Retrograde synaptic signaling by endocannabinoids (eCBs) is a widespread mechanism for activity-dependent inhibition of synaptic strength in the brain. Although prevalent, the conditions for eliciting eCB-mediated synaptic depression vary among brain circuits. As yet, relatively little is known about the molecular mechanisms underlying this variation, although the initial signaling events are likely dictated by postsynaptic proteins. SAP90/PSD-95-associated proteins (SAPAPs) are a family of postsynaptic proteins unique to excitatory synapses. Using Sapap3 knock-out (KO) mice, we find that, in the absence of SAPAP3, striatal medium spiny neuron (MSN) excitatory synapses exhibit eCB-mediated synaptic depression under conditions that do not normally activate this process. The anomalous synaptic plasticity requires type 5 metabotropic glutamate receptors (mGluR5s), which we find are dysregulated in Sapap3 KO MSNs. Both surface expression and activity of mGluR5s are increased in Sapap3 KO MSNs, suggesting that enhanced mGluR5 activity may drive the anomalous synaptic plasticity. In direct support of this possibility, we find that, in wild-type (WT) MSNs, pharmacological enhancement of mGluR5 by a positive allosteric modulator is sufficient to reproduce the increased synaptic depression seen in Sapap3 KO MSNs. The same pharmacologic treatment, however, fails to elicit further depression in KO MSNs. Under conditions that are sufficient to engage eCB-mediated synaptic depression in WT MSNs, Sapap3 deletion does not alter the magnitude of the response. These results identify a role for SAPAP3 in the regulation of postsynaptic mGluRs and eCB-mediated synaptic plasticity. SAPAPs, through their effect on mGluR activity, may serve as regulatory molecules gating the threshold for inducing eCB-mediated synaptic plasticity.
The postsynaptic scaffold of proteins is important for synapse formation and dynamic modification of postsynaptic signaling. As such, these proteins can regulate neurotransmitter receptor activity. Scaffold proteins effectively link the major neurotransmitter receptors to intracellular signaling cascades and cytoskeletal elements (Sheng and Pak, 1999; Rumbaugh et al., 2003; Kim and Sheng, 2004). SAP90/PSD-95-associated proteins (SAPAPs; also referred to as guanylate kinase-associated proteins or GKAPs) are a family of postsynaptic scaffold proteins that localize to an intermediate layer of the PSD (Kim et al., 1997; Takeuchi et al., 1997; Valtschanoff and Weinberg, 2001) and are unique to excitatory synapses (Welch et al., 2004). By virtue of the protein complexes in which SAPAPs have been identified in vitro (Naisbitt et al., 1997; Boeckers et al., 1999; Tu et al., 1999; Hirao et al., 2000; Romorini et al., 2004), SAPAPs might influence the activity of ionotropic and metabotropic glutamate receptors and/or the intracellular signaling cascades with which these receptors interact. However, to date, very little is known about whether SAPAPs influence synaptic activity.
SAPAPs are encoded by a family of four genes that are widely expressed throughout the nervous system (Takeuchi et al., 1997; Kindler et al., 2004; Welch et al., 2004). In prior studies, we have shown that Sapap3 KO mice have obsessive compulsive disorder (OCD)-like behaviors (excessive self-grooming, facial lesions, anxiety-like behaviors, and therapeutic response to fluoxetine) and altered basal striatal neurotransmission (Welch et al., 2007). SAPAP3 is the only SAPAP that is highly expressed in the striatum (Welch et al., 2004, 2007). Viral rescue of SAPAP3 expression in the striatum of Sapap3 KO mice prevents the behavioral abnormalities and reverses the striatal neurotransmission defects demonstrating that striatal loss of SAPAP3 activity is critical for the expression of the pathological behaviors. These findings indicate that, at striatal synapses, SAPAP3 has little functional redundancy with other SAPAPs. Accordingly, the study of excitatory synaptic function in the striatum of Sapap3 KO mice provides a unique platform for elucidating the role of SAPAPs in synaptic function.
In this study, we investigate excitatory synaptic transmission of striatal medium spiny neurons (MSNs) in acute brain slices from Sapap3 KO mice. We find that loss of SAPAP3 results in abnormal endocannabinoid-mediated synaptic plasticity. Striatal excitatory synapses of Sapap3 KO mice engage eCB-mediated, short-term synaptic depression under conditions that are insufficient to activate this process in wild-type (WT) synapses. This is likely due to an increase in group 1 mGluR activity. Group 1 mGluR activity and mGluR5 surface expression are increased in Sapap3 KO MSNs. Pharmacological enhancement of mGluR5 activity in WT MSNs mimics the abnormal synaptic depression of KO MSNs. These findings provide the first functional evidence for a role of SAPAP3 in the regulation of postsynaptic mGluRs and eCB-mediated synaptic plasticity.
Materials and Methods
Generation of Sapap3 KO mice, Drd1a-tdTomato (D1) transgenic mice, and Drd2-EGFP (D2) transgenic mice was described previously (Gong et al., 2003; Welch et al., 2007; Shuen et al., 2008). Experimental mice were the progeny of Sapap3+/− by Sapap3+/− breedings in which only one of the parents expressed the Drd1a-tdTomato or Drd2-EGFP transgenes so that no experimental subjects were homozygous for the transgenic fluorescent markers. In experiments testing genotype effects, littermate control animals were used. All experiments were performed and analyzed with the experimenter blind to the identity of the experimental variable being tested (i.e., genotype or drug effect). Type 1 endocannabinoid receptor (CB1R) KO mice were provided by Dr. Rui Costa (Champalimaud Neuroscience Programme, Oeiras, Portugal) and generated by Dr. Andreas Zimmer (University of Bonn, Bonn, Germany) (Zimmer et al., 1999). All animal procedures were done according to protocols approved by the Institutional Animal Care and Use Committee of Duke University.
Brain slice preparation.
Coronal brain slices (300 μm) were used for all recording and imaging experiments. In extracellular recording experiments, Sapap3+/+ and Sapap3−/− mice aged 3–12 weeks and Sapap3−/−/CB1R−/− and Sapap3−/−/CB1R+/+ mice aged 24-52 weeks were used. Sapap3+/+ and Sapap3−/− mice coexpressing Drd1a-tdTomato and/or Drd2-EGFP transgenes were used for whole-cell recording experiments (3 weeks old) and calcium dye imaging experiments (2.5 weeks old). All electrophysiological results are from 3-week-old mice unless specified otherwise. Young mice were anesthetized with isoflurane before decapitation. Older mice (>8 weeks old) were anesthetized with tribromoethanol and transcardially perfused with oxygenated, ice-cold artificial CSF (ACSF) before decapitation. The brains were rapidly removed and sliced in oxygenated, ice-cold sucrose ACSF with a Vibratome 1500 (Leica). Sucrose ACSF contained the following (in mm): 194 sucrose, 30 NaCl, 2.5 KCl, 1.2 NaH2PO4, 0.2 CaCl2, 2 MgCl2, 26 NaHCO3, and 10 d-(+)-glucose (Choi and Lovinger, 1997). Slices recovered in standard ACSF saturated with 95% O2 and 5% CO2 at room temperature for at least 1 h before recording. Standard ACSF contained the following (in mm): 124 NaCl, 2.5 KCl, 1.2 NaH2PO4, 2 CaCl2, 1 MgCl2, 26 NaHCO3, and 10 d-(+)-glucose.
Slices were continuously perfused at a constant rate of 2–3 ml/min with standard ACSF saturated with 95% O2 and 5% CO2. Recordings were made in the dorsolateral region of striatum and stimulation was given in the corpus callosum. Stimuli were programmed by pClamp software (Molecular Devices) and delivered using an Iso-flex stimulus isolator (A.M.P.I.) and a twisted tungsten wire bipolar electrode (A-M Systems; outer diameter, 0.005 inches) (Lovinger and McCool, 1995). A MultiClamp 700B amplifier and DigiData 1322A (Molecular Devices) were used for signal acquisition. Signals were sampled at 20 kHz and low-pass filtered at 2 kHz. Picrotoxin (50 μm) was present in all experiments to exclude GABAA inhibitory synaptic activity. Experiments were performed at 30–32°C unless specified otherwise.
For field potential recordings, the recording pipette was filled with 2 m NaCl, and pipette resistance was ∼1–3 MΩ. The basal stimulation intensity was selected by identifying the lowest intensity that elicited the maximal field potential amplitude determined by an input/output survey performed at room temperature. Field potential responses were evoked using paired stimuli [interpulse interval (IPI), 50 ms]. For whole-cell, voltage-clamp recordings, membrane potential was held at −70 mV unless specified otherwise. Recording pipettes had a resistance of 2.5–4.5 MΩ when filled with internal pipette solution containing the following (in mm): 120 Cs-MeSO3, 5 NaCl, 10 tetraethyl ammonium chloride, 10 HEPES, 0.3 EGTA, 4 MgATP, 0.3 Na3GTP, and 5 QX314-Cl. The osmolarity was adjusted to ∼290 mOsm/L with sucrose, and the pH was adjusted to 7.3 with cesium hydroxide. Cells were visualized with a 40× water-immersion objective (LUMPlanFI, 40×/0.80 water immersion) under an Olympus BX51WI microscope equipped with infrared differential interference contrast optics, reflected fluorescence system, and OLY-150 camera (Olympus). Series resistance (Rs) was monitored continuously by a 5 mV voltage step following each pair of evoked EPSCs (eEPSCs). Average series resistance (Rs), input membrane resistance (Rm), holding current (Ihold), and membrane capacitance (Cm) were as follows: Rs, WT, 16.8 ± 2.0 MΩ, n = 13; KO, 20.3 ± 1.6 MΩ, n = 14; p = 0.2; Rm, WT, 240 ± 20 MΩ, n = 13; KO, 307 ± 35 MΩ, n = 14; p = 0.1; Ihold, WT, −27 ± 9 pA, n = 13; KO, −13 ± 7 pA, n = 14; p = 0.2; Cm, WT, 100 ± 4 pF, n = 13; KO, 87 ± 6 pF, n = 14; p = 0.1. Recordings with Rs > 30 MΩ or a change of >20% were excluded. EPSCs were evoked using paired stimuli (IPI, 50 ms). The basal stimulation intensity was adjusted to elicit baseline EPSC amplitudes between 200 and 400 pA. Paired-pulse ratios (PPRs) were calculated by the ratio of the averaged peak of the second EPSC [or population spike (PS)] to the averaged peak of the first EPSC (or PS) for each 60 s time period. All responses were normalized to the average value during the 7.5 min period of 90 s interval stimulation just before stimulation interval change (“baseline”). Summary bar graphs of activity-dependent depression protocol data present the average value during the last 10 min of 10 s stimulation normalized to baseline. The dose eliciting the submaximal effect of (R)-(+)-[2,3-dihydro-5-methyl-3- (4-morpholinylmethyl)pyrrolo[1,2,3-de]-1,4-benzoxazin-6-yl]-1- naphthalenylmethanone mesylate (WIN 55212-2) was identified in pilot experiments that tested a range of drug concentrations (0.3–6 μm) to determine a concentration of WIN 55212-2 that produced a submaximal level of depression. At 6 μm, a maximal degree of depression (eEPSC amplitude relative to baseline, 55 ± 9%; n = 4) was observed. Long-term depression (LTD) experiments were performed at 30 s basal stimulation interval. LTD was induced by four 100 Hz trains of 1 s duration paired with depolarization to 0 mV (Choi and Lovinger, 1997; Kreitzer and Malenka, 2005). The intertrain interval was 10 s. Stimulus intensity during high-frequency stimulation (HFS) was increased to the minimum intensity that evoked the maximal field response (Kreitzer and Malenka, 2005). Sample traces are the average of five consecutive sweeps. In all experiments reporting response amplitude, the values refer to the amplitude of the first response elicited by the pair of 50 ms IPI stimuli.
Calcium imaging in acute brain slices.
Acute coronal brain slices (300 μm) were prepared from postnatal day 17 (P17)–P19 WT and KO littermates expressing both Drd1a-tdTomato and Drd2-EGFP transgenes. One hour after sectioning, slices were transferred to incubation chambers containing 3 ml standard ACSF equilibrated with 95% O2 and 5% CO2 and maintained at 36°C for calcium indicator bulk loading (MacLean and Yuste, 2009). The fura-2 AM dye solution was prepared daily by dissolving 50 μg of fura-2 AM (Invitrogen) in 22 μl DMSO/Pluronic F-127 (0.02%; Invitrogen). Nine microliters of filtered dye solution were added to each incubation chamber for bulk loading of the slices. After 30 min, the slices were transferred to a holding chamber containing standard ACSF at room temperature for at least 45 min before imaging. All imaging was completed within 3 h of the loading procedure. Dye preparation as well as loading, recovery, and imaging of the slices were performed in the dark to minimize photobleaching of the fura-2 AM.
After recovery from the loading procedure, slices were transferred to the imaging chamber on an upright Ultima microscope (Prairie Technologies). Fields of view 50 ± 20 μm deep in the dorsolateral striatum were selected for imaging, and a femtosecond laser (Chameleon Ultra I; Coherent Technologies) was used to excite the transgenic and calcium indicator fluorophores. The laser wavelength was tuned to 1040 nm for tdTomato and 900 nm for EGFP. Fura-2 AM was imaged with a single wavelength (800 nm) that predominately excites fura-2 in a calcium-free state (Wokosin et al., 2004; Nikolenko et al., 2008). Fluorescence images were acquired using Prairie View image acquisition software (Prairie Technologies). Emitted photons were collected by a 20×, 1.0 NA XLUMPlanFL N water-immersion objective and an Aplanat/Acrhromatic 1.4 NA oil-immersion condenser (Olympus), and redirected to both epifluorescence and transfluorescence photomultiplier systems where emitted light was split into two channels by a 575 nm dichroic mirror. EGFP and fura-2 AM signals were imaged a 525/70 nm bandpass filter. TdTomato signal was imaged using a 607/45 nm bandpass filter.
After acquiring EGFP and tdTomato reference images, fura-2 fluorescence raster-scan images were acquired at a rate of 1 Hz throughout the experimental procedure, which included a no-drug baseline (10 s duration) and response to 100 μm (RS)-3,5-dihydroxyphenylglycine (DHPG; 20 s period following drug application). All solutions used during imaging contained standard ACSF and the sodium channel blocker tetrodotoxin (1 μm). Slices received continuous bath and local perfusion of solutions at room temperature throughout the imaging procedure. Drugs were applied locally via a VC-6M Mini-Valve System-controlled perfusion system (Warner Instruments). NMDA (500 μm) was applied at the end of each experiment as a positive control for correct placement of local perfusion tube.
Fluo-4 calcium imaging of corticostriatal cocultured neurons.
Corticostriatal cocultures were prepared from P1 littermate Sapap3 WT and KO mice. Following decapitation, striatal and cortical tissues were dissected in ice-cold dissection solution containing the following (in mm): 161 NaCl, 5 KCl, 0.5 MgSO4, 2.9 CaCl2, 5 HEPES, 5.6 d-(+)-glucose, and 0.006 phenol red at pH 7.4. Tissue was enzymatically digested for 20 min at 37°C in dissection solution, which also contained 0.2 mg/ml L-cysteine; 1 mm CaCl2; 0.5 mm EDTA, pH 8; 4,000 units/ml DNase I from bovine pancreas (all from Sigma-Aldrich); and 10 units/ml papain (Worthington Biochemical), pH 8.0. The digestion was inactivated by removal of the supernatant and addition of dissection solution containing 2.5 mg/ml bovine serum albumin (BSA), 2.5 mg/ml trypsin inhibitor, and 4,000 units/ml DNase I from bovine pancreas (all from Sigma-Aldrich). Tissue was triturated with low-resistance, fire-polished Pasteur pipettes, passed through 70 μm cell sieves (BD Biosciences), and pelleted at 200 × g for 8 min before resuspending in serum media that consisted of: Minimum Essential Medium with Earle's salts without l-glutamine (Sigma-Aldrich), 5% fetal bovine serum, 25 mm glucose, and 0.02% Mito+ Serum Extender (BD Biosciences). Cortical and striatal cells were plated at a 1:1 ratio at a density of 16 to 20 × 103 cells/well on Greiner 96 well poly-d-lysine-coated glass-bottom plates (Sigma-Aldrich) in 80% Neurobasal A media (NBA) containing 1× B-27 serum-free supplement (Invitrogen) and 1× GlutaMAX-1 supplement (Invitrogen) and 20% serum media.
On the day of imaging (DIV 15–17), cultures received a 20% media exchange with NBA containing 100 mm HEPES, pH 7.4, and were transferred to the cell imaging facility where they were imaged using the BD Pathway 855 high-content, high-throughput imaging system and an Olympus Plan Fluo 0.75 NA, 20× objective. All experiments were performed at 36–37°C. Cells were loaded with fluorometric imaging plate reader (FLIPR) dye for 1 h by replacing media with 90 μl of filtered imaging buffer containing 1× Hank's balanced salt solution (without phenol red or sodium bicarbonate), 4 mm HEPES, 0.5 mg/ml probenecid, 0.1 mg/ml ascorbic acid, 2 μm tetrodotoxin citrate (TTX), and 1.5× FLIPR calcium assay kit component A (Invitrogen), pH 7.4. Cells were maintained at 37°C during both dye loading and imaging. Imaging was completed within 1 h following completion of dye loading. Fluorescent images were obtained at 1 Hz during a no-drug baseline (10 s duration), response to 100 μm DHPG (20 s period following drug application), and subsequent response to 1–2 mm NMDA (20 s period), the latter done to identify live neurons.
Calcium dye image analysis.
Image processing for both fura-2 AM and Fluo-4 dye experiments was performed using Image J software (http://rsb.info.nih.gov/ij/). For acute slice data, regions of interest (ROIs) with a somatic area of 17–53 μm2 were selected from the tdTomato image (for D1 MSNs) or the EGFP image (for D2 MSNs). For the culture data, somatic ROIs from MSNs were selected from maximum intensity projections and cell type was confirmed using morphological features observed in the corresponding transmitted light images (somatic diameter, ≤14 μm; demonstrated somatic circularity, ≥0.8). Morphological selection criteria were derived from EGFP-fluorescent cells in striatal cultures prepared from Drd2-EGFP transgenic mice.
Fluorescence for all ROIs in both the culture and acute slice was normalized to the mean fluorescence intensity of the ROI during the 10 s baseline period preceding application of DHPG and is expressed as the change in fluorescence/mean baseline fluorescence (ΔF/F). Cells were excluded from analysis if they demonstrated an unstable baseline fluorescence intensity (SD of >0.03 ΔF/F) or showed no positive physiological response to the NMDA application (max response of <0.1 ΔF/F for cultures; <0.15 ΔF/F for acute slices) These objective criteria resulted in exclusion of 10% of WT D1, 9% WT D2, 14% KO D1, and 11% KO D2 cells for the acute slice data. WT and KO culture data were each derived from 13 wells and three separate culture preparations. Acute slice data were derived from seven slices from three animals (WT) and eight slices from two animals (KO). Additionally, <1.5% of transgene positive cells in the acute slices expressed both EGFP and tdTomato, and, when present, such cells were excluded from the analysis. Data are represented as the mean maximal response (within 20 s after DHPG application) ± SEM. The n values represent the number of cells per condition.
Immunocytochemistry and image analysis.
Corticostriatal cocultures prepared as described above were used to measure the surface expression of mGluR5. On DIV 13–14, cultures were fixed with 4% paraformaldehyde in 4% sucrose for 15 min at room temperature. Cultures were then rinsed with PBS and blocked in PBS with 0.2% gelatin, 2% BSA, 2% glycine, and 50 mm NH4Cl for 30 min at room temperature. Cultures were stained with anti-mGluR5 N-terminus rabbit polyclonal antibodies (1:50; AGC-007; Alomone Labs) overnight at 4°C. Cultures were then rinsed in PBS and permeabilized in PBS with 0.25% Triton X-100 for 5 min at room temperature. Cultures were rinsed with PBS, blocked, and then incubated with monoclonal antibodies against rat DARPP-32 (1:40; MAB4230; R & D Systems), a marker used to identify MSNs, for 2 h at room temperature. Cultures were then rinsed and incubated with the appropriate secondary antibodies (Alexa Fluor 568-conjugated goat anti-rabbit IgG, 1:800, Invitrogen; Cy5-conjugated donkey anti-rat IgG, 1:200, Jackson ImmunoResearch) for 2 h at room temperature. After the final washing with PBS, Fluoromount-G (Southern Biotech) was used to mount coverslips. Surface immunostaining, conditions were confirmed by lack of staining with DARPP-32 when incubated under nonpermeabilizing conditions. Images were collected with a 40× objective on an LSM 510 confocal microscope (Zeiss). Alexa Fluor 568 dye was excited with a 561 nm diode laser. Cy5 dye was excited with a 633 nm HeNe laser. No cross talk between channels was detected under these settings. All mGluR5 images were collected under identical acquisition conditions and subsaturating conditions. Images were collected from three independent batches of cultures with two coverslips per group for each batch. Six images were taken from each coverslip. Two dendritic regions (∼40 μm long) were selected for analysis by a separate experimenter who used only the DARPP-32-immunostained and transmitted light images for selection. Dendrites were selected if they were clearly part of a cell with high-intensity DARPP-32 staining (an indicator of medium spiny neurons) and did not have other dendrites overlying. For quantitative analysis, the average mGluR5 immunostaining intensity (total intensity/total area of dendritic region, in arbitrary units per pixel) was measured on DARPP-32-positive dendritic regions with MetaMorph software (Molecular Devices). In data not shown, an analysis of total mGluR5 intensity normalized to the total vGluT1 (Synaptic Systems; catalog #135511) intensity in dendritic regions of interest resulted in similar statistically significant conclusions.
All drugs were made as concentrated stock solutions and diluted in ACSF to their final concentration on the experiment day. N-(piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide (AM251), WIN 55212-2, 2-methyl-6-(phenylethynyl)-pyridine hydrochloride (MPEP), DHPG, 3-cyano-N-(1,3-diphenyl-1H-pyrazol-5-yl) benzamide (CDPPB), TTX, NMDA, and nifedipine were purchased from Tocris Bioscience. Picrotoxin and BSA were purchased from Sigma. BSA (0.05%) was used as a carrier for AM251, WIN 55212-2, and CDPPB experiments and their corresponding vehicle control experiments. For AM251 experiments (see Fig. 2), slices were preincubated with AM251 for 1 h before recording and continuously perfused with antagonist during recording. All other drug treatments were applied during bath perfusion only.
All electrophysiological recording data were analyzed by Clampfit 10.0 software (Molecular Devices). The peak amplitude of field population spikes and evoked ESPCs was measured and data were transferred to Microsoft Excel and GraphPad Prism for analysis and graphing. Data are presented as mean ± SEM, and the n value given for each experiment refers to the number of cells analyzed unless noted otherwise. Bar graph data present the average value of the response during the last 10 min of the 10 s stimulation interval period normalized to the baseline response value. Two-way repeated measures ANOVA (rmANOVA) was performed for the entire time course before post hoc comparisons were made with Student's t test or the Mann–Whitney rank test (JMP version 8.0 software; SAS institute). The significance level for all tests was p < 0.05. For linear regression analysis, R2 values are reported.
Activity-dependent synaptic depression of striatal excitatory synapses is increased in Sapap3 KO mice
Abnormal activity-dependent depression in Sapap3 KO mice was initially suspected because of an observation that evoked extracellular field potentials in Sapap3 KO striatal slices were preferentially unstable (rundown) or lost. However, stable responses could be obtained for both WT and KO genotypes by reducing the frequency of stimulation, suggesting that rundown of responses in KO brain slices might be caused by activity-dependent synaptic depression. To test this hypothesis, we evaluated the changes in field potential PS amplitude and PS PPR that occurred as a consequence of increasing stimulation rate. Stable baseline PS field potential responses were first obtained in dorsolateral striatum by stimulating each 90 s. Thereafter, the stimulation interval was reduced to 10 s. In response to the reduction in stimulation interval, PS amplitude in Sapap3 KO slices depressed markedly (PS amplitude relative to baseline, WT, 74 ± 4%, n = 10; KO, 43 ± 7%, n = 14; p = 0.003, t test) (Fig. 1A–C). Concomitantly, the PPR of the PS amplitude increased (PPR relative to baseline, WT, 124 ± 3%, n = 10; KO, 182 ± 22%, n = 14; p = 0.038, t test) (Fig. 1D,E). These findings were also present and of similar magnitude in mature adult mice (8 weeks and older; PS amplitude relative to baseline, WT, 71 ± 5%, n = 10; KO, 40 ± 4%, n = 11; p = 0.0002, t test). Although the amplitudes of basal field potential responses were lower in Sapap3 KO mice than WT mice (baseline PS amplitude, WT, −1.2 ± 0.1 mV, n = 20; KO, −0.9 ± 0.1 mV, n = 25; p = 0.035, t test), there was no correlation between the degree of activity-dependent depression and initial PS amplitude or PPR (percentage of depression vs initial PS amplitude, WT, R2 = 0.077; KO, R2 = 0.0004; percentage depression vs initial PPR, WT, R2 = 0.002; KO, R2 = 0.110).
Because the PPR of synaptic responses varies inversely with release probability, the frequency-dependent increase in PPR of Sapap3 KO PS responses suggests that a decrease in presynaptic release probability may underlie the PS depression. However, PS can also be influenced by nonsynaptic contributions. To isolate the synaptic contributions, whole-cell, voltage-clamp recordings were performed. In 3-week-old mice, whole-cell, voltage-clamp recordings were obtained from defined subpopulations of MSNs based on their expression of BAC transgenes encoding fluorescent reporters, Drd2-EGFP (Gong et al., 2003) or Drd1a-tdTomato (Shuen et al., 2008). These two populations reflect MSNs projecting to the indirect and direct pathways of the basal ganglia, respectively, and have distinct synaptic properties (Kreitzer and Malenka, 2007; Cepeda et al., 2008; Gertler et al., 2008). Using a similar stimulation paradigm, eEPSCs in D2 MSNs from Sapap3 KO mice showed a significant increase in activity-dependent depression relative to D2 MSNs from WT mice (eEPSC amplitude relative to baseline, WT, 53 ± 3%, n = 13; KO, 40 ± 2%, n = 14; p = 0.002, t test) (Fig. 1F–H). Paired-pulse ratios of eEPSCs were also significantly increased in D2 MSNs of Sapap3 KO mice relative to WT mice (PPR relative to baseline, WT, 113 ± 2%, n = 13; KO, 131 ± 5%, n = 14; p = 0.003, t test) (Fig. 1I,J). In D1 MSNs, qualitatively similar findings were present in Sapap3 KO mice. However, WT D1 MSNs had a greater degree of activity-dependent depression than WT D2 MSNs in this stimulation paradigm (eEPSC amplitude relative to baseline, D1 WT, 47 ± 3%, n = 16; D2 WT, 53 ± 3%, n = 13). As a result, smaller increases in synaptic depression and PPR were observed in D1 MSNs of Sapap3 KO mice compared to WTs (eEPSC amplitude relative to baseline, WT, 47 ± 3%, n = 16; KO, 39 ± 5%, n = 12; p = 0.138, rmANOVA; p = 0.143, t test; PPR, WT, 118 ± 5%, n = 16; KO, 129 ± 5%, n = 12; p = 0.02, rmANOVA; p = 0.117, t test). For this reason, in subsequent analyses, we focused on D2 MSNs for whole-cell recording experiments. Together, these experiments show that MSN excitatory synapses of Sapap3 KO mice have increased activity-dependent synaptic depression that is associated with a change in PPR, suggesting a change in release probability.
Excessive synaptic depression at Sapap3 KO excitatory synapses requires endocannabinoid signaling
It is intriguing to consider that SAPAP3, a postsynaptic protein, may cause enhanced synaptic depression through a presynaptic mechanism. We hypothesized that retrograde signaling by eCBs was involved because abnormalities in eCB signaling could arise from postsynaptic dysfunction and be expressed as a change in presynaptic function. To test this hypothesis, we performed three sets of experiments. First, preincubation of striatal slices with AM251 (3 μm), a potent CB1R antagonist, abolished the genotypic differences in activity-dependent depression of both field potential responses (Fig. 2A–C) and D2 MSN eEPSCs (Fig. 2G–I) between Sapap3 KO and WT mice. Second, we found that the activity-dependent depression was absent at room temperature (PS amplitude relative to baseline, WT, 99 ± 3%, n = 6; KO, 93 ± 6%, n = 6; p = 0.391, t test) (Fig. 2D). Although temperature may have diverse effects on neurotransmission, these data are consistent with a role for eCBs because, in the striatum, room temperature can inhibit eCB-mediated synaptic depression (Kreitzer and Malenka, 2005; Adermark and Lovinger, 2007; Adermark et al., 2009), presumably due to decreased activity of the plasma membrane transporter required for the extracellular transport of eCBs (Beltramo et al., 1997; Hillard and Jarrahian, 2000; Ronesi et al., 2004; Adermark and Lovinger, 2007; Hillard et al., 2007). Third, genetic deletion of CB1R significantly decreased striatal activity-dependent depression and PPR changes in Sapap3 KO mice to levels that were no longer distinguishable from Sapap3 WT mice (PS amplitude relative to baseline, Sapap3 KO/CB1R KO, 81 ± 6%, n = 11; WT, 74 ± 4%, n = 10; p = 0.364; PS PPR relative to baseline, Sapap3 KO/CB1R KO, 120 ± 4%, n = 11; WT, 124 ± 3%, n = 10; p = 0.357) (Figs. 1B,D, 2E,F). Thus, together, three lines of evidence show that eCB signaling mediates the excessive depression of MSN excitatory synapses in Sapap3 KO mice.
To investigate whether signaling through CB1Rs was altered in Sapap3 KO MSNs, CB1Rs were directly activated with agonist, WIN 55212-2. At concentrations eliciting submaximal levels of synaptic depression, WIN 55212-2 (1 μm) similarly depressed eEPSCs (eEPSC amplitude relative to baseline, WT, 69 ± 4%, n = 10; KO, 69 ± 2%, n = 14; p = 0.970, t test) (Fig. 2J) and increased PPR in WT and KO MSNs (PPR relative to baseline, WT, 116 ± 3%, n = 10; KO, 120 ± 6%, n = 14; p = 0.559, t test). Therefore, although CB1Rs are required for the activity-dependent depression, we found no evidence for an alteration in CB1R signaling itself in Sapap3 KO mice.
Sapap3 KO MSNs engage endocannabinoid signaling under anomalous conditions
When eCB signaling is blocked, a degree of synaptic depression persists in both genotypes (Fig. 2B,E,H). Therefore, we investigated the degree of synaptic depression that is mediated by eCBs in each genotype by comparing depression in the absence and presence of AM251 within genotypes. At WT striatal excitatory synapses, AM251 does not alter the degree of eEPSC depression (Fig. 2K), indicating that eCB-mediated synaptic depression is not normally engaged under these conditions. By contrast, an AM251-sensitive component of synaptic depression was readily observed at striatal excitatory synapses lacking SAPAP3 (Fig. 2L). Furthermore, the eCB-independent component of depression is similar in KO and WT MSNs (Fig. 1, 2). These findings reveal that loss of SAPAP3 engages eCB-mediated synaptic plasticity when it is not normally expressed.
Anomalous endocannabinoid synaptic plasticity at Sapap3 KO synapses requires mGluR5 activity
Because we observed eCB-mediated synaptic plasticity under anomalous conditions in Sapap3 KO mice, we evaluated whether the upstream signaling events that are normally required for eCB activity at striatal excitatory synapses were also required in this aberrant context. Endocannabinoid-mediated LTD is a well-described form of eCB-mediated synaptic plasticity at striatal MSN excitatory synapses. Striatal eCB-LTD requires both L-type voltage gated calcium channels and group 1 mGluRs (Calabresi et al., 1994; Sung et al., 2001; Kreitzer and Malenka, 2005; Wang et al., 2006). Short-term eCB-mediated synaptic plasticity as a result of afferent axonal stimulation alone has not been described at these synapses. However, these synapses do appear to be capable of short-term eCB-mediated synaptic plasticity if G-protein coupled receptors (GPCRs) critical for triggering endocannabinoid release, such as group 1 mGluRs, are pharmacologically stimulated (Kreitzer and Malenka, 2005; Narushima et al., 2006; Yin and Lovinger, 2006).
In Sapap3 KO MSNs, the mGluR5 antagonist, MPEP (40 μm), significantly reduced the degree of activity-dependent synaptic depression and PPR increase (Fig. 3A,B). In the presence of MPEP, levels of activity-dependent synaptic depression and PPR in Sapap3 KO mice were indistinguishable from WT mice (Fig. 3C,D). By contrast, the L-type calcium channel antagonist, nifedipine (10 μm), had no effect on synaptic depression or PPR at Sapap3 KO D2 MSNs (Fig. 3E,F). These results indicate that the anomalous eCB-mediated plasticity of Sapap3 KO striatal excitatory synapses requires signaling by mGluR5.
Group 1 mGluR signaling and mGluR5 surface expression are increased in MSNs of Sapap3 KO mice
Because mGluR5 activity is required for the abnormal synaptic depression in Sapap3 KO mice, and SAPAP3 might alter group 1 mGluR activity by virtue of its biochemical interaction with Shank, a protein found in a complex with group 1 mGluRs (Tu et al., 1999; Sala et al., 2001; Hwang et al., 2005), we investigated whether Sapap3 deletion altered the activity of group I mGluRs. Activity of these receptors was monitored by measuring changes in intracellular calcium levels in response to application of DHPG, a nonselective group 1 mGluR agonist. Group 1 mGluRs are coupled to Gαq/11-type G-proteins that increase cytosolic calcium when activated as a consequence of phospholipase C activation and inositol-1,4,5-trisphosphate production. In this assay, DHPG (100 μm) was applied to acute striatal brain slices expressing fluorescent reporters to differentiate D1 and D2 MSNs and in the presence of tetrodotoxin to prevent action potentials. Intracellular calcium levels were monitored in the somata of striatal MSNs using the calcium indicator dye, fura-2 AM and two-photon microscopic imaging. In response to DHPG, both D1 and D2 MSNs from Sapap3 KO mice had larger calcium transients compared to WT MSNs (Fig. 4).
We next investigated whether surface levels of group 1 mGluRs were also altered by Sapap3 deletion. Immunocytochemical methods were used to measure mGluR5 surface levels on dendrites of striatal neurons (indicated by high-intensity DARPP-32 immunostaining) in corticostriatal cocultures. We first confirmed that the cultured striatal neurons from Sapap3 KO mice faithfully reproduced the intracellular calcium signaling abnormality in response to DHPG that was observed in acute brain slice MSNs (Fig. 5A,B). Using this culture preparation, the average mGluR5 immunostaining intensity for a given dendritic region was measured. The average intensity of mGluR5 immunostaining on DARPP-32-positive dendrites was significantly increased in Sapap3 KO neurons as compared to WT (WT, 10 ± 0.8 a.u. per pixel, n = 71 dendritic regions; KO, 14 ± 1 a.u. per pixel, n = 75 dendritic regions; p = 0.003, Mann–Whitney test) (Fig. 5C,D). Analysis of the distribution of average mGluR5 intensities over the population of dendrites examined in each group further revealed that a uniform rightward shift was responsible for this increase, rather than a discrete change in a subpopulation of dendrites (Fig. 5E).
Enhancing synaptic mGluR5 activity at WT synapses increases activity-dependent depression
Thus far, our results suggest that an increase in the number and/or activity of type 5 mGluRs may underlie the enhanced eCB-mediated synaptic plasticity observed at MSN excitatory synapses of Sapap3 KO mice. To directly test this possibility, we examined the role of mGluR5 activation in this process. If increased activation of mGluR5 was responsible for triggering eCB-mediated synaptic depression in Sapap3 KO mice, increasing the number of activated type 5 mGluRs at WT synapses might be sufficient to reproduce the KO phenotype. Likewise, the same manipulation might be occluded at KO synapses.
To selectively increase the activity of only those mGluRs that are endogenously exposed to glutamate upon synaptic stimulation, the positive allosteric modulator of mGluR5, CDPPB was used (Lindsley et al., 2004; Kinney et al., 2005). In the activity-dependent stimulation paradigm, potentiation of synaptically activated mGluR5 by CDPPB increased the level of WT synaptic depression to that of KO synapses [WT + CDPPB, 55 ± 8%, n = 8; KO + CDPPB, 53 ± 5%, n = 10; p = 0.836, t test; WT + CDPPB (Fig. 6B) vs KO (Fig. 1C); 43 ± 7%; p = 0.328, t test] (Fig. 6A,B). By contrast, at KO synapses, CDPPB did not increase depression [KO + CDPPB (Fig. 6B) vs KO (Fig. 1C); 43 ± 7%; p = 0.350, t test]. Similar effects on field PS PPR were observed (Fig. 6C,D).
These results demonstrate that in WT mice, enhanced mGluR5 activation is sufficient to reproduce the degree of activity-dependent depression observed in KO mice. In addition, KO synapses do not express further activity-dependent depression with CDPPB, suggesting that the anomalous depression in KO synapses is due to enhanced mGluR5 activity.
Anomalous endocannabinoid signaling at Sapap3 KO striatal excitatory synapses produces short-term eCB plasticity
To investigate whether the 90 s/10 s interval activity-dependent depression protocol resulted in short- or long-lasting synaptic depression of KO synapses, we performed two experiments. First, after 20 min of paired stimulation at 10 s intervals, AM251 was capable of reversing the depression and PPR changes of Sapap3 KO synapses to levels indistinguishable from WT (Fig. 7A,B). Second, when electrical stimulation was paused briefly (10 min) and resumed at the reduced rate of each 90 s interval, synaptic responses returned to basal values (100 ± 7% of baseline; n = 5; p = 0.963, t test compared with initial baseline) (Fig. 7C). These two experimental paradigms indicate that the anomalous eCB-mediated synaptic plasticity observed in Sapap3 KO synapses is a form of short-term plasticity. Short-term eCB-mediated synaptic plasticity in the absence of pharmacological stimulation of GPCRs has not been reported at these synapses. Thus, Sapap3 deletion alters the conditions required for engaging eCB-mediated synaptic depression.
To test whether Sapap3 deletion alters the magnitude of eCB-mediated synaptic depression, we induced eCB LTD, a readily observed form of eCB synaptic plasticity at D2 MSN excitatory synapses. Endocannabinoid LTD was induced by a commonly used protocol, delivering four high-frequency stimulation trains paired with postsynaptic depolarization (Choi and Lovinger, 1997; Kreitzer and Malenka, 2005; Wang et al., 2006). No differences in the magnitude of LTD were observed between genotypes (eEPSC amplitude relative to baseline, WT, 54 ± 7%, n = 6; KO, 57 ± 6%, n = 6; p = 0.723, t test) (Fig. 7D). Thus, although Sapap3 deletion alters the conditions for engaging eCB-mediated synaptic depression, we did not find evidence that Sapap3 deletion alters the magnitude of such depression once engaged (Fig. 6A,B, 7D).
In the brain, mGluR-dependent endocannabinoid signaling is a common mechanism for activity-dependent regulation of synaptic strength. Although the pathway stimulating the synthesis of endocannabinoids has been well studied (Kano et al., 2009), little is known about postsynaptic mechanisms regulating the induction of this pathway. Here, we investigate excitatory synaptic transmission of striatal MSNs in Sapap3 KO mice and identify a role for SAPAPs in regulating the induction of eCB-mediated synaptic plasticity. Our results indicate that the normal role of SAPAPs is to negatively regulate mGluR5 activity and thereby restrict the conditions under which eCB-mediated synaptic depression occurs (Fig. 7E). These results identify SAPAPs for the first-time as potential mediators of endocannabinoid metaplasticity and further suggest that altered mGluR and/or eCB signaling may contribute to the pathological OCD-like behaviors of Sapap3 mutant mice.
Loss of SAPAP3 promotes eCB-mediated synaptic plasticity at MSN excitatory synapses
Endocannabinoids mediate diverse forms of synaptic plasticity throughout the nervous system. Conditions for the induction and expression of eCB-mediated synaptic plasticity can vary among circuits and between activity states (for review, see Wilson and Nicoll, 2002; Heifets and Castillo, 2009; Kano et al., 2009; Lovinger, 2010). Presynaptic factors have been identified that can influence whether short- or long-term inhibition of presynaptic release occurs as a consequence of CB1R activation (Ronesi et al., 2004; Chevaleyre et al., 2007; Singla et al., 2007). Although there is also evidence that postsynaptic mechanisms are at play, little is known about specific postsynaptic molecules that regulate eCB-mediated synaptic plasticity (Edwards et al., 2008; Kim and Alger, 2010; Roloff et al., 2010). In the present study, we observe that excitatory striatal synapses of Sapap3 KO mice engage eCB signaling under conditions that are not normally sufficient to elicit eCB-mediated synaptic depression. Although both genotypes exhibit synaptic depression under the activity-dependent protocol, only Sapap3 KO mice express an eCB-dependent component. Our results further provide evidence that (1) there is a threshold level of mGluR activation that is required to trigger eCB-mediated synaptic depression, (2) KO synapses reach threshold under conditions that WT synapses do not, and (3) once eCB-mediated synaptic depression is triggered, there are not additional significant differences in the magnitude of the effect at KO synapses in comparison to WT synapses. These results indicate that SAPAP3 may primarily influence the context under which eCB-mediated synaptic depression is activated. In this study, we show that genetic deletion of Sapap3 lowers the threshold for activating eCB-mediated synaptic plasticity by increasing mGluR activity. In future studies, it will be of interest to determine how SAPAP3 activity is modulated in space and time to regulate mGluR activity and eCB-mediated synaptic plasticity.
Loss of SAPAP3 increases group 1 mGluR activity and mGluR5 cell surface expression
Group 1 mGluRs are known triggers for eCB synthesis (Maejima et al., 2001; Varma et al., 2001; Ohno-Shosaku et al., 2002). In Sapap3 KO neurons, we observe both a functional increase in group 1 mGluR activity (intracellular calcium transients in response to DHPG) and a physical increase in mGluR5 surface receptors. These findings provide a mechanism for increased eCB-mediated synaptic plasticity at Sapap3 KO synapses. Although it is logical to speculate that the increased surface receptors are responsible for the increased mGluR activity, our data do not exclude the possibility that mGluR activation or coupling to downstream signaling events is enhanced independent of the increase in surface receptors. Such changes in group 1 mGluR activity have been reported when mGluR scaffolding is altered (Ango et al., 2001; Sala et al., 2005; Kammermeier and Worley, 2007). Likewise, although most of the experiments specifically evaluate mGluR5, the predominant group 1 mGluR expressed in MSNs, we cannot exclude the possibility that Sapap3 deletion also increases mGluR1 activity. Our findings of enhanced group 1 mGluR activity in Sapap3 KO mice indicate that other downstream sequelae of group 1 mGluR activation may also be altered.
One potential mechanism for the increase in surface mGluR5 in Sapap3 KO mice is through an impairment of mGluR endocytosis. This possibility is supported by prior studies that implicate Homer proteins in receptor endocytosis and localization of endocytic zones to dendritic spines (Gray et al., 2003; Davidkova and Carroll, 2007; Lu et al., 2007; Petrini et al., 2009). Homer proteins form a ternary complex with SAPAP3 and Shank3, providing a biochemical basis by which loss of SAPAP3 could alter Homer activity (Tu et al., 1999; Sala et al., 2001). A specific role for Homer in regulating group 1 mGluR endocytosis and surface levels has been demonstrated in experiments over-expressing Homer 1a, a short form of Homer that disrupts the interaction of long forms of Homer proteins with other scaffold proteins (Minami et al., 2003). Several recent studies demonstrate that Homer proteins can also alter group 1 mGluR-dependent endocannabinoid signaling (Fourgeaud et al., 2004; Jung et al., 2007; Roloff et al., 2010). Together these findings suggest that Shank and Homer may be functionally important in mediating the effects of SAPAP3 on group 1 mGluRs and endocannabinoids. Further experiments directly testing the role of these protein interactions are necessary.
Intriguingly, two clinical disorders that share a core feature of compulsive-repetitive behaviors, autism and Fragile X mental retardation, might also disrupt signaling through a putative mGluR/Homer/Shank/SAPAP pathway. Shank3 mutation is associated with a form of autism (Durand et al., 2007). Shank proteins interact biochemically with SAPAP and Homer (Tu et al., 1999). In mouse models for Fragile X mental retardation, SAPAP3 levels are altered (Narayanan et al., 2008), and both excessive group 1 mGluR and endocannabinoid activity have been reported (Huber et al., 2002; Bear et al., 2004; Maccarrone et al., 2010; Zhang and Alger, 2010). These behavioral and molecular similarities with Sapap3 KO mice suggest that increased mGluR5 activity and/or eCB-mediated synaptic plasticity in the striatum may represent a common mechanism for the expression of compulsive-repetitive behaviors.
This work was supported by National Institute of Neurological Disorders and Stroke Grants T32NS051156 (K.A.) and NS064577 and NS054840 (N.C.), the Tourette Syndrome Association (N.C.), the Klingenstein Foundation (N.C.), NARSAD (N.C.), National Institute of Mental Health Grants F32MH084460 (J.T.) and MH081201 (G.F.), The Simons Foundation Autism Research Initiative (G.F.), and the Hartwell Foundation (G.F.). We gratefully acknowledge Samantha Tracy, David Pan, and Audrey Nelson for outstanding technical assistance, members of the Calakos lab for critical discussions, Serena Dudek and Ben Philpot for comments on the manuscript, Rui Costa and Andreas Zimmer for providing CB1R KO mice, and Bryan Roth, Noah Sciaky, and John Allen for critical support in BD pathway imaging experiments.
- Correspondence should be addressed to Nicole Calakos, P.O. Box 2900, Duke University Medical Center, Durham, NC 27710.