Cell surface protein aggregation triggers endocytosis to maintain plasma membrane proteostasis

Abstract

The ability of cells to manage consequences of exogenous proteotoxicity is key to cellular homeostasis. While a plethora of well-characterised machinery aids intracellular proteostasis, mechanisms involved in the response to denaturation of extracellular proteins remain elusive. Here we show that aggregation of protein ectodomains triggers their endocytosis via a macroendocytic route, and subsequent lysosomal degradation. Using ERBB2/HER2-specific antibodies we reveal that their cross-linking ability triggers specific and fast endocytosis of the receptor, independent of clathrin and dynamin. Upon aggregation, canonical clathrin-dependent cargoes are redirected into the aggregation-dependent endocytosis (ADE) pathway. ADE is an actin-driven process, which morphologically resembles macropinocytosis. Physical and chemical stress-induced aggregation of surface proteins also triggers ADE, facilitating their degradation in the lysosome. This study pinpoints aggregation of extracellular domains as a trigger for rapid uptake and lysosomal clearance which besides its proteostatic function has potential implications for the uptake of pathological protein aggregates and antibody-based therapies.

Fig. 1. Endocytosis of HER2-specific biparatopic antibody BS4 is not dependent on clathrin and dynamin but requires F-actin.

a Schematic representation of anti-HER2 biparatopic antibody BS4. The Trastuzumab (Tz) single-chain variable fragment (scFv) is attached to the N terminus of the heavy chain of 39 S IgG1 resulting in four antigen binding sites per molecule. Mutations in the Fc region (*) reduce binding to Fc gamma receptor. b, c Surface binding and endocytosis of dylight650-labelled monotopic antibodies Tz and 39 S and biparatopic BS4 (all at 3 µg/ml) after 1 h of uptake by confocal microscopy (b) and over a period of 6 h by flow cytometry (c), (means ± SD, n = 3 independent experiments). d, e Endocytosis of BS4 is independent of clathrin and dynamin. SkBr3 cells transfected with dominant-negative AP180ct (for clathrin-mediated endocytosis) and DynaminS45N N-terminally GFP-tagged expression constructs were incubated with dylight650-labelled BS4 and AlexaFluor546-transferrin for indicated times over a period of 4 h and analysed by flow cytometry. Endocytosis of transferrin and BS4 in cells from the same well expressing or not dominant-negative constructs (AP180ct⁺ AP180ct^-, DynS45N⁺, DynS45N⁻ in (d) and (e)), (means ± SD, n = 4 independent experiments). f BS4 uptake requires actin polymerisation. SkBr3 cells endocytosing pHrodo-labelled BS4 were treated with 10 µM CytoD, which was subsequently washed out as indicated, (means ± SEM, n = 6 from three independent experiments, two replicate wells each). Scale bars: 10 µm. Source data are provided as a Source Data file.

Fig. 2. Plasma membrane rearrangements during BS4 uptake are Rac1-dependent and morphologically resemble macroendocytosis.

a, b BS4 stimulates fluid-phase uptake. SkBr3 cells were co-incubated with dextran (70 kDa)-TMR and BS4-dylight650 for 10 min and after fixation analysed by confocal microscopy (a). The TMR channel is false-colour-coded in green, and individual cells are outlined. b Quantification of dextran (70 kDa)-TMR endocytosis in absence (mock) and presence of BS4 (dots represent measurements from individual cells, red lines indicate the median; n ≥ 50 cells from three independent experiments, ****P < 0.0001, two-tailed unpaired Student’s t test). c BS4-positive endocytic carriers exhibit dextran-filled lumina. d Concentration of surface-bound BS4 by inwards moving lamellipodium wave. Cells expressing GPI membrane-bound GFP (green) were imaged immediately after addition of BS4-dylight650 (magenta) (see also Supplementary Movies 3 and 4). 3D views (x, y, z dimensions) are shown on the right. The arrows indicate the position of the plasma membrane edge at the 40 s timepoint. e Rac1 re-localises to BS4-induced HER2 cell-surface aggregates. SkBr3 cells were incubated with BS4 for 2.5 min, fixed, stained for surface HER2 (red) and endogenous RAC (green) and analysed by confocal microscopy. BS4 endocytosis depends on Rac1 GTPase activity (f, g). SkBr3 cells transfected with Rac1 wt or dominant-negative T17N C-terminally GFP-tagged expression constructs were incubated with dylight650-labelled BS4 for 10 min. Samples were fixed, surface-bound BS4 was counterstained and cell sections analysed by confocal microscopy (f). Subtraction of surface from total BS4 signal shows endocytosed pool (BS4 uptake), transfected cells are outlined. Results are quantified in panel g, dots representing measurements from individual cells, red lines indicate the median; n ≥ 50 cells from three independent experiments, ns (non-significant) P > 0.05, ****P < 0.0001; one-way ANOVA with Dunnett’s multiple comparison test. BS4 is internalised via macroendocytic cups (h). SkBr3 cells expressing GFP-GPI (green) were imaged after addition of BS4-dylight650 (magenta) (see also Supplementary Movie 6). Maximum intensity projections are shown and 3D views with surface rendering of the plasma membrane (green) below each panel (see also Supplementary Movie 7). The arrows indicate the concentration of BS4 in endocytic vesicles after internalisation. Scale bars: 10 µm (a, e, f), 1 µm (c), 5 µm (d, h). Source data are provided as a Source Data file.

Fig. 3. BS4-induced aggregation is the trigger for HER2 endocytosis.

a BS4 clusters HER2 in the plasma membrane prior to endocytosis. SkBr3 cells were incubated with HER2-specific, dylight650-labelled biparatopic antibody BS4 or monotopic antibody 39 S for indicated times. After fixation, total HER2 in cell sections was stained using an antibody against the cytoplasmic domain and samples were analysed by confocal microscopy. b, c Aggregation of HER2 receptors by BS4 precedes endocytosis. After surface biotinylation, SkBr3 cells were incubated with BS4 for indicated times, surface remaining biotin was removed and samples lysed in low detergent buffer. A fraction was spun to separate soluble (Sup) from insoluble/aggregated proteins (Pellet). Protein in the remaining sample was solubilised (see “Methods” for protocol) and endocytosed biotinylated proteins concentrated on Streptavidin beads (uptake). Samples were assayed by immunoblot for the HER2 protein (b). c Time dependence of BS4-triggered aggregation and endocytosis of HER2 receptor is quantified (means ± SD, n = 3 independent experiments). d BS4-triggered cross-linking and endocytosis of HER2 is abrogated for mutant HER2 lacking the Tz-binding site. CHO cells expressing full-length HER2 receptor, either wt or lacking the Tz-binding site, were surface biotinylated and incubated with BS4 for 10 min. Samples were analysed as described in (b). e, f Cross-linking of both monotopic antibodies phenocopies the effect of BS4 on HER2 aggregation and endocytosis. Cells were incubated with equal amounts of indicated antibodies, with or without a cross-linking anti-human Alexa488 antibody (2nd) for 1 h. After fixation surface-bound, dylight650-labelled antibody was quenched and cell sections analysed by confocal microscopy (e) or antibody uptake quantified by flow cytometry (f) (means ± SD, n = 3 independent experiments, ns (non-significant) P > 0.05, **P = 0.0025, ****P < 0.0001, two-way ANOVA with Sidak’s multiple comparison test). Scale bars: 10 µm. Source data are provided as a Source Data file.

Fig. 4. Aggregation-dependent endocytosis re-routes transferrin receptor.

a Antibody-mediated aggregation and endocytosis of TfR. After surface biotinylation, SkBr3 cells ±  CytochalasinD (CytoD) were incubated with anti-transferrin-receptor antibody and cross-linking secondary (2nd) antibody for 30 min, surface remaining biotin was removed. Samples were processed, as described in the “Methods” section and assayed by immunoblot for transferrin receptor (TfR). b, c Cross-linked transferrin-receptor endocytosis requires actin polymerisation. SkBr3 cells ±  CytoD were incubated with dylight650-labelled anti-transferrin-receptor antibody 289 (red) ± secondary antibody-AlexaFluor488 (green) (2nd) for 30 min and analysed by confocal microscopy (b) or flow cytometry (c) means ± SD, n = 4 independent experiments, ns (non-significant) P > 0.05, ****P < 0.0001, two-way ANOVA with Sidak’s multiple comparison test. d, e Endocytosis of cross-linked transferrin receptor is independent of clathrin and dynamin. SkBr3 cells transfected with control (GFP) and dominant-negative AP180ct and DynaminS45N N-terminally GFP-tagged expression constructs were incubated with dylight650-labelled anti-transferrin-receptor antibody ± secondary antibody (2nd) for 30 min and analysed by flow cytometry (d) or confocal microscopy (e). Means ± SD, n = 3 independent experiments, ****P < 0.0001; two-way ANOVA with Dunnett’s multiple comparison test shown in (d). DynS45N transfected cells are outlined in lower panels (e). f, g Cross-linked transferrin-receptor endocytosis stimulates fluid-phase uptake. SkBr3 cells were co-incubated with dextran (70 kDa)-TMR (green) and dylight650-labelled anti-transferrin-receptor antibody (red) ± secondary antibody (2nd) for 30 min and analysed by confocal microscopy. Results are quantified in (g) dots represent measurements from individual cells, red lines indicate the median; n ≥ 50 cells from three independent experiments, ****P < 0.0001, two-tailed unpaired Student’s t test. h Cross-linked transferrin-receptor-positive endocytic carriers exhibit dextran-filled lumina. Cross-linked transferrin-receptor endocytosis is inhibited by 5‐(N‐ethyl‐N‐isopropyl)amiloride (EIPA) and dominant-negative Rac1 (i). HeLa cells transfected with dominant-negative Rac1 (T17N) for 16 h, or treated with 50 µM EIPA for 30 min, were incubated with dylight650-labelled anti-transferrin-receptor antibody ± secondary antibody (2nd) for 30 min. After fixation, surface-bound antibody was counterstained and samples analysed by confocal microscopy. Quantification of anti-transferrin-receptor antibody ± secondary antibody (2nd) endocytosis (after surface subtraction) is shown (dots represent measurements from individual cells, red lines indicate the median; n ≥ 50 from three independent experiments, ****P < 0.0001, one-way ANOVA with Sidak’s multiple comparison test). Scale bars: 10 µm (b, e, f), 1 µm (h). Source data are provided as a Source Data file.

Fig. 5. Aggregation-dependent endocytosis is induced by cross-linking molecules.

a WGA aggregates and induces endocytosis of glycosylated proteins in cells. After surface biotinylation, SkBr3 cells ±  CytochalasinD (CytoD) were incubated with WGA for 10 min, and remaining surface biotin was removed. Samples were harvested after processing, as described in the “Methods” section, assayed by immunoblot for receptor tyrosine-protein kinase erbB-2 (HER2), epidermal growth factor receptor (EGFR), sodium/potassium-transporting ATPase Alpha1 (Na/K-ATPase) and transferrin-receptor 1 (TfR). Numbers indicate relative levels of uptake normalised to the control sample. b WGA uptake is mediated by both clathrin/dynamin-dependent and actin-dependent endocytic pathways. Cells transfected with dominant-negative DynaminS45N N-terminally GFP-tagged expression construct were incubated ± CytochalasinD (CytoD) with dylight650-labelled WGA and analysed by confocal microscopy. Scale bars: 10 µm. Source data are provided as a Source Data file.

Fig. 6. BS4-triggered aggregation and actin-dependent macroendocytosis is HER2-specific.

a, b Depletion of HER2 and EGFR surface receptor levels by treatment with BS4 and EGF, respectively. SkBr3 cells were incubated for 10 min with BS4 or EGF, and ± CytochalasinD (CytoD) followed by cell-surface biotinylation and concentration on Streptavidin beads (surface). Samples were analysed by immunoblot for HER2, EGFR and HER3, TfR as negative controls (a). A quantitation of the results is shown in (b) means ± SD, n = 4 independent experiments, ns (non-significant) P > 0.05, **P = 0.0051, ***P = 0.0009, ***P < 0.0001; two-way ANOVA with Sidak’s multiple comparison test. c, d BS4 clusters HER2 but no other receptors in the plasma membrane resulting in HER2-specific endocytosis. SkBr3 cells were incubated with HER2-specific, dylight650-labelled biparatopic antibody BS4 for 2.5 min. After fixation, total receptor tyrosine-protein kinase ErbB-2 (HER2), receptor tyrosine-protein kinase ErbB-3 (HER3), epidermal growth factor receptor (EGFR), sodium/potassium-transporting ATPase Alpha1 (Na/K-ATPase) and Transferrin-receptor 1 (TfR) in cell sections were stained and samples were analysed by confocal microscopy. The spot overlap of fluorescent signal for BS4 and respective receptors is quantified in (d); means ± SD, n = 6 randomly chosen fields of view with a total of at least 50 cells per condition, ***P < 0.001, one-way ANOVA with Dunnett’s multiple comparison test. e Specific aggregation of HER2 receptor by BS4. SkBr3 cells were incubated with BS4 for indicated times and samples lysed in the low detergent buffer. Samples were spun to separate soluble (Sup) from insoluble/aggregated proteins (Pellet) and immunoblotted for indicated cell-surface receptors. f BS4 specifically endocytoses HER2 receptor. After surface biotinylation, cells were incubated with BS4 for indicated times. Protein was fully solubilised (see “Methods” for protocol) and endocytosed biotinylated proteins concentrated on Streptavidin beads (uptake). Samples were assayed by immunoblot for the HER2, HER3, EGFR, Na/K-ATPase and TfR. Scale bars: 10 µm. Source data are provided as a Source Data file.

Fig. 7. BS4-triggered ADE of HER2 is independent of Ras and PI3K.

BS4 induces HER2 autophosphorylation (a, b). SkBr3 cells ± dual EGFR/HER2 kinase inhibitor lapatinib (HER1/2i) were incubated with BS4 for 2.5 min as indicated. Samples were analysed by immunoblot for total HER2 and phospho-HER2 (pHER2) (a). SkBr3 cells were incubated with BS4-dylight650 (red) for 2.5 min, after fixation stained for phosphorylated tyrosine residues (p-Y, green) and analysed by confocal microscopy (b). Localisation of PIP3 to BS4-induced HER2 aggregates (c). SkBr3 cells were transfected with a vector expressing the PIP3 sensor PH-AKT-GFP (green) for 16 h treated or not with the PI3K inhibitor LY294002 (PI3Ki) followed by incubation with BS4-dylight650 (red) for 2.5 min. After fixation samples were analysed by confocal microscopy. BS4 uptake is reduced by HER2 kinase but not PI3-kinase inhibition (d). SkBr3 cells were treated with CytoD, lapatinib (HER1/2i) or LY294002 (PI3Ki) as indicated and incubated with BS4-dylight650 for 30 min. After fixation BS4 uptake was quantified by flow cytometry; means ± SD, n = 6 independent experiments, ns (non-significant) P = 0.252, ****P < 0.0001; one-way ANOVA with Dunnett’s multiple comparison test. BS4-induced ADE is independent of Ras (e, f). SkBr3 cells were transfected with a plasmid expressing a dominant-negative mutant (S17N) of Ras for 16 h followed by incubation with BS4-dylight650 for 30 min. After fixation surface-bound BS4 was counterstained and samples analysed by confocal microscopy. Subtraction of surface from total antibody signal yielded the endocytosed pool (e) with transfected cell outlined. Results are quantified in (f); dots represent measurements from individual cells, red lines indicate the median; n ≥ 50 cells from three independent experiments, ns (non-significant) P = 0.532; two-tailed unpaired Student’s t test. Localisation of VAV2 to BS4-induced HER2 aggregates (g). SkBr3 cells were incubated with BS4-dylight650 (red) for 2.5 min. After fixation, samples were stained for endogenous VAV2 and analysed by confocal microscopy. Scale bars: 10 µm (b, e, g), 3 µm (c). Source data are provided as a Source Data file.

Fig. 8. BS4-triggered ADE of HER2 critically requires VAV.

BS4-induced interaction of HER2 with VAV2 is phosphotyrosine dependent (a). SkBr3 cells ± lapatinib (HER1/2i) or LY294002 (PI3Ki) were incubated with BS4 for 2.5 min. BS4-HER2 aggregates were immunoprecipitated and samples analysed by western blot. VAV2 is recruited to BS4-HER2 aggregates via its SH2 domain (b). SkBr3 cells were transfected with VAV2 wt or lacking residues 665–772 (ΔSH2) C-terminally GFP-tagged expression vector (green) for 16 h followed by incubation with BS4-dylight650 (red) for 10 min. After fixation samples were analysed by confocal microscopy. BS4-induced ADE requires VAV2 GEF activity (c, d). SkBr3 cells were transfected with VAV2 wt, comprising amino acids 546–878, or lacking catalytic residues (Δ341–347) for 16 h followed by incubation with BS4-dylight650 (red) for 30 min. After fixation, surface-bound BS4 antibody was counterstained and samples were analysed by confocal microscopy. Subtraction of surface from total BS4 signal shows an endocytosed pool (BS4 uptake), transfected cells are outlined. Results are quantified in (d). VAV proteins are required for BS4 uptake (e, f). SkBr3 wt or VAV1-3 knockout (KO) cells were incubated with BS4-dylight650 (red) for 30 min. After fixation, the surface-bound BS4 antibody was counterstained (green) and samples were analysed by confocal microscopy (e). Subtraction of surface from total BS4 signal shows endocytosed pool (BS4 uptake). Results are quantified in (f). Rescue of BS4 uptake in SkBr3 VAV1-3 knockout cells by expression of VAV1-3 (g). SkBr3 wt or VAV1-3 knockout (KO) cells were transfected with C-terminally HA-tagged VAV1-3 expression vectors for 16 h followed by incubation with BS4-dylight650 (red) for 30 min. Samples were processed, analysed and BS4 uptake quantified as described in (e). Rescue of BS4 uptake in SkBr3 VAV1-3 knockout cells depends on the SH2 domain in VAV2 (h). SkBr3 wt or VAV1-3 knockout (KO) cells were transfected with VAV2 wt or a lacking residue 665–772 (ΔSH2) C-terminally GFP-tagged expression vectors for 16 h followed by incubation with BS4-dylight650 for 30 min. Samples were processed, analysed and BS4 uptake quantified as described in (e). Quantifications in (d, f, g, h): dots represent measurements from individual cells, red lines indicate the median; n ≥ 50 cells from three independent experiments, ns (non-significant) P > 0.05, ****P < 0.0001; one-way ANOVA with Dunnett’s multiple comparison test. Scale bars: 10 µm (b, c, e). Source data are provided as a Source Data file.

Fig. 9. Stress-induced cell-surface receptor aggregates are endocytosed by ADE.

a HER2 receptor aggregation and endocytosis upon chemical and physical stress. After surface biotinylation, SkBr3 cells ± CytochalasinD (CytoD) were washed in an acidic buffer (pH 3.5), heat shocked for 5 min (at 50 °C), or treated with BS4. After incubation of all samples at 37 °C for 30 min, surface remaining biotin was removed and samples lysed in low detergent buffer. A fraction was spun to precipitate insoluble/aggregated proteins (Pellet). Protein in the remaining sample was solubilised (see “Methods” for protocol) and endocytosed biotinylated proteins concentrated on Streptavidin beads (uptake). Samples were assayed by immunoblot for HER2. b Stress-induced aggregation-dependent endocytosis occurs independent of dynamin but requires actin polymerisation. SkBr3 cells transfected with RFP control, or dominant-negative DynaminS45N N-terminally RFP tagged expression vector (RFP channel displayed in white) were heat shocked for 5 min (at 50 °C) or washed in an acidic buffer (pH 3.5) ± CytochalasinD (CytoD) followed by 30 min incubation at 37 °C. After fixation surface HER2 was stained (false-colour-coded in red) before permeabilisation and staining for total HER2 (green channel). Cell sections were analysed by confocal microscopy and post-processed by subtraction of surface from total HER2 staining, which allowed for specific visualisation of intracellular HER2. c, d Heat stress induces fluid-phase uptake. HeLa cells ± CytochalasinD (CytoD) or ± 5‐(N‐ethyl‐N‐isopropyl)amiloride (EIPA) were incubated with dextran (70 kDa)-fluorescein incubated for 5 min at 50 °C followed by 30 min at 37 °C. After fixation samples were analysed by confocal microscopy (c). Quantification of heat stress-induced dextran uptake is shown in (d). Dots represent measurements from individual cells, red lines indicate the median; n ≥ 50 cells from three independent experiments, ****P < 0.0001; one-way ANOVA with Sidaks’s multiple comparison test. Scale bars: 10 µm (b, c), 1 µm insets panel c. Source data are provided as a Source Data file.

Fig. 10. ADE facilitates lysosomal degradation of endocytosed receptor aggregates.

a Time-dependent degradation of endocytosed antibody-receptor aggregates. Western blot analysis of steady-state HER2 protein and antibody levels after exposure to monotopic antibodies Tz and 39 S as well as biparatopic antibody BS4 for increasing length of time. b, c Stress-induced aggregates are degraded in the lysosome after endocytosis. SkBr3 cells ± CytochalasinD (CytoD) or BafilomycinA1 (BafA1) were surface biotinylated, heat shocked for 5 min (at 50 °C) and subsequently incubated for the indicated length of time at 37 °C. Samples were lysed in a low detergent buffer, insoluble/aggregated proteins precipitated by centrifugation, biotinylated receptors isolated as described in materials and methods and analysed by immunoblot for HER2, EGFR and TfR. Total cell lysates analysed for HER2 and actin are shown as controls. c The amount of intracellular HER2 aggregates at different times points is quantified. Means ± SD, n = 3 independent experiments, **P ≤ 0.0067, ***P = 0.0004; one-way ANOVA with Sidaks’s multiple comparison test. d Presence of extracellular receptor aggregates negatively affect cell growth. SkBr3 cells were incubated with monotopic antibody Tz or biparatopic antibody BS4 ± CytochalasinD (CytoD) for 4 h at 37 °C. Cell confluence was determined hourly using an Incucyte live-cell imager. Data normalised to the 4-h timepoint, at which antibodies and CytoD were removed, are displayed. The representative result of three independent experiments is shown. means ± SD, n = 8 replicate wells, significance shown for CytoD vs CytoD+BS4 (red) and CytoD vs CytoD+Tz (turquoise) 5–12 h time points ns P > 0.05, ****P < 0.0001; two-way ANOVA with Tukey’s multiple comparison test. Source data are provided as a Source Data file.