Abstract
Dendritic pruning and loss of synaptic contacts are early events in many neurodegenerative diseases. These effects are dynamic and seem to differ mechanistically from the cell death process. Cannabinoids modulate synaptic activity and afford protection in some neurotoxicity models. We investigated the effects of cannabinoids on activity-induced changes in the number of synapses between rat hippocampal neurons in culture. Morphology and synapses were visualized by confocal imaging of neurons expressing DsRed2 and postsynaptic density protein 95 (PSD95) fused to enhanced green fluorescent protein (GFP). Reducing the extracellular Mg2+ concentration to 0.1 mM for 4 h induced intense synaptic activity, which decreased the number of PSD95-GFP puncta by 45 ± 13%. Synapse loss was an early event, required activation of N-methyl-d-aspartate receptors, and was mediated by the ubiquitin-proteasome pathway. The cannabinoid receptor full agonist WIN55,212-2 [(R)-(+)-[2,3-dihydro-5-methyl-3-[(4-morpholinyl)-methyl] pyrrolo-[1,2,3-de]-1,4-benzoxazin-6-yl](1-napthalenyl)-methanone monomethanesulfonate] (EC50 = 2.5 ± 0.5 nM) and the partial agonist Δ9-tetrahydrocannabinol (THC; EC50 = 9 ± 3 nM) inhibited PSD loss in a manner reversed by the CB1 receptor antagonist rimonabant [N-piperidino-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-3-pyrazole-carboxamide]. The protection was mimicked by inhibition of presynaptic Ca2+ channels, and WIN55,212-2 did not prevent PSD loss elicited by direct application of glutamate, suggesting a presynaptic mechanism. Prolonged exposure to WIN55,212-2, but not THC, desensitized the protective effect. Treating cells that had undergone PSD loss with WIN55,212-2 reversed the loss and enabled recovery of a full compliment of synapses. The modulation of synaptic number by acute and prolonged exposure to cannabinoids may account for some of the effects of these drugs on the plasticity, survival, and function of neural networks.
Δ9-Tetrahydrocannabinol (THC), the principal psychoactive ingredient in marijuana, produces euphoria and relaxation and impairs motor coordination, time sense, and short-term memory (Ameri, 1999). Clinically useful attributes of the cannabinoids (CBs) include their ability to produce analgesia, reduce chemotherapy-induced emesis, stimulate appetite, and attenuate seizures and spasticity (Croxford, 2003). The central actions of CBs are mediated by activation of CB1 receptors located primarily on presynaptic nerve terminals, where they interact with heterotrimeric G proteins (Gi/Go) to inhibit neurotransmitter release via inhibition of voltage-gated Ca2+ channels and activation of K+ channels (Howlett et al., 2004). Endocannabinoids (eCBs) are arachidonic acid derivatives produced in response to postsynaptic stimulation; they diffuse in a retrograde fashion across the synapse to act on presynaptic CB1 receptors to exert a similar inhibition of neurotransmitter release (Diana and Marty, 2004).
Activation of CB1 receptors by either exogenous or endogenous ligands reduces the strength of the affected synapse. The eCB system mediates a long-term depression of projections from prefrontal cortex to nucleus accumbens neurons (Robbe et al., 2002) and long-lasting autoinhibition of neocortical GABAergic interneurons (Bacci et al., 2004). In hippocampus, CBs inhibit GABA release from a subset of interneurons and inhibit glutamate release from principal neurons (Katona et al., 1999; Domenici et al., 2006). Thus, cannabinoids produce both rapid and long-term changes in synaptic transmission.
Cannabinoids protect hippocampal neurons from excitotoxicity (Shen and Thayer, 1998b) and from oxygen and glucose deprivation (Nagayama et al., 1999). Cannabinoids also provide protection in a variety of models of acute neurodegeneration in vivo (Nagayama et al., 1999; Panikashvili et al., 2001; van der Stelt et al., 2001). These effects have mostly been ascribed to inhibition of glutamatergic synaptic transmission (Shen et al., 1996; Gilbert et al., 2007), consistent with localization of CB1 receptors on glutamatergic nerve terminals (Domenici et al., 2006). Indeed, the eCB system seems to serve as an on-demand mechanism to reduce excitatory synaptic activity and thus protect from overstimulation (Marsicano et al., 2003; Alger, 2004).
Changes in dendritic morphology accompany changes in synaptic strength (Harms and Dunaevsky, 2006). Indeed, stimuli that induce long-term potentiation of synaptic transmission alter the shape of synaptic spines (Carlisle and Kennedy, 2005) and increase the number of synapses (Ma et al., 1999). Changes in dendritic morphology also appear in the early stages of neurodegenerative disorders (Sa et al., 2004). Dendritic spines undergo some of the earliest structural changes during excitotoxic injury to neurons (Olney et al., 1979), and changes in spine density are well documented in a variety of epilepsy models (Fiala et al., 2002). Synaptic loss is not necessarily a step on the path toward death but instead may be a mechanism that enables the cell to adapt to excessive excitatory input (Finkbeiner et al., 2006; Waataja et al., 2008). If synaptic loss is a coping mechanism, then pharmacologic strategies to preserve network function may be important compliments to drugs that improve survival.
The postsynaptic density protein (PSD95) is a scaffolding protein that anchors receptors and downstream signaling molecules to the postsynaptic density (Kim and Sheng, 2004) and is widely used as a marker for synaptic sites (Okabe et al., 1999). PSD95 is removed from synapses in response to stimuli that induce long-term depression of synaptic transmission (Colledge et al., 2003) and during epileptic activity (Zha et al., 2005).
Here, we studied changes in the number of synapses between hippocampal neurons identified by fluorescent puncta produced by expression of PSD95 fused to enhanced green fluorescent protein (GFP). We show that intense excitatory synaptic activity reduced the number of synapses between hippocampal neurons in culture. We tested the hypothesis that cannabinoids prevent and reverse synaptic loss via presynaptic inhibition of glutamate release.
Materials and Methods
Materials. Materials were obtained from the following sources. The GFP-PSD95 expression vector was kindly provided by Donald B. Arnold; the expression vector for DsRed2 (pDsRed2-N1) was obtained from Clontech (Mountain View, CA); rimonabant (SR141716) and THC were from the National Institute on Drug Abuse drug supply system (Bethesda, MD); fura-2 acetoxymethyl ester, Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum, and horse serum were from Invitrogen (Carlsbad, CA); WIN55,212-2, l-glutamate, MK801, MG132, and all other reagents were obtained from Sigma-Aldrich (St. Louis, MO).
Cell Culture. Rat hippocampal neurons were grown in primary culture as described previously (Shen and Thayer, 1998a) with minor modifications. Fetuses were removed on embryonic day 17 from maternal rats, anesthetized with CO2, and sacrificed by decapitation. Hippocampi were dissected and placed in Ca2+- and Mg2+-free HEPES-buffered Hanks' salt solution (HHSS), pH 7.45. HHSS was composed of the following: 20 mM HEPES, 137 mM NaCl, 1.3 mM CaCl2, 0.4 mM MgSO4, 0.5 mM MgCl2, 5.0 mM KCl, 0.4 mM KH2PO4, 0.6 mM Na2HPO4, 3.0 mM NaHCO3, and 5.6 mM glucose. Cells were dissociated by trituration through a 5-ml pipette and a flame-narrowed Pasteur pipette. Cells were pelleted and resuspended in DMEM without glutamine, supplemented with 10% fetal bovine serum and penicillin/streptomycin (100 U/ml and 100 μg/ml, respectively). Dissociated cells were then plated at a density of 15,000 to 20,000 cells/dish onto a 25-mm-round cover glass (no. 1) glued to cover a 19-mm-diameter opening drilled through the bottom of a 35-mm Petri dish. The coverslip was coated with poly-d-lysine (0.1 mg/ml) and washed with H2O. Neurons were grown in a humidified atmosphere of 10% CO2 and 90% air (pH 7.4) at 37°C, and fed at days 1 and 6 by exchange of 75% of the media with DMEM, supplemented with 10% horse serum and penicillin/streptomycin. Cells used in these experiments were cultured without mitotic inhibitors for a minimum of 12 days.
[Ca2+]i Imaging. Cells were loaded with 3 μM fura-2 acetoxymethyl ester at 37°C for 45 min in HHSS containing 0.5% BSA. Coverslips with cells were then transferred to a recording chamber, placed on the stage of an Olympus IX71 microscope (Melville, NY), and viewed through a 40× objective. Excitation wavelength was selected with a galvanometer driven monochromator (8-nm slit width) coupled to a 75-W xenon arc lamp (Optoscan; Cairn Research Limited, Faversham, Kent, UK). [Ca2+]i was monitored in a field of cells using sequential excitation of fura-2 at 340 and 380 nm to enable the calculation of ratio-based [Ca2+]i every 1 s. Fluorescence images, 510/40 nm, were projected onto a cooled charge-coupled device camera (Cascade 512B; Roper Scientific, Trenton, NJ) controlled by MetaFluor software (Molecular Devices, Sunnyvale, CA). Changes in fura-2 fluorescence were converted to [Ca2+]i using the formula [Ca2+]i = Kdβ(R - Rmin)/(Rmax - R), where R is the 340-/380-nm fluorescent intensity ratio. The dissociation constant (Kd) used for fura-2 was 225 nM, and β was the ratio of emitted fluorescence at 380 nm in the absence and presence of calcium. Rmin, Rmax, and β were determined in ionomycin-permeabilized cells in calcium-free (1 mM EGTA) and 5 mM Ca2+ buffers. Values of Rmin, Rmax, and β were 0.443, 6.481, and 8.294, respectively.
Transfection. Rat hippocampal neurons were transfected between 10 and 13 days in vitro using a modification of a protocol described previously (Waataja et al., 2008). In brief, hippocampal cultures were incubated for at least 20 min in DMEM supplemented with 1 mM kynurenic acid, 10 mM MgCl2, and 5 mM HEPES, to reduce neurotoxicity. A DNA/calcium phosphate precipitate containing 1 μg of plasmid DNA/well was prepared, allowed to form for 30 min at room temperature and added to the culture. After a 90-min incubation, cells were washed once with DMEM supplemented with MgCl2 and HEPES and then returned to conditioned media, saved at the beginning of the procedure. The transfection efficiency ranged from 10 to 12%.
Confocal Imaging. Forty-eight hours after transfection, neurons were transferred to the stage of a confocal microscope (Olympus Fluoview 300) and viewed through a 60× oil-immersion objective (numerical aperture, 1.40). For experiments in which the same neurons were imaged before and after a 4-h interval, the locations of individual cells were recorded using micrometers attached to the stage of the microscope. Multiple optical sections spanning 8 μmin the z-dimension were collected (1-μm steps), and these optical sections were combined through the z-axis into a compressed z-stack. GFP was excited at 488 nm with an argon ion laser and emission collected at 530 nm (10-nm band pass). The excitation and emission wavelengths for DsRed2 were 543 and >605 nm, respectively.
Image Processing. To count and label PSD95-GFP puncta an automated algorithm was created using MetaMorph 6.2 image-processing software described previously (Waataja et al., 2008). In brief, maximal z-projection images were created from the DsRed2 and GFP image stacks. Next, a threshold set 1 S.D. above the image mean was applied to the DsRed2 image. This created a 1-bit image that was used as a mask via a logical AND function with the GFP maximal z-projection. A top-hat filter (80 pixels) was applied to the masked PSD95-GFP image. A threshold set 1.5 S.D. above the mean intensity inside the mask was then applied to the contrast enhanced image. Structures between 8 and 80 pixels (approximately 0.37–3.12 μm in diameter) were counted as PSDs. The structures were then dilated and superimposed on the DsRed2 maximal z-projection for visualization. PSD counts were presented as mean ± S.E.M., where n is the number of cells, each from a separate cover glass over multiple cultures. We used Student's two-tailed t test for single or ANOVA with Bonferroni's post-test for multiple statistical comparisons.
Results
Visualizing Changes in Synaptic Sites. We developed an imaging-based method to quantify the number of synaptic sites on hippocampal neurons in culture based on the detection of clusters of the scaffolding protein PSD95 fused to enhanced green fluorescent protein (PSD95-GFP) (Waataja et al., 2008). In Fig. 1, we show hippocampal neurons in culture that were transfected with expression plasmids for PSD95-GFP and DsRed2. Confocal imaging of transfected neurons was used to identify postsynaptic sites, as indicated by PSD95-GFP puncta (Fig. 1A) and to determine cell morphology, as revealed by red fluorescence (Fig. 1B). We quantified postsynaptic sites by using an image-processing algorithm that applied the DsRed2 image as a binary mask to the PSD95-GFP image, then identified puncta by locating intensity peaks of the appropriate size (approximately 0.37–3.12 μm in diameter). Figure 1C shows a processed image with counted puncta enlarged and superimposed on the DsRed2 image. PSD95-GFP puncta were localized on dendritic shafts and spines. Analysis of DsRed images of cells with spines, which were identified by an observer using morphological criteria (mushroom shape with length between 1 and 2 μm, base width between 0.4 and 0.7 μm, and body width between 0.8 and 1.6 μm) before running the image-processing algorithm to identify puncta revealed that 87 ± 5% of spines (n = 8 cells) also contained PSD95-GFP puncta. We have shown previously that fluorescent puncta represent functional synapses as indicated by localized increases in [Ca2+]i induced by NMDA, their close apposition to presynaptic sites of vesicular release labeled with FM4-64 and colocalization with NR2A and 2B immunoreactivity (Waataja et al., 2008). These results are consistent with previous studies showing that a PSD95-GFP fusion protein expressed in hippocampal pyramidal neurons was clustered at postsynaptic sites (Okabe et al., 1999).
Aberrant Excitatory Synaptic Activity Induces the Loss of Synaptic Sites. Reducing the extracellular Mg2+ concentration ([Mg2+]o) in the medium bathing hippocampal cultures elicits an intense pattern of excitatory electrical activity that produces repetitive increases in the intracellular Ca2+ concentration ([Ca2+]i spikes) (McLeod et al., 1998). When sustained, this aberrant pattern of neurotransmission results in synaptically mediated neuronal death. We imaged changes in [Ca2+]i with fura-2 to verify that exposure to 0.1 mM [Mg2+]o induced repetitive [Ca2+]i spiking in the hippocampal cultures studied here (Fig. 1, D–F). [Ca2+]i spiking was driven by glutamatergic synaptic transmission because it was significantly reduced (68 ± 4% inhibition, n = 41) by treatment with the NMDA receptor antagonist MK801 (10 μM) (Fig. 1D). Exposure to 0.1 mM [Mg2+]o for 4 h decreased the number of PSD95-GFP puncta by 45 ± 13% (n = 8) (Fig. 1, G and H). Analysis of puncta localized to spines revealed that PSD loss accompanied spine loss. After application of 0.1 mM [Mg2+]o, there was a 78 ± 5% decrease in PSDs on spines and an 80 ± 4% loss of spines (n = 8 cells). Pretreatment with MK801 (10 μM) for 30 min prevented the 0.1 mM [Mg2+]o-induced loss of PSDs (Fig. 1I), indicating that the 0.1 mM [Mg2+]o-induced loss of PSD95-GFP is dependent on the NMDA receptor. NMDA receptor activation can induce PSD95 degradation via the ubiquitin-proteasome pathway. We found that intense excitatory synaptic activity also targets PSD95-GFP to the proteasome as indicated by protection from 0.1 mM [Mg2+]o-induced synapse loss by the proteasome inhibitor MG132 (50 μM) (Fig. 1I). These results suggest that intense excitatory synaptic activity causes the loss of synaptic connections.
We next determined the time course for changes in the number of PSD95-GFP puncta and cell survival. Survival was defined as the retention of cytoplasmic DsRed2 protein. As shown in Fig. 2A, cell survival was stable for 24 h under control conditions, and the number of PSD95-GFP puncta increased initially. The precise mechanism for this increase is not known, although the data in Fig. 2C indicated that it develops in a graded fashion. It was triggered by the media exchange as experiments in which cells remain in growth media showed a significantly smaller increase (6 ± 5% versus 49 ± 10%). New synapse formation required protein synthesis because it was inhibited by cycloheximide (10 μM) (-2 ± 13%) but did not require synaptic activity as indicated by a lack of effect of tetrodotoxin (40 ± 18%). Exposure to 0.1 mM [Mg2+]o caused a significant loss of PSD95-GFP puncta by 1 h (-30 ± 4%, p < 0.05, n = 8), and the magnitude of the loss increased to -39 ± 15% by 8 h. In contrast, no cells died after 1 h exposure to 0.1 mM [Mg2+]o, and the number of dead cells slowly increased to 29 ± 18% after 8 h in 0.1 mM [Mg2+]o and increased further to 71 ± 18% death by 24 h (n = 7). Thus, 0.1 mM [Mg2+]o-induced synaptic activity produces a time-dependent loss of PSDs that precedes overt cell death.
Cannabinoid Receptor Agonists Prevent 0.1 mM [Mg2+]o-Induced PSD Loss. We have previously shown that cannabinoid receptor agonists inhibit both the frequency of [Ca2+]i spiking induced by reducing [Mg2+]o and glutamatergic EPSCs elicited by stimulation of the presynaptic neuron (Shen et al., 1996). We tested the effect of cannabinoid agonists on 0.1 mM [Mg2+]o-induced PSD loss. As shown in Fig. 3, the cannabinoid receptor full agonist WIN55,212-2 (100 nM) reduced PSD loss by 81 ± 25% (n = 7) (Fig. 3, A, B, and E). The cannabinoid receptor partial agonist THC, at the same concentration, reduced PSD loss by 66 ± 11% (Fig. 3, C–E). The selective CB1 receptor antagonist rimonabant completely blocked the actions of both drugs (Fig. 3E). Rimonabant alone did not affect 0.1 mM [Mg2+]o-induced PSD loss (-46 ± 16%, p < 0.01, n = 4). The cannabimimetic drugs produced a concentration-dependent protection from the PSD loss induced by 0.1 mM [Mg2+]o (Fig. 4). THC reduced the loss of PSD95-GFP puncta with an EC50 of 9 ± 3 nM. WIN55,212-2 was more potent than THC with an EC50 of 2.5 ± 0.5 nM.
WIN55,212-2 Acts Presynaptically. Cannabinoid receptor agonists inhibit glutamatergic synaptic transmission by acting presynaptically to inhibit the release of glutamate (Shen et al., 1996). To test whether the effects of cannabimimetic drugs on synaptic sites were mediated via actions at pre- or postsynaptic targets, we examined the effects of WIN55,212-2 on changes in PSDs evoked by treatment with glutamate. Application of glutamate (4 h, 100 μM) to cultured hippocampal neurons reduced the number of PSD95-GFP puncta by 63 ± 14% (Fig. 5A). In contrast to the effects of cannabimimetics on the synaptically mediated PSD loss induced by 0.1 mM [Mg2+]o, WIN55,212-2 did not prevent the PSD loss elicited by the direct application of glutamate (Fig. 5A). Thus, cannabinoid receptor agonists act presynaptically to inhibit 0.1 mM [Mg2+]o-induced PSD loss.
A presynaptic site of action for the CB1 agonists would be consistent with an inhibition of N- and P/Q-type Ca2+ channels (Shen and Thayer, 1998a). We tested this hypothesis by treating hippocampal cultures for 5 min with 1 μM ω-conotoxin GVIA, a selective blocker of N-type Ca2+ channels, before application of 0.1 mM [Mg2+]o. ω-Conotoxin GVIA completely protected hippocampal neurons from 0.1 mM [Mg2+]o induced PSD loss (n = 10) (Fig. 5B). Pretreatment with 1 μM ω-agatoxin IVA, a selective blocker of P/Q-type Ca2+ channels, for 20 min also significantly reduced the loss of PSDs. Neither channel blocker significantly affected the number of PSDs under control conditions (Fig. 5B). Thus, blocking presynaptic Ca2+ channels required for glutamate release prevents the loss of postsynaptic sites induced by intense excitatory synaptic activity.
Prolonged Exposure to WIN55,212-2 Diminishes the Protection of PSDs by Cannabinoids. Prolonged exposure to WIN55,212-2 or THC desensitized CB1 receptor-mediated inhibition of synaptic transmission (Kouznetsova et al., 2002; Lundberg et al., 2005). Here, we investigated whether prolonged exposure to cannabinoid receptor agonists diminished the protective effect of cannabimimetic drugs on 0.1 mM [Mg2+]o-induced loss of PSDs. When cultures were pretreated with 1 μM WIN55,212-2 for 24 h, the protection afforded by cannabimimetic drugs was reduced. In cultures pretreated with WIN55,212-2, subsequent treatment with 0.1 mM [Mg2+]o in the continued presence of 100 nM WIN55,212-2 resulted in a 49 ± 14% decrease in PSDs. Likewise, in cells pretreated with WIN55,212-2 (1 μM) for 24 h, subsequent application of 0.1 mM [Mg2+]o and 100 nM THC resulted in a 18 ± 8% loss of PSDs (Fig. 6), which was not significantly different from the loss induced by 0.1 mM [Mg2+]o in the absence of drug. Prolonged exposure to a maximally effective concentration of THC produces less functional desensitization of CB1-mediated inhibition of excitatory synaptic transmission than that produced by a maximally effective concentration of WIN55,212-2 (Lundberg et al., 2005). Thus, we examined the possibility that the reduced 0.1 mM [Mg2+]o-induced PSD loss produced by THC might be more sustained than that of WIN55,212-2. In cultures pretreated with 1 μM THC for 24 h, subsequent treatment with 0.1 mM [Mg2+]o in the continued presence of 100 nM THC resulted in a 58 ± 29% increase in PSDs, significantly different from the loss seen in the absence of drug (Fig. 6). Thus, even after 24-h pretreatment, THC significantly attenuated 0.1 mM [Mg2+]o-induced PSD loss.
Cannabinoids Reverse Synaptically Driven Synapse Loss. If PSD loss proceeds via a path that is independent from that leading to cell death, then synapse loss might be reversible. We next examined whether WIN55,212-2 could reverse synaptically driven PSD loss after exposure to 0.1 mM [Mg2+]o. Hippocampal neurons expressing PSD95-GFP and DsRed2 were exposed to 0.1 mM [Mg2+]o for 1 h, which produced a 39 ± 6% loss of PSDs (Fig. 7, A and B). WIN55,212-2 (100 nM) was added at this time, and 3 and 23 h after application of the drug in the continued presence of 0.1 mM [Mg2+]o, the cells were imaged again (4- and 24-h elapsed time). After 3 h in the presence of WIN55,212-2, neurons recovered their PSDs (10 ± 10% increase relative to before 0.1 mM [Mg2+]o, p < 0.001, n = 8) (Fig. 7B). This recovery persisted for 23 h after application of WIN55,212-2. Synapse loss induced by 1-h treatment with 0.1 mM [Mg2+]o was also reversed by returning [Mg2+]o to 0.9 mM (data not shown). It is interesting to note that hippocampal neurons that were exposed to 0.1 mM [Mg2+]o for 4 h before addition of WIN55,212-2 did not recover their PSDs (35 ± 11% decrease relative to before 0.1 mM [Mg2+]o, n = 4) (Fig. 7C). Therefore, there is a window of at least 1 h but less than 4 h during which activation of CB1R can reverse synaptic loss resulting from epileptic patterns of synaptic activity.
Discussion
We used an in vitro model to quantify changes in the number of individual synapses in a synaptic network over time. The intense synaptic activity evoked by reducing [Mg2+]o to 0.1 mM decreased the number of PSD95-GFP puncta by 45 ± 13%. This observation is consistent with previous reports showing a decrease in spine density in hippocampal neurons exposed to glutamatergic agonists or epileptiform activity (Müller et al., 1993; Colledge et al., 2003; Zha et al., 2005). Cannabinoids acted potently on presynaptic CB1 receptors to prevent or even reverse PSD loss that resulted from intense synaptic activation of NMDA receptors. Thus, presynaptic inhibition attenuated synaptic transmission sufficient to prevent long-term changes in the number of synapses. Cannabinoid modulation of synapse loss desensitized, raising interesting questions about the changing role of the endocannabinoid system in neuroprotection after chronic exposure to drugs. PSD loss is an early event in neurodegenerative disease; its reversal by cannabinoids has implications for protection of network function, modulation of synaptic plasticity, and regulation of neuronal survival.
We used PSD95, a protein that serves as a scaffold for assembling signaling molecules at excitatory synapses, as a marker for synapses between hippocampal neurons in culture. Glutamate produces a local [Ca2+]i increase at PSD95-GFP puncta, these puncta are in close apposition to presynaptic sites of vesicular release, and they colocalize with NR2A and 2B immunoreactivity (Waataja et al., 2008). Thus, PSD95-GFP puncta represented functional postsynaptic sites. The formation and loss of synaptic puncta were highly dynamic, consistent with previously reported high turnover rates for synapses in culture (Okabe et al., 1999). The increase in synaptic number induced by media exchange required protein synthesis. Thus, the increase was not an imaging artifact but instead was a physiological response to the experimental protocol. The increase did not impede quantitative analysis and statistical testing of hypotheses because it was reproducible and in the opposite direction from activity-induced changes. Epileptic activity induced a robust loss of puncta that was highly reproducible and statistically different from the control conditions. This loss was seen in PSDs located on dendritic shafts and spines. Zha et al. (2005) have described a similar activity-induced loss of PSDs on spines, but in their study, tetrodotoxin induced a loss of PSDs on dendritic shafts. The dissociated cultures used in this study were derived from fetal animals, whereas Zha et al. (2005) used postnatal slice cultures. Overall, our data are in good agreement with previous studies that found a loss of excitatory synapses associated with epileptic activity (Müller et al., 1993; Colledge et al., 2003; Zha et al., 2005); we extended these previous studies to show pharmacological modulation of the process by cannabinoids. This assay was particularly well suited to detect the recovery of synapses after loss because quantitative data could be obtained from the same cell over time. Loss of synaptic puncta was dependent on activation of NMDA receptors and mediated by the ubiquitin-proteasome pathway, in good agreement with a study showing that PSD95 is ubiquitinated and rapidly removed from synaptic sites in response to NMDA receptor activation (Colledge et al., 2003). Thus, intense synaptic activity leads to proteasome-mediated loss of PSD95-GFP from synaptic sites.
Loss of postsynaptic proteins and dendritic spines are early events in many neurodegenerative disorders (Gylys et al., 2004). Loss of synaptic terminals in the brains of Alzheimer's patients more closely correlates with decreased cognitive function than cell death, supporting the hypothesis that disappearance of synapses is a key event in early cognitive decline (Terry et al., 1991). Here, we showed that intense glutamatergic synaptic activity decreased the number of synapses long before cells succumbed to excitotoxic death. This loss was reversible and mediated by the ubiquitin-proteasome pathway, attributes not generally associated with cell death pathways. Synapse loss may be a coping mechanism that is distinct from cell death processes because loss of PSDs seems to protect neurons from overstimulation and subsequent toxicity (Waataja et al., 2008). Loss of synaptic function as a result of degenerative disease, pharmacologic block of NMDA receptors, or cannabinoid inhibition of glutamate release all produce cognitive impairment and all seem to improve neuronal survival during excitotoxic insult. Thus, neuroprotective strategies need to balance protecting PSDs, which can exacerbate overstimulation of the postsynaptic cell, with allowing synapse down-regulation, which may improve survival at the cost of reduced participation in the synaptic network.
Here, we showed for the first time that inhibiting presynaptic Ca2+ channels either directly or by activation CB1 receptors prevented synapse loss resulting from epileptic activity. Ca2+ channel blockers are neuroprotective in some models of neurodegeneration (Valentino et al., 1993), and cannabinoids have antiepileptic properties (Croxford, 2003; Marsicano et al., 2003). However, this is the first report to show that either class of compounds protects network integrity from an excitotoxic insult. As a neuroprotective strategy, drugs that inhibit glutamate release have the advantage of both preventing synapse loss and improving survival. A disadvantage of this approach is the potential for psychoactive side effects.
Cannabinoids acted potently on presynaptic CB1 receptors to prevent excitotoxic synapse loss. These actions were in good agreement with previous reports from our laboratory (Shen et al., 1996; Shen and Thayer, 1999) and others (Hajos et al., 2001) showing cannabinoid inhibition of glutamatergic synaptic transmission. The reversal of the protective actions of the cannabinoids by rimonabant suggests that this drug might modify seizure threshold and network function in circuits with strong eCB tone (Marsicano et al., 2003). It is interesting to note that THC completely blocked PSD loss even though it reduces synaptic activity by less than 50% (Shen and Thayer, 1999), consistent with the idea that attenuation without complete block is sufficient to protect synaptic sites. This finding and the observation that the network displays a modest level of synaptic activity under control conditions suggest that the loss of PSDs requires sustained intense synaptic activity. Reducing [Mg2+]o to 0.1 mM evoked an aberrant pattern of synaptic activity that approximates epileptic discharges (McLeod et al., 1998). Down-regulation of synapses may underlie long-term depression of synaptic transmission (Colledge et al., 2003), in which case the threshold level of activity to induce PSD loss would presumably not be toxic. Perhaps the dramatic synapse loss described here in response to intense stimulation and the sustained moderate activity required to induce LTD lie on a continuum of activity that produces corresponding changes in synaptic input to stabilize neuronal excitability (Desai, 2003).
The cannabinoid receptor agonist WIN55,212-2 reversed the synaptic loss induced by 0.1 mM [Mg2+]o in the continued presence of the excitatory stimulus. This observation suggests that pharmacologic treatments might be used to improve or protect network function independent of treating the underlying cause of the synaptic loss. We speculate that protecting synapses might be important to pursue in parallel to strategies aimed at improving survival. Thus, the effective protection of synapses by cannabinoid agonists and calcium channel blockers suggests that network protection is in principle a reasonable approach. It will be interesting to evaluate other targets for synapse protection such as inhibition of postsynaptic proteolysis. If synapse loss is indeed a coping strategy, then synapse protection may increase the risk of excitotoxicity and thus will need to be part of an approach that also improves survival. Our results also indicate that there is a window during which synapse loss can be reversed. The cell culture model used for these studies is highly dynamic and would appear to undergo changes more rapidly than synapses studied in vivo (Grutzendler et al., 2002). Thus, the opportunity to protect network function may be longer in vivo where both loss and recovery may proceed on a slower time scale. It is clear that more in vivo studies of synaptic loss and recovery are warranted.
Cannabinoid-induced presynaptic inhibition of glutamatergic neurotransmission desensitizes after prolonged exposure to CB1 receptor agonists (Lundberg et al., 2005). Prolonged exposure to WIN55,212-2 diminished the protective effect of cannabimimetic drugs. However, even when tested at a maximally effective concentration, the desensitization produced by THC was less pronounced than that produced by WIN55,212-2. Thus, THC exerted a longer lasting modulation of synaptic function than the more potent and efficacious agonist. It is interesting to note that chronic WIN55,212-2 treatment did not itself affect the number of synapses (Fig. 6) but instead prevented subsequent protection afforded by CB1 receptor agonists (Fig. 6). Thus, prolonged exposure to cannabinoid agonists might diminish on-demand protection of synapses mediated by eCBs. These observations suggest that therapeutic strategies that enhance eCB function without chronic CB1 receptor activation, for example by inhibiting metabolism, might prove most effective for long-term protection of synaptic function. This approach would also avoid the desensitization that accompanies prolonged exposure to agonists.
In conclusion, we showed that intense excitatory synaptic activity decreases synapses between hippocampal neurons and that attenuation of this activity by cannabinoids prevented synapse loss. Excitatory neurons maintain a precarious balance between full integration into excitatory networks, with the associated risk of overstimulation and withdrawal from the network to increase survival, with the associated loss of network function. Modulation of presynaptic CB1 receptors either directly or through modulation of the eCB system may prove to be a useful pharmacologic approach for maintaining this balance during neurodegenerative disease.
Acknowledgments
We thank Donald B. Arnold (University of Southern California) for providing the PSD95-GFP expression plasmid.
Footnotes
-
This study was supported by the National Institute on Drug Abuse (Grants DA07304, DA11806, and DA24428) and National Institute on Drug Abuse Training Grant T32 07234 (to J.J.W.).
-
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
-
doi:10.1124/jpet.107.131607.
-
ABBREVIATIONS: THC, Δ9-tetrahydrocannabinol; CB, cannabinoid; eCB, endocannabinoid; PSD95, postsynaptic density protein 95; GFP, green fluorescent protein; rimonabant, N-piperidino-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-3-pyrazole-carboxamide; DMEM, Dulbecco's modified Eagle's medium; WIN55,212-2, (R)-(+)-[2,3-dihydro-5-methyl-3-[(4-morpholinyl)methyl] pyrrolo-[1,2,3-de]-1,4-benzoxazin-6-yl](1-napthalenyl)methanone monomethanesulfonate; MK801, (5R,10S)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine hydrogen maleate; MG132, Z-Leu-Leu-Leu-al; HHSS, HEPES-buffered Hanks' salt solution; BSA, bovine serum albumin; ANOVA, analysis of variance; NMDA, N-methyl-d-aspartate.
- Received September 12, 2007.
- Accepted February 27, 2008.
- The American Society for Pharmacology and Experimental Therapeutics