dSNX16 CC variants regulate endosomal localization of Tkv to promote BMP signaling. (A) Quantification of the mean bouton number on muscle 4 in animals expressing VGlut-driven UAS-Tkv^Q199D and UAS-dSnx16-SNAP variants. (B) Representative images of α-pMad–stained animals expressing VGlut-driven Tkv^Q199D and dSnx16-SNAP variants in the ventral nerve cords. (C) Quantification of pMad intensity measured from single, central slices of dSNX16-positive cell bodies. (D) Representative images of animals expressing VGlut-driven Tkv-mCherry and dSnx16-SNAP variants at the muscle 4 NMJ, proximal axon, and cell body. (E) Representative axonal transport kymograph of VGlut-driven Tkv-mCherry and dSnx16-GFP (corresponds to Video 1). Arrowhead indicates a dSNX16 particle containing Tkv, and yellow box highlights the retrograde movement of this particle over time. (F) PCCs between Tkv-mCherry and dSNX16 CC variants. (G) Mean intensity quantification of Tkv-mCherry. (H) Representative SIM images of animals expressing VGlut-driven Tkv-mCherry and the indicated dSNX16-GFP in the cell body. dSNX16^3A-GFP (low) and dSNX16^3A-GFP (high) lines correspond to dSNX16^3A-GFPIIA and dSNX16^3A-GFPIIIF lines in Fig. S2. Arrowheads indicate tubular dSNX16 compartments that do not contain Tkv. Quantification is from ≥21 NMJs, 37 axons, or 61 cell bodies analyzed using Kruskal–Wallis tests followed by Dunn’s multiple comparisons test. All intensity measurements were normalized to mean intensity in the wild-type dSNX16 condition. All images show 2D maximum intensity projections of confocal stacks. Data are shown as box-and-whisker plots with all data points superimposed. *, P < 0.05; **, P < 0.01; ***, P < 0.001. ns, not significant. Scale bars, 10 µm; or 2 µm in the zoomed-in view of H.