Human breast cancer cell lines (MDA-MB-231 and MCF-7) and human normal mammary epithelial cell line MCF-10A were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). B3GALT4-overexpressed or -silenced MDA-MB-231 and MCF-10A cells were previously established in our lab (27). Transfected, parental MDA-MB-231 and MCF-7 cells were cultured in DMEM (Biological Industries, Beit Haemek, Israel) containing 10% fetal bovine serum (Biological Industries) and 1% penicillin/streptomycin (Beyotime, Haimen, Jiangsu, China). Transfected and parental MCF-10A cells were cultured in DMEM/F12 (Gibco, Thermo Fisher Scientific; San Jose, CA, USA) supplemented with 5% horse serum (Thermo Fisher Scientific), hydrocortisone, epidermal growth factor (EGF, Peprotech; Rocky Hill, NJ, USA), cholera toxin, recombinant human insulin (Sigma-Aldrich; St. Louis, MO, USA), and 1% penicillin/streptomycin. All cells were cultured at 37°C in 5% CO₂ atmosphere.

sEV was isolated from culture supernatant as previously described (28). Cells were cultured in exosome-free DMEM medium for 48 h. The conditioned medium was collected and centrifuged at 500 g for 10 min, 2,000 g for 20 min, and 11,000 g for 30 min at 4°C, respectively. The supernatant was collected and subjected to ultracentrifugation at 110,000 g for 70 min (Optima XE-100; Beckman Coulter Life Sciences; Indianapolis, IN, USA). sEV pellets were collected and ultracentrifuged at 110,000 g for 70 min again. sEV pellets were resuspended in 100 μl of PBS and stored at −80°C.

sEV was separated by modified OptiPrep density gradient centrifugation (29). The OptiPrep stock solution [60% (w/v) aqueous iodixanol, Axis-Shield PoC; AS; Oslo, Norway] was diluted to 40%, 20%, 10%, and 5% (w/v), and each component was continuously loaded into 14 × 89 mm Ultra-Clear tubes (Beckman Coulter). sEVs obtained from ultracentrifugation were carefully added to the top layer. Tubes were placed in sw-41 rotor (Beckman Coulter) and centrifuged at 100,000 g for 18 h. Twelve fractions were collected. Each fraction was diluted in PBS, ultracentrifuged at 110,000 g for 3 h, and resuspended with PBS.

TEM assay of sEV was modified based on a previously described procedure (30, 31). Purified sEV was applied to carbon-coated 400-mesh grids (Electron Microscopy Sciences; Fort Washington, PA, USA) for 5 min, washed with PBS, and stained with 1% uranyl acetate (Sigma-Aldrich) for 2 min. The excess liquid was carefully absorbed with filter paper. sEV was observed and photographed by TEM (model H-7650; Hitachi; Tokyo, Japan) at 80 kV.

Cells were lysed with RIPA buffer containing 1% protease inhibitor cocktail, 50 mmol/L Tris-HCl, 150 mmol/L NaCl, 1% sodium deoxycholate, 1% Triton X-100, and 0.1% SDS. Cell lysates were centrifuged, and supernatants were collected. Proteins were loaded and separated by SDS-PAGE. After electrophoresis, proteins were transferred onto polyvinylidene fluoride (PVDF) membranes (Bio-Rad; Hercules, CA, USA). Membranes were blocked with 3% bovine serum albumin (BSA, Beyotime) in TBST for 1 h at 37°C, and probed with primary antibodies overnight at 4°C, and incubated with appropriate HRP-conjugated secondary antibodies. Bands were visualized by enhanced chemiluminescence (ECL; Vazyme Biotech, Nanjing, China). The commercially available antibodies used in this study were listed as follows: primary antibodies against CD63 (#ab134045), TSG101 (#ab83) (Abcam; Cambridge Cambridgeshire, UK), Alix (#2171S), Calnexin (#2679S) (Cell Signaling Technology, Danvers, MA, USA), E-cadherin (#610181) (Becton, Dickinson and Company; Sparks, MD, USA), CD81 (#sc-23962), fibronectin (#sc-271098), vimentin (#sc-6260) (Santa Cruz Biotechnology; Santa Cruz, CA, USA), and GAPDH (#G9545, Sigma-Aldrich).

Cells were digested with trypsin, washed twice with PBS, fixed with 4% fresh paraformaldehyde, blocked with 1% BSA in PBS, and incubated with fluorescein isothiocyanate-conjugated Cholera Toxin B subunit (CTB-FITC, #C1655, Sigma-Aldrich) at 37°C for 30 min. Fluorescence signals were detected and analyzed using flow cytometry (ACEA Biosciences, San Diego, CA, USA).

Cells were cultured until they reached 70%–80% confluence. Cells were digested with trypsin, washed with PBS, resuspended in serum-free medium, and inoculated in the upper chamber of a 0.8-μm transwell filter (Corning; Cambridge, MA, USA). Complete medium was added into the lower chamber. After 24-h culture, migrated cells were dyed with 0.1% crystal violet and photographed under a microscope.

To elucidate the cell–cell adhesion, the hanging drop aggregation was performed as previously described (32). Briefly, cells that reached a confluency of 80% were detached by trypsin and resuspended in complete medium. The single-cell suspension of 2×10⁴ cells in 30 μl of medium was suspended as a hanging drop from the lid of a 6-well plate and allowed to aggregate overnight at 37°C in 5% CO₂. Cells were subjected to shear force by passing through a 200-μl pipette tip 10 times, and photographed after mechanical stress.

To elucidate the adhesion strength between cells and extracellular matrix (ECM), the cell adhesion assay was performed as previously described (33). Briefly, the single-cell suspension of 5×10⁴ cells in 500 μl of medium was inoculated in a 48-well plate, and incubated for 4 h at 37°C in 5% CO₂. Attached cells were rinsed with PBS, stained with 0.1% crystal violet, and photographed under a microscope.

GSLs were extracted and analyzed as previously described (34, 35). Cells (pellets harvested by centrifugation) were extracted twice with 2 ml of isopropanol/hexane/water (55:25:20, v/v/v) with vortexing and sonication, and centrifuged. Extracts were evaporated to dry under nitrogen stream. The residue was incubated with 2 ml of 0.1 mol/L NaOH in methanol at 40°C for 2 h to hydrolyze phospholipids, neutralized with 1 mol/L HCl, and added with 2 ml of hexane. The upper layer was removed from the lower layer. GSLs in the lower phase were dried, solubilized in 1 ml of distilled water, and desalted with SepPak C18 cartridge (Waters Corporation, Milford, MA, USA). Glycan components of GSLs were released by EGCase treatment (36) and subjected to LC-MS analysis.

sEV from breast cancer cells was incubated with CFSE (#21888, Sigma-Aldrich) at 37°C for 30 min in the dark (8), purified by ultracentrifugation at 110,000 g for 70 min to remove excess CFSE, and incubated with MCF-10A cells at 37°C for 30 min. Images were acquired by laser confocal microscopy (model TCS SP8; Leica; Weztlar, Germany). The sEV uptake was quantified by calculating the integrated optical density of the green signal in the field using Image Pro Plus software.

GM1 labeling was performed as previously described (37, 38). Briefly, sEV from MDA-MB-231 cells was incubated with biotinylated CTB (CTB-Biotin, #C9972, Sigma-Aldrich) on ice for 1 h, purified by ultracentrifugation at 110,000 g for 70 min to remove excess CTB-Biotin, and incubated with MCF-10A at 37°C for 6 h. Western blot was performed to explore the transfer of biotin-labeled vesicular GM1 in recipient cells.

All experiments were performed in triplicate and data were analyzed by GraphPad Prism (version 9.0). The experimental data between two groups were analyzed using two-tailed Student’s t-test. Differences with p < 0.05 were considered as statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001.

GSLs are integral components of mammalian cell membranes, and play essential roles in cell adhesion and signaling activation. We have previously demonstrated that GM1 can regulate cell proliferation in human mammary epithelial cells under high-density conditions via EGFR signaling (27). Effects of GM1 on the EMT process and its underlying mechanism need further study. In the present study, immunohistochemical analysis showed that the expression of B3GALT4 was higher in breast cancer tissues than in paired normal tissues (Figure S1A). To investigate the biological function of GM1, glycosyltransferase B3GALT4 (GM1 synthase) gene was overexpressed and silenced in MCF-10A cells, respectively. High level of GM1 in B3GALT4-overexpressed cells and low level of GM1 in B3GALT4-silenced cells were confirmed by flow cytometry (Figure 1A), immunofluorescence (Figure S1B), and Western blot (Figure S1C). B3GALT4-overexpressed cells presented a spindle-like morphology and loose cell-to-cell adhesion. B3GALT4-silenced cells were more rounded and tightly adhered (Figure 1B). The EMT process of MCF-10A was facilitated by high level of GM1, as evidenced by E-cadherin downregulation, and fibronectin and vimentin upregulation, and was suppressed by low level of GM1 (Figures 1C, S1C). β-catenin is a component of cell adhesion complex, and nuclear translocation of β-catenin activates the related transcription factors, transcribes Wnt target genes, and mediates the occurrence of EMT (39). Increased nuclear expression of β-catenin is the hallmark of the activation of Wnt/β-catenin signaling pathway. Immunofluorescence analysis showed that the β-catenin was translocated from the cytoplasm to the nucleus by high GM1 level (Figure 1D).

Cell–cell adhesion was suppressed by B3GALT4 overexpression in MCF-10A cells, as evidenced by sensitivity to relatively strong mechanical stress using cell adhesion assay, and was enhanced by B3GALT4 knockdown (Figure 1E). Furthermore, cell-matrix adhesion was significantly enhanced by B3GALT4 overexpression, but suppressed by B3GALT4 knockdown (Figure 1F). Similarly, the migratory ability of MCF-10A was increased by B3GALT4 overexpression, but reduced by B3GALT4 knockdown (Figure 1G). Together, these results suggested that high GM1 level facilitated the EMT process, accompanied by elevated migratory ability in MCF-10A.

We also investigated the effect of GM1 on the TGF-β-induced EMT process. TGF-β-induced EMT was facilitated by endogenous B3GALT4 overexpression with increased expression of mesenchymal marker and decreased epithelial marker (Figures S2A, B), but not by B3GALT4 silencing. Besides GM1, B3GALT4 is also responsible for the synthesis of GD1b and GT1b. To examine the ganglioside expression pattern in B3GALT4-transfected cell, glycan components of GSLs from parental and B3GALT4-silenced MCF-10A cells were profiled by mass spectrometry (Figure S2C). We found that a total of 9 distinct GSLs were identified in both cell lines. Glycan structures and relative intensity of identified GSLs were summarized in Table S1. These 9 GSLs, including GM1, were all significantly downregulated in B3GALT4-silenced MCF-10A cells compared to parental cells (Figure S2D). To investigate the biological function of GM1 other than other GSLs produced by B3GALT4, MCF10A cells were treated with TGF-β, one typical EMT inducer, and fed with exogenous GM1. The mesenchymal marker expression was increased, and epithelial marker expression was decreased by exogenous GM1 treatment in treated MCF10A cells (Figure S2E). These data showed that GM1 of 9 decreased GSLs is mainly responsible for the EMT process.

It has been reported that secretions derived from tumor cells can induce the EMT process in recipient cells (40). In the present study, we explored the effect of GM1 on the biological function of secretions by treating human normal mammary epithelial MCF-10A cells and breast cancer MCF-7 cells with conditioned medium (CM) from B3GALT4-overexpressed (termed as B3GALT4-CM) and -silenced (shB3GALT4-CM) MDA-MB-231 cells (Figure 2A). GM1 expression in those cells were examined by flow cytometry (Figure 2B) and immunofluorescence (Figure S3A). The migratory ability of MCF-10A and MCF-7 cells was enhanced by B3GALT4-CM, but suppressed by shB3GALT4-CM treatment (Figures 2C, D). Western blot results showed an increased expression of the mesenchymal marker of fibronectin and vimentin, and decreased expression of E-cadherin in B3GALT4-CM-treated MCF-10A and MCF-7 cells, and shB3GALT4-CM-treated cells displayed opposite changes in EMT marker expression (Figures 2E, F).

We speculated that GM1 might be transferred into recipient cells through secretions; thus, levels of GM1 in recipient cells were detected by flow cytometry. Compared to the treatment of CM from MDA-MB-231 cells (termed as MB231-CM), levels of GM1 in recipient cells were significantly enhanced by B3GALT4-CM, but decreased by shB3GALT4-CM treatment (Figures 2G, H). Taken together, these results demonstrated that phenotypes of recipient cells could be influenced by secretions from B3GALT4-overexpressed and -silenced MDA-MB-231 cells.

sEV could mediate exchange of biomolecules derived from parental cells, which regulated biological characteristics of recipient cells (8). GM1 was documented to be located in MSC-sEV (41), and we determined here if GM1 was presented on sEV derived from breast cancer cells. sEV from parental or B3GALT4-overexpressed or -silenced MDA-MB-231 cells (respectively termed MB231-sEV, B3GALT4-sEV, and shB3GALT4-sEV) was isolated by a well-established differential centrifugation method. They both displayed the typical cup-shaped morphology visualized by an electron microscope (Figures 3A, S3B) and clear expression of sEV markers (CD63, Alix, and TSG101) (Figures 3B, S3C). Nanoparticle Tracking Analysis (NTA) showed that there were no significant differences in the size distribution of two kinds of sEV, which both exhibited a diameter of 30–100 nm (Figures 3C, S3D).

Density gradient centrifugation showed that GM1 was presented in CD63- and TSG101-positive fractions (Figure 3D), which were primarily located on sEV by immunoelectron microscopy (Figure 3A). GM1 levels were significantly higher in B3GALT4-sEV than MB231-sEV by dot blot analysis (Figure 3E). sEV was labeled with CFSE and fed to MCF-10A cells for measurement of cellular uptake, and we found that they were both internalized (Figure 3F). Notably, sEV from B3GALT4-overexpressed MDA-MB-231 cells was more efficiently internalized by recipient cells (Figure 3F). However, B3GALT4 silence suppressed sEV internalization (Figure S3E), indicating that sEV uptake might be determined by GM1 levels on sEV. Vesicular GM1 labeled with CTB-Biotin (Figure S3F) was efficiently internalized by recipient cells (Figure S3G), demonstrating that GM1 could be transferred from donor cells to recipient cells via sEV. The expression of CD63 was significantly increased in B3GALT4-overexpressed cells, examined by Western blot (Figure 3G) and flow cytometry (Figure 3H). Since CD63 was documented to be associated with sEV secretion (42), these data might indicate an enhanced sEV secretion in MDA-MB-231 cells. Meanwhile, sEV was isolated from equal numbers of B3GALT4-overexpressed and parental MDA-MB-231 cells with equal amounts of GAPDH in total cell lysates. Western blot demonstrated that the expression of TSG101 in B3GALT4-sEV was significantly higher than in sEV from parental MDA-MB-231 cells (Figure 3I). Collectively, these results showed that GM1 was sorted onto sEV, and elevated GM1 levels might enhance the sEV secretion and facilitate sEV internalization.

Heparanase was documented to enhance sEV secretion, and alter sEV cargos and functions in several human cancer cell lines (43). Our data showed that GM1 might also enhance sEV secretion in MDA-MB-231 cells (Figures 3G–I). To explore the effect of GM1 on sEV secretion, proteomic analysis of parental and B3GALT4-transfected MDA-MB-231 cells was performed. A total of 6,086 proteins were identified, of which 637 proteins were differentially expressed (Table S2). Expression patterns of dysregulated proteins in B3GALT4-transfected cells compared to parental cells were represented in a volcano plot (Figure 4A). A total of 341 proteins were significantly upregulated in B3GALT4-transfected cells, and 296 proteins were significantly downregulated. Correlation analysis demonstrated that the differentially expressed protein MS intensities of parental and B3GALT4-overexpressed MDA-MB-231 cells were highly correlated among biological replicates (Pearson correction factor > 0.9), suggesting that the quantification analysis had high reliability (Figure S4A). To illustrate the difference between the two groups, a heatmap was generated using the differentially expressed proteins (DEPs) (Figure 4B). Gene Ontology analysis revealed that DEPs were enriched in cell components, such as cytoplasm, nucleus, and EVs (Figure 4C). Our attention was drawn to proteins involved in sEV secretion, and 11 proteins were filtered by overlapping DEPs and proteins involved in the sEV secretion process (44) (Figure 4D). Eight proteins related to sEV secretion, namely, YKT6, CD9, VAMP7, CD63, RAB9A, STAM, NDRG1, and LITAF, were upregulated in B3GALT4-overexpressed cells, and 3 proteins, namely, TP53, NAPG, and VPS4B, were downregulated (Figure 4E). Proteomic analysis revealed that 11 differentially expressed sEV secretion-related proteins might be responsible for the altered sEV secretion. Furthermore, proteomic analysis of B3GALT4-overexpressed and parental MDA-MB-231 cells identified 26 DEPs defined as oncogenic proteins (45) (Figure S4B). Of these proteins, 15 identified oncogenic proteins were documented to be presented in sEV (46) (Figure S4C), and 8 upregulated oncogenic proteins were associated with the EMT process, indicating that these oncogenic proteins might be transferred to recipient cells via sEV. The data indicated that enhanced GM1 levels might elevate oncogenic protein levels in donor and recipient cells, and thus facilitate the EMT process.

To explore the biological function of vesicular GM1, sEV with different levels of GM1 was fed to MCF-10A cells. Compared to the control group, the migratory ability of MCF-10A cells was promoted by MB231-sEV, and further promoted by B3GALT4-sEV, but suppressed by shB3GALT4-sEV treatment (Figures 5A, B). Increased expression of the mesenchymal marker of fibronectin and vimentin, and decreased expression of E-cadherin were detected in B3GALT4-sEV-treated MCF-10A. shB3GALT4-sEV-treated cells displayed opposite changes in EMT marker expression (Figures 5C, D). The expression of GM1 in MCF-10A cells was increased after B3GALT4-sEV treatment compared to MB231-sEV treatment, but decreased by shB3GALT4-sEV treatment evaluated by flow cytometry (Figures 5E, F). These data indicated that vesicular GM1 could be transferred into recipient cells and further promote cell migration and induce the EMT process.