Temperature shift redistributes AP-3 to the TGN independent of any change in the localization of TGN proteins. Related to Fig. 7. (A) PC12 cells were incubated either at 16°C for 30 min or 18°C for 4 h, fixed immediately, and immunostained. The images were acquired by confocal (left) or SIM (right) and reconstructed as in Fig. 1. Representative 3D SIM volume rendering shows distribution of TGN38 at the indicated temperature. Scatterplot indicates the area occupied by TGN38⁺ structures (n = 20 cells from three independent experiments). Scale bars, 5 µm (left) and 1 µm (right). (B) PC12 cells were incubated at 37°C, 16°C for 30 min, or 18°C for 4 h and immunostained for AP-3δ (green) and TGN38 (magenta). Magnified images show the colocalization of AP-3 with TGN38. Scatterplots show the Pearson’s correlation coefficient of AP-3 with TGN (top) and the Mander’s overlap coefficient for the proportion of pixels labeled strongly for TGN that also stained for AP-3 (TGN/AP-3, middle) and vice versa (AP-3/TGN, bottom) using autothreshold (n = 15 cells from three independent experiments). Scale bar, 5 µm. (C) PC12 cells were incubated at 37°C, 16°C for 30 min, or 18°C for 4 h and immunostained for AP-3δ (green) and TGN38 (red). The Mander’s overlap coefficient was used to quantify AP-3/TGN using autothreshold (above) or manual threshold (below). White puncta indicated the colocalization of AP-3 and TGN38 by auto- or manual threshold. Scatterplot indicates the Mander’s overlap coefficient for AP-3/TGN using a manual threshold. Scale bar, 5 µm. Error bars indicate mean ± SEM. ****, P < 0.0001 relative to control by one-way ANOVA with Tukey’s multiple comparisons test. Data S1 contains statistics source data.