Figure 3.

Reliable detection of putative TC contacts at the light level. The same spines are examined both in confocal microscopy stacks (A) and in subsequent reconstructions of 70 nm serial sections imaged by SEM (B). C–D, Example of a true positive (confirmed TC contact), the spine of which is indicated by the white arrowheads in A and B. Under confocal microscopy (C), the spine (red fluorescence) apposes a VGluT2⁺ vesicle pool (green fluorescence). The corresponding ultrastructure (D) reveals the spine (black punctate stain) with its PSD (red arrowheads) apposed to a VGluT2⁺ vesicle pool (dark gray diffuse stain, green arrowheads). E–J, Examples of a true positive, false positive and true negative from different areas of the same dendrite shown in A and B. E, F, True positive, legend same as for C and D. G, H, False positive. The 70 nm section (H) in which the labeled spine (black punctate stain, red arrowhead) is closest to the VGluT2⁺ vesicle pool (green arrowhead) does not contain a PSD and no synapse is formed between both labeled structures. The labeled spine's PSD is located in a different section and does not contact the VGluT2⁺ vesicle pool. Note that this ultrathin section contains only a small, terminal portion of the spine. I, J, True negative. Nearby, but nonapposed presynaptic and postsynaptic structures visualized under confocal microscopy (green, red fluorescence in I) are separated by ∼50 nm. This spine fails to satisfy the criteria for putative TC contacts; examination under SEM confirms that this is not a synapse (green arrowhead in J corresponds to the VGluT2⁺ vesicle pool in green fluorescence in I). The labeled spine head (black punctate stain, red arrowhead indicates PSD) forms a synapse with a separate terminal containing a VGluT2⁻ vesicle pool (blue arrowhead).