Fig 2: CD166 is endocytosed in a clathrin- and dynamin-independent manner.HeLa cells treated with negative control (siCtrl, a–c), or µ2-adaptin (siAP2M1, a), dynamin-2 (siDNM2, b–c), or clathrin heavy chain (siCLTC, c) siRNAs. a, b Measure of anti-CD166 and transferrin (Tf) endocytosis by loss of surface assay in flow cytometry. a Number of independent experiments: anti-CD166, n = 3; Tf, n = 4. b Three independent experiments. NS, not significant. ****P < 0.0001 (RM two-way ANOVA with Bonferroni’s multiple comparison test). c Continuous uptake of anti-CD166 antibody for 10 and 30 min. Internal fluorescence quantified from confocal images and plotted as the relative percentage of siCtrl condition at 10 min. n cells, two independent experiments. NS not significant. ****P < 0.0001 (Kruskal–Wallis test with Dunn’s multiple comparison post hoc test, two-sided). The Western blots against a-adaptin (a), dynamin (b, c), and clathrin heavy chain (c) document the efficiency of siRNAs. Loading control, a-tubulin. d Continuous co-uptake of anti-CD166 antibody (red) and fluorescent Tf (green) for 5 min in HeLa cells. No co-localization between the two markers (see insets). Representative of three independent experiments. e–h Endocytosis imaged by lattice light-sheet microscopy on live cells. Full 3D volume of 60 planes per U2OS cell acquired within 3 s. Average endocytic intensity traces per lifetime, and classification of endocytic events for anti-CD166 (e) and Tf (g) uptake. Green, endogenous genome-edited CLTA-mRFP; Orange, cargo. Lifetime cohorts of endocytic trajectories show a “dome”-shaped intensity profile of cargoes and co-tracking or not with clathrin. f, h 2D projections of acquired 3D stacks of cargo (red), clathrin (green), and overlay (yellow). Representative of ten experiments. All ten experiments were used for quantifications (e, g). Scale bars, 5 µm (d), 10 µm (f, h). Data are mean ± s.e.m. Source data are provided as a Source Data file and in Supplementary Fig. 11 (blots).

Fig 3: Endophilin-A3 controls the clathrin-independent uptake of CD166.HeLa (a–e, g, h) or BSC-1 (f) cells treated with siRNAs or transfected with GFP constructs, as indicated. a, b, e, Treatment with negative control (siCtrl) or endoA1, A2, or A3 (siEndoA1, siEndoA2, or siEndoA3) siRNAs. c, f–h, Immunodetection of endogenous endoA1 and endoA2 with isoform-specific antibodies, or stable low-level expression of endoA3-GFP. a Continuous uptake of anti-CD166 antibody for 10 and 30 min. Internal fluorescence quantified from confocal images and plotted as the relative percentage of siCtrl condition at 10 min. n cells, two independent experiments. NS, not significant. *P < 0.05, ****P < 0.0001 (Kruskal–Wallis test with Dunn’s multiple comparison post hoc test, two-sided). b Correlations between the percentage of anti-CD166 endocytosis (after 40 min, Supplementary Fig. 3a) and the relative expression level of individual endoA1 (blue), A2 (red), and A3 (green) isoforms (quantified from blots, Supplementary Fig. 3b). c Airyscan images of cells fixed after 10 min of incubation with anti-CD166 antibody (red). White arrows, co-localization. Graph, quantifications of co-localization (n = 15 cells, three independent experiments). NS, not significant. ****P < 0.0001 (ordinary one-way ANOVA with Bonferroni’s multiple comparison test). d Pull-down experiments with cytosolic tail of CD166 on endoA3-GFP- or endoA2-GFP-expressing cell lysates. Negative control, GST alone. Positive control, ezrin. Representative of four independent experiments. e Incubation for 45 min at 37 °C with 50 nM STxB-Cy3. Fluorescence intensity in perinuclear Golgi quantified from tiled wide-field images, and normalized to siCtrl condition. Number of cells: siCtrl, n = 5918; siEndoA1, n = 6147; siEndoA2, n = 4791; siEndoA3, n = 6290. Four independent experiments. NS, not significant. **P < 0.01 (ordinary one-way ANOVA with Dunnett’s multiple comparison test and a single pooled variance). Representative images, Supplementary Fig. 7a. f Uptake of ß1-adrenergic receptor (ß1-AR) after addition of 10 µM dobutamine for 4 min at 37 °C. ß1-AR (red) and endoA2 or endoA3 (green). Graph, quantifications of co-localization (n = 30 cells, three independent experiments). ****P < 0.001 (two-tailed Mann–Whitney test). g, STED images of endoA isoforms (A1, red; A2, blue; A3, green). Equivalent confocal images, Supplementary Fig. 7c. h, Quantification of co-localization of endoA isoforms from confocal images (Supplementary Fig. 7c). Positive control, clathrin heavy chain, (CHC) and dynamin-2 (DNM2). Number of cells per condition: n = 15. Three independent experiments. ****P < 0.0001 (ordinary one-way ANOVA with Dunnett’s multiple comparison test). Scale bars, 2 µm (c, f) and 1 µm (g). Data are mean ± s.e.m. (c, e, h), mean ± 95% CI (b), or median ± 95% CI (a, f). Source data are provided as a Source Data file and in Supplementary Fig. 11 (blots).

Fig 4: Galectin-8 drives the uptake of CD166 and induces the recruitment of Endophilin-A3 to endocytic sites at the plasma membrane.HeLa cells. a Co-incubation with 5 µg ml-1 anti-CD166 antibody and 50 nM fluorescent Galectin-8 (Gal8) for 10 min at 37 °C. Graph, quantifications of co-localization (n = 15 images with 5–15 cells per image). Three independent experiments. ****P < 0.0001 (two-tailed unpaired t test with equal variances). Galectin-1 (Gal1) condition, Supplementary Fig. 9c. b Cells treated with control (siCtrl, red) or Gal8 (siGal8, blue) siRNAs incubated with 5 µg ml-1 anti-CD166 antibody for 10 min at 37 °C in various conditions (graph: ±10% FBS, ±purified Gal8). Intracellular accumulation of anti-CD166 quantified from confocal images. n cells, four independent experiments. NS not significant. ****P < 0.0001 (Kruskal–Wallis test with Dunn’s multiple comparison post hoc test, two-sided). c Cells stably expressing endoA3-GFP incubated with 50 nM fluorescent Gal8 for 10 min at 37 °C. White arrows, co-localization. Graph, number of co-localization events per cell. Number of cells: Gal8, n = 35; Gal1, n = 31 cells. Three independent experiments. ****P < 0.0001 (two-tailed unpaired t test with Welch’s correction). Gal1 condition, Supplementary Fig. 9f. d–f Gal8-induced recruitment of endoA3-GFP at plasma membrane. d Fluid-FM coupled to the confocal setup. Fluorescent galectin-coated gold nanoparticles trapped by microchanneled cantilever and approached to endoA3-GFP-expressing cells. Simultaneous monitoring of galectin (red) and endoA3 (green) fluorescence by fast-scanning confocal microscopy. e Representative images of Fluid-FM/confocal experiments (ten independent measurements). t = 0 s, initial cell exposure to Gal8. TL, transmitted light (cantilever). Insets: area surrounding the beads at the indicated time points. Graph: green fluorescence intensity around beads versus time. Images and plots for Gal1, dye control, bare probe, and endoA2-GFP controls, respectively, Supplementary Fig. 9i–l. f Quantification of fluorescence over time (slopes of linear regressions fitted on intensity vs. time curves). n-independent measures. NS not significant. **P < 0.01, ***P < 0.001 (two-tailed unpaired t tests). Scale bars, 10 µm (a), 1 µm (c), and 5 µm (e). Data are mean ± s.e.m. (a, c, f) or median ± 95% CI. b Source data are provided as a Source Data file.