MAGE-1 is a gene that encodes an antigen on a melanoma cell line that is recognized by cytolytic T-cells. We have used a reverse transcription-polymerase chain reaction assay to analyze expression of the MAGE-1 gene by cell lines from different types of tumors, melanomas from different stages of disease progression, normal diploid cell lines, and melanocyte and nevus tissue from which malignant melanomas are derived. MAGE-1 is expressed by melanoma tissue from all stages of disease, but not melanocytes, nevus tissue, or any normal diploid cell line tested. A fraction of tumor lines derived from various epithelial and neuroectodermal malignancies expressed MAGE-1 but not peripheral blood cells from patients with melanoma. 5-Aza-2'-deoxycytidine (DAC), a demethylating agent, was capable of inducing MAGE-1 expression by a MAGE-1-negative melanoma cell line 888-mel as well as by a number of other melanoma cell lines. At an optimum concentration of 1 microM DAC, MAGE-1 expression was detectable by 24 h, plateaued by 72 h, but remained high for two weeks after removal of DAC from treated 888-mel cells, consistent with induction by demethylation. With the exception of tumor-infiltrating leukocytes, no normal diploid cell line could be induced with DAC to upregulate MAGE-1 expression. DAC-treated 888-mel cells were lysed by a MAGE-1-specific major histocompatibility complex restricted cytolytic T-cell clone, whereas control untreated cells were not, suggesting that production of the antigen encoded by the MAGE-1 gene was induced by DAC and that it was presented in association with major histocompatibility complex class I molecules at the cell surface for T-cell recognition.

A cell line derived from the mouse colon adenocarcinoma, MC-38, has been transduced with a retroviral construct containing complementary DNA encoding the human carcinoembryonic antigen (CEA) gene. MC-38, which forms tumors in syngeneic C57BL/6 mice, has been extensively studied as a target for active immunotherapy. Individual transduced clones that express high levels of cell surface CEA were isolated, and two clones, termed MC-38-ceal and MC-38-cea2, were extensively characterized. The levels of CEA found on the surface of these clones were considerably higher than that found in a moderately differentiated human colon carcinoma cell line (WiDr) and were comparable to those found on the human colon carcinoma cell lines GEO and CBS (among the highest CEA-expressing cells reported). Further analysis demonstrated that the CEA expressed in the MC-38-cea1 clone had a similar molecular weight to native CEA (Mr 180,000), but the MC-38-cea2 cell line expressed a single Mr 70,000 glycosylated immunoreactive product. Seven anti-CEA monoclonal antibodies were found to react with both clones. The CEA gene present in the MC-38-cea2 clone was partially sequenced and was found to contain a deletion of two of the three repeated domains present in CEA. These results provide a basis for future studies to map immunodominant epitopes of CEA and to develop a syngeneic model system that may aid in the design of reagents and protocols to study active and passive immunotherapy directed against a carcinoma expressing human CEA.

Using computer-assisted digital image-analysis, the level of ras and gli messenger RNA (mRNA) in neoplasms of meningeal and glial origin was determined in an attempt to correlate these parameters with histological tumor severity. We used single-stranded, asymmetric, radiolabelled RNA probes to detect the amount of ras and gli mRNA present in formalin-fixed, paraffin-embedded tumor material. Such archival material can provide an immediate, larger sample base than with fresh samples. The extent of ras mRNA expression in 25 brain tumors was not significantly different than normal cerebellum. However, five of 74 astrocytomas of relatively high malignant potential demonstrated gli mRNA levels above normal cerebellum.

We found a direct correlation between increasing ras p21 protein immunopositivity and severity of human glioma using computer-assisted, digital-image processing to quantify the amount of p21 immunoreactive to the monoclonal antibody RAP-5. We determined that there was a significant difference in reactivity between glioblastoma multiformes and more-differentiated astrocytomas (experiment-wise error less than 0.05). This result confirmed the conclusions made on the same tumors using standard light microscopy and visual examination. Immunohistochemistry quantized by automated image analysis may be a useful adjunct to current histopathological strategies since it decreases assay subjectivity and variation.