Cadherins are a family of cell adhesion molecules that enable cells to seclude themselves from cells of other tissues, as well as to interact and communicate with the same kind of cells. The extracellular regions of cadherins are responsible for homotypic binding to the ectodomains of other cadherin molecules of the same isoform, which are presented on neighboring cells [1
]. Additionally, they may interact with other receptors. For example, vascular endothelial cadherin (VE-cadherin) can associate with vascular endothelial growth factor (VEGF) receptor II (VEGFRII, also known as Flk1 or KDR) to reduce its proliferative signaling [2
Intracellularly, the cadherin cytoplasmic tails interact with several proteins, such as p120, and α- and β-catenins, thereby connecting intercellular contacts with the actin cytoskeleton and cell-signaling pathways [3
]. All classical cadherins share similar structures, comprising five extracellular cadherin repeats, a transmembrane (TM) domain, and an intracellular domain (ICD). They can be sub-divided into type I cadherin that have a histidine–alanine–valine (HAV) motif present in the first EC domain, and type II cadherins without such a HAV motif. While E- and N-cadherin are type I cadherins, VE-cadherin is a type II cadherin [4
]. E- and N-cadherin only differ in the catenin isoforms that bind to their ICDs—E-cadherin binds with the shorter isoform of p120 catenin while N-cadherin interacts with the longer isoform [5
]. Of great importance in cancer pathophysiology, changes in these two-cell surface cadherins, including switches between the cadherin types, occur during epithelial-to-mesenchymal transition (EMT), thereby altering cell migration and tumor invasiveness. EMT is an initial step of metastatic expansion, in which tumor cells can acquire stem cell phenotypes and become resistant to cancer therapy [6
]. During EMT, loss, decrease, or dysfunction of E-cadherin is consistently observed in most of the advanced, undifferentiated, and aggressive carcinomas of the mammary gland and other epithelial tissues [9
]. Consequently, β-catenin is released from the cancer cell membrane and, after its translocation into the nucleus, regulates transcription of several genes. For example, the β-catenin/TCF pathway activates the vimentin promoter, resulting in epithelial cell migration and tumor cell dissemination and invasion [11
]. Re-expression of E-cadherin in these cancer cells reverts EMT [12
]. Hence, E-cadherin–based cell interaction is an essential factor in tumor invasiveness [14
]. However, in some tumors, decreased expression of E-cadherin does not necessarily correlate with increased cancer cell motility or invasion [16
]. Instead, N-cadherin may also promote an invasive phenotype in breast cancer cells despite their high E-cadherin expression [16
VE-cadherin, initially described to be typical for intercellular junctions between endothelial cells (ECs) [17
], can also be expressed by cancer cells, as detected in human invasive breast carcinoma sections [18
]. There, VE-cadherin is found at three subcellular sites: in the cytoplasm, at the cell membrane, and in the nucleus. Forced expression of VE-cadherin in breast cancer cells induced collective MDA-MB-231 breast cancer cell migration and promoted their integration into endothelial monolayer as well as the formation of functionally competent cell junctions. VE-cadherin expression reportedly increases during EMT in v-Ha-Ras-transformed mammary epithelial cells and enhances tumor growth in vivo [19
]. However, very little is known about the mechanisms of increased VE-cadherin expression in the cancer cells and about the impact it may have on interactions between ECs and tumor cells (TCs), especially in breast cancer cells.
Orchestrating TC survival and progression, the tumor microenvironment (TME) is characterized by the biochemical composition and biophysical properties of the extracellular matrix (ECM) by ECM-sequestered cytokines, as well as by the mutual interactions of the different cellular components, such as TCs, cancer-associated fibroblasts, immune cells, and ECs [20
]. Within the TME, all these cells communicate through several juxtacrine and paracrine mechanisms. The paracrine mechanisms include communication via soluble factors, such as cytokine and extracellular vesicles (EVs). Cells form EVs by outward budding of the plasma membrane or by an intracellular endocytic trafficking pathway involving the fusion of multivesicular late endocytic compartments with the plasma membrane [22
]. Exosomes, a subclass of EVs, are extracellular nanovesicles with a typical diameter of 50–150 nm, being characterized by different membrane proteins, such as CD63. Secreted from cancer cells, they can carry nucleic acids, lipids, and proteins [24
]. The uptake of exosomes by the acceptor cells can happen through phagocytosis, a type of endocytosis [25
As a consequence, the exosome-transported cargos are released intracellularly into the acceptor cell and may actively influence its phenotype [26
]. Thus, other cells of the TME cells acquire a prometastatic phenotype and support tumor functions such as tumor angiogenesis [27
]. Moreover, other types of TME cells, such as ECs, secrete exosomes with EC-typical cargos, which may influence TCs to acquire EC-like characteristics such as formation of vasculogenic mimicry vessels [28
]. Aggressive tumors possess mosaic and vasculogenic mimicry vessels, which are lined by both ECs and TCs, of which the latter integrate into the endothelium and replace ECs partially or entirely, respectively [20
To understand the molecular mechanism in terms of how tumor cells can form close intercellular contacts with ECs, we hypothesized that juxtacrine or paracrine signaling between ECs and TCs could induce ectopic VE-cadherin expression in cancer cells and thereby support the molecular mimicry of TCs to contact neighboring ECs. The present work provides novel insights into the mutual communication and exchange of material between TCs and ECs, which enables breast cancer cells to express endothelial-specific markers such as VE-cadherin. Aggressive and non-aggressive breast cancer cells show neo-expression of VE-cadherin in a mechanism that depends on EVs released by ECs. This highlights a novel type of mutual interaction between these two cell types within the TME. In contrast to the type II cadherin (cadherin-11), the expression of type I cadherin (E-cadherin) inhibits the integration of TCs into the EC layer. VE-cadherin expression in non-aggressive TCs decreases E-cadherin exposure on the cell surface. Thus, the ratio of VE-cadherin to E-cadherin on the cell surface likely determines the integration of the TCs into the EC layer and allows better cohesion with ECs in 3D tumor spheroids. Therefore, the expression of VE-cadherin in non-invasive cancer cells promotes cancer progression to a more aggressive and invasive phenotype.
In aggressive tumor tissues, VE-cadherin is localized in the cytoplasm of all VE-cadherin-expressing TCs, with additional nuclear and cell membrane localization in some cases [18
]. However, the process mediating this neo-expression has not been elucidated. Our present work demonstrates that VE-cadherin is expressed in breast cancer cells neighboring ECs. Moreover, it unravels possible biological mechanisms of this EC-induced VE-cadherin expression in TCs. The cadherin molecules at adheren junctions are diverse, some of which influence growth factor receptor signaling and Rho GTPases to promote cell motility and invasion [37
]. E-cadherin is preferentially expressed in cells of epithelial origin, while VE-cadherin is specific to ECs. VE-cadherin regulates the intercellular contacts between ECs, thereby determining the integrity of blood vessels and barrier function [17
Here, we demonstrated that (i) HUVECs release VE-cadherin-containing EVs, the composition of which changes if cancer cells are close by. Moreover, we showed that (ii) breast cancer cells take up EC-produced VE-cadherin molecules of distinct lengths, which undergo different fates within the cancer cells. Furthermore, (iii) cancer cells carry out efferocytosis towards neighboring ECs, which, albeit supportive to cancer cell proliferation, undergo necroptosis. (iv) Via different mechanisms, EC induces VE-cadherin expression in breast cancer cells, which is accompanied by a decrease of originally expressed cadherin type. As a consequence, (v) this cadherin switch, especially the increase of VE-cadherin in combination with a loss of E-cadherin, enables breast cancer cells to integrate into endothelial tubes without significant loss of vessel integrity and barrier function. The full-length VE-cadherin in cancer cells, both endogenously produced and transferred from ECs via EVs, localized within intracellular vesicles and at intercellular junctions.
In our studies, two phenotypically different types of breast cancer cells (MCF7 and MDA-MB-231) were employed in co-culture models with HUVEC cells. While the MCF7 breast cancer cell line was derived from an invasive ductal carcinoma patient sample and represented a luminal A breast cancer subtype, MDA-MB-231 was isolated from a breast cancer adenocarcinoma, triple-negative subtype [38
]. When co-cultured with ECs, ectopic expression of VE-cadherin enriched at cell–cell junctions can be observed in both MCF7 and MDA-MB-231 for at least 1 week after isolation. Competition between cadherins for their clustering at intercellular junctions in the same cell was shown in ECs between VE-cadherin and N-cadherin as a regulatory mechanism for modulating cadherin function and signaling [39
]. In TCs, VE-cadherin presents structural features that are responsible for its ability to exclude E-cadherin and cadherin-11 from cell–cell contacts in both MCF7 and MDA-MB-231, respectively, as well as to induce their internalization and downregulation. However, in a 3D culture model, ECs induced EMT in breast cancer cells and were able to switch from E-cadherin to N-cadherin expression [40
]. Therefore, we cannot rule out the fact that factors other than the increase of VE-cadherin expression might contribute to the decrease of E-cadherin expression in the TCs that neighbor ECs.
Furthermore, E-cadherin downregulation led to β-catenin dislocation from MCF7 cell–cell contacts. Conversely, the expression of VE-cadherin was progressively upregulated. As the increase of VE-cadherin lagged behind the decrease of E-cadherin, it is conceivable that β-catenin transiently disappeared from the intracellular junction and reappeared there only 72 h after co-culturing. This goes in line with the cadherin switch described by other researchers [37
]. For instance, such a cadherin switch was described during EMT, in which E-cadherin disappeared and N-cadherin appeared in epithelial cancers, and together with morphological changes, they increased expression of the mesenchymal marker vimentin and enhanced cell motility. Furthermore, downregulation of the originally expressed cadherin type led to carcinoma cell disaggregation from their homotypic neighboring cells. Degradation of endogenous cadherin was considered as a consequence of competition for binding to p120-catenin [41
The effect of VE-cadherin neo-expression seems to be more relevant for the E-cadherin-expressing cell line MCF7, as compared to MDA-MB-231. E-cadherin belongs to type I classical cadherins while cadherin-11 and VE-cadherin belong to type II classical cadherins. In vitro cell aggregation assays showed that type II cadherins mediate both homophilic adhesive interactions between cells expressing identical cadherins and selective heterophilic interactions between cells expressing different cadherins [42
]. Therefore, our data suggest that in MCF7 cells, not only the acquisition of VE-cadherin, but also the decrease of E-cadherin plays a vital role in diminishing the homotypic cell–cell cohesion and switching to a heterotypic cell–cell interaction. VE-cadherin expression has been observed in specific cancer types, including aggressive melanoma associated with vasculogenic mimicry and with trans-differentiation and stem-like phenotype [43
]. VE-cadherin expression in MCF7 cells leads to a change in phenotype, both morphologically and functionally. MCF7 co-culture with ECs leads to morphologic changes towards an elongated cell shape, transient displacement of β-catenin from the cell–cell junction, an increase of vimentin expression, an increase of VE-cadherin expression, and a concomitant decrease in E-cadherin expression. This cadherin switch may contribute to vascular mimicry as it enables TCs to interact with ECs and to integrate into the endothelial monolayer and consequently increase TC invasion, as confirmed by the CAM assay.
Our data reveal the potential mechanisms of how TCs acquire VE-cadherin expression. We discovered that VE-cadherin could be transferred from ECs via EVs to cancer cells. EVs contain different cytosolic and membrane proteins derived from the parent cell. They can transfer functional proteins, nucleic acids, and lipids between cells in vitro and in vivo, and therefore are capable of changing the composition and function of recipient cells. Exosomes are the smallest subset of EVs, with a size ranging from 30–150 nm, and apoptotic bodies are the largest of all EVs (up to 5000 nm) and are released as membrane blebs of cells undergoing apoptosis [44
]. By labeling HUVECs with GFP-tagged VE-cadherin or GFP, we showed that EVs of different sizes originate from HUVECs and are incorporated into MCF7 and MDA-MB-231 cells. EVs generated by ECs could induce the expression of VE-cadherin in breast cancer cells. In target cells, EVs can influence the physiology of cells by transcription of competent RNA molecules [45
]. Moreover, EVs can transport proteins related to essential signaling pathways such as mitogen-activated protein kinase (MAPK), nuclear factor κB (NFκB), and protein kinase B (AKT) to the recipient cells. These proteins and miRNAs play essential roles in processes associated with cell proliferation [46
Among the mutual interactions between TCs and ECs, we pinpointed a second VE-cadherin transfer mechanism, in which material transfer from ECs occurs due to the influence of TCs on EC viability, as highlighted in our co-culture experiments, and previously suggested by others [47
]. Our results demonstrated that TCs cause programmed necrosis (necroptosis) [47
] of ECs and that debris of HUVECs are then incorporated by TCs via efferocytosis, thereby enabling the TCs to obtain EC-typical proteins such as VE-cadherin. Prior to efferocytosis, TCs that are in contact with ECs upregulate the mannose receptor, which is a crucial receptor that enables TCs to endocytose cell debris and material. For this reason, the mannose receptor is a potential cell maker for this process and could be used to identify TCs capable of performing efferocytosis and, as a consequence, receive EC-secreted VE-cadherin-containing material.
Lastly, we could identify a third mechanism by which TCs receive VE-cadherin, which is not the transmembrane full-length form of VE-cadherin, but the soluble VE-cadherin ectodomain with its characteristic molecular mass of 90 kDa. This VE-cadherin side product is a consequence of shedding by enzymes such as ADAM10, as previously shown [48
]. The generation of sVE-cadherin was associated with inflammation-induced breakdown of endothelial barrier functions [48
]. Interestingly, the presence of full-length VE-cadherin could only be found in the supernatant of the co-culture and the lysate of HUVECs, whereas sVE-cadherin was also detected in the supernatant of monocultured HUVECs. Obviously, the TCs can influence the composition of EVs, which are released by the ECs. Interestingly, only the co-culture supernatant with the TC-induced EVs of HUVECs promoted the expression of the VE-cadherin gene in MCF7 cells. While full-length VE-cadherin could be detected in TC lysates for at least 1 week, the sVE-cadherin was transiently detected and was lost after 24 h of treatment with HUVEC supernatant. Mechanistically, sVE-cadherin does not seem to influence TCs directly but can destabilize the ECs monolayer and indirectly contribute to a more invasive TC phenotype [48
Although it is not clear if these different mechanisms of EC-induced VE-cadherin expression in TCs happen simultaneously or sequentially, the release of full-length VE-cadherin depends on juxtacrine cell communication between ECs and TCs. If not direct cell–cell contacts, at least a close proximity of ECs and TCs are required for the EC-induced VE-cadherin expression and the subsequent cadherin switch in TCs. These conditions are met in the TME.
4. Materials and Methods
4.1. Cell Culture
Breast carcinoma cell lines (MCF7, T47D, HCC1806, HCC1937, BT549, MDA-MB-453, SUM149, MDA-MB-468, SKBR3, BT20, and BT549) were kindly provided by Dr. M. Götte (Department of Gynecology and Obstetrics, Münster, Germany) and Dr. B. Greve (Department of Radiation Oncology, Münster, Germany). Cells were maintained in DMEM/high glucose medium (Lonza, Basel, Switzerland) (SKBr3, MDA-MB-453, and MDA-MB-468) or in RPMI medium (MCF7, T47D, HCC1806, HCC1937, BT20, and BT549). Both media were supplemented with 10% fetal calf serum (FCS; Gibco, Waltham, MA, USA) and 100 U/mL penicillin–streptomycin (PS; Gibco). SUM149 cells were cultured in Dulbecco’s modified Eagle’s medium-F12 (1:1) with 5% FCS, insulin (5 μg/mL) (Merk, Darmstadt, Germany), and hydrocortisone (1 μg/mL; Qiagen, Hilden, Germany). MDA-MB-231 cell line was obtained from DSMZ (Leibniz Institute DSMZ–German Collection of Microorganisms and Cell Cultures) and was grown in DMEM/high glucose medium containing 10% FCS and 100 U/mL PS. HUVECs were kindly provided by Dr. D. Vestweber (Max Planck Institute of Molecular Biomedicine, Münster, Germany) or purchased from Promocell (Heidelberg, Germany) and cultured up to passage five in ECGM-2 medium supplemented with SupplementPack (PromoCell).
4.2. PiggyBac Transposon-Based Reporter Expression in MCF7 Cells
MCF7 cells were trypsinized and centrifuged at 200× g for 10 min. One million cells were resuspended in 100 μL Nucleofector Kit V reagent (Lonza, Basel, Switzerland). Then, the cells were mixed with 0.5 μg transposase plasmid and 5 μg transposon plasmid containing VE-cadherin promoter fragments from positions −3394 to +39 followed by tdTomato, kindly provided by Dr. I. Slukvin (Department of Pathology and Laboratory Medicine, Madison, WI, USA), and were then electroporated using program Q-001 according to the manufacturer’s protocol (Amaxa, Cologne, Germany). The electroporated cells were resuspended with RPMI growth medium in a 6-well plate. Selection with 100 μg/mL zeocin (Invitrogen, Toulouse, France) began 3 days after electroporation.
4.3. Lentiviral Transduction of Target Cells and Generation of Stable Cell Lines
Viral particles were produced by transient co-transfection of 293T cells with lentivirus encoded in the psPAX2 plasmid, the envelope elements from the SVS (somatitis virus) encoded in the pMD2.G plasmid, and the vector genome encoded in the pLV-CMV-MCS-SV40-Puro and pLenti-puro transfer plasmids, with GeneJammer transfection reagents (Agilent, Waldbronn, Germany). Conditioned medium containing lentivirus was harvested 72 h after transfection, cleared by low-speed centrifugation, filtered through 0.45 μm pore-size cellulose acetate filters, and concentrated with Lenti-X concentrator (Clontech, Saint-Germain-en-Laye, France). For transduction, cells were seeded in a 6-well plate and incubated overnight in 2 mL culture medium. Concentrated virus particles in 100 μL suspension with 10 μg/mL polybrene were diluted in 500 μL ECGM-2 medium, added to the cells, and incubated for 24 h. After 48 h, cells were selected with 1 μg/mL puromycin (Toku-E, Sint-Denijs-Westrem, Belgium). Transduction efficacy was assessed by flow cytometry analysis. Plasmids psPAX2 and pMD2.G were kindly provided by Dr. D. Trono (Lausanne, Switzerland). Transfer plasmids (pLenti-EGFP, pLenti-mCherry, and pLenti-LifeAct-mCherry) were generous gifts from Dr. S. Huveneers (AMC, Amsterdam, The Netherlands), and pLVX-IRES-Puro-human VE-cadherin-EGFP was provided by Dr. D. Vestweber (Max Planck Institute of Molecular Biomedicine, Münster, Germany).
4.4. Analysis of Cell–Cell Interaction at Single-Cell Resolution in a 3D Environment
MCF7 cells (MCF7-mCherry or MCF7-lifeact-mCherry cell lines) and HUVEC-GFP cells were co-encapsulated in hydrogel beads as described previously [49
]. Briefly, for the co-encapsulation, 75 μL of HUVEC cell suspension (16,000 cells μL−1
) and 75 μL of MCF7 cell suspension (13,000 cells μL−1
) were mixed in phosphate-buffered saline (PBS). Cell suspension (150 μL) was then mixed with 150 μL 3% (w/v) SeaPrep agarose (Lonza). Agarose droplets containing cells were generated in a bead formation chip by using syringe pumps (EVORION Biotechnologies, Münster, Germany). After agarose droplet gelation, beads containing cells were transferred and immobilized into a trapping chip by using a control unit (EVORION Biotechnologies). After immobilization of hydrogel beads, the trapping chip was perfused with medium (ECGM-2 and RPMI mixture (50:50) supplemented with 20% FCS and 20 mM HEPES). TC–EC interaction was monitored by time-lapse confocal microscopy (LSM800) at 37 °C with 10× magnification (Figure S7
). Images were taken with 1 h time intervals over a total of 61–68 h. Images were analyzed with the open-source software CellProfiler (www.cellprofiler.org
). A Cell Profiler pipeline consisting of several individual imaging analysis modules was constructed to identify and quantify objects in every image automatically. The pipeline first segments the images into objects (MCF7-mCherry or MCF7-lifeact-mCherry cells—red channel, and HUVEC-GFP cells—green channel) and then makes a comparison between the objects in each channel to measure colocalization. The detected MCF7 cells were marked as the region of the interest, and the area of GFP pixel (HUVEC-GFP), as well as the number of green particles, were calculated within the region of interest to measure the HUVEC-derived extracellular vesicles taken up by MCF7 cells.
4.5. Homo- and Heterospheroid Formation
Spheroids were formed as previously described [51
]. The cancer cells used in this assay—MCF7-mCherry, MDA-MB-231-mCherry, T47D-mCherry, BT549-mCherry, MCF7-GFP, and MCF7-VE-GFP—were co-cultured with HUVEC-GFP cells of passage 2–3. Each spheroid contained 5000 cells. Cells were resuspended in a solution composed of one-quarter volume of a 6 mg/mL methylcellulose (Sigma-Aldrich, Deisenhofen, Germany) solution and three-quarter volume of cell culture media. This solution was distributed in a round-bottom 96-well plate, 100 μL per well, and incubated at 37 °C for 24 h. For heterospheroid formation, cancer cells were mixed in a 1:2 ratio with HUVECs. Images were acquired using a confocal microscope (LSM 700, Zeiss, Oberkochen, Germany) after 24, 48, and 72 h.
4.6. Chick Chorioallantoic Membrane (CAM) Assay with MCF7 and MDA-MB-231 Cells
Freshly fertilized chicken eggs were purchased from Brinkschulte GmbH (Senden, Germany). Eggs were incubated for 72 h at >60% relative humidity and 37 °C. On developmental day 3, the eggs were cracked open, and the embryo was carefully transferred into a plastic square weighing boat (89 × 89 × 25 mm). The weighing boat was placed in a round transparent glass Petri dish of 100 × 20 mm, and 40 mL of purified water was added to ensure sufficient humidity. On day 10, 106
cancer cells (MCF7-GFP, MCF7-VE-GFP, or MDA-MB-231-GFP) were embedded in collagen gel, as described previously [52
]. A Teflon ring was placed on the CAM membrane, and the embedded cancer cells were grafted into the ring. The images of the region containing the disc on the CAM were taken with the stereomicroscope (Nikon, Tokyo, Japan). The tumors were dissected on day 17 of chick development. Tumors were fixed in 4% formaldehyde for 15 min, washed thrice in PBS, and transferred into 10% sucrose for 3 h at 4 °C and 30% sucrose overnight at 4 °C. Tumors were then embedded in tissue-freezing medium and cut with a cryotome (Leica Biosystems, Leica Biosystems, Wetzlar, Germany) into 8 μm thick sections. The experiments were performed according to the guidelines of the European Parliament (2010/63/EU) and the council for the protection of animals in research (§14 TierSchVersV).
4.7. Immunofluorescence Staining of Cells and Frozen Tumor Sections
Cells (30 × 103) were seeded on 8-well chamber ibiTreat Surface slides (ibidi, Martinsried, Germany), and incubated at 37 °C and 5% CO2. Cells were fixed with freshly prepared 2% formaldehyde dissolved in PBS, and 24 h later were washed with PBS for 10 min, then permeabilized with 0.1% Triton X-100 for 10 min at 4 °C. Cells were stained with the primary antibody diluted in blocking buffer (2% horse serum and 1% bovine serum albumin in PBS) at 4 °C, overnight. After washing three times, cells were incubated with Alexa Fluor-conjugated secondary antibodies (1:200; Invitrogen, Karlsruhe, Germany) for 1 h at room temperature. Nuclei were stained with 200 ng/mL DAPI (4′-6-diamidino-2-phenylindole; Sigma, Deisenhofen, Germany) in PBS. The cells were finally washed twice in PBS and mounted in a fluorescent mounting medium (Dako, Hamburg, Germany). Pictures were acquired using a confocal microscope LSM800 with an oil immersion 40× objective (Zeiss, Oberkochen, Germany). For fluorescent staining of frozen tissue, frozen 8 μm sections were rehydrated in PBS, pH 7.3, for 5 min, and stained as described above.
The primary antibodies were goat anti-human VE-cadherin (1:50; Santa Cruz Biotechnology, Inc., Dallas, TX, USA), mouse anti-cadherin 11 (1:50; Invitrogen. Eugene, OR, USA), mouse anti-E-cadherin (1:100; BD Biosciences, San Jose, CA, USA), mouse anti-β-catenin (1:100; BD Biosciences), rabbit anti-N-cadherin (1:50; Abcam, Cambridge, MA, USA), mouse anti-vimentin (1:50; BD Biosciences), rabbit anti-mannose receptor antibody (1:50; Abcam), rat anti-NRP1 Ab (1:20; Pineda, Berlin, Germany), mouse Anti-CD63 (1:50; Abcam), and rabbit anti-VEGF receptor II (1:20; Abcam).
4.8. Immunoblot Analysis
An equal number of cells was plated in 6 cm dishes and cultivated for 48 h. The cells were first rinsed with cold PBS. Then, 250 μL of lysis buffer (10mM Tris, 1mM EDTA, 150mM NaCl, 0.5% NP-40) supplemented with Complete protease inhibitor cocktail (Roche, Basel, Switzerland) were added to the cells and incubated on ice for 5 min. Cells were scraped off on ice, and cell lysates were collected. The protein concentration was determined with the BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Lysate proteins were separated by SDS-PAGE in a 10% polyacrylamide gel and transferred by wet blotting onto nitrocellulose membranes (Whatman, Dassel, Germany). Membranes were incubated with the following primary antibodies: goat anti-VE-cadherin (C-19) (1:100; Santa Cruz, Dallas, TX, USA), mouse anti-VE-cadherin (BV9) (1:100; Santa Cruz), mouse anti-Cadherin 11 (1:50; Invitrogen), mouse anti-E-cadherin (1:100; BD Biosciences), or mouse anti-vimentin (1:50; BD Biosciences), followed by incubation with HRP-conjugated secondary IgG (1:1000; Dako, Hamburg, Germany).
For detection of immunoblots with fluorescently labelled antibodies, we incubated membranes with IRDye 600 CW and 800 CW infrared secondary anti-mouse- and anti-rabbit-IgG antibodies (1:10,000, Li-COR Biotechnology, Bad Homburg, Germany). Immunoreactive bands were detected with the Li-COR Infrared Reading System according to the manufacturer’s instructions (Li-Cor Odyssey Infrared Reading System, Homburg, Germany).
4.9. RNA Isolation and Reverse Transcription PCR Analysis
Total RNA was isolated from cell lysates using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and was reverse transcribed using a Reverse Transcriptase Kit (Qiagen). Real-time PCR reactions were performed with a RotorGene SYBR Green PCR Kit in the RotorGene Cycler (both Qiagen). The following primers were used: E-cadherin-fw, 5′- GACCGGTGCAATCTTCAAA -3′ and E-cadherin-rev, 5′-TTGACGCCGAGAGCTACAC-3′; cadherin-11-fw, 5′-TTGGTCACTCAACAAATGACAA-3′ and cadherin-11-rev, 5′-GTTGCGTCCACCCTCAAG-3′; VE-cadherin-fw, 5′-CATCTTCCCAGGAGGAACAG-3′ and VE-cadherin-rev, 5′-AGAGCTCCACTCACGCTCAG-3′; VEGFRI-fw, 5′-TTTGCCTGAAATGGTGAGTAAGG-3′ and VEGFRI -rev, 5′-TGGTTTGCTTGAGCTGTGTTC -3′; VEGFRII-fw, 5′-GGCCCAATAATCAGAGTGGCA-3′ and VEGFRII-rev, 5′-CCAGTGTCATTTCCGATCACTTT-3′; NRP1-fw, 5′-TTGCAGTCTCTGTCCTCCAA 3′ and NRP1-rev, 5′-GAAAAATGCGAATGGCTGAT-3′; and TOP1-fw, 5′-CCAGACGGAAGCTCGGAAAC-3′ and TOP1-rev, 5′-GTCCAGGAGGCTCTATCTTGAA -3′. Cycle threshold (Ct) values were normalized by the ∆∆Ct method [53
], and TOP1 was used as a housekeeping gene.
4.10. Electric Cell-Substrate Impedance Sensing
HUVECs were seeded onto L-cysteine-reduced, fibronectin-coated 8W10E electrodes (Applied Biophysics, Troy, NY, USA). Electrical impedance was measured at 4000 Hz in real time at 37 °C and 5% CO2 using the ECIS ΖΘ system (Applied Biophysics, Troy, NY, USA). After 24 h, 10 × 103 cancer cells (MCF7 or MDA-MB-231) were added to each well on the confluent HUVEC layer for 72 h.
4.11. Endothelial Tube Formation
This assay was performed as described previously [54
]. In a μ-Slide angiogenesis chamber (ibidi), 10 μL of Matrigel were solidified at 37 °C within 30–60 min. On this Matrigel, 5 × 103
HUVECs were cultured at 37 °C in ECGM-2 medium (PromoCell, Heidelberg, Germany), and tube formation was recorded for 6–18 h with images acquired every 10 min (IncuCyte ZOOM, Essen BioScience, Welwyn Garden City, United Kingdom). To set up the tumor–endothelial cell tube formation, we mixed cancer cells (MCF7 or MDA-MB-231) with HUVECs in a 1:1 ratio and seeded them onto Matrigel-coated ibdi chambers.
4.12. Flow Cytometry
After co-culturing, cells were harvested with accutase according to the manufacturer’s instructions (Chemicon, Millipore, Darmstadt, Germany). Cell sorting and analysis were performed on a FACSAria IIIu cell sorter (BD Biosciences) using an 85 µm nozzle. Strict forward scatter pulse height vs. pulse width gating was used to exclude cell doublets from the analysis. Flow cytometric data were analyzed using FlowJo software (BD Biosciences).
4.13. Apoptosis Assays
After the TCs (MCF7-mCherry or MDA-MB-231-mCherry) were co-cultured with HUVEC-GFP cells over different time points (24, 48, and 72 h), we harvested cells with accutase according to the manufacturer’s instructions (Chemicon, Millipore, Darmstadt, Germany), centrifuged them at 250× g for 4 min, and resuspended them in the 1× annexin V binding buffer (BD Biosciences). Cells were stained with Alexa Fluor 647-labeled annexin V (1:20, Biolegend, San Diego, CA, USA) and 1.43 μM DAPI to distinguish apoptotic from necrotic cells. All samples were analyzed on a FACSAria IIIu cell sorter (BD Biosciences). Data analysis was performed using FlowJo software (BD Biosciences).
4.14. Exosome Isolation from Cell Culture Medium (CCM)
Exo-spin precipitation was carried out according to the manufacturer’s instructions (Cell Guidance Systems, Cambridge, United Kingdom). Briefly, 10 mL cell culture medium was collected and centrifuged at 300× g for 10 min to remove cells. The supernatant was then transferred to a new centrifuge tube and spun at 16,000× g for 30 min to remove any remaining cell debris. Exo-spin buffer at a 2:1 ratio was mixed with clarified CCM and incubated overnight at 4 °C. The sample was then spun at 20,000× g for 30 min, the supernatant was discarded, and the pellet was resuspended in 100 µL of PBS. The sample was further purified using the provided columns, and exosomes were eluted in 200 µL of PBS.
Statistical analyses were performed using GraphPad Prism 6 software (GraphPad Software, La Jolla, CA, USA). Comparisons between two groups were conducted with Student’s t-test, and analyses among more than two groups were performed using one-way ANOVA followed by Tukey’s post hoc test. Differences with p values < 0.05 were considered significant.