Figure 4. Initial LM-to-EM axon matching.

(a,b) LM image plane showing an axon (green) overlaid with its LM-based skeleton reconstruction (ax1_LM, black) and a nearby blood vessel (B.v., blue) bifurcation (a, https://wklink.org/1399). Dashed rectangle, region magnified in (b, https://wklink.org/5406). (c,d) EM image plane showing the blood vessel (B.v.) corresponding to the vessel shown in (a,b) overlaid with the coarse affine transformation (see Figure 3) of the LM-based axon reconstruction ax1_LM^AT; magnified in d. Note that ax1_LM^AT does not match the ultrastructural features at high-resolution EM (d) due to the limited precision of AT_BV. (e,f), EM plane overlaid with ax1_LM^AT (as shown in c) with bounding box for dense axon reconstruction (light blue) of edge length ${\displaystyle {\lambda }_{align}}$ = 5 µm; demagnified in f, dashed rectangle denotes region shown in e. (g) Skeleton reconstructions of all n = 193 ax1_EM candidates (light grey) traversing the search bounding box (as in e-f), https://wklink.org/5897 (h) Reconstructed ax1_EM candidates (light grey) and the transformed fluorescent axon skeleton ax1_LM^AT (black). Five candidate EM reconstructed axons with the most similar trajectories as ax1_LM^AT highlighted in different colors. (i) Number of EM-reconstructed axons persistently within the ${\displaystyle {\lambda }_{align}}$ neighborhood of the LM-based template axon ax1_LM^AT. Note that at a distance of 43 µm from the seed center (e-h), only one possible ax1_EM candidate (green) remains. (j) Overlay of ax1_LM^AT and the most likely ax1_EM candidate reconstruction. Axonal varicosities (filled circles) were independently annotated in LM and EM. Note the strong resemblance of the varicosity arrangements indicating that ax1_EM is the correctly matched EM equivalent of the LM signal (see Figure 4—figure supplement 1) for statistical analysis). (k) EM plane with overlaid ax1_EM skeleton (dark green) and volume annotation of the corresponding EM axon ultrastructure (transparent green), https://wklink.org/3268.

Figure 4—figure supplement 1. Similarity of matched varicosity patterns between LM and EM.

(a) Independently measured varicosities (red and black circles) highlighted on the skeleton overlay of the initial affine transformed LM axon (ax1_LM^AT, red) and the EM match candidate (ax1_EM, black) (compare Figure 4j). (b) Overlay as in (a) but with ax1_EM varicosity positions (grey circles) randomly sampled. (c) Distribution of the average varicosity pair distance between the measured varicosities on ax1_LM^AT and the respective closest available varicosity on ax1_EM for the measured (black line) and randomly sampled (grey histogram bars) positions. The probability of randomly picking (n = 10,000 draws) varicosity positions with an average bouton pair distance as small as or smaller than the measured positions (black line, d = 0.67 µm) is p≤10⁻⁵ (see Materials and methods: Statistical tests).