Figure 1. Breast cancer cells form protrusive structures during extravasation

(A?D) Lattice light-sheet microscopy of MDA-MB-231 cells expressing LifeAct Ruby injected intravenously in zebrafish xenografts. Yellow arrows point toward rolling BC cells in the blood. White arrows point to: (A and B) protrusions toward endothelium; (C) protrusions into the endothelium; (D) protrusions formed by tumor cell after extravasation. Scale bar, 10 ?m. Time in min.

(E) Representative images of Ctrl, srGAP1 KD #1, and srGAP1 KD #2 MDA-MB-231-GFP cells injected into Tg(kdrl:RFP) zebrafish embryos and imaged 24 h post injection (hpi). Scale bar, 20 ?m.

(F) Percent of extravasated MDA-MB-231 cells from zebrafish with extravasation events. n = 19 (Ctrl), 37 (srGAP KD #1), and 21 (srGAP1 KD #2) zebrafish. Mann Whitney U test, two-tailed. Data represented as a box and whisker plot, with median and interquartile range in the box and min and max as whiskers.

(G) Time-lapse images of MDA-MB-231 Ctrl and srGAP1 KD #1 cells in zebrafish vasculature. Yellow arrows point to protrusive structures. Scale bar, 20 ?m.

(H) Tiled images showing extravasation events at the tailbud of fish. MDA-MB-231 cells. (Top) Ctrl cells imaged 24 h post injection (hpi). (Middle) srGAP KD #1 cells imaged 20 hpi; (bottom) srGAP KD #1 cells imaged 24 hpi. Three representative fish tails shown. Scale bar, 100 ?m (left and middle panels), 25 ?m (right).

(I) Schematic of zebrafish xenograft setup.

Figure 2. srGAP1 depletion increases matrix degradation by forming more mature invadopodia

(A) Immunofluorescence demonstrating srGAP1, cortactin, and F-actin localization at invadopodia structures using direct STORM of MDA-MB-231 cells. Individual invadopodia noted by numbers and white boxes; arrows point to srGAP1 staining. Scale bars, 5 ?m (for cell), 100 nm (for invadopodia #1, 4), 200 nm (for invadopodia #2, 3). For invadopodia #1, respective curve fitting of a region through invadopodia (bottom right).

(B) Time-lapse images of control (Ctrl) or srGAP1 knockdown Lifeact-Ruby MDA-MB-231 cells using siRNAs (srGAP si #1, srGAP1 si #2) and respective matrix degradation on gelatin at 0, 4, and 8 h. Scale bar, 10 ?m.

(C) Matrix degradation of Ctrl, srGAP1 si #1, and srGAP1 si #2 MDA-MB-231 cells. Cells plated on fluorescent gelatin for 4 h. n = 114 (Ctrl), 116 (srGAP1 si #1), and 109 (srGAP1 si #2) fields quantified. Data from three independent experiments and represented as a violin plot, horizontal lines are median and interquartile range. Mann-Whitney U test, two-tailed.

(D) Representative images of Ctrl, srGAP1 si #1, and srGAP1 si #2 MDA-MB-231 cells plated on fibrillar collagen I overlaid on fluorescent gelatin for 16 h. Left: cortactin; middle: gelatin matrix; right: cortactin (red)-, F-actin (green)-, DAPI (blue)-labeled tumor cells on the matrix. Scale bar, 10 ?m.

(E) Matrix degradation of Ctrl, srGAP1 si #1, and srGAP1 si #2 MDA-MB-231 cells plated on fibrillar collagen I overlaid on fluorescent gelatin for 16 h. n = 119 (Ctrl), 101 (srGAP1 si #1), and 113 (srGAP1 si #2) fields quantified. Data are from three independent experiments and represented as a violin plot; horizontal lines are median and interquartile range. Mann-Whitney U test, two-tailed.

(F) Percent of Ctrl, srGAP1 si #1, and srGAP1 si #2 MDA-MB-231 cells with or without invadopodia at 4 h. n = 226 cells from 64 fields (Ctrl), 166 cells from 73 fields (srGAP1 si #1), 147 cells from 72 fields (srGAP1 si #2) pooled over three independent experiments. Chi-squared test.

(G) Representative images of Ctrl and srGAP1 si #1 MDA-MB-231 cells on gelatin for 4 h. Left: cortactin; right: Tks5 staining. Scale bar, 10 ?m. Inset scale bar, 1 ?m. Top left: cortactin; bottom left: Tks5; bottom right: gelatin, Top right: merge of cortactin and Tks5. Mature invadopodia are labeled with cortactin and Tks5; black spots on gelatin matrix represent matrix degradation (yellow arrows point to colocalization of invadopodia and matrix degradation).

(H) Number of mature invadopodia of Ctrl, srGAP1 si #1, and srGAP1 si #2 MDA-MB-231 cells at 4 h, n = 25 (Ctrl), n = 33 (srGAP1 si #1), n = 28 (srGAP1 si #2) fields. Data are from three independent experiments and represented as a violin plot; horizontal lines are median and interquartile range. Mann-Whitney U test, two-tailed.

(I) Time-lapse images of invadopodia lifetime at 0 and 30 min. (Left) Lifeact-Ruby Ctrl and srGAP1 si #1 MDA-MB-231 cells on fluorescent gelatin at 0 min. Scale bar, 20 ?m. Right panel: insets of mature invadopodia (red fluorescent puncta overlaid on black degradation spots) over time (0 and 30 min).

(J) Cumulative distribution of invadopodia lifetimes of Ctrl, srGAP1 si #1, and srGAP1 si #2 MDA-MB-231 cells. n = 122 invadopodia from 10 cells (Ctrl), 317 invadopodia from 11 cells (srGAP1 si #1), and 169 invadopodia from 11 (srGAP1 si #2) cells. Data from three independent experiments. Kolmogorov-Smirnov test.

Figure 3. In vitro and in vivo invasion are regulated by srGAP1 expression

(A) Representative images of Ctrl, srGAP1 si #1, and srGAP1 si #2 MDA-MB-231 cells 3 days after being seeded on top of collagen I plugs. Images taken at 15, 30, and 45 ?m in the z direction (deeper into collagen plug away from surface). Cells are labeled with phalloidin and DAPI, collagen I is fluorescently labeled. Scale bar, 50 ?m.

(B) x-z images of representative collagen I plugs from (A). Scale bar, 10 ?m.

(C) Maximum distance (>15 ?m) Ctrl and srGAP1 si #1 cells (left) or Ctrl and srGAP1 si #2 MDA-MB-231 cells (right) invade into the collagen I plug. Each value is a single cell/field. Left: n = 37 (Ctrl), 34 (srGAP1 si #1) cells. Right: n = 38 (Ctrl), 35 (srGAP1 si #2) cells. Data pooled over five independent experiments and represented as a violin plot; horizontal lines are median and interquartile range. Mann-Whitney U test, two-tailed.

(D) Schematic of intravital imaging setup.

(E) Three individual time frames from intravital imaging movies of Ctrl and srGAP1 KD #1 MDA-MB-231 primary tumors. Tumor cells in green, collagen fibers in red through second harmonic generation (SHG). Blue and yellow arrows indicate in vivo cell motility over time. One z slice shown at 0, 16, 28 min. Scale bar, 25 ?m.

(F) Average mean speed of tumor cells moving in ?m/min in vivo from (C). Data pooled from 3 Ctrl and 3 srGAP1 KD #1 mice, n = 94 (Ctrl) and 102 (srGAP1 KD #1) MDA-MB-231 tumor cells. Mann-Whitney U test, two-tailed. Data represented as a box and whisker plot, with median and interquartile range in the box and min and max as whiskers.

(G) Average path length of tumor cells that appear in a minimum of 10 frames of intravital imaging movies (t = 20?28 min). Data pooled from 3 Ctrl and 3 srGAP1 KD #1 mice, n = 27 (Ctrl) and 33 (srGAP1 KD #1) MDA-MB-231 tumor cells. Mann-Whitney U test, two-tailed. Data represented as a box and whisker plot, with median and interquartile range in the box and min and max as whiskers.

(H) Directional persistence of tumor cells that appear in a minimum of 10 frames of intravital imaging movies (t = 20?28 min). Data pooled from 3 Ctrl and 3 srGAP1 KD #1 mice, n = 27 (Ctrl) and 33 (srGAP1 KD #1) MDA-MB-231 tumor cells. Mann-Whitney U test, two-tailed. Data represented as a box and whisker plot, with median and interquartile range in the box and min and max as whiskers.

(I) Top: schematic of atomic force microscopy setup. Bottom: elasticity heatmaps of representative Ctrl, srGAP1 KD #1, and srGAP1 KD #2 MDA-MB-231 cells.

(J) Modulus of elasticity of Ctrl, srGAP1 KD #1, and srGAP1 KD #2 MDA-MB-231 cells using AFM. n = 100 (Ctrl), 200 (KD #1), and 200 (KD #2) measurements over 4, 8, and 8 cells, respectively. Mann-Whitney U test, two-tailed. Data represented as a box and whisker plot, with median and interquartile range in the box and min and max as whiskers.

Figure 4. srGAP1^low cells are capable of seeding the lung but have reduced metastatic outgrowth

(A) Number of seeding events (single MDA-MB-231 tumor cells and clusters) relative to primary tumor mass (grams). Data pooled over two Ctrl and two srGAP1 KD #1 mice. Three 450 × 450 ?m sections quantified per mouse. Unpaired t test, two-tailed, mean ± SEM.

(B) Number of seeding events (single MDA-MB-231 tumor cells and clusters) relative to primary tumor mass (grams). Data pooled over two Ctrl and two srGAP1 KD #2 mice. Three 450 × 450 ?m sections quantified per mouse. Unpaired t test, two-tailed, mean ± SEM.

(C) Representative images of MDA-MB-231 lung metastasis in Ctrl, srGAP1 KD #1, and srGAP1 KD #2 mice. Maximum intensity z projection, tumor cells in green and collagen fibers in blue through second harmonic generation (SHG). Scale bar, 100 ?m.

(D) Percent area of MDA-MB-231 metastasis per lung area measured (in mm²). One H&E section measured per mouse, n = 5 (Ctrl), 5 (srGAP1 KD #1), and 4 (srGAP1 KD #2) mice. Unpaired t test, two-tailed, mean ± SEM.

(E) Quantification of the proportion of MDA-MB-231 single cells relative to mass of primary tumor per lung section. Data pooled over two Ctrl and two srGAP1 KD #1 mice. Three 450 × 450 ?m sections quantified per mouse. Unpaired t test, two-tailed, mean ± SEM.

(F) Quantification of the proportion of MDA-MB-231 single cells relative to mass of primary tumor per lung section. Data pooled over two Ctrl and two srGAP1 KD #2 mice. Three 450 × 450 ?m sections quantified per mouse. Unpaired t test, two-tailed, mean ± SEM.

Figure 5. Solitary disseminated tumor cells have low srGAP1 expression

(A) Tissue staining of MDA-MB-231 single tumor cells and clusters in Ctrl mice lungs. Yellow arrow points to single cell. Top: GFP-labeled tumor cell, DAPI for nuclei. Bottom: srGAP1 staining. Scale bar, 10 ?m.

(B) srGAP1 intensity of MDA-MB-231 tumor cells and clusters in Ctrl mice lung (mean intensity in a defined area calculated for individual tumor cells and clusters). Data pooled over two Ctrl mice, three 450 × 450 ?m sections quantified per mouse. Mann Whitney U test, two-tailed. Scale bar, 10 ?m. Data represented as a violin plot; horizontal lines are median and interquartile range.

(C) Top: time-lapse imaging of MDA-MB-231 Ctrl and srGAP1 KD #1 cells expressing H2B-RFP and a DHB:Venus sensor for Cdk2 activity. Frames from live-cell imaging of cells plated on fibrillar collagen, time in hours. Scale bar, 10 ?m. Bottom: schematic representing cells expressing DHB:Venus sensor: nuclear expression is G0/G1, nuclear and cytoplasmic expression is S, and cytoplasmic expression is G2/M.

(D) Quantification of cytoplasmic to nuclear (C/N) ratio of DHB:Venus sensor for Cdk2 activity over time for MDA-MB-231 Ctrl and srGAP1 KD #1 cells from (C). n = 24 (Ctrl), 34 (srGAP1 KD #1) cells, two independent experiments. C/N ratio calculated for five frames (2 h). Data represented as mean ± SEM.

(E) Average cytoplasmic to nuclear ratio of MDA-MB-231 Ctrl and KD #1 cells expressing the DHB:Venus sensor for Cdk2 activity and H2B-RFP. C/N calculated from live-cell imaging (averaged over 2 h) of cells plated on fibrillar collagen. n = 24 (Ctrl), 34 (srGAP1 KD #1) cells, two independent experiments. Unpaired t test, two-tailed.

(F) Images of lungs from Ctrl and srGAP1 KD #1 MDA-MB-231 xenografts expressing H2B-RFP and a DHB:Venus sensor for Cdk2 activity. Scale bars, 50 ?m (left), 10 ?m (right).

(G) Number of MDA-MB-231 cells in the lung with nuclear DHB:Venus sensor expression relative to lesion area. Ten fields of view (255 × 255 ?m) per mouse, data pooled over one Ctrl and one srGAP1 KD #1 mouse, each point is a field. Mann-Whitney U test, two-tailed. Data represented as a box and whisker plot, with median and interquartile range in the box and min and max as whiskers.

Figure 6. srGAP1^low cells have increased Smad2 activation and TGF-?2 secretion

(A) Left: KEGG pathway enrichment analysis; right: reactome pathway analysis of differentially expressed genes in srGAP1 KD #1 MDA-MB-231 tumors in relation to Ctrl tumors, ^?p_adj < 0.05.

(B) Top: immunoblot of Smad2 in Ctrl and srGAP1 KD #1 MDA-MB-231 cells using an anti-Smad2/3 antibody. Loading control, anti-?-actin. Bottom: densitometry analysis of Smad2 protein levels in Ctrl and KD #1 cells, three biological replicates.

(C) Immunofluorescence of Smad2/3 in Ctrl and srGAP1 KD #1 MDA-MB-231 cells. Left: Smad2/3. Middle: phalloidin staining for F-actin. Right: DAPI for nuclear staining. Scale bar, 10 ?m.

(D) Smad2/3 nuclear intensity per cell. Left: n = 170 (Ctrl), 171 (srGAP1 KD #1) MDA-MB-231 cells. Right: n = 151 (Ctrl) cells, n = 169 (srGAP1 KD #2) MDA-MB-231 cells. Data over three independent experiments, Mann-Whitney U test, two-tailed. Data represented as a violin plot; horizontal lines are median and interquartile range.

(E) pSmad2/total Smad2 nuclear intensity ratio. Left: n = 90 (Ctrl), 94 (srGAP1 KD #1) MDA-MB-231 cells. Right: n = 80 (Ctrl), 99 (srGAP1 KD #2) MDA-MB-231 cells. Data over three independent experiments, Mann-Whitney U test, two-tailed. Data represented as a violin plot; horizontal lines are median and interquartile range.

(F) Differential expression of proteins from secretome analysis of conditioned medium from srGAP1 KD #1 MDA-MB-231 cells in relation to conditioned medium from Ctrl cells. Data visualized as a Venn diagram, three biological replicates and two technical duplicates.

(G) Pathway enrichment analysis of secretome (proteomics of conditioned medium) from srGAP1 KD #1 MDA-MB-231 cells in relation to conditioned medium from Ctrl cells.

(H) Expression of TGF-? ligands from proteomics data comparing conditioned medium from Ctrl and srGAP1 KD #1 MDA-MB-231 cells. Each data point is a biological replicate (averaged technical duplicates).

(I) Matrix degradation of MDA-MB-231 Ctrl cells plated on gelatin and stimulated with 10 ng/mL TGF-?2, 16 h. Left: cortactin. Middle: phalloidin stain for F-actin. Right: fluorescent gelatin matrix. Scale bar, 10 ?m.

(J) Matrix degradation of MDA-MB-231 Ctrl cells stimulated with 10 ng/mL TGF-?2 plated on gelatin, 16 h. Data from three independent experiments and represented as a violin plot, horizontal lines are median and interquartile range. n = 41 (Ctrl), 42 (TGF-?2-stimulated) fields quantified. Mann-Whitney U test, two-tailed.

(K) Nuclear p27 staining intensity of MDA-MB-231 Ctrl and srGAP1 KD #2 primary tumors. Top: p27. Bottom: DAPI for nuclei. Scale bar, 10 ?m.

(L) Nuclear p27 staining intensity of MDA-MB-231 primary tumors. Top: data pooled over two Ctrl and two srGAP1 KD #1 mice. n = 39 (Ctrl) cells, n = 48 (srGAP1 KD #1) cells. Two 246 × 246 ?m sections quantified per mouse. Mann-Whitney U test, two-tailed. Bottom: data pooled over two Ctrl and two srGAP1 KD #2 mice. One 388 × 388 ?m section quantified per mouse. n = 38 (Ctrl) cells, n = 46 (srGAP1 KD #2) cells. Mann-Whitney U test, two-tailed. Data represented as a violin plot; horizontal lines are median and interquartile range.

Figure 7. TGF-?2-mediated signaling regulates srGAP1^low phenotypes

(A) Matrix degradation of Ctrl and srGAP1 si #1 MDA-MB-231 cells with or without depletion of TGF-?2 by siRNA. Cells plated on fluorescent gelatin for 16 h. n = 51 (Ctrl), 47 (Ctrl + TGF-?2 siRNA), 52 (srGAP1 si #1), and 58 (srGAP1 si #1 + TGF-?2 siRNA) fields quantified. Data are from three independent experiments and represented as a violin plot, horizontal lines are median and interquartile range. Kruskal-Wallis test, corrected with Dunn?s test.

(B) Immunoblot of Ctrl and srGAP1 si #1 MDA-MB-231 cells with or without depletion of TGF-?2 by siRNA. Top blot: anti-srGAP1. Middle blot: anti-TGF-?2, ^? represents the TGF-?2 band. Bottom blot: loading control: anti-?-tubulin.

(C) Matrix degradation of Ctrl and srGAP1 si #1 MDA-MB-231 cells with DMSO or 5 ?M LY364947. Cells treated in serum-free medium with DMSO or 5 ?M LY364947 for 24 h, then plated on fluorescent gelatin for 16 h with DMSO or 5 ?M LY364947. n = 45 (Ctrl + DMSO), 38 (Ctrl +5 ?M LY364947), 52 (srGAP1 si #1 + DMSO), and 61 (srGAP1 si #1 + 5 ?M LY364947) fields quantified. Data are from three independent experiments and represented as a violin plot, horizontal lines are median and interquartile range. Kruskal-Wallis test, corrected with Dunn?s test.

(D) Representative images of cells seeding the lung after tail-vein injections. Left panel: GFP-labeled MDA-MB-231 tumor cells. Middle panel: fibrillar collagen (second harmonic generation). Right panel: merge, tumor cells in green and fibrillar collagen in red. Scale bar, 20 ?m.

(E) Area of tumor cells that have extravasated into the lung per field containing cells. GFP-labeled MDA-MB-231 Ctrl and srGAP1 si #1 MDA-MB-231 cells with or without depletion of TGF-?2 by siRNA. Cells injected into the tail-vein of mice. n = 31 (Ctrl), 31 (Ctrl + TGF-?2 siRNA), 29 (srGAP1 si #1), and 25 (srGAP1 si #1 + TGF-?2 siRNA) fields quantified, pooled over 5, 5, 4, and 5 mice, respectively. Data are represented as a box and whisker plot; horizontal lines are median and interquartile range. Kruskal-Wallis test, corrected with Dunnett T3 test.