Immune synapse components ALCAM and ICAM1 are retrograde transport cargoes that rely on EndoA3 and retromer

A Illustration of the SNAP-tag-based BG-labelled antibody uptake assay to study membrane protein endocytosis and retrograde transport. B Confocal images of GalT-GFP-SNAP (green) and TMR-Star (red) in HeLa cells stably expressing the Golgi-resident GFP-fused SNAP-tag construct (HeLa GalT-GFP-SNAP). Actin (phalloidin, white) and nuclei (DAPI, blue) were also stained. Fluorescence intensity profile was made along the dashed line region in enlarged cropped area, and shows the colocalization of both signals. Scale bar: 20 μm. C-F Retrograde transport of ALCAM and ICAM1. Continuous BG-labelled anti-ALCAM (C and E) and anti-ICAM1 (D and F) antibody uptake for 4h at 37°C in HeLa GalT-GFP-SNAP cells. (C-D) Western blot analysis of HeLa GalT-GFP-SNAP cells transfected for 72h with siRNAs: negative control (siCtrl) or against retromer subunits (siVPS35 and siVPS26A). Immunodetection made with anti-SNAP, anti-VPS35, anti-VPS26A and anti-α-Tubulin (loading control) antibodies. Quantification of the covalent IgG-SNAP-GFP-GalT complex is shown as fractions of siCtrl condition (histogram). Quantification of VPS35 and VPS26A depletion is shown in Fig. S1C. (E-F) Western blot analysis of HeLa GalT-GFP-SNAP cells transfected for 72h with siRNAs: negative control (siCtrl) or against EndoA3 (siEndoA3). Immunodetection made with anti-SNAP, anti-EndoA3 and anti-clathrin heavy chain (CHC, loading control) antibodies. Quantification of the covalent IgG-SNAP-GFP-GalT complex (IB:anti-SNAP) is shown as fractions of siCtrl condition (histogram). Quantification of EndoA3 depletion is shown in Fig. S1H. C Data information: In (B), images are from a single experiment. Quantification data (C-F) are pooled from three independent experiments. Data are presented as mean ± SEM. *P < 0.05, **P < 0.01. One sample t and Wilcoxon test. Source data are available online for this figure.

EndoA3-dependent CIE mediates the uptake of immune synapse components ALCAM and ICAM1 in cancer cells

A-B Live-cell TIRF images of EndoA3-GFP (stable) and ICAM1-mScarlet (transient) in HeLa (A) and LB33-MEL (B) cells. Time series show enlarged cropped areas corresponding to region 1 in the full-size images, and are extracted from Movie S1 (A) and Movie S2 (B). White arrows indicate dynamic co-distribution of both signals. Kymographs were made along the dashed lines in the enlarged cropped areas corresponding to region 2 (A-B; Movie S3-S4). Scale bars: 10 μm (full size image), 2 μm (enlarged cropped areas). C-E Continuous uptake of anti-ALCAM antibody for 15 min at 37°C in the following LB33-MEL cell lines: wild-type (WT, D), stably transfected with empty plasmid (Φ, D), or stably expressing EndoA3-GFP (LB33-MEL EndoA3+, C-E). In (E), cells were transfected with siRNAs: negative control (siCtrl) or against EndoA3 (siEndoA3). Quantification of EndoA3 depletion by western blots in Fig. S3C. (C) Airyscan images of Anti-ALCAM (red) and EndoA3-GFP (green). White arrowheads show colocalization between ALCAM and EndoA3. Scale bars: 10 μm (full size image), 1 μm (enlarged cropped areas). (D-E) Quantifications of anti-ALCAM internalization, expressed as fractions of WT condition (D) or siCtrl condition (E). (D) n cells: WT, n = 270; Φ, n = 279; EndoA3+, n = 274. (E) n cells: siCtrl, n = 350; siEndoA3, n = 234. Representative image examples in Fig. S3A-B. Data information: All images (A-C) are representative of two independent experiments. In (D-E), data are pooled from three independent experiments. Data are presented as median and quartiles. ns, not significant. ****P < 0.0001 (D, Kruskal-Wallis test with Dunn’s multiple comparison test; E, Mann-Whitney test). Source data are available online for this figure.

Inhibition of EndoA3-dependent endocytosis and retrograde transport impairs T cell activation but increases their lytic activity

A Scheme of flow cytometry analysis of cytokine production inside CD8 T cells stimulated by siRNA transfected LB33-MEL EndoA3+ cells. BFA, Brefeldin A. B-C Flow cytometry analysis of CD8 T cell intracellular cytokine production after co-culture for 4h at 37°C with: no LB33-MEL (T only), LB33-MEL EndoA3+ transfected with negative control siRNA (siCtrl) or with siRNA against EndoA3 (siEndoA3). Quantification of EndoA3 depletion is shown in Fig. S4A. (B) Representative examples and quantifications (scatter plots) of the percentages of cytokine producing CD8 T cells (up, IL-2; middle, TNFα; bottom, IFNγ) after being stimulated by siCtrl or siEndoA3 treated LB33-MEL EndoA3+ cells. (C) Representative examples and quantifications (scatter plots) of the absolute amount of cytokines produced by CD8 T cells (up, IL-2; middle, TNFα; bottom, IFNγ; presented as median fluorescence intensity, MFI) after stimulation by siCtrl or siEndoA3 treated LB33-MEL EndoA3+ cells. D Scheme of ELISA analysis of cytokine secretion from CD8 T cells stimulated with siRNA transfected LB33-MEL EndoA3+ cells. E Quantification of IFNγ secretion (detected by ELISA) from CD8 T cells, cultured alone (T only) or co-cultured for 20h with the following LB33-MEL cell lines: wild-type LB33-MEL cells (WT), LB33-MEL cells stably transfected with empty (Φ) or EndoA3-GFP encoding plasmid (EndoA3+), treated with negative control siRNA (siCtrl) or with EndoA3-targeting siRNA (siEndoA3). Data are presented as fractions of WT siCtrl condition. The absolute concentration of secreted IFNγ in the supernatant of WT siCtrl condition is 2203 ± 201 (mean ± SEM) pg/mL. F Quantification of CD8 T cell cytolytic efficiency against LB33-MEL EndoA3+ cells transfected with negative control siRNA (siCtrl) or siRNA against EndoA3 (siEndoA3). CD8 T cell killing efficiency was determined by Chrome 51 release assay and is presented as percentage of lysed LB33-MEL EndoA3+ cells. Different T cell:Target cell ratios were tested. G Quantification of IFNγ secretion (detected by ELISA) from CD8 T cells, cultured alone (T only) or co-cultured for 20h at 37°C with EndoA3+ LB33-MEL cells transfected with siRNAs: negative control (siCtrl), or against retromer subunits (siVPS35 or siVPS26A). Data are presented as fractions of siCtrl condition. The absolute concentration of secreted IFNγ in the supernatant of siCtrl condition is 4077 ± 99.62 (mean ± SEM) pg/mL. Quantification of VPS35 and VPS26A depletion is shown in Fig. S4L. Data information: In (B-C), data are pooled from five independent experiments. In (E), data are pooled from six independent experiments. In (F-G), data are pooled from three independent experiments. In (E,G), data are presented as mean ± SEM. ns, not significant. *P < 0.05, **P < 0.01, ***P < 0.001 (B and C, paired t test; E, RM one-way ANOVA with Tukey’s multiple comparison test; F, two-way ANOVA with Sidak’s multiple comparison test; G, RM one-way ANOVA with Dunnett’s multiple comparison test). Source data are available online for this figure.

ICAM1 is directionally transported to the immune synapse via vesicular trafficking in cancer cells

A Representative live-cell spinning-disk confocal image of an immune synapse-like conjugate formed between a CD8 T cell (red) and an adherent, stable HLA-A*68012–expressing HeLa cell transiently expressing ICAM1-mScarlet (green) (Movie S8). Three regions were selected for quantitative analysis: region 1 at the T cell contact site, and regions 2 and 3 away from the conjugate (quantified in panels C-G). The T cell is delineated by a dotted line. Scale bar, 20 μm. B Time series of the enlarged cropped area (region 1′) from panel A, extracted from Movie S8 and shown in Movie S9. White arrowheads indicate ICAM1-positive tubulo-vesicular carriers moving toward the contact area and fusing with the conjugate membrane. Scale bar, 5 μm. C-E Tracking of ICAM1-positive carriers in the three regions defined in panel A-region 1 (C, with immune synapse) and regions 2-3 (D-E, without immune synapse). For each region, all recorded tracks are displayed on the cell image (left) and in corresponding X–Y displacement plots (right). Scale bar, 10 μm. F Quantification of track density in each region. Track density is markedly higher in region 1 where is the CD8 T cell contact. G Quantification of track directionality and proportion across regions. A higher proportion of tracks is oriented toward the T cell contact site in region 1. Data information: Images and quantifications are from a single movie, representative of two independent experiments.

Inhibition of EndoA3-mediated endocytosis and retrograde transport affects ICAM1 recruitment to and structure of immune synapses

A Airyscan images of an immune synapse-like conjugate formed between a CD8 T cell (stained for CD45, yellow) and a stable HLA-A*68012-expressing HeLa cell (stained for actin, green) transiently expressing ICAM1-mScarlet (red) in suspension. Cell segmentation (white contour) was based on the actin staining for further quantifications (B). Scale bar: 5 μm. B Quantification of relative ICAM1 recruitment to the vicinity of CD8 T cell when an immune synapse-like conjugate is formed between a CD8 T cell and a stable HLA-A*68012-expressing HeLa cell transiently expressing ICAM1-mScarlet in suspension. HLA-A*68012-expressing HeLa cells were transfected for 72h with siRNAs: negative control (siCtrl) or against EndoA3 (siEndoA3). Western blot analysis of EndoA3 depletion is shown in Fig. S5E. n conjuagates: siCtrl, n = 85; siEndoA3, n = 75. C,E Airyscan images of immune synapse-like conjugates formed between CD8 T cells (stained for CD45, yellow) and stable EndoA3-GFP-expressing LB33-MEL cells (EndoA3-GFP, green) transfected for 72h with different siRNAs: negative control (siCtrl), siRNAs against EndoA3 (siEndoA3, C) or against VPS26A (siVPS26A, E). Actin (phalloidin, red) and nuclei (DAPI, blue) were also stained. Two examples are displayed per condition. Scale bar: 5 μm. D,F Quantifications of the sizes of immune synapse-like conjugates formed between CD8 T cells and stable EndoA3-GFP-expressing LB33-MEL cells transfected with different siRNAs from images in panels C and E. Western blot analyses of EndoA3 and VPS26A depletion are shown in Fig. S6C and S6D. (D) n conjugates: siCtrl, n = 48; siEndoA3, n = 58. (F) n conjugates: siCtrl, n = 74; siVPS26A, n = 78. G Working model. Immune synapse components ALCAM and ICAM1 are endocytosed into cancer cells through EndoA3-mediated CIE and are subsequently transported to the TGN in a retromer-dependent retrograde manner. Upon immune synapse formation, immune synapse components, such as ALCAM and ICAM1, are efficiently recruited to the contact area with CD8 T cells, likely from the TGN by polarized re-distribution, which can stabilize the immune synapse and promote cytokine production/secretion by CD8 T cells (left). In contrast, disruption of EndoA3-mediated CIE or retromer-dependent retrograde transport in cancer cells impairs the efficient recruitment of immune synapse components to the contact area with CD8 T cells, resulting in a larger but less stable immune synapses. Although CD8 T cells attempt to compensate by expanding their contact areas with cancer cells, this is insufficient to support efficient cytokine production and secretion. However, the reduced stability of the immune synapse may facilitate more rapid detachment and re-engagement of CD8 T cells, thereby enhancing their lytic activity, potentially through increased serial killing (right). Data information: In (A), (C) and (E), images are representative of three independent experiments. In (B), (D) and (F), data are pooled from three independent experiments. Data are presented as median and quartiles. *P < 0.05, **P < 0.01, ***P < 0.001 (B, Kolmogorov-Smirnov test; D and F, Mann-Whitney test). Source data are available online for this figure.

Immune synapse components ALCAM and ICAM1 are retrograde transport cargoes that rely on EndoA3, retromer and Rab6

A Confocal images of GalT-GFP-SNAP (green) and TMR-Star (red) in LB33-MEL cells stably expressing the Golgi-resident GFP-fused SNAP-tag construct (LB33-MEL GalT-GFP-SNAP). Actin (phalloidin, white) and nuclei (DAPI, blue) were also stained. The fluorescence intensity profile was made along the dashed line region in enlarged cropped area, and shows the colocalization of both signals. Scale bar: 20 μm. B Confocal images of GalT-GFP-SNAP (green) and TGN46 (red) in LB33-MEL GalT-GFP-SNAP cells. Nuclei (DAPI, blue) were also stained. The fluorescence intensity profile was made along the dashed line region in enlarged cropped area, and shows the colocalization of both signals, indicating that GalT-GFP-SNAP is correctly localized at the TGN. Scale bar: 10 μm. C Quantifications of the immunoblots shown in Fig. 1C-D (IB: anti-VPS35 and anti-VPS26A) confirm depletion efficiency of VPS35 and VPS26A in HeLa GalT-GFP-SNAP cells. D Retrograde transport of ICAM-1. Continuous BG-labelled anti-ICAM1 antibody uptake for 4h at 37°C in LB33-MEL GalT-GFP-SNAP cells. Western blot analysis of LB33-MEL GalT-GFP-SNAP cells transfected with siRNAs: negative control (siCtrl) or against retromer subunits (siVPS35 and siVPS26A). Immunodetection with anti-SNAP, anti-VPS35, anti-VPS26A and anti-α-Tubulin (loading control) antibodies. Quantification of the covalent IgG-SNAP complex is shown as fractions of siCtrl condition (histogram). Quantification of VPS35 and VPS26A depletion is shown in Fig. S1E. E Quantifications of the immunoblots shown in Fig. S1D confirm depletion efficiency of VPS35 and VPS26A in LB33-MEL GalT-GFP-SNAP cells. F Retrograde transport of ALCAM. Continuous BG-labelled anti-ALCAM antibody uptake for 4h at 37°C in HeLa GalT-GFP-SNAP cells. Western blot analysis of HeLa GalT-GFP-SNAP cells transfected for 72h with siRNAs: negative control (siCtrl) or against Rab6 (siRab6). Immunodetection made with anti-SNAP, anti-Rab6 and anti-clathrin heavy chain (CHC, loading control) antibodies. Quantification of the covalent IgG-SNAP-GFP-GalT complex is shown as fractions of siCtrl condition (histogram). Quantification of Rab6 depletion is shown in Fig. S1G. G Quantifications of the immunoblots shown in Fig. S1F confirm depletion efficiency of Rab6 in HeLa GalT-GFP-SNAP cells. H Quantifications of the immunoblots shown in Fig. 1E-F (IB: anti-EndoA3) confirm depletion efficiency of EndoA3 in HeLa GalT-GFP-SNAP cells. I Retrograde transport of ALCAM. Continuous BG-labelled anti-ALCAM antibody uptake for 4h at 37°C in LB33-MEL GalT-GFP-SNAP cells. Western blot analysis of LB33-MEL GalT-GFP-SNAP cells transfected (or not) with plasmids encoding free GFP (GFP+) or EndoA3-GFP (EndoA3+). Immunodetection with anti-SNAP, anti-EndoA3 and anti-α-Tubulin (loading control) antibodies. Quantification of the covalent IgG-SNAP-GFP-GalT complex is shown as fractions of the GFP+ condition (histogram). Data information: In (A-B), images are from a single experiment. In (C), data are pooled from six independent experiments. In (D-H), data are pooled from three independent experiments. In (I), data are pooled from two independent experiments. Data are presented as mean ± SEM. *P<0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. One sample t and Wilcoxon test. Source data are available online for this figure.

ICAM1 is an EndoA3-dependent endocytic cargo in cancer cells

A-B Supplementary representative live-cell TIRF images of EndoA3-GFP (stable) and ICAM1-mScarlet (transient) in HeLa (A) and LB33-MEL (B) cells. Time series show enlarged cropped areas extracted from Movie S5 (A) and Movie S6 (B). White arrowheads indicate dynamic co-distribution of both signals. Kymograph in (B) was made along dashed line in the enlarged cropped area corresponding to region 2 (Movie S7). Scale bars: 10 μm (full size image), 2 μm (enlarged cropped area). Data information: images are representative of two independent experiments.

EndoA3-dependent CIE mediates the uptake of immune synapse components ALCAM in LB33-MEL cells

A-B Confocal images of anti-ALCAM (white spots) internalization in LB33-MEL cells stained for actin (phalloidin, yellow) and nuclei (DAPI, blue). In (A), wild-type (WT), stably transfected with empty plasmid (Φ), or stably expressing EndoA3-GFP (EndoA3+) LB33-MEL cells were used. In (B), LB33-MEL EndoA3+ cells were transfected with negative control (siCtrl) siRNA or EndoA3 targeting (siEndoA3) siRNA. Quantification for anti-ALCAM internalization shown in Fig. 2D-E. Scale bars: 20 μm. C Western blot analysis of LB33-MEL cells stably expressing EndoA3-GFP (EndoA3+ LB33-MEL) transfected for 72h with siRNAs: negative control (siCtrl) or against EndoA3 (siEndoA3). Immunodetection made with anti-EndoA3 and anti-α-Tubulin (loading control) antibodies. Quantification of immunoblots shows depletion efficiency of EndoA3 (histogram). Data information: In (A-B), images are representative of three independent experiments. In (C), Western blot images are representative of four independent experiments, from which quantitative data are pooled. Data are presented as mean ± SEM. ****P < 0.0001 (One sample t and Wilcoxon test). Source data are available online for this figure.

Effect of EndoA3 depletion on morphological parameters and surface levels of several markers in LB33-MEL EndoA3+ cells or CD8 T cells

A Flow cytometry analysis of LB33-MEL cells stably expressing EndoA3-GFP (LB33-MEL EndoA3+) transfected with siRNAs: negative control (siCtrl) or against EndoA3 (siEndoA3) (related to Fig. 3B-C). Quantification (scatter plot) based on GFP signal and presented as normalized MFI of GFP relative to siCtrl condition. B-D Quantification of physical parameters (surface area, aspect ratio and roundness) of LB33-MEL EndoA3+ cells transfected with siRNAs: negative control (siCtrl) or against EndoA3 (siEndoA3). E-F Quantification of surface ICAM1 (E) and ALCAM (F) levels on LB33-MEL EndoA3+ cells transfected with siRNAs: negative control (siCtrl) or against EndoA3 (siEndoA3). Quantified by flow cytometry analysis and presented as normalized MFI of ICAM1/ALCAM signal in siCtrl condition. G-I Quantification of T cell surface activation marker levels (PD-1, G; CD137, H; TIM-3, I) after 4 days of stimulation by LB33-MEL EndoA3+ cells transfected with either control siRNAs (siCtrl) or siRNAs against EndoA3 (siEndoA3). Quantified by flow cytometry analysis and presented as absolute MFI. J Quantification of CD8 T cell proliferation after 4 days of stimulation by LB33-MEL EndoA3+ cells transfected with either control siRNAs (siCtrl) or siRNAs against EndoA3 (siEndoA3). Quantified by flow cytometry analysis and presented as proliferation index. K Quantification of CD8 T cell degranulation percentage after a 10 min stimulation by LB33-MEL EndoA3+ cells transfected with either control siRNAs (siCtrl) or siRNAs against EndoA3 (siEndoA3). Quantified by flow cytometry analysis as the percentage of CD107a-positive cells. L Western blot analysis of LB33-MEL EndoA3+ cells transfected with siRNAs: negative control (siCtrl), or against the retromer subunits (siVPS35 or siVPS26A) (related to Fig. 3G). Immunodetection with anti-VPS35, anti-VPS26A and anti-α-Tubulin (loading control) antibodies. Data information: In (A), data are pooled from twelve independent experiments. In (B-D), data are pooled from three independent experiments. In (E-F), data are pooled from eight independent experiments. In (G-J), data are pooled from four independent experiments. In (B-D), data are presented as median and quartiles. In (A, E-F), data are presented as mean ± SEM. ns, not significant. *P < 0.05, **P < 0.01, ****P < 0.0001 (A and G-K, paired t test; B-D, Mann Whitney test; E-F, one sample t test). In (L), the Western blot image is representative of three independent experiments. Source data are available online for this figure.

HLA-A*68012-expressing HeLa cells stimulate the activation of anti-MUM-3 CD8 T cells in the presence of MUM-3 peptide

A Quantification of IFNγ secretion (detected by ELISA) from anti-MUM-3 CD8 T cells, co-cultured for 20h with wild-type HeLa cells (WT) or stable HLA-A*68012-expressing HeLa cells (HLA+), with or without the presence of 300 ng/mL MUM-3 antigenic peptide. Data show absolute concentration of secreted IFNγ. B Flow cytometry analysis of anti-MUM-3 CD8 T cell proliferation after culture alone, or co-culture with LG-2 EBV feeder cells, or co-culture with both HLA-A*68012-expressing HeLa cells and LG-2 EBV feeder cells, with or without the presence of MUM-3 antigenic peptide, for 4 days. Proliferation was measured by the progressive dilution of CMFDA signal in T cells over generations. C Airyscan images of an immune synapse-like conjugate formed between a CD8 T cell (stained for CD45, yellow) and an adherent stable HLA-A*68012-expressing HeLa cell (stained for microtubules, green) transiently expressing ICAM1-mScarlet (red). White arrowheads indicate recruitment of ICAM1 to the vicinity of CD8 T cell. The fluorescence intensity profile was made along the dashed line region in enlarged cropped area. Scale bars: 20 μm (full size image), 5 μm (enlarged cropped areas). D Airyscan images of another immune synapse-like conjugate formed between a CD8 T cell (stained for CD45, yellow) and a stable HLA-A*68012-expressing HeLa cell (stained for actins, green) transiently expressing ICAM1-mScarlet (red) in suspension. The fluorescence intensity profile was made along the dashed line region. Cell segmentation (white contour) was made based on the actin staining for quantification (Fig. 5B). Scale bar: 5 μm. E Western blots analysis of stable HLA-A*68012-expressing HeLa cells transfected with siRNAs: negative control (siCtrl) or against EndoA3 (siEndoA3) (Fig. 5B). Immunodetection made with anti-EndoA3 and anti-clathrin heavy chain (CHC, loading control) antibodies. Data information: In (A-B), data and images are from single experiments. In (C-E), images are representative of three independent experiments.

Inhibition of EndoA3-mediated endocytosis and retrograde transport causes enlarged immune synapses

A-B Separate channels of Airyscan images from Fig. 4D and 4F, respectively. CD45 (yellow), EndoA3 (green), actin (red) and nucleus (blue) are shown. C, cancer cell; T, CD8 T cell. Scale bar: 5 μm. C-D Western blot analyses of stable EndoA3-GFP-expressing LB33-MEL cells transfected with siRNAs: negative control (siCtrl), siRNAs against EndoA3 (siEndoA3) (Fig. 4E) or against VPS26A (siVPS26A) (Fig. 4G). In (C), immunodetection was made with an anti-EndoA3 antibody. In (D), immunodetection was made with anti-VPS35 and anti-VPS26A antibodies. In both panels, anti-α-Tubulin was used as a loading control. Data information: images are representative of three independent experiments.