Using differential RT-PCR techniques, the expression of EGF-related genes was compared in three ovarian carcinoma cell lines and their normal counterpart, ovarian surface epithelial (OSE) cell strains. Two carcinoma cell lines, even when grown in serum-free medium, concurrently and simultaneously expressed high levels of mRNA for Transforming Growth Factor-Alpha (TGF-a), Epidermal Growth Factor Receptor ( EGFR), and erbB-2 receptor. mRNA expression was not detected for any of the EGF Supergene Family ligands or receptors in three OSE cell strains even when grown in the presence of serum. These results are consistent with an autocrine involvement of these three gene products in in vitro ovarian carcinoma cell growth. The expression of the EGF family of ligands and receptors was also examined in the several histologic stages of human ovarian carcinogenesis. Primary and metastatic ovarian cystadenocarcinomas, ovarian carcinomas of low malignant potential (borderline tumors), benign ovarian cystadenomas and normal ovaries were compared for immunoperoxidase detection of the ligands Epidermal Growth Factor (EGF), TGF-a, amphiregulin (AR), cripto, and the receptors EGFR and c-erbB-2. In the case of AR and TGF-a, this matrix analysis indicated a specific pattern of ligand detection with a particular ovarian histologic category in that AR immunostaining was detected almost exclusively in borderline tumors. Secondly, TGF-a differed from EGF in the borderline tumors, in that EGF was uniformly positive while TGF-a was least frequently positive. Thirdly, TGF-a immunoreactivity in the absence of coexpression of cripto or EGF appeared to be associated only with adenocarcinomas of high grade and stage. For example, all six of the stage I/II adenocarcinomas were immunoreactive for at least two of three of cripto, EGF, or TGF-a; in contrast, six of seven of the stage III carcinomas were immunopositive for only TGF-a. Interestingly, immunostaining for cripto, which is thought to be a """"""""tumor marker"""""""", was more common in normal ovarian surface epithelium than in cystadenomas or cystadenocarcinomas. The preferential association of AR immunoreactivity in ovarian carcinomas of low malignant potential and TGF-a in advanced carcinomas was then evaluated using various RT-PCR methods. Differential display PCR, using GAPDH as the housekeeping reference gene, proved to be only semi-quantitative when 1) using densitometry after hybridization with specific DIG-labelled probes or 2) scoring with fluorescence staining of PCR cDNA products. However, the PCR cycle number within which efficient amplification occurs for these mRNAs was established by the latter method. Currently, we use a quantitative RT-PCR method which uses an internal plasmid RNA competitive reference standard with identical sequences to our target genes of interest.
