Microcell-mediated chromosome transfer identifies EPB41L3 as a functional suppressor of epithelial ovarian cancers

Abstract

We used a functional complementation approach to identify tumor-suppressor genes and putative therapeutic targets for ovarian cancer. Microcell-mediated transfer of chromosome 18 in the ovarian cancer cell line TOV21G induced in vitro and in vivo neoplastic suppression. Gene expression microarray profiling in TOV21G(+18) hybrids identified 14 candidate genes on chromosome 18 that were significantly overexpressed and therefore associated with neoplastic suppression. Further analysis of messenger RNA and protein expression for these genes in additional ovarian cancer cell lines indicated that EPB41L3 (erythrocyte membrane protein band 4.1-like 3, alternative names DAL-1 and 4.1B) was a candidate ovarian cancer-suppressor gene. Immunoblot analysis showed that EPB41L3 was activated in TOV21G(+18) hybrids, expressed in normal ovarian epithelial cell lines, but was absent in 15 (78%) of 19 ovarian cancer cell lines. Using immunohistochemistry, 66% of 794 invasive ovarian tumors showed no EPB41L3 expression compared with only 24% of benign ovarian tumors and 0% of normal ovarian epithelial tissues. EPB41L3 was extensively methylated in ovarian cancer cell lines and primary ovarian tumors compared with normal tissues (P = .00004), suggesting this may be the mechanism of gene inactivation in ovarian cancers. Constitutive reexpression of EPB41L3 in a three-dimensional multicellular spheroid model of ovarian cancer caused significant growth suppression and induced apoptosis. Transmission and scanning electron microscopy demonstrated many similarities between EPB41L3-expressing cells and chromosome 18 donor-recipient hybrids, suggesting that EPB41L3 is the gene responsible for neoplastic suppression after chromosome 18 transfer. Finally, an inducible model of EPB41L3 expression in three-dimensional spheroids confirmed that reexpression of EPB41L3 induces extensive apoptotic cell death in ovarian cancers.

Figure 1

(A) Genomic mapping and candidate gene identification in TOV21G⁺¹⁸ hybrids confirm the transfer of chromosome 18 material in parental cancer cell lines. (i) Metaphase fluorescence in situ hybridization (FISH) using a chromosome 18 paint of TOV21G cells (top panel) and TOV21G⁺¹⁸ cells (bottom panel) confirms the transfer of a single additional copy of chromosome 18 (green fluorescent staining) in the hybrid cells. (ii) Microarray CGH analysis confirms the results of FISH and provide more detailed mapping information of TOV21G⁺¹⁸ cells. Microarray CGH profiles show copy number difference between parental and hybrid cell line for a tiling path of bacterial artificial chromosomes (BACs) spanning the length of chromosome 18. Green spots indicate single copy number gain. Mapping showed the same regions transferred for both hybrids used in expression analysis of each cell line. (iii) The location of genes on chromosome 18 identified by differential gene expressionmicroarray analysis of the TOV21G cancer cell line compared with twoMMCT hybrids. The candidate gene EPB41L3, which was taken forward for further analyzes, is highlighted in red. (iv) The expression fold change for the EPB41L3 gene and flanking genes in TOV21G cells suggest the activation of EPB41L3. (B) Volcano plot showing the gene expression data from chromosome 18 in TOV21G⁺¹⁸ hybrids compared with TOV21G cancer cells. The color point showing the magnitude of fold change in EPB41L3 gene expression (x-axis), coupled with statistical significance (y-axis:-log₁₀ of the P value). (C) Immunoblot analysis using a EPB41L3monoclonal antibody suggests a lack of expression in TOV21G cells but strong expression in both TOV21G⁺¹⁸ hybrids (MMCT18G1 and MMCT18G2), suggesting that EPB41L3 is activated in hybrid cells. EPB41L3 is also highly expressed in two primary normal ovarian surface epithelial cell cultures (NOSE11 and NOSE19) and in an immortal ovarian surface epithelial cell line (IOSE4); 110 kDa is the expected and observed size of the EPB41L3 protein band.

Figure 2

Evaluating EPB41L3 status in primary ovarian cancers, ovarian cancer cell lines, and normal ovarian epithelial cell lines and tissues by immunohistochemistry and methylation analyses. (A) Immunoblot analysis for EPB41L3 in ovarian cancer cell lines shows absence of expression in 15 (79%) of 19 cell lines. (B) Illustration of EPB41L3 expression analyzed in 794 invasive ovarian cancer specimens (stratified by histologic subtype) and 33 benign ovarian tumors. Staining values: 0, less than 5% of neoplastic cells stain positive; 1, 5% to 20%; 2, 20% to 50%; 3, more than 50%. Immunohistochemistry was performed for invasive ovarian cancers and benign tumors established as tissue arrays taken from three ovarian tumor tissue collections: the Danish MALOVA study (488 ovarian tumors) [23], the Derby City Hospital tumor array (263 ovarian tumors), and the Newcastle ovarian cancer tissue micro array (160 invasive tumors) [41]. Normal primary epithelial tissues and additional invasive ovarian tumors were provided by the University College London Hospital ovarian tumor tissue biobank. (C) Examples of immunohistochemical staining with EPB41L3. Where we observed EPB41L3 staining, it suggested that EPB41L3 protein expression occurs uniformly throughout the cytoplasm and in cell membranes of both tumor and normal cells. All panels are 200x magnification: (i) 100% expression in an endometrioid ovarian cancer, (ii) 90% expression in a serous ovarian cancer, (iii) 0% expression in an endometrioid ovarian cancer, (iv) 50% expression in an endometrioid ovarian tumor, (v) 10% expression in a serous ovarian tumor (vi), 20% expression in a serous ovarian cancer, (vii) 30% expression in an endometrioid ovarian cancer, (viii) 100% expression in normal ovarian surface epithelial cells, and (ix) 100% expression in normal ovarian epithelial cells in an inclusion cyst. IC indicates inclusion cyst; NOE, normal ovarian epithelium; NSE, negative staining epithelium; PSE, positive staining epithelium. (D) Heat map illustrating methylation analysis of the EPB41L3 promoter in normal ovarian surface epithelial cell lines and 16 ovarian cancer cell lines. Each feature represents the methylation status at a CpG dinucleotide in the promoter region. The darker the feature, the more hypermethylated the CpG. Combining the data, for all CpGs, 74% of CpGs showed greater than 60% methylation in ovarian cancer cell lines; only 1% of CpGs were more than 60% methylated in the normal ovarian epithelial cell lines.

Figure 3

Functional effects of reexpressing EPB41L3 cDNA in the TOV21G ovarian cancer cell line. (A) Spheroid formation in (i) TOV21G cells, (ii) TOV21G^+GFP cells, (iii) TOV21G^+GFP cells under fluorescent microscopy showing expression of green fluorescent protein, (iv) TOV21G⁺¹⁸ hybrid, and (v) TOV21G^+EPB41L3 cells. (B) Analysis of average spheroid volume and area for six replicates in two independent experiments in the ovarian cancer cell lines TOV21G, INTOV2, and SCOV3; their respective GFP-transfected lines TOV21G^+GFP, INTOV2^+GFP, and SCOV3^+GFP; EPB41L3-expressing spheroids TOV21G^+EPB41L3, INTOV2^+EPB41L3, and SCOV3^+EPB41L3; and the chromosome 18-transferred spheroids, TOV21G⁺¹⁸. EPB41L3-expressing and chromosome 18-transferred spheroids show significantly reduced size and volume compared with parent and GFP-expressing cancer cell lines. (C) SEM/TEM analysis of TOV21G spheroids: (i) TEM analysis of TOV21G spheroids shows a well-defined compact structure. (ii) TEM analysis of TOV21G^+EPB41L3 spheroids shows a more diffuse structure. (iii) At higher magnification, TOV21G spheroids display a smooth outer surface, indicative of good cellular interactions and secretion of extracellularmatrix. (iv) In contrast, TOV21G^+EPB41L3 spheroids show degenerating cellmembranes, characteristic of apoptosis. (v) Electron micrographs of sections of TOV21G spheroids show polarized cells and the formation of lumens (L) (a characteristic of epithelial cells). (vi) Micrographs TOV21G^+EPB41L3 sections show chromatin condensation, large numbers of phagocytosomes, and vacuoles (black arrows), which are characteristics of apoptotic cells. (vii, viii) Striking similarities between TOV21G⁺¹⁸ and TOV21G^+EPB41L3 spheroids, which include large cell surface protrusions (filopodia and microspikes). (D) FACS analysis for simultaneous staining of annexin V (apoptotic cells) and propidium iodide (necrotic cells) show increased levels of apoptosis in TOV21G⁺¹⁸ hybrids and EPB41L3-expressing cells compared with untransfected TOV21G, INTOV2, and SCOV3 cell lines and their respective GFP-expressing cells. (E) Inducing EPB41L3 expression in TOV21G spheroids. Fluorescent live-dead viability assays compare the proportion of live (viable) spheroids (green fluorescence) and dead (apoptotic) spheroids (red fluorescence). (i) Two days after induction of EPB41L3, more than 90%of spheroids are viable (ratio of live to dead spheroids is 15:1). (ii) Ten days after induction, more than 70% of the spheroids had undergone apoptosis (live-dead ratio 1:3). (iii) Some apoptosiswas also seen in noninduced spheroids after 10 days (<20%; live-dead ratio 6:1). (v) Higher magnification shows live-dead spheroids 10 days after induction of EPB41L3 expression.