Care and handling of animals were in accordance with NIH Animal Care and Use Committee Guidelines. Postnatal day (P) 15-21 Sprague-Dawley rats, injected with 1-1.2% CTB bilaterally through the superior colliculus, were maintained on a 12:12-hr light:dark cycle.

1. Retinal Tissue Fixation

1. Assemble the following materials and tools: A dissecting microscope, 2 forceps with very fine tips, scissors, cellulose filter paper, plastic pipette and a microscope slide.
2. Anesthetize the rat in a closed chamber with 2.0 ml halothane (an inhalant anesthetic). Determine adequate anesthetization by these methods: lack of withdrawal of rear paw after toe pinch, or lack of blink reflex. Then decapitate immediately with a guillotine. Remove the eyes with a pair of iris scissors and place in a glass dish containing 4% paraformaldehyde in 0.1 M phosphate buffer (PB) at pH 7.4.
3. Using the dissecting microscope, remove the cornea by cutting off the front of the eyeball. Remove the lens and vitreous from the inner retinal surface with forceps.
4. Peel the sclera with the two forceps until the retina is isolated from the eyecup.
5. Cut the retina immediately into 100 – 200 µm-thick strips with a razor, and subject to pH-shift fixation.
6. Fix retina strips in 4% paraformaldehyde in 0.1 M PB at pH 6.0 for 20 – 30 min and then in 4% paraformaldehyde plus 0.01% glutaraldehyde at pH 10.5 for 10 – 20 min at RT.
7. After several washes in PB with 0.15 mM CaCl₂ (pH 7.4 at 4°C), cryoprotect the retinal strips with glycerol (60 min each in 10%, 20%, 30%, then O/N in 30%) in 0.1 M PB prior to freeze substitution.

2. Freeze-substitution

NOTE: This freeze-substitution method is modified from an earlier published protocol ^19,20. Also, it is crucial that the instruments are very cold (wear gloves); otherwise, the tissue may thaw partially when touched with the instruments. All of these steps are done within the AFS chamber and the instruments are never allowed to move above the rim of the chamber. Similarly, proper cooling of all chemicals used in the AFS is necessary.

1. Plunge-freeze the retinal strips in liquid propane at -184°C in an EM cryopreservation instrument (CPC). Using a fine brush, place the samples on small pieces of double-stick tape attached to the metal stubs that go on the end of the plunging rods (wick off extra buffer using the brush).
2. Plunge the rods into the liquid propane and keep there for about 5 sec, and then transfer to the automatic freeze-substitution instrument (AFS) using a small transport chamber that is filled with liquid nitrogen ('LN₂ cooled cryo transfer container').
3. After placing the frozen sample and instruments (forceps and scalpel) into the AFS, cool the instruments in the chamber for several minutes before touching the tissue, or cool them by placing the tips of the instruments for a few seconds into the small transport chamber (filled with liquid nitrogen ('LN₂ cooled cryo transfer container')).  
4. Once in the AFS, remove the sample and tape from the stub using a fine scalpel. Also, keep the nitrogen gas flow control, TF (TF is described as a 'regulator control for LN₂ vaporiser') open during these procedures.
5. Prior to placing the frozen tissue into the flat-embedding holders (i.e., before setting up the AFS), cut a thin circle from a clear plastic sheet and place it into the bottom of the holder so as to line the bottom of each well. This allows relatively easier removal of the polymerized specimen blocks when finished.
   NOTE: Previously, we used an alternative method to the flat embedding holders, and employing double Reichert+gelatin capsules (see Petralia and Wenthold, 1999) ²⁰.
6. Use the following automatic sequence in the AFS (using instrument terminology): T1 = -90°C for 32 hr, S1 = increase temperature by 4°C/hr (11.3 hr), T2 = -45°C for 50 hr, S2 = increase temperature by 5°C/hr (9 hr), T3 = 0°C for 40 hr (total = 142.3 hr). It may take 2 – 4 hr to place samples in the AFS (at -90°C) prior to starting the timed sequence.
7. Immerse the frozen sections in 1.5% uranyl acetate in methanol at -90°C for >32 hr in the AFS. Place two pieces of tissues (typically) in each of the seven wells in the flat-embedding aluminum holder (three fit into the AFS).
   1. Place the tissue+tape into the uranyl acetate/methanol in the wells and remove the tape later, just prior to beginning embedding medium (such as Lowicryl HM20) infiltration, if it is too difficult to remove the tissue from the tape.
8. Then increase the temperature stepwise to -45°C (+4°C per hr; in the automatic sequence).
9. Wash the samples three times in precooled methanol, by using a fine-tipped plastic pipette to remove the old solution from each flat-embedding holder, and then using another standard plastic pipette to add the precooled fresh methanol.
10. Then, using the same method, infiltrate the samples progressively with low temperature embedding resins such as embedding medium (HM20/methanol at 1:1 and 2:1, each for 2 hr, followed by pure resin for 2 hr and then change the resins and keep O/N).
11. Change the resin again the next day, adjusting the level of the resin to reach just to the top edge of the wells.
12. Polymerize the samples (-45°C to 0°C in automatic sequence; +5°C per hr) with UV light for 40 hr.
13. Then remove the samples from the AFS. Typically, sample blocks still show some pinkness in color. Place the samples close to the fluorescent light in the chemical fume hood, at RT O/N or longer until they appear completely clear.

3. Postembedding EM Immunogold Immunocytochemistry

NOTE: Postembedding immunocytochemistry is performed as described ^23,24,25.

1. Cut 1 µm sections with ultramicrotome, stain sections with 1% toluidine-blue, and examine them for section orientation; orient the tissue block to achieve the optimal transverse plane of sectioning.
2. Cut 70 nm thick ultrathin sections with ultramicrotome and collect them on Formvar-carbon-coated nickel slot grids.
3. Wash grids one time in distilled H₂O followed by a Tris-buffered saline three times (TBS, 0.05 M Tris buffer, 0.7% NaCl, pH 7.6) wash.
4. Incubate grids in 20 µl drops of 5% BSA in TBS for 30 min , and then in 20 µl drops of a mixture containing goat anti-CTB (1:3,000) and an antibody to one NMDAR subunit (anti-rabbit GluN2A 1:50, GluN2B 1:30), or an anti-rabbit GluA2/3 (1:30) in TBS-Triton (TBST, 0.01% Triton X-100, pH 7.6) with 2% BSA and 0.02 M NaN₃ O/N at RT.
5. Wash grids on three separate drops (20 µl) of TBS (pH 7.6) for 10, 10, and 20 min, followed by TBS (pH 8.2) for 5 min.
6. Incubate grids for 2 hr on drops (20 µl) of a mixture containing donkey anti-rabbit IgG (1:20) coupled to 10 nm gold particles and donkey anti-goat IgG (1:20) coupled to 18 nm gold particles in TBST (pH 8.2) with 2% BSA and 0.02 M NaN₃.
7. Wash grids in 20 µl drops of TBS (pH 7.6) for 5, 5, and 10 min, then in ultrapure water and then dry them.
8. Counterstain grids with 5% uranyl acetate and 0.3% lead citrate in distilled H₂O for 8 and 5 min under dark conditions, respectively.
9. For triple-labeling experiments, incubate grids O/N at RT with 20 µl drops of a mixture of anti-goat CTB (1:3,000), anti-mouse PSD-95 (1:100), and anti-rabbit GluN2A (1:50). Then incubate grids for 2 hr on 20 µl drops of a mixture of IgGs coupled to 18, 10, and 5 nm gold particles in TBST (pH 8.2) with 2% BSA and 0.02 M NaN₃. Keep the same procedures as double labeling for washing and counterstaining between and after antibody incubation.
10. Controls were performed in which the secondary antibodies were applied alone.
11. View grids on an EM and digitalize images. Process final figures in Adobe Photoshop 6.0 only for brightness and contrast if it is necessary²⁴.

4. Quantification

1. Manually select areas of the inner plexiform layer (IPL) without holes or cracks at 8,000X magnification, then randomly photomontage the full depth of IPL at 25,000X magnification.
2. Identify RGC dendrites at cone bipolar dyads when they a) contain the retrogradely transported CTB signal (large gold particles), b) exhibit well-defined membranes, clefts, and postsynaptic densities, c) contain at least two small gold particles within the PSD, or more than one small gold particle along the extrasynaptic plasma membrane.
3. Collect 45 – 55 RGC dendrites, based on above criteria, for quantitative analysis.
4. Measure the lengths of the PSD and the extrasynaptic plasma membrane of individual RGC dendrites with NIH ImageJ software. Count gold particles within 10 nm of the membrane as membrane-associated, based on an average thickness of 7 – 9 nm for the plasma membrane ²⁶.
5. Calculate particle density as the number of gold particles per linear micrometer (gold/µm). Measure the distance between the center of each gold particle and the middle of the PSD (for synaptic location) or the edge of the PSD (for extrasynaptic location).
6. Perform statistical analysis by software. Use two-tailed t-tests to compare means. Conclude significance when P <0.05. Unless indicated otherwise, report values as mean ± SEM¹⁸.