Author contributions: X.-H.Z. and T.-M.G.: conception and design and financial support; X.C.: performed the experiments; L.-P.L. and S.-J.L.: contributed to the cell culture; X.-H.Q. and Q.W.: performed the immunofluorescence staining and PCR; Y.-Y.F.: performed the stereotaxic microinjections; M.Z., L.-R.S., and W.-C.X.: contributed to the ATP measurements; H.-H.H.: performed the calcium imaging; X.-W.L. and Y.-B.G.: contributed to the animal care and genotyping; X.-H.Z., T.-M.G., and X.C.: manuscript writing and data analysis and interpretation.
Astrocytes are key components of the niche for neural stem cells (NSCs) in the adult hippocampus and play a vital role in regulating NSC proliferation and differentiation. However, the exact molecular mechanisms by which astrocytes modulate NSC proliferation have not been identified. Here, we identified adenosine 5′-triphosphate (ATP) as a proliferative factor required for astrocyte-mediated proliferation of NSCs in the adult hippocampus. Our results indicate that ATP is necessary and sufficient for astrocytes to promote NSC proliferation in vitro. The lack of inositol 1,4,5-trisphosphate receptor type 2 and transgenic blockage of vesicular gliotransmission induced deficient ATP release from astrocytes. This deficiency led to a dysfunction in NSC proliferation that could be rescued via the administration of exogenous ATP. Moreover, P2Y1-mediated purinergic signaling is involved in the astrocyte promotion of NSC proliferation. As adult hippocampal neurogenesis is potentially involved in major mood disorder, our results might offer mechanistic insights into this disease. STEM Cells 2013;31:1633–1643
Active adult neurogenesis has been unambiguously identified to occur in various species, including humans, in two discrete areas of the brain: the subventricular (SVZ) and subgranular (SGZ) zones of the dentate gyrus (DG) in the hippocampus. In the SGZ, the newly generated neuronal cells arise from neural stem cells (NSCs) and migrate into the granular cell layer, where they differentiate into granular neurons and integrate and participate in the local network [1, 2]. Hippocampal neurogenesis dynamically responds to a multitude of extrinsic stimuli and may be important for cognition and mood such as learning and memory [3, 4], pattern separation [5, 6], aging , major depression disorder [8, 9], and Alzheimer's disease .
The NSCs reside in a specialized microenvironment, or “niche.” Astrocytes, endothelial cells, and mature neurons are among the major cellular components of the adult neurogenic niche . The niche provides structural support and functionally regulates NSC proliferation and differentiation ; however, the cellular and molecular mechanisms by which an individual niche component controls the maintenance, self-renewal, and distinct developmental stages of the NSCs remain unclear. In the adult SGZ niche, there is accumulating evidence supporting the idea that astrocytes are key niche components and that astrocyte-derived factors have instructive effects that promote the neuronal differentiation of adult NSCs [1, 12-14]. Several factors such as Wnt3, IL-6, and IL-1β have been found to regulate NSC differentiation in vitro and in vivo [14, 15]; however, the exact molecular mechanisms by which astrocytes modulate NSC proliferation are unknown.
Here, we show that adenosine 5′-triphosphate (ATP) is a proliferative factor that induces astrocyte-mediated neurogenesis in the adult hippocampus. We found that ATP is necessary and sufficient for astrocytes to promote NSC proliferation in vitro. A lack of inositol 1,4,5-trisphosphate receptor type 2 (IP3R2) and transgenic blockage of vesicular gliotransmission inhibited ATP release from astrocytes and thus resulted in deficient NSC proliferation in the adult DG that could be rescued by the addition of exogenous ATP.
Tissue preparation, bromodeoxyuridine (BrdU) labeling, RNA analyses, and cell culture are described in detail in Supporting Information Materials and Methods.
Mice (three to four) were housed in a plastic cage (300 × 170 × 120 mm3) at 24°C ± 1°C under standard laboratory conditions: 12-hour light/dark cycle (lights on from 6:00 a.m. to 6:00 p.m.) with free access to food and water. All experiments were conducted in accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals (China). Male C57BL/6J mice (aged 10–12 weeks) and Wistar rats (postnatal day 0 or older than 6 weeks) were obtained from the Southern Medical University Animal Center (Guangzhou, China, http://www.fimmu.com).
IP3R2 knockout (IP3R2−/−) mice were generated by crossing germline heterozygous null mutant IP3R2+/− mice, a kind gift from Dr. Ju Chen (University of California, San Diego), as previously described . The offsprings were genotyped using PCR analysis of mouse tail DNA with WT (forward, GCTGTGCCCAAAATCCTAGCACTG; reverse, CATGCAGAGGTCGTGTCAGTCATT) and mutant allele-specific primers (neo-specific primer: forward, AGTGATACAGGGCAAGTTCATAC; reverse, AATGGGCTGACCGCTTCCTCGT). The PCR products were visualized after resolution on agarose gels with ethidium bromide staining.
The proopiomelanocortin (POMC)-enhanced green fluorescent protein (EGFP) transgenic mice were purchased from Jackson Laboratory (Cat # 009593, Sacramento, CA, http://www.jax.org) and generated as described . Male heterozygotes (3 months old) were used in this study.
Philip G. Haydon (Tufts University School of Medicine, Boston Campus) generously provided the GFAP-tTA and tetO.SNARE mice. The production of dn-SNARE transgenic mice has been described previously . Briefly, the GFAP.tTA mouse (tTA mouse) contains a glial fibrillary acidic protein (GFAP) promoter that drives the expression of the tet-off tetracycline transactivator. The tetO.SNARE line contains a Tet operator-regulated dominant-negative soluble N-ethylmaleimide-sensitive factor attachment protein receptor (dn-SNARE) domain of synaptobrevin 2 (corresponding to amino acids 1–96 of synaptobrevin 2) as well as lacZ and EGFP reporter genes. The dn-SNARE transgenic mice, in which dn-SNARE, LacZ, and EGFP transgenes are expressed in GFAP-positive astrocytes and suppressed by the application of doxycycline (DOX), were generated by crossing the GFAP-tTA and tetO.SNARE lines. The mice were mated and reared until 3 weeks of age in the presence of DOX (25 μg/ml in the drinking water) to prevent the potential developmental effects of transgene expression; thereafter, DOX administration was terminated.
For the fluorocitrate (FC) microinjection, C57BL/6J mice were anesthetized with chloral hydrate (0.35 g/kg b.wt., i.p.) and placed on a stereotaxic frame (Stoelting, Wooddale, IL, http://www.stoeltingco.com). The FC solution was prepared as previously described . FC (0.25 mM/1 μl, delivered at a rate of 0.1 μl/minute) or vehicle was injected into the hilus in the left side of DG (bregma: 2.0 mm anteroposterior, 1.2 mm lateral, and 2.0 mm vertical) using a calibrated 5-μl Hamilton syringe fitted with a 33-gauge needle and a syringe injector pump (Cat. 53311, Stoelting). Unless otherwise indicated, all compounds were purchased from Sigma-Aldrich (St. Louis, MO, http://www.sigmaaldrich.com).
For the lateral ventricular infusions (i.c.v.), after anesthetization, an adult IP3R2−/− mouse or control littermate was placed on a stereotaxic frame (Stoelting). A 5-μl Hamilton syringe was inserted perpendicularly into the right cerebral ventricle (bregma: 0.6 mm anteroposterior, 1.5 mm lateral, and 2.0 mm vertical), and an automatic injector was connected to a Hamilton syringe. ATPγs (50 μM), glutamate (5 μM), d-serine (10 μM), or adenosine (1 μM) dissolved in sterilized artificial cerebral spinal fluid (ACSF) was delivered at a rate of 0.5 μl/minute (total volume of 5 μl for each drug) into the lateral cerebral ventricle. BrdU (75 mg/kg b.wt., i.p.) was injected 15 minutes before the operation; 24 hours after BrdU injection, the mice were sacrificed, and the brain slices were examined for BrdU staining (Fig. 5A).
Preparation of the Conditioned Media
After astrocyte purification, the culture medium was replaced with adult NSC (ANSC) expansion medium (Cat. # SCM009, Millipore, Billerica, MA, http://www.millipore.com), and the astrocyte-conditioned medium (CM) was harvested after 2 days and filtered through a 0.2-μm syringe filter. CM collected from cultured adult hippocampal astrocytes, newborn hippocampal astrocytes, or spinal cord astrocytes was abbreviated ADH, NBH, or NBS, respectively. As a control, adult fibroblast condition medium (ASF) was prepared from the third passage of skin fibroblast resuspended using ANSC expansion medium for 2 days.
To estimate the molecular weight of the neurogenic component, the CM was concentrated through a 50 or 5 kDa molecular weight cutoff (MWCO) centrifuge concentrator (Vivascience AG, Hannover, Germany, http://www.sartorius.com) for 15 minutes at 3,000g and 4°C. For the ATPase experiments, the CM was treated with ATPase (30 units/ml, apyrase from potato, No: A7646) for 30 minutes. To investigate the mechanisms underlying the astrocyte ATP release, the astrocytes were treated with Gd3+ (50 μM), tetanus neurotoxin (TeNT, 100 nM), and bafilomycin A1 (BA1, 1 μM), and the CM was collected 2 days later. To test whether ATP release from the astrocytes depended on lysosome exocytosis, the cells were pretreated with glycyl-phenylalanine 2-naphthylamide (GPN, 200 μM) for 15 minutes followed by GPN wash out, and the ANSC expansion medium was added for the CM preparation (Fig. 3A).
In Vitro Proliferation Assay
The ANSCs were resuspended using CM (ASF, NBS, NBH, or ADH) containing BrdU (at a 2.5 μM final concentration) and plated on 24-well plates, which were coated with poly(l-lysine) solution. For the pharmacological experiments, the ANSCs were resuspended using NBS-containing BrdU and compounds. Twenty-four hours after BrdU incorporation, the ANSCs were fixed with ice-cold 4% paraformaldehyde for 15 minutes. The detection of BrdU required treatment in 2 M HCl at 37°C for 30 minutes. The secondary antibody, Alexa Fluor 488 goat anti-mouse IgG (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), was applied for 1 hour at room temperature. The coverslips were mounted with Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI, Vector Laboratories Inc., Burlingame, CA, http://vectorlabs.com) and visualized with fluorescent microscopy (Olympus, Japan, http://www.olympus-global.com). Images were analyzed using Image Pro-Plus software (Olympus, Japan). The number of positive cells was quantified systematically from three independent experiments of parallel cultures (four to six wells in each experiment). For each well, four fields were observed across the coverslip and 40–80 cells per field were counted.
The ANSCs were seeded onto 96-well plates and treated with CM for 1 day. The 3-(4, 5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT) solution (100 μl, 2 mg/ml) was added to the wells and incubated at 37°C for 3 hours. Subsequently, DMSO (100 μ1) was added to each well, and the absorbance at 590 nm was measured with a microplate reader. To measure astrocyte viability, astrocytes treated with FC (0.1 mM) or vehicle or cultured in different media (Dulbecco's modified Eagle's medium [DMEM]-F-12 or ANSC expansion medium) for 2 days were subjected to an MTT assay.
The ANSCs were cultured on glass-bottomed dishes (Mat Tek, Ashland, MA, http://glass-bottom-dishes.com) and incubated in Hanks' balanced saline solution (HBSS) (37.93 mM NaCl, 4.17 mM NaHCO3, 5.33 mM KCl, 0.441 mM KH2PO4, 0.338 mM Na2HPO4, 5.56 mM d-glucose, and 1.3 mM CaCl2, pH 7.4) containing fluo-4/AM at a concentration of 6–8 μM for 30 minutes. After carefully removing the unincorporated dye, variations of intracellular calcium concentration ([Ca2+]i) with time were measured using confocal microscopy (Olympus, Japan). When the [Ca2+]i baseline was stable, 200 μ1 of CM (ASF, ADH, or ADH containing ATPase [30 units/ml]) was added to the glass-bottomed dishes. For the blockage experiments, the ANSCs were preincubated with pyridoxal phosphate-6-azophenyl-2′,4′-disulphonic acid (PPADs, 50 μM), suramin (100 μM), or MRS2179 (10 μM); ADH was added 5 minutes later. To investigate the source of the Ca2+ waves, the medium was replaced with Ca2+-free HBSS containing 2 mM ethylene glycol tetraacetic acid (EGTA). The ANSCs were preincubated with thapsigargin (1 μM), ruthenium red (2 μM), 2-APB (100 μM), or U73122 (5 μM); ADH was added 5 minutes later. All experiments were conducted at room temperature (20°C). The fluo-4 fluorescence data were normalized to the baseline for each cell.
A bioluminescent ATP assay kit (Promega, Madison, WI, http://www.promega.com) was used to determine the amount of ATP as previously described . In brief, a 100-μ1 sample (CM or incubation medium for slices) was added to 100 μ1 of ATP assay mix containing luciferase-luciferin buffer. The ectonucleotidase inhibitor 6-N,N-diethyl-β-γ-dibromomethylene-d adenosine-5-triphosphate FPL 67156 (ARL 67156 trisodium salt) was added to the extracellular solution throughout the experiment to decrease ATP hydrolysis. Luminescence was measured using a luminometer (PerkinElmer, Waltham, MA, http://www.perkinelmer.com). A calibration curve was obtained from standard ATP samples, and the luminescence of the normal culture medium was measured as the background ATP level. The ATP level in detected cells was normalized to protein content. To examine the ATP levels in the hippocampus, acutely isolated hippocampal slices were incubated in the oxygenized ACSF, which was collected for the ATP assay 12 minutes later. For the measure of ADP, the Enzylight ADP Assay Kit (BioAssay Systems, Hayward, CA, http://www.bioassaysys.com) was used according to the manufacturer's instructions.
All the results are expressed as the mean ± SE of the mean. The statistical analyses were performed using SPSS 13.0 software. Potential differences between the mean values were evaluated using a one-way analysis of variance followed by the least significant difference test for post hoc comparisons when equal variances were assumed. Independent sample t tests were used to compare differences between any given two groups throughout the study unless otherwise specified. The significance level for all tests was set at p < .05.
FC Treatment Decreases the Number of BrdU+ Cells In Vivo
Microinjection of Evans blue dye (1 μ1) into the DG region resulted in region-specific diffusion (Fig. 1A); therefore, to investigate whether astrocyte dysfunction alters NSC proliferation in vivo, we injected FC, a gliotoxin used exclusively to selectively inhibit glial metabolism , into the left side of the DG of adult mice. This injection did not induce observable tissue injury and had no toxic effect on the morphology and density of the astrocytes in hippocampus (Supporting Information Fig. S1A, S1B). Then, the mice injected with BrdU (i.p., 75 mg/kg b.wt.) were subjected to stereotaxic microinjection of FC or vehicle and were sacrificed 24 hours after the BrdU injection (Fig. 1B). The unbiased stereological quantification of BrdU+ cells revealed that FC application decreased the number of BrdU+ cells by approximately 90% compared with controls [F(3, 20) = 52.428, p < .001, n = 6] (Fig. 1C). There was no difference between the injected and noninjected sides of the DG in the vehicle-treated mice (p = .112), thus excluding off-target effects of the surgery. These results suggest that astrocytes play an important role in regulating NSC proliferation in the adult hippocampus.
FC Treatment Inhibits BrdU Incorporation In Vitro
We then established in vitro assays to identify and determine the neurogenic effects of the proliferative factors secreted from astrocytes. Most ANSCs were positive for Nestin and actively proliferated in the ANSC expansion medium (Supporting Information Fig. S2). However, ANSCs cannot survive in the regular cultured medium of astrocytes (data not shown); therefore, to prepare the CM, DMEM-F12 was replaced with ANSC expansion medium after astrocyte purification, and CM was harvested after 2 days for the BrdU assay. No effect of this medium transfer was observed on the cultured astrocytes (Supporting Information Fig. S3). Then, we compared the ability of astrocyte-CM prepared from different tissues to promote ANSC proliferation. Immunofluorescence studies revealed that application of NBH or ADH increased the number of BrdU+/DAPI+ cells [F(4, 73) = 24.825, p < .001; n = 14–16 wells], whereas a control CM, ASF [13, 15], or NBS did not (Fig. 1D, 1F). Furthermore, the total number of DAPI+ cells increased 1 day after the application of both NBH and ADH [F(4, 73) = 17.836, p < .001; n = 14–16 wells] (Fig. 1E), indicating that ADH promoted ANSC proliferation. These results are consistent with a previous study in which the astrocyte-feeding layer derived from the adult hippocampus promoted ANSC proliferation . As expected, FC application blocked the increased percent of BrdU+/DAPI+ and number of DAPI+ cells induced by ADH (Fig. 1E, 1F). To exclude the possibility that FC affects the metabolism of adult NSCs and thus inhibits the ANSC proliferation induced by ADH, different dosages of FC (0.1–1.0 mM) were applied to the ANSCs, and BrdU incorporation was analyzed 24 hours later. We found that FC application had no neurogenic effects (Supporting Information Figs. S1C, S1D, S4). To further characterize the proliferative effects of ADH, we performed MTT assays. Consistently, we found that both ADH and NBH increased the MTT values [F(4, 75) = 100.532, p < .001; n = 16 wells], which were completely inhibited by FC application (Fig. 1G). Together, these results suggest that a factor or factors secreted from astrocytes are the active components regulating NSC proliferation.
ATP Is Sufficient to Mimic the ADH-Induced Increase in BrdU Incorporation
To determine which molecules contribute to NSC proliferation, we examined the molecular weight of the ADH component using 50 and 5 kDa MWCO filters. As ANSCs cannot survive in the medium containing only components <50 kDa or <5 kDa (data not shown), we separated the NBS and ADH into two parts using a 50 or 5 kDa filter and cross-mixed the media (Fig. 2A and Supporting Information Fig. S5A). While the medium containing the NBS <50 kDa and the ADH >50 kDa had no effect, the medium containing the NBS >50 kDa and the ADH <50 kDa increased the percent of BrdU+/DAPI+ (Supporting Information Fig. S5). Most interestingly, we found that the activity could be retained at <5 kDa, as the medium containing the NBS >5 kDa and the ADH <5 kDa produced a similar neurogenic effect to the ADH [F(4, 75) = 50.018, p < .001; n = 16 wells] (Fig. 2B and Supporting Information Fig. S6A). Together, these results strongly suggest that gliotransmitters could be active components of the ADH. Subsequently, we considered whether three classical gliotransmitters, glutamate, d-serine, and/or ATP , could promote ANSC proliferation. The addition of ATP increased the number of BrdU+/DAPI+ cells in a dose-dependent manner [F(4, 75) = 4.453, p = .016; n = 16 wells] (Fig. 2C and Supporting Information Fig. S6B, S6C), whereas glutamate, d-serine, or glutamate plus d-serine did not (Supporting Information Fig. S7). Moreover, similar to ATP, ATPγs (a nonhydrolyzable ATP analog) [F(4, 75) = 3.114, p = .032; n = 16 wells] and ADP [F(4, 75) = 8.482, p < .001; n = 16 wells] produced proliferative effects on ANSCs, but not adenosine [F(4, 75) = 0.688, p = .563; n = 16 wells] (Fig. 2C and Supporting Information Fig. S6B, S6C). As ANSCs expressed P2Y receptors (Supporting Information Fig. S8), we determined the effects of antagonists for purinergic P2 receptors, PPADs and suramin, and found that the proliferative effects of ATP, ATPγs, and ADP were completely blocked by PPADs or suramin, whereas PPADs or suramin alone had no neurogenic effect (Supporting Information Fig. S9). These results suggest that ATP might be the active component of ADH.
To test whether the proliferative effect of ADH was specific to ATP, we investigated the effects of the antagonists and found that the application of PPADs, suramin, or MRS2179 (a selective antagonist for the P2Y1 receptor), but not 8-cyclopentyl-1,3-dipropylxanthine (DPCPX, an antagonist for the adenosine A1 receptor), blocked the increased number of BrdU+/DAPI+ cells induced by ADH [F(6, 77) = 16.489, p < .001; n = 12 wells] (Fig. 2D and Supporting Information Fig. S6D), indicating that P2Y1-mediated purinergic signaling is involved in the neurogenic effect of ADH. Next, we exploited the ability of ATPase to catalyze the decomposition of ATP in the ADH and found that the neurogenic effect of ADH was inhibited [F(5, 90) = 81.237, p < .001; n = 16 wells]. To exclude the possibility that the toxicity of ATPase contributes to the inhibition, we used the <5 kDa MWCO strategy because the molecular weight of ATPase is approximately 45 kDa . The ratio of BrdU+/DAPI+ did not increase in the medium containing <5 kDa ATPase-treated ADH and >5 kDa ADH. Notably, this inhibitory effect could be reversed by the addition of ATP (Fig. 2E and Supporting Information Fig. S6E). To exclude the possible contribution of ATP-hydrolyzed products, we further measured the ATP level in the CM using bioluminescent ATP measurements. We found that ATP accumulated more in the NBH and ADH than in the control CMs, and FC treatment significantly decreased the concentration of ATP in the ADH [F(4, 32) = 1,171.113, p < .001; n = 7–8 wells]; there was no detectable ATP in the ATPase-treated ADH (Fig. 2F). Moreover, ADP measurements indicated that ADH contained a lower concentration of ADP relative to ATP. However, the ADP level in the ADH and NBH was not different from that in the NBS [F(2, 25) = 0.177, p = .839; n = 8–12 wells]; the level of ADP in the ATPase-treated ADH was undetectable (Fig. 2F). Together, these results indicate that ATP is necessary and sufficient for the astrocyte-mediated proliferation of ANSCs in vitro.
Exocytotic ATP Release Mediates the Mitogenic Effect of Astrocytes
Multiple mechanisms have been described for ATP release in astrocytes including vesicular exocytosis and nonexocytosis . To analyze which of these mechanisms contributes to the proliferative effect of ADH, we pretreated astrocytes with Gd3+, which has been shown to inhibit gap junction hemi-channels with high potency , and the medium (treated ADH) was collected 48 hours later (Fig. 3A). The application of Gd3+ had no effect on the proliferative effect of ADH [F(5, 66) = 22.882, p < .001; n = 12 wells; p = .617 compared with ADH) (Fig. 3B and Supporting Information Figs. S9, S10). In contrast, the application of TeNT, a specific neurotoxin that blocks the exocytotic release of neurotransmitters , blocked half of the mitogenic effect of ADH. TeNT treatment sensitively impairs glutamate exocytosis , whereas lysosomes are the primary sources of ATP in astrocytes [27, 28]; thus, it only partially reduces ATP release in astrocytes. Therefore, we used GPN to selectively dialyze lysosomes. To this end, astrocytes were treated with GPN for 15 minutes followed by GPN wash out, and the ANSC expansion medium was added for 2 days for the CM preparation (Fig. 3A). As expected, GPN application completely blocked the increased number of BrdU+/DAPI+ cells induced by ADH. Furthermore, application of the V-ATPase inhibitor BA1, which impairs ATP storage in the lysosomes , also abolished the proliferative effect of ADH, while the application of BA1 alone had no effect (Supporting Information Fig. S9). These results further support that the exocytotic release of ATP mediates the neurogenic effect of astrocytes.
IP3R2 Ablation Inhibits Astrocytic ATP Release and Adult Hippocampal Neurogenesis
Astrocytic ATP release is triggered by increases in intracellular calcium ([Ca2+]i), which are predominately elicited via the IP3 pathway. Because IP3R2 is the only astrocyte-specific functional IP3 receptor isoform [27, 29-33], we used IP3R2−/− mice, which have been documented to be a valid tool for the study of the functional role of astrocytes in situ [29, 33] to modulate ATP release from astrocytes in vivo. Previous data showed that the deletion of IP3R2 selectively disrupted the G-protein-coupled receptor-dependent [Ca2+]i responses in astrocytes [29, 33]. Therefore, we examined whether ATP levels were altered in the hippocampus of IP3R2−/− mice. To this end, acute hippocampal slices isolated from adult IP3R2−/− mice and control littermates were incubated in oxygenized ACSF and sampled 12 minutes later for ATP measurements. We observed that the amount of ATP was lower in the media incubating hippocampal slices derived from IP3R2−/− mice [t(1, 10) = 86.0, p < .001, n = 6], suggesting that astrocytic ATP release was deficient in adult hippocampus of IP3R2−/− mice, which was further evidenced by an in vivo microdialysis assay (unpublished data). Moreover, the accumulation of ATP was deficient in the medium of astrocytes isolated from newborn IP3R2−/− mice [t(1, 14) = 63.083, p < .001, n = 8 wells] but not in the medium of cultured neurons [t(1, 14) = 0.99, p = .345, n = 8 wells] (Fig. 4A), further indicating that the astrocytic ATP release was specifically affected in IP3R2−/− mice. We then investigated whether the proliferative effect of astrocytes was altered by the lack of IP3R2. In vitro assays revealed that the NBH derived from IP3R2−/− mice did not show mitogenic effects [F(4, 75) = 96.467, p < .001; n = 16 wells) (Fig. 4B and Supporting Information Fig. S11). For the in vivo studies, the adult IP3R2−/− mice and control littermates were singly injected with BrdU (75 mg/kg b.wt.); brain slices were prepared 24 hours later for BrdU staining. The number of BrdU+ cells in the DG of IP3R2−/− mice decreased by approximately 95% compared with the control littermates [t(1, 14) = 72.628, p < .001; n = 8] (Fig. 4C). To further detect neurogenesis deficiencies in the IP3R2−/− mice, we crossed IP3R2−/− mice with POMC-green fluorescent protein (GFP) transgenic mice, such that newborn immature neurons in the DG expressed GFP . The unbiased stereological quantification of GFP+ cells revealed that the number of GFP+ cells dramatically decreased in the adult IP3R2−/−-POMC mice compared with the control littermates [t(1, 10) = 6.323, p < .001; n = 6] (Fig. 4D). These results suggest that ATP is necessary for astrocytes to regulate NSC proliferation in vivo.
ATP Treatment Reverses the Deficiencies of Proliferation of NSCs in IP3R2−/− Mice and dn-SNARE Mice
To test whether ATP could reverse the deficient NSC proliferation in adult IP3R2−/− mice, we performed lateral ventricular infusions (i.c.v.). To exclude the possible contribution of hydrolyzed ATP products, we used ATPγs. BrdU was singly given 15 minutes before the drug infusion, and mice were sacrificed 24 hours after the BrdU injection (Fig. 5A, top). The ATPγs infusion reversed the decreased number of BrdU+ cells in the DG of IP3R2−/− mice to that found in the control littermates, whereas infusions of the glutamate, d-serine, or adenosine had no effect [F(5, 42) = 6.751, p < .001; n = 8] (Fig. 5A).
The astrocyte-specific transgenic expression of dn-SNARE domain of synaptobrevin 2 is presumed to inhibit astrocyte ATP release . As an i.p. injection of ATP (125 mg/kg) could reverse the hippocampal ATP levels of dn-SNRE mice to that of control as indicated by ATP measurements of the media incubating acute hippocampal slices (Supporting Information Fig. S12), to further determine the role of astrocytic ATP release in promoting NSC proliferation, adult dn-SNARE mice, wild-type, and tetracycline transactivator single transgenic (tTA+) control littermates were injected with ATP (125 mg/kg, i.p.) daily for 7 days followed by a single BrdU injection (75 mg/kg, i.p.) on the seventh day. The brain slices were prepared 24 hours later for BrdU staining (Fig. 5B, top). The number of BrdU+ cells decreased in the DG of the vehicle-treated dn-SNARE mice compared with controls, suggesting that NSC proliferation was deficient in the hippocampus of adult dn-SNARE mice [F(5, 30) = 6.489, p < .001; n = 6]. Importantly, ATP administration reversed the decreased number of BrdU+ cells in the DG of dn-SNARE mice to the levels observed in the control mice, although the ATP injections had no obvious effect on the control mice (Fig. 5B). Together, these results support the hypothesis that astrocyte-derived ATP is necessary and sufficient for astrocytes to regulate NSC proliferation in vivo.
ADH-Induced Calcium Waves Require P2Y1-Mediated Purinergic Signaling and IP3-Mediated Calcium Release in ANSCs
ATP induces transient increases of [Ca2+]i and thus stimulates NSC proliferation during development [34, 35]. To better define the mitogenic effect of astrocytes, we used confocal time-lapse calcium imaging. As a positive control, a physiological concentration of ATP (25 μM)  induced transient increases of [Ca2+]i in ANSCs [F(3, 60) = 17.459, p < .001; n = 16 cells] (Fig. 6A, 6B). Interestingly, ADH, similar to ATP, stimulated [Ca2+]i increases in ANSCs, which was inhibited by the application of PPADs, suramin, or MRS2179 [F(9, 110) = 59.94, p < .001; n = 12 cells]. There was no calcium response to ATPase-treated ADH, whereas the addition of exogenous ATP resulted in a robust [Ca2+]i increase again after ATPase-treated ADH was washed out (Fig. 6B). These results further supported that ATP is the active component of ADH, and P2Y1-mediated purinergic signaling might be involved in the astrocyte-mediated proliferation of NSCs. Moreover, we found that the ADH-stimulated [Ca2+]i increases were not attenuated when the extracellular calcium was removed using the calcium chelator EGTA (2 mM; p = .899 compared with ADH). However, the calcium waves decreased following incubation of the ANSCs with thapsigargin, which depleted intracellular calcium stores (Fig. 6C), suggesting a major role for the release of calcium from the endoplasmic reticulum (ER). Ryanodine and IP3 receptors located within the ER are responsible for intracellular Ca2+ release; therefore, we tested which of these receptors was specifically involved. Incubation of ruthenium red, which inhibits calcium release from ryanodine sensitive stores, had no effect (p = .567 compared with EGTA); in contrast, 2-APB, which blocks IP3-sensitive release, or U73122, a phospholipase C (PLC) inhibitor , decreased the ADH-induced transient increase of [Ca2+]i (Fig. 6C). These findings suggest that P2Y1-PLC-phosphatidylinositol 3-kinase (PI3K) signaling was required for astrocyte-mediated NSC proliferation.
In this study, we identified ATP as a proliferative factor required for astrocyte-mediated NSC proliferation in the adult hippocampus. Using FC injections, we found that astrocytes play an important role in regulating NSC proliferation in the adult hippocampus. Our in vitro studies revealed that ATP is necessary and sufficient for astrocytes to promote ANSC proliferation. Moreover, we demonstrated that exocytotic ATP release mediated the mitogenic effect of astrocytes. Using pharmacological and genetic strategies, we observed that a lack of IP3R2 and transgenic blockage of vesicular gliotransmission resulted in the dysfunction of ATP release from astrocytes and deficient proliferation of NSCs in the adult DG, which could be rescued by adding exogenous ATP. Our collective results together with reports that implicate astrocytes in NSC differentiation [13, 14] indicate that astrocytes play a key role in regulating adult hippocampal neurogenesis.
ATP has been identified as a mitogen for NSC and ATP-mediated purinergic signaling through the P2Y1-PLC-PI3K signal pathway that has been associated with developmental neurogenesis [34, 38]. Recent studies have suggested that extracellular purinergic signaling also promotes the proliferation of NSCs in vitro [35, 39] and in the adult SVZ . However, whether ATP promotes the proliferation of NSCs in the adult DG and the sources of ATP remain unclear. Here, we found that astrocytic ATP release is necessary and sufficient to promote the proliferation of ANSCs in vitro and in the adult hippocampus. Furthermore, we observed that ANSCs express P2Y receptors and that ADH induced a transient increase in [Ca2+]i in ANSCs that could be blocked by antagonists for P2Y1 receptors, PPADs, suramin, and MRS2179. In addition, ADH stimulated [Ca2+]i waves required PLC activation and IP3-mediated calcium release. These results suggest that P2Y1-PLC-PI3K signaling is involved in the astrocyte-mediated NSC proliferation in the adult DG.
There are two types of dividing cells in the adult SGZ: radial glia-like cells (RGLs) with low levels of proliferation and amplifying neural progenitors (ANPs) with high levels of proliferation. Clonal analysis identified that RGLs give rise to ANPs through asymmetric division [4, 41]. Studies have demonstrated that the modulation of RGLs could affect neuron production, neuronal migration, and overall cortical architecture during development . However, a histological analysis has shown that there are no obvious structural abnormalities in the brains of adult IP3R2−/− mice [29, 33], suggesting that RGL properties may not be affected by the lack of IP3R2. In this study, we observed that ATP infusion could reverse the decrease in the number of BrdU+ cells in the DG of adult IP3R2−/− mice, suggesting that the deficiency of astrocytic ATP release might be the cause of the dysfunction of hippocampal neurogenesis in adult IP3R2−/− mice rather than changes in the NSCs themself. Future experiments should examine which type of dividing cells in the adult SGZ are modulated by astrocytic ATP.
Extracellular ATP is hydrolyzed rapidly (approximately 200 ms) from ATP to ADP to AMP to adenosine by ectonucleotidases, which are highly expressed in the adult DG [35, 42]. In this study, we found that both ATP and ADP promoted ANSCs proliferation and that ADH contained ADP. However, ADP might not contribute to the mitogenic effect of ADH, possibly due to a low concentration of ADP in the ADH; because NBS containing similar ADP level was not able to promote ANSCs proliferation. We also found that it was ATP, not adenosine, which rescued the deficiency of adult neurogenesis in IP3R2−/− mice, suggesting that astrocytes and NSCs need to be very close in proximity. In fact, in the DG, astrocytes are intimately associated with the NSCs . Therefore, we propose that in the hippocampal niche, astrocytes maintain NSC proliferation by releasing ATP, which acts through the P2Y1 receptor expressed on the NSCs. After NSCs transition from their original sites, the mitogenic effect of astrocytes is immediately delimited by local ectonucleotidases, and other factors secreted from the astrocytes, such as Wnt3, might allow NSCs to proceed to neuronal differentiation. Of cause, a previous report showed that NSCs isolated from E13 mice were themselves the sources of local ATP ; we cannot exclude the possibility that NSCs themselves release ATP, triggering the release of ATP from astrocytes promoting NSC proliferation in the adult DG.
In the adult brain, NSCs were eventually determined to be focused in two discrete regions: SVZ and SGZ. These cells are characterized by the capabilities of self-renewal and the production of neurons and glial cells; however, NSCs in these two regions behave differently in many ways including cell type composition, moving structure, destinations, and even molecular modulations . For example, NSCs isolated from the adult SVZ respond to adenosine stimulation (1–50 μM) ; however, in this study, we did not observe any effect of adenosine on the proliferation of ANSCs isolated from the adult hippocampus (Fig. 2C). Furthermore, compared with neurogenesis in the adult SVZ, adult hippocampal neurogenesis may be important in major mood disorder (MDD) . There is accumulating evidence that a variety of chronic stressors reduce cell proliferation in the adult DG; in contrast, multiple classic antidepressants increase the proliferation of NSCs in a chronic, but not acute, time course. Importantly, hippocampal neurogenesis-deficient mice display a depressive phenotype , and blockage of adult hippocampal, not SVZ, neurogenesis abolishes the behavioral effects of most available antidepressants [8, 10], suggesting that decreases in adult hippocampal neurogenesis and MDD may be causally related. In addition, a large body of studies has implicated glial dysfunction in MDD pathophysiology [44–47]. However, until now, the molecular mechanism through which astrocytes modulate depressive behaviors has been largely uncharacterized. In this study, we found that astrocyte-derived ATP is necessary and sufficient to promote NSC proliferation in the adult DG. Therefore, future studies should test whether ATP release from astrocytes is important for modulating the expression of behaviors related to MDD.
In summary, we found that ATP release from astrocytes is necessary and sufficient to promote NSC proliferation in vitro and in the adult hippocampus. As adult hippocampal neurogenesis is causally related to MDD, we suggest that future studies focus on how stress specifically results in decreased astrocyte ATP release, which will further the field's understanding of MDD pathophysiology.
We thank Dr. Ju Chen and Dr. Philip G. Haydon for kindly providing the mouse lines. This work was supported in part by grants from the National Natural Science Foundation of China (Nos. 81171276, 8103002, and U1201225), the Key Project of Guangdong Province (Nos. 9351051501000003, CXZB1018, and 2011A032100001), the Guangzhou Science and Technology Project (No. 7411802013939), the Program for Changjiang Scholars and Innovative Research Team in University (No. IRT1142), the Major State Basic Research Program of China (No. 2012CB518203), and GDUPS (2011).
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.