Animals.

All animal experiments were approved by the Hokkaido University Animal Care and Use Committee. The dorsolateral portion of the striatum in adult male C57BL/6N, NL2-KO (Jackson Laboratory), CAST-KO (43), D₁R-KO (44), and D₂R-KO (45) mice (1–3 mo of age) was analyzed. D₁R-KO and D₂R-KO mice were homozygous for a knock-in mutation in which the human interleukin-2 receptor α-subunit gene was inserted into the D₁R and D₂R genes, respectively. Wild-type and KO littermates were analyzed for comparison. Both the qualitative and quantitative data were obtained from three mice in each animal group.

Fixation.

Under deep pentobarbital anesthesia (100 mg/kg of body weight, i.p.), mice were perfused transcardially with 60–75 mL of 4% (wt/vol) paraformaldehyde in 0.1 M phosphate buffer (PB, pH 7.2) for 10 min, unless otherwise noted. Striatal tissues perfused with 4% paraformaldehyde/0.1% glutaraldehyde in PB or 2% (wt/vol) paraformaldehyde/0.2% picric acid in PB were also used to detect GABA or dopamine receptors, respectively, by postembedding immunoelectron microscopy.

Section Preparation.

For conventional immunofluorescence and pre-embedding immunoelectron microscopy, sections (20 or 50 μm in thickness) were prepared on a microslicer (VT1000S; Leica) and processed for free-floating incubations. Ultrathin sections (100 nm) were prepared using an ultracryomicrotome (EM-FCS; Leica) after cryoprotection with 0.32M sucrose in PB, and mounted on silane-coated glass slides. For postembedding immunogold electron microscopy, microslicer sections (300 µm) were cryoprotected with 30% (vol/vol) glycerol in PB and frozen rapidly with liquid propane in the EM CPC unit (Leica). Frozen sections were immersed in 0.5% uranyl acetate in methanol at −90 °C in the AFS freeze-substitution unit (Leica), infiltrated at −45 °C with Lowicryl HM20 resin (Electron Microscopy Sciences), and polymerized with UV irradiation. Ultrathin sections were cut with an ultramicrotome (Ultracut; Leica) and mounted on formvar-coated nickel mesh grids. For fluorescent in situ hybridization, brains were removed from the skull under deep pentobarbital anesthesia and frozen immediately in powdered dry ice. Fresh-frozen sections (20 μm) were cut with a cryostat (CM1900; Leica).

Fluorescent in Situ Hybridization.

We used the following fluorescein- or digoxigenin-labeled riboprobes: mouse D₁R (1-950, NM_010076), mouse D₂R (416-1105, NM_010077), and mouse dopamine- and cAMP-regulated neuronal phosphoprotein or protein phosphatase DARPP32 (298-1000; NM_144828). Riboprobe synthesis and in situ hybridization were performed following the protocol described previously (36). Nuclear counterstaining was performed using TOTO-3 (Life Technologies). Images were captured using a confocal laser-scanning microscope equipped with a HeNe/Ar laser, and UPLSAPO 10×/0.40 and 20×/0.75 objective lenses (FV1000; Olympus). The specificity of hybridization signals was confirmed by a lack of any significant labeling using the sense riboprobes.

Antibody.

Information on the antigens, host species, sources, specificity, and source or references of primary antibodies is summarized in Fig. S7 and Table S1. Briefly, overall patterns of immunofluorescence labeling in coronal striatal sections are provided for TH (Fig. 1A), plasmalemmal DAT (Fig. 1B), VMAT2 (Fig. S7A), GAD65/67 (Fig. S7B), and VIAAT (Fig. S7C). The specificity of main primary antibodies is shown for cytomatrix protein at the active zone CAST by blank labelings by immunoblot and immunohistochemistry in CAST-KO mice (Fig. 1H and Fig. S7 D–F); Nrxns by detection of >150-kDa protein bands in the brain and HEK293T cell lysates transfected with Nrxn1α–3α (Fig. S7 G and H); D₁R and D₂R by the lack of immunolabeling in D₁R-KO and D₂R-KO mice (Fig. S7 I–L); DARPP32 by selective detection of a 32-kDa protein band in the brain and HEK293T cell lysates transfected with DARPP32, similar patterns of DARPP32 immunohistochemistry and in situ hybridization, and lack of DARPP32 immunolabeling in striatal interneurons (Fig. S7 M–Q); gepherin by extensive overlap with NL2 clusters in the striatum and vice versa (Fig. S7S); and NL3 by overlap with NL2 clusters (yellow arrows) and wider distribution than NL2 (white arrows) in the striatum (Fig. S7T).

Immunoblot.

Whole brains were homogenized using a Potter homogenizer with 15 strokes at 1,000 rpm in 10 volumes of ice-cold homogenizing buffer containing 0.32 M sucrose, 1 mM EDTA, 1 mM EGTA, 10 mM Tris⋅HCl (pH 7.2), and 0.4 mM phenylmethylsulfonyl fluoride. Homogenized samples were centrifuged at 1,000 × g for 10 min to remove nuclear fractions. To prepare lysates of transfected cells, HEK293T cells (2 × 10⁶) were transfected with mammalian expression vectors using the calcium phosphate precipitation method, harvested 24–48 h after transfection, and sonicated in 10 volumes of ice-cold homogenizing buffer. Mammalian expression plasmid vectors pCAG-HA-Nrxn1α, pCAG-HA-Nrxn2α, and pCAG-HA-Nrxn3α were kind gifts from Kensuke Futai (University of Massachusetts Medical School, Worcester, MA). NL1–4, GABA_ARα1, and DARPP32 cDNAs were amplified from a mouse cDNA library by PCR, subcloned into pTracer-GFP (Life Technologies), and sequenced. The lysates were denatured in 50 mM (±)-DTT at 65 °C for 30 min. Proteins were separated by SDS/PAGE and electroblotted onto transfer membranes (Immobilon-P; Millipore). Membranes were successively incubated with 5% (wt/vol) skimmed milk for 30 min, primary antibody (1 μg/mL) for 1 h, and peroxidase-linked secondary antibody (1:10,000; Jackson ImmunoResearch) for 1 h. Tris-buffered saline (pH 7.5) containing 0.1% Tween 20 was used as a diluent and washing buffer. Proteins were visualized with ECL prime Western blotting detection reagent (GE Healthcare).

Immunofluorescence.

All incubations were performed at room temperature. Sections were incubated successively with 10% (vol/vol) normal donkey serum for 20 min, a mixture of primary antibodies overnight (1 µg/mL), and a mixture of species-specific secondary antibodies for 2 h at a dilution of 1:200 (Life Technologies; Jackson ImmunoResearch; Abcam; Biotium). Combinations of primary and secondary antibodies for multiple immunofluorescence are listed in Table S2. For the detection of GABA_ARα1, microslicer sections were subjected to antigen exposure pretreatment before normal donkey serum blocking: 1 mg/mL pepsin in 0.2 N HCl at 37 °C for 30–60 s. PBS containing 0.1% Tween 20 was used as a dilution and washing buffer. Images were taken with a confocal laser-scanning microscope equipped with 405-, 473-, 559-, and 647-nm diode laser lines and UPLSAPO 10× (NA, 0.4), PLAPON 60×OSC2 (NA, 1.4; oil immersion), and UPLSAPO 100×S (NA, 1.35; silicone oil immersion) objective lenses (FV1200, Olympus) under the following microscopic settings: zoom factor, 1, 4, 6, and 10; image size, 640 × 640 pixels; pinhole size, 1 airy unit. Figs. S1H and S2 G–I were taken with an FV-OSR software module (Olympus) to obtain superresolution images. To obtain high-resolution images shown in Fig. 1 A and B, and Figs. S4A and S7 A–C, a series of images were taken with a UPLSAPO 10× objective lens (zoom facter 1.3, 256 × 256 pixels) and were stitched using FV10-ASW 3.0 (Olympus).

Quantitative Assessment of Dopamine and GABAergic Synapses on GFP-Labeled Dendrites.

To assess the density and molecular expression of dopamine and GABAergic synapses on dendrites of given neuronal types, we labeled single neurons with GFP by injecting lentiviral vector to the striatum of mice at 4 wk of age under chloral hydrate anesthesia (350 mg/kg of body weight, i.p.). The lentiviral vector system with murine stem cell virus promoter (MSCV) was kindly provided by St. Jude Children’s Research Hospital. The original lentivector pCL20c-MSCV-GFP was modified by adding the woodchuck hepatitis posttranscriptional regulatory element (WPRE) sequence downstream of GFP to enhance GFP expression. Injected mice were fixed 1–2 mo later by transcardial perfusion of fixative and processed for multiple immunofluorescence.

We randomly selected GFP-labeled dendrites that ran near and parallel to the section surface, and obtained their z-stacked image series consisting of 20–50 optical sections in steps of 0.2 µm with PLAPON 60×OSC2 and a zoom factor of 6 or 10 for 3D-reconstruction with Imaris (Bitplane) or ImageJ (imagej.nih.gov) software. Dopamine and GABAergic synapses were defined as NL2⁺ or gephyrin⁺ clusters on GFP-labeled dendrites in contact with TH-labeled dopaminergic terminals or VIAAT-labeled GABAergic terminals, respectively. Apposition of the three fluorescences was confirmed by images of the X–Y, X–Z, and Y–Z planes. For quantification, we sampled 5–10 dendrites (>10 µm in length and 0.75–1.5 µm in diameter) from each of three mice in each group. The mean length (µm) and the mean diameter (µm) of analyzed dendrites (mean ± SEM) were as follows: 17.7 ± 1.4 and 1.08 ± 0.03 for d-MSNs (Fig. 4B), 18.0 ± 1.4 and 1.13 ± 0.04 for i-MSNs (Fig. 4C), 17.5 ± 0.87 and 1.09 ± 0.06 for PV⁺ interneurons (Fig. S4C), 18.9 ± 2.1 and 0.98 ± 0.04 for nNOS⁺ interneurons (Fig. S4D), 15.1 ± 2.0 and 1.10 ± 0.06 for CHT⁺ interneurons (Fig. S4E), 20.1 ± 1.2 and 1.04 ± 0.03 for control neurons (Fig. 6A and Fig. S6 D and J), 17.5 ± 2.8 and 0.96 ± 0.03 for NL2-KD#1 neurons (Fig. 6B and Fig. S6 E and K), and 21.8 ± 2.8 and 1.02 ± 0.03 for NL2-KD#2 neurons (Fig. 6C and Fig. S6 F and L). Dendritic spines were also sampled from six spiny dendrites from each of three mice. The mean length (µm) of dendritic spines was 1.45 ± 0.03 for d-MSNs and 1.52 ± 0.03 for i-MSNs (Fig. 4D, filled columns). The density of synapses was shown as mean ± SEM of synapse numbers per 10 µm of dendritic shafts or spines. Fluorescent intensities for gephyrin, NL2, and NL3 clusters on GFP-labeled dendritic shafts (mean ± SEM) were measured from three to four dendrites in each of three mice, and the background level, as estimated from those in the nucleus, was subtracted (Fig. S6 H, I, and M). All of the images presented in this study were taken as single optical sections.

Criteria of Synapses.

Synapses were identified by electron microscopy as interneuronal junctions with the following specializations: rigid and parallel apposition of the apposed membranes, the extracellular space stuffed with some electron-dense materials, accumulation of synaptic vesicles and fuzzy active zone in presynaptic elements, and attachment of the postsynaptic density (PSD) to the cytoplasmic side of the postsynaptic membrane. Asymmetric synapses were defined as synapses whose postsynaptic membrane is attached by thick PSD, by which the postsynaptic specialization appeared more dense than the presynaptic specialization (1). In comparison, symmetrical synapses were defined as those having pale and thin PSD, with almost comparable thickness between the postsynaptic and presynaptic specializations.

By postembedding immunogold electron microscopy, we identified dopamine synapses as TH-labeled symmetric synapses, GABAergic synapses as VIAAT-labeled symmetrical synapses, and glutamatergic synapses as asymmetrical synapses without TH or VIAAT labeling. To minimize the sampling of false-positive dopamine and GABAergic synapses, we only analyzed symmetric synapses having five or more metal particles for TH or VIAAT, respectively, on the presynaptic cytoplasm excluding the mitochondria. The mean and SEM of data obtained from postembedding immunogold analysis were calculated from the mean values of three mice in each group, as shown in Table S3.

Pre-Embedding Immunoelectron Microscopy.

For single-label pre-embedding immunogold electron microscopy for D₁R (Fig. 2A), D₂R (Fig. 2B), DAT (Fig. S5 F and G), and VMAT2 (Fig. S5 K and L), microslicer sections were dipped in Blocking Solutions (Aurion) for 30 min, and incubated overnight with primary antibodies to D₁R (guinea pig), D₂R (rabbit), DAT (rabbit), and VMAT2 (rabbit) (1 µg/mL) diluted with 1% BSA/0.004% saponin in PBS and then with secondary antibodies linked to 1.4-nm gold particles (1:100; Nanogold; Nanoprobes) for 2 h. After postfixaion with 1% glutaraldehyde in PBS for 15 min, immunogold was intensified with a silver enhancement kit (R-GENT SE-EM; Aurion) for 30–60 min.

For double-label pre-embedding immunoelectron microscopy (Fig. 2 D and E), sections labeled for guinea pig D₁R or D₂R antibody (metal particles) were further incubated for TH labeling. After blocking with 10% (vol/vol) normal guinea pig serum for 30 min, sections were incubated for rabbit TH antibody (1 µg/mL) overnight, biotin-conjugated donkey anti-rabbit secondary antibody (1:200; Jackson ImmunoResearch) for 2 h, and peroxidase-conjugated streptoavidin (Nichirei) for 30 min. TH labeling was visualized as 3,3′-diaminobenzidine (DAB) precipitates with a DAB substrate kit (SK-4100; Vector). Immunolabeled sections were treated with 1% osmium tetroxide for 15 min, block-stained with 2% (wt/vol) uranyl acetate for 20 min, dehydrated with a graded ethanol series and propylene oxide twice for 10 min each, and embedded in Epon 812 (TAAB). Ultrathin sections were prepared with an ultramicrotome (Ultracut; Leica) and mounted on copper-mesh grids.

Quantitative Measurement for Pre-Embedding Immunoelectron Microscopy.

To avoid uneven antibody penetration, electron micrographs were taken within 5 µm (D₁R or D₂R in Fig. 2 A, B, D, and E; VMAT2 in Fig. S5 K and L) or 10 µm (DAT in Fig. S5 F and G) from the surface of sections subjected to pre-embedding immunoelectron microscopy. To quantify metal particle labeling for D₁R or D₂R (Fig. 2C), 3 × 4 montage images (∼3.6 µm × 4.8 µm) were randomly taken at a magnification of 20,000×. In double-label pre-embedding immunoelectron microscopy for TH and D₁R or D₂R (Fig. 2G), we searched dendrites, whose profiles formed TH-labeled dopamine synapses and had three or more metal particles for D₁R or D₂R, at a magnification of 4,000×. Then, 3 × 3 montage images (∼3.6 µm × 3.6 µm) over dopamine synapses were taken at a magnification of 20,000×, and four to seven dopamine synapses were sampled from each of three mice in each group. The density of labeling was calculated per 1 µm of the synaptic, perisynaptic, and extrasynaptic membranes.

Metal particles for VMAT2 were also measured at dopamine synapses as the density per 1 µm² of the cytoplasm (Fig. S5M) by sampling nerve terminals, which contained five or more particles for VMAT2 and formed symmetric synapses with dendritic shafts and spines. We took 3 × 3 montage images (∼3.6 µm × 3.6 µm) at a magnification of 20,000×, with five to eight terminals from each of three mice in each group.

Postembedding Immunoelectron Microscopy.

Double-label postembedding immunogold electron microscopy was used to quantify molecular expression at TH-labeled dopamine synapses (Figs. 1 E, G, and I and 3 D–F and Figs. S2 J and K, S3 D and E, and S5 P, Q, U, and V), at VIAAT-labeled GABAergic synapses (Fig. 1D and Figs. S1 I and K and S3 A–C), or glutamatergic synapses without TH or VIAAT labeling (Fig. 1D and Figs. S1 J and L and S3 D and E). Ultrathin sections on nickel grids were successively treated with the following solutions: saturated sodium ethoxide for 2 s, 50 mM glycine in incubation solution (0.03% Triton X-100 in Tris-buffered saline, pH 7.4; TBST) for 10 min, blocking solution containing 2% (vol/vol) normal goat serum in TBST (GTBST) for 10 min, primary antibodies (1:1,000 dilution for GABA antibody and 20 μg/mL for CAST D₁R, D₂R, GABA_ARα1, gephyrin, NL2, NL3, Nrxn pan-AMPAR, PSD-95) diluted in GTBST overnight, and colloidal gold-conjugated (10-nm diameter) secondary antibodies (1:100; British BioCell International) in GTBST for 2 h. After blocking with 2% (vol/vol) normal rabbit or guinea pig serum in TBST, sections were subjected successively to mouse or rabbit TH antibody or rabbit or guinea pig VIAAT antibody in GTBST (20 or 10 µg/mL, respectively) for 6 h and colloidal gold-conjugated (15-nm diameter) secondary antibodies (1:100; British BioCell International) in GTBST for 2 h. Sections were washed thoroughly in distilled water, and stained with 5% (wt/vol) uranyl acetate in 40% (vol/vol) EtOH for 90 s and Reynold’s lead citrate solution for 90 s Photographs were taken with a JEM1400 electron microscope (JEOL).

Quantitative Measurement for Postembedding Immunoelectron Microscopy.

First we searched asymmetrical synapses or symmetrical synapses with immunogold labeling for TH or VIAAT at a magnification of 5,000×, and took 3 × 3 montage images (∼2.4 µm × 2.4 µm) of glutamatergic, dopamine, and GABAergic synapses at a magnification of 30,000×. To measure the density of presynaptic labeling for Nrxn (Fig. 1J) and postsynaptic labeling for GABA_ARα1, gephyrin, NL2, NL3, PSD-95, and AMPAR (Fig. 3 G–I and Figs. S3 F and G and S5 R and W), immunogold particles <20 nm apart from the center of the cell membrane were measured. In assessment of presynaptic labeling for CAST (Fig. 1H), we counted immunogold particles <40 nm from the center of the presynaptic membrane. We examined 5–15 dopamine, GABAergic, or glutamatergic synapses from each of three mice in each group, with the number of analyzed synapses ranging fewer than seven synapses in the same groups. Metal particles for GABA were also measured at dopamine, GABAergic, or glutamatergic synapses as the density per 1 µm² of the cytoplasm excluding the mitochondria (Fig. 1F).

The density of D₁R and D₂R labelings on the synaptic, perisynaptic, and extrasynapti membranes at around dopamine synapses (Fig. S2 L and M) was calculated by measuring immunogold particles <20 nm apart from the center of the cell membrane in contact with TH-labeled dopamine terminals. From each of three mice, 17–20 dendritic profiles forming dopamine synapses were analyzed. The density of immunogold labeling at given synapse types or compartments was calculated as that per 1 µm of the synaptic, perisynaptic, and extrasynaptic membranes.

Dopamine Synapse Density.

The density of dopamine synapses in the striatum was compared between NL2-KO and wild-type mice by measuring the nearest neighbor distance (Fig. S5H). Dopamine synapses, which were labeled with five or more metal particles for DAT and formed symmetric synapses on dendritic shafts and spines, were sampled, and their 5 × 5 montage images (∼11 µm × 11 µm) were taken at a magnification of 10,000×. We measured distances from the center of the postsynaptic membrane of given dopamine synapses to that of the nearest ones, and the mean nearest neighbor distance was calculated by sampling five to eight dopamine synapses from each of three mice in each group.

Knockdown by MicroRNA.

For miR-based knockdown, we used the BLOCK-iT Pol II miR RNAi Expression Vector Kits (Life Technologies) according to the manufacturer’s instructions. pCL20c-MSCV-GFP-miR-WPRE was obtained by inserting the following miR sequences into pCL20c-MSCV-GFP-WPRE: 5′-GTGTACATCCTGGTCCACTAGCGTTTTGGCCACTGACTGACGCTAGTGGCAGGATGTACAC-3′ (NL2-miR#1); 5′-GTGAACTTGGTGTCCTGTGGCAGTTTTGGCCACTGACTGACTGCCACAGCACCAAGTTCA-3′ (NL2-miR#2); 5′-GAAATGTACTGCGCGTGGAGACGTTTTGGCCACTGACTGACGTCTCCACGCAGTACATTT-3′ (control-miR).

Lentiviral vectors were produced by cotransfection of HEK293T cells (5 × 10⁶ per one dish) with a mixture of pCL20c-MSCV-GFP-WPRE or pCL20c-MSCV-GFP-miR-WPRE (20 µg per 1 dish), pCAG-kGP4.1R (12 µg), pCAG4-RTR2 (4 µg), and pCAGGS-VSVG (4 µg) using the calcium phosphate precipitation method. HEK293T cells were incubated in DMEM (D5796; Sigma) supplemented with 10% (vol/vol) FBS (Cell Culture Bioscience), 1× penicillin/streptomycin (P4333; Sigma), 1 mM sodium pyruvate (S8636; Sigma), 1× MEM nonessential amino acid solution (M7145; Sigma), and 1× GlutaMAX (#35050061; Life Technologies) at 37 °C in a 10% CO₂ atmosphere. Culture medium was replaced with Ultraculture (12-725F, Lonza) 16 h after transfection. Supernatant containing viral particles was collected 40 h after transfection, centrifuged at 1,000 × g for 10 min, filtered through 0.45-μm membranes, and concentrated using an Amicon Ultra filter unit 100k (Millipore). Viral solution was further concentrated by centrifuging at 6,000 × g for 16 h, before suspending in PBS. Viral titer [expressed as transducing units (TU)] was calculated as 1.0 × 10^8–9 TU/mL by measuring the number of infected HEK293T cells.

For viral injection in newborn pups, deep anesthesia was obtained by lowering the body temperature on ice, and the skull was fastened with a handmade stereotaxic instrument. A glass pipette (G-1.2; Narishige) filled with 0.5 μL of viral solution was inserted into the striatum. Viral solutions were injected by air pressure (Pneumatic Picopump; World Precision Instruments). Injected mice were fixed 2 mo later by transcardial perfusion of fixative, as described above.

Coculture Assay.

Coculture experiments were performed according to previous methods with modifications (32). Briefly, the midbrain and striatum from newborn mice were collected in ice-cold Hank’s balanced solution (HBS), and dissociated in 0.25% trypsin/HBS for 10 min and 0.25% trypsin/0.1% DNaseI/HBS for 5 min at 37 °C. Dissociated neurons were suspended in neuronal basal medium supplemented with B-27 solution (both from Life Technologies) and GlutaMAX, seeded on plastic dishes coated with 5% (vol/vol) Matrigel (BD Biosciences) at a density of 1–3 × 10⁴ cells/cm², and incubated at 5% CO₂. After 9–11 d in vitro, HEK293T cells transfected with pTracer-NL1, pTracer-NL2, pTracer-NL3, pTracer-GABA_ARα1, and pEGFP-N1 (Clontech) were seeded on the primary midbrain or striatal culture at a density of 2 × 10⁴ cells/cm². After 48 h, cells were fixed with 4% (wt/vol) paraformaldehyde in 0.1 M PB for 10 min at room temperature for immunofluorescence.

Statistical Analysis.

For quantitative comparison, immunofluorescence data from three mice in each group were pooled together and are presented as the mean ± SEM. Statistical significance was assessed by Mann–Whitney u test (for comparison of two independent samples) or one-way ANOVA with Tukey’s post hoc test (for comparison of more than three independent samples). Statistical significance of quantitative electron microscopic data were assessed by unpaired t test, because no significant differences were found in variances between two independent samples by F-test.