Fig 2: Decreased expression levels of eGPC3 in patients with gastro-esophageal adenocarcinomas (GEA). (A) Nanoparticle tracking analysis (NTA) diagram of nanoparticle concentration and average particle size from serum samples of patients with GEA (n = 9; blue line), healthy donors (n = 9; red line) and patients with non-malignant disease (n = 9; green line). (B) Transmission electron microscopy (TEM) images of serum derived vesicles (black arrowheads). (C) Quantification of exosome-associated proteins CD9, CD63 and TSG101 determined by fluorescent NTA (p < 0.05, two-paired student´s t-test). (D) Protein expression analyses of exosome-derived GPC3, TSG101, CD81, CD63, and CD9. Calreticulin was used as a marker for cytoplasm-derived contamination (bottom panel). Representative quantification of GPC3 levels of serum-derived exosomes from patients with GEA compared to serum-derived exosomes from healthy donors and patients with non-malignant disease (right panel, p < 0.05 two-paired student´s t-test).

Fig 3: Decreased serum levels of eGPC3 in GEA patients. (A) Dot plot analysis of the percentage of GPC3 positive exosome-bound latex beads in serum from patients with GEA, patients with non-malignant disease and healthy donors. (B) Receiver operating characteristics (ROC) curve analysis of eGPC3 (blue), exosome concentration (black), and exosome size (dashed line) for patients with GEA vs. patients with non-malignant disease and healthy donors. (C) ROC curve analysis showing the area under the curve (AUC) for eGPC3, exosome concentration, exosome size, serum CA 19-9, serum CA 72-4, serum CEA and a combined score (eGPC3* CA 72-4)/CA 19-9) for patients with GEA vs. patients with non-malignant disease and healthy donors. (D) Protein expression analysis of GPC3 and CD9 in total serum and corresponding exosomes from five patients with GEA (left panel). Quantification of GPC3 and CD9 protein expression is shown on the left panel (right panel; p < 0.05).

Fig 4: TFPI binds to GPC3 on the membrane of CLL cells. (A) Western blotting and (B) quantification of TFPI protein in the cell lysates of CLL patients, relative to ß-actin, after 0, 50, 100 and 200 ng/mL rTFPI treatment. n = 3 individuals; *P < 0.05. (C) After 24 h treatment of 200 ng/mL rTFPI, co-localization of TFPI (red) and GPC3 (green) on the surface of human CLL cell line HG3 was assessed by immunofluorescence staining. Nuclei were stained with DAPI (blue). (D) Co-IP was performed in the HG3 and K562 cells. Immunoblots showed the bait protein TFPI and the pull-down of the prey protein GPC3. (E) Western blotting and (F) quantification of CXCR7 expression, relative to ß-actin, after CLL cells were pre-treated with GPC3 antibody HS20 (100 µg/mL) for 1 h prior to 24 h treatment of 200 ng/mL rTFPI. Human IgG was used as control since HS20 is isolated from human serum. n = 4 individuals; *P < 0.05, **P < 0.01. (G) After CLL cells were pre-treated with HS20 before rTFPI treatment, the cells were washed and applied to TEM assay. Human IgG was used as control. Results were presented as we described in Fig. 2A. n = 10 individuals; **P < 0.01.