Directional cell movement through tissues is controlled by exosome secretion

Abstract

Directional cell movement through tissues is critical for multiple biological processes and requires maintenance of polarity in the face of complex environmental cues. Here we use intravital imaging to demonstrate that secretion of exosomes from late endosomes is required for directionally persistent and efficient in vivo movement of cancer cells. Inhibiting exosome secretion or biogenesis leads to defective tumour cell migration associated with increased formation of unstable protrusions and excessive directional switching. In vitro rescue experiments with purified exosomes and matrix coating identify adhesion assembly as a critical exosome function that promotes efficient cell motility. Live-cell imaging reveals that exosome secretion directly precedes and promotes adhesion assembly. Fibronectin is found to be a critical motility-promoting cargo whose sorting into exosomes depends on binding to integrins. We propose that autocrine secretion of exosomes powerfully promotes directionally persistent and effective cell motility by reinforcing otherwise transient polarization states and promoting adhesion assembly.

Figure 1. Endolysosomal secretion controls cancer cell motility in vivo

(a) Western blots for the expression of synaptotagmin-7 (Syt7) and Rab27a in GFP-expressing HT1080 cells. Sc, scrambled control; KD, knockdown. (b) A schematic cartoon of the chick CAM model. GFP-expressing HT1080 human fibrosarcoma are injected either directly into the CAM or i.v. (c) Primary tumours formed on the CAM through subcutaneous injection. Note the lack of cell migration away from Syt7- and Rab27a-KD tumours. Representative images from ≥13 tumours for each condition. (d) Quantitation of the length of the migration front, defined as longest distance away from the primary tumour, as shown in image on left. (e) Migration away from colonies formed by extravasated cancer cells on the CAM through i.v. injection. Representative images from ≥26 fields. (f) Quantitation of colony number and size/image from thresholded images similar to shown example. Each image for KD is representative of two different targeting shRNA-cell lines (KD1, KD2) and all data are from ≥3 independent experiments. Box and whiskers plots are used where the box indicates the 25–75th percentile and whiskers indicate the 5–95th percentile and the black line indicates the median. *P<0.05; **P<0.01; ***P<0.001 compared with Sc using Mann–Whitney U-test. Scale bars, 200 μm.

Figure 2. Endolysosomal secretion is critical for persistent and fast migration

(a) Representative images of HT1080 cells migrating within the CAM. Representative cell tracks are shown. Scale bar, 200 μm. See also Supplementary Movies 1 and 2. (b) Depiction of total path length (Length, blue curved line) and productive migration (Displacement, red arrow) calculations from cell tracks. (c) Length and displacement plots. (d) Speed, calculated by length divided by time. Displacement rate, calculated by displacement divided by time. (e) Persistence index, calculated as displacement rate divided by speed. (f) Wind–Rose plots from cell tracks. Magnification of boxed area shown on right. Error bars, s.e.m. from four independent experiments (n>130 cells for each cell line). ***P<0.001 compared with Sc by Student's t-test.

Figure 3. Endolysosomal secretion controls protrusion dynamics in vivo and in vitro

(a–e) In vivo protrusion dynamics of HT1080 cells in the CAM (a) Representative images. Arrowheads indicate protrusions. Movies were recorded every 2 min for 60 min. Scale bar, 50 μm. See also Supplementary Movie 3. (b) Aspect ratio was calculated as length of major axis divided by length of minor axis of cell. Representative cell morphologies for high (3.47) and low (1.65) aspect ratios are shown at the left of the bar graphs. (c) Number of protrusions per cell was counted at each time frame (2 min) and averaged over movies (60 min). (d) Length of each protrusion. (e) Persistence of each protrusion was measured as the time between appearance and disappearance of each protrusion. n>30 cells for each cell line. (f,g) In vitro protrusion dynamics of HT1080 cells on tissue culture-treated dishes. (f) Representative kymographs. Note definitions of protrusion persistence (length of time that a protrusion lasts before retraction, Δx) and length of protrusion (Δy). Scale bar on Sc cell image indicates 30 μm. Scale bars on kymograph indicate time and distance axes. (g) Quantitation of number of protrusions/time, length of protrusion and protrusion persistence. n≥12 cells for each condition. Error bars, s.e.m. from three independent experiments. *P<0.05; **P<0.01; ***P<0.001 compared with Sc by Student's t-test.

Figure 4. Exosome secretion promotes cell migration.

(a) Average number of exosomes secreted from endolysosomal secretion-inhibited HT1080 cell lines from n≥3 experiments. (b) Western blots of exosome cargos. FLOT, flotillin. (c) Migration speed of Rab27a-knockdown (KD) HT1080 cells in exosome-depleted media±add-back of purified exosomes (Exo), as indicated on tissue culture-treated plates. (d) Speed of Rab27a-KD cells on exosome-coated plates. (e) Persistence index calculated from experiments shown in d. (f) Wind–Rose plots for data shown in d, Exo 0. Magnification of boxed area is shown at the right. (g) Western blot for FN-depleted exosomes. Numbers below blot represent relative intensity (Int.) of FN (normalized to FLOT) from that blot. WCL, whole-cell lysates; (h) Single-cell migration assays on FN-depleted exosome-coated plates. Error bars, s.e.m., from n≥29 cells for each cell line from three independent experiments. *P<0.05; **P<0.01; ***P<0.001 compared with Sc by Student t-test. ^#P<0.05; ^##P<0.01; ^###P<0.001 compared with same cell line on Exo 0 by Student's t-test. NS, not significant.

Figure 5. Density gradient-purified exosomes have bound FN and rescue Rab27a-KD cell motility defects

Extracellular vesicles were purified by ultracentrifugation. Microvesicle (Microvesicles) and exosome (Ultracentrifuged exosomes) fractions were analysed by NanoSight (a). Part of the Ultracentrifuged exosomes fraction was further purified by Optiprep density gradient centrifugation. (b) Western blot analysis of the fractions (Density gradient fractions 1–12) for exosome markers HSP70 and flotillin identified Fraction 6 as the major peak containing exosomes. NanoSight analysis of this fraction ((a): Density gradient fraction 6) showed a typical exosome size profile similar to the Ultracentrifuged exosomes profile, whereas Microvesicles fraction contained larger vesicles. Both FN and its receptor α5 integrin were present in the Fraction 6 exosome peak. FN was also present separate from exosome markers in Fraction 9. Conditioned media (CM) contained FN but the microvesicle (MV) fraction did not. T=Total cell lysate. Representative blots from n=2. (c) Western blot analysis of MVs, ultracentrifuge-purified exosomes (UC-Exo) and exosomes present in Fraction 6 density gradients (DG-Exo) from control cells cultured in normal media (Regular) or in FN-depleted media for 10 days. Equal numbers of vesicles were loaded in the gel lanes. Note that FN, HSP70 and CD63 were uniquely present in the exosome samples. Although the lanes are demarcated from each other due to intervening lanes not being shown, the regular and FN-depleted Exo lanes came from the same blot and are directly comparable for all antibodies. (d) The ability of various extracellular vesicles to rescue motility of Rab27a-KD cells was tested by coating 10 μg ml⁻¹ UC-Exo, microvesicles purified from an equivalent number of cells that produced 10 μg ml⁻¹ UC-Exo or 10 or 20 μg ml⁻¹ DG-Exo. Similar to results with UC-Exo (Fig. 4d,h), twice as many DG-Exo purified from FN-depleted cells were required to rescue Rab27a-KD speed as Regular DG-Exo. Data from ≥30 cells from ≥3 independent experiments for each condition. Error bars indicate s.e.m. ***P<0.001 compared with Sc by Student's t-test. ^#P<0.05; ^##P<0.01; ^###P<0.001 compared with the same cell line on Exo 0 by Student's t-test.

Figure 6. Adhesive exosome trails are left behind migrating cells

(a) Confocal image of mCherry-CD63 (red) and GFP-paxillin (green)-expressing HT1080 cell from a representative movie from n=17 movies on FN-coated (1 μg ml⁻¹) MatTek dishes. Arrows indicate co-localization of mCherry-CD63 with adhesions, whereas an arrowhead indicates localization of paxillin within endosomes. Magnification of boxed area is shown right. See also Supplementary Movie 5. (b) Image series from a representative epifluorescence movie (1 frame per 2 min, from n=10 movies) of a migrating HT1080 cell on FN-coated (1 μg ml⁻¹) MatTek dishes transiently transfected with pHLuorin-CD63 to show sites of exosome secretion. Note CD63 fluorescence in ruffles at the front of the cell (arrowheads) and in an adhesive trail left behind the cell. Zoom is brightened and shows adhesive trail. See also Supplementary Movie 6. (c) Image from a representative epifluorescence movie with a rapid frame rate (1 frame per 6 s, from n=20 movies) to capture the dynamics of pHLuorin-CD63 in protrusions. Filled arrowheads indicate fluorescence in ruffles, whereas open arrowhead points to adhesive trail. The arrow points to rear fluorescence near retracting fibres. Heat map-coloured intensity is shown right. Grayscale and coloured kymographs show protrusion over time across the perpendicular line drawn in heat map image. Note lack of enhanced fluorescence at flat protrusion membrane edge, which is outlined by drawn white lines. See also Supplementary Movie 7. Scale bars, 30 μm.

Figure 7. Exosomes promote adhesion assembly

(a) Image from TIRF movie of pHLuorin-CD63 (green) and mCherry-paxillin (red)-expressing HT1080 cell on FN-coated (1 μg ml⁻¹) MatTek dishes. Time series shows bursts of pHLuorin-CD63 fluorescence (arrowheads) preceding adhesion formation (indicated by *, red paxillin fluorescence visible at that time). Two adhesion formation events are, respectively, indicated with white and yellow markers. Representative of n=7 movies. See also Supplementary Movie 8. (b) Quantitation of per cent of adhesions with a pHLuorin-CD63 fluorescence burst before or coincident with their formation. (c) Quantitation of the number of minutes between pHLuorin-CD63 appearance and adhesion formation. (d) Representative time series of GFP-paxillin-marked adhesion assembly and disassembly for control (Sc) and Rab27a-KD HT1080 cells on FN-coated (1 μg ml⁻¹) MatTek dishes. Magnifications of boxed area are shown at right by heat map coloration to show intensity. 0 min indicates start of adhesion formation. White arrows indicate example adhesions and yellow arrows indicate the peak adhesion intensity. Asterisk indicates a newly forming adhesion. Note that a longer period of time is required for adhesion assembly in KD cells. (e,f) Box-and-whisker plots of adhesion assembly (e) and disassembly (f) rates. Each box shows the 25–75th percentile and whiskers indicate the 5–95th percentile and the black line indicates the median. Analysis from >30 adhesions for e and >40 adhesions for f from three to eight cells per condition. See also Supplementary Movie 9. (g) Western blot analysis of FN association with exosomes. GRGESP (100 μM; RGE, control) or GRGDSP (100 μM; RGD, integrin-binding) peptides were added to cultured HT1080 control cells (Sc) for 48 h before exosome isolation. Equal exosome numbers were loaded on western blots. Quantitations show FN intensity normalized to flotillin intensity as a loading control from n≥4 western blots. α5-KD, integrin α5-knockdown-HT1080. *P<0.05; **P<0.01; ***P<0.001 compared with Sc using Mann–Whitney U-test. Scale bars in a and d represent 30 μm.

Figure 8. Proposed model for exosome control of directionally persistent in vivo cell migration

(a) In the absence of exosome secretion, cells have unstable protrusions (blue arrows) and are unable to migrate effectively. (b) Secretion of exosomes (yellow) allows cells to effectively respond to directional cues by reinforcing nascent adhesions (orange) and stabilizing protrusions. (c) Positive feedback from exosomes promotes effective and directionally persistent motility, through adhesion enhancement and potentially additional unknown mechanisms. (d) ECM cargoes, such as FN, are carried on exosomes through a process involving endocytosis of ECM-integrin complexes, such as FN-α5β1, and subsequent sorting into MVEs. (e,f) Autocrine secretion of FN-coated exosomes at the leading edge may allow decoration of collagen fibrils with FN-bound exosomes that can then interact with cellular integrin receptors. Locally elevated concentration of FN-bound exosomes via secretion could facilitate integrin clustering and strong adhesion formation leading to accelerated migration. ECM carried on autocrine-secreted exosomes may be particularly effective at promoting cellular adhesion by matching matrix ligands to the adhesion receptor repertoire of the cell.