. The main feature of all Down’s syndrome cases is intellectual disability, which is combined with a large range of variable traits (Rachidi and Lopes, 2008). The cognitive impairment that characterizes subjects with Down’s syndrome is thought to be related to the reduction in the size of various brain regions, including the cerebral cortex, the hippocampus and the cerebellum, and to the reduction in cortical neuron density (Rachidi and Lopes, 2008). Accumulating evidence in individuals with Down’s syndrome and Down’s syndrome mouse models shows that brain hypotrophy is because of neurogenesis impairment that starts from foetal developmental stages. Prenatal and early post-natal proliferation defects have been documented in the neocortex, dentate gyrus and cerebellum of mouse models (Haydar et al., 2000; Lorenzi and Reeves, 2006; Roper et al., 2006; Chakrabarti et al., 2007; Contestabile et al., 2007) and in the hippocampal region and cerebellum of foetuses with Down’s syndrome (Contestabile et al., 2007; Guidi et al., 2008, 2011).

Most of the pharmacotherapies designed to improve cognitive performance in Down’s syndrome have been attempted in mouse models during adult life stages (Bartesaghi et al., 2011) and there are ongoing clinical trials in adults and adolescents affected with Down’s syndrome. However, prenatal testing makes it possible to intervene prenatally to improve brain development in foetuses affected with Down’s syndrome, provided that there are safe compounds with a positive impact on brain development (Guedj and Bianchi, 2013). Interestingly, recent evidence shows that a prenatal therapy with neuroprotective peptides restores learning in the Ts65Dn mouse model of Down’s syndrome, indicating that a prenatal therapy may have a positive outcome (Incerti et al., 2012). Save for this study, no other investigations have explored the possibility of pharmacologically correcting neurodevelopmental defects in Down’s syndrome during the most critical time windows for brain development.

Individuals with Down’s syndrome are characterized by defects in the serotonergic system (Whitaker-Azmitia, 2001). The serotonin 5-HT1A receptor (5-HT1A-R) decreases to below normal levels by birth (Bar-Peled et al., 1991) and reduced serotonin levels are present in adults with Down’s syndrome (Risser et al., 1997). Similar to individuals with Down’s syndrome, Ts65Dn mice exhibit a reduced expression of the 5-HT1A-R at neonatal life stages and reduced serotonin levels in adulthood (Bianchi et al., 2010; Guidi et al., 2013). In view of the key role of serotonin in neurogenesis and dendritic development (Faber and Haring, 1999; Whitaker-Azmitia, 2001), the alteration of the serotonergic system in the trisomic brain may be a key determinant of neurogenesis and neuron maturation impairment. If so, treatments that increase serotonin availability may prove useful in compensating for this defect.

Antidepressant drugs, including fluoxetine, a selective serotonin re-uptake inhibitor, increase cell proliferation in the adult dentate gyrus of normal animals (Malberg and Blendy, 2005) and inhibition of serotonin re-uptake by fluoxetine stimulates dendritic maturation of newborn hippocampal granule cells (Wang et al., 2008). Recent evidence shows that selective serotonin re-uptake inhibitors, including fluoxetine, reduce the expression of p21, an inhibitor of cell cycle progression, which may explain the positive effects of selective serotonin re-uptake inhibitors on cell proliferation. Based on evidence that the serotonergic system is altered in the trisomic brain and that treatment with fluoxetine is able to restore cell proliferation in the dentate gyrus of adult Ts65Dn mice (Clark et al., 2006), in a previous study we treated neonate Ts65Dn mice with fluoxetine and found that this treatment fully restored neurogenesis and dendritic development in the hippocampal dentate gyrus (a structure where neurogenesis takes place mainly post-natally) and hippocampus-dependent learning (Bianchi et al., 2010; Guidi et al., 2013). As in all other brain regions neurogenesis is mainly a prenatal event (Chan et al., 2002; Brazel et al., 2003), treatments aimed at correcting the widespread neurogenesis failure that characterizes Down’s syndrome should be administered during pregnancy. Correction of neurogenesis at the very first steps of brain formation may, in turn, rescue improper brain wiring. It is expected that once the brain is properly formed it can retain its basic features. Thus, early interventions capable of correcting the aberrant developmental programs of the trisomic brain may have enduring effects, potentially across life.

As there is no evidence that it is possible to rescue brain development in Down’s syndrome during embryonic life stages, the aim of our study was to establish whether it is possible to correct the various neurodevelopmental alterations that characterize the trisomic brain through a prenatal pharmacotherapy with fluoxetine, a drug that is able to restore post-natal hippocampal neurogenesis (Clark et al., 2006; Bianchi et al., 2010). For our study we have used the Ts65Dn mouse, a model of Down’s syndrome that recapitulates many neurodevelopmental defects of the human condition.

Materials and methods

Mouse colony

Female Ts65Dn mice carrying a partial trisomy of chromosome 16 (Reeves et al., 1995) were obtained from Jackson Laboratories and maintained on the original genetic background by mating them with C57BL/6JEi x C3SnHeSnJ (B6EiC3) F1 males. Animals were karyotyped as previously described (Reinholdt et al., 2011). The day of birth was designed as post-natal Day (P) zero. A total of 110 mice were used. The animals had access to water and food ad libitum and lived in a room with a 12:12 h dark/light cycle. Experiments were performed in accordance with the Italian and European Community law for the use of experimental animals and were approved by Bologna University Bioethical Committee. All efforts were made to minimize animal suffering and to keep the number of animals used to a minimum.

Experimental protocol

Ts65Dn females (n = 29) were bred with C57BL/6JEi x C3SnHeSnJ (B6EiC3) F1 males (n = 29). Conception was determined by examining the vaginal plug. Pregnant females received a daily subcutaneous injection of either fluoxetine (Sigma-Aldrich) in 0.9% NaCl solution (dose: 10 mg/kg) (n = 14) or saline (n = 15) from the embryonic Day 10 to the day of birth (embryonic Days 20/21). On post-natal Day 2 the progeny of treated and untreated females received an intraperitoneal injection (150 µg/g body weight) of 5-bromo-2-deoxyuridine (BrdU; Sigma) in 0.9% NaCl solution and were sacrificed either after 2 h (on post-natal Day 2; six to nine animals for each group) or after 43 days (on post-natal Day 45; five to six animals for each group). In the period post-natal Days 40–44, these animals were behaviourally tested (see below). Additional animals (n = 58) were not injected with BrdU and were sacrificed on post-natal Day 45. All these animals were used for behavioural testing in the period post-natal Days 40–44. The brain of some of these animals (n = 4 for each experimental condition) was used for western blotting and the head of other animals (n = 4 for each experimental condition) were stored in PBS and used for a magnetic resonance study and evaluation of skull bone density (see Supplementary material). The body weight of post-natal Days 2 and 45 animals was recorded before sacrifice. After sacrifice, brain was excised and weighed. The number of animals used for the experimental procedures described below is summarized in Supplementary Table 1.

Histological procedures

Post-natal Day 2 animals were decapitated and the brain was removed. The rostral brain (forebrain plus mesencephalon) was separated from the hindbrain (cerebellum plus pons and medulla). The rostral brain was cut along the midline and fixed by immersion in Glyo-Fixx as previously described (Bianchi et al., 2010). Each brain part was embedded in paraffin and cut in series of 8-µm thick coronal (rostral brain) or parasagittal (hindbrain) sections that were attached to poly-lysine coated slides. Post-natal Day 45 animals that had received BrdU on post-natal Day 2 were anaesthetized with ether, transcardially perfused and the different brain parts were frozen and stored at −80°C, as previously described (Bianchi et al., 2010). The right hemisphere was cut with a freezing microtome in 30-µm thick coronal sections that were serially collected in PBS. The brain coordinates indicated below for post-natal Day 2 mice refer to the coordinates for post-natal Day 0 mice reported in the Atlas of the developing mouse brain (Paxinos et al., 2007) and the brain coordinates for post-natal Day 45 mice refer to the coordinates in the atlas The mouse brain in stereotaxic coordinates (Franklin and Paxinos, 1997).

Immunohistochemistry

Immunohistochemistry was carried out as previously described (Contestabile et al., 2007; Bianchi et al., 2010; Guidi et al., 2013). The type and dilution of the antibodies used for immunohistochemistry and their purpose are summarized in Supplementary Table 2. To test the specificity of mouse monoclonal antibodies used, sections were submitted to the same immunohistochemistry protocol but without using the primary antibody. Immunoreactivity in these control sections was not observed. A brief outline of the procedures is reported below. More details are reported in the Supplementary material.

BrdU, 5-HT1A receptor, cleaved caspase-3 and calbindin immunohistochemistry

One of 20 sections from the brains of post-natal Day 2 mice were incubated with the appropriate antibodies, as indicated in Supplementary Table 2. For BrdU, 5-HT1A receptor and cleaved caspase-3 immunohistochemistry, sections (n = 14–18) were taken from the beginning of the lateral ventricle to the end of the hippocampal formation (brain coordinates: 2.19–4.83 mm). BrdU, cleaved caspase-3 and calbindin immunohistochemistry was additionally carried out in sections close to the cerebellar midline (n = 4–6 sections).

Double fluorescence immunohistochemistry

One of six sections (n = 10–17) from the hippocampal formation of post-natal Day 45 mice (brain coordinates: from −0.38 to −3.78) was incubated with a primary anti-BrdU antibody and either a neuronal-specific nuclear protein (NeuN) or glial fibrillary acidic protein (GFAP) primary antibody as indicated in Supplementary Table 2. Detection was performed with the secondary antibodies indicated in Supplementary Table 2.

Ki-67 immunohistochemistry

One of 12 sections (n = 10–12) from the beginning of the lateral ventricle to the end of the hippocampal formation of post-natal Day 45 mice (brain coordinates: from +1.42 to −3.78 mm) was incubated with the antibodies indicated in Supplementary Table 2.

DCX, SYN, PSD95 and S100B immunohistochemistry

One out of six sections from the beginning to the end of the hippocampal formation of post-natal Day 45 mice [n = 10–16 sections for doublecortin (DCX) and n = 3 sections for synaptophysin (SYN), postsynaptic density 95 (PSD95) and S100B (S100 calcium binding protein B) immunohistochemistry] was incubated with the antibodies indicated in Supplementary Table 2.

Nissl method

One of 20 sections from the rostral brain and cerebellum of post-natal Day 2 mice and one of six sections from the hippocampal formation of post-natal Day 45 mice were stained with toluidine blue according to the Nissl method.

Western blotting

Samples from the hippocampal formation were used to analyse the levels of Erk1/2 and phosphorylated-Erk1/2, p21, PSD95 and SYN, as indicated in Supplementary material. The antibodies used are summarized in Supplementary Table 2.

Measurements

Number of BrdU-positive cells

BrdU-positive cells were sampled in the subventricular zone, dentate gyrus, neocortex, striatum, thalamus, hypothalamus, mesencephalon and cerebellum of post-natal Day 2 and in the dentate gyrus of post-natal Day 45 mice. As different regions of the ventricular zone/subventricular zone give origin to neurons destined to different telecephalic areas (Brazel et al., 2003), we evaluated the effects of treatment in different subregions of the subventricular zone separately, named here as follows: the rostral subventricular zone is the region that stretches from the rostral horn of the lateral ventricle to the beginning of the hippocampal formation (brain coordinates: 2.19–3.15 mm) and the caudal subventricular zone is the region that stretches from the beginning to the end of the hippocampal formation (brain coordinates: 3.27–4.84 mm). The subventricular zone regions corresponding to the dorso-lateral wall and medial wall of the lateral ventricle are called dorso-lateral subventricular zone and medial subventricular zone, respectively (Fig. 1). Cells were counted within manually-traced areas enclosing the dorso-lateral subventricular zone and medial subventricular zone. In the dentate gyrus of post-natal Day 2 mice, cells were separately counted in the hilus + subgranular zone + granule cell layer and in the molecular layer (Fig. 2). In the dentate gyrus of post-natal Day 45 mice cells were separately counted in the hilus + subgranular zone and in the granule cell layer. Cells were counted within manually-traced areas enclosing the indicated layers. BrdU-positive cells were additionally sampled in: (i) the neocortex overlying the rostral subventricular zone (Fig. 3A; brain coordinates: 2.19–3.15 mm) and caudal subventricular zone (Fig. 3B; brain coordinates: 3.27–4.84 mm); (ii) the striatum, in sections that encroached the rostral subventricular zone (Fig. 3A; brain coordinates: 2.19–3.15 mm); (iii) the dorsal one-half of the thalamus (Fig. 3B; brain coordinates: 3.27–4.23); (iv) the hypothalamus (Fig. 3B; brain coordinates: 2.67–3.87 mm); (v) the dorsal one-third of the mesencephalon (Fig. 3C; brain coordinates: 4.59–4.95 mm); (vi) the cerebellum (Fig. 3D); and (vii) the pons (Fig. 3D). BrdU-positive cells were counted within the manually traced areas indicated in Fig. 3A–D. In the cerebellum, cells were counted at the top of lobes 2 and 3 (Fig. 3D) in the outer external granular layer and inner external granular layer, within manually traced areas of ∼1200 µm2 and 1000 µm2, respectively. Cell density in the regions of interest was calculated by dividing the number of sampled cells by the volume of the sampled region. The latter was obtained by multiplying the traced areas by the nominal section thickness (post-natal Day 2: 8 µm; post-natal Day 45: 30 µm). The number of BrdU-positive cells was expressed as cells/mm3. We additionally estimated the total number of BrdU-positive cells in the subventricular zone and dentate gyrus by multiplying the number counted in the sampled sections by the inverse of the section sampling fraction (section sampling fraction = 1/20, in post-natal Day 2 animals; 1/6 in post-natal Day 45 animals).

Figure 1

Effect of embryonic treatment with fluoxetine (Fluo) on precursor proliferation in the subventricular zone of post-natal Day 2 Ts65Dn and euploid mice. (A–C and F–H) Examples of sections immunostained for BrdU and counterstained with haematoxylin from the rostral subventricular zone (A–C) and caudal subventricular zone (F–H) of an animal from each experimental group. These animals received one injection of BrdU on post-natal Day 2 and were sacrificed after 2 h. Proliferating cells were present both in the medial (mSVZ) and dorso-lateral (dlSVZ) parts of the subventricular zone. The high-magnification images in A, C, F and H were taken from the medial subventricular zone (A) and dorso-lateral subventricular zone (C) of the rostral subventricular zone and from the medial subventricular zone (F) and dorso-lateral subventricular zone (H) of the caudal subventricular zone, in the region indicated in B and G by a star and a diamond, respectively. Scale bars: A, C, F and H = 50 µm; B and G = 500 µm. (D, E, I and J) BrdU-positive cells (number per mm3) (D and I) and total number of BrdU-positive cells (E and J) in the medial subventricular zone and dorso-lateral subventricular zone of the rostral (D and E) and caudal (I and J) subventricular zone of untreated euploid (Eu; n = 9) and Ts65Dn (n = 9) mice and euploid (n = 7) and Ts65Dn (n = 6) mice treated with fluoxetine. Values in E and J represent totals for one hemisphere. Values in D, E, I and J represent mean ± standard error (SE). *P < 0.05; **P < 0.01; ***P < 0.001 (Duncan’s test after ANOVA). DG = dentate gyrus; FI = fimbria; LV = lateral ventricle.

Figure 2

Effect of embryonic treatment with fluoxetine on precursor proliferation in the dentate gyrus of post-natal Day 2 Ts65Dn and euploid mice. (A–D) Examples of sections immunostained for BrdU and counterstained with haematoxylin from the dentate gyrus of an animal from each experimental group. These animals received one injection of BrdU on post-natal Day 2 and were sacrificed after 2 h. Scale bar = 100 µm. (E–H) BrdU-positive cells (number per mm3) (E and F) and total number of BrdU-positive cells (G and H) in the hilus + subgranular zone + granule cell layer (H + SGZ + GR) (E and G) and in the molecular layer (MOL) (F and H) of the dentate gyrus of untreated euploid (n = 9) and Ts65Dn (n = 9) mice and euploid (n = 7) and Ts65Dn (n = 6) mice treated with fluoxetine. Values in (G and H) represent totals for one hemisphere. Values in E–H represent mean ± SE. *P < 0.05; **P < 0.01; ***P < 0.001, Duncan’s test after ANOVA.

Figure 3

Effect of embryonic treatment with fluoxetine on precursor proliferation in different brain regions of post-natal Day 2 Ts65Dn and euploid mice. (A–C) Examples of sections immunostained for BrdU and counterstained with haematoxylin (star) from the rostral neocortex and striatum (A), the thalamus and hypothalamus (B) and the mesencephalon (C) of post-natal Day 2 euploid mice. The Nissl-stained coronal sections in A–C show the brain regions (stippled areas) where BrdU-positive cells were sampled. The ordinal numbers in (A) indicate the cortical layers. Scale bar = 50 µm for BrdU immunostained sections and 500 µm for Nissl-stained sections. (D) Sagittal section (left) across the cerebellum and pons of a post-natal Day 2 euploid mouse immunostained for BrdU and counterstained with haematoxylin. The stippled area and the rectangles indicate the regions where BrdU-positive cells were sampled. The image on the right shows the cerebellar layers. The external granular layer is subdivided into two zones: the outer external granular layer (oEGL) that is formed by actively dividing granule cell precursors and the inner external granular layer (iEGL) that contains post-mitotic granule cells en route to the internal granular layer (IGL) and scattered granule cell precursors. Scale bar = 500 µm for the left image and 20 µm for the right image. (E–M) BrdU-positive cells (number per mm3) in the rostral neocortex (E), caudal neocortex (F), striatum (G), thalamus (H), hypothalamus (I), mesencephalon (J), outer external granular layer (K), inner external granular layer (L) and pons (M) of untreated euploid (n = 9) and Ts65Dn (n = 9) mice and euploid (n = 7) and Ts65Dn (n = 6) mice treated with fluoxetine. Values in E–M represent mean ± SE. (*)P < 0.06; *P < 0.05; **P < 0.01; ***P < 0.001, Duncan’s test after ANOVA. cCX = caudal cortex; HF = hippocampal formation; HYP = hypothalmus; L = lobe; MES = mesencephalon; MOL = molecular layer; PO = pons; rCX = rostral cortex; STR = striatum; TH = thalamus.

Analysis of phenotypes

Sections from the dentate gyrus of post-natal Day 45 animals were analysed for co-expression of either BrdU/NeuN or BrdU/GFAP. The number of cells of undetermined phenotype was calculated by subtracting from the total number of BrdU-positive cells the number of BrdU/NeuN-positive cells plus the number of BrdU/GFAP-positive cells. In each section, the area of the granule cell layer and hilus + subgranular zone were manually traced. Cell density (cells/mm3) was calculated by dividing the number of sampled cells by the volume of the sampled region. The latter was obtained by multiplying the traced areas by the nominal section thickness (30 µm). The total number of cells of each phenotype was additionally estimated by multiplying the total number counted in the series of sections by the inverse of the section sampling fraction (section sampling fraction = 1/6).

Stereology

In Nissl-stained sections from post-natal Day 2 mice the volume of the granule cell layer of the dentate gyrus and fields CA3 and CA1, cell numerical density and number of neurons were estimated as previously described (Bianchi et al., 2010). Counting frames (disectors) with a side length of 30 µm for the granule cell layer and 20 µm for the hippocampal fields and a height of 8 µm spaced in a 100 µm square grid (fractionator) were systematically used. The average number of disectors hitting the granule cell layer of the dentate gyrus, the pyramidal layer of CA3 and CA1 was 17.8 ± 0.3 in the dentate gyrus, 30.5 ± 0.5 in CA3 and 31.3 ± 0.5 in CA1, respectively. Calculation of coefficient of error (CE) of cell numerical density (West and Gundersen, 1990) gave values between 0.040 and 0.061. In post-natal Day 45 mice stereology was carried out for the dentate gyrus as previously described (Bianchi et al., 2010). In Nissl-stained sections from post-natal Day 2 mice, we evaluated cell numerical density in layers II, III and IV/VI of the neocortex overlying the rostral subventricular zone, in the striatum, thalamus, hypothalamus and mesencephalon. Cells were counted approximately in the same regions where BrdU-positive cells were sampled. Counting frames with a side length of 30 µm and a height of 8 µm spaced in a 150 µm square grid were systematically used. The average number of disectors in cortical layers II, III, and IV/VI was 22.1 ± 0.4, 21.8 ± 0.3 and 41.8 ± 0.3, respectively, and the CE of cell numerical density had values between 0.043 and 0.059. The average number of disectors in the striatum, thalamus, hypothalamus and mesencephalon was 23.2 ± 0.5, 20.6 ± 0.5, 31.5 ± 0.7 and 32.4 ± 1.0, respectively, and the CE of cell numerical density had values between 0.059 and 0.066. Cell density was expressed as number of cells/mm3. The thickness of the rostral cortex was measured by tracing at four to five locations radial lines across all cellular layers. The thickness of the external granular layer, internal granular layer and Purkinje cell layer and cell density in each layer were measured in cerebellar sections immunostained for calbindin and counterstained with the Nissl method (Fig. 5F). Cells were counted in manually traced areas at the level of the crown of lobes 2 and 3. Cell density was calculated by dividing the number of counted cells by the volume of the sampled region. The latter was obtained by multiplying the area by the section thickness (8 µm).

Dendritic morphology

The dendritic tree of DCX-positive granule cells of the dentate gyrus was traced with a dedicated software, custom-designed for dendritic reconstruction (Immagini Computer), interfaced with Image Pro Plus, as previously described (Guidi et al., 2013). The dendritic tree was reconstructed in 16–21 neurons per animal.

Synaptic proteins

Intensity of SYN or PSD95 immunoreactivity was determined by optical densitometry of immunohistochemically stained sections. Densitometric analysis was carried out as previously reported (Guidi et al., 2013). A box of 4000 µm2 was placed in the cortical layers II, III and IV-VI and a box of 400 µm2 was placed in stratum radiatum of CA1 and stratum moleculare of the dentate gyrus. Six to ten measurements were taken in each region of interest. Images immunoprocessed for SYN or PSD95 were acquired with a confocal microscope (Nikon Ti-E fluorescence microscope coupled with an A1R confocal system, Nikon). The zoom factor was 6; software Nis-Elements AR 3.2 was used and image size was 512 × 512 pixels. In each section three images from the regions of interest indicated above were captured and the density of individual puncta exhibiting SYN or PSD95 immunoreactivity was evaluated.

Behavioural testing

For the open field test, a 46 × 46 × 35 plexiglass activity cage with white walls and floor and video-tracking software (Any-Maze, Stoelting) were used (UgoBasile). The animals were placed into the activity cage and behaviour was examined for 10 min. The surface of the box was divided into two zones: the arena zone included the entire area of the floor and the standing zone was at the periphery of the floor close to the walls. The software analysed total distance traveled, average speed, total number and duration of rearing behaviour. For the novel object recognition test mice were habituated in an open-square dark-grey arena (46 × 46 × 30 cm; Ugo Basile). The arena illumination was 70 lx. Mice activity was recorded with the software ANY-maze (Stoelting). In the habituation trial animals were placed into the empty arena for 10 min. Next day, mice were placed into the same arena containing two identical objects (familiarization phase), placed in the front wall, 12 cm from the corner. Mice were placed at the midpoint of the opposite wall. After allowing 10 min to explore the objects, the mouse was returned to the home cage. Exploratory behaviour was analysed at a distance <2.0 cm from objects. Exploratory behaviour was analysed by calculating the investigation time on both objects. Tests were performed 4 h and 24 h after the familiarization phase (test trials). The time spent exploring the two objects was recorded for 10 min. Memory was expressed as exploratory time: [(time on novel object)/(total time on objects)] × 100. Contextual fear conditioning was administered as previously described (Bianchi et al., 2010).

Statistical analysis

Data from single animals were the unity of analysis. The software SPSS was used for statistical analysis. Statistical testing was performed with ANOVA followed by post hoc comparisons with Duncan's test. A probability level of P < 0.05 was considered to be statistically significant.

Results

Effect of embryonic treatment with fluoxetine on pup viability and number

Ts65Dn females treated with fluoxetine during pregnancy did not exhibit a higher abortion rate and gave birth to a number of pups similar to that of the untreated counterparts. The survival rate of the progeny was similar to that of untreated females (Table 1). This indicates that fluoxetine has no patently adverse effects on pregnancy and pup viability.

View this table:
Table 1

Effect of embryonic treatment with fluoxetine on perinatal death and litter size

Outcome
TreatmentnSpontaneous abortion (n, mean ± SE)Litter size (n, mean ± SE)Post-natal death (%; mean ± SE)
Saline150.00 ± 0.005.47 ± 0.4214.02 ± 7.04
Fluoxetine140.00 ± 0.005.00 ± 0.4812.69 ± 8.12
P-valuen.s.n.s.n.s.
  • Pregnancy outcome for Ts65Dn females that were injected with either saline (n = 15) or fluoxetine (n = 14) from gestational Day 10 to delivery. Post-natal death is expressed as percentage of deaths over number of births. Data are mean ± SE; n.s. = not significant (Duncan’s test after ANOVA).

Effect of embryonic treatment with fluoxetine on somatic development in Ts65Dn and euploid mice

To establish the overall short- and long-term effect of prenatal treatment with fluoxetine we evaluated the body weight of post-natal Days 2 and 45 mice. At both ages Ts65Dn mice had a lower body weight than euploid mice (Table 2). Fluoxetine had no effect on body weight in post-natal Day 2 euploid and Ts65Dn mice and in post-natal Day 45 Ts65Dn mice (Table 2). In contrast, treated post-natal Day 45 euploid mice had a lower body weight than the untreated counterparts (Table 2). As selective serotonin re-uptake inhibitors cause bone mineral loss (Warden et al., 2008) we examined bone mineral density in the skulls of treated mice. Confirming previous evidence (Blazek et al., 2010) the skulls of Ts65Dn mice had a lower mineral density in comparison with euploid mice. Treatment had no effect on bone mineral density either in euploid or in Ts65Dn mice (Supplementary Fig. 1).

View this table:
Table 2

Effect of embryonic treatment with fluoxetine on body weight

ConditionnMean ± SEConditionnMean ± SEP-value
Post-natal Day 2
Euploid + Sal91.90 ± 0.12Euploid + Fluo72.04 ± 0.10n.s.
Ts65Dn + Sal91.44 ± 0.11Ts65Dn + Fluo61.69 ± 0.06n.s.
P-value<0.05<0.05
Post-natal Day 45
Euploid + Sal2021.60 ± 0.71Euploid + Fluo1518.18 ± 0.77<0.01
Ts65Dn + Sal1417.72 ± 0.78Ts65Dn + Fluo1417.84 ± 0.96n.s
P-value<0.001n.s
  • Body weight (mean ± SE) in grams, of euploid and Ts65Dn mice that received either saline (Sal) or fluoxetine (Fluo) between embryonic Days 10–20/21, measured on post-natal Days 2 and 45. The P-value in the row below each age group refers to the comparison between untreated euploid (Euploid + Sal) and Ts65Dn (Ts65Dn + Sal) mice and treated euploid (Euploid + Fluo) and Ts65Dn (Ts65Dn + Fluo) mice. The P-value in the column on the right refers to the comparison between untreated and treated mice of the same genotype. n.s. = not significant (Duncan’s test after ANOVA).

Gross effects of embryonic treatment with fluoxetine on the brain of Ts65Dn and euploid mice

Untreated Ts65Dn mice had a reduced brain weight compared with euploid mice at post-natal Days 2 and 45 (Table 3). This difference disappeared after treatment with fluoxetine (Table 3). No statistically significant brain weight difference was previously detected between euploid and Ts65Dn mice at post-natal Day 8 (Chakrabarti et al., 2007) and post-natal Day 22 (Belichenko et al., 2004), though the weight values were lower in Ts65Dn mice. The discrepancy with our results may be because of the lower number of sampled brains in comparison with those examined here. On the other hand, we found no genotype-related brain weight differences at 3–4 months of age (unpublished observations), which fits with data obtained in mice aged from 3–21 months (Baxter et al., 2000) and suggests that trisomy-linked brain differences may disappear with age. To establish the effects of treatment on the size of different brain structures, we used MRI automated region detection algorithms and volume quantification (Supplementary material). We found that post-natal Day 45 Ts65Dn mice had a reduced volume in comparison with euploid mice in the hippocampus, cerebellum and neocortex. Treatment with fluoxetine impacted each brain region positively in Ts65Dn mice, without impacting on those of euploid mice (Supplementary Fig. 2).

View this table:
Table 3

Effect of embryonic treatment with fluoxetine on brain weight

ConditionnMean ± SEConditionnMean ± SEP-value
Post-natal Day 2
Euploid + Sal90.11 ± 0.002Euploid + Fluo70.11 ± 0.005n.s
Ts65Dn + Sal90.09 ± 0.003Ts65Dn + Fluo60.10 ± 0.006n.s.
P-value<0.01n.s
Post-natal Day 45
Euploid + Sal200.44 ± 0.006Euploid + Fluo150.43 ± 0.006n.s.
Ts65Dn + Sal140.41 ± 0.009Ts65Dn + Fluo140.43 ± 0.005n.s
P-value<0.05n.s
  • Brain weight (mean ± SE) in grams of euploid and Ts65Dn mice that received either saline (Sal) or fluoxetine (Fluo) between embryonic Days 10–20/21, measured on post-natal Days 2 and 45. The P-value in the row below each age group refers to the comparison between untreated euploid (Euploid + Sal) and Ts65Dn (Ts65Dn+ Sal) mice and treated euploid (Euploid + Fluo) and Ts65Dn (Ts65Dn + Fluo) mice. The P-value in the column on the right refers to the comparison between untreated and treated mice of the same genotype. n.s. = not significant (Duncan’s test after ANOVA).

Effect of embryonic treatment with fluoxetine on the proliferation rate of different populations of neural precursors in Ts65Dn mice

Neurons forming the CNS derive from neural precursors in the ventricular zone and subventricular zone surrounding the cerebral ventricles. In the mouse, which is born 20–21 days after conception, neurons forming the neocortex are mainly born during embryonic Days 11–17 and those forming the hippocampus are born between embryonic Days 10 and 18 (Angevine, 1965, 1975; Takahashi et al., 1996). Most of the neurons forming the striatum, diencephalons and mesencephalon are born between embryonic Days 11–19 (Fentress et al., 1981; Bayer et al., 1995; Suzuki-Hirano et al., 2011; Ishii and Bouret, 2012). Neurogenesis in the hippocampal dentate gyrus starts at embryonic Day 10, though most of the granule neurons (∼80%) are generated by the post-natal subgranular zone (Altman and Bayer, 1975). The Purkinje neurons of the cerebellum become post-mitotic by embryonic Day 12, whereas the precursors of the granule neurons in the external granular layer are generated starting from embryonic Day 12.5 and continue to proliferate up to post-natal Day 16 (Sudarov and Joyner, 2007). In view of the time course of neurogenesis, we treated Ts65Dn mice with fluoxetine from embryonic Day 10 to birth, with the aim to restore the bulk of neurogenesis. Neuroblasts from the embryonic ventricular zone/subventricular zone retain proliferation capacity during migration and their target regions exhibit mitotic activity that extends into the early post-natal period (Brazel et al., 2003). To establish the efficacy of embryonic treatment with fluoxetine on proliferation potency, we evaluated the number of proliferating precursors (BrdU-positive cells) in various brain regions shortly after birth (post-natal Day 2), a time at which precursor cells still exhibit a high proliferation capacity (Figs 1–3).

As there are no systematic studies on neurogenesis failure in the trisomic brain, we were interested in establishing (i) whether neurogenesis impairment involves the whole brain; and (ii) whether prenatal treatment with fluoxetine leads to a generalized rescue of neurogenesis. For this reason we examined neurogenesis in the subventricular zone, dentate gyrus, neocortex, striatum, thalamus, hypothalamus, mesencephalon, cerebellum and pons. We found that in all examined regions Ts65Dn mice had a reduced proliferation rate that ranged from −23% to −65% in comparison with euploid mice (Figs 1–3). Importantly, embryonic treatment with fluoxetine restored neurogenesis in all subregions of the subventricular zone (Fig. 1), all layers of the dentate gyrus (Fig. 2), in the rostral (Fig. 3E) and caudal (Fig. 3F) neocortex, striatum (Fig. 3G), thalamus (Fig. 3H), hypothalamus (Fig. 3I), mesencephalon (Fig. 3J), inner (Fig. 3K) and outer (Fig. 3L) external granular layer of the cerebellum, and pons (Fig. 3M). Taken together these data indicate that in the trisomic brain proliferation impairment spans from the forebrain to the hindbrain and that prenatal administration of fluoxetine has a generalized beneficial effect on neurogenesis throughout the brain.

Effect of embryonic treatment with fluoxetine on brain cellularity in Ts65Dn mice

As embryonic treatment with fluoxetine restored the number of proliferating cells throughout the brain of Ts65Dn mice, we expected the final outcome to be restoration of brain cellularity. To clarify this issue, we evaluated the number of cells in different brain regions of post-natal Day 2 mice.

Neocortex

In untreated Ts65Dn mice, the thickness of layers II–VI of the rostral neocortex [Fig. 4A(2)] and cell density in layers II [Fig. 4A(3)], III [Fig. 4A(4)], and IV–VI [Fig. 4A(5)] were reduced in comparison with euploid mice. After prenatal treatment with fluoxetine the cortical thickness and cell density became similar to those of untreated euploid mice [Fig. 4A(2–5)]. Similar findings were obtained in the caudal neocortex (data not shown).

Figure 4

Effect of embryonic treatment with fluoxetine on the stereology of the neocortex and hippocampal formation in post-natal Day 2 Ts65Dn and euploid mice. (A) Nissl-stained coronal sections across the rostral neocortex of an animal from each experimental group (A1), total thickness of layers II–VI (A2) and cell number (cells per mm3) in layer II (A3), layer III (A4) and layers IV–VI (A5) in untreated euploid (n = 9) and Ts65Dn (n = 9) mice and euploid (n = 7) and Ts65Dn (n = 6) mice treated with fluoxetine. The ordinal numbers in A1 indicate the cortical layers. (B) Nissl-stained coronal sections across the hippocampal formation of an animal of each experimental group (B1). Note the reduced size of the hippocampal formation in untreated Ts65Dn mice. Scale bar = 500 µm. (B2–B4) Volume of the granule cell layer, density of granule cells (number per mm3) and total number of granule cells of the dentate gyrus (B2), volume of the pyramidal layer, density of pyramidal neurons (number per mm3) and total number of pyramidal cells of field CA3 (B3) and field CA1 (B4) in untreated euploid (n = 9) and Ts65Dn (n = 9) mice and euploid (n = 7) and Ts65Dn (n = 6) mice treated with fluoxetine. Values in B2–B4 refer to one hemisphere. Values in A2–A5 and B2–B4 represent mean ± SE. (*)P < 0.06; *P < 0.05; **P < 0.01; ***P < 0.001, Duncan’s test after ANOVA.

Hippocampal region

In untreated Ts65Dn mice the granule cell layer of the dentate gyrus had a smaller volume, a reduced granule cell density and a reduced number of granule cells in comparison with euploid mice [Fig. 4B(2)]. Similarly to the dentate gyrus, the pyramidal layer of fields CA3 [Fig. 4B(3)] and CA1 [Fig. 4B(4)] of Ts65Dn mice had a smaller volume, a reduced pyramidal cell density and a reduced number of pyramidal neurons in comparison with untreated euploid mice. After prenatal treatment with fluoxetine the volume of the cellular layers, neuronal density and total neuron number in the dentate gyrus, CA3 and CA1 became similar to those of untreated euploid mice [Fig. 4B(2–4)].

Striatum, thalamus, hypothalamus and mesencephalon

Untreated Ts65Dn mice had fewer cells in the striatum (Fig. 5A), thalamus (Fig. 5B), hypothalamus (Fig. 5C) and mesencephalon (Fig. 5D). In all these regions, treatment with fluoxetine completely restored cell number that became similar to that of untreated euploid mice (Fig. 5A–D).

Figure 5

Effect of embryonic treatment with fluoxetine on cellularity in different brain regions in post-natal Day 2 Ts65Dn and euploid mice. (A–D) Cell number in the striatum (A), thalamus (B), hypothalamus (C) and mesencephalon (D) in untreated euploid (n = 9) and Ts65Dn (n = 9) mice and euploid (n = 7) and Ts65Dn (n = 6) mice treated with fluoxetine. Cell number was evaluated in Nissl-stained sections. (E) Sagittal sections processed for calbindin immunohistochemistry, in order to label the Purkinje cells, and counterstained with haematoxylin across the cerebellum of an animal from each experimental group (E1), thickness of the cerebellar layers (E2) and cell number in each layer (E3) in untreated euploid (n = 7) and Ts65Dn (n = 7) mice and euploid (n = 5) and Ts65Dn (n = 5) mice treated with fluoxetine. The stars in (E1) indicate the depth of the cerebellar fissures. Calibration = 50 µm. Cell number in A–D and E3 is expressed as cells per mm3. Values in A–D, E2 and E3 represent mean ± SE. EGL = external granular layer; HYP = hypothalamus; IGL = internal granular layer; L = lobe; MES = mesencephalon; PURK = Purkinje cell layer; STR = striatum; TH = thalamus.

Cerebellum

The internal granular layer is the final destination of post-mitotic granule cell precursors migrated from the external granular layer. In untreated Ts65Dn mice the thickness of the external and internal granular layers [Fig. 5E(2)] and cell density in the external granular, Purkinje cell and internal granular layers [Fig. 5E(3)] were reduced in comparison with euploid mice. After prenatal treatment with fluoxetine the thickness of the external and internal granular layers [Fig. 5E(2)] and cell density in the external granular layer and internal granular layer [Fig. 5E(3)] became similar to those of euploid mice [Fig. 5E(2)]. No effect was found on Purkinje cell density [Fig. 5E(3)]. This is consistent with the timing of treatment that began at the end of the period of Purkinje cell precursor proliferation. Development of cerebellar folds requires massive proliferation and migration of granule cell precursors (Sillitoe and Joyner, 2007; Sudarov and Joyner, 2007). In agreement with the reduced proliferation rate of their granule cell precursors, Ts65Dn mice had cerebellar lobes with shallower fissures [Fig. 5E(1)]. This aberrant morphology was fully rescued by treatment [Fig. 5E(1)].

Effect of embryonic treatment with fluoxetine on proliferation rate and cellularity in euploid mice

In all examined regions, with the exception of the external granular layer of the cerebellum, treated euploid mice underwent an increase in the number of proliferating cells in comparison with untreated euploid mice (Figs 1–3), but this increase was proportionally less prominent than in treated Ts65Dn mice. Consistently with this moderate proliferation increase, in treated euploid mice there was a moderate or non-existent increase in the number of neurons forming the structures described in the preceding section (Figs 4 and 5).

Effects of embryonic treatment with fluoxetine on the serotonin receptor 5-HT1A in Ts65Dn and euploid mice

In view of the key role of the 5-HT1A-R in normal neurogenesis (Santarelli et al., 2003; Banasr et al., 2004; Encinas et al., 2006) and synaptogenesis (Mogha et al., 2012) we examined its expression in the dentate gyrus and subventricular zone of post-natal Day 2 mice. We found that Ts65Dn mice had a reduced expression of the 5-HT1A-R in the dentate gyrus, which is in line with previous evidence in post-natal Day 2 and 15 Ts65Dn mice (Bianchi et al., 2010; Guidi et al., 2013), as well as in the subventricular zone. Importantly, embryonic treatment with fluoxetine restored the expression of the 5-HT1A-R in both these regions of Ts65Dn mice (Supplementary Fig. 3), suggesting that normalization of the 5-HT1A-R may underlie restoration of neural precursor proliferation and dendrite/synapse development (see below). In euploid mice, treatment caused a marginal increase in 5-HT1A-R expression (Supplementary Fig. 3).

Effect of embryonic treatment with fluoxetine on apoptotic cell death

Evaluation of the number of apoptotic cells in the subventricular zone, subgranular zone and external granular layer (Supplementary material), the regions of more intense neurogenesis, showed no differences between post-natal Day 2 euploid and Ts65Dn mice and no effect of treatment on apoptotic cell death (Supplementary Table 3). This suggests that the increase in the number of proliferating cells and the rescue of brain cellularity in treated Ts65Dn mice were not because of a reduction in cell death.

Long-term effects of embryonic treatment with fluoxetine on neural precursor proliferation

The subventricular zone and the subgranular zone are the two major neurogenic niches of the adult brain. We used immunohistochemistry for Ki-67 (Supplementary material), an endogenous marker of proliferating cells (Scholzen and Gerdes, 2000), to estimate the size of the population of actively dividing cells in the subventricular zone and subgranular zone, 45 days after the end of treatment with fluoxetine. Untreated post-natal Day 45 Ts65Dn mice had fewer Ki-67-positive cells both in the subventricular zone and subgranular zone (Supplementary Fig. 4A and B). After treatment, the number of Ki-67-positive cells underwent an increase and became similar to that of untreated euploid mice (Supplementary Fig. 4A and B). This indicates that the restoration of the number of proliferating cells observed in the subventricular zone and subgranular zone of Ts65Dn mice aged 2 days (Figs 1 and 2) is a long-lasting effect that is retained well after treatment cessation. Treated euploid mice also had more Ki-67-positive cells than their untreated counterparts, though this effect was less prominent (Supplementary Fig. 4A and B).

Recent evidence shows that antidepressants, including fluoxetine, stimulate proliferation by reducing the levels of p21 (cip1/WAF1) (Pechnick et al., 2011), a cyclin-dependent kinase inhibitor that inhibits cell cycle progression. Embryos and infants with Down’s syndrome exhibit higher brain levels of p21 (Engidawork et al., 2001; Park et al., 2010), suggesting that excessive p21 levels may contribute to neurogenesis impairment. We found that in the hippocampus of Ts65Dn mice aged 45 days there were higher levels of p21 in comparison with euploid mice. After treatment with fluoxetine, p21 levels underwent a large reduction both in Ts65Dn and euploid mice (Supplementary Fig. 4C and D), suggesting that this reduction may take part in the treatment-induced proliferation increase. The mechanisms whereby antidepressants reduce p21 levels remain to be elucidated. The observation that p21 levels were reduced in 45-day-old mice, i.e. well after treatment cessation, can explain the long-lasting effect of treatment on precursor proliferation.

Long-term effects of embryonic treatment with fluoxetine on cell survival and phenotype

The dentate gyrus is an ideal structure for establishing the long-term effects of treatment on cell fate since the final destination of cells born in the subgranular zone is the overlying granule cell layer. Thus, in animals injected with BrdU at post-natal Day 2 and sacrificed at 45 days of age, we evaluated the density, number and phenotype of BrdU-positive cells present in the dentate gyrus. In untreated Ts65Dn mice the density [Fig. 6A(1)] and total number [Fig. 6A(4)] of surviving cells and of cells migrated to the granule cell layer [Fig. 6A(2 and 5)] were reduced in comparison with euploid mice, consistently with their reduced proliferation rate (Fig. 2). After treatment with fluoxetine, the density and number of surviving cells became similar to that of untreated euploid mice [Fig. 6A(1 and 4)], indicating that the surplus of cells born at post-natal Day 2 in treated Ts65Dn mice was not offset by a reduction in their survival rate. An analysis of the phenotype of the surviving cells showed that untreated Ts65Dn mice had fewer new neurons [NeuN/BrdU cells; Fig. 6B(1 and 4)] and fewer new astrocytes [GFAP/BrdU cells; Fig. 6B(2 and 5)] than euploid mice, with no difference in the number of cells with an undetermined phenotype [neither/BrdU; Fig. 6B(3 and 6)]. Treatment increased the density and number of new neurons that became similar to that of untreated euploid mice [Fig. 6B(1 and 4)], without affecting the density and number of new astrocytes [Fig. 6B(2 and 5)] and cells with an undetermined phenotype [Fig. 6B(3 and 6)]. In euploid mice treatment increased the density and number of new neurons [Fig. 6B(1 and 4)] and cells with an undetermined phenotype [Fig. 6B(3 and 6)], with no effect on the number of new astrocytes [Fig. 6B(2 and 5)]. These findings suggest that treatment favours neurogenesis without affecting astrogliogenesis, thereby correcting the higher astrocyte/neuron ratio that characterizes the trisomic condition (Guidi et al., 2008; Lu et al., 2012).

Figure 6

Long-term effect of embryonic treatment with fluoxetine on the dentate gyrus of post-natal Day 45 Ts65Dn and euploid mice. (A) BrdU-positive cells (number per mm3) (A1–A3) and total number of BrdU-positive cells (A4–A6) in the granule cell layer plus subgranular zone/hilus (A1 and A4; Gr + H), in the granule cell layer (A2 and A5) and in the subgranular zone/hilus (A3 and A6) of untreated euploid (n = 6) and Ts65Dn (n = 5) mice and euploid (n = 6) and Ts65Dn (n = 5) mice treated with fluoxetine. These animals received one BrdU injection on post-natal Day 2 and were sacrificed on post-natal Day 45. (B) BrdU-positive cells (number per mm3) (B1–B3) and total number of BrdU-positive cells (B4–B6) with a neuronal phenotype (B1 and B4; NeuN/BrdU), an astrocytic phenotype (B2 and B5; GFAP/BrdU) and an undetermined phenotype (B3 and B6; Neither/BrdU) in the dentate gyrus (Gr + H) of untreated euploid and Ts65Dn mice and euploid and Ts65Dn mice treated with fluoxetine. Same animals as in A. (C) Volume of the granule cell layer (C1), density (number per mm3) of granule cells (C2) and total number of granule cells (C3) in post-natal Day 45 untreated euploid (n = 6) and Ts65Dn (n = 5) mice and euploid (n = 6) and Ts65Dn (n = 5) mice treated with fluoxetine. Stereology was carried out in Nissl-stained sections. Values in A4–A6 and B4–B6 refer to one hemisphere. Values in A–C represent mean ± SE. *P < 0.05; **P < 0.01; ***P < 0.001, Duncan’s test after ANOVA. H = subgranular zone/hilus.

Long-term effects of embryonic treatment with fluoxetine on post-natal granule cell production

As the granule cells of the dentate gyrus are mainly born post-natally, we exploited this feature for establishing whether a prenatal treatment had long-term effects on neurogenesis. In untreated 45-day-old Ts65Dn mice the granule cell layer had a smaller volume, a reduced granule cell density and a reduced granule cell number in comparison with untreated euploid mice (Fig. 6C). In treated Ts65Dn mice the volume of the granule cell layer [Fig. 6C(1)], granule cell density [Fig. 6C(2)] and total granule cell number [Fig. 6C(3)] became similar to those of untreated euploid mice, indicating that the rescue of neurogenesis in the dentate gyrus of post-natal Day 2 Ts65Dn mice (Fig. 2) outlasted treatment cessation. In euploid mice, treatment with fluoxetine had no effect on the stereology of the dentate gyrus (Fig. 6C).

Long-term effects of embryonic treatment with fluoxetine on cortical cellularity

Evaluation of total cell number in the neocortex of post-natal Day 45 mice (Supplementary material) showed that treated Ts65Dn mice still had more cells than their untreated counterparts (Supplementary Fig. 5A), indicating that the cellularity increase observed in post-natal Day 2 Ts65Dn mice (Fig. 5A) was retained over time. Evaluation of the number of astrocytes (Supplementary material) showed that untreated Ts65Dn mice had a similar number of astrocytes as euploid mice, but a higher ratio of astrocytes over total cell number (Supplementary Fig. 5B and C). In treated Ts65Dn mice there was a reduction in the ratio of astrocytes over total cell number (Supplementary Fig. 5B and C), suggesting that treatment specifically favours cortical neurogenesis.

Long-term effects of embryonic treatment with fluoxetine on dendritogenesis in Ts65Dn and euploid mice

We exploited the mainly post-natal generation of granule neurons in the dentate gyrus to establish whether prenatal treatment with fluoxetine has a positive impact on the dendritic development of neurons born after treatment cessation. Granule cell dendritic morphology was analysed in post-natal Day 45 mice with immunohistochemistry for DCX, a protein present in the cytoplasm of granule neurons during the period of neurite elongation (from 1–4 weeks after neuron birth; Couillard-Despres et al., 2005). In view of the time-course of DCX expression, DCX-positive cells present in post-natal Day 45 animals are cells that were born after post-natal Day 16. In untreated Ts65Dn mice, the dendritic tree had a reduced length and fewer branches whereas in treated Ts65Dn mice there was an increase in both parameters that became even larger than those of untreated euploid mice (Fig. 7B). The dendrograms in Fig. 7C show that in Ts65Dn mice dendritic hypotrophy was because of the lack of branches of order higher than six and that after treatment this aberrant phenotype was completely corrected. This evidence indicates that in Ts65Dn mice prenatal treatment with fluoxetine has a long-term positive impact on dendritic development. In euploid mice treatment also had a positive impact on dendritic development, though this effect was comparatively less prominent (Fig. 7B and C).

Figure 7

Long-term effect of embryonic treatment with fluoxetine on the dendritic architecture of newborn granule cells in post-natal Day 45 Ts65Dn and euploid mice. (A) Examples of the dendritic tree of granule cells immunostained for DCX of an animal from each experimental group. Numbers indicate the different dendritic orders. Scale bar = 20 µm. (B) Mean total dendritic length and mean number of dendritic segments of newborn granule cells immunostained for DCX in untreated euploid (n = 6) and Ts65Dn (n = 5) mice and euploid (n = 6) and Ts65Dn (n = 5) mice treated with fluoxetine. Values represent mean ± SE. *P < 0.05; **P < 0.01; ***P < 0.001, Duncan’s test after ANOVA. (C) Dendrograms of the granule cells immunostained for DCX reconstructed for each experimental group. Dendrograms were obtained by averaging the mean length and mean number of branches of each order. The number of branches was approximated to the nearest integer value (thick lines). Thin lines have been used to indicate a number of branches ranging from 0.1 to 0.5. Scale bar = 10 µm.

Long-term effects of embryonic treatment with fluoxetine on synapse development in Ts65Dn and euploid mice

To establish whether treatment had a long-term effect on synapse development, we examined the expression of SYN, a synaptic vesicle glycoprotein that is a marker of presynaptic terminals and of PSD95, a marker of the postsynaptic regions, in post-natal Day 45 mice. Evaluation by western blot analysis of SYN and PSD95 protein levels in the hippocampal formation showed that Ts65Dn mice had reduced SYN and PSD95 levels and that treatment restored both SYN and PSD95 levels (Supplementary Fig. 6A–D). Treatment increased SYN and PSD95 levels in euploid mice too (Supplementary Fig. 6A–D). Evaluation of SYN (Fig. 8A and B) and PSD95 (Fig. 8E and F) immunoreactivity showed a reduction in the neocortex, hippocampus and dentate gyrus of Ts65Dn mice in comparison with euploid mice. In treated Ts65Dn mice, the immunoreactivity of SYN (Fig. 8B) and PSD95 (Fig. 8F) underwent an increase and became similar to that of untreated euploid mice. An increase in SYN and PSD95 immunoreactivity also took place in treated euploid mice (Fig. 8B and F). We examined the number of individual SYN or PSD95 puncta to establish whether the effects of genotype and treatment on SYN and PSD95 immunoreactivity were because of changes in synaptic protein levels in individual synapses or to a change in the number of synapses. We found that SYN (Fig. 8D) and PSD95 (Fig. 8H) puncta were reduced in untreated Ts65Dn mice in all examined regions and that after treatment the number of puncta became similar to that of untreated euploid mice. This evidence indicates that Ts65Dn mice had fewer synaptic contacts and that treatment restored the presynaptic terminals and their postsynaptic targets.

Figure 8

Long-term effect of embryonic treatment with fluoxetine on synapse development in post-natal Day 45 Ts65Dn and euploid mice. (A and E) Sections processed for SYN (A) and PSD95 (E) immunofluorescence from the neocortex, hippocampal field CA1 and dentate gyrus of an animal from each experimental group. Scale bar = 200 µm. (B and F) Optical density of SYN (B) and PSD95 (F) immunoreactivity in layers II, III and IV–VI of the neocortex, stratum radiatum of field CA1 and middle and outer third of the molecular layer of the dentate gyrus of untreated euploid (n = 6) and Ts65Dn (n = 5) mice and euploid (n = 6) and Ts65Dn (n = 5) mice treated with fluoxetine. For each region, data of SYN and PSD95 immunoreactivity are given as fold difference versus untreated euploid mice. (C and G) Images, taken with the confocal microscope, of sections processed for SYN (C) and PSD95 (G) immunofluorescence from the neocortex from an animal of each experimental group. Scale bar = 5 µm. (D and H) Number of puncta per µm2 exhibiting SYN (D) and PSD95 (H) immunoreactivity in layers II, III and IV–VI of the neocortex, stratum radiatum of field CA1 and middle and outer third of the molecular layer of the dentate gyrus of untreated euploid (n = 6), untreated Ts65Dn (n = 5), and euploid (n = 6) and Ts65Dn (n = 5) mice treated with fluoxetine. Values in B, F, D and H represent mean ± SE *P < 0.05; **P < 0.01; ***P < 0.001, Duncan’s test after ANOVA. DG = dentate gyrus; GR = granule cell layer; LM = stratum lacunosum-moleculare; MOL = molecular layer; OR = stratum oriens; PYR = pyramidal layer; RAD = stratum radiatun.

As activation of the 5-HT1A-R or administration of fluoxetine increases PSD95 expression and synapse formation through activation (phosphorylation) of Erk1/2 (Mogha et al., 2012), we examined the effects of treatment on phosphorylated-Erk1/2 level in hippocampal lysates. Consistent with previous evidence (Siarey et al., 2006; Altafaj et al., 2013) we found reduced levels of phosphorylated-Erk1/2 in untreated trisomic mice. After treatment with fluoxetine, the phosphorylated-Erk1/2 levels of Ts65Dn mice underwent an increase, becoming similar to those of euploid mice (Supplementary Fig. 6E and G). An increase also took place in treated euploid mice. No difference among groups was found in Erk1/2 levels (Supplementary Fig. 6F and G). This evidence suggests that the positive impact of treatment on synaptogenesis may be mediated by an increase in phosphorylated-Erk1/2 levels. The treatment-induced restoration of the 5-HT1A-R in Ts65Dn mice (see above) is very likely a key determinant for normalization of phosphorylated-Erk1/2 levels.

Long-term effects of embryonic treatment with fluoxetine on fibre tracts in Ts65Dn and euploid mice

We used diffusion MRI and fractional anisotropy (Supplementary material) to examine white matter integrity in post-natal Day 45 mice. Of the examined fibre tracts only the middle cerebellar peduncle had a reduced fractional anisotropy in Ts65Dn mice, a difference that disappeared after treatment (Supplementary Fig. 7A). In addition, in treated Ts65Dn mice there was an increase in the fractional anisotropy of the rostrum of the corpus callosum, anterior commissure and hippocampal commissure (Supplementary Fig. 7A). We additionally measured connectivity by propagation of streamlines along fibre tracts in diffusion-weighted images (tractography, see Supplementary material). We found a reduced connectivity of the hippocampal commissure and middle cerebellar peduncle of Ts65Dn mice and that these defects were rescued by treatment (Supplementary Fig. 7B).

Long-term behavioural effects of embryonic treatment with fluoxetine in Ts65Dn and euploid mice

Consistent with previous evidence (Kleschevnikov et al., 2012), trisomic mice were characterized by spontaneous locomotor hyperactivity (Fig. 9A–D). In treated Ts65Dn mice, locomotor activity was normalized and became similar to that of untreated euploid mice (Fig. 9A–D). No effects of treatment were found on spontaneous motor behaviour of euploid mice (Fig. 9A–D). To investigate the effects of treatment on hippocampus-dependent learning and memory we used the novel object recognition and contextual fear conditioning paradigms. We found no effect of genotype on short-term (4 h) and long-term (24 h) object recognition memory (data not shown). Normal short-term memory in the novel object recognition task has been documented in Ts65Dn mice aged 2–3 months, though long-term memory appears to be impaired after a retention period of 24 h (Kleschevnikov et al., 2012). The discrepancy of our results may be because of the fact that the novel object recognition test may not reveal trisomy-linked long-term memory damage at such a young age as that of the animals used here. The contextual fear conditioning paradigm is a test that allows for estimation of both hippocampus-independent (cued) and hippocampus-dependent (contextual) memory (McHugh et al., 2007). We found that cued memory was normal in Ts65Dn mice (Fig. 9F). In contrast, untreated Ts56Dn mice showed a significantly lower freezing behaviour in the old context compared to euploid mice (Fig. 9E), indicating a poorer memory for the context. Treatment with fluoxetine completely restored freezing, suggesting that treatment had restored long-term memory (Fig. 9E).

Figure 9

Long-term effects of embryonic treatment with fluoxetine on behaviour in post-natal Day 45 Ts65Dn and euploid mice. (A–D) Open field test. Distance traveled in the open field (A), average speed (B), number (C) and duration (D) of vertical movements (rearing) of untreated euploid (n = 6) and Ts65Dn (n = 5) mice and euploid (n = 5) and Ts65Dn (n = 5) mice treated with fluoxetine (Fluo). (E and F) Contextual fear conditioning test. Percentage freezing in the conditioning context (old context) (E) and after the conditioning sound (cued cage) in the new environment of the test session (F) in untreated euploid (n = 16) and Ts65Dn (n = 10) mice and euploid (n = 11) and Ts65Dn (n = 10) mice treated with fluoxetine. Values are mean ± SE. (*)P < 0.06; *P < 0.05; **P < 0.01; ***P < 0.001, Duncan’s test after ANOVA).

Discussion

Our study in a mouse model of Down’s syndrome shows neurogenesis impairment across the whole trisomic brain and provides novel evidence that a pharmacotherapy with fluoxetine during embryonic development is able to fully restore neurogenesis, overall brain cellularity, dendrite and synapse development, brain volume (Table 4) and behaviour.

View this table:
Table 4

Summary of the effects of embryonic treatment with fluoxetine on Ts65Dn and euploid mice

Post-natal Day 2 micePost-natal Day 45 mice
MeasureRegionTs + SalTs + FluoEu + FluoMeasureRegionTs + SalTs + FluoEu + Fluo
ProliferationSVZ<=>Cycling cells, nSVZ<=>
DG<>>DG<==
NC<=>Surviving cells, nDG<=>
STR<=>New neurons, nDG<=>
TH<==New astrocytes, nDG<==
HYP<>>CellularityNC<>>
MES<=>DG<==
CB<==Dendrites (length and number)DG<>>
PO<=>Fibre tract sizeCC===
CellularityNC<=>AC=>>
DG<==HC<==
CA1<==MCP<==
CA3<==SYN punctaNC<==
STR<==CA1<==
TH<==DG<==
HYP<==PSD95 punctaNC<>>
MES<==CA1<>>
CB<==DG<>>
SYN levelsHF<=>
PSD95 levelsHF<=>
Cell deathSVZ===p21 (cip1/WAF1) levelsHF><<
DG===p-Erk1/2 levelsHF<=>
CB===Brain volumeNC<==
5-HT1A-R expressionDG<==HF<==
SVZ<==CB<==
Brain weight<==Brain weight<==
Body weight<<=Body weight<<<
BMD<<=
  • Effect of genotype and embryonic treatment with fluoxetine on different aspects of brain and body development (Measure column) in mice aged 2 days (post-natal Day 2) and 45 days (post-natal Day 45). These mice were the progeny of Ts65Dn females that received either saline (Sal) or fluoxetine (Fluo) from embryonic Day 10 to delivery. The regions where the different measurements were carried out are indicated (Region column). The effect of genotype and treatment refers to untreated euploid mice. The symbols < and > indicate a lower and higher value, respectively, in comparison with untreated euploid mice. The symbol = indicates a value similar to that of untreated euploid mice. AC = anterior commissure; BMD = bone mineral density; CB = cerebellum; CC = corpus callosum; DG = dentate gyrus; HC = hippocampal commissure; HF = hippocampal formation; HYP = hypothalamus; MCP = medial cerebellar peduncle; MES = mesencephalon; NC = neocortex; PO = pons; STR = striatum; SVZ = subventricular zone; TH = thalamus.

The trisomic brain is characterized by widespread neurogenesis defects and hypocellularity

We show here that the trisomic brain is characterized by neurogenesis impairment in the telencephalon (the whole extent of the subventricular zone, neocortex, striatum, subgranular zone), diencephalon (thalamus and hypothalamus), mesencephalon and metencephalon (cerebellum and pons), which suggests that trisomic neural precursor cells that give origin to different brain regions share the same proliferation defect. Consistently with their widespread neurogenesis impairment, neonate Ts65Dn mice were characterized by hypocellularity throughout the brain (neocortex, striatum, dentate gyrus, hippocampus, diencephalon, mesencephalon and cerebellum). Mouse models of Down’s syndrome and individuals with Down’s syndrome show impairment in (i) sensory-motor and cognitive behaviour, that is, functions that require the participation of the neocortex, basal ganglia, thalamus, mesencephalon and cerebellum; (ii) long-term memory, a function that requires the participation of the hippocampal region; and (iii) regulation of the wake–sleep cycle and metabolism, functions that require the participation of the hypothalamus. The widespread hypocellularity observed in the neonate trisomic brain could account for this constellation of neurological disturbances.

Embryonic treatment with fluoxetine restores neurogenesis and brain cellularity in Ts65Dn mice

As the trisomic brain is characterized by neurogenesis defects starting from the earliest phases of development, we wondered whether an embryonic pharmacotherapy might be a tool to drastically correct this defect. As we had hypothesized, we found that fluoxetine restored neural precursor proliferation across the whole brain. This effect led to restoration of cellularity in all examined brain parts and, consequently, at birth treated trisomic mice had a similar number of cells as their euploid counterparts. Consistent with the generalized cellularity increase induced by fluoxetine in Ts65Dn mice, their brain weight and the volume of various brain regions became similar to those of untreated euploid mice at birth.

Concerning the duration of the effects of the embryonic treatment with fluoxetine, it is important to note that in treated trisomic mice aged 45 days, the number of cycling cells in the subventricular zone and subgranular zone and the number of granule cells added to the dentate gyrus were still higher in comparison with their untreated counterparts indicating that the short-term effects of fluoxetine on cell proliferation observed at post-natal Day 2 were followed by a positive and enduring effect that lasted into adulthood. The reduced proliferation potency that characterizes the trisomic brain is worsened by a reduction in the number of cells differentiating into neurons and an increase in the number of cells differentiating into astrocytes (Contestabile et al., 2007, 2009; Guidi et al., 2008). In Ts65Dn mice prenatally treated with fluoxetine the differentiation programme of cells born at post-natal Day 2 and surviving until post-natal Day 45 was corrected, indicating a long-lasting effect of treatment on cell fate.

Embryonic treatment with fluoxetine restores dendrite and synapse development in Ts65Dn mice

Dendritic hypotrophy and reduction in the number of synaptic contacts are typical features of the trisomic condition (Bartesaghi et al., 2011). In Ts65Dn mice embryonically treated with fluoxetine the reduced number of pre- and postsynaptic terminals in the neocortex, hippocampus and dentate gyrus became similar to those of euploid mice. The increase in the number of presynaptic terminals is consistent with the increase in the number of neurons and, hence, of axons. The increase in the number of postsynaptic terminals implies a parallel increase in the dendritic surface. Accordingly, we found that the severe dendritic hypotrophy of newborn granule neurons was fully corrected by treatment. In addition to the intracortical restoration of synapse development in treated Ts65Dn mice, we found an increase in the size of major fibre tracts. Again, this effect is consistent with the fluoxetine-induced increase in brain cellularity.

Embryonic treatment with fluoxetine restores behaviour in Ts65Dn mice

The poor contextual memory of Ts65Dn mice was rescued by treatment, suggesting treatment-induced restoration of hippocampus-dependent long-term memory. This is in agreement with the rescue of hippocampal cellularity, dendrite and synapse development in treated Ts65Dn mice. As cortical signals are fundamental for hippocampus-dependent cognitive functions, the long-term restoration of cortical cellularity and synapse development in treated Ts65Dn mice is an additional important effect for the rescue of hippocampal functions. We found that treatment also normalized the spontaneous hyperactivity that characterizes trisomic mice. The striatum appears to play a key role in the generation of stereotypies and hyperactivity in rodents (Adriani et al., 2012) and reduced striatal volume has been found in children with the attention deficit hyperactivity disorder (Carmona et al., 2009). We found that treatment rescued striatal neurogenesis and cellularity in Ts65Dn mice, suggesting that this effect may contribute to the restoration of spontaneous motor behaviour.

Embryonic treatment with fluoxetine has moderate effects on brain development in euploid mice

In euploid mice treatment had proportionally more moderate effects than in Ts65Dn mice on cell proliferation, cellularity, dendrite and synapse development and had no effect on memory performance. The finding that prenatal treatment with fluoxetine has relatively scarce advantages in normal animals is in line with similar evidence in mice post-natally treated with fluoxetine (Bianchi et al., 2010; Guidi et al., 2013) and suggests that fluoxetine may help brain development under abnormal but not normal brain conditions.

Early pharmacotherapy with fluoxetine: a tool for correcting brain development in Down’s syndrome?

Previous findings showed the rescue of hippocampal development in Ts65Dn mice by an early 13-day long post-natal treatment with fluoxetine (Bianchi et al., 2010; Guidi et al., 2013) and current findings show that it is possible to rescue the whole brain development with an embryonic pharmacotherapy with fluoxetine. A study in adult Ts65Dn mice shows that a prolonged treatment with fluoxetine (6 weeks) has adverse effects on brain function (Heinen et al., 2012). However, the effects of chronic treatment with fluoxetine appear to be age-dependent and treatment induces adverse effects in adult but not adolescent mice (Bouet et al., 2012). This suggests that the timing and duration of a pharmacotherapy with fluoxetine may be crucial and that fluoxetine may be a therapy of choice during the earliest life stages.

No effective therapies are available at present to rescue neurogenesis and intellectual disability in individuals with Down’s syndrome. The current study shows that a pharmacotherapy with fluoxetine during the embryonic period can rescue cell proliferation and cellularity throughout the brain in the Ts65Dn mouse model. These effects are accompanied by the rescue of dendritic and synaptic development that outlasts treatment cessation. The ultimate effect of treatment is the rescue of behaviour in adulthood. As we used Ts65Dn (and euploid) mice that were the progeny of treated Ts65Dn females (i.e. trisomic females) it cannot be ruled out that treatment with fluoxetine may have exerted a beneficial effect on the mothers too. This may affect development of the pups as a result of, for instance, more normal nurturing. However, the finding that restoration of neurogenesis and dendritogenesis took place in Ts65Dn pups directly treated with fluoxetine (Bianchi et al., 2010; Guidi et al., 2013) suggests that a possible positive impact of treatment on the dams may not be a crucial factor underlying restoration of brain development in embryonically-treated mice.

It is important to observe that in treated Ts65Dn mice neurogenesis, dendritogenesis and synapses were still fully normalized well after treatment cessation. Concerning the mechanisms whereby an embryonic therapy with fluoxetine can lead to these long-lasting effects, the treatment-induced restoration of the defective 5-HT1A-R expression is most likely a key determinant. Based on current results we can envisage the following chain of events: (i) the fluoxetine-induced increase in serotonin bioavailability during embryonic life stages favours neurogenesis and neurite elongation; (ii) consequently, by birth, the brain has a normal number of neurons with normal dendritic processes; (iii) through a mechanism that remains to be clarified, treatment also restores 5-HT1A-R expression; (iv) in view of events i–iii, a correct number of serotonergic terminals can be established on neurons appropriately endowed with 5-HT1A receptors; and (v) because of restoration of the serotonergic system, the processes of neurogenesis, neurite elongation and synapse formation remain normalized, with no further need of treatment after birth.

Fluoxetine is a widely-used antidepressant also prescribed in children (Boylan et al., 2007), suggesting that a pharmacotherapy for Down’s syndrome using fluoxetine may be practicable during gestation. Though it seems that fluoxetine administered during pregnancy has no serious adverse effects on the somatic development of the neonate (Einarson et al., 2009; Pedersen et al., 2010; Hayes et al., 2012; Olivier et al., 2013), some studies suggest the potential risk of pulmonary hypertension (Chambers et al., 2006). Yet, the latter risk is still controversial (Olivier et al., 2013). It must also be observed that in utero exposure to serotonin re-uptake inhibitors may result in a neonatal withdrawal syndrome (Moses-Kolko et al., 2005; Sanz et al., 2005). Though the neonatal behavioural syndrome is generally self-limited and can be managed with supportive care, the withdrawal effects must be taken into account. However, considering the impressive effects of fluoxetine on trisomic brain development, the side effects of prenatal exposure to fluoxetine may be considered as a minor problem in the face of the possible rescue of cognitive disability.

The Ts65Dn mouse is considered a good model of Down’s syndrome because it exhibits several abnormal phenotypes that are similar to those seen in human trisomy 21. It must be observed, however, that this and an increasing variety of new mouse models pose unavoidable limitations because they are trisomic for different sets of genes orthologous to those of Hsa21 (Liu et al., 2011). Caution, therefore, must be exercised in the translation of results from animal models to the human disorder because the outcome may be different. Yet, considering that the serotonergic system is altered both in individuals with Down’s syndrome and in the Ts65Dn mouse, it may not be unreasonable to hypothesize that fluoxetine may be an effective therapy for the rescue of brain development in individuals with Down’s syndrome.

Funding

This work was supported by a grant to R. B. and a grant to R. M. from the ‘Fondation Jérôme Lejeune, France’.

Acknowledgements

Elliott Dolan-Evans, Nargis Reza and Michael Nelson provided assistance with MRI analysis.

Footnotes

  • *These authors contributed equally to this work.

Abbreviations
BrdU
5-bromo-2-deoxyuridine
Erk 1/2
extracellular signal-regulated kinase 1/2
PSD95
postsynaptic density protein 95
SYN
synaptophysin
5-HT
5-hydroxytryptamine
5-HT1A-R
5-HT1A serotonin receptor

References

View Abstract