The functional significance of adult neural stem and progenitor cells in hippocampal-dependent learning and memory has been well documented. Although adult neural stem and progenitor cells in the subventricular zone are known to migrate to, maintain and reorganize the olfactory bulb, it is less clear whether they are functionally required for other processes. Using a conditional transgenic mouse model, selective ablation of adult neural stem and progenitor cells in the subventricular zone induced a dramatic increase in morbidity and mortality of central nervous system disorders characterized by excitotoxicity-induced cell death accompanied by reactive inflammation, such as 4-aminopyridine-induced epilepsy and ischaemic stroke. To test the role of subventricular zone adult neural stem and progenitor cells in protecting central nervous system tissue from glutamatergic excitotoxicity, neurophysiological recordings of spontaneous excitatory postsynaptic currents from single medium spiny striatal neurons were measured on acute brain slices. Indeed, lipopolysaccharide-stimulated, but not unstimulated, subventricular zone adult neural stem and progenitor cells reverted the increased frequency and duration of spontaneous excitatory postsynaptic currents by secreting the endocannabinod arachidonoyl ethanolamide, a molecule that regulates glutamatergic tone through type 1 cannabinoid receptor (CB1) binding. In vivo restoration of cannabinoid levels, either by administration of the type 1 cannabinoid receptor agonist HU210 or the inhibitor of the principal catabolic enzyme fatty acid amide hydrolase, URB597, completely reverted the increased morbidity and mortality of adult neural stem and progenitor cell-ablated mice suffering from epilepsy and ischaemic stroke. Our results provide the first evidence that adult neural stem and progenitor cells located within the subventricular zone exert an ‘innate’ homeostatic regulatory role by protecting striatal neurons from glutamate-mediated excitotoxicity.
In the adult rodent brain, neurogenesis continues lifelong mainly in two regions: the subventricular zone of the lateral ventricles and the subgranular zone of the hippocampal dentate gyrus (Alvarez-Buylla and Lim, 2004; Zhao et al., 2008). Newly formed adult neural stem/progenitor cells (NPCs) of the subgranular zone migrate short distances and differentiate into dentate granule cells within the homonymous layer. These newly formed dentate granule cells are necessary for the modulation and refinement of the existing neuronal circuits involved in hippocampus-dependent memory processing and behaviour (Imayoshi et al., 2008; Kitamura et al., 2009; Singer et al., 2011). On the other hand, adult NPCs in the rodent subventricular zone migrate along the rostral migratory stream to the olfactory bulb, where they differentiate into gamma-aminobutyric acidergic (GABAergic) interneurons and integrate within the granule and glomerular cell layers in order to maintain and reorganize the olfactory bulb system (Imayoshi et al., 2008).
Importantly, neurogenesis is also implicated in central nervous system (CNS) disorders. A long-lasting neurogenic response to injury has been observed in both patients (Marti-Fabregas et al., 2010) and animal models (Bengzon et al., 1997; Parent et al. 1997; Jin et al., 2001, 2010; Goings et al., 2004) of several neurological diseases (Martino et al., 2011). Nonetheless, replacement of damaged cells seems not to be the prevailing aim of subventricular zone neurogenesis occurring in response to tissue damage, and its exact functional significance still needs to be fully elucidated. In fact, in stroke ∼80–90% of newly generated subventricular zone-derived neuroblasts die within a few weeks and do not integrate into the spared neuronal circuitry (Arvidsson et al., 2002). In corpus callosum demyelination, <4% of newly formed subventricular zone-derived cells differentiate into myelinating oligodendrocytes (Menn et al., 2006). In status epilepticus, subventricular zone-derived cells migrating toward the hippocampus terminally differentiate into glial, but not neuronal, cells (Parent et al., 2006).
Interestingly, from transplantation studies it has been demonstrated that, while remaining undifferentiated, donor subventricular zone-derived adult NPCs promote CNS tissue healing via the secretion of immunomodulatory and neuroprotective molecules, exerting a so-called ‘bystander effect’ that suppresses detrimental tissue responses (Ourednik et al., 2002; Pluchino et al., 2003; Martino and Pluchino, 2006; Bacigaluppi et al., 2009). A protective role of transplanted adult NPCs was further demonstrated by studies revealing that an immortalized adult NPC line protects spinal cord cultures from experimentally induced excitotoxic damage (Llado et al., 2004). Thus, while the replacement of damaged cells seems not to be the main goal of subventricular zone neurogenesis occurring in response to tissue damage, it is very likely that the specific features of the extracellular environment in the course of inflammation trigger adult NPCs to carry out other functions (Ekdahl et al., 2009) complementary to cell replacement. As a matter of fact, adult NPCs within the subventricular zone are located in a strategic position being in close contact with the ventricular system, directly participating with the blood–brain barrier (Rolls et al., 2007; Mirzadeh et al., 2008; Tavazoie et al., 2008) and being juxtaposed to the striatum, an area integrating and modulating various signals originating in the cortex (Koos and Tepper, 1999). It is therefore appropriate to consider the possibility that adult NPCs are capable of playing a surveillance function by integrating danger signals coming from the blood and/or from the CSF.
Here, we show, by in vitro and in vivo experiments, that subventricular zone adult NPCs might act as regulators of neuronal homeostasis in adjacent striatal circuits in pathological conditions characterized by glutamatergic excitotoxicity, such as experimental epilepsy and ischaemic stroke.
Materials and methods
Further detailed information is provided in the online Supplementary material.
Generation of transgenic mouse lines
We used a third generation self inactivating lentiviral vector (Follenzi et al., 2000) to generate a NestinfloxGFPfloxTK targeting lentivirus. All experiments were conducted on the 7457 transgenic mouse line. Selective ablation experiments were performed in the double transgenic mice obtained by crossing the NestfloxGFPfloxTK with the CMV-Cre mouse line (present in our laboratory) (Su et al., 2002) thus generating the hereafter named NestinTK mouse line. In this transgenic mouse, green fluorescent protein (GFP) and thymidine kinase 1 (TK) gene expression is controlled by the second intron enhancer of the nestin gene (Supplementary Fig. 1) (Zimmerman et al., 1994).
All efforts were made to minimize animal suffering and to reduce the number of mice used, in accordance with the European Communities Council Directive of 24 November 1986 (86/609/EEC). All procedures involving animals were executed according to the guidelines of the Institutional Animal Care and Use Committee of the San Raffaele Scientific Institute, Milan, Italy.
We used two different protocols of ganciclovir (GCV, Cytovene®, Roche) administration. In the first protocol, adult mice received intraperitoneally (i.p.) 100 mg/kg of GCV dissolved in distilled water each other day for 10 days. In the second protocol, mice received 100 mg/kg/day of GCV dissolved in distilled water administered by subcutaneously implanted osmotic minipumps (ALZET® Model 2002, DURECT Corporation) as previously described (Garcia et al., 2004). Implanted osmotic minipumps were replaced with new ones after 14 days for a total of 28 days of treatment. NestinTK mice treated with GCV are named NestinTK-GCV+, while PBS treated NestinTK mice are called NestinTK-GCV−. As control we used C57Bl/6 mice untreated or treated with GCV (named C57Bl/6-GCV− and C57Bl/6-GCV+, respectively). NestinTK-GCV−, C57Bl/6-GCV− and C57Bl/6-GCV+ were altogether named controls.
Cytosine β-d-arabinofuranoside hydrochloride (AraC, Sigma) 2% dissolved in PBS or PBS only was intracerebroventricularly administered via a brain catheter (Brain Infusion Kit 3, ALZET®) connected to a subcutaneously osmotic AraC containing minipump (ALZET®, model 1007) for 7 days prior to analyses or other treatments.
CB1 receptor agonist HU210 (1 mg/kg/day; Tocris Bioscience) dissolved in dimethyl sulphoxide (Sigma), saline and Tween 0.5% (Sigma) or 0.16 mg/kg/day of fatty acid amide hydrolase inhibitor URB597 (Tocris Bioscience) in dimethyl sulphoxide and saline was intrastriatally continuously administered [stereotaxic coordinates, 1 mm anterior, 1.8 mm lateral and 3.0 mm deep to Bregma (Paxinos and Franklin, 2000)] for 24 h, starting 12 h prior to other treatments (induction of epilepsy or cerebral ischaemia) using a catheter (Brain Infusion Kit 3) connected to a subcutaneous implanted osmotic mini pump (ALZET®, model 1007). Implantation was performed under general anaesthesia. The AMPA receptor antagonist NBQX (2,3-dihydroxy-6-nitro-7-sulphamoyl-benzo[f]quinoxaline-2,3-dione) di-sodium salt (Tocris Bioscience) dissolved in saline was injected at the dosage of 60 mg/kg intraperitoneally 20 min before the induction of epilepsy or stroke. A second injection at the same dosage was given 10 min after induction of middle cerebral artery occlusion (MCAO). Lipopolysaccharide (Sigma) at the dosage of 1 mg/kg dissolved in saline was intraperitoneally administered 30 min before sacrifice of mice and brain extraction. 4-Aminopyridine at the dosage of 8 mg/kg dissolved in saline was intraperitoneally injected in the mice 20 min before sacrifice.
Histology and pathological analysis
For immunohistochemistry/fluorescence studies coronal 10 µm thick cryostat sections were stained with the following primary antibodies: goat anti-doublecortin (DCX; 1:100, Santa Cruz); rabbit anti-GFAP (1:500, Dako); rabbit anti-GFP (1:500, Molecular Probes); mouse anti-GFP (1:1000, Abcam); chicken anti-GFP (1:500, Abcam); mouse anti-rat Nestin (1:100, BD Pharmigen); rabbit anti-c-fos (1:200, Santa Cruz); rabbit anti-Id1 (1:100, Biocheck Inc); rat anti-BrdU (1:100, Abcam); goat anti-TLR4 (1:50, Santa Cruz); mouse anti-S100β (1:100, Sigma); rat anti-CD31 (1:100, BD Pharmigen); rabbit anti-Iba1 (1:400, Wako); rabbit anti-NG2 (1:200, Santa Cruz). For the immunofluorescence on differentiated adult NPCs we used as primary antibodies: rabbit anti-β-tubulin (1:2000, Covance); mouse anti-O4 IgM (1:100, Chemicon) and rabbit anti-GFAP (1:600, Dako). Appropriate fluorophore-conjugated (Alexa Fluor® 488, 546 and 633; Molecular Probes) or biotinylated secondary antibodies were used according to manufacturer’s instructions.
Epilepsy induction via 4-aminopyridine administration
C57Bl/6-GCV−, C57Bl/6-GCV+, NestinTK-GCV+ and NestinTK-GCV− mice, aged 6–8 weeks (18–20 g), were subjected to 4-aminopyridine (Sigma) induced seizures. Animals were treated for 28 days with 100 mg/kg/day of GCV (subcutaneously administered through mini-osmotic pumps) up to 48 h prior to 4-aminopyridine injection, to allow drug washout. Intracerebroventricular AraC (C57Bl/6-AraC+) or saline (C57Bl/6-AraC−) were administered for 7 days before 4-aminopyridine injection as described previously. HU210 and URB597 were provided starting 12 h prior to 4-aminopyridine administration; NBQX was injected 20 min before the 4-aminopyridine administration, as described earlier.
Induction of transient middle cerebral artery occlusion
Male C57Bl/6 and NestinTK mice, aged 8–10 weeks (20–25 g), either treated with GCV or untreated, underwent 45 min of MCAO. For stroke studies, GCV animals were treated for 28 days with 100 mg/kg/day of subcutaneous GCV (administered through mini-osmotic pumps) up to 48 h prior to surgery, when GCV treatment was stopped to allow wash-out. HU210 and URB597 were administered starting 12 h prior to surgery and up to 24 h after surgery; NBQX was injected 20 min before the stroke induction and 10 min after MCAO onset, as described above. For the survival studies, animals were observed for 7 days and then sacrificed. Because of the high mortality in the NestinTK-GCV+ treatment group animals were sacrificed 3 days after ischaemia in order to conduct behavioural investigations and ischaemic volume measurements.
Coronal slices (200 µm) at the level of Bregma (Centonze et al., 2005, 2007) were transferred to a recording chamber and submerged in a continuously flowing artificial CSF (34°C, 2–3 ml/min) gassed with 95% O2–5% CO2. Presynaptic alterations were induced by treating acute brain slices with 100 µM of 4-aminopyridine (Sigma), as previously described (Flores-Hernandez et al., 1994). Postsynaptic modifications were induced by treating acute brain slices for 120 min with 0.6 µM of tumour necrosis factor-α (Sigma), as previously described (Centonze et al., 2009).
To study the effects of lipopolysaccharide-stimulated cells (LPS+aNPCs), adult NPCs were incubated for 16 h with 100 mg/ml of lipopolysaccharide (026:B6 Sigma). After extensive washing, 106 adult NPCs (either stimulated with lipopolysaccharide or not) were gently placed onto the surface of the striatum of the brain slice 30 min before the electrophysiological recordings. To study in vitro the effect of the cells treated with the fatty acid amide hydrolase inhibitor, URB597, (URB597+aNPCs), cells were incubated for 16 h with 30 nM of URB597 (Sigma) (Aguado et al., 2006). To study the effect of adult NPC-derived conditioned medium (from either LPS+aNPCs or LPS−aNPCs) acute brain slices were incubated 30 min prior to electrophysiological recordings in the conditioned medium, and were then, during registration, continuously perfused with CSF containing the conditioned medium (dilution of 10:1, CSF:conditioned medium). To verify the specific effect of adult NPCs, acute brain slices were incubated 30 min prior to electrophysiological recordings with dead LPS+aNPCs; after lipopolysaccharide stimulation cells were fixed for 15 min with 4% paraformaldehyde and then washed extensively with PBS. The cannabinoid agonist HU210 was added to the acute brain slices at the final concentration of 1 μM, whereas the cannabinoid antagonist AM251 was added to the acute brain slices at the final concentration of 10 μM.
Measurement of endocannabinoids
Lipids were extracted from cells or brain homogenates with chloroform/methanol (2:1, v/v), in the presence of arachidonoyl ethanolamide-d8 (AEA-8) and 2-arachidonoyl glycerol-d8 as internal standards (Giuffrida et al., 2000). The organic phase was dried and then analysed by liquid chromatography–electrospray ionization mass spectrometry (LC–ESI–MS), using a Perkin Elmer LC system in conjunction with a single quadrupole API-150EX mass spectrometer (Applied Biosystems). Quantitative analysis was performed by selected ion recordings over the respective sodiated molecular ions, as previously reported (Francavilla et al., 2009). The synthesis of 3H-AEA through the activity of N-acyl phosphatidylethanolamine phospholipase D was assayed in cells or homogenates of the striatum (50 µg/test), using 100 µM N-3H-arachidonoyl-phosphatidylethanolamine (3H-NArPE), as reported (Fezza et al., 2005). The hydrolysis of 3H-AEA by fatty acid amide hydrolase was assayed in cells or brain extracts (20 µg/test), by measuring the release of 3H-arachidonic acid from 10 µM 3H-AEA through reversed phase high performance liquid chromatography (Maccarrone et al., 2008).
For statistical analyses, we used a standard software package (GraphPad Prism version 4.00). As indicated within the text and the figure legends, data were evaluated by either unpaired two-tailed t-tests, Mann–Whitney test or not repeated ANOVA followed by Bonferroni or by Tukey honestly significant difference as post hoc test.
Selective ablation of adult neural stem/progenitor cells in the subventricular zone
To investigate whether endogenous adult NPCs in the subventricular zone may exert homeostatic regulatory functions other than neurogenesis in response to CNS injury, we first generated a transgenic mouse line (i.e. Nestin-floxGFPflox-TK-IRES-LacZ) in which adult NPCs can be labelled and selectively ablated by GCV administration. To specifically study subventricular zone adult NPCs we performed experiments on one of the transgenic lines generated (i.e. line 7457) in which, due to a positional effect (data not shown), both the GFP reporter protein and the thymidine kinase gene are prevalently, if not exclusively, expressed in the subventricular zone but not in the subgranular zone (Fig. 1A and B and Supplementary Fig. 1). In this mouse line, all GFP+ cells also stained positive for nestin, 42.4 ± 1.3% for Id1 and 18.4 ± 4.7% for GFAP (mean ± SEM) (Supplementary Fig. 1). None of the GFP+cells were positive for either the neuroblast marker doublecortin (DCX) (Supplementary Fig. 1), the endothelial marker CD31, the microglial marker Iba1, or the immature oligodendrocyte precursor and pericyte marker, NG2 (data not shown). One quarter (25.6 ± 5.3%) of the GFP+ cells abutted with the lateral ventricles and expressed the ependymal marker S100β (data not shown). Finally, some of the GFP+ cells were proliferating, being BrdU+ (10 h labelling protocol) (18.5 ± 6.3%). Altogether, these data indicate that our transgenic mouse GFP+ cells mostly comprise neural stem cells (or type B cells, encompassing both non-proliferating B− cells and proliferating B+ cells) and transit-amplifying cells (or type C cells).
We next examined the efficiency of thymidine kinase mediated adult NPC ablation in the double transgenic mouse line (NestinTK) obtained by crossing our line 7457 with a CMV-cre line (Supplementary Fig. 1). Selective ablation of subventricular zone, but not hippocampal adult NPCs was achieved in NestinTK mice by administering 100 mg/kg/day of GCV either intraperitoneally for 10 days or subcutaneously for 28 days (NestinTK-GCV+) (Fig. 1). PBS-treated NestinTK mice (NestinTK-GCV−) and wild-type GCV treated mice (C57Bl/6-GCV+) were used as controls.
In NestinTK-GCV+ mice treated for 28 days with GCV, we found a significant reduction of Dlx2 RNA expressing cells in the subventricular zone (Fig. 1I and J) despite the presence of only few terminal deoxynucleotidyl transferase dUTP nick end labelling positive apoptotic cells (Fig. 1K and L). In the same mice, we also found a sharp decrease of Id1+ (labelling only B+ cells), BrdU+ and DCX+ cells, either after 10 or 28 days of GCV administration (Fig. 1C–F and M–O). On the other hand, astrocytes and microglia (Fig. 1P) were not affected by GCV treatment. We also did not find any histological evidence of microglia activation in the subventricular zone or in the striatum of NestinTK-GCV+ mice (Fig. 1G and H).
To exclude any non-specific peripheral tissue damage as a consequence of GCV treatment, blood counts, serum electrolytes and serum glucose were measured in C57Bl/6-GCV+ and NestinTK-GCV+ mice. No signs of overt systemic toxicity were observed although a reduced body weight increase (but no weight loss) was measured in NestinTK-GCV+ mice compared with control mice (see Supplementary material). Hippocampal functions as well as striatal behavioural test, focusing on motor coordination, revealed no differences between NestinTK-GCV+ and control mice (Supplementary Figs 2 and 3).
At the same time, to exclude any non-specific CNS tissue damage due to GCV administration, messenger RNA levels of interleukin 1β, interleukin 6, interferon γ, tumour necrosis factor α, CD68, major histocompatibility complex II, CD11b, Emr1 and of the microglia marker Iba1 were measured in the striatum and in the subventricular zone of C57Bl/6-GCV−, C57Bl/6-GCV+, NestinTK-GCV− and NestinTK-GCV+ mice. No differences between groups were measured (Supplementary Table 1) thus confirming previous evidence indicating that GCV-treated mice display no increase in microglial activity (Singer et al., 2009, 2011; Jin et al., 2010).
Upon GCV withdrawal, impaired subventricular zone neurogenesis almost completely recovered over time in NestinTK-GCV+ mice treated with GCV for 10 days but, as expected, only partially in NestinTK-GCV+ mice treated for 28 days. Indeed in the latter, DCX+ cells were still significantly reduced compared with baseline values up to 30 days after GCV withdrawal (Fig. 1Q and R).
To confirm the specificity of the GCV-mediated adult NPC ablation in NestinTK-GCV+ mice, we further quantified the different subventricular zone cell subpopulations according to their morphological appearance by electron microscopy as previously described (Doetsch et al., 1997). We found a significant decrease of type A and type C cells but not of astrocytes (including both B+ and B− cells), ependymal cells and microglial cells after GCV treatment of NestinTK mice (Supplementary Fig. 1S–U) (Doetsch et al., 1997; Kokovay et al., 2010).
Finally, adult NPCs derived from NestinTK mice and cultivated in vitro as neurospheres were treated with different concentrations of GCV. GCV concentrations higher than 0.1 μM specifically impaired the survival and growth of neurospheres from NestinTK but not from NestinfloxGFPfloxTK and wild-type C57BL/6 mice (Supplementary Fig. 4).
Altogether, these data strongly indicate that GCV treatment in double transgenic NestinTK mice, generated from line 7457, provide a robust tool to specifically and transiently ablate the endogenous proliferating adult NPCs residing within the subventricular zone, without affecting hippocampal neurogenesis.
Mice with ablated subventricular zone adult neural stem/progenitor cells have altered striatal medium spiny neurons glutamatergic currents
Considering that subventricular zone adult NPCs are juxtaposed to the striatum, we hypothesized that any homeostatic regulatory function exerted by subventricular zone adult NPCs, alternative to neurogenesis, might affect striatal neurons. Thus, we sought to determine whether striatal neurons from NestinTK-GCV+ mice display functional alterations. To this aim, glutamate-mediated spontaneous excitatory postsynaptic currents (spontaneous EPSCs) from single medium spiny neurons were recorded by whole-cell patch clamp in acute slices using potassium-based pipettes. Experiments were conducted at− 80 mV holding potential and in the presence of bicuculline. Spontaneous EPSCs in NestinTK-GCV+mice had an increased frequency, increased decay time and an augmented half width, but no differences in the rise time (Fig. 2A–E). The specificity of the results was first confirmed by restoration of baseline spontaneous EPSC values in NestinTK-GCV+, 10 days after GCV wash out, a time required to proper restore neurogenesis in the subventricular zone of NestinTK mice (Fig.1Q and R).
To further confirm the role of subventricular zone-resident adult NPCs in regulating spontaneous EPSCs and to exclude any bystander effect due to thymidine kinase leakiness, spontaneous EPSCs were also recorded from striatal neurons in acute slices of wild-type C57Bl/6 mice treated intracerebroventricularly with the antimitotic AraC, a drug known to be effective in ablating adult NPCs in the subventricular zone (Fig. 2F–J and Supplementary Fig. 5) (Doetsch et al., 1999). Similar to NestinTK-GCV+, C57Bl/6-AraC+ mice displayed an increase of spontaneous EPSC frequency, decay time and half width but not of rise time. Once again, 10 days of AraC wash out restored baseline values in C57Bl/6-AraC+.
Altogether, our data support the idea that selective ablation of subventricular zone adult NPCs may induce deregulation of spontaneous EPSCs in medium spiny neurons, which most likely occurs in response to the activation of corticostriatal fibres. This is because spontaneous EPSCs recorded from single neurons are mediated by the synaptic release of glutamate and not by the release of glutamate from other sources, such as astrocyte, microglia and activated leucocytes. Furthermore, alterations of spontaneous EPSC kinetic properties reflect alterations of AMPA receptor sensitivity to synaptically released glutamate, while the increase of spontaneous EPSC frequency reflects the increase of presynaptic release.
Subventricular zone adult neural stem/progenitor cell-ablated mice display increased susceptibility to epilepsy and stroke
To validate in vivo the hypothesis that endogenous subventricular zone-resident adult NPCs might play a homeostatic regulatory function on striatal neurons in response to glutamatergic excitoxicity, NestinTK-GCV+ mice were subjected to 4-aminopyridine-induced epileptic seizures (Ayala and Tapia, 2005) and 45 min transient MCAO-induced ischaemic stroke (Hazell, 2007). These disease models were selected because they display pathological signs of glutamatergic excitotoxicity and profoundly affect striatal connections. MCAO-induced ischaemic stroke mainly affects striatal and cortical tissue and 4-aminopyridine is a selective potassium channel blocker causing presynaptic depolarizations and an increase in neurotransmitter release within the striatum (Deransart et al., 1998; Alexander and Peters, 2000; Kovacs et al., 2003) (Supplementary Fig. 6).
Ninety minutes after intraperitoneal injection of 4-aminopyridine, NestinTK-GCV+ mice displayed significantly reduced survival, more severe epileptic seizures and a significantly increased number of seizures compared with control mice (Fig. 3A–G). Similarly, NestinTK-GCV+ mice subjected to 45 min MCAO displayed a significantly decreased survival (Fig. 3H). Moreover, surviving NestinTK-GCV+ suffered an increased neurological disability 3 days after stroke (Fig. 3I). NestinTK-GCV+ retained less weight (Fig. 3J) and had an increased ischaemic lesion volume compared with control groups (Fig. 3K–M). Within the control group we purposely included both NestinTK-GCV− and C57Bl/6-GCV+ mice as they showed superimposable behaviour (Supplementary Fig. 7). To exclude that this phenomenon was due to neurogenesis, we measured the number of newly formed neuronal cells in the subventricular zone from both 4-aminopyridine treated and MCAO mice. In 4-aminopyridine mice we did not detect an increase of endogenous neurogenesis within the short observation period of 90 min (data not shown). Similarly, we observed only an initial increase of BrdU+ and DCX+ cells within the ipsilesional subventricular zone 3 days after MCAO in ischaemic C57Bl/6-GCV+ but not in NestinTK-GCV+ mice (Supplementary Fig. 8).
To confirm in vivo the role of subventricular zone-resident adult NPCs in protecting from excitotoxic damage, epileptic seizures and stroke were also induced in wild-type C57Bl/6 mice treated intracerebroventricularly with the antimitotic AraC (C57Bl/6-AraC+) (Doetsch et al., 1999). C57Bl/6-AraC+ mice displayed a significant increase in the severity and a reduced survival in response to 4-aminopyridine induced seizures (Supplementary Fig. 5). Similarly, 3 days after ischaemic stroke (45 min of MCAO) C57Bl/6-AraC+ had worse neurological disability on the modified Neurological Severity Score and reduced survival (Supplementary Fig. 5).
Finally, the use of the selective blocker of AMPA receptors, NBQX, in Nestin-TK-GCV+ mice was effective in reducing the increased mortality rate of epilepsy and the ischaemic volume of stroke down to the level of control mice, thus further confirming the key role of glutamatergic excitotoxicity in our disease models (Fig. 6).
Altogether, these results support the notion that endogenous adult NPCs in the subventricular zone exert a homeostatic regulatory role in protecting the CNS from glutamatergic excitotoxic damage.
Adult neural stem/progenitor cells attenuate striatal glutamatergic-mediated excitotoxicity
Depletion of subventricular zone adult NPCs increased in vivo morbidity and mortality to epilepsy and ischaemic stroke, two pathological conditions whose neurophysiopathology is characterized by glutamate-mediated excitotoxicity (Kovacs et al., 2003; Hossmann, 2006; Maroso et al., 2010). To gain insight into the molecular mechanism(s) underlying this phenomenon, patch-clamp recordings of spontaneous EPSCs from subventricular zone adjacent single medium spiny neurons were collected on acute brain slices from C57Bl/6 mice. To mimic in vitro the neurophysiological alterations measured in vivo, brain slices were pretreated with 4-aminopyridine or TNFα since it is well known that these molecules induce glutamatergic excitotoxicity via increasing spontaneous EPSCs frequencies (Barone et al., 1997; Centonze et al., 2009) and duration (Flores-Hernandez et al., 1994), respectively. Additionally, adult NPCs (in vitro passage 4–6) were pretreated with lipopolysaccharide (LPS+aNPCs) because it is known that binding of this molecule to toll-like receptor 4—a receptor expressed by adult NPCs (Rolls et al., 2007) (Supplementary Fig. 4)—mimics the effect of different danger (or recognition) signals triggering the inflammatory response. The addition of LPS+aNPCs, but not of LPS−aNPCs, to brain slices reverted the increased frequency (Fig. 4A and B), decay time and half width (Fig. 4C–F). Interestingly, we also found that conditioned medium from LPS+aNPCs, but not from LPS−aNPCs, reverted the increased spontaneous EPSC frequency. Moreover, we did not find any difference in spontaneous EPSC frequency when paraformaldehyde-treated LPS+aNPCs (i.e. dead cells) were added to brain slices (Fig. 4B). These latter results further suggest that only activated adult NPCs can restore the 4-aminopyridine-induced spontaneous EPSC alterations, measured in medium spiny neurons, through the secretion of a soluble mediator.
Adult neural stem/progenitor cells secrete the endogenous cannabinoid anandamide that regulates striatal glutamatergic tone in inflammatory conditions
The capacity of subventricular zone adult NPCs to regulate both the frequency and the duration of spontaneous EPSCs in striatal medium spiny neurons, possibly via the release of a soluble mediator, strongly suggests regulation occurring at pre- and postsynaptic level. Among neuromodulators acting pre- and postsynaptically as an ‘on demand’ protection system against glutamatergic excitotoxicity, cannabinoids, signalling through type 1 (CB1) and type 2 (CB2) cannabinoid receptors, are thought to play a relevant role (van der Stelt et al., 2002; Marsicano et al., 2003; Centonze et al., 2007; Docagne et al., 2007; Palazuelos et al., 2009; Navarrete and Araque, 2010; Di Marzo, 2011). Besides, adult NPCs not only express cannabinoid receptors but also secrete endogenous cannabinoids (Aguado et al., 2005; Oudin et al., 2011) that seem to be involved in the migratory and proliferative behaviour of neuroblasts (Galve-Roperh et al., 2007; Oudin et al., 2011). For these reasons, we considered that the endogenous cannabinoid system might be involved in adult NPC-mediated regulation of striatal glutamatergic transmission.
Through LC–ESI–MS (Marsicano et al., 2003; Francavilla et al., 2009), the level of the striatal endocannabinod AEA was found to be reduced in NestinTK-GCV+ mice, in agreement with recent evidence suggesting preferential expression and activity of AEA at striatal glutamatergic synapses (Maccarrone et al., 2008). Although not statistically relevant, a trend toward an increase of AEA levels was observed in C57Bl/6-GCV+, but not in NestinTK-GCV+ mice, upon in vivo challenge with lipopolysaccharide and in NestinTK-GCV− treated with 4-aminopyridine (Fig. 5A). Levels of 2-arachidonoyl glycerol were not changed in striatal tissue of treatment groups (11.4 ± 1.3 nM/g in C57Bl/6-GCV−, 14.8 ± 0.6 nM/g in C57Bl/6-GCV+, 14.4 ± 1 nM/g of tissue in C57Bl/6-GCV+ + LPS, 13.4 ± 1.5 nM/g in NestinTK-GCV+, 16.1 ± 0.7 nM/g in NestinTK-GCV+ + LPS, 9.3 ± 0.9 nM/g in NestinTK-GCV− + 4-aminopyridine and 8.9 ± 1.4 nM/g in NestinTK-GCV+ + 4-aminopyridine).
In vitro analyses by LC–ESI–MS in cultured subventricular zone-derived adult NPCs, confirmed that these cells constitutively secrete significant levels of AEA: these levels almost double upon lipopolysaccharide stimulation (Fig. 5B). In contrast, 2-arachidonoyl glycerol levels were undetectable in both LPS−aNPCs and LPS+aNPCs (data not shown). Moreover, LPS+aNPCs had increased levels of the AEA biosynthetic enzyme N-acyl phosphatidylethanolamine phospholipase D (Fig. 5C) and decreased levels of the major AEA catabolic enzyme fatty acid amide hydrolase (Fig. 5D). Altogether these data suggest that adult NPCs are able to secrete the endogenous cannabinoid AEA in response to danger signals.
In vitro and in vivo replenishment of endogenous levels of cannabinoids reverts the increased susceptibility to excitotoxicity of adult neural stem/progenitor cell-ablated mice
To corroborate the idea that AEA secreted by adult NPCs mediates the homeostatic regulatory effect exerted by subventricular zone adult NPCs on glutamatergic current, ex vivo and in vivo experiments were performed using the CB1 receptor antagonist AM251, the CB1 receptor agonist HU210 or URB597, the inhibitor of the primary degradation enzyme of endocannabinoids fatty acid amide hydrolase.
In vitro, we found that HU210 was indeed effective, as LPS+aNPCs or URB597+ pretreated adult NPCs (URB597+aNPCs) in reverting altered spontaneous EPSCs frequency induced by 4-aminopyridine. Moreover, the addition of the antagonist AM251 reverted the restorative effect on spontaneous EPSC frequency induced by LPS+aNPCs, URB597+aNPCs or by LPS+aNPCs-derived conditioned medium (Fig. 5E and F). AM251 also abolished the effect on spontaneous EPSC half width and decay time exerted by LPS+aNPCs on TNFα-treated slices (Fig. 5G–J). These data indicate that adult NPCs can produce AEA that, in turn, regulates spontaneous EPSCs in medium spiny neurons.
In vivo, intrastriatal administration of HU210 or URB597 in NestinTK-GCV+ mice was effective in reverting the increased mortality and disability observed in experimental epilepsy and stroke, respectively. As supposed, either HU210 or URB597 were highly effective in rescuing the high mortality rate induced by intraperitoneal administration of 4-aminopyridine (Fig. 6A and B) and in reverting the increased lesion volume observed upon induction of 45 min MCAO stroke in NestinTK-GCV+ (Fig. 6C).
These data indicate that adult NPCs exert a regulatory role on striatal glutamatergic currents through cannabinoid secretion, thus explaining the increased disability obtained in NestinTK-GCV+ mice suffering from epilepsy and stroke.
The peculiar position of the subventricular zone—several millimetres away from the olfactory bulbs—and the inconclusive evidence of the presence of a fully-formed rostral migratory stream in non-human primates and humans (Sanai et al., 2004, 2007, 2011; Wang et al., 2011) question the exclusive view that adult NPCs in the subventricular zone solely act as a source of newly formed neurons able to replace granule cells in the olfactory bulb upon migration along the rostral migratory stream (Imayoshi et al., 2008). On one side, subventricular zone adult NPCs are tightly apposed to blood vessels and are in contact with the CSF through apical processes, thus being in communication with two different microenvironments (Sawamoto et al., 2006; Mirzadeh et al., 2008; Tavazoie et al., 2008). On the other side, subventricular zone adult NPCs are very close to crucial areas of the forebrain (i.e. basal ganglia and striatum) containing GABAergic neurons capable of efficiently regulating and modulating interconnections between several cortical and sub-cortical brain areas (Koos and Tepper, 1999). Thus, we focused our attention on the role subventricular zone adult NPCs might exert on striatal neuronal functions.
Because direct anatomical connections between the hippocampus (where the other main neurogenic region of the brain is located) and the striatum are abundant, complex and well systematized (Voorn et al., 2004), we established a transgenic system to selectively ablate the sole subventricular zone adult NPCs to avoid confounding factors. Using this transgenic model, here we provide in vivo and in vitro evidence of a novel homeostatic regulatory mechanism exerted by subventricular zone adult NPCs, which can account for the positioning of such cells within the brain not just as an evolutionary relic. Our results strongly suggest that subventricular zone adult NPCs protect striatal medium spiny neurons from glutamatergic excitotoxicity by secreting the endogenous cannabinoid AEA that in turn, modulates corticostriatal glutamatergic currents by binding to the CB1 receptor. The in vitro demonstration that adult NPCs tune up AEA production when sensing danger signals further supports our idea of a homeostatic regulatory (innate) role exerted by subventricular zone adult NPCs in the early phase of the inflammatory process. TLR4 is triggered by several types of endogenous and exogenous danger signals, including those released after glutamate-induced excitoxicity [e.g. heat shock proteins 70, saturated fatty acids, high mobility group box 1 proteins (HMGB1) and β-defensins] characterizing the early (innate) phases of both ischaemic stroke and epilepsy (Rordorf et al., 1991; Mizukami et al., 2012; Zhang et al., 2012). We cannot, however, exclude that other signalling pathways other than toll-like receptors, might trigger homeostatic regulatory function exerted by subventricular zone adult NPCs. A recent report showed that subventricular zone adult NPCs exert a physiological phagocytic activity that requires intracellular engulfment protein ELMO1 to promote Rac activation downstream of phagocytic receptors (Lu et al., 2011). Moreover, newly generated neuroblasts residing within the subgranular zone are able to behave as antidepressant by regulating, through glucocorticoid release buffering, the hypothalamus–pituitary axis (Snyder et al., 2011).
We observed that subventricular zone adult NPCs exert their protective function in pathological conditions characterized by glutamate-mediated cellular excitoxicity leading to inflammation, such as epilepsy and ischaemic stroke. Our results are supported by recent studies demonstrating a direct link between adult NPCs dysfunctional activity and stroke. Evolving stroke worsens in aged mice where adult NPCs are significantly decreased (Darsalia et al., 2005), in mice with deficient subventricular zone adult NPCs migration abilities (Won et al., 2006), and in mice in which adult NPCs have been genetically ablated (Jin et al., 2010). Evidence supporting our findings is also emerging in experimental models of epilepsy; it has been recently shown that kainic acid-induced epilepsy is exacerbated in mice in which subventricular zone adult NPCs apoptosis is increased (Coremans et al., 2010). The next question is which type of adult NPCs mainly contribute to the observed particular protective effect? Within the subventricular zone, there are two subpopulations of bona fide neural stem cells, called type B+ and B− cells, which are in contact with the ependymal cell layer; the so-called type B− population do not actively proliferate and do not express the epidermal growth factor receptor (EGFR), while type B+ cells actively proliferate, express EGFR and Id1, and are mainly localized at the interface between the striatal parenchyma and the ventricular wall (Doetsch et al., 1997; Kokovay et al., 2010). Only type B+ cells give rise to actively proliferating (transit-amplifying) type C cells that function as the transit amplifying progenitors in the adult brain subventricular zone and which are scattered along the network of migrating (type A) neuroblasts (Alvarez-Buylla and Lim, 2004; Sanai et al., 2004, 2007, 2011; Imayoshi et al., 2008; Zhao et al., 2008; Wang et al., 2011). Our in vivo and in vitro data sustain the primary involvement of type B+, C and A cells since the transgenic system we used specifically ablates proliferating cells. Indeed, short and long term GCV-mediated ablation procedures, which induce a decrease of subventricular zone type B+, C and A cells, was sufficient to induce the electrophysiological glutamatergic alterations causing the increase of mortality and morbidity that we observed in subventricular zone adult NPC-ablated mice suffering from ischaemic stroke and epilepsy. On the other hand, type B− cells were not affected by the cell ablation procedure as the complete recovery of the cytoarchitecture of the subventricular zone occurred within 30 days after GCV washout. Finally, neurospheres (mainly composed of type C and A cells) were capable of regulating excitotoxicity and producing relevant amounts of cannabinoids (i.e. AEA) in vitro, when stimulated with lipopolysaccharide.
Finally, our results suggest that the homeostatic regulatory effect exerted by subventricular zone adult NPCs on glutamatergic excitotoxicity is mediated by endogenous endocannabinoids (AEA) released by subventricular zone adult NPCs and binding to their specific receptors (CB1 and CB2) (Marsicano et al., 2002, 2003; Di Marzo, 2011). The neuroprotective effect of the endocannabinoid system is not unexpected. Previous studies have clearly demonstrated that exogenous natural and synthetic cannabinoids may exert neuroprotective functions in several different models of neurotoxicity (van der Stelt et al., 2001) and neurological disorders (Nagayama et al., 1999; Baker et al., 2003; Croxford et al., 2008). The evidence that subventricular zone adult NPCs promoted AEA secretion upon sensing danger signals released by cells undergoing glutamatergic excitoxicity is also in agreement with the observation that endogenous cannabinoids are secreted as a physiological, on demand, protection system against the consequences of excessive neuronal activity (Marsicano et al., 2003). Lastly, the use of AEA as regulatory molecule by subventricular zone adult NPCs is noteworthy. Compared to other endogenous cannabinoids, AEA seems to preferentially modulate striatal glutamatergic transmission by binding to both pre- and postsynaptic CB1 receptor (Maccarrone et al., 2008; Rossi et al., 2011). Although in our experimental system 2-arachidonoyl glycerol was found below the detection limit, we cannot exclude that this endocannabinoid might be involved in the homeostatic control adult NPCs might exert on neuronal cells. It has been previously shown that adult NPCs from a different origin (e.g. embryonic rat cortices) might produce in vitro discrete amounts of 2-arachidonoyl glycerol (Aguado et al., 2005).
In conclusion, our results indicate for the first time that endogenous subventricular zone adult NPCs might exert a homeostatic (innate) regulatory role by protecting striatal neurons from glutamatergic excitotoxicity via the release of endogenous endocannabinoids. Although we cannot exclude that other factors rather than endocannabinoids may concur to the observed phenomenon (e.g. basic fibroblast growth factor) (Aguado et al., 2007), our study supports the concept that endogenous subventricular zone adult NPCs might act as guardians of the brain, since they are capable of sensing danger signals initiating an inflammatory process involving the innate arm of the immune system, and respond to them by down-modulating glutamatergic excitotoxic currents.
This work was supported in part by the Italian Multiple Sclerosis Foundation (FISM, to G.M.), the National Multiple Sclerosis Society (NMSS, RG 3591-A-1 to G.M.), the TargetBrain (EU Framework 7 project HEALTH-F2-2012-279017 to G.M.), the Italian Ministry of Research and University (MIUR), Italian Ministry of Health, BMW Italy Group, and CARIPLO Foundation. L.T. is supported by a fellowship from FISM. The founders had no role in study design, data collection, analysis, decision to publish or manuscript preparation.
Supplementary material is available at Brain online.
The authors thank Flavia Valtorta for helpful discussion. The authors acknowledge the technical help of C. Alfaro-Cervello, N. Battista, A. Bergamaschi, V. Bianchi, E. Vismara, P. Podini, M. Ruggieri, M. Amatruda, S. Sandrone, M. Gallizioli, L. Peruzzotti-Jametti and C. Marinaro.
cytosine β-d-arabinofuranoside hydrochloride
excitatory post-synaptic current
liquid chromatography–electrospray ionization mass spectrometry
middle cerebral artery occlusion
neural stem/progenitor cells