Figure 1.
Culture and characterization of normal ovarian epithelial and fallopian tube epithelial cells.
A, Replicate plates of normal fallopian tube epithelial cells were cultured in WIT-fo medium (green and purple lines) or standard medium (red, grey lines, 1:1 mixture of Dulbecco’s Modified Eagle’s Medium (DMEM) and Ham’s F12, supplemented with 0.1% BSA, 5% serum). In WIT-fo medium, normal fallopian tube epithelial cells from two different patients divided continuously for at least 30 days and > 100 days and reached at least 12 and >20 population doublings respectively (green and purple lines). In contrast, matched cells from the same donors growth arrested in the DMEM/Ham’s F12 medium (red, grey lines). Normal fallopian tube epithelial cells were isolated from additional patients 3 and 4. B, Normal ovarian epithelial cells were cultured in WIT-fo medium (purple and blue lines) or standard medium (red line, MCDB 105/Medium 199 (M199) (1∶1 mixture) with 10% FBS and 2 mm l-glutamine). Primary normal ovarian epithelial cells were from patients 2 and 3 (cells cultured in standard medium were from patient 3). Matched cells from patient 3 growth arrested in the MCDB105/M199 medium (red line). C-D, Normal human ovarian tissue; immunoperoxidase staining of formalin-fixed paraffin embedded (FFPE) sections with PAX8 demonstrates that ovarian inclusion cyst epithelium is PAX8+ (brown nuclear stain) (C-D) while ovarian surface epithelium is in general PAX8 negative (C), rare presence of rare Pax8 positive cells have been reposted on the ovarian surface. E, Normal human fallopian tube tissue; double immunoperoxidase staining of FFPE sections shows ciliated cells are FOXJ1+ (nuclear brown) and non-ciliated cells are PAX8+ (nuclear red). F, Normal human fallopian tube tissue; double immunoperoxidase staining of FFPE sections shows that ciliated cells are FOXJ1+ (nuclear brown) and non-ciliated cells are CK7+ (cytoplasmic red) (scale bar (C-F) = 20 µM). G, Summary of cell type specific characterization markers.
Figure 2.
Immortalization of ovarian and fallopian tube epithelial cells with hTERT.
A, Immunoblotting of whole cell extracts demonstrates that hTERT immortalized ovarian epithelium (OCE) and fallopian tube non-ciliated epithelium (FNE) were CK7+/PAX8+/FOXJ1¯. Pig kidney cells (LLC-PK1) were included as positive controls for FOXJ1. Each bar marks 52 kDa. B, OCE cells from patient 1 were cultured in WIT-fo medium (blue line) or standard medium (red line, MCDB 105/Medium 199 (M199) (1∶1 mixture) with 10% fetal bovine serum and 2 mm l-glutamine). In WIT-fo medium OCE cells divided continuously for >150 days and reached 70 population doublings (blue line) while matched cells from the same donor growth arrested in the control medium (red line). C, FNE cells from patient 1 were cultured in WIT-fo medium (blue line) or in standard medium (red line, 1∶1 mixture of Dulbecco’s Modified Eagle’s Medium (DMEM) and Ham’s F12, supplemented with 0.1% BSA and 5% serum). In WIT-fo medium FNE cells divided continuously for >150 days and reached 40 population doublings (blue line). In contrast, matched cells from the same donor growth arrested in the standard medium (red line). R2 values indicate the fit of a polynomial line to the data points.
Figure 3.
Validation of FNE versus OCE cell-of-origin gene signature in two independent ovarian cancer datasets.
A, Density plot shows a bimodal distribution of signature scores for 99 ovarian tumor samples and four normal OSE scrapings from the Wu et al. dataset. Histograms demonstrate the classification of these samples into normal ovarian epithelial like (OV-like) and normal fallopian tube epithelial like (FT-like) subgroups (top pane), and show the range of signature scores by tumor histological subtype (lower panes). B, Association of the OV-like/FT-like tumor classification in the Wu and Tothill datasets with clinical characteristics (P-values from logistic regression (grade, stage as ordinal variables) and Fisher’s Exact test (histological subtype)). C, Density plot demonstrates a slight left skewing of the signature scores in the Tothill et al. dataset which suggests a small subpopulation (arrow) of OV-like tumors. Histograms demonstrate the distribution of signature scores in endometrioid and serous tumors. D, Kaplan-Meier plots demonstrate significant differences in the disease-free survival and overall survival between OV-like and FT-like subgroups in the Tothill dataset (univariate P-values from the log-rank test are displayed).
Figure 4.
OCLER and FNLER tumor histopathology in immunodeficient nude (Nu/Nu) mice.
A-B, Hematoxylin and eosin (H&E) stains of representative formalin-fixed paraffin-embedded (FFPE) tumor sections from OCLER (A) and FNLER (B) xenografts revealed focal micropapillary structures. The predominant morphology was diffuse sheets of cells with a poorly differentiated tumor architecture (scale bar = 20 µM). C-D, PAX8 immunoperoxidase stains of representative FFPE tumor sections from OCLER (C) and FNLER (D) xenografts confirmed that xenografts retained their PAX8 expression (scale bar = 20 µM).
Table 1.
Tumor formation, tumor burden and ascites in the OCLER and FNLER xenograft model.