ARTICLE https://doi.org/10.1038/s41467-020-18081-9 OPEN ITGB3-mediated uptake of small extracellular vesicles facilitates intercellular communication in breast cancer cells Pedro Fuentes 1,2,Marta Sesé 1,2, Pedro J. Guijarro1,5, Marta Emperador 1,4,5, Sara Sánchez-Redondo3, Héctor Peinado 3, Stefan Hümmer 1,2 ✉ & Santiago Ramón y. Cajal 1,2 ✉ 1234567890():,; Metastasis, the spread of malignant cells from a primary tumour to distant sites, causes 90% of cancer-related deaths. The integrin ITGB3 has been previously described to play an essential role in breast cancer metastasis, but the precise mechanisms remain undefined. We have now uncovered essential and thus far unknown roles of ITGB3 in vesicle uptake. The functional requirement for ITGB3 derives from its interactions with heparan sulfate pro- teoglycans (HSPGs) and the process of integrin endocytosis, allowing the capture of extra- cellular vesicles and their endocytosis-mediated internalization. Key for the function of ITGB3 is the interaction and activation of focal adhesion kinase (FAK), which is required for endocytosis of these vesicles. Thus, ITGB3 has a central role in intracellular communication via extracellular vesicles, proposed to be critical for cancer metastasis. 1 Translational Molecular Pathology, Vall d’Hebron Research Institute (VHIR), Universitat Autònoma de Barcelona (UAB), Barcelona, Spain. 2 Spanish Biomedical Research Network Centre in Oncology (CIBERONC), Madrid, Spain. 3 Microenvironment & Metastasis Group, Molecular Oncology Program, Spanish National Cancer Research Centre (CNIO), Madrid, Spain. 4Present address: Tumor Biomarkers Group, Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain. 5 These authors contributed equally: Pedro J. Guijarro, Marta Emperador. ✉email: stefan.hummer@vhir.org; sramon@vhebron.net NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 M etastasis accounts for 90% of cancer deaths. It is a We—and others—have proposed that ITGB3 may be required for multistage process culminating in colonisation of a new the formation of macro-metastatic foci26,27. Following on from environment1, and each step relies on the tumour cells’ this, we sought to investigate the mechanisms of cellular ITGB3 interactions with their microenvironment. Extracellular vesicles in exosome biogenesis and uptake in relation to tumour (EVs) secreted from primary tumours are key in this metastasis. interaction2,3. EVs contain different molecular cargo (protein, Our work now demonstrates that EV-stimulated clonal growth RNA, or lipids) that can modify the local and distant environ- relies on ITGB3 in EV-receiving cells and that ITGB3 plays a ment, enabling primary tumours to evolve, establish pre- central and thus far unknown role in EV uptake. The requirement metastatic niches and metastasise4,5. for ITGB3 derives from the process of integrin endocytosis, EVs are a heterogeneous group of secreted membranous vesi- allowing the internalisation of EVs captured by ITGB3- cles including microvesicles, ectosomes, and exosomes6. They interacting HSPGs. This process is furthermore regulated by the have become valuable biomarkers in liquid biopsies, and existing ITGB3-interacting FAK, which is activated by EVs in an ITGB3- research has focused on their characterization in different cancer dependent manner and required for endocytosis of these vesicles. types7,8. However, the mechanisms underlying their biosynthesis, Thus, the transmembrane protein ITGB3 is required both for release from donor cells, and uptake into target cells remain extracellular recognition of EVs by interacting with HSPGs and poorly understood. A major difference among vesicle subtypes is for intracellular FAK activation. The stimulated endocytosis of their origin: while microvesicles and ectosomes bud from the integrins, known from the process of focal adhesion disassembly, plasma membrane, exosome formation begins on early endo- results in the coordinated cellular uptake of ITGB3- and HSPG- somes9. After maturation in multivesicular bodies through the associated EVs. Thus, the central role of ITGB3 in intracellular invagination of the endosomal membrane and formation of communication via EVs and the proposed function of EVs in intraluminal vesicles (ILVs), exosomes are released by fusion of cancer metastasis might explain the requirement for ITGB3 in ILVs with the plasma membrane10. The specific proteins incor- breast cancer metastasis. porated into ILVs are regulated mainly by the ESCRT (endoso- mal-sorting complex required for transport) of which there are 4: ESCRT-0, I, II, and III. Interference with protein function within Results these complexes has been reported to block exosome biogenesis11. ITGB3 is required for EV-induced colony formation. We pre- Once released from the donor cell, EVs induce cell signalling, viously demonstrated that ITGB3 is required for lung metastasis either by interacting with target cell–surface proteins or being formation26, but follow-up experiments revealed no defect in the taken up into the receiving cell6,9. This uptake is currently the initial homing of shITGB3 cells to the lung (Supplementary least understood step in vesicle-based intercellular communica- Fig. 1a–c). The next step in metastasis—formation of colonies tion, and proposed mechanisms range from passive membrane within the distant organ—involves clonal growth from a single fusion to active uptake via macropinocytosis or endocytosis12. cell28, which is dependent on factors secreted by other cells; in the Interestingly, this bears similarity to the mechanisms described laboratory, conditioned medium (CM) is generally used to gen- for virus endocytosis13, which begins when viral membrane gly- erate clonal cell lines29–32. To assess the role of secreted factors on coproteins bind to glycoprotein attachment factors, such as the colony-forming capacity of MDA.MB.231 cells, we measured heparan sulfate proteoglycans (HSPGs), on the target cell surface. the CM requirements for anchorage-dependent clonal growth HSPGs have also been demonstrated to be essential for uptake of (Fig. 1a, b). We seeded 500,000 cells, let them grow for 48 h, then EVs14–16. collected the medium. As expected, in the presence of CM, the Integrins are multifunctional heterodimeric cell–surface colony-forming capacity of MDA.MB.231 cells increased more receptor molecules that serve as entry receptors for a plethora of than 2.5-fold, while shITGB3 cells had only a 1.3-fold increase viruses. Uptake of herpes viruses has been reported to be (Fig. 1a, b). This difference can be attributed to the CM, since the dependent on αvβ3 integrin, and in Kaposi’s sarcoma-associated colony-forming capacity of parental and shITGB3 MDA. herpesvirus (KSHV), binding of the virus to the cell surface via MB.231 cells was not significantly altered after 7 days in normal interactions with heparan sulfate results in temporary interaction growth medium (Fig. 1a, b). Similar results were obtained with a between HSGPs and αvβ317. This activates the integrin, enabling second shRNA construct, excluding a possible off-target effect binding and activation of focal adhesion kinase (FAK) and sub- (Supplementary Fig. 2). To test if this effect was due to soluble sequent assembly of endocytic machinery17,18. Interplay between factors or EVs in the CM, we depleted the CM of EVs (100,000g HSGPs, integrins, FAK and endocytosis also regulates the turn- spin)33,34, and found that this, but not the removal of large EVs over of focal adhesion sites19–22. This turnover allows the only (10,000 g spin), completely negated the CM’s increased dynamic attachment of cells to surfaces and enables cell migra- colony-forming effect on the treated MDA.MB.231 cells (Fig. 1a, tion. Key for this process is integrin recycling, enabling local b; Supplementary Fig. 3a, b). Furthermore, the CM, but not the control of FAK activity. HSPGs of the syndecan family, particu- EV-depleted CM, derived from shITGB3 cells also increased the larly syndecan-4, play a crucial role in this process by regulating colony-forming capacity of MDA.MB231 cells. As in the case of integrin recycling. MDA.MB.231-derived CM, this effect was not observed in Early integrin studies in cancer research focused on their role shITGB3 cells. These results therefore indicate that ITGB3 is as cell adhesion receptors, though new roles have since been required for increased colony-forming capacity in vesicle- uncovered, particularly in the regulation of metastatic processes receiving cells only. by exosomes23,24. But while the role of integrins within exosomes During metastatic dissemination and homing within a different has been studied, little is known about their role in acceptor cells. organ, neoplastic cells are exposed to EVs derived from a variety The fundamental role of integrins in tumour progression—par- of cell types. In the metastatic mouse model, we previously ticularly metastasis—has long been recognised, but their exact observed that the capacity of MDA.MB.231 cells to form molecular functions remain incompletely understood25. We metastatic colonies in the lung is reduced in shITGB326. We recently identified integrin beta 3 (ITGB3) as a factor involved in therefore asked if, besides MDA.MB.231-derived EVs, EVs from tumour stress resistance and demonstrated its essential role in lung tissue cells could also stimulate colony growth of MDA. metastatic progression of triple-negative breast cancer (TNBC)26. MB.231 cells. We used lung-derived IMR90 fibroblasts: CM from 2 NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 ARTICLE a Conditioned medium b Non-CM shCON shITGB3 IMR90 MDA.MB.231 shCON cells 3.5 p-value < 0.0001 p-value 0.001 p-value < 0.0001 p-value 0.0032 p-value 0.0002 3.0 p-value 0.02 Conditioned medium wo EVs shCON shITGB3 IMR90 2.5 Relative to control 2.0 Conditioned medium 1.5 Non-CM shCON shITGB3 IMR90 shITGB3 cells MDA.MB.231 1.0 0.5 Conditioned medium wo EVs shCON shITGB3 IMR90 0.0 Non-CM CM CM CM CM CM CM shCON wo exos shITGB3 wo exos IMR90 wo exos shCON shITGB3 IMR90 MDA.MB.231 shCON cells MDA.MB.231 shITGB3 cells c p-value < 0.0001 d 75 80 p-value < 0.0001 60 50 EVs uptake p-value < 0.0001 EVs uptake p-value 0.0002 40 25 20 0 0 h h h h h h t6 t1 t3 t6 t1 t3 MDA.MB.231 Cells + WT EVs IMR90 cells + WT EVs MDA.MB.231 shITGB3 cells + WT EVs IMR90 cells + shITGB3 EVs MDA.MB.231 cells + shITGB3 EVs MDA.MB.231 shITGB3 cells + shITGB3 EVs Fig. 1 ITGB3 is required for EV-induced colony formation in MDA.MB.231 cells. a Representative pictures are shown for each cell population and condition. b Conditioned medium (CM) was collected from exponentially growing MDA.MB.231 shCON, shITGB3, or IMR90 cells. In the case of vesicle- depleted CM (CM without exosomes), the same CM was split into two parts: one was used directly and the other was depleted of vesicles by ultracentrifugation before use, to establish the clonal cell growth response to the CM. Data are normalized to the non-conditioned control medium (DMEM). c, d Flow cytometry analysis for MDA.MB.231 shCON, MDA.MB.231 shITGB3 cells and IMR90 cells after incubating with 2–5 µg mL−1 fluorescently labelled EVs derived from both MDA.MB.231 shCON and MDA.MB.231 shITGB3 cells at different time points. Source data are provided as a Source Data file.(*p value <0.05, **p value <0.01, ***p value <0.001, ****p value <0.0001; n = 7 for non-CM condition; n = 4 for CM shCON, CM shITGB3, and CM IMR90; n = 3 for CM wo EVs shCON and CM wo EVs IMR90. n = 3 for panel 1c, d). Data are represented as mean ± SD in (b–d). Statistical analysis including two-way ANOVA multiple comparisons was carried out using GraphPad Prism 6.01. these cells, but not EV-depleted CM, stimulated colony growth in ITGB3 is required for vesicle uptake. An observed inability of MDA.MB.231 cells in an ITGB3-dependent manner (Fig. 1a, b). shITGB3 cells to respond to the presence of EVs prompted us to Taken together, these results show that the EV fraction of CM speculate that these cells may have defective vesicle uptake and in promotes clonal growth in MDA.MB.231 cells and indicates an turn defective delivery of the cargo molecules responsible for essential role of ITGB3 in vesicle-receiving cells. increased colony-forming capacity. To test this, we measured EV NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 uptake. EVs derived from shCON or shITGB3 cells were isolated In line with previous reports14,42,43, EV uptake at 4 °C was also by ultracentrifugation and fluorescently labelled with PKH26, and severely blocked in MDA.MB.231 cells (Supplementary Fig. 7), their uptake into both cell lines was measured by FACS at various suggesting that the underlying mechanism may be an energy- time points after incubation. As shown in Fig. 1c, in the shCON dependent process such as endocytosis. To test this, we blocked cell line, EVs were taken up over time, and at 6 h up to 60% of DYNAMIN activity using Dyngo-4a44 and blocked clathrin cells were EV-positive. The portion of EV-positive cells (25% at 6 activity using Pitstop-245. As endocytosis is an essential cell h) was lower in shITGB3 cells than in shCON cells. These results process, inhibitor treatments were reduced to 30 min, followed by were confirmed by the use of an independent shRNA construct incubation with fluorescently labelled vesicles. Measurements of targeting ITGB3 (Supplementary Fig. 4). To exclude the possi- cell fluorescence intensity after 3 h revealed that vesicle uptake in bility of a cell-line-specific function of ITGB3, we generated a MDA.MB.231 cells was significantly reduced in Dyngo-4a but not stable shRNA-mediated knockdown of ITGB3 in MCF7 cells and Pitstop-2 treated cells. Depletion of DYNAMIN 2 by shRNA demonstrated that EV uptake was also dependent on ITGB3 (shDyn2) or the overexpression of a DYNAMIN 2 dominant (Supplementary Fig. 5). Importantly, the origin of the vesicles— negative mutant (Dyn2-44K) confirmed the results obtained for either from shCON or shITGB3 cells—did not affect their uptake Dyngo-4a treatment (Fig. 2c–e). into shCON and shITGB3 cells (Fig. 1c, d). These results were Interestingly, the reduction of EVs in Dyngo-4a treated cells was confirmed in IMR90 cells, which incorporated EVs from both cell accompanied by an increase in the surface abundance of ITGB3 lines with a similar efficiency (Fig. 1d). To further support the (αvβ3), as determined by FACS (Supplementary Fig. 8a). These role of ITGB3 in EV uptake into receiving cells, we performed results are in line with previous reports that examined recycling of confocal microscopy of PKH26-labelled EVs on PFA-fixed and integrins as a key mechanism to regulate their function at the cell ITGB3 (αvβ3)-stained cells. These experiments were carried out surface21. In addition, we detected co-localization of ITGB3 (αvβ3) with shITGB3 cell-derived EVs to avoid staining of EV-localized with EEA1 on confocal microscopy, and ITGB3 was co-localized to ITGB3. Under these experimental conditions, we detected co- internalized EVs derived from shITGB3 cells (Supplementary localisation between cellular ITGB3 and EVs at the surface of Fig. 9). Altogether, these results imply that not solely the presence non-permeabilized cells (Supplementary Fig. 6). of integrin beta 3 on the cell surface, but the active internalisation These results demonstrate that the uptake of a significant of the integrin in a DYNAMIN-dependent manner appears to be portion of EVs relies on ITGB3 in the recipient cell. However, not crucial for EV uptake. all vesicles were dependent on ITGB3 (approximately 50%). This As described above, the uptake of a significant fraction of EVs may have been due to incomplete knockdown of ITGB3, into MDA.MB.231 cells relies on ITGB3, HSPGs and DYNAMIN. redundant pathways for vesicle uptake, or ITGB3 being required To determine whether the remaining vesicles detected by FACS for the uptake of certain vesicle subtypes only. As we could not analysis were taken up into the recipient cells or attached to the rigorously exclude any of these possibilities, we looked further at cell surface, we included an additional wash step with citric acid- the mechanism of ITGB3-dependent EV uptake. containing buffer (CAB wash) prior to FACS analysis46. In shCON and shITGB3 cells, CAB wash only slightly reduced the EV-derived fluorescence signal (Fig. 3a), indicating that vesicles ITGB3-dependent endocytosis of EVs requires HSPGs. Having do not accumulate at the cell surface before being taken up. established that ITGB3 is fundamental for EV uptake in breast Supporting this, heparin treatment, which already interferes with cancer cell lines, we explored the underlying mechanisms. First, the capture of EVs at the cell surface, resulted in a similar we analysed how EVs might be captured at the cell surface. The reduction in EVs following the CAB wash (Fig. 3b). Furthermore, described interactions of ITGB3 with other proteins include its interference with the DYNAMIN-dependent internalisation of selective binding to syndecans through the extracellular FERM EVs through Dyngo-4a treatment did not result in accumulation domain35–40. Syndecans are a family of heparan sulfate-decorated of EVs at the cell surface (Fig. 3b). Thus, EV recognition and cell–surface proteoglycans that recognise “heparin-binding” uptake appear to be intimately linked, preventing the accumula- motifs. Treatment of different cancer cell lines with heparin, a tion of EVs at the cell surface when uptake is impaired. highly sulfated glycosaminoglycan mimetic of heparan sulfate, To provide further evidence for these results independent of has been shown to block EV uptake. To test if this occurred in the FACS-based measurements, we assessed EV uptake into MDA.MB.231 cells, we pretreated MDA.MB.231 shCON and recipient cells on confocal microscopy. Cells treated with PKH26- MDA.MB.231 shITGB3 cells with fluorescently labelled vesicles labelled EVs were fixed, and alpha-tubulin staining of permea- derived from MDA.MB.231 cells. FACS analysis revealed that bilized cells was used for 3D reconstruction (Fig. 3c, d), to heparin blocked vesicle uptake to a similar degree as observed discriminate between surface-bound and internalized vesicles. with shITGB3 knockdown. Furthermore, heparin treatment of Analysis of fluorescently labelled particles firstly showed a strong shITGB3 cells had no additive effect on vesicle uptake, indicating reduction in shITGB3 and shDyn2 cells (Fig. 3e). Furthermore, that ITGB3 and HSPGs act on the same pathway (Fig. 2a). To the ratio of surface-bound and internalized vesicles resembled our strengthen our finding that HSPGs are required for EV uptake, results obtained with the CAB wash in the FACS-based cells were treated with heparinase prior to the addition of measurements of EV uptake (Fig. 2a). Our data therefore suggest PKH26-labelled EVs. As occurred with heparin treatment, EV that ITGB3 is essential for EV uptake in recipient cells. uptake was reduced by approximately 50% (Fig. 2b). As synde- Furthermore, analysis of the PKH26-labelled EV uptake in cans have been reported to directly interact with ITGB3, it is shDyn2 cells revealed a strong reduction in the number of tempting to speculate that this interaction provides a physical link fluorescent particles (Fig. 3c–e); the proportion of surface-bound between exosomes and ITGB3. and internalized vesicles was similar to control and shITGB3 cells. Next, we asked how those captured vesicles are internalized. Thus, FACS analysis combined with CAB-wash and confocal Previously proposed mechanisms for EV uptake and release of microscopy demonstrate that, when DYNAMIN-dependent EV cargo into different target cell lines include fusion of vesicles with uptake is impaired, EVs do not accumulate at the cell surface. the plasma membrane and uptake of entire vesicles by Therefore, the interaction of EVs with the cell surface appears to endocytosis, phagocytosis or macropinocytosis9,10,41. We there- be a highly dynamic process; if attachment of EVs is not fore began by exploring which mechanisms of vesicle uptake were intimately linked to endocytosis, EVs may detach again from the used in MDA.MB.231 cells. cell surface. 4 NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 ARTICLE a 1.25 p-value 0.0010 b 1.25 p-value < 0.0001 p-value 0.0021 p-value < 0.0001 p-value 0.001 p-value < 0.0001 1.00 1.00 0.75 0.75 MFI MFI 0.50 0.50 0.25 0.25 0.00 0.00 shCON cells + EVs shCON cells + EVs shCON cells + Heparin + EVs shCON cells + heparinase + EVs shITGB3 cells + EVs shITGB3 cells + EVs shITGB3 cells + Heparin + EVs shITGB3 cells + heparinase + EVs c d p-value 0.02 p-value 0.01 p-value 0.0015 1.25 1.25 p-value < 0.0001 e Dyn2 Dyn2 p-value < 0.0001 CON WT K44A shDyn2 1.00 1.00 GFP-Dynamin 2 127 KDa 0.75 0.75 Dynamin 2 100 KDa MFI 0.50 0.50 ACTIN 42 KDa 0.25 0.25 0.00 0.00 shCON cells + EVs shCON cells + EVs shITGB3 cells + EVs Dyn2 WT cells + EVs shCON cells + EVs + Dyngo4a Dyn2K44A cells + EVs shCON cells + EVs + PitStop2 shDyn2 cells + EVs Fig. 2 Endocytic uptake of EVs depends on ITGB3, HSPGs and DYNAMIN activity. Internalization of 2–5 µg mL−1 PKH26-labelled EVs derived from MDA.MB.231 cells by MDA.MB.231 shCON or MDA.MB.231 shITGB3 cells, measured by FACS. a Uptake of vesicles was measured after pretreatment with 10 μg mL−1 heparin as a competitive inhibitor for cell–surface heparan-sulfated proteins for 30 min. b Uptake was measured in cells that were untreated or pretreated with 1.2 mIU mL−1 heparinase I and 0.6 mIU mL−1 heparinase III for 3 h before EV incubation. c Uptake of vesicles was measured after pretreatment with 20 μM Dyngo4a or 10 μM Pitstop 2 for 30 min in shCON cells. d Uptake of vesicles measured in MDA.MB.231 cells infected with Dyn2WT− Dyn2K44A− and shDyn2 constructs. e Representative immunoblot showing endogenous and exogenous levels of DYNAMIN 2 protein. Source data are provided as a Source Data file. (*p value < 0.05, **p value < 0.01, ***p value < 0.001, ****p value < 0.0001, n = 3 for panel 2a, d; n = 4 for panel 2c; n = 7 for panel 2b). Data are represented as mean ± SD in (a–d). Statistical analysis including two-tailed unpaired Student’s t test data was carried out using GraphPad Prism 6.01. ITGB3-dependent alterations in the EV secretome. Having vesicle populations by size. In line with previous reports, vesicles established the involvement of ITGB3 in the endocytosis-driven between 150 and 200 nm comprised the peak fraction of MDA. uptake of EVs, we wondered how this might be reflected in the MB.231 cells47. This vesicle population was reduced in shITGB3 secretome of shITGB3 cells. To test this, shCON and shITGB3 cells, while those measuring 50–125 nm were significantly MDA.MB.231 cells were maintained in low-serum-containing increased (Fig. 4b, c) (Supplementary Fig. 10a–c). We confirmed medium (exosome-depleted 0.5% fetal bovine serum (FBS)) for these results on cryo-electron microscopy (Fig. 4d). In summary, 48 h and vesicles in the cell culture supernatant were isolated by vesicles in the supernatant of shITGB3 displayed an altered size differential ultracentrifugation (Fig. 4a). The vesicles were char- distribution. acterized by Nanosight and cryo-electron microscopy (Fig. 4) and To test if these changes were accompanied by alterations in their proteome was determined by liquid chromatography–mass the proteome of the respective EVs, we performed LC–MS/MS. spectrometry (LC–MS)/MS and Western blot (Fig. 5). The overall We identified 3204 proteins(Supplementary Data 1), 1195 of abundance of EVs, measured by Nanosight and normalized to the which exhibited a statistically significant difference in abun- number of EV-producing cells, revealed no significant difference dance between EVs isolated from shCON and shITGB3 cells between shCON and shITGB3 cells (Fig. 4b–d). Similar results (q.value < 0.05) (Fig. 5a, b). The fold change in protein were obtained when the protein quantity in the EV fraction after abundance was calculated using the shCON condition as a ultracentrifugation was measured by bicinchoninic acid (BCA) reference, i.e., log2(shITGB3/shCON). The heat map in Fig. 5a protein assay (Supplementary Fig. 10c). However, we did detect summarizes the results from the independent biological clear differences in EV quantity when we analysed different replicates and highlights the striking differences in the NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 a 1.25 p-value 0.0003 b 1.25 p-value 0.0007 p-value < 0.0001 ns p-value < 0.0001 1.00 1.00 0.75 ns 0.75 MFI MFI 0.50 0.50 + EVs + EVs + citric washes 0.25 shCON cells + EVs 0.25 + Dyngo + EVs shCON cells + EVs + citric washes + Dyngo + EVs + citric washes shITGB3 cells + EVs + Heparin + EVs shITGB3 cells + EVs + citric washes + Heparin + EVs + citric washes 0.00 0.00 c TUBULIN EVs EVs/TUBULIN d e MDA.MB.231 shCON MDA.MB.231 1.0 shCON TOTAL EV 0.8 EV IN EV OUT Signal/cell relative to control 0.6 MDA.MB.231 shITGB3 0.4 0.20 0.15 0.10 MDA.MB.231 shDyn2 0.05 0.00 ShCON ShITGB3 ShDyn2 Fig. 3 Interaction of EVs with the cell surface is a highly dynamic process. Internalization of 2–5 µg mL−1 PKH26-labelled EVs derived from MDA. MB.231 cells by MDA.MB.231 shCON or MDA.MB.231 shITGB3 cells, measured by FACS. a Uptake of vesicles was measured by FACS after subjecting the cells to acid washing. b Internalization of 2–5 µg mL−1 PKH26-labelled EVs derived from MDA.MB.231 cells by MDA.MB.231 WT cells, pretreated with 20 μM Dyngo4a and pre-treated with 10 μg mL−1 heparin, with or without subjecting the cells to acid washing. c Representative confocal pictures used for 3D reconstruction. d Representative 3D reconstruction picture of MDA.MB.231 shCON cells. Bar represents 5 µm. e Analysis of EV internalization in MDA. MB.231 shCON, shITGB3 and shDyn2 cells after 3D reconstruction. Source data are provided as a Source Data file. (***p value < 0.001, ****p value < 0.0001, n = 3). Data are represented as mean ± SD in (a, b, e). Statistical analysis including two-way ANOVA multiple comparisons was carried out using GraphPad Prism 6.01. proteome of vesicles derived from shCON and shITGB3 cells. line-specific phenomenon. Finally, we performed iodixanol These differences were furthermore supported by GO-term and gradient purification of EVs isolated by ultracentrifugation from KEGG pathway analysis (Supplementary Fig. 11). shCON and shITGB3 cells. Western blot analysis confirmed that To validate our findings, we performed Western blot analysis while flotillin and actin were equally present in the EV fractions on proteins from the EV fractions derived from shCON and (fraction 6/7) from shCON and shITGB3 cells, CD81 and shITGB3 cells (Fig. 5c). This confirmed the knockdown of ITGB3 TSG101 were strongly reduced in those fractions from shITGB3 in EVs from the shITGB3 cell line, accompanied by a reduction in cells (Fig. 5d, e). ITGAV, the alpha subunit of the αvβ3 heterodimer of ITGB3. As described above, EVs from both MDA.MB231 shCON and However, the levels of ITGB1, also described to be present in EVs, shITGB3 cells were equally capable of promoting colony growth were almost indistinguishable between vesicle fractions derived of recipient MDA.MB231 cells in an ITGB3-dependent manner. from shCON and shITGB3 cells. These results not only confirm However, the striking differences in the EV secretome of shCON the specificity of our knockdown, but also indicate that different and shITGB3 cells prompted us to speculate that other cellular integrins might be associated with different types of EV. In line processes might be altered. To test this, we used an indirect with our proteomic analysis, we also confirmed a strong coculture system48 to measure the migratory capacity of IMR90 reduction in TSG101, a member of the ESCRT family, and the cells in the presence of MDA.MB.231 shCON or shITGB3 cells. tetraspanin CD81, a bona fide exosome marker. Importantly, this In line with the reported pro-migratory effect of MDA.MB231- defect appears to be specific to exosomes, as general EV marker derived exosomes49, IMR90 cells cocultured with MDA. proteins such as flotillin-1 or actin were not affected by the MB.231 cells displayed an increased migratory capacity. Such knockdown of ITGB3. In contrast, certain proteins, such as migration was not induced in IMR90 cells cocultured with EIF4E, were exclusively associated with vesicles derived from shITGB3 cells (Supplementary Fig. 13a, b). These results were not shITGB3 cells. These results were confirmed with another shRNA due to impaired uptake of vesicles derived from shITGB3 cells, construct, excluding the possibility of an off-target effect since isolated vesicles from both shCON and shITGB3 cells were (Supplementary Fig. 4). Furthermore, we obtained the same incorporated into IMR90 cells with similar kinetics (Fig. 1d). results in MCF7 cells (Supplementary Fig. 12), excluding a cell- Thus, while secreted factors from shITGB3 cells did not promote 6 NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 ARTICLE a Culture Media b 500G; 10’ Pellet: cells and dead cells. 7.0 x 109 Supernatant 6.0 x 109 MDA.MB.231 shCON-DERIVED VESICLES 12.000G; 20’ MDA.MB.231 shITGB3-DERIVED VESICLES [Particles/mL/106 cells] 9 5.0 x 10 Pellet: cell debris, apoptotic bodies, microvesicles. 4.0 x 109 Supernatant 3.0 x 109 100.000G; 70’ Supernatant: soluble 2.0 x 109 factors. 1.0 x 109 Pellet: PBS wash 0.0 100.000G; 70’ 0 100 200 300 400 Size (nm) Nanosight CryoEM LC-MS/MS c Size: 0–125 nm Size: 150–200 nm Size: 125–500 nm Size: 0–500 nm p-value 0.04 10 10 10 10 1.2×10 3.0×10 5.0×10 5.0×10 ns p-value 0.01 ns 2.5×1010 1.0×1010 4.0×1010 4.0×1010 [Particles/mL/106 cells] [Particles/mL/106 cells] [Particles/mL/106 cells] [Particles/mL/106 cells] 2.0×1010 7.5×1009 3.0×1010 3.0×1010 1.5×1010 5.0×10 09 2.0×1010 2.0×1010 1.0×1010 2.5×1009 1.0×1010 1.0×1010 5.0×1009 0 0 0 0 MDA.MB.231 shCON-DERIVED EVs MDA.MB.231 shITGB3-DERIVED EVs d DERIVED-VESICLES DERIVED-VESICLES MDA.MB.231 shCONTROL MDA.MB.231 shITGB3 0.5 µm 0.5 µm 0.5 µm 0.5 µm Fig. 4 Characterization of EVs isolated from shCON and shITGB3 cells. a Schematic workflow for the preparation of EV samples for LC–MS/MS analysis. b Vesicle size distribution based on NTA data. c Particle concentration of isolated vesicle fractions normalized per million cells. d Representative CryoEM images of EV isolated from cultured MDA.MB.231 shCON and MDA.MB.231 shITGB3 cells. Source data are provided as a Source Data file. Scheme created with BioRender (https://biorender.com/). (*p value < 0.05, n = 3). Data are represented as mean ± SD in (b, c). Statistical analysis including two-tailed unpaired Student’s t test data was carried out using GraphPad Prism 6.01. cell migration in IMR90 cells, MDA.MB231 colony growth was In summary, the overall number of vesicles was not stimulated similarly by EVs from shCON and shITGB3 cells, significantly altered between shCON and shITGB3 cells but indicating functional diversity within the heterogeneous popula- the EVs isolated from the cell culture supernatant differed in tion of EVs. size and protein composition. Particularly exosomes, defined as NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 a b c MDA.MB.231 Heat map for q values under 0.05 40 LYSATE EVs All candidates shlTGB3 - + - + Log TOP3 area MDA.MB.231 shiTGB3 –2 0 1 Row Z-Score 2 35 FC<-3 110 KDa lTGB3 30 FC>3 CD81 lTGAV 135 KDa 25 TSG101 20 lTGB1 130 KDa FLOTILLIN 1 15 ITGB3 49 KDa 10 ITGA5 TSG101 ITGB1 49 KDa 5 #1 #2 #3 #1 #2 #3 EIF4E MDA. MB.231 MDA. MB.231 0 0 5 10 15 20 25 30 35 40 23 KDa shCON EVs shITGB3 EVs CD81 Log TOP3 area MDA.MB.231 shCON 23 KDa d MDA.MB.231 shCON DERIVED-EVs e MDA.MB.231 shITGB3 MDA.MB.231 shCON 1.05 1.19 FLOTILLIN 1 49 KDa INPUT Density DERIVED-EVs DERIVED-EVs (g/mL) F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 ITGB3 110 KDa ACTIN 42 KDa TSG101 49 KDa INPUT INPUT FLOTILLIN 1 49 KDa F6 F7 F6 F7 CD81 23 KDa lTGB3 110 KDa ACTIN 42 KDa TSG101 49 KDa MDA.MB.231 shITGB3 DERIVED-EVs FLOTILLIN 1 49 KDa 1.05 1.19 Density INPUT (g/mL) F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 CD81 23 KDa ITGB3 110 KDa ACTIN 42 KDa TSG101 49 KDa FLOTILLIN 1 49 KDa CD81 23 KDa ACTIN 42 KDa Fig. 5 Alterations in the proteome of shITGB3-derived EVs. a Heat map representation of the normalized protein abundance (Log TOP 3 area) (q.value < 0.05) for the comparison of experimental replicates. b Scatter plot comparison of the abundance (Log TOP3 area) of proteins in the LC–MS/MS analysis of MDA.MB.231 shCON− and MDA.MB.231 shITGB3-derived EVs. The proteins validated by Western blot are highlighted. c Representative immunoblot analysis of candidate proteins identified in the proteomic survey. d Representative immunoblots after separation of vesicles in an iodixanol gradient for both MDA.MB.231 shCON (upper panel) and MDA.MB.231 shITGB3 (lower panel). e Representative immunoblot analysis of input and fractions 6 and 7. Source data are provided as a Source Data file. CD81/TSG101-positive vesicles, appear strongly diminished in that their uptake is particularly dependent on ITGB3. As we could the case of shITGB3. The cause of this phenotype might be due to not rigorously rule out if defects in exosome biogenesis or EV defective exosome biogenesis in shITGB3 cells or linked to the uptake were responsible for the alterations in the EV secretome of role of ITGB3 in vesicle uptake. Regarding a putative role in shCON and shITGB3 cells, we looked further at the mechanism exosome biogenesis, it is important to mention that the ITGB3- of ITGB3-dependent EV uptake. interacting protein SRC has previously been shown to be required for exosome biogenesis34. Linking the described alterations in the Uptake of EVs by MDA.MB.231 cells is dependent on FAK. As EV secretome of shITGB3 cells to the role of ITGB3 in EV described above, DYNAMIN-dependent endocytic recycling is uptake, the abundance of small vesicles (50–125 nm) indicates integral to the regulation of integrin function50. In the case of 8 NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 ARTICLE ITGB3, this recycling enables the assembly and disassembly of blocked when DYNAMIN-dependent endocytosis was inhibited focal adhesions to allow cell migration. A key regulator of this (Dyngo), but not when clathrin-dependent endocytosis was process is FAK, which binds to and regulates the function of both inhibited (Pitstop) (Fig. 6g). Notably, among all tested signalling proteins: DYNAMIN and ITGB351–53. pathways reported to be downstream of ITGB3, only FAK To elaborate a putative link between FAK and ITGB3 in EV- activity was stimulated by EVs (Supplementary Fig. 14). related functions, we first monitored the activity of FAK in cells in Together with our previous observation that vesicle uptake different growth conditions. We prepared protein extracts from relies on DYNAMIN-dependent endocytosis (Fig. 2b, c), these cells grown as colonies in the absence or presence of CM and results suggest that DYNAMIN and FAK are mutually normally growing cells, and monitored FAK activity by Western dependent and together regulate EV uptake in an ITGB3- blot using the Y397 autophosphorylation as readout. In dependent manner. Notably, DYNAMIN and FAK have been comparison to normal growth conditions, FAK activity was shown to be interdependent for regulating focal adhesion strongly increased when cells were grown as individual clones, turnover51,52, a scenario reminiscent of our observations. particularly under CM-stimulated conditions (Fig. 6a). This FAK In summary, we postulate that the transmembrane cell-surface activity correlates with its described role in clonogenic protein ITGB3 plays a central role in the uptake of a subset of growth54,55. However, FAK as a multifunctional protein kinase EVs. In this function, ITGB3 does not act as an EV receptor but is is required for clonogenic growth under basal conditions, critical for mediating the link between EV capture at the cell complicating further analysis to establish a link to the ITGB3- surface and their uptake by endocytosis. Through the critical dependent function in CM-stimulated colony formation. position of ITGB3 at the interface of extracellular and cellular Nevertheless, the described interaction between FAK and interactions, the recognition of EVs at the cell surface by ITGB3- ITGB3 prompted us to examine in more detail if FAK activity is interacting HSPGs can be transmitted to the cytoplasm to trigger also required for EV uptake. To test this, cells were pre-treated the interplay between DYNAMIN and FAK which ultimately with the small-molecule inhibitor FAK14 for 16 h to inactivate allows the endocytic uptake of EVs (Fig. 6h). the kinase, then incubated with fluorescently labelled EVs. After 3 h, FACS analysis of internalized vesicles revealed a reduction in fluorescent intensity of approximately 60% compared to cells Discussion treated with DMSO as solvent control (Fig. 6b). Knockdown of Building on the findings26,27,56 that ITGB3 is required for lung FAK by siRNA confirmed these results (Fig. 6b, c). These data metastasis in MDA.MB.231 cells, we have demonstrated that the suggest that FAK activity is required for EV uptake into recipient underlying mechanism may relate to the role ITGB3 plays in EV cells. To test if FAK activity alone was sufficient for EV uptake in uptake and exosome biogenesis. In light of the preserved lung- the absence of ITGB3, we transiently expressed FAK-GFP or the homing capacity of shITGB3 cells in animal models, we focused pCDNA3-GFP vector as a negative control in shCON and on the next step in metastasis—clonal growth—and found that it shITGB3 cells. Cells were treated with PKH26-labelled EVs for 3 was largely dependent on secreted factors in the cellular envir- h and the uptake into GFP-positive cells was determined by onment; shITGB3 cells had reduced EV uptake, rendering them FACS. While we did observe a slight increase in EV uptake in insensitive to these factors. Our data indicate that this need for shCON cells expressing FAK-GFP, the defect in vesicle uptake in ITGB3 may be due to its dual role in connecting the capture of shITGB3 cells was similar to the vector control (Fig. 6d). The vesicles at the cell surface by HSPG-modified proteins like syn- activity of the overexpressed FAK-GFP was confirmed by decans to the local activation of FAK-induced DYNAMIN-driven Western blot, using the autophosphorylation of FAK on Y397 endocytosis (Fig. 6h). as readout (Fig. 6e). These results therefore suggest that, while As depicted in our model, endocytosis of EVs is linked to integrin FAK is required for EV uptake, ITGB3 interaction with HSPGs is internalisation. Integrins are established key players in the inter- required to link vesicle recognition at the cell surface to FAK action between cells and their environment. This interaction must activation in the cytoplasm. be tightly regulated for normal cell development; the deregulated As described above for the knockdown of ITGB3, a significant expression of various integrins correlates with disease progression portion of vesicles were still taken up following FAK14 treatment in several cancers. As transmembrane cell–surface receptors, or knockdown of FAK by siRNA. We therefore wondered if FAK integrins are critically positioned for cellular–extracellular interac- inhibition would result in the same alterations in EVs in the cell tions. Together with other scaffold proteins and signalling mole- culture supernatant as we observed in shITGB3 cells. To test this, cules they form focal adhesion complexes that link the actin control and FAK-14-treated MDA.MB.231 cells were maintained cytoskeleton to the extracellular environment. These complexes in low-serum-containing medium (exosome-depleted 0.5% FBS) form following integrin clustering induced by interactions with the for 48 h, and vesicles in the cell culture supernatant were isolated extracellular matrix. The attachment of EVs to integrin-linked cell- by differential ultracentrifugation (Fig. 4a). Western blot analysis surface receptors may induce a similar effect. The existence of such of the obtained EVs revealed that CD81 and TSG101 were “endocytic hotspots57” for EV uptake on filopodia has been significantly reduced, while flotillin-1 and actin remained described previously. Cell spreading and motility requires active unaffected (Fig. 6f). turnover of cell adhesion complexes, achieved mainly through Considering these parallelisms in the EV-related functions of endocytic recycling of integrins, a process that has been studied in ITGB3 and FAK, we next wondered if ITGB3-mediated detail in recent years21,58. Many common molecular players are recognition of EVs was directly linked to the activation of involved, and certain specific integrin variations have been descri- FAK. To demonstrate this, we first serum-starved cells for 24 h bed, mainly relating to the highly variable C-terminal tail of beta to eliminate residual FAK activity19,52, then added purified EVs. integrins59,60. These sequence motifs define specific interactions As shown in Fig. 6g, the purified vesicles resulted in FAK with proteins, determining whether the uptake will be dependent activation, as demonstrated by autophosphorylation at Y397. upon caveolae, clathrin, neither, of both. In the case of ITGB3, The specificity of this phosphorylation event was confirmed by endocytosis has been reported to rely solely on DYNAMIN49—our FAK-14 treatment. Importantly, this activation was strongly findings support this. Finally it is important to mention that the reduced in shITGB3 cells, pointing towards an ITGB3- accurate localisation of individual EVs and the discrimination dependent activation of FAK. Moreover, FAK activation was between different vesicle types among the heterogeneous population NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 a b 1.25 c p-value 0.0006 p-value 0.0027 Clonogenic siFAK - - + Non clonogenic p-value 0.0008 growth FAK-14 - + - 1.00 growth Non CM pFAK (Y397) 135 KDa CM 0.75 MFI Tot-FAK 135 KDa pFAK (Y397) 135 KDa 0.50 ACTIN 42 KDa Tot-FAK 135 KDa 0.25 ACTIN 42 KDa f shITGB3 FAK-14 shCON 0.00 shCON cells + EVs d 1.75 shITGB3 cells + EVs siFAK cells + EVs 1.50 FAK-14 cells + EVs TSG101 49 KDa e shCON shITGB3 1.25 23 KDa Cells Cells CD81 1.00 NT FAK NT FAK MFI FLOTILLIN-1 49 KDa GFP-pFAK (Y397) 150 KDa 0.75 pFAK (Y397) 135 KDa 0.50 ACTIN 42 KDa ACTIN 42 KDa 0.25 0.00 h Heparinase FAK-14 shCON pCDNA3-GFP cells Heparin shITGB3 siFAK shCON FAK-GFP cells Extracellular shITGB3 pCDNA3-GFP cells space shITGB3 FAK-GFP cells (a) (b) (c) g t0 + EVs 60’ Vesicle Endocytosis docking induction shITGB3 shITGB3 FAK-14 shCON shCON PitStop Alternative Dyngo entry routes shDynamin (d) Dyn2K44A Dyngo Endocytosis ITGB3 110 Kda HSPGs ITGB3 p-FAK (Y397) 135 Kda Small EV Dynamin Cytoplasm Tot-FAK 135 Kda Heterogeneous p-FAK EV population Endosome ACTIN 42 Kda Fig. 6 EV-induced activation of FAK is required for ITGB3-dependent EV uptake. a Representative immunoblot analysis of pFAK activity either in normal conditions or clonogenic growth conditions with and without conditional medium (CM). b Internalization of 2–5 µg mL−1 PKH26-labelled exosomes derived from MDA.MB.231 cells by MDA.MB.231 shCON, shITGB3, siFAK or shCON pretreated with 15 µM FAK 14 for 14–16 h, measured by FACS. c Representative immunoblot analysis confirming FAK knockdown and pFAK activity reduction after FAK-14 treatment. d Internalization of 2–5 µg mL−1 PKH26-labelled exosomes derived from MDA.MB.231 cells by MDA.MB.231 shCON pCDNA3-GFP, shCON FAK-GFP, shITGB3 pCDNA3-GFP, and shITGB3 FAK-GFP, in GFP-positive cells measured by FACS. e Representative immunoblot analysis of pFAK activity in shCON pCDNA3-GFP, shCON FAK-GFP, shITGB3 pCDNA3- GFP, and shITGB3 FAK-GFP. f Representative immunoblot analysis of TSG101, CD81 and flotilin-1 on EVs derived from untreated MDA.MB.231 shCON or MDA.MB.231 shITGB3 cells or from control or FAK-14 inhibitor (15 μM) treated MDA.MB.231 shCON cells. g Representative immunoblot analysis of the autophosphorylation of FAK at Y397 as readout for its kinase activity. MDA.MB.231 shCON or shITGB3 serum-starved cells were pretreated with 20 μM Dyngo4a, 10 μM Pitstop 2 or 10 μg mL−1 heparin for 30 min or with 15 μM FAK-14 for 14–16 h followed by addition of EVs for 60 min. h Model for the proposed role of ITGB3 in vesicle uptake and exosome biogenesis. (a) EV-HSPG interaction; (b) αvβ3 recruitment; (c) pFAK-DYNAMIN recruitment to endocytosis complex; (d) Dynamin-mediated internalization of EVs and EE formation. Source data are provided as a Source Data file. Model created with BioRender (https://biorender.com/). (**p value < 0.01, ***p value < 0.001 n = 3). Data are represented as mean ± SD in (b) and (d). Statistical analysis including two-way ANOVA multiple comparisons was carried out using GraphPad Prism 6.01. 10 NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 ARTICLE of EVs is still hindered by the current limitations of the applied have been described. Finally, certain receptor cell-specific adap- microscopy techniques10. This is particularly important, when tations to this mechanism may also occur, such as the use of considering the role ITGB3 in the selective uptake of small EVs. different integrin receptors. The most likely candidates would be While NTA can readily discriminate those different sized EV integrins that can interact with heparinised cell-surface receptors populations, the resolution limit of conventional confocal micro- and contain a β-subunit capable of binding and activating FAK. scopy does not permit visualisation of those vesicles and to dis- We found that ITGB3 knockdown rendered breast cancer cells criminate among EV subpopulations at reliable resolution in vivo. insensitive to secreted factors in their environment, reducing their Current advancements of FACS (asymmetric-flow field-flow frac- clonal growth capacity. Intercellular communication represents tionation technology) and microscope-based techniques will com- an important target for future cancer treatment strategies. Cur- plement the NTA based technologies in the future, which will allow rent approaches target cancer cells, yet the past decade of research the precise determination of vesicle size and localisation in vivo10,61. has demonstrated the importance of the cell environment and As described above, FAK and DYNAMIN are key players in intercellular communication in disease progression and treatment the ITGB3-dependent endocytosis of EVs. Supporting this, the resistance: blocking these interactions seems a sensible strategy to interaction between FAK and the C-terminal tail of ITGB3 is well improve current treatments. established60,62. FAK-dependent phosphorylation and regulation of the effector protein DYNAMIN is required for the turnover of Methods focal adhesions19,51,52. Interestingly, our studies indicate that Cell culture and reagents. Breast cancer cell lines were purchased from the FAK and DYNAMIN are mutually dependent on each other for American Type Culture Collection (ATCC) and authenticated by DNA profiling using short tandem repeat (GenePrint® 10 System, Promega) at Genomics Core the ITGB3-dependent uptake of EVs. This interdependency Facility, Instituto de Investigaciones Biomédicas Alberto Sols CSIC-UAM. MDA. might be due to the reported direct interaction between the two MB.231 and MCF7 cells were maintained in Dulbecco’s modified Eagle’s medium proteins for the endocytic uptake of ITGB1. However, given the (DMEM) (Invitrogen) supplemented with 10% heat-inactivated FBS (Life Tech- marked differences in the cytoplasmic tail of ITGB3 and ITGB1, nologies) and antibiotics (100 U mL−1 penicillin, 100 μg mL−1 streptomycin) (Life detailed molecular analysis exploring the involvement of other Technologies). Cells were maintained at 37 °C in a 5% CO2 humidified incubator. The FBS used for exosome purification experiments was depleted of EVs by cen- adapter proteins or the reported interaction with signalling trifugation at 100,000g for 1 h 10 min at 4 °C. For the preparation of medium with molecules like SRC will be required to delineate the exact mole- EV-depleted 0.5% FBS, EV-depleted FBS was added to DMEM to a final con- cular mechanisms. As for ITGB3, we cannot rigorously exclude centration of 0.5% v/v. the possibility that FAK might also be required for exosome biogenesis. While our data strongly support a role of FAK in the Cell culture migration assays in co-culture. Cells were cultured in a double ITGB3-dependent uptake of EVs, a closer examination of the chamber coculture system allowing cell types to grow in close proximity to one another without mixing (TC insert for 24-well plate, PET membrane bottom, potential role of FAK in exosome biogenesis might present an translucent, pore size 0.4 µm, sterile, non-pyrogenic/endotoxin-free, non-cytotoxic, interesting focus for future investigation. 1 pc./blister (Sarstedt). The upper part is a transwell comprised of 24 cell culture Interestingly, our proposed model for ITGB3-dependent EV inserts with a 0.4-μm pore size to avoid mixing the two cell types. The lower part is uptake closely resembles the uptake of herpes viruses. Viral a well of the 24-well plate. IMR90 cells were plated in the lower culture 24-well plate area in triplicate. Either MDA.MB.231 shCONTROL or MDA. endocytosis begins with the viral capsid binding to integrin- MB.231 shITGB3 cells were plated in triplicate in the upper inserts. After 48 h associated HSPGs at the cell surface. Following attachment and growing in shared medium, IMR90 reached a monolayer and were treated over- receptor clustering, activation of receptor tyrosine kinases (FAK, night with mitomycin C (5 μg mL−1, Santa Cruz Biotechnology). Then, a wound SRC) drive specific signalling pathways that activate membrane was made in the monolayer with a pipette tip, the medium was replaced, and the fission and fusion factors. To date, a variety of mechanisms have cells were incubated in normal conditions. Pictures of the wounds were taken after 24 h, and wound closure was measured using ImageJ software. been described for EV uptake. A similar variety has been described for the uptake of different viruses (20–300 nm) into Cell-surface flow cytometry staining protocol. For cell-surface levels, after 1 h specific cell types. Drawing a further comparison between the treatment with heparin (Sigma Aldrich), Dyngo-4a (Abcam) or Pitstop-2 (Abcam) cell-specificity of viruses and that of EVs, we demonstrated that a or after an overnight treatment in the case of FAK-14 inhibitor, cells were detached specific fraction of EVs (approximately 50% of the total vesicles) using enzyme-free phosphate-buffered saline (PBS)-based Cell Dissociation Buffer were taken up in an ITGB3-dependent manner. Since this effect (Gibco). Cells were washed in PBS solution and incubated with the FITC- conjugated ITGB3 antibody at room temperature for 1 h in non-permeabilized was not increased by combining treatments (shITGB3 + conditions. Cells were washed in PBS solution, suspended in FACS buffer (EDTA heparin), we conclude that a certain type of EVs are taken up via 2.5 mM, 1% bovine serum albumin (BSA) in PBS), and analysed on a FACSCalibur this route. Analysis of the EVs in cell culture supernatant revealed instrument with integrated FACSDiva (BD Biosciences) v8.0.1 Software. a significant increase in vesicles with a diameter between 50 and 100 nm in shITGB3 cells, which might indicate that the uptake of Clonal cell growth assay. Cells were seeded in six-well plates (500 cells/well) and those vesicles is particularly dependent on ITGB3. Interestingly, cultured for 7–9 days at 37 °C and 5% CO2. The supernatant was discarded and the cells were gently washed with 1 mL of PBS. Clonal cell growth was assessed fol- vesicles of this size (termed exomeres or P200 exosomes) are lowing staining with 0.5% crystal violet in ddH2O. The crystal violet staining of described to contain less CD81 than other EVs63,64, which might cells from each well was solubilized using 15% acetic acid and the optical density of explain why the relative increase in these vesicles in the total the solution was measured. isolated EVs results in a reduction of CD81 in the cell culture supernatant of shITGB3 cells. However, detailed molecular ana- Cryogenic electron microscopy (CryoEM). EVs collected by ultracentrifugation lysis using recently developed techniques like asymmetric flow were analysed at similar dilutions on electron cryomicroscopy (CryoEM) in the field-flow fractionation of EVs are required to determine the exact microscopy facility at the Universitat Autònoma de Barcelona. Vitrified specimens were prepared by placing 3 μL of a sample on a holey carbon TEM grid, blotted to a molecular identity of those vesicles61. Although we cannot rig- thin film and plunged into liquid ethane-N2(l) in the Leica EM GP cryoworksta- orously exclude that the differences in the EV secretome of tion. The grids were transferred to a 626 Gatan cryoholder and maintained at shITGB3 cells might be caused by alterations in exosome bio- −179 °C. The grids were analysed with a Jeol JEM 2011 transmission electron genesis, the data presented here on the role of ITGB3 in vesicle microscope operating at an accelerating voltage of 200 kV. Images were recorded on a Gatan Ultrascan 2000 cooled charge-coupled device (CCD) camera with the uptake and the parallelism in the phenotype with the ITGB3- Digital Micrograph software package (Gatan). interacting protein FAK rather suggest a selective defect in the uptake of small vesicles devoid of CD81. The recently discovered Enzymatic treatments. The protocol followed for heparan sulfate enzymatic variety in EV size and molecular composition is reminiscent of digestion experiments was described in Christianson et al.14. Briefly, cells were the variety seen in viruses, for which different uptake mechanisms cultured in digestion buffer (DMEM supplemented with 0.5% [wt vol−1] BSA and NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 20 mM Hepes-HCl, pH 7.4) and either untreated or treated with 1.2 mIU mL−1 Immunofluorescence. Cells were grown on cover glasses for 1 day, rinsed with heparinase I and 0.6 mIU mL−1 heparinase III (Sigma-Aldrich) for 3 h at 37 °C. TBS at room temperature and fixed for 10 min with 4% paraformaldehyde. After Then, EVs were added and incubated for 3 h, followed by flow cytometry analysis rinsing with TBS, in the case of permeabilized conditions, cells were incubated with (as described later). 0.1% Triton X-100 in TBS for 30 min followed by 30 min in blocking buffer (BSA 2% in TBS-T 0.1% Triton X-100) at room temperature. For non-permeabilized conditions, cells were directly incubated in blocking buffer (BSA 2% in TBS) for 1 h EV labelling. For EV-uptake experiments, purified EVs were fluorescently labelled at room temperature. Next, cells were incubated in the suitable primary antibodies using PKH26 Red Fluorescent Cell Linker Midi Kit (Sigma-Aldrich). EVs isolated for 1 h at room temperature and rinsed repeatedly with TBS before incubating in by ultracentrifugation and resuspended in 1 mL of PBS were labelled with the the appropriate fluorescein-labelled secondary antibody for 1 h at room tempera- PKH26 dye (2 µM final concentration) according to the manufacturer’s ture. Cells were then washed extensively with TBS and mounted on a glass slide in instructions. Prolong Diamond Antifade mountant with DAPI (Pierce, Thermo Fisher Scien- tific). The following primary antibodies were used: anti-EEA1 (1:1000) (Cell Sig- EV purification. For all experiments, EVs were isolated by ultracentrifugation. nalling), anti-LAMP-1 (1:1000) (Santa Cruz), anti-avb3 (1:50) clone LM609 Cells were cultured in media supplemented with 10% exosome-depleted FBS and (Merck), and anti-tubulin (1:1000) (DM1a-FTIC, Sigma). The following secondary after 48 h were changed to media supplemented with 0.5% exosome-depleted FBS. antibodies were used: goat anti-mouse IgG (H + L) Alexa Fluor Plus 488 and goat Equivalent amounts of supernatant fractions collected after 48–72 h cell culture anti-rabbit IgG (H + L) Alexa Fluor 633 (Thermo Fisher Scientific) (1:1000). were pelleted by centrifugation at 500 g for 10 min, to remove cells. The super- natant was centrifuged at 12,000 g for 20 min, to remove cell debris and dead cells, Microscopy and image analysis. Cells were observed under a confocal laser and finally, EVs were collected by centrifugation at 100,000 g for 70 min (Thermo scanning microscope LSM980 (Carl Zeiss, Germany) with a 63× 1.4 NA oil Scientific, mx ultra-series centrifuge). The EV pellet was resuspended in 20 mL of immersion lens. Images for EV location analysis were captured with a pixel size of phosphate-buffered saline (PBS) and collected by ultracentrifugation at 100,000 g 0.07 × 0.07 × 0.31 px µm−1 (xyz, respectively). Images for co-localization analysis for 70 min (Thermo Scientific, mx ultra-series centrifuge), and the final pellet was were optimally sampled (Nyquist–Shannon theorem) at 0.03 × 0.03 × 0.16 px µm−1 resuspended in 100 µL of PBS. Protein concentration in the final EV pellet was (xyz, respectively). Images were firstly deconvoluted using Zeiss Zen 3.0 program. measured by BCA (Pierce, Thermo Fisher Scientific) according to the manu- EV location and counting was done using Imaris software (7.2.3) though 3D cell facturer’s instructions. A detailed protocol for EV preparation is available at Pro- reconstruction. Co-localization was done using Zeiss Zen 3.0 program. tocol Exchange (Nature Research, https://doi.org/10.21203/rs.3.pex-1044/v1). Modulation of expression using plasmid transient transfection. Wild-pGFP EV gradient separation. A discontinuous iodixanol gradient was prepared by FAK was a gift from Kenneth Yamada (Addgene plasmid # 50515; http://n2t.net/ diluting a stock solution of OptiPrep™ (60% w/v) with 0.25 M sucrose/10 mM Tris, addgene:50515; RRID:Addgene_50515)65. Transfections using TransIT-BrCa pH 7.5, to generate 40, 20, 10, and 5% w/v iodixanol solutions. The discontinuous (Mirrus) were performed according to the manufacturer’s protocol, at a final iodixanol gradient was generated by sequentially layering 3 mL each of 40, 20, and concentration of 3 µg of plasmid DNA in a single well of a 6-well plate. Western 10% (w/v) iodixanol solutions, followed by 2.5 mL of the 5% iodixanol solution. blotting with anti-phospho-FAK (Y397) antibody was used to confirm plasmids EVs, previously purified by ultracentrifugation, were loaded on the discontinuous were expressing the wild type. iodixanol gradient and centrifuged for 16 h at 100,000 g at 4 °C. Fractions of 1 mL were collected from the top of the gradient, and recentrifuged at 100,000 g for 2 h at Expression modulation by retroviral and lentiviral infection. For lentiviral 4 °C. The resulting pellets were resuspended in 25 µL Laemmli buffer 1× and shRNA of ITGB3, pLKO.1-puro-shITGB3 was constructed by annealing the oli- analysed by Western blot. gonucleotides 5′-CCGGGCCAAGACTCATATAGCATTGCTCGA-3′ and 5′-AAT TCAAAAAGCCAAGACTCATATAGCATT-3′ and cloning them into a pLKO1 FAK activity. Starved cells were treated with Dyngo-4a or Pitstop-2 30 min prior to vector. The shITGB3 #3 was obtained from Dharmacon. We also used pLKO1 as adding EVs; FAK 14 inhibitor was added the night before the EVs. Acceptor cells shCTL. Sigma Mission shRNA clone TRCN0000006649 plasmid was used for were incubated with EVs (10–25 µg mL−1) for 1 h, then washed with cold PBS and targeting DYNAMIN 2. WT DYNAMIN 2 pEGFP (Addgene plasmid # 34686; collected on ice for immunoblotting. http://n2t.net/addgene:34686; RRID:Addgene_34686) and K44A DYNAMIN 2 pEGFP (Addgene plasmid # 34687; http://n2t.net/addgene:34687; RRID: Addgene_34687), a gift from Sandra Schmid, were used for subcloning (HindIII/ Flow cytometry. For EV uptake studies, acceptor cells were incubated with PKH- NotI) into pLPCX retroviral plasmid. Cell monolayers were incubated with virus labelled EVs (1–5 µg mL−1) for 1–6 h. Cells were detached using enzyme-free PBS- containing cell culture supernatant in the presence of 4 µg/ml polybrene (Sigma- based Cell Dissociation Buffer (Gibco). Cells were washed in PBS solution, sus- Aldrich) for 24 h. Infected MDA.MB.231 and MCF7 cells were selected with 0.7 or pended in FACS buffer (EDTA 2.5 mM, 1% BSA in PBS), and analysed on a 1.5 μg mL−1 puromycin, respectively, for 3–4 days. Viral production and infection FACSCalibur instrument with integrated FACSDiva (BD Biosciences) v8.0.1. FCS were performed at 37 °C. All plasmids were sequenced twice from both ends to Express 4 Flow Research was used to perform Flow cytometry analysis. In blocking ensure expression of the correct coding sequence. experiments, heparin, Dyngo-4a or Pitstop-2 were added 30 min prior to EVs; FAK 14 inhibitor was added the night before the EVs. Expression modulation by small interfering RNA knockdown. siRNA trans- fections were performed in Opti-MEM medium (Life Technologies) using Lipo- Homing of MDA.MB.231 cells to lung. Female athymic nude mice (Harlan fectamine RNA-iMAX (Life Technologies), following manufacturer’s instructions. Interfauna Iberica, Barcelona, Spain) were kept in pathogen-free conditions and Unless otherwise indicated, transfections were performed for 48 h, with 50 nM used at 7 weeks of age. Temperature and relative humidity is continuously recorded siRNA. The following siRNAs were used: non-silencing control siRNA (D-001810- in the animal housing. The temperature ranged from 23 to 25 °C. Relative humidity 01-05) from Dharmacon and siFAK from Sigma-Aldrich (SASI_Hs01_00035697). ranged between 45 and 65%. Lighting was artificial, from an automatic controlled supply. The cycle gradually simulates twilight and sunset from 7:30 to 8:00 a.m. and Nanoparticle tracking analysis (NTA). EVs collected by ultracentrifugation were from 19:30 to 20:00 p.m., respectively, giving 12 h of light with an intensity of 300 analysed at similar dilutions in a Nanosight NS-300 instrument (Malvern Instru- Lux and 12 h of darkness for each 24 h period. Animal care was handled in ments) for real-time characterization of the vesicles. Using the Automatic Syringe accordance with the Guide for the Care and Use of Laboratory Animals of the Vall Pump (Malvern) at a pump speed of 40, three videos of 60 s were recorded for each d’Hebron University Hospital Animal Facility, and the experimental procedures sample at 20 °C and a concentration of 10–55 particles per frame and used to were approved by the Animal Experimentation Ethical Committee of the institu- calculate mean particle concentration and size. The measurements were analysed tion (76/17 CEEA). using a detection threshold of 5. For all analyses, NTA version 3.1 Build 3.1.45 For the homing assays, we used shCON and shITGB3 MDA.MB.231 cells that was used. were labelled with DiIC18 (D7757; Molecular Probes) according to the All NTA and CryoEM studies were performed at the CIBER-BBN manufacturer instructions; 100,000 cells were injected intravenously into the left (Bioingenieria, Biomateriales y Nanomedicina), Institut de Ciència de Materials de caudal tail vein of each mouse. Mice were sacrificed 12 h later for lung harvest. The Barcelona (ICMAB-CSIC, Barcelona, Spain) (http://www.nanbiosis.es/u6-e04- whole lungs were minced with scalpels and incubated in 2 mL of freshly prepared nanosight-lm-20-for-nanoparticle-tracking-analysis/). DMEM containing 0.8 mg mL−1 dispase (Gibco) and 1.5 mg mL−1 collagenase P (Roche) at 37 °C in agitation for 15 min. After incubation, 10 mL of DMEM was added, and tissue fragments and cell suspensions were spun at 500 g for 5 min. Preparation of EVs and cell lysates. EV pellets prepared by differential high- Pellets were incubated with 0.05% trypsin-EDTA at 37 °C for 2 min. Single-cell speed centrifugation were resuspended in PBS and a known concentration was suspensions were filtered through 70-µm nylon filter and washed in PBS with 2 loaded on SDS–polyacrylamide gels in Laemmli buffer. The corresponding cell mM EDTA and 1% BSA. Cells were blocked and stained with CD45 Monoclonal layers were washed in cold PBS and scraped on ice in lysis buffer (50 mM Tris-HCl, Antibody (30-F11) (1:200), FITC (eBioscience) for flow cytometry, and absolute pH 7.4, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 1 counts of CD45-negative DiIC18-positive cells were obtained and analysed on a mM EDTA) supplemented with PhosSTOP and Complete Phosphatase/Protease FACSCalibur instrument with integrated FACSDiva (BD Biosciences) v8.0.1. Inhibitor Cocktails (Roche Diagnostics GmbH, Mannheim, Germany). The whole 12 NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 ARTICLE cellular lysate was centrifuged at 16,000 g for 15 min to clear cell debris then stored quantification. Proteins observed in all replicates of one condition and in none of at −20 °C. Protein concentration in the cellular extracts was determined using the the other conditions were considered significantly different (presence/absence). BCA method (Pierce, Thermo Fisher Scientific) according to the manufacturer’s instructions. Protein extracts (40–50 μg per sample) were loaded onto sodium Statistics. Results are expressed as means + standard errors of the means. Two- dodecyl sulfate-polyacrylamide gel electrophoresis gels and transferred electro- tailed Student’s t test and two-way ANOVA multiple comparisons were carried out phoretically to PVDF membranes, and immunodetection of proteins was per- as appropriate, both using GraphPad Prism 6.01. p < 0.05 was considered formed using ECL™ Western Blotting Detection Reagents (GE Healthcare, significant. Buckinghamshire, UK). The following primary antibodies were used: anti-CD81 (1:5000), anti-pFAK (1:1000) and anti-FAK (1:1000), anti-DYNAMIN 2 (1:500), anti-LAMP1 (1:500) and anti-ERK (1:1000) (Santa Cruz Biotechnology); anti- Reporting summary. Further information on research design is available in the Nature Vinculin (1:1000) (Sigma-Aldrich) anti-TSG101 (1:2500), anti-αv integrin (1:2000) Research Reporting Summary linked to this article. and anti-β1 (1:5000) integrin (Abcam); anti-β3 integrin (1:500), anti-pERK (1:1000), anti-pSrc (1:500), anti-Src (1:500), anti-pAKT (1:1000), anti-AKT (1:1000), anti-EEA1 (1:2500) and anti-EIF4E (1:1000) (Cell Signalling); anti- Data availability The mass spectrometry raw data have been deposited into the PRIDE69 repository with flotillin-1 (1:1500) (Novus Biologicals) and anti-β-actin (1:500; Sigma-Aldrich). Anti-mouse and anti-rabbit HRP secondary antibodies were from Pierce (1:10000). the dataset identifier PXD013489. All the data supporting the findings of this study are Bound antibodies were visualized with an enhanced chemiluminescence detection available within the article and its supplementary information files and from the kit (Amersham Pharma-Biotech). corresponding author upon reasonable request. A reporting summary for this article is available as a Supplementary Information file. Source data are provided with this paper. Proteomics sample preparation. Eluted proteins were reduced, alkylated and Received: 25 April 2019; Accepted: 31 July 2020; digested to peptide mixes according to the filter-aided sample preparation66 method using LysC 1:10 ratio (w:w; enzyme:substrate) at 37 °C overnight, followed by trypsin 1:10 ratio (w:w; enzyme:substrate) at 37 °C for 8 h. Tryptic peptide mixtures were desalted using a C18 UltraMicroSpin column67. Protein extracts (15 µg) were reduced with dithiothreitol (30 nmol, 37 °C, 60 min) and alkylated in the dark with iodoacetamide (60 nmol, 25 °C, 30 min). Samples were first diluted to 2 References M urea with 200 mM NH4HCO3 for digestion with endoproteinase LysC (1:10 w- 1. Massagué, J. & Obenauf, A. C. Metastatic colonization. Nature 529, 298–306 w, 37 °C, o/n, Wako), and then diluted 2-fold with 200 mM NH4HCO3 for trypsin (2016). digestion (1:10 w-w, 37 °C, 8 h, Promega). After digestion, peptide mix was acid- 2. Wan, Z. et al. Exosome-mediated cell-cell communication in tumor ified with formic acid and desalted with a MicroSpin C18 column (The Nest progression. Am. J. Cancer Res. 8, 1661–1673 (2018). Group, Inc) prior to LC–MS/MS analysis. 3. Peinado, H. et al. Pre-metastatic niches: organ-specific homes for metastases. Nat. Rev. Cancer 17, 302–317 (2017). 4. Costa-Silva, B. et al. Pancreatic cancer exosomes initiate pre-metastatic niche Chromatographic and mass spectrometric analysis. Samples were analysed formation in the liver. Nat. Cell Biol. 17, 816–826 (2015). using an LTQ-Orbitrap Velos Pro mass spectrometer (Thermo Fisher Scientific, 5. Becker, A. et al. Extracellular vesicles in cancer: cell-to-cell mediators of San Jose, CA, USA) coupled to an EASY-nLC 1000 (Thermo Fisher Scientific metastasis. Cancer Cell 30, 836–848 (2016). [Proxeon], Odense, Denmark). Peptides were loaded onto the 2-cm Nano Trap 6. Mathieu, M., Martin-Jaular, L., Lavieu, G. & Théry, C. Specificities of secretion column with an inner diameter of 100 μm packed with C18 particles of 5 μm and uptake of exosomes and other extracellular vesicles for cell-to-cell particle size (Thermo Fisher Scientific) and were separated by reversed-phase communication. Nat. Cell Biol. 21, 9–17 (2019). chromatography using a 25-cm column with an inner diameter of 75 μm, packed 7. Choi, D. S., Kim, D. K., Kim, Y. K. & Gho, Y. S. Proteomics, transcriptomics with 1.9 μm C18 particles (Nikkyo Technos Co., Ltd. Japan). Chromatographic and lipidomics of exosomes and ectosomes. Proteomics 13, 1554–1571 (2013). gradients started at 93% buffer A and 7% buffer B with a flow rate of 250 nL min−1 8. Green, T. M., Alpaugh, M. L., Barsky, S. H., Rappa, G. & Lorico, A. Breast for 5 min and were gradually adjusted to 65% buffer A and 35% buffer B over 120 cancer-derived extracellular vesicles: Characterization and contribution to the min. After each analysis, the column was washed for 15 min with 10% buffer A and metastatic phenotype. Biomed. Res. Int. 2015, 6977–6994 (2015). 90% buffer B. Buffer A: 0.1% formic acid in water. Buffer B: 0.1% formic acid in 9. Van Niel, G., D’Angelo, G. & Raposo, G. Shedding light on the cell biology of acetonitrile. The mass spectrometer was operated in positive ionization mode with extracellular vesicles. Nat. Rev. Mol. Cell Biol. 19, 213–228 (2018). nanospray voltage set at 2.1 kV and source temperature at 300 °C. Ultramark 1621 10. Kalluri, R. & LeBleu, V. S. The biology, function, and biomedical applications was used for external calibration of the FT mass analyser before the analyses, and of exosomes. Science 367, 640–655 (2020). an internal calibration was performed using the background polysiloxane ion signal 11. Colombo, M. et al. Analysis of ESCRT functions in exosome biogenesis, at m/z 445.1200. The acquisition was performed in data-dependent acquisition composition and secretion highlights the heterogeneity of extracellular (DDA) mode, and full MS scans with 1 micro scans at a resolution of 60,000 were vesicles. J. Cell Sci. 126, 5553–5565 (2013). used over a mass range of m/z 350–2000 with detection in the Orbitrap. Auto gain 12. Mulcahy, L. A., Pink, R. C. & Carter, D. R. F. Routes and mechanisms of control (AGC) was set to 1E6, dynamic exclusion (60 s), and charge state filtering extracellular vesicle uptake. J. Extracell. Vesicles 3, 24641–24656 (2014). disqualifying singly charged peptides was activated. In each cycle of DDA analysis, 13. Yamauchi, Y. & Helenius, A. Virus entry at a glance. J. Cell Sci. 126, following each survey scan, the 20 most intense ions with multiple charged ions 1289–1295 (2013). above a threshold ion count of 5000 were selected for fragmentation. Fragment ion 14. Christianson, H. C., Svensson, K. J., van Kuppevelt, T. H., Li, J.-P. & Belting, spectra were produced via collision-induced dissociation (CID) at normalized M. Cancer cell exosomes depend on cell-surface heparan sulfate proteoglycans collision energy of 35% and acquired in the ion trap mass analyser. AGC was set to for their internalization and functional activity. Proc. Natl Acad. Sci. USA 110, 1E4, with an isolation window of 2.0 m/z, an activation time of 10 ms and a 17380–17385 (2013). maximum injection time of 100 ms. All data were acquired with Xcalibur software 15. Friand, V., David, G. & Zimmermann, P. Syntenin and syndecan in the v2.2. Digested bovine serum albumin (New England biolabs cat # P8108S) was biogenesis of exosomes. Biol. Cell 107, 331–341 (2015). analysed between each sample to avoid sample carryover and to assure stability of the instrument, and QCloud66 was used to control instrument longitudinal per- 16. Christianson, H. C. & Belting, M. Heparan sulfate proteoglycan as a cell- formance during the project. surface endocytosis receptor. Matrix Biol. 35, 51–55 (2014). 17. Kumar, B. & Chandran, B. KSHV entry and trafficking in target cells— hijacking of cell signal pathways, actin and membrane dynamics. Viruses 8, Data analysis. Acquired spectra were analysed using the Proteome Discoverer 305–328 (2016). software suite (v1.4, Thermo Fisher Scientific) and the Mascot search engine (v2.5, 18. Cynthia, Y. A. & Patricio, I. M. Usage of heparan sulfate, integrins, and FAK Matrix Science67). The data were searched against a Swiss-Prot human database in HPV16 infection. Virology 403, 1–16 (2010). plus a list68 of common contaminants and all the corresponding decoy entries. For 19. Burridge, K. Foot in mouth: do focal adhesions disassemble by endocytosis? peptide identification a precursor ion mass tolerance of 7 ppm was used for MS1 Nat. Cell Biol. 7, 545–547 (2005). level; trypsin was chosen as the enzyme and up to three missed cleavages were 20. Woods, A. J., White, D. P., Caswell, P. T. & Norman, J. C. PKD1/PKC?? allowed. The fragment ion mass tolerance was set to 0.5 Da for MS2 spectra. promotes??v??3 integrin recycling and delivery to nascent focal adhesions. Oxidation of methionine and N-terminal protein acetylation were used as variable EMBO J. 23, 2531–2543 (2004). modifications whereas carbamidomethylation on cysteines was set as a fixed 21. Moreno-Layseca, P., Icha, J., Hamidi, H. & Ivaska, J. Integrin trafficking in modification. False discovery rate in peptide identification was set to a maximum of cells and tissues. Nat. Cell Biol. https://doi.org/10.1038/s41556-018-0223-z 5%. Protein abundance was estimated using the area under the chromatographic (2019). peak of the three most intense peptides per protein. Data were log transformed and 22. Roper, J. A., Williamson, R. C. & Bass, M. D. Syndecan and integrin normalized by equalizing the median of the total protein abundance per sample. interactomes: large complexes in small spaces. Curr. Opin. Struct. Biol. 22, Fold changes, p values and q-values were calculated to assess protein relative 583–590 (2012). NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 23. Hamidi, H. & Ivaska, J. Every step of the way: integrins in cancer progression 52. Ezratty, E. J., Partridge, M. A. & Gundersen, G. G. Microtubule-induced focal and metastasis. Nat. Rev. Cancer https://doi.org/10.1038/s41568-018-0038-z adhesion disassembly is mediated by dynamin and focal adhesion kinase. Nat. (2018). Cell Biol. 7, 581–590 (2005). 24. Paolillo, M. & Schinelli, S. Integrins and exosomes, a dangerous liaison in 53. Alanko, J. et al. Integrin endosomal signalling suppresses anoikis. Nat. Cell cancer progression. Cancers 9, 95–103 (2017). Biol. 17, 1412–1421 (2015). 25. Jay, S. D. & Cherech, D. A. Integrins in cancer: biological implications in 54. Johnson, T. R. et al. Focal adhesion kinase controls aggressive phenotype therapeutic opportunities. Cancer Nat. Rev. 10, 9–22 (2015). of androgen-independent prostate cancer. Mol. Cancer Res. 6, 1639–1648 26. Sesé, M. et al. Hypoxia-mediated translational activation of ITGB3 in breast (2008). cancer cells enhances TGF-β signaling and malignant features in vitro and 55. Begum, A. et al. The extracellular matrix and focal adhesion kinase signaling in vivo. Oncotarget 8, 114856–114876 (2017). regulate cancer stem cell function in pancreatic ductal adenocarcinoma. PLoS 27. Silencing β3 Integrin by Targeted ECO_siRNA Nanoparticles Inhibits EMT ONE 12, 1–21 (2017). and Metastasis of Triple Negative Breast Cancer.pdf. 56. Liu, Z. et al. EGFRvIII/integrin β3 interaction in hypoxic and 28. Lambert, A. W., Pattabiraman, D. R. & Weinberg, R. A. Emerging biological vitronectinenriching microenvironment promote GBM progression and principles of metastasis. Cell 168, 670–691 (2017). metastasis. Oncotarget 7, 4680–4694 (2016). 29. Tóth, L., Pásti, G., Sárváry, A., Balázs, M. & Ádány, R. Effect of tumor- 57. Heusermann, W. et al. Exosomes surf on filopodia to enter cells at endocytic conditioned medium on intercellular communication and proliferation of hot spots, traffic within endosomes, and are targeted to the ER. J. Cell Biol. Balb/c 3T3 cells. Cancer Lett. 151, 57–61 (2000). 213, 173–184 (2016). 30. Cellular, I. V. Effects of various conditioned media on proliferation of an 58. Schmid, S. L. Reciprocal regulation of signaling and endocytosis: implications isolated single cell from insect cell lines author(s): Toshinori Itoh and Jun for the evolving cancer cell. J. Cell Biol. 216, 2623–2632 (2017). Mitsuhashi Source: In Vitro Cellular & Developmental Biology. Animal 31, 59. Arias-Salgado, E. G. et al. Src kinase activation by direct interaction with the 814–816 (2016). (Dec., 1995), pp. integrin β cytoplasmic domain. Proc. Natl Acad. Sci. USA 100, 13298–13302 31. Chen, Y. H. & Pruett-Miller, S. M. Improving single-cell cloning workflow for (2003). gene editing in human pluripotent stem cells. Stem Cell Res. 31, 186–192 60. Wang, R., Shattil, S. J., Ambruso, D. R. & Newman, P. J. Truncation of the (2018). cytoplasmic domain of β3 in a variant form of glanzmann thrombasthenia 32. Zitzmann, J. et al. Single-cell cloning enables the selection of more productive abrogates signaling through the integrin α(IIb)β3 complex. J. Clin. Investig. Drosophila melanogaster S2 cells for recombinant protein expression. 100, 2393–2403 (1997). Biotechnol. Rep. 19, e00272 (2018). 61. Zhang, H. et al. Identification of distinct nanoparticles and subsets of 33. Takahashi, A. et al. Small extracellular vesicles secreted from senescent cells extracellular vesicles by asymmetric flow field-flow fractionation. Nat. Cell promote cancer cell proliferation through EphA2. Nat. Commun. 8, 15729 Biol. 20, 332–343 (2018). (2017). 62. Schaffner-Reckinger, E., Gouon, V., Melchior, C., Plançon, S. & Kieffer, N. 34. Imjeti, N. S. et al. Syntenin mediates SRC function in exosomal cell-to-cell Distinct involvement of β3 integrin cytoplasmic domain tyrosine residues 747 communication. Proc. Natl Acad. Sci. USA 114, 201713433 (2017). and 759 in integrin-mediated cytoskeletal assembly and phosphotyrosine 35. Beauvais, D. M., Ell, B. J., McWhorter, A. R. & Rapraeger, A. C. Syndecan-1 signaling. J. Biol. Chem. 273, 12623–12632 (1998). regulates αvÎ23 and αvÎ25 integrin activation during angiogenesis and is 63. Zhang, Q. et al. Transfer of functional cargo in exomeres article transfer of blocked by synstatin, a novel peptide inhibitor. J. Exp. Med. 206, 691–705 functional cargo in exomeres. Cell Rep. 27, 1–15 (2019). (2009). 64. Lee, S.-S. et al. A novel population of extracellular vesicles smaller than 36. Rapraeger, A. C. Synstatin: a selective inhibitor of the syndecan-1-coupled exosomes promotes cell proliferation. Cell Commun. Signal 17, 1–15 IGF1R-αvβ3 integrin complex in tumorigenesis and angiogenesis. FEBS J. 280, (2019). 2207–2215 (2013). 65. Gu, J. et al. Modulated by PTEN. Cell 146, 389–403 (1999). 37. Yang, N. & Friedl, A. Syndecan-1-induced ECM fiber alignment requires 66. Chiva, C. et al. QCloud: a cloud-based quality control system for mass integrin αvβ3 and syndecan-1 ectodomain and heparan sulfate chains. PLoS spectrometry-based proteomics laboratories. PLoS ONE 13, 1–14 (2018). ONE 11, e015032–57 (2016). 67. Pappin, D. J. C., Creasy, D. M., Cottrell, J. S. & Perkins, D. N. Probability- 38. Gharbaran, R. Advances in the molecular functions of syndecan-1 (SDC1/ based protein identification by searching sequence databases using mass CD138) in the pathogenesis of malignancies. Crit. Rev. Oncol. Hematol. 94, spectrometry data. Electrophoresis 20, 3551–3567 (1999). 1–17 (2015). 68. Beer, L. A., Liu, P., Ky, B., Barnhart, K. T. & Speicher, D. W. Serum/plasma. 39. Beauvais, D. M., Burbach, B. J. & Rapraeger, A. C. The syndecan-1 Proteomics 728, 339–352 (2011). ectodomain regulates??v??3 integrin activily in human mammary carcinoma 69. Vizcaíno, J. A. et al. 2016 update of the PRIDE database and its related tools. cells. J. Cell Biol. 167, 171–181 (2004). Nucleic Acids Res. 44, D447–D456 (2016). 40. Afratis, N. A. et al. Syndecans—key regulators of cell signaling and biological functions. FEBS J. 284, 27–41 (2017). 41. Colombo, M., Raposo, G. & Théry, C. Biogenesis, secretion, and intercellular interactions of exosomes and other extracellular vesicles. Annu. Rev. Cell Dev. Acknowledgements The authors thank Hector Peinado and his laboratory team (Microenvironment & Biol. 30, 255–289 (2014). Metastasis Group, Molecular Oncology Program, Spanish National Cancer Research 42. Emam, S. E. et al. Liposome co-incubation with cancer cells secreted exosomes Centre (CNIO)). The authors thank Dr. Manel Bosch for helpful discussion and assis- (extracellular vesicles) with different proteins expressions and different uptake tance with confocal microscopy. The authors thank Dr. Anne Spang and Dr Jordi pathways. Sci. Rep. 8, 1–11 (2018). Berenguer for helpful suggestions and discussion on the paper, and Jane Marshall for 43. Schneider, D. J. et al. Mechanisms and modulation of microvesicle uptake in a language review and editing support. This work was supported by Fondo de Investiga- model of alveolar cell communication. J. Biol. Chem. 292, 20897–20910 ciones Sanitarias (PI14/01320), Redes Temáticas de Investigación Cooperativa en Salud (2017). (RD12/0036/0057) and CIBERONC (CB16/12/00363). SRC acknowledges support from 44. Mccluskey, A. et al. Building a better dynasore: the dyngo compounds potently the Generalitat de Catalunya (2014 SGR 1131). The CRG/UPF Proteomics Unit is part of inhibit dynamin and endocytosis. Traffic 14, 1272–1289 (2013). the Spanish Infrastructure for Omics Technologies (ICTS OmicsTech) and it is a member 45. Caswell, P. T., Vadrevu, S. & Norman, J. C. Integrins: masters and slaves of of the ProteoRed PRB3 consortium which is supported by grant PT17/0019 of the PE I + endocytic transport. Nat. Rev. Mol. Cell Biol. 10, 843–853 (2009). D + i 2013-2016 from the Instituto de Salud Carlos III (ISCIII) and ERDF. 46. Feng, D. et al. Cellular internalization of exosomes occurs through phagocytosis. Traffic 11, 675–687 (2010). 47. Yoshioka, Y. et al. Comparative marker analysis of extracellular vesicles in different human cancer types. J. Extracell. Vesicles 2, 1–9 (2013). Author contributions 48. Knol, J. C. et al. Sensing of latent EBV infection through exosomal transfer of Concept and design: P. Fuentes, S. Hümmer, M. Sesé, S. Ramón y Cajal. Development of 5′pppRNA. Proc. Natl Acad. Sci. USA 113, E587–E596 (2016). methodology: P. Fuentes, S. Hümmer, M. Sesé, H. Peinado. Data acquisition: P. Fuentes, 49. Westbroek, W. et al. Exosomes released from breast cancer carcinomas P.J. Guijarro, M. Emperador, S. Sánchez-Redondo H. Peinado. Data analysis and inter- stimulate cell movement. PLoS ONE 10, e0117495 (2015). pretation: P. Fuentes, S. Hümmer, M. Sesé, H. Peinado. Writing: P. Fuentes, S. Hümmer. 50. Lee, M. Y. et al. Dynamin 2 regulation of integrin endocytosis, but not VEGF Review of paper: P. Fuentes, S. Hümmer, M. Sesé, H. Peinado, S. Ramón y Cajal. Study signaling, is crucial for developmental angiogenesis. Development 141, supervision: S. Ramón y Cajal, M. Sesé, S. Hümmer. 1465–1472 (2014). 51. Wang, Y., Cao, H., Chen, J. & McNiven, M. A. A direct interaction between the large GTPase dynamin-2 and FAK regulates focal adhesion dynamics in Competing interests response to active Src. Mol. Biol. Cell 22, 1529–1538 (2011). The authors declare no competing interests. 14 NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18081-9 ARTICLE Additional information Open Access This article is licensed under a Creative Commons Supplementary information is available for this paper at https://doi.org/10.1038/s41467- Attribution 4.0 International License, which permits use, sharing, 020-18081-9. adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Correspondence and requests for materials should be addressed to S.Hüm. or S.Rón.y.C. Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless Peer review information Nature Communications thanks the anonymous reviewer(s) for indicated otherwise in a credit line to the material. If material is not included in the their contribution to the peer review of this work. Peer reviewer reports are available. article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from Reprints and permission information is available at http://www.nature.com/reprints the copyright holder. To view a copy of this license, visit http://creativecommons.org/ licenses/by/4.0/. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. © The Author(s) 2020 NATURE COMMUNICATIONS | (2020)11:4261 | https://doi.org/10.1038/s41467-020-18081-9 | www.nature.com/naturecommunications 15