Cancer Immunity 12:3 (2012)

When Toshitada Takahashi was in Nagoya, Japan, in 1972, one year after he left New York, he received a phone call from Lloyd Old, asking him to return to the Sloan-Kettering Institute (SKI); Dr. Old wished to organize a human tumor immunology group. By the early 1970s, tumor-associated antigens such as carcinoembryonic antigen (CEA) and -fetoprotein had been defined by analysis of heteroimmune sera, and the association of Epstein-Barr virus (EBV) with Burkitt’s lymphoma and nasopharyngeal carcinoma had been demonstrated. However, the most important unanswered question was whether tumor cells express antigens that are able to induce host immune responses, ultimately resulting in tumor regression. Around that time, attempts were made by various investigators to define such antigens, predominantly by analyses of the reactions obtained by allogeneic combination of cultured tumor cells and sera and/or lymphocytes from patients. At SKI, the members of the cell-mediated immunity subgroup under Herbert F. Oettgen, Michael A. Bean, and Yoshihisa Kodera were already very actively working on melanoma patients. So, the newly organized virology subgroup of Gaetano Giraldo and our serology subgroup joined together to form the human tumor immunology group headed by Dr. Old, who was just promoted to vice president of Memorial Sloan-Kettering Cancer Center (MSKCC) and associate director of SKI. When Dr. Takahashi met Dr. Old for the first time in his office to initiate a group of human cancer serology in 1973, Dr. Old told him that there was not yet a textbook for human tumor immunology, and that they would be exploring a new research field. There was a concern that analyses based on the allogeneic combination of cultured tumor cells and sera and/or lymphocytes from patients may result in detection of alloantigens, such as blood group antigens and histocompatibility antigens, rather than tumor antigens. Consequently, we decided to apply our approach of autologous typing with serological techniques; in other words, we restricted our analysis to the study of autologous reactions (reaction between sera and tumor cells from the same patient). We primarily chose analyses of serology rather than cellmediated immunity because of clarity of specificity (1). We also had much experience, in the 1960s, in serology for murine immunogenetic research. For serological techniques, we used rosette assays, such as the immune adherence (IA) assay to detect complement-dependent antibody, predominantly IgM, and mixed hemadsorption assay (MHA) for IgG detection; both techniques are very sensitive and convenient ways to detect surface antigens on cells grown in monolayers, and both are superior to complement-dependent cytotoxicity microassays. The most difficult part of autologous typing was deriving cultures of target tumor cells from solid tumors, but we were very lucky to have Lois A. Resnick working as a technical assistant in our laboratory, a very hard worker with ‘magic hands’ for cell culture. Thanks to the careful arrangement by Dr. Oettgen, we were able to obtain and cultivate various types of tumors, and found that the highest success rate was with melanoma (25-30%), with slightly lower success with renal cancer and glioblastoma (10-20%). The success rates for other tumors were much lower, and we therefore chose these three tumor types for further study. We conducted autologous typing of sera from 75 melanoma patients. Four patients were found to have antibodies identifying individually distinct tumor antigens (class I), and five patients had antibodies detecting shared tumor antigens (class II). Among class I, the AU antigen, defined by Thomas E. Carey (2), was further studied, since antibody titers residing in the IgG fraction were relatively high, up to 1/256 by MHA; however, characterization of the AU antigen by the conventional radioimmunoprecipitation technique was not possible. Instead, we used antibody inhibition assays to follow antigen solubilization and characterization. AU antigen is easily solubilized by papain and has a molecular weight in the range of 20,000-50,000; unfortunately, the gene encoding for this antigen was not isolated (3). Among class II, the AH antigen was most extensively analyzed. This antigen was defined by Hiroshi Shiku from Nagoya, who joined as a member of the serology group (4, 5), while working with Eiichi Nakayama on Lyt-phenotyping of T cell functions. AH antibody was present in the IgM fraction of the sera of a melanoma patient who had remained alive for six years after resection of recurrent melanoma. This antigen has been found on 70% of melanomas and almost all glioblastomas, but not on normal cultured cells. Biochemical characterization of AH antigen was attempted, but it proved very difficult because of our limited knowledge and experience of membrane chemistry. We asked Kenneth O. Lloyd, from the Texas Tech University School of Medicine in Lubbock, to join as a member of the human tumor immunology group; he kindly accepted our offer in 1976. In 1982, Takeo Watanabe and Kenneth Lloyd demonstrated that the AH antigen molecule resides on a glycolipid molecule, GD2 (6). It is noteworthy in this context that analysis of mouse monoclonal antibodies against melanoma also demonstrated that GD2 and other gangliosides such as GD3 and GM2 are tumor-associated antigens (7). The order of immunogenicity of gangliosides in humans appears to be first GD2, then GM2, and last GD3, whereas in mice, GD3 is the