Fig 2: SORLA promotes HER2 recycling. a, b Confocal microscopy images (a) and quantification (b) of HER2 staining after SORLA silencing (shSORLA #1 and shSORLA #4) in BT474 cells (n = 31 shCTRL, 25 shSORLA #1 and 20 shSORLA #4 cells from two experiments; analysis performed on 8-bit images; statistical analysis: Mann-Whitney test). c, d Confocal microscopy images (c) and quantification (d) of HER2 in vehicle- and primaquine-treated (60 min) BT474 cells (n = 34 (vehicle), 28 (0.1 mM primaquine) and 34 (0.3 mM primaquine) cells; analysis performed on 16-bit images; statistical analysis: Mann-Whitney test). e, f Microscopy analysis (e) and quantification (f) of AlexaFluor 568-labelled trastuzumab (Tz-568) internalization in MDA-MB-361 cells silenced with SORLA (siSORLA #3) or scramble (siCTRL) siRNA at the indicated time points (mean ± s.e.m; n = 64, 77, 86 and 64 siCTRL cells and 111, 83, 103 and 87 siSORLA #3 cells at the 0, 15, 30 and 60 min time points, respectively, from two independent experiments; statistical analysis: Mann-Whitney test; a.u. arbitrary units). g Microscopy-based HER2 recycling assay in control or SORLA siRNA-treated MDA-MB-361 cells. Labelled HER2 recycling back to the plasma membrane was monitored over 30 min after an internalization step (45 min) and imaged with a confocal microscope. Ratio of surface/internalized Tz-568 signal is displayed as box plots (n = 34 and 57 siCTRL cells and 45 and 47 siSORLA cells for 0 and 45 min time points, respectively, from two independent experiments; statistical analysis: Nonparametric Kruskal-Wallis). h Immunoblotting analysis of biotin-labelled cell-surface HER2 internalization in JIMT-1 cells overexpressing SORLA-GFP (or control GFP; GFP-CTRL), and quantification of internalized HER2 relative to total HER2 (data are mean ± s.d.; n = 4 independent experiments; statistical analysis: unpaired Student’s t test). i Quantification of HER2 recycling rate (% return of internalized biotinylated cell-surface HER2 back to the plasma membrane after 10 min) in JIMT-1 cells transfected with GFP-CTRL or SORLA-GFP following 30 min of endocytosis (data are mean ± s.d.; n = 3 independent experiments; statistical analysis: unpaired Student’s t test). Scale bars: 10 µm. Box plots represent median and IQR and whiskers extend to maximum and minimum values. Where micrographs are shown, these are representative of n = 3 independent experiments; ROI magnified region of interest

Fig 3: Schematic illustrating the role of SORLA in the oncogenic fitness of HER2 in cancer cells. SORLA, through interactions at its extracellular domain, is in a complex with HER2 and co-traffics with HER2, facilitating HER2 recycling to the plasma membrane to support HER2 downstream signalling. In the absence of SORLA, HER2 becomes localized to enlarged, partially dysfunctional lysosomes resulting in defective HER2 signalling and increased sensitivity to cationic amphiphilic drugs (CADs) like ebastine. End endosome, Lys lysosome

Fig 4: SORLA is highly expressed in HER2-amplified breast cancer cells and co-traffics with HER2. a Western blot analysis of SORLA and HER2 protein levels in breast cancer cell lines. a-tubulin is a loading control. b Quantification of SORLA cell-surface levels by FACS in MDA-MB-361, BT474 and JIMT-1 cells (n = 3 independent experiments; data are geo mean fluorescence intensity (MFI) ± standard error of mean (s.e.m.). c Confocal microscopy imaging of HER2 (magenta) and EEA-1 (green) in BT474, MDA-MB-361 and JIMT-1 cells. Co-localisation of the HER2 and EEA1 signals is indicated in white in the merged panels. d Endogenous SORLA (green), HER2 (magenta) and EEA-1 (white) staining in MDA-MB-361 cells (top panel). Endogenous HER2 (magenta) and VPS35 (white) staining in JIMT-1 cells expressing SORLA-GFP (green) (bottom panel). e Co-immunoprecipitation of endogenous SORLA with endogenous HER2 in MDA-MB-361 and BT474 cells. f Schematic of the SORLA protein domains and summary of the constructs used. g Co-immunoprecipitation of endogenous HER2 with different SORLA-GFP fragments in MDA-MB-361 cells. Scale bars: 10 µm. Where immunoblots and micrographs are shown, these are representative of n = 3 independent experiments; IB immunoblotting, IP immunoprecipitation. ECD extracellular domain, TM transmembrane domain, CD cytosolic domain

Fig 5: Silencing SORLA induces HER2 accumulation in dysfunctional lysosomes. a Immunofluorescence imaging of LAMP1 (white) and HER2 (red) in shCTRL and shSORLA MDA-MB-361 cells (n = 3 independent experiments). b Immunofluorescence imaging of LAMP1 (green) and CD63 (LAMP3; green) in MDA-MB-361 cells (blue is DAPI) after scramble (siCTRL) or SORLA (siSORLA #3 and siSORLA #4) siRNA silencing (n = 3 independent experiments). c Quantification of late endosomes/lysosome aggregation after SORLA silencing in MDA-MB-361 cells. LAMP1-positive structures = 5 µm2 were considered as lysosome aggregates (n = 73 siCTRL, 79 siSORLA #3 and 67 siSORLA #4 cells from three independent experiments; statistical analysis: Mann-Whitney test). d Immunofluorescence imaging and quantification of lysosomal aggregation in MDA-MB-361 cells treated with the indicated siRNA (n fields of view analysed, total cells = 11, 131 (siCTRL); 12, 182 (siSORLA #3); 11, 169 (siSORLA #3 + siHER2 #2); 10, 133 (siSORLA #4); 10, 121 (siSORLA#4 + siHER2 #2) from two independent experiments; statistical analysis: Mann-Whitney test). e Transmission electron microscopy imaging of lysosomes in siCTRL or siSORLA MDA-MB-361 and BT474 cells. Red arrows indicate the maturation defect in late endosome/lysosome structures (n = 3 technical replicates). f Flow cytometry analysis of the fluorescence signal in DQ Red BSA-loaded (24 h) MDA-MB-361 cells after scramble (siCTRL) or SORLA (siSORLA #3 and siSORLA #4) silencing. Cells loaded with DQ Red BSA (4 h) and treated with bafilomycin (or vehicle) are included as controls (bafilomycin blocks lysosome function) (mean ± s.d.; n = 5 independent experiments; statistical analysis: unpaired Student’s t test). g Cell viability assay to determine ebastine (48 h treatment) IC50 values in SORLA- or control-silenced BT474 and MDA-MB-361 cells (mean ± s.d.; n = 12, four technical replicates, three independent experiments; statistical analysis: unpaired Student’s t test). h, i Immunoblotting (h) and quantification (i) of cleaved PARP1 in ebastine-treated (15 µM, 48 h) siCTRL and siSORLA #3 or siSORLA #4 MDA-MB-361 cells. a-tubulin is a loading control (mean ± s.d.; n = 3 independent experiments; statistical analysis: unpaired Student’s t test). Scale bars: 10 µm (a, b, d) and 1 µm (e). Box plots represent median and IQR and whiskers extend to maximum and minimum values. Nu nucleus