Ultrastructural analysis of Dscam2^null boutons reveals endosomal defects. (A and B) Representative electron micrographs of control (A) and Dscam2^null (B) boutons. Arrowheads indicate examples of large vesicles (red), tubular endosomes (blue), DCVs (green), and MVBs (yellow). Mitochondria (M) and SSR indicated in white text. Scale bar, 500 nm. (C–F) High-magnification representative images of DCVs (green arrowheads, C), tubular endosomes (blue arrowheads, D), MVBs (yellow arrowhead, E), and large vesicles (red arrowhead, F). Scale bar, 100 nm. (G–K) Quantification of MVBs (G), large vesicle density (H), tubular endosome density (I), DCVs (J), and synaptic vesicle (SV) density (K). All measurements are the average per bouton. (L and M) Representative electron micrographs of control (L) and Dscam2^null (M) synaptic vesicles surrounding active zones (T-bars indicated by yellow arrowheads). Scale bar, 100 nm. (N and O) Quantification of SV diameter (N) and the number of SVs within 200 nm of T-bar (O). (P) Model. In wild-type nerve terminals, Dscam2 regulates CenG1A, which in turn enhances PI3K activity. This promotes PI(3)P deposition (green inositol ring) on early endosome (blue) membrane and inhibits the deposition of synaptic vesicles (brown) at active zones. Magenta inositol rings, PIP₂;RP, releasable pool; RRP, readily releasable pool.(Q) In Dscam2 mutants, decreases in CenG1A and PI3K activity lead to an increase in PIP₂ at the plasma membrane (magenta inositol ring) and a decrease in PI(3)P on early endosome (blue) membranes. Endosomal intermediates accumulate as 70–80-nm vesicles (gray), and more synaptic vesicles (brown) are deposited at active zones, which increases neurotransmitter release. Note that these pathways appear to function primarily under conditions of moderate neuronal activity. Data shown as mean ± SEM; n indicated in graph; Mann–Whitney rank-sum test. *, P > 0.05; **, P < 0.01; ****, P < 0.0001 for all panels.