New blocking antibodies impede adhesion, migration and survival of ovarian cancer cells, highlighting MFGE8 as a potential therapeutic target of human ovarian carcinoma

Abstract

Milk Fat Globule--EGF--factor VIII (MFGE8), also called lactadherin, is a secreted protein, which binds extracellularly to phosphatidylserine and to αvβ3 and αvβ5 integrins. On human and mouse cells expressing these integrins, such as endothelial cells, phagocytes and some tumors, MFGE8/lactadherin has been shown to promote survival, epithelial to mesenchymal transition and phagocytosis. A protumoral function of MFGE8 has consequently been documented for a few types of human cancers, including melanoma, a subtype of breast cancers, and bladder carcinoma. Inhibiting the functions of MFGE8 could thus represent a new type of therapy for human cancers. Here, we show by immunohistochemistry on a collection of human ovarian cancers that MFGE8 is overexpressed in 45% of these tumors, and we confirm that it is specifically overexpressed in the triple-negative subtype of human breast cancers. We have established new in vitro assays to measure the effect of MFGE8 on survival, adhesion and migration of human ovarian and triple-negative breast cancer cell lines. Using these assays, we could identify new MFGE8-specific monoclonal antibodies, which efficiently blocked these three tumor-promoting effects of MFGE8. Our results suggest future use of MFGE8-blocking antibodies as new anti-cancer therapeutics in subgroups of ovarian carcinoma, and triple-negative breast carcinoma patients.

Figure 1. Expression of MFGE8 in biopsies of human ovarian carcinoma.

(a) Examples of Human MFGE8 immunohistochemistry on ovarian carcinoma sections showing: no staining of tumor cells (score 0), but MFGE8-positive endothelial cells (black arrow) in the left panel. Areas of weak expression (w) and of medium expression (m) of MFGE8 in the same section, thus ranked as score 2 in the middle panel. Strong MFGE8 expression (s) throughout the tumor, ranked as score 3 (right panel). (b) Scoring of MFGE8 expression (y-axis) in 48 ovarian tumor biopsies from Institut Curie’s patients, as a function of tumor grade (x-axis). No statistically significant difference was observed in distribution of tumors overexpressing MFGE8 (score 2-3) or not (score 0-1) in grade 2 versus grade 3 tumors.

Figure 2. Expression of MFGE8 and its receptors in human ovarian cancer cell lines.

(a) Analysis of mRNA expression levels (Log₂(Affymetrix U133plus2.0 signal)) of MFGE8 in public results of microarray data of the Broad-Novartis Cancer Cell Line Encyclopedia. The expression level threshold behind which it could be considered as noise (MAS5 “Absence” flag status, gene not expressed) is 6.1 for this probeset. Black arrows indicate the SKOV-3 and IGROV-1 cell lines. (b) Quantification of MFGE8 by ELISA in SKOV-3, IGROV-1 and SHIN-3 conditioned medium (24 hr). Secreted MFGE8 is expressed in ng/mL per 10⁶ cultured cells. Mean of two experiments + SD. (c) Analysis of mRNA expression levels (Log₂(Affymetrix U133plus2.0 signal)) of ITGB3 (red) and ITGB5 (green) in public results of microarray data of the Broad-Novartis Cancer Cell Line Encyclopedia. The expression level threshold behind which it could be considered as noise (MAS5 “Absence” flag status, gene not expressed) is 4.8 for the ITGB3 probeset. (d) FACS analysis of αvβ3 (upper panel) and αvβ5 (lower panel) integrins expression at the surface of SKOV-3, IGROV-1 and SHIN-3 cells. Specific antibody (blue line). Isotype control (red line).

Figure 3. In vitro adhesion, migration and survival assays to measure effect of MFGE8 on SKOV-3 cells.

(a–e) Assay for adhesion of SKOV-3 to MFGE8 using the xCELLigence system. (a) Example of cell index (CI) as a function of time, on PBS or human MFGE8 (huMFGE8), in the presence of anti-huMFGE8 or control IgG. Blue and red vertical lines indicate the time points used for calculating slope values. (b) Adhesion of SKOV-3 to MFGE8, as compared to PBS (Ctrl), quantified by slope value between 0 and 2h. (c) Inhibition of adhesion to 5 µg/mL MFGE8 by 10 µg/mL anti-huMFGE8 (hMc3) or a negative control (Rituximab, Ritux). (d) Effect on adhesion to 5 µg/mL MFGE8 (Ctrl) of polyclonal anti-huMFGE8 sera (1/100) specific for the RGD-domain (anti RGD), the C1 domain (anti C1) or the RGD domain of mouse Mfge8 (anti mouse RGD). (e) Inhibition of adhesion to 5 µg/mL MFGE8 (Ctrl) by anti-αvβ3 and/or -αvβ5 integrin antibodies (5µg/mL each). (f–g) Assay for migration of SKOV-3 using xCELLigence. Cells were seeded in the upper chamber, and MFGE8 and antibodies in the lower chamber of CIM-plates. (f) Example of cell index (y axis) as a function of time (x axis), in similar conditions as (a). Blue and red vertical lines indicate the time points used for calculating slope values. (g) Dose-dependent effect of MFGE8 on SKOV-3 migration, as compared to PBS (Ctrl), quantified by slope value between 0 and 18 hours. (h) Inhibition of migration to 5 µg/mL huMFGE8 by 10µg/mL anti-huMFGE8 hMc3, as compared to negative control (Ritux). (i–j) Survival assay of SKOV-3 cultured for 96 hours in the presence of 0.1% serum. Cell number is quantified as fluorescence arbitrary units (A.U.) as compared to the control condition corresponding to 100%. (i) Dose-dependent effect of MFGE8 on SKOV-3 survival. 100% = basal level of A.U. in the absence of exogenous MFGE8 (Ctrl). (j) Inhibition of survival in the presence of 10 µg/mL huMFGE8, by 10µg/mL hMc3 or the negative antibody. 100% = basal level of A.U. in the presence of MFGE8. Data are represented as Box-and-whiskers graphs showing the minimum (lower error bar), 25^th percentile-median-75^th percentile and maximum (upper error bar) values. N=4 (except 3j: N=5). *** = P<0.001; ** = P<0.01; * = P<0.05 compared to Ctrl = PBS in 3b, g; to 0.312 µg/ml MFGE8 in 3i; or to Ctrl = MFGE8 in 3c, d, e, h, j..

Figure 4. Identification of new antibodies blocking either adhesion, migration or survival induced by MFGE8.

(a) Effect of 10 new anti-MFGE8 antibodies (10 or 50 µg/mL) on adhesion to MFGE8 (adhesion assay), as compared to hMC3. 100% adhesion corresponds to 5 µg/mL MFGE8 coating without antibodies (ctrl). Example of the screening experiments, performed once with each of the 40 new antibodies. Black arrows indicate the antibodies selected for further characterization. (b) Dose–response effect of the 5 selected antibodies, as compared to hMC3 in the adhesion assay. (c) Dose–response effect of the 5 selected antibodies, as compared to hMC3 in the migration assay. 100% migration corresponds to 5 µg/mL MFGE8 without antibodies. (d) Dose–response effect of the 5 selected antibodies, as compared to hMC3 in the survival assay. 100% survival corresponds to 10 µg/mL recombinant MFGE8 without antibodies. Box-and-whiskers graphs as in Figure 3. N=4 (except in (c) for 215A9 -10 µg/mL and 311A7 -0.625 µg/mL: N=3 and for hMc3: N=2; in (b) for 215A9 -5µg/mL: N=2). *** = P<0.001; ** = P<0.01; * = P<0.05 compared to the lowest antibody concentration (0.5 or 0.625 µg/ml) for each individual antibody in 4b, c, d.

Figure 5. Effect of MFGE8 and anti-MFGE8 antibodies on IGROV-1 and SHIN-3 ovarian cancer cells.

(a–b) Adhesion of IGROV-1 (a) and SHIN-3 (b) to 5 µg/mL MFGE8, as compared to PBS, quantified by slope value between 0 and 2h. The effect of three selected antibodies delivered at two doses (2,5 and 10 µg/mL) on MFGE8-coated wells is also shown. (c–d) IGROV-1 (c) and SHIN-3 (d) migration assay to 5µg/mL MFGE8 as in Figure 3-4, and inhibition by antibodies as in panels a and b for adhesion. Slopes values calculated between 0 and 18 hours. (d–e) IGROV-1 (d) and SHIN-3 (e) survival assay in the presence of 10µg/mL MFGE8, as in Figure 3-4, and inhibition by 2,5 or 10µg/mL antibodies. 100% survival corresponds to 10 µg/mL MFGE8 without antibodies. Box-and-whiskers graphs as in Figure 3. N=4 (except in (b): 399A12 -2.5 µg/mL, 346B6 -10 µg/mL, 311A7 -10 µg/mL: N = 3; in (e, f) PBS and MFGE8 : N=5; in (f) MFGE8 + 311A7: N=3). *** = P<0.001; ** = P<0.01; * = P<0.05, compared to Ctrl = MFGE8.

Figure 6. Expression of MFGE8 in human breast carcinoma and effect of MFGE8 on MDA-MB-231 cells.

(a–b) Scoring of MFGE8 expression, analyzed by immunohistochemistry as in Figure 1, in 59 breast tumor biopsies from Institut Curie’s patients, as a function of tumor grade (a) or phenotype in terms of Hormon Receptor (HR) or HER2 expression (b). ***: P = 0.0001 for distribution of tumors overexpressing (score 2-3) or not (score 0-1) MFGE8 between grade 2 versus grade 3 tumors (a), or between HR+/HER2- versus triple-negative tumors (b). (c) Analysis of mRNA expression levels (Log₂(Affymetrix U133plus2.0 signal)) of MFGE8 in public results of Affymetrix microarray data of the Broad-Novartis Cancer Cell Line Encyclopedia as in Figure 2. Black arrow indicates the MDA-MB-231 cell line (HR-/HER2- triple negative phenotype) selected for subsequent experiments. (d) Quantification of MFGE8 by ELISA in MDA-MB-231 conditioned medium (24 hr). Secreted MFGE8 is expressed in ng/mL per 10⁶ cultured cells. Mean of two experiments + SD. (e) FACS analysis of αvβ3 (left panel) and αvβ5 (right panel) integrin expression at the surface of MDA-MB-231 cells. Specific antibody (blue line). Isotype control (red line). (f) MDA-MB-231 adhesion assay on 5µg/mL MFGE8, as in Figure 3-5, and inhibition by 10µg/mL hMC3, represented as slope values between 0 and 1 hour. (g) MDA-MB-231 migration assay to 5µg/mL MFGE8 as in Figure 3-5, and inhibition by 10µg/mL 215A9 antibody. Data represented as slope values calculated between 0 and 18 hours. (h) MDA-MB-231 survival assay in the presence of 10 µg/mL MFGE8, as in Figure 3-5, and inhibition by 10 µg/mL 399A12. 100% survival corresponds to 5µg/mL MFGE8 without antibodies. Box-and-whiskers graphs as in Figure 3. N=4 (except in (g): PBS N=6, MFGE8 + 215A9 N=5, and in (h): ctrl N=6, MFGE8: N=5). ***: P<0.001, **: P<0.01, compared to Ctrl = MFGE8.