Synaptogenesis by Cholinergic Stimulation of Astrocytes

Astrocytes release numerous factors known to contribute to the process of synaptogenesis, yet knowledge about the signals that control their release is limited. We hypothesized that neuron-derived signals stimulate astrocytes, which respond by signaling back to neurons through the modulation of astrocyte-released synaptogenic factors. Here we investigate the effect of cholinergic stimulation of astrocytes on synaptogenesis in co-cultured neurons. Using a culture system where primary rat astrocytes and primary rat neurons are first grown separately allowed us to independently manipulate astrocyte cholinergic signaling. Subsequent co-culture of pre-stimulated astrocytes with naïve neurons enabled us to assess how prior stimulation of astrocyte acetylcholine receptors uniquely modulates neuronal synapse formation. Pre-treatment of astrocytes with the acetylcholine receptor agonist carbachol increased the expression of synaptic proteins, the number of pre- and postsynaptic puncta, and the number of functional synapses in hippocampal neurons after 24 hours in co-culture. Astrocyte secretion of the synaptogenic protein thrombospondin-1 increased after cholinergic stimulation and the inhibition of the target receptor for thrombospondins prevented the observed increase in neuronal synaptic structures. Thus, we identified a novel mechanism of neuron-astrocyte-neuron communication, i.e., neuronal release of acetylcholine stimulates astrocytes to release synaptogenic proteins leading to increased synaptogenesis in neurons. This study provides new insights into the role of neurotransmitter receptors in developing astrocytes and into our understanding of the modulation of astrocyte-induced synaptogenesis.

Additionally, astrocyte released secreted protein acidic and rich in cysteine (SPARC) inhibits the synaptogenic effect of hevin [16]. Despite the increasing identi cation of astrocyte-released synaptogenic factors, knowledge of the signals controlling their release remains for the most part unknown.
The fact that astrocytes are integral to neuronal development and that they express ACh receptors (AChRs) and respond to synaptic release of ACh suggests that stimulation of astrocyte AChRs may promote neuronal development by modulating the expression and release of factors that create an environment conducive to it. Here we test the hypothesis that the selective cholinergic stimulation of astrocytes increases synaptogenesis through the release, by astrocytes, of factors that potentiate the formation of synaptic structures in co-cultured neurons. Our results show that pre-treatment of astrocyte cultures with a cholinergic agonist leads to enhanced expression of neuronal synaptic proteins, an increase in the number of pre-and postsynaptic puncta, and an increase in the number of structural and functional synapses in hippocampal neurons co-cultured with carbachol-pre-treated astrocytes. This enhanced synaptogenic function of astrocytes is prevented by the inhibition of TSP receptors in the coculture system and mimicked by exogenous TSP1, an extracellular matrix protein whose release is increased after cholinergic stimulation of astrocytes. The elucidation of the signals that regulate changes in astrocyte-secreted proteins involved in synaptogenesis is a key step toward the understanding of normal and pathological neurodevelopmental processes.
To hinder astrocyte proliferation, ARAC was added at a nal concentration of 2.5 µM to each well after three days in culture. One-third of the medium was changed every three to four days post ARAC treatment. On day in culture (DIC) 12-13, neurons were either co-cultured with primary astrocytes or treated with human TSP1 for 24 hours (see below).

Primary Astrocyte Preparation and Culture
Primary cortical astrocytes were dissected and prepared from E21 Sprague Dawley rat pups, as previously described [39,48,49] and cultured in 75 cm 2 asks coated with 40 µg/mL PDL at an initial density of 2.5-3.0 x 10 6 cells/ ask. Astrocytes were grown to con uency in Dulbecco's Modi ed Eagle's Medium supplemented with 10% fetal bovine serum and penicillin (100 U/mL) and streptomycin (100µg/mL) (P/S).

Astrocyte Treatments and Neuronal Co-Culture
Primary cortical astrocytes were trypsinized from established cultures and sub-cultured in 24 well plates (250,000 cells/mL) for synaptogenesis and electrophysiology experiments, or on the underside of 0.4 µm mesh 6-well plate inserts for protein expression experiments. Wells and inserts were coated with 40 µg/mL PDL. Forty-eight hours prior to co-culture with neurons, astrocytes were serum deprived (DMEM + 0.1% bovine serum albumin and P/S) for 24 hours, and then treated with or without carbachol (0.01, 0.10, 1 mM) for an additional 24 hours. Astrocytes were washed with PBS and incubated in serum-free medium for three hours prior to co-culture with neurons. In some experiments, 30 minutes prior to the addition of carbachol, astrocytes were treated with 10 µM of the acetylcholine receptor antagonists, mecamylamine, gallamine, or 4-DAMP. After treatment washout and medium conditioning, primary hippocampal neurons (12)(13), grown on glass coverslips were inverted over the pre-treated astrocyte monolayer for immunocytochemistry or electrophysiology experiments. Neurons were never in direct contact with astrocytes, nor were they exposed to astrocyte treatments. For Western blot experiments, astrocytes plated on the underside of porous inserts and their medium were transferred to 6-well plates containing primary hippocampal neurons.
To block TSP1 signaling in neurons, after astrocyte carbachol pre-treatment and washout, neurons and astrocytes were pre-treated for half hour with gabapentin (15 or 30 µM) prior to co-culturing; gabapentin remained throughout the co-culture incubation. For all co-culture experiments, astrocytes and neurons were co-incubated for 24 hours.

Immunocytochemistry, Image Acquisition and Analysis
Neurons were xed in 4% PFA for 20 minutes at 37°C, blocked and permeabilized for 30 minutes (0.1% Triton-X, 5% bovine serum albumin in PBS) and co-labeled with primary antibodies against synaptophysin (1:250) and PSD-95 (1:200), or synaptotagmin (1:250) and NR2B (1:200) for at least 18 hours at 4°C under slow rocking. Coverslips were then incubated for 1 hour at room temperature with uorescent secondary donkey anti-rabbit Alexa 488, and donkey anti-mouse Alexa 555 antibodies (1:500) and the nuclear dye Hoechst 33342 (1 µg/mL). Coverslips were mounted on glass slides using Vectashield, topped with a cover glass, and sealed with nail polish. Individual, healthy neurons, identi ed with Hoechst stain, and located at least two cell bodies from their nearest neighbor, were imaged using confocal microscopy (Olympus Fluoview-1000) Confocal images were acquired using a 60X oil immersion objective, a 1024 x 1024 image size with a 2X optical zoom, yielding a 0.103 µm/pixel resolution. An average of 12-18 slices per neuron were acquired through the z-plane, using a step size of 0.30 µm from 5 neurons per coverslip. To capture the entire range of uorescent signal, the slide from the treatment group with the greatest intensity was used to set the imaging parameters for each channel and the settings remained constant during acquisition of images from each treatment groups within an experiment. Images were deconvolved using Huygens Professional software (Scienti c Volume Imaging). For object analysis, the surface of synaptic puncta was rendered three-dimensionally based on a size and intensity threshold which was determined by obtaining the mean intensity of 30 manually selected, size appropriate puncta from 15 neuronal elds of each channel (350-450 total) using Image J, from the treatment group used to set confocal imaging parameters. Channel thresholds were held constant for all treatment groups within an experiment. The number of individual pre and postsynaptic objects and those overlapping between channels were automatically calculated and recorded using Huygens Object Analysis software.
For immunocytochemical labeling of TSP1 on the astrocyte surface, astrocytes were plated on PDLcoated glass coverslips at a density of 250,000 per well, in 24 well plates. Control and carbachol-treated astrocytes were incubated for 24-hours, washed in PBS, and xed in 4% PFA for 20 minutes at 36°C, but not permeabilized in order to label only cell-surface TSP1. Astrocytes were then blocked in 10% goat serum and incubated in anti-mouse TSP1 primary antibody (1:80 in PBS) for 3 hours at room temperature, followed by donkey anti-mouse Alexa 488 (1:75, PBS, 3% BSA) secondary antibody and the nuclei stained with Hoechst 33342 (1:2000 in PBS). Cells were mounted on slides using Vectashield and imaged using confocal imaging with parameters held constant between treatment groups. Confocal images were acquired using a 40X oil immersion lens, at a step-size of 0.50 µm for a total of 18-21 planes per eld. Nine to ten elds were imaged per treatment group from two coverslips. For each eld, the integrated optical density per plane was determined using Metamorph software (Molecular Devices), totaled, normalized to cell number and averaged for each treatment group. For presentation of a representative image, the maximum intensity Z-projection was obtained from confocal images using Image J, and channels were merged.

Western blot
Neurons were lysed in 1% SDS lysis buffer, sonicated twice at 3.5 power for 5 seconds and the protein quanti ed using the bicinchoninic acid assay (BCA). Equal amounts of protein were loaded into freshly cast bis-trisacrylamide gels (10%) and proteins were separated by gel electrophoresis then transferred to PVDF membranes. Membranes were blocked in TBST (5% milk) for 1 hour, then probed overnight at 4˚C with primary antibodies for synaptophysin (1:500), or PSD-95 (Neuromab, 1:500). Membranes were incubated for 1 hour with HRP-conjugated secondary antibodies (1:1000) and developed. Band densitometry was determined using Image J software, and results normalized to total protein as determined by Coomassie blue protein stain.
For measurement of astrocyte TSP1 levels, astrocytes were plated on PDL (40 µg/mL) pre-coated 100 mm plates at a density of 2.5 x 10 6 per plate and cultured for 4 days, serum-deprived for 24 hours, then treated for 24 hours with carbachol (1 mM) prepared in serum free medium. Medium was collected and cells were lysed in 1% SDS lysis buffer from treated and untreated plates immediately after the 24h treatment (time 0), and 6 and 24 hours after treatment washout. Intracellular protein was quanti ed from lysate samples using the BCA method and equal amounts were loaded into 3-8% Tris-Acetate gels and separated by gel electrophoresis. After transfer, PDVF membranes were blocked in TBST (3% BSA) for 1 hour and probed overnight in goat anti-mouse TSP1 (1:1000), then incubated for 1 hour with HRPconjugated secondary antibodies (1:1000) and developed. TSP1 was detected by chemiluminescence and normalized to β-actin levels using densitometric analysis. For measurements of TSP1 in the astrocytic medium, 7 mL of media from each time point was concentrated to 200 µl using Pierce concentrators (MWCO 9kDa) centrifuged at 4000 x g for 25 minutes at 25 C. After the addition of protease inhibitors, reducing reagents, and sample buffer, samples were denatured by heating (70°C for 10 minutes) and equal volumes (30 µl) of concentrated sample were loaded into 3-8% Tris-Acetate precast gels. After transfer, PDVF membranes were blocked in TBST (3% BSA) for 1-3 hours and incubated overnight in goat anti-mouse TSP1 (1:500). Band densitometry was determined using Image J software and normalized to corresponding cell lysate protein concentrations.

Electrophysiology
Spontaneous miniature excitatory postsynaptic currents (mEPSCs) were recorded at room temperature (21.5-23.5˚C) from hippocampal neurons after co-culture with carbachol (1 mM) pre-treated astrocytes, or untreated astrocyte controls using whole cell patch clamp techniques. Currents were recorded using a MultiClamp 700B by Axon Instruments; cells were voltage clamped at a holding potential of -70 mV. Patch pipettes were pulled from borosilicate capillary glass (2.4-5.9 MΩ). Dissociated hippocampal neurons were bathed in arti cial cerebral spinal uid (ACSF) (119 mM NaCl, 2.5 mM KCl, 4 mM CaCl2, 4 mM MgCl2, 26 mM NaHCO3, 1 mM NaH2PO4, 11 mM glucose at pH of 7.4 and gassed with 5% CO2 and 95% O2). Tetrodotoxin (1 uM) was added to the circulating bath to isolate spontaneous miniature postsynaptic currents. Internal solution contained: 115 mM CsMeSO 4 , 20 mM CsCl, 2.5 mM MgCl 2 , 10 mM HEPES, 4 mM Na 2 ATP, 0.4 mM Na 3 GTP, 10 mM Na-phosphocreatine, and-0.6 mM EGTA, pH 7.25 (CsOH). Activity was recorded for up to ten minutes and events were manually selected using Synaptosoft Mini Analysis software. Traces from individual neurons of each treatment group were combined into arrays and the inter-event interval (IEI), amplitude, and time to decay, cumulative distributions were plotted (7-9 neurons per treatment group array from 3 independent experiments). The frequency of events, as measured by inter-event-intervals, was used to represent the relative number of functional synapses.

Statistical Analysis
Student's t-test was used for statistical comparison of the means in experiments containing two treatment groups and One-way ANOVA with Dunnett's Multiple Comparison test was used where multiple concentrations were assessed relative to control. To compare multiple treatments, one-way ANOVA followed by Bonferroni's Multiple Comparison tests was performed. Electrophysiology analysis was performed using Synaptosoft Mini Analysis software and Graph Pad Prism. Kolmogorov-Smirnov Two Sample Analysis was used to compare the cumulative distributions from arrayed electrophysiology recordings. In gures where individual data points are presented using dot plots, the means and standard error of the means (SEM) are shown. For experiments with a high number of data points, the distribution of the data, the median, and the interquartile range, or the 25th and 75th percentiles, are presented using violin plots and signi cant differences of the means are indicated.

Cholinergic-stimulation of astrocytes enhances astrocyteinduced synapse formation
To test whether cholinergic stimulation of astrocytes affects synapse formation we treated astrocyte cultures for 24 hours with increasing concentrations of the acetylcholine receptor (AChR) agonist carbachol (0.01, 0.1, or 1 mM). At treatment end, carbachol was removed and fresh treatment-free medium was added. Primary hippocampal neurons, grown separately on glass coverslips, were then inverted over the pre-treated astrocytes for an additional 24 hours. In this co-culture system neurons were neither directly exposed to carbachol nor were they in contact with the astrocytes, limiting any effect on neuronal synapse formation to the modulation of astrocyte-secreted factors.
To quantify the effect of astrocyte cholinergic stimulation on the localization of synaptic proteins into excitatory synaptic structures, we generated three-dimensional surface renderings of the clustered synaptophysin and PSD95 protein puncta from deconvolved confocal images and the number of pre and postsynaptic specializations were automatically counted for each neuronal eld. Astrocytes pre-treated with carbachol increased the number of individual presynaptic synaptophysin (Fig. 1B) and postsynaptic PSD-95 (Fig. 1C) puncta in neurons in a dose-dependent manner. An almost two-fold increase in synaptophysin puncta and a 2.2-fold increase in PSD-95 puncta were observed in neurons after co-culture with astrocytes pre-treated with 0.1 mM carbachol, when compared to control astrocytes. Astrocyte pretreatment with 1mM carbachol induced a 1.6-and 2.3-fold increase in the number of individual synaptophysin and PSD-95 puncta, respectively (Fig. 1B, C). Because synapses are structures consisting of aligned pre and postsynaptic specializations, we used the number of overlapping synaptophysin and PSD-95 puncta as a measure of the number of structural synapses in each treatment group. Stimulation of astrocytes with carbachol induced the formation of a greater number of synapses than those induced by untreated astrocytes. Pre-treatment of astrocytes with 1 mM carbachol caused a 3.2-fold increase in the number of aligned pre and postsynaptic puncta (structural synapses) above the number of synapses observed when neurons were co-cultured with unstimulated astrocytes (Fig. 1D). Representative deconvolved images of hippocampal neurons after co-culture for 24 hours with astrocytes pretreated with 1 mM carbachol or control astrocytes are shown in Fig. 1A with synaptophysin shown in green, PSD-95 shown in red, and the overlapping puncta shown in yellow. Together, these results indicate that astrocyteinduced hippocampal synaptic structures are greatly increased when neurons are co-cultured in the presence of astrocytes previously stimulated with the cholinergic agonist carbachol.
To determine the cholinergic receptor subtype(s) involved in carbachol-stimulated astrocyte-induced synaptogenesis, astrocyte cultures were pre-treated with carbachol (1 mM) in the presence of muscarinic and nicotinic AChR antagonists for 24 hours before co-culturing the astrocytes with hippocampal neurons. As rat cortical astrocytes do not express M1 or M4 muscarinic receptors [39], we investigated the role of M2 and M3 muscarinic receptors. Inhibiting the M2 muscarinic AChR with gallamine (10 µM) had no effect on carbachol-induced increase in synapse number, while the M3 muscarinic antagonist 4-DAMP reduced the increase in the number of structural synapses observed (Fig. 2). The effect of carbacholtreated astrocytes on synapse formation was also inhibited by the nicotinic antagonist mecamylamine (10 µM) (Fig. 2). These data suggest that cholinergic signaling through both astrocytic M3 muscarinic and nicotinic receptors were responsible for the observed increase in hippocampal neuron synapse number.
Using Western blotting techniques, we examined total expression of the pre-and postsynaptic proteins synaptophysin and PSD-95. Neurons co-cultured with 1 mM carbachol pre-treated astrocytes showed a robust increase in synaptophysin and PSD-95 protein levels relative to control (Fig. 3A, B). These ndings suggest that the increase in puncta size and intensity may be due, at least in part, to increased expression of neuronal synaptophysin and PSD95 expression in neurons.
We also immunolabeled neurons with two other markers of synaptic structures, the presynaptic protein synaptotagmin, and the postsynaptic NMDA receptor NR2B subunit. We observed a 2.3-fold increase in synaptotagmin puncta (Fig. 4A), a 2.8-fold increase in the NR2B puncta (Fig. 4B), and a 4.6-fold increase in synaptotagmin/NR2B overlapping puncta (Fig. 4C) in neurons co-cultured with carbachol-pre-treated astrocytes, compared to untreated astrocytes. Together this data indicates that cholinergic stimulation of astrocytes enhances synapse formation in co-cultured hippocampal neurons.

Cholinergic stimulation of astrocytes increases the number of functional synapses
We measured the frequency of spontaneous miniature excitatory postsynaptic currents (mEPSCs) [50] in neurons co-cultured with astrocytes pre-treated with carbachol compared to neurons co-cultured with control astrocytes and found a signi cantly greater frequency of mEPSCs, as measured by the mEPSCs inter-event interval (Fig. 5A), indicating a greater number of functional synapses in neurons co-cultured with carbachol-stimulated astrocytes. A slight, but statistically signi cant increase in the amplitude was also observed (Fig. 5B), suggesting a potential increase in the number of postsynaptic receptors in neurons co-cultured with carbachol astrocytes. Additionally, although there was no difference in the mean time to half decay between treatment groups, neurons co-cultured with control astrocytes relative to carbachol pre-treated astrocytes, showed faster receptor kinetics in a proportion of the synapses, but slower kinetics in the remainder (Fig. 5C). This observation suggests potential differences in the composition of the excitatory receptors at the synapses in each treatment group, though further studies are needed to fully understand this observation.
Overall, these results indicate that astrocytes pre-treated with the AChR agonist carbachol enhance the formation of functional synapses.

Astrocyte-released TSP1 is involved in the increase in synaptic structures after cholinergic stimulation of astrocytes
We investigated whether astrocytes induce protracted expression and secretion of TSP1 after carbachol pre-treatment. Astrocytes were treated with carbachol for 24 hours followed by treatment washout and the subsequent replacement with treatment-free culture medium. We collected astrocyte medium and cells lysate at the end of the 24h incubations (t0), and 6h and 24h after treatment wash-out; levels of secreted and total cellular TSP1 were quanti ed by Western blot.
Medium from astrocytes treated with carbachol showed a 2.4-fold increase in TSP1 compared to medium from untreated astrocytes at the end of the 24h incubation with carbachol (Fig. 6A, B) con rming our previous ndings [51]. Six hours after replacing the treatment medium with carbachol-free medium, TSP1 was detectable, though at lower levels, suggesting continuous basal release of TSP1. Importantly, 6h after treatment wash-out, but not 24h after, the amount of TSP1 released by carbachol-treated astrocytes was still greater than that released by untreated astrocytes (Fig. 6A, B). These post-treatment time points are within the time-frame of astrocyte-neuron cocultures (see Figs. 1-3). We used immunocytochemistry and confocal imaging to measure the surface expression of TSP1 in non-permeabilized cells. Twentyfour hours after carbachol treatments, astrocytes showed a 3.35-fold increase in TSP1 surface uorescence relative to untreated astrocytes (Fig. 6C, D), while total TSP1 levels, measured by Western blot in the astrocyte cell lysate, were not altered (Fig. 6E, F). Taken together, these results indicate that stimulation of the cholinergic pathway in astrocytes enhances the secretion of TSP1, an effect that lasts several hours after the end of carbachol treatments and could therefore be in part responsible for carbachol-treated astrocyte-induced synaptogenesis.
Hippocampal neurons were then directly treated with TSP1 (5 or 10 µg/mL) for 24-hours. The number of synaptophysin and PSD-95 puncta was not increased by treatment of neurons with TSP1 (Fig. 7A, B). However, the number of co-localized synaptophysin/PSD-95 puncta increased up to 2-fold relative to untreated neurons (Fig. 7C) indicating an increase in the formation of apposed pre-and postsynaptic structures.
To test whether TSP1-activated signaling in neurons is necessary for the enhanced synaptogenic effect of carbachol pre-treated astrocytes, we incubated astrocyte-neuron co-cultures with gabapentin, an inhibitor of the α2δ1 subunit of neuronal voltage-gated calcium channels that mediates the synaptogenic effects of TSPs on retinal ganglion cells [18,52]. The addition of gabapentin (15 and 30 µM) to the coculture system containing both neurons and carbachol pre-treated astrocytes blocked the upregulation of synaptic puncta induced by carbachol-treated astrocytes (Fig. 7C).
Together, these results indicate that TSP1, whose secretion by astrocytes is increased by carbachol treatments, is involved in the increased synaptogenesis observed after co-culture of neurons with carbachol-stimulated astrocytes.
In this study we characterized the effects of cholinergic stimulation in astrocytes on the formation of active synapses. We report that cholinergic stimulation in astrocyte potentiates the formation of synapses (Figs. [1][2][3][4][5] in part by enhancing astrocyte secretion of the synaptogenic factor TSP1, which persists after the ACh receptor agonist has been removed (Fig. 6), and that the inhibition of TSP signaling by gabapentin abolishes the synaptogenic effects of carbachol-stimulated astrocytes (Fig. 7).
Several studies have focused on the role of ACh in brain development independent of its function as neurotransmitter; nicotinic and muscarinic ACh receptors and acetylcholinesterase are expressed in the brain before synapses are formed [43][44][45]. Neuronal acetylcholine release in uences neuronal migration, differentiation, synapse formation and function [32,[38][39][40][41][42], as well as oligodendrocyte development and function [55]. However, this is the rst study to report that cholinergic stimulation in astrocytes directly modulates synaptogenesis in neurons.
We found that synapse formation induced by carbachol-stimulated astrocytes is inhibited by the M3-AChR antagonist 4-DAMP (Fig. 2). We previously characterized the robust Gq-coupled signaling induced by cholinergic stimulation of muscarinic M3-AChR in astrocyte cultures [60-65]. We also identi ed the promoting effect of astrocyte cholinergic stimulation on neurite outgrowth in co-cultured neurons and in organotypic hippocampal cultures during early phases of neuronal maturation; these effects were mediated in part by an increased release by carbachol-stimulated astrocytes of extracellular matrix proteins laminin and bronectin and by the modulation of the plasminogen activator extracellular matrix protease system [1,2,66]. Signaling through GPCRs in astrocytes is typically long-lasting because GPCRs have a slower desensitization time course than ionotropic receptors [67].
Somewhat surprisingly, the nicotinic receptor antagonist mecamylamine also attenuated the carbachol effect (Fig. 2). While ionotropic receptor signaling in neurons is usually localized and short-lived, the activation of astrocyte-expressed α7 ionotropic nAChRs has been shown to increase permeability to Ca 2+ , which in turn causes a prolonged release of Ca 2+ from intracellular stores [23] and plays a role in synapse function by recruiting AMPA receptors to the postsynaptic surface [68].
We also found that cholinergic stimulation of astrocytes increased the frequency of mEPSCs (Fig. 5), indicating an increase in the number of functional synapses [50]. Importantly, we found that the number of puncta containing the NR2B-subunit of the NMDA receptor increased when neurons where co-cultured with ACh pre-treated astrocytes (Fig. 4). NMDARs containing this subunit are expressed during early development [69], and have been shown to be required for synaptic plasticity and synaptogenesis [70].
In a previous proteomic study we have reported that carbachol increased the release of TSP1 from astrocytes [51]; in this study, we have shown that TSP1, in the absence of astrocytes, increased synapse numbers (i.e., the colocalization of presynaptic and postsynaptic puncta), but did not increase the number of pre-or postsynaptic puncta (Fig. 7), suggesting that TSP1 may facilitate the alignment of pre and postsynaptic puncta into structural synapses. Indeed, it has been proposed that the TSPs, acting through the postsynaptic α2δ1 receptor, may exert their synaptogenic effects through the recruitment of cell adhesion and scaffolding proteins, promoting the assembly of apposing pre-and postsynaptic structural synapses [14,17,18,52]. Consistent with these ndings, we report that inhibition of α2δ1 signaling by gabapentin prevents carbachol-treated-astrocyte-induced increased synapses (Fig. 7). TSP1 is expressed [59] and released by astrocytes during synaptogenesis [3] and exogenous addition of TSP1 and TSP2 to retinal ganglion cells increased the number of synapses which are presynaptically active, but postsynaptically silent [3]. TSP1 also increases the speed of synapse formation in hippocampal neurons [71]. Additionally, ATP activation of astrocytes results in increased expression of TSP1 in a time and concentration-dependent manner [72].
In conclusion our study uncovers a new mechanism of astrocyte-neuron communication by which a signal from neurons, ACh, through the stimulation of a robust signaling pathway in astrocytes [64], leads to the increased release by these cells of synaptogenic proteins and increased formation of functional synapses. Figure 8 is a schematic representation of our ndings. This study expands our understanding of the signals that regulate the release of synaptogenic proteins from astrocytes and contributes a novel mechanism by which the cholinergic system is involved in brain development.   The inhibition of M3 muscarinic and nicotinic receptors in astrocytes reduces carbachol-stimulated astrocyte-induced increase in neuronal synaptic structure formation. Astrocytes were incubated with 1 mM carbachol in the presence or absence of nicotinic receptor antagonist mecamylamine, M2-muscarinic receptor antagonist gallamine, and M3-muscarinic antagonist 4-DAMP (10 µM each) for 24 hours. After treatment washout, hippocampal neurons were co-cultured with astrocytes for 24 hours and synaptic structures, i.e., overlapping synaptophysin and PSD-95 puncta, were quanti ed. The distribution of the data is shown. The bold, dashed line represents the median; the upper and lower dotted lines represent the 75 th and 25 th percentile, respectively; n= 25-33 neurons from 3 independent experiments. Statistical comparison of the means was carried out by one-way ANOVA followed by Bonferroni's Multiple Comparison test. ***: p < 0.001 vs. control; ###: p < 0.001 vs. carbachol.   Neurons co-cultured with carbachol pre-treated astrocytes have a higher frequency of mEPSC events.
Whole cell patch clamp techniques were used to record spontaneous, miniature excitatory postsynaptic currents (mEPSCs) from dissociated hippocampal neurons after 24 hours of co-culture with carbachol pre-treated or untreated astrocytes. Cumulative distributions of arrayed spontaneous mEPSC inter-event interval (IEI) (A), amplitude (B) and time to decay (C) are shown on the left. Inset A: Example traces of neurons co-cultured with control (top) and carbachol pre-treated astrocytes (bottom) (Scale bar: 20 pA, 1 sec). Graphs of the mean IEI (A), amplitude (B), and time to half decay (C) from each array are shown on the right. Control array n = 9, carb array n= 7 neurons from 3 independent experiments. Cumulative distributions of IEI, amplitude, and time to half decay differ signi cantly (all < 0.0001) as measured by the Kolmogorov-Smirnov Two Sample Test (IEI: K-S D = 0.0714; amplitude: K-S D = 0.1048; decay time: K-S D = 0.1215). Student's t-test was performed to compare the means of the arrayed data per treatment group. ***: p < 0.0001 vs control.

Figure 6
Cholinergic stimulation of astrocytes induces TSP1 release from astrocytes with no effect on TSP1 protein expression.