WDFY2 localization in the endocytic pathway. a SIM image showing representative WDFY2 localization in relation to different markers of the early endocytic pathway. APPL1 (gray) was used as a marker for early vesicles and EEA1 (red) marks early endosome. GFP-WDFY2 (green) localizes to an independent vesicle pool (inset 2) and to subdomains on EEA1-labeled endosomes (inset 3). There is only limited overlap with APPL1 endosomes (inset 1). Scale bar: 10 µm, insets 1 µm. Representative image of 14 cells. b SIM images showing the localization of GFP-WDFY2 in relation to mCherry-RAB5, mCherry-RAB4, RAB7 or mCherry-RAB11. Shown are representative images of 12(RAB5), 13(RAB4), 14(RAB7, RAB11) cells. Scale bar: 1 µm, insets: 0.5 µm. c Colocalization using Manders’ colocalization coefficient (MCC) analysis of hTERT-RPE1 cells stably expressing GFP-WDFY2 shows overlap between WDFY2 and RAB4 and RAB5, but only moderate overlap with RAB7 and RAB11. n = 3 experiments (Rab4, Rab5, and Rab11) and n = 4 (Rab7) with 70 cells per condition. Shown are individual experiments and the mean ± 95% CI. d Confocal image showing localization of endogenously-tagged WDFY2 (green) and EEA1 (red). Scale bar: 10 µm, insets: 1 µm. Representative image from 14 cells. Source data are provided as a Source Data file

WDFY2 localizes to endosomal tubules. a Sequential images showing GFP-WDFY2 localization to newly formed tubular structures and GFP-WDFY2 accumulation on the base of the tubules. Shown are frames from a timelapse sequence with images acquired every 2 s. Representative image from 200 observed endosomes. Scale bar: 0.5 µm. b dSTORM image showing GFP-WDFY2 localization to tubules. Representative image of five cells. Scale bar: 0.5 µm. c DNA-PAINT image showing GFP-WDFY2 and HRS localization to endosomes. Representative image of three cells. Scale bar: 0.5 µm. d Sequential images of endogenously NLAP-tagged WDFY2 showing localization to endosomes positive for mCherry-EEA1 and endosomal tubules. Shown are images from a time-lapse movie with images acquired every second. Representative image from 31 movies. Scale bar: 5 µm. e Localization of GFP-WDFY2 and mCherry-CORONIN1B on a tubulating endosome. CORONIN1B localizes to the base of WDFY2-positive tubules. The graph shows the normalized fluorescence intensity of GFP-WDFY2 and mCherry-CORONIN1B along the indicated line. Representative image from 50 observed endosomes. Scale bar 10 µm/1 µm (inset). f Localization of GFP-WDFY2 and mCherry-WASH on a tubulating endosome. WASH localizes to the base of WDFY2-positive tubules. The graph shows the normalized fluorescence intensity of GFP-WDFY2 and mCherry-WASH along the indicated line. Representative data from 50 observed endosomes. Scale bar 10 µm/1 µm (inset). g Localization of mCherry-WDFY2 and GFP-FAM21 on a tubulating endosome. FAM21 localizes to the base of WDFY2-positive tubules. The graph shows the normalized fluorescence intensity of mCherry-WDFY2 and GFP-FAM21 along the indicated line. Representative data from 50 observed endosomes. Scale bar 10 µm/1 µm (inset). h Quantification of WDFY2 tubules containing CORONIN1B, WASH, or FAM21. Results are shown as percentage of WDFY2 tubules positive for CORONIN1B, WASH or FAM21 per cell (20 cells per condition, CORONIN1B = 332 tubules, WASH = 338 tubules, FAM21 = 309 tubules). Plotted are the percentage of double-positive tubules per cell and the mean ± 95% CI. Source data are provided as a Source Data file

WDFY2 depends on PI3P for endosome localization. a Protein–lipid overlay assay using purified full-length WDFY2. WDFY2 binds with high selectivity to PtdIns3P. Shown is a representative plot from two experiments. b Deconvolved widefield image showing GFP-WDFY2 localization to endosomes. GFP-WDFY2-R315A, a mutation in the binding site for PtdIns3P, abolishes the localization to endosomes and the protein is cytosolic. Representative image from 10 cells per condition. Scale bar: 10 µm. c hTERT-RPE1 cells stably expressing GFP-WDFY2 was treated with SAR405, Wortmannin or DMSO (as a bleaching control) with a final concentration of 6 µM. Cells were imaged every 5 s for 15 min WDFY2 spots were quantified per time point (5 cells per condition). Shown are mean ± 95% CI. d Distribution of GFP-2xFYVE_WDFY2 and mCherry-2xFYVE_HRS on a tubulating endosome. A WDFY2-derived 2xFYVE probe shows a preference for tubulating membranes. The graph shows the normalized fluorescence intensity of GFP-2xFYVE_WDFY2 and mCherry-2xFYVE_HRS along the indicated line. Representative image from 50 observed endosomes. Scale bar: 1 µm. e Coomassie Brilliant Blue stained gel showing top fractions from liposome flotation assays using His-MPB-fused of 2xFYVE_WDFY2 and 2xFYVE_HRS with differently sized liposomes. Shown is one representative gel from three independent experiments. f Plots showing the relative binding of 2xFYVE_WDFY2 and 2xFYVE_HRS to differently sized liposomes. Data shown are derived from three experiments. Shown are mean ± S.E.M. Source data are provided as a Source Data file

WDFY2 interacts and colocalizes with VAMP3. a Volcano plot identifying enriched proteins isolated from cells expressing GFP-WDFY2 by GFP-trap affinity purification of cell lysates in comparison to GFP-expressing control cells. Label-free quantification (LFQ) ratios of GFP-WDFY2/GFP pulldowns were plotted against the p-value, with the permutation-based FDR threshold (<0.01) indicated by the black lines. b GFP affinity purification of GFP-VAMP3 and mCherry-WDFY2. Affinity isolation of GFP-VAMP3 coprecipitates mCherry-WDFY2, whereas isolation of GFP alone does not coprecipitate mCherry-WDFY2. One representative blot from three experiments is shown. c Representative SIM image showing GFP-VAMP3 and mCherry-WDFY2 localization to endosomes and endosomal tubules. Scale bar overview image: 1 µm, scale bar of inset image: 0.5 µm. Representative image from 13 observed cells. d Sequential images showing GFP-WDFY2 and PA-mCherry-VAMP3 localization to endosomes and endosomal tubule. The tubule pinches off and remains positive for VAMP3. Shown are frames from a time-lapse movie with images acquired every second. Representative image from 20 cells. Scale bar: 1 µm. Source data are provided as a Source Data file

WDFY2 controls intracellular VAMP3 distribution. a Deconvolved widefield images of WT and WDFY2(−/−) cells transiently expressing GFP-VAMP3 and mCherry-CORTACTIN. Scale bar: 10 µm (representative image of 30 cells per condition. Hexagonal superpixel image showing VAMP3 distribution in the cells shown in the upper panel. Mean intensities of hexagonal ROIs were extracted and the ROI filled with the corresponding mean value, thereby generating superpixels. b Distribution of VAMP3 from the nucleus to the leading edge. Cells were transfected with VAMP3 and stained for CORTACTIN to identify leading edges. A line ROI from the nucleus to the leading edge was drawn, and evenly spaced ROIs were automatically generated. Mean intensity of VAMP3 in each ROI was extracted, normalized for each cell, and then plotted. 30 cells for each condition, shown are mean and 95% CI. c TIRF micrograph of WT and WDFY2(−/−) cells transfected with VAMP3-pHluorin. Individual secretion events (summed over a 2-min interval) are indicated with a red circle. Representative image from 30 cells per condition. Scale bar: 10 µm. d Quantification of VAMP3-pHluorin secretion in WT and WDFY2(−/−) cells, as shown in (c). Shown are events from three experiments, 10 cells per experiment per condition. Secretion events per minute are shown normalized to cell area. Student’s unpaired t test, p = 0.00205. Shown are the median, quartiles (boxes), and 1.5 times the interquartile range (whiskers). *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. e Quantification of GFP-VAMP3 intensity in EEA1-positive endosomes in WT and WDFY2(−/−) cells. Plotted is the mean intensity within EEA1-positive endosomes per cell from four independent experiments (10, 12, 15, 10 cells (WT), 10, 10, 15, 10 cells (WDFY2(−/−)). Student’s unpaired t test, p = 0.00015. Shown are the median, quartiles (boxes), and 1.5 times the interquartile range (whiskers). *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. f Quantification of GFP-VAMP3 intensity in LAMP1-positive vesicles in RPE1 (WT) and WDFY2(−/−) cells. Plotted is the mean intensity within LAMP1-positive vesicles per cell from four experiments (10, 12, 15, 10 cells (WT), 10, 10, 15, 10 cells (WDFY2(−/−)). Student’s unpaired t test, p = 0.00083. Shown are the median, quartiles (boxes), and 1.5 times the interquartile range (whiskers). *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. Source data are provided as a Source Data file

WDFY2 controls MT1-MMP trafficking. a Confocal images showing localization of MT1-MMP and VAMP3 to GFP-WDFY2-positive endosome. Representative image from 10 cells. Scale bar: 10 µm, inset: 1 µm. b Confocal images showing localization of MT1-MMP and RAB4 to GFP-WDFY2-positive endosome. Representative image from 10 cells. Scale bar: 10 µm, inset: 1 µm. c TIRF micrograph of WT and WDFY2(−/−) cells transfected with pHluorin-MT1-MMP. Individual exocytic events (summed over two minutes) are indicated with a red circle. Representative image from 30 observed cells per condition. Scale bar: 10 µm, d Quantification of pHluorin-MT1-MMP exocytosis in WT and WDFY2(−/−) cells. Shown are events from three experiments, 10 cells per experiments per condition. Exocytosis events per minute are shown normalized to cell area. Student’s unpaired t test, p = 0.0026. Shown are the median, quartiles (boxes), and 1.5 times the interquartile range (whiskers). *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. e Confocal micrographs of RPE1 (WT) and RPE1 cells stably expressing GFP-WDFY2. Cells overexpressing WDFY2 accumulates MT1-MMP in endosome positive for WDFY2. Scale bar: 10 µm, insets: 5 µm. f Quantification of MT1-MMP spots based on high-content microscopy show a significant increase in total intensity of MT1-MMP in hTERT-RPE1 GFP-WDFY2 cells when compared to hTERT-RPE1 parental cells. Shown are quantifications from three independent experiments, cells in total: hTERT-RPE1: 3570, hTERT-RPE1 GFP-WDFY2: 5250. Plotted are individual experiments and the mean ± 95% CI. p = 0.02007. *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. g Quantification of pHlourin-MT1-MMP exocytosis in WDFY2(−/−) cells treated with siControl or siVAMP3. Shown are events from three experiments, ten cells per experiments per condition. Exocytosis events per minute are shown normalized to cell area. Student’s unpaired t-test, p = 0.0149. *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. Source data are provided as a Source Data file

WDFY2 controls extracellular matrix degradation. a Confocal micrographs showing degradation of a fluorescent gelatin layer, indicated by dark areas, by WT, WFDY2(-/-) and MT1-MMP-inhibitor treated WFDY2(−/−) cells. Scale bar: 5 µm. b Confocal micrographs showing degradation of a fluorescent gelatin layer, indicated by dark areas, in WDFY2(−/−) cells treated with control siRNA and MT1-MMP siRNA. Scale bar: 5 µm. c Quantification of gelatin degradation shown in (a). The graph shows average area of gelatin degradation per cell. n = 4 experiments, representing in total 815 cells (WT), 741 cells (WDFY2(−/−)) and 221 cells (WDFY2(−/−) + MMP inhibitor). Data points indicate the mean degradation per experiment, also shown is the mean ± 95% CI. Student’s unpaired t test, p = 0.00057. *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. d Quantification of gelatin degradation shown in (b). The graph shows average area of gelatin degradation per cell n = 3 experiments, representing 543 cells (WDFY2(−/−) siControl) and 455 cells (WDFY2(−/−) MT1-MMP siRNA) in total. Data points indicate the mean degradation per experiment, also shown is the mean ± 95% CI. Student’s unpaired t test, p = 1.028e−07. *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. e Quantification of gelatin degradation in WDFY2(−/−) cells treated with control siRNA and VAMP3 siRNA. The graph shows average area of gelatin degradation per cell n = 3 experiments, representing 618 cells (WDFY2(−/−) siControl) and 610 cells (WDFY2(−/−) VAMP3 siRNA) in total. Data points indicate the mean degradation per experiment, also shown is the mean ± 95% CI. p = 0.0002164. Student’s unpaired t test, *p < 0.05, **p < 0.01, ***p < 0.001, n.s. not statistically significant. Source data are provided as a Source Data file

WDFY2 controls invasive cell migration. a Optical sections (Δz = 10 µm) of WT and WDFY2(−/−) cells stained with Calcein-AM invading into fibronectin-supplemented Matrigel^TM. The red line indicates the z-axis threshold (50 µm) defining invading cells. b Orthogonal view of a 3D reconstruction of WT and WDFY2(−/−) invading into fibronectin-supplemented Matrigel^TM. The red line indicates the z-axis threshold (50 µm) defining invading cells. c Quantification of invasion of WT and WDFY2(−/−) cells into fibronectin-supplemented Matrigel^TM. Plotted points indicate the mean invasion per experiments (15 z-stacks per experiment, n = 3 experiments), also shown is the mean ± 95% CI. Student’s unpaired t test, p = 0.0162. *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. d Quantification of invasion of WT and WDFY2(−/−) cells into fibronectin-supplemented acid extracted type 1 Collagen. Plotted points indicate the mean invasion per experiments (15 z-stacks per experiment, n = 3 experiments), shown is the mean ± 95% CI. Student’s unpaired t test, p = 0.0389. *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. Source data are provided as a Source Data file

WDFY2 controls the invasivity of cancer cells. a Deconvolved widefield image showing localization of mCherry-WDFY2 to endosomes and endosomal tubules in MDA-MB231 cells. Representative image from 10 cells. Scalebar: 10 µm, inset: 1 µm. b Confocal image of MDA-MB231 cells showing colocalization of MT1-MMP (green) in EEA1 (blue) and mCherry-WDFY2-positive endosomes. Representative image from 8 cells. Scalebar: 10 µm, inset: 1 µm. c Quantification of invasion of MDA-MB231 cells transfected with nontargeting siRNA (siControl) and siRNA targeting WDFY2 (siWDFY2) and MDA-MB231 cells expressing siRNA-resistant GFP-WDFY2 transfected with non-targeting siRNA (siControl rescue) and siRNA targeting WDFY2 (siWDFY2 rescue) invading in fibronectin-supplemented Matrigel^TM. Plotted points indicate the mean invasion derived from 15 z-stacks per experiment, n = 3 experiments, shown is the mean ± 95% CI. Anova with Bonferroni post test p value: 0.003. *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. d Optical sections (Δz = 10 µm) of the conditions described in (c), cells stained with Calcein-AM invading into fibronectin-supplemented Matrigel^TM. The red line indicates the z-axis threshold (50 µm) defining invading cells. e Deconvolved widefield image showing localization of WDFY2 to endosomes and endosomal tubules in PC3 cells. Scalebar: 10 µm, inset: 1 µm. f Quantification of invasion of  PC3 and PC3 cells stably expressing GFP-WDFY2 cells into fibronectin-supplemented Matrigel^TM. Plotted points indicate the mean invasion per experiments derived from 15 z-stacks per experiment, n = 3 experiments, shown is the mean ± 95% CI. Student’s t test, p = 0.0029. *p < 0.05, **p < 0.01, ***p < 0.001,  n.s. not statistically significant. g Optical sections (Δz = 10 µm) of PC3 and PC3 stably expressing GFP-WDFY2 stained with Calcein-AM invading into fibronectin-supplemented Matrigel^TM. The red line indicates the z-axis threshold (50 µm) defining invading cells. Source data are provided as a Source Data file

WDFY2 controls cancer cell invasion. a Model of WDFY2 action in the endocytic system. In WT cells, WDFY2 interacts with VAMP3 and prevents sorting into bulk recycling tubules. This limits the amounts of VAMP3-positive recycling vesicles transporting MT1-MMP. In cells lacking WDFY2, more VAMP3 can be sorted into recycling vesicles, allowing more MT1-MMP to be recycled to the cell surface. b WDFY2 is frequently deep deleted in cancer samples. The graph shows data extracted from the CBioportal for Cancer Genomics, alteration frequency is grouped by cancer type and plotted