Protein expression and purification
Total mRNA was isolated from HUVECs with TRI Reagent (Thermo Fisher Scientific, Waltham, MI, USA); and cDNA was generated from the mRNA with Hight-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. Human full-length TRAX  and the C-terminal domain of PLCβ1 (PLCβ1-C)  were expressed with N-terminal modifications. The cDNA fragments encoding the modified TRAX and PLCβ1 were prepared from HUVECs and produced by PCR using the primers (TRAX F′-primer: AAGAATTCGAATGAGCAACAAAGAAGGATCAGG; R′-primer: TTAAGCTTCTAGTGGTGGTGGTGGTGGTGGTGGTGGTGAG AAATGCCCTC TTCTTGATC; PLCβ1C F′-primer: AAGGATCCGA GCACCTGCCAAAACAGAAG; R′-primer: TTAAGCTTCTAGTGGTGGTG GTGGTGGTGGTGGTGGTGGATCTTTCCTTTCATGGCTTC) and inserted into the EcoRI/HindIII and BamHI/HindIII-digested pET-21d (Addgene, Watertown, MA, USA) expression vectors, respectively. The F′-primers were designed to add nine histidine residues at the N-terminus to facilitate purification. The expression plasmids were introduced into Escherichia coli strain BL21 (DE3) pLys (NEB, Ipswich, MA, USA).
To produce the plasmid pPET-21d-TRAXQ219A, Q223A for mutation experiment, site-directed mutagenesis was carried out using standard full plasmid amplification by DpnI-based PCR strategy with KAPA HiFi HotStart DNA polymerase (Roche, Basel, Switzerland) using pET-21d-TRAX as template. In all cases, fidelity of the amplified DNA was verified by DNA sequencing.
The human recombinant wild-type TRAX, the TRAXQ219A, Q223A protein, and the PLCβ1-C were expressed and extracted from E. coli, and purified by ÄKTA pure chromatography system (Cytiva, Marlborough, MA, USA). Extract was centrifuged and the soluble fraction was applied to a HisTrap FF column (5 mL) (Cytiva) equilibrated with buffer A (50 mM potassium phosphate, pH 7.4, 500 mM sodium acetate, 20% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, 1% sodium cholate, 1% Tween 20 and 0.1 mM phenylmethanesulfonyl fluoride); and then the column was washed with 75 mL of buffer A containing 40 mM imidazole and with 20 mL of buffer B (50 mM potassium phosphate, pH 7.4, 20% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, 40 mM imidazole, 1% sodium cholate, 1% Tween 20, 0.1 mM ATP, and 0.1 mM phenylmethanesulfonyl fluoride). Proteins were eluted with buffer C (200 mM imidazole acetate, pH 7.4, 20% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, 1% sodium cholate, and 1% Tween 20). The eluted fractions were combined and diluted with five volumes of buffer D (20 mM potassium phosphate, pH 7.4, 20% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, and 1% sodium cholate) and applied to a DEAE-Sepharose column (30 × 50 mm, Cytiva) equilibrated with buffer E (20 mM potassium phosphate, pH 7.4, 20% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, 10 mM imidazole, 1% sodium cholate, and 0.1% Tween 20). Finally, the column was washed with 40 mL of buffer E. Pass-through fractions were then applied to an SP-Sepharose Fast Flow column (30 × 40 mm) (Cytiva) equilibrated with buffer F (20 mM potassium phosphate, pH 7.4, 20% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, 10 mM imidazole, and 1% sodium cholate). The column was washed with 20 mL of buffer F and eluted with a 0–125 mM NaCl gradient in buffer G (40 mM potassium phosphate, pH 7.4, 20% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, 10 mM imidazole, and 1% sodium cholate). The major eluted fractions were combined, repeatedly concentrated, and diluted using a centrifugal tube (Vivaspin 20) (Cytiva) to replace the buffer with buffer H (50 mM potassium phosphate, pH 7.4, 20% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, 1% sodium cholate, and 0.05% Tween 20).
GSLs and antibody
GHCer was prepared by modification of previously reported methods . In brief, lactosyl sphingosine was assembled with required sugars by enzymatic synthesis to give Globo H-Sph, and SSEA3-Sph which were coupled with fatty acid, respectively, to yield GHCer (molecular weight: 1535.81) and SSEA3Cer (molecular weight: 1389.67).
Galactosylceramide, lactosylceramide, and Gb4-ceramide were purchased from Matreya LLC (State College, PA, USA). All GSLs in experiments were dissolved in PBS and sonicated right before use. The mAb VK9 hybridoma which was kindly provided by Dr. Ragupathi, was used to produce anti-GHCer antibody .
Enzyme-linked immunosorbent assay (ELISA)
The wells of microtiter plate (Nunc Maxisorp™) (Thermo Fisher Scientific) were coated with 50 μL of GHCer (10 μg/mL in ethanol) or TRAX (10 μg/mL in PBS), followed by incubation overnight at 4 °C. The plate was washed three times with PBS and blocked with 200 μL/well of 1% BSA in PBS for 1 h at room temperature. Empty wells were also blocked as controls. The plate was washed again and incubated with 50 μL/well of TRAX (10 μg/mL to 15.6 ng/mL) at GHCer coated-wells or GHCer (2 μg/mL to 15.6 ng/mL) at TRAX coated-wells for 1 h at room temperature. The plate was washed three times with PBS-T (0.05% Tween 20) and incubated with anti-TRAX antibody or anti-PLCβ1 for 1 h at room temperature. The plate was washed three times with PBS-T before addition of alkaline phosphatase (AP)-conjugated goat anti-mouse IgG diluted 1/4000 in PBS-T and incubation for 45 min at room temperature. The plate was washed again with PBS-T and incubated for 20 min at 37 °C with 100 μL/well of substrate pNPP (Sigma-Aldrich, Darmastadt, Germany). The reaction was stopped with 50 μL of 3 N NaOH and the optical density was measured at 420 nm.
Biacore assays for protein-glycan and protein–protein interactions were performed using Biacore X platform (Cytiva). Generally, proteins were immobilized using a standard amine-coupling protocol provided by the manufacturer. The carboxylic acid groups on CM5 chips (Cytiva) were first activated using a mixture of EDC and NHS solutions (Cytiva) at 25 °C for 7 min at a flow rate of 10 μL/min. Subsequently, about 5 μg of recombinant TRAX protein dissolved in 100 μL of sodium acetate buffer (10 mM; pH 5.0) was injected, resulting about 2500-RU responses of proteins which were covalently immobilized on the chip. Finally, the chip was deactivated by ethanolamine and applied to the Biacore X system for binding assays. HBS-P buffer (Cytiva) was used as the running buffer; a 1-min pulse of 0.005% (w/v) sodium dodecyl sulfate dissolved in the running buffer was used to regenerate the chip surface. GSLs were dissolved in methanol as stock solutions (1 mM) and were diluted with HBS-P buffer before Biacore assays. Binding affinities determined by Biacore system were calculated using BIAevaluation Software (version 4.1) (Cytiva).
Cell culture of HUVECs
HUVECs were obtained from Lonza and maintained in EGM-2 medium (Lonza, Basel, Switzerland) supplemented with 5% fetal bovine serum (Gibco, Waltham, MI, USA), 3 ng/mL bFGF, 5 U/mL heparin, 100 U/mL penicillin, and 0.1 mg/mL streptomycin (Gibco). All cells were cultured for less than 10 passages after resuscitation for experiments.
Measurement of intracellular free calcium mobilization
Intracellular free calcium mobilization in HUVECs was determined with the fluorescent calcium indicator fluo-4/AM (Thermo Fisher Scientific) . Briefly, 4 × 104 HUVECs were seeded into 96-well black plates with clear bottom plates (Corning, NY, USA) and loaded with fluo-4/AM (40 μM) (Thermo Fisher Scientific) in serum-free EGM-2 at 37 °C for 30 min. The cells were then washed twice with calcium-free Locke solution (158.4 mM NaCl, 5.6 mM KCl, 1.2 mM MgCl2, 0.2 mM EGTA, 5 mM HEPES, and 10 mM glucose, pH 7.3) to remove extracellular dye. Probenecid (2.5 mM) (Sigma-Aldrich) was added to both the loading medium and the washing solution to prevent dye leakage. Cells were exposed to 60 μM GHCer or SSEA3Cer. Fluorescence intensity was determined with a Victor3 (Perkin Elmer, Waltham, MA, USA) using an alternative wavelength time scanning method.
Tube formation assay
HUVECs were cultured for 4 h under serum-free conditions. Then, growth factor reduced Matrigel (BD Discovery Labware, Bedford, MA, USA) was placed in 12 well plates (Corning) and allowed to polymerize for 30 min at 37 °C. Next, 6 × 104 cells per well were seeded on Matrigel and stimulated overnight in the presence of GHCer, SSEA3Cer, or PBS. Images were taken under microscope (LECIA DFC310 FX, Wetzlar, Germany) in 5–8 fields per well and analyzed by Image J software (National Institute of Health, Bethesda, MA, USA). By using Angiogenesis Analyzer (Gilles Carpentier), a toolset of Image J software, tube area, total length of tube-like structures, number of branches, and number of meshes were quantified on phase contrast images of HUVECs after 16 h of culture.
Total RNA from the cells was purified with a TRIzol Reagent (Thermo Fisher Scientific) following the manufacturer’s instruction. The purity and quantity of the RNA were measured with a spectrophotometer, Nanodrop (Thermo Fisher Scientific). RNA was subjected to high-capacity cDNA reverse transcription kit (Applied Biosystems, Waltham, MA, USA). The cDNA product was amplified and quantified with 7500 Real-time PCR system (Applied Biosystems) using SYBR® Green PCR Master Mix (Applied Biosystems). The primer sets used for PCR amplification were listed below: Vascular endothelial growth factor (VEGF) A F′-primer: TTGCCTTGCTGCTCTACCTCCA; R′-primer: GATGGCAGTAGCTGCGCTGATA. KDR F′-primer: GGAACCTCACTATCCGCAGAGT; R′-primer: CCAAGTTCGTCTTTTCCTGGGC. FGF13 F′-primer: GGGTCAAACTCTTCGGCTCCAA; R′-primer: GGTGCCATCAATGGTTCCATCC FGF2 F′-primer: AGCGGCTGTACTGCAAAAACGG; R′-primer: CCTTTGATAGACACAACTCCTCTC. The thermal cycling program consisted of 2 min at 50 °C, 10 min at 95 °C, followed by 40 cycles for 15 s at 95 °C, and 1 min at 60 °C. The value was normalized with the ratio of mRNA of the target gene to mRNA of the internal reference gene, GADPH, in each sample. Fold change was calculated as the ratio of the normalized values of the cells as compared to the PBS treatment.
To generate TRAX-GSL binding modes, molecular docking of GSLs against TRAX was performed using Dock (version 5.1.1; https://dock.compbio.ucsf.edu) software . Subsequently, HotLig  software was used to evaluate TRAX–GSL interactions. To predict the binding pocket for GSLs, the cavities on TRAX were first detected using PscanMS  software, then these cavities were subjected to analysis using Dock software for docking calculations. Kollman partial charges were applied to protein models for the force field calculations with Dock. Structures of GSLs were first created using Marvin software (version 5.2.2; https://chemaxon.com/marvin) (ChemAxon, Budapest, Hungary) and the three-dimensional coordinates with energy optimization were built using Balloon software (version 0.6; http://users.abo.fi/mivainio/balloon/index.php) . Additionally, the Gasteiger partial charges on the atoms of GSLs were calculated using OpenBabel (version 2.2.3; http://openbabel.org/wiki/Main_Page) . In addition, we also used MD simulation (see below) to generate multiple conformations of GHCer beforehand for docking with TRAX. The parameters for Dock were set to generate 2000 orientations and 200 conformers per cycle of conformational search in the binding pocket with the parameter of “anchor size” set to 1. The GSL conformers docked to TRAX were then scored and ranked by HotLig  to predict molecular interactions between GSLs and TRAX. The figures for molecular modeling and interactions were generated using Chimera (version 1.16; https://www.cgl.ucsf.edu/chimera) .
Studies of MD were performed using GROMACS (version 4.5.7-1; https://www.gromacs.org)  run on a CentOS (release 6.5; https://www.centos.org) Linux system. We used CHARMM27 force field for the generation of topologies for protein structures, processing of energy minimization, and MD simulations. Structures of GSLs were built using doGlycans packages (https://bitbucket.org/biophys-uh/doglycans) ; the topologies with CHARMM force fields were generated using SwissParam (https://www.swissparam.ch) . Subsequently, the GSLs or the protein-GSL complexes were solvated with water using the TIP3P model defined in GROMACS, forming an in-solution system for simulation. Sodium ions were then added to neutralize the system according to the charges of the complexes.
First, the molecular structures in the system were refined by an energy minimization process until the maximum force lower than 100 kJ/mol/nm. Position restrained MD were performed for 20 ps to equilibrate the distribution of the water molecules. Subsequently, MD for the whole system were simulated at a temperature of 300 K, with 500 steps per ps for simulation of atomic motion. The coordinates of molecules were written to trajectory files every 2 ps for analysis of conformational changes and molecular interactions. VMD software (http://www.ks.uiuc.edu/Research/vmd/) was used for making movies of molecular trajectories.
We examined FRET analysis by confocal microscope (Leica) to study the interaction of PLCβ1 and TRAX . Briefly, HUVECs were incubated with or without 20 μM GHCer or SSEA3Cer for 3 h, 100 μg EVs, EVs + mAb VK9 (5 μg), or EVs + isotype antibody for 16 h, at 37 °C then washed twice in cold PBS. After fixation and permeabilization, cells were stained for PLCβ1 with unlabeled mouse antibody (Santa Curz, Dallas, TX, USA) and a saturating amount of donor Alexa488 labeled anti-mouse IgG (Biolegend, San Diego, CA, USA). TRAX was detected by rabbit anti-TRAX antibodies (Abcam, Cambridge, UK), followed by saturating concentrations of acceptor Alexa555-tagged anti-rabbit IgG (Biolegend). After staining, the cells were washed and analyzed under a confocal microscope (Leica) with 555 laser power-off.
Human clinical breast cancer specimens were obtained from patients at the time of initial surgery and were fully encoded to protect patient confidentiality. Clinical specimens were utilized under a protocol approved by the Institutional Review Board of the Human Subjects Research Ethics Committee of Chang Gung Memorial Hospital. For GHCer staining, tissue sections were deparaffinized followed by antigen retrieval by autoclaving at 121 °C for 5 min in AR-10 solution (Biogenex, Fremont, CA, USA). Slides were incubated with mAb VK9 (1:100 antibody dilution buffer) (Ventana Medical Systems, Inc., Oro Valley, AZ, USA) overnight at 4 °C, followed by antibody detection using a polymer-HRP IHC detection system (Biogenex). The slides were counter stained with hematoxylin and mounted. Digital images were captured using an Aperio ScanScope XT Slide Scanner (Aperio Technologies, Vista, CA, USA) under 20× magnification. The expression of Globo H and the morphology of tumor blood vessels were confirmed by pathologists.
EV isolation, quantitation, and characterization
MCF-7 (Bioresource Collection and Research Center, Hsinchu, Taiwan) cells were cultured and grown in 15 cm dishes (Corning). When cells reached approximately 80% confluence, cells were washed with phosphate buffered saline (PBS) (Gibco) and transferred to fresh media containing 1% EV-free serum (Gibco) for 24 h. Media was then collected and subjected to centrifugation at 500×g for 5 min to pellet cells, then 2000×g for 15 min to remove cellular debris (Eppendorf, Hamburg, Germany). The supernatant was then transferred to ultracentrifuge tubes (Beckman Coulter, Pasadena, CA, USA) and ultracentrifuged (Beckman Coulter) at 10,000×g for 30 min twice to pellet and remove large vesicles. The supernatant was then transferred to new ultracentrifuge tubes and ultracentrifuged twice at 100,000×g for 60 min to pellet EVs. EVs were resuspended in PBS and then stored at 4 °C. Protein content of EVs was determined by BCA (Thermo Fisher Scientific) and treatment dosing was determined by EV protein concentration. Particle size was measured using qNano Gold (Izon, Science Ltd., Burnside, New Zealand) by Tunable Resistive Pulse Sensing. Data was recorded and analyzed using Izon Control Suite Software (version 188.8.131.52). Particle size reported are representative of three separate measurements of EVs.
Transmission electron microscopy and immunogold labelling
EV samples were fixed by adding an equal volume of 4% paraformaldehyde (Merck KGaA, Darmstadt, Germany) to the EV suspension and incubated at room temperature for 10 min. Then, 5 µL of EV suspension were added to carbon-coated 300-mesh copper grids (EMS, Hatfield, PA, USA) for 20 min. Grids were then floated on 100 µL drops of PBS on parafilm for 2 min. Grids were then transferred to 50 µL drops of 1% glutaraldehyde (Merck KGaA) for 5 min. Next, grids were washed on 100 µL drops of PBS for a total of 8 washes and 2 min per wash. Grids were then transferred to 50 µL drops of uranyl oxalate (Sigma-Aldrich), pH = 7.4, for 5 min, and finally to 50 µL drops of 2% methylcellulose (Sigma-Aldrich) and 4% uranyl acetate (EMS) for 10 min on ice. Excess fluid was then blotted off on filter paper; and grids were dried at room temperature for 20 min prior to imaging or storage. Immunolabelling was performed by mounting the concentrated samples on carbon-coated, glow discharged 300 mesh Ni grids (EMS) for 30 s and washed 3 times with PBS. Grids were blocked with 0.5% bovine serum albumin (BSA) (Sigma-Aldrich) in PBS and incubated with primary mAb VK9, anti-GHCer antibody: anti-SSEA3Cer antibody 1:50 in 0.5% BSA in PBS for overnight at 4 °C. After incubation, grids were washed 3 times with PBS and incubated with secondary antibody goat anti-mouse conjugated with 4 nm colloidal gold (Jackson ImmunoResearch, West Grove, PA, USA) 1:20 in 0.5% BSA in PBS. Samples and secondary antibody were incubated for 1 h at room temperature. The grids were then washed with 3 drops of PBS. The grids were finally washed with 3 drops of PBS before staining with 2 drops of methyl cellulose/uranyl acetate and blotted dry. Images were obtained with a transmission electron microscope (HITACHI HT-7800, Krefeld, Germany).
EV fluorescent labeling
EVs were labeled with 0.05 µL anti-CD9-FITC, anti-CD63-PE, or anti-CD81-APC (Thermo Fisher Scientific) in 100 µL PBS for 30 min on ice in the dark. To avoid false positive events, all antibodies used were run in PBS alone to ensure antibody clumps were not present. To avoid carry‐over effects between each sample measurement we performed a washing step with filtered double distilled water for 1 min at an increased flow rate of 60 µL/min. Flow cytometry analysis was performed on CytoFLEX LX Flow Cytometer (Beckman Coulter). Data is analyzed by FlowJo™ V10 software (BD, Ashland, OR, USA).
Matrigel plug angiogenesis in vivo assay
Mice were maintained at the animal facility of Chang Gung University (IACUC number: CGU105-027). Animal studies were conducted by the guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee. Balb/c mice (National Laboratory Animal Center, Taipei, Taiwan), weighing 20 ± 2 g at the beginning of the experiment, were housed in a room maintained at 23 °C with a 12-h light–dark cycle. To determine the sample size of animal experiments, we used power analysis assuming (difference in means)/(standard deviation) is > 2.5. A total of 30 male mice were anesthetized by isoflurane (USP TERRELL, Lexington, KY, USA) inhalation and randomized into different groups, and given a subcutaneous injection of Matrigel (500 μL/injection) (BD) with a 27-gauge needle (BD). Matrigel plus PBS served as a negative control. Matrigel containing GHCer, SSEA3Cer, GHCer + mAb VK9, EVs, or EVs + mAb VK9 was the test substance. Five mice were used for each group. After 14 days, all mice were sacrificed; the Matrigel plugs were carefully dissected out and analyzed for hemoglobin content. The Matrigel plugs were weighed and homogenized for 5 to 10 min on ice. Supernatants (50 μL) were mixed with 950 μL Drabkin reagent (Sigma-Aldrich) and incubated at room temperature for 30 min, and the absorbance was read with an ELISA reader at 540 nm. Matrigel plugs were fixed in formaldehyde and embedded in paraffin. The Matrigel sections were deparaffinized followed by hematoxylin and eosin (H&E) staining and visualized using Aperio ScanScope XT Slide Scanner (Aperio Technologies). The capillary structure was counted. All mice in each group were examined.
Dot-blot assay for determination of GHCer concentration
Supernatants from confluent monolayer cultures of MCF-7 cells (Bioresource Collection and Research Center, Hsinchu, Taiwan) were collected on day 5, centrifuged at 1200×g for 10 min, and passed through 0.22 μm filter. The culture medium was concentrated (Eppendorf) before loading into membrane. The PVDF membrane was activated in MeOH and washed in PBS and blocked for 30 min with 3% BSA in PBS. Membrane was incubated with anti-GHCer mAb VK9. After incubation with alkaline phosphatase conjugated anti-mouse IgG, immune-reactive GSL dots were detected by enhanced chemiluminescence reagents (Amersham Pharmacia Biotech, Amersham, UK) and analyzed by Typhoon (Cytiva). The optical density of dots detected by dot-blotting was calculated with ImageQuantTL (Cytiva).
Bioassays were replicated three times. Data analysis for tube formation and expression of mRNA was conducted by one-way ANOVA in GraphPad Prism (version 9) (San Diego, CA, USA).